Equation-oriented system: an efficient programming approach to solve multilinear and polynomial equations by the conjugate gradient algorithm

被引:14
作者
Wang, JH
Hopke, PK [1 ]
机构
[1] Clarkson Univ, Dept Chem Engn, Potsdam, NY 13699 USA
[2] Clarkson Univ, Dept Chem, Potsdam, NY 13699 USA
关键词
PARAFAC; bilinear model; trilinear model; polynomial equation; conjugate gradient (CG) algorithm; equation-oriented system (EOS);
D O I
10.1016/S0169-7439(00)00110-6
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
The factor analysis problem can be conceptualized as an expansion of polynomial equations that are solvable using least-squares methods. The equation-oriented system (EOS) is introduced as a method for solving polynomial equations using a preconditioned conjugate gradient (CG) algorithm for the normal equations, EOS is a fast, easy to program, low computer memory requirement method for accomplishing this task. EOS can be used to solve multilinear and PARAFAC problems. The practical aspects of implementing EOS in MATLAB are discussed. (C) 2001 Elsevier Science B.V. All rights reserved.
引用
收藏
页码:13 / 22
页数:10
相关论文
共 6 条
[1]  
[Anonymous], 1995, FRONTIERS APPL MATH
[2]   PARAFAC - PARALLEL FACTOR-ANALYSIS [J].
HARSHMAN, RA ;
LUNDY, ME .
COMPUTATIONAL STATISTICS & DATA ANALYSIS, 1994, 18 (01) :39-72
[3]  
*MATLAB, 1999, MATHW
[4]  
MUELLER FJ, 1969, ELEMENTS ALGEBRA
[5]   The multilinear engine -: A table-driven, least squares program for solving multilinear problems, including the n-way parallel factor analysis model [J].
Paatero, P .
JOURNAL OF COMPUTATIONAL AND GRAPHICAL STATISTICS, 1999, 8 (04) :854-888
[6]  
Saad Y., 1996, Iterative Methods for Sparse Linear Systems