MathLabs
TheoremProved

Bolzano–Weierstrass theorem (bisection proof)

Statement

Every bounded sequence (xn)(x_n) of real numbers has a convergent subsequence (xnk)(x_{n_k}).

Why is it true?

This is the key compactness fact that makes real analysis work: it guarantees that a bounded process cannot wander forever without accumulating somewhere, and it underlies proofs of the extreme value theorem, existence of minimizers in optimization, and completeness arguments throughout analysis.

Proof sketch

Let (xn)(x_n) be bounded, so there exist a<ba<b with xn∈[a,b]x_n\in[a,b] for every nn. We build a nested sequence of intervals by repeated bisection. Split [a,b][a,b] into its two halves [a,a+b2][a,\tfrac{a+b}{2}] and [a+b2,b][\tfrac{a+b}{2},b]. Since the sequence has infinitely many terms (counted with index) and only two halves are available, by the pigeonhole principle at least one half must contain xnx_n for infinitely many indices nn; call that half [ak,bk][a_k,b_k] with k=1k=1.

Repeat the same bisection on [a1,b1][a_1,b_1]: split it in two, and again by pigeonhole at least one half contains xnx_n for infinitely many nn; call it [a2,b2][a_2,b_2]. Continuing forever produces a nested chain [a,b]⊃[a1,b1]⊃[a2,b2]⊃⋯[a,b]\supset[a_1,b_1]\supset[a_2,b_2]\supset\cdots, each containing xnx_n for infinitely many indices, and each half the width of the previous one, so the width of [ak,bk][a_k,b_k] is exactly (b−a)/2k(b-a)/2^{k}, which tends to 00 as k→∞k\to\infty.

Now build the subsequence: since [a1,b1][a_1,b_1] contains infinitely many terms of the sequence, pick any index n1n_1 with xn1∈[a1,b1]x_{n_1}\in[a_1,b_1]. Since [a2,b2][a_2,b_2] also contains infinitely many terms (all but finitely many indices remain available), pick n2>n1n_2>n_1 with xn2∈[a2,b2]x_{n_2}\in[a_2,b_2]. Continuing inductively, at each step kk choose nk>nk−1n_k>n_{k-1} with xnk∈[ak,bk]x_{n_k}\in[a_k,b_k]; this is always possible because [ak,bk][a_k,b_k] contains infinitely many terms, so infinitely many indices beyond nk−1n_{k-1} remain.

By the nested interval property of the real numbers (each [ak,bk][a_k,b_k] is closed, nested, and their widths shrink to 00), the intersection is a single point: ⋂k=1∞[ak,bk]={L}\bigcap_{k=1}^{\infty} [a_k,b_k] = \{L\} for some L∈[a,b]L\in[a,b]. Since both xnkx_{n_k} and LL lie in [ak,bk][a_k,b_k], whose width is (b−a)/2k(b-a)/2^{k}, we get ∣xnk−L∣≤(b−a)/2k→0|x_{n_k}-L|\le (b-a)/2^{k}\to 0. As k→∞k\to\infty, the right-hand side tends to 00, forcing xnk→Lx_{n_k}\to L. Thus (xnk)(x_{n_k}) is a convergent subsequence of (xn)(x_n), as required.

Topics that use this theorem

Step-by-step proofs

No step-by-step proof yet for this theorem.

References

  1. Morris Kline (1980). Mathematics: The Loss of Certainty
  2. Maryna Viazovska (2016). The sphere packing problem in dimension 8 · arXiv:1603.04246
  3. DeepMind (2024). AI solves IMO problems at silver medal level