Monday, May 10

Recent Advances in Direct Search Methods

10:45 AM-12:45 PM
Room: Atlanta 4

Direct search methods, which do not use or explicitly estimate a function's derivatives, have been used to solve a wide variety of real-world applications. The application of these methods is motivated by their simplicity and problem characteristics such as derivative is unknown and expensive derivative estimation and function calculation that is noisy or contains physical inaccuracies. Additionally, many direct search methods can exploit parallel computational resources. The presentations in this minisymposium will describe new advances in our understanding of the theory and application of direct search methods.

Organizer: William E. Hart
Sandia National Laboratories, Albuquerque

10:45-11:10 Direct Search Methods and Approximate Gradients
C. T. Kelley, North Carolina State University
11:15-11:40 Exploiting the Full Flexibility of Pattern Search
UpdatedElizabeth Dolan and Virginia Torczon, College of William & Mary
11:45-12:10 An Asynchronous Parallel Direct Search Algorithm for Nonlinear Optimization
Patricia D. Hough, Sandia National Laboratories, Livermore; Tamara G. Kolda, Oak Ridge National Laboratory; and Virginia Torczon, College of William & Mary
12:15-12:40 Convergence Properties of Evolutionary Methods for Nonlinear Optimization
William E. Hart, Organizer

OP99 Home


Program Updates

Speaker Index




MMD, 3/26/99