  


Inexact primaldual pathfollowing algorithms for a special class of convex quadratic SDP and related problems
K. C. Toh (mattohkcnus.edu.sg) Abstract: We propose a primaldual pathfollowing Mehrotratype predictorcorrector method for solving convex quadratic semidefinite programming (QSDP) problems. For the special case when the quadratic term has the form $\frac{1}{2} X \bul (UXU)$, we compute the search direction at each iteration from the Schur complement equation. We are able to solve the Schur complement equation efficiently via the preconditioned symmetric quasiminimal residual iterative solver with two appropriately constructed preconditioners. Numerical experiments on a variety of QSDPs with matrices of dimensions up to 2000 are performed and the computational results show that our methods are efficient and robust. Our methods can also be extended to linear SDP problems with upper bound constraints on primal matrix variables. Keywords: semidefinite programming, semidefinite least squares, pathfollowing methods, NesterovTodd scaling, symmetric quasiminimum residual iteration Category 1: Linear, Cone and Semidefinite Programming (Semidefinite Programming ) Citation: Technical Report 1421, School of Operations Research and Industrial Engineering, Cornell University, Ithaca, NY 14853, February 2005. Download: [PDF] Entry Submitted: 03/11/2005 Modify/Update this entry  
Visitors  Authors  More about us  Links  
Subscribe, Unsubscribe Digest Archive Search, Browse the Repository

Submit Update Policies 
Coordinator's Board Classification Scheme Credits Give us feedback 
Optimization Journals, Sites, Societies  