An improved algorithm for the estimation of the mean first passage time of ordinary stochastic differential equations

✍ Scribed by M. Seeβelberg; F. Petruccione

• Publisher: Elsevier Science
• Year: 1993
• Tongue: English
• Weight: 723 KB
• Volume: 74
• Category: Article
• ISSN: 0010-4655
• DOI: 10.1016/0010-4655(93)90095-t

✦ Synopsis

We investigate the estimation of the mean first passage time of a stochastic differential equation by numerical methods. In order to determine the mean first passage time correctly, one needs a numerical procedure to generate trajectories which converge in the mean square limit to exact solutions of the stochastic differential equation. First we briefly review a suitable algorithm for this purpose which uses Gaussian distributed random numbers. Then we show that the algorithm remains appropriate for the estimation of the mean first passage time even if one replaces the Gaussian random numbers by uniformly distributed ones. Exploiting this fact it is possible to reduce substantially the computation time for the estimation of the mean first passage time.