Eigen Systems

Prepared by: Naresh Vasant Afre

[email protected]

M.Sc, M.Phil, JRF-NET, Mah-SET in Mathematics.

Contents
1 Eigenvalues and Eigenvectors 2
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Eigenvalues . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Eigenvectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2 Diagonalization 14
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2 Similarity of matrices . . . . . . . . . . . . . . . . . . . . . . . 15
2.3 Diagonalizable matrices . . . . . . . . . . . . . . . . . . . . . . 16

3 Cayley Hamilton theorem 22

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2 Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

4 Functions of square matrix 27

4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.2 Using diagonalizability . . . . . . . . . . . . . . . . . . . . . . 27
4.3 Using Cayley-Hamilton theorem . . . . . . . . . . . . . . . . . 30

1
1 Eigenvalues and Eigenvectors
1.1 Introduction
Sometimes, the effect of multiplying a vector by a matrix is to produce simply
a scalar multiple
of thatvector.
For example,
3 2 1 5 1
= =5 .
−4 9 1 5 1
We often want to know to what extent this phenomenon occurs for a given
matrix. That is, we ask for all the instances of the equation AX = λX, where
A is given matrix but we know neither the vector X nor the scalar λ. Because
there are two unknowns on the right multiplied together, this is not a linear
problem! If λ were fixed, and we only wanted to find all the appropriate
X − vectors, it would be a standard linear problem. Even easier is the case
when x is known, and we only want to determine λ, each of these is a linear
problem. But it is nonlinear if λ and X are both unknown. Therefore, we
must expect to encounter some obstacles in solving this problem.

2
1.2 Eigenvalues
First we note that the vector 0 always solve the equation AX = λX.
We rule out that case as uninteresting.
The properly posed question is therefore:
For what scalars λ is there a nontrivial solution (X ̸= 0) to the equation
AX = λX where the matrix A is given.
This is known as the eigenvalue problem for the given matrix A.

Definition 1.1. Let A be any real square matrix.

A number λ is an eigenvalue of A if the equation AX = λX is true for some
nonzero vector X.
The vector X is an eigen vector associated with the eigenvalue λ.

In the example given in the introduction λ = 5 and X = (1, 1)t .

(For notational convenience we will denote (x1 , x2 , x3 )t := (x1 , x2 , x3 ).
In matrix theory, some terminology has changed overtime.
For example, eigenvalues have been known as characteristic values, latent
roots and eigenwerte.

Procedure to find eigenvalues: If λ is an eigenvalue of A, then the equa-

tion AX = λX has a nontrivial solution.
Consequently, the equation (A − λI)X = 0 has a nontrivial solution.
Hence we have Det(A − λI) = 0.
So for a given matrix A we are interested in finding the values of λ for which
Det(A − λI) = 0.
Det(A − λI) is called characteristic polynomial of A, which will be of
degree n, if A is an n × n matrix.
The equation Det(A − λI) = 0 is called characteristic equation of A, solv-
ing which gives n eigenvalues.

Challenge
1.1. Find the characteristic equation and the eigenvalues of the
2 1
matrix .
4 −1
Solution: The
characteristicequation of A is
2−λ 1
Det(A − λI) = = 0. Next we expand the determinant and
4 −1 − λ
obtain: (2 − λ)(−1 − λ) − 4 = 0. This leads to λ2 − λ − 6 = 0.

3
Therefore λ2 − λ − 6 = 0 is the required characteristic equation of A solving
it for λ gives the eigenvalues of A.
Hence, the given matrix A has precisely two eigenvalues λ = 3 and λ = −2.

 
7 0 0
Challenge 1.2. Find the eigenvalues of the matrix  −8 3 0  .
0 0 0

Solution: The
characteristic equation
of A is
7−λ 0 0

Det(A − λI) = −8 3 − λ 0 = 0.
0 0 0−λ
On expanding we obtain: (7 − λ)(3 − λ)(−λ) = 0.
Hence, the given matrix A has three eigenvalues λ = 7, 3, 0.

Another way to find eigenvalues for 2 × 2 and 3 × 3 matrices:

Note: T race(A) denotes the sum of elements on the main diagonal of A.

1. If A is a 2 × 2 matrix then the characteristic polynomial of A is given by

λ2 − T race(A)λ + Det(A). (1)

2. If A is a 3 × 3 matrix then the characteristic polynomial of A is given by

λ3 − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A). (2)

Where Aii is the deteminant of a 2 × 2 matrix obtained from A by deleting

ith row and ith column.

Solving Challenge 1.1 using this method we get,

λ2 − T race(A)λ + Det(A) = λ2 − 1λ + (−6).
Which gives λ2 − λ − 6 = 0 as the characteristic equation.

4
 
4 6 6
Challenge 1.3. Find the eigenvalues of the matrix  1 3 2 .
−1 −5 −2
Solution: The characteristic polynomial is
λ3 − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A)
= λ3 − 5λ2 + (4 + (−2) + 6)λ − 4
= λ3 − 5λ2 + 8λ − 4.
Solving λ3 − 5λ2 + 8λ − 4 = 0 gives λ = 1, 2, 2 as the eigenvalues.

0 1
Challenge 1.4. Find the eigenvalues of the matrix .
−1 0

Solution: The characteristic equation of A is

2
λ − T race(A)λ + Det(A)
= λ2 − 0λ + 1.
Solving λ2 + 1 = 0 gives λ = i, −i as the eigenvalues.

Observation: A real matrix may have complex eigenvalues.

We will focus only on the matrices having real eigenvalues.

More eigenvalues challenges in the next section.

5
1.3 Eigenvectors
In previous section we learned how to find the eigenvalues of the given matrix
A. Here in this section, we are going to find the eigenvectors of matrix A by
using its eigenvalues. i.e., we will solve AX = λX in which A and λ(which
we are able to compute now) are known, so this is a linear problem to solve.

1 2
Challenge 1.5. Find the eigenvalues and eigenvectors of A = .
4 3
Solution: The characteristic polynomial is
λ2 − T race(A)λ + Det(A) = λ2 − 4λ − 5.
Solving λ2 − 4λ − 5 = 0 gives λ = 5, −1 as the eigenvalues.
Now we fix one eigenvalue and find the corresponding eigenvector/s.
Let λ = 5 be fixed,
we want to find a vector which satisfy AX = λX
which (A −λI)X
is equivalent to =0
1−5 2 x1 0
i.e., = ,
4 3 − 5 x2 0
−4 2 x1 0
= which gives x2 = 2x1 .
4 −2 x2 0
Therefore X = (x1 , x2 )t = (x1 , 2x1 )t = x1 (1, 2)t .
So X1 = (1, 2)t is an eigenvector corresponding to λ = 5.
Note: Any nonzero multiple of X1 is also an eigenvector of A corresponding
to λ = 5.
Now fix λ = −1,
1+1 2 x1 0
consider (A − λI)X = 0 i.e., = ,
4 3 + 1 x 2 0
2 2 x1 0
= which gives x2 = −x1 .
4 4 x2 0
Ttherefore X = (x1 , x2 )t = (x1 , −x1 )t = x1 (1, −1)t .
So X2 = (1, −1)t is an eigenvector corresponding to λ = −1.

Observation: Any nonzero multiple of eigenvector corresponding to λ is

again an eigenvector corresponding to the same λ.

6
 
2 −1 1
Challenge 1.6. Find the eigenvalues and eigenvectors of A =  1 2 −1  .
1 −1 2
Solution: The characteristic polynomial is
λ3 − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A)
= λ3 − 6λ2 + (3 + 3 + 5)λ − 6
= λ3 − 6λ2 + 11λ − 6.
Solving λ3 − 6λ2 + 11λ − 6 = 0 gives λ = 1, 2, 3 as the eigenvalues.
For λ = 1,     
2 − 1 −1 1 x1 0
consider (A − I)X = 0,  1 2 − 1 −1   x2 = 0  ,
 
    1  −1 2 − 1 x3 0
1 −1 1 x1 0
 1 1 −1   x2  =  0  , R2 − R1 , R3 − R1 gives
 1 −1 1   x3   0 
1 −1 1 x1 0
 0 2 −2   x2  =  0  .
0 0 0 x3 0
This implies that x1 − x2 + x3 = 0 and x2 = x3 substituting in the first
equation gives x1 = 0.
Therefore X = (x1 , x2 , x3 ) = (0, x2 , x2 ) = x2 (0, 1, 1).
Hence X1 = (0, 1, 1) is an eigenvector corresponding to λ = 1.
Note: As the rank of the coefficient matrix is 2 and the number of unknowns
are 3 therefore there is 3 − 2 = 1 eigenvector corresponding to λ = 2.
For λ = 2,     
2 − 2 −1 1 x1 0
consider (A − 2I)X = 0,  1 2 − 2 −1   x2  =  0  ,
    1  −1 2 − 2 x3 0
0 −1 1 x1 0
 1 0 −1   x2  =  0  .
1 −1 0 x3 0
This implies that x2 = x3 , x1 = x3 and x1 = x2 .
Therefore X = (x1 , x2 , x3 ) = (x1 , x1 , x1 ) = x1 (1, 1, 1).
Hence X2 = (1, 1, 1) is an eigenvector corresponding to λ = 2.
For λ = 3,     
2 − 3 −1 1 x1 0
consider (A − 3I)X = 0,  1 2 − 3 −1   x2  =  0  ,
1 −1 2 − 3 x3 0

7
    
−1 −1 1 x1 0
 1 −1 −1   x2  =  0  , R2 + R1 , R3 − R2 gives
 1 −1 −1  x3  0
−1 −1 1 x1 0
 0 −2 0   x2  =  0  .
0 0 0 x3 0
This implies that x1 + x2 − x3 = 0 and x2 = 0 substituting in first equation
gives x1 = x3 .
Therefore X = (x1 , x2 , x3 ) = (x3 , 0, x3 ) = x3 (1, 0, 1).
Hence X3 = (1, 0, 1) is an eigenvector corresponding to λ = 3.
 
3 2 4
Challenge 1.7. Find the eigenvalues and eigenvectors of A = 2  0 2 .
4 2 3
Solution: The characteristic polynomial is
λ − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A)
3

= λ3 − 6λ2 + ((−4) + (−7) + (−4))λ − 8

= λ3 − 6λ2 − 15λ − 8.
Solving λ3 − 6λ2 − 15λ − 8 = 0 gives λ = −1, −1, 8 as the eigenvalues.
For λ = 8,     
3−8 2 4 x1 0
consider (A − 8I)X = 0,  2 0−8 2   x2  =  0  ,
   4  2 3−8 x3 0
−5 2 4 x1 0
 2 −8 2   x2  =  0  , R2 + 2 R1 , R3 − 2R1 gives
5

 4 2 −5 x 3
   0
−5 2 4 x1 0
 0 − 36 18   x2  =  0  R3 + 5 R2 gives
5 5 2

 0 18 −9  x 3
   0
−5 2 4 x1 0
 0 − 36 18   x2  =  0  .
5 5
0 0 0 x3 0
This implies that −5x1 + 2x2 + 4x3 = 0 and x3 = 2x2 substituting in the first
equation gives x1 = 2x2 .
Therefore X = (x1 , x2 , x3 ) = (2x2 , x2 , 2x2 ) = x2 (2, 1, 2).
Hence X1 = (2, 1, 2) is an eigenvector corresponding to λ = 8.
For λ = −1,

8
    
3+1 2 4 x1 0
consider (A + I)X = 0,  2 0+1 2   x2 = 0  ,
 
    4 2 3+1 x3 0
4 2 4 x1 0
 2 1 2   x2  =  0  , R2 − 1 R1 , R3 − R1 gives
2

 4 2 4  x 3
   0
4 2 4 x1 0
 0 0 0   x2  =  0  .
0 0 0 x3 0
This implies that 4x1 + 2x2 + 4x3 = 0 i.e., x2 = −2x1 − 2x3 .
Therefore X = (x1 , x2 , x3 ) = (x1 , −2x1 −2x3 , x3 ) = x1 (1, −2, 0)+x2 (0, −2, 1).
Hence X2 = (1, −2, 0) and X3 = (0, −2, 1) are eigenvectors corresponding to
λ = −1.
Note: As the rank of the coefficient matrix (A + I) is 1 and the number of
unknowns are 3 therefore there are 3 − 1 = 2 eigenvectors corresponding to
λ = −1.

Observation: We can have more than one eigenvectors corresponding to

an eigenvalue.
 
3 −1 1
Challenge 1.8. Find the eigenvalues and eigenvectors of A =  7 −5 1  .
6 −6 2
Solution: The characteristic polynomial is
λ − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A)
3

= λ3 − 0λ2 + ((−4) + 0 + (−8))λ − (−16)

= λ3 − 12λ + 16.
Solving λ3 − 12λ + 16 = 0 gives λ = 2, 2, −4 as the eigenvalues.
For λ = −4,     
3+4 −1 1 x1 0
consider (A − (−4)I)X = 0,  7 −5 + 4 1   x2  =  0  ,
     6 −6 2+4 x3 0
7 −1 1 x1 0
 7 −1 1   x2  =  0  , R2 − R1 , R3 − 6 R1 gives
7

 6 −6 6 x 3 0
  
7 −1 1 x1 0
 0 0 0   x2  =  0  .
0 − 367
36
7
x3 0

9
This implies that 7x1 − x2 + x3 = 0 and x3 = x2 substituting in the first
equation gives x1 = 0.
Therefore X = (x1 , x2 , x3 ) = (0, x2 , x2 ) = x2 (0, 1, 1).
Hence X1 = (0, 1, 1) is an eigenvector corresponding to λ = −4
For λ = 2,     
3−2 −1 1 x1 0
consider (A − 2I)X = 0,  7 −5 − 2 1   x2  =  0  ,
     6 −6 2−2 x3 0
1 −1 1 x1 0
 7 −7 1   x2  =  0  , R2 − 7R1 , R3 − 6R1 gives
 6 −6 0  x3  0 
1 −1 1 x1 0
 0 0 0   x2 = 0  .
 
0 0 −6 x3 0
This implies that x1 − x2 + x3 = 0 and x3 = 0 substituting in the first equa-
tion gives x1 = x2 .
Therefore X = (x1 , x2 , x3 ) = (x2 , x2 , 0) = x2 (1, 1, 0).
Hence X2 = (1, 1, 0) is an eigenvector corresponding to λ = 2.

 
1 0 0
Challenge 1.9. Find the eigenvalues and eigenvectors of A =  0 1 0  .
0 0 1
Solution: The characteristic polynomial is
λ − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A)
3

= λ3 − 3λ2 + (1 + 1 + 1)λ − 1
= λ3 − 3λ2 + 3λ − 3.
Solving λ3 − 3λ2 + 3λ − 3 = 0 gives λ = 1, 1, 1 as the eigenvalues.
For λ = 1,     
1−1 0 0 x1 0
consider (A − I)X = 0,  0 1−1 0   x2  =  0  ,
    0 0 1−1 x3 0
0 0 0 x1 0
 0 0 0   x2  =  0  .
0 0 0 x3 0
This means there is no condition on x1 , x2 , x3 .
Therefore X = (x1 , x2 , x3 ) = x1 (1, 0, 0) + x2 (0, 1, 0) + x3 (0, 0, 1).
Hence X1 = (1, 0, 0), X2 = (0, 1, 0) and X3 = (0, 0, 1) are eigenvectors corre-

10
sponding to λ = 1.
Note: As the rank of the coefficient matrix is 0 and the number of unknowns
are 3, therefore there are 3 − 0 = 3 eigenvectors corresponding to λ = 1,

Some facts:

ˆ Eigenvalues of Triangular(upper or lower) matrices are the diagonal

elements.

ˆ If λ1 , λ2 , · · · λn are the eigenvalues and X1 , X2 , · · · , Xn are eigenvectors

of matrix A then
[1] λk1 , λk2 , · · · λkn are the eigenvalues of Ak , with the same eigenvec-
tors X1 , X2 , · · · , Xn .
[2] λ11 , λ12 , · · · , λ1n are the eigenvalues of A−1 , if λ′i s are non-zero,
with the same eigenvectors X1 , X2 , · · · , Xn .
[3] If P (x) ∈ Z[x] is any polynomial then eigenvalues of P (A) are
p(λ1 ), p(λ2 ), · · · , p(λn ), with the same eigenvectors X1 , X2 , · · · , Xn .
|A| |A|
[4] , ,···
λ1 λ2
, |A|
λn
are the eigenvalues of Adj(A),
[5] T race(A) = λ1 + λ2 + · · · + λn . (Sum of the eigenvalues)
[6] Det(A) = λ1 λ2 · · · λn . (Product of the eigenvalues)

ˆ The Det(A) = 0 iff 0 is an eigenvalue of A.

 
−1 0 0
Challenge 1.10. If A =  2 −3 0  , then find the eigenvalues and
1 4 −2
2
eigenvectors of A + 5A − 2I.

Solution: The characteristic polynomial is

λ3 − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A)
= λ3 + 6λ2 + (6 + 2 + 3)λ + 6
= λ3 + 6λ2 + 11λ + 6.
Solving λ3 + 6λ2 + 11λ + 6 = 0 gives λ = −1, −2, −3 as the eigenvalues.

Or

11
As the given matrix is lower triangular, so the eigenvalues are the diagonal
elements i.e., λ = −1, −2, −3. are the eigenvalues of A.
For λ = −1,     
−1 + 1 0 0 x1 0
consider (A + I)X = 0,  2 −3 + 1 0   x2  =  0  ,
    1 4 −2 + 1 x3 0
0 0 0 x1 0
 2 −2 0   x2  =  0  .
1 4 −1 x3 0
This implies that x1 + 4x2 − x3 = 0 and x1 = x2 substituting in the first
equation gives x3 = 5x2 .
Therefore X = (x1 , x2 , x3 ) = (x2 , x2 , 5x2 ) = x2 (1, 1, 5).
Hence X1 = (1, 1, 5) is an eigenvector of A corresponding to λ = −1.
For λ = −2,     
−1 + 2 0 0 x1 0
consider (A + 2I)X = 0,  2 −3 + 2 0   x2 = 0  ,
 
    1 4 −2 + 2 x3 0
1 0 0 x1 0
 2 −1 0   x2  =  0  .
1 4 0 x3 0
This implies that x1 = 0 and 2x1 − x2 = 0 which gives x2 = 0 and there is
no condition on x3 .
Therefore X = (x1 , x2 , x3 ) = (0, 0, x3 ) = x3 (0, 0, 1).
Hence X2 = (0, 0, 1) is an eigenvector of A corresponding to λ = −2.
For λ = −3,     
−1 + 3 0 0 x1 0
consider (A + 3I)X = 0,  2 −3 + 3 0   x2  =  0  ,
     1 4 −2 + 3 x3 0
2 0 0 x1 0
 2 0 0   x2  =  0  .
1 4 1 x3 0
This implies that x1 + 4x2 + x3 = 0 and x1 = 0 substituting in the first
equation gives x3 = −4x2 .
Therefore X = (x1 , x2 , x3 ) = (0, x2 , −4x2 ) = x2 (0, 1, −4).
Hence X3 = (0, 1, −4) is an eigenvector of A corresponding to λ = −3.
As the eigenvalues of A are −1, −2, −3.
So the eigenvalues of A2 + 5A − 2I are
(−1)2 + 5(−1) − 2, (−2)2 + 5(−2) − 2, (−3)2 + 5(−3) − 2

12
i.e., the eigenvalues of A2 + 5A − 2I are −6, − 8, − 8.
We know that eigenvectors of A2 + 5A − 2I are same as that of A.
So eigenvectors of A2 + 5A − 2I are X1 = (1, 1, 5), X2 = (0, 0, 1) and X3 =
(0, 1, −4).
More Challenges:
Q.1. Find the eigenvalues and eigenvectors of the following matrices
 
8 −8 −2
1.  4 −3 −2  . Answer: λ = 1, 2, 3.
3 −4 1
 
2 1 1
2.  2 3 2  . Answer: λ = 1, 1, 7.
3 3 4
 
1 2 2
3.  2 1 2  . Answer: λ = −1, −1, 5.
2 2 1
 
2 3 4
4.  0 2 −1  . Answer: λ = 1, 2, 2.
0 0 1
 
−2 2 −3
5.  2 1 −6  . Ans: λ = 5, −3, −3. X1 = (1, 2, −1), X2 = (2, −1, 0), X3 =
−1 −2 0
(3, 0, 1).
 
1/3 2/3 2/3
6.  2/3 1/3 −2/3  . Ans: λ = 1, 1, −1. X1 = (1, −1, 0), X2 =
2/3 −2/3 1/3
(1, 0, −1), X3 = (1, −1, −1).

Q.2. Two of the eigenvalues of a 3 × 3 matrix are −1, 2. If the determinant

of the matrixis 4, find
its third eigenvalue.
2 4
Q.3. If A = , then find the eigenvalues and eigenvectors of A2 .
 0 3 
2 3 4
Q.4. If A =  0 4 2  , then find the eigenvalues of adj(A) and A2 −2A+I.
0 0 3

13
2 Diagonalization
2.1 Introduction
Diagonalizable matrices are of interest because diagonal matrices are espe-
cially easy to handle. In this section we will try to express the given matrix A
as A = P DP −1 where D is diagonal matrix and P is some invertible matrix.
Not all matrices are expressible in such way but the matrices which has such
expression are called diagonalizable matrix, we will define diagonalizability
formally ahead. Once we have such expression then we can easily find An as
An = (P DP −1 )(P DP −1 ) · · · (P DP −1 )
= P D(P −1 P )D(P −1 P )D · · · D(P −1 P )P −1 = P Dn P −1 . Similarly we will be
able to compute eA , tan(A) and manymore functions of A which we will dis-
cuss in detail.

14
2.2 Similarity of matrices
Definition 2.1 (Similar). Two n × n matrices A and B are called similar if
there exists an n × n invertible matrix P such that B = P −1 AP.

2 −3 1 0
For example: A = is similar to B = because
1 −2 0 −1
3 1
P = works. To check B = P −1 AP, we check that P B = AP and
1 1
3 −1
observe that P B = = AP.
1 −1
Some properties of similar matrices:
If A and B are two similar matrices then the following holds:

1. Rank(A)=Rank(B).

2. Characteristic polynomial of A is same as characteristic polynomial of

B.

3. Eigenvalues of A is same as eigenvalues of B, which implies that Det(A) =

Det(B) and T race(A) = T race(B).

1 2 3 0
Challenge 2.1. Is the matrix A = is similar to B = ?
0 3 1 2

Solution: As we know that, If A and B are similar then T race(A) =

T race(B), here T race(A) = 4 and T race(B) = 5 not same.
Therefore A and B are not similar.

−1 6 3 0
Challenge 2.2. Is the matrix A = is similar to B = ?
0 6 0 2

Solution: No, as Det(A) = −6 and Det(B) = 6 not same.

Therefore A and B are not similar.

In this section we saw some examples which tells us that there are matrices
which are similar and there are matrices which are not similar.
The next section focuses on the matrices which are similar to diagonal ma-
trices to which we call as diagonalizable matrices.

15
2.3 Diagonalizable matrices
Definition 2.2 (Diagonalizable). A square matrix A is said to be diagonal-
izable if it is similar to diagonal matrix. i.e., if there exist a diagonal matrix
D and an invertible matrix P (called modal matrix) such that A = P DP −1 .

We state and use Necessary and sufficient condition for diagonal-

izability to answer whether the given matrix is diagonalizable.

Definition 2.3 (Algebraic multiplicity). Let A be any square matrix and

let λ be an eigenvalue of A then the algebraic multiplicity of λ is denoted
as A.M (λ) and is defined as A.M (λ) = the number of times λ appears as a
root of the characteristic polynomial of A.

Definition 2.4 (Geometric multiplicity). Let A be any square matrix and

let λ be an eigenvalue of A then the geometric multiplicity of λ is denoted
as G.M (λ) and is defined as G.M (λ) = the number of linearly independent
eigenvectors associated with the eigenvalue λ.

Necessary and sufficient condition for diagonalizability: A square

matrix A is diagonalizable if and only if A.M (λ) = G.M (λ) for all the eigen-
values λ of A.

Some facts:

1. Every diagonal matrix A is diagonalizable. (since P = I and D = A

works.)

2. Eigenvectors corresponding to distinct eigenvalues are linearly indepen-

dent.

3. Necessary and sufficient condition and above point ensures that there
are n linearly independent eigenvectors for a diagonalizable matrix.

4. The matrix D is obtained by placing the eigenvalues of matrix A along

the diagonals and P is obtained by placing the eigenvectors as columns
in the same order as that of the eigenvalues placed in D.

5. A.M (λ) ≥ G.M (λ) ≥ 1, for any eigenvalue. λ.

6. If A.M (λ) = 1 then by above point G.M (λ) = 1.

16
7. Suppose matrix A has distinct eigenvalues i.e., A.M (λ) = 1 for all the
eigenvalues of A then G.M (λ) = 1 for all the eigenvalues of A by above
point. This implies that A.M (λ) = G.M (λ) = 1, for all the eigenvalues
of A. So we can state the following
If a matrix A has distinct eigenvalues then it is diagonalizable,
converse of this statement is not true.

We are now ready to face the challenges related to the diagonalizablity.

 
−2 2 −3
Challenge 2.3. Check whether the matrix A =  2 1 −6  is diago-
−1 −2 0
nalizable? If yes, find the diagonal matrix and modal matrix.

Solution: The characteristic polynomial is

λ3 − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A)
= λ3 − (−1)λ2 + ((−12) + (−3) + (−6))λ − 45
= λ3 + λ2 − 21λ − 45.
Solving λ3 + λ2 − 21λ − 45 = 0 gives λ = −3, −3, 5 as the eigenvalues.
This implies that A.M (5) = 1 and A.M (−3) = 2.
For λ = −3,     
−2 + 3 2 −3 x1 0
consider (A + 3I)X = 0,  2 1 + 3 −6   x2  =  0  ,
   −1  −2 0 + 3 x3 0
1 2 −3 x1 0
 2 4 −6   x2  =  0  , R2 − 2R1 , R3 + R1 gives
 −1 −2 3  x3  0
1 2 −3 x1 0
 0 0 0   x2  =  0  .
0 0 0 x3 0
This implies that x1 + 2x2 − 3x3 = 0 i.e., x1 = 3x3 − 2x2 .
Therefore X = (x1 , x2 , x3 ) = (3x3 − 2x2 , x2 , x3 ) = x2 (−2, 1, 0) + x3 (3, 0, 1).
Hence X1 = (−2, 1, 0) and X2 = (3, 0, 1) are eigenvectors corresponding to
λ = −3.
By construction X1 and X2 are linearly independent.
 λ = 5, consider (A −5I)X
For  =0,  
−2 − 5 2 −3 x1 0
 2 1 − 5 −6   x2 = 0  ,
 
−1 −2 0 − 5 x3 0

17
    
−7 2 −3 x1 0
 2 −4 −6   x2  =  0  , R1 − 7R3 , R2 + 2R3 gives
 −1 −2 −5  x3  0 
0 16 32 x1 0
 0 −8 −16   x2  =  0  , R1 − 2R2 gives
 −1 −2 −5   x3   0 
0 0 0 x1 0
 0 −8 −16   x2  =  0  .
−1 −2 −5 x3 0
This implies that x2 = −2x3 and −x1 − 2x2 − 5x3 = 0 substituting first
equation in second equation gives x1 = −x3 . Therefore X = (x1 , x2 , x3 ) =
(−x3 , −2x3 , x3 ) = x3 (−1, −2, 1).
Hence X3 = (−1, −2, 1) is an eigenvector corresponding to λ = 5.
We obtained 2 linearly independent eigenvectors corresponding to λ = −3
and 1 eigenvector corresponding to λ = 5,
therefore G.M (−3) = 2 and G.M (5) = 1.
Henceby necessary and  sufficient
 condition, A is diagonalizable.
5 0 0 −1 −2 3
D =  0 −3 0  , P =  −2 1 0  are the required diagonal and
0 0 −3 1 0 1
modal matrix respectively.

Observation: Matrix P and D are not unique. since we could have

placed the eigenvalues in the construction of matrix D in a different way,
which will change P as well.
 
8 −8 −2
Challenge 2.4. Check whether the matrix A =  4 −3 −2  is diago-
3 −4 1
nalizable? If yes, find the diagonal matrix and modal matrix.
Solution: The characteristic polynomial is
λ − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A)
3

= λ3 − (6)λ2 + ((−11) + 14 + 8)λ − 6

= λ3 − 6λ2 + 11λ − 6.
Solving λ3 − 6λ2 + 11λ − 6 = 0 gives λ = 1, 2, 3 as the eigenvalues.
As the eigenvalues are distinct, so A is diagonalizable.
Now to find modal matrix we need eigenvectors corresponding to λ = 1, 2, 3.
For λ = 1,

18
    
8−1 −8 −2 x1 0
consider (A − I)X = 0,  4 −3 − 1 −2   x2 = 0  ,
 
    3  −4 1−1 x3 0
7 −8 −2 x1 0
 4 −4 −2   x2  =  0  , R1 − R2 , R2 − 4 R3 gives
3

 3 −4 0  x 3
   0
3 −4 0 x1 0
 0 4 −2   x2  =  0  , R1 − R3 gives
3

 3 −4 0   x3   0 
0 0 0 x1 0
 0 4 −2   x2  =  0  .
3
3 −4 0 x3 0
4
This implies that 3 x2 = −2x3 i.e., 2x2 = 3x3 and 3x1 = 4x2 ,
if we substitute x3 = 2t then x2 = 3t and x1 = 4t,
then X = (x1 , x2 , x3 ) = (4t, 3t, 2t) = t(4, 3, 2).
Hence X1 = (4, 3, 2) is an eigenvector corresponding to λ = 1.
For λ = 2,     
8−2 −8 −2 x1 0
consider (A − 2I)X = 0,  4 −3 − 2 −2   x2  =  0  ,
    3 −4 1−2 x3 0
6 −8 −2 x1 0
 4 −5 −2   x2  =  0  , R1 − 2R3 , R2 − 4 R3 gives
3

 3 −4 −1  x 3
   0
0 0 0 x1 0
 0 1 − 2   x2  =  0  .
3 3
3 −4 −1 x3 0
This implies that 3x1 − 4x2 − x3 = 0 and x2 = 2x3 substituting in the first
equation gives x1 = 3x3 .
Therefore X = (x1 , x2 , x3 ) = (3x3 , 2x3 , x3 ) = x3 (3, 2, 1).
Hence X1 = (3, 2, 1) is an eigenvector corresponding to λ = 2.
For λ = 3,     
8−3 −8 −2 x1 0
consider (A − 3I)X = 0,  4 −3 − 3 −2   x2 = 0  ,
 
    3 −4 1−3 x3 0
5 −8 −2 x1 0
 4 −6 −2   x2  =  0  , R1 − 2R3 , R2 − 4 R3 gives
3
3 −4 −2 x3 0

19
    
5 −8 −2 x1 0
 0 2 − 2   x2  =  0  .
5 5
0 45 − 54 x3 0
This implies that 5x1 − 8x2 − 2x3 = 0 and x2 = x3 substituting in the first
equation gives x1 = 2x3 .
Therefore X = (x1 , x2 , x3 ) = (2x3 , x3 , x3 ) = x3 (2, 1, 1).
HenceX1 = (2, 1,1) is aneigenvector  corresponding to λ = 3.
1 0 0 4 3 2
D =  0 2 0  , P =  3 2 1  are the required diagonal and modal
0 0 3 2 1 1
matrix respectively.
By referring to the problems solved in previous section we observe that

1. Challenge 1.5 and Challenge 1.6 are diagonalizable as we have dis-

tinct eigenvalues.

2. Challenge 1.7 is diagonalizable as A.M (−1) = 2 = G.M (−1) and

A.M (8) = 1 = G.M (8).

3. Challenge 1.8 is not diagonalizable as A.M (2) = 2 but G.M (2) = 1.

4. Challenge 1.9 is diagonalizable as A.M (1) = 3 = G.M (1). Infact we

know all diagonal matrices are diagonalizable.

More challenges:
Check whether the following matrices are diagonalizable? If it is, then find
the diagonal matrix and modal matrix.
     
4 2 −2 1 0 0 2 1 0
1.  −5 3 2  Ans: D =  0 2 0  , P =  1 1 1  .
−2 4 1 0 0 5 4 2 1
 
2 1 0
2.  0 2 1  Ans: λ = 2, 2, 2. Here A.M (2) = 3 but G.M (2) = 1.
0 0 2
Hence the given matrix is not diagonalizable.
 
−17 18 −6
3.  −18 19 −6  Ans: λ = 2, 1, 1. Here A.M (1) = 2 but G.M (1) = 1
−9 9 2
Hence the given matrix is not diagonalizable.

20
 
2 −1 1
4.  2 2 −1  Ans: λ = 1, 1, 1. Here A.M (1) = 3 but G.M (1) = 1.
1 2 −1
Hence the given matrix is not diagonalizable.

21
3 Cayley Hamilton theorem
3.1 Introduction
We turn now to a remarkable theorem, named after Arthur Cayley and Sir
William Rowan Hamilton. This deals with the polynomials of the square
matrices. The theorem helps us to find A−1 if exists and also gives answer
for the higher degree polynomials of the matrix.

3.2 Theorem
Theorem 3.1 (Cayley-Hamilton). Every square matrix satisfies its own
characteristic polynomial, i.e., If p(λ) is characteristic polynomial of a square
matrix A then p(A) = 0.
 
7 2 −2
Challenge 3.1. Verify Cayley-Hamilton theorem for A =  −6 −1 2 
6 2 −1
−1
and hence find A if exists.
Solution: The characteristic polynomial of A is given by
λ − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A)
3

= λ3 − 5λ2 + ((−3) + 5 + 5)λ − 3

= λ3 − 5λ2 + 7λ − 3.
Therefore p(λ) = λ3 − 5λ2 + 7λ − 3.
To verify Cayley-Hamilton theorem we need to check whether P (A) = 0,
i.e., To check A3 − 5A2 + 7A − 3I = 0.
For this
 we need values
of A2 and A3 .   
7 2 −2 7 2 −2 25 8 −8
A2 =  −6 −1 2   −6 −1 2  =  −24 −7 8  .
6 2 −1 6  2 −1  8 −7
24 
25 8 −8 7 2 −2 79 26 −26
A3 = A2 A =  −24 −7 8   −6 −1 2  =  −78 −25 26  .
24 8 −7 6 2 −1   78 26 −25
79 26 −26 25 8 −8
Now A3 − 5A2 + 7A − 3I =  −78 −25 26  − 5  −24 −7 8  +
   78 26 −25 24 8 −7
7 2 −2 1 0 0
7  −6 −1 2  − 3  0 1 0 
6 2 −1 0 0 1

22
   
79 − 125 + 49 − 3 26 − 40 + 14 −26 + 40 − 14 0 0 0
=  −78 + 120 − 42 −25 + 35 − 7 26 − 40 + 14  =  0 0 0  .
78 − 120 + 42 26 − 40 + 14 −25 + 35 − 7 0 0 0
In the process of finding characteristic polynomial of A, we found that Det(A) =
3, so A−1 exists.
We verified that A3 − 5A2 + 7A − 3I = 0.
Multiplying both side by A−1 we get,
A−1 A3 − 5A−1 A2 + 7A−1 A − 3A−1 I = 0 i.e., A2 − 5A + 7I − 3A−1 = 0.
−1 1 2
Hence A = 3 (−A + 5A− 7I)   
25 8 −8 7 2 −2 1 0 0
= 31 −  −24 −7 8  + 5  −6 −1 2  − 7  0 1 0 
 24 8 −7 6 2 −1 0 0 1
−3 −2 2
1
=3 6 5 −2  .
−6 −2 5
 
1 2 −2
Challenge 3.2. Verify Cayley-Hamilton theorem for A =  −1 3 0 
0 −2 1
−1 4
and hence find A if exists, and A .

Solution: The characteristic polynomial of A is given by

λ3 − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A)
= λ3 − 5λ2 + (3 + 1 + 5)λ − 1
= λ3 − 5λ2 + 9λ − 3.
Therefore p(λ) = λ3 − 5λ2 + 9λ − 3.
To verify Cayley-Hamilton theorem we need to check whether A3 − 5A2 +
9A − I = 0.
For this
 we need values
of A2 and A3 .   
1 2 −2 1 2 −2 −1 12 −4
A2 =  −1 3 0   −1 3 0  =  −4 7 2 .
0  −2 1 0 −2 1 2  −8 1 
−1 12 −4 1 2 −2 −13 42 −2
A3 = A2 A =  −4 7 2   −1 3 0  =  −11 9 10 
2 −8 1  0 −2 1  10 −22 −3 

−13 42 −2 −1 12 −4
3 2
Now A − 5A + 9A − I =  −11 9 10  − 5 −4 7
 2 +
10 −22 −3 2 −8 1

23
   
1 2 −2 1 0 0
9  −1 3 0 − 0 1 0 
 0 −2 1 0 0 1   
−13 + 5 + 9 − 1 42 − 60 + 18 −2 + 20 − 18 0 0 0
=  −11 + 20 − 9 9 − 35 + 27 − 1 10 − 10  =  0 0 0 .
10 − 10 −22 + 40 − 18 −3 − 5 + 9 − 1 0 0 0
In the process of finding characteristic polynomial of A,
we found that Det(A) = 1, so A−1 exists.
We verified that A3 − 5A2 + 9A − I = 0.
Multiplying both side by A−1 we get,
A−1 A3 − 5A−1 A2 + 9A−1 A − A−1 I = 0 i.e., A2 − 5A + 9I − A−1 = 0.
−1 2
Hence A = −A +  5A −9I     
−1 12 −4 1 2 −2 1 0 0 3 2 6
= −  −4 7 2 +5  −1 3 0 −9  0 1 0  =  1 1 2  .
2 −8 1 0 −2 1 0 0 1 2 2 5
4 3 2
To find A multiply the verified equation A − 5A + 9A − I = 0 by A,
4 3 2
 A = 5A − 9A + A 
we get   
−13 42 −2 −1 12 −4 1 2 −2
= 5  −11 9 10  − 9  −4 7 2  +  −1 3 0 
 10 −22 −3  2 −8 1 0 −2 1
−55 104 24
= −20 −15 32  .

32 −42 13
 
2 1 1
Challenge 3.3. Find the characteristic equation of the matrix A =  0 1 0 
1 1 2
and hence find the matrix represented by A8 − 5A7 + 7A6 − 3A5 + A4 − 5A3 +
8A2 − 2A + I.
Solution: The characteristic polynomial of A is given by
λ3 − T race(A)λ2 + (A11 + A22 + A33 )λ − Det(A)
= λ3 − 5λ2 + (2 + 3 + 2)λ − 3
= λ3 − 5λ2 + 7λ − 3.
Therefore λ3 − 5λ2 + 7λ − 3 = 0 is the required characteristic equation.
Now dividing the polynomial λ8 − 5λ7 + 7λ6 − 3λ5 + λ4 − 5λ3 + 8λ2 − 2λ + 1
by the characteristic polynomial λ3 − 5λ2 + 7λ − 3,
we get the quotient λ5 + λ and the remainder λ2 + λ + 1.
So λ8 − 5λ7 + 7λ6 − 3λ5 + λ4 − 5λ3 + 8λ2 − 2λ + 1 = (λ3 − 5λ2 + 7λ − 3)(λ5 +

24
λ) + λ2 + λ + 1.
In terms of A we have,
A8 − 5A7 + 7A6 − 3A5 + A4 − 5A3 + 8A2 − 2A + I = (A3 − 5A2 + 7A − 3I)(A5 +
A) + A2 + A + I.
But by Cayley-Hamilton theorem A3 − 5A2 + 7A − 3I = 0.
8 7 6 5 4 3 2 2
 A − 5A
Therefore  + 7A − 3A +A − 5A + 8A − 2A + I = A + A + I.
2 1 1 2 1 1 5 4 4
2
A = 0 1 0
   0 1 0 = 0 1 0 .
 
1 1 2  1 1 2   4 4 5    
5 4 4 2 1 1 1 0 0 8 5 5
Hence A2 +A+I =  0 1 0 + 0 1 0 + 0 1 0  =  0 3 0  .
4 4 5 1 1 2 0 0 1 5 5 8

1 2
Challenge 3.4. Apply Cayley-Hamilton theorem to A = and
2 −1
deduce that A8 = 625I.
Solution: The characteristic polynomial of A is given by
2
λ − T race(A)λ + Det(A)
= λ2 − 0λ − 5 = λ2 − 5.
So by Cayley-Hamilton theorem A2 − 5I = 0 i.e., A2 = 5I.
Therefore A8 = 54 I = 625I.
 
1 2 0
Challenge 3.5. Verify Cayley-Hamilton theorem for A =  2 −1 0 
0 0 −1
−2
and hence find A if exists.
Solution: The characteristic equation is
λ + λ2 − 5λ − 5 = 0.
3
3 2 2
 whether A + A − 5A − 5I= 0For this we
To check  need A and A3 .
1 2 0 1 2 0 5 0 0
A2 =  2 −1 0   2 −1 0  =  0 5 0 
0 0 −1 0  0 −1  0 0 1 
5 0 0 1 2 0 5 10 0
A3 = A2 A =  0 5 0   2 −1 0  =  10 −5 0 
0 0 1 0 0 −1   0 0 −1  
5 10 0 5 0 0 1 2 0
A3 + A2 − 5A − 5I =  10 −5 0  +  0 5 0  − 5  2 −1 0  −
0 0 −1 0 0 1 0 0 −1

25
 
1 0 0
5 0 1 0 
0 0 1   
5+5−5−5 10 − 10 0 0 0 0
= 10 − 10 −5 + 5 + 5 − 5 0 = 0 0 0 
0 0 −1 + 1 + 5 − 5 0 0 0
Multiply the verified equation A3 + A2 − 5A − 5I = 0 by A−1 (which exists
as Det(A) = 5)
−1 −1
we get, 2
 A + A −5I − 5A = 0 i.e.,A  = 51 (A2 + A
 − 5I)  
5 0 0 1 2 0 1 0 0 1 2 0
= 15  0 5 0  +  2 −1 0  − 5  0 1 0  = 15  2 −1 0  .
0 0 1 0 0 −1   0 0 1  0 0 −5 
1 2 0 1 2 0 5 0 0
Now A−2 = (A−1 )2 = 51  2 −1 0  15  2 −1 0  = 25 1 
0 5 0 .
0 0 −5 0 0 −5 0 0 25
More challenges:  
2 −1 1
Q.1. Find the characteristic equation of the matrix A =  −1 2 −1 
1 −1 2
−1
and verify Cayley-Hamilton theorem and hence, obtain A  if exists. 
1 3 7
Q.2. Find the characteristic equation of the matrix A = 4 2 3  and

1 2 1
hence, find the matrix represented by A − 4A − 20A − 34A − 4A3 − 20A2 −
7 6 5 4

33A + I.
cos θ sin θ
Q.3. Verify Cayley-Hamilton theorem for A = and obtain
− sin θ cos θ
A−1 .
1 2
Q.4. Verify Cayley-Hamilton theorem for A = and evaluate 2A4 −
2 2
5A3 − 7A + 6I.

26
4 Functions of square matrix
4.1 Introduction
In this section, we use the concepts which were discussed in the previous
sections and apply it to find the function f (A) = An , eA , tan(A), etc.
where A is given.

4.2 Using diagonalizability

 
λ1 0 · · · 0
 0 λ2 · · · 0 
We note the following, if D =  ..  is a diagonal matrix then
 
.. . .
 . . . 0 
0 0 · · · λk
 
λn1 0 · · · 0
n
 0 λn2 · · · 0 
1. D =  .
 
.. .. . .
 . . . 0 
0 0 ··· λnk
 
e λ1 0 · · · 0
 0 e λ2 · · · 0 
2. eD =  .. .
 
.. . .
 . . . 0 
0 0 ··· e λn
 
tan(λ1 ) 0 ··· 0
 0 tan(λ2 ) ··· 0 
3. tan(D) = 
 
.. .. ... 
 . . 0 
0 0 ··· tan(λn )
We know that if matrix A is diagonalizable then there exists an invertible
matrix P and an diagonal matrix D such that P −1 AP = D i.e., A = P DP −1 .
So
1. An = (P DP −1 )(P DP −1 ) · · · (P DP −1 ) = P Dn P −1 .
2. eA = I + A + 2!1 A2 + 3!1 A3 + · · · = P P −1 + P DP −1 + 1
2!
P D2 P −1 +
1
3!
P D3 P −1 + · · ·
= P (I + D + 2!1 D2 + 3!1 D3 + · · · )P −1 = P eD P −1 .

27
3. cA = P cD P −1 , where c is constant.

4. sin A = P (sin D)P −1 .

5. cos A = P (cos D)P −1 .

6. tan A = P (tan D)P −1 .

So we keep f (A) = P f (D)P −1 in mind.

2 1
Challenge 4.1. If A = , find A50 .
1 2
Solution: The characteristic equation of A is
λ2 − T race(A)λ + Det(A)
= λ2 − 4λ + 3.
Solving λ2 − 4λ + 3 = 0 gives λ = 1, 3 as the eigenvalues.
For λ = 1,
2−1 1 x1 0
consider (A − I)X = 0 i.e., = .
1 2−1 x2 0
This imples that x1 = −x2 .
Therefore X = (x1 , x2 ) = (−x2 , x2 ) = x2 (−1, 1).
Hence X1 = (−1, 1) is an eigenvector corresponding to λ = 1.
For λ = 3,
2−3 1 x1 0
consider (A − 3I)X = 0 i.e., = .
1 2−3 x2 0
This imples that x1 = x2 .
Therefore X = (x1 , x2 ) = (x2 , x2 ) = x2 (1, 1).
Hence X2 = (1, 1) is an eigenvector corresponding to λ =3.
1 0 −1 1 −1 1
So D = ,P = and P −1 = 21 .
0 3 1 1 1 1
Therefore
50 50 −1 −1 1 1 0 1 −1 1
A = PD P =
1 1 0 350 2 1 1
50 50
1+3 −1 + 3
= 12 50 .
−1 + 3 1 + 350

28
3 1

Challenge 4.2. If A = 2
1
2
3 , find eA and 4A .
2 2

Solution: The characteristic equation of A is

2
λ − T race(A)λ + Det(A)
= λ2 − 3λ + 2.
Solving λ2 − 3λ + 2 = 0 gives λ = 1, 2 as the eigenvalues.
For λ = 1, 3 1

−1 x1 0
consider (A − I)X = 0 i.e., 2
1 3
2 = .
2 2
−1 x2 0
This imples that x1 = −x2 .
Therefore X = (x1 , x2 ) = (−x2 , x2 ) = x2 (−1, 1).
Hence X1 = (−1, 1) is an eigenvector corresponding to λ = 1.
For λ = 2, 3 1

− 2 x 1 0
consider (A − 2I)X = 0 i.e., 2 1 3
2 = .
2 2
− 2 x 2 0
This imples that x1 = x2 .
Therefore X = (x1 , x2 ) = (x2 , x2 ) = x2 (1, 1).
Hence X2 = (1, 1) is an eigenvector
corresponding to λ =2.
1 0 −1 1 −1 1
So D = ,P = and P −1 = 12 .
0 2 1 1 1 1
Therefore 1
A D −1 −1 1 e 0 1 −1 1
e = Pe P =
1 1 0 e2 2 1 1
2 2
e+e −e + e
= 12 .
−e + e2 e + e2 1
A D −1 −1 1 4 0 1 −1 1
4 = P4 P = 2 2
1 1 0 4 1 1
20 12 10 6
= 12 = .
12 20 6 10

−1 4
Challenge 4.3. If A = then prove that 3 tan(A) = (tan 3)A.
2 1
Solution: The characteristic equation is
2
λ − 0λ − 9 = 0 which implies λ = 3, −3.
For λ = 3,
−1 − 3 4 x1 0
consider (A − 3I)X = 0 i.e., = .
2 1−3 x2 0
This imples that x1 = x2 .

29
Therefore X = (x1 , x2 ) = (x2 , x2 ) = x2 (1, 1).
Hence X1 = (1, 1) is an eigenvector corresponding to λ = 3.
For λ = −3,
−1 + 3 4 x1 0
consider (A + 3I)X = 0 i.e., = .
2 1+3 x2 0
This imples that x1 = −2x2 .
Therefore X = (x1 , x2 ) = (−2x2 , x2 ) = x2 (−2, 1).
Hence X2 = (−2, 1) is an eigenvector corresponding to λ =−3.
3 0 1 −2 1 2
So D = ,P = and P −1 = 13 .
0 −3 1 1 −1 1
−1
Therefore
tan(A)
= P tan(D)P
1 −2 tan 3 0 1 1 2
=
1 1 0 tan(−3)

3 −1 1
− tan 3 4 tan 3 −1 4
= 13 = tan3 3 .
2 tan 3 tan 3 2 1
Hence 3 tan(A) = (tan 3)A.

4.3 Using Cayley-Hamilton theorem

Previous method can be used only if the given matrix is diagonalizable, so
that method has limitation.
Cayley-Hamilton theorem can be used even if the matrix is not diagonaliz-
able.
By dividing f (λ) = λn , eλ , tan(λ), etc by the characteristic polynomial
ch(λ) of A, we get some quotient q(λ) and the remainder r(λ).
Then f (λ) = ch(λ)q(λ)+r(λ) writing this in terms of A and applying Cayley-
Hamilton theorem we get, f (A) = r(A) (since ch(A) = 0.)
Now we know that either r(λ) = 0 or degree(r(λ)) < degree(Ch(λ)).
Therefore f (A) = a2 A2 +a1 A+a0 I if A is of order 3×3 and f (A) = a1 A+a0 I
if A is of order 2 × 2.
 
1 2 −2
Challenge 4.4. If A =  0 2 1  , find A100 .
0 0 −1
Solution: The characteristic equation is
λ3 − 2λ2 + ((−2) + (−1) + 2)λ − (−2) = 0
i.e., λ3 − 2λ2 − λ + 2 = 0 which gives λ = 1, 2, −1 as the eigenvalues.

30
(Note that the given matrix is upper triangular, so eigenvalues are the diag-
onal elements.)
Here f (A) = A100 and A is of order 3 × 3.
So let A100 = a2 A2 + a1 A + a0 I, in terms of λ the equation becomes
λ100 = a2 λ2 + a1 λ + a0 .
The above equality is satisfied by the eigenvalues of A.
Putting λ = 1, 2, −1 we get the equations

a2 + a1 + a0 = 1 (3)

4a2 + 2a1 + a0 = 2100 (4)

a2 − a1 + a0 = 1 (5)
Adding equation 3 and equation 5 we get

a2 + a0 = 1 (6)

Subtracting equation 3 and equation 5 we get

a1 = 0 (7)

On resubstituting, we get a2 + a0 = 1 and 4a2 + a0 = 2100 .

100 100
On solving, we get a2 = 2 3−1 and a0 = 2 3−4 . 
2.2100 −4 4(2100 −1) 2(2100 −1)
100 −1 100 −4
3 3 3
Therefore A100 = ( 2 )A2 + ( 2 )I =  5.2100 −8 2100 −1
0 .
 
3 3 3 3
2.2100 −4
0 0 3

2 3 50 −149 −150
Challenge 4.5. If A = , prove that A = .
−3 −4 150 151

Solution: The characteristic equation is λ2 + 2λ + 1 = 0 which implies

λ = −1, −1.
Here f (A) = A50 and A is of order 2 × 2.
So let A50 = a1 A + a0 I, in terms of λ the equation becomes
λ50 = a1 λ + a0 .
The above equality is satisfied by the eigenvalues of A.
Putting λ = −1, we get
1 = −a1 + a0 (8)

31
Since the eigenvalue is repeated, so the eigenvalue will also satisfy the deriva-
tive of λ50 = a1 λ + a0 i.e., 50λ49 = a1
Putting λ = −1, we get
a1 = −50 (9)
Solving both the equation gives a0 = −49 and a1 = −50.
50 −100 −150 49 0
∴ A = a1 A + a0 I = −50A − 49I = − =
150 200 0 49
−149 −150
.
150 151

π π4

Challenge 4.6. If A = , find cos A.
0 π2
Solution: The given matrix is upper triangular, so the eigenvalues of the
matrix A are the diagonal elements, i.e., λ = π, π2 .
Let cos A = a1 A + a0 I in terms of λ the equation becomes
cos λ = a1 λ + a0 .
The above equation is satisfied by the eigenvalues of A.
Putting λ = π, we get
a1 π + a0 = −1 (10)
Putting λ = π2 , we get
a1 π
+ a0 = 0 (11)
2
On solving, we get a1 = − π2 and a0 = 1.
π π4 −1 − 12

2 2 1 0
Therefore cos A = − π A + I = − π + =
0 π2 0 1 0 0
 
1 0 0
Challenge 4.7. If A =  1 0 1  , find A50 .
0 1 0
Solution: The characteristic equation is
λ − λ2 + ((−1) + 0 + 0)λ − (−1) = 0 i.e., λ3 − λ2 − λ + 1 = 0
3

which gives λ = 1, 1, −1 as the eigenvalues.

Now as A is of order 3 × 3.
Let A50 = a2 A2 + a1 A + a0 I in terms of λ the equation becomes
λ50 = a2 λ2 + a1 λ + a0 .

32
The above equality is satisfied by the eigenvalues of A. Putting λ = −1, we
get
a2 − a1 + a0 = 1 (12)
Putting λ = 1, we get
a2 + a1 + a0 = 1 (13)
Now as the eigen values λ = 1 is repeated, so λ = 1 will satify the derivative
of λ50 = a2 λ2 + a1 λ + a0 = 0 i.e., 50λ49 = 2a2 λ + a1 . Putting λ = 1, we get

50 = 2a2 + a1 (14)

Solving these equations, we get a2= 25, a1 =0and a0 =−24  

1 0 0 1 0 0 1 0 0
50 2
Therefore A = 25A − 24I = 25 1 0 1    1 0 1  − 24  0 1 0 
    0 1 0 0 1 0 0 0 1
1 0 0 1 0 0 1 0 0
= 25  1 1 0  − 24  0 1 0  =  25 1 0  .
1 0 1 0 0 1 25 0 1
More Challenges:

2 3
Q.1. If A = , find A100 .
−3 −4
π/2 π
Q.2. If A = , find sin A.
0 3π/2
3 1
Q.3. If A = , find eA and 5A .
1 3

33
References
[1] Linear Algebra, Theory and application, Ward Cheney and David Kin-
caid.

[2] Schaum’s outlines Linear Algebra, Fourth edition, Seymour Lipschutz

and Marc Lipson.

34