Computational and Applied Math Proseminar
joint with
Mathematics Colloquium

Department of Mathematics, Arizona State University

Thursday, March 2, 2000, 3:40 p.m. in PSA Room 103

Stephen Wright

Mathematics and Computer Science Division, Argonne National Labs

Solving Large Optimization Problems on Computational Grids

Abstract We discuss our experiences in implementing algorithms for very large optimization problems that run on grid computing platforms made up of a loosely connected, distributed, heterogeneous network of workstations. Such platforms have the advantage that they make use of idle cycles that are otherwise wasted and are powerful and essentially free, in contrast to a conventional supercomputer. By nature, however, grid platforms are difficult to use, so much of our effort has gone into developing the software frameworks that allow algorithms to be implemented effectively. We discuss the optimization problems that we have addressed so far, outline the algorithms and software that we use to solve them, and present some current results. (This research has been carried out under the metaNEOS project, a collaborative effort involving researchers at Argonne, Northwestern, Wisconsin, and Columbia.)

For further information please contact: mittelmann@asu.edu