IMSC (PHY. & CHEM.) - IV SEM.

MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

TOPICS COVERED: Method of separation of variables and its application in solving one
dimensional wave and heat equations

3.1 Introduction
Problems of fluid flow, vibrations of a string, heat flow etc. give rise to partial
differential equations (pdes) of second order, hence, it is important to study methods
to solve second order pdes according to the given physical problem. In case of
physical or engineering problems, pdes are always accompanied by certain
conditions (boundary and initial conditions), so as to form the unique (or particular)
solution of the given pde. If the given conditions are specified at the boundary of the
region, then these are referred to as boundary conditions (BCs) and if conditions
are specified at time t = 0, then these are referred to as initial conditions (ICs). For
solving pdes with given initial & boundary conditions, method of separation of
variables is the most efficient analytical tool.

3.2 Classification of Linear Partial differential Equation of Second

order
If we consider PDE of second order in two independent variables x and y as:

2z 2z 2z  z  z 

A  B  C  f  x , y , z , , 0 ------------- (*)
x 2  x y y 2   x y 

where A, B, C are either constants or functions of x, y and A is positive. Then equation

(*) can be classified by the value B 2  4 AC in the following way:

I. if B 2  4 AC  0, then PDE is called Hyperbolic.

II. if B 2  4 AC  0 , then PDE is called Parabolic.

III. if B 2  4 AC  0 , then PDE is called Elliptic.

1
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

Note: If A, B, C in the given equation (*) are constants, then the nature of equation is
same for all values of x and y, if A, B, C are functions of x and y, then nature of
equation will not be same for all values of x and y.
PROBLEMS

 2u  2u  2u  2u  2u  2u
Problem 3.1: Classify the operator i) 2   ii) 5 2  9 4 2
t x t x 2 x xt t

Problem 3.2: Show that the equation u xx  xu yy  u y  0 is elliptic for x  0 & hyperbolic for x  0. .

3.3 Method of Separation of Variables (or Product Rule)

In this method of solving pde of second order, it can be assumed that the solution is
a product of two functions, each of which involves only one variable e.g. for given
 2u  2u
pde,  , suppose that the solution u( x, t )  X ( x)T (t ) . Substituting this in
t 2 x 2
given p.d.e. will result into two ordinary differential equations. Solution of these
ordinary differential equations in accordance to given boundary and initial
conditions will give X (x ) and T (t ) , hence we get the required solution u ( x, t )  X ( x)T (t )
.

Basics Concepts:

Principle of superposition: If u1 , u2 ,....,un ,...... be an infinite set of solutions for any

homogeneous (RHS = 0), second order linear p.d.e. of the form
 2u  2u  2u u u
A 2 B C 2  D E  Fu  0
x  x y y x y

then u   ui (linear combination of solutions) will also be its solution.
n 1

2
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

Fourier Sine series: If it is required to expand any function f (x) as a Fourier series
of sine terms in 0  x  l , then its expansion will be given as:
 nx 2l n x
f ( x )   bn sin , where bn   f ( x ) sin dx
n 1 l l 0 l

Fourier cosine series: If it is required to expand any function f (x) as a Fourier

series of cosine terms in 0  x  l , then its expansion will be given as:
a0  nx 2l 2l nx
f ( x)    an cos , where a 0   f ( x )dx , a n   f ( x ) cos dx
2 n 1 l l 0 l 0 l

3.4 One - Dimensional Wave equation – Vibrations of a Stretched String

One dimensional wave equation is of the form:
 2u 2  u
2
c
t 2 x 2
T
with c 2  , where T is the tension of the string and m is mass per unit length of the
m
string.
Physical Problem: This PDE governs the transverse vibrations of an elastic string,
such as a violin string. The string is placed along the x-axis, stretched it to length L
and then, fixed at the ends x  0 and x  L . We then distort it and at some instant,
say t  0, then it is released and allowed to vibrate. The problem is to determine the
vibrations of the string, that is, to find the deflection u( x, t ) at any point x and at any
time t.
Note:
i) Wave equation is hyperbolic type, being B 2  4 AC  0  4.1. c 2   4c 2  0

3
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

ii) For the wave equation u( x, t )  0 will always be the solution of this equation,
called trivial solution. But, it is of interest to find the solution u( x, t )  0, called
non trivial solution.
The above physical problem can be mathematically formulated in the following
form, with boundary and initial conditions in order to find the particular solution.

Problem for Vibrations of a String

The governing PDE i.e. Wave equation will be of the form:

 2u 2  u
2
 c --------- (*)
t 2 x 2
Boundary Conditions: I) u(0, t )  0 i.e. u  0 when x  0 for all values of t
II) u( L, t )  0 i.e. u  0 when x  L for all values of t

Initial Conditions: I) u( x,0)  f ( x) i.e. u  f (x) at t  0

 u  u
II)  g ( x ) i.e. velocity = g (x ) at t  0
 t  t
t 0

Solution of Wave Equation (*) satisfying BCs & ICs

STEP-1: Applying method of separation of variables to get two odes & their
solutions
Let
u( x, t )  X ( x)T (t ) --------- (1)
which are product of two functions, where X is a function of x only and T is a
function of t only, be a solution of the Wave Equation (*). Then from equation (1)
we get,
 2u  2u
 XT and 2  X ' 'T ----------- (2)
''
t 2
x

4
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

'
where denotes differentiation with respect to their corresponding independent
variable. Substituting these values in Wave equation (*), we get:
XT   c 2 X T ----------- (3)
On separating the variables in equation (3), we get:
X '' 1 T ''
 2 ---------- (4)
X c T
Since x and t are independent variables, so equation (4) will hold only, if both sides
reduces to a constant ,say, k (called separation constant), hence we get equation (4)
as:
X '' 1 T ''
 2 k ------------ (5)
X c T
From this we get two ordinary differential equations as:
X '' d2X
k  kX  0 ------- (6)
X dx 2
and
1 T '' d 2T
2
 k  2  c 2 kT  0 -------- (7)
c T dt
Here, the constant k is arbitrary and we consider three cases for values of k .
Case- 1: If k is positive i.e. k  p 2 (say), then ode’s (6) and (7) reduce to:

d2X
2
 p 2 X  0. Solution of this ode is X  c1e px  c2 e  px ------ (8)
dx
d 2T
2
 c 2 p 2T  0 , Solution of this ode is: T  c3e pct  c4 e  pct ----- (9)
dt
From these solutions (8) and (9), we get the solution u( x, t )  X ( x)T (t ) as:

 
u( x, t )  X ( x )T (t )  c1e px  c2 e  px ( c3e pct  c4 e  pct ) ----- (A)

5
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

( k is +ve i.e. k  p 2 )
Case- 2: If k is zero i.e. k  0, then ode’s (6) and (7) reduce to:

d2X
 0 , Solution of this ode is: X  c1  c2 x ------ (10)
dx 2
d 2T
 0 , Solution of this ode is T  c3  c4 t ----- (11)
dt 2
From these solutions (10) and (11), we get the solution u ( x, t )  X ( x)T (t ) as:

u( x, t )  X ( x )T (t )  c1  c2 x ( c3  c4t ) ----- (B)

( k is equal to zero)
Case-3: If k is negative i.e. k   p 2 (say), then ode’s (6) & (7) reduce to:

d2X
2
 p 2 X  0 , Solution of this ode is X  c1 cos px  c2 sin px ------ (12)
dx
d 2T
2
 c 2 p 2T  0 , Solution of this ode is T  c3 cos cpt  c4 sin cpt ----- (13)
dt
From these solutions (12) & (13) we get the solution u ( x, t )  X ( x)T (t ) as:

u( x, t )  X ( x )T (t )  c1 cos px  c2 sin px ( c3 cos cpt  c4 sin cpt ) -----(C)

( k is –ve i.e. k   p 2 )
So various values of k , the various possible solutions of wave equation are:

 
A) u( x, t )  X ( x )T (t )  c1e px  c2 e  px ( c3e pct  c4 e  pct )
B) u( x, t )  X ( x )T (t )  c1  c2 x ( c3  c4 t )
C) u( x, t )  X ( x )T (t )  c1 cos px  c2 sin px ( c3 cos cpt  c4 sin cpt )
Of these three possible solutions A, B & C, we have to choose the solution which is
in consistent with the physical nature of the problem (also gives non-trivial solution

6
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

with BCs and ICs) since we are dealing with a problem on vibrations, hence u must
be periodic function of x and t. . So, solution should have trigonometric terms.
So,
u( x, t )  X ( x )T (t )  c1 cos px  c2 sin px ( c3 cos cpt  c4 sin cpt ) -------- (14)

is the only suitable solution of the wave equation , corresponding to k  p 2 .

STEP -2: Applying BCs to the obtained suitable solution from Step -1

Suitable solution is
u( x, t )  X ( x )T (t )  c1 cos px  c2 sin px ( c3 cos cpt  c4 sin cpt ) ---- (14)
BCs are I) u(0, t )  0 II) u( L, t )  0
i) Applying BC I) to the equation (14) we get:
u(0, t )  0  c1 (c3 cos cpt  c4 sin cpt )  c1 (c3 cos cpt  c4 sin cpt )  0
which gives c1  0
(since c3 cos cpt  c4 sin cpt  0 , since sine and cosine terms cannot be zero
simultaneously).
ii) Applying BC II) to the equation (14), with c1  0 , we get:

u( L, t )  0  (c2 sin pL )( c3 cos cpt  c4 sin cpt )  c2 sin pL(c3 cos cpt  c4 sin cpt )  0

which gives sin pL  0

(since c3 cos cpt  c4 sin cpt  0 and c2  0 as if c2 also becomes zero along
with c1 , we get u( x, t )  0 (trivial solution), that is not required.)
sin pL  0 implies that
sin pL  sin n  pL  n ( n  I ) or we get

7
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

n
or p
L
where n  1,2,3,......( n  0 , as p  0 & discarded negative integers as we are
dealing with physical problem(frequencies)).
Hence, from these BCs, we get the following form of equation (14) as:
nx  nc nct 
u( x, t )  X ( x )T (t )  c2 sin  c3 cos t  c4 sin 
L  L L 
 nc nct  nx
  an cos t  bn sin  sin
 L L  L
where, an  c2 c3 and bn  c2 c4 are the new values for constants, hence we have:

 nc nct  nx

u( x, t )   an cos t  bn sin  sin --------------- (15)
 L L  L
This equation (15) represents solution for any one particular value of n , but we will
get such type of solutions for various values of n ( n  1,2,3,...... ), hence by
superposition theorem, adding up all such solutions we get:
   nc nct  nx

u ( x, t )   u( x, t )   u( x, t )    an cos t  bn sin  sin ------- (16)
n 1 n 1 n 1 L L  L
Here, solutions for different values of n ( n  1,2,3,......) are called eigen functions
n
and n  are called eigenvalues.
L
STEP -3: Applying ICs to the obtained suitable solution from Step -2
After step 2, we get
   nc nct  nx

u ( x, t )   u ( x, t )   u ( x , t )    an cos L t  bn sin L  sin L --------- (16)
n 1 n 1 n 1 

 u 
Now, we have ICs as: I) u( x,0)  f ( x) II)    g ( x)
 t  t  0

8
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

i) Applying IC: u( x,0)  f ( x) to the equation (16), we get:
 nx
u( x,0)  f ( x )   an sin L
----------- (17)
n 1

Equation represents Fourier sine series (Half Range) in interval (0, L ),

therefore,
2 L nx
an  
L 0
f ( x ) sin
L
---------- (18)

 u 
ii) Applying IC    g ( x ) to the equation (16), we get
 t  t  0
First differentiating equation (16) partially w.r.t. t , we get:

u  nc nc nc nc  nx
    an sin t bn cos t  sin
t n 1 L L L L  L
Now, putting t  0 , we get
u  nc nx
 
 g ( x)    bn  sin -------------- (19)
t t  0 n 1 L  L
This also represents Fourier sine series (Half Range) in interval (0, L ),
therefore,
nc 2L nx
bn   f ( x ) sin dx
L L0 L

or
2 L
nx
bn 
nc
 f ( x ) sin
L
dx ---------- (20)
0

Hence, finally we get the required solution of Wave equation with given BCs & ICs
as:

9
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

 nc nct  nx

u ( x, t )    an cos t  bn sin  sin -------------------
n 1 L L  L
(**)
2 L nx 2 L nx
where, an   f ( x ) sin and bn   f ( x ) sin dx
L 0 L nc 0 L

 u 
Corollary: If initial velocity is zero, i.e.    0, then bn  0. So we get
 t  t  0
solution as:
 nc nx 2L nx
u( x, t )   an cos t sin where, a n   f ( x ) sin dx
n 1 L L L0 L

3.5 One - Dimensional Heat equation – Heat flow in a Bar

u 2  u
2
One dimensional heat equation is of the form: c
t x 2
K
with c 2  , where c 2 is called diffusivity of the material of the bar.
S
Physical Problem: This PDE governs the temperature in a long thin bar or wire of
length L oriented along the x  -axis and is perfectly insulated laterally, so that heat
flows in x  direction only. Here, the temperature u depends only on x and time t.
The problem is to determine the temperature u( x, t ) in a bar at any point x and at any
time t . Here the solution of this heat equation should be of transient type (i.e.
temperature u decreases as time t increases)
The heat equation is of parabolic type, being B 2  4 AC  0  4.0.  c 2  0.  

10
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

Steady State: A condition (phenomena) is said to be in steady state if the dependent
variable is independent of time e.g. for heat equation in steady state, temperature u
2
u d u
will be independent of time i.e.  0 and heat equation reduces to 0
t dx 2

Problem for Heat flow in a Bar of a String

The governing PDE i.e. Wave equation will be of the form:

u  2u
 c2 2 --------- (*)
t x
Boundary Conditions: I) u(0, t )  0 i.e. u  0 when x  0 for all values of t
II) u( L, t )  0 i.e. u  0 when x  L for all values of t
Initial Conditions: I) u( x,0)  f ( x) i.e. u  f (x) at t  0

Here, we need only one initial condition as compared to two in Wave Equation.

Solution of Heat Equation (*) satisfying BCs & ICs

STEP-1: Applying method of separation of variables to get two ode’s & their solutions
Let
u( x, t )  X ( x)T (t ) --------- (1)
which are product of two functions , where X is a function of x only and T is a
function of t only, be a solution of the Heat Equation (*).
Then from equation (1) we get,

u dT  2u d 2 X
X  XT and 2 
'
T  X ' 'T ----------- (2)
t dt x dx 2

11
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

where dash denotes differentiation with respect to their corresponding independent
variable. Substituting these values in heat equation (*), we get:
XT '  c 2 X ' 'T ----------- (3)
On separating the variables in equation (3) and equating to separation constant k , we
get:
X '' 1 T'
 2 k ---------- (4)
X c T
From this we get two ordinary differential equations as:
X '' d2X
k  kX  0 ------- (5)
X dx 2
and
1 T' dT
2
 k   c 2 kT  0 -------- (6)
c T dt
Here, the constant k is arbitrary and so, we can consider three cases for values of k.
Case- 1: If k is positive i.e. k  p 2 (say), then odes (5) and (6) reduce into:

d2X
2
 p 2 X  0. Solution of this ode is X  c1e px  c2 e  px ------ (7)
dx
dT 2 2
 c 2 p 2T  0 , Solution of this ode is T  c3e p c t -------- (8)
dt
From these solutions (7) & (8) we get the solution u( x, t )  X ( x)T (t ) as:

 
u( x, t )  X ( x )T (t )  c1e px  c2 e  px (c3e p
2 2
c t
) ----- (A)

( k is +ve i.e. k  p 2 )
Case- 2: If k is zero i.e. k  0 , then odes (5) and (6) reduce into:

12
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

d2X
 0, , Solution of this ode is X  c1  c2 x ------ (9)
dx 2
dT
 0, , Solution of this ode is T  c3 ----- (10)
dt
From these solutions (9) & (10) we get the solution u( x, t )  X ( x)T (t ) as:

u( x, t )  X ( x )T (t )  c1  c2 x c3 ----- (B)

( k is equal to zero)
Case-3: If k is negative i.e. k   p 2 (say), then odes (5) and (6) reduces into:

d2X
2
 p 2 X  0 , Solution of this ode is: X  c1 cos px  c2 sin px ------ (11)
dx
dT
 c 2 p 2T  0 , Solution of this ode is T  c3e  p c t ----- (12)
2 2

dt
From these solutions (12) & (13) we get the solution u ( x, t )  X ( x)T (t ) as:

u( x, t )  X ( x )T (t )  c1 cos px  c2 sin px c3e  p

2 2
c t
----- (C) ( k is – ve

i.e. k   p 2 )
So for various values of k , we get various possible solutions of heat equation as:

 
A) u( x, t )  X ( x )T (t )  c1e px  c2 e  px c3e p
2 2
c t

B) u( x, t )  X ( x )T (t )  c1  c2 x c3

C) u( x, t )  X ( x )T (t )  c1 cos px  c2 sin px c3e  p

2 2
c t

Of these three possible solutions A, B & C, we have to choose the solution which is
in consistent with the physical nature of the problem, since u decreases as time t
increases (transient solution).
So,

13
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

u( x, t )  X ( x )T (t )  c1 cos px  c2 sin px c3e  p

2 2
c t
-------- (13)

is the suitable solution of the heat equation, corresponding to k   p 2 . .

STEP -2: Applying BCs to the obtained suitable solution from Step -1
Suitable solution is:

u( x, t )  X ( x )T (t )  c1c3 cos px  c2 c3 sin px e  p   An cos px  Bn sin px e  p

2 2 2 2
c t c t

------- (14)
BCs are I) u(0, t )  0 II) u( L, t )  0
i) Applying BC I) to the equation (14) we get:

u(0, t )  0  An e  p
2 2
c t

(as e  p
2 2
It implies that An  0 c t
 0)
ii) Applying BC II) to the equation (14), taking An  0 , we get

u( L, t )  0  ( Bn sin pL )e  p  Bn sin pLe  p

2 2 2 2
c t c t
0
which gives
sin pL  0 (as Bn  0 (to avoid trivial solution) and

e p
2 2
c t
 0)
sin pL  sin n  pL  n ( n I )
or
n
p where n  1,2,3,......
L
Hence, from these BCs, we get the following form of equation (14) as:
n 2 2 c 2
nx  t
u( x, t )  X ( x )T (t )  Bn sin e L2 ----------------- (15)
L

14
ABHINAV TANDON
IMSC (PHY. & CHEM.) - IV SEM. MA207R1 MATHEMATICS - IV

MODULE III: PARTIAL DIFFERENTIAL EQUATIONS

This equation (15) represents solution for any one particular value of n, but we will
get such type of solutions for various values of n ( n  1,2,3,......). Hence by
superposition theorem, adding up all such solutions we get
n 2 2 c 2
  nx  t
u ( x, t )   u( x, t )   Bn sin L
e L2 ------------------ (16)
n 1 n 1

STEP -3: Applying ICs to the obtained suitable solution from Step -2
n 2 2 c 2
 nx  t
After step 2, we get u( x, t )   Bn sin e L2 -------- (16)
n 1 L
Now, we have ICs as: u( x,0)  f ( x)
Applying IC: u( x,0)  f ( x) to the equation (16), we get
 nx
u( x,0)  f ( x )   Bn sin L
----------- (17)
n 1

This equation represents Fourier sine series (Half Range) in interval (0, L), therefore,

2L nx
Bn   f ( x ) sin dx --------- (18)
L0 L

Finally, we get the solution of the heat equation (*) as:

n 2 2 c 2
 nx  t
u( x, t )   Bn sin L
e L2 ------------------- (**)
n 1

2L nx
where, Bn   f ( x ) sin dx
L0 L

Note: If it is given that both ends of the rod are insulated, no heat flows through
u u
them and then the corresponding BCs are:  0 and 0
x x 0 x xL

15
ABHINAV TANDON