Repository navigation
Open
Labels
agent-queueAgent follow-up; worked through by a backlog sweepAgent follow-up; worked through by a backlog sweep
Description
Tracker for spin-channel and orbital-layout defects, moved here from #306 and #307 during triage. Each line is one item for the backlog sweep. Line numbers are on origin/dev at ce6a550 unless the line says it quotes the source issue.
From #307 ("Spin and orbital layout"):
- QE non-collinear DOS has 1 spin channel, not 4 (Parser, spin and phase defects found during the #283-#305 round #307). The parsed DOS is (7435, 1, 5, 32), j-resolved, while the core contract (
check_projected_layout, CLAUDE.md) is 4 channels: total, Sx, Sy, Sz. -
DensityOfStates.is_non_collinearkeys on the orbital count (Parser, spin and phase defects found during the #283-#305 round #307).pyprocar/core/dos.py:448-455returns True for any 32-orbital (2+2+4+4+6+6+8) DOS, so the 1-channel QE spin-orbit DOS reports non-collinear with spin names it cannot index.ProjwfcDOS.is_non_colinear(pyprocar/io/qe/projwfc/projwfc_dos.py:240) checks onlyj, so QE non-collinear without spin-orbit is probably read as collinear (inferred in Parser, spin and phase defects found during the #283-#305 round #307, not run). - One core orbital order at the parser boundary (Parser, spin and phase defects found during the #283-#305 round #307). Each parser hands its own column order to the core; reorder columns into one core order when parsing, so selections mean the same orbitals for every code.
- t2g/eg groups and orbital names are wrong for non-VASP codes (Parser, spin and phase defects found during the #283-#305 round #307).
AZIMUTHAL_ORBITAL_ORDERatpyprocar/core/atomic_orbital_index.py:18spellsdzx/dzy(and wrong f names at :19), and thet2g(4, 5, 7) andeg(5, 6, 8) index groups assume VASP column order, which mislabels QE and Elk selections. - VASP DOS orbital names come from a fixed list (Parser, spin and phase defects found during the #283-#305 round #307), while VASP bands read the PROCAR header.
- Elk task-10 DOS keeps
Y{l}{m}names in the irrep basis (Parser, spin and phase defects found during the #283-#305 round #307). Withlmirepon, the columns are in the irrep basis; fix: resolve the #285 follow-ups (DOS labels, autobandsplot report, BS2D grid, overlay weights, Elk lmirep) #304 renamed only the task-22 band characters. - Lobster DOS is probably all zeros on real files (Parser, spin and phase defects found during the #283-#305 round #307).
DoscarLobster.projected_dos(pyprocar/io/lobster/doscar_lobster.py:163) matches labels likesexactly, while real headers likely carry principal numbers (4s 3d_xy). Inferred in Parser, spin and phase defects found during the #283-#305 round #307, not run.
From #306 (autobandsplot spin-down branches read spin up):
-
simple_energy_windowtakes spin-down bands from spin up (Plotting and script defects found during the #284-#305 round #306).b_down = self.bands_up[kpoint]atpyprocar/scripts/scriptAutoBandsplot.py:123. -
write_reporttakes spin-down defect states from spin up (Plotting and script defects found during the #284-#305 round #306).states_down = self.defect_states[i][0]atpyprocar/scripts/scriptAutoBandsplot.py:353.
Related user reports (spin-orbit and non-collinear channels, orbital labels): #211, #212, #216, #219, #220, #221.
Metadata
Metadata
Assignees
Labels
agent-queueAgent follow-up; worked through by a backlog sweepAgent follow-up; worked through by a backlog sweep
Activity