(597i) A Data-Driven Closure for Polymer Integral Equation Theory

Polymer reference interaction site model (PRISM) theory is a powerful technique that can rapidly predict the structure and thermodynamics of liquid-like polymer systems. However, PRISM theory can be difficult to implement, and can suffer issues with accuracy and numerical stability for complex systems of industrial and technological relevance. Typically, the analytical closure relation used to generate a numerical solution to the PRISM equations is considered the key barrier toward broader applicability of the theory. In this work, we describe our efforts to develop a data-driven machine learning (ML)-based closure relationship for PRISM theory. We prepared an initial dataset for model training using coarse-grained molecular dynamics simulations of homopolymer melts and solutions across a range of chain lengths, intermolecular interaction strengths, and thermodynamic conditions. Then, we evaluated multiple approaches to develop an ML-based closure to predict the structural correlation functions needed to solve a PRISM theory calculation. Our preliminary results show that the ML closure performs favorably in comparison to widely-used analytical closures (e.g., Percus-Yevick closure). We then describe our ongoing work to evaluate the ML closure's transferability to more complex systems such as blends, copolymers, and polymer nanocomposites.