A queer problem about MBONDDEF parameter in reax lib

Hi,everyone! When I use reax potential to simulate a system which just includes a triclinic four layer Ni(111) slab (this is a test for the problem),
, I find a queer problem. I set the simulation box to be triclinic and the boundary is set periodic in x and y direction,
then the err “STOP Too many bonds on atom. Increase MBONDDEF” is threw.While when I set the simulation box to be orthogonal
and keep the boundary condition,the err don’t be threw.Why is it? And which one .cpp file is the code to calculate the number of bonds
connected to one atom?? Aidan have told a LAMMPS user in previous letter the reason of the err is the unphysical configuration. But I
think the reason for my case is not what Aidan said. Is it ture???

Thank you for your time and your replying


Have you tried to plot your atomic configuration? It looks strange that some atom has number of bonds more than 20 (MBONDDEF defined in lib/reax/reax_refs.h), as if it had too much neighboring atoms (an unphysical configuration then).

As long as you use reax, it is probably not cpp, but Fortran code that determine bonds connectivity, I suppose reax_connect.F.


16.10.12, 13:03, "Z.Y.Qiu" <[email protected]>":