Assignment-10

(Due 11/21)

1. Consider the sequence of functions fn (x) = xn on [0, 1].

(a) Show that each function fn is uniformly continuous on [0, 1].

Solution: Any continuous function on a compact set is uniformly continuous.

(b) For a sequence of functions fn on [0, 1] write what it means for them to NOT be an
equicontinuous sequence.

Solution: There exists ε > 0 such that for each k, there exists points xk , yk and
functions fnk such that

1
|xk − yk | < but |fnk (xk ) − fnk (yk )| > ε.
k

(c) Using only part (b) (and not Ascoli-Arzela theorem) show that the sequence of functions
fn is not equicontinuous.

Solution: Let xn = 1 − 1/n and yn = 1, then

 1 n n→∞
|fn (yn ) − fn (xn )| = 1 − 1 − −−−→ 1 − e−1 > 0.
n
So if we take ε > 0 such that ε < 1 − e−1 (for instance we can take ε = (1 − e−1 )/2),
then there exists N such that for all n > N ,

|fn (yn ) − fn (xn )| > ε,

and so by part(b), {fn } is not an equicontinuous family.

(d) Now prove the same statement using Ascoli-Arzela.

Solution: The family {fn } is clearly pointwise bounded by 1. So if the family was
equicontinuous, then by Ascoli-Arzela, there will exist a uniformly convergent subse-
quence. BUt the pointwise limit of the sequence (and hence also the sub-sequence)
is (
0, x 6= 1
f (x) =
1, x = 1,

1
 which is not continuous. This is a contradiction since uniform limits of continuous
functions have to be continuous.

2. Suppose {fn } is an equicontinuous sequence of functions on a compact set K ⊂ R. If {fn }

converges pointwise on K, show that it converges uniformly on K.

Solution: We show that the sequence {fn } is uniformly Cauchy. So let ε > 0. By equicon-
tinuity, there exists a δ > 0 such that for all n,
ε
|x − y| < δ =⇒ |fn (x) − fn (y)| < .
3
Since K is compact, there exist finitely many points {p1 , · · · , pm } such that

K ⊂ ∪m
k=1 Bδ (pk ).

By the Cauhy criteria for pointwise convergence, for each k = 1, 2, · · · , m, there exists Nk
such that whenever n, m > Nk ,
ε
|fn (pk ) − fm (pk )| < .
3
Let N = max(N1 , · · · , Nm ). Now for any x ∈ K, there exists pk such that |x − pk | < δ.
Then for n, m > N ,
ε
|fn (x) − fm (x)| ≤ |fn (x) − fn (pk )| + |fn (pk ) − fm (pk )| + |fm (pk ) − fm (x)| < 3 · = ε.
3
This completes the proof of the claim, and hence it follows that the sequence {fn } converges
uniformly.

3. Recall that C 0 [0, 1] denotes the set of continuous functions on [0, 1]. We endow it with the
usual metric
d(f, g) = sup |f (t) − g(t)|.
t∈[0,1]

Define a function T : C 0 [0, 1] → C 0 [0, 1] by

Z x
T [f ](x) = f (t) dt.
0

Let K ⊂ C 0 [0, 1] be a bounded set.

(a) Show that T is a continuous function. Is it injective?

Solution: For f, g ∈ C 0 [0, 1], and any t ∈ [0, 1], by definition,

|f (t) − g(t)| ≤ d(f, g).

2
 So for any x ∈ [0, 1],
Z x
|T [f ](x) − T [g](x)| = [f (t) − g(t)] dt
0
Z x
≤ |f (t) − g(t)| dt
0
Z x
≤ d(f, g) dt
0
≤ d(f, g).

Taking supremum over all x ∈ [0, 1], we see that

d(T [f ], T [g]) ≤ d(f, g).

So given an ε > 0, let δ = ε. Then

d(f, g) < δ =⇒ d(T [f ], T [g]) < ε.

This shows that T is a continuous map. We claim that the function is injective. It is
enough to show that if T [f ](x) = 0 for all x ∈ [0, 1], then f (t) = 0 for all t ∈ [0, 1]. To
see this, note that if T [f ](x) = 0 for all x, then for any x, y ∈ [0, 1], x < y,
Z y
f (t) dt = 0.
x

Now, suppose f (t0 ) > 0 at some point t0 ∈ [0, 1]. Since f is continuous, there exists
δ > 0 such that for any t ∈ (t0 − δ, t0 + δ), f (t) > 0. But then
Z t0 +δ
f (t) dt > 0,
t0 −δ

which is a contradiction. So there is not point where f is strictly positive. Similarly

there is no point where f is strictly negative. Hence f has to be identically zero.

(b) Show that the set T (K) is a compact subset of C 0 [0, 1]. Hint. Use the Version-2 of
Ascoli-Arzela.

Solution: Since K is a bounded set, there exists an M such that d(f, 0) < M for all
f ∈ K. Here 0 denotes the zero function, that is the function that vanished on all of
[0, 1]. By the definition of d, this means that

|f (t)| < M, ∀t ∈ [0, 1], ∀f ∈ K.

• Closed. Trivially since it is the closure of a set.

3
• Bounded. For any f ∈ K, since |f (t)| < M for all t ∈ [0, 1], we see that
Z x
|T [f ](x)| = f (t) dt
Z x0
≤ |f (t)| dt
0
< M,

and so
d(T [f ], 0) < M.
This shows that T (K) is bounded, and hence T (K) is also bounded.

• Equicontinuous. Let ε > 0, and δ = δ(ε) to be picked later. For any f ∈

C 0 [0, 1], by the fundamental theorem of calculus, T [f ] is differentiable on [0, 1],
and moreover,
T [f ]0 (x) = f (x).
So if f ∈ K, we have that
|T [f ]0 (x)| < M
for all x ∈ [0, 1]. Then by the mean value theorem, for any x, y ∈ [0, 1],

|T [f ](x) − T [f ](y)| < M |x − y|.

Another way to see this is to directly estimate the integral. That is if x, y ∈ [0, 1]
with say x > y, then
Z x
|T [f ](x) − T [f ](y)| = f (t) dt
y
Z x
≤ |f (t)| dt
y
< M |x − y|.

In any case, if we take δ = ε/M , then

|x − y| < δ =⇒ |T [f ](x) − T [f ](y)| < ε.

This shows that T (K) is equicontinuous. To see that the closure is also equicon-
tinuous, we use the ε/3 trick. So let ε > 0, and δ = ε/3M . Then by the abvove
argument for any f ∈ K,
ε
|x − y| < δ =⇒ |T [f ](x) − T [f ](y)| < .
3

Now if g ∈ T (K), there exists an f ∈ K such that d(T [f ], g) < ε/3. That is for
any x ∈ [0, 1],
ε
|T [f ](x) − g(x)| < .
3

4
 Then if |x − y| < δ,

|g(x) − g(y)| ≤ |g(x) − T [f ](x)| + |T [f (x) − T [f ](y)| + |T [f ](y) − g(y)|

ε ε ε
≤ + + = ε.
3 3 3
So given an ε > 0, there exists a δ = δ(ε) (in this case δ = ε/3M works) such
that for any g ∈ T (K),

|x − y| < δ =⇒ |g(x) − g(y)| < ε,

and hence T (K) is equicontinuous.

Then by the version-2 of Arzela-Ascoli, the set T (K) is compact.

4. Suppose f : [0, 1] → R is a continuous function such that

Z 1
f (t)tn dt = 0
0

for all n = 0, 1, 2, · · · .R Show that f (t) = 0 for all t ∈ [0, 1]. Hint. Use Stone-Weierstrass
1
theorem to show that 0 f 2 (t) dt = 0.

Solution: Note that by the hypothesis, if P is any polynomial,

Z 1
f (t)P (t) dt = 0.
0

By Stone-Wieirstrass theorem, there exists a sequence of polynomials Pn → f uniformly on

[0,1]. Since f is bounded on [0, 1], this also implies that Pn (t)f (t) → f (t)2 uniformly. (Note
that this would not have been true if f were not bounded). By the theorem on uniform
convergence and integration,
Z 1 Z 1
2
f (t) dt = lim Pn (t)f (t) dt = 0.
0 n→∞ 0

We then claim that f (t) = 0 for all t ∈ [0, 1]. If not, then there exists a point p ∈ [0, 1]
such that f (p) 6= 0, and so f 2 (p) > 0. By continuity, there is a δ > 0 such that f 2 (t) > 0
for all t ∈ (p − δ, p + δ). Since f 2 (t) is at least non-negative at other points, by comparison
principle for integrals,
Z p+δ Z 1
2
0< f (t) dt ≤ f 2 (t) dt = 0,
p−δ 0

a contradiction.