Math 313: Lecture 17

M. K. Vemuri
Last homework

https://chatgpt.com/c/67f9690e-fa38-8012-8555-2ce2b1c821f9

It is expected that you have already read articles 3.2, 3.3 and 3.4
on pp48-51 in Rudin.
Subsequences

Definition
Given a sequence {pn }, consider a sequence {nk } of positive
integers, such that n1 < n2 < n3 < · · · . Then the sequence {pnk }
is called a subsequence of {pn }. If {pnk } converges, its limit is
called a subsequential limit of {pn }.

Proposition
Let X be a metric space and p ∈ X . If a sequence {pn } converges
to p then every subsequence of {pn } converges to p.

Proof.
Assume pn → p and suppose n1 < n2 < n3 < · · · is an increasing
sequence of positive integers. Observe that nk ≥ k (proof by
induction). Let ε > 0 be given. Since pn → p, there exists N such
that n ≥ N =⇒ d(pn , p) < ε. Therefore
k ≥ N =⇒ d(pnk , p) < ε. It follows that pnk → p.
Subsequences (contd)
n+1
Example. In the metric space R, let pn = (−1)n


n . Thus

3 4 5 6 7
{pn }∞
n=1 = {−2, , − , , − , , . . . }.
2 3 4 5 6
Let nk = 2k. Thus {nk }∞
k=1 = {2, 4, 6, 8, . . . } and

3 5 7 9
{pnk }∞
k=1 = { , , , , . . . }.
2 4 6 8
The sequence pnk = p2k = 2k+1 1
2k = 1 + 2k converges to 1, so 1 is a
subsequential limit of the sequence {pn }.
Let mk = 2k − 1. Thus {mk }∞ k=1 = {1, 3, 5, 7, . . . } and
2k 1
pmk = p2k−1 = − 2k−1 = −1 − 2k−1 converges to −1, so −1 is a
subsequential limit of the sequence {pn }.
We can conclude that the sequence {pn } does not converge by the
previous proposition.
Subsequences (contd)
Theorem (Bolzano-Weierstrass)
(a). If {pn } is a sequence in a compact metric space X , then some
subsequence of {pn } converges to a point of X .
(b). Every bounded sequence in Rk contains a convergent
subsequence.

Proof. (a) Let E be the range of {pn }. If E is finite, then there is

a p ∈ E and a sequence of positive integers n1 < n2 < · · · , such
that pnk = p for all k (by the pigeon-hole principle). The
subsequece {pnk } evidently converges to p.
If E is infinite, then E has a limit point p ∈ X . Therefore there
exists n1 such that pn1 ∈ E ∩ N1 (p). Since E ∩ N1/2 is infinite,
there exists n2 > n1 such that pn2 ∈ E ∩ N1/2 . Since E ∩ N1/3 is
infinite, there exists n3 > n2 such that pn3 ∈ E ∩ N1/3 . Continuing
in this way, we obtain a subsequece {pnk } such that
d(pnk , p) < 1/n. By the Archimedian property, pnk → p.
Subsequences (contd)

(b) This follows from (a) because every bounded sequence in Rk is

contained in a compact set.
Theorem
The subsequential limits of a sequence in a metric space form a
closed set.
See book for the proof.
Cauchy sequences

Definition
A sequence {pn } in a metric space X is said to be a Cauchy
sequence if for all ε > 0 there exists an integer N such that
m, n ≥ N implies d(pn , pm ) < ε.

Theorem
If a sequence in a metric space X converges, then it is a Cauchy
sequence.

Proof.
Let {pn } be a sequence in X and suppose pn → p ∈ X . Let ε > 0
be given. Then there exists a positive integer N such that
n ≥ N =⇒ d(pn , p) < ε/2. Therefore m, n ≥ N implies

d(pn , pm ) ≤ d(pn , p) + d(p, pm ) < ε/2 + ε/2 = ε.

Cauchy sequences (contd)
Example
Let X be the metric space (0, ∞) with the usual metric. Let
pn = 1/n. Then {pn } is Cauchy but does not converge.
Here are the details:
(Cauchy) Let ε > 0 be given. By the Archimedian Property
there exists a positive integer N > 2/ε. If m, n > N then
|1/n − 1/m| ≤ 1/m + 1/n < ε/2 + ε/2 = ε.

(Does not converge) Let p ∈ X . Let ε = p/2. Then ε > 0. By the

Archimedian Property, there exists a positive integer D > 1/ε. If
n ≥ D then
|pn − p| = p − pn > p − ε = ε.

https://chatgpt.com/c/67f9690e-fa38-8012-8555-2ce2b1c821f9
Proposition
Every subsequence of a Cauchy sequence is a Cauchy sequence.
The proof is left to the reader.
Cauchy sequences (contd)
Theorem
Let {pn } be a Cauchy sequence in a metric space X . If a
subsequence {pnk } converges to a point p ∈ X , then the entire
sequence {pn } converges to p.
Proof. Let ε > 0 be given. Since {pn } is Cauchy, there exists N
such that
m, n ≥ N =⇒ d(pn , pm ) < ε/2.
Since pnk → p, there exists K1 such that
k ≥ K1 =⇒ d(pnk , p) < ε/2.

Since {nk } is an increasing sequence of positive integers, there

exists K2 such that k ≥ K2 =⇒ nk ≥ N (proof?). Let
K = max{K1 , K2 }. Then n ≥ N implies
d(pn , p) ≤ d(pn , pnK ) + d(pnK , p) < ε/2 + ε/2 = ε.
Cauchy sequences (contd)

Theorem
(a). In a compact metric space, every Cauchy sequence converges.
(b). In Rk , every Cauchy sequence converges.

Proof.
By the Bolzano-Weierstrass theorem, the given sequence has a
convergent subsequence. Since the sequence is Cauchy, the
previous theorem implies that the entire sequence converges.

Definition
A metric space in which every Cauchy sequence converges is said
to be complete.
The previous theorem says that all compact metric spaces and all
Euclidean spaces are complete. We have seen that (0, ∞) is not
complete.
Cauchy sequences (contd)
Proposition
Let (X , d) be a complete metric space, and Y ⊆ X . The metric
space (Y , d) is complete iff Y is a closed subset of X .

Proof.
Assume Y is a closed subset of X . Suppose {pn } is a Cauchy
sequence in Y . Then {pn } is a Cauchy sequence in X . Since X is
complete, there exists p ∈ X such that pn → p. Since Y is closed,
p ∈ Y.
Assume Y is not closed. Then there exists p ∈ Y 0 \ Y . Therefore
there is a sequence {pn } in Y such that pn → p in the metric
space (X , d). Since {pn } converges in the metric space (X , d), it
is Cauchy. Since the limit of a sequence is unique, there is no point
p 0 ∈ Y such that pn → p 0 . Therefore the sequence {pn } does not
converge in (Y , d).
It follows immediately from this theorem that Q is not complete
with the usual metric.
Some homework

Read articles 3.13 and 3.14 on pp55 in Rudin.