Parallel numerical algorithms for simulation of multidimensional ill-posed nonlinear diffusion-advection-reaction models

Authors:
Raimondas Čiegis;Andrej Bugajev
Affiliations:
Vilnius Gediminas Technical University, Vilnius, Lithuania;Vilnius Gediminas Technical University, Vilnius, Lithuania
Venue:
PARA'12 Proceedings of the 11th international conference on Applied Parallel and Scientific Computing
Year:
2012

Citing 2
Cited 0

Introduction to parallel computing: design and analysis of algorithms

Introduction to parallel computing: design and analysis of algorithms
Robust numerical methods for taxis-diffusion-reaction systems: Applications to biomedical problems

Mathematical and Computer Modelling: An International Journal

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents finite difference approximations of two dimensional in space mathematical model of a bacterial self-organization. Due to the chemotaxis process some instability of the solution can be developed in the system, in this paper we show that such instability can be connected to the ill-posed problem defined by the backward in time diffusion process. The ADI and splitting type methods are used to construct robust parallel numerical approximations. Domain decomposition method is applied to distribute subtasks among processors. The scalability analysis of the parallel algorithm is done and results of computational experiments are presented.