I am just finishing teaching a course in Set Theory, and was thinking about the ZFC axioms, when the following axiom occurred to me:

Axiom: Given a sentence $S$ with free variables among $x,y,A_1,\ldots,A_k$

$\forall X,A_1,\ldots,A_k$:


$\forall x\in X, \exists y$ such that $S(x,y,A_1,\ldots,A_k)$

then there exists a function $f$ with domain $X$ such that

$\forall x\in X, S(x,f(x),A_1,\ldots,A_k)$.

(I guess this is really an axiom schema.)

This seems to imply the axiom of choice: Take $k=1$, let $A_1$ be a family of nonempty sets and let $X$ be the indexing set of the family. Let $S$ be the sentence that says that $y$ is in the set indexed by $x$. Then the function guaranteed by the Axiom is a choice function for the family. This also seems to imply the Axiom of Replacement, too, and in fact the new Axiom seems to be equivalent to Replacement plus Choice.

I like the Axiom because it says something I find very intuitive: If you know that for all x there exists a y such that BLAH, you can find a function sending each x to a y such that BLAH.

This is probably not of great importance, but I thought I would ask, in case anyone is aware of an approach that combines the axioms like this.

  • 3
    $\begingroup$ A function $f$ as in your schema is said to uniformize the relation $S$. As you say, requiring uniformizing functions for arbitrary definable relations is just equivalent to Replacement + Choice. Things become more interesting, however, if we require $f$ to be nice in some way. For example, descriptive set theory has many theorems saying that a relation of a given complexity can be uniformized by a function of the same complexity. $\endgroup$ – Miha Habič Apr 20 '16 at 18:49
  • $\begingroup$ Thanks, this is the kind of thing I was curious about. $\endgroup$ – Nathan Reading Apr 21 '16 at 13:08

Taking the clause $\exists y\;[S(x,y,...)]$ and substituting $S'$ for $S$, where $S'$ is $S\land \forall z\;[S(x,z,...)\implies z=y]$ gives us Replacement.

When $S(x,y,...)$ is $(y\in x)\lor (x=y=\phi)$ we obtain a function $f:X\to \{\phi\}\cup (\cup X)$ such that $\forall x\in X\; (x\ne \phi \implies f(x)\in x)$, which is equivalent to AC.

  • $\begingroup$ Right, as I said, it's pretty clear that my Axiom is equivalent to Choice + Replacement. My question was about whether this Axiom is meaningful/useful. Miha Habič's comment actually answers my question fairly well. $\endgroup$ – Nathan Reading Apr 24 '16 at 18:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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