Decomposition of two functions in the orthogonality equation (Q291275)

Suppose that \(H\) and \(K\) are two Hilbert spaces and \(f,g:H\to K\) are two unknown functions such that they give a solution of the generalization orthogonality equation \(\langle f(x), g(y)\rangle=\langle x, y\rangle\) for each \(x, y\in H\). \textit{J. Chmieliński} [Aequationes Math. 90, No. 1, 11--23 (2016; Zbl 1342.39023)] proved some properties of \(f\) and \(g\). In this paper, the authors show that \(f\) and \(g\) satisfy the equation above if and only if there exist suitable closed subspaces \(M_1,M_2,M_3\subset K\) such that \(K=M_1+M_2+M_3,M_k\perp M_j\) for \(k\neq j\) and \(f\) and \(g\) can be written as the following decomposition \(f=A+\phi, \quad g=(A^*)^{-1}+\gamma\) for some invertible bounded linear mapping \(A:H\to M_1\) and for some mappings \(\phi:H\to M_2\) and \(\gamma :H\to M_3\).