The spherical gap of operators (Q5961692)

For operators \(A\), \(B\) acting in a Hilbert space, the gap \(\theta(A, B)\) is defined by \(\theta(A,B)= \theta(G(A), G(B))\), where \(G(C)\) is the graph of the operator \(C\) and \[ \theta(M, N)= \max\{\sup_{x\in S(N)}\text{dist}(x,M), \sup_{y\in S(M)}\text{dist}(y, N)\} \] where \(S(N)\), \(S(M)\) are the unit spheres in the subspaces \(N\), \(M\). Equivalently \(\theta(A,B)= \| P_A- P_B\|\), where \(P_C\) is the projection operator onto \(G(C)\). The spherical gap \(\widetilde\theta(A, B)\) is defined by \(\widetilde\theta(A, B)= \widetilde\theta(G(A), G(B))\), where \[ \widetilde\theta(M, N)= \max\{\sup_{x\in S(N)}\text{dist}(x, S(M)), \sup_{y\in S(M)}\text{dist}(y, S(N))\} \] for subspaces \(M\), \(N\). The author determines a relationship between the gap and spherical gap, and gives a geometrical interpretation of the spherical gap.