End Semester Examination: MA 201 Mathematics III

Department of Mathematics, IIT Guwahati

Date: November 23, 2022 Time: 9:00–12:00 hours Max Marks: 50

Full Name: Roll No:

Instructions:
(I) Write your name and roll number above as soon as you receive the question paper.
(II) Make sure that you are in the designated room as per the seating plan.
(III) Enter the Question number and the corresponding page number(s) in the front page of the
booklet as instructed.
(IV) Use of calculator is not permitted.
(V) Not adhering to Instructions above will invite penalty.
(VI) The question paper contains 15 questions with question numbers 1, 14 and 15 having sub-parts
(a) and (b).
(VII) Meanings of the symbols used in the question paper are the same as followed in the course.
(VIII) Questions are self-explanatory. No query will be entertained by the invigilators.

1. Use ∫residue theorem to evaluate the following integrals:

∫ ∞
2π
cos(3θ) x sin(πx)
(a) dθ, (b) dx. [3+4]
0 5 − 4 cos(θ) −∞ x 2 + 2x + 5

Solution:
(a) Putting z = eiθ we have
∫ 2π ∫
cos 3θ 1
dθ = − f (z) dz,
0 5 − 4 cos θ 2i |z|=1

z6 + 1 21 65
where f (z) = . Now Res(f, 0) = and Res(f, 12 ) = − .
z (2z − 1)(z − 2)
3 8 24
∫ ( )
1 21 65 π
So f (z) dz = − · 2πi − = .
|z|=1 2i 8 24 12
∫ iπz
ze
(b) Consider 2
dz, where ΓR = [−R, R] ∪ γR with γR = {Reiθ : θ ∈ [0, π]} . So
ΓR z + 2z + 5
∫ R ∫ ∫ ( )
xeiπx zeiπz zeiπz zeiπz
2
dx+ 2
dz = 2
dz = 2πiRes , −1 + 2i
−R x + 2x + 5 γR z + 2z + 5 ΓR z + 2z + 5 z 2 + 2z + 5

Taking R → ∞ and using Jordan’s lemma we get

∫ ∞ ∫ ∞
x cos πx x sin πx π
2
dx + 2
dx = e−2π − iπe−2π
−∞ x + 2x + 5 −∞ x + 2x + 5 2
∫ ∞
x sin πx
Equating the imaginary parts we get 2
dx = −πe−2π .
−∞ x + 2x + 5
iz − 1
2. Find the image of the lower half plane Im(z)< 0 under the map w = . [3]
i−z

1
 iz − 1
Solution: Let w(z) = . Then w(0) = i, w(−i) = 0 and w(−1) = −1. Let z = x + iy.
i−z
Then
i(i + z) x2 + (y + 1)2
|w(z)| = = 2 < 1,
i−z x + (y − 1)2
i+z
for y < 0. So the image of the lower half plane Im(z)< 0 under the map w = , is the open
i−z
unit disc.

3. Find a first-order linear partial differential equation (PDE) whose characteristic curves are rep-
resented by a one-parameter family of circles x2 + y 2 = R2 . [1]

Solution: The general form of a linear first-oder PDE in 2D is a(x, y)ux + b(x, y)uy = c(x, y)u +
d(x, y).
dx dy
The characteristic equation of this PDE is given by, = .
a b
Now, the given characteristic curves are: x2 + y 2 = R2 .
dx dy
Differentiating we get, 2xdx + 2ydy = 0 ⇒ = , the characteristic equation.
y −x
Comparing these two characteristic equations we have, a = y and b = −x.
So, the PDE whose characteris curves are circle is, yux − xuy = c(x, y)u + d(x, y).
( )
4. Obtain the surface that is orthogonal to the one-parameter family of surfaces u = cxy x2 + y 2
(c is a non-zero parameter) and passes through the hyperbola x2 − y 2 = a2 , u = b, (a, b > 0). [4]

Solution: Any surface passing through the integral curves of fx p + fy q = fu is the surface that
will be orthogonal to f (x, y, u) = c.
( )
xy x2 + y 2 1
For the given surface, f (x, y, u) = = .
( ) ( )u c ( )
y 3x2 + y 2 x x2 + 3y 2 −xy x2 + y 2
So, fx = , fy = and fu = .
u u u2
dx dy du
The auxiliary equations of fx p + fy q = fu is = =
fx fy fu
udx udy u2 du
implies, = =
y (3x2 + y 2 ) x (x2 + 3y 2 ) −xy (x2 + y 2 )
dx dy du
implies, = =
yu (3x2 + y 2 ) xu (x2 + 3y 2 ) −xy (x2 + y 2 )

For the first surface:

xdx + ydy −4udu

2 2
=
4xyu (x + y ) 4xyu (x2 + y 2 )
implies x2 + y 2 + 4u2 = C1 .
For the second surface:

xdx + ydy xdx − ydy

=
2 2
4xyu (x + y ) 2xyu (x2 − y 2 )
( )2
x2 − y 2
implies = C2 .
(x2 + y 2 )
The general solution of fx p + fy q = fu is F (ϕ, ψ) = 0, where
( )2
2 2 2 x2 − y 2
ϕ(x, y, u) = x + y + 4u and ψ(x, y, u) = .
(x2 + y 2 )

2
 The integral surface passes through the hyperbola: x2 − y 2 = a2 , u = b, (a, b > 0).
a4 a4
From the second surface we get, = C2 ⇒ x2
+ y 2
= .
(x2 + y 2 ) C2
a4 ( )
From the first surface, + 4b2 = C1 ⇒ C1 − 4b2 C2 = a4 .
C2
So, the desired surface is
( )( 2 )2 ( )
x2 + y 2 + 4u2 − 4b2 x − y 2 = a4 x 2 + y 2 .

x2 y2 u2
5. Find a second-order PDE arising from the ellipsoid + + = 1. Here, x and y are the
a2 b2 c2
independent variables, u = u(x, y) and a, b, c are arbitrary non-zero constants. [2]

x2 y 2 u2
Solution: The given surface is + 2 + 2 = 1.
a2 b c
2x 2uux
Differentiating the given equation with respect to x one gets, 2 + 2 = 0.
a c

Further, differentiating the above equation with respect to y we have,

2ux uy 2uuxy
2
+ = 0.
c c2
So, the desired PDE is uuxy + ux uy = 0.

Other two solutions are:

xuuxx + xu2x − uux = 0. (differentiang two times w.r.t. x)

yuuyy + yu2y − uuy = 0. (differentiang two times w.r.t. y)

6. Write the general form of a second-order linear homogeneous PDE in two dimensions involving
one dependent variable only. Considering the PDE as a hyperbolic one, reduce it to a normal
form which will be free from mixed derivatives. [1+3]

Solution: The general form of a second-order linear homogeneous PDE in 2D is

auxx + buxy + cuyy + dux + euy + f u = 0,

coefficients are functions of x and y only.

Introducing new variables ξ and η by substitution, ξ = ξ(x, y), η = η(x, y) [such that the
Jacobian ξx ηy − ξy ηx ̸= 0], the transformed equation can be written as,

Auξξ + Buξη + Cuηη + Duξ + Euη + F u = 0,

where,
A = aξx2 + bξx ξy + cξy2 ,
B = 2aξx ηx + b (ξx ηy + ξy ηx ) + 2cξy ηy ,
C = aηx2 + bηx ηy + cηy2 ,
D = aξxx + bξxy + cξyy + dξx + eξy ,
E = aηxx + bηxy + cηyy + dηx + eξy ,
F = f.

3
 The equation is of hyperbolic type, and let ξ =constant and η =constant be the two families of
charcteristics. This implies that A = 0 and C = 0.
So the cannonical form of the equation will be

uξη + Duξ + Euη + F u = 0.

By a further linear transformation, α = ξ + η and β = ξ − η, one can get the final form of the
equation, which is free from mixed derivatives, as,

uαα − uββ + D̂uα + Êuβ + F̂ u = 0,

where, D̂ = D + E, Ê = D − E, F̂ = F.

7. Find the general solution of the PDE y 2 uxx − 2yuxy + uyy = ux + 6y by reducing to its canonical
form. [3]

Solution: Here, b2 − 4ac = 4y 2 − 4y 2 = 0 ⇒ Parabolic.

dy −2y 1 y2
The characteristic is = = − ⇒ − + c = x.
dx 2y 2 y 2
y2
Let ξ(x, y) = + x. The other non-parallel function can be chosen as η(x, y) = y.
2
The Jacobian ξx ηy − ξy ηx = 1 ̸= 0.
Substituting these functions into the given PDE, it becomes, uηη = 6y.
The cannonical form is uηη = 6η.
Solving this we get, u = η 3 + ηf (ξ) + g(ξ).
So, the general solution is
( 2 ) ( 2 )
3 y y
u(x, y) = y + yf +x +g +x .
2 2

Note: In questions 8, 9, 10 and 12 on next page, writing/finding the IBVP/BVP

and the corresponding solution for the general problem will not carry any marks.
Solutions to the specific problem as asked will only carry marks.

8. Consider the following initial boundary value problem for u(x, t) governed by one-dimensional
wave equation for a flexible string of unit length:

utt = uxx , 0 < x < 1, t > 0;

u(0, t) = 0, u(1, t) = 0, t ≥ 0;
u(x, 0) = sin(πx), ut (x, 0) = 0, 0 < x < 1.

Find the displacement of the string at location x = 1/2 and time t = 2. [3]
Solution: By using the given boundary conditions and the second initial condition,
∞
∑
u(x, t) = An sin(nπx) cos(nπt),
n=1
∫ 1
with An = 2 sin(πx) sin(nπx)dx. Here λn = nπ. Using orthogonality of sine function over
0
[0, 1], An = 0 for all n ̸= 1. Therefore, the only non-zero coefficient is A1 = 1. The solution,
therefore, is
u(x, t) = sin(πx) cos(πt).
Therefore, u(0.5, 2) = 1.

4
 9. Consider solving the one-dimensional heat conduction equation ut = uxx for a thin metal rod
of length π with an initial temperature distribution u(x, 0) = sin2 x, 0 < x < π. Further,
homogeneous Neumann conditions are prescribed at the ends x = 0, π for t ≥ 0. Obtain the
solution of the corresponding IBVP. [4]
Solution: The IBVP is

ut = uxx , 0 < x < π, t > 0,

ux (0, t) = 0 = ux (π, t), t ≥ 0,
u(x, 0) = sin2 x.

By looking at the boundary conditions, the solution will be

A0 ∑
π
An e−n t cos nx.
2
u(x, t) = +
2
n=1

where ∫ π
2
An = sin2 x cos nx dx.
π 0
Evaluating the integral to find A0 and An , we get A0 = 1 and A2 = −1/2. All other An = 0 for
n ̸= 0, 2.
Hence the solution of the IBVP is
1 1 −4t
u(x, t) = − e cos 2x.
2 2

10. By considering the one-dimensional wave equation for finite spatial domain along with suitable
boundary conditions and initial conditions, discuss the basic idea behind Duhamel’s principle.
[2]
Solution: Suppose the one-dimensional wave equation for string problem is given as

utt = uxx + F (x, t), 0 < x < L, t > 0;

u(0, t) = 0, u(L, t) = 0, t ≥ 0;
u(x, 0) = 0, ut (x, 0) = 0, 0 < x < L.

According to Duhamel’s principle, the solution can be found as

∫ t
u(x, t) = v(x, τ − t, t)dτ,
0

in which v(x, t, s) is the solution of the IBVP

vtt = vxx , 0 < x < L, t > 0;

v(0, s, t) = 0, v(L, s, t) = 0, t ≥ 0;
v(x, s, 0) = 0, vt (x, s, 0) = F (x, s), 0 < x < L.

In other words, the source term in the problem for u has been shifted to the second initial
condition in the problem for v.

11. Suppose it is required to solve the steady-state heat conduction equation (without source) in
a rectangular plate 0 ≤ x ≤ a, 0 ≤ y ≤ b subject to two homogeneous Dirichlet boundary
conditions along x = 0, x = a and two non-homogeneous Dirichlet conditions along y = 0, y = b.
Write the resulting boundary value problem clearly. Show how you would restructure the problem
so that the method of separation of variables can be appropriately used. [1+2]

5
 (You need not solve.)
Solution: The BVP is:

uxx + uyy = 0, 0 < x < a, 0 < y < b,

u(0, y) = 0, u(a, y) = 0, u(x, 0) = f1 (x), u(x, b) = f2 (x).

(The last two conditions can be taken as Neumann conditions also.)

Split u(x, y) = v(x, y) + w(x, y) to get the following BVPs:

vxx + vyy = 0, 0 < x < a, 0 < y < b,

v(0, y) = 0, v(a, y) = 0, v(x, 0) = f1 (x), v(x, b) = 0;
and
wxx + wyy = 0, 0 < x < a, 0 < y < b,
w(0, y) = 0, w(a, y) = 0, w(x, 0) = 0, w(x, b) = f2 (x).

12. Consider a circular planar disk of unit radius. Solve the corresponding BVP outside the disk
with the given boundary condition: [3]
1 1
urr + ur + 2 uθθ = 0, r > 1, −π < θ < π,
r r
u(1, θ) = θ, −π < θ < π.

Solution: In general, the solution can be written as

∞
A0 ∑ [ ]
u(r, θ) = + An sin(nθ) + Bn cos(nθ) [Cn rn + Dn r−n ].
2
n=1

with
∫ π
1
An = u(a, θ) sin(nθ)dθ,
an π −π
∫ π
1
Bn = n u(a, θ) cos(nθ)dθ.
a π −π
Here a = 1 and the solution is to be obtained only for r > 1 with u(a, θ) = θ. Therefore, the
solution reduces to
∞
∑
u(r, θ) = An sin(nθ)r−n .
n=1
with
∫
1 π
An = θ sin(nθ)dθ
π −π
2
= (−π cos nπ)
nπ
2(−1)n+1
= .
n
Thus, the solution is
∞
∑ ( n+1
u(r, θ) = 2 1) n sin(nθ)r−n .
−
n=1

13. Obtain the Fourier series expansion of the function f given by [3]

π , −π < x ≤ 0,
 1 + 2x
f (x) =

1 − 2x
π , 0 < x < π.

6
 Solution: Clearly f is even and hence Bn = 0. For an , we use the definition to have
4 [ ]
A0 = 0 & An = 2 2 1 − (−1)n , n = 1, 2, 3, . . . .
n π
That is, { 8
n2 π 2
, n odd,
An =
0, n even.
Therefore, the Fourier series for f is given by

8( )
∞
8 ∑ cos((2n − 1))x 1 1
= cos x + cos 3x + cos 5x + · · · .
π2 (2n − 1)2 π2 32 52
n=1

14. (a) By considering an appropriate Fourier transform ∫ ∞ of the function f (t) = e−βt , t > 0, where
σ sin(σt)
β > 0 is a constant, evaluate the improper integral dσ. [3]
0 β 2 + σ2
Solution: Consider an odd extension in (−∞, ∞) so that
∫ ∞
1
F{fo (t)} = √ e−iστ fo (τ )dτ
2π −∞
[∫ 0 ∫ ∞ ]
1 aτ −iστ −aτ −iστ
= √ (−e )e dτ + (e )e dτ
2π −∞ 0
1 (−2iσ)
= √ = g(σ).
2π a2 + σ 2
Inverting and using oddness and evenness of the functions
∫ ∞
1
fo (t) = √ g(σ)eiσt dσ
2π −∞
∫
1 ∞ (−iσ) iσt
= e dσ
π −∞ a2 + σ 2
∫
2 ∞ σ sin σt
= dσ
π 0 a2 + σ 2

Now,
∫
−at 2 ∞ σ sin σt
e = dσ
π 0 a2 + σ 2
∫ ∞
σ sin σt π
⇒ 2 2
dσ = e−at .
0 a +σ 2
(b) Consider the heat conduction in an infinite thin metal rod having thermal diffusivity α
with an initial temperature distribution ϕ(x). By using Fourier transform, find the temperature
distribution u(x, t) at any point of the rod at any subsequent time t > 0. How would you
approach this problem if the rod is considered to be semi-infinite? [2+1]
Solution: The IVP is
∂u ∂2u
= α 2 , −∞ < x < ∞, t > 0,
∂t ∂x
u(x, 0) = ϕ(x).
d
Applying Fourier transform in x variable, we get, (F{u(x, t)} = ū(σ, t)) dt ū(σ, t) = −ασ 2 ū(σ, t)
and F{u(x, 0)} = ū(σ, 0) = f¯(σ). On solving the ODE in t variable, we get ū(σ, t) = Ce−ασ t .
2

So
ū(σ, t) = f¯(σ)e−ασ t .
2

7
 Taking inverse Fourier transform, we get
∫ ∞
1
f¯(σ)e−ασ t eiσx dσ.
2
u(x, t) =
2π 0
If the rod is of semi-infinite length, we would have a boundary condition at x = 0. Depending
on if the boundary condition is Dirichlet or Neumann, we will use sine or cosine transform,
respectively.
15. (a) Find the Laplace transform of f (t) = 2H(sin πt) − 1 where H is Heaviside unit step function.
[2]
Solution: From definition of unit step function
{
1, sin πt > 0,
H(sin πt) =
0, sin πt < 0.
This will give the given function as +1 between 0 and 1, 2 and 3, and so on whereas it will be
−1 between 1 and 2, 3 and 4, and so on. That is
∫ 1 ∫ 2 ∫ 3 ∫ 4
−st −st −st
L{2H(sin πt) − 1} = e dt − e dt + e dt − e−st dt + · · ·
0 1 2 3
e−st −st −st e−st
1 2 3 4
e e
= − + − + ···
−s 0 −s 1 −s 2 −s 3
1[ ]
= (1 − e−s ) + (e−2s − e−s ) + (e−2s − e−3s ) + (e−4s − e−3s ) + · · ·
s
1
= [1 − 2e−s (1 − e−s + e−2s − e−3s + · · · )]
s[ ]
1 −s 1
= 1 − 2e
s 1 + e−s
1 1 − e−s
=
s 1 + e−s
1 s
= tanh
s 2
d2 x dx
(b) By using Laplace transform, solve the second-order ODE 2 + 2p + qx = f (t), t > 0
dt dt
dx
subject to the initial conditions x(0) = a, (0) = b for the case q − p2 > 0, where p, q, a and b
dt
are constants, and f is piecewise continuous and of exponential order. [3]
Solution: Notation: L{x} = X(s), L{f (t)} = f (s). Taking Laplace transform
dx
s2 X(s) − sx(0) −
(0) + 2p(sX(s) − x(0)) + qX(s) = f (s)
dt
⇒ (s2 + 2ps + q)X(s) = as + 2pa + b + f (s).
Therefore
as + 2pa + b + f (s)
X(s) =
(s2 + 2ps + q)
a(s + p) + ap + b + f (s)
=
(s + p)2 + q − p2
a(s + p) ap + b f (s) √
= + + , n = q − p2 .
(s + p)2 + n2 (s + p)2 + n2 (s + p)2 + n2
Taking inverse
∫ t
−pt ap + b −pt 1
x(t) = ae cos nt + e sin nt + f (t − τ )e−pτ sin nτ dτ.
n n 0