EECS Publication
Performance of Parallel Eigensolvers on Electronic Structure Calculations
Robert C. Ward, Yihua Bai and Justin Pratt
Many models employed to solve problems in quantum mechanics, such as electronic structure calculations, result in nonlinear eigenproblems. The solution to these problemstypically involves iterative schemes requiring the solution of a large symmetric linear eigenproblem during each iteration. This paper evaluates the performance of various popular and new parallel symmetric linear eigensolvers applied to the Self-Consistent Field procedure in electronic structure calculations on the distributed memory supercomputers at the OakRidge National Laboratory. Results using established routines from ScaLAPACK and vendor optimized packages, are presented, as well as from two recently developed parallel eigensolvers, the method of Multiple Relatively Robust Representations using PLAPACK support routines and the block divide-and-conquer algorithm.
Published 2005-02-01 05:00:00 as ut-cs-05-560 (ID:162)