\(A_{\infty}\)-coderivations and the Gerstenhaber bracket on Hochschild cohomology (Q2032757): Difference between revisions

Let \(A\) be an algebra over a field \(\mathbf{k}\). It is well known that the Hochschild cohomology \(HH^{\bullet}(A) = \operatorname{Ext}^{\bullet}_{A^{\operatorname{e}}}(A,A)\) of \(A\), carries a natural structure of \textit{Gerstenhaber algebra}, i.e., a graded commutative algebra together with a Lie algebra structure on \(HH^{\bullet}(A)[-1]\) (following the conventions of the article) such that the bracket is a biderivation with respect to the algebra structure (see [\textit{M. Gertenhaber}, Ann. Math. (2) 78, 267--288 (1963; Zbl 0131.27302)]). The Lie bracket is usually called the \textit{Gerstenhaber bracket} of \(A\), and it is typically defined by making use of a particular projective resolution of the regular \(A\)-bimodule \(A\), called the \textit{bar resolution}. In the article under review the authors provide an intrinsic definition of the Gertenhaber algebra structure on \(HH^{\bullet}(A)\) using any projective resolution \(P'\) of the regular \(A\)-bimodule \(A\). More precisely, they first show that the shift \(P = P'[-1]\) of a projective resolution \(P'\) of the regular \(A\)-bimodule \(A\) has a natural structure of weakly counital (topological) \(A_{\infty}\)-coalgebra in the monoidal category \(\mathcal{Q}\) of dg \(A\)-bimodules (see Thm. 3.1.1). Given any general such \(A_{\infty}\)-coalgebra \(C\), one may construct its (topological) cobar construction \(\Omega C\) and the associated space of (continuous graded) derivations \(\operatorname{Der}_{\mathcal{Q}}(\Omega C)\), which is naturally a dg Lie algebra for the commutator bracket. The space of cocycles \(\operatorname{Coder}^{\infty}_{\mathcal{Q}}(C)\) of \(\operatorname{Der}_{\mathcal{Q}}(\Omega C)\) is called the space of \(A_{\infty}\)-\textit{coderivations} of \(C\) and its space of coboundaries \(\operatorname{Inn}^{\infty}_{\mathcal{Q}}(C)\) is called the space \textit{interior} \(A_{\infty}\)-coderivations of \(C\). The former is naturally a graded Lie algebra and the latter is a Lie ideal of the former. On the other hand, the space \(\operatorname{Der}_{\mathcal{Q}}(\Omega C)[1]\) has a natural structure of \(A_{\infty}\)-algebra coming from the convolution construction (see Def. 2.4.3 and Lemma 2.4.4), which induces a Gerstenhaber algebra structure on \((\operatorname{Coder}^{\infty}_{\mathcal{Q}}(C)/\operatorname{Inn}^{\infty}_{\mathcal{Q}}(C))[1]\) (see Thm. 2.4.7). The main result of the article then tells us that \(HH^{\bullet}(A)\) and \((\operatorname{Coder}^{\infty}_{\mathcal{Q}}(P)/\operatorname{Inn}^{\infty}_{\mathcal{Q}}(P))[1]\) are isomorphic Gertenhaber algebras (see Thm. 4.1.1). Most proofs are constructive and the authors provide some simple examples, mainly for the algebra \(A = \mathbf{k}[x]/(x^{n})\) with \(n = 2, 3\) (see Sections 3.2 and 4.7), even though the computations seem to become rather hard for more general algebras.