Zur NavigationLogin

DE EN

B7 (N): Machine learning for multiscale simulations

The goal of the project is to examine the application of machine learning methods, in particular modern, multi-scale representation learning methods based on deep neural networks to multi-scale simulations of soft-matter systems. We will consider two application scenarios: (i) Regression to compute complex interactions: here we will use polarization as a test case for a non-additive interaction, and develop a regression scheme that learns a mapping function from the configuration of particles directly to forces. (ii) Generative probabilistic models: we will consider the problem of back-mapping coarse-grained simulations to higher-resolution representations. From a machine learning perspective, this requires conditional generative statistical models.

Funding for this project has started in July 2018.


Adversarial reverse mapping of equilibrated condensed-phase molecular structures
Marc Stieffenhofer, Michael Wand, Tristan Bereau
Machine Learning: Science and Technology 1, 045014 (2020);
doi:10.1088/2632-2153/abb6d4

A tight and consistent link between resolutions is crucial to further expand the impact of multiscale modeling for complex materials. We herein tackle the generation of condensed molecular structures as a refinement—backmapping—of a coarse-grained (CG) structure. Traditional schemes start from a rough coarse-to-fine mapping and perform further energy minimization and molecular dynamics simulations to equilibrate the system. In this study we introduce DeepBackmap: A deep neural network based approach to directly predict equilibrated molecular structures for condensed-phase systems. We use generative adversarial networks to learn the Boltzmann distribution from training data and realize reverse mapping by using the CG structure as a conditional input. We apply our method to a challenging condensed-phase polymeric system. We observe that the model trained in a melt has remarkable transferability to the crystalline phase. The combination of data-driven and physics-based aspects of our architecture help reach temperature transferability with only limited training data.

Kernel-Based Machine Learning for Efficient Simulations of Molecular Liquids
Christoph Scherer, René Scheid, Denis Andrienko, Tristan Bereau
Journal of Chemical Theory and Computation 16 (5), 3194-3204 (2020);
doi:10.1021/acs.jctc.9b01256

Contact

This page uses cookies. By continuing to visit this site you agree to our use of cookies.

More info