Lecture

3
Basic Properties of Differentiable Functions

Proposition 1. Let f :    be defined in some
neighbourhood of a and differentiable at a. Then,
f f
( a ), ( a ) exist and satisfy
x y
f f
( a )   i ( a ) . (*)
x y

Notes: 1. For a function  :    , the first order partial

derivatives  x (also denoted as ) and  y (also denoted as
x

) at a point   ( ,  ) are defined as , for t   ,
y
 (  t ,  )   ( ,  )  ( ,   t )   ( ,  )
 x  lim ,  y  lim
t 0 t t 0 t

2. For a function f :    , given by f  u  i v , the
first order partial derivatives f x and f y , at a point
a    i  , are defined as, for t   ,

f (  t ,  )  f ( ,  )
f x  lim  ux  i vx ,
t 0 t
f ( ,   t )  f ( ,  )
f y  lim  u y  i v y .
t 0 t



Proof of Proposition 1. Let a    i  , t  , t  0. Then,

f (a  t )  f (a ) f (  t ,  )  f ( ,  ) f
f ( a )  lim  lim  (a )
t 0 t t 0 t x

f (a  it )  f ( a ) f ( ,   t )  f ( ,  ) f
f (a )  lim  lim  i ( a ) .
t 0 it t 0 it y

The identity (*) follows by the above identities.



Proposition 2. Let f = u + i v be differentiable at a   . Then,
u v v u
( a )  ( a ), ( a )   ( a ) .
x y x y

Proof. By Prop. 1,

f f
(a )   i (a )
x y
u v u v
 ( a )  i ( a )  i ( ( a )  i ( a ))
x x y y

Equating real and imaginary parts of the above identity we
get the result of Proposition 2.


Definition. Any one of the following equations

f f
  i
x y
u v v u
  , 
x y x y

are called Cauchy Riemann equations.


Note that the above equations are equivalent.

Example. The function

 ( z )2
 if z  0
f z   z
0 if z  0


satisfies CR equations at 0, but is not diff. at 0.

f is not diff. at 0:

__
f ( z ) z x  i y 2
lim  lim ( )2  lim ( )
 z  0 z z  0  z z 0 x  i y
1 if y  0
 1 if x  0
1 i 2
( ) if x  y
1 i

CR Equations are satisfied at 0:

Note that

u( x,0)  Re f ( x,0)  Re x  x
( i y ) 2 y 2
u(0, y )  Re f (0, y )  Re  Re(i )  0
i y y
v ( x,0)  Im f ( x,0)  Im x  0
v (0, y )  Im f (0, y )  Im(i y )  y .

Consequently,

u ( x,0)  u (0,0) x
  1 as x  0
x x
u (0, y )  u (0,0) 0
  0 as y  0
y y
v ( x,0)  v (0,0) 0
  0 as x  0
x x
v (0, y )  v (0,0) y
  1 as y  0 .
y y

Therefore,

u v v u
(0)  (0)  1; (0)   (0)  0 .
x y x y

Exercise. Prove that for the following functions CR equations
are satisfied at 0 but the functions are not differentiable at 0:
 z5
 4 if z  0
(i) f  z    z

0 if z  0
 x 3 (1  i )  y 3 (1  i )
 if z  0
(ii) f  z    x y
2 2
0 if z  0



Definition. Any function  :    having continuous partial
derivatives up to second order and satisfying the equation
 2  2
   2  2  0 (*) is called a Harmonic Function or
2
x y
Potential Function. Equation (*) is called Laplace Equation
and 2 is called Laplacian of the function  .

Harmonic Functions are widely used in the study of steady
state temperatures, wave theory, two dimensional
electrostatics, fluid flow, robotics etc.

Since real and imaginary parts u and v of a complex
differentiable function f  u  i v satisfy CR‐equations, it
easily follows that 2u  0, 2 v  0 . Thus, real and imaginary
parts of a complex differentiable function are Harmonic
functions.
The converse of Prop.1, proved below, holds under the
additional hypothesis that f / x, f / y exist, are continuous
and satisfying CR equations f / x  i f / y .


Theorem. Let, for f ( z ) defined in a domain G, f / x, f / y
exist, are continuous and CR Equation f / x  i f / y is
satisfied at any point z0  ( x0 , y0 )  G . Then, f ( z0 ) exists and
is given by f ( z0 )  u x ( x0 , y0 )  i v x ( x0 , y0 ) .

Proof. Let,
z0  x0  iy0  G & h  s  it be such that h   and z0  h  G.

Then,

f ( z0  h )  f ( z0 ) u( x0  s, y0  t )  u( x0 , y0 )
 
h h (1)
v ( x0  s, y0  t )  v( x0 , y0 )
i
h








Now,

u ( x0  s, y0  t )  u ( x0 , y0 )
h

u ( x0  s, y0  t )  u( x0 , y0  t ) u ( x0 , y0  t )  u( x0 , y0 )
 
h h
 ( s, t ) s t
  u x ( x0 , y0 )  u y ( x0 , y0 )
h h h . (2)
( say , i.e. define  ( s, t ) by this identity )

Then,

 ( s, t ) s t
 u x ( x0  s1 , y0  t )  u y ( x0 , y0  t1 )
h h h

s t
 u x ( x0 , y0 )  u y ( x0 , y0 )  By MVT &(2) 
h h

s t
 0 as h  0 ( u x & u y are continuous at z0 and  1,  1)
h h
Similarly,

v ( x0  s, y0  t )  v ( x0 , y0 )  ( s, t ) s t
  v x ( x0 , y0 )  v y ( x0 , y0 ),
h h h h

 ( s, t )
where, 0
h


 By (1),

f ( z0  h )  f ( z0 ) s t   i
 [u x  i v x ]  [u y  i v y ] 
h h h h

s t   i
 [u x  i v x ]  [  v x  i u x ]  ( since u x  v y & u y   v x )
h h h
s it   i
 [u x  i v x ]  [u x  i v x ] 
h h h
  i
 ux  i vx 
h
 u x  i v x as h  0

f ( z0  h )  f ( z0 )
 lim exists & equals u x  i v x at z0 .
h 0 h
Cauchy Riemann Equations in Polar Coordinates

Let f ( z )  u( r, )  i v ( r, ) . Then,
r ur  v , r vr  u
are Cauchy Riemann equations in Polar coordinates.

The above equations can be easily obtained from CR
equations in Cartesian coordinates as follows:

With x  r cos  , y  r sin  ,
ur  u x xr  u y yr  u x (cos  )  u y (sin  )
v  v x x  v y y  v x (  r sin  )  v y ( r cos  )
 r ur  v
by CR Equations u x  v y , u y   v x in cartesian coordinates .

Similarly, r vr  u .

The following proposition readily follows from the above
theorem:
 f f
Proposition 4. Let f ( z )  u( r, )  i v ( r, ) . If
, exist,
r 
are continuous in some domain G, and satisfy the following
Cauchy Riemann equations in polar coordinates at a point
z0  ( r0 ,0 )  G ,

r ur ( r0 ,0 )  v ( r0 ,0 )
r vr ( r0 ,0 )  u ( r0 ,0 )

then, f ( z0 ) exists and is given by
u ( r , )  i v ( r , )
f ( z0 )  r 0 0 i r 0 0 .
e 0
y
Proof. Since, r 2  x 2  y 2 , tan   ,
x

1
u x  ur rx  u  x  ur (cos  )  u (sin  )
r
 ur (cos  )  vr (sin  )
1
v x  vr rx  v  x  vr (cos  )  v (sin  )
r
 vr (cos  )  ur (sin  )

Therefore, using the above theorem,

 f ( z0 )  u x ( x0 , y0 )  i v x ( x0 , y0 )
u ( r , )  i v ( r , )
 r 0 0 i r 0 0
e 0
Elementary Functions

Logarithmic Function. Define

log z  ln z  i arg z

Since arg z takes multiple values for every z, log z also takes
multiple values for every z and is therefore not a function.

To make it a well‐defined function, the range of arg z has to
be so restricted that it takes a unique value for every value of
z in its domain of definition.

To this end, define

Log z  ln z  i Arg z

where,   Argz   . Arg z takes a unique value for every
z  0 . Consequently, Log z is a well‐defined function for every
z  0 . Log z is called the Principal Branch of Logarithmic
Function.

 Log z maps   {0}onto the strip {w :   Im z   }.


 Log z is continuous at all points in its domain of
definition except at the points on the negative real axis,
since Arg z is a continuous at all the points in its domain
of definition, except the points on negative real axis.