Supplementary dataset to: "A Simple and Scalable Kernel Density Approach for Reliable Uncertainty Quantification in Atomistic Machine Learning"

Supplementary dataset for: "A Simple and Scalable Kernel Density Approach for Reliable Uncertainty Quantification in Atomistic Machine Learning" Python Scripts: - 1_DESC.py – Generates descriptors from a training database and performs PCA to reduce data dimensionality. - 2_CALIBRATION.py – Calculates the standard deviation of k nearest neighbors to determine the bandwidth for density estimation. - 3_DENSITY.py – Computes the KDE density for a given structure or trajectory, indicating the coverage and uncertainty relative to the training database. Case Study Files: - Pt_silica – Contains training database descriptors and trajectories used in the study, including Pt_5 clusters in CHA and Pt_6 clusters on silicatene from molecular dynamics simulations. - H-MFI – MD trajectories for MFI with Si/Al ratios of 95 and 11, each with water loading of one H2O molecule per Al atom. - NMR – Cell and .magres files for NMR calculations of zeolites, along with CASTEP setup files for the corresponding simulations.