Let $\mathcal{C}$ and $\mathcal{D}$ be pretopoi, and let $f: \mathcal{C} \rightarrow \mathcal{D}$ be a pretopos functor (that is, a functor which preserves finite coproducts, finite limits, and epimorphisms). Let $M( \mathcal{C} )$ be the category of models of $\mathcal{C}$ (that is, pretopos functors from $\mathcal{C}$ to the category of sets) and define $M( \mathcal{D} )$-similarly.

The conceptual completeness theorem of Makkai-Reyes asserts that if $f$ induces an equivalence of categories $M(f): M( \mathcal{D} ) \rightarrow M( \mathcal{C} )$, then $f$ is itself an equivalence of categories.

I am wondering about the following more general situation. Suppose that the functor $M(f)$ is an op-fibration in sets (in other words, that the category $M( \mathcal{D} )$ can be obtained by applying the Grothendieck construction to a functor from $M( \mathcal{C} )$ to the category of sets). I would like to conclude that $\mathcal{D}$ can be obtained as a filtered colimit of pretopoi of the form $\mathcal{C}_{ / C}$ (in other words, that $\mathcal{D}$ is the pretopos associated to a Pro-object of $\mathcal{C}$).

Is something like this true, and/or available in the categorical logic literature?

  • 1
    $\begingroup$ Might Breiner's 'Logical schemes' approach (andrew.cmu.edu/user/awodey/students/breiner.pdf) be helpful. He considers conceptual completeness in section 4.4. "We reframe Makkai & Reyes’ conceptual completeness theorem as a theorem about schemes…The theorem follows immediately from our scheme construction." (p. 9). $\endgroup$ Jun 30, 2016 at 8:23

1 Answer 1


I believe the answer is:

"Yes" for pretopoi associated to classical first-order theories.

"No, but close" in general. (Close means: you cannot conclude that $\mathcal{D}$ is pro-etale over $\mathcal{C}$, but you can "cover" $\mathcal{D}$ by objects which are pro-etale over both.)

  • 1
    $\begingroup$ Since you are developing an $\infty$-categorical version of Makkai-Reyes conceptual completeness in A.9 of Spectral Algebraic Geometry, what can be said of the analogue of your question and answer? $\endgroup$ Jun 30, 2016 at 11:40
  • 1
    $\begingroup$ @David: You can ask the same question for infty-pretopoi. The answer is "yes" when C is initial (which motivated the question). That is, a bounded coherent infty-topos X is profinite iff all maps between geometric morphisms f_*: S -> X are invertible. The answer is "no" in general (the 1-categorical counterexample gives an infty-categorical counterexample), but I think the idea can be salvaged by using a more liberal notion of pro-etale (a similar phenomenon appears in the work of Bhatt-Scholze on pro-etale maps in algebraic geometry). $\endgroup$ Jul 3, 2016 at 6:46
  • $\begingroup$ Thanks! As to whether there's a categorical logician who might be able to speak to the original question (and perhaps the infty-analogue), Steve Awodey (supervisor of the Breiner I mentioned above) should be well-placed. There's a 4 page summary of their scheme-theoretic approach here: andrew.cmu.edu/user/sbreiner/documents/…, including suggestions towards non-affine logical schemes and pseudoelementary classes. $\endgroup$ Jul 4, 2016 at 13:42
  • $\begingroup$ I'd be curious to see this "more liberal notion of pro-etale map". $\endgroup$
    – David Roberts
    Jul 27, 2016 at 14:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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