18.

100B Problem Set 9 Solutions

Sawyer Tabony
1) First we need to show that
1
fn (x) =
nx + 1
converges pointwise but not uniformly on (0, 1). If we fix some x ∈ (0, 1), we have that
1
lim fn (x) = lim = 0.
n→∞ n→∞ nx + 1

Thus the fn converge pointwise. However, consider ε = 14 , and let N ∈ N. Then

1 1 1
fN ( )= 1 = �< ε.
N N ( N + 1) 2
So the fn do not converge to zero uniformly. Now we consider
x
gn = .
nx + 1
Given ε > 0, let N ∈ N be larger than 1ε . Then for any x ∈ (0, 1), M ≥ N , we have
x 1 1 1
gM (x) = = 1 < ≤ < ε.
Mx + 1 M+ x
M N
Thus, gn converges to 0 uniformly on (0,1).

2) The functions fn are defined on R by

x
fn = .
1 + nx2
1
First we show fn −→ 0. For any ε > 0, choose N > . Then for n > N , if |x| ≤ ε,
ε2
|x| < ε(1 + nx2 ) so |fn (x)| < ε. If |x| > ε,
� � � �
2 2
� x � � x �
|nεx| > nε > 1 =⇒ |nεx | > |x| =⇒ ε > � 2 � > �
� � � � = |fn (x)|.
nx 1 + nx2 �
Thus fn converges uniformly to 0 = f .
Differentiating, we find
(1 + nx2 ) · 1 − x · (2nx) 1 − nx2
fn� (x) = = .
(1 + nx2 )2 (1 + nx2 )2
Thus,
� 1 − nx2 � 2 2
� �
|fn� (x)| = �� � = |1 − nx | ≤ 1 + nx = 1
.
2
(1 + nx )2 � 2
(1 + nx ) 2 2
(1 + nx ) 2 1 + nx2
� 0, fn� (x) −→ 0, but fn (0) = 1, ∀n ∈ N. So if g(x) = lim fn (x), g(0) = 1 =
So for all x = � 0=
� �
f (0). But f (x) = 0, and thus exists, for all x.
� 0, f � (x) = 0 = g(x). We showed above that fn −→ f uniformly on all of R. Finally,
For all x =
fn� −→ g uniformly away from zero, that is, on any interval that does not have zero as a limit
point. This is true because the denominator of fn� can be made arbitrarily large compared to the
numerator, for |x| > ε.
1
3) So we have fn −→ f uniformly for fn bounded. Then for ε = 1, ∃N such that ∀n > N , x ∈ E,
|fn (x)−f (x)| < 1. Thus if fn is bounded by Bn , ∀x ∈ E, |f (x)| < 1+BN which implies that ∀n >
N , x ∈ E, fn (x) < BN + 2. Thus (fn ) is uniformly bounded by max{B1 , B2 , ..., BN −1 , BN + 2}.
If the fn are converging pointwise, f need not be bounded. For example, on (0, 1), if
1
fn (x) = ,
x + n1
1 1
then fn −→ pointwise, and each fn is bounded by n, but of course is unbounded on (0, 1).
x x

4) We have fn −→ f and gn −→ g uniformly.

ε
a) Given ε > 0, ∃M, N ∈ N such that for any m > M , n > N , |fm − f | < 2 and |gn − g| < 2ε . So
for n, m > max{M, N }
ε ε
|(fn + gn ) − (f + g)| = |(fn − f ) + (gn − g)| ≤ |fn − f | + |gn − g| < + = ε.
2 2
So (fn + gn ) −→ (f + g) uniformly.
b) If each fn and gn is bounded on E, by the previous problem they are uniformly bounded, say
by A and B. Say without loss of generality A ≥ B. Then for ε > 0, choose N such that if
n > N , |fn − f | < 2εA and |gn − g| < 2εA . Then
ε ε
|fn gn − f g| = |fn gn − f gn + f gn − f g| ≤ |fn − f ||gn | + |f ||gn − g| < A+ A = ε.
2A 2A
So fn gn converges to f g uniformly.

5) Define �
0 if x ∈/Q
f (x) = x, g (x) =
q if, in lowest terms, x = p/q
so that � �
1 1
fn (x) = f (x) 1 + , and gn (x) = g (x) + .
n n
On any interval [a, b], with M = max (|a|, |b|) we have
|x| M 1
|fn (x) − f (x)| = ≤ , |gn (x) − g (x)| =
n n n
so that fn → f and gn → g uniformly.
On the other hand, with m = min (|a|, |b|) we have
|fn (x) gn (x) − f (x) g (x)| = |(fn (x) − f (x)) gn (x) + f (x) (gn (x) − g (x))|
� �
f (x) n+1 m
= g (x) + ≥ g (x) .
n n n
k
Now notice that if L ∈ N is larger than b − a then there is an integer k such that L ∈ [a, b],
n
choosing L ∈ N larger than m and larger than b − a we get
� �
m k m
�fn gn − f g� ≥ g = L>1
n L n
and hence fn gn does not converge to f g uniformly.
6) We want to show g ◦ fn −→ g ◦ f uniformly, for g continuous on [−M, M ]. Since g is continuous
on a compact set, it is uniformly continuous. So given ε > 0, ∃δ > 0 such that if |x − y| < δ,
|g(x) − g(y)| < ε. Now since fn −→ f uniformly, ∃N ∈ N such that for any n > N , x ∈ E,
|fn (x) − f (x)| < δ. Thus |g(fn (x)) − g(f (x))| < ε, so |g ◦ fn (x) − g ◦ f (x)| < ε. Thus we have
shown (g ◦ fn ) −→ g ◦ f uniformly on E.

7) a) First we claim that, for every x ∈ [0, 1],

√
0 ≤ Pn (x) ≤ Pn+1 (x) ≤ x.
This is clearly true for n = 0, so assume inductively that it is true for n = k and notice that
√ √
� �
1�
x − Pk (x)2
�
x − Pk+1 (x) = x − Pk (x) +
2
√ 1 �√ � �√ �
= x − Pk (x) − x − Pk (x) x + Pk (x)
�2 �
�√ � 1 �√ �
= x − Pk (x) 1 − x + Pk (x) ≥ 0.
2
√
It follows that Pk+1 (x) ≤ x, and then
1� �
Pk+1 (x) − Pk (x) = x − Pk (x)2 ≥ 0
2
shows that Pk (x) ≤ Pk+1 (x), and proves the claim.
Notice that√for every fixed x, the sequence (Pn (x)) is monotone increasing and bounded
above (by x), it follows that this sequence converges, to say f (x). This function f (x) is
non-negative and satisfies
� �
1
f (x) = lim Pn+1 (x) = lim Pn (x) + (x − Pn (x) ) = f (x) + (x − f (x)2 )
2
n−→∞ n−→∞ 2
√ √
which implies f (x) = x, and hence the polynomials converge pointwise to x on [0, 1].
Since they are continuous and converge monotonically to a continuous function on [0, 1], a
compact set, they converge uniformly (by Dini’s theorem).
b) Here we use a something similar to problem 6 along with the above work to show that
Pn (x2 ) −→ |x| on [−1, 1]. The difference between this and problem 6 is we need (fn√◦ g) −→
(f ◦ g), but this is easier. Given ε > 0, ∃N
√ ∈ N with ∀x ∈ [0, 1], √ n > N , |Pn (x) − x| < ε.
So for all x ∈ [−1, 1], n > N , |Pn (x2 ) − x2 | < ε. Since |x| = x2 , we have show that the
polynomials Pn (x2 ) converge uniformly to |x| on [−1, 1].
MIT OpenCourseWare
http://ocw.mit.edu

18.100B Analysis I
Fall 2010

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.