Dear all,
1) a perennial recommendation of mine is indeed to make sure the cutoffs
for wfcs and rho are both enough to have the stress well converged -
this was one key critiations for the SSSP (that you can find here
https://www.materialscloud.org/discover/sssp/). Reasonably tuned
parameters and appropriate pseudopotentials for any input structure can
be obtained from here:
https://www.materialscloud.org/work/tools/qeinputgenerator
Your mileage might vary, and you might be able to get away with looser
thresholds, or need tighter ones, but that is a recommended starting
point, to understand what you need.
2) a second recommendaiton in dealing with solids is to use PBEsol
rather than PBE as a first default choice.
As always, test crtically these input parameters to understand how they
are affecting your calculations.
nicola
On 26/05/2021 10:28, Stefano de Gironcoli wrote:
more than with a different set of k-points, the last scf is done
resetting the spherical cutoff according to the last cell (during the
vc-relaxation the G vectors in the list are deformed but the miller
indices included in the list are not updated and the sphere actually
deforms into an ellipsoid ).
As the pressure converges slowly with cutoff if the pressure changes
significantly this is an indication that the cutoff might have been too
low or that the cell has changed a lot.
stefano
On 26/05/21 09:40, mkondrin wrote:
Dear Rencong,
I believe that the last step in vc-relax calculations is made in
slightly different Brillouin zone (calculated in relaxed variable cell
parameters), so the pressure calculated in the last step may be
different from whose obtained in the previous step (where Brillouin
zone of the original cell was used convergence criteria with desired
pressure is checked).
It also seems to me that your pressure is rather small. May be it has
sense to increase the pressure to at least 10 kbar.
Sincerely yours,
M. V. Kondrin
On 26.05.2021 05:48, 连云龙 wrote:
Dear Rencong,
I had similar experiences when I use QE pw.x to relax structures
under pressure and I want to follow this conversation.
My experience is that, the behaviour and result of relaxation depend
largely on the pesudopotential.
It is also helpful to investigate the energy and pressure during the
relaxation.
In order for other experts to help you, could you please provide:
(1) the pseudopotentials used in the calculation
(2) the entire output file of pw.x, or at least the section contains
the final coordinates? (search "End of BFGS" in the output file)
(3) your name and affiliation (this is the rule of the mailing list)
Best,
Yunlong
--
Yunlong LIAN ORCID : https://orcid.org/0000-0002-9474-2181 CSNS,
Dongguan Neutron Science Center Dongguan, Guangdong, P. R. China 连云
龙 中国散裂中子源 中子科学部 A1-629 (内线 303) 中国广东省东莞市
-----Original Messages-----
From:Amadeus <[email protected]>
Sent Time:2021-05-25 15:22:02 (Tuesday)
To: users <[email protected]>
Cc:
Subject: [QE-users] under pressure PbTaSe2 vc-relax calculation
Dear QE developer
I did the structural relaxation calculation for Pb at pressure of
3kbar， this is relax.in
&CONTROL
prefix='PbTaSe2',
calculation='vc-relax', pseudo_dir='../../', outdir='./tmp',
verbosity='high',
tprnfor=.true., tstress=.true., forc_conv_thr=1.0d-4, nstep=100,
/
&SYSTEM
ibrav= 0, nat= 4, ntyp= 3,
occupations = 'smearing', smearing = 'gauss', degauss = 1.0d-2,
ecutwfc = 50, ecutrho = 500,
/
&ELECTRONS
conv_thr = 1.0d-8
mixing_beta = 0.7d0
/
&IONS
/
&CELL
cell_dynamics = "bfgs"
press = 2.70000e+00
press_conv_thr =0.1
/
ATOMIC_SPECIES
Ta 180.9479 Ta.UPF
Pb 207.2 Pb.UPF
Se 78.96 Se.UPF
CELL_PARAMETERS (angstrom)
3.4834700000 0.0000000000 0.0000000000
-1.7417350000 3.0167735133 0.0000000000
0.0000000000 0.0000000000 9.4755300000
ATOMIC_POSITIONS (crystal)
Pb 0.0000000000 0.0000000000 0.0000000000
Se 0.0000000000 0.0000000000 0.3246710000
Ta 0.3333330000 0.6666670000 0.5000000000
Se 0.0000000000 0.0000000000 0.6753290000
K_POINTS {automatic}
15 15 5 0 0 0
In relax.in, press=3.0kbar. But after pw.x is done, relax.out show
that total stress (Ry/bohr**3) (kbar) P=-0.83, and This is
inconsistent with press=3. this is part of relax.out
[rencong@mu01 phonon]$ grep 'Computing stress (Cartesian axis) and
pressure' -A 15 relax.out
....
Computing stress (Cartesian axis) and pressure
total stress (Ry/bohr**3) (kbar) P= 2.93
0.00002030 -0.00000000 0.00000000 2.99 -0.00 0.00
-0.00000000 0.00002030 -0.00000000 -0.00 2.99 -0.00
-0.00000000 0.00000000 0.00001921 -0.00 0.00 2.83
kinetic stress (kbar) 26877.23 0.00 0.00
0.00 26877.23 -0.00
0.00 0.00 26796.57
local stress (kbar)-103340.53 0.03 0.00
0.03-103340.50 0.00
0.00 0.00 89394.48
nonloc. stress (kbar) 7797.15 0.00 0.00
--
Computing stress (Cartesian axis) and pressure
total stress (Ry/bohr**3) (kbar) P= 2.96
0.00002009 -0.00000000 0.00000000 2.96 -0.00 0.00
-0.00000000 0.00002009 -0.00000000 -0.00 2.96 -0.00
0.00000000 -0.00000000 0.00002028 0.00 -0.00 2.98
kinetic stress (kbar) 26879.49 0.00 0.00
0.00 26879.49 -0.00
0.00 -0.00 26798.86
local stress (kbar)-103299.75 0.03 0.00
0.03-103299.72 0.00
0.00 0.00 89348.84
nonloc. stress (kbar) 7797.79 -0.00 0.00
--
Computing stress (Cartesian axis) and pressure
total stress (Ry/bohr**3) (kbar) P= -0.83
-0.00000653 -0.00000000 -0.00000000 -0.96 -0.00 -0.00
0.00000000 -0.00000653 0.00000000 0.00 -0.96 0.00
0.00000000 0.00000000 -0.00000384 0.00 0.00 -0.56
kinetic stress (kbar) 26883.07 0.00 -0.00
0.00 26883.07 -0.00
-0.00 -0.00 26802.38
local stress (kbar)-103303.36 0.03 0.00
0.03-103303.33 0.00
0.00 0.00 89345.22
nonloc. stress (kbar) 7795.83 -0.00 0.00
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
users mailing list [email protected]
https://lists.quantum-espresso.org/mailman/listinfo/users
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
users mailing list [email protected]
https://lists.quantum-espresso.org/mailman/listinfo/users
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
users mailing list [email protected]
https://lists.quantum-espresso.org/mailman/listinfo/users
--
----------------------------------------------------------------------
Prof Nicola Marzari, Chair of Theory and Simulation of Materials, EPFL
Director, National Centre for Competence in Research NCCR MARVEL, EPFL
http://theossrv1.epfl.ch/Main/Contact http://nccr-marvel.ch/en/project
_______________________________________________
Quantum ESPRESSO is supported by MaX (www.max-centre.eu)
users mailing list [email protected]
https://lists.quantum-espresso.org/mailman/listinfo/users
