janosh/pymatviz

What it solves

pymatviz is a toolkit designed to simplify the visualization of data in materials informatics. It provides specialized tools to visualize chemical compositions, crystal structures, phonon properties, and other materials-science-specific data that are often difficult to represent using standard plotting libraries.

How it works

The library offers a suite of plotting functions and interactive widgets that integrate with common materials science data formats (such as pymatgen, ASE, and PhonopyAtoms). It leverages Plotly for interactive 2D plots and a custom backend called MatterViz for 3D structure and trajectory visualizations.

Who it’s for

It is built for researchers and data scientists working in materials informatics who need to perform exploratory data analysis (EDA) on material properties, compositions, and atomic structures.

Highlights

• Periodic Table Visualizations: Interactive heatmaps, histograms, and scatter plots mapped directly onto the periodic table.
• Composition Clustering: Tools to visualize relationships between chemical compositions and properties using embeddings (One-hot, Magpie, Matscholar, MEGNet) and dimensionality reduction (PCA, t-SNE, UMAP, Isomap, Kernel PCA).
• Phonon Analysis: Specialized plots for phonon bands and density of states (DOS).
• Interactive Widgets: 3D structure and molecular dynamics trajectory viewers for Jupyter, Marimo, and VSCode notebooks.
• Materials-Specific Plots: Built-in support for rendering 3D Brillouin zones, X-Ray Diffraction (XRD) patterns, and Radial Distribution Functions (RDF).
• Headless Export: Ability to export interactive widgets to standalone HTML or static images (PNG/JPEG/SVG/PDF) for use in CI/CD pipelines or agent workflows.