Asymptotic behavior of Helmberg-Kojima-Monteiro (HKM) paths in interior-point methods for monotone semidefinite linear complementarity problems: general theory

C. K. Sim, G. Zhao

Research output: Contribution to journalArticlepeer-review

Abstract

An interior-point method (IPM) defines a search direction at each interior point of the feasible region. These search directions form a direction field, which in turn gives rise to a system of ordinary differential equations (ODEs). Thus, it is natural to define the underlying paths of the IPM as the solutions of the system of ODEs. In Sim and Zhao (Math. Program. Ser. A, [2007], to appear), these off-central paths are shown to be well-defined analytic curves and any of their accumulation points is a solution to the given monotone semidefinite linear complementarity problem (SDLCP). Off-central paths for a simple example are also studied in Sim and Zhao (Math. Program. Ser. A, [2007], to appear) and their asymptotic behavior near the solution of the example is analyzed. In this paper, which is an extension of Sim and Zhao (Math. Program. Ser. A, [2007], to appear), we study the asymptotic behavior of the off-central paths for general SDLCPs using the dual HKM direction. We give a necessary and sufficient condition for when an off-central path is analytic as a function of μ √ at a solution of the SDLCP. Then, we show that, if the given SDLCP has a unique solution, the first derivative of its off-central path, as a function of μ √ , is bounded. We work under the assumption that the given SDLCP satisfies the strict complementarity condition.
Original languageEnglish
Pages (from-to)11-25
JournalJournal of Optimization Theory and Applications
Volume137
Issue number1
DOIs
Publication statusPublished - Apr 2008

Keywords

  • semidefinite linear complementarity problems
  • ordinary differential equations
  • off-central paths
  • HKM direction
  • analyticity

Fingerprint

Dive into the research topics of 'Asymptotic behavior of Helmberg-Kojima-Monteiro (HKM) paths in interior-point methods for monotone semidefinite linear complementarity problems: general theory'. Together they form a unique fingerprint.

Cite this