Chapter 8

Hyperbolic Equations

Consider the one dimensional wave equation

∂2u 2
2∂ u
= a (8.1)
∂t2 ∂x2
It has been shown in the earlier chapter that, two real characteristics exists for the
equation (8.1) along which the given differential equation takes a simpler form given
by
∂2u
= 0 (8.2)
∂ξ∂η
where ξ = x+at and η = x−at. By integrating on both sides of (8.2), it can solved to
get u(ξ, η) = φ(ξ)+ψ(η) where φ and ψ are two arbitrary functions (uξη = 0 ⇒ uξ =
F (ξ) ⇒ u = φ(ξ) + ψ(η)). Therefore, the solution of (8.1) can be written as
u = φ(ξ) + ψ(η) ⇒ u(x, y) = φ(x + at) + ψ(x − at) (8.3)

8.1 D’Alembert’s Solution

Include the initial conditions (for an initial value or Cauchy problem)
∂u
u(x, 0) = f (x) and |t=0 = g(x) (8.4)
∂t
and eliminate the arbitrary functions from (8.3) gives the solution of the wave equation
in the domain −∞ < x < ∞, t > 0 with initial conditions (8.4) at t = 0. The
D’Alembert’s solution is given by
u(x, t) = φ(x + at) + ψ(x − at)
u(x, 0) = φ(x) + ψ(x) = f (x)
∂u
= aφ (x + at) − aψ  (x − at)
∂t
∂u
(x, 0) = aφ (x) − aψ  (x) = g(x)
∂t

91
92 Sanyasiraju V S S Yedida sryedida@iitm.ac.in

Integrating the last equation in the limits x0 to x gives

 x  x
 
(aφ (x) − aψ (x)) dx = g(x) dx
x0 x0
 x
1
φ(x) − ψ(x) = φ(x0 ) + ψ(x0 ) + g(x) dx
a x0

We also have

φ(x) + ψ(x) = f (x)

Solving these two equations for φ and ψ gives

 x
φ(x0 ) + ψ(x0 ) + f (x) 1
φ(x) = + g(x) dx
2 2a x0
 x
φ(x0 ) + ψ(x0 ) − f (x) 1
ψ(x) = − − g(x) dx
2 2a x0

Therefore
 x+at
φ(x0 ) + ψ(x0 ) + f (x + at) 1
φ(x + at) = + g(x) dx
2 2a x0
 x−at
φ(x0 ) + ψ(x0 ) − f (x − at) 1
ψ(x − at) = − − g(x) dx
2 2a x0

Finally, the D’Alembert’s solution for the wave equation is

u(x, y) = φ(x + at) + ψ(x − at)

 x+at
f (x + at) + f (x − at) 1
= + g(x) dx (8.5)
2 2a x−at

It is clear from the solution (8.5) that, the solution of the wave equation at any
point (x, t) depends on f at the two points (x − at, 0) and (x + at, 0) and also on g
between x − at and x + at, as shown in the Figure (8.1), which is called the domain
of dependence. That is, the region on which, the solution at any point say, P (x, y)
depends on is bounded by the lines x − at = 0, x + at = 0 and the x-axis. Therefore,
any changes in the initial conditions outside this region can’t influence the solution
at the point P . Similarly, the domain influence of the point P is the region bounded
by the extension of the lines x − at = 0 and x + at = 0 for all the later times of t.

8.1.1 Numerical Examples

1. Find the vertical displacement u(x, t) in a plane of an infinitely long elastic
string which is started from rest and having an initial displacement K sin πx,
where K is a constant (Assume a2 = Tρ = 1, where T is the tension and ρ is the
Lecture Notes MA2020 Differential Equations 93

P(x,t)

Domain of
Dependence

P1(x−at,0) P (x+at,0) t=0

2

Figure 8.1: Domain of dependence

density).

Solution: By D’Alembert’s solution, we have f (x) = K sin πx and g(x) = 0,

therefore, the solution is

K
u(x, t) = (sin π(x + t) + sin π(x − t)) = K sin πx cos πt
2

2. Find the vertical displacement u(x, t) in a plane of an infinitely long elastic

string which is started with a constant velocity g and having an initial displace-
ment K sin πx, where K is a constant (Assume a2 = Tρ = 1, where T is the
tension and ρ is the density).

Solution: By D’Alembert’s solution, we have f (x) = K sin πx and g(x) = g,

therefore, the solution is
 x+t
K 1
u(x, t) = (sin π(x + t) + sin π(x − t)) + g dx = K sin πx cos πt + gt
2 2 x−t

8.1.2 Problems to workout

Solve the wave equation using the following initial conditions x ∈ (−∞, ∞)

1. f (x) = Kx(1 − x) and g(x) = 0

2. f (x) = K(x − x3 ) and g(x) = 0

3. f (x) = K(1 − cos 2πx) and g(x) = 0

2
4. f (x) = 0 and g(x) = −4xe−x
94 Sanyasiraju V S S Yedida sryedida@iitm.ac.in

8.2 Fourier Series based solutions for Hyperbolic

Equations
Hyperbolic equations defined in finite domains with boundary conditions, that is
initial-boundary value hyperbolic equations can be solved using a method called Sep-
aration of variables in which series solutions can be obtained with the help of Fourier
series. Before looking at such problems, in the next section, we look at some of the
important concepts, which we need in our discussion, from Fourier series.

8.2.1 Fourier Series

Due to the orthogonal nature of sin nπx
L
and cos nπx
L
, in the interval 0 to L, given by
 L 
nπx mπx 0 m = n
sin sin dx = L (8.6)
0 L L 2
m=n
⎧
 L ⎨ 0 m = n
nπx mπx L
cos cos dx = m=n (8.7)
0 L L ⎩ 2
L m=n=0
any piecewise smooth function f on the interval −L ≤ x ≤ L can be expressed as

∞
nπx 
∞
nπx
f (x) ∼ a0 + an cos + bn sin (8.8)
n=1
L n=1
L

where
 L
1
a0 = f (x) dx (8.9)
2L −L

1 L nπx
an = f (x) cos dx (8.10)
L −L L

1 L nπx
bn = f (x) sin dx (8.11)
L −L L
for n = 1, 2, . . ..

Some Convergence Results

1. The, fourier series, (right hand side of (8.8)), converges to
(a) the periodic extension of f (x) if f is continuous at x
(b) the average of the two limits, that is 12 (f (x+) + f (x−)) at the points of
jump discontinuity
2. The Fourier series of f (x) is continuous and converge to f (x) for −L ≤ x ≤ L
if and only if f (x) is continuous and f (−L) = f (L). The corresponding results
for cosine and sine series (the series obtained with even and odd extensions of
f (x) in 0 ≤ x ≤ L, respectively) are
Lecture Notes MA2020 Differential Equations 95

(a) The Fourier cosine series of f (x) is continuous and converge to f (x) for
0 ≤ x ≤ L if and only if f (x) is continuous
(b) The Fourier sine series of f (x) is continuous and converge to f (x) for
0 ≤ x ≤ L if and only if f (x) is continuous and f (0) = f (L) = 0
3. A Fourier series that is continuous can be differentiated term by term if f  (x)
is piecewise smooth. Therefore, if f (x) is piecewise smooth then f (x) can be
differentiated term by term if f (−L) = f (L).

Counter Example: 2 ∞ L
n=1 nπ (−1)
n+1
sin nπx
L
is the Fourier sine series of x
inthe interval 0 ≤ x < L however, the series with its termwise derivatives
2 ∞ n=1 (−1)
n+1
cos nπx
L
is not a cosine series of f (x) = 1 (the cosine series of 1
is 1 itself).
4. If f  (x) is piecewise smooth, then the Fourier cosine series of a continuous
function f (x) can be differentiated term by term.
5. If f  (x) is piecewise smooth, then the Fourier sine series of a continuous function
f (x) can be differentiated term by term if f (0) = f (L) = 0.
From the results given above, since the solutions of the wave equation are twice
differentiable (space direction), therefore, Fourier series along with term by term
differentiation exists for the solutions of Hyperbolic equations with homogeneous
boundary conditions.

8.2.2 Fixed Oscillations of a Thin Homogeneous String

A thin perfectly flexible homogenous string under uniform tension is tied at the ends
x = 0 and x = L. The string is pulled in a planar region and released. In the absence
of any external forces, the problem is modeled by
PDE utt = a2 uxx 0 < x < L, t > 0
Initial Conditions u(x, 0) = f (x), ut (x, 0) = g(x) 0 ≤ x ≤ L (8.12)
Boundary Conditions u(0, t) = u(L, t) = 0 t>0
Computation of u : To separate the variables, assume
u(x, t) = X(x)T (t)
Substituting in the differential equation gives
1 d2 T 1 d2 X
=
a2 T dt2 X dx2
Since the left hand side of the above equation is a function of t alone and similarly,
the right hand side is function of x alone, therefore, the equality of these two enforces
the constant nature of these two. That is
1 d2 T 1 d2 X
= =k
a2 T dt2 X dx2
96 Sanyasiraju V S S Yedida sryedida@iitm.ac.in

where k is a constant. Equivalently, we have

d2 T d2 X
− ka2 T = 0, = kX = 0
dt2 dx2

Further, the boundary conditions on u at x = 0 and x = L gives

X(0)T (t) = X(L)T (t) = 0
that is, either X is zero at x = 0 and x = L or T ≡ 0. Since the latter condition
makes u ≡ 0 for all times, for a possibility of any non-zero solution choose X is zero
at x = 0 and x = L.

1. Case i: k = λ2 (a positive constant)

Solving the ODE for X gives X(x) = c1 eλx + c2 e−λx , which gives X ≡ 0 for the
boundary conditions X is zero at x = 0 and x = L.
2. Case ii: k = 0
Once again solving for X gives X(x) = c1 + c2 x, which again gives X ≡ 0 for
the boundary conditions X is zero at x = 0 and x = L.
3. Case iii: k = −λ2 (a negative constant)
Solving the ODE for X gives X(x) = c1 cos λx + c2 sin λx
X(0) = 0 ⇒ c1 = 0
X(L) = 0 ⇒ c2 sin λL = 0
If c2 = 0, then once again X ≡ 0, therefore, sin λL = 0 must be zero. That is,
nπ
λL = nπ ⇒ λ =
L
for n = 0, ±1, ±2, · · · .

Now using the facts that sine function is odd and sin 0 = 0, we have
nπ
λ= for n = 1, 2, · · ·
L
nπ
Now, solving for T gives (after replacing λ with L
)
nπat nπat
T (t) = A cos + B sin for n = 1, 2, · · ·
L L
where A and B are constants. Now, using the superposition principle, the solution
for u(x, t) can be written as
∞

nπx nπat nπat

u(x, t) = sin An cos + Bn sin
n=1
L L L
Lecture Notes MA2020 Differential Equations 97

where An and Bn for n = 0, ±1, ±2, · · · are constants. Finally, the initial conditions
must be applied to find the remaining constants in the solution. That is,


∞

nπx nπa(0) nπa(0)

u(x, 0) = sin An cos + Bn sin = f (x)
n=1
L L L

∞
nπx
An sin = f (x)
n=1
L

The last equation is a Fourier sine series, therefore, An can obtained as

 L
2 nπx
An = f (x) sin dx
L 0 L

for n = 1, 2, · · · .

The second initial condition is ut (x, 0) = g(x), that is


∞

nπa nπx nπa(0) nπa(0)

sin −An sin + Bn cos = g(x)
n=1
L L L L

∞
nπa nπx
sin Bn = g(x)
n=1
L L
∞
nπa nπx
Bn sin = g(x)
n=1
L L

∞
nπx nπa
Bn∗ sin = g(x), where Bn∗ = Bn
n=1
L L

Again solving the Fourier sine series gives

 L
nπa 2 nπx
Bn∗ = Bn = g(x) sin dx
L L 0 L

for n = 1, 2, · · ·

The required solution is


∞

nπx nπat nπat

u(x, t) = sin An cos + Bn sin (8.13)
n=1
L L L

2 L nπx
An = f (x) sin dx (8.14)
L 0 L
 L
2 nπx
Bn = g(x) sin dx (8.15)
nπa 0 L
98 Sanyasiraju V S S Yedida sryedida@iitm.ac.in

8.2.3 Problems to Workout

1. Show that the above solution satisfies the given initial and boundary conditions.

2. Solve the wave equation with homogeneous boundary conditions and the fol-
lowing initial conditions f (x) = sin3 πx
2
and g(x) = 0 (take a = 1 in the PDE).

3. Assuming g(x) = 0, show that the above solution gives the D’Alemberts solution
of a suitable problem

4. Solve the wave equation with homogeneous

 boundary conditions and the fol-
2k
L
x if 0 < x < L2
lowing initial conditions f (x) = 2k , g(x) = 0 (take
L
(L − x) if L2 < x < L
a = 1 in the PDE).

5. Solve the wave equation with homogeneous boundary conditions, L = π, a = 1,

and with the following initial deflection and velocities

(a) f (x) = k(sin x − 12 sin 2x), g(x) = 0

(b) f (x) = .1x(π 2 − x2 ), g(x) = 0

0.01x if 0 ≤ x ≤ π2
(c) f (x) = 0, g(x) = ,
0.01(π − x) if π2 ≤ x ≤ L

8.2.4 Forced Vibrations

PDE utt − a2 uxx = F (x, t) 0 < x < L, t > 0

Initial Conditions u(x, 0) = f (x), ut (x, 0) = g(x) 0 ≤ x ≤ L (8.16)
Boundary Conditions u(0, t) = u(L, t) = 0 t>0

Define u = u1 + u2 such that u1 is the solution of (8.12) and u2 is the solution of the
problem

PDE utt − a2 uxx = F (x, t) 0 < x < L, t > 0

Initial Conditions u(x, 0) = 0, ut (x, 0) = 0 0 ≤ x ≤ L (8.17)
Boundary Conditions u(0, t) = u(L, t) = 0 t>0

then u satisfies the given equation and also the initial and boundary conditions. Since
u1 is already known from the earlier problem (problem of free vibrations of the string),
it is complete if we compute u2 . Let assume (from the understanding of the problem
of the free vibrations)


∞
nπx
u2 (x, t) = φn (t) sin (8.18)
n=1
L
Lecture Notes MA2020 Differential Equations 99

(8.17) already satisfies the boundary conditions of (8.16) and it will satisfy the initial
conditions if, φn (0) = φn (0) = 0, n = 1, 2, · · · . Substituting (8.17) in the differential
equation gives
∞ 2


d φn (t) n2 π 2 a2 nπx
2
+ 2
φn (t) sin = F (x, t)
n=1
dt L L
∞ 2

d φn (t) 2 nπx nπa

2
+ ωn φn (t) sin = F (x, t), ωn =
n=1
dt L L

Multiplying with sin kπx

L
on both sides and then integrating in the limits x ∈ (0, L)
gives (using the orthogonal property)
2
 L
d φk (t) 2 2 kπx
2
+ ωk φk (t) = F (x, t) sin dx
dt L 0 L
d2 φk (t)
+ ωk2 φk (t) = F̄k (t)
dt2

L
where F̄k (t) = L2 0 F (x, t) sin kπx L
dx subject to the boundary conditions φk (0) =

φk (0) = 0 for k = 1, 2, · · · . Solving this problem with variation of parameters gives
the solution
 t
1
φk (t) = F̄k (s) sin(ωk (t − s)) ds
ωk 0
Therefore, by superposition principle, the solution u2 is given by
∞   t 
1 nπx
u2 (x, t) = F̄k (s) sin(ωk (t − s)) ds sin
n=1
ωk 0 L

Finally, the solution of the given forced vibrations problem is given by


∞
nπx
u(x, t) = u1 + u2 = {An cos ωn t + Bn sin ωn t} sin
n=1
L
∞   t 
1 nπx
+ F̄k (s) sin(ωk (t − s)) ds sin
n=1
ωk 0 L

where

2 L nπx
An = f (x) sin dx
L 0 L
 L
2 nπx
Bn = g(x) sin dx
nπa 0 L
100 Sanyasiraju V S S Yedida sryedida@iitm.ac.in