Recipe of dimer method

[erogen]

Hello, Dr Henkelman.

I would like to ask a question for dimer method.
Even though I read all of the threads in the board, I can't grab the details in a right way.

1) Starting from NEBM
If I know the two optimized states after NEB, I can interpolate the tentative intermediate stages.
interpolate.pl POSCAR_initial POSCAR_saddle_point 0.5

This would be a best candidate if I know two endpoints.
Why do we need to follow this scheme if I know all of initial, final, and transition states?

2) Random guess of TS from IS
If I know a local minimum configuration which is well optimized,

a) I need to guess the good transition state by interpolation or something like that
b) I should make a good guess with an intuition :(


I decribe a dimer method which I understand in the board.
Do I have myself undertood in a right way?

3) The detail (input file) to run a dimer method in VASP

a) I should have two folders. i.e. "01" & "02"
b) In two folders, I should put "POSCAR_initial" in folder "01" and "POSCAR_saddle_point or equivalent" in folder "02" w/o any other file.
c) In home directory (just below "01" and "02"), I should have "INCAR (w/ dimer input), KPOINTS, POTCAR, POSCAR_initial".
d) Run!

Could you point out what is wrong in order to set up input files in VASP?

[andri]

For a dimer run, you need two different folders, 01 and 02. In each one you have one of the dimer image, called POSCAR. If you place the same POSCAR in directories 01 and 02, VASP will generate a dimer configuration with a random orientation. You can also try the diminit.pl script which is probably better. You should not need a guess for the transition state if you start from a minimum. All you really need is to randomly displace the minimum configuration as an initial guess for the dimer.

Best thing though is to start from a converged NEB run and use the neb2dim.pl to generate the dimer configuration. It should all happen automatically.

In the same directory as you have 01 02, you place the appropriate KPOINTS, INCAR and POTCAR files.

[staar]

I would like to ask you, do we need to place POSCAR file in the main directory apart from folders 01 and 02?

Secondly, do we need to use the same POSCAR file in both the folders 01 and 02?

I am using the regular VASP code modified with your tools. And also I have
almost a transition state and not a transition state. So I guess I can start with this POSCAR.

I request you to provide a clear description of how to run dimer method in this board and it is very useful for all novice users. I mean the exact steps, like creating folders, input files and where to place them etc.,

Thank you very much for the support you are providing.

Amjad

[graeme]

The setup for the dimer method has recently changed. It is a very good suggestion that we could clearify how it works and give steps for setting up a calculation, as well as an example.

I've updated the page http://theory.cm.utexas.edu/vtsttools/dimer/
with these suggestions. Please let me know if this is not clear.

Thanks for the feedback.