The modified alternating group explicit (M.A.G.E.) method (Q1108741): Difference between revisions
From MaRDI portal
Added link to MaRDI item. |
Removed claim: author (P16): Item:Q579832 |
||
Property / author | |||
Property / author: David J. Evans / rank | |||
Revision as of 18:07, 19 February 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | The modified alternating group explicit (M.A.G.E.) method |
scientific article |
Statements
The modified alternating group explicit (M.A.G.E.) method (English)
0 references
1988
0 references
The authors describe an extension of the classical A.D.I. algorithm of Peaceman and Rachford for the solution of a system of linear equations (*) \(Au=b\). The present method is an extension of one which they presented at the 11th IMACS' World Congress, see pp. 213-220 of Vol. I of the Proceedings (1985). The equations (*) are assumed to have arisen from the standard discretization of the two point boundary value problem \(-U''+qU=f\), \(q\geq 0\), defined on a finite interval with Dirichlet boundary conditions. The method is an iteration scheme of the form \((\bar G_ 1+rI)u^{(k+1/2)}=b_ 1-(\bar G_ 1-rI)u^{(k)},\) \((\bar G_ 2+rI)u^{(k+1)}=b_ 2-(\bar G_ 2-rI)u^{(k+1/2)},\) where \(b=b_ 1+b_ 2\). \(\bar G_ 1\), and \(\bar G_ 2\) are block diagonal matrices with, at most, (2\(\times 2)\) blocks on the diagonal. It is shown that \(u^{(k)}\to y\) and \(u^{(k+1/2)}\to z\) as \(k\to \infty\) and the solution \(u=y+z.\) A rather complicated example is examined in detail. There is a brief discussion of the computational advantages of M.A.G.E. of S.O.R. and their previous method. However, no comparisons are made with the standard A.D.I. method or with the direct methods.
0 references
modified alternating group explicit method
0 references
tridiagonal linear system
0 references
finite difference equation
0 references
explicit solution
0 references
block system
0 references
Peaceman- Rachford algorithm
0 references
alternating direction implicit method
0 references
successive overrelaxation
0 references
A.D.I. algorithm
0 references