An extension of sums of squares relaxations to polynomial optimization problems over symmetric cones (Q879967): Difference between revisions

The authors consider a polynomial optimization problem (POP) over a symmetric cone in an \(N\)-dimensional Euclidean Jordan algebra with compact feasible set. Besides the (standard) polynomial optimization problem over the nonnegative orthant in the \(N\)-dimensional Euclidean space, the POP considered in this paper includes, as very important particular cases, the polynomial semidefinite programming (SDP) problem and the polynomial second order cone programming (SOCP) problem. The authors proof a lemma concerning the representation of a positive polynomial on the nonempty and compact feasible set of POP in terms of a specific cone of polynomials. Based on that lemma, a sum of squares (SOS) relaxation for POP is proposed to generate a sequence of SOS optimization problems whose optimal objective values converge to the optimal value of POP. Several SOS and SDP relaxations of POP based on a dual approach, and SDP relaxations based on a primal approach are derived. Also, convergence of optimal values of the relaxations to the optimal value of POP and some relationships among the relaxations are shown. The paper contains a useful brief introduction to Euclidean Jordan algebras, symmetric cones and Euclidean Jordan algebra-valued polynomials.