Re: [Phonopy-users] [help] Query for translation imaginary modes taken into account by phonopy

[Aldo R. <Ald...@ma...>]

Asif

I am sorry but I believe you are wrong… DFPT as implemented in VASP also needs convergence with respect to supercell size…. If you compare DFPT in Quantum Espresso and ABINIT, you will see that in those codes, you have to do a convergence over the Q-mesh but the calculation is done in the unit cell.. in VASP it is not the case… keep that in mind.. Alfonso pointed this out.. he said that the supercell ( independently if you use DFPT or frozen core), needs to be converged because the number of Q points are calculated based on the supercell size and the IFCs can be tricky in many systems… if you check the phonon database published by the ABINIT group, you will see that some systems only converged with Q-meshes of around 8x8x8, this would be the supercell size…

Cheers

-aldo.


______________________::::__________________________________________
Prof. Aldo Humberto Romero
Fellow American Physical Society
Assistant Editor EPJB
Member of the Editor Board of  “Materials”
Physics and Astronomy Department
West Virginia University
Phone: (304) 2936317
email: alr...@ma...


From: BHATTI Asif <Asi...@ls...>
Reply-To: "pho...@li..." <pho...@li...>
Date: Thursday, February 20, 2020 at 2:19 PM
To: "pho...@li..." <pho...@li...>
Subject: Re: [Phonopy-users] [help] Query for translation imaginary modes taken into account by phonopy

Hi,

I tried both approaches I did not see any difference.  Well I will use --fc-symmetry option.

Just to tell you I don't think you need a large supercell for DFPT calculation only if you use frozen phonon calculation then you need to consider a big cell because of the perturbative field effect resulting from the displacement of the atoms within a small cell. The mathematical formalism of DFPT barred us in using the large cell. I think we should read the original paper regarding the assumption being laid.

However, these translation imaginary could be removed I think if I use ADDGRID or increase the NGKX grid to accurate.

Regards,
Asif
________________________________
From: Alex Kent [ake...@gm...]
Sent: Thursday, February 20, 2020 2:36 PM
To: pho...@li...
Subject: Re: [Phonopy-users] [help] Query for translation imaginary modes taken into account by phonopy
Dear Asif,

As prof. Togo should you should use finite displacement method. Also make sure to use large enough super cell.

Also you should use --fc-symmetry option when plotting the phonon dispersion. This may improve the translation invariance, and may improve the phonon dispersion.


HTH...

Alex

On Mon, Feb 10, 2020, 2:23 PM BHATTI Asif <Asi...@ls...<mailto:Asi...@ls...>> wrote:
Hello,

I have a disordered structure and want to know the stability and thermal properties of the structure and this can be obtained through vibrational analysis.

I optimized my structure with VASP until the forces were E=-1E-4. I used DFPT to obtain the Hessian matrix (for this aproach we don't need to build larger supercells). When I checked the OUTCAR there were three imaginary modes (the last three can be ignored because they are translationary modes). Now when I generate the FORCE_CONSTANTS using phonopy it takes into account these modes which can be seen in the DOS and BAND structure.

Should I remove these imaginary translationary modes before submitting them to Phonopy? vasprun.xml doesn't display any imaginary modes only OUTCAR does.

Is this the right procedure? The acoustic soft modes are flat near the Gamma point.

NB: I am attaching a file so you can see where the problem is.

Regards,
Asif

_______________________________________________
Phonopy-users mailing list
Pho...@li...<mailto:Pho...@li...>
https://lists.sourceforge.net/lists/listinfo/phonopy-users