Skip to content

VASPBERRY 1.2.0 — Fukui-Hatsugai Z2 and Hall transport

Choose a tag to compare

@github-actions github-actions released this 04 Sep 21:05

VASPBERRY 1.2.0 — Fukui-Hatsugai Z2 and Hall transport

VASPBERRY 1.2.0 brings the two-dimensional Fukui-Hatsugai Z2 calculation,
energy- and valley-resolved Hall transport tools, and documented serial/MPI
builds into a versioned release. Berry curvature, Chern number, circular
dichroism, and Gamma-point real-space wavefunction output remain available.

Main features

  • Two-dimensional Z2: -z2 1 calculates the invariant from a full, even,
    unshifted Gamma-centered SOC spinor mesh. Time-reversal reconstruction uses
    explicit reciprocal-G mappings and Kramers partners; overlap conditioning
    is measured by SVD and determinant phases are evaluated by LU factorization.
  • Results and plotting: Z2_FIELD.csv records the fundamental n-field,
    numerical checks, half-zone parities, and the invariant. A final PASS CSV is
    published only after the guarded output sequence completes. The plotting
    helper draws the current field as PDF/PNG.
  • Hall transport: opt-in Python tools export high-precision Fukui maps
    and energy- and valley-resolved transport, including cumulative occupied
    subspaces, with explicit isolation and numerical checks.
  • Examples and documentation: Bi and 1H-MoS2 examples are organized under
    examples/. The Bi workflow includes SCF-to-CHGCAR and fixed-charge
    full-mesh input templates, a reference CSV, and an n-field figure. Built-in
    -h help documents the Z2 options and input requirements.
  • Builds: Makefile targets cover GNU serial/Open MPI and Intel compiler
    recipes, with direct-access record-length and BLAS/LAPACK guidance.

Get started

Install Git LFS, GNU Fortran, and LP64 BLAS/LAPACK, then:

git clone --branch v1.2.0 https://github.com/Infant83/VASPBERRY.git
cd VASPBERRY
git lfs pull --include='examples/Bi_Z2/WAVECAR'
make serial
./build/vaspberry-gfortran -h
python3 -m pip install -r requirements-transport.txt
VASPBERRY_BIN="$PWD/build/vaspberry-gfortran" ./examples/Bi_Z2/scripts/run_z2.sh

The Bi runner writes its calculation and plots under
examples/Bi_Z2/results-z2/. Its 12 x 12 reference gives top/bottom sums
-3/+3, matching parities 1/1, and Z2=1 on this mesh.

The Bi WAVECAR is stored with Git LFS. For a runnable example, use the clone
and LFS commands above; downloaded source archives may contain an LFS pointer
depending on GitHub archive settings. Licensed VASP POTCAR files are not
included in this release; supply them locally for new VASP calculations.

Validation and scope

The release is pinned to commit
de942a6.
CI passed
Python 3.10/3.12 regression tests, GNU Fortran 11/13 serial and Open MPI
checks, production-linked Z2 helper tests, and Intel serial compile/help
checks with ifx 2025.0 and ifort 2021.10.
Bi validation
passed the full 12 x 12 calculation in GNU serial and two-rank Open MPI and
confirmed matching discrete fields and agreement of continuous diagnostics.
Intel numerical runs and Intel MPI still require validation on the target
system.

Z2 input must describe a nonmagnetic, time-reversal-symmetric 2D insulator
with an isolated even-rank occupied subspace and even Nx,Ny >= 4. PASS
covers the numerical self-consistency of the time-reversal reconstruction;
physical input symmetry, the band gap, and mesh convergence remain separate
checks. The direct WAVECAR overlap backend omits PAW augmentation terms. The
bundled Bi WAVECAR supports post-processing reproduction; its complete parent
SCF provenance is unavailable.

Documentation and references

Core methods: Fukui, Hatsugai, and Suzuki, JPSJ 74, 1674 (2005)
and Fukui and Hatsugai, JPSJ 76, 053702 (2007).
Additional method and backend references are listed in the guides.

Cite VASPBERRY version 1.2.0 and the methods used in your calculation;
CITATION.cff
contains the software metadata. The 2018 DOI 10.5281/zenodo.1402593
identifies version 1.0 and is not a version-specific DOI for this release.
An importable Python library port is recorded as future work in the roadmap.