Partial Differential Equations

Naga Raju Gande

gnagaraju@mth.vnit.ac.in

Department of Mathematics,
Visversvaraya National Institute of Technology,Nagpur
October 20, 2020
Course Details
Course name: Integral Transforms and Partial
Differential Equations
Course code: MAL-201
Credits:
B.Tech. Second Year
Visvesvaraya National Institute of Technology,
Nagpur
For more detials check the course book
Exams details can be found from Dean's note
available on the website for students

2
PREREQUISITES
Pre-requisite

The first step before studing the PDE’s

4
Pre-requisite

The first step before studing the PDE’s

Ordinary Differential Equations:
,→ Linear first order ODE
,→ Second order constant coefficient ODE
,→ The regular Strum- Liouville equation
,→ Cauchy-Euler equation

4
Pre-requisite

The first step before studing the PDE’s

Ordinary Differential Equations:
,→ Linear first order ODE
,→ Second order constant coefficient ODE
,→ The regular Strum- Liouville equation
,→ Cauchy-Euler equation

Second step towards the solution of PDE’s

4
Pre-requisite

The first step before studing the PDE’s

Ordinary Differential Equations:
,→ Linear first order ODE
,→ Second order constant coefficient ODE
,→ The regular Strum- Liouville equation
,→ Cauchy-Euler equation

Second step towards the solution of PDE’s

Fourier Series:
,→ Bounded domain
,→ Half range Fourier series

4
INTRODUCTION TO
PARTIAL
DIFFERENTIAL
EQUATIONS (PDE)
Partial Differential Equations

What is a partial differential equation (PDE)?

Definition should me more intuitive:

6
Partial Differential Equations

What is a partial differential equation (PDE)?

Definition should me more intuitive:

Definition
Let u : R2 → R. Then an equation involving the partial
derivatives of u is known as partial differential
equation (PDE), say,

F(x, y, u, ux , uy , uxx , uxy , ...) = 0

6
Linear second order PDE

Definition
A linear 2nd order PDE satisfied by a function u(x, y)
depending on two independent variables x and y can
be written as:

auxx + 2buxy + cuyy + dux + euy + fu = g(x, y)

7
Linear second order PDE

Definition
A linear 2nd order PDE satisfied by a function u(x, y)
depending on two independent variables x and y can
be written as:

auxx + 2buxy + cuyy + dux + euy + fu = g(x, y)

,→ a, b, c, ..., f are constants or functions of x and/or y.

7
Linear second order PDE

Definition
A linear 2nd order PDE satisfied by a function u(x, y)
depending on two independent variables x and y can
be written as:

auxx + 2buxy + cuyy + dux + euy + fu = g(x, y)

,→ a, b, c, ..., f are constants or functions of x and/or y.

,→ Define D := b2 − ac

7
Classification

Hyperbolic : D > 0;
Wave Equation

8
Classification

Hyperbolic : D > 0; utt = c2 uxx

Wave Equation

8
Classification

Hyperbolic : D > 0; utt = c2 uxx

Wave Equation

Parabolic: D = 0;
Heat Equation

8
Classification

Hyperbolic : D > 0; utt = c2 uxx

Wave Equation

Parabolic: D = 0;
ut = κuxx
Heat Equation

8
Classification

Hyperbolic : D > 0; utt = c2 uxx

Wave Equation

Parabolic: D = 0;
ut = κuxx
Heat Equation

Elliptic : D < 0;
Laplace Equation

8
Classification

Hyperbolic : D > 0; utt = c2 uxx

Wave Equation

Parabolic: D = 0;
ut = κuxx
Heat Equation

Elliptic : D < 0;
Laplace Equation uxx + uyy = 0

8
Classification

Hyperbolic : D > 0; utt = c2 uxx

Wave Equation

Parabolic: D = 0;
ut = κuxx
Heat Equation

Elliptic : D < 0;
Laplace Equation uxx + uyy = 0
Why ?

8
Classification

Hyperbolic : D > 0; utt = c2 uxx

Wave Equation

Parabolic: D = 0;
ut = κuxx
Heat Equation

Elliptic : D < 0;
Laplace Equation uxx + uyy = 0
Why ?
Remember: ax 2 + 2hxy + by 2 + cx + dy + e = 0
8
Examples: Classify the following PDEs

1) x 2 uxx − y 2 uyy − 2yuy = 0.

9
Examples: Classify the following PDEs

1) x 2 uxx − y 2 uyy − 2yuy = 0.

2) y 2 uxx + 2xyuxy + 2x 2 uyy + yuy = 0.

9
Examples: Classify the following PDEs

1) x 2 uxx − y 2 uyy − 2yuy = 0.

2) y 2 uxx + 2xyuxy + 2x 2 uyy + yuy = 0.

3) Tricomi equation yuxx + uyy = 0.

9
Examples: Classify the following PDEs

1) x 2 uxx − y 2 uyy − 2yuy = 0.

2) y 2 uxx + 2xyuxy + 2x 2 uyy + yuy = 0.

3) Tricomi equation yuxx + uyy = 0.

4) x 2 uxx + 2xuxy + uyy = uy .

9
WAVE EQUATION
Bounded Domain
Wave Equation
The pde:

utt = c2 uxx , 0 < x < l, t > 0,

11
Wave Equation
The pde:

utt = c2 uxx , 0 < x < l, t > 0,

Boundary conditions (BC):

u(0, t) = 0, u(l, t) = 0, t ≥ 0,

11
Wave Equation
The pde:

utt = c2 uxx , 0 < x < l, t > 0,

Boundary conditions (BC):

u(0, t) = 0, u(l, t) = 0, t ≥ 0,

Intial conditions (IC):

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

11
Wave Equation
The pde:

utt = c2 uxx , 0 < x < l, t > 0,

Boundary conditions (BC):

u(0, t) = 0, u(l, t) = 0, t ≥ 0,

Intial conditions (IC):

(
u(x, 0) = f (x),
ut (x, 0) = g(x),

11
Wave Equation
The pde:

utt = c2 uxx , 0 < x < l, t > 0,

Boundary conditions (BC):

u(0, t) = 0, u(l, t) = 0, t ≥ 0,

Intial conditions (IC):

(
u(x, 0) = f (x),
0 ≤ x ≤ l.
ut (x, 0) = g(x),

11
Examples
Example 1:
Consider the problem with zero initial velocity, i.e.,

ut (x, 0) = 0

12
Examples
Example 1:
Consider the problem with zero initial velocity, i.e.,

ut (x, 0) = 0

and with initial displacement


x l
for 0 ≤ x ≤

u(x, 0) = f (x) = 2



12
Examples
Example 1:
Consider the problem with zero initial velocity, i.e.,

ut (x, 0) = 0

and with initial displacement


x l
for 0 ≤ x ≤

u(x, 0) = f (x) = 2
l − x l
for <x≤l

2

Solution procedure:
Fourier method/Separation of variable method
12
Important Steps...
Suppose u(x, t) = X(x)T(t)
X 00 T̈
From pde: = 2
X c T

13
Important Steps...
Suppose u(x, t) = X(x)T(t)
X 00 T̈
From pde: = 2 = −λ
X c T
From BCs and from above equations, we’ve a BVP:

X 00 + λX = 0; X(0) = X(l) = 0

n2 π 2
The eigenvalues for the BVP are: λn = .
l2
And the eigenfunctions are non-zero constant
multiples of
 nπx 
Xn (x) = sin , for n = 1, 2, ....
l
Ex2

13
Important Steps...
Now for T(t) : from zero initial condition
 nπc 2
T̈ + T = 0; Ṫ(0) = 0.
l
The general solution for T(t) is:
 
nπct
Tn (t) = cn cos
l

for each n ∈ N, with cn as yet to determine.

Now we’ve for n = 1, 2, ...
 nπx   
nπct
un (x, t) = cn sin cos .
l l

14
Important Steps...
In order to satisfy the non-zero initial condition we
take infinite superposition
∞  nπx   
X nπct
u(x, t) = cn sin cos .
n=1
l l

Choose the cn0 s to satisfy

∞
X  nπx 
u(x, 0) = f (x) = cn sin .
n=1
l

We can do this!
The r.h.s. is the Fourier sine expansion of f (x) on [0, l].

15
Important Steps...
Thus we obtain the solution:
∞  nπx   
X nπct
u(x, t) = cn sin cos .
n=1
l l

The cn0 s the Fourier sine coefficients

Zl  nπx 
2
cn = f (x) sin dx
l l
0

4l  nπ 
Thus cn = 2 2 sin .
nπ 2

16
Examples
Example 2:
Consider the problem with zero initial displacement,
i.e.,
u(x, 0) = 0.

17
Examples
Example 2:
Consider the problem with zero initial displacement,
i.e.,
u(x, 0) = 0.

and with initial velocity

  πx 
ut (x, 0) = g(x) = x 1 + cos , for 0 ≤ x ≤ l.
l

17
Examples
Example 2:
Consider the problem with zero initial displacement,
i.e.,
u(x, 0) = 0.

and with initial velocity

  πx 
ut (x, 0) = g(x) = x 1 + cos , for 0 ≤ x ≤ l.
l
Solution procedure:
Fourier method/Separation of variable method FM

17
Ex. 2 Cont.
Now for T(t) : from zero initial condition
 nπc 2
T̈ + T = 0; T(0) = 0.
l
The general solution for T(t) is:
 
nπct
Tn (t) = cn sin
l

for each n ∈ N, with cn as yet to determine.

Now we’ve for n = 1, 2, ...
 nπx   
nπct
un (x, t) = cn sin sin .
l l

18
Ex. 2 Cont...
In order to satisfy the non-zero initial condition we
take infinite superposition
∞  nπx   
X nπct
u(x, t) = cn sin sin .
n=1
l l

Choose the cn0 s to satisfy

∞
X nπc  nπx 
ut (x, 0) = g(x) = cn sin .
n=1
l l

We can do this!
The r.h.s. is the Fourier sine expansion of g(x) on [0, l].

19
Ex. 2 Cont...
nπc
Thus cn are the Fourier sine coefficients
l
And
Zl
nπc 2  nπx 
cn = g(x) sin dx
l l l
0

We’ve with
  πx 
g(x) = x 1 + cos , for 0 ≤ x ≤ l.
l

(−1)n l2

if n 6= 1,


 nπ(n2 − 1)


cn =
2

 3l


if n = 1.

4π
20
Ex. 2 Cont...

Thus we obtain the solution:

2 3l2
   πx   
πct
u(x, t) = sin sin
πc 4π l l
∞ n 2
 
2 X (−1) l  nπx  nπct
+ sin sin .
πc n=2 n2 π(n2 − 1) l l

21
Solution of a wave equation
With initial displacement and velocity.

How do we get the solution, if we’ve non-zero initial

displacement as well as non-zero initial velocity ?

Linearity !
If we let v(x, t) the solution with zero initial velocity
7→ non-zero initial displacement.

Let w(x, t) the solution with zero initial displacement

7→ non-zero initial velocity.

The the solution will be u(x, t) = v(x, t) + w(x, t).

22
Example 3
Ques. Solve the IBVP

utt = 9uxx , 0 < x < 2, t > 0,

u(0, t) = 0 = u(2, t), t ≥ 0,

u(x, 0) = x(x − 2), 0 ≤ x ≤ 2,


1
3 ≤ x ≤ 1,
ut (x, 0) = 2 0 ≤ x ≤ 2.
0 otherwise,

using separation of variable of method.

23
Non-homogeneous wave equation
Ques. Solve the IBVP

utt = uxx − cos x, 0 < x < 2π, t > 0,

u(0, t) = 0 = u(2π, t), t ≥ 0,

u(x, 0) = 0, ut (x, 0) = 0, 0 ≤ x ≤ 2π,

using separation of variable of method.

Ans. Let u(x, t) = w(x, t) + ψ(x).
Choose ψ(x) such that the whole problem reduces to
the one that we’ve already solved/known.

24
Non-homogeneous wave equation

If we choose ψ(x) = 1 − cos x,

then we have for w(x, t) :

wtt = wxx , 0 < x < 2π, t > 0,

w(0, t) = 0 = w(2π, t), t ≥ 0,
w(x, 0) = cos x − 1, wt (x, 0) = 0, 0 ≤ x ≤ 2π.

Now we can employ, separation of variable method.

If we let w(x, t) = X(x)T(t),

25
Non-homogeneous wave equation

The problem for X(x) is:

X 00 + λX = 0; X(0) = 0 = X(2π)

Thus the eigenvalues are

n2
λn =
4
and the eigenfunctions are any constant multiple of
 nx 
Xn (x) = sin
2

26
Non-homogeneous wave equation
The problem for T(t) is:

n2
T̈ + T = 0; Ṫ(0) = 0
4
Thus  
nt
Tn (t) = bn cos
2
Taking infinite superposition of solution
∞  nx   
X nt
w(x, t) = cn sin cos .
n=1
2 2

27
Non-homogeneous wave equation
Appling the non-zero initial condition
∞
X n 
w(x, 0) = cos x − 1 = bn sin x
n=1
2

where b0n s are Fourier sine coefficients:

Z2π n 
2
bn = (cos x − 1) sin x dx
2π 2
0

Thus w(x, t) =
∞    
16 X 1 2n − 1 2n − 1
sin x cos t
π n=1 (2n − 1) ((2n − 1)2 − 4) 2 2

28
Non-homogeneous wave equation

Finally the solution of the give problem is:

u(x, t) = w(x, t) + 1 − cos x.

29
Non-homogeneous wave equation
Ques. Solve the IBVP

utt = uxx + Ax, 0 < x < l, t > 0,

u(0, t) = 0 = u(l, t), t ≥ 0,

u(x, 0) = 0, ut (x, 0) = 1, 0 ≤ x ≤ l,

using separation of variable of method.

Ans. Let u(x, t) = w(x, t) + ψ(x).
Choose ψ(x) such that the whole problem reduces to
the one that we’ve already solved/known.

30
Non-homogeneous wave equation
Recall that we want to have for w(x, t) :

wtt = wxx , 0 < x < l, t > 0,

w(0, t) = 0 = w(l, t), t ≥ 0,

In order to have this we can choose ψ(x) so that

ψ 00 (x) + Ax = 0; ψ(0) = 0, ψ(l) = 0.

1
That is, ψ(x) = Ax (l2 − x 2 ) .
6
Finally it’s time to take care of ICs:
I u(x, 0) = 0 ⇒ w(x, 0) = −ψ(x).
I ut (x, 0) = 1 ⇒ wt (x, 0) = 1.
The solution procedure is similar to Example 3.
31
Non-homogeneous wave equation
The solution procedure is similar to Example 3.
Thus we’ve:
 
∞ Z l
2 X nπs nπx ! nπt
!  

u(x, t) =
 
s2 −l2


 As sin ds
 sin cos 
l n=1

 l  l l
0
 
∞ l
nπs nπx ! nπt
Z
2X1 !  

 
+  sin ds
 sin sin .
π n=1 n 
 l  l l
0
∞
2Al3 X (−1)n  nπx   
nπt
= 3 sin cos
π n=1 n3 l l
∞
2l X 1 − (−1)n  nπx   
nπt
+ 2 sin sin .
π n=1 n2 l l

32
WAVE EQUATION
Unbounded Domain
Wave equation on entire line.

The Cauchy problem for wave equation

utt = c2 uxx , x ∈ R, t > 0,

u(x, 0) = f (x), ut (x, 0) = g(x), x ∈ R.

The lines x − ct = k1 , x + ct = k2 with k1 , k2 are any real

constants, are called characteristics of wave equation.

Define the change of coordinates

ξ = x − ct, η = x + ct

This transformation is invertible !

34
D'Alembert's solution
Define w(ξ, η) = u(x(ξ, η), y(ξ, η)).
See that: utt − c2 uxx = 4c2 wξη
Since

ux = wξ + wη
uxx = wξξ + 2wξη + wηη
ut = c (−wξ + wη )
utt = c2 (wξξ − 2wξη + wηη )

In the new coordinate system the wave equation is

wξη = 0.

35
D'Alembert's solution
After integration we’ve:
Z
w(ξ, η) = h(η)dη + F(ξ)

= G(η) + F(ξ)

F and G are arbitrary functions of ξ and η resp.

and twice continuously differentiable
Thus the general solution of utt − c2 uxx = 0 is:

u(x, t) = F(x + ct) + G(x − ct). (1)

Choose F and G such that they satisfy the Cauchy or

initial conditions.

36
D'Alembert's solution

u(x, 0) = F(x) + G(x) = f (x) (2)

and
ut (x, 0) = cF 0 (x) − cG0 (x) = g(x) (3)
Integrating (3):
Zx
1
F(x) − G(x) = g(s)ds + F(0) − G(0)
c
0

Adding this to (2),

Zx
1 1 1 1
F(x) = f (x) + g(s)ds + F(0) − G(0)
2 2c 2 2
0

37
D'Alembert's solution
Again from (2),

G(x) = f (x) − F(x)

Zx
1 1 1 1
= f (x) − g(s)ds − F(0) + G(0)
2 2c 2 2
0

Thus
x+ct
Z
1 1
u(x, t) = (f (x − ct) + f (x + ct)) + g(s)ds.
2 2c
x−ct

Which the d’Alembert’s formula.

38
HEAT EQUATION
Bounded Domain
Heat Equation
Ends of the bar kept at zero temperature
The pde:

ut = κuxx , 0 < x < l, t > 0,

40
Heat Equation
Ends of the bar kept at zero temperature
The pde:

ut = κuxx , 0 < x < l, t > 0,

Boundary conditions (BC):

u(0, t) = 0, u(l, t) = 0, t ≥ 0,

40
Heat Equation
Ends of the bar kept at zero temperature
The pde:

ut = κuxx , 0 < x < l, t > 0,

Boundary conditions (BC):

u(0, t) = 0, u(l, t) = 0, t ≥ 0,

Intial condition (IC):

u(x, 0) = f (x), 0 ≤ x ≤ l.

40
Examples

Example 1:
Consider the problem with constant temperature,
say,
u(x, 0) = f (x) = C for all x ∈ [0, l].

41
Examples

Example 1:
Consider the problem with constant temperature,
say,
u(x, 0) = f (x) = C for all x ∈ [0, l].

Solution procedure:
Fourier method/Separation of variable method

41
Important Steps...
Suppose u(x, t) = X(x)T(t)
X 00 Ṫ
From pde: =
X κT

42
Important Steps...
Suppose u(x, t) = X(x)T(t)
X 00 Ṫ
From pde: = = −λ
X κT
From BCs and from above equations, we’ve a BVP:

X 00 + λX = 0; X(0) = X(l) = 0

The eigenvalues for the BVP are:

n2 π 2
λn = , n = 1, 2, 3, ....
l2
And the eigenfunctions are non-zero constant
multiples of
 nπx 
Xn (x) = sin , for n = 1, 2, ....
l
42
Important Steps...
Now for T(t) :
 nπ 2
Ṫ + κT = 0.
l
The general solution for T(t) is:
 2 2 
n π κt
Tn (t) = cn exp − 2
l

for each n ∈ N, with cn as yet to determine.

Now we’ve for n = 1, 2, ...
 nπx   2 2 
n π κt
un (x, t) = cn sin exp − 2 .
l l

43
Important Steps...
In order to satisfy the non-zero initial condition we
take infinite superposition
∞
n2 π 2 κt
X  nπx   
u(x, t) = cn sin exp − 2 .
n=1
l l

Choose the cn0 s to satisfy

∞
X  nπx 
u(x, 0) = f (x) = K = cn sin .
n=1
l

We can do this!
The r.h.s. is the Fourier sine expansion of f (x) = K on
[0, l].
44
Important Steps...
Thus we obtain the solution:
∞
n2 π 2 κt
X  nπx   
u(x, t) = cn sin exp − 2 .
n=1
l l

The cn0 s the Fourier sine coefficients

Zl  nπx 
2 2K
cn = K sin dx = (1 − cos(nπ)) .
l l nπ
0

0 for n even.
Thus cn = 4K
 for n odd.
nπ

45
Example 2
Temperature in a bar with insulated ends
The pde:

ut = κuxx , 0 < x < l, t > 0,

46
Example 2
Temperature in a bar with insulated ends
The pde:

ut = κuxx , 0 < x < l, t > 0,

Boundary conditions (BC):

ux (0, t) = 0, ux (l, t) = 0, t ≥ 0,

46
Example 2
Temperature in a bar with insulated ends
The pde:

ut = κuxx , 0 < x < l, t > 0,

Boundary conditions (BC):

ux (0, t) = 0, ux (l, t) = 0, t ≥ 0,

Intial condition (IC):


K l
for 0 ≤ x ≤

u(x, 0) = f (x) = 2 .
l
0 for < x ≤ l.

2
46
Important Steps...

Solution procedure:
Fourier method/Separation of variable method
Suppose u(x, t) = X(x)T(t)

X 00 Ṫ
From pde: =
X κT

47
Important Steps...

Solution procedure:
Fourier method/Separation of variable method
Suppose u(x, t) = X(x)T(t)

X 00 Ṫ
From pde: = = −λ
X κT

From BCs and from above equations, we’ve a BVP:

X 00 + λX = 0; X 0 (0) = X 0 (l) = 0

47
Important Steps...
The eigenvalues for the BVP are:

n2 π 2
λn = , n = 0, 1, 2, 3, ....
l2
And the eigenfunctions are non-zero constant
multiples of
 nπx 
Xn (x) = cos , for n = 0, 1, 2, ....
l
Now for T(t) :
 nπ 2
Ṫ + κT = 0, n = 0, 1, 2, ...
l

48
Important Steps...

The general solution for T(t) is:


constant
 2 2  for n = 0
Tn (t) = n π κt .
cn exp − 2 for n = 1, 2, ...
l

with cn as yet to determine.

Now we’ve for n = 0, 1, 2, ...

 nπx   2 2 
n π κt
un (x, t) = cn cos exp − 2 .
l l

49
Important Steps...
In order to satisfy the non-zero initial condition we
take infinite superposition
∞  2 2 
c0 X  nπx  n π κt
u(x, t) = + cn cos exp − 2 .
2 n=1
l l

Choose the cn0 s to satisfy

∞
c0 X  nπx 
u(x, 0) = f (x) = + cn cos .
2 n=1
l

We can do this!
The r.h.s. is the Fourier cosine expansion of f (x) on
[0, l].
50
Important Steps...
Thus
Zl/2
2
c0 = Kdx = K,
l
0

and
Zl/2  nπx   nπ 
2 2K
cn = K cos dx = sin .
l l nπ 2
0

Note that: (
 nπ  0 for n even,
sin =
2 (−1)m+1 for m = 2n − 1 odd.

51
Important Steps...

Hence the solution for the temperature distribution

is:

K
u(x, t) = +
2
∞
2K X (−1)2n+1 (2n − 1)2 π 2
   
(2n − 1) π
cos x exp − κt .
π n=1 (2n − 1) l l2

52
Bar with ends at different temperatures
The pde:

ut = κuxx , 0 < x < l, t > 0,

53
Bar with ends at different temperatures
The pde:

ut = κuxx , 0 < x < l, t > 0,

Boundary conditions (BC):

u(0, t) = T1 , u(l, t) = T2 , t ≥ 0,

53
Bar with ends at different temperatures
The pde:

ut = κuxx , 0 < x < l, t > 0,

Boundary conditions (BC):

u(0, t) = T1 , u(l, t) = T2 , t ≥ 0,

Intial condition (IC):

u(x, 0) = f (x) for 0 ≤ x ≤ l.

53
Solution procedure

1
Let ψ(x) = (T2 − T1 )x + T1 a linear transformation.
L

Set u(x, t) = w(x, t) + ψ(x), then we’ve

wt = κwxx , 0 < x < l, t > 0,

w(0, t) = 0 = w(l, t), t ≥ 0
w(x, 0) = f (x) − ψ(x), 0 ≤ x ≤ l.

Whose solution procedure is already known, which

follows from example 1.

54
Solution
The solution is
∞
n2 π 2 κt
X  nπx   
w(x, t) = cn sin exp − 2 .
n=1
l l

The cn0 s the Fourier sine coefficients

Zl  nπx 
2
cn = (f (x) − ψ(x)) sin dx.
l l
0

Or
Zl  nπx 
2 1
cn = (f (x) − (T2 − T1 )x − T1 ) sin dx.
l l l
0

55
Example 3
Consider the case:
3
T1 = 1, T2 = 2 and f (x) = , 0 ≤ x ≤ l.
2
Thus
Zl    nπx 
2 3 1
cn = − (2 − 1)x − 1 sin dx
l 2 l l
0
(1 + (−1)n )
= .
nπ
Hence u(x, t) =

∞
(1 + (−1)n ) n2 π 2
 nπx   
X 1
sin exp − 2 κt + x + 1.
n=1
nπ l l l
56
LAPLACE EQUATION
Bounded Domain
Laplace Equation
Basic equation
In Cartesian coordinates:

uxx + uyy = 0, (x, y) ∈ Ω.

58
Laplace Equation
Basic equation
In Cartesian coordinates:

uxx + uyy = 0, (x, y) ∈ Ω.

In polar coordinates:
1 1
urr + ur + 2 uθθ = 0, (r, θ) ∈ Ω.
r r

58
Laplace Equation
Basic equation
In Cartesian coordinates:

uxx + uyy = 0, (x, y) ∈ Ω.

In polar coordinates:
1 1
urr + ur + 2 uθθ = 0, (r, θ) ∈ Ω.
r r

It is associated with the BCs:

m Dirchlet
m Neumann
m Mixed or Robin
58
Dirchlet problem for rectangle

Example 1:

uxx + uyy = 0, 0 < x < l, 0 < y < k,

u(0, y) = 0 = u(l, y), 0 ≤ y ≤ k,
(
u(x, 0) = 0
, 0 ≤ x ≤ l.
u(x, k) = (l − k) sin x

Solution procedure:
Fourier method/Separation of variable method

59
Important Steps...
Suppose u(x, t) = X(x)Y (y)
X 00 Ÿ
From pde: =−
X Y

60
Important Steps...
Suppose u(x, t) = X(x)Y (y)
X 00 Ÿ
From pde: = − = −λ
X Y
From BCs and from above equations, we’ve a BVP:

X 00 + λX = 0; X(0) = X(l) = 0

n2 π 2
The eigenvalues for the BVP are: λn = .
l2
And the eigenfunctions are non-zero constant
multiples of
 nπx 
Xn (x) = sin , for n = 1, 2, ....
l

60
Important Steps...
Now for Y (y) : from zero boundary condition
 nπ 2
Ÿ − Y = 0; Y (0) = 0.
l
The general solution for Y (y) is:
 nπy 
Yn (y) = cn sinh
l
for each n ∈ N, with cn as yet to determine.
Now we’ve for n = 1, 2, ...
 nπx   nπy 
un (x, t) = cn sin sinh .
l l
which are harmonic on the rectangle, and satisfies
zero boundary conditions.
61
Important Steps...
In order to satisfy the non-zero boundary condition
we take infinite superposition
∞
X  nπx   nπy 
u(x, t) = cn sin sinh .
n=1
l l

Choose the cn0 s to satisfy

∞  nπx   
X nπk
u(x, k) = cn sin sinh = (l − k) sin x.
n=1
l l

Tada, once again, it is doable!

The l.h.s. is the Fourier sine expansion of (l − x) sin x
on [0, l].
62
Important Steps...
Thus Fourier sine coefficients
  Zl
nπk 2  nπx 
cn sinh = (l − x) sin(x) sin dx
l l l
0

4l2 [1 − (−1)n cos(l)] nπ

Thus cn =   , 6= 1.
nπk (l2 − n2 π 2 ) l
sinh
l
Thus we obtain the solution:
∞
X  nπx   nπy 
u(x, t) = cn sin sinh .
n=1
l l

63
Dirchlet problem for semicircle
Example 2: Obtain the steady state temperature
distribution in a semicircular plate of radius R whose
bounding diameter is kept at 0o C, while the
circumference is kept at 50o C.
The mathematical formulation of the problem is:
1 1
urr + ur + 2 uθθ = 0, 0 < r < a, 0 < θ < π,
r r
u(r, 0) = 0 = u(r, π), 0 ≤ r ≤ a,
u(a, θ) = 50, 0 ≤ θ ≤ π.

Solution procedure:
Fourier method/Separation of variable method !?

64
Important Steps...
Suppose u(r, θ) = R(r)Θ(θ)
r 2 R00 + rR0 Θ̈
From pde: − = = −λ
R Θ
From zero BCs and from above equation, we’ve a BVP:

Θ̈ + λΘ = 0; Θ(0) = Θ(π) = 0
The eigenvalues for the BVP are: λn = n2 .
And the eigenfunctions are non-zero constant
multiples of

Θn (θ) = sin(nθ), for n = 1, 2, ....

65
Important Steps...
Now for R(r) :

r 2 R00 + rR0 − n2 R = 0.

The general solution is:

R(r) = cn r n + dn r −n

Choose dn = 0, so that the solution is bounded at

origin.
Now we’ve for n = 1, 2, ...

un (r, θ) = cn r n sin (nθ) .

66
Important Steps...
In order to satisfy the non-zero boundary condition
we take infinite superposition
∞
X
u(r, θ) = cn r n sin (nθ) .
n=1

Choose the cn0 s to satisfy

∞
X
u(a, θ) = cn an sin(nθ) = 50.
n=1

Tada, once again, it is doable!

The l.h.s. is the Fourier sine expansion of 50 on [0, π].

67
Important Steps...

Thus Fourier sine coefficients


Zπ  4(50)
2 for n odd,
cn an = 50 sin (nθ) dθ = nπ
π 0 for n even.
0

Thus we’ve the solution:

∞
X 200  r 2n−1
u(r, θ) = sin (2n − 1) θ.
n=1
(2n − 1) a

68
 RUN FOR
SOME
QUESTIONS ? 69