Problem running TB2J with siesta

[islam]

Dear all,

I successfully ran Siesta with the but while running TB2J using the script 

siesta2J.py --fdf_fname Fe6P3.fdf --elements Fe --kmesh 12 12 15 --rcut 7 --emin -9 --nz 100

I am getting the error:

Reading Siesta hamiltonian: colinear spin.
Starting to calculate exchange.
Traceback (most recent call last):
  File "/home/r.islam/.local/bin/siesta2J.py", line 139, in <module>
    run_siesta2J()
  File "/home/r.islam/.local/bin/siesta2J.py", line 120, in run_siesta2J
    gen_exchange_siesta(
  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/manager.py", line 279, in gen_exchange_siesta
    exchange = ExchangeCL2(
  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/exchange.py", line 72, in __init__
    self.set_tbmodels(tbmodels)
  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/exchangeCL2.py", line 30, in set_tbmodels
    self.Gup = TBGreen(self.tbmodel_up,
  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/green.py", line 92, in __init__
    self._prepare_eigen()
  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/green.py", line 149, in _prepare_eigen
    for ik, result in enumerate(results):
  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/sisl_wrapper.py", line 163, in HSE_k
    evalue, evec = eigh(Hk, Sk)
  File "/com/python/3.9.5/lib/python3.9/site-packages/scipy/linalg/decomp.py", line 578, in eigh
    raise LinAlgError('The leading minor of order {} of B is not '
numpy.linalg.LinAlgError: The leading minor of order 154 of B is not positive definite. The factorization of B could not be completed and no eigenvalues or eigenvectors were computed.

I have attached the Fe6P3.fdf and the script job.sh here.

Please help me to solve this issue.

Best regards

Riyajul Islam

[Xu He]

Hi, 

which version of SIESTA did you use? It seems either the Hamiltonian is positive definite or it is not correctly written to the output file, or loaded to TB2J. 

I have encountered some similar problem with one version of SIESTA (don't remember which one) because the hamiltonian file was not written correctly. 

Perhaps you can first try to run with a new SIESTA version. 

Best wishes,

HeXu

[islam]

Hi,

The SIESTA version that I am using is 4.1-b4.

Surprisingly, it used to work well before.

However, I also tried with SIESTA version 4.1.5 and still got the same error.

Best regards,

Riyajul Islam

[Xu He]

Hi, 

Could you send your input files (+pseudopotential) so I can take a look? 

Best regards,

HeXu

--
You received this message because you are subscribed to the Google Groups "TB2J" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tb2j+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tb2j/6c67e77d-af74-49db-ab84-503bf1dd5039n%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[islam]

Here I have attached the input and pseudopotential files.

Can you please tell me which version of SIESTA you are using?

Best regards,

Riyajul Islam

[Xu He]

Hi, 

I run SIESTA and TB2J with your input files and there is no problem.  I am using the most recent version on gitlab. 

https://gitlab.com/siesta-project/siesta/-/tree/master

But I guess most new versions should be fine. The version 4.1 a few years old and there are a lot of bug fixes/improvements since then. I think it is worth upgrading. 

Best regards,

HeXu

[Xu He]

Hi, 

There are some nice suggestion on how to get high-quality wannier functions which can be useful. 

 https://www.wanniertools.org/tutorials/high-quality-wfs

For the siesta version. I would recommend versions with psml support. Note that psml is now mandatory in the main siesta version.

Best wishes,

HeXu 

On Tue, Jun 27, 2023 at 2:12 PM Riyajul Islam <riya...@gmail.com> wrote:

For me, the problem with wannier90 is that since I want to specifically target Mn-3d orbitals for projections, I get the error "dis_windows: Energy window contains fewer states than number of target WFs" if I fix the dis_win to 

dis_win_min = -15
dis_win_max = 25
dis_froz_min = 2
dis_froz_max = 16

The Mn-3d bands however exist between -5 to 10 eV (Fermi level 0 eV). But no matter how wide the dis_win values are I always get the error. So in the previous calculations, I commented out the dis_win_min and dis_froz_min values that might be the issue with the calculations.

For the siesta compilation should I use the PSML version of siesta. I was facing difficulty installing the libpsml so I compile the siesta without psml. Is that could be the case?

Best regards,

Riyajul Islam

On Tue, 27 Jun 2023 at 13:52, Xu He <mail...@gmail.com> wrote:

Hi, 

- Could you send the siesta compilation configurations? If possible send the siesta output files (Hamiltonian, etc). 

 And upgrading sisl might also help as this is the package for reading the Hamiltonian from the siesta output. 

With the Wannier Hamiltonian I get the same error. This is because there is no band from the Wannier Hamiltonian in the energy range defined by the emin (about -10 below Efermi) to the Fermi energy. Indeed from the Wannier center it seems the Wannier functions are quite far away from the atoms and are somehow random.  Could you check if they can well reproduce the DFT band structure? 

Best regards,

HeXu

On Tue, Jun 27, 2023 at 1:27 PM Riyajul Islam <riya...@gmail.com> wrote:

Thank you for the suggestion.

However; the problem still persists even with the new version. Moreover, I am also getting the same error using wannier90 after successfully running wannier90 for Mn13Ga11.

 input.rar

TB2J version 0.7.4.3
Copyright (C) 2018-2020  TB2J group.
This software is distributed with the 2-Clause BSD License, without any warranty. For more details, see the LICENSE file delivered with this software.


Reading atomic structure from file ./POSCAR.
Reading Wannier90 hamiltonian: spin up.
Reading Wannier90 hamiltonian: spin down.

Starting to calculate exchange.
Traceback (most recent call last):

  File "/home/r.islam/.local/bin/wann2J.py", line 186, in <module>
    run_wann2J()
  File "/home/r.islam/.local/bin/wann2J.py", line 156, in run_wann2J
    gen_exchange(path=args.path,
  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/manager.py", line 98, in gen_exchange
    exchange = ExchangeCL2(tbmodels=(tbmodel_up, tbmodel_dn),

  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/exchange.py", line 72, in __init__
    self.set_tbmodels(tbmodels)
  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/exchangeCL2.py", line 30, in set_tbmodels
    self.Gup = TBGreen(self.tbmodel_up,
  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/green.py", line 92, in __init__
    self._prepare_eigen()

  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/green.py", line 156, in _prepare_eigen
    self.evals, self.evecs = self._reduce_eigens(self.evals,
  File "/home/r.islam/.local/lib/python3.9/site-packages/TB2J/green.py", line 98, in _reduce_eigens
    istart, iend = ts[0], ts[-1] + 1
IndexError: index 0 is out of bounds for axis 0 with size 0

Here I have attached the input files.

Please suggest something to solve this issue.

Best regards,

Riyajul Islam

To view this discussion on the web visit https://groups.google.com/d/msgid/tb2j/6f015c54-e85b-4222-ba07-9f2ec3dbc475n%40googlegroups.com.