\(\varGamma \)-commuting triples of positive definite matrices and midpoint equations on unitary matrices (Q2479516)

A triple of positive definite matrices is called \(\Gamma\)-commuting if it can be transformed to a commuting triple of positive definite matrices by a congruence transformation. In this paper, \(\Gamma\)-commuting triples of positive definite matrices are described and characterized by means of a variety of equivalences in a general algebraic and geometric framework. The set \({\mathcal S}_{A,B}\) of positive definite matrices \(X\) of which the triple \( [X,A,B] \) is \(\Gamma\)-commuting is studied for two given positive definite matrices \(A\) and \(B\). The solution set \({\mathcal S}_{A,B}\) forms a geodesic submanifold of the Riemannian symmetric space. Also, a family of unitary matrices \(V(A,B)\) generating the special unitary group \(SU(n)\) is considered. The associated midpoint equation on this generating set is the one of the form \(V(A,B) = V(A,X)V(X,B)\), an analog of the matrix space case. The solution set of the midpoint equation coincides with \({\mathcal S}_{A,B}\) in the \(\Gamma\)-commuting problem. This is obtained by an equivalence between the midpoint equation and the geometric mean equation \((A \# X)X^{-1}(X \# B)=A \# B\), where \(A \# B\) denotes the geometric mean of \(A\) and \(B\).