Computational and Applied Math Proseminar

Tuesday, December 6, 2005, 12:15 p.m. GWC 409

Hans Mittelmann

Dept. Math. & Stats.

Using State-of-the-Art NLP Solvers in System Identification and Beyond

Abstract Our work is lately predominantly in optimization and it has two substantial components, one of a service nature and one in interdisciplinary research. We will give an overview of the first and report in some detail about an ongoing research project in system identification. On the service side there are two major aspects of our work. We provide the two webpages which rank first in a Google search for "optimization software". The first is a guide to mostly "free-for-research" sofware, the second a collection of comparisons of such software. The other service we provide is one third of all interactive solvers for optimization problems accessible through the NEOS gateway (neos.mcs.anl.gov).

Starting with a CLAS interdiscplinary fellowship we established an interdiscplinary research program in the area of system identification as it is needed in chemical, electrical, and other engineering discplines. We will give an overview of the mathematical problems faced in this work and how we solved them computationally. This then naturally links to our knowledge about state-of-the-art optimization algorithms which are essential in this and, in turn, the problems we solve provide challenging tests for these algorithms.

For further information please contact: mittelmann@asu.edu