Lecture Notes for Partial Differential Equations(3)

By

Dr. S. K. Ray

NB : These notes are provided to the students of ME, 3rd Semester (JGEC) as study materials . In case
of any query, they can contact me over Mobile No. 9474116742 after 7 :30 pm . Some problems, not
solved here, are left to them as Assignments.

Classification of Second Order Linear Partial Differential Equations and Solutions of Some Important
Equations .

The most general 2nd order linear PDE in two independent variables 𝑥, 𝑦 is of the form

𝜕2𝑢 𝜕2𝑢 𝜕2𝑢 𝜕𝑢 𝜕𝑢

𝐴 𝜕𝑥 2 + 𝐵 𝜕𝑥𝜕𝑦 + 𝐶 𝜕𝑦 2 + 𝐷 𝜕𝑥 + 𝐸 𝜕𝑦 + 𝐹𝑢 = 𝐺 … … … … 2)

where 𝐴 ,𝐵, 𝐶, 𝐷, 𝐸, 𝐹, 𝐺 are functions of 𝑥, 𝑦 or constants. If 𝐺(𝑥, 𝑦) = 0 , the Equation 2) is said to

be homogeneous; otherwise it is said to be inhomogeneous .

The classification of a PDE depends only on the highest order derivatives present.

The equation 2) is said to be Hyperbolic at (𝑥, 𝑦) if 𝐵2 − 4𝐴𝐶 > 0, Parabolic at (𝑥, 𝑦) if 𝐵2 − 4𝐴𝐶 =
0 and Elliptic at (𝑥, 𝑦) if 𝐵2 − 4𝐴𝐶 < 0. If this is true at all points in a domain D, then equation 2) is said
to be hyperbolic ,parabolic or elliptic in D.

Some important 2nd order linear PDEs are

𝜕2𝑢 𝜕2 𝑢
i) = 𝑐 2 𝜕𝑥 2 ( One-dimensional wave equation )
𝜕𝑡 2

𝜕𝑢 𝜕2𝑢
ii) = 𝑐 2 𝜕𝑥 2 ( One-dimensional heat equation )
𝜕𝑡

𝜕2𝑢 𝜕2𝑢
iii) + 𝜕𝑦 2 = 0 ( Two- dimensional Laplace equation )
𝜕𝑥 2

𝜕2 𝑢 𝜕2𝑢
iv) + = 𝑓(𝑥, 𝑦) (Two- dimensional Poisson equation )
𝜕𝑥 2 𝜕𝑦 2

where 𝑐 is a constant and 𝑡 is the time variable. Equation iv) with 𝑓(𝑥, 𝑦) ≠ 0 is inhomogeneous, while
the other equations are homogeneous.

Note that the equation i) is hyperbolic, ii) is parabolic and iii) is elliptic.

D’Alembert’s Solution of One-dimensional Wave Equation :

Consider the one dimensional wave equation

𝜕2𝑦 𝜕2𝑦
1) = 𝑎2 𝜕𝑥 2
𝜕𝑡 2

𝜕 𝜕
Writing 𝐷 ≡ and 𝐷′ ≡ 𝜕𝑥 , equation 1) becomes
𝜕𝑡

2) (𝐷2 − 𝑎2 𝐷′2 )𝑦 = 0

which is a homogeneous linear PDE with constant coefficients.

The Auxiliary Equation is 𝑚2 − 𝑎2 = 0 .  𝑚=±𝑎

 𝐶. 𝐹. = 𝜑1 (𝑥 + 𝑎𝑡) + 𝜑2 (𝑥 − 𝑎𝑡) where 𝜑1 , 𝜑2 are arbitrary functions .

Since right side of 2) is zero , 𝑃. 𝐼. = 0 .

 The general solution of 1) is

3) 𝑦(𝑥, 𝑡) = 𝜑1 (𝑥 + 𝑎𝑡) + 𝜑2 (𝑥 − 𝑎𝑡) where 𝜑1 , 𝜑2 are arbitrary functions .

The equation 3) is known as the D’Alembert’s solution of one-dimensional wave equation.

The functions 𝜑1 , 𝜑2 can be determined if initial conditions are prescribed.

Exercise 1. Obtain the D’Alembert’s solution of one-dimensional wave equation

𝜕2𝑦 𝜕2𝑦
1) = 𝑎2 𝜕𝑥 2
𝜕𝑡 2

subject to

𝜕𝑦(𝑥,0)
Initial Conditions : 𝑦(𝑥, 0) = 𝑓(𝑥) and =0
𝜕𝑡

Left to the students

Hints. Apply initial conditions to equation 3) above to obtain

4) 𝜑1 (𝑥) + 𝜑2 (𝑥) = 𝑓(𝑥)

and 5) 𝜑′1 (𝑥) − 𝜑′2 (𝑥) = 0

Integrate 5) between 𝑥0 and 𝑥 to obtain

6) 𝜑1 (𝑥) − 𝜑2 (𝑥) = 𝜑1 (𝑥0 ) − 𝜑2 (𝑥0 ) where 𝑥0 is arbitrary.

Solve 4) and 6) to obtain 𝜑1 (𝑥) and 𝜑2 (𝑥) and use in equation 3) to obtain
1
𝑦(𝑥, 𝑡) = [ 𝑓(𝑥 + 𝑎𝑡) + 𝑓(𝑥 − 𝑎𝑡)]
2
Duhamel’s Principle for One-dimensional Wave Equation :

If 𝑣(𝑥, 0, 𝜏) for each fixed 𝜏 satisfies the one-dimensional wave equation

1) 𝑣𝑡𝑡 = 𝑎2 𝑣𝑥𝑥 , 𝑡 > 0

with initial conditions 𝑣(𝑥, 0, 𝜏) = 0 and 𝑣𝑡 (𝑥, 0, 𝜏) = 𝐹(𝑥, 𝑡)

𝑡
then 𝑢(𝑥, 𝑡) defined by 𝑢(𝑥, 𝑡) = ∫0 𝑣(𝑥, 𝑡 − 𝜏, 𝜏)𝑑𝜏

satisfies the inhomogeneous problem

2) 𝑢𝑡𝑡 = 𝑎2 𝑢𝑥𝑥 + 𝐹(𝑥, 𝑡) , 𝑡 > 0

with initial conditions 𝑢(𝑥, 0) = 0 and 𝑢𝑡 (𝑥, 0) = 0

where 𝐹(𝑥, 𝑡) is an external driving force.

Solution of one- dimensional wave equation by the method of Separation of Variables.

The transverse displacement 𝑦(𝑥, 𝑡) of a vibrating string of length L fixed at 𝑥 = 0 and 𝑥 = 𝐿 satisfies
the differential equation

𝜕2𝑦 𝜕2𝑦
1) = 𝑎2 𝜕𝑥 2
𝜕𝑡 2

subject to

Boundary Conditions : 𝑦(0, 𝑡) = 0, 𝑦(𝐿, 𝑡) = 0

𝜕𝑦(𝑥,0)
and Initial Conditions : 𝑦(𝑥, 0) = 𝑓(𝑥) and =0
𝜕𝑡

where 𝑓(𝑥) is the initial position of the string .

The main assumption in the method of separation of variables is to assume the solution in the form

𝑦(𝑥, 𝑡) = 𝑋(𝑥)𝑇(𝑡)

where 𝑋(𝑥) is a function of 𝑥 alone and 𝑇(𝑡) is a function of 𝑡 alone . Substituting this into the given
equation 1), we obtain

𝑋(𝑥)𝑇’’(t) = 𝑎2 𝑋 ′′ (𝑥)𝑇(𝑡)
𝑋′′ 𝑇′′
or = 𝑎2 𝑇 = (constant) = −β2 , say
𝑋

We choose  = −β2 < 0 to obtain solution consistent with the physical nature of the problem.

 𝑋 ′′ + β2 X = 0 and 𝑇 ′′ + β2 𝑎2 𝑇 = 0
The solution of 𝑋 ′′ + β2 X = 0 is 𝑋(𝑥) = 𝐴 cos 𝛽𝑥 + 𝐵 sin 𝛽𝑥 .

The boundary condition 𝑦(0, 𝑡) = 0 gives 𝑋(0)𝑇(𝑡) = 0 which implies 𝑋(0) = 0 [ since 𝑇(𝑡) ≠ 0 ]

Now 𝑋(0) = 0 gives 𝐴 cos 0 + 𝐵 sin 0 = 0 ,i.e., 𝐴 = 0 .

 𝑋(𝑥) = 𝐵 sin 𝛽𝑥

The boundary condition 𝑦(𝐿, 𝑡) = 0 gives 𝑋(𝐿)𝑇(𝑡) = 0 which implies 𝑋(𝐿) = 0.

 𝑋(𝐿) = 𝐵 sin 𝛽𝐿 = 0
𝑛𝜋
 sin 𝛽𝐿 = 0  𝛽𝐿 = 𝑛𝜋 or 𝛽 = , 𝑛 = 1, 2, 3, ………
𝐿

𝑛𝜋
 For each 𝑛, we get a value of 𝛽 and a different solution 𝑋𝑛 (𝑥) = 𝐵𝑛 sin 𝑥 .
𝐿

The solution of 𝑇 ′′ + β2 𝑎2 𝑇 = 0 is 𝑇(𝑡) = 𝐶 cos 𝛽𝑎𝑡 + 𝐷 sin 𝛽𝑎𝑡

and its derivative is

𝑇 ′ (𝑡) = −𝛽𝑎𝐶 sin 𝛽𝑎𝑡 + 𝛽𝑎𝐷 cos 𝛽𝑎𝑡

𝜕𝑦(𝑥,0)
The initial condition = 0 gives 𝑋(𝑥)𝑇 ′ (0) = 0 and hence 𝑇 ′ (0) = 0 .
𝜕𝑡

Thus 𝑇 ′ (0) = −𝛽𝑎𝐶 sin 0 + 𝛽𝑎𝐷 cos 0 = 0 ,i.e., 𝐷 = 0

 𝑇(𝑡) = 𝐶 cos 𝛽𝑎𝑡

𝑛𝜋
Since 𝛽 = , we get
𝐿

𝑛𝜋
𝑇𝑛 (𝑡) = 𝐶𝑛 cos 𝑎𝑡
𝐿

𝑛𝜋 𝑛𝜋
 𝑋𝑛 (𝑥)𝑇𝑛 (𝑡) = 𝐵𝑛 sin 𝑥 𝐶𝑛 cos 𝑎𝑡
𝐿 𝐿

𝑛𝜋 𝑛𝜋
= 𝑏𝑛 sin 𝑥 cos 𝑎𝑡 (𝑏𝑛 = 𝐵𝑛 𝐶𝑛 )
𝐿 𝐿

Since equation 1) is linear, any linear combination of solutions is a solution and hence the series
𝑛𝜋 𝑛𝜋
𝑦(𝑥, 𝑡) = ∑∞
𝑛=1 𝑏𝑛 sin 𝑥 cos 𝑎𝑡 … … … … 2)
𝐿 𝐿

is a solution where 𝑏𝑛 are obtained from Initial Condition 𝑦(𝑥, 0) = 𝑓(𝑥) .

𝑛𝜋
Now 𝑦(𝑥, 0) = 𝑓(𝑥) gives 𝑓(𝑥) = ∑∞
𝑛=1 𝑏𝑛 sin 𝑥, 0<𝑥<𝐿
𝐿
2 𝐿 𝑛𝜋
where 𝑏𝑛 = ∫ 𝑓(𝑥) sin 𝑥 𝑑𝑥 … … …3) , 𝑏𝑛 being the Fourier coefficients of half range sine
𝐿 0 𝐿
series .

 The solution of 1) is given by 2) where 𝑏𝑛 are given by 3).

Example 2. A string is stretched and fastened to two points L apart. Motion is started by displacing the
𝜋𝑥
string in the form 𝑦 = 𝑐 sin from which it is released at time 𝑡 = 0. Show that the displacement of
𝐿
𝜋𝑥 𝜋𝑎𝑡
any point at a distance 𝑥 from one end at time 𝑡 is given by 𝑦(𝑥, 𝑡) = 𝑐 sin 𝐿
cos 𝐿 .

Solution

The displacement 𝑦(𝑥, 𝑡) satisfies the one-dimensional wave equation

𝜕2𝑦 𝜕2𝑦
1) = 𝑎2
𝜕𝑡 2 𝜕𝑥 2

subject to

2) Boundary Conditions : 𝑦(0, 𝑡) = 0, 𝑦(𝐿, 𝑡) = 0

𝜋𝑥 𝜕𝑦(𝑥,0)
and 3) Initial Conditions : 𝑦(𝑥, 0) = 𝑓(𝑥) = 𝑐 sin and =0
𝐿 𝜕𝑡

The solution of 1) satisfying the conditions 2) and 3) is given by

𝑛𝜋 𝑛𝜋
𝑦(𝑥, 𝑡) = ∑∞
𝑛=1 𝑏𝑛 sin 𝑥 cos 𝑎𝑡
𝐿 𝐿

2 𝐿 𝜋𝑥 𝑛𝜋
where 𝑏𝑛 = ∫ (𝑐 sin 𝐿 ) sin 𝐿 𝑥 𝑑𝑥 which vanishes ∀ 𝑛 except when 𝑛 = 1
𝐿 0

2 𝐿 𝜋𝑥 𝜋𝑥 𝑐 𝐿 𝜋𝑥 𝑐
Now 𝑏1 = ∫ (𝑐 sin 𝐿 ) sin 𝐿 𝑑𝑥 = 𝐿 ∫0 2𝑠𝑖𝑛2 𝑑𝑥 = 𝐿 𝐿 = 𝑐
𝐿 0 𝐿

𝑛𝜋 𝑛𝜋
 𝑦(𝑥, 𝑡) = ∑∞
𝑛=1 𝑏𝑛 cos 𝑎𝑡 sin 𝑥
𝐿 𝐿

𝜋𝑎𝑡 𝜋𝑥
= 𝑏1 cos sin , since 𝑏𝑛 vanishes for 𝑛 > 1
𝐿 𝐿

𝜋𝑥 𝜋𝑎𝑡
= 𝑐 sin cos
𝐿 𝐿

Exercise 2. A tightly stretched string has its ends fixed at 𝑥 = 0 and 𝑥 = 𝑙. At time 𝑡 = 0 , the string is
given a shape defined by 𝑓(𝑥) = 𝜇𝑥(𝐿 − 𝑥), where 𝜇 is a constant, and then released. Find the
displacement of any point 𝑥 of the string at any time 𝑡 > 0.

Left to the students

Solution of one- dimensional Heat equation by the method of Separation of Variables

The temperature distribution 𝑢(𝑥, 𝑡) in a thin insulated bar or wire of length L of which the ends 𝑥 = 0
and 𝑥 = 𝐿 are kept at zero temperature satisfies the differential equation

𝜕𝑢 𝜕2𝑢
= 𝑎2 𝜕𝑥 2
𝜕𝑡

subject to

2) Boundary Conditions : 𝑢(0, 𝑡) = 0, 𝑦(𝐿, 𝑡) = 0

and 3) Initial Conditions : 𝑢(𝑥, 0) = 𝑓(𝑥)

where 𝑓(𝑥) is the initial temperature in the bar.

Let 𝑢(𝑥, 𝑡) = 𝑋(𝑥)𝑇(𝑡) be a solution of 1). Then 1) becomes

𝑋(𝑥)𝑇 ′ (𝑡) = 𝑎2 𝑋′′(𝑥)𝑇(𝑡)

𝑋′′ 𝑇′
or = 𝑎2 𝑇 =  = −β2
𝑋

We choose  = −β2 < 0 to obtain solution consistent with the physical nature of the problem.

 𝑋 ′′ + β2 X = 0 and 𝑇 ′ + β2 𝑎2 𝑇 = 0

The solution of 𝑋 ′′ + β2 X = 0 is 𝑋(𝑥) = 𝐴 cos 𝛽𝑥 + 𝐵 sin 𝛽𝑥 .

The boundary condition 𝑢(0, 𝑡) = 0 gives 𝑋(0)𝑇(𝑡) = 0 which implies 𝑋(0) = 0 and this gives
𝐴 cos 0 + 𝐵 sin 0 = 0 i.e., 𝐴 = 0 .

 𝑋(𝑥) = 𝐵 sin 𝛽𝑥

The boundary condition 𝑢(𝐿, 𝑡) = 0 gives 𝑋(𝐿)𝑇(𝑡) = 0 which implies 𝑋(𝐿) = 0 and this gives
𝑛𝜋
𝐵 sin 𝛽𝐿 = 0  sin 𝛽𝐿 = 0  𝛽𝐿 = 𝑛𝜋 or 𝛽 = , 𝑛 = 1, 2, 3, ………
𝐿

𝑛𝜋
 For each 𝑛, we get a value of 𝛽 and a different solution 𝑋𝑛 (𝑥) = 𝐵𝑛 sin .
𝐿

2 a2 t
The solution of 𝑇 ′ + β2 𝑎2 𝑇 = 0 is 𝑇(𝑡) = 𝐶 e−𝛽
𝑛𝜋
Since 𝛽 = , we get
𝐿

𝑛2 𝜋2 2
− a t
𝑇𝑛 (𝑡) = 𝐶𝑛 e 𝐿2
𝑛2 𝜋2 2
− a t 𝑛𝜋
 𝑋𝑛 (𝑥)𝑇𝑛 (𝑡) = 𝑎𝑛 e 𝐿2 sin 𝑥 , 𝑎𝑛 = 𝐵𝑛 𝐶𝑛
𝐿

The solution of 1) is then obtained as

𝑛2𝜋2 2
− a t 𝑛𝜋
𝑢(𝑥, 𝑡) = ∑∞
𝑛=1 𝑎𝑛 e 𝐿2 sin 𝑥 … … … … 2)
𝐿

where 𝑎𝑛 are obtained from the Initial Condition 𝑢(𝑥, 0) = 𝑓(𝑥)

𝑛𝜋
The initial condition 𝑢(𝑥, 0) = 𝑓(𝑥) gives 𝑓(𝑥) = ∑∞
𝑛=1 𝑎𝑛 sin 𝑥, 0<𝑥<𝐿
𝐿

2 𝐿 𝑛𝜋
 𝑎𝑛 = ∫ 𝑓(𝑥) sin 𝐿 𝑥 𝑑𝑥 … … …3)
𝐿 0

The solution of 1) is then given by 2) where 𝑎𝑛 are given by 3) .

Example 3. A homogeneous rod of conducting material of length L cm is initially at a uniform

temperature 𝑢0 . Its ends are suddenly cooled at 00 𝐶 and are kept at that temperature. Find the
temperature 𝑢(𝑥, 𝑡) at any time 𝑡 .

Solution:

The temperature distribution 𝑢(𝑥, 𝑡) in the rod is governed by

𝜕𝑢 𝜕2𝑢
1) = 𝑎2 𝜕𝑥 2
𝜕𝑡

subject to

2) Boundary Conditions : 𝑢(0, 𝑡) = 0, 𝑦(𝐿, 𝑡) = 0

and 3) Initial Conditions : 𝑢(𝑥, 0) = 𝑓(𝑥) = 𝑢0

The solution 𝑢(𝑥, 𝑡) of 1) is

𝑛2 𝜋2 2
− a t 𝑛𝜋
𝑢(𝑥, 𝑡) = ∑∞
𝑛=1 𝑎𝑛 e 𝐿2 sin 𝑥 … … … … 2)
𝐿

2 𝐿 𝑛𝜋 2 𝐿 𝑛𝜋 2𝑢0 𝐿 𝑛𝜋
where 𝑎𝑛 = ∫ 𝑓(𝑥) sin 𝐿 𝑥 𝑑𝑥 = 𝐿 ∫0 𝑢0 sin 𝑥 𝑑𝑥 = ∫0 sin 𝑥 𝑑𝑥
𝐿 0 𝐿 𝐿 𝐿

2𝑢0 𝐿 𝑛𝜋𝑥 2𝑢0 𝐿 2𝑢0

= [ − cos ] 𝐿0 = [ − (−1)𝑛 + 1 ] = [ 1 − (−1)𝑛 ]
𝐿 𝑛𝜋 𝐿 𝐿 𝑛𝜋 𝑛𝜋

0 𝑤ℎ𝑒𝑛 𝑛 𝑖𝑠 𝑒𝑣𝑒𝑛
= {4𝑢0 𝑤ℎ𝑒𝑛 𝑛 𝑖𝑠 𝑜𝑑𝑑
𝑛𝜋
𝑛2𝜋2 2
4𝑢0 − a t 𝑛𝜋
 𝑢(𝑥, 𝑡) = ∑∞
𝑛=1,3,5,…… e 𝐿2 sin 𝑥=
𝑛𝜋 𝐿
(2𝑛−1)2 𝜋2 a2
4𝑢0 1 − t (2𝑛−1)𝜋
∑∞
𝑛=1 e 𝐿2 sin 𝑥
𝜋 2𝑛−1 𝐿

Exercise 3. A homogeneous rod of conducting material of length L cm has its ends kept at zero
temperature and the initial temperature is

𝐿
𝑥 ,0 ≤ 𝑥 ≤ 2
𝑢(𝑥, 𝑡) = { 𝐿
𝐿−𝑥, ≤𝑥≤𝐿
2

Find the temperature 𝑢(𝑥, 𝑡) at any time 𝑡 .

Left to the students