7/28/2019 AEMC15
1/37
15 Integral Transform Method
Exercises 15.1
1. (a) The result follows by letting = u2 or u =
in erf(
t ) =2
t0
eu2
du.
(b) Using {t1/2} =
s1/2and the first translation theorem, it follows from the convolution theorem that
erf(
t)
=1
t0
e
d
=
1
{1}
t1/2et
=1
1
s
t1/2
ss+1
=
1
1
s
s + 1 =1
ss + 1 .
2. Since erfc(
t ) = 1 erf(t ) we have
erfc(
t )
= {1}
erf(
t )
=1
s 1
s
s + 1=
1
s
1 1
s + 1
.
3. By the first translation theorem,
et erf(
t )
=
erf(
t )
ss1
=1
s
s + 1
ss1
=1
s (s
1)
.
4. By the first translation theorem and the result of Problem 2,
et erfc(
t )
=
erfc(
t )
ss1=
1
s 1
s
s + 1
ss1
=1
s 1 1
s (s 1)
=
s 1
s (s 1) =
s 1s (
s + 1)(
s 1) =1
s (
s + 1).
5. From table entry 3 and the first translation theorem we have
eGt/Cerfx2RCt = eGt/C1 erfcx2RCt =
eGt/C
eGt/Cerfc
x
2
RC
t
=1
s + G/C e
xRC
s
s
ss+G/C
=1
s + G/C e
xRC
s+G/C
s + G/C=
C
Cs + G
1 ex
RCs+RG
.
702
7/28/2019 AEMC15
2/37
Exercises 15.1
6. We first compute
sinh a
s
s sinh
s=
eas ea
s
s(es es) =
e(a1)s e(a+1)
s
s(1 e2s )
=
e(a1)s
s 1 + e2s + e4s + e(a+1)s
s 1 + e2s + e4s + =
e(1a)
s
s+
e(3a)s
s+
e(5a)s
s+
e(1+a)s
s+
e(3+a)s
s+
e(5+a)s
s+
=n=0
e(2n+1a)
s
s e
(2n+1+a)s
s
.
Then
sinh a
s
s sinhs =
n=0
e(2n+1a)s
s e(2n+1+a)s
s =
n=0
erfc
2n + 1 a
2
t
erfc
2n + 1 + a
2
t
=n=0
1 erf
2n + 1 a
2
t
1 erf
2n + 1 + a
2
t
=n=0
erf
2n + 1 + a
2
t
erf
2n + 1 a
2
t
.
7. Taking the Laplace transform of both sides of the equation we obtain
{y(t)} = {1} t0
y()t d
Y(s) =
1
s Y(s)
s
s +
sY(s) =
1
s
Y(s) =1
s (
s +
).
Thus
y(t) = 1
s (
s +
) = et erfc(
t ). By entry 5 in the table
8. Using entries 3 and 5 in the table, we haveeabeb2t erfc
b
t +a
2
t
+ erfc
a
2
t
=
eabeb2t erfc
b
t +a
2
t
+
a
2
t
= eas
s (
s + b)+
eas
s
703
7/28/2019 AEMC15
3/37
-10 -5 5 10x
-2
-1
1
2
y
erfHxLerfcHxL
Exercises 15.1
= eas
1
s 1
s (
s + b)
= ea
s
1
s
s
s (
s + b)
= eas
s + b s
s (s + b) =bea
s
s (s + b).
9.
ba
eu2
du =
0a
eu2
du +
b0
eu2
du =
b0
eu2
dua
0
eu2
du
=
2erf(b)
2erf(a) =
2[erf(b) erf(a)]
10. Since f(x) = ex2
is an even function, aa
eu2
du = 2
a0
eu2
du.
Therefore,
aa
eu2
du = erf(a).
11. The function erf(x) is symmetric with respect to the origin,
while erfc(x) appears to be symmetric with respect to the point
(0, 1). From the graph it appears that limx erf(x) = 1and limx erfc(x) = 2.
Exercises 15.2
1. The boundary-value problem is
a22u
x2=
2u
t2, 0 < x < L, t > 0,
u(0, t) = 0, u(L, t) = 0, t > 0,
u(x, 0) = A sin
Lx,
u
t
t=0
= 0.
Transforming the partial differential equation gives
d2U
dx2 s
a
2U = s
a2A sin
Lx.
Using undetermined coefficients we obtain
U(x, s) = c1 coshs
ax + c2 sinh
s
ax +
As
s2 + a22/L2sin
Lx.
The transformed boundary conditions, U(0, s) = 0, U(L, s) = 0 give in turn c1 = 0 and c2 = 0. Therefore
U(x, s) =As
s2 + a22/L2sin
Lx
704
7/28/2019 AEMC15
4/37
Exercises 15.2
and
u(x, t) = A
s
s2 + a22/L2
sin
Lx = A cos
a
Lt sin
Lx.
2. The transformed equation isd2U
dx2 s2U = 2sin x 4sin3xand so
U(x, s) = c1 cosh sx + c2 sinh sx +2
s2 + 2sin x +
4
s2 + 92sin3x.
The transformed boundary conditions, U(0, s) = 0 and U(1, s) = 0 give c1 = 0 and c2 = 0. Thus
U(x, s) =2
s2 + 2sin x +
4
s2 + 92sin3x
and
u(x, t) = 2
1
s2 + 2
sin x + 4
1
s2 + 92
sin3x
=2
sin t sin x +
4
3sin3t sin3x.
3. The solution of
a2d2U
dx2 s2U = 0
is in this case
U(x, s) = c1e(x/a)s + c2e(x/a)s.
Since limx u(x, t) = 0 we have limx U(x, s) = 0. Thus c2 = 0 and
U(x, s) = c1e(x/a)s.
If {u(0, t)} = {f(t)} = F(s) then U(0, s) = F(s). From this we have c1 = F(s) and
U(x, s) = F(s)e(x/a)s
.
Hence, by the second translation theorem,
u(x, t) = f
t xa
t x
a
.
4. Expressing f(t) in the form (sin t)[1 (t 1)] and using the result of Problem 3 we find
u(x, t) = f
t xa
t x
a
= sin
t x
a
1
t x
a 1
t x
a
= sin t
x
a t x
a tx
a t x
a 1
= sin
t xa
t x
a
t x
a 1
Now
t xa
t x
a 1
=
0, 0 t < x/a1, x/a t x/a + 10, t > x/a + 1
=
0, x < a(t 1) or x > at1, a(t 1) x at
705
7/28/2019 AEMC15
5/37
Exercises 15.2
so
u(x, t) =
0, x < a(t 1) or x > atsin (t x/a), a(t 1) x at.
The graph is shown for t > 1.
5. We use
U(x, s) = c1e(x/a)s g
s3.
Now
{u(0, t)} = U(0, s) = As2 + 2
and so
U(0, s) = c1 gs3
=A
s2 + 2or c1 =
g
s3+
A
s2 + 2.
Therefore
U(x, s) =A
s2 + 2e(x/a)s +
g
s3e(x/a)s g
s3
and
u(x, t) = A
e(x/a)s
s2 + 2
+ g
e(x/a)s
s3
g
1
s3
= A sin
t xa
t x
a
+
1
2g
t xa
2 t x
a
1
2gt2.
6. Transforming the partial differential equation gives
d2U
dx2 s2U =
s2 + 2 sin x.
Using undetermined coefficients we obtain
U(x, s) = c1 cosh sx + c2 sinh sx +
(s2 + 2)(s2 + 2)sin x.
The transformed boundary conditions U(0, s) = 0 and U(1, s) = 0 give, in turn, c1 = 0 and c2 = 0. Therefore
U(x, s) =
(s2 + 2)(s2 + 2)sin x
and
u(x, t) = sin x 1(s2 + 2)(s2 + 2)=
2 2 sin x
1
s2 + 2 1
s2 + 2
=
(2 2) sin t sin x 1
2 2 sin t sin x.
7. We use
U(x, s) = c1 coshs
ax +2 sinh
s
ax.
706
7/28/2019 AEMC15
6/37
Exercises 15.2
Now U(0, s) = 0 implies c1 = 0, so U(x, s) = c2 sinh(s/a)x. The condition EdU
dx
x=L
= F0 then yields
c2 = F0a/Es cosh(s/a)L and so
U(x, s) =aF0Es
sinh(s/a)x
cosh(s/a)L=
aF0Es
e(s/a)x e(s/a)xe(s/a)L + e(s/a)L
=aF0Es
e(s/a)(xL) e(s/a)(x+L)1 + e2sL/a
=aF0
E
e(s/a)(Lx)
s e
(s/a)(3Lx)
s+
e(s/a)(5Lx)
s
aF0E
e(s/a)(L+x)
s e
(s/a)(3L+x)
s+
e(s/a)(5L+x)
s
=aF0
E
n=0
(1)n
e(s/a)(2nL+Lx)
s e
(s/a)(2nL+L+x)
s
and
u(x, t) = aF0
En=0
(1)n e(s/a)(2nL+Lx)s
e(s/a)(2nL+L+x)s
=aF0
E
n=0
(1)n
t 2nL + L xa
t 2nL + L x
a
t 2nL + L + xa
t 2nL + L + x
a
.
8. We use
U(x, s) = c1e(x/a)s + c2e(x/a)s v0
s2.
Now limxdU
dx = 0 implies c2 = 0, and U(0, s) = 0 then gives c1 = v0/s2. Hence
U(x, s) =v0s2
e(x/a)s v0s2
and
u(x, t) = v0
t x
a
t x
a
v0t.
9. Transforming the partial differential equation gives
d2U
dx2 s2U = sxex.
Using undetermined coefficients we obtain
U(x, s) = c1esx + c2esx 2s
(s2
1)2
ex +s
s2
1
xex.
The transformed boundary conditions limx U(x, s) = 0 and U(0, s) = 0 give, in turn, c2 = 0 andc1 = 2s/(s2 1)2. Therefore
U(x, s) =2s
(s2 1)2 esx 2s
(s2 1)2 ex +
s
s2 1 xex.
From entries (13) and (26) in the table we obtain
u(x, t) =
2s
(s2 1)2 esx 2s
(s2 1)2 ex +
s
s2 1 xex
= 2(t x)sinh(t x) (t x) tex sinh t + xex cosh t.
707
7/28/2019 AEMC15
7/37
0 2 4 6
8 10x
510
15
t
0
25
50
75
100
uHx,tL
2 4
5 00 10 00 15 00 2 00 0t
20
40
60
80
100
uH10,tL
Exercises 15.2
10. We use
U(x, s) = c1exs + c2exs +
s
s2 1 ex.
Now limx u(x, t) = 0 implies limx U(x, s) = 0, so we define c2 = 0. Then
U(x, s) = c1exs +s
s2 1 ex.Finally, U(0, s) = 1/s gives c1 = 1/s s/(s2 1). Thus
U(x, s) =1
s s
s2 1 exs +
s
s2 1 ex
and
u(x, t) =
s
s2 1 e(x/a)s
+
s
s2 1
ex
= cosh
t xa
t x
a
+ ex cosh t.
11. (a) We useU(x, s) = c1e
s/k x + c2es/k x.
Now limx u(x, t) = 0 implies limx U(x, s) = 0, so we define c2 = 0. Then
U(x, s) = c1es/k x.
Finally, from U(0, s) = u0/s we obtain c1 = u0/s. Thus
U(x, s) = u0es/k x
s
and
u(x, t) = u0 es/k x
s = u0 e(x/
k )s
s = u0 erfc x2kt .
Since erfc(0) = 1,
limt
u(x, t) = limt
u0 erfc(x/2
kt ) = u0.
(b)
12. (a) Transforming the partial differential equation and using the initial condition gives
kd2U
dx2 sU = 0.
Since the domain of the variable x is an infinite interval we write the general solution of this differential
equation as
U(x, s) = c1es/k x + c2e
s/k x.
708
7/28/2019 AEMC15
8/37
0 2 4 6 8 10x
510
15
t
0
10
20
30
40
uHx,tL
2 4
5 00 1 00 0 1 50 0 2 00 0t
100
200
300
400
500
uH10,tL
Exercises 15.2
Transforming the boundary conditions gives U(0, s) = A/s and limx
U(x, s) = 0. Hence we find c2 = 0
and c1 = A
k/s
s . From
U(x, s) = A
kes/k x
s
s
we see that
u(x, t) = A
k
es/k x
s
s
.
With the identification a = x/
k it follows from entry 49 of the table in Appendix III that
u(x, t) = A
k
2
t
ex
2/4kt xk
x/2
kt
= 2A
kt
ex
2/4kt Ax erfc
x/2
kt
.
Since erfc(0) = 1,
limt
u(x, t) = limt
2Akt
ex
2/4kt Ax erfc x2
kt
= .
(b)
13. We use
U(x, s) = c1es x + c2e
sx + u1s
.
The condition limx u(x, t) = u1 implies limx U(x, s) = u1/s, so we define c2 = 0. Then
U(x, s) = c1es x +
u1s
.
From U(0, s) = u0/s we obtain c1 = (u0 u1)/s. Thus
U(x, s) = (u0 u1) es x
s+
u1s
and
u(x, t) = (u0 u1)
exs
s
+ u1
1
s= (u0 u1) erfc
x
2
t+ u1.
14. We use
U(x, s) = c1es x + c2e
sx +
u1x
s.
The condition limx u(x, t)/x = u1 implies limx U(x, s)/x = u1/s, so we define c2 = 0. Then
U(x, s) = c1es x +
u1x
s.
From U(0, s) = u0/s we obtain c1 = u0/s. Hence
U(x, s) = u0e
s x
s+
u1x
s
709
7/28/2019 AEMC15
9/37
Exercises 15.2
and
u(x, t) = u0
ex
s
s
+ u1x
1
s
= u0 erfc
x
2
t
+ u1x.
15. We use
U(x, s) = c1es x + c2esx +u
0s .
The condition limx u(x, t) = u0 implies limx U(x, s) = u0/s, so we define c2 = 0. Then
U(x, s) = c1es x +
u0s
.
The transform of the remaining boundary conditions gives
dU
dx
x=0
= U(0, s).
This condition yields c1 = u0/s(s + 1). Thus
U(x, s) =
u0
es x
s(s + 1)+
u0
sand
u(x, t) = u0
exs
s(
s + 1)
+ u0
1
s
= u0ex+t erfc
t +
x
2
t
u0 erfc
x
2
t
+ u0 By (5) in the table in 15.1.
16. We use
U(x, s) = c1es x + c2e
s x.
The condition limx u(x, t) = 0 implies limx U(x, s) = 0, so we define c2 = 0. Hence
U(x, s) = c1es x.
The remaining boundary condition transforms into
dU
dx
x=0
= U(0, s) 50s
.
This condition gives c1 = 50/s(
s + 1). Therefore
U(x, s) = 50e
s x
s(
s + 1)
and
u(x, t) = 50 ex
s
s(s + 1) = 50ex+t erfc
t +
x
2t + 50 erfcx
2t .17. We use
U(x, s) = c1es x + c2e
s x.
The condition limx u(x, t) = 0 implies limx U(x, s) = 0, so we define c2 = 0. Hence
U(x, s) = c1es x.
The transform of u(0, t) = f(t) is U(0, s) = F(s). Therefore
U(x, s) = F(s)es x
710
7/28/2019 AEMC15
10/37
Exercises 15.2
and
u(x, t) =
F(s)exs
=x
2
t0
f(t )ex2/43/2
d.
18. We use
U(x, s) = c1es x
+ c2e
s x
.
The condition limx u(x, t) = 0 implies limx U(x, s) = 0, so we define c2 = 0. Then U(x, s) = c1esx.
The transform of the remaining boundary condition gives
dU
dx
x=0
= F(s)
where F(s) = {f(t)}. This condition yields c1 = F(s)/s . Thus
U(x, s) = F(s)e
s x
s
.
Using entry (44) of the table and the convolution theorem we obtain
u(x, t) =
F(s) es x
s
=
1
t0
f()ex2/4(t)
t d.
19. Transforming the partial differential equation gives
d2U
dx2 sU = 60.
Using undetermied coefficients we obtain
U(x, s) = c1es x + c2e
s x +
60
s.
The condition limx u(x, t) = 60 implies limx U(x, s) = 60/s, so we define c2 = 0. The transform of the
remaining boundary condition gives
U(0, s) =60
s+
40
se2s.
This condition yields c1 =40
se2s. Thus
U(x, s) =60
s+ 40e2s
es x
s.
Using entry (46) of the table in Appendix III and the second translation theorem we obtain
u(x, t) =
60
s+ 40e2s
es x
s
= 60 + 40erfc
x
2
t 2
(t 2).
20. The solution of the transformed equationd2U
dx2 sU = 100
by undetermined coefficients is
U(x, s) = c1esx + c2e
s x +100
s.
From the fact that limx
U(x, s) = 100/s we see that c1 = 0. Thus
U(x, s) = c2es x +
100
s. (1)
711
7/28/2019 AEMC15
11/37
Exercises 15.2
Now the transform of the boundary condition at x = 0 is
U(0, s) = 20
1
s 1
ses
.
It follows from (1) that
20s 20
ses = c2 + 100
sor c2 = 80
s 20
ses
and so
U(x, s) =
80
s 20
ses
e
sx +
100
s
=100
s 80
se
s x 20
se
s xes.
Thus
u(x, t) = 100
1
s
80
e
s x
s
20
e
s x
ses
= 100 80 erfcx/2t 20 erfcx/2t 1 (t 1).21. Transforming the partial differential equation gives
d2U
dx2 sU = 0
and so
U(x, s) = c1es x + c2e
s x.
The condition limx u(x, t) = 0 implies limx U(x, s) = 0, so we define c1 = 0. The transform of the
remaining boundary condition givesdU
dx x=1=100
s U(1, s).
This condition yields
c2
s es =
100
s c2e
s
from which it follows that
c2 =100
s(
s + 1)e
s.
Thus
U(x, s) = 100e(1x)
s
s(
s + 1).
Using entry (49) of the table in Appendix III we obtain
u(x, t) = 100e(1x)s
s(
s + 1)
= 100
e1x+t erfc
t + 1 xt
+ erfc
1 x2
t
.
22. Transforming the partial differential equation gives
kd2U
dx2 sU = r
s.
Using undetermined coefficients we obtian
U(x, s) = c1es/k x + c2e
s/k x +
r
s2.
712
7/28/2019 AEMC15
12/37
Exercises 15.2
The condition limx
u
x= 0 implies lim
xdU
dx= 0, so we define c2 = 0. The transform of the remaining boundary
condition gives U(0, s) = 0. This condition yields c1 = r/s2. Thus
U(x, s) = r 1s2 es/k x
s2 .Using entries (3) and (46) of the table in Appendix III and the convolution theorem we obtain
u(x, t) = r
1
s2 1
s e
s/k x
s
= rt r
t0
erfc
x
2
k
d.
23. The solution ofd2U
dx2 sU = u0 u0 sin
Lx
is
U(x, s) = c1 cosh(
s x) + c2 sinh(
s x) +u0s
+u0
s + 2/L2sin
Lx.
The transformed boundary conditions U(0, s) = u0/s and U(L, s) = u0/s give, in turn, c1 = 0 and c2 = 0.
Therefore
U(x, s) =u0s
+u0
s + 2/L2sin
Lx
and
u(x, t) = u0
1
s
+ u0
1
s + 2/L2
sin
Lx = u0 + u0e
2t/L2 sin
Lx.
24. The transform of the partial differential equation is
kd2U
dx2 hU + h um
s= sU u0
or
kd2U
dx2 (h + s)U = h um
s u0.
By undetermined coefficients we find
U(x, s) = c1e
(h+s)/k x + c2e
(h+s)/k x +hum + u0s
s(s + h).
The transformed boundary conditions are U(0, s) = 0 and U(L, s) = 0. These conditions imply c1 = 0 and
c2 = 0. By partial fractions we then get
U(x, s) =hum + u0s
s(s + h)=
ums um
s + h+
u0s + h
.
Therefore,
u(x, t) = um
1
s
um
1
s + h
+ u0
1
s + h
= um umeht + u0eht.
25. We use
U(x, s) = c1 cosh
s
kx + c2 sinh
s
kx +
u0s
.
The transformed boundary conditionsdU
dx
x=0
= 0 and U(1, s) = 0 give, in turn, c2 = 0 and
713
7/28/2019 AEMC15
13/37
Exercises 15.2
c1 = u0/s cosh
s/k . Therefore
U(x, s) =u0s u0 cosh
s/k x
s cosh
s/k=
u0s u0 e
s/k x + e
s/k x
s(es/k + e
s/k )
= u0s u0 es/k (x1) + es/k (x+1)
s(1 + e2s/k )
=u0s u0
es/k (1x)
s e
s/k (3x)
s+
es/k (5x)
s
u0
es/k (1+x)
s e
s/k (3+x)
s+
es/k (5+x)
s
=u0s u0
n=0
(1)n
e(2n+1x)s/k
s+
e(2n+1+x)s/k
s
and
u(x, t) = u0
1
s
u0
n=0
(1)n
e(2n+1x)s/k
s
e(2n+1+x)
s/k
s
= u0 u0n=0
(1)n
erfc
2n + 1 x
2
kt
erfc
2n + 1 + x
2
kt
.
26. We use
c(x, s) = c1 cosh
s
Dx + c2 sinh
s
Dx.
The transform of the two boundary conditions are c(0, s) = c0/s and c(1, s) = c0/s. From these conditions we
obtain c1 = c0/s and
c2 = c0(1 coshs/D )/s sinhs/D .Therefore
c(x, s) = c0
cosh
s/D x
s+
(1 cosh
s/D )
s sinh
s/Dsinh
s/D x
= c0
sinh
s/D (1 x)
s sinh
s/D+
sin
s/Dx
s sinh
s/D
= c0
es/D (1x) e
s/D (1x)
s(es/D e
s/D )
+es/Dx e
s/D x
s(es/D e
s/D )
= c0es/D x es/D (2x)
s(1 e2s/D )
+es/D (x1) es/D (x+1)
s(1 e2s/D )
= c0(e
s/Dx e
s/D (2x))
s
1 + e2
s/D + e4
s/D +
+ c0(es/D (x1) e
s/D (x+1)
s
1 + e2
s/D + e4
s/D +
714
7/28/2019 AEMC15
14/37
Exercises 15.2
= c0
n=0
e(2n+x)
s/D
s e
(2n+2x)s/D
s
+ c0
n=0
e(2n+1x)s/D
s e(2n+1+x)
s/D
s and
c(x, t) = c0
n=0
e
(2n+x)D
s
s
e
(2n+2x)D
s
s
+ c0
n=0
e
(2n+1x)D
s
s
e
(2n+1+x)D
s
s
= c0
n=0
erfc
2n + x
2
Dt
erfc
2n + 2 x
2
Dt
+ c0
n=0
erfc
2n + 1 x
2
Dt
erfc
2n + 1 + x
2
Dt
.
Now using erfc(x) = 1 erf(x) we get
c(x, t) = c0
n=0
erf
2n + 2 x
2
Dt
erf
2n + x
2
Dt
+ c0
n=0
erf
2n + 1 + x
2
Dt
erf
2n + 1 x
2
Dt
.
27. We use
U(x, s) = c1eRCs+RGx + c2eRCs+RG + Cu0Cs + G
.
The condition limx u/x = 0 implies limx dU/dx = 0, so we define c2 = 0. Applying U(0, s) = 0 to
U(x, s) = c1eRCsRGx +
Cu0Cs + G
gives c1 = Cu0/(Cs + G). Therefore
U(x, s) = Cu0 eRCs+RGx
Cs + G+
Cu0Cs + G
and
u(x, t) = u0 1s + G/C u0 ex
RCs+G/C
s + G/C = u0e
Gt/C u0eGt/Cerfc
x
RC
2
t
= u0eGt/C
1 erfc
x
2
RC
t
= u0eGt/Cerf
x
2
RC
t
.
715
7/28/2019 AEMC15
15/37
x
cHx,tL
t=0.1
t=0.5
t=1t=2t=5
Exercises 15.2
28. We use
U(x, s) = c1es+hx + c2e
s+hx.
The condition limx u(x, t) = 0 implies limx U(x, s) = 0, so we take c2 = 0. Therefore
U(x, s) = c1es+hx.
The Laplace transform of u(0, t) = u0 is U(0, s) = u0/s and so
U(x, s) = u0e
s+hx
s
and
u(x, t) = u0
e
s+hx
s
= u0
1
se
s+hx
.
From the first translation theorem,e
s+hx
= eht {ex
s} = eht x
2
t3ex
2/4t.
Thus, from the convolution theorem we obtain
u(x, s) =u0x
2
t0
ehx2/4
3/2d.
29. (a) Letting C(x, s) = {c(x, t)} we obtaind2C
dx2 s
kC = 0 subject to
dC
dx
x=0
= A.
The solution of this initial-value problem is
C(x, s) = A
ke(x/
k )s
s
,
so that
c(x, t) = Ak
tex2/4kt.
(b)
(c)
0
c(x, t)dx = Ak erfx
2
kt
0
= Ak(1 0) = Ak
30. (a) We use
U(x, s) = c1e(s/a)x + c2e(s/a)x +
v20F0(a2 v20)s2
e(s/v0)x.
The condition limx u(x, t) = 0 implies limx U(x, s) = 0, so we must define c2 = 0. Consequently
U(x, s) = c1e(s/a)x +
v20 F0(a2 v20 )s2
e(s/v0)x.
716
7/28/2019 AEMC15
16/37
Exercises 15.3
The remaining boundary condition transforms into U(0, s) = 0. From this we find
c1 = v20F0/(a2 v20)s2.
Therefore, by the second translation theorem
U(x, s) = v20F0(a2 v20)s2
e(s/a)x +v20F0
(a2 v20 )s2e(s/v0)x
and
u(x, t) =v20 F0
a2 v20
e(x/v0)s
s2
e(x/a)s
s2
=v20 F0
a2 v20
t x
v0
t x
v0
t xa
t x
a
.
(b) In the case when v0 = a the solution of the transformed equation is
U(x, s) = c1e(s/a)x + c2e(s/a)x
F0
2as
xe(s/a)x.
The usual analysis then leads to c1 = 0 and c2 = 0. Therefore
U(x, s) = F02as
xe(s/a)x
and
u(x, t) = xF02a
e(x/a)s
s
= xF0
2a
t x
a
.
Exercises 15.3
1. From formulas (5) and (6) in the text,
A() =
01
(1)cos xdx +1
0
(2)cos xdx = sin
+ 2sin
=
sin
and
B() =
01
(1) sin xdx +1
0
(2)sin xdx
=1 cos
2 cos 1
=
3(1 cos )
.
Hencef(x) =
1
0
sin cos x + 3(1 cos )sin x
d.
2. From formulas (5) and (6) in the text,
A() =
2
4cos xdx = 4sin2 sin
and
B() =
2
4sin xdx = 4cos cos2
.
717
7/28/2019 AEMC15
17/37
Exercises 15.3
Hence
f(x) =4
0
(sin2 sin )cos x + (cos cos2)sin x
d
=4
0
sin2 cos x cos2 sin x sin cos x + cos sin x
d
=4
0
sin (2 x) sin ( x)
d.
3. From formulas (5) and (6) in the text,
A() =
30
x cos xdx =x sin x
30
1
30
sin xdx
=3sin3
+
cos x
2
30
=3 sin3 + cos 3 1
2
and
B() =3
0
x sin xdx = x cos x
30
+1
30
cos xdx
= 3cos3
+sin x
2
30
=sin3 3 cos3
2.
Hence
f(x) =1
0
(3 sin3 + cos 3 1) cos x + (sin 3 3 cos3)sin x2
d
=1
0
3(sin3 cos x cos3 sin x) + cos 3 cos x + sin 3 sin x cos x2
d
= 1
0
3 sin (3 x) + cos (3 x) cos x2
d.
4. From formulas (5) and (6) in the text,
A() =
f(x)cos xdx
=
0
0 cos xdx +
0
sin x cos xdx +
0 cos xdx
=1
2
0
[sin(1 + )x + sin(1 )x] dx
=1
2
cos(1 + )x
1 + cos(1 )x
1
0
= 12
cos(1 + ) 1
1 + +
cos(1 ) 11
= 12
cos(1 + ) cos(1 + ) + cos(1 ) + cos(1 ) 2
1 2
=1 + cos
1 2 ,
718
7/28/2019 AEMC15
18/37
Exercises 15.3
and
B() =
0
sin x sin xdx =1
2
0
[cos(1 )x cos(1 + )] dx
=1
2 sin(1 )
1
sin(1 + )
1 + =sin
1
2.
Hence
f(x) =1
0
cos x + cos x cos + sin x sin
1 2 d
=1
0
cos x + cos (x )1 2 d.
5. From formula (5) in the text,
A() =
0
ex cos xdx.
Recall {cos kt} = s/(s2 + k2). If we set s = 1 and k = we obtain
A() =1
1 + 2.
Now
B() =
0
ex sin xdx.
Recall {sin kt} = k/(s2 + k2). If we set s = 1 and k = we obtainB() =
1 + 2.
Hence
f(x) =1
0
cos x + sin x
1 + 2d.
6. From formulas (5) and (6) in the text,
A() =11
ex cos xdx
=e(cos + sin ) e1(cos sin )
1 + 2
=2(sinh1) cos 2(cosh 1) sin
1 + 2
and
B() =
11
ex sin xdx
=e(sin cos ) e1( sin cos )
1 + 2
=2(cosh 1) sin 2(sinh 1) cos
1 + 2.
Hence
f(x) =1
0
[A()cos x + B()sin x] d.
7. The function is odd. Thus from formula (11) in the text
B() = 5
10
sin xdx =5(1 cos )
.
719
7/28/2019 AEMC15
19/37
Exercises 15.3
Hence from formula (10) in the text,
f(x) =10
0
(1 cos )sin x
d.
8. The function is even. Thus from formula (9) in the text
A() =
21
cos xdx =
sin2 sin
.
Hence from formula (8) in the text,
f(x) = 2
0
(sin2 sin )cos x
d.
9. The function is even. Thus from formula (9) in the text
A() =
0
x cos xdx =x sin x
0
1
0
sin xdx
=
sin
+
1
2 cos x
0 =
sin + cos
1
2 .
Hence from formula (8) in the text
f(x) =2
0
( sin + cos 1) cos x2
d.
10. The function is odd. Thus from formula (11) in the text
B() =
0
x sin xdx = x cos x
0
+1
0
cos xdx
= cos
+1
2sin x
0
= cos + sin
2.
Hence from formula (10) in the text,
f(x) =2
0
( cos + sin )sin x2
d.
11. The function is odd. Thus from formula (11) in the text
B() =
0
(ex sin x)sin xdx
=1
2
0
ex[cos(1 )x cos(1 + )x] dx
=1
2
0
ex cos(1 )x dx 12
0
ex cos(1 + )x, dx.
Now recall
{cos kt} =
0
est cos ktdt = s/(s2 + k2).
If we set s = 1, and in turn, k = 1 and then k = 1 + , we obtain
B() =1
2
1
1 + (1 )2 1
2
1
1 + (1 + )2=
1
2
(1 + )2 (1 )2[1 + (1 )2][1 + (1 + )2] .
Simplifying the last expression gives
B() =2
4 + 4.
720
7/28/2019 AEMC15
20/37
Exercises 15.3
Hence from formula (10) in the text
f(x) =4
0
sin x
4 + 4d.
12. The function is odd. Thus from formula (11) in the text
B() =
0
xex sin xdx.
Now recall
{t sin kt} = dds
{sin kt} = 2ks/(s2 + k2)2.If we set s = 1 and k = we obtain
B() =2
(1 + 2)2.
Hence from formula (10) in the text
f(x) =4
0
sin x
(1 + 2)2d.
13. For the cosine integral,
A() =
0
ekx cos xdx =k
k2 + 2.
Hence
f(x) =2
0
k cos x
k2 + 2=
2k
0
cos x
k2 + 2d.
For the sine integral,
B() =
0
ekx sin xdx =
k2 + 2.
Hence
f(x) =2
0
sin x
k2 + 2d.
14. From Problem 13 the cosine and sine integral representations of ekx, k > 0, are respectively,
ekx =2k
0
cos x
k2 + 2d and ekx =
2
0
sin x
k2 + 2d.
Hence, the cosine integral representation of f(x) = ex e3x is
ex e3x = 2
0
cos x
1 + 2d 2(3)
0
cos x
9 + 2d =
4
0
3 2(1 + 2) (9 + 2)
cos xd.
The sine integral representation of f is
ex e3x = 2
0
sin x
1 + 2d 2
0
sin x
9 + 2d =
16
0
sin x
(1 + 2) (9 + 2)d.
15. For the cosine integral,
A() =
0
xe2x cos x dx.
But we know
{t cos kt} = dds
s
(s2 + k2)=
(s2 k2)(s2 + k2)2
.
If we set s = 2 and k = we obtain
A() =4 2
(4 + 2)2.
721
7/28/2019 AEMC15
21/37
Exercises 15.3
Hence
f(x) =2
0
(4 2)cos x(4 + 2)2
d.
For the sine integral,
B() = 0
xe2x sin x dx.
From Problem 12, we know
{t sin kt} = 2ks(s2 + k2)2
.
If we set s = 2 and k = we obtain
B() =4
(4 + 2)2.
Hence
f(x) =8
0
sin x
(4 + 2)2d.
16. For the cosine integral,
A() =
0
ex cos x cos xdx
=1
2
0
ex[cos(1 + )x + cos(1 )x] dx
=1
2
1
1 + (1 + )2+
1
2
1
1 + (1 )2
=1
2
1 + (1 )2 + 1 + (1 + )2[1 + (1 + )2][1 + (1 )2]
= 2 + 2
4 + 4.
Hence
f(x) =2
0
(2 + 2)cos x
4 + 4d.
For the sine integral,
B() =
0
ex cos x sin xdx
=1
2
0
ex[sin(1 + )x sin(1 )x] dx
= 12
1 + 1 + (1 + )2
12
1 1 + (1 )2
=1
2
(1 + )[1 + (1 )2] (1 )[1 + (1 + )2]
[1 + (1 + )2][1 + (1 )2]
=3
4 + 4.
Hence
f(x) =2
0
3 sin x
4 + 4d.
722
7/28/2019 AEMC15
22/37
Exercises 15.4
17. By formula (8) in the text
f(x) = 2
0
e cos xd =2
1
1 + x2, x > 0.
18. From the formula for sine integral of f(x) we have
f(x) =2
0
0
f(x)sin xdx
sin xdx
=2
10
1 sin xd +
1
0 sin xd
=2
( cos x)x
10
=2
1 cos xx
.
19. (a) From formula (7) in the text with x = 2, we have
1
2
=2
0
sin cos
d =1
0
sin2
d.
If we let = x we obtain 0
sin2x
xdx =
2.
(b) If we now let 2x = kt where k > 0, then dx = (k/2)dt and the integral in part (a) becomes0
sin kt
kt/2(k/2) dt =
0
sin kt
tdt =
2.
20. With f(x) = e|x|, formula (16) in the text is
C() =
e|x|eixdx =
e|x| cos xdx + i
e|x| sin xdx.
The imaginary part in the last line is zero since the integrand is an odd function of x. Therefore,
C() =
e|x| cos xdx = 2
0
ex cos xdx =2
1 + 2
and so from formula (15) in the text,
f(x) =1
cos x
1 + 2d =
2
0
cos x
1 + 2d.
This is the same result obtained from formulas (8) and (9) in the text.
Exercises 15.4
For the boundary-value problems in this section it is sometimes useful to note that the identities
ei = cos + i sin and ei = cos i sin imply
ei + ei = 2cos and ei ei = 2i sin .
723
7/28/2019 AEMC15
23/37
Exercises 15.4
1. Using the Fourier transform, the partial differential equation becomes
dU
dt+ k2U = 0 and so U(, t) = cek
2t.
Now
{u(x, 0)} = U(, 0) = e|x| .We have
e|x|
=
e|x|eixdx =
e|x|(cos x + i sin x) dx =
e|x| cos x dx.
The integral
e|x| sin xdx = 0
since the integrand is an odd function of x. Continuing we obtain
e|x|
= 2
0
ex cos xdx =2
1 + 2.
But U(, 0) = c =2
1 + 2gives
U(, t) =2ek
2t
1 + 2
and so
u(x, t) =2
2
ek2teix
1 + 2d =
1
ek2t
1 + 2(cos x i sin x)d
=1
ek2t cos x
1 + 2d =
2
0
ek2t cos x
1 + 2d.
2. Since the domain of x is (,) we transform the differential equation using the Fourier transform:k2U(, t) = du
dtdu
dt+ k2U(, t) = 0
U(, t) = cek2t. (1)
The transform of the initial condition is
{u(x, 0)} =
u(x, 0)eixdx =
01
(100eix)dx +1
0
100eixdx
= 1001
ei
i + 100
ei
1
i = 100
ei + ei
2
i
= 1002cos 2
i= 200
cos 1i
.
Thus
U(, 0) = 200cos 1
i,
and since c = U(, 0) in (1) we have
U(, t) = 200cos 1
iek
2t.
724
7/28/2019 AEMC15
24/37
Exercises 15.4
Applying the inverse Fourier transform we obtain
u(x, t) = 1{U(, t)}
=1
2
200cos 1
iek
2teix d
=100
200cos 1
iek
2t(cos x i sin x) d
=100
cos x(cos 1)i
ek2t
odd function
d 100
sin x(cos 1)
ek2t
even function
d
=200
0
sin x(1 cos )
ek2td.
3. Using the Fourier transform, the partial differential equation equation becomes
dUdt
+ k2U = 0 and so U(, t) = cek2t.
Now
{u(x, 0)} = U(, 0) = e2/4
by the given result. This gives c =
e2/4 and so
U(, t) =
e(14+kt)
2
.
Using the given Fourier transform again we obtain
u(x, t) =
1
{e(1+4kt)
2/4
}=
1
1 + 4ktex
2/(1+4kt).
4. (a) We use U(, t) = cek2t. The Fourier transform of the boundary condition is U(, 0) = F(). This gives
c = F() and so U(, t) = F()ek2t. By the convolution theorem and the given result, we obtain
u(x, t) = 1{F() ek2t} = 12
kt
f()e(x)2/4kt d.
(b) Using the definition of f and the solution is part (a) we obtain
u(x, t) =u0
2
kt
11
e(x)2/4ktd.
If u = x 2
kt, then d = 2ktdu and the integral becomes
u(x, t) =u0
(x+1)/2t(x1)/2
kt
eu2
du.
Using the result in Problem 9, Exercises 15.1, we have
u(x, t) =u02
erf
x + 1
2
kt
erf
x 12
kt
.
725
7/28/2019 AEMC15
25/37
-4-2
02
4x
2
4
6
t
0
20
40uHx,tL
4-2
02
-4 -2 2 4t
20
40
60
80
100
u
t=0.05
t=15
Exercises 15.4
5.
Since erf(0) = 0 and limx erf(x) = 1, we have
limt
u(x, t) = 50[erf(0) erf(0)] = 0
and
limx
u(x, t) = 50[erf() erf()] = 50[1 1] = 0.
6. (a) Using the Fourier sine transform, the partial differential equation becomes
dU
dt+ k2U = ku0.
The general solution of this linear equation is
U(, t) = cek2t +
u0
.
But U(, 0) = 0 implies c = u0/ and soU(, t) = u0
1 ek2t
and
u(x, t) =2u0
0
1 ek2t
sin x d.
(b) The solution of part (a) can be written
u(x, t) =2u0
0
sin x
d 2u0
0
sin x
ek
2t d.
Using
0
sin x
d = /2 the last line becomes
u(x, t) = u0 2u0
0
sin x
ek
2t d.
7. Using the Fourier sine transform we find
U(, t) = cek2t.
Now
S{u(x, 0)} = U(, 0) =1
0
sin xdx =1 cos
.
726
7/28/2019 AEMC15
26/37
Exercises 15.4
From this we find c = (1 cos )/ and so
U(, t) =1 cos
ek
2t
and
u(x, t) =
2
0
1
cos
ek2t
sin xd.
8. Since the domain of x is (0,) and the condition at x = 0 involves u/x we use the Fourier cosine transform:
k2U(, t) kux(0, t) = dUdt
dU
dt+ k2U(, t) = kA
U(, t) = cek2t +
A
2.
Since
{u(x, 0)} = U(, 0) = 0
we find c = A/2, so thatU(, t) = A
1 ek2t2
.
Applying the inverse Fourier cosine transform we obtain
u(x, t) = C1{U(, t)} = 2A
0
1 ek2t2
cos xd.
9. Using the Fourier cosine transform we find
U(, t) = cek2t.
Now
C{u(x, 0)} = 10
cos xdx =sin
= U(, 0).
From this we obtain c = (sin )/ and so
U(, t) =sin
ek
2t
and
u(x, t) =2
0
sin
ek
2t cos xd.
10. Using the Fourier sine transform we find
U(, t) = cek2t +
1
.
NowS{u(x, 0)} = S
ex
=
0
ex sin xdx =
1 + 2= U(, 0).
From this we obtain c = /(1 + 2) 1/. Therefore
U(, t) =
1 + 2 1
ek
2t +1
=
1
e
k2t
(1 + 2)
and
u(x, t) =2
0
1
e
k2t
(1 + 2)
sin xd.
727
7/28/2019 AEMC15
27/37
Exercises 15.4
11. (a) Using the Fourier transform we obtain
U(, t) = c1 cos at + c2 sin at.
If we write
{u(x, 0)} = {f(x)} = F()and
{ut(x, 0)} = {g(x)} = G()
we first obtain
c1 = F() from U(, 0) = F() and then c2 = G()/a fromdU
dt
t=0
= G(). Thus
U(, t) = F()cos at +G()
asin at
and
u(x, t) =
1
2
F()cos at + G()a sin at eixd.(b) If g(x) = 0 then c2 = 0 and
u(x, t) =1
2
F()cos ateixd
=1
2
F()
eati + eati
2
eixd
=1
2
1
2
F()ei(xat)d +1
2
F()ei(x+at)d
=
1
2 [f(x at) + f(x + at)] .12. Using the Fourier sine transform we obtain
U(, t) = c1 cos at + c2 sin at.
Now
S{u(x, 0)} =
xex
=
0
xex sin xdx =2
(1 + 2)2= U(, 0).
Also,
S{ut(x, 0)} = dUdt
t=0
= 0.
This last condition gives c2 = 0. Then U(, 0) = 2/(1 + 2
)2
yields c1 = 2/(1 + 2
)2
. Therefore
U(, t) =2
(1 + 2)2cos at
and
u(x, t) =4
0
cos at
(1 + 2)2sin x d.
13. Using the Fourier cosine transform we obtain
U(x, ) = c1 cosh x + c2 sinh x.
728
7/28/2019 AEMC15
28/37
Exercises 15.4
Now the Fourier cosine transforms of u(0, y) = ey and u(, y) = 0 are, respectively, U(0, ) = 1/(1 + 2) andU(, ) = 0. The first of these conditions gives c1 = 1/(1 +
2). The second condition gives
c2 = cosh (1 + 2) sinh
.
Hence
U(x, ) =cosh x
1 + 2 cosh sinh x
(1 + 2)sinh =
sinh cosh cosh sinh x(1 + 2) sinh
=sinh ( x)
(1 + 2) sinh
and
u(x, t) =2
0
sinh ( x)(1 + 2)sinh
cos y d.
14. Since the boundary condition at y = 0 now involves u(x, 0) rather than u(x, 0), we use the Fourier sine
transform. The transform of the partial differential equation is then
d2U
dx2 2U + u(x, 0) = 0 or d
2U
dx2 2U = .
The solution of this differential equation is
U(x, ) = c1 cosh x + c2 sinh x +1
.
The transforms of the boundary conditions at x = 0 and x = in turn imply that c1 = 1/ and
c2 =cosh
sinh 1
sinh +
(1 + 2)sinh .
Hence
U(, x) =1
cosh x
+
cosh
sinh sinh x sinh x
sinh +
sinh x
(1 + 2)sinh
= 1 sinh ( x) sinh sinh x(1 + 2) sinh .Taking the inverse transform it follows that
u(x, y) =2
0
1
sinh ( x)
sinh sinh x
(1 + 2) sinh
sin y d.
15. Using the Fourier cosine transform with respect to x gives
U(, y) = c1ey + c2ey.
Since we expect u(x, y) to be bounded as y we define c2 = 0. Thus
U(, y) = c1ey.
Now
C{u(x, 0)} =1
0
50 cos xdx = 50sin
and so
U(, y) = 50sin
ey
and
u(x, y) =100
0
sin
ey cos xd.
729
7/28/2019 AEMC15
29/37
Exercises 15.4
16. The boundary condition u(0, y) = 0 indicates that we now use the Fourier sine transform. We still have
U(, y) = c1ey, but
S{u(x, 0)} =1
0
50sin xdx = 50(1 cos )/ = U(, 0).
This gives c1 = 50(1 cos )/ and so
U(, y) = 501 cos
ey
and
u(x, y) =100
0
1 cos
ey sin xd.
17. We use the Fourier sine transform with respect to x to obtain
U(, y) = c1 cosh y + c2 sinh y.
The transforms of u(x, 0) = f(x) and u(x, 2) = 0 give, in turn, U(, 0) = F() and U(, 2) = 0. The first
condition gives c1 = F() and the second condition then yields
c2 = F()cosh2sinh2
.
Hence
U(, y) = F() cosh y F()cosh2 sinh ysinh2
= F()sinh2 cosh y cosh2 sinh y
sinh2
= F()sinh (2 y)
sinh2
and
u(x, y) =2
0
F()sinh (2 y)
sinh2sin x d.
18. The domain of y and the boundary condition at y = 0 suggest that we use a Fourier cosine transform. The
transformed equation is
d2U
dx2 2U uy(x, 0) = 0 or d
2U
dx2 2U = 0.
Because the domain of the variable x is a finite interval we choose to write the general solution of the latter
equation as
U(x, ) = c1 cosh x + c2 sinh x.
Now U(0, ) = F(), where F() is the Fourier cosine transform of f(y), and U(, ) = 0 imply c1 = F()
and c2 = F() sinh / cosh . Thus
U(x, ) = F() cosh x F() sinh cosh
sinh x = F()cosh ( x)
cosh .
Using the inverse transform we find that a solution to the problem is
u(x, y) =2
0
F()cosh ( x)
cosh cos y d.
730
7/28/2019 AEMC15
30/37
Exercises 15.4
19. We solve two boundary-value problems:
Using the Fourier sine transform with respect to y gives
u1(x, y) =2
0
ex
1 + 2sin y d.
The Fourier sine transform with respect to x yields the solution to the second problem:
u2(x, y) =2
0
ey
1 + 2sin xd.
We define the solution of the original problem to be
u(x, y) = u1(x, y) + u2(x, y) = 2
0
1 + 2
ex sin y + ey sin x d.
20. We solve the three boundary-value problems:
Using separation of variables we find the solution of the first problem is
u1(x, y) =n=1
Aneny sin nx where An = 2
0
f(x)sin nxdx.
Using the Fourier sine transform with respect to y gives the solution of the second problem:
u2(x, y) =200
0
(1 cos )sinh ( x) sinh
sin y d.
Also, the Fourier sine transform with respect to y gives the solution of the third problem:
u3(x, y) =2
0
sinh x
(1 + 2) sinh sin y d.
The solution of the original problem is
u(x, y) = u1(x, y) + u2(x, y) + u3(x, y).
21. Using the Fourier transform with respect to x gives
U(, y) = c1 cosh y + c2 sinh y.
The transform of the boundary conditionu
y
y=0
= 0 isdU
dy
y=0
= 0. This condition gives c2 = 0. Hence
U(, y) = c1 cosh y.
Now by the given information the transform of the boundary condition u(x, 1) = ex2
is
731
7/28/2019 AEMC15
31/37
Exercises 15.4
U(, 1) =
e2/4. This condition then gives c1 =
e
2/4 cosh . Therefore
U(, y) =
e
2/4 cosh y
cosh
and
U(x, y) = 12
e2
/4 cosh ycosh
eixd
=1
2
e2/4 cosh y
cosh cos xd
=1
0
e2/4 cosh y
cosh cos xd.
22. Entries 42 and 43 of the table in Appendix III imply0
estsin at
tdt = arctan
a
s
and 0
estsin at cos bt
tdt =
1
2arctan
a + b
s+
1
2arctan
a bs
.
Identifying = t, x = a, and y = s, the solution of Problem 16 is
u(x, y) =100
0
1 cos
ey sin xd
=100
0
sin x
eyd
0
sin x cos
eyd
=100
arctanx
y 1
2arctan
x + 1
y 1
2arctan
x 1y .
Exercises 15.5
1. We show that 14 F4F4 = I:
1
4F4F4 =
1
4
1 1 1 1
1 i 1 i1 1 1 11 i 1 i
1 1 1 1
1 i 1 i1 1 1 11 i 1 i
=1
4
4 0 0 0
0 4 0 0
0 0 4 0
0 0 0 4
= I.
Thus F14 =14
F4.
2. We have
f(x)(x a)dx = 12
a+a
f(x)dx =1
2f(c)(2) = f(c)
by the mean value theorem for integrals.
3. By the sifting property,
{(x)} =
(x)eixdx = ei0 = 1.
732
7/28/2019 AEMC15
32/37
Exercises 15.5
4. We already know that f = f. Then, by the sifting property,
(f )(x) =
f()(x )d =
f()( x)d = f(x).
5. Using integration by parts with u = f(x) and dv = (x
a) we find
f(x)(x a)dx =
f(x)(x a)dx = f(a)
by the sifting property.
6. Using a CAS we find
{g(x)} = 12
[sign(A ) + sign(A + )]where sign(t) = 1 if t > 0 and sign t = 1 if t < 0. Thus
{g(x)} =
1, A < < A0, elsewhere.
7. Using
8 =
2
2+ i
2
2
28 = i
38 =
2
2+ i
2
2
48 = 1
58 =
2
2 i
2
2
68 = i
78 =
2
2 i
2
2
88 = 1
we have
F8 =
1 1 1 1 1 1 1 1
1 22 + i2
2i 22 + i22 1 22 i22 i 22 i22
1 i 1 i 1 i 1 i1
2
2 + i
22 i
2
2 + i
22 1
2
2 i
22
i
22 i
2
2
1 1 1 1 1 1 1 11
2
2 i
22
i
22 i
2
2 1
22 + i
2
2 i
22 + i
2
2
1 i 1 i 1 i 1 i1
2
2 i
22 i
2
2 i
22 1
2
2 + i
22
i
22 + i
2
2
.
In factored form
F8 = I4 D4I4 D4
F4 00 F4
P,where I4 is the 4 4 identity matrix.
D4 =
1 0 0 0
0
2/2 + i
2/2 0 0
0 0 i 0
0 0 0 2/2 + i2/2
,
and P is the 8 8 matrix with 1 in positions (1, 1), (2, 3), (3, 5), (4, 7), (5, 2), (6, 4), (7, 6), and (8, 8).
733
7/28/2019 AEMC15
33/37
-1 -0.5 0.5 1 x
-1
-0.5
0.5
1
y
-2 -1 1 2
0.2
0.4
0.6
0.8
Exercises 15.5
8. The 8th roots of unity, 18, 28, . . .,
88 are shown in the solution
of Problem 7 above. The points in the complex plane are equally
spaced on the perimeter of the unit circle.
9. The Fourier transform of g(x) = (sin 2x)/x is
G() =
1, 2 < < 20, elsewhere.
This implies that (f g)(x) = 1{F()G()} is band-limited.The graph ofF()G(), which is identical to the graph of (f g), is shown.
10. For N = 6,
F6 =
1 1 1 1 1 1
1 1/2 +
3i/2 1/2 + 3i/2 1 1/2 3i/2 1/2 3i/21 1/2 + 3i/2 1/2 3i/2 1 1/2 + 3i/2 1/2 3i/21 1 1 1 1 11 1/2 3i/2 1/2 + 3i/2 1 1/2 3i/2 1/2 + 3i/21 1/2 3i/2 1/2 3i/2 1 1/2 + 3i/2 1/2 + 3i/2
.
If, for example, f = (2, 0, 1, 6, 2, 3), then
c =1
6F6f =
7/3
2/3 +
3i/3
5/6 3i/62/3
5/6 3i/62/3 i/3
.
Chapter 5 Review Exercises
1. The partial differential equation and the boundary conditions indicate that the Fourier cosine transform is
appropriate for the problem. We find in this case
u(x, y) =2
0
sinh y
(1 + 2)cosh cos xd.
2. We use the Laplace transform and undetermined coefficients to obtain
U(x, s) = c1 cosh
s x + c2 sinh
s x +50
s + 42sin2x.
The transformed boundary conditions U(0, s) = 0 and U(1, s) = 0 give, in turn, c1 = 0 and c2 = 0. Hence
U(x, s) =50
s + 42sin2x
734
7/28/2019 AEMC15
34/37
Chapter 5 Review Exercises
and
u(x, t) = 50 sin 2x
1
s + 42
= 50e4
2t sin2x.
3. The Laplace transform gives
U(x, s) = c1e
s+hx
+ c2e
s+hx
+
u0
s + h .The condition limx u/x = 0 implies limx dU/dx = 0 and so we define c2 = 0. Thus
U(x, s) = c1es+hx +
u0s + h
.
The condition U(0, s) = 0 then gives c1 = u0/(s + h) and so
U(x, s) =u0
s + h u0 e
s+hx
s + h.
With the help of the first translation theorem we then obtain
u(x, t) = u0 1
s + h u0
e
s+hx
s + h
= u0e
ht u0eht erfcx
2
t= u0e
ht
1 erfc
x
2
t
= u0e
hterf
x
2
t
.
4. Using the Fourier transform and the result
e|x|
= 1/(1 + 2) we find
u(x, t) =1
2
1 e2t2(1 + 2)
eixd
=1
2
1 e2t2(1 + 2)
cos xd
=1
0
1 e2t
2
(1 + 2
)
cos xd.
5. The Laplace transform gives
U(x, s) = c1es x + c2e
s x.
The condition limx u(x, t) = 0 implies limx U(x, s) = 0 and so we define c2 = 0. Thus
U(x, s) = c1es x.
The transform of the remaining boundary condition is U(0, s) = 1/s2. This gives c1 = 1/s2. Hence
U(x, s) =e
s x
s2and u(x, t) =
1
s
es x
s
.
Using 1
s
= 1 and
es x
s
= erfc
x
2
t
,
it follows from the convolution theorem that
u(x, t) =
t0
erfc
x
2
d.
6. The Laplace transform and undetermined coefficients gives
U(x, s) = c1 cosh sx + c2 sinh sx +s 1
s2 + 2sin x.
735
7/28/2019 AEMC15
35/37
Chapter 5 Review Exercises
The conditions U(0, s) = 0 and U(1, s) = 0 give, in turn, c1 = 0 and c2 = 0. Thus
U(x, s) =s 1
s2 + 2sin x
and
u(x, t) = sin x ss2 + 2 1 sin x s2 + 2= (sin x)cos t 1
(sin x)sin t.
7. The Fourier transform gives the solution
u(x, t) =u02
ei 1
i
eixek
2td
=u02
ei(x) eixi
ek2td
=u02
cos ( x) + i sin ( x) cos x + i sin xi
ek2td.
Since the imaginary part of the integrand of the last integral is an odd function of , we obtain
u(x, t) =u02
sin ( x) + sin x
ek2td.
8. Using the Fourier cosine transform we obtain
U(x, ) = c1 cosh x + c2 sinh x.
The condition U(0, ) = 0 gives c1 = 0. Thus
U(x, ) = c2 sinh x.
Now
C{u(, y)} = 2
1 cos ydy =
sin2
sin
= U(, ).
This last condition gives c2 = (sin 2 sin )/ sinh . Hence
U(x, ) =sin2 sin
sinh sinh x
and
u(x, y) =2
0
sin2 sin sinh
sinh x cos y d.
9. We solve the two problems
2u1x2
+2u1y2
= 0, x > 0, y > 0,
u1(0, y) = 0, y > 0,
u1(x, 0) =
100, 0 < x < 1
0, x > 1
and2u2x2
+2u2y2
= 0, x > 0, y > 0,
u2(0, y) =
50, 0 < y < 1
0, y > 1
u2(x, 0) = 0.
736
7/28/2019 AEMC15
36/37
Chapter 5 Review Exercises
Using the Fourier sine transform with respect to x we find
u1(x, y) =200
0
1 cos
ey sin x d.
Using the Fourier sine transform with respect to y we find
u2(x, y) =100
0
1 cos
ex sin y d.
The solution of the problem is then
u(x, y) = u1(x, y) + u2(x, y).
10. The Laplace transform gives
U(x, s) = c1 cosh
s x + c2 sinh
s x +r
s2.
The conditionu
x
x=0
= 0 transforms intodU
dx
x=0
= 0. This gives c2 = 0. The remaining condition u(1, t) = 0
transforms into U(1, s) = 0. This condition then implies c1 = r/s2 coshs . Hence
U(x, s) =r
s2 r cosh
s x
s2 cosh
s.
Using geometric series and the convolution theorem we obtain
u(x, t) = r
1
s2
r
cosh
s x
s2 cosh
s
= rt rn=0
(1)nt
0
erfc
2n + 1 x
2
d +
t0
erfc
2n + 1 + x
2
d
.
11. The Fourier sine transform with respect to x and undetermined coefficients give
U(, y) = c1 cosh y + c2 sinh y +A
.
The transforms of the boundary conditions are
dU
dy
y=0
= 0 anddU
dy
y=
=B
1 + 2.
The first of these conditions gives c2 = 0 and so
U(, y) = c1 cosh y +A
.
The second transformed boundary condition yields c1 = B/(1 +
2
) sinh . Therefore
U(, y) =B cosh y
(1 + 2) sinh +
A
and
u(x, y) =2
0
B cosh y
(1 + 2) sinh +
A
sin xd.
12. Using the Laplace transform gives
U(x, s) = c1 cosh
s x + c2 sinh
s x.
737
7/28/2019 AEMC15
37/37
Chapter 5 Review Exercises
The condition u(0, t) = u0 transforms into U(0, s) = u0/s. This gives c1 = u0/s. The condition u(1, t) = u0
transforms into U(1, s) = u0/s. This implies that c2 = u0(1 cosh
s )/s sinh
s . Hence
U(x, s) =u0s
cosh
s x + u0
1 coshs
s sinh
s
sinh
s x
= u0 sinhs coshs x cosh sinhs sinhs x + sinhs x
s sinh
s
= u0
sinh
s (1 x) + sinhs x
s sinh
s
= u0
sinh
s (1 x)
s sinh
s+
sinh
s x
s sinh
s
and
u(x, t) = u0
sinh
s (1 x)
s sinh
s
+
sinh
s x
s sinh
s
= u0
n=0
erf2n + 2 x2t erf2n + x2t + u0
n=0
erf
2n + 1 + x
2
t
erf
2n + 1 x
2
t
.
13. Using the Fourier transform gives
U(, t) = c1ek2t.
Now
u(, 0) =
0
exeix dx =e(i1)x
i 1
0
= 0 1i 1 =
1
1 i = c1so
U(, t) =
1 + i
1 + 2 ek2t
and
u(x, t) =1
2
1 + i
1 + 2ek
2teix d.
Since1 + i
1 + 2(cos x i sin x) = cos x + sin x
1 + 2+
i( cos x sin x)1 + 2
and the integral of the product of the second term with ek2t is 0 (it is an odd function), we have
u(x, t) =1
2
cos x + sin x
1 + 2ek
2t d.