Transition State Calculations

Message

anagaw · #1 Post by **anagaw** » Mon Jul 04, 2005 9:27 pm

Hi everyone,
I am not able to use the VASP code for Elastic Band Method calculations to
find the Transition states , Barrier. I would appreciate if any one tell me what the problems are.
I used the following just for two POSCARs which are written in two
subdirectories named 01 and 02

SYSTEM = Elastic Band Method
LSCALAPACK= .FALSE.
IBRION=3
POTIM=0.0
ISYM=0
ICHAIN=2
IMAGES=2

But it did not work
It instead says

hus no job control in this shell.
master mpi start node is dirac
nodes are:
dirac
langmuir
nodecount is 2
vasp.4.6.20 12Dec03 complex
ERROR: the following files does not exist POSCAR

Thank you.

<span class='smallblacktext'>[ Edited ]</span>

#2 Post by **admin** » Tue Jul 05, 2005 2:31 pm

1) the parameters to be set in INCAR to do an NEB cakculation are
SPRING = -5
IMAGES = # of intermediate geometries
ICHAIN is not a valid input parameter for vasp.4.6

2) by default, you need directories named 00 -- 0X:
00 and 0X contain the POSCARS of the initial and final state geometries, all other directories (01 -- 0(X-1) ) contain the POSCARS of the geometries interpolated between initial and final state.
the error message says that the POSCARS were not found where they had
been expected, probably because the NEB method was not started (SPRING
is missing in INCAR)

graeme · #3 Post by **graeme** » Tue Jul 05, 2005 6:45 pm

The ICHAIN variable is used to control transition state methods which have not yet been incorporated into the main vasp distribution. The ICHAIN=2 is a setting to run the dimer saddle point finding method, for which only two images in the 01 and 02 directories are requried. The additional code required to run this method is available at http://theory.cm.utexas.edu/vtsttools/ . If you rebuild vasp with the additional code, your INCAR settings are appropriate for running the dimer method.

Vasp Masters: Would you considering incorporating these new methods into the main vasp distribution? The dimer method, and the climbing image NEB combined with an improved tangent are now mature methods which are used by folks in the vasp community. Also, it would be nice to have a version of the conjugate gradient optimizer which does not require the energy, to speed up convergence of the NEB. The combination of these changes will enable vasp to find transition states more efficiently.

anagaw · #4 Post by **anagaw** » Tue Jul 05, 2005 9:44 pm

I have still problem in running the job. I have created four subdirectories
from 00 ..03 in the main directory in which vasp is started. Each
subdirectories has one POSCAR file. But when I execute the program with the following
INCAR files

SYSTEM = Elastic Band Method
SPRING= -5
IMAGES= 2
IBRION=3
POTIM=0.0

It syays vasp.4.6.20 12Dec03 complex
ERROR: the following files does not exist POSCAR

<span class='smallblacktext'>[ Edited Tue Jul 05 2005, 11:46PM ]</span>

#5 Post by **admin** » Fri Jul 08, 2005 9:23 am

apart from the choice of POTIM (it is not reasonable to set it to 0) your input
should work.
please check the following
1) did you really use the parallel executable of VASP ( this is required for the
implemented version of the NEB method)
2) did your jobfile generate all subdirectories and copy the POSCARs
without errors ?

wsyjsrrp · #6 Post by **wsyjsrrp** » Tue Nov 14, 2006 7:48 am

[quote author= Elastic Band Method
LSCALAPACK= .FALSE.
IBRION=3
POTIM=0.0
ISYM=0
ICHAIN=2
IMAGES=2

But it did not work
It instead says

hus no job control in this shell.
master mpi start node is dirac
nodes are:
dirac
langmuir
nodecount is 2
vasp.4.6.20 12Dec03 complex
ERROR: the following files does not exist POSCAR

Thank you.

[/quote]

wsyjsrrp · #7 Post by **wsyjsrrp** » Tue Nov 14, 2006 7:53 am

I think that your POTIM=0.0 is error, POTIM should not 0.0,if that, structure will not change.so you can't find transition states

#8 Post by **admin** » Tue Nov 14, 2006 8:43 am

you forgot to set SPRING=-5
[ Edited Tue Nov 14 2006, 09:43AM ]

Haley · #9 Post by **Haley** » Thu Jul 03, 2008 7:53 am

The implementation of the nudged elastic band(NEB) method in VASP is based on parallel machine, it does not work on serial machines.

safak · #10 Post by **safak** » Wed Feb 25, 2009 2:00 pm

Hi
I have an additional question. I have initial and final POSCARS but i have not been able to figure out how to use interpolatePOS. I guess i have to use it because my system is strictly two dimensional.
thanks

#11 Post by **admin** » Tue Mar 17, 2009 12:07 pm

interpolatePOS should work for any system, not just a 2D-one, all it does is to calculate the positions by interpolation between initial and final state. Please note that
1) the unit cells (Bravais matrices) for initial and final states have to be the same
2) take care whether all atoms are 'on the same side' of the cell (ie, none of the atoms re-enters the cell from the opposite side, due to the periodic boundary conditions)