VC-NEB incorrectly transforms first/last images ...

[Vladimir Timoshevskii]

Dear USPEX team,

I am trying to do VC-NEB calculations for a MOF structure, and I am facing a strange behavior of the code. My first and last images are perfectly relaxed with QE, but USPEX shows quite large forces for them. A closer examination showed that USPEX changes initial and final structures while generating a series of images. I may guess that the code tries to find a symmetry group and then rotate the coordinate system to make computations more efficient. Indeed, a visualization shows (at first glance) the same structure, but rotated. However, a closer look showed that some groups of atoms are just incorrectly placed in this transformed first image (either too close or too far from each other). As a result, if I take this new image and try to relax it again with QE, the MOF system just breaks into pieces. Could it be a bug in the code, or I am misunderstanding something here? Also, is there a way to turn-off this first/last image transformations and just use my relaxed structures "As is"? Any advices are highly appreciated! I am using version 10.5 + QE as DFT engine. Thanks a lot!

Best,

Vladimir.

[Vladimir Timoshevskii]

Hi colleagues,

Any ideas how to solve this problem? 

Thanks!

Vladimir.