An Implementation and Evaluation of the AMLS Method for SparseEigenvalue Problems

Journal Article · Tue Feb 14 00:00:00 EST 2006 · ACM Transactions on Mathematical Software

OSTI ID:929006

Gao, Weiguo; Li, Xiaoye S; Yang, Chao; Bai, Zhaojun

We describe an efficient implementation and present aperformance study of an algebraic multilevel sub-structuring (AMLS)method for sparse eigenvalue problems. We assess the time and memoryrequirements associated with the key steps of the algorithm, and compareitwith the shift-and-invert Lanczos algorithm in computational cost. Oureigenvalue problems come from two very different application areas: theaccelerator cavity design and the normal mode vibrational analysis of thepolyethylene particles. We show that the AMLS method, when implementedcarefully, is very competitive with the traditional method in broadapplication areas, especially when large numbers of eigenvalues aresought.

• Research Organization: Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United States)
• Sponsoring Organization: USDOE Director. Office of Science. Advanced ScientificComputing Research
• DOE Contract Number: DE-AC02-05CH11231
• OSTI ID: 929006
• Report Number(s): LBNL-57438; R&D Project: KS1210; BnR: KJ0101010; TRN: US0804069
• Journal Information: ACM Transactions on Mathematical Software, Vol. 34, Issue 4; Related Information: Journal Publication Date: 09/19/2007
• Country of Publication: United States
• Language: English

Related Subjects

99
ACCELERATORS
ALGORITHMS
DESIGN
EIGENVALUES
EVALUATION
IMPLEMENTATION
PERFORMANCE
POLYETHYLENES
multilevel substructuring method sparse eigenvalueproblems