RISTO MALCHESKI

SAVA GROZDEV
KATERINA ANEVSKA

GEOMETRY
OF COMPLEX NUMBERS

2015

Reviewers: Prof. Dr. Lidia Ilievska,
Assoc. Prof. Dr. Veselin Nenkov

© Arhimed, Publisher, 2015

© Risto Malcheski, Sava Grozdev, Katerina Anevska, Authors, 2015
© Angelina Avramova, Graphic design, 2015

ISBN 978-954-799-

Contents
Preface ....................................................................................................... 5

CHAPTER I
COMPLEX NUMBERS

1. The concept of complex number, basic properties ....................... 7

2. Algebraic notation of a complex number .................................. 10
3. A conjugate complex number ..................................................... 11
4. Geometric presentation of a complex number . .......................... 16
5. Extended complex plane. Reimann interpretation
of complex number ..................................................................... 19
6. Trigonometric entry of a complex number ................................ 22
7. Roots of a complex number ........................................................ 24
8. Exponential entry of a complex number . ................................... 28
9. The set Cn. .................................................................................. 32

CHAPTER II
TRANSFORMATION IN EUCLEDIAN PLANE
1. Line equation. Parallel and perpendicular lines . ........................ 36
2. Distance from a point to a line . .................................................. 40
3. Circle equation . ......................................................................... 42
4. Direct similarities . ...................................................................... 46
5. Motions ....................................................................................... 51
6. Homothety .................................................................................. 56
7. Indirect similarity . ...................................................................... 62
8. Inversion .................................................................................... 70
9. Mőbius transformation . .............................................................. 80
10. Geometric properties of a Mőbius transformation . .................... 83

CHAPTER III
GEOMETRY OF CIRCLE AND TRIANGLE
1. Central and inscribed angles ....................................................... 89
2. Power of a point with respect to a circle...................................... 91
3. Radical axis and radical center ................................................... 95
4. A pencil and a bundle of circles . ................................................ 99
5. Orthocenter and centroid of a triangle ..................................... 103
6. Right angled triangle . ............................................................... 109

7. Euler line and Euler circle ........................................................ 111
8. Menelaus’ theorem . ................................................................. 114
9. Pascal’s and Desargues’ theorem .............................................. 116
10. Triangular coordinates .............................................................. 118
11. Ceva’s and Van Aubel’s theorem .............................................. 120
12. Area of a triangle ...................................................................... 123
13. Incircles and excircles of a triangle ......................................... 127
14. Stewart’s theorem ..................................................................... 133
15. Simpson line.............................................................................. 136
16. Ptolemy’s theorem .................................................................... 138
17. Inner product . ........................................................................... 140

CHAPTER IV
EXAMPLES AND EXERCISES
1. Examples (Chapter I) ................................................................ 144
2. Exercises (Chapter I) ................................................................ 159
3. Examples (Chapter II, Chapter III) ........................................... 164
4. Exercises (Chapter II, Chapter III) ........................................... 222

References . ............................................................................................ 236

PREFACE

No research of people could be named true science

if it is not supported by a mathematical proof.
Reliability of the assertions in different subjects is
problematic when application of any mathematical
domain is missing, i. e. when there is no connection
with mathematics.
Leonardo da Vinci

The present book is devoted to students of the last school grades, university
students, teachers, lecturers and all lovers of mathematics who want to enrich their
knowledge and skills in complex numbers and their numerous applications in Euclidean
Geometry. Few countries in the world include complex numbers in their secondary
school curriculum but even if included the volume of the corresponding content is quite
insufficient consisting of elementary operations and geometric representation at most. The
significance of the complex numbers is far from a real recognition in known textbooks
and scholar literature. The applications not only in mathematics but also in many other
subjects are considerable and the present book is a strong proof of such a statement.
Mainly, the book will be useful for outstanding students with high potentialities
in mathematics preparing themselves for successful participation in mathematical
competitions and Olympiads. Other target groups are not excluded, namely those, whose
representatives like to meet real challenges, connected with unexpected circumstances in
problem solving.
The material in the book is divided into four chapters. The first one contains
basic properties of the complex numbers, their algebraic notation, the notion of a
conjugate complex number, geometric, trigonometric and exponential presentations, also
interesting facts in connection with Reimann interpretation and the set Cn. The second
chapter includes various transformations of complex numbers in the Euclidean plane
like similarity, homothety, inversion and Mőbius transformation. The third chapter
is dedicated to the geometry of circle and triangle on the base of complex numbers.
Numerous theorems are proposed, namely: Menelau’s theorem, Pascal’s and Desargue’s
theorem, Ceva’s and Van Aubel’s theorem, Stewart’s theorem, Ptolemy’s theorem and
others. Exercises and problems are included in the Fourth chapter: 122 examples with
solutions and 161 solved problems pare proposed. Together with all the 138 theorems,
lemmas and corollaries accompanied by 64 examples and 88 figures, the book turns out
to be a rather exhaustive collection of the complex number applications in Euclidean
Geometry.

A high just appraisal of the book is due to the numerous non-standard problems
in it taken from the National Olympiads of Bulgaria, China, Iran, Japan, Korea, Poland,
Romania, Russia, Serbia, Turkey, Ukraine and others but also from Several International
and Balkan Mathematical Olympiads.
The authors express their sincere thanks to the Editorial House “Archimedes 2”
for the decision to accept the manuscript and to support the appearance of the present
book. Also, sincere thanks to the reviewers Prof. Dr. Lidia Ilievska and Assoc. Prof.
Dr. Veselin Nenkov for their helpful criticism, removal of mistakes and well-wishing
advices, which contributed to the final quality of the book. Of course, different lapses are
possible and we will be grateful to the readers in case they notice such and bring them to
the attention of the Editor.

February, 2015 The authors

Skopje and Sofia

CHAPTER I
COMPLEX NUMBERS

1. THE CONCEPT OF COMPLEX NUMBER,

BASIC PROPERTIES

1.1. Definition. Complex number z = (a, b) is the ordered pair of real numbers
a and b.
The set of complex numbers is denoted by C, i.e. C = {(a, b) | a, b ∈ R} .
The complex numbers (0,0) and (1,0) are denoted by n and e, respectively.
The definition of a complex number directly implies that two complex numbers
z1 = (a1 , b1 ) and z2 = (a2 , b2 ) are equal if a1 = a2 and b1 = b2 .

1.2. Definition. Sum of two complex numbers z1 = (a1 , b1 ) and z2 = (a2 , b2 ) is

the complex number
z1 + z2 = (a1 + a2 , b1 + b2 ) .

1.3. Definition. Product of two complex numbers z1 = (a1 , b1 ) and z2 = (a2 , b2 )

is the complex number
z1 ⋅ z2 = (a1a2 − b1b2 , a1b2 + a2b1 ) .

1.4. Theorem. The addition and multiplication of complex numbers, satisfy the
already known laws of arithmetic. Namely,:
i) z1 + z2 = z2 + z1 , commutative property of addition,
ii) ( z1 + z2 ) + z3 = z1 + ( z2 + z3 ) , associative property of addition,
iii) z1 z2 = z2 z1 , commutative property of multiplication,
iv) ( z1 z2 ) z3 = z1 ( z2 z3 ) , associative property of multiplication, and
v) ( z1 + z2 ) z3 = z1 z3 + z2 z3 , distributive property.
hold true for all complex numbers z1 , z2 , z3 .
Proof. i) Let z1 = (a1 , b1 ) and z2 = (a2 , b2 ) be any complex numbers. Thus,
(a1 , b1 ) + (a2 , b2 ) = (a1 + a2 , b1 + b2 ) = (a2 + a1 , b2 + b1 ) = (a2 , b2 ) + (a1 , b1 )
i.e., z1 + z2 = z2 + z1 .
The properties ii), iii), iv) and v) can be proven analogously. ■

Let’s state that when proving Theorem 1.4 we explicitly used (by coordinates)
the commutative, associative and distributive properties of addition and multiplication of
real numbers.

1.5. Theorem. Any complex number z satisfies the following equalities:
z + n = z , z ⋅ n = n and z ⋅ e = z. , where n denotes the additive identity, and e denotes the
multiplicative identity.
Proof. Indeed, if z = (a, b) is an arbitrary comblex number, then
z + n = (a, b) + (0, 0) = (a + 0, b + 0) = (a, b) = z ,
z ⋅ n = (a, b) ⋅ (0, 0) = (a ⋅ 0 − b ⋅ 0, a ⋅ 0 + b ⋅ 0) = (0, 0) = n , and
z ⋅ e = (a, b) ⋅ (1, 0) = (a ⋅ 1 − b ⋅ 0, a ⋅ 0 + 1 ⋅ b) = (a, b) = z. ■

1.6. Theorem. If z1 + z3 = z2 + z3 , then z1 = z2 .

Proof. Let z1 = (a1 , b1 ), z2 = (a2 , b2 ) and z3 = (a3 , b3 ) be complex numbers.
Then,
z1 + z3 = (a1 , b1 ) + (a3 , b3 ) = (a1 + a3 , b1 + b3 ) and
z2 + z3 = (a2 , b2 ) + (a3 , b3 ) = (a2 + a3 , b2 + b3 ) .
Since the given equality z1 + z3 = z2 + z3 and the Definition 1.1 we get the fol-
lowing
a1 + a3 = a2 + a3 and b1 + b3 = b2 + b3 .
Furthermore, the properties of real numbersimply that a1 = a2 and b1 = b2 , and since
Definition 1.1. ■

1.7. Theorem. For each complex number z there exists one and only one com-
plex number w, so that z + w = n .
Proof. Let z = (a, b) be an arbitrary complex number, and w be defined as
w = (− a, −b) . We get,
z + w = (a, b) + (− a, −b) = (a + (− a ), b + (−b)) = (0, 0) = n .
So, we proved the existence of a complex number w. The uniqueness is directly implied
by Theorem 1.6. ■

In our further consideration, the complex number w, so that z + w = n , will be

denoted by w = − z , and w is called to bean opposite complex number of z.
Let z and w be arbitrary complex numbers. The complex number z + (− w) is
called to be a substraction of the numbers z and w, and is denoted by z − w .

1.8. Theorem. For all complex numbers z1 and z2 , the equality

(− z1 ) ⋅ z2 = z1 ⋅ (− z2 ) = −( z1 z2 ) = (−e) ⋅ ( z1 z2 ) ,
holds true and there is no ambiguity in the notation −z1 z2 .
Proof. Let z1 = (a1 , b1 ) and z2 = (a2 , b2 ) be arbitrary complex numbers So,
(− z1 ) ⋅ z2 = (− a1 , −b1 ) ⋅ (a2 , b2 ) = (− a1a2 + b1b2 , − a1b2 − b1a2 ) =
= −(a1a2 − b1b2 , a1b2 + b1a2 ) = −( z1 z2 ).

But, the commutativelaw of multiplication holds true, therefore the above equality im-
plies that
−( z1 z2 ) = −( z2 z1 ) = (− z2 ) z1 = z1 (− z2 ).
Finally, since the Theorem 1.5 and the already proved equalities we get that
(−e)( z1 z2 ) = −(e( z1 z2 )) = −( z1 z2 ) = (− z1 ) z2 = z1 (− z2 ) . ■

1.9. Definition. The absolute value of the complex number z = (a, b) is defined
by
| z | = a 2 + b2 .
Thus, the absolute value of each complex number z is a non-negative real number.

1.10. Theorem. a) If z ≠ n , then | z | > 0 and | n | = 0 .

b) | z1 ⋅ z | =| z1 | ⋅ | z | , for all complex numbers z and z1 .
Proof. Let z = (a, b) and z1 = (a1 , b1 ) be any complex numbers
a) It is obvious that
| n | = 02 + 02 = 0 .
If z ≠ n , then a ≠ 0 or b ≠ 0 , i.e. a 2 > 0 or b 2 > 0 . Thus,
| z |2 = a 2 + b 2 > 0.
b) Since,
| z1 z |2 = | (a1a − b1b, a1b + b1a ) |2 = (a1a − b1b) 2 + (a1b + b1a ) 2
= a12 a 2 + b12b 2 + a12b 2 + b12 a 2 = (a12 + b12 )(a 2 + b 2 ) =| z1 |2 | z |2 ,
we get that | z1 ⋅ z | = | z1 | ⋅ | z | . ■

1.11. Remark. Theorem 1.10. b) and the principle of mathematical induction

directly imply the following:
| z1 z2 ...zn | = | z1 | ⋅ | z2 | ⋅  ⋅ | zn | . ■ (1)

1.12. Theorem. If zw = n , then z = n or w = n .

Proof. If zw = n , then Theorem 1.10 implies
| z | ⋅ | w | = | zw | = | n | = 0.
But, | z | and | w | are real numbers, and thus | z | = 0 and | w | = 0 , i.e. z = n or w = n . ■

1.13. Theorem. If z ≠ n and zw = zw1 , then w = w1 .

Proof. The given condition zw = zw1 implies that − zw = − zw1 . So,
n = zw − zw1 = z ( w − w1 ) .
Acording to Theorem 1.12, z ≠ n implies that w − w1 = n , i.e. w = w1 . ■

1.14. Theorem. For each complex number z ≠ n there exists one and only one
complex number w, denoted by ez , so that zw = e holds.
Proof. Firstly, we will prove the existence of the complex number w = ez . Let
z = (a, b) ≠ n be an arbitrary complex number . Let
(
w = 2 a 2 , 2−b 2 ,
a +b a +b )
So,
zw = (a, b) ⋅ ( a , −b
a 2 +b 2 a 2 +b 2 ) = (1, 0) = e.
The uniqueness is implied immideately by Theorem 1.13. ■

1.15. Theorem. If z ≠ n , then for each complex number w there exists one and
only one complex number u, so that zu = w holds.
Proof. By Theorem 1.14, for any complex number z ≠ n there exists one and
only one complex number ez , so that z ⋅ ez = e holds. Let u = e
z
⋅ w . Thus we get a unique
complex number u such that satisfies the following
zu = z ⋅ ez w = w . ■

2. ALGEBRAIC NOTATION OF A COMPLEX NUMBER

2.1. In the previous considerations we discussed the arithmetics of complex num-

bers, but the ususl symbol i was not presented, yet. Now, we will prove that the notation
(a,b) is equivalent to the usual notation for a complex number a + ib .
The proofs of the stataments in Theorem 2.2 are elementary therefore the ones
will not be done.

2.2. Theorem. For all real numbers a and b the following equalities are satis-
fied:
a) (a, 0) + (b, 0) = (a + b, 0),
b) (a, 0)(b, 0) = (ab, 0) ,
c) | (a, 0) | = | a | , where | a | is the absolute value of the real number a,

( b ,0 )( )
d) ( a,0) = a , 0 , for b ≠ 0 . ■
b

2.3. The statements given in Theorem 2.2 immideately imply that the mapping
f : R → C defined by f (a ) = (a, 0) is a bijection between R and A = {(a, 0) | a ∈ R} ⊆ C,
and such that inhire the operations. So, the set of real numbers R might be reviewed as a
subset of the set of complex numbers C.

10
According to this, the complex numbers e and n correspond to the notation 1 and
0, respectively. So, they will be used in our further consideration

2.4. Definition. The complex number i = (0,1) is called to be imaginary unit.

The imaginary unit satisfies the following equality
i 2 = (0,1) ⋅ (0,1) = (0 ⋅ 0 − 1 ⋅ 1, 0 ⋅ 1 + 0 ⋅ 1) = (−1, 0) = −1.

2.5. The obvious equation

(0,1) ⋅ (b, 0) = (0, b)
implies that
z = (a, b) = (a, 0) + (0, b) = (a, 0) + (0,1) ⋅ (b, 0) = a + ib
holds true for any complex number z = (a, b) .

Definition. The notation z = a + ib is called to be an algebraic notation of the

complex number z = (a, b) .

The addition and multiplication of complex numbers, by using the algebraic no-
tations of complex numbers, are written as following:
(a1 + ib1 ) + (a2 + ib2 ) = (a1 + a2 ) + i (b1 + b2 ) ,
(a1 + ib1 ) ⋅ (a2 + ib2 ) = (a1a2 − b1b2 ) + i (a1b2 + a2b1 ).

2.6. Definition. The components a and b of a complex number z = a + ib are

called to be real and imaginary part of z, respectively, and we use the following notations
a = Re z and b = Im z to denote them.

3. A CONJUGATE COMPLEX NUMBER

3.1. Definition. The complex number a − ib is called to be the complex conju-

gate of z = a + ib and is denoted as z .

3.2. Theorem. a) z = z , for each complex number z.

b) z1 + z2 = z1 + z2 , for all complex numbers z1 and z2 ,
c) z1 z2 = z1 ⋅ z2 , for all complex numbers z1 and z2 ,
d) z + z = 2 Re z , for each complex number z,
e) z − z = 2 Im z , for each complex number z, and
f) z ⋅ z = a 2 + b 2 = | z |2 ≥ 0 , for each complex number z.
Proof. The definition of a conjugate complex number, also addition and multili-
cation of complex numbers directly imply the validity of the above theorem. ■

11
3.3. Remark. The equality z − z = 2 Im z directly implies that the complex
number z is a real number if and only if z = z .

3.4. Remark. The validity of the equality (1) can be proved since the equality
z1 z2 = z1 ⋅ z2 holds and also by applying the principle of mathematical induction,
z1 z2 z3 ...zn = z1 ⋅ z2 ⋅ ... ⋅ zn , (1)
for each n ∈ N and any complex numbers z1 , z2 ,..., zn . If zk = z for each k = 1, 2,..., n ,
then (1) implies that z n = z n .

3.5. Remark Theorem 3.2. f) and а) holds, therefore it is true that

| z |2 = z ⋅ z = z ⋅ z = | z |2 , i.е. | z | = | z | .
Let z = x + iy . The equality | z |2 = z z = x 2 + y 2 directly implies the inequalities
− | z | ≤ Re z ≤ | z | and − | z | ≤ Im z ≤ | z | (2)

3.6. Example. Let

P ( z ) = a0 z n + a1 z n −1 + ... + an −1 z + an
be a real polynomial. If w is a root of P(z), then w is also a root of the same polynomial
P(z). Prove it!
Solution. Since w is a root of the polynomial P(z), P ( w) = 0 . Further,
n n −1
P ( w) = a0 w + a1 w + ... + an −1 w + an

= a0 wn + a1wn −1 + ... + an −1w + an = P ( w) = 0.

The latter means that w is also a root of the same polynomial P(z). ■

3.7. Example. Let’s consider the polynomial

P ( z ) = a0 z n + a1 z n −1 + ... + an −1 z + an , a0 ≠ 0 ,
written as the following
P ( z ) = a0 ( z − z1 )( z − z2 )...( z − zn ) ,
where zi , i = 1, 2,..., n are the roots of P(z).
The identity
a0 ( z − z1 )( z − z2 )...( z − zn ) = a0 z n + a1 z n −1 + ... + an −1 z + an
implies that
( )
a0 z n − ( z1 + z2 + ... + zn ) z n −1 + (−1) n z1 z2 ...zn = a0 z n + a1 z n −1 + ... + an −1 z + an .
By equating the coefficients of the corresponding degrees we get the following formu-
lae:

12
a
z1 + z2 + ... + zn = − a1
0
a
z1 z2 + ... + z1 zn + z2 z3 + ... + z2 zn + ... + zn −1 zn = − a2
0
...................................................................................................
a
z1 z2 ...zk + ... + z1...zk −1 zn + ... + zn −k +1 zn −k + 2 ...zn = (−1) k ak
0
.....................................................................................................
a
z1 z2 ...zn = (−1) n an
0
These formulae are known to be the Vieta’s formulae. ■

3.8. Remark. In Theorem 1.15 we proved that if z ≠ n , then for each complex
number w there exists a unique complex number u so that zu = w holds. The complex
number u is called to be a quotient of the complex numbers w and z and is denoted by
u = wz .
Hence, if z = a2 + ib2 and w = a1 + ib1 , then the quotient is expressed as the fol-
lowing
( a +ib )( a −ib ) a a +b b a b −a b
u = wz = wz = ( a 1 +ib1 )( a2 −ib2 ) = 1 22 12 2 + i 2 21 12 2 .
zz 2 2 2 2 a2 +b2 a2 +b2

For the quotient of the complex numbers z1 and z2 ( )=

z1
z2
z1
z2
holds.

z +z
3.9. Example. If | z1 | = | z2 | = 1 and z1 z2 ≠ −1 , then 1+1 z z2 is a real number.
1 2
Prove it!
Solution. By multiplying both the numerator and the denominator by the conju-
gate of 1 + z1 z2 we get
z1 + z2 z +z 1+ z1z2 ( z1 + z2 )(1+ z1 z2 )
1+ z1z2
= 1+1 z z2 ⋅ =
1 2 1+ z1z2 |1+ z1z2 |2
z1 + z2 + z1 z1 z2 + z1z2 z2 z1 + z1 z2 + z2
= = + .
|1+ z1z2 |2 |1+ z1z2 |2 |1+ z1z2 |2
z +z
But both, z1 + z1 and z2 + z2 are real numbers, therefore 1+1 z z2 is also a real num­-
1 2
ber. ■

3.10. Example. Prove that if the absolute value of a complex number is 1, than
such a complex number can be expressed as cc+−ii , where c is a real number.
Solution. If | z | = 1 and c = i zz +−11 , then z = cc+−ii and cc+−ii = 1 , i.e.
c + i ⋅ c −i = 1.
c −i c + i
By reducing the last equality, we get that c − c = 0 . Hence, c ∈ R and z = cc+−ii ,
which actually was supposed to be proven. ■

13
3.11. Theorem. All complex numbers z and w ≠ 0 satisfy the equality
z | z|
w
= | w| .
Proof. By applying the above stated we get the following,

( )
2 | z| 2
z = wz ⋅ wz = wz ⋅ z = z z = 2 ,
w w ww | w |
| z|
that is wz = | w| . ■

3.12. Example. Determine all the complex numbers z such that satisfy the fol-
lowing equalities:
z −12 = 5 z −4 = 1
z −8i 3 and z −8 .
Solution. Let z = x + iy . Since the given condition we get that
2 | z −12 |2 ( x −12)2 + y 2 2 | z −4| ( x − 4) + y 2 2 2
z −12 = = = 25 and zz −−84 = = = 1.
z −8i | z −8i | 2
x + ( y −8) 9
2 2
| z −8|2 ( x −8)2 + y 2
By reducing we get the following system of equations
2 x 2 + 2 y 2 + 27 x − 50 y + 38 = 0

 x = 6
whose solutions are x = 6, y = 17 and x = 6, y = 8 . Hence, the required complex num-
bers are z = 6 + 17i and z = 6 + 8i . ■

3.13. Example. If a, b and c are complex numbers such that

|a| = |b| = |c| = r , r > 0
then
| ab + bc + ca | = r | a + b + c | .
Prove it!

Solution. Since r 2 = | a |2 = aa , we get that 1a = a2 . Analogously, b1 = b2 and

r r
1 = c . Therefore,
c r2
(
ab + bc + ca = abc 1a + b1 + 1c = abc a2 + b2 + c2 = abc)2 (
⋅a +b + c .
r r r ) r
The latter implies that,
| abc |
| ab + bc + ca | = a+b+c =r|a+b+c|
r2
holds true, which actually was supposed to be proven. ■

3.14. Theorem. All complex numbers z1 and z2 satisfy the following

| z1 + z2 | ≤ | z1 | + | z2 | and | z1 − z2 | ≥ | z1 | − | z2 | .
Proof. Since Theorem 3.2, the following holds true for all complex numbers z1
and z2 :
| z1 + z2 |2 = ( z1 + z2 )( z1 + z2 ) =| z1 |2 + | z2 |2 +2 Re z1 z2 (3)

14
| z1 − z2 |2 = ( z1 − z2 )( z1 − z2 ) =| z1 |2 + | z2 |2 −2 Re z1 z2 . (4)
tothe equalities (3) and (4), if we apply the first inequality in (2), imply the following
inequalities:
| z1 + z2 | ≤ | z1 | + | z2 | (5)
| z1 − z2 | ≥ | z1 | − | z2 | . (6)
The inequality (5) is known to be the triangle inequality. ■

3.15. Corollary. For all complex numbers z1 , z2 ,..., zn the following inequality
holds true:
n n
| ∑ zi | ≤ ∑ | zi |.
i =1 i =1
Proof. The proof is directly implied by the inequality (5) and the principle of
mathematical induction. ■

3.16. Example. For all complex numbers z1 and z2

(
| z1 + z2 |2 + | z1 − z2 |2 = 2 | z1 |2 + | z2 |2 . ) (7)
holds true. Prove it!
Solution. If we summarize the equalities (3) and (4) as given in Theorem 3.14 (the
ones hold for all complex numbers z1 and z2 ) we get the required identity. This identity is
known to be the parallelogram identity. ■

3.17. Example. Prove that all complex numbers z1 and z2 satisfy

2 2
| 1 − z1 z2 |2 − | z1 − z2 |2 = (1 + | z1 z2 |) − (| z1 | + | z2 |) .

Solution. The identity z z = | z |2 , directly implies the following

| 1 − z1 z2 |2 − | z1 − z2 |2 = (1 − z1 z2 )(1 − z1 z2 ) − ( z1 − z2 )( z1 − z2 )
= (1 − z1 z2 )(1 − z2 z1 ) − ( z1 − z2 )( z1 − z2 )
= 1 − z1 z2 − z2 z1 + z1 z1 z2 z2 − z1 z1 − z2 z2 + z1 z2 − z2 z1
= 1 + | z1 z2 |2 − | z1 |2 − | z2 |2 =
2 2
= (1 + | z1 z2 |) − (| z1 | + | z2 |) ,
which was suppesed to be proven. ■

3.18. Theorem. Let zi , wi , i = 1, 2,..., n be complex numbers. Thus,

n n n
| ∑ zi wi |2 ≤ ∑ | zi |2 ∑ | wi |2 . (8)
i =1 i =1 i =1

15
Proof. Let
n n n
C = ∑ wi zi , A = ∑ | zi |2 and B = ∑ | wi |2 .
i =1 i =1 i =1
If A = 0 , then | zi | = 0 , for all i = 1, 2,..., n . that is, zi = 0 , for all i = 1, 2,..., n . So,
also C = 0 . Thus, the inequlity (8) is satisfied.
If A ≠ 0 , then the inequality (8) is implied by the obvious equality
n 2 n
∑ C zi − Awi = ∑ (C zi − Awi )(Czi − Awi ) = A( AB − | C |2 )
i =1 i =1
and the inequalities
n 2
∑ C zi − Awi ≥ 0, A> 0.
i =1
The inequality (8) is known to be the Cauchy-Schwarz-Bunyakovsky inequality
for complex numbers. ■

4. GEOMETRIC PRESENTATION
OF A COMPLEX NUMBER

The Eucledian plane with Cartesian coordinates is denoted by R 2 . Each com-

plex number z = x + iy is an ordered pair of real numbers (x,y). Since it exists a one-
to-one correspondence between the set of the ordered
pairs of real numbers (x,y) and R 2 , we get that to each
point A∈ R 2 may be adjoined a complex number
z = x + iy , and conversely (figure 1). The complex
number z such that it corresponds to the point A is
called to be the affix of the point A. This correspond-
ence between the complex numbers and the points of
the Euclidean plane is bijection. Thereby, the real part
of the complex numbers mapps onto the points of the
Figure 1 x-axis (abscissa), while the imaginary part of the com-
plex numbers mapps onto the points of the y-axis (ordi-
nate). So, the real numbers map onto the points onto the abscissa, and the pure imaginary
numbers map onto the points of the ordinate. Thus, the abscissa is called to be the real
axis, and the ordinate is called to be the imaginary axis. Thus, the Euclidean plane R 2 ,
is naturaly called to be the complex plane, and the complex numbers to be points in this
plane.
Clearly, the points z and −z are symmetric with respect to the origin, and z and
z are symmetric with respect to the real axis. Namely, if z = x + iy then

16
− z = (− x) + i (− y ) and z = x + (− y )i .
Obviously, the complex number z corre-
sponds to the vector with tail in the origin
point O and head in the point z. Clearly,
this correspondence between the complex
numbers and the vectors in the complex
plane with tails in O is bijective. Thus,
the vector which determines a complex
number z, will be denoted by the same
letter z.
By using the vector interpretation
of the complex numbers we can demon- Figure 2
strate the addition and the subtraction of
complex numbers. Since 1.2 we get that
the number z1 + z2 corresponds to the
vector obtained by adding the vectors z1
and z2 (figure 2). The vector z1 − z2 is
constructed as the sum of the vectors z1
and −z2 (figure 3).
The already stated and also using
figure 3 implies that the distance between
the points z1 and z2 is equal to the length
of the vector z1 − z2 , i.e. it is equal to
| z1 − z2 | . Clearly, the absolute value | z |
is equal to the length of the correspond-
ing vector of the point z If we consider Figure 3
the triangles whose vertices are
O, z1 , z1 + z2 and O, z1 , z1 − z2 ,
then the geometric sence of the inequalities (5) and (6) from paragraph 1 is obvious.

4.2. Dividing a line segment in a given ratio. Let A and B be given points
with affixes z1 and, z2 respectively, and C be a point on the line segment AB, such
  
that C divides AB in a given ratio λ : µ ≠ −1 , i.e. µ AC = λCB . Since AC = z − z1 and
 λz +µz
CB = z2 − z , we get that µ( z − z1 ) = λ ( z2 − z ) . Hence, the affix of C is z = λ2+µ 1 . If
z2 + z1
λ : µ = 1 then z = 2
is an affix of C, the midpoint of the line segment AB.

Example А. Let a = 1 + i and b = 3 + 5i be the affixes of the endpoints of the line

segment AB. Then
(1+i ) + (3+5i )
c = 1⋅a1++11⋅b = 2
= 1 + 3i .
is the affix of C, the midpoint of the line segment AB. ■

17
Example B. Let ABCD be a quadrilateral and M, N, P, Q, K, L be the midpoints
of the line segments AB, BC, CD, DA, BD, CA, respectively. Show that the line segments
MN, NQ, KL concur at such a point T that bisects each of them.
Solution. Let the affixes of the points A, B, C, D, M¸ N, P, Q, K, L are denoted
by lowcase letters a, b, c, d, m, n, p, q, k, l, respectively, then
m = a +2 b , n = b +2 c , p = c +2d , q = a +2d , k = b +2d , l = a 2+c .
The affixes of the midpoints of MP, NQ, KL are
t = a +b +4c + d , t1 = b +c +4a + d , t2 = b + d 4+ a +c ,
respectively and since t = t1 = t2 , we get that MP, NQ, KL are concurrent at the point
which bisects each of these segments. ■

4.3. Example. а) The set of points z, such that satisfy the equation | z − z0 | = R ,
is a circle centered at z0 and radius R. Namely, | z − z0 | is the distance between the
points A and B with affixes z and z0 , respectively.
b) The equation
| z − z1 | − | z − z2 | = 2a ,
where
a < 12 | z1 − z2 | ,
is a hyperbola, whose foci are at points whose affixes are z1 and z2 and a real semi-axis
whose length is a, (why?).
c) The set of points z, such that satisfy the equation
| z − z1 | = | z − z2 | ,
is a set of points equidistanced form z1 and z2 . Thus,
| z − z1 | = | z − z2 |
is an equation of the bisector of the segment, whose extremities have affixes z1 and z2 .
d) The set of points z, such that satisfy the equation
| z − z1 | + | z − z2 | = 2a ,
1
where a > 2 | z1 − z2 | , is an ellipse with focal points z1 , z2 , and major semi-axis a, there-
by | z − z1 | + | z − z2 | is the sum of the distances between the point M with affix z and the
points A and B with affixes z1 and z2 , respectively. ■

4.4. Example. Determine the set of points, which corresponds to such complex
numbers z, that satisfy the following condition
| 2z | ≥ |1 + z2 | .
Solution. Let z = x + iy . Thus,
1 + z 2 = 1 + ( x + iy ) 2 = (1 + x 2 − y 2 ) 2 + 4 x 2 y 2 .

18
Hence,
4( x 2 + y 2 ) ≥ 1 + x 4 + y 4 + 2 x 2 − 2 y 2 − 2 x 2 y 2 + 4 x 2 y 2 ,
1 + x4 + y 4 − 2 x2 − 2 y 2 + 2 x2 y 2 − 4 y 2 ≤ 0 ,
( x 2 + y 2 − 1) 2 − 4 y 2 ≤ 0 ,
( x 2 + y 2 − 1 − 2 y )( x 2 + y 2 − 1 + 2 y ) ≤ 0
or
( x2 + ( y − 1)2 − 2 ) ( x2 + ( y + 1)2 − 2 ) ≤ 0 .
The last inequality is satisfied if and only if
x 2 + ( y − 1) 2 − 2 ≥ 0 and
x 2 + ( y + 1) 2 − 2 ≤ 0
or
x 2 + ( y − 1) 2 − 2 ≤ 0 and
x 2 + ( y + 1) 2 − 2 ≥ 0 .
The required set of points is shown in figure 4. ■ Figure 4

5. EXTENDED COMPLEX PLANE. REIMANN INTERPRETA-

TION OF COMPLEX NUMBERS

5.1. Stereographic projection. In Euclidean space R 3 with Cartesian coordi-

( )
nates ξ, η, ζ, consider the sphere S which is centered at 0, 0, 12 and its radius is 12 :
2 2 2
ξ + η + ζ − ζ = 0 . (1)
The plane ζ = 0 coincides with the complex plane C, the real axis Im z = 0 coincides
with the axis η = 0 , ζ = 0 , and the imaginary axis Re z = 0 coincides with the axis
ζ = 0 , ξ = 0.
We draw a line through P(0,0,1),
such that it meets the sphere S at a point
M(ξ,η,ζ), which differs from P. The in-
tersection of the line PM and the complex
plane is denoted by z = x + iy . The point
M(ξ,η,ζ) is called to be the stereographic
projection of the complex number (point)
z onto the sphere S with pole P (figure 5).
The stereographic projection de-
fines a bijection between the points in the
complex plane C and the points on the
sphere S, except the pole P. Therefore,
each point of the sphere S, except the pole Figure 5

19
P, may be considered as a point in the complex plane. Such the interpretation of complex
numbers is called to be Reimann interpretation of complex numbers, and the sphere S is
called to be Reimann sphere.
The points P(0,0,1), M(ξ,η,ζ) and z are collinear, so
ξ η ζ −1
x
= y = −1
or
ξ η ξ+iη
x = 1−ζ , y = 1−ζ , z = 1−ζ . (2)
ξ2 + η2 ζ
Hence, | z |2 = . If we substitute in (1) we get | z |2 = 1−ζ , i.e.
(1−ζ )2
| z |2
ζ= . (3)
1+| z |2
If we substitute in (2) and also have on mind the fact that x = z +2 z , y = z2−iz we get
ξ = z + z 2 , η = z − z 2 . (4)
2(1+| z | ) 2i (1+| z | )
The formulae (3) and (4) are known to be formulae of the stereographic projection.

5.2. Extended complex plane. In 5.1 we defined a bijection between the com-
plex plane and the Reimann sphere S without the pole P. If we add the “ideal complex
number” z = ∞ to the set of complex numbers C, and complete the complex plane by
adjoining the unique infinity point, denoted by ∞, then it exists a bijection between the
Reimann sphere S and the set C ∪ {∞}, whereby the pole P corresponds to the infinity
point ∞.
The complex plane, together with the infinity point, is called to be an extended
complex plane and is denoted by C∞ = C ∪ {∞} . Let us state that the infinity point is not
involved in the algebraic operations with complex numbers.

5.3. Distance in the extended complex plane. In the complex plane C the dis-
tance between the points z and z ' is defined by | z − z ' | . In C∞ we define the distance
between the points z and z ' , d ( z , z ') as a distance between appropriate stereographic
projections of the points z and z ' . Namely, if M(ξ,η,ζ) and M '(ξ ', η ', ζ ') are the stere-
ographic projections of the points z ≠ ∞ and z ' ≠ ∞ , respectively, then
| z − z '|
d ( z , z ') = (ξ − ξ ') 2 + (η − η ') 2 + (ζ − ζ ') 2 =
1+| z|2 1+| z '|2
and if z ' = ∞ , then
d ( z , ∞) = 1
.
1+| z |2

5.4. Theorem. Under the stereographic projection each circle in the complex
plane maps to a circle on the Reimann sphere,such that it does not pass through the pole,
and conversely.

20
Proof. Let
x 2 + y 2 + Ax + By + C = 0, A, B, C ∈ R (5)
be any circle in the complex plane xOy. Since (2) and (5) we get that
ζ ξ η
+ A 1−ζ + B 1−ζ + C = 0 , i.e.
1−ζ
Aξ + Bη + (1 − C )ζ + C = 0 . (6)
The equation (6) is the equation of a plane which does not pass through the pole P(0,0,1).
So, the coordinates ξ, η, ζ satisfy the equalities (1) and (6). Thus, the points (ξ,η,ζ) lie on
the sphere (1) and the plane (6), i.e. they create a circle on the Reimann sphere such that
it does not pass through the pole.
Conversely, each circle on the Reimann sphere (1) such that it does not pass
through the pole by the stereographic projection is mapped to a circle in the complex
plane,thereby, by using arbitrary numbers A, B, C the equation of the intersecting plane
may be expressed as (6). ■

5.5. Theorem. Each line in the complex plane under the stereographic projection
maps to a circle through the pole and vice versa.
Proof. Let
Ax + By + C = 0, A, B, C ∈ R (7)
be any line in the complex plane xOy. Since the identities (2) and (7), we get that
ξ η
A 1−ζ + B 1−ζ + C = 0
i.e.
Aξ + Bη + −C ζ + C = 0 . (8)
The equation (8) is the equation of a plane such that it passes through the pole P(0,0,1).
Therefore, the coordinates ξ, η, ζ satisfy the equalities (1) and (8). So, the points (ξ,η,ζ)
lie on the sphere (1) and in the plane (8), i.e. they create a circle on the Reimann sphere
which passes through the pole.
Conversely, each circle on the Reimann sphere (1) which passes through the
pole by the stereographic projection is mapped to a circle in the complex plane, thereby
the arbitrariness of numbers A, B, C allows the equation of the intersecting plane to be
represented as (8). ■

5.6. Let l and q be two distinct curves on the Reimann sphere (1) such that they
meet at a point M. Through such a point it can be drawn tangents to the curves l and q.
Let α be the angle formed by the tangents. Let l ', q ' and M ' be the images of l, q and
M, respectively, under the stereograph projectionon the complex plane. It is easy to prove
that the angle create by the tangents to the curves l ' and q ' through M ' is congruent
to α. The proof of this statement will not be elaborated, thereby, it is beyond our main
considerations.

21
6. TRIGONOMETRIC ENTRY
OF A COMPLEX NUMBER

6.1. Argument of a complex number. The angle ϕ created by the positive part
of the real axis and the position vector of the point z, is called to be the argument of
the complex number z, and is denoted as
ϕ = Arg z , (figure 6). The argument is either
positive or negative, depending on the orien-
tation of the angle ϕ. The argument is called
to be positive if it is oriented from the posi-
tive direction of the real axis to the positive
direction of the imaginary axis or negative if
the one is oriented from the positive direction
Figure 6 of the real axis to the negative direction of the
imaginary axis.
For z = 0 the argument is not specified. So, in our further discussion about argu-
ments, we assume that z ≠ 0 .
The position of the point z in the complex plane is uniquely determined by its
Cartesian coordinates x, y and by its polar coordinates r = | z | and ϕ = Arg z . The rela-
tion between these two types of coordinates is given by the following formulae:
x = r cos ϕ , y = r sin ϕ . (1)
For a given point z, its absolute value is uniquely determined, while the argument
is determined by accuracy of up to a summand 2kπ, k = 0, ±1, ±2,... . The value of the ar-
gument, such that it satisfies the condition 0 < Arg z ≤ 2π is called to be the main value
of the argument and is denoted by arg z. Most commonly, in our further considerations
we will use the main value of the argument.

6.2. Trigonometric entry of a complex number. Using formulae (1), which

refer to the Cartesian and polar coordinates of z, we get the so called trigonometric rep-
resentation of a complex number
z = | z | ( cos(arg z ) + i sin(arg z ) ) . (2)
By the notation (2) for the product of two complex numbers
z1 = | z1 | (cos ϕ1 + i sin ϕ1 ) and z2 = | z2 | (cos ϕ2 + i sin ϕ2 )
we get
z1 z2 = | z1 | ⋅ | z2 | ( cos(ϕ1 + ϕ2 ) + i sin(ϕ1 + ϕ2 ) ) . (3)
Further, by 1.10 and the definition for the argument of a complex numbers it is true that
z1 z2 = | z1 z2 | ( cos ( arg( z1 z2 ) ) + i sin ( arg( z1 z2 ) ) ) .
The above statement implies that
arg( z1 z2 ) = arg z1 + arg z2 + 2k π , k = 0, ±1, ±2,... (4)

22
Analogously, by the equality z1 = z2 z3 , for z2 ≠ 0 , using 3.11 we get
z
arg z1 = arg z1 − arg z2 + 2k π , k = 0, ±1, ±2,...
2

6.3. De Moivre’s formula. Using mathematical induction, the formulaes (3) and
(4) could be easily generalized for a finite number of multiples z1 , z2 ,...., zn . Namely,
arg( z1 z2 ...zn ) = arg z1 + arg z2 + ... + arg zn + 2k π , k = 0, ±1, ±2,... (5)
Particularly, for z1 = z2 = ... = zn we get
n
| z n | = | z |n and arg z = n arg z + 2k π , k = 0, ±1, ±2,...
i.e.
z n = | z |n ( cos( n arg z ) + i sin( n arg z ) ) . (6)
The formula (6) is known to be the De Moivre’s formula.

6.4. Example. Compute the difference

( −1 + i 3 )9 − (1 + i 3 )9 .
Solution. Since,
− 1 + i 3 = 2 and arg ( −1 + i 3 ) = 23π + 2k π , k = 0, ±1, ±2,...
we get
(
−1 + i 3 = 2 cos 23π + i sin 23π . )
Using De Moivre’s formula, we obtain:
( −1 + i 3 )9 = 29 ( cos 23π ⋅ 9 + i sin 23π ⋅ 9 ) = 29 .
Analogously,
(1 + i 3 )9 = 29 ( cos 93π + i sin 93π ) = −29 .
Hence,
( −1 + i 3 )9 − (1 + i 3 )9 = 29 − ( −29 ) = 210 . ■
6.5. Example. а) Find the exact value of the expression:
(1 − i 3 )3 (1 + i)10 .
b) Let
( ) ( )
n n
f (n) = 1+i + 1−i .
2 2
Determine the sum f (1990) + f (1994) .
Solution. а) We have that:
( ) (
1 − i 3 = 2 cos π3 − i sin π3 and 1 + i = 2 cos π4 + i sin π4 . )
Therefore,
(1 − i 3 )3 (1 + i)10 = 23 (cos π − i sin π) ⋅ 25 ( cos 52π + i sin 52π )
= 28 (−1 − i ⋅ 0)(0 + i ) = −256i.

23
b) Since,
1+i = cos π4 + i sin π4 and 1−i = cos π4 − i sin π4
2 2
we express the function f in terms of cos as follows
f (n) = 2 cos n4π .
Hence,
f (1990) + f (1994) = 2 cos 1990 π + 2 cos 1994 π = 2 cos 995π + 2 cos 997 π = 0 . ■
4 4 2 2

6.6. Example. If z + 1z = 1, then find the exact value of

z158 + z152 + 122
2 .
z
Solution. The given condition z + 1z = 1 implies z 2 − z + 1 = 0 , i.e.

( )
z = 12 ± i 23 = cos ± π3 + i sin ± π3 . ( )
Thus,

z 3 3 3 3 3(
2 = cos 158π ± i sin 158π + cos 152 π ± i sin 158π − 2 − cos 122 π ± i sin 122 π
z158 + z152 + 122 3 )
(
= cos 23π ± i sin 23π + cos 23π ± i sin 23π − 2 − cos 23π ± i sin 23π )
= 4 cos 23π = −2 .■

7. ROOTS OF A COMPLEX NUMBER

7.1. Definition. Let z ≠ 0 be given complex number and n be a positive integer.

The n-th root of z is defined as such a complex number w that
wn = z . (1)
n
We denote w = z .

Since applying the De Moivre formula

z n = | z |n ( cos( n arg z ) + i sin( n arg z ) )
and the trigonometric notations
z = | z | ( cos(arg z ) + i sin(arg z ) ) and w = | w | ( cos(arg w) + i sin(arg w) )
we get
| w |n ( cos n(arg w) + i sin n(arg w) ) = | z | ( cos(arg z ) + i sin(arg z ) )
i.e.
| w |n = | z | and n(arg w) = arg z + 2k π, k = 0, ±1, ±2,... (2)
Therefore,
arg z +2 k π
| w | = n | z | , arg w = n
, k = 0, ±1, ±2,... ,
i.e.
(
w = n z = n | z | cos
arg z + 2 k π
n
+ i sin
arg z + 2 k π
n ), (3)

24
By letting k = 0,1, 2,..., n − 1 in (3), we obtain n distinct complex numbers w0 , w1 ,..., wn −1
as w for k = n , and also thereby the periodicity of the trigonometric functions we obtain
w0 , etc. Thus, the n-th root of the complex number z has exactly n distinct values, ob-
tained by the formula (3) for k = 0,1, 2,..., n − 1 .

3
7.2. Example. Find 27i5 .
Solution. Since,
i 5 = i 4⋅ i = i = cos π2 + i sin π2
we get
3
(
27i5 = 3 27i = 3 27 cos π2 + i sin π2 = 3 cos ) ( 2 k π+ π2
3
+ i sin
2 k π+ π2
3 )
(
= 3 cos
( 4 k +1) π
6
+ i sin
( 4 k +1) π
6 ),
for k = 0,1, 2. ■

7.3. Example. Prove that all complex numbers a and b satisfy

2 (| a | + | b |) = a + b − 2 ab + a + b + 2 ab ,
where ab denotes one of the two roots of ab.
Solution. Since,
2 (| a | + | b | ) = 2 ( a
2
+ b
2
)
we obtain
2 2
2 (| a | + | b | ) = a+ b + a− b =

= ( a + b) +
2
( a − b)
2

= a + b + 2 ab + a + b − 2 ab .
So, the required equality is proved. ■

7.4. Example. Prove that all complex numbers a and b satisfy

| a + b | + | a − b | = a + a 2 − b2 + a − a 2 − b2 ,

where a 2 − b 2 denotes one of the two roots of a 2 − b 2 .

Solution. By example 3.16, we get

( a+ )
2
a 2 − b2 + a − a 2 − b2 =
2 2
= a + a 2 − b2 + a − a 2 − b2 + 2 a + a 2 − b2 ⋅ a − a 2 − b2
2 2
= a + a 2 − b2 + a − a 2 − b2 + 2 a 2 − (a 2 − b 2 )
 
( )
2
= 2  a 2 + a 2 − b2  + 2 b2 = 2 a 2 + b2 + 2 a 2 − b2

25

= | a − b |2 + | a + b |2 + 2 | a − b | ⋅ | a + b | = (| a − b | + | a + b |) 2
= a + a 2 − b2 + a − a 2 − b2 + 2 a + a 2 − b2 ⋅ a − a 2 − b2
2 2
= a + a 2 − b2 + a − a 2 − b2 + 2 a 2 − (a 2 − b 2 )
 
( )
2
= 2  a 2 + a 2 − b2  + 2 b2 = 2 a 2 + b2 + 2 a 2 − b2

= | a − b |2 + | a + b |2 + 2 | a − b | ⋅ | a + b | = (| a − b | + | a + b |) 2
i.e.
| a + b | + | a − b | = a + a 2 − b2 + a − a 2 − b2 . ■

7.5. Example. Solve the equation

( x + i ) n + ( x − i ) n = 0 , n ∈ N, n > 1.
Solution. Since, x ≠ i , the given equation is equivalent to

( xx+−ii )
n
= −1 .
Therefore,
x +i
x −i
= n −1 = n cos π + i sin π = cos π+n2k π + i sin π+n2k π , k = 0,1,..., n − 1 .
By using the last equation we get that
x +i − 1 = cos π+n2k π + i sin π+n2k π − 1,
x −i
2i
x −i (
= 2i sin π+22nk π cos π+22nk π − 1i sin π+n2k π , )
x−i = 1 ,
(
sin π +22nk π cos π +22nk π +i sin π +n2 k π )
cos π +22nk π −i sin π +n2 k π
x−i = ,
sin π +22nk π

x − i = ctg π+22nk π − i .
that is,
x = ctg π+22nk π , k = 0,1,..., n − 1 . ■

7.6. The n-th roots of the unity. In 7.1 we discussed roots of complex numbers.
If z = 1, then arg z = 0 and by using (3) the n different roots of the unity are expessed
as
uk = cos 2knπ + i sin 2knπ , k = 0,1, 2,..., n − 1. (4)
If
u = u1 = cos 2nπ + i sin 2nπ ,
then, since the DeMoivre’s formula
uk = u k , k = 0,1,..., n − 1.

Let us consider that in geometric terms, for n ≥ 3, the points in the of complex
plane with affixes the n-th roots of unity, form a regular n-gon inscribed in the unit circle
and one of the polygon’s vertices coincides to the point with affix z = 1.

26
n −1
p
7.8. Example. Let S p = ∑ uk be the sum of the p-th exponents of the n-th
k =0
roots of the unity and n ∈ N . Prove that
n, for n | p
Sp = 
0, for n /| p.
Solution. Since,
uk = u k , k = 0,1,..., n − 1
and
u = cos 2nπ + i sin 2nπ ,
we get
S p = 1 + u p + u 2 p + ... + u ( n −1) p . (5)
p
If n | p and n = m, then
u p = u mn = (u n ) m = 1m = 1
and moreover, thereby (5) we deduce that S p = n.
Let n /| p . Thus
u np = (u n ) p = 1 p = 1 holds.
Since n /| p , it follows that u p − 1 ≠ 0. Therefore,
np
S p = 1 + u p + u 2 p + ... + u ( n −1) p = u p −1 = 0. ■
u −1

7.9. Example. Prove the following identities

2( n −1) π
а) cos 2nπ + cos 4nπ + ... + cos n
= −1 , for n = 2, 3,... and
2( n −1) π
b) sin 2nπ + sin 4nπ + ... + sin n
= 0 for n = 2, 3,...

Solution. The equation z n − 1 = 0 has n roots. The before stated roots are the
n-th roots of the unity
uk = u k , k = 0,1,..., n − 1 and u = cos 2nπ + i sin 2nπ .
Example 7.8 implies that their sum is equal to zero. Hence,
n −1
∑ uk = 0 ,
k =0
i.e.
cos 2nπ + cos 4nπ + ... + cos
2( n −1) π
(
+ i sin 2nπ + sin 4nπ + ... + sin n
n
2( n −1) π
= −1 . )
The latter is equivalent to a) and b). So, the required identities are proven. ■

7.10. Definition. The complex number u is called to be primitive n-th root of the
unity, if u n = 1 and there is no any lower exponent of u, which is equal to 1.

27
7.11. Example. Let
uk = u k , k = 0,1,..., n − 1
be the n-th roots of the unity. Prove that uk is the primitive n-th root of unity if and only
if n and k are co-prime numbers.
Solution. Let n and k be co-prime numbers and let there exists r < n such that
satisfies ukr = 1. According to the De Moivre’s formula we get
1 = ukr = cos 2kr
n
π + i sin 2 kr π .
n
According to the last equality, we get
cos 2kr
n
π = 1, sin 2 kr π = 0.
n
Hence, krn
∈ Z and since, n and k are co-prime numbers, we get that n | r , which is con-
tradiction, thereby r < n . Thus, uk is the primitive n-the root of the unity.
Conversely, let uk be the primitive n-th root of the unity. Let assump the greatest
common divisor of n and k is d, d > 1. Let k = k1d , n = n1d . Then
n nk k dn kn
uk1 = (u1k ) n1 = u1 1 = u1 1 1 = u1 1 = (u1n ) k1 = 1k1 = 1.
The latter contradicts to the fact that uk is a primitive n-the root of unity, thereby
n1 < n . ■

8. EXPONENTIAL ENTRY
OF A COMPLEX NUMBER

8.1. In our previous discussion we have presented the algebraic and trigonomet-
ric representations of complex numbers. In this section we will focus on the so called
exponential entry of complex numbers.

Theorem. Let the function f : R → C be defined by

f (α) = cos α + i sin α , for each α ∈ R .
Then,
a) f (0) = 1 and f (α) ≠ 0 , for each α ∈ R .
b) f (α + β) = f (α) f (β) , for all α, β ∈ R .
c) f (−α) = f (1α ) , for each α ∈ R .

Proof. а) Clearly, f (0) = cos 0 + i sin 0 = 1 . Let there exist α ∈ R , such that
f (α) = 0 . Thus, it exists α ∈ R , so that cos α + i sin α = 0 , i.e. cos α = sin α = 0 , which
contradicts to the basic trigonometric identity
cos 2 α + sin 2 α = 1 .

28
b) For all α, β ∈ R
f (α + β) = cos(α + β) + i sin(α + β)
= cos α cos β − sin α sin β + i (sin α cos β + cos α sin β)
= (cos α + i sin α)(cos β + i sin β) = f (α) f (β).
c) For each α ∈ R
f (−α) = cos( −α) + i sin(−α) = cos α − i sin α
(cos α −i sin α )(cos α +i sin α )
= cos α +i sin α
= cos α +1i sin α = f (1α ) . ■

8.2. In the previous theorem we proved that the function f satisfies the ordi-
nary properties of the exponential function, so it is natural to introduce the notation
f (α) = eiα , for each α ∈ R . Therefore, the properties b) and c) of the theorem could be
stated as the following:
eiα eiβ = ei (α +β) (1)
e −iα = 1iα . (2)
e
By using the identities (1) and (2) and also the principle of mathematical induction, we
obtain that
(eiα ) n = einα , for n = 0, ±1, ±2,... . (3)

8.3. Euler’s formulae. The above stated, implies that each complex number z,
such that | z | = 1 and ϕ = arg z may be denoted as
z = cos ϕ + i sin ϕ = eiϕ . (4)
πi 3 πi
Thus, e 2πi = 1, e πi = −1, e2 = i, e2 = −i . If we change ϕ by − ϕ we obtain that
cos ϕ − i sin ϕ = e−iϕ . (5)
By using the dentities (4) and (5) we obtain the well known Euler’s formulae:
iϕ − iϕ iϕ − iϕ
cos ϕ = e +2e , sin ϕ = e −2e . (6)
These formulae allow the trigonometric functions cos and sin to be expessed in terms of
the exponential function.
At the present moment we shall state that in Theorem 1 we did not give the proof
of the formula (4)., we only gave its “acceptable” explanation.

8.4. The formula (4) and the trigonometric form of complex numbers imply that
each complex number z ≠ 0 may be written as
z = reiϕ , (7)
where r = | z | and ϕ = arg z . The notation (7) of a complex number z ≠ 0 is called to be
exponential representation of z.

29
Using formulae (1) and (2), we obtain the exponential forms of the formulae for
multiplication and division of complex numbers, i.e.
z1 z2 = r1eiϕ1 r2 eiϕ2 = r1r2 ei (ϕ1 +ϕ2 ) , (8)
iϕ1
z1 r1e r
z2
= i ϕ2
= r1 ei (ϕ1 −ϕ2 ) . (9)
r2e 2

Let z = reiϕ . Since (4) and (5) the expression for z_ is the following z = re −iϕ .
Hence, if ϕ = arg z , then −ϕ = arg z .

8.5. Example. Find the sums:

a) A = cos x + cos( x + α) + cos( x + 2α) + ... + cos( x + nα) , and
b) B = sin x + sin( x + α) + sin( x + 2α) + ... + sin( x + nα) .
Solution. Let S = A + iB . Thus
S = eix + ei ( x +α ) + ei ( x + 2α ) + ... + ei ( x + nα )
ix ( i ( n +1) α )
= eix (1 + eiα + ei 2α + ... + einα ) = e e iα −1 .
e −1
Since, A = Re S and B = Im S . by applying the last formula we express A and B. If the
iα
numerator and the denominator be divided by e 2 , then thereby the Euler’s formulae we
get that the denominator is 2i sin α2 , and the numerator is

( ( ) ) ( ) ( ( ( )) (
cos x + n + 12 α − cos x − α2 + i sin x + n + 12 α − sin x − α2 )) =
= 2 sin
( n +1)α
2 ( − sin ( x + n2α ) + i cos ( x + n2α )).
Hence,
sin
( n +1) α
cos ( x+ n2α ) and B =
sin
( n +1) α
2
sin ( x+ n2α ) . ■
A= 2
sin α2 sin α2

8.6. Remark. If x = 0 in example 8.5 we get

( n +1) α
sin cos n +2α
1 + cos α + cos 2α + ... + cos nα = 2
, and
sin α2
( n +1) α
sin 2 sin n +2α
sin α + sin 2α + ... + sin nα = .■
sin α2

8.7. Let E be the point with affix 1. Consider the points A and A' with affixes
a = ρeiθ and a ' = ρ ' eiθ ' ,
respectively (figure 7). The product b = aa ' corresponds to a point B, obtained as the
third vertex of the triangle OA ' B , if this triangle is constructed as a similar one to the
triangle OEA.

30
Figure 7 Figure 8

Indeed, the similarity of these triangles implies ∠EOA = ∠A ' OB , i.e. arg b = θ + θ '
. For the same reason ρ : 1 = | b |: ρ '' holds, i.e. b = ρρ' holds. Therefore, b = aa ' .
The point Z whose affix is the
complex number z = aa' is obtained by
construction the triangle OZA' similar
to the triangle ОЕА.
Indeed, the similarity of these
triangles implies az = a ' . Therefore,
z = aa' (figure 8).
By using the relation a n = a n −1a
and consecutively applying the proce-
dures for constructing the affixes of the
product and the quotient of two complex
numbers, we obtain the points
..., A−3 , A−2 , A−1 , E , A1 , A2 , A3 ,...
whose affixes are the complex numbers
..., a −3 , a −2 , a −1 , a 0 , a1 , a 2 , a3 ,... ,
respectively.
Let be r > 1 and 0 < α < π . The
points A2 , A3 ,... (figure 9), with affixes
a 2 , a3 ,... are obtained by a consecutive
construction of the similar triangles
OEA1 , OA1 A2 , OA2 A3 ,... .
Constructing the similar triangles
OEA1 , OA−1E , OA−2 A−1 , OA−3 A−2 ,... Figure 9

31
by applying the same procedure, but in the opposite direction, we get the points
A−1 , A−2 , A−3 ,...
with affixes
a −1 , a −2 , a −3 ,... .
n
For ρ = r , θ = nα and after eliminating n in the above equations we obtain that
θ
ρ = r α . Thus, each powers a n lie on the curve, which is presented in polar form by the
previous relation. This curve is known to be the logarithmic (Bernoulli) spiral. Clearly,
in the previous considerations the absolute values of the powers increase or decrease like
a geometric progression, while the arguments – like an arithmetic progression.
Obviously, for r < 1 and 0 < α < π , or r > 1 and − π < α < 0 , the logarithmic
spiral is in opposite direction of the spiral given in figure 9 and it wraps around the origin
when θ decreases. Likewise if r < 1 and − π < α < 0 , then the logarithmic spiral has the
same appearance as shown in figure 9.

9. THE SET Cn

9.1. Definition. The notation a = (a1 , a2 ,..., an ) , for ai ∈C , and i = 1, 2,..., n is

called to be an ordered n-tuple of complex numbers.
Thereby, we accept the notation
Cn = {(a1 , a2 ,..., an ) | ai ∈ C, i = 1, 2,..., n} .

9.2. Definition. Let a = (a1 , a2 ,..., an ) and b = (b1 , b2 ,..., bn ) be two ordered n-
tuples of complex numbers, i.e. a, b ∈ Cn . The sum of a and b is the ordered n-tuple
c = (a1 + b1 , a2 + b2 ,..., an + bn ) .
Thereby, we accept the notation c = a + b .

9.3. Theorem. а) a + b = b + a , for all a, b ∈ Cn .

b) (a + b) + c = a + (b + c) , for all a, b, c ∈ Cn .
c) There exists o ∈ Cn such that a + o = a , for each a ∈ Cn .
d) For each a ∈ Cn there exists b ∈ Cn sothat a + b = o .
Proof. а) Let a = (a1 , a2 ,..., an ) and b = (b1 , b2 ,..., bn ) . By applying the com-
mutative and the associative lows of the addition of complex numbers, the definition 9.2
implies
a + b = (a1 + b1 , a2 + b2 ,..., an + bn )
= (b1 + a1 , b2 + a2 ,..., bn + an )
= b + a.

32
b) The proof is analogical to the previous one. We should use the associative law
of addition of complex numbers.
c) For an ordered n-tuple o = (0, 0,..., 0) and for each a ∈ Cn
a + o = (a1 , a2 ,..., an ) + (0, 0,..., 0)
= (a1 + 0, a2 + 0,..., an + 0)
= (a1 , a2 , ..., an ) = a.
holds true.
d) Let a = (a1 , a2 ,..., an ) be a given ordered n-tuple. By letting
b = (− a1 , − a2 ,..., − an ) , we get
a + b = (a1 − a1 , a2 − a2 ,..., an − an )
= (0, 0,..., 0) = o.
Hence, for each a ∈ C ther exists b ∈ Cn so that a + b = o . Moreover, the ordered n-
n

tuple b is called to be an opposite n-tuple of a and is denoted by b = − a . ■

9.4. Definition. Let be an ordered n-tuple and λ ∈C . The product of the ordered
n-tuple a and the complex number λ is the ordered n-tuple c = (λa1 , λa2 ,..., λan ) .
Therefore we accept the notation c = λa .

9.5. Theorem. а) λ (a + b) = λa + λb , for all a, b ∈ Cn and λ ∈C .

b) (λ + µ)a = λa + µa , for all λ, µ ∈C and a ∈ Cn .
c) (λµ)a = λ (µa) , for all λ, µ ∈C and a ∈ Cn .
d) 1⋅ a = a for each a ∈ Cn .
Proof. а) Let a = (a1 , a2 ,..., an ) , b = (b1 , b2 ,..., bn ) and λ ∈C . Since the defini-
tion 9.4 and by applying the distributive laws of addition and the multiplication to the
coordinates it follows that
λ (a + b) = λ ( (a1 , a2 ,..., an ) + (b1 , b2 ,..., bn ) )
= λ (a1 + b1 , a2 + b2 ,..., an + bn )
= ( λ (a1 + b1 ), λ (a2 + b2 ),..., λ (an + bn ) )
= (λa1 + λb1 , λa2 + λb2 , ..., λan + λbn )
= (λa1 , λa2 ,..., λan ) + (λb1 , λb2 ,..., λbn )
= λ (a1 , a2 , ..., an ) + λ (b1 , b2 ,..., bn ) = λa + λb.
The proofs for the other statements are analogous. While proving them it is nec-
essary to apply the distributive and the associative lows and also the fact that 1⋅ z = z , for
each z ∈C . ■

33
9.6. Definition. Scalar (inner) product of an ordered n-tuples a = (a1 , a2 ,..., an )
and b = (b1 , b2 ,..., bn ) is the complex number
f = a1 b1 + a2 b2 + ... + an bn .
Therefore we accept the notation f = (a, b) = a ⋅ b .
9.7. Theorem. а) (a, a) ∈ R + for each a ∈ Cn .
b) (a, b) = (b, a) , for all a, b ∈ Cn .
c) (a + b, c) = (a, c) + (b, c) for all a, b, c ∈ Cn .
d) (a, b + c) = (a, b) + (a, c) for all a, b, c ∈ Cn .
e) (λa, b) = λ (a, b) and (a, λb) = λ (a, b) for all a, b ∈ Cn and λ ∈C .
Proof. а) Let a = (a1 , a2 ,..., an ) . Then
(a, a) = a1 a1 + a2 a2 + ... + an an =| a1 |2 + | a2 |2 +...+ | an |2∈ R + .
b) By using the properties of conjugate complex numbers we obtain
(a, b) = a1 b1 + a2 b2 + ... + an bn
= a1b1 + a2b2 + ... + an bn = (b, a) .
The proofs of the other statements are direct implications of the scalar product
definition, the distributive and the associative laws of the operations in the set of comlex
numbers and the statement b). ■

9.8. Remark. The sum of ordered n-tuples and also the product of an ordered n-
tuple and real numbers are ordered n-tuples. On the other hand the scalar product of two
ordered n-tupes is a complex number.

9.9. We define the mapping T: Cn → Cn by

Ta = (a2 , a3 ,..., an , a1 ) , for each a = (a1 , a2 ,..., an ) .
By induction we determine T ma = TT m−1a , for m ≥ 2 . For example,
T 2a = T (a2 , a3 ,..., an , a1 ) = (a3 , a4 ,..., an , a1 , a2 ) .
We define a mapping T1 from Cn to Cn , such that each ordered n-tu-
ple a = (a1 , a2 ,..., an ) maps to an ordered n-tuple T1a = (an , a1 ,..., an −1 ) . Obviously,
T1Ta = TT1a = a , for each a ∈ Cn , i.e. the mappings T and T1 are inverse to each other.
Therefore, T1 = T −1 and by induction we determine
T − ma = T −1T −( m−1)a , for m ≥ 2 .

9.10. Theorem. а) T (a + b) = Ta + Tb for all a, b ∈ Cn .

b) T (λa) = λTa for each and each λ ∈C .
c) If a = (α, α,..., α) , α ∈C , then Ta = a .

34
d) T n a = T − n a = a , for each a ∈ Cn , (or more generally T nk a = T − nk a = a for
each a ∈ Cn and each k ∈ N ).
e) (T ma, T s b) = (T m+ k a, T s + k b) for all a, b ∈ Cn and all m, k , s ∈ N .
Proof а) Let a = (a1 , a2 ,..., an ) and b = (b1 , b2 ,..., bn ) be arbitrary elements of
n
C . Then,
T (a + b) = T ( (a1 , a2 ,..., an ) + (b1 , b2 ,..., bn ) )
= T (a1 + b1 , a2 + b2 ,..., an + bn )
= (a2 + b2 , a3 + b3 ,..., an + bn , a1 + b1 )
= (a2 , a3 ,..., an , a1 ) + (b2 , b3 ,..., bn , b1 )
= Ta + Tb.
The other statements can by proved analogously, by applying the definition of the map-
ping T and the properties of the arithmetic operations in Cn . ■

9.11. Corollary. For each a ∈ Cn

(Ta, a) = (T 2a, Ta) = ... holds.
Proof. The proof is directly implied by Theorem 9.10. e). ■

35
CHAPTER II
TRANSFORMATION IN EUCLEDIAN PLANE

In this chapter we will firstly discuss few elementary transformations in the complex
plane.Special attention will be paid to the similarities, their group propertiesand classification.
Furthermore, the inversion and the Möbius transformation, treated as the most important
elementary transformation of the complex plane, will be elaborated in separate paragraphs.

1. LINE EQUATION. PARALLEL

AND PERPEDNICULAR LINES

1.1. Let the line (p) do not pass through the origin and let the point A, with affix
a, be symmetric to the origin O with respect to (p). Then, a point B, with affix z, is on the
line (p) if and only if OB = AB , i.e. | z | = | z − a | , that is
z z = ( z − a )( z − a ) .
The last equality may be transformed and rewritten as the following
az + az = aa . (1)
If (p) passes through the origin and the points A and A' , with affixes a and a ' ,
respectively, are symmetric to each other with respect to the origin O and to the line (p),
then any arbitrary point B with affix z, such that B lies on (p) satisfies the following rela-
tion AB = A ' B , i.e. | z + a | = | z − a | , that is
( z + a )( z + a ) = ( z − a )( z − a ) .
The last equality may be transformed and rewritten as the following
az + az = 0 . (2)
iϕ −iϕ
If a = re , then a = re . Hence, if we divide the equalities (1) and (2) by a
we obtain the following equations
z = η z + a (3)
and
z = η z , (4)
where η = − a = −e 2iϕ . The number η is called to be complex gradient of the line (p), and
a
the point A is called to be mirror point of the line (p). Obviously, each line (p), which
does not pass through the origin, is determined by the mirror point A, with affix a = reiϕ
and the complex gradient η = −e 2iϕ . Each line (p), which passes through the origin, is
uniquely determined by its complex gradient. It is easy to prove that in both cases, the
angle between the line (p) and the positive part of the real axis is α = ϕ − π2 .

36
The above stated implies the validity of the following theorem.

Theorem. If A, with affix a, is a symmetric point to the origin with respect to a

given line (p), such that it does not pass through the origin and if ϕ is the oriented (directed)
angle between the real axis and the line through the origin and perpendicular to (p), then
the equation of (p) is given by (3), where η = −e 2iϕ . If (p) passes through the origin, then
its equation is given by (4). ■

1.2. Theorem. The equation of a line (p), which passes through two distinct
points A and B with affixes z0 and z1 , respectively, is
z1 − z0
z − z0 = ( z − z0 ) (5)
z1 − z0
and its complex gradient is
z1 − z0
η= . (6)
z1 − z0

Proof. Let z0 and z1 be the affixes of A and B, respectively. By substituting

these affixes into the equation (3) we get that z0 = η z0 + a and z1 = η z1 + a . Further, by
subtracting the last two equalities we obtain the complex gradient of the line as following
z1 − z0
η= , i.e. the equality (6) holds true. If the so obtained expression for complex
z1 − z0
gradient η is substitute z0 = η z0 + a we obtain the following :
z1 − z0
a = z0 − z ,
z1 − z0 0
Moreover, if the above determined values for η and a, we substitute into the equation (3)
we get the equation (5). ■

1.3. Corollary. The points z0 , z1 and z2 are collinear if and only if

z2 − z0 z2 − z0
= . (7)
z1 − z0 z1 − z0

Proof. According to Theorem 1.2, the equation of a line (p), such that it passes
through z0 and z1 , is given by (5). The points z0 , z1 and z2 are collinear if and only if
z2 satisfies the equality (5), that is if and only if the equality which is equivalent to (7).
z1 − z0
z2 − z0 = ( z2 − z0 ) ,
z1 − z0
is satisfied. ■

z −z
1.4. Corollary. The points z0 , z1 and z2 are collinear if and only if z2 − z 0 is a
1 0
real number.
Proof. The proof is directly implicated by Corollary 1.3 and the properties of the
complex numbers.■

37
z1 − z0
1.5. Remark. Since | η | = , the equation (5) can be rewritten as
z1 − z0

z − z0 = η( z − z0 ) , | η | = 1 ������
(8)
Conversely, each
���������������������������������������������
equation of type (8) is a line equation.
Indeed, thereby | η | = 1, it follows that it exists ϕ ∈[0, π) so that η = e 2iϕ . Thus, the
equation of the line through z0 and z1 = z0 + eiϕ , shall be as (8).

1.6. Remark. According to theorem 1.1 the equation of a line which passes
through the origin and the point z0 ≠ 0 is the following
z0
z = ηz , η = = e 2iϕ (9)
z0
The line (9) passes through points whose affixes are the square roots of the complex
gradient η.
Indeed, by substituting one of the two values of the square root of η into (9) we
obtain the following
2 2
η η= η η = η⋅ η = η,
i.e. the points with affixes η satisfy the equation (9).

1.7. Theorem. The oriented angle ϕ between the lines (p) and (q) with complex
η
gradients η1 = −e 2iϕ1 and η2 = −e 2iϕ2 , respectively, is given by the formula e 2iϕ = η1 .
2

Proof. Let ( p ') and (q ') be perpendicular to (p) and (q), respectively. Then, ac-
cording to theorem 1.1 the lines ( p ') and (q ') and the positive part of the real axis create
oriented angles ϕ1 and ϕ2 , respectively. Thus, the oriented angle of the above lines is
ϕ = ϕ2 − ϕ1 and it is congruent to the angle of (p) and (q) (as angles with perpendicular
rays). Thus, the statement given in the theorem is implicated by the relation
η1 2 iϕ1
η2
= −e2iϕ = e 2i (ϕ2 −ϕ1 ) = e 2iϕ . ■
−e 2

1.8. The equality ϕ = 0 is equivalent to the equality η1 = η2 , and ϕ = π2 is equiv-

alent to η1 = −η2 . Hence, the following corollary holds true.

Corollary А. а) Two lines are parallel if and only if their complex gradients are
equal.
b) Two lines are perpendicular if and only if their complex gradients are opposite
numbers. ■

38
Corollary B. а) Let the points M i , i = 1, 2, 3, 4 have affixes zi , i = 1, 2, 3, 4 . The
z −z
lines M1M 2 and M 3 M 4 are perpendicular to each other if and only if z1 − z2 ∈ iR*,
3 4
where R* = R \ {0} .
b) The lines M1M 2 and M 3 M 4 are perpendicular to each other if and only if
z1 − z2
z −z
∈ iR*,.
3 4

z −z
Proof. а) The complex gradients of the lines M1M 2 and M 3 M 4 are 1 2 and
z1 − z2
z3 − z4
, respectively. Since Corollary А, the lines M1M 2 and M 3 M 4 are perpendicular
z3 − z4
z1 − z2 z3 − z4 z −z z −z
to each other if and only if =− , i.e. if and only if z1 − z2 = − z1 − z2 , that is if
z1 − z2 z3 − z4 3 4 3 4
z −z
and only if z1 − z2 ∈ iR*,.
3 4

b) The proof is a direct implication of the statement a). ■

Corollary C. а) The equation of a line ( p ') through a point M with affix m and
is such that it is parallel to ( p ) : z = η z + a is the following z − m = η( z − m) .
b) The equation of a line ( p ') through a point M with affix m and is such that it
is perpendicular to ( p ) : z = η z + a is the following z − m = −η( z − m) .
Proof. The proof is a direct implication of Remark 1.5 and Corollary А. ■

1.9. Example. Let A and B be two distinct points in the complex plane with
affixes z1 and z2 , respectively. Determine the affix p ' of the point P ' , symmetric to P
with affix p, with respect to the line AB.
Solution. Through the point P we draw the line l, perpendicular to the line AB and
we find P1 , the point of intersection () of l and AB. Thus, P1 ( p1 ) is the midpoint of the line
p+ p '
segment PP ' , i.e. p1 = 2 that is p ' = 2 p1 − p , and the point P1 is the projection of P
onto the line AB.
The equation of the line through the points А ����
and B is the following
z2 − z1
z − z1 = ( z − z1 ) . (10)
z2 − z1
z −z
The complex gradient of the line l is η1 = − 2 1 . Thus, its equation is
z2 − z1
z −z
z − p = − 2 1 ( z − p ) . (11)
z2 − z1
If we add the last two equations (10) and (11) we obtain the affix p1 of the point P1 :
( p − z1 )( z2 − z1 ) + ( z2 − z1 )( p + z1 )
p1 = . (12)
2( z2 − z1 )

39
By substituting the so obtained expression for p1 in p ' = 2 p1 − p , we find the affix p '
of the point P′:
p ( z2 − z1 ) + z2 z1 − z2 z1
p' = .■
z2 − z1

1.10. Example. Determine the locus of the points which are equidistant from two
given points A and B.
Solution. Let a and b be the affixes of the points A and B, respectively, and let
the point M with affix z belong to the required locus. Then | z − a |2 = | z − b |2 thereby
MA = MB . The last equation is equivalent to
(
z − a +2 b = − b −a z − a +2 b .
b−a
)
Thus, the required locus is a line which passes through the midpoint of the line segment
AB and is perpendicular to AB. ■

1.11. Example. Let ABC be a given triangle and let K and H be such points on the
sides AB and AC, that AK = 1p AB and AH = p1+1 AC , respectively. Prove the following
statement: for each p, p > 0 the lines KH passthrough a unique point.
Solution. Let 0, b, c, k, h be the affixes of A, B, C, K, H, respectively. Then
k = bp , h = pc+1 . Since Corollary 1.4, the point M with affix z lies on the line KH if and only
if z − k = t ∈ R . Hence,
h−k
(
z = 1p b + pt+1 ( (c − b) p − b ) . )
If t = p +1 , then we get that z = c − b . Therefore, each line KH consists of the point X
with affix c − b . ■

2. DISTANCE FROM A POINT TO A LINE

2.1. Lemma. The line equation

z1 − z0
z − z0 = ( z − z0 )
z1 − z0
can be written as
Az + Bz + C = 0 , where C ∈ R and B = A ≠ 0 . (1)
Conversely, each equation as (1) is a line equation.
Proof. Let be given the equation
z1 − z0
z − z0 = ( z − z0 ) .
z1 − z0

40
So,
z ( z1 − z0 ) − z ( z1 − z0 ) + z0 z1 − z0 z1 = 0 .
If we multiply the latter by i, we get that
i ( z1 − z0 ) z − i ( z1 − z0 ) z + i ( z0 z1 − z0 z1 ) = 0 .
Let
A = i ( z1 − z0 ), B = −i ( z1 − z0 ), C = i ( z0 z1 − z0 z1 ) ,
the equation of a line through the points M and N with affixes z0 and z1 , respectively is
as (1).
Conversely, let the equation (1) be given. If (1) is divided by A and and then set
that η = − BA , a = − CA , we obtain the equation as z = η z + a, | η | = 1 . Since Theorem 1.1,
the letter is a line equation. ■

2.2. Definition.The line equation (1) is called to be a self-conjugate line equa-

tion.

2.3. Let line (p) be given by its self-conjugated equation (1) and the point z0 . If
(1) is rewritten as
z = − BA z − CA ,
then the complex gradient of an arbitrary line perpendicular to (p) is η' = BA . Thus, the
equation of (q) such that it passes through the point z0 and is perpendicular to (p) is the
following
z − z0 = BA ( z − z0 ) ,
i.e.
Az − Bz − z0 A + z0 B = 0 . (2)
By adding the equations (1) and (2), we obtain that 2 Az ' = Az0 − Bz0 − C is the intersec-
tion of (p) and (q), i.e. the projection z0 of onto the line (p), that is
Az0 − B z0 − C
z' = 2A
.
Thus,
Az0 + B z0 + C
z0 − z ' = ,
2A
so, the distance from a point z0 to a line (p), given by its self-conjugate equation (1) is
| Az0 + B z0 + C |
d ( z0 ,( p ) ) = | 2 A|
.

41
3. CIRCLE EQUATION

3.1. As already stated, | z − z0 | = R is the equation of a circle centered at S (with

affix z0 ) and radius R. In this section we will discuss circles in the complex plane.

3.2. Example. Let P1 and P2 be arbitrary points in the complex plane with
affixes z1 and z2 , respectively. Prove that the circumcircle of the line segment P1P2 ,
viewed as its diameter, has the following equation
| 2 z − z1 − z2 | = | z1 − z2 | . (1)
Solution. Since, the radius of a circumscircle of the line segment P1P2 ,
| z −z |
viewed as its diameter, is R = 1 2 2 and P0 (the midpoint of the line segment P1P2 ,
z1 + z2
with affix z0 = 2
) is its center, we get that the equation of the considered circle is
z1 + z2 | z1 − z2 |
z− 2
= 2 . If we multiply the last equation by 2, then we obtain the equation
which is equivalent to the latter, that is we obtain the equation (1). ■

3.3. Example. Let A, B and C be three distinct points in a plane. Determine the
locus of the points equidistant to points A, B and C.
Solution. Let a, b and c be the affixes of the points A, B and C, respectively. Since
Example 1.10, the locus of the points equidistant to the points A and B, B and C, A and C,
are the bisectors of the line segments AB, BC and CA and their equations are

b−a
(
z − a +2 b = − b −a z − a +2 b ) (2)

z − b +2 c = − −c
b
b −c
( z − b+2c ) (3)

z − a +2 c = − a −c
a −c
( z − a2+c ) (4)
respectively. We will discuss two cases:
а) If the points A, B and C are collinear, then Corollary 1.3 implies that the bisec-
tors of the line segments AB, BC and CA have equal complex gradients, and since Corollary
1.8 it implies that the lines are parallel. But, the fact that the points A, B and C differ from
each other, implies that the midpoints of the line segments AB, BC and CA, also differ from
each other. The latter means that it does not exist any point such that it satisfies the given
conditions.
b) If the points A, B and C are non-collinear, then the bisectors of the line segments
AB, BC and CA parwise intersect . If we subtract the equations (3) from (2), we obtain the
affix o of O, the point of intersection of the bisectors of the segments AB and BC,
aa (c −b ) +bb ( a −c ) + cc (b − a )
o= .
ab +bc + ca − ab −bc −ca
By direct checking we prove that the point O lies on the line CA. Therefore, the required
locus is the point O with affix o. ■

42
3.4. Remark. In the previous example, we actually proved that through three dis-
tinct non-collinear points A, B and C it passes exactly one i.e. a unique circle which is cen-
tered at O and whose radius is R = | a − o | . That is we proved that for each triangle there
exists a unique circumcircle, and the center of the such circle is the point of intersection of
the side bisectors of the given triangle.

3.5. Example. Prove that all complex numbers so that | z − 1 | = 2 | z + 1 | is satis-

fied, lie on a same circle. Determine the centre and the radius of that circle.
Solution. For any complex number z = x + iy , its absolute value is given by
| z |= x 2 + y 2 .
This fact applied to the given identity | z − 1 | = 2 | z + 1 | implies that

( x − 1) 2 + y 2 = 2 ( x + 1) 2 + y 2 .

( ) ()
2 2
After reducing, we obtain the following expression: x + 53 + y 2 = 43 , which obvi-

( )
ously is the equation of a circle centred at − 53 , 0 and whose radius is 43 . ■

3.6. As above stated, the equation of a circle with centre z0 and radius R is
| z − z0 | = R . But, it is useful also to emphasize the circle equation similar to the self-
conjugate line equation. Therefore, we will show that
z z + Az + Az + B = 0, B ∈ R , A ∈ C, | A |2 − B > 0 (5)
is a circle equation.
Indeed, if
z0 = − A and R 2 = z0 z0 − B = | A |2 − B > 0 ,
and we substitute in (5) we obtain the following
z z − z0 z − z z0 + z0 z0 = R 2 ,
i.e. the equation | z − z0 | = R , which is the equation of a circle with centre z0 and ra-
dius R. Hence, the equation (5) is the equation of a circle with centre z0 and radius
R = | A |2 − B , which is called to be a self-conjugate circle equation.

3.7. Remark. In the above discussions , in chapter 1, we proved that the ste-
reographic projections of a line and a circle in the extended complex plane, i.e. their
Reimann interpretation, are circles which consist of or do not consist of the pole, respec-
tively. This is one of the reasons the lines and the circles in the extended complex plane
are called as circles, and the circles in the complex plane are called as true circles. In the
following example we will give one more argument which enforces this terminology.

3.8. Example. (Apollonius circle). Let A and B be arbitrary points in the plane.
The locus of the point M so that MA : MB = k , (k > 0, k ≠ 1) is a circle. Prove it!

43
Solution. We will consider the case where k > 1 . Firstly, let set a coordinate sys-
tem xOy such that the x-axis coincides with the line AB, and the origin with the midpoint
of the line segment AB. Hence, A(a,0) and B(−a,0), i.e. the affixes of the points A and
B are z1 = a and z2 = − a , respectively. If the point M, which belongs to the considered
| z −a |
locus, has affix z, then the given condition implies that k = | z + a | , i.e.
2
z z + a k 2 +1 ( z + z ) + a 2 = 0 . (6)
k −1
The constants
2
A = a k 2 +1 , B = a 2
k −1
satisfy the condition
| A |2 − B > 0 .
Therefore, (6) is the equation of a circle with centre
2
z0 = − a k 2 +1 ,
k −1
and radius
R = | A |2 − B = 22ak .
k −1
The case where 0 < k < 1 can be considered analogously. ■

3.9. Relationship between a line and a circle. Let z − z0 = η( z − z0 ) and

| z − z1 | = R be the equations of a given line (p) and a circle (K), respectively. Through
the centre of the circle with affix we draw the line ( p ') perpendicular to (p). The line
equation of ( p ') is z − z1 = −η( z − z1 ) . If we add the equations of the lines (p) and ( p ')
then we find the affix of the point of intersection of these two lines
η( z1 − z0 ) + z1 + z0
z* = 2
.
So, the distance between the centre of the circle and the line (p) is the following
η( z1 − z0 ) + z1 − z0
d ( z*, z0 ) = | z * − z0 | = 2
.
The above stated implies that:
η( z − z ) + z − z0
- if 1 0
2
1
= R , then the line (p) is tangent to the circle (K) and the point
of touching has affix z*;
η( z1 − z0 ) + z1 − z0
- if 2
< R , then the line (p) and the circle (K) have two points of
intersection;
η( z − z0 ) + z1 − z0
- if 1
2
> R , then the line (p) and the circle (K) do not have any
common points.

44
3.10. Example. Determine the relationship between the line (p) and the circle (K)
whose equations are z = z + 3i and | z + 4 − 2i | = 3 , respectively.
Solution. Since the line equation z = z + 3i , we get that z0 = 32i . Further, since
the circle equation
| z + 4 − 2i | = 3 ,
we obtain that z1 = − 4 + 2i and R = 3 . Therefore,
η( z1 − z0 ) + z1 − z0 ( −4−2i + 32i )+( −4+ 2i − 32i ) | −8|
d= 2
= = 2 =4>3= R,
2
and thereby 3.9, the line (p) and the circle (K) do not have any common points i.e. they
do not have any points of inetrsection.■

3.11. Example. Let (K): | z − z0 | = R be a given circle and z1 be a point which is

placed on the circle. Find the equation of the tangent to the circle (K) at the point z1 .
Solution. The equation of a line (p), such that it passes through points z0 and
z1 , is the following
z1 − z0
z − z0 = ( z − z0 ) .
z1 − z0
So, the equation of the tangent ( p ') to (K) at the point z1 is the following
z1 − z0
z − z1 = − ( z − z1 ) . ■
z1 − z0

3.12. Remark. а) If (K): | z | = 1 is the unit circle and z1 is a point on (K), then
the equation of the tangent to (K) at the point z1 is the following z + z12 z = 2 z1 .
b) If A, B, C and D with affixes a, b, c and d, respectively, are on the unit circle
(K): | z | = 1, then a = a −1 , b = b −1 , c = c −1 and d = d −1 . Thereby Corollary 1.8, the
chords AB and CD are parallel if and only if
(b − a )(d − c) = (b − a )(d − c)
holds true, that is if and only if ab = cd . Analogously, the chords AB and CD are perpen-
dicular to each other if and only if ab + cd = 0 .
Apparently, if A and B with affixes a and b, respectively are points on the unit
circle, then a = a −1 , b = b −1 , and therefore
a −b = a −b = − ab .
a −b a −1 −b −1
Further, if M with affix m is a point on the chord AB, then by Corollary 1.3 it is true that
m−a
= m−a = ma −−ba ab = − mab −b ,
b−a
b−a b−a
and by equivalent transformations, we express m as the following m = a +ab b−m .

c) Let A, B, C and D with affixes a, b, c and d, respectively, be points on the

unit circle (K): | z | = 1, and AB ∩ CD = {S }. The equations of the lines AB and CD are

45
z + abz = a + b and z + cd z = c + d , respectively. By eliminating z from the last two
( a +b )cd −(c + d ) ab
equations, we obtain that the affix of S is s = cd − ab
.
d) Let A and B, with affixes a and b, respectively, be points on the unit circle, so
that the line segment AB is not its diameter. According to the statement a), the equations
of tangents (t A ) and (t B ) are z + a 2 z = 2a and z + b 2 z = 2b , respectively. If we elimi-
nate z from the last two equations, we get that s = a2+abb , is the affix of the intersection
point S = (t A ) ∩ (t B ) .
e) Let the line (p) meet the unit circle at two points A and B, whose affixes are a
and b, respectively, and let M, whose affix is m, be an arbitrary point on the plane. It is
easy to prove that the affix of the orthogonal projection of M onto the line (p), is given
by c = a +b + m2 −abm .

4. DIRECT SIMILARITIES

4.1. Definition. The mapping S : C → C defined by

w = S ( z ) = az + b, a, b ∈ C, a ≠ 0 (1)
is called to be a direct similarity.

4.2. Theorem. The set of the direct similarities DS under the operation composi-
tion of mappings is a noncommutative group.
Proof. If S1 ,S2 ∈DS , then
S1 ( z ) = az + b, a, b ∈ C, a ≠ 0 and S2 ( z ) = cz + d , c, d ∈ C, d ≠ 0 .
Hence,
S1 ( S2 ( z )) = S1 (cz + d ) = a (cz + d ) + b = (ac) z + (ad + b), ac, ad + b ∈ C, ac ≠ 0
i.e. S1  S2 ∈DS . Thus, the set DS is closed with respect to the composition of mappings
and in general the following holds true
S1 ( S2 ( z )) = (ac) z + (ad + b) ≠ (ac) z + (bc + d ) = S2 ( S1 ( z )) .
Let S1 ,S2 , S3 ∈DS . With direct checking the following could be proved
S1  ( S2  S3 )( z ) = ( S1  S2 )  S3 )( z ) , for each z ∈C .
Therefore, S1  ( S2  S3 ) = ( S1  S2 )  S3 , i.e. the associative law holds true.
The mapping E ( z ) = z , for each z ∈C is an element of DS and further
E  S = S  E = S , for each S ∈DS .
Let S ( z ) = az + b, a, b ∈ C, a ≠ 0 be an arbitrary direct similarity. The mapping
defined by S1 ( z ) = 1a z − ba is a direct similarity and furthermore, the following holds
true
S ( S1 ( z ) ) = S1 ( S ( z ) ) , for z ∈C , i.e. S −1 = S1 ∈ DS . ■

46
4.3. Theorem. Each direct similarity is uniquely determined by two pairs of cor-
responding points.
Proof. Let S be an arbitrary direct similarity so that S ( z1 ) = w1 and S ( z2 ) = w2 .
Then S ( z ) = az + b, where a, b ∈ C, a ≠ 0 are coefficients which should be obtained.
Since Theorem 4.2 each direct similarity is bijection, and therefore z1 ≠ z2 implies that
w1 ≠ w1 . By substituting in S ( z ) = az + b we get the following system of linear equations
of the variables a and b
 w1 = az1 + b
 (2)
 w2 = az2 + b
By solving the system (2) with respect the variables a and b, we obtain
w −w z w −z w
a = z1 − z 2 , b = 1 z2 − z 2 1
1 2 1 2

and a ≠ 0 , i.e. the coefficients a and b of the direct similarity S ( z ) = az + b are com-
pletely determined by two pairs of corresponding points ( z1 , S ( z1 ) ) and ( z2 , S ( z2 ) ) . ■

4.4. Theorem. а) The image of a line (p) under a direct similarity is a line ( p ') .
b) Two parallel lines under direct similarity map to parallel lines.
c) Two perpendicular lines under direct similarity map to perpendicular lines.
Proof. а) Let be given the direct similarity (1) and a equation of (p) as z = η z + c .
According to (1), we get that z = wa−b and if we substitute in the line equation we get

(a )
w = a η w + ac + b − ab η .
a
aη = 1 implies that an image of line (p) under a direct similarity is the line
Further,
a
( p ') with complex gradient a η .
a
The proofs of the statements b) and c) are direct implications of the statement a)
and Corallary 1.8 A. ■

4.5. Theorem. The image of a circle (K) under direct similarity is a circle ( K ') .
Proof. Let be the given direct similarity (1) and a circle (K) with equation
| z − c | = R . According to (1), we get that z = wa−b and if we substitute in the circle
equation we obtain | w − (ac + b) | = | a | R . The latter actually means that the image of the
circle (K) under the given similarity (1) is the circle ( K ') . The centre of the image circle
has an affix ac + b , and the length of its radius is | a | R . ■

4.6. Theorem. If A, B are arbitrary distinct points, A ', B ' are their images under
the direct similarity (1), respectively, and if a = reiϕ , then A ' B ' = r AB , and the lines AB
and A ' B ' form an oriented angle ϕ.

47
Proof. Let z1 , z2 , w1 , w2 be the affixes of the points A, B, A ', B ' , respectively.
Hence, z2 − z1 = ABeiα and w2 − w1 = A ' B 'eiα1 , where α and α1 are the angles formed
 
by the real axis and the vectors AB and A ' B ' , respectively. By the equalities w1 = az1 + b
and w2 = az2 + b , we get the following equality
w2 − w1 = a ( z1 − z2 ) ,
i.e. the equality
A ' B 'eiα1 = r ABei (α +ϕ) ,
which implies that A ' B ' = r AB and ϕ = α1 − α . ■

The real number r is called to be the ratio of the direct similarity (stretching
factor or direct similarity coefficient) (1), and the angle ϕ is called to be an angle of the
direct similarity (1).

4.7. Definition. Two figures are said to be directly similar if there exists a direct
similarity under which one of the figures is mapped to the other one.

4.8. Corollary. If ABC and A ' B ' C ' are directly similar triangles, then
A ' B ' : A ' C ' = AB : AC and ∠A ' B ' C ' = ∠ABC .
Proof. The proof is directly implicated by Theorem 4.6.. ■

4.9. Theorem. Let z1 , z2 , z3 , w1 , w2 , w3 , be the affixes of A, B, C, A ', B ', C '

respectively. The triangles ABC and A ' B ' C ' are directly similar if and only if
z1 ( w2 − w3 ) + z2 ( w3 − w1 ) + z3 ( w1 − w2 ) = 0 (3)
holds.
Proof. The triangles ABC and A ' B ' C ' are directly similar if and only if there
exists a direct similarity (1) so that wi = azi + b , for i = 1, 2, 3 . The last equalities imply
the following
w1 − w2 = a ( z1 − z2 ) and w1 − w3 = a ( z1 − z3 ) .
If we divide the first equality by the other one, we actually obtain the following equality
w1 − w2 z −z
w1 − w3
= z1 − z2 . (3’)
1 3
The latter is equivalent to (3). ■

4.10. Remark. The condition (3), i.e. the condition (3’) of the previous Theorem
is equivalent to the following condition
1 1 1
z1 z2 z3 = 0 .
w1 w2 w3

48
Indeed, the condition (3’) and the properties of the determinants imply
1 1 1
z −z z1 − z3
z1 z2 z3 = 1 2
w1 − w2 w1 − w3
w1 w2 w3
= ( z1 − z2 )( w1 − w3 ) − ( z1 − z3 )( w1 − w2 ) = 0.

4.11. Definition. A point z is said to be a fixed point under the direct similarity
(1) if it satisfies the condition z = az + b .
Apparently, for a ≠ 1, the direct similarity (1) has a unique fixed point and its
affix is z1 = 1−ba . If a = 1 , then b = 0 , i.e. the direct similarity (1) is the identity mapping
and each point of the complex plane is a fixed point.
The point C with affix c = 1−ba is said to be center of the direct similarity
S ( z ) = az + b .

4.12. Let (p): z − z0 = η ( z − z0 ) be a tangent to the circle (K): | z − z1 | = R and

let w = S ( z ) = az + b, a, b ∈ C, a ≠ 0 be a direct similarity. Since 3.9 ,
η( z1 − z0 ) + z1 − z0
2
= R (4)
holds. Further, according to Theorem 4.5 the image of the circle (K) is a circle ( K ') with
equation | w − (az1 + b) | = | a | R . Analogously to the proof of Theorem 4.4 а) if we sub-
stitute that z = wa−b in the equation of (p) and we get that the image of a line (p) is a line
( p ') with the following equation
ηa
w − (az0 + b) = ( w − az0 + b) .
a
By using the equation (4) we find that the circle ( K ') and the line ( p ') satisfy the fol-
lowing
ηa ηa
( az1 +b − az0 +b ) + az1 +b −( az0 +b ) a ( z1 − z0 ) + a ( z1 − z0 )
a a
2
= 2
η( z1 − z0 ) + ( z1 − z0 )
=|a| 2
=|a|R
Hence, 3.9 implies that the line ( p ') is tangent to the circle ( K ') .
Thus, we proved the following theorem.

Theorem. Let (p) be a tangent line to the circle (K) and ( p ') and ( K ') be their
images under the direct similarity (1), respectively. Then ( p ') is a tangent line to the
circle ( K ') . ■

4.13. Example. Let ABCD be a given parallelogram. On its sides CD and CB,
similar and same oriented triangles (directly similar) CDE and FBC, are constructed.

49
Prove that the triangle FAE is similar and is also the same oriented to the triangles CDE
and FBC.
Solution. Let the origin be the point of intersection of the parallelogram diago-
nals. Then c = − a and d = −b . The triangles CDE and FBC are similar and a same ori-
ented. Thus, Theorem 4.9 implies that bc−−bf = ed−−dc . Therefore,
2 2
f = be+ce−−dbc −cd = bee++ba .
Hence,
(b − a )(e − a )
f −a= , c − d = c + b , d − e = −(b + e) and b − a = c − d ,
e +b
therefore
( b − a )( e − a )
f −a
= −(be−+ab) = cd−−de ,
a −e
= e+b
a −e
The latter according to Theorem 4.9, means that the triangles FAE and CDE are directly
similar.■

Example 4.14. On the sides AB, BC and CA of a triangle ABC, such pairwise
similar triangles ABK, BCL and ACM are constructed,
that the first two are out of the triangle ABC and the
third one is in (see figure 1). Prove that the quadrilateral
BLMK is parallelogram.
Solution. The triangles AKB and BLC are di-
rectly similar. Therefore,
k −a = cl −−bb , i.e. l = b + (k − a ) bc −−ba .
b−a
The triangles AKB and AMC are directly similar, and
Figure 1 therefore bk −−aa = mc −−aa , i.e.
m = a + (k − a ) bc −−aa .
Thus, 
BL = l − b = (k − a ) bc −−ba
and

(
KM = m − k = a + (k − a ) bc −−aa − k = (k − a ) bc −−aa − 1 = (k − a ) bc −−ba , )
 
i.e. BL = KM . The latter means that the quadrilateral BLMK is parallelogram. ■

50
5. MOTIONS

5.1. In the previous section we considered and discussed the direct similarities
and we also proved several properties about them. In this section the focus of our interest
is one of the most important classes of direct similaritiesand their classification.

5.2. Definition. The direct similarity S ( z ) = az + b, for | a | = 1 , for | a | = 1 is

said to be motion.

5.3. Theorem. The set of motions D under the operation composition of mappings
is a subgroup of the direct similarity group DS.
Proof. If S1 , S2 ∈D , then
S1 ( z ) = az + b, S2 ( z ) = cz + d , | a | = | d | = 1 .
Thus,
S1 ( S2 ( z ) ) = S1 (cz + d ) = a (cz + d ) + b = (ac) z + (ad + b), | ac | = 1 ,
therefore, S1  S2 ∈D . The latter means that the set D is closed under the composition of
mappings.
If S1 , S2 , S3 ∈D , then S1 , S2 , S3 ∈DS , therefore
S1  ( S2  S3 ) = ( S1  S2 )  S3 ,
i.e. the associative law holds true.
If a = 1, b = 0 then 1 ⋅ z + 0 = E ( z ) ∈ D .
Let S ( z ) = az + b, | a |= 1 be an arbitrary motion. The mapping
S1 ( z ) = 1a z − ba , 1a = | 1a | = 1
is motion and moreover the following holds true
S ( S1 ( z ) ) = S1 ( S ( z ) ) = z , for each z ∈C , i.e. S −1 = S1 ∈ D . ■

5.4. Definition. The motion S ( z ) = z + b is said to be translation for the vector

b, and it is denoted by Sb .

5.5. Theorem. The translation which is not identity mapping has no fix points.
Proof. The proof is directly implicated by 4.10. ■

5.6. Theorem. The set of translations T under the composition of mappings is a

commutative subgroup of the group of motions D.
Proof. If S1 , S2 ∈T , then S1 ( z ) = z + b, S2 ( z ) = z + d . Thus,
S1 ( S2 ( z ) ) = S1 ( z + d ) = ( z + d ) + b = z + (d + b),
therefore, S1  S2 ∈T . The latter means that the set T is closed under the composition of
mappings.

51
If S1 , S2 , S3 ∈T , then S1 , S2 , S3 ∈D , Hence,
S1  ( S2  S3 ) = ( S1  S2 )  S3 ,
i.e. the associative law holds true.
Let S1 , S2 ∈T , then S1 ( z ) = z + b, S2 ( z ) = z + d . Hence,
S1 ( S2 ( z ) ) = S1 ( z + d ) = ( z + d ) + b = ( z + b) + d
= S2 ( z + b) = S2 ( S1 ( z ) ) ,
for each z ∈C , i.e. the commutative law holds true.
If b = 0 then 1 ⋅ z + 0 = E ( z ) ∈T .
Let S ( z ) = z + b be translation. The mapping S1 ( z ) = z − b is also translation and
furthermore the following holds true
S ( S1 ( z ) ) = S1 ( S ( z ) ) = z , for each z ∈C , i.e. S −1 = S1 ∈ D . ■

5.7. Definition. The direct similarity with ratio 1 and angle π is said to be a point
reflection.
The shape F is said to be a point reflection iff there exists such a reflection S, that
S(F ) = F .
Thus, S ( z ) = az + b, a, b ∈ C, a ≠ 0 is a point reflection if a = −1. Due to the last,
S ( z ) = b − z denotes a point reflection.
Moreover 4.10, implies that the point reflection S ( z ) = b − z has center C with
affix c = b2 and is denoted by S = SC . In our further considerations the set of point reflec-
tions will be denoted by CS.
Let A(a) be any arbitrary point in a plane. The image of that point under the point
reflection S ( z ) = b − z is the point A '(b − a ) . Since the centre C of the point reflection
has an affix c = b2 , we get that  
AC = b2 − a = (b − a ) − b2 = CA ' .
So, the centre C of the point reflection is the midpoint of the line segment AA' .

5.8. Theorem. а) The composition of two point reflections is translation.

b) The composition of a point reflection and a translation is point reflection.
Proof. а) Let S1 ( z ) = b − z and S2 ( z ) = d − z , b, d ∈C be arbitrary point reflec-
tions. Then,
S1 ( S2 ( z ) ) = S1 (d − z ) = b − (d − z ) = z + (b − d ) .
So, the composition S1  S2 is translation for the vector b − d .
б) Let S1 ( z ) = b − z and S2 ( z ) = z + d , b, d ∈C be arbitrary point reflection and
translation, respectively. Therefore,
S1 ( S2 ( z ) ) = S1 ( z + d ) = b − ( z + d ) = b − d − z
and
S2 ( S1 ( z ) ) = S2 (b − z ) = d + (b − z ) = b + d − z ,
i.e. S1  S2 and S2  S1 are point reflections with centers b −2d and b +2d , respectively. ■

52
5.9. Definition. The mapping S : C → C is said to be involuntary iff the mapping
is invertible, i.e. there exists such S −1 that S −1 = S .

5.10. Theorem. The direct similarity which is not identity is involuntary if and
only if it is a point reflection.
Proof. Theorem 4.2 implies that the direct similarity is involuntary if and only
if
az + b = z −ab , for each z ∈C ,
i.e. if and only if a = 1a and b = − ba . The last two equalities are satisfied if and only if
a = −1. Therefore, S is involuntary if and only if S is a point reflection. ■

5.11. Corollary. The set T ∪ CS under the operation composition of mappings

is a non-commutative subgroup of the group of motions D.
Proof. The proof is a directly implicated by Theorem 5.6 and Theorem 5.8. ■

5.12. Definition. The motion which is not translation is said to be rotation.

Thereby each rotation S ( z ) = az + b, | a | = 1 satisfies that a ≠ 1, we deduce that
each rotation has center C with affix c = 1−ba . If C and α are the center of the rotation and
the angle of the rotation, respectively,we use to say that we have a rotation about C with
angle α and we use to write S = SC ,α . In our further discussion the set of the rotations
will be denoted by R. Apparently, the point reflections are rotations with angle π, and
therefore CS ⊂ R .
Let S ( z ) = az + b, | a |= 1, a ≠ 1 be rotation about C with angle α. Then, the in-
verse mapping S −1 defined by S −1 ( z ) = az − ab is rotation about C with angle −α.

5.13. Theorem. а) The composition of two rotations is either rotation or transla-

tion.
b) The composition of rotation and translation is rotation.
Proof. а) Let
S1 ( z ) = az + b, | a | = 1 , a ≠ 1 and S2 ( z ) = cz + d , | c | = 1 , c ≠ 1
be two rotations. So,
S1 ( S2 ( z ) ) = S1 (cz + d ) = a (cz + d ) + b = (ac) z + (ad + b) . (1)
Apparently, if ac = 1 , then S1  S2 is translation; if ac ≠ 1 , then S1  S2 is rotation about
C, with affix ad +b , and with angle α + α , where α and α are the angles of the
1− ac 1 2 1 2
rotations S1 and S2 , respectively.
b) Let
S1 ( z ) = z + b and S2 ( z ) = cz + d , | c | = 1 , c ≠ 1
be an arbitrary translation and a rotation, respectively.

53
Since,
S1 ( S2 ( z ) ) = S1 (cz + d ) = cz + (d + b)
it follows that S1  S2 is rotation about C with affix d1−+cb and with angle α 2 . Further,
S2 ( S1 ( z ) ) = S2 ( z + b) = cz + (d + bc)
implies that S2  S1 is rotation about C ' with affix d1+−bc
c
and with angle α 2 . ■

5.14. Consider the rotations

S1 ( z ) = az + b, | a | = 1 , a ≠ 1 and S2 ( z ) = cz + d , | c | = 1 , c ≠ 1.
While proving Theorem 5.13 we established that the composition S1  S2 is either rotation
or translation, depending on whether ac ≠ 1 or ac = 1 , respectively. Apparently,
S2 ( S1 ( z ) ) = S2 (az + b) = c(az + b) + d = (ac) z + (bc + d ) , (2)
implies that S2  S1 is either rotation or translation, too, depending on whether ac ≠ 1 or
ac = 1 , respectively. Logically the following question arises: whether and under which
conditions S1 and S2 commute under to the composition of mappings, i.e. when does
the following relation hold true
S2 ( S1 ( z ) ) = S1 ( S2 ( z ) ) , for each z ∈C . (3)
If we substitute (1) and (2) into (3), after reducing, we obtain that S1 and S2 commute
if and only if ad + b = bc + d . The latter actually means that S1 and S2 commute if and
only if 1−ba = 1−dc . So, we proved the following theorem.

Theorem. Two rotations commute if and only if their centers of rotations

coincide. ■

5.15. Consider the rotations about a common centre. Let

S1 ( z ) = az + b, | a | = 1 , a ≠ 1 and S2 ( z ) = cz + d , | c | = 1 , c ≠ 1,
be so that 1−ba = 1−dc holds. Then,
S2 ( S1 ( z ) ) = (ac) z + (bc + d ) , a ≠ 1, c ≠ 1, | a | = 1 and | c | = 1 .
Obviously, | ac | = 1 . If ac = 1 , then | c | = 1 implies cc = 1, and therefore a = c = 1c . By
substituting in 1−ba = 1−dc , we obtain that bc + d = 0 , that is S2 ( S1 ( z ) ) = z = E ( z ) . If
ac ≠ 1 , then the condition 1−ba = 1−dc is equivalent to bc + d = d , i.e. the composition
1− ac 1−c
S2  S1 is rotation with center which coincides with the centers of the rotations S1 and
S2 . Finally, if have on mind the fact that the identity mapping can be understood as a
rotation about an arbitrary centre, then the previous considerations and also Theorem
5.14 imply the validity of the following Theorem.

54
Theorem. The set of rotations about a common centre under the operation com-
position of mappings is a commutative sub-group of the group of motions D. ■

5.16. Example. Let ABCDEF be a regular hexagon, K be the midpoint of the

diagonal BD, and M be the midpoint of the side EF. Prove that  AMK is an equilateral
triangle.
Solution. Let the hexagon ABCDEF
be inscribed into the unit circle (figure 2). Then
iπ i 2π i 4π i 5π
1, e 3 , e 3 , eiπ , e 3 , e 3 are the affixes of the ver-
tices C, D, E, F, A, B, respectively. Thus the affixes
of the points K and M are
iπ i 5π iπ i 2π
k = e +2e = 12 and m = e 2+e = − 34 + i 43 ,
3 3 3

respectively. Further,
( k − a )e
i π3
(
+ a = 12 − e
i 4π
3
)e i π3
+e
i 43π
=
Figure 2
π i 53π i 43π
= 1 ei 3 −e +e = − 34 + i 43 = m,
2
iπ
i.e. (k − a )e 3 = m − a implies, that the side MA of  AMK is obtained when the side AK
is rotated about the vertex A with angle π3 . So,  AMK is an equilateral triangle. ■

5.17. Example. Let M k , k = 1, 2, 3, 4 with affixes zk , k = 1, 2, 3, 4 , respectively,

be given distinct points in the complex plane. Prove the following statement:
The identity
z2 − z1 = ±i ( z4 − z3 ) (1)
holds true if and only if
M1M 2 = M 3 M 4 and M1M 2 ⊥ M 3 M 4 (2)
Solution. Since the condition (1) we get | z2 − z1 | = | z4 − z3 | , that is
M1M 2 = M 3 M 4 . Likewise,
±i π
z2 − z1 = ±i ( z4 − z3 ) = e 2 ( z4 − z3 ) ,
that is the number z2 − z1 is obtained by rotation of the number z4 − z3 about the origin
O with angle ± π2 . Thus, M1M 2 ⊥ M 3 M 4 . Therefore, the condition (2) is implied by the
condition (2).
Conversely, since
M1M 2 = | z2 − z1 | , M 3 M 4 = | z4 − z3 | and M1M 2 = M 3 M 4

the following is satisfied z2 − z1 = reit , z4 − z3 = reis . Therefore

z2 − z1 = ei (t − s ) ( z4 − z3 ) . (3)

55
The second condition in (2) implies
t − s = ± π2 + 2k π, k ∈ Z .
By substituting in (3) we get z2 − z1 = ±i ( z4 − z3 ) . Thus, the condition (2) implies the
condition (1). ■

6. HOMOTHETY

6.1. Definition. The direct similarity with angle 0 or π, which is not translation
is called to be homothety.
Due to this, the direct similarity S ( z ) = az + b, a, b ∈ C, a ≠ 0 is homothety
if and only if a ∈ R \ {0,1} . Ratio of homothety is a real (not a complex number as it
was for generally directly similarities) number a, such that a, (a ≠ 0,1) . Apparently, the
point reflections are homotheties with ratio 1a . Theorem 4.2, implies that the inverse
mapping of a homothety with ratio a is also a homothety, but the ratio is 1a . In our further
considerations the set of the homotheties will be denoted by H.

6.2. Theorem. а) The composition of two homotheties is either a homothety or

a translation.
b) The composition of homothety and translation is homothety.
Proof. а) Let
S1 ( z ) = a1 z + b1 , a1 ∈ R \ {0,1}
and
S2 ( z ) = a2 z + b2 , a2 ∈ R \ {0,1}
be two homotheties. Then,
S2 ( S1 ( z ) ) = a1a2 z + a2b1 + b2 .
Obviously, if a1a2 = 1 , then S2  S1 is translation for the vector a2b1 + b2 , and if
a1a2 ≠ 1 , then S2  S1 is homothety with center C and a ratio of homothety a1a2 . The
a b +b
affix of such the center is 12−a1 a 2 .
1 2
b) Let the homothety and the translation be given by
S1 ( z ) = a1 z + b1 , a1 ∈ R \ {0,1} and S2 ( z ) = z + b2 , respectively.
Thereby,
S2 ( S1 ( z ) ) = a1 z + b1 + b2 , a1 ∈ R \ {0,1}
holds true, we get that S2  S1 is homothety with center C, the affix of the center is
b1 +b2
1− a
, and the ratio of the homothety is a1 . Since,
1
S1 ( S2 ( z ) ) = a1 z + b1 + a1b2 , a1 ∈ R \ {0,1}

56
holds true,we get that S1  S2 is a homothety with center C, the affix of the center is
a1b2 +b1
1− a
and the ratio of the homothety is a1 . ■
1

6.3. Corollary. The set T ∪ H is a non-commutative subgroup of the direct

similarity group DS under the composition of mappings.
Proof. The proof is directly implied by Definition 6.1 and Theorems 4.2, 5.6 and
6.2. ■

6.4. Theorem. Any two homotheties and their composition, if it is not transla-
tion, have collinear centers.
Proof. а) Let
S1 ( z ) = a1 z + b1 , a1 ∈ R \ {0,1} and S2 ( z ) = a2 z + b2 , a2 ∈ R \ {0,1}
b b
be two homotheties with centers C1 and C2 , whose affixes are c1 = 1−1a and c2 = 1−2a ,
1 2
respectively, and also let the composition
S2 ( S1 ( z ) ) = a1a2 z + a2b1 + b2
a b +b
be homothety with center C, whose affix is 12−a1 a 2 . Then,
1 2
b2 a2b1 +b2
−
c2 −c 1− a2 1− a1a2 a a −a
c1 −c
= b1 a2b1 +b2
= 11−2a 2
− 2
1− a1 1− a1a2
is real number, and thereby Corollary 1.4, the points C1 , C2 and C are collinear. ■

6.5. Theorem. The line (p) under a direct similarity S ( z ) = az + b is mapped to a

parallel line ( p ') if and only if the direct similarity is either homothety or translation.
Proof. If the line (p) has complex gradient η, then its image ( p ') under the direct
similarity S ( z ) = az + b has complex gradient η a .The straight lines (p) and ( p ') are
a
parallel if and only if η a = η , i.e. if and only if a = a , or in other words if and only if
a
a ∈ R . Hence, the direct similarity S ( z ) = az + b maps the line (p) to a parallel line ( p ')
if and only if the direct similarity is either a homothety or a translation. ■

6.6. Let | z − c1 | = R1 and | z − c2 | = R2 be the equations of circles ( K1 ) and

( K 2 ) , respectively.
If R1 ≠ R2 , then the mapping S : C → C defined by
R R1c2 − R2c1
w = S ( z ) = R2 z + R1
(1)
1
R R1c2 − R2c1
is a homothety with ratio R2 and center R1 − R2
. Since (1), we obtain the following
1
expression for z

57
R1w− R1c2 + R2c1
z= R2
and if we substitute it into the ( K1 ) circle equation, we get the following equation
R1w− R1c2 + R2c1
R2
− c1 = R1 ,
which in fact is equivalent to the ( K 2 ) circle equation.
If R1 = R2 , the mapping (1) is translation for the vector c2 − c1 and furthermore
it maps ( K1 ) onto ( K 2 ) .
Analogously, the mapping S : C → C defined by
R R1c2 + R2c1
w = S ( z ) = − R2 z + R1
(2)
1
R R1c2 + R2c1
is homothety with ratio − R2 and center R1 + R2
( K1 ) onto
, and furthermore it maps
1
(K2 ) .
The above statement implies the validity of the following Theorem.

Theorem. Any two circles are homotethic, i.e. there exists a homotethy which
maps one of the circles to the other one. ■

6.7. Example. Let B and C be arbitrary distinct points on a given circle, such that
they are not diametrically opposite and let the tangents to the given circle at these points
intersect at point A. Let P be an arbitrary point on the circle. Let A1 , B1 , C1 be the feet
of the perpendiculars from P to the lines BC, CA, AB, respectively.
Hence, PA12 = PB 1⋅ PC 1 . Prove it!
Solution. Without loss of generality we may assume that the given circle is the
unit circle, and 1 is the affix of P (why?).
Let b and c be the affixes of the points B and C, respectively. According to Remark
3.12 e) and d) it follows, that the affixes of the points A and A1 are the following
a = b2+bcc and a1 = b +c +21−bc ,
respectively. In order to determine the affix b1 of the point B1 we will use the fact that the
point B1 is on the line AC and furthermore that PB1 is perpendicular to AC. So,
b1 −c b 1−c b −1 b −1
a −c
= and a1−c = 1 .
a −c a −c
By substituting the expression for a in the last two equations, and after reducing, we get
the following system:
b1 + b 1c 2 = 2c
 2 2
b1 − b 1c = 1 − c
2 2
therefore b1 = 1+ 2c2−c . Analogously, c1 = 1+ 2b2−b . Due to this,

58
PA12 = | 1 − a1 |2 = 14 | bc − b − c + 1 |2 = 12 | b − 1 |2 ⋅ 12 | c − 1 |2
= | 1 − b1 | ⋅ | 1 − c1 | = PB 1⋅ PC 1. 

6.8. According to Theorem 5.5 and Definition 4.11, the direct similarity which
is neither identity nor translation, has exactly one fixed point, and such fixed point is the
center of the direct similarity. The line (p): z = η z + c is called to be a fixed line for the
direct similarity S if S ( p ) = p , i.e. if the direct similarity maps (p) to itself. The circle
(K): | z − c | = R is called to be a fixed circle for the direct similarity S if S ( K ) = K .

6.9. Theorem 4.5 implies that the image of the circle (K): | z − c | = R under the
direct similarity w = S ( z ) = az + b is the circle ( K ') : | z − (ac + b) | = R⋅ | a | . So, the circle
(K) is fixed under the direct similarity if and only if | a | = 1 and c = 1−ba , i.e. if and only if
S is a motion which is not a translation and the center of the such motion coincides with
the center of the circle.
So, we proved the following theorem.

Theorem. а) The direct similarity has a fixed circle if and only if the direct
similarity is a motion which is not translation.
b) The The only fixed circles under a motion which is not translation, are the
circles centered at the center of the motion. ■

6.10. Theorem 4.4 implies that the image of the line (p): z = η z + c under the
direct similarity w = S ( z ) = az + b is the line ( p ') :
(a )
w = a η w + b + ac − a bη .
a
Therefore, the line (p) is fixed under the direct similarity S if and only if
a η = η and b + ac − a bη = c ,
a a
i.e. if and only if a ∈ R and
b + ac − bη = c .
The already stated assertion implies that the line (p): z = η z + c is a fixed line
under the direct similarity if and only if a = 1 and b = bη or a ≠ 1 and b2 = η b2 + c , i.e. if
and only if either S is translation and the line (p) is parallel to the translation vector or S
is a homothety and the line (p) passes through its center.
Thus, we proved the following theorem.

Theorem. A fixed line under a direct similarity exists if and only if the direct
similarity is:
а) translation – the fixed line is each line parallel to the translation vector,
б) homothety – the fixed line is each line which passes through its center. ■

59
6.11. Remark. The proof of Theorem 6.6 implies that two concentric circles
have one and only one center of similarity which coincides
with the center of the circles (figure 3), and non-concentric
circles have either one or two centers of similarity, depending
on either their radii are congruent or not, respectively. In case
when the radii of non-concentric circles are not congruent, the
center of similarity of homothety (1) is said to be an outer
center of similarity, and the center of homothety (2) is said to
Figure 3 be an inner center of similarity (figure 4).

6.12. Remark. By Theorem 6.5, each homothety an arbitrary line maps to a par-
allel line. This statement enables the center of similarity of a homoethety to be construct-
ed in case where we have non-concen-
tric circles ( K1 ) and ( K 2 ) . Through
the center S1 we draw a diameter AB of
the circle ( K1 ) and through the center
S2 we draw a radius S2C parallel to
the diameter AB (figure 4). If R1 ≠ R2 ,
Figure 4 then the lines AC and BC meet the line
S1S2 at points O1 and O2 , which are
an outer and an inner center of the considered homotheties, respectively. If R1 = R2 , then
AC is parallel to S1S2 , and the lines BC and S1S2 intersect at the inner center of similar-
ity.
For the common tangent (t) to the circles ( K1 ) and ( K 2 ) , we get:
1. If R1 = R2 , then (t) is parallel to S1S2 , and
2. If R1 ≠ R2 , then (t) passes through one of the centers of similarity (why?).
So, if R1 ≠ R2 , then in order to construct the common tangents to ( K1 ) and
( K 2 ) , we should firstly determine their centers of similarity and then draw
the tangents to one of the circles.

6.13. Consider the circles ( Ki ), i = 1, 2, 3 given by the following equations

| z − ci | = Ri , i = 1, 2, 3,
respectively, Ri ≠ R j , for i ≠ j , and their centers are not collinear (figure 5). The proof
of Theorem 6.6 implies, that
R1c2 − R2c1 R2c3 − R3c2 R1c3 − R3c1
R1 − R2
, R −R and R1 − R3
2 3
are the affixes of the homothety centers S12 , S23 and S13 , where ( K1 ) is mapped to
( K 2 ) , ( K 2 ) to ( K3 ) , and ( K1 ) to ( K3 ) , respectively. Further,

60
R1c3 − R3c1 R1c2 − R2c1
R1 − R3
− R −R R (R −R )
1 2
R2c3 − R3c2 R1c2 − R2c1
= R1 ( R2 − R3 ) ∈ R .
R2 − R3
− R1 − R2
2 1 3

So, by corollary 1.4, the points

S12 , S23 and S13 are collinear.
Analogously, the following can
be proven:
- the points S12 , S '23 , S '13
are collinear
- the points S '12 , S '23 , S13
are collinear
- the points S '12 , S23 , S '13
are collinear.
Hence, we proved the following
theorem. Figure 5

Theorem. If the centers of the circles ( Ki ), i = 1, 2, 3 , whose radii are not

congruent, are non-collinear, then the centers of homothety S12 , S23 , S13 , S '12 , S '23 , S '13
are on four lines, so that each line consists of exactly three of themy. ■

6.14. Example. Construct a circle which passes through a given point and touches
two different given lines.
Solution. Let (a)
and (b) be given lines and
A be a given point. We will
consider only the case where
the lines (a) and (b) intersect,
and the point A is on neither
one of the lines (a) and (b). A
is also not on the bisector of
the angle formed by the lines
Figure 6
(a) and (b) (as shown in fig-
ure 6). The other cases are left as exercises.
Let S = (a ) ∩ (b) and let K(O,r) be the required circle. Since (a) and (b) are outer
tangents to the circle (K), it is true that the center O of the circle is on the bisector (s) of
the angle formed by the lines (a) and (b), (we consider the angle for which A is an inner
point). If Η is a homothety with center S and arbitrary ratio of homothety, then Η(a ) = a,
Η(b) = b and Η( K ) = K is a circle which touches the lines (a) and (b). So, if we want to
construct the circle (K), we must firstly construct an arbitrary circle ( K ) which touches

61
the lines (a) and (b). Let A1 and A2 be the points where the circle ( K ) meets the line
SA. If Η1 and Η 2 are homotheties with center S and ratio of homotheties OA : OA 1
and OA : OA 2 , respectively, then Η1 ( A1 ) = A and Η 2 ( A2 ) = A . Therefore, the circles
Η1 ( K ) = K ' and Η 2 ( K ) = K '' pass through the point А and
���������������������
touch the lines (a) and (b).
Their centers are Η1 (O) = O ' and Η 2 (O) = O '' , respectively. Due to this, it is necessary
to draw parallel lines to the lines OA1 and ���� OA2 . The points where these parallel lines
intersect the line (s), are actually the points O ' ����
and O '' .
According to the previous considerations we conclude that the given problem has
two solutions.■

7. INDIRECT SIMILARITY

7.1. Definition. The mapping S : C → C defined

����������
by
S ( z ) = az + b, a, b ∈ C, a ≠ 0 (1)
is said to be indirect similarity. In our further discussion the set of indirect similarities
will be denoted by IS.

7.2. Theorem. The indirect similarity S : C → C defined

������������������������ ���������������������������������
as (1) is bijection. Its
−1
inverse mapping S : C → C ��������������
is defined by
S −1 ( z ) = 1 z − b , a, b ∈ C, a ≠ 0 , (2)
a a
and furthermore S −1 ∈ IS .
Proof. If S ( z1 ) = S ( z2 ) , ����
then
az1 + b = az 2 + b ,
therefore z1 = z2 , i.e. S is an injection. If w∈C , then z = w−b satisfies
a
S ( z) = S ( )
w −b
a
= a wa−b +b= w
i.e. S is a surjection. So, S is a bijection.
The mapping S1 ( z ) = 1 z − b is an indirect similarity and furthermore the fol-
a a
lowing holds true S ( S1 ( z ) ) = S1 ( S ( z ) ) = z , i.e. S −1 = S1 ∈ IS . ■

7.3. Theorem. The composition of two indirect similarities is also direct similarity,
and the composition of a direct and an indirect similarity is indirect similarity.
Proof. If S1 , S2 ∈ IS , then the following holds true
S1 ( z ) = az + b, a, b ∈ C, a ≠ 0 and S2 ( z ) = cz + d , c, d ∈ C, c ≠ 0 .

62
Thus,
S1 ( S2 ( z ) ) = S1 (cz + d ) = a (cz + d ) + b = (ac) z + (ad + b),
where ac, ad + b ∈ C, ac ≠ 0 , i.e. S1  S2 ∈DS .
So, the composition of two indirect similarities is a direct similarity.
If S1 ∈DS and S2 ∈ IS , then
S1 ( z ) = az + b, a, b ∈ C, a ≠ 0 and S2 ( z ) = cz + d , c, d ∈ C, c ≠ 0 .
Hence,
S1 ( S2 ( z ) ) = S1 (cz + d ) = a (cz + d ) + b = (ac) z + (ad + b),
where ac, ad + b ∈ C, ac ≠ 0 , i.e. S1  S2 ∈ IS . Analogously, S2  S1 ∈ IS .
So, the composition of a direct and an indirect similarity is indirect similarity. ■

7.4. Both the direct and indirect similarities are commonly said to be similarities.
In our further discussion the set of the similarities will be denoted by S. With direct
checking, the associative law for similarities with respect to composition of mappings can
be proved. The above stated assertion and Theorems 4.2, 7.2 and 7.3 imply the validity
of the following Theorem.

Theorem. The set of the similarities S is non-commutative group under the

composition of mappings. ■

7.5. Theorem. Each indirect similarity is exactly determined by two pairs of

corresponding points.
Proof. Let S be an arbitrary indirect similarity such that S ( z1 ) = w1 and
S ( z2 ) = w2 . So, S ( z ) = az + b, where a, b ∈ C, a ≠ 0 are coefficients that should be
determined. According to Theorem 7.2 each indirect similarity is a bijection, and therefore
z1 ≠ z2 implies w1 ≠ w2 . By substituting in S ( z ) = az + b , we get the following system
of equations
 w1 = az1 + b
 (3)
 w2 = az 2 + b
By solving the system (3) with respect to a and b, we get
w1 − w2 z1w2 − z 2 w1
a= , b= and a ≠ 0 ,
z1 − z 2 z1 − z 2
i.e. the ratios of the indirect similarity are completely determined by two pairs of
corresponding points ( z1 , S ( z1 ) ) and ( z2 , S ( z2 ) ) . ■

7.6. Theorem. The image of a line (p) under indirect similarity is the line ( p ') .
Proof. Let S1 ( z ) = az + b, a, b ∈ C, a ≠ 0 be an indirect similarity and
(p): z = η z + c be a given line. Hence, z = w−b and by substituting into the line equation
a

63
of (p), we get that line (p) is mapped to the line ( p ') with the following line equation
(a ) (
w = a η w + b − acη − ab η . ■
a )
7.7. Theorem. The image of a circle (K) under an indirect similarity is the circle
( K ') .
Proof. Let S1 ( z ) = az + b, a, b ∈ C, a ≠ 0 be an indirect similarity and
(K): | z − c | = R be a given circle. Hence, z = w−b and by substituting into the circle
a
equation of (K), we get that the circle (K) is mapped to the circle ( K ') with the following
circle equation
| w − (b + ac) | = R⋅ | a | . ■

7.8. Theorem. If A, B are arbitrary distinct points and A ', B ' are their images,
respectively, under the indirect similarity (1) and if a = reiϕ , then A ' B ' = r AB .
Furthermore, if α and α1 are the directed angles between the real axis and the lines AB
and A ' B ' , respectively, then α + α1 = ϕ .
Proof. Let z1 , z2 , w1 , w2 be affixes of the points A, B, A ', B ' , respectively. The
following equalities are satisfied,
z2 − z1 = AB ⋅ eiα and w2 − w1 = A ' B ' ⋅ eiα1 ,  
where α and α1 are the angles formed by the real axis and the vectors AB and A ' B ' ,
respectively. Thereby w1 = az1 + b and w2 = az 2 + b , we get the following equation
w2 − w1 = a ( z2 − z1 ) ,
i.e.
A ' B ' ⋅ eiα1 = r AB ⋅ ei (ϕ−α ) ,
which implies that A ' B ' = r AB and α + α1 = ϕ . ■

7.9. Definition. Two forms are indirect similar if there exists an indirect similarity
so that under that similarity one of the forms is mapped to the other one.
The real number r given in the previous theorem is called to be the ratio of the
indirect similarity (1).

7.10. Corollary. If ABC and A ' B ' C ' are indirect similar triangles, then
A ' B ' : A ' C ' = AB : AC and ∠A ' B ' C ' = −∠ABC .
Proof. The proof is directly implicated by Theorem 7.8. ■

7.11. Theorem. Let z1 , z2 , z3 , w1 , w2 , w3 be affixes of the points A, B, C , A ', B ', C

A, B, C , A ', B ', C ', respectively. The triangles ABC and A ' B ' C ' are indirectly similar if and only
if the following holds true
z1 ( w2 − w3 ) + z2 ( w3 − w1 ) + z3 ( w1 − w2 ) = 0 . (4)

64
Proof. The triangles ABC and A ' B ' C ' are indirectly similar if and only if there
exists an indirect similarity like (1) such that the following holds true
wi = az i + b, for i = 1, 2, 3 .
By the last equations we obtain the following
w1 − w2 = a ( z1 − z2 ) and w1 − w3 = a ( z1 − z3 ) .
After dividing and reducing the last two equations we get:
w1 − w2 z1 − z2
w1 − w3
= ,
z1 − z3
which is equivalent to (4). ■

7.12. Example. Let ABCD be a given rectangle and M and N be the midpoints of
the sides AD and BC, respectively. Let P be a point on the extension of DC through D and
Q is the intersection point of the lines PM and AC. Prove that ∠QMN = ∠MNP .
Solution. Let the origin coincides with the point
N and B (B−(x−),x),C C ( x(),x), DD ( x(+
x+iyiy
), ),A(A−(x−+
x+iyiy
) ) , P ( x + iz )
(figure 7). Hence, M(iy) and the equation of the line PM is
z ( x − ip + iy ) − z ( x + ip − iy ) − 2ixy = 0 , i.e.
x +ip −iy 2ixy
z − z x −ip +iy − x −ip +iy = 0 . (*)

( iy )
Let S 2 be the intersection of the lines MN and AC.
Then AC coincides to AS, thus the equation of AC is
( iy
) ( iy
)
z x + 2 − z x − 2 − ixy = 0 , i.e.
2 x −iy 2ixy
z − z 2 x +iy − 2 x +iy = 0 . (**)
Since Q = AC ∩ PM , by utilizing the identities (*) and (**) we Figure 7
find the affix of Q to be the following
xy yp xy yp
q = y + 2 p + i y + 2 p , i.e. q = y + 2 p − i y + 2 p .
Further, if K denotes the projection of the point Q on the x-axis, then its affix is
xy
k = y +2 p . In order to prove the statement, we should only prove that the triangle CPN is
indirectly similar to the triangle KQN. Therefore by Theorem 7.11 it is sufficient to check
the validity of the following equality
xy yp xy
x +ip − x y+2 p
+i y + 2 p − y + 2 p
0− x
= xy
,
0− y + 2 p
which is obviously satisfied. ■

7.13. Definition. An indirect similarity with ratio | a | = 1 is called to be an

indirect isometry. The motions and the indirect isometries are called to be isometries.

65
In our further discussion the set of the direct isometries will be denoted by II,
and the set of the isometries by I. The already stated assertion implies the validity of the
following Theorem.

Theorem. The set I of the isomeries with the composition of mappings is a non-
commutative group. ■

7.14. Theorem. The indirect similarity is involutory if and only if it is a reflec-

tion.
Proof. Let the indirect similarity S : C → C be involutory. Hence, S ( z ) = S −1 ( z ) ,
for each z ∈C , and therefore
az + b = 1 z − b , for each z ∈C .
a a
The latter implies | a | = 1 and b = − a . Thus, the equation z = az + b is a line equation.
b
Let us consider the point z and its image w = S ( z ) . We have w = az + b , and according to
Example 1.9 the points are symmetric with respect to the line z = az + b , i.e. S : C → C
is a reflection.
Conversely, it is enough to apply Example 1.9 directly. ■

7.15. The point z is fixed point of the indirect similarity (1) if and only if
z = az + b . Hence, z = az + b and applying the previous equation we get
z (1− aa ) = ab + b . (5)
There are three possibilities:
1) If aa ≠ 1, then (1) is not an isometry and it has only one fixed point z,
z = ab +b .
1− aa
2) If aa = 1 and ab + b ≠ 0 , then (1) is an indirect isometry, but it is not a
reflection and there are no fixed points.
3) If aa = 1 and ab + b = 0 , then | a | = 1 and b = − a . By the proof of Theorem
b
7.13 we deduce that (1) is a reflection. Furthermore, z = az + b implies that
the point z is on the line of reflection.
The above statement implies the validity of the following Theorem.

Theorem. The indirect similarity which is not an isometry has a unique fixed
point z = ab +b . If the indirect isometry is not a reflection, then there are no fixed points.
1− aa
If the indirect isometry is a reflection, then the only fixed points are points on the line of
reflection. ■

66
7.16. Definition. If the indirect similarity (1) is not an isometry, then the fixed
ab +b
point is called to be a center of the indirect similarity.
1− aa

7.17. Definition. The line (p): z = η z + c is called to be a fixed line under the
indirect similarity (1) if S ( p ) = p , i.e. if the indirect similarity maps the line (p) to itself.
The circle (K): | z − c | = R is called to be a fixed circle under the indirect similarity (1)
if S ( K ) = K .

7.18. Theorem. а) If the indirect similarity (1) is a reflection, then the line of
reflection and its perpendicular lines are the only fixed lines under (1).
b) If the indirect similarity (1) is a reflection, then only the circles with center on
the line of reflection are fixed circles under (1).
Proof. а) Let (1) be a reflection, (p): z = az + b be the line of reflection, | a | = 1 ,
b = − a and (q): z = η z + c be an arbitrary line. Hence w = az + b . We obtain z = w−b .
b a
Now, substituting in the equation of (q) we get
w −b = η wa−b + c
a
i.e.
w= a w+ b +b− c .
ηa ηa ηa
The line (q) is fixed line under the reflection (1) if and only if
a = η and b + b − c = c .
ηa ηa ηa
Since a = η and | a | = 1 , we have η = ±a . Furthermore, if η = a , then
ηa
b +b− c =c
ηa ηa
implies that b = c . If η = −a , then the equation
b +b− c =c
ηa ηa
is satisfied for each c ∈C and (q ) ⊥ ( p ) .
b) Let (1) be a reflection with a line of reflection (p): z = az + b , | a | = 1 , b = − a
b
and let (K): | z − c | = R be an arbitrary circle. Since w = az + b , we obtain z = w −b . By
a
substituting in the equation of the circle (K), we get | w − (b + ac) | = R . The circle (K) is
fixed under the reflection (1) if and only if c = ac + b , or in other words if and only if its
center is on the line of reflection. ■

7.19. Example. Construct a circle which passes through two distinct given points
and touches a given line.

67
Solution. Let the points A, B and the line (c) be given. Obviously, if A and B are in
different semi-planes with respect to the line (c) or both are on the line (c), then the given
problem has no solution. If A ∈ (c) and B ∉ (c) , then the problem has a unique solution.
The center O of the required circle will be the
intersection of the bisector of the segment AB
and the line (p), which passes through A and
is perpendicular to (c). If the points A and B
are in the same semi-plane with respect to
the line (c) and AB is parallel to (c), then the
given problem has two solutions. The one is
the circle which passes through the points A,
Figure 8 B and M, where M is the intersection of the
bisector of the segment AB and the line (c).
The other one is the line AB (figure 8).
Let the points A and B be in the same
semi-plane with respect to the line (c) and the
line AB be not parallel to the line (c) (figure 9).
Since, the required circle K(O,R) passes
through the points A and B, its center will
be on the bisector (l) of the segment AB. Let
σl be a reflection with a line of reflection (l).
By the previous theorem, σl ( K ) = K . Since
Figure 9
the line (c) is tangent to the circle K(O,R),
Corollary 7.10 implies that σl (c) = c ' is
tangent to K(O,R) . If the lines (c) and (c ')
are not parallel, then the problem can be
transformed to Example 6.14.
If the lines (c) and (c ') are parallel,
then the point A is between them. Let d be
the distance between the lines (c) and (c ') .
( )
The circle K A, d2 meets the bisector (s) in
points O1 and O2 . So, the required circles are
Figure 10
( ) ( )
K1 O1 , d2 and K 2 O2 , d2 , (figure 10). ■

7.20. It is naturally to wonder, if there are any fixed lines under an indirect
similarity, which is not a reflection, i.e. | a | ≠ 1 .
According to Theorem 7.6, the image of the line (p): z = η z + c under the indirect
similarity (1) is the line ( p ') with equation

(a ) (
w = a η w + b − acη − ab η .
a )
The lines (p) and ( p ') coincide if and only if
aη=η and b − acη − ab η = c .
a a

68
Reducing the last two equalities, we get
2
η2 = a = a 2 , i.e. η1 = | aa | , η2 = − | aa |
a |a|
and
ba −b| a | ba +b| a |
c1 = , c2 = ,
a (1+| a |) a (1−| a |)
respectively. Thus, the indirect similarity which is not an isometry has two fixed lines
( p1 ) and ( p2 ) such that
ba −b| a | ba +b| a |
( p1 ) : z = | aa | z + and ( p2 ) : z = − | aa | z + .
a (1+| a |) a (1−| a |)
According to Corollary 1.8. А b) the lines and are perpendicular.
Since

( )+
a ab +b
| a | 1− aa
ba −b| a |
a (1+| a |)
= | aa | ba +b2 +
1−| a |
ba −b| a |
a (1+| a |)
| a |(ba +b ) + (ba −b| a |)(1−| a |)
=
a (1−| a |2 )

= ba +baa2 = b +ba2
a (1−| a | ) 1−| a |

we get that the center b +ba2 of the indirect similarity (1), which is not an isometry, is on
1−| a |
the line ( p1 ) . Analogously,

( )
− | aa | ab +b +
1− aa
ba +b| a |
= − | aa | ba +b2 +
a (1−| a |)
ba +b| a |
1−| a | a (1−| a |)
−| a |(ba +b ) + (ba +b| a |)(1+| a |)
=
a (1−| a |2 )

= ba +baa2 = b +ba2
a (1−| a | ) 1−| a |

i.e. the center b +ba of the indirect similarity (1), which is not an isometry, is on the line
1−| a |2
( p2 ) .
Thus, we proved the following theorem.

Theorem. The indirect similarity (1), which is not isometry, has such two
perpendicular fixed lines that pass through the center of similarity. ■

7.21. Definition Let (1) be an indirect similarity, which is not isometry. The lines
( p1 ) and ( p2 ) with equations
ba −b| a | ba +b| a |
z = | aa | z + and z = − | aa | z + ,
a (1+| a |) a (1−| a |)
respectively, are called to be lines of the indirect similarity (1).
Clearly, by Theorem 7.19 the lines of the indirect similarity are the only fixed
lines under indirect similarity which is not an isometry.

69
8. INVERSION

8.1. Let m ∈ R , m > 0 and a ∈C and let I : C \ {a} → C \ {a} be a mapping

defined by
I ( z ) = a + m . (1)
z −a
i) If I ( z1 ) = I ( z2 ) , then
a+ m =a+ m
z1 − a z 2 −a
implies z1 = z2 , i.e. I is an injection.
ii) For w ∈C \ {a}, there exists such z = a + m that I ( z ) = w , i.e. I is a surjec-
w− a
tion.
Now, i) and ii) imply that I is a bijection.

Definition. The mapping

I : C \ {a} → C \ {a}
defined by (1) is called to be an inversion with center a and radius m.

8.2. The point z is fixed point under the inversion (1) if and only if the following
equality is satisfied
z=a+ m ,
z −a
or in other words, if and only if
|z−a| = m .
Thus we proved the following theorem.

Theorem. The point z is fixed point under the inversion (1) if and only if z is on
the circle | z − a | = m . ■

8.3. Definition. The circle ( K 0 ) : | z − a | = m , is called to be the inversion

circle of (1).

8.4. Theorem. The inversion is involutory mapping.

Proof. Let an inversion be defined by (1). Thus, for each z ∈C the following
holds true
(
I ( I ( z)) = I a + m = a +
z −a )m
m
= z = E( z)
a + z − a −a
−1
i.e. I 2 = E and since I is a bijection, then I = I , i.e. the inversion is involutory. ■

8.5. Let O be the center of inversion (1), A be an arbitrary point in the plane,
such that A differs from O, and Ι( A) = A ' . Let a, z and z* be the affixes of the points O,
A and A' , respectively. Thus, it holds true
z * −a = a + m − a = m = m 2 ( z − a) .
z −a z −a | z −a |

70
The latter implies
arg( z * − a ) = arg( z − a ) and | z * − a | ⋅ | z − a | = m .
Thus, we have proved the following theorem.

Theorem. а) Under inversion each point A, distinct from O (the center of the
inversion (1)), is mapped to (inverts to) a point A' which is on the ray OA→ and
OA ⋅ OA ' = m . (2)
b) The points A and A' , with affixes z and z*, respectively, are inverse with respect
to the circle | z − a | = m if and only if
( z * − a )( z − a ) = m . ■

8.6. Theorem. Each inner point of a referent circle inverts to an outer point of the
referent circle, and vice versa.
Proof. If A is an inner point of the referent circle K 0 ( O, m ) and Ι( A) = A ' ,
then OA < m and according to (2) it follows, that OA ' > m . So, A' is an outer point
for the circle K 0 .
The converse statement can be proved analogously. ■

8.7. Let us explain the construction of A' as an inverse of A under the inversion
(1). Let A , with affix z0 , be an inner point of a referent circle K 0 ( O, m ) . Therefore,
| z0 − a |2 < m . The equation of the line OA is
z0 − a
z−a= ( z − a) .
z 0−a
Through the point A we draw a line (q) perpendicular to OA. The equation of that line is
the following
z0 − a
z − z0 = − ( z − z 0) .
z 0−a

The points of intersection of the line (q) and the circle K 0 ( O, m ) are obtained as
solutions of the following system
 z − z = − z0 −a ( z − z )
 0 z 0−a 0

| z − a | = m

One of them is the point T with affix
m −| z0 − a |2
z1 = z0 + i | z0 − a |
( z0 − a ) .
Through the point T we draw a tangent (t) to
the circle K 0 ( O, m ) . The equation of that
tangent is Figure 11

71
z1 − a
z − z1 = − (z − z1 ) .
z 1− a
Further, we determine the point B, as intersection point of the tangent (t) and the line OA.
Its affix is
z=a+ m .
z 0 −a
Thus, Ι( A) = B = A ' .
Let A be an outer point for the circle K 0 ( O, m ) . According to Theorem 8.4 the
following holds true Ι 2 = E . The latter implies the construction Ι( A) = A ' . We draw the
tangent (t) to the circle K 0 through the A. The orthogonal projection of T, the point where
the tangent (t) touches K 0 , on the line OA is the point Ι( A) = A ' .

8.8. Let c and b, be the affixes of the points A and B respectively. The affixes of
their images under the inversion (1) are the following:
c ' = a + m and b ' = a + m ,
c −a b−a
respectively. The complex gradients of the
lines OA, OB, AB, OA' , OB ' and A ' B '
are
η1 = c −a , η2 = b −a , η3 = b −c ,
c −a b−a b −c
η4 = η1 , η5 = η2 ,
(b − a )(c − a )(b −c )
η6 =
(b − a )(c − a )(b −c )
η η η η
respectively. Since η2 = η6 and η1 = η6
3 4 3 5
and also theorem 1.7, we deduce, that
Figure 12 (figure 12):
∠ OBA = ∠ B ' A ' O and
∠ OAB = ∠ A ' B ' O .
Thus, we have proved the following theorem.

Theorem. Let O be the center of the inversion (1), А ����

and B be arbitrary points
and A' and B ' are their images under the inversion (1). Then,
∠ OBA = ∠ B ' A ' O and ∠ OAB = ∠ A ' B ' O . ■

8.9. Under the inversion (1) the line ( p ) with a self-conjugate equation
Az + Bz + C = 0, C ∈ R , B = A
is mapped to the curve with equation
Aa + Ba + C + Am + wBm −a
= 0 . (3)
w− a
Two following cases are possible:

72
i) If Aa + Ba + C = 0 , i.e. the line passes through the inversion center, then (3)
implies that the image of (p) is the line
Aw + Bw + C = 0, C ∈ R , B = A .
The latter is actually the equation of the line (p).
ii) If Aa + Ba + C ≠ 0 , i.e. the line does not pass through the inversion center,
then (3) implies that the image of (p) is the circle
ww + A1w + A1 w + B1 = 0 ,
where
A1 = Bm − a, B1 = aa − Aam+ Bma .
Aa + Ba +C Aa + Ba +C
By direct checking we get aa + A1a + A1 a + B1 = 0 . The latter means that the image of a
line (p) which does not pass through the center of the inversion (1) is circle which passes
through the inversion center.
The above stated assertion implies the validity of the following Theorem.

Theorem. Each line through the inversion center O is mapped to itself, and each
line, which does not pass through O is mapped to a circle through O. ■

8.10. If the line (p) does not pass through the inversion center O, then the proof of
Theorem 8.9 implies that the center O1 of the circle, in which (p) maps to, has affix
z1 = − A1 = a − Bm .
Aa + Ba +C
Let P be the orthogonal projection of the O on the line
(p). According to 2.3 the affix of the point P is
z0 = Aa −2Ba
A
−C ,

Therefore the affix of its image P ' = Ι( P ) is

z* = a − 2 Bm .
Aa + Ba +C
Since,
z*+ a =a− Bm = z1
2 Aa + Ba +C Figure 13
we deduce that O1 is the midpoint of OP ' .
The above stated assertion defines
the construction of the circle Ι(p) if the
line (p) does not pass through the center
O of an inversion Ι. Firstly, we determine
the orthogonal projection P of the inversion
center O on the line (p). Thus we get the point
P ' = Ι( P ) , (figures 13 and 14). Further, we
construct a circle with diameter OP ' . So,
we get the circle Ι(p). Figure 14

73
8.11. Example. Let (p) and (q) be given lines. Is there an inversion Ι such that
Ι( p ) = q ?
Solution. According to Theorem 8.9 such an inversion exists if and only if the
lines (p) and (q) coincide. Further, for each inversion with center on (p) and arbitrary
radius the following holds true Ι( p ) = q . ■

8.12. Theorem. If a circle K1 passes through the center O of the inversion Ι, then
Ι( K1 ) is a line which does not pass through О.
Proof. The
���������������������������������������������
proof is directly implicated by Theorems 8.4
��������
and 8.9.
������
■

8.13. Example. Given are a line (p) and a circle K1 (O1 , R ) . Is there an inversion
Ι such that Ι( p ) = K1 ?
Solution. According to Theorem 8.9 if the required inversion Ι exists, then its
center O is on the circle K1 (O1 , R ) and the line OO1 is perpendicular to the line (p).
The following three cases are possible:
i) The line (p) and the circle K1 meet at points M and N. Starting from the center
O1 of the circle K1 we draw a line (q) perpendicular to (p) and find the points Πand ���� O '
as intersections of (q) and K1 . The discussion in 8.10, implies that the inversions Ι and Ι1
determined by the circles K (O, OM ) and K '(O ', O ' M ) satisfy the given conditions.
ii) The line (p) touches the circle K1 at the point M. Starting from the center
O1 of the circle K1 we draw a line (q) perpendicular to (p). The point O is found as
intersection of (q) and K1 . The discussion in 8.10 implies that the inversion Ι determined
by the circles K (O, OM ) satisfies the given conditions.
iii) The line (p) and the circle K1
have no common points. Starting from the
center O1 of the circle K1 ���������������
�������������� we draw a line
(q) perpendicular to (p). The point P is
determined as intersection of (q) and (p).
The points O and P ' are the intersection
of (q) and K1 such that O1 , P ' , P are
positioned in that order (figure 15). We
construct a semicircle with diameter OP,
through P ' we draw a line perpendicular
to OP and determine the point T as their
intersection. The discussion in 8.10 implies
that the inversion Ι determined by the circle
Figure 15 K (O, OT ) satisfies the given conditions. ■

74
8.14. Now, we have to consider the image of the circle K1 (O1 , R ) under inversion
Ι when K1 does not pass through the inversion center O.
Let K1 : | z − b | = R be a given circle, which does not pass through the center of
the inversion (1), i.e. | a − b | ≠ R . By substituting
z=a+ m
w− a
in the equation of the circle K1 , and after equivalent transformations we get that the
circle K1 inverts to a circle Ι( K1 ) with equation
ww + A1w + A1 w + B1 = 0 ,
where
m (b − a ) m 2 + ma (b − a ) + ma (b − a )
A1 = − a, B1 =| a |2 −
R 2 −| b − a |2 R 2 −| b − a |2 .
With direct checking we get
aa + A1a + A1 a + B1 ≠ 0 .
The above implies the validity of the following Theorem.

Theorem. If a circle K1 does not pass through the center O of the inversion Ι,
then Ι( K1 ) is circle such that it does not pass through O. ■

8.15. If the circle K1 does not pass through the inversion center O, then the proof
of Theorem 8.14 implies that the center O '1 of the circle Ι( K1 ) has affix
m (b − a )
z1 = − A1 = a − .
R 2 −| b − a |2
The affixes z1 , a and b of the points O '1 , O and O1 satisfy
z1 − a m
b−a
= 2 2
∈R .
| b−a | − R
According to Corollary 1.4, the above means that they are collinear. But, the line OO1
is fixed under the inversion Ι. Therefore according to the discussion in 8.5 we conclude
that the diameter of K1 which is on this line inverts to a diameter of Ι( K1 ) which is on
OO1 .
The above stated assertion implies the following construction of the circle
Ι( K1 ) , when K1 does not pass through
the center O of the inversion Ι. We draw a
straight line OO1 , we find the points M, N
(points where K1 meets OO1 ) and further we
obtain M ' = Ι( M ) and N ' = Ι( N ) . After that,
we construct a circle with diameter M ' N ' .
Hence, we get the circle Ι( K1 ) (figure 16).
Figure 16

75
8.16. Example. Given are the circles K1 (O1 , R ) and K 2 (O2 , R ) . Is there an
inversion Ι so that Ι( K1 ) = K 2 ?
Solution. Let the equations of the circles K1 and K 2 be | z − c1 | = R1 and
| z − c2 | = R2 , respectively. We will consider five cases:
i) If c1 = c2 , i.e. the circles are concentric, then the mapping
R1R2
Ι( z ) = c1 +
z −c 1
is inversion, so that Ι( K1 ) = K 2 . Clearly, c1 is the center of such the inversion and R1R2
is the radius. The referent circle can be constructed if we draw an arbitrary half-line and
use the fact that the common points of the circles K1 and K 2 are inverse.
ii) If c1 ≠ c2 , R1 ≠ R2 and | c1 − c2 | ≠ | R1 − R2 | , then the mapping
R1R2 ( R2 − R1 )2 −| c1 − c2 |2
c R −c R ( R2 − R1 )2
Ι( z ) = 1 R2 − R2 1 + c R −c R
2 1 z − 1 R2 − R2 1
2 1
is such an inversion, that Ι( K1 ) = K 2 . We notice that the inversion center coincides with
the outer center of homothety, given by 6.6, which can be constructed as explained in
Remark 6.7. When constructing the inversion circle, it is necessary to follow the procedure
given in Example 8.13.
iii) If c1 ≠ c2 , R1 ≠ R2 and | c1 − c2 | = | R1 − R2 | , then the mapping
R1R2 ( R2 + R1 )2 −| c1 − c2 |2
c R +c R ( R2 + R1 )2
Ι( z ) = 1 R2 + R2 1 + c R +c R
2 1 z − 1 R2 + R2 1
2 1
is such an inversion, that Ι( K1 ) = K 2 . We notice that the inversion center coincides with
the inner center of homothety, given by 6.6, which can be constructed as explained in
Remark 6.7. When constructing the inversion circle, it is necessary to follow the procedure
given in Example 8.13.
iv) If c1 ≠ c2 , R1 = R2 = R and | c1 − c2 | < 2 R , then the mapping
|c1 − c2 |2 |
c1 + c2 R2 −
Ι( z ) = 2
+ c +c
4
z− 1 2 2
is such an inversion, that Ι( K1 ) = K 2 . We notice that the inversion center is the midpoint
of segment O1O2 , and the common points of the circles K1 and K 2 are fixed. So, we
can construct the referent circle.
v) If c1 ≠ c2 , R1 = R2 = R and | c1 − c2 | > 2 R , then Theorem 8.4 implies that the
required inversion does not exist.■

8.17. Consider the following theorem.

Theorem. If two lines, a line and a circle, or two circles have no common points,
they are tangent, or they have two common points, then their images under the inversion

76
Ι have no common points, they are tangent, or they have two common points, respec-
tively. ■

8.18. Definition. Let the line (p) and the circle K meet in the points M and N. We
draw a tangent (t) to the circle K through M. Let α be the smaller angle between (p) and
(t). The angle α is called to be angle of intersection of the circle K and the line (p).
Let the circles K and K* meet in the points M and N. We draw tangents (t1 ) and
(t2 ) to the circles K and K* through M. Let α be the smaller angle between the lines (t1 )
and (t2 ) . The angle α is called to be angle of intersection of the circles K and K*.
We say that the circle K meets the circle K* orthogonally if the measure of the
angle of intersection of K and K* is π2 . K and K* are said to be orthogonal.
Theorem 8.2 proves that the fixed points under the inversion Ι with circle K 0
are the points on K 0 only. This means that the circle is fixed under Ι. Theorem 8.8
implies that there does not exist any line which is fixed under Ι, but each line through the
inversion center is fixed under Ι.
The question now is whether there exist
any other circle (different from K 0 ), which is fixed
under the inversion Ι. By Theorem 8.6, it follows
that if there exists such a circle K1 : | z − b | = R ,
then it is mandatory to meet the reference circle
K 0 . Therefore, it must have two fixed points. One
of them is denoted by T with affix z1 (figure 17).
Now, the proof of Theorem 8.14 implies that Figure 17
K1 : | z − b | = R is fixed under inversion (1) if and
only if
m (b − a )
b=a− ,
R 2 −|b − a|2
z1 − a z1 −b
i.e. if and only if m + R 2 = | b − a |2 . The last equality is equivalent to =− . Due
z1 − a z1 −b
to this, K1 is fixed under inversion (1) if and only if the tangents to K1 and K 0 through
T are perpendicular to each other.
So, we proved the following Theorem.

Theorem. A circle K1 , different from K 0 is fixed under inversion Ι if and only

if K1 intersects K 0 orthogonally. ■

8.19. Theorem. An angle between two lines, a line and a circle or between two
circles is preserved under inversions.
Proof. Let an inversion be given by (1) and let (p) and (q) be two lines with the
following self-conjugate equations:
Az + Bz + C = 0, B = A, C ∈ R and A1 z + B1 z + C1 = 0, B1 = A1 , C1 ∈ R
respectively. The following two cases are possible:

77
i) Let the lines (p) and (q) pass through the inversion center. It follows by the
proof of Theorem 8.9, that (p) and (q) are fixed, and therefore the angle between them is
preserved.
ii) Let one of the lines, for example (p), passe through the inversion center and
let the other one (q) does not pass through the inversion center. The proof of Theorem 8.9
implies that the line (p) is fixed, and the line (q) inverts to the circle
B1m m| A1|
Ι(q ) : z − a + = .
A1a + B1 a +C1 | A1a + B1 a +C1|
Let the line (p) and the circle Ι(q) meet at the inversion center with affix a. The
equation of the tangent (q1 ) to the circle Ι(q) through a is the following
A1 z + B1 z + C1 − aA1 − aB1 = 0 .
We check directly that the complex gradients η1 , η2 , η3 of the lines (p), (q), (q1 ) satisfy
η1 η
η
= η1 . Further, the statement given in this Theorem is implied by Theorem 1.7.
2 3
iii) The lines (p) and (q) do not pass through the center of inversion. The proof
of Theorem 8.9 implies that the inverses of the lines (p) and (q) are circles with the
following equations
m| A| B1m m| A1|
z−a+ Bm = and z − a + = ,
Aa + Ba +C | Aa + Ba +C| A1a + B1 a +C1 | A1a + B1 a +C1|
respectively. The circles Ι(p) and Ι(q) meet in the inversion center with affix е a. The
equations of the tangents ( p1 ) and (q1 ) to the circles Ι(p) and Ι(q) through a are the
following
Az + Bz + C − aA − aB = 0 and A1 z + B1 z + C1 − aA1 − aB1 = 0 ,
respectively. The complex gradients η1 , η2 , η3 , η4 of the lines (p), (q), ( p1 ) , (q1 )
η η
satisfy the following equality η1 = η3 . Due to this, Theorem 1.7 implies the validity of
2 4
the given Theorem.
The remaining part of Theorem is proved analogously, using Theorems 8.9 8.12
and 8.14. ■

8.20. Example. Let the line (p) and the circle

K(O,R) have no common points. Prove that there exists
an inversion Ι such that Ι(p) and Ι(K) are two concentric
circles.
Solution. Through the center O of the circle K we
draw a line (q) perpendicular to (p) and let P = ( p ) ∩ (q ) ,
(figure18). Let T be an arbitrary point on the circle K, so
that T is not on (q), and S be one of the common points of
the circle K1 ( P, PT ) and the line (q).
Figure 18

78
We define an inversion Ι with center S and arbitrary radius. According to Theorems
8.9 and 8.14, the inverses (images) Ι(p) and Ι(K) are circles. We have to determine their
centers.
The line (p) meets orthogonally the line (q) and the circle K1 . Due to this,
Theorem 8.19 implies that Ι(p) meets orthogonally Ι(q) and Ι( K1 ) . Theorem 8.9 implies
that Ι(q ) = q , and Theorem 8.12 implies that Ι( K1 ) is a line. Since, the lines (q) and
Ι( K1 ) meet the circle Ι(p) orthogonally, we get that its center O '' is a common point of
these two lines. Hence, O '' = (q ) ∩ Ι( K1 ) .
The circle K meets orthogonally the line (q) and the circle K1 . Due to this,
Theorem 8.19 implies that Ι(K) meets Ι(q) and Ι( K1 ) orthogonally. This means that Ι(K)
meets (q) and Ι( K1 ) orthogonally. Hence, the center O ' of the circle Ι(K) is obtained as
O ' = (q ) ∩ Ι( K1 ) .
Finally, O ' ≡ O '' , i.e. Ι(p) and Ι(K) are two concentric circles.■

8.21. We solve the problem below by applying homothety (Example 7.18). In

this section we will give the solution by applying inversion.

Example. Construct a circle which passes through two given points and touches
a given line.
Solution. Let points A and B and a
line c be given. We suppose that the points
A and B lie in the same semi-plane with
respect to the line c and that the line AB is
not parallel to the line c.
Let Ι be an inversion with center A
2
and radius AB . Hence, Ι( B ) = B and Ι(c)
is the circle K1 which passes through A.
The required circle K passes through the
point А, ����������
therefore Ι(K) is a line which passes
through B and touches the circle K1 , i.e.
Ι(K) is the tangent to the circle K1 through Figure 19
the point B. The already stated assertion
implies the following construction (figure 19):
2
- We define an inversion Ι with center A and radius AB .
- We construct the circle K1 = Ι(c) .
- Through B we draw the tangents (t ') and (t '') to the circle .
- The required circles are K ' = Ι(t ') and K '' = Ι(t '') . ■

79
9. MÖBIUS TRANSFORMATION

9.1�������������
. Definition. The mapping
az +b , ad − bc ≠ 0 ,
S ( z ) = cz (1)
+d
where a, b, c, d are complex numbers, is called to be a Möbius transformation.
The Möbius transformation is determined for each z ≠ − dc , ∞ . If c = 0 , then
the Möbius transformation is determined for each finite z. If c ≠ 0 , then we extend the
definition by
( )
S − dc = ∞ and S (∞) = ac . (2)
Thus, S : C∞ → C∞ is a well defined
If c = 0 , then it is sufficient to let S(∞) = ∞ . ������
mapping.
Note that, the condition ad − bc ≠ 0 is equivalent to S(z) is an injection (show
it!).

9.2. Example. Determine the image of the unit circle | z | = 1 under the
transformation
w = u z −v , u , v ∈ C and | v | ≠ 1.
vz −1
Solution. Since
ww = uu z z −v z − zv +vv
vvz z + zv − zv +1
for z z = 1 we get ww = uu . That means, that the unit circle is mapped to the circle
| w| = |u |. ■

9.3. Theorem. The Möbius transformation defined by (1) and (2) is a bijection
from C∞ to C∞ .
Proof. Let S be a Möbius transformation defined by (1) and (2).
If w∈C and w ≠ ac , then since w = cz
az +b we get z = dw−b . So, when w ≠ a , ∞
+d − cw+ a c
there exists z = −dw −b such that S ( z ) = w .
cw+ a
( )
If w = ∞ , then S − dc = ∞ , and if w = ac , then S (∞) = ac . Due to that, S is
surjection. But, we already noted that S is injection, and therefore S is bijection.
The above stated implies that the mapping S −1 : C∞ → C∞ defined by
S −1 ( z ) = dz −b , if z ≠ a and by S −1 (∞) = − d , S
− cz + a c
−1 a
c c
=∞( )
is well defined, and furthermore
S ( S −1 ( z ) ) = S −1 ( S ( z ) ) = z holds.
Finally, S −1 is the inverse mapping to S and since
�������������������
ad − (−b)(−c) = ad − bc ≠ 0
we get that it is Möbius transformation. ■

80
9.4. Theorem. The family M of all Möbius transformations under the composi-
tion of mapping is a group.
Proof. Let
az +b , ad − bc ≠ 0 and S ( z ) = ez + f
S1 ( z ) = cz +d 2 gz + h
, eh − fg ≠ 0
be two Möbius transformations. Then,

( )
ez + f
ez + f a +b ( ae + gb ) z + ( af +bh )
S ( z ) = ( S1  S2 )( z ) = S1 ( S2 ( z ) ) = S1 gz + h = ezgz++fh = (ce+ gd ) z +(cf + dh)
c gz + h + d
and
(ae + gb)(cf + dh) − (af + bh)(ce + gd ) = (ad − vc)(eh − fg ) ≠ 0 .
So, the mapping S = S1  S2 is Möbius transformation, i.e. (M , ) is a groupoid.
а) Associativity. For all S1 , S2 , S3 ∈ M we have
S1  ( S2  S3 ) = ( S1  S2 )  S3 . (3)
Namely, both sides of (3) are equal to the Möbius transformation S1 ( S2 ( S3 ( z ) ) ) .
Therefore, (M , ) is a semi-group.
b) Existence of identity. Obviously, the identity mapping E ( z ) = z is Möbius
transformation, that E ( z ) = z is an identity in the semi-group (M , ) .
c) Thereby
�����������������������������������������������������
Theorem 9.3, it is true that each element of (M , ) �����������������
has its inverse.
So, statements a), b) and c) imply that (M , ) is group. ■

9.5. Remark. The group (M , ) is a non-abelian.

Indeed, for
S1 ( z ) = z + a, a ≠ 0 and S2 ( z ) = 1z
the following holds true
S1 ( S2 ( z ) ) = 1z + a and S2 ( S1 ( z ) ) = z +1 a , i.e. S1  S2 ≠ S2  S1 .

9.6. Theorem. If S is Möbius transformation, then S is a composition of elemen-

tary transformations in a complex plane.
Proof. Let
az +b , ad − bc ≠ 0
S ( z ) = cz +d
be an arbitrary Möbius transformation.
If c = 0 , then
S ( z ) = da z + db .
Therefore, if
S1 ( z ) = da z and S2 ( z ) = z + db
we get S = S2  S1 , i.e. S is a composition of elementary transformations.
If c ≠ 0 , then for
S1 ( z ) = z + dc , S2 ( z ) = 1z , S3 ( z ) = bc −cad z , S4 ( z ) = z + ac
we get S = S4  S3  S2  S1 . Thus, also in this case S is composition of elementary trans-
formations. ■

81
9.7. Example. Let w = ii +− zz be a given transformation. Determine the image of:
а) the real axis, b) the circle | z | = 1
under the given transformation.
Solution. The given transformation can be rewritten as w = −1 + i 2+iz .
а) The equation of the real axis is z − z = 0 , and under the transformation
w1 = z + i , it is mapped to the line w1 − w1 = 2i . Further, the line w1 − w1 = 2i is mapped
to the circle w2 + 2i = 12 under w2 = w1 , and under the transformation w3 = 2iw2 the
1
circle w2 + 2i = 12 is mapped to the circle | w3 − 1 | = 1 . Finally, under the transformation
w = −1 + w3 the circle | w3 − 1 | = 1 is mapped to the circle | w | = 1. The above means that
the given transformation maps the real axis to the circle | w | = 1.
b) The translation w1 = z + i maps the circle | z | = 1 to the circle | w1 − i | = 1 .
Further, the transformation w2 = w1 maps the circle | w1 − i | = 1 to the line
1
w2 − w2 = −i , and the transformation w3 = 2iw2 maps the line w2 − w2 = −i to the line
w3 + w3 = 2 . Finally, the transformation w = −1 + w3 maps the line w3 + w3 = 2 to the
line w + w = 0 . The above means that the given Möbius transformation maps the circle
| z | = 1 to the line w + w = 0 . ■

9.8. Consider the Möbius transformation (1). The mapping S1 ( z ) = − dc + 1 d

z+
c
is an inversion with respect to the circle z + dc = 1, and the mapping S2 ( z ) = pz + q ,
where
p = bc −2ad , q = bc −2ad d + ac
c c c
is an indirect similarity. Further, since
S2 ( S1 ( z ) ) = S2 (− dc + 1 d ) = bc −2ad ⋅  − dc + 1 d  + bc −2ad d + ac
z+ c  z+  c c
c  c 

= bc −2ad ⋅  − d + 1 d  + bc −2ad d + ac
c  c z+ c  c c

= − bc −2ad d + bc −2ad d + bc −cad cz1+ d + ac

c c c c
bc − ad + a (cz + d ) az +b
= c (cz + d )
= cz + d ,
it follows the validity of the following Theorem.

Theorem. The Möbius transformation (1) can be expressed as composition

S = S2  S1 of the inversion S1 ( z ) = − dc + 1 d with respect to the circle z + dc = 1 and
z+
c
the indirect similarity S2 ( z ) = pz + q , where
p = bc −2ad , q = bc −2ad d + ac . ■
c c c

82
10. GEOMETRIC PROPERTIES OF A
MÖBIUS TRANSFORMATION

10.1. Theorem��. Arbitrary Möbius transformation maps each circle of C∞ to a

circle of C∞ .
Proof. According to Theorem 9.6, each Möbius transformation is a composition of
elementary transformations in the complex plane and moreover the elementary transforma-
tion S2 ( z ) = 1z given in the proof of Theorem 9.6 is composition of the inversion I ( z ) = 1
z
and the reflection S ( z ) = z . Then, the validity of the Theorem����������������������������
is implied
������������������������
directly by the
previously proved properties of the elementary transformations in the complex plane. ■

10.2. Applying the properties of elementary transformations in a complex plane,

analogously to the proof of Theorem 8.19, can be proved the following very important
Theorem.

Theorem. Arbitrary Möbius transformation preserves the angle between circles

in the extended complex plane C∞ . ■

10.3. Definition. Consider the circle K(O,R). The points M and M* are said to be
symmetric with respect to the circle K, if Ι( M ) = M * , where Ι is an inversion determined
by the circle K.

10.4. Before discussing the properties of the symmetric points with respect to
a circle, and related to the Möbius transformation we will give the following Lemma,
which characterizes the symmetric points M and M* ���������������������������
with respect to the circle K(O,R).

Lemma. The points M and M* are symmetric with respect to the circle K(O,R) if
and only if each circle γ through these points meets orthogonally the circle K(O,R).
Proof. Let the points M and M* be symmetric with respect to the circle K(O,R)
(figure 20).
Consider the circle γ through the
points M and M* and K ∩ γ = {P} . According
to Theorem 8.14, the circle γ under inversion Ι,
determined by the circle K, is mapped to a circle
γ1 , which passes through the points M, M* and
P. Therefore, the circles γ and γ1 coincide, i.e.
the circle γ is fixed under the inversion Ι. Now,
the validity of the given Theorem is directly
implicated by Theorem 8.18.
Conversely, let each circle γ, which
passes through the points M and M*, meets Figure 20

83
orthogonally the circle K (figure 20). The Theorem 8.18 implies that the circle γ is fixed
under the inversion Ι, determined by the circle K. Then the line (in the extended complex
plane viewed as a circle), which passes through the points M and M*, also meets the
circle K orthogonally i.e. the line passes through the center O of the circle K. But, γ
is fixed under the inversion Ι, and therefore Ι( M ) = M * , i.e. the points M and M* are
symmetric with respect to the circle K. ■

10.5. Theorem. A Möbius transformation maps a pair of symmetric points with

respect to a circle, to a pair of symmetric points with respect to the image of that circle.
Proof. Let the points z and z* be symmetric with respect to the circle K and let
w = S ( z ) be an arbitrary Möbius transformation. According to Theorem 10.1 the image
K ~ of the circle K is a circle. We have to prove that the points w and w* are symmetric
with respect to K ~ . According to Lemma 10.4 it is sufficient to prove that each circle
γ ~ , which passes through the points w and w*, crosses K ~ at right angle.
The inverse image of the circle γ ~ under the Möbius transformation w = S ( z )
is a circle which passes through the points z and z*. This circle crosses the circle K at
Therefore, γ ~ crosses K ~ at right angle, because Theorem 10.2 states that
right angle�������������
. �����������
the Möbius transformation preserves the angle between intersecting circles at any point
in the extended complex plane. ■

10.6. Further, we will prove one important Theorem about Möbius transforma-
tion.

Theorem. A Möbius transformation S : C∞ → C∞ maps the unit circle | z | ≤ 1

to the unit circle | w | ≤ 1 if and only if
w = S ( z ) = eiϕ z −v , v ∈ C, 0 ≤ ϕ < 2π, | v | < 1 . (1)
1−vz
Proof. Let z ∈ C, | z | ≤ 1 . Likewise, the following holds true:
| z −v |2
| w |2 = ww = eiϕ z −v e −iϕ z −v = 2
≤ 1,
1−vz 1− zv |1−vz |
i.e. the Möbius transformation (1) maps the unit circle | z | ≤ 1 to the unit circle | w | ≤ 1.
Conversely, let S : C∞ → C∞ be Möbius transformation such that it maps the
unit circle | z | ≤ 1 to the unit circle | w | ≤ 1 and let us suppose that there exists such a
point v, v ≠ 0, | v | < 1 that is mapped to the point w = 0 . The point symmetric to null
with respect to the circle | w | = 1 is the infinity distanced point. According to Theorem
10.5 it follows that w = ∞ when z = 1 , and therefore the required Möbius transformation
v
is the following

84
w = k z −v1 ,
z−
v

where k is a constant. The latter can be transformed and rewritten as it follows

w = − kv z −v = k ' z −v . (2)
1−vz 1−vz
Since z = 1z when | z | = 1 we get
| 1 − vz | = 1z − v = | z − v | = | z − v | .
But, the circle | z | = 1 is mapped to the circle | w | = 1, and therefore | k ' | = 1, i.e.
k ' = eiϕ , for some 0 ≤ ϕ < 2π , i.e. the formula (2) transforms to (1). Clearly, likewise the
formula (2) holds true when v = 0 . ■

10.7. Definition. A point z is said to be a fixed point under the Möbius transfor-
mation
az +b if z = S ( z ) , i.e. z = az +b .
S ( z ) = cz +d cz + d
az +b
Clearly, z is a fixed point under the Möbius transformation if S ( z ) = cz +d
cz 2 + (d − a ) z − b = 0 . (3)
If c ≠ 0 , then the fixed points are the following:
a − d ± ( a − d )2 + 4bc
z1 / 2 = 2c
. (4)
If c = 0 , then the fixed points are z1 = b and z = ∞ . Further, if b = c = 0 and a = d ,
d −a 2
then the Möbius transformation is the identity mapping S ( z ) = z and therefore each point
of C∞ is fixed point.
By (4), if (a − d ) 2 + 4bc = 0 , then z1 = z2 . The last means that we have a re-
peated or a double fixed point i.e. the two fixed points coincide. When c = 0 , the condition
for repeated points implies that a = d , and in this case we get that z = ∞ is a double fixed
point for the translation S ( z ) = z + db .

10.8. Comment. When defining the Möbius transformation

az +b ,
S ( z ) = cz ,
+ d ad − bc ≠ 0
four complex numbers a, b, c and dare used. But, one of c or d differs from 0, and therefore
if we divide both, the numerator and the denominator, by this number, we get that the
Möbius transformation can be expressed using three coefficients. Therefore, it is naturally
to expect that the images of three given points determine a unique Möbius transformation.
The following Theorem confirms our assumtion..

85
10.9. Theorem. There is a unique Möbius transformation S, such that the points
z1 , z2 , z3 ( zi ≠ z j , i ≠ j )
under such transformation, are mapped to the points w1 , w2 , w3 ( wi ≠ w j , i ≠ j ) , respec-
tively.
Proof. Existence. The mappings S1 and S2 defined by
( z − z )( z − z ) ( w− w )( w − w )
S1 ( z ) = ( z − z1 )( z3 − z2 ) and S1 ( w) = ( w− w1 )( w3 − w2 ) , (5)
2 3 1 2 3 1

map the points z1 , z2 , z3 in the plane z and the points w1 , w2 , w3 in the plane w, to the
points 0, ∞, 1 in the plane ς , respectively. Finally, the mapping
S = S2−1  S1 (6)
which is determinedfrom the plane z to the plane w, S ( z ) = w , as
z − z1 z3 − z2 w− w w −w
⋅
z − z2 z3 − z1
= w− w1 ⋅ w3 − w2 (7)
2 3 1

is Möbius transformation, which maps the points z1 , z2 , z3 to the points w1 , w2 , w3 ,

respectively.
Uniqueness. Let λ, λ ( zi ) = wi , i = 1, 2, 3 be an arbitrary Möbius transformation.
Consider the mapping µ = S2  λ  S1−1 , where S1 and S2 are mappings defined by (5).
Clearly, µ is Möbius transformation and the points 0, ∞, 1 are fixed points under this
transformation. Since the condition µ(∞) = ∞ it follows that µ(ς) = ας + β . The condi-
tion µ(0) = 0 implies that β = 0 , and the condition µ(1) = 1 implies that α = 1. Therefore,
µ(ς) = ς , i.e.
S2  λ  S1−1 = E .
Since (M , ) is a group, we get λ = S2−1  S1 , i.e. λ = S . ■

10.10. Remark. In the equality (7), each of the points zi and wi appears
exactly twice, once in the numerator and once in the denominator. It is easy to prove
that the equality holds true when one of the points zi or wi (either one zi and one wi )
��������
is the infinity point. Then, it is necessary the numerator and denominator, where this point
appears, to be replaced by 1. For example, if z2 = w1 = ∞ , then the formula (7) can be
transformed and rewritten as
z − z1 w3 − w2
1
⋅ z 1− z = w−1w ⋅ 1
.
3 1 2

Therefore, Theorem 10.9 holds true for any point in the extended complex plane C∞ .

10.1�������������
1. Corollary. It exists a unique Möbius transformation S, such that it maps
the points z1 , z2 , z3 , z4 ( zi ≠ z j , i ≠ j ) to the points w1 , w2 , w3 , w4 ( wi ≠ w j , i ≠ j ) if
and only if

86
z4 − z1 z3 − z2 w −w w −w
⋅
z4 − z2 z3 − z1
= w4 − w1 ⋅ w3 − w2 . (7’)
4 2 3 1

Proof. Let the equality (7’) be satisfied and let S be such the Möbius transformation
that maps the points z1 , z2 , z3 ( zi ≠ z j , i ≠ j ) to the points w1 , w2 , w3 ( wi ≠ w j , i ≠ j ) ,
respectively. Then the Möbius transformation is defined by (7), and therefore
z4 − z1 z3 − z2 S ( z )− w w −w
⋅
z4 − z2 z3 − z1
= S ( z 4 )− w1 ⋅ w3 − w2 .
4 2 3 1
Thus,
S ( z4 ) − w1 w3 − w2 w −w w −w
⋅
S ( z4 ) − w2 w3 − w1
= w4 − w1 ⋅ w3 − w2 ,
4 2 3 1
i.e. S ( z4 ) = w4 .
If there exists a Möbius transformation with the stated properties, then that
transformation maps the points z1 , z2 , z3 ( zi ≠ z j , i ≠ j ) to the points w1 , w2 , w3
( wi ≠ w j , i ≠ j ) , and therefore the Möbius transformation is as (7)������
, and S ( z4 ) = w4 . So,
(7’) holds true. ■

10.12. Remark. Theorems 10.9 and 10.1 imply that each circle K in C∞ can be
mapped to a circle K* in C∞ . It is sufficient to map any three distinct points of K to any
three distinct points of K*.

10.13. Example. Determine the Möbius transformation such that it maps the
points −1, i, 1 + i to the points
а) 0, 2i, 1 − i ; б) i, ∞, 1 ,
respectively.
Solution. According to Theorem 10.9 and Remark 10.10 the required Möbius
transformations are
−2i ( z +1) (1+ 2i ) z + 6−3i
а) w = −4 z −1−5i , б) w = 5 ( z −i )
.■

10.14. Remark. The proof of Theorem 10.9 implies that each Möbius transfor-
mation S have up to two fixed points z1 , z2 , �����������������
i.e. such points z1 , z2 , ��������
so that S ( z1 ) = z1 ,
S ( z2 ) = z2 , when S ≠ E . The Möbius transformation with two fixed points z1 , z2 is
determined by
w− z1 z−z
w− z 2
= A z − z1 , z1 , z2 ≠ ∞ (8)
2
or
w − z1 = A1 ( z − z1 ), z2 = ∞ . (9)
The coefficients A and A1 are given by

87
z −z w −z w −z
A = z3 − z2 ⋅ w3− z1 , A = w3− z1 , (10)
3 1 3 2 3 2
az +b
and they do not depend on the choice of z3 when w3 = cz 3+ d . It is easy to express the
3
coefficients A and A1 in terms of a, b, c and d, if (10) holds true and we set w3 = db
when z3 = 0 .

10.15. At the end of this section we will provide a proof of the following Theo-
rem.

Theorem. The set of Möbius transformations F such that the unit circle | z | ≤ 1
they mapped to the unit circle | w | ≤ 1 is a subgroup of the group of Möbius transforma-
tions.
Proof. Let S1 , S2 ∈F . Then, Theorem 10.6 implies that
S1 ( z ) = eiα z −a , | a | < 1 and S2 ( z ) = eiβ z −b , | b | < 1 ,
1− az 1−bz
therefore,
iβ
iβ z − eiβ b + a
S1 ( S2 ( z ) ) = ei (α −β) e + ab e + ab
,
eiβ + ab iβ
1− eiβ b + a  z

 e + ab 
where
eiβb + a iβ iβ
= b + aeiβ < 1 and ei (α −β) e + ab = 1.
eiβ + ab 1+ ae b eiβ + ab

Applying Theorem 10.6 once again, we get that S1  S2 ∈F . So, the set F is closed under
the composition of mappings.
Let S1 , S2 , S3 ∈F . Then S1 , S2 , S3 ∈ M , therefore
S1  ( S2  S3 ) = ( S1  S2 )  S3
i.e. the associative law holds true.
Letting v = 0 and ϕ = 0 in Theorem 10.6 we get
F ∋ E ( z ) = z = ei⋅0 z −0 .
1− z⋅0
Let S ∈F . Then, S ( z ) = eiϕ z −v , | v |< 1 . Consider the transformation
1−vz
iϕ
i ( − ϕ) z −( − ve )
S1 ( z ) = e , | −veiϕ | < 1 .
1−( − veiϕ )z
Clearly, S1 ∈F and the following holds true
S ( S1 ( z ) ) = S1 ( S ( z ) ) , i.e. S −1 = S1 ∈ F . ■

88
CHAPTER III
GEOMETRY OF CIRCLE AND TRIANGLE

1. CENTRAL AND INSCRIBED ANGLES

1.1. Definition. An angle whose vertex coincides with the center of a given circle
K is called a central angle.

1.2. Theorem. If two central angles in a same circle are congruent, then their
corresponding arcs are congruent, too.
Proof. Without loss of generality, we consider
K as a unit circle. Let ∠ AOB = ∠ COD (figure 1) and
the affixes of points A, B, C and D be a, b, c and d,
respectively. Therefore
b = d , i.e. c = d .
a c a b
Since
S (a ) = ac a = c and S (b) = ac b = db b = d
we get that under the mapping S ( z ) = ac z , which in
fact is a rotation around ∠ AOC , the point A maps to a Figure 1
point C, and B to D. Therefore, the arc AB maps to the arc CD. ■

1.3. Similarly the reverse theorem of theorem 1.2 can be proved. The proof is left
as an exercise.

Theorem. If two arcs in a same circle are congruent, then their corresponding
central angles are congruent, too. ■

1.4. Definition. An angle whose vertex is on a given circle K, and its rays meet
the circle is called an inscribed angle.

1.5. Theorem. The size of an inscribed angle is

half of the size of its corresponding central angle.
Proof. Without loss of generality, we consider K
as a unit circle. Let consider the arc  AB where points A
iϕ
and B has affixes 1 and e , ϕ ∈ (0, π) , respectively. Then,
∠ AOB = ϕ . Let M be a point of the complementary arc
of the circle (figure 2), i.e.it has affix eiθ , θ ∈ (ϕ, 2π) .
Figure 2

89
Then,
ϕ ϕ ϕ
i i i ( θ− )
iϕ iθ e 2 ( e 2 −e 2 )
∠ AMB = arg e −ieθ = arg
1−e iθ
−2ie 2 sin θ2
ϕ ϕ ϕ
i i ( θ − ) −i ( θ − )
e 2 (e 2 2 −e 2 2 )
= arg
2i sin θ2
ϕ ϕ ϕ
i 2 sin( θ2 − 2 ) i2 ϕ
= arg e = arg e =2
sin θ2
∠ AOB
i.e. ∠ AMB = 2
.■

1.6. Corollary. All inscribed angles with congruent arcs in a given circle K, are
�����������������������������������������������������������
congruent.
Proof. The proof is directly implied by Theorem 1.5. ■

1.7. Corollary (Thales’ Theorem). Each inscribed angle on a diameter of a

circle is a right angle.
Proof. The proof is directly implied by Theorem 1.5. ■

1.8. Remark. In paragraph II 8.7 during the construction of the image of an

arbitrary point under inversion, we have drawn a tangent to a circle at point which is
out of that circle. The effective construction of the tangent is implied by the Thales’
Theorem.
Let a circle K(O,R) and a point M which
is outer for the circle K be given. The tangent at
M to K is constructed with the following proce-
dure (figure 3).
а) we construct the midpoint O1 of the
line segment OM,
b) we construct a circle K1 (O1 , MO1 ) ,
c) we find the intersecting points
Figure 3 K ∩ K1 = {T , T1} and
d) we draw lines MT and MT1 .

1.9. The above discussed line segments MT and MT1 , are called tangent segments
to the circle K(O,R) at point M. The tangent segments satisfy the following Theorem.

Theorem. Let M be a point out of the circle K(O,R). If MT and MT1 are tangent
segments drawn at point M to the circle K, then MT = MT 1 .

90
Proof. Without loss of generality we consider K as a unit circle (why?). Let the
affixes of points T and T1 be t and t1 , respectively. The lines MT and MT1 are tangents,
and since Remark II 3.13, their equations are
z + t 2 z = 2t and z + t12 z = 2t1
2tt
and affix of M is m = t +t1 . Finally,
1
2tt t ( t −t ) t1 (t −t1 ) 2tt
MT = | m − t |= t +t1 − t = t +1 t = t +t1
= t +t1 − t1 = | m − t1 | = MT 1 . ■
1 1 1

1.10. Theorem. The angle between a chord AB and the tangent to the circle (t)
��������������������������
drawn at one of the points A or B is congruent to the inscribed angle on the chord AB.
Proof. Without loss of generality we can consider the unit circle K(O,R). Let
affixes of points A and B be a and b, respectively. The equation of the tangent (t) drawn
at the point B is as following
z = −b 2 z + 2b ,
And the equation of a line AB is
z = − abz + a + b .
Therefore, the angle β between line А� B and the tangent (t)
2iβ b
satisfies the following e = a . Since the proof of Theorem
II 1.4, the inscribed angle α satisfies that
e 2iα = ba = e 2iβ , Figure 4
So, α = β . ■

2. POWER OF A POINT WITH RESPECT TO A CIRCLE

2.1. Let K(O,R) be a given circle and M be an arbitrary point of a plane. Through
M we draw an arbitrary line (p) which meets the circle K at points A and B. We will prove
that the product MA ⋅ MB does not depend on the choice of the line (p).
If the point M is on a circle K, then M is one of the points А or
��� B. Therefore one
of MA or MB is null. So, MA ⋅ MB = 0 .
Let point M be out of the circle K (figure
5) and (t) be one of the tangents to K drawn
at M, and T be the tangent point. The circle
K1 ( M , MT ) orthogonally crosses the circle K,
therefore K is fixed under the inversion defined
by the circle K1 . Therefore, from the definition
of inversion we have that
Figure 5

91
2
MA ⋅ MB = MD ⋅ MC = ( MO − R )( MO + R ) = MO − R 2 .
In this case, since the properties of inversion (Theorem 8.5), it is true that
2
MA ⋅ MB = MT .
Let the point M be an inner point for the circle K and let (p) be an arbitrary line
which passes through M. Without loss of generality we assume that the center of the
circle coincides with the origin. Then, if m is an affix of M, the equation of line (p) is
the following z − m = η( z − m) , and the equation of the circle is the following z z = R 2 .
From the equation of (p) we express z , and when we substitute in the equation of the
circle we obtain a quadratic equation as following
η z 2 + ( m − η m) z − R 2 = 0
whose solutions
− m + η m ± ( m −η m ) 2 + 4η R 2
z1 / 2 =
2η
are the affixes of the points A and B, points of intersection between the line and the circle.
Due to this,
2
MA ⋅ MB = | m − z1 | ⋅ | m − z2 | = | mm − R 2 | = R 2 − MO .
The arbitrariness of (p) implies that the product MA ⋅ MB does not depend on the choice
of the line (p) through the point M.
The already stated implies that the product MA ⋅ MB does not depend on the
choice of the line (p) through M. It depends only on the length of the radius R of a circle
K and on a distance d = MO between the point M and the center O of the circle K. The
value (the real number) d 2 − R 2 is called as power of the point M with respect to the
circle K. Clearly, if M is a point on K, then the power is 0, furthermore if M is point
outside the circle K, then the power is a positive real number, and if M is within the circle
K, then the power is a negative real number.

2.2. Definition. Let K1 (O1 , R1 ) and K 2 (O2 , R2 ) be given circles and let Η be
homothety with center S so that Η( K1 ) = K 2 . If (p) is line such that it passes through S and
meets the circles K1 and K 2 at points P1 , P2 and Q1 , Q2 respectively and if Η( P1 ) = Q1 ,
Η( P2 ) = Q2 , then the points P1 and Q2 ( P2 and Q1 ) are said to be antihomothetic
(figure 6).
If the circle K touches the circles
K1 and K 2 , both internally or externally,
then we shall say K touches K1 and K 2
in a same way, and if it touches one
internally and the other one externally ,
then we shall say that K touches K1 and
K 2 in a different way.
Figure 6

92
2.3. Lemma. The product of the distances between the center of homothety of
two circles and two antihomothetic points is a constant value.
Proof. Let S be external center of a homothety Η of the circles K1 (O1 , R1 )
and K 2 (O2 , R2 ) , ( R1 ≠ R2 ) and let points P1 and Q2 ( P2 and Q1 ) be antihomothetic
(figure 6). Then, the homothety ratio is
SQ1 SQ 2
a= =
SP1 SP 2
So,
SQ 2
SP1 ⋅ SQ 2 = SP1 ⋅ SP 2 ⋅ = SP1 ⋅ SP 2 ⋅ a = const
SP 2
Thus, SP1 ⋅ SP 2 is a power of the point S with respect to the circle K1 . ■

2.4. Lemma. Let K(O,R) be such a circle that touches the circles K1 (O1 , R1 ) and
K 2 (O2 , R2 ) . Then,
а) if K touches K1 and K 2 in a same way, then the points where K touches K1
and K 2 are antihomothetic with respect to the external center of similarity S for circles
K1 and K 2 ,
b) if K touches K1 and K 2 in a different way, then the points where K touches
K1 and K 2 are antihomothetic with respect to the internal center of similarity S for
circles K1 and K 2 ,
Proof. b) Let a circle K touches the circles K1 and K 2 at points P1 and Q2 in
a different way, respectively (figure 7) and let the affixes of O, O1 , O2 , P1 and Q2 be
c, c1 , c2 , p and q, respectively. Then the affix of internal center of homothety is
R c +R c
s = 1R2 + R 2 1 .
1 2
The points O, O1 , P1 are collinear and it holds true that O1P1 = R1 , therefore the following
equalities are satisfied
c1 −c
p − c1 = ( p − c1 )
c1 −c
and
( p − c1 )( p − c1 ) = R12
By reducing we get
c1 −c 2
( p − c1 ) 2 = R ,
c1 −c 1
i.e.
(c1 −c )2 Figure 7
( p − c1 ) 2 = R12 .
|c1 −c|2
Thus, the affix p of the point P1 is as following
c −c c −c
p = c1 − |c1 −c| R1 = c1 − R1+ R R1 .
1 1
Analogously, the affix q of point Q2 is as following

93
c −c c −c
q = c2 + |c2 −c| R2 = c2 + R2− R R2 .
2 2
Finally,
s− p R ( R−R )
s −q
= − R1 ( R + R2 ) ∈ R ,
2 1
i.e. the points S , P1 and Q2 are collinear. Due to this, the points P1 and Q2 are antiho-
mothetic.
The statement а) can be proved analogously. The details are left to be proven as
an exercise. ■

2.5. Lemma. Each center of homothety of circles K1 (O1 , R1 ) and K 2 (O2 , R2 )

has equal power with respect to each circle which touches the circles K1 and K 2 .
Proof. Since Lemma 2.4, if points where K touches the circles K1 and K 2 are P1
and Q2 respectively, then P1 and Q2 are antihomothetic points (figure 23 and figure 24).
If S is homothety center, then due to Lemma 2.3 the product SP1 ⋅ SQ 2 is a constant
value, and that is actually the power of a point S with respect to the circle K. ■

2.6. Example. Construct a circle such that it runs through the points A and B and
touches the line (c).
Solution. We will consider the most general case when the points A and B are on
a same semi-plane with respect to the line (c) and furthermore, the lines (c) and AB are
not parallel to each other (figure 8).

Figure 8

The power of the point M = AB ∩ (c) with respect to the arbitrary circle which
crosses through the points A and B is MA ⋅ MB . If K O, AB
2
2 ( )
and if (t) is a tangent to (K)
drawn at the point M, then MT is a power of M with respect to (K), and therefore it is
actually the power of M with respect to the required circle. That means, the points P1 and
P2 where the line (c) meets the circle ( M , MT ) in fact are points where line (c) meets
the circles which passes through the points A and B. If ( p1 ) and ( p2 ) are perpendicular

94
to (c) and P1 and P2 respectively, are foots of perpendicular and if (s) is a bisector of
the line segment AB, then ( p1 ) ∩ ( s ) = O ' and ( p2 ) ∩ ( s ) = O '' are centers of required
circles. Clearly, the given problem has two solutions. ■

3. RADICAL AXIS AND RADICAL CENTER

3.1. Let two circles K '(O ', R ') and K ''(O '', R '') be given. Let’s determine the
����������������
locus Γ such that it has an equal power with respect to the circles ( K ') and ( K '') .
Let affixes of the centers O ' and O '' be c1 and c2 , respectively. Clearly, M ∈ Γ
if and only if
2 2
MO ' − R '2 = MO '' − R ''2 ,
i.e. if and only if
| z − c1 |2 − R '2 =| z − c2 |2 − R ''2 .
The last equality is equivalent to
c2 −c1 |c2 |2 −|c1|2 + R '2 − R ''2
z=− z+ .
c 2 −c1 c 2 −c1
Finally, the required locus is a line perpen-
dicular to O ' O '' (figure 9). This line shall be
called the radical axis of K1 and K 2 . Figure 9

3.2. Lemma. A ��������������������������������������������������������������������

radical axis of two circles with no common points and also a part
of a radical axis of two crossing circles (outer for the intersecting circles) is a locus of the
centres of circles which orthogonally crosses the given circles.
Proof. ���
If K(O,R) orthogonally crosses the circles K '(O ', R ') ���� and K ''(O '', R '') ,
then the tangents at O to ( K ') and ���� ( K '') are congruent, i.e. O has
�������������������� ������������������������
an equal power with
respect to ( K ') ����
and ( K '') . �������
Hence, O ��������������������������
is on the radical axis of ( K ') ����
and ( K '') , figure 10.

a) b)
Figure 10

Let K(O,R) be a circle centred at a point on the radical axis of the circles ( K ')
and ( K '') ����
and

95
R = T ' O = T '' O .
Therefore, OT ' ���� are tangents to ( K ') ����
and OT '' ���������������� and ( K '') , ���������������������������
respectively, the circles (K)
and ( K ') , ����������
and also (K) and ( K '') ����������������������������
cross at right angle, i.e. (K) orthogonally crosses ( K ')
and ( K '') . ■

3.3. Definition. The circle K(O,R) halves the circle K '(O ', R ') , ����
����������� if (K) intersects
( K ') at
����������������������������������
two diametric opposite points.
Let us state that if the circle K(O,R) crosses
the circle K '(O ', R ') at
���������������������������������
two diametric opposite points
A and B (figure 11), ����������������������������
then the power of the point O ' with
�����
respect to the circle K(O,R) implies that
2
O ' C ⋅ O ' D = O1 A
Therefore, when writing d = O ' O �����������
we get that
����
2
i.e. R = R ' + d 2 .
( R − d )( R + d ) = R ' , ����� 2 2
Figure 11
3.4. Lemma. The inner part of a radical axis of intersecting circles K '(O ', R ')
and K ''(O '', R '') is a locus of the centers of circles (K), such that both ( K ') and ( K '')
halve (K).
Proof. Clearly, if M is a center of
circle which is halved by two given circles
( K ') and ( K '') , then M must be an inner
point for circles ( K ') and ( K '') (figure
12), therefore, such a circle exists only if
( K ') and ( K '') intersect each other. The
comment after definition 3.3 implies that
R 2 = R '2 − d '2 and R 2 = R ''2 − d ''2 ,
Figure 12 thus
R '2 − d '2 = R ''2 − d ''2 ,
i.e. M has the equal power with respect to the both circles ( K ') and ( K '') , i.е. M is placed
on the inner part of the radical axis of ( K ') and ( K '') . ■

3.5. Let be given three circles Ki (Oi , Ri ) , i = 1, 2, 3 . We will determine the locus
of points in a plane which has an equal power with respect to the three given circles. Let
p12 , p23 , p13 be the radical axis of ( K1 ) and ( K 2 ) , ( K 2 ) and ( K3 ) , ( K3 ) and ( K1 ) ,
respectively. Hence, if there is any point P with an equal power with respect to the circles
( K1 ) , ( K 2 ) and ( K3 ) , then that point must be on a radical axis ( p12 ) and ( p23 ) . Two
cases are possible.
а) If centers Oi , i = 1, 2, 3 of the circles are not collinear (figure 13), then the
radical axis ( p12 ) and ( p23 ) intersect each other. The point P = ( p12 ) ∩ ( p23 ) has an
equal power with respect to the circles ( Ki ) , i = 1, 2, 3 . Therefore the radical axis ( p13 )

96
Figure 13 Figure 14
which passes through the point P, a point with equal power with respect to the circles
( Ki ) , i = 1, 2, 3 . This point is to be called a radical center of ( Ki ) , i = 1, 2, 3 .
b) If the centers Oi , i = 1, 2, 3 , of the circles are collinear, then the racial axes are
parallel, and furthermore each of them is either different or coincide. When the first case
is satisfied, there is no any point with required property, when the second case is satisfied
the required locus is a line.

3.6. The already stated implies

the effective construction of radical axis
of two circles. Namely, if circles ( K1 )
and ( K 2 ) intersect at A and B, then the
radical axis is a straight line AB (figure
14), and if circles touch at T, then the
radical axis is a common tangent to Figure 15
( K1 ) and ( K 2 ) at T (figure 15).
If circles ( K1 ) and ( K 2 ) do not
intersect, then we construct an arbitrary
circle ( K3 ) such that it intersects both
( K1 ) and ( K 2 ) . The intersecting point
of radical axes ( p13 ) and ( p23 ) is a
radical center P of the circles ( Ki ) ,
i = 1, 2, 3 , and therefore the radical axis
is a line which passes through P and is
perpendicular to O1O2 (figure 16). Figure 16

3.7. Remark. Let us notice that we can discuss about a radical axis of a point and
a circle, and likewise about a radical axis of two points. Clearly, the radical axis of two
points A and B is a bisector of the line segment AB. When talking about the radical axis of
point A and circle ( K1 ) , if A ∈ ( K1 ) , then it is a tangent at the point A, and if A is outside
the circle, then the radical axis can be constructed by using the radical center of A, ( K1 )

97
and a circle ( K3 ) such that it passes
through A and intersects ( K1 ) (figure
17). Namely, the radical axis of A and
( K3 ) in fact is a tangent (t) to ( K3 ) at A,
and the radical axis ( p13 ) passes through
the points of intersection M and N. So,
the radical center P is an intersection of
(t) and ( p13 ) , and therefore the required
radical axis of A and ( K1 ) is a straight
Figure 17
line (n) at P, perpendicular to AO1 .
Analogously, we can discuss about the radical center of a point and two circles,
of two points and a circle and also about radical center of three points. Clearly, the radical
center of three non-collinear points A, B and C is a center of circle which passes through
A, B and C.

3.8. Lemma. If Ki (Oi , Ri ) , i = 1, 2, 3 are

����������������������������������������������
three not concurrent circles with centers
which are not collinear, then it exists a unique circle (K) such that either (K) intersects all
three circles or all three circles halve the circle (K).
Proof. Let
���� P be a radical center of Ki (Oi , Ri ) , i = 1, 2, 3 . Then, P is either inside
or outside of each three circles. When the first case is satisfied the power of the circle
is −m 2 and the circle K(P,m) halves the circles Ki (Oi , Ri ) , i = 1, 2, 3 , and when the
second case is satisfied the power of the circle is m 2 and the circle K(P,m) orthogonally
intersects the circles Ki (Oi , Ri ) , i = 1, 2, 3 . ■

3.9. Example. Construct a circle which passes through the points A and B and
touches ( K1 ) .
Solution. Through the points A and B, we draw an arbitrary circle (K) such that
(K) meets the circle ( K1 ) at C and D (figure 18).
Then, the radical axis of (K) and ( K1 ) is a straight
line CD, and the radical axis of (K) and the required
circle (K*) is a line А� B. Thus, the intersection P of
lines AB and CD, if such point exists, is radical
center of (K), (K*) and ( K1 ) , which implies that
the radical axis of ( K1 ) and (K*) is a tangent to
( K1 ) at P.
Hence, drawing the tangents (t1 ) and (t2 )
at P to ( K1 ) , we find the points of touching T1 and
T2 between the required circles and ( K1 ) . So, the
given problem has at most two solutions. ■

Figure 18

98
4. A PENCIL AND A ������������������
BUNDLE �����������
OF CIRCLES

4.1. ������������
Let circles K1 (O1 , R1 ) and K 2 (O2 , R2 ) ������������������
be given and let (p) be their radical
axis. The set of circles such that (p) is radical axis for each two of them ((p) is considered
as a circle) is said to be a pencil of circles. The straight line (p) is said to be a radical axis
of the pencil.
Since 3.1 the centres of each circle of the pencil are on the line O1O2 which is
and is said to be a central line of a pencil.
perpendicular to (p)���������������������
����������������

4.2. Clearly, each pencil is defined by two circles, but our further discussion will
imply that pencil of circles is likewise fully defined by a radical axis and a circle. Let
M be an arbitrary point of a plane. If M is on a radical axis, then it exists a circle which
passes through M. Let M not be �������������������������������������������������������������
on a radical axis. Likewise in 3.5 we will consider three
cases:
а) If circles ( K1 ) and ( K 2 ) intersect at A and B, then
������������� the unique circle ( K 0 )
�����������������������
passes through the points M, A and B. It is easy to conclude that the circle ( K 0 ) belongs
to a pencil of circles determined by the circles ( K1 ) and ( K 2 ) .
b) If circles ( K1 ) and ( K 2 ) ��������� then the circle K 0 (O, OA) (O is
touch at A, ����������������
intersection of bisector of line segment AM and the central line of the pencil), belongs to
a pencil of circles determined by ( K1 ) and ( K 2 ) .
c) Let ( K1 ) and ( K 2 ) have no intersection, m be an affix of M and T with
���������������������� affix t
�����������
be an arbitrary point on a radical axis. Then, it exists a unique point M1 ����������
with affix
|t −o1|2 − R12
m1 = t + (m − t ) .
|m −t|2
Since,
|t −o1|2 − R12
TM ⋅ TM 1= | m − t | ⋅ t + (m − t ) − t
|m −t|2

= | t − o1 |2 − R12 = TO 12 − R12 ,
the circle K 0 (O, OM ) (������������������������������������������������
is intersection of the bisector of line segment MM1 and the
central line of a pencil), belongs to a pencil of circles determined by ( K1 ) and ( K 2 ) .
Thus, we proved the following theorem.

Theorem. Through each point of a plane passes exactly one and only one circle
of a given pencil of circles. ■

4.3. Let
���������������
the pencil Рbe������������������������
defined by a radical
axis (p) and a circle ( K 0 ) . Since
���������������������������
the proof of Theorem
relationship between �(p) and
4.2, depending on the ����������������������
( K 0 ) , ������������������������������������������
there are three types of pencils. Namely,
а) If (p) and ( K 0 ) �������������
intersect at A and B (fig-
each circle K of
ure ����������������������
19), then
����������������� ��� Рintersects
�����������������������
the radical
axis at A ����
and B �����
the Poncelet points of the pencil������
(���� ) and Figure 19

99
vice verse, each circle which passes through A and B be- ���
longs to a pencil Π. Thus,
�����������������������������������������
the pencil consists of each circle
which passes through A and B. So, �������������������������
we will say that the
pencil Π has
��������������������������������������������������
two basic points or in other words is said to
be a�� hyperbolic pencil .
����(p) and ( K 0 ) touch
b) If at A (figure 20), ����������
��������� then each
Figure 20 circle (K) of Π touches the radical axis at A and vice verse,
����������������������������
each circle such that it touches (p) at A belongs
�����������
to Π. Thus,
������
the pencil of circles consists of each circle which touches
(p) at A. So, we will say that the pencil Рhas ��������������
one basic
point or in other words is said to be a�� parabolic pencil .
If (p) and ( K 0 ) ����������������������������
c) ���� have not any sharing points
(figure �������������������������������
21), ��������������������������
then each other circle of ���������������
has no sharing
Figure 21 points with (p) and furthermore, each two circles of Рdo ���
not have any sharing points. So, we will say that the pencil �����������������������������
has no any basic point or in
��� elliptic pencil.
other words is said to be an

4.4. Lemma. �������������������

The set of circles Π1 (���������������������������������������������
each of them is orthogonal to the circles of
a pencil Π) �����������������������������
is a pencil of circles�������
, too.
Proof. ����Let Ki (Oi , Ri ) , i = 1, 2 �������
be two
arbitrary circles of the pencil Рwith �������������
radical
axis (p) and central line (�q), �������� and let
K 'i (O 'i , R 'i ) , i = 1, 2 ���������������������
be two arbitrary cir-
cles of the set � Π1 (figure 22). ��������������
Since the cir-
cles K 'i (O 'i , R 'i ) , i = 1, 2 are
������������������
orthogonal to
Ki (Oi , Ri ) , i = 1, 2 ��������������
their centres O 'i , i = 1, 2
are on a radical axis (p) of K 'i (O 'i , R 'i ) ,
i = 1, 2 . ������������������������������
On the other hand, the points
Figure 22 Oi , i = 1, 2 are
����������������������������������
with equal power with respect
to K 'i (O 'i , R 'i ) , i = 1, 2 , ������
i.e. (�q) �������������
is a radical
axis of K 'i (O 'i , R 'i ) , i = 1, 2 . ����������������������������������������������������������
Now, the statement is implied by the arbitrariness of the
circles K 'i (O 'i , R 'i ) , i = 1, 2 . ■

4.5. Definition. If
����������������������������������������
each circle of the pencil of circles Π is
�����������������������������
orthogonal to the pencil �
Π1 , ����������������� and Π1 shall
then the pencils ���� be called conjugate pencil of circles.
����������������

4.6. Lemma. а) If �����������������������������������������������������������������

one of two conjugate pencils of circles is elliptic, then the
other one is hyperbolic, and vice versa.
b) If one of two conjugate pencils of circles is parabolic, then the other one is
parabolic, too.

100
Proof. а) Let Π be ���������������
an elliptic
pencil with radical (p) and central line
(q) (figure 23). The intersection P of
�����������������
(p) and �(q) is����������������������������
outer for each circle of
the pencil Π, and
�����������������������������
therefore it is a centre
of ( K ') , which belongs to a conjugate
Since ( K ') ������������������
pencil Π1 . ������ meets the radical
axis (�q) ��������������
of the pencil Π1 ���at A ����
and B, we Figure 23
get that the pencil Π1 ���������������
is hyperbolic.
Conversely, let Π1 be hyperbolic pencil with radical axis �(q) ������������������
and central line (p)
(figure ������������������������������
23). ������������������������� centred at P = ( p ) ∩ (q ) �����������������������
The circle of the pencil Π1 ����������� meets the radical axis
at A and B. If K1 (O1 , R1 ) ��������������������������
(q) ��� and T a point of intersection
be an arbitrary circle of Π, ���� �������������
between ( K1 ) and���� ( K ') , then
����� O1T < O1P , and therefore ( K1 ) has no any intersection
��������������
point with the radical axis (p) ��� of Π. ��������������������������
So, the pencil of circles �������������
is elliptic.
b) The statement is directly implied by the Definition of parabolic pencil of
circles. The details are left as an exercise. ■

4.7. Example. ��������������������������������� with a radical axis �(p) ������

Let be given a pencil of circles ��������������������� and a
circle K1 (O1 , R1 ) . �������������������
Construct a circle K(O,R) such that it belongs to the pencil ����and
touches the given circle K 2 (O2 , R2 ) .
Solution. Let
���� ( p12 ) be a radical axis of the circles ( K1 ) and ( K 2 ) . Since
��������������������������������� ����������
the
required circle (K) belongs to a pencil Π, the radical axis of ( K1 ) and (K) will
�������������������� �����������������
be the line
(p). So, the radical centre of (K),� ( K1 ) and ( K 2 ) ���is P = ( p ) ∩ ( p12 ) , ������������������
and therefore the
radical axis of (K) and ( K 2 ) ����������������
is a tangent to ( K 2 ) ����������������������������
such that it passes through P. ������������
After that,
we construct the tangents to ( K 2 ) through
�������� P, if
���������������������������������������������
such tangent exist, and further construct
the required circle (K) centred at the perpendicular of the tangent drawn at the point of
touching to ( K 2 ) . Then we find the central line of the pencil determined by ( K1 ) ���� and
(p). ■

4.8. Example. Given a pencil Π with a radical axis (p) and a circle K1 (O1 , R1 ) .
Construct a circle K(O,R) such that it belongs to the pencil Π and touches the given line
(a) which differs from (p).
Solution. If (a) is parallel to (p), then the exersize given problem may be
considered as construction of a circle that belongs to the pencil Π and passes through the
point (a ) ∩ (q ) , where (q) is the central line of the tensile Π.
Therefore, let’s assume that the line (a) intersects the line (p) and let M be the
point of intersection. The tangent distance from M to ( K1 ) is equivalent to the tangent
distance from M to the required circle (K), thus the circle (K) tangents the line (a) at the
point P so that MP = MT , where T is the point of tangent of the tangent drawn from M to
( K1 ) . Now the center of (K) is in the intersection of the perpendicular to (a) at the point
P and the central line (q) of the pencil Π. ■

101
4.9. Definition. The set of all the circles, such that each three of them have a
common radical center P, and all the lines such that pass through the point P, is called
a bundle of circles. The point P is called the radical center of the bundle and the degree
of the point P with respect to an arbitrary circle of that bundle is called the degree of the
bundle.

4.10. Every bundle of circles Γ is determined by

- center and degree,
- a center and a circle
- a degree and two circles or
- three circles.
Depending on the mutual position of the center with
respect to the circles of the bundle Γ, we can distinguish
between three types of bundles such that:
а) If the degree is m 2 > 0 , then the center P is an
external point to every circle of the bundle Γ and according
Figure 24 to the lemma 13.8 the circle K(P,m) intersects orthogonally
every circle of Γ. According to this, Γ consists of every
circle and every line which orthogonally intersect the circle
K(P,m) (figure 24).
b) If the degree is m 2 = 0 , then the bundle Γ
consists of every circle and every line that pass through the
centre P.
c) If the degree is m 2 < 0 , then the centre P is an
internal point for every circle of the bundle Γ and according
to lemma 13.8 every circles of Γ half the circle K(P,m).
Therefore, Γ consists of every circle and every line which
Figure 25 intersect the circle K(P,m) in diametrically opposite points
(figure 25).

4.11. Remark. Sometimes a bundle of circles is regarded as a set of all ������������

circles
whose centers are on a given line (p) and every line perpendicular to (p), i.e. the set of
every circle and every line which orthogonally intersect the line (p) and this bundle is a
bundle of the first type, because as we said before the line in an extended complex plane
might be regarded as a circle.

4.12. Lemma. The intersection of two bundles of circles is a pencil of circles or

pencil of lines.
Proof. Let Γ1 and Γ 2 be two bundles of circles with centers P and Q and degrees
m and n, respectively. If P ≡/ Q , then Γ1 ∩ Γ 2 is a pencil of circles with a radical axis PQ,
and if P ≡ Q ≡ O , then it is a pencil of lines with a center at O. We will consider three
different cases where P ≡/ Q .

102
а) If m = n = 0 , then Γ1 and Γ 2 are the sets of every circle which passes through
P and Q, respectively, thus Γ1 ∩ Γ 2 is the set of every circle which passes through the
points P and Q, i.e it is a hyperbolic pencil of circles
b) If m > 0 and n > 0 , then Γ1 ∩ Γ 2 is the set of all circles which orthogonally
intersect the circles K ( P, m ) and K * ( Q, n ) . According to the lemma 4.4 Γ1 ∩ Γ 2
is a pencil of circles.
c) If m = 0 and n > 0 , then Γ1 ∩ Γ 2 is the set of all the circles which pass
through the point P and orthogonally intersect the circle K * ( Q, n ) , and that is a pencil
of circles with at least one base point, i.e. it is a hyperbolic or parabolic pencil of circles
depending on whether P ∉ K * ( Q, n ) or P ∈ K * ( Q, n ) .
The other possible cases are left as an exercise for the reader. ■

5. ORTHOCENTAR AND CENTROID OF A TRIANGLE

5.1. Let��������������
consider the  ABC, whose vertices A, B and C have affixes a, b and c,
respectively. In the example II 3.3 we proved that
aa (c −b ) +bb ( a −c ) + cc (b − a )
o=
a a 1
b b 1
c c 1

is the affix of O circumcenter of  ABC. Clearly, the radius of the circum circle of  ABC
is R = | a − o | . The mapping S : C → C determined by S ( z ) = R1 ( z − o) is a direct
similarity which maps  ABC into  A ' B ' C ' . According to consequence 4.8 we get that
A ' B ' : A ' C ' = AB : AC and ∠ A ' B ' C ' = ∠ ABC .
Let a ', b ', c ' be the affixes of the vertices A ', B ', C ' , respectively, and let t be one of
the three square roots of the complex number a ' b ' c ' . The mapping S1 : C → C deter-
mined by S1 ( z ) = tz is a movement which maps  A ' B ' C ' into  A '' B '' C '' . Moreover, if
a '', b '', c '' are the affixes of A '', B '', C '' , respectively, then
a '' b '' c '' = t 3a ' b ' c ' = 1
and according to the consequence 4.8 we get that
A '' B '' : A '' C '' = A ' B ' : A ' C ' and ∠ A ' B ' C ' = ∠ A '' B '' C '' .
The above stated implies that when consider the triangle with no loss of generality
we take that its vertexes A, B and C with the affixes a, b and c, respectively, are on
the unit center circle centered at the origin. Therefore | a | = | b | = | c | = 1 . Moreover, the
coordinate system is chosen such that abc = 1.

103
In our further consideration, unless it is not mentioned differently, we will
consider that  ABC is inscribed in the unit circle centered at the origin O and that for the
affixes a, b and c and the vertices A, B and C is true that abc = 1.

5.2. Let’s consider the  ABC whose vertices A, B and C have the affixes a, b
and c, respectively. According to the theorem 1.2 the equations of the lines AB, BC and
CA are such that
z + abz = a + b , z + bcz = b + c , z + caz = c + a , (1)
respectively, i.e. their complex angle coefficients are − ab, − bc and − ca , respectively.
According to consequence 1.8 the lines that go through C, A and B, and are perpendicular
to AB, BC and CA are expressed as following
cz − z = c 2 − ab , az − z = a 2 − bc , bz − z = b 2 − ca , (2)
respectively. The lines whose equations are given in (2) are to be called altitudes of
 ABC, drawn from the vertices C, A and B, respectively.
Similarly, the equations of the bisectors of the sides AB, BC and CA are the
following
z − abz = 0 , z − bcz = 0 , z − caz = 0 , (3)
respectively.
Therefore a ≠ b we get that the system of equations
az − z = a 2 − bc
 2
bz − z = b − ca
has a solution h = a + b + c . With a direct check we can prove that the complex number h
satisfies the equation of the altitude drawn from the vertex C as well.
Thus, we proved the following theorem.

Theorem. The altitudes at  ABC concur at the point H with affix

h=a+b+c. ■

5.3. Definition. The point N discussed in the theorem5.2 is called the orthocen­ter
of  ABC.

5.4. Remark. а) For  ABC whose vertexes A, B and C have the affixes a, b
and c, respectively, and which is not inscribed in the unit circle centered at the origin, it
can be proven that for the affixes h and o of the orthocenter H and the circumcenter O,
respectively, the following is true
h + 2o = a + b + c .
b) Let’s consider the at  OXY such that one of its vertexes coincides with the
origin and the other X and Y have the affixes x and y, respectively. Then, for the affix o of
O1 circum center of the  OXY we get that

104
00( y − x ) + x x (0− y ) + y y ( x −0) y yx − x xy xy ( y − x )
o= = =
0 0 1 x y − xy x y− xy
x x 1
y y 1
.
Moreover, for the affix h of the orthocenter H of  OXY we get that
xy ( y − x ) x 2 y − x xy + y yx − y 2 x − 2 y yx + 2 x xy
h = 0 + x + y − 2o = x + y − 2 =
x y − xy x y− xy
x 2 y − y 2 x − y yx + x xy x y ( x − y )+ y x ( x − y ) ( x − y )( x y + y x )
= = = .
x y − xy x y − xy x y − xy

5.5. Example. Draw a line perpendicular to the

diameter of the circle K(O,R) from a point M which is not
on that circle.
Solution. We draw lines from the point M to
the ends of the diameter A and B. Then, the lines AM
and BM meet the circle K(O,R) at the points C and D,
respectively. In accordance with consequence 1.7 the
lines AD and BC are altitudes of the triangle whose
sides are on the lines AC and BD. So the theorem 5.2
implies that the line MP is the required perpendicular
drawn at the point M to the diameter AB (figure 26). ■ Figure 26

5.5. Example. If H and O are the orthocenter and the circum center of the  ABC,
respectively, then
2 2 2 2
OH = 9 R 2 − ( AB + AC + BC ) ,
where R is the length of the circumradius. Prove this!
Solution. Without loss of generality, we can assume that  ABC is inscribed in a
circle centered at the origin and a radius R. If the vertices A, B and C have the affixes a,
b and c, then | a | = | b | = | c | = R and the remark 15.4 implies that:
2
OH = | a + b + c |2 = (a + b + c)(a + b + c)
= aa + bb + cc + ab + ab + bc + bc + ca + ca
(
= 3(a a + bb + cc) − | a − b |2 + | b − c |2 + | c − a |2 )
2 2 2
= 9 R 2 − ( AB + AC + BC ),
Which was exactly supposed to be proven. ■

5.6. Theorem. If H is the orthocenter of  ABC and are the points symmetric
to H with respect to the lines BC, CA, AB, respectively, then the points A4 , B4 , C4 lie on
the circle circumscribed around  ABC.

105
Proof. The example II 1.9 and the theorem 5.2 it follows that the affixes of the
points A4 , B4 , C4 symmetric to the point H with respect to the lines BC, CA, AB are
− b 2 c 2 , − c 2 a 2 , − a 2b 2 , respectively. From
| a | = | b | = | c | =1
it follows that
| −b 2 c 2 | = | −c 2 a 2 | = | − a 2b 2 | = 1 ,
i.e. the points A4 , B4 , C4 are on the circumcircle of the  ABC. ■

5.7. Consequence. The projections A2 , B2 , C2 of the vertexes A, B, C on the

2 2 2 2 2 2
sides BC, CA, AB of  ABC have the affixes h −b2 c , h −c2 a , h −a2 b , respectively.
Proof. According to the theorem 5.6 the points A2 , B2 , C2 are midpoints of the
line segments HA4 , HB4 , HC4 , respectively. Now, the proof by the fact that the affix
of the orthocenter H is h, and the affixes of the points A4 , B4 , C4 are −b 2 c 2 , − c 2 a 2 ,
− a 2b 2 respectively. ■

5.8. Let  ABC be given, and let its vertices A, B, C have the affixes a, b, c,
respectively. The affixes of the midpoints A1 , B1 , C1 of the sides BC, CA, AB are
b + c , c + a , a +b , respectively, which means that the equations of the lines AA , BB , CC
2 2 2 1 1 1
are
z − a = b +c −2a ( z − a ), z − b = c + a −2b ( z − b), z − c = a +b −2c ( z − c), (4)
b+c −2a c + a − 2b a +b − 2c
The system of equations
 z − a = b+c −2a ( z − a)
 b+c −2a

 z − b = + a − 2b ( z − b )
c
 c + a − 2b
has the solution
t = a +3b +c .
By a direct check we prove that the complex number t satisfies the equality of the line
CC1 as well. The point whose affix is the complex number t is denoted by T. Further-
more, we get that
AT = a +3b +c − a = b +c3−2a = 2 b +c6−2a |= 2 b +2 c − a +3b +c = 2 A1T .
Analogously we prove that
BT = 2 B1T and CT = 2C1T .
Thus we proved the following theorem.

Theorem. If A1 , B1 , C1 are the midpoints of the sides BC, CA, AB of the  ABC,
then the lines AA1 , BB1 , CC1 concur at a point T whose affix is t = a +3b +c and the point
T divides the line segments AA1 , BB1 , CC1 in a ratio of 2: 1. ■

106
5.9. Definition. The point T defined as in theorem 5.8 is called a centroid of
 ABC, and the lines AA1 , BB1 , CC1 are its medians.

5.10. Example А. Let be given a quadrangle ABCD and let Ta , Tb , Tc , Td be the

centroids of the triangles BCD, ACD, BAD, ABC, respectively. Prove that the line seg-
ments ATa , BTb , CTc , DTd intersect in one point, and each of them is divided in a ratio
of 3 :1 starting at the vertices of the quadrangle.
Solution. Due to the theorem 15.8 we get that
ta = b +c3+ d , tb = a +c3+ d , tc = a +b3+ d and td = a +3b +c .
Let A ', B ', C ', D ' be the points which divide the line segments ATa , BTb , CTc , DTd in
a ratio of 3:1 starting from the vertices of the quadrangle, respectively. I 4.2. implies that
a ' = b ' = c ' = d ' = a +b +4c + d , which means that ATa , BTb , CTc , DTd concur at a point T
with an affix t = a +b +4c + d , and each of them is divided by in a ratio of 3:1 starting from
the vertices of the quadrangle. ■

Comment. The point T discussed in the previous example is called centroid

ABCD. The example А shows how we can define the centroid of a pentagon. Namely,
we consider the line segments which connect a vertex of a pentagon with a centroid of a
quadrlateral formed by the other four vertices of the pentagon and thus we get five line
segments which intersect at T, which is to be called a vertex of a pentagon. It is easy
to prove that if the affixes of the vertices of the pentagon ABCDE are a ', b ', c ', d ', e ' ,
respectively, then the affix t of its centroid T is t = a +b +5c + d +e . On a similar way we can
define the centroid T of n-gon A1 A2 ... An and we can prove that its affix is
a + a +...+ an
t = 1 2n ,
where ai , i = 1, 2,..., n are the affixes of the vertices Ai , i = 1, 2,..., n , respectively.

Example B. Let S be the center of a circumcenter, аnd H be the orthocenter of

 ABC. Furthermore, let the point Q be such that S is the midpoint of the line segment
HQ and let T1 , T2 and T3 be the centroids of  BCQ,  CAQ and  ABQ, respectively.
Prove that
AT1 = BT2 = CT3 = 43 R ,
where R is the circumradius of  ABC.
Solution. Without loss of generality, we can say that the circumcircle of  ABC is
h+ q
the unit circle, i.e. that o = 0 and | a | = | b | = | c | = 1 . We have h = a + b + c and o = 2 ,
b+c+ q
and therefore q = − h = − a − b − c . Furthermore, t1 = 3 = − a3 and similar to this we
get that t2 = − b3 , t3 = − 3c . Now we have that

107
AT1 = | a − t1 | = a + a3 = 43a = 43 = 43 R , BT2 = CT3 = 4 = 4 R ,
3 3
Which was actually supposed to be proven. ■

5.11. Theorem (Leibniz). If T is the centroid of  ABC and P is an arbitrary

point of the plane of the triangle, then
2 2 2 2 2 2 2
PA + PB + PC = 3PT + TA + TB + TC .
Proof. Let a, b, c be the affixes of the vertexes A, B, C respectively and let p be
the affix of the point P. From the theorem 5.8, we get
2 2 2 2 2 2 2 2
a + b + c −3 p
3PT + TA + TB + TC = 3 3
+ b +c3−2a + a +c3−2b + a +b3−2c

= 3 p p + aa + bb + cc − pa − pa − pb − pb − pc − pc
= | p − a |2 + | p − b |2 + | p − c |2
2 2 2
= PA + PB + PC ,
Which was supposed to be proven. ■

5.12. Example. If T is the centroid of  ABC, then

2 2 2 2 2 2
TA + TB + TC = 13 ( AB + BC + CA ) .
Prove this!
Solution. It is sufficient to get that P ≡ A , P ≡ B and P ≡ C in the equality of
theorem 5.11 and after that to summaries all the obtained equalities. ■

5.13. Let it be given  ABC and let consider the homothety

. w = − 12 z + h2 . (5)
The equality z = − 12 z + h2 implies that z = h3 , i.e. the center of the homothety (5) is the
centroid T of  ABC. The point A under the homothety (5) maps to a point with an affix
− a2 + h2 = b +2 c ,
This means that at A1 maps in the midpoint of the side BC. Analogously it can be
proven that the points B and C map at the midpoints B1 and C1 of the sides AC and AB,
respectively. Thus, we proved the following theorem.

Theorem. If T is the centroid of  ABC and A1 , B1 and C1 are the midpoints of

the sides BC, AC and AB, respectively, then the homothety with a center in T and a ration
− 12 maps  ABC to  A1B1C1 . ■

5.14. Consequence. If A1 , B1 and C1 are the midpoints of the sides BC, AC and
AB of  ABC, then the following holds true

108
A1B1 || AB, B1C1 || BC , C1 A1 || CA,
2 A1B1 = AB, 2 B1C1 = BC , 2C1 A1 = CA .
Proof. The theorems 5.13 and II 6.5 imply that
A1B1 || AB, B1C1 || BC , C1 A1 || CA,
and the theorems 5.12 and 4.6 imply that
2 A1B1 = AB, 2 B1C1 = BC , 2C1 A1 = CA . ■

5.15. Definition. Let A1 , B1 and C1 and be the midpoints of the sides BC,
AC and AB, respectively, of  ABC. The line segments A1B1 , B1C1 , C1 A1 are called the
medians of the sides AB, BC, CA, respectively.

5.16. Remark. In consequence 5.14 we proved that the medians of the triangle
are parallel to the suitable sides of the triangle and that the length of each one Is half of
the length of the suitable side.

6. Right angled triangle

6.1. We call  ABC right angled triangle if its orthocenter H coincides with one
of the vertices A, B or C. Due to this,  ABC is a right angled triangle if and only if
| h | = 1 , i.e if and only if
(a + b + c)(a + b + c) = 1.
The last equality is equivalent to the equality
(a + b)(b + c)(c + a ) = 0 .
Which implies that the  ABC is a right angled triangle if and only if either
a + b = 0 or b + c = 0 or c + a = 0 .
Thus, we proved the following theorem.

Theorem. The triangle ABC is a right angled triangle if and only if either a + b = 0
or b + c = 0 or c + a = 0 . ■

6.2. Consequence. The triangle ABC is a right triangle if and only if one of the
sides AB, BC or CA is the diameter of the circle circumscribed around it.
Proof. It is directly implied by the theorem 6.1. ■

The side of the right angled traingle ABC which is the diameter of its circumcircle
is called a hypotenuse, and the remaining two sides are called legs of  ABC.

109
6.3. Theorem (Pythagoras). For every right angled triangle the square of the
length of the hypotenuse is equal to the sum of the squares of the length of its legs.
Proof. Let AB be the hypotenuse of the right triangle ABC. According to theorem
16.1 we get that a + b = 0 i.e. b = − a , thus
2 2
AC + BC = | c − a |2 + | c − b |2
= | c − a |2 + | c + a |2
= (c − a )(c − a ) + (c − a )(c + a )
= 2 | c |2 + 2 | a |2 = 4 | b |2
2
= | 2b |2 = | b − a |2 = AB ,
which was supposed to be proven. ■

6.4. Example. If the hypotenuse of the right triangle is divided in three equal
parts and the point of division are connected with the vertex of the right angle, then the
sum of the squared of the length of the sides of so obtained triangle is equal to 23 of the
square of the hypotenuse. Prove it!
Solution. Without loss of generality, we take that ABC is a right angled triangle
with a right angle in the vertex C, such that it is inscribed in the unit circle. Let the affixes
on the vertices A, B, C are a, b, c respectively. If D and E are points of the hypotenuse
AB such that
AD = DE = EB ,
then their affixes are 2a3+b and a +2
3
b , respectively. According to theorem 6.1 we get
that
2 2 2 2 2 2
CD + DE + EC = 2a +3b −3c + b −3a + a + 23b −3c
2 2 2
= a −33c + 23a + a +33c

(
= 2a a3+6cc = 23 | a |2 +3 | c |2 )
( )
= 23 | a |2 +3 | a |2 = 23 | 2a |2
2
= 23 | b − a |2 = 23 AB ,
which was supposed to be proven. ■

110
7. EULER LINE AND EULER CIRCLE

7.1. Theorem. The circum center O, the centroid T and the orthocenter H of
 ABC are on a same line and furthermore OH = 3OT .
Proof. For the affixes h and t of the orthocenter H and the centroid T we get
that
h = a + b + c = 3 a +3b +c = 3t ,
which means that O, T and H are collinear and thus OH = 3OT . ■

7.2. Definition. The line on which are the circum center O, the centroid T and the
orthocenter H of  ABC is called Euler line for  ABC.

7.3. Example. If T and O are the centroid and the circum center of  ABC,
respectively, then
2 2 2 2
OT = R 2 − 19 ( AB + BC + CA ) ,
where R is the length of the circumradius of the  ABC.
2 2
Solution. According to the theorem 7.1 we get OH = 9OT . So, the statement
is implied directly by example 5.5. ■

7.4. According to theorem II 2.2 the equation of the Euler line OH is z = h z ,

h
which means that its complex angle coefficient is h . Let t1 , t2 , t3 and t4 be the complex
h
angle coefficients of the lines BC, CA, AB and the Euler line of  ABC. We get that,
t1 = −bc, t2 = −ca, t3 = − ab and t4 = h . Thus,
h
t1t2 + t1t3 + t1t4 + t2t3 + t2t4 + t4t4 = (t1 + t2 + t3 )t4 + t2t3 + t1t3 + t1t2
= −(bc + ca + ab) h + abc(a + b + c)
h
= −h h + 1⋅ h = 0
h
For any movement the complex angle coefficients of the lines are multiplied by the same
constant, thus if the equality
t1t2 + t1t3 + t1t4 + t2t3 + t2t4 + t4t4 = 0 (1)
is true for a  ABC which is inscribed in the unit circle centered at the origin O, and for
the affixes a, b, c of its vertices A, B, C it is true that abc = 1, then this is true for any
triangle. Thus, we proved the following theorem.

Theorem. For the complex angle coefficients t1 , t2 , t3 and t4 of the sides and the
Euler line of an arbitrary triangle the equality (1) holds true. ■

111
7.5. Theorem. If H is the orthocenter of  ABC, A1 , B1 , C1 are the midpoints
of the sides BC, CA, AB, respectively, A2 , B2 , C2 are foots of the altitudes and
A3 , B3 , C3 are the midpoints of the line segments AH, BH, CH, then the points
A1 , B1 , C1 , A2 , B2 , C2 , A3 , B3 , C3 are on a same circle.

Proof. The midpoints A1 , B1 , C1 of the sides BC, CA, AB have the affixes b +2 c ,
c + a , a +b . According to consequence 5.7 A2 , B2 , C2 the foots of the altitudes have
2 2
2 2 2 2 2 2
the affixes h −b2 c , h −a2 c , h −b2 a , and since A3 , B3 , C3 are the midpoints of the line

segments AH, BH, CH we get that their affixes are a +2 h , b +2 h , c +2h .

Due to the example II 3.3 we get that O9 the circumcircle of the  A1B1C1 has
the affix
ε = a +2b +c = h2 ,
and its radius is
R1 = ε − a +2 b = 12 .
By a direct check we can conclude that the points A2 , B2 , C2 , A3 , B3 , C3 are on the same
circle centered at O9 and with radius R1 = 12 , which means that the points A1 , B1 , C1 ,
A2 , B2 , C2 , A3 , B3 , C3 are on a same circle. ■

7.6. Definition. A circle centered at O9 and with radius R1 = O9 A1 , is called

Euler circle, and the point O9 is called Euler point for  ABC.

7.7. Remark. Clearly, for any  ABC the Euler point is a midpoint of the line
segment OH, i.e. it is placed on the Euler line, and the radius of the Euler circle is equal
to R2 , where R is the radius of the circumcircle of the  ABC.

7.8. Let L, M, N be the points with the affixes

l = b + c, m = c + a , n = a + b ,
respectively. then, the line segments BC and OL have a common midpoint A1 , so the
point L is symmetrical to the center O of the circumcircle of the  ABC with respect to
the A1 , i.e with respect to the line BC. Since
a +l
= b +2m = c +2n = h2
2
we get that the line segments AL, BM, CL, HO have common midpoint, and that is the
Euler point O9 . The equalities
h = l + a , b = l − c, c = l − b
and the fact that | a | = | b | = | c | = 1 , we conclude that the points H, B, C are on the circle
with a center at L and a length radius 1. Similarly, H, C, A are on the circle centered at M
and a length radius of 1, and H, B, A are on the circle centered at N and a length radius

112
of 1. The symmetry with respect to the point O9 implies that the other statements of the
following theorem are true.

Theorem. If L, M, N are symmetrical points to the center O of the circumcircle

of the  ABC to the lines BC, CA, AB, respectively, аnd H is the orthocenter of  ABC,
then the quadrilaterals ABCH and LMNO are symmetrical with respect to the Euler point
O9 . The triangles ABC, BCH, CAH, ABH, LMN, MNO, NLO, LMO subsequently have
the orthocenters H, A, B, C, O, L, M, N and the circumcircle with congruent radii and
centers at the points O, L, M, N, H, A, B, C, respectively. ■

7.9. Example. Let H be the orthocenter of  ABC. Prove that the Euler lines of
the triangles ABC, ABH, BCH and CAH intersect at a unique point.
Solution. Without loss of generality we can say that  ABC is inscribed into the
unit circle. The orthocenter H of the triangle has an affix h = a + b + c . The point O ' with
an affix o ' = a + b is symmetrical to the center O of the circumcenter with respect to the
line AB. Moreover,
O ' A = | a + b − a | = | a | = 1,
O ' B = | a + b − b | = | b | = 1 and
O ' H = | a + b + c − a − b | = | c | = 1,
thus O ' is the circumcenter of the  ABH.
Analogously, the points O '' and O ''' with the
affixes o '' = b + c and o ''' = a + c are circum-
centers of the triangles BCH and ACH, respec-
tively. If T ' is the centroid of the triangle ABH,
then its affix is
a +b + ( a +b + c )
t'= 3
= 2a +32b +c .
The Euler lines of the triangles ABH and ABC Figure 27
are the lines T ' O ' and OH, respectively.
( a +b ) + ( a +b + c ) +0
Thereby, t ' = 3
we get that T ' is the centroid of  HOO ' , thus the line
T ' O ' intersects the line segment OH at point E with an affix e = a +2b +c . Therefore, the
point E is the intersection of the Euler lines of the triangles ABH and ABC.
Similarly, we can prove that the point E is the intersection of the Euler lines of the
triangles BCH and ABC, that is the triangles CAH and ABC, which means that the Euler
lines of the triangles ABC, ABH, BCH and CAH intersect in the same point. ■

113
8. MENELAU’S THEOREM
 
8.1. If AB and CD are collinear vectors, then it exists a real number λ such that
  
AB = λCD . In the following consideration we will get that λ = AB
 .
  CD  
Since the equality AB = λCD is equivalent to the equality CD = λ1 AB we get

that CD
 = 1 .
AB λ

8.2. Definition. Let the side AB of  ABC be on the line (p). The point P is called
Menelaus point of the side AB if P ∈ ( p ) and P ≡/ A, B . Analogously, we define the
Menelaus points of the sides BC and CA of the  ABC.

8.3. Theorem (Menelaus). Let D, E and F be the Menelaus` points of the sides
BC, CA and AB of any  ABC, respectively. The points D, E and F are collinear if and
only if it holds true that
  
 ⋅ CE
BD  ⋅ AF
 = −1 . (1)
DC EA FB

Proof 1. Let D, E and F with affixes p, q and r respectively be the Menelaus`

points of the sides BC, CA and AB. If it holds true that
  
BD
 = λ, CE
 = µ, AF
 = ν ,
DC EA FB
then for the affixes p, q and r of the points D, E and F we get that
c +µa
p = b1++λλc , q = 1+µ , r = a1++ννb . (2)
The points D, E and F are collinear if and only if
p −q r −q
= .
p −q r −q
If in the last equality (2) we substitute the values of p, q and r and the obtained equality
we multiply by (1 + λ )(1 + µ)(1 + ν) , we get that
(1 + λµν)(ab + bc + ca − ba − cb − ac) = 0 . (3)
Therefore, the points D, E and F are collinear if and only if the equality (3) is satisfied.
Lastly, the points D, E and F are collinear if and only if 1 + λµν = 0 (why?), that is if and
only the condition (1) is satisfied
Proof 2. Let the condition (1) be satisfied , i.e µ = − λν
1 . Then

p = b1++λλc , r = a1++ννb and q = λν c −a ,

λν −1
thus
 (1+λ ) a +(λν −1)b −λ (1+ν )c
DF = (1+ ν )(1+ λ )
,
 (1+λ ) a +(λν −1)b −λ (1+ν )c
DE = (1−λν )(1+ λ )
,

114

which means that DE  = 1+ λ ∈ R , since for 1 − λν = 0
DF 1−λν
we get that FD || AC , which contradicts the finiteness
of the point E. Therefore, DE || DF , which implies that
the points D, E and F are collinear.
Let the points D, E and F be collinear and let
the projections of the points A, B, C on the line ED be
the points A ', B ', C ' , respectively (see the figure). Then
the triangles BB ' D and CC ' D are directly similar, and
thus Figure 28
b ' −b p −b
= cp'−−cc , i.e. bc ''−−bc =
p −b
. p −c
Analogously, the direct similarity of the triangles AA ' E and CC ' E implies that
a '− a q −a
c '− c
= q −c and the direct similarity of the triangles AA ' F and BB ' F implies that
b ' −b = rr −−ba . Finally,
a '− a   
 ⋅ CE
BD
DC EA FB
 ⋅ AF
 =
b− p c −q a −r
⋅ ⋅
p − c q − a r −b ( )( )( )
= − bc ''−−bc − ac ''−−ca − ab ''−−ba − 1 . ■

8.4. Example. Given is the  ABC and the points D and E on the sides BC and CA,
respectively, such that BD = CE = AB . We draw a line (l) through the point D, parallel to
3
AB. If M = (l ) ∩ BE and F = CM ∩ AB , then AB = AE ⋅ FB ⋅ CD . Prove this!
Solution. Let’s consider the  ACF (figure 29). The points E, M and B are
Menelaus` points of the sides AC, CF and AF, respectively, and under a condition, they
are collinear. From the Menelaus’ theorem , we get that
  
AB
  ⋅ CE
⋅ FM  = −1 ,
BF MC EA
which implies that
AB ⋅ FM ⋅ CE =1.
BF MC EA

Due to DM || BF we get that FM = BD . If we substitute in

MC DC
the previous equality, we get that
AB ⋅ BD ⋅ CE =1
BF DC EA
3 Figure 29
And due to BD = CE = AB we get AB = AE ⋅ FB ⋅ CD . ■

115
9. PASCAL`S AND DESARGUES` THEOREM

9.1. In this section, by using the Menelay’s theorem, we are going to prove the
Desarg theorem which is a fundamental result in projective geometry. We are also going
to prove the Pascal`s theorem for a hexagon inscribed in a circle.

Definition. The triangles ABC and A ' B ' C ' are called copolar if the lines
AA ', BB ' and CC ' are concurrent.
The triangles ABC and A ' B ' C ' are called coax if the points of intersection of the
lines BC and B ' C ' , CA and C ' A ' , AB and A ' B ' lie on the same line.

9.2. Theorem (Desargue). The triangles ABC and A ' B ' C ' are copolar if and
only if they are coax.
Proof. Let the triangles ABC and A ' B ' C ' be copolar and let the lines AA ', BB '
and CC ' intersect in the point O. Let’s denote the points of intersection of the lines
BC and B ' C ' , CA and C ' A ' , AB and A ' B ' by P, Q, R respectively (figure 30). The
Menelaus’ theorem, applied to the triangles, BCO, CAO and AOB implies
  
 ⋅ CC
BP ' ⋅ OB
' = −1 ,
PC C ' O B ' B
  
CQ AA ' OC '
 ⋅  ⋅  = −1
QA A ' O C ' C
  
AR ' ⋅ OA
 ⋅ BB ' = −1
RB B ' O A ' A
If we multiply the above equalities, we
get
  
CQ
BP
 ⋅  ⋅ AR
 = −1 .
PC QA RB
Thus, from the Menelaus’ theorem we
conclude that the points P, Q and R are
collinear. Therefore, the triangles ABC
and A ' B ' C ' are coax.
Figure 30 Reversely, let’s assume that P,
Q and R are collinear and let the lines
AA' and BB ' intersect in the point O. Now, the triangles AQA' and BPB ' are copolar,
and therefore, coax. According to this, the points O, C and C ' are collinear, which means
that the coax triangles are copolar. ■

9.3. Theorem (Pascal). Let the hexagon ABCDEF, whose opposite sides are not
collinear, be inscribed in a circle. Let denote by L, M, N the points of intersection of the
three pairs of opposite sides AB and ED, BC and EF, FA and CD, respectively. Then, the
points L, M, N are collinear.

116
Figure 31

Proof 1. Let X, Y, Z be the points of intersection of AB and CD, CD and EF, EF

and AB, respectively (figure 31). The points D, E, L; F, A, N; B, C, M are the Menelaus’
points for  XYZ, and from the Menelaus’ theorem we get that
        
XL
 ⋅ ZE
 ⋅ YD
 = −1, XA  ⋅ YN
 ⋅ ZF  = −1 , XB  ⋅ YC
 ⋅ ZM  = −1.
LZ EY DX AZ FY NX BZ MY CX
If we multiply the above equalities, we get that

( )
        
XL
  ⋅ YN
⋅ ZM  ⋅ ZE
 ⋅ YD
 ⋅ XA
 ⋅ ZF  ⋅ YC
 ⋅ XB  = −1 . (1)
LZ MY NX EY DX AZ FY BZ CX
Furthermore, we apply the power of the points X, Y, Z with respect to the circle, then the
geometrical interpretation of complex numbers, implies that
           
ZE ⋅ ZY = AZ ⋅ BZ , EY ⋅ FY = YD ⋅ YC , CX ⋅ DX = XA ⋅ XB .
If we substitute in (1) we get
  
XL
  ⋅ YN
⋅ ZM  = −1,
LZ MY NX
which according to the Menelaus’ theorem means that the points L, M and N are
collinear.
Proof 2. Without loss of generality, we can say that the hexagon ABCDEF is
inscribed in a unit circle. The affixes l, m, n of the points L = AB ∩ DE , M = BC ∩ FE
and N = CD ∩ AF are
a +b −( d + e ) b + c −( e + f ) c + d −( f + a )
l= ab − de
, m= bc −ef
and n = cd − fa
.
Furthermore,
(b −e)(bc −cd + de −ef + fa − ab ) (c − f )(cd − de + ef − fa + ab −bc )
l−m= ( ab − de)(bc −ef )
and m − n = (bc −ef )(cd − fa )
,
So,
l −m (b −e)(cd − fa )
= ( f −c )( ab −de) .
m−n
Finally, if we take that for every point of a unit circle it is true that x = 1x , by applying the
properties of the complex numbers we get that

117
l −m (b −e)(cd − fa ) ( b1 − 1e )( cd1 − fa1 ) (e−b)( fa−cd ) l −m
= = 1 1 1 1 = (c − f )( de−ab) = ,
m−n ( f −c )( ab − de) ( f − c )( ab − de ) m−n

which means that m l − m is a real number, therefore the consequence 1.4 implies that the
−n
points L, M and N are collinear . ■

10. TRIANGULAR COORDINATES

10.1. Lemma. If for the complex numbers a, b and c holds true

λa + µb + νc = 0 (1)
where
λ + µ + ν = 0 (2)
and λ, µ, ν are nonzero real numbers, then the points A, B and C whose affixes are a, b
and c, respectively, are collinear, and vice versa.
Proof. Truly, from (1) and (2) it follows that
µ
a+ λ b
c= µ
, (3)
1+ λ
µ
i.e. the point C divides the line segment AB in a ratio λ , which means that the points A,
B and C are collinear.
Reversely, if the points A, B and C lay on a same line and if the point C divides
µ
the line segment AB in a ration of λ , then from (3) in
ν = −(λ + µ )
we obtain the equalities (1) and (2). ■

10.2. Remark. Equality (2) implies that the numbers λ, µ and ν may not have the
same sign, i.e. one of them must have an opposite sign from the other two. The point to
which in (1) corresponds this number is between the other two points.
Thus, for example, from the equality
3a − b − 2c = 0
according to lemma 10.1 we get that the point whose affix is the complex number a is
between the points whose affixes are the complex numbers b and c.

10.3. Lemma. Let A, B and C, with affixes a, b and c, respectively, are three
non-collinear points in the plane. Then, to every point in the plane D, with an affix d,
correspond three real numbers λ, µ and ν such that

118
λa + µb + νc = d (4)
where
λ + µ + ν = 1 (5)
Proof. Let’s connect the point D, for example, with the point A and the intersection
of the lines AD and BC be denoted by D ' . Then, the point D divides the line segment
AD ' in a ratio of α : λ = DD ' : DA , and the numbers α and λ can be chosen so that
α + λ = 1 . the lemma 20.1 implies that
− d + λa + αd ' = 0 and −1 + λ + α = 0 (6)
Furthermore, the point D ' divides the line segment BC with a ratio
ν : µ = CD ' : D ' B
These numbers can be chosen so that
ν+µ=α.
According to the lemma 10.1 we get that
−αd '+ µb + νc = 0 and −α + µ + ν = 0 . (7)
Now the equalities (4) are (5) are implied directly by the equalities (6) and (7).
Reversely, for µ and ν it exists a sole point D ' on the line BC whose affix is
determined by
µb + νc
d ' = µ+ν .
Now, on the line AD ' there is a unique point D such that
λa + (µ + ν ) d '
d= λ +µ + ν
= λa + µb + νc , λ + µ + ν = 1 ,
This means that (4) and (5) determine a unique point D in the plane. ■

10.4. Remark. The numbers λ, µ

and ν uniquely determine the position of
the point D with respect to the  ABC.
Therefore, the ordered triple (λ,µ,ν) is called
triangle coordinates of the point D with
respect to  ABC. From (5) we get that all
three numbers λ, µ and ν cannot be negative,
and the position of the point D with respect
to  ABC is determined by the signs of the
numbers λ, µ and ν (figure 32).

Figure 32

119
11. CEVA`S AND VAN AUBEL`S THEOREM

11.1. Theorem (Ceva). Let D ' , D '' and D ''' lie on the sides BC, AC and AB of
 ABC or their extensions, respectively. The lines AD ', BD '' and CD ''' intersect in one
point if and only if the following equality is satisfied
  
' ⋅ CD
BD '' ⋅ AD
''' = 1. (1)
D ' C D '' A D ''' B
Proof. Let the lines AD ', BD '' and CD ''' intersect in a point Q and let (p) be a
line which goes through the point A and is parallel to the line BC. Let BD ''∩ ( p ) = {K }
and CD '''∩ ( p ) = {L} , figure 33. The triangle D ' QB is directly similar to the triangle
q −d ' q −a
AQK, thus b −d ' = k −a . The triangle D ' CQ is directly similar to the triangle ALQ, thus
c − d ' = l − a . From the last two equalities we get that
q −d ' q −a
b−d '
= kl −−aa . (2)
c −d '
Furthermore, the triangle CD '' B is directly simi-
lar to the triangle AD '' K , therefore the equality
d ''−c = d ''− a holds true, and it is equivalent to the
b −c k −a
equality
d ''−c
= kb−−ac , (3)
d ''− a
Figure 33 The triangle BCD ''' is directly similar to the triangle
ALD ''' , and therefore the equality dc'''−−bb = dl'''−−aa
holds true, and is equivalent to the equality
d '''− a = cl −−ab . (4)
d '''−b
Finally, the equalities (2), (3) and (4) imply that
  
' ⋅ CD
BD ( )( )(
= dc −'−db' ⋅ ad−''−dc'' ⋅ bd−'''d−a''' = − kl −−aa − kb−−ac − cl −−ab = 1,
'' ⋅ AD
D ' C D '' A D ''' B
''' )
i.e. the equality (1) holds true.
On the other hand, if BD ''∩ CD ''' = {Q} and AQ ∩ BC = { A '} , then, the previously
     
' ⋅ CD
proven, implies that BA '' ⋅ AD ' ⋅ CD
''' = 1 and by assumption that BD '' ⋅ AD
''' = 1
A ' C D ''
 A D ''' B D ' C D '' A D ''' B
holds true we get that BD ' BA
 = ' . The latter implies that the points A' and D ' coincide,
D 'C A 'C
i.e. Q ∈ AD ' . ■

11.2. Remark. According to the Ceva theorem the medians of  ABC intersect
in one point. Namely, if A1 , B1 and C1 are the midpoints of the sides BC, CA and AB
     
respectively, then BA1 = A1C , CB1 = B1 A and AC1 = C1B , thus
  
BA CB AC
1 ⋅ 1 ⋅ 1 = 1 ,
A1C B1 A C1B

120
which according to the Ceva theorem means that the medians AA1 , BB1 and CC1 intersect
in one point.

11.3. Definition. The line segments (the lines) AD ', BD '' and CD ''' from theorem
11.1 are called Ceva line segments (lines) for  ABC.

11.4. Theorem. The lines which connect the midpoints of the sides of a triangle
with the suitable Ceva line segments intersect in a unique point.
Proof. Let in  ABC (figure 34) AD, BE and CF be arbitrary lines which intersect
in the point M; A ', B ' and C ' are midpoints of the lines BC, CA and AB respectively and
P, K and L are midpoints of the line segments AD, BE and CF respectively. By applying
the Ceva theorem to AD, BE and CF, we get the following
AE ⋅ CD ⋅ BF =1,
EC DB FA
thus,
AE CD BF
2
EC
⋅ 2
DB
⋅ 2
FA
=1
2 2 2
i.e
C 'K ⋅ B ' P ⋅ A ' L = 1.
KA ' PC ' LB '
Finally, from the Ceva theorem applied to
 A ' B ' C ' we get that the lines A ' P, B ' K and
Figure 34
C ' L intersect in one and only one point. ■

11.5. Consequence. The lines which connect the midpoints of the sides of a
triangle and the midpoints of its altitudes intersect in one and only one point.
Proof. It is implied directly from the theorems 5.1 and 11.4. ■

11.6. Theorem (Van Aubel). If A', B ', C ' are points on the sides BC, CA, AB
respectively, of the triangle ABC such that the lines AA ', BB ', CC ' intersect in the point
Q, then the following holds true
AQ
= AC ' + AB ' .
QA ' C 'B B 'C
Proof. Let the affixes of points be labeled by
a suitable small letter. Let (r) be the linе which goes
through the point C and is parallel to the line AA' , and
(s) be the line which goes through the point B and is
parallel to the line AA' and let (r ) ∩ BB ' = {L} and
( s ) ∩ CC ' = {K } , figure 35. Figure 35

121
q −a
The triangle AQC ' is directly similar to the triangle BKC ' , thus c '−a = ck'−−bb ,
q −a
i.e. cc ''−−ba = k −b . Furthermore, the triangle CB ' L is directly similar to the triangle AB ' Q ,
b '− c b '− a q −a
thus l −c = q −a , i.e bb ''−−ac = l −c ; the triangle BKC is directly similar to the triangle
q −a ' q −a '
A ' QC , thus kc −−bb = c −a ' , i.e. cc−−ab' = k −b and the triangle CLB is directly similar to the
q −a ' q −a '
triangle A ' QB , thus bl −−cc = b −a ' , i.е. bb−−ac' = l −c . Now,
AC '
C 'B
| c '− a | | b '− a | | q − a | | q − a |
(
+ AB ' = | b −c '| + | b '−c | = | k −b | + | l −c | =| q − a | | k 1−b | + | l −1c |
B 'C )
( | c − a '| | b − a '|
)
= | q − a | | q −1a '| | c −b | + | q −1a '| | b −c | = | q −a '| ⋅
| q − a | | c − a '|+| b − a '|
| b −c |

| q −a | AQ
= | q −a '| = . ■
QA '

11.7. Comment. In the Van Aubel theorem the lines AA ', BB ', CC ' intersect in
the point Q, therefore by the Ceva theorem we get that BA ' ⋅ CB ' ⋅ AC ' = 1, which means
A 'C B ' A C ' B
p
that there exist real numbers m, n, p such that BA ' = = mn . Then, the n
, CB ' = mp , AC '
A 'C B' A C 'B
nb + pc
affix of the point A', is determined by a ' = n + p and thereby the Van Aubel theorem
AQ p n+ p
holds true that = AC ' + AB ' = n + = m , therefore the affix of the point Q is as
QA ' C ' B B ' C m m
following
ma + ( n + p ) a ' ma + nb + pc
q= m+ n+ p
= m+ n + p . (5)

11.8. Theorem. If A', B ', C ' are points on the sides BC, CA, AB, respectively,
of the triangle ABC such that
p
= n , CB ' = mp , AC ' = mn .
BA '
A 'C
B' A C 'B
Then the lines AA ', BB ', CC ' intersect in a point Q whose affix is determined in (5).
Proof. Thereby, it holds true that
p
BA ' = n , CB ' = mp and AC ' = mn ,
A 'C B' A C 'B
we get that
nb + pc pc + ma
a ' = n + p , b ' = p + m and c ' = ma + nb .
m+ n
Clearly, the first part of the statement implies from the Ceva theorem. We are going to
show that the point Q whose affix is given in (5) lies on the line AA' . We have
ma + nb + pc
q −a m+n+ p
−a n+ p nb + pc −( n + p ) a n+ p
a '− a
= nb + pc
= m+ n + p ⋅ nb + pc −( n + p ) a = m+ n + p ∈ R ,
n+ p
−a

122
which according to the consequence 1.4 means that the points A, Q and B, Q, B ' are
collinear. Analogously, we prove that the points are collinear and that the points C , Q, C '
are collinear as well. ■

11.9. The last theorem can be used to find the affixes of some important points
of a triangle, such as the centroid, the center of the in-circle, the Jargon point( will be
discussed later) and so. For example, for the midpoints A', B ', C ' of the sides BC, CA,
AB of the triangle, it holds true that
BA ' = 1, CB ' = 1, AC ' = 1 ,
A 'C B' A C 'B
i.e. m = n = p = 1 , thus, from the theorem 11.8 we get that the medians intersect at a point
T with affix t = a +3b +c . ■

12. AREA OF A TRIANGLE

12.1. Let be given a  ABC and let the affixes of the vertices A, B, C be a, b, c
respectively. We plot a line through the vertices B and C, whose auto conjugated equation
is as following
i (c − b)z − i (c − b) z + i (cb − cb) = 0 . (1)
The distance from the point A to the line (1), i.e. the length of the altitude of  ABC plot
at the vertex A, is
i (c −b )a −i (c −b ) a +i (cb −cb )
hBC = 2| c −b |
,
i.e.
− (c −b ) a + (c −b ) a −cb + cb
hBC = 2| c −b |
.
Therefore the area of the  ABC is
BC ⋅hBC − (c −b ) a + (c −b ) a −cb + cb
P ABC = = . (2)
2 4
Since the arbitrary complex numbers u and v the number uv − vu is an imaginary number,
the equality(2) transforms as following
− (c −b ) a + (c −b ) a −cb + cb
P ABC = ±i ,
4
i.e
a a 1
P ABC = ± 4i b b 1 . (3)
c c 1
12.2. Definition. Let be given  ABC and let the affixes of the vertices A, B, C
be a, b, c respectively. We shall say that the  ABC is positively oriented with respect to

123
the considered coordinate system, if its area, calculated by applying the formula (3), is
obtained when we multiply by +i, and negatively oriented if its area is obtained when the
formula (3) is multiplied by −i.

12.3. Remark. If  ABC is a right angled triangle, with a right angle in the vertex
C, then for the affixes in the vertices A and B it holds true b = − a , thus from (3) for the
area of the triangle we get that
| ca −ca |
P ABC = 2
.

12.4. Remark. The affix of the point P ' , symmetrical to the point P with an affix
p, with respect to the line which passes through the points with affixes a and b, can be
determined by using the following condition
a a 1
b b 1 =0,
m m 1
p+ p '
where m = 2 . The details are left to the reader as an exercise.

12.5. Definition. We shall say that the n-gon A1 A2 ... An is positively oriented if
 A1 A2 A3 is positively oriented.
We shall say that the n-gon A1 A2 ... An is negatively oriented if  A1 A2 A3 is
negatively oriented.

12.6. Theorem. If A1 A2 ... An is a convex polygon whose vertices A1 , A2 ,... , An

have affixes a1 , a2 ,... , an respectively, and S is its area, then S = ± 12 Im(Ta, a) , where
a = (a1 , a2 ,..., an ) and T is the mapping given in paragraph I 9.
Proof. The area of the polygon is calculated as a sum of the areas of  A1 A2 A3 ,
 A1 A3 A4 ,... ,  A1 An −2 An −1 and  A1 An −1 An . Furthermore, these triangles are equivalently
oriented, therefore
a1 a1 1 a1 a1 1 a1 a1 1 a1 a1 1 a1 a1 1
i a2 a2 1 i a3 a3 1 i a4 a4 1 i an − 2 an − 2 1 i an −1 an −1 1
a3 a3 1 a4 a4 1 a5 a5 1 an −1 an −1 1 an an 1
S =± 4
± 4
± 4
± .... ± 4
± 4

= ± 4i (a1 a2 − a2 a1 + a2 a3 − a3 a2 + a3 a4 − a4 a3 + ... + an a1 − a1 an )
= ± 4i (2i Im a1 a2 + 2i Im a2 a3 + 2i Im a3 a4 + ... + 2i Im an a1 )
= ± 4i 2i Im(a1 a2 + a2 a3 + a3 a4 + ... + an a1 )

= ± 12 Im(a, Ta) = ± 12 Im(Ta, a) . ■

124
12.7. Remark. The formula for calculating the area of a convex polygon, given
in the previous theorem, applies as well when the triangle is not convex.

12.8. Example. Let be given a convex pentagon A1 A2 A3 A4 A5 . If we connect the

midpoints of its sides consequently, we get a new pentagon. Following the procedure, we
get a series of pentagons and let S0 , S1 , S2 ,... be their areas. Prove that
16 Sn + 2 − 12 Sn +1 + Sn = 0 .
Solution. It is sufficient to prove that
16 S2 − 12 S1 + S0 = 0 (4)
For this purpose we will first calculate S1 and S2 .
±8S1 = Im(Ta + T 2a, a + Ta) = 2 Im(Ta, a) + Im(T 2a, a) and

(( 2
) 2
±2 S2 = Im T a + 2Ta4+T a , a + 2Ta4+T a , )
i.e.
±32 S2 = 5 Im(Ta, a) + 3 Im(T 2a, a) .
If we eliminate Im(T 2a, a) and take that ± S0 = Im(Ta, a) we get the equality (5). ■

12.9. Example. Prove that if the odd vertices of a n-gon are translated for the
same vector, then the area of the new n-gon is equivalent to the area of the given n-gon.
Solution. Let a be the oriented n-tuple of the affixes of the n-gon`s vertices. The
area of the n-gon is as following
S = ± 12 Im(Ta, a) ,
For the areа of the new n-gon we get the following
S ' = ± 12 Im (T (a + h), a + h )
where h is one of the following ordered n-tuples
h = (α, 0, α, 0,..., α, 0) or h = (α, 0, α, 0,..., α) , α ∈C ,
depending whether n is an even or an odd number, respectively.
We get that
±2S ' = Im (T (a + h), a + h ) = Im {(Ta, a) + (Ta, h) + (Th, a) + (Th, h)} .
From
(Ta, h) = (a, Th) = (Ta, T 2h) = (Ta, h)
we get the following
Im {(Ta, h) + (a, Th)} = 0
and if (Th, h) = 0 we get
±2 S ' = Im (T (a + h), a + h ) = Im(Ta, a) = ±2 S .
According to this, S = S ' , which was supposed to be proven. ■

125
12.10. Let’s consider a  ABC, whose vertices A, B and C have the affixes a, b
and c, respectively. In the example II 3.3 we proved that
aa (c −b ) +bb ( a −c ) + cc (b − a )
o=
a a 1
b b 1
c c 1
Is the affix of the center O of the circumcicrle of the  ABC. Clearly, the radius of the
circumcircle of the  ABC is R = | a − o | . According to this, for the radius of the circum
radius if of  ABC is as following
| a −b | ⋅ | b − c | ⋅ | a − c |
R = |o−a| =
a a 1
b b 1
c c 1
Since, AB = | a − b |, BC = | b − c |, CA = | c − a | , and also applying 12.1 we get the
following formula
AB ⋅ BC ⋅ CA
R = 4P .
 ABC

12.11. Theorem. The ratio between the areas of two similar triangles is equivalent
to the ratio of the squares of the respected sides.
Proof. Let the triangles ABC and A1B1C1 , whose affixes of the vertices are a, b,
c and a1 , b1 , c1 respectively, be similar. There are two possible cases.
а) There is a direct similarity S ( z ) = dz + e which maps  ABC into  A1B1C1
and therefore we get
a1 a1 1 da + e da + e 1 a a 1
b1 b1 1 db + e db + e 1 b b 1
c1 c1 1 dc + e dc + e 1 c c 1
P A1B1C1 = 4
= 4
= | d |2 = 4
= | d |2 P ABC .
b) There is an indirect similarity S ( z ) = dz + e which maps  ABC to  A1B1C1 .
Analogously as in а) we prove that
P A1B1C1 = | d |2 P ABC .
The statement of the theorem is implied by the theorems 4.6 and 7.8. ■

12.12. Consequence. The ratio between the areas of two similar n-gons is
equivalent to the ratio of the squares of the respected sides.
Proof. Theorems 12.6 and 12.11 directly imply the above given consequence. ■

126
13. INCIRCLES AND EXCIRCLES OF A TRIANGLE

13.1. Let’s consider the  ABC whose vertices A, B and C have affixes a, b and
c, respectively. In lemma 10.4 we proved that any point D with an affix d is uniquely
determined by the real numbers λ, µ, ν such that
λa + µb + νc = d (1)
where
λ + µ + ν = 1 . (2)
Let P, P1 , P2 , P3 be the areas of the
triangles ABC, DBC, DAC and DBA, taken with an
appropriate sign depending on the orientation of
the triangles (figure 36). From (1) and (2) and the
conjugated equation of (1) we obtain the system:
λ + µ + ν = 1

λa + µb + νc = d

 λ a + µb + ν c = d
The solution of the above system is Figure 36
P P P
λ = P1 , µ = P2 , ν = P3 . (3)
So, we proved the following lemma.

Lemma. The numbers λ, µ and ν which according to the formulae (1) and (2)
determine the position of the point D with respect to the  ABC are proportional to the
areas of P1 , P2 and P3 , i.e. they are determined by the relation (3). ■

13.2. Let be given a point I, with an affix z and let the distances between the point
I to the sides BC, CA and AB of the  ABC be r1 , r2 and r3 respectively. The lemmas
13.1 and 10.4 imply that
2µP
r1 = |2bλ−Pc | , r2 = | c −a | , r3 = | 2aν−Pb | , (4)
where the numbers λ, µ and ν are uniquely determined. Clearly, if
| b −c | | c −a | | a −b |
λ = | a −b |+| b −c |+| c −a | , µ = | a −b |+| b −c |+| c −a | , ν = | a −b |+| b −c |+| c −a | ,
then
r1 = r2 = r3 = r = | a −b |+| b2−Pc |+| c −a |
and for the affix of the point I we get that
| b −c |⋅a +| c − a |⋅b +| a −b |⋅c
z= | a −b |+| b −c |+| c − a |
. (5)
Since the numbers λ, µ and ν are positive numbers, the remark10.4 implies that the point
I is inside the  ABC. So, we proved the following lemma.

127
Lemma. For any triangle, there is a unique point inside of it which is on the same
distance from the sides of the triangle. ■

13.3. Remark. Clearly, the circle with a center in I, whose affix is given in (5),
and radius
r = | a −b |+| b2−Pc |+| c −a |
touches the sides of the  ABC, i.e. it is an incircle of the  ABC.

13.4. Let A ', B ', C ' be the points in which the circle K(I,r) meets the sides BC,
CA, AB of the  ABC. Then, using the degree of the
points A, B, C with respect to the circle K(I,r) we get
that
AC ' = AB ', BC ' = BA ', CA ' = CB '
thus
  
' ⋅ CA
AB ' ⋅ BC
' = 1.
B 'C A ' B C'A
Now, according to the Ceva theorem we get that
the lines AA ', BB ' and CC ' intersect at a point M
Figure 37 (figure 37). So, we proved the following lemma.

Lemma. The lines which connect the vertices A, B, C of  ABC to the points
A ', B ', C ' at which the sides meet the incircle K(I,r) intersect in a point M, which is
called a Gergonne point for  ABC.

13.5. Remark. Analoguosly, we prove that there exist three circles which are
excircle of  ABC. The affixes of their centers are as following
−| b −c |⋅a +| c − a |⋅b +| a −b |⋅c | b −c |⋅a −| c − a |⋅b +| a −b |⋅c | b −c |⋅a +| c − a |⋅b −| a −b |⋅c
z ' = −| a −b |+| b −c |+| c −a | , z '' = | a −b |−| b −c |+| c −a | , z = | a −b |+| b −c |−| c −a | ,
their radii are
r ' = −| a −b |+|2bP−c |+| c −a | , r '' = | a −b |−| b2−Pc |+| c −a | , r ''' = | a −b |+| b2−Pc |−| c −a | ,
(figure 38), respectively.

Figure 38

128
13.6. Lemma. The lines which connect the vertices A, B, C of the  ABC to K, F,
L the points where the sides of the triangle meet the excircle concur in a point N which is
called Nagel point for the  ABC.
Proof. Firstly we are going to prove that
BF ' = BM = AB + BC
2
+CA

(figure 39). There is

BF ' = BA + AF ' = BA + AF
and
BM = BC + CM = BC + CF .
If we add the last two equations and take that
AC = AF + CF
we get the required equality. According to this, Figure 39

AF = AF ' = BF ' − BA = AB + BC
2
+CA − BA = − AB + BC +CA .
2
Analogously to this, we prove that
CK = LA = AB + BC
2
−CA , BL = FC = AB − BC +CA , KB = − AB + BC +CA .
2 2
Now, the statement in Lemma is implied from Ceva`s theorem, thereby
  
AF
 ⋅ CK
 ⋅ BL
 = 1. ■
FC KB LA

13.7. Remark. Clearly, the lines AI, BI and CI are bisectors of the inside angles
of the  ABC and their equations are
| c − a |⋅(b − a ) +| a −b |⋅(c − a )
z−a= ( z − a) ,
| c − a |⋅(b − a ) +| a −b |⋅(c − a )
| b −c |⋅( a −b ) +| a −b |⋅(c −b )
z −b = ( z − b) and
| b −c |⋅( a −b ) +| a −b |⋅(c −b )
| b −c |⋅( a −c ) +| c − a |⋅(b −c )
z −c = ( z − c) ,
| b −c |⋅( a −c ) +| c − a |⋅(b −c )
respectively. Analogously, we can determine the bisectors of the outside angles of the
 ABC.

13.8. Let A1 be the point where the bisector AI meets the side BC. Its affix is
a1 = b + λc
. Since the points A, I and A1 are collinear, by applying the consequence II 1.3
1+ λ
we get that
| c − a |⋅(b − a ) +| a −b |⋅(c − a ) b −a +λ (c −a )
= .
| c − a |⋅(b − a ) +| a −b |⋅(c − a ) b − a + λ(c − a )
The latter is equivalent to
(λ − )( | a −b |
| c −a |
c −a
c −a
− b−a = 0
b−a )
129
| a −b |
Now, since the points A, B and C are not collinear we get that λ = | c −a | , i.e. the affix of
the point A1 is
| c − a |⋅b +| a −b |⋅c
a1 = | c −a |+| a −b | .

Analogously, the affixes of B1 and C1 the points where the bisectors BI and CI meet the
sides AC and AB are
| b −c |⋅a +| a −b |⋅c | c − a |⋅b +| b −c |⋅a
b1 = | b −c |+| a −b | and c1 = | c −a |+| b −c | ,
respectively. Thus
BA1 = AB⋅BC , CA1 = AC ⋅BC (6)
CA+ AB CA+ AB
i.e.
BA1
= AB . (7)
CA1 AC
Therefore, we proved the following lemma.

Lemma. If A1 is the point where AI, the bisector of the inside angle at the vertex
A of the  ABC, meets the side BC, then the equalities (6) and (7) are satisfied. ■

13.9. Remark. Clearly, the analogous equalities to the equalitity (6) hold true
for the bisector BI and CI of the inside angles at the vertices B and C of the  ABC.
Furthermore, if B1 and C1 are the points of where the bisector meets the sides CA and
AB respectively, then
AB1 AC1 CA
= BA and = .
CB1 BC BC1 CB

13.10. Theorem (Euler). Let O be I the circumcenter and the incenter of the
 ABC, and R and r be their radii, respectively. Then,
2
OI = R 2 − 2 Rr .
Proof. Without loss of generality, we can say that the center of the circumscribed
circle coincides with the origin. If a, b, c are the affixes of the vertices of the  ABC, then
| a | = | b | = | c | = R and according to the proof of lemma 13.2 the affix of the incenter is
as following
| b −c |⋅a +| c − a |⋅b +| a −b |⋅c
z= | a −b |+| b −c |+| c − a |
.
Therefore,
2 (| b −c |⋅a +| c − a |⋅b +| a −b |⋅c )(| b −c |⋅a +| c − a |⋅b +| a −b |⋅c )
OI = .
(| a −b |+| b −c |+| c − a |)2
Now, the statement in Lemma is directly implied by the operations of complex numbers,
the formula for the radius of circumcircle (discussed in 12.10) and the formula for the
radius of incircle (discussed in Remark 13.3)

130
13.11. Remark. Analogously, as in theorem 13.10 it can be proved that
2
OI ' = R 2 + 2 Rr ' , where I ' and r ' are the center and the radius of the excircle of the
 ABC, аnd O and R are the center and the radius of the inscribed circle of the  ABC.

13.12. Remark. Let the unit circle be inscribed in the  ABC whose vertices A,
B, C have the affixes a, b, c, respectively, and let it meets the sides BC, CA, AB at the
points P, Q, R with affixes p, q, r, respectively. Then, according to the remark II 3.12 d)
we get that
2 qr 2 rp 2 pq
a = q + r , b = r + p and c = p + q .
Furthermore, the example II 3.3 implies that the circum center of  ABC has the affix
2 pqr ( p + q + r )
o = ( p + q )( q + r )( r + p ) ,
The remark 5.4 implies that the orthocenter of the  ABC has an affix
2( p 2 q 2 + q 2 r 2 + r 2 p 2 + pqr ( p + q + r ) )
h= ( p + q )( q + r )( r + p )
.

13.13. At the end of this part let’s note that the following theorem can be proved.

Theorem. Let  ABC, whose vertices A, B and C have the affixes a, b and c,
respectively, be inscribed in the unit circle. Then, there are complex numbers u, v, w such
that a = u 2 , b = v 2 , c = w2 and the midpoints of the arcs AB, BC, CA which don’t consist
of the points C, A, B are points with affixes − uv, − vw, − wu , respectively. Therefore,
the affix of the incenter L of the  ABC is l = −(uv + vw + wu ) . ■

13.14. Example. Let L be the incenter of the  ABC, and the lines AL, BL, CL
meets the circums circle of the  ABC at the points A1 , B1 , C1 , respectively. If R is the
radius of the circumscribed, and r the radius of the inscribed circle of the  ABC prove
that:
LA1⋅LC1 P
а) = R , b) LA⋅LB = 2r , c) P ∆ABC = 2Rr .
LB LC1 ∆A1B1C1

Solution. Let the circumscribed circle of the  ABC be the unite circle and let u,
v, w are the complex numbers as in the theorem 13.13. According to that theorem, we get
that l = −(uv + vw + wu ) and a = u 2 , b = v 2 , c = w2 and the midpoints of the arcs AB,
BC, CA which don’t consist of the points C, A, B are points with affixes − uv, − vw, − wu ,
respectively. Furthermore, since,
l −( − vw) a −( − vw) u 2 + vw
= −uv1 −uw = vwu 2 and = 1 1 = u 2 vw
l −( − vw) 1
− uv − uw a −( − vw) + vw
u2
we get that the points with affixes a, l and −vw are collinear, which means that the point
A1 has an affix a1 = −vw . Similarly, the affixes of the points B1 and C1 are b1 = −uw
and c1 = −uv , respectively.

131
а) the statement is implied by the equality
LA1⋅LC1 | l − a1|⋅| l −c1| | u (v + w)|⋅|w(u + v )| | v + w |⋅| u + v |
= | l −b |
= = |(u +v )( w+v )| = 1 = R .
LB | uv +uw+ vw+ v 2 |

b) If z is the affix of the point where the incircle meets the side BC, then z is the
affix of the foot of the perpendicular drawn from the point L to the side BC. Thus its affix
is
z = 12 (b + c + l − bcl )
thus
(u + v )(v + w)( w+u )
r = | l − z | = 12 u
= 12 | (u + v)(v + w)( w + u ) | .
Therefore,
LA⋅LB |(u + v )(u + w)|⋅|(u + v )(v + w)|
= |w(u + v )|
LC1
= | (u + v)(v + w)( w + u ) | = 2r.
c) The areas of the triangles are
u2 1/ u 2 1 vw u 1
P ABC = ± 4i v 2
1/ v 2
1 and i
P A1B1C1 = ± 4uvw uw v 1 ,
w2 1 / w2 1 uv w 1

thus
P∆ABC 4 2 4 2 4 2 4 2
− u 4 v 2 − w 4u 2
P∆A1B1C1
= u w 2+ w v 2+v u2 −v w
2 2 2
v w+ vu +uw −uv −vw − wu

(v −u 2 )(v 2u 2 + w4 − w2u 2 − w2v 2 )

2
=
(u −v )(uv + w2 − wu − wv )

(u −v )(u + v ) (uv + w2 )2 −( wu + wv )2 
=
(u −v )(uv + w2 − wu − wv )

(u −v )(u + v )(uv + w2 − wu − wv )(uv + w2 + wu + wv )

=
(u −v )(uv + w2 − wu − wv )

= | (u + v)(uv + w2 + wu + wv) |
= | (u + v)(v + w)( w + u ) | = 2Rr ,
which was required to be proven. ■

132
14. STEWART`S THEOREM

14.1. Let a, b, c be the affixes of the vertices A, B, C of the  ABC and let
the point D be on the side BC (figure 40). Then, the affix of D is d = b1++λλc , for some
λ ∈ (0,1) . According to this,
AB = | b − a |, AC = | c − a |, BC = | b − c |,
λ | b −c | | b −c | | b − a + λ (c − a )|
BD = 1+λ , CD = 1+λ , AD = 1+ λ
Thus
2 2 | b −c |
AC ⋅ BD + AB ⋅ CD − BC ⋅ CD ⋅ BD = 1+λ (λ | c − a |2 + | b − a |2 − 1+λλ | b − c |2 )
λ 2 (c − a )(c − a ) + (b − a )(b − a ) + λ (c − a )(c − a ) + λ (b − a )(b − a ) −λ (b −c )(b −c )
= |b−c|
(1+ λ )2
λ 2 (c − a )(c − a ) + (b − a )(b − a ) + λ (c − a )(b − a ) + λ((b − a )(c − a )
= |b−c|
(1+ λ )2
| b − a + λ (c − a )|2 2
= |b−c| = AD ⋅ BC.
(1+ λ )2

So, we proved the following theorem.

Theorem (Stewart). If D is a point on the side BC of the  ABC, such that O is

between the points B and C, the following equality holds true
2 2 2
AC ⋅ BD + AB ⋅ CD − BC ⋅ CD ⋅ BD = AD ⋅ BC. ■ (1)

14.2. Example. Let m A , mB , mC be the lengths of the medians of the  ABC

drawn from the vertices A, B, C respectively. Prove that
2 2 2 2 2 2 2 2 2
m 2A = AC 2+ AB − BC
4
, mB2 = BC 2+ BA − AC 4
, mC2 = CA +2CB − AB 4
. (2)
Solution. We shall only prove the first equality in (2). The other two equalities
can be proved analogously.
Let A' be the midpoint of the side BC. Since
CA ' = BA ' = BC
2
and m A = AA ' ,
By applying the Stewart theorem, we get the following equality
2 2
m 2A ⋅ BC = AC ⋅ BC
2
+ AB ⋅ BC
2
− BC ⋅ BC
2
⋅ BC
2
If we divide it by BC we get the following equality
2 2 2
m 2A = AC 2+ AB − BC
4
.■

14.3. Example. The lengths of the medians of the  ABC are m A = 9 cm,
mB = 12 cm and mC = 15 cm . Calculate the lengths of its sides.

133
Solution. From the equalities (2) using the length of the medians we can find the
length of the sides of the  ABC.
2 8( m 2 + m 2 ) − 4 m 2 2 8( mB2 + mC2 ) − 4 m 2A 2 8( mC2 + m 2A ) − 4 mB2
AB = A
9
b C
, BC = 9
, CA = 9
. (3)
Using the equality (3) we get that
AB = 10 cm, BC = 2 73 cm and CA = 4 13 cm . ■

14.4. Example. Express the length of the bisectors of the interior angles of the
 ABC using the length of its sides.
Solution. Let D be the point where, the bisector l A of the angle at the vertex A,
meets the side BC. In lemma 13.8 we proved that
BD = AB⋅BC and CD = AC ⋅BC .
CA+ AB CA+ AB
According to the Stewart theorem, we get that
 2

l A2 = AB ⋅ AC 1 − BC 2  .
 ( AB + AC ) 
It can be proved analogously that
 2
  2

lB2 = AB ⋅ BC 1 − AC 2  and lC2 = AC ⋅ BC 1 − AB 2  . ■
 ( AB + BC )   ( AC + BC ) 

14.5. Definition. The line symmetrical to the median with respect to the bisector
of the angle drawn from the same vertex is called symmedian.

14.6. Let AA1 , AA ' and AA'' be the bisectors of the angle, the median and the
symmedian of the  ABC, drawn from the vertex A (figure 41). The equations of the
median and the symmetry are the following
| c − a |(b − a ) +| a −b |(c − a )
z − a = b +c −2a ( z − a ) and z − a = ( z − a) ,
b+c −2a | c − a |(b − a ) +| a −b |(c − a )
respectively.
The indirect similarity
| c − a |(b − a ) +| a −b |(c − a )
S ( z) = ( z − a ) + a (4)
| c − a |(b − a ) +| a −b |(c − a )

is a line symmetry and the line of symmetry is a bisector

of the angle at the vertex A by direct calculations we get
that the line with an equation
| c − a |2 (b − a ) +| a −b |2 (c − a )
z−a= ( z − a) (5)
| c − a |2 (b − a ) +| a −b |2 (c − a )
is an image of the median е under the line symmetry (4).
According to this, (5) is the equation of the symmedian
Figure 41 drawn from the vertex A. A'' the point where the

134
symmedian and the side BC has the affix a '' = b1++λλc . But, the point A'' lies on the line
(5), thus the following equality is satisfied
| c − a |2 (b − a ) +| a −b |2 (c − a ) b −a +λ (c −a )
2 2
=
| c − a | (b − a ) +| a −b | (c − a ) b − a + λ(c − a )
which is equivalent to the equality


| a −b |2  c − a b − a
 λ − | c −a |2  c −a − b −a = 0

( )
| a −b |2
and since the points A, B and C are not collinear we get that λ = , i.e. the affix of
| c − a |2
the point A'' is
| c − a |2 b +| a −b |2 c
a '' = .
| c − a |2 +| a −b |2
Analogously to this, for the points of intersection B '' and C '' of the other symmedians
and the sides AC and AB we get that
| b −c |2 a +| a −b |2 c | c − a |2 b +| b −c |2 a
b '' = and c '' = ,
| b −c |2 +| a −b |2 | c − a |2 +| b −c |2
respectively.
The above stated impies that, thus,
2 2
BA '' = AB2 ⋅BC2 and CA '' = AC2 ⋅BC2 (6)
so AB +CA AB +CA
2
BA '' = AB2 . (7)
CA '' CA
So we proved the following lemma.

Lemma. If A'' is the point of intersection between the symmedian at the vertex
A of the  ABC and the side BC, then the equalities (6) and (7) are satisfied. ■

14.7. Consequence. The symmedians AA '', BB '' and CC '' of the  ABC are
concurrent.
Proof. The proof is directly implied by lemma 14.6 and the Ceva theorem. ■

14.8. Example. Express the lengths of the symmedians of the  ABC using the
lengths of the sides of the triangle.
Solution. By using the equalities (6), and the Stewart theorem for the length of
the symmedian AA'' we get the following
2 2 2 BC
2

A '' A = AC ⋅ AB  2 − 2 2 2 
.
 ( AC + AB ) 
Analogously to this, for the symmedians BB '' and CC '' the following holds true
2 2 2 AC
2
 2 2 2 AB
2

B '' B = BC ⋅ AB  2 − 2 2 2  and C '' C = AC ⋅ BC  2 − 2 2 2 
.■
 ( BC + AB )   ( AC + BC ) 

135
15. SIMPSON LINE

15.1. ������������������
Theorem (Simpson). Let D be point on the circumcircle of  ABC. Then
the foots of the perpendiculars drawn from the point D to the sides of the  ABC are
collinear.
Proof. With no loss of generality, we can
say that the  ABC is inscribed in the unit circle. Let
a ', b ', c ' be the affixes of A ', B ', C ' the foots of
perpendicular drawn from the point D to the sides BC,
CA and AB, respectively (figure 42). If we take that
a = 1a , b = b1 , c = 1c and d = d1 ,
Then the example 1.9 implies that
(
a ' = 12 b + c + d − bc
d ) (
, b ' = 12 c + a + d − ac
d
and )
1
( ab
c' = 2 a + b + d − d . )
Thus,
Figure 42
a '−c ' =
b + c + d − bc
d ( d) = (c−a)(d −b) = (a−c)(b−d )
− a +b + d − ab
b '−c ' c + a + d − ac
d ( a+b+d − d ) (c−b)(d −a) (b−c)(a−d )
− ab

(c − a )(b − d ) b + c + d − d −( a +b + d − d ) a '−c '

bc ab
= (c −b)( d −a ) = = b '− c ' ,
d ( d )
c + a + d − ac − a +b + d − ab

thus the consequence 1.4 implies that the points A ', B ', C ' are collinear. ■

15.2. Definition. The line such that the points A ', B ', C ' (defined as in theorem
15.1) lay on it, is called Simpson line of the point D with respect to the  ABC.

15.3. We shall give the equation of Simpson line for the point D about the  ABC.
If Z is an arbitrary point on the Simpson line, then the points Z , A ', C ' are collinear,
therefore az'−−cc'' = z −c ' , and for the Simpson line we get that
a '−c '
z − z a '−c ' + a ' c '−c ' a ' = 0 .
a '−c ' a '−c '
Furthermore, according to the proof of theorem 15.1 we get that
( )
a '− c ' = 12 (c − a ) 1 − db , a ' − c ' = 21ac (a − c) 1 − db , ( )
therefore a '−c ' = acb
d
. In order to determine the constant term m = a ' c '−c ' a ' we shall take
a '−c ' a '−c '
that the point B ' is on the Simpson line, i.e.
1
2 ( c + a + d − acd ) − 12 ( c + a + d − acd ) acbd + m = 0 .
So,

136
m = abc
2d
(a + b + c + d ) − 12 (a + b + c + d )
thus the equation of Simpson line of the point D with respect to the  ABC is
z − z acb
d
+ abc
2d
(a + b + c + d ) − 12 (a + b + c + d ) = 0 . (1)

15.4. Example. Let the points A, B, C, D lie on a same circle. Prove that the point
X where the Simpson line of A for the  BCD meets the Simpson line of B for the  ACD
is on the line which passes through the point C and the orthocenter H of the  ABD.
Solution. The quadrilateral ABCD is cyclic, so we can say that it is inscribed in
a unit circle. Let a ', a '', a ''' be the affixes of A ', A '', A ''' the foot of the perpendiculars
drawn from the point A to the lines BC, CD, DB, respectively, and b ', b '', b ''' are the
affixes of the food of perpendicular B ', B '', B ''' of the normal lines drawn from the point
B of the lines AC, CD, DA, respectively. So,
(
a ' = 12 a + b + c − bc
a ) (
, a '' = 12 a + b + d − bd
a )
, a ''' = 12 a + c + d − cd
a (
, )
b ' = 12 ( b + a + c − ac
b )
, b '' = 12 ( b + c + d − cd
b )
, a ''' = 12 ( b + d + a − ad
b )
.
The equation of Simpson line of the point A is: z − a ' = a ''−a ' ( z − a ') , i.e.
a ''− a '
z − a ' = bcd
a
( z − a ') , (2)
and the equation of Simpson line of the point B is: z − b ' = b ''−b ' ( z − b ') , i.e.
b ''−b '
z − b ' = acd
b
( z − b ') . (3)
By solving the system of equations (2) and (3) we get the affix x of the point of intersec-
tion X, as following
x = 12 (a + b + c + d ) .
Moreover, the affix of the orthocenter H of the  ABD is h = a + c + d and since
1 ( a +b + c + d ) −c
h −c = a +b + d −c = a +b +c + d −2c = 12 = x −c ,
h −c a +b + d −c a +b + c + d − 2c 2
( a +b + c + d ) −c x −c
According to the consequence II 1.3 we get that the points C, H and X are collinear. ■

15.5. Example. Let l(N,PQR) be the SImson line of the point N of the  PQR.
Let the points A, B, C, D be on a same circle. Prove that the lines l(A,BCD), l(B,ACD),
l(C,ABD), l(D,ABC) concur.
Solution. We shall say that the points A, B, C, D lie on a unit circle. According to
the example 15.4 the lines l(A,BCD) and l(B,ACD) intersect at the point X with an affix
x = 12 (a + b + c + d ) . The right part of the last equality is symmetrical with respect to the
affixes a, b, c, d of the points A, B, C, D, thus the point X is the point of intersection of the
Simpsons lines l(A,BCD), l(B,ACD), l(C,ABD), l(D,ABC). ■

137
15.6. Example. Let l(P,ABC) and l(Q,ABC) be the Simpsons lines of the points
P and Q with respect to the  ABC and O be the center of the circum circle of the  ABC.
Prove that ∠ ( l ( P, ABC ), l (Q, ABC ) ) = 12 ∠ POQ .
Solution. We shall say that  ABC is inscribed in the unit circle. 15.3 implies that
the equations of the lines l(P,ABC) and l(Q,ABC) are
z − z acb
p
+ abc
2p
(a + b + c + p ) − 12 (a + b + c + p ) = 0 and

z − z acb
q
+ abc
2q
(a + b + c + q ) − 12 (a + b + c + q ) = 0 ,
respectively. The complex angle coefficients of the lines l(P,ABC) and l(Q,ABC) are
η = acb
p
and η' = acb
q
, respectively. According to theorem II 1.7 the oriented angleϕ
η q
between the lines l(P,ABC) and l(Q,ABC) is given by the formula e 2iϕ = η ' = p , which
according to I 8.7 means that
∠ ( l ( P, ABC ), l (Q, ABC ) ) = 12 ∠ POQ . ■

16. PTOLEMY`S THEOREM

16.1. Lemma. If z j , j = 1, 2, 3, 4 are the affixes of the consecutive vertices of a

cyclic quadrilateral, then
( z1 − z2 )( z3 − z4 )
( z1 − z4 )( z2 − z3 )
> 0 (1)

Proof. With no loss of generality we shall say that the center of the circum circle
iϕ
coincides with the origin, and the radius of the circle is r. Then, z j = re j , j = 1, 2, 3, 4 .
Also, we can assume that consecutively of the vertices is equivalent to the condition
ϕ1 < ϕ2 < ϕ3 < ϕ4 < ϕ1 + 2π . (2)
Thus, it holds true that:
( z1 − z2 )( z3 − z4 ) (eiϕ1 −eiϕ2 )(eiϕ3 −eiϕ4 )
( z1 − z4 )( z2 − z3 )
=
(eiϕ1 −eiϕ4 )(eiϕ2 −eiϕ3 )

(
= e
ϕ −ϕ
i 1 2
2 −e
ϕ −ϕ
−i 1 2
2 )(e ϕ −ϕ
i 3 4
2 −e
ϕ −ϕ
−i 3 4
2 )
(e ϕ −ϕ
i 1 4
2 −e
ϕ −ϕ
−i 1 4
2 )(e ϕ −ϕ
i 2 3
2 −e
ϕ −ϕ
−i 2 3
2 )
ϕ1 − ϕ2 ϕ3 − ϕ4
sin sin
= 2
ϕ −ϕ
2
ϕ −ϕ
>0
sin 1 2 4 sin 2 2 3

therefore according to (2), each argument of sin belongs to the interval (−π, 0) . Thus the
inequality (1). is proved. ■

138
16.2. Theorem (Ptolemy). The product of the
lengths of the diagonals of a cyclic quadrilateral is equal
to the sum of the products of its lengths of the opposite
sides.
Proof. Let z j , j = 1, 2, 3, 4 be the affixes of the
vertices A, B, C, D of a cyclic quadrilateral ABCD, figure
43. The statement of the theorem is equal to the equality
AC ⋅ BD = AB ⋅ CD + BC ⋅ AD ,
i.e. the equality Figure 43
| z1 − z3 | ⋅ | z2 − z4 | = | z1 − z2 | ⋅ | z3 − z4 | + | z1 − z4 | ⋅ | z2 − z3 | . (3)
According to the lemma 26.1 we have that
| ( z1 − z2 )( z3 − z4 ) | + | ( z1 − z4 )( z2 − z3 ) | = | ( z1 − z2 )( z3 − z4 ) + ( z1 − z4 )( z2 − z3 ) |
= | − z1 z4 − z2 z3 + z1 z2 + z3 z4 |
= | ( z1 − z3 )( z2 − z4 ) |,
which means that the equality (3) is satisfied. ■

( p − s )( r − q )
16.3. Theorem. If for the complex numbers p, q, r, s it is true that ( p −q )( r − s ) ∈ R ,
then they are the affixes of the consecutive vertices P, Q, R, S of a cyclic quadrilateral
(cyclic) or they are collinear.
Proof. In extended complex plane the points Q, R, S, determine a circle (K).
A Möbius transformation determined by f (q ) = ∞ , f (r ) = 1, f ( s ) = 0 maps the circle
( z − s )( r − q )
(K) on the real axis and thereby theorem II 10.8 it is defined by f ( z ) = ( z −q )( r − s ) . The
point P is on the circle (K) if and only if its image is on the real axis i.e. if and only if
( p − s )( r − q )
f ( p ) = ( p −q )( r − s ) ∈ R . ■

16.4. Example. An equilateral triangle ABC is inscribed in a circle. An

 on which doesn’t belong the point A. Prove that
arbitrary point M is on the arc BC
BM + CM = AM .
Solution. By applying the Ptolemy`s theorem on the cyclic quadrilateral ABMC
we get that
BM ⋅ CA + CM ⋅ AB = BC ⋅ AM . (4)
But, the triangle ABC is equilateral, AB = BC = CA , so thereby (4) holds true we get
that
BM ⋅ AB + CM ⋅ AB = AB ⋅ AM
and if we divide the last equality by AB we get the necessary equality. ■

139
16.5. Example. Given are circles k1 , k2 , k3 , k4 so required that
k1 ∩ k2 = { A1 , B1} , k2 ∩ k3 = { A2 , B2 }, k3 ∩ k4 = { A3 , B3} and
k4 ∩ k1 = { A4 , B4 } .
If the points A1 , A2 , A3 , A4 are concyclic (they lie on a same circle too) or are collinear,
then the points B1 , B2 , B3 , B4 are concyclic or collinear. Prove it!
Solution. The points
A1 , B1 , A2 , B2 ; A2 , B2 , A3 , B3 ; A3 , B3 , A4 , B4 and A4 , B4 , A1 , B1
are concyclic, so thereby lemma 25.1 we get that the numbers
( a1 − a2 )(b2 −b1 ) ( a − a )(b −b ) ( a − a )(b −b ) ( a − a )(b −b )
, ( a2 −b3 )(b3 −a2 ) , ( a3 −b4)(b 4−a3 ) , ( a4 −b1 )(b1 −a4 )
( a1 −b1 )(b2 − a2 )
2 2 3 3 3 3 4 4 4 4 1 1
are real numbers. The product of the first and the third number divided by the product of
the second and the forth number in (1) is equal to
( a1 − a2 )( a3 − a4 ) (b2 −b1 )(b4 −b3 )
⋅
( a2 − a3 )( a4 − a1 ) (b3 −b2 )(b1 −b4 )
and it is a real number too. According to the condition of the example, the points
( a − a )( a − a )
A1 , A2 , A3 , A4 are concyclic, so from the lemma 16.1 we get that ( a1 −a2 )( a3 −a4 ) is a real
2 3 4 1
(b −b )(b −b )
number. But, that means that the number (b2 −b1 )(b4 −b3 ) is a real number, so the theorem
3 2 1 4
16.3 we get that the points B1 , B2 , B3 , B4 are concyclic or collinear. ■

16.6. Lemma (Ptolemy`s inequality). For arbitrary points A, B, C, D in a plane,

the following inequality holds true
AB ⋅ CD + BC ⋅ AD ≥ AC ⋅ BD . (5)
Proof. Let a, b, c, d be the affixes of the vertices A, B, C, D respectively. Then
(a − b)(c − d ) + (b − c)(a − d ) = (a − c)(b − d )
and if we apply the triangle inequality, we get the following inequality
| (a − b)(c − d ) | + | (b − c)(a − d ) | ≥ | (a − c)(b − d ) | ,
which is actually the inequality (5) written in terms of the affixes of the vertices of the
quadrilateral ABCD. ■

17. INNER PRODUCT

Let the complex numbers a = | a | eiα and b = | b | eiβ be given. Then

17.1. ������������������������
ab = | a | e−iα | b | eiβ = | a | ⋅ | b | ei (β−α ) = | a | ⋅ | b | ( cos(β − α) + i sin(β − α) ) ,
so
| a | ⋅ | b | cos(β − α) = Re ab and | a | ⋅ | b | sin(β − α) = Im ab .  
Furthermore, the complex numbers a and b correspond to the vectors a and b with
tale at the origin and held at the points A and B with affixes a and b . Moreover, the

140
      
scalar
  product of the vectors a and b is defined by ab = | a | ⋅ | b| cos ∠( , b) and since
a
∠(a, b) = ± (β − α) we get that the scalar product of the vectors a and b (the complex
numbers a and b) holds true that ab = Re ab ( a ⋅ b = Re ab ). The following properties of
a scalar product are very simple, and therefore they are left on the reads.

Theorem. The following statements hold true:

1) a ⋅ a = | a |2 , for each a ∈C ,
2) a ⋅ b = b ⋅ a , for all a, b ∈C ,
3) a ⋅ (b + c) = a ⋅ b + a ⋅ c , for all a, b, c ∈C ,
4) (ka ) ⋅ b = k (a ⋅ b) = a ⋅ (kb) , for all a, b ∈C and for each k ∈ R ,
5) (ac) ⋅ (bc) = | c |2 a ⋅ b , for all a, b ∈C , and
6) a ⋅ b = 0 if and only if OA ⊥ OB . ■

17.2. Lemma. Let A, B, C, D be four distinct points with affixes a, b, c, d,

respectively. Then AB ⊥ CD if and only if (b − a ) ⋅ (d − c) = 0 .

 Proof. The complex numbers b − a and d − c correspond to the vectors AB and
CD . The statement directly implies the statement 6) of the previous theorem. ■

17.3. Example. Let O be the circumcenter of the  ABC, C ' be the

midpoint of the side AB and T be the centroid. Prove that OT ⊥ CC ' if and only if
2 2 2
BC + AC = 2 AB .
Solution. With no loss of generality, we shall say that the triangle is inscribed in
the unit circle. Then OT ⊥ CC ' if and only if
t ⋅ (c '− c) = 0 ⇔

( )
a +b + c ⋅ a +b − c = 0 ⇔
3 2
(a + b + c) ⋅ (a + b − 2c) = 0 ⇔
| a |2 + | b |2 − 2 | c |2 + 2a ⋅ b − a ⋅ c − b ⋅ c = 0 ⇔
2a ⋅ b − a ⋅ c − b ⋅ c = 0 .
Furthermore,
2 2 2
BC + AC − 2 AB = | c − b |2 + | a − c |2 − 2 | a − b |2
= (c − b) ⋅ (c − b) + (a − c) ⋅ (a − c) − 2(a − b) ⋅ (a − b)
= 2 | c |2 − | a |2 − | b |2 + 4a ⋅ b − 2a ⋅ c − 2b ⋅ c
= 2(2a ⋅ b − a ⋅ c − b ⋅ c).
Therefore, OT ⊥ CC ' if and only if 2a ⋅ b − a ⋅ c − b ⋅ c = 0 if and only if
2 2 2
BC + AC = 2 AB . ■

141
17.4. Theorem (Apollonius). Let M be a point on the side BC of the triangle
ABC such that BM : MC = m : n . Then
2 2 2 2 2
n AB + m AC = mCM + nBM + (m + n) AM .
Proof. Let a, b, c, z be determine the affixes of the points A, B, C, M, respec­tively.
Then z = nbm++mc
n
, thus
2
AB = | b − a |2 = (b − a ) ⋅ (b − a ) = | a |2 − 2a ⋅ b + | b |2 ,
2
AC = | c − a |2 = (c − a ) ⋅ (c − a ) = | a |2 − 2a ⋅ c + | c |2 ,
2 n2 n2
CM = | b − c |2 = (| b |2 − 2b ⋅ c + | c |2 ),
( m+ n)2 ( m+ n)2
2 m2 m2
BM = | b − c |2 = (| b |2 −2b ⋅ c + | c |2 ),
( m+ n)2 ( m+ n)2
2 n2 2
AM = | a |2 + | b |2 + m | c |2 + 2mn 2 b ⋅ c − m2+nn a ⋅ b − m2+mn a ⋅ c.
( m+ n)2 ( m+ n)2 ( m+ n)
So,
2 2 2
mCM + nBM + (m + n) AM = mmn
+n
| b |2 + mmn
+n
| c |2 − 2 mmn
+n
b⋅c
2 2
+ (m + n) | a |2 + mn+ n | b |2 + mm+ n | c |2
2 mn b ⋅ c − 2na ⋅ b − 2ma ⋅ c
+m +n
= m(| a |2 −2a ⋅ c + | c |2 ) + n(| a |2 −2a ⋅ b + | b |2 )
2 2
= n AB + m AC . ■

17.5. Let’s consider distinct points Ai , i = 1, 2,..., n in a plane with

affixes ai , i = 1, 2,..., n and let ki , i = 1, 2,..., n be nonzero numbers, such that
k1 + k2 + ... + kn = k ≠ 0 . A barycenter or centroid of the system composed of the points
Ai , i = 1, 2,..., n with centroids ki , i = 1, 2,..., n is called the point T with an affix
k1a1 + k2 a2 +...+ kn an
t= k1 + k2 +...+ kn
.
If ki =k1,i , i = 1, 2,..., n , then the point T is called the equibarycenter or centroid of the set
of points Ai , i = 1, 2,..., n .

17.6. Theorem (Lagrange). Let the points Ai , i = 1, 2,..., n be given and

the centroids ki , i = 1, 2,..., n , be non-zero, such that k1 + k2 + ... + kn = k ≠ 0 . If T
is the barycenter of the system composed of the points Ai , i = 1, 2,..., n and centroids
ki , i = 1, 2,..., n , then for each point M with affix z the following holds true:
n 2 n
∑ ki MA i2 = k MT + ∑ ki TA i2 . (1)
i =1 i =1

142
Proof . We have that:
( (t − z ) + (ai − t )) ⋅ ( (t − z ) + (ai − t ) ) = | t − z |2 + | ai − t |2 + 2(ai − t ) ⋅ (t − z ) ,
MA i2 =
thus
n n n n
∑ ki MA i2 = | t − z |2 ∑ ki + ∑ ki | ai − t |2 + 2(t − z ) ∑ ki (ai − t )
i =1 i =1 i =1 i =1
2 n  n  n
= k MT + ∑ ki TA i2 + 2(t − z ) ⋅  ∑ ki ai − t ∑ ki 
i =1  i =1 i =1 
2 n 2 n
= k MT + ∑ ki TA i2 + 2(t − z ) ⋅ 0 = k MT + ∑ ki TA i2 ,
i =1 i =1
i.e. the equality (1) holds true . ■

17.7. Consequence (Leibnez). Let the points Ai , i = 1, 2,..., n be given and let T
be the centroid. Then for any point M in the plane the following holds true
n 2 n
. ∑ MA i2 = nMT + ∑ TA i2 (2)
i =1 i =1
Proof. The equality (2) is implied directly from the equality (1) for ki = 1,
ki , i = 1, 2,..., n . ■

17.8. Comment а) If the points Ai , i = 1, 2,..., n be on a circle centered at O and

with radius R and if the point M corresponds with the center of the circle, we get the
formula
2 n
nR 2 = nOT + ∑ TA i2 .
i =1
b) If Ai , i = 1, 2,..., n are the vertices of a regular n-gon inscribed in the circle
| z | = R , then its centroid shall be the origin (why?). Now, according to the Leibniz
theorem we get that
n 2
∑ MA i2 = nMO + nR 2 ,
i =1
and if the point M is on the circumcircle of this polygon, then the previous formula
implies the following
n
∑ MA i2 = nR 2 + nR 2 = 2nR 2 .
i =1

143
IV CHAPTER
EXAMPLES AND EXERCISES

1. EXAMPLES (CHAPTER I)

1. ��������
Without ����
any ���������������
transformation �����
in a ��������������
trigonometric ������
form, �����
find ����
the set
���� of
��� the
���� second
�������
roots of the complex number z = a + ib. Hence, find the following
4 −7 + 24i.
2
Solution. Since ( x + iy ) = a + ib holds true, we obtain the following system of
equations
x 2 − y 2 = a, 2xy = b,
which implies
 2 2 2 2 
x + iy = ±  a + a2 +b + i − a + 2a +b  .
 
The fourth roots of −7 + 24i are the following 2 + i, − 2 − i, 1 − 2i, − 1 + 2i. ■

2. Determine the set of points z in the complex plane for which it exists a real number
c such that it satisfies the following z = 2cc−−ii .
Solution. Let z = x + iy . Since z = 2cc−−ii we get that
− y + ( x −1)i 2 x −1− 2iy (1− 2 x ) y + 2( x −1) y + (( x −1)( 2 x −1) + 2 y 2 ) i
c = 2ziz−−i1 = 2 x −1+ 2iy ⋅ 2 x −1−2iy =
( 2 x −1)2 + 4 y 2
Now, (2 x − 1) 2 + 4 y 2 ≠ 0 , thus x ≠ 12 and y ≠ 0 . Furthermore, it should be

( )
2
( x − 1)(2 x − 1) + 2 y 2 = 0 , i.e. x − 34 + y 2 = 16
1 .

So, the required set of points is the following

{ (
S = z = x + iy | x − 34 )
2
+ y 2 = 16 ( )}
1 , ( x, y ) ≠ 1 , 0
2

{
= z | | z − a | = 14 , a = 34 , z ≠ 12 . ■ }
3. Let a, b, c be complex numbers such that they satisfy the following
| a | = | b | = | c | = 1 . Prove the following
| ab + bc + ca | = | a + b + c | .
Solution. Since the condition of the given problem, it is true that aa = bb = cc = 1 ,
thus a = 1a , b = b1 , c = 1c . Hence,
| ab + bc + ca | = abc 1a + b1 + 1c( ) =| a | ⋅ | b | ⋅ | c | ⋅ | a + b + c |
= |a+b+c| = |a+b+c|. ■

144
4. If z and w are complex numbers such that they satisfy the following Re z > 0 and
Re w > 0, then z − w < 1. Prove that!
z+w
Solution. I solution. Let z = x + iy, w = a + ib. Then:
z −w | z − w| |( x − a ) + ( y −b )i | ( x − a ) 2 + ( y −b ) 2
= = |( x + a )+( − y +b)i | = .
z+w | z + w| ( x + a ) 2 + (b − y ) 2

If x > 0 and a > 0, then ( x − a ) 2 < ( x + a ) 2 , and ( y − b) 2 = (b − y ) 2 , the radicand

of the numerator is always smaller then the radicand of the denominator.
II solution. Since Re z > 0 , Re w > 0, and the properties of complex numbers, we
have that
| z − w |2 − | z + w |2 = ( z − w)( z − w) − ( z + w)( z + w)
= z z − zw − zw + ww − z z − zw − zw − ww
= −[ z ( w + w) + z ( w + w)]
= −( w + w)( z + z ) = −4 Re z ⋅ Re w < 0,
i.e. | z − w | < | z + w | , which implies that z − w < 1. ■
2 2
z+w

5. Prove the following inequality

n
(a1 + a2 + ... + an ) 2 + (b1 + b2 + ... + bn ) 2 ≤ ∑ ai2 + bi2
i =1
for ai , bi ∈ R , i = 1, 2,..., n .
Solution. Let consider complex numbers zi = ai + ibi , for i = 1, 2,..., n . By substitu-
tion in the inequality
| z1 + z2 + ... + zn | ≤ | z1 | + | z2 | +...+ | zn |
we obtain the required inequality. ■

5. Find the smallest possible value of the following expression z − 1 , z ∈C , if

z
given | z | = 2 .
Solution. If | z | = 2 , then
2 2 2 2
2 | z −1| | z −1| | z |−1 | z | −1
z − 1z = z z−1 = | z | = 2 ≥ 2 = 2 = 32 .
The above stated implies that the required minimal value is equal to 32 and it is achieved
for z 2 = 4 , i.e. z = ±2 . ■

6. Determine and show in a complex plane the set

{
z = 3tt−+ii : t ∈ R . }
Solution. Let
2
z = 3tt−+ii = 3t2 −1 + i 24t = x + iy.
t +1 t +1

145
2 y
Therefore, x = 3t2 −1 , y = 24t . Hence, t = 3− x , that is
t +1 t +1
( x − 1) 2 + y 2 = 4, x ≠ 3.
So, the required set is a circle centered at (1,0) and radius 2, except the point (3,0). ■

7. Solve the equation

2(1 + i ) z 2 − 4(2 − i ) z − 5 − 3i = 0 .
Solution. By solving the given quadratic equation for z we obtain the following
4( 2−i ) ± 16( 2−i )2 +8(1+i )(5+3i )
z= 4(1+i )
,
hence, z1 = 14+−ii = 3−25i and − i 1+
z2 = 1+i = − 2 . ■i

8. Determine all complex numbers z such that they satisfy the following
| z | = | 1z | = | z − 1 | .
Solution. Firstly, let’s notice that the given expressions are defined for z ≠ 0 . Since
| z | = | 1z | , we get that | z | = 1 and | z | = | z −1 | , imply that | z − 1 | = 1 . If z = x + iy , then
the above stated conditions imply that
x 2 + y 2 = 1 and ( x − 1) 2 + y 2 = 1 ,
that is x 2 = ( x − 1) 2 , thus x = 12 , and y = 23 or y = − 23 .
Finally, the solutions of the given equation are: 12 + i 23 and 12 − i 23 . ■

9. Solve the equation in a field of complex numbers

( x − 3) 4 + ( x − 4) 4 = (2 x − 7) 4 .
Solution. Let y = x − 72 . The given equation can be written as following
112 y 4 − 24 y 2 − 1 = 0 ,
thus y 2 = 14 and y 2 = − 28
1 , i.e.

thus
{
y ∈ 12 , − 12 , i 28
1 , −i 1 ,
28 }
{
x ∈ 3, 4, 72 + i 28
1 ,7 −i 1
2 28 }. ■
( )
n
10. Find all complex numbers n, so that the number z = 32+−ii is a real number.
Solution. Since
( 2 −i )
n
z = 3+i ⋅ 2+i = (1 + i ) n ,
2+i
n
we have that z = (2i ) , so the number z is real if and only if z 2 ≥ 0 holds true, and that
2

it is true if and only if n is factored by 4. ■

146
11. Given a complex number u. Find all complex numbers z, such that a = u1−−uz
z
is
a real number.

Solution. A number a is a real number if and only if a = a . Thus, u1−−uz

z
is a real
u − uz
number if and only if 1− z = u − zu , i.e. if and only if
1− z
(u − u )(1 − z z ) = 0 .
Hence, if u is a real number, then a solution is each complex number z ≠ 1 , and if u is not
a real number, then a solution of the given problem is each complex number z ≠ 1 such
that z z = 1 , i.e. a solution is each complex number z such that | z | = 1, z ≠ 1 . ■

12. Let a1 ,..., an be given complex numbers, such that

| a1 | = ... = | an | = 1 and a1 + a2 + ... + an = 0 .
Prove that for each complex number z
| a1 − z | + | a2 − z | +...+ | an − z | ≥ n holds true.
n
Solution. Since | ai | = | ai | , for each i = 1, 2,..., n and ∑ ai = 0 , we get that
i =1
n n n n n
2
n = ∑ | ai | = ∑ ai ai = ∑ ai ai − z ∑ ai = ∑ (ai ai − zai )
i =1 i =1 i =1 i =1 i =1
n n n
= ∑ ( ai − z ) ai ≤ ∑ | ai − z | ⋅ | ai | = ∑ | ai − z | . ■
i =1 i =1 i =1

13. Given complex numbers z1 , z2 ,..., z2n +1 such that | zi | = 1 and Im zi ≥ 0 , for
i = 1, 2,..., 2n + 1 . Prove that the following holds true
2 n +1
∑ zi ≥ 1 .
i =1
Solution. The inequality can be proven by the principle of mathematical induction.
Clearly, the inequality holds true for n = 0 . Let assume that it holds true for all 2n − 1
complex numbers which satisfy the condition of the given problem.
Let z1 , z2 ,..., z2n +1 be complex numbers which satisfy the conditions of the problem.
Without loss of the generality we assume
arg z1 ≤ arg z2 ≤ ... ≤ arg z2n +1 .
Let position a new coordinate system in a complex plane so that the imaginary axis is a
bisector of ∠ z1Oz2 , and the real axis passes through the point O(0,0). In a so-defined
coordinate system the points are denoted by
zk = xk + iyk , k = 1, 2,..., 2n + 1 .
We have
yk ≥ 0 and x1 = − x2n +1 , y1 = y2n +1 ,
The above stated and the inductive assumption imply that:

147
2 n +1
∑ zi = ( x1 + x2 + ... + x2n + x2n+1 )2 + ( y1 + y2 + ... + y2n + y2n+1 )2
i =1

= (xx2 + ... + x2n ) 2 + ( y1 + y2 + ... + y2n + y2n +1 ) 2

≥ ( x2 + ... + x2n ) 2 + ( y2 + ... + y2n ) 2
= | z2 + ... + z2n |2 ≥ 1.
Finally, by applying the principle of mathematical induction we get that each odd number
of complex numbers satisfies the conditions in the given problem. ■

14. Let a0 , a1 ,..., an be complex numbers such that if z ∈C, | z | ≤ 1 , then

| an z n + an −1 z n −1 + ... + a1 z + a0 | ≤ 1.
Prove the following
| ak | ≤ 1 and | a0 + a1 + ... + an − (n + 1)ak | ≤ n , for each k = 0,1, 2,..., n .
Solution. Let
P ( z ) = an z n + an −1 z n −1 + ... + a1 z + a0
and wi , i = 0,1,..., n be the (n +1)-th roots of the one. But,
n
∑ wik = 0, if k is not factored by n +1 and
i =0
n
∑ wik = n + 1 if k is factored by n +1.
i =0
So,
n
∑ wik P( wi ) = (n + 1)an−k , for each k ∈{0,1,..., n}.
i =0
Hence,
n n n
(n + 1) | an −k | = ∑ wik P ( wi ) ≤ ∑ | wik P ( wi ) | = ∑ | P ( wi ) | ≤ 1
+ + ...
1 + 1 = n + 1 ,
i =0 i =0 i =0 n +1
The last implies that | an −k | ≤ 1, for each k ∈{0,1,..., n}.
For the second part of the statement it is true that
n n n
∑ wik P( wi ) = ∑ wik P( wi ) − P(1) = (n + 1)an−k − ∑ ai and
i =1 i =0 i =1
n n n
∑ wik P( wi ) ≤ ∑ | wik P( wi ) | = ∑ | P( wi ) | ≤ n
i =1 i =1 i =1
thus,
n
(n + 1)an −k − ∑ ai ≤ n , for each k ∈{0,1,..., n}. ■
i =1

148
15. Let n be a positive integer. Prove that the polynomial
P ( z ) = z n +1 − z n − 1
has a root w such that | w | = 1 if and only if 6 | (n + 2) .
Solution. Let | w | = 1 and wn +1 − wn − 1 = 0 . Then wn ( w − 1) = 1 and since | w | = 1
we get that | w − 1 | = 1 . Hence, w is one of the intersection of the circles | z | = 1 and
| z − 1 | = 1 , that is w = cos π3 ± i sin π3 . Moreover, w − 1 = cos 23π ± i sin 23π = w2 . Finally,
( n + 2) π ( n + 2) π
1 = wn ( w − 1) = wn + 2 = cos 3
± i sin 3
,
( n + 2) π
The last means that 3
= 2k π, for some k ∈ N , thus n + 2 = 6k , for some k ∈ N , i.e.
6 | (n + 2) .
Conversely, if 6 | (n + 2) , i.e. n + 2 = 6k , for some k ∈ N , then for w = cos π3 ± i sin π3
holds true that | w | = 1, w2 = w − 1 and wn+ 2 = 1 , thus
wn +1 − wn − 1 = wn ( w − 1) − 1 = wn w2 − 1 = wn + 2 − 1 = 0 . ■

16. Let z1 , z2 ,..., zn be arbitrary complex numbers. Prove that positive integers
i1 ,...., ik can be chosen such that they satisfy the followings 1 ≤ i1 < ... < ik ≤ n and
| zi1 + zi2 + ... + zik | ≥ 2 (| z1 | + | z2 | +...+ | zn |) .
4 2
Solution. Let z j = x j + iy j , x j , y j ∈ R , j = 1,..., n . Lets
S1 = { j | x j ≥ 0, y j ≥ 0}, S2 = { j | x j < 0, y j ≥ 0},
S3 = { j | x j < 0, y j < 0} , S4 = { j | x j ≥ 0, y j < 0}.
Then,
n 4
∑ | zj |= ∑ ∑ | zj |
j =1 j =1 j∈Sk
by applying the principle of Dirichlet, we get that for some k ∈ {1, 2, 3, 4} the following
inequality holds true
n
∑ | z j | ≥ 14 ∑ | z j | .
j∈Sk j =1
For such a number k we get the following
n
1 1 | = 1 ∑ | x j + iy j | ≤ 1 ∑ (| x j | + | y j |)
∑ | zj |≤
4 2 j =1
∑ | zj
2 j∈S 2 2
k j∈Sk j∈Sk

= 1  ∑ xj + ∑ yj 
2  j∈S j∈Sk 
 k 
1
∑ x j + ∑ y j  = ∑ z j . ■
2 2
≤ 2 2
2
 j∈Sk j∈Sk  j∈Sk

149
17. Let z1 , z2 ,..., zn be complex numbers such that
| z1 | + | z2 | +...+ | zn | = 1 .
Prove that it exists a set S ⊆ {z1 , z2 ,..., zn } such that
∑ zi ≥ 16 .
zi ∈S

Solution. Let zk = xk + yk , k = 1, 2,..., n . Then

n n n
1 = ∑ | zk | = ∑ xk2 + yk2 ≤ ∑ (| xk | + | yk |)
k =1 k =1 k =1
= ∑ | xk | + ∑ | xk | + ∑ | yk | + ∑ | yk |.
xk ≥0 xk <0 yk ≥ 0 yk < 0
by applying the principle of Dirichlet, we get that at least one of the four sums of the
right side of the equality is equal or greater to 14 . Let assume that ∑ | xk | ≥ 14 .So we
xk <0
get that
∑ zi ≥ ∑ xk = ∑ | xk | ≥ 14 > 16 . ■
xk <0 xk <0 xk <0

18. Let a, b, c be any complex numbers and w = −1+2i 3 . Prove that the following
holds true
a3 + b3 + c3 − 3abc = (a + b + c)(a + bw + cw2 )(a + bw2 + cw).
Solution. It can be directly checked that
w3 = 1, w4 = w, w + w2 = −1,
thus,
(a + b + c)(a + bw + cw2 )(a + bw2 + cw) =
(
= (a + b + c) a 2 + b 2 + c 2 + w(ab + bc + ca ) + w2 (ab + bc + ca ) )
= (a + b + c) ( a 2 + b 2 + c 2 + ( w + w2 )(ab + bc + ca ) )

= (a + b + c) ( a 2 + b 2 + c 2 − (ab + bc + ca ) ) = a3 + b3 + c3 − 3abc,
which was required to be proven. ■

19. Let a and b be positive real numbers. Determine the minimum of the expression
x+ y
, if x and y are complex numbers, so that | x | = a, | y | = b .
1+ x y
Solution. We have
2 | x |2 +| y |2 +2 Re x y
x+ y x+ y x+ y
= ⋅ =
1+ x y 1+ x y 1+ xy 1+| x y |2 +2 Re x y
| x |2 +| y |2 −1−| x y|2 ( a 2 −1)(b 2 −1)
=1+ 2
=1− ,
1+| x y | +2 Re x y 1+| x y |2 +2 Re x y
whereby

150
min{Re x y :| x | = a,| y | = b} = − ab,
max{Re x y :| x | = a,| y | = b} = ab.
If at least one of the numbers a and b is equal to 1, then
x+ y
min =1
x = a , y =b + x y
1
If both of the numbers a and b are either greater or less than 1, then
x+ y ( a 2 −1)(b 2 −1)
min = 1− = 1a−−ab
b
| x |= a ,| y |=b 1+ x y 1+ a 2b 2 − 2 ab
If one of the numbers a and b is greater than 1, and the other is less than 1, then
x+ y ( a 2 −1)(b 2 −1)
min = 1− = 1a++ab
b .■
| x |= a ,| y |=b + x y
1 1+ a 2b 2 + 2 ab

( 2 + 2i )
10
20. Compute z = .
( 3 +i )
8

Solution. The numbers

z1 = 3 + 1 and z2 = 2 + 2i
shall be rewritten using Euler’s formulas. So,
2 2 2
r1 = 3 + 12 = 2 , r1 = 2 + 2 = 2,
ϕ1 = arctg = 6 , ϕ2 = arctg1 = π4 ,
1 π
3
i π6 iπ
thus z1 = 2e and z2 = 2e 4 . Therefore,
( )
10
iπ i 10 π
( 2 + 2i )10 10
= 2e = 2 e 8π
4 4
z=
( 3 +i )8
( 2e )
8 i
iπ 28 e 6
6

= 22 e
i 5π
2
= 4e
i ( 52π − 43π ) = 4ei 76π
i 4 π
e 3
( ) (
= 4 cos 76π + i sin 76π = 4 − 23 − i 12 = −2 3 − 2i. ■ )
21. Let z1 , z2 , z3 be distinct complex numbers with equal modulus. If the numbers
z1 + z2 z3 , z2 + z3 z1 and z3 + z1 z2 are real, then the equality z1 z2 z3 = 1. holds true.
Prove it!
Solution. Let zk = r (cos ϕk + i sin ϕk ), k = 1, 2, 3. By direct computations we get
that
z1 z2 z3 = r 3 (cos ϕ1 + i sin ϕ1 )(cos ϕ2 + i sin ϕ2 )(cos ϕ3 + i sin ϕ3 )
= r 3 (cos ϕ + i sin ϕ),
where ϕ = ϕ1 + ϕ2 + ϕ3 .

151
The condition of the given problem therefore
sin ϕk + r sin(ϕ − ϕk ) = 0, k = 1, 2, 3,
that is
sin ϕk (1 − r cos ϕ) + cos ϕk ⋅ r sin ϕ = 0, k = 1, 2, 3.
Let assume that 1 − r cos ϕ ≠ 0. Then
r sin ϕ
tg ϕk = r cos ϕ−1 , k = 1, 2, 3,
which is not possible thereby ϕk ∈[0, 2π) . These numbers differ each other, hence
1 − r cos ϕ = 0 and sin ϕ = 0, thus cos ϕ = 1, r = 1. Finally,
z1 z2 z3 = r 3 (cos ϕ + i sin ϕ) = 1. ■

22. Determine the maximal value of | z | if given that the complex number z satisfies
the following condition z + 1z = 1?
iϕ
Solution. Let z = re , where eiϕ = cos ϕ + i sin ϕ . So,
z + 1z
2
( )(
= reiϕ + 1r e −iϕ re −iϕ + 1r eiϕ )
= r 2 + 12 + e 2iϕ + e −2iϕ = r 2 + 12 + 2 cos 2ϕ.
r r
The equation
r 2 + 12 + 2 cos 2ϕ = 1
r
implies
2 −2 cos 2ϕ+1± ( 2 cos 2ϕ−1)2 −4
r = 2
.
This implies that r 2 , as well as r will be maximal if
cos 2ϕ = −1.
Then
−2( −1) +1+ ( 2( −1)−1)2 −4
rmax = 2
= 3+2 5 = 1+2 5 . ■

23. Prove that z + 1z = 2cos θ implies z m + 1m = 2cos mθ .

z
Solution. The equation z + = 2cos θ implies z 2 − 2 z cos θ + 1 = 0 . The solutions
1
z
of the last quadratic equations are z = cos θ ± i sin θ . Now 1z = cos θ  i sin θ . Therefore,
z m = cos mθ ± i sin mθ and 1m = cos θ  i sin mθ .
z
By adding the last two equalities we get
z m + 1m = 2cos mθ . ■
z

24. Let t ∈ R and z = 11+−itit . Prove that

z n + z n = 2 cos( 2n arctg t ) .

152
2
Solution. We express z as following z = 11+−itit = 1−t 2 − i 2t2 . If t = tg 2x , i.e.
1+t 1+t
1− tg 2 2x 2 2 tg 2x
x = 2arctg t , then cos x = = 1−t 2 and sin x = = 2t 2 , thus
1+ tg 2 2x 1+t 1+ tg 2 2x 1+t

z n + z n = (cos x + i sin x) n + (cos x − i sin x) n

= cos nx − i sin nx + conx − i sin nx
= 2 cos nx = 2 cos(2n arctg t ) . ■

25. Solve the eqaution:

iz 6 +8 = 3.
8i − z 6
Solution. If 8i − z 6 = 0 , then the fraction which is on the left side of the equation is
not defined. If 8i − z 6 ≠ 0 , i.e. z 6 ≠ 8i , then we multiply by 8i − z 6 the both sides of the
equation and after grouping the terms which consist of z on the left side, and the other
ones on the right side of the equation we get the following
z 6 ( 3 + i ) = −8 + 8 3i .
Thus,
( )( ) ( )( )
z 6 = −8+8 3i = −8+8 3i 3 −i = 8 −1+i 43 3 −i = 8i .
3 +i ( 3 +i )( 3 −i )
6
But, we previously found that z ≠ 8i , so, we find that the given equation has no solu­
tion. ■

26. On the field of complex numbers find the solution of the given system
 x 4 + 6 x 2 y 2 + y 4 = 5
 3
 x y + xy 3 = 1
Solution. The given system is equivalent to the system whose second equation is
multiplied by 4, that is equivalent to the following system
 x 4 + 6 x 2 y 2 + y 4 = 5
 3 3
4 x y + 4 xy = 4
Thus, it is equivalent to the system which consists of the sum and the difference of these
two equations, that is equivalent to the following system
( x + y ) 4 = 9
 4
( x − y ) = 1
Since the above equations we get that
α⋅ 3 +β α⋅ 3 −β
x = 2 , y = 2 , α, β ∈{1, − 1, i, − i} ,
that is, 16 solutions. It is necessary to percieve that these solutions differed from each
other. Namely, if

153
α1⋅ 3 +β1 α 2 ⋅ 3 +β 2
= ,
2 2
then
(α1 − α 2 ) 3 = (β2 − β1 ) ,
so α1 = α 2 , which means that β1 = β2 , i.e.that is the same solution, since if it is con-
versely 3 can not be expressed as r + is , where r , s ∈Q , therefore 3 is an irrational
num­ber. ■

27. Solve the system in the field of complex numbers

 z19 w25 = 1,

 z 5 w7 = 1,
 4 4
 z + w = 2.
Solution. If we cube the second equation of the system we get that z15 w21 = 1 . Since
z19 w25 = 1 and z15 w21 = 1 it is true that z 4 w4 = 1 . But, z 4 + w4 = 2 , and the Vieta’s for-
mulas we get that z 4 and w4 are solutions of the quadratic equation t 2 − 2t + 1 = 0 , i.e.
4
z 4 = 1 and w = 1 . Moreover, z 5 w8 = w implies zz 4 ( w4 ) 2 = w , i.e. z = w . Finally, so-
lutions of the given system might be only the ordered pairs (1,1), (−1, −1), (i, i ), (−i, −i ) .
With direct check it is easy to be proven that these pairs are truly solutions of the given
system. ■

28. If given that w0 , w1 ,..., wn −1 are the n-th roots of 1, calculate the following
sums:
n n n
а) ∑ kwk −1 b) ∑ k 2 wk −1 c) ∑ k 3 wk −1
k =1 k =1 k =1
Guidelines. а) The n-th roots of 1 can be written as following wk = wk −1 , where
w = cos 2nπ + i sin 2nπ . The required sum can be written as
n n
S = ∑ kwk −1 = ∑ kwk −1
k =1 k =1
Since 1 − w ≠ 0 , the following equalities hold true
S = S1−−Sw = 1+ 2 w+3w2 +...+ nwn −1 − w− 2 w2 −3w3 −...− nwn
w 1− w
1− wn − nwn
n −1 n 1−( n +1) wn + nwn +1
= 1+ w+...+1−ww −nw = 1− w
1− w
= .
(1− w)2
Finally, the required sum can be found by substituting for w as following
w = cos 2nπ + i sin 2nπ
and by using the De Moivre’s formula.
The examples b) and c) can be solved analogously as already explaned example a). ■

154
29. Calculate the sums:
n n
а) ∑ k cos 2knπ , b) ∑ k sin 2knπ
k =1 k =1
Guidelines. Let
n n
A = ∑ k cos 2knπ , B = ∑ k sin 2knπ
k =1 k =1
calculate S = 1 + A + iB . Then, apply the example 28 а) and further find
A = Re( S − 1) and B = Im( S − 1) . ■

30. Prove that

n
a) ∑ Cnk cos( k + 1)α = 2n cos n α2 cos n +2 2 α
k =0
n
b) ∑ Cnk sin( k + 1)α = 2n cos n α2 sin n +2 2 α .
k =0
Guidelines. Let z = cos α + i sin α , hence apply the binomial formula for
(1+ z ) n . ■

31. Prove that

s = cos π7 − cos 27π + cos 37π = 12 .

Solution. Firstly, let declare that z = cos πn + i sin πn implies

z n = −1, z 2n = 1 . (1)
Let w = cos ϕ + i sin ϕ . Hence w = cos ϕ − i sin ϕ = 1 , thus
w
cos ϕ = 12 w + w (
2w )
1 = w2 +1 , sin ϕ = 1 w − 1 = w2 −1
2i w 2iw . ( )
Further, by applying De Moivre’s formula we get
k
wk = cos k ϕ + i sin k ϕ and w = cos k ϕ − i sin k ϕ = 1k ,
w
thus
2k 2k
cos k ϕ = w k+1 , sin k ϕ = w −k1 . (2)
2w 2iw
Let z = cos π7 + i sin π7 . By taking for k = 1, 2, 3 in (2) and hence by applying (1) we
gat that
z 7 +1 + z 3
2 4 6 ( z 6 − z 5 + z 4 − z 3 + z 2 − z +1) + z 3 3
s = z2 +z 1 − z +21 + z +31 = = z +1
= z 3 = 12 . ■
2z 2z 2 z3 2z 3
2z

32. Compute
p = cos π7 cos 47π cos 57π .

155
Solution. Let z = cos π7 + i sin π7 . Then (2) and (1) in the example 31 therefore
2 4 6
cos π7 = z2 +z 1 , cos 27π = z +21 , cos 37π = z +31 ,
2z 2z
thus
( z 2 +1)( z 4 +1)( z 6 +1)
p = cos π7 cos 47π cos 57π = cos π7 cos 37π cos 27π =
8z6
12 10 8 6 4 2 5 3 6
6 4 2
= z + z + z + 26z + z + z +1 = − z − z − z + z +6 z + z +1+ z
8z 8z
6 5 4 3 2 6 z 7 +1 + z 6
= z − z + z − z 6+ z − z +1+ z = z +1
6
= 18 . ■
8z 8z

33. Let α, β and γ be angles of any triangle. Then,

cos 2 α + cos 2 β + cos 2 γ + 2 cos α cos β cos γ = 1 ,
holds. Prove it!
Solution. Let the left side of the identity be denoted by S and let
z = cos α + i sin α, w = cos β + i sin β . Then, the trigonometric entry of complex number
implies zw = cos(α + β) + i sin(α + β) , and since the equalities (2) in the example 31 we
get
cos 2 α + cos 2 β + cos 2 γ = cos 2 α + cos 2 β + cos 2 (π − (α + β))
= cos 2 α + cos 2 β + cos 2 (α + β)

( ) + ( w2w+1 ) + ( z 2wzw+1 )
2 2 2 2 2 2 2
= z2 +z 1
2 4 2 4 2 2 2 2 4 4
= w z + z w +6 z w2 +2z + w + z w +1 ,
4z w
and
2 2 2 2
2 cos α cos β cos γ = −2 cos α cos β cos(α + β) = −2 z2 +z 1 w2 w+1 z 2wzw+1
2 4 2 4 2 2 2 2 4 4
= − w z − z w − 2 z 2w 2 − z − w − z w .
4z w
Finally, S = 4 z 2 w2 = 1. ■
4 z 2 w2

34. Solve the equation

cos 2 x + cos 2 2 x + cos 2 3 x = 1 .
Solution. By using that
cos 2 t = 1+cos 2t ,
2
the given equation can be transformed as following
3 + 1 (cos 2 x + cos 4 x + cos 6 x ) = 1,
2 2
i.e.
cos 2 x + cos 4 x + cos 6 x = −1 .

156
Let z = cos x + i sin x . If we apply the equality (2) of example 31, we obtain the following
z 4 +1 + z8 +1 + z12 +1 = −1 ,
2z2 2z4 2 z6
and by using the well known mathematical transformation we get
(1 + z 2 + z 4 + 6 + z8 + z10 + z12 ) + z 6 = 0,
z14 −1 + z 6 = 0 ,
z 2 −1
8 6
( z − 1)( z + 1) = 0 .
In the above transformation we used the fact that z 2 ≠ 1 , thereby x = 0 and x = π are
not the solution of the given equation. Since ( z8 − 1)( z 6 − 1) = 0 , we get that z 6 + 1 = 0
or z 2 + 1 = 0 or z 4 + 1 = 0 . Furthermore, z 6 + 1 = 0 therefore cos 6 x = −1 , i.e.
x = π6 + k3π , k ∈ Z .
Similarly one can obtain the other solutions
x = π4 + k2π , k ∈ Z and x = π2 + k π, k ∈ Z . ■

35. Prove the identity

1
sin 2 x
+ sin14 x + ... + 1 = ctg x − ctg 2n x ,
sin 2n x
for n ∈ N, x ≠ λπk , k = 0,1, 2,..., n, λ ∈ Z
2
Solution. Let z = cos x + i sin x . The equality (2) in the example 31 implies
2k 2k
i ( w −1)
tg k ϕ = w 2 k−1 , ctg k ϕ = 2 k . (1)
i ( w +1) w −1
Further, thereby (1) for n = 2 s it is true that
s s +1 s s +1 s s +1
1 2iz 2 z2 + 2 z 2 +1− z 2 −1 ( z 2 +1)2 −( z 2 +1)
= =i =i
sin 2s x z 2s +1
−1 z 2s +1
−1 z 2s +1
−1 (2)
2s 2s +1
= i z s +1 − i z +1 = ctg 2 s −1 x − ctg 2 s x.
2 2s +1
z −1 z −1
Finally, if in (2) consequently we set s = 1, 2, 4,..., 2n and add the such obtained equalities
we get that
1 + 1 + ... + 1 =
sin 2 x sin 4 x n
sin 2 x
= (ctg x − ctg 2 x) + (ctg 2 x − ctg 4 x) + ... + (ctg 2n −1 x − ctg 2n x)
= ctg x − ctg 2n x,
which was supposed to be proven. ■

36. Let α, β and γ be angles of any triangle. The triangle is right angled triangle if
and only if
cos 2α + cos 2β + cos 2 γ = −1 .
Prove it!

157
Guidelines. Firstly, prove that for any triangle
cos 2α + cos 2β + cos 2 γ = −1 − 4 cos α cos β cos γ
holds. ■

37. Find the sum:

Sn = 1 + 2 cos x + 22 cos 2 x + ... + 2n cos nx .
Solution. Let
Tn = i (sin x + 22 sin 2 x + ... + 2n sin nx) and z = 2(cos x + i sin x) .
Then
Sn + Tn = 1 + 2(cos x + i sin x) + 22 (cos 2 x + i sin 2 x) + ... + 2n (cos nx + i sin nx)
n +1 2n +1[cos( n +1) x +i sin( n +1) x ]−1
= 1 + z + z 2 + ... + z n = z z −1−1 = 2(cos x −i sin x )
,
thus
 2n +1[cos( n +1) x +i sin( n +1) x ]−1  2n + 2 cos nx −2n +1 cos( n +1) x −2 cos x +1 . ■
Sn = Re  2(cos x −i sin x )  = 5− 4 cos x

38. Let a1 , a2 ,..., an be real numbers such that for each real number x is true that
1 + a1 cos x + a2 cos 2 x + ... + an cos nx ≥ 0. (1)
Prove that
a1 + a2 + ... + an ≤ n . (2)
Solution. Let ϕ = n2+π1 . Then for z = eiϕ we get
2 ( n +1) k π 2 ( n +1) k π
( n +1) k 1−cos −i sin n +1
1 + z k + z 2k + ... + z nk = 1− z = n +1
=0,
1− z k 1−cos 2nk+π1 −i sin 2nk+π1
for k = 1, 2,..., n , therefore
1 + cos k ϕ + cos 2k ϕ + ... + cos nk ϕ = 0 , (3)
for k = 1, 2,..., n . If in the inequality (1) we set consecutive
x = ϕ, x = 2ϕ, ..., x = nϕ ,
we get n inequalities, which after adding, by using the (3) give the inequality
n − a1 − a2 − ... − an ≥ 0 , which is equivalent to the inequality (2). ■

158
2. EXERCISES (CHAPTER 1)

1. Compute the following

а) i n , n ∈ Z , b) (1 + i ) n + (1 − i ) n , n ∈ N , and

( )
n
c) − 1 + i 3 , n∈N .
2 2

2. Find the exact value of the following expression

(1+i )2000
2000
.
(1−i ) −(1+i )2000

3. Determine the modulus of a complex number

x 2 − y 2 + 2 xyi
.
xy 2 +i x 4 + y 4

4. Let A = {z1 , z2 ,..., z1996 } be a set of complex numbers and let for each i = 1, 2,...,1996
hold true that A = {z1 zi , z2 zi ,..., z1996 zi } .
a) For each i = 1, 2,...,1996 it holds true that | zi | = 1 . Prove it!
b) If z ∈ A then z ∈ A . Prove it!

5. Let * be an operation in a field of complex numbers defined as following:

z * z1 = zz1 + i ( z + z1 ) − (1 + i ) . Prove that * is commutative and associative. Determine
the identity element e, i.e. determine an element e ∈C such that e * z = z * e = z is
satisfied for each z ∈C . Prove that there exists a unique complex number which
does not have its inverse ( z '∈C is an inverse element of z ∈C if zz ' = z ' z = e ).

6. Prove that
| b0 |2 + | b1 |2 + | b2 |2 = 3(a02 + a12 + a22 ) ,
for
bk = a0 + a1wk + a2 wk , k = 0,1, 2; a0 , a1 , a2 ∈ R and w = − 12 + i 23 .

7. Prove that x + iy = ( s + it ) n , x, ys, t ∈ R implies that

( x 2 + y 2 ) = ( s 2 + t 2 )n .

8. Graphically show the set of complex numbers z such that holds zz +−33 = 2 .

9. Solve the equations:

а) | z | + z = 4 − i , and b) | z + i | + | z − i | = 2 .

159
10. Let a1 , a2 ..., an be given complex numbers, such that | a1 | = ... = | an | = r .
Let Tns be the sum of each products of s pairwise distinct numbers. For example
Tn2 = a1a2 + a1a3 + ... + a1an + a2 a3 + ..... + an −1an .
Prove the following
|Tnn − s |
= r s , s = 1, 2, 3,..., n − 1 .
|Tns |

11. Let all roots of a polynomial

P ( x) = a0 x n + a1 x n −1 + ... + an −1 x + an
be on the circle | z | = r . Prove the following
| an −i | = r i | ai | , i = 0,1, 2,..., n .

12. Prove the following:

а) If | α | < 1 , then z −α ≤ 1 if and only if | z | < 1 .
1− zα
b) If | α | > 1 , then z −α ≥ 1 if and only if | z | < 1 .
1− zα

13. Express in trigonometric form the following complex numbers:

а) − 2 , b) −1 + i , c) 2 − i 3
d) 1+ cos α + i sin α , e) sin α + i (1 + cos α ) .

14. Find the exact value of (1+ w) n , if given that n ∈ Z and w = − 12 + i 23 .

15. Compute:
а) 2i , b) −8i , c) 4 + 3i ,
d) 3 −1 + i , e) 4 −2 3 − 2i .

16. 1 + i 3 + 1 − i 3 = 6 . Prove it!

( )
n
17. If aa +−ii = 1 , then a = ctg knπ , k = 0,1, 2,..., n − 1 . Prove it!

18. Solve the following equations, for n ∈ N .

а) ( x + 1) n − ( x − 1) n = 0 ,
b) ( x + 5i ) n − ( x − 5i ) n = 0 ,
c) ( x + 3i ) n + i ( x − 3i ) n = 0 ,

160
( )
n ctgα +i
d) ii +− xx = ctg±-i , α ∈ R , and
e) ( x + αi ) n + (cos θ + i sin θ)( x − αi ) n = 0, θ ≠ 2k π, α ≠ 0, α, θ ∈ R .

19. Prove the following identities:

π + cos 3π + ... + cos 9 π = 1 ,
а) cos 11 11 11 2

b) cos 211π + cos 411π + ... + cos 10

11
π =−1 .
2

20. Find the sums:

а) 1 − Cn2 + Cn4 − Cn6 + ... , and
b) C1n − Cn3 + Cn5 − Cn7 + ....

21. Find the sum:

4k
1 + Cn4 + Cn8 + C12
n + ... + Cn + ... .

22. Prove the equality:

(
1 + Cn3 + Cn6 + Cn9 + ... = 23 2n −1 + cos π3n . )
23. Prove that for m = 2, 3, 4,... it is true that:
( m −1) π
π sin 2 π sin 3π ...sin
sin m m m m
= mm−1 .
2

24. Prove the identity:

n
∑ (−1)k cos n πnk = 2nn−1 .
k =0
25. Let
Fn = a n sin nA + b n sin nB + c n sin nC ,
where a, b, c, A, B, C ∈ R and A + B + C = 2k π , for some k ∈ Z . Then, F1 = F2 = 0
implies Fn = 0 for each n ∈ N . Prove it!

26. Let x ≠ 2k π , x ∈ R , k ∈ Z and n ∈ N . Find the sum

1 + 2 cos x + 3 cos 2 x + 4 cos 3 x + ... + (n + 1) cos nx .

27. Let x, y ∈ R and n ∈ N . Calculate the sums:

а) cos x + cos( x + 2 y ) + ... + cos( x + 2ny ) , and
b) sin x + sin( x + 2 y ) + ... + sin( x + 2ky ) .

161
28. If given that cos π4 = 22 , then calculate cos 12
π .

29. Given that tg π8 = 2 − 1 . Prove the following: tg 24

π = 2 + 6 − 3 −2.

30. Prove the following:

а) tg π5 = 5 − 2 5 , b) tg 25π = 5 + 2 5 ,

c) tg 35π = − 5 + 2 5 , d) tg 45π = − 5 − 2 5

31. Calculate:
π sin 8π sin π ,
а) sin 21 21 7
π 2 π 3
b) tg 7 tg 7 tg 7 , π

π sin 2 π sin 3π sin 4 π sin 5π sin 6 π .

c) sin 14 14 14 14 14 14

32. Prove the identities:

а) sin 3 x = 34 sin x − 14 sin 3 x , and
b) cos 4 x = 18 cos 4 x + 12 cos 2 x + 83 .

33. Prove the identities:

а) cos 5 x = 16 cos5 x − 20 cos3 x + 5 cos x , and
b) sin 5 x = sin x(16 cos 4 x − 12 cos 2 x + 1) .

34. A complex number a is called an algebraic, if it is a root of a polynomial

P ( z ) = a0 z n + a1 z n −1 + ... + an −1 + an
with integer coeficients. The numbers wich are not algebraic are called as
transcendental numbers. For example π and e are transcendental numbers. But, there
exist numbers, such as eπ and e + π , for which it is still not found either they are
algebraic or trancedental numbers.
Prove that the numbers
а) 3 + 2 and b) 3 4 − 2i
are algebraic numbers.

35. Solve the eqation:

x n − nax n −1 − Cn2 a 2 x n −2 − Cn3a3 x n −3 ... − a n = 0, a ≠ 0 .

162
36. Let n ≥ 2 be a positive integer. Find all the solutions x0 of the equation
x n − x n −2 − x + 2 = 0 such that | x0 | = 1 .

37. Find the solutions for x:

cos a + C1n x cos(a + b) + Cn2 x 2 cos( a + 2b) + ... + Cnn x n cos( a + nb) = 0 .

38. Solve the equations:

а) sin 2 x + cos 2 x = 2 , and
б) tg 2 x + ctg 2 x = 6 .

39. Let f ( x, y ) =befa(rational

y, x) function with real coefficients. If the function f is symmetric,
i.e. if f ( x, y ) = f ( y, x) , then f (a, a ) ∈ R , for each a ∈C . Prove it!

40. Let [t] be the maximal integer which is not greater than t. Let z = x + iy . Prove the
following:
[ n / 2]
( )
а) Re z n = ∑ (−1) k 2nk x n −2k y 2k ,
k =0
[ n / 2]
( )
b) Im z n = ∑ (−1) k 2kn+1 x n −2k −1 y 2k +1 ,
k =0
n
( )
c) Re z n + Im z n = ∑ (−1)[ k / 2] nk x n −k y k .
k =0
Find the proper expression for Re z n − Im z n ?

41. Each zero of the following polynomial is in the field of complex numbers.
а) Re( x + i ) n ,
b) Im( x + i ) n ,
c) Re( x + i ) n ± Im( x + i ) n
For each of the above polynomial prove the given statement and find the zeros.

163
3. EXAMPLES (CHAPTER II, CHAPTER III)

1. Construct a trapezoid if given its bases and diagonals.

Solution. Let’s suppose that the given problem is solved and let ABCD be the required
trapezoid (figure 1). If a, b, c, d are the affixes
 of the points A, B, C, D respectively, then
DC = c − d . Let’s consider the translation
S ( z) = z + c − d
and let the point B1 be the image of B under
this translation. Further, S (d ) = d + c − d = c
therefore, the point C is the image of D under
this translation, thus CB 1 = DB . Therefore, all
Figure 1 three sides of the triangle AB1C are already
determined ( AB 1 = AB + DC , AC and
CB 1 = DB ), so it might be constructed. Since the base AB is already determined, the
point B can be found, and the point D is the inverse image of C. ■

2. Construct a circle such that it passes through a given point and tangents two
parallel lines.
Solution. Let’s assume that the given problem is solved (figure 2). If a is a vector
parallel to the line (p), then the translation S ( z ) = z + a maps the circle (K) to a circle
( K ') , such that it tangents the lines (p) and (q), but
it does not pass through A. So, we have to construct
a circle ( K ') such that it tangents the lines (p) and
(q) and by the translation we map that circle to the
required circle
The number of solutions depends on the
Figure 2 relationship between the point A and the lines (p) and
(q). Namely:
- If the point A is placed between the lines (p) and (q), then the given problem has
two solutions;
- If the point A is placed on any of the lines, then the given problem has a unique
solution, and
- In any other case the given problem has no solution. ■

3. Given the circles K '(o ', R ') and K ''(o '', R '') and the line (p). Construct a line (q)
parallel to (p), so that the circles ( K ') and ( K '') intersept congruent line segments on
the line (q).
Solution. Let’s assume that the given problem is solved (figure 3). Let’s plot a line
(a) perpendicular to (p) and a line O ' O perpendicular to (a), i.e. parallel to (p). Hence,
  
A ' A '' = B ' B '' = O ' O .

164
Due to this, if
S ( z) = z + o − o '

is translation for vector O ' O , then
S (a ') = a '' , S (b ') = b '' and S (o ') = o ,
and the circle ( K ') maps to a circle
K (o,| o − a '' |) and further A ''(a '') and
B ''(b '') are midpoints of the circles (K)
and ( K '') .
Whether the problem has a solution
or not depends on the relationship of the Figure 3
circles ( K ') and (K). Due to that, the given problem has a solution, and furthermore:
- if | R ''− R ' | ≤ | o ''− o | = OO '' ≤ R '+ R '' , then the given problem has a unique solu-
tion,
- if | R ''− R ' | > | o ''− o | = OO '' or | o ''− o | = OO '' > R '+ R '' , then the given problem
has no solution. ■

4. Let be given two A, B and let S = u − z be a poin reglection. If S ( A) = A '

 points
and S ( B ) = B ' , then AB = B ' A ' . Prove it!
Solution. Let a, b be the affixes of the points A, B, respectively. The affixes of the
points A ', B ' are a ' = S (a ) = u − a and b ' = S (b) = u − b , respectively. Thus,
b '− a ' = u − b − (u − a ) = a − b
 
The latter means that AB = B ' A ' . ■

5. If O ' and O '' are centers of a symmetry of the figure F, prove that O = SO '' (O ')
is also a center of a symmetry of the figure F.
Solution. Let o2' and o2'' be affixes of the points O ' and O '' respectively, and let
A(a) be any point on the figure F. Then, the affix of O is
( )
o = SO '' o2' = o ''− o2' .
Thereby the condition, the points A1 , A2 , A3 whose affixes are
a1 = SO '' (a ) = o ''− a,
a2 = SO ' (a1 ) = o '− (o ''− a ) = o '− o ''+ a,
a3 = SO '' (a2 ) = o ''− (o '− o ''+ a ) = 2o ''− o '− a,
belong on the figure F. Finally, the arbitrariness of the
point A and the equality
SO (a ) = 2o ''− o '− a = a3
imply that O = SO '' (O ') is also a center of a symmetry of
the figure F (figure 4). ■ Figure 4

165
6. Given two lines (p) and (q) and a point A. Through A draw a line (a), so that A is
the midpoint of the line segment MN, where M = (a ) ∩ ( p ) and N = (a ) ∩ (q ) .
Guidelines. Consider a poin reglection centered at A. ■

7. Let the circles K '(O ', R ') and K ''(O '', R '') meet at a point A. Through the point
A draw a line (a) on which the circles intersept
congruent line segments.
Solution. Let’s suppose that the given problem
is solved and let (a) be the required line (figure 5).
Thereby AB = AC , a point A is the midpoint of a
line segment BC. It means that C = S A ( B ) . A point
B ∈ ( K ') , thus the point C = S A ( B ) will be on the
circle ( K ) = S A ( K ') . So, C ∈ ( K ) ∩ ( K '') .
Figure 5 The problem has a unique solution. ■

8. four points A, B, C, D, so that AB = CD,

Given
but AB ≠ CD . Prove that it exists a rotation S such
that S ( A) = C , S ( B ) = D !
How many such rotations exist?
Guidelines. Consider the case when a line AC
is parallel to a line BD, (figure 6) and the case when
Figure 6 a line AC is not parallel to a line BD, (figure 7),
separately. ■

9. Let ( K ') and ( K '') be circles with congruent

radii and let they intersect at A and B. Prove that
it exists a rotation S around the point A, so that
S ( K ') = K '' . Moreover, if X ∈ ( K ') and S ( X ) = X ' ,
then the line XX ' passes through the point B.
Solution. Since r ' = r '' , the line AB is a bisector of
Figure 7 the line segment O ' O '' , thus A is a center of the rotation
S A,α such that S A,α ( K ') = K ' , where α = ∠O ' AO '' .
Let X ∈ ( K ') be any point and X ' = XB ∩ ( K '') ,
(figure 8). Since ∠ AXB and ∠ AX ' B are inscribed
angles of congruent arcs, the angles are conguent,
and thus AX = AX ' . If S A,α ( X ) = X * , we get that
AX = AX * and X * ∈ ( K '') , therefore X ' = X * .
Finally, for any point X ∈ ( K ') , the points X, B
and S A,α ( X ) are collinear. ■
Figure 8

166
10. Given lines (p) and (q) and a point A. Construct an equilateral triangle ABC, such
that B ∈ ( p ) and C ∈ (q ) .
Solution. Let assume that the given problem
is solved and ABC is the required triangle, (fig-
ure 9). Since the triangle ABC is an equilateral,
we obtain that AB = AC and ∠ BAC = 60° .
That is, S A,60° ( B ) = C or S A,−60° ( B ) = C .
The point B is on the line (p), and therefore the
point S A,60° ( B ) = C is on the line S A,60° ( p ) .
On the other hand C is on the line (q), thus
C = (q ) ∩ S A,60° ( p ) . For the rotation S A,−60° the
latter holds true, analogously.
Figure 9
Let
S A,60° ( p ) = ( p1 ) and S A,−60° ( p ) = ( p2 ) .
If
C1 = (q ) ∩ ( p1 ) , C2 = (q ) ∩ ( p2 ) ,
S A,−60° (C1 ) = B1 and S A,60° (C2 ) = B2 ,
then AB1C1 and AB2C2 are the required trian-
gles, (figure 10). Further, at least one of the lines
S A,60° ( p ) and S A,−60° ( p ) meet the line (q),
which means that the problem always has at least
Figure 10
one possible solution. ■

11. Given lines (p) and (q) and a point O. Construct a

square ABCD centered at the point O, such that two adjacent
vertices are on the lines (p) and (q), respectively.
Solution. Let assume that the given problem is already
solved and ABCD is the required square, (figure 11). We have
∠ AOB = − 90° , therefore SO,−90° ( A) = B . ­Further, A ∈ ( p ) Figure 11
implies that B ∈ (q ) ∩ SO,−90° ( p ) .
Let’s consider the rotations SO,90° and SO,−90° .
So,
( p1 ) = SO,90° ( p ) , ( p2 ) = SO,−90° ( p ) ,
B1 = ( p1 ) ∩ (q ), B2 = ( p2 ) ∩ (q ),
A1 = SO,−90° ( B1 ) , A2 = SO,90° ( B2 ) .
Then the required squares are squares with sides
A1B1 и A2 B2 , respectively.
Let state that the above problem may have ei-
ther two solutions or none. Namely, if the lines (p)
and (q) are perpendicular to each other, then the lines Figure 12

167
( p1 ) and ( p2 ) are parallel with (q) and the given problem does not have any solution,
if lines (p) and (q) are not perpendicular, then the given problem has exactly two solu­-
tions. ■

12. Given lines (p) and (q) and a point O.

Construct an equilateral triangle ABC centered at
the point O, so that its two vertices are placed on
the lines (p) and (q), respectively.
Solution. Let assume that the given prob-
lem is solved and let the triangle ABC be the
Figure 13 required triangle. Thereby ∠ AOC = −120° ,
we get that SO,−120° ( A) = C , therefore
C = SO,−120° ( p ) ∩ (q ) , (figure 13).
Let’s consider the rotations SO,120° and
SO,−120° .Therefore
( p1 ) = SO,120° ( p ) , ( p2 ) = SO,−120° ( p ) ,
C1 = ( p1 ) ∩ (q ) , C2 = ( p2 ) ∩ (q ) ,
A1 = SO,−120° (C1 ) , A2 = SO,120° (C2 ) .
Then the required triangles have sides A1C1 and
A2C2 , respectively (figure 14).
The given problem may have two solutions,
Figure 14 one solution or none. ■

13. Let S ( z ) = az + b , a ∈ R \ {0,1} be a homothety and M be any point. Prove that

the center of a homothety C, the point M and its image N are collinear.
Solution. Let the affix of the point M be z. Then the affixes of the center of a
homothety C and a point N are c = 1−ba and w = az + b . Then,
w− z b + ( a −1) z b + ( a −1) z
= a zb+b − z = (1 − a ) = 1 − a = b +( a −1) z (1 − a ) = azb+b − z = wc −−zz ,
c− z −z b + ( a −1) z −z
1− a
1− a
and since corollary 1.4 we have that the points C, M and N are collinear. ■

14. Given a circle K(O,r) and a point A on the circle. Determine the locus of midpoints
of the chords of the circle (K) at the point A.
Solution. Let AX be an arbitrary chord of a circle K(O,r) and let Y be its midpoint,
(figure 15). If the affixes of the points A, X and Y are a, z and w, respectively, then since
the condition of the problem it is true that
w = 12 ( z + a ) = 12 z + 12 a ,
The latter means that the required locus is the image of the circle (K) under the similar-
ity

168
S ( z ) = 12 z + 12 a ,
with coefficient 12 and center
1a
c= 2
=a.
1− 12
Finally, the required locus is the image of a circle (K) under
the homothety with center at A and coefficient 12 , i.e. that is
the circle with diameter OA. ■
Figure 15
15. On the bases AB and DC of the trapezoid
ABCD on the same side of them, are constructed
equilateral triangles ABM and DCN. Prove that a
line MN passes through the intersection point O
of the extensions of the legs.
Guideline. Prove that the homothety with
center at O and coefficient AB maps the point
DC
N at M, (figure 16), and further apply the Exam- Figure 16
ple 13. ■

16. Let ABCD be trapezoid with bases AB and CD and let M be the midpoint of AB,
N be the midpoint of CD, P be the intersection of diagonals and Q the intersection of
extensions of the legs. Prove that the points M, N, P and Q are collinear.
Guideline. Prove that it exists a homothety Η with
center at Q and coefficient DC so that Η( A) = D and
AB
Η( B ) = C , further, conclude that Η( M ) = N , (fig-
ure 17), and then apply the example 13. Provethat there
exists a homothety Η1 with center P and coefficient
− DC so that Η1 ( A) = C and Η1 ( B ) = D , and further
AB
conclude that Η( M ) = N , (figure 17), and then apply
Figure 17
the example 13. ■

17. Given two concentric circles K(O,R) and K '(O ', R ') , R > R ' . Draw a line (p)
which consecutively meets the circles at A, B, C and D, so that AB = BC = CD .
Solution. Let assume that the given problem is
solved and let (p) be the required line, (figure 18).
Then AB = 13 , thus under homothety Η with cen-
AD
ter at A and coefficient 13 the point D maps at B.
This means that B is on the circle Η(K). That is,
B ∈ Η( K ) ∩ ( K ') . Figure 18

169
Finally, if we fix a point A on a circle (K) and let B ∈ Η( K ) ∩ ( K ') , then the required
line ( p ) = AB , (make a figure). ■

18. Construct a triangle ABC if given α, β and hc .

Solution. Let assume that the given problem is solved
and ABC is a required triangle (figure 19). If A1B1C1 is
any triangle with angles α and β, then the triangles ABC
and A1B1C1 are homothetic with center of homothety at
hc
D and coefficient , thus the required triangle can be
C1D
constructed if we take an arbitrary triangle with angles
α and β and map it under homothety Η with center D and
Figure 19 coefficient A1B1C1 , (make a figure). ■

19. In a triangle ABC, inscribe a triangle PQR whose sides are perpendicular to the
sides of ABC.
Guideline. Take any
point K on the side AB, and
construct the triangle KMN so
that NK ⊥ AB, MN ⊥ AC .
Further, draw the line AM and
find the point Q = AM ∩ BC
. Now, the required triangle
PQR is an image of the triangle
KMN under homothety Η with
Figure 20 AQ
center A and coefficient
AM
(figure 20). ■

20. Prove that if two circles touch each other, their centers and the point of touch are
collinear.
Guideline. Determine the affix c of a point of touch of the circles with equations
| z | = 1 and | z − a | = R and prove that the points with affixes 0, a and c are collinear. ■

21. Given points A and B. Let A' be a point on the line OB, B ' be a point on the line
OA and Z be a point on AB. Construct a point Z ' which divides the line segment A ' B '
in a same ratio as a point Z divides the line segment AB.
Solution. Let a, a ', b, b ', z be the affixes of points A, B, A ', B ', Z , respectively. If
BZ = λ , then
ZA
b − z = λ( z − a ) (1)
Hence, we should obtain the affix z ' of a point Z so that

170

b '− z ' = λ ( z '− a ') . (2)
Let Z1 be the point where
the OA meets the line which
passes through the point Z and
is parallel with BB ' , and Z 2
be the point where the line OB
meets the line which passes
through the point Z and is
parallel with AA' . Then, the
similarity of triangles AZZ1
and ABB ' , i.e. the similarity Figure 21
of BZZ 2 and BBA' implies that
b '− z1 = λ ( z1 − a ) and b − z2 = λ( z2 − a ') . (3)
By reducing a and b, in (1) and (3), we obtain that
b '− ( z1 + z2 − z ) = λ[( z1 + z2 − z ) − a '] ,
Since (2),we get that
z ' = z1 + z2 − z π .
Hence, the point Z ' is obtained by construction, (show in figure 21). ■

22. Prove that the sum of interior angles of any triangle is p.

Solution. Let a, b, c be affixes of vertices A, B, C of a triangle. Then
∠ CBA = arg ac −−bb , ∠ BAC = arg bc −−aa , ∠ ACB = arg ba−−cc .
Since each of the interior angles of a triangle is strictly smaller then π, we get that their
sum is strictly smaller than 3π, and thus
∠ CBA + ∠ BAC + ∠ ACB = arg ac −−bb + arg bc −−aa + arg ba−−cc
= arg ac −−bb bc −−aa ba−−cc = arg(−1) = π,
which was supposed to be proven. ■

23. Let ABCD be strictly convex quadrilateral and let the points Ta , Tb , Tc , Td be
centroids of the triangles BCD, ACD, BAD, ABC, respectively. Prove that the medians of
the quadrilaterals ABCD and TaTbTcTd are concurrent.
Solution. According to Example I 4.2. B) the medians MP and NQ of the quadrilateral
ABCD intersect at a point T with affix t = a +b +4c + d , which according to Example 15.10 А)
is the centroid of the quadrilateral ABCD. Analogously, the medians of the quadrilateral
TaTbTcTd intersect at its centroid, whose affix is
b +c + d + a +c + d + a +b + d + a +b + c
ta +tb +tc +td
t'= 4
= 3 3
4
3 3
= a +b +4c + d .
Finally since, t = t ' the statement of the given problem is proven. ■

171
24. а) Given two vertices of an equilateral trianglein a complex plane. Determine its
third vertex.
b) Obtain a point z3 so that the points z1 = 2 + 2i , z2 = 3 + i and z3 create an
equilateral triangle.
Solution. а) Let be given points A and B with affixes a and b. The given problem has
two possible solutions C and C ' :  ABC is positively oriented
 and  ABC ' is negatively
oriented. Thereby, a point C is obtained
 if the vector AB is rotated around a point A at
π , and C ' is obtained when aif the AB is rotated around a point A at − π . Therefore
3 3
i π3 −i π3
c = a + (b − a )e and c ' = a + (b − a )e .
b) Since solution a),
i π3 i π3
z3' = z1 + ( z2 − z1 )e = 2 + 2i + (1 − i )e = 2 + 2i + (1 − i ) 1+i2 3 = 5+2 3 + i 3+2 3
−i π3 −i π3
z3'' = z1 + ( z2 − z1 )e = 2 + 2i + (1 − i )e = 2 + 2i + (1 − i ) 1−i2 3 = 5−2 3 + i 3−2 3 . ■

25. Given an equilateral triangle  ABC let a be an affix of the vertex A. Determine
the affix of the vertex B if the origin coincides with:
а) the vertex C,
b) the centroid T of the  ABC,
c) A1 , the foot of the altitude at the vertex A to the line segment BC.
Solution. а) Since C coincides with the origin, c = 0 . If  ABC is positively oriented,
then B is obtained by rotation of the point A at π3 around the point C and therefore
i π3
b = ae = a 1+i2 3 .
If  ABC is negatively oriented, then B ' is obtained by rotation of the point A at − π3
around the point C and therefore
−i π3
b ' = ae = a 1−i2 3 .
b) Since T coincides with the origin, t = 0 . If  ABC is positively oriented,
thenthe point B is obtained by rotation of the point A at 23π around T and therefore
i 23π
b = ae = a −1+2i 3 . If  ABC is negatively oriented, then the point B ' is obtained by
−i 2 π
rotation of the point A at − 23π around C and therefore b ' = ae 3 = a −1−2i 3 .
c) Since A1 coincides with the origin, a1 = 0 . Due to A1 A = A1B 3 , we obtain the
point B by rotation of the point A around A1 at π2 if  ABC is positively oriented, that is
− π2 if  ABC is negatively oriented and the both cases, the obtained results are divided
by 3 . Thus, the given problem has two solutions: b = ai and b ' = − ai . ■
3 3

172
26. If a, b, c are the affixes of vertices of an equilateral triangle, then
a + b 2 + c 2 = ab + bc + ca . Prove it!
2

Solution. Let t be affix of the centroid of an equilateral triangle, (with vertices whose
i 23π
affixes are a, b, c) such that a, b, c are the affixes, of its vertices,and let u = e . Then
b = t + (a − t )u and c = t + (a − t )u 2 . The equality
a 2 + b 2 + c 2 = ab + bc + ca
is equivalent to the equality
(a − b) 2 + (b − c) 2 + (c − a ) 2 = 0 . (1)
We will prove the equality (1). Thus
(a − b) 2 + (b − c) 2 + (c − a ) 2 = [(1 − u ) 2 + (u − u 2 ) 2 + (u 2 − 1) 2 ](a − t ) 2
= [1 + u 2 + (u + 1) 2 ](1 − u ) 2 (a − t ) 2
= 2(1 + u + u 2 )(1 − u ) 2 (a − t ) 2
= 2(1 − u 3 )(1 − u )(a − t ) 2
= 2(1 − e 2iπ )(1 − u )(a − t ) 2
= 2(1 − 1)(1 − u )(a − t ) 2 = 0. ■

27. If a, b and c are complex numbers so that they satisfy a 2 + b 2 + c 2 = ab + bc + ca ,

then either a = b = c or a, b and c are affixes of the vertices of an equilateral triangle.
Prove it!
Solution. The equality a 2 + b 2 + c 2 = ab + bc + ca is equivalent to the equality
(b − c) 2 = (c − a )(a − b) , and thus
| b − c |2 = | c − a | ⋅ | a − b | .
Analogously, it can be proven that
| c − a |2 = | a − b | ⋅ | b − c | and | a − b |2 = | b − c | ⋅ | c − a | .
If the last three equalities are multiplied by | b − c |, | c − a | and | a − b | , respectively,
then, we obtain the following equalities
| a − b |3 = | b − c |3 = | c − a |3 = | a − b | ⋅ | b − c | ⋅ | c − a | ,
The latter explaines the following | a − b | = | b − c | = | c − a | . The above means that either
a = b = c or a, b and c are affixes of the vertices of an equilateral triangle. ■
Comment. The statements in the Examples 26 and 27 can be transformed as fol-
lowing
A triangle ABC, where a, b, c are affixes of its vertices is an equilateral triangle if
and only if a 2 + b 2 + c 2 = ab + bc + ca holds true.

173
Indeed, a triangle ABC is equilateral if it and only if it is directly similar with the
triangle BCA, that is if and only if α = β, β = γ, γ = α ⇔ α = β = γ = π3 , which actually
means that if and only if
1 1 1
a b c = 0 ⇔ a 2 + b 2 + c 2 = ab + bc + ca . ■
b c a

28. On the sides of a triangle ABC three equilateral triangles are constructed, such
that the triangles A ' BC and B ' AC are constructed on an interior side and C ' AB is
constructed on an exterior side. If M is the center of
a triangle C ' AB , then prove that the triangle A ' B ' M
is an isosceles triangle and furthermore ∠ A ' MB ' = 23π
holds true.
Solution. Let a, b, c be the affixes of vertices A, B,
C respectively. Then the condition of the given problem
implies that
iπ iπ
a' = b −ce 3
, b' = c − ae3
and
i π i π
1−e 3 1−e 3
iπ
b − ae 3 i π3 i 23π
c' = = be − ae .
iπ
1−e 3
Figure 22 Hence,
m = a +b3+c ' = 13 a 1 − e
i ( 2π
3 ) + 13 b (1 + ei ) = a+2b + i b−6a
π
3 3.
Finally,
(a '− m)e
i 23π
+ m = a 'e
i 23π
+ m 1− e ( i 23π )
) ( 32 − i 23 )
iπ
= b −ceπ e
−i
3 i 23π
(
+ a +2 b + i b −6 a 3
e 3

i π3
= −b + ce + 3a +3b4+b −a + i b −a −4 a −b 3

( )
iπ
i π3 iπ i 2π
− a − 12 + i 23 = ce 3 − ae 3 = c −aeπ = b ',
3
= ce
−i
e 3
the latter implies the statement of the given problem. ■

29. Given a triangle A1 A2 A3 and a point P0 on its plane. Let As = As −3 , s ≥ 4 . We

construct consecutive points P1 , P2 , P3 ,... such that the point Pk , under rotation around
the point Ak +1 at − 23π , maps at Pk +1 . If P2013 = P0 , then the triangle A1 A2 A3 is an
equilateral triangle. Prove it!

174
Solution. By applying the given rotation, we obtain the following
S A ,− 2 π ( p0 ) = p1 , S A ,− 2 π ( p1 ) = p2 ,..., S A ,− 2 π ( p2012 ) = p0 ,
1 2 2013
thus 3 3 3

( S A ,−  S A ,−  ...  S A ,−  S A ,− ) ( p0 ) = p0 .
2013
2π
3 2012
2π
3 2
2π
3 1
2π
3

Further, 3 ( − 23π ) = −2π , so the theorem 5.13 implies that

S A ,− 2 π  S A ,− 2 π  S A ,− 2 π = Sv , (1)
3 3 2 3 1 3
where S w is translation for a vector v. But,
S A ,− 2 π  S A ,− 2 π  ...  S A ,− 2 π  S A ,− 2 π = ( Sv )671 = S671v = S w ,
2013 3 2012 3 2 3 1 3

i.e. S w ( p0 ) = p0 . This means that S w is a translation which has a fixed point. Now the
theorem 5.5. implies that S w is identity i.e. w = 0 , thus v = 0 , which according to (1)
means that S A ,− 2 π  S A ,− 2 π = S A , 2 π . The points A1 and A2 are fixed points for the
2 3 1 3 3 3

rotations S A ,− 2 π and S A ,− 2 π . Thus

1 3 2 3

S A ,− 2 π = e
−i 23π (
z + 1− e
−i 23π ) a1 and S A ,− 2π =e
−i 23π (
z + 1− e
−i 23π ) a2 ,
1 3 2 3
which according to the theorem 5.13 means that an affix a3 of the center of rotation
S A , 2 π satisfies the following equality
3 3

e
−i 2 π
3 (1−e )a +(1−e )a
−i 2 π
3
−i 2 π
3
(
, i.e. a3 1 + e
−i 23π ) = e −i 2π
3 a + a1 .
a3 = 2 1
2
−i 2 π −i 2 π
1−e 3 e 3

The last equality is equivalent to the following equality

iπ
a1 − a3 = (a2 − a3 )e 3 ,
The latter implies that the triangle A1 A2 A3 is an equilateral triangle. ■

30. Determine the points c and d, which together with the points a = 1 + i and
b = 2 + 3i form such a square onto the Oxy plane, that a and b are its two adjacent vertices
and one of the other two vertices is placed on the second quadrant.
Solution. c = 4i and d = −1 + 2i . ■

31. The points a = 1 + i and c = −1 + i 3 are opposite vertices of a square. Determine

the other two vertices of that square.

Solution. The affix of the center of that square is o = i 1+2 3 , and hence

b = 32−1 + i 3+2 3 and d = 1−2 3 + i −1+2 3 . ■

175
32. The complex numbers a = 1 + i and b = 2 + 2i are adjacent vertices of a square.
Determine the other two vertices c and dof that square.
Solution. The given problem has two possible solutions:
c ' = 1 + 3i, d ' = 2i and c '' = 3 + i, d '' = 2 . ■

33. Given a square ABCD and a as the affix of A. Determine the affixes b, c, d of B,
C, D if the origin coincides with:
а) the vertex B,
b) the vertex C,
c) the center of the square.
Solution. а) The given problem has two possible solutions and in both of them
b = 0.
If the square is positively oriented, then the point C is
obtained by rotation of the
π
point A at − 2 around B, thus c ' = −ia . Now, thereby AD = BC we get that d '− a = c '− b ,
thus d ' = a + c ' = (1 − i )a .
If the square is negatively oriented, then the point C is obtained by rotation of a point
��� π2 around B, thus c '' = ia and d '' = a + c '' = (1 + i )a .
А at
b) The given problem has two possible solutions, and in both cases c = 0 , since the
center O of the square is the midpoint of its diagonal, we get that the affix of the center
is o = a2 . If the square is positively oriented then the point B is obtained by rotation of A
around O at π2 , and D by rotation at − π2 , thus b = a2 + i a2 and d = a2 − i a2 . If the square
ABCD is negatively oriented, then b and d will only change their positions.
c) To determine the points B, C, D we only rotate the point A around O at π2 , p, 32π
respectively, and thus b = ia, c = − a, d = −ia . ■

34. Squares are constructed to the outside of

the parallelogram ABCD on each side. Prove that
their centers form a square.
Solution. Let the intersection of diagonals of a
square ABCD coincide with the origin and let a, b,
c, d be the affixes of A, B, C, D, respectively, and
a ', b ', c ', d ', be the affixes of A ', B ', C ', D ' (the
centers of the constructed squares), respectively
(see the figure 23). Then c = − a, d = −b , thus
−i π2
(a − a ')e + a ' = b , i.e. a ' = b1++ai
i
.
Figure 23 Similarly,

176
b ' = c1++bii , c ' = d1++ci
i
and d ' = a1++di
i
,
and hence
b ' = −1a++ibi , c ' = −1b+−iai , d ' = a1−+bi
i
.
Finally, since
− ( a + b ) + (b − a )i
b '− a ' = 1+i
= d '− c '
and
a −b − ( b + a ) i − ( a + b ) + (b − a )i
c '− b ' = 1+i
=i 1+i
= i (b '− a ') ,
we get that the quadrilateral is a square. ■

35. Squares are constructed to the outside of the quadrilateral ABCD on each side. If
A' , B ' , C ' and D ' are centers of the squares constructed on the sides AB, BC, CD and
DA, respectively, E is the midpoint of A ' C ' , F is the midpoint of BD, G is the midpoint
of B ' D ' and H is the midpoint of AC, prove that the quadrilateral EFGH is a square.
Solution. Let the affixes of points
be denoted by the appropriate lower case
letters. The properties of rotation and the
condition of the given problem, imply the
following
a ' = a1−−bi
i
, b ' = b1−−cii , c ' = c1−−di
i
, d ' = d1−−ai
i
.
Further,
a + c −(b + d )i
e = a '2+c ' = 2(1−i ) ,
b + d −( a + c )i
g = b '+2d ' = 2(1−i )
,

f = b +2d , h = a +2 c .
Hence, Figure 24
a + c −(b + d )i
f − e = b +2d − 2(1−i )
= 2b(1+−di ) − 2a(1+−ci ) ,
b + d −( a + c )i
g −h= 2(1−i )
− a +2 c = 2b(1+−di ) − 2a(1+−ci ) , and
b + d −( a + c )i
g− f = 2(1−i )
− b +2d = ( 2b(1+−di ) − 2a(1+−ci ) ) = i ( f − e) ,

therefore quadrilateral EFGH is a square. ■

36. Squares centered at P, Q, R are constructed to the outside of the triangle ABC on
each side. Squares centered at A ', B ', C ' are constructed to the inside of the PQR on each
side. Prove that the points A ', B ', C ' are midpoints of the sides of the triangle ABC.

177
Solution. Let the affixes of points be
denoted by appropriate lower case letters
and let A ', B ', C ' be centers of the squares
constructed on the sides QR, RP, PQ,
respectively (see the figure 25). Then, the
condition of the given problem and the
properties of rotation imply that
p = b1−−cii , q = c1−−ai
i
, r = a1−−bi
i
, (1)
r − qi p − ri q − pi
a ' = 1−i , b ' = 1−i , c ' = 1−i . (2)
If we substitute the expressions for p, q, r, in
(2) we get that
a ' = b +2 c , b ' = c +2a , c ' = a +2 b ,
Figure 25 which was supposed to be proven. ■

37. Let a, b be complex,but not real numbers, and are such that they satisfy the
following | a − b | = 2 and ab = 1. Prove that the quadrilateral ABCD whose vertices have
affixes −1, a, 1, b, respectively, is an isosceles trapezoid.
Solution. Since AC = | 1 − (−1) | = 2 = | b − a | = BD it is sufficient to prove that the
quadrilateral ABCD is a cyclic. But,
[b −( −1)](1− a )
= 1−2a(+bb−−aab
[1−( −1)](b − a ) )
= 2(bb−−aa ) = 12 ∈ R ,
thus, the remark 3.6 b) implies that the points −1, a, 1, b are placed on a same circle,
but they are not on a same line, because in that case it must be the real axis, which is
contradictory with the given condition. Hence, the quadrilateral ABCD is an isosceles
trapezoid. ■

38. Let ABCD be a cyclic quadrilateral and let H A , H B , H C and H D be the

orthocenters of the triangles BCD, CDA, DAB and ABC, respectively. Prove that the
quadrilaterals ABCD and H A H B H C H D are congruent.
Solution. Without loss of generality we assume that the circumcircle of the
quadrilateral ABCD is the unit circle. So, we have that
ha = b + c + d , hb = c + d + a, hc = d + a + b, hd = a + b + c .
To prove that quadrilaterals ABCD and H A H B H C H D are congruent it is sufficient to
prove that for all x, y ∈{a, b, c, d }
| x − y | = | hx − hy |
holds true (why?), which is easy to be proven. Indeed, for example
| ha − hb | = | b + c + d − (c + d + a ) | = | b − a | = | a − b | . ■

178
39. Let a, b, c be complex numbers such that they satisfy the following
a + b + c = 0 and | a | = | b | = | c | .
Then a, b, c are vertices of an equilateral triangle. Prove it!
Solution.
| a − b |2 = (a − b)(a − b) = aa − ab − ba + bb = 4 − aa − ab − ba − bb
= 4 − (a + b)(a + b) = 4 − | a + b |2 = 4 − | −c |2 = 4 − | c |2 = 3.
Hence, | a − b | = 3 . Similarly, it can be proven that | b − c | = 3 and | c − a | = 3 ,
which actually means that a, b, c are vertices of an equilateral triangle. ■

40. Let the complex numbers a, b, c have equal modules and let a, b, c be affixes of
vertices of an equilateral triangle. Prove that the complex numbers ab, bc, ca are also the
affixes of vertices of an equilateral triangle.
Solution. Let | a | = | b | = | c | = r and | a − b | = | b − c | = | c − a | = x hold true.
Hence,
| ab − bc | = | b | ⋅ | a − c | = xr ,
| bc − ca | = | c | ⋅ | b − a | = xr ,
| ca − ab | = | a | ⋅ | c − b | = xr ,
the above imply that ab, bc, ca are affixes of vertices of an equilateral triangle. ■

41. The squares BCDE, CAFG and ABHI are constructed to the outside of the triangle
 ABC, on each side BC, CA and AB. Let GCDQ and EBHP be parallelograms. Prove
that the  APQ is an isosceles right angled triangle.
Solution. The point h is obtained by rotation of a point a around b at π2 in a positive
direction, which means that
iπ
h = a + (b − a )e 2 = (1 − i )a + ib .
Similarly, d = (1 − i )b + ic and g = (1 − i )c + ia . The quadrilateral BCDE is a square, thus
the midpoints of sides CE and BD coincide with each other, which implies that d + b = e + c ,
thus e = (1 + i )b − ic . Analogously, g = (1 + i )c − ia . Further, the quadrilaterals BEPH and
CGQD are parallelograms, thus p + b = e + h and c + q = g + d , that is p = ia + b − ic
and q = −ia + ib + c . Finally, by rotation of the point p around a at π2 we get that
i π2
a + ( p − a )e = a + i (ia + b − ic − a ) = a − a + ib + c − ia = −ia + ib + c = q .
Finally, the point Q is obtained by rotation of the point P around A at π2 , hence the
triangle  APQ is isosceles right angled triangle.■

42. The equilateral triangles BCB1 , CDC1 , DAD1 are constructed to the outside of
a convex quadrilateral ABCD, on each side BC, CD, DA. If the points P, Q and R are the
midpoints of the sides B1C1 , C1D1 and AB, respectively, prove that the triangle PQR is
an equilateral triangle.

179
Solution. The points B1 , C1 , D1 are obtained by rotation of the points B, C, D around
i π3
C, D, A at π3 in a positive direction, respectively. Hence, by letting the ε = e we get
that
b1 = c + (b − c)ε, c1 = d + (c − d )ε, d1 = a + (d − a )ε .
Further, thereby P is the midpoint of B1C1 we get that
b +c bε + c + (1−ε ) d
p = 12 1 = 2
.
cε + d + (1−ε ) a a + b
Similarly, q = 2
. Clearly, r = 2 . Hence,

r + ( p − r )ε = a +2 b + (
bε + c + (1−ε ) d a +b
2
− 2 ε )
cε + a (1−ε ) + d (ε −ε 2 ) +b (1−ε + ε 2 )
= 2
cε + d + (1−ε ) a
= 2
=q
thereby ε 2 − ε + 1 = 0 , (why?). Hence, the point Q is obtained by rotation of the point P
around R. Therefore, the triangle PQR is equilateral triangle. ■

43. Let ABCD be a convex quadrilateral so that AC = BD . On the exterior side of

the quadrilateral on its sides are constructed equilateral triangles. Let O1 , O2 , O3 , O4 be
the centers of triangles constructed on the sides AB, BC, CD, DA, respectively. Prove that
the lines O1O3 and O2O4 are perpendicular to each other.
Solution. Since, a point A is obtained by rotation of B around O1 at 23π in a positive
i 2π
direction, and by taking that ε = e 3 we get that a = o1 + (b − o1 )ε , i.e. o1 = a1−−bεε .
Analogously, o2 = b1−−cεε , o3 = c1−−dεε and o4 = d1−−aεε . Further, to prove that O1O3 ⊥ O2O4
o1 −o3 o −o
it is sufficient to prove that = − 2 4 , i.e. it is sufficient to prove that
o1 −o3 o2 −o4
a −c −(b − d ) ε b − d −( c − a ) ε
=− .
a −c −(b − d ) ε b − d −( c − a ) ε
We can be directly assured in the validity of the latter by using the εε = 1 , i.e. ε = 1ε and
(a − c)(a − c) = | a − c |2 = | b − d |2 = (b − d )(b − d ) . ■

44. Let M and N be two distinct points on a plane of a triangle  ABC so that
AM : BM : CM = AN : BN : CN .
Prove that the line MN passes through the circumcenter of the triangle  ABC.
Solution. Without loss of generality we consider the circumcircle of the triangle
 ABC as the unit circle. Then o = 0 and a = 1a , b = b1 and c = 1c . The proportion
AM : BM = AN : BN is written as following
|a − m|⋅|b − n|
1 = |a −n|⋅|b −m| ,

180
thus,
|a − m|2 |b − n|2 ( a − m )( a − m )(b − n )(b − n )
1= = . (1)
|a − n|2 |b − m|2 ( a − n )( a − n )(b − m )(b − m )
Further,
(a − m)(a − m)(b − n)(b − n) = 1 − m( a )( )
− am + mm 1 − bn − bn + nn ,

(a − n)(a − n)(b − m)(b − m) = (1 − an − an + nn ) (1 − m

b
− bm + mm )
If we substitute in (1) we obtain the following equality
(1 − ma − am + mm ) (1 − bn − bn + nn ) = (1 − an − an + nn ) (1 − mb − bm + mm )
After reducing and dividing the latter by a − b we get that
m n + ( a +b ) mn ( a +b ) mn
ab
− m − ab ab
− mmn
ab
+n− ab
+ mmn + mnn
ab
− mnn = 0 . (2)
Analogously, since the proportion AM : CM = AN : CN , whereby in (2) b is substitute by
c, and thereby symmetry we get that
m n + ( a + c ) mn ( a + c ) mn
ac
− m − ac ac
− mmn
ac
+ n − ac + mmn + mnn ac
− mnn = 0 . (3)
If we subtract (3) from (2), after reducing and dividing the so-obtained equality by b − c
we get the following
m + n − mn + mmn + mn − mnn = 0 .
− abc (4)
abc bc abc bc abc
Further, thereby the symmetry, reapplying the same procedure to the proportions
AM : BM = AN : BN and BM : CM = BN : CN we obtain the following equality
m + n − mn + mmn + mn − mnn = 0 .
− abc (5)
abc ac abc ac abc
Finally, if we subtract (5) from (4), and further the so-obtained equality we divide by
1 − 1 we get that mn − nm = 0 , which is equivalent to
ac bc
m −o = n −o ,
m −o n −o
therefore, the points M, N and O are collinear. ■

45. The quadrilateral ABCD is inscribed into a circle, such that AC is its diameter.
Lines AB and CD meet at M, and the tangents at B and D meet at N. Prove that
MN ⊥ AC .
Solution. Let the quadrilateral ABCD be inscribed into a unit circle. Since AC is a
diameter we have that c = − a . Further, thereby Remark 3.13 holds true we have that affix
of M is
( a +b )cd −(c + d ) ab
m= cd − ab
= 2bd d+ ad
+b
− ab

and affix of N is n = b2+bdd . Further, since

a = 1a , b = b1 and d = d1
we get that
a ( d −b ) a ( d −b )
m − n = b+ d and m − n = = a (bb−+dd ) ,
b+ d

181
The latter means that
m−n = a2 .
m−n
But,
a −c = − 2a = −a 2 ,
a −c 2a
Therefore, the complex gradient of MN and AC holds true
m−n = a 2 = − a −c ,
m−n a −c
which means that MN ⊥ AC . ■

46. Let H be the orthocenter of  ABC and let P be placed on its circumcircle. Let E
be the foot of the altitude BH, the quadrilaterals PAQB and PARC be parallelograms and
AD and HR meet at X. Prove that EX and AP are parallel.
Solution. Without loss of generality we get that the circumcircle of  ABC is the unit
circle. Since Theorem 15.2 we have that h = a + b + c , and since solution of Example 1.9
we get that the affix of E is the following e = 12 a + b + c − ac ( )
b . Further, the equilateral
PAQB is parallelogram, and therefore the midpoints of the line segments PQ and AB
coincide, i.e. q = a + b − p . Analogously, since the quadrilateral PARC is parallelogram
we have that r = a + c − p . But, A, Q, X are collinear, and therefore
x −a a −q p −b pb + a 2 − ax
= = = − pb , i.e. x = abp
.
x −a a −q p −b
Analogously, the points H, R, X are collinear, and therefore
bp bp
x −h p +b x − a −b − c + p + a + c
= h−r = = pb , i.e. x = ab
.
x −h h−r p +b
By equating the obtained equalities for x we express x
( bp
x = 12 2a + b + c − p − c . )
Finally, to prove that EX and AP are parallel, it is sufficient to prove that
e− x a− p
= = − ap
e− x a− p
holds true. The latter can be directly checked if we consider that
( bp
e − x = 12 p + c − a − ac
b
= 2bc )
(b + c )(bp − ac )
.■

47. Let ABCD be a cyclic quadrilateral. The points P and Q are symmetric to C
with respect to the lines AB and AD, respectively. Prove that PQ passes through the
orthocenter of the triangle ABD.
Solution. Without loss of generality we can consider that the quadrilateral ABCD is
inscribed into the unit circle. Since the solution of Example 1.9, the affixes of P and Q
are
p = a + b − ab
c
, q = a + d − ad
c
. (1)

182
The Theorem 15.2 implies that the orthocenter of  ABD has affix h = a + b + d , and
therefore (1) implies that
p −h a +b − ab − a −b − d a + d − ad − a −b − d q −h
= c
1 +1− c −1 −1− 1
= abd
c
= 1 1
c
c 1 1 1
= ,
p −h a b ab a b d a
+ d − ad − a − b − d q −h
The latter means that the line PQ passes through the orthocenter of the triangle ABD. ■

48. Let ABC be a given triangle, H be the orthocenter, O be the sircumcenter and R
be the circumradius of its circumcircle. Let the point D be symmetric to A with respect to
the line BC, the point E be symmetric to B with respect to CA and F be symmetric to C
with respect to AB. Prove that the points D, E and F are collinear if and only if OH = 2 R .
Solution. Without loss of the generality we can consider that the triangle is inscribed
into the unit circle. Then o = 0 , R = 1 and since 15.2 we have h = a + b + c . Thereby
Example 1.9 the affixes of the points D, E and F are the following
d = b + c − bc
a
, e = a + c − ac
b
, f = a + b − ab
c
. (1)
Further, the points D, E and F are collinear if and only if
d −e f −e
= . (2)
d −e f −e
holds true. If (1) is substituted in (2), and after reducing we get that the points D, E and
F are collinear if and only if
(c − a )(abc − a 2b − ab 2 − a 2 c − ac 2 − b 2 c − bc 2 ) = 0 ,
Thereby c − a ≠ 0 , we get that the points D, E and F are collinear if and only if
abc − a 2b − ab 2 − a 2 c − ac 2 − b 2 c − bc 2 = 0 ⇔
a 2b + ab 2 + abc + a 2c + ac 2 + abc +b 2c +bc 2 + abc = 4 ⇔
abc
ab ( a +b + c ) + ac ( a +b + c ) +bc ( a +b + c )
abc
= 4 ⇔
( 1 1
)1
(a + b + c) a + b + c = 4 ⇔ (a + b + c)(a + b + c) = 4 ⇔

| h − o |2 = hh = 4 R 2 ⇔ OH = 2 R . ■

49. Let ABC be a triangle so that the tangent of its circumcirle at the vertex A meets
the midsegment of a triangle (parallel to BC) at A1 . The points B1 and C1 are defined
analogously. Prove that the points A1 , B1 and C1 are collinear and moreover the line
which passes through these points is perpendicular to the Euler line of  ABC.
Solution. Without loss of generality we can consider that the triangle  ABC is
inscribed into the unit circle. Then, h = a + b + c and according to 17.4 the equation of
the Euler line is the following z = h z . Further, if A ', B ', C ' are the midpoints of the line
h
segments BC, CA, AB respectively, then their affixes are
a ' = b +2 c , b ' = c +2a , c ' = a +2 b .

183
Therefore, the equation of the midsegment B ' C ' parallel to BC is
(
z − c +2a = −bc z − a2+acc , ) (1)
And the equation of the tangent at the vertex A is
z + a 2 z = 2a . (2)
Since (2) we obtain the following expression z = 2a − z
and by substituting in (1) we get
a2
the following equation
z − c +2a = −bc 2a −2 z − a2+acc ,
a ( )
a 2 ( a +b + c ) −3abc
whose solution a1 = is the affix of the point A1 . Symmetrically,
2( a 2 −bc )
b 2 ( a +b + c ) −3abc c 2 ( a +b + c ) −3abc
b1 = 2
and c1 = .
2(b −ca ) 2(c 2 − ab )
Further,
a 2 ( a +b + c ) −3abc b 2 ( a +b + c ) −3abc c ( a −b )3 ( a + b + c )
a1 − b1 = − =−
2( a 2 −bc ) 2(b 2 −ca ) 2( a 2 −bc )(b 2 −ca )
it is easy to check that following holds true
a1 −b1 ( a +b + c ) abc
= − ab +bc +ca = − h ,
a1 −b1 h
the latter means that the line A1B1 is perpendicular to the Euler line. Symmetrically, the
line B1C1 is perpendicular to the Euler line, and therefore the points A1 , B1 and C1 are
collinear. ■

50. Let H be the orthocenter of  ABC. Prove that the Euler circles of the triangles
ABC, ABH, BCH, CAH coincide.
Solution. Without loss of generality we can consider that the triangle  ABC is
inscribed into the unit circle. The center of the Euler circle of the triangle  ABC is the
point E with affix e = a +2b +c . The solution of Example 17.9 implies that the circumradius
of the triangles ABC and ABH are congruent and the point O ' with affix o ' = a + b is the
circumcenter of the triangle ABH. Thereby CH ⊥ AB and BC ⊥ AH , C is the orthocenter
of the triangle ABH. If E ' is the center of the Euler circle of the triangle ABH, and H '(h ')
is its orthocenter, we get that h ' = c and E ' is the midpoint of the line segment O ' H ' , i.e.
( a +b ) + c
O ' C , and therefore its affix is e ' = 2
= e . The latter means that the points E and E '
coincide. But, the radius of the Euler circle is half of the length of the circumradius. The
above stated implies that the Euler circles of the triangles ABC and ABH coincide.
Analogously, it can be proven that the Euler circles of triangles ABC and BCH, i.e. of
triangles ABC and CAH coincide.
Finally, the above stated implies that the Euler circles of triangles ABC, ABH, BCH,
CAH coincide. ■

184
51. Let ABCD be a cyclic quadrilateral. Prove that
а) the Euler circles of the triangles ABC, BCD, CDA, DAB meet at a unique point.
b) the centers of the Euler circles of the triangles ABC, BCD, CDA, DAB are vertices
of a cyclic quadrilateral.
Solution. а) Without loss of generality we can consider that the quadrilateral ABCD
is inscribed into the unit circle. The quadrilateral ABCD is cyclic, and therefore the cen-
ters of circumcircles of the triangles ABC, BCD, CDA, DAB coincide. If E1 , E2 , E3 , E4
are the centers of the Euler circles of the triangles ABC, BCD, CDA, DAB, then their af-
fixes will be
e1 = a +2b +c , e2 = b +c2+ d , e3 = c + d2 + a , e4 = d +2a +b ,
respectively. If H1 , H 2 , H 3 , H 4 are the orthocenters of the triangles ABC, BCD, CDA,
DAB, then their affixes will be
h1 = a + b + c, h2 = b + c + d , h3 = c + d + a, h4 = d + a + b ,
respectively. The point E with affix e = a +b +2c + d is midpoint of line segments
DH1 , AH 2 , BH 3 , CH 4 and the following hold true
| e − e1 | = a +b +2c + d − a +2b +c = d2 = 12 ,

| e − e2 | = a +b +2c + d − b +c2+ d = a2 = 12 ,

| e − e3 | = a +b +2c + d − c + d2 + a = b2 = 12 ,

| e − e4 | = a +b +2c + d − d +2a +b = 2c = 12 .
The above means that it belongs to the Euler circles of the triangles ABC, BCD, CDA,
DAB.
b) The proof is directly implied by the following equalities
| e − e1 | = | e − e2 | = | e − e3 | = | e − e4 | = 12 . ■

52. Let AA1 , BB1 and CC1 be the altitudes of  ABC and let AB ≠ AC . Let M be
the midpoint of BC, H be the orthocenter of  ABC and D be the point of intersection of
BC and B1C1 . Prove that DH ⊥ AM .
Solution. Let the circumcircle of  ABC be the unit circle. The condition of the
given problem implies that
(
b1 = 12 a + b + c − ac
b ) (
and c1 = 12 a + b + c − ab
c )
, m = b +2 c and h = a + b + c .
The equation of the line BC is z − b = c −b ( z −b) , i.e.
c −b
z − b = −bcz + c . (1)
c −b
The equation of the line B1C1 is z − b1 = 1 1 ( z −b1 ) , i.e.
c1 −b1
z − b1 = − a 2 ( z −b1 ) . (2)

185
Since (1), z = c +bc
b − z . By substituting in (2) and after reducing we obtain the following

expression for the affix of D

2 2 2 2 2 2
d = a b + a c + ab + ac2 −b c −bc −2abc .
2( a −bc )
Finally, to prove that DH ⊥ AM it is sufficient to check that
d −h = − m−a ,
d −h m−a
(b + c − 2 a )( ab +bc + ca + a 2 )
where d − h = and m − a = b +c2−2a . The details are left as an
2( a 2 −bc )
exercise. ■

53. Let ABC be an acute triangle, so that BC > CA and let O be the circumcenter, H
be the orthocenter and F be the foot of the altitude CH. If the line through F, perpendicular
to OF, intersects the side CA at P, then prove that ∠ FHP = ∠ BAC .
Solution. Without loss of generality we consider the circumcircle of triangle  ABC
as a unit circle. The affix of F is the following f = 12 a + b + c − ab
c (
. The equation of the )
line CA is the following
z − a = c −a ( z − a ) , i.e. z + acz = a + c ,
c −a
and the equation of the line which passes through F and is perpendicular to OF is the
following
f
z − f = − (z − f ) .
f
By solving the system of the last two equations we obtain the affix of P as following

p= f
2 ac f −( a + c )
=
( a+b+c− abc )c2 .
ac f − f b2 +c2

Let ∠ PHF = ϕ and ∠ BAC = α . Then

f −h p − h 2iϕ
= e and c −a = b −a e 2iα ,
f −h p −h c −a b−a

i.e.
( f − h )( p − h ) (c − a )(b − a ) c
e 2iϕ = and e 2iα = = .
( f −h) p −h (c − a )(b − a ) b

Then thereby
2 2
p − h = −b ab +bc2 +ca2 +c , p − h = −c ab +bc +2 ca +2 c ,
b +c ab (b + c )
2
f −h= ab + bc + ca + c and f − h = +bcabc
ab + ca + c 2
2c
it is true that e 2iϕ = bc = e 2iα . The latter implies that ϕ = α
or α = ϕ + π . But,
 ABC is an acute triangle, and therefore α = ϕ + π is not possible. So, ϕ = α , i.e.
∠ FHP = ∠ BAC . ■

186
54. If a Symson line l(P,ABC) passes through Q which is diametrically opposite of
P, then it passes through the centroid of  ABC. Prove that!
Solution. Without loss of generality we consider the circumcircle of triangle  ABC
as the unit circle. According to the condition of a problem, the line l(P,ABC) consists
of a point Q with affix q = − p . Further, according to Example 24.4 the line l(P,ABC)
a +b + c + p
consists of a point O p with affix o p = 2
, and the centroid T of  ABC has affix
t = a +3b +c . Then,
a +b + c + p
t −q a +b + c +3 p
o p −q
= 3
a +b + c + p
= 23 a +b +c +3 p = 23 .
2
+p
Since Corollary 1.4. we get that points Q, T and are collinear, which actually means that
T is placed on the line l(P,ABC). ■

55. Prove that the Symson line of any point P (P is a point placed on a circumcircle
of  ABC) bisects a line segment PH where H is the orthocenter of a  ABC.
Solution. Without loss of generality we consider the circumcircle of the triangle
 ABC as a unit circle. The orthocenter H of  ABC has affix h = a + b + c , and therefore
a +b + c + p
the midpoint Q of the line segment PH has affix q = 2
, which obviously satisfies
the equation
z − z acb
p
+ abc
2p
(a + b + c + p ) − 12 (a + b + c + p ) = 0
of the Symson line l(P,ABC). The latter means that l(P,ABC) bisects the line segment
PH. ■

56. Let  ABC be a triangle and let D be on the circumcircle of the triangle  ABC.
Determine the locus of meeting points of the Symson lines l(A,BCD), l(B,ACD), l(C,ABD),
l(D,ABC), when D moves on a circumcircle of  ABC.
Solution. Without loss of generality we consider the circumcircle of triangle  ABC
as a unit circle. If a, b, c, d are the affixes of A, B, C, D, respectively, then according
to the Example 23.4 the point of intersection of lines l(A,BCD), l(B,ACD), l(C,ABD),
l(D,ABC) has affix x = 12 (a + b + c + d ) So, the required locus of points is a set of all
points x = 12 (a + b + c + d ) , when d moves on a circle. That actually is a circle with
radius 12 and center a +2b +c , i.e. it is a circle centered at the midpoint of the line segment
whose ends are the orthocenter and the circumenter of  ABC, and the radius is congruent
to half of the circumradius.■

57. Let  ABC be such a triangle that AB ≠ AC and let D be a point of intersection
of the tangent to the circumcircle of  ABC at A and the line BC. If E and F are such
points of bisectors of line segments AB and AC, respectively, that the lines BE and CF are
perpendicular to BC, then the points D, E and F are collinear. Prove it!

187
Solution. Without loss of generality we consider the circumcircle of triangle  ABC
as a unit circle. The equation of the line BC is the following
z − b = c −b ( z − b) , i.e. z + bcz = b + c ,
c −b
and the equation of the tangent at A is z + a 2 z = 2a . By solving the system of the last
two equations we obtain d, the affix of point of intersection between the line BC and the
tangent to the circle at A, as following
a 2 (b + c ) − 2 abc
d= .
a 2 −bc
The point E is placed on the bisector of the line segment AB, and thus OE ⊥ AB ,
therefore e−o = − a −b , i.e. e = ab
e . Further, BE ⊥ BC implies that e −b = − c −b , and
e −o a −b e −b c −b
c + e − b e c + e −b a ( c −b ) a (b −c )
therefore e = bc . So, ab = bc
, thus e = c −a
. Analogously, f = b−a .
Finally,
a 2 (b + c ) − 2 abc a (b −c ) ab ( a −c )(b + c − 2 a )
d− f = − b−a = and
a 2 −bc ( a 2 −bc )(b − a )
a 2 (b + c ) − 2 abc a ( c −b ) ac ( a −b )(b + c − 2 a )
d −e= − c −a =
a 2 −bc ( a 2 −bc )(c − a )
imply
d− f b ( a −c ) 2 a 2b 2 c ( c − a ) 2 b ( a −c ) 2 d− f
= 2
= 2 2
= d −e , 2
= 2
d −e
c ( a −b ) c a b (b − a ) c ( a −b )
the above means that the points D, E and F are collinear. ■

58. (Brokar theorem). Let ABCD be a cyclic quadrilateral. The lines AB and CD
intersect at E, the lines AD and BC intersect at F and the lines AC and BD intersect at
G. Prove that the circumcenter O of the quadrilateral coincides to the orthocenter of
 EFG.
Solution. Let assume that ABCD is inscribed into the unit circle. According to the
Remark 3.13 c) the affixes of E, F and G are
ab (c + d ) −cd ( a +b ) ad (b + c ) −bc ( a + d ) ac (b + d ) −bd ( a + c )
e= ab −cd
, f = ad −bc
, g= ac −bd
. (1)
To prove that O is the orthocenter of  EFG it is sufficient to prove that OF ⊥ EG and
OG ⊥ EF . Since (1) it is easy to find that
f −o ad (b + c ) −bc ( a + d )
= a + d −(b + c )
, (2)
f −o
( a − d )(b −c )[(b + c ) ad −( a + d )bc ]
e−g = ( ab −cd )( ac −bd ) (3)
( a − d )(b −c )(b + c −( a + d ))
e−g = ( ab −cd )( ac −bd ) . (4)
Now, (2), (3) and (4) imply that

188
( a − d )( b − c )[( b + c ) ad − ( a + d ) bc ]
e− g ( ab − cd )( ac −bd ) (b + c ) ad −( a + d )bc ad (b +cc ) −bc ( a + d ) f −o
= ( a − d )( b − c )(( b + c − ( a + d ))
= b + c −( a + d )
=− a + d −(b + c )
=− ,
e− g f −o
( ab − cd )( ac −bd )
The latter means that OF ⊥ EG . Since the symmetry we conclude that OG ⊥ EF , i.e. O
is the orthocenter of  EFG. ■

59. Let ABC be an isosceles triangle, AB = AC . Let P be a point on the extension

of the side BC and X and Y be the points on the sides AB and AC, respectively, so that
 , then PT ⊥ XY . Prove it!
PX || AC , PY || AB . If T is the midpoint of the arc BC
Solution. Let the circumcircle of the triangle  ABC be the unit circle and a = 1 .
Then c = b and t = −1 . Further, since P is on the line BC, its affix p satisfies the following
p = b + b1 − p . Further, since X is a point on the side AB it is true that x = 1+bb− x , and since
b ( p +1)
PX || AC we get that x = p + bp − bx . So, x = b +1 thereby the last three equations.
1+ c − y p +1
Analogously, y = c
and y = p + cp − cy , thus y = b +1 . Finally,
( p +1)( b −1)
x− y b +1 p +1 p −t
= ( p +1)( b −1)
=− =− ,
x− y − b +1 p +1 p −t

implies that PT ⊥ XY . ■

60. Let ABCD be a cyclic quadrilateral and let K, L, M, N be the midpoints of the
sides AB, BC, CD, DA, respectively. Prove that the orthocenters of the triangles AKN,
BKL, CLM, DMN form a parallelogram.
Solution. Let the circumcircle of a quadrilateral be the unit circle. The affixes of the
points K, L, M, N are
k = a +2 b , l = b +2 c , m = c +2d , n = d +2 a .
We have to determine the affix h1 of the orthocenter H1 of the triangle AKN. Since
KH1 ⊥ AN , NH1 ⊥ AK , the following holds true
k − h1 n − h1
= − a −n = − a −d = ad and = − a −k = − a −b = ab ,
k − h1 a −n a −d n − h1 a −k a −b
that is
kad − k + h1 nab − n + h1
h1 = ad
and h1 = ab
,
which imply that
h1 = 2a +2b + d .
Analogously, the affixes of the orthocenters of the triangles BKL, CLM, DMN are
h2 = 2b +2c + a , h3 = 2c +2b + d , h4 = 2d +2a +c ,
respectively. Finally, since
h1 + h3 h2 + h4
2
=a+b+c+d = 2
,

189
the midpoints of the diagonals of the quadrilateral coincide. The latter means that the
quadrilateral is a parallelogram. ■

61. The incircle of a  ABC centered at O tangents the sides AB, BC, CA at M, K, E. If
P = MK ∩ AC , then OP ⊥ BE . Prove it!
Solution. Let the incircle of  ABC be the unit circle. Then, according to the Remark
3.13 d), it is true that
a = e2+em
m
2 mk .
and b = m +k
Since P is on the chord MK we get that P, M, K are collinear, and therefore their affixes
satisfy the following
m+ k − p
p= mk
.
Further, P is on the line AC. Thereby this line tangents the circle at E, we get that PE ⊥ OE ,
and therefore
e− p 2e − p
= − e−o = −e 2 , i.e. p =
e− p e −o e2
holds true. By equating the last two expressions for p and after reducing, we obtain that
the affix of P is the following
( m + k )e2 − 2 mke
p= .
e2 − mk
Finally, it is easy to be checked that the affixes o, p, b, e of the points O, P, B, E
satisfy the following
p −o
= − e −b ,
p −o e −b
(check it!). The latter means that OP ⊥ BE . ■

62. A circle centered at O is incircle of a quadrilateral ABCD and tangents the sides
AB, BC, CD, DA at K, L, M, N respectively. The lines KL and MN meet at S. Prove that
OS ⊥ BD .
Solution. Let the incircle be the unit circle. Then Remark 13.3 implies that
a = n2+nkk , b = k2+kll , c = l2+lm 2 mn , s = kl ( m + n ) − mn ( k +l ) .
,d=m
m +n kl − mn
Further,
kl ( m + n ) − mn ( k +l ) k +l −( m + n )
s= = kl −mn ,
kl − mn
2 kl ( m + n ) − 2 mn ( k +l )
b−d = ( k +l )( m + n )
and
2( m + n ) − 2( k +l )
b − d = ( k +l )( m+ n)
and therefore

190
kl ( m + n ) − mn ( k + l ) 2 kl ( m + n ) − 2 mn ( k + l )
s −o kl ( m + n ) − mn ( k +l )
= kl − mn
k +l −( m + n )
= − ( m + n ) −( k +l ) =− ( k + l )( m + n )
2( m + n )−2( k +l )
= − b−d ,
s −o b−d
kl − mn ( k + l )( m + n )
The latter actually means that OS ⊥ BD . ■

63. Let ABC be an acute triangle, whose incircle tangents the sides AB and AC at Q
and R, respectively. Let X and Y be the points of intersection between the bisectors of the
angles ∠ ACB and ∠ ABC with the line QR, respectively and let Z be the midpoint of
the line segment BC. Prove that the triangle XYZ is an equilateral triangle if and only if
∠ BAC = π3 .
Solution. Without loss of generality we consider the incircle of the triangle as a unit
circle. Let P be the point where the line BC tangents the incircle. Then,
2 qr 2 pr 2 pq pr pq
a = q + r , b = p + r , c = p + q and z = b +2 c = p + r + p + q .
Since the bisector of ∠ ACB passes through the center of the incircle, we get that the points
B, O and X are collinear, and therefore the affix x of the point X satisfies the following
2 pq 2 pr
x = αc = α p + q , α ∈ R . Similarly, the affix y of Y is y = β p + r , β∈ R . Further, the
constants α and β are determined by the following given conditions that X , Y ∈ QR , i.e.
the points X, Q, R are collinear likewise Y, Q, R. So,
q −r q −r y −q
= x −r and = ,
q −r x −r q −r y −q
By direct calculation we get that,
( p + q )( q + r ) ( p + r )( q + r )
α = 2q ( p + r ) and β = 2r ( p + q ) ,
thus
p(q+r ) p(q+r )
x = p+r and y = p + q .
We have to prove that
∠ BAC = π3 if and only if  XYZ is an equilateral triangle.
i 2π
The first condition is equivalent to ∠ QOR = 23π , i.e. q = re 3 . The second condition is
iπ
equivalent to y − z = ( x − z )e 3 . So,
p(q+r )
( pr
y − z = p + q − p + r + p + q = ( p + q )( p + r ) ,
pq
) pr ( r − q )

x − z = p+r
p(q+r )
−(
pr
p+r
+ p+q
pq
)= pq ( q − r )
( p + q )( p + r )
.
Therefore,
i π3 pr ( r − q ) pq ( q − r ) i π3
y − z = ( x − z )e ⇔ ( p + q )( p + r ) = ( p + q )( p + r ) e ⇔
i π3 i 23π
r = − qe ⇔ q = re ,

191
i 23π
The last equivalence is implied if the equation before the last is multiplied by e and
also having on mind that eiπ = −1 . ■

64. (Newton theorem). Let ABCD be a cyclic quadrilateral. Let M and N be the
midpoints of the diagonals AC and BD and S be the center of its incircle. Prove that M,
N and S are collinear.
Solution. Let the quadrilateral be inscribed into the unit circle and let P, Q, R, S be
the points where the circles tangents the sides AB, BC, CD, DA, respectively. Then,
2 ps 2 pq 2 qr
a = p + s , b = p + q , c = q + r , d = r2+rss ,
and therefore
pqs + prs + pqr + qrs p+q+r +s
m = a +2 c = ( p + s )( q + r )
, m = ( p + s )( q + r ) ,
pqr + pqs + prs + qrs p+q+r +s
n = b +2d = ( p + q )( r + s )
, n = ( p + q )( r + s ) .
Thus,
m −o pqr + pqs + prs + qrs
= p+q+r +s
= n −o ,
m −o n −o
The latter actually means that M, N and S are collinear. ■

65. Let ABCD be a quadrilateral and let its incircle tangents the sides AB, BC, CD,
DA at points M, N, P, Q, respectively. Prove that the lines AC, BD, MP, NQ are concurrent
(meet at a unique point).
Solution. Let the incircle of the quadrilateral ABCD be the unit circle. Therefore,
2 mn , d = 2 pq .
b= m +n p+q
If X = MP ∩ NQ , then
mp ( n + q ) − nq ( m + p )
x= mp − nq
.
So,
mn ( p + q ) − pq ( m + n ) p + q −( m + n )
b−d =2 ( m + n )( p + q )
, b − d = 2 ( m+ n)( p + q ) ,
( m − n )[ mn ( p + q ) − pq ( m + n )] ( m − n )[ p + q −( m + n )]
b−x= ( m + n )( mp − nq )
, b−x= ( m + n )( mp − nq )
,
Thus,
mn ( p + q ) − pq ( m + n )
mn ( p + q ) − pq ( m + n ) 2
b− x = = ( m + n )( p + q )
= b−d ,
b− x p + q −( m + n ) p + q −( m + n )
2 ( m + n )( p + q ) b−d

The above means that X is placed on the line BD. Further, by applying the symmetry, we
conclude that X is placed on the line AC. The above stated means that the lines AC, BD,
MP, NQ are concurrent. ■

192
66. The incircle of a triangle ABC tangents the sides BC, CA, AB at D, E, F,
respectively, and X, Y, Z are the midpoints of the sides EF, FD, DE, respectively. Prove
that the center of incircle is placed on the line determined by the circumcenters of the
triangles XYZ and ABC.
Solution. Let the incircle of  ABC be the unit circle. According to the Remark
22.13 the affix o of the circumcenter of the triangle ABC is the following
2 def ( d + e + f )
o = ( d +e)(e+ f )( f + d ) .
Further, the affixes of points X, Y, Z are
e+ f d+ f
x = 2 , y = 2 , z = d 2+e ,
thereby the Remark 3.4 and Example 3.3 b),the affix o ' of circumcenter of  XYZ is:
x x ( z − y )+ y y ( x − z )+ z z ( y − x ) d +e+ f
o' = = 2
.
x y + y z + z x − xy − yz − zx
So,
2 def ( d + e + f ) 2( de + ef + fd )
o − i = ( d +e)(e+ f )( f + d ) , o − i = ( d +e)(e+ f )( f + d ) ,
d +e+ f de + ef + fd
o '− i = 2
, o' − i = 2 def
,
thus,
o −i def ( d + e + f )
= de+ef + fe = o '−i .
o −i o '−i
The above means that the points I , O, O ' are collinear, which actually was supposed to
be proven. ■

67. The incircle of a triangle ABC, centered at I, tangents the sides BC, CA, AB at
D, E, F, respectively. Let AI ∩ EF = K , ED ∩ KC = N and DF ∩ KB = M . Prove that
MN || BC .
Solution. Let the triangle ABC be inscribed into the unit circle. Then,
2 fe 2 fd
a = f +e , b = f + d and c = e2+edd .
e+ f
Further, the affix of the midpoint of the line segment EF is 2
and since
2 ef f +e
a −o e+ f −o
= 2 ef
= ef = 2
f +e
a −o −o
e+ f 2
e+ f
we get that k = 2 . Further, the equations of the lines DF and KB are
d− f
z−d = ( z − d ) and z − k = k −b ( z − k ) , (1)
d− f k −b
respectively, and if we substitute for
e+ f 2 fd
k = 2 and b = f + d ,
by reducing the system (1) we obtain the following expression for the affix of M

193
4ef 2 d + efd 2 −e2 d 2 −e2 f 2 − 2 f 2 d 2 − f 3e
m= .
6efd −e2 d −ed 2 −ef 2 −e2 f − d 2 f − df 2
The equations of the lines ED and KC are
z − d = d −e ( z − d ) and z − k = k −c ( z − k ) , (2)
d −e k −c
e+ f
respectively, and if we substitute the following expressions for k = 2 and c = e2+edd ,
and by solving the system (2) we obtain the following expression for the affix of N
4e2 fd + efd 2 − f 2 d 2 −e2 f 2 − 2e2 d 2 −e3 f
n= .
6efd −e2 d −ed 2 −ef 2 −e2 f − d 2 f − df 2
Finally, it is sufficient to prove that MN ⊥ ID , namely to prove that
m − n = d −o = − d 2 .
m−n d −o
Details are left as your exercise. ■

68. Let  ABC be any triangle, with orthocenter H, circumcenter O, incenter I

and K the point where the side BC tangents the incircle of  ABC. If IO || BC , then
AO || HK . Prove it!
Solution. Let the incircle of  ABC be the unit and let it tangent BC, CA, AB at K, L,
M, respectively. According to the Remark 22.13 it is true that
2 klm ( k +l + m ) 2( k 2l 2 +l 2 m 2 + m 2 k 2 + klm ( k +l + m ) )
o = ( k +l )(l + m)( m+ k ) and h = ( k +l )(l + m )( m + k )
.

Further, IO || BC implies that IO ⊥ KL , and therefore

o −i = − k −i = − k 2
o −i k −i
By substituting for o and o , and after reducing we obtain that
klm(k + l + m) + k 2 (kl + lk + mk ) = 0 . (1)
We will prove that if the condition (1) is satisfied, then AO || HK . The affix of A is the
2 ml , thus
following a = m +l

a−o = m 2 ml − 2 klm ( k +l + m ) = 2 m 2l 2 2
and a − o = ( k +l )(l2+km)( m+ k ) .
+l ( k +l )(l + m )( m + k ) ( k +l )(l + m )( m + k )
On the other hand, if we use the condition (1) we get that
( kl +lm + mk )2 [( k +l + m )2 + k 2 ] ( k +l + m ) 2 + k 2
h−k = 2
and h − k = ( k +l )(l + m)( m+ k ) .
( k +l + m ) ( k +l )(l + m )( m + k )
Thus,
h−k ( kl +lm + mk )2 2 2
= 2
= (acording to (1))= m 2l = a −o .
h−k ( k +l + m ) k a −o
So, AO || HK . ■

194
69. Let AH1 , BH 2 , CH 3 be the altitudes of an acute triangle  ABC. The incircle
of the  ABC tangents the sides BC, CA, AB at points T1 , T2 , T3 , respectively. Let the
lines l1 , l2 , l3 be symmetric to the lines H 2 H 3 , H 3 H1 , H1H 2 with respect to the lines
T2T3 , T3T1 , T1T2 , respectively. Prove that the lines l1 , l2 , l3 form a triangle whose vertices
are on the incircle of the  ABC.
2t t
Solution. Let the incircle of the  ABC be the unit circle. Thus c = t +1t2 . Let’s
1 2
determine the affix h3 of the point H 3 . Since the given conditions H 3T3 ⊥ T3 I and
H 3C || T3 I it is true that
h3 −t t3 −o h3 −c t3 −o
=− = −t32 and = = t32 .
h3 −t3 t3 −o h3 −c t3 −o
By solving the system of the last two equations we get the following expression for h3
t t −t 2
h3 = 12 (2t3 + c − ct32 ) = t3 + 1t 2+t 3 .
1 2
t t −t 2
Analogously h2 = t2 + 1t 3+t 2 . Further in order to determine the line l1 which is symmetric
1 3
to H 2 H 3 with respect to the line T2T3 , it is sufficient to determine the points P2 and
P3 which are symmetric to H 2 and H 3 with respect to the line T2T3 , respectively. The
equation of the line T2T3 is
t3 −t2
z − t2 = ( z − t2 ) .
t3 −t2
Since the example 1.9, the affix of P2 is
h2 (t2 −t3 ) +t2t3 −t2 t3 t (t 2 + t 2 )
p2 = = t1 (t2 +t3 ) .
t2 −t3 2 1 3

Analogously, the affix of P3 is

h3 (t2 −t3 ) +t2t3 −t2 t3 t (t 2 + t 2 )
p3 = = t1 (t2 +t3 ) .
t2 −t3 3 1 2
Further,
t (t 2 + t 2 ) t (t 2 + t 2 ) t 2 (t 2 +t 2 )(t −t )
p2 − p3 = t1 (t2 +t3 ) − t1 (t2 +t3 ) = t1t (2t +3t )(t3 +t2 ) ,
2 1 3 3 1 2 23 1 2 1 3

thus, the equation of the line l1 is

p2 − p3
z − p2 = ( z − p2 ) ,
p2 − p3
i.e.
t (t 2 + t 2 )  t 2 +t 2 
z − t1 (t2 +t3 ) = −t12  z − t t 2(t +3t )  . (1)
2 1 3  23 1 3 

Analogously, the equation of the line l2 which is symmetric to H 3 H1 with respect to the
line T3T1 is
t (t 2 + t 2 )  t 2 +t 2 
z − t2 (t3 +t1 ) = −t22  z − t t 3(t +1 t )  . (2)
3 2 1  31 2 1 

195
tt
By solving the system of the equations (1) and (2) we get that m1 = 1t 2 is affix of the point
3
t t
of intersection of the lines l1 and l2 . Analogously, m2 = 2t 3 is affix of point of intersection
1
tt
of the lines l2 and l3 has affix and m3 = t3 1 is affix of the point of intersection of the
2
lines l3 and l1 . Finally, the statement is implied by the fact that | m1 | = | m2 | = | m3 | = 1 .
Prove it ■

70. Let O and R be the circumcenter and the circumradius of  ABC, and Z and r be
the incenter and the inradius of  ABC, respectively. If K is the centroid of the triangle
whose vertices are the points where the incircle tangents the sides of  ABC prove that
Z ∈ OK and also that OZ : ZK = 3rR .
Solution. Let the incircle of the triangle  ABC be the unit circle and let d, e, f be
the affixes of its tangent points with the sides BC, CA, AB respectively. According to
2 def ( d + e + f ) d +e+ f
the Remark 22.13 it is true that o = ( d +e)(e+ f )( f + d ) . Since Theorem 15.8, k = 3
.
Thus,
2 def ( d + e + f ) d +e+ f
−0 −0
o− z = ( d + e )( e + f )( f + d )
= def = 3
= k −z
o− z 2( d +e+ f ) d +e+ f k −z
( d + e )( e + f )( f + d )
−0 3 def
−0
Therefore the points K, Z and O are collinear. Further,
2 def ( d + e + f )
OZ |o − z| ( d + e )( e + f )( f + d )
= |z −k | = d +e+ f
= 1
3 = 3rR .
ZK 2
( d + e)(e + f )( f + d )
3
The latter was supposed to be proven. ■

71. Let P be the intersection of the diagonals of a convex quadrilateral ABCD so

that AB = AC = BD and let O and I be the circumcenter and the incenter of  ABP,
respectively. Prove that if O ≠ I , then OI ⊥ CD .
Solution. Let  ABP be inscribed into the unit circle and let u, v, w be complex
numbers as given in Theorem 22.14, so that a = u 2 , b = v 2 , p = w2 , hold true. Then,
according to the stated theorem i = −uv − vw − wu . But, AB = AC , and therefore for
α = ∠CAB it is true that
c − a = eiα (b − a ) , (1)
(make a figure). Further, the points A, C and P are collinear, and therefore
α = ∠ CAB = ∠ PAB . The latter means
− vw−u 2 2i α2 v 2 −u 2
=e , i.e. eiα = − wv .
− vw−u 2 v 2 −u 2
By substituting in (1) we get c − u 2 = − wv (v 2 − u 2 ) , i.e.

196
2 2 2
c = u w+uv v −v w . (2)
Analogously,
2 2 2
d = v w+vuu −u w . (3)
Finally, (2) and (3) imply
(u 2 −v 2 )(uv + vw+ wu )
c−d = uv
,
therefore
c −d = − uvu++vw + wu uvw = − i −o
v+ w .
c −d i −o
The latter means that OI ⊥ CD . ■

72. Let I be the circumcenter of  ABC, AB ≠ AC . The point O1 is symmetric to

O, the circumcenter of  ABC, with respect to the line BC. Prove that the points A, I and
O1 are collinear if and only if ∠ BAC = 60° .
Solution. Let the circumcircle of  ABC be the unit circle. According to the Theorem
22.14 there exist complex numbers u, v, w so that
a = u 2 , b = v 2 , c = w2 and i = −uv − vw − wu .
According to the Example 1.9, the affix of O1 is the following
0(b −c ) +bc −bc
o1 = = b + c = v 2 + w2 .
b −c
Further, the points A, I and O1 are collinear if and only if
a −o1
= a −i ,
a −o1 a −i
i.e. if and only if
v 2 + w 2 −u 2 2
= u +uv +vw+ wu ⇔
v 2 + u 2 −u 2 u 2 +uv + vw+ wu
v 2 + w 2 −u 2 u (u + v + w) + vw 2
u 2 v 2 w2 = u vw ⇔
u 2 ( v 2 + w2 ) − v 2 w2 vw+uw+uv +u 2
v 2 + w 2 −u 2 vw = 1 ⇔
u ( v 2 + w2 ) − v 2 w2
2

wv3 + v 2 w + vw3 − vwu 2 − u 2 v 2 − u 2 w2 = 0 ⇔

( wv − u 2 )(vw + v 2 + w2 ) = 0.
That is, the points A, I and O1 are collinear if and only if either u 2 = vw or
vw + v 2 + w2 = 0 . If u 2 = vw , then
u 2 −o o −( − vw)
= u 4 = v 2 w2 = ,
u 2 −o o −( − vw)
which means that A, O and O1 are collinear, thus AB ≠ AC , which is contradictory. So,
the points A, O and O1 are collinear if and only if the following holds true

197
vw + v 2 + w2 = 0 ,
i.e.
(v + w) 2 = vw ,
thus AB ≠ AC , which is contradictory. So, the points A, O and O1 are collinear if and
only if the following holds true
vw + v 2 + w2 = 0 ,
i.e.
(v + w) 2 = vw ,
therefore
| v + w |2 = | vw | = 1 ⇔ | vw + w2 | = | vw | = 1 ⇔ | w2 − (−vw) | = | o − (−vw) | ,
i.e. if and only if the triangle with vertices w2 , − vw, o is an equilateral, that is if and only
if ∠ BAC = 60° . ■

73. Let  ABC be any triangle. Let A1 , B1 , C1 be the midpoints of the sides BC, CA,
AB, respectively, P, Q, R be the points where the incircle k tangents the sides BC, CA, AB;
P1 , Q1 , R1 be the midpoints of the arcs QR, RP, PQ and P2 , Q2 , R2 be the midpoints of
arcs QPR, RPQ, PRQ, respectively. Prove that both lines A1P1 , B1Q1 and C1R1 , and lines
A1P2 , B1Q2 and C1R2 are concurrent.
Guidelines. Let the incircle be a unit circle. Since Theorem 22.14 there exist complex
numbers u, v, w so that
p = u 2 , q = v 2 , r = w2 and p1 = −vw, q1 = − wu , r1 = −uv .
The points P2 , Q2 , R2 are symmetric with respect to the center of k with the points
P1 , Q1 , R1 , thus
p2 = vw, q2 = wu , r2 = uv .
Further,
2 2 2 2 2 2
a = 22v w 2 , b = 2 w2 u 2 , c = 22u v 2 ,
v +w w +u u +v
therefore,
2 2 2 2 2 2 2 2 2 2 2 2
a1 = w2 u 2 + u2 v 2 , b = v2 w 2 + u2 v 2 , c = w2 u 2 + v2 w 2 .
w +u u +v v +w u +v w +u v +w
Use that the equations of lines A1P1 , B1Q1 are
a −p b −q
z − a1 = 1 1 ( z − a1 ) and z − b1 = 1 1 ( z − b1 ) ,
a1 − p1 b1 − q1
Determine the affix n of the point of intersection and verify whether it satisfies the
equation of the line C1R1 .
The second part of the statement should be proved analogously. ■

74. The squares ABB ' B '' , ACC ' C '' and BCXY are constructed on the outside of a
triangle  ABC. Let P be the center of the square BCXY. Prove that the lines CB '' , BC ''
and AP are concurrent.

198
Solution. Let the point A coincide with the origin, i.e. a = 0 . Thus
c ''− a = eiπ / 2 (c − a ) , i.e. c '' = ic .
Similarly,
b '' = −ib , x − c = eiπ / 2 (b − c) , i.e. x = (1 − i )c + ib
And since P is the midpoint of BX we get that
p = 1+2 i b + 1−2 i c .
The equation of the lines BC '' and AP are
z − b = b −c '' ( z − b) (1)
b −c ''
a− p
z−a= ( z − a ) (2)
a− p
By solving the system of equations (1) and (2), we obtain the affix of Q, point of
intersection between the lines BC '' and AP as following
(bc +bc )[(1+i )b + (1−i )c ]
q= .
(b −ic )(b +ic )
The equation of line B '' C is following
z − b '' = b ''−c ( z − b '') (3)
b ''−c
By solving the system of equations (2) and (3), we obtain the affix of Q ' , the point of
intersection of the lines B '' C and AP as following
(bc +bc )[(1+i )b + (1−i )c ]
q' = .
(b −ic )(b +ic )
Finally, the statement of the given problem is implied by the equality q ' = q . ■

75. Let ABCD be any quadrilateral, O be the intersection of its diagonals, M be the
midpoint of the side AB and N be the midpoint of the side CD. Prove that if OM ⊥ CD
and ON ⊥ AB , then ABCD is a cyclic quadrilateral.
Solution. Let the intersection of the diagonals coincide with the origin, i.e. o = 0
. The points A, O and C are collinear and also the points B, O and D are collinear, and
thus ac = ca and bd = db . Further, m = a +2 b and n = c +2d . Thereby, OM ⊥ CD and
ON ⊥ AB the following holds true
c + d −o a + b −o
2
= − a −b and 2
= − c −d ,
c + d −o a −b a + b −o c −d
2 2
i.e.
da ( ab − 2bb + ab ) da ( ab + 2bb + ab )
c= and c = ,
b ( ab − 2 aa + ab ) b ( ab + 2 aa + ab )
therefore
(ab + ab)(aa − bb) = 0 . (1)

199
We have to prove that the condition (1) is sufficient for the points A, B, C, D to be on a
same circle, which according to Remark 25.3 means that the condition (1) is sufficient to
(c − d )(b − a )
(b − d )(c − a )
∈R , that is
(c − d )(b − a ) (b − d )(c − a )
= (2)
(c − d )(b − a ) (b − d )(c − a )

The points B, O and D are collinear, and thus b −d = b and also points A, O and C are
b−d b
collinear and thus a −c a
= . If ab + ab = 0 , then
a −c a
2 ab ( a −b )
c−d =d ,
b ( ab − 2 aa + ab )
and if aa − bb = 0 , then
d ( a −b )( ab + ab )
c−d = .
b ( ab − 2 aa + ab )
By direct checking we assure that in both cases the condition (2) is satisfied. The latter
means that the points A, B, C, D are placed on the same circle. Details are left as your
exercise. ■

76. Let F be the point on the base AB of a trapezoid ABCD, such that DF = CF ,
E = AC ∩ BD and O1 and O2 be the circumcenters of the triangles ADF and FBC,
respectively. Prove that FE ⊥ O1O2 .
Solution. Let the origin coincide with the point F, i.e. f = o and let d = c . CD || AF
implies that
a− f
= c −d = −1,
a− f c −d
i.e. a = − a and similarly, b = −b . Further, the above statated and the Example 3.3 imply
that
ad ( d − a ) c( a +c) cb (c −b ) c ( c +b )
o1 = = and o2 = = .
ad − ad c +c bc −bc c +c
The equations of the lines AC and BD are
z − a = c −a ( z − a ) and z − b = d −b ( z − b) .
c −a d −b
The solution of the system consists of the last two equations is the affix of the point E,
thus
e = ac −bc .
a + c −b − c
Finally,
o1 − o2 = ca −cb
c +c
By direct checking we get that the following holds true
o1 −o2 e− f
=− ,
o1 −o2 e− f
therefore FE ⊥ O1O2 . ■

200
77. Let the diagonals of a convex quadrilateral ABCD meet at O and let T1 and T2
be the centroids of the triangles AOD and BOC, and H1 and H 2 be the orthocenters of
the triangles AOD and BOC, respectively. Prove that T1T2 ⊥ H1H 2 .
Solution. Let the point O and the origin coincide. Then the affixes of the orthocenters
H1 and H 2 and centroids T1 and T2 are
( a −b )( ab + ab ) (c − d )(cd + cd )
h1 = , h2 = , t1 = a +3d and t2 = b +3 c .
ab − ab cd −cd
The points A, C and O are collinear, and also B, D and O are collinear, therefore c = ca
a
and d = dbb , that is
(c − d )(ba +ba )
h2 = .
ab − ab
Further,
( a + d −b −c )( ab + ab )
h1 − h2 = , t1 − t2 = a + d 3−b −c
ab − ab
By direct checking we obtain that
t1 −t2 h1 − h2
=− ,
t1 −t2 h1 − h2
therefore T1T2 ⊥ H1H 2 . The details are left as your exercise. ■

78. Let the tangents of a circle Γ at points A and B meet at C. The circle Γ1 is such
a circle that passes through C, tangents the line AB at B and meets Γ at M. Prove that the
line AM bisects the line segment BC.
Solution. Let Γ be the unit circle. Then c = a2+abb . Let O1 be the center of the circle
Γ1 . Then O1B ⊥ AB , thus
o1 −b
= − a −b = ab ,
o1 −b a −b
o1 + a −b
So, we obtain that o1 = ab
. Further, | o1 − b | = | o1 − c | , and by squaring we get
o1
(o1 − b)(o1 − b) = (o1 − c)(o1 − c) , i.e. o1 = − b(aa−+bb) .
b2
Thus,
o1 + a −b o1
ab
= − b(aa−+bb) , i.e. o1 = aab
+b
+b.
b2
1 and since it is placed on the
The point M is placed on the unit circle Γ, therefore m = m
circle whose radius is O1B and is centered at O1 we get that
(o
| o1 − b | = | o1 − m | , i.e. o1m 2 − b1 + o1b m + o1 = 0 )
holds. The solutions of the last quadratic equation are m and b, and thereby the Viet
formulas it is true that
o1
b+m= + b , i.e. m = b 2aa++2bb .
o1b

201
Further, the affix of the midpoint of the line segment BC is b +2 c . Finally, to prove that the
line AM bisects the line segment BC it is sufficient to prove that
a − b +2 c
= a −m = − am ,
a − b +2 c a −m

The validity of the latter could be easily checked. Details are left as your exercise. ■

79. Let Γ be a given circle, and AB be its diameter. Let P be an arbitrary point on Γ
distinct of A and B. The projection of the point P to AB is a point Q. A circle centered at
P and radius PQ meets Γ at points C and D. The lines CD and PQ intersect at a point E.
Let F be the midpoint of AQ, and G be the foot of the perpendicular at F to CD. Prove
that the points A, G and P are collinear and furthermore
EP = EQ = EG .
Solution. Let Γ be the unit circle and let b = 1. Then a = −1 and thereby P ∈ Γ we
( )
get that p = 1p . Further, the affix of the point Q is q = 12 p + 1p , and the affix of F is

f =
1
2 ( p+ 1p )−1 = ( p−1)2 . (1)
2 4p
The point C is placed on the circle centered at P and radius PQ, thus | p − q | = | p − c | ,
The latter implies that
( p − q )( p − q ) = ( p − c)( p − c) . (2)
But, C ∈ Γ , and thus c = 1c and thereby
p − q = 12 p − 1p ( )
by substituting in (2) we obtain that
4 pc 2 − ( p 4 + 6 p 2 + 1)c + 4 p3 = 0 . (3)
The equation (3) is a quadratic equation with variable c and since the point D satisfies the
same conditions as the conditions applied when determined the affix of C, we get that d
is the second solution of (3). Now, by applying the Viet roles we get the following
p 4 + 6 p 2 +1
c+d = , cd = p 2 .
4 p3
The point G is placed on the chord CD, therefore C, D, G are collinear, thus we get that
c+d − g
g= , and thereby FG ⊥ CD we have that
cd
g− f
= − c −d = cd = p 2 .
g− f c −d
By solving the system consisting of the last two equations where f is the expression given
in (1), we get that
p3 +3 p 2 − p +1
g= 4p
.
To prove that the points A, G and P are collinear it is sufficient to prove that

202
g −a a− p
=− = p.
g −a a− p
The last can be easily checked if we consider that
p3 +3 p 2 +3 p +1 p3 +3 p 2 +3 p +1
g −a= 4p
and g − a = .
4 p2
The point E is placed on the chord CD, therefore C, D and E are collinear, i.e.
e= c + d −e
. So, PE ⊥ AB implies that
cd
e− p
= − a −b = −1, i.e. e = p + 1p − e .
e− p a −b
That is,
p + 1p − e = e = c +cd
d −e ,

therefore
3 p 2 +1
e = 4p .
Now,
3 p 2 +1 1− p 2 p 2 −1 p − p3
e − p = 4 p − p = 4 p , e − q = 4 p and e − g = 4 p
Further, since | p | = 1, we get that
| e − p | = | e − q | = | e − g | , i.e. EP = EQ = EG . ■

80. Let H be the orthocenter of a  ABC. The tangents at A to the circle whose diameter
is BC touch the circle at P and Q. Prove that the points P, Q and H are collinear.
Solution. Let the circle over the diameter BC be a unit circle and let b = −1 . Then,
c = 1 and the origin is the midpoint of the line segment BC. The point P lies on the unit
circle, thus p = 1p and since PA ⊥ PO we get that
a− p p −o
=− = − p2 ,
a− p p −o
The latter implies
ap 2 − 2 p + a = 0 . (1)
The equation (1) is a quadratic equation with a variable p and thereby the point Q satisfies
the same conditions as the ones used when determining the point P, we get that q is the
second solution of (1). Now, by applying the Viet formulae we get the following
p + q = a2 , pq = a .
a
Let H ' be the intersection of the line through A perpendicular to the side BC and the line
PQ. the points P, Q and H ' are collinear, so
p + q −h '
h' = pq
, i.e. h ' = 2−aah ' .
But, AH ' ⊥ BC , therefore
a −h ' = − b −c = −1,
a −h ' b −c
The latter means that h ' = a + a − h ' . Thus,

203
a + a − h ' = h ' = 2−aah ' ,
So, we find
2
h ' = aa + a −2 .
a −a
We will prove that h = h ' , therefore the statement in the given problem shall be implied.
In order to do that, it is sufficient to prove that CH ' ⊥ AB , that is, to prove that
c −h ' = − a −b ,
c −h ' a −b
holds true (why?). We can be assured in validity of the last equality by direct checking if
we use that
2 ( a +1)( a + a − 2)
h '− c = h '− 1 = aa + a −2−a + a = and a − b = a +1.
a −a a −a
Details are left as an exercise. ■

81. Let P be a point on the extension of the diagonal AC of a rectangle ABCD

through the point C, so that ∠ BPD = ∠ CBP . Determine the ratio PB : PC .
Solution. Let the intersection of diagonals O of the rectangle be the origin and let the
line AB be parallel to the real axis. Then a + c = 0, b + d = 0, c = b and d = a . Further,
the points P, A, O are collinear, and therefore
p
= a , i.e. p = − ba p .
p a
Let ∠ DPB = ∠ PBC = ϕ . Then,
d−p b− p p −b
= e 2iϕ and = e 2iϕ c −b .
d−p b− p p −b c −b
If we multiply the last two equalities, and express the obtained equality in terms of a and
b, we get
p +b a ( p +b ) 2
= .
bp + a 2 (bp − a 2 )2
Further, if we express the above equality as a polynomial of p we get the following
(b − a )[bp3 + (a 2 + 3ab + b 2 ) p 2 − a (a 2 + 3ab + b 2 ) p − a3b] = 0 ,
i.e.
bp3 + (a 2 + 3ab + b 2 ) p 2 − a (a 2 + 3ab + b 2 ) p − a3b = 0 . (1)
But the point A satisfies ∠ DAB = ∠ ABC = π2 , thus one of the points which satisfies
the given condition in the given problem is the point A. So, a is one of the roots of the
polynomial (1). The latter implies that the polynomial can be divided by p − a and p is a
root of the such obtained quotient (why?), i.e.
bp 2 + (a 2 + 4ab + b 2 ) p + a 2b = 0 . (2)
holds true. Finally, by applying the condition (2) we get

204
PB
2
=
| p −b|2
=
( p −b )( p −b )
=
(
( p −b ) − ba p + a )
PC
2
| p −c|2 ( p −c )( p −c ) (
( p + a ) − ba p −b )
bp 2 −( a 2 +b 2 ) p + a 2b −2( a 2 + 4 ab +b 2 ) p
= = =2
bp 2 + 2 abp + a 2b − ( a 2 + 2 ab +b 2 ) p

The latter implies that PB : PC = 2 . ■

82. In a convex quadrilateral ABCD the diagonal BD is not a bisector neither of

∠ ABC nor of ∠CDA . A point P placed into the ABCD is such that ∠ PBC = ∠ DBA and
∠ PDC = ∠ BDA . Prove that the quadrilateral ABCD is cyclic if and only if AP = CP .
Solution. Let the quadrilateral ABCD be cyclic and let the circumcircle be the unit
circle. If ∠ PBC = ∠ ABD = ϕ and ∠ PDC = ∠ BDA = θ , then
p −b c − d p −d b−d
d −b = e 2iϕ a −b , c −b = e 2iϕ , = e 2iθ , = e 2iθ a −d
d −b a −b c −b p −b c −d p −d b−d a −d

and thereby a = 1a , b = b1 , c = 1c , d = d1 by using the first equality, we obtain that

e 2iϕ = da , and since the fourth one, e 2iθ = ba . By substituting at the second and the third
equality we get that
a p −b = −bc and ba = −cd ,
p −d
d p −b p −d
since which
p = acb ++bd
d
.
Further,
a − p = ab + adb +−dac −bd , a − p = bc +accd(b−+acd−) bd ,

c − p = bc +bdb +−dac −bd , c − p = ab +acad(b−+acd −) bd ,

thus,
2
AP = | a − p |2 = (a − p )(a − p ) = ab + adb +−dac −bd ⋅ bc +accd(b−+acd−) bd
2
= bc +bdb +−dac −bd ⋅ ab +acad(b−+acd −) bd = (c − p )(c − p ) = | c − p |2 = CP ,
i.e. AP = CP .
Let AP = CP , i.e.
| a − p | = | c − p | (*)
and let suppose that the circumcircle of the triangle ABC is the unit circle. This means
that a = 1a , b = b1 , c = 1c . The condition (*), after squaring and reducing, implies that
p p
ap+ a =cp+ c ,
that is
(
(a − c) p − ac = 0 ,
p
)
205
p
therefore p = ac . Let D ' denote the intersection of a side CD and a unit circle. Then
c − d = c − d ' = −cd ' ,
c −d c −d '
thus
d = c +cd
d '− d .
'
Thereby the condition of the problem, we have that
∠ CBP = ∠ DBA = ϕ and ∠ PDC = ∠ ADB = θ ,
thus
p −b p −d b−d
a −b = e 2iϕ d −b , = e 2iϕ c −b , c −d = e 2iθ , = e 2iθ a −d . (1)
a −b d −b p −b c −b c −d p −d b−d a −d
Thereby the first two equations in (1) we get that
p −b d −b
= a −b c −b = ab 2 c ,
p −b d −b a −b c −b
And by substitution for d and p , and after reducing we obtain following
p = c bdd '+ acd '−abd '− abc + abd −b 2 d ' . (2)
2 2 2
cd ' d −b d '+b d −b c
Now, the third and the fourth equality in (1) and c −d = −cd ' imply
c −d
a − d p −d b−d
−cd ' = . (3)
a −d p −d b−d
If in the latter we substitute the above expression for p and p , then after reducing
we obtain a polynomial of P(d), which is obviously at most quartic. By comparing
the coefficients of d 4 we get that the polynomial P(d) is exactly a cubic polynomial.
Clearly, two of its zeros are a and b. We shall prove that its third zero is d ' , and therefore
d = d ' , i.e. the quadrilateral ABCD is cyclic. Indeed, if d = d ' , then
a −d p −d p −d '
= − ad ' , b −d = −bd ' and = pd '−ac acd ' .
a −d b−d p −d
Thus the equality (3) is equivalent to the p = acd+'+bd
b
' , which is obviously satisfied, and is

obtained by letting d = d ' in (2). ■

83. Three triangles KPQ, QLP and PQM, so that ∠ QPM = ∠ PQL = α ,
∠ PQM = ∠ QPK = β and ∠ PQK = ∠ QPL = γ for α < β < γ and α + β + γ = 180° , are
constructed on a same side of a line segment PQ. Prove that the triangle KLM is similar
to, and moreover, is the same oriented with the triangles KPQ, QLP and PQM.
Solution. Let p = 0 and q = 1 . Since ∠ MPQ = α
m− p q− p
= e 2iα ,
m− p q− p
holds thus m = e 2iα . Further, ∠ PQM = β implies
m
m−q p −q
e 2iβ = ,
m−q p −q

206
thus e 2iβ m−1 = 1 . If we remember that e 2i (α +β+ γ ) = 1 , then by solving the system
m −1
e 2iα = m
 m
 2iβ
e m−1 = 1
 m −1
we get
2i ( α + γ )
m= e −1 .
e2i γ −1
2i (β + γ ) 2i ( α +β )
Analogously l = e 2 iβ
−1 and k = e −1 . According to the theorem 4.9 in order to
e −1 e2iα −1
prove that the triangle KLM is simmilar and same oriented as the triangle KPQ, it is
sufficient to prove that
k −l k− p
= p −q = −k , l −m
in which we can be convinced by immediate validation. Finally, since the triangles KPQ,
QLP and PQM are similar and same oriented, we get that each of the four triangles is
similar and same oriented as the other ones. ■

84. Prove that the area of the triangle whose vertices are feet of the perpendiculars
at any vertex of a cyclic pentagon to its sides does not depend on the choice of the vertex
of the pentagon.
Solution. Let the circumcircle of the pentagon ABCDE be the unit circle and let X, Y,
Z be the feet of the perpendiculars at the vertex A to the sides BC, CD, DE, respectively.
(
Then x = 12 a + b + c − bc
a )
, y = 12 a + c + d − cd
a (
and z = 12 a + d + e − ed
a
, thus ) ( )
a + b + c − ab a + b + c − bc 1
x x 1 a a

P XYZ = ± 4i y i a + c + d − cd
y 1 = ± 16 a
a + c + d − cd 1
a
z z 1
a + e + d − ed
a
a + e + d − ed 1
a

a + b + c − ab
a
a + b + c − bc 1
a
i ( a −c )( d −b ) ( a −c )( d −b )
= ± 16 a
0
a
(e −c )( a − d ) (e −c )( a − d )
a
0
a

a + b + c − ab
a
a + b + c − bc 1
a
i ( a −c )( d −b ) ( a −c )( d −b )
= ± 16 a bcd
0
(e −c )( a − d ) (e −c )( a − d )
a ced
0

207
a + b + c − ab
a
a + b + c − bc 1
a
i ( a −c )( d −b )(e −c )( a − d ) 1 1
=± 16 a bcd
0
1 1 0
a ced
i ( a −c )( d −b )(e −c )( a − d ) 1 1 )
=± 16
( aced − abcd
i ( a −c )(c −e)(e −b )(b − d )( a − d )
=± 16 abcde
.
Therefore, the area is the sixteenth of the product of the pentagon diagonals, so it does not
depend on the choice of the pentagon vertex. ■

85. The points A1 , B1 , C1 are positioned on the altitudes of the  ABC plotted at the
vertices A, B, C, respectively, and H is the orthocenter of  ABC. If
P ABC1 + P BCA1 + PCAB1 = P ABC , (1)
prove that the quadrilateral A1B1C1H is cyclic.
Solution. Let the circumcircle of the  ABC be the unit circle. Let A' be the foot of
the perpendicular at the vertex A to the side BC. Then
|b −c|⋅|a1 − a '| |b −c|⋅|a − a '|
P BCA1 = 12 BC ⋅ A1 A ' = 2
and P ABC = 12 BC ⋅ AA ' = 2
,
thus
P BCA1 |a − a '| |a − a '|−|a − a | |a − a | a −a
= |a1−a '| = |a −a '| 1 = 1 − |a −a1'| = 1 − a −a1' ,
P ABC
The latter means that the equality (1) can be transformed and rewritten as the following
a − a1 b −b c −c
a −a '
+ b −b1' + c −c1' = 2 . (2)
Further,
(
a ' = 12 a + b + c − bc
a
, )
thus
( a −b )( a −c )
a − a' = 2a
.
Analogously,
(b −c )(b − a ) (c − a )(c −b )
b − b' = 2b
, c − c' = 2c
.
If we substitute in (2), and after equivalent transformations, we get that the above
condition is equivalent to
aa1 (b − c) + bb1 (c − a ) + cc1 (a − b) = 0 . (3)
According to the Remark 25.3, in order to prove that the quadrilateral A1B1C1H is
cyclic, it is sufficient to prove that
a1 −c1 b1 − h b1 −c1 a1 − h
= . (4)
a1 −c1 b1 − h b1 −c1 a1 − h
holds.

208
The point H is the orthocenter of the  ABC, thus h = a + b + c and since A1H ⊥ BC
a1 − h b1 − h
we get that = − b −c = bc and similary = ac . Further, thereby A1 A ⊥ BC , we
a1 − h b −c b1 − h
a1 − a bc + aa1 − a 2 ac +bb1 −b 2 ab + cc1 −c 2
get = bc , thus a1 = bc
and similarly b1 = ac
and c1 = ab
.
a1 − a
Finally, if we apply the obtained equalities and the condition (4), we immideatly check
the validity of (4), which means that the quadrilateral A1B1C1H is cyclic. The details are
left as an exercise for the reader. ■

86. The feet of the altitudes at the vertices A, B and C of a  ABC are D, E and F,
respectively. The line through D is parallel to EF and meets the lines А�
C and AB at Q and
R, respectively. The line EF meets the line BC at P. Prove that the circumcircle of  PQR
consists of the midpoint of the side BC.
Solution. Let the circumcircle of the  ABC be the unit circle. So,
(
d = 12 a + b + c − bc
a )
, e = 12 a + b + c − ac
b (, f = 12 a + b + c − bc
a )
, a1 = b +2 c , ( )
a +c −q
where A1 is the midpoint of BC. Since Q is placed on AC we get that q = ac
. But,
QD || EF , thus
q −d e− f
= = −a 2 .
q −d e− f
By solving the system consisting of the last two equations, we get that
3 2
abc −b 2c .
q = a + a b2+ab
Similarly,
3 2
abc −bc 2 .
r = a + a c2+ac
Moreover, P ∈ BC , thus
b+c− p
p= bc
And since P ∈ EF , we get that
p −e e− f
= = −a 2 .
p −e e− f
By solving the system of the last two equations, we obtain
a (b −c ) 2
p = b +2 c + .
2( a 2 −bc )
We have to prove that the points P, Q, R and A1 are concyclic, which according to
Remark 25.3 means that we have to prove the equality
p − a1 q − r q − a1 p − r
= ,
p − a1 q − r q − a1 p − r
in order to do this it is sufficient to apply that
a (c −b )( a 2 +bc ) ( a 2 −c 2 )(b 2c + abc − a3 − a 2c )
q−r = 2 abc
, p−r = ,
2 ac ( a 2 −bc )

209
2
3 2
b 2c − ab 2 and p − a = a (b −c ) .
q − a1 = a + a b2−ab 1 2
2( a −bc )
The details are left to the reader as an exercise. ■

87. Given two circles Γ1 and Γ 2 on the plane. Let А be their common point. On the
circles Γ1 and Γ 2 , with constant velocities the points M1 and M 2 move, respectively.
They pass through A at the same moment of time. Prove that it exists a fixed point P
which at every moment in time is on a same distance of the points M1 and M 2 .
Solution. Let B and C be the centers of the circles Γ1 and Γ 2 and let BC be the real
axis. If the points M1 and M 2 move in the same direction, then
m1 − b = (a − b)eiϕ and m2 − c = (a − c)eiϕ .
The existance of a point P with the desirable property is consecutively equivalent to
the following conditions
| p − m1 | = | p − m2 | , ( p − m1 )( p − m1 ) = ( p − m2 )( p − m2 ) ,
m1 m1 − m2 m2 − p ( m1 − m2 )
p= m1 − m2
. (1)

Let eiϕ = z . If we apply that b = b , c = c and z = 1z , we get that the condition (1) is
equivalent to
(b + c − a − p ) z 2 − [2(b + c) − a − a − p − p ]z + b + c − a − p = 0 , (2)
The latter means that the right side of the polynomial, in (2), with a variable z must be
identically equal to null. Therefore all of its coefficients must be nulls. Since the free
term, we find that p = b + c − a and clearly the coefficients of the linear and the quadratic
terms are null.(Check it!)
The completely identically procedure is applied in case when the points M1 and
M 2 move in opposite directions. The details are left for reader as an exercise. ■

88. Given a square ABCD and a circle Γ with diameter AB. Let P be any point on the
side CD, M and N be the points where the line segments AP and BP meet Γ which differs
from A and B, and Q be the intersection of the lines DM and CN. Prove that Q ∈ Γ and
further that AQ : BQ = DP : CP .
Solution. Let Γ be the unit circle and let a = −1. Then b = 1, c = 1 + 2i and
d = −1 + 2i . Further, the points A, P, M are collinear, therefore
a− p
= a −m = − am = m ,
a− p a −m
the latter implies that
p +1− m
p= m
.
But the points C, D, P are collinear, therefore
c− p
= c −d = 1 ,
c− p c −d

210
therefore p = p − 4i . That is,
p +1− m 4im − 1.
= p = p − 4i , i.е. p = m
m −1
Similarly, the points B, N, P are collinear, and therefore
c− p
= c −d = 1 ,
c− p c −d
that is
b− p m (1− 2i ) −1
n=− = 1+ 2i −m .
b− p
Let Q ' = Γ ∩ DM . So,
q '− m
q ' q ' = 1 and d −m = = −q ' m ,
d −m q '− m
therefore
q ' = − mm(1++12−i2)+i 1 .
Further,
q '− c q '− c m (1− 2i ) −1 q '− n
= mm(1++12−i2)+i 1 ⋅ 1+ 2i −m = − nq ' =
= q' ,
q '−c 1− q ' c q '− n
The latter means that the points Q ', C , N are collinear, which implies that
Q ' = CN ∩ DM = Q .
The equality AQ : BQ = DP : CP is equivalent to the equality
| q − a |⋅| p − c | = | d − p |⋅| b − q | ,
Its validity can be proven by immediate checking and by applying the following
4im − 1 − 1 − 2i = 2 m (i −1) +1+i ,
| q − a | = mm(1++12−i2)+i 1 + 1 = 2 m(1m++21i )+1 , | p − c | = m −1 m −1

| d − p | = −1 + 2i − m4im + 1 = 2 m +1 , | b − q | = 1 + m +1− 2i = 2 m (1+i ) +1−i

−1 m −1 m (1+ 2i ) +1 m (1+ 2i ) +1
and apply that
i[m(1 + i ) + 1 − i ] = m(i − 1) + i + 1 . ■

89. Given a  ABC and a circle such that it passes through B and C and remeets the
sides AB and AC at the points C ' and B ' respectively. Prove that the lines BB ', CC ' and
HH ' are concurrent (H and H ' are the orthcenters of the triangles ABC and A ' B ' C ' ,
respectively).
Solution. Let the circumcircle of the quadrilateral BCB ' C ' be the unit circle. The
intersection of lines BB ' and CC ' is a point X with affix
bb '(c + c ') −cc '(b +b ')
x= bb '−cc '
.
Further, since BH ⊥ CB ' and CH ⊥ BC ' we get
b−h = − b '−c = −bc ' and c −h = − b −c ' = bc ' ,
b−h b '−c c −h b −c '
therefore
b 2 + cb ' and h = ch −c 2 +bc ' ,
h = bh −bb 'c bc ' c

211
So,
2 2
bh −b 2 + cb ' c 2 +bc ' , i.e. h = b ' c '(b −c ) +b c '−b ' c .
= h = ch −bc
bb ' c 'c bc '−cb '
Analogously,
bc (b '−c ') +b '2 c −bc '2
h' = b ' c −c ' b
.
Finally, in the order to prove the statement, it is sufficient to prove that the points
H , H ', X are collinear, i.e. to prove that
h−h ' = h− x ,
h−h ' h− x
holds true. We can be convinced in the validity by immediate check if we use that
(b +b '−c −c ')(bc '+ cb ') b ' c '(b 2 −c 2 )(b '+b −c '−c )
h − h' = bc '−cb '
and h − x = (bc '−cb ')(bb '−cc ')
.
The details are left to the reader as an exercise. ■

90. Let ABCDEF be a convex hexagon such that

∠ B + ∠ D + ∠ F = 360° and AB ⋅ CD ⋅ EF = BC ⋅ DE ⋅ FA .
Then
BC ⋅ AE ⋅ FD = CA ⋅ EF ⋅ DB .
Prove it!
Solution. Let ∠ A = α, ∠ B = β, ∠ C = γ, ∠ D = δ, ∠ E = ε, ∠ F = ϕ . So,
c −b = eiβ a −b , e − d = eiδ c − d , a − f = eiϕ e − f .
|c −b| |a −b| |e − d | |c − d | |a − f | |e − f |
If we multiply the last three equalities and consider that
β + δ + ϕ = 360° and | a − b | ⋅ | c − d | ⋅ | e − f | = | b − c | ⋅ | d − e | ⋅ | f − a |
we get
(c − b)(e − d )(a − f ) = (a − b)(c − d )(e − f ) .
So, it is easy to conclude that
(b − c)(a − e)( f − d ) = (c − a )(e − f )(d − b) ,
If we take modulus in the last equality we obtain the required equality. ■

91. Given a triangle  ABC and points X and Y on the sides BC and CA, respectively.
 
Let R = AX ∩ BY and AY
 = p, AR
 = q , for
YP  RX
0 < p < q . Determine the ratio BX .
XC
Solution. Let consider the  AXC. The points
B, R and Y are Menelaus’ points of the sides CX, AX
and AC, respectively. The are collinear thereby the
condition R = AX ∩ BY (figure 26). According to
the Menelaus’ theorem it is true that
  
AR  ⋅ CY
 ⋅ XB  = −1.
Figure 26 RX BC YA

212
So,
  
BC q
 ⋅ CY
= AR  = −

XB
RX YA p
    
and thereby BC = BX + XC and XB = − BX by substitution in the last equality we get
 
BX+ q
XC = p
BX
that is

p
BX
 = q− p . ■
XC

92. Given a right angled triangle  ABC whose right angle is at B and sides AB = 4,
BC = 3 . A point E is the midpoint of the side AB, and the point D is on placed the side
AC and moreover DA = 1 . Let F = DE ∩ BC . Determine the length of the line segment
BF.
Solution. Let’s consider the triangle  ABC
(figure 27). The points D, E and F are Menelaus’
points of the sides CA, AB and BC, respectively,
so they are collinear. The Menelaus’ theorem
implies
AE ⋅ FB ⋅ CD =1.
(1)
EB FC DA
Since the condition of the given problem,
FC = FB + CB = FB + 3 ,
DA = 1 and AE = EB = 2 .
Moreover the Pithagora’s theorem implies Figure 27
2 2
CA = BC + AB = 5 . Due to this, CD = CA − DA = 4 and if we substitute in (2) and
after reducing, we get that FB = 1. ■

93. Let A0 A1 A2 A3 A4 A5 A6 be a regular heptagon. Prove that

1 = 1 + 1 . (1)
A0 A1 A0 A2 A0 A3

Solution. Without loss of generality, we consider the case the regular heptagon
is inscribed into the unit circle and (1,0) is the affix of the vertex A0 . So, the affixes
of the vertices Ak , k = 0,1, 2, 3, 4, 5, 6 are ak = wk , k = 0,1, 2, 3, 4, 5, 6 , respectively, for
i 2π
w = e 7 . Further, since the properties of the regular heptagon, we get that if the point A1
is rotated at A0 around 27π and the point A2 is rotated at A0 around 14 2 π we get points

i 2π
which are collinear with the points A0 and A3 . Let ε = e 14 , w = ε 2 , thus
a '1 = 1 + (a1 − 1) w and a '2 = 1 + (a1 − 1)ε .
To prove the equality (1) it is sufficient to prove that

213
1 = 1 + 1 ,
a '1 −1 a '2 −1 a3 −1
(why?). The last equality is equivalent to the equality
1 = 14 + 61 ,
2 2
ε (ε −1) ε (ε −1) ε −1
Which after reducing, can be rewritten as below
ε 6 + ε 4 + ε 2 + 1 = ε5 + ε3 + ε .
But,
ε5 = −ε12 , ε3 = −ε10 , ε = −ε8 .
Therefore the last equality is equivalent to the equality
ε12 + ε10 + ε8 + ε6 + ε 4 + ε 2 + 1 = 0 ,
i.e. to the equality
w6 + w5 + w4 + w3 + w2 + w + 1 = 0 ,
which is obviously true thereby w7 = 1 . ■

94. Let A0 A1... A13 A14 be a regular 15-gon. Prove that

1 = 1 + 1 + 1 , (1)
A0 A1 A0 A2 A0 A4 A0 A7
holds.
Solution. Without loss of generality we consider ak = wk , k = 0,1, 2,...,14 , for
i 2π
w = e 15 . Further, by rotation of the points A1 , A2 , A4 at A0 around 615π , 515π , 15 3π ,

respectively, we get points with affixes a '1 , a '2 , a '4 which are collinear with the points
A0 and A7 . Therefore, to prove the equality (1) it is sufficient to prove that
1 = 1 + 1 + 1 . (2)
a '1 −1 a '2 −1 a '4 −1 a7 −1
holds. We set that
iπ
ε = e 15 , w = ε 2 , ε30 = 1
and obtain that
a '1 = 1 + (a1 − 1)ε6 , a '2 = 1 + (a2 − 1)ε5 and a '4 = 1 + (a4 − 1)ε3 ,
The latter means that the equality (2) is equivalent to the following equality
1 = 1 + 1 ε16 .
− 16
ε (ε 2 −1)
6
ε (ε 4 −1)
5
ε (ε8 −1)
3
ε −1
2
If the last is multiplied by ε − 1 ≠ 0 , and after reducing we obtain the following
equality
ε14 + ε12 + ε10 + ε8 + ε6 + ε 4 + ε 2 + 1 = ε(ε12 + ε8 + ε 4 + 1) + ε3 (ε8 + 1) − ε 22 . (3)
But, ε15 = eiπ = −1 = −ε30 , therefore ε15−k = −ε30−k , which implies
ε13 = −ε 28 , ε9 = −ε 24 , ε5 = −ε 20 , ε = −ε16 , ε11 = −ε 26 , ε3 = −ε18 ,
so, the equality (3) is equivalent to the equality

214
ε 28 + ε 26 + ε 24 + ε 22 + ε 20 + ε18 + ε16 + ε14 + ε12 + ε10 + ε8 + ε6 + ε 4 + ε 2 + 1 = 0 ,
30
which is obviously true, thereby the left side of the last equality is equal to ε 2 −1 = 0 . ■
ε −1

95. Given a cyclic quadrilateral ABCD. The points A ', B ', C ', D ' are the centroids of
the triangles BCD, ACD, BAD, ABC, respectively. Prove that the quadrilateral A ' B ' C ' D '
is also a cyclic quadrilateral.
Solution. The quadrilateral ABCD is cyclic, so ac −−bb ⋅ ac −−dd ∈ R *. Further, a ' = b +c3+ d , b ' = a +c3+
= b +c3+ d , b ' = a +c3+ d , c ' = a +b3+ d , d ' = a +3b +c , so
b −c d −a
c ' −b ' ⋅ a ' − d ' = 3
⋅ 3
= ac −−bb ⋅ ac −−dd ∈ R*,
a ' −b ' c ' − d ' b−a d −c
3 3
the latter implies that the quadrilateral A ' B ' C ' D ' is cyclic. ■

96. Given a triangle ABC and points P, N, M positioned on the sides AB, BC, CA,
respectively. Prove that the circumcircles of the triangles APN, BMP, CNM meet at a
unique point.
Solution. Let Q be the other point of intersection
of the circumcircles of the triangles APN and BMP
(see the figure 28). The points A, P, Q, N are on a same
q− p q −m b− p
circle, thus a − p ⋅ aq −−nn ∈ R*, simillary b −m ⋅ q − p ∈ R * .
q −m b− p q − p q −m b− p
So, q −n ⋅ aa−−np ⋅ b −m = a − p ⋅ aq −−nn ⋅ b −m ⋅ q − p ∈ R * , which
q −m p −b
implies q −n ⋅ mn −−cc ⋅ mn −−cc ⋅ np−−aa ⋅ m−b ∈ R * . But, Figure 28

( p −b
) p −b
arg mn −−cc ⋅ np−−aa ⋅ m−b = arg mn −−cc + arg np−−aa + arg m−b

= ∠ MCN + ∠ NAP + ∠ PBM

= ∠ ABC + ∠ BCA + ∠ CAB = π,
p −b q −m
thus mn −−cc ⋅ np−−aa ⋅ m−b ∈ R*, which means that q −n ⋅ mn −−cc ∈ R * , i.e. the points Q, M, N, C
lie on a same circle. Finally, the circumcircles of the APN, BMP, CNM meet at Q. ■

97. Four distinct lines intersect each other, so that they form four triangles. Prove
that the four circumcircles of these triangles have a common point.
Solution. Since the condition of the given problem, three of the given lines are not
concurrent. Let A, B, C, D, E, F, be the point of intersection of the lines, see the figure.
Let the circumcircles of the triangles ABC and EFC intersect at the point P. We will prove
that the points E, P, A, D are concyclic. It is true that,
p − a b −c c − f e− p
⋅ ,
b − a p −c e − f
⋅ c − p ∈ R * . (1)

215
By dividing these two numbers, we get
p − a b −c e − f
⋅ ⋅
b−a e− p c − f
∈ R * . Further, the points E, F,
D are collinear, and so are the points B, A, D
e− f
therefore e−d = t ∈ R * and db −−aa = t ' ∈ R * . If the
last two equalities we substitute in (1) we obtain
a − p e−d
⋅
e− p a −d
⋅ bf −−cc ⋅ tt' ∈ R * . Since the points B, C, F
a− p
are collinear, bf −−cc ∈ R * holds, thus e− p ⋅ ae−−dd ∈ R * ,
which means that the points A, D, E, P are concyclic.
Analogously, it can be proven that the points B, D,
Figure 29 F, P are concyclic. So, the four circles consist of the
point P. ■

98. In a convex quadrilateral ABCD the sides AB and CD are congruent.

а) The lines AB and CD with the line which connect the midpoints of the sides AD
and BC form congruent angles. Prove it!
b) The lines AB and CD with the line which connect the midpoints of the diagonals
AC and BD form congruent angles. Prove it!
Solution. а) Let 0, r, c, d where r ∈ R + , c, d ∈C , be the affixes of A, B, C, D. The
points N and M are the midpoints of the line segments AD and BC, respectively, thus
n = d2 and m = r +2 c . So,
2 r  (c −d )
(b − a )  (m − n) = r  r +c2−d = r2 + 2
therefore | c − d | = r implies
r  (c −d ) |c − d |2
(c − d )  (m − n) = (c − d )  r +c2−d = 2
+ 2
r  (c −d ) 2
= 2
+ r2 = (b − a )  (m − n)
and since | c − d | = | b − a | = r and the previously stated implies
∠( AB, NM ) = ∠( NM , DC ) .
b) Let L and K be the midpoints of AC and BD, respectively. Then l = 2c and
k = r +2d . Thus,
(k − l )  (m − n) = r + d2 −c  r +c2−d
2 ( c − d )  ( d −c )
= r4 + 4
2 |c − d |2
= r4 − 4 = 0,
so, KL ⊥ MN . The latter and the statement a)
imply
Figure 30 ∠( AB, KL) = ∠( KL, DC ) . ■

216
99. In an acute triangle ABC, the orthocentar H satisfies the following HC = AB .
Determine the angle at C.
Solution. Let the triangle be inscribed into the unit circle. Then, h = a + b + c
therefore HC = | a + b | and AB = | a − b | . Hence,
| a + b |2 = | a − b |2 , i.e. (a + b)  (a + b) = (a − b)  (a − b) ,
therefore, a  b = 0 , that is that OA ⊥ OB , i.e. ∠ AOB = π2 . But, the triangle ABC is an
acute triangle, and thereby the measure of inscribed angle is half of the measure of its
corresponding central angle, we get that ∠ ACB = π4 . ■

100. In a convex quadrilateral ABCD the points P and Q are the midpoints of the
diagonals AC and BD, respectively. Prove that
2 2 2 2 2 2 2
AB + BC + CD + DA = AC + BD + 4 PQ .
Solution. The points P and Q are midpoints of the diagonals AC and BD, thus
p= a +c
and q = b +2d . Further,
2
2 2
AB = | b − a |2 = (b − a )  (b − a ) = | a |2 − 2a  b + | b |2 , BC = | b |2 − 2b  c + | c |2 ,
2 2
CD = | c |2 − 2c  d + | d |2 , DA = | d |2 − 2d  a + | a |2 ,
2 2
AC = | a |2 − 2a  c + | c |2 , BD = | b |2 − 2b  d + | d |2 ,
2
4 PQ = 4(q − p )  (q − p ) = (b + d − a − c)  (b + d − a − c)
= | a |2 + | b |2 + | c |2 + | d |2 + 2a  c + 2b  d − 2a  b − 2b  c − 2c  d − 2a  d ,
thus
2 2 2 2
AB + BC + CD + DA = 2 | a |2 + 2 | b |2 + 2 | c |2 + 2 | d |2 −
− 2a  b − 2b  c − 2c  d − 2a  d
2 2 2
= AC + BD + 4 PQ ,
which was supposed to be proven. ■

101. Let H be the orthocenter of an acute triangle ABC. A circle through H and
centered at the midpoint of the line segment BC, meets the line BC at points A1 and
A2 . Analogously, a circle through H and centetred at the midpoint of the line segment
CA, meets the line CA at points B1 and B2 , a circle through H and centetred at the mid-
point of the line segment AB meets the line AB at points C1 and C2 . Prove that the points
A1 , A2 , B1 , B2 , C1 , C2 lie on a same circle.
Solution. Let the triangle be inscribed into the unit circle, A0 be the midpoint of
the line segment BC and let a, b, c, a0 , a1 , a2 , h be the affixes of points A, B, C, A0 ,

217
A1 , A2 , H, respectively. Then, h = a + b + c , aa = bb = cc = 1 and a0 = b +2 c . Since the
triangles OA1 A0 and A2OA0 are right angled triangles,
OA12 = OA 22 = OA 02 + A0 A12 = OA 02 + A0 H 2

( )(
= b +2 c ⋅ b +2 c + a + b + c − b +2 c a + b + c − b +2 c )
= aa + bb
2
+ cc
2
+ ab + ab +bc2+cb +ca +ca

= 2 + ab + ab +bc2+cb +ca +ca .

The last expression is symmetric by a, b, c, and by cyclic substitution of the variables we
obtain the following OA12 = OA 22 = OB 12 = OB 22 = OC 12 = OC 22 , which means that the
points A1 , A2 , B1 , B2 , C1 , C2 are on a same circle. ■

102. Let I be the incenter, and Γ be the circumcircle of a triangle  ABC. Let the line
AI meets Γ at points A and D. Let E be a point on the arc BDC  , and F be a point on the
line segment BC such that ∠ BAF = ∠ CAE < 12 ∠ BAC holds true. Let G be the midpoint
of the line segment IF. Prove that the intersection of DG and EI belongs to Γ.
Solution. Let  ABC be inscribed into the unit circle. Since Theorem 13.3 there
exist complex numbers a, b, c such that the points A, B, C have affixes a 2 , b 2 , c 2 ,
respectively, and the midpoints of the arcs BC  , CA
,  AB which do not consist of the
points A, B, C have affixes −bc, − ca, − ab , respectively and the incenter I has affix
s = − ab − bc − ca . Therefore I is the orthocenter of the triangle whose vertices are the
midpoints of the arcs BC , CA
,  AB which do not consist of the points A, B, C.
Without loss of generality we get that the points B and C are symmetric with respect
to the real axis. Let F ' be the point of intersection of AF and Γ, which differs from
A and let the points D, E , F , F ', I have affixes d , e, f , f ', s , respectively. Hence,
2 2
| a | = | b | = | c | = 1 , c = b, d = −1, f ' = b ec = 1e and s = −1 − a (b + b) . Thereby F is a
point of intersection of the lines AF and BC we get that
f −b 2 2 2
f −a 2 a 2 − 1e 2
2
= b2 −b = −1 and 2
= 2
= − ae ,
f −b b −b 2 f −a b −1
e
2 2
which imply the following f + f = b 2 + b and f + ae f = a 2 + 1e , thus
2
f (a 2 − e) = a 2 (b 2 + b ) − a 2 e − 1 .
Let y be the affix of the point of intersection of IE and Γ (which differs from E), and
x be the affix of the point of intersection of DG and Γ (which differs from D). Then
x = 1x , y = 1y , thus

218
y −e
− ye = = s −e , i.e. y = s −e .
y −e s −e 1−es
Further,
2
− a (b + b) = 1 + s , a 2 s = − a 2 − a 2 b +a b = 1 − a 2 + s , a 2 − e = 12 − 1e = − a 2−e ,
2 a a e
( f + s + 2)(a 2 − e) = a 2 (b 2 + b ) − a 2 e − 1 + 2a 2 + sa 2 − es − 2e
= a 2 (b + b) 2 − 1 + a 2 ( s − e) − 2e − es
= (1 + s ) 2 − 1 − 2e − es + a 2 ( s − e) = ( s − e)(2 + s + a 2 ),
2 2
( f + s + 2)(a 2 − e) = ( s − e)(2 + s + a 2 ) = − 1−ees ⋅ 2a + a2 s +1
a
2 (1−es )( 2+ s + a 2 )
2
= − 1−ees ⋅ 2a +1−2a + s +1 = − 2
,
a a e
imply the following
f +s
g +1 +1 f +s+2 ( f + s + 2)( a 2 −e) ( f + s + 2)( a 2 −e)
x = 1+ x = = 2
= = = − 12 ⋅
1+ x g +1 f +g
+1 f +s+2 2
( f + s + 2)( a −e) a e ( f + s + 2)( a 2 −e)
2
2
( s −e)( 2+ s + a )
= − 12 ⋅ 2
= s −e = y ,
a e − (1− es )( 2 + s + a ) 1−es
2
a e
which actually was supposed to be proven. ■

103. Let P be a point in the inner part of a triangle ABC and let the lines AP, BP, CP
remeet the circumcircle Γ of the triangle ABC at points K, L, M, respectively. The tangent
of the circle Γ at C meets the line AB at S. Let SC = SP . Prove that MK = ML .
Solution. Without loss of generality we consider that the triangle ABC is inscribed
into the unit circle and let 1 be the affix of ABC. If to the points correspond the affixes
denoted by the appropriate lower case letters then | a | = | b | = | k | = | l | = | m | = 1 = c ,
a− p p −a p −b p −1
= a −k = − ak , thus k = and symmetrically l = , m= . The point S is
a− p a −k 1− a p 1−b p 1− p
the point of intersection of the line AB and the tangent to Γ at C, thus s + s = 2 and
s + abs = a + b , therefore we obtain that s = a +1b−−ab2ab . Let T be the midpoint of PC, i.e.
p +1
t = 2 . Since the condition of the problem SC = SP , that is T is the foot of the perpendicular

at S to PC, i.e. to MC, which implies that t −1 = m−1 = − m and t − s = − m−1 = m , i.e.
t −1 m −1 t −s m −1
p + 1 = 2t = m + 1 + s − ms , i.e. p = m + s − ms = m + s − m(2 − s ) = s − m + ms .
Thereby
p −1 p−s
= m = s −1
1− p
it is true that

219
2 p − sp −1
p − s = sp − p − s + 1 + ( p − s ) p , i.e. p = p−s
.
Thus,
p−s p − a +1b−−ab2 ab (1− ab ) p − a −b + 2 ab (1− ab ) p − a −b + 2 ab
m = s −1 = a + b − 2 ab −1
= a +b −2ab −1+ ab = − (1− a )(1−b )
,
1− ab

k=
p −a ( p − a )( p − s )
= p −2ap + a +( ap −1) s =
(
( p − a ) p − a + b − 2 ab
1− ab )
2 p − sp −1 p − 2 ap + a + ( ap −1)⋅ a +1b−−ab2 ab
1− a⋅ p−s
( p − a )[(1− ab ) p − a −b + 2 ab ]
=
p − 2 ap + a − abp + 2 a 2bp − a 2b + a 2 p + abp − 2 a 2bp − a −b + 2 ab
( p − a )[(1− ab ) p − a −b + 2 ab ] ( p − a )[(1− ab ) p − a −b + 2 ab ]
= =
p − 2 ap + a 2 p − a 2b −b + 2 ab p (1− a )2 −b (1− a )2
( p − a )[(1− ab ) p − a −b + 2 ab ]
=
( p −b )(1− a )2
and symmetrically,
( p −b )[(1− ab ) p − a −b + 2 ab ]
l= ,
( p − a )(1−b )2
So, m 2 = kl , which implies the statement of the problem.■

104. Let ABC be an acute scalene triangle so that AC > BC , O be the circumcenter
H be the orthocenter and F be the foot of the altitude at the vertex C. Let P be a point on
the line AB, which differs from A, so that AF = PF , and M be the midpoint of the line
segment AC. Let X be the point of intersection of the lines PH and BC, Y be the point of
intersection of the lines OM and FX, and Z be the point of intersection of the lines OF and
AC. Prove that the points F, M, Y and Z lie on a same circle.
Solution. Without loss of generality we consider the case where the triangle ABC is
inscribed into the unit circle. Let the affixes of the points A, B, C, H, F, P, X be a, b, c, h,
f, p, x, respectively. So, h = a + b + c and | a | = | b | = | c | = 1 , holds true.
Thereby F is placed on AB, and CF is perpendicular to AB we get that
f −a f −c
= b −a = − ab = − , i.e. f + ab f = a + b
and f − ab f = c − abc , thus
f −a b−a f −c
 
f = a + b + c − abc . p = 2 f − a = b + c − abc , thereby AF = FP . The point X is placed on
2
BC, hence x −b = b −c = −bc , i.e. x = b +bc
c − x . But, X is placed on PH, hence
x −b b −c
p− x p −h ac (b + c ) 2
= = b +c −abc −a −b −c = − = acb ,
p− x p −h b −c − abc − a −b −c − ab (c +b )
therefore

( )
2 2 2
2
c − x = − ac 2 + a 2 x , i.e. x = c p + ac = 2 fc .
p − x = acb b + c − abc − b +bc 2 2 2 2 2
c a +c a +c

220
f −0 f −x
It is sufficient to prove that OF ⊥ FX , which is equivalent to =− , i.e.
f −0 f −x
x f + x f = 2 | f |2 , which is obviously true, thereby

( ) = 2| f | . ■
2
2 fc 2 2fc c2 2
x f + xf = 2 2
f + 2 2
f = 2 | f |2 + 2a 2
a +c a +c a +c2
2
a +c2

105. Let  ABC not be an isosceles triangle and let AD, BF and CF be the bisectors
of its angles ( D ∈ BC , E ∈ AC , F ∈ AB ). Let K a , Kb , K c be points of incircle
of the triangle  ABC such that DK a , EKb , FK c are tangents of the incircle and
K a ∉ BC , Kb ∉ AC , K c ∉ AB . Let A1 , B1 , C1 be the midpoints of sides BC, CA, AB,
respectively. Prove that lines A1K a , B1Kb , C1K c concure at the incircle of the triangle
 ABC.
Solution. Without loss of generality we consider the incircle of the triangle  ABC as
the unit circle and let the circle tangents the sides BC, CA, AB at A ', B ', C ' , respectively.
Let S be the incenter whose affix is 0. If the affixes of the points are denoted by correspond-
ing lowercase letters, we get that | a ' | = | b ' | = | c ' | = 1 and a = b2b'+' cc '' , b = a2a'+' cc '' , c = 2aa'+' bb '' ,
2 2
thus a1 = b +2 c = a ' b( a'+'+ab' ')(ca'+'+2ca')' b ' c ' . Since DK a is a tangent of an incircle we get that

∠ ASK a = ∠ A ' SA and thereby | ka | = 1 , it implies that ka = a ' , so, ka = a1' ⋅ a = ba' c' '
a a a ( )
. The point of intersection X, of the incircle and the line A1K a satisfies the following
x −k
1 a
x −k
( )
| x | = 1 and a −ka = a −ka , that is (a1 − ka )( x − ka ) = 1x − k1 (a1 − ka ) , and thereby
1 a a
( )
x ≠ ka , it is true that a1 − ka = − xk1 (a1 − ka ) . Since
a
2 2 ( a '2 −b ' c ')(aa ' b '+ a ' c '+b ' c ')
a1 − ka = a ' b( a'+'+ab' ')(ca'+'+2ca')' b ' c ' − ba' c' ' = a '( a '+b ')( a '+ c ')
and
b ' c ' − a '2 ⋅ a ' + b ' + c '
a '2 b ' c ' a ' b ' c ' (b ' c '− a '2 )( a '+b '+ c ')
a1 − ka = ( a ' + b ')( a ' + c ')
= b ' c '( a '+b ')( a '+c ') ,
a '3 b ' c '
and a '2 ≠ b ' c ' , since  ABC is not an isosceles triangle,
a1 − ka
x = − k1 ⋅ = a ' ba'+'+bb' c'+'+cc' ' a ' .
a a1 − ka
Thereby the obtained expression is symmetric at a ', b ', c ' , the lines B1Kb and C1K c
meet the incircle at X. ■

221
4. EXERCISES (CHAPTER 2 AND 3)

1. Dermine the relationship between the points A and B with affixes a and b,
respectively, if given that
а) Re ab = 0 , b) Im ab = 0 ,
c) Re ab = 0 , d) Im ab = 0 .

2. Let A' be the projection of a point

A on a real axis. Determine the
point A'' such that it is symmetric
to A' with respect to the line OA
(figure 31).
Figure 31

3. Let OA = 1 be the diameter of a semicircle (fig-

ure 32). At B and C such that OB = 14 and OC = 34
are drawn perpendiculars to x-axis and E and D are
the points of intersection of these perpendiculars
and the semicircle. Find the complex number such
that it is the affix of M, the point of intersection of
the lines OE and BD. Figure 32

π
4. Let be given a point C with affix c = 2e 6 . Determine the affixes a and b of the
points A and B such that they are symmetric with respect to the line OC, the distance
between each of them and the point C is 1 unit and satisfy the following:
а) | a − b | = 2 , b) | a − b | = 2 .

5. Given the points A, B, C and Z with affixes a, b = a + eiα , c = a + eiβ and z,

respectively. Determine the distance between the symmetric points Z ' and Z '' to
the point Z with respect to the lines AB and AC.

6. We shall say that α is a viewing angle for a line segment AB of a point M a line
segment AB is viewed at an angle α of a point M if ∠ AMB = α . Let be given
points A and B with affixes a and b, respectively. Let W be the point on the bisector
of AB and furthermore the viewing angle for the line segment AB be α. Prove that
iα
w = aeiα −b is the affix of W.
e −1

222
7. The square of the diagonal is equal to the sum of the square of the leg and the
product of the length of the bases in an isosceles trapezoid. Prove it!

8. Let a quadrilateral ABCD be parallelogram and let N be the point of intersection

of the semiline AD and the circumcircle of the triangle ABC. Prove that
2 2
AD ⋅ AM = AC − AB .

9. A circle (K) is a circumcircle of a regular pentagon ABCDE. Let M be a point on the

arc 
AE . Prove that
MA + MC + ME = MB + MD .

10. Construct a trapezoid if given all its sides.

11. Given circles K '(o ', R ') and K ''(o '', R '') and a line segment AB. Construct a line
segment CD parallel and congruent to AB, such that C ∈ ( K ') and D ∈ ( K '') .

12. Given a line (p), a circle (K) and a line segment AB. Construct a line segment CD
parallel and congruent to AB, such that C ∈ ( K ) and D ∈ ( p ) .

13. Given lines (p) and (q) and a line segment AB. Construct a line segment CD parallel
and congruent to AB, such that C ∈ ( p ) and D ∈ (q ) .

14. Let А, B, C, D be four given points and let

S A ( D) = D1 , S B ( D1 ) = D2 , SC ( D2 ) = D3 ,
S A ( D3 ) = D4 , S B ( D4 ) = D5 , SC ( D5 ) = D6 .
Prove that D = D6 !

15. Given points Oi , i = 1, 2, 3, 4 and a line segment A0 B0 . Let Si , i = 1, 2, 3, 4 be a

point reflection centered at Oi , i = 1, 2, 3, 4 and let
Ai Bi = Si ( Ai −1Bi −1 ), i = 1, 2, 3, 4 .
Prove that A0 A 4 = B0 B 4 .

16. Does the figure F = { A, B, C} have a center of reflection?

17. In which case a figure consisting of two semilines is a point reflective figure?

18. Given circles K '(O ', R ') and K ''(O '', R '') . In which case a figure consisting of the
circles ( K ') and ( K '') is a point reflective figure?

223
19. If a figure F is point reflective, then it has either a unique or, infinitely many centers
of reflection. Prove it!

20. Given circles K '(O ', R ') and K ''(O '', R '') and point A. Draw a line (a) through
A so that A is the midpoint of the line segment MN, for M ∈ (a ) ∩ ( K ') and
N ∈ (a ) ∩ ( K '') .

21. Given four distinct points A, B, C, D on a circle (K) and a point M on the chord CD.
Determine a point X on the circle (K), so that the lines AX and BX on the chord CD
intercept a line segment ST for which M is the middle point.

22. Given a rotation SC,α , α ≠ 0, π . Are there any lines which are fixed lines under this
rotation?

23. Given a rotation SC,α . Prove that a circle K(O,R) is fixed if and only if O ≡ C .

24. Given lines (p) and (q). In which case there exists a rotation SC,α , so that
SC,α ( p ) = q ?

25. Given circles K(O,R) and K '(O ', R ') . In which case there exists a rotation SC,α , so
that SC,α ( K ) = K ' ?

26. Given two circles ( K ') and ( K '') and a point A. Construct an equilateral triangle
ABC, so that B ∈ ( K ') and C ∈ ( K '') .

27. Given three parallel lines (p), (q) and (r). Construct an equilateral triangle ABC, so
that A ∈ ( p ), B ∈ (q ), C ∈ (r ) .

28. Given three concentric ( K ') , ( K '') and ( K ''') . Construct an equilateral triangle
ABC, so that A ∈ ( K '), B ∈ ( K ''), C ∈ ( K ''') .

29. Given a line (p), a circle (K) and a point O. Construct an equilateral triangle ABC
centered at O, so that two of its vertices are on (p) and (K), respectively.

30. Given two circles (K), ( K ') and a point O. Construct an equilateral triangle ABC
centered at O, so that two of its vertices are on (K) and ( K ') , respectively.

224
31. In a triangle ABC inscribe a rhombus with acute angle α = 60° , so that its two
adjacent vertices are on the side AB, and the other two on the sides BC and AC,
respectively.

32. In a circle K(O,R) inscribe a triangle ABC, which is similar to a given triangle
PQR.

33. Given intersecting lines (p), (q) and a circle (K). Construct a circle so that it tangents
both lines (p) and (q) and a circle (K).

34. Let Η and Η1 be homotheties with a common (mutual) center O and coefficients
a and a1 , respectively. Prove that Η  Η1 is also a homothety and Η  Η1 = Η1  Η
holds true.

35. а) Prove that the composition of a point reflection and a homothety with coefficient
a ≠ −1 is homothety.
b) Prove that the composition of a homothety with coefficient a ≠ −1 and a point
reflection is homothety.

36. Prove that a composition of rotation around α ≠ 0°,180° and a homothety is a

similarity such that it is not a homothety.

37. Prove that a composition of two reflections is either translation or rotation.

38. Prove that each translation can be expressed as a composition of two reflections

39. Prove that each rotation can be expressed as a composition of two reflections.

40. Let (a), (b) and (c) be three parallel lines. Prove that the composition of the reflections
σa , σb and σc is a reflection.

41. Prove that a composition of a reflection and a homothety is similarity such that it is
not homothety.

42. Prove that there does not exist any similarity such that it is not: movement, ho-
mothety, composition of rotation and homothety, composition of reflection and ho-
mothety.

225
43. If z1 , z2 , z3 and z4 are four distinct points on a circle, then their дворазмер is a real
number. Prove it!

44. Determine the set of points z such that under the Möbius transformation w = 2ziz+ 2−1i
map to the following set {w | | w | = 1} .

45. Detrmine such a Möbius transformation that a semiplane {z | Im z > 0} maps

to the circle {z | | z | < 1} and furthermore the points z = i and z = ∞ under that
transformation will be mapped to w = 0 and w = −1 , respectively.

46. Find a condition which has to be satisfied, so that under the Möbius transfor-
az +b , the circle circle {z | | z | < 1} will be mapped to the semiplane
mation w = cz +d
{z | Im z > 0} .

47. Determine the Möbius transformation so that the points 0, −i, −1 map at i, 1, 0,
respectively.

48. Determine the Möbius transformation so that the points i, −i, 1 map at 0, 1, ∞,
respectively.

49. Prove that any three distinct points on a circle are not collinear.

50. Given the chords AB and CD of a circle so that AC = BD . Prove that either or
AB || CD or AD || BC .

51. In an acute triangle ABC, B ' and C ' are feet t of the altitudes at the vertices B and
C, respectively. The circle with diameter AB meets the line CC ' at M and N, and the
circle with diameter AC meets the line BB ' at P and Q. Prove that the quadrilateral
MNPQ is cyclic.

52. Let ABCD be a quadrilateral such that the inner angles at the vertices A, B and C
are congruent. Prove that the point D, the circumcentar and the orthocenter of the
triangle ABC are collinear.

53. In a circle k is inscribed a hexagon ABCDEF, so that the sides AB, CD and EF are
congruent with the radius of k. Prove that the midpoints of the other three sides are
vertices of an equilateral triangle.

226
54. Isoscaled triangles BCD, CAE and ABF, whose bases are BC, CA and AB, respectively,
are constructed on the outer part of a triangle  ABC. Prove that the perpendiculars
drawn at the vertices A, B and C to the lines EF, FD and DE, respectively, are
concurent.

55. Let the quadrilateral ABCD be a cyclic and let E and F be the feet of the perpendiculars
plot at the intersection of diagonals to the sides AB and CD, respectively. Prove that
the line EF is perpendicular to the line which passes through the midpoints of the
sides AD and BC.

56. Prove that the midpoints of the altitudes of a triangle are collinear if and only if the
triangle is right angled triangle.

57. The feet of the perpendiculars in an acute triangle  ABC are A' , B ' and C ' . If
A '', B '' and C '' are the touching points of the incircle of the triangle  A ' B ' C ' , then
prove that the Euler lines of  ABC and  A '' B '' C '' coincide.

58. Let ABCD be a convex quadrilateral such that its diagonals AC and BD are
perpendicular to each other and let E = AC ∩ BD . Prove that the points symmetric
to E with respect to the lines AB, BC, CD and DA form a cyclic quadrilateral.

59. Let AK, BL, CM be the altitudes ofa triangle ABC, H be its orthocenter and P the
midpoint of the line segment AH. If BH ∩ MK = S and LP ∩ AM = S , then
TS ⊥ BC . Prove it!

60. Let AD, BE, CF be the altitudes of a triangle ABC. Let A' , B ' , C ' be such that
AA ' = k AD, BB ' = k BE , CC ' = kCF , for each k ∈ R , k ≠ 0 . Determine all k such
that for any non-isoscale triangle ABC the triangles ABC and A ' B ' C ' are simillar.

61. Given a triangle  ABC and points D, E, F on its altitudes BC, CA, AB, respectively,
so that
BD = CE = AF = 1− k , k ∈ R .
DC EA FB k
Determine the locus of points of the circumcenters of DEF for k ∈ R .

62. Let H ' and H '' be the feet of the perpendiculars at the orthocenter H of  ABC to
the bisector of the outer and the inner angle at C. Prove that the line H ' H '' consists
of the midpoint of the side AB.

227
63. Given an acute triangle ABC and a point D on its inner part, so that
∠ ADB = ∠ ACB + 90° and AB ⋅ CD = AD ⋅ BC . Determine
AB⋅CD .
AC ⋅BD

64. Tangents AM and AN and a line which crosses the circle at k and L are constructed
on a circle k at the point A (which is positioned out of the circle k). Let l be any line
which is parallel to AM and let KM and LM meet l at the points P and Q, respectively.
Prove that the line MN bisects the line segment PQ.

65. On the sides BC, CA and AB of a triangle ABC are given points D, E and F,
respectively, so that BD = CE = AF . Prove that the triangles ABC and DEF have
common circumcenter if and only if the triangle ABC is an equilateral triangle.

66. Let be given a cyclic quadrelateral ABCD. Prove that the incenters of the triangles
ABC, BCD, CDA, DAB are vertices of a rectangle.

67. Let I be the incenter of atriangle ABC and let D and E be the midpoints of the
sides AC and AB, respectively. Let AB ∩ DI = S and AC ∩ EI = Q . Provе that
AP ⋅ AQ = AB ⋅ AC if and only if ∠ CAB = 60° .

68. Let M be a point of the interior part of the square ABCD and A ', B ', C ', D ' be
the intersections of the lines AM, BM, CM, DM and the circumcircle of the square
ABCD, respectively. Prove that
A ' B ' ⋅ C ' D ' = A ' D ' ⋅ B 'C ' .

69. Let ABCD be a cyclic quadrilateral and let F = AC ∩ BD and E = AD ∩ BC . If

M and N are the midpoints of the sides AB and CD, respectively, then prove that
MN = 12 AB − CD .
EF CD AB

70. The points A ', B ', C ' are symmetric to the points A, B, C with respect to the sides
BC, CA, AB, respectively. Determine the type of the triangle ABC, so that the triangle
A ' B ' C ' is an equilateral triangle?

71. Let O be the circumcenter and R be the circumradius of a triangle ABC. The in-
circle of the triangle ABC, with radius r, touches the sides BC, CA, AB at points
A ', B ', C ' , respectively. Let the lines determined by the midpoints of the line
segments AB ' and AC ' , BA' and BC ' , CA' and CB ' intersect at C '' , A'' and

228
B '' . Prove that the circumcenter of the triangle A '' B '' C '' is O and the circumradius
is R + 2r .

72. Let the trapezoid ABCD, AB || CD, AB > CD , not be isoscaled and let it be
circumscribed about a circle centered at I. The incircle tangents the side CD at E.
Let M be the midpoint of the side AB and moreover MI and CD intersect at F. Prove
that DE = FC if and only if AB = 2CD .

73. Given a cyclic hexagon ABCDEF so that AB = CD = EF and the diagonals AD, BE

( ) . Prove it!
2
and CF are concurrent. If p = AD ∩ CE , then CP = AC
PE CE

74. Given a triangle ABC. A ', B ', C ' are the midpoints of the arcs BC  , CA
,  AB , such
that each of them does not consist of the point A, B, C, respectively. The lines
A ' B ', B ' C ', C ' A ' divide the sides of the triangle in six parts. Prove that “the middle”
parts are congruent if and only if the triangle ABC is an equilateral triangle.

75. Let  ABC be such a triangle that ∠ ABC = 60° . Let the line IF be parallel to AC
(I is the incenter, and F lies on the side AB). The point P is on the side BC and
3BP = BC . Prove that ∠ BFP = 12 ∠ ABC .

76. The angle at A is the smallest angle in a  ABC. The points B and C divide the
circumcircle of the triangle in two arcs. Let U be the interior point of the arc between
B and C which does not consist of A. The bisectors of the line segments AB and AC
meet the line AU at points V and W, respectively. The lines BV and BW meet at T.
Prove that AU = TB + TC .

77. Let ABCD be a convex quadrilateral so that AB is not parallel to CD and AD is not
parallel to BC. The points P, Q, R, S are such chosen on the sides AB, BC, CD, DA,
respectively, that the quadrilateral PQRS is parallelogram. Find the locus of the
intersections of all such quadrilaterals PQRS.

78. The incircle of a triangle ABC tangents the sides BC, CA, AB at the points E, F, G,
respectively. Let AA ', BB ', CC ' be the intercepts of the bisectors of the inner angles
of the triangle ABC. Let K A , K B , KC be the points where the second tangents to
the incircle drawn at A ', B ', C ' , respectively. Let P, Q, R be the midpoints of the
sides BC, CA, AB, respectively. Prove that the lines PK A , QK B , RKC concure on
the incircle of the triangle ABC.

229
79. Let AD, BE, CF be the altitudes of the triangle ABC, and A ', B ', C ' are points on
them respectively, so that AA ' = BB ' = CC ' = k holds. Determine each values of k so
AD BE CF
that the triangles ABC and A ' B ' C ' are similar.

80. (Gauss’s theorem). If the line l meets the lines which consist of the sides BC, CA,
AB of the triangle ABC at A ', B ', C ' , respectively, then prove that the midpoints of
the line segments AA ', BB ', CC ' are collinear.

81. Given a triangle ABC and a point T. Let P and Q be the feet of the perpendiculars at T
to the lines AB and AC, respectively, and let R and S be the feet of the perpendiculars
at A to the lines TC and TB, respectively. Prove that the intersection of the lines PR
and QS lies on the line BC.

82. Let PQRS be a cyclic quadrilateral, such that the lines PQ and RS are not parallel.
Consider the set of all circles through P and Q and the set of all circles through R
and S. Determine the set of all touching points between the circles which belong to
these two sets.

83. Given a circle k and a point P positioned in the outer part of the circle. A variable
line s, such that it consists of the point P, meets the circle at the points A and B. Let
M and N be the midpoints of the arcs determined by the points A and B and let C be
2
a point positioned on the line segment AB so that PC = PA ⋅ PB holds. Prove that
the angle ∠ MCN does not depend on the choise of the line s.

84. Two circles k1 and k2 touch at a point M. The radius of k1 is greater than the radius
of k2 . Let A be any point on the circle k2 such that it is not placed on the line which
connects the centers of the circles, B and C be points on k1 so that AB and AC are
its tangents. The lines BM and CM remeet k2 at E and F, respectively, and D is the
intersection of the tangent to k2 at A and the line EF. Prove that the locus of the
point D, when A moves on k2 , is a line.

85. On a plane are given two circles k1 and k2 such that they meet at points A and B.
The tangents to k1 at A and B intersect at K. Let M be any point of the circle k1 and
let
MA ∩ k2 = { A, P} , MK ∩ k1 = {M , C} and CA ∩ k2 = { A, Q} .
Prove that the midpoints of the line segment PQ is placed on the line MC and PQ
passes through a fixed point when M moves round the circle k1 .

230
86. Let ABC be a triangle so that ∠ ACB = 2∠ ABC and let D be a point on the line
segment BC so that CD = 2 BD holds. The line segment AD is extended through D
to the point E so that AD = DE holds. Prove that the following is satisfied
∠ ECB + 180° = 2∠ EBC .

87. Given a triangle A1 A2 A3 and a line p which passes through a point P and meets
the sides A2 A3 , A3 A1 , A1 A2 at X1 , X 2 , X 3 , respectively. Let Ai P meetthe
circumcircle of the triangle A1 A2 A3 at a point Ri , for i = 1, 2, 3 . Prove that the lines
X1R1 , X 2 R2 , X 3 R3 concur at a point which belongs on the circumcircle of the
triangle A1 A2 A3 .

88. Two circle with defferent radii meet at points A and B. Their mutual tangents are
MN and ST. Prove that the orthocenters of the triangles AMN, BMN, AST, BST are
vertices of a rectangle.

89. Given a cyclic quadrilateral ABCD. The lines AD and BC meet at a point E, so that
C is between B and E. The diagonals AC and BD meet at F. Let M be the midpoint
of CD and let N ≡/ M be the point on the circumcircle of the triangle ABM such that
AN = AM . Prove that points E, F and N are collinear.
BN BM

90. The diameter of a circle k is placed on a line l. Let C and D be points on k. The tangents
to k at C and D consecutively meet the line l at B and A, so that the center of the
circle is between B and A. Let E = AC ∩ BD and F be the foot of the perpendicular
at E to l. Prove that EF is the bisector of ∠CFD .

91. Let ABCD be a convex quadrilateral whose sides BC and AD are congruent, but not
parallel. Let E and F be interior points of the sides BC and AD, respectively, so that
BE = DF . The lines AC and BD intersect at P, the lines BD and EF intersect at Q
and the lines EF and AC intersect at R. Let’s consider the triangles PQR which are get
for all points E and F. Prove that the circumcircles of these triangles have a common
point, such that it differs from P.

92. Let O be an interior point for the acute triangle  ABC. The circles centered at the
midpoints of the sides of the triangle  ABC, such that each of them passes through
O, concur at K, L, M (K, L, M differ from O). Prove that O is the incenter of the
triangle  KLM if and only if O is the circumcenter of the triangle  ABC.

231
93. Let M and N be points of the interior of a triangle  ABC such that ∠ MAB = ∠ NAC
and ∠ MBA = ∠ NBC hold. Prove that
AM ⋅ AN + BM ⋅BN + CM ⋅CN = 1.
AB⋅ AC BA⋅BC CA⋅CB

94. Let  ABC be such a triangle that ∠ A = 90° and ∠ B < ∠ C hold. The tangent at
A of its circumcircle Γ meets the line BC at D. Let E be the image of A under line
symmetry with respect to the line BC, X be the foot of the perpendicular at A to BE
and Y be the midpoint of the line segment AX. Let the line BY remeets Γ at a point Z.
Prove that the line BD is a tangent to the circumcircle of the  ADZ.

95. Given  ABC and points A1 ∈ BC , B1 ∈ AC , C1 ∈ AB such that  ABC and  A1B1C1
are simmilar. If the orthocenters or the incenters of  ABC and  A1B1C1 coincide,
then  ABC is an equilateral triangle. Prove it!

96. Let be given points A, B and C. Determine the locus of a point D so that
DA ⋅ DB ⋅ AB + DB ⋅ DC ⋅ BC + DC ⋅ DA ⋅ CA = AB ⋅ BC ⋅ CA
holds.

97. Prove that the length of the side of a regular nonagon is equal to the difference of the
lengths of its longest and shortest diagonal.

98. Prove that for any regular n-gon inscribed into a circle with radius r the, product of
n n ( n −1)
all sides and diagonals is equal to n 2 r 2
.

99. On the circumcircle of a regular 2n-gon A1 A2 ... A2n is chosen an arbitrary point P.
Prove that the sum of the squared distances between the point P and the vertices
whose indexis are even numbers is equal to the sum of the squared distances between
the point P and the vertices whose indexis are odd numbers.

100. Let A0 A1 A2 ... A2n be a regular polygon, P be an arbitrary point of the smaller arc

A0 A2n of its circumcircle and m be a positive integer, 0 ≤ m < n . Prove that
n n
∑ PA 22mk +1 = ∑ PA 22mk −+11 .
k =0 k =1

101. Let A0 A1... An −1 be a regular n-gon inscribed in a circle whose radius is r. Prove
that for any point P of the circumcircle and a positive integer m < n ,

232
n −1

k =0
∑ PAk
2m
( )
= 2mm nr 2m
holds true.

102. Let h1 , h2 ,..., hn be the distances between an arbitrary point P of the smaller
arc A0 An −1 of the circumcircle of regular n-gon A0 A1... An −1 and the lines
A0 A1 , A1 A2 ,..., An −1 A0 . Prove that
1 + 1 + ... + 1 = 1 .
h1 h2 h hn −1 n

103. Given a regular n-gon A1 A2 ... An and a point P of the smaller arc 
A1 An . Let d k be
the distance between the points P and Ak . Prove that
1 + 1 + ... + 1 = d 1d .
dd d d
1 2 2 3d d n −1 n 1 n

104. Let P be any point on the circumcircle of a regular 2n-gon A1 A2 ... A2n . If
p1 , p2 ,..., p2n are the distances between the point P and the lines which consists
of the sides A1 A2 , A2 A3 ,..., A2n A1 , respectively then p1 p3 ... p2n −1 = p2 p4 ... p2n .
Prove it!

105. Let n be a prime number and let H1 be a convex n-gon. The polygons H 2 , H 3 ,..., H n
are constructed consecutive: the vertices of the polygon H k +1 are obtained by
applying the the symmetry through the k-th adjacent vertex to the vertices of the
polygon H k in a positive direction. Prove that the polygons H1 and H n are
similar.

106. Let A0 , A1 ,..., A2k be cosequtive points on a circle, such that they divide the circle
in 2k + 1 congruent arcs. The point A0 is plot by chords with each other points.
These 2k chords divide the circle in 2k + 1 parts. These parts are alternately colored
with white and black color, such that the number of the white parts is greater for
one than the number of the black ones. Prove that the black area is greater than the
white one.

107. The vertices of a regular n-gon are coloured with a few colors (each vertex with
only one colour) so that the vertices coloured with the same colour form a regular
polygon. Prove that two of these polygons are similar.

108. Let the points A, B, C, D and E be such that ABCD is a parallelogram, and BCED
is a cyclic quadrilateral. Let l be a line which consists of the point A and intersects
the line segment DC at an inner point F, and the line BC at C. If EF = EG = EC ,
then l is the bisector of the angle DAB. Prove it!

233
109. Let H be the orthocenter of acute triangle ABC. The circle centered at the midpoint
of the line segment BC, such that it consists of the point H, meets the line BC at A1
and A2 . Analogously, the circle centered at the midpoint of the line segment CA,
such that it consists of the point H, meets the line CA at B1 and B2 , and the circle
centered at the midpoint of the line segment AB, such that it consists of the point
H, meets the line AB at C1 and C2 . Prove that the points A1 , A2 , B1 , B2 , C1 and
C2 belong on a same circle.

110. Let ABCD be convex quadrilateral so that BA ≠ BC and k1 and k2 be the incircles
of the triangles ABC and ADC, respectively. Let it exists a circle k such that it
touches the extension of the side BA at A and the extension of the side BC at C, and
it likewise touches the lines AD and CD. Prove that the common outer tangents to
the circles k1 and k2 intersect at a point on the circle k.

111. Let O be the circumcenter of triangle ABC, P and Q be inner point for the line
segments CA and AB, respectively, K, L and M be the midpoints of the line segments
BP, CQ and PQ, respectively and Γ be a circle which consists of the points K, L
and M. If the line PQ is a tangent to the circle Γ, then OP = OQ . Prove it!

112. Let I be the incenter, and Γ be the circumcircle of a triangle  ABC. Let the line AI
 , and F be a point on
intersect Γ at A and D, and let E be a point on the arc BDC
the line segment BC so that
∠ BAF = ∠ CAE < 12 ∠ BAC .
Let G be the midpoint of the line segment IF. Prove that the lines DG and EI
intersect at a point of the circle Γ.

113. Let P be a inside point of triangle  ABC and the lines AP, BP and CP remeet the
circumcircle Γ of the  ABC at K, L and M, respectively. The tangent to the circle Γ at
the point C meets the line AB at S. Let SC = SP . Prove that MK = ML .

114. Let ABC be an acute triangle and let Γ be its circumcircle. Let l be any tangent to
the circle Γ and let la , lb and lc be lines symmetric to l with respect to BC, CA
and AB, respectively. Prove that the circumcircle of the triangle determined by the
lines la , lb and lc touches the circle Γ.

115. Let ABC be a scalene acut triangle such that AC > BC satisfies. Let O be the
circumcenter, H be the orthocenter, and F the foot of the altitude at the vertex C.

234
Let P be a point on the line AB, such that it differes from A, and AF = PF holds,
and M be the midpoint of the line segment AC. Let X be the intersection of PH and
BC, Y be the intersection of OM and FX, and Z be the intersection of OF and AC.
Prove that the points F, M, Y and Z are on a same circle.

116. In  ABC, M and N are points on the sides AB and AC, respectively, so that the line
MN is parallel to the side BC. Let P be the intersection of the lines BN and CM. The
circumcircles of the triangles  BMP and  CNP meet at two distinct points P and
Q. Prove that ∠ BAQ = ∠ CAP .

117. Let  ABC be not isoscaled triangle. Let AD, BE, CF be the bisector of the angles
of this triangle ( D ∈ BC , E ∈ AC , F ∈ AB ). Let K a , Kb , K c be points on the
incircle of the  ABC so that DK a , EKb , FK c are tengents to the incircle and
K a ∉ BC , Kb ∉ AC , K c ∉ AB . Let A1 , B1 , C1 be the midpoints of the sides BC,
CA, AB. Prove that the lines A1K a , B1Kb , C1K c are concurrent on the incircle of
 ABC.

118. Let  ABC not be isoscaled triangle and k be its incircle centered at S. The circle
k tangents the sides BC, CA, AB at points P, Q, R, respectively. The line QR meets
BC at M. Let a circle which consists of the points B and C touches k at N. The
circumcircle of the triangle MNP meets the line AP at L which differes from P.
Prove that the points S, L and M are collinear.

119. In an acute triangle  ABC a point M is the midpoint of the side BC, and points D,
E, F are feet of the altitudes at vertices A, B, C, respectively. Let H be the ortho-
center of the triangle  ABC, S be the midpoint of the line segment AH, and G be
the intersection of the line segments FE and AH. If N is point of intersection of the
median AH and the circumcircle of the  BCH, prove that ∠ HMA = ∠ GNS .

120. In  ABC, M and N are points on the sides AB and AC, respectively, so that the
line MN is parallel to BC. Let P be the point of intersection of BN and CM. The
circumcircles of the triangles  BMP and  CNP meet at two distinct points P and
Q. Prove that ∠ BAQ = ∠ CAP .

235
REFERENCES

1. Alfors, L. V.: Complex analysis, second edition, New-York-St. Louis-San Fran­cis­co-

Toronto-London-Sydney, 1966
2. Anevska, K., Gogovska, V., Malcheski, R.: The role of complex numbers in
interdisciplinary integration in mathematics teaching, Elsevier, Procedia - Social and
Behavioral Sciences, 2014
3. Anevska, K.: Methodical approach for introduction of an exponential entry of complex
numbers in secondary education, Proc. V Congres of SMM, Ohrid, September,
2014
4. Arslanagić, Š.: Matematika za nadarene (drugo izdanje), Bosanska riječ, Sarajevo,
2005
5. Ašić, M. i dr.: Međunarodne matematičke olimpijade, DM Srbije, Beograd, 1986
6. Conway, J. B.: Functions of One Complex Variable, Springer-Werlag, New-York-
Haidelberg-Berlin, 1987
7. Đurković, R.: Matematička takmičenja srednjoškolaca u Jugoslaviji 1990, DM Srbije,
Beograd, 1991
8. Engel, A.: Problem-Solving Strategies, Springer-Verlag, New York-Berlin-
Heidelberg, 1997
9. Govedarica, V.: Matematička takmičenja u Republici Srpskoj, Zavod za udžbenike I
nastavna sredstva, Istočno Sarajevo, 2007
10. Grozdev, S., V. Nenkov.: Design of Information-educational Environment for the
Establishment of New Facts in Mathematics, Procedia – Social and Behavioral
Sciences, volume 128, (22 April 2014), International Conference: Education and
psychology challenges – teachers for the knowledge society - 2nd edition epc – tks
2013, 211-216, ISSN 1877-0428,�
11. Kadelburg, Z., Mladenović, P.: Savezna takmičenja iz matematike, DMS, Beograd,
1990
12. Mateljević, M.: Kompleksne funkcije 1 & 2, DMS, Beograd, 2006
13. Mitrinović, D. S.: Kompleksna analiza, zbornik zadataka i problema, Naučna knji­ga,
Beograd, 1972
14. Mitrović, M.; Ognjanović, S.; Veljković, M.; Petković, Lj.; Lazarević, N.: Geometrija
za I razred Matematičke gimnazija, Krug, Beograd, 1998
15. Mladenović, P.; Ognjanović, S.: Pripremni zadaci za matematička takmičenja za
učenike srednjih škola, DMS, Beograd, 1991

236
16. Pavković, B.; Veljan, D.: Elementarna matematika I, Tehnička knjiga, Zagreb, 1992
17. Spiegel, A. M.: Theory and problems of complex variabbles, McGraw-Hill Book
Co., Singapore, 1981
18. Бойваленков, П.; Николов, Н.: Национални олимпиади по математика 2008,
УНИМАТ СМБ, София, 2008
19. Гроздев, С., В. Ненков. Геометрични места при движение на вписан
четириъгълник. Математика и математическо образование, 2013, 42, 373-382,
ISSN� ����������
1313-3330.
20. Гроздев, С., В. Ненков. Допирателни окръжности, породени от инциденти точка
и права, 2012, 2, 151 – 160,����������������
���������������
ISSN� ����������
1310-2230�.
21. Гроздев, С., В. Ненков. Една зависимост, породена от конични сечения. Мате­
матика и математическо образование, 2008, 37, 312 – 319, ���������������
ISSN� ����������
1313-3330�.
22. Гроздев, С., В. Ненков.: Около ортоцентъра в равнината и пространството.
София, Архимед 2, 2012, ISBN 978-954-779-145-9.
23. Гроздев, С., В. Ненков.: Ортоцентър на вписан четириъгълник, Математика
плюс, 2011, 4, 63 – 69, ISSN 0861-8321.
24. Јанковић, З.; Каделбург, З.; Младеновић, П.: Међународне и балканске матема­
тичке олимпијаде 1984-1995, ДМС, Београд, 1995
25. Карамата, Ј.: Комплексан број, са применом на елементарну геоме­трију, Научна
књига, Београд, 1950
26. Кендеров, П., Табов, Й.: Бьлгарски олимпиади по математика, Народ­на просвета,
София, 1990
27. Кртинић, Ђ.: Математичке олимпијаде средњошколаца 2007-2012 године, ДМ
Србије, 2012
28. Мадески, Ж.; Самарџиски, А.; Целакоски, Н.: Збирка задачи по геометрија,
Просветно дело, Скопје, 1981
29. Малчески, Р., Аневска, К.: Хомотетија, Сигма, 2014, Скопје
30. Малчески, Р.: Основи на математичка анализа, Ун. Св. Кирил и Методиј, Скопје,
2001
31. Малчески, Р.: Теорема на Менелај, Сигма 43, Скопје, 1999
32. Малчески, Р.; Малчески, А.: Пресликувања во рамнина преку комплекс­ни
броеви I, Сигма 49, Скопје, 2000
33. Малчески, Р.; Малчески, А.: Пресликувања во рамнина преку комплекс­ни
броеви II, Сигма 50, Скопје, 2000

237
34. Малчески, Р.; Малчески, А.: Пресликувања во рамнина преку комплекс­ни
броеви III, Сигма 51, Скопје, 2001
35. Малчески, Р.; Малчески, А.: Пресликувања во рамнина преку комплекс­ни
броеви IV, Сигма 42, Скопје, 2001
36. Моденов, П. Ц.: Задачи по геоме­трии, Наука, Москва, 1979
37. Ненков, В.: Върху една задача и нейното пълно решение, Математика и
информатика, 1998, 3, 70 – 72, ISSN 1310-2230.
38. Ненков, В.: Комплексни числа и педални окръжности, Математика и
информатика, 1997, 5-6, 37 – 41, ISSN 1310-2230.
39. Ненков, В.: Някои геометрични места в равнината на триъгълника, Математика
плюс, 2005, 1, 53 – 59, ISSN 0861-8321.
40. Ненков, В.: Четири Ойлерови прави в равнината на триъгълник, Математика и
информатика, 2005, 2, 68 – 80, ISSN 1310-2230
41. Рудин, У.: Основи на математическая анализ, Наука и изкуство, София, 1973
42. Самарџиски, А.: Хомотетија, ин­вер­зија и задачите на Аполониј, ПМФ, Скопје,
1988
43. Сидоров, Ю. В.; Федорюк, М. В.; Шабунин, М. И.: Лекции по тории функций
комплексного переменного, Наука, Москва, 1989
44. Тонов, И. К.; Сидеров, П. Н.: Прило­жение на комплексните числа в геометрията,
Наука, София, 1981
45. Тренчевски, К.; Урумов, В.: Меѓународни олимпијади по математика, Природно-
математички факултет, Скопје, 2000
46. Шабат, Б. В.: Введение в комплексный анализ, часть I, Наука, Москва, 1976

238
GEOMETRY OF COMPLEX NUMBERS
Risto Malcheski, Sava Grozdev, Katerina Anevska

Bulgarian. First edition in english

Printed by Alliance print – Sofia

240