Skip to content
Tutorials Center
My learning

Topics

Lesson 22 of 36 · Compactness

Sequential Compactness and Limit Points

Compare open-cover compactness with sequential compactness and limit point compactness. See the bisection proof of Bolzano–Weierstrass in code.

Open covers are the official definition of compactness, but when you work with real numbers or metric spaces you usually think in terms of sequences: does every sequence have a subsequence that converges? This property is called sequential compactness. In metric spaces it is equivalent to compactness, which is why analysts can mostly forget about covers.

In this lesson you will meet three related properties, learn how they connect, and run the bisection argument behind the Bolzano–Weierstrass theorem.

Three definitions

Let X be a topological space.

  • X is compact if every open cover has a finite subcover.
  • X is sequentially compact if every sequence in X has a subsequence that converges to a point of X.
  • X is limit point compact if every infinite subset of X has a limit point in X. (A limit point p of a set S is a point such that every neighborhood of p contains a point of S other than p.)

A subsequence of (x_n) is a sequence (x_{n_k}) where n_1 < n_2 < n_3 < .... It keeps some terms in the original order and drops the others.

How they relate

In general topological spaces:

Implication Always true?
compact ⇒ limit point compact yes
sequentially compact ⇒ limit point compact yes
compact ⇒ sequentially compact no (needs extra conditions)
sequentially compact ⇒ compact no

In metric spaces all three are equivalent. That is the theorem you will actually use.

For a subset of ℝⁿ, all of these mean the same thing as "closed and bounded". Pick whichever formulation makes your proof easiest.

Bolzano–Weierstrass

Theorem. Every bounded sequence in ℝⁿ has a convergent subsequence.

The proof for n = 1 uses bisection. Suppose (x_k) lies in [a, b]. Split into two halves. At least one half contains x_k for infinitely many k. Keep that half and split again. After m steps you have an interval of length (b − a)/2^m containing infinitely many terms. Choose one term from each interval, with strictly increasing indices. The chosen terms get squeezed together, so they form a Cauchy sequence, and it converges because the reals are complete.

Python Bisection: building a convergent subsequence

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

Program input · 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.

Look at how the interval widths halve each time and the picked terms get squeezed into a shrinking window. Continuing forever would produce a subsequence converging to a point of [−1, 1].

Why the property needs closedness

The sequence x_n = 1/n in (0, 1] is bounded, but every subsequence converges to 0, which is not in the space. So (0, 1] is not sequentially compact. The unbounded sequence x_n = n in ℝ has no convergent subsequence at all. These are the same two failures as in the Heine–Borel lesson, now seen through sequences.

A metric space where bounded is not enough

Take the infinite-dimensional space ℓ² of sequences with Σ x_i² < ∞ and distance ‖x − y‖ = √(Σ (x_i − y_i)²). The unit vectors e_n (one in position n, zero elsewhere) all have norm 1, but ‖e_m − e_n‖ = √2 for m ≠ n. No subsequence can be Cauchy, so none converges. The closed unit ball in ℓ² is closed and bounded but not sequentially compact, hence not compact.

Python Unit vectors stay far apart: the closed unit ball is not compact in infinite dimensions

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

Program input · 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 distances never shrink, so in the limit d → ∞ the sequence e_1, e_2, ... has no convergent subsequence. The ball is closed and bounded, but compactness fails.

A space with sequences but without compactness

There are spaces where sequential compactness and compactness come apart. The set [0, 1]^ℝ (all functions from ℝ to [0, 1] with the product topology) is compact by Tychonoff's theorem, but it is not sequentially compact. In the other direction, the space of countable ordinals [0, ω₁) with the order topology is sequentially compact but not compact. These are advanced examples; the point is that for general spaces you must say which notion you mean. For metric spaces, no such care is needed.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. In a metric space, which statement is true?

    For metric spaces, compactness, sequential compactness and limit point compactness coincide.
  2. 2. Why does the sequence x_n = 1/n show that (0, 1] is not sequentially compact?

    Subsequences of 1/n still tend to 0, and 0 is not a point of the space.
  3. 3. What does the bisection proof of Bolzano–Weierstrass use about the real numbers?

    The chosen terms are squeezed into intervals of length (b−a)/2^m, so they form a Cauchy sequence, and completeness gives a limit.
  4. 4. What goes wrong with the unit vectors e_n in the infinite-dimensional space ℓ²?

    All distances between different unit vectors equal √2, which prevents any subsequence from converging even though the ball is closed and bounded.

Summary

  • Compact, sequentially compact and limit point compact are three related properties; in metric spaces they are equivalent.
  • Bolzano–Weierstrass: every bounded sequence in ℝⁿ has a convergent subsequence (proved by bisection).
  • Subsets of ℝⁿ: compact means closed and bounded, in every formulation.
  • Closed bounded sets in infinite-dimensional spaces need not be compact; the unit vectors e_n show it.
  • In general topological spaces the notions can differ, so name the one you mean.

Next: Connected Spaces