Block SOR preconditioned projection methods for Kronecker structured Markovian representations

被引:11
作者
Buchholz, P [1 ]
Dayar, T
机构
[1] Univ Dortmund, D-44221 Dortmund, Germany
[2] Bilkent Univ, Dept Comp Engn, TR-06800 Ankara, Turkey
[3] Tech Univ Dresden, Dresden, Germany
关键词
Markov chains; Kronecker structured numerical techniques; block SOR; preconditioning; projection methods; real Schur factorization; COLAMD ordering;
D O I
10.1137/S1064827503425882
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
Kronecker structured representations are used to cope with the state space explosion problem in Markovian modeling and analysis. Currently, an open research problem is that of devising strong preconditioners to be used with projection methods for the computation of the stationary vector of Markov chains (MCs) underlying such representations. This paper proposes a block successive overrelaxation (BSOR) preconditioner for hierarchical Markovian models (HMMs1) that are composed of multiple low-level models and a high-level model that defines the interaction among low-level models. The Kronecker structure of an HMM yields nested block partitionings in its underlying continuous-time MC which may be used in the BSOR preconditioner. The computation of the BSOR preconditioned residual in each iteration of a preconditioned projection method becomes the problem of solving multiple nonsingular linear systems whose coefficient matrices are the diagonal blocks of the chosen partitioning. The proposed BSOR preconditioner solves these systems using sparse LU or real Schur factors of diagonal blocks. The fill-in of sparse LU factorized diagonal blocks is reduced using the column approximate minimum degree (COLAMD) ordering. A set of numerical experiments is presented to show the merits of the proposed BSOR preconditioner.
引用
收藏
页码:1289 / 1313
页数:25
相关论文
共 44 条
[21]  
Fernandes P, 1998, RAIRO-RECH OPER, V32, P325
[22]   On the Use of Two QMR Algorithms for Solving Singular Systems and Applications in Markov Chain Modeling [J].
Freund, Roland W. ;
Hochbruck, Marlis .
NUMERICAL LINEAR ALGEBRA WITH APPLICATIONS, 1994, 1 (04) :403-420
[23]   A TRANSPOSE-FREE QUASI-MINIMAL RESIDUAL ALGORITHM FOR NON-HERMITIAN LINEAR-SYSTEMS [J].
FREUND, RW .
SIAM JOURNAL ON SCIENTIFIC COMPUTING, 1993, 14 (02) :470-482
[24]  
Greenbaum A., 1997, Iterative methods for solving linear systems
[25]   Numerical analysis of superposed GSPNs [J].
Kemper, P .
IEEE TRANSACTIONS ON SOFTWARE ENGINEERING, 1996, 22 (09) :615-628
[26]  
LANGVILLE A, 2002, THESIS N CAROLINA ST
[27]   Kronecker product approximate preconditioner for SANs [J].
Langville, AN ;
Stewart, WJ .
NUMERICAL LINEAR ALGEBRA WITH APPLICATIONS, 2004, 11 (8-9) :723-752
[28]  
Migallon V, 1996, NUMER LINEAR ALGEBR, V3, P413, DOI 10.1002/(SICI)1099-1506(199609/10)3:5<413::AID-NLA91>3.0.CO
[29]  
2-S
[30]   ANALYSIS OF A KANBAN DISCIPLINE FOR CELL COORDINATION IN PRODUCTION LINES .2. STOCHASTIC DEMANDS [J].
MITRA, D ;
MITRANI, I .
OPERATIONS RESEARCH, 1991, 39 (05) :807-823