Grid-free simulation of diffusion using random wall methods (Q1077906)

The authors investigate the simulation of diffusion of a set of particles by means of the random walk displacement. They consider a type of statistical approach that differs from classical random walk methods both of the discrete and of the floating type. The method they expose can be called of ''gradient random walk'', where particles following a Brownian motion transport flux elements while their timely displacements are constructed to approximate, in statistical sense, the gradient of the concentration distribution; this latter is then obtained by repeated integration. The method can be considered as originated by the researches of Chorin et al. [see e.g. \textit{A. J. Chorin}, Commun. Pure Appl. Math. 34, 853-866 (1981; Zbl 0458.76042)]. The paper contains also some examples of applications and some considerations about the advantages that this method presents from various angles in comparison with classical Monte Carlo methods.