nLab
presentations of (infinity,1)-sheaf (infinity,1)-toposes

Context

(,1)(\infty,1)-Topos Theory

(∞,1)-topos theory

Background

Definitions

Characterization

Morphisms

Extra stuff, structure and property

Models

Constructions

structures in a cohesive (∞,1)-topos

Contents

Idea

An (∞,1)-sheaf – often called a ∞-stack – is the (∞,1)-categorical analog of a sheaf. Just like a category of sheaves is a topos, an (∞,1)-category of (∞,1)-sheaves is an (∞,1)-topos.

There is good motivation for sheaves, cohomology and higher stacks.

Here we recall basic definitions and then concentrate on 1-categorical models that present (∞,1)-categories of ∞-stacks.

What we describe is effectively the old theory of the model structure on simplicial presheaves seen in the new light of Higher Topos Theory.

Plan

We proceed as follow.

Sheaf toposes

It is helpful to briefly recall the story that we want to tell in the category theory context, because in the full higher category theory context it will be literally the same with all notions such as adjoint functor, exact functor etc suitably regarded in the context of (∞,1)-functors.

Presheaves

Consider a category CC that we want to think as a category of “test spaces”. Classical choices would be C=C = Top, the category of topological spaces, C=C = Diff, the category of smooth manifolds or C=Op(X)C = Op(X), the category of open subsets of some topological space XX.

Let Set be the category of sets. We write

PSh(C):=[C op,Set]:=Func(C op,Set) PSh(C) := [C^{op}, Set] := Func(C^{op}, Set)

for the category of presheaves on CC. This is like a category of very general spaces modeled on CC as described at motivation for sheaves, cohomology and higher stacks.

Sheaves

In fact, this is a bit too general for most purposes: the objects of PSh(C)PSh(C) may be very non-local in that they don’t respect the way test objects in CC are supposed to glue together. The full subcategory on those presheaves that do respect some kind of gluing of test objects is the category of sheaves.

Definition

A category of sheaves on CC is a category Sh(C)Sh(C) equipped with a geometric embedding into PSh(C)PSh(C)

Sh(C)PSh(C). Sh(C) \stackrel{\leftarrow}{\to} PSh(C) \,.

Recall that this means that

in other words that

In view of our models for \infty-sheaves it is of importance that this implies an equivalence characterization

Proposition

The category Sh(C)Sh(C) is equivalent to the full subcategory of SS-local presheaves, where SS is the set of local isomorphisms.

Another useful kind of geometric embeddings is that of the point:

let *{*} be the category with a single morphism (the identity on a single object). Then PSh(*)Sh(*)SetPSh({*}) \simeq Sh({*}) \simeq Set . Geometric embeddings

x:Sh(*)Sh(C) x : Sh({*}) \stackrel{\leftarrow}{\to} Sh(C)

are called points of Sh(C)Sh(C). We say that Sh(C)Sh(C) has enough points if isomorphisms of sheaves can be tested on points

(f:AB)Sh(C)x:(x *f:x *Ax *B). (f : A \stackrel{\simeq}{\to} B)\in Sh(C) \;\; \Leftrightarrow \;\; \forall x : (x^* f : x^* A \stackrel{\simeq}{\to} x^* B) \,.

This is the situation we shall concentrate on here.

  • The topos Sh(Diff)Sh(Diff) has enough points, one for every nn \in \mathbb{N}.

  • The topos Sh(Op(X))Sh(Op(X)) has enough points: one for every ordinary point of XX.

If Sh(C)Sh(C) has enough points, we may characterize sheaves in yet another way, which is the one that directly suggests the local model structure on simplicial presheaves discussed below:

Proposition

Let SMor(PSh(C))S \subset Mor(PSh(C)) be the set of stalkwise isomorphisms, i.e. those morphisms f:ABf : A \to B of presheaves such that for all points xx the morphism x *f:x *Ax *Bx^* f : x^* A \to x^* B is an isomorphism (of sets).

If Sh(C)Sh(C) has enough points, then Sh(C)Sh(C) is equivalent to the full subcategory of SS-local presheaves.

The local model structure on simplicial presheaves that we are going to describe is obtained from this description of sheaves by

So the model structures we shall encounter are plausible guesses. What is less trivial is that this plausible structure indeed presents the fully general notion of (∞,1)-sheaf/∞-stack.

This fully general notion we introduce now.

(,1)(\infty,1)-categories and their presentation

An ordinary locally small category is a category enriched over the category Set of sets.

An (∞,0)-category is an ∞-groupoid which we think of as modeled by a simplicial set that is a Kan complex.

Recall that there is a notion of nerve and realization

N:SSet-CatSSet: N : SSet\text{-}Cat \stackrel{\leftarrow}{\to} SSet : |-|

for SSet-enriched categories induced by a cosimplicial simplicially enriched category

Δ SSetCat:ΔSSetCat \Delta_{SSet-Cat} : \Delta \to SSet-Cat

where the nerve operation NN is called the homotopy coherent nerve of simplicially enriched categories.

Definition ((,1)(\infty,1)-category)

An (∞,1)-category is a category enriched over \infty-groupoids, i.e. an SSet-enriched category all whose hom-objects happen to be Kan complexes.

Given two (,1)(\infty,1)-categories C\mathbf{C} and D\mathbf{D} the (∞,1)-functor (,1)(\infty,1)-category is

Func(C,D):=SSet(N(C),N(D)). Func(\mathbf{C}, \mathbf{D}) := |SSet(N(\mathbf{C}), N(\mathbf{D}))| \,.

This is indeed itself an (,1)(\infty,1)-category (HTT, prop 1.2.7.3).

The (∞,1)-category of (∞,1)-categories (,1)Cat(\infty,1)Cat is that whose

  • objects are (,1)(\infty,1)-categories;

  • for C\mathbf{C} and D\mathbf{D} two (,1)(\infty,1)-categories the \infty-groupoid (,1)Cat(C,D)(\infty,1)Cat(\mathbf{C}, \mathbf{D}) is the maximal Kan complex inside the simplicial set of maps between the homotopy coherent nerves

    (,1)Cat(C,D):=Core(SSet(N(C),N(D))). (\infty,1)Cat(\mathbf{C}, \mathbf{D}) := Core( SSet(N(\mathbf{C}), N(\mathbf{D})) ) \,.

Examples

  • Using the monoidal embedding const:SetGrpdfSSetconst : Set \hookrightarrow \infty Grpdf \subset SSet every ordinary category is an (,1)(\infty,1)-category.

  • The (,1)(\infty,1)-category Grpd\infty Grpd (∞Grpd) is the full [[SSet]-subcategory of SSet on Kan complexes.

Definition (homotopy category)

The simplicial connected components functor

π 0:SSetSet \pi_0 : SSet \to Set

is strong monoidal and hence induces a functor

H:(,1)CatCat. H : (\infty,1)Cat \to Cat \,.

The image H(C)H(\mathbf{C}) of an (,1)(\infty,1)-category C\mathbf{C} with H(C)(x,y)=π 0(C(x,y))H(\mathbf{C})(x,y) = \pi_0(\mathbf{C}(x,y)) is the homotopy category of an (∞,1)-category.

Two (,1)(\infty,1)-categories C\mathbf{C} and D\mathbf{D} are equivalent if they are isomorphic in H((,1)Cat)H((\infty,1)Cat)

(f:CDisequivalence)(H (,1)Cat(f):CDisisomorphism). (f : \mathbf{C} \to \mathbf{D} \;is equivalence) \;\; \Leftrightarrow \;\; (H_{(\infty,1)Cat}(f) : \mathbf{C} \to \mathbf{D} \; is isomorphism) \,.

Presentations

It is often convenient to present (,1)(\infty,1)-categories by 1-categorical models.

Definition ((,1)(\infty,1)-category presented by a model category)

For A\mathbf{A} a combinatorial simplicial model category, the (,1)(\infty,1)-category presented by it is the full subcategory A A\mathbf{A}^\circ \subset \mathbf{A} on objects that are both cofibrant and fibrant.

Remark The axioms of a simplicial model category ensure that the hom-simplicial sets of A \mathbf{A}^\circ are indeed Kan complexes. (for instance HTT, remark 3.1.8).

Proposition (HTT, remark A.3.7.7)

Let A\mathbf{A} and B\mathbf{B} be combinatorial simplicial model categories. Then the corresponding (,1)(\infty,1)-categories A \mathbf{A}^\circ and B \mathbf{B}^\circ are equivalent precisely if there is a sequence of SSet-enriched Quillen equivalences

AB. \mathbf{A} \stackrel{\leftarrow}{\to} \stackrel{\to}{\leftarrow} \stackrel{\leftarrow}{\to} \cdots \mathbf{B} \,.

(,1)(\infty,1)-Sheaf (,1)(\infty,1)-toposes

There is now an obvious definition of (,1)(\infty,1)-categories of (,1)(\infty,1)-presheaves and of (,1)(\infty,1)-sheaves by interpreting the 1-categorical story in the (,1)(\infty,1)-categorical context.

(,1)(\infty,1)-Presheaves

Now we generalize the above from sheaves to (∞,1)-sheaves also known as ∞-stacks.

Definition ((,1)(\infty,1)-presheaves)

The (,1)(\infty,1)-category of (∞,1)-presheaves on CC is

PSh (C):=[C op,Grpd]=Func(C op,Grpd). PSh_\infty(C) := [C^{op}, \infty Grpd] = Func( C^{op}, \infty Grpd ) \,.
Proposition (models for (,1)(\infty,1)-presheaves) (HTT, prop. 4.2.4.4,

see also HTT, prop. 5.1.1.1)

The (,1)(\infty,1)-category presented by the global model structure on simplicial presheaves SPSh(C) projSPSh(C)_{proj} on CC (either the projective or the injective one) is equivalent to that of (,1)(\infty,1)-presheaves on CC:

(SPSh(C) proj) (SPSh(C) inj) PSh (C). (SPSh(C)_{proj})^{\circ} \simeq (SPSh(C)_{inj})^{\circ} \simeq PSh_\infty(C) \,.

(,1)(\infty,1)-sheaves

There are (,1)(\infty,1)-category analogs of all the familiar notions from category theory, in particular

Using this we obtain a definition of geometric embedding of (,1)(\infty,1)-toposes , i.e. left exaxt reflective (∞,1)-subcategories by literally copying the 1-categorical definition.

Definition ((,1)(\infty,1)-sheaves) (HTT, def. 6.1.0.4)

An (∞,1)-category of (∞,1)-sheaves is a geometric embedding into an (∞,1)-category of (∞,1)-presheaves

Sh (C)PSh (C). Sh_\infty(C) \stackrel{\leftarrow}{\to} PSh_\infty(C) \,.
Proposition (models for reflective (,1)(\infty,1)-subcategories)

Let the combinatorial simplicial model category B\mathbf{B} be a left Bousfield localization of the combinatorial simplicial model category A\mathbf{A} then

B A \mathbf{B}^\circ \stackrel{\leftarrow}{\to} \mathbf{A}^\circ

is the inclusion of a reflective (∞,1)-subcategory.

Proof

By HTT, prop A.3.7.4 every combinatorial simplicial left Bousfield localization is given by a set SS of cofibrations such that

  • the fibrant objects of B\mathbf{B} are precisely the fibrant objects in A\mathbf{A} that are SS-local object;

  • the weak equivalences of B\mathbf{B} are the SS-local morphisms in A\mathbf{A}.

Accordingly B \mathbf{B}^\circ is the full Grpd\infty Grpd-enriched subcategory of A \mathbf{A}^\circ on SS-local objects. (see also HTT, prop 6.5.2.14).

By HTT, prop. 5.5.4.15 this means that B\mathbf{B} is a reflective (∞,1)-subcategory of A\mathbf{A}.

Remark Notice that this does not yet say that the localization is left exact .

But this makes at least plausible that the local model structure on simplicial presheaves is a presentation for an (∞,1)-category of (∞,1)-sheaves.

That this is indeed the case is

Proposition (model for hypercomplete (,1)(\infty,1)-sheaves) (HTT, prop. 6.5.2.14)

The local model structure on simplicial presheaves SSh(C) proj llocSSh(C)^{l loc}_{proj} presents the hypercompleted version of the (∞,1)-category of (∞,1)-sheaves Sh hc(C)Sh^{hc}(C) on CC.

(SPSh(C) proj loc) Sh hc(C). (SPSh(C)_{proj}^{loc})^\circ \simeq Sh^{hc}(C) \,.

Remark See the discussion at Čech cohomology for the role of hypercompletion.

Applications

Abelian sheaf cohomology as special case of \infty-stackification

The nerve operation of the Dold-Kan correspondence

N:Ch +SimpAbGrpd N : Ch_+ \to SimpAb \subset \infty Grpd

embeds sheaves with values in non-negatively graded chain complexes of abelian groups into simplicial sheaves as those simplicial sheaves with values in Kan complexes that carry a struict abelian group structure. This way homological algebra and abelian sheaf cohomology are realized as special cases of models for \infty-stacks: a complex of abelian sheaves presents a stably abelian \infty-stack.

Proposition

Under the Dold-Kan correspondence abelian sheaf cohomology identifies with the hom-set of the homotopy category corresponding infinity-stack (infinity,1)-topos.

More precisely, let

  • the underlying site be the category of open subsets C=Op(X)C = Op(X) of a topological space XX,

  • let ASh(X)A \in Sh(X) be a sheaf with values in abelian groups on XX;

  • let B nASh(X,SSet)\mathbf{B}^n A \in Sh(X,SSet) be the image of the complex of sheaves A[n]A[-n] concentrated in degree nn under the Dold-Kan nerve;

  • write XSh(X)X \in Sh(X) for the terminal object sheaf in Sh(X)Sh(X) (the sheaf constant on the singleton set).

Then degree nn abelian sheaf cohomology of XX with coefficients in AA is homotopy classes of maps from XX to B nA\mathbf{B}^n A:

H n(X,A)Ho SSh(X)(X,B nA). H^n(X,A) \simeq Ho_{SSh(X)}(X, \mathbf{B}^n A) \,.
Proof

The original proof was given in BrownAHT in terms of the category of fibrant objects structure on locally Kan simplicial sheaves.

The analogous arguments in terms of the full injective model structure were given by Jardine. See section 6 of his lecture notes.

Revised on January 17, 2011 14:03:08 by Urs Schreiber (131.211.233.246)