D’Alembert’s solution to the 1-D Wave equation and

its physical interpretation

1 A quick look back

Before embarking on detailed discussion on hyperbolic partial differential equations (PDEs),
specifically the 1-D Wave equation, let us quickly browse through some of the relevant
concepts.
A partial differential equation (PDE) is an equation involving a function u of several
variables and its partial derivatives. For example,

ut = uxx (heat equation in one-dimension),

ut = uxx + uyy (heat equation in two-dimensions),

1 1
urr + ur + 2 uθθ = 0 (Laplace’s equation in Polar coordinates),
r r
utt = uxx + uyy + uzz (wave equation in three-dimensions)
and
utt = uxx + αut + βu (telegraph equation)
are all PDEs.
Again a PDE of the form

Auxx + Buxy + Cuzz + Dux + Euy + F u = G (1)

where A, B, C, D, E, F and G are continuous functions of x and y possessing continuous

derivatives is called a second order linear PDE in two variables. Equation (1) is
called parabolic, elliptic or hyperbolic according as B 2 − 4AC = 0, < 0 or > 0.
Obviously, the 1-D wave equation

∂ 2u 2
2∂ u
=c (2)
∂t2 ∂x2
is a Hyperbolic one. Such equations represent wave motions and vibrating systems.

2 The solution
2.1 The travelling wave
Let us consider (2) as a purely initial value problem (transverse vibration of an infinite
string or a moving wave where the constant c represents the velocity with which the wave is
traversing, as would be seen later) in the (unbounded) domain 0 < t < ∞, −∞ < x < ∞.
with initial conditions

u(x, 0) = f (x), ut (x, 0) = g(x) −∞<x<∞

 t

( x0 t0 ) depends on the initial

’ displacement at two points

x−ct= x0 − ct0 x+ct= x0+ ct0

x
( x0 − ct0 0 ) ( x0+ ct0, 0 )
’

Figure 1: The characteristic triangle.

It will be seen that the solution (which we shall try to get with a slightly different
approach than the conventional ones) is
1 1 Z x+ct
u(x, t) = [f (x + ct) + f (x − ct)] + g(s)ds (3)
2 2c x−ct
Equation (3) clearly shows that the solution at any point (x0 , t0 ) depends upon
Z the value
of the initial displacement and the integral of the initial velocity viz, G(s) = g(s)ds at
the points x0 + ct0 and x0 − ct0 . So let us try to track these two points now, and with
these construct the triangle in Fig 1. This triangle will be providing us a lot of useful
information.

2.2 Problem 1: zero initial velocity

Consider equation (2) with initial conditions
(
1 if − 1 < x < 1
u(x, 0) =
0 everywhere else

ut (x, 0) = 0. The solution will then evolve as shown in Fig 2.

 u

1
t=0 −1 1
x

1/2
t=1/2c
u x

t=1/c
x

t=3/2c
x

t=2/c x
1
t=−

t=1
x+c

x−

x−
x+c

ct=

ct=
1

1−
Figure 2: The evolution of the solution of the problem with zero velocity.

The solution in the xt-plane is shown in Fig. 3.

t

x−ct=−1
x+ct=1
u(x,t)=0
u(x,t)=0

x+ct=−1
x−ct=1
u=1/2 u=1/2

u(x,t)=0 u(x,t)=0

u=1

x
−1 0 1

Figure 3: The solution of Problem 1 in the xt-plane.

2.3 Problem 2: zero initial displacement
We consider the same problem interchanging the initial velocity and the initial displace-
ment. The initial conditions now become:
u(x, 0) = 0.
(
1 if − 1 < x < 1
ut (x, 0) =
0 everywhere else
The problem has a solution in the xt-plane as illustrated in Fig. 4. It corresponds to
imposing an initial impulse (velocity=1) on the string for −1 < x < 1 and then watching
the resulting wave-motion. To find out the displacement, we compute the D’Alembert
t

Region 1 Region 2 Region 3 Region 4 Region 5

u=1/c

x+ct=−1
x−ct=1

1+x+ct 1−x+ct
u= u=
2c 2c
u(x,t)=0 u(x,t)=0

Region 6
u=t
x
−1 0 1

Figure 4: The solution of Problem 2 in the xt-plane.

solution as
1 Z x+ct
u(x, t) = g(s)ds (4)
2c x−ct
1 Z x+ct
= 0ds ∀(x, t) ∈ Region 1
2c x−ct
1 Z x+ct
= ds ∀(x, t) ∈ Region 2
2c −1
1 Z1
= ds ∀(x, t) ∈ Region 3
2c −1
1 Z1
= ds ∀(x, t) ∈ Region 4
2c x−ct
1 Z x+ct
= 0ds ∀(x, t) ∈ Region 5
2c x−ct
1 Z x+ct
= ds ∀(x, t) ∈ Region 6
2c x−ct
2.4 The standing wave
We are now ready to consider equation (2) as an initial-boundary value problem which
represents the motion of a vibrating string of finite length L. The solution which represents
the displacement of the string at subsequent time, has a very interesting interpretation in
terms of travelling waves. The problem can be redefined as:

∂ 2u 2
2∂ u
= c , 0 < x < L, t>0 (5)
∂t2 ∂x2
u(0, t) = 0 and u(L, t) = 0 ∀ t > 0 (6)
∂u
u(x, 0) = f (x) and (x, 0) = g(x) ∀ 0 < x < L (7)
∂t
The solution to this is
1 ∗ ∗ 1 Z x+ct ∗
u(x, t) = [f (x + ct) + f (x − ct)] + g (s)ds (8)
2 2c x−ct
It seems that this solution is similar to that of the travelling wave. But there is a slight
difference with f ∗ coming in place of f and g ∗ in place of g. They represent the odd
periodic extensions of f and g respectively.

2.5 Problem 3: zero initial velocity.

Consider equations (5)-(7) with L = 1, c = π1 ,
( 3
10
x if 0 ≤ x ≤ 31
f (x) = 3(1−x)
20
if 31 ≤ x ≤ 1.
π 2π
and g(x) = 0. Let us try to find out the shape of the string at time, say t = 3
and t = 3
,
which we would find as given in Fig. 5.
 (a)
f*(x)

0.1
0.05
x
−2/3−1/3 0 1/3 2/3 4/3 5/3

−0.1

f*(x+1/3) f*(x+2/3)

0.1 0.1
0.05
x x
0 1/3 2/3 1 0 1/3 2/3 1

f*(x−1/3) f*(x−2/3)
0.1 0.1
0.05

0 x 0 1/3 2/3 1 x
1/3 2/3 1
−0.05
−0.1 −0.1

u(x, π/3) u(x, 2π/3)

0.05
1/3 2/3 1
x 0 x
0 1/3 2/3 1
−0.05
−0.1

(b) (c)
π 2π
Figure 5: The shape of the string in Problem 3 at t = 3
and t = 3
.
2.6 Problem 4: zero initial displacement.
Consider equations (5)-(7) with c = 1, L = 1, f (x)Z = 0, and g(x) = x for 0 < x < 1.
x
An antiderivative G of g ∗ is then given by G(x) = g ∗ (s)ds (Fig. 6.). This results in
−1
(
1 2
x− 12 if − 1 < x < 1
G(x) = 2
G(x + 2) otherwise

Let us try to draw the graphs of g ∗ and G(x), and interpret the solution. Note how this

(a) g*(x) (b)

G(x)
1
x2/2−1/2

−2 −1 0 1 2 x −3 −1 0 1 3 x

−1
−1/2

Figure 6: The shape of g ∗ (x) and G(x) in Problem 4.

yields the zero initial position when t = 0.

3 Numerical solution
Although we are going to specifically deal with the numerical solution of hyperbolic type
of equations, still it would be worthwhile to recap some of the basic concepts while solving
PDEs numerically in the finite difference set up. The terms consistency, convergence,
stability and order are likely to crop up quite frequently.
A finite difference equation (FDE) is consistent with a PDE if the difference between
the FDE and the PDE (ie, the truncation error) vanishes as the size of the grid spacing
goes to zero independently. An FD method is said to be convergent if the solution of the
FDE approaches the exact solution of the PDE as the sizes of the grid spacings go to zero.
Also an FDE is stable if it produces a bounded solution when applied to a PDE, and
unstable if it produces an unbounded solution. Lastly, the order of a FD approximation
to a PDE is the rate at which the global error of the FD solution approaches zero as
the grid spacings approach zero. Another theorem that is very much in use in numerical
solution of PDEs is the Lax’s equivalence theorem which is as follows:
Given a properly posed linear initial value problem and a finite difference approximation
to it that is consistent, stability is the necessary and sufficient condition for convergence.
We are now in a position to discuss a few numerical schemes for solving the 1-D ad-
vection equation ft + ufx = 0 (again hyperbolic type) through which we shall try to il-
lustrate the concept of stability. As a tool to perform the error analysis, we shall use
the von Neumann approach. It will be seen that for this equation, the FTCS method is
unconditionally unstable whereas most of the methods including that of Lax’s, upwind,
Leapfrog, Lax-Wendroff one-step method are conditionally stable if the convection num-
u∆x
ber (also called CFL number) C = ≤ 1, where ∆x and ∆t are the space and time
∆t
steps of discretization respectively.
Finally we would numerically solve the 1-D wave (diffusion) equation, which interest-
ingly will be seen to produce a conditionally stable result for 0 < s ≤ 1 and an unstable
(∆x)2
one for s > 1 where s = c2 where c is as in equation (3) and ∆x and ∆t are just as
(∆t)2
described.