Solving Large-scale Eigenvalue Problems in SciDac Applications

Chao Yang, LBNL

Large-scale eigenvalue problems arise from several DOE applications. In this talk, I will describe the state-of-art algorithms currently being used to tackle some SciDac applications. I will also discuss our latest progress in developing new algorithms to meet a variety of different application needs. In particular, I will focus on the development of algebraic sub-structuring techniques in the context of linear accelerator design and the use of a special constrained minimization algorithm for solving nonlinear eigenvalue problems arising from electronic structure calculation.