[gmx-users] pressure coupling questions

Mu Yuguang (Dr) YGMu at ntu.edu.sg
Fri Oct 17 11:38:01 CEST 2003


It is normal.

-----Original Message-----
From: Chng Choon Peng [mailto:cpchng at bii.a-star.edu.sg] 
Sent: Friday, October 17, 2003 5:34 PM
To: gmx-users at gromacs.org
Subject: Re: [gmx-users] pressure coupling questions

Hi Erik,

  Along the same line, I have observed pretty large fluctuations in
pressure
in my NVT + NPT simulation for 10ns. Below are some numbers:

Energy           Average       RMSD     Fluct.      Drift  Tot-Drift
-------------------------------------------------------------------
Temperature      309.915    2.76174    2.76174 -4.33312e-07 -0.00216656
Pressure (bar)   1.00917      230.1      230.1 6.10213e-05   0.305106
Density (SI)     997.553    3.64684    3.64397 -0.000100151  -0.500756
Volume           70.0431   0.255603   0.255401 7.04046e-06  0.0352023

My reference T is 310K and my reference P is 1atm. However,
there's a fluctuation of +/- 230 around the desired average value of
1.0.
But density fluctuations are only around 0.3%.

Is this considered normal?
Or I should turn off pressure-coupling after say 1ns as what you
suggested?

Thanks!


Regards,
Choon Peng
Bioinformatics Institute, Singapore

On 10/1/03 1:53 PM, "Erik Lindahl" <lindahl at csb.stanford.edu> wrote:

> Hi,
> 
> First, NVT and NPT are well-defined ensembles, and not all
> pressure/temperature coupling methods reproduce them exactly.
> (Nose-Hoover for temperature and Parinello-Rahman for pressure).
> 
> "N" means constant number of particles.
> "V" means constant volume (no pressure coupling) and "P" constant
> pressure (=coupling).
> "T" means constant temperature and "E" constant energy (no temperature
> coupling).
> 
> Now, you might be tempted to always use the well-defined ensemble
> coupling methods, but that's not always a good solution. The methods
> above (often called "extended ensemble") lead to larger fluctuations
> and can sometimes oscillate. For instance, if you want to simulate at
> constant pressure it is usually better to first run a short simulation
> (say 100ps) with berendsen pressure scaling which is guaranteed to
> converge rapidly. Then you just turn off the pressure scaling and keep
> the box constant during the rest of the run.
> 
> Gromacs reports pressure in bar. A negative pressure means the box
> would like to contract, i.e. your density is slightly too low.
Pressure
> will fluctuate a lot in a microscopic system (thousands of bars), so
> you can't tell anything until you have an average over a couple of
> picoseconds. Even then,  -1000 isn't horribly off for a condensed
phase
> system - your density will probably only change by a couple of percent
> if you enable pressure coupling.
> 
> Actually - as soon as you enable pressure coupling there will be a
term
> in the energy file for the density (measured in SI units = kg/m3).
> 
> Cheers,
> 
> Erik
> 
> 
> On Tuesday, September 30, 2003, at 07:17 AM, Q.Z. wrote:
> 
>> Dear All,
>> I am new to md and gromacs. According to the manual and some the
>> tutorial, one can set either temp coupling or pressure coupling or
>> both at the same time. Does the temp coupling alone mean to use NVT
>> ensemble and pressure alone NPT ensemble? What is it when use both
>> coupling?
>> If I turned off pressure coupling in my md but temp coupling on, I
got
>> about average pressure of -1000 bar. What does that mean? Is the
>> density of my system too high? If it is, how should I change it?
>> 
>> Thanks for any input.
>> 
>> Q. Zou Ph D.
>> Indiana Univ.
>> 
> 
> _______________________________________________
> gmx-users mailing list
> gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> 

_______________________________________________
gmx-users mailing list
gmx-users at gromacs.org
http://www.gromacs.org/mailman/listinfo/gmx-users
Please don't post (un)subscribe requests to the list. Use the 
www interface or send it to gmx-users-request at gromacs.org.




More information about the gromacs.org_gmx-users mailing list