Page 1 of 2
Fine q-grid convergence
Posted: Fri Feb 16, 2018 5:28 pm
by Nandan
Hi,
For GaAs I have used two fine q-mesh grids: 1. 50x50x50 and 2. 80x80x80
keeping the fine k-grid the same (a path along L-G-X with 25000 points). But I get two different
results for Im(Sigma) as shown in the link below. They look similar but the magnitudes differ significantly.
http://www.mediafire.com/?f22v94bes7rwaAll other details of the calculation are identical including the machine on which the job was run.
Any idea why this is happening?
Thanks and regards,
Nandan.
Re: Fine q-grid convergence
Posted: Wed Feb 21, 2018 8:34 pm
by Nandan
If it helps, I am using the following input file:
--
&inputepw
prefix = 'gaas'
amass(1) = 69.72300
amass(2) = 74.92160
outdir = './'
! ephwrite = .true.
elph = .true.
kmaps = .true.
epbwrite = .false.
epbread = .false.
epwwrite = .false.
epwread = .true.
etf_mem = 0
nbndsub = 12
nbndskip = 0
dis_win_max = 29
dis_froz_max= 14
dis_froz_min= -8
restart = .true.
restart_freq= 2000
wannierize = .false.
lpolar = .true.
num_iter = 5000
iprint = 2
proj(1) = 'random'
proj(2) = 'As:sp3'
! efermi_read = .false.
! fermi_energy= 5.4304
elecselfen = .true.
phonselfen = .false.
band_plot = .true.
parallel_k = .true.
parallel_q = .false.
iverbosity = 3
fsthick = 25
eptemp = 300.d0
degaussw = 0.005
dvscf_dir = '../phonons/save'
filukk = './gaas.ukk'
filkf = './LGX-25K.pwscf'
nk1 = 6
nk2 = 6
nk3 = 6
nq1 = 6
nq2 = 6
nq3 = 6
! nkf1 = 20
! nkf2 = 20
! nkf3 = 20
!
nqf1 = 50
nqf2 = 50
nqf3 = 50
! rand_k = .true.
! rand_nk = 8000
!mp_mesh_k = .true.
! rand_q = .true.
! rand_nq = 125000
/
16 cartesian
0.0000000 0.0000000 0.0000000 0.0092593
-0.1666667 0.1666667 -0.1666667 0.0740741
-0.3333333 0.3333333 -0.3333333 0.0740741
0.5000000 -0.5000000 0.5000000 0.0370370
0.0000000 0.3333333 0.0000000 0.0555556
-0.1666667 0.5000000 -0.1666667 0.2222222
0.6666667 -0.3333333 0.6666667 0.2222222
0.5000000 -0.1666667 0.5000000 0.2222222
0.3333333 0.0000000 0.3333333 0.1111111
0.0000000 0.6666667 0.0000000 0.0555556
0.8333333 -0.1666667 0.8333333 0.2222222
0.6666667 -0.0000000 0.6666667 0.1111111
0.0000000 -1.0000000 0.0000000 0.0277778
0.6666667 -0.3333333 1.0000000 0.2222222
0.5000000 -0.1666667 0.8333333 0.2222222
-0.3333333 -1.0000000 0.0000000 0.1111111
Nandan.
Re: Fine q-grid convergence
Posted: Wed Feb 21, 2018 10:28 pm
by sponce
Hello Nandan,
I do not see any obvious issue.
Do you really need 25,000 points along two lines ?
What I would do is to reduce the number of k-point to 250 instead and then do a few q-grids (homogeneous and random).
Note that in polar materials it will converge slowly with the q-grids (you probably need 300x300x300 or so).
You should use random q-grid or better some grids that over-sample the divergence in Gamma (for example a log grid).
Make sure that the weights are correct.
Best,
Samuel
Re: Fine q-grid convergence
Posted: Thu Feb 22, 2018 4:42 pm
by Nandan
Hello Samuel,
Thanks for the reply.
I had not considered the weights of the k-points at all. I have been using constant weights for
the k-points in the list. For N-kpoints, the weight for each is 1/N.
How do I create a k-list with correct weights?
Nandan.
Re: Fine q-grid convergence
Posted: Fri Feb 23, 2018 2:53 pm
by sponce
If you are interested in linewidths (so values for a given k-point) it does not matter.
However the weight for the q-points are important since you do a q-point integration.
The weight of the q-points should be equal to their volume in the BZ.
For a log grid, you will have much more points close to gamma and therefore their weight should be smaller.
Best,
Samuel
Re: Fine q-grid convergence
Posted: Tue Feb 27, 2018 4:06 pm
by Nandan
Hi Samuel,
So after doing another calculation with a denser fine q-grid of 10^6 points and a coarse k-grid of 250 (red) points (or 5000:green) along L-G-X, I get the
scattering rates (in terms of Im(Sigma) meV) as shown below:
http://www.mediafire.com/file/l70neisng ... -1mill.epsIf this is compared with Phys. Rev. B 94, 201201(R) (2016), Fig 1, then the magnitudes of Im(sigma) are atleast an order higher
in my calculation.
Where as for a coarse grid as I had done earlier in
http://www.mediafire.com/file/9173kpgq2 ... h-comp.eps,
the magnitude is comparable to the reference.
I am trying to understand why is the magnitude so different in the two cases compared to the above reference?
Nandan.
Re: Fine q-grid convergence
Posted: Tue Feb 27, 2018 4:31 pm
by carla.verdi
Dear Nandan,
Have you checked that the interpolated band structure and phonons that you get from epw are correct?
Best,
Carla
Re: Fine q-grid convergence
Posted: Tue Feb 27, 2018 8:43 pm
by Nandan
Hi Cara,
The interpolated electrons and phonons look fine. There is a small negative frequency near Gamma,
but overall it looks alright.
phonons:
http://www.mediafire.com/file/hxye197xq ... phonon.epselectrons:
http://www.mediafire.com/file/9rihxg4mw ... ctrons.epsNandan.
Re: Fine q-grid convergence
Posted: Tue Feb 27, 2018 8:45 pm
by Nandan
Carla,
Sorry for spelling your name wrong earlier.
nandan.
Re: Fine q-grid convergence
Posted: Tue Mar 06, 2018 9:34 pm
by Nandan
I am wondering if this could be compiler related issue?
Or do the negative phonon frequencies near Gamma cause this behavior in Im(sigma)?
Thanks and regards,
Nandan.