A full Nesterov-Todd step feasible interior-point algorithm for semidefinite optimization based on a new hyperbolic barrier function

Le résumé fourni par la source

This study concerns solving semidefinite programming (SDP) problems using a new kernel-based primal-dual interior-point method (IPM). We propose a parameterized kernel function (KF) that has a hyperbolic barrier term. Taking advantage of the exponential convexity property of the new KF, we prove that the corresponding algorithm has a complexity of order n logn logne O(√n · log n · log(n/ε)) for large-update methods. To the best of our knowledge, this is the first hyperbolic KF for SDP to reach the best-known iteration bound for such methods.Preliminary numerical experiments indicate that the new KF is efficient compared with other existing KFs in the literature.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.