I have asked this question before, but I believe I did not make myself clear and point of query was lost.
I am using EPW 5.0.0 version with QE 6.3
Here are my input files
--
&inputepw
prefix = 'cdte'
amass(1) = -----,
amass(2) = ------,
outdir = './'
iverbosity = 3
restart = .true.
restart_freq = 5000
elph = .true.
epbwrite = .false.
epbread = .true.
epwwrite = .false.
epwread = .true.
nbndsub = 8
nbndskip = 48
kmaps = .true.
wannierize = .false.
num_iter = 8000
iprint = 2
dis_win_max = 4.5
dis_win_min = -11.0
dis_froz_min = -5.8
dis_froz_max= -4.50
proj(1) = 'Cd:pz'
wdata(1) = 'bands_plot = true'
wdata(2) = 'begin kpoint_path'
wdata(3) = 'G 0.0 0.0 0.0 X 0.5 0.0 0.0'
wdata(4) = 'X 0.5 0.0 0.0 A 0.5 0.5 0.0'
wdata(5) = 'A 0.5 0.5 0.0 Y 0.0 0.5 0.0'
wdata(6) = 'Y 0.0 0.5 0.0 G 0.0 0.0 0.0'
wdata(7) = 'G 0.0 0.0 0.0 A 0.5 0.5 0.0'
wdata(8) = 'end kpoint_path'
wdata(9) = 'num_print_cycles = 50'
wdata(10)= 'dis_conv_tol = 1.0E-10' !!!!!!!!!!!!!!!!!!!!!!!!!!!!
wdata(11)= 'dis_mix_ratio = 0.6'
wdata(12)= 'dis_num_iter = 8000'
wdata(13) = 'conv_window = 2'
wdata(15) = 'conv_tol = 1.0E-10'!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
wdata(16) = 'exclude_bands = 1-48'
wdata(17) = 'guiding_centres = true'
wdata(18) = 'num_guide_cycles = 1'
wdata(19) = 'trial_step = 10 '
elecselfen = .true.
phonselfen = .false.
a2f = .false.
system_2D = .true.
fsthick = 5.0
efermi_read = .true.
fermi_energy = -5.0946
eptemp = 00 ! K (same as PRB 76, 165108)
degaussw = 0.025 ! eV
dvscf_dir = '../../phonon/save'
filukk = './cdte.ukk'
rand_q = .true.
rand_nq = 50000
rand_k = .true.
rand_nk = 50000
nk1 = 9
nk2 = 12
nk3 = 1
nq1 = 9
nq2 = 12
nq3 = 1
/
35 cartesian
0.0000000 0.0000000 0.0000000 0.0185185
0.0000000 0.1241107 0.0000000 0.0370370
------
and output looks like
ik = 1 coord.: 0.7643841 0.7986299 0.0000000
-------------------------------------------------------------------
E( 1 )= -2.1456 eV Re[Sigma]= -105.959970 meV Im[Sigma]= 1531.068704 meV Z= -0.216803 lam= -5.612474
E( 2 )= -2.1144 eV Re[Sigma]= 85.033292 meV Im[Sigma]= 1362.941486 meV Z= -0.556214 lam= -2.797870
E( 3 )= -1.9462 eV Re[Sigma]= 101.942915 meV Im[Sigma]= 1115.004743 meV Z= -1.192632 lam= -1.838481
E( 4 )= -1.8811 eV Re[Sigma]= 30.205088 meV Im[Sigma]= 1349.102199 meV Z= -1.711502 lam= -1.584282
E( 5 )= -1.1866 eV Re[Sigma]= 465.610257 meV Im[Sigma]= 1303.605972 meV Z= 1.496201 lam= -0.331640
E( 6 )= -0.8064 eV Re[Sigma]= 76.803729 meV Im[Sigma]= 1377.742857 meV Z= -0.590501 lam= -2.693476
E( 7 )= -0.4996 eV Re[Sigma]= 993.438657 meV Im[Sigma]= 2293.397168 meV Z= -0.110621 lam= -10.039843
E( 8 )= -0.2284 eV Re[Sigma]= 625.702646 meV Im[Sigma]= 2480.589282 meV Z= -0.797771 lam= -2.253492
-------------------------------------------------------------------
ik = 2 coord.: 0.7928197 0.2263995 0.0000000
-------------------------------------------------------------------
E( 1 )= -2.3547 eV Re[Sigma]= 106.053337 meV Im[Sigma]= 1803.898127 meV Z= -0.145076 lam= -7.892919
E( 2 )= -2.2934 eV Re[Sigma]= 76.745732 meV Im[Sigma]= 1166.982974 meV Z= -1.631489 lam= -1.612937
E( 3 )= -2.1547 eV Re[Sigma]= 255.235261 meV Im[Sigma]= 1474.367241 meV Z= -0.221136 lam= -5.522105
E( 4 )= -2.0407 eV Re[Sigma]= 122.220751 meV Im[Sigma]= 1424.201067 meV Z= -0.528531 lam= -2.892038
E( 5 )= -1.3591 eV Re[Sigma]= 502.388187 meV Im[Sigma]= 1969.069922 meV Z= -0.189611 lam= -6.273957
E( 6 )= -0.9261 eV Re[Sigma]= 155.964231 meV Im[Sigma]= 1485.002753 meV Z= -0.282598 lam= -4.538601
E( 7 )= -0.5822 eV Re[Sigma]= 572.682013 meV Im[Sigma]= 2426.862067 meV Z= -0.142072 lam= -8.038685
E( 8 )= -0.2320 eV Re[Sigma]= 767.578750 meV Im[Sigma]= 2747.968693 meV Z= -0.242122 lam= -5.130144
-------------------------------------------------------------------
I don't understand the reason of huge Sigma. I have cross checked electronic and phonon band structure from QE and EPW, both looks fine.
I read there were problems in iverbosity = 3 with restart option in past. I checked this as well, this possibility can be ruled out.
My colleague suggested me to increase coarse k-mesh in the nscf step. It worked in one case (not in above case). I would agree with this in case of semi-metal or metal but in case of semiconductor we don't very dense coarse mesh. In literature, 8x8x8 or 12x12x12 has been used for bulk cases. So I believe 9x12 is good enough since it can reproduce the plane wave bandstructure. And increasing coarse grid will make calculations even longer.
Another suggestion from the forum (http://epwforum.uk/viewtopic.php?f=3&t=739) is to switch from random k/q to regular k/q mesh. I am checking this possibility, it will take few days to finish the calculations. This happens irrespective of "assume_isolated = '2D'" flag, in pw calculations. Since there was no conclusive answer in that discussion. I thought of asking the question on the forum again.
Since the electron and phonon bandstructure are well reproduced, so as the electron eigenfunction. I would expect at least in the frozen window the scattering rates are not that high. What could be the possible reason for the error?
Looking forward for the replies.
Best wishes,