(401a) Development of Coarse-Grained (CG) Embedded Atom Method (EAM) Potentials for FCC Metals Using Machine Learning and Bayesian Uncertainty Quantification

To develop a fundamental understanding of large and complex systems, use of coarse-grained (CG) molecular dynamics (MD) has become increasingly popular as a way to perform MD simulations in an efficient and economical way due to its lower resolution, and faster dynamics. However, the determination of the parameters for accurate interatomic potentials (force field parameters) for any organic and/or inorganic materials systems remains a challenging task. Here we have developed new CG embedded atom method (EAM) potentials to describe the interatomic interactions for face-centered cubic (FCC) metals viz. Gold (Au), Palladium (Pd), Nickel (Ni), Copper (Cu), Aluminum (Al) and Platinum (Pt). Specifically, particle swarm optimization (PSO) integrated CG MD approach was utilized to explore 14 parameter space defined using EAM potential to reproduce physical, mechanical, and thermodynamic properties of their respective metals. Furthermore, a computational framework was developed to determine the uncertainty and accuracy of these developed FF parameters. Particularly, this framework consists of the following steps: (i) A sensitivity analysis using Sobol sampling was performed using the data obtained via the CG model simulations and varied parameter possibilities. (ii) A machine learning model (Gaussian Process regression (GPR)) developed on this sobol data was used as a surrogate model for MD simulations. (iii) Bayesian parameter estimation and further uncertainty quantification (UQ) of the model were then accomplished using an integration between the GP model and PEUQSE, a Bayesian tool for parameter estimation. Predictions of metal properties using the Bayesian refined CG model were within reasonable confidence levels of the experimental data. This robust computational framework will allow for the development of interatomic potentials that are highly accurate and reliable for both hard and soft materials.