[Wien] LMBJ failure for 2D BaSe (Error in NLVDW)

[karina Khan]

  Dear Wien2k community and developers,

I am encountering an issue with LMBJ calculations for 2D BaSe monolayer
using Wien2k 23.2. The nlvdw step crashes immediately with "Error in
NLVDW",
while the identical workflow works perfectly for CaSe and SrSe 2D
monolayers.

== System Details ==

Material: 2D BaSe monolayer (PbO-type structure)
Space group: 123 (P4/mmm)
Lattice: a = b = 8.587093 Bohr, c = 42.038689 Bohr (~22 Å vacuum)
Atoms: Ba at (0, 0, 0.5), Se at (0.5, 0.5, 0.5)
RMT: Ba = 2.5, Se = 2.3
Spin-unpolarized calculation

== Key Parameters ==

case.in1:
  RKmax = 7.0, MAX L IN WF = 10, V-NMT = 4
  Local orbitals included for Ba 4d (-6.21 Ry), 5s (-2.17 Ry),
  5p (-1.03 Ry), and Se 3d (-3.57 Ry)
  Energy cutoff used in lstart: -8.0 Ry

case.in2:
  EMIN = -14.0 Ry, NE = 36
  Lattice harmonics: max L = 6
  GMAX = 12.00

case.innlvdw (identical to working CaSe):
  kernel type: 1
  kernel parameter: -0.8491
  GMAX: 25
  rhoc: 0.3
  potential: T
  GMAXpot: 10
  sigma: 3.78 bohr

== What Works ==

PBE SCF: Converges cleanly in 11 iterations.
  - :NEC = 90.00000 / 90.00000000 (zero deviation)
  - :ENE change < 1e-7 Ry between last iterations
  - :DIS = 0.0000276 (well below threshold)
  - :GAP (global) = 0.148086 Ry = 2.015 eV
  - No warnings

CaSe and SrSe 2D monolayers: LMBJ works without any issues
(same workflow, same parameter ranges, same Wien2k version).

== The Problem ==

After:
  save_lapw BaSe_2D_pbe
  init_mbj_lapw (chose: 2 for lmBJ, then 0 for default Koller 2012)
  run_lapw -i 1 -NI -p
  save_lapw PBE
  init_mbj_lapw (chose: 2 for lmBJ, then 0)
  run_lapw -lmbj -i 100 -ec 0.0001 -cc 0.0001 -p

The dayfile shows:
  cycle 1 (...) (100/99 to go)
  > lapw0 -grr 5.144u 0.121s 0:01.71 307.6%
  > nlvdw -lmbj 0.007u 0.008s 0:00.01 0.0% <-- crashes immediately
  > stop error

The nlvdw.error file contains only: "Error in NLVDW"

== Diagnostic Information ==

- nlvdw binary exists at /opt/WIEN_23/nlvdw
- All required input files exist with correct sizes:
  BaSe_2D.r2v: 565,653 bytes
  BaSe_2D.r2vdn: 565,653 bytes (copy of r2v for spin-unpolarized)
  BaSe_2D.tausum: 598,913 bytes
  BaSe_2D.innlvdw: parameters identical to working CaSe/SrSe
- Removed all empty/stale tauup, taudn files
- Tested both serial and parallel runs - same error
- case.in0 has XC_LMBJ correctly set
- 2*lnsmax (=8) >= lmax in case.in2 (=6), so this is not the
  lmax-in-in2 vs. lnsmax issue from the older nlvdw thread (msg16382, 2017)

== What I Have Tried ==

1. Compared every parameter with working CaSe directory - identical except
   for atom-specific values (EMIN, NE, GMAX which is 12 vs 16).
2. Tried with -p and without -p (parallel/serial)
3. Verified all files have non-zero sizes
4. Manually copied r2v_grr -> r2v (Wien2k 23.2 uses _grr suffix)
5. Restored tausum from saved PBE state

== Question ==

Could this be a numerical issue specific to heavy elements with deep
semicore d-states (Ba 4d at -6.21 Ry)? The fact that CaSe/SrSe work
but BaSe doesn't, with the same workflow and parameter ranges,
suggests a Ba-specific limitation in nlvdw -lmbj.

Is there any known workaround for LMBJ with heavy alkaline earth
elements like Ba? Or should I expect regular mBJ to give comparable
results given my 22 Å vacuum thickness?

I am happy to provide case.struct, case.in0, case.in1, case.in2,
case.innlvdw or any other files if needed for diagnosis.

Thank you for any guidance!

*Thanks and Regards*

*Dr. Karina *

*| **Google Scholar
<https://scholar.google.com/citations?user=2_K6k2MAAAAJ&hl=en> |*
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20260519/4a42dd27/attachment.htm>