Section 5.

1 x Limits of Functions 52

This work is protected by United States copyright laws and is provided solely for
the use of instructors in teaching their courses and assessing student learning.
Dissemination or sale of any part of this work (including on the World Wide Web)
will destroy the integrity of the work and is not permitted. The work and materials
from it should never be made available to students except by instructors using
the accompanying text in their classes. All recipients of this work are expected to
abide by these restrictions and to honor the intended pedagogical purposes and
the needs of other instructors who rely on these materials.

Analysis
with an Introduction to Proof
5th Edition

by Steven R. Lay
[email protected]

Chapter 5 – Limits and Continuity

Solutions to Exercises

The following notations are used throughout this document:

 = the set of natural numbers {1, 2, 3, 4, …}
 = the set of rational numbers

= the set of real numbers
 = “for every”
 = “there exists”
† = “such that”

Section 5.1 – Limits of Functions

1. (a) False: Definition 5.1.1 requires 0  | x – c |  G .

(b) False: The neighborhoods are reversed.
(c) True: Theorem 5.1.8.
(d) True: Theorem 5.1.10.

Copyright © 2014 Pearson Education, Inc.

 Section 5.1 x Limits of Functions 53
2. (a) True: Example 5.1.14.
(b) False: This is true only if Q(c) z 0.
(c) False: The points are less than a.
(d) True: comments before the Review of Key Terms.

3. Answers in book: (a) 4; (b) 5/2; (c) 1/2; (d) –2; (e) 0; (f ) 1/4; (g)  4; (h) 2.

4. Let G = 1/28. Then for x such that | x – 2 | < G we have | x + 4 | = | x – 2 + 6 | d | x – 2 | + 6 < 1/28 + 6 < 7,
so that | x2 + 2x – 8 | = | x – 2 | | x + 4 | < (1/28)(7) = 1/4.

5. Hint in book: G = 1/24 will work.

Proof: Let G = 1/24. Then for x such that | x + 2 | < G we have | x – 5 | = | x +2 – 7 | d | x + 2 | + | –7 | < 1/24 + 7 < 8,
so that | x2 – 3x – 10 | = | x + 2 || x – 5 | < (1/24)(8) = 1/3.

6. (a) | x2  5x + 1 + 5 | = | x2  5x + 6 | = | x – 3 | | x  2 |. If | x  3 |  1, then | x – 2 | = | x  3 + 1 | d | x  3 | + 1  2.
Thus given H ! 0, let G = min {1, H/2}.
(b) | x2 + 3x + 8  8 | = | x2 + 3x | = | x | | x + 3 |. If | x + 3 |  1, then | x | = | x + 3  3 | d | x + 3 | + | 3 |  1 + 3 = 4.
Thus given H ! 0, let G = min {1, H/4}.
(c) | x3  8 | = | x  2 | | x2 + 2x + 4 |. If | x  2 |  1, then | x |  3 so that | x2 + 2x + 4 | d | x2 | + | 2x | + 4  9 + 6 + 4 = 19.
Thus given H ! 0, let G = min {1, H/19}.

7. (a) Given H !0, let G = H. Then if | x  0 |  G , we have | | x |  0| = | x | = | x  0 |  G = H. Thus limx o 0 | x | = 0.

x2 x2
(b) If x z 0, then x . Thus lim x o 0 0 by part (a).
x x
(c) Hint in book: Let (sn) be a positive sequence converging to c and use Example 4.2.6.
Proof: Let (sn) be a positive sequence converging to c. Then by Example 4.2.6,
lim x o c x lim n of sn lim n of sn c.
8. (a) The triangle inequality implies that | | f (x) |  | L | | d | f (x)  L |, and the desired result follows immediately.
(b) Let sn o c with sn z c and sn  D  n. Then lim f (sn) = L, so that lim f sn L by Example 4.2.6.

Hence lim x o c f x L.

9. Hint in book: (a) and (b) do not exist; in (c), the limit is 0.
(a) Let f (x) = 1/x and let sn = 1/n. Then sn o 0 with sn > 0  n, but lim f (sn) = lim n = + f, so f (sn) is divergent.
Thus by Theorem 5.1.10, limx o 0+ f (x) does not exist.
(b) Use the same sequence as in Example 5.1.11 to show that the limit does not exist.
(c) Given H ! 0, if | x |  H, then x sin 1x x ˜ sin 1x d x  H . Thus the limit is 0.
10. (a) Suppose L1 and L2 are both limits of f at c. Then given any H ! 0,  G ! 0 † | f (x) – L1 |  H/2 and | f (x) – L2 | 
H/2 whenever 0  | x  c |  G and x  D. But then
L1  L2 d L1  f ( x)  f ( x)  L2  H2  H2 H.
Since this holds for all H ! 0, we must have L1 = L2.
(b) Suppose L1 and L2 are both limits of f at c. Suppose sn o c with sn  D and sn z c  n. Then by Theorem
5.1.8, lim f (sn) = L1 and lim f (sn) = L2. By Theorem 4.1.14, L1 = L2.

11. Hint in book: To prove that (a) º (b), suppose that (b) is false. Let (sn) be a sequence in D with sn o c and sn z c
for all n. Since (b) is false, ( f (sn)) converges to some value, say L. Before we can use Theorem 5.1.8, we must
show that given any sequence (tn) in D with tn o c and tn z c for all n, we have lim f (tn) = L. We only know
from the negation of (b) that ( f (tn)) is convergent. To see that lim f (tn) = L, consider the sequence (un) = (s1, t1,
s2, t2, …) and note that ( f (sn)) and ( f (tn)) are both subsequences of ( f (un)).

Copyright © 2014 Pearson Education, Inc.

 Section 5.1 x Limits of Functions 54
Proof: It follows that lim f (un) exists by the negation of (b). Since ( f (sn)) and ( f (tn)) are subsequences of
( f (un)), we must have lim f (sn) = lim f (tn) = lim f (un) = L. Now by Theorem 5.1.8 we conclude that lim x o c
f (x) = L, which contradicts (a). To prove (b) º (a), suppose (a) is false. Then f has a limit at c. The negation of
(b) follows from Theorem 5.1.8 directly.

12. The parts of this proof are all straightforward.

13. Hint in book: Use Exercise 4.1.13.

14. Let sn o c with sn  D and sn z c  n. Let lim x o c f (x) = L. Then lim n o f f (sn) = L. But a d f (sn) d b  n, so
Exercise 4.2.18 implies that a d lim n o f f (sn) d b. Hence a d lim x o c f (x) d b.

15. Hint in book: Use Exercises 4.1.12 and 4.1.16.

Proof: Let sn o c with sn  D and sn z c  n. Exercise 4.1.16 implies that  N   † sn  U whenever n > N.
Thus the sequence (g(sn)) is bounded by max {M, | g(s1) |, …, | g(sN) | }. Since lim f (sn) = 0, Exercise 4.1.12 implies
that lim f (sn) g(sn) = 0. Thus lim x o c ( f g)(x) exists and is equal to 0.

16. Let lim x o c f (x) = L. Then L > 0, so that V = 1

2
L, 32 L is a neighborhood of L (of radius 12 L). Thus by Theorem

5.1.2,  a deleted neighborhood U of c † f (x)  V whenever x  U ˆ D. But if f (x)  V, then f (x) ! 0, since L ! 0.

17. Hint in book: To show that limx o c f (x) does not exist for c z 0, look at a sequence of rational numbers
converging to c. Then look at a sequence of irrational numbers converging to c.
Proof: Claim lim x o 0 f (x) = 0. Given H ! 0, if | x – 0 |  0, then | f (x) – 0 | d | x |  H. Thus f has a limit at c if
c = 0. Conversely, suppose that c z 0 and that lim x o c f (x) exists. Say lim x o c f (x) = L. Let (sn) be a sequence of
rationals † sn o c. Then L = lim f (sn) = lim sn = c. Now let (tn) be a sequence of irrationals † tn o c. Then L =
lim f (tn) = lim 0 = 0. This contradicts c z 0.

18. Let lim x o c f (x) = L and let H = 1. Then V = (L – 1, L + 1) is a neighborhood of L. By Theorem 5.1.2  a deleted
neighborhood U* of c † f (x)  V whenever x  U* ˆ D. If c  D, let M = max {| L – 1|, | L + 1|, f (c)}. If c  D,
let M = max {| L – 1|, | L + 1|}. Then let U = U*‰{c}. It follows that | f (x)| d M for all x  U ˆ D.

19. Hint in book: First show that f (n x) = n f (x) for all n   and all x  , so that f ( y/n) = f ( y)/n for all n  



and all y  . Then show that if lim x o 0 f (x) exists, it is equal to 0. Also show that f (x) = f (x – c) + f (c), so that
f (x)  f (c) = f (x – c), for all x, c  .
Proof: We have f (2x) = f (x) + f (x) = 2 f (x). Then by induction, f (nx) = n f (x). Let nx = y. Thus f ( y) =
n f ( y/n) and f ( y/n) = f ( y)/n. In particular, f (1/n) = f (1)/n. If lim x o 0 f (x) = L exists, then L = lim n o f f (1/n)
= lim n o f f (1)/n = 0. Now lim x o c [ f (x)  f (c)] = lim x o c f (x  c) = lim y o 0 f ( y) = 0. Hence lim x o c f (x) exists
and is equal to f (c).
The converse is trivial.

20. Suppose lim x o c+ f (x) = L and lim x o c f (x) = L. Then given H ! 0  G 1 ! 0 and G 2 ! 0 such that c  x  c + G 1
implies | f (x)  L|  H and c  G 2  x  c implies | f (x)  L|  H. Let G = min {G 1, G 2}. Then 0  | x  c |  G implies
| f (x)  L|  H. Hence lim x o c f (x) = L. The converse is trivial.

21. Since lim f ( x) = L and L > 0, given any M > 0  G1 † f (x) > L/2 if 0 < |x – c| < G1. Likewise, G 2 † g(x) > 2M/L
xoc

§ L · § 2M ·
if 0 < |x – c| < G 2. Let G = min {G 1, G 2}. Then if 0 < |x – c| < G we have ( fg )( x) f ( x) g ( x) ! ¨ ¸ ¨ ¸ M.
© 2 ¹© L ¹

Copyright © 2014 Pearson Education, Inc.

 Section 5.2 x Continuous Functions 55
Section 5.2 – Continuous Functions

1. (a) True: Definition 5.2.1.

(b) False: The Dirichlet function (Example 5.2.8) is bounded on \ but not continuous anywhere.
(c) True: See the comment after Definition 5.2.1.
(d) False: Let xn = (1)n 2 and f (x) = x2. Then f (xn) = 4 for all n, so f (xn) o 4, but (xn) is divergent.
(e) False: f (U ˆ D) Ž V.

2. (a) True: Theorem 5.2.2(d).

(b) True: Example 5.2.3.
(c) False: Example 5.2.7.
(d) False: See the modified Dirichlet function.
(e) True: Theorem 5.2.12.

3. Answer in book: Define f (3) = 10.

4. Given H ! 0, let G = min {1, H/10}. Then if | x 3 |  G , we have | x – 3 |  1, so that | x + 6 | = | x – 3 + 9 | d

| x – 3 | + 9 10. Thus for | x 3 |  G we have
| f (x)  f (3) | = | x2 + 3x – 5 13 | = | x2 + 3x – 18 | = | x + 6 | ˜ | x – 3 | 10| x 3 |  10G d H.
Hence f is continuous at 3.

5. Hint in book: Adapt Example 5.2.8 so that the two parts of the function “come together” at one point.
Example: Define f (x) = x if x   and f (x) = x  . Then f is continuous only at 0. Or, use the function in
Exercise 5.1.17.

6. (a) True. Define g(x) = k  x  D. Then g is continuous since it is a polynomial (Example 5.2.3). But f g = k f, so
k f is continuous by Theorem 5.2.10.
(b) True. If f is continuous on D, then so is – f, by part (a). Since g = ( f + g) + (– f ), g is continuous by Theorem
5.2.10.
­0 if x 0
(c) False. Define g ( x) ® and f (x) = x so that f g (x) = 2x. Then f and f g are continuous on \, but g
¯ 2 if x z 0
is not continuous at x = 0.
­ 1 if x  _
(d) False. Define f ( x) ® . Then f 2 is continuous on \, but f is discontinuous everywhere.
¯ 1 if x  _
(e) False. Define f : (0, 1) o \ by f (x) = 1/x. Then f is not bounded on (0, 1).
­0 if x d 0 ­1 if x d 0
(f ) False. Define f ( x) ® and g ( x) ® . Then f and g are both discontinuous at 0, but the
¯1 if x ! 0 ¯0 if x ! 0
function ( f + g)(x) = 1  x is continuous on \.
(g) False. Same example as part (f ). We have f g(x) = 0  x, which is continuous on \.
­ x if x  [1,1]
(h) False. Define f ( x) g ( x)
® . Then (g q f )(x) = x  x is continuous on \, but f and g are
¯ x otherwise
both discontinuous at 1 and –1.

7. Hint in book: True.

Answer: Any sequence is a function f : o. Since  consists entirely of isolated points, f is automatically
continuous.

8. f (x) = 1 if | x | ! 1, f (x) = 0 if | x | 1, f (x) = 1/2 if x = 1, and f (1) is not defined. f is continuous  x except
x = r1.

Copyright © 2014 Pearson Education, Inc.

 Section 5.2 x Continuous Functions 56
9. Hint in book: f is also continuous at 3.
Answer: If xn o 1 and each xn is rational, then lim f (xn) = 5. If yn o 1 and each yn is irrational, then lim f ( yn) = 7.
Hence f is not continuous at 1. To prove that f is continuous at 2, let H ! 0 be given. Note that f (2) = 10. Let G =
min {1, H/5}. Then if | x – 2 |  G d 1, we have | x + 2 |  5. Thus if x is rational, then
| f (x)  f (2) | = | 5x 10 | = 5 | x 2 | < 5G d 5 (H/5) = H.
If x is irrational, then
| f (x)  f (2) | = | x2 + 6 10 | = | x2  4 | = | x + 2 | ˜ |x  2|  5 | x  2 |  5 (H/5) = H.
If we set 5x = x2 + 6, we get x2 –5x + 6 = 0, or (x 2)(x 3) = 0. We see that f is also continuous at x = 3.

10. (a) Let (xn) be a sequence in D converging to c. Since f is continuous at c, Theorem 5.2.2 implies that lim f (xn) =
f (c). Thus lim | f | (xn) = lim | f (xn) | = | f (c) | = | f | (c) by Exercise 4.1.9. Hence | f | is continuous at c.
­1 if x  _
(b) False. Define f ( x) ® . Then | f | is continuous on \, but f is continuous nowhere.
¯1 if x  _

11. Hint in book: Show that for all m, n  , max {m, n} 1

2
( m  n)  1
2
m  n . Do this by considering the two cases
m t n and m < n.

12. This is similar to Exercise 10, except it uses Example 4.2.6.

13. Hint in book: Let D 1

2
f (c).
Solution: Since f (c) ! 0, D 12 f (c) ! 0. Let V = ( f (c) – D, f (c) + D). Since f is continuous at c,
Theorem 5.2.2 implies that  a neighborhood U of c † f (U ˆ D) Ž V. But then f (x) > D ,  x  U ˆ D.

14. If c is an accumulation point of D, then the result follows from Theorem 5.2.2 and Exercise 5.1.18. If c is an
isolated point of D,  a neighborhood U of c † U ˆ D = {c}. In this case, let M = | f (c) |.

15. If x  H ˆ D, then x  U (c) for some c in f –1 (G ). Thus f (x)  G since f (U (c) ˆ D) Ž V Ž G, and x  f –1 (G ).
Conversely, if x  f –1 (G ), then x  U (x) Ž H. Also, since x  f –1 (G ), x  D. Thus x  H ˆ D.

16. Suppose f is continuous on  and let H be a closed set. Then I\H is open by Theorem 3.4.7. By Corollary 5.2.15,
f 1(I\H ) is open. But f 1(I\H ) = I\ f 1(H ) by Theorem 2.3.16 (g). Thus \ f 1(H ) is open and f 1(H ) is
closed by Theorem 3.4.7.
The converse is similar.

17. Hint in book: See the hint for Exercise 5.1.19. Show that for any rational number m/n, f (m/n) = (m/n) f (1).
Solution: From the answer to Exercise 5.1.19 we have f m
n m ˜ f 1n m ˜ f (1). So let k = f (1). Given any
n
x  , let (xn) be a sequence of rationals converging to x. Then f (x) = lim f (xn) = lim [xn f (1)] = [ f (1)] lim xn = k x.

18. Given x  (a, b), let (xn) be a sequence of rationals in (a, b) converging to x. Since f is continuous at x, f (x) = lim
f (xn) = lim n o f 0 = 0.

19. (a) If c = 1, then s1 = 1 = sn +1  n, and (sn) is constant and hence increasing. If c ! 1, we use induction to establish
sn > sn +1  n. We have s2 c s1 c c ! c1 c s1 . Now suppose sk > sk +1 for some k  `. Since c ! 1, it
sk 1 sk
follows that sk  2 c !c sk 1 . Thus sn > sn +1 for all n and (sn) is an increasing sequence.
(b) We claim that sn d 3  n. Clearly, s1 c  3 3  3 . Now suppose sk d 3 for some k  `. Then since c d 3 3 ,
sk 3
we have sk 1 c d c3 d 3
3 3 . Hence by induction, sn d 3  n  `.

Copyright © 2014 Pearson Education, Inc.

 Section 5.2 x Continuous Functions 57
(c) From the Monotone Convergence Theorem (4.3.3) we know that (sn) converges to some number, say b. Since
x
sn 1 c sn and f (x) = c is continuous, we have
b lim sn 1 lim c sn cb .
n of n of
b
(d) The given continued power is the sequence when c 2  3 3 . It’s limit b must satisfy b 2 , that is,
b2 2b . This implies b = 2 or b = 4. But from part (b) we know that b d 3. Thus b = 2.

Section 5.3 – Properties of Continuous Functions

1. (a) True: Theorem 5.3.2.

(b) False: Corollary 5.3.3 requires D to be compact. For example, let f (x) = x2 and D = [0,2).
(c) False: Practice 5.3.1.

2. (a) True: Intermediate Value Theorem (5.3.6).

(b) True: Intermediate Value Theorem (5.3.6). If k = f (a), then c = a.
(c) False: Practice 5.3.4.

3. Hint in book: Only (g) and (i) are true. Note that (g) is true for all functions. To prove (i), first show that a
nonempty subset S of  is an interval iff [a, b] Ž S whenever a, b  S and a  b. Do this by considering cases
depending on whether or not S is bounded above or bounded below.
Here are some possible counterexamples for the other parts:
(a) f (x) = x2 maps (1, 1) onto [0, 1); or f (x) = 1  x maps (0, 1) onto {1}.
(b) f (x) = 1/(1 + x2) maps [0, f) onto (0, 1]; or f (x) = 1/x maps [1, f) onto (0, 1].
(c) f (x) = x2 maps {1} * (0, 2) onto (0, 4).
(d) f (x) = x2 maps (1, 2] onto [0, 4].
(e) f (x) = x /(1 + x2) maps [0, f) onto [0, 1]; or same as (d).
(f) f (x) = 1  x maps  onto {1}; or f (x) = 1/x maps  into [0, 1].
(h) f (x) = 1  x maps  onto {1}.
(j) f (x) = (x2 sin x)/(1 + x2) maps [0, f) onto (1, 1).
x
4. Let f (x) = 5x  3 . Then f is continuous, f (0) = 1, and f (1) = 2. Hence by Theorem 5.3.6,  c  (0, 1) † f (c) = 0.
c
That is 2 = 3 c.
x
5. Let f (x) = 5  x4. Then f is continuous, f (1) =  4/5, and f (0) = 1. Hence by Theorem 5.3.6,  c  (1, 0) †
c
f (c) = 0. That is, 5 = c4.

6. Let p(x) = ao + a1x + … + an xn and suppose an ! 0. Then lim x o f p (x) = + f and lim x o  f p(x) =  f. (This may be
verified by using the triangle inequality.) In particular,  x † p (x) > 0 and  y † p ( y) < 0. Thus by Theorem 5.3.6,
 z between x and y † p(z) = 0.

7. Hint in book: Consider the function g (x) = f (x) – x.

The proof is routine.

8. Let h (x) = f (x)  g (x). Then h (a) d 0 and h (b) t 0. The Intermediate Value Theorem (5.3.6) implies that  c 
[a, b] † h (c) = 0. That is f (c) = g (c).

9. Theorem 5.3.10 implies that f ([a, b]) is a compact interval in \. If f ([a, b]) contained more than one point, it
would have to contain irrational points, a contradiction. Thus f ([a, b]) is a single point and f is constant on [a, b].

Copyright © 2014 Pearson Education, Inc.

 Section 5.3 x Properties of Continuous Functions 58

­ x  12 if x  [ 12 ,1],
°°
10. (a) Let f ( x) 1 if x 1 and n 3, 4, 5, ! ,
® n 1 n
°
°̄ x otherwise.
Then f is bijective from [1/2, 1] onto [0, 1/2] and f is bijective from [0, 1/2) onto [0, 1/2].
(b) Suppose f were continuous. Then f attains its maximum at two distinct points, say M1 and M2. Likewise, f
attains its minimum at two distinct points, say m1 and m2. Consider the possible orderings of the m’s and the
M ’s. Each leads to the existence of some value that is attained three times.

11. (a) From the triangle inequality | a + b | d | a | + | b |, we have | a + b | – | a | d | b |. Let a = x – c and b = c – p, so that
a + b = x – p. Thus we have | x – p | – | c – p | d | x – c |. Similarly, by letting a = c – x and b = x – p so that
a + b = c – p, we obtain | c – p | – | x – p | d | c – x | = | x – c |. Combining these inequalities we have
| | x – p | – | c – p | | d | x – c |. It follows that | f (x) – f (c)| = | | x – p | – | c – p | | d | x – c |, and continuity follows
directly with G = H.
(b) Define the function f : \ o \ as in part (a). Since S is compact, the continuous function f has a minimum
on S. That is, there exists a point q in S such that f ( q) d f (x) for all x  S. Then | q – p | d | x – p | for all x  S.
Since q  S, this means | q – p | = inf { | x – p |: x  S}.

12. From the HeineBorel theorem it suffices to show that f (D) is closed and bounded. To this end, suppose that f (D)
is not bounded. Then for each n   there exists a point xn  D such that | f (xn)| ! n. Since D is bounded, the
BolzanoWeierstrass theorem for sequences (4.4.7) implies that the sequence (xn) has a subsequence (xnk)
converging to some x0  . Since D is closed, we must have x0  D. (See Exercise 4.1.15.) Since f is therefore
continuous at x0, f (xnk) converges to f (x0). In particular, f (xnk) must be bounded. But this contradicts the fact
that | f (xnk)| > nk t k for all k  . We conclude that f (D) is bounded.
It remains to show that f (D) is closed. Let ( yn) be a convergent sequence in f (D) and let y0 = lim yn. It
suffices (Exercise 4.1.15) to show that y0  f (D). Since yn  f (D) for each n, there exists a sequence (xn) in D such
that f (xn) = yn for all n. Since D is bounded and closed, it follows as in the first part of the proof that there exists
a subsequence (xnk) converging to some point x0 in D. Since f is continuous at x0, we have
f ( x0 ) lim f ( xn ) lim yn . But ( ynk) is a subsequence of ( yn), so lim k o f ynk = y0. Thus f (x0) = y0 and
k of k k of k

y0  f (D).

13. (a) Suppose f is continuous and not monotone. Then  x, y, z  I † x  y  z and either f (x)  f ( y) and f ( y) !
f (z) or f (x) ! f ( y) and f ( y)  f (z). Suppose, for example, that f (x)  f (z)  f ( y). Then by the Intermediate
Value Theorem  w  (x, y) † f (w) = f (z). This contradicts f being injective. The other cases are similar.
(b) Let c  I and suppose c is not an endpoint of I. (If c is an endpoint, then the proof must be altered slightly.) It
follows that f (c) is not an endpoint of f (I ), so  r ! 0 † ( f (c)  r, f (c) + r) Ž f (I ). Given any H !0, let D =
min {H, r/2}. Since f (I ) is an interval,  y and z in I † f ( y) = f (c)  D and f (z) = f (c) + D . Since f is strictly
increasing we have y  c  z. Furthermore, if x  ( y, z), then f ( y)  f (x)  f (z). That is, if x  ( y, z), then
| f (x)  f (c)|  D d H . Thus we let G = min {c  y, z  c}. Then | x  c |  G implies that x  ( y, z) so that
| f (x)  f (c)| H . Hence f is continuous at c.
The strictly increasing property of f 1 follows easily, and this in turn implies that f 1 is continuous.

14. (a) Example 5.1.11 shows that lim x o 0 f (x) does not exist. Hence f cannot be continuous at 0.
(b) If 0  x  y, then the intermediate value property for x and y follows from the continuity of f on [x, y]. The
case x  y  0 is similar. Note that f assumes every value between 1 and 1 in any neighborhood of 0. If
x  0  y, let H = min {x, y}. Then f assumes every value between f (x) and f ( y) in the neighborhood (H , H)
of 0. If 0 = x  y, then  z  (x, y) † f (z) = 0. Then the intermediate value property follows from the
continuity of f on [z, y]. The case x  y = 0 is similar.

Copyright © 2014 Pearson Education, Inc.

 Section 5.3 x Properties of Continuous Functions 59
15. (a) Hint in book: For every x  D, there exists a neighborhood Ux of x and a number Mx such that f is bounded
by Mx on Ux. Now {Ux: x  D} is an open cover for D. Since D is compact, there exist x1, …, xn in D such
that D Ž Ux1 ‰ … ‰ Uxn . Let M = max {Mx1, …, Mxn}. Then f is bounded by M on D.
(b) Let f (x) = 1/x and D = (0, f).
(c) Follows from part (a) and Exercise 5.1.18.

16. We shall prove the contrapositive. Suppose f (S) is not connected. Then  disjoint open sets U and V † f (S) Ž
U ‰ V, f (S) ˆ U z ‡, and f (S) ˆ V z ‡. By Corollary 5.2.15, f 1(U ) and f 1(V ) are open sets, and they clearly
must be disjoint (since U and V are disjoint). But S Ž f 1(U ) ‰ f 1(V ), S ˆ f 1(U ) z ‡, and S ˆ f 1(V ) z ‡.
Hence S is not connected.

Section 5.4 – Uniform Continuity

1. (a) False: The H and the G have been reversed.

(b) False: Technically, f is uniformly continuous on D, but we never speak of uniform continuity at a point.
(c) True: Theorem 5.4.6.

2. (a) True: Definition 5.4.1. (b) False: Practice 5.4.7.

(c) True: It is uniformly continuous on [a, b] by Theorem 5.4.9, and hence also on the subset (a, b).

3. Hint in book: (a), (c), (d), (g), and (h) are uniformly continuous.
(a) Yes, because [2, 5] is compact. (b) No, because limx o 0+ ex/x = + f.
(c) Yes, because [0, 5] is compact. (d) Yes, by part (c) and because (1, 4) Ž [0, 5].
(e) No, because limx o 0+ 1/x2 = + f. (f ) No, by part (e) and because (0, 1) Ž (0, f).
(g) Yes. Define f (2) 4 and f (4) 6. (h) Yes. Define f (0) 0 and f (1) sin1.

4. (a) | x3  y3 | = | x  y | ˜ | x2 + xy + y2 | d 12 | x  y | for x, y  [0, 2].

1 1 yx 1 1
(b)  x y d x  y for x, y  [2, f).
x y xy xy 4
x 1 y 1 2 x y
(c)  d 2 x  y for x, y  [0, f).
x 1 y 1 ( x  1)( y  1)

5. Hint in book: Note that f ( x) x is uniformly continuous on [0, 2] by Theorem 5.4.6. Then prove f is
uniformly continuous on [1, f) using Definition 5.4.1. Given H! 0, let G 1 be the G that works on [0, 2] and let G 2
be the G that works on [1, f). Then let G 3 = min {1, G 1, G 2}. Show that G 3 satisfies Definition 5.4.1 on [0, f).

6. Given H ! 0  G 1 ! 0 † | f (x)  f ( y) |  H/2 whenever | x  y |  G 1 and x, y  D. Likewise  G 2 ! 0 † | g (x)  g ( y) | 

H/2 whenever | x  y |  G 2 and x, y  D. Let G = min {G 1, G 2}. Then for x, y  D with | x  y |  G we have
| ( f + g)(x)  ( f + g)( y) | d | f (x)  f ( y) | + | g (x)  g ( y) | < H/2 + H/2 = H. Hence f + g is uniformly continuous
on D.

7. Let H ! 0 be given. If k z 0,  G ! 0 † | f (x)  f ( y) | < H/| k | whenever | x  y |  G and x, y  D. Thus for x, y  D

H
with | x  y |  G we have | (k f )(x)  (k f )( y)| = | k | ˜ | f (x)  f ( y)|  k ˜ = H. If k = 0, then  G ! 0 and all
k
x, y  D, it follows that | (k f )(x)  (k f )( y) | = 0  H.

8. (a) Let f (x) = 1 and g (x) = x for x  (0,1). Then ( f /g)(x) = 1/x is not uniformly continuous on (0, 1).
(b) f /g is continuous on D by Theorem 5.2.10. Since D is compact, f /g must be uniformly continuous on D
(Theorem 5.4.6).

Copyright © 2014 Pearson Education, Inc.

 Section 5.4 x Uniform Continuity 60
9. Hint in book: True.
Proof: Suppose f : A o B and g: B o C are uniformly continuous. Given any H ! 0  G 1 ! 0 † | g (u)  g (v)|  H
whenever u, v  B and | u v |  G 1. And  G 2 ! 0 † | f (x)  f ( y) | < G 1 whenever x, y  A and | x  y |  G 2. Thus
when x, y  A and | x  y |  G 2 we have f (x), f ( y)  B and | f (x)  f ( y)|  G 1, so that | g( f (x))  g( f ( y))|  H . Thus
g q f : A o C is uniformly continuous.

10. Let f (x) = g (x) = x,  x  I. Then f and g are uniformly continuous on , but their product ( f g)(x) = x2 is not.
See Examples 5.4.2 and 5.4.4.

11. Hint in book: Suppose that f (D) is not bounded. Then there exists a sequence (sn) in D such that | f (x) | t n for
each n. Now use Theorems 4.4.6 and 5.4.8.

12. (a) Since f is bounded on D, there exists M1 > 0 such that | f (x)| d M1 for all x  D. Since g is bounded on D, there
exists M 2 > 0 such that | g (x)| d M 2 for all x  D. Let H > 0 be given. Since f is uniformly continuous on D,
H
there exists G 1 > 0 such that f ( x)  f ( y )  whenever | x – y | < G 1 and x, y  D. Likewise, since g is
2M 2
H
uniformly continuous on D, there exists G 2 > 0 such that g ( x)  g ( y )  whenever |x – y| < G 2 and x, y  D.
2M1
Let G = min {G 1, G 2}. Then whenever | x – y | < G and x, y  D, we have
| f (x) g(x) – f ( y) g( y)| = | f (x) g(x) – f (x) g( y) + f (x) g( y) – f ( y) g( y) |
d | f (x) g(x) – f (x) g( y)| + | f (x) g( y) – f ( y) g( y) |
= | f (x)| | g(x) – g( y)| + | g( y)| | f (x) – f ( y)|
d M 1 | g(x) – g( y)| + M 2 | f (x) – f ( y)|
§ H · § H · H H
 M1 ¨ ¸  M2 ¨ ¸  H.
© 2 M 1 ¹ © 2M 2 ¹ 2 2
Thus f g is uniformly continuous on D.
(b) Apply Exercise 11 to conclude that f and g are bounded on D. Then apply part (a).

13. Given H ! 0,  G 1 ! 0 † | f (x) – f ( y)| < H/2 whenever x, y  [a, b] and | x – y |  G 1. Likewise,  G 2 ! 0 †
| f (x) – f ( y)| < H/2 whenever x, y  [b, c] and | x – y |  G 2. Let G = min {G 1, G 2}. Then given any x, y  [a, c]
such that | x – y |  G , if x, y  [a, b] or x, y  [b, c] we are done.
Otherwise, suppose without loss of generality that x  b  y and | x – y |  G . Then we have | x – b |  G 1 and
| b – y |  G 2 so that | f (x) – f ( y)| d | f (x) – f ( b)| + | f (b) – f ( y)| < H/2 + H/2 = H.

14. (a) Follows from Definition 5.4.1.

(b) Since D is compact, D is bounded and Theorem 4.4.7 applies. Since D is closed, x  D.
(c) ynk  x d ynk  xnk  xnk  x o 0.

(d) Since f is continuous at x, Theorem 5.2.2 implies that lim k of f ( xnk ) lim k of f ( ynk ) . This contradicts

f ( xnk )  f ( ynk ) t H .

15. Since the interval [k, k] is compact, f is uniformly continuous on [k, k]. Thus given any H ! 0  G 1 ! 0 †
| f (x)  f ( y)|  H whenever x, y  [k, k] and | x  y |  G 1. Now let G = min {G 1, k/2} and suppose | x  y |  G .
Now  an integer n † x  n k  [ k/2, k/2]. But then y  n k  [ k, k] since | x  y |  G d k/2. Thus | f (x)  f ( y) | =
| f (x + n k)  f ( y +n k)|  H, as required. Thus f is uniformly continuous on .
Since [0, k] is compact, f is bounded on [0, k], say by M. But then  x    an integer n † x + n k  [0, k].
Thus | f (x) | = | f (x + n k) | d M.

Copyright © 2014 Pearson Education, Inc.

 Section 5.5 x Continuity in Metric Spaces 61
Section 5.5 – Continuity in Metric Spaces

1. (a) True: comment before Example 5.5.2.

(b) True: Example 5.5.4.
(c) False: Let X1 = X2 = , let xn = (1)n 2, and let f (x) = x 2. Then f (xn) = 4 for all n, so f (xn) o 4, but (xn) is
divergent.

2. (a) False: See Example 5.5.6. For example, let X1 = X2 = , let f (x) = 3 for all x, and let U = (1, 2). Then U is
open in , but f (U ) = {3} is not open in .
(b) True: Theorem 5.5.9.
(c) True: Theorem 5.5.14.

3. Hints in book: (a) sn o (0, 6); (c) sn o (0, 3).

1
Additional answers: (a) sn o (0, 6). d(sn, (0, 6)) = o 0.
4n
1 1 2
(b) sn o (0, 0). d(sn, (0, 0)) d  d 2 o 0.
n 2 n3 n
1 2 3
(c) sn o (0, 3). d(sn, (0, 3)) d  d o 0.
n  2 n 1 n 1
1 5 6
(d) sn o (0, 2). d(sn, (0, 2)) d  d o 0.
n 2n  1 2n  1

4. Let (X, d) be a metric space, let p  X, and let N( p; H) be an arbitrary neighborhood of p. Define f : X o  by
f (x) = d (x, p). By Example 5.5.4, f is continuous. Now N( p; H) = f –1(I ), where I = (1, H). Since I is open in ,
Theorem 5.5.7 implies that N( p; H) is open in X.

5. Hint in book: The proof is similar to the proof of Theorem 5.2.14, but it is simpler since the statement of Theorem
5.5.5(c) is less cumbersome than the statement of Theorem 5.2.2(c). This comes from having the domain of f be
equal to X1 and not just a subset of X1.
Proof: Suppose f is continuous on X1 and let G be an open subset of X2. We must show that every point in
f – 1 (G ) is an interior point of f – 1 (G ). To this end, let x  f – 1 (G ) so that f (x)  G. Since G is open, there exists a
neighborhood V of f (x) such that V Ž G. Since f is continuous at x, Theorem 5.5.5 implies the existence of a
neighborhood U of x such that f (U ) Ž V. But f (U ) Ž V Ž G implies that U Ž f – 1 (G ). Hence x is an interior
point of f – 1 (G ).
Conversely, suppose f – 1(G ) is open in X1 whenever G is open in X2. Let x  X1 and let V be a neighborhood
of f (x). To show that f is continuous at x, it suffices by Theorem 5.5.5 to show that there exists a neighborhood U
of x such that f (U ) Ž V. Now since V is a neighborhood, V is open (Theorem 3.6.6). Hence f – 1(V ) is an open set
containing x. Thus there exists a neighborhood U of x such that U Ž f – 1(V ). But then f (U ) Ž V, so f is
continuous on X1.

6. Let F be an open cover of f (D). Since f is continuous on X1, Theorem 5.5.7 implies that, for each U  F, f – 1(U )
is open in X1. Furthermore, since f (D) Ž * F , it follows that D Ž * { f – 1(U ): U  F}. That is, { f – 1(U ): U  F }
is an open cover for D. Since D is compact, there exist finitely many sets U1, U2, …, Un in F such that
D Ž f 1 (U1 ) ‰ f 1(U 2 ) ‰ " ‰ f 1 (U n ).
Since f ( f (B )) Ž B for every B Ž X2, we must have
–1

f ( D ) Ž U1 ‰ U 2 ‰ " ‰ U n .
Thus {U1, U2, …, Un} is a finite subcover of F for f (D), and f (D) is compact.

7. From Theorem 5.5.9, f (D) is compact in . Lemma 3.5.4 implies that f (D) has a minimum, say m, and a
maximum, say M. Since m, M  f (D),  x1, x2  D † f (x1) = m and f (x2) = M. It follows that f (x1) d f (x) d f (x2)
 x  D.

Copyright © 2014 Pearson Education, Inc.

 Section 5.5 x Continuity in Metric Spaces 62
8. Let p  cl D. Then  a sequence (sn) with sn  D  n such that sn o p. Since sn  D  n, f (sn) = g(sn)  n. Thus
since f and g are both continuous on S, Theorem 5.5.5 implies that
f ( p) = lim f (sn) = lim g(sn) = g( p).

9. Hint in book: Use Theorems 5.5.7 and 5.5.9.

(a) Since f is bijective, ( f 1) 1 = f. Thus by Theorem 5.5.7 it suffices to show that f (U ) is open in X2 whenever
U is open in X1. Now if U is open in X1, then X1 \U is closed. Since X1 is compact, X1 \U is compact
(Exercise 3.6.12), and so by Theorem 5.5.9, f (X1 \U ) is also compact. Since f is bijective, f (X1 \U ) =
X2 \ f (U ). That is X2 \ f (U ) is compact. This implies that X2 \ f (U ) is closed and f (U ) is open.
(b) Define f (t) = (cos t, sin t). Then f : [0, 2S) o C is bijective and continuous. But f 1 is not continuous
at (1, 0) = f (0).

10. Let U be an open set in X3 and let h = g q f. We must show that h 1 ( U ) is open in X1. Since g is continuous on X,
g 1 (U ) is open in X2. Since f is continuous on X1, f 1 (g 1 (U )) is open in X1. But f 1(g 1 (U )) = (g q f ) 1(U ) =
h 1(U ), so by Theorem 5.5.7, h is continuous on X1.

11. Given (X1, d1), (X2, d2), and (X3, d3), let x, y  X2. Given H ! 0,  G ! 0 † d3(g(x), g( y))  H whenever d2(x, y)  G ,
since g is uniformly continuous on X2. Since f is uniformly continuous on X1,  J ! 0 † d2( f ( p), f (q)) < G
whenever p, q  X1 and d1( p, q)  J. Thus whenever d1(p, q)  J , we have d2( f ( p), f (q))  G and d3[g( f ( p)),
g( f (q))]  H. Hence g q f is uniformly continuous on X1.

12. (a) Since g is continuous and X2 is compact, Exercise 9(a) implies that g 1 is continuous. Thus f = g 1 q h is
continuous by Exercise 10.
(b) Since X2 is compact and g is continuous, X3 is compact. Thus g 1 is uniformly continuous, since its domain X3
is compact. But f = g 1 q h and h is uniformly continuous, so f is also uniformly continuous (Exercise 11).

13. (a) If x  cl D, then every neighborhood of x intersects D. Thus given any H ! 0  a point p0 in D † d(x, p0)  H.
Thus d(x, D) d H  H ! 0, and so d(x, D) d 0 by Theorem 3.2.6. But d(x, D) t 0, since it is the infimum of
nonnegative numbers. Thus d(x, D) = 0.
(b) In Example 5.5.4 we showed that the function f (x) = d(x, p) is continuous. By Corollary 5.5.10, this function
assumes its minimum on the compact set D.
(c) Hint in book: Follow the same strategy as in Example 5.5.4 to show that | d(x, D)  d( y, D)| d d(x, y) for all
x, y  X.
Proof: For all p  D we have d(x, D) d d(x, p) d d(x, y) + d( y, p). Thus d(x, D) d d(x, y) + d( y, D)
and d(x, D)  d( y, D) d d(x, y). The inequality d( y, D)  d(x, D) d d(x, y) follows in a similar way. Thus
| d(x, D)  d( y, D)| d d(x, y) for all x, y  S, so f (x) = d(x, D) is uniformly continuous as in Example 5.5.4 and
Practice 5.5.12.

14. (a) Let A = {(x1, x2): x1x2 = 1 and x1 ! 0} and B = {(x1, x2): x2 = 0}. Then A and B are disjoint and closed, but
d(A, B) = 0.
(b) Since A is closed and A ˆ B = ‡,  b  B  H b ! 0 † N(b; H b) ˆ A = ‡. Now F = {N (b; H b/2): b  B}
is an open cover of B, so  b1, …, bm in B † {N (bi ; H bi /2): i = 1, …, m} is a finite subcover of B. Let
H = min {H bi /2 : i = 1, …, m}. Then no point of A is within H of any point of B. Thus d(A, B) t H ! 0.

15. Hint in book: Prove the contrapositive.

Proof: We will establish the contrapositive. That is, suppose f (B) is disconnected. Then  disjoint open sets U and
V in X2 † f (B) Ž U ‰ V, f (B) ˆ U z ‡, and f (B) ˆ V z ‡. Let E = f 1 (U ) and F = f 1 (V). Then E and F are
open sets since f is continuous. Furthermore, E and F are disjoint since U and V are disjoint. We also have B Ž
E ‰ F, B ˆ E z ‡, and B ˆ F z ‡. Thus B is disconnected.

16. d(x  p, y p) = ([(x1 + p1)  ( y1 + p1)]2 + … + [(xn + pn)  (yn + pn)]2 )1/2
= ([x1  y1]2 + … + [xn  yn]2 )1/2 = d(x, y)

Copyright © 2014 Pearson Education, Inc.

 Section 5.5 x Continuity in Metric Spaces 63
17. This follows from Exercise 16.

18. d(Ox, Oy) = [(Ox1  Oy1)2 + … + (Oxn  Oyn)2 ]1/2

= (O2)1/2[(x1  y1)2 + … + (xn  yn)2]1/2 = | O | d(x, y).

19. If O = 0, then f is continuous by Example 5.5.6. If O z 0, then continuity follows from Exercise 18.

20. Define g(x) = Ox and h(x) = p  x. Then f = h q g. Since g and h are continuous, so is f.

21. (a) For any p  n and O z 0, the function f (x) = p + Ox is continuous by Exercise 20. Its inverse f 1 ( y) =
(1/O) p + (1/O) y is also continuous. Thus Theorem 5.5.7 implies that the original function f = ( f 1) 1 maps
open sets onto open sets.
(b) For O z 0, the function f (x) = p + Ox is injective. Thus f (n \A) = n \ f (A) for each A Ž n. If A is closed,
then n \A is open. Part (a) implies that f (n \A) = n \ f (A) is open, so f (A) is closed.
(c) If p  A  B, then  a0  A and b0  B † p = a0 + b0. But then p  a0 + B, so that p  *a  A (a + B).
Conversely, if p  *a  A (a + B), then  a0  A † p  a0 + B. But then  b0  B † p = a0 + b0. Thus
p  A + B. The second equality is similar.
(d) If A is open, then A + B = *b  B (A + b) is open, since it is a union of open sets.
(e) False. Let A = {(x1, x2): x1 x2 = 1 and x1 ! 0} and B = {(x1, x2): x2 = 0}. Then A + B = {(x1, x2): x2 ! 0} is not
closed, even though A and B are both closed.

Copyright © 2014 Pearson Education, Inc.