Strauss PDEs 2e: Section 3.

4 - Exercise 2 Page 1 of 11

Exercise 2
Solve utt = c2 uxx + eax , u(x, 0) = 0, ut (x, 0) = 0.

Solution

Solution by Operator Factorization

Bring c2 uxx to the other side.

utt − c2 uxx = eax
Write the left side as an operator acting on u.

(∂t2 − c2 ∂x2 )u = eax

The operator is a difference of squares, so it can be factored.

(∂t + c∂x )(∂t − c∂x )u = eax

Let
v = (∂t − c∂x )u
so that the PDE becomes
(∂t + c∂x )v = eax .
The second-order PDE we started with has thus been reduced to the following system of
first-order PDEs that can be solved with the method of characteristics.

ut − cux = v (1)
ax
vt + cvx = e (2)

We will solve the second one for v first, and once that is known, the first equation for u will be
solved. For a function of two variables φ = φ(x, t), its differential is defined as

∂φ ∂φ
dφ = dt + dx.
∂t ∂x
If we divide both sides by dt, then we get the relationship between the ordinary derivative of φ
and its partial derivatives.
dφ ∂φ ∂φ dx
= + (3)
dt ∂t ∂x dt
Comparing this with equation (2), we see that along the curves in the xt-plane that satisfy

dx
= c, (4)
dt
the PDE for v(x, t) reduces to an ODE.
dv
= eax (5)
dt
Because c is a constant, equation (4) can be solved by integrating both sides with respect to t.

x = ct + ξ, (6)

www.stemjock.com
Strauss PDEs 2e: Section 3.4 - Exercise 2 Page 2 of 11

where ξ is a characteristic coordinate. Substitute this expression for x into equation (5) to obtain
an ODE that only involves t (ξ is regarded as a constant).

dv
= ea(ct+ξ)
dt
Distribute a and split the exponential function into two.
dv
= eact eaξ
dt
Integrate both sides with respect to t.

eaξ act
v(ξ, t) = e + f (ξ),
ac
where f is an arbitrary function of the characteristic coordinate ξ. In order to write v in terms of
x and t, solve equation (6) for ξ.

x = ct + ξ → ξ = x − ct

Hence,

ea(x−ct) act
v(x, t) = e + f (x − ct)
ac
eax
= + f (x − ct).
ac
As a result, equation (1) becomes

eax
ut − cux = + f (x − ct).
ac
Comparing this equation with equation (3), we see that along the curves in the xt-plane that
satisfy
dx
= −c, (7)
dt
the PDE for u(x, t) reduces to an ODE.

du eax
= + f (x − ct) (8)
dt ac
Because c is a constant, equation (7) can be solved by integrating both sides with respect to t.

x = −ct + η, (9)

where η is another characteristic coordinate. Substitute this expression for x into equation (8) to
obtain an ODE that only involves t (η is regarded as a constant).

du ea(−ct+η)
= + f (−ct + η − ct)
dt ac
du eaη −act
= e + f (η − 2ct)
dt ac

www.stemjock.com
Strauss PDEs 2e: Section 3.4 - Exercise 2 Page 3 of 11

Integrate both sides with respect to t.

ˆ t
eaη
u(η, t) = − 2 2 e−act + f (η − 2cs) ds + g(η),
a c
where g is an arbitrary function of the characteristic coordinate η. The integral of an arbitrary
function is another arbitrary function.
eaη −act
u(η, t) = − e + F (η − 2ct) + g(η),
a2 c2
In order to write u in terms of x and t, solve equation (9) for η.

x = −ct + η → η = x + ct

Hence,

ea(x+ct) −act
u(x, t) = − e + F (x + ct − 2ct) + g(x + ct)
a2 c2
e ax
= − 2 2 + F (x − ct) + g(x + ct).
a c
This is the general solution to utt = c2 uxx + eax . If we apply the two initial conditions, we can
determine F and g. Before doing so, take a derivative of the solution with respect to t.

ut (x, t) = −cF 0 (x − ct) + cg 0 (x + ct)

From the initial conditions we obtain the following system of equations.

eax
u(x, 0) = − + F (x) + g(x) = 0
a2 c2
ut (x, 0) = −cF 0 (x) + cg 0 (x) = 0

Even though this system is in terms of x, it’s really in terms of w, where w is any expression we
choose.
eaw
− + F (w) + g(w) = 0
a2 c2
−cF 0 (w) + cg 0 (w) = 0

Differentiating both sides of the first equation with respect to w, we get

eaw eaw
− + F 0 (w) + g 0 (w) = 0 → g 0 (w) = − F 0 (w).
ac2 ac2
Plug this expression for g 0 (w) into the second equation.
 aw
eaw eaw

e
−cF 0 (w) + c − F 0
(w) = 0 → −2cF 0
(w) + =0 → F 0 (w) = .
ac2 ac 2ac2

Solve for F (w) and obtain an expression for F (x − ct).

eaw ea(x−ct)
F (w) = + C1 ⇒ F (x − ct) = + C1
2a2 c2 2a2 c2

www.stemjock.com
Strauss PDEs 2e: Section 3.4 - Exercise 2 Page 4 of 11

Use the first equation to solve for g(w) and obtain an expression for g(x + ct).

eaw
g(w) = − F (w)
a2 c2
e aw eaw
= 2 2 − 2 2 − C1
a c 2a c
e aw ea(x+ct)
= 2 2 − C1 ⇒ g(x + ct) = − C1
2a c 2a2 c2
The general solution for u(x, t) becomes

eax
u(x, t) = − + F (x − ct) + g(x + ct)
a2 c2
eax ea(x−ct) ea(x+ct)
=− 2 2 + + C1 + −
C
1
a c 2a2 c2  2a 2 c2
eax

1 1
= 2 2 −1 + e−act + eact
a c 2 2
eax
= 2 2 (−1 + cosh act)
a c
Therefore,
eax
u(x, t) = (cosh act − 1).
a2 c2

www.stemjock.com
Strauss PDEs 2e: Section 3.4 - Exercise 2 Page 5 of 11

Solution by the Method of Characteristics

Bring c2 uxx to the left side of the PDE.

utt − c2 uxx = eax
Comparing this with the general form of a second-order PDE,
Autt + Buxt + Cuxx + Dut + Eux + F u = G,
we see that A = 1, B = 0, C = −c2 , D = 0, E = 0, F = 0, and G = eax . The characteristic
equations for a second-order PDE are given by
dx 1 p
= (B ± B 2 − 4AC),
dt 2A
the solutions of which are known as the characteristics. Since B 2 − 4AC = 4c2 > 0, the PDE is
hyperbolic, so the solutions to these equations are two real and distinct families of characteristic
curves in the xt-plane.
dx 1 √
= (± 4c2 )
dt 2
dx 1
= (±2c)
dt 2
dx dx
= c or = −c
dt dt
Integrate both sides of each equation with respect to t.
x = ct + C2 or x = −ct + C3
Now make the substitutions,
ξ = x − ct = C2
η = x + ct = C3 ,
so that the PDE takes the simplest form. The aim is to write utt , uxx , and eax in terms of the
new variables, ξ and η. Solving these two equations for x and t with elimination gives
1
x = (η + ξ)
2
1
t = (η − ξ).
2c
Use the chain rule to write the old derivatives in terms of the new variables.
∂u ∂u ∂ξ ∂u ∂η
= + = uξ (−c) + uη (c) = c(uη − uξ )
∂t ∂ξ ∂t ∂η ∂t
∂u ∂u ∂ξ ∂u ∂η
= + = uξ (1) + uη (1) = uξ + uη
∂x ∂ξ ∂x ∂η ∂x
Find the second derivatives by using the chain rule again.
∂2u
   
∂ ∂ξ ∂ ∂η ∂ ∂ξ ∂ ∂η ∂
= c (uη − uξ ) = c + (uη − uξ ) = c (uη − uξ ) + (uη − uξ )
∂t2 ∂t ∂t ∂ξ ∂t ∂η ∂t ∂ξ ∂t ∂η
∂2u
 
∂ ∂ξ ∂ ∂η ∂ ∂ξ ∂ ∂η ∂
2
= (uξ + uη ) = + (uξ + uη ) = (uξ + uη ) + (uξ + uη )
∂x ∂x ∂x ∂ξ ∂x ∂η ∂x ∂ξ ∂x ∂η

www.stemjock.com
Strauss PDEs 2e: Section 3.4 - Exercise 2 Page 6 of 11

Hence,

∂2u
= c[(−c)(uξη − uξξ ) + (c)(uηη − uξη )] = c2 (uξξ − 2uξη + uηη )
∂t2
∂2u
= (1)(uξξ + uξη ) + (1)(uξη + uηη ) = uξξ + 2uξη + uηη .
∂x2
Substituting these expressions into the PDE, utt − c2 uxx = eax , we obtain
ha i
c2 (ξξ − 2uξη + H
u uηη ) − c2
(uξξ
 + 2uξη + u
Hηη ) = exp (η + ξ) .
H  H 2
Simplify the left side. ha i
−4c2 uξη = exp (η + ξ)
2
Divide both sides by −4c2 .
∂2u 1 ha i
= − 2 exp (η + ξ)
∂ξ∂η 4c 2
This is known as the first canonical form of the PDE. Integrate both sides of it partially with
respect to η.
ˆ η 2  ˆ η
∂ u  1 ha i
ds = − exp (s + ξ) ds + f (ξ),
∂ξ∂η η=s 4c2 2

where f is an arbitrary function of ξ.

∂u 1 2 ha iη
= − 2 exp (s + ξ)  + f (ξ)
∂ξ 4c a 2
∂u 1 ha i
=− exp (η + ξ) + f (ξ)
∂ξ 2ac2 2
Now integrate both sides partially with respect to ξ.
ˆ ξ  ˆ ξ 
∂u  1 ha i
ds = − exp (η + s) + f (s) ds + g(η),
∂ξ ξ=s 2ac2 2

where g is an arbitrary function of η.

 ξ
1 2 ha i 
u(ξ, η) = − exp (η + s) + F (s)  + g(η)
2ac2 a 2 
1 ha i
u(ξ, η) = − 2 2 exp (η + ξ) + F (ξ) + g(η)
a c 2
Since u has been solved for, change back to the original variables, x and t, by substituting the
expressions for ξ and η.
eax
u(x, t) = − 2 2 + F (x − ct) + g(x + ct)
a c
This is the general solution to utt = c2 uxx + eax . If we apply the two initial conditions, we can
determine F and g. Before doing so, take a derivative of the solution with respect to t.

ut (x, t) = −cF 0 (x − ct) + cg 0 (x + ct)

www.stemjock.com
Strauss PDEs 2e: Section 3.4 - Exercise 2 Page 7 of 11

From the initial conditions we obtain the following system of equations.

eax
u(x, 0) = − + F (x) + g(x) = 0
a2 c2
ut (x, 0) = −cF 0 (x) + cg 0 (x) = 0

Even though this system is in terms of x, it’s really in terms of w, where w is any expression we
choose.
eaw
− + F (w) + g(w) = 0
a2 c2
−cF 0 (w) + cg 0 (w) = 0

Differentiating both sides of the first equation with respect to w, we get

eaw eaw
− + F 0 (w) + g 0 (w) = 0 → g 0 (w) = − F 0 (w).
ac2 ac2
Plug this expression for g 0 (w) into the second equation.
 aw
eaw eaw

0 e 0 0
−cF (w) + c − F (w) = 0 → −2cF (w) + =0 → F 0 (w) = .
ac2 ac 2ac2

Solve for F (w) and obtain an expression for F (x − ct).

eaw ea(x−ct)
F (w) = + C1 ⇒ F (x − ct) = + C1
2a2 c2 2a2 c2
Use the first equation to solve for g(w) and obtain an expression for g(x + ct).

eaw
g(w) = − F (w)
a2 c2
eaw eaw
= 2 2 − 2 2 − C1
a c 2a c
e aw ea(x+ct)
= 2 2 − C1 ⇒ g(x + ct) = − C1
2a c 2a2 c2
The general solution for u(x, t) becomes

eax
u(x, t) = − + F (x − ct) + g(x + ct)
a2 c2
eax ea(x−ct) ea(x+ct)
=− 2 2 + + C
1
 + −
C
1
a c 2a2 c2 2a 2 c2
ax

e 1 1
= 2 2 −1 + e−act + eact
a c 2 2
eax
= 2 2 (−1 + cosh act)
a c
Therefore,
eax
u(x, t) = (cosh act − 1).
a2 c2

www.stemjock.com
Strauss PDEs 2e: Section 3.4 - Exercise 2 Page 8 of 11

Solution by Green’s Theorem

utt − c2 uxx = eax , −∞ < x < ∞, t > 0
u(x, 0) = 0 ut (x, 0) = 0
The characteristics were found to be straight lines, ξ = x − ct and η = x + ct, with slopes ±c.
Suppose (x0 , t0 ) is the point in the xt-plane we want to evaluate u at. The equations of the lines
going through this point are
x − x0 = c(t − t0 )
x − x0 = −c(t − t0 ).
Integrate both sides of the inhomogeneous wave equation over the triangular domain D enclosed
by these lines (from left to right as indicated below).

Write the double integral explicitly on the right side.

¨ ˆ t0 ˆ x0 −c(t−t0 )
2
(utt − c uxx ) dA = eax dx dt
0 x0 +c(t−t0 )
D

Rewrite the left side.

¨   ˆ t0 ˆ x0 −c(t−t0 )
∂ 2 ∂
− (c ux ) − (ut ) dA = eax dx dt
∂x ∂t 0 x0 +c(t−t0 )
D

Multiply both sides by −1.

¨   ˆ t0 ˆ x0 −c(t−t0 )
∂ 2 ∂
(c ux ) − (ut ) dA = − eax dx dt
∂x ∂t 0 x0 +c(t−t0 )
D

Apply Green’s theorem (essentially the divergence theorem in two dimensions) to the double
integral on the left to turn it into a counterclockwise line integral around the triangle’s boundary
bdy D.
ffi ˆ t0 ˆ x0 −c(t−t0 )
2
(ut dx + c ux dt) = − eax dx dt
0 x0 +c(t−t0 )
bdy D

www.stemjock.com
Strauss PDEs 2e: Section 3.4 - Exercise 2 Page 9 of 11

Let L1 , L2 , and L3 represent the legs of the triangle.

The line integral is the sum of three integrals, one over each leg.
ˆ ˆ ˆ ˆ t0 ˆ x0 −c(t−t0 )
(ut dx + c2 ux dt) + (ut dx + c2 ux dt) + (ut dx + c2 ux dt) = − eax dx dt
L1 L2 L3 0 x0 +c(t−t0 )

On L1 On L2 On L3
t=0 x − x0 = −c(t − t0 ) x − x0 = c(t − t0 )
dt = 0 dx = −c dt dx = c dt

Replace the differentials in the integrals over L2 and L3 .

ˆ x0 +ct0 ˆ    ˆ    ˆ t0 ˆ x0 −c(t−t0 )
2 dx 2 dx
ut (x, 0) dx + ut (−c dt) + c ux − + ut (c dt) + c ux =− eax dx dt
x0 −ct0 L2 c L3 c 0 x0 +c(t−t0 )

In this exercise ut (x, 0) = 0, so the integral over L1 vanishes.

ˆ   ˆ   ˆ t0 ˆ x0 −c(t−t0 )
∂u ∂u ∂u ∂u
−c dt + dx + c dt + dx = − eax dx dt
L2 ∂t ∂x L3 ∂t ∂x 0 x0 +c(t−t0 )

The integrands are how the differential of u = u(x, t) is defined.

ˆ ˆ ˆ t0 ˆ x0 −c(t−t0 )
−c du + c du = − eax dx dt
L2 L3 0 x0 +c(t−t0 )

Evaluate the integrals on the left side.

ˆ t0 ˆ x0 −c(t−t0 )
−c[u(x0 , t0 ) − u(x0 + ct0 , 0)] + c[u(x0 − ct0 , 0) − u(x0 , t0 )] = − eax dx dt
0 x0 +c(t−t0 )

In this exercise u(x, 0) = 0, so u(x0 + ct0 , 0) = 0 and u(x0 − ct0 , 0) = 0.

ˆ t0 ˆ x0 −c(t−t0 )
−2cu(x0 , t0 ) = − eax dx dt
0 x0 +c(t−t0 )

Divide both sides by −2c.

ˆ t0 ˆ x0 −c(t−t0 )
1
u(x0 , t0 ) = eax dx dt
2c 0 x0 +c(t−t0 )

www.stemjock.com
Strauss PDEs 2e: Section 3.4 - Exercise 2 Page 10 of 11

Finally, switch the roles of x and t with those of x0 and t0 , respectively.

ˆ tˆ x−c(t0 −t)
1
u(x, t) = eax0 dx0 dt0
2c 0 x+c(t0 −t)

Proceed to evaluate the last integral.

ˆ tˆ x+c(t−t0 )
1
u(x, t) = eax0 dx0 dt0
2c 0 x−c(t−t0 )
ˆ t ax0 x+c(t−t0 )

1 e
=  dt0
2c 0 a x−c(t−t0 )
ˆ t
1
= {ea[x+c(t−t0 )] − ea[x−c(t−t0 )] } dt0
2ac 0
 ˆ t ˆ t 
1 a(x+ct) −act0 a(x−ct) act0
= e e dt0 − e e dt0
2ac 0 0
1 − e−act
    act 
1 e −1
= ea(x+ct) − ea(x−ct)
2ac ac ac
1
= [ea(x+ct) − eax − eax + ea(x−ct) ]
2a2 c2
eax act
= (e + e−act − 2)
2a2 c2
eax
= (2 cosh act − 2)
2a2 c2
Therefore,
eax
u(x, t) = (cosh act − 1).
a2 c2

www.stemjock.com
Strauss PDEs 2e: Section 3.4 - Exercise 2 Page 11 of 11

Solution by Duhamel’s Principle

utt − c2 uxx = eax , −∞ < x < ∞, t > 0
u(x, 0) = 0 ut (x, 0) = 0
According to Duhamel’s principle, the solution to the inhomogeneous wave equation is
ˆ t
u(x, t) = U (x, t − s; s) ds,
0
where U = U (x, t; s) is the solution to the associated homogeneous equation with a particular
choice for the initial conditions.
Utt − c2 Uxx = 0, −∞ < x < ∞, t > 0
U (x, 0; s) = 0 Ut (x, 0; s) = eax
The solution for U is given by d’Alembert’s formula in section 2.1 on page 36.
ˆ
1 x+ct ar
U (x, t; s) = e dr
2c x−ct
1 ear x+ct

=
2c a  x−ct
1 a(x+ct)
= [e − ea(x−ct) ]
2ac
eax act
= (e − e−act )
2ac
eax
= sinh act
ac
The solution to the inhomogeneous wave equation is then
ˆ t ax
e
u(x, t) = sinh[ac(t − s)] ds
0 ac
ˆ
eax t
= {− sinh[ac(s − t)]} ds
ac 0
eax cosh act − 1
 
=
ac ac
e ax
= (cosh act − 1).
ac
We can check that the Duhamel solution satisfies the wave equation. Use the Leibnitz rule to
differentiate the integrals.
 ˆ t  2 ˆ t
2 ∂ ∂ 2 ∂
utt − c uxx = U (x, t − s; s) ds − c U (x, t − s; s) ds
∂t ∂t 0 ∂x2 0
ˆ t  ˆ t
∂ ∂
= U (x, t − s; s) ds + U (x, 0; t) ·1 − U (x, t; 0) · 0 − c2 Uxx (x, t − s; s) ds
∂t 0 ∂t | {z } 0
=0
ˆ t 2 ˆ t
∂ 2
= 2
U (x, t − s; s) ds + Ut (x, 0; t) · 1 − Ut (x, t; 0) · 0 − c Uxx (x, t − s; s) ds
0 ∂t 0
ˆ t
= [Utt (x, t − s; s) − c2 Uxx (x, t − s; s)] ds + Ut (x, 0; t) = eax
0 | {z }
=0

www.stemjock.com