A parallel structured banded DC algorithm for symmetric eigenvalue problems

被引：0
作者：
Shengguo Li
Xia Liao
Yutong Lu
Jose E. Roman
Xiaoqiang Yue
机构：
[1] National University of Defense Technology,College of Computer Science
[2] Sun Yatsen University,National Supercomputer Center in Guangzhou, and the School of Data and Computer Science
[3] Universitat Politècnica de València,D. Sistemes Informàtics i Computació
[4] Xiangtan University,Hunan Key Laboratory for Computation and Simulation in Science and Engineering, Key Laboratory of Intelligent Computing and Information Processing of Ministry of Education
来源：
CCF Transactions on High Performance Computing | 2023年 / 5卷
关键词：
ScaLAPACK; Divide-and-conquer; PSMMA; PBSDC; Distributed-memory parallel algorithm; 65F15; 68W10;
D O I：
暂无
中图分类号：
学科分类号：
摘要：
In this paper, a novel parallel structured divide-and-conquer (DC) algorithm is proposed for symmetric banded eigenvalue problems, denoted by PBSDC, which modifies the classical parallel banded DC (PBDC) algorithm by reducing its computational cost. The main tool that PBSDC uses is a parallel structured matrix multiplication algorithm (PSMMA), which can be much faster than the general dense matrix multiplication ScaLAPACK routine PDGEMM. Numerous experiments have been performed on Tianhe-2 supercomputer to compare PBSDC with PBDC and ELPA. For matrices with few deflations, PBSDC can be much faster than PBDC since computations are saved. For matrices with many deflations and/or small bandwidths, PBSDC can be faster than the tridiagonalization-based DC implemented in LAPACK and ELPA. However, PBSDC would become slower than ELPA for matrices with relatively large bandwidths.
引用
收藏
页码：116 / 128
页数：12
相关论文
共 128 条
[1]  
Ambikasaran S(2013)An J. Sci. Comput. 57 477-501
[2]  
Darve E(1992) fast direct solver for partial hierarchically semi-separable matrices Parallel Comput. 18 1105-1128
[3]  
Arbenz P(2011)Divide-and-conquer algorithms for the bandsymmetric eigenvalue problem Parallel Comput. 37 783-794
[4]  
Auckenthaler T(2007)Parallel solution of partial symmetric eigenvalue problems from electronic structure calculations ACM Trans. Math. Softw. 33 1-23
[5]  
Blum V(2000)A parallel symmetric block-tridiagonal divide-and-conquer algorithm ACM Trans. Math. Softw. 26 581-601
[6]  
Bungartz HJ(2000)A framework for symmetric band reduction ACM Trans. Math. Softw. 26 602-616
[7]  
Huckle T(1978)Algorithm 807: the SBR toolbox-software for successive band reduction Numer. Math. 31 31-48
[8]  
Johanni R(2005)Rank one modification of the symmetric eigenproblem SIAM J. Matrix Anal. Appl. 27 341-364
[9]  
Krämer L(2006)Some fast algorithms for sequentially semiseparable representation SIAM J. Matrix Anal. Appl. 29 67-81
[10]  
Lang B(1994)A fast solver for HSS representations via sparse matrices Concurr. Comput. Pract. Exper. 6 543-570
12345678910