ABINIT Discussion Forums

Hello, my aim is to calculate the Eliashberg function for MgB2 but in the file *_ep_A2F I get nothing but NaN for its values. PHDOS comes out finite so I think the problem is the GKK elements. I calculated this on a 2x2x2 qgrid and got something finite but no where near accurate. This run is for a 4x4x4 qgrid. My acell values are experimental and rprim was chosen to be compatible with ANADDB's brav=4. I would really appreciate it if a more experienced user could look over my inputs and provide comments... I'm new to ABINIT and am working half blind here

ABINIT input...
ndtset 23

tolwfr1 1.0d-14
nline1 8
rfphon1 0
nqpt1 0
prtwf1 1
getwfk1 0

qpt2
...
qpt23

rfatpol 1 3
rfdir 1 1 1
rfphon 1
prtwf 0
tolvrs 1.0e-5
getwfk 1
nqpt 1
prepgkk 1
prtgkk 1

ngkpt 16 16 16
kptopt 3
nshiftk 1
shiftk 0.0 0.0 0.5
ecut 60.0
acell 2*5.831694401
6.659394384
rprim
1.0 0.0 0.0
-0.5 0.86602540378443864676 0.0
0.0 0.0 1.0
nband 12
occopt 6
tsmear 0.025
natom 3
typat 1 2 2
xred 0.00 0.00 0.00
0.33 0.66 0.5
0.66 0.33 0.5
nstep 800
ntypat 2
znucl 12 5

ANADDB input
elphflag 1
nqpath 2
qpath
0.0 0.0 0.0
# 1/2 1/2 0.0
# 1 1 1
1/2 1/2 1/2
# 1/2 1/2 0.0
# 1/2 3/4 1/4
# 1/2 1/2 1/2
mustar 0.136
ngqpt 4 4 4

asr 2
dipdip 0
brav 4

nqshft 1
q1shft 0.0 0.0 0.0
#
ifcflag 1
ifcana 1
ifcout 200

natifc 1
atifc 1 2 3
prtdos 1
ng2qpt 4 4 4
dieflag 0

eivec 1

nph1l 1
qph1l
0.00000000E+00 0.00000000E+00 0.00000000E+00 1
symdynmat 0

Hi!

What are the three dots for, between qpt2 and qpt23?

Igor L.

your atomic positions are not precise at all, so that's a start, but it shouldn't make things NaN. Use 1/3 2/3 and so on.

your tsmear is huge, and you may bias results or get washed out results this way. Need to be careful

for el-ph calculations you may not use a shiftk. It may work more or less (especially if your k-grid is super converged) but there can be errors (and NaNs) with the present production version of abinit

tolvrs 1.0e-5 is not very tight convergence. You need more if you want good numbers.

as Igor says - what is you qpt2 .... and why do you have so many? You can use symmetry to reduce the q-point grid, and with prepgkk there is no need to specify 1 perturbation at a time: 1 dataset for each q-point is now the rule.

still nothing conclusive, but you need to give us real information on the error and output... Read the nettiquette, please

matthieu