Skip to content
Tutorials Center
My learning

Topics

Lesson 13 of 37 ยท Sequences and Limits

Subsequences and the Bolzano-Weierstrass Theorem

Learn subsequences, limit points, limsup and liminf, and prove the Bolzano-Weierstrass theorem: every bounded sequence has a convergent subsequence.

A sequence may bounce around and never converge, yet part of it may settle down. For example (-1)โฟ diverges, but its even-numbered terms are all 1 and its odd-numbered terms are all -1. Choosing such infinite portions is formalized by subsequences. The Bolzano-Weierstrass theorem guarantees that any bounded sequence has at least one convergent subsequence, a result with huge consequences for continuity and optimization.

Subsequences

Given a sequence (a_n) and indices n_1 < n_2 < n_3 < ..., the sequence (a_{n_k}) is a subsequence. You keep terms in their original order but skip some (infinitely many are kept).

Examples for a_n = (-1)โฟ/n:

  • even terms: a_{2k} = 1/(2k)
  • odd terms: a_{2kโˆ’1} = โˆ’1/(2kโˆ’1)
  • terms at squares: a_{kยฒ}

Fact. If a_n โ†’ L, then every subsequence also converges to L. (Since n_k โ‰ฅ k, once k โ‰ฅ N we have n_k โ‰ฅ N.)

Divergence criterion. If two subsequences of (a_n) converge to different limits, then (a_n) diverges. This is the cleanest way to show (-1)โฟ or sin(n ฯ€/2) diverges.

Python Two subsequences, two limits

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The Bolzano-Weierstrass theorem

Theorem. Every bounded sequence of real numbers has a convergent subsequence.

Proof (by nested intervals). Let |a_n| โ‰ค M, so all terms lie in I_0 = [โˆ’M, M]. Bisect I_0 into two halves. At least one half contains a_n for infinitely many n; call it I_1. Bisect I_1 the same way, giving I_2 with infinitely many terms, and so on. We get nested closed intervals I_k of length 2M/2^k โ†’ 0. By the Nested Interval Property there is exactly one point x in all of them. Choose indices n_1 < n_2 < ... with a_{n_k} โˆˆ I_k (always possible since each I_k contains infinitely many terms). Then |a_{n_k} โˆ’ x| โ‰ค length(I_k) โ†’ 0, so a_{n_k} โ†’ x. โˆŽ

An alternative proof: every sequence has a monotone subsequence, which by MCT converges when bounded.

Python Bisection search for a convergent subsequence

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The printout is an illustration with a finite sample. The theorem's proof uses "infinitely many" at each stage, and works for any bounded sequence.

Limit points, limsup and liminf

A number x is a subsequential limit (limit point) of (a_n) if some subsequence converges to x. For a bounded sequence, define

  • limsup a_n = the largest subsequential limit,
  • liminf a_n = the smallest subsequential limit.

Equivalently, limsup a_n = lim_{nโ†’โˆž} sup{a_k : k โ‰ฅ n}; the sequence of tail suprema is decreasing and bounded, so the limit exists by MCT.

Theorem. A bounded sequence converges if and only if limsup a_n = liminf a_n, and then the common value is the limit.

Example: for a_n = (โˆ’1)โฟ(1 + 1/n), the subsequential limits are 1 (even terms) and โˆ’1 (odd terms): limsup = 1, liminf = โˆ’1, so the sequence diverges.

Python Tail suprema approach limsup

Press Run (or Ctrl + Enter) to see the result here.

Program input ยท what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The theorem needs boundedness. The sequence a_n = n is unbounded and has no convergent subsequence at all.

flowchart TD
  A["Bounded sequence in [โˆ’M, M]"] --> B["Bisect: half with infinitely many terms"]
  B --> C["Nested intervals shrinking to a point x"]
  C --> D["Subsequence converging to x"]

The bisection proof of Bolzano-Weierstrass.

Check your understanding

Quick check ยท 4 questions

of 4 answered You scored / 4 ยท perfect!

  1. 1. If a_n โ†’ L, what can be said about any subsequence?

    Later terms of a subsequence are later terms of the original sequence.
  2. 2. How can you show a sequence diverges using subsequences?

    A convergent sequence's subsequences would all share its limit.
  3. 3. The Bolzano-Weierstrass theorem says

    Boundedness is exactly what guarantees a convergent subsequence.
  4. 4. What are limsup and liminf of (โˆ’1)^n?

    The subsequential limits are 1 and โˆ’1.

Summary

  • A subsequence keeps infinitely many terms in the original order; subsequences of a convergent sequence share its limit.
  • Two subsequences with different limits prove divergence.
  • Bolzano-Weierstrass: every bounded sequence has a convergent subsequence (bisection plus nested intervals).
  • limsup and liminf are the largest and smallest subsequential limits; a bounded sequence converges iff they are equal.

Next: Cauchy Sequences