On-Top Pair Density Predictions for Organic Molecules via Machine Learning

A machine learning model predicts the CASSCF-quality on-top pair density, a quantum chemical property indicating electron correlation, for organic molecules. The model was trained on the GDB11-AD-3165 database, bypassing the need for ab-initio computations. It was developed by Alberto Fabrizio at École Polytechnique Fédérale de Lausanne.

Use Cases

• Predicting electron correlation effects in organic molecules based on the on-top pair density property.
• Visualizing bond-breaking in real-space based on the on-top ratio metric.
• Training surrogate models for density functional theory and multireference wavefunction theory applications.

Strengths

• Model trained on the GDB11-AD-3165 database, a known source for molecular structures.
• Predicts with minimal error according to the description, bypassing computationally expensive ab-initio methods.
• Includes a specialized basis set for fitting the on-top pair density in an atom-centered expansion.

Limitations

• Row count and dataset size are unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect bias inherent to the GDB11-AD-3165 database and the paperswithcode platform.

Provenance

Source

École Polytechnique Fédérale de Lausanne

Collection Method

Machine learning model trained on the GDB11-AD-3165 database.