A general finite element preconditioning for the conjugate gradient method
conjugate gradient methodconvergencefinite elementincomplete Cholesky decompositionNeumann problemnumerical experimentspreconditioningreverse Cuthill-McKee algorithm
Boundary value problems for second-order elliptic equations (35J25) Iterative numerical methods for linear systems (65F10) Numerical computation of matrix norms, conditioning, scaling (65F35) Computational methods for sparse matrices (65F50) Finite element, Rayleigh-Ritz and Galerkin methods for boundary value problems involving PDEs (65N30)
Some preconditioning techniques like the (modified) incomplete Cholesky decomposition (MIC and IC) should not be applied to finite element discretizations of elliptic boundary value problems, when the FEM leads to a matrix which lacks the property M (f.i. quadratic triangular elements), because the existence of a preconditioning matrix cannot be proved then. The author introduces a method that constructs a preconditioning matrix for all finite element methods if only a mild condition for the node numbering is fulfilled: No global node should have a maximal number, that means, if all the neighbours of a node have a lower number, the node or one of its neighbours must be on the Dirichlet- type portion of the boundary. Such a numbering can be constructed using the reverse Cuthill-McKee algorithm starting with a node on the Dirichlet-type portion of the boundary. The main idea of the preconditioning method is to write the contribution of each finite element to the global matrix as a product of a lower triangular, a diagonal and an upper triangular matrix and assemble these matrices. The preconditioning matrix is defined as the product of the assembled global matrices. The idea can be generalized by assembling the sum of the factors of contributions of several finite elements to the global matrix. It is proved for a certain variant of the Cuthill-McKee algorithm, which fulfil a necessary and sufficient condition for the existence of the preconditioning matrix, that the smallest eigenvalue of this matrix is equal to 1. The application of the method to the (singular) Neumann problem is considered. The numerical experiments show, that the IC and MIC preconditioners give the fastest convergence of the conjugate gradient method in most cases, but for strange problems like a rhombus with vertex angle \(5\pi\) /6 the preconditioning matrix for the IC does not exist. The conclusion is that the MIC preconditioning for the conjugate gradient method is very effective even in the situation when the coefficient matrix is not an M-matrix. But the less efficient finite element preconditiong proposed by the author is robust and promising for higher order or non-symmetric problems.
- Numerical comparison of preconditionings for large sparse finite element problems
- On Preconditioning for Finite Element Equations on Irregular Grids
- A preconditioning technique based on element matrix factorizations
- Two-level hierarchically preconditioned conjugate gradient methods for solving linear elasticity finite element equations
- Preconditioning Second-Order Elliptic Operators: Experiment and Theory
- A practical termination criterion for the conjugate gradient method
- A preconditioning technique based on element matrix factorizations
- A survey of preconditioned iterative methods for linear systems of algebraic equations
- An Iterative Solution Method for Linear Systems of Which the Coefficient Matrix is a Symmetric M-Matrix
- Block Preconditioning for the Conjugate Gradient Method
- Comparative Analysis of the Cuthill–McKee and the Reverse Cuthill–McKee Ordering Algorithms for Sparse Matrices
- Efficient Implementation of a Class of Preconditioned Conjugate Gradient Methods
- Element Preconditioning Using Splitting Techniques
- Guidelines for the usage of incomplete decompositions in solving sets of linear equations as they occur in practical problems
- scientific article; zbMATH DE number 3854346 (Why is no real title available?)
- scientific article; zbMATH DE number 3686229 (Why is no real title available?)
- scientific article; zbMATH DE number 3720254 (Why is no real title available?)
- scientific article; zbMATH DE number 41268 (Why is no real title available?)
- scientific article; zbMATH DE number 176173 (Why is no real title available?)
- scientific article; zbMATH DE number 3596197 (Why is no real title available?)
- scientific article; zbMATH DE number 3892457 (Why is no real title available?)
- Methods of conjugate gradients for solving linear systems
- On recurring theorems on diagonal dominance
- On the rate of convergence of the preconditioned conjugate gradient method
- Preconditioned conjugate gradients for solving singular systems
- The rate of convergence of conjugate gradients
- A preconditioning technique based on element matrix factorizations
- Conjugate gradient type methods and preconditioning
- An element-by-element preconditioned conjugate gradient method implemented on a vector computer
- A new mixed preconditioning method for finite element computations
- Scalable preconditioned conjugate gradient inversion of vector finite element mass matrices
- Preconditioned conjugate gradient solvers for the generalized finite element method
- Large preconditioning symmetric finite volume element systems on triangular mesh
- scientific article; zbMATH DE number 4166704 (Why is no real title available?)
- A restarted version of a generalized preconditioned conjugate gradient method
- Preconditioned conjugate gradient method for finding minimal energy surfaces on Powell-Sabin triangulations
- On vectorizing the preconditioned generalized conjugate residual methods
- Multi-order initialization of CG solution of FE discretizations
- On Preconditioning for Finite Element Equations on Irregular Grids
- Extended preconditioners for the FETI method applied to constrained problems
- Preconditioning Complicated Finite Elements by Simple Finite Elements
- A coarse/fine preconditioner for very ill‐conditioned finite element problems
- Using the modified 2nd order incomplete Cholesky decomposition as the conjugate gradient preconditioning
- Preconditioned conjugate gradient method for the sparse generalized eigenvalue problem in electronic structure calculations
- Conjugate gradient solution of finite element equations on the IBM 3090 vector computer utilizing polynomial preconditionings
This page was built for publication: A general finite element preconditioning for the conjugate gradient method
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q582001)
