Minimax Theory and its Applications 04 (2019), No. 2, 329--339
Copyright Heldermann Verlag 2019
An Elementary Proof for the Exact Relaxation for Rank One Moment Matrices in Multi-Polynomial SOS Relaxation
Shenglong Hu
Department of Mathematics, School of Science, Hangzhou Dianzi University, Hangzhou 310018, China
shenglonghu@hdu.edu.cn
We present an elementary proof for the fact that an optimal rank one moment matrix in the multi-polynomial SOS relaxation gives an exact relaxation. This fact is a fundamental result in multi-polynomial SOS relaxation method for the class of multi-polynomial optimization problems. The multi-polynomial SOS relaxation method is designed by exploring the special structures of the class of multi-polynomial optimization problems, which has the advantage for giving an SDP with size about half of that for the classical SOS relaxation in the general formulation.
Keywords: Multi-polynomial, multi-form, SOS, exact relaxation.
MSC: 15A18, 15A42, 15A69
[ Fulltext-pdf (110 KB)] for subscribers only.