Friday, 24th April, 2020, 10.30

Soft-matter materials modeling in the data-driven era

Advanced statistical methods are rapidly impregnating many scientific fields, offering new perspectives on long-standing problems. In materials science, data-driven methods are already bearing fruit in various disciplines, such as hard condensed matter or inorganic chemistry, while comparatively little has happened in soft matter.

I will describe how we use data-driven methods to leverage molecular simulations in soft matter. We aim at establishing structure-property relationships for complex thermodynamic processes across the chemical space of small molecules. Akin to screening experiments, we devise a high-throughput coarse-grained simulation framework. Coarse-graining is an appealing screening strategy for two main reasons: it significantly reduces the size of chemical space and it can suggest a low-dimensional representation of the structure-property relationship.

I will illustrate these aspects through the passive translocation of small molecules across a phospholipid bilayer, identifying interpretable structure-property relations, as well as recent results on the screening of small molecules to drive phase transitions in lipid mixtures. Further applications of recent machine learning architectures to molecular simulations will be described: representation learning of a free-energy landscape using Gaussian-mixture variational autoencoders, and the systematic backmapping of coarse-grained configurations to an atomistic level using conditional generative adversarial networks.

Speaker: Tristan Bereau, Van't Hoff Institute for Molecular Sciences and Informatics Institute, University of Amsterdam, Netherlands

Room Charles Darwin, PRBB Innner square