In many cases, filtered colimits commute with forgetful functors to $\mathbf{Set}$, for example with $\mathbf{CRing} \to \mathbf{Set}$ or $R-\mathbf{Mod} \to \mathbf{Set}$. Is there a slick way of showing this?

I am mainly interested in this because you use this fact for the computation of stalks of a sheaf, and usually this is proven by saying: "check it if you don't believe it"

  • $\begingroup$ I don't know about "slick", but there is certainly a general way of showing this for categories of finitary algebraic structures. $\endgroup$ – Zhen Lin Nov 11 '14 at 11:02
  • $\begingroup$ I would be interested in this way. Do you have any resources? The nLab wasn't very helpful (or I overlooked something) $\endgroup$ – Andre Knispel Nov 11 '14 at 18:05
  • $\begingroup$ Well, if you know how to do it for a few examples it is easy to extrapolate (provided you understand what finitary algebraic structures are). $\endgroup$ – Zhen Lin Nov 11 '14 at 18:35

1) Let $T$ be a monad on a cocomplete category $C$. It is a general and trivial fact that the forgetful functor $\mathsf{Mod}(T) \to C$ preserves all colimits which $T$ presveres. In particular, if $T$ preserves filtered colimits (one then says that $T$ is finitary), then the forgetful functor does so. If $T$ is given by a theory of finite operations, $C$ has products which preserve filtered colimits in each variable, then $T$ is finitary.

2) Stalks of sheaves on $X$ commute with colimits because the stalk functor is left adjoint; in fact it is given by $i^*$ where $i : \{x\} \to X$, with right adjoint $i_*$.


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.