Some experiments on numerical simulations of stochastic differential equations and a new algorithm (Q1335598)

This paper reports the results of numerical experiments involving numerical methods for approximating the solutions of Itô stochastic differential equations. A method of Mil'shtein, a method of Talay, and a new Runge-Kutta type method were applied to four test problems and then compared with respect to accuracy and speed under different stepsizes, sample sizes of the associated Monte Carlo simulations, and type of approximation of the increment of Brownian motion.