Dear LAMMPS users,
I’m currently running simulations on LAMMPS (29 Sep 2021) with rods under DLVO potential (added to the input file with pair style table).
For the limit kD → 0 it behaves as a coulombic potential (long range).
I would like to know if anyone has already tried and succeeding in implementing Ewald summation for this kind of potential.
Thank you for your attention,
Laura
Laura,
most people that implement features into LAMMPS don’t tell any of the developers or users about it, until they are “finished” and related publications are written and accepted for publication. Which also makes it unlikely that they respond to a request like yours (there are exceptions, but they are rare).
You may have a little more success submitting a feature request issue on github, but since your use case is rather special, I would not get my hopes up, if I were you.
@stamoor may have some suggestions for you about what to look out for, in case you plan to implement it yourself.
Since LAMMPS is a project that mostly “lives” from contributions of its users, there is very little “control” and exchange of information about who is doing what and when.
Sorry,
Axel.
I’d be quite surprised if anyone is working on a long range solver for DLVO, and that’s because DLVO (as I understand) is a theory about how electrostatic interactions become screened by solvent ions – and hence become short ranged.
So I would expect much more bang for buck if you just expanded your cutoffs to reach the desired accuracy – exponential forces drop off very quickly with distance. That way you merely have slower simulations, instead of weeks of development time (and possible bugs!).
That said, the math and coding aren’t that hard. Your main job is to calculate the appropriate force and energy kernels in Fourier space, and code them up. There is a good, modern exposition in Todd Gingrich’s 2010 thesis on ionic simulations (Chapter 3).
Also, I believe (1) LAMMPS has implemented the oxDNA2 coarse-grained DNA model which (2) uses Debye-Huckel electrostatics, which should be very similar to what you want.