An inexact primal-dual path following algorithm for convex quadratic SDP (Q995786): Difference between revisions

This is a follow-up to the paper [\textit{K. C. Toh, R. H. Tütüncü, M. J. Todd}, Pac. J. Optim. 3, No. 1, 135-164 (2007; Zbl 1136.90026)] where the authors described interior-point algorithms for solving convex quadratic semidefinite programming (QSDP) problems, under a definiteness assumption on the quadratic objective function. In the paper under review, the author allows the objective function to be positive semidefinite (instead of positive definite), therefore covering problems such as the nearest Euclidean distance matrix problem. Under the assumption that the linear map appearing in the constraints is sparse or structured, the author focuses on preconditioning techniques applied to the large-scale linear system to be solved at each iteration of the interior-point algorithm. The study is illustrated with an extensive set of numerical experiments.