[Pw_forum] NVT dynamics

[Nicola Marzari]

Nichols A. Romero wrote:
> Hi,
> 
> Is the c.o.m. velocity a conserved quantity in NVT dynamics?
> For example, if the c.o.m. velocity is initially zero should it stay 
> that way for the rest of the simulation?
> 
> Would a larger system be more susceptible to "forces" introduced by a 
> finite tolerance
> on the SCF cycle?


Dear Nichols,


as you mention, the finite tolerance on self-consistency in 
Born-Oppenheimer molecular dynamics implies that the sum of all the
forces is not exactly zero - i.e. at every step you have a very small
random force on the total system. The larger the system, the more
difficult it becomes to screen very accurately the 1/G^2 term in the
electrostatic potential, so in principle (especially in a metal)
things could go slightly worse. In practice, a good strategy would be to
apply an "acoustic sum rule" to the scf forces that you calculate, and
make sure they sum to zero. You can think at it as a strategy to make
the approximate SCF result less approximate, in a way that conserves
a known symmetry of the system.

In a Car-Parrinello dynamics, you can have transfer of momentum
between the ionic degrees of freedom and the electronic ones,
and so it can be even more relevant.

Not sure what the current status is in either pw or cp - cp used to
enforce the sum rule on the ionic forces.

				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