Recently, I used both of the morse potential and the lj potential and find
that cutoff has a great influence on my results when I set the cutoff of
the morse and lj potential as 7 and 8.5 respectively as follows:
pair_style hybrid tersoff morse 7 eam/alloy lj/cut 8.5
pair_coeff * * eam/alloy Cu.eam.alloy NULL Cu NULL
pair_coeff * * tersoff C.tersoff C NULL NULL
pair_coeff 1 2 lj/cut 0.019996 3.225
pair_coeff 1 3 lj/cut 0.002967 3.407
pair_coeff 2 3 morse 0.087 1.7 2.2
pair_coeff 3 3 none
it will form the twinning.
what does this set of parameters represent?
what does this have to do with twinning?
I also read the manual that the cutoff of the morse is optional, if not
specified, the global morse cutoff is used.
this is a reference to the fact, that for pairwise additive potentials,
you may enter different cutoffs for different pairs of atom types in the
pair_coeff statement and thus override the value from the pair_style
I also found that somebody said that the cutoff will be set as 12
...and you'll find somebody saying, that 14 or 15 is better and others
recommending less. for certain models, it must be much less.
I am confused how to set the cutoff of the morse and the lj potential to
make the result correct.
how do you determine, whether your result is correct or not?
how do you know it is not due to using a hybrid potential setup for a
system where this is not a reasonable model or some other reason?
a cutoff for either LJ or Morse *is* an approximation, and you can easily
determine how much of an approximation by computing the remaining force at
the respective cutoff. you will see, that the approximation gets better, if
the cutoff is larger. for some geometries, these remaining forces mostly
cancel, and the resulting error is small for a given cutoff, for other
geometries it is not as small.
typically, the paper describing the parameterization where you take the
potential from should give some recommendation for the system the potential
parameters are suitable for and what a meaningful choice of cutoff is.