A parallel structured banded DC algorithm for symmetric eigenvalue problems

被引：2
|
作者：
Li, Shengguo [1 ]
Liao, Xia [1 ]
Lu, Yutong [2 ,3 ]
Roman, Jose E. [4 ]
Yue, Xiaoqiang [5 ]
机构：
[1] Natl Univ Def Technol, Coll Comp Sci, Changsha 410073, Peoples R China
[2] Sun Yatsen Univ, Natl Supercomp Ctr Guangzhou, Guangzhou 510006, Peoples R China
[3] Sun Yatsen Univ, Sch Data & Comp Sci, Guangzhou 510006, Peoples R China
[4] Univ Politecn Valencia, D Sistemes Informat & Comp, Cami Vera S-N, Valencia 46022, Spain
[5] Xiangtan Univ, Hunan Key Lab Computat & Simulat Sci & Engn, Key Lab Intelligent Comp & Informat Proc, Minist Educ, Xiangtan 411105, Peoples R China
来源：
CCF TRANSACTIONS ON HIGH PERFORMANCE COMPUTING | 2023年 / 5卷 / 02期
关键词：
ScaLAPACK; Divide-and-conquer; PSMMA; PBSDC; Distributed-memory parallel algorithm; RANK-ONE MODIFICATION; CONQUER ALGORITHM; DIVIDE; FRAMEWORK;
D O I：
10.1007/s42514-022-00117-9
中图分类号：
TP3 [计算技术、计算机技术];
学科分类号：
0812 ;
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：13
相关论文
共 50 条
  • [11] A parallel Lanczos method for symmetric generalized eigenvalue problems
    Wu, Kesheng
    Simon, Horst
    Computing and Visualization in Science, 1999, 2 (01): : 37 - 46
  • [12] An efficient parallel algorithm for the symmetric tridiagonal eigenvalue problem
    Forjaz, MA
    Ralha, R
    VECTOR AND PARALLEL PROCESSING - VECPAR 2000, 2001, 1981 : 369 - 379
  • [13] A class of inverse eigenvalue problems for real symmetric banded matrices with odd bandwidth
    Li, Jicheng
    Dong, Liqiang
    Li, Guo
    LINEAR ALGEBRA AND ITS APPLICATIONS, 2018, 541 : 131 - 162
  • [14] A PARALLEL CONTINUATION ALGORITHM FOR NONLINEAR EIGENVALUE PROBLEMS
    CHEN, MJ
    CHEN, TZ
    CHIEN, CS
    YU, MS
    COMPUTERS & MATHEMATICS WITH APPLICATIONS, 1994, 27 (07) : 83 - 97
  • [15] A Parallel Method for the Banded Eigenvalue Problem
    Arabian Journal for Science and Engineering, 22 (2-C):
  • [16] A parallel method for the banded eigenvalue problem
    Li, KY
    Li, TY
    ARABIAN JOURNAL FOR SCIENCE AND ENGINEERING, 1997, 22 (2C) : 31 - 40
  • [17] Toward a parallel solver for generalized complex symmetric eigenvalue problems
    Schabauer, Hannes
    Pacher, Christoph
    Sunderland, Andrew G.
    Gansterer, Wilfried N.
    ICCS 2010 - INTERNATIONAL CONFERENCE ON COMPUTATIONAL SCIENCE, PROCEEDINGS, 2010, 1 (01): : 437 - 445
  • [18] AN OVERVIEW OF PARALLEL ALGORITHMS FOR THE SINGULAR VALUE AND SYMMETRIC EIGENVALUE PROBLEMS
    BERRY, M
    SAMEH, A
    JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS, 1989, 27 (1-2) : 191 - 213
  • [19] USING PARALLEL BANDED LINEAR-SYSTEM SOLVERS IN GENERALIZED EIGENVALUE PROBLEMS
    ZHANG, H
    MOSS, WF
    PARALLEL COMPUTING, 1994, 20 (08) : 1089 - 1105
  • [20] An efficient parallel algorithm design for symmetric triple diagonal eigenvalue
    Chen, Changhua
    Li, Bin
    2011 International Conference on Electrical and Control Engineering, ICECE 2011 - Proceedings, 2011, : 3816 - 3819
12345