[CPMD-list] EHAM decreases in NVE MD

[Axel Kohlmeyer]

On Wed, 11 Apr 2007, liu liuxiandong wrote:

LL> Dear all,

dear liu,

LL> I performed NVE CP MD to a mineral phase. I have optimized the wavefunctions
LL> to 1.0d-6 and then set the following in the input file,
LL> 
LL> TEMPCONTROL IONS
LL>   400.0 30.0

sorry to disagree. with TEMPCONTROL IONS you do _not_ do an MD 
simulation in an NVE ensemble. in fact, you are not in any of 
the standard ensebles. please consider carefully what TEMPCONTROL 
does. every time the _instantaneous_ kinetic energy in your system
gets higher than the equivalent of 430K, the kinetic energy will
be rescaled to 400K (the same will happen at 370K).

as ari pointed out. your system is not yet in equilibrium, in
fact it has too high a potential energy for 400K, so there will
be a lot of rescaling happening until the system has reached 
equilibrium. ideally, you'll want to keep going with tempcontrol
until the temperature oscillates regularly between the boundaries.
 
LL> But I find the 400.0 K is the lower limit during the MD run and the EHAM
LL> decreases.
LL> I guess that the decreasing EHAM was because some hydroxyl groups in the
LL> input stucture are a little far away from the equilibrium positions. Then ,
LL> is this output result normal  ? And is there some quicker way to attain the
LL> EHAM conservation ? Thanks so much.

EHAM should be conserved if you do not use TEMPCONTROL, but then you'll
probably end up with a temperature that you do not want. ;-)

the real problem is to get rid of the excess potential energy and for
that, as ari suggested, doing a (partial) geometry optimization might 
be faster. but you should be careful, if you optimize the geometry too
much, you'll have to put back some of the energy that you extracted.

please take also note of the fact, that having a conserved EHAM does
not mean that your system is in equilibrium. with using tempcontrol 
it very frequently happens (particularly for the small systems that 
you can handle with CPMD), that you excite only a few vibrational
degrees of freedom and your ergodicity is low. so after you have reached
somewhat of an equilibrium with TEMPCONTROL, it is highly recommended
to run with NOSE IONS MASSIVE (i.e. one _separate_ thermostat for each
degree of freedom) to have better equipartitioning of the vibrational
energy and a better chance to reach equilibrium (you can always go 
back to a regular nose-hoover chain thermostats later). otherwise, your
system my have a sudden drop in potential energy during production
and you'll have to throw away part of your trajectory.

hope this helps,
   axel.

p.s.: i would appreciate it a lot, if people would state their name and
affiliation in e-mails (just configure your mailer to add a proper 
signature. it makes it _so_ much easier to respond, if you know who you
are talking to, especially if it goes beyond simple technical problems.
thanks in advance.


-- 
=======================================================================
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.