CHAPTER 4

Section 4.1
1. (a) Using integration by parts twice, the integral can be written as
Z Z Z
2 2 2
x sin x dx = −x cos x + 2x cos x dx = −x cos x + 2x sin x − 2 sin x dx

= −x2 cos x + 2x sin x + 2 cos x + C = 2x sin x − x2 − 2 cos x + C

(b) Making the substitution u = x2 so that du = 2xdx, the integral can be written as
Z Z Z
x 1 2x 1 1 1 1
4
dx = 2 dx = 2
du = tan−1 u+C = tan−1 x2 +C
1+x 2 1 + (x2 ) 2 1+u 2 2

(c) Using partial fraction expansion, we can write

Z Z   Z Z
1 1 1 dx dx
dx = − + dx = − +
(x − 1) (x − 2) x−1 x−2 x−1 x−2
x−2
= − ln (x − 1) + ln (x − 2) + C = ln +C
x−1
√
(d) Making the substitution u = x − 1 so that 2udu = dx, the integral can be
written as
Z  
−1 + 1 + u
Z Z Z
1 u 1
√ dx = 2 du = 2 du = 2 − + 1 du
1+ x−1 1+u 1+u 1+u
Z Z
du
= −2 + 2 du = −2 ln (1 + u) + 2u + C
1+u
√ √

= 2 x − 1 − ln 1 + x − 1 + C

2. (a) Making the substitution x = sin θ so that dx = cos θdθ and using the identity
sin2 θ + cos2 θ = 1, the integral can be written as
Z 1√ Z 1 π/2 π/2
1 1
Z
2 2 θ sin 2θ
1 − x dx = cos θ dθ = (1 + cos 2θ) dθ = +
0 0 2 0 2 0 4 0
π/2 π/2
θ cos θ sin θ π
= + =
2 0 2 0 4

(b) Using the identity sin mx sin nx = (1/2) cos[(m − n)x] − (1/2) cos[(m + n)x], the
integral can be written as
Z π
1 π 1 π 1 π
Z Z Z
sin 2x sin 3x dx = (cos x − cos 5x) dx = cos x dx − cos 5x dx
0 2 0 2 0 2 0
π π
sin x sin 5x
= − =0
2 0 10 0

1
 (c) Using integration by parts twice, the integral can be written as
Z 1 Z 1
2

x 2

x1
2x − 3x + 1 e dx = 2x − 3x + 1 e 0 − (4x − 3) ex dx
0 0
Z 1
2 x 1 x 1
4ex dx


= 2x − 3x + 1 e 0 − (4x − 3) e 0 +
0

x1 x 1 1
= 2x − 3x + 1 e 0 − (4x − 3) e 0 + 4ex 0 = 3e − 8
2

(d) Using integration by parts, the fact that (d/dx) tan−1 x = 1/(1 + x2 ) and making
the substitution u = x2 so that du = 2xdx, the integral can be written as
Z 1 Z 1
1 1 2x
Z
−1 −1 1 x −1 1
tan x dx = x tan x 0 − 2
dx = x tan x 0 − dx
0 0 1+x 2 0 1 + x2
1
1 1 du
Z
−1 1 1 ln (1 + u)
= x tan x 0 − = x tan−1 x 0 −
2 0 1+u 2 0
1
1 ln (1 + x2 ) π 1
= x tan−1 x 0 − = + ln √
2 0 4 2

3. (a) Making the substitution x = sin θ so that dx = cos θdθ and using the identity
sin2 θ + cos2 θ = 1, the integral can be written as
Z 1 Z 1 Z π/2 Z π/2
dx dx cos θ
√ = lim √ = lim p dθ = lim+ dθ
0 1 − x2 b→0+ b 1 − x2 b→0+ b 1 − sin2 θ b→0 b
π  π
π/2
= lim+ θ b = lim+ −b =
b→0 b→0 2 2
(b) Making the substitution u = −x so that du = −dx, the integral can be written
as
Z ∞ Z b Z b
−x −x b
eu du = lim −eu 0 = lim −eb + 1 = 1


e dx = lim e dx = lim −
0 b→∞ 0 b→−∞ 0 b→−∞ b→−∞

(c) Using integration by parts, the integral can be written as

Z 1 Z 1 Z 1
1 1
ln x dx = lim+ ln x dx = lim+ x ln x b − lim+ dx = lim+ (x ln x − x) b
0 b→0 b b→0 b→0 b b→0

= lim+ (−1 − b ln b + b) = −1 − lim+ b ln b = −1

b→0 b→0

where the last step follows from the fact that

ln b LH 1/b
lim+ b ln b = lim+ = lim+ = lim+ −b = 0
b→0 b→0 1/b b→0 −1/b2 b→0

using L’Hopital’s rule.

2
(d) Making the substitutions x = tan θ so that dx = sec2 θdθ, 2u = θ so that 2du = dθ,
v = cos u so that dv = − sin udu and w = sin u so that dw = cos udu and the
identities 1 + tan2 θ = sec2 θ and sin 2θ = 2 sin θ cos θ, the integral can be written
as
Z ∞ Z b Z b
dx dx sec2 θ
√ = lim √ = lim √ dθ
1 x 1 + x2 b→∞ 1 x 1 + x2 b→π/2 π/4 tan θ 1 + tan2 θ
Z b Z b Z b
dθ 2du du
= lim = lim = lim
b→π/2 π/4 sin θ b→π/4 π/8 sin 2u b→π/4 π/8 sin u cos u
Z b
sin2 u + cos2 u
= lim du
b→π/4 π/8 sin u cos u
Z b Z b
sin u cos u
= lim du + lim du
b→π/4 π/8 cos u b→π/4 π/8 sin u
Z b Z b
dv dw
= lim √ √ √ − + lim √ √ √
b→ 2/2 2+ 2/2 v b→ 2/2 2− 2/2 w
b√
+ lim ln w √ √
b
= lim − ln v
√
√
2+ 2/2 √ 2− 2/2
b→ 2/2 b→ 2/2
p √ p √
2+ 2 2− 2
= lim√ − ln b + ln + lim
√ ln b − ln
b→ 2/2 2 b→ 2/2 2
p √
2+ 2 1  √  1  √ 2  √ 
= ln p √ = ln 3 + 2 2 = ln 1 + 2 = ln 1 + 2
2− 2 2 2

(e) Using integration by parts twice and making the substitution u = −x so that
du = −dx, the integral can be written as
Z ∞ Z b Z b
2 −x 2 −x
x e dx = lim x e dx = lim −u2 eu du
0 b→∞ 0 b→−∞ 0
Z b
2 u b
= lim −u e 0 + lim 2ueu du
b→−∞ b→−∞ 0
Z b
2 u b u b
= lim −u e 0 + lim 2ue 0 − lim 2eu du
b→−∞ b→−∞ b→−∞ 0
b b b
= lim −u2 eu 0 + lim 2ueu 0
− lim 2eu 0
b→−∞ b→−∞ b→−∞
2 b b b


= lim −b e + 2be − 2e + 2 = 2
b→−∞

where the last step follows from employing L’Hopital’s rule:

b2 LH 2b LH 2
lim −b2 eb = lim − −b
= lim −b = lim − −b = 0
b→−∞ b→−∞ e b→−∞ e b→−∞ e

3
 (f) Using integration by parts, the integral can be written as
Z ∞ Z b b Z b
ln x ln x ln x dx
2
dx = lim 2
dx = lim − + lim
1 x b→∞ 1 x b→∞ x 1 b→∞ 1 x2
b b  
ln x 1 ln b 1
= lim − − lim = lim − − +1 =1
b→∞ x 1
b→∞ x
1
b→∞ b b
where the last step follows from employing L’Hopital’s rule:
ln b LH 1/b 1
lim − = lim − = lim − = 0
b→∞ b b→∞ 1 b→∞ b
4. (a)
Z 1 Z 0 Z 1 Z b Z 1
dx dx dx dx dx
= + = lim + lim
−1 x1/3 −1 x1/3 0 x1/3 b→0− −1 x1/3 b→0+ b x1/3
b 1
3 3
= lim− x2/3 + lim+ x2/3
b→0 2 −1 b→0 2 b
3 2/3 3
1 − b2/3 = 0



= lim− b − 1 + lim+
b→0 2 b→0 2

(b)
Z 1 Z 0 Z 1 Z b Z 1
dx dx dx dx dx
= + = lim− + lim
−1 x3 −1 x3 0 x3 b→0 −1 x3 b→0+ b x3
b 1
1 1
= lim− − 4 + lim+ 4
b→0 4x −1 b→0 4x b
1 1
−b−4 + 1 + lim+ 1 − b−4 = −∞



= lim−
b→0 4 b→0 4

Hence, the integral is divergent.

(c) Making the substitution x = tan θ so that dx = sec2 θdθ, the integral can be
written as
Z ∞ Z b Z b Z b
dx dx sec2 θ b
2
= lim 2
= lim 2
dθ = lim dθ = lim θ 0
0 1 + x b→∞ 0 1 + x b→π/2 0 1 + tan θ b→π/2 0 b→π/2
π
= lim b =
b→π/2 2
(d) Using a partial fraction expansion, the integral can be written as
Z ∞ 2 Z 1 2 Z b 2
x −x−1 x −x−1 x −x−1
3
dx = lim+ 3
dx + lim dx
0 x (x + 1) b→0 b x (x + 1) b→∞ 1 x (x3 + 1)
Z 1 
4x − 2 1 1
= lim+ − − dx + . . .
b→0 b 3 (x2 − x + 1) 3 (x + 1) x
1 1 4x − 2 1 1 dx
Z Z Z 1
dx
= lim+ dx − lim − lim + ...
b→0 3 b x2 − x + 1 b→0+ 3 b x + 1 b→0+ b x

4
 It is clear to see that the first two integrals (obtained by a partial fraction expan-
sion) belonging to the first partial integral converge. However, the third diverges:
Z 1
dx 1
lim+ = lim+ ln x b = − lim+ ln b = ∞
b→0 b x b→0 b→0

Hence, since the first partial integral diverges, we conclude that the original inte-
gral is divergent.
(e)
Z ∞ Z b
b
sin x dx = lim sin x dx = lim − cos x 0
= lim (− cos b + 1)
0 b→∞ 0 b→∞ b→∞

Since limb→∞ cos b does not exist the integral is divergent.

(f) Making the substitution u = cosh x so that du = sinh xdx, the integral can be
written as
Z ∞ Z b Z b Z b
(1 − tanh x) dx = lim (1 − tanh x) dx = lim dx − lim tanh x dx
0 b→∞ 0 b→∞ 0 b→∞ 0
Z b
b sinh x
= lim x 0 − lim dx
b→∞ b→∞ 0 cosh x
Z cosh b
b du
= lim x 0 − lim
b→∞ b→∞ 1 u
b cosh b
= lim x 0 − lim ln u 1
b→∞ b→∞
= lim (b − ln cosh b) = ln 2
b→∞

where the last step follows from the fact that

lim (b − ln cosh b) = lim ln eb − ln eb + e−b + ln 2 = lim ln 2 − ln 1 + e−2b

 

b→∞ b→∞ b→∞
= ln 2

5. (a) The curves y = 0, y = 1 − x2 intersect at the point (−1, 0), (1, 0). Hence, the
area between the curves is given by
Z 1 1
x3

2

4
A= 1 − x dx = x − =
−1 3 −1 3

(b) The curves y = x3 , y = x1/3 intersect at the points (−1, −1), (0, 0), (1, 1). Hence
the area between the curves is given by
Z 1  4/3 1
1/3 3

3x x4
A=2 x − x dx = − =1
0 2 2 0
Note that we have used the fact that the intersection of the two curves is anti-
symmetric with respect to the y-axis, and so in order to calculate the total area
we can simply integrate from x = 0 to x = 1 and multiply the result by two.

5
 (c) The curves y = 6 sin−1 x, y = π sin πx intersect at the points (−1/2, −π), (0, 0), (1/2, π).
Hence, the area between the curves is given by
Z 1/2
π sin πx − 6 sin−1 x dx


A=2
0
Z 1/2 Z 1/2
= 2π sin πx dx − 12 sin−1 x dx
0 0
Z 1/2
1/2 −1 1/2 12x
= −2 cos πx 0 − 12x sin x 0 + √ dx
0 1 − x2
Z 3/4
du
=2−π−6 √
1 u
Z 3/4
=2−π−6 u−1/2 du
1
√ 3/4 √
= 2 − π − 12 u 1 = 14 − π − 6 3

where we have used integration by parts and the substitution u = 1 − x2 so that

du = −2xdx in order to solve the second integral. Note that we have used the
fact that the intersection of the two curves is anti-symmetric with respect to the
y-axis, and so in order to calculate the total area we can simply integrate from
x = 0 to x = 1/2 and multiply the result by two.

6. (a)
Z b Z π/2 π/2
1 2 2 cos x 2
f (x) dx = sin x dx = − =
b−a a π 0 π 0 π
(b)
Z b Z 0 0 −π/2
1 2 2 cos x 2 cos x 2
f (x) dx = sin x dx = − = =−
b−a a π −π/2 π −π/2 π 0 π

(c) Using the identities sin2 x + cos2 x = 1, cos 2x = 2 cos2 x − 1, we find

b π/2
2 π/2
Z Z Z
1 2 2
1 − cos2 x dx


f (x) dx = sin x dx =
b−a a π 0 π 0
1 π/2
Z
= (1 − cos 2x) dx
π 0
 π/2
1 sin 2x 1
= x− =
π 2 0 2

6
 (d)
Z b Z x2  2 x 2
1 1 1 ax
f (x) dx = (ax + b) dx = + bx
b−a a x2 − x1 x1 x2 − x1 2 x1
a
= b + (x1 + x2 )
2
7. Let f (x) and g(x) be continuous for a ≤ x ≤ b and |g(x) − f (x)| ≤  for a ≤ x ≤ b.
Defining h(x) = g(x) − f (x) so that |h(x)| ≤  and using (4.6) we then find
Z b Z b Z b Z b
h(x) dx = [g(x) − f (x)] dx = g(x) dx − f (x) dx ≤  (b − a)
a a a a

8. (a)
1 1 1
x6
 3
x7
Z Z  
2 2 x 13
sin x dx u x − dx = − = u 0.3095
0 0 6 3 42 0 42
The worst error is approximately 0.0081 at the point x = 1.
(b)
1 1 1
x4 x3 x5
Z Z   
−x2 2 23
e dx u 1−x + dx = x − + = u 0.7667
0 0 2 3 10 0 30
The worst error is approximately 0.1321 at the point x = 1.
9. Let f (x) be continuous for 0 ≤ x ≤ 1. Then (4.20) may be used to approximate the
integral of f (x) numerically:
Z 1
1
f (x) dx ∼ [f (0) + 2f (x1 ) + 2f (x2 ) + · · · + 2f (xn−1 ) + f (1)]
0 2n
where 0 < x1 < x2 · · · < xn−1 < 1. If we then let n → ∞ and choose x1 = 1/n, x2 =
2/n, . . . , xn−1 = (n − 1)/n, xn = n/n such that the endpoints converge to x = 0 and
x = 1 respectively, while at the same time choosing an infinite number of equally
spaced, but infinitely close interior points x1 , x2 , . . . , xn−1 the finite sum converges to:
1
lim [f (0) + 2f (x1 ) + 2f (x2 ) + · · · + 2f (xn−1 ) + f (1)] =
n→∞ 2n
1
  
1
 
2

n−1
  n  Z 1
lim f +f + ··· + f +f = f (x) dx
n→∞ n n n n n 0

Note that the end points of the first and second limits differ by a factor of 1/2. However,
since 2∞ = ∞ this difference is of no importance.
10. (a)
1 1
x2
 
1 + 2 + ··· + n n−1 n
Z
1 1 2 1
lim 2
= lim + + ··· + + = x dx = =
n→∞ n n→∞ n n n n n 0 2 0 2

7
 (b)
"  2   #
2  2
12 + 22 + · · · + n2

1 1 2 n−1 n 2
lim = lim + + · · · + +
n→∞ n3 n→∞ n n n n n
Z 1 1
x3 1
= x2 dx = =
0 3 0 3

(c) Provided that P ≥ 0

"  P   #
P P P P  P 
1 + 2 + ··· + n 1 1 2 n−1 n P
lim = lim + + · · · + +
n→∞ nP +1 n→∞ n n n n n
Z 1 1
xP +1 1
= xP dx = =
0 P +1 0 P +1

(d) Taking the natural log of both sides of the equation gives ln(4/e) = ln 4 − 1 and
 
1 1/n
ln lim [(n + 1) (n + 2) . . . (2n)] =L
n→∞ n

Then, manipulating the left-hand side further, we find

   
1 1/n 1 1
L = lim ln [(n + 1) (n + 2) . . . (2n)] = lim ln n (n + 1) (n + 2) . . . (2n)
n→∞ n n→∞ n n
 
1 n+1n+2 2n
= lim ln + ··· +
n→∞ n n n n
 
1 n+1 n+2 n+n
= lim ln + ln + · · · + ln
n→∞ n n n n
       
1 1 2 n−1  n
= lim ln 1 + + ln 1 + + · · · + ln 1 + + ln 1 +
n→∞ n n n n n
Z 1 Z 1 Z 2
1 x u−1
= ln (1 + x) dx = x ln (1 + x) 0 − dx = ln 2 − du
0 0 1+x 1 u
Z 2 
1 4
= ln 2 − 1− du = ln 2 − [u − ln u]21 = ln 4 − 1 = ln
1 u e

where we have used integration by parts to solve the first integral and the substi-
tution u = 1 + x so that du = dx to solve the second integral.

11. Let f (x) be a continuous function for a ≤ x ≤ b and let it be a given fact that
Z b1
f (x) dx = 0
a1

8
 for every interval a1 ≤ x ≤ b1 contained in the interval a ≤ x ≤ b. Next, let us choose
a fixed point x0 such that a1 ≤ x0 , x0 + δ ≤ b1 , where δ > 0. Then by (4.13) we find
Z x0 +δ
f (x) dx = f (x∗ ) δ = 0 for x0 ≤ x∗ ≤ x0 + δ
x0

Now if we let δ → 0 we get

Z x0 +δ
1
lim f (x) dx = f (x0 ) = 0
δ→0 δ x0

As x0 was chosen arbitrarily within a1 ≤ x ≤ b1 , we conclude that f (x) ≡ 0.

12. Let f (x) be a continuous function for a ≤ x ≤ b, f (x) ≥ 0 on the interval and
Z b
f (x) dx = 0
a
Rc Rb
Next, let c be such that a < c < b. Then the integrals a f (x) dx, c f (x) dx are either
positive or zero. However, since their sum must be zero, the only option is that in fact
they both are zero. The interval of each partial integral thus obtained can in turn be
subdivided into smaller intervals over which to individually integrate f (x) and since
again f (x) is either positive or zero on this new sub interval, but the total integral
over a ≤ x ≤ b must be zero, we conclude that each partial integral must be zero
over the relevant sub interval. We can continue to apply this argument indefinitely
for every
R b1 smaller sub interval obtained from a larger sub interval and so we conclude
that a1 f (x) dx = 0 for every choice a1 , b1 on the interval a ≤ x ≤ b. Hence, by
Problem 11, f (x) ≡ 0.

Section 4.2
Rx
1. (a) Let f (x) = x and F (x) = 0
x dx. Then
R x+1 Rx
x x x
t dt 0
t dt = F (x)
0 0 0.5 0.0
1 1 1.5 0.5
2 2 2.5 2.0
3 3 3.5 4.5
4 4 4.5 8.0
5 5 5.5 12.5
6 6 6.5 18.0
7 7 7.5 24.5
8 8 8.5 32.0
9 9 9.5 40.5
10 10 50.0

9
 2 R1 2
(b) Let f (x) = e−x and F (x) = 0
e−x dx. Then
2 R x+0.1 −t2 R x −t2
x e−x x
e dt 0
e dt = F (x)
0 1.0 0.100 0.00
0.1 0.99 0.098 0.100
0.2 0.96 0.094 0.197
0.3 0.91 0.088 0.291
0.4 0.85 0.082 0.379
0.5 0.78 0.074 0.460
0.6 0.70 0.066 0.534
0.7 0.61 0.057 0.600
0.8 0.53 0.049 0.657
0.9 0.44 0.041 0.705
1.0 0.37 0.746
R1
(c) Let f (x) = cos x and F (x) = 0 cos x dx. Then
R x+0.1 Rx
x cos x x cos t dt 0 cos t dt = F (x)
0 1.00 0.100 0.00
0.1 1.00 0.099 0.100
0.2 0.98 0.097 0.199
0.3 0.95 0.094 0.295
0.4 0.92 0.090 0.389
0.5 0.88 0.085 0.479
0.6 0.83 0.080 0.564
0.7 0.76 0.073 0.644
0.8 0.70 0.066 0.717
0.9 0.62 0.058 0.783
1.0 0.54 0.841
R 1
(d) Let f (x) = 1/(1 + x3 ) and F (x) = 0 dx/(1 + x3 ). Then
R x+0.1 Rx
x 1/(1 + x3 ) x dt/(1 + t3 ) 0 dt/(1 + t3 ) = F (x)
0 1.00 0.100 0.00
0.1 1.00 0.100 0.100
0.2 0.99 0.098 0.200
0.3 0.97 0.096 0.298
0.4 0.94 0.091 0.393
0.5 0.89 0.086 0.485
0.6 0.82 0.078 0.570
0.7 0.75 0.070 0.649
0.8 0.66 0.062 0.719
0.9 0.58 0.054 0.781
1.0 0.50 0.835

10
 √ R0 √
(e) Let f (x) = 1 − x3 and F (x) = 0 .5 1 − x3 dx. Then
√ R x+0.1 √ Rx√
x 1 − x3 x
1 − t3 dt
0
1 − t3 dt = F (x)
0 1.00 0.100 0.00
0.1 1.00 0.100 0.100
0.2 1.00 0.099 0.200
0.3 0.99 0.098 0.299
0.4 0.97 0.095 0.397
0.5 0.04 0.492

2. Let f (x) be continuous for a ≤ x ≤ b

(a) By (4.15) and (4.9) we find

Z b  Z x 
d d d dF
f (t) dt = − f (t) dt = [−F (x)] = − = −f (x)
dx x dx b dx dx

(b) Let us make the substitution u = x2 and note that a ≤ u ≤ b. Then

Z x2 Z u
d d d dF du
= 2xf (u) = 2xf x2


f (t) dt = f (t) dt = F (u) =
dx a dx a dx du dx

(c) Let us make the substitution u = x2 and note that a ≤ u ≤ b. Then

Z b " Z 2 #
x  Z u 
d d d d
f (t) dt = − f (t) dt = − f (t) dt = [−F (u)]
dx x2 dx b dx b dx
dF du
=−
du dx
= −2xf (u) = −2xf x2

(d) Let us make the substitutions u = x2 , v = x3 and note that a ≤ u, v ≤ b.

Furthermore, let c be a fixed point x2 ≤ c ≤ x3 . Then
Z x3 Z x3 Z c !
d d
f (t) dt = f (t) dt + f (t) dt
dx x2 dx c x2
Z v Z c
d d
= f (t) dt + f (t) dt
dx c dx u
Z v Z u
d d
= f (t) dt − f (t) dt
dx c dx c
dF dv dF du
= 3x2 f x3 − 2xf x2



= −
dv dx du dx

11
3. (a) By (4.20) we find approximately
Z 1
dt 1−1
ln 1 = ∼ [1 + 1] = 0
1 t 4
Z 2
dt 1
ln 2 = ∼ (1 + 1.818 + 1.667 + 1.538 + · · · + 1.053 + 0.5) u 0.694
1 t 20
Z 1/2 Z 1
dt dt 1
ln 0.5 = =− ∼ − (2 + 3.636 + 3.333 + · · · + 2.105 + 1) u −0.694
1 t 1/2 t 40

(b) Using the definition Z x

dt
ln x = x>0
1 t
let F (x) = ln x. Then by (4.10) we find
Z x
dF d d dt 1
= ln x = = x>0
dx dx dx 1 t x
Hence, the first derivative of ln x, x > 0 exists and as such, ln x is defined and
continuous for 0 < x < ∞.
(c) Let F (x) = ln ax − ln x for a > 0 and x > 0. Next, let us make the substitution
u = ax so that du = adx. Then by (b) we find
Z u Z x
dF d d d d dt d dt
= (ln ax − ln x) = a ln u − ln x = a −
dx dx du dx du 1 t dx 1 t
a 1 a 1
= − = − =0
u x ax x
Hence, F 0 (x) ≡ 0 so that F (x) ≡ const = ln a. And so

F (x) = ln a = ln ax − ln x =⇒ ln ax = ln a + ln x for a, x > 0

4. Let an ellipse be given by the parametric equations: x = a cos φ, y = b sin φ, b > a > 0.
Then by (3.53) the element of arc ds on the curve traced out by the ellipse is defined
as ds2 = dx2 + dy 2 . Hence, the length of arc from φ = 0 to φ = α is given by
Z α Z αp Z αq
s= ds = dx2 + dy 2 = a2 sin2 φ + b2 cos2 φ dφ
0 0
Z0 α q
a2 sin2 φ + b2 1 − sin2 φ dφ


=
Z0 α q
= b2 − (b2 − a2 ) sin2 φ dφ
0
Z α r Z αq
b 2 − a2 2
= b 1− 2
sin φ dφ = b 1 − k 2 sin2 φ dφ
0 b 0

12
5. (a) Let F (x) be as in (4.24). Then by (4.10) we find
Z x
dF d dt 1
= p =p
dx dx 0 2
1 − k 2 sin t 1 − k 2 sin2 x

Hence, the first derivative of F (x), 0 < k 2 < 1 exists for all x and as such, F (x)
is defined and continuous for all x.
(b) Let x2 > x1 . Then since F 0 (x) > 0 for 0 < k 2 < 1 it follows from the very
definition of the derivative that F (x2 ) > F (x1 ). Hence, we conclude that as x
increases, F (x) increases.
(c) Let F (x) be as in (4.24). Then to show that F (x + π) − F (x) = const we use
(4.10) to find
Z x+π Z x
d d dt d dt
[F (x + π) − F (x)] = p − p
dx dx 0 1 − k 2 sin2 t dx 0 1 − k 2 sin2 t
1 1
=p 2
−p
2
1 − k sin (x + π) 1 − k 2 sin2 x
1 1
=q −p =0
1 − k 2 (− sin x)2 1 − k 2 sin2 x

Hence, since F 0 (x+π)−F 0 (x) ≡ 0, we conclude that the quantity F (x+π)−F (x) =
2K, where K > 0 is some positive constant. The fact that K must be positive
and non-zero follows from (b).
(d) We know from (b) that as x increases, F (x) increases. Furthermore, since F (x) ≥
0 for 0 < k 2 < 1 it then follows that limx→∞ F (x) = ∞. Next, to show that
limx→−∞ F (x) = −∞ we write
Z x Z 0
dt dt
lim F (x) = lim p = lim −p
x→−∞ x→−∞ 0
1 − k 2 sin2 t x→−∞ x 1 − k 2 sin2 t
Z x
dt
= lim −p
x→∞ 0
1 − k 2 sin2 t
= − lim F (x) = −∞
x→∞

6. Let x = am(y) be the inverse of the function y = F (x) of (4.24).

(a) Let y2 > y1 . Furthermore, from (a) and (b) of Problem 5 we know that F 0 (x) > 0
for 0 < k 2 < 1 so that F (x2 ) > F (x1 ) for x2 > x1 . Since y = F (x) this implies
that y2 = F (x2 ), y1 = F (x1 ). Then noting that x = am(y) is defined as the
inverse of y = F (x) we can write am(y2 ) = x2 , am(y1 ) = x1 . Now since x2 > x1
we conclude that am(y2 ) > am(y1 ) for y2 > y1 .

13
 (b) From (c) of Problem 5 we know that F (x + π) = F (x) + 2K = y + 2K. Using the
fact that x = am(y) is defined as the inverse of y = F (x) we then find

am (y + 2K) = am [F (x + π)] = x + π = am (y) + π

p
(c) From (a) of Problem 5 we know that dF/dx = dy/dx = 1/ 1 − kp2 sin2 x. Since
(dy/dx)(dx/dy) ≡ 1 we thus conclude that am0 (y) = dx/dy = 1 − k 2 sin2 x.
Hence, the first derivative of am(y), 0 < k 2 < 1 exists for all y and as such,
am(y) is defined and continuous for all y.

7. Let the functions sn(y), cn(y), dn(y) be defined in terms of the function of Problem 6:
q
sn (y) = sin [am (y)] cn (y) = cos [am (y)] dn (y) = 1 − k 2 sin2 y

Then

(a)
sn2 (y) + cn2 (y) = sin2 [am (y)] + cos2 [am (y)] = sin2 x + cos2 x = 1
(b)

d d d dx
sn (y) = sin [am (y)] = cos [am (y)] am (y) = cn (y)
dy dy dy dy
p
= cn (y) 1 − k 2 sin2 x
= cn (y) dn (y)

(c)

d d d dx
cn (y) = cos [am (y)] = − sin [am (y)] am (y) = −sn (y)
dy dy dy dy
p
= −sn (y) 1 − k 2 sin2 x
= −sn (y) dn (y)

(d)

sn (y + 4K) = sin [am (y + 4K)] = sin [am (y) + 2π] = sin [am (y)] = sn (y)

(e)

cn (y + 4K) = cos [am (y + 4K)] = cos [am (y) + 2π] = cos [am (y)] = cs (y)

14
 (f)
p q
dn (y + 2K) = 1 − k 2 sn2 (y + 2K) = 1 − k 2 sin2 [am (y + 2K)]
q
= 1 − k 2 sin2 [am (y) + π]
q
= 1 − k 2 (− sin [am (y)])2
q
= 1 − k 2 sin2 [am (y)]
p
= 1 − k 2 sn2 (y) = dn (y)

8. Let the error function y = erf (x) be defined by the equation

Z x
2
y = erf (x) = e−t dt
0

(a) Using the definition of y = erf (x) form above, then by (4.10) we find
Z x
dy d d 2 2
= erf (x) = e−t dt = e−x
dx dx dx 0

Hence, the first derivative of y = erf (x) exists for all x and as such, y = erf (x)
is defined and continuous for all x.
(b) Z −x Z 0 Z x
−t2 −t2 2
erf (−x) = e dt = − e dt = − e−t dt = −erf (x)
0 −x 0

(c) Let us first consider the case where x ≥ 0. Then we find

Z x Z 1 Z x
1 x d −t2
Z
−t2 −t2
erf (x) = e dt ≤ dt + te dt = 1 − e dt
0 0 1 2 1 dt
1 d x −t2
Z
=1− e dt
2 dt 1
2
e−x e−1
=1− + <1
2 2
Note that we can move the differentiation operation outside of the integral, since
the limits of integration are assumed constant relative to the integration variable
t. From (b) it then follows immediately that −1 < erf (x) < 1.

15
Section 4.5
1. (a) If R is a triangle with vertices (0, 0), (1, 0) (1, 1), so that 0 ≤ x ≤ 1, 0 ≤ y ≤ x,
then
Z 1Z x Z 1 x
y3
ZZ
2 2 2 2 2



x + y dx dy = x + y dy dx = x y+ dx
R 0 0 0 3 0
Z 1
x3

3
= x + dx
0 3
 4 1
x x4 1
= + =
4 12 0 3
2 2
√
(b) Let R
√ be the region: u + v ≤ 1, 0 ≤ w ≤ 1, so that −1 ≤ x ≤ 1, − 1 − u2 ≤
v ≤ 1 − u2 , 0 ≤ w ≤ 1. Then
ZZZ Z 1 Z √1−u2 Z 1
u2 v 2 w du dv dw = √
u2 v 2 w dw dv du
R −1 − 1−u2 0
√
Z 1Z 1−u2 Z 1
=4 u2 v 2 w dw dv du
0 0 0
√ √
Z 1 Z 1−u2 Z 1 Z 1−u2
2 2 1
=2 uv w2 0
dv du = 2 u2 v 2 dv du
Z0 1 0 √ Z 1 0 0
2 1−u2 2
3/2
= u2 v 3 0 du = u2 1 − u2 du
3 0 3 0
2 π/2 2
Z Z π/2
4 1
= sin θ cos θ dθ = sin2 2θ (1 + 2 cos 2θ) dθ
3 0 12 0
Z π/2
1
= (2 + cos 2θ − 2 cos 4θ − cos 6θ) dθ
48 0
 π/2
1 sin 2θ sin 4θ sin 6θ π
= 2θ + − − =
48 2 2 6 0 48

where we have used the identities: 2 sin2 θ = 1 − cos 2θ, 2 cos2 θ = 1 + cos 2θ,
2 cos A cos B = cos(A + B) + cos(A − B).
(c) If R is the region: 1 ≤ r ≤ 2, (π/4) ≤ θ ≤ π, then
ZZ Z 2Z π Z 2 Z 2
√
3 3 π 2 3
r cos θ dr dθ = r cos θ dθ dr = r3 sin θ π/4
dr = − r dr
R 1 π/4 1 1 2
√ 2 √
2 4 15 2
=− r =−
8 1 8

(d) Let R be a tetrahedron with vertices (0, 0, 0), (1, 0, 0), (0, 2, 0), (0, 0, 3). To de-
termine the x and y limits we can use the triangle in the xy-plane with vertices

16
 (0, 0), (1, 0), (0, 2), since this is the projection of the tetrahedron onto the xy-
plane. Hence, we find 0 ≤ x ≤ 1, 0 ≤ y ≤ 2 − 2x. To find the z limit we need to
find the equation for the plane passing through the points (1, 0, 0), (0, 2, 0), (0, 0, 3)
as this gives the top surface of the tetrahedron. To this end, we first form the two
planar vectors u = −i + 2j, v = −i + 3k. Next, to find the normal to the plane
we compute u × v = 6i + 3j + 2k. Then by (1.23) the equation for the plane is
given by z = 3 − 3x − (3/2)y. Hence,
ZZZ Z 1 Z 2−2x Z 3−3x−(3/2)y
(x + z) dV = (x + z) dz dy dx
R 0 0 0
1 2−2x 3−3x−(3/2)y
z2
Z Z 
= xz + dy dx
0 0 2 0
Z 1 Z 2−2x
1
12x2 + 24xy − 48x + 9y 2 − 36y + 36 dy dx


=
8 0 0
1 1
Z
2−2x
= 12x2 y + 12xy 2 − 48xy + 3y 3 − 18y 2 + 36y 0 dx
8 0
Z 1
=3 (−x + 1)2 dx = 1
0

2. (a) Let z = f (x, y) = ex cos y and 0 ≤ x ≤ 1, 0 ≤ y ≤ π/2. Then

ZZ Z 1 Z π/2 Z 1 Z 1
x x π/2
V = f (x, y) dx dy = e cos y dy dx = e sin y 0
dx = ex dx
R 0 0 0 0
x 1
= e 0 =e−1

(b) Let z = f (x, y) = x2 e−x−y and 0 ≤ x ≤ 1, 0 ≤ y ≤ 2. Next, let u = e−x so that

−u−1 du = dx. Then
ZZ Z 1Z 2 Z 1
2 −x−y 2
V = f (x, y) dx dy = xe dy dx = − x2 e−x−y 0 dx
R 0 0 0
Z 1 Z e−1
−x −x−2 −2
2
ln2 u du



= x e −e dx = e −1
0
" #1
Z e−1
e−1
= e−2 − 1 u ln2 u


1
−2 ln u du
1
!
Z e−1
e−1
= e−2 − 1 e−1 − 2u ln u


1
+2 du
1

 e−1

= e−2 − 1 3e−1 + 2u = 1 − e−2 2 − 5e−1



1

17
 (c) Let z = f (x, y) = x2 y and 0 ≤ x ≤ 1, x + 1 ≤ y ≤ x + 2. Then
Z 1 Z x+2
1 1 2 2 x+2
ZZ Z
2
V = f (x, y) dx dy = x y dy dx = x y x+1 dx
R 0 x+1 2 0
1 1
Z
2x3 + 3x2 dx


=
2 0
1 4 1 3
= x + 2x3 0 =
4 4
p
(d) Let z = f (x, y) = x2 − y 2 and x2 − y 2 ≥ 0, 0 ≤ x ≤ 1. Next, let sin u = y/x so
that cos u du = dy/x. Then
ZZ Z 1Z x p
V = f (x, y) dx dy = x2 − y 2 dy dx
R 0 −x
Z 1Z x
1 1 2 x
Z Z
2 2
= x cos u du dx = x (1 + cos 2u) du dx
0 −x 2 0 −x
x
1 1 2
Z 
sin 2u
= x u+ dx
2 0 2 −x
1 1 2
Z   x π 1 2
Z
−1 y 1 −1 y

= x sin + sin 2 sin = x dx
2 0 x 2 x −x 2 0
1
π π
= x2 =
6 0 6

3. (a) Let 1 ≤ x ≤ 2, 1 − x ≤ y ≤ 1 + x. Then

ZZ Z 2Z 1+x
f (x, y) dA = f (x, ) dy dx =
R 1 1−x

(b) Let y 2 + x(x − 1) ≤ 0. Then

ZZ Z 1 Z √x(1−x)
f (x, y) dA = √ f (x, y) dy dx
R 0 − x(1−x)

4. (a) Let R be the cube of vertices (0, 0, 0), (1, 0, 0), (0, 1, 0), (0, 0, 1), (1, 1, 0), (1, 0, 1),

18
 (0, 1, 1), (1, 1, 1).
1 1 1 1 1 1 √
√
ZZZ Z Z Z Z Z Z
f (x, y, z) dV = x + y + z dz dy dx = u du dy dx
R 0 0 0 0 0 0 |{z}
u=x+y+z
Z 1 Z 1
2 1
= (x + y + z)3/2 0 dy dx
3 0 0
2 1 1h
Z Z i
3/2 3/2
= (1 + x + y) − (x + y) dy dx
3 0 0
Z 1h
4 i1
= (1 + x + y)5/2 − (x + y)5/2 dx
15 0 0
Z 1h
4 i
= (2 + x)5/2 − 2 (1 + x)5/2 + x5/2 dx
15 0
8 h i1 8  √ √ 
= (2 + x)7/2 − 2 (1 + x)7/2 + x7/2 = 9 3−8 2+1
105 0 35

(b) Let R be the pyramid of vertices (±1, ±1, 0) and (0, 0, 1). In order to determine
the limits of integration we will use the cross-section method. Let 0 ≤ z ≤ 1.
Then a plane perpendicular to the z-axis whose boundaries satisfy the inequalities
−1 + z ≤ y ≤ 1 − z, −1 + z ≤ x ≤ 1 − z will represents the cross-sectional area
of the pyramid for a given value of z. Hence,
ZZZ Z 1 Z 1−z Z 1−z
x2 + z 2 dx dy dz


f (x, y, z) dV =
R 0 −1+z −1+z
Z 1 Z 1−z  3
1−z
x
= + xz 2 dy dz
0 −1+z 3 −1+z
Z 1
1−z
Z
2 2 3
= 1 − 3z + 6z − 4z dy dz
3 0 −1+z
2 1
Z

1−z
= 1 − 3z + 6z 2 − 4z 3 y −1+z dz
3 0
4 1
Z
1 − 4z + 9z 2 − 10z 3 + 4z 4 dz


=
3 0
 1
4 2 3 5 4 4 5 2
= z − 2z + 3z − z + z =
3 2 5 0 5

5. (a) Let the integral Z 1 Z 1−x

f (x, y) dy dx
1/2 0

be given. It then follows that 1/2 ≤ x ≤ 1, 0 ≤ y ≤ 1 − x and so the region

R is the triangle with vertices (1/2, 0), (1, 0), (1/2, 1/2) in the xy-plane. Hence,

19
 interchanging the order of integration, the integral can also be written as
Z 1/2 Z 1−y
f (x, y) dx dy
0 1/2

(b) Let the integral √

Z 1 Z 1−x2
f (x, y) dy dx
0 0
√
be given. It then follows that 0 ≤ x ≤ 1, 0 ≤ y ≤ 1 − x2 and so the region R is
the quarter circle in the positive quadrant of the xy-plane. Hence, interchanging
the order of integration, the integral can also be written as
Z Z √ 1 1−y 2
f (x, y) dx dy
0 0

(c) Let the integral Z 1 Z 0

f (x, y) dx dy
0 y−1

be given. It then follows that y − 1 ≤ x ≤ 0, 0 ≤ y ≤ 1 and so the region R is the

triangle with vertices (0, 0), (−1, 0), (0, 1) in the xy-plane. Hence, interchanging
the order of integration, the integral can also be written as
Z 0 Z x+1
f (x, y) dy dx
−1 0

(d) Let the integral Z 1 Z 1+x

f (x, y) dy dx
0 1−x

be given. It then follows that 0 ≤ x ≤ 1, 1 − x ≤ y ≤ 1 + x and so the region R is

the triangle with vertices (1, 0), (0, 1), (1, 1) in the xy-plane. Hence, interchanging
the order of integration, the integral can also be written as
Z 2Z 1
f (x, y) dx dy
0 |y−1|

6. (a) First of all, let us define a diametral plane of a sphere as a plane that cuts the
sphere into two equal halves, hence, containing the midpoint of the sphere. Next,
let n = N/|N|, where N = Ai+Bj+Ck, be a unit vector normal to the diametral
plane and let P = (x0 , y0 , z0 ) be an arbitrary point of the plane so that by (1.23)
the equation for the plane can be written as Ax + By + Cz + D = 0, where
D = −Ax0 − By0 − Cz0 and x, y, z are expected to be on the place. Then the
distance of the point P to another arbitrary point Q = (x1 , y1 , z1 ) is the length of

20
 −→
the projection of the vector P Q onto the normal unit vector n, or in other words
−→
by (1.12), as the component of P Q in the direction of n:
−→ −→ −→ |A (x1 − x0 ) + B (y1 − y0 ) + C (z1 − z0 )|
d = compn P Q = |P Q| cos α = |P Q · n| = √
A2 + B 2 + C 2
|Ax1 + By1 + Cz1 + D|
= √
A2 + B 2 + C 2
Recognizing that the point Q was chosen arbitrarily then by (2.84) the distance
of any point to the diametral plane can be written as
|Aρ sin φ cos θ + Bρ sin φ sin θ + Cρ cos φ + D|
d (ρ, φ, θ) = √
A2 + B 2 + C 2
in spherical coordinates. Hence, the mass of a sphere whose density is proportional
to the distance from one diametral plane is given by
ZZZ Z 2π Z π Z R
M =k d (ρ, φ, θ) dV = k d (ρ, φ, θ) ρ2 sin φ dρ dφ dθ
R 0 0 0
Z 2π Z π Z R
=κ |Aρ sin φ cos θ + Bρ sin φ sin θ + Cρ cos φ + D|ρ2 sin φ dρ dφ dθ
0 0 0
√
where R is the radius of the sphere, κ = k/ A2 + B 2 + C 2 , k is the constant of
proportionality and dV = ρ2 sin φ dρ dφ dθ.
(b) By (4.54), the coordinates of the center of mass of the sphere of part (a) are given
by
ZZZ Z 2π Z π Z R
k k
x̄ = xd (ρ, φ, θ) dV = d (ρ, φ, θ) ρ3 sin2 φ cos θ dρ dφ dθ
M R M 0 0 0
Z 2π Z π Z R
κ
= |Aρ sin φ cos θ + Bρ sin φ sin θ + Cρ cos φ + D|ρ3 sin2 φ cos θ dρ dφ dθ
M 0 0 0
ZZZ Z 2π Z π Z R
k k
ȳ = yd (ρ, φ, θ) dV = d (ρ, φ, θ) ρ3 sin2 φ sin θ dρ dφ dθ
M R M 0 0 0
Z 2π Z π Z R
κ
= |Aρ sin φ cos θ + Bρ sin φ sin θ + Cρ cos φ + D|ρ3 sin2 φ sin θ dρ dφ dθ
M 0 0 0

ZZZ Z 2π Z π Z R
k k
z̄ = zd (ρ, φ, θ) dV = d (ρ, φ, θ) ρ3 sin φ cos φ dρ dφ dθ
M R M 0 0 0
Z 2π Z π Z R
κ
= |Aρ sin φ cos θ + Bρ sin φ sin θ + Cρ cos φ + D|ρ3 sin φ cos φ dρ dφ dθ
M 0 0 0
√
where again R is the radius of the sphere, κ = k/ A2 + B 2 + C 2 , k is the constant
of proportionality and dV = ρ2 sin φ dρ dφ dθ.

21
 (c) The moment of inertia about the x-axis of a solid filling the region 0 ≤ z ≤
1 − x2 − y 2 , 0 ≤ x ≤ 1, 0 ≤ y ≤ 1 − x and having density proportional to xy is,
using (4.55), given by
ZZZ Z 1 Z 1−x Z 1−x2 −y 2
2 2
xy y 2 + z 2 dz dy dx



Ix = y +z f (x, y, z) dx dy dz = k
R 0 0 0

where k is the constant of proportionality.

7. The moment of inertia of a solid about an arbitrary line L is defined as

ZZZ
IL = d2 f (x, y, z) dx dy dz
R

where f is density and d is the distance from a general point (x, y, z) of the solid to
the line L. Next, let L̄ be a line parallel to L that coincides with the z-axis, such that
the center of mass is located at the origin. Since L̄ and L are parallel, we can define
the distance between L̄ and L by the constant h. We may assume, without loss of
generality, that in a Cartesian coordinate system the distance between the lines L and
L̄ lies along the x-axis. Hence, the square of the distance from a general point (x, y, z
of the solid to the line L may be written as d2 = (x + h)2 + y 2 . Substituting for this
in the equation for IL then gives
ZZZ
(x + h)2 + y 2 f (x, y, z) dx dy dz
 
IL =
ZR
ZZ
x2 + y 2 + h2 + 2xh f (x, y, z) dx dy dz


=
ZR
ZZ ZZZ
2 2 2


= x +y f (x, y, z) dx dy dz +h f (x, y, z) dx dy dz
R R
| {z } | {z }
(4.55) (4.52)
ZZZ
+ 2h xf (x, y, z) dx dy dz = IL̄ + M h2 + 2M hx̄ = IL̄ + M h2
R
| {z }
(4.54)

Note that the term 2M hx̄ = 0, since this is a multiple of the x-coordinate of the center
of mass, which is located at the origin.

8. Let L be a line through the origin O with directions cosines l, m, n. Then we can
define the vector n = li + mj + nk as a unit vector parallel to the line L. Furthermore,
we can also define the vector r = xi + yj + zk as the position vector of a point in
space. The shortest distance from an arbitrary point in space P to the line L is then

22
 given by the length of the perpendicular line from P to the line L, which may be
expressed as d = |r| sin θ, where 0 ≤ θ ≤ π is the angle between the vectors n and
r. Since n is a unit vector (i.e. a vector with magnitude 1), we note that by (1.16)
|r| sin θ = |r||n| sin θ = |r × n|. The moment of inertia of a solid about the line L can
then be defined as
ZZZ ZZZ
2
IL = d f (x, y, z) dx dy dz = |r × n|2 f (x, y, z) dx dy dz
ZR
ZZ R

(ny − mz)2 + (lz − nx)2 + (mx − ly)2 dx dy dz

 
=
| {z }
R (1.20)
ZZZ ZZZ ZZZ
2 2 2 2 2 2 2
x2 + y 2 dx dy dz




=l y + z dx dy dz + m x + z dx dy dz + n
R
ZZZ Z ZRZ ZZZ R

− 2lm xy dx dy dz − 2mn yz dx dy dz − 2ln xz dx dy dz

R R R
2 2 2
= Ix l + Iy m + Iz n − 2Ixy lm − 2Iyz mn − 2Ixz ln

9. Let us consider a tetrahedron whose base (i.e. one of its sides) is in the xy-plane such
that the centroid of its base is located at the origin O and let the three vertices of
the base be given by the points (x1 , y1 ), (x2 , y2 ), (x3 , y3 ). As such, the centroid of the
triangular base satisfies the equations
x1 + x2 + x3 y1 + y2 + y3
=0 =0
3 3
Next, the z-coordinate of the only vertex that is not in the xy-plane can be de-
fined as the constant z = h. Then the coordinates of the tetrahedron are given by
(x1 , y1 , 0), (x2 , y2 , 0), (x3 , y3 , 0), (x4 , y4 , h). Next, let us consider the line L passing
through the origin and the point (x4 , y4 , h). The centroid of the tetrahedron thus is
given by
x1 + x2 + x 3 + x4 y1 + y2 + y3 + y4 z1 + z2 + z3 + z4
x̄ = ȳ = z̄ =
4 4 4
Using the fact that x1 + x2 + x3 = y1 + y2 + y3 = z1 + z2 + z3 = 0 the three equations
above reduce to x̄ = x4 /4, ȳ = y4 /4, z̄ = h/4. Denoting the vertex not in the xy-plane
by P = (x4 , y4 , h) and the centroid of the tetrahedron by G = (1/4)(x4 , y4 , h) we then
find that
1 3 3
P − G = (x4 , y4 , h) − (x4 , y4 , h) = (x4 , y4 , h) = (P − O)
4 4 4

23
10. (a) Let F(t) = t2 i − et j + k/(1 + t). Then
1 1 1 1 1
t3
Z Z Z Z
2 t dt 1 1
F (t) dt = t dti − e dtj + k= i − et 0 j + ln (1 + t) 0 k
0 0 0 0 1+t 3 0
1
= i + (1 − e) j + ln 2k
3

(b) Let R be the triangular region enclosed by the triangle of vertices (0, 0), (1, 0), (0, 1)
and F(x, y) = x2 yi + xy 2 j. Then
ZZ Z 1 Z 1−x Z 1 Z 1−x
2
F (x, y) dA = x y dy dxi + xy 2 dy dxj
R 0
Z 10 Z0 1 0
1 1−x 1 1−x
= x2 y 2 0 dxi + xy 3 0 dxj
2 0 3 0
Z 1
1 1
Z
1 2
= 2
x (1 − x) dxi + x (1 − x)3 dxj
2 0 3 0
1 1
1 x3 x4 x5 1 x2 3x4 x5
 
3 1
= − + i+ −x + − j= (i + j)
2 3 2 5 0 3 2 4 5 0 60

11. Let F(t) = f (t)i + g(t)j + h(t)k be continuous for a ≤ t ≤ b and let q be a constant
vector. Then

(a)
Z b Z b
q · F (t) dt = (qx i + qy j + qz k) · [f (t) i + g (t) j + h (t) k] dt
a a
Z b
= [qx f (t) + qy g (t) + qz h (t)] dt
a
Z b Z b Z b
= qx f (t) dt + qy g (t) dt + qz h (t) dt
a a a
Z b Z b Z b
= qx f (t) dt + qy g (t) dt + qz h (t) dt
a a a
Z b Z b Z b 
= (qx i + qy j + qz k) · f (t) dti + g (t) dtj + h (t) dtk
a a a
| {z }
(4.57)
Z b
=q· F (t) dt
a

24
 (b)
Z b Z b
q × F (t) dt = (qx i + qy j + qz k) × [f (t) i + g (t) j + h (t) k] dt
a a
Z b
= [(qy h (t) − qz g (t)) i + (qz f (t) − qx h (t)) j + (qx g (t) − qy f (t)) k] dt
a
Z b Z b
= [qy h (t) − qz g (t)] dti + [qz f (t) − qx h (t)] dtj
a a
Z b
+ [qx g (t) − qy f (t)] dtk
a
 Z b Z b   Z b Z b 
= qy h (t) dt − qz g (t) dt i + qz f (t) dt − qx h (t) dt j
a a a a
 Z b Z b 
+ qx g (t) dt − qy f (t) dt k
a a
Z b Z b Z b 
= (qx i + qy j + qz k) × f (t) dti + g (t) dtj + h (t) dtk
a a a
Z b
=q× F (t) dt
a

Section 4.6
1. (a) Let x = sin θ so that dx = cos θ dθ. Using the identity 2 cos2 θ = 1 + cos 2θ, then
1 π/2
1 π/2
Z Z Z
2 3/2 4
(1 + cos 2θ)2 dθ


1−x dx = cos θ dθ =
0 0 4 0
1 π/2
Z
1 + 2 cos 2θ + cos2 θ dθ


=
4 0
1 π/2
Z
= (3 + 4 cos 2θ + cos 4θ) dθ
8 0
1 3π
= [12θ + 8 sin 2θ + sin 4θ]π/2
0 =
32 16

(b) Let x = u2 − 1 so that dx = 2u du and v = 1 + u so that dv = du. Then

√ √
1 2 1+ 2 √
v−1
Z Z Z
dx u
√ =2 du = 2 dv = 2 [v − ln v]1+
2
2

0 1+ 1+x 1 1+u 2 v
√  √ 
= 2 2 − 2 + 2 ln 2 2 − 2

√ √
(c) Let t = tan(x/2) so that dt = (1/2) sec2 x dx and 2u = t + 1 so that 2du = dt.

25
 Then
Z π/2 1
2 cos2 (x/2)
Z
dx
= dt
0 sin x + cos x + 2 0 sin x + cos x + 2
Z 1
2 cos2 (x/2)
= 2
dt
0 2 sin (x/2) cos (x/2) + 2 cos (x/2) + 1
Z 1
2 cos2 (x/2)
= 2 2
dt
0 2 tan (x/2) cos (x/2) + 2 cos (x/2) + 1
Z 1
2 cos2 (x/2)
= 2 2 2 dt
0 2 tan (x/2) cos (x/2) + 3 cos (x/2) + sin (x/2)
Z 1 Z 1
2 dt
= 2
dt = 2 2
0 tan (x/2) + 2 tan (x/2) + 3 0 t + 2t + 3
√
Z 1 √ Z 2 √ √
dt du −1 2
=2 2 = 2 √ 2
= 2 tan u √2/2
0 (t + 1) + 2 2/2 u + 1
√ !
√ √ 2
= 2 tan−1 2 − tan−1
2

(d) Let t = 1 + x cos x so that dt = (cos x − x sin x) dx. Then

Z π/4 Z 1+c  
x cos x (x sin x − cos x) 1
dx = − 1 dt = [ln t − t]1+c
1 = ln (1 + c)−c
0 1 + x cos x 1 t
√
where c = π/(4 2).
2. Let φ(u) be a function of u. Next, we consider the integral
Z u2 Z u2 Z u2
0 dφ
φ (u) du = du = dφ = φ (u2 ) − φ (u1 )
u1 u1 du u1

which is equal to (4.60) when setting f (x) = f [x(u)] ≡ 1.

3. (a) Let f (x) be continuous for x1 ≤ x ≤ x2 . Next, let us choose a u0 and x0 = x(u0 )
such that u1 < u0 < u2 . Replacing u2 for u0 in (4.60) then gives
Z x0 Z u0
dx
f (x) dx = f [x (u)] du
x1 u1 du
which remains to hold. Then if we let u0 → u2 we get
Z u0
dx
lim f [x (u)] du = lim (F [x (u0 )] − F [x (u1 )]) = F [x (u2 )] − F [x (u1 )]
u0 →u2 u
1
du u0 →u2

= F (x2 ) − F (x1 )
which is the same as the value of the left-hand side of (4.60). Hence, if the right-
hand side of (4.60) has a limit, then the other side has a limit also and the two
limits are equal.

26
 (b) Let u = 1/x so that du = −dx/x2 . Then
Z ∞ Z b Z 1
1 1 1 1 1
sinh dx = lim sinh dx = sinh u du = cosh u 0
= cosh 1 − 1
1 x2 x b→∞ 1 x2 x 0

(c) Let u = tanh x so that du = (1 − tanh2 x) dx. Then

Z ∞ Z b Z 1
du 1
(1 − tanh x) dx = lim (1 − tanh x) dx = = ln (1 + u) 0
= ln 2
0 b→∞ 0 0 1+u

4. (a) Let Rxy be the region x2 + y 2 ≤ 1 and x = r cos θ, y = r sin θ. Then by (4.64) we
find
ZZ Z 1 Z √1−x2
1 − x2 − y 2 dx dy = 1 − x2 − y 2 dy dx



√
Rxy 0 − 1−x2
Z 1 Z 2π Z 1
2 2π
1 − r2 rθ



= 1−r r dθ dr = 0
dr
0 0 0
1 1
r4
Z 
2

2 π
= 2π 1 − r r dr = π r − =
0 2 0 2
(b) Let R be the region 1 ≤ x ≤ 2, 0 ≤ y ≤ x and x = r cos θ, y = r sin θ. Further-
more, let u = sec3 θ so that du = 3 sec3 θ tan θ dθ. Then by (4.64) we find
ZZ p 2 Z 2Z x p 2 Z π/4 Z 2 sec θ
y x + y2 y x + y2
dx dy = dy dx = r2 tan θ dr dθ
R x 1 0 x 0 sec θ
Z π/4 Z π/4
1 2 sec θ 7
= r3 tan θ sec θ dθ = sec3 θ tan θ dθ
3 0 9 0
Z √ √
7 2 2 7 2√2 14 2 − 7
= du = u 1 =
9 1 9 9
(c) Let Rxy be the parallelogram with successive vertices (π, 0), (2π, π), (π, 2π), (0, π)
and u = x−y, v = x+y so that Ruv is the square region with vertices (π, π), (π, 3π),
(−π, 3π), (−π, π). Then by (4.61)
ZZ Z 2π Z 2π−|π−x|
2 2
(x − y) sin (x + y) dx dy = (x − y)2 sin2 (x + y) dy dx
Rxy 0 |π−x|
Z π Z 3π
1
= u2 sin2 v dv du
2 −π π
1 π 3π 2
Z Z
= u (1 − cos 2v) dv du
4 −π π
1 π 2 π π 2
Z Z
3π
= u [2v − sin 2v]π du = u du
8 −π 2 −π
π
πu3 π4
= =
6 −π 3

27
 (d) Let R be the trapezoidal region bounded by the lines x + y = 1, x + y = 2 in the
first quadrant and u = 1 + x + y, v = x − y. Then by (4.61)

(x − y)2 1
(x − y)2
2−x
(x − y)2
ZZ Z Z Z 2 Z 2−x
dx dy = dy dx + dy dx
R 1+x+y 0 1−x 1 + x + y 1 0 1+x+y
−1+u
1 3 −1+u v 2 1 3 v3
Z Z Z
= dv du = du
2 2 1−u u 6 2 u 1−u
3
1 3 2 1 u3 3u2
Z 
−1


= u − 3u + 3 − u du = − + 3u − ln u
3 2 3 3 2 2
11 1 2
= + ln
18 3 3

(e) Let R be the region bounded by the ellipse 5x2 +2xy +2y 2 = 1 and x = u + v, y =
−2u + v. Then by (4.61)
ZZ p Z √ Z (√2−9x2 −x)/2 p
2/3
2 2
5x + 2xy + 2y dx dy = 5x2 + 2xy + 2y 2 dy dx
√ √
R − 2/3 −( 2−9x2 −x)/2
Z 1/3 Z √(1/9)−u2 √
=9 √ u2 + v 2 dv du
−1/3 − (1/9)−u2
Z 2π Z 1/3 Z 2π
2 1/3
=9 r dr dθ = 3 r3 0
dθ
0 0 0
Z 2π
3 2π
= dθ =
27 0 9

5. According to Section 2.7, the transformation u = ex cos y, v = ex sin y can be regarded

as a mapping from the xy-plane to the uv-plane. Since the inverse mapping exists and
is given by x = (1/2) ln(u2 + v 2 ), y = tan−1 (v/u), hence, taking each point of a region
in the uv-plane to a unique point in the xy-plane, the mapping is one-to-one. Now let
us consider the rectangle Rxy : 0 ≤ x ≤ 1, 0 ≤ y ≤ π/2. Then u and v are defined
and continuous for each point (x, y) ∈ Rxy (i.e. their derivatives exist). Hence, the
given transformation defines a one-to-one mapping of the rectangle Rxy onto a region
of the uv-plane. To find the boundaries of the region in the uv-plane we consider each
of the sides of the rectangle Rxy in turn. For the bottom side of the rectangle given
by 0 ≤ x ≤ 1, y = 0 we find 1 ≤ u ≤ e, v = 0. For the top side of the rectangle
given by 0 ≤ x ≤ 1, y = π/2 we find u = 0, 1 ≤ v ≤ e. The left side of the rectangle
given by x = 0, 0 ≤ y ≤ π/2 transforms as u = cos y, v = sin y, whereas the right side
of the rectangle given by x = 1, 0 ≤ y ≤ π/2 transforms as u = e cos y, v = e sin y.
Hence, the rectangle Rxy transforms to the ring Ruv as shown in the figure below

28
 As such, by (4.61) we find that
Z π/2 Z 1
e2x e2x
ZZ
dx dy = 2 dx dy
1 + e4x cos2 y sin2 y 0
4x 2
0 1 + e cos y sin y
Rxy
√ √
Z 1 Z e2 −u2 Z eZ e2 −u2
dv du dv du
= √
+
1−u2 1 + u2 v 2 1 + u2 v 2
Z0Z 1 0
du dv
=
1 + u2 v 2
Ruv

6. The inverse mapping is given by

q √ q √
x = ± (1/2)(v ± u2 + v 2 ) y = ± (1/2)(−v ± u2 + v 2 )

Since the inverse mapping is well-defined, the mapping from the xy-plane to the
uv−plane is one-to-one. Now we consider the square region 0 ≤ x ≤ 1, 0 ≤ y ≤ 1.
Then u and v are defined and continuous for each point (x, y) of the square. Hence, the
given transformation defines a one-to-one mapping of the square 0 ≤ x ≤ 1, 0 ≤ y ≤ 1
onto a region of the uv-plane. To find the boundaries of the region in the uv-plane we
consider each of the sides of the square in the xy-plane in turn. For the bottom side of
the square given by 0 ≤ x ≤ 1, y = 0 we find u = 0, v = x2 . Hence, the bottom side of
the square in the xy-plane maps to the vertical line segment with endpoints (0, 0), (0, 1)
in the uv-plane. For the top side of the square given by 0 ≤ x ≤ 1, y = 1 we find
u = 2x, v = x2 −1. Treating this as a parametric equation in the variable t this becomes
u = 2t, v = t2 −1 where 0 ≤ t ≤ 1. Then solving for v gives v = (1/4)u2 −1, 0 ≤ u ≤ 2.
Hence, the top side of the square in the xy-plane maps to the line segment given by
the aforementioned equation with endpoints (0, −1), (2, 0) in the uv-plane. For the

29
 left side of the square given by x = 0, 0 ≤ y ≤ 1 we find u = 0, v = −y 2 . Hence,
the left side of the square in the xy-plane maps to the vertical line segment with end-
points (0, 0), (0, −1) in the uv-plane. Lastly, for the right side of the square given by
x = 1, 0 ≤ y ≤ 1 we find u = 2y, v = 1 − y 2 . Treating this as a parametric equation
in t this becomes u = 2t, v = 1 − t2 where 0 ≤ t ≤ 1. Solving for v then gives v =
1 − (1/4)u2 , 0 ≤ u ≤ 2. Hence, the right side of the square in the xy-plane maps to the
line segment given by the aforementioned equation, having endpoints (0, 1), (2, 0) in the
uv-plane. To summarize, the figure below shows the transformed region in the uv-plane.

As such, by (4.61) we find that

ZZ p Z 1 Z 1 p
3 4 2 2 4 3
x − 6x y + y dx dy = x4 − 6x2 y 2 + y 4 dx dy
0 0
Rxy
2 1−(1/4)u2
√
3 ZZ √
3
v 2 − u2 v 2 − u2
Z Z
1 1
= √ dv du = √ du dv
4 0 −1+(1/4)u2 v 2 + u2 4 v 2 + u2
Ruv

7. (a) Let Rxy be the triangular region bounded by the lines y = x, 0 ≤ x ≤ 1 and
x = 1, 0 ≤ y ≤ 1 in the first quadrant of the xy-plane. And let the transformation
x = u + v, y = u − v be given. Then the inverse transformation is given by
1 1
u= (x + y) v= (x − y)
2 2
so that it defines a one-to-one mapping of the triangle Rxy onto the triangle
Ruv , which boundaries are given by the two lines v = u, 0 ≤ u ≤ 1/2 and
v = 1 − u, 1/2 ≤ u ≤ 1 in the first quadrant of the uv-plane. As such, by (4.61)

30
 we find
ZZ Z 1 Z x
2 2
ln 1 + x2 + y 2 dy dx



ln 1 + x + y dx dy =
0 0
Rxy
Z 1/2 Z 1−v
ln 1 + 2u2 + 2v 2 du dv


=2
Z0Z v

ln 1 + 2u2 + 2v 2 du dv


=2
Ruv

(b) Let Rxy be the triangular region bounded by the lines y = 1 − x, y = 1 + x, 0 ≤

x ≤ 1 and x = 1, 0 ≤ y ≤ 2 in the first quadrant of the xy-plane. And let the
transformation x = u, y = u + v be given. Then the inverse transformation is
given by

u=x v =y−x

so that it defines a one-to-one mapping of the triangle Rxy onto the triangle Ruv ,
which boundaries are given by the lines v = 1−2u, 0 ≤ u ≤ 1, v = 1+u, 0 ≤ u ≤ 1
and u = 1, −1 ≤ v ≤ 1 in the first and fourth quadrants of the uv-plane. As
such, by (4.61) we find
ZZ p Z 1 Z 1+x p Z 1Z 1 q
1 + x2 y 2 dx dy = 1 + x2 y 2 dy dx = 1 + u2 (u + v)2 dv du
0 1−x 0 1−2u
Rxy
ZZ q
= 1 + u2 (u + v)2 du dv
Ruv

8. Let Ruv be the square 0 ≤ u ≤ 1, 0 ≤ v ≤ 1.

(a) Let the transformation x = u + u2 , y = ev be given. It then defines a one-

to-one mapping of the square Ruv onto the rectangle Rxy . The bottom side of
the square given by 0 ≤ u ≤ 1, v = 0 maps to the line segment y = 1, 0 ≤
x ≤ 2 forming the bottom side of the rectangle. The top side of the square
given by 0 ≤ u ≤ 1, v = 1 maps to the line segment y = e, 0 ≤ x ≤ 2
forming the top side of the rectangle. The left side of the square given by
u = 0, 0 ≤ v ≤ 1 maps to the line segment x = 0, 1 ≤ y ≤ e forming
the left side of the rectangle. Finally, the right side of the square given by
u = 1, 0 ≤ v ≤ 1 maps to the line segment x = 2, 1 ≤ y ≤ e forming the right
side of the rectangle. The below figure shows the graph for the rectangle Rxy .

31
(b) Let the transformation x = uev , y = ev be given. The bottom side of the square
given by 0 ≤ u ≤ 1, v = 0 maps to the line segment 0 ≤ x ≤ 1, y = 1. The
top side of the square given by 0 ≤ u ≤ 1, v = 1 maps to the line segment
0 ≤ x ≤ e, y = e. The left side of the square given by u = 0, 0 ≤ v ≤ 1
maps to the line segment x = 0, 1 ≤ y ≤ e. Finally, the right side of the
square given by u = 1, 0 ≤ v ≤ 1 maps to the line segment x = y, 1 ≤
x ≤ e. The below figure shows the graph for the region Rxy in the xy-plane.

(c) Let the transformation x = 2u − v 2 , y = v + uv be given. The bottom side

of the square maps to the line segment 0 ≤ x ≤ 2, y = 0. The top side of
the square maps to the line segment y = (1/2)(3 + x), −1 ≤ x ≤ 1. The left
side of the square maps to the line segment x = −y 2 , 0 ≤ y ≤ 1. Finally,
the right side of the square maps to the line segment x = (1/4)(8 − y 2 ), 1 ≤

32
 x ≤ 2. The below figure shows the graph for the region Rxy in the xy-plane.

(d) Let the transformation x = 5u − u2 + v 2 , y = 5v + 10uv be given. The bottom

side of the square maps to the line segment 0 ≤ x ≤ 4, y = 0. The top side of the
square maps to the line segment x = −(1/100)y 2 +(3/5)y−(7/4), 5 ≤ y ≤ 15. The
left side of the square maps to the line segment x = (1/25)y 2 , 0 ≤ y ≤ 5. Finally,
the right side of the square maps to the line segment x = 4 + (1/225)y 2 , 0 ≤ y ≤
15. The below figure shows the graph for the given region Rxy in the xy-plane.

9. Cylindrical coordinates are defined by the transformation x = r cos θ, y = r sin θ, z =

33
 z. Hence, we find

cos θ −r sin θ 0
∂ (x, y, z)
J= = sin θ r cos θ 0 = r
∂ (r, θ, z)
0 0 1

And so by (4.66)
ZZZ ZZZ ZZZ
∂ (x, y, z)
f (x, y, z) dx dy dz = F (r, θ, z) dr dθ dz = F (r, θ, z) r dr dθ dz
∂ (r, θ, z)
Rxyz Rrθz Rrθz

The element of volume is approximately a rectangular box with sides r∆θ, ∆r and ∆z:
∆V ∼ r∆θ∆r∆z.

10. Spherical coordinates are defined by the transformation x = ρ sin φ cos θ, y = ρ sin φ sin θ, z =
ρ cos φ. Hence, we find

sin φ cos θ ρ cos φ cos θ −ρ sin φ sin θ

∂ (x, y, z)
J= = sin φ sin θ ρ cos φ sin θ ρ sin φ cos θ = ρ2 sin φ
∂ (ρ, φ, θ)
cos φ −ρ sin φ 0

And so by (4.66)
ZZZ ZZZ
∂ (x, y, z)
f (x, y, z) dx dy dz = F (ρ, φ, θ) dρ dφ dθ
∂ (ρ, φ, θ)
Rxyz Rρφθ
ZZZ
= F (ρ, φ, θ) ρ2 sin φ dρ dφ dθ
Rρφθ

The element of volume is approximately a rectangular box with sides ρ∆φ, ∆ρ and
ρ sin φ∆θ: V ∼ ρ2 sin φ∆φ∆ρ∆θ.

11. Let cylindrical coordinates be given: x = r cos θ, y = r sin θ, z = z.

(a)
ZZZ Z 1 Z 1 Z √1−y2 Z 1 Z 2π Z 1
2 2
x y dx dy dz = √ x y dx dy dz = r4 cos2 θ sin θ dr dθ dz
0 0 − 1−y 2 0 0 0
Rxyz
ZZZ
= r4 cos2 θ sin θ dr dθ dz
Rrθz

where Rxyz is the region x2 + y 2 ≤ 1, 0 ≤ z ≤ 1.

34
 (b)
√
Z 1 Z 1−x2 Z 1+x+y Z π/2 Z 1 Z 1+r(cos θ+sin θ)
2 2
r3 cos 2θ dz dr dθ


x −y dz dy dx =
0 0 0 0 0 0

12. Let spherical coordinates be given by x = ρ sin φ cos θ, y = ρ sin φ sin θ, z = ρ cos φ.

(a)
ZZZ Z 2π Z π Z a
2
x y dx dy dz = ρ5 sin4 φ cos2 θ sin θ dρ dφ dθ
0 0 0
Rxyz

where Rxyz is the spherical region x2 + y 2 + z 2 ≤ a2 .

√ √
(b) From the boundsp−1 ≤ x ≤ 1, − 1 − x2 ≤ y ≤ 1 − x2 follows 0 ≤ θ ≤ 2π.
From the bound x2 + y 2 ≤ z ≤ 1 follows 0 ≤ φ ≤ π/4, 0 ≤ ρ ≤ sec φ. Hence,
√
Z 1 Z 1−x2 Z 1 Z 2π Z π/4 Z sec φ
2 2 2
ρ4 sin φ dρ dφ dθ


√ √ x +y +z dz dy dx =
−1 − 1−x2 x2 +y 2 0 0 0

Section 4.7
1. (a) Let x = a cos θ, y = a sin θ. Then by (4.71) the length of the circumference of the
circle is given by
s
Z 2π  2  2 Z 2π
dx dy
s= + dθ = a dθ = 2πa
0 dθ dθ 0

(b) Let x = a(1 − t2 )/(1 + t2 ), y = 2at/(1 + t2 ). Then by (4.70) the length of the
circumference of the circle is given by
s
Z t2  2  2 Z 1 s 2  2
dx dy 4at 2a (1 − t2 )
s= + dt = 2 − + dt
t1 dt dt −1 (1 + t2 )2 (1 + t2 )2
Z 1
1 1
= 4a 2
dt = 4a tan−1 −1 = 2πa
−1 1 + t
p
2. (a) Let the equation z = ± a2 − x2 − y 2 be given. Furthermore let us use the
substitution x = r cos θ, y = r sin θ so that J = r. Then by (4.72) and (4.64) the

35
 area of the surface of a sphere is given by
s  2  2
ZZ
∂z ∂z
S= 1+ + dx dy
∂x ∂y
Rxy
Z 1 Z √1−x2 s
x2 y2
=2 √
1 + + dx dy
0 − 1−x2 a2 − x 2 − y 2 a2 − x 2 − y 2
Z 1 Z √1−x2 Z 2π Z a
1 r
= 2a √
p dx dy = 2a √ dr dθ
0 − 1−x2 a2 − x 2 − y 2 0 0 a2 − r 2
Z a Z a2
r √ a2
= 4πa √ dr = 2πa u−1/2 du = 4πa u 0 = 4πa2
0 a2 − r 2 0

where for the last step we have made use of the substitution u = a2 − r2 .
(b) Let the parametric equations x = a sin φ cos θ, y = a sin φ sin θ, z = a cos φ be
given. Then by (4.74) and (4.75) the area of the surface of a sphere is given by
ZZ √ Z 2π Z π Z 2π
2 2 2 π
S= EG − F dφ dθ = a sin φ dφ dθ = −a cos φ 0 dθ
0 0 0
Rφθ
Z 2π
2
= 2a dθ = 4πa2
0

3. (a) Let a surface in space be given by the parametric equations x = (b+a cos v) cos u, y =
(b + a cos v) sin u, z = a sin v, 0 ≤ u ≤ 2π, 0 ≤ v ≤ 2π, where a and b are con-
stants, 0 < a < b (torus). Then by (4.74) and (4.75) the surface area is given
by
ZZ √ Z 2π Z 2π
S= EG − F 2 du dv = a (b + a cos v) du dv
0 0
Ruv
Z 2π Z 2π
2π
=a (b + a cos v) u 0
dv = 2πa (b + a cos v) dv
0 0
= 2πa [bv + a sin v]2π 2
0 = 4π ab

(b) Let a surface in space be given by the parametric equations x = u cos v, y =

u sin v, z = u2 sin 2v, 0 ≤ u ≤ 1, 0 ≤ v ≤ π/2 (portion of saddle surface z = 2xy).
Then by (4.74) and (4.75) the surface area is given by
ZZ √ Z 1 Z π/2 √ Z 1 √
2 2 π/2
S= EG − F du dv = u 1 + 4u dv du = uv 1 + 4u2 0 du
Ruv 0 0 0
Z 1 √ Z 5
π π √ π 5
= u 1 + 4u2 du = w dw = w3/2 1
2 0 16 1 24
√

π 5 5−1
=
24
36
4. Let us use the parametric equations x = a sin φ cos θ, y = a sin φ sin θ, z = a cos φ to
approximate the surface area of earth. The area of the United States then may be
approximated by the rectangle bounded by the parallels 30◦ N and 47◦ N and meridians
75◦ W and 122◦ W. Furthermore, we take the radius of the earth to be 4000 miles, such
that a = 4000. Hence, by (4.74) we find for the area of the United States
ZZ √ Z 122◦ Z 47◦ Z 122◦
2 2 2 47◦
A= EG − F dφ dθ = 4000 sin φ dφ dθ = − (4000) cos φ 30◦ dθ
75◦ 30◦ 75◦
Rφθ
122◦
47◦ π
Z  
2 ◦ ◦ 2 ◦ ◦
= − (4000) (cos 47 − cos 30 ) dθ = (4000) (cos 47 − cos 30 )
75◦ 180◦
≈ 2415332 mi2

5. (a) Let the vectors a and b be the sides of a parallelogram in space. Then by (1.35),
the area of the parallelogram is given by A = |a × b|, where by (1.20) the vector
a × b is perpendicular to both a and b. Next, let the vector c be a unit vector
(i.e. |c| = 1) perpendicular to an arbitrary plane C in space. Hence, by (1.9) and
(1.12) the dot product a × b · c = |a × b| cos θ = A cos θ = compc a × b, where
θ = ^(a × b, c), 0 ≤ θ ≤ π. As such, we conclude that this dot product equals
plus or minus the area of the projection of the parallelogram on the plane C.
(b) As already discussed for part (a), the area of the parallelogram is given by S = A =
|a × b| and the angle between the vectors a × b and c as γ = θ = ^(a × b, c), 0 ≤
γ ≤ π. Hence, by (1.9) a × b · c = S cos γ.
(c) Let Syz = a × b · i, Szx = a × b · j, Sxy = a × b · k be the areas of the projections
of the parallelogram on the yz-plane, zx-plane and xy-plane respectively. Then
by (1.13) we find
q r  2
Syz + Szx + Sxy = ([a × b]x ) + [a × b]y + ([a × b]z )2 = |a × b| = S
2 2 2 2

6. Let a surface of revolution be given by rotating a curve z = f (x), y = 0 in the xz-plane

about the z-axis.
(a) Next, we consider the transformation x = r cos θ, y = r sin θ, thus introducing
curvilinear, polar coordinates in the xy-plane. As such, we find that y = r sin θ =
0 =⇒ θ = 0. And so x = r cos 0 = r. Hence, by rotating z = f (x) = f (r)
around the z-axis one will obtain the same surface of revolution.
(b) By (4.64) and (4.72), the surface area then is given by
s  2  2
ZZ ZZ p
∂z ∂z
S= 1+ + dx dy = 1 + f 0 (r)2 r dr dθ
∂x ∂y
Rxy Rrθ
Z 2π Z bp Z bp Z bp
2π
= 1 + f 0 (r)2 r dr dθ = 1 + f 0 (r)2 rθ 0
dr = 2π 1 + f 0 (r)2 r dr
0 a a a

37
 where 0 ≤ θ ≤ 2π, since the curve is rotated around the z-axis once fully to obtain
the revolution of the surface and a ≤ r ≤ b, since the radius r will attain both an
identical minimum value a and maximum value b in each plane perpendicular to
the plane of rotation (e.g. the xz-plane).

7. By (4.72) the area of a surface z = f (x, y) is given by

s  2  2
ZZ
∂z ∂z
S= 1+ + dx dy
∂x ∂y
Rxy

Now as (4.73) states, a surface in space can be represented parametrically by the

equations

x = x (u, v) y = y (u, v) z = z (u, v)

where u and v vary in a region Ruv of the uv-plane. Furthermore, since

∂ (x, y)
J=
∂ (u, v)

it follows that (see the solution to Problem 4 following section 2.12)

∂u 1 ∂y ∂u 1 ∂x ∂v 1 ∂y ∂v 1 ∂x
= =− =− =
∂x J ∂v ∂y J ∂v ∂x J ∂u ∂y J ∂u
Next, by (4.61) and using the chain rule it follows that
s  2  2
ZZ
∂z ∂z
S= 1+ + dx dy
∂x ∂y
Rxy
s  2  2
ZZ
∂z ∂u ∂z ∂v ∂z ∂u ∂z ∂v ∂ (x, y)
= 1+ + + + du dv
∂u ∂x ∂v ∂x ∂u ∂y ∂v ∂y ∂ (u, v)
Ruv
s  2  2
ZZ
1 ∂z ∂y ∂y ∂z 1 ∂x ∂z ∂z ∂x ∂ (x, y)
= 1+ 2 − + 2 − du dv
J ∂u ∂v ∂u ∂v J ∂u ∂v ∂u ∂v ∂ (u, v)
Ruv
s 2  2  2
ZZ
∂x ∂y ∂y ∂x ∂z ∂y ∂y ∂z ∂x ∂z ∂z ∂x
= − + − + − du dv
∂u ∂v ∂u ∂v ∂u ∂v ∂u ∂v ∂u ∂v ∂u ∂v
Ruv
ZZ √
= EG − F 2 du dv
Ruv

where E, F, G are given by (4.75).

38
 8. If u, v are curvilinear coordinates in a plane area Rxy then this implies that ∂z/∂u =
∂z/∂v ≡ 0. Hence, we see from the second equality of the previous problem that (4.74)
reduces to
s  2  2
ZZ
∂z ∂z
S= 1+ + dx dy
∂x ∂y
Rxy
s  2  2
ZZ
∂z ∂u ∂z ∂v ∂z ∂u ∂z ∂v ∂ (x, y)
= 1+ + + + du dv
∂u ∂x ∂v ∂x ∂u ∂y ∂v ∂y ∂ (u, v)
Ruv
ZZ ZZ
∂ (x, y)
= du dv = dx dy
∂ (u, v)
Ruv Rxy

9. Let a surface z = f (x, y) be given in implicit form F (x, y, z) = 0. Then by (2.59) we

find that
∂z Fx ∂z Fy
=− =−
∂x Fz ∂y Fz
And so (4.72) becomes
s  2  2 ZZ s
F 2 Fy2
ZZ
∂z ∂z
S= 1+ + dx dy = 1 + x2 + 2 dx dy
∂x ∂y Fz Fz
Rxy Rxy
ZZ p 2
Fx + Fy2 + Fz2
= dx dy
|Fz |
Rxy

CHAPTER 4
Section 4.8
1. As stated in the text, the error integral is given by
Z ∞
2
e−x dx
0

One way to evaluate this is to use the equations

Z ∞ 2 Z ∞ Z ∞ Z ∞Z ∞ ZZ
−x2 −x2 −y 2 −x2 −y 2 2 −y 2
e dx = e dx e dy = e dx dy = e−x dx dy
0 0 0 0 0
R

where R is the unbounded closed region in the positive quadrant of the xy-plane.
2 2
Next, we note that f (x, y) = e−x −y > 0 and is continuous outside and on a circle

39
 x2 + y 2 = a2 . Hence, the double integral above can be expressed as the sum of two
double integrals:
ZZ ZZ ZZ
−x2 −y 2 −x2 −y 2 2 2
e dx dy = e dx dy + e−x −y dx dy
R R1 R2

Here the two regions R1 , R2 overlap only on the boundary x2 + y 2 = a2 and the region
R2 = Rk is the region a2 ≤ x2 + y 2 ≤ k 2 with k → ∞. Introducing polar coordinates
and noting that |J| = r, then by (4.61) we can evaluate the square of the error integral
as
Z ∞ 2 Z Z ZZ ZZ
−x2 −x2 −y 2 −x2 −y 2 2 2
e dx = e dx dy = e dx dy + e−x −y dx dy
0
R R1 R2
Z π/2 Z a Z π/2 Z k
2 2
= e−r r dr dθ + lim e−r r dr dθ
0 0 k→∞ 0 a
Z π/2 Z π/2
1 2 a 1 2 k
= − e−r 0 dθ + lim − e−r a dθ
0 2 0 k→∞ 2
Z π/2  Z π/2
1 2
 1  2 2

= 1 − e−a dθ + lim e−a − e−k dθ
2 0 2 0 k→∞
Z π/2 
1 π/2
Z
1 −k2
 π
= 1 − lim e dθ = dθ =
2 0 k→∞ 2 0 4
where we have used the substitution u = −r2 , so that du = −2r dr. From this it
follows that Z ∞ √
−x2 π
e dx =
0 2
2. Let the integral ZZ p
ln x2 + y 2 dx dy
R
p
where R is the region x2 + y 2 ≤ 1 be given. Now f (x, y) = ln x2 + y 2 is continuous
in R, except at the point (0, 0). As such, we isolate the point (0, 0) by integrating up
to a small circle of radius h about (0, 0) and then letting h approach zero. Employing
polar coordinates, integration by parts and (4.61) we thus find
Z 2π Z 1 1
1 2π r2
ZZ p Z 
2 2 2
ln x + y dx dy = lim r ln r dr dθ = lim r ln (r) − dθ
R h→0 0 h 2 0 h→0 2 h
1 2π h2 1 2π
Z   Z
1 2 π
= lim − − h ln (h) + dθ = − dθ = −
2 0 h→0 2 2 4 0 2
where the limit limh→0 h2 ln h is evaluated using L’Hospital’s rule:
ln h H 1/h h2
lim h2 ln h = lim = lim = lim − =0
h→0 h→0 1/h2 h→0 −2/h3 h→0 2

40
3. (a) Let the integral
ZZZ
1 p
dx dy dz r= x2 + y 2 + z 2 , p > 0
rp
R

where R is the spherical region x2 + y 2 + z 2 ≤ 1 be given. As for the previous

problem, we integrate up to a small circle of radius h about (0, 0, 0) and then let
h approach zero. Employing spherical coordinates and (4.61) we thus find
ZZZ Z 2π Z π Z 1
1 1 2
lim p
dx dy dz = lim p
r sin φ dr dφ dθ
h→0 r h→0 0 0 h r
R
Z 2π Z π 1
r3−p
= lim sin φ dφ dθ
0 0 h→0 3 − p h
1 − h3−p 2π π
Z Z
= lim sin φ dφ dθ
h→0 3 − p 0 0
1 − h3−p 2π 1 − h3−p 2π
Z Z
π
= lim − cos φ 0 dθ = lim 2 dθ
h→0 3 − p 0 h→0 3 − p 0
4π (1 − h3−p )
= lim
h→0 3−p
From this it follows that the integral converges to the value 4π/(3 − p) for p < 3
and diverges for p ≥ 3.
(b) For the integral of part (a), let R be the exterior region x2 + y 2 + z 2 ≥ 1, thus
forming an unbounded closed region. Since f (x, y, z) = 1/rp ≥ 1 (i.e. is positive
for the region given), the integral over R can be defined as the limit
ZZZ Z 2π Z π Z k
1 1 2
lim p
dx dy dz = lim p
r sin φ dr dφ dθ
k→∞ r k→∞ 0 0 1 r
R
2π π k
r3−p
Z Z
= limsin φ dφ dθ
0 0 k→∞ 3 − p 1
k 3−p − 1 2π π
Z Z
= lim sin φ dφ dθ
k→∞ 3 − p 0 0
k 3−p − 1 2π k 3−p − 1 2π
Z Z
π
= lim − cos φ 0 dθ = lim 2 dθ
k→∞ 3 − p 0 k→∞ 3 − p 0
4π (k 3−p − 1)
= lim
k→∞ 3−p
From this it follows that the integral converges to the value 4π/(3 − p) for p > 3.
4. (a) Let the integral ZZ
1
dx dy
x2 + y2
R

41
 where R is the square |x| < 1, |y| < 1 be given. Since R is an unbounded open
region the integral will diverge.
(b) Let the integral
ln (x2 + y 2 )
ZZ
p dx dy
x2 + y 2
R
2 2
where R is the circle x + y ≤ 1 be given. p Here the region R is a unbounded
2 2
closed region and f (x, y) = ln(x + y )/ x2 + y 2 ≤ 0 and continuous in R. As
such, employing polar coordinates, integration by parts and (4.61) the integral
over R can be defined as the limit
Z 2π Z 1 Z 2π
ln (x2 + y 2 )
ZZ
2
1
lim r ln r2 − 2r h dθ

lim p dx dy = lim ln r dr dθ =
h→0 x2 + y 2 h→0 0 h 0 h→0
R
= 2π lim 2h − h ln h2 − 2 = −4π


h→0

where the limit limh→0 h ln h2 is evaluated using L’Hospital’s rule:

ln h2 H 2/h
lim h ln h2 = lim = lim = lim −2h = 0
h→0 h→0 1/h h→0 −1/h2 h→0

Hence, the integral converges.

(c) Let the integral ZZ
ln x2 + y 2 dx dy

where R is the region x2 + y 2 ≥ 1 be given. Here R is an unbounded closed region

and f (x, y) = ln(x2 + y 2 ) ≥ 0 and continuous in R. Again, employing polar
coordinates, integration by parts and (4.61) the integral over R can be defined as
the limit
ZZ Z 2π Z k
2 2
r ln r2 dr dθ


lim ln x + y dx dy = lim
k→∞ k→∞ 0 1
R
1 2π
Z
k
lim r2 ln r2 − r2 1 dθ

=
2 0 k→∞
= π lim k 2 ln k 2 − k 2 + 1 = −∞


k→∞

where the limit limk→∞ k 2 ln k 2 is evaluated using L’Hospital’s rule:

ln k 2 2/k
lim k 2 ln k 2 = lim 2
= lim = lim −k 2 = −∞
k→∞ k→∞ 1/k k→∞ −2/k 3 k→∞

Hence, the integral diverges.

42
(d) Let the integral
ZZ p 2
x + xy + y 2
dx dy
x2 + y 2
R
22
where R is the region x + yp ≤ 1 be given. Here the region R is a unbounded
closed region and f (x, y) = x2 + xy + y 2 /(x2 + y 2 ) ≥ 0 and continuous in R.
Again, employing polar coordinates and (4.61) the integral over R can be defined
as the limit
ZZ p 2 Z 2π Z 1 √
x + xy + y 2
lim dx dy = lim 1 + sin θ cos θ dr dθ
h→0 x2 + y 2 h→0 0 h
R
Z 2π √ 1
= lim 1 + sin θ cos θr h dθ
h→0
Z0 2π √
= 1 + sin θ cos θ lim (1 + h) dθ
0 h→0
Z 2π √
= 1 + sin θ cos θ dθ
0

The resulting ordinary integral in θ has a finite value, hence, the original double
integral converges.
(e) Let the integral ZZZ
ln x2 + y 2 + z 2 dx dy dz

where R is the solid x + y 2 + z 2 ≤ 1 be given. Here R is an unbounded closed

2

region and f (x, y, z) = ln(x2 + y 2 + z 2 ) ≤ 0 and continuous in R. Employing

spherical coordinates, integration by parts and (4.66) the integral over R can be
defined as the limit
ZZZ Z 2π Z π Z 1
2 2 2
ρ2 sin φ ln ρ2 dρ dφ dθ



lim ln x + y + z dx dy dz = lim
h→0 h→0 0 0 h
R
1
2 2π π ρ3
Z Z 
3
= sin φ lim ρ ln ρ − dφ dθ
3 0 0 h→0 3 h
2 2π π
Z Z  3 
h 3 1
= sin φ lim − h ln h − dφ dθ
3 0 0 h→0 3 3
2 2π π 2 2π
Z Z Z
π
=− sin φ dφ dθ = cos 0 dθ
9 0 9 0
Z 2π 0
4 8π
=− dθ = −
9 0 9
Hence, the integral converges.

43
Section 4.9
1. (a) Z π Z π Z π
d cos xt ∂ cos xt
dx = dx = − sin xt dx
dt π/2 x π/2 ∂t x π/2

(b)
2 2 2
x2 x2 x3
Z Z Z
d ∂
dx = dx = 2 dx
dt 1 (1 − tx)2 1 ∂t (1 − tx)2 1 (1 − tx)3
(c) Z 2 Z 2 Z 2
d ∂ dx
ln (xu) dx = ln (xu) dx =
du 1 1 ∂u 1 u
(d)
2 2 2
dn ∂ n sin x
Z Z Z
sin x sin x
dx = dx = n! dx
dy n 1 x−y 1 ∂y n x − y 1 (x − y)n+1

2. (a) By (4.10) Z x
d
t2 dt = x2
dx 1

(b) By (4.10) and the first term of (4.95)

Z t2 Z b(t)
d d
sin x dx = 2
sin x2 dx = sin [b (t)]2 b0 (t) = 2t sin t4
dt 1 dt 1

(c) By (4.10) and the first term of (4.95)

2 t3
d b(t)
Z Z Z
d 2 d 2
ln 1 + x2 dx




ln 1 + x dx = − ln 1 + x dx = −
dt t3 dt 2 dt 2
= −3t2 ln 1 + t6

(d) By (4.10) and the first term of (4.95)

Z tan x Z a Z tan x
d −t2 d −t2 d 2
e dt = e dt + e−t dt
dx x dx x dx a
Z x Z tan x
d −t2 d 2
=− e dt + e−t dt
dx a dx a
2 2
= e− tan x
sec2 x − e−x

44
3. (a) By (4.95)
Z cos α
d
ln (x + α) dx = − sin α ln (cos α + α) − cos α ln (sin α + α)
dα sin α
Z cos α
∂
+ ln (x + α) dx
sin α ∂α
Z cos α
dx
= − sin α ln (cos α + α) − cos α ln (sin α + α) +
sin α x + α
cos α
= − sin α ln (cos α + α) − cos α ln (sin α + α) + ln (x + α) sin α
cos α + α
= ln − [sin α ln (cos α + α) + cos α ln (sin α + α)]
sin α + α
(b) By (4.94) and (4.95) and using integration by parts
Z π/(2u) Z π/(2u)
d π ∂
u sin ux dx = − + u sin ux dx
du 0 2u 0 ∂u
Z π/(2u)
π
=− + (sin ux + ux cos ux) dx
2u 0
π/(2u) π/(2u)
π cos ux π/(2u) cos ux
=− − + x sin ux 0
+ =0
2u u 0 u 0

(c) By (4.95)
Z y2 Z y2
d −x2 y 2 −y 6 −y 4 ∂ −x2 y2
e dx = 2ye −e + e dx
dy y y ∂y
Z y2
−y 6 −y 4 2 y2
= 2ye −e − 2y x2 e−x dx
y

R1
4. (a) Using the fact that when f (x) = ax then f 0 (x) = ax ln a and that 0
xn dx =
1/(n + 1), n > −1 we find
Z 1
d d 1
xn dx =
dn 0 dn n + 1
Z 1
d n 1
x dx = −
0 dn (n + 1)2
Z 1
1
xn ln x dx = −
0 (n + 1)2

(b) Assuming a > 0 we firstly note that

Z ∞ Z b
−ax 1 b 1
1
e dx = lim e−ax dx = − lim e−ax 0
=− lim e−ab − 1 =
0 b→∞ 0 a b→∞ a b→∞ a

45
 and hence,
∞ ∞
dn ∂ n −ax dn 1
Z Z
−ax
e dx = e dx =
dan 0 0 ∂an dan a
Z ∞
n!
xn e−ax dx = n+1
0 a

(c) Using the substitution u = x2 , we firstly note that (assuming x > 0 and so u > 0)
Z ∞ Z b Z b b
dy dy 1 dy 1 y
= lim = 2 lim = √ lim tan−1 √
0 u + y 2 b→∞ 0 u+y 2 x b→∞ 0
2
1 + y /u u b→∞ u 0
1 b
= √ lim tan−1 √
u b→∞ u
π 1
= √
2 u

Next, we find
Z ∞ Z ∞ n−1 Z ∞
dn−1 dy ∂ dy n−1 dy
= = (−1) (n − 1)!
du n−1
0 u+y 2
0 ∂u n−1 u+y 2
0 (u + y 2 )n

for n = 1, 2, . . . . Also

dn−1 π 1 π (n − 1)! [1 × 3 × · · · × (2n − 3)] 1

n−1
√ = (−1)n−1 √
du 2 u 2 2 × 4 × · · · × (2n − 2) u2n−1
Equating both terms and substituting back for x then finally gives
Z ∞
dy π 1 × 3 × · · · × (2n − 3) 1
2 2 n =
0 (x + y ) 2 2 × 4 × · · · × (2n − 2) x2n−1

5. (a) Firstly, we note that (assuming a > 0)

Z Z
dx dx 1 −1 x
= = √ tan √ + C1
x2 + a 1 + x2 /a a a

Hence,

dn−1 ∂ n−1 dx ∂ n−1

Z Z  
dx 1 −1 x
= − C = n−1 √ tan √ + C1
dan−1 2
x +a ∂an−1 x2 + a ∂a a a
n−1
Z  
n−1 dx ∂ 1 −1 x
(−1) (n − 1)! = n−1 √ tan √ +C
(x2 + a)n ∂a a a
(−1)n−1 ∂ n−1
Z  
dx 1 −1 x
= √ tan √ +C
(x2 + a)n (n − 1)! ∂an−1 a a

46
 (b) Firstly, we note that Z
sin ax
cos ax dx = + C1
a
Hence,
dn ∂n ∂ n sin ax
Z Z  
cos ax dx = cos ax dx − C = n + C1
dan ∂an ∂a a
∂ n sin ax
Z
an cos ax dx = n +C
∂a a
for n = 4, 8, 12, . . . .
R
(c) Let f (x, t) dx = F (x, t) + C, so that ∂F/∂x = f (x, t). Substituting for f (x, t)
in Eq. (a) then gives
Z Z
∂ ∂
f (x, t) dx + C = f (x, t) dx
∂t ∂t
Z Z
∂ ∂F ∂ ∂F
dx + C = dx
∂t ∂x ∂x ∂t
Z Z
∂ ∂F ∂ ∂F
dx + C = dx
∂t ∂x ∂x ∂t
Z 2 Z 2
∂ F ∂ F
dx + C = dx
∂t∂x ∂x∂t
This last form is equivalent to the indefinite integral of
∂ 2F ∂ 2F
=
∂t∂x ∂x∂t
6. As the problem states, it is known that
Z 2π √
cos θ 1 − 1 − a2
dθ = 2π √
0 1 − a cos θ a 1 − a2
R 2π
where a is a constant, 0 < a < 1. Next, let g(a) be defined as g(a) = 0 ln(1 −
R 2π
a cos θ) dθ. Then g 0 (a) = − 0 cos θ/(1 − a cos θ) dθ. Hence, using the substitution
√
u = 1 − a2 so that du = −a(1 − a2 )−1/2 da:
Z Z 2π √
− 1 − a2
Z Z
cos θ 1
g (a) = g 0 (a) da = − da = −2π √ da
0 1 − a cos θ a 1 − a 2
Z Z Z
da da du
= 2π − 2π √ = 2π ln a + C1 − 2π
a 2 1 − u2
Za 1 − a Z
du du
= 2π ln a + C1 − π −π
1−u 1+u
 √   √ 
2
= 2π ln a − π ln 1 − 1 − a + π ln 1 + 1 − a + C 2
 √ 
= 2π ln 1 + 1 − a2 + C

47
 Using the continuity of g for a = 0 and g(0) = 0 it follows that C = −2π ln 2. As such,
we find
√
Z 2π  √  1 + 1 − a2
ln (1 − a cos θ) dθ = 2π ln 1 + 1 − a2 − 2π ln 2 = 2π ln
0 2

7. Let f (x, t) be a scalar associated with the flow of a fluid along the x-axis. The Stokes
derivative then is given by
Df ∂f dx ∂f ∂f ∂f
= + =v +
Dt ∂x dt ∂t ∂x ∂t
As the problem states, a piece of the fluid occupying an interval a0 ≤ x ≤ b0 when t = 0
will occupy an interval a(t) ≤ x ≤ b(t) at time t, where da/dt = v(a, t), db/dt = v(b, t).
The integral
Z b(t)
F (t) = f (x, t) dx
a(t)

is then an integral of f over a definite piece of the fluid, whose position varies with time.
Using (4.60), this integral can be written as an integral over the region a0 ≤ x ≤ b0 as
Z b(t) Z b0
dx
f (x, t) dx = f (x0 , t) dx0
a(t) a0 dx0
Differentiating both sides of this equation and applying Leibnitz’s rule and the chain
rule (since f (x, t) = f [φ(t), t] so that df /dt = ∂f /∂t + (∂f /∂φ)(dφ/dt)) then gives

d b(t) d b0
Z Z Z b(t)  
dx d dx
f (x, t) dx = f (x0 , t) dx0 = f (x0 , t) dx0
dt a(t) dt a0 dx0 a(t) dt dx0
Z b0  
∂f dx ∂f dx0 dx d dx
= (x0 , t) + + f (x0 , t) dx0
a0 ∂t dx0 ∂x0 dt dx0 dt dx0
Z b0  
∂f dx ∂f dv
= (x0 , t) +v + f (x0 , t) dx0
a0 ∂t dx0 ∂x0 dx0
Z b0  
∂f dx ∂
= (x0 , t) + (f v) dx0
a0 ∂t dx0 ∂x0
Z b(t)   Z b(t)  
∂f ∂ dx0 ∂f ∂
= (x, t) + (f v) dx = (x, t) + (f v) dx
a(t) ∂t ∂x0 dx a(t) ∂t ∂x
Z b(t)  
Df dv
= +f dx
a(t) Dt dx

8. Let f (α) be continuous for 0 ≤ α ≤ 2π and let

Z 2π
1 1 − r2
u (r, θ) = f (α) dα
2π 0 1 + r2 − 2r cos (θ − α)

48
 for r < 1, r and θ being polar coordinates. Next, let w = 1 + r2 − 2r cos(θ − α) and
v(r, θ, α) = (1 − r2 )w−1 . Applying Leibnitz’s rule to the right-hand side of the integral
equation above and using (2.138) we find
1 2 2π
 2  Z 2π
1 ∂2
Z
2 1 ∂ 1 ∂
∇ u= ∇ f (α) v dα = + + f (α) v dα
2π 0 2π ∂r2 r2 ∂θ2 r ∂r 0
Z 2π  2
1 ∂2

1 ∂ 1 ∂
= f (α) + + v dα
2π 0 ∂r2 r2 ∂θ2 r ∂r
Z 2π  2
1 ∂ 2 v 1 ∂v

1 ∂ v
= f (α) + + dα
2π 0 ∂r2 r2 ∂θ2 r ∂r
Z 2π
1
= f (α) ∇2 v dα
2π 0
Next, we compute all required partial derivatives:
vr = −2rw−1 − 1 − r2 w−2 wr

vrr = −2w−1 + 2rw−2 wr + 2rw−2 wr + 2 1 − r2 w−3 wr2 − 1 − r2 w−2 wrr

vθθ = 2 1 − r2 w−3 wθ2 − 1 − r2 w−2 wθθ

which gives
vθθ vr −1 −1

−2
∇2 v = vrr + + = −4w + 5r − r w wr
r2 r
+ r2 − 1 w−2 wrr − 2w−3 wr2 + r−2 w−2 wθθ − 2r−2 w−3 wθ2

Multiplying both sides by r2 w3 and substituting for w, wr . . . then gives

r2 w3 ∇2 v = −4r2 w2 + 5r2 − 1 rwwr + r2 − 1 r2 wwrr − 2r2 wr2 + wwθθ − 2wθ2

= (−4 + 10 + 2 − 8) r6 + (16 − 30 − 4 + 16 + 2) r5 cos (θ − α)

+ (−8 − 8 + 8 + 8) r4 + (16 − 4 + 4 − 16) r3 cos (θ − α) + (−4 − 2 − 2 + 8) r2
+ (2 − 2) r cos (θ − α)
=0
In conclusion, since r2 w3 ∇2 v = 0 so will ∇2 v = 0.

Section 4.11
1. (a) Let u = (u1 , u2 , . . . , un ) and v = (v1 , v2 , . . . , vn ) be two vectors in V n . Then we
have
|u| = |u − v + v| ≤ |u − v| + |v| ⇒ |u| − |v| ≤ |u − v|
|v| = |v − u + u| ≤ |v − u| + |u| ⇒ |u| − |v| ≥ −|u − v|
And so
−|u − v| ≤ |u| − |v| ≤ |u − v| ⇒ |u| − |v| ≤ |u − v|

49
 (b) Let x = (x1 , x2 , . . . , xn ) be a vector in E n , where 0 ≤ x1 , x2 , . . . , xn < ∞. Then
it follows immediately that
q
f (x) = |x| = x21 + x22 + · · · + x2n

it a continuous mapping of E n into E 1 . In order to determine if this mapping is

uniformly continuous, we look for a δ > 0 such that c = |x2 − x1 | < δ implies
|f (x1 ) − f (x2 )| <  for two different vectors x1 , x2 in E n . If |x2 − x1 | = c,
then, using part (a), |f (x1 ) − f (x2 )| = |x1 | − |x2 | ≤ |x1 − x2 | = c. Hence,
since |f (x1 ) − f (x2 )| ≤ c, we can always find a δ > 0 such that c < δ implies
|f (x1 )−f (x2 )| <  and so we conclude that the mapping f (x) is in fact a uniformly
continuous mapping of E n into E 1 for 0 ≤ x1 , x2 , . . . , xn < ∞.

2. (a) By Theorem K, the function y = f (x) = ex , 0 ≤ x ≤ 1 is uniformly continuous,

since y = f (x) is a continuous mapping of the bounded closed set G : 0 ≤ x ≤ 1
into E 1 .
(b) The function y = f (x) = ln x, 0 < x ≤ 1, is a continuous mapping of the set
G : 0 < x ≤ 1 into E 1 . However, this mapping is not uniformly continuous. In
order to prove this, let us take  = 1 and ask how close x1 , x2 must be so that
|f (x1 ) − f (x2 )| < 1. Substituting for f (x) gives
x1
|f (x1 ) − f (x2 )| = |ln x1 − ln x2 | < 1 or < e, 0 < x ≤ 1
x2
But we can choose x1 , x2 as close together as we wish while x1 /x2 is arbitrarily
large. Thus there is no δ > 0 such that c = |x1 −x2 | < δ implies |f (x1 )−f (x2 )| < 1.
For instance, let us approximate c = |x1 −x2 | as c = |x1 −x21 |, while letting x1 → 0.
Clearly c → 0, so that x1 , x2 are as close together as possibly desired. However,
x1 x1 1
lim = lim 2 = lim =∞>e
x1 →0 x2 x 1 →0 x1 x 1 →0 x1

(c) The function y = f (x) = ln x, 1 ≤ x < ∞, is a continuous mapping of the set

G : 1 ≤ x < ∞ into E 1 . This mapping is also uniformly continuous, since for the
set G : 1 ≤ x < ∞ we can always find a δ > 0 such that c = |x1 − x2 | < δ implies
|f (x1 ) − f (x2 )| < . To this end, let us define x2 = x1 + h, where 0 < h < δ, so
that c = h < δ. Again, taking  = 1 and substituting for f (x) gives
x1
|f (x1 ) − f (x2 )| = |ln x1 − ln (x1 + h)| < 1 or < e, 1 ≤ x < ∞
x1 + h
Now we let x1 → ∞ and let h → 0 and take limits to get
x1 1 1
lim
x1 →∞
= xlim
→∞
= =1<e
h→0
x1 + h 1
h→0
1 + h/x1 1 + (0) (0)

50
 Similarly, we can choose to define x1 = x2 + h and let x2 → ∞, which gives
 
x2 + h h
lim
x2 →∞
= xlim 1+ = 1 + (0) (0) = 1 < e
x2 2 →∞ x2
h→0 h→0

This concludes the prove for the set G being right-open. Furthermore, G is
bounded and closed from the left and hence, by Theorem K and the above analysis
we may conclude that y = f (x) is uniformly continuous.
(d) The function y = f (x) = sin x, −∞ < x < ∞, is a continuous mapping of the set
G : −∞ < x < ∞ onto E 1 . Now since the function f (x) = sin x is periodic with
period 2π, it suffices to show that f (x) is uniformly continuous for the bounded
closed (compact) set H ⊂ G : 0 ≤ x ≤ 2π. But then by Theorem K it follows
immediately that y = f (x) is uniformly continuous.

3. Let f (x) be defined for a < x < b, f 0 (x) be continuous for a < x < b and let f 0 (x)
be bounded. Then since f 0 (x) is bounded, there exists a scalar K > 0 such that
|f 0 (x)| ≤ K. Hence, utilising the mean value theorem: f (x2 ) − f (x1 ) = f 0 (ξ)(x2 − x1 ),
where x1 < ξ < x2 we find |f (x1 ) − f (x2 )| ≤ K|x2 − x1 | = Kc. As such, choosing
c < /K we can always find a δ > 0, such that c < δ implies |f (x1 ) − f (x2 )| < , i.e. δ
depends only on  and the bound K > 0 and not on the point x.

51