Want to take part in these discussions? Sign in if you have an account, or apply for one below
Vanilla 1.1.10 is a product of Lussumo. More Information: Documentation, Community Support.
Have internal sup-lattices in quasitoposes been studied much?
Of course they’ve been studied in toposes to a pretty fair extent. In a topos , the covariant power object functor (where for , we define ) carries a structure of monad whose unit is the singleton map and whose multiplication is internal union . In fact this is a lax idempotent or KZ monad in some suitable internal sense, and we define a sup-lattice to be a -algebra. This is all well-studied, and figures heavily in the Joyal-Tierney monograph on a generalized Galois theory à la Grothendieck.
It seems to me that a certain chunk of this can be developed in quasitoposes in parallel fashion. For example, for let me construct by hand. There is a canonical regular subobject classified by the evaluation map . Form the composite
and then take its epi-(regular mono) factorization, using the fact that quasitoposes are coregular. Take the regular image and its classifying map . Curry that to a map ; this is what I call .
Now I’ll construct a map . Form the diagram below where the square is a pullback
and take the regular image of the top composite, say . Curry its classifying map to get the map .
All this is parallel to similar constructions in topos theory; we’re just being mindful to take regular images instead of just images at the appropriate steps. And so if I had to guess, I’d guess that there’s not a great deal of difference: we get a KZ monad on a quasitopos just as we do in topos theory, and there are theorems like -algebra maps are the same as internal left adjoints between algebras, and there’s an adjoint functor theorem that if is a map between sup-lattices that preserves infs, then it has a left adjoint.
Does anyone know about this?
Good question! I don’t recall seeing this developed anywhere, but you could check for instance Wyler’s book on quasitopoi. The only thing that I would worry about at first glance is to what extent the adjoint functor theorem depends on the function comprehension principle, which is false for the logic of regular subobjects in a quasitopos.
Thanks for that heads-up! The adjoint functor theorem I had in mind is a very algebraic form which uses basically the structures available for lax idempotent monads, and so I think such a blunt instrument as the general adjoint functor theorem wouldn’t be needed for the context I had in mind.
The reason I’m asking has to do with an answer I’ve just written up to an MO query by Qiaochu Yuan. Luckily, I saw in the meantime how to circumvent a lot of the complications in getting such a theory of sup-lattices up and running.
1 to 3 of 3