Skip to content

VASPBERRY 1.6.6 — readable help and CLI consistency

Latest
Latest

Choose a tag to compare

View all tags
@github-actions github-actions released this 04 Oct 06:12
v1.6.6
2e7067b

VASPBERRY 1.6.6 — report reproduction, CLI consistency and readable help

This patch retains the WAVEDER-default charge-Kubo protocol introduced in
1.6.5 and makes commands, input validation, examples and report statements
consistent. The scientific kernels and principal Chern/Kubo reference
results are unchanged.

The maintainer requested the help update of 2026-10-09 in the existing
1.6.6 release. The release body identifies the updated source commit and
its checks. The original 2026-10-04 source remains available at commit
de84ed3.
Record the exact commit together with version 1.6.6 when citing a run.

Readable help — updated 2026-10-09

  • Organize all native task and option pages into plain-text sections for
    commands, inputs, output quantities/units, defaults and validity conditions.
    Keep lines within 80 columns and preserve topic aliases and redirection.
  • Correct legacy flag names, output labels, optical spectrum quantities,
    repeat dimensions and defaults. Distinguish WAVEDER and canonical WAVECAR
    Kubo output schemas and task-specific file replacement rules.
  • Group Python options and expose required inputs, useful defaults/units,
    supported artifacts and scientific scope in each subcommand's own help.
    Clarify direct WAVEDER calculations, native WAVECAR execution and the
    files consumed by plotting commands.
  • Preserve numerical kernels, parsing/selection rules, defaults and output
    schemas. Existing material reference data retain their producer metadata.

For example, run vaspberry --help, vaspberry --help task,
vaspberry --help kubo, or vaspberry --help spin-chern.
For Python, use python tools/vaspberry_kubo.py kubo-hall --help.
See native commands and the
postprocessing reference.

Command and workflow fixes

  • Reject wavefunction, optical and folding options that a standard WAVEDER
    Kubo task does not use; warn when a later band selector overrides an earlier
    one. Keep native --mesh integration separate from legacy mesh settings.
  • Align CLI and INI chemical-potential scans: equal endpoints require one
    point, and multi-point scans require increasing endpoints. Reject a
    nondefault --mu-chunk for the WAVEDER path, where it has no effect.
  • Explain that INI check evaluates the requested WAVEDER response without
    writing results; show resolved optical files and distinguish these checks
    from material convergence. Correct output names, rescan/cache guidance,
    protocol links and the two-endpoint pair-band cutoff description.
  • Stabilize the Intel MPI CI launcher's signal policy for fatal-input tests.
    Controlled exit status, expected diagnostics and output guards remain
    required; native calculations and numerical comparisons are unchanged.
  • Make a best-effort flush of diagnostic streams before fatal termination,
    preserving the existing MPI_ABORT call and error code. Linux MPICH archive
    validation explicitly retains per-process stdout/stderr logs and checks
    them for the required diagnostics. Numerical kernels and rejection guards
    are unchanged.

The standard protocol remains the same. For
supported charge-Kubo calculations, keep matching WAVEDER, WAVECAR, INCAR
and OUTCAR from the same optical run. --input-dir DIR selects their directory;
its default is the working directory. A per-file override affects only that
file. The canonical WAVECAR approximation requires --kubo-source wavecar.
Other task and operator routes retain their own documented inputs.

Report and reference corrections

The VASPBERRY and associated analysis calculations behind all sixteen report
figures, Section 3.6 and five numerical result tables were repeated from
retained inputs. Numerical contents in 42 MoS₂ and 63 native topology/spin
CSV tables matched their references exactly. Fixed Wannier-model analyses
and all five MnBi₂Te₄ Hall quadratures were rerun; one independent postw90
quadrature was also repeated. Other external reference points retain their
historical provenance.

Eight supplementary conditioning cells in four Bi rows now use their actual
48-band source instead of a 64-band fallback. Main Chern/Kubo results and
Figure 12 are unchanged. Two rounded report values now use approximate
wording, and the Figure 2 operator label is fully visible. See the
dated Bi correction,
report Appendix C,
PDF, and
reproduction guide.

Validation and scope

The validation record separates the report recheck
from exact-commit release testing. Reproduction used retained VASP outputs,
operator matrices and fixed Wannier models; it did not regenerate VASP
SCF/NSCF states, rerun operator production or relocalize Wannier functions.
Existing unresolved mesh, band-window and operator limits remain.
Historical producer versions and tags preceding v1.6.6 are preserved.
The help update is a maintainer-approved same-version exception; it does
not repeat or relabel the numerical report reproduction.

Updated source and checks — 2026-10-09

Source commit: 2e7067b.

All thirteen jobs checked this exact updated source, including Intel MPI
(ifx and ifort), actual Bi serial/MPI checks and clean Linux/macOS
source-archive installations. The help update does not repeat or
relabel the original report-wide convergence/reproduction scope.

Original publication checks — 2026-10-04

Source commit: de84ed3.

These four original workflows checked the original de84ed3 source. Historical material-convergence studies retain their original provenance.

Assets 2
Loading