Nudged elastic band calculation..

[cyc99]

Hello,

I have questions about NEB(nudged band theory) calculations implemented in VASP.

1) During the NEB calculations, Is the image driven up to the saddle point?

2) If there is local minimum on the pathway between the initial and final positions of the NEB calculations, what happens? one of the image goes to the local minimum?

3) How do you check if there is local minimum on the pathway or not? Do you have any knowhow? or Should I minimize the structures with various stating points on the pathway?

Thanks.

[lunatic]

I don't think there is a reason to start a new thread so I'll just add my questions about NEB here together with cyc99's.
1) Before the NEB calculation, one has to make sure that all atoms are on the same order and that the center of mass of the SuperCell is the same for all the images and the endpoints.
I'm running a calculation with 3 images and I noticed that after an ionic step, some atoms in one of the images are moved by one period. So if they were in one side of the SC, they move to the other.
I know that this is not a problem for a regular calculation of course. But is it a problem for a NEB calculation? Is there any way to prevent this from happening? I saw this from the CONTCAR

2) I use linear interpolation between the images. I am facing a problem with my middle image. The middle image should be very close to the saddle point because it's right in the middle of the initial and final positions. However, it relaxes
to a very structure. I don't think this is correct... This is happening always with the middle image. For example if I use 3 images, then instead of a gaussian shape for the energies, I'm getting a 2 bumps shape.
I guess this is somehow related to cyc99's question about the local minimum. My concern is that the structure I'm getting in the end has nothing to do to what I started with...

For my calculations I'm using SPRING=-5. Also I'm using the built-in method and not the VTST tools.