regarding size missmatch while doing epw calculations

Dear all,
I am running test calculations for phonon line widths for 2D materials (similar to MgB2). For this, I started with less kpoints (8X8X1) and qpoints (4X4X1) for scf and phonon calculations respectively and successfully finished with positive frequencies. On the other hand, while running epw calculations I struck with the below error

[quote:364f3yb9] Error in routine calbec (3):
size mismatch[/quote:364f3yb9]

Sincearly
S. Appalakondaiah,
Research Schloar
SAINT, SKKU

Here with I am attaching my input files for the referance
[u:364f3yb9]scf.in[/u:364f3yb9]
[code:364f3yb9]&control
calculation=‘scf’,
prefix=‘test’,
pseudo_dir = ‘/home/saint/Softwares/pslibrary.1.0.0/pbe/PSEUDOPOTENTIALS’,
outdir=‘./’,
tprnfor = .true.,
verbosity =‘high’
tstress = .true.,
etot_conv_thr = 1.0d-8
forc_conv_thr = 1.0d-6
/
&system
ibrav= 0, A=3.4446,nat= 3, ntyp= 2,
ecutwfc =60, ecutrho=720,
degauss=.01469972360,
smearing=‘gauss’,
occupations=‘smearing’,
/
&electrons
diagonalization = ‘david’
mixing_mode = ‘plain’
mixing_beta = 0.3
conv_thr = 1.0d-8
/
ATOMIC_SPECIES

CELL_PARAMETERS
0.873521802 -0.504328048 0.000000000
0.000000000 1.008656096 0.000000000
0.000000000 0.000000000 6.976744186
K_POINTS AUTOMATIC
8 8 1 0 0 0 [/code:364f3yb9]

[u:364f3yb9]nscf.in[/u:364f3yb9]
&control
[code:364f3yb9] calculation=‘nscf’,
prefix=‘test’,
pseudo_dir = ‘/home/saint/Softwares/pslibrary.1.0.0/pbe/PSEUDOPOTENTIALS’,
outdir=‘./’,
tprnfor = .true.,
verbosity =‘high’
tstress = .true.,
etot_conv_thr = 1.0d-8
forc_conv_thr = 1.0d-6
/
&system
ibrav= 0, A=3.4446,nat= 3, ntyp= 2,
ecutwfc =60, ecutrho=720,
degauss=.01469972360,
smearing=‘gauss’,
occupations=‘smearing’,
/
&electrons
diagonalization = ‘david’
mixing_mode = ‘plain’
mixing_beta = 0.3
conv_thr = 1.0d-8
/
ATOMIC_SPECIES

CELL_PARAMETERS
0.873521802 -0.504328048 0.000000000
0.000000000 1.008656096 0.000000000
0.000000000 0.000000000 6.976744186
K_POINTS crystal_b
4
0.0 0.0 0.0 50
0.5 0.0 0.0 20
0.3333 0.3333 0.0 20
0.0 0.0 0.0 50[/code:364f3yb9]

[u:364f3yb9]epw.in[/u:364f3yb9]

[code:364f3yb9]&inputepw
prefix = ‘NbSe2’,
amass(1) = 92.91,
amass(2) = 78.96,
outdir = ‘./’

ep_coupling = .true.
elph = .true.
kmaps = .false.
epbwrite = .true.
epbread = .false.

epwwrite = .true.
epwread = .false.

etf_mem = .true.

nbndsub = 1,
nbndskip = 0

wannierize = .true.
num_iter = 500
dis_froz_max= 2
proj(1) = ‘random’

iverbosity = 2

elinterp = .true.
phinterp = .true.

tshuffle2 = .true.
tphases = .false.

parallel_k = .true.
parallel_q = .false.

eps_acustic = 2.0 ! Lowest boundary for the
ephwrite = .true. ! Writes .ephmat files used when wliasberg = .true.

fsthick = 0.4 ! eV
eptemp = 5 ! K
degaussw = 0.10 ! eV
nsmear = 1
delta_smear = 0.04 ! eV

degaussq = 0.5 ! meV
nqstep = 500

eliashberg = .true.

laniso = .true.
limag = .true.
lpade = .true.

conv_thr_iaxis = 1.0d-2

wscut = 1.0 ! eV Upper limit over frequency integration/summation in the Elisashberg eq

nstemp = 1
tempsmin = 15.00
tempsmax = 60.00

nsiter = 500

muc = 0.16

dvscf_dir = ‘../phonons/save’

nk1 = 4
nk2 = 4
nk3 = 1

nq1 = 4
nq2 = 4
nq3 = 1

mp_mesh_k = .true.
nkf1 = 16
nkf2 = 16
nkf3 = 1

nqf1 = 16
nqf2 = 16
nqf3 = 1
/
4 cartesian
0.000000000 0.000000000 0.000000000 1.000000000
0.143098890 0.247854547 0.000000000 1.000000000
-0.286197779 -0.495709094 0.000000000 1.000000000
0.429296669 0.247854547 0.000000000 1.000000000[/code:364f3yb9]

Dear S. Appalakondaiah,

This error should be raised when the max number of plane wave is larger than the current one (on that node) or when some Kleiman-Bylander momentum are too large.

Therefore this might be related to the use of PSP. Which kind of psp did you use?

Did you use the same ecut for the phonon calculation as for the scf,nscf and epw run?

Best,

Samuel

Dear Prof Samuel,

Yes. I am using same ecut for scf, nscf in epw and phonon calculations. Also, scalar relativistic RRKJ ultrasoft pseudopotentails for all the calculations.

Sincearly
S. Appalakondaiah,
Research Schloar
SAINT, SKKU

Dear S. Appalakondaiah,

Are they PBE xc functionals?

If so, could you try a small calculation with LDA just to see if you get the same error message?

Best,

Samuel

Hi,

I also noticed that you run the nscf using only 4 k points - instead, you need a uniform 4x4x1 grid in the input (this is a requirement for the wannierization afterwards).

Best
Carla

Yep !

Well spotted Carla.

You need a uniform positively defined [0:1] homogeneous k-grid for nscf.in

Dear Prof. Samuel and Prof. Carla

Thank you for the reply. Now I changed according to your suggestions and fixed the PWs for both scf.in and nscf.in. But the issue changed now into [quote:13v6ftsx]coarse k-mesh needs to be strictly positive in 1st BZ[/quote:13v6ftsx]. Below I am adding my inputs. Suggest me for the next (I hope this issue is not related to my phonon calculations !!!)
[u:13v6ftsx]SCF.in[/u:13v6ftsx]
[code:13v6ftsx]K_POINTS AUTOMATIC
8 8 2 0 0 0 [/code:13v6ftsx]
[u:13v6ftsx]NSCF.in[/u:13v6ftsx]
[code:13v6ftsx]K_POINTS AUTOMATIC
8 8 2 0 0 0 [/code:13v6ftsx]
[u:13v6ftsx]epw.in[/u:13v6ftsx]
[code:13v6ftsx] nk1 = 8
nk2 = 8
nk3 = 2

nq1 = 8
nq2 = 8
nq3 = 2

mp_mesh_k = .true.
nkf1 = 16
nkf2 = 16
nkf3 = 2

nqf1 = 16
nqf2 = 16
nqf3 = 2
[/code:13v6ftsx]

Hello,

You have to manually input the point in the nscf.in and they have to be between 0 and 1 for each value.

Indeed if you put
[code:2kcmmy6w]K_POINTS AUTOMATIC
8 8 2 0 0 0 [/code:2kcmmy6w]

Then you do not have positive values (you can look at nscf.out).

You can look the input nscf.in files in the EPW example to see how this should be done.

Best,

Samuel

Dear Prof..

I too agree with you.. In examples, Kpoints are in manual. But How it is ?? I tried with kpoints.x tool in QE, generation using xcrysden, kpoints automatic selection and other script which is in below. In all ways, i am not able to solve my issue. can any one suggest me for generation of kpoints in manual (specifically points should between 0 to 1)

[code:1ttirlm6]#!/bin/sh
cat > script.m << EOF
nk1 = 16;
nk2 = 16;
nk3 = 4;
fid = fopen(‘klist.dat’,‘w’);
for i1=0:nk1-1;
for i2=0:nk2-1;
for i3=0:nk3-1;
kvec = [i1/nk1 i2/nk2 i3/nk3 1/(nk1nk2nk3)];
fprintf(fid,‘%12.6f %12.6f %12.6f %12.6f\n’,kvec’);
end;
end;
end;
fclose(fid);
EOF
octave < script.m
rm script.m
[/code:1ttirlm6]

Hello,

You can use the following python script (for your particular problem):

[code:3t5wqwm0]import numpy as np

wgt = 1.0/(882)

for i in np.arange(0,1,1.0/8):
for j in np.arange(0,1,1.0/8):
for k in np.arange(0,1,1.0/2):
print i, j, k, wgt[/code:3t5wqwm0]

Dear Prof. Samuel,
Now solved the issue with changing the PSPs to norm conserving and XC to LDA. But the issues are with now Wannier90 interpolation and got the below error

[quote:3pjilyp8] Unable to satisfy B1 with any of the first 12 shells
Your cell might be very long, or you may have an irregular MP grid
Try increasing the parameter search_shells in the win file (default=12)
[/quote:3pjilyp8]

I also checked abt this error and recompiled with newer version. I may be missed something in input for epw.in. Any suggestions regarding the same ??? Below I am attaching my input
[code:3pjilyp8]

&inputepw
prefix = ‘NbSe2’,
amass(1) = 92.91,
amass(2) = 78.96,
outdir = ‘./’

ep_coupling = .true.
elph = .true.
kmaps = .false.
epbwrite = .true.
epbread = .false.

epwwrite = .true.
epwread = .false.

etf_mem = .true.
! system_2d = .true.

nbndsub = 15,
nbndskip = 0

wannierize = .true.
num_iter = 500
dis_win_min = -10
dis_win_max = 10
dis_froz_min = -10
dis_froz_max = 10.000

proj(1) = ‘Nb:l=0;l=1;l=2’
proj(2) = ‘Se:l=1’

iverbosity = 2

elinterp = .true.
phinterp = .true.

tshuffle2 = .true.
tphases = .false.

parallel_k = .true.
parallel_q = .false.

eps_acustic = 2.0 ! Lowest boundary for the
ephwrite = .true. ! Writes .ephmat files used when wliasberg = .true.

fsthick = 0.4 ! eV
eptemp = 250 ! K
degaussw = 0.10 ! eV
nsmear = 1
delta_smear = 0.01 ! eV

degaussq = 0.05 ! meV
nqstep = 500

eliashberg = .true.

laniso = .true.
limag = .true.
lpade = .true.

conv_thr_iaxis = 1.0d-4

wscut = 1.0 ! eV Upper limit over frequency integration/summation in the Elisashberg eq

nstemp = 1
tempsmin = 15.00
tempsmax = 60.00

nsiter = 500

muc = 0.16

broyden_beta = 0.3

dvscf_dir = ‘../phonons/save’

nk1 = 8
nk2 = 8
nk3 = 1

nq1 = 8
nq2 = 8
nq3 = 1

mp_mesh_k = .true.
nkf1 = 8
nkf2 = 8
nkf3 = 1

nqf1 = 8
nqf2 = 8
nqf3 = 1
/
4 cartesian
0.000000000 0.000000000 0.000000000 1.000000000
0.146257668 0.253325712 0.000000000 1.000000000
-0.292515336 -0.506651424 0.000000000 1.000000000
0.438773004 0.253325712 0.000000000 1.000000000[/code:3pjilyp8]

Dear appalakondaiah,

I have never seen this warning/crash report before from Wannier90.

It might be related with your structural information.

Maybe the cell is highly distorted. If you do calculation on a 2d materials, is the vacuum large enough etc ??

Best,

Samuel

Dear appalakondaiah,

I see in the epw.in you now have nk1=8, nk2=8, nk3=1, and same for nq, however you had 8x8x2 grids previously and you should have generated a 8x8x2 grid for the nscf calculation according to Samuel’s script. If that’s the case you need to change nk3 and nq3 to 2, otherwise please ignore this comment.

Best
Carla

Hi,

I have the same problem with the wannier90. I found a solution in wannier forum by adding kmesh_tol=0.00001. (http://mailman.qe-forge.org/pipermail/w … 00155.html)

How can I do it from the input of epw input? There is no such kind of parameter can be set.

With many thanks,
Loc

Hello,

You can add

wdata(1) = ‘kmesh_tol=0.00001’

to your epw input file and it should work.

Best,

Samuel

Hi Samuel,

Thank you for your kind instruction. I can add more input for wannier90.

The issue is not solved by adding kmesh_tol but increasing the search_shells= to 80 works for me.

With many thanks,
Loc
P/S: Do you see my signature in the post? I can not see and modify my signature in the forum.