Thank you very much for so a detailed reply
and the historical information on the reaction filed usage.
At the moment I will stick with that description and
maybe sometime later will look more into more advanced options.
i strongly recommend to first implement a variant
of pair style coul/cut and then use pair style
hybrid/overlay to add reactionfield coulomb
to any nonbonded pair style. that will reduce
effort and increase flexibility.
I guess it would still work if I do a variant of
lj/cut/coul/cut with additional parameter (on top of 2 LJ)
for the reaction field?
That would be less flexible, I understand, but might be a little bit
more suitable for my needs.