scientific article; zbMATH DE number 3843938

From MaRDI portal
Publication:3314006





Let k be a commutative ring and A a k-algebra which is finitely generated and projective as a k-module. Dualizing the multiplication map and using the fact that \(A\otimes A^*=End_ kA\) yields a coassociative A-linear homomorphism \(\Delta:End_ kA\to End_ kA\otimes_ AEnd_ kA\). Let \(A\otimes \bar A\) denote the algebra extension of \(A\otimes A^{opp}\). Although \(End_ kA\otimes_ AEnd_ kA\) is in general not an algebra, one may restrict to a certain submodule \(End_ kA\times_ AEnd_ kA\) of \(End_ k A \otimes_ A End_ k A\) (which in this case is just the image of \(\Delta)\), which is an \(A\otimes \bar A\)-algebra for the component-wise multiplication and such that \(\Delta:End_ kA\to End_ kA\times_ AEnd_ kA\) is an \(A\otimes \bar A\) algebra morphism. We say that \(\Delta\) induces a bialgebra structure on \(End_ kA.\) As the author points out, it is clear that this construction does not work in general (i.e. A not necessarily finitely generated or projective). Yet there usually exist nice subalgebras B of End A which do possess a bialgebra structure \(\Delta:B\to B\times_ AB\), induced by the multiplication on A. These bialgebras in endomorphism rings were first studied by Sweedler in the commutative case and some of his results were later generalized by Takeuchi to the noncommutative case. The purpose of the present paper is to continue their study in the general case. Although bialgebras B in End A are defined by a comultiplication \(\Delta:B\to B\times_ AB\), there is sometimes another structural morphism \(S:B\to(B^{opp}\times B)^{opp}\), which comes into the picture. Indeed, if B is a bialgebra in End A with bijective \(\Delta\), then \(xA\) induces a monoid structure on the set of morphism classes of \(A\otimes \bar A\) algebras which are \(A\otimes \bar A\)-bimodule isomorphic to B. The group of units of this monoid is denoted by G(B). It was pointed out by Sweedler and Takeuchi that if B is a bialgebra in End A with bijective \(\Delta\) and S, which is flat as a left A-module and such that for all nonzero \(A\otimes \bar A\) subbimodules M of B we have \(M^{opp}\times_ AB\neq 0\), then A is a simple B-module if and only if B is a simple algebra if and only if any U representing a class in G(B) is simple. The author generalizes this by giving for any U in G(B) a canonical one-to-one correspondence between the ideals of U and the B- submodules of A (modulo some harmless assumptions on B, satisfied e.g. by all examples mentioned by Sweedler and Takeuchi). A nice class of bialgebras in End A contains the so called inductively defined ones. One starts from a direct system \(\{L_ i\}\) of A- submodules of \({}_ AA\otimes A_ A\) satisfying some characteristic assumptions and then one proves that the \(f\in End A\) satisfying \(L_ if=0\) for some index i form a bialgebra in End A. Conversely, if A is semisimple then one may show that any bialgebra in End A is inductively defined and if A is just a Dedekind domain, then a bialgebra in End A is inductively defined if and only if B is a saturated left A-submodule of End A. From this one may deduce that for A semisimple or a Dedekind domain, there exists a maximal bialgebra and a maximal bialgebra with S in End A. The author also generalizes the Jacobson-Bourbaki theorem for skew fields to the semisimple case as follows: a subalgebra B of End A which is finitely generated over A is of the form \(End_ UA\) for some subalgebra U of A if and only if B is a bialgebra in End A with S. In the last part of the text, the author studies so-called B-invariant modules. These are the objects of a monoidal category B-\underbar{Inv} and the morphism groups of B-\underbar{Inv} possess an A-bimodule structure, induced by the inclusion \(B-\underline{inv}\subset A-mod-A\). Functors which respect this A-bimodule structure are said to be linear. One may prove that (upto isomorphism) linear monoidal autoequivalences of B-\underbar{Inv} are of the form \(U\times_ A\)- for a so-called B- invertible algebra U. If we denote by \(G^*(B)\) the group of isomorphism classes of B-invertible algebras, then one may prove (i) that G and \(G^*\) are functorial; (ii) that the ideal submodule correspondences still hold for algebras U in \(G^*(B)\). These methods are then applied to derive some old and new results on Brauer groups and Azumaya algebras.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3314006)