The small cardinality selection axiom (SCSA) is a weak form of the axiom of choice, which asserts in a certain precise way that choice fails “in at most a small way”. It was introduced by Michael Makkai for the study of anafunctors, and thus it also has consequences for the existence of stack completions.
The “global version” of SCSA states:
Therefore, we have a class function assigning to each set , another set and a bijection , in such a way that takes only set-many values on each isomorphism class in Set.
There is also a “local version” of SCSA which can be stated as follows. Given a function , say that a function equipped with is a local pullback of if there is a pair of pullback squares in which is surjective. Then the local SCSA states:
In more set-theoretic language, given a family of sets , define a family of sets equipped with to be a local reindexing of if for each there exists an isomorphism (but there is not necessarily a function picking out such an isomorphism). Then the local SCSA states that for any has a local reindexing with such that for any other local reindexing with , there exists a function such that and a specified family of isomorphisms .
To deduce the local version from the global version, given let be the set of pairs such that for some , with the first projection and . Then given with such that , define .
In the presence of the axiom of global choice in material set theory, the category Set has a skeleton, namely the category of von Neumann cardinals, which of course implies the global SCSA. Similarly, the ordinary axiom of choice implies that every set is bijective to a unique von Neumann cardinal, and we can choose such a bijection uniformly for any small family of sets, which implies the local SCSA.
Global SCSA also follows from the “global” version of the axiom of small violations of choice, as proven in Makkai’s paper (attributed to the referee). Makkai claims there is a local form of SCSA that follows from the local SVC. I have not attempted to show that this is true for the above local SCSA, although it is a guess at the sort of axiom Makkai might have had in mind.
SCSA implies that the anabicategory of categories and anafunctors is locally essentially small and cartesian closed. Specifically, local SCSA implies that for any two categories there is a small set of anafunctors from to such that any such anafunctor is isomorphic to one in this set. The idea is that we can replace any anafunctor by a saturated one, in which case the set of “specifications that ” for any is, if nonempty, a torsor over , and hence isomorphic to it. Thus, applying local SCSA to the family , we obtain a “universal” set of specifications for saturated anafunctors.
This implies that there exists a full subcategory of whose inclusion is a weak equivalence, and that an exponential exists. If local SCSA holds in “functional” form, meaning there is a class-function specifying weakly terminal local reindexings, then this implies that is weakly equivalent (i.e. ana-equivalent) to a locally small bicategory and that it admits a functor . (Of course, global SCSA implies functional local SCSA.)
Makkai states only the global SCSA, but claims that there is a local version of it that holds in all Grothendieck toposes. The local form of SCSA stated above is a guess at the one Makkai might have had in mind. I have not attempted to verify that it is true in all Grothendieck toposes.
Last revised on May 23, 2021 at 13:39:14. See the history of this page for a list of all contributions to it.