Response function negative frequency result with PAW+LDA PSP

Phonons, DFPT, electron-phonon, electric-field response, mechanical response…

Moderators: mverstra, joaocarloscabreu

Locked
suvas
Posts: 15
Joined: Sun Nov 11, 2012 4:00 am

Response function negative frequency result with PAW+LDA PSP

Post by suvas » Fri Aug 30, 2013 3:44 pm

Hi Samauel/Boris,

My apology for posting this issue under your solved topic. I could not upload output file due its big size. I have posted the calculated phonon freq with LO/TO splitting. Regarding question of Boris, new latest version abinit 7.4.1does have electric perturbation implemented withing DFPT. I recently installed and checked myself. My problem is with negative frequency.

Thanks

Suvas


Phonon wavevector (reduced coordinates) : 0.00000 0.00000 0.00000
Phonon energies in Hartree :
-1.620450E-03 -1.513718E-03 -1.512787E-03 -1.451621E-03 2.759276E-04
3.775872E-04 4.306877E-04 5.900446E-04 6.964165E-04 6.981539E-04
9.053279E-04 1.065183E-03 1.126469E-03 1.275009E-03 1.285303E-03
1.313353E-03 1.323778E-03 1.326713E-03 1.355389E-03 1.441340E-03
1.470240E-03 1.641299E-03 1.803550E-03 1.805975E-03 1.852514E-03
2.105351E-03 2.177800E-03 2.193856E-03 2.261542E-03 2.278722E-03
2.368199E-03 2.536005E-03 2.571721E-03 2.896839E-03 3.167745E-03
3.169526E-03 3.681068E-03 4.380026E-03 4.623580E-03 4.649283E-03
4.713596E-03 4.727524E-03
Phonon frequencies in cm-1 :
- -3.556476E+02 -3.322226E+02 -3.320184E+02 -3.185939E+02 6.055910E+01
- 8.287080E+01 9.452503E+01 1.294998E+02 1.528458E+02 1.532271E+02
- 1.986965E+02 2.337806E+02 2.472313E+02 2.798320E+02 2.820914E+02
- 2.882477E+02 2.905358E+02 2.911798E+02 2.974736E+02 3.163376E+02
- 3.226804E+02 3.602236E+02 3.958335E+02 3.963656E+02 4.065799E+02
- 4.620712E+02 4.779718E+02 4.814958E+02 4.963512E+02 5.001217E+02
- 5.197596E+02 5.565887E+02 5.644276E+02 6.357827E+02 6.952396E+02
- 6.956306E+02 8.079011E+02 9.613045E+02 1.014759E+03 1.020400E+03
- 1.034515E+03 1.037572E+03

Phonon at Gamma, with non-analyticity in the
direction (cartesian coordinates) 1.00000 0.00000 0.00000
Phonon energies in Hartree :
-1.620450E-03 -1.512787E-03 -1.452258E-03 -1.397813E-03 2.759276E-04
3.775872E-04 5.862242E-04 6.964165E-04 6.981539E-04 9.053279E-04
1.065183E-03 1.126469E-03 1.275009E-03 1.285303E-03 1.313353E-03
1.323778E-03 1.326713E-03 1.355389E-03 1.366846E-03 1.469079E-03
1.597045E-03 1.641299E-03 1.803550E-03 1.852514E-03 2.091009E-03
2.177800E-03 2.193856E-03 2.261542E-03 2.278722E-03 2.368192E-03
2.536005E-03 2.571721E-03 2.896839E-03 3.167745E-03 3.169526E-03
3.617468E-03 4.380026E-03 4.618823E-03 4.649283E-03 4.713596E-03
4.727498E-03 4.943744E-03
Phonon frequencies in cm-1 :
- -3.556476E+02 -3.320184E+02 -3.187337E+02 -3.067845E+02 6.055910E+01
- 8.287080E+01 1.286613E+02 1.528458E+02 1.532271E+02 1.986965E+02
- 2.337806E+02 2.472313E+02 2.798320E+02 2.820914E+02 2.882477E+02
- 2.905358E+02 2.911798E+02 2.974736E+02 2.999880E+02 3.224256E+02
- 3.505109E+02 3.602236E+02 3.958335E+02 4.065799E+02 4.589234E+02
- 4.779718E+02 4.814958E+02 4.963512E+02 5.001217E+02 5.197580E+02
- 5.565887E+02 5.644276E+02 6.357827E+02 6.952396E+02 6.956306E+02
- 7.939425E+02 9.613045E+02 1.013715E+03 1.020400E+03 1.034515E+03
- 1.037566E+03 1.085026E+03

Phonon at Gamma, with non-analyticity in the
direction (cartesian coordinates) 0.00000 1.00000 0.00000
Phonon energies in Hartree :
-1.620450E-03 -1.513718E-03 -1.512787E-03 -1.451621E-03 3.775872E-04
4.306877E-04 5.900446E-04 6.964165E-04 6.981532E-04 9.053279E-04
1.057977E-03 1.126368E-03 1.275009E-03 1.285303E-03 1.297643E-03
1.323103E-03 1.326713E-03 1.355389E-03 1.441340E-03 1.470240E-03
1.641299E-03 1.803550E-03 1.805975E-03 1.852514E-03 2.062593E-03
2.105351E-03 2.177800E-03 2.193856E-03 2.261542E-03 2.368199E-03
2.536005E-03 2.571721E-03 2.896839E-03 3.167730E-03 3.169526E-03
3.681068E-03 4.380026E-03 4.623580E-03 4.649283E-03 4.713596E-03
4.727524E-03 5.214976E-03
Phonon frequencies in cm-1 :
- -3.556476E+02 -3.322226E+02 -3.320184E+02 -3.185939E+02 8.287080E+01
- 9.452503E+01 1.294998E+02 1.528458E+02 1.532269E+02 1.986965E+02
- 2.321991E+02 2.472091E+02 2.798320E+02 2.820914E+02 2.847997E+02
- 2.903875E+02 2.911798E+02 2.974736E+02 3.163376E+02 3.226804E+02
- 3.602236E+02 3.958335E+02 3.963656E+02 4.065799E+02 4.526869E+02
- 4.620712E+02 4.779718E+02 4.814958E+02 4.963512E+02 5.197596E+02
- 5.565887E+02 5.644276E+02 6.357827E+02 6.952364E+02 6.956306E+02
- 8.079011E+02 9.613045E+02 1.014759E+03 1.020400E+03 1.034515E+03
- 1.037572E+03 1.144555E+03

Phonon at Gamma, with non-analyticity in the
direction (cartesian coordinates) 0.00000 0.00000 1.00000
Phonon energies in Hartree :
-1.620435E-03 -1.513718E-03 -1.451621E-03 -1.312054E-03 2.759276E-04
3.775872E-04 4.306877E-04 5.900446E-04 6.964165E-04 6.981539E-04
1.065183E-03 1.126469E-03 1.274489E-03 1.285303E-03 1.313353E-03
1.323778E-03 1.326713E-03 1.355297E-03 1.441340E-03 1.470240E-03
1.623364E-03 1.803477E-03 1.805975E-03 1.852500E-03 2.105351E-03
2.177800E-03 2.193853E-03 2.261542E-03 2.278722E-03 2.368199E-03
2.535992E-03 2.566078E-03 2.894180E-03 3.167745E-03 3.169526E-03
3.681068E-03 3.851559E-03 4.380026E-03 4.623580E-03 4.713232E-03
4.727524E-03 6.042338E-03
Phonon frequencies in cm-1 :
- -3.556443E+02 -3.322226E+02 -3.185939E+02 -2.879626E+02 6.055910E+01
- 8.287080E+01 9.452503E+01 1.294998E+02 1.528458E+02 1.532271E+02
- 2.337806E+02 2.472313E+02 2.797180E+02 2.820914E+02 2.882477E+02
- 2.905358E+02 2.911798E+02 2.974534E+02 3.163376E+02 3.226804E+02
- 3.562873E+02 3.958174E+02 3.963656E+02 4.065767E+02 4.620712E+02
- 4.779718E+02 4.814950E+02 4.963512E+02 5.001217E+02 5.197596E+02
- 5.565860E+02 5.631891E+02 6.351991E+02 6.952396E+02 6.956306E+02
- 8.079011E+02 8.453195E+02 9.613045E+02 1.014759E+03 1.034435E+03
- 1.037572E+03 1.326140E+03

Boris
Posts: 128
Joined: Tue Feb 16, 2010 10:13 am
Location: France

Re: Response function negative frequency result with PAW+LDA

Post by Boris » Fri Aug 30, 2013 9:20 pm

Hi

You seem to have quite a large cell, so issues may arise because PAW+DFPT is not fully operational right now.

Are these the results from ANADDB or from ABINIT ? If they are from ABINIT, try setting RFASR = 1, because frequencies at gamma need to be corrected to account for the acoustic sum rule.

If they are from ANADDB and/or you have already set RFASR = 1, then I suggest you stay tuned for updates on DFPT because we are working on trying to solve this issue of negative frequencies that appear for several systems that we've been studying.

Hope this helps
Boris
----------------------------------------------------------
Boris Dorado
Atomic Energy Commission
France
----------------------------------------------------------

suvas
Posts: 15
Joined: Sun Nov 11, 2012 4:00 am

Re: Response function negative frequency result with PAW+LDA

Post by suvas » Mon Sep 09, 2013 3:46 pm

Hi Boris,

Thanks for your reply. I got negative frequency reslut in abiniit output. I also used anaddb analysis, the results is still same. By the way, new abinit 7.4.2 is released today, is there improvement on PAW phonon code?
I will check my calculation once again with this new version.

Thanks,

Suvas

Boris
Posts: 128
Joined: Tue Feb 16, 2010 10:13 am
Location: France

Re: Response function negative frequency result with PAW+LDA

Post by Boris » Mon Sep 09, 2013 6:29 pm

Hi Suvas

A bug yielding large negative frequencies has been recently identified and corrected. Hopefully, it will solve your problem (it solved mine). However, it won't be corrected in the 7.4.2 version, you will have to wait for the next release.

Boris
----------------------------------------------------------
Boris Dorado
Atomic Energy Commission
France
----------------------------------------------------------

suvas
Posts: 15
Joined: Sun Nov 11, 2012 4:00 am

Re: Response function negative frequency result with PAW+LDA

Post by suvas » Mon Sep 09, 2013 7:38 pm

Hi Boris,

Thanks once again for your quick response. Is there I can do change or modification in subroutine that has bugs associated with. I would be appreciate if you could give me a clue for that.


Thanks,

Suvas

Boris
Posts: 128
Joined: Tue Feb 16, 2010 10:13 am
Location: France

Re: Response function negative frequency result with PAW+LDA

Post by Boris » Wed Sep 11, 2013 1:42 pm

Hi

There is a bug in the kpoint distribution: abinit sometimes makes a funny distribution depending on the number of cpus, the number of kpoints in the ground state calculation, the reduced number of kpoint in the DFPT calculation, and the number of spin polarizations. Unfortunately, we could not find a simple way to get around that, and the kpoint parallelisation has to be rewritten.

It should be available soon.
Boris
----------------------------------------------------------
Boris Dorado
Atomic Energy Commission
France
----------------------------------------------------------

Locked