No label defined (Q3022244)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 2182372
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 2182372 |
Statements
23 June 2005
0 references
generalized conjugate residual method
0 references
generalized minimal residual method
0 references
variable preconditioning
0 references
inner-loop and outer-loop
0 references
large sparse liner system
0 references
Helmholtz equation
0 references
stationary diffusion equation
0 references
numerical examples
0 references
finite difference method
0 references
iterative method
0 references
0.8459805
0 references
0.8458841
0 references
0.83927006
0 references
0.7952776
0 references
0 references
0.7654452
0 references
0.76297605
0 references
The authors propose a variant of variable preconditioning for Generalized Conjugate Residual (GCR)-like methods. For solving the large and sparse system \(Ax=b\) by a Krylov subspace method, instead of using a fix preconditioner the authors propose to obtain an approximate solution of the linear system \(Az=r\), where \(r\) is the residual. They prove that if the approximation of the solution satisfies the inequality \(\| r_k - Az_k \| _2 \leq \theta_k \| r_k \| _2\), where \(0<\theta_k<1\), then the norms of the residuals form a decreasing sequence. Then they propose to use as stopping criteria \(\| r_{k+1} - Az_{k+1}^{(l)}\| _2 /\| r_{k+1} \| _2 \leq \delta\) for the preconditioning step, if it is solved by a method that computes the residual or if the preconditioning step is solved by an stationary iterative method (like Gauss-Seidel, SOR, \dots) they propose to use \(\| z_{k+1}^{(l)} - z_{k+1}^{(l-1)}\| _\infty /\| z_{k+1}^{(l)} \| _\infty \leq \delta\). In any case the number of iterations of the inner loop can not exceed a maximum. The differences between the proposed Variable Preconditioning GCR (VPGCR) method, the FGMRES method proposed by \textit{Y.\ Saad} [SIAM J.\ Sci.\ Stat.\ Comput. 14, No. 2, 461--469 (1993; Zbl 0780.65022)] and GMRESR proposed by \textit{H.\ A.\ van der Vorst} and \textit{C.\ Vuik} [Numer.\ Linear Algebra Appl. 1, 369--386 (1994; Zbl 0839.65040)] are summarized. Restarted versions of these methods and CGR are compared on two problems. The VPGCR algorithm is preconditioned using the SOR method, and the other using ILU(0) and ILU(1).
0 references