pair_style reax/c implementation

Hi!

While using reaxff force field for Li and Si (developed by Van Duin, et al.), by using pair_style reax/c I am getting different results for potential energy against distance, as well as for force against distance, than the results I obtain using pair_style reax, specially when the distance between the atoms is low. I have attached the results in the accompanying slides. When using pair_style reax/c I have also used the “fix qeq/reax” command for charge equilibrations.

I have included the different results for potential energy v/s distance and force v/s distance curves. Also in order to isolate the causes of these differences, I compared the different components of the energy for Li-Li interaction, and I see a huge difference in the Van der Waals contribution for reax and reax/c implementation. I have included the curves for the same in the attached .jpg file.

I am looking forward to any suggestion regarding what might be a reason for this, and what I kind of modification should be made in the reax/c code to address this?

Thanks,
Swastik

vdwcontr.jpg

reaxcurves.pptx (476 KB)

Hi Swastik,

I remember the discussion of some discontinuity of energy in reax/c implementation: https://github.com/lammps/lammps/issues/71 . You may try to revert the changes from the commit Axel mentions in the discussion (get rid of “numbonds”) and check if the problem is solved.

Regards,
Oleg

27.01.2017, 16:51, “Swastik Basu” <basus5@…707…>: