On the Solution of Block-Tridiagonal Systems Arising from Certain Finite-Difference Equations
- 1 October 1972
- journal article
- Published by JSTOR in Mathematics of Computation
- Vol. 26 (120) , 859-868
- https://doi.org/10.2307/2005868
Abstract
We consider the solution of the linear systems arising from certain implicit finite-difference approximations to systems of linear differential equations. In particular, we consider those schemes which lead to matrices of block-tridiagonal form. There are two common methods for solving such equations: using a block-tridiagonal factorization (blocksolve), or treating the matrix as a band matrix (bandsolve). First, we discuss conditions for ensuring the numerical stability of the block-tridiagonal factorization for general matrices of this form. Then, we compare the two methods for general block-tridiagonal matrices (including matrices arising from the Crank-Nicolson scheme for systems of parabolic equations) and for a more specialized block-tridiagonal matrix which arises from schemes of H. B. Keller for systems of two-point boundary value problems and parabolic equations.Keywords
This publication has 8 references indexed in Scilit:
- Nested Dissection of a Regular Finite Element MeshSIAM Journal on Numerical Analysis, 1973
- A NEW DIFFERENCE SCHEME FOR PARABOLIC PROBLEMS**This work was supported by the U. S. Army Research Office, Durham, under Contract DAHC 04-68-C-0006.Published by Elsevier ,1971
- Accurate Difference Methods for Linear Ordinary Differential Systems Subject to Linear ConstraintsSIAM Journal on Numerical Analysis, 1969
- Difference Methods for Initial-Value ProblemsMathematics of Computation, 1968
- Theoretical Numerical AnalysisMathematics of Computation, 1967
- The Theory of Matrices in Numerical Analysis.The American Mathematical Monthly, 1966
- Block diagonally dominant matrices and generalizations of the Gerschgorin circle theoremPacific Journal of Mathematics, 1962
- Error Analysis of Direct Methods of Matrix InversionJournal of the ACM, 1961