Nanotech 2001 Vol. 1
Nanotech 2001 Vol. 1
Technical Proceedings of the 2001 International Conference on Modeling and Simulation of Microsystems

Numerics, Algorithms Chapter 2

Variance Reduction in Monte Carlo Device Simulation by Means of Event Biasing

Authors: H. Kosina, M. Nedjalkov and S. Selberherr

Affilation: Technical University of Vienna, Austria

Pages: 11 - 14

Keywords: event biasing, variance reduction, Monte Carlo method, Boltzmann equation, device simulation

Abstract:
A theoretical analysis of the Monte Carlo (MC) method for the solution of the stationary boundary value problem defined by the Boltzmann equation is briefly outlined. This analysis clearly shows how event biasing can be used withing the well-known One-Particle MC method. To enhance statistics in sparsely populated regions of interest, artificial carier heating is introduced by increasing the probability for phonon emission. The distribution of the scattering angle is biased to induce artificial carrier diffusion. Having identified the random variable whose realizations are statistically independent new variance estimates are proposed.

Variance Reduction in Monte Carlo Device Simulation by Means of Event Biasing

ISBN: 0-9708275-0-4
Pages: 638