Berlin 2015 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

DY: Fachverband Dynamik und Statistische Physik

DY 25: Nonlinear Stochastic Systems

DY 25.3: Vortrag

Dienstag, 17. März 2015, 15:30–15:45, BH-N 128

Simulating stochastic differential equations using truncated Markov chains — •Rüdiger Kürsten and Ulrich Behn — Institut für Theoretische Physik, Universität Leipzig, Brüderstr. 16 D-04103 Leipzig

Markov chains with detailed balance are widely used to sample high-dimensional distributions, a prominent example is the Monte Carlo simulation. To sample critical phenomena or rare events with high precision modified techniques such as, for example, the multicanonical method are successfully applied. The methods use more or less tacitly a truncated Markov chain where transition probabilities between certain regions of state space are manipulated. Solution trajectories of stochastic differential equations (SDE) can be sampled naively for instance by an Euler-Mayurama scheme. Here we develop an efficient simulation technique for SDEs that is based on a variant of the truncated Markov chain. An extension for systems without detailed balance is proposed.