Proving lower bounds on a minimization problem over positive semidefinite matrices with a bounded maximal rank

18 Views Asked by At

Given a minimization optimization problem of a linear target function over the set of positive semidefinite matrices of some fixed maximal rank, subject to affine constraints, what are (analytical) methods to find lower bounds on the optimal value of that problem? If we had no rank constraints, the answer would be "convex optimization duality", but the rank constraint makes the problem non-convex.