El-ph calculation on spin-polarized (LSDA) system

Run-time issues concerning Yambo that are not covered in the above forums.

Moderators: Daniele Varsano, andrea marini, Conor Hogan, myrta gruning

User avatar
amolina
Posts: 111
Joined: Fri Jul 15, 2011 11:23 am
Location: Braga, Portugal
Contact:

Re: El-ph calculation on spin-polarized (LSDA) system

Post by amolina » Tue Oct 24, 2017 8:50 am

Dear Faozan,
I have reproduced your error. It seems a bug to me. I will report and we will investigate it. We keep you informed of this issue.
Best,
Alejandro.
Alejandro Molina-Sánchez
INL, Braga
Portugal

User avatar
amolina
Posts: 111
Joined: Fri Jul 15, 2011 11:23 am
Location: Braga, Portugal
Contact:

Re: El-ph calculation on spin-polarized (LSDA) system

Post by amolina » Tue Oct 24, 2017 9:49 am

Dear Faozan,

an option to avoid this problem is to use a random q-grid and to use the noncollinear spin.

I have substitute nspin=2 by
noncolin = .true.
lspinorb = .true.

and now the ph input file is this
&inputph
fildvscf = 'dvscf'
fildyn = 'cuo.dyn'
iverbosity = 1
ldisp = .false.
alpha_mix(1) = 0.4
tr2_ph = 5e-1
prefix = 'cuo'
electron_phonon = 'dvscf'
trans = .true.
qplot = .true.
/
3
0.00000000 0.00000000 0.00000000 1
-0.06424970 -0.41081430 -0.14263660 1
0.46910030 0.19915260 -0.06829270 1

Doing that ypp_ph works and if you are interested in doing el-ph calculations I think is a good solution.

Tell me what you think.
Alejandro Molina-Sánchez
INL, Braga
Portugal

User avatar
amolina
Posts: 111
Joined: Fri Jul 15, 2011 11:23 am
Location: Braga, Portugal
Contact:

Re: El-ph calculation on spin-polarized (LSDA) system

Post by amolina » Tue Oct 24, 2017 9:49 am

Dear Faozan,

an option to avoid this problem is to use a random q-grid and to use the noncollinear spin.

I have substitute nspin=2 by
noncolin = .true.
lspinorb = .true.

and now the ph input file is this
&inputph
fildvscf = 'dvscf'
fildyn = 'cuo.dyn'
iverbosity = 1
ldisp = .false.
alpha_mix(1) = 0.4
tr2_ph = 5e-1
prefix = 'cuo'
electron_phonon = 'dvscf'
trans = .true.
qplot = .true.
/
3
0.00000000 0.00000000 0.00000000 1
-0.06424970 -0.41081430 -0.14263660 1
0.46910030 0.19915260 -0.06829270 1

Doing that ypp_ph works and if you are interested in doing el-ph calculations I think is a good solution.

Tell me what you think.
Alejandro Molina-Sánchez
INL, Braga
Portugal

faozan
Posts: 9
Joined: Sun Sep 17, 2017 2:09 pm

Re: El-ph calculation on spin-polarized (LSDA) system

Post by faozan » Tue Oct 24, 2017 11:30 am

Dear Dr. Alejandro,
amolina wrote: I have reproduced your error. It seems a bug to me. I will report and we will investigate it. We keep you informed of this issue.
Best,
Alejandro.
Yes, I think there is a bug, may be for all cases with nspin=2, will get the same error. Besides my material, I have tried the same case on titanium dioxide.

Best,
Faozan
Faozan Ahmad
PhD Student, ITB, Indonesia

faozan
Posts: 9
Joined: Sun Sep 17, 2017 2:09 pm

Re: El-ph calculation on spin-polarized (LSDA) system

Post by faozan » Tue Oct 24, 2017 11:39 am

Dear Dr. Alejandro,
amolina wrote:Dear Faozan,
an option to avoid this problem is to use a random q-grid and to use the noncollinear spin.
Thank you so much for your proposed solution and your time. I have never used a noncolinear-spin approach, but I will try it soon and hope to get the appropriate electronic structure.

Best regards,
Faozan
Faozan Ahmad
PhD Student, ITB, Indonesia

User avatar
amolina
Posts: 111
Joined: Fri Jul 15, 2011 11:23 am
Location: Braga, Portugal
Contact:

Re: El-ph calculation on spin-polarized (LSDA) system

Post by amolina » Wed Jun 06, 2018 7:44 am

Dear Faozan,
I think your error comes from the number of q-points. You have 36, not 8! This is because using magnetization you are breaking most of the symmetries (if not all of them).
Try to redo your calculations using last_q = 36

Cheers,
Alejandro.
Alejandro Molina-Sánchez
INL, Braga
Portugal

User avatar
luonn
Posts: 11
Joined: Mon Nov 12, 2018 4:52 pm

Re: El-ph calculation on spin-polarized (LSDA) system

Post by luonn » Fri Mar 15, 2019 3:48 am

Dear Faozan:
Have you solved the k-point problem? Actually, you can generate your own random list of q-points.
1. Get a SAVE fold after the nscf calculation, initialize it.
2. ypp_ph -k r, get ypp.in, modify it with: NoWeights, cooOut="alat", BZ_random_NK=n(how many q points you want).
ypp_py -F ypp.in, get o.random_k_pts.
3. Replace the first q point in o.random_k_pts with gamma point, and use the list of q-points in the following ph calculations.
Thus, the problem can be solved.

bests

Nannan Luo
Nannan Luo
Low-Dimensional Materials and Devices Laboratory,
Tsinghua-Berkeley Shenzhen Institute (TBSI), Tsinghua University
C2-7, iPark, Nanshan district, Shenzhen, Guangdong province, China 518055
Tel: (+86)18811369291
E-mail: luo-nn@sz.tsinghua.edu.cn

Post Reply