atomic position relaxation

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

Moderator: bguster

Locked
hamideh
Posts: 13
Joined: Sat Sep 18, 2010 3:13 pm

atomic position relaxation

Post by hamideh » Sat Sep 18, 2010 3:39 pm

hi ,
I want to relax a Graphene layer , but i dont know all the relaxation variable completely , i want to know how can i relax atomic position .
i couldn't find any relaxation example in abinit site or other sites
would you please help me & introduce some simple example about relaxation in abinit ?

spamrefuse
Posts: 35
Joined: Wed Jan 20, 2010 3:08 am

Re: atomic position relaxation

Post by spamrefuse » Sun Sep 19, 2010 8:04 am

I think you should use the proper optcell parameter, e.g. opcell = 9 leaves the
third lattice vector unchanged, while optimizing the first two.

Rob.

fariba
Posts: 15
Joined: Sat Sep 18, 2010 3:18 pm

relaxation

Post by fariba » Sun Sep 19, 2010 2:04 pm

Hi
I have the same problem in relaxation & I know that relaxation's input variables are much more than "optcell" so I was wondering if you introduce some examples to me.

tnx

spamrefuse
Posts: 35
Joined: Wed Jan 20, 2010 3:08 am

Re: atomic position relaxation

Post by spamrefuse » Sun Sep 19, 2010 4:30 pm

I wonder if following could do what you want:
optcell 9
ionmov 0
acell A1 A2 A3
rprim sqrt(3/4) -0.5 0.0
sqrt(3/4) 0.5 0.0
0 0 1.0
xred 0.0 0.0 0.0
1/3 1/3 0.0


where A1 = A2 and should be some reasonable values for the
graphene sheet. Then A3 should be large, to neglect interactions
between layers.

I'm not an expert; I simply guess that this might work for you.
It's a simple system (2 atoms per unit cell), so easy to check it out.

I think that Abinit will recognize the symmetry and thus keep the
values of A1 and A2 equal througout optimization.

R.

hamideh
Posts: 13
Joined: Sat Sep 18, 2010 3:13 pm

Re: atomic position relaxation

Post by hamideh » Mon Sep 20, 2010 7:55 am

Hi Rob ,
first >>> thank's for your help ....
second >>> I think , there is some differece between atomic relaxation & optimization .
" optcell " means : optimize the CELL SHAPE and DIMENTIONS

is it possible to relax atomic position by ionmov = 0 ???
I don't think so ......
I mean : ionmov=0 => do not move ions .

ionmov is a so important variable .

regards ,

Robin
Posts: 61
Joined: Thu Jan 14, 2010 3:18 am

Re: atomic position relaxation

Post by Robin » Mon Sep 20, 2010 9:45 am

Dear hamideh,

The following is what I employed in the geometrical optimization of Graphene.

Sincerely,
Guangfu Luo

#-----------------------------------------------------------
acell 2.46 2.46 10 Angstrom
angdeg 90 90 120
ntypat 1
znucl 6
natom 2
typat 2*1
xred 1.3333333333333335 1.6666666666666665 0.5000000000000000
-1.3333333333333335 -1.6666666666666663 0.5000000000000000

chksymbreak 0

#-----------------------------------------------------------
ecut 800 eV
ixc 3
iscf 7
nstep 1000
iprcel 45
nband 8 ####### 4 No Conducting Bands
occopt 1
kptopt 1
ngkpt 36 36 1

#-----------------------------------------------------------
getwfk -1
optcell 9
ionmov 3
ntime 200
dilatmx 1.05
ecutsm 0.5 eV
tolmxf 1.85d-7 #1.85d-4Ha/Bohr==0.01ev/A
toldfe 1.0d-10

fariba
Posts: 15
Joined: Sat Sep 18, 2010 3:18 pm

Re: atomic position relaxation

Post by fariba » Tue Sep 21, 2010 3:03 pm

Hi
I want to relax just some atoms of the graphene layer,but I don't know how I can determine them in "Xcart" part;
I know in VASP package they are determined with "F F T" or "T T T",....
Is there any thing same in abinit?

tnx

Robin
Posts: 61
Joined: Thu Jan 14, 2010 3:18 am

Re: atomic position relaxation

Post by Robin » Wed Sep 22, 2010 9:20 am

Dear hamideh,

You can check the usage of the following two parameters:

natfix
iatfix

Sincerely,
Guangfu Luo

mverstra
Posts: 655
Joined: Wed Aug 19, 2009 12:01 pm

Re: atomic position relaxation

Post by mverstra » Mon Oct 11, 2010 12:13 pm

Setting optcell but leaving ionmov 0 will not relax anything. Set ionmov 2 (usually the best choice).

If you also want to fix atoms see other posts and documentation on natfix / iatfix input variables

Matthieu
Matthieu Verstraete
University of Liege, Belgium

Locked