How to calculate Coulomb potential without cutoff?

LAMMPS LAMMPS Mailing List Mirror
1

Hi all,

1 Done anyone know how to calculate Coulomb potential without cutoff?

2 If I use pair_style coul/cut with a cutoff LARGER than the box size, is that OK for LAMMPS to achieve the goal in question 1?

3 If I have lj/cut and coul/cut both in my system, does neigh_modify give the same behavior for changing neighbor lists for lj/cut and coul/cut?

2

For a periodic system, you must use a cut-off, which may or may not be
combined with a kspace calculation. In LAMMPS, the cutoff can exceed
the box dimensions, but making the cut-off very large will slow down
the calculation a lot.

For non-periodic systems, you can use a cutoff that is effectively infinite.

The last question about multiple neighbor lists is too vague. Read
the doc pages, and do some tests.

For more information on this fairly complicated topic, read the doc
pages more thoroughly and/or try a text book.

3

Perhaps the "fast multipole method" (FMM) would be useful to you,
especially if your system is not periodic.
I don't think you can use this algorithm in LAMMPS, but found a link
suggesting that it may be available if you use the DLPOLY program:
ftp://ftp.dl.ac.uk/ccp5.newsletter/35/ps/smith.ps
http://math.nyu.edu/faculty/greengar/shortcourse_fmm.pdf
(I don't use dlpoly, so I am not very helpful.)

I know there are other algorithms as well, but I do not know much
about them or which software uses them.
http://lammps.sandia.gov/threads/pdfDYOr4duqpQ.pdf

I don't know this helps you.
Cheers
Andrew

4

(I should mention that credit goes to Ashok Cutkosky, who told me
about this method. There's also a nice description of it in his
thesis.)

5

MSM, which is available in lammps, implements similar ideas, but has less problems.

Axel

6

http://lammps.sandia.gov/doc/kspace_style.html

I overlooked that. Thank you. Nice discussion.

I have a very strange question: does the MSM (kspace) style crash for
non-zero total charge? (non-periodic systems)

(I suspect Aiqun has something like this in mind. Changing the
direction of the forces, this might be useful for people simulating
gravitational stuff. I actually had something else I wanted to try.)
Cheers

Andrew

7

nevermind that. I'll try it and find out.
Cheers