Re: system becomes very large when set LangevinPiston on

From: Xiaoqing Wang (namdwxq_at_126.com)
Date: Wed Jul 27 2005 - 21:32:01 CDT

Hi Erik,

Yes, I compiled NAMD on AMD64 opteron cluster which I mentioned on the
first email:
http://www.ks.uiuc.edu/Research/namd/mailing_list/namd-l/2579.html
and run "namd2 equil.namd>equil.log" and then the problem happens.

I try running other example files which without constant pressure
control and only the values of pressure(in log file) is different from
the given example file.

I don't know why this happens and how to check and solve it.

Any tips would be greatly appreciated!

Sincerely,
Xiaoqing

在2005年07月27日的19:37,nordgren_at_sas.upenn.edu写道:
> Hi Xiaoqing,
>
> > Thank you very much for your explanation.
>
> Sure, you're welcome!
>
>
> > The PRESSAVG and GPRESSAVG values are close to 1 atm in the end. But I
> > don't know why the system becomes dispersing completely. The input file
> > "equil.namd" and xst file "equil_out.xst" are enclosed. Does it arise
> > from the pressure or some others?
> >
> > When I calculate the same input files by using "namd2" which I
> > downloaded binary of "Linux-i686" from web site :
> > http://www.ks.uiuc.edu/Development/Download/download.cgi?PackageName=NAMD
> > the unitcell of the system is close to the original value.
>
> I don't see anything wrong with your input file offhand.
>
> But are you saying that this problem (of your system "exploding") only
> happens when you run a certain version of NAMD, and with other versions of
> the code the problem disappears? If so, which version(s) are you referring
> to?
>
> - Erik
>
>
>
> > > The NAMD units of pressure are indeed in bars.
> > >
> > > However, in a typical constant-pressure simulation, the short-term
> > average
> > > pressure of the system (PRESSAVG) will typically fluctuate by hundreds
> > of
> > > bars, and the instantaneous value (PRESSURE) will often be in the
> > thousands.
> > >
> > > Moreover, you *may* see PRESSURE values that can look quite large (say,
> > in
> > > the tens of thousands) -- if you are using multiple-timestepping, but
> > > printing energies every step. This is because the long-range forces are
> > not
> > > being computed every step, and so when you look at the instantaneous
> > pressure
> > > during those off-steps, the numbers look pretty wild. (But it all comes
> > out
> > > correctly in the end.)
> > >
> > > IMO, the only parameter really worth paying much attention to is
> > "PRESSAVG"
> > > (or better, "GPRESSAVG"), and even then it is most meaningful when you
> > look
> > > at its longer-term average trend (which *should* be close to your
> > specified
> > > value of 1 atm or whatever).
> > >
> > > hope this helps...
> > >
> > > C. Erik Nordgren, Ph.D.
> > > Department of Chemistry
> > > University of Pennsylvania
> > >
> > >
> > > Quoting Xiaoqing Wang <namdwxq_at_126.com>:
> > >
> > > > Hi,
> > > >
> > > > What's the unit of the item PRESSURE of output file?
> > > > In the equil.namd input script file the LangevinPistonTarget is set to
> > > > 1.01325(bar). In my output file the PRESSURE is about 0 in the end.
> > But
> > > > in the output(equil.log) file which I downloaded from
> > > >
> > http://www.ks.uiuc.edu/Research/namd/tutorial/NCSA2002/hands-on/sample/
> > > > the PRESSURE value is about 1000 in the end.
> > > >
> > > > Sincerely,
> > > > Xiaoqing
> > > >
> > > > > Dear all,
> > > > >
> > > > > I downloaded "fftw-linux-amd64.tar.gz" and "tcl-linux-amd64.tar.gz"
> > from
> > > > > http://www.ks.uiuc.edu/Research/namd/libraries/ and plugins for
> > > > > LINUXAMD64 and "charm-net-linux-amd64.tar.gz" from
> > > > > http://charm.cs.uiuc.edu/autobuild/bin/ and NAMD_2.5_Source.tar.gz.
> > > > >
> > > > > I compiled NAMD with
> > > > > ./config tcl fftw plugins Linux-amd64-g++
> > > > > ./make
> > > > >
> > > > > I download input files from web site :
> > > > >
> > http://www.ks.uiuc.edu/Research/namd/tutorial/NCSA2002/hands-on/sample/
> > > > > and run "namd2 equil.namd>equil.log" on AMD64 opteron.
> > > > >
> > > > > When run equilibrate with langevinPiston on
> > > > > ---------------------------------------------
> > > > > # equilibrate volume with CAs restrained
> > > > > langevinPiston on
> > > > > run 5000
> > > > > output equil_ca
> > > > > ---------------------------------------------
> > > > > the systerm becomes very large and disperse.
> > > > >
> > > > > I compiled many times and run many other examples and found that all
> > of
> > > > > the systerms become disperse when set "langevinpiston on". I don't
> > know
> > > > > why the pressure value is incorrect and the pressure control can't
> > work.
> > > > >
> > > > >
> > > > > Any tips would be greatly appreciated!
> > > > >
> > > > --Xiaoqing
> > > > Wuhan University
> > > > Wuhan 430072
> > > > China
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:39:45 CST