[QE-users] Convergence and restart of hp.x
Timrov Iurii
iurii.timrov at epfl.ch
Mon Aug 17 12:09:36 CEST 2020
Dear Prasenjit,
> It is metallic at PBE level and it is expected to be so.
Ok, so I would use degauss=0.02 Ry and the 'MV' smearing method.
> Let me try with your suggestions and see if I can make the calculations converge.
Ok
Greetings,
Iurii
--
Dr. Iurii TIMROV
Postdoctoral Researcher
STI - IMX - THEOS and NCCR - MARVEL
Swiss Federal Institute of Technology Lausanne (EPFL)
CH-1015 Lausanne, Switzerland
+41 21 69 34 881
http://people.epfl.ch/265334
________________________________
From: users <users-bounces at lists.quantum-espresso.org> on behalf of Prasenjit Ghosh <prasenjit.jnc at gmail.com>
Sent: Monday, August 17, 2020 11:38:44 AM
To: Quantum ESPRESSO users Forum
Subject: Re: [QE-users] Convergence and restart of hp.x
Dear Iurii,
Thanks for your response.
For the HP calculation you can try alpha_mix(1)=0.1 or even 0.05. Also try to use niter_max = 200 or 300.
I have submitted one calculation with niter_max=300. It is running. Let's see what happens.
For the SCF calculation degauss=0.004 Ry is very small. The system is gapped or metallic at the PBE level? Maybe you can try larger values for degauss? Also you can try higher values for ecutwfc and ecutrho.
It is metallic at PBE level and it is expected to be so. The actual system that I am interested in is a composite of this one with another material. However, that involves a larger unit cell. So I thought that it would be best to do it with this smaller one and understand the method before going to a larger system.
> When I go to higher k-mesh, some of the k-points are not converging (within 100 iterations).
Do you mean some q points that do not converge?
Yes. For my system with 2x2x2 q-grid I get the following q-points:
1 0.000000000 0.000000000 0.000000000 0.125000000
2 0.000000000 0.000000000 -0.500000000 0.375000000
3 0.000000000 -0.500000000 -0.500000000 0.375000000
4 -0.500000000 -0.500000000 -0.500000000 0.125000000
The hp.x calculation converged for all the q-points.
Using a 3x3x3 q-grid, I am getting the following q-points:
1 0.000000000 0.000000000 0.000000000 0.037037037
2 0.000000000 0.000000000 0.333333333 0.222222222
3 0.000000000 0.333333333 0.333333333 0.444444444
4 0.333333333 0.333333333 0.333333333 0.296296296
So at q=(0,0,0) and q=(0,0.1/3), the calculation converged. However, for the third one it is not converging.
Let me try with your suggestions and see if I can make the calculations converge. I will keep you posted.
With regards,
Prasenjit
Greetings,
Iurii
--
Dr. Iurii Timrov
Postdoctoral Researcher
STI - IMX - THEOS and NCCR - MARVEL
Swiss Federal Institute of Technology Lausanne (EPFL)
CH-1015 Lausanne, Switzerland
+41 21 69 34 881
http://people.epfl.ch/265334
________________________________
From: users <users-bounces at lists.quantum-espresso.org<mailto:users-bounces at lists.quantum-espresso.org>> on behalf of Prasenjit Ghosh <prasenjit.jnc at gmail.com<mailto:prasenjit.jnc at gmail.com>>
Sent: Monday, August 17, 2020 10:57:41 AM
To: Quantum ESPRESSO users Forum
Subject: Re: [QE-users] Convergence and restart of hp.x
Dear Iurii,
Thanks for your prompt response. I would just like to point out that the calculation is converging for a smaller q-mesh, 2x2x2 in my case. When I go to higher k-mesh, some of the k-points are not converging (within 100 iterations). I have not yet tried increasing the no. of iterations.
Given below is the input scf file:
&control
calculation = 'scf'
restart_mode='from_scratch',
prefix='tifesb',
pseudo_dir = './',
outdir= './tmp_tifesb'
/
&system
ibrav= 1, celldm(1)= 11.343477089,
nat=12, ntyp= 3,
ecutwfc = 60, ecutrho=500,
occupations='smearing'
smearing='mv'
degauss=0.004
lda_plus_u=.true.
lda_plus_u_kind=0
U_projection_type='atomic'
Hubbard_U(2)=1.0d-10
Hubbard_U(3)=1.0d-10
/
&electrons
diagonalization='david'
mixing_mode = 'local-TF'
mixing_beta = 0.3
conv_thr = 1.0d-07
/
&ions
/
&cell
cell_dofree='xyz'
/
ATOMIC_SPECIES
Sb 50.9415 Sb.pbe-n-rrkjus_psl.1.0.0.UPF
Fe 26.981539 Fe.pbe-sp-van_mit.UPF
Ti 26.981539 Ti.pbe-spn-rrkjus_psl.1.0.0.UPF
ATOMIC_POSITIONS (crystal)
Ti 0.000000000 0.000000000 0.000000000
Ti 0.500000000 0.500000000 0.000000000
Ti 0.000000000 0.500000000 0.500000000
Ti 0.500000000 0.000000000 0.500000000
Fe 0.250000000 0.250000000 0.250000000
Fe 0.750000000 0.750000000 0.250000000
Fe 0.250000000 0.750000000 0.750000000
Fe 0.750000000 0.250000000 0.750000000
Sb 0.500000000 0.500000000 0.500000000
Sb 1.000000000 1.000000000 0.500000000
Sb 0.500000000 1.000000000 1.000000000
Sb 1.000000000 0.500000000 1.000000000
K_POINTS {automatic}
12 12 12 1 1 1
The input for the hp.x is given below:
&inputhp
prefix='tifesb'
outdir='./tmp_tifesb'
nq1=3, nq2=3, nq3=3
conv_thr_chi = 1.0d-5
iverbosity = 2
alpha_mix(1)=0.2
/
With regards,
Prasenjit
On Mon, 17 Aug 2020 at 13:58, Timrov Iurii <iurii.timrov at epfl.ch<mailto:iurii.timrov at epfl.ch>> wrote:
Dear Prasenjit Ghosh,
> Can someone please indicate other input parameters that can be tuned to reach convergence?
You can check the HP documentation, but there are not that many parameters that can help to converge the HP calculation. Typically, the reason why HP does not converge is because there are problems with the SCF calculation. Can you share with us your input files for PWscf and HP? This will help us to understand what can be the problem.
> Also is there a way to restart the hp.x calculations?
Yes. Check examples 7 in the HP repository. You can restart from a q point which was not converged (note that for this q point the calculation will be done from scratch).
Greetings,
Iurii
--
Dr. Iurii Timrov
Postdoctoral Researcher
STI - IMX - THEOS and NCCR - MARVEL
Swiss Federal Institute of Technology Lausanne (EPFL)
CH-1015 Lausanne, Switzerland
+41 21 69 34 881
http://people.epfl.ch/265334
________________________________
From: users <users-bounces at lists.quantum-espresso.org<mailto:users-bounces at lists.quantum-espresso.org>> on behalf of Prasenjit Ghosh <prasenjit.jnc at gmail.com<mailto:prasenjit.jnc at gmail.com>>
Sent: Monday, August 17, 2020 5:10:56 AM
To: Quantum Espresso users Forum
Subject: [QE-users] Convergence and restart of hp.x
Dear all,
I am using hp.x to compute the U value. I have two types of Hubbard atoms in my unit cell. I am using qe-6.5.
For my system with a 2x2x2 q-grid, the calculations converge. However, the moment I use a denser grid, the calculations don't converge, even after playing around with the mixing parameter.
Can someone please indicate other input parameters that can be tuned to reach convergence?
Also is there a way to restart the hp.x calculations?
With regards,
Prasenjit
--
PRASENJIT GHOSH,
IISER Pune,
Dr. Homi Bhabha Road, Pashan
Pune, Maharashtra 411008, India
Phone: +91 (20) 2590 8203
Fax: +91 (20) 2589 9790
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso<http://www.max-centre.eu/quantum-espresso>)
users mailing list users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>
https://lists.quantum-espresso.org/mailman/listinfo/users
--
PRASENJIT GHOSH,
IISER Pune,
Dr. Homi Bhabha Road, Pashan
Pune, Maharashtra 411008, India
Phone: +91 (20) 2590 8203
Fax: +91 (20) 2589 9790
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso<http://www.max-centre.eu/quantum-espresso>)
users mailing list users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>
https://lists.quantum-espresso.org/mailman/listinfo/users
--
PRASENJIT GHOSH,
IISER Pune,
Dr. Homi Bhabha Road, Pashan
Pune, Maharashtra 411008, India
Phone: +91 (20) 2590 8203
Fax: +91 (20) 2589 9790
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200817/1c5f9122/attachment.html>
More information about the users
mailing list