Berlin 2018 – wissenschaftliches Programm

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

DY: Fachverband Dynamik und Statistische Physik

DY 72: Poster: Stoch. and Nonl. Dy., Modeling, Compl. Sys.

DY 72.4: Poster

Donnerstag, 15. März 2018, 15:30–18:00, Poster A

Using Neural Networks to Obtain the Dynamics on the Normally Hyperbolic Invariant Manifold of Driven Systems — •Martin Tschöpe, Johannes Reiff, Robin Bardakcioglu, Matthias Feldmaier, Jörg Main, and Günter Wunner — Institut für Theoretische Physik 1 Universität Stuttgart, Deutschland

The semi-classical reaction dynamics in Transition State Theory (TST) for periodically driven systems has been solved numerically in the past. It has been shown that the reaction rate can be determined by integrating a large number of trajectories and analyzing phase space structures.

An alternative way to obtain a reaction rate for these systems might be to analyze the dynamics on the Normally Hyperbolic Invariant Manifold (NHIM) itself. This task is especially challenging, due to the critical properties of such trajectories. Our work will tackle this disadvantage by using neural networks as a tool to stabilize the numerical integration on the NHIM.