Kontsevich's graph complex, GRT, and the deformation complex of the sheaf of polyvector fields (Q894310): Difference between revisions

V. Drinfeld introduced the pro-unipotent algebraic Grothendieck-Teichmüller group \textbf{GRT}, closely related to \(\text{Gal}(\overline{\mathbb Q}/\mathbb Q)\), and appearing in the study of moduli of algebraic curves, motives, the Kashiwara-Vergne problem, and formal quantization procedures. The Lie algebra of \textbf{GRT}, \(\mathfrak{grt}\), has a natural \(\mathbb N\)-grading and a nonzero vector \(\sigma_n\) for every odd degree \(n\geq 3\), the Deligne-Drinfeld elements. The article considers a link between the graph complex \textsl{GC} introduced by M. Kontsevich and the Lie algebra \(\mathfrak{grt}\), it is shown that \(H^0({GD})\cong\mathfrak{grt}\). M. Kontsevich conjectured that \(\mathfrak{grt}\) reveals itself in the extended moduli of deformations of an algebraic variety \(X\) by an application of the action of odd components of the Chern character of \(X\) on the cohomology of the sheaf of polyvector fields \(H^\bullet(X,\mathcal T_{\text{poly}}).\) The isomorphism \(H^0({GD})\cong\mathfrak{grt}\) is used to prove this for an arbitrary smooth algebraic variety \(X\) over an algebraically closed field \(\mathbb K\) of characteristic zero, generalizing Kontsevich's construction from the case of affine space to the case of an arbitrary smooth algebraic variety. A link between the graph complex \textsl{GC} and the deformation complex of the operad \textsl{Ger} gives that every cocycle \(\gamma\in{GC}\) gives a derivation of the Gerstenhaber algebra \(H^\bullet(X,\mathcal T_{\text{poly}}).\) Combining this with \(H^0({GD})\cong\mathfrak{grt}\) a natural action of \(\mathfrak{grt}\) on the cohomology \(H^\bullet(X,\mathcal T_{\text{poly}})\) is obtained, and this is compatible with the Gerstenhaber algbra structure. The authors show that the action of \(\sigma_n\) in \(\mathfrak{grt}\) on \(H^\bullet(X,\mathcal T_{\text{poly}})\) is given by a nonzero multiple of the contraction with the \(n\)-th component of the Cern character of \(X\). Thus \textbf{GRT} is re-discovered in the extended moduli of deformations of \(X\) just as M. Kontsevich had predicted. The results in this article imply that the contraction of polyvector fields with any odd component of the Chern character induces a derivation of \(H^\bullet(X,\mathcal T_{\text{poly}})\) with respect to the cup-product. Also, the article contains a proof that the \(\hat A\)-genus in the Calaque-Van den Bergh formula for the isomorphism between harmonic and Hochschild structures can be replaced by a generalized \(\hat A\)-genus. Examples are given of algebraic varieties where odd components of the Chern character acts nontrivially on \(H^\bullet(X,\mathcal T_{\text{poly}})\). The main result of the paper is, more or less verbatim: {Theorem 8.1:} Let \(n\geq 3\) be odd. Then the action of \(\mathcal D_n\) on \(H^\bullet(X,\mathcal T_{\text{poly}})\) is a nonzero scalar multiple of the constraction with the \(n\)-th component of the Chern character of the tangent bundle of \(X\), regardless of the choice of Deligne-Drinfeld elements (Lie words). This result is used to prove that smooth Calabi-Yau complete intersections in projective spaces provides many nontrivial representations of \(\mathfrak{grt}\). Corollaries of Theorem 8.1 also give information about the Gerstenhaber algebra structure of \(H^\bullet(X,\mathcal T_{\text{poly}})\) when \(X\) is a complete intersection in a projective space. Towards the main results, the authors consider the Fedosov resolution of the sheaf of tensor fields on a smooth algebraic variety, the version given is a modification of the constriction proposed by Van den Bergh which is streamlined by avoiding the use of formal schemes and jets. This makes it possible for the authors to give an explicit representative of the Atiyah class of \(X\) in the Fedosov resolution of the tensor algebra. The operad \textsl{Gra}, the full graph complex \textsl{fGC} of Kontsevich graph complex \textsl{GC} is reviewed, and related results given. The authors construct a map \(\Theta\) of dg Lie algebras from \textsl{GC} to the deformation complex of dg sheaf \(\mathcal{FR}\), quasi-isomorphic to the sheaf of polyvector fields on \(X\) (\(\mathcal{FR}\) is viewed as a sheaf of dg Gerstenhaber algebras). Results of D. Calaque and M. Van den Bergh on harmonic and Hochschild structures of a smooth algebraic variety are generalized, and several examples given which show that the article's main results are nontrivial. The article ends with appendices. A: The homotopy \(O\)-algebra and the deformation complex. B: Sheaves of algebras on an operad, Thom-Sullivan normalization and global sections of a dg sheaf \(\mathcal A\) of operadic algebras and the deformation complex of \(\mathcal A\). C: Twisting of shifted Lie algebras and Gerstenhaber algebras by Maurer-Cartan elements. All in all, this is a detailed and readable article. It gives a thorough insight to operads and their applications to deformation theory in the derived setting. It gives a good understanding of the applications of this theory to polyvector fields, or similar deformation problems. The appendices can stand alone as introductions to each of its themes.