[lammps-users] about tersoff/zbl potential

LAMMPS users,

Hi, I am confused about how the tersoff potential was splined to zbl potential in the code. I read the manual about it and thought that the zbl potential only takes effect for small distance and when the distance is large enough, the whole potential is almost the same with the tersoff potential. But in the code, zbl potential was only splined to the repusive part of the tersoff potential---- ters_fc, and I don’t understand where have the ters_far and the three body part have gone? And most importantly, we need to make the total potential smoothly right? Then where is the connection point? and what is the formula that zbl has to connect with? Thank you.



I don't know the answers to these Qs. I suggest reading
the tersoff/zbl doc page which has a lot of details and paper
citations. Also it has the names of several people who
contributed to the code module, so I suggest contacting
them directly. Aidan may have additional thoughts.


Thank you for your reply. Do you mean that the ZBL potential is switched on while rzblcut? If the potential is ‘F_fermi(’ pair_tersoff_zbl.cpp) while r<zblcut, then what is the formula while r>zblcut (tersoff potential is a many body potential)?And if I want to check whether my choice of zblcut and zblexpscale parameters are right, what should I do?

2011/1/17 Thompson, Aidan <[email protected]>

The documentation file pair_tersoff_zbl.html is correct and should address
your questions. The three-body terms in Tersoff are not directly modified by
the Fermi function, but the overall energies, both attractive and repulsive
are scaled by F_fermi.


Sorry, I still do not understand what should I do if I want to check the zbl entry parameters. Can you explain it with the exact formulas?


2011/1/17 Thompson, Aidan <[email protected]>

How do you verify that a potential in LAMMPS? What I like to do is read in a
geometry in to LAMMPS, compute forces and energy ('run 0') and compare with
the "gold standard" code. This "gold standard" could be another
implementation in LAMMPS, or a completely separate code. This test often
catches small differences that might have gone unnoticed otherwise.