Branching processes and the Boltzmann equation - Computational aspects

A modification of a Monte Carlo algorithm based on branching process modeling is proposed for estimating the functionals of solutions to the nonlinear Boltzmann equation. The algorithm allows straight-through calculation over a certain time interval, with no intermediate results needed to be remembered, and provides a simple way of error estimation. A description of the algorithm is presented, and its applicability to the Cauchy problem is demonstrated. The numerical results obtained are compared with an exact BKW solution.