Tuesday, May 21
12:45-2:15 PM
Salon A
MS22
Methods for Large Scale Nonlinear Programming (Invited Minisymposium)
Nonlinear Programming is the problem of minimizing a nonlinear function subject to nonlinear constraints. It is important in a wide range of applications in which it permits much more realistic models to be built.
Interest in solution methods centers around SQP methods, augmented Lagrangian methods and interior point methods. The first two rely on good QP software, but can be very effective on small and medium to large problems. Interior point methods may be better able to avoid combinatorial difficulties in finding the correct active set on very large problems.
The speakers will outline current research issues relating to how the first two methods can be made to work well for large systems and will discuss the structure and stability of systems that arise in the Interior point approach.
Organizer: Roger Fletcher,
University of Dundee, Scotland
- Large Issues in Large Scale SQP
- Roger Fletcher, Organizer
- Limited Memory Methods for Constrained Optimization
- Jorge Nocedal and Ciyou Zhu, Northwestern University; and Todd Plantenga, Sandia National Laboratories
- Stability of Systems Arising in Interior Methods for Constrained Optimization
- Anders Forsgren, Royal Institute of Technology, Sweden; Philip R. Gill and Joseph R. Shinnerl, University of California, San Diego
MEM, 3/18/96