TY - JOUR
T1 - A sparse matrix library with automatic selection of iterative solvers and preconditioners
AU - Sakurai, Takao
AU - Katagiri, Takahiro
AU - Kuroda, Hisayasu
AU - Naono, Ken
AU - Igai, Mitsuyoshi
AU - Ohshima, Satoshi
N1 - Funding Information:
This study was supported by the Seamless and Highly Productive Parallel Programming Environment for High-Performance Computing program of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan and by JSPS KAKENHI Grant Number 24300004. Authors appreciate Professors Shoji Itoh and Kengo Nakajima of the University of Tokyo for fruitful discussions on the design and development of OpenATLib and Xabclib.
PY - 2013
Y1 - 2013
N2 - Many iterative solvers and preconditioners have recently been proposed for linear iterative matrix libraries. Currently, library users have to manually select the solvers and preconditioners to solve their target matrix. However, if they select the wrong combination of the two, they have to spend a lot of time on calculations or they cannot obtain the solution. Therefore, an approach for the automatic selection of solvers and preconditioners is needed. We have developed a function that automatically selects an effective solver/preconditioner combination by referencing the history of relative residuals at runtime to predict whether the solver will converge or stagnate. Numerical evaluation with 50 Florida matrices showed that the proposed function can select effective combinations in all matrices. This suggests that our function can play a significant role in sparse iterative matrix computations.
AB - Many iterative solvers and preconditioners have recently been proposed for linear iterative matrix libraries. Currently, library users have to manually select the solvers and preconditioners to solve their target matrix. However, if they select the wrong combination of the two, they have to spend a lot of time on calculations or they cannot obtain the solution. Therefore, an approach for the automatic selection of solvers and preconditioners is needed. We have developed a function that automatically selects an effective solver/preconditioner combination by referencing the history of relative residuals at runtime to predict whether the solver will converge or stagnate. Numerical evaluation with 50 Florida matrices showed that the proposed function can select effective combinations in all matrices. This suggests that our function can play a significant role in sparse iterative matrix computations.
UR - http://www.scopus.com/inward/record.url?scp=84896984818&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84896984818&partnerID=8YFLogxK
U2 - 10.1016/j.procs.2013.05.300
DO - 10.1016/j.procs.2013.05.300
M3 - Conference article
AN - SCOPUS:84896984818
SN - 1877-0509
VL - 18
SP - 1332
EP - 1341
JO - Procedia Computer Science
JF - Procedia Computer Science
T2 - 13th Annual International Conference on Computational Science, ICCS 2013
Y2 - 5 June 2013 through 7 June 2013
ER -