Published September 1977 | Version v1
Report Open

The ICCG method for the iterative solution of large sparse linear systems

  • 1. Atomic Energy Board, Pelindaba, Pretoria (South Africa). Physics Div.

Description

A modified version of the conjugate gradient method (CG) for the iterative solution of large sparse linear systems is described. The method is first applied to symmetric positive definite matrices. Preceding the application of CG by an incomplete Choleski decomposition which is computationally inexpensive, has the effect of improving the convergence of CG considerably thereby rendering the algorithm computationally attractive. An extention to non-symmetric matrices as well as computer routines for the symmetric case are given

Availability note (English)

MF available from INIS under the Report Number.

Abstract (Afrikaans)

'n Gewysigde vorm van die toegevoegde gradient-algoritme (TG) vir die iteratiewe oplossing van groot yl stelsels lineere vergelykings word beskryf. Die algoritme word eers op simmetriese positief defnitiewe matrikse toegepas. Deurdat die toepassing van TG voorafgegaan word deur 'n onvolledige Choleski-ontbinding wat relatief min rekenwerk vereis, word die konvergensie van TG aansienlik versnel en die algoritme ekonomies t.o.v. rekenkoste. 'n Uitbreiding na die nie-simmetriese geval asook 'n rekenaarimplementering van die simmetriese geval word gegee

Files

8344340.pdf

Files (443.1 kB)

Name Size Download all
md5:5b7c7f167c444739cb90046cbc756d5c
443.1 kB Preview Download

Additional details

Publishing Information

ISBN
0 86960 665 4
Imprint Pagination
15 p.
Report number
PER--19

INIS

Country of Publication
South Africa
Country of Input or Organization
South Africa
INIS RN
8344340
Subject category
S99: GENERAL AND MISCELLANEOUS;
Descriptors DEI
COMPUTER CODES; DIFFERENTIAL EQUATIONS; EQUATIONS; ITERATIVE METHODS; MATRICES; NUMERICAL SOLUTION; SERIES EXPANSION; VECTORS
Descriptors DEC
TENSORS