This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Hochschild cohomology for functors on linear symmetric monoidal categories

Nadia Romero111[email protected]
Departamento de Matemáticas,
Universidad de Guanajuato.
Abstract

Let RR be a commutative ring with unit. We develop a Hochschild cohomology theory in the category \mathcal{F} of linear functors defined from an essentially small symmetric monoidal category enriched in RR-Mod, to RR-Mod. The category \mathcal{F} is known to be symmetric monoidal too, so one can consider monoids in \mathcal{F} and modules over these monoids, which allows for the possibility of a Hochschild cohomology theory. The emphasis of the article is in considering natural hom constructions appearing in this context. These homs, together with the abelian structure of \mathcal{F} lead to nice definitions and provide effective tools to prove the main properties and results of the classical Hochschild cohomology theory.

Keywords: Hochschild cohomology, enriched monoidal categories, linear functors.

AMS MSC (2020): 18M05, 18D20, 18G90.

1 Introduction

We consider an essentially small symmetric monoidal category, 𝒳\mathcal{X}, enriched in RR-Mod with RR a commutative ring with unit, and then RR-linear functors from 𝒳\mathcal{X} to RR-Mod. This category of functors, denoted by \mathcal{F}, is an abelian, symmetric monoidal, closed category, via Day’s convolution. Given a monoid AA in \mathcal{F}, we have then a category a AA-modules, AA-Mod. We refer the reader to [9] or [7] for the generalities on monoids and modules over them. The main example we have in mind when considering these hypothesis is 𝒳\mathcal{X} as the biset category and \mathcal{F} as the category of biset functors (see, for example, [3]). Monoids in this case are called Green biset functors and have been extensively studied in the last years, in particular, in [4] the commutant and the center of a Green biset functor are studied. As we will see, the commutant of a Green biset functor is the Hochschild cohomology functor of degree zero. So, it is a natural question to ask for a Hochschild cohomology theory of Green biset functors. The purpose of the paper is to develop the main results and properties of this theory for monoids in \mathcal{F}, in order to apply them to biset functors in a forthcoming paper in collaboration with Serge Bouc.

Even though there are recent articles about Hochschild cohomology in monoidal categories (see for example [1] and [6]), none of them is suited for a direct application to our case. So, we take a different and new approach by making use of the internal hom functor in \mathcal{F}, defined in Section 3. The advantage of working with functors from 𝒳\mathcal{X} to RR-Mod is that we can give an explicit construction of the internal hom functor, via the Yoneda-Dress construction (see chapters 2 and 3 of [7] for the abstract definition). With this, the definition of the Hochschild cochain complex of AA appears in a natural way in the category \mathcal{F}. In particular, we can give an explicit description of its arrows, in terms of these homs, which allows for a better understanding of the Hochschild cohomology functors, denoted by Hi(A,M)\mathcal{H}H^{i}(A,\,M), for MM an AA-bimodule and ii a natural number.

The article is inspired in Loday’s presentation of the classical Hochschild cohomology theory (see [8]). So, in Section 4 we introduce the bar resolution of a monoid AA in \mathcal{F} and we define the Hochschild cochain complex of AA. In Section 5 we deal with the important example of a separable monoid. The main result is Theorem 5.6, in which we show that if AA is a separable monoid in \mathcal{F}, then, for any AA-bimodule MM, the functors Hi(A,M)\mathcal{H}H^{i}(A,\,M) are zero for very i>0i>0 and, conversely, if AA is a monoid in \mathcal{F} such that the first Hochschild cohomology functor H1(A,M)\mathcal{H}H^{1}(A,\,M) is zero for any AA-bimodule MM, then AA is separable. As a corollary, we obtain that the Hochschild cohomology functors, for i>0i>0, of the Burnside biset functor, RBRB, as well as those of the biset functor of rational representations, kRkR_{\mathbb{Q}}, over a filed kk of characteristic 0, are all zero.

Finally, in Section 6, we describe the Hochschild cohomology functors of degrees 0, 1 and 2. As in the classical case, the degree 1 can be described in terms of derivations and the degree 2 is described in terms of extensions. Also, following [5], for a monoid AA in \mathcal{F}, we endow the coproduct iHi(A,A)\bigoplus_{i\in\mathbb{N}}\mathcal{H}H^{i}(A,\,A) with a structure of a graded monoid and call it the Hochschild cohomology monoid. We try too keep the proofs of the statements in this last section as brief as possible, since there are many straightforward calculations that work as in the classical case.

2 Preliminaries

For an object xx in a category 𝒞\mathcal{C}, we denote the identity morphism of xx simply as xx.

In what follows RR is a commutative ring with identity and (𝒳,, 1,α,λ,ρ,s)(\mathcal{X},\,\diamond,\,\mathbf{1},\,\alpha,\,\lambda,\,\rho,\,s) is an essentially small symmetric monoidal category enriched in RR-Mod, in the sense of Definition 3.1.51 of [7]. In particular the functor __:𝒳×𝒳𝒳\,\_\,\diamond\,\_\,:\mathcal{X}\times\mathcal{X}\rightarrow\mathcal{X} is RR-bilinear. The category of RR-linear functors from 𝒳\mathcal{X} to RR-Mod is denoted by \mathcal{F}. Recall that \mathcal{F} is an abelian, symmetric monoidal, closed category with identity given by I=𝒳(𝟏,_)I=\mathcal{X}(\mathbf{1},\,\_\,) (see for example Section 3.3 in [7]). We denote the tensor product of two objects MM and NN in \mathcal{F} as MNM\otimes N. With this, the complete notation for \mathcal{F} is (,,I,α,λ,ρ,S)(\mathcal{F},\,\otimes,\,I,\,\alpha^{\mathcal{F}},\,\lambda^{\mathcal{F}},\,\rho^{\mathcal{F}},\,S). As it is customary, we will avoid, whenever is possible, the parenthesis of association in tensor product of several objects in \mathcal{F}.

Given a monoid AA in \mathcal{F}, we denote its product by μ:AAA\mu:A\otimes A\rightarrow A, in case of confusion we may write μA\mu_{A} instead of μ\mu. We recall that we have a morphism e:IAe:I\rightarrow A, or eAe_{A}, and the following commuting diagrams,

1.
A(AA)αAμAAμA(AA)AμAAAμ
2.IAeAλAAμAIAeρA
1.\quad\lx@xy@svg{\hbox{\raise 0.0pt\hbox{\kern 30.36105pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\kern 0.0pt\offinterlineskip\halign{\entry@#!@&&\entry@@#!@\cr&\\&&\\&\crcr}}}\ignorespaces{\hbox{\kern-30.36105pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A\otimes(A\otimes A)\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 0.0pt\raise-35.91666pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.75pt\hbox{$\scriptstyle{\cong}$}}}\kern 3.0pt}}}}}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern-13.08902pt\raise-35.91666pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.19028pt\hbox{$\scriptstyle{\alpha^{\mathcal{F}}}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 0.0pt\raise-66.33331pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 31.9049pt\raise 6.07222pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.71112pt\hbox{$\scriptstyle{A\otimes\mu}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 54.36105pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 54.36105pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A\otimes A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 92.67004pt\raise-13.20189pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-0.8264pt\hbox{$\scriptstyle{\mu}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 111.58322pt\raise-32.94547pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern-3.0pt\raise-38.41666pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 67.97214pt\raise-38.41666pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 111.58322pt\raise-38.41666pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A}$}}}}}}}{\hbox{\kern-30.36105pt\raise-76.83331pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{(A\otimes A)\otimes A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 31.9049pt\raise-70.7611pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.71112pt\hbox{$\scriptstyle{\mu\otimes A}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 54.36105pt\raise-76.83331pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 54.36105pt\raise-76.83331pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A\otimes A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 97.89554pt\raise-59.39581pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-0.8264pt\hbox{$\scriptstyle{\mu}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 114.63596pt\raise-41.41666pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces}}}}\ignorespaces\quad 2.\quad\lx@xy@svg{\hbox{\raise 0.0pt\hbox{\kern 15.45134pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\kern 0.0pt\offinterlineskip\halign{\entry@#!@&&\entry@@#!@\cr&&\\&&\crcr}}}\ignorespaces{\hbox{\kern-15.45134pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{I\otimes A\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 17.47443pt\raise 5.68333pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.1pt\hbox{$\scriptstyle{e\otimes A}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 39.45134pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 25.64903pt\raise-12.15373pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.75pt\hbox{$\scriptstyle{\cong}$}}}\kern 3.0pt}}}}}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 16.9765pt\raise-24.4176pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-3.11389pt\hbox{$\scriptstyle{\lambda^{\mathcal{F}}}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 49.31242pt\raise-32.77414pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern 39.45134pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A\otimes A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 56.06242pt\raise-15.62498pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-0.8264pt\hbox{$\scriptstyle{\mu}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 56.06242pt\raise-27.41666pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 96.67351pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A\otimes I\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 74.6966pt\raise 5.68333pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.1pt\hbox{$\scriptstyle{A\otimes e}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 72.67351pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 75.22394pt\raise-12.15044pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.75pt\hbox{$\scriptstyle{\cong}$}}}\kern 3.0pt}}}}}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 82.53pt\raise-24.17125pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.50974pt\hbox{$\scriptstyle{\rho^{\mathcal{F}}}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 62.81244pt\raise-32.76755pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern-3.0pt\raise-37.24997pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 49.31242pt\raise-37.24997pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A}$}}}}}}}{\hbox{\kern 109.12485pt\raise-37.24997pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces

in \mathcal{F}.

In what follows AA and CC are monoids in \mathcal{F}. A morphism of monoids from AA to CC is an arrow f:ACf:A\rightarrow C in \mathcal{F} such that the following diagram commutes

AA\textstyle{A\otimes A\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}μA\scriptstyle{\mu_{A}}ff\scriptstyle{f\otimes f}A\textstyle{A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}f\scriptstyle{f}CC\textstyle{C\otimes C\ignorespaces\ignorespaces\ignorespaces\ignorespaces}μC\scriptstyle{\mu_{C}}C\textstyle{C}

and feA=eCf\circ e_{A}=e_{C}.

The tensor product ACA\otimes C, of AA and CC, is a monoid with product given by

(AC)(AC)\textstyle{(A\otimes C)\otimes(A\otimes C)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}AS2, 3C\scriptstyle{A\otimes S_{2,\,3}\otimes C}(AA)(CC)\textstyle{(A\otimes A)\otimes(C\otimes C)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}μAμC\scriptstyle{\mu_{A}\otimes\mu_{C}}AC,\textstyle{A\otimes C,}

where S2, 3S_{2,\,3} is the symmetry CAACC\otimes A\rightarrow A\otimes C. When needed, as it was in this case, we will denote the symmetry with subindex.

A left AA-module is an object MM in \mathcal{F} together with an arrow ν:AMM\nu:A\otimes M\rightarrow M such that the following diagrams commute

1.
A(AM)αAνAMνM(AA)MμMAMν
2.IMeMλAMνM
1.\quad\lx@xy@svg{\hbox{\raise 0.0pt\hbox{\kern 32.00688pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\kern 0.0pt\offinterlineskip\halign{\entry@#!@&&\entry@@#!@\cr&\\&&\\&\crcr}}}\ignorespaces{\hbox{\kern-32.00688pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A\otimes(A\otimes M)\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 0.0pt\raise-35.91666pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.75pt\hbox{$\scriptstyle{\cong}$}}}\kern 3.0pt}}}}}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern-13.08902pt\raise-35.91666pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.19028pt\hbox{$\scriptstyle{\alpha^{\mathcal{F}}}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 0.0pt\raise-66.33331pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 33.93073pt\raise 5.68333pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.1pt\hbox{$\scriptstyle{A\otimes\nu}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 56.00688pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 56.00688pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A\otimes M\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 96.77193pt\raise-13.43681pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.50694pt\hbox{$\scriptstyle{\nu}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 116.52068pt\raise-32.0542pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern-3.0pt\raise-38.41666pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 71.26378pt\raise-38.41666pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 116.52068pt\raise-38.41666pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{M}$}}}}}}}{\hbox{\kern-32.00688pt\raise-76.83331pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{(A\otimes A)\otimes M\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 32.39867pt\raise-70.7611pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.71112pt\hbox{$\scriptstyle{\mu\otimes M}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 56.00688pt\raise-76.83331pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 56.00688pt\raise-76.83331pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A\otimes M\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 102.94923pt\raise-58.71527pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.50694pt\hbox{$\scriptstyle{\nu}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 120.96143pt\raise-41.41666pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces}}}}\ignorespaces\quad 2.\quad\lx@xy@svg{\hbox{\raise 0.0pt\hbox{\kern 17.09717pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\kern 0.0pt\offinterlineskip\halign{\entry@#!@&&\entry@@#!@\cr&\\&&\crcr}}}\ignorespaces{\hbox{\kern-17.09717pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{I\otimes M\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 17.96817pt\raise 5.68333pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.1pt\hbox{$\scriptstyle{e\otimes M}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 41.09717pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 26.43353pt\raise-11.75977pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.75pt\hbox{$\scriptstyle{\cong}$}}}\kern 3.0pt}}}}}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 18.19289pt\raise-24.02364pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-3.11389pt\hbox{$\scriptstyle{\lambda^{\mathcal{F}}}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 50.95824pt\raise-31.98619pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}\ignorespaces\ignorespaces{\hbox{\lx@xy@drawline@}}\ignorespaces{\hbox{\lx@xy@drawline@}}{\hbox{\kern 41.09717pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A\otimes M\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 59.35406pt\raise-15.62498pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-1.50694pt\hbox{$\scriptstyle{\nu}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 59.35406pt\raise-27.41666pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern-3.0pt\raise-37.24997pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{}$}}}}}}}{\hbox{\kern 50.95824pt\raise-37.24997pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{M}$}}}}}}}{\hbox{\kern 101.61096pt\raise-37.24997pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{}$}}}}}}}\ignorespaces}}}}\ignorespaces

in \mathcal{F}. Right AA-modules are defined in an analogous way. Morphisms of AA-modules are defined in the obvious way and the subcategory of \mathcal{F} of left AA-modules is denoted by AA-Mod.

We denote the opposite of AA in \mathcal{F} as AopA^{op}. That is, AopA^{op} denotes the object AA with opposite monoidal structure, μS:AAA\mu\circ S:A\otimes A\rightarrow A. The monoid AA is called commutative if it is equal to its opposite, i.e. μS=μ\mu\circ S=\mu.

Recall that CopC^{op}-modules identify with right CC-modules and thus that (A,C)(A,\,C)-bimodules identify with ACopA\otimes C^{op}-modules in the following way.

Remark 2.1.

If MM is an (A,C)(A,\,C)-bimodule, with compatible actions lA:AMMl_{A}:A\otimes M\rightarrow M and rC:MCMr_{C}:M\otimes C\rightarrow M, then the action of ACopA\otimes C^{op} on MM is given by

(ACop)M\textstyle{(A\otimes C^{op})\otimes M\ignorespaces\ignorespaces\ignorespaces\ignorespaces}AS2, 3\scriptstyle{A\otimes S_{2,\,3}}AMC\textstyle{A\otimes M\otimes C\ignorespaces\ignorespaces\ignorespaces\ignorespaces}lAC\scriptstyle{l_{A}\otimes C}MC\textstyle{M\otimes C\ignorespaces\ignorespaces\ignorespaces\ignorespaces}rC\scriptstyle{r_{C}}M.\textstyle{M.}

The symmetry is actually S2, 3:CMMCS_{2,\,3}:C\otimes M\rightarrow M\otimes C, but in ACopA\otimes C^{op} we consider CC with the opposite product, as above. On the other direction, if MM is an ACopA\otimes C^{op}-module we obtain the actions of AA and CC by composing with AeCop:AIACopA\otimes e_{C^{op}}:A\otimes I\rightarrow A\otimes C^{op} and eACop:ICopACope_{A}\otimes C^{op}:I\otimes C^{op}\rightarrow A\otimes C^{op}.

We denote the monoid AAopA\otimes A^{op} by AeA^{e}. We may indistinctly write AeA^{e}-module, (A,A)(A,\,A)-bimodule or even AA-bimodule.

Example 2.2.

Let 𝒞\mathcal{C} be the biset category for the class of finite groups over the ring RR, as defined in [3], and let 𝒞,R\mathcal{F}_{\mathcal{C},\,R} be the category of biset functors, that is, RR-linear functors from 𝒞\mathcal{C} to RR-Mod. Then 𝒞\mathcal{C}, with the direct product of groups and the trivial group, satisfies the hypothesis of 𝒳\mathcal{X} and so 𝒞,R\mathcal{F}_{\mathcal{C},R} those of \mathcal{F}. In this case, the functor II is the Burnside functor RBRB.

3 Hom functors

We begin by recalling the construction of the internal hom object in \mathcal{F}. The general definition of this object can be found in Section 3 of [7] but, since we are working with functors on RR-Mod, we follow the lines of [3], to give an explicit construction.

For each object x𝒳x\in\mathcal{X}, we have an RR-linear functor px:𝒳𝒳\mathrm{p}_{x}:\mathcal{X}\rightarrow\mathcal{X}, given by the monoidal structure of 𝒳\mathcal{X}, that is, it sends an object yy to yxy\diamond x and an arrow φ\varphi to φx\varphi\diamond x. This in turn allows us to define the endofunctor Px:\mathrm{P}_{x}:\mathcal{F}\rightarrow\mathcal{F} called the Yoneda-Dress construction. In an object FF, it is defined as Fx:=FpxF_{x}:=F\circ\mathrm{p}_{x} and in an arrow f:MNf:M\rightarrow N as (Px(f))y=fyx=:fx(y)(P_{x}(f))_{y}=f_{y\diamond x}=:f_{x}(y), for yy an object of 𝒳\mathcal{X}.

The following lemma is a generalization of Lemma 8.2.4. Its proof is straightforward.

Lemma 3.1.

Let xx, yy and xx be objects in 𝒳\mathcal{X}.

  1. 1.

    The functors PyPxP_{y}\circ P_{x} and PyxP_{y\diamond x} are naturally isomorphic.

  2. 2.

    Given an arrow φ\varphi in 𝒳(x,y)\mathcal{X}(x,\,y), it induces a natural transformation Pφ:PxPy\mathrm{P}_{\varphi}:\mathrm{P}_{x}\rightarrow\mathrm{P}_{y}. In a functor FF, the arrow Pφ,F:FxFy\mathrm{P}_{\varphi,\,F}:F_{x}\rightarrow F_{y}, denoted by FφF_{\varphi}, is defined at an object ww in 𝒳\mathcal{X} as Fφ,w=F(wφ)F_{\varphi,w}=F(w\diamond\varphi).

  3. 3.

    If ϕ𝒳(x,y)\phi\in\mathcal{X}(x,\,y) and ψ𝒳(y,z)\psi\in\mathcal{X}(y,\,z), then PψPϕ=PψϕP_{\psi}\circ P_{\phi}=P_{\psi\circ\phi}.

  4. 4.

    The correspondence P_P_{\,\_\,} sending an object xx of 𝒳\mathcal{X} to PxP_{x} and an arrow φ\varphi in 𝒳\mathcal{X} to PφP_{\varphi} is an RR-linear functor from 𝒳\mathcal{X} to the category FunR(,)Fun_{R}(\mathcal{F},\,\mathcal{F}) of RR-linear endofunctors of \mathcal{F}.

By point 4 of the previous lemma, for every object FF in \mathcal{F}, we have an RR-linear functor F_:𝒳F_{\,\_\,}:\mathcal{X}\rightarrow\mathcal{F}.

Remark 3.2.

For the biset category, the Yoneda-Dress construction PGP_{G} at a group GG, satisfies also that PGP_{G} is a self-adjoint functor. We will not need this extra condition in what follows.

Definition 3.3.

Let MM and NN be objects in \mathcal{F}, we denote by (M,N)\mathcal{H}(M,\,N) the object in \mathcal{F} defined by the following composition of functors

(M,N)=Hom(M,N_).\mathcal{H}(M,\,N)=\textrm{Hom}_{\mathcal{F}}(M,\,N_{\,\_\,}).

The correspondence (_,_):op×\mathcal{H}(\,\_\,,\,\_\,):\mathcal{F}^{op}\times\mathcal{F}\rightarrow\mathcal{F} is a bilinear functor, called the internal hom functor.

Lemma 3.4.

We have a Yoneda Lemma for the internal hom functor. If x𝒳x\in\mathcal{X}, then

(𝒳(x,_),F)Fx.\mathcal{H}(\mathcal{X}(x,\,\_\,),\,F)\cong F_{x}.
Proof.

The known isomorphism of RR-modules, Hom(𝒳(x,_),F)F(x)\mathrm{Hom}_{\mathcal{F}}(\mathcal{X}(x,\,\_\,),\,F)\leftrightarrow F(x) defines a natural equivalence between the functor Hom(𝒳(x,_),_)\mathrm{Hom}_{\mathcal{F}}(\mathcal{X}(x,\,\_\,),\,\_\,) and the evaluation functor, evxev_{x}, going from \mathcal{F} to RR-Mod. By precomposition with the functor F_F_{\,\_\,} we obtain an equivalence between (𝒳(x,_),F)\mathcal{H}(\mathcal{X}(x,\,\_\,),\,F) and evxF_ev_{x}\circ F_{\,\_\,} in \mathcal{F}. Finally, by composing with the symmetry ss of 𝒳\mathcal{X}, we obtain an equivalence between evxF_ev_{x}\circ F_{\,\_\,} and FxF_{x}. ∎

The rest of the section is devoted to define a hom functor for modules over monoids in \mathcal{F}. We begin by stating the universal property of tensor products. It is easy to see that the proof of Proposition 8.4.2 in [3] holds in our case, so Remark 8.4.3 in [3] can also be generalized.

Remark 3.5.

Let MM, NN and TT be objects in \mathcal{F}. Consider the functors (M,N)(M,\,N) and TT from 𝒳×𝒳\mathcal{X}\times\mathcal{X} to RR-Mod given by sending (x,y)(x,\,y) to M(x)RN(y)M(x)\otimes_{R}N(y) and T(xy)T(x\diamond y) respectively, and in an obvious way in arrows. Then, there is a one-to-one correspondence between Hom(MN,T){}_{\mathcal{F}}(M\otimes N,\,T) and Hom((M,N),T)\mathrm{Hom}_{\mathcal{F}}((M,\,N),T) which is natural in every variable.

In other words, a natural transformation from MNM\otimes N to TT is given by a collection of RR-bilinear maps

M(x)×N(y)T(xy)M(x)\times N(y)\rightarrow T(x\diamond y)

satisfying obvious functoriality conditions.

Notice that this remark can be generalized to an arrow from a finite product i=1nMi\bigotimes_{i=1}^{n}M_{i} to TT.

Let AA be a monoid in \mathcal{F}. The Yoneda Lemma,

Hom(𝒳(𝟏,_),A)A(𝟏)\mathrm{Hom}_{\mathcal{F}}(\mathcal{X}(\mathbf{1},\,\_\,),\,A)\cong A(\mathbf{1})

together with the previous remark allows us to see the monoid AA as in Definition 8.5.1 of [3].

Let xx, yy and zz be objects in 𝒳\mathcal{X}, consider αx,y,z:x(yz)(xy)z\alpha_{x,\,y,\,z}:x\diamond(y\diamond z)\rightarrow(x\diamond y)\diamond z, λx:𝟏xx\lambda_{x}:\mathbf{1}\diamond x\rightarrow x and ρx:x𝟏x\rho_{x}:x\diamond\mathbf{1}\rightarrow x. Then, AA is a monoid in \mathcal{F} if for every pair x,yx,\,y of objects in 𝒳\mathcal{X}, there exist bilinear maps

A(x)×A(y)A(xy),A(x)\times A(y)\rightarrow A(x\diamond y),

denoted simply by (a,b)a×b(a,\,b)\mapsto a\times b, and an element εAA(𝟏)\varepsilon_{A}\in A(\mathbf{1}) satisfying the following conditions.

  • Associativity. For xx, yy and zz, objects in 𝒳\mathcal{X}, and for any aA(x)a\in A(x), bA(y)b\in A(y) and cA(z)c\in A(z),

    (a×b)×c=A(αx,y,z)(a×(b×c)).(a\times b)\times c=A(\alpha_{x,\,y,\,z})(a\times(b\times c)).
  • Identity element. For an object xx in 𝒳\mathcal{X} and any aA(x)a\in A(x),

    a=A(λx)(εA×a)=A(ρx)(a×εA).a=A(\lambda_{x})(\varepsilon_{A}\times a)=A(\rho_{x})(a\times\varepsilon_{A}).
  • Functoriality. Let φ:xx\varphi:x\rightarrow x^{\prime} and ψ:yy\psi:y\rightarrow y^{\prime} be arrows in 𝒳\mathcal{X}. Then for any aA(x)a\in A(x) and bA(y)b\in A(y),

    A(φψ)(a×b)=A(φ)(a)×A(ψ)(b).A(\varphi\diamond\psi)(a\times b)=A(\varphi)(a)\times A(\psi)(b).

Also, a module over AA can be seen as an in Definition 8.5.5 of [3]. That is, a left AA-module is an object MM in \mathcal{F} such that for every pair x,yx,\,y of objects in 𝒳\mathcal{X}, there exist bilinear maps

A(x)×M(y)M(xy),A(x)\times M(y)\rightarrow M(x\diamond y),

which we continue to denote by (a,m)a×m(a,\,m)\mapsto a\times m, satisfying corresponding conditions of associativity, identity element (on the left only) and functoriality. In a similar fashion we can define right modules and bimodules.

Morphisms of AA-modules can also be rewritten in these terms. Let MM and NN be AA-modules, a morphism of AA-modules from MM to NN is an arrow f:MNf:M\rightarrow N in \mathcal{F} such that for any objects xx and yy in 𝒳\mathcal{X}, we have fxy(a×m)=a×fy(m)f_{x\diamond y}(a\times m)=a\times f_{y}(m) for all aA(x)a\in A(x) and mM(y)m\in M(y).

The proof of the following lemma is straightforward.

Lemma 3.6.

Let AA be a monoid in \mathcal{F} and MM be an AA-module, then we have an isomorphism of RR-modules,

HomAMod(A,M)M(𝟏),\mathrm{Hom}_{A-\mathrm{Mod}}(A,\,M)\cong M(\mathbf{1}),

given by sending a morphism of AA-modules, ff, to f𝟏(εA)f_{\mathbf{1}}(\varepsilon_{A}).

Lemma 3.7.

Let xx and yy be objects in 𝒳\mathcal{X} and α:xy\alpha:x\rightarrow y be an arrow. Consider Px\mathrm{P}_{x} and Pα\mathrm{P}_{\alpha} as defined at the beginning of the section. Then P_\mathrm{P}_{\,\_\,} defines a functor from 𝒳\mathcal{X} to the category of RR-linear endofunctors of AA-Mod\mathrm{Mod}.

Proof.

Let xx and φ\varphi be an object and an arrow in 𝒳\mathcal{X}. By Lemma 3.1, we only need to verify that Px\mathrm{P}_{x} restricts from AA-Mod to AA-Mod and that the components of the transformation Pφ:PxPy\mathrm{P}_{\varphi}:\mathrm{P}_{x}\rightarrow\mathrm{P}_{y}, as defined before, are morphisms of AA-modules.

Let MM be an AA-module, then

A(y)×Mx(z)Mx(yz),A(y)\times M_{x}(z)\rightarrow M_{x}(y\diamond z),

given by the original action of AA on MM, endows MxM_{x} with a structure of AA-module. Also, if f:MNf:M\rightarrow N is a morphism of AA-modules, then it is immediate to see that Px(f)P_{x}(f), is a morphism of AA-modules. Hence we can restrict Px:AP_{x}:A-ModA\textrm{Mod}\rightarrow A-Mod. Now let MM be an AA-module, the functoriality of the action of AA on MM shows that Pα,M:MxMy\textrm{P}_{\alpha,\,M}:M_{x}\rightarrow M_{y} is a morphism of AA-modules. ∎

Remark 3.8.

As said in Section 2, NN is a right AA-module if and only if it is an AopA^{op}-module. The two structures are related by the following equation

a×n=N(szyyz)(n×a),a\times n=N(s_{z\diamond y}^{y\diamond z})(n\times a),

for aA(y)a\in A(y), nN(z)n\in N(z) and where szyyz:zyyzs_{z\diamond y}^{y\diamond z}:z\diamond y\rightarrow y\diamond z is the symmetry in 𝒳\mathcal{X}. Now, if we begin with a right AA-module, MM, then it is an AopA^{op}-module and then N=MxN=M_{x} is an AopA^{op}-module, as in the previous lemma. After some straightforward computations we see that its structure of right AA-module is given by

Mx(z)×A(y)Mx(zy),(m,a)M(szxyzyx)(m×a).M_{x}(z)\times A(y)\rightarrow M_{x}(z\diamond y),\,(m,\,a)\mapsto M(s_{z\diamond x\diamond y}^{z\diamond y\diamond x})(m\times a).

A word of caution is needed, since, in general we will write m×am\times a for the action of AA in any right AA-module NN regardless of whether it is shifted or not. However, if NN is explicitly a shifted module MxM_{x}, we will write the corresponding symmetry.

Given two AA-modules MM and NN, we have that HomAMod(M,N)\textrm{Hom}_{A-\mathrm{Mod}}(M,\,N) is an RR-submodule of Hom(M,N)\textrm{Hom}_{\mathcal{F}}(M,\,N). Moreover, for an AA-module MM, we have a functor HomAMod(M,_):A\textrm{Hom}_{A-\mathrm{Mod}}(M,\,\_):A-ModR\mathrm{Mod}\rightarrow R-Mod\mathrm{Mod}.

Definition 3.9.

Let MM and NN be objects in AA-Mod, we denote by A(M,N)\mathcal{H}_{A}(M,\,N) the object in \mathcal{F} defined by

A(M,N)=HomAMod(M,N_).\mathcal{H}_{A}(M,\,N)=\textrm{Hom}_{A-\mathrm{Mod}}(M,\,N_{\,\_\,}).
Lemma 3.10.

The correspondence A(_,_):(A\mathcal{H}_{A}(\,\_\,,\,\_\,):(A-Mod)op×A\mathrm{Mod})^{op}\times A-Mod\mathrm{Mod}\rightarrow\mathcal{F} is a bilinear functor.

Proof.

This follows from Lemma 3.7. ∎

Corollary 3.11.

For AA and MM as before, we have an isomorphism in \mathcal{F},

A(A,M)M.\mathcal{H}_{A}(A,\,M)\cong M.
Proof.

The bijection of Lemma 3.6 defines a natural equivalence between the functor HomAMod(A,_)\mathrm{Hom}_{A-\mathrm{Mod}}(A,\,\_\,) and the evaluation functor, ev𝟏Aev^{A}_{\mathbf{1}}, going from AA-Mod to RR-Mod. By precomposing with M_M_{\,\_\,} we obtain an equivalence between A(A,M)\mathcal{H}_{A}(A,\,M) and ev𝟏AM_ev^{A}_{\mathbf{1}}\circ M_{\,\_\,}, but clearly the last one is equivalent to MM.

4 The Hochschild cochain complex

4.1 Bar resolution of a monoid

Let AA be a monoid in \mathcal{F}. We omit the morphism α\alpha of diagram 1 and write AnA^{\otimes n} for the tensor product of AA with itself nn times.

Consider the following sequence in \mathcal{F},

Cbar(A):An+1bAnbbA2C_{\ast}^{bar}(A):\lx@xy@svg{\hbox{\raise 0.0pt\hbox{\kern 6.75pt\hbox{\ignorespaces\ignorespaces\ignorespaces\hbox{\vtop{\kern 0.0pt\offinterlineskip\halign{\entry@#!@&&\entry@@#!@\cr&&&&\crcr}}}\ignorespaces{\hbox{\kern-6.75pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{\ldots\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces{\hbox{\kern 30.75pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 30.75pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A^{\otimes n+1}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 64.8315pt\raise 5.43056pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.43056pt\hbox{$\scriptstyle{b}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 83.12247pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 83.12247pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A^{\otimes n}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 109.90805pt\raise 5.43056pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.43056pt\hbox{$\scriptstyle{b}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 128.33936pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 128.33936pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{\ldots\ignorespaces\ignorespaces\ignorespaces\ignorespaces}$}}}}}}}\ignorespaces\ignorespaces\ignorespaces\ignorespaces{}{\hbox{\lx@xy@droprule}}\ignorespaces\ignorespaces\ignorespaces{\hbox{\kern 151.12617pt\raise 5.43056pt\hbox{{}\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\hbox{\hbox{\kern 0.0pt\raise-2.43056pt\hbox{$\scriptstyle{b}$}}}\kern 3.0pt}}}}}}\ignorespaces{\hbox{\kern 165.83936pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\lx@xy@tip{1}\lx@xy@tip{-1}}}}}}{\hbox{\lx@xy@droprule}}{\hbox{\lx@xy@droprule}}{\hbox{\kern 165.83936pt\raise 0.0pt\hbox{\hbox{\kern 0.0pt\raise 0.0pt\hbox{\hbox{\kern 3.0pt\raise 0.0pt\hbox{$\textstyle{A^{\otimes 2}}$}}}}}}}\ignorespaces}}}}\ignorespaces

where A2A^{\otimes 2} is in degree zero and b:An+1Anb:A^{\otimes n+1}\rightarrow A^{\otimes n}, for n2n\geq 2, is defined in the following way. Given n2n\geq 2, define di:An+1And_{i}:A^{\otimes n+1}\rightarrow A^{\otimes n}, for 1in1\leq i\leq n, as

di=AAμAA,d_{i}=A\otimes\cdots\otimes A\otimes\mu\otimes A\otimes\cdots\otimes A,

where μ\mu appears in the position ii of the tensor product. That is, d1=μAn1d_{1}=\mu\otimes A^{\otimes n-1}, di=Ai1μAnid_{i}=A^{\otimes i-1}\otimes\mu\otimes A^{n-i} if 1<i<n1<i<n and dn=An1μd_{n}=A^{\otimes n-1}\otimes\mu. With this, b:=i=1n(1)i+1dib:=\sum_{i=1}^{n}(-1)^{i+1}d_{i}. When considering the augmented complex, having AA in degree -1, with morphism b=μ:A2Ab=\mu:A^{\otimes 2}\rightarrow A and 0 in all degrees j<1j<-1, we will write Cbar(A)μC_{\ast}^{bar}(A)_{\mu}.

Lemma 4.1.

The sequence Cbar(A)C_{\ast}^{bar}(A) is a presimplicial object in \mathcal{F}, that is, the arrows did_{i} defined above satisfy

didj=dj1di1i<jn.d_{i}\circ d_{j}=d_{j-1}\circ d_{i}\quad 1\leq i<j\leq n.

Moreover, Cbar(A)μC_{\ast}^{bar}(A)_{\mu} is a complex in \mathcal{F}.

Proof.

Notice that if i+1<ji+1<j, then we clearly have didj=dj1did_{i}\circ d_{j}=d_{j-1}\circ d_{i}. On the other hand, the commutativity of diagram 1 implies the previous equality if j=i+1j=i+1. It also implies that μb=0\mu\circ b=0. With this, the proof of bb=0b\circ b=0 is standard (see for example Lemma 1.0.7 of [8]). ∎

The structure of AA-bimodule of AA is given by diagram 1. Now consider AnA^{\otimes n} with n2n\geq 2. The structure of AA-bimodule of AnA^{\otimes n} is given by dnd1:An+2And_{n}\circ d_{1}:A^{\otimes n+2}\rightarrow A^{\otimes n}, which is equal to d1dn+1d_{1}\circ d_{n+1}.

Proposition 4.2.

Let AA be a monoid in \mathcal{F}, the complex Cbar(A)C_{\ast}^{bar}(A) is a resolution of the AeA^{e}-module AA. It is called the bar resolution of AA.

Proof.

We will first show that the complex Cbar(A)μC_{\ast}^{bar}(A)_{\mu} is exact and then we will verify that it is a complex of AeA^{e}-modules.

Using diagram 1, one can show that μ\mu is an epimorphism in \mathcal{F} but we prefer to define the operators of extra degeneracy and show that the complex is acyclic. Consider first q:AAAq:A\rightarrow A\otimes A as the morphism

A\textstyle{A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}λ1\scriptstyle{\lambda^{-1}}IA\textstyle{I\otimes A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}eA\scriptstyle{e\otimes A}AA\textstyle{A\otimes A}

from diagram 2. Next, for n2n\geq 2, define

q:AnAn+1,q=((eA)λ1)An1.q:A^{\otimes n}\rightarrow A^{\otimes n+1},\quad q=((e\otimes A)\circ\lambda^{-1})\otimes A^{\otimes n-1}.

It is easy to see that if 2in2\leq i\leq n, then diq=qdi1d_{i}\circ q=q\circ d_{i-1}. On the other hand, diagram 2 shows that for any n1n\geq 1 we have d1q=And_{1}\circ q=A^{\otimes n}. This implies bq+qb=Anb\circ q+q\circ b=A^{\otimes n} for n2n\geq 2 and μq=A\mu\circ q=A. Hence qq is a contracting homotopy and Cbar(A)μC_{\ast}^{bar}(A)_{\mu} is acyclic.

Now we see that Cbar(A)μC_{\ast}^{bar}(A)_{\mu} is a complex of AeA^{e}-modules. The arrow μ\mu is a morphism of AeA^{e}-modules thanks to the associativity of the product, i.e. diagram 1. Next, let n2n\geq 2 and di:An+1And_{i}:A^{\otimes n+1}\rightarrow A^{\otimes n} with 1in1\leq i\leq n. To see that the action of AeA^{e} commutes with bb, we verify it commutes with the arrows did_{i}, that is, the commutativity of the following diagram

An+3\textstyle{A^{\otimes n+3}\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}dn+1d1\scriptstyle{d_{n+1}\circ\,d_{1}}Adi+1A\scriptstyle{A\otimes d_{i+1}\otimes A}An+1\textstyle{A^{\otimes n+1}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}di\scriptstyle{d_{i}}An+2\textstyle{A^{\otimes n+2}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}dnd1\scriptstyle{d_{n}\circ\,d_{1}}An,\textstyle{A^{\otimes n},}

where the horizontal arrows correspond to the structures of AA-bimodules of An+1A^{\otimes n+1} and AnA^{\otimes n}. Notice that Adi+1A=di+1A\otimes d_{i+1}\otimes A=d_{i+1}. By Lemma 4.1 we have

didn+1d1=dnd1di+1.d_{i}\circ d_{n+1}\circ d_{1}=d_{n}\circ d_{1}\circ d_{i+1}.

Hence, Cbar(A)μC_{\ast}^{bar}(A)_{\mu} is a complex of AeA^{e}-modules.

4.2 Hochschild cochain complex

Definition 4.3.

Let AA be a monoid in \mathcal{F} and MM be an AA-bimodule. The Hochschild cochain complex is the complex in \mathcal{F} given by Ae(Cbar(A),M)\mathcal{H}_{A^{e}}(C^{bar}_{\ast}(A),\,M). Since \mathcal{F} is an abelian category, we define the Hochschild cohomology of AA with coefficients in MM as the object in \mathcal{F} given by

Hn(A,M):=Hn(Ae(Cbar(A),M))\mathcal{H}H^{n}(A,\,M):=H^{n}(\mathcal{H}_{A^{e}}(C^{bar}_{\ast}(A),\,M))

It is easy to see that Hn(A,_)\mathcal{H}H^{n}(A,\,\_\,) defines a functor from AeA^{e}-Mod to \mathcal{F}.

The next definition generalizes Definition 18 in [4].

Definition 4.4.

Let AA be a monoid in \mathcal{F} and MM be an AA-bimodule. The commutant of MM at x𝒳x\in\mathcal{X} is

CMA(x)={mM(x)a×m=M(sxyyx)(m×a),aA(y),y𝒳},CM_{A}(x)=\{m\in M(x)\mid a\times m=M(s_{x\diamond y}^{y\diamond x})(m\times a),\forall a\in A(y),\forall y\in\mathcal{X}\},

where sxyyx:xyyxs_{x\diamond y}^{y\diamond x}:x\diamond y\rightarrow y\diamond x is the symmetry in 𝒳\mathcal{X}. If the AA-bimodule is AA itself, we keep the notation CACA of [4].

Since syxxys_{y\diamond x}^{x\diamond y} is a natural isomorphism, it is easy to see that CMACM_{A} is a subfunctor of MM. Also, we will see in the last section that the functors Hn(A,M)\mathcal{H}H^{n}(A,\,M) are all CACA-modules.

To describe the Hochschild cochain complex we need the following lemma.

Lemma 4.5.

Let AA be a monoid in \mathcal{F}, MM be an AA-bimodule and n0n\geq 0 be an integer. We have the following isomorphism in \mathcal{F},

Ae(An+2,M)(An,M),\mathcal{H}_{A^{e}}(A^{\otimes n+2},\,M)\cong\mathcal{H}(A^{\otimes n},\,M),

where, if n=0n=0, then A0A^{\otimes 0} denotes 𝒳(𝟏,_)\mathcal{X}(\mathbf{1},\,\_\,). Also, Ae(A,M)CMA\mathcal{H}_{A^{e}}(A,\,M)\cong CM_{A} in \mathcal{F}.

Proof.

For the first statement, consider n=0n=0. Using Remark 2.1, it is easy to see that A2A^{\otimes 2} and AeA^{e} are isomorphic as AeA^{e}-modules. Then, by lemmas 3.4 and 3.11, we have the following isomorphisms in \mathcal{F},

Ae(A2,M)M(𝒳(𝟏,_),M).\mathcal{H}_{A^{e}}(A^{\otimes 2},\,M)\cong M\cong\mathcal{H}(\mathcal{X}(\mathbf{1},\,\_\,),\,M).

Hence Ae(A2,M)(A0,M)\mathcal{H}_{A^{e}}(A^{\otimes 2},\,M)\cong\mathcal{H}(A^{\otimes 0},\,M).

Now suppose n1n\geq 1 and let NN be an AA-bimodule. Then, by Remark 3.5, an arrow f:An+2Nf:A^{\otimes n+2}\rightarrow N, of AeA^{e}-modules, is given by a corresponding collection of maps

f~w,y,z:A(w)×An(y)×A(z)N(wyz)\tilde{f}_{w,\,y,\,z}:A(w)\times A^{\otimes n}(y)\times A(z)\rightarrow N(w\diamond y\diamond z)

linear and functorial in each entry. These maps also satisfy the following identity

f~uw,y,zv(α×a,c,b×β)=α×f~w,y,z(a,c,b)×β\tilde{f}_{u\diamond w,\,y,\,z\diamond v}(\alpha\times a,\,c,\,b\times\beta)=\alpha\times\tilde{f}_{w,\,y,\,z}(a,\,c,\,b)\times\beta

for all objects uu, vv, ww, yy and zz and elements aa, bb, cc, α\alpha and β\beta in the corresponding evaluations. In particular

f~w,y,z(a,c,b)=a×f~𝟏,y, 1(εA,c,εA)×b,\tilde{f}_{w,\,y,\,z}(a,\,c,\,b)=a\times\tilde{f}_{\mathbf{1},\,y,\,\mathbf{1}}(\varepsilon_{A},\,c,\,\varepsilon_{A})\times b,

and f~𝟏,y, 1\tilde{f}_{\mathbf{1},\,y,\,\mathbf{1}} is functorial in yy and linear in the variable cc. Hence, if we consider the functors HomAeMod(An+2,_)\mathrm{Hom}_{A^{e}-\mathrm{Mod}}(A^{\otimes n+2},\,\_\,) and Hom(An,_)\mathrm{Hom}_{\mathcal{F}}(A^{\otimes n},\,\_\,) from AeA^{e}-Mod to RR-Mod, we obtain a natural isomorphism between them. Indeed, by sending ff to f~𝟏,_,𝟏(εA,_,εA)\tilde{f}_{\mathbf{1},\,\_\,,\mathbf{1}}(\varepsilon_{A},\,\_,\,\varepsilon_{A}) we define an isomorphism of RR-modules from HomAeMod(An+2,N)\mathrm{Hom}_{A^{e}-\mathrm{Mod}}(A^{\otimes n+2},\,N) to Hom(An,N)\mathrm{Hom}_{\mathcal{F}}(A^{\otimes n},\,N), which is clearly natural in NN. By precomposition with the functor M_:𝒳AeM_{\,\_\,}:\mathcal{X}\rightarrow A^{e}-Mod, we obtain the result.

The second statement follows as in the classical case. That is, since μ:AAA\mu:A\otimes A\rightarrow A is an epimorphism, we have a monomorphism in \mathcal{F},

Ae(μ,M):Ae(A,M)Ae(A2,M)M\mathcal{H}_{A^{e}}(\mu,\,M):\mathcal{H}_{A^{e}}(A,\,M)\rightarrow\mathcal{H}_{A^{e}}(A^{\otimes 2},\,M)\cong M

defined in x𝒳x\in\mathcal{X} by sending fHomAeMod(A,Mx)f\in\mathrm{Hom}_{A^{e}-\mathrm{Mod}}(A,\,M_{x}) to (fμ)𝟏(εAe)(f\circ\mu)_{\mathbf{1}}(\varepsilon_{A^{e}}). Given that μ\mu is a morphism of monoids and that εAe=εAA\varepsilon_{A^{e}}=\varepsilon_{A\otimes A}, we have μ𝟏(εAe)=εA\mu_{\mathbf{1}}(\varepsilon_{A^{e}})=\varepsilon_{A}, hence (fμ)𝟏(εAe)=f𝟏(εA)(f\circ\mu)_{\mathbf{1}}(\varepsilon_{A^{e}})=f_{\mathbf{1}}(\varepsilon_{A}). But, since ff is a morphism of AA-bimodules, it is easy to see that f𝟏(εA)f_{\mathbf{1}}(\varepsilon_{A}) must be in CMA(x)CM_{A}(x). ∎

With this, the complex Ae(Cbar(A)μ,M)\mathcal{H}_{A^{e}}(C^{bar}_{\ast}(A)_{\mu},\,M) now looks like

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}CMA\textstyle{CM_{A}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}M\textstyle{M\ignorespaces\ignorespaces\ignorespaces\ignorespaces}β\scriptstyle{\beta}(A,M)\textstyle{\mathcal{H}(A,\,M)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}β\scriptstyle{\beta}(A2,M)\textstyle{\mathcal{H}(A^{\otimes 2},\,M)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}β\scriptstyle{\beta},\textstyle{\ldots,}

with MM in degree zero. The second arrow from the left is just the inclusion of CMACM_{A} in MM and the arrows β\beta are described as follows.

Let xx be an object in 𝒳\mathcal{X}. In the complex HomAeMod(Cbar(A),Mx)\mathrm{Hom}_{A^{e}-\mathrm{Mod}}(C^{bar}_{\ast}(A),\,M_{x}), an arrow φ\varphi in HomAeMod(An+2,Mx)\mathrm{Hom}_{A^{e}-\mathrm{Mod}}(A^{\otimes n+2},\,M_{x}), with n0n\geq 0, is sent to φb\varphi\circ b in HomAeMod(An+3,Mx)\mathrm{Hom}_{A^{e}-\mathrm{Mod}}(A^{\otimes n+3},\,M_{x}). We will see what this means in terms of the corresponding linear maps of Remark 3.5.

Suppose first n1n\geq 1. Denote by φ~\tilde{\varphi} the n+2n+2-linear map that corresponds to φ\varphi by Remark 3.5. As in the previous lemma, the corresponding nn-linear map

φ~𝟏,x12,,xnn+1,𝟏:A(x1)××A(xn)Mx(x1xn)\tilde{\varphi}_{\mathbf{1},x_{1}^{2},\ldots,x_{n}^{n+1},\mathbf{1}}:A(x_{1})\times\cdots\times A(x_{n})\longrightarrow M_{x}(x_{1}\diamond\cdots\diamond x_{n})

is given by φ~𝟏,x12,,xnn+1,𝟏(εA,_,,_,εA)\tilde{\varphi}_{\mathbf{1},x_{1}^{2},\ldots,x_{n}^{n+1},\mathbf{1}}(\varepsilon_{A},\,\_\,,\ldots,\,\_\,,\varepsilon_{A}). In the variables xjx_{j} we add the superscript ii to denote the position of xjx_{j} in φ~\tilde{\varphi}. If di:An+3An+2d_{i}:A^{\otimes n+3}\rightarrow A^{\otimes n+2}, with 2in+12\leq i\leq n+1, is one of the morphisms defined in Section 4.1, then it is not hard to see that

(φdi)~𝟏,x12,,xn+1n+2,𝟏:A(x1)××A(xn+1)Mx(x1xn+1)\widetilde{(\varphi\circ d_{i})}_{\mathbf{1},x_{1}^{2},\ldots,x_{n+1}^{n+2},\mathbf{1}}:A(x_{1})\times\cdots\times A(x_{n+1})\longrightarrow M_{x}(x_{1}\diamond\cdots\diamond x_{n+1})

sends an element (a1,,an+1)(a_{1},\ldots,a_{n+1}) to

φ~𝟏,x12,,xi1ixii+1,,xn+1n+1,𝟏(εA,a1,,ai1×ai,,an+1,εA).\tilde{\varphi}_{\mathbf{1},x_{1}^{2},\ldots,x_{i-1}^{i}\diamond x_{i}^{i+1},\ldots,x_{n+1}^{n+1},\mathbf{1}}(\varepsilon_{A},a_{1},\ldots,a_{i-1}\times a_{i},\ldots,a_{n+1},\varepsilon_{A}).

Also,

(φd1~)𝟏,x12,,xn+1n+2,𝟏(a1,,an+1)=φ~x11,,xn+1n+1,𝟏(a1,,an+1,εA)\widetilde{(\varphi\circ d_{1}})_{\mathbf{1},x_{1}^{2},\ldots,x_{n+1}^{n+2},\mathbf{1}}(a_{1},\ldots,a_{n+1})=\tilde{\varphi}_{x_{1}^{1},\ldots,x_{n+1}^{n+1},\mathbf{1}}(a_{1},\ldots,a_{n+1},\,\varepsilon_{A})

and

(φdn+2~)𝟏,x12,,xn+1n+2,𝟏(a1,,an+1)=φ~𝟏,x12,,xn+1n+2(εA,a1,,an+1).\widetilde{(\varphi\circ d_{n+2}})_{\mathbf{1},x_{1}^{2},\ldots,x_{n+1}^{n+2},\mathbf{1}}(a_{1},\ldots,a_{n+1})=\tilde{\varphi}_{\mathbf{1},x_{1}^{2},\ldots,x_{n+1}^{n+2}}(\varepsilon_{A},\,a_{1},\ldots,a_{n+1}).

Thus, if we let fHom(An,Mx)f\in\mathrm{Hom}_{\mathcal{F}}(A^{\otimes n},\,M_{x}) to be the arrow corresponding to φ\varphi and βx(f)\beta_{x}(f) be the arrow in Hom(An+1,Mx)\mathrm{Hom}_{\mathcal{F}}(A^{\otimes n+1},\,M_{x}) corresponding to φb\varphi\circ b, then

βx(f)~x1,,xn+1:A(x1)××A(xn+1)Mx(x1xn+1)\widetilde{\beta_{x}(f)}_{x_{1},\ldots,x_{n+1}}:A(x_{1})\times\cdots\times A(x_{n+1})\longrightarrow M_{x}(x_{1}\diamond\cdots\diamond x_{n+1})

sends (a1,,an+1)(a_{1},\ldots,a_{n+1}) to

a1×f~x2,,xn+1(a2,,an+1)+i=1n(1)if~x1,,xixi+1,,xn+1(a1,,ai×ai+1,,an+1)+(1)n+1M(sx1xnxxn+1x1xn+1x)(f~x1,,xn(a1,,an)×an+1),\begin{array}[]{l}a_{1}\times\tilde{f}_{x_{2},\ldots,x_{n+1}}(a_{2},\ldots,a_{n+1})\\ +\sum_{i=1}^{n}(-1)^{i}\tilde{f}_{x_{1},\ldots,x_{i}\diamond x_{i+1},\ldots,x_{n+1}}(a_{1},\ldots,a_{i}\times a_{i+1},\ldots,a_{n+1})\\ +(-1)^{n+1}M(s_{x_{1}\diamond\ldots\diamond x_{n}\diamond x\diamond x_{n+1}}^{x_{1}\diamond\ldots\diamond x_{n+1}\diamond x})(\tilde{f}_{x_{1},\ldots,x_{n}}(a_{1},\ldots,a_{n})\times a_{n+1}),\end{array}

where the last symmetry comes from Remark 3.8.

For the case n=0n=0, we see immediately that βx:M(x)Hom(A,Mx)\beta_{x}:M(x)\rightarrow\mathrm{Hom}_{\mathcal{F}}(A,\,M_{x}) is given by

βx(m)y:A(y)M(yx),aa×mM(sxyyx)(m×a).\beta_{x}(m)_{y}:A(y)\rightarrow M(y\diamond x),\quad a\mapsto a\times m-M(s_{x\diamond y}^{y\diamond x})(m\times a).

5 Relative projectivity and separable monoids

Given a monoid AA in \mathcal{F}, we consider the functors

A:A-ModandA:A-Mod,\mathcal{R}_{A}:A\textrm{-Mod}\rightarrow\mathcal{F}\quad\textnormal{and}\quad\mathcal{I}_{A}:\mathcal{F}\rightarrow A\textrm{-Mod},

the first one is just the forgetful functor and the second one sends a functor FF to AFA\otimes F and an arrow ff to AfA\otimes f. It is easy to see that A\mathcal{I}_{A} is a left adjoint of A\mathcal{R}_{A}.

Following Definition 4.1 in [2] we say that an AA-module MM is projective with respect to the restriction A\mathcal{R}_{A} if for any diagram

M\textstyle{M\ignorespaces\ignorespaces\ignorespaces\ignorespaces}β\scriptstyle{\beta}X\textstyle{X\ignorespaces\ignorespaces\ignorespaces\ignorespaces}α\scriptstyle{\alpha}Y\textstyle{Y}

in AA-Mod such that A(α)\mathcal{R}_{A}(\alpha) is a split epimorphism, there exists a morphism φ:MX\varphi:M\rightarrow X in AA-Mod such that αφ=β\alpha\varphi=\beta.

Now consider Ae\mathcal{R}_{A^{e}} and Ae\mathcal{I}_{A^{e}}. Lemma 4.6 in [2] translates in the following way.

Lemma 5.1.

For every object MM in AeA^{e}-Mod there exists a resolution

\textstyle{\cdots\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Li\textstyle{L_{i}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Li1\textstyle{L_{i-1}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}\textstyle{\cdots\ignorespaces\ignorespaces\ignorespaces\ignorespaces}L0\textstyle{L_{0}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}M\textstyle{M\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0\textstyle{0}

where the LiL_{i} are projective with respect to the restriction Ae\mathcal{R}_{A^{e}} and such that

\textstyle{\cdots\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(Li)\textstyle{\mathcal{R}_{A^{e}}(L_{i})\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(Li1)\textstyle{\mathcal{R}_{A^{e}}(L_{i-1})\ignorespaces\ignorespaces\ignorespaces\ignorespaces}\textstyle{\cdots\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(L0)\textstyle{\mathcal{R}_{A^{e}}(L_{0})\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(M)\textstyle{\mathcal{R}_{A^{e}}(M)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0\textstyle{0}

is an exact split complex. Moreover, such a resolution for MM is unique up to homotopy.

Remark 5.2.

By point 3 of Lemma 4.3 in [2], every AiA^{\otimes i} with i2i\geq 2, is projective with respect to A\mathcal{R}_{A}. Also, by the proof of Proposition 4.2, the restriction of the bar resolution, Ae(Cbar(A)μ)\mathcal{R}_{A^{e}}(C^{bar}_{\ast}(A)_{\mu}), is an exact split complex. Hence, the bar resolution is a resolution for AA as in the previous lemma.

This remark will also help us to prove the following result.

Lemma 5.3.

Let EE be a short exact sequence in AeA^{e}-Mod,

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}K\textstyle{K\ignorespaces\ignorespaces\ignorespaces\ignorespaces}j\scriptstyle{j}M\textstyle{M\ignorespaces\ignorespaces\ignorespaces\ignorespaces}z\scriptstyle{z}N\textstyle{N\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0\textstyle{0}

such that Ae(z)\mathcal{R}_{A^{e}}(z) is a split epimorphism. Then we have a short exact sequence of complexes in \mathcal{F},

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(Cbar(A),K)\textstyle{\mathcal{H}_{A^{e}}(C^{bar}_{\ast}(A),\,K)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(Cbar(A),M)\textstyle{\mathcal{H}_{A^{e}}(C^{bar}_{\ast}(A),\,M)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(Cbar(A),N)\textstyle{\mathcal{H}_{A^{e}}(C^{bar}_{\ast}(A),\,N)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0.\textstyle{0.}
Proof.

Observe first that, for an object xx in 𝒳\mathcal{X}, the shifted sequence ExE_{x},

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Kx\textstyle{K_{x}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}jx\scriptstyle{j_{x}}Mx\textstyle{M_{x}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}zx\scriptstyle{z_{x}}Nx\textstyle{N_{x}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0\textstyle{0}

is also a short exact sequence in AeA^{e}-Mod. Also (Ae(L))x=Ae(Lx)(\mathcal{R}_{A^{e}}(L))_{x}=\mathcal{R}_{A^{e}}(L_{x}) and (Ae(l))x=Ae(lx)(\mathcal{R}_{A^{e}}(l))_{x}=\mathcal{R}_{A^{e}}(l_{x}) for any AA-bimodule LL and any morphism of AA-bimodules ll. Finally, if Ae(z)\mathcal{R}_{A^{e}}(z) is a split epimorphism, then Ae(zx)\mathcal{R}_{A^{e}}(z_{x}) is also a split epimorphism.

It is easy to see that we have a sequence of complexes in \mathcal{F},

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(Cbar(A),K)\textstyle{\mathcal{H}_{A^{e}}(C^{bar}_{\ast}(A),\,K)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(Cbar(A),M)\textstyle{\mathcal{H}_{A^{e}}(C^{bar}_{\ast}(A),\,M)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(Cbar(A),N)\textstyle{\mathcal{H}_{A^{e}}(C^{bar}_{\ast}(A),\,N)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0.\textstyle{0.}

Now, Let i2i\geq 2. The hom functor Ae(Ai,_)\mathcal{H}_{A^{e}}(A^{\otimes i},\,\_\,) is clearly left exact, so to show that the sequence

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}HomAe(Ai,Kx)\textstyle{\mathrm{Hom}_{A^{e}}(A^{\otimes i},\,K_{x})\ignorespaces\ignorespaces\ignorespaces\ignorespaces}HomAe(Ai,Mx)\textstyle{\mathrm{Hom}_{A^{e}}(A^{\otimes i},\,M_{x})\ignorespaces\ignorespaces\ignorespaces\ignorespaces}HomAe(Ai,Nx)\textstyle{\mathrm{Hom}_{A^{e}}(A^{\otimes i},\,N_{x})\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0\textstyle{0}

(we abbreviate HomAeMod\mathrm{Hom}_{A^{e}-\textrm{Mod}} as HomAe\mathrm{Hom}_{A^{e}}) is exact, it suffices to show that the morphism HomAe(Ai,zx)\mathrm{Hom}_{A^{e}}(A^{\otimes i},\,z_{x}) from HomAe(Ai,Mx)\mathrm{Hom}_{A^{e}}(A^{\otimes i},\,M_{x}) to HomAe(Ai,Nx)\mathrm{Hom}_{A^{e}}(A^{\otimes i},\,N_{x}) is surjective. Let fHomAe(Ai,Nx)f\in\mathrm{Hom}_{A^{e}}(A^{\otimes i},\,N_{x}). As in the previous remark, every AiA^{\otimes i} with i2i\geq 2 is projective with respect to Ae\mathcal{R}_{A^{e}}, so, by definition, there exists φ:AiMx\varphi:A^{\otimes i}\rightarrow M_{x} in AeA^{e}-Mod such that zxφ=fxz_{x}\circ\varphi=f_{x}. ∎

We have immediately the following corollary.

Corollary 5.4.

If EE is a short exact sequence in AeA^{e}-Mod as in the previous lemma, then we have a long exact sequence

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}H0(A,K)\textstyle{\mathcal{H}H^{0}(A,\,K)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}H0(A,M)\textstyle{\mathcal{H}H^{0}(A,\,M)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}H0(A,N)\textstyle{\mathcal{H}H^{0}(A,\,N)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}H1(A,K)\textstyle{\mathcal{H}H^{1}(A,\,K)\ldots}

in \mathcal{F}.

Since for any AA-bimodule MM, the functor Ae(_,M)\mathcal{H}_{A^{e}}(\,\_\,,\,M) is additive, we can calculate the Hochschild cohomology with coefficients in MM from any resolution that satisfies the conditions of Lemma 5.1, as it is the case in the following example.

Definition 5.5.

Let AA be a monoid in \mathcal{F}. We say that AA is separable if μ:AAA\mu:A\otimes A\rightarrow A is a split epimorphism in AeA^{e}-Mod.

Theorem 5.6.

If AA is a separable monoid in \mathcal{F}, then for any AA-bimodule MM we have Hi(A,M)=0\mathcal{H}H^{i}(A,\,M)=0 for every i>0i>0. Conversely, if AA is a monoid in \mathcal{F} such that H1(A,M)=0\mathcal{H}H^{1}(A,\,M)=0 for any AA-bimodule MM, then AA is separable.

Proof.

If AA is separable, we have a split short exact sequence in AeA^{e}-Mod

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Kerμ\textstyle{Ker\mu\ignorespaces\ignorespaces\ignorespaces\ignorespaces}AA\textstyle{A\otimes A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}μ\scriptstyle{\mu}A\textstyle{A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0.\textstyle{0.}

As in Remark 5.2, AAA\otimes A is projective with respect to the restriction Ae\mathcal{R}_{A^{e}} and by Lemma 4.2 in [2], we have that KerμKer\mu is also projective with respect to Ae\mathcal{R}_{A^{e}}. Hence, this short exact sequence gives a resolution for AA as in Lemma 5.1. Since this resolution is contractible, we have that the bar resolution is contractible and thus, the complex Ae(Cbar(A)μ,M)\mathcal{H}_{A^{e}}(C^{bar}_{\ast}(A)_{\mu},\,M) is contractible.

For the other direction, consider the short exact sequence in AeA^{e}-Mod,

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Kerμ\textstyle{Ker\mu\ignorespaces\ignorespaces\ignorespaces\ignorespaces}AA\textstyle{A\otimes A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}μ\scriptstyle{\mu}A\textstyle{A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0\textstyle{0}

and let I=KerμI=Ker\mu. By the first lines of the proof of Proposition 4.2, we have that Ae(μ)\mathcal{R}_{A^{e}}(\mu) is a split epimorphism. Then, by Corollary 5.4, we have the following exact sequence in \mathcal{F}

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}H0(A,I)\textstyle{\mathcal{H}H^{0}(A,\,I)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}H0(A,AA)\textstyle{\mathcal{H}H^{0}(A,\,A\otimes A)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}H0(A,A)\textstyle{\mathcal{H}H^{0}(A,\,A)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0,\textstyle{0,}

where the last zero corresponds to H1(A,I)=0\mathcal{H}H^{1}(A,\,I)=0. But, by the last lines of the previous section, we have that H0(A,M)CMA\mathcal{H}H^{0}(A,\,M)\cong CM_{A} for any AA-bimodule MM, hence, by Lemma 4.5, we have the following exact sequence in RR-Mod,

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(A,I)(𝟏)\textstyle{\mathcal{H}_{A^{e}}(A,\,I)(\mathbf{1})\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(A,AA)(𝟏)\textstyle{\mathcal{H}_{A^{e}}(A,\,A\otimes A)(\mathbf{1})\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ae(A,A)(𝟏)\textstyle{\mathcal{H}_{A^{e}}(A,\,A)(\mathbf{1})\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0.\textstyle{0.}

So, for the identity morphism, AA in Ae(A,A)(𝟏)\mathcal{H}_{A^{e}}(A,\,A)(\mathbf{1}), there exists a morphism fAe(A,AA)(𝟏)f\in\mathcal{H}_{A^{e}}(A,\,A\otimes A)(\mathbf{1}) such that μf=A\mu\circ f=A. ∎

We will finish the section with an example but, first, we need the following lemmas. The first one generalizes what happens in the classical case.

Lemma 5.7.

Let AA be a monoid in \mathcal{F}. Then AA is separable if and only if there exists an element ξC(AA)A(𝟏)\xi\in C(A\otimes A)_{A}(\mathbf{1}) such that μ𝟏(ξ)=ε\mu_{\mathbf{1}}(\xi)=\varepsilon. Moreover, ξ\xi determines the morphism of separability t:AAAt:A\rightarrow A\otimes A.

Proof.

Suppose first that AA is separable. Then we have t:AAAt:A\rightarrow A\otimes A in AeA^{e}-Mod that satisfies μt=A\mu\circ t=A. Let ξ=t𝟏(ε)\xi=t_{\mathbf{1}}(\varepsilon). For an object xx in 𝒳\mathcal{X} and aA(x)a\in A(x), the action on the left gives

a×ξ=a×t𝟏(ε)=tx𝟏(a×ε)=tx𝟏(A(ρx1)(a))=(AA)(ρx1)(tx(a))a\times\xi=a\times t_{\mathbf{1}}(\varepsilon)=t_{x\diamond\mathbf{1}}(a\times\varepsilon)=t_{x\diamond\mathbf{1}}(A(\rho_{x}^{-1})(a))=(A\otimes A)(\rho_{x}^{-1})(t_{x}(a))

and on the right

ξ×a=t𝟏(ε)×a=t𝟏x(ε×a)=t𝟏x(A(λx1)(a))=(AA)(λx1)(tx(a)),\xi\times a=t_{\mathbf{1}}(\varepsilon)\times a=t_{\mathbf{1}\diamond x}(\varepsilon\times a)=t_{\mathbf{1}\diamond x}(A(\lambda_{x}^{-1})(a))=(A\otimes A)(\lambda_{x}^{-1})(t_{x}(a)),

but ρx1=s𝟏xx𝟏λx1\rho_{x}^{-1}=s_{\mathbf{1}\diamond x}^{x\diamond\mathbf{1}}\circ\lambda_{x}^{-1}. This shows that ξC(AA)A(𝟏)\xi\in C(A\otimes A)_{A}(\mathbf{1}) and that it determines the morphism tt.

Now suppose we have ξC(AA)A(𝟏)\xi\in C(A\otimes A)_{A}(\mathbf{1}) as in the hypothesis of the lemma and define t:AAAt:A\rightarrow A\otimes A in an object xx as

tx(a):=(AA)(ρx)(a×ξ),t_{x}(a):=(A\otimes A)(\rho_{x})(a\times\xi),

where a×ξa\times\xi is the left action of AA on AAA\otimes A. Since ξC(AA)A(𝟏)\xi\in C(A\otimes A)_{A}(\mathbf{1}) this is equal to (AA)(λx)(ξ×a)(A\otimes A)(\lambda_{x})(\xi\times a). This clearly defines a natural transformation from AA to AAA\otimes A and the functoriality of the bi-action of AA on AAA\otimes A shows that tt is a morphism of bimodules. Finally, we clearly have t𝟏(ε)=ξt_{\mathbf{1}}(\varepsilon)=\xi and thus μ𝟏t𝟏(ε)=ε\mu_{\mathbf{1}}\circ t_{\mathbf{1}}(\varepsilon)=\varepsilon. So, for any object xx in 𝒳\mathcal{X} and any aA(x)a\in A(x), since μ\mu is a morphism of bimodules, we have

a=A(ρx)(ε×a)=A(ρx)(μ𝟏xt𝟏x(ε×a))=μxtx(A(ρx)(ε×a)),a=A(\rho_{x})(\varepsilon\times a)=A(\rho_{x})(\mu_{\mathbf{1}\diamond x}t_{\mathbf{1}\diamond x}(\varepsilon\times a))=\mu_{x}t_{x}(A(\rho_{x})(\varepsilon\times a)),

the last equality comes from the naturality of ρ\rho. Thus μt=A\mu\circ t=A. ∎

Lemma 5.8.

Let BB and AA be monoids in \mathcal{F} and suppose f:BAf:B\rightarrow A is a morphism of monoids that is an epimorphism in \mathcal{F}. If BB is separable, then AA is separable.

Proof.

We notice first that ff (and we do not need it to be an epimorphism for this) induces a structure of BB-bimodule in AA, given by

(BA)B\textstyle{(B\otimes A)\otimes B\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}\scriptstyle{\cong}(fA)f\scriptstyle{(f\otimes A)\otimes f}(AA)A\textstyle{(A\otimes A)\otimes A\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}\scriptstyle{\cong}μAA\scriptstyle{\mu_{A}\otimes A}AA\textstyle{A\otimes A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}μA\scriptstyle{\mu_{A}}A\textstyle{A}B(AB)\textstyle{B\otimes(A\otimes B)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}f(Af)\scriptstyle{f\otimes(A\otimes f)}A(AA)\textstyle{A\otimes(A\otimes A)\ignorespaces\ignorespaces\ignorespaces\ignorespaces}AμA\scriptstyle{A\otimes\mu_{A}}AA.\textstyle{A\otimes A.\ignorespaces\ignorespaces\ignorespaces\ignorespaces}μA\scriptstyle{\mu_{A}}

Since ff is a morphism of monoids, it is easy to see it is also a morphism of BB-bimodules. Then clearly, AAA\otimes A is a BB-bimodule and fff\otimes f is a morphism of bimodules.

Let ξB\xi_{B} be an element in C(BB)B(𝟏)C(B\otimes B)_{B}(\mathbf{1}) as in the previous lemma and consider ξA:=(ff)(ξB)\xi_{A}:=(f\otimes f)(\xi_{B}). We have

μA,𝟏(ξA)=μA,𝟏(ff)𝟏(ξB)=f𝟏μB,𝟏(ξB)=f𝟏(εB)=εA,\mu_{A,\mathbf{1}}(\xi_{A})=\mu_{A,\mathbf{1}}(f\otimes f)_{\mathbf{1}}(\xi_{B})=f_{\mathbf{1}}\mu_{B,\mathbf{1}}(\xi_{B})=f_{\mathbf{1}}(\varepsilon_{B})=\varepsilon_{A},

since ff is a morphism of monoids.

Now let xx be an object of 𝒳\mathcal{X} and aA(x)a\in A(x). Then there exists bB(x)b\in B(x) such that fx(b)=af_{x}(b)=a and the left action of a×ξAa\times\xi_{A} is equal to fx(b)×(ff)𝟏(ξB)f_{x}(b)\times(f\otimes f)_{\mathbf{1}}(\xi_{B}). But this is precisely the left action of BB on AAA\otimes A and, since fff\otimes f is a morphism of bimodules, we have that this is equal to

(ff)x𝟏(b×ξB)=(ff)x𝟏((BB)(s𝟏xx𝟏)(ξB×b)),(f\otimes f)_{x\diamond\mathbf{1}}(b\times\xi_{B})=(f\otimes f)_{x\diamond\mathbf{1}}((B\otimes B)(s_{\mathbf{1}\diamond x}^{x\diamond\mathbf{1}})(\xi_{B}\times b)),

which, by functoriality, is equal to

(AA)(s𝟏xx𝟏)((ff)𝟏x(ξB×b)),(A\otimes A)(s_{\mathbf{1}\diamond x}^{x\diamond\mathbf{1}})((f\otimes f)_{\mathbf{1}\diamond x}(\xi_{B}\times b)),

which is equal to (AA)(s𝟏xx𝟏)(ξA×a)(A\otimes A)(s_{\mathbf{1}\diamond x}^{x\diamond\mathbf{1}})(\xi_{A}\times a). Hence ξAC(AA)A(𝟏)\xi_{A}\in C(A\otimes A)_{A}(\mathbf{1}). ∎

Corollary 5.9.

For very object FF in \mathcal{F} and every i>0i>0, we have Hi(I,F)=0\mathcal{H}H^{i}(I,\,F)=0. Moreover if AA is a monoid such that there exists a morphism of monoids IAI\rightarrow A which is an epimorphism in \mathcal{F}, then for every AA-bimodule MM and every i>0i>0 we have Hi(A,M)=0\mathcal{H}H^{i}(A,\,M)=0.

Example 5.10.

The previous corollary implies that for every biset functor FF and every i>0i>0, we have Hi(RB,F)=0\mathcal{H}H^{i}(RB,\,F)=0. Also, if we take R=kR=k as a field of characteristic 0 and A=kRA=kR_{\mathbb{Q}} as the biset functor of rational representations, we know that kRkR_{\mathbb{Q}} is a commutative monoid and that the linearization morphism kBkRkB\rightarrow kR_{\mathbb{Q}} provides an arrow in 𝒞,R\mathcal{F}_{\mathcal{C},\,R} as in the previous corollary. Hence, for every kRkR_{\mathbb{Q}}-module MM and every i>0i>0, we have Hi(kR,M)=0\mathcal{H}H^{i}(kR_{\mathbb{Q}},\,M)=0.

6 Hochschild cohomology functors

Through this section AA is a monoid in \mathcal{F}, MM and NN are AA-bimodules and xx, yy, zz and ww are objects of 𝒳\mathcal{X}.

Let PP, QQ and TT be objects in \mathcal{F}. To simplify the notation, in what follows we will work indistinctly with Hom(PQ,T)\mathrm{Hom}_{\mathcal{F}}(P\otimes Q,\,T) and the corresponding set of bilinear maps given by Remark 3.5. In particular, given an arrow and f:PQTf:P\otimes Q\rightarrow T in \mathcal{F} we will avoid the notation f~\tilde{f}, used in Section 4.2.

6.1 n=0n=0

As said before, in this case we clearly have

H0(A,M)CMAAe(A,M).\mathcal{H}H^{0}(A,\,M)\cong CM_{A}\cong\mathcal{H}_{A^{e}}(A,\,M).

6.2 n=1n=1

Definition 6.1.

An arrow fHom(A,Mx)f\in\mathrm{Hom}_{\mathcal{F}}(A,\,M_{x}) is called a derivation from AA to MxM_{x} if for any aA(y)a\in A(y) and bA(z)b\in A(z), we have

fyz(a×b)=(a×fz(b))+M(syxzyzx)(fy(a)×b).f_{y\diamond z}(a\times b)=(a\times f_{z}(b))+M(s_{y\diamond x\diamond z}^{y\diamond z\diamond x})(f_{y}(a)\times b).

The arrow ff is called an inner derivation if there exist mM(x)m\in M(x) such that the morphisms fyf_{y} are given by sending aA(y)a\in A(y) to a×mM(sxyyx)(m×a)a\times m-M(s_{x\diamond y}^{y\diamond x})(m\times a).

By the last paragraphs of Section 4.2, we have that the kernel of

βx:Hom(A,Mx)Hom(A2,Mx)\beta_{x}:\mathrm{Hom}_{\mathcal{F}}(A,\,M_{x})\rightarrow\mathrm{Hom}_{\mathcal{F}}(A^{\otimes 2},\,M_{x})

corresponds to the derivations from AA to MxM_{x} and that the image of βx:M(x)Hom(A,Mx)\beta_{x}:M(x)\rightarrow\mathrm{Hom}_{\mathcal{F}}(A,\,M_{x}) corresponds to the inner derivations from AA to MxM_{x}. In particular, we can define two subfunctors of (A,M)\mathcal{H}(A,\,M) in \mathcal{F}. First, that of derivations Der(A,M)\textrm{Der}(A,\,M), defined in xx as the derivations from AA to MxM_{x}, and then that of inner derivations Inn(A,M)\textrm{Inn}(A,\,M), defined accordingly. With this

H1(A,M)=Der(A,M)Inn(A,M).\mathcal{H}H^{1}(A,\,M)=\frac{\textrm{Der}(A,\,M)}{\textrm{Inn}(A,\,M)}.

As in the classical case, we can endow H1(A,A)\mathcal{H}H^{1}(A,\,A) with a Lie-type structure in the following way. If d¯H1(A,A)(x)\overline{d}\in\mathcal{H}H^{1}(A,\,A)(x) and d¯H1(A,A)(x)\overline{d^{\prime}}\in\mathcal{H}H^{1}(A,\,A)(x^{\prime}), consider the arrow [d,d]:AAxx[d,\,d^{\prime}]:A\rightarrow A_{x\diamond x^{\prime}}, given as

[d,d]y=(dyxdy)A(syxxyxx)(dyxdy).[d,\,d^{\prime}]_{y}=(d^{\prime}_{y\diamond x}\circ d_{y})-A(s_{y\diamond x^{\prime}\diamond x}^{y\diamond x\diamond x^{\prime}})(d_{y\diamond x^{\prime}}\circ d^{\prime}_{y}).

Defining [d¯,d¯][\overline{d},\,\overline{d^{\prime}}] as [d,d]¯H1(A,A)(xx)\overline{[d,\,d^{\prime}]}\in\mathcal{H}H^{1}(A,\,A)(x\diamond x^{\prime}) is well defined and the bracket satisfies the corresponding Lie-type axioms.

Remark 6.2.

Serge Bouc has suggested that if AA is a Green biset functor, then we can call H1(A,A)\mathcal{H}H^{1}(A,\,A) a Lie biset functor.

6.3 n=2n=2

Definition 6.3.

A square-zero extension of AA consists of a monoid EE in \mathcal{F} and a monoid homomorphism π:EA\pi:E\rightarrow A which is an epimorphism in \mathcal{F} and such that KerπKer\pi is an ideal of EE of square zero, that is, the bilinear maps

Kerπ(y)×Kerπ(z)Kerπ(yz)Ker\pi(y)\times Ker\pi(z)\rightarrow Ker\pi(y\diamond z)

are zero for all yy and zz.

In this case, it is easy to see that KerπKer\pi becomes an AA-bimodule. If KerπKer\pi is isomorphic to NN, as AA-bimodules, we call this extension a square-zero extension of AA by NN.

Definition 6.4.

A square-zero extension of AA is called a Hochschild extension if it is a split extension in \mathcal{F}, i.e., there exists an arrow σ:AE\sigma:A\rightarrow E in \mathcal{F} such that πσ=A\pi\sigma=A.

Two Hochschild extensions, π:EA\pi:E\rightarrow A and π:EA\pi^{\prime}:E^{\prime}\rightarrow A, of AA by NN are said to be equivalent if there exists a monoid morphism φ:EE\varphi:E\rightarrow E^{\prime} such that the squares in the following diagram

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}N\textstyle{N\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}N\scriptstyle{N}E\textstyle{E\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}φ\scriptstyle{\varphi}π\scriptstyle{\pi}A\textstyle{A\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces\ignorespaces}A\scriptstyle{A}0\textstyle{0}0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}N\textstyle{N\ignorespaces\ignorespaces\ignorespaces\ignorespaces}E\textstyle{E^{\prime}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}π\scriptstyle{\pi^{\prime}}A\textstyle{A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0\textstyle{0}

commute. This defines an equivalence relation in the collection of all the Hochschild extensions of AA by NN, we denote the collection of equivalence classes by Ext(A,N)\mathrm{Ext}(A,\,N).

Notice that Ext(A,N)\mathrm{Ext}(A,\,N) is not empty, since we have the class of the semidirect product NAN\rtimes A. As an object in \mathcal{F}, it is given by the coproduct of NN with AA, that is (NA)(y)=N(y)A(y)(N\rtimes A)(y)=N(y)\oplus A(y). As a monoid, the product is given by

((m1,a1),(m2,a2))(N(szyyz)(m1×a2)+(a1×m2),a1×a2),((m_{1},\,a_{1}),(m_{2},\,a_{2}))\mapsto(N(s_{z\diamond y}^{y\diamond z})(m_{1}\times a_{2})+(a_{1}\times m_{2}),\,a_{1}\times a_{2}),

for (m1,a1)N(y)A(y)(m_{1},\,a_{1})\in N(y)\oplus A(y) and (m2,a2)N(z)A(z)(m_{2},\,a_{2})\in N(z)\oplus A(z).

Theorem 6.5.

There exists a bijection

H2(A,M)(x)Ext(A,Mx).\mathcal{H}H^{2}(A,\,M)(x)\leftrightarrow\mathrm{Ext}(A,\,M_{x}).
Proof.

By the last paragraphs of Section 4.2, we have that the kernel of

βx:Hom(A2,Mx)Hom(A3,Mx)\beta_{x}:\mathrm{Hom}_{\mathcal{F}}(A^{\otimes 2},\,M_{x})\rightarrow\mathrm{Hom}_{\mathcal{F}}(A^{\otimes 3},\,M_{x})

consists of arrows fHom(A2,Mx)f\in\mathrm{Hom}_{\mathcal{F}}(A^{\otimes 2},\,M_{x}) that satisfy that for any (a1,a2,a3)(a_{1},\,a_{2},\,a_{3}) in A(y)×A(z)×A(w)A(y)\times A(z)\times A(w), the expression

a1×fz,w(a2,a3)fyz,w(a1×a2,a3)+fy,zw(a1,a2×a3)M(yzsxwwx)(fy,z(a1,a2)×a3)a_{1}\times f_{z,\,w}(a_{2},\,a_{3})-f_{y\diamond z,\,w}(a_{1}\times a_{2},\,a_{3})+f_{y,\,z\diamond w}(a_{1},\,a_{2}\times a_{3})-M(y\diamond z\diamond s_{x\diamond w}^{w\diamond x})(f_{y,\,z}(a_{1},\,a_{2})\times a_{3})

is equal to zero.

For such an ff, we consider Ef=MxAE_{f}=M_{x}\oplus A in \mathcal{F} with the following product

((m1,a1),(m2,a2))(M(syxzyzx)(m1×a2)+(a1×m2)+fy,z(a1,a2),a1×a2),((m_{1},\,a_{1}),(m_{2},a_{2}))\mapsto(M(s_{y\diamond x\diamond z}^{y\diamond z\diamond x})(m_{1}\times a_{2})+(a_{1}\times m_{2})+f_{y,\,z}(a_{1},\,a_{2}),\,a_{1}\times a_{2}),

for (m1,a1)Mx(y)A(y)(m_{1},\,a_{1})\in M_{x}(y)\oplus A(y) and (m2,a2)Mx(z)A(z)(m_{2},\,a_{2})\in M_{x}(z)\oplus A(z). The equality to zero of the expression above allows us to show that this product is associative. Also, it shows that if we take mε=f𝟏, 1(ε,ε)m_{\varepsilon}=-f_{\mathbf{1},\,\mathbf{1}}(\varepsilon,\,\varepsilon), then (mε,ε)Mx(𝟏)A(𝟏)(m_{\varepsilon},\,\varepsilon)\in M_{x}(\mathbf{1})\oplus A(\mathbf{1}) is the identity element for the product. Finally, it is easy to verify that this product is functorial. With this, it is immediate to see that the extension

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Mx\textstyle{M_{x}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Ef\textstyle{E_{f}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}A\textstyle{A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0\textstyle{0}

is a Hochschild extension. Now consider f=fβx(g)f^{\prime}=f-\beta_{x}(g), where gHom(A,Mx)g\in\mathrm{Hom}_{\mathcal{F}}(A,M_{x}) and

βx:Hom(A,Mx)Hom(A2,Mx).\beta_{x}:\mathrm{Hom}_{\mathcal{F}}(A,\,M_{x})\rightarrow\mathrm{Hom}_{\mathcal{F}}(A^{\otimes 2},\,M_{x}).

Then, the extensions given by EfE_{f} and EfE_{f^{\prime}} are equivalent. Indeed, the isomorphism of monoids φ:EfEf\varphi:E_{f}\rightarrow E_{f^{\prime}} is given by sending (m,a)Mx(y)A(y)(m,\,a)\in M_{x}(y)\oplus A(y) to (m+gy(a),a)Mx(y)A(y)(m+g_{y}(a),\,a)\in M_{x}(y)\oplus A(y).

On the other direction, given a Hochschild extension

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Mx\textstyle{M_{x}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}λ\scriptstyle{\lambda}E\textstyle{E\ignorespaces\ignorespaces\ignorespaces\ignorespaces}π\scriptstyle{\pi}A\textstyle{A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0,\textstyle{0,}

since there exists σ\sigma in \mathcal{F} such that πσ=A\pi\sigma=A, then EMxAE\cong M_{x}\oplus A in \mathcal{F}. Hence, we can obtain the monoid structure of EE from this direct sum,

(λy(m1),σy(a1))×(λz(m2),σz(a2)),(\lambda_{y}(m_{1}),\,\sigma_{y}(a_{1}))\times(\lambda_{z}(m_{2}),\,\sigma_{z}(a_{2})),

for (m1,a1)(MxA)(y)(m_{1},\,a_{1})\in(M_{x}\oplus A)(y) and (m2,a2)(MxA)(z)(m_{2},\,a_{2})\in(M_{x}\oplus A)(z). To simplify the notation, let n1=λy(m1)n_{1}=\lambda_{y}(m_{1}), b1=σy(a1)b_{1}=\sigma_{y}(a_{1}), n2=λz(m2)n_{2}=\lambda_{z}(m_{2}) and b2=σz(a2)b_{2}=\sigma_{z}(a_{2}). Then, since the product is bilinear and MxM_{x} is of square zero, it is easy to see that the product above is equal to

(Imλ(syxzyzx)(n1×b2)+(b1×n2)+fy,z0(b1,b2),b1×b2),\left(\textrm{Im}\lambda(s_{y\diamond x\diamond z}^{y\diamond z\diamond x})(n_{1}\times b_{2})+(b_{1}\times n_{2})+f^{0}_{y,\,z}(b_{1},\,b_{2}),b_{1}\times b_{2}\right),

where fyz0(b1,b2)f^{0}_{y\diamond z}(b_{1},\,b_{2}) is the unique element in Imλ(yz)\textrm{Im}\lambda(y\diamond z) that corresponds to the product (0,b1)×(0,b2)(0,\,b_{1})\times(0,\,b_{2}). This allows us to define a map fy,zE:A(y)×A(z)Mx(yz)f^{E}_{y,\,z}:A(y)\times A(z)\rightarrow M_{x}(y\diamond z) sending (a1,a2)(a_{1},\,a_{2}) to λyz1fyz0(σy(a1),σz(a2))\lambda_{y\diamond z}^{-1}f^{0}_{y\diamond z}(\sigma_{y}(a_{1}),\,\sigma_{z}(a_{2})). In turn, this defines an arrow fEHom(A2,Mx)f^{E}\in\mathrm{Hom}_{\mathcal{F}}(A^{\otimes 2},\,M_{x}) and the associativity of the product of EE implies that fEf^{E} is in the kernel of βx\beta_{x}. Finally, suppose we have an equivalent extension

0\textstyle{0\ignorespaces\ignorespaces\ignorespaces\ignorespaces}Mx\textstyle{M_{x}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}λ\scriptstyle{\lambda^{\prime}}E\textstyle{E^{\prime}\ignorespaces\ignorespaces\ignorespaces\ignorespaces}π\scriptstyle{\pi^{\prime}}A\textstyle{A\ignorespaces\ignorespaces\ignorespaces\ignorespaces}0,\textstyle{0,}

with splitting morphism σ:AE\sigma^{\prime}:A\rightarrow E^{\prime}. Let φ:EE\varphi:E\rightarrow E^{\prime} be the equivalence morphism. Then, as before, for a1A(y)a_{1}\in A(y) there exist a unique fy0(σy(a1))Imλ(y)f^{0}_{y}(\sigma_{y}(a_{1}))\in\mathrm{Im}\lambda^{\prime}(y) that corresponds to φy(0,σy(a1))E(y)\varphi_{y}(0,\,\sigma_{y}(a_{1}))\in E^{\prime}(y). Hence we can define gy:A(y)Mx(y)g_{y}:A(y)\rightarrow M_{x}(y) by sending a1a_{1} to (λ)y1fy0(σy(a1))(\lambda^{\prime})^{-1}_{y}f^{0}_{y}(\sigma_{y}(a_{1})). This defines an arrow in Hom(A,Mx)\mathrm{Hom}_{\mathcal{F}}(A,\,M_{x}). Also, using the fact that φ\varphi is a morphism of monoids, if we take the morphism defined before fEHom(A2,Mx)f^{E^{\prime}}\in\mathrm{Hom}_{\mathcal{F}}(A^{\otimes 2},\,M_{x}) corresponding to EE^{\prime}, then we obtain fEfE=βx(g)f^{E}-f^{E^{\prime}}=\beta_{x}(g). ∎

Corollary 6.6.

Ext(A,M_)\mathrm{Ext}(A,\,M_{\,\_\,}) is a functor in \mathcal{F}.

Proof.

Indeed, since H2(A,M)(x)\mathcal{H}H^{2}(A,\,M)(x) is an RR-module, the previous bijection endows Ext(A,Mx)\mathrm{Ext}(A,\,M_{x}) with a structure of RR-module. The functorial structure is also given by the bijection. That is, given an arrow α:xy\alpha:x\rightarrow y in 𝒳\mathcal{X}, then the map

Ext(A,Mx)Ext(A,My)\mathrm{Ext}(A,\,M_{x})\rightarrow\mathrm{Ext}(A,\,M_{y})

sends (the class of) the extension EfE_{f}, corresponding to the cocycle ff, to EMαfE_{M_{\alpha}\circ f}. ∎

Remark 6.7.

It is not hard to see that the sum we obtain in Ext(A,Mx)\mathrm{Ext}(A,\,M_{x}), by the previous corollary, corresponds to the Baer sum of extensions. Given two Hochschild extensions, π:EA\pi:E\rightarrow A and π:EA\pi^{\prime}:E^{\prime}\rightarrow A, of AA by NN, their Baer sum is defined in yy by taking the pullback of πy\pi_{y} and πy\pi^{\prime}_{y}, that is

Γ(y)={(a,b)E(y)E(y)πy(a)=πy(b)},\Gamma(y)=\{(a,\,b)\in E(y)\oplus E^{\prime}(y)\mid\pi_{y}(a)=\pi^{\prime}_{y}(b)\},

and then making the quotient over {(λy(n),λy(n))nN(y)}\{(\lambda_{y}(n),\,-\lambda^{\prime}_{y}(n))\mid n\in N(y)\}, if λ:NE\lambda:N\rightarrow E and λ:NE\lambda^{\prime}:N\rightarrow E^{\prime} are the corresponding morphisms. It is defined in an obvious way in arrows. This construction yields a Hochschild extension of AA by NN.

6.4 The Hochschild cohomology monoid

Definition 6.8.

An \mathbb{N}-graded monoid in \mathcal{F} is a lax monoidal functor G:G:\mathbb{N}\rightarrow\mathcal{F}, where \mathbb{N} is seen as a discrete monoidal category.

That is, an \mathbb{N}-graded monoid consists of the following:

  • For each ii\in\mathbb{N}, a functor FiF_{i} in \mathcal{F}.

  • For every i,ji,\,j\in\mathbb{N} and every xx and yy objects in 𝒳\mathcal{X}, a bilinear map

    Fi(x)×Fj(y)Fi+j(xy),F_{i}(x)\times F_{j}(y)\rightarrow F_{i+j}(x\diamond y),

    that is associative in an obvious way and functorial in xx and yy (analogous conditions to those appearing in Section 3 for a monoid).

  • An element εF0F0(𝟏)\varepsilon_{F_{0}}\in F_{0}(\mathbf{1}) such that for every ii\in\mathbb{N} and aFi(x)a\in F_{i}(x),

    a=Fi(λx)(εF0×a)=Fi(ρx)(a×εF0).a=F_{i}(\lambda_{x})(\varepsilon_{F_{0}}\times a)=F_{i}(\rho_{x})(a\times\varepsilon_{F_{0}}).

In this case, F0F_{0} is a monoid in \mathcal{F} and every FiF_{i} is an F0F_{0}-bimodule.

Since \mathcal{F} is an abelian category, we can consider the coproduct

H(A):=iHi(A,A)\mathcal{H}H(A):=\bigoplus_{i\in\mathbb{N}}\mathcal{H}H^{i}(A,\,A)

in \mathcal{F}. We will endow H(A)\mathcal{H}H(A) with a structure of graded monoid. We begin by defining, for i,ji,\,j\in\mathbb{N}, a cup product,

__:(Ai,A)(x)×(Aj,A)(y)(Ai+j,A)(xy),\_\,\smile\,\_:\mathcal{H}(A^{\otimes i},\,A)(x)\times\mathcal{H}(A^{\otimes j},\,A)(y)\rightarrow\mathcal{H}(A^{\otimes i+j},\,A)(x\diamond y),

sending (fi,fj)(f^{i},\,f^{j}), with fiHom(Ai,Ax)f^{i}\in\textrm{Hom}_{\mathcal{F}}(A^{\otimes i},\,A_{x}) and fjHom(Aj,Ay)f^{j}\in\textrm{Hom}_{\mathcal{F}}(A^{\otimes j},\,A_{y}), to

fifj:A(x1)××A(xi)×A(y1)××A(yj)Axy(x1xiy1yj),f^{i}\smile f^{j}:A(x_{1})\times\cdots\times A(x_{i})\times A(y_{1})\times\cdots\times A(y_{j})\rightarrow A_{x\diamond y}(x_{1}\diamond\cdots\diamond x_{i}\diamond y_{1}\diamond\cdots\diamond y_{j}),

which sends an (i+j)(i+j)-tuple (a1,,ai,b1,,bj)(a_{1},\ldots,a_{i},\,b_{1},\ldots,b_{j}) to

A(s(i)x(j)y(i+j)xy)(fx1,,xii(a1,,ai)×fy1,,yjj(b1,,bj)),A\left(s_{(i)\diamond x\diamond(j)\diamond y}^{(i+j)\diamond x\diamond y}\right)\left(f^{i}_{x_{1},\ldots,x_{i}}(a_{1},\ldots,a_{i})\times f^{j}_{y_{1},\ldots,y_{j}}(b_{1},\ldots,b_{j})\right),

here s(i)x(j)y(i+j)xys_{(i)\diamond x\diamond(j)\diamond y}^{(i+j)\diamond x\diamond y} is the symmetry in 𝒳\mathcal{X} sending x1xixy1yjyx_{1}\diamond\cdots\diamond x_{i}\diamond x\diamond y_{1}\diamond\cdots\diamond y_{j}\diamond y to x1xiy1yjxyx_{1}\diamond\cdots\diamond x_{i}\diamond y_{1}\diamond\cdots\diamond y_{j}\diamond x\diamond y.

This product makes i=0n(Ai,A)\bigoplus_{i=0}^{n}\mathcal{H}(A^{\otimes i},\,A) an \mathbb{N}-graded monoid in \mathcal{F}, with identity εAA(𝟏)(A0,A)(𝟏)\varepsilon_{A}\in A(\mathbf{1})\cong\mathcal{H}(A^{\otimes 0},\,A)(\mathbf{1}). Next, consider the complex (A,A)\mathcal{H}(A^{\otimes*},\,A) with morphisms β\beta as in Section 4.2. A series of straightforward computations (in particular one must pay attention to the symmetries that appear in the expressions) show that for any i,ji,\,j\in\mathbb{N} and fif^{i} and fjf^{j} as before,

β(fifj)=βfifj+(1)ifiβfj.\beta(f^{i}\smile f^{j})=\beta f^{i}\smile f^{j}+(-1)^{i}f^{i}\smile\beta f^{j}.

This formula allows us to extend the cup product to iHi(A,A)\bigoplus_{i\in\mathbb{N}}\mathcal{H}H^{i}(A,\,A), just as in the classical case. Also, it is easy to see that the product satisfies the conditions for a graded monoid, with εACA(𝟏)H0(A,A)(𝟏)\varepsilon_{A}\in CA(\mathbf{1})\cong\mathcal{H}H^{0}(A,\,A)(\mathbf{1}).

Remark 6.9.

If AA is a Green biset functor, then we can call H(A)\mathcal{H}H(A) a graded Green biset functor.

Remark 6.10.

The cup product just described works exactly the same if we consider first

__:A(x)×(Aj,M)(y)(Aj,M)(xy),\_\,\smile\,\_:A(x)\times\mathcal{H}(A^{\otimes j},\,M)(y)\rightarrow\mathcal{H}(A^{\otimes j},\,M)(x\diamond y),

via the isomorphism A(x)(A0,A)(x)A(x)\cong\mathcal{H}(A^{\otimes 0},\,A)(x) and then

__:CA(x)×Hj(A,M)(y)Hj(A,M)(xy).\_\,\smile\,\_:CA(x)\times\mathcal{H}H^{j}(A,\,M)(y)\rightarrow\mathcal{H}H^{j}(A,\,M)(x\diamond y).

This shows that each Hj(A,M)\mathcal{H}H^{j}(A,\,M) is a CACA-module.

6.5 Further results

The author believes that the following results, not considered in this paper, can also be extended to our context.

  • i)

    The description of H3(A,M)\mathcal{H}H^{3}(A,\,M) in terms of crossed bimodules (see E.1.5.1 in [8]).

  • ii)

    The definition of a bracket giving H(A)\mathcal{H}H(A) a structure of graded Lie monoid, with the grading shifted by -1, as in the classical case.

Acknowledgments

The contents of this article were developed during a sabbatical year the author did at the laboratory LAMFA of the Université de Picardie, in Amiens, France, from August 2022 to July 2023. The author thanks the staff and colleagues at LAMFA for all the support she received during her stay, which led to the successful development of the project. Special thanks to Serge Bouc, host during the sabbatical, for all the ideas, suggestions and stimulating conversations.

References

  • [1] A. Ardizzoni, C. Menini, and D. Ştefan. Hochschild cohomology and “smoothnes” in monoidal categories. Journal of pure and applied algebra, 208:297–330, 2007.
  • [2] S. Bouc. Résolutions de foncteurs de Mackey. In Group representations: cohomology, group actions and topology (Seattle, WA, 1996), volume 63 of Proc. Sympos. Pure Math., pages 31–83. Amer. Math. Soc., Providence, RI, 1998.
  • [3] Serge Bouc. Biset functors for finite groups. Springer, Berlin, 2010.
  • [4] Serge Bouc and Nadia Romero. The center of a Green biset functor. Pacific J. Math., 303(2):459–490, 2019.
  • [5] Murray Gerstenhaber. The cohomology structure of an associative ring. Annals of mathematics, 78:267–288, 1963.
  • [6] Magnus Hellstrøm-Finnsen. Hochschild cohomology of ring objets in monoidal categories. Communications in algebra, 46:5202–5233, 2018.
  • [7] Michael A. Hill, Michael J. Hopkins, and Douglas C. Ravenel. Equivariant stable homotopy theory and the Kervaire invariant problem, volume 40 of New Mathematical Monographs. Cambridge University Press, Cambridge, 2021.
  • [8] Jean-Louis Loday. Cyclic homology. Springer-Verlag, Berlin, 2nd edition, 1998.
  • [9] Saunders Mac Lane. Categories for the working mathematician. Springer, Berlin, 1971.