Batalin-Vilkovisky algebras and the noncommutative Poincaré duality of Koszul Calabi-Yau algebras (Q5964095)

Koszul Calabi-Yau algebras \(A\) comes from (among other theories) deformation theory, and can be constructed from \textit{a potential} which is of recent interest. The authors show that for a Koszul Calabi-Yau algebra \(A\), its Hochshchild cohomology is isomorphic to the Hochschild cohomology of its Koszul dual algebra \(A^!\) as \textit{Batalin-Vilkoviky algebras}, i.e. \(\text{HH}^\bullet(A;A)\simeq\text{HH}^\bullet(A^!;A^!)\). A lot of results are known about these algebras, in particular the Batalin-Vilkovisky algebra structure on both sides. A (bi-)graded algebra \(A\) is \textit{Calabi-Yau} of dimension \(n\) if \(A\) is homologically smooth, and if there is an isomorphism \(\eta:\text{RHom}_{A^e}(A,A\otimes A)\rightarrow\sum^n A(l)\) in the derived category of (bi-)graded left bimodules \(\mathcal D(A^e)\). \(A\) is homologically smooth if it is perfect as \(A^e\)-module, and \(A^e=A\otimes_kA^{op}\). Ginzburg proved that there is a Batalin-Vilkoviski algebra structure on the Hochshild cohomology of a Calabi-Yau algebra \(A\). On the other hand, if \(A\) is a (differential graded) cyclic associative algebra, i.e. an associative algebra with a non-degenerate cyclically invariant pairing, Tradler constructed a Batalin-Vilkoviski structure on its Hochschild cohomology. If the Calabi-Yau algebra \(A\) is Koszul, then its Koszul dual algebra \(A^!\) has a non-degenerate pairing, and Tradler's construction can be applied to it. It follows from work of Buchweitz, Beilinson, Ginzburg and Soergel, and Keller, that \(\text{HH}^{\bullet}(A;A)\simeq\text{HH}^{\bullet}(A^{!};A^{!})\) as Gerstenhaber algebras, and Ginzburg conjectured that this is an isomorphism as Batalin-Vilkovisky algebras. The main goal of this article, is to prove that this conjecture is holds true: Here we give the main result verbatim: {Theorem } (Rouquier's conjecture). Suppose that \(A\) is a Koszul Calabi-Yau algebra, and let \(A^!\) be its Koszul dual algebra. Then there is an isomorphism \(\text{HH}^{\bullet}(A;A)\simeq\text{HH}^{\bullet}(A^{!};A^{!})\) of Batalin-Vilkoviski algebras between the Hochshild cohomology of \(A\) and \(A^!\). The main ingredients for proving the theorem is as follows: Let \(A\) be an associative algebra and \(A^{\text{!`}}\) its Koszul dual. Then the chain complex \((A\otimes A^{\text{!`}},b)\) computes the Hochschild homology of \(A\) and \(A^{\text{!`}}\) simultaneously, \(\text{HH}_{\bullet}(A)\simeq\text{H}_{\bullet}(A\otimes A^{\text{!`}},b)\simeq\text{HH}_{\bullet}(A^{\text{!`}})\). The differential \(b\) is appropriately constructed. The Connes differential operator exists on both \(\text{HH}_{\bullet}(A)\) and \(\text{HH}_{\bullet}(A^{\text{!`}})\). This is not sufficient to define a Connes operator on \((A\otimes A^{\text{!`}},b)\), but it is proved that the above isomorphism commutes with the existing Connes operators. The authors prove that for the Koszul algebra \(A\), there is a canonical complex \((A\otimes A^{!},\delta)\) which computes the Hochschild cohomology of \(A\) and \(A^!\) simultaneously, \(A^!\) is the Koszul dual algebra of \(A\). This means that there are canonical isomorphisms \(\text{HH}^{\bullet}(A;A)\simeq\text{H}^{\bullet}(A\otimes A^{!},\delta)\simeq\text{HH}^{\bullet}(A^{!},A^{!})\). Two versions of Poincaré duality, the first by Van den Bergh, the second by Tradler, says that \(\text{HH}^{\bullet}(A;A)\simeq\text{HH}_{n-\bullet}(A)\), and \(\text{HH}^{\bullet}(A^{\text{!`}},A^{\text{!`}})\simeq\text{HH}{n-\bullet}(A^{\text{!`}})\). Together, these parts imply the main result that \(\text{HH}^{\bullet}(A;A)\simeq\text{HH}^{\bullet}(A^{!};A^{!})\), where the Batalin-Vilkovisky operator on each side is the pull-back of the Connes operator via the Poincaré duality. The authors point out that the above isomorphisms does not respect the usual grading, but are isomorphisms with respect to a certain bigrading. The article is mostly self contained, and contains the necessary tools. It starts with recalling the bar/cobar construction and twisting morphisms, and this is done explicitly. So is the treatment of Hochschild (co)homology and cyclic (co)homology of algebras. Also, the authors treat algebras and coalgebras simultaneously, so that the two constructions can be more easily compared. The article contains the explicit definitions of Gerstenhaber algebras, Koszul algebras and their duals, and give their (co)homology. Nice examples are given. Based on all the (needed) preliminaries, the definition of \textit{Koszul Calabi-Yau} algebras can be given. Their dimension and the noncommutative Poincaré duality are defined, and necessary results are recalled. In particular, a new proof of a result of Van den Bergh stating that a Koszul algebra \(A\) is \(n\)-Calabi-Yau if and only of \(A^!\) is cyclic of degree \(n\), is given. The authors give nice examples by enveloping algebras. The final section defines \textit{Batalin-Vilkovisky algebras} and prove the main theorem of the article. This is a complete and explicit proof, illustrating methods and making possible an application to cyclic homology in the end. The article gives an important contribution to the noncommutative homology theory used in noncommutative algebraic geometry, in particular deformation theory.