ConceptComplete

The Axiom of Choice and Its Formulations

The axiom of choice asserts that every family of nonempty sets admits a choice function. Despite its intuitive appeal, it has surprising consequences and is independent of the other ZF axioms.


The Axiom

Definition6.1Axiom of choice (AC)

For every family {Ai}i∈I\{A_i\}_{i \in I} of nonempty sets, there exists a choice function f:I→⋃i∈IAif: I \to \bigcup_{i \in I} A_i with f(i)∈Aif(i) \in A_i for all i∈Ii \in I.

Equivalently, if F\mathcal{F} is a set of nonempty sets, there exists a function f:F→⋃Ff: \mathcal{F} \to \bigcup \mathcal{F} with f(A)∈Af(A) \in A for all A∈FA \in \mathcal{F}.

RemarkWhy AC is controversial

For finite families, the existence of choice functions is provable in ZF. For countable families of nonempty subsets of R\mathbb{R}, one can sometimes construct explicit choices. But for arbitrary uncountable families of arbitrary nonempty sets, no construction is possible in general -- the axiom simply postulates existence. AC implies the existence of non-measurable sets (Vitali), well-orderings of R\mathbb{R}, and the Banach-Tarski paradox, all of which are counterintuitive.


Equivalent Formulations

Theorem6.1Equivalents of the axiom of choice

The following are equivalent over ZF:

  1. Axiom of choice (AC).
  2. Well-ordering theorem: Every set can be well-ordered.
  3. Zorn's lemma: If every chain in a partially ordered set has an upper bound, then the set has a maximal element.
  4. Hausdorff maximal principle: Every chain in a partially ordered set is contained in a maximal chain.
  5. Tukey's lemma: Every set of finite character has a maximal element under inclusion.
  6. For every surjection f:Aβ† Bf: A \twoheadrightarrow B, there exists a right inverse g:Bβ†’Ag: B \to A with f∘g=idBf \circ g = \mathrm{id}_B.
  7. The Cartesian product of any family of nonempty sets is nonempty.

Weaker Forms

Definition6.2Weaker choice principles

Several weaker forms of AC are used in analysis and topology:

  • Countable choice (ACΟ‰_\omega): Every countable family of nonempty sets has a choice function.
  • Dependent choice (DC): If RR is a binary relation on a nonempty set AA such that for every a∈Aa \in A there exists bb with aRbaRb, then there exists a sequence (an)(a_n) with anRan+1a_n R a_{n+1} for all nn.
  • Boolean prime ideal theorem (BPI): Every Boolean algebra has a prime ideal (equivalent to the ultrafilter lemma).

The implications are: AC β€…β€ŠβŸΉβ€…β€Š\implies DC β€…β€ŠβŸΉβ€…β€Š\implies ACΟ‰_\omega, and none reverses.

ExampleUses of AC in mathematics
  • Every vector space has a basis: requires full AC.
  • Hahn-Banach theorem: requires only BPI.
  • Baire category theorem for complete metric spaces: requires only DC.
  • Every countable union of countable sets is countable: requires ACΟ‰_\omega.
  • Tychonoff's theorem (arbitrary products of compact spaces are compact): equivalent to AC.