At least for weak forms of Choice, I've found df-acn 
<https://us.metamath.org/mpeuni/df-acn.html> to be pretty helpful. 
Countable Choice would correspond to an antecedent of AC_ _om = _V ("you 
can get a choice function for a countable collection of arbitrary-sized 
sets").

On Friday, November 21, 2025 at 10:54:45 PM UTC-5 [email protected] wrote:

> Oooh, nice historical note.
>
> Also makes me muse a bit about exploring axioms via $a (as in ax-cc ) or 
> via including them as explicit hypotheses/antecedents, as in notations like 
> CHOICE (set.mm and iset.mm) or CCHOICE (iset.mm). The definition checker 
> would complain if DV conditions were missing from 
> https://us.metamath.org/ileuni/df-cc.html .
> On 11/21/25 18:53, Matthew House wrote:
>
> I'm mainly just putting this up in case someone else notices this, since I 
> couldn't find anything else about it. I've recently been trawling through 
> old versions of set.mm, and I noticed that from 2013 to 2016, ax-cc 
> <https://us.metamath.org/mpeuni/ax-cc.html> as written was inconsistent 
> with the rest of the ZFC axioms. As first introduced 
> <https://github.com/metamath/set.mm/commit/03160ffca94aec05c482f455f6140102d44cc48b>
>  
> to set.mm, it was written:
>
>   ${
>     $( The axiom of countable choice (CC).  It is clearly a special case of
>        ~ ac5 , but is weak enough that it can be proven using DC (see
>        ~ axcc ).  It is, however, strictly stronger than ZF and cannot be
>        proven in ZF. $)
>     ax-cc $a |- ( x ~~ om ->
>        E. f A. z e. x ( z =/= (/) -> ( f ` z ) e. z ) ) $.
>   $}
>
> Notice that this has no DV conditions, and thus it includes the statement |- 
> ( x ~~ om -> E. z A. z e. x ( z =/= (/) -> ( z ` z ) e. z ) ), to which 
> there are obvious counterexamples if we assume ax-inf 
> <https://us.metamath.org/mpeuni/ax-inf.html> or ax-inf2 
> <https://us.metamath.org/mpeuni/ax-inf2.html>. This was quietly rectified 
> in a 2016 commit 
> <https://github.com/metamath/set.mm/commit/cfb23de8be111e40084f4921a3718263dba63077>
>  
> by NM, which added the missing DV condition:
>
>    ${
> +    $d f n x z y $.
>      $( The axiom of countable choice (CC), also known as the axiom of
>         denumerable choice.  It is clearly a special case of ~ ac5 , but 
> is weak
>         enough that it can be proven using DC (see ~ axcc ).  It is, 
> however,
>         strictly stronger than ZF and cannot be proven in ZF. It states 
> that any
>         countable collection of non-empty sets must have a choice function.
>         (Contributed by Mario Carneiro, 9-Feb-2013.) $)
>      ax-cc $a |- ( x ~~ om ->
>         E. f A. z e. x ( z =/= (/) -> ( f ` z ) e. z ) ) $.
>    $}
>
> This appears to be the only historical inconsistency in set.mm that was 
> not directly marked as such.
>
> Matthew House
>
> -- 
> You received this message because you are subscribed to the Google Groups 
> "Metamath" group.
> To unsubscribe from this group and stop receiving emails from it, send an 
> email to [email protected].
> To view this discussion visit 
> https://groups.google.com/d/msgid/metamath/c0cd4d1c-9d87-4ad1-840c-630351b95a88n%40googlegroups.com
>  
> <https://groups.google.com/d/msgid/metamath/c0cd4d1c-9d87-4ad1-840c-630351b95a88n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"Metamath" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion visit 
https://groups.google.com/d/msgid/metamath/2b78d2ad-3cab-4cb6-91a3-e44616019681n%40googlegroups.com.

Reply via email to