Thank you, David. I did use PMEMD and it was NVT so density was fixed.
I changed the safety from 1.25 to 10.25 and it works, but what does this
"safety" do? Does making it bigger slow down the simulation?
On Thu, Feb 18, 2010 at 9:43 AM, case <case.biomaps.rutgers.edu> wrote:
> On Thu, Feb 18, 2010, Dian Jiao wrote:
> >
> > I was running simulation with amoeba force field. It is a very small
> > system ~2000 atoms.
> >
> > The first 2ns was done ok. But when I restarted the simulation after
> that,
> > it kept dying due to "Too many dipole_dipole interactions for allocated".
> > And everytime I restarted the job where it stopped, it would run for a
> while
> > and then died again for the same reason. So in principle, I could
> continue
> > the simulation to no matter how long I want but it just does not look
> right
> > to me. What is causing that? Is it memory related?
>
> Is the density changing during the run?
>
> My suggestions, in order of preference:
>
> 1. use pmemd.amba, not sander. It is faster, better tested, and parallel
> to
> boot.
>
> 2. try increasing the "safety" parameter in amoeba_direct.f in sander. The
> code is trying to estimate how many intereactions will be needed, and it
> may
> be making too conservative an estimate.
>
> ....dac
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Feb 18 2010 - 12:30:02 PST
