A cookbook for finite element methods for nonlocal problems, including quadrature rules and approximate Euclidean balls

05/21/2020
by   Marta D'Elia, et al.
0

The implementation of finite element methods (FEMs) for nonlocal models with a finite range of interaction poses challenges not faced in the partial differential equations (PDEs) setting. For example, one has to deal with weak forms involving double integrals which lead to discrete systems having higher assembly and solving costs due to possibly much lower sparsity compared to that of FEMs for PDEs. In addition, one may encounter non-smooth integrands. In many nonlocal models, nonlocal interactions are limited to bounded neighborhoods that are ubiquitously chosen to be Euclidean balls, resulting in the challenge of dealing with intersections of such balls with the finite elements. We focus on developing recipes for the efficient assembly of FEM stiffness matrices and on the choice of quadrature rules for the double integrals that contribute to the assembly efficiency and also posses sufficient accuracy. A major feature of our recipes is the use of approximate balls, e.g., several polygonal approximations of Euclidean balls, that, among other advantages, mitigate the challenge of dealing with ball-element intersections. We provide numerical illustrations of the relative accuracy and efficiency of the several approaches we develop.

READ FULL TEXT
research
02/09/2018

GPU Accelerated Finite Element Assembly with Runtime Compilation

In recent years, high performance scientific computing on graphics proce...
research
02/15/2023

High performance implementation of 3D FEM for nonlocal Poisson problem with different ball approximation strategies

Nonlocality brings many challenges to the implementation of finite eleme...
research
04/08/2022

On the fast assemblage of finite element matrices with application to nonlinear heat transfer problems

The finite element method is a well-established method for the numerical...
research
08/09/2023

Finite Element Operator Network for Solving Parametric PDEs

Partial differential equations (PDEs) underlie our understanding and pre...
research
11/04/2019

Abstractions and automated algorithms for mixed domain finite element methods

Mixed dimensional partial differential equations (PDEs) are equations co...
research
03/18/2022

Convex Optimization-Based Structure-Preserving Filter For Multidimensional Finite Element Simulations

In simulation sciences, it is desirable to capture the real-world proble...
research
05/09/2018

MPI+X: task-based parallelization and dynamic load balance of finite element assembly

The main computing tasks of a finite element code(FE) for solving partia...

Please sign up or login with your details

Forgot password? Click here to reset