From: Cesar Luis Avila (cavila_at_fbqf.unt.edu.ar)
Date: Mon Aug 20 2007 - 16:47:37 CDT
This kind of crash does not mean that your simulation is wrong. As I
told you before, this parameter is related to space partitioning, ie, on
how namd manages to assign a subspace of the simulation to different
processors. You could prevent it to happen using larger margin values,
but this will probably result on worst scaling. What you should do in
these cases is to restart the simulation from a previous restart point.
This way namd will be able to reassing the subspaces taking into account
the current dimensions.
I am sorry if I messed it all up on my explanations. Perhaps someone on
the list can give a better explanation?
Regards
Cesar
I think there was something on NAMD WIKI about this?
Adam Fraser escribió:
> Thanks again Cesar. I did notice the box fluctuations, in fact they
> were causing my simulations to crash , which is why someone suggested
> using margin 1-3 in the first place. I was able to prevent the
> crashes by making a slightly larger waterbox in the first place, but I
> still don't understand the margin 1-3 and can't find anything
> documenting it on the web. I'm going to try NPT on a much larger
> system to see what happens with the energies and pressure there.
> Interesting that you say the pressure fluctuations are normal though,
> since NPT implies constant pressure. I'm learning anyway.
>
> -Adam
>
> On 8/20/07, *Cesar Luis Avila* <cavila_at_fbqf.unt.edu.ar
> <mailto:cavila_at_fbqf.unt.edu.ar>> wrote:
>
> In the simulations I have performed so far, instant pressure always
> presents large fluctuations. What you should monitor is the box size
> fluctuations. You should get to a constant volume. I don't know how to
> properly describe the margin keyword, but I will give it a try. I has
> something to do with the partitioning of the total simulation
> space into
> patches. If you expect the volume to change you should give a rather
> large margin, for example 5. When your box reaches a constant volume
> you can change it to 1 or even 0 if you switch to NVT.
> I don't know what might be the relationship between margin and PME
> that
> you mention.
> Regards
> Cesar
>
> Adam Fraser escribió:
> > Thanks Cesar,
> > We've been trying to equilibrate in NPT actually by using
> Nosé-Hoover
> > Langevin piston pressure control, but are having trouble. The
> > pressure reports throughout the equil vary +/-1000 bar. The average
> > pressure is 1bar, but the fluctuations are extremely large and I
> don't
> > know why.
> >
> > I submitted this question to the group before and received a few
> > different suggestions... amongst them was one suggesting to use
> margin
> > 1.0-3.0 because the fluctuations were an artifact of faulty ewald
> > summation due to coarse grid size in ewald. I haven't tried
> this as I
> > don't yet understand it completely, but it does make sense
> because the
> > system's energy is also fluctuating strangely.
> >
> > Any input you might have on this would be great.
> >
> > Thanks very much,
> > Adam
> >
> > On 8/20/07, *Cesar Luis Avila* < cavila_at_fbqf.unt.edu.ar
> <mailto:cavila_at_fbqf.unt.edu.ar>
> > <mailto:cavila_at_fbqf.unt.edu.ar <mailto:cavila_at_fbqf.unt.edu.ar>>>
> wrote:
> >
> > Dear Adam,
> > I think that the solvation box was built with a rather low
> density to
> > avoid getting high VdW forces when building the system. You are
> > expected
> > to run a few equilibration steps using NPT until you get to the
> > desired
> > density. Then you can switch to the ensemble you want to use.
> > Hope this helps
> > Cesar
> >
> >
> > Adam Fraser escribió:
> > > I was wondering if there is a way to specify a solvation
> density of
> > > 1.0 (rather than 0.93) using vmd.
> > >
> > > This is the script I'm currently using in vmd:
> > >
> > > resetpsf
> > > readpsf 2HEX.psf
> > > coordpdb 2HEX.pdb
> > > resetpsf
> > > package require solvate 1.2
> > > solvate 2HEX.psf 2HEX.pdb -o 2HEX_solv2 -s WT -b 2.4
> -minmax { {-7.5
> > > -12.5 -12.5} {47.5 12.5 12.5} }
> > >
> > > Any help or suggestions would be great. Thanks,
> > > Adam
> >
> >
>
>
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:45:07 CST