Fig. 2: The Bayesian optimization workflow for refining coarse-grained topologies.
From: Refining coarse-grained molecular topologies: a Bayesian optimization approach
This flowchart illustrates the iterative process where properties from all-atom molecular dynamics (AAMD) serve as the ground truth. A Gaussian Process Regression (GPR) surrogate model and an acquisition function are used to intelligently select new coarse-grained (CG) topology parameters to evaluate, progressively minimizing the difference between CG and AA properties to find an optimal topology.
