## Which band and K-points contribute more for lambda

General discussion around the EPW software

Moderator: hlee

zega14
Posts: 15
Joined: Wed Oct 21, 2020 8:05 pm
Affiliation: University of Arkansas

### Which band and K-points contribute more for lambda

Dear All

I hope everyone is doing great. I have calculated lambda ( ele-phon coupling constant) and I would like to know which bands and k-points are contributed more for lambda. How can I find this information? Thank you.

Sincerely, Zegnet

hpaudya1
Posts: 95
Joined: Tue Mar 21, 2017 7:11 pm
Affiliation:

### Re: Which band and K-points contribute more for lambda

Hi Zegnet,

I am not sure how (which approach) did you calculate lambda, however if you have used "eliashberg = .true.", you can find "prefix.lambda_aniso" file which contains 4 columns "# Enk-Ef[eV] lambda_nk #kpt #band".

Best,
Hari Paudyal

zega14
Posts: 15
Joined: Wed Oct 21, 2020 8:05 pm
Affiliation: University of Arkansas

### Re: Which band and K-points contribute more for lambda

Dear Hari

Thank you very much! I have not used eliashberg to calculate lambda. I used the normal method (elph=.true.). Do I have to use eliashberg=.true. variable to get information about the band contribution to lambda?

Sincerely, Zegnet

hpaudya1
Posts: 95
Joined: Tue Mar 21, 2017 7:11 pm
Affiliation:

### Re: Which band and K-points contribute more for lambda

Hi Zegnet,

Best,
Hari Paudyal

zega14
Posts: 15
Joined: Wed Oct 21, 2020 8:05 pm
Affiliation: University of Arkansas

### Re: Which band and K-points contribute more for lambda

Hi Hari

Here is the input

&inputepw

elph = .true.
kmaps = .false.
epbwrite = .true.

epwwrite = .true.

nbndsub = 30
! bands_skipped = 'exclude_bands = 1-10'

wannierize = .true.
num_iter = 300

proj(1) = 'l=0; l=1'
proj(2) = 'l=0; l=1'

wdata(1) = 'bands_plot = .true.'
wdata(2) = 'bands_num_points=20'
wdata(3) = 'bands_plot_format = gnuplot'
wdata(4) = 'begin kpoint_path'

wdata(10) = 'end kpoint_path'

iverbosity = 0

elecselfen = .true.
phonselfen = .true.

fermi_energy = 4.6

fsthick = 2 ! eV
temps = 0.075 ! K
degaussw = 0.2 ! eV
degaussq = 0.05 !meV

a2f = .true.

dvscf_dir = '../../../phonons/save'

nk1 = 6
nk2 = 6
nk3 = 4

nq1 = 6
nq2 = 6
nq3 = 4

nkf1 = 60
nkf2 = 60
nkf3 = 60

nqf1 = 16
nqf2 = 16
nqf3 = 16

out put

Memory usage: VmHWM = 769Mb
VmPeak = 1464Mb
===================================================================

===================================================================
Eliashberg Spectral Function in the Migdal Approximation
===================================================================

lambda : 0.3423983
lambda_tr : 0.3797309

Estimated Allen-Dynes Tc

logavg = 0.0004335 l_a2f = 0.3424619
mu = 0.10 Tc = 0.103596615530 K
mu = 0.12 Tc = 0.047641097098 K
mu = 0.14 Tc = 0.017616035963 K
mu = 0.16 Tc = 0.004706846907 K
mu = 0.18 Tc = 0.000751339318 K
mu = 0.20 Tc = 0.000049259594 K

I am now running anisotropic eliashberg calculation but when I check the out put file it stuck on:

Size of allocated memory per pool: ~= 1.5452 Gb

Size of allocated memory per pool: ~= 1.5346 Gb

Max nr of q-points = 3072

The calculation is still running. I used default iverbosity = 0. Any idea why it is not updating? Thank you.

hpaudya1
Posts: 95
Joined: Tue Mar 21, 2017 7:11 pm
Affiliation:

### Re: Which band and K-points contribute more for lambda

Hi Zegnet,

I can only guess what might have happened.

Code is calculating a2f and lambda after reading freq, egnv, ikmap, and ephmat files. This might take minutes to hours depending on your input setting and nr. of cores used.

I see you have fsthick = 2 eV, do you have any reason to take this large fermi window? I recommend using four times your maximum phonon frequency.
nkf/nqf = 60^3/16^3 (Code does not work with non commensurate fine k/q meshes). I recommend try smaller commensurate meshes eg. 30x30x20.

Best,
Hari