Galerkin Projection Methods for Solving Multiple Linear Systems
The paper is devoted to solve multiple linear systems \(A^{(i)}x^{(i)}=b^{(i)}\), \(1 \leq i \leq s\), with different coefficient matrices and different right-hand sides. Two important practical situations are considered, namely, the eigenvectors of the coefficient matrices are the same and the coefficient matrices differ by rank-1 or rank-2 matrices. The authors propose a methodology to solve these related multiple systems effectively. They focus on the seed projection method, where one seed system is selected and solved by the conjugate gradient (CG) method. Then a Galerkin projection of the residuals onto the Krylov subspace, generated by the seed system, is performed to obtain approximate solutions for the unsolved systems. These approximate solutions are then refined by the CG method. The process is repeated until all the systems are solved. A theoretical error bound is given for the approximation obtained from a projection process onto the Krylov subspace generated from solving a previous linear system. Numerical examples of multiple linear systems arising from image restorations and recursive least squares approximations are given to illustrate the effectiveness of the method.
- Analysis of Projection Methods for Solving Linear Systems with Multiple Right-Hand Sides
- MINRES seed projection methods for solving symmetric linear systems with multiple right-hand sides
- Block projection methods for linear systems
- A block seed method for solving linear systems with multiple right-hand sides
- Global least squares method (Gl-LSQR) for solving general linear systems with several right-hand sides
- A projection method for solving nonsymmetric linear systems on multiprocessors
- Restarted Hessenberg method for solving shifted nonsymmetric linear systems
- MINRES seed projection methods for solving symmetric linear systems with multiple right-hand sides
- A FEM for an optimal control problem subject to the fractional Laplace equation
- An automatic Krylov subspaces recycling technique for the construction of a global solution basis of non-affine parametric linear systems
- Numerical investigation of ensemble methods with block iterative solvers for evolution problems
- A local character based method for solving linear systems of radiation diffusion problems
- Solving shifted linear systems with restarted GMRES augmented with error approximations
- Accelerated GCRO-DR method for solving sequences of systems of linear equations
- An accelerated subspaces recycling strategy for the deflation of parametric linear systems based on model order reduction
- Fast Algorithms for Hyperspectral Diffuse Optical Tomography
- Analysis of Projection Methods for Solving Linear Systems with Multiple Right-Hand Sides
- Multisplitting for regularized least squares with Krylov subspace recycling.
- A survey of subspace recycling iterative methods
- A rational filtering algorithm for sequences of shifted symmetric linear systems with applications to frequency response analysis
- Block cross-interactive residual smoothing for Lanczos-type solvers for linear systems with multiple right-hand sides
- Optimal fourth-order staggered-grid finite-difference scheme for 3D frequency-domain viscoelastic wave modeling
This page was built for publication: Galerkin Projection Methods for Solving Multiple Linear Systems
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4702413)
