1 Chapter 2.

1 Analytic Polynomials (continued)

osition 1.1: If f and g are differentiable at z, then so is h1 = f + g and h2 = f g. Furthermore, h3 = f /g is also

differentiable at z provided that g(z) 6= 0.
0 0
It also follows that h01 (z) = f 0 (z) + g 0 (z), h02 (z) = f 0 (z)g(z) + f (z)g 0 (z) and h03 (z) = g(z)f (z)−g
[g(z)]2
(z)f (z)

Proof : We will only prove this statement for h2 , the rest is left as an exercise.

f (z + h)g(z + h) − f (z)g(z) f (z + h)g(z + h) + f (z + h)g(z) − f (z + h)g(z) − f (z)g(z)

h02 (z) = lim = lim
h→0 h h→0 h
g(z + h) − g(z) f (z + h) − f (z)
= lim f (z + h) + g(z)
h→0 h h
0 0
= f (z)g (z) + g(z)f (z)
where the last step is valid since f and g are differentiable.

2 Chapter 2.2: Power Series

ition 2.2: Limit Superior
The limit supremum (denoted by lim sup or lim) of (an ) is the defined as
Let (an ) be any real sequence.
lim sup an = lim supk≥n ak .
n→∞ n→∞

Note that this limit always exist. In case the sequence is not bounded above for every tail {ak | k ≥ n}
of terms, this will of course be an infinite limit. Furthermore, the sequence supk≥n an is non-increasing,
hence it is either ∞ or a real number.
We are interested in the following properties of lim sup. If lim sup an = L, then:
n→∞
(i) for each N and each ε > 0, there is some k > N such that ak ≥ N − ε
(ii) for each ε > 0, there is some some N ∈ N such that ak ≤ L + ε for each k ≥ N
(iii) lim sup can = cL if c is any non-negative constant.
n→∞

ple 2.3: Consider the sequences (an ) = (1, 0, 2, 0, 3, 0, 4, 0, . . . , ) and (bn ) = ((−1)n (1− n1 )) and (cn ), where cn = 1
n
if n is odd and cn = −1 otherwise.

(a) lim sup an = ∞, since supk≥n ak = ∞ for each n.

n→∞
(b) lim sup bn = 1, since supk≥n bk = 1 for each n.
n→∞
1 1
(c) lim sup cn = 0. For n odd: supk≥n bk = n and if n is even, then supk≥n bk = n+1 . Either way as
n→∞
n → ∞, both quantities approach zero.
ition 2.4: Power Series P∞
A power series in z is an infinite series of the form k=0 Ck z k , where for each k, Ck is some complex
number.
1
rem 2.5: Suppose lim sup |Ck | k = L.
n→∞
P∞
(a) If L = 0, then k=0 Ck z k converges for all z ∈ C.
P∞
(b) If L = ∞, then k=0 Ck z k converges only for z = 0.
P∞
(c) If 0 < L < ∞, then for R = 1/L, we have that k=0 Ck z k converges for z such that |z| < R and
diverges for |z| > R. (Nothing can be said about |z| = R)
1 1
Proof : Note that lim sup |Ck | k = lim supk≥n (|Ck | k )
n→∞ n→∞
1 1
(a) If L = 0, then lim sup |Ck | k = 0, then for each z we have: lim sup |Ck | k |z| = 0 as |z| ≥ 0.
n→∞ n→∞
1 1
Hence lim sup |Ck | k |z| = lim supk≥n (|Ck | k |z|) = 0. Choosing ε = 12 , we must have some N ∈ N
n→∞ n→∞
1 1
such that for each k ≥ N : supk≥n |Ck | k |z| ≤ 12 . Hence of course |Ck | k |z| ≤ 12 and consequently
|Ck ||z|k < 21k . It follows that the series converges absolutely and hence converges.

1
1 1
(b) If L = ∞, then if z 6= 0, choosing ε = |z| , we must have |Ck | k > ε for infinitely many values of k.
If this was not the case, then supk≥n |Ck |1/k wouldbe bounded and L < ∞. We then immediately
have that |Ck z k | ≥ 1 for infinitely many k and hence can not converge. That the series converges
to 0 for z = 0 is rather trivial.
(c) Suppose 0 < L < ∞ and first assume |z| < R = 1/L. Let δ = 1−|z|L 2 , then |z| = R(1 − 2δ) and
1
since lim supk≥n (|Ck | k )|z| = L|z| = 1 − 2δ. It follows that |Ck |1/k |z| < 1 − δ for sufficiently large
n→∞
k and the series is absolutely convergent.

1
On the other hand, if |z| > R, or equivalently L > |z| for infinitely many k, hence Ck z k has
absolute value greater than 1 for infinitely many values, hence the series diverges.


rk 2.6:  ∞ if L = 0
Note that R is called the radius of convergence of the series and is defined as follows: R = 0 if L = ∞
1/L if 0 < L < ∞

P∞
Furthermore, if k=0 Ck z k has radius of convergence P∞ R, then the series
P∞ converges uniformly in any disc
k k
smaller than the radius: Explicitly, ∀0 < δ < R: k=0 |C k z | ≤ k=0 |C k |R which also converges. It
follows that a power series is continuous inside a disc of radius R.

ple 2.7:
P∞
(a) The series n=0 nz n converges for |z| < 1 and diverges for |z| > 1 since lim sup |n|1/n = 1. For
n→∞
|z| = 1, we have that it diverges since the sequence an = nz n does not converge to 0.
P∞
(b) The series n=0 z n /n2 also has radius of convergence 1, but here it converges for all z with |z| = 1.
P∞ n
(c) The series n=1 zn converges for all |z| < 1 and also converges for all z with modulus 1, except at
Pk n
z = 1. It diverges for all |z| > 1. To see this, show that for each k ∈ N we have: (1 − z) n=1 zn =
Pn ∞ n
1
1 + k=2 ( n1 − n−1 )z k to conclude that 1 − z n=0 zn converges.
P
P∞ zn 1
(d) n=0 n! converges for all z since n→∞
lim n!1/n → 0.
P∞ n n n 1 n n
(e) n=0 [1+(−1) ] z has radius of convergence 2 , since lim supn→∞ [1+(−1) ] = lim sup (ak ) = 2.
n→∞ k≥n
Where does the above series converge and diverge?
P∞ n2
(f ) n=0 z has a radius of convergence 1.
P∞
(g) Any series n=0 Cn z n with Cn = ±1 for all n has radius of convergence 1.
P∞ n
P∞ n
(h) It
P can easily be nshown that the sum of two convergent power series n=0 an z + n=0 bn z =
(a
n=0 n + b n )z is convergent.
P∞
The Cauchy
Pn product, of the two abovementioned series, is defined as the series n=0 cn z n , where
cn = k=0 ak bn−k . In short, the sequence of partial sums is simply the product of the sequence of
partial sums of the two original series.