[CPMD-list] TEMPCONTROL ELECTRONS

[Axel Kohlmeyer]

On Tue, 29 Nov 2005, Eduard Schreiner wrote:

eddi, younes,

ES> On Tuesday 29 November 2005 12:12, Younes Ansari wrote:
ES> > Dear CPMD-list:
ES> >   As I know the TEMPCONTROL ELECTRONS is the fictitious kinetic energy of
ES> > the electrons in (a.u.). Would you please tell me which is the best way to
ES> > find a credible value for the system.

actually, you _don't_ want to control the fictitious kinetic energy of 
the electrons at. it should be small and conserved. but see below for 
more comments.
 
ES> >
ES> >
ES> > ---------------------------------
ES> >  Yahoo! Music Unlimited - Access over 1 million songs. Try it free.
ES> 
ES> Hey Younes,
ES> in order to find a value for the electronic temperature you can thermostat at, 
ES> you could do a nonthermostatted CPMD using a short time step, say 2 au,
ES> and see what is the native electronic kinetic energy. Then use that as the 
ES> target temperature.

this is mainly for very long trajectories where you always accumulate 
a little numerical noise and that will show as a small gradual heating.
then you have to use a thermostat to maintain adiabacity.
note, that using a thermostat on the electrons can hide the fact, that
you have a too small band gap or that your time step is to large and 
_then_ even if your energies look ok, you may have a trajectory that
is completely nonsensical. there was a recent discussion on that
topic in the q-espresso mailing list, please have a look at:
http://www.democritos.it/pipermail/pw_forum/2005-November/003289.html
and the responses to it.

ES> Why do you want to use TEMPCONTROL instead of  NOSE ELECTRONS?

in case you are doing some kind of annealing or have a high energy 
starting structure, TEMPCONTROL ELECTRONS can be useful as 'safety net' 
that quenches the electrons whenever their kinetic energy is about 
to go haywire. e.g. in the form:
  TEMPCONTROL ELECTRONS
     0.005 0.02
which should only get triggered if EKINC goes beyond 0.025 a.u.
cpmd will print a message, whenever a rescaling of the velocities
of the orbitals has occured. you should get worried, if it happens
more than a few times. note that, as eddi has stated, the 'target'
'temperature' depends on the number of electronic degress of freedom
and the fictitious electron mass and thus changes from system to system.

same as with TEMPCONTROL IONS you don't want to have 
this in a production run, as both don't preserve a statistical
mechanical ensemble.

axel.

ES> 
ES> 
ES> Eddi
ES> _______________________________________________
ES> CPMD-list mailing list
ES> CPMD-list at cpmd.org
ES> http://cpmd.org/mailman/listinfo/cpmd-list
ES> 

-- 
=======================================================================
Axel Kohlmeyer   akohlmey at cmm.chem.upenn.edu   http://www.cmm.upenn.edu
   Center for Molecular Modeling   --   University of Pennsylvania
Department of Chemistry, 231 S.34th Street, Philadelphia, PA 19104-6323
tel: 1-215-898-1582,  fax: 1-215-573-6233,  office-tel: 1-215-898-5425
=======================================================================
If you make something idiot-proof, the universe creates a better idiot.