Answers Manual

Students Manual
Essential Mathematics for

Economic Analysis
3
rd
edition
Knut Sydster
Arne Strm
Peter Hammond
For further supporting resources please visit:
www.pearsoned.co.uk/sydsaeter
Preface
This students solutions manual accompanies Essential Mathematics for Economic Analysis (3rd edition, FT
Prentice Hall, 2008). Its main purpose is to provide more detailed solutions to the problems marked
SM
in the
text. This Manual should be used in conjunction with the answers in the text. In some few cases only a part
of the problem is done in detail, because the rest follows the same pattern. We are grateful to Carren Pindiriri
for help with the proofreading. We would appreciate suggestions for improvements from our readers as well
as help in weeding out inaccuracies and errors.
Oslo and Coventry, July 2008
Knut Sydster ([email protected])
Arne Strm ([email protected])
Peter Hammond ([email protected])
Version: 9 July 2008
Contents
1 Introductory Topics I: Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Introductory Topics II: Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3 Introductory Topics III: Miscellaneous . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
4 Functions of One Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
5 Properties of Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
6 Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
7 Derivatives in Use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
8 Single-Variable Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
9 Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
10 Interest Rates and Present Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
11 Functions of Many Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
12 Tools for Comparative Statics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
13 Multivariable Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
14 Constrained Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
15 Matrix and Vector Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
16 Determinants and Inverse Matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
17 Linear Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Knut Sydster, Arne Strm, and Peter Hammond 2008
CHAPTER 1 I NTRODUCTORY TOPI CS I : AL GEBRA 1
Chapter 1 Introductory Topics I: Algebra
1.1
1. (a) True (b) False. 5 is smaller than 3, so on the number line it is to the left of 3. (See Fig. 1.1.1
in the book.) (c) False. 13 is an integer, but not a natural number. (d) True. Any natural number is
rational. For example 5 = 5/1. (e) False, since 3.1415 = 31415/10000, the quotient of two integers.
(f) False. Counterexample:

2 +(
2) = 0. (g) True. (h) True.

1.3
9. (a) (2t 1)(t
2
2t +1) = 2t (t
2
2t +1)(t
2
2t +1) = 2t
3
4t
2
+2t t
2
+2t 1 = 2t
3
5t
2
+4t 1
(b) (a +1)
2
+(a 1)
2
2(a +1)(a 1) = a
2
+2a +1 +a
2
2a +1 2a
2
+2 = 4
(c) (x + y + z)
2
= (x + y + z)(x + y + z) = x(x + y + z) + y(x + y + z) + z(x + y + z) =
x
2
+xy +xz +yx +y
2
+yz +zx +zy +z
2
= x
2
+y
2
+z
2
+2xy +2xz +2yz (d) (x y z)
2
=
(xyz)(xyz) = x
2
xyxzxy+y
2
yzxzyz+z
2
, so (x+y+z)
2
(xyz)
2
= 4xy+4xz
13. (a) a
2
+4ab+4b
2
= (a+2b)
2
using the rst quadratic identity. (d) 9z
2
16w
2
= (3z4w)(3z+4w),
according to the difference-of-squares formula. (e)
1
5
x
2
+2xy 5y
2
=
1
5
(x
2
10xy +25y
2
) =
1
5
(x 5y)
2
(f) a
4
b
4
= (a
2
b
2
)(a
2
+ b
2
), using the difference-of-squares formula. Since
a
2
b
2
= (a b)(a +b), the answer in the book follows.
1.4
5. (a)
1
x 2

1
x +2
=
x +2
(x 2)(x +2)
x 2
(x +2)(x 2)
=
x +2 x +2
(x 2)(x +2)
=
4
x
2
4
(b) Since 4x +2 = 2(2x +1) and 4x
2
1 = (2x +1)(2x 1), the LCD is 2(2x +1)(2x 1). Then
6x +25
4x +2

6x
2
+x 2
4x
2
1
=
(6x +25)(2x 1) 2(6x
2
+x 2)
2(2x +1)(2x 1)
=
21(2x 1)
2(2x +1)(2x 1)
=
21
2(2x +1)
(c)
18b
2
a
2
9b
2

a
a +3b
+2 =
18b
2
a(a 3b) +2(a
2
9b
2
)
(a +3b)(a 3b)
=
a(a +3b)
(a +3b)(a 3b)
=
a
a 3b
(d)
1
8ab
1
8b(a +2)
+
1
b(a
2
4)
=
a
2
4 a(a 2) +8a
8ab(a
2
4)
=
2(5a 2)
8ab(a
2
4)
=
5a 2
4ab(a
2
4)
(e)
2t t
2
t +2

_
5t
t 2

2t
t 2
_
=
t (2 t )
t +2

3t
t 2
=
t (t 2)
t +2

3t
t 2
=
3t
2
t +2
(f)
a
_
1
1
2a
_
0.25
=
a
1
2
1
4
= 4a 2, so 2
a
_
1
1
2a
_
0.25
= 2 (4a 2) = 4 4a = 4(1 a)
6. (a)
2
x
+
1
x +1
3 =
2(x +1) +x 3x(x +1)
x(x +1)
=
2 3x
2
x(x +1)
(b)
t
2t +1

t
2t 1
=
t (2t 1) t (2t +1)
(2t +1)(2t 1)
=
2t
4t
2
1
(c)
3x
x +2

4x
2 x
2x 1
(x 2)(x +2)
=
3x(x 2) +4x(x +2) (2x 1)
(x 2)(x +2)
=
7x
2
+1
x
2
4
2 CHAPTER 1 I NTRODUCTORY TOPI CS I : AL GEBRA
(d)
1
x
+
1
y
1
xy
=
_
1
x
+
1
y
_
xy
1
xy
xy
=
y +x
1
= x +y (e)
1
x
2

1
y
2
1
x
2
+
1
y
2
=
_
1
x
2

1
y
2
_
x
2
y
2
_
1
x
2
+
1
y
2
_
x
2
y
2
=
y
2
x
2
x
2
+y
2
(f) Multiply the numerator and the denominator by xy. Then the fraction reduces to
a(y x)
a(y +x)
=
y x
y +x
.
8. (a)
1
4

1
5
=
5
20

4
20
=
1
20
, so
_
1
4

1
5
_
2
=
_
1
20
_
2
= 20
2
= 400
(b) n
n
1
1
n
= n
n n
_
1
1
n
_
n
= n
n
2
n 1
=
n(n 1) n
2
n 1
=
n
n 1
(c) If u = x
pq
, then
1
1 +x
pq
+
1
1 +x
qp
=
1
1 +u
+
1
1 +1/u
=
1
1 +u
+
u
1 +u
= 1
(d)
_
1
x 1
+
1
x
2
1
_
(x
2
1)
_
x
2
x +1
_
(x
2
1)
=
x +1 +1
x
3
x 2x +2
=
x +2
(x +2)(x
2
2x +1)
=
1
(x 1)
2
(e)
1
(x +h)
2

1
x
2
=
x
2
(x +h)
2
x
2
(x +h)
2
=
2xh h
2
x
2
(x +h)
2
, so
1
(x +h)
2

1
x
2
h
=
2x h
x
2
(x +h)
2
(f) Multiplying denominator and numerator by x
2
1 = (x +1)(x 1) yields
10x
2
5x(x 1)
=
2x
x 1
1.5
5. (Needs some hints.) Multiply the denominator and the numerator by: (a)

7

5 (b)

5

3
(c)
3 +2 (d) x
y y
x (e)
x +h +
x (f) 1
x +1
7. The answers will depend on the calculator you use.
12. (a) For x = 1 the left-hand side is 4 and the right-hand side is 2. (In fact, (2
x
)
2
= 2
2x
.) (b) Correct
because a
pq
= a
p
/a
q
(c) Correct because a
p
= 1/a
p
(d) For x = 1 it says 5 = 1/5, which is
absurd. (e) For x = y = 1, it says that a
2
= 2a, which is usually wrong. (In fact, a
x+y
= a
x
a
y
.)
(f) 2
x
2
y
= 2
x+
y
, not 2
xy
.
1.6
4. (a) 2 <
3x +1
2x +4
has the same solutions as
3x +1
2x +4
2 > 0, or
3x +1 2(2x +4)
2x +4
> 0, or
x 7
2x +4
> 0
A sign diagram reveals that the inequality is satised for 7 < x < 2. A serious error is to multiply
the inequality by 2x + 4, without assuming that 2x + 4 > 0. When multiplying with 2x + 4 when this
number is negative, the inequality sign must be reversed. (It might be a good idea to test the inequality
for some values of x. For example, for x = 0 it is not true. What about x = 5?)
(b) The inequality is equivalent to
120
n
3
4
0, i.e.
3(160 n)
4n
0. A sign diagram reveals that the
inequality is satised for n < 0 and for n 160. (Note that for n = 0 the inequality makes no sense.
For n = 160, we have equality.)
CHAPTER 1 I NTRODUCTORY TOPI CS I : AL GEBRA 3
(c) Easy: g(g 2) 0 etc. (d) Note that p
2
4p + 4 = (p 2)
2
, and the inequality reduces to
p +1
(p 2)
2
0. The fraction makes no sense if p = 2. The conclusion follows.
(e) The inequality is equivalent to
n 2
n +4
2 0, i.e.
n 2 2n 8
n +4
0, or
3n 10
n +4
0, etc.
(f) See the text and use a sign diagram. (Dont cancel x
2
. If you do, x = 0 appears as a false solution.)
5. (a) Use a sign diagram. (b) The inequality is not satised for x = 1. If x = 1, it is obviously satised
for x + 4 > 0, i.e. x > 4 (because (x 1)
2
is positive when x = 1). (c) Use a sign diagram.
(d) The inequality is not satised for x = 1/5. If x = 1/5, it is obviously satised for x < 1.
(e) Use a sign diagram. ((5x 1)
11
< 0 if x < 1/5, > 0 if x > 1/5.)
(f)
3x 1
x
> x +3,
3x 1
x
(x +3) > 0,
(1 +x
2
)
x
> 0, so x < 0. (1 +x
2
is always positive.)
(g)
x 3
x +3
> 2x 1 or
x 3
x +3
(2x 1) < 0 or
2x(x +2)
x +3
< 0. Then use a sign diagram.
(h) Use the hint and a sign diagram. (Actually, this problemand the next could be postponed to Section 2.3
if you have forgotten your high school algebra.) (i) Use the hint and a sign diagram.
Review Problems for Chapter 1
4. (a) (2x)
4
= 2
4
x
4
= 16x
4
(b) 2
1
4
1
= 1/2 1/4 = 1/4, so (2
1
4
1
)
1
= 4
(c) Cancel the common factor 4x
2
yz
2
. (d) (ab
3
)
3
= (1)
3
a
3
b
9
= a
3
b
9
, so
[(ab
3
)
3
(a
6
b
6
)
2
]
3
= [a
3
b
9
a
12
b
12
]
3
= [a
9
b
3
]
3
= a
27
b
9
(e)
a
5
a
3
a
2
a
3
a
6
=
a
6
a
3
= a
3
(f)
__
x
2
_
3
8
x
2
_
3
=
_
x
3
8
8 x
2
_
3
= (x
5
)
3
= x
15
8. All are straightforward, except (c), (g), and (h): (c)
3
_
3
6
_
= 3 +
6 = 3 +
2
= 3 + 3
2 (g) (1 + x + x
2
+ x
3
)(1 x) = (1 + x + x
2
+ x
3
) (1 + x + x
2
+ x
3
)x = 1 x
4
(h) (1 +x)
4
= (1 +x)
2
(1 +x)
2
= (1 +2x +x
2
)(1 +2x +x
2
) a.s.o.
11. (a) and (b) are easy. (c) ax +ay +2x +2y = ax +2x +ay +2y = (a+2)x +(a+2)y = (a+2)(x +y)
(d) 2x
2
5yz +10xz xy = 2x
2
+10xz (xy +5yz) = 2x(x +5z) y(x +5z) = (2x y)(x +5z)
(e) p
2
q
2
+pq = (pq)(p+q) +(pq) = (pq)(p+q +1) (f) See the answer in the book.
15. (a)
s
2s 1

s
2s +1
=
s(2s +1) s(2s 1)
(2s 1)(2s +1)
=
2s
4s
2
1
(b)
x
3 x
1 x
x +3

24
x
2
9
=
x(x +3) (1 x)(x 3) 24
(x 3)(x +3)
=
7(x +3)
(x 3)(x +3)
=
7
x 3
(c) Multiplying numerator and denominator by x
2
y
2
yields,
y x
y
2
x
2
=
y x
(y x)(y +x)
=
1
x +y
16. (a) Cancel the factor 25ab. (b) x
2
y
2
= (x + y)(x y). Cancel x + y. (c) The fraction can be
written
(2a 3b)
2
(2a 3b)(2a +3b)
=
2a 3b
2a +3b
. (d)
4x x
3
4 4x +x
2
=
x(2 x)(2 +x)
(2 x)
2
=
x(2 +x)
2 x
4 CHAPTER 2 I NTRODUCTORY TOPI CS I I : EQUATI ONS
Chapter 2 Introductory Topics II: Equations
2.1
3. (a) We note rst that x = 3 and x = 4 both make the equation absurd. Multiplying the equation by
the common denominator, (x + 3)(x + 4), yields (x 3)(x + 4) = (x + 3)(x 4), and thus x = 0.
(b) Multiplying by the common denominator (x 3)(x + 3) yields 3(x + 3) 2(x 3) = 9, from
which we get x = 6. (c) Multiplying by the common denominator 15x (assuming that x = 0, yields
18x
2
75 = 10x
2
15x +8x
2
, from which we get x = 5.
5. (a) Multiplying by the common denominator 12 yields 9y 3 4 +4y +24 = 36y, and so y = 17/23.
(b) Multiplying by 2x(x + 2) yields 8(x + 2) + 6x = 2(2x + 2) + 7x, from which we nd x = 4.
(c) Multiplying the numerator and the denominator in the rst fraction by 1z, leads to
2 2z z
(1 z)(1 +z)
=
6
2z +1
. Multiplying by (1 z
2
)(2z +1) yields (2 3z)(2z +1) = 6 6z
2
, and so z = 4.
(d) Expanding the parentheses we get
p
4

3
8

1
4
+
p
12

1
3
+
p
3
=
1
3
. Multiplying by the common
denominator 24 gives an equation with the solution p = 15/16.
2.2
2. (a) Multiply both sides by abx to obtain b+a = 2abx. Hence, x =
a +b
2ab
=
a
2ab
+
b
2ab
=
1
2
_
1
a
+
1
b
_
.
(b) Multiply the equation by cx + d to obtain ax + b = cAx + dA, or (a cA)x = dA b, and thus
x = (dAb)/(a cA). (c) Multiply the equation by x
1/2
to obtain
1
2
p = wx
1/2
, thus x
1/2
= p/2w,
so, by squaring each side, x = p
2
/4w
2
. (d) Multiply each side by
1 +x to obtain 1 +x +ax = 0,
so x = 1/(1 +a). (e) x
2
= b
2
/a
2
, so x = b/a. (f) We see immediately that x = 0.
4. (a) x a = x b ( )x = a b, so x = (a b)/( ).
(b) Squaring each side of
pq = 3q +5 yields pq = (3q +5)

2
, so p = (3q +5)
2
/q.
(c) Y = 94 +0.2(Y (20 +0.5Y)) = 94 +0.2Y 4 0.1Y, so 0.9Y = 90, and then Y = 100.
(d) Raise each side to the 4th power: K
2 1
2
r
w
K = Q
4
, so K
3
= 2wQ
4
/r, and hence K =
_
2wQ
4
/r
_
1/3
.
(e) Multiplying numerator and denominator in the left-hand fraction by 4K
1/2
L
3/4
, leads to 2L/K = r/w,
fromwhich we get L = rK/2w. (f) Raise each side to the 4th power:
1
16
p
4
K
1
_
1
2
r
w
_
= r
4
. It follows
that K
1
= 32r
3
w/p
4
, so K =
1
32
p
4
r
3
w
1
.
5. (a)
1
s
=
1
t
1
T
=
T t
t T
, so s =
t T
T t
. (b)

KLM = B + L, so KLM = (B + L)
2
, and
so M = (B +L)
2
/KL. (c) Multiplying each side by x z yields, x 2y + xz = 4xy 4yz, or
(x+4y)z = 4xyx+2y, and so z = (4xyx+2y)/(x+4y). (d) V = CCT/N, so CT/N = CV
and thus T = N(1 V/C).
2.3
5. (a) See the answer in the book. (b) If the rst natural number is n, then the next is n + 1, so the
requirement is that n
2
+(n +1)
2
= 13, which reduces to 2n
2
+2n 12 = 0, i.e. n
2
+n 6 = 0. This
second-order equation has the solutions n = 3 and n = 2, so the two numbers are 2 and 3. (If we asked
for integer solutions, we would have 3 and 2 in addition.)
CHAPTER 2 I NTRODUCTORY TOPI CS I I : EQUATI ONS 5
(c) If the shortest side is x, the other is x + 14, so according to Pythagoras Theorem (see page 633
and draw a picture), x
2
+ (x + 14)
2
= (34)
2
, or x
2
+ 14x 480 = 0. The solutions are x = 16 and
x = 30, so the shortest side is 16 cm and the longest is 30 cm. (d) If the usual driving speed is x
km/h and the usual time spent is t hours, then xt = 80. 16 minutes is 16/60 = 4/15 hours, so driving at
the speed x +10 for t 4/15 hours gives (x +10)(t 4/15) = 80. From the rst equation, t = 80/x.
Inserting this into the second equation, we get (x + 10)(80/x 4/15) = 80. Rearranging, we obtain
x
2
+10x 3000 = 0, whose positive solution is x = 50. So his usual driving speed is 50 km/h.
2.4
4. (a) If the two numbers are x and y, then x + y = 52 and x y = 26. Adding the two equations gives
2x = 78, so x = 39, and then y = 52 39 = 13. (b) Let the cost of one chair be $x and the cost of
one table $y. Then 5x +20y = 1800 and 2x +3y = 420. Solving this system yields x = 120, y = 60.
(c) Units produced of B: x. Then x +
1
2
x =
3
2
x units are produced of A, and 300
3
2
x +200x = 13 000,
or 650x = 13 000, so x = 20. Thus, 30 of quality A and 20 of quality B should be produced.
(d) If she invests $x at 15 % and $y at 20 %, then x +y = 1500 and 0.15x +0.2y = 275. The solution
is x = 8000 and y = 2000.
2.5
2. (a) The numerator 5 +x
2
is never 0, so there are no solutions. (b) The equation is obviously equivalent
to
x
2
+1 +2x
x
2
+1
= 0, or
(x +1)
2
x
2
+1
= 0, so x = 1. (c) x = 1 is clearly no solution. Multiply
the equation by (x + 1)
2/3
. Then the denominator becomes x + 1
1
3
x, which is 0 for x = 3/2.
(d) Multiplying by x 1 and rearranging yields x(2x 1) = 0, and so x = 0 or x = 1/2.
3. (a) z = 0 satises the equation. If z = 0, canceling z
2
yields z a = za +zb, or z(1 (a +b)) = a. If
a +b = 1 we have a contradiction. If a +b = 1, z = a/(1 (a +b)). (b) The equation is equivalent
to (1 + )(x y) = 0, so = 1, = 0, or x = y. (c) = 1 makes the equation meaningless.
Multiplying the equation by 1
2
yields (1 ) = , or (2 ) = 0, so = 0 or = 2.
(d) The equation is equivalent to b(1 +)(a 2) = 0, so b = 0, = 1, or a = 2.
2. See Problem 2.1.3.
3. (a) x =
2
3
(y 3) +y =
2
3
y 2 +y =
5
3
y 2, or
5
3
y = x +2, so y =
3
5
(x +2).
(b) ax cx = b +d, or (a c)x = b +d, so x = (b +d)/(a c).
(c)
L = Y
0
/AK, so squaring each side yields L = (Y
0
/AK)
2
. (d) qy = mpx, so y = (mpx)/q.
(e) and (f): See the answers in the text.
5. (a) Multiply the equation by 5K
1/2
to obtain K
1/2
= 15L
1/3
. Squaring each side gives K = 225L
2/3
.
(b) Raise each side to the power 1/t to obtain 1 +r/100 = 2
1/t
, and so r = 100(2
1/t
1).
(c) abx
b1
0
= p, so x
b1
0
= p/ab. Now raise each side to the power 1/(b 1).
(d) Raise each side to the power to get (1 )a
+b
= c
, or b
=
1
(c
(1 )a
).
Now raise each side to the power 1/.
9. (a) See the answer in the text. (b) Let u = 1/x and v = 1/y. Then the system reduces to 3u +2v = 2,
2u 3v = 1/4, with solution u = 1/2, v = 1/4. It follows that x = 1/u = 2 and y = 1/v = 4.
(c) See the answer in the text.
6 CHAPTER 3 I NTRODUCTORY TOPI CS I I : MI SCEL L ANEOUS
Chapter 3 Introductory Topics II: Miscellaneous
3.1
3. (a)(d): Look at the last term and replace n by k. Sum over k from 1 to n. (e) The coefcients are the
powers 3
n
for n = 1, 2, 3, 4, 5, so the general term is 3
n
x
n
. (f) and (g) see answers in the text.
(h) This is tricky. One has to see that each term is 198 larger that the previous term. (The problem is
related to the story about Gauss on page 56.)
7. (a)
n
k=1
ck
2
= c 1
2
+c 2
2
+ +c n
2
= c(1
2
+2
2
+ +n
2
) = c
n
k=1
k
2
(b) Wrong even for n = 2: The left-hand side is (a
1
+a
2
)
2
= a
2
1
+2a
1
a
2
+a
2
2
, but the right-hand side
is a
2
1
+a
2
2
. (c) Both sides equal b
1
+b
2
+ +b
N
. (d) Both sides equal 5
1
+5
2
+5
3
+5
4
+5
5
.
(e) Both sides equal a
2
0,j
+ +a
2
n1,j
. (f) Wrong even for n = 2: The left-hand side is a
1
+a
2
/2, but
the right-hand side is (1/k)(a
1
+a
2
).
3.2
5. One does not have to use summation signs. The sum is a +(a +d) +(a +2d) + +(a +(n 1)d.
There are n terms. The sum of all the as is na. The rest is d(1 +2 + +n 1). Then use formula (4).
3.3
1. (a) See the text. (b)
2
s=0
4
r=2
_
rs
r+s
_
2
=
2
s=0
__
2s
2+s
_
2
+
_
3s
3+s
_
2
+
_
4s
4+s
_
2
_
=
_
2
3
_
2
+
_
3
4
_
2
+
_
4
5
_
2
+
_
4
4
_
2
+
_
6
5
_
2
+
_
8
6
_
2
= 5 +
3113
3600
(c)
m
i=1
n
j=1
i j
2
=
m
i=1
i
n
j=1
j
2
=
1
2
m(m+1)
1
6
n(n+1)(2n+1) =
1
12
m(m+1)n(n+1)(2n+1),
where we used (4) and (5).
4. a is the mean of the a
s
s because a =
1
n
n
s=1
_
1
m
m
r=1
a
rs
_
=
1
n
n
s=1
a
s
.
To prove (), note that because a
rj
a is independent of the summation index s, it is a common factor
when we sum over s, so
m
s=1
(a
rj
a)(a
sj
a) = (a
rj
a)
m
s=1
(a
sj
a) for each r. Next, summing
over r gives
m
r=1
m
s=1
(a
rj
a)(a
sj
a) =
_
m
r=1
(a
rj
a)
__
m
s=1
(a
sj
a)
_
()
Using the properties of sums and the denition of a
j
, we have
m
r=1
(a
rj
a) =
m
r=1
a
rj

m
r=1
a = m a
j
m a = m( a
j
a)
Similarly, replacing r with s as the index of summation, one also has

m
s=1
(a
sj
a) = m( a
j
a).
Substituting these values into () then conrms ().
3.4
6. (a) If (i)
x 4 =
x +5 9, then also (ii) x 4 = (
x +5 9)
2
, which we get by squaring both
sides in (i). Calculating the square on the right-hand side of (ii) gives
x +5 = 5, and so x +5 = 25,
CHAPTER 3 I NTRODUCTORY TOPI CS I I : MI SCEL L ANEOUS 7
i.e. x = 20. This shows that if x is a solution of (i), then x = 20. No other value of x can satisfy (i).
But if we check this solution, we nd that with x = 20 the LHS of (i) becomes
16 = 4, and the RHS

becomes
25 9 = 5 9 = 4. Thus the LHS and the RHS are different. This means that equation (i)
actually has no solutions at all. (But note that 4
2
= (4)
2
, i.e. the square of the LHS equals the square
of the RHS. That is how the spurious solution x = 20 managed to sneak in.)
(b) If x is a solution of (iii)
x 4 = 9
x +5, then just as in part (a) we nd that x must be a solution

of (iv) x 4 = (9
x +5 )
2
. Now, (9
x +5 )
2
= (
x +5 9)
2
, so equation (iv) is equivalent
to equation (ii) in part (a). This means that (iv) has exactly one solution, namely x = 20. Inserting this
value of x into equation (iii), we nd that x = 20 is a solution of (iii).
A geometric explanation of the results can be given with reference to the following gure.
y
-5
5
x
5 10 15 20 25
y = 9
x +5
y =
x 4
y =
x +5 9
Figure SM3.4.6
We see that the two solid curves in the gure have no point in common, that is, the expressions
x 4
and
x +59 are not equal for any value of x. (In fact, the difference
x 4(
x +59) increases
with x, so there is no point of intersection farther to the right, either.) This explains why the equation
in (a) has no solution. The dashed curve y = 9
x +5, on the other hand, intersects y =
x +5 for
x = 20 (and only there), and this corresponds to the solution in part (b).
Comment: In part (a) it was necessary to check the result, because the transition from (i) to (ii) is only
an implication, not an equivalence. Similarly, it was necessary to check the result in part (b), since the
transition from (iii) to (iv) also is only an implication at least, it is not clear that it is an equivalence.
(Afterwards, it turned out to be an equivalence, but we could not know that until we had solved the
equation.)
7. (a) Here we have iff since
4 = 2. (b) It is easy to see by means of a sign diagram that x(x +3) < 0
precisely when x lies in the open interval (3, 0). Therefore we have an implication from left to right
(that is, only if), but not in the other direction. (For example, if x = 10, then x(x + 3) = 130.)
(c) x
2
< 9 3 < x < 3, so x
2
< 9 only if x < 3. If x = 5, for instance, we have x < 3 but
x
2
> 9. Hence we cannot have if here. (d) x
2
+ 1 is never 0, so we have iff here. (e) If x > 0,
then x
2
> 0, but x
2
> 0 also when x < 0. (f) x
4
+y
4
= 0 x = 0 and y = 0. If x = 0 and, say,
y = 1, then x
4
+y
4
= 1, so we cannot have if here.
9. (a) If x and y are not both nonnegative, at leat one of them must be negative, i.e. x < 0 or y < 0.
(b) If not all x are greater than or equal to a, at least one x must be less than a. (c) At least one of them
is less than 5. (Would it be easier if the statement to negate were Neither John nor Diana is less than
5 years old?) (d)(f) See the answers in the text.
8 CHAPTER 4 F UNCTI ONS OF ONE VARI ABL E
3.7
3. For n = 1, both sides are 1/2. Suppose () is true for n = k. Then the sum of the rst k +1 terms is
1
1 2
+
1
2 3
+
1
3 4
+ +
1
k(k +1)
+
1
(k +1)(k +2)
=
k
k +1
+
1
(k +1)(k +2)
But
k
k +1
+
1
(k +1)(k +2)
=
(k +1)
2
(k +1)(k +2)
=
k +1
k +2
, which is () for n = k +1. Thus, by induction,
() is true for all n.
4. The claim is true for n = 1. As the induction hypothesis, suppose k
3
+(k +1)
3
+(k +2)
3
is divisible
by 9. Note that (k + 1)
3
+ (k + 2)
3
+ (k + 3)
3
= (k + 1)
3
+ (k + 2)
3
+ k
3
+ 9k
2
+ 27k + 27 =
k
3
+(k +1)
3
+(k +2)
3
+9(k
2
+3k +3). This is divisible by 9 because the induction hypothesis implies
that the sum of the rst three terms is divisible by 9, whereas the last term is also obviously divisible by 9.
6. (b) false (because x
2
= 16 also has the solution x = 4), true, because if x = 4, then x
2
= 16.
(c) true, false because with y > 2 and x = 3, (x 3)
2
(y + 2) = 0. (d) and both true,
since the equation x
3
= 8 has the solution x = 2 and no others. (In the terminology of Section 6.3,
f (x) = x
3
is strictly increasing. See Problem 6.3.3 and see the graph Fig. 7, page 88.)
9. Consider Fig. A3.6.8, page 643 in the book, and let n
k
denote the number of students in the set marked S
k
,
for k = 1, 2, . . . , 8. Sets A, B, and C refer to those who study English, French, and Spanish, respectively.
Since 10 students take all three languages, n
7
= 10. There are 15 who take French and Spanish, so
15 = n
2
+ n
7
, and thus n
2
= 5. Furthermore, 32 = n
3
+ n
7
, so n
3
= 22. Also, 110 = n
1
+ n
7
, so
n
1
= 100. The rest of the information implies that 52 = n
2
+n
3
+n
6
+n
7
, so n
6
= 5252210 = 15.
Moreover, 220 = n
1
+n
2
+n
5
+n
7
, so n
5
= 220100510 = 105. Finally, 780 = n
1
+n
3
+n
4
+n
7
,
so n
4
= 780 100 22 10 = 648. The answers to the problems are:
(a): n
1
= 100 (b): n
3
+n
4
= 648 +22 = 670 (c) 1000
8
i=1
n
i
= 1000 905 = 95
Chapter 4 Functions of One Variable
4.2
1. (a) f (0) = 0
2
+ 1 = 1, f (1) = (1)
2
+ 1 = 2, f (1/2) = (1/2)
2
+ 1 = 1/4 + 1 = 5/4,
and f (
2) = (
2)
2
+ 1 = 2 + 1 = 3. (b) (i) Since (x)
2
= x
2
, f (x) = f (x) for all x.
(ii) f (x+1) = (x+1)
2
+1 = x
2
+2x+1+1 = x
2
+2x+2 and f (x)+f (1) = x
2
+1+2 = x
2
+3. Thus
equality holds if and only if x
2
+2x +2 = x
2
+3, i.e. if and only if x = 1/2. (iii) f (2x) = (2x)
2
+1 =
4x
2
+1 and 2f (x) = 2x
2
+2. Now, 4x
2
+1 = 2x
2
+2 x
2
= 1/2 x =
1/2 =
1
2
2.
10. (a) No: f (2 +1) = f (3) = 18, whereas f (2) +f (1) = 10. (b) Yes: f (2 +1) = f (2) +f (1) = 9.
(c) No: f (2 +1) = f (3) =
3 1.73, whereas f (2) +f (1) =
2 +1 2.41.
13. (a) We must require 5 x 0, so x 5. (b) The denominator x
2
x = x(x 1) must be different
from 0, so x = 0 and x = 1. (c) To begin with, the denominator must be nonzero, so we must require
x = 2 and x = 3. Moreover, since we can only take the square root of a nonnegative number, the
fraction (x 1)/(x 2)(x +3) must be 0. A sign diagram reveals that D
f
= (3, 1] (2, ). Note
in particular that the function is dened with value 0 at x = 1.
CHAPTER 4 F UNCTI ONS OF ONE VARI ABL E 9
15. Since g is obviously dened for x 2, D
g
= [2, ). Note that g(2) = 1, and g(x) 1 for all
x D
f
. As x increases from 2 to , g(x) decreases from 1 to , so R
g
= (, 1].
4.4
3. If D = a + bP, then 200 = a + 10b, and 150 = a + 15b. Solving for a and b yields a = 300 and
b = 10, so D = 300 10P.
4. L
1
: The slope is obviously 1, and the point-slope formula with (x
1
, y
1
) = (0, 2) and a = 1 give y = x+2.
L
2
: Using the point-point formula with (x
1
, y
1
) = (0, 3) and (x
2
, y
2
) = (5, 0) yields: y 3 =
0 3
5 0
x,
or y =
3
5
x +3. L
3
: Has slope 0 and equation y = 1. For L
4
and L
5
see the text.
10. The set of points that satisfy the inequality 3x + 4y 12 are those on or below the straight line
3x +4y = 12 as explained in Example 6 for a similar inequality. Those points that satisfy the inequality
x y 1, or equivalently, y x 1, are those on or above the straight line x y = 1. Finally, those
points that satisfy the inequality 3x + y 3, or equivalently, y 3 3x, are those on or above the
straight line 3x + y = 3. The set of points that satisfy all these three inequalities simultaneously, is the
set shown in Fig. A.4.4.10.
4.5
3. The pointpoint formula gives C 200 =
275 200
150 100
(x 100), or C =
3
2
x +50.
4.6
2. Complementing the answers in the text. (c) Formula (4) with a =
1
2
and b = 1 gives x = 1 as the
maximumpoint. (Alternatively, completingthe square, f (x) =
1
2
(x
2
+2x3) =
1
2
(x
2
+2x+14) =
1
2
(x +1)
2
+2, fromwhich we see immediately that f (x) has maximum2 at x = 1.) (e) Use (2.3.5),
or expand, to verify the formula for f (x). Use a sign diagram to study the sign variation of f (x).
6. Expanding we get U(x) = (1 + r
2
)x
2
+ 8(r 1)x. Then apply (4.6.4) with a = (1 + r
2
) and
b = 8(r 1).
9. (b) If B
2
4AC > 0, then according to formula (2.3.4), the equation f (x) = Ax
2
+ Bx + C = 0
would have two distinct solutions, which is impossible when f (x) 0 for all x. We nd that A =
a
2
1
+a
2
2
+ +a
2
n
, B = 2(a
1
b
1
+a
2
b
2
+ +a
n
b
n
), and C = b
2
1
+b
2
2
+ +b
2
n
, so the conclusion
follows easily.
4.7
1. (a) Integer roots must divide 6. Thus 1, 2, 3, and 6 are the only possible integer solutions. We
nd that 2, 1, 1, 3 all are roots, and since there can be no more than 4 roots in a polynomial equation
of degree 4, we have found them all.
(b) The same possible integer solutions. Only 6 and 1 are integer solutions. (The third root is 1/2.)
(c) Neither 1 nor 1 satises the equation, so there are no integer roots.
(d) First multiply the equation by 4 to have integer coefcients. Then 1, 2, and 4 are seen to be the
only possible integer solutions. In fact, 1, 2, 2 are all solutions.
10 CHAPTER 4 F UNCTI ONS OF ONE VARI ABL E
3. (a) The answer is 2x
2
+2x +4 +3/(x 1), because
(2x
3
+2x 1) (x 1) = 2x
2
+2x +4
2x
3
2x
2
2x
2
+2x 1
2x
2
2x
4x 1
4x 4
3 remainder
(b) The answer is x
2
+1, because
(x
4
+x
3
+x
2
+x) (x
2
+x) = x
2
+1
x
4
+x
3
x
2
+x
x
2
+x
0 no remainder
(c) The answer is x
3
4x
2
+3x +1 4x/(x
2
+x +1), because
(x
5
3x
4
+1) (x
2
+x +1) = x
3
4x
2
+3x +1
x
5
+ x
4
+ x
3
4x
4
x
3
+1
4x
4
4x
3
4x
2
3x
3
+4x
2
+1
3x
3
+3x
2
+3x
x
2
3x +1
x
2
+ x +1
4x remainder
(d) The answer is 3x
5
+6x
3
3x
2
+12x 12 +(28x
2
36x +13)/(x
3
2x +1), because
(3x
8
x
2
+ 1) (x
3
2x +1) = 3x
5
+6x
3
3x
2
+12x 12
3x
8
6x
6
+3x
5
6x
6
3x
5
+ x
2
+ 1
6x
6
12x
4
+ 6x
3
3x
5
+12x
4
6x
3
+ x
2
+ 1
3x
5
+ 6x
3
3x
2
12x
4
12x
3
+ 4x
2
+ 1
12x
4
24x
2
+12x
12x
3
+28x
2
12x + 1
12x
3
+24x 12
28x
2
36x +13 remainder
CHAPTER 4 F UNCTI ONS OF ONE VARI ABL E 11
4. (a) y =
1
2
(x + 1)(x 3). (Since the graph intersects the x-axis at the two points x = 1 and x =
3, we try the quadratic function f (x) = a(x + 1)(x 3). Then f (1) = 4a, and since the graph
passes through the point (1, 2),. f (1) = 2 = 4a. But the a = 1/2.) (b) Because the equation
f (x) = 0 has roots x = 3, 1, 2, we try the cubic function f (x) = b(x + 3)(x 1)(x 2). Then
f (0) = 6b. According to the graph, f (0) = 12. So b = 2, and hence y = 2(x +3)(x 1)(x 2).
(c) y =
1
2
(x +3)(x 2)
2
. (We try a polynomial of the form f (x) = c(x 2)
2
(x +3), with x = 2 as a
double root. Then f (0) = 12c. From the graph we see that f (0) = 6, and so a = 1/2.)
8. Polynomial division gives
(x
2
x ) (x +) = x ( + )
x
2
+x
( + )x
( + )x ( + )
( + ) remainder
and so
E =
_
x ( + ) +
( + )
x +
_
= x ( + ) +
( + )
x +
4.8
4. (a) C. The graph is a parabola and since the coefcient in front of x
2
is positive, it has a minimum point.
(b) D. The function is dened for x 2 and crosses the y-axis at y = 2
2 2.8.
(c) E. The graph is a parabola and since the coefcient in front of x
2
is negative, it has a maximum point.
(d) B. When x increases, y decreases, and y becomes close to 2 when x is large.
(e) A. The function is dened for x 2 and increases as x increases.
(f) F. Let y = 2 (
1
2
)
x
. Then y increases as x increases. For large values of x, y is close to 2.
5. (a) See the answer in the text. (b) 9
t
= (3
2
)
t
= 3
2t
and (27)
1/5
/3 = (3
3
)
1/5
/3 = 3
3/5
/3 = 3
2/5
, and
then 2t = 2/5, so t = 1/5.
4.9
10. Suppose y = Ab
x
, with b > 0. Then in (a), since the graph passes through the points (x, y) = (0, 2) and
(x, y) = (2, 8), we get 2 = Ab
0
, or A = 2, and 8 = 2b
2
, so b = 2. Hence, y = 2 2
x
.
In (b),
2
3
= Ab
1
and 6 = Ab. It follows that A = 2 and b = 3, and so y = 2 3
x
.
In (c), 4 = Ab
0
and 1/4 = Ab
4
. It follows that A = 4 and b
4
= 1/16, and so b = 1/2. Thus, y = 4(
1
2
)
x
.
4.10
3. (a) and (c) see the text. (b) Since ln x
2
= 2 ln x, 7 ln x = 6, so ln x = 6/7, and thus x = e
6/7
.
4. (a) ln(Ae
rt
) = ln(Be
st
), so ln A + rt = ln B + st , or (r s)t = ln(B/A), and so t =
1
r s
ln
B
A
.
(b) t =
1
0.09 0.02
ln
5.6 10
12
1.2 10
12
=
1
0.07
ln
14
3
22.
According to this, the two countries would have the same GNP in approximately 22 years, so in 2012.
12 CHAPTER 5 PROPERTI ES OF F UNCTI ONS
4. (a) We must have x
2
1, i.e. x 1 or x 1. (Look at Fig. 4.3.6, page 88.)
(b) The square root is dened if x 4, but x = 4 makes the denominator 0, so we must require x > 4.
(c) We must have (x 3)(5 x) 0, i.e. 3 x 5 (using a sign diagram).
7. (a) The point-slope formula gives y 3 = 3(x +2), or y = 3x 3.
(b) The two-point formula gives: y 5 =
7 5
2 (3)
(x (3)), or y = 2x/5 +31/5.
(c) y b =
3b b
2a a
(x a), or y = (2b/a)x b.
10. (1, 3) belongs to the graph if 3 = a + b + c, (0, 6) belongs to the graph if 6 = c, and (3, 15)
belongs to the graph if 15 = 9a +3b +c. It follows that a = 2, b = 1, and c = 6.
14. (a) p(x) = x(x
2
+x 12) = x(x 3)(x +4), because x
2
+x 12 = 0 for x = 3 and x = 4.
(b) 1, 2, 4, 8 are the only possible integer zeros. By trial and error we nd that q(2) = q(4) =
0, so 2(x 2)(x + 4) = 2x
2
+ 4x 16 is a factor for q(x). By polynomial division we nd that
q(x) (2x
2
+4x 16) = x 1/2, so q(x) = 2(x 2)(x +4)(x 1/2).
16. We use (4.7.5) and denote each polynomial by p(x). (a) p(2) = 8 2k = 0 for k = 4.
(b) p(2) = 4k
2
+2k 6 = 0 for k = 3/2 and k = 1. (c) p(2) = 26 +k = 0 for k = 26.
(d) p(1) = k
2
3k 4 = 0 for k = 1 and k = 4.
17. Since p(2) = 0, x 2 is a factor in p(x). We nd that p(x) (x 2) =
1
4
(x
2
2x 15) =
1
4
(x +3)(x 5), so x = 3 and x = 5 are the two other zeros. (Alternative: q(x) has the same zeros as
4p(x) = x
3
4x
2
11x +30. This polynomial can only have 1, 2, 3, 5, 10, 15, and 30 as
integer zeros. It is tedious work to nd the zeros in this way.)
21. (a) ln(x/e
2
) = ln x ln e
2
= ln x 2 (b) ln(xz/y) = ln(xz) ln y = ln x + ln z ln y
(c) ln(e
3
x
2
) = ln e
3
+ln x
2
= 3 +2 ln x for x > 0. (In general, ln x
2
= 2 ln |x|.) (d) See the text.
Chapter 5 Properties of Functions
5.1
3. The equilibrium condition is 106 P = 10 + 2P, and thus P = 32. The corresponding quantity is
Q = 106 32 = 74. See the graph in the answer section of the text.
6. f (y
d) = f (y
) c gives A(y
d) +B(y
d)
2
= Ay
+B(y
)
2
c, or Ay
Ad +B(y
)
2
2Bdy
+Bd
2
= Ay
+B(y
)
2
c. It follows that y
= [Bd
2
Ad +c]/2Bd.
5.2
4. If f (x) = 3x +7, then f (f (x)) = f (3x +7) = 3(3x +7) +7 = 9x +28. f (f (x
)) = 100 requires
9x
+28 = 100, and so x
= 8.
5.3
4. (a) f does have an inverse since it is one-to-one. This is shown in the table by the fact that all numbers in
the second row, the domain of f
1
, are different. The inverse assigns to each number in the second row,
the corresponding number in the rst row. (b) Since f (0) = 4 and f (x) increases by 2 for each unit
increase in x, f (x) = 2x +4. Solving y = 2x +4 for x yields x =
1
2
y 2, and thus f
1
(x) =
1
2
x 2.
CHAPTER 5 PROPERTI ES OF F UNCTI ONS 13
9. (a) (x
3
1)
1/3
= y x
3
1 = y
3
x = (y
3
+1)
1/3
. If we use x as the independent variable,
f
1
(x) = (x
3
+1)
1/3
. is the domain and range for both f and f
1
.
(b) The domain of f is all x = 2 Hence (with x = 2),
x +1
x 2
= y x + 1 = y(x 2)
(1 y)x = 2y 1 x =
2y 1
1 y
=
2y +1
y 1
. Using x as the independent variable, f
1
(x) =
(2x +1)/(x 1). The domain of the inverse is all x = 1.
(c) Here y = (1 x
3
)
1/5
+ 2 y 2 = (1 x
3
)
1/5
(y 2)
5
= 1 x
3
x
3
=
1 (y 2)
5
x = (1 (y 2)
5
)
1/3
. With x as the free variable, f
1
(x) = (1 (x 2)
5
)
1/3
.
is the domain and range for both f and f
1
.
10. (a) The domain is and the range is (0, ), so the inverse is dened on (0, ). From y = e
x+4
,
ln y = x + 4, so x = ln y 4, y > 0. (b) The range is , which is the domain of the inverse. From
y = ln x 4, ln x = y + 4, and then x = e
y+4
. (c) The domain is . y is increasing and when
x , y ln 2. Moreover, y as x . So the range of the function is (ln 2, ). From
y = ln
_
2 +e
x3
_
, 2 +e
x3
= e
y
, so e
x3
= e
y
3, and thus x = 3 +ln(e
y
3), y > ln 2.
11. We must solve x =
1
2
(e
y
e
y
) for y. Multiplythe equationbye
y
toget
1
2
e
2y
1
2
= xe
y
or e
2y
2xe
y
1 =
0. Letting e
y
= z yields z
2
2xz 1 = 0, with solution z = x
x
2
+1. The minus sign makes z
negative, so z = e
y
= x +
x
2
+1. This gives y = ln
_
x +
x
2
+1
_
as the inverse function.
5.4
1. (a) It is natural rst to see if the curve intersects the axes, by putting x = 0, and then y = 0. This gives 4
points. Then choose some values of
6 < x <
6, and compute the corresponding values of y. Argue

why the graph is symmetric about the x-axis and the y-axis. (The curve is called an ellipse. See the next
section.) (b) This graph is also symmetric about the x-axis and the y-axis. (If (a, b) lies on the graph,
so does (a, b), (a, b), and (a, b). (The graph is a hyperbola. See the next section.)
2. We see that we must have x 0 and y 0. If (a, b) lies on the graph, so does (b, a), so the graph is
symmetric about the line y = x. See the answer in the text.
5.5
4. (a) See the text. (b) Since the circle has centre at (2, 5), its equation is (x 2)
2
+(x 5)
2
= r
2
. Since
(1, 3) lies on the circle, (1 2)
2
+(3 5)
2
= r
2
, so r
2
= 13.
8. x
2
+y
2
+Ax+By+C = 0 x
2
+Ax+y
2
+By+C = 0 x
2
+Ax+
_
1
2
A
_
2
+y
2
+By+
_
1
2
B
_
2
=
1
4
(A
2
+ B
2
4C)
_
x +
1
2
A
_
2
+
_
y +
1
2
B
_
2
=
1
4
(A
2
+ B
2
4C). The last is the equation of a
circle centred at
_
1
2
A,
1
2
B
_
with radius
1
2
A
2
+B
2
4C. If A
2
+B
2
= 4C, the graph consists only
of the point
_
1
2
A,
1
2
B
_
. For A
2
+B
2
< 4C, the solution set is empty.
5.6
1. In each case, except (c), the rule denes a function because it associates with each member of the original
set a unique member in the target set. For instance, in (d), if the surface area of a sphere is given, its
volume is uniquely determined: From S = 4r
2
, r = (S/4)
1/2
, and then V =
4
3
r
3
=
4
3
(S/4)
3/2
(for the formulas for the surface area and the volume of a sphere of radius r, see page 632.)
14 CHAPTER 6 DI F F ERENTI ATI ON
3. (a) Equilibrium condition: 150
1
2
P
= 20 +2P
, which gives P
= 52 and Q
= 20 +2P
= 124.
For (b) and (c) se answers in the text.
7. (a) f is dened and strictly increasing for e
x
> 2, i.e. x > ln 2. Its range is . (f (x) as
x ln 2
+
, and f (x) as x .) From y = 3 +ln(e
x
2), we get ln(e
x
2) = y 3, and so
e
x
2 = e
y3
, or e
x
= 2 +e
y3
, so x = ln(2 +e
y3
). Hence f
1
(x) = ln(2 +e
x3
), x .
(b) Note that f is strictly increasing. Moreover, e
x
as x , and e
x
0 as x .
Therefore, f (x) 0 as x , and f (x) 1 as x . So the range of f , and therefore the
domain of f
1
, is (0, 1). From y =
a
e
x
+a
we get e
x
+ a = a/y, so e
x
= a(1/y 1), or
x = ln a + ln(1/y 1). Thus x = (1/) ln a (1/) ln(1/y 1), and therefore the inverse is
f
1
(x) = (1/) ln a (1/) ln(1/x 1), with x (0, 1).
Chapter 6 Differentiation
6.2
5. (a) We start by using the recipe in (6.2.3) to nd the slope of the tangent.
(a) (A): f (a + h) = f (0 + h) = 3h + 2 (B): f (a + h) f (a) = f (h) f (0) = 3h + 2 2 = 3h
(C)(D): [f (h) f (0)]/h = 3 (E): [f (h) f (0)]/h = 3 3 as h 0, so f
(0) = 3. The slope of

the tangent at (0, 2) is 3.
(b) (A): f (a+h) = f (1+h) = (1+h)
2
1 = 1+2h+h
2
1 = 2h+h
2
(B): f (1+h)f (1) = 2h+h
2
(C)(D): [f (1+h)f (1)]/h = 2+h (E): [f (1+h)f (1)]/h = 2+h 2 as h 0, so f
(1) = 2.
(c) (A): f (3+h) = 2+3/(3+h) (B): f (3+h) f (3) = 2+3/(3+h) 3 = h/(3+h) (C)(D):
[f (3 +h) f (3)]/h = 1/(3 +h) (E): [f (3 +h) f (3)]/h = 1/(3 +h) 1/3 as h 0, so
f
(3) = 1/3. (d) [f (h) f (0)]/h = (h

3
2h)/h = h
2
2 2 as h 0, so f
(0) = 2.
(e)
f (1 +h) f (1)
h
=
1 +h +1/(1 +h) +2
h
=
h
1 +h
0 as h 0, so f
(0) = 0.
(f)
f (1 +h) f (1)
h
=
(1 +h)
4
1
h
=
h
4
+4h
3
+6h
2
+4h +1 1
h
= h
3
+ 4h
2
+ 6h + 4 4 as
h 0, so f
(1) = 4.
8. (a)
_
x +h
x
__
x +h+
x
_
= (
x +h)
2
+
x +h
x +h(
x)
2
= x+hx = h.
(b)
f (x +h) f (x)
h
=
(
x +h
x)(
x +h +
x)
h(
x +h +
x)
=
h
h(
x +h +
x)
=
1
x +h +
x
(c) This follows from (b).
6.5
5. (a)
1/3 2/3h
h 2
=
3h
_
1/3 2/3h
_
3h(h 2)
=
h 2
3h(h 2)
=
1
3h
1
6
as h 2
(b) When x 0, x
2
1 1 and x
2
, so the fraction has no limit, but tends to .
(c)
32t 96
t
2
2t 3
=
32(t 3)
(t 3)(t +1)
=
32
t +1
8, as t 3, so
3
_
32t 96
t
2
2t 3

3
8 = 2 as t 3.
(d)
h +3
3
h
=
(
h +3
3)(
h +3 +
3)
h(
h +3 +
3)
=
h +3 3
h(
h +3 +
3)
=
1
h +3 +
3

1
2
3
as h 0.
CHAPTER 6 DI F F ERENTI ATI ON 15
(e)
t
2
4
t
2
+10t +16
=
(t +2)(t 2)
(t +2)(t +8)
=
t 2
t +8

2
3
as t 2.
(f) Observe that 4 x = (2 +
x)(2
x), so lim
x4
2
x
4 x
= lim
x4
1
2 +
x
=
1
4
.
6. (a)
f (x) f (1)
x 1
=
x
2
+2x 3
x 1
=
(x 1)(x +3)
x 1
= x +3 4 as x 1.
(b)
f (x) f (1)
x 1
= x +3 5 as x 2.
(c)
f (2 +h) f (2)
h
=
(2 +h)
2
+2(2 +h) 8
h
=
h
2
+6h
h
= h +6 6 as h 0.
(d)
f (a +h) f (a)
h
=
(a +h)
2
+2(a +h) a
2
2a
h
= 2a +2 +h 2a +2 as h 0.
(e) Same answer as in (d) putting x a = h.
(f)
f (a +h) f (a h)
h
=
(a +h)
2
+2a +2h (a h)
2
2a +2h
h
= 4a +4 4a +4 as h 0.
7. (a) x
3
8 = 0 has the solution x = 2, and polynomial division yields x
3
8 = (x 2)(x
2
+2x +4).
(b) and (c): see the text.
6.6
7. (a) With f (x) = x
2
, lim
h0
f (a +h) f (a)
h
= lim
h0
(5 +h)
2
5
2
h
= f
(5). On the other hand, f
(x) =
2x, so f
(5) = 10, and the limit is therefore 10. (b) and (c): see the text.
6.7
3. (a) y =
1
x
6
= x
6
y
= 6x
7
, using the power rule (6.6.4).
(b) y = x
1
(x
2
+ 1)
x = x
1
x
2
x
1/2
+ x
1
x
1/2
= x
3/2
+ x
1/2
y
=
3
2
x
1/2
1
2
x
3/2
(c) y = x
3/2
y
=
3
2
x
5/2
(d) y =
x +1
x 1
y
=
1 (x 1) (x +1) 1
(x 1)
2
=
2
(x 1)
2
(e) y =
x
x
5
+
1
x
5
= x
4
+x
5
y
=
4
x
5

5
x
6
(f) y =
3x 5
2x +8

3(2x +8) 2(3x 5)
(2x +8)
2
=
34
(2x +8)
2
(g) y = 3x
11
y
= 33x
12
(h) y =
3x 1
x
2
+x +1
y
=
3(x
2
+x +1) (3x 1)(2x +1)
(x
2
+x +1)
2
=
3x
2
+2x +4
(x
2
+x +1)
2
6. (a) f
(x) = 6x 12 = 6(x 2) 0 x 2, so f is increasing in [2, ). (b) f
(x) = x
3
3x =
x(x
2
3) = x(x
3)(x +
3), so (using a sign diagram) f is increasing in

_
3, 0
_
and in
_
3,
_
.
(c) f
(x) =
2(2 x
2
)
(x
2
+2)
2
=
2(2
2)(2 +
2)
(x
2
+2)
2
, so f is increasing in [
2,
2]. (d) See the text.

7. (a) y
= 1 2x = 3 when x = 1, so the slope of the tangent is 3. Since y = 1 when x = 1,

the point-slope formula gives y 1 = 3(x 1), or y = 3x + 4. (b) y
= 4x/(x
2
+ 1)
2
= 1 and
y = 0 when x = 1, so y = x 1. (c) y = x
2
x
2
, so y
= 2x +2x
3
= 17/4 and y = 15/4 when
x = 2, so y = (17/4)x 19/4. (d) y
=
4x
3
(x
3
+3x
2
+x +3) (x
4
+1)(3x
2
+6x +1)
[(x
2
+1)(x +3)]
2
=
1
19
and y = 1/3 when x = 0, so y = (x 3)/9.
16 CHAPTER 6 DI F F ERENTI ATI ON
9. (a) We use the quotient rule: y =
at +b
ct +d
y
=
a(ct +d) (at +b)c
(ct +d)
2
=
ad bc
(ct +d)
2
(b) y = t
n
_
a
t +b
_
= at
n+1/2
+bt
n
y
= (n +1/2)at
n1/2
+nbt
n1
.
(c) y =
1
at
2
+bt +c
y
=
0 (at
2
+bt +c) 1 (2at +b)
(at
2
+bt +c)
2
=
2at b
(at
2
+bt +c)
2
12. This is rather tricky because the denominator is 0 at x
1,2
= 2
2. A sign diagram shows that f (x) > 0

only in (, 0) and in (x
1
, x
2
).
6.8
3. (a) y = (x
2
+ x + 1)
5
= u
5
, where u = x
2
+ x + 1. By the chain rule, y
= (5)u
6
u
=
5(2x + 1)(x
2
+ x + 1)
6
. (b) With u = x +
_
x +
x, y =

u = u
1/2
, so y
=
1
2
u
1/2
u
. Now,
u = x + v
1/2
, with v = x + x
1/2
. Then u
= 1 +
1
2
v
1/2
v
, where v
= 1 +
1
2
x
1/2
. Thus, all in all,
y
=
1
2
u
1/2
u
=
1
2
_
x +(x +x
1/2
)
1/2
_
1/2
(1 +
1
2
(x +x
1/2
)
1/2
(1 +
1
2
x
1/2
)). (c) See the text.
6. x = b
ap c = b
u, with u = ap c. Then
dx
dp
=
1
2
u
u
=
a
2
ap c
.
12. (a), (e), and (g) are easy. For the others, you need the chain rule. In (d) you need the differentiation rules
for sum, product as well as the chain rule. See the text.
6.9
4. g
(t ) =
2t (t 1) t
2
(t 1)
2
=
t
2
2t
(t 1)
2
, g
(t ) =
(2t 2)(t 1)
2
(t
2
2t )2(t 1)
(t 1)
4
=
2(t 1)
(t 1)
4
=
2
(t 1)
3
, so g
(2) = 2.
5. With simplied notation: y
= f
g + fg
, y
= f
g + f
+ f
+ fg
= f
g + 2f
+ fg
,
y
= f
g +f
+2f
+2f
+f
+fg
= f
g +3f
+3f
+fg
6.10
2. (a) dx/dt = (b +2ct )e
t
+(a +bt +ct
2
)e
t
= (a +b +(b +2c)t +ct
2
)e
t
(b)
dx
dt
=
3qt
2
t e
t
(p +qt
3
)(e
t
+t e
t
)
t
2
e
2t
=
(qt
4
+2qt
3
pt p)e
t
t
2
e
2t
(c) See the text.
4. (a) y
= 3x
2
+2e
2x
is obviously positive everywhere, so y increases in (, ).
(b) y
= 10xe
4x
+5x
2
(4)e
4x
= 10x(12x)e
4x
. Asign diagramshows that y increases in [0, 1/2].
(c) y
= 2xe
x
2
+ x
2
(2x)e
x
2
= 2x(1 x)(1 + x)e
x
2
. A sign diagram shows that y increases in
(, 1] and in [0, 1]. (The answer in the text is wrong.)
6.11
3. For these problems we need the chain rule. That is an important rule! In particular, we need the fact that
d
dx
ln f (x) =
1
f (x)
f
(x) =
f
(x)
f (x)
when f is a differentiable function (with f (x) > 0).
(a) y = ln(ln x) = ln u y
=
1
u
u
=
1
ln x
1
x
=
1
x ln x
.
(b) y = ln
_
1 x
2
= ln u y
=
1
u
u
=
1
1 x
2
2x
2
1 x
2
=
x
1 x
2
.
(Alternatively:

1 x
2
= (1 x
2
)
1/2
y =
1
2
ln(1 x
2
), and so on.)
CHAPTER 6 DI F F ERENTI ATI ON 17
(c) y = e
x
ln x y
= e
x
ln x +e
x
1
x
= e
x
_
ln x +
1
x
_
.
(d) y = e
x
3
ln x
2
y
= 3x
2
e
x
3
ln x
2
+e
x
3 1
x
2
2x = e
x
3
_
3x
2
ln x
2
+
2
x
_
.
(e) y = ln(e
x
+1) y
=
e
x
e
x
+1
. (f) y = ln(x
2
+3x 1) y
=
2x +3
x
2
+3x 1
.
4. (a) ln u is dened for u > 0, so we must require x +1 > 0, i.e. x > 0.
(b) We must have 1x = 0 for the fraction to be dened, and
3x 1
1 x
> 0 for the logarithmto be dened.
A sign diagram (see below) shows that
3x 1
1 x
is dened and positive if and only if 1/3 < x < 1.
(c) ln |x| is dened |x| > 0 x = 0.
x
1/3 1 0
3x 1
1 x
3x 1
1 x
5. (a) One must have x
2
> 1, i.e. x > 1 or x < 1. (See Fig. 4.3.6 in the text.) (b) ln(ln x) is dened
when ln x is dened and positive, that is, for x > 1. (c) The fraction
1
ln(ln x) 1
is dened when ln(ln x)
is dened and different from 1. From (b), ln(ln x) is dened when x > 1. Further, ln(ln x) = 1
ln x = e x = e
e
. Conclusion:
1
ln(ln x) 1
is dened x > 1 and x = e
e
.
9. In these problems we can use logarithmic differentiation. Alternatively we can write the functions in the
form f (x) = e
g(x)
and then use the fact that f
(x) = e
g(x)
g
(x) = f (x)g
(x).
(a) Let f (x) = (2x)
x
. Then ln f (x) = x ln(2x), so
f
(x)
f (x)
= 1 ln(2x) +x
1
2x
2 = ln(2x) +1. Hence,
f
(x) = f (x)(ln(2x) +1) = (2x)

x
(ln x +ln 2 +1).
(b) f (x) = x
x
=
_
e
ln x
_
x
= e
x ln x
, so f
(x) = e
x ln x
d
dx
(
x ln x) = x
x
_
ln x
2
x
+
x
x
_
(c) ln f (x) = x ln
x =
1
2
x ln x, so f
(x)/f (x) =
1
2
(ln x +1), which gives f
(x) =
1
2
(
x )
x
(ln x +1).
10. ln y = v ln u, so y
/y = v
ln u +(v/u)u
, and so y
= u
v
_
v
ln u +
vu
u
_
.
11. (a) See the answer in the text. (b) Let f (x) = ln(1 +x)
1
2
x. Then f (0) = 0 and moreover f
(x) =
1/(x +1)
1
2
= (1 x)/2(x +1), which is positive in (0, 1), so f (x) > 0 in (0, 1), and the left-hand
inequality is established. To prove the other inequality, put g(x) = x ln(1 + x). Then g(0) = 0 and
g
(x) = 1 1/(x +1) = x/(x +1) > 0 in (0, 1), so the conclusion follows.
(c) Let f (x) = 2(
x 1) ln x. Then f (1) = 0 and f
(x) = (1/
x) 1/x = (x
x)/x
x =
(
x 1)/x, which is positive for x > 1. The conclusion follows.

18 CHAPTER 7 DERI VATI VES I N USE
5. (a) y = 3 and y
= 6x = 6 at x = 1, so y (3) = (6)(x 1), or y = 6x +3.

(b) y = 14 and y
= 1/2
x 2x = 31/4 at x = 4, so y = (31/4)x +17.

(c) y = 0 and y
= (2x
3
8x
2
+6x)/(x +3)
2
= 1/4 at x = 1, so y = (1/4)(x 1).
7. (a) f (x) = x
3
+x, etc. (b) Easy. (c) h(y) = y(y
2
1) = y
3
y, etc. (d)(f): use the quotient rule.
15. (a) y
=
2
x
ln x 0 if x 1. (b) y
=
e
x
e
x
e
x
+e
x
0 e
x
e
x
e
2x
0 x 0
(c) y
= 1
3x
x
2
+2
=
(x 1)(x 2)
x
2
+2
0 x 1 or x 2. (Use a sign diagram.)
Chapter 7 Derivatives in Use
7.1
2. Implicit differentiation yields () 2xy + x
2
(dy/dx) = 0, and so dy/dx = 2y/x. Differentiating ()
implicitly w.r.t. x gives 2y +2x(dy/dx)+2x(dy/dx)+x
2
(d
2
/dx
2
) = 0. Inserting the result for dy/dx,
and simplifying yields d
2
/dx
2
= 6y/x
2
. (Alternatively, we can differentiate 2y/x as a fraction.) These
results follows more easily by differentiating y = x
2
twice.
3. (a) Implicit differentiation w.r.t. x yields () 1 y
+ 3y + 3xy
= 0. Solving for y
yields y
=
(1 + 3y)/(1 3x) = 5/(1 3x)
2
. Differentiating () w.r.t. x gives y
+ 3y
+ 3y
+ 3xy
= 0.
Inserting y
= (1 +3y)/(1 3x) and solving for y
gives y
= 6y
/(1 3x) = 30/(1 3x)

3
(c) Implicit differentiation w.r.t. x yields () 5y
4
y
= 6x
5
, so y
= 6x
5
/5y
4
= (6/5)x
1/5
. Differentiating
() w.r.t. x gives 20y
3
(y
)
2
+ 5y
4
y
= 30x
4
. Inserting y
= 6x
5
/5y
4
and solving for y
yields y
=
6x
4
y
4
(144/25)x
10
y
9
= (6/25)x
4/5
.
6. (a) 2x +2yy
= 0, and solve for y
. (b) 1/2
x +y
/2
y = 0, and solve for y
.
(c) 4x
3
4y
3
y
= 2xy
3
+x
2
3y
2
y
, and solve for y
.
8. (a) y +xy
= g
(x) +3y
2
y
, and solve for y
. (b) g
(x +y)(1 +y
) = 2x +2yy
, and solve for y
.
(c) 2(xy +1)(y +xy
) = f
(x
2
y)(2xy +x
2
y
), and solve for y
. (How did we differentiate f (x

2
y) w.r.t.
x? Well, if z = f (u) and u = x
2
y, then z
= f
(u)u
where u is a product of two functions that both

depend on x. So u
= 2xy +x
2
y
.)
10. (a) 2(x
2
+y
2
)(2x+2yy
) = a
2
(2x2yy
), and solve for y
. (b) Note that y
= 0 when x
2
+y
2
= a
2
/2,
or y
2
=
1
2
a
2
x
2
. Inserting this into the given equation yields x =
1
4
a
6. This yields the four points

on the graph at which the tangent is horizontal.
7.2
1. Implicit differentiation w.r.t. P, with Q as a function of P, yields
dQ
dP
P
1/2
+ Q
1
2
P
1/2
= 0.
Thus
dQ
dP
=
1
2
QP
1
=
19
P
3/2
.
5. Differentiating () w.r.t. P yields f
(P +t )
_
dP
dt
+1
_
2
+f
(P +t )
d
2
P
dt
2
= g
(P)
_
dP
dt
_
2
+g
(P)
d
2
P
dt
2
.
With simplied notation f
(P
+ 1)
2
+ f
= g
(P
)
2
+ g
. Substituting P
= f
/(g
) and
solving for P
, we get P
= [f
(g
)
2
g
(f
)
2
]/(g
)
3
.
CHAPTER 7 DERI VATI VES I N USE 19
7.3
2. (a) f
(x) = x
2
_
4 x
2
+
1
3
x
3
2x
2
4 x
2
=
4x
2
(3 x
2
)
3
4 x
2
. For the rest, see the answer in the text.
5. (a) See the text. (b) dy/dx = e
x
/(e
x
+3), so dx/dy = (e
x
+3)/e
x
= 1 3e
x
(c) Implicit differentiation w.r.t. x yields y
3
+x3y
2
(dy/dx) 3x
2
y x
3
(dy/dx) = 2. Solve for dy/dx,
and then invert.
7.4
3. (a) f (0) = 1 and f
(x) = (1 +x)
2
, so f
(0) = 1. Then f (x) f (0) +f
(0)x = 1 x.
(b) f (0) = 1 and f
(x) = 5(1 +x)

4
, so f
(0) = 5. Then f (x) f (0) +f
(0)x = 1 +5x.
(c) f (0) = 1 and f
(x) =
1
4
(1 x)
3/4
, so f
(0) =
1
4
. Then f (x) f (0) +f
(0)x = 1
1
4
x.
4. F(1) = A and F
(K) = AK
1
, so F
(1) = A. Then F(K) F(1) +F
(1)(K 1)=
A +A(K 1) = A(1 +A(K 1)).
9. 3e
xy
2
+ 3xe
xy
2
(y
2
+ x2yy
) 2y
= 6x + 2yy
. For x = 1, y = 0 this reduces to 3 2y
= 6, so
y
= 3/2. (b) y(x) y(1) +y
(1)(x 1) =
3
2
(x 1)
7.5
2. f
(x) = (1 + x)
1
, f
(x) = (1 + x)
2
, f
(x) = 2(1 + x)
3
, f
(iv)
(x) = 6(1 + x)
4
, f
(v)
(x) =
24(1+x)
5
. Then f (0) = 0, f
(0) = 1, f
(0) = 1, f
(0) = 2, f
(iv)
(0) = 6, f
(v)
(0) = 24, and so
f (x) f (0) +
1
1!
f
(0)x +
1
2!
f
(0)x +
1
3!
f
(0)x
3
+
1
4!
f
(iv)
(0)x
4
+
1
5!
f
(v)
(0)x
5
= x
1
2
x
2
+
1
3
x
3
1
4
x
4
+
1
5
x
5
3. With f (x) = 5(ln(1+x)
1 +x ) = 5 ln(1+x)5(1+x)
1/2
we get f
(x) = 5(1+x)
1
5
2
(1+x)
1/2
,
f
(x) = 5(1 + x)
2
+
5
4
(1 + x)
3/2
, and so f (0) = 5, f
(0) =
5
2
, f
(0) =
15
4
. and the Taylor
polynomial of degree 2 about x = 0 is f (0) +f
(0)x +
1
2
f
(0)x
2
= 5 +
5
2
x
15
8
x
2
.
9. h
(x) =
(px
p1
qx
q1
)(x
p
+x
q
) (x
p
x
q
)(px
p1
+qx
q1
)
(x
p
+x
q
)
2
=
2(p q)x
p+q1
(x
p
+x
q
)
2
, so h
(1) =
1
2
(p q). Since h(1) = 0, we get h(x) h(1) +h
(1)(x 1) =
1
2
(p q)(x 1).
7.6
1. From Problem 7.5.2, f (0) = 0, f
(0) = 1, f
(0) = 1, and f
(c) = 2(1 + c)
3
. Then (3) gives
f (x) = f (0) +
1
1!
f
(0)x +
1
2!
f
(0)x +
1
3!
f
(c)x
3
= x
1
2
x
2
+
1
3
(1 +c)
3
x
3
.
4. (a) With g(x) = (1+x)
1/3
, g
(x) =
1
3
(1+x)
2/3
, g
(x) =
2
9
(1+x)
5/3
, and g
(x) =
10
27
(1+x)
8/3
,
so g(0) = 1, g
(0) =
1
3
, g
(0) =
2
9
, g
(c) =
10
27
(1 +c)
8/3
, so
g(x) = 1 +
1
3
x
1
9
x
2
+R
3
(x), where R
3
(x) =
1
6!
10
27
(1 +c)
8/3
x
3
=
5
81
(1 +c)
8/3
x
3
(b) c (0, x) and x 0, so (1 +c)
8/3
1, and the inequality follows.
(c)
3
1003 = 10(1+3 10
3
)
1/3
10.0099900, using part (a) to approximate (1+3 10
3
)
1/3
. The error
in (b) is |R
3
(x)|
5
81
(3 10
3
)
3
=
5
3
10
9
. So the error in
3
1003 is 10|R
3
(x)| =
50
3
10
9
< 2 10
8
,
and the answer is correct to 7 decimal places.
20 CHAPTER 7 DERI VATI VES I N USE
7.7
4. (a) El
x
e
ax
= (x/e
ax
)ae
ax
= ax (b) El
x
ln x = (x/ ln x)(1/x) = 1/ ln x
(c) El
x
(x
p
e
ax
) =
x
x
p
e
ax
(px
p1
e
ax
+x
p
ae
ax
) = p +ax
(d) El
x
(x
p
ln x) =
x
x
p
ln x
(px
p1
ln x +x
p
(1/x)) = p +1/ ln x
9. (a) El
x
A =
x
A
dA
dx
= 0 (b) El
x
(fg) =
x
fg
(fg)
=
x
fg
(f
g +fg
) =
xf
f
+
xg
g
= El
x
f +El
x
g
(c) El
x
f
g
=
x
(f/g)
_
f
g
_
=
xg
f
_
gf
fg
g
2
_
=
xf
xg
g
= El
x
f El
x
g
(d) See the answer in the text. (e) Is like (d), but with +g replaced by g, and +g
by g
.
(f) z = f (g(u)), u = g(x) El
x
z =
x
z
dz
dx
=
x
u
u
z
dz
du
du
dx
= El
u
f (u) El
x
u
7.8
3. Using (7.8.4), the functions are continuous wherever they are dened. So (a) and (d) are dened every-
where. In (b) we must exclude x = 1, in (c) the function is dened for x < 2, in (e) we must exclude
x =
31, because the denominator is 0 for these values of x. Finally, in (f), the rst fraction requires
x > 0, and then the other fraction is also dened.
7.9
1. (b) |x| = x for x < 0. Hence, lim
x0
x +|x|
x
= lim
x0
x x
x
= lim
x0
0 = 0.
(c) |x| = x for x > 0. Hence, lim
x0
+
x +|x|
x
= lim
x0
+
x +x
x
= lim
x0
+
2 = 2.
(d) When x 0
+
,

x 0, so 1/
x . (e) When x 3
+
, x 3 0
+
, and so
x/(x 3) . (f) When x 3
, x 3 0
, and so x/(x 3) .
4. (a) Vertical asymptote, x = 1. Moreover, x
2
(x + 1) = x 1 + 1/(x + 1), so y = x 1 is an
asymptote as x . (b) No vertical asymptote. Moreover. (2x
3
3x
2
+ 3x 6) (x
2
+ 1) =
2x 3 +(x 3)/(x
2
+1), so y = 2x 3 is an asymptote as x . (c) Vertical asymptote, x = 1.
Moreover, (3x
2
+ 2x) (x 1) = 3x + 5 + 5/(x 1), so y = 3x + 5 is an asymptote as x .
(d) Vertical asymptote, x = 1. Moreover, (5x
4
3x
2
+1) (x
3
1) = 5x +(3x
2
+5x +1)/(x
3
1),
so y = 5x is an asymptote as x .
7.10
4. Recall from Note 4.7.2 that any integer root of the equation f (x) = x
4
+3x
3
3x
2
8x +3 = 0 must
be a factor of the constant term 3. The way to see this directly is to notice that we must have
3 = x
4
3x
3
+3x
2
+8x = x(x
3
3x
2
+3x +8)
and if x is an integer then the bracketed expression is also an integer. Thus, the only possible integer
solutions are 1 and 3. Trying each of these possibilities, we nd that only 3 is an integer solution.
We are told in the problem that there are three other real roots, with approximate values x
0
=
1.9, y
0
= 0.4, and z
0
= 1.5. If we use Newtons method once for each of these roots we get new
approximations
CHAPTER 7 DERI VATI VES I N USE 21
x
1
= 1.9
f (1.9)
f
(1.9)
= 1.9
0.1749
8.454
1.9 +0.021 = 1.879
y
1
= 0.4
f (0.4)
f
(0.4)
= 0.4
0.4624
8.704
0.4 0.053 = 0.347
z
1
= 1.5
f (1.5)
f
(1.5)
= 1.5
0.5625
16.75
1.5 +0.034 = 1.534
It can be shown by more precise calculations that the actual roots, rounded to six decimals, are 1.879385,
0.347296, and 1.532089.
7.11
2. (a) When n , 2/n 0 and so 5 2/n 5. (b) When n ,
n
2
1
n
= n 1/n .
(c) When n ,
3n
2n
2
1
=
3n
n
_
2 1/n
2
=
3
_
2 1/n
2

3
2
=
3
2
3
.
7.12
2. LHpitals rule yields lim
xa
x
2
a
2
x a
=
0
0
= lim
xa
2x
1
= 2a. But note that we dont really need
lHpitals rule here, because x
2
a
2
= (x +a)(x a), and therefore lim
xa
x
2
a
2
x a
= lim
xa
(x +a) = 2a.
(b) lim
x0
2(1 +x)
1/2
2 x
2(1 +x +x
2
)
1/2
2 x
=
0
0
= lim
x0
(1 +x)
1/2
1
(1 +2x)(1 +x +x
2
)
1/2
1
=
0
0
=
lim
x0
1
2
(1 +x)
3/2
2(1 +x +x
2
)
1/2
+(1 +2x)
2
(
1
2
)(1 +x +x
2
)
3/2
=
1
3
7. L = lim
xa
f (x)
g(x)
= lim
xa
1/g(x)
1/f (x)
=
0
0
= lim
xa
1/(g(x))
2
1/(f (x))
2

g
(x)
f
(x)
= lim
xa
(f (x))
2
(g(x))
2

g
(x)
f
(x)
=
L
2
lim
xa
g
(x)
f
(x)
= L
2
lim
xa
1
f
(x)/g
(x)
. The conclusion follows. (Here, we have ignored problems with
division by 0, when either f
(x) or g
(x) tends to 0 as x tends to a.)

2. 5y
4
y
y
2
2xyy
= 0, so y
=
y
2
5y
4
2xy
=
y
5y
3
2x
. Because y = 0 makes the given equation
meaningless, y
is never 0.
6. y
= 0 when 1 +
1
5
ln x = 0, i.e. ln x = 5, and then x = e
5
.
7. (a) We must have
1 +x
1 x
> 0, i.e 1 < x < 1. When x 1
, f (x) . When x 1
,
f (x) . Since f
(x) = 1/(1 x
2
) > 0 when 1 < x < 1, f is strictly increasing and the range
of f is . (b) From y =
1
2
ln
1 +x
1 x
, ln
1 +x
1 x
= 2y, so
1 +x
1 x
= e
2y
. Then solve for x.
9. (a) f (0) = ln 4 and f
(x) = 2/(2x +4), so f
(0) = 1/2. Then f (x) f (0) +f
(0)x = ln 4 +x/2.
(b) g(0) = 1 and g
(x) = (1/2)(1 +x)

3/2
, so g
(0) = 1/2. Then g(x) g(0) +g
(0)x = 1 x/2.
(c) h(0) = 0 and h
(x) = e
2x
+2xe
2x
, so h
(0) = 1. Then h(x) h(0) +h
(0)x = x.
22 CHAPTER 8 SI NGL E- VARI ABL E OPTI MI ZATI ON
12. With x =
1
2
and n = 5, formula (7.6.6) yields, e
1
2
= 1 +
1
2
1!
+
(
1
2
)
2
2!
+
(
1
2
)
3
3!
+
(
1
2
)
4
4!
+
(
1
2
)
5
5!
+
(
1
2
)
6
6!
e
c
, where c
is some number between 0 and
1
2
. Now, R
6
(
1
2
) =
(
1
2
)
6
6!
e
c
<
(
1
2
)
6
6!
2 =
1
23040
0.0004340, where we used
the fact that since c <
1
2
, e
c
< e
1
2
< 2. Thus it follows that
e
1
2
1 +
1
2
1!
+
(
1
2
)
2
2!
+
(
1
2
)
3
3!
+
(
1
2
)
4
4!
+
(
1
2
)
5
5!
= 1 +
1
2
+
1
8
+
1
48
+
1
384
+
1
3840
1.6486979. The error is less
than 0.000043, and e
1
2
1.649 correct to 3 decimals.
14. y
+ (1/y)y
= 1, or () yy
+ y
= y. When y = 1, y
= 1/2. Differentiating () w.r.t, x yields

(y
)
2
+yy
+y
= y
. With y = 1 and y
= 1/2, we nd y
= 1/8, so y(x) 1 +
1
2
x +
1
16
x
2
.
21. (a) lim
x0
(2 x)e
x
x 2
x
3
=
0
0
= lim
x0
e
x
+(2 x)e
x
1
3x
2
=
0
0
= lim
x0
e
x
e
x
+(2 x)e
x
6x
=
lim
x0
xe
x
6x
= lim
x0
e
x
6
=
1
6
. (By canceling x, we needed to use lHpitals rule only twice.)
(b) lim
x3
_
1
x 3

5
x
2
x 6
_
= lim
x3
x
2
6x +9
x
3
4x
2
3x +18
=
0
0
= lim
x3
2x 6
3x
2
8x 3
=
0
0
=
lim
x3
2
6x 8
=
1
5
(c) lim
x4
x 4
2x
2
32
=
0
0
= lim
x4
1
4x
=
1
16
. (Can you nd another way?)
23. (a) lim
x1
ln x x +1
(x 1)
2
=
0
0
= lim
x1
(1/x) 1
2(x 1)
=
0
0
= lim
x1
(1/x
2
)
2
=
1
2
(b) lim
x1
1
x 1
ln
_
7x +1
4x +4
_
= lim
x1
ln(7x +1) ln(4x +4)
x 1
=
0
0
= lim
x1
7
7x +1

4
4x +4
1
=
3
8
(c) lim
x1
x
x
x
1 x +ln x
=
0
0
= lim
x1
x
x
(ln x +1) 1
1 +1/x
=
0
0
= lim
x1
x
x
(ln x +1)
2
+x
x
(1/x)
1/x
2
= 2
(using Example 6.11.4 to differentiate x
x
).
24. When x 0, the denominator tends to

b
d and the numerator to 0, so the limit does not exist

when b = d. If b = d, see the text.
Chapter 8 Single-Variable Optimization
8.1
1. (a) f (0) = 2 and f (x) 2 for all x (we divide 8 by a number larger than 2), so x = 0 maximizes f (x).
(b) g(2) = 3 and g(x) 3 for all x, so x = 2 minimizes g(x). g(x) as x , so there
is no maximum. (c) h(x) has its largest value 1 when 1 +x
2
is the smallest, namely for x = 0, and h(x)
has its smallest value 1/2 when 1 +x
2
is the largest, namely for x = 1.
8.2
2. See the text, but note that the sign variation alone is not sufcient to conclude that the two stationary
points are extreme points. It is important to point out that h(x) 0 as x . Sketch the graph.
(For example, f (x) = 3x x
3
is decreasing in (, 1], increasing in [1, 1], and decreasing in
[1, ) but has no maximum or minimum, since f (x) as x , and f (x) as x .
In fact, x = 1 is a local minimum point and x = 1 is a local maximum point.)
3. h
(t ) = 1/2
t
1
2
= (1
t )/2
t . We see that h
(t ) 0 in [0, 1] and h
(t ) 0 in [1, ). According
to Theorem 8.2.1(a), t = 1 maximizes h(t ).
CHAPTER 8 SI NGL E- VARI ABL E OPTI MI ZATI ON 23
5. f
(x) = 3x
2
ln x + x
3
/x = 3x
2
(ln x +
1
3
). f
(x) = 0 when ln x =
1
3
, i.e. x = e
1/3
. We see that
f
(x) 0 in (0, e
1/3
] and f
(x) 0 in [e
1/3
, ), so x = e
1/3
minimizes f (x). Since f (x)
as x , there is no maximum.
8. (a) y
= e
x
2e
2x
, y
= e
x
+ 4e
2x
. We see that y
= 0 when e
x
= 2e
2x
, or e
3x
= 2, i.e.
x =
1
3
ln 2. Since y
> 0 everywhere, this is minimum point. (b) y
= 2(x a) 4(x b) = 0
when x =
1
3
(a +2b). This is a maximum point since y
= 6 for all x.
(c) y
= 1/x 5 = 0 when x =
1
5
. This is a maximum point since y
= 1/x
2
< 0 for all x > 0.
10. (a) f
(x) = k Ae
x
= 0 when x
0
= (1/) ln(A/k). Note that x
0
> 0 iff A > k. Moreover,
f
(x) < 0 if x < x

0
and f
(x) > 0 if x > x

0
, so x
0
solves the minimization problem.
(b) Substituting for Ain the answer to (a) gives the expression for the optimal height x
0
. Its value increases
as p
0
(probability of ooding) or V (cost of ooding) increases, but decreases as (interest rate) or k
(marginal construction costs) increases. The signs of these responses are obviously what an economist
would expect. (Not only an economist, actually.)
8.3
2. (a) (Q) = Q(aQ)kQ = Q
2
+(ak)Q, so
(Q) = 2Q+(ak) = 0 for Q
=
1
2
(ak). This
maximizes because
(Q) < 0. The monopoly prot is (Q
) = (
1
2
(a k))
2
+(a k)
1
2
(a k) =
1
4
(a k)
2
. (b) d(Q
)/dk =
1
2
(a k) = Q
, as in Example 3. (c) The new prot function is

(Q) = (Q) +sQ = Q
2
+(a k)Q+sQ.
(Q) = 2Q+a k +s = 0 when

Q =
1
2
(a k +s).
Now

Q =
1
2
(a k + s) = a k provided s = a k, which is the subsidy required to induce the
monopolist to produce a k units. (The answer in the text has the wrong sign.)
5. T
(W) = a
pb(bW +c)
p1
W (bW +c)
p
W
2
= a(bW +c)
p1
pbW bW c
W
2
, e.t.c.
8.4
2. In all cases the maximum and minimum exist by the extreme value theorem. Follow the recipe in (8.4.1).
(a) f (x) is strictly decreasing so maximum is at x = 0, minimum at x = 3.
(b) f (1) = f (2) = 10 and f
(x) = 3x
2
3 = 0 at x = 1. f (1) = 6.
(c) f (x) = x +1/x. f (1/2) = f (2) = 5/2 and f
(x) = 1 1/x
2
= 0 at x = 1. f (1) = 2.
(d) f (1) = 4, f (
5) = 0, and f
(x) = 5x
2
(x
2
3) = 0 at x = 0 and x =
3. f (0) = 0, f (
3) =
6
3. (e) f (0) = 0, f (3000) = 4.5 10

9
, f
(x) = 3(x
2
3000x +2 10
6
= 3(x 500)(x 2000).
f (1000) = 2.5 10
9
, f (2000) = 2 10
9
.
4. (a) When there are 60 + x passengers, the charter company earns 800 10x from each, so they earn
$(60 +x)(800 10x). The sports club earns 1/10 of that amount. (b) See the text.
6. (a) (f (2) f (1))/(2 1) = (4 1)/1 = 3 and f
(x) = 2x, so 2x
= 3, and thus x
= 3/2.
(b) (f (1) f (0))/1 = 1 and f
(x) = 2x/
1 x
2
, so 2x
/
_
1 (x)
2
= 1, so x
=

5/5.
(From 2x
/
_
1 (x)
2
= 1,
_
1 (x)
2
= 2x
, and then 1 (x
)
2
= 4(x
)
2
. The positive solution is
x
5/5. (c) (f (6) f (2))/4 = 1/6 and f
(x) = 2/x
2
, so 2/(x
)
2
= 1/6, so x
12.
(d) (f (4) f (0))/4 = 1/4 and f
(x) = x/
9 +x
2
, so x/
_
9 +(x
)
2
= 1/4, so x
3.
24 CHAPTER 8 SI NGL E- VARI ABL E OPTI MI ZATI ON
8.5
1. (Q) = 10Q
1
1000
Q
2
(5000 + 2Q) = 8Q
1
1000
Q
2
5000. Since
(Q) = 8
1
500
Q = 0 for
Q = 4000, and
(Q) =
1
500
< 0, Q = 4000 maximizes prots.
4. (i) (Q) = 1840Q(2Q
2
+40Q+5000) = 1800Q2Q
2
5000. Since
(Q) = 1800 4Q = 0
for Q = 450, and
(Q) = 4 < 0, Q = 450 maximizes prots.

(ii) (Q) = 2200Q2Q
2
5000. Since
(Q) = 22004Q = 0 for Q = 550, and
(Q) = 4 < 0,
Q = 550 maximizes prots..
(iii) (Q) = 2Q
2
100Q5000 is negative for all Q 0, so Q = 0 obviously maximizes prots.
6.
(Q) = P abQ
b1
= 0 when Q
b1
= P/ab, i.e. Q = (P/ab)
1/(b1)
. Moreover,
(Q) =
ab(b 1)Q
b2
< 0 for all Q > 0, so this is a maximum point.
8.6
2. (a) Strictly decreasing, so no extreme points. (b) f
(x) = 3x
2
3 = 0 for x = 1. With f
(x) = 6x,
f
(1) = 6 and f
(1) = 6, so x = 1 is a loc. maximum point, and x = 1 is a loc. minimum point.

(c) f
(x) = 1 1/x
2
= 0 for x = 1. With f
(x) = 2/x
3
, f
(1) = 2 and f
(1) = 2, so x = 1
is a local maximum point, and x = 1 is a local minimum point. (d)(f): see the text.
3. (a) f (x) is dened if and only if x = 0 and x 6. f (x) = 0 at x = 6 and at x = 2. At any other
point x in the domain, f (x) has the same sign as (x +2)/x, so f (x) > 0 if x (6, 2) or x (0, ).
(b) We rst nd the derivative of f :
f
(x) =
2
x
2
x +6 +
x +2
x
1
2
x +6
=
4x 24 +x
2
+2x
2x
2
x +6
=
x
2
2x 24
2x
2
x +6
=
(x +4)(x 6)
2x
2
x +6
By means of a sign diagram we see that f
(x) > 0 if 6 < x < 4, f
(x) < 0 if 4 < x < 0,

f
(x) < 0 if 0 < x < 6, f
(x) > 0 if 6 < x. It follows that f is strictly increasing in [6, 4],

decreasing in [4, 0), decreasing in (0, 6], and increasing [6, ). It follows from the rst-derivative test
(Thm. 8.6.1) that f has two local minimum points, x
1
= 6 and x
2
= 6, and one local maximum point,
x
3
= 4, with f (6) = 0, f (6) =
4
3
8 = 8
2/3, and f (4) =

1
2
2.
(c) Since lim
x0
x +6 = 6 > 0, while lim

x0
(1 + 2/x) = and lim
x0
+(1 + 2/x) = , we
see that lim
x0
f (x) = and lim
x0
+ f (x) = . Furthermore,
lim
x
f
(x) = lim
x
_
x
2
2x 24
2x
2

1
x +6
_
=
1
2
0 = 0
y
4
2
2
4
6
8
x
6 4 2 2 4 6 8 10
Figure SM8.6.3
CHAPTER 8 SI NGL E- VARI ABL E OPTI MI ZATI ON 25
4. Look at the point a. Since f
(x) is graphed, f
(x) < 0 to the left of a, f
(a) = 0, and f
(x) > 0 to the

right of a, so a is a local minimum point. At the points b and e, f
(x) > 0 on both sides of the points, so

they cannot be extreme points.
6. (a) f
(x) = x
2
e
x
(3 + x). Use a sign diagram. (x = 0 is not an extreme point, but an inection point.)
(b) See the text and use a sign diagram for g
(x), or check the sign of g
(x) = 2
x
(2+4x ln x +x
2
(ln 2)
2
)
at the stationary points.
7. f (x) = x
3
+ax +b as x , and f (x) as x . Thus f (x) has at least one real
root. We have f
(x) = 3x
2
+a. Thus, for a 0, f
(x) > 0 for all x = 0, so f is strictly increasing, and

there is only one real root. Note that for a 0, 4a
3
+27b
2
0. Assume next that a < 0. Then f
(x) = 0
for x =
a/3 =
p, where p = a/3 > 0. Then f has a local maximum at (
p, b +2p
p)
and a local minimum at (
p, b 2p
p). If one of the local extreme values is 0, the equation has a

double root, and this is the case iff 4p
3
= b
2
, that is, iff 4a
3
+ 27b
2
= 0. The equation has three real
roots iff the local maximum value is positive and the local minimum value is negative. This occurs iff
|b| < 2p
p or iff b
2
< 4p
3
or iff 4a
3
+27b
2
< 0.
8.7
1. (a) f
(x) = 3x
2
+3x6 = 3(x1)(x+2). Use a sign diagramand see the text. (b) f
(x) = 6x+3 = 0
for x = 1/2 and f
(x) changes sign around x = 1/2, so this is an inection point.

3. Straightforward by using these derivatives: (a) y
= e
x
(1 +x), y
= xe
x
(b) y
=
x 1
x
2
, y
=
2 x
x
3
(c) y
= x
2
e
x
(3 x), y
= xe
x
(x
2
6x +6)
(d) y
=
1 2 ln x
x
3
, y
=
6 ln x 5
x
4
(e) y
= 2e
x
(e
x
1), y
= e
x
(2e
x
1)
(f) y
= 2e
x
(2 x
2
), y
= e
x
(x
2
2x 2)
2. (a) Q
(L) = 24L
3
20
L
2
= 3L(8
1
20
L) = 0 for L
= 160, and Q(L) is increasing in [0, 160],

decreasing in [160, 200], so Q
= 160 maximizes Q(L). Output per worker is Q(L)/L = 12L

1
20
L
2
,
and this quadratic function has maximum at L
= 120. (b) See the text.

3. = 0.0016Q
2
+ 44Q 0.0004Q
2
8Q 64 000 = 0.002Q
2
+ 36Q 64 000, and Q = 9000
maximizes this quadratic function.
(b) El
Q
C(Q) =
Q
C(Q)
C
(Q) =
0.0008Q
2
+8Q
0.0004Q
2
+8Q+64 000
0.12 when Q = 1000.
4. (a) See Problem 8.7.3(c). (b) lim
x
f (x) = 0 according to (7.12.3), page 253 in the text.
lim
x
f (x) = because x
3
and e
x
. (See Fig. A8.R.4, page 671 in the text.)
5. (a) See the text. (b) Asign diagramshows that f
(x) 0 in (1, 1] and f
(x) 0 in [1, ). Hence x = 1

is a maximum point. f
(x) =
x(x
2
+x 1)
(x +1)
2
= 0 for x = 0 and x =
1
2
(
5 1). (x =
1
2
(
5 1)
is outside the domain.) Since f
(x) changes sign around these points, they are both inection points.
26 9 I NTEGRATI ON
6. (a) h
(x) =
e
x
(2 +e
2x
) e
x
2e
2x
(2 +e
2x
)
2
=
e
x
(2 e
2x
)
(2 +e
2x
)
2
. See the text.
(b) h is strictly increasing in (, 0], lim
x
h(x) = 0, and h(0) = 1/3. Thus, h dened on (, 0] has
an inverse dened on (0, 1/3] with values in (, 0]. To nd the inverse, note that
e
x
2 +e
2x
= y
y(e
x
)
2
e
x
+2y = 0. This quadratic equationine
x
has the roots e
x
= [1
_
1 8y
2
]/2y. Since y = 1/3
for x = 0, see that we must have e
x
= [1
_
1 8y
2
]/2y, and thus x = ln(1
1 8x
2
) ln(2x).
Using x as the free variable, h
1
(x) = ln(1
1 8x
2
) ln(2x). The function and its inverse are
graphed in Fig. SM8.R.6.
y
1.5
1.0
0.5
0.5
x
1.5 1.0 0.5 0.5
h
h
1
Figure SM8.R.6
8. f
(x) =
6x
2
(x
2
3)(x
2
+2)
(x
4
+x
2
+2)
2
, so f is stationary when x = 0 and when x =
3. x =
3 is a local
(and global) maximum point, x =
3 is a local (and global) minimum point, and x = 0 is neither.

(It is an inection point.) The graph of f is shown in Fig. A8.R.8 in the text, page 671.
9 Integration
9.1
1. This should be straightforward, since all the integrands are powers of x. Note that x
x = x x
1/2
= x
3/2
,
1/
x = x
1/2
, and
_
x
_
x
x =
_
x
x
3/2
=
x x
3/4
=
x
7/4
= x
7/8
.
4. (a)
_
(t
3
+2t 3) dt =
_
t
3
dt +
_
2t dt
_
3 dt =
1
4
t
4
+t
2
3t +C
(b)
_
(x1)
2
dx =
_
(x
2
2x+1) dx =
1
3
x
3
x
2
+x+C. Alternative: Since
d
dx
(x1)
3
= 3(x1)
2
,
we have
_
(x 1)
2
dx =
1
3
(x 1)
3
+C
1
. This agrees with the rst answer, with C
1
= C +1/3.
(c)
_
(x 1)(x +2) dx =
_
(x
2
+x 2) dx =
1
3
x
3
+
1
2
x
2
2x +C
(d) Either rst evaluate (x +2)
3
= x
3
+6x
2
+12x +8, to get
_
(x +2)
3
dx =
1
4
x
4
+2x
3
+6x
2
+8x +C,
9 I NTEGRATI ON 27
or:
_
(x +2)
3
=
1
4
(x +2)
4
+C
1
. (e)
_
(e
3x
e
2x
+e
x
) dx =
1
3
e
3x
1
2
e
2x
+e
x
+C
(f)
_
x
3
3x +4
x
dx =
_
_
x
2
3 +
4
x
_
dx =
1
3
x
3
3x +4 ln |x| +C
5. (a) First simplify the integrand:
(y 2)
2
y
=
y
2
4y +4
y
= y
3/2
4y
1/2
+ 4y
1/2
. From this we get
_
(y 2)
2
y
dy =
_
(y
3/2
4y
1/2
+4y
1/2
) dy =
2
5
y
5/2
8
3
y
3/2
+8y
1/2
+C.
(b) Polynomial division:
x
3
x +1
= x
2
x +1
1
x +1
, so
_
x
3
x +1
dx =
1
3
x
3
1
2
x
2
+x ln |x +1| +C.
(c) Since
d
dx
(1 +x
2
)
16
= 16(1 +x
2
)
15
2x = 32x(1 +x
2
)
15
,
_
x(1 +x
2
)
15
dx =
1
32
(1 +x
2
)
16
+C.
10. (a) Easy. (b) (ii)
x +2 = (x +2)
1/2
, and use (a). (iii)
1
4 x
= (4 x)
1/2
, and use (a).
11. F(x) =
_
(
1
2
e
x
2x) dx =
1
2
e
x
x
2
+C. F(0) =
1
2
implies C = 0.
(b) F(x) =
_
(x x
3
) dx =
1
2
x
2
1
4
x
4
+C. F(1) =
5
12
implies C =
1
6
.
13. f
(x) =
_
(x
2
+x
3
+2) dx = x
1
+
1
4
x
4
+2x+C. With f
(1) = 1/4 we have 1/4 = 1+

1
4
+2+C,
so C = 1. New integration yields f (x) =
_
(x
1
+
1
4
x
4
+2x 1) dx = ln x +
1
20
x
5
+x
2
x +D.
With f (1) = 0 we have 0 = ln 1 +
1
20
+1 1 +D, so D = 1/20.
9.2
5. We do only (c) and (f):
_
3
2
_
1
2
x
2
1
3
x
3
_
dx =
3
2
(
1
6
x
3
1
12
x
4
) =
3
2
(
1
12
x
3
(2 x) =
27
12
+
32
12
=
5
12
.
(f)
_
3
2
_
1
t 1
+t
_
dt =
3
2
_
ln(t 1) +
1
2
t
2
_
= ln 2 +
9
2

4
2
= ln 2 +
5
2
6. (a) A sign diagram shows that f (x) > 0 when 0 < x < 1 and when x > 2. (b) f (x) = x
3
3x
2
+2x,
hence f
(x) = 3x
2
6x +2 = 0 for x
0
= 1
3/3 and x
1
= 1 +
3/3. We see that f
(x) > 0
x < x
0
or x > x
1
. Also, f
(x) < 0 x
0
< x < x
1
. So f is (strictly) increasing in (, x
0
] and in
[x
1
, ), and (strictly) decreasing in [x
0
, x
1
]. Hence x
0
is a local maximumpoint and x
1
is a local minimum
point. (c) See the graphinthe text.
_
1
0
f (x) dx =
_
1
0
(x
3
3x
2
+2x) dx =
1
0
_
x
4
4
x
3
+x
2
_
=
1
4
0 =
1
4
.
7. (a) f
(x) = 1 +
3000 000
x
2
= 0 for x =

3000 000 = 1000
3. (Recall x > 0.) For the rest, see the

text.
9.3
2. (a)
_
1
0
(x
p+q
+x
p+r
) dx =
1
0
x
p+q+1
p +q +1
+
x
p+r+1
p +r +1
=
1
p +q +1
+
1
p +r +1
(b) f
(1) = 6 implies a +b = 6. Since f
(x) = 2ax +b, f
(1) = 18 implies 2a +b = 18. It follows

that a = 12 and b = 6, so f
(x) = 12x
2
6x. But then f (x) =
_
(12x
2
6x) dx = 4x
3
3x
2
+ C,
and since we want
_
2
0
(4x
3
3x
2
+C) = 18, we must have
2
0
(x
4
x
3
+Cx) = 18, i.e. C = 5.
3. (a) See text. (b)
_
1
0
(x
2
+2)
2
dx =
_
1
0
(x
4
+4x
2
+4) dx =
1
0
(
1
5
x
5
+
4
3
x
3
+4x) = 83/15
(c)
_
1
0
x
2
+x +
x +1
x +1
dx =
_
1
0
x(x +1) +(x +1)
1/2
x +1
dx =
_
1
0
(x +(x +1)
1/2
) dx =
28 9 I NTEGRATI ON
1
0
(
1
2
x
2
+2(x +1)
1/2
) dx = 2
2
3
2
(d) A
x +b
x +c
+
d
x
= A
x +c +b c
x +c
+
d
x
= A +
A(b c)
x +c
+
d
x
. Now integrate.
6. From y
2
= 3x we get x =
1
3
y
2
, which inserted into the other equation gives y + 1 = (
1
3
y
2
1)
2
, or
y(y
3
6y 9) = 0. Here y
3
6y 9 = (y 3)(y
2
+ 3y + 3), with y
2
+ 3y + 3 never 0. So (0, 0)
and (3, 3) are the only points of intersection. See the text.
7. W(T ) = K(1 e
T
)/T . Here W(T ) 0 as T , and using lHpitals rule, W(T ) K as
T 0
+
. For T > 0, we nd W
(T ) = Ke
T
(1 + T e
T
)/T
2
< 0 because e
T
> 1 + T (see
Problem 6.11.11). We conclude that W(T ) is strictly decreasing and that W(T ) (0, K).
8. (a) f
(x) =
2
x +4 (
x +4 2)
> 0 for x > 0 and f has range (, ), so f has an inverse
dened on (, ). We nd that the inverse is g(x) = e
x/2
+ 4e
x/4
. (y = 4 ln(
x +4 2)
ln(
x +4 2) = y/4

x +4 = e
y/4
+2 x +4 = (e
y/4
+2)
2
x = e
y/2
+4e
y/4
.)
(b) See Fig. A9.3.8. (c) In Fig. A9.3.8 the graphs of f and g are symmetric about the line y = x, so
area A = area B. But area B is the area of a rectangle with base a and height 10 minus the area below the
graph of g over the interval [0, a]. Therefore, B = 10a
_
a
0
(e
x/2
+4e
x/4
) dx = 10a+182e
a/2
16e
a/4
.
Because a = f (10) = 4 ln(
14 2), this simplies to 10a +14 8
14 6.26.
9.4
2. (a) Let n be the total number of individuals. The number of individuals with income in the interval [b, 2b]
is then N = n
_
2b
b
Br
2
dr = n
2b
b
Br
1
=
nB
2b
. Their total income is M = n
_
2b
b
Br
2
r dr =
n
_
2b
b
Br
1
dr = n
2b
b
B ln r = nB ln 2. Hence the mean income is m = M/N = 2b ln 2.
(b) Total demand is x(p) =
_
2b
b
nD(p, r)f (r) dr =
_
2b
b
nAp
Br
2
dr = nABp
_
2b
b
r
2
dr =
nABp
2b
b
r
1
1
= nABp
b
1
2
1
1
1
.
5. (a) See Fig. A9.4.5. (b)
_
t
0
_
g() f ()
_
d =
_
t
0
_
2
3
30
2
+100) d =
1
2
t
2
(t 10)
2
0 for
all t .
(c)
_
10
0
p(t )f (t ) =
_
10
0
_
t
3
+9t
2
+11t 11 +11/(t +1)
_
dt = 940 +11 ln 11 966.38,
_
10
0
p(t )g(t )dt =
_
10
0
_
t
3
19t
2
+79t +121 121/(t +1)
_
dt = 3980/3 121 ln 11 1036.52.
Prole g should be chosen.
9.5
1. (a)
_
x
f
e
x
dx = x
f
(e
x
)
_
1
(e
x
)
g
dx = xe
x
+
_
e
x
dx = xe
x
e
x
+C
(b)
_
3xe
4x
dx = 3x
1
4
e
4x
_
3
1
4
e
4x
dx =
3
4
xe
4x
3
16
e
4x
+C
9 I NTEGRATI ON 29
(c)
_
(1 +x
2
)e
x
dx = (1 +x
2
)(e
x
)
_
2x(e
x
) dx = (1 +x
2
)e
x
+2
_
xe
x
dx
= (1 +x
2
)e
x
2xe
x
2e
x
+C = (x
2
+2x +3)e
x
+C
(d)
_
xln x dx =
1
2
x
2
ln x
_
1
2
x
2
1
x
dx =
1
2
x
2
ln x
_
1
2
x dx =
1
2
x
2
ln x
1
4
x
2
+C
2. (a) See the text.
(b) Recall that
d
dx
2
x
= 2
x
ln 2, and therefore 2
x
/ ln 2 is the indenite integral of 2
x
. If follows that
_
2
0
x2
x
dx =
2
0
x
2
x
ln 2
_
2
0
2
x
ln 2
dx =
8
ln 2
2
0
2
x
(ln 2)
2
=
8
ln 2
_
4
(ln 2)
2
1
(ln 2)
2
_
=
8
ln 2
3
(ln 2)
2
(c) First use integration by parts on the indenite integral: with f (x) = x
2
and g(x) = e
x
,
()
_
x
2
e
x
dx = x
2
e
x
_
2xe
x
dx. To evaluate the last integral we must use integration by parts once
more: with f (x) = 2x and g(x) = e
x
,
_
2xe
x
dx = 2xe
x
_
2e
x
dx = 2xe
x
(2e
x
+C). Inserted into
() this gives
_
x
2
e
x
dx = x
2
e
x
2xe
x
+2e
x
+C, and hence,
_
1
0
x
2
e
x
dx =
1
0
(x
2
e
x
2xe
x
+2e
x
) =
(e 2e +2e) (0 0 +2) = e 2. Alternatively, more compactly using formula (9.5.2):
_
1
0
x
2
e
x
dx =
1
0
x
2
e
x
2
_
1
0
xe
x
dx = e 2
_ 1
0
xe
x
_
1
0
e
x
dx
_
= e 2[e
1
0
e
x
] = e 2
5. (a) By formula (9.5.2),
_
T
0
t e
rt
dt =
T
0
t
1
r
e
rt
_
T
0
1
r
e
rt
dt =
T
r
e
rT
+
1
r
_
T
0
e
rt
dt
=
T
r
e
rT
+
1
r
T
0
1
r
e
rt
=
1
r
2
(1 (1 +rT )e
rT
). Multiply this expression by b.
(b)
_
T
0
(a +bt )e
rt
dt = a
_
T
0
e
rt
dt +b
_
T
0
t e
rt
dt , a.s.o. using (a).
(c)
_
T
0
(a bt +ct
2
)e
rt
dt = a
_
T
0
e
rt
dt b
_
T
0
t e
rt
dt +c
_
T
0
t
2
e
rt
dt . Use the previous results
and
_
T
0
t
2
e
rt
dt =
T
0
t
2
(1/r)e
rt
_
T
0
2t (1/r)e
rt
dt = (1/r)T
2
e
rT
+(2/r)
_
T
0
t e
rt
dt.
9.6
2. (a) See the text. (b) With u = x
3
+2 we get du = 3x
2
dx and
_
x
2
e
x
3
+2
dx =
_
1
3
e
u
du =
1
3
e
u
+C =
1
3
e
x
3
+2
+C.
(c) First attempt: u = x +2, which gives du = dx and
_
ln(x +2)
2x +4
dx =
_
ln u
2u
du.
This does not look very much simpler than the original integral. Abetter idea is to substitute u = ln(x+2).
Then du =
dx
x +2
and
_
ln(x +2)
2x +4
dx =
_
1
2
u du =
1
4
(u)
2
+C =
1
4
(ln(x +2))
2
+C.
(d) First attempt: u = 1+x. Then, du = dx, and
_
x
1 +x dx =
_
(u1)
u du =
_
(u
3/2
u
1/2
) du
=
2
5
u
5/2
2
3
u
3/2
+C =
2
5
(1+x)
5/2
2
3
(1+x)
3/2
+C. Second attempt: u =
1 +x. Then u
2
= 1+x
and 2udu = dx. Then the integral is
_
x
1 +x dx =
_
(u
2
1)u2u du =
_
(2u
4
2u
3
) du e.t.c. Check
that you get the same answer. Actually, even integration by parts works in this case. Put f (x) = x and
g
(x) =
1 +x, and choose g(x) =

2
3
(1 +x)
3/2
. (The answer looks different, but is not.)
30 9 I NTEGRATI ON
(e) With u = 1 + x
2
, x
2
= u 1, and du = 2xdx, so
_
x
3
(1 +x
2
)
3
dx =
_
x
2
x
(1 +x
2
)
3
dx =
1
2
_
u 1
u
3
du =
1
2
_
(u
2
u
3
) du =
1
2
u
1
+
1
4
u
2
+C =
1
2(1 +x
2
)
+
1
4(1 +x
2
)
2
+C.
(f) With u =
4 x
3
, u
2
= 4 x
3
, and 2udu = 3x
2
dx, so
_
x
5
_
4 x
3
dx =
_
x
3
_
4 x
3
x
2
dx
=
_
(4u
2
) u (
2
3
)u du =
_
(
8
3
u
2
+
2
3
u
4
) du =
8
9
u
3
+
2
15
u
5
+C =
8
9
(4x
3
)
3/2
+
2
15
(4x
3
)
5/2
+C
6. (a) I =
_
1
0
(x
4
x
9
)(x
5
1)
12
dx =
_
1
0
x
4
(x
5
1)
13
dx. Introduce u = x
5
1. Then du = 5x
4
dx,
and when x = 0, u = 1, when x = 1, u = 0, and thus I =
_
0
1
1
5
u
13
du =
0
1
1
70
u
14
=
1
70
.
(b) With u =
x, u
2
= x, and 2udu = dx, and so
_
ln x
x
dx = 2
_
ln u
2
du = 4
_
ln u du = 4(u ln u u) +C = 4
x ln
x 4
x +C = 2
x ln x
4
x +C. (Integration by parts also works in this case with f (x) = ln x and g
(x) = 1/
x.)
(c) With u = 1 +
x, u 1 =
x, or (u 1)
2
= x, so 2(u 1)du = dx. When x = 0, u = 1, when
x = 4, u = 3. Hence,
_
4
0
dx
_
1 +
x
=
_
3
1
2(u 1)
u
du = 2
_
3
1
(u
1/2
u
1/2
) du = 2
3
1
(
2
3
(u
3/2
2u
1/2
) du =
8
3
(The substitution u =
_
1 +
x also works.)
7. (a) With u = 1 + e
x
, u > 0, and du =
1
2
x
e
x
dx. When x = 1, u = 1 + e and x = 4 gives
u = 1 +e
2
. Thus we get (note how we carry over the limits of integration):
_
4
1
e
x (1 +e
x
)
dx =
_
1+e
2
1+e
2 du
u
= 2
1+e
2
1+e
ln u = 2 ln(1 +e
2
) 2 ln(1 +e)
(b) Anatural substitution is u = e
x
+1, du = e
x
dx, and dx = du/e
x
= du/(u1). When x = 0, u = 2,
when x = 1/3, u = e
1/3
+1. Thus,
_
1/3
0
dx
e
x
+1
=
_
e
1/3
+1
2
1
u(u 1)
du =
_
e
1/3
+1
2
_
1
u 1

1
u
_
du =
e
1/3
+1
2
_
ln |u 1| ln |u|
_
=
1
3
ln(e
1/3
+1) +ln 2 = ln 2 ln(e
1/3
+1). (Verify the last equality.)
Rewriting the integrand as
e
x
1 +e
x
, the suggested substitution t = e
x
(or even better u = 1 + e
x
),
dt = e
x
dx works well. Verify that you get the same answer.
9.7
3. (a) See answer in the text. Using a simplied notation and the result in Example 1(a), we have:
(b)
_
0
(x 1/)
2
e
x
dx =
0
(x 1/)
2
e
x
+
_
0
2 (x 1/) e
x
dx
= 1/
2
+2
_
0
xe
x
dx (2/)
_
0
e
x
dx = 1/
2
+2/
2
2/
2
= 1/
2
(c)
_
0
(x 1/)
3
e
x
dx =
0
(x 1/)
3
e
x
+
_
0
3 (x 1/)
2
e
x
dx
= 1/
3
+(3/)
_
0
(x 1/)
2
e
x
dx = 1/
3
+(3/)(1/
2
) = 2/
3
5. (a) f
(x) = (1 3 ln x)/x
4
= 0 at x = e
1/3
, and f
(x) > 0 for x < e

1/3
and f
(x) < 0 for x > e

1/3
.
Hence f has a maximum at (e
1/3
, 1/3e). Since f (x) as x 0
+
, there is no minimum. Note
that f (x) 0 as x . (Use lHpitals rule.)
9 I NTEGRATI ON 31
(b)
_
b
a
x
3
ln x dx =
b
a
1
2
x
2
ln x +
_
b
a
1
2
x
3
dx =
b
a
(
1
2
x
2
ln x
1
4
x
2
). This diverges when b = 1
and a 0. But
_
1
x
3
ln x dx = 1/4.
7. If both limits exist, the integral is the sumof the following two limits: I
1
= lim
0
+
_
3
2+
_
1/
x +2
_
dx
and I
2
= lim
0
+
_
3
2
_
1/
3 x
_
dx. Here I
1
= lim
0
+
3
2+
_
2
x +2
_
= lim
0
+
_
2
5 2
_
=
2
5, and I
2
= lim
0
+
3
2
_
2
3 x
_
= lim
0
+
_
2
+2
5
_
= 2
5.
12. (a) The suggested substitution gives
_
+
f (x) dx =
1
_
+
e
u
2
du = 1, by (9.7.8).
(b) Here
_
+
xf (x) dx =
1
_
+
( +
2u)e
u
2
du = , using part (a) and Example 3.
(c) Here I =
_
+
x
2
f (x) dx =
1
_
+
(2
2
u
2
+2
2u +
2
)e
u
2
du
=
2
2
_
+
u
2
e
u
2
du +
2
_
+
ue
u
2
du +

2
_
+
e
u
2
du =
2
+ 0 +
2
. (Note how
integration by parts gives
_
u
2
e
u
2
du =
1
2
ue
u
2
+
_
1
2
e
u
2
du, so
_
+
u
2
e
u
2
du =
1
2
.)
9.8
5. P(10) = 705 gives 641e
10k
= 705, or e
10k
= 705/641. Taking the natural logarithm of both sides yields
10k = ln(705/641), so k = 0.1 ln(705/641).
7. Straightforward. Note that in (9.8.10), if b = 0, there are always two constant solutions, x 0 and
x a/b. The latter is obtained by letting A = 0 in (9.8.10). So in addition to the answer in the text, in
(e) add x 0 as a solution, in (f) add K 0.
9. (a) Use (9.8.7). (Using (9.8.10), and then using N(0) = 1 to determine the constant, is less efcient.)
(b) 800 =
1000
1 +999e
0.39t
999e
0.39t
=
1
4
, so e
0.39t
= 1/3996, and so 0.39t
= ln 3996, etc.
9.9
2. (a) dx/dt = e
2t
/x
2
. Separate:
_
x
2
dx =
_
e
2t
dt . Integrate:
1
3
x
3
=
1
2
e
2t
+C
1
. Solve for x:
x
3
=
3
2
e
2t
+ 3C
1
=
3
2
e
2t
+ C, with C = 3C
1
. Hence, x =
3
_
3
2
e
2t
+C. (You cannot wait to put the
constant in the end. Wrong:
1
3
x
3
=
1
2
e
2t
, x
3
=
3
2
e
2t
, x =
3
_
3
2
e
2t
+C. This is not a solution!)
(b)
_
e
x
dx =
_
e
t
dt . Integrate: e
x
= e
t
+ C
1
. Solve for x: e
x
= e
t
+ C, with C = C
1
.
Hence, x = ln(e
t
+C), so x = ln(e
t
+C). (c) Directly from (9.9.3). (d) Similar to (a).
(e) By (9.9.5), x = Ce
2t
+ e
2t
_
(t )e
2t
dt = Ce
2t
e
2t
_
t e
2t
dt . Here
_
t e
2t
dt = t (
1
2
)e
2t
+
1
2
_
e
2t
dt = (
1
2
t
1
4
)e
2t
and thus x = Ce
2t
e
2t
(
1
2
t
1
4
)e
2t
= Ce
2t
+
1
2
t +
1
4
.
(f) Formula (9.9.5): x = Ce
3t
+e
3t
_
e
3t
t e
t
2
3t
dt = Ce
3t
+e
3t
_
t e
t
2
dt = Ce
3t
+
1
2
e
t
2
3t
3. The equation is separable: dk/k = se
t
dt , so ln k =
s
e
t
+ C
1
, or k = e
s
e
t
e
C
1
= Ce
s
e
t
. With
k(0) = k
0
, we have k
0
= Ce
s
, and thus k = k
0
e
s
(e
t
1)
.
32 9 I NTEGRATI ON
5. (a) See the text. (b)
_
K
dK =
_
L
0
e
t
dt , so
1
1
K
1
=
L
0
e
t
+C
1
.
Hence, K
1
=
L
0
(1 )
e
t
+(1 )C
1
. For t = 0, K
1
0
=
L
0
(1 )
+(1 )C
1
, so
K
1
=
(1 ) L
0
(e
t
1) +K
1
0
, from which we nd K.
6.
t
x
dx
dt
= a is separable:
dx
x
= a
dt
t
, so
_
dx
x
= a
_
dt
t
. Integrating yields, ln x = a ln t + C
1
, so
x = e
a ln t +C
1
= (e
ln t
)
a
e
C
1
= Ct
a
, with C = e
C
1
.
3. (a)
_
12
0
50 dx =
12
0
50x = 600 (b)
_
2
0
(x
1
2
x
2
) dx =
2
0
(
1
2
x
2
1
6
x
3
) =
2
3
(c)
_
3
3
(u +1)
2
du =
3
3
1
3
(u +1)
3
du = 24 (d)
_
5
1
2
z
dz =
5
1
2 ln z = 2 ln 5
(e)
_
12
2
3 dt
t +4
dt =
12
2
3 ln(t +4) = 3 ln(8/3) (f)
_
4
0
v
_
v
2
+9 dv =
4
0
1
3
(v
2
+9)
3/2
= 98/3
5. (a) With u = 9 +
x, x = (u 9)
2
and dx = 2(u 9) du. When x = 0, u = 9 and x = 25 gives
u = 14. Thus,
_
25
0
1
9 +
x
dx =
_
14
9
2(u 9)
u
du =
_
14
9
_
2
18
u
_
du = 10 18 ln
14
9
.
(b) With u =
t +2, t = u
2
2, and dt = 2u du. When t = 2, u = 2, and t = 7 gives u = 3. Hence,
_
7
2
t
t +2 dt =
_
3
2
(u
2
2)u 2u du = 2
_
3
2
(u
4
2u
2
) du = 2
3
2
_
1
5
u
5
2
3
u
3
_
= 886/15
(c) With u =
3
19x
3
+8, u
3
= 19x
3
+ 8, so 3u
2
du = 57x
2
dx. When x = 0, u = 2 and x = 1 gives
u = 3. Then
_
1
0
57x
2
3
_
19x
3
+8 dx =
_
3
2
3u
3
du =
3
2
3
4
u
4
= 195/4.
10. Equilibrium when 50/(Q
+5) = 10 +Q
, i.e. (Q
)
2
+50Q
275 = 0. The only positive solution

is Q
= 5, and then P
= 5. CS =
_
5
0
_
50
Q+5
5
_
dQ =
5
0
[50 ln(Q+5) 5Q] = 50 ln 2 25,
PS =
_
5
0
(5 4.5 0.1Q) dQ = 1.25.
11. (a) f
(t ) = 4
2 ln t (1/t ) t (ln t )
2
1
t
2
= 4
(2 ln t ) ln t
t
2
, and
f
(t ) = 4
(2 (1/t ) 2 ln t (1/t )) t
2
(2 ln t (ln t )
2
) 2t
t
4
= 8
(ln t )
2
3 ln t +1
t
3
(b) Stationary points: f
(t ) = 0 ln t (2 ln t ) = 0 ln t = 2 or ln t = 0
t = e
2
or t = 1 Since f
(1) = 8 > 0 and f
(e
2
) = 8e
6
, t = 1 is a local minimum point and
t = e
2
7.4 is a local maximum point. We nd f (1) = 0 and f (e
2
) = 16e
2
2.2.
(c)
d
dt
_
4
3
(ln t )
3
_
=
4
3
3(ln t )
2
1
t
= f (t ), so
_
f (t ) dt =
4
3
(ln t )
3
+ C. Since f (t ) 0 for all t > 0,
the area is
_
e
2
1
f (t ) dt =
e
2
1
4
3
(ln t )
3
=
4
3
2
3
0 =
32
3
.
12. Straightforward by using (9.8.8)(9.8.10).
10 I NTEREST RATES AND PRESENT VAL UES 33
13. (a) Separable.
_
x
2
dx =
_
t dt , and so 1/x =
1
2
t
2
+C
1
, or x = 1/(C
1
2
t
2
), (with C = C
1
).
(b) and (c): Direct use of (9.9.3). (d) Using (9.9.5), x = Ce
5t
+ 10e
5t
_
t e
5t
dt . Here
_
t e
5t
dt =
t
1
5
e
5t
1
5
_
e
5t
dt =
1
5
t e
5t
1
25
e
5t
. Thus x = Ce
5t
+10e
5t
(
1
5
t e
5t
1
25
e
5t
) = Ce
5t
+2t
2
5
.
(e) x = Ce
t /2
+e
t /2
_
e
t /2
e
t
dt = Ce
t /2
+e
t /2
_
e
3t /2
dt = Ce
t /2
+e
t /2 2
3
e
3t /2
= Ce
t /2
+
2
3
e
t
.
(e) x = Ce
3t
+e
3t
_
t
2
e
3t
dt = Ce
3t
+e
3t
(
1
3
t
2
e
3t
2
3
_
t e
3t
dt )
= Ce
3t
+
1
3
t
2
2
3
e
3t
(
1
3
t e
3t
1
3
_
e
3t
dt ) = Ce
3t
+
1
3
t
2
2
9
t +
2
27
.
16. (a) and (b), see the text. (c) F
(x) = f
(x) = a
2
e
x
(e
x
a)(e
x
+a)
3
. Note that F
(x) = 0
for e
x
= a, i.e. for x
0
= ln a/. Since F
(x) changes sign about x

0
= ln a/, this is an inection
point. F(x
0
) = F(ln a/) = a/(a +a) = 1/2. See the graph in Fig. A9.R.16. (d)
_
f (x) dx =
lim
a
_
0
a
f (x)dx + lim
b
_
b
0
f (x) dx = lim
a
[F(0) F(a)] + lim
b
[F(b) F(0)] = 1, by (a).
10 Interest Rates and Present Values
10.1
3. We solve (1 + p/100)
100
= 100 for p. Raising each side to 1/100, 1 + p/100 =
100
100, so p =
100(
100
100 1) 100(1.047 1) = 4.7.

5. Use formula (10.1.2). (i) R = (1 +0.17/2)
2
1 = (1 +0.085)
2
1 = 0.177225 or 17.72%
For (ii) ansd (iii) see the answers in the text.
10.2
4. If it looses 90% of its value, then e
0.1t
= 1/10, so 0.1t
= ln 10, hence t
= (ln 10)/0.1 23.

6. With g(x) = (1 +r/x)
x
for all x > 0, then ln g(x) = x ln(1 +r/x). Differentiation gives g
(x)/g(x) =
ln(1 +r/x) +x(x/r
2
)/(1 +r/x) = ln(1 +r/x) (x/r)/(1 +r/x), as claimed in the problem. Then
see the answer in the text.
10.3
3. (a) We nd f
(t ) = 0.05(t +5)(35 t )e
t
. Obviously, f
(t ) > 0 for t < 35 and f
(t ) < 0 for t > 35,

so t = 35 maximizes f (with f (35) 278). (b) f (t ) 0 as t . See the graph in Fig. A10.3.3.
10.4
2. We use formula (10.4.5): (a)
1
5
1
1
5
=
1
4
(b)
0.1
1 0.1
=
0.1
0.9
=
1
9
(c)
517
1 1/1.1
=
517 1.1
0.1
= 5687
(d)
a
1 1/(1 +a)
= 1 +a (e)
5
1 3/7
=
35
4
6. Let x denote the number of years beyond 1971 that the extractable resources of iron will last. Then
794+7941.05+ +794(1.05)
x
= 24910
3
. Using (10.4.3), 794[1(1.05)
x+1
]/(11.05) = 24910
3
or (1.05)
x+1
= 249 10
3
0.05/794 16.68. Using a calculator, we nd x (ln 16.68/ ln 1.05) 1
56.68, so the resources will be exhausted part way through the year 2028.
34 10 I NTEREST RATES AND PRESENT VAL UES
8. (a) The quotient of this innite series is e
rt
, so the sum is f (t ) =
P(t )e
rt
1 e
rt
=
P(t )
e
rt
1
(b) f
(t ) =
P
(t )(e
rt
1) P(t )re
rt
(e
rt
1)
2
, and t
> 0 can only maximize f (t ) if f
(t
) = 0, that is, if
P
(t
)(e
rt
1) = rP(t
)e
rt
, which implies that

P
(t
)
P(t
)
=
r
1 e
rt
.
(c) lim
r0
r
1 e
rt
=
0
0
= lim
r0
1
t
e
rt

1
t
10.5
4. Offer (a) is better, because the second offer has present value 4600
1 (1.06)
5
1 (1.06)
1
20 540.
7. This is a geometric series with rst term a = D/(1 +r) and quotient k = (1 +g)/(1 +r). It converges
iff k < 1, i.e. iff 1 +g < 1 +r, or g < r. The sum is
a
1 k
=
D/(1 +r)
1 (1 +g)/(1 +r)
=
D
r g
.
10.6
4. Schedule (b) has present value
12 000 1.115
0.115
[1 (1.115)
8
] 67 644.42.
Schedule (c) has present value 22 000 +
7000
0.115
[1 (1.115)
12
] 66 384.08.
Thus schedule (c) is cheapest. When the interest rate becomes 12.5 %, schedules (b) and (c) have present
values equal to 65907.61 and 64374.33, respectively.
10.7
5. After dividing all the amounts by $ 10 000, the equation is f (s) = s
20
+s
19
+ +s
2
+s 10 = 0. Then
f (0) = 10 andf (1) = 10, sobythe intermediate value theorem(Theorem7.10.1), there exists a number
s
such that f (s
) = 0. Here s
is unique because f
(s) > 0. In fact f (s) = 10+(s s

21
)/(1s), and
f (s
) = 0 (s
)
10
11s
+10 = 0. Then s
= 0.928 is an approximate root, which corresponds

to an internal rate of return of about 7.8 %. (See Problem 7.R.26.)
10.8
3. Equilibrium requires P
t
= P
t +1
, or P
t +1
= (/)P
t
+( + )/. Using (10.8.4) we obtain
the answer in the text.
3. If you borrow $ a at the annual interest rate of 11% with interest paid yearly, then the dept after 1 year
is a(1 +11/100) = a(1.11); if you borrow at annual interest rate 10% with interest paid monthly, your
dept after 1 year will be a(1 +10/12 100)
12
1.1047a, so schedule (ii) is preferable.
6. We use formula (10.4.5): (a)
44
1 0.56
= 100 (b) The rst term is 20 and the quotient is 1/1.2, so the
sum is
20
1 1/1.2
= 120 (c)
3
1 2/5
= 5 (d) The rst term is (1/20)
2
= 400 and the quotient is
1/20, so the sum is
400
1 1/20)
= 8000/19
11 F UNCTI ONS OF MANY VARI ABL ES 35
8. (a) See the text. (b) We use formula (10.5.3) on the future value of an annuity. See the text.
(c) See the text.
11. (a) f
(t ) = 100e
t /2
e
rt
_
1
4
t
r
_
. We see that f
(t ) = 0 for t
= 1/16r
2
. Since f
(t ) > 0 for t < t
and f
(t ) < 0 for t > t
, t
maximizes f (t ).
(b) f
(t ) = 200e
1/t
e
rt
_
1
t
2
r
_
. We see that f
(t ) = 0 for t
= 1/
r. Since f
(t ) > 0 for t < t
and f
(t ) < 0 for t > t
, t
maximizes f (t ).
12. (a) F(10) F(0) =
_
10
0
(1 +0.4t ) dt =
10
0
(t +0.2t
2
) = 30. (Note: the total revenue is F(10) F(0),
not F(10).) (b) See Example 9.5.3.
11 Functions of Many Variables
11.1
6. (a) The denominator must not be 0, so the function is dened for those (x, y) where y = x 2.
(b) Must require 2 (x
2
+y
2
) 0, i.e. x
2
+y
2
2.
(c) Put a = x
2
+y
2
. We must require (4 a)(a 1) 0, i.e. 1 a 4. (Use a sign diagram.)
7. For (a) and (c) see the text. (b) Since (x a)
2
0 and (y b)
2
0, it sufces to assume that x = a
and b = b, because then we take ln of positive numbers.
11.2
3. (a) and (b) are straightforward. (c) f (x, y) = (x
2
2y
2
)
5
= u
5
, where u = x
2
2y
2
. Then
f
1
(x, y) = 5u
4
u
1
= 5(x
2
2y
2
)
4
2x = 10x(x
2
2y
2
)
4
. In the same way, f
2
(x, y) = 5u
4
u
2
=
5(x
2
2y
2
)
4
(4y) = 20y(x
2
2y
2
)
4
. Finally, f
12
(x, y) is the derivative of f
1
(x, y) w.r.t. y, keeping
x constant, so f
12
(x, y) = (/y)(10x(x
2
2y
2
)
4
) = 10x4(x
2
2y
2
)
3
(4y) = 160xy(x
2
2y
2
)
3
.
5. (a)(c) are easy. (d) z = x
y
= (e
ln x
)
y
= e
y ln x
= e
u
with u = y ln x. Then z
x
= e
u
u
x
= x
y
(y/x) =
yx
y1
. In the same way, z
y
= e
u
u
y
= x
y
ln x. Moreover, z
xx
= (/x)(yx
y1
) = y(y 1)x
y2
.
(When differentiating x
y1
partially w.r.t. x, y is a constant, so the rule dx
a
/dx = ax
a1
applies.)
z
yy
= (/y)(x
y
ln x) = x
y
(ln x)
2
. Finally, z
xy
= (/y)(yx
y1
) = x
y1
+ yx
y1
ln x. (Note that if
w = x
y1
= x
v
, with v = y 1, then w
y
= x
v
ln x 1 = x
y1
ln x. Or: w = x
y1
= (1/x)x
y
etc.)
11.3
8. (a) The point (2, 3) lies on the level curve z = 8, so f (2, 3) = 8. The points (x, 3) are those on the line
y = 3 parallel to the x-axis. This line intersects the level curve z = 8 when x = 2 and x = 5.
(b) See the text. (c) If at A you look in the direction of the positive x-axis, you meet level sets whose
values increase, so f
1
(x, y) > 0. A rough estimate of f
1
(x, y) at A is 2, because if you go one unit in
the positive x-axis direction from A, then f (x, y) increases from 8 to 10.
9. (a) It might help to regard the gure as a map of a mountain. At P the terrain is raising in the direction
of the positive x-axis, so f
x
(P) > 0. The terrain is sloping downwards in the direction of the positive
y-axis so f
y
(P) < 0. (b) (i) The line x = 1 has no point in common with any of the given level curves.
36 11 F UNCTI ONS OF MANY VARI ABL ES
(ii) The line y = 2 intersects the level curve z = 2 at x = 2 and x = 6 (approximately).
(c) If you start at the point (6, 0) and move up along the line 2x +3y = 12, you rst meet the level curve
z = f (x, y) = 1. Moving further you meet level curves with higher z-values. The level curve with the
highest z-value you meet is z = 3, where the straight just touches the level curve.
10. F(1, 0) = F(0, 0) +
_
1
0
F
1
(x, 0) dx
_
1
0
2 dx = 2, F(2, 0) = F(1, 0) +
_
2
1
F
1
(x, 0) dx F(1, 0) +2,
F(0, 1) = F(0, 0) +
_
1
0
F
2
(0, y) dy 1, F(1, 1) = F(0, 1) +
_
1
0
F
1
(x, 1) dx F(0, 1) +2,
F(1, 1) = F(1, 0) +
_
1
0
F
2
(1, y) dy F(1, 0) +1.
11.6
2. (a)(d) are routine. (e) f (x, y, z) = (x
2
+y
3
+z
4
)
6
= u
6
, with u = x
2
+y
3
+z
4
. Then f
1
= 6u
5
u
1
=
6(x
2
+y
3
+z
4
)
5
2x = 12x(x
2
+y
3
+z
4
)
5
, f
2
= 6u
5
u
2
= 6(x
2
+y
3
+z
4
)
5
3y
2
= 18y
2
(x
2
+y
3
+z
4
)
5
,
f
3
= 6u
5
u
3
= 6(x
2
+y
3
+z
4
)
5
4z
3
= 24z
3
(x
2
+y
3
+z
4
)
5
(f) f (x, y, z) = e
xyz
= e
u
, with u = xyz,
gives f
1
= e
u
u
1
= e
xyz
yz. Similarly, f
2
= e
u
u
2
= e
xyz
xz, and f
3
= e
u
u
3
= e
xyz
xy
5. When r and w are constants, so is (1/r +1/w), and thus /p =
1
2
p(1/r +1/w).
10. From f = x
y
z
we get () ln f = y
z
ln x. Differentiating () w.r.t x yields f
x
/f = y
z
/x, and so
f
x
= fy
z
/x = x
y
z
y
z
/x = y
z
x
y
z
1
. Differentiating () w.r.t y yields f
y
/f = zy
z1
ln x, and so f
y
=
zy
z1
(ln x)x
y
z
. Differentiating () w.r.t z yields f
z
/f = y
z
(ln y)(ln x), and so f
z
= y
z
(ln x)(ln y)x
y
z
.
11.7
2. (a) Y
K
= aAK
a1
and Y
K
= aBL
a1
, so KY
K
+LY
L
= aAK
a
+aBL
a
= a(AK
a
+BL
a
) = aY
(b) KY
K
+LY
L
= KaAK
a1
L
b
+LAK
a
bL
b1
= aAK
a
L
b
+bAK
a
L
a
= (a +b)AK
a
L
b
= (a +b)Y
(c) Y
K
=
2aKL
5
bK
4
L
2
(aL
3
+bK
3
)
2
and Y
L
=
2bK
5
L aK
2
L
4
(aL
3
+bK
3
)
2
, so
KY
K
+ LY
L
=
2aK
2
L
5
bK
5
L
2
+2bK
5
L
2
aK
2
L
5
(aL
3
+bK
3
)
2
=
K
2
L
2
(aL
3
+bK
3
)
(aL
3
+bK
3
)
2
=
K
2
L
2
aL
3
+bK
3
= Y.
(According to Section 12.6 these functions are homogeneous of degrees a, a +b, and 1, respectively, so
the results we obtained are immediate consequences of Eulers Theorem, (12.6.2).)
7. Y
K
= (m/)a()K
1
Ae
t
_
aK
+bL
_
(m/)1
= maK
1
Ae
t
_
aK
+bL
_
(m/)1
,
Y
L
= (m/)b()L
1
Ae
t
_
aK
+bL
_
(m/)1
= mbL
1
Ae
t
_
aK
+bL
_
(m/)1
.
Thus, KY
K
+LY
L
= m(aK
+bL
)Ae
t
_
aK
+bL
_
(m/)1
= mY. (This function is homo-
geneous of degree m, so the result is an immediate consequences of Eulers Theorem, (12.6.2).)
11.8
4.

m
_
pD
m
_
= p
mD
m
D
m
2
=
p
m
2
(mD
m
D) =
pD
m
2
[El
m
D 1] > 0 iff El
m
D > 1,
so pD/m increases with m if El
m
D > 1. (Using the formulas in Problem 7.7.9, the result also follows
from the fact that El
m
(pD/m) = El
m
p +El
m
D El
m
m = El
m
D 1.)
7. (a) z = (x
2
y
4
+2)
5
= u
5
, with u = x
2
y
4
+2, so
z
x
= 5u
4
u
x
= 5(x
2
y
4
+2)
4
2xy
4
= 10xy
4
(x
2
y
4
+2)
4
For the rest see the text.
12 TOOL S F OR COMPARATI VE STATI CS 37
11. (b) We want to nd all (x, y) that satisfy both equations (i) 4x
3
8xy = 0 and (ii) 4y 4x
2
+ 4 = 0.
From (i), 4x(x
2
2y) = 0, which means that x = 0, or x
2
= 2y. For x = 0, (ii) yields y = 1, so
(x, y) = (0, 1) is one solution. With x
2
= 2y, (ii) reduces to 4y 8y + 4 = 0, or y = 1. But then
x
2
= 2, such that x =
2. So the two additional solutions are (x, y) = (
2, 1).
12 Tools for Comparative Statics
12.1
5. We look at (c) and (d). (c): If z = F(x, y) = xy with x = f (t ) and y = g(t ), then F
1
(x, y) = y,
F
2
(x, y) = x, dx/dt = f
(t ), and dy/dt = g
(t ), so formula (12.1.1) gives

dz/dt = F
1
(x, y)(dx/dt ) +F
2
(x, y)(dy/dt ) = yf
(t ) +xg
(t ) = g(t )f
(t ) +f (t )g
(t ).
(d): If z = F(x, y) =
x
y
with x = f (t ) and y = g(t ), then F
1
(x, y) =
1
y
, F
2
(x, y) =
x
y
2
,
dx
dt
= f
(t ),
and
dy
dt
= g
(t ), so formula (12.1.1) gives

dz
dt
= F
1
(x, y)
dx
dt
+ F
2
(x, y)
dy
dt
=
1
y
f
(t )
x
y
2
g
(t ) =
yf
(t ) xg
(t )
y
2
=
g(t )f
(t ) f (t )g
(t )
(g(t ))
2
.
6. Let U(x) = u(x, h(x)) = ln[x
+(ax
4
+b)
/3
]

3
ln(ax
4
+b).
Then U
(x) =
x
1
(3b ax
4
)
3[x
+(ax
4
+b)
/3
](ax
4
+b)
. So U
(x
) = 0 at x
=
4
3b/a, whereas U
(x) > 0 for

x < x
and U
(x) < 0 for x > x
. Hence x
maximizes U.
7. Differentiating (12.1.1) w.r.t. t yields, d
2
z/dt
2
= (d/dt )[F
1
(x, y) dx/dt ] + (d/dt )[F
2
(x, y) dy/dt ].
Here (d/dt )[F
1
(x, y) dx/dt ] = [F
11
(x, y) dx/dt +F
12
(x, y) dy/dt ]dx/dt +F
1
(x, y) d
2
x/dt
2
,
(d/dt )[F
2
(x, y) dy/dt ] = [F
21
(x, y) dx/dt + F
22
(x, y) dy/dt ] dy/dt + F
2
(x, y) d
2
y/dt
2
. Assuming
F
12
= F
21
, the conclusion follows.
12.2
2. (a) Let z = F(x, y) = xy
2
with x = t + s
2
and y = t
2
s. Then F
1
(x, y) = y
2
, F
2
(x, y) = 2xy,
x/t = 1, and y/t = 2t s. Then (12.2.1) gives z/t = F
1
(x, y)(x/t ) + F
2
(x, y)(y/t ) =
y
2
+2xy2t s = (t
2
s)
2
+2(t +s
2
)t
2
s2t s = t
3
s
2
(5t +4s
2
). z/s is found in the same way.
(b)
z
t
= F
1
(x, y)
x
t
+F
2
(x, y)
y
t
=
2y
(x +y)
2
e
t +s
+
2sx
(x +y)
2
e
t s
, etc.
3. It is important that you can do these problems, because in economic applications, functions are frequently
not completely specied. (a) z
r
= F
u
u
r
+F
v
v
r
+F
w
w
r
= F
u
2r +F
v
0+F
w
(1/r) = 2rF
u
+(1/r)F
w
.
Check that you get the same answers as in the text.
7. Use the formulas in (12.2.1) and see the text. Only the notation is different.
8. (a) Let u = ln v, where v = x
3
+ y
3
+ z
3
3xyz. Then u/x = (1/v)(v/x) = (3x
2
3yz)/v.
Similarly, u/y = (3y
2
3xz)/v, and u/z = (3z
2
3xy)/v. Hence,
x
u
x
+y
u
y
+z
u
z
=
1
v
(3x
3
3xyz) +
1
v
(3y
3
3xyz) +
1
v
(3z
3
3xyz) =
3v
v
= 3
38 12 TOOL S F OR COMPARATI VE STATI CS
which proves (i). Equation (ii) is then proved by elementary algebra.
(b) Note that f is here a function of one variable. With z = f (u) where u = x
2
y, we get z/x =
f
(u)u
x
= 2xyf
(x
2
y). Likewise, z/y = x
2
f
(x
2
y), so xz/x = 2x
2
yf
(x
2
y) = 2yz/y.
12.3
2. (a) See the text. (b) Put F(x, y) = x y +3xy. Then F
1
= 1 +3y, F
2
= 1 +3x, F
11
= 0, F
12
= 3,
F
22
= 0. In particular, y
= F
1
/F
2
= (1 +3y)/(1 +3x). Moreover, using equation (12.3.3),
y
=
1
(F
2
)
3
_
F
11
(F
2
)
2
2F
12
F
1
F
2
+F
22
(F
1
)
2
_
=
6(1 +3y)(1 +3x)
(1 +3x)
3
) =
6(1 +3y)
(1 +3x)
2
.
(c) Put F(x, y) = y
5
x
6
. Then F
1
= 6x
5
, F
2
= 5y
4
, F
11
= 30x
4
, F
12
= 0, F
22
= 20y
3
, so
y
= F
1
/F
2
= (6x
5
/5y
4
) = 6x
5
/5y
4
. Moreover, using equation (12.3.3),
y
=
1
(5y
4
)
3
_
(30x
4
)(5y
4
)
2
+20y
3
(6x
5
)
2
_
=
6x
4
y
4

144x
10
25y
9
.
3. (a) With F(x, y) = 2x
2
+ xy + y
2
, y
= F
1
/F
2
= (4x + y)/(x + 2y) = 4 at (2, 0). Moreover,
y
= (28x
2
+14y
2
+14xy)/(x +2y)
3
= 14 at (2, 0). The tangent has the equation y = 4x +8.
(b) y
= 0 requires y = 4x. Inserting this into the original equation gives the two points.
4. If we dene F(x, y) = 3x
2
3xy
2
+ y
3
+ 3y
2
, the given equation is F(x, y) = 4. Now, F
1
(x, y) =
6x3y
2
andF
2
(x, y) = 6xy+3y
2
+6y, soaccordingto(12.3.1), y
= (6x3y
2
)/(6xy+3y
2
+6y).
12.4
1. (a) and (b) are easy. (c) Dening F(x, y, z) = e
xyz
3xyz, the given equation is F(x, y, z) = 0.
Now, F
x
(x, y, z) = yze
xyz
3yz, F
z
(x, y, z) = xye
xyz
3xy, so (12.4.1) gives, z
x
= F
x
/F
z
=
(yze
xyz
3yz)/(xye
xyz
3xy) = yz(e
xyz
3)/xy(e
xyz
3) = z/x. (Actually, the equation
e
xyz
= 3xyz has two constant solutions. From xyz = c we nd z
x
much easier.)
3. (a) Equation () is here P/2
= w. Solve for L
. (b) The rst-order condition is now

Pf
(L
) C
L
(L
, w) = 0 ()
Differentiate () partially w.r.t. P keeping in mind that L
depends on P. To nd the partial derivative

of Pf
(L
) w.r.t. P use the product rule to get 1 f
(L
) + Pf
(L
)(L
/P). The partial deriva-

tive of C
L
(L
, w) w.r.t. P is C
LL
(L
, w)(L
/P). So, all in all, f
(L
) + Pf
(L
)(L
/P)
C
LL
(L
, w)(L
/P) = 0. Then solve for L
/P.
Differentiating () w.r.t. w gives, Pf
(L
)(L
/w) C
LL
(L
, w)(L
/w) C
Lw
(L
, w) = 0.
Then solve for L
/w.
6. (a) F
1
(x, y) = e
y3
+y
2
and F
2
(x, y) = xe
y3
+2xy 2. Hence, the slope of the tangent to the level
curve F(x, y) = 4 at the point (1, 3) is y
= F
1
(1, 3)/F
2
(1, 3) = 10/5 = 2.
(b) Taking the logarithm of both sides, we get (1 +c ln y) ln y = ln A+ ln K + ln L. Differentiation
with respect to K gives
c
y
y
K
ln y +(1+c ln y)
1
y
y
K
=

K
. Solving for y/K yields the given answer.
y/K is found in the same way.
12.5
3. With F(K, L) = AK
a
L
b
, F
K
= aF/K, F
L
= bF/L, F
KK
= a(a 1)F/K
2
, F
KL
= abF/KL, and
F
LL
= b(b1)F/L
2
. Also, F
K
F
L
(KF
K
+LF
L
) = (aF/K)(bF/L)(a+b)F = ab(a+b)F
3
/KL.
Moreover, KL
_
(F
L
)
2
F
KK
2F
K
F
2
F
KL
+(F
K
)
2
F
LL
_
= ab(a +b)F
3
/KL. It follows that
KL
= 1.
12.6
3. (2): xf
1
(x, y) +yf
2
(x, y) = x(y
2
+3x
2
) +y2xy = 3(x
3
+xy
2
) = 3f (x, y)
(3): It is easy to see that f
1
(x, y) = y
2
+3x
2
and f
2
(x, y) = 2xy are homogeneous of degree 2.
(4): f (x, y) = x
3
+xy
2
= x
3
[1 +(y/x)
2
] = y
3
[(x/y)
3
+x/y]
(5): x
2
f
11
+2xyf
12
+y
2
f
22
= x
2
(6x) +2xy(2y) +y
2
(2x) = 6x
3
+4xy
2
+2xy
2
= 3 2f (x, y)
4. Using the results in Example 11.2.1(b), x
f
x
+ y
f
y
=
xy
3
x
3
y +x
3
y xy
3
(x
2
+y
2
)
2
= 0 = 0 f ,
so f is homogeneous of degree 0.
8. From (), with k = 1, we get f
11
= (y/x)f
12
and f
22
= (x/y)f
21
. With f
12
= f
21
we get
f
11
f
22
(f
12
)
2
= (y/x)f
12
(x/y)f
12
(f
12
)
2
= 0.
12.7
1. (a) and (f) are easy. In (b) you can use Eulers theorem, as in (e) below.
(c) h(t x, ty, t z) =
t x +
ty +
t z
t x +ty +t z
=
t (
x +
y +
z )
t (x +y +z)
= t
1/2
h(x, y, z) for all t > 0, so
his homogeneous of degree 1/2. (d) G(t x, ty) =
t xty ln
(t x)
2
+(ty)
2
t xty
= t
xy ln
t
2
(x
2
+y
2
)
t
2
xy
=
t G(x, y) for all t > 0, so G is homogeneous of degree 1. (e) xH
x
+ yH
y
= x(1/x) + y(1/y) = 2.
Since 2 is not equal to k(ln x + ln y) for any constant k, by Eulers theorem, H is not homogeneous of
any degree.
2. (a) f (t x
1
, t x
2
, t x
3
) =
(t x
1
t x
2
t x
3
)
2
(t x
1
)
4
+(t x
2
)
4
+(t x
3
)
4
_
1
t x
1
+
1
t x
2
+
1
t x
3
_
=
t
6
(x
1
x
2
x
3
)
2
t
4
(x
4
1
+x
4
2
+x
4
3
)
_
1
t
__
1
x
1
+
1
x
2
+
1
x
3
_
= tf (x
1
, x
2
, x
3
), so f is homogeneous of degree 1.
(b) x(t v
1
, t v
2
, . . . , t v
n
) = A
_
1
(t v
1
)
+
2
(t v
2
)
+ +
n
(t v
n
)
_
/
=
A
_
t
(
1
v
1
+
2
v
2
+ +
n
v
n
)
_
/
= (t
)
/
A
_
1
v
1
+
2
v
2
+ +
n
v
n
_
/
=
t
A
_
1
v
1
+
2
v
2
+ +
n
v
n
_
/
= t
x(x
1
, x
2
, x
3
), so x is homogeneous of degree .
12.8
1. In both (a) and (b) we use the approximation f (x, y) f (0, 0) +f
1
(0, 0)x +f
2
(0, 0)y.
(a) For f (x, y) =

1 +x +y, f (0, 0) = 1, and f
1
(x, y) = f
2
(x, y) =
1
2
1 +x +y
, so f
1
(0, 0) =
f
2
(0, 0) = 1/2, and the linear approximation to f (x, y) about (0, 0) is f (x, y) 1 +
1
2
x +
1
2
y.
(b) For f (x, y) = e
x
ln(1 + y), f
1
(x, y) = e
x
ln(1 + y) and f
2
(x, y) =
e
x
1 +y
. Here, f (0, 0) = 0,
f
1
(0, 0) = e
0
ln 1 = 0 and f
2
(0, 0) = 1. That yields f (x, y) = e
x
ln(1 +y) 0 +0 x +1 y = y.
40 12 TOOL S F OR COMPARATI VE STATI CS
3.
g
=
1
1
_
(1 +)(1 +)
_
1/(1)1
(1 +)
=
1
1
_
(1 +)(1 +)
_
/(1)
(1 +)
,
g
=
1
1
_
(1 +)(1 +)
_
/(1)
(1 +)(1 +)
1
. See the text.
7. We use formula (12.8.3). (a) Here, z/x = 2x and z/y = 2y. At (1, 2, 5), we get z/y = 2 and
z/x = 4, so the tangent plane has the equation z 5 = 2(x 1) +4(y 2) z = 2x +4y 5.
(b) Fromz = (y x
2
)(y 2x
2
) = y
2
3x
2
y +2x
4
we get z/x = 6xy +8x
3
and z/y = 2y 3x
2
.
Thus, at (1, 3, 2) we have z/x = 10 and z/y = 3. The tangent plane is given by the equation
z 2 = 10(x 1) +3(y 3) z = 10x +3y +3.
12.9
2. We can either use the denition of the differential, (12.9.1), or the rules for differentials as we do here.
(a) dz = d(x
3
) + d(y
3
) = 3x
2
dx + 3y
2
dy (b) dz = (dx)e
y
2
+ x(de
y
2
). Here de
y
2
= e
y
2
dy
2
=
e
y
2
2ydy, so dz = e
y
2
dx +2xye
y
2
dy = e
y
2
(dx +2xydy).
(c) dz = d ln u, where u = x
2
y
2
. Then dz =
1
u
du =
2xdx 2ydy
x
2
y
2
.
5. d(Ue
U
) = d(x
y), and so e
U
dU +Ue
U
dU = dx
y +(x/2
y)dy. Solving for dU yields the answer.

12.11
3. Since we are asked to nd the partials of y
1
and y
2
w.r.t. x
1
only, we might as well differentiate the system
partially w.r.t. x
1
:
(i) 3
y
1
x
1
9y
2
2
y
2
x
1
= 0 (ii) 3x
2
1
+6y
2
1
y
1
x
1
y
2
x
1
= 0
Solve for the partials and see the text.
(An alternative, in particular if one needs all the partials, is to use total differentiation:
(i) 3 dx
1
+2x
2
dx
2
dy
1
9y
2
2
dy
2
= 0, (ii) 3x
2
1
dx
1
2dx
2
+6y
2
1
dy
1
dy
2
= 0
Letting dx
2
= 0 and solving for dy
1
and dy
2
leads to dy
1
= Adx
1
and dy
2
= Bdx
1
, where A = y
1
/x
1
and B = y
2
/x
1
.)
4. Differentiation with respect to M gives, (i) I
(r)r
M
= S
(Y)Y
M
, (ii) aY
M
+L
(r)r
M
= 1.
(Remember that Y and r are functions of the independent variables a and M.) Writing this as a linear
equation system on standard form, we get
S
(Y)Y
M
+I
(r)r
M
= 0
aY
M
+L
(r)r
M
= 1
Cramers rule (or use ordinary elimination) gives
Y
M
=
0 I
(r)
1 L
(r)
(Y) I
(r)
a L
(r)
=
I
(r)
S
(Y)L
(r) +aI
(r)
and r
M
=
S
(Y)
S
(Y)L
(r) +aI
(r)
5. Differentiation w.r.t. x yields, y +u
x
v +uv
x
= 0, u +xu
x
+yv
x
= 0. Solving this system for u
x
and
v
x
we get
u
x
=
u
2
y
2
yv xu
=
u
2
y
2
2yv
, v
x
=
xy uv
yv xu
=
2xy 1
2yv
where we substituted xu = yv and uv = 1 xy. Differentiating u
x
w.r.t. x nally yields
u
xx
=

2
u
x
2
=

x
u
x
=
2uu
x
2yv (u
2
y
2
)2yv
x
4y
2
v
2
=
(u
2
y
2
)(4uv 1)
4y
2
v
3
9. (a) Differentiation yields 2uvdu +u
2
dv du = 3x
2
dx +6y
2
dy, and e
ux
(udx +xdu) = vdy +ydv.
At P these equations become 3du + 4dv = 6dy and dv = 2dx dy. Hence du = 2dy (4/3)dv =
(8/3) dx +(10/3) dy. So u/y = 10/3, v/x = 2. (b) See the text.
4. X = Ng(u), where u = (N)/N). Then du/dN = [
(N)N (N)]/N
2
= (1/N)(
(N) u), and

by the product rule and the chain rule,
dX
dN
= g(u) +Ng
(u)
du
dN
= g(u) +g
(u)(
(N) u), u =
(N)
N
Differentiating g(u) +g
(u)(
(N) u) w.r.t. N gives

d
2
X
dN
2
= g
(u)
du
dN
+g
(u)
du
dN
(
(N) u) +g
(u)(
(N)
du
dN
)
=
1
N
g
_
(N)/N
__
(N) (N)/N
_
2
+g
_
(N)/N
_
(N)
5. (a) Take the natural logarithm, ln E = ln A a ln p +b ln m, and then differentiate. (b) See the text.
11. El
x
(y
2
e
x
e
1/y
) = El
x
y
2
+ El
x
e
x
+ El
x
e
1/y
= 0. Here El
x
y
2
= 2 El
x
y and El
x
e
x
= x. Moreover,
El
x
e
1/y
= El
x
e
u
, where u = 1/y, so El
x
e
u
= u El
x
(1/y) = (1/y)(El
x
1 El
x
y) = (1/y) El
x
y.
All in all, 2 El
x
y +x (1/y) El
x
y = 0, so El
x
y = xy/(1 2y). (We used the rules for elasticities in
Problem 7.7.9. If you are not comfortable with these rules, nd y
by implicit differentiation and then

use El
x
y = (x/y)y
.
16. (a) Differentiating and then gathering all terms in dp and dL on the left-hand side, yields
(i) F
(L) dp +pF
(L) dL = dw (ii) F(L) dp +(pF
(L) w) dL = Ldw +dB

Since we know that pF
(L) = w, (ii) implies that dp = (Ldw + dB)/F(L). Substituting this into (i)
and solving for dL, we obtain dL = [(F(L) LF
(L))dw F
(L)dB]/pF(L)F
(L). It follows that

p
w
=
L
F(L)
,
p
B
=
1
F(L)
,
L
w
=
F(L) LF
(L)
pF(L)F
(L)
,
L
B
=
F
(L)
pF(L)F
(L)
(b) We know that p > 0, F
(L) > 0, and F
(L) < 0. Also, F(L) = (wL + B)/p > 0. Hence, it is

clear that p/w > 0, p/B > 0, and L/B > 0.
42 13 MUL T I VARI ABL E OPTI MI ZATI ON
To nd the sign of L/w, we need the sign of F(L)LF
(L). Fromthe equations in the model, we

get F
(L) = w/p and F(L) = (wL +B)/p, so F(L) LF
(L) = B/p > 0. Therefore L/w < 0.

13 Multivariable Optimization
13.1
3. F
K
= 2(K 3) (L 6) and F
L
= 4(L 6) (K 3), so the rst-order conditions yield
2(K 3) (L 6) = 0.65, 4(L 6) (K 3) = 1.2. The solution is (K, L) = (2.8, 5.75).
4. (b) First-order conditions: P
x
= 2x +22 = 0, P
y
= 2y +18 = 0. It follows that x = 11 and y = 9.
13.2
3. Note that U = (1083y4z)yz. ThenU/y = 108z6yz4z
2
= 0andU/z = 108y3y
2
8yz =
0. Because y and z are assumed to be positive, these two equations reduce to 6y + 4z = 108 and
3y + 8z = 108, with solution y = 12 and z = 9. Theorem 13.2.1 cannot be used directly to prove
optimality. However, it can be applied to the equivalent problemof maximizing ln z. See Theorem13.6.3.)
7. Solve the constraint for z: z = 4x +2y 5. Then minimize P(x, y) = x
2
+y
2
+(4x +2y 5)
2
w.r.t. x
and y. The rst-order conditions are: P
1
= 34x + 16y 40 = 0, P
2
= 16x + 10y 20 = 0, with
solution x = 20/21, y = 10/21. Since P
11
= 34, P
12
= 16, and P
22
= 10, we see that the second-order
conditions for minimum are satised.
13.3
3. (a) V
t
(t, x) = f
t
(t, x)e
rt
rf (t, x)e
rt
= 0, V
x
(t, x) = f
x
(t, x)e
rt
1 = 0, so at the optimum,
f
t
(t
, x
) = rf (t
, x
) and f
x
(t
, x
) = e
rt
. (b) See the text and (c).

(c) V(t, x) = g(t )h(x)e
rt
x, so V
t
= h(x)(g
(t ) rg(t ))e
rt
, V
x
= g(t )h
(x)e
rt
1. Moreover,
V
t t
= h(x)(g
(t ) 2rg
(t ) + r
2
g(t ))e
rt
, V
t x
= h
(x)(g
(t ) rg(t ))e
rt
, and V
xx
= g(t )h
(x)e
rt
.
At (t
, x
), V
t x
= 0, V
xx
< 0, and V
t t
= h(x
)[g
(t
) 2rg
(t
) + r
2
g(t
)]e
rt
. Because g
(t
) =
rg(t
), we obtain V
t t
= h(x
)[g
(t
) r
2
g(t
)]e
rt
< 0. Thus (t
, x
) is a local maximum point.

(d) The rst-order conditions in (b) reduce to e
/2
= re
, so t
= 1/4r
2
, and 1/(x
+ 1) =
e
1/4r
/e
1/2r
, or x
= e
1/4r
1. The two conditions in (c) are satised. (Note that in part (c) of this
problemin the text, the second condition should be h
(x
) < 0.) Obviously, h
(x
) = (1+x
)
2
< 0.
Moreover, g
(t
) =
1
4t
1) = r
2
(1 2r)e
1/2r
< r
2
e
1/2r
, which is true when r > 0.
5. (a) We need to have 1+x
2
y > 0. When x = 0, f (0, y) = 0. For x = 0, 1+x
2
y > 0 y > 1/x
2
.
(The gure in the text shows a part of the graph of f . Note that f = 0 on the x-axis and on the y-axis.)
(b) See the text. (c) f
11
(x, y) =
2y 2x
2
y
2
(1 +x
2
y)
2
, f
12
(x, y) =
2x
(1 +x
2
y)
2
, and f
22
(x, y) =
x
4
(1 +x
2
y)
2
.
The second-order derivatives at all points of the form (0, b) are f
11
(0, b) = 2b, f
12
(0, b) = 0, and
f
22
(0, b) = 0. Since f
11
f
22
(f
12
)
2
= 0 at all the stationary points, the second-derivative test tells us
nothing about the stationary points. See the text.
13 MUL TI VARI ABL E OPTI MI ZATI ON 43
13.4
2. (a) See the text. (b) The new prot function is = bp
2
dp
2
+(a +b)p+(c +d)p(a +c)
and the price which maximizes prots is easily seen to be p =
a +c +(b +d)
2(b +d)
.
(c) If = 0, then p
=
a
2b
, q
=
c
2d
, and (p
, q
) =
a
2
4b
+
c
2
4d
. Moreover, p =
a +c
2(b +d)
with
( p) =
(a +c)
2
4(b +d)
, and (p
, q
) ( p) =
(ad bc)
2
4bd(b +d)
0. Note that the difference is 0 when
ad = bc, in which case p
= q
, so the rm wants to charge the same price in each market anyway.

3. Imposing a tax of t per unit sold in market area 1 means that the new prot function is (Q
1
, Q
2
) =
(Q
1
, Q
2
) t Q
1
. The optimal choice of production in market area 1 is then

Q
1
= (a
1
t )/2b
1
(see the text), and the tax revenue is T (t ) = t (a
1
t )/2b
1
= [t (a
1
) t
2
]/2b
1
. This quadratic
polynomial has maximum when T
(t ) = 0, so t =
1
2
(a
1
).
4. (a) Let (x
0
, y
0
) = (0, 11.29), (x
1
, y
1
) = (1, 11.40), (x
2
, y
2
) = (2, 11.49), and (x
3
, y
3
) = (3, 11.61),
so that x
0
corresponds to 1970, etc. (The numbers y
t
are approximate, as are most subsequent results.)
We nd that
x
=
1
4
(0 + 1 + 2 + 3) = 1.5,
y
=
1
4
(11.29 + 11.40 + 11.49 + 11.61) = 11.45, and
xx
=
1
4
[(0 1.5)
2
+(1 1.5)
2
+(2 1.5)
2
+(3 1.5)
2
] = 1.25. Moreover, we nd
xy
= 0.13125,
and so a =
xy
/
xx
= 0.105 and

b =
y
a
x
11.45 0.105 1.5 = 11.29.
(b) With z
0
= ln 274, z
1
= ln 307, z
2
= ln 436, and z
3
= ln 524, we have (x
0
, z
0
) = (0, 5.61),
(x
1
, z
1
) = (1, 5.73), (x
2
, z
2
) = (2, 6.08), and (x
3
, z
3
) = (3, 6.26). As before,
x
= 1.5 and
xx
= 1.25.
Moreover,
z
=
1
4
(5.61 +5.73 +6.08 +6.26) = 5.92 and
xz
0.2875. Hence c =
xz
/
xx
= 0.23,
d =
z
c
x
= 5.92 0.23 1.5 = 5.575.
(c) With ln (GNP) = 0.105x + 11.25, GNP = e
11.25
e
0.105x
= 80017e
0.105x
. Likewise, FA = 256e
0.23x
.
The requirement that FA = 0.01 GNP implies that e
0.23x0.105x
= 80017/25600, and so 0.125x =
ln(80017/25600). Thus x = ln(80017/25600)/0.125 = 9.12. Since x = 0 corresponds to 1970, the
goal would have been reached in 1979.
5. (a) See the text. (b) FirmAs prot is now
A
(p) = px5x = p(295p+4q)529+5p4q =
34p 5p
2
+ 4pq 4q 34, with q xed. This quadratic polynomial is maximized at p = p
A
(q) =
1
5
(2q + 17). Likewise, rm Bs prot is now
B
(q) = qy 3 2y = 28q 6q
2
+ 4pq 8p 35,
with p xed. This quadratic polynomial is maximized at q = q
B
(p) =
1
3
(p + 7). For (c) and (d) see
the text.
y
x
4
4
(3, 3)
Figure SM13.5.2
13.5
2. (a) The continuous function f is dened on a closed, bounded set S (see Fig. SM13.5.2), so the extreme
value theoremensures that f attains both a maximumand a minimumover S. Stationary points are where
(i) f
1
(x, y) = 3x
2
9y = 0 and (ii) f
2
(x, y) = 3y
2
9x = 0. From (i), y =
1
3
x
2
, which inserted
into (ii) yields
1
3
x(x
3
27) = 0. The only solutions are x = 0 and x = 3. Thus the only stationary
point in the interior of S is (x, y) = (3, 3). We proceed by examining the behaviour of f (x, y) along the
boundary of S, i.e. along the four edges of S.
(I) y = 0, x [0, 4]. Then f (x, 0) = x
3
+27, which has minimum at x = 0, and maximum at x = 4.
(II) x = 4, y [0, 4]. Then f (4, y) = y
3
36y +91. The function g(y) = y
3
36y +91, y [0, 4]
has g
(y) = 3y
2
36 = 0 at y =
12. Possible extreme points along (II) are therefore (4, 0), (4,
12),
and (4, 4).
(III) y = 4, x [0, 4]. Then f (x, 4) = x
3
36x +91, and as in (II) we see that possible extreme points
are (0, 4), (
12, 4), and (4, 4).

(IV) x = 0, y [0, 4]. As in case (I) we obtain the possible extreme points (0, 0) and (0, 4).
This results in six candidates, and f (3, 3) = 0, f (0, 0) = 27, f (4, 0) = f (0, 4) = 91, f (4,
12) =
f (
12, 4) = 91 24
12 7.7, f (0, 0) = 27. The conclusion follows.

(b) The constraint set S =
_
(x, y) : x
2
+ y
2
1
_
are all points that lie on or inside a circle around
the origin with radius 1. This is a closed and bounded set, and f (x, y) = x
2
+ 2y
2
x is continuous.
Therefore the extreme value theorem ensures that f attains both a maximum and a minimum over S.
Stationary points for f where f
x
(x, y) = 2x 1 = 0 and f
y
(x, y) = 4y = 0. So the only stationary
point for f is (x
1
, y
1
) = (1/2, 0), which is an interior point of S.
An extreme point that does not lie in the interior of S must lie on the boundary of S, that is, on the
circle x
2
+y
2
= 1. Along this circle we have y
2
= 1 x
2
, and therefore
f (x, y) = x
2
+2y
2
x = x
2
+2(1 x
2
) x = 2 x x
2
where x runs through the interval [1, 1]. (It is a common error to ignore this restriction.) The function
g(x) = 2 x x
2
has one stationary point in the interior of [1, 1], namely x = 1/2, so any extreme
values of g(x) must occur either for this value of x or at one the endpoints 1 of the interval [1, 1].
Any extreme points for f (x, y) on the boundary of S must therefore be among the points
(x
2
, y
2
) = (
1
2
,
1
2
3), (x
3
, y
3
) = (
1
2
,
1
2
3), (x
4
, y
4
) = (1, 0), (x
5
, y
5
) = (1, 0)
Now, f (
1
2
, 0) =
1
4
, f (
1
2
,
1
2
3) =
9
4
, f (1, 0) = 0, and f (1, 0) = 2. The conclusion follows.
3. The set S is shown in Fig. A13.5.3 in the book. It is clearly closed and bounded, so the continuous
function f has a maximum in S. The stationary points are where f/x = 9 12(x + y) = 0 and
f/y = 8 12(x +y) = 0. But 12(x +y) = 9 and 12(x +y) = 8 give a contradiction. Hence, there
are no stationary points at all. The maximum value of f must therefore occur on the boundary, which
consists of ve parts. Either the maximum value occurs at one of the ve corners or extreme points of
the boundary, or else at an interior point of one of ve straight edges. The function values at the ve
corners are f (0, 0) = 0, f (5, 0) = 105, f (5, 3) = 315, f (4, 3) = 234, and f (0, 1) = 2.
We proceed to examine the behaviour of f at interior points of each of the ve edges in Fig. A13.5.3.
(I) y = 0, x (0, 5). The behaviour of f is determined by the function g
1
(x) = f (x, 0) = 9x 6x
2
for
x (0, 5). If this function of one variable has a maximum in (0, 5), it must occur at a stationary point
where g
1
(x) = 9 12x = 0, and so at x = 3/4. We nd that g
1
(3/4) = f (3/4, 0) = 27/8.
(II), x = 5, y (0, 3). Dene g
2
(y) = f (5, y) = 45 + 8y 6(5 + y)
2
for y (0, 3). Here g
2
(y) =
52 12y, which is negative throughout (0, 3), so there are no stationary points on this edge.
(III) y = 3, x (4, 5). Dene g
3
(x) = f (x, 3) = 9x + 24 6(x + 3)
2
for x (4, 5). Here g
3
(x) =
27 12x, which is negative throughout (4, 5), so there are no stationary points on this edge either.
(IV) x + 2y = 2, or y = x/2 + 1, with x (0, 4). Dene the function g
4
(x) = f (x, x/2 + 1) =
27x
2
/2 5x +2 for x (0, 4). Here g
4
(x) = 27x 5, which is negative in (0, 4), so there are no
stationary points here.
(V) x = 0, y (0, 1). Dene g
5
(y) = f (0, y) = 8y 6y
2
. Then g
5
(y) = 8 12y = 0 at y = 2/3,
with g
5
(2/3) = f (0, 2/3) = 8/3.
After comparing the values of f at the ve corners of the boundary and at the points found on the
edges labeled (I) and (V), we conclude that the maximumvalue of f is 27/8, which is achieved at (3/4, 0).
5. (a) f
1
(x, y) = e
x
(1 x)(y 4)y, f
2
(x, y) = 2xe
x
(y 2). It follows that the stationary points are
(1, 2), (0, 0) and (0, 4). (Make sure you understand why!) Moreover, f
11
(x, y) = e
x
(x 2)(y
2
4y),
f
12
(x, y) = e
x
(1 x)(2y 4), and f
22
= 2xe
x
. Classication of the stationary points:
(x, y) A B C AC B
2
Type of point
(1, 2) 4e
1
> 0 0 2e
1
8e
2
> 0 Loc. min. point
(0, 0) 0 4 0 16 < 0 Saddle point
(0, 4) 0 4 0 16 < 0 Saddle point
(b) We see that f (1, y) = e
1
(y
2
4y) as y . This shows that f has no global maximum
point. Since f (1, y) = e(y
2
4y) as y , f has no global minimum point either.
y
x
y
x
(1, 2) (II) (IV)
(I)
(III)
(c) The set S is obviously bounded. The boundary of S consists of
the four edges of the rectangle, and all points on these line segments
belong to S. Hence S is closed. Since f is continuous, the extreme
value theoremtells us that f has global maximumandminimumpoints
in S. These global extreme points must be either stationary points if f
in the interior of S, or points on the boundary of S. The only stationary
point of f in the interior of S is (1, 2). The function value at this point
is f (1, 2) = 4e
1
1.4715.
The four edges are most easily investigated separately:
(i) Along (I), y = 0 and f (x, y) = f (x, 0) is identically 0.
(ii) Along (II), x = 5 and f (x, y) = 5e
5
(y
2
4y), which has its least value for y = 2 and its greatest
value for y = 0 and for y = 4. (Note that y [0, 4] for all points (x, y) on the line segment (II).)
The values are f (5, 2) = 20e
5
0.1348 and f (5, 0) = f (5, 4) = 0.
(iii) On edge (III), y = 4 and f (x, y) = f (x, 4) = 0.
(iv) Finally, along (IV), x = 0 and f (x, y) = f (0, y) = 0.
Collecting all these results, we see that f attains its least value (on S) at the point (1, 2) and its
greatest value (namely 0) at all points of the line segments (I), (III) and (IV).
(d) y
=
f
1
(x, y)
f
2
(x, y)
=
e
x
(1 x)(y 4)y
2xe
x
(y 2)
=
(x 1)(y 4)y
2x(y 2)
= 0 when x = 1.
13.6
1. (a) f
x
(x, y, z) = 22x = 0, f
y
(x, y, z) = 102y = 0, f
x
(x, y, z) = 2z = 0. The conclusionfollows.
(b) f
x
(x, y, z) = 2x 2y 2z = 0, f
y
(x, y, z) = 4y 2x = 0, f
z
(x, y, z) = 6z 2x = 0.
From the two last equations we get y =
1
2
x and z =
1
3
x. Inserting this into the rst equation we get
2x +x +
2
3
x = 0, and thus x = 0, and then y = z = 0.
4. To calculate f
x
is routine. To differentiate f w.r.t. y and z we use (9.3.7) and (9.3.6). The derivative of
_
z
y
e
t
2
dt w.r.t. y, keeping z constant is according to (9.3.7), e
y
2
. The derivative of
_
z
y
e
t
2
dt w.r.t. z,
keeping y constant is according to (9.3.6), e
z
2
. Thus f
y
= 2 e
y
2
and f
z
= 3 + e
z
2
. Since each of
the three partials depends only on one variable and is 0 for two different values of that variable, there are
eight stationary point. See the text.
13.7
2. (a) First-order conditions:
K
=
2
3
pK
1/3
r = 0,
L
=
1
2
pL
1/2
w = 0,
T
=
1
3
pT
2/3
q = 0.
Thus, K
1/3
= 3r/2p, L
1/2
= 2w/p, and T
2/3
= 3q/p. Raising each side of K
1/3
= 3r/2p to the
power of 3 yields, K = (3r/2p)
3
= (2p/3r)
3
= (8/27)p
3
r
3
. In a similar way we nd L, and T .
(b) See the text.
4. Differentiating pF
K
(K
, L
) = r using the product rule gives dp F
K
(K
, L
) + pd(F
K
(K
, L
)) =
dr. Moreover, d(F
K
(K
, L
)) = F
KK
(K
, L
) dK
+ F
KL
(K
, L
) dL
. (To see why, note that

dg(K
, L
) = g
K
(K
, L
) dK
+ g
L
(K
, L
) dL
. Then let g = F
K
.) This explains the rst dis-
played equation (replacing dK by dK
and dL by dL
). The second is derived in the same way.

(b) Rearrange the equation systemby moving the differentials of the exogenous variables to the right-hand
side, suppressing the fact that the partials are evaluated at (K
, L
):
pF
KK
dK
+pF
KL
dL
= dr F
K
dp
pF
LK
dK
+pF
LL
dL
= dw F
L
dp
We use Cramers rule to express the differentials dK
and dL
in terms of dp, dr, and dw. Putting

= F
KK
F
LL
F
KL
F
LK
= F
KK
F
LL
(F
KL
)
2
, we get
dK
=
1
p
2
dr F
K
dp pF
KL
dw F
L
dp pF
LL
=
F
K
F
LL
+F
L
F
KL
p
dp +
F
LL
p
dr +
F
KL
p
dw
In the same way
dL
=
1
p
2
pF
KK
dr F
K
dp
pF
LK
dw F
L
dp
=
F
L
F
KK
+F
K
F
LK
p
dp +
F
LK
p
dr +
F
KK
p
dw
We can now read off the required partials. (Note that there is an error in the expression for K
/p in
the answer to this problem in the text. In the numerator, replace F
KK
by F
LL
. (c) See the text. (Recall
that F
LL
< 0 follows from () in Example 13.3.3.)
5. (a) (i) R
1
(x
1
, x
2
) +s = C
1
(x
1
, x
2
) (marginal revenue plus subsidy equal marginal cost)
(ii) R
2
(x
1
, x
2
) = C
2
(x
1
, x
2
) +t = 0 (marginal revenue equals marginal cost plus tax). For (b) see the
text. (c) Taking the total differentials of (i) and (ii) yields
(R
11
C
11
)dx
1
+(R
12
C
12
)dx
2
= ds, (R
21
C
21
)dx
1
+(R
22
C
22
)dx
2
= dt
Solving for dx
1
and dx
2
yields, after rearranging,
dx
1
=
(R
22
C
22
)ds (R
12
C
12
)dt
D
, dx
2
=
(R
21
C
21
)ds +(R
11
C
11
)dt
D
From this we nd that the partial derivatives are
x
1
s
=
R
22
+C
22
D
> 0,
x
1
t
=
R
12
+C
12
D
> 0,
x
2
s
=
R
21
C
21
D
< 0,
x
2
t
=
R
11
C
11
D
< 0
where the signs follow from the assumptions in the problem and the fact that D > 0 from (b). Note
that these signs accord with economic intuition. For example, if the tax on good 2 increases, then the
production of good 1 increases, while the production of good 2 decreases.
(d) Follows from the expressions in (c) because R
12
= R
21
and C
12
= C
21
.
2. (a) The prot function is (Q
1
, Q
2
) = 120Q
1
+ 90Q
2
0.1Q
2
1
0.1Q
1
Q
2
0.1Q
2
2
. First-order
conditions for maximal prot are:
1
(Q
1
, Q
2
) = 120 0.2Q
1
0.1Q
2
= 0 and
2
(Q
1
, Q
2
) =
90 0.1Q
1
0.2Q
2
= 0. We nd (Q
1
, Q
2
) = (500, 200). Moreover,
11
(Q
1
, Q
2
) = 0.2 0,
12
(Q
1
, Q
2
) = 0.1, and
22
(Q
1
, Q
2
) = 0.2 0. Since also
11
22
(
12
)
2
= 0.03 0,
(500, 200) maximizes prots.
(b) The prot function is now, (Q
1
, Q
2
) = P
1
Q
1
+90Q
2
0.1Q
2
1
0.1Q
1
Q
2
0.1Q
2
2
. First-order
conditions for maximal prot:
1
= P
1
0.2Q
1
0.1Q
2
= 0,
2
= 900.1Q
1
0.2Q
2
= 0. If we need
to have Q
1
= 400, the rst-order conditions reduce to P
1
80 0.1Q
2
= 0 and 90 40 0.2Q
2
= 0.
It follows that P
1
= 105.
3. (a) Stationary points where P
1
(x, y) = 0.2x 0.2y +47 = 0 and P
2
(x, y) = 0.2x 0.4y +48y = 0.
It follows that x = 230 and y = 5. Moreover, P
11
= 0.2 0, P
12
= 0.2, and P
22
= 0.4 0.
Since also P
11
P
22
(P
12
)
2
= 0.04 0, (230, 5) maximizes prots. (b) With x + y = 200, and so
y = 200 x, the new prot function is (x) = f (x, 200 x) = 0.1x
2
+39x +1000. This function
is easily seen to have maximum at x = 195. Then y = 200 195 = 5.
4. (a) Stationary points:(i) f
1
(x, y) = 3x
2
2xy = x(3x 2y) = 0, (ii) f
2
(x, y) = x
2
+2y = 0. From
(i), x = 0 or 3x = 2y. If x = 0, then (ii) gives y = 0. If 3x = 2y, then (ii) gives 3x = x
2
, and so x = 0
or x = 3. If x = 3, then (ii) gives y = x
2
/2 = 9/2. So the stationary points are (0, 0) and (3, 9/2).
(b) (i) f
1
(x, y) = ye
4x
2
5xy+y
2
(8x
2
5xy +1) = 0, (ii) f
2
(x, y) = xe
4x
2
5xy+y
2
(2y
2
5xy +1) = 0.
If y = 0, then (i) is satised and (ii) is only satised when x = 0. If x = 0, then (ii) is satised and
(i) is only satised when y = 0. Thus, in addition to (0, 0), any other stationary point must satisfy
8x
2
5xy + 1 = 0 and 2y
2
5xy + 1 = 0. Subtracting the second from the rst yields 8x
2
= 2y
2
,
or y = 2x. Inserting y = 2x into 8x
2
5xy +1 = 0 yields 18x
2
+1 = 0, which has no solutions.
Inserting y = 2x into 8x
2
5xy + 1 = 0 yields x
2
=
1
2
, and so x =
1
2
2. We conclude that the

stationary points are: (0, 0) and (
1
2
2,
2), (
1
2
2,
2)
5. (a) The rst-order conditions for (K
, L
, T
) to maximize are:
K
= pa/K
r = 0,
L
=
pb/L
w = 0,
T
= pc/T
q = 0. Hence, K
= ap/r, L
= bp/w, T
= cp/q.
(b)
= pa ln(ap) pa ln r +pb ln(bp/w) +pc ln(cp/q) ap bp cp = pa ln r plus terms that

do not depend on r. So
/r = pa/r = K
. (c) See the text.

48 14 CONSTRAI NED OPTI MI ZATI ON
7. (a) f
1
(x, y) = 2x y 3x
2
, f
2
(x, y) = 2y x, f
11
(x, y) = 26x, f
12
(x, y) = 1, f
22
(x, y) = 2.
Stationary points where 2x y 3x
2
= 0 and 2y x = 0. The last equation yields y = x/2, which
inserted into the second equation gives x(
5
6
x) = 0. It follows that there are two stationary points,
(x
1
, y
1
) = (0, 0) and (x
2
, y
2
) = (5/6, 5/12). These points are classied in the following table:
(x, y) A B C AC B
2
Type of point
(0, 0) 2 1 2 5 Saddle point
(
5
6
,
5
12
) 3 1 2 5 Lok. max. point
(b) f is concave in the domain where f
11
0, f
22
0, and f
11
f
22
(f
12
)
2
0, i.e. where 2 6x 0,
2 0, and (2 6x)(2) (1)
2
0. These conditions are equivalent to x 1/3 and x 5/12.
In particular, one must x 5/12. Since 5/12 > 1/3, f is concave in the set S consisting of all (x, y)
where x 5/12.
(c) The stationary point (x
2
, y
2
) = (5/6, 5/12) found in (a) does belong to S. Since f concave in S,
this is a (global) maximum point for f in S. f
max
= 125/432.
8. (a) We nd f
1
(x, y) = x 1 + ay, f
2
(x, y) = a(x 1) y
2
+ 2a
2
y. Stationary points require that
x 1 = ay and a(x 1) = y
2
2a
2
y. These two equations yield a
2
y = y
2
2a
2
y, and so a
2
y = y
2
.
Hence y = 0 or y = a
2
. Since x = 1 ay, the stationary points are (1, 0) and (1 a
3
, a
2
). (Since we
were asked only to show that (1 a
3
, a
2
) is a stationary point, it would sufce to verify that it makes
both partials equal to 0.) (b) Note that the partial derivative of f w.r.t. a, keeping x and y constant is
f/a = y(x 1) + 2ay
2
. Evaluated at x = 1 a
3
, y = a
2
, this partial derivative is also a
5
, thus
conrming the envelope theorem. (c) See the text.
9. For (a)(c), see the text. (d)
2
/p
2
= 2b,
2
/q
2
= 2 , and
2
/py = + c. The direct
partials of order 2 are negative and = (
2
/p
2
)(
2
/q
2
) (
2
/pq)
2
= 4 b ( + c)
2
, so
the conclusion follows.
14 Constrained Optimization
14.1
4. (a) With L(x, y) = x
2
+ y
2
(x + 2y 4), the rst-order conditions are L
1
= 2x = 0 and
L
2
= 2y 2 = 0. From these equations we get 2x = y, which inserted into the constraint gives
x +4x = 4. So x = 4/5 and y = 2x = 8/5, with = 2x = 8/5.
(b) The same method as in (a) gives 2x = 0 and 4y = 0, so x = 2y. From the constraint we
get x = 8 and y = 4, with = 16. (c) The rst-order conditions imply that 2x +3y = = 3x +2y,
which implies x = y. So the solution is (x, y) = (50, 50) with = 250.
5. The budget constraint is 2x +4y = 1000, so with L(x, y) = 100xy +x +2y (2x +4y 1000), the
rst-order conditions are L
1
= 100y +12 = 0 and L
2
= 100x +24 = 0. From these equations,
by eliminating , we get x = 2y, which inserted into the constraint gives 2x +2x = 1000. So x = 250
and y = 125.
7. The problem is: max 0.1x
2
0.2xy 0.2y
2
+ 47x + 48y 600 subject to x + y = 200.
With L(x, y) = 0.1x
2
0.2xy 0.2y
2
+47x +48y 600(x +y 200), the rst-order conditions
14 CONSTRAI NED OPTI MI ZATI ON 49
are L
1
= 0.2x 0.2y + 47 = 0 and L
2
= 0.2x 0.4y + 48 = 0. Eliminating yields
y = 5, and then the budget constraint gives x = 195.
9. (a) With L(x, y) = 100 e
x
e
y
(px +qy m), L
x
= L
y
= 0 when e
x
= p and e
y
= q.
Hence, x = ln(p) = ln ln p, y = ln ln q. Inserting these expressions for x and y
into the constraint, then solving for ln , yields ln = (m + p ln p + q ln q)/(p + q). Therefore
x(p, q, m) = [m+q ln(q/p)]/(p +q), y(p, q, m) = [m+p ln(p/q)]/(p +q).
(b) x(tp, t q, t m) = [t m + t q ln(t q/tp)]/(tp + t q) = x(p, q, m), so x is homogeneous of degree 0.
In the same way we see that y(p, q, m) is homogeneous of degree 0.
14.2
3. (a) Solving x+2y = a for y yields y =
1
2
a
1
2
x, and then x
2
+y
2
= x
2
+(
1
2
a
1
2
x)
2
=
5
4
x
2
1
2
ax+
1
4
a
2
.
This quadratic function certainly has a minimum at x = a/5. (b) L(x, y) = x
2
+y
2
(x +2y a).
The necessary conditions are L
1
= 2x = 0, L
2
= 2y 2 = 0, implying that 2x = y. From the
constraint, x = a/5 and then y = 2a/5, = 2a/5.
The value function f
(a) = (a/5)
2
+ (2a/5)
2
= a
2
/5, so df
(a)/da = 2a/5, which is also the

value of the Lagrangian multiplier. Equation (2) is conrmed. (c) See the text.
4. (a) With L(x, y) =
x +y (x +4y 100), the rst-order conditions for (x
, y
) to solve the problem

are: (i) L/x = 1/2
= 0 (ii) L/y = 1 4 = 0. From (ii), = 1/4, which inserted into

(i) yields
= 2, so x
= 4. Then y
= 25
1
4
4 = 24, and maximal utility is U
+y
= 26.
(b) Denote the new optimal values of x and y by x and y. If 100 is changed to 101, still = 1/4 and
x = 4. The constraint now gives 4 +4 y = 101, so that y = 97/4 = 24.25, with

U =
x + y = 26.25.
The increase in the maximum utility as 100 is increased to 101, is thus

U U
= 0.25 = . (In general,

the increase in utility is approximately equal to the value of the Lagrange multiplier.)
(c) The necessary conditions for optimality are nowL/x = 1/2
p = 0, L/y = 1q = 0.
Proceeding in the same way as in (a), we nd = 1/q,

x
= q/2p, and so x
= q
2
/4p
2
, with
y
= m/q q/4p. (Note that y
> 0 m > q
2
/4p.) (If we solve the constraint for y, the utility
function is u(x) =

x + (m px)/q. We see that u
(x) = 1/2
x p/q = 0 for x
= q
2
/4p
2
and
u
(x) = (1/4)x
3/2
< 0 when x > 0. So we have found the maximum.)
5. (a) px
= pa+/ and qy
= qb+/ give m = px
+qy
= pa+qb+(+)/ = pa+qb+1/,
so 1/ = m(pa +qb). The expressions given in () are now easily established. (If we think of a and
b as kind of existence minimum of the two goods, the assumption pa +qb < mmeans that the consumer
can afford to buy (a, b).) (b) With the U
given in the answer, since + = 1,

U
m
=

m(pa +qb)
+

m(pa +qb)
=
1
m(pa +qb)
= > 0. Moreover,
U
p
=
a
m(pa +qb)
p
+
a
m(pa +qb)
=
a
m(pa +qb)
p
= a
p
, and
U
m
x
=
(a +

p
) = a

p
, so
U
p
=
U
m
x
. The last equality is shown in the same way.

6. f (x, T ) = x
_
T
0
[t
3
+ (T
2
+ T 1)t
2
+ (T T
3
)t ] dt = x
T
0
[
1
4
t
4
+ (T
2
+ T 1)
1
3
t
3
+
(T T
3
)
1
2
t
2
] =
1
6
xT
5
+
1
12
xT
4
+
1
6
xT
3
. In the same way, g(x, T ) =
1
6
xT
3
. The solution of
() is x = 384
3
M, T = 1/4, with f
(M) = M + M/16. (The easiest way to solve the problem

is to note that because
1
6
xT
3
= M, the problem reduces to that of maximizing M +
1
2
MT MT
2
for T 0. For T = 0, the expression is equal to M, and its maximum is attained for T = 1/4.)
Alternatively, eliminating the Lagrange multiplier fromthe rst-order conditions f
1
= g
1
and f
2
= g
2
,
we eventually obtain T = 1/4. The Lagrange multiplier is = 1 +1/16. Clearly, f
(M)/M = ,
which conrms (2).
14.3
1. (a) With L(x, y) = 3xy (x
2
+y
2
8), the rst-order conditions are L
1
= 3y 2x = 0 and L
2
=
3x 2y = 0. Since (0, 0) does not satisfy the constraint, from these equations we get x
2
= y
2
. Inserted
into the constraint this yields x
2
= 4, and so x = 2, and the solution candidates are: (2, 2), (2, 2),
(2, 2), (2, 2). Here f (2, 2) = f (2, 2) = 12 and f (2, 2) = f (2, 2) = 12. So (2, 2) and
(2, 2) solves the maximization problem, and (2, 2) and (2, 2) solves the minimization problem,
because the extreme value theorem ensures that solutions exist. (f is continuous and the constraint curve
is a closed bounded set (a circle).)
(b) With L = x + y (x
2
+ 3xy + 3y
2
3), the rst-order conditions are 1 2x 3y = 0
and 1 3x 6y = 0. From these equations we get 2x + 3y = 3x + 6y, or (3y + x) = 0.
Here = 0 is impossible, so x = 3y. Inserted into the constraint we have (3, 1) and (3, 1) as the
only possible solutions of the maximization and minimization problems, respectively. The extreme value
theorem ensures that solutions exist. (The objective function is continuous and the constraint curve is a
closed bounded set (an ellipse, see (5.5.5)).
2. (a) With L = x
2
+y
2
2x+1(x
2
+4y
2
16), the rst-order conditions are (i) 2x22x = 0 and
(ii) 2y 8y = 0. Equation (i) yields = 1 1/x (why can we be sure that x = 0?), and equation (ii)
shows that y = 0 or = 1/4. If y = 0, then x
2
= 164y
2
= 16, so x = 4, and then gives = 11/4.
If y = 0, then = 1/4 and (i) gives 2x 2 x/2 = 0, so x = 4/3. The constraint x
2
+ 4y
2
= 16
now yields 4y
2
= 16 16/9 = 128/9, so y =
32/9 = 4
2/3. Thus, there are four solution

candidates: (i) (x, y, ) = (4, 0, 3/4), (ii) (x, y, ) = (4, 0, 5/4), (iii) (x, y, ) = (4/3, 4
2/3, 1/4),
and (iv) (x, y, ) = (4/3, 4
2/3, 1/4). Of these, the second is the maximum point (while (iii) and (iv)
are the minimum points).
(b) The Lagrangian is L = ln (2 +x
2
)+y
2
(x
2
+2y2). Hence, the necessary rst-order conditions
for (x, y) to be a minimum point are (i) L/x = 2x/(2 +x
2
) 2x = 0 (ii) L/y = 2y 2 = 0,
(iii) x
2
+2y = 2. From (i) we get x
_
1/(2 +x
2
)
_
= 0, so x = 0 or = 1/(2 +x
2
).
(I) If x = 0, then (iii) gives y = 1, so (x
1
, y
1
) = (0, 1) is a candidate.
(II) If x = 0, then y = = 1/(2 +x
2
) , where we used (ii). Inserting y = 1/(2 +x
2
) into (iii) gives
x
2
+2/(2 +x
2
) = 2 2x
2
+x
4
+2 = 4 +2x
2
x
4
= 2 x =
4
2.
From (iii), y = 1
1
2
x
2
= 1
1
2
2. Thus, (x
2
, y
2
) = (
4
2, 1
1
2
2) and (x
3
, y
3
) = (
4
2, 1
1
2
2)
are candidates. Now f (x
1
, y
1
) = f (0, 1) = ln 2 + 1 1.69, f (x
2
, y
2
) = f (x
3
, y
3
) = ln (2 +
2 ) +
(1
1
2
2 )
2
= ln (2 +
2 ) +
3
2

2 1.31. Hence, the minimum points for f (x, y) (subject to

x
2
+2y = 2) are (x
2
, y
2
) and (x
3
, y
3
).
4. (a) With L = 24x x
2
+ 16y 2y
2
(x
2
+ 2y
2
44), the rst-order conditions are (i) L
1
=
242x 2x = 0 and (ii) L
2
= 164y 4y = 0. From(i) x(1+) = 12 and from(ii) y(1+) = 4.
Eliminating from (i) and (ii) we get x = 3y. Inserted into the constraint, 11y
2
= 44, so y = 2, and
then x = 6. So there are two candidates, (x, y) = (6, 2) and (6, 2), with = 1. Computing the
objective function at these two points, the only possible solution is (x, y) = (6, 2). Since the objective
function is continuous and the constraint curve is closed and bounded (an ellipse), the extreme value
theorem assures us that the optimum is found. (b) According to (14.2.3) the approximate change is
1 = 1.
14.4
4. The minimum is 1 at (x, y) = (1, 0). Actually, this problem is quite tricky. The Lagrangian can be
written in the form L = (x +2)
2
+(1 )y
2
+x(x +1)
2
. The only stationary point which satises
the constraint is (0, 0), with = 4, and with f (0, 0) = 4. (In fact, L
2
= 0 only if = 1 or y = 0. For
= 1, L
1
= 3(x +1)
2
+2 > 0 for all x. For y = 0, the constraint gives x = 0 or x = 1. But x = 1
gives L
1
= 2, so x = 0 is necessary for a stationary point.) Yet at (1, 0) both g
1
(1, 0) and g
2
(1, 0)
are 0, and the Lagrange multiplier method fails. The given problem is to minimize (the square of) the
distance from (2, 0) to a point on the graph of g(x, y) = 0. But the graph consists of the isolated point
(1, 0) and a smooth curve, as illustrated in Fig. SM14.4.4.
y
6
5
4
3
2
1
1
2
3
4
5
6
x
321 1 2 3 4
(x, y)
y
2
= x(x +1)
2
d
Figure SM14.4.4
14.5
4. U
11
(x, y) = a(a 1)x
a2
0, U
22
(x, y) = a(a 1)y
a2
0, and U
12
(x, y) = 0, so U is concave.
With L = x
a
+ y
a
(px + qy m), the rst-order conditions are L
1
= ax
a1
p = 0 and
L
2
= ay
a1
q = 0. Hence, ax
a1
= p and ay
a1
q. Eliminating we get (x/y)
a1
= p/q,
and so x = y(p/q)
1/(a1)
. Inserted into the budget constraint we get
px + qy = py(p/q)
1/(a1)
+ qy = yp
a/(a1)
q
1/(a1)
+ qy = yq
1/(a1)
[p
a/(a1)
+ q
a/(a1)
] = m,
and so y = mq
1/(a1)
/[p
a/(a1)
+q
a/(a1)
]. A similar expression is obtained for x.
14.6
1. (a) L
x
= 2x = 0, L
y
= 2y = 0, L
z
= 2z = 0. It follows that x = y = z, etc. See the text.
3. (a) The Lagrangianis L = ln x+ ln y+(1) ln(L)(px+qywm), whichis stationary
when: (i) L
x
= /x
p = 0; (ii) L
y
= /y
q = 0; (iii) L
= (1)/(L
) +w = 0.
From (i) and (ii), qy
= (/)px
, while (i) and (iii) yield
w = wL[(1 )/]px
. Insertion
into the budget constraint and solving for x
yields the answer in the text. The corresponding values for

y
and
follows. The assumption m [(1 )/]wL ensures that
0. (b) See the text.

6. The Lagrangian is L = x +y (x
2
+2y
2
+z
2
1) (x +y +z 1), which is stationary when:
(i) L
x
= 1 2x = 0; (ii) L
y
= 1 4y = 0; (iii) L
z
= 2z = 0. From (i) and
(ii), (x 2y) = 0. If = 0, then (ii) and (iii) yield the contradiction = 1 and = 0. Therefore
x = 2y instead. Substituting this value for x into the constraints gives 6y
2
+ z
2
= 1, 3y + z = 1.
Thus z = 1 3y and 1 = 6y
2
+ (1 3y)
2
= 15y
2
6y + 1. Hence y = 0 or y = 2/5, implying
that x = 0 or 4/5, and that z = 1 or 1/5. The only two solution candidates are (x, y, z) = (0, 0, 1)
with = 1/2, = 1, and (x, y, z) = (4/5, 2/5, 1/5) with = 1/2, = 1/5. Because x + y is
0 at (0, 0, 1) and 6/5 at (4/5, 2/5, 1/5), these are respectively the minimum and the maximum. (The
constraints determine geometrically the curve which is the intersection of an ellipsoid (see Fig. 11.4.2)
and a plane. The continuous function x + y does attain a maximum and a minimum over this closed
bounded set.)
7. (a) With a CobbDouglas utility function, U
k
(x) =
k
U(x)/x
k
, so from (6) (with j = 1), we have
p
k
/p
1
= U
k
(x)/U
1
(x) =
k
x
1
/
1
x
k
. Thus p
k
x
k
= (a
k
/a
1
)p
1
x
1
. Inserted into the budget constraint, we
have p
1
x
1
+ (a
2
/a
1
)p
1
x
1
+ + (a
n
/a
1
)p
1
x
1
= m, which implies that p
1
x
1
= a
1
m/(a
1
+ + a
n
).
Similarly, p
k
x
k
= a
k
m/(a
1
+ +a
n
) for k = 1, . . . , n.
(b) From (6) (with j = 1), we get x
a1
k
/x
a1
1
= p
k
/p
1
and so x
k
/x
1
= (p
k
/p
1
)
1/(1a)
or p
k
x
k
/p
1
x
1
=
(p
k
/p
1
)
11/(1a)
= (p
k
/p
1
)
a/(1a)
. The budget constraint gives
p
1
x
1
_
1 +(p
2
/p
1
)
a/(1a)
+ +(p
n
/p
1
)
a/(1a)
_
= m. So p
1
x
1
= mp
a/(1a)
1
_
n
i=1
p
a/(1a)
i
.
Arguing similarly for each k, one has x
k
= mp
1/(1a)
k
_
n
i=1
p
a/(1a)
i
for k = 1, . . . , n.
14.7
2. Here L = x +4y +3z (x
2
+2y
2
+
1
3
z
2
b). So necessary conditions are:
(i) L
1
= 1 2x = 0; (ii) L
2
= 4 4y = 0; (iii) L
3
= 3
2
3
z = 0. It follows that = 0,
and so x = 1/2, y = 1/, z = 9/2. Inserting these values into the constraint yields
2
= 9/b, so
= 3/
b. The value of the objective function is x +4y +3z = 18/, so = 3/
b determines the
minimum point. This is (x, y, z) = (a, 2a, 9a), where a =
b/6. See the text.

4. With L = x
2
+y
2
+z (x
2
+2y
2
+4z
2
1), necessary conditions are: (i) L/x = 2x 2x = 0,
(ii) L/y = 2y 4y = 0, (iii) L/z = 1 8z = 0. From (i), 2x(1 ) = 0, so there are two
possibilities: x = 0 or = 1.
(A): x = 0. From (ii), 2y(1 2) = 0, so y = 0 or = 1/2.
If (A.1), y = 0, then the constraint gives 4z
2
= 1, so z
2
= 1/4, or z = 1/2. Equation (iii) gives
= 1/8z, so we have two solution candidates: P
1
= (0, 0, 1/2) with = 1/4 and P
2
= (0, 0, 1/2)
with = 1/4.
(A.2) If = 1/2, then (iii) gives z = 1/8 = 1/4. It follows from the constraint that 2y
2
= 3/4 (recall
that we assumed x = 0), and hence y =
3/8 =
6/4. So newcandidates are: P

3
= (0,
6/4, 1/4)
with = 1/2, P
4
= (0,
6/4, 1/4) with = 1/2.

(B): Suppose = 1. Equation (iii) yields z = 1/8, and (ii) gives y = 0. Fromthe constraint, x
2
= 15/16,
so x =
15/4. Candidates: P
5
= (
15/4, 0, 1/8) with = 1, P

6
= (
15/4, 0, 1/8) with = 1.

By computing the values of the criterion function, f (0, 0, 1/2) = 1/2, f (0, 0, 1/2) = 1/2,
f (0,
6/4, 1/4) = 5/8, f (0,
6/4, 1/4) = 5/8, f (
15/4, 0, 1/8) = 17/16, f (
15/4, 0, 1/8) =
17/16, we obtain the conclusion in the text. (c) See the text.
5. The Lagrangian is L = rK +wL (K
1/2
L
1/4
Q), so necessary conditions are:
(i) L
K
= r
1
2
K
1/2
L
1/4
= 0, (ii) L
L
= w
1
4
K
1/2
L
3/4
= 0, (iii) K
1/2
L
1/4
= Q. From
(i) and (ii) we get by eliminating , L =
1
2
(rK/w). Inserting this into the constraint and solving for K
yields the answer given in the text.
14.8
2. (a) Conditions (2)(3): (i) 2x 1 2x = 0; (ii) 4y 2y = 0; (iii) 0, but = 0 if x
2
+y
2
< 1.
(b) From (ii), y(2 ) = 0, so either (I) y = 0 or (II) = 2.
(I) y = 0. If = 0, then from (i), x = 1/2 and (x, y) = (1/2, 0) is a candidate for optimum (since
it satises all the KuhnTucker conditions). If y = 0 and > 0, then from (iii) and x
2
+ y
2
1,
x
2
+y
2
= 1. But then x = 1, and (x, y) = (1, 0) are candidates, with = 1/2 and 3/2, respectively.
(II) = 2. Then from (i), x = 1/2 and (iii) gives y
2
= 3/4, so y =
3/2. So (1/2,
3/2) are
the two remaining candidates. For the conclusion, see the text.
3. (a) The Kuhn-Tucker conditions are: (i) 2(x 1) 2x = 0; (ii) 2ye
y
2
2y = 0;
(iii) 0, with = 0 if 4x
2
+ y
2
< a. From (i), x = (1 + )
1
, and (ii) reduces to y(e
y
2
+ ) = 0,
and so y = 0 (because e
y
2
+ is always positive).
(I): Assume that = 0. Then equation (i) gives x = 1. In this case we must have a x
2
+y
2
= 1.
(II): Assume that > 0. Then (iii) gives x
2
+y
2
= a, and so x =
a (remember that y = 0). Because

x = 1/(1 +) and > 0 we must have 0 < x < 1, so x =
a and a = x
2
< 1. It remains to nd the
value of and check that it is > 0. From equation (i) we get =
2(x 1)
2x
=
1
x
1 =
1
a
1 > 0.
Conclusion: The only point that satises the KuhnTucker conditions is (x, y) = (1, 0) if a 1 and
(
a, 0) if 0 < a < 1. The corresponding value of is 0 or

1
a
1, respectively. In both cases it follows
from Theorem 14.8.1 that we have found the maximum point, because Lis concave in (x, y), as we can
see by studying the Hessian
_
L
11
L
12
L
21
L
22
_
=
_
2 2 0
0 e
y
2
(2 +4y
2
) 2
_
.
(b) If a (0, 1) we have f
(a) = f (
a, 0) = 2 (
a 1)
2
1 = 2
a a, and for a 1 we
get f
(a) = f (1, 0) = 1 (not 2, as it says in the answer section of the rst printing of the book). The
derivative of f
is as given in the book, but note that in order to nd the derivative df
(a)/da when
a = 1, we need to show that the right and left derivatives (see page 242 in the book)
(f
(1
+
) = lim
h0
+
f
(1 +h) f
(1)
h
and (f
(1
) = lim
h0
(1 +h) f
(1)
h
exist and are equal. The right derivative is obviously 0, since f
(a) = 1 for all a 1. To nd the left

derivative we need to calculate lim
h0
(2
1 +h (1 + h) 1)/h. A straightforward application of

lHpitals rule shows that this limit is also 0. Hence (f
(1) exists and equals 0.

14.9
2. (a) The admissible set is the shaded region in Fig. A14.9.2 in the text.
(b) With the constraints g
1
(x, y) = x y 4, g
2
(x, y) = x 1, g
3
(x, y) = y 1,
the Lagrangian is L = x + y e
x
e
x+y

1
(x y + 4)
2
(x 1)
3
(y + 1). The
rst-order conditions are that there exist nonnegative numbers
1
,
2
, and
3
such that (i) L
x
=
1 e
x
e
x+y
+
1
+
2
= 0; (ii) L
y
= 1 e
x+y
+
1
+
3
= 0; (iii)
1
(x y + 4) = 0;
(iv)
2
(x 1) = 0; (v)
3
(y + 1) = 0. (We formulate the complementary slackness conditions as
in (14.8.5).) From (ii), e
x+y
= 1 +
1
+
3
. Inserting this into (i) yields
2
= e
x
+
3
e
x
> 0.
Because
2
> 0, (iv) implies that x = 1. So any solution must lie on the line (II) in the gure, which
shows that the third constraint must be slack. (Algebraically, because x + y 4 and x = 1, we have
y 4 x = 5 > 1.) So from (v) we get
3
= 0, and then (ii) gives
1
= e
x+y
1 e
4
1 > 0.
Thus from (iii), the rst constraint is active, so y = 4 x = 5. Hence the only possible solution is
(x
, y
) = (1, 5). Because L(x, y) is concave, we have found the optimal point.
3. (a) The feasible set is shown in Fig. A14.9.3 in the book. (The function to be maximized is f (x, y) =
x +ay. The level curves of this function are straight lines with slope 1/a if a = 0, and vertical lines if
a = 0. The dashed line in the gure is such a level curve (for a 0.25). The maximum point for f is
that point in the feasible region that we shall nd if we make a parallel displacement of this line as far to
the right as possible (why to the right?) without losing contact with the shaded region.)
The Lagrangian is L(x, y) = x + ay
1
(x
2
+ y
2
1) +
2
(x + y) (the second constraint must
be written as x y 0), so the KuhnTucker conditions are:
(i) L
1
(x, y) = 1 2
1
x +
2
= 0; (ii) L
2
(x, y) = a 2
1
y +
2
= 0;
(iii)
1
0, with
1
= 0 if x
2
+y
2
< 1; (iv)
2
0, with
2
= 0 if x +y > 0.
(b) From (i), 2
1
x = 1 +
2
1 > 0, so since from (iii)
1
0, we must have
1
> 0 and also x > 0.
Because
1
> 0, it follows from (iii) that x
2
+y
2
= 1, so any maximum point must lie on the circle.
(I) First assume that x +y = 0. Then y = x, and since x
2
+y
2
= 1, we get x =
1
2
2 (recall that we
have seen that x must be positive) and y =
1
2
2. Adding equations (i) and (ii), we get

1 +a 2
1
(x +y) +2
2
= 0
and since x +y = 0, we nd that
2
= (1 +a)/2. Now,
2
must be 0, and therefore we must have
a 1 in this case. Equation (i) gives
1
= 1 +
2
/2x = 1 a/4x = 1 a/2
2.
(II) Then consider the case x +y > 0. Then
2
= 0, and we get 1 2
1
x = 0 and a 2
1
y = 0,
which gives x = 1/(2
1
) and y = a/(2
1
). Since (x, y) must lie on the circle, we then get 1 = x
2
+y
2
=
1 +a
2
/4
2
1
, and therefore 2
1
=

1 +a
2
. This gives x =
1
1 +a
2
and y =
a
1 +a
2
. Because
x + y = (1 + a)/(2
1
), and because x + y is now assumed to be positive, we must have a > 1 in
this case. Conclusion: The only points satisfying the KuhnTucker conditions are the ones given in the
text. Since the feasible set is closed and bounded and f is continuous, it follows from the extreme value
theorem that extreme points exists.
4. (a) The Lagrangian is L = y x
2
+ y + (y x + 2) (y
2
x), which is stationary when
(i) 2x + = 0; (ii) 1 + + 2y = 0. Complementary slackness requires in addition,
(iii) 0, with = 0 if y > 0; (iv) 0, with = 0 if y x > 2; (v) 0, with = 0 if y
2
< x.
From (ii), 2y = 1 + + > 0, so y > 0. Then (iii) implies = 0, and 2y = 1 + . From (i),
x =
1
2
( ). But x y
2
> 0, so > 0, and from (v), y
2
= x.
Suppose > 0. Then y x + 2 = y y
2
+ 2 = 0 with roots y = 1 and y = 2. Only y = 2 is
feasible. Then x = y
2
= 4. Because = 0, conditions (i) and (ii) become + = 8 and 4 = 1,
so = 7/3, which contradicts 0, so (x, y) = (4, 2) is not a candidate. Therefore = 0 after
all. Thus x =
1
2
= y
2
and, by (ii), 1 = 2y = 4y
3
. Hence y = 4
1/3
, x = 4
2/3
. This is the only
remaining candidate. It is the solution with = 0, = 0, and = 1/2y = 4
1/6
.
(b) We write the problem as max xe
yx
2ey subject to y 1 +
1
2
x, x 0, y 0. The Lagrangian is
L = xe
yx
2ey (y 1 x/2), so the rst-order conditions (14.9.4) and (14.9.5) are:
(i) L
x
= e
yx
xe
yx
+
1
2
0 (= 0 if x > 0); (ii) L
y
= xe
yx
2e 0 (= 0 if y > 0);
(iii) 0 with = 0 if y < 1 +
1
2
x.
From (i) we have x 1 +
1
2
e
xy
1, so (iv) (x 1)e
yx
=
1
2
. Suppose > 0. Then (iii)
implies (v) y = 1 +
1
2
x > 0. From (ii) and (iv) we then have xe
yx
= 2e + = e
yx
+
1
2
and so
= 2e
yx
4e = 2e(e
1
2
x
2), by (v). But then > 0 implies that e
1
2
x
> 2, which contradicts x 0.
When = 0, (iv) gives x = 1. If y > 0, then (ii) yields e
y1
= 2e, and so y 1 = ln(2e) >
1
2
x when
x = 1. Thus feasibility requires that y = 0, so we see that (x, y) = (1, 0) is the only point satisfying all
the conditions, with = 0.
5. Afeasible triple (x
1
, x
2
, k
) solves the problemiff there exist numbers and such that (i) 12x
1
0
(= 0 if x
1
> 0); (ii) 3 2x
2
0 (= 0 if x
2
> 0); (iii) 2k
+ + 0 (= 0 if k
> 0);
(iv) 0 with = 0 if x
1
< k
; and (v) 0 with = 0 if x
2
< k
.
If k
= 0, then feasibility requires x
1
= 0 and x
2
= 0, and so (i) and (iii) imply that 1 and
3, which contradicts (iii). Thus, k
> 0. Next, if = 0, then x
2
3/2 and = 2k
> 0. So
x
1
= k
= 1/4, contradicting x
2
k
. So > 0, which implies that x
2
= k
. Now, if x
1
= 0 < k
,
then = 0, which contradicts (i). So 0 < x
1
=
1
2
(1 ). Next, if > 0, then x
1
= k
= x
2
=
1
2
(1 ) =
1
2
(3 ) =
1
2
( + ). But the last two equalities are only satised when = 1/3 and
= 5/3, which contradicts 0. So = 0 after all, with x
2
= k
> 0, > 0, x
1
=
1
2
(1 ) =
1
2
.
Now, from (iii) it follows that = 2k
and so, from (ii), that 3 = 2x
2
+ = 4k
. The only possible

solution is, therefore, (x
1
, x
2
, k
) = (1/2, 3/4, 3/4), with = 0 and = 3/2. (The Lagrangian is

concave in (x
1
, x
2
, k). See FMEA, Theorem 3.2.4.)
6. A minus sign has disappeared in the objective function which should be: (x +
1
2
)
2
1
2
y
2
.
(a) See Fig. A14.9.6 in the text. Note that for (x, y) to be admissible, e
x
y 2/3, and so e
x
3/2.
(b) The Lagrangian is L = (x +
1
2
)
2
1
2
y
2
1
(e
x
y)
2
(y
2
3
), and the rst-order conditions are:
(i) (2x + 1) +
1
e
x
= 0; (ii) y +
1

2
= 0; (iii)
1
0, with
1
= 0 if e
x
< y;
(iv)
2
0, with
2
= 0 if y < 2/3. From (i),
1
= (2x + 1)e
x
3/2, because of (a). From (ii),
2
=
1
y 3/2 2/3 > 0, so y = 2/3 because of (iii). Solution: (x
, y
) = (ln(3/2), 2/3), with
1
= 3[ln(3/2) +1/2],
2
= 3 ln(3/2) +5/6. The Lagrangian is concave so this is the solution.
Alternative argument: Suppose
1
= 0. Then from (ii), y =
2
0, contradicting y e
x
.
So
1
> 0, and (iii) gives y = e
x
. Suppose
2
= 0. Then from (ii),
1
= y = e
x
and (i) gives
e
2x
= 2x + 1. Dene g(x) = 2x + 1 e
2x
. Then g(0) = 0 and g
(x) = 2 + 2e
2x
> 0. So the
equation e
2x
= 2x +1 has no solution except x = 0. Thus
2
> 0, etc.
3. (a) Interpretation of the rst-order condition p(x
) = C
1
(x
, y
) x
(x
): How much is gained by

selling one ton extra of the rst commodity? p(x
), because this is the price obtained for one ton. How

much is lost? First, selling one ton extra of the rst commodity incurs the extra cost C(x
+ 1, y
)
C(x
, y
), which is approximately C
1
(x
, y
). But since presumably p
(x) < 0, producing one ton extra

leads to a decrease in income which is approximately x
(x
) (the number of tons sold times the

decrease in the price. So what we loose by increasing production by 1 ton (C
1
(x
, y
) x
(x
)) is
approximately what we gain (p(x
)). The other rst-order condition, q(y
) = C
2
(x
, y
) y
(y
)
has a similiar interpretation.
(b) See the text. If we assume that the restriction is x +y m, we have to add the condition 0, with
= 0 if x + y < m.
4. See the text. If one were to nd the partial derivatives of x and y w.r.t. p as well, it would be better to
calculate differentials, which would yield the equations (i) y dp + p dy = 24 dw wdx x dw and
(ii) U
1
dp + p(U
11
dx + U
12
dy) = U
2
dw + wU
21
dx + wU
22
dy, and then solve these equations for
dx and dy in terms of dp and dw.
5. (a) With L = x
2
+y
2
2x +1 (
1
4
x
2
+y
2
b), the rst-order conditions are:
(i) L
1
= 2x 2
1
2
x = 0; (ii) L
2
= 2y 2y = 0; (iii)
1
4
x
2
+y
2
= b.
From (ii), (1 )y = 0, and thus = 1 or y = 0.
(I) Suppose rst that = 1. Then(i) gives x =
4
3
, andfrom(iii) we have y
2
= b
1
4
x
2
= b
4
9
, whichgives
y =
_
b
4
9
. This gives two candidates: (x
1
, y
1
) = (4/3,
_
b
4
9
) and (x
2
, y
2
) = (4/3,
_
b
4
9
).
(II) If y = 0, then from (iii), x
2
= 4b, i.e. x = 2
b. This gives two further candidates: (x

3
, y
3
) =
(2
b, 0) and (x
4
, y
4
) = (2
b, 0). The objective function evaluated at the candidates are: f (x

1
, y
1
) =
f (x
2
, y
2
) = b1/3, f (x
3
, y
3
) = (2
b1)
2
= 4b4
b+1, f (x
4
, y
4
) = (2
b1)
2
= 4b+4
b+1
Clearly, (x
4
, y
4
) is the maximum point. To decide which of the points (x
3
, y
3
), (x
1
, y
1
), or (x
2
, y
2
) give
the minimum, we have to decide which of 4b 4
b + 1 and b
1
3
is the largest. The difference is
4b 4
b +1
_
b
1
3
_
= 3
_
b
4
3
b +
4
9
_
= 3
_
b
2
3
_
2
> 0 since b >
4
9
. Thus the minimum occurs
at (x
1
, y
1
) and (x
2
, y
2
).
The constraint x
2
/4 +y
2
= b is the ellipse indicated in the gure. The objective function f (x, y) =
(x 1)
2
+y
2
is the square of the distance between (x, y) and the point (1, 0). The level curves for f are
therefore circles centred at (1, 0), and in the gure we see those two that passes through the maximum
and minimum points. (b) See the text.
y
x
-4
-2
2
4
-4 -2 2 4 6
(x
1
, y
1
)
(x
3
, y
3
)
(x
2
, y
2
)
(x
4
, y
4
)
For Problem 14.R.5
7. (a) With L = x
2
2x +1 +y
2
2y [(x +y)
x +y +b 2
a], the rst-order conditions are:

(i) L
1
= 2x 2 [
x +y +b +(x +y)/
x +y +b ] = 0,
(ii) L
2
= 2y 2 [
x +y +b + (x + y)/
x +y +b ] = 0. From these equations it follows

immediately that 2x 2 = 2y 2, so x = y. See the text.
15 MATRI X AND VECTOR AL GEBRA 57
(b) Differentiation yields: (i) dx = dy; (ii) 6x
2
dx + x
2
db + 2bx dx = da. From these equations we
easily read off the rst-order partials of x and y w.r.t. a and b. Further,
2
x
a
2
=

a
_
x
a
_
=

a
1
6x
2
+2bx
=
12x +2b
(6x
2
+2bx)
2
x
a
=
12x +2b
(6x
2
+2bx)
3
=
6x +b
4(3x
2
+bx)
3
8. (a) The Lagrangian is L = xy
1
(x
2
+ry
2
m)
2
(x), and the necessary KuhnTucker conditions
for (x
, y
) to solve the problem are:

(i) L
1
= y
2
1
x
+
2
= 0;
(ii) L
2
= x
2r
1
y
= 0;
(iii)
1
0, with
1
= 0 if (x
)
2
+r(y
)
2
< m;
(iv)
2
0, with
2
= 0 if x
> 1;
(v) (x
)
2
+r(y
)
2
m;
(vi) x
1
(b) From (ii) and (vi) we see that
1
= 0 is impossible. Thus
1
> 0, and from (iii) and (v),
(vii) (x
)
2
+r(y
)
2
= m.
(I): Assume
2
= 0. Then from (i) and (ii), y
= 2
1
x
and x
= 2
1
ry
, so y
= 4
2
1
ry
. If
y
= 0, then (ii) implies x
= 0, which is impossible. Hence,

2
1
= 1/4r and thus
1
= 1/2
r. Then
y
= x
r, which inserted into (vii) and solved for x
yields x
m/2 and then y
m/2r. Note
that x
1

m/2 1 m 2. Thus for m 2, x
=

m/2 and y
=

m/2r, with
1
= 1/2
r and
2
= 0 is a solution candidate.
(II): Assume
2
> 0. Then x
= 1 and from (vii) we have r(y
)
2
= m 1, so y
=

(m1)/r
(y
(m1)/r contradicts (ii)). Inserting these values for x
and y
into (i) and (ii) and solving

for
1
and
2
yields
1
= 1/2
r(m1) and furthermore,

2
= (2 m)/
r(m1). Note that
2
> 0 1 < m < 2. Thus, for 1 < m < 2, the only solution candidate is x
= 1, y
(m1)/r, with
1
= 1/2
r(m1) and
2
= (2 m)/
r(m1).
The objective function is continuous and the constraint set is obviously closed and bounded, so by the
extreme value theorem there has to be a maximum. The solution candidates we have found are therefore
optimal. (Alternatively, L
11
= 2
1
0, L
22
= 2r
1
0, and = L
11
L
22
(L
12
)
2
= 4r
2
1
1.
In the case m 2, = 0, and in the case 1 < m < 2, = 1/(m1) > 0. Thus in both cases, L(x, y)
is concave.)
15 Matrix and Vector Algebra
15.1
2. Here is one method: Adding the 4 equations gives x
1
+x
2
+x
3
+x
4
=
1
3
(b
1
+b
2
+b
3
+b
4
), after dividing
by 3. From this equation, subtracting each original equation in turn gives x
1
=
2
3
b
1
+
1
3
(b
2
+b
3
+b
4
),
x
2
=
2
3
b
2
+
1
3
(b
1
+b
3
+b
4
), x
3
=
2
3
b
3
+
1
3
(b
1
+b
2
+b
4
), x
4
=
2
3
b
4
+
1
3
(b
1
+b
2
+b
3
).
Systematic elimination of the variables starting by eliminating (say) x
4
is an alternative.
6. Suggestion: Solve the rst equation for y. Insert this expression for y and x = 93.53 into the third
equation. Solve it for s. Insert the results into the second equation and solve for c, etc.
58 15 MATRI X AND VECTOR AL GEBRA
15.3
1. (a)
_
0 2
3 1
__
1 4
1 5
_
=
_
0 (1) +(2) 1 0 4 +(2) 5
3 (1) + 1 1 3 4 +1 5
_
=
_
2 10
2 17
_
The rest goes in the same way. Note that in (d) AB is not dened because the number of columns in A is
not equal to the number of rows in B.
5. (a) We know that A is an m n matrix. Let B be a p q matrix. The matrix product AB is dened if
and only if n = p, and BA is dened if and only if q = m. So for both AB and BA to be dened, it is
necessary and sufcient that B is an n m matrix.
(b) We know from part (a) that if BA and AB are dened, then B must be a 2 2 matrix. So let B =
_
x y
z w
_
. Then BA =
_
x y
z w
__
1 2
2 3
_
=
_
x +2y 2x +3y
z +2w 2z +3w
_
, and AB =
_
1 2
2 3
__
x y
z w
_
=
_
x +2z y +2w
2x +3z 2y +3w
_
. Hence, BA = AB iff (i) x + 2y = x + 2z, (ii) 2x + 3y = y + 2w,
(iii) z + 2w = 2x + 3z, and (iv) 2z + 3w = 2y + 3w. The rst and last of these four equations
are true if and only if y = z, and if y = z, then the second and third are true if and only if x = w y.
Hence, the matrices B that commute with A are precisely the matrices of the form
B =
_
w y y
y w
_
= w
_
1 0
0 1
_
+y
_
1 1
1 0
_
where y and w can be any real numbers.
15.4
2. We start by performing the multiplication
a d e
d b f
e f c
x
y
z
ax +dy +ez
dx +by +f z
ex +fy +cz
. Next,
(x, y, z)
ax +dy +ez
dx +by +f z
ex +fy +cz
= (ax
2
+by
2
+cz
2
+2dxy +2exz +2fyz), which is a 1 1 matrix.
7. (a) Direct verication yields (i) A
2
= (a +d)A (ad bc)I
2
=
_
a
2
+bc ab +bd
ac +cd bc +d
2
_
(b) For the matrix A in (a), A
2
= 0 if a + d = 0 and ad = bc, so one example with A = 0 is
A =
_
1 1
1 1
_
.
(c) Multiplying (i) in (a) by A and using A
3
= 0 yields (ii) (a + d)A
2
= (ad bc)A. Multiplying
by A once more gives (ad bc)A
2
= 0. If ad bc = 0, then A
2
= 0. If ad bc = 0, (ii) yields
(a + d)A
2
= 0, and if a + d = 0, again A
2
= 0. Finally, if ad bc = a + d = 0, then (i) implies
A
2
= 0.
15.5
6. In general, for any natural number n > 3, one has ((A
1
A
2
A
n1
)A
n
)
= A
n
(A
1
A
2
A
n1
)
.
As the induction hypothesis, suppose the result is true for n 1. Then the last expression becomes
A
n
A
n1
, A
2
A
1
, so the result is true for n.
15 MATRI X AND VECTOR AL GEBRA 59
8. (a) TS = S is shown in the text. A similar argument shows that T
2
=
1
2
T +
1
2
S. To prove the last
equality, we do not have to consider the individual elements: T
3
= TT
2
= T(
1
2
T+
1
2
S) =
1
2
T
2
+
1
2
TS =
1
2
(
1
2
T +
1
2
S) +
1
2
S =
1
4
T +
3
4
S.
(b) The appropriate formula is () T
n
= 2
1n
T + (1 2
1n
)S. This formula is correct for n = 1
(and for n = 2, 3). Suppose () is true for n = k. Then using the two rst equalities in (a), T
k+1
=
TT
k
= T(2
1k
T + (1 2
1k
)S) = 2
1k
T
2
+ (1 2
1k
)TS = 2
1k
(
1
2
T +
1
2
S) + (1 2
1k
)S =
2
k
T +2
k
S +S 2 2
k
S = 2
k
T +(1 2
k
)S, which is formula () for n = k +1.
15.6
3. By using elementary operations, we nd that
w x y z
2 1 4 3 1
1 3 2 1 3c
1 1 2 1 c
2
1 0 2 2 1 c
2
0 1 0 1 2c
2
1
0 0 0 0 5c
2
+3c +2
We can tell from the last matrix that the system has solutions if and only if 5c
2
+3c +2 = 0, that is, if
and only if c = 1 or c = 2/5. For these particular values of c we get the solutions in the text.
4. (a) After moving the rst row down to row number three, Gaussian elimination yields the matrix
1 2 1 b
2
0 1 2
3
2
b
2
1
2
b
3
0 0 3 4a b
1
+(2a
3
2
)b
2
+(
1
2
a)b
3
. Obviously, there is a unique solution iff a = 3/4.

(b) Put a = 3/4 in part (a). Then the last row in the matrix in (a) becomes (0, 0, 0, b
1
1
4
b
3
). It follows
that if b
1
=
1
4
b
3
there is no solution. If b
1
=
1
4
b
3
there are an innite number of solutions. In fact,
x = 2b
2
+b
3
5t , y =
3
2
b
2
1
2
b
3
+2t , z = t , with t .
15.7
3. Using the denitions of vector addition and multiplication of a vector by a real number, we get
3(x, y, z) + 5(1, 2, 3) = (4, 1, 3) (3x 5, 3y + 10, 3z + 15) = (4, 1, 3). Since two vectors
are equal if and only if they are component-wise equal, this vector equation is equivalent to the equation
system 3x 5 = 4, 3y +10 = 1, and 3z +15 = 3, with the obvious solution x = 3, y = 3 , z = 4.
5. We need to nd numbers t and s such that t (2, 1)+s(1, 4) = (4, 11). This vector equation is equivalent
to (2t + s, t + 4s) = (4, 11), which in turn is equivalent to the equation system (i) 2t + s = 4
(ii) t +4s = 11. This system has the solution t = 3, s = 2, so (4, 11) = 3(2, 1) 2(1, 4).
15.8
2. (a) See the text. (b) As runs through [0, 1], the vector x will run through all points on the line segment
S between a and b. In fact, according to the pointpoint formula, the line L through (3, 1) and (1, 2)
has the equation x
2
=
1
4
x
1
+
7
4
or x
1
+ 4x
2
= 7. The line segment S is traced out by having x
1
run
through [3, 1] as x
2
runs through [1, 2]. Now, (1 )a + b = (3 4, 1 + ). Any point (x
1
, x
2
)
on L satises x
1
+ 4x
2
= 7 and equals (3 4, 1 + ) for =
1
4
(3 x
1
) = x
2
1. Any point on the
segment of this line between a = (3, 1) and b = (1, 2) equals (3 4, 1 +) for some [0, 1].
60 15 MATRI X AND VECTOR AL GEBRA
8. (||a|| +||b||)
2
||a +b||
2
= ||a||
2
+2||a|| ||b|| +||b||
2
(a +b)(a +b) = 2(||a|| ||b|| a b)
2(||a|| ||b|| |a b|) 0 by the CauchySchwarz inequality (2).
15.9
3. One method: (5, 2, 1) (3, 4, 3) = (2, 2, 4) and (2, 1, 4) (3, 4, 3) = (1, 5, 7) are two
vectors in the plane. The normal (p
1
, p
2
, p
3
) must be orthogonal to both these vectors, so (2, 2, 4)
(p
1
, p
2
, p
3
) = 2p
1
2p
2
+ 4p
3
= 0 and (1, 5, 7) (p
1
, p
2
, p
3
) = p
1
5p
2
+ 7p
3
= 0. One
solution of these two equations is (p
1
, p
2
, p
3
) = (1, 3, 2). Then using formula (4) with (a
1
, a
2
, a
3
) =
(2, 1, 4) yields (1, 3, 2) (x
1
2, x
2
+1, x
3
4) = 0, or x
1
3x
2
2x
3
= 3.
A more pedestrian approach is to assume that the equation is ax +by +cz = d and require the three
points to satisfy the equation: a +2c = d, 5a +2b +c = d, 2a b +4c = d. Solve for a, b, and c in
terms of d, insert the results into the equation ax +by +cz = d and cancel d.
7. (a)
_
1 4 1
2 2 8
_
2

_
1 4 1
0 6 6
_
1/6

_
1 4 1
0 1 1
_

4

_
1 0 5
0 1 1
_
The solution is x
1
= 5, x
2
= 1.
(b)
2 2 1 2
1 3 1 0
3 4 1 1
1 3 1 0
2 2 1 2
3 4 1 1
2 3
1 3 1 0
0 8 3 2
0 13 4 1
1/8
1 3 1 0
0 1 3/8 1/4
0 13 4 1
13
1 3 1 0
0 1 3/8 1/4
0 0 7/8 9/4
8/7
1 3 1 0
0 1 3/8 1/4
0 0 1 18/7
1 0 1/8 3/4
0 1 3/8 1/4
0 0 1 18/7
3/8 1/8
1 0 0 3/7
0 1 0 5/7
0 0 1 18/7
. The solution is x
1
= 3/7, x
2
= 5/7, x
3
= 18/7.
(c)
_
1 3 4 0
5 1 1 0
_
5

_
1 3 4 0
0 14 19 0
_
1/14
_
1 3 4 0
0 1 19/14 0
_

3

_
1 0 1/14 0
0 1 19/14 0
_
The solution is x
1
= (1/14)x
3
, x
2
= (19/14)x
3
, where x
3
is arbitrary. (One degree of freedom.)
10. (a) See the text.
(b) In (a) we saw that a can be produced even without throwing away outputs. For b to be possible if we
are allowed to throw away output, there must exist a in [0, 1] such that 6 +2 7, 2 +6 5, and
6 +10 5. These inequalities reduce to 5/6, 1/2, 5/6, which are incompatible.
(c) Revenue = R() = p
1
x
1
+p
2
x
2
+p
3
x
3
= (6p
1
2p
2
6p
3
)+2p
1
+6p
2
+10p
3
. If the constant
slope 6p
1
2p
2
6p
3
is > 0, then R() is maximized at = 1; if 6p
1
2p
2
6p
3
is < 0, then R()
is maximized at = 0. Only in the special case where 6p
1
2p
2
6p
3
= 0 can the two plants both
remain in use.
16 DETERMI NANTS AND I NVERSE MATRI CES 61
11. If PQ QP = P, then PQ = QP + P, and so P
2
Q = P(PQ) = P(QP + P) = (PQ)P + P
2
=
(QP+P)P+P
2
= QP
2
+2P
2
. Thus, P
2
QQP
2
= 2P
2
. Moreover, P
3
Q = P(P
2
Q) = P(QP
2
+2P
2
) =
(PQ)P
2
+2P
3
= (QP +P)P
2
+2P
3
= QP
3
+3P
3
Hence, P
3
QQP
3
= 3P
3
.
16 Determinants and Inverse Matrices
16.1
3. (a) Cramers rule gives x =
8 1
5 2
3 1
1 2
=
16 +5
6 +1
=
11
5
, y =
3 8
1 5
3 1
1 2
=
15 8
5
=
7
5
.
(b) and (c) are done in the same way.
7. (b) Note that because c
1
is the proportion of income consumed, we can assume that 0 c
1
1. Likewise,
0 c
2
1. Because m
1
0 and m
2
0, we see that D > 0 (excluding the case c
1
= c
2
= 1).
(c) Y
2
depends linearly on A
1
. Increasing A
1
by one unit changes Y
2
by the factor m
1
/D 0, so Y
2
increases when A
1
increases.
Here is an economic explanation: An increase in A
1
increases nation 1s income, Y
1
. This in turn
increases nation 1s imports, M
1
. However, nation 1s imports are nation 2s exports, so this causes nation
2s income, Y
2
, to increase, and so on.
16.2
1. (a) Sarruss rule yields:
1 1 0
1 3 2
1 0 0
= 0 2 +0 0 0 0 = 2.
(b)
1 1 0
1 3 2
1 2 1
= 3 2 0 0 4 (1) = 2
(c) 5 of the 6 products have 0 as a factor. The only product that does not include 0 as a factor is the product
of the terms on the main diagonal. The determinant is therefore adf .
(d)
a 0 b
0 e 0
c 0 d
= aed +0 +0 bec 0 0 = e(ad bc)

3. (a) The determinant of the coefcient matrix is |A| =
1 1 1
1 1 1
1 1 1
= 4.
The numerators in (16.2.4) are (verify!)
2 1 1
0 1 1
6 1 1
= 4 ,
1 2 1
1 0 1
1 6 1
= 8 ,
1 1 2
1 1 0
1 1 6
= 12
Hence, (4) yields the solution x
1
= 1, x
2
= 2, and x
3
= 3. Inserting this into the original system of
equations conrms that this is a correct answer.
(b) The determinant of the coefcient matrix is equal to 2, and the numerators in (16.2.4) are all 0, so
the unique solution is x
1
= x
2
= x
3
= 0. (c). Follow the pattern in (a).
62 16 DETERMI NANTS AND I NVERSE MATRI CES
6. (a) Substituting T = d +t Y into the expression for C gives C = a bd +b(1 t )Y. Substituting for
C in the expression for Y then yields Y = a +b(Y d t Y) +A
0
. Then solve for Y, T , and C in turn
to derive the answers given in (b) below.
(b) We write the system as
1 1 0
b 1 b
t 0 1
Y
C
T
A
0
a
d
. Then Cramers rule yields

Y =
A
0
1 0
a 1 b
d 0 1
1 1 0
b 1 b
t 0 1
=
a bd +A
0
1 b(1 t )
, C =
1 A
0
0
b a b
t d 1
1 1 0
b 1 b
t 0 1
=
a bd +A
0
b(1 t )
1 b(1 t )
T =
1 1 A
0
b 1 a
t 0 d
1 1 0
b 1 b
t 0 1
=
t (a +A
0
) +(1 b)d
1 b(1 t )
(This problem is meant to train you in using Cramers rule. Note how systematic elimination is much
more efcient.)
16.3
1. Each of the determinants is a sum of 4! = 24 terms. In (a) there is only one nonzero term. In fact,
according to (16.3.4), the value of the determinant is abcd. (b) Only two terms in the sum are nonzero:
The product of the elements on the main diagonal, which is 1 1 1 d, with a plus sign, and the term
shown here:
1 0 0 1
0 1 0 0
0 0 1 0
a b c d
Since there are 5 rising lines between the pairs, the sign of the product 1 1 1 a must be minus. So the
value of the determinant is d a. (c) 4 terms are nonzero. See the text.
16.4
10. (a) and (b) see the text. (c) We have (I
n
A)(I
n
+A) = I
n
I
n
AI
n
+I
n
AAA = I
n
A+AA
2
=
I
n
A
2
, and this expression equals 0 if and only if A
2
= I
n
.
12. The description in the answer in the text amounts to the following:
D
n
=
a +b a a
a a +b a
.
.
.
.
.
.
.
.
.
.
.
.
a a a

1
.
.
.
1
=
na +b na +b na +b
a a +b a
.
.
.
.
.
.
.
.
.
.
.
.
a a a +b
= (na +b)
1 1 1
a a +b a
.
.
.
.
.
.
.
.
.
.
.
.
a a a +b
a a

.
.
.
= (na +b)
1 1 1
0 b 0
.
.
.
.
.
.
.
.
.
.
.
.
0 0 b
According to (16.3.4), the last determinant is b

n1
. Thus D
n
= (na +b)b
n1
.
16.5
1. (a) See the answer. (b) One possibility is to expand by the second row or the third column (because they
have both two zero entries). But it is easier rst to use elementary operations to get a row or a column
with one more zero. For instance in this way:
1 2 3 4
0 1 0 11
2 1 0 3
2 0 1 3
2 2
1 2 3 4
0 1 0 11
0 5 6 5
0 4 5 11
1 0 11
5 6 5
4 5 11
5 4
1 0 11
0 6 60
0 5 55
= 1
6 60
5 55
= (330 +300) = 30
When computing determinants one can use elementary row operations as well as column operations, but
columnoperations become meaningless whensolvinglinear equationsystems usingGaussianelimination.
When elementary operations have produced a row or column with only one non-zero element, then it is
natural to expand the determinant by that row or column. (c) See the answer in the text.
16.6
6. (b) From (a), A
3
2A
2
+A = I, or A(A
2
2A+I) = I, so using Theorem 16.6.2, A
1
= A
2
2A+I.
The last expression can also be written (A I)
2
. (c) See the text.
9. B
2
+B =
_
3/2 5
1/4 3/2
_
+
_
1/2 5
1/4 1/2
_
=
_
1 0
0 1
_
= I. One can verify directly that B
3
2B+
I = 0, but here is an alternative that makes use of B
2
+B = I: B
3
2B+I = B
3
+B
2
B
2
2B+I =
B(B
2
+B) B
2
2B +I = B B
2
2B +I = (B
2
+B) +I = 0.
16.7
1. (a) The determinant is 1012 = 2, and the adjoint is
_
C
11
C
21
C
12
C
22
_
=
_
5 3
4 2
_
, so the inverse is
1
2
_
5 3
4 2
_
=
_
5/2 3/2
2 1
_
(b) If we denote the matrix by A, the adjoint is
adj A =
C
11
C
21
C
31
C
12
C
22
C
32
C
13
C
23
C
33
1 4 2
2 1 4
4 2 1
and the determinant is |A| = a

11
C
11
+a
21
C
21
+a
31
C
31
= 1 1 +2 4 +0 2 = 9, (by expansion along
the rst column). Hence,
A
1
=
1
9
(adj A) =
1
9
1 4 2
2 1 4
4 2 1
(c) Since the determinant is 0 there is no inverse.

3. The determinant of I A is |I A| = 0.496, and the adjoint is adj(I A) =
0.72 0.64 0.40

0.08 0.76 0.32
0.16 0.28 0.64
.
Hence (I A)
1
=
1
0.496
adj(I A)
1.45161 1.29032 0.80645

0.16129 1.53226 0.64516
0.32258 0.56452 1.29032
, rounded to ve decimal
places. If you want an exact answer, note that
1000
496
=
125
62
and adj(I A) =
0.72 0.64 0.40

0.08 0.76 0.32
0.16 0.28 0.64
=
1
25
18 16 10
2 19 8
4 7 16
. This gives (I A)
1
=
5
62
18 16 10
2 19 8
4 7 16
.
4. Let B denote the n p matrix whose ith column has the elements b
1i
, b
2i
, . . . , b
ni
. The p systems of
n equations in n unknowns can be expressed as AX = B, where A is n n and X is n p. Following
the method illustrated in Example 2, exactly the same row operations that transform the n 2n matrix
(A : I) into (I : A
1
) will also transform the n (n + p) matrix (A : B) into (I : B
), where
B
is the matrix with elements b
ij
. (In fact, because these row operations are together equivalent to
premultiplication by A
1
, it must be true that B
= A
1
B.) When k = r, the solution to the system is
x
1
= b
1r
, x
2
= b
2r
, . . . , x
n
= b
nr
.
5. (a) The following shows that the inverse is
_
2 1
3
2

1
2
_
:
_
1 2 1 0
3 4 0 1
_
3

_
1 2 1 0
0 2 3 1
_

1
2
_
1 2 1 0
0 1
3
2

1
2
_
2

_
1 2 2 1
0 1
3
2

1
2
_
(b)
1 2 3 1 0 0
2 4 5 0 1 0
3 5 6 0 0 1
2 3
1 2 3 1 0 0
0 0 1 2 1 0
0 1 3 3 0 1
1 2 3 1 0 0
0 1 3 3 0 1
0 0 1 2 1 0
1
1
1 2 3 1 0 0
0 1 3 3 0 1
0 0 1 2 1 0
1 0 3 5 0 2
0 1 3 3 0 1
0 0 1 2 1 0
3 3
1 0 0 1 3 2
0 1 0 3 3 1
0 0 1 2 1 0
(c) We see that the third row is the st row multiplied by 3, so the matrix has no inverse.
16.8
1. (a) The determinant |A| of the coefcient matrix is |A| =
1 2 1
2 1 1
1 1 3
= 19.
The determinants in (16.8.2) are (verify!)
5 2 1
6 1 1
3 1 3
= 19 ,
1 5 1
2 6 1
1 3 3
= 38 ,
1 2 5
2 1 6
1 1 3
= 38
According to (16.8.4) the solution is x = 19/19 = 1, y = 38/19 = 2, and z = 38/19 = 2. Inserting
this into the original system of equations conrms that this is the correct answer.
(b) The determinant |A| of the coefcient matrix is
1 1 0 0
1 0 1 0
0 1 1 1
0 1 0 1
= 1.
The determinants in (16.8.2) are (verify!)
3 1 0 0
2 0 1 0
6 1 1 1
1 1 0 1
= 3 ,
1 3 0 0
1 2 1 0
0 6 1 1
0 1 0 1
= 6 ,
1 1 3 0
1 0 2 0
0 1 6 1
0 1 1 1
= 5 ,
1 1 0 2
1 0 1 3
0 1 1 6
0 1 0 1
= 5
According to (16.8.4) the solution is x = 3, y = 6, z = 5, and u = 5. Inserting this into the original
system of equations conrms that this is the correct answer.
3. According to Theorem 16.8.2, the system has nontrivial solutions iff the determinant of the coefcient
equal to 0. Expansion according to the rst row gives
a b c
b c a
c a b
= a
c a
a b
b a
c b
+c
b c
c a
= a(bc a
2
) b(b
2
ac) +c(ab c
2
) = 3abc a
3
b
3
c
3
.
Thus the system has nontrivial solutions iff 3abc a
3
b
3
c
3
= 0.
16.9
4. The equation system is obtained directly from (16.9.4).
3. It is a bad idea to use brute force here. Note instead that rows 1 and 3 and rows 2 and 4 in the determinant
have much in common. So begin by subtracting row 3 from row 1, and row 4 from row 2. According
to Theorem 16.4.1(F), this does not change the value of the determinant. This gives, if we thereafter use
Theorem 16.4.1(C),
0 a b 0 b a
b a 0 a b 0
x b x a
a x b x
= (a b)
2
0 1 0 1
1 0 1 0
x b x a
a x b x
= (a b)
2
0 1 0 0
1 0 1 0
x b x a +b
a x b 2x
The last equality is obtained by adding row 2 to row 4 in the middle determinant. If we expand the last
determinant by the row 1, we end up with an easy 3 3 determinant to evaluate. The equation becomes
(a b)
2
(4x
2
(a +b)
2
) = (a b)
2
(2x +(a +b))(2x (a +b)) = 0. The conclusion follows.
5. (a) Expanding by column 3: |A| =
q 1 q 2
1 p 2 p
2 1 0
= (q 2)
1 p
2 1
(2 p)
q 1
2 1
=
(q 2)(1 +2p) (2 p)(q +2) = (q 2)(p +1), but there are many other ways.
|A + E| =
q +1 0 q 1
2 1 p 3 p
3 0 1
= (1 p)
q +1 q 1
3 1
= (1 p)[q + 1 3(q 1)]

= 2(p 1)(q 2). For the rest see the text.
8. (a) Note that
U
2
=
1 1 . . . 1
1 1 . . . 1
.
.
.
.
.
.
.
.
.
.
.
.
1 1 . . . 1
1 1 . . . 1
1 1 . . . 1
.
.
.
.
.
.
.
.
.
.
.
.
1 1 . . . 1
n n . . . n
n n . . . n
.
.
.
.
.
.
.
.
.
.
.
.
n n . . . n
= nU
(b) The trick is to note that
A =
4 3 3
3 4 3
3 3 4
1 0 0
0 1 0
0 0 1
3 3 3
3 3 3
3 3 3
= I
3
+3U
From (a), (I
3
+3U)(I
3
+bU) = I
3
+(3 +b +3 3bU) = I
3
+(3 +10b)U, which is equal to I
3
if we
choose b = 3/10. It follows that
A
1
= (I
3
+3U)
1
= I
3
(3/10)U =
1 0 0
0 1 0
0 0 1
3
10
3
10
3
10
3
10
3
10
3
10
3
10
3
10
3
10
=
1
10
7 3 3
3 7 3
3 3 7

17 L I NEAR PROGRAMMI NG 67
11. (a) Gauss-elimination yields
a 1 4 2
2 1 a
2
2
1 0 3 a
1 0 3 a
2 1 a
2
2
a 1 4 2
2 a
1 0 3 a
0 1 a
2
+6 2a +2
0 1 3a +4 a
2
+2
1 0 3 a
0 1 a
2
+6 2a +2
0 0 a
2
+3a 2 a
2
+2a
It follows that the system has unique solution iff a

2
+3a 2 = 0, i.e. iff a = 1 and a = 2.
If a = 2, the last row consists only of 0s so there ia an innite number of solutions, while if a = 1,
there are no solutions.
(b) If we perform the same elementary operations as in (a) on the associated extended matrix, we get
1 0 3 b
3
0 1 a
2
+6 b
2
2b
3
0 0 a
2
+3a 2 b
1
b
2
+(2 a)b
3
,
We see that there are innitely many solutions iff all elements in the last row are 0, i.e. iff a = 1 and
b
1
b
2
+b
3
= 0, or when a = 2 and b
1
= b
2
.
13. For once we use unsystematic elimination. Solve the rst equation for y, the second for z, and the fourth
for u, using the expression found for y. Insert all this into the third equation. This gives: a(b 2)x =
2a+2b+3. There is a unique solution provided a(b2) = 1. We easily verify the solutions in the text.
15. We prove the result for 3 3 matrices that differ only in the rst row:
a
11
+x a
12
+y a
13
+z
a
21
a
22
a
23
a
31
a
32
a
33
a
11
a
12
a
13
a
21
a
22
a
23
a
31
a
32
a
33
x y z
a
21
a
22
a
23
a
31
a
32
a
33
()
We expand the rst determinant in () by the rst row, where C
11
, C
12
, C
13
are the complements of the
three entries in the rst row, and we get
(a
11
+x)C
11
+(a
12
+y)C
12
+(a
13
+z)C
13
= [a
11
C
11
+a
12
C
12
+a
13
C
13
] +[xC
11
+yC
12
+zC
13
]
The sums in square brackets are the two last determinants in ().
17 Linear Programming
17.1
3. The set A is the shaded set in Fig. SM17.1.3.
(a) The solution is obviously at the point P in the gure because it has the largest x
2
coordinate among
68 17 L I NE AR PROGRAMMI NG
all points in A. P is the point where the two lines 2x
1
+ x
2
= 2 and x
1
+ 2x
2
= 8 intersect, and the
solution of these two equations is (x
1
, x
2
) = (4/5, 18/5).
x
2
x
1
P
A
Q
8
4
2x
1
+x
2
= 2
3x
1
+2x
2
= c
x
1
+2x
2
= 8
Figure SM17.1.3
(b) The point in A with the largest x
1
coordinate is obviously Q = (8, 0).
(c) One of the lines 3x
1
+ 2x
2
= c is the dashed line in Fig. SM17.1.3. As c increases, the line moves
out farther and farther to the north-east. The line that has the largest value of c and still has a point in
common with A, is the point Q in the gure.
(d) The line 2x
1
2x
2
= c (or x
2
= x
1
c/2) makes a 45
angle with the x

1
axis, and intersects the x
1
axis at c/2. As c decreases, the line moves to the left. The line that has the smallest value of c and still
has a point in common with A, is the point P in the gure.
(e) The line 2x
1
+4x
2
= c is parallel to the line x
1
+2x
2
= 8 in the gure. As c increases, the line moves
out farther and farther to the north-east. The line with points in common with A that has the largest value
of c is obviously obtained when the line covers the line x
1
+2x
2
= 8. So all points on the line segment
between P and Q are solutions.
(f) The line 3x
1
2x
2
= c is parallel to the dashed line in the gure, and intersects the x
1
axis at c/3.
As c decreases, the line moves out farther and farther to the north-east, so the solution is at Q = (8, 0).
(We could also argue like this: Minimizing 3x
1
2x
2
subject to (x
1
, x
2
) A must occur at the same
point as maximizing 3x
1
+2x
2
subject to (x
1
, x
2
) A.)
17.2
1. (a) See Fig. A17.1.1(a) in the text. When 3x
1
+ 2x
2
6 is replaced by 3x
1
+ 2x
2
7 in Problem
17.1.1, the feasible set expands because the steepest line is moved to the right. The new optimal point
is at the intersection of the lines 3x
1
+ 2x
2
= 7 and x
1
+ 4x
2
= 4, and it follows that the solution is
(x
1
, x
2
) = (2, 1/2). The old maximum value of the objective function was 36/5. The new optimal value
is 3 2 +4
1
2
= 8 = 40/5, and the difference in optimal value is u
1
= 5/4.
(b) When x
1
+4x
2
4 is replaced by x
1
+4x
2
5, the feasible set expands because the line x
1
+4x
2
= 4
is moved up. The newoptimal point is at the intersection of the lines 3x
1
+2x
2
= 6 and x
1
+4x
2
= 5, and
it follows that the solution is (x
1
, x
2
) = (7/5, 9/10). The old maximum value of the objective function
was 36/5. The new optimal value is 39/5, and the difference in optimal value is u
2
= 3/5.
(c) See the text.
17.3
1. (a) From Fig. A17.3.1(a) in the text it is clear as c increases, the dashed line moves out farther and farther
to the northeast. The line that has the largest value of c and still has a point in common with the feasible
set is the point P, which has coordinates (x, y) = (0, 3).
(b) In Fig. A17.3.1(b), as c decreases, the dashed line moves farther and farther to the south west. The
line that has the smallest value of c and still has a point in common with the feasible set is the point P,
which has coordinates (u
1
, u
2
) = (0, 1). The associated minimum value is 20u
1
+ 21u
2
= 21. This is
the maximum value in the primal problem, so the answer to question (c) is yes.
2. Actually not much to add. From the easily produced gure we can read off the solution.
3. See Fig. SM17.3.3 and the text.
x
2
5
x
1
5 10
2x
1
+x
2
= 16
400x
1
+500x
2
= konst.
x
1
+2x
2
= 11
(x
1
, x
2
) = (7, 2)
x
1
+4x
2
= 16
Figure SM17.3.3
17.4
2. (a) The problemis similar to Problem17.3.3. See the answer in the text. Note that 300x
1
+200x
2
= 2800.
(b) The dual problem is
min (54u
1
+48u
2
+50u
3
) subject to
6u
1
+4u
2
+5u
3
300
3u
1
+6u
2
+5u
3
200
u
1
, u
2
, u
3
0
70 17 L I NE AR PROGRAMMI NG
y
100
200
300
400
500
600
700
800
900
1000
x
100 200 300 400 500 600 700 800 900 1000
(2)
(1)
(3)
Figure SM17.5.3
The optimal solution of the primal is x
1
= 8, x
2
= 2. Since they are both positive, the rst two constraints
in the dual is satised with equality at the optimal triple (u
1
, u
2
, u
3
). Since the second constraint in the
primal is satised with strict inequality: 4x
1
+ 6x
2
= 44 < 48, u
2
= 0. So 6u
1
+ 5u
3
= 300 and
3u
1
+5u
3
= 200. It follow that u
1
= 100/3, u
2
= 0, and u
3
= 20, with 54u
1
+48u
2
+50u
3
= 2800.
(c) See the text.
17.5
3. (a) See the text. (b) The dual is given in the text. Look at Fig. SM17.5.3. We see from the gure that
optimum occurs at the point where the rst and the third constraint are satised with equality i.e. where
10x
1
+ 20x
2
= 10 000 and 20x
1
+ 20x
2
= 11 000. The solution is x
1
= 100 and x
2
= 450. The
maximum value of the criterion is 300 100 +500 450 = 255 000.
By complementary slackness, the constraints in the dual problem must in optimum be satised with
equality. Since the second constraint in the primal in the optimum has a slack (20 100 + 10 450 <
8000), then y
2
= 0. Hence 10y
1
+ 20y
3
= 300, 20y
1
+ 20y
3
= 500. It follows that the solution
of the dual problem is y
1
= 20, y
2
= 0, y
3
= 5. The minimum value of the objective function is
10 000 20 +8 000 0 +11 000 5 = 255 000.
(c) If the cost per hour in factory 1 increases by 100, the maximum in the primal problem would increase
by y
1
= 20. (The numbers y
1
, y
2
, and y
3
are the shadowprices for the resources in the primal.) Increasing
the costs in factory 1 by 100 will therefore increase the maximum in the primal by 100 20 = 2000.
(We assume that the optimal point in the primal does not change.) Since the maximum in the primal is
the minimum in the dual, it follows that the minimum costs in the dual will increase by 2000 if the cost
per hour in factory 1 increases by 100.
2. (a) Regard the given LP problem as the primal and denote it by (P). Its dual is shown in answer section
and is denoted by (D). If you draw the feasible set for (D) and a line x
1
+ x
2
= c, you see that as c
increases, the line moves to the northwest, and the line that has the largest value of c and still has a point
in common with the feasible set is the point (0, 8), which is the solution to (D).
(b) We see that when x
1
= 0 and x
2
= 8, the second and fourth constraint in (D) are satised with strict
inequality, so in the optimum in (P), y
2
= y
4
= 0. Also, since x
2
= 8 > 0, the second constraint in (P) is
in optimum satised with equality, i.e. 2y
1
y
3
= 1. But then we see that the objective function in (P)
is 16y
1
+6y
2
8y
3
15y
4
= 16y
1
8y
3
= 8(2y
1
y
3
) = 8. If we put y
3
= b, y
1
=
1
2
(1 +b), and we
conclude that (y
1
, y
2
, y
3
, y
4
) = (
1
2
(1+b), 0, b, 0) must be a solution of (P) provided b is chosen such that
y
1
=
1
2
(1 +b) 0, i.e. b 1, and y
3
= b 0, and the rst constraints in (P) is satised. (The second
constraint we already knowis satised with equality.) The rst constraint reduces to
1
2
(1+b)2b 1,
or b
1
5
. We conclude that (
1
2
(1 +b), 0, b, 0) is optimal provided 0 b
1
5
.
(c) The objective function in (D) is now kx
1
+x
2
. If k 0, there is no solution. The condition for (0, 8)
to be the solution is that k is less or equal to the slope of the constraint x
1
+2x
2
= 16, i.e. k 1/2.
4. (a) See the text. (b) With the Lagrangian L = (500 ax
1
)x
1
+250x
2
1
(0.04x
1
+0.03x
2
100)
2
(0.025x
1
+0.05x
2
100)
3
(0.05x
1
100)
4
(0.08x
2
100), the KuhnTucker conditions (with
nonnegativity constraints) are: there exist numbers
1
,
2
,
3
, and
4
, such that
L/x
1
= 500 2ax
1
0.04
1
0.025
2
0.05
3
0 (= 0 if x
1
> 0) (i)
L/x
2
= 250 0.03
1
0.05
2
0.08
4
0 (= 0 if x
2
> 0) (ii)
1
0, and
1
= 0 if 0.04x
1
+0.03x
2
< 100 (iii)
2
0, and
2
= 0 if 0.025x
1
+0.05x
2
< 100 (iv)
3
0, and
3
= 0 if 0.05x
1
< 100 (v)
4
0, and
4
= 0 if 0.08x
2
< 100 (vi)
(c) The KuhnTucker conditions are sufcient for optimality since the Lagrangian is easily seen to be
concave in (x
1
, x
2
) for a 0. If (x
1
, x
2
) = (2000, 2000/3) is optimal, then (i) and (ii) are satised
with equality. Moreover, the inequalities in (iv) and (vi) are strict when x
1
= 2000 and x
2
= 2000/3,
so
2
=
4
= 0. Then (ii) gives
1
= 25000/3. It remains to check for which values of a that
3
0.
From (i), 0.05
3
= 500 4000a 0.04(25000/3) = 500/3 4000a 0 iff a 1/24.

Answers Manual

Uploaded by

Copyright:

Available Formats

Answers Manual

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Answers Manual

Uploaded by

Copyright:

Available Formats

Students Manual

Essential Mathematics for

2) = 0. (g) True. (h) True.

pq = 3q +5 yields pq = (3q +5)

x +5 9, then also (ii) x 4 = (

16 = 4, and the RHS

x +5, then just as in part (a) we nd that x must be a solution

x +5, on the other hand, intersects y =

3 1.73, whereas f (2) +f (1) =

+28 = 100, and so x

6, and compute the corresponding values of y. Argue

(0) = 3. The slope of

(3) = 1/3. (d) [f (h) f (0)]/h = (h

(5). On the other hand, f

(x) = 6x 12 = 6(x 2) 0 x 2, so f is increasing in [2, ). (b) f

3), so (using a sign diagram) f is increasing in

2]. (d) See the text.

= 1 2x = 3 when x = 1, so the slope of the tangent is 3. Since y = 1 when x = 1,

2. A sign diagram shows that f (x) > 0

(x) = f (x)(ln(2x) +1) = (2x)

x 1) ln x. Then f (1) = 0 and f

x 1)/x, which is positive for x > 1. The conclusion follows.

= 6x = 6 at x = 1, so y (3) = (6)(x 1), or y = 6x +3.

x 2x = 31/4 at x = 4, so y = (31/4)x +17.

= (1 +3y)/(1 3x) and solving for y

/(1 3x) = 30/(1 3x)

= 0, and solve for y

y = 0, and solve for y

, and solve for y

, and solve for y

, and solve for y

), and solve for y

. (How did we differentiate f (x

where u is a product of two functions that both

), and solve for y

. (b) Note that y

6. This yields the four points

(0) = 1. Then f (x) f (0) +f

(x) = 5(1 +x)

(0) = 5. Then f (x) f (0) +f

(1) = A. Then F(K) F(1) +F

. For x = 1, y = 0 this reduces to 3 2y

= 3/2. (b) y(x) y(1) +y

(x) tends to 0 as x tends to a.)

(x) = 2/(2x +4), so f

(0) = 1/2. Then f (x) f (0) +f

(x) = (1/2)(1 +x)

(0) = 1/2. Then g(x) g(0) +g

(0) = 1. Then h(x) h(0) +h

= 1/2. Differentiating () w.r.t, x yields

d and the numerator to 0, so the limit does not exist

> 0 everywhere, this is minimum point. (b) y

(x) < 0 if x < x

(x) > 0 if x > x

(Q) = 2Q+(ak) = 0 for Q

(Q) < 0. The monopoly prot is (Q

, as in Example 3. (c) The new prot function is

(Q) = 2Q+a k +s = 0 when

3. (e) f (0) = 0, f (3000) = 4.5 10

5/5. (c) (f (6) f (2))/4 = 1/6 and f