Greetings all,
I would like to know what happens if there is an overlap of particles
when using the lubricateU pair_style. Does the simulation stop, use
an alternate separation distance, etc.?
No - it will probably crash due to the potential blowing up.
From the doc page:
The Asq (squeeze) term is the strongest and is included as long as
flagHI is set to 1 (default). It scales as 1/gap where gap is the
separation between the surfaces of the 2 particles. The Ash (shear)
and Apu (pump) terms are only included if flaglog is set to 1. They
are the next strongest interactions, and the only other singular
interaction, and scale as log(gap).
Both terms go bad as gap->0.0.
This is no different than highly overlapped LJ particles. It's up
to you to build a model that is consistent with the potential
you are using.
Steve
This pair_style appears to be poorly documented, in that the
resistance matrix specifics do not appear in the reference Kumar and
Higdon PRE (2010). In fact, they merely refer to the isotropic term
as the short time diffusivity, and refer to an unpublished manuscript
for derivation of this novel term. The authors also quote the
accuracy as being comparable to Stokesian Dynamics, again referencing
the unpublished manuscript. I feel the lammps community should have
reservations about using potentials that do not have readily available
documentation and benchmarking.
Is there a different document where the specifics of the formulation are shown?
Well, you can send Kumar or Higdon an email with questions.
This work was part of Kumar's thesis and Higdon is a well known
prof in the SD community. So I doubt there are conceptual
issues, though there may be bugs in the implementation.
I will say that this pair style is meant to be used with one
or more other pair styles which provide the particle-particle
interactions. E.g. with pair_style colloid which insures no
overlap between 2 particles.
If you can clarify anything in the doc pages, we'd be happy
to post updates.
Steve
Thanks for the suggestions.