Title :
Experimental Study on Parallel Methods for Solving Systems of Equations
Author :
Maruster, S. ; Negru, Viorel ; Mafteiu-Scai, L.O.
Author_Institution :
Comput. Sci. Dept., West Univ. of Timisoara, Timisoara, Romania
Abstract :
The paper presents experimental results on parallel variants of some classical methods for solving systems of equations. The following four methods are studied: Newton method, Chebyshev method, Gradient method with Fridman control sequence, a method of conjugate gradient type. The main conclusion is that the precondition of the sparse systems (both linear and nonlinear) improves in a great extent the performance of the parallel algorithms.
Keywords :
Chebyshev approximation; Newton method; conjugate gradient methods; nonlinear equations; parallel algorithms; Chebyshev method; Fridman control sequence; Newton method; conjugate gradient method; nonlinear systems; parallel algorithms; parallel methods; sparse systems; systems of equations; Chebyshev approximation; Convergence; Equations; Gradient methods; Jacobian matrices; Mathematical model; Newton method; Nonlinear equations; parallel methods;
Conference_Titel :
Symbolic and Numeric Algorithms for Scientific Computing (SYNASC), 2012 14th International Symposium on
Conference_Location :
Timisoara
Print_ISBN :
978-1-4673-5026-6
DOI :
10.1109/SYNASC.2012.73
