Symmetries and ICHARG mode

[the-hampel]

Hi, Hampel,
Thank you very much for this timely information. Here are some more questions.

1. Symmetry adaption in the converting step. Here are the details.
   For the DFT calculation, I used vasp.6.5.1 and turn the symmetry on (#ISYM=-1) in the INCAR, which can output the vaspout.h5 file. But during the converting step, an error occurs as follows:

Warning: could not identify MPI environment!
Starting serial run at: 2025-09-28 20:59:15.815476
[WARNING]: Error reading Efermi from LOCPROJ, trying DOSCAR...
Read parameters: LOCPROJ
0  ->  {'label': 'dxy', 'isite': 2, 'l': 2, 'm': 0}
1  ->  {'label': 'dyz', 'isite': 2, 'l': 2, 'm': 1}
2  ->  {'label': 'dz2', 'isite': 2, 'l': 2, 'm': 2}
3  ->  {'label': 'dxz', 'isite': 2, 'l': 2, 'm': 3}
4  ->  {'label': 'dx2-y2', 'isite': 2, 'l': 2, 'm': 4}
  Found POSCAR, title line: SrVO3
  Total number of ions: 5
  Number of types: 3
  Number of ions for each type: [1, 1, 3]

    Total number of k-points: 35
   Total number of tetrahedra: 353
Traceback (most recent call last):
  File "/svo-csc-plo/scf/converter.py", line 5, in <module>
    plo_converter.generate_and_output_as_text('plo.cfg', vasp_dir='./')
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/anaconda3/envs/triqs/lib/python3.13/site-packages/triqs_dft_tools/converters/plovasp/converter.py", line 73, in generate_and_output_as_text
    el_struct = ElectronicStructure(vasp_data)
  File "/anaconda3/envs/triqs/lib/python3.13/site-packages/triqs_dft_tools/converters/plovasp/elstruct.py", line 89, in __init__
    assert nk_plo == self.nktot, "PLOCAR is inconsistent with IBZKPT (number of k-points)"
           ^^^^^^^^^^^^^^^^^^^^
AssertionError: PLOCAR is inconsistent with IBZKPT (number of k-points)

The input of the converter.py is

from triqs_dft_tools.converters.vasp import *
import triqs_dft_tools.converters.plovasp.converter as plo_converter

# Generate and store PLOs
plo_converter.generate_and_output_as_text('plo.cfg', vasp_dir='./')

# run the converter
Converter = VaspConverter(filename = 'vasp')
Converter.convert_dft_input()

and the plo.cfg is

[General]
#DOSMESH = -3.0 5.0 4001

[Group 1]
SHELLS = 1
NORMALIZE = True
EWINDOW = -1.4 2.0

[Shell 1]
LSHELL = 2
IONS = 2

TRANSFORM = 1.0  0.0  0.0  0.0  0.0
            0.0  1.0  0.0  0.0  0.0
            0.0  0.0  0.0  1.0  0.0

So I wonder if it's still not available to consider the symmetry of VASP so far?

2. VASP-ICHARG=5 problems. Actually, I am not sure when use turn this tag on, and I tried for two cases. Here are the details.
   2a) Without initial run with ICHARG=2, using only INCAR, KPOINTS, POTCAR and POSCAR to run, I got stuck but no error has been reported. Here are part of the log, and you may see the whole test in the attachment (initial-ICHARG5 folder):

 Reading from existing POTCAR
 LDA part: xc-table for (Slater+PW92), standard interpolation
 POSCAR, INCAR and KPOINTS ok, starting setup
 FFT: planning ... GRIDC
 FFT: planning ... GRID_SOFT
 FFT: planning ... GRID
 WAVECAR not read
 WARNING: chargedensity file is incomplete
 entering main loop
       N       E                     dE             d eps       ncg     rms          rms(c)
 ELPA: falling back to scaLAPACK
DAV:   1     0.194678777216E+03    0.19468E+03   -0.11478E+04  1716   0.105E+03
DAV:   2    -0.364410937904E+02   -0.23112E+03   -0.22584E+03  2148   0.269E+02
DAV:   3    -0.491754035442E+02   -0.12734E+02   -0.12702E+02  1992   0.787E+01
DAV:   4    -0.495377321012E+02   -0.36233E+00   -0.36235E+00  2760   0.114E+01

2b) Run with ICHARG=5 after an initial run with ICHARG=2, either GAMMA file nor vaspgamma.h5 is generated, and the code also got stuck, but at the first iteration (initial-ICHARG2 folder). Here is the log file:

 Reading from existing POTCAR
 LDA part: xc-table for (Slater+PW92), standard interpolation
 found WAVECAR, reading the header

 POSCAR, INCAR and KPOINTS ok, starting setup
 FFT: planning ... GRIDC
 FFT: planning ... GRID_SOFT
 FFT: planning ... GRID
 reading WAVECAR
 the WAVECAR file was read successfully
 charge-density read from file: SrVO3
 entering main loop
       N       E                     dE             d eps       ncg     rms          rms(c)
 ELPA: falling back to scaLAPACK

initial-ICHARG2.zip

initial-ICHARG5.zip

Originally posted by @Fenice-Jin in #108

[the-hampel]

Hi @Fenice-Jin,

I moved this to a new issue. Please do open a new issue for each question. This is easier for me to answer.

1. So I wonder if it's still not available to consider the symmetry of VASP so far?

The option to use symmetries in VASP together with the TRIQS interface is a new development on both sides. It is supported in VASP 6.5.x , but only the unstable branch of dft_tools has the PR merged: TRIQS/dft_tools#262 . You can test the feature if you install the unstable branch of dft_tools manually.

2. VASP-ICHARG=5 problems

ICHARG=5 is solely indented to be used when running a full CSC calculation: https://triqs.github.io/solid_dmft/tutorials/PrNiO3_csc_vasp_plo_cthyb/tutorial.html or see here https://triqs.github.io/dft_tools/unstable/guide/dftdmft_selfcons.html#vasp-plovasp . Running VASP without solid_dmft is little meaningful here, because VASP will wait for the dmft code to "communicate". Can you rerun the ICHARG=5 example once by starting it through solid_dmft as intended? However this does not explain why your ICHARG=2 calculation stops. Is this the same if you startk without WAVECAR / CHGCAR present?

Best,
Alex

[the-hampel]

Hi @Fenice-Jin ,

is still a problem for you? If so please give a ping here and we can work on it again.

[Fenice-Jin]

Hi @Fenice-Jin ,

is still a problem for you? If so please give a ping here and we can work on it again.

Hi, Alex,
Thank you very much for your reply. I followed the following instructions (https://triqs.github.io/solid_dmft/tutorials/PrNiO3_csc_vasp_plo_cthyb/tutorial.html) and the results seems good enough compared with the references.
Here are some information when I modify the VASP source code following this page (https://triqs.github.io/dft_tools/latest/guide/conv_vasp.html#remarks-on-the-vasp-version). The LPRJ_WRITE still exists in the main.F for VASP-6.1.2 but no longer exists for VASP-6.4.2. Instead, the LPRJ_WRITE exists in the mlwf.F. I modified the code in VASP-6.4.2, and the modified vasp_std code for ICHARG=1 in the INCAR only gives zero value of E-fermi in the LOCPROJ, but if I start the CSC DMFT calculation after the ICHARG=1 calculation with ICHARG=5, the E-fermi in the LOCPROJ file can be well written.
Besides, here is a small problem regarding the parallel computation of VASP in the solid_dmft calculation. Only n_cores = 1 set in the dmft_config.toml can work. Using n_cores more than 1 would be stuck when reading the WAVECAR. As you've mentioned using OMP_NUM_THREADS instead, I am not sure how to set this parameter. Do I need to set OMP_NUM_THREADS equal to the desired number of cores in the job submission script and also set n_cores equal to the desired number?
Best regards,
Jin

[the-hampel]

Hi @Fenice-Jin ,
thank you for the feedback. I think what should work is to use a job script like this:

#!/bin/bash

#SBATCH --partition=nameofpart
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=2
#SBATCH --cpus-per-task=8
#SBATCH --cpu-bind=cores
#SBATCH --time=12:00:00

module load ...

export OMP_NUM_THREADS=$SLURM_CPUS_PER_TASK
export OMP_STACKSIZE=2048m
export OMP_PLACES=cores
export OMP_PROC_BIND=close


srun solid_dmft

For that VASP needs to be compiled with -D_OPENMP .