[gmx-users] confused about rcoulomb<=rlist

[David Mobley]

Dear all,

I desperately need some clarification about how GROMACS handles
coulomb interactions with rcoulomb and rlist, and why it makes sense
to do things this way.

In particular, there was a thread on the revisions list
(http://www.gromacs.org/pipermail/gmx-developers/2006-April/001581.html)
a while back explaining that in 3.3.1, grompp checks to ensure that
rcoulomb=rlist for PME/Ewald/PPM. I was very surprised by this, as I'd
always been using rlist>rcoulomb, which I thought was best. Can
someone explain why one would want rcoulomb=rlist? I don't quite
understand the explanation on the revisions/developers list.

My confusion is exacerbated by the fact that the GROMACS 3.3 manual
includes the following statement:
"The neighbor search cut-off rlist should be 0.1 to 0.3 nm larger than rvdw
 to accommodate for the size of charge groups and diffusion between neighbor
 list updates."

This is exactly the same logic I was applying to rcoulomb when I
concluded that I wanted to make sure rlist was larger than rcoulomb --
basically, that I didn't want things outside rlist diffusing to within
rcoulomb between neighbor list updates, and hence not being counted in
the real-space interactions when they should be. Is there some other
concern that outweighs this or means that this point is irrelevant?

Thanks,
David Mobley
UCSF