scientific article; zbMATH DE number 3930229

From MaRDI portal
Publication:3703183





Let \(x_ k(k=1,2,...,\ell)\) be vectors (the k-th one of dimension \(m_ k)\), satisfying the system of equations \(B_ 1x_ 1-C_ 1x_ 2=f_ 1\), \(-A_ kx_{k-1}+B_ kx_ k-C_ kx_{k+1}=f_ k\), \(k=2,3,...,l- 1\), \(-A^{\ell}x_{l-1}+B_ lx_ l=f_ l\) where \(B_ k\) are square monotonic tridiagonal matrices of \(m_ k\)-th order, \(A_ k\) and \(C_ k\) matrices with nonnegative elements. Suppose that the blockwise tridiagonal matrix M of the entire system has the properties of diagonal dominance and of monotonicity (i.e. \(M^{-1}\) is a matrix with nonnegative elements). Such systems may be obtained, for example, when using a 5-point difference approximation for elliptic-type boundary-value problems. The author proposes a class of iterative algorithms to solve such systems of linear algebraic equations with monotonic blockwise-tridiagonal matrices. The class contains - as particular cases - well known blockwise methods of Jacobi and Seidel, Buleev method of incomplete factorization and also the iterationless matrix progonka method.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3703183)