[QE-users] Alpha_mix not resolving ph.x convergence failure
Jessica Johnson
jj611 at soe.rutgers.edu
Mon Jul 11 21:41:46 CEST 2022
Hello,
I wanted to follow up on this thread with a solution to my previous issue and a related, but new issue.
For the system I posted about previously, (the 1T phase of MoS2 with H adsorption), I was able to get the phonons calculation to converge by reducing conv_thr in the scf calculation even further. It was 1e-10; lowering it to 1e-16 led my subsequent phonons calculation to complete.
I am also looking at the 1H phase of MoS2. I am racking my brain, but I cannot figure out what I am doing wrong. I've tried: 1) adjusting alpha_mix(i), 2) lowering the convergence threshold in the scf step, 3) gaussian smearing or no gaussian smearing, 4) increasing the threshold in the ph.x step, 5) repeating the relaxation step with a tighter conv_thr, 6) increasing "maxter" in .f90 file, and possibly some other parameters I am forgetting at the moment.
I am using the norm-conserving, scalar relativistic PBE potential from pseudo dojo (was my choice in pp or functional the mistake?). Please find my input files below:
&CONTROL
calculation = 'scf'
prefix = '1H_0'
pseudo_dir = '.'
title = '1H layers'
verbosity = 'high'
/
&SYSTEM
A = 6.385866
B = 6.385866
C = 14
cosAB = -0.5
cosAC = 0
cosBC = 0
!degauss = 0.005
ecutrho = 368
ecutwfc = 92
ibrav = 4
input_dft = 'BEEF-vdW'
nat = 12
ntyp = 2
!occupations = 'smearing'
!smearing = 'gaussian'
/
&ELECTRONS
conv_thr = 5e-15
/
&IONS
/
&CELL
cell_dofree = 'epitaxial_ab'
/
ATOMIC_SPECIES
Mo 95.95 Mo_HER.upf
S 32.066 S_HER.upf
ATOMIC_POSITIONS (crystal)
Mo -0.0000000000 -0.0000000000 0.2500001541
S 0.1666664325 0.3333333650 0.3617461091
S 0.1666664465 0.3333333930 0.1382538265
Mo 0.0000001753 0.5000000876 0.2499999935
S 0.1666664325 0.8333335675 0.3617461091
S 0.1666664465 0.8333335535 0.1382538265
Mo 0.4999999124 -0.0000001753 0.2499999935
S 0.6666665000 0.3333335000 0.3617463074
S 0.6666665000 0.3333335000 0.1382537513
Mo 0.4999999124 0.5000000876 0.2499999935
S 0.6666666350 0.8333335675 0.3617461091
S 0.6666666070 0.8333335535 0.1382538265
K_POINTS (automatic)
8 8 1 0 0 0
=============================================================
Phonon calculation 1H_0
&inputph
prefix = '1H_0'
fildyn = '1H_0.1.G'
tr2_ph = 1.0d-12
alpha_mix(1) = .4
alpha_mix(10) = .2
alpha_mix(15) = .1
/
0.0 0.0 0.0
Thank you again for your insights.
Jessica Johnson
graduate student
Materials Science and Engineering Department
Rutgers University, New Brunswick, New Jersey
________________________________
From: Jessica Johnson <jj611 at soe.rutgers.edu>
Sent: Monday, May 16, 2022 3:46 PM
To: users at lists.quantum-espresso.org <users at lists.quantum-espresso.org>
Subject: Re: [QE-users] Alpha_mix not resolving ph.x convergence failure
I see- that may do the trick. My university platform will be down for a couple of days for scheduled maintenance, but I will implement this suggestion as soon as it is back up.
Thank you very much for your help,
Jessica
________________________________
From: users <users-bounces at lists.quantum-espresso.org> on behalf of Pietro Davide Delugas <pdelugas at sissa.it>
Sent: Friday, May 13, 2022 2:46 AM
To: users at lists.quantum-espresso.org <users at lists.quantum-espresso.org>
Subject: Re: [QE-users] Alpha_mix not resolving ph.x convergence failure
Hello
many times it works better if you reduce alpha_mix gradually. For example with
alpha_mix(10) = 0.3
and
alpha_mix (15) = 0.1
you use default value for alpha_mix in first 10 iterations, then you reduce it to 0.3, and to 0.1 after 15th iteration.
Pietro
On 5/13/22 00:06, Jessica Johnson wrote:
Hello QE users,
I am having trouble converging a phonons calculation. The archives indicate that many users were able to solve this problem reducing the default value of alpha_mix(1) to .1 or .2. I have tried many values from .01 - .9 and I end up with the calculation being unable to converge. In addition, I have lowered the convergence threshold for the scf calculation thinking that would set up the phonons calculation to converge more easily. I have also varied the nq1, nq2, k1 and k2 values. Neither of the latter two changes made much difference
I am using espresso + d3q (version number I am unsure of, however the release date is Dec. 12, 2021).
My ultimate goal is to use the d3_q2r.x and d3_qha.x executables to compute adsorption of H onto MoS2 layers.
Here are my pw.x and ph.x input files for H adsorbed onto the Mo site:
________________________________________________________
&CONTROL
calculation = 'scf'
prefix = '1T_H_Mo_0'
pseudo_dir = '.'
title = '1T layers'
verbosity = 'high'
/
&SYSTEM
A = 6.385866
B = 6.385866
C = 14
cosAB = -0.5
cosAC = 0
cosBC = 0
degauss = 0.005
ecutrho = 368
ecutwfc = 92
ibrav = 4
input_dft = 'BEEF-vdW'
nat = 13
ntyp = 3
occupations = 'smearing'
smearing = 'gaussian'
/
&ELECTRONS
conv_thr = 1e-10
/
&IONS
/
&CELL
cell_dofree = 'epitaxial_ab'
/
ATOMIC_SPECIES
Mo 95.95 Mo_HER.upf
S 32.066 S_HER.upf
H 1.00794 H_HER.upf
ATOMIC_POSITIONS crystal
Mo 0.0000010000 -0.0000010000 0.2432527978
S 0.3387936710 0.1693960855 0.1415931532
S 0.1716523937 0.3433032874 0.3652990281
H -0.0000005000 0.0000000000 0.3685376338
Mo 0.4770138961 -0.0459752078 0.2481206178
S 0.8306049145 0.1693960855 0.1415931532
S 0.6666670000 0.3333325000 0.3420643699
Mo 0.4770138961 0.5229861039 0.2481206178
S 0.8306049145 0.6612073290 0.1415931532
S 0.6566962126 0.8283471063 0.3652990281
Mo 0.0459752078 0.5229861039 0.2481206178
S 0.3333345000 0.6666665000 0.1232508011
S 0.1716523937 0.8283471063 0.3652990281
K_POINTS automatic
4 4 1 0 0 0
_____________________________________________________________
Phonon calculation 1T_H_Mo_0
&inputph
prefix = '1T_H_Mo_0'
fildyn = '1T_H_Mo_0.1.G'
tr2_ph = 1.0d-14
alpha_mix(1) = .1
ldisp = .true.
nq1 = 4
nq2 = 4
nq3 = 1
/
___________________________________________________________
Thank you in advance for your help on this matter, as it is one of the tougher ones I have encountered using QE.
Jessica Johnson
graduate student
Materials Science and Engineering Department
Rutgers University, New Brunswick, New Jersey
_______________________________________________
The Quantum ESPRESSO community stands by the Ukrainian
people and expresses its concerns about the devastating
effects that the Russian military offensive has on their
country and on the free and peaceful scientific, cultural,
and economic cooperation amongst peoples
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu<https://nam02.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.max-centre.eu%2F&data=05%7C01%7Cjj611%40soe.rutgers.edu%7C57bc9cfdd4cb48b300c708da34ac64ce%7Cb92d2b234d35447093ff69aca6632ffe%7C1%7C0%7C637880212705115756%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=ashc21r4MAZKHTA2u1H9UAHdnEkhp7OxFlw%2FSpgAocM%3D&reserved=0>)
users mailing list users at lists.quantum-espresso.org<mailto:users at lists.quantum-espresso.org>
https://lists.quantum-espresso.org/mailman/listinfo/users<https://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.quantum-espresso.org%2Fmailman%2Flistinfo%2Fusers&data=05%7C01%7Cjj611%40soe.rutgers.edu%7C57bc9cfdd4cb48b300c708da34ac64ce%7Cb92d2b234d35447093ff69aca6632ffe%7C1%7C0%7C637880212705115756%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=3dcqIXC0jHqSnK78RlTrnkoXCq6wO%2B6%2BsZTEYxtB5N8%3D&reserved=0>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20220711/a18f30ab/attachment.html>
More information about the users
mailing list