The uniqueness of the adjoint operation. II (Q1097484)

Let H be a complex, finite-dimensional Hilbert space, and let L(H) denote the set of linear transformations mapping H into itself. For certain interesting subsets A(H) of L(H), the functions h:A(H)\(\to L(H)\) which have the properties \(h(ST)=h(T)h(S)\), h(S)S\(\geq 0\), and h(I)\(\neq 0\) are characterized. Examples: If \(A(H)=\{unitary\) transformations with determinant \(1\}\), then \(h(S)=S^*\) for all \(S\in A(H)\); if \(A(H)=\{non\sin gular\) transformations\(\}\), then either \(h(S)=\sigma (\det (S))S^*\) for all \(S\in A(H)\) or \(h(S)=\sigma (\det (S))S^{-1}\) for all \(S\in A(H)\), where \(\sigma\) is a multiplicative homomorphism from the nonzero complex numbers to the positive real numbers. [For part I see Integral Equations Oper. Theory 9, 623-634 (1986; Zbl 0608.47002).]