V.

Ravichandran
Differential Equations
MAIR22 Complex Analysis and

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

 MAIR22 Complex Analysis and Differential Equations by V. Ravichandran
Preface
This notes is prepared for use in teaching the course MAIR22 Com-
plex Analysis and Differential Equations at NIT Trichy. My earlier notes
on complex integration posted on the institute course material site has
more information than what is presented in these notes. The portion
of application of Laplace transformation is also very brief. One should
work out more problem for all the topics from some other sources as I
have not included any problems for practice. These notes has not been
proof-read carefully. You have now a wonderful opportunity to correct
my mistakes.

V. Ravichandran
June 16, 2020
 Contents

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1. Complex Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1. Analytic functions and C-R equations . . . . . . . . . . . . . . 5
2. Sufficient condition for analyticity . . . . . . . . . . . . . . . . . 9
3. Properties of analytic functions . . . . . . . . . . . . . . . . . . . 13
4. Harmonic functions and harmonic conjugate . . . . . . . . . . 15
5. Möbius transformation . . . . . . . . . . . . . . . . . . . . . . . . . 20
6. Cross-ratio of four points . . . . . . . . . . . . . . . . . . . . . . . 22

2. Complex Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
7. Integral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
8. Cauchy’s theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
9. Applications of Cauchy’s integral formula . . . . . . . . . . . . 36
10. Taylor’s series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
11. Laurent’s series . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
12. Singularities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
13. Cauchy’s residue theorem . . . . . . . . . . . . . . . . . . . . . . 56
14. Evaluation of Improper Integrals . . . . . . . . . . . . . . . . . 57

3. Differential Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
15. Review: Equations of First Order . . . . . . . . . . . . . . . . . 59
16. Linear Homogeneous Equation with Constant Coefficients 67
17. Linear Non-homogeneous Equation of Order n . . . . . . . . 75
18. The Euler-Cauchy Differential Equation . . . . . . . . . . . . 82
19. Method of Variation of Parameters . . . . . . . . . . . . . . . . 87

4. Laplace Transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
20. Laplace transform: examples and existence . . . . . . . . . . 91
21. Properties: linearity and shifting . . . . . . . . . . . . . . . . . . 95
22. Properties: derivatives and integrals . . . . . . . . . . . . . . . 100
23. Periodic functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
24. Inverse Laplace transform . . . . . . . . . . . . . . . . . . . . . . 110
3
4 CONTENTS

25. Convolution theorem . . . . . . . . . . . . . . . . . . . . . . . . . 113

26. Ordinary differential equations . . . . . . . . . . . . . . . . . . . 115
27. Integral equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5. Partial Differential Equations . . . . . . . . . . . . . . . . . . . . . . 117
28. First and Second Order PDE . . . . . . . . . . . . . . . . . . . . 117
29. Standard type of first order equations . . . . . . . . . . . . . . 120
30. The Lagrange’s linear equation Pp + Qq = R . . . . . . . . . 123

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

31. Second Order PDEs . . . . . . . . . . . . . . . . . . . . . . . . . . 125
32. Method of Separation of Variables . . . . . . . . . . . . . . . . 129
 1. Complex Analysis

1. Analytic functions and C-R equations

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

The set C of complex numbers are the set of numbers z = a + ib
where i2 = −1. For two complex numbers z1 = a1 + ib1 and z2 = a2 +
ib2 , the addition and the multiplication are defined as follows:

z1 + z2 = (a1 + a2 ) + i(b1 + b2 )
z1 z2 = a1 a2 − b1 b2 + i(a1 b2 + a2 b1 ).

Indeed, the product is obtained in the usual way using i2 = −1:

z1 z2 = (a1 + ib1 )(a2 + ib2 )

= a1 a2 + ia1 b2 + ia2 b1 + i2 b1 b2
= a1 a2 − b1 b2 + i(a1 b2 + a2 b1 ).
√
If z = a+ib, the non-negative number |z| = a2 + b2 is its modulus. The
complex number z = a − ib is the conjugate of z = a + ib. Clearly |z|2 =
zz. For z 6= 0, the solution of the equation a = |z| cos θ , b = |z| sin θ is an
argument of z and is written as arg z. Any two argument of a complex
number differs by a integral multiple of 2π. The following known as De
Moivre’s theorem is useful:

(cos θ + i sin θ )n = cos nθ + i sin nθ .

Let z0 ∈ C. The limit of a function f : C → C as z → z0 , written as

lim f (z) = l or f (z) → l as z → z0 ,

z→z0

is the unique number l if, for a given ε > 0, there is a δ = δ (ε, z0 ) > 0
such that
| f (z) − l| < ε whenever 0 < |z − z0 | < δ .
The limit of a function can be defined at z0 even if it is not defined at
z = z0 . The limit of a function f = u + iv at z0 = x0 + iy0 can be obtained
5
6 1. COMPLEX ANALYSIS

from the limits of the real and imaginary parts: limz→z0 f (z) = l1 + il2 if
and only if
lim u(x, y) = l1
x→x0
and lim v(x, y) = l2 .
x→x0
y→y0 y→y0

Example 1.1. Consider the function f (z) defined by

|z|

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

f (z) = (z ∈ C − {0}).
z
If z → 0 through the straight lines passing through origin, i.e. when
y = mx, we have
p √ √
|z| x2 + y2 x 2 + m2 x 2 1 + m2
= = =
z x + iy x + imx 1 + im
which depends on the slope of the straight line. Thus the limit of f (z) as
z → 0 does not exist. 

Example 1.2. Consider the function f (z) defined by

z
f (z) = (z ∈ C − {0}).
z
If z → 0 through the straight lines passing through origin, i.e. when
y = mx, we have
z x − iy x − imx 1 − im
= = =
z x + iy x + imx 1 + im
which depends on the slope of the straight line. Thus the limit of f (z) as
z → 0 does not exist. 

Definition 1.3. A function f is differentiable at z = z0 if the limit

f (z) − f (z0 )
lim
z→z0 z − z0
exists. This limit, denoted by f 0 (z0 ), is the derivative of f at z = z0 .

Definition 1.4. A function f is analytic at z = z0 if the limit it has

derivatives at z0 and at every point z with |z − z0 | < R for some R > 0.
The function f is analytic in a region R if it is analytic at all points of
the region R.
 1. ANALYTIC FUNCTIONS AND C-R EQUATIONS 7

Theorem 1.5 (Necessary condition for analyticity). If f (z) =

u(x, y) + iv(x, y) (z = x + iy) is analytic, then the four partial derivatives
ux , uy , vx and vy exist and satisfy the Cauchy-Riemann equations:

ux = vy , uy = −vx .
Proof. By the definition,

f (z0 + h) − f (z0 )

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

f 0 (z0 ) = lim .
∆h→0 h
By choosing real values for h, we have

f (z0 + h) − f (z0 ) ∂ f
f 0 (z) = lim = .
h→0 h ∂x
If we choose purely imaginary values for h, that is, h = ik, k ∈ R, we
have
f (z0 + ik) − f (z0 ) 1 ∂ f
f 0 (z) = lim = .
k→0 ik i ∂y
Comparing the real and imaginary parts of the two expressions for f 0 ,
we get the Cauchy-Riemann equations.
As a consequence, we see that the derivative f 0 (z) of a differentiable
function f is given by the following:

∂f ∂f
f 0 (z) = , f 0 (z) = −i ,
∂x ∂y
∂v
f 0 (z) = ux + i , f 0 (z) = vy − iuy ,
∂x
∂u
f 0 (z) = ux − i , f 0 (z) = vx + ivy .
∂y

Note that the C-R equations in polar form are

∂P 1 ∂Q ∂Q 1 ∂P
= and =
∂r r ∂θ ∂r r ∂θ

where f (z) = P(r, θ ) + iQ(r, θ ), (z = reiθ ).

Example 1.6. The functions f (z) = Re z, f (z) = Im z, f (z) = z and
f (z) = |z| are not differentiable at all the points of C.
8 1. COMPLEX ANALYSIS

We first show that the function f (z) = Re z = x is not differentiable at

all the points of C. In this case, the real and imaginary parts are given
by
u(x, y) = x and v(x, y) = 0.
The partial derivatives are given by
ux = 1, uy = 0, vx = 0, vy = 0.

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Therefore
ux = 1 6= 0 = vy ,
and thus the first of the Cauchy-Riemann equations is not satisfied.
Hence f (z) = Re z is not differentiable at any point of C. Similarly the
function f (z) = Im z is not differentiable at any point of C.
For the function f (z) = z = x − iy, the real and imaginary parts are
given by u(x, y) = x and v(x, y) = −y. Since
ux = 1 6= −1 = vy ,
the first of the Cauchy-Riemann equations is not satisfied. Hence f (z) =
z is not differentiable at any point of C.
For the function f (z) = |z|, the real and imaginary parts are given
by
p
u(x, y) = x2 + y2 , and v(x, y) = 0
and therefore
x y ∂v
ux = p , uy = p , vx = 0, = 0.
x2 + y2 x2 + y2 ∂y
Now for x 6= 0, y 6= 0, we have
x y
ux = p 6= 0 = vy , uy = p 6= 0 = −vx
x + y2
2 x + y2
2

and thus the Cauchy-Riemann equations are not satisfied except at z = 0.

Hence f (z) = |z| is not differentiable at any point of C other than z = 0.
A direct computation shows that
f (z) − f (0) |z|
=
z−0 z
and this has no limit when z → 0 by Example 1.1. Hence f (z) = |z| is
not differentiable even at the origin. 
 2. SUFFICIENT CONDITION FOR ANALYTICITY 9

2. Sufficient condition for analyticity

The following theorem gives a sufficient condition for analyticity:
Theorem 1.7. If f (z) = u(x, y) + iv(x, y) (z = x + iy) is continuous and
the four partial derivatives ux , uy , vx and vy are continuous and satisfy
the Cauchy-Riemann equations:

ux = vy , uy = −vx ,

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

then f (z) is analytic.
More generally, we have
Theorem 1.8. If f = u + iv, the functions u and v has continuous first
order partial derivatives at z = z0 and if the Cauchy-Riemann equations

ux = vy , uy = −vx

are satisfied at z = z0 , then f is differentiable at z = z0 .

Proof. Write z0 = x0 + iy0 and ∆z0 = z − z0 . From advanced calculus,
for functions with continuous first order partial derivatives, we have
u(x, y) = u(x0 , y0 ) + (x − x0 )ux (x0 , y0 )
+ (y − y0 )uy (x0 , y0 ) + η1 |∆z0 |
v(x, y) = v(x0 , y0 ) + (x − x0 )vx (x0 , y0 )
+ (y − y0 )vy (x0 , y0 ) + η2 |∆z0 |
where η1 , η2 → 0 when |∆z0 | → 0. By using this and the Cauchy-
Riemann equations, we have
f (z) − f (z0 )
= u(x, y) − u(x0 , y0 ) + i [v(x, y) − v(x0 , y0 )]
= (x − x0 )ux (x0 , y0 ) + (y − y0 )uy (x0 , y0 ) + η1 |∆z0 |
+ i ((x − x0 )vx (x0 , y0 ) + (y − y0 )vy (x0 , y0 ) + η2 |∆z0 |)
= (x − x0 )ux (x0 , y0 ) + i2 (y − y0 )vx (x0 , y0 ) + η1 |∆z0 |
+ i ((x − x0 )vx (x0 , y0 ) + (y − y0 )ux (x0 , y0 ) + η2 |∆z0 |)
= (z − z0 ) [ux (x0 , y0 ) + ivx (x0 , y0 )] + (η1 + iη2 )|∆z0 |,
or
f (z) − f (z0 ) |∆z0 |
= ux (x0 , y0 ) + ivx (x0 , y0 ) + (η1 + iη2 ) .
z − z0 ∆z0
10 1. COMPLEX ANALYSIS

Since |∆z0 |/∆z0 has unit modulus, we see that the last term in the above
expression tends to zero as ∆z0 → 0. Therefore, by letting ∆z0 → 0 in
the last equation, we get

f 0 (z0 ) = ux (x0 , y0 ) + ivx (x0 , y0 ).

Thus f 0 (z0 ) exists.

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Example 1.9. The function f (z) = z2 is differentiable at every point of
C. For this function f (z), we have

f (z) = (x + iy)2 = x2 − y2 + 2xyi

and therefore the real and imaginary parts are given by

u(x, y) = x2 − y2 and v(x, y) = 2xy.

The partial derivatives of u(x, y) and v(x, y) satisfy the equations

ux = 2x = vy

and

uy = −2y = −vx .

Thus the Cauchy-Riemann equations are satisfied by the function f (z) =

z2 . The derivative can be computed as follows:

∂f
f 0 (z) = = 2x + 2yi = 2(x + iy) = 2z,
∂x
or

∂f
f 0 (z) = −i = −i[−2y + 2xi] = 2(x + iy) = 2z,
∂y
or

f 0 (z) = vx + ivy = 2yi + 2x = 2(x + iy) = 2z.


 2. SUFFICIENT CONDITION FOR ANALYTICITY 11

Example 1.10. Consider the function

f (z) = ez := ex+iy = ex (cos y + i sin y).

For this function,

u(x, y) = ex cos y and v(x, y) = ex sin y

and therefore

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

ux = ex cos y, uy = −ex sin y,
vx = ex sin y, vy = ex cos y.

These partial derivatives of u and v are continuous and satisfy the Cau-
chy-Riemann equations:

ux = ex cos y = vy

and
uy = −ex sin y = −vx .
Thus the function f (z) = ez is analytic in C. Also the derivative is given
by

f 0 (z) = ux + ivx
= ex cos y + iex sin y
= ex+iy = ez .

Example 1.11. Consider the function f (z) = sin z. Since

sin(iy) = i sinh y, cos(iy) = cosh y,

we have

f (z) = sin(x + iy)

= sin x cos(iy) + cos x sin(iy)
= sin x cosh y + i cos x sinh y.

Thus we have

u(x, y) = sin x cosh y and v(x, y) = cos x sinh y

12 1. COMPLEX ANALYSIS

and therefore
ux = cos x cosh y, uy = sin x sinh y,
vx = − sin x sinh y, vy = cos x cosh y.
These partial derivatives of u and v are continuous and satisfies the
Cauchy-Riemann equations:
ux = cos x cosh y = vy

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

and
uy = − sin x sinh y = −vx .
Thus the function f (z) = sin z is analytic in C. Also the derivative is
given by
f 0 (z) = ux + ivx
= cos x cosh y − i sin x sinh y
= cos x cos(iy) − sin x sinh(iy)
= cos(x + iy) = cos z.


Example 1.12. The function f (z) = |z|2 is differentiable at z = 0 but

not analytic at z = 0. Note that
f (z) − f (0) |z|2 − 0 zz
= = = z → 0 as z → 0.
z−0 z−0 z
Thus the derivative of f (z) = |z| at z = 0 is given by f 0 (0) = 0. For this
2

function
f (z) = |z|2 = x2 + y2 ,
the real and imaginary parts are given by
u(x, y) = x2 + y2 and v(x, y) = 0.
Clearly the partial derivatives of u(x, y) and v(x, y) satisfy
ux = 2x 6= 0 = vy
and
uy = 2y 6= 0 = −vx
for any z 6= 0. Thus the function f (z) = |z|2 is not differentiable at any
point other than z = 0. Hence f (z) = |z|2 is differentiable at z = 0 but
not analytic at z = 0. 
 3. PROPERTIES OF ANALYTIC FUNCTIONS 13

3. Properties of analytic functions

Theorem 1.13. A differentiable functions having derivative that vanish
everywhere on a region G is a constant.
Proof. Let f = u + iv be differentiable. Since f 0 = ux + ivx = 0, we have
ux = vx = 0. By Cauchy-Riemann equations, we have uy = −vx = 0 and
vy = ux = 0. Since ux = uy = 0 on G, the function u is constant on G and
similarly v is constant on G. Therefore, f = u + iv is also a constant on

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

G.
Corollary 1.14. If f is differentiable on region G, and either Re f , Im f
or | f | is constant, then f is constant.
Proof. Let f = u + iv. We prove that in each of the case f 0 = 0. Once
this is done, the result follows from the previous theorem. If Re f = u is
constant, then ux = 0 = uy and so f 0 = ux − iuy = 0. The proof is similar
if Im f is constant.
Let | f | is a constant c, then u2 + v2 = c2 . Differentiating with respect
to x and y, we get (upon cancelling 2), uux + vvx = 0 and uuy + vvy = 0.
Eliminating u and v, we get
ux vx
= 0.
uy vy
Using Cauchy-Riemann equations ux = vy and uy = −vx , we have
ux −uy
=0
uy ux
or u2x + u2y = 0. This shows that ux = uy = 0 and so f 0 = ux − iuy = 0.
The angle between a curve γ and the real axis at a point z = z0 in the
curve γ is the angle between the tangent of γ at z = z0 and the straight
line through z = z0 which is parallel to the real axis. The angle between
two curves intersecting at a given point is the angle between their tan-
gents at that point.
Definition 1.15. Two families of curves u(x, y) = c1 and v(x, y) = c2
form an orthogonal system if they intersect at right angles at each of
their points of intersection.

Example 1.16. Consider the families of curves x2 − y2 = c1 and 2xy =

c2 . These two families of curves are obtained by considering the real and
the imaginary parts of the analytic function f (z) = z2 = x2 − y2 = 2xyi.
14 1. COMPLEX ANALYSIS

Let the two families of curves intersect at a point z = z0 , z0 6= 0. The

derivative of a function in the first family is given by

dy dy x
x−y = 0 or = .
dx dx y

Similarly the derivative of a function in the second family is given by

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

dy dy y
y+x = 0 or =− .
dx dx x
If z = z0 is not on the real or the imaginary axis, it is clear that the prod-
uct of these two derivatives at z = z0 is −1 and hence the two families of
curves are orthogonal at z = z0 . If z = z0 is on the real or the imaginary
axis, then the value of the derivatives are 0 for one function and ∞ for
the other. In this case also, the two families of curves are orthogonal at
the point of intersection. 

Theorem 1.17. If the function f = u + iv is analytic in a domain D,

and z0 ∈ D is a point common to both u(x, y) = c1 and v(x, y) = c2 and
f 0 (z0 ) 6= 0, then the curves form an orthogonal system.
Proof. Let u(x, y) = c1 and v(x, y) = c2 intersect at z = z0 . Since f 0 (z0 ) 6=
0, it follows that either ux or vx is nonzero at z = z0 . Let us first consider
the case where both ux and vx are nonzero. By differentiating u(x, y) = c1
with respect x,
dy
ux + uy =0
dx
or
dy ux
=− .
dx uy
Therefore the slope m1 of u(x, y) = c1 at z = z0 is given by

ux (x0 , y0 )
m1 = −
uy (x0 , y0 )

and similarly the slope m2 of v(x, y) = c2 is given by

vx (x0 , y0 )
m2 = − .
vy (x0 , y0 )
 4. HARMONIC FUNCTIONS AND HARMONIC CONJUGATE 15

By using the Cauchy-Riemann equations, we see that the product of the

slopes at z = z0 is given by

ux (x0 , y0 ) vx (x0 , y0 )
m1 m2 = − ×− = −1.
uy (x0 , y0 ) vy (x0 , y0 )

Hence the curves are orthogonal at z = z0 .

If either ux or vx is zero, then the slopes are 0 for one function and ∞

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

for the other and hence the curves are orthogonal at z = z0 .

4. Harmonic functions and harmonic conjugate

Definition 1.18. A function φ (x, y) is harmonic if (i) the second partial
derivatives of φ (x, y) are continuous, (ii) φ (x, y) satisfies the Laplace
equation

5φ = φxx + φyy = 0.

Definition 1.19. If f (z) = u + iv is analytic function, then v is the con-

jugate harmonic function of u.

Theorem 1.20. The real and imaginary parts of an analytic functions

are harmonic functions.
Proof. Since f is analytic, the Cauchy-Riemann equations ux = vy and
uy = −vx holds and differentiating them with respect to x and y respec-
tively we get uxx = vxy and uyy = −vyx . Since the second order par-
tial derivative of v are continuous, vxy = vyx and therefore uxx + uyy =
vxy − vyx = 0. Thus, u is harmonic and similarly v is harmonic.

Example 1.21. Consider the function

u(x, y) = x2 − y2 + 2xy.

The partial derivatives are given by

ux (x, y) = 2x + 2y,
uy (x, y) = −2y + 2x,
uxx = 2,
uyy = −2
uxy = 2.
16 1. COMPLEX ANALYSIS

All the partial derivatives are continuous in C and

uxx (x, y) + uyy (x, y) = 2 − 2 = 0.

Thus u(x, y) is a harmonic function in C. 

Given a harmonic function u(x, y), one of the following methods
can be used to find the conjugate harmonic function and in particular
the analytic function f (z).

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Method I. Since f (z) is analytic, the differential dv can be written
as

dv = vx dx + vy dy
∂u
= −uy dx + dy
∂x
= Mdx + Ndy,

where
M = −uy and N = ux .
Note that the differential Mdx + Ndy is exact if there is a function h
having first order partial derivatives such that Mdx + Ndy = dh. Since
dh = hx dx + hy dy, it follows that M = hx and N = hy . Thus we must
have
My = hyx = hxy = Nx
provided the second order partial derivatives are all continuous. It is also
known that if M and N are continuous and has continuous first order
partial derivatives satisfying

My = Nx ,

then Mdx + Ndy is exact. Since u(x, y) is harmonic, we have ∇2 u = 0

and therefore
My = −uyy = uxx = Nx .
Hence v(x, y) can be found by integrating the differential dv and thus we
have
Z
v= −uy dx + ux dy + ic
 4. HARMONIC FUNCTIONS AND HARMONIC CONJUGATE 17
R
where c is a real constant. If Mdx + Ndy is exact, then Mdx + Ndy
is the sum of integral of the function M with respect to x keeping y as
constant plus the integral (with respect to y) of those terms in N which
involves only y.
Method II. (Milne Thompson method) Since

f 0 (z) = ux − iuy

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

we have
Z
f (z) = [ux − iuy ] dz + c

where c is a real constant.

Method III. We know that

f 0 (z) = ux − iuy .

Writing
z+z z−z
x= and y = ,
2 2i
we have
   
0 z+z z−z z+z z−z
f (z) = ux , − iuy , .
2 2i 2 2i
By replacing z by z, we get

f 0 (z) = ux (z, 0) − iuy (z, 0) .

and hence
Z
f (z) = [ux (z, 0) − iuy (z, 0)] dz + ic

where c is a real constant.

Method IV. By using
z+z z−z
x= and y = ,
2 2i
write
2u(x, y) = f (z) + f (z).
Then the required analytic function is f (z).
We illustrate these ideas in the following:
18 1. COMPLEX ANALYSIS

Example 1.22. Consider the function u(x, y) = x2 − y2 . For this func-

tion, we have

ux = 2x, uy = −2y.

Also

uxx = 2, uyy = −2

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

and therefore

uxx + uyy = 0.

Therefore the function u(x, y) is harmonic in C.

By using Method I, we get
Z Z
v= 2ydx + 2xdy = 2 d(xy) = 2xy + c.

Hence we get

f (z) = x2 − y2 + i2xy + ic = z2 + ic.

where c is a real constant.

By Method II, we have
Z Z
f (z) = (2x + 2yi)dz + ic = 2zdz = z2 + ic.

By Method III, we have

Z
f (z) = 2zdz + ic = z2 + ic.

By Method IV, we have

2u(x, y) = 2(x2 − y2 )
"  #
z+z 2 z−z 2
 
=2 −
2 2i
= z2 + z2

and therefore f (z) = z2 and a purely imaginary constant can be added.


 4. HARMONIC FUNCTIONS AND HARMONIC CONJUGATE 19

Example 1.23. Consider the function u(x, y) = ex cos y. Note that

ux = ex cos y, uy = −ex sin y.

Also
uxx = ex cos y, uyy = −ex cos y
and therefore

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

uxx + uyy = 0.
Therefore the function u(x, y) is harmonic in C.
By using Method I, we get
Z Z
v= ex sin ydx + ex cos ydy = d(ex sin y) = ex sin y + c.

Hence we get

f (z) = ex cos y + iex sin y + c

= ex (cos y + i sin y) + c
= ex eiy + c
= ex+iy + c
= ez + c

where c is a real constant.

By Method II, we have
Z
f (z) = (ex cos y + iex sin y)dz + ic
Z
= ez dz + ic
= ez + ic.

Note that

ux (z, 0) = ez cos 0 = ez , uy (z, 0) = −ez sin 0 = 0.

By Method III, we have

Z
f (z) = ez dz + ic = ez + ic.
20 1. COMPLEX ANALYSIS

Thus by Method IV, we have

 
x ( z+z
) z−z
2u = 2e cos y = 2e 2 cos .
2i
By writing cos z = (eiz + e−iz )/2, we see that
z+z
h z−z z−z
i
2u(x, y) = e 2 e 2 + e− 2
= ez + ez .

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Therefore f (z) = ez and a purely imaginary constant can be added. 

5. Möbius transformation
Definition 1.24. A mapping f (z) which preserves angles both in mag-
nitude and direction between every pair of curves through a point is
conformal at that point. If the mapping preserves the magnitude only,
then the mapping is isogonal.
An analytic function f (z) is conformal at a point z = z0 if f 0 (z0 ) 6= 0.
Definition 1.25. A point z = z0 is cirtical point of w = f (z) if f 0 (z0 ) =
0 or ∞. Critical points are those points at which the mapping w = f (z)
is not conformal.
In calculus, the set of real numbers was extended by adding two
symbols ∞ and −∞. In case of complex numbers, we just add the sym-
bol ∞ to the complex plane and the resulting set is called the extended
complex plane. We denote the extended complex plane by C∞ .
The transformation
az + b
w= (ad − bc 6= 0)
cz + d
is called the bilinear transformation or the linear fractional transfor-
mation or the Möbius transformation. When ad − bc = 0 or b/a =
d/c, we have

a z + ab a
w= = ,
c z + dc c
a constant. The mapping is defined for all z 6= −d/c in C. At z = −d/c
it takes the value ∞. Also
a + b/z a
w= → as z → ∞.
c + d/z c
 5. MÖBIUS TRANSFORMATION 21

Thus w is a mapping from C into C. The inverse transformation of

the bilinear transformation w = (az + b)/(cz + d) is given by

−dw + b
z= .
cw − a
Since the inverse exits, it is one-to-one and onto mapping of C∞ to C∞ .
It is also a bilinear transformation. When w 6= a/c, it is a function into

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

C. It maps w = a/c to ∞ and the point ∞ is transformed to −d/c.
The derivative of the bilinear transformation is given by

dw a(cz + d) − c(az + b) ad − bc
= 2
= 6= 0.
dz (cz + d) (cz + d)2

Therefore the mapping is conformal at all points where it is finite.

Theorem 1.26. A bilinear transformation other than w = z has one or
two fixed points.
Proof. For the bilinear transformation w = z, every point in the complex
plane is a fixed point. Note that w = z corresponds to the case where
b = c = 0 and a = d. Hence consider any other bilinear transformation

az + b
w= , (ad − bc 6= 0).
cz + d
Rewriting the equation
az + b
= z,
cz + d
we get

cz2 + (d − a)z − b = 0,

which is a polynomial equation of degree one or two. This has one or

two roots. This shows that a bilinear transformation has one or two fixed
points.

Example 1.27. The function w = (2z + 1)/(z + 1) has the two fixed
2
√ of the equation (2z + 1)/(z + 1) = z or z −
points; these are the roots
z − 1 = 0 or z = (−1 ± 5)/2. The fixed point of w = (z + 1)/2 is given
by z + 1 = 2z or z = 1. Every point in the complex plane is clearly a
fixed point of the mapping w = z. 
22 1. COMPLEX ANALYSIS

The bilinear transformation contains some of the transforms we have

seen earlier as special cases. For example, when c = 0, d = 1, we get the
transform w = az + b which includes as special cases the transform w =
z+a, w = az. Also if a = 0 = d, b = 1 = c, we get the transform w = 1/z.
In fact, the bilinear transformation can be obtained as a composition of
these transforms. We first write
a
az + b (cz + d) + b − ad a bc − ad 1

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

w= = c c
= + .
cz + d cz + d c c cz + d

Consider the mappings defined by

w1 = cz
w2 = w1 + d = cz + d
1 1
w3 = =
w2 cz + d
bc − ad bc − ad 1
w4 = w3 =
c c cz + d
a a bc − ad 1 az + b
w5 = + w4 = + = = w.
c c c cz + d cz + d

Thus w is the composition of the mappings w1 , w2 , . . . , w5 . Since each

of the mapping w1 , w2 , . . . , w5 maps “circles” into “circles”, we see that
the bilinear transformation also maps “circles” into “circles”.
Since each of one of these transformation maps circles to circles, we
have
Theorem 1.28. The bilinear transformation maps “circles” into “cir-
cles”.

6. Cross-ratio of four points

Definition 1.29. The cross-ratio (z1 , z2 , z3 , z4 ) of four distinct complex
numbers z1 , z2 , z3 , z4 is defined by

(z1 − z2 )(z3 − z4 )
(z1 , z2 , z3 , z4 ) = .
(z1 − z4 )(z3 − z2 )

If any one of these four numbers is ∞, say z j , then the two terms on the
right-hand side containing z j is taken to be 1.
 6. CROSS-RATIO OF FOUR POINTS 23

Theorem 1.30 (Invariance of cross-ratio). If w1 , w2 , w3 , w4 are the im-

ages of the four distinct points z1 , z2 , z3 , z4 under the bilinear transfor-
mation
az + b
w= ,
cz + d
then

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

(w1 , w2 , w3 , w4 ) = (z1 , z2 , z3 , z4 ).
In other words, the cross-ratio is invariant under the bilinear transfor-
mation.
Proof. Assume that none of the four numbers z1 , z2 , z3 , z4 is ∞. Since wi
is the image of zi under the bilinear transformation, we have
azi + b
wi = , (i = 1, 2, 3, 4).
czi + d
Now
azi + b az j + b
wi − w j = −
czi + d cz j + d
(azi + b)(cz j + d) − (az j + b)(czi + d)
=
(czi + d)(cz j + d)
(ad − bc)(zi − z j )
= .
(czi + d)(cz j + d)
Therefore
(w1 − w2 )(w3 − w4 )
(w1 , w2 , w3 , w4 ) =
(w1 − w4 )(w3 − w2 )
(ad − bc)(z1 − z2 ) (ad − bc)(z3 − z4 )
=
(cz1 + d)(cz2 + d) (cz3 + d)(cz4 + d)
(cz1 + d)(cz4 + d) (cz3 + d)(cz2 + d)
×
(ad − bc)(z1 − z4 ) (ad − bc)(z3 − z2 )
(z1 − z2 )(z3 − z4 )
=
(z1 − z4 )(z3 − z2 )
= (z1 , z2 , z3 , z4 ).
Assume that one of the four numbers z1 , z2 , z3 , z4 is ∞, say z1 = ∞.
In this case, we see that w1 = a/c and
a a azi + b ad − bc
w1 − wi = − wi = − =
c c czi + d c(czi + d)
24 1. COMPLEX ANALYSIS

and therefore
a
w1 − wi c − wi cz j + d
= a = .
w1 − w j c −wj czi + d
This shows that
(w1 − w2 )(w3 − w4 )
(w1 , w2 , w3 , w4 ) =
(w1 − w4 )(w3 − w2 )

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

(cz4 + d) (ad − bc)(z3 − z4 ) (cz3 + d)(cz2 + d)
=
(cz2 + d) (cz3 + d)(cz4 + d) (ad − bc)(z3 − z2 )
(z3 − z4 )
=
(z3 − z2 )
= (z1 , z2 , z3 , z4 ).
Thus the cross-ratio is invariant under the bilinear transformation.
As cross-ratio is preserved under Möbius transformation, we have
Theorem 1.31. The bilinear transformation
w − w1 w2 − w3 z − z1 z2 − z3
=
w − w3 w2 − w1 z − z3 z2 − z1
maps z1 , z2 and z3 onto w1 , w2 and w3 respectively.

Example 1.32. Find a bilinear transformation that maps z1 = 1, z2 = i

and z3 = −1 to the points w1 = 1, w2 = 0 and w3 = −1.
By using (??), we obtain
(1 − 0)(−1 − w) (1 − i)(−1 − z) i − 1 − z + iz
= =
(1 − w)(−1 − 0) (1 − z)(−1 − i) −1 − i + z + iz
or
1+w i − 1 − z + iz
= .
1 − w −1 − i + z + iz
Therefore
(1 + w) − (1 − w)
w=
(1 + w) + (1 − w)
1+w
1−w +1
= 1+w
1−w −1
(i − 1 − z + iz) + (−1 − i + z + iz)
=
(i − 1 − z + iz) + (−1 − i + z + iz)
 6. CROSS-RATIO OF FOUR POINTS 25

i−z
=
iz − 1
1 + iz
= .
i+z
Since
(1 + iz)(z − i)
Im w = Im
|z + i|2

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

−i + z + z + i|z|2
= Im
|z + i|2
−(1 − |z|2 )
= <0
|z + i|2

when |z| < 1, we see that this transform maps the unit disk |z| < 1 onto
the half-plane Im w < 0. 

Example 1.33. Find a bilinear transformation that maps z1 = 0, z2 =

−1 and z3 = 1 to the points w1 = 1, w2 = 0 and w3 = ∞.
In this case, the formula for finding the required bilinear transfor-
mation should be written without the term w3 . The formula is

w1 − w2 (z1 − z2 )(z3 − z)
= .
w1 − w (z1 − z)(z3 − z2 )

For the given choices of zi and wi , this formula becomes

1 1−z
=
1−w −2z
or
2z 1+z
w = 1+ = .
1−z 1−z
Since
1 + z 1 − |z|2
Re w = Re = >0
1 − z |1 − z|2
for |z| < 1, it is clear that the disk |z| < 1 is mapped by this transform to
the right half-plane Re w > 0. 
MAIR22 Complex Analysis and Differential Equations by V. Ravichandran
 2. Complex Integration

7. Integral

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

A curve or path in C is the image of a complex-valued continuous
function of a real variable. The interval [a, b] is the parametric interval
of the curve γ. The function γ : [a, b] → C is called a parametric rep-
resentation of the curve. The curve γ is an open curve if γ(a) 6= γ(b). It
is a closed curve if γ(a) = γ(b). The curve is smooth if g0 is continuous
and γ 0 (t) 6= 0 for all t ∈ [a, b].
Example 2.1. A parametric representation of the line segment joining
the point z = z1 to z = z2 is

γ(t) = z1 + (z2 − z1 )t (0 ≤ t ≤ 1).

The parametric interval is [0, 1]. When the parametric interval is [a, b], a
parametric representation of the line joining the points z = z1 and z = z2
is
z1 (b − t) + z2 (t − a)
γ(t) = .
b−a


Example 2.2. A parametric representation for the circle

|z − z0 | = r,

whose center is at z = z0 and radius r, is

γ(t) = z0 + reit (0 ≤ t ≤ 2π),

where the parametric interval is [0, 2π]. The function

2π(a−t)i
γ(t) = z0 + re a−b (a ≤ t ≤ b)

is also a parametric representation of the circle |z − z0 | = r where the

parametric interval is [a, b]. 
27
28 2. COMPLEX INTEGRATION

For a complex-valued function f : [a, b] → C, given by f (t) = u(t) +

iv(t), we define the integral ab f (t)dt by
R

Z b Z b Z b
f (t)dt = u(t)dt + i v(t)dt.
a a a

For any two continuous functions f : [a, b] → C and g : [a, b] → C, we

have the following:

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

(1) ab (α f + β g)(t)dt = α ab f (t)dt + β ab g(t)dt,
R R R

(2) ab f (t)dt = ac f (t)dt + γb f (t)dt for every c ∈ (a, b),

R R R

(3) Re ab f (t)dt = ab Re f (t)dt,

R R

(4) Im ab f (t)dt = ab Im f (t)dt,

R R

(5) ab f (t)dt = ab f (t)dt,

R R

(6) ab f (t)dt = F(b) − F(a) if F 0 (t) = f (t) for all t ∈ [a, b],
R
Rb Rb
(7) a f (t)dt ≤ a | f (t)|dt.

Definition 2.3. Let f be a continuous function defined on domain D,

and let γ : [a, b] → C be a smooth curve with γ([a, b]) ⊂ D. The line
integral of the function f (z) over the curve γ is defined by the following
equation
Z Z b
f (z)dz = f (γ(t))γ 0 (t)dt.
γ a

If the parametric form of γ is z = z(t), a ≤ t ≤ b, then the length of

γ is given by
Z Z b
l(γ) = |dz| = |γ 0 (t)|dt.
γ a

Example 2.4. Consider the circle of radius r given by

γ(t) = reit (t ∈ [0, 2π]).

The length of the circle is given by

Z 2π Z 2π
l(γ) = |rieit |dt = rdt = 2πr.
0 0


 7. INTEGRAL 29

Theorem 2.5 (ML inequality). If the function f (z) is continuous on a

curve γ, | f (z)| ≤ M on γ and the length of the curve γ is L, then
Z
f (z)dz ≤ ML.
γ

Proof. If
Z
f (z)dz = 0,

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

γ
then the inequality is trivial. So assume that
Z
f (z)dz 6= 0.
γ

Then there is an α ∈ [0, 2π] such that

Z Z
f (z)dz = f (z)dz eiα .
γ γ
Now
Z Z
−iα
f (z)dz = e f (z)dz
γ γ
Z
= e−iα f (z)dz
γ
Z b
= e−iα f (γ(t))γ 0 (t)dt
a
Z b
= Re(e−iα f (γ(t))γ 0 (t))dt
a
Z b
≤ |e−iα f (γ(t))γ 0 (t)|dt
a
Z b
= | f (γ(t))| |γ 0 (t)|dt
a
Z b
≤ M|γ 0 (t)|dt
a
Z b
=M |γ 0 (t)|dt
a
= ML.
Note that the imaginary part of the integral
Z b
e−iα f (γ(t))γ 0 (t)dt
a
30 2. COMPLEX INTEGRATION

given by
Z b
Im(e−iα f (γ(t))γ 0 (t))dt
a
is zero.

Example 2.6. Let f (z) = zn , n 6= −1 be an integer. Let γ be the circular

arc given by

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

|z| = r and t0 ≤ arg z ≤ t1 .
The parametric representation for this curve is
γ(t) = reit (t0 ≤ t ≤ t1 ).
Let
a = reit0 and b = reit1 ,
so that γ is the circular arc of radius r joining a to b. Thus
Z Z t1
zn dz = rn+1 ei(n+1)t idt
γ t0
t1
rn+1 ei(n+1)t
=
n+1
t0
n+1
r e i(n+1)t1 − r n+1 ei(n+1)t0
=
n+1
bn+1 − an+1
= .
n+1
In particular, when γ is the circle |z| = r, we have t0 = 0, t1 = 2π, or
a = b. Therefore when γ is the circle |z| = r, we have
I
zn dz = 0,
γ

and, in general, for any polynomial p(z) = a0 + a1 z + · · · + an zn ,

I I I I
p(z)dz = a0 1dz + a1 zdz + · · · + an zn dz = 0.
γ γ γ γ

It is also true that

I
zn dz = 0 (n 6= −1)
γ
 7. INTEGRAL 31

when γ : [a, b] → C is any closed curve (γ should not pass through z = 0

when n < −1). In this case, the proof is as follows:
I Z b
n
z dz = [γ(t)]n γ 0 (t)dt
γ a
b
[γ(t)]n+1
=
n+1 a

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

[γ(b)]n+1 [γ(a)]n+1
= −
n+1 n+1
= 0.

Example 2.7. Consider the function

1
f (z) = (z 6= z0 ).
z − z0
Let γ be the circular arc given by

|z − z0 | = r and t0 ≤ arg z ≤ t1 .

The parametric representation for this curve is

γ(t) = z0 + reit , t0 ≤ t ≤ t1 .

Then
γ 0 (t) = rieit
and
Z t1
1
I
dz = idt = it|tt10 = (t1 − t0 )i.
γ z − z0 t0

In particular, when γ is the circle |z − z0 | = r, we have t0 = 0, t1 = 2π.

Therefore when γ is the circle |z − z0 | = r, we have
1
I
dz = 2πi.
γ z − z0

Notice that the function f (z) = 1/(z − z0 ) is not the derivative of some
function F(z) in any domain containing z0 . 
32 2. COMPLEX INTEGRATION

Example 2.8. Consider the integral

I
Re zdz
γ

where γ is the unit circle |z| = 1. On γ, we have

|z|2 = 1

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

or
zz = 1
or
z = 1/z
and therefore
 
z+z 1 1
Re z = = z+ .
2 2 z
Thus
I 
1 1 1
I I
Re zdz = zdz + dz = (0 + 2πi) = πi.
γ 2 γ γ z 2


8. Cauchy’s theorems
The following theorem, stated without proof, is needed to prove
Cauchy’s integral theorem.
Theorem 2.9 (Green’s theorem). Let M(x, y), N(x, y) be functions
with continuous first order partial derivatives on a simply connected
domain D and γ a positively oriented curve in D enclosing the domain
R. Then
ZZ  
∂N ∂M
I
Mdx + Ndy = − dxdy.
γ R ∂x ∂y
Theorem 2.10 (Cauchy’s integral theorem). If f (z) is analytic and
has continuous derivative f 0 (z) inside and on a simple closed curve γ,
then
I
f (z)dz = 0.
γ
 8. CAUCHY’S THEOREMS 33

Proof. Since
I I
f (z)dz = [u(x, y)dx − v(x, y)dy]
γ γ
I
+i [u(x, y)dy + v(x, y)dx],
γ

Green’s Theorem and the Cauchy-Riemann equations yield

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

ZZ  
∂v ∂u
I
f (z)dz = − − dxdy
γ R ∂x ∂y
ZZ  
∂u ∂v
+i − dxdy = 0.
R ∂x ∂y
The above theorem is valid without the assumption that f 0 (z) is con-
tinuous inside and on γ.
Theorem 2.11 (Cauchy-Goursat theorem). If the function f (z) is an-
alytic inside and on a simple closed curve γ, then
I
f (z)dz = 0.
γ

Example 2.12. Let p(z) be a nonconstant polynomial. Then for any α,

z − α divides p(z) − p(α) and therefore it can be written as

p(z) − p(α) = (z − α)q(z)

where q(z) is a polynomial. Thus

p(z) p(α)
= q(z) +
z−α z−α
and hence, for any positively oriented curve γ enclosing α, we have
1 p(z) 1 1 p(α)
I I I
dz = q(z)dz + dz = p(α).
2πi γ z−α 2πi γ 2πi γ z−α
Thus
1 p(z)
I
p(α) = dz
2πi γ z−α
for any point α enclosed by γ. In other words, the value of the polyno-
mial at a point α ∈ I(γ) is computed by using the value of the polynomial
on the curve γ. 
34 2. COMPLEX INTEGRATION

γ1

γ2
A C
D B

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Theorem 2.13 (Cauchy’s Integral Formula). If f (z) is analytic at all
points inside and on a closed curve γ and a is a point inside γ, then

1 f (z)
Z
f (a) = dz.
2πi γ z−a

In general, we have

n! f (z)
Z
(n)
f (a) = dz.
2πi γ (z − a)n+1
Proof. Assume that γ2 lies inside the domain bounded by γ1 . Let A, B
be two points on γ1 and C and D be two points in γ2 as in the Fig. ??.
Join the points A and C by a line segment and similarly join B and D.
Denote the upper portion of the curve γ1 from B to A by γ1u and the
lower portion of the curve γ1 from A to B by γ1d . Similarly, let γ2u and
γ2d denote respectively the upper and lower portions of the curve γ2 .
Let γ3 be the curve formed by γ1d , the line segment BD, γ2d (tra-
versed from D to C) and the line segment CA. Similarly let γ4 be the
curve formed by γ1u , the line segment AC, γ2u (traversed from C to D)
and the line segment DB. Thus
γ3 = γ1d + BD − γ2d +CA,
γ4 = γ1u + AC − γ2u + DB,
γ1 = γ1u + γ1d ,
γ2 = γ2u + γ2d .
Then the function f (z) is analytic inside and on the two simple closed
curve γ3 and γ4 . Thus by applying the Cauchy-Goursat Theorem for
each of these simple closed curves, we have
I I
f (z)dz = f (z)dz = 0
γ3 γ4
 8. CAUCHY’S THEOREMS 35

and therefore
Z Z Z Z
f (z)dz + f (z)dz − f (z)dz + f (z)dz
γ1d BD γ2d CA
I
= f (z)dz = 0
γ3

and

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Z Z Z Z
f (z)dz + f (z)dz − f (z)dz + f (z)dz
γ1u AC γ2u DB
I
= f (z)dz = 0.
γ4

Adding the last two equations and noting that

Z Z
f (z)dz + f (z)dz = 0
γ −γ

for any curve γ, we get

Z Z Z Z
f (z)dz − f (z)dz + f (z)dz − f (z)dz = 0
γ1d γ2d γ1u γ2u

or
I I
f (z)dz = f (z)dz
γ1d +γ1u γ2d +γ2u

or
I I
f (z)dz = f (z)dz.
γ1 γ2

Let γ1 be the circle |z − a| = r where the radius r is chosen so that

γ1 lies inside the domain bounded by γ (see Fig. ??). The function F(z)
defined by
f (z) − f (a)
F(z) = (z 6= a)
z−a
is analytic inside the annulus bounded by the curves γ and γ1 . Thus by
Cauchy’s theorem for annulus, we have
I I
F(z)dz = F(z)dz
γ γ1
36 2. COMPLEX INTEGRATION

or
f (z) − f (a) f (z) − f (a)
I I
(8.1) dz = dz.
γ z−a γ1 z−a
Note that f (z) is continuous and therefore for a given ε > 0 there is an
δ > 0 such that
ε
| f (z) − f (a)| < whenever |z − a| < δ .
2π

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Note that on γ1 , we have |z − a| = r. By taking r < δ and making use of
the ML inequality, we obtain
f (z) − f (a)
I
ε
dz ≤ 2πr = ε whenever |z − a| < δ .
γ1 z−a 2πr
Therefore
f (z) − f (a)
I
dz = 0 when r → 0.
γ1 z−a
Thus by letting r → 0 in (8.1), we get
f (z) − f (a)
I
dz = 0
γ z−a
or
f (z) f (a)
I I
dz = dz = 2πi f (a).
γ z−a γ z−a
This proves Cauchy’s integral formula.

9. Applications of Cauchy’s integral formula

Example 2.14. Consider the integral
eaz
I
dz
γ z
where γ is the unit circle |z| = 1. By Cauchy’s integral formula, we have
eaz
I
dz = 2πi.
γ z
Let a be a real number. Since the parametric representation of the unit
circle is given by
γ(t) = eit (0 ≤ t ≤ 2π),
 9. APPLICATIONS OF CAUCHY’S INTEGRAL FORMULA 37

we have
Z 2π aeit
eaz
Z 2π
e
I
it it
2πi = dz = it
e idt = eae idt
γ z 0 e 0
and therefore
Z 2π
it
eae dt = 2π.
0

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Since
it
eae = ea cost+ai sint = ea cost [cos(a sint) + i sin(a sint)],
by taking the real part of the above integral, we get
Z 2π
ea cost cos(a sint)dt = 2π.
0


Example 2.15. Consider the integral

Re z
I
dz
γ z

where γ is the unit circle |z| = 1. Since

 
1 1
Re z = z+
2 z
on |z| = 1, we have
I  
Re z 1 1
I
dz = 1 + 2 dz = 0.
γ z 2 γ z
Also if 0 < |α| < 1, then
z2 + 1
I  
Re z 1
I
dz = dz
γ z−α 2 γ z(z − α)
Resolving the integrand into partial fractions, we have
1 z2 + 1 1 1 1 + α2 1
 
1
= 1− + .
2 z(z − α) 2 αz α z−α
Using this, we obtain
1 1 + α2
 
Re z
I
dz = πi − + = παi.
γ z−α α α

38 2. COMPLEX INTEGRATION

Example 2.16. Consider

eaz
I
dz
γ zn+1
where γ is the unit circle |z| = 1 and a be a real number. Let
f (z) = eaz .
Then

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

f (n) (z) = an eaz
and therefore

f (n) (0) = an .
By Cauchy’s integral formula for the nth derivative,
eaz 2πian
I
dz = .
γ zn+1 n!
On the other hand, since the parametric representation of the unit circle
is given by
γ(t) = eit (0 ≤ t ≤ 2π),
we have
it
eaz eae
I Z 2π Z 2π
it −nit
dz = eit idt = eae idt
γ zn+1 0 ei(n+1)t 0

and therefore
2πan
Z 2π
it −nit
eae dt = .
0 n!
Since
it −nit
eae = ea cost+(a sint−nt)i = ea cost [cos(a sint − nt) + i sin(a sint − nt)],
by taking the real part, imaginary part of the above integral, we get
2πan
Z 2π
ea cost cos(a sint − nt)dt =
0 n!
and
Z 2π
ea cost sin(a sint − nt)dt = 0.
0

 9. APPLICATIONS OF CAUCHY’S INTEGRAL FORMULA 39

Example 2.17. Consider the integral

dθ
Z π
.
0 5 + 4 cos θ
First we convert the integral to an integral over [0, 2π]. By the substitu-
tion θ = 2π − t and noting that cos(2π − t) = cost, we have
Z 2π Z 0
dθ −dt
=
5 + 4 cos θ π 5 + 4 cos(2π − t)

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

π
dt
Z π
=
0 5 + 4 cost
dθ
Z π
= .
0 5 + 4 cos θ
Thus we have
Z 2π Z 2π
dθ dθ dθ
Z π
= +
0 5 + 4 cos θ 0 5 + 4 cos θ π 5 + 4 cos θ
dθ
Z π
=2
0 5 + 4 cos θ
or
dθ 1 2π dθ
Z π Z
= .
0 5 + 4 cos θ 2 0 5 + 4 cos θ
To evaluate this later integral, consider the contour γ which is just the
unit circle |z| = 1. By converting the given integral into contour integral
by using (??) and using Cauchy’s integral formula (with f (z) = 1/(z +
2) and a = −1/2), we have
dθ 1 1 dz
Z π Z
= 2
0 5 + 4 cos θ 2 γ 5 + 2(z + 1)/z iz
i dz
Z
=− 2
2 γ 2z + 5z + 2
i dz
Z
=−
2 γ (2z + 1)(z + 2)
1
i
Z
z+2
=− dz
4 γ z + 1/2
i 1
= − 2πi
4 −1/2 + 2
π
= .
3

40 2. COMPLEX INTEGRATION

Example 2.18. Consider the integral

dθ
Z π
(0 < a < 1).
0 1 + a2 − 2a cos θ
By the substitution θ = 2π − t, we have
Z 2π Z 0
dθ −dt
2
=
π 1 + a − 2a cos θ π 1 + a2 − 2a cost

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

dθ
Z π
= .
0 1 + a2 − 2a cos θ
Thus we have
Z 2π Z Z 2π 
dθ π dθ
2
= +
0 1 + a − 2a cos θ 0 π 1 + a2 − 2a cos θ
dθ
Z π
=2
0 1 + a2 − 2a cos θ
or
Z 2π
dθ 1 dθ
Z π
2
= .
0 1 + a − 2a cos θ 2 0 1 + a2 − 2a cos θ
Converting the integral in the right side of the above equation, we get
dθ 1 dz
Z π I
2
=
0 1 + a − 2a cos θ 2 |z|=1 iz(1 + a − a(z2 + 1)/z)
2

i dz
I
=
2 |z|=1 az2 − (1 + a2 )z + a
i dz
I
=
2 |z|=1 (az − 1)(z − a)
i f (z)
I
= dz
2 |z|=1 z − a
where
1
f (z) = .
az − 1
Since 0 < a < 1, the function f (z) is analytic inside and on the unit
circle |z| = 1. By using Cauchy’s integral formula for this function f (z),
we get
dθ i f (z)
Z π I
2
= dz
0 1 + a − 2a cos θ 2 |z|=1 z − a
 9. APPLICATIONS OF CAUCHY’S INTEGRAL FORMULA 41

i 2πi
=
2 a2 − 1
π
= .
1 − a2


Example 2.19. Consider the integral

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Z 2π
dθ
(0 < a < 1).
0 1 + a sin θ
By writing this integral as a contour integral, we get
Z 2π
dθ dz
I
= −1 2
0 1 + a sin θ iz(1 + a z 2iz
|z|=1 )
dz
I
=2 2
|z|=1 az + 2iz − a
dz
I
=2
|z|=1 a(z − αi)(z − β i)

where αi and β i are the roots of the equation

az2 + 2iz − a = 0.

In fact, we have
√ √
−1 + 1 − a2 −1 − 1 − a2
α= , β= .
a a
Since 0 < a < 1, the inequality |α| < 1 is equivalent to
√
−1 + 1 − a2
−1 < <1
a
or
p
1−a < 1 − a2 < 1 + a.

Squaring the inequality and rewriting, we see that |α| < 1 is equivalent
to 2a(a − 1) < 0 < 2a(a + 1). The last inequality holds for 0 < a < 1.
Thus it follows that |α| < 1. Since αβ = 1, it follows that |β | > 1. Thus
42 2. COMPLEX INTEGRATION

we see that αi lies inside the unit disk and β i lies outside the unit disk.
Now we write
Z 2π
dθ f (z)
I
= dz
0 1 + a sin θ |z|=1 z − αi

where
2
f (z) = .
a(z − β i)

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

The function f (z) is analytic inside and on the unit circle |z| = 1 and by
using Cauchy’s integral formula for this function f (z), we get
Z 2π
dθ 2 2π
= 2πi =√ .
0 1 + a sin θ a(α − β )i 1 − a2


Example 2.20. Consider the integral

Z 2π
dθ
(0 < b < a).
0 (a + b cos θ )2
By writing this integral as a contour integral using (??), we get
Z 2π
dθ 1 dz
I
= 2
0 (a + b cos θ )2 |z|=1 (a + b z +1 )2 iz
2z
zdz
I
= −4i
|z|=1 (bz2 + 2az + b)2
zdz
I
= −4i
|z|=1 b2 (z − α)2 (z − β )2
where α and β are the roots of the equation

bz2 + 2az + b = 0.

In fact, we have
√ √
−a + a2 − b2 −a − a2 − b2
α= , β= .
b b
Note that
√
2a 2 a2 − b2
α +β = − , α −β = .
b b
 10. TAYLOR’S SERIES 43

Since 0 < b < a, it can be shown that |α| < 1. Since αβ = 1, it follows
that |β | > 1. Thus we see that α lies inside the unit disk and β lies
outside the unit disk. Now we write
Z 2π
dθ f (z)
I
= dz
0 (a + b cos θ )2 |z|=1 (z − α)
2

where

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

4zi
f (z) = − .
b2 (z − β )2
Note that
4i(z + β )
f 0 (z) =
b2 (z − β )3
and hence
4i(α + β ) ai
f 0 (α) = 2 3
=− 2 .
b (α − β ) (a − b2 )3/2
The function f (z) is analytic inside and on the unit circle |z| = 1 and by
using Cauchy’s integral formula for the first derivative of the function
f (z), we get
Z 2π
dθ 2πa
2
= 2πi f 0 (α) = 2 .
0 (a + b cos θ ) (a − b2 )3/2


10. Taylor’s series

Theorem 2.21 (Taylor’s Theorem). If f (z) is analytic at all points
inside and on a circle γ with center at a, then

f 0 (a) f 00 (a) 2 f 000 (a)

f (z) = f (a) + (z − a) + (z − a) + (z − a)3 + . . . .
1! 2! 3!
Proof. Let D(a, δ ) denote the disk centered at the point z = a with radius
δ . Let z ∈ D(a, δ ) be a fixed complex number. Then there is an r0 such
that |z − a| = r0 and r := δ +r 0
2 . Consider the circle C centered at z = a
and radius r. Notice that
1 1 (z − a) (z − a)2 ∞
(z − a)n
= + + + · · · = ∑
η − z η − a (η − a)2 (η − a)3 n=0 (η − a)
n+1
44 2. COMPLEX INTEGRATION

whenever |z−a| < |η −a| = r. Since the series ∑∞ n n+1 = 1 ∞ (r /r)n

n=0 r0 /r r ∑n=0 0
is convergent to (1/r)/(1 − r0 /r) = 1/(r − r0 ) and
(z − a)n r0n
= ,
(η − a)n+1 rn+1
(z−a) n
we see, by Theorem ??, that the series ∑∞ n=0 (η−a)n+1 converges uni-
formly. Therefore this series can be integrated term by term. Thus we

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

have
1 f (η) 1 f (η)
Z Z
= dη
2πi C η − z 2πi C η − a
(z − a) f (η)
Z
+
2πi C (η − a)2
(z − a)2 f (η)
Z
+ 3
+··· .
2πi C (η − a)

By using Cauchy’s integral formula

1 f (η) f (k) (a)
Z
k+1
dη = ,
2πi C (η − a) k!
we have
∞
f (k) (a)
f (z) = ∑ (z − a)k .
k=0 k!

Example 2.22. Consider the function f (z) = ez . Then f (k) (z) = ez for
any nonnegative integer k. Thus, for any z ∈ C, we have
∞
ea ∞
(z − a)k
ez = ∑ k! (z − a)k
= ea
∑ k! .
k=0 k=0

This shows that

∞
(z − a)k
ez−a = ∑ k!
k=0
and thus
∞
zk
ez = ∑ (z ∈ C).
k=0 k!

 10. TAYLOR’S SERIES 45

Example 2.23. The derivatives of the function f (z) = sin z are given by
 π
f 0 (z) = cos z = sin z +
2 
00
 π  2π
f (z) = cos z + = sin z +
2 2
   
000 2π 3π
f (z) = cos z + = sin z +
2 2

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

and in general
   
(k) (k − 1)π kπ
f (z) = cos z + = sin z + .
2 2

Thus we have
 
(k) kπ
f (a) = sin a +
2

and therefore

sin a + kπ
∞


sin z = ∑ 2
(z − a)k .
k=0 k!

In particular, when a = 0, we have

kπ
∞


sin
sin z = ∑ 2
zk .
k=0 k!

Since
(
kπ 0 (k even)
sin =
2 (−1)(k−1)/2 (k odd).

we have
∞
(−1)k 2k+1
sin z = ∑ z .
k=0 (2k + 1)!


46 2. COMPLEX INTEGRATION

1
Example 2.24. The function f (z) = 1−z has the derivatives given by
1
f 0 (z) =
(1 − z)2
1·2
f 00 (z) =
(1 − z)3
and in general

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

k!
f (k) (z) = .
(1 − z)k+1
Thus we have
k!
f (k) (−1) =
2k+1
and hence, for z satisfying |z + 1| < 2, we have
∞
1 1
= ∑ k+1 (z + 1)k .
1 − z k=0 2
This can also be obtained by writing
1 1 1
= .
1 − z 2 1 − 1+z
2
By using
1
= 1 + η + η2 + ·· · (|η| < 1)
1−η
we have
∞
1 1
= ∑ k+1 (z + 1)k
1 − z k=0 2
1+z
whenever 2 < 1 or |z + 1| < 2. 

Example 2.25. The function

1
f (z) =
1 − z2
has the Maclaurin series given by
f (z) = 1 + z2 + z4 + · · ·
valid in the disk |z| < 1. 
 10. TAYLOR’S SERIES 47

Example 2.26. Consider the function f (z) = (1 + z)n (n ∈ N). In this

case we have

f 0 (z) = n(1 + z)n−1

f 00 (z) = n(n − 1)(1 + z)n−2

and in general, for k ≤ n,

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

f (k) (z) = n(n − 1) · · · (n − k + 1)(1 + z)n−k
n!
= (1 + z)n−k .
(n − k)!

Note that f (k) (z) = 0 for k > n. Thus we have

(
n!
for k ≤ n
f (k) (0) = (n−k)!
0 for k > n

and therefore we have, for any z ∈ C,

∞ ∞  
n n! k n k
(1 + z) = ∑ z =∑ z.
k=0 (n − k)!k! k=0 k

1
Example 2.27. Consider the function f (z) = (1−z)n (n ∈ N). In this
case we have
n
f 0 (z) =
(1 − z)n+1
n(n + 1)
f 00 (z) =
(1 − z)n+2
and in general

n(n + 1) · · · (n + k − 1)
f (k) (z) = .
(1 + z)n+k
Thus we have
f (k) (0) = n(n + 1) · · · (n + k − 1).
48 2. COMPLEX INTEGRATION

Therefore, for any z with |z| < 1, we have

∞
1 n(n + 1) · · · (n + k − 1) k
n
=∑ z.
(1 − z) k=0 k!

In particular, we have (by taking n = 1, 2)

∞
1
= ∑ zk

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

1 − z k=0
∞
1
= ∑ (k + 1)zk .
(1 − z)2 k=0

Also we have
∞ ∞
1+z 2
= −1 + = −1 + 2 ∑ zk = 1 + ∑ 2zk .
1−z 1−z k=0 k=0

11. Laurent’s series

Definition 2.28. The series ∑∞ n
k=−∞ cn (z − a) is the Laurent expan-
sion of the function f (z) if it converges to the function f (z). The series
∑−1 n
k=−∞ cn (z − a) is called the principal part of the function f (z) and the
series ∑k=0 cn (z − a)n is called the analytic part of the function f (z).
∞

Theorem 2.29 (Laurent series). Let the function f (z) be analytic in

the annulus
D := {z : 0 < r1 < |z − a| < r2 }.
Then
∞
f (z) = ∑ ck (z − a)k
k=−∞

where
1 f (η)
Z
(11.1) ck = k+1
dη (−∞ < n < ∞)
2πi C (η − a)

and C is a closed curve in D enclosing the circle |z − a| = r1 .

 11. LAURENT’S SERIES 49

Example 2.30. Consider the function

a−b
f (z) = .
(z − a)(z − b)
By resolving the function into partial fraction, we get
1 1
f (z) = + .
z−a b−z

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

We consider the following cases.
Case (1) (|z| < |a|, |z| < |b|) In this case, we have
1 1 1 1
f (z) = −
b 1 − z/b a 1 − z/a
∞
zk ∞
zk
= ∑ k+1
−∑ k+1
k=0 b k=0 a
∞  
1 1
= ∑ − zk .
k=0 bk+1 ak+1

Case (2) (|z| < |a|, |z| > |b|). In this case, we have
1 1 1 1
f (z) = − −
a 1 − z/a z 1 − b/z
∞
zk ∞
bk
=−∑ k+1
−∑ k+1
.
k=0 a k=0 z

Case (3) (|z| > |a|, |z| < |b|). In this case, we have already shown
that
∞
ak ∞
zk
f (z) = ∑ k+1
+∑ k+1
.
k=0 z k=0 b

Case (4) (|z| > |a|, |z| > |b|). In this case, we have
1 1 1 1
f (z) = −
z 1 − a/z z 1 − b/z
∞
ak ∞
bk
= ∑ k+1
−∑ k+1
k=0 z k=0 z
∞ k
a − bk
= ∑ zk+1 .
k=0


50 2. COMPLEX INTEGRATION

Example 2.31. Consider the function

1
f (z) = .
z(z2 + 1)

The function is analytic in any domain not containing the points z =

0, i, −i.
First let us consider the annular region 0 < |z| < 1. In this case

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

the function can be expanded into a series; by using binomial series
expansion:

1
f (z) = (1 − z2 + z4 + . . .)
z
1
= − z + z3 − z5 + . . .
z
∞
1
= + ∑ (−1)n+1 z2n+1 .
z n=0

Similarly when 1 < |z| < ∞, we have

!
1 1
f (z) = 3
z 1 + z12
1 1 1
= (1 − + − . . .)
z3 z2 z4
1 1 1
= 3 − 5 + 7 −...
z z z
∞
(−1) n+1
= ∑ 2n+1 .
n=1 z

12. Singularities
Definition 2.32. If a function f (z), analytic in a region R, is zero at
point z = z0 in R, then z = z0 is a zero of f (z). That is, if f (z0 ) = 0, then
z0 is a zero of f (z).
If f (z0 ) = 0, f 0 (z0 ) = 0, f 00 (z0 ) = 0, . . ., f (n−1) (z0 ) = 0 and f (n) (z0 ) 6=
0, then z = z0 is a zero of order n.
 12. SINGULARITIES 51

Definition 2.33. A point z = z0 is a singular point (or singularity) of

f (z) if f (z) is not analytic at that point. The singular point z = z0 is
isolated if there is no other singular point in a neighbourhood of the
point z = z0 . Otherwise it is a non-isolated singular point.

Example 2.34. The function

1
f (z) =

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

z
is analytic at all point in C expect at z = 0 and thus it has a singular
point at z = 0. Similarly the function
sin z
f (z) =
z
has a singular point at z = 0. The function
1
f (z) = ,
(z − 1)(z − 2)
is analytic at all point in C except at z = 1, 2. It has singular points at
z = 1, 2 respectively.
The function f (z) = Re z, Im z, arg z have no singular points as they
are nowhere analytic in C. 

Example 2.35. Consider the function

1
f (z) = .
sin( 1z )

The points where sin( 1z ) = 0 are all singular points. Thus the points
1
z = nπ (n = ±1, ±2, . . .) are all singular points of f (z). Also the point
1
z = 0 is a singular point. The singular points z = nπ (n = ±1, ±2, . . .)
are all isolated singular points.
The singular point z = 0 is not isolated, for, in every neighbourhood
of z = 0, there are points where the function f (z) is not analytic. For
example, if we consider the deleted neighbourhood 0 < |z| < ε, then the
1 1
points nπ belongs to the neighbourhood for all n > επ . 

Definition 2.36 (Removable Singular Point). A singular point z = a

is removable singular point if limz→a f (z) exists.
52 2. COMPLEX INTEGRATION

From the definition of isolated singularity, it is clear that a func-

tion having isolated singularity at z = z0 is analytic in an annular region
0 < |z − z0 | < R. In this region the function f (z) has a Laurent series
expansion
∞
f (z) = ∑ cn (z − z0 )n
n=−∞

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

where
1 f (z)
Z
cn = n+1
dz (n = 0, ±1, . . .)
2πi c (z − z0 )

and C is a closed curve enclosing z = z0 in |z − z0 | < R. At an isolated

singular point z = z0 , the principal part ∑−1 n
−∞ cn (z − z0 ) of the Laurent
series expansion of f (z) may have
(1) no term. In this case, the singularity is called a removable
singularity.
(2) finite number of terms. In this case, the singularity is called a
pole. If the principal part is
c−n c−n+1 c−1
+ + · · · + ,
(z − z0 )n (z − z0 )n−1 (z − z0 )

c−n 6= 0, then the pole is of order n. A pole of order 1 is called

a simple pole and a pole of order 2 is called a double pole.
(3) infinite number of terms. In this case, the singular is called an
essential singularity.

Example 2.37. Consider the function

1 − cos z
f (z) =
z2
2 4 6
1 − (1 − z2! + z4! − z6! + · · · )
=
z2
1 z2 z4
= − + −··· .
2! 4! 6!
The singular point z = 0 of this function is isolated and also removable.
Since
z3 z5
sin z = z − + −...,
3! 5!
 12. SINGULARITIES 53

we have
sin z z2 z4
= 1− + −....
z 3! 5!
Thus the isolated singular point z = 0 of the function sin z/z is also re-
movable. 

Example 2.38. The function

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

1
f (z) =
z(z − 1)

has two singular points z = 0 and z = 1. They were isolated. In 0 <

|z| < 1, we have
1 1
f (z) = − (1 + z + z2 + · · · ) = − − 1 − z − · · ·
z z
and therefore z = 0 is a simple pole of the function f (z). Similarly, in
0 < |z − 1| < 1, we have
1
f (z) = (−1 + 1 − z)
(1 − z)
1
=− [1 + (1 − z) + (1 − z)2 + · · · ]
(1 − z)
1
=− − 1 − (1 − z) − · · ·
1−z
and therefore z = 1 is also a simple pole of the function f (z). 

Example 2.39. The function

1 1 1
f (z) = e z = 1 + + 2 + · · ·
z 2!z
has only z = 0 as a singular point. This singular point is clearly an iso-
lated singular point. Since the principal part of f (z) has infinite num-
ber of terms, the singular point z = 0 is an essential singular point of
f (z). 

Definition 2.40. Let f (z) be a function defined except possibly at cer-

tain points. If the point z = z0 is an isolated singular point of the function
54 2. COMPLEX INTEGRATION

f (z), then the residue of the function f (z) at the point z = z0 , denoted
by Res[ f (z), z0 ], is the coefficient of (z − z0 )−1 in the Laurent expansion
of the function f (z) around the point z = z0 . Equivalently, by taking
k = −1 in equation (11.1), we see that the residue of the function f (z)
at the point z = z0 is given by
1
Z
Res[ f (z), z0 ] = f (z)dz
2πi C

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

where C is a closed curve enclosing the point z = z0 that lie in a deleted
neighbourhood of the point z = z0 in which the function f (z) is analytic.

Theorem 2.41. Let the point z = z0 be a pole of order n of the function

f (z). Then the residue of the function f (z) at the pole z = z0 is given by

φ (n−1) (z0 )
Res[ f (z), z0 ] = lim
z→z0 (n − 1)!

where
φ (z) = (z − z0 )n f (z).
In particular, if the point z = z0 is a simple pole of the function f (z), the
residue of f (z) at the simple pole z = z0 is given by

Res[ f (z), z0 ] = lim (z − z0 ) f (z).

z→z0

Proof. Since the point z = z0 is a pole of order n for the function f (z),
we can write
bn bn−1 b1
f (z) = n
+ n−1
+···+ + a0 + a1 (z − z0 ) + · · ·
(z − z0 ) (z − z0 ) z − z0
in a deleted neighbourhood of z = z0 . Thus we have

φ (z) = bn + bn−1 (z − z0 ) + · · · + b1 (z − z0 )n−1 + a0 (z − z0 )n + · · ·

By differentiating φ (z), we have

φ 0 (z) = bn−1 +2bn−2 (z−z0 )+· · ·+(n−1)b1 (z−z0 )n−2 +na0 (z−z0 )n−1 +· · ·

and in general
n! (n + 1)!
φ (n−1) (z) = (n − 1)!b1 + a0 (z − z0 ) + a2 (z − z0 )2 + · · ·
1! 2!
 12. SINGULARITIES 55

Hence by taking limit as z → z0 , we have

lim φ (n−1) (z) = (n − 1)!b1

z→z0

or since b1 = Res[ f (z), z0 ],

φ (n−1) (z)
Res[ f (z), z0 ] = lim .
z→z0 (n − 1)!

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

In particular, when the pole z = z0 is a simple pole, we have φ (z) =
(z − z0 ) f (z) and hence the residue of the function f (z) at the simple
pole z = z0 is given by

Res[ f (z), z0 ] = lim φ (0) (z)

z→z0
= lim φ (z)
z→z0
= lim (z − z0 ) f (z).
z→z0

Example 2.42. Consider the function

z2 + 1
f (z) =
(z − 1)(z − 2)(z − 3)2 .

The poles of the function f (z) are z = 1, 2, 3. The poles z = 1, 2 are

simple poles of the function f (z). The residue of the function f (z) at
these poles are given by

(z − 1)(z2 + 1) 1
Res[ f (z), 1] = lim = −
z→1 (z − 1)(z − 2)(z − 3)2 2

and

(z − 2)(z2 + 1)
Res[ f (z), 2] = lim =5
z→2 (z − 1)(z − 2)(z − 3)2

The pole z = 3 is a pole of order 2. To compute the residue of the function

f (z) at the pole z = 3, we first write

z2 + 1 z2 + 1
φ (z) = (z − 3)3 f (z) = = 2
(z − 1)(z − 2) z − 3z + 2
56 2. COMPLEX INTEGRATION

The residue is given by φ 0 (3). Since

2z(z2 − 3z + 2) − (z2 + 1)(2z − 3)
φ 0 (z) =
(z2 − 3z + 2)2
−3z2 + 2z + 3
=
(z2 − 3z + 2)2
we have

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

9
Res[ f (z), 3] = φ 0 (3) = − .
2


13. Cauchy’s residue theorem

Theorem 2.43 (Cauchy’s residue theorem). If the function f (z) is an-
alytic inside and on a positively oriented simple closed curve C except
at a finite number of singular points α1 , α2 , . . ., αn inside C, then
Z n
f (z)dz = 2πi ∑ Res[ f (z), αk ].
C k=1

Example 2.44. Consider the integral

dz
I

γ (z − α)(z − β )n
where n is a positive integer and α ∈ I(γ) and β ∈ E(γ). Let
1
f (z) = .
(z − α)(z − β )n
Note that z = α and z = β are poles of the integrand f (z). The pole
z = α is simple pole and lies inside γ where the pole z = β lies outside
γ. The residue of f (z) at z = α is given by
1 1
Res[ f (z), α] = lim (z − α) f (z) = lim n
= .
z→α z→α (z − β ) (α − β )n
Hence by Cauchy’s residue theorem, we have
dz 2πi
Z
n
= .
γ (z − α)(z − β ) (α − β )n

 14. EVALUATION OF IMPROPER INTEGRALS 57

14. Evaluation of Improper Integrals

We shall use the following lemmas without proof.
Lemma 2.45 (Cauchy’s Lemma ). If f (z) is uniformly continuous and
|(z − a) f (z)| → 0 as |(z − a)| → 0, then
Z
f (z) dz → 0
γ

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

as r → 0 where γ is circle |z − a| = r.

Example 2.46. Evaluate the integral

dx
Z ∞
(a > 0, b > 0, a 6= b).
0 (x2 + a2 )(x2 + b2 )

Consider the function

1
f (z) = (a > 0, b > 0, a 6= b).
(z2 + a2 )(z2 + b2 )

The poles of the function f (z) are

z = ±ai, ±bi.

Since a, b are positive real numbers, the poles z = ai, bi are in the upper
half-plane. The residue at these simple poles are given by
1
Res[ f (z), ai] = lim (z − ai)
z→ai (z2 + a2 )(z2 + b2 )
1
= lim
z→ai (z + ai)(z2 + b2 )
1
=
2ai(b − a2 )
2

and
1
Res[ f (z), bi] = .
2bi(a2 − b2 )

By Cauchy’s residue theorem, we have

 
dx 1 1
Z ∞
p.v. 2 2 2 2
= 2πi +
−∞ (x + a )(x + b ) 2ai(b2 − a2 ) 2bi(a2 − b2 )
58 2. COMPLEX INTEGRATION

2πi
= [b − a]
2abi(b2 − a2 )
π
= .
ab(a + b)

Since the integrand is even, if follows that

dx
Z ∞
π
= .

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

0 (x2 + a2 )(x2 + b2 ) 2ab(a + b)


 3. Differential Equations

15. Review: Equations of First Order

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

(This section is omitted for examination!)
Differential equation is an equation in which differential coefficients
occur. Differential equations are of two types: ordinary differential
equation, and partial differential equation. An ordinary differential equa-
tion is one in which a single independent variable enters, either explicitly
or implicitly. A partial differential equation is one in which at least two
independent variables occur and the partial differential coefficients oc-
curring in them have reference to anyone of these variables. The order
of the differential equation is the order of the highest derivative, occur-
ring in it. If a differential equation can be expressed as a polynomial
in derivatives, the power of the highest derivative is the degree of the
differential equation.
A solution of differential equation is a function that satisfies the
given differential equation. The functions y = A cos nx, y = B sin nx and
more generally y = A cos nx + B sin nx are solutions of
d2y
+ n2 y = 0.
dx2
The solution, in which the number of the arbitrary constants occurring
is the same as the order of the equation is called as general solution or
complete integral. Particular solution is obtained by giving particular
values to the constant occurring in the general solution. For example
y = 2 cos nx − sin nx is a particular solution of the differential equation
y00 + n2 y = 0.
15.1. Variable Separable Equations. Suppose an Equation is in
the form of f (x)dx + g(y)dy = 0, we can directly integrate this equation
and the solution is
Z Z
f (x)dx + g(y)dy = C,

where C is a arbitrary constant.

59
60 3. DIFFERENTIAL EQUATIONS

Example 3.1. Consider

tan y sec2 xdx + tan x sec2 ydy = 0.

By dividing the equation by tan x tan y, and integrating, we have

sec2 x sec2 y
Z Z
dx + dy = 0,
tan x tan y

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

or the solution is given implicitly by

log(tan x) + log(tan y) = C.

This solution can also be written as tan x tan y = eC = A. 

15.2. Homogeneous Equations. Consider

dy f1 (x, y)
=
dx f2 (x, y)
where f1 and f2 are homogeneous functions of the same degree in x and
y, that is, f1 can be written as x φ x and f2 can be written as xn ψ xy ,
n y

where n is the degree of homogeneity. Put y = vx, then

dy dv
= v+x· .
dx dx
Using these, we see that the given equation becomes
dv φ (v)
v+x =
dx ψ(v)
and hence,
dx ψ(v)dv
= .
x φ (v) − vψ(v)
The solution is obtained by integrating the above and replacing v by xy .
Example 3.2. The equation
dy dy
y2 + x2 = xy
dx dx
is homogenous equation and we rewrite it as
dy y2
= .
dx xy − x2
 15. REVIEW: EQUATIONS OF FIRST ORDER 61

Substituting y = vx, we get

dv v2 x2 v2
v+x· = 2 =
dx vx − x2 v − 1
or
dv v2 v
x· = −v = .
dx v − 1 v−1

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Integrating, we have
v−1 dx
Z Z
dv = .
v x
This leads to v − log v = log(x) + logC or v = log(vx) +C. By substitut-
ing v = y/x, we get y/x = log(y) + logC, or ey/x = yC. 

15.3. Non-homogeneous equations. Non-homogeneous equations

of first degree in x and y. Consider the equation
dy
(ax + by + c) = Ax + By +C, where a, b, c, A, B,C are constants.
dx
Put x = X + h and y = Y + k, The equation becomes
dy
(15.1) (aX + bY + ah + bk + c) = AX + BY + Ah + Bk +C.
dx
If h, k be chosen to satisfy

(15.2) ah + bk + c = 0
(15.3) Ah + Bk + c = 0,

then the above equation (15.1) becomes

dy
(aX + bY ) = AX + BY
dx
This is a homogeneous equation in X and Y .
15.4. Linear homogenous equation. The linear equation of first
order is of the form of
dy
+ P(x)y = 0
dx
R
is known as linear homogenous equation. The solution is y e Pdx = C,
where C is a arbitrary constant.
62 3. DIFFERENTIAL EQUATIONS

15.5. Linear non-homogenous equation. An equation of the form

dy
+ P(x)y = Q(x)
dx
is known as linear non-homogenous equation. The solution is given by
R Z R
P(x)dx Pdx
ye = Q(x)e dx +C,

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

where C is a arbitrary constant.
15.6. Bernoulli’s equation. Consider the equation
dy
+ P(x)y = Q(x)yn
dx
is known as Bernoulli’s equation. This can be reduced into the linear
form. Dividing by yn , we get
dy
y−n + P(x)y1−n = Q.
dx
Put z = y1−n , the equation reduces to
dz
+ (1 − n)P(x)z = (1 − n)Q(x).
dx
This is a linear non-homogenous equation.
Example 3.3. Solve

dy sin x cos2 x
− (tan x)y = .
dx y2

Multiplying by y2 , we get
dy
(15.4) y2 − (tan x)y3 = sin x cos2 x.
dx
Put z = y3 , so that
dz dy
= 3y2 .
dx dx
The given equation becomes
dz
− 3(tan x)z = 3 sin x cos2 x.
dx
 15. REVIEW: EQUATIONS OF FIRST ORDER 63

Let
P(x) = −3 tan x, Q(x) = 3 sin x cos2 x.
Then
R 3 x)
e P(x)·dx
= elog(cos = cos3 x.
Hence the solution is given by
− cos6 x

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Z
3
z cos x = 3 sin x cos5 xdx + c = +c
2
or
cos6 x
y3 cos3 x = − + c. 
2
15.7. Exact equation. The equation M · dx + N · dy = 0 is exact if
∂M ∂N
= .
∂y ∂x
The solution can be got by the following steps:
(1) Integrate M · dx keeping y as constant,
(2) Integrate those terms in N · dy which do not contain any x,
(3) Equate to a constant, the sum of these integrates.

Example 3.4. Solve (x2 − 4xy − 2y2 )dx + (y2 − 4xy − 2x2 )dy = 0. The
functions
M = x2 − 4xy − 2y2 and N = y2 − 4xy − 2x2
satisfy
∂M ∂N
= .
∂y ∂x
Therefore the given equation is exact and
x3
Z
M · dx = − 2x2 y − 2y2 x,
3
y3
Z
N · dy = (leaving terms involving x).
3
Hence the complete solution is
x3 y3
− 2x2 y − 2xy2 + = C. 
3 3
We now discuss first order higher degree equations. We shall denote
dy/dx by p for convenience.
64 3. DIFFERENTIAL EQUATIONS

15.8. Equations solvable for p. Let the equation of the first order
and the nth degree in p be

(15.5) pn + P1 pn−1 + P2 pn−2 + . . . + Pn = 0

where P1 , P2 . . . Pn denote the functions of x and y. The equation (15.5)

can be resolved into factors:

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

(p − R1 )(p − R2 ) . . . (p − Rn ) = 0

Let the solution of p − Ri = 0 be φi (x, y, ci ) = 0. Then the solution of

(15.5) is

φ1 (x, y, c1 )φ2 (x, y, c2 ) . . . φn (x, y, cn ) = 0.

Example 3.5. Solve x2 p2 + 3xyp + 2y2 = 0. The equation is solvable

for p. Solving for p, we get p = y/x or p = −2y/x. The equations
dy/dx = y/x gives xy = c1 , and dy/dx = −2y/x give x2 y = c2 . The
solution is given by (xy − c1 )(x2 y − c2 ) = 0. 

15.9. Equations solvable for y. In this case, f (x, y, p) = 0 can be

put in the form

(15.6) y = g(x, p).

Differentiating with respect to x, we get

 
dp
p = φ x, p, .
dx

This being an equation in two variables p and x, can be solved by any of

the foregoing methods to get a solution

(15.7) ψ(x, p, c) = 0.

Eliminating p from (15.6) and (15.7), we get the solution for the given
equation.
Example 3.6. Consider xp2 − 2yp + x = 0. Solving for y, we have

x(p2 + 1)
y= .
2p
 15. REVIEW: EQUATIONS OF FIRST ORDER 65

Differentiating with reference to x, we get

p2 + 1 (p2 − 1) d p
p= +x .
2p 2p2 dx
Rewriting it, we have
p2 − 1 d p x(p2 − 1)
=
2p dx 2p2

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

or
dx d p
= .
x p
Integrating both sides, we obtain

p = cx.

Eliminating p from this and the given equation, we get the solution

2cy = c2 x2 + 1. 

15.10. Equations solvable for x. In this case, the equation is of the

form

(15.8) x = F(y, p).

Differentiate with respect to y to get

 
1 dp
= φ y, p, .
p dy
Solving this differential equation, we get

(15.9) ψ(y, p, c) = 0.

By eliminating p from equation (15.8) and (15.9), we get the solution.

Example 3.7. Solve

(15.10) x = y2 + log p.

Differentiate with respect to y to get

1 1 dp
= 2y + .
p p dy
66 3. DIFFERENTIAL EQUATIONS

Hence
dp
+ 2py = 1.
dy
Now the solution is of this equation is
Z
y2 2
(15.11) pe = ey · dy + c.
Using (15.10) and (15.11), we get

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Z
2 +logp 2 2
ex = ey = pey = ey dy + c.
which is the required solution. 

15.11. Clairaut’s equation. The differential equation of the form

y = px + f (p)
is known as Clairaut’s equation. Differentiate with reference to x to get
dp
p = p + (x + f 0 (p)) .
dx
0
Therefore either d p/dx = 0 or x + f (p) = 0. Since d p/dx = 0 gives
p = c, a constant, we get the general solution y = x + f (c). The singular
solution can be obtained by eliminating x + f 0 (p) = 0 and the given
equation.
Example 3.8. Solve
(15.12) y = px + p2 .
The general solution, obtained by putting p = c in given equation, is
y = cx + c2 .
To find the singular solution, differentiate the given equation with
respect to x. This gives
dp dp
p = p + x + 2p .
dx dx
This shows that x + 2p = 0 or p = −x/2. Substituting this value in the
given equation, we get
x2 x2
y=− +
2 4
x2
=−
4
2
and this gives 4y + x = 0 which is the singular solution of the given
equation. 
 16. LINEAR HOMOGENEOUS EQUATION WITH CONSTANT COEFFICIENTS 67

16. Linear Homogeneous Equation with Constant Coefficients

An equation of the form

dny d n−1 y
(16.1) a0 n + a1 n−1 + . . . + an y = b(x).
dx dx
where a0 6= 0, a1 , . . . an are constants and b(x) is a function of x is a linear
differential equation. of nth order with constant coefficients. By dividing

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

a0 we can arrive at the equation of the same form with a0 replaced by 1.
We can always assume a0 = 1 in the equation. Let us take

dny d n−1 y
L(y) = a0 + a 1 + . . . + an y
dxn dxn−1
then equation (1) becomes

L(y) = b(x).

If b(x) = 0 for all x, then the corresponding equation L(y) = 0 is called

as homogeneous equation. If b(x) 6= 0 for some x, then L(y) = b(x) is
called a non-homogeneous equation. The operator L can be written as
n
L := ∑ ak Dn−k where Dk y = d k y/dxk .
k=0

In solving these equation we need the following theorem.

Theorem 3.9. If φ1 , φ2 , . . . , φm are solutions of homogeneous equation
L(y) = 0 and c1 , c2 , . . . , cn are constants, then φ = ∑m
i=1 ci φi is also a
solution of L(y) = 0.
Proof. Since φi is a solution of L(y) = 0, we have ∑nk=0 ak Dn−k φi = 0
and so
n n
∑ ak Dn−k ∑ ciφi


L(φ ) =
k=0 i=1
n n
∑ ak ∑ ciDn−k φi


=
k=0 i=1
n n
∑ ak Dn−k φi


= ∑ ci
i=1 k=0
n
= ∑ ci × 0 = 0.
i=1
68 3. DIFFERENTIAL EQUATIONS

Theorem 3.10. The function y = emx is a solution of the homogeneous

differential equation
n
(16.2) L(y) = ∑ ak Dn−k y = 0,
k=0

provided m satisfies the auxiliary equation ∑nk=0 ak mn−k = 0.

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Proof. If y = emx , then Dk y = mk emx . Using these in equation (16.2), we
get
n
L[emx ] = ( ∑ ak mn−k )emx = 0.
k=0

Hence emx is a solution of L(y) = 0 if the auxiliary equation ∑nk=0 ak mn−k =

0 is satisfied.
Corollary 3.11. If m1 , m2 , . . . mn are all distinct roots of the auxiliary
equation ∑nk=0 ak mn−k = 0, then y = c1 em1 x + c2 em2 x + . . . + cn emn x is
the general solution for the equation L(y) = ∑nk=0 ak Dn−k y = 0, where
c1 , c2 , . . . , cn are arbitrary constants.

Proof. Since m1 , m2 , . . . mn are all distinct roots of the auxiliary equation

∑nk=0 ak mn−k = 0, the functions em1 x , em2 x , . . . emn x are solutions of the
homogeneous equation L(y) = ∑nk=0 ak Dn−k y = 0. By previous theorem,
it follows that y = c1 em1 x + c2 em2 x + . . . + cn emn x is the general solution
of (16.2).

Example 3.12. Solve y00 − 3y0 + 2y = 0.

Since the auxiliary equation

m2 − 3m + 2 = 0

has roots m1 = 1, m2 = 2, the general solution is y = c1 ex + c2 e2x where

c1 , c2 are arbitrary constants. 

Corollary 3.13. Let m1 = α + iβ and m2 = α − iβ be two complex

roots of the auxiliary equation and m1 , m2 , m3 , . . . , mn be distinct. Then
the general solution is

y = eαx (c1 cos β x + c2 sin β x) + c3 em3 x + . . . cn emn x .

 16. LINEAR HOMOGENEOUS EQUATION WITH CONSTANT COEFFICIENTS 69

Proof. Since
C1 e(α+iβ )x +C2 e(α−iβ )x
= C1 eαx (cos β x + i sin β x) +C2 eαx (cos β x − i sin β x)
= eαx (c1 cos β x + c2 sin β x),
we can write the general solution as
y = eαx (c1 cos β x + c2 sin β x) + c3 em3 x + . . . cn emn x .

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Similarly, we have the following:
Corollary 3.14. Let m1 , m2 = α1 ± iβ1 and, m3 , m4 = α2 ± iβ2 be two
pair complex roots of the auxiliary equation and m1 , m2 , m3 , . . . , mn be
distinct. Then the general solution is

y = eα1 x (c1 cos β1 x + c2 sin β1 x) + eα2 x (c3 cos β2 x + c4 sin β2 x)

+ c5 em3 x + . . . cn emn x .

Example 3.15. Solve y00 + a2 y = 0.

The auxiliary equation m2 + a2 = 0 has roots given by m = ±ai.
Thus, the general solution is y = e0x (c1 cos ax + c2 sin ax) = c1 cos ax +
c2 sin ax, where c1 , c2 are arbitrary constants. 

Example 3.16. Solve y(4) − y = 0.

The auxiliary equation m4 − 1 = (m2 + 1)(m + 1)(m − 1) = 0 has
roots given by m = ±i, 1, and −1. Thus, the general solution is
y = (c1 cos x + c2 sin x) + c2 ex + c4 e−x ,
where c1 –c4 are arbitrary constants. 

Example 3.17. Solve y(4) + y00 + y = 0.

The auxiliary equation
m4 + m2 + 1 = (m2 + 1)2 − m2 = (m2 + m + 1)(m2 − m + 1) = 0
has roots given by
√
±1 ± 3i
m= .
2
Thus, the general solution is
 √ √   √ √ 
y = ex c1 cos( 3x) + c2 sin( 3x) +e−x c3 cos( 3x) + c4 sin( 3x) ,
where c1 –c4 are arbitrary constants. 
70 3. DIFFERENTIAL EQUATIONS

Corollary 3.18. If the auxiliary equation ∑nk=0 ak mn−k = 0 has k equal

roots, say, m1 = m2 = · · · = mk = m, and other roots mk+1 , · · · , mn
are distinct and different from m, then the solution of the differential
equation L(y) = ∑nk=0 ak Dn−k y = 0 is

y = (c1 + c2 x + · · · + ck xk−1 )emx + ck+1 emk+1 x + . . . + cn emn x ,

where c1 , c2 , . . . , cn are arbitrary constants.

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Proof. We consider the case k = 2 and the general case is similar. If we
consider the solution

y = (c1 + c2 )em1 x + c3 em2 x + . . . + cn emn x ,

it has only n − 1 arbitrary constants which is less than the order of the
differential equation. To get a solution with n arbitrary constants, we
first write the differential equation as

(D − m)(D − m)(D − m3 ) . . . (D − mn )y = 0.

Consider the equation (D − m)(D − m)y = 0. Let (D − m)y = u. Then

(D − m)u = 0 which has the solution u = c2 emx . Substituting u value in
(D − m)y = u, the equation becomes

(D − m)y = c2 emx .
R
Recall that
R the solution of y0 + P(x)y = Q(x) is given by y e P(x)dx =
Q(x)e Pdx dx +C, where C is a arbitrary constant. Using this, we get
R

Z
−mx
ye = c2 emx e−mx dx + c1 = c1 + c2 x

or

y = (c1 + c2 x)emx .

The solutions of (D − m3 ) . . . (D − mn )y = 0 are em3 x , . . . , emn x . Thus,

the general solution of (D − m)2 (D − m3 ) . . . (D − mn )y = 0 is given by

y = (c1 + c2 x)emx + c3 em3 x + . . . + cn emn x

where c1 –cn are arbitrary constants.

 16. LINEAR HOMOGENEOUS EQUATION WITH CONSTANT COEFFICIENTS 71

Example 3.19. Solve y000 − 3y00 + 3y0 − y = 0.

The auxiliary equation m3 − 3m + 3m − 1 = (m − 1)3 = 0 has a root
m = 1 of multiplicity 3. Therefore, the general solution is
y = (c1 + c2 x + c3 x2 )ex ,
where c1 , c2 , c3 are arbitrary constants. 

Example 3.20. Solve y(4) − 2y00 + y = 0.

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

The auxiliary equation
m4 − 2m2 + 1 = (m2 − 1)2 = (m + 1)2 (m − 1)2 = 0
has roots m = 1, −1, both of multiplicity 2. Thus, the general solution is
y = (c1 + c2 x)ex + (c3 + c4 x)e−x ,
where c1 –c4 are arbitrary constants. 

Example 3.21. Solve

d4y d2y
+ 6 + 9y = 0.
dx4 dx2
The auxiliary equation
m4 + 6m2 + 9 = (m2 + 3)2 = 0
has roots given by m = ±3i, ±3i. Therefore, the solution is given by
√ √
y = (c1 x + c2 ) cos 3x + (c3 x + c4 ) cos 3x
where c1 –c4 are arbitrary constants. 

Example 3.22. Solve

d2q dq 1
L 2
+R + q = 0
dt dt C
4L
given that R2 − =0
C
The auxiliary equation is given by Lm2 + Rm + C1 = 0 and its roots
are given by
q
−R ± R2 − 4L C −R
m= =
2L 2L
72 3. DIFFERENTIAL EQUATIONS

(we have used the given condition R2 − 4L

C = 0). Since the roots are real
and of multiplicity 2, the solution is
R
y = (c1 + c2t)e− 2L t

where c1 , c2 are arbitrary constants. 

Corollary 3.23. If the auxiliary equation of a differential equation hav-

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

ing real coefficients has a complex root α + iβ of multiplicity n, then
corresponding part of the general solution is

y = (c1 + c2 x + c3 x2 + · · · + cn xn−1 )eαx cos(β x)

+ (cn+1 + cn+2 x + · · · + c2n xn−1 )eαx sin(β x)

where c1 –c2n are arbitrary constants.

Proof. Since the auxiliary equation has a complex root α + iβ of multi-

plicity n, the solutions of the given differential equations are

e(α+iβ )x , xe(α+iβ )x , x2 e(α+iβ )x , . . . xn−1 e(α+iβ )x

and the corresponding conjugate root is α − iβ gives the solutions

e(α−iβ )x , xe(α−iβ )x , x2 e(α−iβ )x , x3 e(α−iβ )x , . . . xn−1 e(α−iβ )x .

Thus the corresponding part of the general solution is

y = (c1 + c2 x + c3 x2 + · · · + cn xn−1 )eαx cos(β x)

+ (cn+1 + cn+2 x + · · · + c2n xn−1 )eαx sin(β x)

where c1 –c2n are arbitrary constants.

Example 3.24. Solve y(4) + 2y00 + y = 0

The auxiliary equation is m4 + 2m2 + 1 = (m2 + 1)2 = 0. Since ±i
is a root of multiplicity 2, the general solution is

y = (c1 + c2 x) cos x + (c3 + c4 x) sin x

where c1 –c4 are arbitrary constants. 

 16. LINEAR HOMOGENEOUS EQUATION WITH CONSTANT COEFFICIENTS 73

16.1. Linearly Independent Solutions. (This section is omitted

for examination!)
The solutions of a differential equation ϕ1 , ϕ2 , . . . ϕn are said to be
linearly dependent over an interval I if there exists a set of n constants
c1 , c2 , . . . cn , at least one of which is different from zero, such that the
equation
c1 ϕ1 + c2 ϕ2 + . . . + cn ϕn = 0.

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

holds for all x in I. They are linearly independent over I if

c1 ϕ1 + c2 ϕ2 + . . . + cn ϕn = 0 implies c1 = c2 = . . . cn = 0.

If ϕ1 , ϕ2 , . . . ϕn are functions of x, then the Wronskian determinant,

denoted by W [ϕ1 , ϕ2 , . . . ϕn ], is defined as
ϕ1 ϕ2 ··· ϕn
ϕ10 ϕ20 ··· ϕn0
W [ϕ1 , ϕ2 , . . . ϕn ] = .. .. ..
. . ··· .
(n−1) (n−1) (n−1)
ϕ1 ϕ2 · · · ϕn

Theorem 3.25. If the roots m1 , . . . , mn of the auxiliary equations of the

differential equation L(y) = 0 are distinct, then solutions em1 x , em2 x , . . .,
emn x are linearly independent.
Proof. Suppose that the solutions em1 x , em2 x , . . ., emn x are linearly de-
pendent, then

(16.3) α1 em1 x + α2 em2 x + . . . + αn emn x = 0,

where at least one of αi 6= 0, say, αn 6= 0. Dividing equation (16.3) by

em1 x and differentiating with respect to x, we get

(16.4) α2 (m2 − m1 )e(m2 −m1 )x + . . . + αn (mn − m1 )e(mn −m1 )x = 0,

which is a linear equation in (n − 1) exponential functions of the form

e px . Again dividing equation (16.4) by e(m2 −m1 )x and differentiating with
respect to x, we obtain α3 (m2 − m1 )(m3 − m1 )e(m3 −m2 )x + . . . + αn (mA −
m1 )(mn − m2 )emn −m2 = 0. Continuing this process (n − 1) times, we get

αn (mn − m1 )(mn − m2 ) . . . (mn − mn−1 )e(mn −mn−1 )x = 0.

Since all the m0i s are distinct we have αn = 0, contradicting the assump-
tion that αn 6= 0. Thus, em1 x , em2 x , . . . emn x are linearly independent.
74 3. DIFFERENTIAL EQUATIONS

Theorem 3.26. If the functions of ϕ1 , ϕ2 , . . . ϕn are linearly dependent

over the interval I and if each functions possesses derivatives up to the
order n − 1, then the determinant W [ϕ1 , ϕ2 , . . . ϕn ] is equal to zero at all
points of I.
Proof. Given the functions ϕ1 , ϕ2 , . . . ϕn are linearly dependent over I,
there is a set of constants c1 , c2 , . . . cn , with at least one of them different
from zero, such that

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

(16.5) c1 ϕ1 + c2 ϕ2 + . . . + cn ϕn = 0
at all the points of I. By the hypothesis each of the functions ϕ posses
derivatives at least up to order n − 1. Hence, by repeated differentiation
of (16.5), we obtain the following equations
c1 ϕ10 + c2 ϕ20 + . . . + cn ϕn0 = 0
..
.
(n−1) (n−1) (n−1)
c1 ϕ1 + c2 ϕ2 + . . . cn ϕn = 0.
Equation (16.5) and the (n − 1) equations above constitute a set of n
homogeneous linear equations in the n quantities c1 , c2 , . . . cn , and this
system of equations is satisfied by a set of values at least one of which
is different from zero. Thus, the determinant of the coefficients of the
system is equal to zero. Hence the Wronskian determinant, being the
determinant formed by the coefficients of this system of equations, is
equal to zero.
Corollary 3.27. The solutions ϕ1 , ϕ2 , . . . , ϕn over the internal I of a
homogeneous equation L(y) = 0 are linearly dependent if and only if
Wronskian determinant of the functions ϕ1 , ϕ2 , . . . , ϕn is non-zero.

Example 3.28. The converse of the theorem need not be true if the func-
tions ϕ1 , · · · , ϕn are not the solutions of same linear equation L(y) = 0.
Consider the functions ϕ1 and ϕ2 defined by
(
−x2 x ≤ 0
ϕ1 = x2 ϕ2 2
x x > 0.
We will show that Wronskian of ϕ1 and ϕ2 is zero. Clearly ϕ10 (x) = 2x
for all value of x, while
(
−2x x ≤ 0
ϕ20 (x) =
2x x>0
 17. LINEAR NON-HOMOGENEOUS EQUATION OF ORDER n 75

Therefore for x ≤ 0, the Wronskian of ϕ1 , ϕ2 is

x2 −x2
W (ϕ1 , ϕ2 ) = =0
2x −2x
and for x > 0 the Wronskian is

x2 x2
W (ϕ1 , ϕ2 ) = = 0.
2x 2x

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

However the functions ϕ1 and ϕ2 are linearly independent over any in-
terval. In fact, if
c1 ϕ1 + c2 ϕ2 = 0,
then

c1 ϕ1 (1) + c2 ϕ2 (1) = c1 + c2 = 0,
c1 ϕ1 (1) + c2 ϕ2 (−1) = c1 − c2 = 0.

Hence c1 = c2 = 0 and ϕ1 , ϕ2 are linearly independent. 

Theorem 3.29. If the Wronskian of ϕ1 and ϕ2 vanishes identically over

the interval I and if one of these functions is different from zero at all
points of I, then the functions are linearly dependent on I
Proof. Suppose W (ϕ1 , ϕ2 ) = ϕ1 ϕ20 − ϕ2 ϕ10 ≡ 0 on I and at least one of
the function say ϕ1 is different from zero at all points of I. Hence we
may divide the Wronskian relation by ϕ12 to get
 0
ϕ2 ϕ1 ϕ20 − ϕ2 ϕ10
= ≡0
ϕ1 ϕ12

Therefore, ϕ2 /ϕ1 = k which shows that ϕ1 and ϕ2 are linearly depen-

dent.

17. Linear Non-homogeneous Equation of Order n

An nth order linear non-homogeneous equation is of the form
n
(17.1) L(y) = ∑ ak Dn−k y = b(x)
k=0

where b(x) 6= 0. Given the non-homogeneous equation L(y) = b(x), the

equation L(y) = 0 is the corresponding homogenous equation.
76 3. DIFFERENTIAL EQUATIONS

Theorem 3.30. If y = y p is a particular solution of the non-

homogeneous equation L(y) = b(x), then any solution y of L(y) = b(x) is
y = yc + y p where y = yc is a solution of the corresponding homogenous
equation L(y) = 0.
The function yc is called as the complementary function of L(y) =
b(x).

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Proof. Since L(y p ) = b(x) and L(y) = 0, we have

L(y − y p ) = L(y) − L(y p ) = b(x) − b(x) = 0.

This proves that yc = y − y p is a solution of L(y) = 0.

Theorem 3.31. If y = y p1 , y p2 are respectively a particular solution of

the non-homogeneous equation L(y) = b1 (x) and L(y) = b2 (x), then any
solution y of L(y) = b1 (x) + b2 (x) is y = yc + y p1 + y p2 where y = yc is a
solution of the corresponding homogenous equation L(y) = 0.

Proof. Similar to the proof of Theorem 3.30

We denote by D the differential operator d/dx and define D−1 as the

inverse of this operator D; in this case D−1 is the integration operator.
In general, for an operator F(D), the inverse operator 1/F(D) is defined
by
1 1
F(D)X = X = F(D) X
F(D) F(D)

for any appropriate X. For example, let us consider the operator D − α.

Let
1
X = y.
D−α
Then (D − α)y = X and this gives y = eαx e−αx Xdx and therefore we
R

have
1
Z
X = eαx e−αx Xdx.
D−α
The following theorem gives rules to find particular integral when
b(x) = eax .
 17. LINEAR NON-HOMOGENEOUS EQUATION OF ORDER n 77

Theorem 3.32. The particular integral of the differential equation Ly =

eax where L = F(D) = ∑nk=0 ak Dn−k has constant coefficients is given by

eax eax
yp = =
F(D) F(a)

provided F(a) 6= 0. If F(a) = 0, then

eax xm eax

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

yp = =
G(D)(D − a)m m!G(a)

where F(D) = G(D)(D − a)m .

Proof. Since Dn eax = an eax , we have F(D)eax = F(a)eax and therefore
the function y p = eax /F(a) satisfies F(D)y p = F(D)(eax /F(a)) = eax .
This proves that y p = eax /F(a) is the particular integral of Ly = eax .
If F(a) = 0, then F(D) = G(D)(D − a)m for some integer m > 0
where G(a) 6= 0. Note that

(D − a)(xm eax ) = (mxm−1 + axm )eax − axm eax = mxm−1 eax

and, by mathematical induction,

(D − a)m (xm eax ) = m!eax .

Therefore

F(D)(xm eax ) = G(D)(D − a)m (xm eax ) = G(D)(m!eax ) = m!G(a)eax

or
xm eax
 
F(D) = eax
m!G(a)
Therefore, the function y p given by
eax eax
yp = =
F(D) m!G(a)
is the particular integral Ly = eax .

Example 3.33. Solve

d2y dy
− 4 + 3y = e−2x .
dx2 dx
78 3. DIFFERENTIAL EQUATIONS

Since the auxiliary equation m2 − 4m + 3 = 0 has roots m = 1, 3,

the complementary function yc = c1 ex + c2 e3x . The particular integral
is given by
e−2x e−2x e−2x
yp = = = .
D2 − 4D + 3 (−2)2 − 4(−2) + 3 15
Therefore, the complete solutions is given by
e−2x

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

y = yc + y p = c1 ex + c2 e3x +
15
where c1 and c2 are arbitrary constants. 
Note that the particular integral of Ly = sin ax, cos ax can be found
by writing sin ax, cos ax as the imaginary and real part of eiax .
Example 3.34. Solve (D2 + 2D + 2)y = ex + cos x.
Since the auxiliary equation m2 + 2m + 2 = 0 has roots m = −1 ± i,
the complementary function is given by
yc = e−x (c1 cos x + c2 sin x).
The particular integral y p1 corresponding to ex is given by
ex ex
y p1 = = .
D2 + 2D + 2 5
The particular integral y p2 corresponding to cos x is given by
eix
y p2 = Re
D2 + 2D + 2
eix
= Re
−1 + 2i + 2
eix
= Re
1 + 2i
(1 − 2i)eix
= Re
5
1
= (cos x + 2 sin x).
5
Therefore, the complete solution is given by
y = yc + y p1 + y p2
−x ex 1
= e (c1 cos x + c2 sin x) + + (cos x + 2 sin x)
5 5
where c1 and c2 are arbitrary constants. 
 17. LINEAR NON-HOMOGENEOUS EQUATION OF ORDER n 79

Alternately, we can use the following method to find particular in-

tegral when b(x) = sin ax or cos ax. Since D2 (sin ax) = −a2 sin ax and
D2 (cos ax) = −a2 cos ax, we have

sin ax sin ax cos ax cos ax

= and = .
F(D2 ) F(−a2 ) F(D2 ) F(−a2 )

Example 3.35. Find particular integral of (D2 + 2D + 2)y = cos x.

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

We have
cos x
yp =
D2 + 2D + 2
cos x
=
−1 + 2D + 2
cos x
=
2D + 1
(2D − 1) cos x
=
4D2 − 1
(−2 sin x − cos x)
=
4(−1) − 1
cos x + 2 sin x
= . 
5
If b(x) = xn , then we use the following method: Write 1/F(D) =
a0 + a1 D + · · · and find the values of ai ’s by formally finding the product
F(D)(a0 + a1 D + · · · ) = 1 and comparing the like powers of D. Then
we get
xn
= (a0 + a1 D + · · · )xn
F(D)
= a0 xn + a1 nxn−1 + a2 n(n − 1)xn−2 + · · · + an n!.

When c 6= 0, we have the following:

1
aD2 + bD + c
  −1
1 b a 2
= 1+ D+ D
c c c
!
a 2 2 a 2 3
     
1 b a 2 b b
= 1− D+ D + D+ D − D+ D +...
c c c c c c c
80 3. DIFFERENTIAL EQUATIONS

a 2 b2 2 2ab 3 b3 3
 
1 b
= 1− D− D + 2 D + 2 D − 3 D +...
c c c c c c
b2 − ac 2 2abc − b3 3
 
1 b
= 1− D+ D + D +...
c c c2 c3

Example 3.36. Solve (D2 + 1)y = x3 .

The complementary function is given by yc = c1 cos x + c2 sin x. The

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

particular integral y p2 corresponding to x3 is given by

x3
y p1 = 2
= (1 − D2 + D4 − · · · )x3 = x3 − 6x.
1+D
Therefore the complete solution is

y = yc + y p = c1 cos x + c2 sin x + x3 − 6x

where c1 and c2 are arbitrary constants. 

Example 3.37. Solve (D2 − 3D + 2)y = x3 − 3x2 .

The complementary function is given by yc = c1 ex + c2 e2x . The par-
ticular integral y p2 corresponding to x3 − 3x2 is given by

x3 − 3x2
y p1 =
D2 − 3D + 2
 
1 3 7 2 15 3
= 1+ D+ D + D +···
2 2 4 8
 
1 3 2 3 2 7 15
= x − 3x + (3x − 6x) + (6x − 6) + (6)
2 2 4 8
1 3 3 3
= x3 + x2 + x +
2 4 4 8
Therefore the complete solution is
1 3 3 3
y = yc + y p = c1 ex + c2 e2x + x3 + x2 + x +
2 4 4 8
where c1 and c2 are arbitrary constants. 
We now consider b(x) = eaxV (x). Since

D(eax X) = eax DX + aeax X = eax (D + a)X,

 17. LINEAR NON-HOMOGENEOUS EQUATION OF ORDER n 81

by mathematical induction, we have

Dn (eax X) = eax (D + a)n X.

This gives

F(D)(eax X) = eax F(D + a)X

and therefore

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

eax F(D + a)X
eax X = .
F(D)
Writing F(D + a)X = V , we get

eaxV (x) V (x)

= eax .
F(D) F(D + a)

Example 3.38. Solve (D2 + 2D + 5)y = xex .

The auxiliary equation m2 + 2m + 5 = 0 has roots −1 ± 2i and so
the complementary function is given by

yc = e−x (c1 cos 2x + c2 sin 2x).

The particular integral y p2 corresponding to xex is given by

xex
y p1 = 2
D + 2D + 5
x
= ex
(D + 1)2 + 2(D + 1) + 5
x
= ex 2
D + 4D + 8
 
x1 1
=e 1− D+··· x
8 2
1 1
= (x − )ex .
8 2
Therefore the complete solution is
 
−x x 1
y = e (c1 cos 2x + c2 sin 2x) + − ex . 
8 16
82 3. DIFFERENTIAL EQUATIONS

Example 3.39. Solve (D2 + 2D + 1)y = ex sin x.

The auxiliary equation m2 + 2m + 1 = 0 has roots −1, −1 and so the
complementary function is given by

yc = (c1 + c2 x)e−x .

The particular integral y p2 corresponding to ex sin x is given by

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

ex sin x
y p1 =
D2 + 2D + 1
sin x
= ex 2
(D + 1) + 2(D + 1) + 1
sin x
= ex 2
D + 4D + 4
sin x
= ex
−1 + 4D + 4
(4D − 3) sin x
= ex
14D2 − 9
4 cos x − 3 sin x
= ex
−25
3 sin x − 4 cos x
= ex .
25
Therefore the complete solution is

3 sin x − 4 cos x x
y = (c1 + c2 x)e−x + e. 
25

18. The Euler-Cauchy Differential Equation

The Euler-Cauchy equation

dny n−1 d
n−1 y dy
(18.1) a0 x n n
+ a 1 x n−1
+ . . . an−1 x + an y = f (x).
dx dx dx
has variable coefficients and it can always be transformed into a linear
equation with constant coefficients by using simple change of indepen-
dent variable |x| = ez .
Let
d d
D= , and D = .
dx dz
 18. THE EULER-CAUCHY DIFFERENTIAL EQUATION 83

Then, for the first derivative, we have

dy dy dz 1
Dy = = · = Dy
dx dz dx x
or
xDy = Dy.
Differentiating with respect to x, we have

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

1 1
xD2 y + Dy = DDy = (xD)Dy = D 2 y.
x x
Therefore, using xDy = Dy, we have
x2 D2 y = (D 2 − D)y = D(D − 1)y.
Proceeding this way, we get
(18.2) xn Dn y = D(D − 1) . . . (D − n + 1)y.
We can prove this by mathematical induction. The formula (18.2) is
true for n = 1 and n = 2. We assume it to be true for n = k, that is,
(18.3) xk Dk y = D(D − 1) . . . (D − k + 1)y.
and prove it to be true for n = k + 1. By differentiating both sides of this
equation with respect to x, and using D = 1x D, we get
1
D(xk Dk y) = D(D(D − 1) . . . (D − k + 1)y)
x
or
1
xk Dk+1 y + kxk−1 Dk y = DD(D − 1) . . . (D − k + 1)y.
x
Multiplying by x, and applying the inductive hypothesis (18.3) to the
second term on the left and rearranging the result, we have
xk+1 Dk+1 y = DD(D − 1) . . . (D − k + 1)y − kD(D − 1) . . . (D − k + 1)y
= D(D − 1) . . . (D − k + 1)(D − k)y.
This proves the result is true for n = k + 1.
Using the substitution |x| = ez and (18.2), the equation (18.1) be-
comes a differential equation with constant coefficients. We then solve
this transformed differential equation. Again by substituting z by log |x|,
we get the solution of (18.1).
84 3. DIFFERENTIAL EQUATIONS

Example 3.40. Solve x2 y00 + xy0 + 9y = 0.

Let |x| = ez . Then z = log |x|, xy0 = D and x2 y00 = (D 2 − D). Using
these we see that the given equation becomes

D(D − 1)y + Dy + 9y = 0

or (D 2 + 9)y = 0. Therefore, the solution is given by

y = c1 cos 3z + c2 sin 3z = c1 cos(3 log |x|) + c2 sin(3 log |x|). 

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Example 3.41. Find a complete solution of

d3y 2
2d y dy
x3 + 4x − 5x − 15y = x4 .
dx3 dx2 dx
Let |x| = ez . Then z = log|x| and
dy
x = Dy,
dx
d2y
x2 2 = D(D − 1)y,
dx
d3y
x3 3 = D(D − 1)(D − 2)y.
dx
Using these we see that the given equation becomes

[D(D − 1)(D − 2) + 4D(D − 1) − 5Dy − 15]y = e4z

or

(D 3 + D 2 − 7D − 15)y = e4z .

The auxiliary equation m3 + m2 − 7m − 15 = (m − 3)(m2 + 4m + 5) = 0

has roots m = 3, −2 ± i and so the complementary function yc is given
by
yc = c1 e3z + e−2z (c2 cos z + c3 sin z).
The particular integral is given by

e4z
yp =
D 3 + D 2 − 7D − 15
e4z
= 3
4 + 42 − 7 × 4 − 15
 18. THE EULER-CAUCHY DIFFERENTIAL EQUATION 85

e4z
= .
37
Therefore the complete solution is

e4z
y = c1 e3z + e−2z (c2 cos z + c3 sin z) + .
37
Replacing z by log |x|, we get the solution given equation as

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

e4 log |x|
y = c1 e3 log |x| + e−2 log |x| (c2 cos(log |x|) + c3 sin(log |x|)) +
37
1 x4
= c1 |x|3 + (c 2 cos(log |x|) + c3 sin(log |x|)) + . 
x2 37

Example 3.42. Solve (x2 D2 + 4xD + 2)y = x2 + x12 .

Put |x| = ez or z = log |x|. Then xD = D and x2 D2 = D(D − 1).
Therefore the given equation becomes

(D(D − 1) + 4D + 2)y = e2z + e−2z

or

(D 2 + 3D + 2) y = e2z + e−2z .

The complementary function is given by

yc = c1 e−z + c2 e−2z .

The particular integral corresponding to e2z is given by

e2z
y p1 =
D 2 + 3D + 2
e2z
= 2
2 +3×2+2
e2z
= .
12
and the particular integral corresponding to e−2z is given by

e−2z
y p2 =
D 2 + 3D + 2
86 3. DIFFERENTIAL EQUATIONS

e−2z
=
(D + 2)(D + 1)
ze2z
=− .
3
The general solution y = yc + y p1 + y p2 is given by
e2z ze2z
y = c1 e−z + c2 e−2z +
−

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

12 3
c1 c2 x 2 log |x|
= + 2+ − 2 . 
|x| x 12 x

Example 3.43. Solve (x2 D2 − 2xD − 4)y = 32(log x)2 .

Put x = ez or z = log |x|. Then xD = D and x2 D2 = D(D − 1).
Therefore the given equation becomes
(D 2 − 3D − 4)y = 32z2 .
The auxiliary equation m2 − 3m − 4 = 0 has roots m = 4, −1 and so the
complementary function is given by
yc = c1 e4z + c2 e−z .
The particular integral is given by
32z2
yp =
D 2 − 3D − 4
 
1 3 13 2
=− 1 − D + D + · · · (32z2 )
4 4 16
= −8z2 + 12z − 13.

y = c1 e4z + c2 e−z − 8z2 + 12z − 13

c1 c2
= 4 + − 8(log |x|)2 + 12(log |x|) − 13.
x |x|


Example 3.44. Solve (x3 D3 + 3x2 D2 + xD + 1)y = sin(log x).

Putting |x| = ez and z = log x, we have xD = D, x2 D2 = D(D − 1),
x3 D3 = D(D − 1)(D − 2). The given equation becomes
(D 3 + 1)Y = sin z.
 19. METHOD OF VARIATION OF PARAMETERS 87

3 + 1 = (m + 1)(m2 − m + 1) = 0 has roots

The auxiliary equation m√
given by m = −1 or (1 ± 3i)/2 and so the complementary function is
given by

√ √ !
3 3
yc = c1 e−z + e1/2z c2 cos z + c3 sin .
2 2

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

The particular integral is given by

sin z
yp =
D3 + 1
sin z
=
−D + 1
(D + 1) sin z
=
1 − D2
cos z + sin z
= .
2

Therefore, the complete solution is given by

√ √ !
3 3 cos z + sin z
y = c1 e−z + e1/2z c2 cos z + c3 sin +
2 2 2
√ ! √ !
c1 1 3 1 3
= + c2 |x| 2 cos log |x| + c3 |x| 2 sin log |x|
|x| 2 2
1
+ [sin(log |x|) + cos(log |x|)]. 
2

19. Method of Variation of Parameters

By using method of variation of parameters we can find complete
solution of nonhomogeneous linear equation. The method for second
order equation is given in the following theorem.
88 3. DIFFERENTIAL EQUATIONS

Theorem 3.45. If y1 and y2 are linearly independent solutions of

(19.1) y00 + P1 (x)y0 + P2 (x)y = 0,

then the solution of

(19.2) y00 + P1 (x)y0 + P2 (x)y = R(x)

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

is given by
y = (u1 + c1 )y1 + (u2 + c2 )y2
where
y2 R(x) y1 R(x)
Z Z
(19.3) u1 = − dx, and u2 = dx
y1 y02 − y2 y01 y1 y02 − y2 y01

Proof. Since y1 and y2 are solutions of the equation (19.1), the function
y = u1 y1 + u2 y2 is the general solution of (19.1) where u1 and u2 are
constant. The idea is to find functions u1 and u2 such that y = u1 y1 +u2 y2
is a particular integral for equation (19.2). Note that

y0 = u1 y01 + u2 y02 + u01 y1 + u02 y2 .

Another differentiation will clearly introduce second derivatives of the

unknown functions u1 and u2 and so we assume that

(19.4) u01 y1 + u02 y2 = 0.

Therefore

y0 = u1 y01 + u2 y02 and y00 = (u1 y001 + u01 y01 ) + u2 y002 + u02 y02

Substituting y, y0 , y00 into equation (19.2), we get

u1 [y001 + P1 (x)y01 + P2 (x)y1 ] + u2 [y002 + P1 (x)y02 + P2 (x)y2 ]

+ u01 y01 + u02 y02 = R(x).

The expression in the brackets vanish because y1 and y2 are the solution
of the homogeneous equation (19.2). Hence we get

(19.5) u01 y01 + u02 y02 = R(x).

 19. METHOD OF VARIATION OF PARAMETERS 89

Since y1 and y2 are linearly independent, we have y02 y1 − y2 y01 6= 0. Solv-

ing (19.4) and (19.5), we get

y2 R(x) y1 R(x)
(19.6) u01 = − and u02 = .
y1 y02 − y2 y01 y1 y02 − y2 y01

Therefore, the complete solution is given by

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

y = (u1 + c1 )y1 + (u2 + c2 )y2

where
y2 R(x) y1 R(x)
Z Z
u1 = − dx, and u2 = dx.
y1 y02 − y2 y01 y1 y02 − y2 y01

Example 3.46. Solve y00 + y = sec x.

The solution of y00 + y = 0 is given by y = c1 cos x + c2 sin x. To solve
the given equation, we compute u01 , u02 using (19.6):

− sin x sec x cos x sec x

u01 = = − tan x, and u02 = = 1.
cos2 x + sin2 x cos2 x + sin2 x
Therefore
Z
u1 = − tan dx = log(cos x), and u2 = x.

The complete solution is

y = c1 cos x + c2 sin x + cos x log(cos x) + x sin x. 

Example 3.47. Solve y00 + a2 y = tan ax.

The solution of y00 + a2 y = 0 is given by y = c1 cos ax + c2 sin ax. To
solve the given equation, we compute u01 , u02 using (19.6):

− sin ax tan ax
u01 =
a cos2 ax + a sin2 ax
1 sin2 ax
=−
a cos ax
1
= (cos ax − sec ax),
a
90 3. DIFFERENTIAL EQUATIONS

and
cos ax tan ax
u02 =
a cos2 ax + a sin2 ax
1
= sin ax.
a
Therefore

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

1
Z
u1 = (cos ax − sec ax)dx
a
1
= 2 (sin ax − log(sec ax + tan ax)),
a
and
1
u2 = − cos ax.
a2
The complete solution is

y = c1 cos ax + c2 sin ax
1 1
+ 2 (sin ax − log(sec ax + tan ax)) cos ax − 2 cos ax sin ax
a a
= c1 cos ax + c2 sin ax − log(sec ax + tan ax). 
 4. Laplace Transform

20. Laplace transform: examples and existence

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

The integral of the function f : [0, ∞) → R is the limit of the integral
over a finite interval [0, a] as a → ∞:
Z ∞ Z a
f (t) dt = lim f (t) dt.
0 a→∞ 0

Definition 4.1. The Laplace transform of a function f : [0, ∞) → R,

denoted by L [ f (t)], is defined by
Z ∞
L [ f (t)] := e−st f (t)dt
0

if it exists.

Example 4.2. The Laplace transform of eat is given by

Z ∞ Z ∞
L [eat ] = e−st eat dt = e−(s−a)t dt
0 0
∞
e−(s−a)t 1
=− =
s−a s−a
0

provided s > a. In particular, L [1] = 1/s for s > 0. 

Example 4.3. The Laplace transform of f (t) = t a , a > −1 is given by

Z ∞
L [t ] =
a
e−st t a dt
0
1
Z ∞
= e−u ua+1−1 du
sa+1 0
Γ(a + 1)
=
sa+1
91
92 4. LAPLACE TRANSFORM

for s > 0. (Recall that the gamma function Γ is defined by Γ(x) :=

R ∞ x−1 −u
0 u e du and, for any natural number n, Γ(n + 1) = n!.) In partic-
ular, we have
1
L [1] = ,
s
1
L [t] = 2
s

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

and more generally, for any non-negative integer n,
n!
L [t n ] = .
sn+1
Also, we have
1  π 1/2  π 1/2
L [t 1/2 ] = , L [t −1/2 ] = .
2 s3 s


Example 4.4. Using the formula

ebt 
Z 
cos(at)ebt dt = b cos(at) + a sin(at) ,
a2 + b2
we see that the Laplace transform of cos(at) is given by
Z ∞
L [cos(at)] = e−st cos(at) dt
0
e−st   ∞
= 2 − s cos(at) + a sin(at)
s + a2 0
s
= 2
s + a2
for s > 0. Note that, for s > 0, |e−st cos(at)| ≤ e−st → 0 as t → ∞ and
so e−st cos(at) → 0 as t → ∞. Similarly, e−st sin(at) → 0 as t → ∞.
Similarly, using
ebt 
Z 
sin(at)ebt dt = 2 b sin(at) − a cos(at) ,
a + b2
we see that the Laplace transform of sin(at) is given by
a
Z ∞
L [sin(at)] = e−st sin(at) dt =
0 s2 + a2
for s > 0. 
 20. LAPLACE TRANSFORM: EXAMPLES AND EXISTENCE 93

Example 4.5. For 0 < a < ∞, the unit step function (or Heaviside func-
tion) ua : [0, ∞) → R is the function defined by
(
0 (t < a)
ua (t) =
1 (t ≥ a).

For s > 0, the Laplace transform of ua is given by

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Z ∞
L [ua (t)] = e−st f (t) dt
Z0∞
= e−st dt
a
e−as
= .
s


Definition 4.6. The function f : [a, b] → R is piecewise continuous

if there are finite number of points ti with a = t0 < t1 < t2 < · · · <
tn = b such that the function is continuous on each interval (ti−1 ,ti ),
i = 1, 2, · · · , n, and has finite limits as t → ti , i = 0, 2, · · · , n. In other
words, the function f is piecewise continuous if it is continuous at all
points except for a finite number of points where the function has fi-
nite discontinuity. A function f : [0, ∞) → R is piecewise continuous on
[0, ∞) if it is piecewise continuous in [0, a] for all a > 0.

Example 4.7. The function f : [0, ∞) → R defined by

(
t 2, 0 < t ≤ 1,
f (t) =
1 + t, 1 < t < ∞

is piecewise continuous. 

Example 4.8. The function f : [0, ∞) → R defined by

 1 , 0 ≤ t < 1,


f (t) = 1 − t
1+t 2 ≤ t∞,


is not piecewise continuous as f (t) → ∞ as t → 1−. 

94 4. LAPLACE TRANSFORM

Definition 4.9. The function f : [0, ∞) → R is of exponential order if

there are constants M, and α satisfying

| f (t)| ≤ Meαt (t ≥ 0).

Example 4.10. Any polynomial p(t) = ∑nk=0 akt k is of exponential or-

der. Since, for t ≥ 0,

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

tk
∞
tk
et = ∑ k! k! =⇒ t k ≤ k!et
≥
k=0

and so
!
n n
p(t) = ∑ akt k ≤ ∑ k!ak et = Met
k=0 k=0

where M := ∑nk=0 k!ak . 

A function f : [0, ∞) → R is periodic with period c if f (t + c) = f (t)
for all t ≥ 0.
Example 4.11. Every bounded function f : [0, ∞) → R is of exponential
order. For, if f is bounded by M, then | f (t)| ≤ M ≤ Met . Also, every
continuous periodic function is of exponential order as t → ∞. This
follows because such functions, being continuous, are bounded on [0, c]
and so on entire [0, ∞). 

2
Example 4.12. The function f : [0, ∞) → R defined by f (t) = et is not
of exponential order. 

Theorem 4.13 (Existence of Laplace transform). If the function f :

[0, ∞) → R is piecewise continuous and is of exponential order, then the
Laplace transform of the function f exists for s > α, where α is a real
number that depends on f .
Proof. Since the function f : [0, ∞) → R is of exponential order, there
are constants M, and α satisfying

| f (t)| ≤ Meαt (t ≥ 0).

We note that, for t ≥ 0,

|e−st f (t)| ≤ Me−(s−α)t .

 21. PROPERTIES: LINEARITY AND SHIFTING 95

Thus, for s > α, we have

M
Z ∞ Z ∞ Z ∞
−st −(s−α)t
| f (t)e | dt ≤ M e dt ≤ M e−(s−α)t dt = .
0 0 0 s−α
This proves that the integral converges absolutely for s > c. Thus, the
Laplace transform of f exists for s > α.
The functions s/(s + 1), sin(as), s2 /(1 + s2 ) are not Laplace trans-

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

form of any function.
The above condition is not necessary. As an example, consider the
function f : [0, ∞) → R defined by f (t) = t a where −1 < a < 0. Since
f (t) → ∞ as t → 0+, the function f is not piecewise continuous in [0, ∞).
However, for s > 0, we have already seen that

Γ(a + 1)
L [t a ] = .
sa+1
Corollary 4.14. If the function f : [0, ∞) → R is piecewise continuous,
is of exponential order, and L [ f (t)] = F(s), then F(s) → 0 as s → ∞.
Proof. Under this assumption, it was shown that
M
Z ∞
|F(s)| ≤ e−st | f (t)| dt ≤ , s > α.
0 s−α
Letting s → ∞, we see that |F(s)| → 0 and hence F(s) → 0 as s → ∞.

21. Properties: linearity and shifting

Theorem 4.15 (Linearity property). If the Laplace transforms of the
functions f , g : [0, ∞) → R exist, then the Laplace transform of α f + β g
exists for any two real number α, β and

L [α f (t) + β g(t)] = αL [ f (t)] + β L [g(t)].

Proof. The Laplace transform is linear follows from the linearity of the
integral:
Z ∞
L [a f (t) + bg(t)] = [a f (t) + bg(t)]e−st dt
0
Z ∞ Z ∞
−st
=a f (t)e dt + b g(t)e−st dt
0 0
= aL [ f (t)] + bL [g(t)].
96 4. LAPLACE TRANSFORM

Example 4.16. Using linearity of Laplace transform, we have

L [(1 + t)2 ] = L 1 + 2t + t 2
 

1 2 2
= + 2+ 3
s s s
2
s + 2s + 2
= .
s3

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran



Example 4.17. Using linearity of Laplace transform, we have

 
1 at −at
L [cosh(at)] = L (e + e )
2
1 
= L (eat ) + L (e−at )
2 
1 1 1
= +
2 s−a s+a
s
= 2 .
s − a2
Similarly,
 
1 at −at
L [sinh(at)] = L (e − e )
2
 
1 1 1
= −
2 s−a s+a
a
= 2 .
s − a2


Example 4.18. Since sin2 t = (1 − cos 2t)/2, we have

1
L [sin2 (at)] = (L [1 − cos(2at)])
2
1
= (L [1] − L [sin(2at)])
2 
1 1 s
= −
2 s s2 + 4a2
2a2
= .
s(s2 + 4a2 )
 21. PROPERTIES: LINEARITY AND SHIFTING 97

Also, using cos2 (at) = 1 − sin2 (at),

L [cos2 (at)] = L [1 − sin2 (at)]

1 2a2
= − 2
s s(s + 4a2 )
s2 + 2a2
= .
s(s2 + 4a2 )

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran



Example 4.19. Since sin3 t = (3 sint − sin 3t)/4, we have

1
L [sin3 (at)] = (L [3 sin(at) − sin(3at)])
4
1
= (3L [sin(at)] − L [sin(3at)])
4 
1 a 3a
= 3 2 −
4 s + a2 s2 + 9a2
6a3
= .
(s2 + a2 ) (s2 + 9a2 )

Theorem 4.20 (First Shifting Theorem). If the Laplace transform
F(s) := L [ f (t)] of f : [0, ∞) → R exists for s > α, then the Laplace
transform of eat f (t) exists for s > a + α and

L [eat f (t)] = F(s − a).

Proof. Since F(s) = L [ f (t)] exist for s > k, we have, for s − a > k,
Z ∞ Z ∞
L [eat f (t)] = e−st eat f (t) dt = e−(s−a)t f (t) dt = F(s − a).
0 0

Example 4.21. For b > −1, we have

Γ(b + 1)
L [t b ] =
sb+1
and therefore
Γ(b + 1)
L [eat t b ] = .
(s − a)b+1
98 4. LAPLACE TRANSFORM

Also,
1 
L [t b cosh(at)] = L [t b eat + t b e−at ]
2  
Γ(b + 1) 1 1
= + .
2 (s − a)b+1 (s − a)b+1


MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Example 4.22. Since
b s
L [sin(bt)] = and L [cos(bt)] =
s2 + b2 s2 + b2
we have
b s−a
L [eat sin(bt)] = and L [eat sin(bt)] = .
(s − a)2 + b2 (s − a)2 + b2

Theorem 4.23 (Second Shifting Theorem). If the Laplace transform
of f : [0, ∞) → R exists, then the Laplace transform of f (t − a)ua (t)
exists and
L [ f (t − a)ua (t)] = e−as L [ f (t)]
where ua is unit step function.
Proof. From the definition of Laplace transform, we have
Z ∞
L [ua (t) f (t − a)] = e−st ua (t) f (t − a) dt
0
Z ∞
= e−st f (t − a) dt
a
and, using the substitution t − a = u,
Z ∞
−as
L [ua (t) f (t − a)] = e e−su f (u) du
0
= e−as L [ f (t)].

Example 4.24. Let g : [0, ∞) → R be given by g(t) = 0 for 0 ≤ t ≤ 2

and g(t) = 3 for t > 2. Then g(t) equals 3u2 (t). Hence, using the second
shifting theorem,
L [g(t)] = L [3u2 (t)] = e−2s L [3] = 3e−2s /s.

 21. PROPERTIES: LINEARITY AND SHIFTING 99

Example 4.25. Let g : [0, ∞) → R be given by g(t) = 0 for 0 ≤ t ≤ 1

and g(t) = t for t > 1. Then g(t) equals tu1 (t). Hence, using the second
shifting theorem,

L [g(t)] = L [tu1 (t)]

= L [(t − 1)u1 (t)] + L [u1 (t)]
e−s
= e−s L [t] +

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

s
e−s e−s
= +
s2 s
(s + 1) −s
= e .
s2


Example 4.26. Let g : [0, ∞) → R be given by

(
e−t 0 ≤ t ≤ 4
g(t) = .
0 t >4

Then g(t) equals e−t (1 − u4 (t)). Hence, using the second shifting theo-
rem,

L [g(t)] = L [e−t (1 − u4 (t))]

= L [e−t ] − L [e−(t−4)−4 u4 (t)]
1
= − e−4 e4s L [e−t ]
s+1
1 e4(s−1)
= −
s+1 s+1
1−e 4(s−1)
= .
s+1

Theorem 4.27. If the Laplace transform of f : [0, ∞) → R exists for
s > α, and F(s) = L f (t), then the Laplace transform of f (at) exists
for s > aα and
1 s
L [ f (at)] = F( ).
a a
100 4. LAPLACE TRANSFORM

Example 4.28. Since

1
L [sin(t)] =
s2 + 1
we have
1 1 a
L [sin(at)] = = 2 .
a (s/a) + 1 s + a2
2

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran



22. Properties: derivatives and integrals

Theorem 4.29 (Transform of derivative). If f : [0, ∞) → R is differ-
entiable, f 0 is piecewise continuous on any finite interval in [0, ∞) and
f is of exponential order, then the Laplace transform of f 0 exists and

L [ f 0 (t)] = sL [ f (t)] − f (0).

Proof. We prove this in the case when f 0 is continuous in [0, ∞). In this
case,
Z a Z a
−st 0 −st a
e f (t) = e f (t) 0
+s e−st f (t) dt
0 0
Z a
−sa
(22.1) =e f (a) − f (0) + s e−st f (t) dt
0

Since f is of exponential order, there are constants M, and α satisfying

| f (t)| ≤ Meαt (t ≥ 0).

This implies, for s > α,

|e−st f (t)| ≤ Me−(s−α)t → 0 (as t → ∞).

Using this and by letting a → ∞ in (22.1), we get

L [ f 0 (t)] = sL [ f (t)] − f (0).

Example 4.30. When f (t) = eat , we have f 0 (t) = aeat and hence the
formula

L [ f 0 (t)] = sL [ f (t)] − f (0)

 22. PROPERTIES: DERIVATIVES AND INTEGRALS 101

becomes

aL [eat ] = sL [eat ] − 1

and this gives

1
L [eat ] = .
s−a


MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Theorem 4.31. Let f : [0, ∞) → R and its derivatives f 0 , f 00 , . . ., f (n−1)
be continuous on [0, ∞). If f is of exponential order (of order α), and
f (n) is piecewise continuous on every finite interval in [0, ∞), then the
Laplace transform of f (n) exists and

L [ f (n) (t)] = sn L [ f (t)] − sn−1 f (0) − sn−2 f 0 (0) − . . . − f (n−1) (0)

for s > α.
Proof. For n = 1, the result becomes L [ f 0 (t)] = sL [ f (t)] − f (0). For
n = k, this result is

L [ f (k) (t)] = sk L [ f (t)] − sk−1 f (0) − sk−2 f 0 (0) − . . . − f (k−1) (0)

and using this we get

L [ f (k+1) (t)] = L [( f (k) (t))0 ]

= sL [ f (k) (t)] − f (k) (0)
 
= s sk L [ f (t)] − sk−1 f (0) − . . . − f (k−1) (0) − f (k) (0)

= sk+1 L [ f (t)] − sk f (0) − sk−2 f 0 (0) − . . . − f (k) (0).

By mathematical induction, the result follows for all n.

Example 4.32. When f (t) = cos(at), we have f 0 (t) = −a sin(at) and

f 00 (t) = −a2 cos(at). In this case, the formula

L [ f 00 (t)] = s2 L [ f (t)] − s f (0) − f 0 (0)

becomes

−a2 L [cos(at)] = s2 L [cos(at)] − s

102 4. LAPLACE TRANSFORM

and this gives

s
L [cos(at)] = .
s2 + a2


Example 4.33. When f (t) = cosh(at), we have f 0 (t) = a sinh(at) and

f 00 (t) = a2 cosh(at). In this case, we get

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

a2 L [cosh(at)] = s2 L [cosh(at)] − s

and this gives

s
L [cosh(at)] = .
s2 − a2


Theorem 4.34. If f : [0, ∞) → R is piecewise continuous and is of ex-

ponential order and F(s) = L f (t), then

L [t n f (t)] = (−1)n F (n) (s).

Proof. By differentiating under the integral sign, we have

d
Z ∞
0
F (s) = e−st f (t) dt
ds 0
Z ∞
∂
e−st f (t) dt


=
∂s
Z0∞
= e−st (−t f (t)) dt
0
= L [−t f (t)].

Repeating the process, the general formula follows.

Example 4.35. For f (t) = 1, we have F(s) = 1/s and the formula
L [t n f (t)] = (−1)n F (n) (s) becomes

dn 1 n!
L [t n ] = (−1)n = .
dsn s sn+1

 22. PROPERTIES: DERIVATIVES AND INTEGRALS 103

Example 4.36. Let f (t) = sin(bt). Then

b 0 2bs
F(s) = and so F (s) = − .
s2 + b2 (s2 + b2 )2

Hence, using L [t f (t)] = −F 0 (s), we get

2bs
L [t sin(bt)] = .

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

(s2 + b2 )2

From this, it follows that

2b(s − a)
L [teat sin(bt)] = .
((s − a)2 + b2 )2

In particular,
2(s + 1)
L [te−t sint] = .
(s2 + 2s + 2)2


Example 4.37. Let f (t) = cos(bt). Then

s 0 b2 − s2
F(s) = and so F (s) = .
s2 + b2 (s2 + b2 )2

Hence, using L [t f (t)] = −F 0 (s), we get

s2 − b2
L [t cos(bt)] = .
(s2 + b2 )2

and so
(s − a)2 − b2
L [teat cos(bt)] = .
((s − a)2 + b2 )2
In particular,
s(s + 2)
L [te−t cost] = .
(s2 + 2s + 2)2

104 4. LAPLACE TRANSFORM

Example 4.38. Let f (t) = sinh(bt). Then

b 2bs
F(s) = and so F 0 (s) = − .
s2 − b2 (s2 − b2 )2

Hence, using L [t f (t)] = −F 0 (s), we get

2bs
L [t sinh(bt)] = .

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

(s2 − b2 )2

From this, it follows that

2b(s − a)
L [teat sinh(bt)] = .
((s − a)2 − b2 )2

Theorem 4.39. If f : [0, ∞) → R is piecewise continuous, is of expo-

nential order and L [ f (t)] = F(s), then
  Z∞
f (t)
L = F(τ)dτ.
t s
R ∞ −st
Proof. By definition, F(s) = 0 e f (t)dt and so
Z ∞ Z ∞ Z ∞
F(τ)dτ = e−τt f (t)dtdτ
s t=0
Zτ=s
∞ Z ∞
= e−τt f (t)dτdt
t=0 τ=s
f (t)
Z ∞
−τs
= e dt
t=0 t
 
f (t)
=L .
t

Example 4.40. Since

1 1
L [eat − ebt ] = −
s−a s−b
we have

eat − ebt
  Z ∞ 
1 1
L = −
t s s−a s−b
 22. PROPERTIES: DERIVATIVES AND INTEGRALS 105

s−a ∞
 
= log
s−b s
 
s−b
= log .
s−a


Example 4.41. Since

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

a
L [sin(at)] =
s2 + a2
we have
  Z∞
sin(at) a
L = 2 2
ds
t s s +a
∞
= tan−1 (s/a) s
π
= − tan−1 (s/a)
2
= cot−1 (s/a).


Example 4.42. Since

s
L [cos(at)] =
s2 + a2
we have
  Z ∞ 
cos(at) − cos(bt) s s
L = − ds
t s s2 + a2 s2 + b2
 2 ∞
1 s + a2
= log 2
2 s + b2 s
 2
s + b2

1
= log 2 .
2 s + a2
In particular, we have
√
s2 + a2
 
1 − cos(at)
L = log .
t s
Since sin2 (at) = (1 − cos(2at))/2, we have
√
1 s2 + 4a2 1 s2 + 4a2
L [sin2 (at)] = log = log .
2 s 4 s2

106 4. LAPLACE TRANSFORM

Theorem 4.43. If f : [0, ∞) → R is piecewise continuous and is of ex-

ponential order, then

L [ f (t)]
Z t 
L f (t)dt = .
0 s
Proof. Since the function f is piecewise continuous, the function g :
[0, ∞) → R defined by

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Z t
g(t) = f (τ) dτ
0

is continuous. Since f is of exponential order, there are constants M,

and α satisfying
| f (t)| ≤ Meαt (t ≥ 0).
Hence,
Z t
M αt M
|g(t)| ≤ M eατ dτ = (e − 1) ≤ eαt .
0 α α
Thus, g is continuous and is of exponential order. Hence, Laplace trans-
form of g exists. Further g0 (t) = f (t) and g(0) = 0. Using the formula
for the Laplace transform of the derivative
R t L[g0 (t)] = sL [g(t)] − g(0),
we have L [ f (t)] = sL [g(t)] or L 0 f (t)dt = L [ f (t)]/s.

Example 4.44. Since

s
L [cos(at)] =
s2 + a2
we have
t
Z 
1
=L cos(at)dt
s2 + a2 0
 
sin(at)
=L
a

and so
a
L [sin(at)] = .
s2 + a2

 22. PROPERTIES: DERIVATIVES AND INTEGRALS 107

Theorem 4.45 (Initial Value Theorem). If f : [0, ∞) → R piece-wise

continuous and is of exponential order, the Laplace transform of its de-
rivative f 0 exist, and F(s) = L f (t), then

lim sF(s) = lim f (t)

s→∞ t→0+

provided the limit limt→0+ f (t) exists.

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Proof. Since
Z ∞
sF(s) − f (0) = L [ f 0 (t)] = e−st f 0 (t)dt
0

we have
lim sF(s) − f (0) = 0.
s→∞

Theorem 4.46 (Final Value Theorem). If the Laplace transforms of

f : [0, ∞) → R, its derivative f 0 exist and F(s) = L f (t) then

lim sF(s) = lim f (t)

s→0 t→∞

provided the two limits exit.

Proof. Since
Z ∞
0
sF(s) − f (0) = L [ f (t)] = e−st f 0 (t)dt
0

we have
Z ∞
lim sF(s) − f (0) = f 0 (t)dt = lim f (t) − f (0)
s→0 0 t→∞

and so lims→0 sF(s) = limt→∞ f (t).

Example 4.47. Let f : [0, ∞) → R be given by cos at. Then its Laplace
transform is given by F(s) = s/(s2 + a2 ).
1
lim sF(s) = lim s2 /s2 + a2 = lim = 1 = lim f (t).
s→∞ s→∞ s→∞ 1 + a2 /s2 t→0

Also, we have lims→0 sF(s) = 0 but the limit limt→∞ f (t) does not exist.

108 4. LAPLACE TRANSFORM

23. Periodic functions

Theorem 4.48. If f : [0, ∞) → R is piecewise continuous and periodic
with period T , then
Z T
1
L [ f (t)] = e−st f (t)dt.
1 − e−sT 0
Proof. We have

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Z ∞
F(s) = f (t)e−st dt
0
∞ Z (n+1)T
= ∑ e−st f (t) dt
n=0 nT
and using the substitution u = t − nT we have
∞ Z T
F(s) = ∑ e−su−snT f (u + nT ) du
n=0 0
∞ Z T
−snT −su
= ∑e e f (u) du
n=0 0
!Z
∞
n T
e−sT e−su f (u) du


= ∑
n=0 0
Z T
1
= e−st f (t)dt.
1 − e−sT 0

Example 4.49. The function f : [0, ∞) defined by f (t) = sin(at) is peri-

odic of period 2π/a as
f (t + 2π/a) = sin(a(t + 2π/a)) = sin(at + 2π) = sin(at) = f (t).
Therefore, we have
Z 2π/a
1
L [sin(at)] = e−st sin(at) dt
1 − e−2πs/a 0
2π/a
1 e−st  
= − s sin(at) − a cos(at)
1 − e−2πs/a a2 + s2 0
a 1 − e−2πs/a
=
1 − e−2πs/a
s2 + a2
a
= 2 .
s + a2

 23. PERIODIC FUNCTIONS 109

Example 4.50. Let f : [0, ∞) → R be defined by f (t) = t for 0 ≤ t < 1

and extended periodically by f (t + 1) = f (t) for t ≥ 0. This function
is known as a saw-tooth function. Its period is T = 1. Therefore, the
Laplace transform F of f is given by
Z 1
1
F(s) = te−st dt
1 − e−s 0
1
1 e−st (st + 1)

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

=−
1 − e−s s2 0
1 − e−s (1 + s)
= 2
s (1 − e−s )
1 e−s
= 2− .
s s(1 − e−s )

Example 4.51. Let f : [0, ∞) → R be defined by

(
t, 0 ≤ t < 1,
f (t) =
2 − t, 1 ≤ t < 2

and extended periodically by f (t + 2) = f (t) for t ≥ 0. Since this func-

tion f is a periodic of period T = 2, the Laplace transform F of f is
given by

1
Z 2
F(s) = f (t)e−st dt
1 − e−2s 0
Z 1 Z 2 
1 −st −st
= te dt + (2 − t)e dt
1 − e−2s 0 1

 −st 1 2
1 e (st + 1) −st
e (s(t − 2) + 1) 
= − −2s 2
+
1−e s 0 s2
1

1 − e−s (1 + s) e−2s − e−s (1 − s)

 
1
=− +
1 − e−2s s2 s2
(1 − e−s )2
=
s2 (1 − e−2s )
1 − e−s
= 2
s (1 + e−s )
110 4. LAPLACE TRANSFORM

1
= tanh(s/2).
s2


24. Inverse Laplace transform

If F is the Laplace transform of a function f , then f is called the

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

inverse Laplace transform of F and is denoted by

F(s) = L −1 [ f (t)].

The inverse Laplace transform is also linear.

For a > −1, we have

ta
 
Γ(a + 1) −1 1
L [t ] =
a
⇒ L = .
sa+1 sa+1 Γ(a + 1)

In particular, we have

t n−1
     
−1 1 −1 1 −1 1
L = 1, L = t, and L = .
s s2 sn (n − 1)!

Also,
 
1 −1 1
L [e ] =
at
⇒L = eat .
s−a s−a
 
a −1 a
L [sin(at)] = 2 ⇒L = sin(at)
s + a2 s2 + a2
 
s −1 s
L [cos(at)] = 2 ⇒L = cos(at)
s + a2 s2 + a2
 
a −1 a
L [sinh(at)] = 2 ⇒L = sinh(at)
s − a2 s2 − a2
 
s −1 s
L [cosh(at)] = 2 ⇒L = cosh(at)
s − a2 s2 − a2
 
b −1 b
L [eat sin(bt)] = ⇒L = eat sin(bt),
(s − a)2 + b2 (s − a)2 + b2
 
s−a −1 s−a
L [eat sin(bt)] = ⇒L = eat sin(bt).
(s − a)2 + b2 (s − a)2 + b2
 24. INVERSE LAPLACE TRANSFORM 111

Theorem 4.52 (Linearity property). If F and G are Laplace trans-

forms of some functions, then for any two real number α, β , we have

L −1 [αF(s) + β G(s)] = αL −1 [F(s)] + β L −1 [G(s)].

Proof. Let F and G are Laplace transforms of the functions f and g.
Then, from
L [α f (t) + β g(t)] = αL [ f (t)] + β L [g(t)] = αF(s) + β G(s),

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

we get, by definition of inverse Laplace transform,
L −1 [αF(s) + β G(s)] = α f (t) + β g(t)
= αL −1 [F(s)] + β L −1 [G(s)].

Example 4.53. To find the inverse Laplace transform of 1/s(s + 1), we

first note that
1 1 1
= − .
s(s + 1) s s + 1
Using linearity of the inverse transform, we get
     
−1 1 −1 1 −1 1
L =L +L = 1 − e−t .
s(s + 1) s s+1


Example 4.54. We show that

 
−1 s
L = cosh(at).
s2 − a2
Since
 
s 1 1 1
2 2
= + .
s −a 2 s−a s+a
using linearity, we get
      
−1 s 1 −1 1 −1 1
L = L +L
s2 − a2 2 s−a s+a
at
e +e −at
=
2
= cosh(at).

112 4. LAPLACE TRANSFORM

Theorem 4.55. If F is the Laplace transform of function, then

L −1 [F(s + a)] = e−at L −1 [F(s)].

Proof. By replacing a by −a in first shifting theorem (Theorem 4.20),
we get

F(s) := L [ f (t)] ⇒ L [e−at f (t)] = F(s + a)

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

and by inverting we have

L −1 [F(s + a)] = e−at f (t)] = e−at L −1 [F(s)].

Example 4.56. Find the inverse Laplace transform of 1/s(s + 1)2 .

Solution. Since
1 1
L [t] = 2
=⇒ L [te−t ] =
s (s + 1)2

Hence for f (t) = te−t , we have F(s) = 1/(s + 1)2 . Thus,

  Zt
1 F(s) −1 1
= =⇒ L = τe−τ dτ = 1−(t +1)e−t
s(s + 1)2 s s(s + 1)2 0

Theorem 4.57. If F(s) is the Laplace transform of f , then

(24.1) L [−t f (t)] = F 0 (s), and L −1 [F 0 (s)] = −t f (t).

Example 4.58. Find the inverse Laplace transform of

 
s−a
F(s) = ln
s−b

Solution. If L [ f (t)] = F(s), then L [t f (t)] = −F 0 (s). Hence

1 1 ebt − eat
L [t f (t)] = − = L [ebt − eat ] =⇒ f (t) = .
s−b s−a t

 25. CONVOLUTION THEOREM 113

Example 4.59. Consider the same problem as in Example 4.58, i.e.

inverse Laplace transform of
 
s−a
F(s) = ln
s−b
Solution. Note that
  Z∞  bt   at 
s−a 1 1 e e
Z ∞
L [ f (t)] = ln = dp = L −L

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

d p−
s−b s s−b s s−a t t
Hence,
ebt − eat ebt − eat
 
L [ f (t)] = L =⇒ f (t) = .
t t


25. Convolution theorem

Definition 4.60. (Convolution) For two functions f , g : [0, ∞) → R, the
convolution of f and g, denoted by f ∗ g, is defined by
Z t
( f ∗ g)(t) = f (τ)g(t − τ) dτ.
0

Theorem 4.61 (Convolution Theorem). If f , g : [0, ∞) → R are piece-

wise continuous and is of exponential order and L [ f (t)] = F(s),
L [g(t)] = G(s), then L [ f (t) ∗ g(t)] = F(s)G(s).
Proof. Using definition, we find
  Z∞
L ( f ∗ g)(t) = ( f ∗ g)(t)e−st dt
0
Z ∞ Z t 
= f (τ)g(t − τ) dτ e−st dt
0 0

The region of integration is the area in the first quadrant of tτ-plane

bounded by the t-axis and the line τ = t. The limit of the inner integral
varies from τ = 0 to τ = t. Changing the order of integration, we see
that τ vary from τ = 0 to τ = ∞ and the limit of t vary from t = τ to
t = ∞. Hence, we have, using t − τ = u,
  Z ∞ Z ∞ 
−st
L ( f ∗ g)(t) = e g(t − τ) dt f (τ) dτ
0 τ
114 4. LAPLACE TRANSFORM
Z ∞ Z ∞ 
= e g(u) du f (τ)e−sτ dτ
−su
0 0
Z ∞  Z ∞ 
−su −sτ
= e g(u) du e f (τ) dτ
0 0
= F(s)G(s).

By taking g : [0, ∞) → R the constant function, g(t) = 1, we see

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

that G(s) = 1/s and therefore the convolution theorem becomes Theo-
rem 4.43:
L [ f (t)]
Z t 
L f (t)dt = .
0 s
Replacing f by f 0 in this, we get
t
Z 
0
L [ f (t)] = sL f (t)dt
0
= sL [ f (t) − f (0)]
= sL [ f (t)] − f (0).

We shall use the convolution theorem in finding inverse Laplace trans-

form. Note that
Z t
−1
L [F(s)G(s)] = f (t) ∗ g(t) = f (τ)g(t − τ) dτ.
0

Example 4.62. To find the inverse Laplace transform of 1/(s(s + 1)2 ),

we write it as F(s)G(s), where F(s) = 1/s and G(s) = 1/(s + 1)2 . The
inverse Laplace tranform of F and G are f (t) = 1 and g(t) = te−t .
Hence, using convolution theorem, we find
  Zt
−1 1
L = f (t − τ)g(τ) dτ
s(s + 1)2 0
Z t
= τe−τ dτ
0
t
= −(1 + τ)e−τ 0
= 1 − (1 + t)e−t .


 27. INTEGRAL EQUATIONS 115

Example 4.63. To find the inverse Laplace transform of 1/(s2 + a2 )2

write F(s) = 1/(s2 + a2 ) = G(s). Thus, f (t) = sin(at)/a = g(t). Hence,
we have
1 t
 
1
Z
−1
L = 2 sin(aτ) sin(a(t − τ))dτ.
(s2 + a2 )2 a 0
Using 2 sin x sin y = cos(x − y) − cos(x + y), we have
2 sin(aτ) sin(a(t − τ)) = cos(a(2τ − t)) − cos(at)

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

andZ so
t 1
2 sin(aτ) sin(a(t − τ))dτ = (sin(at) − sin(−at)) − t cos(at).
0 2a
Using this we have
 
−1 1 1
L 2 2 2
= 3 [sin(at) − at cos(at)].
(s + a ) 2a


26. Ordinary differential equations

Example 4.64. Solve the IVP
y00 + y = t, y(0) = 0, y0 (0) = 2

Solution. Take Laplace transform on both sides. This gives
1 1 2
s2Y − 2 +Y = 2 =⇒ Y = 2 2 + 2
s s (s + 1) s + 1
Using partial fraction, we find
1 1
Y= 2+ 2 =⇒ y(t) = t + sint
s s +1
27. Integral equations
Example 4.65. (Integral equation) Solve
Z t
0
y+ y(t − τ)e−2τ dτ = 1, y(0) = 1.
0

Solution. Take Laplace Transform on both sides, we find
Y 1 s+2 2 1
sY − y(0) + = =⇒ Y = =⇒ Y = −
s+2 s s(s + 1) s s+1
Hence,
y(t) = 2 − e−t
MAIR22 Complex Analysis and Differential Equations by V. Ravichandran
 5. Partial Differential Equations

28. First and Second Order PDE

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

We consider a function z = z(x, y) of two independent variables x and
y. We use the following notation for the first order partial derivatives:
∂z ∂z
p = zx = , q = zy = ,
∂x ∂y
and for the second order partial derivatives:

∂ 2z ∂ 2z ∂ 2z
r = zxx = , s = zxy = , t = zyy = .
∂ x2 ∂ x∂ y ∂ y2
The general form of a first order partial differential equation is

f (x, y, z, p, q) = 0.

For example, p − q = 0, xp + yq = z, xp − yq = x2 − y2 and z = px + qy

are all examples of first order partial differential equation.
Example 5.1. The partial derivatives of the function z = (x + a)(y + b)
are p = y + b and q = x + a. Therefore, the partial differential equation
satisfied by the function z is

z = (x + a)(y + b) = pq.

Example 5.2. Consider the function z defined implicity by ax2 + by2 +

z2 = 1. By differentiating the given equation with respect to x and y and
then diving the equations by the number 2, we get

ax + zp = 0 and by + zq = 0.

By multiplying the equations by x and y respectively and adding, we get

ax2 + by2 + xzp + yzq = 0.

117
118 5. PARTIAL DIFFERENTIAL EQUATIONS

This together with the given equation gives

1 − z2 + xzp + yzq = 0.
Therefore, the partial differential equation satisfied by the function z is
given by xzp + yzq = z2 − 1. 

Example 5.3. The function z = f (x2 − y2 ) gives p = 2x f 0 (x2 − y2 ) and

q = −2y f 0 (x2 − y2 ). Therefore, by adding the two equations, we see that

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

the partial differential equation satisfied by the function z is
yp + xq = 0.
The function z = f (x2 − y2 ) can be written in equivalent form as F(x2 −
y2 , z) = 0. 

Example 5.4. Consider the function z = z(x, y) defined by f (x + y +

z, x2 + y2 + z2 ) = 0. This can be written in equivalent form as x + y + z =
g(x2 + y2 + z2 ). By differentiating this equation with respect to x and y,
we get
1 + p = g0 (x2 + y2 + z2 )(2x + 2zp),
1 + q = g0 (x2 + y2 + z2 )(2y + 2zq).
These two equations immediately gives
(1 + p)(y + zq) = (1 + q)(x + zp).
Rewriting the equation, we see that the partial differential equation sat-
isfied by the function z is
(y − z)p + (z − x)q = x − y.


Example 5.5. Consider the function z defined by f (xy, y/z) = 0. This

can be written as z/y = g(xy) or z = yg(xy). Differentiation with respect
to x and y respectively, we get p = y2 g0 (xy) and q = g(xy) + xyg0 (xy).
Therefore, we have
yq = yg(xy) + xy2 g0 (xy) = z + px.
Thus, the partial differential equation satisfied by the function z is
yq − xp = z.

 28. FIRST AND SECOND ORDER PDE 119

A function z = z(x, y, a, b) satisfying the given differential equation

f (x, y, z, p, q) = 0 is called a solution (or integral) of the partial dif-
ferential equation f (x, y, z, p, q) = 0. A solution z = z(x, y, a, b) that
depends on the independent variables x, y and two arbitrary constants
a, b of the partial differential equation f (x, y, z, p, q) = 0 is its complete
solution (or complete integral). Example 5.1 shows that the function
z = (x + a)(y + b) satisfied pq = z and therefore it is a complete so-
lution of pq = z. From Example 5.2, the partial differential equation

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

xzp + yzq = z2 − 1 has the function implicity defined by ax2 + by2 + z2 =
1 as its complete solution.
The functions in Examples 5.4-5.5 are of the form F(u, v) = 0 where
u, v are functions of x, y, z and they give rise to partial differential equa-
tions. These functions are solutions of the respective partial differential
equations. A solution of the partial differential equation f (x, y, z, p, q) =
0 given by F(u, v) = 0 is known as general solution or general integral.
General solution can be obtained from complete integral given by
f (x, y, z, a, b) = 0 by writing the second arbitrary constant b by an arbi-
trary function b = φ (a). This gives f (x, y, z, a, φ (a)) = 0. Differentiating
this equation partially with respect to a and removing the constant a, the
general solution is obtained. We illustrate this method with few simple
examples.
Example 5.6. The function z = ax + (1 − a)y + b is a complete integral
of the equation p+q = 1. Write b = φ (a) to get z = ax+(1−a)y+φ (a).
Differentiating this equation with respect to a, we get 0 = x − y + φ 0 (a).
Solving for a, we get a = h(x −y). Using this in z = ax +(1−a)y+φ (a),
we get

z − y = a(x − y) + φ (h(x − y)) = f (x − y).

This is the general integral of the equation p + q = 1. It can also be

written in an equivalent form as F(z − y, x − y) = 0. 

Example 5.7. The function z = a(x2 + y) + b is a complete integral of

the equation p = 2qx. Write b = φ (a) to get z = a(x2 + y) + φ (a). Dif-
ferentiating this equation with respect to a, we get 0 = x2 + y + φ 0 (a).
Solving for a, we get a = h(x2 + y). Using this in z = a(x2 + y) + φ (a),
we get

z = a(x2 + y) + φ (h(x2 + y)) = f (x2 + y).

120 5. PARTIAL DIFFERENTIAL EQUATIONS

This is the general integral of the equation p = 2qx. It can also be

written in an equivalent form as F(x2 + y, z) = 0. 

Example 5.8. A simple calculation shows that the function z = a(x +

y) + 31 (x3 − y3 ) + b is a complete integral of the equation p − x2 = q + y2 .
Write b = φ (a) to get z = a(x + y) + 31 (x3 − y3 ) + φ (a). Differentiating
this equation with respect to a, we get 0 = x + y + φ 0 (a). Solving for a,
we get a = h(x + y). Using this, we see that the general integral is

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

1
z = a(x + y) + (x3 − y3 ) + φ (a)
3
1
= (x + y)h(x + y) + (x3 − y3 ) + φ (h(x + y))
3
1 3
= (x − y3 ) + f (x + y)
3
or in equivalent form, F x + y, z − 13 (x3 − y3 ) = 0.




29. Standard type of first order equations

The equation f (p, q) = 0 Some simple first order equations can be
solved easily. We first consider the partial differential equation f (p, q) =
0 that involves only p and q. In this case, we try the function z = ax +
by + c as a solution. Since p = a and q = b, the equation f (p, q) = 0 is
satisfied by the function z = ax + by + c if f (a, b) = 0. Solving for b, we
get b = g(a) and hence the function z = ax + g(a)y + c is a solution of
f (p, q) = 0.
Example 5.9. Solve p2 + q = p. Since this equation, involves only p
and q, the function z = ax + by + c is a solution provided a2 + b = a.
Solving for b, we have b = a − a2 and hence z = ax + a(1 − a)y + c is a
solution of p2 + q = p. 

Example 5.10. Solve p − q = pq. Since this equation, involves only

p and q, the function z = ax + by + c is a solution provided a − b = ab.
Solving for b, we have b = a/(1+a) and hence z = ax+(a/(1+a))y+c
is a solution of p − q = pq. 
Clairaut’s equation The equation z = px + qy + f (p, q) is known
as the Clairaut’s equation. The general solution of this equation is given
by z = ax + by + f (a, b). For this function, p = a, q = b and hence the
equation z = ax + by + f (a, b) becomes z = px + qy + f (p, q).
 29. STANDARD TYPE OF FIRST ORDER EQUATIONS 121

Example 5.11. Solve z = px + qy + p2 − q2 . Since this equation is the

Clairaut’s equation, the function z = ax + by + a2 − b2 is the general
solution. 
The variable separable equationA partial differential equation that
can be written in the form f (x, p) = g(y, q) is called a variable separable
equation. These equations can be solved by writing f (x, p) = a and
g(y, q) = a and solving for p, q respectively. Let p = φ (x, a) and q =

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

ψ(y, a). Then

dz = pdx + qdy = φ (x, a)dx + ψ(y, a)dy.

Integrating, we get the complete integral as

Z
z= φ (x, a)dx + ψ(y, a)dy + b.

Example 5.12. The equation p − x2 = q + y2 is a variable separable

equation. Write p − x2 = a and q + y2 = a. Then a complete integral is
Z
z= pdx + qdy
Z
= (a + x2 )dx + (a − y2 )dy
1
= a(x + y) + (x3 − y3 ) + b.
3


Example 5.13. The equation pq = xy is a variable separable equation

as it can be written as p/x = y/q. Write p/x = a and y/q = a. Then
Z
z= pdx + qdy
Z
= axdx + y/ady

= ax2 /2 + y2 /(2a) + b/2.

Thus, a complete integral is given by 2z = ax2 + y2 /a + b. 

The partial differential equation of the form

f (x, p, q) = 0 or f (y, p, q) = 0
122 5. PARTIAL DIFFERENTIAL EQUATIONS

are both special cases of the variable separable equations. By solving

for q, the equation f (x, p, q) = 0 can be written as g(x, p) = q. Thus, by
taking q = a and g(x, p) = a, we can solve for
Z Z
z= φ (x, a)dx + ady + b = φ (x, y)dx + ay + b

where p = φ (x, a) is obtained by solving g(x, p) = a for p.

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Example 5.14. The equation p = q cos x is of the form f (x, p, q) = 0.
Writing q = a and solving for p, we get p = a cos x and hence a complete
integral is given by
Z
z= pdx + qdy
Z
= a cos xdx + ady = a sin x + ay + b.


The equation f (z, p, q) = 0 The equation f (z, p, q) = 0 can be solved
by converting it to an ordinary differential equation by writing z = z(u)
where u = x + ay. With this substitution, we have p = z0 (u), q = az0 (u)
and the given differential equation becomes
 
dz dz
f z, , a = 0.
du du

Solving this, we get z as z = z(u) = z(x + ay).

Example 5.15. The equation z = p2 + q2 is of the form f (z, p, q) = 0.
Let u = x + ay and z = z(u). Then the given equation becomes z =
dz 2
( du ) (1 + a2 ) and hence
p dz √
1 + a2 = z.
du
Solving this ordinary differential equation, we get the complete integral
as
p √
2 1 + a2 z = u + b or 4(1 + a2 )z = (x + ay + b)2 .


 30. THE LAGRANGE’S LINEAR EQUATION Pp + Qq = R 123

Example 5.16. The equation p(1 + q) = qz is of the form f (z, p, q) = 0.

Let u = x + ay and z = z(u). Then the given equation becomes
 
dz dz dz
1+a = az .
du du du
This equation is satisfied if either
dz dz
=0 or a − az = −1.

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

du du
The first one gives z = a while the second equation gives

log(z − 1) = u + b = x + ay + b.

Solving this ordinary differential equation, we get the complete integral

as
p √
2 1 + a2 z = u + b or 4(1 + a2 )z = (x + ay + b)2 .


30. The Lagrange’s linear equation Pp + Qq = R

From Examples 5.3-5.5, we see that elimination of the arbitrary
function from f (u, v) = 0 where u, v are functions of x, y, z give rise to
differential equation of the form Pp + Qq = R. Here P, Q and R are also
functions of x, y, z. This holds not just for those functions in Examples
5.3-5.5. Indeed, if f (u, v) = 0, write it in equivalent form as u = g(v)
and then differentiate partially with respect to x and y to get
 
∂u ∂u 0 ∂v ∂v
+ p = g (v) + p ,
∂x ∂z ∂x ∂z
 
∂u ∂u 0 ∂v ∂v
+ q = g (v) + q .
∂y ∂z ∂y ∂z
Eliminating the function g0 (v), we get
∂u ∂u ∂v ∂v
∂x + ∂z p ∂x + ∂z p = 0.
∂u ∂u ∂v ∂v
∂y + ∂z q ∂y + ∂zq

Using the notation

∂ (u, v) ∂ u ∂ v ∂ u ∂ v
= −
∂ (x, y) ∂ x ∂ y ∂ y ∂ x
124 5. PARTIAL DIFFERENTIAL EQUATIONS

and simplifying the determinant above, we get the equation Pp+Qq = R

where
∂ (u, v) ∂ (u, v) ∂ (u, v)
P= , Q= , R= .
∂ (y, z) ∂ (z, x) ∂ (x, y)
If u = a and v = b, then we get du = dv = 0. This becomes

∂u ∂u ∂u ∂v ∂v ∂v

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

dx + dy + dz = 0 and dx + dy + dz = 0.
∂x ∂y ∂z ∂x ∂y ∂z

These equations hold if

dx dy dz
∂u ∂v
= =
∂y ∂z − ∂∂ uz ∂∂ vy ∂u ∂v
∂z ∂x − ∂∂ ux ∂∂ vz ∂u ∂v
∂x ∂y − ∂∂ uy ∂∂ xv

or
dx dy dz
(30.1) = = .
P Q R

Therefore, f (u, v) = 0 is the general solution of Pp+Qq = R if u = a and

v = b are solutions of (30.1). The equation (30.1) is called the auxiliary
equation corresponding to Pp + Qq = R.
Example 5.17. The auxiliary equation corresponding to yp + xq = 0 is

dx dy dz
= = .
y x 0

Clearly, z = a is a solution of the equation. From the first equality,

we have xdx − ydy = 0 and integration gives x2 − y2 = b. Therefore,
f (z, x2 − y2 ) = 0 is a general solution of yp + xq = 0. 

Example 5.18. The auxiliary equation corresponding to xzp − yzq =

y2 − x2 is
dx dy dz
= = 2 .
xz −yz x − y2
From the first equality, we have

dx dy
+ =0
x y
 31. SECOND ORDER PDES 125

and integration gives log x + log y = log a or xy = a. Also, we have

xdx + ydy dz
= 2
z(x − y ) x − y2
2 2

and hence
xdx + ydy + zdz = 0.

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

Integration gives x2 + y2 + z2 = b. Therefore, f (xy, x2 + y2 + z2 ) = 0 is
the general integral of the given equation. 

31. Second Order PDEs

Second order partial differential equations arise when we eliminate
the arbitrary functions or when there are more than three arbitrary con-
stants.
Example 5.19. Let z = ax2 +2bxy+cy2 . Then p = 2ax+2by, q = 2bx+
2cy. Also, we have r = 2a, s = 2b and t = 2c. Eliminating the constants
a, b, c, we get the partial differential equation 2z = x2 r + 2xys + y2t. 

Example 5.20. Consider the function z = f (x + αy) + g(x − αy). The

partial derivatives of z are given by

p = f 0 (x + αy) + g0 (x − αy),
q = α f 0 (x + αy) − αg0 (x − αy),
r = f 00 (x + αy) + g00 (x − αy),
t = α 2 f 0 (x + αy) + α 2 g0 (x − αy).

If follows that the differential equation satisfied by the function z = f (x+

αy) + g(x − αy) is α 2 r − t = 0 or

∂ 2z ∂ 2z
α2 − = 0.
∂ x2 ∂ y2


Example 5.21. Consider the function z = f (x2 + y) + g(x2 − y). The

partial derivatives of z are given by

p = 2x[ f 0 (x2 + y) + g0 (x2 − y)],

126 5. PARTIAL DIFFERENTIAL EQUATIONS

q = f 0 (x2 + y) − g0 (x2 − y),

r = 4x2 [ f 00 (x2 + y) + g00 (x2 − y)] + 2[ f 0 (x2 + y) + g0 (x2 − y)],
t = f 00 (x2 + y) + g00 (x2 − y).
If follows that the differential equation satisfied by the function z =
f (x2 + y) + g(x2 − y) is
p
r − 4x2t = 2[ f 0 (x2 + y) + g0 (x2 − y)] =

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

x
or
∂ 2z 2
2∂ z 1 ∂z
2
− 4x 2
= .
∂x ∂y x ∂x

When a physical problem is modelled into a partial differential equa-
tion, there are variables associated with space occupied by physical ob-
ject and variable associated with time. A condition satisfied boundary
points of the space is boundary condition and the condition satisfied at
the initial time, usually t = 0, is known as initial condition. We illustrate
it by two examples.
Example 5.22 (One-dimensional wave equation). Consider a per-
fectly elastic string of length l placed along the x-axis, and fasten at
the ends x = 0 and x = l. Let us allow the string to vibrate on xu-plane
and stretch the string to the position f (x) at the time t = 0 and release
with a velocity g(x) at that instant. Let u(x,t) be deflection of the elastic
string at a point x and time t. Then the deflection u(x,t) satisfy the one-
dimensional wave equation, discovered by the French scientist Jean le
Rond d’Alembert1
∂ 2u 2
2∂ u
= c
∂t 2 ∂ x2
This equation is derived under physical assumptions that the mass of
the string per unit length µ is constant, and the string is stretched by
a tension T so that the effect of the gravitation on the string is negligi-
ble. The constant related to these physical quantities by c2 = T /ρ. The
deflection u(x,t) satisfies the conditions
u(0,t) = 0, and u(l,t) = 0, (t ≥ 0)
1He was born in 1717 and was also a music theorist. The one-dimensional wave
equation is also known as the d’Alembert’s equation.
 31. SECOND ORDER PDES 127

and also the conditons

u(x, 0) = f (x), and ut (x, 0) = g(x), (0 ≤ x ≤ l).
The conditions u(0,t) = 0, u(l,t) = 0 are satisfied as there is no deflec-
tion at the end points. The problem has one independent space vari-
able x and one independent time variable t. The conditions u(0,t) = 0,
u(l,t) = 0 are known as the boundary conditions; these are the condi-
tions satisfied at the end points of the string. The conditions u(x, 0) =

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

f (x) and ut (x, 0) = g(x) are known the initial conditions and these are
the conditions satisfied at time t = 0. 

Example 5.23 (One-dimensional heat equation). Consider a long but

thin metal rod of uniform thickness made of homogeneous material placed
on the x-axis. Let one end of the rod at x = 0 and the other end be at
x = l. Assume that the rod is insulated and the heat flow is along x-
axis only. The temperature u(x,t) at a point x on the rod and at time t
satisfies the one-dimensional heat equation
∂u ∂ 2u
= α 2.
∂t ∂x
The constant α is the thermal diffusivity given by α = k/(cρ) where k
is the thermal conductivity, c is the specific heat capacity and ρ is the
density of the material. We may assume that the temperature is constant
at the two ends of the rod and, for example, you can take
u(0,t) = 0, and u(l,t) = 0, (t ≥ 0).
These conditions are the boundary conditions for the problem. We also
assume that the temperature distribution at t = 0 is f (x), that is,
u(x, 0) = f (x), (0 ≤ x ≤ l).
This is the initial conditions for the problem. 

Example 5.24 (Two-dimensional heat equation). Consider a thin metal

rectangular plate of uniform thickness made of homogeneous material
placed on the xy-plane. Let the corners of the plate are at (0, 0), (0, b),
(a, 0), (a, b). For an infinite plate, b = ∞. The temperature u(x, y,t) at a
point (x, y) on the plate and at time t satisfies the two-dimensional heat
equation
 2
∂ u ∂ 2u

∂u
=α + .
∂t ∂ x2 ∂ y2
128 5. PARTIAL DIFFERENTIAL EQUATIONS

If the temperature does not depend on time, then u = u(x, y) satisfy the
Laplace equation
∂ 2u ∂ 2u
+ = 0.
∂ x2 ∂ y2
A steady-state heat problem consists of this equation with certain bound-
ary conditions. We can for example assume that the temperature at the
three sides of the plate is zero and on fourth side it is a function of x:

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

u(0, y) = 0, and u(a, y) = 0 (0 ≤ y ≤ b),
and
u(x, 0) = 0, and u(x, b) = f (x), (0 ≤ x ≤ a).

The second order partial differential equation of the form
∂ 2z ∂ 2z ∂ 2z
A(x, y) + 2B(x, y) +C(x, y) + f (x, y, z, zx , zy ) = 0
∂ x2 ∂ x∂ y ∂ y2
is classified as
(1) elliptic at the point (x, y) if B(x, y)2 − A(x, y)C(x, y) < 0,
(2) parabolic at the point (x, y) if B(x, y)2 − A(x, y)C(x, y) = 0,
(3) hyperbolic at the point (x, y) if B(x, y)2 − A(x, y)C(x, y) > 0.

Example 5.25. The following are some standard second order partial
differential equations.
(1) The wave equation c2 uxx = utt , c > 0, is hyperbolic. In this
case, A = c2 , B = 0,C = −1 and hence B2 − AC = c2 > 0.
(2) The one dimensional heat flow equation uxx = ut is parabolic.
In this case, A = 1, B = 0,C = 0 and hence B2 − AC = 0.
(3) The Laplace equation uxx + uyy = 0 and the Poisson equation
uxx + uyy = f (x, y) are elliptic. In these two cases, A = B = 1
and B = 0 and B2 −AC = −1 < 0. The Laplace equation is also
known as the two dimensional steady-state heat flow equation.


Example 5.26. For the equation

∂ 2z 2
2∂ z 1 ∂z
2
− 4x 2
= ,
∂x ∂y x ∂x
 32. METHOD OF SEPARATION OF VARIABLES 129

we have A = 1, B = 0,C = −4x2 , and B2 − AC = −16x4 ≤ 0. Thus, the

equation is parabolic at all points where x = 0 and elliptic at all points
where x 6= 0. 

32. Method of Separation of Variables

Certain partial differential equation f (x, y, z, p, q, r, s,t) = 0 can be
converted into an equation of the form P(X, X 0 , X 00 ; x) = Q(Y,Y 0 ,Y 00 ; y)

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

by using a substitution z(x, y) = X(x)Y (y). In such case, by solving
P(X, X 0 , X 00 ; x) = k and Q(Y,Y 0 ,Y 00 ; y) = k where k is a constant, we get
X and Y . This in turn gives the complete integral z = X(x)Y (y) of the
given equation.
Example 5.27. Consider the equation yp + xq = 0. Let z = X(x)Y (y)
so that p = X 0Y and q = XY 0 . With this substitution, the given equation
becomes yX 0Y + xXY 0 = 0 or
X0 Y0
=− .
xX yY
Writing
X0 Y0
= 2b, − = 2b,
xX yY
we get
X 0 − 2bxX = 0, Y 0 + 2byY = 0.
2 2
Solving these equations, we get X = Aebx and Y = Be−by . This shows
that a complete integral of the given equation is
2 −y2 ) 2 −y2 )
z = XY = ABeb(x = aeb(x

Example 5.28. The substitution z = X(x)Y (y) reduce the equatio p = q

into two ordinary differential equation
X0 Y0
= b, = b.
X Y
Solving these equations, we get X = Aebx and Y = Beky and hence z =
ABeb(x+y) = aeb(x+y) is a complete integral of the equation p = q. 
130 5. PARTIAL DIFFERENTIAL EQUATIONS

We use the method of separation of variables to solve the Laplace

equation.
Example 5.29. Consider the Laplace equation uxx + uyy = 0. Let u =
X(x)Y (y) so that uxx = X 00Y and uyy = XY 00 . With this substitution, the
Laplace equation becomes X 00Y + XY 00 = 0 or

X 00 Y 00
=− .

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

X Y
Writing X 00 /X = k and −Y 00 /Y = k, we get

(32.1) X 00 − kX = 0, Y 00 + kY = 0.

We consider three different possibilities for k.

Case (i). (k = 0). If k = 0, the solutions of (32.1) are X = Ax +B and
Y = Cy + D. This shows that u(x, y) = (Ax + B)(Cy + D) is a complete
integral of the Laplace equation.
Case (ii). (k = p2 > 0, p > 0). If k = p2 > 0, the solutions of (32.1)
are X = Ae px + Be−px and Y = C cos py + D sin py. This shows that
u(x, y) = (Ae px + Be−px )(C cos py + D sin py) is a complete integral of
the Laplace equation.
Case (iii). (k = −p2 < 0, p > 0). If k = −p2 < 0, the solutions of
(32.1) are X = A cos px + B sin px and Y = Ce py + De−py . This shows
that u(x, y) = (A cos px + B sin px)(Ce py + De−py ) is a complete integral
of the Laplace equation.
In summary, a complete integral of the Laplace equation uxx + uyy =
0 is of the form

(32.2) u(x, y) = (Ax + B)(Cy + D),

(32.3) u(x, y) = (Ae px + Be−px )(C cos py + D sin py),
(32.4) u(x, y) = (A cos px + B sin px)(Ce py + De−py ).

Example 5.30. Solve the Laplace equation uxx + uyy = 0 satisfying the
conditions u(x, y) → 0 as x → ∞ and u(0, y) = α cos my.
The solutions of the Laplace equation are given by (32.2)-(32.4). If
u(x, y) → 0 as x → ∞, and u is given by (32.2), then A = 0 and either
B = 0 or Cy + D = 0. This leads to u(x, y) = 0 and this does not satisfy
the other condition. Similarly, if u is given by (32.4), then u(x, y) = 0.
 32. METHOD OF SEPARATION OF VARIABLES 131

This leaves the possibility that u is given by (32.3). In this case, the
condition u(x, y) → 0 as x → ∞ implies that A = 0 and so u(x, y) =
Be−px (C cos py + D sin py). If we use the condition u(0, y) = α cos my,
then we get α cos my = B(C cos py+D sin py). This gives C = 0, BD = α
and p = m. Thus, u(x, y) = αe−px cos my is the solution of the problem.


Example 5.31. Solve the Laplace equation uxx + uyy = 0 satisfying the

MAIR22 Complex Analysis and Differential Equations by V. Ravichandran

conditions u(0, y) = u(a, y) = 0, u(x, 0) = 0 and u(x, b) = sin(nπx/a).
The solutions of the Laplace equation are given by (32.2)-(32.4).
Consider the function u given by (32.4):

u(x, y) = (A cos px + B sin px)(Ce py + De−py ).

The condition u(0, y) = 0 shows that A must be taken to be zero and so

u(x, y) = sin px(Ce py + De−py ) where B is absorbed in C and D. The
condition u(a, y) = 0 shows that 0 = sin pa(Ce py + De−py ) and so pa =
nπ for some integer n. This gives the solution

u(x, y) = sin(nπx/a)(Cenπy/a + De−nπy/a ).

The condition u(x, 0) = 0 leads to C + D = 0 or

u(x, y) = C sin(nπx/a)(enπy/a − e−nπy/a )

= 2C sin(nπx/a) cosh(nπy/a).

Using the condition u(x, b) = sin(nπx/a), we get

sin(nπx/a) = 2C sin(nπx/a) cosh(nπb/a)

and this give 2C = 1/ cosh(nπb/a). Thus the solution to the given prob-
lem is
sin(nπx/a) cosh(nπy/a)
u(x, y) = .
cosh(nπb/a)