psdpy

Python implementation for computing pore size distribution in materials using voxelization and the Euclidean distance transform.

Usage

Run the analysis on a structure file such as .pdb or .gro:

python psd_from_structure.py <structure-file> --dx 1.0 --use-pbc --output-dir <output-dir>

Common options:

• --dx: grid spacing in Angstrom
• --probe-radius: probe radius added to atomic radii
• --mode: voxel or maximal_sphere
• --selection: MDAnalysis atom selection string
• --align-axis: axis to align the longest principal axis onto (x, y, or z)
• --no-align: disable the PCA-based alignment step

Outputs

Each run writes a result directory containing:

• psd_outputs.json: serialized PDF, CDF, frame metrics, and run parameters
• psd_outputs.npz: NumPy archive with the same numeric arrays
• psd_pdf.png: pore size distribution plot
• psd_cdf.png: cumulative distribution plot
• psd_frame_metrics.png: per-frame porosity and mean radius trace
• psd_slice_x.png: EDT slice along the x axis
• psd_slice_y.png: EDT slice along the y axis
• psd_slice_z.png: EDT slice along the z axis

Examples

See examples/README.md for bundled input structures and saved outputs.

Reference

Bhattacharya, Supriyo, and Keith E. Gubbins. "Fast method for computing pore size distributions of model materials." Langmuir 22, no. 18 (2006): 7726-7731.