[Pw_forum] Problem in energy conservation in NVE MD using cp.x

Nicola Marzari marzari at MIT.EDU
Wed Jul 15 16:21:32 CEST 2009




Dear Hiroshi,


thanks again for your observations - it turns out that in recent times 
(years) a small modification had arisen in the way the center of mass
is treated that gave rise (in the very exotic case that you have been
looking at) to the observed behavior. Roberto Car and his group 
(Federico Zipoli and Zhaofeng Li) has also noticed this a few months
ago, but it has percolated back into the code just now.

You can download the working version from the cvs, with these two commands:

cvs -d :pserver:anonymous at scm.qe-forge.org:/cvsroot/q-e login
cvs -d :pserver:anonymous at scm.qe-forge.org:/cvsroot/q-e checkout espresso

or wait for a few days for the new release of the code, coinciding
with the Santa Barbara school.

If you apply it to your example, you'll notice that the constant
of motion of the CP lagrangian is perfectly conserved, the physical
constant of motion is very well conserved (as it should be), and also
that after a fraction of a ps the water molecule will take up some 
kinetic energy (average around 10K, i.e. osciallting between 0 and 20K)
due to the transfer of momentum between the fictitious system and the
physical one. This is normal, and the relatively large amount is due to
this very unusual case, using on top a lot of plane waves. It would be 
negligible in any large system.

All the best,

			nicola


-- 
---------------------------------------------------------------------
Prof Nicola Marzari   Department of Materials Science and Engineering
13-5066   MIT   77 Massachusetts Avenue   Cambridge MA 02139-4307 USA
tel 617.4522758 fax 2586534 marzari at mit.edu http://quasiamore.mit.edu



More information about the users mailing list