Dataset for Differential Learning for Robust Prediction of Thermal Stability with Application to Energetic Materials

This dataset contains descriptors for 931 energetic material-like molecules. This is for SupportingInformation for our manuscript “Differential Learning for Robust Prediction of Thermal Stabilitywith Application to Energetic Materials”, where we train machine learning models to predictdecomposition temperatures of these molecules. These molecules were collected from two openliterature sources1,2 and are present in the dataset in the form of SMILES strings.Elemental composition of the dataset is limited to hydrogen, carbon, nitrogen and oxygen. Noperformance metrics are provided. Descriptors include several derived from Density FunctionalTheory calculations, including internal energy, bond dissociation enthalpies, heats of explosion,etc. Other descriptors are principally computed using the cheminformatics package RDkit, e.g.fr_nitro (number of nitro groups), mol_wt (molecular weight), etc.