An analytic center cutting plane method to determine complete positivity of a matrix

Riley Badenbroek*, Etienne de Klerk

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

73 Downloads (Pure)

Abstract

We propose an analytic center cutting plane method to determine whether a matrix is completely positive and return a cut that separates it from the completely positive cone if not. This was stated as an open (computational) problem by Berman et al. [Berman A, Dur M, Shaked-Monderer N (2015) Open problems in the theory of completely positive and copositive matrices. Electronic 1. Linear Algebra 29(1):46-58]. Our method optimizes over the intersection of a ball and the copositive cone, where membership is determined by solving a mixed-integer linear program suggested by Xia et al. [Xia W, Vera JC, Zuluaga LF (2020) Globally solving nonconvex quadratic programs via linear integer programming techniques. INFORMS J. Comput 32(1):40-561 Thus, our algorithm can, more generally, be used to solve any copositive optimization problem, provided one knows the radius of a ball containing an optimal solution. Numerical experiments show that the number of oracle calls (matrix copositivity checks) for our implementation scales well with the matrix size, growing roughly like O(d(2)) for d x d matrices. The method is implemented in Julia and available at https://github.com/rileybadenbroek/CopositiveAnalyticCenter.jl. Summary of Contribution: Completely positive matrices play an important role in operations research. They allow many NP-hard problems to be formulated as optimization problems over a proper cone, which enables them to benefit from the duality theory of convex programming. We propose an analytic center cutting plane method to determine whether a matrix is completely positive by solving an optimization problem over the copositive cone. In fact, we can use our method to solve any copositive optimization problem, provided we know the radius of a ball containing an optimal solution. We emphasize numerical performance and stability in developing this method. A software implementation in Julia is provided.
Original languageEnglish
Pages (from-to)1115-1125
JournalINFORMS Journal on Computing
Volume34
Issue number2
DOIs
Publication statusPublished - Mar 2022

Keywords

  • analytic center cutting plane method
  • completely positive matrices
  • copositive optimization

Fingerprint

Dive into the research topics of 'An analytic center cutting plane method to determine complete positivity of a matrix'. Together they form a unique fingerprint.

Cite this