The Atomistic Reconstruction of Coarse-Grained Polymeric Systems via Machine Learning Techniques
Loading...
Date
2020
Authors
Olsen, Jake Dylan
Journal Title
Journal ISSN
Volume Title
Publisher
University of Oregon
Abstract
The development of a statistically accurate backmapping procedure, coupled with an accurate coarse-graining (CG) method, is necessary as it would allow a system to freely transform between varying degrees of CG. This ability allows for the computational gain of CG with the resolution of atomistic simulations. Therefore, using state-of-the-art machine learning techniques coupled with atomistic simulation data, we have developed a backmapping procedure for CG polymeric systems. Specifically, we used a gated recurrent unit (GRU) to learn the atomistic structure within a single CG site of a polyethylene system. A categorical cross-entropy loss function was used to allow for more flexibility in the model. The model’s training yielded consistent loss and validation loss demonstrating that the model did not overfit the data. Furthermore, the model was able to accurately reproduce a variety of structural quantities, such as the bond angle, bond length, dihedral angle, mean-square internal distance (MSID), end-to-end distance, and distribution about the center-of-mass.
Description
46 pages
Keywords
Physical Chemistry, Chemistry, Course-graining, Backmapping, Machine Learning, Molecular Dynamics, Polymers