Page 1 of 1

neb restart

Posted: Thu May 22, 2008 9:09 am
by lanjh
Hi,
I am doing the neb calculation. Now I have a question about restart of neb jobs. During my neb calculation, I first set a samll value to the NSW and found the NSW is not enough to converge the job. So I want to know whether I can restart the neb calculation by the setting 'ISTART=1'!
I still have another question. Can you tell me how to determine the proper value for NSW? In addition, what is the proper initial value for POTIM, POTIM = 0.1 or 0.05?
I am eager to your reply ! Thanks !

Re: neb restart

Posted: Thu May 22, 2008 3:52 pm
by graeme
To restart the NEB, all the CONTCARs need to be copied to the POSCAR. This is not done by setting ISTART. Instead, we provide the vfin.pl script which takes all the important files from your calculation, zips them, and moves them to a specified directory. For example, if you calculation is done in a directory nebrun, with image directories nebrun/00 .. nebrun/NN, then once the calculation is done, you can run (from the nebrun directory) 'vfin.pl run1' and then restart your job.

The value of POTIM depends upon your optimizer. A reasonable strategy is to use IBRION=3, POTIM=0.1 until the forces drop below 1 eV/Ang. If your initial band is really crazy, with atoms running into each other, you may need to drop the value of POTIM for a controlled descent from the high forces.

Once the forces are reasonable, there are several efficient optimizers for the NEB. Options are listed at http://theory.cm.utexas.edu/vtsttools/optimizers/ , and have been compared in a recent paper: http://theory.cm.utexas.edu/henkelman/p ... 134106.pdf