Symmetric coalgebras. (Q1887596)

The authors define the notion of a symmetric coalgebra, study such coalgebras, and describe Hopf algebras with symmetric coalgebra structure. They work over a field \(k\). A coalgebra \(C\) is called symmetric if there is an injective morphism \(\alpha\colon C\to C^*\) of \((C^*,C^*)\)-bimodules. Since \(C\) is left (right) co-Frobenius if and only if there exists an injective morphism of left (right) \(C^*\)-modules from \(C\) to \(C^*\), then clearly a symmetric coalgebra is left and right co-Frobenius. If \(C\) is cocommutative, then the notions of symmetric and co-Frobenius coincide. An equivalent condition for \(C\) to be symmetric is the existence of a symmetric, \(C^*\)-balanced, nondegenerate bilinear form \(B\colon C\times C\to k\). A third equivalent condition for \(C\) to be symmetric involves the multiplication on \(C\) induced by the Casimir element and is an analogue for the condition that a finite dimensional algebra \(A\) is symmetric if and only if there is a \(k\)-linear map \(f\colon A\to k\) such that \(f(xy)=f(yx)\) and \(\text{Ker}(f)\) contains no non-zero left ideals. A cosemisimple coalgebra is symmetric, but there exist infinite dimensional symmetric coalgebras which are not cosemisimple. Any semiperfect coalgebra \(C\) can be embedded in a symmetric coalgebra; this result dualizes the fact that any finite dimensional algebra is isomorphic to a quotient of a symmetric algebra. Also for \(C\) a semiperfect coalgebra, and \(G,H\) functors defined by \(G\colon{\mathcal M}^C\to\text{mod-}C^*\), where \(G(M)=\Hom_{C^*}({_{C^*}M},{_{C^*}C_{C^*}})\) and \(H\colon{\mathcal M}^C\to\text{mod-}C^*\) where \(H(M)=\Hom_{C^*}({_{C^*}M},{_{C^*}C^*_{C^*}})\), then \(C\) is symmetric if and only if \(G,H\) are naturally equivalent. Finally the authors consider Hopf algebras which are symmetric as coalgebras and prove that a Hopf algebra \(H\) is symmetric as a coalgebra if and only if \(H\) is unimodular and \(S^{*2}\) is an inner automorphism of \(H^*\) where \(S\) is the antipode of \(H\). The proofs in this section use the Nakayama automorphism defined on a co-Frobenius coalgebra \(C\) via \(B\), a nondegenerate \(C^*\)-balanced bilinear form as above; the Nakayama automorphism is a ring automorphism of the ring \(C\) where \(C\) has multiplication defined via the Casimir element, i.e. via the multiplication on \(\text{Rat}(C^*)\).