[Pw_forum] Very large value of temperature at the beginning of a nose thrmostat run

[Axel Kohlmeyer]

On Fri, 10 Aug 2007, Lawrence Lee wrote:

LL> Dear all,

dear lawrence, 

LL> I am a newbie in using cp.x of the espresso package. I have a some
LL> problems concerning the use of nose thermostat in temperature
LL> control.

LL> First I do an electronic and ionic minimization on my system. And
LL> finally the output file said convergence has been achieved. Then I
LL> restart from this to make a 3000K run.

LL> Problem then emerge: when I looked at the cp.evp file, I found that
LL> the temperature has been so much higher than 3000K at some steps
LL> near the beginning (~0.079 ps to 0.089 ps, ie, 272 to 307 steps as I
LL> use dt =12), so high that the value is ****** (>10000K). However,
LL> after some 0.5 ps, the value stabilized, and fluctuating about 3000K
LL> with amplitude about a few hundred K.

possibly you have started from a structure that was only a local minimum
with respect to a 3000K system and/or seen a phase transition. it also
is possible that your simulation has gone completely wrong. what are the
values for the fictitious kinetic energy? what happens if you run
born-oppenheimer (cg as electron dynamics)? does the structure look like
what you would expect? does some published data from simulations on your
system exist or for something similar? can you reproduce that?
there are other possible sources of problems, that can only be seen 
from the input and relevant parts of the output. right now, all we can 
do is 'fish in the dark'.

cheers,
   axel.

LL> Other quantites after that about 0.5 ps all look fine, but that
LL> intermediate steps with very high temperature worries me. In fact,
LL> does it matter?

LL> Lawrence Lee Shun Hang
LL> MPhil student of the Chinese University of Hong Kong
LL> tel: 852-61094093
LL> 
LL>        
LL> ---------------------------------
LL> ¨Ï¥ÎYahoo! Mail µL­­¹q¶lÀx¦s¶q¡A¥Î±o·U¦h¡A½à§AµL­­Åå³ß!¥ß§Y¦æ°Ê 

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