Fig. 4: Convergence stage: data-efficient active learning of the GNN.

We illustrate the GP-based active learning of the MACE structures for the N recombination and dissociation process. a Distribution of the maximum uncertainty on the forces \({\sigma }_{max}^{(MACE)}\) (calculated from a committee model) on the simulation performed with MACE before the active learning. The gray dotted line at 90 meV Å⁻¹, represents the chosen threshold for the query-by-committee selection. In the inset (a1), we report the distribution of the same configurations along the collective variable d_N,N. b Distribution of the DFT single point performed with DEAL (red bars), among the ones pre-selected via query-by-committee (blue bars). The number of configurations in each bin is reported on top of the bars. c Distribution of the maximum uncertainty on the forces \({\sigma }_{max}^{(MACE)}\) for initial model (blue histogram) and after DEAL (red histogram). The second distribution is obtained from a new MD with identical parameters performed with the MACE model after DEAL. In the inset (c1), we report the average uncertainty along the reaction coordinate d_N,N for the two simulations generated respectively with the initial MACE model (blu line) and after DEAL (red line).