Eliashberg function output related issue

[anindya@iiita]

Dear Dr. Samuel Ponc,
After the completion of epw.in, I am not getting the correct result of a2f(eliashberg function). I have mentioned my input code below.

--
&inputepw
prefix = 'calc',
amass(1) = 12.0107
outdir = './'

elph = .true.
kmaps = .false.
epbwrite = .true.
epbread = .false.
epwwrite = .true.
epwread = .false.

nbndsub = 8
nbndskip = 0
wannierize = .true.
num_iter = 300
dis_win_max = 11
dis_froz_max= 1
dis_win_min = -3
dis_froz_min= -3
proj(1) = 'random'
system_2d = .true.

wdata(1) = 'bands_plot = .true.'
wdata(2) = 'begin kpoint_path'
wdata(3) = 'G 0.00 0.00 0.00 M 0.50 0.00 0.00'
wdata(4) = 'M 0.50 0.00 0.00 K 0.33 0.33 0.50'
wdata(5) = 'K 0.33 0.33 0.50 G 0.00 0.00 0.00'
wdata(6) = 'end kpoint_path'
wdata(7) = 'bands_plot_format = gnuplot'

iverbosity = 0

elecselfen = .true.
phonselfen = .true.
a2f = .true.
delta_approx= .true.

parallel_k = .true.
parallel_q = .false.

fsthick = 4 ! eV
eptemp = 300 ! K
degaussw = 0.1 ! eV
degaussq = 0.1 ! meV

dvscf_dir = './save'
filukk = './calc.ukk'

nk1 = 16
nk2 = 16
nk3 = 1

nq1 = 8
nq2 = 8
nq3 = 1

nqf1 = 20
nqf2 = 20
nqf3 = 1

nkf1 = 20
nkf2 = 20
nkf3 = 1

/
34 cartesian
0.000000000 0.000000000 0.000000000
0.000000000 0.144337567 0.000000000
0.000000000 0.288675135 0.000000000
0.000000000 0.433012702 0.000000000
0.000000000 -0.577350269 0.000000000
0.125000000 0.072168784 0.000000000
0.125000000 0.216506351 0.000000000
0.125000000 0.360843918 0.000000000
0.125000000 0.505181486 0.000000000
0.125000000 -0.505181486 0.000000000
0.125000000 -0.360843918 0.000000000
0.125000000 -0.216506351 0.000000000
0.125000000 -0.072168784 0.000000000
0.250000000 0.144337567 0.000000000
0.250000000 0.288675135 0.000000000
0.250000000 0.433012702 0.000000000
0.250000000 0.577350269 0.000000000
0.250000000 -0.433012702 0.000000000
0.250000000 -0.288675135 0.000000000
0.250000000 -0.144337567 0.000000000
0.250000000 0.000000000 0.000000000
0.375000000 0.216506351 0.000000000
0.375000000 0.360843918 0.000000000
0.375000000 0.505181486 0.000000000
0.375000000 0.649519053 0.000000000
0.375000000 -0.360843918 0.000000000
0.375000000 -0.216506351 0.000000000
0.375000000 -0.072168784 0.000000000
0.375000000 0.072168784 0.000000000
-0.500000000 -0.288675135 0.000000000
-0.500000000 -0.144337567 0.000000000
-0.500000000 0.000000000 0.000000000
-0.500000000 0.144337567 0.000000000
-0.500000000 -0.866025404 0.000000000

Please suggest me a way to get the converged result.Why the star printed rows are coming? Please help.

Thanks and regards,
Anindya Bose,
JRF, IIIT ALLAHABAD

[anindya@iiita]

Dear epw experts,
Please suggest me a way to achieve convergence in my results.
Thanks and regards,
Anindya Bose

[Vahid]

I am not sure what "Why the star printed rows are coming?" refers to but I see the following areas that you can improve your results/convergence:

1. Going with proj(1) = 'random' could be risky. Did you check to see if the Wannier band structure agrees with the QE's? And if the Wannier functions are real? If so, the projections are likely OK.

2. nkfs and nqfs appear small. Did you try to increase them to see what happens to your results?

3. You may also try different degauss values. There are other threads on this forum with many valuable comments from the developers regarding convergence such as (viewtopic.php?f=6&t=144).

Cheers,

Vahid

Vahid Askarpour
Department of Physics and Atmospheric Science
Dalhousie University,
Halifax, NS, Canada

[anindya@iiita]

Dear Mr Vahid,
Thank you for your response.My wannier band structure of the file prefix_band.dat is entirely different from DFT band structure and also in my case no other projection supported other than projt='random'.Please suggest me how to resolve this issue?

Thanks and regards ,
Anindya Bose

[Vahid]

Dear Anindya,

I find it easier to do the Wannier analysis with the Wannier90 code first. Once I am happy with my Wannier settings (projections, windows, number of bands) and I get agreement between Wannier90 and QE band structures, I move on to EPW.

As for Wannier analysis, you may look into the Wannier90 Examples to see how projections are defined for each system. An understanding of the bonding is essential. So, for example, if you have a covalent bond, then you may define projections centered between bonded atoms. On the other hand, hybrid sp3 bonds may be atom-centered.

I then make sure that the EPW electronic and phonon dispersions agree with QE and then work on the convergence.

I am sure more experienced users have more insight into your questions.

Cheers,

Vahid

[anindya@iiita]

Dear Mr Vahid,
I am a beginner in this field. Can you please share any document or examples where wannier analysis is briefly described?If it require another coding or programming like epw and all then I really like to know about it.I will be waiting for your response.

Thanks and regards,
Anindya Bose

[Vahid]

Dear Anindya,

When you install QE and EPW, the Wannier90 code also gets installed. The Wannier website has a tutorial (http://www.wannier.org/doc/tutorial.pdf) that goes through the examples. The examples come with the installation of QE/EPW. The examples give you some idea as to how to set up energy windows, projections and number of Wannier functions.

Cheers,

Vahid

[anindya@iiita]

Dear Experts,
In my epw calculation(for Graphene), The wannier band structure I am getting is same as pw band structure .Still I am not getting the correct eliashberg function of it.What should I do?Should I play with broadening parameters degaussw and degaussq or should I increase the nkf and nqf values.Which parameter do I need to alter to get correct eliashberg function.Please suggest.

Thanks and regards,
Anindya Bose

[anindya@iiita]

Dear Experts,
Please suggest me a way to get correct converged result of eliashberg function after getting wannier band structure similar to pw band structure.Should I decrease degaussw and degaussq further(degaussw<0.025 eV and degaussq<0.05 meV) and also nkf and nqf grid upto 200*200*1.Can these things give me an appropriate result?I will be waiting for the answer.

Thanks and regards,
Anindya Bose

[anindya@iiita]

Dear Experts,
Please suggest me a way to resolve this issue.

Thanks and regards,

Anindya Bose