Fig. 5
From: Multiobjective genetic training and uncertainty quantification of reactive force fields
a Original MOGA workflow for re-optimizing ReaxFF parameters to reproduce the ground-truth QMD trajectory. This workflow creates N different force-field files corresponding to N different RMD simulations. It then produces the equal number of bond time-series files for H–S, Mo–O and Mo–S, which are compared against the ground-truth QMD time series. b The iMOGA workflow produces N different force-field n-tuples, corresponding to which N different RMD simulations are run, which in turn generates the equal number of bond time-series and compares them with the QMD ground-truth
