The Axiom of Choice is one of the most debated assumptions in mathematics. It states, roughly, that given any collection of non-empty sets, you can always pick one element from each set simultaneously, even if there is no explicit rule telling you how to do so. A separate and important fact from linear algebra is that every vector space has a "basis," meaning a minimal set of building blocks from which every element of the space can be constructed. Mathematicians have long known that these two statements are equivalent: assuming the standard axioms of set theory, the Axiom of Choice holds if and only if every vector space has a basis.
That classical equivalence, however, was proven using another axiom called the Axiom of Foundation, which roughly says that sets cannot contain themselves or form circular membership chains. A natural question is whether Foundation is truly necessary for the equivalence, or whether it is just a convenient tool that happens to be present in standard set theory. This paper answers that question by showing Foundation is not needed: even if you remove it from the standard axioms, the existence of bases for all vector spaces still implies the Axiom of Choice.
The authors focus specifically on vector spaces over fields of characteristic zero (such as the rational or real numbers), showing that even this restricted version of the "every vector space has a basis" statement is enough to recover the full Axiom of Choice, without relying on Foundation. They also extend the result to set theories that include "atoms," which are objects that are not sets but can still be elements of sets. This strengthens our understanding of why the Axiom of Choice is so deeply tied to linear algebra, regardless of which foundational assumptions are in play.