ScalingStacks

5.3. Higher centers

Unlike in classical algebra, for an algebra object in an ∞\infty-category, commutativity is an additional structure rather than a property. More precisely, the forgetful functor from ℰ∞\mathcal{E}_{\infty}-algebras to ℰ1\mathcal{E}_{1}-algebras is conservative but not fully faithful. Given an ℰ∞\mathcal{E}_{\infty}-algebra AA in a symmetric monoidal ∞\infty-category 𝒮\mathcal{S}, the center 𝒵⁡(A)\mathcal{Z}(A) that we have studied to this point does not involve the commutativity of AA. More precisely, it only depends upon the ℰ1\mathcal{E}_{1}-algebra underlying AA.

In this section, we discuss higher versions of the center where we forget less commutativity. For a perfect stack XX, we calculate the ℰn\mathcal{E}_{n}-center of the ℰn\mathcal{E}_{n}-category underlying the ℰ∞\mathcal{E}_{\infty}-category QC⁡(X)\qc(X). A detailed treatment of the foundations of the subject may be found in Chapter 2 of [F1]. Here we summarize only what is required for our consideration of the ℰn\mathcal{E}_{n}-center of QC⁡(X)\qc(X).

0NYF

Remark 5.4. As noted in the introduction, after the completion of this manuscript, the paper [L5] was revised to include a thorough treatment of ∞\infty-categorical operads and their algebras, and the paper [L7] treats in great detail the specific case of the ℰn\mathcal{E}_{n}-operads. We refer the reader to these preprints for further details.

Let FF be a topological operad. We define a topological category ℱ\mathcal{F} with objects finite pointed sets and morphism spaces given by

Mapℱ(J∗,I∗)=∐f:J∗→I∗∏IF(f−1{i}).\Map_{\mathcal{F}}(J_{*},I_{*})=\coprod_{f:J_{*}\rightarrow I_{*}}\prod_{I}F(f^{-1}\{i\}).

We then obtain an ∞\infty-category, also denoted by ℱ\mathcal{F}, by applying the singular functor to the mapping spaces to get a simplicial category and then taking the simplicial nerve.

0NYG

Definition 5.5 ([F1]). An ℱ\mathcal{F}-monoidal structure on an ∞\infty-category 𝒞\mathcal{C} consists of a functor p:ℱ→Cat∞p:\mathcal{F}\rightarrow{\rm Cat}_{\infty} with an identification 𝒞≃p⁡(1∗)\mathcal{C}\simeq p(1_{*}) such that the natural maps p⁡(J∗)→∏Jp⁡(1∗)p(J_{*})\rightarrow\prod_{J}p(1_{*}) resulting from a choice of a point x∈F⁡(J)x\in F(J) is an equivalence.

We will refer to the ∞\infty-category 𝒞\mathcal{C} equipped with an ℱ\mathcal{F}-monoidal structure as an ℱ\mathcal{F}-category. This construction will be of particular interest to us when ℱ\mathcal{F} is the little nn-disks operad ℰn\mathcal{E}_{n}.

To discuss Hochschild cohomology, we must next introduce the notion of operadic modules.

To this end, let Fin∗{{\rm Fin}_{*}} be the category of finite based sets. Let Fin+{\rm Fin}_{+} be the category of “doubly-based sets” with objects finite based sets I∗I_{*} and (I∐+)∗(I\amalg+)_{*}, and morphisms α\alpha of the underlying based sets such that +∈α−1(+)+\in\alpha^{-1}(+), and either α(+)=+\alpha(+)=+ or α(+)=∗\alpha(+)=*.

Given a topological operad FF, recall the ∞\infty-category ℱ\mathcal{F} introduced above. We also define the ∞\infty-category ℱ+\mathcal{F}_{+} to be the fiber product of ∞\infty-categories

ℱ+=ℱ×Fin∗Fin+.\mathcal{F}_{+}=\mathcal{F}\times_{{\rm Fin}_{*}}{\rm Fin}_{+}.
0NYH

Definition 5.6. Let 𝒞\mathcal{C} be an ℱ\mathcal{F}-category. An ℱ\mathcal{F}-𝒞\mathcal{C}-module structure on an ∞\infty-category ℳ\mathcal{M} is a functor q:ℱ+→Cat∞q:\mathcal{F}_{+}\rightarrow{\rm Cat}_{\infty} extending the ℱ\mathcal{F}-monoidal structure on 𝒞\mathcal{C}, together with an identification q⁡(+)≃ℳq(+)\simeq\mathcal{M} such that the natural map q((I∐+)∗)→𝒞I×ℳq((I\amalg+)_{*})\rightarrow\mathcal{C}^{I}\times\mathcal{M} is an equivalence for any II.

0NYI

Example 5.7. When ℱ\mathcal{F} is the ℰ1\mathcal{E}_{1} operad, the notion of an ℱ\mathcal{F}-𝒞\mathcal{C}-module coincides with that of a 𝒞\mathcal{C}-bimodule: it is an ∞\infty-category left and right tensored over 𝒞\mathcal{C}. When ℱ\mathcal{F} is the ℰ∞\mathcal{E}_{\infty} operad, the notion coincides with that of a left (or equivalently right) module.

There is a similar definition of an ℱ\mathcal{F}-algebra and an ℱ\mathcal{F}-AA-module in a symmetric monoidal or ℱ\mathcal{F}-monoidal ∞\infty-category ℳ\mathcal{M} generalizing the definition given above. This requires the notion of an ℱ\mathcal{F}-lax monoidal functor (see [F1, Chapter 2, Definition 3.10]). The simplification of the previous definition is available because Cat∞{\rm Cat}_{\infty} is equipped with the Cartesian monoidal structure. We will avail ourselves of this greater generality in the following definition, although the only case that will concern us in the following is when ℳ\mathcal{M} is either Cat∞{\rm Cat}_{\infty} or 𝒫​rL\mathcal{P}r^{\rm L}.

Now let ℳ\mathcal{M} be a presentable symmetric monoidal ∞\infty-category whose monoidal structure distributes over colimits. Let AA be an ℱ\mathcal{F}-algebra in ℳ\mathcal{M}, and let ModAℱ​(ℳ)\mathrm{Mod}_{A}^{\mathcal{F}}(\mathcal{M}) be the ∞\infty-category of ℱ\mathcal{F}-AA-modules in ℳ\mathcal{M} (see [F1, Chapter 2, Definition 4.3]). Under the above assumptions, the ∞\infty-category ModAℱ​(ℳ)\mathrm{Mod}_{A}^{\mathcal{F}}(\mathcal{M}) is naturally tensored over ℳ\mathcal{M}.

0NYJ

Definition 5.8. For an ℱ\mathcal{F}-algebra AA in ℳ\mathcal{M}, we define the ℱ\mathcal{F}-Hochschild cohomology

HHℱ∗⁡(A)=ℋ​o​mModAℱ​(A,A)\hh^{*}_{\mathcal{F}}(A)={\mathcal{H}om}_{\mathrm{Mod}_{A}^{\mathcal{F}}}(A,A)

to be the object of ℳ\mathcal{M} representing the endomorphisms of AA as an object of ModAℱ​(ℳ)\mathrm{Mod}_{A}^{\mathcal{F}}(\mathcal{M}).

In particular, when ℳ=𝒫​rL\mathcal{M}=\mathcal{P}r^{\rm L}, the ℱ\mathcal{F}-Hochschild cohomology of an ℱ\mathcal{F}-category 𝒞\mathcal{C} is the ∞\infty-category of ℱ\mathcal{F}-𝒞\mathcal{C}-module functors

HHℱ∗⁡(𝒞)=FunMod𝒞ℱ⁡(𝒞,𝒞).\hh^{*}_{\mathcal{F}}(\mathcal{C})=\Fun_{\mathrm{Mod}_{\mathcal{C}}^{\mathcal{F}}}(\mathcal{C},\mathcal{C}).

We now specialize to the case where ℱ\mathcal{F} is the ℰn\mathcal{E}_{n}-operad. Intuitively, an ℰn\mathcal{E}_{n}-algebra structure on an object AA is equivalent to a family of associative algebra structures on AA parameterized by Sn−1S^{n-1} such that antipodal points are associated to opposite multiplications on AA. An ℰn\mathcal{E}_{n}-AA-module structure on MM admits a similar intuitive interpretation as a family of compatible left AA-module structures on MM parameterized by Sn−1S^{n-1}. This intuition leads to the following (to appear in [F2], see also [L7]).

0NYK

Proposition 5.9 ([F2]). Let 𝒞\mathcal{C} be a presentable symmetric monoidal ∞\infty-category whose monoidal structure distributes over colimits.

To an ℱ\mathcal{F}-algebra AA in 𝒞\mathcal{C} there is functorially assigned associative algebra UAU_{A} such that there is a canonical equivalence ModAℱ​(𝒞)≃ModUA​(𝒞)\mathrm{Mod}_{A}^{\mathcal{F}}(\mathcal{C})\simeq\mathrm{Mod}_{U_{A}}(\mathcal{C}) between ℱ\mathcal{F}-AA-modules and left UAU_{A}-modules.

If ℱ\mathcal{F} is the ℰn\mathcal{E}_{n} operad, and the ℰn\mathcal{E}_{n}-algebra structure on AA is obtained by restriction from an ℰ∞\mathcal{E}_{\infty}-algebra structure, then there is a canonical equivalence of associative algebras UA≃Sn−1⊗AU_{A}\simeq S^{n-1}\otimes A.

For n=1n=1, the proposition reduces to the familiar statement that for an ℰ∞\mathcal{E}_{\infty}-algebra AA, an A⊗AopA\otimes A^{\rm op}-module structure (in the form of an ℰ1\mathcal{E}_{1}-AA-module structure) is equivalent to an A⊗AA\otimes A-module structure (in the form of a left UAU_{A}-module structure).

For our current purposes, one can interpret the proposition as furnishing the definition of an ℰn\mathcal{E}_{n}-AA-module. Namely, the reader uncomfortable with the abstractions can take left Sn−1⊗AS^{n-1}\otimes A-modules as the definition of ℰn\mathcal{E}_{n}-AA-modules

0NYL

Example 5.10. Let 𝒞\mathcal{C} be an ℰ1\mathcal{E}_{1}-algebra in 𝒫​rL\mathcal{P}r^{\rm L} (so 𝒞\mathcal{C} is a presentable monoidal ∞\infty-category whose monoidal structure distributes over colimits). Then left modules for the monoidal ∞\infty-category U𝒞U_{\mathcal{C}} are equivalent to 𝒞\mathcal{C}-bimodules. In particular, we have an equivalence U𝒞≃𝒞⊗𝒞opU_{\mathcal{C}}\simeq\mathcal{C}\otimes\mathcal{C}^{\rm op}, where here 𝒞op\mathcal{C}^{\rm op} denotes the ∞\infty-category 𝒞\mathcal{C} equipped with the opposite monoidal structure. As a consequence, we see that in this case, the preceding general definition of ℱ\mathcal{F}-Hochschild cohomology recaptures the notion of the Drinfeld center introduced earlier.

Let AA be an ℰ∞\mathcal{E}_{\infty}-algebra and UAU_{A} be as above. We have a companion definition of ℰn\mathcal{E}_{n}-Hochschild homology.

0NYM

Definition 5.11. For an ℰ∞\mathcal{E}_{\infty}-algebra AA, we define the ℰn\mathcal{E}_{n}-Hochschild homology

HH∗ℰn⁡(A)=A⊗UAA\hh_{*}^{\mathcal{E}_{n}}(A)=A\otimes_{U_{A}}A

to be the tensor product of AA with itself over the algebra UAU_{A}.

We now have the following contribution of this paper to the story.

0NYN

Corollary 5.12. For a perfect stack XX, consider the stable ∞\infty-category QC⁡(X)\qc(X) equipped with its ℰn\mathcal{E}_{n}-tensor product. Then with XSn=Map⁡(Sn,X)X^{S^{n}}=\Map(S^{n},X), there are canonical equivalences

QC⁡(XSn)≃HHℰn∗⁡(QC⁡(X))≃HH∗ℰn⁡(QC⁡(X))\qc(X^{S^{n}})\simeq\hh^{*}_{\mathcal{E}_{n}}(\qc(X))\simeq\hh_{*}^{\mathcal{E}_{n}}(\qc(X))
0NYP

Proof. The result follows from an inductive application of Theorem 4.7 and Corollary 4.12 to the Cartesian diagrams

XSn\textstyle{X^{S^{n}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}X\textstyle{X\ignorespaces\ignorespaces\ignorespaces\ignorespaces}X\textstyle{X\ignorespaces\ignorespaces\ignorespaces\ignorespaces}XSn−1\textstyle{X^{S^{n-1}}}

where the two maps X→XSn−1X\to X^{S^{n-1}} assign to a point of XX the corresponding constant map. ∎

Original mathematics by the credited authors. Source collection and HTML conversion remain in progress.

David Ben-Zvi, John Francis, David Nadler

Original source: arXiv:0805.0157v5