neighbor list-pairwise interactions

Dear Lammps users,
I do have a question regarding the pairwise neighbor list. Could anyone please tell me where in the code it is being built and where it is being updated.

In the pairwise interaction (e.g. pair_lj_cut_coul_long.cpp) that pair interactions are being counted for some pairs twice (e.g i&j and i& image of j). can anyone please help? Why is this needed?

Thanks
emily

Dear Lammps users,
I do have a question regarding the pairwise neighbor list. Could anyone
please tell me where in the code it is being built and where it is being
updated.

in the Neighbor class, of course.

you should have a closer look at this document:
http://lammps.sandia.gov/doc/Developer.pdf
also, learning how to use a tool like grep will
be highly beneficial.

In the pairwise interaction (e.g. pair_lj_cut_coul_long.cpp) that pair
interactions are being counted for some pairs twice (e.g i&j *and* i& image
of j). can anyone please help? Why is this needed?

you must be misreading the code.

where exactly is a reference to "image of j"?

the pair styles are completely agnostic
to what is an image of a particle. this is
all handled outside when the verlet lists
are built.

you probably also want to have a look at
steve plimpton's paper describing the basic
algorithms and parallelization strategies in
LAMMPS (which have not changed significantly
in over a decade).

axel.

you probably also want to have a look at
steve plimpton’s paper describing the basic
algorithms and parallelization strategies in
LAMMPS (which have not changed significantly
in over a decade).

Could you tell us the reference/link to this paper?

Thanks!

Cy.

you probably also want to have a look at
steve plimpton’s paper describing the basic
algorithms and parallelization strategies in
LAMMPS (which have not changed significantly
in over a decade).

Could you tell us the reference/link to this paper?

It is linked prominently on the lammps homepage.

lammps.sandia.gov/cite.html

Axel

Hi,

I am tackling that part of the code too. I found that paper

S. Plimpton, J. Comp. Phys. 117, 1-19, 1995.

particularly useful. To the best of my knowledge, LAMMPS only uses domain/spatial decomposition.

Hope this helps!

Best,

Hu