A local radial basis function differential quadrature semi-discretisation technique for the simulation of time-dependent reaction-diffusion problems. Issue 6 (20th January 2021)

• Record Type: Journal Article
• Authors: Jiwari, Ram; Gerisch, Alf
• Is Part Of: Engineering computations. Volume 38:Issue 6(2021)
• Page Start: 2666
• Page End: 2691
• Publication Date: 2021-01-20
• Subjects: Grey–Scott model -- Meshfree method -- Turing pattern -- Numerical method -- Partial differential equation; Computer-aided engineering -- Periodicals; Computer graphics -- Periodicals; 620.00285
• DOI: 10.1108/EC-05-2020-0291
• Languages: English
• ISSNs: 0264-4401

Abstract

Abstract : Purpose: This paper aims to develop a meshfree algorithm based on local radial basis functions (RBFs) combined with the differential quadrature (DQ) method to provide numerical approximations of the solutions of time-dependent, nonlinear and spatially one-dimensional reaction-diffusion systems and to capture their evolving patterns. The combination of local RBFs and the DQ method is applied to discretize the system in space; implicit multistep methods are subsequently used to discretize in time. Design/methodology/approach: In a method of lines setting, a meshless method for their discretization in space is proposed. This discretization is based on a DQ approach, and RBFs are used as test functions. A local approach is followed where only selected RBFs feature in the computation of a particular DQ weight. Findings: The proposed method is applied on four reaction-diffusion models: Huxley's equation, a linear reaction-diffusion system, the Gray–Scott model and the two-dimensional Brusselator model. The method captured the various patterns of the models similar to available in literature. The method shows second order of convergence in space variables and works reliably and efficiently for the problems. Originality/value: The originality lies in the following facts: A meshless method is proposed for reaction-diffusion models based on local RBFs; the proposed scheme is able to capture patterns of the models for big time T ; the scheme has second order of convergence in both time and space variables and Nuemann boundary conditions are easy to implement in this scheme.