[Pw_forum] temperature oscillations with nose' thermostat

[Axel Kohlmeyer]

On Mon, 20 Jun 2005, Konstantin Kudin wrote:

dear kostya,

KK>  If these oscillations concern you, you could run with the Nose-Hoover
KK> chains in CPV from the current CVS. With the chains the oscillations
KK> are still present, however, the highs are lower and the lows are
KK> higher. Thus the plot of temperature vs. time step looks way nicer!

two more remarks about this. having low temperature fluctuations is
not really an important point here. with few atoms (and for AIMD runs
you basically always have few atoms) the fluctuations _have_ to be
large. with larger systems they average out more and more, but they
are still present if you plot the instantaneous temperature/kinetic
energy of a small subsystem. in fact, you _need_ those fluctuations,
so that your system can overcome energy barriers. so the fluctuations
per se are not a bad thing. you should watch out, however, for a 
temperature _drift_. for an MD simulation of 'only' one picosecond, 
there should be (almost) none (provided you are starting in equilibrium).

using nose-hoover thermostat chains is a Good Thing(TM) nevertheless,
but for a different reason: the nose-hoover chains are a very important
means to ensure ergodicity, i.e. the fact that that you sufficiently 
explore phase space. the 'normal' nose-hoover does not do that.
the smaller fuctuations in the total instantaneous temperature 
are just a side effect of that.
for a system with a small number of atoms (say, less then two hundred),
you should even have one nose-hoover chain for each degree of freedom.
a simple example to illustrate that: consider an MD of a single
water molecule in gas phase. if the starting conditions are selected
in such a way, that only the bending mode is sufficiently populated,
it will take a very long time, until the stretch modes are populated,
since those modes (almost) don't couple to the bending mode.


best regards,
	axel.

KK>  Kostya
KK> 
KK> 
KK> --- Massimiliano Bonomi <massimiliano.bonomi at mi.infn.it> wrote:
KK> 
KK> >   Dear users,
KK> > I'm running molecular dynamics with fpmd (in espresso v. 2.1.3
KK> > package) 
KK> > of a solid assembled with cluster of gold.
KK> > After reaching the electronic ground state i've started the dynamics 
KK> > using nose' thermostat
KK> > both on ions and electron, choosing the right thermostat frequencies 
KK> > (for ions half the higher phonon frequency, for electrons 2/3 times
KK> > the 
KK> > higher phonon frequency).
KK> > 
KK> > During a simulation of 10000 time step (~ 1.0 picoseconds) at T=300K 
KK> > 
KK> > the temperature of the system  made big oscillations around 300K
KK> > (even 
KK> > of 200K!!).
KK> > 
KK> > Is it normal?? How can i reduce the width of these oscillations?
KK> > 
KK> > Best regards and thanks in advance.
KK> > Massimiliano Bonomi.
KK> > 
KK> > -- 
KK> > 
KK> > Massimiliano Bonomi
KK> > Dipartimento di Fisica, Università degli Studi di Milano  
KK> > via Celoria, 16 - 20133 Milano, Italy
KK> > ph. +39 02 50317 717
KK> > e-mail: massimiliano.bonomi at mi.infn.it, massimiliano.bonomi at unimi.it
KK> > 
KK> > 
KK> > _______________________________________________
KK> > Pw_forum mailing list
KK> > Pw_forum at pwscf.org
KK> > http://www.democritos.it/mailman/listinfo/pw_forum
KK> > 
KK> 
KK> 
KK> 
KK> 		
KK> ____________________________________________________ 
KK> Yahoo! Sports 
KK> Rekindle the Rivalries. Sign up for Fantasy Football 
KK> http://football.fantasysports.yahoo.com
KK> _______________________________________________
KK> Pw_forum mailing list
KK> Pw_forum at pwscf.org
KK> http://www.democritos.it/mailman/listinfo/pw_forum
KK> 
KK> 

-- 

=======================================================================
Dr. Axel Kohlmeyer   e-mail: axel.kohlmeyer at theochem.ruhr-uni-bochum.de
Lehrstuhl fuer Theoretische Chemie          Phone: ++49 (0)234/32-26673
Ruhr-Universitaet Bochum - NC 03/53         Fax:   ++49 (0)234/32-14045
D-44780 Bochum  http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.