Skip to content

Small correctness and cleanup follow-ups from the #277-#279 round #285

Closed

Description

@lllangWV

Found by independent verifiers and owners during #280, #281 and #282. All items exist on dev at 4cde607; none was introduced by those PRs.

Correctness:

  • ElectronicBandStructure.build_overlay_orbitals_weights raises AttributeError: 'list' object has no attribute 'get' at core/ebs.py:1467 and :1527.
  • The VASP parser passes the 16-name orbital table (io/vasp/parser.py:345) for 9-orbital PROCARs to both the EBS and the DOS; any code pairing names with the orbital axis can mis-index (the Unfolder no longer does, since fix(unfold): make pyprocar.unfold draw correct unfolded bands #282).
  • io/abinit/output.py:74,82 uses 27.211396641308 instead of HARTREE_TO_EV 27.211386245988 (bands Fermi off by ~1e-4 eV).
  • A spin-polarized dosplot labels both channels "Total - ↑".
  • autobandsplot writes report.txt into the current working directory.
  • Elk 10.7.8 and later use the irreducible-representation basis for task 22 by default (lmirep=.true.), so the Ylm orbital names are wrong for those runs.
  • The legacy io/procarparser.py:153 has the same fragile k-point regex fix(io): resolve the #279 parser follow-ups for Abinit, QE, VASP, Elk and Siesta #280 fixed in io/vasp/procar.py (fused fixed-width coordinates like 0.50000000-0.25000000).
  • Graphene BS2D surfaces (data/examples/bands/2d-bands/graphene) render jagged and spiky; not investigated.

Cleanup:

  • scripts/procar.py subcommands fermi2D, filter and cat look stale against the current API (unchecked).
  • UnfoldPlotMode.SACATTER and the matching member in scriptBandsplot are misspelled.
  • docs/source/api/cfg/index.rst lists yml files that do not exist; api/cfg/unfold.rst shows stale values (e.g. color: black instead of #eeeeee).
  • Saved notebook outputs are stale: the unfolding notebook, and examples/00-band_structure/00-Getting Started.ipynb:590-591 (lists the removed weighted_color/weighted_width).

Activity

  1. lllangWV commented on Oct 4, 2026

    @lllangWV
    MemberAuthor

    Two more pre-existing items found during #281 round 9 (both reproduce on dev 4cde607):

    • A FermiSurface.load(path) without ebs leaves original_ebs as None; the slicer callback then crashes in _active_scalars_info, so no area text appears.
    • k-points with small jitter (e.g. 1e-7 noise) or missing points crash inside FermiSurface.from_ebs instead of raising a clear error.
  2. lllangWV commented on Oct 4, 2026

    @lllangWV
    MemberAuthor

    Two items are fixed in dev by #291 (d6d4aaa):

    • The VASP parser now passes one orbital name per projected orbital, and the ElectronicBandStructure and DensityOfStates constructors raise on any mismatch (check_projected_layout).
    • io/abinit/output.py now uses HARTREE_TO_EV and AU_TO_ANG from pyprocar/utils/units.py; tests/test_unit_constants.py fails on any hand-typed copy.

    The other items stay open.

  3. added 7 commits that reference this issue on Oct 4, 2026
  4. lllangWV commented on Oct 4, 2026

    @lllangWV
    MemberAuthor

    Fixed in dev by #304. All remaining items are done, as are the overlay-weight AttributeError and the Elk lmirep naming released after #298. The verdicts are on #304. The CHANGELOG has an Unreleased section noting autobandsplot report= and the SCATTER rename (SACATTER is kept as a deprecated alias).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions