Final Exam Answers and Hints

(MATH4052 Partial Differential Equations)

Xiaoyu Wei∗
December 1, 2016

Problem 1. What is the type of the equation

uxx − 4uxy + 4uyy = 0?

Show by direct substitution that u(x, y) = f (y + 2x) + xg(y + 2x) is a solution

for arbitrary functions f and g.
Solution. Given u(x, y) = f (y + 2x) + xg(y + 2x),
(
ux = 2f 0 (y + 2x) + g(y + 2x) + 2xg 0 (y + 2x),
uy = f 0 (y + 2x) + xg 0 (y + 2x),

and
00 0 00

 uxx = 4f (y + 2x) + 4g (y + 2x) + 4xg (y + 2x),

uxy = 2f 00 (y + 2x) + g 0 (y + 2x) + 2xg 00 (y + 2x),
uyy = f 00 (y + 2x) + xg 00 (y + 2x).



Plug these into the equation,

uxx − 4uxy + 4uyy = (4 − 8 + 4)f 00 (y + 2x) + (4 − 4)g 0 (y + 2x) + (4 − 8 + 4)xg 00 (y + 2x)

= 0.

That is, u(x, y) = f (y + 2x) + xg(y + 2x) is a solution for arbitrary functions
f and g.
With two arbitrary functions in its expression, u(x, y) = f (y+2x)+xg(y+2x)
should be the general solution to the given equation.
As seen from the expression of u(x, y), the equation has exactly one family
of real characteristics
y + 2x = Constant.
Therefore, the equation is of parabolic type.

∗ For any issues, email me at [email protected].

1
Problem 2. Use the method of separation of variables to solve the following
equation. Explain all your steps.

ut − 2ktuxx = 0, 0 < x < π, t ≥ 0

u(0, t) = u(π, t) = 0, t≥0
u(x, 0) = 2 sin 2x − 5 sin 3x 0≤x≤π

Solution. First of all, the functions of the form v(x, t) = X(x)T (t) s.t.

vt − 2ktvxx = 0, 0 < x < π, t ≥ 0

(1)
v(0, t) = v(π, t) = 0, t≥0

are sought for.

Plug the ansatz into the equation, yielding

X(x)T 0 (t) − 2ktX 00 (x)T (t) = 0

X 00 (x) T 0 (t)
= = −λ,
X(x) 2ktT (t)
where λ is a constant independent of x, t. Then the PDE is reduced to a system
of ODEs

X 00 (x) = −λX(x),
T 0 (t) = −2kλtT (t),

whose general solution is

√ √
X(x) = A sin λx + B cos λx,
−kλt2
T (t) = Ce ,

where A, B, C are arbitrary constants s.t. A2 + B 2 6= 0 and C 6= 0.

To satisfy the boundary conditions,

0 = v(0, t) = X(0)T (t) = BC,

yielding B = 0, and
2 √
0 = v(π, t) = X(π)T (t) = ACe−kλt sin λπ,
√
yielding sin λπ = 0; Therefore, the family of functions within our ansatz
solving (1) is

vn (x, t) = Dn Xn (x)Tn (t), n = 1, 2, 3, . . .

where Dn ’s are arbitrary constants, and

Xn (x) = sin nx,

2
kt2
Tn (t) = e−n .

2
 Next, using the family of functions obtained above to span the solution space,
solution to the given initial-boundary value problem with the form
∞
2
kt2
X
u(x, t) = Un e−n sin nx
n=1

is sought for.
Applying the initial conditions,
∞
X
2 sin 2x − 5 sin 3x = u(x, 0) = Un sin nx,
n=1

yielding 
 2, n = 2,

Un = −5, n = 3,

0, otherwise.


In conclusion,
2 2
u(x, t) = 2e−4kt sin 2x − 5e−9kt sin 3x.

3
Problem 3. Find the harmonic function u in the semidisk {r < 1, 0 < θ < π}
with u vanishing on the diameter ( θ = 0, π) and

u = π sin θ − sin 2θ on r = 1.

Solution. Denote the semidisk region as D. In polar coordinates, the harmonic

function u is determined by the following boundary value problem (refer to the
formula sheet for ∆ in polar coordinates):
1 1
urr + ur + 2 uθθ = 0 in D,
r r
u(r, 0) = u(r, π) = 0, (2)
u(1, θ) = π sin θ − sin 2θ.
Using the method of separation of variables, firstly the functions of the form
v(r, θ) = R(r)Θ(θ) ,s.t.
1 1
vrr + vr + 2 vθθ = 0 in D,
r r
v(r, 0) = v(r, π) = 0, (3)
v(0, θ) = 0,
are sought for.
Plug in the ansatz v(r, θ) = R(r)Θ(θ) into the equation, yielding
1 1
R00 Θ + R0 Θ + 2 RΘ00 = 0
r r
(r2 R00 + rR0 )Θ = −RΘ00
Θ00 r2 R00 + rR0
=− = −λ,
Θ R
where λ is a constant independent of θ, r. Then the PDE is reduced to a system
of ODEs

Θ00 = −λΘ,
r2 R00 + rR0 − λR = 0,

via which the general solution for Θ(θ) can be solved as

√ √
Θ(θ) = A sin λθ + B cos λθ.

Using the boundary conditions in (3), nontrivial solutions to (3) exists iff
√
B = 0, and sin λπ = 0.

Consequently, solutions to (3) can be parametrized with natural numbers

vn (θ, r) = Θn (θ)Rn (r), where

Θn (θ) = An sin nθ, n = 1, 2, 3, . . . ,

4
and Rn (r) solves the ODE, which is of the form of a Cauthy-Euler equation:

r2 Rn00 + rRn0 − n2 Rn = 0,

and is subject to a one-sided boundary condition Rn (0) = 0.

Using the ansatz for Cauthy-Euler equations Rn (r) = rm , it leads to an
algebraic equation

(m − 1)m + m − n2 = 0
m2 = n2 ,

yielding
Rn (r) = Cn rn + Dn r−n ,
where Cn , Dn are arbitrary constants. And the boundary condition Rn (0) = 0
implies that Dn = 0 for all n.
Now the ansatz solutions to (3) above are used to span the solution space of
(2), yielding
∞
X
u(r, θ) = Un rn sin nθ.
n=1

Applying the boundary condition at r = 1,

∞
X
π sin θ − sin 2θ = u(1, θ) = Un sin nθ,
n=1

yields 
 π, n = 1,

Un = −1, n = 2,

0, otherwise.


Therefore,
u(r, θ) = πr sin θ − r2 sin 2θ.

5
Problem 4. Solve the wave equation
utt + uxt − 20uxx = 0, −∞ < x < +∞
with the initial conditions
u(x, 0) = φ(x), ut (x, 0) = ψ(x).
Solution. Write the given equation in operator form
Lu := utt + uxt − 20uxx = 0,
with L = ∂tt + ∂xt − 20∂xx a second order hyperbolic operator.
Factor L into the composition of two first order operators,
L = (∂t − r1 ∂x )(∂t − r2 ∂x ),
where r1 , r2 are two real roots of a quadratic polynomial
r2 + r − 20 = 0,
which solves r1 = 4, r2 = −5.
Now let v(x, t) = ut +5ux , then v(x, t) solves the following (first order) initial
value problem: (
vt − 4vx = 0,
(4)
v(x, 0) = ψ(x) + 5φ0 (x).
This being a linear advection equation, the solution is simply
v(x, t) = v(x + 4t, 0) = ψ(x + 4t) + 5φ0 (x + 4t).
Then u(x, t) can be solved from another (first order) initial value problem:
(
ut + 5ux = ψ(x + 4t) + 5φ0 (x + 4t),
(5)
u(x, 0) = φ(x).
The rate of change of u(x, t) with t as we move along the characteristics
x = 5t + C (C being an arbitrary constant) is
du dx
= ux + ut = ut + 5ux
dt dt
= ψ(x + 4t) + 5φ0 (x + 4t).
Therefore, for any point (X, T ), X ∈ R, T > 0, one can integrate the change
of u(x, t) all the way to t = 0, yielding
Z T
du
u(X, T ) = u(X − 5T, 0) + (5τ + (X − 5T ), τ )dτ
τ =0 dt
Z T
= φ(X − 5T ) + ψ(9τ + X − 5T ) + 5φ0 (9τ + X − 5T )dτ
τ =0
Z T
5
= φ(X − 5T ) + ψ(9τ + X − 5T )dτ + [φ(X + 4T ) − φ(X − 5T )]
τ =0 9
Z T
5 4
= φ(x + 4T ) + φ(x − 5T ) + ψ(9τ + x − 5T )dτ
9 9 τ =0

6
In conclusion, the solution is
Z t
5 4
u(x, t) = φ(x + 4t) + φ(x − 5t, 0) + ψ(9τ + x − 5t)dτ.
9 9 τ =0

7
Problem 5. Prove the uniqueness of the Robin problem
∆u = f in D
with boundary condition
∂u
+ au = h on ∂D
∂n
where D is any domain in three dimensions and where a is a positive constant.
Proof. Suppose that there are two solutions to the given Robin problem, say,
u1 (x, t) and u2 (x, t). Let v(x, t) = u1 (x, t) − u2 (x, t), then v(x, t) solves the
following boundary value problem:

 ∆v = 0 in D,
(6)
 ∂v + av = 0 on ∂D.
∂n
Multiply both sides of the eqution in (6) with v, and integrate over the whole
domain, yielding
Z Z
v∆vdx = 0dx = 0.
D D

With the help of Green’s identity, the identity above can be rewritten as
Z Z
∂v
v dS − |∇v|2 dx = 0.
∂D ∂n D

From the boundary condition in (6),

∂v
= −av on ∂D,
∂n
thus
Z Z
v(−av)dS − |∇v|2 dx = 0.
∂D D

Since given a > 0,

Z Z
2
0≤ |∇v| dx = − av 2 dS ≤ 0,
D ∂D

all terms presented should be zero

Z
|∇v|dx = 0 =⇒ v = constant in D,
ZD
av 2 dS = 0 =⇒ v = 0 on ∂D.
∂D

Therefore, continuity of v at the boundary implies that v(x, t) = 0 every-

where. This proves the uniqueness of the original Robin problem.