UT theoretical chemistry code forum

Thanks Johannes, this looks excellent.

Have you tried to play around with the mesh density?

Thanks for pointing this out. I guess it was a pretty obvious, and embarrassing, error.

That's odd, I just ran nebresults.pl on the files you sent me and everything worked fine. Are you using the latest version of the scripts? I have now updated the nebconverge.pl so that *Forces* now reads *FORCES*. Couple of points: 1) You should have the elements in the first line of the POSCAR, tha...

Can you send me (moller.plesset@gmail.com) all your files from the NEB run?

Ok, I think I see what the problem is. The nebconverge.pl script is looking for "Forces: m", but VASP is writing out "FORCES: m". So if you change this in the script then it should work.

Report back if it does so the cvs version can be updated.

Well, it seems like that for some reason your job didn't start up properly. Did the other VASP files get generated, e.g. XDATCAR, WAVECAR, CHGCAR etc.? If not, then this problem probably has nothing to do with vtstcode but with how VASP is compiled on your system.

1) I would think the 0.1 should be an absolute upper limit on EDDIFG, but I agree that 0.001 can be very optimistic. Usually I start with 0.05, then starting from the previously converged run, I set it to 0.01, then 0.005 etc. until I think the path is properly converged. 2) I am little bit confused...

This has nothing to do with VTSTTools!

I am a little bit confused as to why you think your configuration is a minimum. According to your DIMCAR, your initial configuration starts out slightly too high in energy, which is perfectly reasonable. As the calculation progresses, the energy is lowered as the system approaches the saddle point. ...

It looks like a saddle point to me, i.e.:

1) Force is low (and getting smaller)
2) Torque is vanishing.
3) Curvature is negative.

So, smells like a saddle point!

If the path is curved then 5 A is probably too large of a distance between the images. Remember that the neighboring images are used in a 'finite' difference way to estimate the local tangent. Personally, I prefer never to have the distance exceed 1 A, and try very hard to have it close to 0.5 A. So...

Using for instance the *dist.pl* script, what is the distance between the images? Also, have you tried to generate a .xyz movie (nebmovie.pl) from the the current CONTCARs? Is there something funny going in with the path?

I would perhaps decrease EDIFF to 1.0E-4.

What is the output from *nebef.pl*?

Did your old executable have all the necessary libraries build statically into it? Try to run "ldd <old_executable>" and "ldd <new_executable>" and see if there is any difference.