[gmx-users] parinello-rahman coupling

[Tsjerk Wassenaar]

Hi Loris,

You're always stuck with limitations of simulations one way or the
other. Now, berendsen barostat is disputed in some sense, but it does
a good job usually (and it is so simple). The question would be, how
far off is it from THE thermodynamic ensemble of your system (and how
far off are the others). Then you have to wonder how on earth you're
going to get THE thermodynamic ensemble with an empirical force field.
But then again, you'll probably be close enough for what you want to
see.

Now, the thing with different starting conditions giving you different
trajectories, means that you have to do multiple runs to get a proper
sample from the distribution of possible trajectories (starting from a
certain region in conformational space). You need statistics.

Hope it helps,

Tsjerk


On 3/27/07, lorix <Loris.Moretti at pharm.unige.ch> wrote:
> Hi Tsjerk,
>
> I set the pressure in the parameter file to be 1bar and I get different
> average values ranging from 0.4 to 1.7 bar. Maybe they are not significantly
> different (?) but I try to simulate large conformational changes and
> according to
> what I saw small difference in settings can make different trajectories.
> In my simulations Berendsen barostat can keep the mean value of the
> pressure, but
> according to the manual it does not yield a correct thermodynamic ensemble.
>
> thanx
>
> loris...
>
>
>
> Tsjerk Wassenaar wrote:
>
> > Hi Lorix,
> >
> > Sorry for that. You used the word "maintained" from which I inferred
> > you were referring to the fluctuations (biased by the occurrence of
> > the question obviously). Under normal conditions the equilibration
> > shouldn't take that long. How far off is the mean? Do you still
> > observe a trend? Still note that you're dealing with enormous, both
> > positive and negative, pressures, from which you try to extract an
> > average value. Such an average will usually not be very exact because
> > of the numerical inaccuracies.
> >
> > Best,
> >
> > Tsjerk
> >
> > On 3/27/07, lorix <Loris.Moretti at pharm.unige.ch> wrote:
> >
> >> Hi Tsjerk,
> >>
> >> Thanks for your prompt reply.
> >> I was asking about the average value of the property.
> >> I understood that large fluctuations are reasonable, but I was expecting
> >> that the mean value
> >> over 15 ns could be somehow as the set in the parameter file as happens
> >> with Berendsen pressure coupling.
> >> I checked the archives and I did not find much about the mean value.
> >>
> >> best regards,
> >>
> >> lorix...
> >>
> >>
> >> Tsjerk Wassenaar wrote:
> >>
> >> > Hi Loris,
> >> >
> >> > Enjoying the nice weather here, I'll just give a gentle reiteration of
> >> > previous remarks. What you observe is caused by the fact that you only
> >> > have a very small system from which you try to extract a macroscopic
> >> > property (i.e. defined only for large systems). The large fluctuations
> >> > are normal. It would be a good idea to read some general texts about
> >> > statistical mechanics and molecular dynamics and before posting a
> >> > question to the user list, it would also be good to browse through the
> >> > archives of this list, to see if the question you're about to ask has
> >> > already been answered (and trust me, this one has been, many times).
> >> >
> >> > Good luck with your simulations,
> >> >
> >> > Tsjerk
> >> >
> >> > On 3/27/07, lorix <Loris.Moretti at pharm.unige.ch> wrote:
> >> >
> >> >> Dear gmx-users,
> >> >>
> >> >> I am a new one...
> >> >> coming to the point...I am simulating in parallel different
> >> >> configurations and settings of a protein
> >> >> around 300 residues large plus approximately 12000 water molecules.
> >> >> I am simulating NPT with berendsen for P and T to equilibrate over
> >> 100
> >> >> ps and then the simulation
> >> >> for several ns (5-20) with nose-hoover for T and parinello-rahman
> >> for P.
> >> >> But in the last step I cannot find a value for tau_p to maintained
> >> and
> >> >> average of 1 bar.
> >> >> Reducing the tau_p value (0.06, 0.05 and 0.01) I get many messages in
> >> >> the log file like:
> >> >>
> >> >> Grid: 11 x 10 x 9 cells
> >> >> Grid: 11 x 11 x 9 cells
> >> >>
> >> >> and then the simulation stops randomly without any messages
> >> besides the
> >> >> lam-mpi ones.
> >> >>
> >> >> does anyone have ideas?
> >> >>
> >> >> thanks
> >> >>
> >> >> loris...
> >> >> _______________________________________________
> >> >> gmx-users mailing list    gmx-users at gromacs.org
> >> >> http://www.gromacs.org/mailman/listinfo/gmx-users
> >> >> Please search the archive at http://www.gromacs.org/search before
> >> >> posting!
> >> >> Please don't post (un)subscribe requests to the list. Use the
> >> >> www interface or send it to gmx-users-request at gromacs.org.
> >> >> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> >> >>
> >> >
> >> >
> >>
> >> _______________________________________________
> >> gmx-users mailing list    gmx-users at gromacs.org
> >> http://www.gromacs.org/mailman/listinfo/gmx-users
> >> Please search the archive at http://www.gromacs.org/search before
> >> posting!
> >> Please don't post (un)subscribe requests to the list. Use the
> >> www interface or send it to gmx-users-request at gromacs.org.
> >> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> >>
> >
> >
>
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>


-- 
Tsjerk A. Wassenaar, Ph.D.
Junior UD (post-doc)
Biomolecular NMR, Bijvoet Center
Utrecht University
Padualaan 8
3584 CH Utrecht
The Netherlands
P: +31-30-2539931
F: +31-30-2537623