A fast numerical homogenization algorithm for finite element analysis
Cholesky factorizationelimination treefinite elementmatrix re-orderingnumerical homogenizationparallel computationrecursive triangulation algorithmSchur's complement techniquesshallow waters
Boundary value problems for second-order elliptic equations (35J25) Finite element, Rayleigh-Ritz and Galerkin methods for boundary value problems involving PDEs (65N30) Water waves, gravity waves; dispersion and scattering, nonlinear interaction (76B15) Homogenization applied to problems in fluid mechanics (76M50)
One classical way to solve problems governed by partial differential equations with coefficients that vary arbitrarily in space is to use homogenization techniques. This paper proposes a fast and flexible alternative which consists to homogenize the discrete problem instead of the original continuous problem. This method uses a recursive triangulation algorithm to produce a mesh which fits the homogenization needs; next, a standard finite element discretization on this mesh gives the global rigidity matrix. And then, the ``numerical homogenization is performed on this matrix by using Schur's complement techniques. It is worth to mention that this method can be well adapted to parallel computation. Some examples from shallow waters illustrate the efficiency of the method, step by step. This method works for two-dimensional problems discretized by using triangles. It should be possible to extend it to three-dimensional problems.
- A fast algorithm for reiterated homogenization
- Fast homogenization algorithm based on asymptotic theory and multiscale schemes
- A parallel solver for adaptive finite element discretizations
- An efficient zooming method for finite element analysis
- Fast integration and weight function blending in the extended finite element method
- The best-approximation weighted-residuals method for the steady convection-diffusion-reaction problem
- Multiscale modeling using goal-oriented adaptivity and numerical homogenization. I: Mathematical formulation and numerical results
- Multiscale modeling using goal-oriented adaptivity and numerical homogenization. II: Algorithms for the Moore-Penrose pseudoinverse
- An anisotropic unstructured triangular adaptive mesh algorithm based on error and error gradient information
This page was built for publication: A fast numerical homogenization algorithm for finite element analysis
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4953253)