Approximation algorithms for homogeneous polynomial optimization with quadratic constraints (Q607501)

Constrained maximization of a generic polynomial function \(F(x^1,\dots, x^d)\), \(x^k\in\mathbb{R}^{n(k)}\) for all \(k=1,\dots, d,\) is studied. Two types of constraints are considered: (1) Euclidean spherical constraints of the form \(\| x^k\|= 1\), for all \(k= 1,\dots, d\); (2) general ellipsoidal constraints of the form \((x^k)^TQ^k_{i_k}\,x^k\leq 1\), \(k= 1,\dots, d\), \(i_k= 1,\dots, m_k\). Algorithms for solving the optimization problems are described, a comparison with related papers in the literature is presented. Numerical results are reported illustrating the effectiveness of the proposed approximation algorithms.