Technical Report 47, c4e-Preprint Series, Cambridge

Weighted particle methods for the Smoluchowski coagulation equation

ref: Technical Report 47, c4e-Preprint Series, Cambridge

Associated Themes: Numerics and Nanoparticles

Abstract

The Smoluchowski coagulation with additional linear terms is formulated on an explicitly weighted space. The use of such a space extends the ideas behind the mass flow algorithm [A. Eibeck and W. Wagner, Ann. Appl. Prob. (2001) 11:1137-1165] and allows the derivation of a range of new stochastic particle algorithms. Two such algorithms, which seem particularly natural, are implemented and successfully validated. The weighted algorithms are then compared to each other and to a version direct simulation algorithm, which uses particle doubling as a variance reduction technique. One weighted algorithm is found to consistently outperform the other by a significant amount. Applications for which the better weighted algorithm outperforms direct simulation are demonstrated, along with situations where the converse is true.

Material from this preprint has been published in: Journal of Computational Physics 230, 7456-7472, (2011)

Download

PDF (370.25 KB)