9
$\begingroup$

I have a very simple question regarding geometric morphisms of $\infty$ topoi, but have been unable to find the answer in Lurie's HTT (although it seems likely that its there somewhere and I just can't find it).

Classically, given two sites $C$ and $D$, a morphism of sites $f:C\to D$ induces a geometric morphism of topos $f_{*}:Shv(C)\to Shv(D)$. Does the statement still hold in the $\infty$ context?

What I'm really after is the following: Let $C$ be an $\infty$ site and let $c$ be an object of $C$. Then $C_{/c}$ inherits a topology from $C$ by restricting the covering sieves to objects in the overcategory. The forgetful functor $f:C_{/c}\to C$ is trivially a morphism of $\infty$ sites. Does $f$ induce a functor $$ f^*:Shv(C)\to Shv(C_{/c})$$ preserving finite $\infty$ limits?

$\endgroup$
2
  • $\begingroup$ It might be valuable to record something in the nlab $\endgroup$ Oct 29 '15 at 14:23
  • 1
    $\begingroup$ The particular case you care about is covered by the fact that $Shv(C/c)\simeq Shv(C)/y(c),$ where $y$ is the Yoneda embedding. With this insight, the geometric morphism you seek IS in HTT; it's an etale geometric morphism. The fact I claimed is Proposition 2.2.1 here: arxiv.org/abs/1312.2204 $\endgroup$ Nov 7 '15 at 7:34
12
$\begingroup$

I never found any discussion about this in HTT either, but it turns out to work exactly the same way as in SGA 4.

That is, let $f^* : P(D) \to P(C)$ denote the restriction functor on presheaves, and $f_!$ its left adjoint (given by left Kan extension). Since $f$ preserves covering families, $f_!$ preserves local equivalences, so by adjunction $f^*$ preserves sheaves and induces a functor $$ f^*_s : Sh(D) \to Sh(C). $$ Define $$ f_!^s := a_D f_! i_C : Sh(C) \to Sh(D), $$ where $i_C$ is the inclusion $Sh(C) \hookrightarrow P(C)$ and $a_D : P(D) \to Sh(D)$ is the associated sheaf functor. It is straightforward to check that this is left adjoint to $f^*_s$. The fact that $f_!^s$ is left-exact follows from the fact that $i_C$ commutes with (small) limits, $f_!$ commutes with finite limits, and $a_D$ commutes with finite limits (I am assuming of course that $D$ is small).

The fact that $f_! : P(C) \to P(D)$ commutes with finite limits follows from the pointwise formula for the Kan extension $$ f_! F(d) = \varinjlim_{(c \in C, d \to f(c)) \in (d/f)} F(c) $$ where, since $f$ is left-exact, the comma category $(d/f)$ is filtered. (The commutativity of filtered colimits and finite limits in an $\infty$-topos is Example 7.3.4.7 in HTT.)

$\endgroup$
2
  • $\begingroup$ ah, perfect. I thought the argument might go the same way as in the classical case but I wasn't completely sure. Thanks for the response. $\endgroup$ Oct 28 '15 at 11:05
  • 1
    $\begingroup$ Of course, as in SGA, we may replace the assumption that $f : C \to D$ preserves finite limits with the assumption that $f_! : \mathrm P (C) \to \mathrm P (D)$ does (or, better, that $a_D f_! : \mathrm P (C) \to \mathrm{Sh} (D)$ does) when $C$ does not have all finite limits. $\endgroup$
    – Zhen Lin
    Oct 28 '15 at 11:51

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.