Wednesday, July 12
MS39
Aspects of High-Level Problem Representations for Optimization
10:30 AM-12:30 PM
Rio Mar 5
Sponsored by SIAM Activity Group on Optimization
This session considers several ways of dealing with high-level problem representations, i.e., representations that focus on essential problem details while hiding implementation details. Often several implementations are possible. Representations that hide implementation details permit the people who formulate problems to focus on correct formulations and facilitate experimenting with alternative implementations. The speakers in this session will discuss
several kinds of high-level representations, often in the context of optimization algorithms, such as representations for linear-algebraic operators, functional information about PDE solutions, "logical' optimization constraints, and nonlinear expressions.
Organizers: David M. Gay
Bell Laboratories, Lucent Technologies, USA
Paul T. Boggs
Sandia National Laboratories, USA
- 10:30-10:55 An Object Framework for High-Level Representation of Objective Functions and Constraints in PDE-Based Optimization
- Kevin R. Long, Sandia National Laboratories, USA; Paul T. Boggs, Organizer; Leslea J. Lehoucq and William E. Mason, Sandia National Laboratories, USA
- 11:00-11:25 A Flexible Implementation of an SQP Algorithm
- Paul T. Boggs, Organizer; and Kevin R. Long, Sandia National Laboratories, USA
- 11:30-11:55 Extending Algebraic Modeling Languages for Combinatorial Optimization via Constraint Programming
- Robert Fourer, Northwestern University, USA; and David M. Gay, Organizer
- 12:00-12:25 Toward Convenient yet Efficient Handling of Nonlinear Expressions
- David M. Gay, Organizer