Multi-p Preconditioners
\(p\)-version of the finite element methodcomparison of methodsconjugate gradient methodmulti-\(p\) preconditionermultigrid methodnumerical examples
Boundary value problems for second-order elliptic equations (35J25) Iterative numerical methods for linear systems (65F10) Numerical computation of matrix norms, conditioning, scaling (65F35) Finite element, Rayleigh-Ritz and Galerkin methods for boundary value problems involving PDEs (65N30) Multigrid methods; domain decomposition for boundary value problems involving PDEs (65N55)
In analogy with the multigrid method, the authors present a solution method for problems based on the \(p\)-version of the finite element method whereby degrees of freedom associated with approximation order \(p\) are visited at once and separately from those of approximation order \(p-1\) and \(p+1\). The authors present the method in an algebraic setting, independently of the finite element formulation from which the system arises. They present three variations on the theme: a straightforward ``V-cycle, a nested variation in which lower order levels are visited more frequently than higher order levels, and a combined approach with the nested variation used initially, followed by the V-cycle. Multi-\(p\) methods can be used to precondition the conjugate gradient iterative method. Indeed, the authors remark that an arbitrary preconditioner matrix \(M\) for the conjugate gradient method can be related to a smoothing matrix in a multi-\(p\) method, and the resulting method can be used as a preconditioner for the conjugate gradient method. They go on to demonstrate that this preconditioner reduces the condition number of the iteration and improves its convergence rate. One commercially available computer program employs a particular choice of hierarchical shape functions based on quadrilateral finite elements, solving the resulting system with a direct solver. An alternative approach to solution involves eliminating the so-called internal degrees of freedom using the Schur complement and solving the resulting system by preconditioned conjugate gradients. In this paper, the authors compare the direct solution method, preconditioning by the SSOR method as implemented in ITPACK, and preconditioning by the multi-\(p\) method presented above. Numerical results show multi-\(p\) preconditioning to be superior to the other two methods. The numerical results also indicate that the number of iterations required for solution does not increase as \(p\) increases, in contrast to other available methods.
- Multi-grid solver for the p-version of the FEM.
- Multi-P Methods: Iterative Algorithms for the P-Version of the Finite Element Analysis
- scientific article; zbMATH DE number 4155809
- Solution techniques for the p‐version of the adaptive finite element method
- scientific article; zbMATH DE number 1225872
- Efficient preconditioning for the discontinuous Galerkin finite element method by low-order elements
- A parallel multi-p method
- Algebraic multigrid for higher-order finite elements
- Performance enhancement of the multi-p preconditioner
- Multi-level \(hp\)-adaptivity: high-order mesh adaptivity without the difficulties of constraining hanging nodes
- An improved multigrid solver for the \(p\)-hierarchical basis finite element method using a space decomposition smoother
- Multi-grid solver for the p-version of the FEM.
- Comparison of multigrid algorithms for high-order continuous finite element discretizations.
- Solution techniques for the p‐version of the adaptive finite element method
- Multi-P Methods: Iterative Algorithms for the P-Version of the Finite Element Analysis
This page was built for publication: Multi-p Preconditioners
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4376250)