CHAPTER ONE

ORDINARY DIFFERENTIAL EQUATIONS

1.1 Power Series Method

Recall the power series of the form
∞
X
am (x − x0 )m = a0 + a1 (x − x0 ) + a2 (x − x0 )2 + .... (1.1)
m=0

.S
a0 , a1 , a2 ,...... are constants, called the coefficients of the series, x0 is a

Q
constant called the center of the series, and x is the variable. If x0 = 0, we

aa
Sl
obtain power series in powers of x:
∞
X
am xm = a0 + a1 x + a2 x2 + .... (1.2)
.S
Q

m=0
aa

we assume that all variables and constants are real. Some of the familiar
Sl

examples of power series are the Maclaurin series:

.S

1
P∞
1. = xm = 1 + x + x2 + ....(|x| < 1)
Q

1−x m=0
aa

P∞ xm x2 x3
Sl

2. ex = m=0 m! =1+x+ 2!
+ 3!
+ ....
P∞ (−1)m x2 m x2 x4
3. cosx = m=0 (2m)!
=1− 2!
+ 4!
− +....
.S
Q

P∞ (−1)m x2 m+1 x3 x5
4. sinx = =x− + − +....
aa

m=0 (2m+1)! 3! 5!
Sl

Power series method for solving differential equation is simple and natural.
d2 y dy
Example 1.1. dx2
+ p(x) dx + q(x)y = 0

As a solution we first assume represent p(x) and q(x) by power series in

powers of x or x − x0 if solutions of x − x0 are wanted. Often p(x) and q(x)
are polynomials and then nothing needs to be done in this first step. We
assume a solution in the form of power series with unknown coefficients:
∞
X
y= am xm = a0 + a1 x + a2 x2 + a3 x3 .... (1.3)
m=0
it follows that
∞
dy X
=m am xm−1 = a1 + 2a2 x + 3a3 x2 + .... (1.4)
dx m=1

∞
d2 y X
2
= m(m − 1) am xm−2 = 2a2 + 6a3 x + 12a4 x2 + .... (1.5)
dx m=1

We collect like terms of x and equate the sum of the coefficients of each
occurring power of x to zero, starting with the constant term etc.
dy
Example 1.2. Solve dx
−y =0

.S
Q
aa
In the first step, we insert (1.3) and (1.4)into the equation and we get

Sl
(a1 + 2a2 x + 3a3 x3 + ....) − (a0 + a1 x + a2 x2 + .....) = 0
.S

Then we collect like powers of x, finding

Q
aa

(a1 − a0 ) + (2a2 − a1 )x + (3a3 − a2 )x2 + ...... = 0

Sl

Equating the coefficient of each power of x to zero, we have: a1 − a0 = 0,

2a2 − a1 = 0, 3a3 − a2 = 0, .....
.S
Q

Solving these equations, we may express a1 , a2 ,.... in terms of a0 , which

aa

remains arbitrary, thus a1 = a0 , a2 = a21 = a2!0 , a3 = a32 = a3!0 with these

Sl

coefficients the series (1.3) becomes

a0 2 a0 3
y = a0 + a0 x + x + x + .....
.S

2! 3!
Q
aa

simplifying further we get

Sl

x2 x3
y = a0 (1 + x + + + ....., ) = a0 ex
2! 3!

1.2 Frobenius’ Method

In power series method, we established the solution as a power series in
integral powers of x. Such a solution is not always possible and a more
general method is to assume a trial solution of the form

y = xc (a0 + a1 x + a2 x2 + a3 x3 + ... + ar xr + .....)

2
where a0 is the first coefficient that is not zero. The type of equation that
can be solved by this method is of the form

d2 y dy
+ P + Qy = 0
dx2 dx
where P and Q are functions of x. However, certain conditions have to be
satisfied:

(a) If the functions P and Q are such that both are finite when x is put
equal to zero, x = 0 is called an ordinary point of the equation.

.S
(b) If xP and x2 Q remain finite at x = 0, then x = 0 is called a regular

Q
singular point of the equation.

aa
Sl
(c) If, however, P and Q do not satisfy either of these conditions stated
in (a) and (b), then x = 0 is called an irregular singular point of the
equation and the method of Frobenius cannot be applied.
.S
Q
aa
Sl

In both of the conditions (a) and (b) the method of Frobenius can be applied.

1.2.1 Solution of differential equations by the method of Frobe-

.S

nius
Q
aa
Sl

To solve a given equation, we have to find the coefficients a0 ,a1 ,a2 ,a3 ,..... and
also the index c in the trial solution. Basically, the steps in the method are
as follows:
.S
Q
aa
Sl

(a) Differentiate the trial series as required.

(b) Substitute the results in the given differential equation.

(c) Equate coefficients of corresponding powers powers of x on each side of

the equation.
d y 2 dy
Example 1.3. Find a series solution for the equation 2x dx 2 + dx
+ y = 0.

Assume a solution of the form

y = xc (a0 + a1 x + a2 x2 + a3 x3 + ... + ar xr + .....)

3
Where, a0 6= 0. Differentiating term by term, we get
dy
= a0 cxc−1 + a1 (c + 1)xc + a2 (c + 2)xc+1 + ..... + ar (c + r)xc+r−1 + ......
dx
Repeating the process one stage further, we have

d2 y
= a0 c(c−1)xc−2 +a1 c(c+1)xc−1 +a2 (c+1)(c+2)xc +....+ar (c+r−1)(c+r)xc+r−2 +....
dx2
Considering each term of the equation in turn gives

d2 y
2x = 2a0 c(c − 1)xc−1 + 2a1 c(c + 1)xc + 2a2 (c + 1)(c + 2)xc+1 + ....

.S
dx2

Q
+ ar (c + r − 1)(c + r)xc+r−1 + ....

aa
Sl
dy
= a0 cxc−1 + a1 (c + 1)xc + a2 (c + 2)xc+1 + .... + ar (c + r)xc+r−1 + ...
dx
y = a0 xc + a1 xc+1 + .... + ar xc+r + ....
.S
Q

Adding these three lines to form the left-hand side of the equation, we can
aa

equate the total coefficient of each power of x zero, since the right-hand side
Sl

is zero. xc−1 gives

a0 c(2c − 1) = 0 (1.6)
.S

Similarly, xc gives
Q
aa

a1 (c + 1)(2c + 1) + a0 = 0 (1.7)
Sl

Also xc+1 gives

a2 (c + 2)(2c + 3) + a1 = 0 (1.8)
.S

Note that the coefficient of xc involves all three lines of the expression and,
Q

from then on,a general relationship can be obtained for xc+r , r ≥ 0. In the
aa

d2 y dy
Sl

c+r−1
expression for 2x dx 2 and dx we have terms in x . If we replace r by
c+r
(r + 1), we shall obtain the corresponding terms in x .
d2 y c+r
In the series for 2x dx 2 , this is 2ar+1 (c + r)(c + r + 1)x
dy
In the series for dx , this is ar+1 (c + r + 1)xc+r
In the series for y, this is ar xc+r
Therefore, equating the total coefficient of xc+r to zero, we have 2ar+1 (c +
r)(c + r + 1) + ar+1 (c + r + 1) + ar = 0 which tidies up to

ar+1 [(c + r + 1)(2c + 2r + r + 1)] + ar = 0 (1.9)

4
1.2.2 Indicial Equation

Equation (1.6), formed from the coefficient of the lowest power of x, that
is xc−1 , is called the indicial equation from which the values of c can be
obtained. In the present example a0 (2c − 1) = 0 since a0 6= 0 by definition
c = 0 or c = 21 . Both values of c are valid, so that we have two possible
solutions of the given equation. We will consider each in turn.
a0
Using c = 0 (1.7) gives a1 = −a0 , (1.8) gives a2 = 2×3 and from (1.9)
−ar
ar+1 = (r+1)(2r+1) , r ≥ 0. From the combined series, the term in xc and
all subsequent terms involve all three lines and the coefficient of the general
term can be used.
−ar

.S
So we have a1 = −a0 and ar+1 = (r+1)(2r+1) for r = 0, 1, 2, ..... Therefore,

Q
−a1 a0 −a2 −a0 −a3 a0
a2 = 2×3 = 2×3 , a3 = 3×5 = (2×3)(3×5) , a4 = 4×7 = (2×3×4)(3×5×7) etc.

aa
Sl
Thus,
a0 3 a0
y = x0 [a0 − a0 x + x − x3 + .....]
2×3 (2 × 3)(3 × 5)
.S
Q

or
aa
Sl

x2 x3 x4
y = a0 [1 − x + − + + .....]
2 × 3 (2 × 3)(3 × 5) (2 × 3 × 4)(3 × 5 × 7)
.S

Now we go through the same steps using our second value for c, that is c = 12 .
Q

Putting c = 21 in (1.7) we get a1 = − a30 , similarly (1.8) gives a2 = − a101 = 3×10

a0
aa

−ar
Sl

and from the general relationship, (1.9) we have ar+1 = (r+1)(2r+3)

1
y = xc [a0 + a1 x + a2 x2 + a3 x3 + ... + ar xr + ...] that is y = x 2 [a0 − a30 x +
a0 a0 1 x x2 x3
x2 − (1×2×3)(3×5×7) x3 +...] = a0 x 2 [1− 1×3 + (1×2)(3×5) − (1×2×3)(3×5×7) +
.S

(1×2)(3×5)
...]. Since a0 is an arbitrary (non-zero)constant in each solution, its values
Q
aa

may well be different, A and B say. If we denote the first solution by u(x)
Sl

and the second by v(x), then

x2 x3 x4
u = A[1 − x + − + + .....]
2 × 3 (2 × 3)(3 × 5) (2 × 3 × 4)(3 × 5 × 7)
and

1 x x2 x3
v = Bx 2 [1 − + − + ...]
1 × 3 (1 × 2)(3 × 5) (1 × 2 × 3)(3 × 5 × 7)
x2 x3
The general solution y=u+v is therefore, y = A[1 − x + 2×3
− (2×3)(3×5)
+
x4 1 x x2 x3
(2×3×4)(3×5×7)
+ .....] + Bx [1 −
2
1×3
+ (1×2)(3×5)
− (1×2×3)(3×5×7)
+ ...]

5
 d 2
dy
Example 1.4. Find the series solution for the equation 3x2 dx 2 − x dx + y −

xy = 0.

We proceed just in the same way as in the previous example. Assume

y = xc (a0 + a1 x + a2 x2 + a3 x3 + ... + ar xr + .....)

therefore
dy
= a0 cxc−1 + a1 (c + 1)xc + a2 (c + 2)xc+1 + ..... + ar (c + r)xc+r−1 + ......
dx

.S
and

Q
d2 y

aa
= a0 c(c−1)xc−2 +a1 c(c+1)xc−1 +a2 (c+1)(c+2)xc +....+ar (c+r−1)(c+r)xc+r−2 +....

Sl
dx2
Now we build up the terms in the given equation
.S

d2 y
Q

3x2 = 3a0 c(c − 1)xc + 3a1 (c + 1)cxc+1 + 3a2 (c + 2)(c + 1)xc+2 + ....
aa

dx 2
Sl

+ 3ar (c + r)(c + r − 1)xc+r + ....

dy
−x = −a0 cxc − a1 (c + 1)xc+1 − a2 (c + 2)xc+2 − .... − ar (c + r)xc+r − ...
.S

dx
Q

y = a0 xc + a1 xc+1 + a2 xc+2 + .... + ar xc+r + ...

aa
Sl

−xy = −a0 xc+1 − a1 xc+2 − .... − ar xc+r+1 − ....

The indicial equation, that is equating the coefficient of the lowest power of
x to zero, gives the value of c. Thus, in this case c = 1 or 31 because the
.S
Q

lowest power is xc and the indicial equation is a0 (3c2 − 4c + 1) = 0. The

aa

coefficient of the general term xc+r gives

Sl

3ar (c + r)(c + r − 1) − ar (c + r) + ar − ar−1 = 0 (1.10)

From this equation we can form the recurrence relation

ar−1
ar = (1.11)
(c + r − 1)(3c + 3r − 1)

Using c = 1 the recurrence relation becomes

ar−1
ar =
r(3r + 2)

6
Thus
a0
a1 =
1×5
a0
a2 =
(1 × 2)(5 × 8)
a0
a3 =
(1 × 2 × 3)(5 × 8 × 11)
our first solution is therefore,
a0 x a0 x 2 a0 x 3
y = x1 (a0 + + + + ....)
1 × 5 (1 × 2)(5 × 8) (1 × 2 × 3)(5 × 8 × 11)

.S
x x2 x3
= Ax1 (1 + + + + ....)

Q
1 × 5 (1 × 2)(5 × 8) (1 × 2 × 3)(5 × 8 × 11)

aa
Sl
For the second solution, we put c = 13 . The recurrence relation the becomes
ar−1
ar =
.S

r(3r − 2)
Q
aa

Therefore, we can now determine the coefficients for r = 1, 2, 3, 4, .... and

Sl

complete the second solution

1 x2 x3 x4
.S

y = Bx 3 (1 + x + + + + .....)
2 × 4 (2 × 3)(4 × 7) (2 × 3 × 4)(4 × 7 × 10)
Q
aa

Therefore, the general solution is

Sl

x x2 x3
y = Ax1 (1 + + + + ....)
1 × 5 (1 × 2)(5 × 8) (1 × 2 × 3)(5 × 8 × 11)
.S
Q

1 x2 x3 x4
+ Bx 3 (1 + x + + + + .....)
aa

2 × 4 (2 × 3)(4 × 7) (2 × 3 × 4)(4 × 7 × 10)

Sl

d2 y
Example 1.5. Find the series solution for the equation dx2
− y = 0.

As usual, we start off with the assumed solution

y = xc (a0 + a1 x + a2 x2 + ... + ar xr + ....)

So that we have
dy
= a0 cxc−1 + a1 (c + 1)xc + a2 (c + 2)xc+1 + .... + ar (c + r)xc+r−1 + ......
dx
d2 y
2
= a0 c(c − 1)xc−2 + a1 (c + 1)cxc−1 + a2 (c + 2)(c + 1)xc + .... + ar (c + r)(c + r − 1)xc+r−1 + ..
dx
7
Now we build up the terms in the left-hand side of the equation.
d2 y
dx2
= a0 c(c − 1)xc−2 + a1 (c + 1)cxc−1 + a2 (c + 2)(c + 1)xc + .... + ar (c + r)(c +
r − 1)xc+r−1 + ....
y = xc [a0 + a1 x + a2 x2 + ... + ar xr + ....]
The term in xc+r in the first of these expansion is

ar+2 (c + r + 2)(c + r + 1)xc+r (1.12)

and this is obtained by replacing r by (r + 2) in ar (c + r)(c + r + 1)xc+r−2

Then
d2 y
−y = a0 c(c−1)xc−2 +a1 (c+1)cxc−1 +[a2 (c+2)(c+1)−a0 ]xc +...+[ar+2 (c+r+2)(c+r+1)−ar ]xc+r
dx2

.S
Q
We now equate each coefficient in turn to zero, since the right-hand side of

aa
Sl
the equation is zero. The coefficient of the lowest power of x gives the indicial
equation from which we obtain the values of c. So, in this case c = 0 or 1.
For the term in xc−1 , we have
.S
Q

a1 (c + 1)c = 0 (1.13)
aa
Sl

with c = 1, a1 = 0. But with c = 0, a1 is indeterminate, because any value

of a1 combined with the zero value of c would make the product zero. For
xc term we have
.S

a2 (c + 2)(c + 1) − a0 = 0 (1.14)
Q
aa

a0
therefore a2 = (c+1)(c+2) .
Sl

c+r
For the general term x we have
ar
.S

ar+2 = (1.15)
(c + r + 1)(c + r + 2)
Q
aa
Sl

From the indicial equation, c = 0 or c = 1. When c = 0 a1 is indeterminate,

a2 = a20 ,
In general we have
ar
ar+2 = (1.16)
(r + 1)(r + 2)
a1
thus, a3 = 2×3 , a4 = a4!0
And therefore, one solution is y = x0 [a0 + a1 x + a2!0 x2 + a3!1 x3 + a4!0 x4 + ...] that
2 4 3 5
is, y = a0 [1 + x2! + x4! + ...] + a1 [x + x3! x5! + ...] a0 and a1 are arbitrary constants
depending on the boundary conditions.
a0 ar
Similarly, when c = 1, a1 = 0, a2 = 2×3 and ar+2 = (r+2)(r+3) , therefore,
a0 a1 a0
a1 = 0, a2 = 3! , a3 = 3×4 , a4 = 5! , a5 = 0 etc.
3 5
A second solution with c = 1 is therefore y = a0 [x+ x3! + x5! +...]. Note that this

8
is not, in fact, a separate solution, since it already forms the second series in
the solution for c = 0 obtained previously.Therefore, the first solution, with
its two arbitrary constants gives the general solution. This happen when the
two values of c differ by an integer.
Make the note of the following

(a) If c1 and c2 differ by not an integer then two independent solutions,

y = u(x) and y = v(x), are obtained.The general solution is then
y = Au + Bv.
(b) If c1 and c2 differ by an integer, i.e c2 = c1 + n, and if one coefficient ar
is indeterminate when c = c1 , the complete general solution is given by

.S
using this value of c. Using c = c1 + n gives a series which is a simple

Q
aa
multiple of one of the series in the first solution.

Sl
(c) If the roots c = c1 and c = c1 + n of the indicial equation differ by an
integer and one coefficient ar becomes infinite when c = c1 , the series is
.S

rewritten with a0 replaced by k(c − c1 ). Putting c = c1 in the rewritten

Q

series and that of its derivative with respect to c gives two independent
aa
Sl

solutions.
d y 2dy
Example 1.6. Find the series solution of the equation x dx2 +(2+x) dx −2y =

0.
.S
Q
aa

Using y = xc [a0 + a1 x + a2 x2 + ... + ar xr + ....] and its first two derivatives

Sl

as before will lead us to the indicial equation

a0 (c2 + c) = 0 (1.17)
.S
Q

This will give c = 0 or c = −1 also from the expansions, the total coefficient
aa

−a0 (c−2)
of xc gives a1 = (c+1)(c+2) . From the terms in xc+r , all four expansions are
Sl

involved, so we can form the recurrence relation from the coefficient of xc+r .
−ar (c + r − 2)
ar+1 = (1.18)
(c + r + 1)(c + r + 2)
From the expansion we obtain
−a0 (c − 1)(c − 2)
a2 =
(c + 1)(c + 2)2 (c + 3)
and from the recurrence relation when r = 2 we have
−a0 c(c − 1)(c − 2)
a3 =
(c + 1)(c + 2)2 (c + 3)2 (c + 4)

9
Therefore,

(c − 2) (c − 1)(c − 2) c(c − 1)(c − 2)

y = a0 xc [1− x+ 2
x2 − x3 +....]
(c + 1)(c + 2) (c + 1)(c + 2) (c + 3) (c + 1)(c + 2)2 (c + 3)2 (c + 4)

From the indicial equation above, the values of c are c = 0 and c = −1.
Putting c = 0 we have one solution

x2
y = u = a0 [1 + x + ]
6
Note that coefficients after x2 are zero, because of factor c in the numera-
tor.Putting c = −1, we soon find that coefficients become infinite, because

.S
of the factor (c + 1) in the denominator.

Q
aa
Therefore, we substitute a0 = k(c − c1 ) = k(c − [−1]) = k(c + 1).This will

Sl
give us

−(c − 2) (c − 1)c − 2) c(c − 1)(c − 2)

y = k(c+1)xc [1− x+ x2 − x3 +...]
.S

(c + 1)(c + 2) 2
(c + 1)(c + 2) (c + 3) (c + 1)(c + 2)2 (c + 3)2 (c + 4)
Q
aa

Now putting c = −1 we get

Sl

x3
y = kx−1 [3x + 3x2 + ]
2
.S
Q

all subsequent terms are zero, since the numerators all contain a factor (c+1).
aa
Sl

Thus
x2
y = v = [3 + 3x + ]
2
.S

is a solution.
Q

A solution is also given by

aa

∂y
Sl

=0
∂c
So, starting from

c−2 (c − 1)(c − 2) 2 c(c − 1)(c − 2)

y = kxc [(c + 1) − x+ 2
x − 2 2
x3 + ....]
c+2 (c + 2) (c + 3) (c + 2) (c + 3) (c + 4)

we have
∂y c−2 (c − 1)(c − 2) 2 c(c − 1)(c − 2)
= kxc lnx((c + 1) − x+ 2
x − x3 + ....)
∂c c+2 (c + 2) (c + 3) (c + 2)2 (c + 3)2 (c + 4)
∂ c−2 (c − 1)(c − 2) 2
+ kxc ((c + 1) − x+ x − ....)
∂c c+2 (c + 2)2 (c + 3)

10
We now have to determine the partial derivative of each term:
∂
(c + 1) = 1
∂c
∂ c−2 4
( )=
∂c c + 2 (c + 2)2
(c−1)(c−2) (c−1)(c−2)
now we have to differentiate (c+2)2 (c+3)
, Let t = (c+2)2 (c+3)
it follows that

lnt = ln(c − 1) + ln(c − 2) − 2ln(c + 2) − ln(c + 3)

thus
∂t (c − 1)(c − 2) 1 1 2 1

.S
= ( + − − )

Q
∂c 2
(c + 2) (c + 3) c − 1 c − 2 c + 2 c + 3

aa
Sl
when c = −1,
∂
(c + 1) = 1
.S

∂c
Q

∂ c−2
( )=4
aa

∂c c + 2
Sl

∂ (c − 1)(c − 2)
( )) = −10
∂c (c + 2)2 (c + 3)
.S

Therefore, when c = −1:

Q
aa
Sl

∂y x3
= kx−1 lnx(0 + 3x + 3x2 + + ....) + kx−1 (1 − 4x − 10x2 + ...)
∂c 2
.S

thus another solution is

Q

x2
aa

y = w = C(lnx(3 + 3x + + ...) + x−1 (1 − 4x − 10x2 + ...))

Sl

2
Now we have a problem, for we seem to have three separate series solutions
for second-order differential equations

x2
(a) y = u = A(1 + x + 6
).
x2
(b) y = v = B(3 + 3x + 2
).
x2
(c) y = w = C(lnx(3 + 3x + 2
+ ...) + x−1 (1 − 4x − 10x2 + ...)).

11
But (b) is clear a simple multiple of (a) and thus not distinct solution. So
finally, we have just (a) and (c). The complete solution is then y = u + w.
In general if c1 − c2 = n, where n is a non-zero integer the solution is of the
form:

y = (1 + klnx)xc1 (a0 + a1 x + a2 x2 + ...) + xc2 (b0 + b1 x + b2 x2 + ...)

EXERCISE

A:Use the power series method to solve the following differential equations:

.S
Q
1. y 0 + 3xy = 0.

aa
Sl
2. y 0 − 2xy = 0.

3. y 00 − xy 0 − y = 0
.S
Q

4. y 00 + xy 0 + y = 0
aa
Sl

B: Using the Frobenius method to solve the following differential equations:

.S
Q

1. xy 00 + (2 + x)y 0 − 2y = 0.
aa
Sl

2. xy 00 + y 0 − xy = 0.

3. 4xy 00 + y 0 + y = 0.
.S
Q

4. 6x2 y 00 − xy 0 + y − xy = 0 .
aa
Sl

12
 CHAPTER TWO
PARTIAL DIFFERENTIAL EQUATIONS

2.1 Partial Differential Equations

Definition 2.1. A partial Differential Equation is a relationship between
a dependent variable u and two or more independent variables (x, y, t, ....)
and partial differential coefficients of u with respect to those independent
variables. The solution is therefore, u = f (x, y, t, ...).

.S
Q
2.1.1 Solution by Direct Integration

aa
Sl
The simplest form of partial differential equation is such that a solution can
be determined by direct partial integration.
.S

∂2u
Example 2.7. Solve the equation = 12x2 (t + 1), given that at x = 0,
Q

∂x2
aa

u = cos2t,and ∂u = sint.
Sl

∂x

Solution: Integrating partially with respect to x, we have

.S

∂u
Q

= 4x3 (t + 1) + φ(t)
aa

∂x
Sl

where the arbitrary function φ(t) takes the place of normal arbitrary constant
in ordinary integration. Integrating partially again with respect to x, gives
.S

u = x4 (t + 1) + xφ(t) + θ(t)
Q
aa
Sl

where θ(t) is the second arbitrary function. To find the two arbitrary func-
tions, we apply the given initial conditions that at x = 0, ∂u
∂x
= sint and
∂u
u = cos2t. Thus at x = 0, cos2t = θ(t) and ∂x = φ(t). It follows that
u = x4 (t + 1) + xsint + cos2t.
∂2u
Example 2.8. Solve the equation ∂x∂y
= sin(x + y), given that at y = 0,
∂u
∂x
= 1 and at x = 0, u = (y − 1)2 .

∂2u ∂2u
Solution: ∂x∂y
= ∂y∂x
. Thus,

∂ ∂u
( ) = sin(x + y)
∂y ∂x

13
Integrating partially with respect to y we get
∂u
= −cos(x + y) + φ(x)
∂x
∂u
but ∂x
= 1 when y = 0,this implies that φ(x) = 1 + cosx. Therefore,
∂u
= −cos(x + y) + cosx + 1
∂x
Integrating again partially with respect to x, we have
u = −sin(x + y) + sinx + x + θ(y)
but at x = 0, u = (y − 1)2 these initial conditions will lead us to the solution

.S
u = −sin(x + y) + x + sinx + siny + (y − 1)2

Q
aa
Sl
2.1.2 Initial Conditions and Boundary Conditions

As with any Differential equation, the arbitrary constants or arbitrary func-

.S
Q

tions in any particular case are determined from additional information given
aa

concerning the variables of the equation. These extra facts are called initial
Sl

conditions or more generally boundary conditions since they do not always

refer to zero values of the independent variables.
2
.S

∂ u
Example 2.9. Solve the equation ∂x∂y = sinxcosy subject to the boundary
Q

π ∂u
conditions that at y = 2 , ∂x = 2x and at x = π, u = 2siny.
aa
Sl

∂2u ∂ ∂u
Solution: ∂x∂y
= ( )
∂y ∂x
this implies that
.S

∂ ∂u
( ) = sinxcosy
Q

∂y ∂x
aa
Sl

Integrating partially with respect to y we get

∂u
= sinysinx + φ(x)
∂x
and when y = π2 , ∂u
∂x
= 2x this gives φ(x) = 2x − sinx thus
∂u
= sinysinx + 2x − sinx
∂x
Integrating partially again with respect to x we have u = x2 + cosx +
sinycosx + θ(y), but when x = π , u = 2siny substituting these conditions
to equation we get the solution
u = x2 + (1 − siny)cosx + siny + 1 − π 2

14
2.2 One Dimensional Heat Equation
We assume that temperature is along thin bar or wire of constant cross-
section and homogeneous material which is oriented along the x-axis and is
perfectly insulated laterally, so that heat flows in the x-direction only.
Thus one dimensional heat equation is the equation of the form
∂u ∂ 2u
= c2 2 (2.1)
∂t ∂x
At the end point x = 0 and x = L of the bar are kept at temperature zero;
so that we have the boundary conditions

.S
u(0, t) = 0 (2.2)

Q
aa
Sl
and
u(L, t) = 0 (2.3)
∀t and the initial temperature in the bar at time t = 0 is f (x), so that we
.S

have initial condition

Q
aa

u(x, 0) = f (x) (2.4)

Sl

We shall determine a solution u(x, t) of 2.1 satisfy 2.2 and 2.3 and one ini-
tial condition. We can now formulate two ordinary differential equations as
follows: Let
.S
Q

u(x, t) = F (x)G(t) (2.5)

aa

d2 F
Sl

F G0 = c2 F 00 G with G0 = dG
dt
and F 00 = dx2
. By the method of separation of
variables we have
G0 F 00
= (2.6)
.S

c2 G F
Q

Both sides of 2.6 must be equal to a constant k, for k ≥ 0 the only u = F G

aa

that satisfy 2.2 and 2.3 is u = 0, for negative value of k say k = −p2 we have
Sl

from 2.6
G0 F 00
= = −p2
c2 G F
thus the two ordinary differential equations are

F 00 + p2 F = 0 (2.7)

and
G0 + p2 c2 G = 0 (2.8)
The general solution of 2.7 is

F (x) = Acospx + Bsinpx (2.9)

15
From the boundary condition 2.2 and 2.3 it follows that

u(0, t) = F (0)G(t) = 0

and
u(L, t) = F (L)G(t) = 0
Since G = 0 would give u = 0, we require F (0) = 0, F (L) = 0 and get
F (0) = A = 0 by 2.9 and then F (L) = BsinpL = 0 with B 6= 0(to avoid
F = 0); thus
sinpL = 0
hence
nπ

.S
p=
L

Q
aa
n = 1, 2, 3, 4, ....

Sl
setting B = 1 we thus obtain the following solution of 2.7 satisfying 2.2 and
2.3 as
nπ
Fn (x) = sin x
.S

L
Q

n = 1, 2, 3, ....
aa

Again we solve 2.8 for p = nπ as follows: let λn = cnπ the 2.8 has the general
Sl

L L
solution
Gn = Bn e− λ2c t (2.10)
.S

where Bn is a constant.
Q

Hence the functions

aa
Sl

nπ −λ2c t cnπ
un (x, t) = Fn (x)Gn (t) = Bn sin xe ; (λn = ) (2.11)
L L
.S

To obtain the solution also satisfying the initial condition 2.4 we consider a
Q

series of these eigenfunctions;

aa
Sl

∞ ∞
X X nπ −λ2n t
u(x, t) = un (x, t) = Bn sin xe (2.12)
n=1 n=1
L

From this and 2.4 we have

∞
X nπ
u(x, 0) = Bn sin x = f (x)
n=1
L

Hence for 2.12 to satisfy 2.4 the Bn0 s must be the coefficients of the Fourier
sine series; Thus
2 L
Z
nπ
Bn = f (x)sin xdx
L 0 L

16
2.3 One-Dimensional Wave Equation
We consider a vibrating string and make the following assumptions

1. The mass of the string per unit length is constant(homogeneous string).

The string is perfectly elastic and does not offer any resistance to bend-
ing.

2. The tension caused by stretching the string before fixing it at the ends
is so large that the action of gravitational force on the string can be
neglected.

.S
Q
3. The sting performs small transverse motion in a vertical plane; that

aa
Sl
is, every particle of string moves strictly vertically and so that the
deflection and the slope at every point of the string always remain in
absolute value.
.S
Q
aa

One dimensional wave equation is the equation of the form

Sl

∂ 2u 2
2∂ u
= c (2.13)
∂t2 ∂x2
.S
Q

where u(x, t) is the deflection of the string. To find out how the string moves,
aa

we solve this equation; Since the string is fixed at the ends x = 0 and x = L,
Sl

we have the two boundary conditions

u(0, t) = 0, and u(L, t) = 0, ∀t (2.14)

.S
Q

The form of the motion of the string will depend on the initial deflection
aa

(deflection at t = 0)and on the initial velocity(velocity at t = 0). Denoting

Sl

the initial deflection by f (x) and initial velocity by g(x), we thus obtain two
initial conditions:

u(x, 0) = f (x) (2.15)

∂u
|t=0 = g(x) (2.16)
∂t
We need to find the solution of (2.11) satisfying conditions (2.12) to (2.14)
In the method of separation of variables or product method, we determine
solutions of the wave equation (2.11) of the form:

u(x, t) = F (x)G(t) (2.17)

17
This equation is the product of two functions, each depending on one of the
variables x and t. Differentiating equation above we obtain
∂2u 2
∂t2
= F G00 and ∂∂xu2 = F 00 G
Inserting this into our differential equation (2.11) we have:
00 00
F G00 = c2 F 00 G, separating the variables we get cG2 G = FF . We claim that
both sides must be constant
G00 00
c2 G
= FF = k. This yields immediately two ordinary linear differential
equations, namely
F 00 − kF = 0 (2.18)
and
G00 − c2 kG = 0 (2.19)

.S
Here the constant k is still arbitrary. We shall now determine solutions F

Q
aa
and G of (2.16) and (2.17) so that u = F G satisfies the boundary conditions

Sl
(2.12) that is u(0, t) = F (0)G(t) = 0, u(L, t) = F (L)G(t) = 0 for all t.
Solving (2.16) if F = 0, then u = 0, which is of no interest. Thus,F 6= 0 and
then
.S
Q
aa

F (0) = 0 and
Sl

F(L)=0(2.20)For k = 0, the general solution of (2.16) is F = ax + b and from

(2.18) we obtain a = b = 0. Hence F = 0; which is of no interest because then
u = 0. For positive k = µ2 the general solution of (2.16) is F = Aeµx +Be− µx
.S
Q

and from (2.18) we obtain F = 0, as before. Hence we are left with the
aa

possibility of choosing k negative, say k = −p2 . Then, (2.16) takes the form:
Sl

F 00 + p2 F = 0. Its general solution is F (x) = Acospx + Bsinpx, from this

and (2.18) we have: F (0) = A = 0 and then F (L) = BsinpL = 0. We must
take B 6= 0 since otherwise F = 0. Here sinpL = 0. Thus,
.S
Q
aa

pL = nπ, sothat,
Sl

p=nπ L (n integer) (2.21)

Setting B = 1, we thus obtain infinitely many solutions F (x) = Fn (x) where

nπ
Fn (x) = sin x,
L
(n=1,2,.....)(2.22)These solution satisfy (2.18). Solving (2.17) the constant
k is now restricted to value k = −p2 = −( nπ L
). For this k equation (2.17)
becomes
G00 + λ2n G = 0 where λn = cnπL
. A general solution is Gn (t) = Bn cosλn t +

18
Bn∗ sinλn t. Hence solution of (2.11) satisfying (2.12) are un (x, t) = Fn (x)Gn (t);
written out
nπ
un (x, t) = (Bn cosλn t + Bn∗ sinλn t)sin x (2.23)
L
Since equation (2.11) is linear and homogeneous, it follows that the sum of
infinitely many solutions un is a solution of (2.11). To obtain the solution
that satisfies (2.13) and (2.14), we consider the infinite series (with λn = cnπ L
as before)
∞ ∞
X X nπ
u(x, t) = un (x, t) = (Bn cosλn t + Bn∗ sinλn t)sin x (2.24)
n=1 n=1
L

.S
satisfying initial condition (2.13) (given initial displacement), from (2.22)

Q
and (2.13) we obtain

aa
Sl
∞
X nπ
u(0, t) = Bn sin x = f (x) (2.25)
n=1
L
.S
Q

Hence we must choose Bn0 s so that u(x, 0) become the Fourier series of f (x).
aa
Sl

Thus
2 L
Z
nπ
Bn = f (x)sin xdx,
L 0 L
.S

n=1,2,....(2.26)To satisfy initial condition (2.14) (given initial velocity) we

Q

differentiate P
(2.22) with respect to t to get
aa
Sl

∂u
| = 0 = ∞
∂t t
∗ nπ ∗0
n=1 Bn λn sin L x = g(x). Here we must choose Bn s so that
∂u
for t = 0 the derivative ∂t become the Fourier sine series of g(x). Thus
RL
Bn∗ λn = L2 0 g(x)sin nπ xdx, since λn = cnπ we obtain
.S

L L
Q

Z L
2 nπ
aa

Bn∗ = g(x)sin xdx,

Sl

cnπ 0 L

n=1,2.....(2.27)Therefore,
u(x, t) = ∞ un (x, t) = ∞ ∗ nπ
P P
n=1 n=1 (Bn cosλn t + Bn sinλn t)sin L
x, where
2 L ∗ L
Bn = L 0 f (x)sin L xdx, n = 1, 2, .. and Bn = cnπ 0 g(x)sin nπ
nπ 2
R R
L
xdx, n =
1, 2.....

EXERCISE

1. Find u(x, t) of the string of length L = π when c2 = 1, the initial

velocity is zero, and the initial deflection is k(sinx − 12 sin2x).

19
Find solutions u(x, y) of the following equations by separating variables:

2. ux + uy = 0.
3. ux − uy = 0.
4. y 2 ux − xuy = 0
5. ux + uy = (x + y)u
6. uxx + uyy = 0
7. uxy − u = 0

.S
Q
8. uxx − uyy = 0

aa
Sl
9. xuxy + 2yu = 0 .S

2.4 Two-Dimensional Wave Equation

Q
aa
Sl

In this case we consider the motion of a stretched elastic membrane, such

as the drumhead. This is the two-dimensional analog of the vibrating string
problem.
.S

Physical Assumption:We assume the following:

Q
aa
Sl

1. The mass of the membrane per unit area is constant(homogeneous

membrane).The membrane is perfectly flexible and offer no resistance
to bending.
.S
Q

2. The membrane is stretched and then fixed along its entire boundary
aa

in the xy-plane.The tension per unit length T caused by stretching the

Sl

membrane is the same at all points and in all directions and does not
change during the motion.
4. The deflection u(x, y, t) of the membrane during the motion is small
compared to the size of the membrane, and all angles of inclination are
small.

To solve the problem of a vibrating membrane, we have to determine a solu-

tion u(x, y, t) of the two-dimensional wave equation
∂ 2u 2
2 ∂ u ∂ 2u
= c ( + ),
∂t2 ∂x2 ∂y 2

20
 T
c2 = ρ
(2.28)

that satisfy the boundary condition

u(0, 0, t) = 0 (2.29)

on the boundary of the membrane for all t ≥ 0, and the two initial conditions

u(x, y, 0) = f (x, y) (2.30)

and
∂u
|t=0 = g(x, y) (2.31)
∂t

.S
Q
where f (x, y) and g(x, y) are given initial displacement and velocity respectively.u(x, y, t)

aa
gives the displacement of the point (x, y) of the membrane from rest (u = 0)

Sl
at time t.We see that the conditions (2.27) − (2.29) are similar to those for
the vibrating string.As a first important case, let us consider the rectangular
membrane R shown in figure () By applying the method of separation of
.S
Q
aa

Figure 2.1: The rectangular membrane

Sl

variables we first determine the solution of (2.26) that satisfy the boundary
condition (2.27). We start from
.S
Q
aa

u(x, y, t) = F (x, y)G(t). (2.32)

Sl

By substituting this into the wave equation 2.26 we have

F G00 = c2 (Fxx G + Fyy G) (2.33)

.S
Q
aa

where subscripts denote partial derivatives and dash denote derivative w.r.t
Sl

00
t. To separate the variables, we divide both sides by c2 F G: cG2 G = F1 (Fxx +
Fyy ).Since the left side depends only on t, whereas the right side is indepen-
dent of t, both sides must equal a constant. By a little investigation we see
that only negative values of that constant will lead to solutions that satisfy
2.27 without being identically zero. Denoting that negative constant by −ν 2 ,
00
we have cG2 G = F1 (Fxx + Fyy ) = −ν 2 . This gives two equations for the ”time
function” G(t) we have the ordinary differential equation

G00 + λ2 G (2.34)

and for the ”amplitude function” F(x,y) the partial differential equation

Fxx + Fyy + ν 2 F = 0, (2.35)

21
known as the two-dimensional Helholtz equation.Separation of the Helmholtz
equation is achieved if we set
F (x, y) = H(x)Q(y). (2.36)
d2 H d2 Q
Substitution of this into (2.33) gives dx2
= −(H dy2 +ν 2 HQ).To separate the
2 2
variables, we divide both sides by HQ, finding H1 ddxH2 = − Q1 ( ddyQ2 +ν 2 Q). Both
sides must equal a constant, by the usual argument. This constant must be
negative, say, −k 2 , because only negative values will lead to solutions that
2 2
satisfy (2.27) without being identically zero. Thus H1 ddxH2 = − Q1 ( ddyQ2 +ν 2 Q) =
−k 2 .This yields two ordinary differential equations for H and Q:
d2 H

.S
2
+ k 2 H = 0. (2.37)
dx

Q
aa
and

Sl
d2 Q
+ p2 Q = 0, where
dy 2
p2 = ν 2 − k 2 . (2.38)
.S
Q
aa

The general solution of (2.35) and (2.36) are H(x) = Acoskx + Bsinkx
Sl

and Q(y) = Ccospy + Dsinpy, where A,B,C and D are constants. From
(2.30) and (2.27) it follows that the function F = HQ must be zero on the
.S

boundary, which corresponds to x = 0, x = a, and y = b, this means that

Q

H(0) = 0, H(a) = 0, Q(0) = 0, Q(b) = 0. Therefore, H(0) = A = 0, and

aa

then H(a) = Bsinka = 0.We must take B 6= 0, since otherwise H = 0 and

Sl

F = 0.Hence sinka = mπ, k = mπ a

, m is an integer. In precisely the same
fashion we conclude that C = 0 and p must be restricted to the value p = nπ b
where n is an integer.We thus obtain the solutions Hm (x) = sin mπx
.S

a
and
Q

Qn (y) = sin nπy

b
, m = 1, 2, 3, ...., n = 1, 2, 3, .....
aa

As for the case of the vibrating string, it is not necessary to consider m,n =
Sl

−1, −2, .... since the corresponding solutions are essentially the same as for
positive m and n, except for factor −1. Hence the functions
mπx nπy
Fmn (x, y) = Hm (x)Qn (y) = sin sin ,
a b
m=1,2,3,...andn=1,2,3,....(2.39)are the solutions of (2.35)that are zero on
the boundary of the membrane.
Eigenfunctions and Eigenvaluesp Since p2 = ν 2 − k 2 in (2.36) and λ = cν
in (2.32), we have λ = c k + p . Hence to k = mπ
2 2
a
and p = nπ
b
there
corresponds the value
r
m 2 n2
λ = λmn = cπ + 2 , m = 1, 2, 3, ...andn = 1, 2, 3, ... (2.40)
a2 b

22
in the differential equation (2.32).The corresponding general solution of (2.32)
?
is Gmn (t) = Bmn cosλmn t+Bmn sinλmn t.It follows that the functions umn (x, y, t) =
Fmn (x, y)Gmn (t), written out
? mπx nπy
umn (x, y, t) = (Bmn cosλmn t + Bmn sinλmn t)sin sin (2.41)
a b
with λmn according to (2.38), are the solutions of the wave equation (2.26)
that are zero on the boundary of the rectangular membrane. These functions
are called the Eigenfunctions or characteristic functions, and the numbers
λmn are called the eigenvalues or characteristic values of the vibrating mem-
brane. The frequency of umn is λ2πmn
.
NOTE: It is very interesting that, depending on a and b, several functions

.S
Fmn may correspond to the same eigenvalue. Physically this means that

Q
aa
there may exist vibrations having the same frequency but entirely different

Sl
nodal lines(curves of points on the membrane that do not move).
Example 2.10. Consider the square membrane for which a = b = 1.
.S
Q

From (2.38) we obtain the eigenvalues

aa
Sl

√
λmn = cπ m2 + n2 (2.42)

Hence λmn = λnm but for m 6= n the corresponding functions Fmn =

.S
Q

sinmπxsinnπy and Fnm =√sinnπxsinmπy are certainly different. For ex-

aa

ample, toλ12 = λ21 = cπ 5 there correspond the two functions F12 =

Sl

sinπxsin2πy and F21 = sin2πxsinπy. To obtain the solution that also sat-
isfy the initial conditions (2.28) and (2.29) we consider the double series
.S

∞ X
∞ ∞ X ∞
mπx nπy
Q

X X
?
u(x, y, t) = umn (x, y, t) = (Bmn cosλmn t+Bmn sinλmn t)sin sin
aa

m=1 n=1 m=1 n=1

a b
Sl

(2.43)
From this and (2.28) we obtain
∞ X
∞ ∞ X
∞
X X mπx nπy
(x, y, 0) = umn (x, y, t) = sin
Bmn sin = f (x, y).
m=1 n=1 m=1 n=1
a b
(2.44)
This is called a double Fourier series.Suppose that f (x, y) can be developed
in such a series. Then the Fourier coefficient Bmn of f (x, y) in (2.41) may be
determined as follows.Setting
∞
X nπy
Km (y) = Bmn sin (2.45)
n=1
b

23
we can write (2.41) in the form f (x, y) = ∞ mπx
P
m=1 Km (y)sin a . For fixed y
this is the Fourier sine series of f (x, y), considered as a function of x. From
the concepts of Fourier series we see that the coefficients of this expansion
are
2 a
Z
mπx
Km (y) = f (x, y)sin dx (2.46)
a 0 a
Furthermore, (2.42) is the Fourier sine series of of Km (y), and from Fourier
Rb
series concepts it follows that the coefficients are Bmn = 2b 0 Km (y)sin nπy
b
dy.
From this and (2.43) we obtain the generalized Euler formula
Z bZ a
4 mπx nπy
Bmn = f (x, y)sin sin dxdy, m = 1, 2..., n = 1, 2, 3, ...

.S
ab 0 0 a b

Q
(2.47)

aa
for the Fourier coefficients of f (x, y) in the double Fourier series in (2.41).

Sl
∗
Bmn is obtained
P∞ Pby differentiating (2.40) termwise with respect to t, that is
∞
∂u
|
∂t t=0
= m=1 n=1 (Bmn ∗
λmn sin mπx
a
sin nπy
b
= g(x, y). Suppose that g(x, y)
can be developed in this double Fourier series. Then, proceeding as before,
.S
Q

we may find that the coefficients are

aa
Sl

Z bZ a
∗ 4 mπx nπy
Bmn = g(x, y)sin sin dxdy,
abλmn 0 0 a b
.S

m=1,2...,n=1,2,3,...(2.48)
Q
aa
Sl

EXERCISE
.S

1. Find the vibrations of a rectangular membrane of sides a = 4f t, and b =

Q

2f t. If the tension is 12.5lb/f t, the density is 2.5slugs/f t2 , the initial

aa
Sl

velocity is 0, and initial displacement is f (x, y) = 0.1(4x − x2 )(2y −

y 2 )f t.

2. Find u(x, y, t) for the rectangular membrane with sides a and b and
c = 1 if the initial velocity is zero and the initial displacement is
sin 2πx
a
sin 3πy
b

3. Find the deflection u(x, y, t) of the square membrane with a = b = 1

and c = 1 if the initial velocity is zero and the initial deflection is (a)
0.1sin3πxsin4πy (b)kxy(1 − x)(1 − y)

24
 CHAPTER THREE
COMPLEX VARIABLES

3.1 COMPLEX NUMBERS

Definition 3.2. A complex number z is an ordered pair (x, y) of real numbers
x and y, written z = (x, y). x is called the real part and y is the imaginary
part of z, written as x = Rez and y = Imz. We need to note that two
complex numbers are equal if and only if their real parts are equal and their
imaginary parts are equal.(0, 1) is called imaginary unit and is denoted by i.

.S
Q
aa
3.1.1 ADDITION AND SUBTRACTION OF COMPLEX NUM-

Sl
BERS .S

If z1 = (x1 , y1 ) and z2 = (x2 , y2 ) then z1 ± z2 = (x1 ± x2 , y1 ± y2 )

Q
aa
Sl

3.1.2 MULTIPLICATION

If z1 = (x1 , y1 ) and z2 = (x2 , y2 ) then z1 .z2 = (x1 , y1 )(x2 , y2 ) = (x1 x2 −

.S

y1 y2 , x1 y2 + x2 y1 ). We need to note that if x = 0, then z = iy and is called

Q

purely imaginary and i2 = −1.

aa
Sl

3.1.3 DIVISION OF COMPLEX NUMBER

.S
Q

If z1 = (x1 , y1 ) and z2 = (x2 , y2 ) then zz21 = zz12 zz22 . The conjugate of a complex
aa

number z = x + iy is defined by z = x − iy
Sl

3.1.4 POLAR FORM OF COMPLEX NUMBERS

If x = rcosθ and y = rsinθ, then z = x + iy = r(cosθ + isinθ) p

r is called the modulus or absolute value of z defined by |z| = r = x2 + y 2 ,
θ is called the argument of z and is denoted by argz and θ = tan− 1( xy ). The
value of θ between −π ≤ θ ≤ π is called the principal value of the argument
of z.

25
3.1.5 MULTIPLICATION IN POLAR FORM

Let z1 = r1 (cosθ1 + isinθ1 ) and z2 = r2 (cosθ2 + isinθ2 ). Then z1 z2 = |z1 ||z2 |.

Arg(z1 z2 ) = arg(z1 ) + arg(z2 ).

3.1.6 DIVISION IN POLAR FORM

The quotient z = zz12 is the number z satisfying zz2 = z1 . Hence |zz2 | =

|z1 |
|z||z2 | = |z1 |, arg(zz2 ) = arg(z) + arg(z2 ) = arg(z1 ), this gives | zz21 | = |z 2|
and arg( zz21 ) = arg(z1 ) − arg(z2 ).

.S
Q
3.1.7 De Moivre’s Theorem

a
a
Sl
z n = rn (cosnθ + isinnθ). If |z| = r = 1 we have (cosθ + isinθ)n = cosnθ +
isinnθ
.S
Q

3.1.8 ROOTS
aa
Sl

If z = wn (n = 1, 2, 3, ...), then to each value of w there corresponds one

value of z. We shall see that for z 6= 0 there corresponds precisely n distinct
.S

1
values of w. Each of these values nth root of z, and we write w = z n .
Q

Let z = r(cosθ + isinθ) and w = R(cosφ + isinφ), wn = z becomes wn =

a
a

Rn (cosφ + isinφ) = z = r(cosθ + isinθ), by equating the absolute values

Sl

1
on both sides we have Rn = r thus R = r n .By equating the arguments
nφ = θ + 2kπ ; thus φ = nθ + 2π n
, where k is an integer. For k = 0, 1, 2, ...n − 1
.S

we get n distinct values of w. Further integers of k would give values already

Q

obtained.
aa

1 1
Therefore, z n = r n (cos θ+2kπ + isin θ+2kπ ), k = 0, 1, 2, ....n − 1.
Sl

n n

3.2 COMPLEX FUNCTION

A function f defined on a set S of complex number is a rule that assign
to every z in S a complex number w, called the value of f at z. We write
w = f (z), here z varies in S and is called a complex variable.The set S is
called the domain of definition of f .
Example 3.11. w = f (z) = z 2 + 3z is a complex function defined for all
z; that is domain S is the whole complex plane. The set of all values of a
function f is called the range of f . The set of all values of a function f is

26
called the range of f . w is complex, and we write w = u + iv, w depends on
z = x + iy. Thus w = f (z) = u(x, y) + iv(x, y).

Example 3.12. Let w = f (z) = z 2 + 3z, find u and v and calculate the value
of f at z = 1 + 3i.

Solution:
u = Ref (z) = x2 − y 2 + 3x and v = 2xy + 3y.
f (1 + 3i) = −5 + 15i

Example 3.13. Let w = f (z) = 2iz + 6z. Find u and v and the value of f
at z = 21 + 4i.

.S
Q
a
a
Sl
3.2.1 Limits and Continuity

A function f (z) is said to have the limit L as z approaches a point z0 and if

.S

the values of f are close to L, for all z close to z0 that is limz→z0 f (z) = L.
Q

A function f (z) is said to be continuous at z = z0 if f (z0 ) is defined and

aa
Sl

limz→z0 f (z) = f (z0 ).

3.2.2 Derivative
.S
Q

The derivative of a complex function f at a point z0 is written f 0 (z0 ) and

a
a
Sl

is defined by : f 0 (z0 ) = limδz→0 f (z0 +δz)−f

δz
(z0 )
provided that the limit exist.
Then f is said to be differentiable at z0 .
.S
Q

Example 3.14. The function f (z) = z 2 is differentiable for all z and f 0 (z) =
aa

2z.
Sl

Example 3.15. The function f (z) = z is not differentiable.

Solution: It is important to note that there are many simple functions that
do not have a derivative at any point. For instance, f (z) = z = x − iy is
such a function. Indeed if we write δz = δx + iδy, we have: f (z+δz)−f
δz
(z)
=
(z+δz)−z δx−iδy
δz
= δz
δz
= δx+iδy , if δy = 0 this limit is 1, if δx = 0 this limit is −1.
Thus the limit does not exist at any z.

27
3.2.3 Analytic Functions

These are the functions that are differentiable in some domain, so that we
can do calculus in complex numbers.

Definition 3.3. A function f (z) is said to be analytic in a domain D if f (z)

is defined and differentiable at all points of D. The function f (z) is said to
be analytic at point z = z0 in D if f (z) is analytic in a neighborhood of z0 .

Example of these functions is a polynomial function f (z) = c0 + c1 z + c2 z 2 +

... + cn z n .

.S
Q
a
a
EXERCISE

Sl
1. Find the value of Ref and Imf at indicated point. (a) f = z 2 + 2z + 2
.S

1
at 1 − i (b) f = 1−z at 7 + 2i.
Q
aa
Sl

2. Find the value of the derivatives of the following functions:

(a) z−i
z+i
at i (b) (z − 4i)8 at 5 + 4i (c) 5+3i
z3
at 2 + i (d) (3z 2 + iz)2 at
1 + i.
.S
Q
a
a
Sl

NOTE: The differentiation rules are the same as in real calculus, since their
proofs are literally the same.
.S

3.2.4 Cauchy-Riemann Equations

Q
aa
Sl

These provide a criterion (test) for the analyticity of a complex function

W = f (z) = u(x, y) + iv(x, y).
The function f (z) is analytic in a domain D iff the first partial derivatives
of u and v satisfy the two so-caled Cauchy-Riemann equations:

ux = vy (3.1)

and

28
uy = −vx (3.2)

∂u ∂u
everywhere in the domain D. Here ux = ∂x
and uy = ∂y
.
Theorem 1. Let f (z) = u(x, y) + iv(x, y) be defined and continuous in some
neighborhood of a point z = x + iy and differentiable at z itself, then at
that point, the first order partial derivatives of u and v exist and satisfy the
Cauchy-Riemann equations.

Hence if f (z) is analytic in a domain D, those partial derivatives exist and

satisfy Cauchy-Riemann equations at all points of D.

.S
Theorem 2. If two real-valued continuous function u(x, y) and v(x, y) of two

Q
real variables x and y have continuous first partial derivative that satisfy the

aa
Sl
Cauchy-Riemann equations in some domain D, then the complex function
f (z) = u(x, y) + iv(x, y) is analytic in D.
.S

If we use the polar form of a complex number z = r(cosθ + isinθ) and set
Q
aa

f (z) = u(r, θ) + iv(r, θ), then the Cauchy-Riemann equations becomes:

Sl

ur = 1r vθ (3.3)
.S
Q
aa

and
Sl

vr = − 1r uθ (3.4)
.S
Q
aa
Sl

EXERCISE

1. Show that (a) f (z) = z 2 is analytic for all z (b) f (z) = z ? ,where z ? is
the conjugate of z is not analytic for all z.

2. Derive the Cauchy-Riemann equation in polar form from the algebraic

form of them.
3. Check for the analyticity of the following complex functions:
(a) f (z) = z 6 (b) f (z) = ex (cosy + isiny) (c) f (z) = zz (d) f (z) =
Rez
lnz + iargz (e) f (z) = Imz (f) f (z) = z 3 .

29
3.3 Complex Function and Different Functions
3.3.1 Exponential function

By definition ez = ex (cosy + isiny). This definition is motivated by the re-

quirements that makes ez a natural extension of the real exponential function
ex , namely:

1. ez should reduce to ex when z = x is real

2. ez should be an entire function, that is analytic for all z.

.S
3. Similar to calculus, its derivative should be (ez )0 = ez .

Q
aa
Sl
Properties: ez1 +z2 = ez1 ez2 . If z1 = x, z2 = iy we have ez = ex ei y. If z = iy
we have the so-called Euler’s formula ei y = cosy + isiny.
Hence the polar form of a complex function z = r(cosθ + isinθ) may now be
.S

written as z = rei θ .
Q
aa

NOTE:
Sl

1. e2πi = 1
.S

πi −πi
2. e 2 = i and e = −i.
Q

2
aa

3. eπi = −1 and e−πi = −1.

Sl

4. |eiy | = 1.
.S

5. arg(ez ) = y ± 2nπ , n = 0, 1, 2, 3, ....

Q
aa

6. ez is periodic with period 2πi, that is ez+2πi = ez .

Sl

3.3.2 Trigonometric and Hyperbolic Functions

Given that, eix = cosx + isinx and e−ix = cosx − isinx, by addition and
subtraction we obtain for real cosine and sine;
cosx = 21 (eix + e−ix ) and sinx = 2i1 (eix − e−ix ), this suggest the following
definitions for complex values z = x + iy
cosz = 12 (eiz + e−iz ) and sinz = 2i1 (eiz − e−iz )
For Hyperbolic function coshz = 12 (ez + e−z ) and sinhz = 12 (ez − e−z ). It
follows that coshiz = cosz and sinhiz = isinz, cosiz = coshz and siniz =
isinhz.

30
3.3.3 Logarithmic Function

Given ew = eu+iv = reiθ then, lnz = lnr + iθ. It follows that

1. lnz = lnz ± 2nπi

2. ln1 = 0, ±2πi, ±, ±4πi, ....

3. ln4 = 1.386294 ± 2nπi.

4. ln(−1) = ±πi, ±3πi, ±5πi, ...

.S
Q
5. ln(−4) = 1.386 ± ±(2n + 1)πi and ln4i = 1.386 + πi ± 2nπi, ln(−4i) =

a
2

a
Sl
πi
1.386 − 2 ± 2nπi

6. lni = π2 i, 3π
2
i, 5π
2
i, ...
.S

7. ln(3 − 4i) = ln5 + iarg(3 − 4i)

Q
aa
Sl

EXERCISE
.S

1. Verify that e1.4−0.6i = 3.347 − 2.290i.

Q
aa
Sl

2. Solve ez = 3 + 4i

3. Show that (a) cosz = cosxcoshy − isinxsinhy, (b) sinz = sinxcoshy +

.S

icosxsinhy,
Q

(c) |cosz|2 = cos2 x + sinh2 y, (d) |sinz|2 = sin2 x + sinhy

aa
Sl

4 Find ez (in terms of u + iv) and |ez | if z equals (a) 2 + 3πi (b) 2π(1 + i),
(c) −πi
2
.

5. Show that cosh2 z − sinh2 = 1 and cosh2 z + sinh2 = cosh2z

3.4 MAPPING OF COMPLEX NUMBERS

For a complex function w = f (z) = u(x, y) + iv(x, y), z = x + iy we need two
planes , the z-plane in which we plot values of z and the w-plane in which
we plot the corresponding function value w = f (z).We say that, f defines
mapping of D into the w-plane.

31
Example 3.16. Mapping w = z 2
In polar coordinates we have z = r(cosθ + isinθ), w = R(cosφ + isinφ) =
z 2 = r2 (cos2θ + isin2θ). Comparing the moduli and arguments gives R =
r2 andφ = 2θ, thus r = r0 are mapped onto circles R = r02 and θ = θ0 onto
φ = 2θ0 . Therefore, the region 1 ≤ |z| ≤ 32 , π6 ≤ θ ≤ π3 is mapped onto region
1 ≤ w ≤ 49 , π3 ≤ θ ≤ 2π3
.
In cartesian coordinate we have: z = x + iy and u = Re(z 2 ) = x2 − y 2 ,
v = Im(z 2 ) = 2xy. Hence vertical lines x = c =constant are mapped onto
u = c2 − y 2 , v = 2cy. From this we can eliminate y so that we have v 2 =
4c2 (c2 − u) -(parabolas open to left). Similarly, horizontal lines y = k mapped
onto parabolas opening to right, that is v 2 = 4k 2 (k 2 + u)

.S
Q
3.4.1 Conformal Mapping

aa
Sl
A conformal Mapping is a mapping that preserves angles between any ori-
ented curves both in magnitude and sense. For a curve C in xy-plne a
.S

parametric representation is x = x(t), y = y(t) e.g x = cost, y = sint repre-

Q

sents the circle x2 + y 2 = 1. In the complex plane we can write this as: C:
aa
Sl

z(t) = x(t) + iy(t), z(t) = cost + isint. We assume that C is smooth i.e z(t)
is differentiable and the derivative dz
dt
is continuous and nowhere zero.
Theorem 3. The mapping defined by an analytic function f (z) is conformal,
.S

except at critical points that is, points at which the derivative f 0 (z) is zero.
Q
aa
Sl

Example 3.17. Conformality of w = z n

The mapping w = z n , n = 2, 3, 4, ... is conformal except at z = 0, where
w0 = nz n−1 = 0. For n = 2, at 0 the angles are doubled.The general n
.S

the angles at 0 are multiplied by a factor n under mapping.Hence the sector

Q

0 ≤ θ ≤ πn is mapped by z n onto the upper half plane v ≥ 0.

aa
Sl

Example 3.18. Calculate the points at which the mapping w = z + z1 is not

conformal.

EXERCISE

1. Find all points at which the following mapping are not conformal(a)
w = (z − a)3 , (b) w = z 2 + bz + c.

2. Find the image of the given region under the given mapping w = f (z):
(a) |z| > 1, w = 4z (b) π2 < z < 3π
4
,w = z 2 , (c) 0 < y < 1, w = z 2

32
3.5 COMPLEX INTEGRATION
Integration in the complex plane is important for two reasons:

1. In applications there occur real integrals that can be evaluated by com-

plex integration, where as the usual methods for real integral calculus
fails.

2. Some basic properties of analytic functions can be established by com-

plex integration, but would be difficult to prove by other methods.

.S
3.5.1 Line Integrals

Q
aa
Sl
Complex definite
R integrals are called (complex) Line Integrals. They are
written by C f (z)dz. Here the integrand f (z) is integrated over a given
curve C in the complex plane, called the path of integration.
.S

NOTE:
Q
aa
Sl

1. We may represent C by a parametric representation z(t) = x(t) + iy(t),

a ≤ t ≤ b.
.S

2. The sense of increasing t is called positive sense on C.

Q
aa
Sl

3. C is assumed to be smooth curve ,continuous and non zero derivative

dz
dt
.
.S

Let C be a smooth curve in complex plane and let f (z) be a continuous

Q
aa

function at each point of C. If we partition the interval a ≤ t ≤ b by

Sl

points t0 (= a),t1 ,...,tn−1 ,tn (= b), where t0 < t1 , ..., < tn . To this subdivision
there corresponds a subdivision of C by points z0 ,z1 ,...,zn−1 ,zn (= z), where
zj = z(tj ).On each portion of subdivision of C we choose an arbitrary point
say a point P1 between z0 and z1 (i.e P1 = z(t)) where t satisfy Pn t0 ≤ t ≤ t1 , a
point P2 between z1 and z2 etc. Then we form the sum Sn = m=1 f (Pm )δzm ;
where δzm = zm − zm−1 we do this for each n = 2, 3, ...., in a completely
independent manner, but so that the greatest |δtm | = |tm − tm−1 | approaches
zero as n →. This implies that the greatest |δzm | also approaches zero
because it cannot exceed the length of arc C from zm−1 to zm and the latter
goes to zero since the arc length of the smooth curve C is continuous function
of t. The limit of the sequence of complex numbers S2 , S3 , ... thus obtained
is called Line Integral of f (z) over the oriented curve C. This curve C is

33
called the path of integration.
General Assumption: All paths of integration for complex line integrals
are assumed to be piecewise smooth, that is, they consist of finitely smooth
curves joined end to end.
Basic Properties:

R
1. Linearity-
R Integration
R is a linear operator, C
(k1 f1 (z) + k2 f2 (z))dz =
k1 c f1 (z)dz + k2 c f2 (z)dz

2. Sense reversal inR integration over

R z0the same path, from z0 to z (left) and
z
z to z0 (right): z0 f (z)dz = − z f (z)dz.

.S
R R R
3. Partitioning of path: c f (z)dz = c1 f (z)dz + c2 f (z)dz.

Q
aa
Sl
Theorem 4. Let f (z) be analytic in a simply connected domain D, then
there exists an indefinite integral of f (z) in the domain D,that is, an analytic
function F (z) such that F 0 (z) = f (z) in D, and for all paths in D joining
.S

two
R z points z0 and z1 in D we have:
Q

z 01
f (z)dz = F (z1 ) − F (z0 ).
aa
Sl

R 1+i
Example 3.19. 0 z 2 dz = 31 z 3 |1+i 0 = −2
3
+ 23 i.
R πi
Example 3.20. −πi coszdz = sinz|πi −π = 2sinπi = 2isinhπ.
.S
Q

R 8−3πi z z
4− 3πi 4+ πi
Example 3.21. 8+πi e 2 dz = 2e 2 |8−3π 8+π = 2(e
2 − e 2 ) = 0.
aa
Sl

Ri
Example 3.22. −i dz z
= lnz|i−i = πi.
Theorem 5. Let C be a piecewise smooth path, represented by R z = z(t);
.S
Q

where a ≤ t ≤ b. Let f (z) be continuous function on C. Then C f (z)dz =

aa

Rb
f (z(t))z 0 (t)dt.
Sl

a
1
Example 3.23. Integrate z
around the unit circle.

Solution: We may represent the unit circle in the form: z(t) = cost+isint =
ei t(0 ≤ t ≤ 2π), so that the counterclockwise integration corresponds to an
increase of t from 0 to 2π. By differentiation z 0 = ieit , f (z(t)) = z(t)
1 −it
e we
get:
R dz R 2π −it it R 2π
C z
= 0 e ie dt = i 0 = 2πi (check using z(t) = cost + isint)
Example 3.24. Integral of integer powers: Let f (z) = (z − z0 )m ; where m
is an integer and z0 a constant. Integrate counterclockwise around the circle
C of radius ρ with center at z0 .

34
Solution: We may represent C in the form: z(t) = z0 + ρ(cost + isint) =
zR0 + ρeit (0 ≤ t ≤R2π). Then we have: (z −Rz0 )m = ρm eimt , dz = iρe it
dt, Thus
m 2π m imt it m+1 2π i(m+1)t m+1
R 2π
C
(z−z0 ) dz = 0 ρ e iρe dt = iρ 0
e dt = iρ ( 0 cos(m+
R 2π
1)tdt + i 0 sin(m + 1)tdt).
If m = −1, we have ρm+1 = 1, we thus obtain 2πi. For integer m 6= 1
each of the two integrals is zero because we integrate over the interval of
Rlength 2π,mequal to the period of Rsine and cosine.
m
Hence the result is:
C
(z − z0 ) dz = 2πi, m = −1 or C (z − z0 ) dz = 0, m 6= −1 and inte-
ger.

3.5.2 Dependence in Path

.S
Q
a
If we integrate a given function f (z) from a point z0 to point z1 along different

a
Sl
paths, the integrals will in general have different values. In other words, a
complex line integral depends not only on the end points of the path but in
general also on the path itself.
.S
Q

Example 3.25. Integrate f (z) = Rez = x from 0 to 1 + 2i, along the paths
aa

denoted by (a) C: z(t) = t + 2it (0 ≤ t ≤ 1) (b) C1 and C2 , where C1 :

Sl

z(t) = t, (0 ≤ t ≤ 1), C2 : z(t) = 1 + it, (0 ≤ t ≤ 1)

.S

Solution: (a) Since

R 1C: z(t) = t+2it (0 ≤ t ≤ 1), z 0 (t) = 1+2i and f (z) = t.
Q

Thus C Rezdz = 0 t(1 + 2i)dt = 21 + i.

R
aa
Sl

(b) We have C1 : z(t) = t, (0 ≤ t ≤ 1), so that z 0 (t)

R = 1, C2 : z(t)
R = 1 + it,
0
R
(0 ≤ t ≤ 2), z (t) = i. Therefore, C Rezdz = C1 Rezdz + C1 Rezdz =
R1 R2
0
tdt + 0 idt = 21 + 2i.
.S

NOTE: This result is different from the result in (a).

Q
aa
Sl

3.6 Cauchy’s Integral Theorem

To state Cauchy’s Integral Theorem we need the following two concepts:

1. A simple closed path- is a closed path that does not intersect or touch
itself. eg a circle.

2. A simply connected domain D in the complex plane- is a domain such

that every simple closed path in D encloses only points of D. Example
the interior of a circle (open disk), ellipse or any simple closed curve.

35
Theorem 6. Cauchy’s Integral theorem- If f (z) is analytic
R in a simply con-
nected domain D, then for every simple closed path in D C f (z)dz = 0.

NOTE: A simple closed path is sometimes called contour and integral over
such a path a contour integral.

Example
R z 3.26.
R No singularities(Entire
R n functions)
C
e dz = 0, C
coszdz = 0, C
z dz = 0 for any closed path, since the
functions are entire (analytic for all z).

Example 3.27.R Singularities outside Contour

1
seczdz = 0, C z2dz+4 = 0, where C is the unit circle, secz = cosz
R
is not

.S
C
π 3π
analytic at z = ± 2 , ± 2 ,.... but all these points lie outside C; none lies

Q
aa
on C or inside C. Similarly for the second integral,whose integrand is not

Sl
analytic at z = ±2i outside C .

Example R 3.28. Nonanalytic function

.S
R 2π − it
zdz = e itie dt = 2πi, where C: z(t) = eit is the unit circle.This does
Q

C 0
not contradict Cauchy’s theorem because f (z) = z is not analytic.
aa
Sl

Example
R dz 3.29. Analyticity sufficient, not necessary
C z2
dz = 0, where C is the unit circle. This result does not follow from
Cauchy’s theorem because f (z) = z12 is not analytic at z = 0.Hence the
.S
Q

condition that f (z) be analytic in domain D is sufficient rather than necessary

aa

R
for C f (z)dz = 0 to be true.
Sl

Example
R dz 3.30. Simple connectedness essential
dz = 2πi, for counterclockwise integration around the unit circle. C
.S

C z
lies in the annulus 12 < |z| < 23 where z1 is analytic, but this domain is not
Q

simply connected, so that Cauchy’s theorem cannot be applied. Hence the

aa
Sl

condition that the domain D be simply connected is quite essential. In other

words, by Cauchy’s theorem, if f (z) is analytic on a simple closed path C
and everywhere inside C, with no exception, not even a single point, then
f (z)dz = 0 holds. The point that causes trouble here in z = 0 where z1 is
R
C
not analytic.

Theorem 7. If f (z) is analytic in a simply connected domain D, then the

integral of f (z) is independent of path D.

Proof:
Let z1 and z2 be any points in D. Consider two paths C1 and C2 in D
from z1 to z2 without further common points Denote by C2? the path C2

36
with orientation reversed. Integrate from z1 over C1 to z2 and over c?2 back
to z1 . This is simple closed path, and Cauchy’s theorem applies under our
assumptions
R of
R ? the present theorem
R and gives zero:
R?
C1
f (z)dz + C2
f (z)dz = 0 thus C1
f (z)dz = − C2
f (z)dz. But the minus
sign on the right disappears if we integrate in the reverse direction from
zR1 to z2 whichR shows that integrals of f (z) over C1 and C2 are equal, i.e
C1
f (z)dz = C2 f (z)dz. This proves the theorem for path that have only
the endpoints in common.

3.6.1 Principle of Deformation of path

.S
This is related to path independence. We may imagine that the path C2

Q
was obtained from C1 by continuously moving C1 (with ends fixed) until it

a
a
Sl
coincides with C2 . Hence we may impose a continuous deformation of the
path of an integral, keeping the ends fixed. As long as our deforming path
always contains only points at which f (z) is analytic, the integral retains the
.S

same value.This is called the principle of deformation of path.

Q
aa

Theorem 8. If f (z) is analytic in a simply connected domain D, then there

Sl

exists an indefinite integral F (z) of f (z) in D and for all paths in D joining
any two points
R z1 z0 and z1 in D, the integral of f (z) from z0 to z1 can be
evaluated by z0 = F (z1 ) − F (z0 )
.S
Q

Example 3.31. Given C dz

R
= 0, (a) For what contour C will this hold?(follow
a

z
a

Cauchy’s integral theorem), (b) Can we conclude that the integral is also zero
Sl

over the boundary of the square with vertices 1 + i, −1 + i, −1 − i, and 1 − i

(counterclockwise)?
.S

Example 3.32. If the integral of a function f (z) over the unit circle equals
Q

3 and over the circle |z| = 2 equals 5, can we conclude that f (z) is analytic
aa
Sl

everywhere in the annulus 1 < |z| < 2?

R
Example 3.33. Prove that C ez dz = 0, if C is a unit circle.

3.6.2 Cauchy’s Theorem for Multiply Connected Domains

For doubly connected domain D- Consider doubly connected D with outer

boundary curve C1 and inner curve C2 . If a function f (z) is analytic in
?
any
R domain DR that contains D and its boundary curves, we claim that
C1
f (z)dz = C2 f (z)dz both integrals being taken counterclockwise.
For domains of higher connectivity R the idea remains
R the same.
R Thus, for a
triply connected domain we have C 1 f (z)dz = C 2 f (z)dz = C 3

37
3.7 Cauchy’s Integral Formula
This is the consequence of Cauchy’s integral theorem.

Theorem 9. Let f (z) be analytic in a simply connected domain D. Then

for any point z0 in D and any simple closed path C in D that enclosed z0 .

R f (z)
dz
= 2πif (z0 )
C z−z0
R ez
Example 3.34. C z−2 dz = 2πiez |z=2 = 46.4268i
R 3 −6 3
Example 3.35. C z2z−i dz = 2πiez ( z2 − 3)|z = 2i = π
− 6πi.

.S
8

Q
Example 3.36. Integrate g(z) = (z 2 − 1)−1 tanz around the circle C: |z| = 3

aa
2

Sl
counterclockwise.

Solution: tanz is not analytic at ± π2 , ± 3π ,... but all these points lie outside
.S

2
the contour, (z 2 − 1)−1 = (z−1)(z+1) 1
is not analytic at 1 and −1, z21−1 =
Q
aa

1
( 1 − z+1 1
)
Sl

R2 z−1
tanz 1
R tanz R tanz
2
C z −1
dz = (
2 C z−1
dz − C z+1
dz) = 2πitan1 = 9.785i.
.S
Q

3.8 Taylor’s and Laurent Series

aa
Sl

P∞ n
Power series is powers of z − z0 in a series of the form: n=0 an (z − z0 ) =
2
a0 + a1 (z − z0 ) + a2 (z − z0 ) + ..., where z is a complex variable, a0 , a1 ,
.... are complex(or real) constants called the coefficients of the series, and
.S
Q

z0 is a complex (or real) constant, called the center of the series.

P If z0 = 0,
aa

we obtain as a particular case a power series in powers of z, ∞ a

n=0 n z n
=
Sl

a0 + a1 (z − z0 ) + a2 (z − z0 )2 + ...

3.8.1 Taylor Series

P∞ n
The Taylor series of a function f (z) is f (z) = n=0 an (z − z0 ) ; where
2
an = n!1 f n (z0 ), thus f (z) = f (z0 ) + z−z
1!
f (z0 ) + (z−z
0 0
2!
0)
f 00 (z0 ) + .....
NOTE: A Maclaurian series is a Taylor’s series with center z0 = 0.
1 n!
Example 3.37. f (z) = 1−z ; thenf n (z) = (1−z) n
n , f (0) = n!. Hence the
1 1
P∞ n
Maclaurian expansion of 1−z is the geometric series 1−z = n=0 z = 1 +
z + z 2 + ..., (|z| < 1) f (z) is singular at z = 1

38
Other Common Examples:

P∞ zn z2
1. ez = n=0 n! =1+z+ 2!
+ ...
P∞ n z 2n z2 z4
2. cosz = n=0 (−1) (2n)! =1−z+ 2!
+ 4!
...
P∞ n z 2n+1 z3 z5
1. sinz = n=0 (−1) (2n+1)! =z− 3!
+ 5!
...
P∞ z 2n z2 z4
1. coshz = n=0 (2n)! =1+ 2!
+ 4!
...
P∞ z 2n+1 z3 z5
1. sinhz = n=0 (2n+1)! =z+ 3!
+ 5!
...

.S
z2 z3 z3 z5
− ..., so that ln( 1+z

Q
1. ln(1 + z) = z − 2!
+ 3! 1−z
) = 2(z + 3!
+ 5!
+ ...)

aa
Sl
3.8.2 Laurent series
.S

Definition: The Laurent series is a representation of a complex function

Q
aa

f (z) as a series. Unlike the Taylor series which expresses f (z) as a series
Sl

of terms with non-negative powers of z, Laurent series includes terms with

negative powers. A consequence of this is that a Laurent series may be used in
cases where a Taylor expansion in not possible. Calculating the Laurent
.S

series expansion: To calculate the Laurent series we use the standard and
Q

modified geometric series which are

aa
Sl

1
P∞ n 1
P∞ 1
1−z
= n=0 z , when|z| < 1and 1−z
= − n=1 z n , when|z| > 1. Here f (z) =
1
1−z
is analytic everywhere apart from the singularity at z = 1. Therefore,
these equations represents the expansions for f (z) in the regions inside and
.S

outside the circle of radius 1, centered on z = 0, where |z| < 1 is the region
Q

inside the circle and |z| > 1 is the region outside the circle.
aa
Sl

Theorem 10. If f (z) is analytic on two concentric circles C1 and C2 with

center z0 in the annulars P between them,n then f (z) can be represented by the
Laurent series f (z) = ∞
P∞ bn
n=0 na (z − z0 ) + n=1 (z−z0 )n = a0 + a1 (z − z0 ) +
b1 b2
a2 (z − z0 ) + .... + z−z 0
+ z−z 0
+ ...
1
Example 3.38. Determine the Laurent series for f (z) = (z+5)
that are valid
in the regions (i) z : |z| < 5 and (ii) z : |z| > 5.

Solution:
The region (i) is an open disk inside a circle of radius 5, centered on z = 0,
and the region (ii) is an open annulus outside a circle of radius 5, centered on

39
z = 0. To make the series expansion easier to calculate we can manipulate
1
our f (z) into a form similar to the series expansion for f (z) = 1−z . So
1 1
f (z) = 5(1+ z ) = 5(1−(− z )) . Hence, for (i) the series expansion is f (z) =
5 5
P∞ P∞ (−1) n zn
1 z n
5 n=0 (− 5
) = n=0 5n+1 , |z| < 5 and for (ii) the series expansion is
P∞ (−1)n 5n−1
f (z) = − 15 ∞ 1
P
n=1 (− z )n = −
5
n=1 zn
, |z| > 5

Example 3.39. Find the Laurent series of z −5 sinz with center at 0

P∞ (−1)n 2n−4 z2
Solution: z −5 sinz = n=0 (2n+1) z = 1
z4
− 1
6z 2
+ 1
120
− 5040
..., (|z| > 0).
1
Example 3.40. Find the Laurent series of z 2 e z with center 0.

.S
Q
aa
1 1 1
Solution: z 2 e z = z 2 (1 + + ...) = z 2 + z + 21 + 1 1

Sl
1!z
+ 2!z 2 3!z
+ 4!z 2
+ ....

EXERCISE
.S
Q
aa

1
Sl

1. Determine the Laurent series for f (z) = z(z+5)

valid in the region
z : |z| < 5.
1
.S

2. For the function f (z) = z(z+2) determine the Laurent series that is valid
Q

within the region R = z : 1 < |z − 1| < 3.

aa
Sl

1
3. Obtain the series expansion for f (z) = z 2 +4
valid in the region |z−2i| >
4
.S
Q
aa

3.9 Residue Integration

Sl

The purpose
R of Cauchy’s residue integration method is the evaluation of
integrals C f (z)dz taken around a simple closed path C. If f (z) is analytic
everywhere on C and inside C, such an integral is zero by Cauchy’s integral
theorem.If f (z) has singularity at a point z = z0 inside C, but P is otherwise
analytic on C and inside C, then f (z) has Laurent series f (z) = ∞ n=0 an (z −
n b1 b2
z0 ) + z−z0 + z−z0 + ... that converges for all points near z = z0 (except at
z = z0 itself),in some domain of the form 0 < |z − z0 | < R. The coefficient
1
b1 of the first negative power of z−z 0
of this Laurent series is given by the
1
R R
integral formula b1 = 2πi C f (z)dz this implies that C f (z)dz = 2πib1 . The
coefficient b1 is is called the residue of f (z) at z = z0 and we denote it by
b1 = Resz=z0 f (z)

40
Example 3.41. Integrate the function f (z) = z −4 sinz counterclockwise
around a unit circle C.

3
Solution: f (z) = sinz z4
= z13 − 3!z
1
+ 5!z − z7! + ..., which converges for |z| > 0,
here bR1 = − 31 . This series shows that f (z) has a pole of third order at z = 0.
Thus C sinzz4
dz = 2πib1 = − πi 3
.
1
Example 3.42. Integrate the function f (z) = z 3 −z 4
clockwise around circle
C: |z| = 21 .

Solution: z 3 − z 4 = z 3 (1 − z) shows that f (z) is singular at z = 0 and

.S
z = 1. Now z = 1 lies outside C. Hence it is of no interest here. So we

Q
1
find it from the Laurent series that converges for 0 < |z| < 1. z3 −z 4 =

aa
Sl
1 1 1
z3
+ z2 + z R+ 1 + z + ..., we see that the residue is 1. Clockwise integration
thus yields C z3dz
−z 4
= −2πiResz=0 f (z) = −2πi.
.S

3.9.1 Two Formulae for Residues at simple Poles

Q
aa
Sl

To calculate a residue at a pole we need not to produce a whole Laurent

series but, more economically, we can derive formulas for residues onces and
for all. For a simple pole at z = z0 the Laurent series is:
.S

b1
+ a0 + a1 (z − z0 ) + a2 (z − z0 )2 + ... (0 < |z − z0 | < R). Here
Q

f (z) = z−z 0
aa

b1 6= 0. Multiplying both sides by z − z0 we have (z − z0 )f (z) = b1 + (z −

Sl

z0 )[a0 + a1 (z − z0 ) + ...].We now let z → z0 . Then the right side approaches

b1 because of continuity. Thus, Resz=z0 f (z) = b1 = limz=z0 (z − z0 )f (z).
.S

Example 3.43. Resz=i z(z9z+i 9z+i 9z+i

2 +1) = limz→i (z − i) z(z+i)(z−i) = [ z(z+i) ]z=i =
10i
=
Q

−2
−5i.
aa
Sl

Another, sometimes simpler formula formula for the residue at simple pole
is obtained if we start from f (z) = p(z) q(z)
(p, q analytic), where p(z0 ) 6= 0
and q(z) has a simple zero at z0 , so that f (z) has a simple pole at z0 , by
definition of simple zero, the Taylor series of q(z) with center z0 is of the
form: q(z) = (z − z0 )q 0 (z0 ) + 22!q00z−z 0
(z0 )+.....
, substituting this into f (z) = p(z) q(z)
and then f (z) into the formula Resz=z0 f (z) = b1 = limz=z0 (z − z0 )f (z) we
obtain Resz=z0 f (z) = limz→z0 (z − z0 ) p(z) q(z)
= limz→z0 0
(z−z0 )p(z)
(z−z0 )q 00 (z0 ) ].
(z−z0 )[q (z0 )+ 2
+...
z − z0 cancels. The denominator has the limit q 0 (z0 ), again by continuity.
Hence our second formula for the residue at simple pole is:
Resz=z0 f (z) = Resz=z0 p(z)
q(z)
= qp(z 0)
0 (z ) .
0

41
Example 3.44. Resz=i z(z9z+i 9z+i
2 +1) = [ 3z 2 +1 ]z=i =
10i
−2
= −5i.

This is the residue at simple pole calculated using the second formula for
residue at simple pole. For the residue at pole of any order,let assume f (z)
having the pole of mth order at z = z0 , the residue is given by: Resz=z0 f (z) =
1 dm−1 2
(m−1)!
limz→z0 [ dz m−1 ((z − z0 ) f (z))].

In particular for a second order, that is for m = 2 we have:

Resz=z0 f (z) = limz→z0 [((z − z0 )2 f (z))0 ]
50z
Example 3.45. The function f (z) = (z+4)(z−1) 2 has a pole of second order at

z = 1, and from the formula second order above we obtain the corresponding
d d 50z
[(z − 1)2 f (z)] = limz→1 dz

.S
residue Resz=1 f (z) = limz→1 dz ( z+4 ) = 8.

Q
aa
Sl
.S
Q
aa
Sl
.S
Q
aa
Sl
.S
Q
aa
Sl

42