17
$\begingroup$

In the paper Quantifiers and Sheaves by Lawvere, at the bottom of the second page, the author writes:

"... the condition that every epi splits, which geometrically we would call 0-dimensionality and logically we would call the axiom of choice."

What is meant by this geometric interpretation of choice as zero-dimensionality? I'm guessing it has something to do with fibers but can't figure it out myself.

$\endgroup$

1 Answer 1

23
$\begingroup$

Here are two possible motivating examples.

First, for any topos $\mathcal{E}$, if all epimorphisms in $\mathcal{E}$ split, then $\mathcal{E}$ is a boolean topos. In particular, for a topological space $X$, if all epimorphisms in $\mathbf{Sh} (X)$ split, then every open subset of $X$ is also closed (and vice versa), so $X$ is a disjoint union of indiscrete spaces. The converse is straightforward. Thus, all epimorphisms in $\mathbf{Sh} (X)$ split if and only if $X$ is a disjoint union of indiscrete spaces.

Second, for a commutative ring $A$, if all epimorphisms in $\mathbf{Mod} (A)$ split, then every closed subset of $\operatorname{Spec} A$ is also open (and vice versa), so the Krull dimension of $A$ is zero, therefore $\operatorname{Spec} A$ is finite and discrete, and hence $A$ is a direct product of finitely many fields. Conversely, if $A$ is a direct product of finitely many fields then all epimorphisms in $\mathbf{Mod} (A)$ split.

I suppose the point is that if the category of representations of some geometric object satisfies the axiom of choice, then that object cannot have a very complicated internal structure.

$\endgroup$
1
  • $\begingroup$ I don't see why Spec(A) should be finite and discrete, unless A is noetherian. And how to deduce that A is a finite product of fields, unless A is reduced. $\endgroup$
    – HeinrichD
    Sep 23, 2016 at 11:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.