Preconditioning for diffusion problem with small size high resolution inclusions

2018 ◽  
Vol 33 (4) ◽  
pp. 243-251
Author(s):  
Yuri A. Kuznetsov
Keyword(s):  
Solution Method ◽  
Diffusion Problem ◽  
Lanczos Method ◽  
Point System ◽  
High Contrast ◽  
Finite Element Discretization ◽  
Element Discretization ◽  
Preconditioning Technique ◽  
Saddle Point System ◽  
Diffusion Problems

Abstract In this paper, we propose and investigate a new preconditioning technique for diffusion problems with multiple small size high contrast inclusions. The inclusions partitioned into two groups. In the first group inclusions the value of diffusion coefficient can be very small, and in the inclusions of the second group it can be very large. The classical P1 finite element discretization is converted in the special algebraic saddle point system. The solution method combines elimination of the DOFs from the first group of inclusions with the Preconditioned Lanczos method with block diagonal preconditioner for the rest of the DOFs. Condition number estimates for the proposed preconditioner are given.

New homogenization method for diffusion equations

2018 ◽  
Vol 33 (2) ◽  
pp. 85-93 ◽  
Author(s):  
Yuri A. Kuznetsov
Keyword(s):  
Saddle Point ◽  
Diffusion Coefficients ◽  
Diffusion Problem ◽  
Homogenization Method ◽  
Diffusion Equations ◽  
Point System ◽  
Algebraic Problem ◽  
Saddle Point System ◽  
Saddle Point Matrix ◽  
Diffusion Problems

Abstract In this paper, we propose and investigate a new homogenization method for diffusion problems in domains with multiple inclusions with large values of diffusion coefficients. The diffusion problem is approximated by the P1-finite element method on a triangular mesh. The underlying algebraic problem is replaced by a special system with a saddle point matrix. For the solution of the saddle point system we use the typical asymptotic expansion. We prove the error estimates and convergence of the expanded solutions. Numerical results confirm the theoretical conclusions.

Diagonal PCG Method for FE Solution of 3D Biot's Consolidation Problem

2013 ◽  
Vol 838-841 ◽  
pp. 718-721
Author(s):  
Kun Yong Zhang ◽  
Gui Heng Xie
Keyword(s):  
Linear Systems ◽  
Large Scale ◽  
Solution Method ◽  
Preconditioned Conjugate Gradient ◽  
Finite Element Discretization ◽  
Element Discretization ◽  
Direct Solution Method ◽  
Indefinite Linear Systems ◽  
Pcg Method ◽  
Biot’S Consolidation

To solve large symmetric indefinite linear systems in finite element discretization of 3D Biot's consolidation equations.This paper adopted diagonal preconditioned conjugate gradient method to FE program. Several numerical examples show that the diagonal PCG method are significantly more efficient than direct solution method for large-scale symmetric indefinite linear systems.

Quantum Hall effect in a saddle-point system

1998 ◽  
Vol 2 (1-4) ◽  
pp. 523-526
Author(s):  
M.V Budantsev ◽  
Z.D Kvon ◽  
A.G Pogosov ◽  
E.B Olshanetskii ◽  
D.K Maude ◽  
...  
Keyword(s):  
Saddle Point ◽  
Hall Effect ◽  
Quantum Hall Effect ◽  
Quantum Hall ◽  
Point System ◽  
Saddle Point System

scholarly journals Reduced basis approximations of the solutions to spectral fractional diffusion problems

2020 ◽  
Vol 28 (3) ◽  
pp. 147-160
Author(s):  
Andrea Bonito ◽  
Diane Guignard ◽  
Ashley R. Zhang
Keyword(s):  
Computational Cost ◽  
Fractional Power ◽  
Reaction Diffusion ◽  
Fractional Diffusion ◽  
Diffusion Problem ◽  
Quadrature Point ◽  
Reduced Basis ◽  
Fast Evaluation ◽  
Bottle Neck ◽  
Diffusion Problems

AbstractWe consider the numerical approximation of the spectral fractional diffusion problem based on the so called Balakrishnan representation. The latter consists of an improper integral approximated via quadratures. At each quadrature point, a reaction–diffusion problem must be approximated and is the method bottle neck. In this work, we propose to reduce the computational cost using a reduced basis strategy allowing for a fast evaluation of the reaction–diffusion problems. The reduced basis does not depend on the fractional power s for 0 < smin ⩽ s ⩽ smax < 1. It is built offline once for all and used online irrespectively of the fractional power. We analyze the reduced basis strategy and show its exponential convergence. The analytical results are illustrated with insightful numerical experiments.

ON A RESCALED NONLOCAL DIFFUSION PROBLEM WITH NEUMANN BOUNDARY CONDITIONS

2021 ◽  
Vol 10 (8) ◽  
pp. 3013-3022
Author(s):  
C.A. Gomez ◽  
J.A. Caicedo
Keyword(s):  
Boundary Conditions ◽  
Existence And Uniqueness ◽  
Neumann Boundary ◽  
Diffusion Problem ◽  
Neumann Boundary Conditions ◽  
Nonlocal Diffusion ◽  
Banach Fixed Point Theorem ◽  
Positive Continuous Function ◽  
Normalization Constant ◽  
Diffusion Problems

In this work, we consider the rescaled nonlocal diffusion problem with Neumann Boundary Conditions \[ \begin{cases} u_t^{\epsilon}(x,t)=\displaystyle\frac{1}{\epsilon^2} \int_{\Omega}J_{\epsilon}(x-y)(u^\epsilon(y,t)-u^\epsilon(x,t))dy\\ \qquad \qquad+\displaystyle\frac{1}{\epsilon}\int_{\partial \Omega}G_\epsilon(x-y)g(y,t)dS_y,\\ u^\epsilon(x,0)=u_0(x), \end{cases} \] where $\Omega\subset\mathbb{R}^{N}$ is a bounded, connected and smooth domain, $g$ a positive continuous function, $J_\epsilon(z)=C_1\frac{1}{\epsilon^N}J(\frac{z}{\epsilon}), G_\epsilon(x)=C_1\frac{1}{\epsilon^N}G(\frac{x}{\epsilon}),$ $J$ and $G$ well defined kernels, $C_1$ a normalization constant. The solutions of this model have been used without prove to approximate the solutions of a family of nonlocal diffusion problems to solutions of the respective analogous local problem. We prove existence and uniqueness of the solutions for this problem by using the Banach Fixed Point Theorem. Finally, some conclusions are given.

Sign in / Sign up

Export Citation Format

Share Document