100% found this document useful (5 votes)
7K views9 pages

Bartle Intro To Real Analysis Solutions CH 8 PDF

The document contains solutions to problems from Chapter 8 of Bartle's Introduction to Real Analysis textbook. Problem 8.1-2 shows that the limit of nx/(1+n^2x^2) is 0 for all real x. Problem 8.1-3 evaluates the limit of nx/(1+nx) as x approaches 0 and infinity. Problem 8.1-4 similarly evaluates the limit of x^n/(1+x^n). Problem 8.1-9 shows that the limits of both x^2e^-nx and n^2x^2e^-nx are 0 as x approaches infinity. Problem 8.1-10 finds the limit of cos(πx)^

Uploaded by

doni ajah
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (5 votes)
7K views9 pages

Bartle Intro To Real Analysis Solutions CH 8 PDF

The document contains solutions to problems from Chapter 8 of Bartle's Introduction to Real Analysis textbook. Problem 8.1-2 shows that the limit of nx/(1+n^2x^2) is 0 for all real x. Problem 8.1-3 evaluates the limit of nx/(1+nx) as x approaches 0 and infinity. Problem 8.1-4 similarly evaluates the limit of x^n/(1+x^n). Problem 8.1-9 shows that the limits of both x^2e^-nx and n^2x^2e^-nx are 0 as x approaches infinity. Problem 8.1-10 finds the limit of cos(πx)^

Uploaded by

doni ajah
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

Bartle - Introduction to Real Analysis - Chapter 8 Solutions

Section 8.1
Problem 8.1-2. Show that lim(nx/(1 + n2 x2 )) = 0 for all x ∈ R.

Solution: For x = 0, we have lim(nx/(1 + n2 x2 )) = lim(0/1) = 0, so f (0) = 0. For x ∈ R\{0}, observe that
0 < nx/(1 + n2 x2 ) < nx/(n2 x2 ) = 1/(nx). By the Squeeze Theorem, lim(nx/(1 + n2 x2 )) = 0. Therefore, f (x) = 0 for
all x ∈ R.

Problem 8.1-3. Evaluate lim(nx/(1 + nx)) for x ∈ R, x ≥ 0.

Solution: For x = 0, we have lim(nx/(1 + nx)) = lim(0/1) = 0, so f (0) = 0.


For x ∈ (0, ∞), we have:
   
nx 1 1
lim = lim = = 1,
1 + nx 1/nx + 1 1/x lim(1/n) + 1
from which it follows that f (x) = 1 for x ∈ (0, ∞). Therefore,

0 for x=0
f (x) =
1 for x > 0.

Problem 8.1-4. Evaluate lim(xn /(1 + xn )) for x ∈ R, x ≥ 0.

Solution: For 0 ≤ x < 1, we have lim(xn /(1 + xn )) = 0/1 = 0 by Example 3.1.11(b), so f (x) = 0. For x = 1, we have
lim(xn /(1 + xn )) = 1/2, so f (1) = 1/2. For x > 1, we have lim(xn /(1 + xn )) = lim(1/(1 + 1/xn ) = 1, so f (x) = 1.*
Accordingly, 
 0 for 0 ≤ x < 1
1
f (x) = for x = 1
 2
1 for x > 1.
* Note that for 1/xn with fixed x, given  > 0, if K() = logx (2/), then for n ≥ K(), we have |1/xn | = 1/xn <
1/(2/) = /2 < . Therefore, lim(1/xn ) = 0.

Problem 8.1-9. Show that lim(x2 e−nx ) = 0 and that lim(n2 x2 e−nx ) = 0 for x ∈ R, x ≥ 0.

Solution: Part (i): For x = 0, we have lim(x2 e−nx ) = lim(0·1n ) = 0, so f (0) = 0. For x > 0, observe that 0 < e−x < 1.
From Example 3.1.11(b), it follows that lim(x2 e−nx ) = x2 lim(e−x )n = 0. As a result, f (x) = 0 for x ≥ 0.
Part (ii): We can establish limit of (fn ) = (n2 x2 e−nx ) using L’Hopı̂tal’s Rule and the Sequential Criterion for limits
of functions. Let g(m) = m2 x2 e−mx = m2 x2 /emx . For x ∈ (0, ∞), the limit as m → ∞ is in ∞/∞ indeterminate form,
so we apply L’Hopı̂tal’s Rule twice:

m 2 x2 2mx2 2m2 2
lim mx
= lim mx
= lim 2 mx
= lim mx = 0.
n→∞ e m→∞ xe m→∞ m e m→∞ e

By the Sequential Criterion for limits of functions (Theorem 4.1.8), the limit of g above implies that for any sequence
(yn ) on (0, ∞) that converges to infinity, the sequence (g(yn )) converges to 0. If yn = n for all n ∈ N, then (g(yn )) =
(n2 x2 e−nx ), which is equal to (fn ). It follows that if x > 0, then lim n2 x2 e−nx = 0.
For x = 0, clearly lim n2 x2 e−nx = lim 0 = 0. Accordingly, if x ∈ [0, ∞), then (n2 x2 e−nx ) converges to f (x) = 0.

Problem 8.1-10. Show that lim(cos(πx)2n ) exists for all x ∈ R. What is its limit?
Solution: If x ∈ Z, then cos(πx)2n = (±1)2n = 1, so lim(cos(πx)2n ) = 1. Therefore, f (x) = 1.
If x ∈ R\Z, then 0 ≤ cos2 (πx) < 1, so by Example 3.1.11(b), lim[cos2 (πx)]n = 0. Therefore:

0 for x ∈ Z
f (x) =
1 for x ∈ R\Z.

Problem 8.1-10. Show that if a > 0, then the convergence of the sequence in Exercise 1 is uniform on the interval [0, a], but
is not uniform on the interval [0, ∞).

Solution: Let a > 0 and A = [0, a]. Because fn is continuous, it is bounded on A by Theorem 5.3.2. Suppose f (x) = 0
for x ∈ A. Then kfn − 0kA = sup{x/(x + n) : x ∈ A} = a/(a + n) because fn is increasing on A. Therefore,
lim kfn − 0kA = lim a/(a + n) = 0. By Lemma 8.1.8, (fn ) converges uniformly to f (x) = 0 on A.
Now let A = [0, ∞]. As shown in Theorem 1 below, if (fn ) is uniformly convergent on A, then it must converge
uniformly to f (x) = 0 because this sequence is pointwise convergent to that function on A. We see that kfn − 0kA =
sup{|x/(x + n)| : x ≥ 0} = 1. This is because 0 < x/(x + n) < 1, and for 0 < δ < 2, if x = n(2/δ − 1), then
1−δ < x/(x+n) = 1−δ/2 < 1. Therefore, 1 is the supremum of {|x/(x+n)| : x ≥ 0}. Consequently, lim kfn − 0kA = 1.
By Lemma 8.1.8, (fn ) does not uniformly converge to any f on [0, ∞).
Theorem 1. Suppose (fn ) converges pointwise to f on A ⊆ R. If (fn ) does not uniformly converge to f on A, then (fn )
does not uniformly converge to any function on A.
Proof. Suppose there is a function f 0 : A → R to which (fn ) converges uniformly on A. Now assume that f 0 6= f .
It follows that (fn ) must converges pointwise to f 0 on A. However, by Theorem 3.1.4, the limit function f is uniquely
determined, so we have a contradiction if f 0 6= f . Therefore, it must be that f = f 0 . Accordingly, if (fn ) does not converge
uniformly to f on A, it does not converge uniformly to any function on A.

Problem 8.1-12. Show that if a > 0, then the convergence of the sequence in Exercise 2 is uniform on the interval [a, ∞], but
is not uniform on the interval [0, ∞).

Solution: Let a > 0 and A = [a, ∞). From Exercise 8.1.2, √ (fn ) converges pointwise to f (x) = 0 on A. Observe that
fn0 (x) = n(1 − n2 x2 )/(1 + n2 x2 )2 . We see that if x0 = 1/ n, then fn0 (x0 ) = 0. On either side of this point, fn0 (x) > 0
for x < x0 and fn0 (x) < 0 for x > x0 . By Theorem 6.2.8, the maximum of fn on √ [0, ∞) is at x0 . Moreover, it is clear that
fn0 decreases on (x0 , ∞). It follows that the maximum of fn on A is b = sup{1/ n, a}. We then have:
 
nx
= sup nx : x ∈ A = bn


1 + n2 x2 − 0 1 + n2 x2 ,

A 1 + n2 b2
Since 0 < bn/(1 + n2 b2 ) < bn/(n2 b2 ) = 1/(nb) and lim 1/(nb) = 0, it follows from the Squeeze Theorem that
lim kfn − f kA = 0. By Lemma 8.1.8, (fn ) uniformly converges to f (x) = 0 on [a, ∞). (Note that the limit above holds
true even though b may be a function of n. For the sake of brevity, I cavalierly omitted this point in applying the Squeeze
Theorem.)
Now suppose A = [0, ∞). Let (xk ) be a sequence on A where xk = 1/k and nk = k. For  = 1/4:

k(1/k) 1
|fnk (xk ) − f (xk )| = = > .
1 + k 2 (1/k 2 ) 2
From Lemma 8.1.5, it follows that (fn ) does not converge uniformly on [0, ∞).

Problem 8.1-13. Show that if a > 0, then the convergence of the sequence in Exercise 3 is uniform on the interval [a, ∞), but
is not uniform on the interval [0, ∞).

Solution: Let a > 0 and A = [a, ∞). We know that (fn ) converges pointwise to f (x) = 1 on A. Observe that:
 
nx −1 1 1
kfn (x) − 1)kA = sup
− 1 =
= :x∈A = . (1)
1 + nx 1 + nx 1 + nx 1 + an
It follows that lim kfn − f kA = 0. Therefore, (fn ) uniformly converges to f (x) = 1 on [a, ∞).

Page 2
Now let A = [0, ∞). We then have:
   
: x ∈ (0, ∞) ∪ n · 0 − 0
nx
kfn (x) − f (x)kA = sup
1 + nx − 1 1 + n · 0 = 1. (2)

Consequently, lim kfn − f kA = 1, so by Lemma 8.1.7 (fn ) does not converge uniformly on A.

Problem 8.1-14. Show that if 0 < b < 1, then the convergence of the sequence in Exercise 4 is uniform on the interval [0, b],
but is not uniform on the interval [0, 1].

Solution: Let b ∈ (0, 1) be given and A = [0, b]. We then have:


xn bn
 
xn

1 + xn − 0 = sup
1 + xn : x inA = ,

A 1 + bn
because fn is increasing on A (since fn0 > 0 on that interval). Clearly lim kxn /(1 + xn ) − 0kA = (lim bn )/(1 + lim bn ) = 0
since n ∈ (0, 1). Therefore, (fn ) converges uniformly to f (x) = 0 on x ∈ [0, b].
Now let A = [0, 1]. Let (xk ) be a sequence in A where xk = 2−1/k and nk = k. It follows that for any  > 0 where
 < 1/3:

(2−1/k )k 1/2
|fnk (xk ) − f (xk )| = −1/k k
= = 1/3 > .
1 + (2 ) 1 + 1/2
By Lemma 8.1.5, (fn ) does not uniformly converge on [0, 1].

Problem 8.1-15. Show that if a > 0, then the convergence of the sequence in Exercise 5 is uniform on the interval [a, ∞), but
is not uniform on the interval [0, ∞).

Solution: Suppose a > 0 and A = [a, ∞). From Exercise 8, we know that (fn ) converges pointwise to f (x) = 0 on A. It is
clear that |sin(nx)| ≤ 1. Therefore, 0 ≤ sup{|sin(nx)/(1 + nx) − 0| : x ∈ A} ≤ 1/(1 + an) < 1/an. Since lim 1/an = 0,
it follows from the Squeeze Theorem that lim kfn − f kA = 0. By Lemma 8.1.8, (fn ) converges uniformly to f (x) = 0.
Now let A = [0, ∞). Let (xk ) be a sequence on A where xk = π/(2k) and nk = k. For any positive  where
 < 1/(1 + π/2):

sin(k(π/(2k))) sin π/2 1
1 + k(π/(2k)) − 0 = 1 + π/2 = 1 + π/2 > .

It follows that (fn ) does not uniformly converge on [0, ∞).

Problem 8.1-19. Show that the sequence (x2 e−nx ) converges uniformly on [0, ∞).

Solution: Let A = [0, ∞). From Exercise 8.1.9, (fn ) = (x2 e−nx ) converges pointwise to f (x) = 0. Note that fn0 (x) =
e−nx (2x − nx2 ). The roots of fn0 are at 0 and 2/n. A simple calculation shows that fn (2/n) = 4/(en)2 > fn (0) = 0. In
addition, fn0 (x) > 0 for 0 < x < 4/n and fn0 (x) < 0 for x > 4/n. By Theorem 6.2.8, fn is at an absolute maximum at
x = 4/n.
As a result, 0 < kfn − f kA = fn (2/n) = 4/(en)2 < 4/n. Because lim(4/n) = 0, it follows from the Squeeze Theorem
that lim kfn − f kA = 0. Therefore, (fn ) converges uniformly to f (x) = 0 on A.

Problem 8.1-21. Show that if (fn ), (gn ) converge uniformly on the set A to f , g, respectively, then (fn + gn ) converges
uniformly on A to f + g.

Solution: For  > 0, there are K 0 (/2), K 00 (/2) > 0 such that if n ≥ K 0 (/2), then kfn − f kA < /2, and if n ≥
K 00 (/2), then kgn − gkA < /2. Let K() = sup{K 0 (/2), K 00 (/2)}. By the Triangle Inequality and Theorem 2 below,
for n ≥ K():

Page 3
|k(fn + gn ) − (f + g)kA − 0| ≤ kfn − f kA + kgn − gkA < /2 + /2 = .
It follows that lim k(fn + gn ) − (f + g)kA = 0, so (fn + gn ) converges uniformly on A to f + g.
Theorem 2. Given any φ, ψ : A → R on A ⊆ R, kφ + ψkA ≤ kφkA + kψkA .
Proof. Let s = kφ + ψkA , in which case s = sup{|φ(x) + ψ(x)| : x ∈ A}. Now let t1 = sup{|ψ(x)| : x ∈ A} and
t2 = sup{|ψ(x)| : x ∈ A}. If x ∈ A, then t1 + t2 ≥ |φ(x)| + |ψ(x)| ≥ |φ(x) + ψ(x)|. It follows that t1 + t2 is an upper
bound of {|φ(x) + ψ(x)| : x ∈ A} and therefore is greater than or equal to s. Since t1 + t2 = kφkA + kψkA , we have
shown the desired inequality.

Problem 8.1-22. Show that if fn (x) := x + 1/n and f (x) := x, then (fn ) converges uniformly on R to f , but the sequence
(fn2 ) does not converge uniformly on R. (Thus the product of uniformly convergent sequences of functions may not coverge
uniformly.)

Solution: Note that in contrast to the functions in Exercise 8.1-23, fn is not bounded on R.
For (fn ), we have lim kx + 1/n − xkA = lim k1/nkA = lim 1/n = 0. Therefore, (fn ) converges uniformly to f on R.
Note that lim fn2 = lim(x2 + 2x/n + 1/n2 ) = x2 + lim(2x/n + 1/n2 ) = x2 . Accordingly, (fn2 ) converges pointwise to
f 2 on R.
We will now show that (fn2 ) does not uniformly converge on R. Let (xk ) be a sequence on R such that xk = k and
nk = k. We then have for all k ∈ N:

fn − f 2 = 2xk + 1 = 2 − 1 ≥ 1.
2
k k 2 2
k
2
For positive  where  ≤ 1, we have fn − f 2 ≥ . By Lemma 8.1.5, (fn2 ) does not uniformly converge on R.

Problem 8.1-23. Let (fn ), (gn ) be sequences of bounded functions on A that converge uniformly on A to f , g, respectively.
Show that (fn gn ) converges uniformly on A to f g.

Solution: In order to show the uniform convergence of (fn gn ) on A, we can rewrite the limit we seek under Lemma 8.1.8
as:

lim kfn gn − f gkA = lim k(fn gn − f gn ) − (f g − f gn )kA = lim kgn (fn − f ) + f (gn − g)kA .
Since fn is bounded on A, it must be that f is bounded on A by some M1 ≥ 0. We can prove this by contradiction.
If we assume this were not true, then for some x ∈ A, it would follow that |fn (x) − f (x)| is unbounded, resulting in the
contradiction that the supremum of {|fn (x) − f (x)| : x ∈ A} does not exist.
It must also be that (gn ) is bounded by some M2 ≥ 0 for all x ∈ A. This follows from the fact that lim gn must exist
because (gn ) must pointwise converge to g on A. Theorem 3.2.2 then requires that, (gn ) be bounded.
We may now establish boundaries on kfn gn − f gkA = sup{|fn gn − f g| : x ∈ A}. We have from the Triangle Inequality:

|fn gn − f g| = |gn (fn − f ) + f (gn − g)| ≤ M2 |fn − f | + M1 |gn − g| .


It follows that:
0 ≤ sup{|fn gn − f g| : x ∈ A} < M2 kfn − f kA + M1 kgn − gkA .
By hypothesis:

lim (M2 kfn − f kA + M1 kgn − gkA ) = M2 lim kfn − f kA + M1 lim kgn − gkA = M2 · 0 + M1 · 0 = 0.
By the Squeeze Theorem, lim kfn gn − f gkA = 0. Therefore, (fn gn ) converges uniformly to f g on A.

Problem 8.1-24. Let (fn ) be a sequence of functions that converges uniformly to f on A and that satisfies |fn (x)| ≤ M for
all n ∈ N and for all x ∈ A. If g is continuous on the interval [−M, M ], show that the sequence (g ◦ fn ) converges uniformly
to (g ◦ f ) on A.

Page 4
Solution: Let  > 0 be given. Since g is continuous on [−M, M ], there is a δ > 0 such that if y ∈ [−M, M ] and
|y − c| < δ, then |g(y) − g(c)| < /2. We will let fn and f take the place of y and c to establish our result. Because (fn )
uniformly converges to f on A, there is a K(δ) > 0 such that if n ≥ K(δ), then:

0 ≤ kfn − f kA = sup{|fn (x) − f (x)| : x ∈ A} < δ.


Accordingly, if n ≥ K(δ) and x ∈ A, then |fn (x) − f (x)| < δ. It follows from the continuity of g that:

|(g ◦ fn )(x) − (g ◦ f )(x)| = |g(fn (x)) − g(f (x)| < /2.


Because this is true for all x ∈ A, we have established /2 as an upper bound of kg ◦ fn − g ◦ f kA for n ≥ K(δ).
Consequently, |kg ◦ fn − g ◦ f kA − 0| ≤ /2 <  for n ≥ K(δ), from which it follows that lim kg ◦ fn − g ◦ f kA = 0. We
conclude that (g ◦ fn ) uniformly converges to g ◦ f on A.

Section 8.2
Problem 8.2-2. Prove that the sequence in Example 8.2.1(c) is an example of a sequence of continuous functions that converges
nonuniformly to a continuous limit.

Solution: If x = 0, then lim fn (x) = lim 0 = 0. If x ∈ [0, 2], then let  > 0 and K(, x) = 2/x. For n = K(, x), we have
|fn (x) − 0| = n2 (x − 2/n) = 0 < . If n > K(, x), then fn (x) = 0, from which follows that |fn (x) − 0| < . We then
infer that lim fn (x) = 0.
We will now prove that (fn ) is does not converge uniformly on [0, 2]. Let (xk ) be a sequence on [0, 2] where xk = 1/k
and let nk = k, in each case for all k ∈ N. For a given 0 <  < 1, we have |fnk (xk ) − 0| = k 2 (1/k) = k ≥ 1 > . By
Lemma 8.1.5, (fn ) does not converge uniformly on [0, 2].

Problem 8.2-4. Suppose (fn ) is a sequence of continuous functions on an interval I that converges uniformly on I to a function
f . If (xn ) ⊆ I converges to xo ∈ I, show that lim(fn (xn )) = lim f (x0 ).

Solution: By Theorem 8.2.2, f is continuous on I, so limx→x0 f (x) = f (x0 ). Applying the Sequential Criterion (Theorem
5.1.3), since (xn ) converges to x0 , it follows that lim f (xn ) = f (x0 ). For a given  > 0, there is a K 0 (/2) ∈ N such that
if n ≥ K 0 (/2), then:

|f (xn ) − f (x0 )| <
.
2
Because (fn ) converges uniformly on I, there is also a K 00 (/2) ∈ N such that if n ≥ K 00 (/2), then for x = xn for
any n ∈ N:

|fn (xn ) − f (xn )| < .
2
Let K() = sup{K 0 (/2), K 00 (/2)}, We then have for n ≥ K():

|fn (xn ) − f (x0 )| = |(fn (xn ) − f (xn )) + (f (xn ) − f (x0 )|


 
≤ |fn (xn ) − f (xn )| + |f (xn ) − f (x0 )| < + =
2 2
Therefore, lim fn (xn ) = f (x0 ).

Problem 8.2-5. Let f : R → R be uniformly continuous on R and let fn (x) := f (x + 1/n) for x ∈ R. Show that (fn )
converges uniformly on R to f .

Solution: Because f is uniformly continuous on R, for any given  > 0, there is a δ() > 0 such that for any x, y ∈ R, if
|x − u| < δ(), then |f (x) − f (u)| < . Let K() = 2/δ(). If n ≥ K(), then |(x + 1/n) − x| ≤ δ()/2 < δ(), in which
case |fn (x) − f (x)| = |f (x + 1/n) − f (x)| < . Therefore:

Page 5
kfn − f kR = sup{|fn (x) − f (x)| : x ∈ R} ≤ .
Since  is arbitrary, lim kfn − f kR = 0, and (fn ) uniformly converges to f on R.

Problem 8.2-7. Suppose the sequence (fn ) converges uniformly to f on the set A, and suppose that each fn is bounded on
A. (That is, for each n there is a constant Mn such that |fn (x)| ≤ Mn for all x ∈ A.) Show that the function f is bounded
on A.

Solution: For any  > 0, there is a K ∈ N such that if n ≥ K, then sup{|fn (x) − f (x)| : x ∈ A} < . Consequently,  is
an upper bound on this set, so |fn (x) − f (x)| <  for all x ∈ A. Since  is arbitrary, fn (x) = f (x) for n ≥ K. Because
fK is bounded on A, it follows that |f (x)| ≤ MK for all x inA. The function f is therefore bounded on A.

Problem 8.2-10. Let gn (x) := e−nx /n for x ≥ 0, n ∈ N. Examine the relationship between lim(gn ) and lim(gn0 ).

Solution: Observe that 0 < e−nx ≤ 1 for all x ∈ [0, ∞), so 0 < e−nx /n ≤ 1/n. Therefore, lim e−nx /n = 0 for x ∈ [0, ∞).
Because gn is bounded above by 1/n, it follows from the Squeeze Theorem that lim ke−nx /n − 0kR≥0 = 0. Therefore,
(gn ) converges uniformly to g(x) = 0 on [0, ∞).
We then have gn0 (x) = −e−nx . If x = 0, then lim[−e−nx ] = −1. If x ∈ (0, ∞), then because 0 < e−x < 1, it follows
from the Squeeze Theorem that lim[−e−nx ] = − lim(e−x )n ) = 0. Therefore, (gn0 ) converges to g 0 (0) = −1 and g 0 (x) = 0
for x ∈ (0, ∞). Note that g 0 is discontinuous at x = 0.
Now let  be given where 0 <  < 1/2. Suppose (xk ) is a sequence
on [0, ∞) where xk = − ln(2)/k (note that the
allowed range of  ensures xk > 0 for all k ∈ N) and nk = k. Then −e−kxk = eln 2 = 2 >  for all k ∈ N. By Lemma
8.1.5, (gn0 ) does not uniformly converge on [0, ∞).

Problem 8.2-11. Let I := [a, b] and let (fn ) be a sequence of functions on I → R that converges on I to f . Suppose that each
Rb
derivative fn0 is continuous on I and that the sequence (fn0 ) is uniformly convergent to g on I. Prove that f (x)−f (a) = a g(t)dt
and that f 0 (x) = g(x) for all x ∈ I.

Solution: Because (fn ) converges to f on the bounded interval I and (fn0 ) exists for n ∈ N and converges uniformly to g,
it follows from Theorem 8.2.3 that (fn ) converges uniformly to some function. This function must be f because the limit
of (fn ) is unique. It further follows from Theorem 8.2.3 that f 0 (x) = g(x) for all x ∈ I.
Now let x ∈ [a, b]. Given that (fn0 ) converges uniformly on I, it must converge uniformly on [a, x] (since this result has
not yet been proven, see Theorem 3 below). Because each fn0 is continuous on I, by the Lebesque Cirterion fn ∈ R[a, x].
Applying Theorem 8.2.4 and the fact that (fn0 ) converges to g by hypothesis, we have:
Z x Z x Z x
0
g = lim fn = f 0,
a a a

and g ∈ R[a, x].


Since f 0 exists on all of I, f 0 is continuous on I. Applying the Fundamental Theorem of Calculus, we get:
Z x Z x
g= f 0 = f (x) − f (a).
a a

Theorem 3. Suppose (fn ) converges uniformly to f on [a, b]. If γ ∈ [a, b], then (fn ) also converges uniformly to f on
[a, γ].

Proof. By hypothesis, lim kfn − f k[a,b] = 0. Observe that 0 ≤ kfn − f k[a,γ] ≤ kfn − f k[a,b] for all n ∈ N. By the Squeeze
Theorem, lim kfn − f k[a,γ] = 0. The sequence (fn ) therefore uniformly converges to f on [a, γ].

Problem 8.2-15. Let gn (x) := nx(1 − x)n for x ∈ [0, 1], n ∈ N. Discuss the convergence of (gn ) and (i nt10 gn dx.

Solution: Observe that gn (0) = gn (1) = 0 for all n ∈ N. Now let x ∈ (0, 1). There is a y > 0 such that 1 − x = 1/(1 + y).
By the Binomial Theorem:

Page 6
       
n n n 2 n n 1
(1 + y)n = 1+ y+ y + ··· + y > n(n − 1)y 2 .
0 1 2 n 2
It follows that for n ≥ 2:
nx nx 2x
0 < nx(1 − x)n = < = 2 .
(1 + y)n (1/2)n(n − 1)y 2 y (n − 1)

By the Squeeze Theorem, the 2-tail of (gn ) converges to zero. By Theorem 3.1.9, lim(gn − 0) = 0, so (gn ) converges
to g(x) = 0 on x ∈ [0, 1].
We see that gn0 (x) = n(1 − x)n−1 [1 − (n + 1)x]. Setting gn (x0 ) = 0, we see that gn is at an absolute maximum at
x0 = 1/(n + 1). We then have gn (x0 ) = (n/(n + 1))n+1 < 1. Therefore, 0 ≤ gn (x) < 1 for all x ∈ [0, 1], from which it
follows that kgn k[0,1] < 1 for all n ∈ N. Since gn is continuous on [0, 1], each gn ∈ R[0, 1]; further, g ∈ R[0, 1]. Applying
R1 R1
Theorem 8.2.5, we infer that 0 g = 0 = lim 0 gn .
Note, however, that (gn ) does not uniformly converge on [0, 1] (hence the power of Theorem 8.2.5). Suppose 0 <  <
1/e. Let (xk ) be a sequence on [0, 1] where xk = 1/k and nk = k for all k ∈ N. We then have:
 k
1 1
|gnk (xk )| = nk xk (1 − xk )nk = 1− = > ,
k e
where we have used the result from Exercise 3.3.12(d). By Lemma 8.1.5, (gn ) does not uniformly converge on [0, 1].

Problem 8.2-17. Let fn (x) := 1 for x ∈ (0, 1/n) and fn (x) := 0 elsewhere on [0, 1]. Show that (fn ) is a decreasing sequence
of discontinuous functions that converge to a continuous limit function, but the convergence is not uniform on [0, 1]

Solution: Clearly fn is discontinuous at x = 0 and x = 1/n for all n ∈ N.


Let x ∈ [0, 1] and n ∈ N. Note that 1/(n + 1) < 1/n. If x ∈ (0, 1/(n + 1)), then fn (x) = fn+1 (x) = 1. If
x ∈ [0, 1]\(0, 1/n), then fn (x) = fn+1 (x) = 0. If x ∈ [1/(n + 1), 1/n), then fn (x) = 1 > fn+1 (x) = 0. Therefore, (fn )
is a decreasing sequence of discontinuous functions.
Let x ∈ [0, 1] and  > 0 be given. Suppose K() = 2/x. If n ≥ K(), then 1/n = x/2 < x, so fn (x) = 0. Therefore,
|fn (x) − 0| = 0 <  for all x ∈ [0, 1]. It follows that (fn ) converges pointwise to f (x) = 0 on [0, 1].
The sequence does not, however, converge uniformly on this interval. Suppose (xk ) is a sequence on [0, 1] where
xk = 1/(2k) and nk = k for k ∈ N. Let  = 1/2. Then |fnk (xk ) − 0| = fk (1/(2k) = 1 >  because 0 < 1/(2k) < 1/k for
all k ∈ N. By Lemma 8.1.5, (fn ) does not converge uniformly on [0, 1].

Problem 8.2-18. Let fn (x) := xn for x ∈ [0, 1], n ∈ N. Show that (fn ) is a decreasing sequence of continuous functions that
converges to a function that is not continuous, but the convergence is not uniform on [0, 1].

Solution: Clearly all fn are continuous on [0, 1]. If x ∈ [0, 1], then fn+1 (x) = xn+1 = x · xn = xfn (x) ≤ fn (x).
Accordingly, (fn ) is a decreasing sequence of continuous functions.
If x ∈ [0, 1), then lim fn (x) = lim xn = 0 by Example 3.1.11(b). If x = 1, then lim fn (x) = lim 1n = 1. Therefore,
(fn ) converges on [0, 1] to:

0 for x ∈ [0, 1)
f (x) =
1 for x = 1.
Obviously f is discontinuous at x = 1.
We will show that kfn − f k[0,1] = 1 in all cases. Let n ∈ N be arbitrary. Now observe that fn is bounded above by 1 on
[0, 1]. Clearly |fn (1) − f (1)| = 0, so for the uniform norm to be greater than zero, we must look on [0, 1). Let  ∈ (0, 1).
If we solve 1 −  < xn , then x > (1 − )1/n . Since (1 − )1/n < 1, we can choose x0 ∈ ((1 − )1/n , 1). We then have
1 −  < fn (x0 ). It follows that 1 = sup{|fn (x) − f (x)| : x ∈ [0, 1]}. Therefore, lim kfn − f k[0,1] = 1. We conclude that
(fn ) does not uniformly converge on [0, 1].

Problem 8.2-19. Let fn (x) := x/n for x ∈ [0, ∞), n ∈ N. Show that (fn ) is a decreasing sequence of continuous functions
that converges to a continuous limit function, but the convergence is not uniform on [0, ∞).

Page 7
Solution: Each fn is obviously continuous on the interval of interest. Let n ∈ N and x ∈ [0, ∞). Because fn (x) = x/n ≥
x/(n + 1) = fn+1 (x), the sequence (fn ) is a decreasing sequence of continuous functions.
Now let  > 0 be given. Let K() = 2x/. If n ≥ K(), then |fn (x) − 0| = x/n ≤ /2 < . Therefore, lim fn (x) = 0
for all x ∈ [0, ∞). The sequence (fn ) thus converges pointwise to f (x) = 0 on [0, ∞).
Each fn is obviously unbounded on [0, ∞); hence, for all n ∈ N, the uniform norm kfn − f k[0,∞) = ∞ and is therefore
divergent. Consequently, (fn ) does not uniformly converge on [0, ∞).

Problem 8.2-20. Give an example of a decreasing sequence (fn ) of continuous functions on [0, 1) that converges to a continuous
limit function, but the convergence is not uniform on [0, 1)

Solution: The sequence where:


1
fn (x) = ,
n(1 − x)
for all n ∈ N fits the bill. Notice that fn extends to infinity as x → 1.
Clearly, each fn is continuous on [0, 1). In addition, for any n ∈ N and x ∈ [0, 1), we have fn (x) = 1/[n(1 −
x)] > 1/[(n + 1)(1 − x)] = fn+1 (x). Therefore, (fn ) is a decreasing sequence of continuous functions. Moreover,
lim 1/[n(1 − x)] = 1/(1 − x) lim 1/n = 0. Accordingly, (fn ) converges pointwise to f (x) = 0 on [0, 1), which is continuous
on that interval.
But the sequence does not uniformly converge on [0, 1). As with the sequence in exercise 8.2-19, each fn is unbounded.
Assume for any n ∈ N there is an Mn > 0 such that fn (x) ≤ Mn . Let x0 = 1 − 1/(2Mn n) ∈ [0, 1). It follows that
fn (x0 ) = 2Mn > Mn . Therefore, fn is unbounded on [0, 1). As a result, kfn − f k[0,1) = ∞, from which it follows that the
limit of the uniform norm is divergent. The sequence therefore does not uniformly converge on [0, 1). This result does not
conflict with Dini’s Theorem, however, because the interval of interest here is half-open, whereas Dini’s Theorem requires
a closed, bounded interval.

Section 8.3
Problem 8.3-5. If x ≥ 0 and n ∈ N, show that:

1 (−x)n
= 1 − x + x2 − x3 + · · · + (−x)n−1 + .
x+1 1+x
Use this to show that:
x
x2 x3 xn (−t)n
Z
ln(x + 1) = x − + − · · · + (−1)n + dt
2 3 n 0 1+t
and that:
2 3 n n+1
 
ln(x + 1) − x − x + x − · · · + (−1)n x ≤ x

.
2 3 n n+1

Solution: The first part can be easily shown by multiplying the right-hand side of the expression by (1 + x)/(1 + x).
Now let f (x) = ln(1 + x), so f 0 (x) = 1/(1 + x) on (−1, ∞). Clearly, f is continuous on [0, ∞). If x ≥ 0, then f 0 is
continuous on [0, x], from which it follows that f 0 ∈ R[0, x]. By the Fundamental Theorem of Calculus:
Z x
f (x) − f (0) = ln(1 + x) = f 0 (x)dx
0
x x
(−t)n
Z Z
1 − x + x2 − x3 + · · · + (−x)n−1 dx +
 
= dt.
0 0 1+t
n−1 x
xk+1 (−t)n
X Z
= (−1)k + dt,
k+1 0 1+t
k=0

which is what we set out to prove for the second part.

Page 8
For the third part, rearrange the previous expression and take the absolute value to get:
Z
n−1 k+1 x Z x n
X
k x (−t)n |t|
ln(1 + x) − (−1) = dt = dt.

k + 1 0 1+t 0 1+t
k=0

n n
Observe that |t| /(1 + t) ≤ |t| for t ≥ 0. Applying Theorem 7.1.5(c), we have:
Z
n−1 k+1 x
X x n xn+1
ln(1 + x) − (−1)k ≤ |t| dt = .

k + 1 0 n+1
k=0

Problem 8.3-8. Let f : R → R such that f 0 (x) = f (x) for all x ∈ R. Show that there exists K ∈ R such that f (x) = Kex
for all x ∈ R.

Solution: If f (0) 6= 0, let g : R → R such that g(x) = f (x)/f (0). It follows that g 0 (x) = f 0 (x)/f (0) = g(x) and g(0) = 1.
By Theorem 8.3.4, g(x) = ex because the function with both these properties is unique. Therefore f (x) = f (0)g(x) = Kex
where K = f (0).
Now suppose f (0) = 0. It can be easily shown by induction that f (n) = f for all n ∈ N. Therefore, f (n) (0) = 0 for all
n ∈ N. Let x ∈ R be given. Applying Taylor’s Theorem at x0 = 0, we have for any n:
n
X f (j) (0) f (n+1) (c) n+1 f (n+1) (c) n+1
f (x) = f (0) + xj + x = x ,
j=1
j! (n + 1)! (n + 1)!

for some c between 0 and x0 (non-inclusive).


n+1
Now let L > 0 be such that |f (x)| ≤ L for all x ∈ [0, x]. It follows that |f (x)| ≤ K |x| /(n + 1)!. Since
n+1
lim |x| /(n + 1)! = 0 by Exercise 3.2-19(c), we infer that lim |f (x)| = 0. Since x is arbitrary, we conclude that f (x) = 0
for all x ∈ R. Therefore, f (x) = 0 · ex = 0.

Page 9

You might also like