On the diagonal approximation of full matrices (Q2564278)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | On the diagonal approximation of full matrices |
scientific article |
Statements
On the diagonal approximation of full matrices (English)
0 references
7 January 1997
0 references
For a given \(s\)-by-\(s\) matrix \(A\), the author is interested in constructing diagonal matrices \(D\) such that the spectral radius \(\rho(I-D^{-1}A)=0\), which is equivalent to \(\text{det} (D^{-1}A- \lambda I)=0\) having \(s\) unit zeros, i.e. \(\text{det} (D^{-1}A- \lambda I)=(1-\lambda)^s\). From the latter equation, one can derive \(s\) nonlinear equations for determining the \(s\) diagonal entries \(d_1,\dots,d_s\) of the diagonal matrices \(D^{-1}= \text{diag} (d_1,\dots, d_s)\). In the second part of the paper, this idea is applied to construct appropriate \(D\) matrices for the parallel diagonal-implicit iteration of Runge-Kutta methods, where \(s\) is here the number of stages of the Runge-Kutta method, and ranges from 2 to 8. In order to solve the above mentioned system of nonlinear equations, the author uses the computer algebra system Maple.
0 references
preconditioning
0 references
scaling
0 references
diagonal approximation of inverse matrices
0 references
parallel diagonal-implicit iteration of Runge-Kutta methods
0 references
computer algebra system Maple
0 references