Practical Approaches to Bottom-Up Coarse-Graining for Soft Materials

Deep neural networks have become popular model architectures for fitting coarse-grained molecular dynamics potentials (CGMD) owing to their ability to describe complex features and ease of training against large datasets. However, such architectures are much more complicated than traditional functional forms. This raises the question of whether a similar data-driven approach that used a simpler functional form would have any advantages. Here, we develop a genetic algorithm that optimizes a Lennard-Jones potential for coarse-grained models of various crystal and liquid crystal forming materials based on both structural and thermodynamic information. A detailed description of the genetic algorithm, its loss function, and hyperparameters is presented. The models developed by the algorithm are demonstrated to approximately reproduce the radial distribution function of the experimental crystal structures, maintain high levels of directional order, and realistic pressures or volumes. Simulations of larger systems with these models demonstrate stabilization of the reference crystal structure on longer time scales. Test molecules with known liquid crystalline phases are shown to stabilize a simulated liquid crystalline phase along with liquid crystalline phase transitions, despite this information being absent from model training. These case-studies suggest that simpler functional forms retain untapped potential for CGMD when coupled with data-driven training algorithms.