Chapter 6 Functions of a Complex Variable I

October 3 Cauchy-Riemann conditions

6.1 Complex algebra
Complex number: z  x  iy (both x and y are real, i   1.)
Complex algebra:
z1  z2  x1  iy1   x2  iy 2   x1  x2   i y1  y2  (Anologous to 2d vectors.)
z1 z2  x1  iy1 x2  iy 2   x1 x2  y1 y2   ix1 y2  x2 y1   cz  cx  iy   cx  icy   z1  z2 
Complex conjugation: z *  ( x  iy )*  x  iy
 zz *  ( x  iy )( x  iy )  x 2  y 2

Polar representation: z  x  iy  r (cos   i sin  )  rei

Modulus (magnitude): z  zz  r  x  y  z1 z2  z1 z2
* 2 2

 y
Argument (phase): arg( z )    arctan   ( if z is in the 2nd or 3rd quadrants. )
x
 arg( z1 z 2 )  arg( z1 )  arg( z 2 )

1
Functions of a complex variable:
All elementary functions of real variables may be extended into the complex plane.
 
x x2 xn z z2 zn
Example : e  1      
x
 e  1    
z

1! 2! n 0 n! 1! 2! n 0 n!

A complex function can be resolved into its real part and imaginary part:
f ( z )  u ( x, y )  iv ( x, y )
Examples : z 2  ( x  iy ) 2  ( x 2  y 2 )  i 2 xy
1 1 x y
  2  i
z x  iy x  y 2 x2  y2

Multi-valued functions and branch cuts:

Example 1 : ln z  ln( rei )  ln[ rei ( 2n ) ]  ln r  i(  2n )  u  iv
To remove the ambiguity, we can limit all phases to (-,).
 = - is the branch cut.
lnz with n = 0 is the principle value.


Example 2 : z1 2  (rei )1 2  rei (  2 n ) 
12
 r1 2ei (  2n ) / 2
We can let z move on 2 Riemann sheets so that f ( z )  (re i )1 2 is single valued everywhere.
2
6.2 Cauchy-Riemann conditions
Analytic functions: If f (z) is differentiable at z = z0 and within the neighborhood of
z=z0, f (z) is said to be analytic at z = z0. A function that is analytic in the whole
complex plane is called an entire function.
Cauchy-Riemann conditions for differentiability
df f ( z  z )  f ( z ) f ( z )
f ' ( z)   lim  lim
dz z 0 z z 0 z

In order to let f be differentiable, f '(z) must be the same in any direction of z.
Particularly , it is necessary that
u  iv u v
For z  x, f ' ( z )  lim  i .
x 0 x x x
u  iv u v
For z  iy, f ' ( z )  lim  i  .
y 0 iy y y
Equating them we have
uu vv uu vv
 ,,   Cauchy-Riemann conditions
xx yy yy xx

3
Conversely, if the Cauchy-Riemann conditions are satisfied, f (z) is differentiable:
 u v   u v   u v   v u 
  i x    i y   i x     i y
f ( z )  x x   y y  x x   x x 
 lim 
df
 lim  lim
dz z 0 z z 0 x  iy z 0 x  iy
 u v 
  i x  iy 
x x  u v 1  u v 
 lim    i , and    i .
z 0 x  iy x x i  y y 

More about Cauchy-Riemann conditions:

1) It is a very strong restraint to functions of a complex variable.
df u v v u u v
  i   i 
2) dz x x y y (iy ) (iy ) .  i

u v u v
3)   0  u  v  0  u  v  u  c1  v  c2
x x y y
f
4) Equivalent to  0, so that f ( z,z* ) only depends on z :
z *

f f x f y f 1 f  1  f f  u v   u v 
        0   i  0    i   i  i   0  
z * x z * y z * x 2 y  2i  x y  x x   y y 
e.g., f  x  iy is everywhere continuous but not analytic.
4
Reading: General search for Cauchy-Riemann conditions:
Our Cauchy-Riemann conditions were derived by requiring f '(z) be the same when z
changes along x or y directions. How about other directions?
Here I do a general search for the conditions of differentiability.
 u u   v v   u u dy   v v dy 
 dx  dy   i dx  dy      i  
df du  idv  x y   x y   x y dx   x y dx 
f ' ( z)    
dz dx  idy dx  idy 1 i
dy
dx
dy
Now let  p, the direction of the change of z. We want to find the condition under whic h
dx
f ' ( z ) does not depend on p.
 u u   v v   u v   u u   v v 
  p   i  p    i 1  ip   i  p     p 
df ' ( z ) d  x y   x y    y y   x y   x y 
0
dp dp 1  ip 1  ip 2
 u v   v u   u v That is, if we require f ' ( z ) be the same at all directions ,
    i    
 y x   y x   x y we get the same Cauchy - Riemann conditions .
 
1  ip 2
 u   v
 y x

5
Read: Chapter 6: 1-2
Homework: 6.1.7,6.1.17,6.2.1,6.2.8
Due: October 14

6
October 7 Cauchy’s theorem
6.3 Cauchy’s integral theorem
Contour integral:
z2
z1
f ( z )dz   (u  iv )( dx  idy )   (udx  vdy)  i  (vdx  udy )
C C C

Cauchy’s integral theorem: If f (z) is analytic in a simply connected region R, [and f ′(z)
is continuous throughout this region, ] then for any closed path C in R, the contour
integral of f (z) around C is zero: C f ( z )dz  0

Proof using Stokes’ theorem:  V  dλ     V  dσ

C S

 V Vx 
  V dx  V dy     x
y
 dxdy
y 
x y
C S

 C
f ( z )dz   (udx  vdy)  i  (vdx  udy )
C C

 v u   u v 
     dxdy  i    dxdy
S
 x y  S x
 y 
0

7
Cauchy-Goursat proof: The continuity of f '(z) is not z2
necessary.
Corollary: An open contour integral for an analytic
function is independent of the path, if there is no singular z1
points between the paths.
z2 z1

z1
f ( z )dz  F ( z2 )  F ( z1 )    f ( z )dz
z2

Contour deformation theorem:

A contour of a complex integral can be arbitrarily “nails” z2
C2
deformed through an analytic region without changing
the integral. C1
z1 “rubber bands”
1) It applies to both open and closed contours.
2) One can even split closed contours.
Proof: Deform the contour bit by bit.
Examples: C2
C2 C3
1) Cauchy’s integral theorem.
(Let the contour shrink to a point.) C1 C1
2) Cauchy’s integral formula.
(Let the contour shrink to a small
circle.)
8
6.4 Cauchy’s integral formula
Cauchy’s integral formula:
If f (z) is analytic within and on a closed contour C, then for any point z0 within C,
1 f ( z)
2i C z  z0
f ( z0 )  dz

Proof :
f ( z) f ( z) f ( z) f ( z)
C z  z0 L1 z  z0 C0 z  z0 L2 z  z0
L1 z0
dz  dz  dz  dz  0
L2 C0 C
i
f ( z) f ( z) 0 f ( z  re )
 dz    dz    i
0
i
rie d (Let r  0)
C z  z0 C0 z  z
0
2 re
 2if ( z0 ) Can directly use the contour
deformation theorem.

9
 n! f ( z)
Derivatives of f (z): f ( n ) ( z0 )  
2i z  z0 
C n 1
dz

Corollary: If a function is analytic, then its derivatives of all orders exist.

Corollary: If a function is analytic, then it can be expanded in Taylor series.

Cauchy’s inequality: If f ( z)   an z is analytic and bounded, f ( z ) | z| r  M ,

n

then an r n  M . (That is, an is bounded.)

n! f ( z) 1 f ( z) M
2i | z| r z n 1 
Proof : f ( n ) (0)  n!an  dz  a  dz   a r n
M
2 n 1
n n n
| z|  r z r

Liouville’s theorem: If a function is analytic and bounded in the entire complex plane,
then this function is a constant.
M
Proof : an  n , let r  , then an  0 for n  0. f ( z )  a0 .
r
n
Fundamental theorem of algebra: P( z )   ai z i (n  0, an  0) has n roots.
i 0

Suppose P(z) has no roots, then 1/P(z) is analytic and bounded as z  . Then P(z) is
constant. That is nonsense. Therefore P(z) has at least one root we can divide out.
10
Morera’s theorem: If f (z) is continuous and  f ( z )dz  0 for every closed contour
C
within a simply connected region, then f (z) is analytic in this region.
Proof :
z2
C
f ( z )dz  0   f ( z )dz  F ( z2 )  F ( z1 )  F ' ( z )  f ( z )
z1

 F ( z ) is analytic
 F ' ( z )  f ( z ) is analytic
z2
Why  f ( z )dz  F ( z2 )  F ( z1 ) ?
z1
z2
Let z1
f ( z )dz  G ( z1 , z 2 ), then
G ( z1 , z2 )  G ( z1 ,0)  G (0, z 2 )
 G (0, z1 )  G (0, z2 )   F ( z1 )  F ( z 2 )

11
Read: Chapter 6: 3-4
Homework: 6.4.1,6.4.3,6.4.4
Due: October 14

12
October 10 Analytic continuation
6.5 Laurent expansion
Taylor expansion for functions of a complex variable:
Expanding an analytic function f (z) about z = z0, where z1 is the nearest singular point.
1 f ( z' ) 1 f ( z' )
2i C z ' z 2i C ( z ' z0 )  ( z  z0 )
f ( z)  dz '  dz '
n
 z  z0 


1 f ( z' ) 1
   f ( z ' )
n  0  z ' z 0 

2i C 2i C
 dz '  dz ' z
 z  z0  ( z ' z0 )
( z ' z0 )1  
 z ' z0 
1 
 z  z0  f ( z ' )
n
1 
  f ( z' )
      
n
dz ' z z dz '
2i n 0 ( z ' z0 ) n 1
2i n 0 ( z ' z0 ) n 1
0
C C


f ( n ) ( z0 )
 z  z0 n
n 0 n!

13
Schwarz’s reflection principle:
If f (z) is 1) analytic over a region including the real axis, and 2) real when z is real, then
f * ( z )  f ( z * ).


f ( n ) ( x0 )
Proof : f ( z )   z  x0 n
n 0 n!
 f * ( z)  f ( z* )

Examples: most of the elementary functions.

14
Analytic continuation: Suppose f (z) is analytic around z = z0, we can expand it
about z = z0 in a Taylor series:

f ( m ) ( z0 )
f ( z)   z  z0 m
m 0 m!
a0
This series converges inside a circle with a radius of R0
convergence R0  a 0  z0 , where a0 is the nearest z0
singularity from z = z0.
z1
We can also expand f (z) about another point z = z1 within

f ( n ) ( z1 ) R1
the circle R0: f ( z )   z  z1 n . a1
n 0 n!
In general, the new circle has a radius of convergence R1  a1  z1 and contains points
not within the first circle.

f ( m ) ( z0 )
From the first expansion, f ( z1 )  
(n)
z1  z0 mn
m  n ( m  n)!

f ( m ) ( z0 )z1  z0 
 mn
Plug into the second expansion, f ( z )   z  z1 n
n 0, m  n n!(m  n)!
Consequences:
1) f (z) can be analytically continued over the complex plane, excluding singularities.
2) If f (z) is analytic, its values at one region determines its values everywhere.
15
Read: Chapter 6: 5
Homework: 6.5.2,6.5.5
Due: October 21

16
October 12 Laurent expansion
6.5 Laurent expansion
Problem: Expanding a function f (z) that is analytic in an annular region (between r and R).
1 f ( z ' )dz '
2i C1  L1 C2  L2 z ' z
f ( z)  ~

1 f ( z ' )dz ' 1 f ( z ' )dz '

2i C1 z ' z 2i C2 z ' z
 
L2
1 f ( z ' )dz ' 1 f ( z ' )dz '
2i C1 ( z ' z0 )  ( z  z0 ) 2i C2 ( z ' z0 )  ( z  z0 )
  L1

1 f ( z ' )dz ' 1 f ( z ' )dz '

2i C1  z  z0  2i C2
 
 z ' z0 
( z ' z0 )1   ( z  z0 )1  
 z ' z 0   z  z 0 

1  f ( z ' )dz ' 1  1

 
2i n 0
( z  z 0 ) n
 
C1 ( z ' z0 ) n1 2i m0 ( z  z0 ) m1 C2 ( z ' z 0 ) m
f ( z ' )dz '

1  f ( z ' )dz ' 1  1

 
2i n 0
( z  z 0 ) n
C1 ( z ' z0 ) n1 2i 

m 1 ( z  z 0 )
m C2
( z ' z0 ) m 1 f ( z ' )dz '

1  f ( z ' )dz ' 1  f ( z ' )dz '

 
2i n 0
( z  z 0 ) n
C1 ( z ' z0 ) n1 2i n

 1
( z  z 0 ) n
C2 ( z ' z0 ) n1
1  f ( z ' )dz '
 C ( z ' z0 ) n1
C is any contour that encloses z0 and lies
 ( z  z ) n
17
2i n  
0 between r and R (deformation theorem).
Laurent expansion:

1 f ( z ' )dz '
f ( z)   an ( z  z0 )n , an 
n   2i C ( z ' z0 ) n1
1) Singular points of the integrand.
For n < 0, the singular points are determined by f (z). For n ≥0, the singular
points are determined by both f (z) and 1/(z'-z0)n+1.
2) If f (z) is analytic inside C, then the Laurent series reduces to a Taylor series:
 f ( n ) ( z0 )
 , n  0,
an   n!
0, n  0.

3) Although an has a general contour integral form, In most times we need to use
straight forward complex algebra to find an.

18
Laurent expansion: Examples

z3
Example 1: Expand f ( z )  about z0=1.
z  12

z3 [( z  1)  1]3 ( z  1)3  3( z  1) 2  3( z  1)  1 1 3
     3  ( z  1)
z  12 z  12 z  12 z  12 z  1

1
Example 2: Expand f ( z )  about z0=i.
z 1
2

1 1 1 1  1 1 1 
f ( z)         
z 2  1 2i  z  i z  i  2i  z  i 2i  z  i 
 
 1  1 1  n
1 1 1 1  1
     
2i  z  i 2i 1  z  i  2i z  i 2i 2
    ( z  i)
n 0  2i 
n

 
 2i 
i 1 1 i
   ( z  i)  
2 z i 4 8

19
Read: Chapter 6: 5
Homework: 6.5.8,6.5.9,6.5.10
Due: October 21

20
October 17 Branch points and branch cuts
6.6 Singularities 
Poles: In a Laurent expansion f ( z )  a
m  
m ( z  z0 ) m , if am  0 for m  n  0 and a n  0,

then z0 is said to be a pole of order n.

A pole of order 1 is called a simple pole.
A pole of infinite order (when expanded about z0) is called an essential singularity.
The behavior of a function f (z) at infinity is defined using the behavior of f (1/t) at t = 0.
Examples:
1 1 1 1 1  1 1 1  1 1 1 1
          
z  1 ( z  i)( z  i ) 2i  z  i z  i  2i  z  i 2i  ( z  i ) 
1) 2
2i z  i 4 1  ( z  i ) / 2i
1  z i  z i  
2
1 1
  1      has a single pole at z  i.
2i z  i 4  2i  2i  

(1) n z 2 n1 1  (1) n 1
2) sin z   , sin  
n 0 (2n  1)! t n0 (2n  1)! t 2 n1
sinz thus has an essential singularity at infinity.

3) z 2  1 has a pole of order 2 at infinity.

21
Branch points and branch cuts:
Branch point: A point z0 around which a function f (z) is discontinuous after going a
small circuit. E.g., z0  1 for z-1, z0  0 for ln z.
Branch cut: A curve drawn in the complex plane such that if a path is not allowed to
cross this curve, a multi-valued function along the path will be single valued.
Branch cuts are usually taken between pairs of branch points. E.g., for z-1 , the curve
connects z=1 and z =  can serve as a branch cut.

Examples of branch points and branch cuts:

1. f ( z )  z a  r a (cos a  i sin a )

If a is a rational number, a  p / q, then circling the branch point z = 0 q times will bring
f (z) back to its original value. This branch point is said to be algebraic, and q is called
the order of the branch point.
If a is an irrational number, there will be no number of turns that can bring f (z) back to
its original value. The branch point is said to be logarithmic.

22
2. f ( z)  ( z 1)( z  1)
A
B
We can choose a branch cut from z = -1 to z = 1 (or any
curve connecting these two points). The function will be
single-valued, because both points will be circled.
Alternatively, we can choose a branch cut which
connects each branch point to infinity. The function will
be single-valued, because neither points will be circled.
A
It is notable that these two choices result in different B
functions. E.g., if f (i )  2i, then
f (i )   2i for the first choice and f (i )  2i
for the second choice.

23
Read: Chapter 6: 6
Homework: 6.6.1,6.6.2,6.6.3
Due: October 28

24
October 19 Mapping
6.7 Mapping y v
Mapping: A complex function w( z )  u ( x, y )  iv ( x, y ) z1 w1
can be thought of as describing a mapping from the
complex z-plane into the complex w-plane. z0 w0
In general, a point in the z-plane is mapped into a
point in the w-plane. A curve in the z-plane is mapped z x w u
into a curve in the w-plane. An area in the z-plane is
mapped into an area in the w-plane.

Examples of mapping:
Translation:
w  z  z0

Rotation:
w  zz0 , or
i i i 0   r  r0
e  re  r0e 
     0

25
Inversion:
1
w  , or
z
 1
1   
ei  i   r
re   

In Cartesian coordinates:
 x  u
 u  x 
1 1  x 2  y 2  u 2  v2
w   u  iv   , .
z x  iy v   y y   v
2 
 x y 
2
u 2  v2
A straight line is mapped into a circle:
v au
y  ax  b    b
u v
2 2
u v
2 2

 b(u 2  v 2 )  au  v  0.

26
6.8 Conformal mapping
Conformal mapping: The function w(z) is said to be conformal at z0 if it preserves the
angle between any two curves through z0.
If w(z) is analytic and w'(z0)0, then w(z) is conformal at z0.
Proof: Since w(z) is analytic and w'(z0)0, we can expand w(z) around z = z0 in a
Taylor series:
1
w  w( z0 )  w' ( z0 )( z  z0 )  w' ' ( z0 )( z  z0 ) 2  
2
w  w0
lim  w' ( z0 ), or w  w0  w' ( z0 )( z  z0 ).
z  z 0 0 z  z
0

w  w0  Aeia ( z  z0 )    a     2  1   2  1.

1) At any point where w(z) is conformal, the mapping consists of a rotation and a
dilation.
2) The local amount of rotation and dilation varies from point to point. Therefore a
straight line is usually mapped into a curve.
3) A curvilinear orthogonal coordinate system is mapped to another curvilinear
orthogonal coordinate system .

27
What happens if w'(z0) = 0?
Suppose w (n)(z0) is the first non-vanishing derivative at z0.
 Br n
 
(n)
w ( z0 ) 1
w  w0  ( z  z0 ) n  ei  Be i (re i ) n   n!
n! n!   n  

This means that at z = z0 the angle between any two curves is magnified by a factor n
and then rotated by .

28
Read: Chapter 6: 7-8
Homework: 6.7.1,6.7.2,6.8.1
Due: October 28

29