Numerical learning approximation of time-fractional sub diffusion model on a semi-infinite domain

Abstract

Abstract The propose of this research is to apply a novel numerical learning approximation of time-fractional sub diffusion model on a semi-infinite domain. This model is a nonlinear fractional differential equation in two unbounded dimensions. Combination of Least Squares Support Vector Regression (LSSVR) based on generalized Laguerre Functions kernel and collocation/Galerkin method is applied to obtain the solutions. The marching in time technique is applied for time discretization. Numerical results verify that proposed methods have high convergence and performance.