We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

math.OC

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Mathematics > Optimization and Control

Title: Completely Positive Factorization by Riemannian Smoothing Method

Abstract: Copositive optimization is a special case of convex conic programming, and it optimizes a linear function over the cone of all completely positive matrices under linear constraints. Copositive optimization provides powerful relaxations of NP-hard quadratic problems or combinatorial problems, but there are still many open problems regarding copositive or completely positive matrices. In this paper, we focus on one of such open problems; finding a completely positive (CP) factorization for a given completely positive matrix. We treat it as a nonsmooth Riemannian optimization, i.e., a minimization of a nonsmooth function over the Riemannian manifolds. To solve this problem, we present a general smoothing framework for nonsmooth Riemannian optimization and guarantee convergence to a stationary point of the original problem. An advantage is that we can implement it quickly with minimal effort by directly using the existing standard smooth Riemannian solvers, such as Manopt. Numerical experiments show the efficiency of our method especially for large-scale CP factorizations.
Comments: 18 pages
Subjects: Optimization and Control (math.OC)
Report number: Department of Policy and Planning Sciences Discussion Paper Series No. 1377, University of Tsukuba
Cite as: arXiv:2107.01538 [math.OC]
  (or arXiv:2107.01538v3 [math.OC] for this version)

Submission history

From: Akiko Yoshise [view email]
[v1] Sun, 4 Jul 2021 04:41:42 GMT (373kb,D)
[v2] Tue, 6 Jul 2021 01:15:01 GMT (373kb,D)
[v3] Sun, 12 Dec 2021 04:02:41 GMT (23kb)

Link back to: arXiv, form interface, contact.