RangeBO: Benchmark Data for Specification-Driven Materials Design

Benchmark datasets for "Range-Aware Bayesian Optimization for Discovering Diverse Designs within Target Property Windows" by Jiang, Wu, Schroeder, and Webb. Code: https://github.com/webbtheosim/target-range-optimization This archive contains the processed datasets for 10 benchmarks: bace, esol, freesolv, lipo, qm9, nanoparticle, propensity, toporg, kmc, and tddft. The synthetic benchmarks, ackley, branin, hartmann, and layeb06, are defined in the code and are not included. Contents of benchmark_data.zip: clean/Model-ready features, labels, and output scalers. mordred/Mordred molecular descriptors before PCA. smiles/Raw molecular data in CSV and Excel formats. config/Target values and tolerance settings. README.txtDataset description and provenance. Data flow: smiles/ -> mordred/ -> clean/ The clean/ folder is sufficient to run the benchmarks. To use the data: git clone https://github.com/webbtheosim/target-range-optimization.gitcd target-range-optimizationunzip benchmark_data.zip -d bo_target/data/