Solutions to Problems for the 1-D Wave Equation

18.303 Linear Partial Di¤erential Equations

Matthew J. Hancock

Fall 2004

1 Problem 1
(i) Generalize the derivation of the wave equation where the string is subject to a damping
force b@u=@t per unit length to obtain

@2u 2
2@ u @u
= c 2k (1)
@t2 @x2 @t
All variables will be left in dimensional form in this problem to make things a little di¤erent.
How is the constant k related to b? What are the dimensions of b and k? The constant 2 is
included for later convenience.
Solution: The derivation follows that in Section 1 of WaveEqnI.pdf. Consider an element
of the string between x and x + x. Let T (x; t) be tension and (x; t) be the angle wrt the
horizontal x-axis. Note that
@u
tan (x; t) = slope of tangent at (x; t) in ux-plane = (x; t) : (2)
@x
Newton’s Second Law (F = ma) states that

@2u
F =( x) (3)
@t2
where is the linear density of the string (M L 1 ) and x is the length of the segment. The
force F comes from the tension in the spring and also the damping force (we ignore any
external forces such as gravity). The damping force acts in the opposite direction to the
motion. Recall our assumptions on the string. We assumed the displacements of the string
are su¢ ciently small so that each point on the string moves vertically. Thus the damping

1
force also acts vertically in the opposite direction to the motion. Balancing the forces in the
horizontal direction gives

T (x + x; t) cos (x + x; t) = T (x; t) cos (x; t) = = const (4)

where is the constant horizontal tension. Balancing the forces in the vertical direction
yields
@u
F = T (x + x; t) sin (x + x; t) T (x; t) sin (x; t) b x
@t
@u
= T (x + x; t) cos (x + x; t) tan (x + x; t) T (x; t) cos (x; t) tan (x; t) b x (5)
@t
Substituting (4) and (2) into (5) yields
@u
F = (tan (x + x; t) tan (x; t)) b x
@t
@u @u @u
= (x + x; t) (x; t) b x: (6)
@x @x @t
Substituting F from (3) into Eq. (6) and dividing by x gives
@u @u
@2u @x
(x + x; t) @x
(x; t) @u
( ; t) = b
@t2 x @t
for 2 [x; x + x]. Dividing by and letting x ! 0 gives the 1-D Wave Equation

@2u 2
2@ u @u b
2
= c 2
2k ; c2 = > 0; k= : (7)
@t @x @t 2
h i1=2 h i
Force 1 force
Note that c has units [c] = Density = LT of speed, b has units [b] = distance speed
=
M LT 2 M L 1T 1
1 1 1
L2 T 1 = M L T and k has units [k] = [b] = [ ] = = T . Thus k is proportional
ML 1
to a frequency (i.e. has units of 1/time or Hz).
(ii) Use separation of variables to …nd the normal modes of the damped Wave Equation
(1) subject to the BCs
u (0; t) = 0 = u (l; t) (8)
Impose a restriction on the parameters c, l, k which will guarantee that all solutions are
oscillatory in time. You may assume that the eigenvalues and eigenfunctions are
n2 2 n x
n = ; Xn (x) = sin ; n = 1; 2; 3:::
l2 l
Solution: Separating variables as

u (x; t) = X (x) T (t) ;

2
substituting into the PDE (1) and dividing by c2 X (x) T (t) gives

1 T 00 2k T 0 X 00
+ = = (9)
c2 T c2 T X
From the BCs (8),
X (0) = X (l) = 0
since T (t) cannot be zero for all time to obtain a non-trivial solution. The Sturm-Liouville
problem for X (x) is, from (9),

X 00 + X = 0; X (0) = X (l) = 0:

The non-trivial solutions are the eigenvalues and eigenfunctions

n2 2 n x
n = ; Xn (x) = sin : (10)
l2 l
The solutions for T (t) are, from (9),

T 00 + 2kT 0 + c2 T = 0:

From (9), the values of for T (t) are the same as those for X (x). From (10), the only
values of that lead to non-trivial solutions are = n . Thus for each n , we need to solve
for the corresponding T (t) = Tn (t) to …nd the normal mode un (x; t) = Xn (x) Tn (t). To
…nd the solutions for T (t), we substitute T = ert ,

r2 + 2kr + c2 = 0:

Solving the quadratic equation for r gives

p
r= k k2 c2 (11)

The solutions Tn (t) corresponding to the eigenvalues = n are all oscillatory if r has a
complex part, i.e. if k 2 c2 n < 0 for all n, or

c2 n2 2
c2 2
k 2 < min c2 n = min = :
n 1 n 1 l2 l2
Taking the square root of both sides and rearranging gives the criterion that all the Tn (t)
be oscillatory,
c
> 1: (12)
kl

3
 The solution Tn (t) is a linear combination of ert where r is given in (11) with = n,
p p
Tn (t) = c1 e( k+ k2 c2 n )t + c e(
2
k k 2 c2 n )t
p p
Under condition (12), k2 c2 n = i c2 n k 2 , where c2 n k 2 > 0. Thus
p p
kt c2 k2 t 2 2
Tn (t) = e cn1 ei n
+ cn2 e i c n k t
p p
kt
= e (cn1 + cn2 ) cos c2 n k 2 t + i (cn1 cn2 ) sin c2 n k2t
p p
kt
= e n cos c2 n k 2 t + n sin c2 n k 2 t

where we have rewritten the constants of integration as n = cn1 + cn2 , n = i (cn1 cn2 ).
The corresponding normal modes are
p n x p
un (x; t) = Xn (x) Tn (t) = e kt n cos c2 n k 2 t + n sin c2 n k2t sin
l
(13)
where n = n2 2 =l2 . Notice that with damping (k > 0), the normal mode decays with time
and oscillates (c > kl) as it decays.
(iii) Express the frequency fen of the oscillatory part of the n’th normal mode in terms
of the frequency of the undamped mode fn = nc= (2l). What di¤erence does the damping
make?
Solution: The frequency fen of the normal mode un (x; t), given in (13), is
r s s
1 p 1 cn 2 2 2 cn lk
2
lk
2
e
fn = 2
c n k = 2 2
k = 1 = fn 1 (14)
2 2 l2 2l cn cn

As the damping (k > 0) increases, the frequencies of the normal modes decrease.
(iv) Show that the solution of the damped wave equation (1) subject to the BCs (8) and
the initial condition
@u
u (x; 0) = f (x) ; (x; 0) = 0 (15)
@t
is given by
X
1
n x
u (x; t) = e kt
n cos 2 fen t + n sin 2 fen t sin
n=1
l

Express the constants n , n in terms of the Fourier Sine coe¢ cients Bn of f .

Solution: Summing the normal modes gives the solution
X
1 X
1
n x
u (x; t) = un (x; t) = e kt
n cos 2 fen t + n sin 2 fen t sin (16)
n=1 n=1
l

4
where the frequencies fen are given in (14) and n , n are to be determined from the initial
conditions on u (x; t). Imposing the ICs (15) gives

X
1
n x
f (x) = u (x; 0) = n sin (17)
n=1
l
X
1
n x
0 = ut (x; 0) = k n + 2 fen n sin (18)
n=1
l

Eq. (18) implies

k n + 2 fen n =0 (19)
for all n. Multiplying both sides of (17) by sin m x, integrating from x = 0 to l, and using
the orthogonality properties of sin m x, gives
Z
2 l n x
n = f (x) sin dx: (20)
l 0 l

From (19),
k n
n = (21)
2 fen
From (16), (20) and (21), the complete solution is

X
1
k n x
u (x; t) = e kt
n cos 2 fen t + sin 2 fen t sin (22)
n=1 2 fen l

2 Problem 2
Prove that if a vibrating string is damped, i.e. subject to the PDE in Problem 1(i), then the
energy E (t) is monotone decreasing. You may use the formula we derived in lecture,
Z l
E (t) = u2t + c2 u2x dx (23)
2 0

Also, you may assume Homogeneous Type I BCs for the displacement u (x; t).
Solution: The formula derived in lecture is valid for a system with damping, since the
kinetic and potential energies of the string only depend on the displacement u (x; t) and its
derivatives. Di¤erentiating (23) gives
Z l
dE
= ut utt + c2 ux uxt dx
dt 0

5
Replacing utt using the PDE (1) gives
Z l
dE
= c2 ut uxx 2ku2t + c2 ux uxt dx
dt 0
Z l
= c2 (ut ux )x 2ku2t dx
0
Z l
= c2 [ut ux ]lx=0 2k u2t dx (24)
0

Di¤erentiating the BCs in time t gives

d d
u (0; t) = 0; u (l; t) = 0
dt dt
Therefore, (24) becomes
Z l
dE
= 2k u2t dx (25)
dt 0

Thus dE=dt 0 which shows that the energy E (t) is monotone non-increasing.
We want to show little more, namely that E (t) is monotone decreasing for t > 0. We
must simply show that there is no time interval 0 < t1 t t2 in which dE=dt = 0.
Suppose, for the sake of contradiction, there was a time interval 0 < t1 t t2 in which
2
dE=dt = 0. Then (25) implies ut must vanish identically for all 0 x l and for each time
in this interval,
ut (x; t) = 0; t 2 [t1 ; t2 ] ; x 2 [0; l]
Di¤erentiating in time gives

utt (x; t) = 0; t 2 [t1 ; t2 ] ; x 2 [0; l]

Substituting for utt from the Damped Wave Equation (1), we have

uxx (x; t) = 0; t 2 [t1 ; t2 ] ; x 2 [0; l]

Integrating in x and applying the BCs (8) gives

u (x; t) = 0; t 2 [t1 ; t2 ] ; x 2 [0; l]

Substituting the solution (22) from Problem 1 yields

X
1
k n x
e kt
n cos 2 fen t + sin 2 fen t sin ; t 2 [t1 ; t2 ] ; x 2 [0; l]
n=1 2 fen l

6
From the orthogonality of sin (n x=l), this implies that for all n,
k
e kt
n cos 2 fen t + sin 2 fen t = 0; t 2 [t1 ; t2 ] (26)
2 fen
We assume that f (x) 6= 0, for otherwise the solution is trivial u (x; t) = 0. Thus from Eq.
(20), there exists an n1 such that
Z
2 l n1 x
n1 = f (x) sin dx 6= 0:
l 0 l
For this n1 , n1 6= 0 and (26) implies
k
cos 2 fen t + sin 2 fen t = 0
2 fen
Rearranging yields
2 fen
tan 2 fen t = ; t 2 [t1 ; t2 ]
k
This equation makes no sense because the l.h.s. varies with time t 2 [t1 ; t2 ], while the r.h.s. is
constant. This is a contradiction, and hence there can be no time interval where dE=dt = 0,
unless the solution u (x; t) is trivial, i.e. identically zero.
Therefore, for non-trivial solutions u (x; t) (given by functions f (x) that are not identi-
cally zero), the energy is monotonic decreasing, dE=dt < 0 for all time t > 0.

3 Problem 3
(i) Suppose that an “in…nite string”has an initial displacement
8
>
< x + 1; 1 x 0
u (x; 0) = f (x) = x + 1; 0 x 1
>
:
0; jxj > 1

and zero initial velocity ut (x; 0) = 0. Write down the solution of the wave equation

utt = uxx

with ICs u (x; 0) = f (x) and ut (x; 0) = 0 using D’Alembert’s formula. Illustrate the nature
of the solution by sketching the ux-pro…les z = u (x; t) of the string for t = 0; 1=2; 1; 3=2.
Solution: Step 1. Since ut (x; 0) = 0, D’Alembert’s solution is
f (x t) + f (x + t)
u (x; t) = : (27)
2

7
Notice that df =dx is not continuous, but it turns out that the smoothness criterion can be
relaxed. f (x) is piecewise smooth. Lastly, we can rewrite f (x) as
(
1 jxj ; jxj 1
f (x) = (28)
0; jxj > 1

Step 2. Identify the regions. The function f (x) is a triangle with vertices at x =
1; 0; 1. It is zero for jxj > 1. Thus, the regions of interest are found by plotting the four
characteristics x t = 1 (see plot above). The regions are identi…ed in the plot, and are
given mathematically by

R1 = f(x; t) : 1 x t 1 and 1 x+t 1g

R2 = f(x; t) : 1 x t 1 and x + t 1g
R3 = f(x; t) : x t 1 and 1 x+t 1g (29)
R4 = f(x; t) : x t 1 and x + t 1g
R5 = f(x; t) : x + t 1g ;
R6 = f(x; t) : x t 1g

8
The regions determine where x t and x + t are relative to 1, which tells us what part of
the case function f (x) should be used.
Step 3. Consider the solution in each region. In R1 , combining the inequalities gives
jx tj 1 and hence from (28),

f (x t) = 1 jx tj
f (x + t) = 1 jx + tj

thus (27) becomes

f (x t) + f (x + t) 1
u (x; t) = =1 (jx tj + jx + tj)
2 2
In region R2 ,

f (x + t) = 0
f (x t) = 1 jx tj

and hence (27) becomes

f (x t) 1 1
u (x; t) = = jx tj
2 2 2
In region R3 ,

f (x + t) = 1 jx + tj
f (x t) = 0

and hence (27) becomes

f (x + t) 1 1
u (x; t) = = jx + tj
2 2 2
In regions R4 , R5 and R6 , f (x + t) = 0 = f (x t) and hence u = 0. To summarize,
8 1
>
> 1 2
(jx tj + jx + tj) ; (x; t) 2 R1
>
< 1 1
2 2
jx tj ; (x; t) 2 R2
u (x; t) = (30)
>
>
1 1
jx + tj ; (x; t) 2 R3
>
: 2 2
0; (x; t) 2 R4 ; R5 ; R6

Step 4. For each speci…c time t = t0 , write the x-intervals corresponding to the sets Rn
(i.e. the intersection of the set Rn with ft = t0 g, or in the …gure above, where the line t = t0

9
intersects the region Rn ). As a check, we note that at t = 0,

R1 \ ft = 0g = f 1 x 1g = fjxj 1g
R2 \ ft = 0g = fx = 1g
R3 \ ft = 0g = fx = 1g (31)
R4 \ ft = 0g = ?
R5 \ ft = 0g = fx 1g ;
R6 \ ft = 0g = fx 1g

and (30) becomes (

1 jxj ; jxj 1
u (x; 0) = = f (x)
0; jxj > 1
At t = 1=2,

1 1 3 1 3 1 1
R1 \ t = = x and = x x
2 2 2 2 2 2 2
1 1 3 1 1 3
R2 \ t = = x and x = x
2 2 2 2 2 2
1 1 3 1 3 1
R3 \ t = = x and x = x (32)
2 2 2 2 2 2
1 1 1
R4 \ t = = x and x =?
2 2 2
1 3
R5 \ t = = x ;
2 2
1 3
R6 \ t = = x
2 2

and (30) becomes

8 1 1 1 1 1
>
> 1 2
x 2
+ x+ 2
; 2
x 2
>
< 1 1
1 2 2
x 21 ; 1
2
x 3
2
u x; =
2 >
>
1 1
x + 21 ; 3
x 1
>
: 2 2 2 2
3
0; jxj 2
8 1 1 1
>
> 2
; 2
x 2
>
< 3 1 1 3
4 2
x; 2
x 2
=
>
>
3 1
+ 2 x; 3
x 1
>
: 4 2 2
3
0; jxj 2

10
At t = 1,

R1 \ ft = 1g = f0 x 2 and 2 x 0g = fx = 0g
R2 \ ft = 1g = f0 x 2 and x 0g = f0 x 2g
R3 \ ft = 1g = fx 0 and 2 x 0g = f 2 x 0g (33)
R4 \ ft = 1g = fx 0 and x 0g = fx = 0g
R5 \ ft = 1g = fx 2g ;
R6 \ ft = 1g = fx 2g

and (30) becomes 8

1 1
>
< 2 2
jx 1j ; 0 x 2
1 1
u (x; 1) = jx + 1j ; 2 x 0
>
:
2 2
0; jxj 2
At t = 3=2,
3 1 5 1 5
R1 \ t = = x and =? x
2 2 2 2 2
3 1 5 1 1 5
R2 \ t = = x and x = x
2 2 2 2 2 2
3 1 5 1 5 1
R3 \ t = = x and x = x (34)
2 2 2 2 2 2
3 1 1 1 1
R4 \ t = = x and x = x
2 2 2 2 2
3 5
R5 \ t = = x ;
2 2
3 5
R6 \ t = = x
2 2
and (30) becomes
8 1 1 1
>
> 2 2
jx tj ; 2
x 25
>
< 1 1 5 1
3 2 2
jx + tj ; 2
x 2
u x; =
2 >
> 0; 1
x 21
>
: 2
0: jxj 5=2
These are plotted below.
(ii) Repeat the procedure in (i) for a string that has zero initial displacement but is given
an initial velocity (
2; jxj 1
ut (x; 0) = g (x) =
0; jxj > 1

11
 Solution: Step 1. Since u (x; 0) = 0, D’Alembert’s solution is
Z
1 x+t
u (x; t) = g (s) ds:
2 x t

Step 2. Regions. Since g (x) has the same form as f (x), namely it is a case function of
two cases, one for jxj 1 and one for jxj > 1, the characteristic plot is the same as for (i)
and the regions Rn are also the same (see (29)).
Step 3. Calculate solution in each interval. The integral can be calculated in each region,
since each region tells us where x + t and x t are with respect to 1. In region R1 ,
1 x t 1 so that
Z Z
1 x+t 1 x+t
u (x; t) = g (s) ds = 2ds = 2t
2 x t 2 x t

12
In region R2 , 1 x t 1 and x + t 1 so that
Z
1 x+t
u (x; t) = g (s) ds
2 x t
Z Z
1 1 1 x+t
= g (s) ds + g (s) ds
2 x t 2 1
Z Z
1 1 1 x+t
= 2ds + 0ds
2 x t 2 1
= 1 (x t)

Similarly, in region R3 ,
Z 1 Z x+t Z x+t
1 1 1
u (x; t) = g (s) ds + g (s) ds = 2ds = x + t + 1
2 x t 2 1 2 1

In region R4 ,
Z 1 Z 1 Z x+t
1 1 1
u (x; t) = g (s) ds + g (s) ds + g (s) ds
2 x t 2 1 2 1
Z 1
1
= 2ds = 2
2 1

13
In regions R5 and R6 , u (x; t) = 0. To summarize,
8
>
> 2t; (x; t) 2 R1
>
>
Z >
> (x; t) 2 R2
1 x+t < 1 x + t;
u (x; t) = g (s) ds = 1 + x + t; (x; t) 2 R3 (35)
2 x t >
>
>
> 2; (x; t) 2 R4
>
>
: 0; (x; t) 2 R5 ; R6
Step 4. For each speci…c time t = t0 , write the x-intervals corresponding to the sets Rn
(i.e. the intersection of the set Rn with ft = t0 g, or in the …gure above, where the line t = t0
intersects the region Rn ). Since the characteristics are the same, the intervals for x are the
same for each time. As a check, we note that at t = 0, the x-intervals correponding to each
region are given by (31) and (35) becomes
8
>
> 0; jxj 1
>
>
Z >
>
1 x < 1 x; x = 1
u (x; 0) = g (s) ds = 0 = 1 + x; x = 1 = 0 = f (x)
2 x >
>
>
> 2; ?
>
>
: 0; jxj 1
so this checks! At t = 1=2, the regions Rn are given by (32) and (35) becomes
8
>
> 2 21 ; 1
x 21 8
>
> 2
> 1; 1
x 1
Z x+1=2 >
> 1 x + 1
; 1
x 3 >
> 2 2
1 1 < 2 2 2 < 3
x; 1
x 3
u x; = g (s) ds = 1 + x + 21 ; 3
x 1
= 2 2 2
2 2 x 1=2 >
>
2 2 >
>
3
+ x; 3
x 1
>
> 2; ? >
: 2 2 2
>
> 0; jxj 3
: 0; jxj 23 2

At t = 1, the regions Rn are given by (33) and (35) becomes

8
Z x+1 >
< 2 x; 0 x 2
1
u (x; 1) = g (s) ds = 2 + x; 2 x 0
2 x 1 >
:
0; jxj 2
At t = 3=2, the regions Rn are given by (34) and (35) becomes
8 5 1 5
>
> 2
x; 2
x 2
Z x+3=2 >
< 5 5 1
3 1 2
+ x; 2
x 2
u x; = g (s) ds =
2 2 x 3=2 >
> 2; 1
x 1
>
: 2 2
5
0; jxj 2

These are plotted below.

14
15
4 Problem 4
(i) For an in…nite string (i.e. we don’t worry about boundary conditions), what initial
conditions would give rise to a purely forward wave? Express your answer in terms of the
initial displacement u (x; 0) = f (x) and initial velocity ut (x; 0) = g (x) and their derivatives
f 0 (x), g 0 (x). Interpret the result intuitively.
Solution: Recall in class that we write D’Alembert’s solution as

u (x; t) = P (x t) + Q (x + t) (36)

where
Z x
1
Q (x) = f (x) + g (s) ds + Q (0) P (0) (37)
2 0
Z x
1
P (x) = f (x) g (s) ds Q (0) + P (0) (38)
2 0

To only have a forward wave, we must have

Q (x) = const = q1

Substituting (37) gives

Z x
1
Q (x) = q1 = f (x) + g (s) ds
2 0

Di¤erentiating in x gives
1 df
0= + g (x)
2 dx
Thus
df
g (x) = (39)
dx
Substituting (39) into (37) gives
1
Q (x) = (f (0) + Q (0) P (0))
2
and setting x = 0 yields f (0) P (0) = Q (0). Substituting this and (39) into (38) gives
1
P (x) = (2f (x) f (0) Q (0) + P (0)) = f (x)
2
and hence
u (x; t) = f (x t) :

16
The displacement u (x; t) only contains the forward wave! Intuitively, we have set the initial
velocity of the string in such a way, given by Eq. (39), as to cancel the backward wave.
(ii) Again for an in…nite string, suppose that u (x; 0) = f (x) and ut (x; 0) = g (x) are
zero for jxj > ". Prove that if t + x > " and t x > ", then the displacement u (x; t) of the
string is constant. Relate this constant to g (x).
Solution: D’Alembert’s solution for the wave equation is
Z
1 1 x+t
u (x; t) = (f (x t) + f (x + t)) + g (s) ds
2 2 x t

If x+t > " and t x > ", then jx + tj > " and jx tj > ", so that f (x t) = 0. Furthermore,
x t < ", so that
Z x+t Z " Z 1
g (s) ds = g (s) ds = g (s) ds = c"
x t " 1

Thus c" is just the area under the curve g (x), and

u (x; t) = c" ; x + t > "; t x > ":

17