MuML: Molecular Dipole Moment Prediction Models and Source Data

Source data, models, and scripts for reproducing results from the paper 'Predicting molecular dipole moments by combining atomic partial charges and atomic dipoles'. The model combines symmetry-adapted Gaussian process regression for atomic dipoles with scalar GPR for atomic partial charges. The dataset was created by Max Veit at École Polytechnique Fédérale de Lausanne to accompany the arXiv preprint 2003.12437.

Use Cases

• Reproducing molecular dipole moment predictions based on the provided source data and kernel matrices.
• Benchmarking new quantum chemistry models against the published SA-GPR and scalar GPR results.
• Analyzing model residuals for atomic partial charges and atomic dipoles mentioned in the description.
• Developing new fitting or plotting scripts for molecular property data using the included examples.

Strengths

• Includes all source data, kernel matrices, and model weights necessary for full reproducibility of the published research.
• Provides scripts for both fitting the models and plotting the results, facilitating reuse and extension.

Limitations

• Row count and dataset size are unknown, which may limit suitability assessment for large-scale applications.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

École Polytechnique Fédérale de Lausanne

Collection Method

Created to accompany the research paper 'Predicting molecular dipole moments by combining atomic partial charges and atomic dipoles'.