[gmx-users] neighbor list

[Mark Abraham]

On Mon, Jul 28, 2014 at 8:27 PM, Sikandar Mashayak <symashayak at gmail.com>
wrote:

> Hi,
>
> I need to set up a simulation of LJ system, such that neighbor list is
> built using grid algorithm with a cut-off of lj_cut_off + 0.1 every 20
> timesteps.
>

So, you want a single-range scheme with an explicit buffer. This will be
faster if you use the Verlet scheme in GROMACS >= 4.6, not the group
scheme. Even more so if that will let you use a GPU...


> So I am wondering what should be the accurate settings in the grompp.mdp
> file.
>
> After reading the manual, I understand rvdw must be >= rlist.


No, that'll get you a twin-range scheme.

So how can I
> specify rlist=lj_cut_off+0.1, should I use rlistlong?
>

nstlist
rlist = rvdw + 0.1

and to force mdrun to use your explicit buffer (rather than tune the buffer
size automatically)

verlet-buffer-tolerance = -1

Also, the cut-off scheme 'group' for simple LJ atoms with no charge, is it
> equivalent to building neighbor list using atoms rather than group, since
> each charge group will have only a single atom?
>

Yes, so long as that is what your input topology specifies. "Charge groups"
can be specified even though all the charges are zero, and obviously you
don't want to do this! However, using the Verlet scheme means they're
always ignored anyway.

Mark


>
> Thanks,
> Sikandar
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>