Monday, May 20
4:50-6:50 PM
Salon A
MS12
Semidefinite Programming (Part I of II)
(Invited Minisymposium)
Semidefinite programming (SDP) is a generalization of linear programming in which the variable
space is the set of n x n real symmetric matrices, and the inequality constraint is a
semidefinite matrix constraint, written X _> 0. SDP is a convex optimization problem and it
has been known for some years that it enjoys much of the duality theory of LP. SDP has received
much recent attention because of the realization that the interior point methods developed for
LP are also largely applicable to SDP, and it has been consequently widely applied in application
areas ranging from combinatorial optimization to control theory. In this session, the speakers
will discuss various aspects of SDP, including both theoretical and algorithmic issues.
Organizer: Michael L. Overton
Courant Institute of Mathematical Sciences, New York University
- Using SDP for Non-Lipschitz Eigenvalue Optimization
- James V. Burke, University of Washington; and Michael L. Overton, Organizer
- Exploiting Sparsity in Interior-Point Methods for Semidefinite Programming
- Lieven Vandenberghe, Stanford University
- The Analytical Center, Optimal Face and Convergence of the Central Path of a
Semidefinite Program
- Katya Scheinberg, Columbia University
- Exact Duality Theories for Semidefinite Programming
- Motakuri Ramana, University of Florida
LMH, 3/15/96