Title: Riemannian Optimization and Riemannian Trust-Region Methods

Speaker: Christopher G. Baker, Sandia National Laboratories

Date/Time: Monday, January 7, 2007, 1:00 pm - 2:00 pm

Location: Bldg. 895/Auditorium

Brief Abstract: Riemannian optimization is concerned with the optimization of functions defined on Riemannian manifolds. Current research activity involves the transfer of theory and methods from Euclidean optimization to a Riemannian setting, as well as the description of efficient methods for large scale Riemannian optimization. This talk will briefly survey the field, highlighting some differences between Riemannian and Euclidean optimization. The Riemannian Trust-Region method will be presented as an example for some of the challenges involved in porting classical optimization methods to Riemannian manifolds. Problems from numerical linear algebra will provide application examples.

CSRI POC: Scott Collis, (505) 284-1123