Adaptive kernel density estimation and Monte Carlo sampling

The simulation of electron transport is computationally intensive. In principle, conventional (analog) Monte Carlo methods can be employed to analyze complex configurations to arbitrary accuracy, but in practice, the amount of computation is prohibitive for most applications. Condensed history Monte Carlo (CHMC) methods, which sample the outcome after a large number of electron collisions, have become the standard method for electron transport calculations but suffer well-known deficiencies. This report describes a method based on the kernel density estimator proposed as an alternative to histogramming