A class of modified block SSOR preconditioners for symmetric positive definite systems of linear equations (Q1282329): Difference between revisions

The author considers second-order elliptic problems and finite element method systems with hierarchical bases. The matrix \(A= A^*>0\) has the block form \([A_{ij}]\) \((1\leq i,j\leq N)\), where \(N-1\) relates to the number of grid refinements. Convergence of a modified block symmetric successive overrelaxation (SSOR) iteration with the relaxation parameter is investigated under the condition that \(D\leq \alpha^2A\) where \(D= [A_{ii}]\) is the block-diagonal part of \(A\). Methods with inner iterations for the systems with \(D\) are also considered. It should be noted that even for the two-dimensional case \(\alpha\) depends on \(N\); it means that the iterative methods apply model operators \(B\) (preconditioners) which are almost spectrally equivalent to \(A\).