On numerical testing of the regularity of semidefinite problems

Abstract

This paper is devoted to study regularity of Semidefinite Programming (SDP) problems. Current methods for SDP rely on assumptions of regularity such as constraint qualifications and wellposedness. Absence of regularity may compromise characterization of optimality and algorithms may present numerical difficulties. Prior that solving problems, one should evaluate the expected efficiency of algorithms. Therefore, it is important to have simple procedures that verify regularity. Here we use an algorithm to test regularity of linear SDP problems in terms of Slater’s condition. We present numerical tests using problems from SDPLIB and compare our results with those from others available in literature.