Regensburg 2019 – wissenschaftliches Programm

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

DY: Fachverband Dynamik und Statistische Physik

DY 50: Poster: Nonlinear Systems, Patterns, Flows ..

DY 50.16: Poster

Donnerstag, 4. April 2019, 15:00–18:00, Poster B2

Estimating low dimensional dynamical models for molecules — •Max Philipp Holl and Oliver Kamps — Institut für theoretische Physik, Universität Münster, Wilhelm-Klemm-Straße 9, 48149 Münster

The dynamics of molecules is a complex problem involving many degrees of freedom. However different states of the molecular system can often be described by only a few of them. We combine two data driven techniques to calculate a low dimensional representation for these systems and estimate dynamical models for the molecules. First we reduce the dimensionality of a time series using a nonlinear dimensionality reduction algorithm, Scalable ISOMAP [1]. Secondly we estimate the deterministic drift and stochastic driving on this reduced system using a data-driven estimation method, based on the Langevin equation [2].

[1] P. Das et al. Low-dimensional, free-energy landscapes of protein-folding reactions by non-linear dimensionality reduction. Proceedings of the National Academy of Sciences, 103(26):9885-9890, 2006.

[2] R. Hegger and G. Stock. Multidimensional Langevin model- ing of biomolecular dynamics. The Journal of Chemical Physics, 130(3):034106, jan 2009.