Monte Carlo Simulation and Machine Learning-Assisted Scattering Analysis of Mechanically Driven Polymers

We develop off-lattice Markov Chain Monte Carlo simulations and a Machine Learning Inversion method to analyze the behavior of semiflexible polymers under external forces. We model the polymer as a chain of fixed-length bonds subjected to bending energy, with configurations updated through adaptive non-local Monte Carlo moves. This enables accurate predictions of polymer response under uniaxial stretching and steady shear, overcoming the orientational bias of on-lattice models. The polymer conformation is captured by the scattering function. We apply a Machine Learning inversion method to extract key energy and conformation parameters from the scattering function. Using a dataset generated from Monte Carlo simulations, we train a Gaussian Process Regressor to successfully recover the bending modulus, stretching and shear forces, as well as end-to-end distance, radius of gyration, and the off-diagonal component of the gyration tensor. Our combined approach enhances precision in studying polymer behavior, offering insights into scattering functions and polymer conformation.