The equivariant Atiyah class (Q2019413)

The goal of the excellent paper under review is that of turning into a (number of) theorem(s) the dramatically appealing slogan: ``\textit{the Atiyah class of a \(G\)-equivariant perfect complex on \(X\) is \(G\)-equivariant}''. The slogan does not contradicts the traditional belief that deep beautiful mathematics can be formulated in very simple terms. The Atiyah class has a noble history and not just because was introduced by Sir Michael Atiyah himself in [\textit{M. F. Atiyah}, Trans. Am. Math. Soc. 85, 181--207 (1957; Zbl 0078.16002)]. It is the class of an extension of a vector bundle \(E\) over a complete algebraic scheme \(X\) by its twisting with the cotangent sheaf. This is clearly reminiscent of the first extension of a vector bundle, which the author alludes to. Its vanishing is to the existence of an algebraic connection on \(E\). More generally, the \textit{truncated equivariant Atiyah Class} \(At_E\) of a quasi-coherent sheaf \(E\) is an extension of \(E\) by \(E\otimes \mathbb{L}_X\), in this case \(\mathbb{L}_X\) is the truncated \textit{cotangent complex}. If \(X\) is acted on by an algebraic group \(G\), a shortcut to mean a connected \(\mathbb{C}\)-scheme of finite type, it should not sound surprising that one can speak of \(G\)-equivariant quasi-coherent sheaves as well as complexes out of them. One can then consider the \(G\)-equivariant derived category of the quasi coherente sheaves. The derived category of the category of quasi-coherent sheaves is the category whose objects are complexes of quasi coherent sheaves. In particular is a triangulated category. The Atiyah class is said to be equivariant if the morphism \(E\to E\otimes \mathbb{L}_X[1]\) can be lifted to the derived category of \(G\)-equivariant coherent sheaves. The first main theorem of the paper (Theorem A in the introduction and 55 in the text of the paper) has an astonishingly simple statement: \textit{if \(X\) admits a \(G\)-equivariant embedding in \(G\)-equivariant smooth scheme, then the Atiyah class of any lift of the quasi-coeherent sheaf \(E\) to \(D(QCoh(X)^G)\) (the deried category)is equivariant.} The author remarks that the assumption on \(X\) is satisfied as soon as, for instance, \(X\) is quasi-projective and carries at least one \(G\)-equivariant line bundle. The second main theorem (B in the introduction and 53 in the text) says that \textit{if \(G\) is an affine reductive algebraic group, \(Y\to B\) a \(G\)- invariant smooth projective family of varieties and \(M \to B\) a fine moduli space of simple perfect complexes on the fibres of \(Y\to B\), then the relative obstruction theory on \(M\to B\) is \(G\)-equivariant.} There is no need to warn the reader that to feel the intense flavor of Theorem B one should have a bit more than just some familiarity with the notion of obstruction theory, perfect obstruction theory and other important sophisticated deformations mathematical tools. It is fortunately easy to find excellent references about the above (just google: deformation theory to discover classical and modern books, papers and movies). Nevertheless the author recalling in a few words what a perfect obstruction theory is, providing precise references as well, and so friendly helping the readership. As an application of Theorem A and B, it is shown that, if \(G\) is reductive, the obstruction theory on the fine relative moduli space \(M\to B\) of simple perfect complexes on a \(G\)-invariant smooth projective family \(Y\to B\) is \(G\)- equivariant, This in turn suggests promising new methods for checking the equivariance of the natural obstruction theories on a wide variety of moduli spaces acted on by tori, as those occurring for instance in Donaldson-Thomas theory and Vafa-Witten theory. Just one last remark, before ending the review. Besides the beauty of the theorems, the paper is impressively well written, at the point that many readers, possibly including the reviewer, may even get the impression to substantially follow what is going on. This feeling is however probably comparable with that of somebody looking at a landscape from the window on board of an airplane. He/she may have a good panoramic view but missing the fine details down, on the ground, which are known only to a few masters. Like the author.