Elastic band method

Message

sharma · #1 Post by **sharma** » Thu Jul 28, 2011 4:37 am

Dear Everyone,
I want to calculate dissociation path for O2 molecule on Pt3Ni(111) surface. I have PARALLEL system. I made a folders from 00 to 09 with 8 intermediate points and I put the corresponding POSCAR files inside each folder. First I used 4 nodes with 4CPU each in my calculation but it stopped at 21th interaction step without any messages in stdout file. Just stopped the calculation. So I thought that my computer memory was not enough to calculate 8 parallel points so I reduced to 4 and again calculate but it again stopped at 51th interaction. I have no problem in POSCAR files. Here is my INCAR, please suggest me the way of handling the Elastic band method. also I saw the total energy was about stable energy * no . of Intermediate points. How can I get the energy of particular position? It seems like sum of all .......... I also checked for POTIM = 0.3 but same type problem.

My INCAR

NWRITE = 2
ISTART = 2
ICHARG = 2
INIWAV = 1
ISPIN = 2
MAGMOM = 12*0 4*1 2*1
LORBIT = 11
PREC = Medium
NELM = 100
EDIFF = 1E-05
ENCUT = 400.00 eV
NSW = 100
EDIFFG = 1E-03
IBRION = 3
POTIM = 0.1
ISIF = 2
ISYM = 2
ISMEAR = 1;SIGMA = 0.2
IALGO = 38
LDIAG = T
LREAL = T
VOSKOWN = 1
AMIX = 0.4
AMIX_MAG = 1.6
BMIX = 1.0
BMIX_MAG = 1.0
# Elastic band method
# NPAR = 8
IMAGES = 4
SPRING = -5
# NSIM = 1
LSCALAPACK = .FALSE.
<span class='smallblacktext'>[ Edited ]</span>

kambiz · #2 Post by **kambiz** » Sun Jul 31, 2011 7:31 am

Hi,

Here is what I usually use for my NEB calculations, you may wanna try this:

ISTART = 0
ICHARG = 2
NPAR = ...

ISPIN = 2
MAGMOM = ...

ENCUT = ...
ISMEAR = ...
SIGMA = ...
NELM = ...
EDIFF = 1E-4
ALGO = F

ICHAIN = 0
IMAGES = ...
SPRING = ...
LSCALAPACK = .FALSE.
EFIRST = ...
ELAST = ...

GGA = ...
VOSKOWN = 1
NBANDS = ...
LREAL = .TRUE.
ADDGRID = .TRUE.
PREC = Medium

NSW = ...
IBRION = 3
POTIM = 0.15
EDIFFG = ...

LMAXMIX = 4
ISYM = 0

What do you mean by "How can I get the energy of particular position"?.
If you mean energy of each step during ionic relaxation, just do this:

grep F OSZICAR

cheers

sharma · #3 Post by **sharma** » Mon Aug 01, 2011 7:08 am

Thank you Kambiz