MATH2065: INTRO TO PDEs

Semester 2, 2006
Tutorial Solutions (Week 6)

1. We have the equation

φ00 (x) + λ φ(x) = 0 ,
which is a second-order constant-coefficient, homogeneous ODE. While this is exactly the
equation we have looked at in class, the boundary conditions are different here, and therefore
it is necessary to go through the whole argument to figure out possible values of λ leading
to non-trivial solutions. The basic approach to follow is to attempt a solution of the form
φ(x) = eµx (note: we cannot use a λ here, since there is already one in the problem). This
gives us the characteristic equation
µ2 + λ = 0 .
Clearly, this will have different types of solutions for µ depending on whether λ is negative,
zero, or positive, and so three different cases will need to be addressed.
Case (i): λ < 0
Set λ = −q 2 , where q > 0. (This is just a convenient way of avoiding writing square-roots.)
Thus, µ2 − q 2 = 0, and µ = ±q, leading to
φ(x) = A cosh qx + B sinh qx .
(Using this form rather than the two exponentials eqx and e−qx turns out to be more con-
venient here.) Since φ(0) = 0, A = 0. Since φ0 (L) = 0,
B q cosh qL = 0 .
Now, the cosh function is never zero, and hence it is not possible to find non-trivial solutions.
That is, the only possibility here is B = 0, which means that φ(x) is identically the zero
function. Therefore, there are no negative eigenvalues.
Case (ii): λ = 0
Then the ODE is simply φ00 (x) = 0, which can be integrated twice to give
φ(x) = Ax + B .
Since φ(0) = 0, B = 0. Since φ0 (L) = 0, A = 0. Thus, this yields only the trivial solution,
and therefore λ = 0 is not an eigenvalue.
Case (iii): λ > 0
Set λ = p2 , where p > 0. Thus, µ2 + p2 = 0, and so µ = ±i p. This gives the solutions
cos px and sin px, and hence the general solution
φ(x) = A cos px + B sin px .
Now, we need φ(0) = 0, and therefore A = 0. We then need φ0 (L) = 0, which means that


B p cos px =0 ⇒ cos pL = 0 .
x=L

Now, the cosine function is zero at π/2, and at any points ±π, ±2π, ±3π, etc, away from
this value. That is, we need
π
p L = + n π n = 0, ±1, ±2, ±3, · · ·
2
 However, we have chosen p > 0, and so we must limit ourselves to
π
p L = + n π n = 0, 1, 2, 3, · · ·
2
This means that  2
2 π/2 + nπ
λn = p = n = 0, 1, 2, 3, · · ·
L
are the eigenvalues. The corresponding eigenfunctions are, since A = 0 but B 6= 0,
 
π/2 + nπ
φn (x) = sin x n = 0, 1, 2, · · ·
L
2. We have already solved the heat equation in this form, for a general initial condition
u(x, 0) = f (x). The solution is given by
∞
X nπx −k(nπ/L)2 t
u(x, t) = Bn sin e ,
n=1
L
where
2 L
Z
nπx
Bn = f (x) sin dx (1)
L 0 L
gives the coefficients, which are related to the expansion of f (x) in terms of sine functions
in the form ∞
X nπx
f (x) = Bn sin .
n=1
L
For this problem, we are given that
πx 3πx
f (x) = 3 sin − sin
L L
Note that we do not need to invoke the formula (1), since f is already given in terms of a
expansion in sines. In fact, we can simply read off
B1 = 3 and B3 = −1 ,
with all other Bn s being zero. This is since, for example, B3 is the coefficient of sin 3πx
L
.
Therefore, in the infinite summation representing the solution u(x, t), all but two terms
are zero. Retaining only these two terms (corresponding to n = 1 and n = 3), we get the
solution
kπ 2 t 9kπ 2 t
   
πx 3πx
u(x, t) = 3 sin exp − 2 − sin exp − 2 .
L L L L

3. Please refer the solution to the previous problem. In this case, however, we are not able to
simply read off the values of the Bn s, since f (x) is not given as an expansion over the sines.
We therefore have to employ the formula (1):
Z L/2 Z L !
2 nπx nπx
Bn = 1 sin dx + 2 sin dx
L 0 L L/2 L
L/2 L !
2 cos nπx
L 
 cos nπx
L 

= − nπ  − 2 nπ 
L L x=0 L x=L/2
2 L  nπ nπ 
= − cos − cos 0 + 2 cos nπ − 2 cos
L nπ 2 2
2 h nπ i
= 1 − 2(−1)n + cos .
nπ 2
 There is unfortunately no nice expression for cos nπ
2
. The solution is therefore
∞
2 X 1 − 2(−1)n + cos nπ
2 nπx −k(nπ/L)2 t
u(x, t) = sin e .
π n=1 n L

4. (Very brief solutions will be given; the basic solution technique is essentially the same as in
the previous problems, with the insulated boundary conditions requiring a cosine expansion
rather than a sine one)
The solution is given by (2.4.19), where the coefficients satisfy (2.4.21), and hence (2.4.23-
24). The values of these coefficients is given below.
L
1 L 1 2 L nπx 2 L nπx  2
sin nπ
R R
(a) A0 = L L/2 1 dx = 2 , An = L L/2 cos L dx = L nπ sin L  = − nπ 2
for n 6= 0.
L/2

(b) By inspection A0 = 6, A3 = 4, and all other As are zero.

(c) By inspection, A8 = −3, and all other As are zero.

5. Assume a separation of variables solution of the form u(x, t) = X(x) T (t), which upon
substitution into the PDE leads to
X 00 (x) 1 T 00 (t)
= 2 = −λ
X(x) c T (t)

for a separation constant λ. The fact that λ must be a constant is obtained by realising
that it cannot be a function of t (since it equals the left-hand term above), and moreover
it cannot be a function of x (since it equals the middle term above). Now, examining the
homogeneous boundary and initial conditions, we see that

u(0, t) = 0 ⇒ X(0) T (t) = 0 for all t ⇒ X(0) = 0

u(L, t) = 0 ⇒ X(L) T (t) = 0 for all t ⇒ X(L) = 0
u(x, 0) = 0 ⇒ X(x) T (0) = 0 for all t ⇒ T (0) = 0.

This leads to the ODEs

X 00 (x) = −λ X(x) ; X(0) = X(L) = 0 (2)

T 00 (t) = −c2 λ T (t) ; T (0) = 0.

The ODE for X(x) is the standard boundary value problem, which can be nontrivially
 nπ 2
solved only if λ = λn = for positive integers n, leading to the solutions
L
 nπx 
Xn (x) = sin .
L
The T equation then becomes
 nπc 2
00
T (t) + T (t) = 0
L
from which the basic form
nπct nπct
Tn (t) = A cos + B sin
L L
results. Since T (0) = 0 is required, A must be zero, and the cosine term can be discarded.
Now, since the solution u(x, t) is the product of X and T , for each n the function
 nπx   
nπct
un (x, t) = sin sin
L L

is a solution. The general solution is the superposition of all these:

∞  nπx   
X nπct
u(x, t) = Bn sin sin .
n=1
L L

The inhomogeneous initial condition is all that remains to be satisfied. Taking the t deriva-
tive of the above,
∞  
∂u X  nπc   nπx  nπct
= Bn sin cos .
∂t n=1
L L L

This, when evaluated at t = 0, must be 3c sin 4πx

L
, and so
∞
4πx X nπc  nπx 
3c sin = Bn sin .
L n=1
L L

It should be clear that the condition 3c = B4 4πc

L
for B4 is all that is necessary (all other
0 3L
B s are zero). Thus, B4 = 4π , and Bn = 0 for all other n. Thus, only one term (that
corresponding to n = 4) survives from the superposed solution, and hence

3L 4πx 4πct
u(x, t) = sin sin .
4π L L