Cauchy’s Integral Formula

If f (z)is regular on and inside a closed loop, C, and a is a point inside the loop, it follows
I
1 f (z)
f (a) = dz (1)
2πi C z − a

i.e. the values of f (z) inside the loop are completely determined by the values of f (z) on the loop alone.

Proof(sort of )
Assume that f (z) can be expanded by Taylor’s series (a big assumption) around z = a as

f ′′ (a) f ′′′ (a)

f (z) = f (a) + f ′ (a)(z − a) + (z − a)2 + (z − a)3 + . . . . (2)
2! 3!
By dividing both sides of eq.(6) by (z − a), one obtains

f (z) f (a) f ′′ (a) f ′′′ (a)

= + f ′ (a) + (z − a) + (z − a)2 + . . . (3)
z−a z−a 2! 3!
Integrating both sides of eq.(3) along the loop C, one obtains

I I I I I
f (z) 1 f ′′ (a) f ′′′ (a)
dz = f (a) dz + f ′ (a) 1dz + (z − a)dz + (z − a)2 dz + . . .
C z−a C z−a C 2! C 3! C
= 2πif (a) + 0 + 0 + 0 + . . .
= 2πif (a), (4)

where Cauchy’s theorem was used. This is NOT a mathematically rigorous proof as we have not defined
Taylor’s series expansion for complex functions yet (next Chapter).

Cauchy’s integral formula implies that once a regular function, f (z), is assigned its values on a closed loop,
it automatically determines the values of f (z) at all the points inside the loop.

Generalized Cauchy’s Integral Formulas

If both sides of eq.(6) are divided by (z − a)n+1 instead of (z − a), the following generalized Cauchy’s integral
formulas can be obtained (derivation given in class):

1
 I
n! f (z)
f (n) (a) = dz. (5)
2πi C (z − a)n+1
Again, this result implies that any order of derivative of f (a) is determined completely once the values of
f (z) are given along the contour alone. This further means that if f (z) is regular, it can be differentiated as
many times (∞ times !).
A quick way to derive Cauchy’s integral formulae is to start with the Taylor series of f (z) around z = a as

f ′′ (a) f ′′′ (a)

f (z) = f (a) + f ′ (a)(z − a) + (z − a)2 + (z − a)3 + . . . (6)
2! 3!
To extract f (a), divide both sides of eq.(6) by (z − a) as

f (z) f (a) f ′′ (a) f ′′′ (a)

= + f ′ (a) + (z − a) + (z − a)2 + . . . (7)
z−a z−a 2! 3!
and integrate both sides along a loop that contains z = a as

I I I I I
f (z) f (a) f ′′ (a) f ′′′ (a)
dz = dz + f ′ (a)dz + (z − a)dz + (z − a)2 dz + . . .
C z−a C z − a C C 2! C 3!
I I I I
dz ′ f ′′ (a) f ′′′ (a)
= f (a) + f (a) dz + (z − a)dz + (z − a)2 dz + . . .
C z−a C 2! C 3! C
= 2πif (a) + 0 + 0 + 0 + . . .
= 2πf (a), (8)

so
I
1 f (z)
f (a) = dz. (9)
2πi C z−a
To extract f ′ (a), divide both sides of eq.(6) by (z − a)2 as

f (z) f (a) f ′ (a) f ′′ (a) f ′′′ (a)

= + + + (z − a) + . . . (10)
(z − a)2 (z − a)2 z−a 2! 3!
and integrate both sides along a loop that contains z = a as

I I I I I
f (z) f (a) f ′ (a) f ′′ (a) f ′′′ (a)
dz = dz + dz + dz + (z − a)dz + . . .
C (z − a)2 C (z − a)
2
C z−a C 2! C 3!
I I I I
dz ′ 1 f ′′ (a) f ′′′ (a)
= f (a) + f (a) dz + dz + (z − a)dz + . . .
C (z − a) C z−a
2 2! 3!
C C
= 0 + 2πif ′ (a) + 0 + 0 + . . .
= 2πf ′ (a), (11)

so
I
1 f (z)
f ′ (a) = dz. (12)
2πi C (z − a)2
The generalization of this result is
I
(n) n! f (z)
f (a) = dz. (13)
2πi C (z − a)n+1
Example 1 I
ez
I= dz (14)
C (z − 2)(z + 4)

2
where C is given below. Note that C contains only one of the singular points (z = 2) and the other singular
point, z = −4, is outside the loop.

So if we define
ez
f (z) ≡ (15)
z+4
f (z) is regular on and inside the loop, C, thereby

I
f (z)
I = dz
C z −2
= 2πif (2)
e2
= 2πi
2+4
2
πie
= . (16)
3
Example 2
Same as Example 1 but for the loop, C, shown below:

3
This time, the loop C contains both of the singular points. However, as discussed in class, it is possible to
modify (deform) the loop as long as the topology of the singular points is not altered. Thus

With deformation shown in the figure, the integration can be split into two parts as

I I
f1 (z) f2 (z)
I = dz + dz
C1 z + 4 C2 z −2
= 2πif1 (−4) + 2πif2 (2)
e−4 e2
= 2πi + 2πi (17)
−4 − 2 2+4
where
ez
f1 (z) =
z−2

4
 ez
f2 (z) = , (18)
z+4
Example 3 I
ez
I≡ dz (19)
C z3
where C is a unit circle that encircles the origin. Instead of using the generalized Cauchy’s formula (eq.(5)),
we can expand ez directly by Taylor’s series as

z2 z3 z4
ez = 1 + z + + + + ... (20)
2! 3! 4!
By dividing both sides of the above by z 3 and integrating the result along C, one obtains

I I I I I
ez 1 1 1 1 1
dz = dz + dz + dz + 1dz + . . .
C z3 C z
3
C z
2 2! C z 3! C
1
= 0 + 0 + × 2πi + 0 + . . .
2!
= πi. (21)

Liouville’s theorem
If f (z) is regular and bounded on the entire plane, it must be a constant.

(Proof) Change a to z and z to ζ in eq.(12) to get

I
′ 1 f (ζ)
f (z) = dζ. (22)
2πi C (ζ − z)2

Choose C as a circle centered at z with a radius ρ. Let M be the maximum value of f (ζ) along C. Then,

 I 
 1 f (ζ) 
′
|f (z)| =  dζ 
 2πi (ζ − z) 2
I C 
1  f (ζ) 
=  dζ 
2π C (ζ − z) 2
I  
1  f (ζ) 
≤   dζ
2π C  (ζ − z)2 
I
1 M
≤ dζ
2π ρ2 C
I
1 M
≤ |dζ|
2π ρ2 C
1 M
= 2πρ
2π ρ2
M
= . (23)
ρ

As M in eq.(23) remains finite while ρ in eq.(23) is chosen arbitrarily large, it follows

f ′ (z) → 0 asρ → ∞. (24)

So f ′ (z) = 0 which implies that f (z) = constant. Q.E.D.
According to the Liouville theorem, any analytic function which is not a constant must be unbounded at
some point(s) on the complex plane.

5
Example 1
1
f (x) =
1 + x2

As a real valued function, this function is bounded as seen in the figure above. However, as a complex valued
function, f (z) → ∞ as z → ±i.
Example 2
f (x) = sin x.
This function, as a real valued function, is bounded between -1 and 1. However, as a complex valued function,
it becomes unbounded as the imaginary part of z goes to ∞.

Graph of |sin(x + iy)|. Note that the value increases as the y value increases.

Fundamental Theorem of Algebra

An n-th order algebraic equation, Pn (z) = 0, has at least one root where

Pn (z) ≡ an z n + an−1 z n−1 + . . . a1 z + a0 .

(Proof): Let
1
f (z) ≡ .
P (z)

6
If we assume that Pn (z) is never equal to 0 (i.e. Pn (z) has no zeros), then, f (z) is analytic everywhere (the
denominator never goes to 0) and it is bounded (P (z) → ∞ as z → ∞ along the real axis). Thus, according
to the Liouville theorem, P (z) must be a constant, which is not the case, hence, the assumption (Pn (z) never
goes to 0) is wrong !!! 1
If Pn (z) has a zero, say, a1 , it can be factorized as

Pn (z) = (z − a1 )(bn−1 z n−1 + bn−2 z n−2 + . . . b1 z + b0 ). (25)

Applying the fundamental theorem of calculus recursively, it is shown that an n-th order algebraic equation
has n roots (including multiple roots).

Fundamental Theorem of Calculus

Integration and differentiation are reciprocal to each other.

(Proof): Let ∫ z
F (z) ≡ f (z)dz, (26)
z0

then
∫ z+∆z ∫ z
F (z + ∆z) − F (z) = f (z)dz − f (z)dz
z0 z0
∫ z+∆z
= f (z)dz
z
∼ ∆zf (z)2 (28)

so

F (z + ∆z) − F (z)
∼ f (z), (29)
∆z
or by taking a limit,

F (z + ∆z) − F (z)
lim = f (z). (30)
∆z→0 ∆z
or

F ′ (z) = f (z). (31)

Therefore,

∫ b ∫ b ∫ a
f (z)dz = f (z)dz − f (z)dz
a z0 z0
= F (b) − F (a)
b
= [F (z)]a , (32)

i.e. if one can find F (z) whose derivative is f (z), an integration of f (z) can be computed from F (z). Q.E.D.

1 Unfortunately, the fundamental theorem of algebra does not tell you how to actually solve the algebraic equation.