Dear Kenan,
It seems you may be confusing the "num_wann" and "num_bands" inputs. From the manual:
- "num_bands" is "Number of bands passed to the code"
- "num_wann" is "Number of WF"
In other words, "num_bands" are the bands before disentanglement, and "num_wann" the number after disentanglement. Importantly, the number of bands that need to lie within the energy window is "num_wann", not "num_bands".
The way I understand the error message is this: Of the 96 bands that are passed to the code, there are fewer than "num_wann" inside the given energy window. Increasing "num_bands" can fix this because more bands are passed to Wannier90 to begin with. As an extreme example, if "num_wann" > "num_bands", the calculation will never work regardless of energy windows.
In general, "num_bands" should be the same as the number of bands produced by the first-principles code, minus the number of "exclude_bands". I'm not sure exactly what happens when this is not the case, but maybe it is silently ignored in your case.
Best regards,
Dominik
---
Dominik Gresch
Simulation Engineer
Microsoft Quantum
Date: Thu, 1 Apr 2021 17:58:37 +0300
From: Kenan Song <kenan.s...@kaust.edu.sa>
To: hoangminh nguyenduy <nguyenduyhoangm...@gmail.com>
Cc: "wannier@lists.quantum-espresso.org"
<wannier@lists.quantum-espresso.org>
Subject: Re: [Wannier] Enquiry about the Failure in Wannierization
Calculation
Message-ID:
<CAMbJXZ9w6u-bcg7B5YkymnwTYUL8tive=xHYpZurj=c9ad6...@mail.gmail.com>
Content-Type: text/plain; charset="utf-8"
Dear Hoangminh,
Thank you for the suggestion.
I will check that and change the energy window value, as you suggest.
On the other hand, I really do not understand why this problem happened. I
did the wannierization two times. First time, I set num_bands = 112 and
dis_win_max = 150.0; then, the calculation could be done successfully. I
think that this means that 150 eV includes more bands than 112 bands;
otherwise, the calculation would report the error message (dis_windows:
Energy window contains fewer states than number of target WFs).
Second time, I only changed num_bands parameter to 96 and other parameters
remained the same. I suppose that 150 eV should also include far more bands
than 96 bands. Would you please provide me some possible reasons why the
calculation failed, saying 'dis_windows: Energy window contains fewer
states than number of target WFs'?
Thank you very much again.
Kind regards,
Kenan Song
Post Doctor in KAUST
On Thu, Apr 1, 2021 at 5:06 PM hoangminh nguyenduy <
nguyenduyhoangm...@gmail.com> wrote:
> Dear Kenan,
>
>
>
> Although 150 >> 15, would you mind checking whether the highest level when
> num_bands=96 is still higher than the dis_froz_max value?
>
> If the value of dis_froz_max is fine, may you try decreasing the value of
> dis_win_max so that it is just above the 96th band?
>
>
>
> Best,
>
> Minh.
>
> *From: *Kenan Song <kenan.s...@kaust.edu.sa>
> *Sent: *Thursday, April 1, 2021 6:03 PM
> *To: *Stepan Tsirkin <tsirki...@gmail.com>
> *Cc: *wannier@lists.quantum-espresso.org
> *Subject: *Re: [Wannier] Enquiry about the Failure in Wannierization
> Calculation
>
>
>
> Dear Stepan,
>
>
>
> Thank you for the suggestion.
>
>
>
> I used the same VASP code (version 5.4.4) to do the calculation for these
> two calculations. The only difference is that I set NBANDS flag as 112 and
> 96 respectively in order to interface with WANNIER90 code but I found that
> the first wannierization with num_bands=112 could be done successfully;
> while the second one with num_bands=96 failed. All other parameters in the
> wannierization step are the same.
>
>
>
> In the DFT self-consistent calculation, the fermi energy level is located
> at 8.51 eV and the WANNIER90 code took this value as the Fermi level in its
> calculations. The minimum eigen value/band is located at -55 eV but they
> are just 6 highly localised bands (straight lines) and I am interested in
> bands close to Fermi energy level. This is why I choose dis_win_min as -0.5
> eV.
>
>
>
> I could use larger values (160 ~ 200) for num_bands flag but I am afraid
> that the dis_win_max flag needs to be set up as an even higher value. I am
> not sure whether the same error message (dis_windows: Energy window
> contains fewer states than number of target WFs) would pop up.
>
>
>
> Would you please give me some more suggestions/comments? Thank you so much
> for your help.
>
>
>
> Kind regards,
>
>
>
> Kenan Song
>
> Post Doctor in KAUST
>
>
>
> On Thu, Apr 1, 2021 at 12:06 PM Stepan Tsirkin <tsirki...@gmail.com>
> wrote:
>
> Dear Kenan,
>
> do you meant that hte first (successful) calculation was done with another
> code (not VASP) ?
>
> Different codes may produce bandstructures shifted, also depending on
> pseudopotential there might be a different number of semi-core electrons
> included. Are you sure all your bands are above the dis_win_min = -0.5
>
> If you remove dis_win_min and dis_win_max - then all bands will be
> included.
>
> However, for a good disentanglement, typically a much larger number of
> bands should be included, then the number of Wannier functions. I would
> use at least a double, i.e. num_bands~= 160-200.
>
> Regards,
>
> Stepan Tsirkin,
> University of Zurich.
> http://wannier-berri.org
> <https://urldefense.com/v3/__http:/wannier-berri.org__;!!Nmw4Hv0!mcRhIZFtkkQwIAT8TyKbCdjzSk8IetfWcmjwpRZp-udWAbKcLoQ1Q9Vamt-4HoG-iizG$>
>
> On 01.04.21 10:49, Kenan Song wrote:
>
> Dear All,
>
>
>
> Can I ask a question?
>
>
>
> I am running the wannierization now and it could be done successfully with
> the following parameter settings in the wannier90.win file.
>
>
>
> num_wann = 80
>
> num_bands = 112
>
>
>
> num_iter = 50
>
> guiding_centres = .true.
>
> dis_num_iter = 1000
>
>
>
> dis_froz_min = -0.5
>
> dis_froz_max = 15.0
>
> dis_win_min = -0.5
>
> dis_win_max = 150.0
>
>
>
> spinors = .true.
>
> begin projections
>
> Ni : s; p; d
>
> N : s; p
>
> Mn : s; p; d
>
> end projections
>
>
>
> begin unit_cell_cart
>
> 3.8784000000 0.0000000000 0.0000000000
>
> 0.0000000000 3.8784000000 0.0000000000
>
> 0.0000000000 0.0000000000 3.8784000000
>
> end unit_cell_cart
>
>
>
> begin atoms_cart
>
> Ni 0.000000000 0.000000000 0.000000000
>
> N 0.500000000 0.500000000 0.500000000
>
> Mn 0.000000000 0.500000000 0.500000000
>
> Mn 0.500000000 0.000000000 0.500000000
>
> Mn 0.500000000 0.500000000 0.000000000
>
> end atoms_cart
>
>
>
> write_hr = .true.
>
> bands_plot = .true.
>
> bands_num_points = 100
>
> begin kpoint_path
>
> Gamma 0.0 0.0 0.0 X 0.0 0.5 0.0
>
> X 0.0 0.5 0.0 M 0.5 0.5 0.0
>
> M 0.5 0.5 0.0 R 0.5 0.5 0.5
>
> R 0.5 0.5 0.5 Gamma 0.0 0.0 0.0
>
> Gamma 0.0 0.0 0.0 M 0.5 0.5 0.0
>
> M 0.5 0.5 0.0 R 0.5 0.5 0.5
>
> R 0.5 0.5 0.5 X 0.0 0.5 0.0
>
> end kpoint_path
>
> bands_plot_format = gnuplot
>
>
>
> mp_grid = 21 21 21
>
>
>
> begin kpoints
>
> 0.000000000000 0.000000000000 0.000000000000
>
> 0.047619047619 0.000000000000 0.000000000000
>
> 0.095238095238 0.000000000000 0.000000000000
>
> ...
>
> -0.476190476190 0.476190476190 -0.476190476190
>
> 0.476190476190 0.476190476190 -0.476190476190
>
> -0.476190476190 -0.476190476190 0.476190476190
>
> end kpoints
>
>
>
> Then, I re-did the DFT calculation with VASP and set NBANDS to 96 to
> interface with WANNIER90. In the wannierizaion step, I only changed
> num_bands flag from 112 to 96 and kept all other parameter settings the
> same; however, the calculation failed. The wannier90.wout file has the
> following information at the end.
>
>
>
> Error at k-point 1 ndimwin 76 num_wann 80
>
> Exiting.......
>
> dis_windows: Energy window contains fewer states than number of target WFs
>
>
>
> I do not understand why it is that. I suppose that if the num_bands flag
> with 112 in the previous calculation could be done successfully;
> then, num_bands flag with 96 in this new calculation should be done
> successfully as well because the dis_win_max flag in these two calculations
> always remained 150.0.
>
>
>
> Would anyone please give me some suggestions/comment on this issue? Thank
> you very much.
>
>
>
> Kind regards,
>
>
>
> Kenan Song
>
> Post Doctor in KAUST
>
>
>
_______________________________________________ Wannier mailing list Wannier@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/wannier
