NEB pathways across unit cell boundaries

bhinnema · Post by **bhinnema** » Wed Jun 14, 2006 4:30 pm

I just want to make sure: is it a problem if NEB pathways go across unit cell boundaries? I have a unit cell setup where I cannot avoid going across a boundary/along the unit cell edge. In the VASP forum it was mentioned by the head admin that this could lead to a pathological path, but I have never seen anything strange with it and I cannot see why it should be a problem either with PBCs. I would be interesting in hearing other experiences.

andri · Post by **andri** » Wed Jun 14, 2006 9:17 pm

Our NEB implementation should be airtight in regards to that aspect. I have never had any problems with it, but I guess anything is possible. But I usually try to set up the calculations in such a way that any boundary crossing is unlikely. So if you witness any funny business with our implementation then please report it asap, it would be much appreciated.

bhinnema · Post by **bhinnema** » Thu Jun 15, 2006 12:29 am

I usually try to set up my unit cells up, too, so that no boundary crossing takes place. But I was trying a number of different things and ended up with this pathway. So far, everything looks normal and I have not observed any funny things, e.g. discontinuities, when the adsorbate crosses the cell boundary. Also I have checked the TS I found by vibrational analysis. Thus, so far things seem to work fine with cell crossing but of course I will let you know, should I enounter anything strange. Thanks!

bhinnema · Post by **bhinnema** » Fri Jun 16, 2006 4:56 am

Just to check back in, I actually performed a test on one of my diffusion pathways. I took the converged pathway which crosses a unit cell boundary and changed the unit cell such that it (as much as possible) stays within that cell. No energy and force changes apart from noise (energy changes <0.001eV and forces <0.01eV/A) so it seems indeed to be airtight. I thought you would like to know.

Post by **graeme** » Sat Jun 17, 2006 4:32 am

Thanks, that's good to hear.

I have an off-topic question. As one of the few people who have used both vasp and dacapo extensively, how do they compare in efficiency? Do you notice significant differences when finding transition states?

bhinnema · Post by **bhinnema** » Sun Jun 18, 2006 3:37 pm

I will email you about this.