Skip to content

Projection layout follow-ups: non-collinear spin channels, axis order and orbital names per parser #293

Closed

Description

@lllangWV

The verifiers of #291 (comments 5981491331 and 5981502918) found these older gaps. #291 did not cause them, but its new check_projected_layout cannot catch them yet.

  • QE non-collinear band structures have 1 spin channel. The projections come out as (k, band, 1, 5, 32) instead of 4 spin channels (total, Sx, Sy, Sz). Abinit non-collinear DOS also has 1 spin channel. The check counts axes only, so both pass.
  • The check cannot see swapped axes. A 5-D array with the spin and atom axes swapped passes. A stronger check would compare shape[:2] with the bands array, the atom axis with structure.natoms when a structure is given, and the spin axis with 1, 2 or 4 consistently with the band count.
  • Orbital names are never compared where a parser passes none. That covers QE band structures and the Abinit, QE, Elk and Lobster DOS. Each parser should pass the names of the orbitals it actually projects.
  • VASP names its orbitals by cutting a fixed list, not by reading the PROCAR header. A filtered PROCAR whose header reads o0 (the bands/auto fixture) is labelled ['s'].

Related: #220 and #219 (QE non-collinear parsing).

Activity

  1. 111 remaining items

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