Global quadratic minimization over bivalent constraints: necessary and sufficient global optimality condition (Q415403)

This paper establishes global optimality conditions for quadractic optimization problems with quadratic equality and bivalent constraints. First, a necessary and sufficient condition is presented for a KKT point to be a global minimizer of a quadratic optimization problem with quadratic equality and bivalent constraints. From it, a necessary and sufficient condition for a feasible point to be a global minimizer of the same problem is derived. Then, situations are examined where this optimality condition is equivalent to checking the positive semidefiniteness of a related matrix, and so can be verified in polynomial time by using elementary eigenvalues decomposition techniques. As a consequence, simple sufficient global optimality conditions are presented, which can be verified by solving a linear matrix inequality problem. This extends several results in the existing literature.