Improper setting of KPOINTS led to an error.

[Jike.Wang]

Hello,

I would like to know what impact KPOINTS has on the calculation besides affecting the accuracy?
When I was calculating a certain structure, I found that if the KPOINTS setting was 441, there would be an error message "ZBRENT: fatal error in bracketing", but if it was set to 451, the calculation could proceed normally.

The relevant files for the two calculations have been uploaded.

Thanks!
Jike

[andreas.singraber]

Hello Jike,

thanks for sharing the input/output files! With those I was able to find an issue with your setup which has nothing to do with the KPOINTS setting. Instead, the initial structure in the POSCAR file seems already to be flawed. Here is a screenshot of the structure, looking along the axis perpendicular to the slab:

slab_small.png

The blue box represents the box boundaries and the atoms outside are just periodic images of the ones inside. As you can see there is an easily visible gap between neighboring rows of atoms. This leads to very large forces from the very beginning of the trajectory. If you visualize the trajectory you will find that it is pretty "wild" with atoms moving all around the box. Most likely this is not what you intended in the first place, in particular with the lower half of the slab in fixed positions.

My guess is that the box was rescaled but the Cartesian coordinates were not. So you should either use direct coordinates or also adapt the Cartesian coordinates to the box change. Please review the POSCAR file and see if you get more reasonable results.

All the best,
Andreas Singraber