Endomorphisms of Koszul complexes: formality and application to deformation theory (Q2174825)

The theme of the article under review is the interplay between the absence of obstructions of a deformation problem and the homotopy abelianity of its associated differential graded Lie algebra (DGLA). The authors observe that homotopy abelianity of a DGLA \(L\), i.e. to have a zig-zag of morphisms between \(L\) and some abelian DGLA \(M\) all of which are quasi-isomorphisms, is a notion which is too restrictive. Therefore they shift attention to the weaker notion of \textit{numerically homotopy abelianity} for a DGLA \(L\), which means that there is a zig-zag of morphisms leading to an abelian DGLA \(M\) such that the morphisms pointing in direction of \(L\) are surjective in cohomology, while the morphisms pointing towards \(M\) are injective in cohomology. Observe that this still implies that the bracket of a numerically homotopy abelian DGLA is trivial in cohomology. Carocci and Manetti show that over a general commutative ring, there exist numerically homotopy abelian DGLAs which are not homotopy abelian. The endomorphism DGLA of a Koszul complex of a regular sequence can serve here as an example (except in trivial cases). On the other hand, over a field of characteristic zero, both notions are equivalent. The main goal of the article is to use these considerations in order to locate the obstructions to the deformation problem of a line bundle over a locally complete intersection in a projective scheme and to have a sufficient condition when the associated DGLA is homotopically abelian over a field of characteristic zero. The authors have a section about the relation of their work to the derived setting, abouting to a sufficient condition of homotopically abelianity also in the derived setting.