Overview
Constructing low-dimensional PES representations for reactions involving many coupled internal coordinates.
This workflow:
- Uses PCA to identify dominant structural variations from multiple internal coordinates (e.g., bond lengths, angles, dihedrals),
- Uses the resulting principal components (PC1 and PC2) as physically meaningful scan axes,
- Performs relaxed surface scans using electronic structure software (e.g., Gaussian),
- Visualizes 2D or 3D PESs.
Requirements
Python ≥ 3.8
Required Python packages: numpy pandas scikit-learn matplotlib
Quantum chemistry software for scan calculations: Gaussian 16 (examples provided), but the workflow can be adapted to other packages.
Example Usage
A Jupyter Notebook example is provided in the examples/ directory to walk you through the complete procedure step by step.