LECTURE-6

VED V. DATAR∗

Curves in the complex plane

A parametrized curve (or simply a curve) in a domain Ω ⊂ C is a contin-
uous function z(t) : [a, b] → Ω. Writing
z(t) = x(t) + iy(t),
we say that z(t) is differentiable with derivative z 0 (t), if x0 (t) and y 0 (t) exist
for all t ∈ (a, b), and then we set
z 0 (t) = x0 (t) + iy 0 (t).
We call it regular if z 0 (t) exists and z 0 (t) 6= 0 for all t ∈ (a, b). Geometrically,
the vector
x0 (t)î + y 0 (t)ĵ
gives the tangent vector to the curve at the point (x(t), y(t)), and so the
complex number z 0 (t) encodes the information of the tangent vector. More-
over
|z 0 (t)| = x0 (t)2 + y 0 (t)2 ,
p

measures the speed at which the curve is traversed.

For a parametrised curve as above, the points z(a) and z(b) are called
the initial and final points of the curve respectively, and together they are
referred to as the end points of the curve. The curve is said to be closed if
z(a) = z(b), and is called simple if z(t) is injective on the interval (a, b).
Orientation. The choice of a parametrization fixes the orientation of the
curve. Given a parametrization z(t) : [a, b] → C of the curve C, we say that
w(s) : [c, d] → C is an orientation preserving re-parametrization or that z(t)
and w(s) are equivalent parametrizations, if there exists a strictly increasing
function α : [c, d] → [a, b] with α(c) = a and α(d) = b, such that
w(s) = z(α(s)).
Example 1. Consider the curve defined by z : [0, 2π] → C where
z(t) = R(cos t + i sin t).
The image of the curve is of course a circle of radius R. The tangent vector
is given by
z 0 (t) = R(− sin t + i cos t),

Date: 24 August 2016.

1
and so the speed is |z 0 (t)| = 1. A parametrization is given by w : [0, π] → C,
w(t) = R(cos 2t + i sin 2t),
which traverses the same circle, but with double the speed. On the other
hand the parametrization
z(t) = R(cos t − i sin t)
for t ∈ (0, 2π) also describes the same circle with the same speed, but tra-
versed in a clock-wise direction. A circle is said to be positively oriented if
the parametrization traverses the circle in the anti-clockwise direction, while
negatively oriented otherwise.
We will need to consider slightly more general curves. A curve z(t) : [a, b]
is said to be piecewise regular if there is a partition
a = t0 < t1 < · · · < tn = b
such that z(t) restricted to each (ti−1 , ti ) is a regular curve.

Notation and conventions. If the image of restriction of the curve z(t)

to the interval (ti−1 , ti ) is denoted by Ci , and the image of the full curve is
denoted by C, we then write
n
X
C= Cn .
i=1

We denote by −C, the curve C traced in the opposite direction. For instance,
if z(t) : [a, b] → C is a parametrization for C, a parametrization for −C is
given by z − (s) : [a, b] → C where
z − (s) = z(a + b − s).
For a positive integer a > 0, we denote by aC to be the curve C traversed ‘a’
times. A circle CR (p) or |z − p| = R, unless otherwise specified, will always
mean a circle of radius R centred at p traversed once in the anti-clockwise
direction.

Complex integration
For a continuous function f = u + iv : [a, b] → C of one real variable, we
can extend the definition of integration by defining
Z b Z b Z b
f (t) dt := u(t) dt + i v(t) dt.
a a a
Now, suppose we are given a smooth curve as above, and a function
f : Ω → C, we then define the complex integral along the curve by
Z Z b
f (z) dz := f (z(t))z 0 (t) dt,
C a
2
where C denotes the image of the curve. Note that the multiplication above
is the complex multiplication. It is convenient to think of dz as a complex
differential, representing an infinitesimal complex change, and given by
dz = dx + idy.
So if f = u + iv, then
Z Z Z
f (z) dz = (u dx − v dy) + i (u dy + v dx),
C C C
where the integrals on the right are now the usual line integrals from mul-
tivariable calculus.
Remark 1. Recall that the differential forms dx and dy are defined to be
duals of the vector fields ∂/∂x and ∂/∂y in the xy-plane. That is,
∂  ∂  ∂  ∂ 
dx = dy = 1, dx = dy = 0.
∂x ∂x ∂y ∂x
Then it is easy to compute that dz = dx + idy and dz̄ = dx − idy are dual
complex valued differential forms to the complex valued vector fields ∂/∂z
and ∂/∂ z̄ vector fields defined in the previous lecture.
To make sure the integral is well defined, we need to show that it is
independent of orientation preserving parametrizations.
Lemma 1. Let C be a curve with parametrization z(t). Let w(s) = z(α(s))
be another orientation preserving parametrization, where α : [a, b] → [c, d].
Then Z Z
f (w) dw = f (z) dz.
C C

Proof. By the chain rule, since w0 (s) = z 0 (α(s))α0 (s), we see that
Z Z b Z b
0
f (w) dw = f (w(s))w (s) ds = f (z(α(s))z 0 (α(s))α0 (s) ds.
C a a

Putting t = α(s), we see that α0 (s)ds

= dt, and hence
Z Z b Z
f (w) dw = f (z(t))z 0 (t) dt = f (z) dz.
C a C
This shows that the definition of integration is independent of the parametriza-
tion chosen. 

We can now extend the definition of complex integrals to piecewise smooth

curves by linearity. That is, if C = C1 + · · · + Cn is a piecewise smooth curve
with Cj smooth curve for all j = 1, · · · , n, then we define
Z X n Z
f (z) dz := f (z) dz.
C j=1 Cj
3
 R
Other integrals. Similar to our definition of C f (z) dz we can define the
integral with respect to dz̄ by
Z Z
f (z, z̄) dz̄ := f (z) dz.
C C

We can also define the integral with respect to arc-length. For a complex or
real valued function f (z) and a curve z = z(t) : [a, b] → C we also define
Z Z b
f (z) |dz| := f (z)|z 0 (t)|dt.
C a

We then define the length of the curve by

Z
len(C) = |dz|.
C

We next state, without proof, some basic properties of the complex line
integral. The proofs follow from the definition of the complex integral and
corresponding properties of the Riemann integral.
Proposition 1. Let C be a parametric piecewise regular curve in an open
set Ω ⊂ C.
(1) For any complex numbers a, b and any complex valued functions f
and g we have that
Z Z Z
[af + bg](z) dz = a f (z) dz + b g(z) dz.
C C C

(2) For integers aj ∈ Z and piecewise smooth curves Cj , if we denote by

C = a1 C1 + · · · + an Cn , we have
Z Xn Z
f (z) dz = aj f (z) dz.
C j=1 Cj

In particular,
Z Z
f (z) dz = − f (z) dz.
−C C

(3) (Triangle inequality)

Z Z
f (z) dz ≤ |f (z)| |dz| ≤ sup |f (z)| · length(C),
C C z∈C

with all inequalities replaced with equality if and only if f is a con-

stant real number.
u.c
(4) If fn −−→ f , then
Z Z
lim fn (z) dz = f (z) dz.
n→∞ C C
4
 A fundamental computation
We now compute the integrals
Z
z n dz,
CR

where CR = {z | |z| = R} is the circle of radius R centered at the origin

positively oriented and traversed once. This computation, although elemen-
tary, will play a fundamental role in the rest of course. A parametrization
of the circle is given by z(t) = Reiθ , θ ∈ [0, 2π]. Then dz = iReiθ dθ, and so
Z Z 2π
n
z , dz = Rn einθ (iReiθ ) dθ
CR 0
Z 2π
n+1
= iR ei(n+1)θ dθ
0

Now if n + 1 6= 0, then
Z 2π
1 θ=2π
ei(n+1)θ dθ = ei(n+1)θ = 0,
0 (n + 1)i θ=0

since eiθ is periodic with period 2π. On the other hand, if n = −1, then
Z Z 2π Z 2π
n n+1 i(n+1)θ
z , dz = iR e dθ = i dθ = 2πi.
CR 0 0

So summarizing, we have the following:

(
0, n 6= −1
Z
1 n
z dz =
2πi |z|=R 1, n = −1.

In particular, the integral is independent of the radius R. More generally,

we have the following.

Proposition 2. Let D be any disc in C, and let p be a point not lying

on the boundary circle C = ∂D. If C is traversed only once with positive
orientation, then
(
1, p ∈ D
Z
1 1
n(C; p) := dz = .
2πi C z − p 0, p ∈
/ D.

Moreover, if n 6= −1 and n ∈ Z, then

Z
(z − p)n dz = 0.
C

The number n(C, p) is called the index or the winding number of the circle
C around p.
5
Proof. Without loss of generality, we can assume that z0 = 0. Now suppose
p ∈ D. Then for any z ∈ C, |z| > |p|. Then by the geometric series
expansion, for z ∈ C we have
∞
1 1 1 1 X pn−1
= · = + .
z−p z 1 − p/z z zn
n=2

Integrating both sides (this can be done since convergence is uniform), and
using the above computation, we see that n(C, p) = 1. On the other hand,
if p ∈
/ D, then |z| < |p| for all z ∈ C, and hence
∞
1 1 1 1 X zn
=− · =− .
z−p p 1 − z/p p pn
n=0

Again integrating both sides, we see that n(C, p) = 0, since there are only
positive powers of z on the right. For the second part, if n > 0, then
the integrand is a polynomial and hence the integral is zero by the above
computations. If n = −m < 0, then we can write
 1 m
(z − p)n = .
z−p
Once again using the geometric series expansions above, in both cases, there
will no terms with exponent −1. And hence by the computation above, the
integral will be zero. 

Remark 2. Later in the course, we will define the index n(γ, p) of a general
curve γ around a point p by a similar formula, and we shall prove (rather
indirectly) that the index of any closed curve is always an integer. Assuming
this, we can provide a more conceptual explanation of the above result. It is
clear that the n(C, p), as a function of p defined on the open set C \ C is a
continuous function. But then being integer valued, it must be locally con-
stant. From our elementary observation, n(C, 0) = 1, and hence n(C, p) = 1
for all p ∈ D. On the other hand, clearly as |p| → ∞, n(C, p) approaches
0. Again by virtue of being locally constant, this implies that n(C, p) = 0 for
all p ∈ C \ D.

Primitives
We then have the following theorem, which is a generalization of the
fundamental theorem for line integrals from multivariable calculus.

Proposition 3 (Fundamental theorem for complex integrals). If C is any

curve joining the point p to q, then
Z
F 0 (z) dz = F (q) − F (p).
C
6
Proof. Let z(t) : [0, 1] → C be a parametrization for C such that p = z(0)
and q = z(1). Then
Z Z 1
F 0 (z) = F 0 (z(t))z 0 (t) dt.
C 0

But by Chain rule, if we let g(t) = F (z(t)), then g 0 (t) = F 0 (z(t))z 0 (t), and
so
Z Z 1
dg
F 0 (z) dz = dt = g(1) − g(0),
C 0 dt
where we use the usual one variable fundamental theorem of calculus. But
g(0) = F (z(0)) = F (p) and g(1) = F (q), and this completes the proof. 

Recall that an open set is called connected is any two points can be joined
by a continuous curve lying completely inside the open set. An important
and immediate consequence of the fundamental theorem is the following.
Corollary 1. Let Ω ⊂ C be an open connected subset, and f : Ω → C be
holomorphic. Then f 0 (z) = 0 for all z ∈ Ω if and only if f (z) is a constant.
For a domain Ω ⊂ C, and F, f : Ω → C complex valued functions, we say
that F (z) is a primitive of f (z) if
F 0 (z) = f (z)
for all z ∈ Ω. Then another direct corollary of the above theorem is the
following.
Corollary 2. Suppose f : Ω → C has a primitive on F , then
Z
f (z) dz = 0
C

for every closed curve C ⊂ Ω.

Using the above corollary, we can explain the results of the calculation
above. Recall that for any integer n 6= −1, we have that
 1 d
z n+1 = z n .
n + 1 dz
Or in other words, for n 6= −1, z n has a primitive at least on C \ {0}.
Hence the integral on any closed loop not passing through z = 0, is always
zero. In particular, the integrals around |z| = R are zero. That leaves
the case when n = −1. We have already seen (as a consequence of the
chain rule) that if a holomorphic logarithm log z can be defined, then it is
a natural primitive for 1/z. But we saw some lectures back, we saw that
going around a circle centered at the origin, makes the logarithm function
discontinuous, leave alone non-holomorphic. In fact, combining the corollary
with the calculations above, we have managed to prove the following.
7
Proposition 4. Let Ω ⊂ C \ {0} be an open set containing at least one
circle |z| = r. Then there is no holomorphic function F : Ω → C such that
eF (z) = z.
In particular, there cannot be a holomorphic logarithm defined on all of C∗ ,
or indeed on any punctured neighbourhood Dr (0) \ {0} of 0.
In fact, the theorem can also be used to explain the fact that n(∂D, p) = 0
when p ∈/ D. Again without loss of generality, we assume that D = DR (0).
Recall that log w can indeed be defined via a power series in the region
|w − 1| < 1, namely
∞
X (−1)n−1
log w = (w − 1)n ,
n
n=1
(log w)0
and it satisfies = w−1 . Now putting w = z/p−1, f (z) = log(z/p−1)
is a holomorphic function in the region |z| < |p| with
1
f 0 (z) = .
z−p
Since C is completely contained in the region |z| < |p|, by Corollary 2,
n(C, p) = 0.
∗ Department of Mathematics, Indian Institute of Science
Email address: [email protected]