inner cutoff for SW potential ?

Hi,

In order to simulate high-energy collisions between atoms, I would like to couple the ZBL pair style (for very close interactions) to the Stillinger-Weber potential (for larger distances). Typically, I would need the ZBL interaction for interatomic separations between 0 and 1 angstrom, above which I would impose the Stillinger-Weberpotential. Is there a way to have an inner cutoff for the Stillinger-Weber potential, in addition to the outer cutoff (= sigma*a) ?
Also, I don’t understand the role of the “tol” parameter with the Stillinger-Weber pair style. The following statement in the command page : “In the Stillinger-Weber potential, the interaction energies become negligibly small at atomic separations substantially less than the theoretical cutoff distances.” seems strange to me, because if you plot the two-body term of the Stillinger-Weber potential, you get a curve that diverges to infinity as you approach very small atomic separations. If anyone could help me with this, it would me much appreciated. Thank you in advance,

Paule

Hi,

In order to simulate high-energy collisions between atoms, I would like to
couple the ZBL pair style (for very close interactions) to the
Stillinger-Weber potential (for larger distances). Typically, I would need
the ZBL interaction for interatomic separations between 0 and 1 angstrom,
above which I would impose the Stillinger-Weberpotential. Is there a way to
have an inner cutoff for the Stillinger-Weber potential, in addition to the
outer cutoff (= sigma*a) ?

no. that would not be a very good idea.

the solution would be to create a derived class PairSWZBL and replace
the twobody() method with a custom version.

Also, I don't understand the role of the "tol" parameter with the
Stillinger-Weber pair style. The following statement in the command page :
"In the Stillinger-Weber potential, the interaction energies become
negligibly small at atomic separations substantially less than the
theoretical cutoff distances." seems strange to me, because if you plot the
two-body term of the Stillinger-Weber potential, you get a curve that
diverges to infinity as you approach very small atomic separations. If
anyone could help me with this, it would me much appreciated. Thank you in

you are misinterpreting this statement. it says that the potential
decays fast and that one could use a shorter cutoff for pairwise
interactions at minimal loss of accuracy, but significant speed gain.
however, since the cutoff is part of the parameterization, it cannot
be changed. tol determines which cutoff should be used for the
neighborlist and non-bonded interactions. this is quite easily
understood from reading the comments in the source code. cudos to
aidan. :wink:

I think the description of tol is pretty clear, if you read the whole
thing. tol allows LAMMPS to use a virtual*outer* cutoff that is
smaller than the theoretical cutoff defined by sigma*a. This is an
approximation that becomes more and more accurate as the virtual
cutoff approaches the theoretical.

Regarding ZBL, you can add it to the standard SW potential using pair
hybrid/overlay with pair_styles sw and zbl.

There is no setting in LAMMPS that turns off a potential for r <
r_inner, but it certainly an interesting idea. We might eventually
support this using an extension of hybrid/overlay.