Thermodynamically constrained inverse screening of metal hydride alloys for solid-state hydrogen storage
Description
This dataset supports the manuscript "Thermodynamically constrained inverse screening of metal hydride alloys for solid-state hydrogen storage" submitted to Journal of Alloys and Compounds. The dataset contains standardized and processed data used for data-driven screening of metal hydride hydrogen-storage alloys. It includes the harmonized ML-HydPARK-derived training table, composition descriptors, external TiFe-Cr-Mo and AB2/HEA validation tables, model-output summaries, candidate-screening results, Pareto-ranked candidates, source data for tables and figures, and reproducibility scripts. The release also includes results.json, environment files, documentation of data provenance and preprocessing, and scripts for regenerating processed datasets, model predictions, tables, figures, and candidate lists. The screened alloy compositions should be treated as model-generated hypotheses requiring independent synthesis and experimental validation.
Files
Steps to reproduce
Download and unzip the dataset package. Create the Python environment using environment.yml or install the dependencies listed in requirements.txt. From the project root, run scripts/run_all.py to regenerate processed datasets, model outputs, tables, figures, candidate-screening results, and results.json. The main processed data are stored under data/processed, external validation tables under data/external and data/raw, candidate-screening outputs under results/candidates, and manuscript tables and figure source data under results/tables and results/figures.