LECTURE-9

VED V. DATAR∗

In this lecture, we give three further applications of the Cauchy integral

formula.

The fundamental Theorem of algebra

Recall that in the previous lecture we proved Liouville’s theorem, namely
that there are no bounded, non-constant, entire functions. As a simple and
beautiful consequence of this, we can prove the fundamental theorem of
algebra, namely that any polynomial can be completely factored into linear
factors over complex numbers.
Theorem 1. Any non-constant polynomial p(z) has a complex root, that is
there exists an α ∈ C such that p(α) = 0. As a consequence, any polynomial
is completely factorable, that is we can find α1 , · · · , αn (some of them might
be equal to each other), and c ∈ C such that
p(z) = c(z − α1 ) · · · (z − αn ),
where n = deg(p(z)).
Proof. For concreteness, let
p(z) = an z n + · · · + a1 z + a0 ,
where an 6= 0. Then we claim that there exists an R such that
|an | n
|z| ≤ |p(z)| ≤ 2|an ||z|n ,
2
whenever |z| > R. To see this, by the triangle inequality,
 |a
n−1 | |a0 |  3
|p(z)| ≤ |an ||z|n +· · · |a1 ||z| = |a0 | = |an ||z|n 1+ +· · ·+ n ≤ |an ||z|n
|z| |z| 2
if |z| > R and R is sufficiently big. For the other inequality, we use the other
side of the triangle inequality. That is,
an−1 a0
|p(z)| ≥ |an ||z|n 1 − + ··· + n .
z z
But
an−1 a0 |an−1 | |a0 | 1
+ ··· + n ≤ + ··· + n < ,
z z |z| |z| 2

Date: 24 August 2016.

1
if |z| > R when R is as above. So when |z| > R,
|an ||z|n
|p(z)| ≥ .
2
Now suppose, that p(z) has no root in C. Then
1
f (z) =
p(z)
will be an entire function. Moreover, on |z| > R, by the lower bound above,
|p(z)| > |an ||z|n /2 > |an |Rn /2,
that is |f (z)| ≤ M for some M on |z| > R. On the other hand, we claim
that there exists an ε > 0 such that |p(z)| > ε on |z| ≤ R. If not, then
there is a sequence of points zk ∈ DR (0) such that |p(zk )| → 0. Since DR (0)
is compact, there exists a subsequence, which we continue to denote by zk ,
such that zk → a ∈ DR (0). But then by continuity, p(a) = 0, contradicting
our assumption that there is no root. This proves the claim. The upshot
is that on |z| ≤ R, |f (z)| ≤ 1/ε. This shows that f (z) is a bounded, entire
function, and hence by Liouville, must be a constant, which in turn implies
that p(z) must be a constant. This proves the first part of the theorem.
The second part follows from induction. If p(z) is a non-constant polyno-
mial, let α1 be a root, which is guaranteed to exist by the first part. Then
by the remainder theorem, p(z) is divisible by (z − α1 ). For a proof of the
remainder theorem, see remark below. Then we define a new polynomial
p(z)
p1 (z) = .
z − α1
The crucial observation is that deg(p1 ) is strictly smaller than deg (p). In
finitely many steps we should reach a linear polynomial which is obviously
factorable. 
Remark 1. We used the remainder theorem for the proof of the second part
of the theorem which basically says that for a polynomial p(z),
p(z) − p(a) = (z − a)q(z)
for some polynomial q(z) of a strictly smaller degree. To see this, let
p(z) = an z n + · · · + a1 z + a0 ,
and so
p(z) − p(a) = an (z n − an ) + · · · + a1 (z − a).
Each term is of the form
ak (z k − ak ) = ak (z − a)(z k−1 + z k−2 a + · · · + zak−2 + ak−1 ),
and this completes the proof. It is also easy to see that highest degree term in
q(z) is an z n−1 , and hence in particular, the degree of q(z) is strictly smaller.
2
 Zeroes of a holomorphic function
A complex number a is called a zero of a holomorphic function f : Ω → C
if f (a) = 0. A basic fact is that zeroes of holomorphic functions are isolated.
This follows from the following theorem.
Theorem 2. Let f : Ω → C be a holomorphic function that is not identically
zero, and let a ∈ Ω be a zero of f . Then there exists a disc D around a,
a non vanishing holomorphic function g : D → C (that is, g(z) 6= 0 for all
z ∈ D), and a unique positive integer n such that
f (z) = (z − a)n g(z).
Moreover, we have that
n = min{ν ∈ N | f (ν) (a) 6= 0}.
The positive integer n is called the order or multiplicity of the zero at a.
Proof. By the principle of analytic continuation, if f is not identically zero,
there exists an n such that f (k) (a) = 0 for k = 0, 1, · · · , n−1 but f (n) (a) 6= 0.
Let Dε (a) be a disc such that Dε (a) ⊂ Ω. Then f has a power series
expansion in the disc centered at z = a with the first n terms vanishing. So
we write
f (z) = an (z − a)n + an+1 (z − a)n+1 + · · ·
with an 6= 0. The the theorem is proved with
g(z) = an + an+1 (z − a) + · · · .

As an immediate corollary to the theorem we have the following.
Corollary 1. Let f : Ω → C holomorphic.
(1) The set of zeroes of f is isolated. That is, for every zero a, there
exists a small disc Dε (a) centered at a such that f (z) 6= 0 for all
z ∈ Dε (a) \ {a}.
(2) (strong principle of analytic continuation) If Ω is connected, and
f, g : Ω → C are holomorphic functions such that for some sequence
of points pn ∈ Ω, f (pn ) = g(pn ). Then either f ≡ g or {pn } does
not have a limit point in Ω.
Proof. By the theorem, if a ∈ Ω is a root, then there exists a disc D around
a, and integer m, and a holomorphic function g : D → C such that g(a) 6= 0
and
f (z) = (z − a)m g(z).
Since g(a) 6= 0, by continuity, there is a small disc Dε (a) ⊂ D such that
g(z) 6= 0 for any z ∈ Dε (a). But then on this disc (z − a) is also not zero
anywhere except at a, and hence for any z ∈ Dε (a) \ {a}, f (z) 6= 0 exactly
what we wished to prove. Part (2) is a trivial consequence of the Lemma. 
3
Example 1. Even though the zeroes are isolated, they could converge to the
boundary. For instance, consider the holomorphic function
π 
f (z) = sin
z
on C∗ . Clearly z = 1/n is a sequence of zeroes. They are isolated, but
converge to z = 0 which is not in the domain.
Let f : Ω → C be a holomorphic function. Then by Corollary 1 any disc
D such that D ⊂ Ω contains only finitely many zeroes of the function in the
interior. The next proposition allows one to calculate the number of zeroes
counted with multiplicity.
Corollary 2. Let f : Ω → C be holomorphic, and let D ⊂ Ω be a disc such
that D ⊂ Ω, and C := ∂D contains no zeroes of f . Let p1 . · · · , pk are the
zeroes of f in D with multiplicity n1 . · · · , nk .
(1) There exists a no-where vanishing holomorphic function g : D → C
such that
f (z) = (z − p1 )n1 · · · (z − pk )nk g(z).
(2) The total number of roots (counted with multiplicity) is given by
k
f 0 (z)
Z X
dz = nj .
C f (z)
j=1

Proof. (1) For each j ∈ {1, · · · , k}, there exist a radius rj > 0 and a
nowhere vanishing holomorphic function gj : Drj (pj ) → C such that
for all z ∈ Drj (pj ),
f (z) = (z − pj )nj gj (z).
We can choose rj small enough  so that the discs
 have mutually dis-
k
joint closures. Let U := D \ ∪j=1 Drj /2 (pj ) . Define the function

gj (z)
Πi6=j (z−pi )ni , z ∈ Drj (pj ) where j = 1, · · · , k

g(z) :=
 f (z) nj , z ∈ U.
Πj (z−pj )

Clearly g(z) satisfies all the required properties.

(2) With g(z) as above, for any z 6= pj , a simple computation shows
that
k
f 0 (z) X nj g 0 (z)
= + .
f (z) z − pj g(z)
j=1

Since g(z) is nowhere vanishing, g 0 (z)/g(z) is holomorphic on D,

and hence by Cauchy’s theorem it’s integral on C vanishes. It then
4
 follows that
k
f 0 (z)
Z Z
X dz X
dz = nj = nj .
C f (z) C z − pj
j=1 j

Sequences of holomorphic functions

Recall that if a sequence of differentiable functions fn : I → R converges
uniformly to f : I → R, for some interval I ⊂ R, it does not necessarily
imply that f is also differentiable. For instance, consider fn : [−1, 1] → R
defined by r
1
fn (x) = x2 + .
n
Then it is not difficult to see that fn → |x| uniformly, but of course |x| is not
differentiable at x = 0. It turns out that holomorphic function behave much
better under uniform convergence. We say that a sequence of holomorphic
functions fn : Ω → C converges to f : Ω → C compactly if the convergence
is uniform over compact subsets K ⊂ Ω. That is, for all ε > 0 and K ⊂ Ω
compact, there exists a N = N (ε, K) such that
|fn (x) − f (x)| < ε,
for all n > N and all x ∈ K.
Theorem 3. If {fn }∞ n=1 is a sequence of holomorphic functions on Ω that
converge compactly to f : Ω → C, then f (z) is holomorphic. Moreover
fn(k) → f (k)
compactly on Ω for all k ∈ N.
Proof. Fix a a ∈ Ω, and let D be a disc around a such that it’s closure is
also in Ω. Then for any triangle T ⊂ D, by Goursat’s theorem
Z
fn (z) dz = 0.
T

Since D is compact, fn → f converges uniformly on D, and hence

Z Z
f (z) dz = lim fn (z) dz = 0.
T n→∞ T

But this is true for all triangles in D, and hence by Morera’s theorem f is
holomorphic in D and in particular at a. Since a is arbitrary, this shows
that f is holomorphic on Ω.
We prove that fn0 → f 0 compactly. For k > 1, the result will follow from
induction. There is no loss of generality in assuming that Ω ⊂ C is compact.
First we define
Ωr = {z ∈ Ω | Dr (z) ⊂ Ω}.
5
Geometrically, this is the set of all points in Ω that are at least a distance
r away from the boundary of Ω. Given any compact set K, there exists
a r > 0 such that K ⊂ Ωr , and hence it suffices to show that fn0 → f 0
uniformly on Ωr . The key point is the following estimate.
Claim 1. Let F : Ω → C be holomorphic. Then for any r > 0,
2
sup |F 0 (ζ)| ≤ sup |F (ζ)|.
z∈Ωr r ζ∈Ωr/2

First notice that if z ∈ Ωr then Dr/2 (z) ⊂ Ωr . To see this, let w ∈ Dr/2 (z)
i.e. |z − w| ≤ r/2, and we need to show that w ∈ Ωr/2 or equivalently that
Dr/2 (w) ⊂ Ω. But for any w0 ∈ Dr/2 (w), |w0 − w| ≤ r/2 and hence by
the triangle inequality |w0 − z| ≤ r, that is w0 ∈ Dr (z) which is contained
in Ω by definition, since z ∈ Ωr . This shows that Dr/2 (w) ⊂ Ω and hence
that Dr/2 (z) ⊂ Ωr/2 . Now by Cauchy’s integral formula, if we denote the
boundary of Dr (z) by Cr (z), then for any z ∈ Ω2r ,
Z
0 1 F (ζ)
F (z) = dζ.
2πi Cr/2 (z) (ζ − z)2
By the above observation, if z ∈ Ωr then Cr/2 (z) ⊂ Ωr/2 , and hence by
triangle inequality, for all z ∈ Ωr , since |ζ − z| = r/2 for ζ ∈ Cr/2 (z) we have
1
|F 0 (z)| ≤ sup |F (ζ)|len(Cr/2 (z))
2πr2 ζ∈Cr/2 (z)
2
≤ sup |F (ζ)|
r ζ∈Ωr
This proves the claim. Now given any ε > 0, since Ωr/2 ⊂ Ω is compact (re-
member we are assuming without any loss of generality that Ω is bounded),
there exists an N = N (r, ε) such that for all n > N and all ζ ∈ Ωr/2 ,
εr
|fn (ζ) − f (ζ)| < .
2
But then by the claim, for n > N and for all z ∈ Ωr , we have the estimate
2 εr
|f (z) − fn (z)| ≤ · = ε,
r 2
proving that fn → f uniformly on Ωr , and this completes the proof of the
theorem.

Recall that Weiestrass’ theorem states that any continuous function on
a compact interval is the uniform limit of polynomials. On the other, by
the above theorem, a continuous non-holomorphic function cannot be the
uniform limit of polynomials. Instead we have the following, which we state
without a proof.
6
Theorem 4 (Runge’s thoerem). Let K ⊂ C and let f be a function that is
holomorphic in a neighbourhood of K.
u.c
(1) There exists a sequence of rational functions Rn (z) such that Rn −−→
f on K, and such that the singularities of the rational functions all
lie in K c .
(2) If K c is connected, then one there exists a sequence of polynomials
u.c
pn (z) such that pn −−→ f .

∗ Department of Mathematics, Indian Institute of Science (IISc)

Email address: [email protected]