Removing short range charge effects in Pair_coul_long

Hi

I am attempting to remove forces from atoms that are very close to one another (r < rcut). I did this in pair could cut by simply setting fpair=0 under the corresponding conditions. However, when I made the same correction to pair could long the forces on the atoms did not change. This seems very odd to me because it appears in both codes that the forces are being assigned in exactly the same way (via f[i][0], etc.). What am I missing? is it something with the way ewald is performing the calculation that it is impossible to remove charges before computing the long range forces? I thought ewald was only for the long range terms and the short range ones were still computed directly or is this incorrect?

thanks much in advance
matthew

Hi

I am attempting to remove forces from atoms that are very close to one
another (r < rcut). I did this in pair could cut by simply setting fpair=0
under the corresponding conditions. However, when I made the same correction

that is a pretty "brutal" change to the forces and could lead to a ton
of problems and strange behavior.
is there a reason, you cannot use soft-core potentials instead? like
these: LAMMPS Molecular Dynamics Simulator

to pair could long the forces on the atoms did not change. This seems very
odd to me because it appears in both codes that the forces are being
assigned in exactly the same way (via f[i][0], etc.). What am I missing? is
it something with the way ewald is performing the calculation that it is
impossible to remove charges before computing the long range forces? I
thought ewald was only for the long range terms and the short range ones
were still computed directly or is this incorrect?

please note, that in the coul/long pair styles, the coulomb force
computation is implemented twice: once for use with the analytical
expression (slower) and once for used with tabulated coulomb (faster
at the expense of precision). you probably modified the wrong branch.

axel.

I think this coul/soft is exactly what I ultimately want. the one thing I am not understanding from the manual is how is lambda determined?

The only thing I can see is when r=rcut: Esoft =Eregular.

Does the program then solve for lambda given the values of n and alpha_c?

Is this correct or am I completely misunderstanding the man page?

thanks again
matthew

I think this coul/soft is exactly what I ultimately want. the one thing I am
not understanding from the manual is how is lambda determined?

lambda is an input parameter given with the pair_coeff statement. it
is used, e.g. in FEP calculations to gradually turn a pairwise
interaction on or off.
from the documentation:

The coefficient lambda is an activation parameter. When lambda = 1 the
pair potential is identical to a Lennard-Jones term or a Coulomb term
or a combination of both. When lambda = 0 the interactions are
deactivated. The transition between these two extrema is smoothed by a
soft repulsive core in order to avoid singularities in potential
energy and forces when sites are created or anihilated and can overlap
(Beutler).

axel.