[Pw_forum] minimization in CP molecular dynamics

Nicola Marzari marzari at MIT.EDU
Thu Jul 10 10:20:06 CEST 2008



Shouldn't ekinc be zero ? I forgot the name used, but in a properly
executed CP calculation, in which you do damped dynamics on the
electrons to reach the ground state at fixed ions, the kinetic energy
of the electrons (as opposed to the quantum kinetic energy of your
system, ekin) should be zero.

I suggest to anyone learning CP calculations (a very worthy
endeavour, and one that would be extremely useful in the long
run) to study carefully the review paper on the topic by Marx and
Hutter,

http://www.theochem.ruhr-uni-bochum.de/research/marx/marx.pdf

CP in Q-E now can also do conjugate gradients minimization on
the electrons (more efficient than damped dynamics) and, for
metals, ensemble-dft (i.e. born-oppenheimer dynamics).

For cg/edft, there was one bug (memmory misallocation)
for parallel execution, just found by by Brandon Wood. It'll
be fixed in the cvs in the next few days.


			nicola

Mansoureh Pashangpour wrote:
> Dear Paolo
> I know one of them is in Ry and another one in Ha,but when I compare 
> them they are different
> scf:  total energy              =  -243.33284101 Ry
> 
> CP:  ekinc         :       10.67409      10.67409 (AU)
>    ekin          :      153.58930     153.58930 (AU)
>    epot          :     -458.58466    -458.58466 (AU)
>    totel energy  :     -258.18270    -258.18270 (AU)
>    temperature   :        0.00000       0.00000 (K )
>    enthalpy      :     -258.18270    -258.18270 (AU)
>    econs         :     -258.18270    -258.18270 (AU)
>    pressure      :       -3.47982      -3.47982 (Gpa)
>    volume        :      488.11442     488.11442 (AU)
> Thanks

-- 
---------------------------------------------------------------------
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 Pw_forum mailing list