atomix

Atomistic Modeling Interface for eXploration — a natural language driven toolkit for ab initio / DFT / atomistic modeling workflows.

Installation

Atomix is currently installed from a source checkout. The segmented VASP reader requires the scientific dependencies:

conda activate atomix
pip install -e ".[science]"

Status

Early development. The supported surface is deliberately narrow: read-only inspection of VASP calculations stored as seg01, seg02, ... directories. Other commands and modules remain experimental scaffolding.

Command line

atomix inspect-vasp path/to/calculation
atomix inspect-vasp path/to/calculation --json
atomix inspect-vasp path/to/calculation --diagnostics

The command exits nonzero when the reader finds an error-level ambiguity, such as both seg1 and seg01 representing the same segment number.

Python

from atomix.calculators import VASPSegmentReader

result = VASPSegmentReader("path/to/calculation").read()

for frame in result.frames:
    print(frame.segment_number, frame.time_fs, frame.source_file)

for warning in result.warnings:
    print(warning.level, warning.code, warning.path)

The reader sorts segments numerically, prefers XDATCAR trajectories with an OUTCAR fallback, removes repeated boundary frames, and retains raw OUTCAR and REPORT records with source metadata. It never modifies calculation files.

Development

pip install -e ".[science,dev]"
pytest tests/
ruff check .