⚗️ vitrum

vitrum is a Python package designed for the generation, analysis, and simulation of disordered and glassy atomic structures. It provides a comprehensive suite of tools for structural characterization, diffusion analysis, and tools for machine learning-driven potential development.

🚧 Active development

vitrum is under active development. As of 1.0, the public API follows semantic versioning — breaking changes will be reflected in a major version bump and noted in the changelog.

📖 Documentation

Please see the docs folder for detailed documentation or check the online documentation.

📦 Installation

vitrum is available on PyPI:

pip install vitrum

To install dependencies for simulation workflows (atomate2, fireworks, jobflow):

pip install vitrum[workflows]

For the latest development version, clone the repository and install it in editable mode instead:

git clone https://github.com/R-Chr/vitrum.git
cd vitrum
pip install -e .

🚀 Examples

See the examples folder for runnable Jupyter notebooks demonstrating scattering/RDF analysis, Qn speciation, and random structure generation, among others.

🎯 Scope and Functionality

vitrum offers:

1. Structural Characterization

• Scattering Functions: Calculate partial and total Radial Distribution Functions (RDF) and Structure Factors ($S(q)$) for both Neutron and X-ray scattering (vitrum.scattering).
• Ring Analysis: Analyze ring size distributions and statistics in network glasses (vitrum.rings).
• Topological Analysis: Compute persistent homology to identify medium-range order and topological features (vitrum.persistent_homology).
• Coordination & Angles: Analyze bond angle distributions and coordination environments (vitrum.coordination).

2. Dynamics & Diffusion

• Diffusion Analysis: Calculate Mean Squared Displacement (MSD), diffusion coefficients, and Van Hove correlation functions (vitrum.diffusion).

3. Machine Learning & Workflows

• BALACE Framework: A Batch Active Learning framework for Atomistic Simulations (vitrum.batch_active) (requires workflows dependencies).
  • Automated workflow for training Machine Learning Interatomic Potentials (MLIPs) based on ACE .
  • Integration with VASP and LAMMPS for data generation and active learning loops.
  • Job management via Fireworks and Jobflow.

📑 Citation

If you use vitrum in your work, please cite it. Each GitHub release is archived on Zenodo with a version-specific DOI; see CITATION.cff for the citation metadata (GitHub's "Cite this repository" button uses this file automatically).

(DOI badge above is a placeholder — update it with the DOI Zenodo mints for the next release.)

👥 Author

Rasmus Christensen (rasmusc@bio.aau.dk)

⭐ Acknowledgements

vitrum relies on several powerful open-source packages:

• ASE
• Pymatgen
• NumPy / SciPy / pandas
• scikit-learn
• Dionysus / DioDe
• Atomate2 / Jobflow / Fireworks