NEB image ending up at end points

Message

esag · #1 Post by **esag** » Tue Nov 16, 2010 10:23 pm

I tried running a NEB computation with only one image between the two end points. The image was originally the interpolation of the two end points. The image converged to one of the two end points. I used IBRION=2 during the computation.

Why does the image end up at an end point? I thought the NEB image would be restricted to having equal distance to the two end points using some metric.

alex · #2 Post by **alex** » Wed Nov 17, 2010 8:07 am

Hi esag, the full name is Nudged _Elastic_ Band ...
So if the forces are high, you'll loose.

Cheers,

alex

esag · #3 Post by **esag** » Wed Nov 17, 2010 8:37 am

Thanks for your fast reply Alex!

Do you mean that the forces parallel to the path end1-image-end2 will "overpower" the spring constant intended to span out the images? I thought the way things work with NEB is that only the springs (and not the forces parallel to this path from the gradient of the potential energy surface) were considered when determining the movement of the image parallel to the path, and likewise that only forces from the potential energy surface (and not the springs) were included when determining the movement perpendicular to the path.

alex · #4 Post by **alex** » Wed Nov 17, 2010 12:30 pm

Hm, good point and right from the theory point of view. But, to my experience this does not work in practise that way. It's more like if you are not close to the transition structure of the reaction, you'll drop in some region near a local minimum. At least my â‚¬0.01.

Cheers,

alex