Hi Axel,
So this might be related by my lack of familiarity with the neighbor list
and anglelist. I'm using a modified Mercedes-Benz potential where the
potential looks like
modified in which way?
exp(mag(rij)^2)exp(ni dot rij)exp(nj dot rij)
where ni and vectors normal to i-1,i,i+1 and j-1,j,j+1 respectively. i and
j are c-alpha indicies along a protein backbone and may be on the same
molecule or on separate molecules. The angle potentials I have looked at
seem to allow me to compute a potential / force due to a triple using the
angle list, but not between sets of triples.
the MB model that i know would be a 4-tuple that forms the 3 angles
and thus could be represented by an improper style, for example. even
though impropers are generally meant to keep three atoms connected to
a fourth in a plane formed by all four of them. there is nothing that
keeps you from using it for other purposes.
If I can just pairwise calculation over angles in the anglelist that would
make me happy because that seems very straight forward. I just got the
feeling from looking at the other pair potentials that I must use ilist not
anglelist to determine my i j pairs.
I should note that I'm a first year PhD student new to MD so while I think I
understand what's going on in the code, I could very well be mistaken.
then the first approach should always be to confer with your adviser
or senior colleagues. discussions on such conceptual issues are
difficult to make work via e-mail, particularly when we cannot be
certain that we are talking about the same things.
PS Sorry for calling you Alex in my first reply.
no problem. lots of people do; i barely notice it.
If you need me to elaborate further just let me know.
i am still struggling to really understand what you are asking for and
why you are asking to do it in the most backwards way i can imagine.
that may be the right way, and i am guessing wrong, but in general,
you only want to process neighborlists, if you are dealing with
arbitrary neighbors. as soon as the relationship is fixed by a
topology, then you'll be better off to process that kind of
information in the most direct way. looking up relations backwards,
however, can be very time consuming and also may collide with the
domain decomposition of LAMMPS, as there may be multiple copies of the
same atom id/tag in the same subdomain and one or none of them may be
a local atom.
axel.