In this presentation, I will discuss artificial intelligence (AI) and machine learning (ML) based approaches to study structure and dynamics of soft materials [1,2]. Although the examples will be from soft and biological systems, the methods themselves are general and have broader applicability. The first example is lipids. Their phase behavior is complex, and even single component bilayers still pose challenges. The gel and fluid phases of lipid bilayers are planar and conformationally homogeneous, but the ripple phase, which exists at temperatures between the gel and fluid phases, undulates in an asymmetric sawtooth pattern and consists of heterogeneous conformational clusters. The ML approach here identifies the molecular conformations, and changes in their respective proportions that give rise to the changes. The method is also demonstrated for multicomponent systems.

References:

1. Elucidating Lipid Conformations in the Ripple Phase: Machine Learning Reveals Four Lipid Populations. Davies, Matthew; Reyes-Figueroa, A. D.; Gurtovenko, Andrey A.; Frankel, Daniel; Karttunen, Mikko. Biophys. J. 122, P442-450 (2023).
2. Learning glass transition temperatures via dimensionality reduction with data from computer simulations: Polymers as the pilot case. Glova, Artem; Karttunen, Mikko. J. Chem. Phys. 161, 184902 (2024).