Most algorithms for optimization problems and nonlinear equations require the solution
of linear systems of equations. In many applications, the linear systems are very large
while in others the coefficient matrix is not explicitly available. These difficulties
suggest the use of iterative solvers.
The goal of this minisymposium is to present new ideas concerning the use of iterative
methods in several algorithmic frameworks for solving optimization problems and nonlinear
equations. The presentations given here will address practical as well as theoretical
issues concerning this topic. Some of these issues are the use of Krylov subspace methods,
preconditioning, limited memory Quasi Newton updates, and inexactness.
Organizers: Amr S. El-Bakry, Alexandria University, Egypt; and
Luis N. Vicente, Rice University
- Truncated-Newton Methods for Large-Scale Optimization
- Steven G. Nash, George Mason University
- Preconditioning of Elliptic Variational Inequalities
- Tony Choi, North Carolina State University
- Newton-Krylov Methods
- Homer F. Walker, Utah State University
- Computational Experiments with Iterative Solvers for Primal-Dual Interior-Point
Methods in Nonlinear Programming
- Amr S. El-Bakry, Organizer
LMH, 3/15/96