Trace ratio Maximization involving Complex matrices

Hii Everyone, I am trying to maximize the function V_opt= Tr(V^H*A*V)/Tr(V^H*B*V) subject to a constraint Tr(V^H*V)=1 where A and B are Hermitian symmetric matrices and are known. Can anyone help me with this

Have you been able to formulate a convex or quasi-convex optimization problem? I don’t see what such a formulation would be.

I will assume this problem is non-convex, unless you show otherwise. Perhaps YALMIP could be used with a local or global optimization solver.

1 Like