Hopf algebras and subfactors associated to vertex models (Q1273950)

Let \(V\), \(W\) be finite-dimensional linear spaces over a field \(k\). We consider the equation \[ (t\otimes id) u^{-1}= ((t\otimes id)u)^{-1},\tag{\(*\)} \] where \(u\in{\mathcal L}(V)\otimes{\mathcal L}(W)\) is an invertible element and \(t:{\mathcal L}(V)\to{\mathcal L}(V^*)\) is the transposition, which first appeared in a work by V. F. R. Jones. In particular, if \(H\) is a Hopf algebra whose square of the antipode is the identity, \(v\in{\mathcal L}(V)\otimes H\) is a corepresentation, and if \(\pi: H\to{\mathcal L}(W)\) is a representation, then \(u= (id\otimes \pi)v\) is a solution of the equation \((*)\). Using a universal construction, one can show that every solution \(u\) of \((*)\) is obtained in this way. A more refined construction leads to a triple \((H,v,\pi)\), which is minimal in a certain sense, and such that \(u= (id\otimes \pi)v\). The aim of the present paper is to study this latter type of construction of Hopf algebras. When \(u\) is unitary, the author finds a \(C^*\)-norm on \(H\) and describes, in a new manner, the standard invariant of the subfactor associated to \(u\). The ``twisted'' case is also discussed.