vasprunrs

A fast parser for VASP vasprun.xml files, written in Rust with Python bindings.

vasprunrs is a drop-in replacement for pymatgen.io.vasp.outputs.Vasprun that parses the same files 5–10x faster using a streaming Rust XML parser. It returns native pymatgen objects (Structure, CompleteDos, BandStructure, etc.) so existing code needs only a one-line import change.

Installation

pip install vasprunrs

# with pymatgen compatibility shim (recommended)
pip install "vasprunrs[pymatgen]"

Usage

Drop-in replacement for pymatgen

# Before
from pymatgen.io.vasp import Vasprun

# After
from vasprunrs.pymatgen import Vasprunrs as Vasprun

vr = Vasprun("vasprun.xml")

structure      = vr.final_structure       # pymatgen Structure
dos            = vr.complete_dos          # pymatgen CompleteDos
band_structure = vr.get_band_structure()  # pymatgin BandStructure
eigenvalues    = vr.eigenvalues           # {Spin.up: ndarray, ...}

Native API (no pymatgen required)

from vasprunrs import Vasprun

vr = Vasprun("vasprun.xml", parse_dos=True, parse_eigen=True)

print(vr.version)           # "6.4.2"
print(vr.efermi)            # 4.123 (eV)
print(vr.final_structure)   # dict with lattice, positions, species
print(vr.eigenvalues.shape) # (nspins, nkpts, nbands, 2)
print(vr.dos)               # dict with efermi, total, partial

Features

Performance

vasprunrs uses a streaming XML parser that does not load the entire file into memory. Parse times are typically 5–10x faster than pymatgen's Python parser.

Parse options

Both the native Vasprun and the compatibility Vasprunrs accept the same options:

Known limitations

• converged_electronic and converged_ionic are not tracked separately (both reflect the same value)
• Epsilon static/ionic tensors (epsilon_static, epsilon_ionic) are not parsed from the DFPT block
• Force constants / phonons and Born effective charges are not yet implemented

Note on per-atom magnetic moments: Vasprunrs.magnetization reads the <varray name="magnetization"> block from each ionic step. This block is written by VASP 5.x but is absent in VASP 6.x output, so magnetization will return None for all VASP 6.x files regardless of whether the calculation is magnetic. Per-atom moments for VASP 6.x can be recovered by integrating the partial DOS up to the Fermi energy (spin-up minus spin-down per atom), which requires parse_dos=True and LORBIT ≥ 11 in the INCAR.

Requirements

• Python 3.9+
• numpy
• pymatgen (optional, for vasprunrs[pymatgen])

License

MIT