Kenneth A. Ross
Selected Solutions
Angelo Christopher Limnios
EXERCISE 1.2
Claim:
P(n) = 3 + 11 + + (8n 5) = 4n
2
n n N
Proof :
By induction. Let n = 1. Then 3 = 4(1)
2
(1) = 3, which will serve as the
induction basis. Now for the induction step, we will assume P(n) holds true
and we need to show that P(n +1) holds true. P(n +1) = 3 +11 + +(8n
5) + (8(n + 1) 5) = 4(n + 1)
2
(n + 1).
Now, 3 + 11 + + (8n 5) + (8(n + 1) 5) = 4n
2
n + 8(n + 1) 5. So in
order to show that P(n+1) is true, we need to show that 4(n+1)
2
(n+1) =
4n
2
n + 8(n + 1) 5, which, when evaluated is true, as desired.
EXERCISE 1.3
Claim:
P(n) = 1
3
+ 2
3
+ +n
3
= (1 + 2 + +n)
2
n N
Proof :
By induction. Let n = 1. Then, P(1) = 1
3
= 1 = 1
2
, which will serve as the
induction basis. Now for the induction step, we will assume P(n) holds true
and we need to show that P(n + 1) holds true.
P(n + 1) = 1
3
+ 2
3
+ +n
3
+ (n + 1)
3
= (1 + 2 + +n)
2
+ (n + 1)
3
= (1 + 2 + +n)
2
+ [(n + 1)
3
(n + 1)
2
] + (n + 1)
2
= (1 + 2 + +n)
2
+ (n + 1)
2
[n + 1 1] + (n + 1)
2
= (1 + 2 + +n)
2
+ 2(n + 1)n(n + 1)/2 + (n + 1)
2
= (1 + 2 + +n)
2
+ 2(n + 1)(1 + 2 + +n) + (n + 1)
2
=
1
(1 + 2 + + (n + 1))
2
,
as desired.
1
(a + b)
2
= a
2
+ 2ab + b
2
, where a = 1 + 2 + + n and b = n + 1.
1
EXERCISE 1.4
(a) For
n = 1, the expression is 1.
n = 2, the expression is 4.
n = 3, the expression is 9.
n = 4, the expression is 16.
We note that 1 + 3 + (2n 1) = n
2
, which is the proposed formula.
(b) Claim:
P(n) = 1 + 3 + (2n 1) = n
2
n N
Proof :
By induction. We already showed the case when n = 1. Now for the induction
step, we will assume P(n) holds true and we need to show that P(n + 1) holds
true.
P(n + 1) = 1 + 3 + (2(n + 1) 1) = 1 + 3 + + (2n 1) + (2n + 1)
= n
2
+ 2n + 1 = (n + 1)
2
,
as desired.
EXERCISE 1.6
Claim:
P(n) = (11)
n
4
n
is divisible by 7 when n N
Proof :
By induction. P(1) = 11 4 = 7 is divisible by 7. Now for the induction step,
we will assume P(n) holds true and we need to show that P(n + 1) holds true.
P(n + 1) = (11)
n+1
4
n+1
= 11
n+1
4 11
n
+ 4 11
n
4
n+1
= 11
n
(11 4) + 4(11
n
4
n
) = 7 11
n
+ 4(11
n
4
n
).
Since we assumed that 11
n
4
n
was divisible by 7 because we assumed P(n)
was true, we can see that
11
n+1
4
n+1
is divisible by 7, as desired.
2
EXERCISE 1.9
(a) The inequality does not hold for n = 2, 3, 4. It holds true for all other n N.
(b) It is true by by inspection for n = 1 and 2
4
= 4
2
also holds for n = 4.
Implement the induction step. For n 4, if 2
n
n
2
, then 2
(n+1)
> (n + 1)
2
.
But 2
(n+1)
= 2 2
n
2n
2
> (n + 1)
2
i (n + 1) <
21
=
2 + 1, which includes n N : n 4.
EXERCISE 1.10
Claim:
(2n + 1) + (2n + 3) + (2n + 5) + + (4n 1) = 3n
2
n N.
Proof :
Using the results from 1.4, we can avoid induction. Observe that
(2n+1)+(2n+3)+ +(4n1) = (1+3+ +(4n1))(1+3+ +(2n1))
= (1 + 3 + + 2(2n) 1) (1 + 3 + + (2n 1))
= (2n)
2
n
2
= 4n
2
n
2
= 3n
2
,
as desired.
EXERCISE 1.12 (b) and (c)
(b) Observe the following:
1
k
+
1
n k + 1
=
(n k + 1) +k
k(n k + 1)
=
n + 1
k(n k + 1)
.
Hence,
n!
k!(n k)!
+
n!
(k 1)!(n k + 1)!
=
n!
(k 1)!(n k)!
_
1
k
+
1
n k + 1
_
=
n!
(k 1)!(n k)!
_
n + 1
k(n k + 1)
_
=
(n + 1)!
k!(n k + 1)!
.
(c) For n = 1 : (a +b)
n
= a +b =
_
1
1
_
a +
_
1
1
_
b. Now, for n 1, we get
(a +b)
n
=
n
k=0
_
n
k
_
a
k
b
nk
.
3
Hence,
(a +b)
(n+1)
= (a +b)
n
l=0
_
n
l
_
a
l
b
nl
=
n+1
k=0
__
n
k 1
_
+
_
n
k
__
a
k
b
nk+1
=
n+1
k=0
_
n + 1
k
_
a
k
b
n1+k
.
EXERCISE 2.1
Show that
3 is not rational (The exact same technique can be used to show
the other numbers are not rational).
Claim:
3 / R
Proof :
By Contradiction. If
3 R, then an r Q :
p
q
=
3, where r =
p
q
=
_
p
q
_
2
= 3 = p
2
= 3q
2
= p
2
is divisible by 3 = p is divisible by 3. Let p =
3k. Then 9k
2
= 3q
2
= 3k
2
= q
2
= q
2
is divisible by 3 = q is divisible by 3,
which is a contradiction.
EXERCISE 2.5
Lets assume [3+
2]
2
3
does represent a rational number. Further, lets call this
rational number q. This implies q
3
= [3 +
2]
2
= 9 + 6
2 + 2 = 11 + 6
2,
which in turn implies
2 =
(q
3
11)
6
, which is a rational number. But we know
2 / Q  a contradiction.
4
EXERCISE 3.3
This problem will be done in two parts.
Part I: Show that (a)(b) = ab (a, b) R.
Proof:
(a)(b) = ab (1)
(a)(b) + (ab) = ab + (ab) (2)
(a)(b) + (a)b = ab + (ab) (3)
(a) [(b) +b] = ab + (ab) (4)
(a) [b + (b)] = ab + (ab) (5)
(a)(0) = ab + (ab) (6)
0 = ab + (ab), (7)
which is true by A4. Hence, (a)(b) = ab
2
EXERCISE 3.6
a. Claim: a +b +c a +b +c
Proof: One iteration of the triangle inequality can be used to construct:
a +b +c = (a +b) +c a +b +c .
A second iteration of the triangle inequality yields:
a +b +c a +b +c ,
2
(1) used (i), (2) used (iii), (3) used DL, (4) used A2, (5) used A4, (7) reintroduced
(1) and the conclusion reused (i).
3
(8) was built from multiplying both sides of the equality by the same element to preserve
equality, (9) used M1 and (10) used M4.
5
as desired.
b. We want to show a
1
+a
2
+ +a
n
 a
1
 +a
2
 + a
n

We will use the principal of mathematical induction to show the above is true.
Proof:
P
1
is certainly true, since a
1
 = a
1
, and thus will serve as our basis for induc
tion. Our induction hypothesis is:
Assume an n R : P
n
holds, i.e. P
n
: a
1
+a
2
+ +a
n
 a
1
+a
2
+ a
n
 .
Implementing the induction step and the triangle inequality yield:
a
1
+a
2
+ +a
n
+a
n+1
 a
1
+a
2
+ +a
n
 +a
n+1
 .
The induction step and O4 yield:
a
1
+a
2
+ +a
n
+a
n+1
 a
1
 +a
2
 + +a
n
 +a
n+1
 ,
as desired.
EXERCISE 4.2, (a) through (n)
This exercise is asking us to list three lower bounds for the set; if the set is not
bounded below, it will be labeled NOT BOUNDED BELOW or NBB. The
rule outlined in the text we will follow is Denition 4.2 (b):
If a real number m satises m s s S, then m is called a lower bound of S
and the set S is said to be bounded below.
a. S = [0, 1]; 1, 2 and 3 satisfy Denition 4.2 (b).
b. S = [0, 1); 1, 2 and 3 satisfy Denition 4.2 (b).
c. S = {2, 7}; 1, 2 and 3 satisfy Denition 4.2 (b).
d. S = {, e}; 1, 2 and 3 satisfy Denition 4.2 (b).
e. S =
_
1
n
: n N
_
; 1, 2 and 3 satisfy Denition 4.2 (b).
f. S = {0}; 1, 2 and 3 satisfy Denition 4.2 (b).
6
g. S = [0, 1] [2, 3]; 1, 2 and 3 satisfy Denition 4.2 (b).
h. S =
n=1
[2n, 2n + 1]; 1, 2 and 3 satisfy Denition 4.2 (b).
i. S =
n=1
_
1
n
, 1 +
1
n
n=1
[2n, 2n + 1]; infS = minS = 2.
i. S =
n=1
_
1
n
, 1 +
1
n
2 < r < b
2. Then we have x = r +
2 R \ Q and we have
a < x < b, as desired.
EXERCISE 4.14 (a)
Let A and B be nonempty bounded subsets of R, and let S be the set of all
sums a +b where a A and b B.
Claim: supS = supA + supB.
Proof :
The information we are given is that A and B are both subsets of R which
are nonempty and bounded from above (since they are bounded, this implies
they are bounded from above and below). Since both subsets A and B are both
nonempty and bounded from above, they both possess a least upper bound,
i.e., both supA and supB exist
6
. If S is to be dened as the set whose elements
are all of the sums a + b, where a A and b B and we are given both A
and B bounded, we know S is bounded; more specically bounded from below
and above. Hence, by the Completeness Axiom, S is bounded above and as a
consequence supS exists. If supS exists, then a number O R : r 0 r S
and whenever O
1
< O, r
1
O : r
1
< O
1
..
6
Otherwise, the Completeness Axiom wouldnt hold.
9
If both supA and supB exist, then there exists a number M R : s M s A
and there exists a number N R : t N t B; whenever M
1
< M s
1
A : s
1
< M
1
and whenever N
1
< N t
1
B : t
1
< N
1
.
As a consequence of the above identities, it is obvious that since S set of all
sums a + b where a A and b B, O M + N, and hence supS = supA +
supB, as desired.
EXERCISE 4.15
Claim:
Let a, b R. If a b +
1
n
n N, then a b.
Proof :
By contradiction. Assume a > b. This implies a b > 0. By the Archimedian
Property of R n N : a b >
1
n
. Using this specic n, we have a > b +
1
n
, a
contradiction, as desired.
EXERCISE 4.16
Claim: sup{r Q : r < a} = a for each a R.
Proof :
Let S {r Q : r < a}. From this, we can see that S is certainly bounded
from above, since we can chose any number a + n with n N and Denition
4.2 (b) will still be satised. Now by the denseness of Q in R, we know that
7
if (a, b) R with a < b, then an r Q : a < r < b. So although we know
S is bounded and we can certainly povide many upper bounds for this set, if
we search for a least upper bound for this set, the search will go on indenitely,
since whatever rational upper bound is discovered, it is always possible to nd
one smaller. Thus, although a R, but a / S, this does not disqualify it from
being the supS. However, since the search for a least upper bound for S goes
on indenitely due to the denseness of Q, by the denition of the supremum of
a set, sup{r Q : r < a} = a for each a R, as desired.
EXERCISE 7.3 (b), (d), (f ), (h), (j), (l), (n), (p), (r), (t)
(b) b
n
=
n
2
+3
n
2
3
converges to 1.
(d) t
n
= 1 +
2
n
converges to 1.
(f ) s
n
= (2)
1
n
converges to 1.
(h) d
n
= (1)
n
n diverges.
7
Proof ommitted, as it was done in class
10
(j)
7n
3
+8n
2n
3
31
converges to
7
2
.
(l) sin
_
n
2
_
diverges.
(n) sin
_
2n
3
_
diverges.
(p)
2
n+1
+5
2
n
7
=
2
n
(2)+5
2
n
7
converges to 2.
(r)
_
1 +
1
n
_
2
converges to 1.
(t)
6n+4
9n
2
+7
converges to 0.
EXERCISE 8.4
If (s
n
) is a sequence which converges to 0 and (t
n
) is a bounded sequence, then
the sequence (s
n
t
n
) converges to 0.
We will argue the fact that we are given a sequence (s
n
) which converges to 0
so we know that
lim
n
(s
n
) = 0,
and we are given a sequence (t
n
) which is a bounded sequence, so we know that
a constant M : t
n
 M n N,
which means, in a geometric sense, that we can nd an interval [M, M] that
contains every term in the sequence t
n
.
With the above given, we can almost assuredly argue with what we know about
products that if one sequence which arbitrarily closes in on the value zero is
multiplied by another sequence which is bounded by a constant we can con
clude that their product will eventually close in on the value zero since zero
a constant = zero (eventually).
Claim: Given (s
n
), a sequence which converges to 0 and (t
n
), a bounded se
quence,
lim
n
(s
n
t
n
) 0.
Proof :
Let > 0 be given. Since we know lim
n
s
n
0, we can always nd an
n > N : s
n
0 < <
M
8
, for some constant M > 1. Further, a constant
M : t
n
 M n N. Then, for n > N, it holds that
s
n
t
n
0 = s
n
 t
n
 <
_
M
_
(M) = .
8
Proof ommitted since convergence to 0 is taken as given.
11
EXERCISE 8.7
Claim:
s
n
= cos
_
n
3
_
does not converge.
Proof :
For n = 1, 2, , 6, the terms of the sequence are
1
2
,
1
2
,
1
1
,
1
2
,
1
2
,
1
1
,. . . .
Hence, for any s and any N, we can come up with n > N : s
n
= 1 =
s
n+3
= 1 = s
n
s 1 or s
n+3
s 1, by the triangle inequality, which
proves this sequence does not converge, as desired.
EXERCISE 8.8 (c)
Claim:
lim
n
[
4n
2
+n 2n] =
1
4
.
Proof :
_
4n
2
+n 2n =
n
4n
2
+n + 2n
=
1
2
_
1 +
1
4n
+ 2
If 1 < a, then 1 < a < a
2
which then implies
1 lim
n
_
1 +
1
2n
lim
n
_
1 +
1
2n
_
= 1 + 0 = 1.
Now, applying the limit theorems, we see that
lim
n
1
2
_
1 +
1
4n
+ 2
=
1
2 1 + 2
=
1
4
,
as desired.
12
EXERCISE 8.10
We are given s > a. Let lim
n
s
n
= s.
Claim:
a number N : n > N = s
n
> a.
Proof :
s > a = there is an > 0 : s > a
9
For such chosen N
: n > N
=
s
n
(s , s +) = n > N
= s
n
> s > a, as desired.
EXERCISE 9.4
Let s
1
= 1 and for n 1 let s
n+1
=
s
n
+ 1.
(a) The rst four elements are 1,
2,
_
2 + 1,
_
_
2 + 1 + 1.
(b) Assume s
n
converges.
Claim: lim
n
s
n
=
1
2
(1 +
5).
Proof :
We are given the assumption that s
n
converges. We will name some as
the element s
n
converges to. If s
n
converges, s
n+1
inherently converges; more
specically, s
n+1
converges to , as well. From the problem, we have
(s
n+1
)
2
= s
n
+ 1,
which when utilized with the assumption laid out will converge to
()
2
10
= + 1
2
1 = 0.
Implementing the quadratic formula
11
to nd a pair of solutions to the equation
yields
1
=
1 +
5
2
and
2
=
1
5
2
.
Since the greatest lower bound of the sequence s
n
is 1, the solution we want
to pick will reect that the limit will trivially be positive, eliminating
2
as a
solution.
Hence, lim
n
s
n
=
1
=
1
2
(1 +
5).
9
For example :=
sa
2
.
10
For this, we will use the fact that the limit of a product of sequences is the product of the
limits, i.e. if lims
n
s and limt
n
t, then lim(s
n
t
n
) st.
11
The quadratic formula, learned from principals, is x = b
b
2
4ac
2a
.
13
EXERCISE 9.5
Assume limt
n
exists and is dened to be t. Then limt
n+1
= t as well. For all
n, we get 2t
n
t
n+1
= t
2
n
+ 2. Implementing the limit theorems, we see that
2t
2
= t
2
+ 2 = t =
2.
Since we are given t
1
= 1 and the fact that the sequence stays positive, we can
eliminate
s
n+1
s
n
s
N+1
<
a
2
s
N
 . . .. Thus, by induction, s
N+n
 < a
n
s
N
 n N. To summarize,
lim
n
s
n
 = lim
n
s
N+n
 lim
n
a
n
s
N
 = s
N
 lim
n
a
n
= 0
14
given a < 1.
EXERCISE 9.15
Let s
n
=
a
n
n!
=
s
n+1
s
n
=
a
(n+1)
0 as n = lims
n
= 0
EXERCISE 10.1 (b), (d) and (f )
(b)
(1)
n
n
2
. The rst term, 1, is less than the second term,
1
4
. Hence, the
sequence is not nonincreasing. The second term is greater than the third term,
1
9
. Hence, the sequence is not nondecreasing. Since
(1)
n
n
2
n
2
1 n
N, the sequence is bounded.
(d) sin
_
n
7
_
. The rst term is positive, however, the 7th term is 0 = the
sequence is not nondecreasing. The second term is greater than the rst term
and hence the sequence is not nonincreasing. Since sin x 1 x, the sequence
is bounded.
(f )
n
3
n
. The sequence is nonincreasing since
a
n+1
a
n
< 1. As all terms are positive,
the sequence is bounded from below and since the sequence is nonincreasing, it
is bounded from above = the sequence is bounded.
EXERCISE 10.2
Claim:
All bounded monotone sequences converge.
Proof :
Let (s
n
) be a bounded nonincreasing sequence. Let S be dened as the set
{s
n
: n N} and let v = inf S. S bounded = v exists and is real. We will
now show that lims
n
= u. Let > 0. Since v + is not an upper bound for S,
N : s
N
< v + . Since s
n
is nonincreasing, = s
N
s
n
n > N. s
n
v n
and hence, n > N = v + > s
n
v = s
n
v < = lims
n
= v, as
desired.
EXERCISE 10.5
Claim: If (s
n
) is an unbounded nonincreasing sequence, then lims
n
= .
We want to show that
lims
n
= for each M < 0 a number N : n > N = s
n
< M.
15
Proof :
Let (s
n
) be an unbounded nondecreasing sequence. Let M < 0. Since the set
{s
n
: n N} is unbounded and it is bounded above by s
1
, it must be unbounded
below, since for this sequence to be nonincreasing, the condition s
n
s
n+1
must
be fulilled. Hence, for some N N we have s
N
< M. Clearly, n > N =
M > s
N
s
n
, so
lims
n
= ,
as desired.
EXERCISE 10.6
(a) Let (s
n
) be a sequence such that
s
n+1
s
n
 < 2
n
n N
Claim: (s
n
) is a Cauchy sequence and hence a convergent sequence.
Proof :
Given an arbitrary > 0, for n > N, choose N : 2
N
<
2
. Then m > n > N,
we can see that
s
m
s
n

m1
k=n
s
k+1
s
k
.
Now,
m1
k=n
s
k+1
s
k
m1
k=n
s
k+1
s
k

m1
k=n
2
k
(11)
*Note: The second inequality from line (1) is justied by Let (s
n
) be a sequence
such that s
n+1
s
n
 < 2
n
n N in the rst line of the problem.
m1
k=n
2
k
k=N
2
k
= 2
N+1
< , (12)
as desired.
(b) Is the result in (a) true if we only assume that s
n+1
s
n
 <
1
n
n N?
If we only assume that s
n+1
s
n
 <
1
n
n N, our immediate reaction is to
say that the new constraint makes the result in (a) false, since the harmonic
series,
n=1
1
n
, is taken to be divergent.
As a counterexample, let us choose s
n
=
n=2
1
n1
, which will still fulll
s
n+1
s
n

1
n
, but will inevitably diverge. Thus, as we travel down the se
quence (N ), arbitrary s
m
s and s
n
s are not encroaching towards eachother,
violating the Cauchy criterion.
16
EXERCISE 10.7
Choose n N. Construct s
n
S : sup S s
n
<
1
n
and s
n
> s
n1
for n > 1 =
s
n
an increasing sequence converging to sup S.
Pick s
1
S : sup S 1 < sup S is not an upper bound of S. Implement induc
tion. Assume s
1
< < s
n1
exists. Since sup S / S = s
n1
< sup S =
s
n
S : sup S s
n
> s
n1
and sup S s
n
<
1
n
. This is possible given that
neither s
n1
< sup S nor sup S
1
n
< sup S is an upper bound for S.
EXERCISE 10.8
Let (s
n
) be a nondecreasing sequence of positive numbers and dene
n
=
1
n
(s
1
+s
2
+ +s
n
)
Claim: (
n
) is a nondecreasing sequence.
Proof :
From the assumption of s
n
being a nondecreasing sequence, we will proclaim
s
n
s
n+1
Hence,
ns
n
ns
n+1
(13)
= (s
1
+s
2
+ s
n
) ns
n
n(s
n+1
) (14)
= n(s
1
+s
2
+ s
n
) + (s
1
+s
2
+ s
n
) n(s
1
+s
2
+ s
n
) +n(s
n+1
)
(15)
= (n + 1)(s
1
+s
2
+ s
n
) n(s
1
+s
2
+ s
n
+s
n+1
) (16)
=
1
n
(s
1
+s
2
+ s
n
)
1
n + 1
(s
1
+s
2
+ s
n
+s
n+1
),
(17)
Thus,
n
n+1
,
as desired.
EXERCISE 11.8
(a) Use Denition 10.6 and Exercise 5.4 to prove that lim inf s
n
= lim sup(s
n
).
It is readily observable that
lim sup(s
n
) = lim
N
sup {(s
n
) : n > N} = lim
N
inf {(s
n
) : n > N} = lim inf(s
n
).
17
Hence,
lim sup(s
n
) = lim inf(s
n
) lim inf(s
n
) = lim sup(s
n
),
as desired.
(b) Let (t
k
) be a monotonic subsequence of (s
n
) converging to lim sup(s
n
).
Show that (t
k
) is a monotonic subsequence of (s
n
) converging to lim inf s
n
.
Observe that this completes the proof of Corollary 11.4.
We are given a monotonic subsequence of (s
n
), denoted (t
k
). Then
t
k
= s
n
k
k, (t
k
monotonic) t
k
= s
n
k
k, (t
k
monotonic).
When taken with the results from part (a), the desired result is apparent.
EXERCISE 11.9
(a) Let (x
n
) be a convergent sequence such that a x
n
b n N =
a lim
n
x
n
b = [a, b] is a closed set.
(b) No. We know that the set of any subsequential limits of any set mus be
closed. The interval (0, 1) is not closed.
EXERCISE 11.10
Let (s
n
) be the sequence of numbers in Figure 11.2 listed in the indicated order.
(a) Find the set S of subsequential limits of (s
n
).
Laying out the numbers listed in the indicated order gives the sequence
1,
1
2
, 1, 1,
1
2
,
1
3
,
1
4
,
1
3
,
1
2
, 1, 1, ...
Rearranging,
1/1,
1/2, 1/1,
1/1, 1/2, 1/3,
1/4, 1/3, 1/2, 1/1,
1/1, 1/2, 1/3, 1/4, 1/5,
1/6, 1/5, 1/4, 1/3, 1/2, 1/1,
1/1, 1/2, 1/3, 1/4, 1/5, 1/6, 1/7,
1/8, 1/7, 1/6, 1/5, 1/4, 1/3, 1/2, 1/1,
It seems that 2 sequences, both monotonic, one increasing the other decreas
ing. The even rows of the pyramid represent a monotonically increasing se
quence and the odd rows correspond to a monotonically decreasing sequence.
Upon further inspection, we see that with respect to the decreasing sequences,
18
the denominator of the last entry corresponds to the row number of the pyramid
for which the decreasing sequence is located, i.e., the last entry of row 3 has
1
3
as its last entry. Conversely, with respect to the increasing sequences, the
denominator of the entry for which the increasing sequence commences corre
sponds to the row number of the pyramid for which the increasing sequence is
located, i.e., the rst entry of row 6 is
1
6
. Since the odd rows all commence at
1
1
and progressively converge to
1
n
, we can say that one subsequential limit is
0. Conversely, since the even rows all seem to converge to 1 from a fraction
progressively closer to
1
n
, we could say the other subsquential limit is 1.
Hence the set of subsequential limits is
S =
_
1
n
: n N
_
_
_
0
_
EXERCISE 12.1
Let (s
n
) and (t
n
) be sequences and suppose that there exists N
0
: s
n
t
n
n >
N
0
.
Claim:
liminf s
n
liminf t
n
and limsup s
n
limsup t
n
.
Proof :
Refer to Exercise 4.8 (b). From there, we see that
sup {s
n
: n > N} inf {t
n
: n > N} N N
0
=
lim
N
sup {s
n
: n > N} lim
N
inf {t
n
: n > N}
limsup(s
n
) liminf(t
n
) ()
Since
liminf(s
n
) limsup(s
n
) and
liminf(t
n
) limsup(t
n
),
the desired inequalities are obtained implementing (*), as desired.
19
EXERCISE 12.2
Prove that lim sup s
n
 = 0 if and only if lim s
n
= 0.
We will use Theorem 11.7 throughout this proof.
Claim: lim sup s
n
 = 0 lims
n
= 0.
Proof :
= :
lims
n
= 0 lims
n
 = 0 lim sups
n
= 0.
12
= :
limsup s
n
 = 0 liminf s
n
 = 0, since s
n
 0.
As N , limsup s
n
 lims
n
, since Theorem 11.7 tells us that limsup s
n

is exactly the largest subsequential limit of s
n
. Hence,
limsup s
n
 = 0 lims
n
= 0,
as desired.
EXERCISE 12.3
(a) liminf s
n
+ liminf t
n
= 0 + 0 = 0
(b) liminf(s
n
+t
n
) = 1
(c) liminf s
m
+ limsup t
n
= 0 + 2 = 2
(d) limsup(s
n
+t
n
) = 3
(e) limsup s
n
+ limsup t
n
= 2 + 2 = 4
(f ) liminf s
n
t
n
= 0
(g) limsup s
n
t
n
= 2
EXERCISE 12.4
Claim:
limsup(s
n
+t
n
) limsup s
n
+ limsup t
n
for bounded (s
n
) and (t
n
).
Proof :
For any n > N,
s
n
+t
n
sup {s
n
: n > N} + sup {t
n
: n > N} =
sup {(s
n
+t
n
) : n > N} sup {s
n
: n > N} + sup {t
n
: n > N} .
12
By Theorem 11.7.
20
Hence,
lim
N
sup {(s
n
+t
n
) : n > N} lim
N
(sup {s
n
: n > N} + sup {t
n
: n > N})
= lim
N
sup {s
n
: n > N} + lim
N
sup {t
n
: n > N} ,
where the last equality holds since
_
sup {s
n
: n > N}
_
N=N
0
and
_
sup {t
n
: n > N}
_
N=N
0
are both convergent and monotonic, as desired.
EXERCISE 12.8
Let (s
n
) and (t
n
) be bounded sequences of nonnegative numbers. Prove that
lim sup s
n
t
n
(lim sup s
n
)(lim sup t
n
).
Give an example where the strict inequality holds.
Claim:
lim sup s
n
t
n
(lim sup s
n
)(lim sup t
n
).
Proof :
Since s
n
and t
n
are both bounded sequences of nonnegative numbers,
s
n
sup {s
n
: n > N} n N
and
t
n
sup {t
n
: n > N} n N,
by denition. This implies
(s
n
)(t
n
) (sup {s
n
: n > N})(sup {t
n
: n > N}) n N,
by multiplicativity and since, once again, s
n
and t
n
are both bounded and
nonnegative. Now since the right hand side of the inequality serves as an upper
bound for the left hand side,
sup {s
n
t
n
: n > N} (sup {s
n
: n > N})(sup {t
n
: n > N}) n N
If s
n
and t
n
are both sequences which converge to s and t with s
n
t
n
n N,
we can take limits and conclude
lim sup s
n
t
n
(lim sup s
n
)(lim sup t
n
),
as desired.
21
An example where the strict inequality would hold would be the case when s
n
and t
n
both converge to s and t, respectively, but s
n
< t
n
n N.
EXERCISE 12.10
Claim:
(s
n
) is bounded limsup s
n
 < +
Proof :
limsup = + = a subsequence (s
n
k
) : lims
n
k
 = + =this subsequence
is unbounded. Conversely, (s
n
) unbounded = for any k N s
n
k
: s
n
k
 > k.
Assume that n
1
< n
2
< < n
k
< and thus get a subsequence (s
n
k
) :
lims
n
k
 = +, i.e., limsup s
n
 = +.
EXERCISE 12.11
Claim:
liminf
s
n+1
s
n
liminf s
n

1
n
Proof :
Let = liminf s
n

1
n
. Let L = liminf
s
n+1
s
n
s
n+1
s
n
= lim
n
_
inf
_
s
n+1
s
n
: n > N
__
> L
1
= N N : inf
_
s
n+1
s
n
: n N
_
> L
1
= n N we have
s
n+1
s
n
> L
1
It then follows that
s
n
 > L
nN
1
s
n
 , n > N
s
n
 > L
n
1
_
L
N
1
s
n

_
. .
=
, n > N
s
n
 > L
n
1
, n > N = s
n

1
n
> L
1
1
n
, n > N
Since lim
n
1
n
= 1, the result is that limsup s
n

1
n
L
1
.
22
EXERCISE 13.3 (a)
Let B be the set of all bounded sequences x = (x
1
, x
2
, ...) and dene d(x, y) =
sup {x
j
y
j
 : j = 1, 2, ...}.
Show that d is a metric for B.
To show that d is indeed a metric for B, we need to show that it satises the
three conditions of Denition 13.1. As we will see the rst 2 conditions are
trivial.
D1. d(x, x) = 0, since sup {x
j
x
j
 : j = 1, 2, ...} = sup {0} = 0, and
d(x, y) > 0, since for bounded sequences x = (x
1
, x
2
, ...) and y = (y
1
, y
2
, ...),
sup {x
j
y
j
 : j = 1, 2, ...} > 0.
D2. d(x, y) = sup {x
j
y
j
 : j = 1, 2, ...} = sup {y
j
x
j
 : j = 1, 2, ...} =
d(y, x).
D3. If x, y, z B, then for each j = 1, 2, ..., k,
d(x, z) = sup {x
j
z
j
 : j = 1, 2, ...} (18)
= sup {x
j
y
j
+y
j
z
j
 : j = 1, 2, ...} (19)
= sup {(x
j
y
j
) + (y
j
z
j
) : j = 1, 2, ...} (20)
sup {x
j
y
j
 : j = 1, 2, ...} + sup {y
j
z
j
 : j = 1, 2, ...} (21)
d(x, y) +d(y, z), (22)
where the inequality follows from the triangle inequality and the nature of the
supremum. Hence d is a metric.
EXERCISE 13.6
We will prove each part of Proposition 13.9 in turn.
(a) Claim:
The set E is closed if and only if E = E
.
Proof :
=
Assume E is closed = E is a closed set containing E = the intersection of
all closed sets containing E E = E
= E.
=
Assume E
E, as desired.
(c) Claim:
An element is in E
in this ball = E
o
is not the interior
of E because any (s = x S) the aforementioned ball is not contained in
E
o
, which is a contradiction. Now if x E
\ E
o
= we can create a similar
ball around x: some of the interior of E is missing, another contradiction =
any such point in E
is a limit point.
=
see part (b) above, as desired, lol.
(d) Claim:
A point is in the boundary of E if and only if it belongs to the closure of both
E and its complement.
Proof :
Let E E
o
= S \ E closed = S \ E contains its boundary = any s
boundary of S \ E must also be in the boundary of E, otherwise either some
x: x is neither E or S \ E, or x both E and S \ E, which both contradict
S \ E E
c
. Let E E
(0) := {x : d(x, 0) } ,
WLOG, dene the complement of the closed ball by
B
15
= of all T
n
Recalling the above, T
is closed = T
) =
_
T
c
_
c
and note that we are still in the metric space [0, 1]. Thus, T
c
= [0, 1] \ T
=
T
c
= ([0, 1] \ T
{x [0, 1] : x is a limit point of [0, 1] \ T} [0, 1], we know
T
c
= ([0, 1] \ T
) T
[0, 1].
In other words, the closure of the Cantor Set is T
0
, the interval from 0 to 1.
Hence,
int(T
) =
_
T
c
_
c
[0, 1]
c
,
as desired.
EXERCISE 13.12
(a) Let (S, d) be any metric space.
Claim:
If E is a closed subset of a compact set F, then E is also compact.
Proof :
Let E be a closed subset of a compact set F. Let a collection of open sets U be
an open cover for E. E closed = E
c
open. Now, E
c
+ U an open cover of
F. Given F compact, a nite subcover that covers F. Now given E F and
F covered by some nite subcover, E compact.
14
For every real numbers x and y, with x < y, a rational and an irrational such that
x < < y and x < < y.
15
T
1
n
and G
l
+
1
n
which are clearly E. Furthermore, limL
u
1
n
= L
u
and limG
l
+
1
n
= G
l
. E closed = L
u
, G
l
E. But then by the denition of
28
L
u
and G
l
, (supE, infE) E.
EXERCISE 14.2
(a)
n1
n
2
=
16
n
n
2
1
n
2
=
1
n
1
n
2
. Since
1
n
diverges, the entire
series diverges.
(b) The series
(1)
n
fails to converge because it doesnt satisfy the Cauchy
criterion
17
. In other words, the terms of the sequence a
n
do not arbitrarily grow
closer to eachother as n .
(c)
3n
n
3
=
3
n
2
= 3
1
n
2
, which we know is 3 times a convergent series, thus
the series converges.
(d) If a
n
=
n
3
3
n
, then a
n+1
/a
n
=
(n+1)
3
3
n
3
n+1
n
3
, so lima
n+1
/a
n
 =
1
3
. Hence the
series converges by the Ratio Test.
(e) If a
n
=
n
2
n!
, then a
n+1
/a
n
=
(n+1)
2
n!
(n+1)!n
2
, so lima
n+1
/a
n
 = 0. Hence the series
converges by the Ratio Test.
(f ) If a
n
=
1
n
n
=
_
1
n
_
n
, then lim supa
n

1
n
= 0. Hence the series converges by
the Root Test.
(g) If a
n
=
n
2
n
, then a
n+1
/a
n
=
(n+1)2
n
2
n+1
n
, so lima
n+1
/a
n
 =
1
2
. Hence the series
converges by the Ratio Test.
EXERCISE 14.4
(a) We will use induction and the Comparison Test to show that the series
n=2
1
[n + (1)
n
]
2
converges.
To accomplish this task we need to show
n + (1)
n
1
2
n,
so that
n=2
1
[n + (1)
n
]
2
n=2
4
n
2
= 4
n=2
1
n
2
,
since, by the Comparison Test, this would show the series is convergent.
16
This single series has been split into 2 seperate series. The rule I am follow
ing is that the sum of two series will converge is both of the sums converge. Hence
the series will diverge, if we can show that one of the sums diverges. Reference:
http://www.sosmath.com/calculus/series/examples/examples.html
17
Proof ommitted, as it wasnt required.
29
The summand index commences at n = 2, so this will serve as the induction
basis.
2 + (1)
2
1
2
(2) = 3 1,
which is trivial.
Let us now implement the induction step, n + 1, and show the inequality still
holds.
(n + 1) + (1)
n+1
1
2
(n + 1) (23)
n + 1 + (1)
n
(1)
1
2
n +
1
2
(24)
n + 1 (1)
n
1
2
n +
1
2
(25)
1
2
n +
1
2
(1)
n
0 (26)
1
2
n +
1
2
(1)
n
, (27)
which holds for n 2, as desired. Hence,
n + (1)
n
1
2
n (28)
(n + (1)
n
)
2
_
1
2
n
_
2
(29)
1
(n + (1)
n
)
2
1
_
1
2
n
_
2
=
4
n
2
(30)
n=2
1
[n + (1)
n
]
2
n=2
4
n
2
= 4
n=2
1
n
2
, (31)
as desired.
(b) Since
_
n + 1
n + 1 +
1
2
n + 1
1
2
2n
=
1
2
n
,
which is a divergent pseries
18
, the series diverges.
(c) We will show the series
n!
n
n
18
A pseries is a series of the form
1
n
p
. Such a series converges if p > 1 and diverges if
p 1.
30
converges via the Ratio Test. We want to show
lim
a
n+1
a
n
< 1.
lim
a
n+1
a
n
= lim
(n + 1)!
(n + 1)
n+1
n
n
n!
= lim
n!(n + 1)n
n
(n + 1)
n
(n + 1)n!
= lim
n
n
(n + 1)
n
= lim
_
n
n + 1
_
n
.
Since
lim
_
n + 1
n
_
n
= lim
_
1 +
1
n
_
n
= e,
this suces to show that
lim
_
n
n + 1
_
n
=
1
e
< 1.
Hence, the series converges by the Ratio Test.
EXERCISE 14.7; assume p Z : p > 1
We want to show that if we have a known convergent series
a
n
and raise it
to a power p > 1, it will simply converge quicker. An example would be the
convergent pseries. We know that
1
n
p
converges for values of p > 1. Now if the series is further raised by a power of
p > 1 the original p will be even greater, and thus wll still be convergent.
Claim:
a
n
:= convergent and a
n
0 n N =
(a
n
)
p
converges.
Proof :
If we know that
a
n
is a convergent series, then
lim
n
a
n
= 0 = N N : a
n
(0, 1) n N = (a
n
)
p
< a
n
p > 1 and n N
(a
n
)
p
< a
n
p > 1 and n N =
(a
n
)
p
<
a
n
p > 1 and n N.
Hence,
(a
n
)
p
converges by the Comparison Test, as desired.
31
EXERCISE 14.10
Consider this series:
n=0
2
(1)
n
+n
EXERCISE 15.4
Determine which of the following series converge.
(a)
n=2
1
nlog n
. Since log n <
n for large n =
1
n
<
1
log n
=
1
n
<
1
nlog n
=
n=2
1
n
<
n=2
1
nlog n
, and hence is divergent by comparison
with the harmonic series.
(b)
n=2
log n
n
. This problem trivially diverges when compared
19
to the har
monic series for values of n > 1.
(c)
n=4
1
n(log n)(log log n)
. Implement the integral Test.
_
n=4
1
n(log n)(log log n)
dn
can be evaluated with a substitution. Let u = log log n. The integral now
becomes
_
log log
log log 4
1
u
du =
_
log u
_
log log
log log 4
= log log log log log log 4 = ,
hence the series diverges by the integral test.
(d)
n=2
log n
n
2
. The integral of
log n
n
2
is
(log n+1)
n
, so implementing the integral
test yields:
lim
n
(log x + 1)
x
n
2
=
log n
n
1
n
+
log 2
2
+
1
2
,
which converges to
log 2
2
+
1
2
using LHospitals Rule. Hence, since the integral
converges to an existent nite number, the series converges.
EXERCISE 15.6
(a) A divergent series
a
n
for which
a
2
n
converges is the harmonic series,
n=1
1
n
.
(b) If
a
n
is a convergent series of nonnegative terms, then
a
2
n
also con
verges. This can be shown with a similar method as the proof done in Problem
No. 3.
19
Note: the index on the series begins at n = 2 =log n > 1 for n 2.
32
Claim:
a
n
:= convergent and a
n
0 n N =
(a
n
)
2
converges.
Proof :
If we know that
a
n
is a convergent series, then
lim
n
a
n
= 0 = N N : a
n
(0, 1) n N = (a
n
)
2
< a
n
n N
(a
n
)
2
< a
n
n N =
(a
n
)
2
<
a
n
n N.
Hence,
(a
n
)
2
converges by the Comparison Test, as desired.
(c) Consider this series:
n=1
(1)
n
n
.
EXERCISE 17.5
(a) We will use induction to prove f(x) = x
m
is continuous. This will be done
by rst showing that when m = 1 we are dealing with f(x) = x, which will
be taken as continuous (proof is provided in the Appendix). Then by assuming
that f(x) = x
m
is continuous for some m N, we will show our induction
step, f(x) = x
m+1
is continuous, by noting that f(x) = x
m+1
= x
m
x is hence
continuous given Theorem 17.4 (ii), which will then imply that f(x) = x
m
is
continuous for m N.
Claim:
If m N, then the function f(x) = x
m
is continuous on R.
Proof :
This proof will use mathematical induction. For m = 1, f(x) = x, which
will be taken as continuous (proof provided in the Appendix). f(x) = x will
then serve as our induction basis. We can now assume our induction hypothesis,
f(x) = x
m
is continuous. But we need to show continuity holds for the induction
step, m + 1. But we know f(x) = x
m+1
x
m
x, which is the product of two
continuous functions, our inductions hypothesis, and our induction basis, and
is hence continuous by Theorem 17.4 (ii).
(b) Claim:
Every polynomial function p(x) = a
0
+a
1
x + +a
n
x
n
is continuous on R.
Proof :
Given the solution from part (a), p(x) is simply the sum and product of con
tinuous functions and is hence continuous, by Theorem 17.4 (i) and Theorem
17.3, as desired.
33
EXERCISE 17.6
Claim:
Every rational function is continuous.
Proof :
A rational function is composed of constants, f(x) = c and the continuous func
tion f(x) = x by multiplication, addition and division. Since f(x) = x and
f(x) = c are trivially continuous = rational functions are continuous by the
continuity theorems of sums, products and quotients of continuous functions, as
desired.
EXERCISE 17.7 (b)
Claim:
x is a continuous function on R.
Proof :
x is continuous at any x0 since it coincides with x for x > 0 and x for x < 0.
At x = 0, the function f(x) = x is continuous because for any > 0 we have:
x 0 < = f(x) f(0) = x < .
EXERCISE 17.8
(a) Claim:
min(f, g) =
1
2
(f +g)
1
2
f g
Proof :
Case 1: Let f(x) g(x). Then
min(f, g) = f(x) =
1
2
(f +g)
1
2
(g f) (32)
=
1
2
(f +g)
1
2
f g (33)
Case 2: Let f(x) g(x). Then
min(f, g) = g(x) =
1
2
(f +g)
1
2
(f g) (34)
=
1
2
(f +g)
1
2
f g . (35)
Hence,
min(f, g) =
1
2
(f +g)
1
2
f g ,
as desired.
34
(b) Claim:
min (f, g) = max (f, g)
Proof :
Case 1: Let f(x) g(x) = f(x) g(x)
= min (f, g) = f(x) = (f(x)) = max(f, g)
Case 2: Let f(x) g(x) = f(x) g(x)
= min (f, g) = g(x) = (g(x)) = max(f, g).
Hence,
min (f, g) = max (f, g) ,
as desired.
(c) Claim:
f and g continuous at x
0
R = min(f, g) is continuous at x
0
Proof :
Recall Theorems 17.3 and 17.4 (i):
Theorem 17.3: Let f be a realvalued function with dom(f) R.
If f is continuous at x
0
in dom(f), then f and kf, k R, are
continuous at x
0
.
Theorem 17.4 (i): Let f and g be realvalued functions that are
continuous at x
0
R. Then f +g is continuous at x
0
.
In combination with the results from part (a), i.e.,
min(f, g) =
1
2
(f +g)
1
2
f g ,
we see that min(f, g) is simply the sum, dierence and composition of functions
which are continuous at x
0
, and hence is itself continuous at x
0
, as desired. .
EXERCISE 17.9
(a) Claim:
f(x) = x
2
is continuous at x
0
= 2
Proof :
Let > 0 be given. We want to show that x
2
4 < provided x 2 is
suciently small, i.e., less than some . We observe that x
2
4 = (x+2)(x
2) x + 2 x 2. We need to get a bound for x 2 that doesnt depend
on x. We notice that if x 2 < 1, say, then x + 2 < 5, so it suces to get
x2 5 < . So by setting = min
_
1,
5
_
, we see that f(x) = x
2
is continuous
at x
0
= 2, as desired.
35
(b) Claim:
f(x) =
x is continuous at x
0
= 0
Proof :
Let > 0 be given. We want to show that 
x 0 < provided x 0
is suciently small, i.e., less than some . We observe that 
x 0 =
x.
Since we want this to be less than , we set =
2
. Then x 0 < implies
x <
= , so
x 0 < = f(x) f(0) < ,
as desired.
(c) Claim:
f(x) = xsin
_
1
x
_
for x = 0 and f(0) = 0 is continuous at x
0
= 0
Proof :
Let > 0 be given. We want to show that xsin
_
1
x
_
0 < provided x 0
is suciently small, i.e., less than some . We see that xsin
_
1
x
_
0 x x.
Since we want this to be less than , we set = . Then x 0 < implies
x < = , so
x 0 < =
xsin
_
1
x
_
0
< ,
as desired.
(d) Claim:
g(x) = x
3
is continuous at x
0
arbitrary
Proof :
By the hint given,
x
3
x
3
0
= (x x
0
)(x
2
+x
0
x +x
2
0
) = (x x
0
)
_
(x
2
x
0
)
2
+ 3xx
0
)
gx
0

2
+ 1
,
3
+
3
+
3
= ,
as desired. .
Alternatively,
Assume g(x) = x
2
, x
0
arbitrary. Observe that (x
3
x
3
0
) = (xx
0
)(x
2
+x
0
x+x
2
0
).
Now we will make the assumption that x x
0
 < 1 = x < x
0
 + 1, which
enables us to see
x
2
+x
0
x+x
2
0
 x
2
 +x
0
x +x
2
0
 < x
0

2
+2x
0
 +1+x
0
 x
0
 +1 +x
2
0
 k,
where the material to the right of the last inequality is all greater than 0.
We can then set = min
_
1,
k
_
. Hence,
xx
0
 < = g(x)g(x
0
) = (xx
0
)(x
2
+x
0
x+x
2
0
) < xx
0
k <
k
k = ,
as desired.
EXERCISE 17.12
(a) Let f be a continuous realvalued function with domain (a, b).
Claim:
If f(r) = 0 for each rational number r (a.b), then f(x) = 0 x (a.b).
Proof :
We are given f(x) = 0 x Q. If x R \ Q, then a sequence of rational
numbers, (r
n
), which converges to x. Hence, by continuity, r
n
x = f(r
n
)
f(x). But f(r
n
) = 0 n, given the conditions in the claim, so 0 f(x) =
f(x) = 0 x (a, b), as desired.
(b) Let f and g be continuous realvalued functions on (a, b) : f(r) = g(r) for
each rational number r (a, b).
Claim:
f(x) = g(x) x (a, b).
Proof :
Using the limit concept of sequences, for any x (a, b) a sequence of rational
numbers, (r
n
) : limf
n
= x = f(x) = limf(r
n
) = limg(r
n
) = g(x).
20
1
37
Hence, f(x) = g(x), as desired.
EXERCISE 17.13 (b)
Let h(x) = x x Q and h(x) = 0 x R \ Q.
Claim:
h is continuous at x = 0 and no other point.
Proof :
For any > 0, if x 0 < , then h(x) h(0) is either 0 (if x is R\Q) or x
(if x Q, and thus in both cases < ). Thus h is continuous at x = 0. other
x, consider two sequences with limit x, one (r
n
) Q, and another, (x
n
) R\Q.
Then limh(x
n
) = 0 and limh(r
n
) = x = 0. Hence, disconituity for h at x = 0,
as desired.
EXERCISE 17.14
Claim:
f is continuous at each point of R \ Q and discontinuous at each point of Q.
Proof :
For x =
p
q
Q, dene a sequence x
n
R\Q : limx
n
= x = limf(x
n
) = 0, but
f(x) =
1
q
= 0 = f is discontinuous at x. For an irrational x and any > 0, let
> 0 be dened as the distance from x to the closest irreducible fraction
p
q
with
denominator q
1
: x
x < = f(x
) f(x) < 0 = f
continuous at x R \ Q, as desired.
EXERCISE 18.2
The limit x
0
(or y
0
) of the subsequence x
n
k
(a, b) (or y
n
k
(a, b)) may be
an endpoint a or b of the interval and thus lie outside of the domain of the
function.)
38
EXERCISE 18.4
Let S R and suppose there exists a sequence (x
n
) in S that converges to a
number x
0
/ S.
Claim:
an unbounded continuous nction on S.
Proof :
Let x
0
/ S. We are given a sequence (x
n
) S which converges to x
0
. Then
x
n
x
0
 := the distance to x
0
is continuous and strictly positive on S. Dene
f :=
1
x
n
x
0

= f is welldened and continuous on S = lim
n
f = , as
desired.
EXERCISE 18.6
Claim:
x = cos x for some x
_
0,
2
_
.
Proof :
We know f(x) = xcos x is continuous
_
0,
2
1
2
=
1
1
4
= 4. Dene :=
4
. Then for any x, y
_
1
2
,
_
with x y < ,
f(x) f(y) =
1
x
1
y
y x
xy
= x y
1
xy
< x y 4 = ,
as desired. .
EXERCISE 19.4
(a) Claim:
If f is uniformly continuous on a bounded set S, then f is a bounded function
on S.
Proof :
By contradiction. Assume that f is uniformly continuous on a bounded set S
and an unbounded function on S. Then a sequence, (x
n
) in the domain of
f such that f(x
n
) n. By the BolzanoWeierstrass Theorem, the sequence
contains a convergent subsequence (x
n
k
), since the domain is bounded. A con
vergent subsequence is Cauchy (by denition) and hence the sequence of values
f(x
n
k
) is Cauchy by the property of uniformly continuous functions. But the
sequence f(x
n
k
) n
k
is unbounded, contradicting our assumption in the out
set of this proof. Hence, if f is uniformly continuous on a bounded set S, then
f is a bounded function on S, as desired.
(b) Claim:
1
x
2
is not uniformly continuous on (0, 1).
Proof :
We want to show that if
1
x
2
is not a bounded function on (0, 1), then
1
x
2
is not
uniformly continuous on (0, 1)
21
. So it will suce to show that
f(x) =
1
x
2
is not a bounded function on (0, 1).
Let M > 1 be arbitrary = 0 <
1
M
2
< 1. We want to show that
1
x
2
> M for
some x (0, 1) = x <
1
M
. Let x =
1
M+1
, since 0 <
1
M+1
<
1
M
< 1. But
then for f(x) =
1
x
2
, we see that
1
x
2
=
1
_
1
M+1
_
2
= M + 1 > M,
21
This is the contrapositive of what was proved in part (a).
41
which shows that f(x) =
1
x
2
is not a bounded function on (0, 1), as desired.
EXERCISE 19.6
(a) Claim:
f(x) =
x is uniformly continuous on (0, 1], although f
is unbounded.
Proof :
f
(x) =
1
2
x
+ as x 0 = f
y =
xy

x+
y
. Take := 2. Then
x y < = x y < 2 =
x y
2
x y

x +
y
= 
y < ,
as desired.
EXERCISE 19.10
The limit
lim
x0
g(x)
x
= lim
x0
xsin x
1
x
= 0,
in other words, g is dierentiable at x = 0 (g
(0) = 0). At x = 0,
g
(x) = 2xsin
1
x
+x
2
cos
1
x
_
1
x
2
_
= 2xsin
1
x
cos
1
x
,
which is bounded. Since cos y = 1and sin y y y = for y =
1
x
:
2xsin
1
x
cos
1
x
2
y
sin y cos y
2 + 1 = 3.
Thus, g
b
x b
=
1
2
b
Proof :
b
x b
=
(
b)
x+
x+
b
x b
=
xb
x+
b
x b
=
1
x +
b
Hence,
lim
xb
1
x +
b +
b
=
1
2
b
EXERCISE 20.14
Claim:
lim
x0
+
1
x
= + and lim
x0
1
x
=
We will split this up into 2 cases/claims.
Claim 1:
lim
x0
+
1
x
= +
Proof :
Let M > 0 and :=
1
2M
. Then 0 < x < 0 + = 0 < x < =
f(x) =
1
x
>
1
=
1
1
2M
= 2M > M. Hence, lim
x0
+
1
x
= +, as desired.
Claim 2:
lim
x0
1
x
=
Proof :
Let M < 0 and :=
1
2M
. Then 0 < x < 0 = < x < 0 =
f(x) =
1
x
<
1
=
1
1
2M
= 2M < M. Hence, lim
x0
1
x
= , as desired.
EXERCISE 20.16
Suppose that the limits L
1
= lim
xa
+ f
1
(x) and L
2
= lim
xa
+ f
1
(x) exist.
(a) Claim:
If f
1
(x) f
2
(x) x in some interal (a, b), then L
1
L
2
.
43
Proof :
By contradiction. Let f
1
(x) f
2
(x) and assume L
1
> L
2
. Corollary 20.8 says
lim
xa
+
f(x) = L > 0 > 0 : a < x < a + = f(x) L <
Dene :=
1
2
(L
1
L
2
) = for > 0 :
1
: x (a, a +
1
), f
1
(x) > L
1
= L
1
_
1
2
(L
1
L
2
)
_
=
1
2
(L
1
+L
2
)
2
: x (a, a +
2
), f
2
(x) < L
2
+ = L
2
+
_
1
2
(L
1
L
2
)
_
=
1
2
(L
1
+L
2
).
Let := min {
1
,
2
} =
f
1
(x) >
1
2
(L
1
+L
2
) > f
2
(x),
which is a contradiction. Hence, if f
1
(x) f
2
(x) x in some interal (a, b), then
L
1
L
2
, as desired.
(b) Suppose f
1
(x) < f
2
(x) x in some interval (a, b). Can you conclude that
L
1
< L
2
.
Consider f
1
(x) = x and f
2
(x) = 2x, x (a, b) = f
1
(x) < f
2
(x) x (a, b).
Consider a = 0 and b = 1 = (a, b) = (0, 1). We can plainly see that
lim
x0
f
1
(x) = lim
x0
f
2
(x) = 0
22
,
a contradiction = we cannot conclude that L
1
< L
2
if f
1
(x) < f
2
(x) x in
some interval (a, b).
EXERCISE 20.18
Claim:
For f(x) =
1 + 3x
2
1
x
2
, lim
x0
f(x) =
3
2
.
Proof :
Nonformal. Rearranging f(x),
1 + 3x
2
1
x
2
=
1 + 3x
2
1
x
2
1 + 3x
2
+ 1
1 + 3x
2
+ 1
=
(1 + 3x
2
) 1
(
1 + 3x
2
+ 1)x
2
=
3
1 + 3x
2
+ 1
,
which we see is a composition of continuous functions which happen to behave
well near x = 0. Hence, lim
x0
f(x) f(0) =
3
2
, as desired.
22
Since both f
1
and f
2
extend continuously to 0 where they take on the value 0.
44
EXERCISE 21.6
Let (S
1
, d
1
), (S
2
, d
2
), and (S
3
, d
3
) be metric spaces.
Claim:
f : S
1
S
2
and g : S
2
S
3
continuous = g f continuous from S
1
into S
3
Proof :
From both the claim and Theorem 21.3, f
1
(U) is an open subset of S
1
for every
open subset U of S
2
, i.e., f
1
(U) = {s S
1
: f(s) U} and g
1
(V ) is an open
subset of S
3
for every open subset V of S
3
, i.e., g
1
(V ) = {s S
3
: f(s) V }.
Dene a new mapping, h(S
1
) = g f : S
1
S
3
. Then h
1
(V ) is an open subset
of S
1
for every open subset V of S
3
, i.e. h
1
(V ) = {s S
1
: h(S) V }, by the
continuity of composite functions outlined in Theorem 17.5 and the fact that
g and f are both coninuous throughout S. Thus, by Theorems 17.5 and 21.3,
h(S
1
) := (g f)(S
1
) : S
1
S
3
is continuous, as desired.
EXERCISE 21.10 (b) and 21.11 (b)
For 21.10 (b), in order to show continuous functions mapping (0, 1) R,
consider
f(x) =
log(2x)
1 x
As x 0, f(x) and as x 1, f(x) +. We see f(x) is the com
position of continuous functions and the denominator = 0 through fs domain
and is hence continuous, as desired.
For 21.11 (b), we can use informal contradiction:
f : [0, 1] R = R is compact, since [0, 1] is compact, by Theorem 21.4 (i).
But R is unbounded and thus cannot be compact, which is a contradiction.
Hence, there do not exist continuous functions mapping [0, 1] onto R.
EXERCISE 23.1 (b), (d), (f ) and (h)
(b)
_
x
n
_
n
=
x
n
n
n
=
_
1
n
_
n
x
n
. If a
n
=
_
1
n
_
n
, then limsup a
n

1
n
= 0.
Therefore, = 0, R = + and this series has a radius of convergence + and
hence an interval of convergence of (, +).
(d)
_
n3
3
n
_
x
n
. If a
n
=
n3
3
n
, then
a
n+1
a
n
=
(n+1)
3
3
n+1
3
n
n
3
, so lim
a
n+1
a
n
=
1
3
. There
fore, =
1
3
and R = 3. This series diverges for both x = 3 and x = 3, hence
the radius of convergence is 3 and the interval of convergence is (3, 3).
(f )
_
1
(n+1)
2
2
n
_
x
n
. If a
n
=
1
(n+1)
2
2
n
, then
a
n+1
a
n
=
(n+1)
2
2
n
(n+2)
2
2
n+1
, so lim
a
n+1
a
n
=
1
2
. Therefore, =
1
2
and R = 2. This series converges at both x = 2 and
45
x = 2, hence the radius of convergence is 2 and the interval of convergence is
[2, 2].
(h)
_
(1)
n
n
2
4
n
_
x
n
. If a
n
=
(1)
n
n
2
4
n
, then
a
n+1
a
n
=
(1)
n+1
(n+1)
2
4
n+1
n
2
4
n
(1)
n
, so lim
a
n+1
a
n
=
1
4
. Therefore, =
1
4
and R = 4. This series converges at both x = 4 and x = 4,
hence the radius of convergence is 4 and the interval of convergence is [4, 4].
EXERCISE 23.2
(a)
nx
n
. If a
n
=
n, then
a
n+1
a
n
=
n+1
n
, so lim
n+1
= 1. Therefore,
= 1 and R = 1. This series diverges at x = 1 and x = 1, hence the radius
of convergence is 1 and the interval of convergence is (1, 1).
(b)
1
n
n
x
n
. If a
n
=
1
n
n
, then a
n

1
n
=
1
n
1
and limsup
1
n
1
= 1.
At x = 1, the series converges with comparison to
1
n
p
with p > 1 and for
x = 1, the series diverges by the alternating series theorem since the limit of
a
n
 approaches 1 and not 0. Hence, the radius on convergence is 1 and the
interval of convergence is (1, 1].
(c)
x
n!
. When x 1, the series diverges since limx
n!
does not tend to
0 as n . Now for x < 1, the series converges absolutely by comparison
with
x
m23
. Therefore, = 1 and R = 1. This series diverges at x = 1 and
x = 1, hence the radius of convergence is 1 and the interval of convergence is
(1, 1).
(d)
3
n
n
x
2n+1
. For this series, the radius of convergence is:
R =
1
lim
_
3
n
n
_ 1
2n+1
=
1
3
lim(3n)
1
4n+2
=
1
3
.
When x =
1
3
, the series explodes in both directions
24
. Hence, the radius of
convergence is
1
3
and the interval of convergence is
_
1
3
,
1
3
_
.
EXERCISE 23.6 (b)
An example of such a series is
n>0
(x)
n
n
.
23
Note that
x
n!
=
a
m
x
m
with a
m
= 1 when m = n! and a
m
= 0 when m = n!.
24
The series turns into
3n
.
46
This series converges to ln(1 +x) when x < 1, diverges at x = 1 and con
verges at x = +1.
EXERCISE 23.8
f
n
(x) 0 since
n
1
sin nx
=
1
n
0 as n . But f
n
(x) = cos nx (1)
n
when x = , which has no limit.
EXERCISE 24.2
For x [0, ), let f
n
(x) =
x
n
.
(a) For f
n
(x) =
x
n
, lim
n
f
n
(x) = lim
n
x
n
= 0. Hence, f(x) = 0.
(b) YES.
Claim: f
n
f uniformly on [0, 1].
Proof :
Let > 0 be given. Let N :=
1
x
n
n
=
1
n
< , as desired.
(c) NO.
Claim: f
n
does not converge uniformly to f on [0, ).
Proof :
By contradiction. Using the above, let := 1 = an N :
x
n
< 1 n >
N = x < n n > N = x < n + for some > 0. But since x [0, ),
x is unbounded, contradicting x < n + for some > 0. Hence, f
n
does not
converge uniformly to f on [0, ), as desired.
EXERCISE 24.6
Let f
n
(x) =
_
x
1
n
_
2
for x [0, 1].
(a) YES.
Claim:
f
n
(x) =
_
x
1
n
_
2
for x [0, 1] converges pointwise on the set [0, 1].
Proof :
f
n
(x) =
_
x
1
n
_
2
= x
2
2x
n
+
1
n
2
With x [0, 1], let n grow arbitrarily large
= f
n
(x) x
2
, since with n large,
2x
n
0 and
1
n
2
0 = f
n
(x)
f(x) := x
2
. Hence, given x arbitrary, f
n
(x) converges pointwise for x [0, 1],
as desired.
The limit function will thus be f(x) = x
2
.
(b) YES.
47
Claim:
f
n
(x) =
_
x
1
n
_
2
for x [0, 1] converges uniformly on the set [0, 1].
Proof :
Let > 0 be given and N :=
1
2
. Then for x [0, 1], we have:
_
x
1
n
_
2
x
2
1 2xn
n
2
n
<
1
N
= ,
as desired. .
EXERCISE 24.14
Let f
n
(x) =
nx
1+n
2
x
2
(a) Claim:
f
n
0 pointwise on R.
Proof :
Let x R. Let > 0 be arbitrary. Let N := max
_
x
2
,
_
. Then for n > N,
f
n
0 =
nx
1 +n
2
x
2
x
1
n
+nx
2
x
1
n
+n
2
x
n
2
x
N
2
< .
Hence, f
n
(x) converges uniformly to zero, as desired.
(b) Claim:
f
n
(x) does not converge to 0 uniformly on [0, 1].
Proof :
Implement Remark 24.4. Hence,
f
n
(x) =
n
1 +n
2
x
2
nxn
2
2x
(1 +n
2
x
2
)
2
=
n +n
3
x
2
2n
3
x
2
(1 +n
2
x
2
)
2
=
n n
3
x
2
(1 +n
2
x
2
)
2
Assume, for contradiction,
nn
3
x
2
(1+n
2
x
2
)
2
0, which implies x
2
=
1
n
2
which implies
x =
1
n
. But we see that f
n
_
1
n
_
=
1
2
= 0, a contradiction, since x [0, 1]
satises x =
1
n
. Hence, f
n
(x) does not converge to 0 uniformly on [0, 1] as
desired.
(c) Claim:
f
n
(x) converges to 0 uniformly on [1, ).
48
Proof :
Implement Remark 24.4. With x [1, ), x is unable to always satisfy
1
n
.
Using methods from calculus we can see that f
n
(x) =
nx
1+n
2
x
2
assumes its max
imum at x = 1, which is [1, ). Since f
n
(1) =
n
1+n
2
0 as n , f
n
(x)
converges to 0 uniformly on [1, ) as desired.
EXERCISE 25.2
Let f
n
(x) =
x
n
n
.
Claim: (f
n
) is uniformly convergent on [1, 1].
Proof :
Let > 0 be given. Let N :=
1
x
n
n
0
1
n
25
1
N
= ,
as desired.
The limit function is f
n
(x) f(x) = 0 for large n.
EXERCISE 25.6
(a) Show that if
a
k
 < , then
a
k
x
k
converges uniformly on [0, 1] to a
continuous function.
Implement Theorem 25.5 and the Weierstrass Mtest. Since
a
k
 <
and that
a
k
x
k
a
k
because x [0, 1], we know
a
k
x
k
converges
uniformly on S. Now since the series converges uniformly on S and a
k
x
k
is
continuous, then
a
k
x
k
represents a continuous function on S.
(b) Does
n=1
1
n
2
x
n
represent a continuous function on [1, 1]?
This is a series which converges at both x = 1 (by the alternating series test)
and at x = 1 (convergent pseries). Now consider the interval 1 a 1
and note that
n=1
1
n
2
a
n
converges. Since n
2
x
n
 n
2
a
n
 =
_
a
n
n
2
_
for
x [a, a], the Weierstrass Mtest shows that the series
n=1
1
n
2
x
n
converges
uniformly to a function on [a, a]. Since a can be any number 1, we con
clude that f represents a continuous function on [1, 1]
26
.
25
This function takes its max value at x = 1.
26
Note:
n=1
1
n
2
x
n
2
6
for x [1, 1].
49
EXERCISE 25.14
Claim:
If
g
k
converges uniformly on a set S and if h is a bounded function on S,
then
hg
k
converges uniformly on S.
Proof :
If the series
g
k
converges uniformly on a set S, then
g
k
is uniformly Cauchy
on S. If h is a bounded function on S, then an M : h M. Let > 0 be
given and let N :=
M
. Then
n m > N =
k=m
hg
k
k=m
Mg
k
= M
k=m
g
k
M
M
= ,
as desired.
EXERCISE 27.2
Show that if f is continuous on R, then there exists a sequence (p
n
) of polyno
mials such that p
n
f uniformly on each bounded subset of R. Hint: Arrange
for f(x) p
n
(x) <
1
n
for x n.
Claim:
If f is continuous on R, then there exists a sequence (p
n
) of polynomials such
that p
n
f uniformly on each bounded subset of R.
Proof :
Given x n, suppose I = [n, n], a closed and bounded interval; hence x I
and f : I R is a continuous function. Let g : [0, 1] [n, n] be a bijective
map dened by g(x) = n + x(2n), and hence continuous, i.e., g(0) = n and
g(1) = n. Since f is continuous, the composite function, f g : [0, 1] R
is continuous. Hence, for any > 0, N > 0 such that for any n N, the
Bernstein Polynomial B
n
(f g) satises
(f g)(x) B
n
(f g)(x) < x [0, 1]
Now g is a continuous injective map and so g has a continuous inverse function
dened by
g
1
(x) =
x +n
2n
x [n, n]
Thus, for all x [n, n], f(x) B
N
(f g)(g
1
(x)) < .
Hence,
f(x) B
N
(f g)
_
x +n
2n
_
< x [n, n]
50
Since B
N
(fg) is a polynomial function, p
(x) = B
N
(fg)
_
x+n
2n
_
is a polynomial
function in x and f(x) p
k=0
f g
_
k
n
__
n
k
__
x +n
2n
_
k
_
1
_
x +n
2n
__
nk
(36)
=
n
k=0
(f g)
_
k
n
__
n
k
__
x +n
2n
_
k
_
n x
2n
_
nk
(37)
=
n
k=0
f
_
a +
k
n
(2n)
__
n
k
__
x +n
2n
_
k
_
n x
2n
_
nk
(38)
It then follows from (f g)(x) B
n
(f g)(x) < x [0, 1] that q
n
f
uniformly on [n, n], as desired.
EXERCISE 27.6
Claim:
If B
n
f f uniformly on [0, 1], then f is continuous on [0, 1].
Proof :
If B
n
(f) f uniformly on [0, 1], then
> 0 N x [0, 1] n > N : B
n
f(x) f(x) <
2
Let > 0 be given. Let N := . Then x [0, 1] and x x
0
 < =
f(x)f(x
0
) = B
n
f(x)f(x
0
)+f(x)B
n
f(x) B
n
f(x)f(x
0
)+B
n
f(x)f(x)
2
+
2
= ,
as desired.
EXERCISE 28.2
Use the denition of the derivative to calculate the derivatives of the following
functions at the indicated points.
(a) f(x) = x
3
at x = 2.
f
(2) = lim
x2
x
3
8
x 8
= lim
x2
(x 2)(x
2
+ 2x + 4)
x 2
= lim
x2
x
2
+2x+4 = (2)
2
+2(2)+4 = 12
51
(b) g(x) = x + 2 at x = a.
f
(a) = lim
xa
(x + 2) (a + 2)
x a
= lim
xa
x + 2 a 2
x a
= lim
xa
x a
x a
= 1
(c) f(x) = x
2
cos x at x = 0.
f
(0) = lim
x0
x
2
cos x (0)
2
cos(0)
x 0
= lim
x0
x
2
cos x 0
x 0
= lim
x0
xcos x = 0 1 = 0
(d) r(x) =
3x+4
2x1
at x = 1
r
(1) = lim
x1
3x+4
2x1
7
1
x 1
= lim
x1
11x+11
2x1
x 1
= lim
x1
11x + 11
2x 1
1
x 1
= lim
x1
11
2x 1
= 11
EXERCISE 28.4
Let f(x) = x
2
sin
1
x
for x = 0 and f(0) = 0.
(a) Use Theorems 28.3 and 28.4 to show that f is dierentiable at each a = 0
and calculate f
(a). Use, without proof, the fact that sin x is dierentiable and
that cos x is its derivative.
Using the denition, we see that
f
(a) = lim
xa
x
2
sin
_
1
x
_
a
2
sin
_
1
a
_
x a
= lim
xa
x
2
sin
_
1
x
_
sin
_
1
a
_
x a
+ sin
_
1
a
_
x
2
a
2
x a
,
where the
sin(
1
x
)sin(
1
a
)
xa
limit represents sin
_
1
x
_
. We are given that a = 0,
which gives the function sin
_
1
a
_
some meaning in terms of dierentiability and
hence, using Theorems 28.3 and 28.4 and the given fact that sin
x = cos x,
f
(a) = a
2
sin
_
1
a
_
+ sin
_
1
a
_
(a
2
)
= a
2
cos
_
1
a
__
1
a
2
_
+ 2a sin
_
1
a
_
= 2a sin
_
1
a
_
cos
_
1
a
_
.
(b) Use the denition to show that f is dierentiable at x = 0 and that f
(0) =
0.
lim
x0
f(x) f(0)
x 0
= lim
x0
f(x)
x
= lim
x0
x
2
sin
1
x
x
= lim
x0
xsin
1
x
= 0 ,
where  1, and hence the whole expression equals zero, as desired.
(c) Show that f
is not continuous at x = 0.
52
Implement Theorem 28.4, which implies f is dierentiable everywhere. Further,
f
(x) = xsin
_
1
x
_
cos
_
1
x
_
at x = 0. We know xsin
_
1
x
_
0 as x 0 and
cos
_
1
x
_
doesnt have a limit at x = 0. Hence, the sum f
k=0
_
n
k
_
f
(k)
(a)g
(nk)
(a)
Proof :
By Induction on n
27
. Let Leibniz Rule hold for n = m. Then
(fg)
(m+1)
= (f
g +fg
)
(m)
=
m
k=0
_
m
k
_
f
(k+1)
g
(mk)
+
m
k=0
_
m
k
_
f
(k)
g
(m+1k)
=
m+1
k=0
__
m
k 1
_
+
_
m
k
__
f
(k)
g
(m+1k)
=
m+1
k=0
_
m+ 1
k
_
f
(k)
g
(m+1k)
,
where, using Pascals triangle
28
(from the Binomial Theorem handout), the last
equality holds. Hence, Leibniz Rule holds true, as desired.
EXERCISE 29.4
Let f and g be dierentiable functions on an open interval I. Suppose that a, b
in I satisfy a < b and f(a) = f(b) = 0.
Claim:
f
(x) +f(x)g
(x) = 0.
Dierentiating h(x) yields:
h
(x) = f(x)g
(x)e
g(x)
+f
(x)e
g(x)
= e
g(x)
(f(x)g
(x) +f
(x)) .
Given h
(x) +f
(x)+f
(x) = 0
for some x (a, b), as desired.
27
For n = 1, Leibniz Rule turns into the product rule, i.e., (fg)
= f
g + fg
.
28
m+1
k
m
k1
m
k
.
54
EXERCISE 29.8
Claim:
f is strictly decreasing is f
(x) < 0.
Since x
2
x
1
> 0 and f(x
2
) f(x
1
) < 0 = f(x
2
) < f(x
1
), as desired.
Claim:
f is increasing if f
(x) 0.
Since x
2
x
1
> 0 and f(x
2
) f(x
1
) 0 = f(x
2
) f(x
1
), as desired.
Claim:
f is decreasing if f
(x) 0.
Since x
2
x
1
> 0 and f(x
2
) f(x
1
) 0 = f(x
2
) f(x
1
), as desired.
EXERCISE 29.10
Let f(x) =
x
2
+x
2
sin
1
x
at x = 0, and f(0) = 0.
Claim:
f
(0) > 0, but f is not increasing on any interval containing 0. Compare this
result with Theorem 29.7 (i).
55
Proof :
We know
f
(0) = lim
x0
f(x)
x
= lim
x0
_
1
2
+xsin
1
x
_
=
1
2
+ 0 > 0
Conversely, at x = 0, we get
f
(x) = 2xsin
1
x
+
1
2
cos
1
x
.
The rst term 0 as x 0. The last term oscillates between +1 and 1
innitely many times in any neighborhood of x = 0.
1
2
1 < 0 = in any
neighborhod of x = 0, f
at
x = 0, as desired.
EXERCISE 29.15
We know that (x
m
)
= mx
m1
for m 0,
_
1
x
_
=
1
x
2
and
_
x
1
n
_
= x
1
n
1
. We
can thus calculate
_
x
m
n
_
y=x
m
mx
m1
=
m
n
x
m(
1
n
1)
x
m1
=
m
n
x
m
n
1
.
EXERCISE 29.18
Let f be be dierentiable on R with a = sup {f
(y)(s
n
s
n1
) as
n
s
n1
,
which implies
s
n+1
s
n
 a
n
s
1
s
0
, n > 0,
by induction. Further,
s
m+1
s
n

m
k=n
s
k+1
s
k
 (s
1
s
0
)
m
k=n
a
k
, m n > 0.
56
We know the geometric series
a
k
converges if a < 1 and further, its partial
sums,
n
k=0
form a Cauchy Sequence. Therefore, by the previous estimate, (s
n
)
is a Cauchy sequence, and hence converges, as desired.
EXERCISE 30.2
Find the following limits if they exist.
(a) lim
x0
x
3
sin xx
=
0
0
, which is an indeterminant form, so we apply LHospitals
rule and get lim
x0
3x
2
cos x1
=
0
0
, which is an indeterminant form, so we apply
LHospitals rule and get lim
x0
6x
sin x
= 6 lim
x0
x
sin x
= 6 1 = 6.
(b) lim
x0
x
x
3
=
0
0
, which is an indeterminant form, so we apply LHospitals
rule and get lim
x0
sec
2
x1
3x
2
=
0
0
, which is an indeterminant form, so we apply
LHospitals rule and get lim
x0
2 sec
2
x tan x
6x
=
0
0
, which is an indeterminant
form, so we apply LHospitals rule and get lim
x0
sec
4
x+2 sec
2
x tan
2
x
3
=
1
3
(c) lim
x0
_
1
sin x
1
x
= lim
x0
xsin x
x sin x
=
0
0
, which is an
indeterminant form, so we apply LHospitals rule and get lim
x0
1cos x
x cos x+sin x
=
0
0
, which is an indeterminant form, so we apply LHospitals rule and get
lim
x0
sin x
2 cos xx sin x
=
0
2
= 0.
(d) lim
x0
(cos x)
1
x
2
= 1
e
. Therefore, the limit as x 0 =
1
e
.
EXERCISE 30.4
Let f be a function dened on some interval (0, a), and dene g(y) = f
_
1
y
_
for
y (a
1
, ); here we set a
1
= 0 if a = .
Claim:
lim
x0
+ f(x) exists if and only if lim
y
g(y) exists, in which case they are
equal.
Proof :
In two parts.
lim
x0
+ f(x) exists = lim
y
g(y) exists.
57
Assume lim
x0
+ f(x) exists and is equal to L. Then for each > 0 > 0 : 0 <
x < = f(x) L < . Now, dene y :=
1
x
. Then 0 < x < =
1
< y < .
Then for f dened on an interval (c, ), for each > 0 < : <
1
y
=
f
_
1
y
_
L
< = lim
y
g(y) exists and is equal to L, as desired.
lim
y
g(y) = lim
x0
+ f(x) exists.
lim
y
g(y) = lim
y
f
_
1
y
_
, which when implemented with the same trans
formation, y :=
1
x
= x =
1
y
, we see that lim
y
f
_
1
y
_
lim
x0
+ f(x) and
hence lim
x0
+ f(x) exists and is nite.
EXERCISE 30.7
For x R, let
f(x) = x + cos xsin x and g(x) = e
sin x
(x + cos xsin x).
(a) Since  cos x 1 and  sin x 1 =  cos xsin x 1, we can conclude
lim
x
f(x) = x + cos xsin x lim
x
x 1 = +,
and hence, lim
x
f(x) = x+cos xsin x = +. Since we know
1
e
lim
x
e
sin x
e, this implies
lim
x
1
e
(x + cos xsin x) lim
x
e
sin x
(x + cos xsin x) lim
x
e(x + cos xsin x)
which implies
1
e
lim
x
e
sin x
(x + cos xsin x) e
which implies lim
x
e
sin x
(x + cos xsin x) = +, by the Squeeze Theorem.
(b) Implement Theorem 28.3 and the trigonometric identity sin
2
x+cos
2
x = 1.
f(x) = x + cos xsin x
f
(x) = e
sin x
(2 cos
2
x) + (x + cos xsin x) cos xe
sin x
= e
sin x
(2 cos
2
x) +e
sin x
cos x[f(x)]
= e
sin x
cos x[2 cos x +f(x)]
58
f
(x)
g
(x)
=
2(cos x)
2
e
sin x
cos x[2 cos x +f(x)]
=
2e
sin x
cos
2
x
cos x[2 cos x +f(x)]
=
2e
sin x
cos x
2 cos x +f(x)
(d) Using the fact that  sin x 1 and  cos x 1, we see that
lim
x
2 cos x
e
sin x
(2 cos x +x + cos xsin x)
lim
x
2
2e +xe +e
lim
x
2
e(3 +x)
0 as x ,
so we can conclude that
f
(x)
g
(x)
tends to 0 for large x. However,
lim
x
x + cos xsin x
e
sin x
(x + cos xsin x)
= lim
x
1
e
sin x
,
which will oscillate between e and
1
e
as x , which shows that lim
x
f(x)
g(x)
doesnt exist.
EXERCISE 31.2
sinh x =
n1
x
2n1
(2n 1)!
and cosh x =
n0
x
2n
(2n)!
.
Both of these results follow from the series expansion for e
x
x
n
n!
and the
fact that (sinh x)
x
2n+2
(2n + 2)!
(2n)!
x
2n
= limsup
x
2
2n(2n + 1)
0 < 1,
limsup
x
2n+1
(2n + 1)!
(2n 1)!
x
2n1
= limsup
x
2
(2n + 1)(2n + 2)
0 < 1.
EXERCISE 32.2
Let f(x) = x for rational x and f(x) = 0 for irrational x.
(a) To calculate the upper Darboux integral for f on the interval [0, b], we need
to come up with an upper bound. Consider the partition
P = {0 = t
0
< t
1
< < t
n
= b} where t
k
= b
k
n
for each k.
59
This implies
U(f, P) =
n
k=1
kb
n
b
n
29
=
_
b
n
_
2 n
k=1
k =
_
b
n
_
2
n
2
+n
2
=
1
2
b
2
(1 + 2n
1
).
1
2
b
2
(1 + 2n
1
)
1
2
b
2
as n = U(f)
1
2
b
2
.
We now need to come up with a lower bound. Consider the partition
P = {0 = t
0
< t
1
< < t
n
< b}
which implies
U(f, P) =
n
k=1
t
k
(t
k
t
k1
)
30
k=1
(t
k
+t
k1
)
2
(t
k
t
k1
)
31
=
1
2
n
k=1
(t
2
k
t
2
k1
)
32
=
1
2
b
2
which implies U(f) =
1
2
b
2
.
To calculate the lower Darboux integral for f on the interval [0, b], for each
partition
P = {0 = t
0
< t
1
< < t
n
< b} ,
an irrational number in [t
i1
, t
i
] for each i, which implies the minimal value
of f on [t
i1
, t
i
] is 0 = L(f, P) = 0 for each P = L(f) = 0.
(b) Assume b > 0 = U(f) = L(f) = f is not integrable on [0, b], since the
Theorems of this chapter wont hold for a degenerate interval.
EXERCISE 32.6
Let f be a bounded function on [a, b]. Suppose there exist sequences (U
n
) and
(L
n
) of upper and lower Darboux sums for f such that lim(U
n
L
n
) = 0. Show
f is integrable and
_
b
a
f = limU
n
= limL
n
.
We want to show that limL
n
= limU
n
. Assume that limL
n
< limU
n
. Then
lim(U
n
L
n
) > 0 which is a contradiction. Observe that L
n
L(f) U(f)
U
n
. Taking limits of both sides and implementing the Squeeze Theorem yields
limL
n
L(f) U(f) limU
n
= limL
n
. Thus L(f) = U(f) by the Squeeze
Theorem which implies f is integrable by Theorem 32.9.
29
Each interval has length
b
n
; on the interval [t
i1
, t
i
], the maximum value of f is attained
at t
i
=
i
n
.
30
Each interval has length
b
n
; the supremum of f on [t
i1
, t
i
] is t
i
since you can pick rational
numbers in the interval arbitrarily close to t
i
.
31
Since t
k
> t
k1
, t
k
>
t
k
+t
k1
2
.
32
Note that this is a telescoping series, lol.
60
EXERCISE 33.3 (a)
A function f on [a, b] is called a stepfunction if a partition P = {a = u
0
< u
1
< < u
m
= b}
of [a, b] such that f is constant on each interval (u
j1
, u
j
), say f(x) = c
j
for x
in (u
j1
, u
j
).
Claim:
f is integrable.
Proof :
Consider a subpartition of P, called P
, where
P
=
_
u
0
< u
1
< u
1
< u
1
< < u
n1
< u
n1
< u
n1
< u
n
_
,
where u
i
u
i
< i.
We want to show U(f, P) S < and S L(f, P) < . From P
,
U(f, P
) L(f, P
) =
(u
i
u
i
) c
i
c
i1
 =
c
i
c
i1
 max {c
i
} ,
which will tend to 0 as tends to 0. Hence,
L(f, P
)
n
j=1
c
j
(u
j
u
j1
) U(f, P
),
and since L(f, P
) = U(f, P
) from above,
_
b
a
f =
n
j=1
c
j
(u
j
u
j1
), as de
sired.
EXERCISE 33.4
Give an example of a function f on [0, 1] that is not integrable for which f is
integrable.
Consider the intervale [0, 1] and let f(x) = 1 for rational x [0, 1], and let
f(x) = 1 for irrational x [0, 1]. For any partition
P = {0 = t
0
< t
1
< < t
n
= 1} ,
we have
U(f, P) =
n
k=1
M(f, [t
k1
, t
k
]) (t
k
t
k1
) =
n
k=1
1 (t
k
t
k1
) = 1
and
L(f, P) =
n
k=1
(1) (t
k
t
k1
) = 1.
61
It follows that U(f) = L(f) which implies that f is not integrable. However, if
the absolute value of f is taken, we see that f(x) = 1 x R, and hence will
be integrable since U(f) = L(f).
EXERCISE 33.7
Let f be a bounded function on [a, b], so that B > 0 : f(x) B x [a, b].
(a)Claim:
U(f
2
, P) L(f
2
, P) 2B
_
U(f, P) L(f, P)
_
partitions P of [a, b]
Proof :
n
k=1
_
M(f
2
, [t
k1
, t
k
]) m(f
2
, [t
k1
, t
k
)
(t
k
t
k1
)
=
n
k=1
_
[M(f, [t
k1
, t
k
]) +m(f, [t
k1
, t
k
)] [M(f, [t
k1
, t
k
]) m(f, [t
k1
, t
k
)]
_
(t
k
t
k1
)
2B
n
k=1
_
M(f, [t
k1
, t
k
]) m(f, [t
k1
, t
k
])
_
(t
k
t
k1
),
as desired
33
. .
(b)Claim:
If f is integrable on [a, b], then f
2
is integrable on [a, b].
Proof :
From part (a), we know
U(f
2
, P) L(f
2
, P) 2B
_
U(f, P) L(f, P)
_
partitions P of [a, b],
and by Theorem 32.5, if f is integrable on [a, b], then for > 0, U(f, P)
L(f, P) < .
But,
U(f
2
, P)L(f
2
, P) 2B
_
U(f, P)L(f, P)
_
< =
U(f
2
, P) L(f
2
, P)
2B
< ,
which implies
U(f
2
, P) L(f
2
, P) < 2B.
So dene a new ,
,
33
The rst equality in this proof is from factoring the perfect squares.
62
which implies that if f is integrable on [a, b], then f
2
is integrable on [a, b], as
desired.
EXERCISE 33.8
Let f and g be integrable functions on [a, b].
(a) Claim:
fg is integrable on [a, b].
Proof :
Implement Theorem 33.3. Let f and g be integrable on [a, b]. Then, by Theo
rem 33.3, f + g and f g are integrable on [a, b]. Further, if f + g and f g
are integrable on [a, b], then (f +g)
2
and (f g)
2
are both integrable on [a, b],
as per the results of exercise 33.7 (b). Since
1
4
_
(f +g)
2
(f g)
2
= fg, and
Theorem 33.3 tells us a constant times an integrable function is integrable, this
shows that if f and g are integrable on [a, b], then fg is integrable on [a, b], as
desired.
(b) Claim:
max(f, g) and min(f, g) are integrable on [a, b].
Proof :
Implement Theorem(s) 33.3 and 33.5. We know
max(f, g) =
1
2
(f +g) +
1
2
f g
min(f, g) =
1
2
(f +g)
1
2
f g,
which are compositions of functions and constants integrable on [a, b] and hence
are integrable by Theorems 33.3 and 33.5, as desired. .
EXERCISE 33.10
Let f(x) = sin
_
1
x
_
for x = 0 and f(0) = 0. Show that f is integrable on [1, 1].
Let > 0 be given. Since f is piecewise continuous, by Denition 33.7, on [
4
, 1],
a partition P
1
of [
4
, 1] : U(f
1
, P)L(f
1
, P) <
2
. Similarly, a partition P
2
of
[1,
4
] : U(f
2
, P) L(f
2
, P) <
2
. Dene P = P
[
1 P
2
, a partition of [1, 1].
Since _
M(f,
_
4
,
4
_
) m(f,
_
4
,
4
_
)
_
4
_
4
__
< ,
which, when combined with Theorem 32.5, shows f is integrable on [1, 1].
63
EXERCISE 33.14
Suppose f and g are continuous functions on [a, b] and that g(x) 0 x [a, b].
(a) Claim:
x [a, b] :
_
b
a
f(t)g(t)dt = f(x)
_
b
a
g(t)dt.
Proof :
Given that
_
b
a
f(t)g(t)dt =
i=1
f(t
i
)g(t
i
) (t
i
t
i1
),
where t
i
[u
i1
, u
i
] and f(t
i
) = a
i
and g(t
i
) = b
i
. Further,
a
i
b
i
min {a
i
} b
i
= a
i
min {a
i
}
a
i
b
i
max {a
i
} b
i
= a
i
max {a
i
} .
Then,
_
b
a
f(t)g(t)dt f
min
_
b
a
g(t)dt
_
b
a
f(t)g(t)dt f
max
_
b
a
g(t)dt.
Note: if b
i
= 0 then the equality holds, trivially. Hence, by the Intermediate
Value Theorem, x [a, b] :
_
b
a
f(t)g(t)dt = f(x)
_
b
a
g(t)dt., as desired. .
(b) Let g(x) :=
1
ba
. Then
1
b a
_
b
a
f(t)dt =
_
b
a
f(t)g(t)dt = f(x)
_
b
a
g(t)dt = f(x) 1 = f(x).
EXERCISE 34.2
(a)
lim
x0
1
x
_
x
0
e
t
2
dt
To calculate this integral, let us use the formulation
F(x) F(x
0
)
x x
0
=
1
x x
0
_
x
x
0
f(t)dt for x = x
0
Hence,
f(x
0
) =
1
x x
0
_
x
x
0
f(x
0
)dt.
64
Thus, by the Fundamental Theorem of Calculus II,
lim
x0
1
x
_
x
0
e
t
2
dt = e
0
2
= 1.
(b)
lim
h0
1
h
_
3+h
3
e
t
2
dt
Using the same argument as above,
lim
h0
1
h
_
3+h
3
e
t
2
dt = lim
h0
1
h
_
3
0
e
(t+3)
2
dt = e
3
2
= e
9
.
EXERCISE 34.6
Let f be a continuous function on R and dene
G(x) =
_
sin x
0
f(t)dt for x R.
If f is continuous at x R, then f(sin x) is dierentiable on R as a com
position of dierentiable functions. Hence, f is dierentiable at sin x and
G
_
d
a
f(x)dx =
_
b
a
f(x)dx.
It suces to show that if f is bounded by some number B, then
_
d
a
f(x)dx
_
b
a
f(x)dx
B(b d).
Hence, we want to show that for b d < = B(b d) < . Choose :=
B
.
Then
B(b d) = B(b d) < B = .
65
Hence as d approaches b from the left, the dierence between the integrals con
verges to zero, showing they are equivalent.
EXERCISE 36.6
Let f and g be continuous functions on (a, b) : 0 f(x) g(x) x (a, b); a
can be and b can be +.
(a) Claim:
_
b
a
g(x)dx < =
_
b
a
f(x)dx <
Proof :
We know
_
b
a
f(x)dx =
_
0
a
f(x)dx +
_
b
0
f(x)dx,
by Theorem 33.6. Likewise,
_
b
a
g(x)dx =
_
0
a
g(x)dx +
_
b
0
g(x)dx.
From the fact that 0 f(x) g(x) x (a, b), we can assume
_
_
0
a
f(x)dx +
_
b
0
f(x)dx
_
_
_
0
a
g(x)dx +
_
b
0
g(x)dx
_
.
Implementing Denition 36.1 and Theorem 19.6 (Extensions from bounded in
tervals to unbounded intervals being uniformly continuous and hence integrable):
_
lim
a
_
0
a
f(x)dx + lim
b+
_
b
0
f(x)dx
_
_
lim
a
_
0
a
g(x)dx + lim
b+
_
b
0
g(x)dx
_
.
Dene a continuous function h(x) := g(x) f(x) = h(x) 0, since g(x)
f(x). Hence,
_
h(x)dx :=
_
_
g(x) f(x)
_
dx =
_
g(x)dx =
_
_
h(x) +f(x)
_
dx.
We can now see that
_
g(x)dx < =
_
_
h(x)+f(x)
_
dx < =
_
f(x)dx < , since
_
h(x)dx 0,
as desired.
(b) Claim:
_
b
a
f(x)dx = =
_
b
a
g(x)dx =
66
Proof :
Using the result from part (a),
_
f(x)dx =
_
g(x)dx
_
h(x)dx,
and therefore,
_
f(x)dx = =
_
g(x)dx
_
h(x)dx = =
_
g(x)dx = ,
as desired. .
67