Deformation theory via deviations (Q1340005): Difference between revisions

The present paper was inspired by Drinfeld's introduction and study of quasi-Hopf algebras. This is an object \(A\) akin to an ordinary Hopf algebra but where the comultiplication is only associative up to a conjugacy by an element \(\Phi\in A\otimes A\otimes A\). This is exactly the structure which is needed to turn the category of modules over \(A\) into a monoidal one. Drinfeld's proof of the existence of quasi- triangular quasi-Hopf algebras starting with initial ``classical'' data was reminiscent of arguments in the ordinary deformation theory of algebras, but there was notably lacking an appropriate complex controlling of the relevant theory. The authors give a suitable generalization of the cobar resolution for coalgebras to describe a cohomology theory related to deformations of Drinfeld algebras with an arbitrary \(\Phi\), leaving the multiplication and comultiplication fixed. The classical approach is step-by-step integration of ``partial'' deformations \(\overline {\Phi}= \Phi+ t\Phi_ 1+ \dots+ t^ n \Phi_ n\). So the authors work with diagrams of \(k [[t ]]\)-modules, which are commutative modulo \(t^{n+1}\) and consider deviations from commutativity modulo \(t^{n+2}\). They develop the deviation calculus for such two- dimensional diagrams which is based on the fact that the deviations are ``added'' under pasting of diagrams. Application of the deviation calculus makes more visual the result of the Gerstenhaber-Schack theory that the obstruction to the integrability of a partial deformation of an algebra (bialgebra) is an element of a certain third cohomology group. The analogous result for Drinfeld algebras is obtained. This work should be regarded as parallel to the work of \textit{S. Shnider} and \textit{S. Sternberg} [J. Algebra 169, No. 2, 343-366 (1994; Zbl 0818.17016)]. See also the article of \textit{M. Markl} and \textit{S. Shnider} [Cohomology of Drinfeld algebras: a general nonsense approach, q- alg/9505002 e-print archive q-alg\@eprints.math.duke.edu] for further progress in this subject.