Failing to relax a 2D structure

Total energy, geometry optimization, DFT+U, spin....

Moderator: bguster

Locked
CHILL
Posts: 25
Joined: Sun Oct 08, 2017 1:37 pm

Failing to relax a 2D structure

Post by CHILL » Mon Jan 27, 2020 2:30 pm

Dear experts and users,

I've been trying to relax a 2D Ga structure with no success. Could you please indicate what might be wrong?
The structure has been successfully calculated by me or my colleagues in other DFT packages (QE, VASP).
Thus, I started off with an almost relaxed structure, and I expected it to be fully relaxed in ABINIT within just a few steps.
Indeed, the first step indicated forces of 10^(-5) Ha/bohr, but after a couple of quickly converging relaxation steps the situation got quite bad: all the following steps are not converging well (oscillating residm/vres) and forces are going bad (I terminate the job when forces deteriorate to 10^(-1) Ha/bohr ).
I tried ionmov = 2,3, different versions of Abinit, increasing nband, ecut, ngkpt. Nothing seems to work.

Below I provide an example of my input file:

Code: Select all

optcell  9
ionmov   2
tolmxf   5.0d-5
strfact  10
ntime    500
dilatmx  1.1
ecutsm   0.5
npulayit 25
nline 6
autoparal 1

acell 1 1 1 angstrom
rprim
 +4.7440200000  +0.0000000000  +0.0000000000 
 +0.0000000000  +4.9204393662  +0.0000000000 
 +0.0000000000  +0.0000000000 +15.0000000000 


ntypat 1
znucl  31

natom 4
typat 1 1 1 1
xangst
 +0.0000000000  +1.9081898149  +5.2271844041 
 +0.0000000000  +4.3684094980  +6.4204537083 
 +2.3720100000  +0.5520298682  +5.2271844041 
 +2.3720100000  +3.0122495508  +6.4204537083

ecut   60

nband  40

kptopt 1
ngkpt  32 32 1 
nshiftk 1
shiftk 0.0  0.0  0.0

nstep 500 
toldfe 1.0d-8
iscf 7
occopt 3
tsmear 0.01
#iprcel 45
Thank you very much for the attention.
Best regards, Mikhail

MasoudMS
Posts: 5
Joined: Sun Jun 02, 2019 3:51 pm

Re: Failing to relax a 2D structure

Post by MasoudMS » Wed Jan 29, 2020 6:05 pm

Hi Mikhail
check your shiftk and use optforces and nbdbuf keyword keywords. And check your pseudo potentials files.
Best regards
Masoud

ebousquet
Posts: 469
Joined: Tue Apr 19, 2011 11:13 am
Location: University of Liege, Belgium

Re: Failing to relax a 2D structure

Post by ebousquet » Sun Feb 09, 2020 12:00 am

Dear Mikhail,
Can we see how bad is your convergence (from grep ETOT outputfile, for example)?
Also, what about if you start to relax the internal positions only?
Thanks,
Eric

CHILL
Posts: 25
Joined: Sun Oct 08, 2017 1:37 pm

Re: Failing to relax a 2D structure

Post by CHILL » Tue Mar 17, 2020 1:53 pm

Thank you all for your help.
I translated my vectors to angstroms manually, and switched to reduced atomic coordinates.
That fixed the problem.

Locked