Numerical comparison of three stochastic methods for nonlinear PN junction problems (Q2259127)

The authors deal with the problem of simulating a PN junction in which the anod surface is randomly rough. The mathematical model consists of a system of two stationary drift-diffusion equations, one for electrons and one for holes, coupled through the Poisson equation for the self-consistent potential. They use stochastic mapping to transform the deterministic governing system of equations in a random domain into a stochastic system in a deterministic domain. They discretize the two-dimensional physical space using the finite element method, with the help of AFEPack software package. The resulting equations are solved by means of the approximate Newton iterative method. Comparisons among numerical simulations obtained by means of the Monte Carlo, Galerkin and stochastic collocation methods are made. These simulations show that, for such a complicated nonlinear problem, the stochastic Galerkin method is accurate but the number of basis functions grows quickly with the increasing order, while the stochastic collocation method combines the accuracy of the stochastic Galerkin method and the easy implementation of the Monte Carlo method.