[Pw_forum] CP for all systems

Nicola Marzari marzari at MIT.EDU
Sat Nov 26 17:07:12 CET 2005


Hi Paul,


these are all good ideas. My sugggestion would be the following:
we are giving CP some time to rest and debug (e.g., we have just
found a bug in the conjugate-gradients in parallel, that didn't
exist until a couple of weeks ago...). The plan is to use the time
until the end of December to iron out everything.

At that point, we'll also define some guidelines on how people should
contribute back - i.e. who to contact, how to submit an upgrade, and
what to write as a description. In this specific case, a description, 
and a reference to the literature, would also be very very important
Then, it would be ideal if you could yourself contribute that feature 
(seems an easy one), or coordinate with some of the "guiding centers"...


				nicola


Paul Tangney wrote:

> I have a further suggestion for developers:
> 
> How about providing a means of rescaling the masses of the
> ions in the code, as Sandro Scandolo and I showed was necessary
> for the correct calculation of temperature, thermal pressure,
>  and dynamical properties ?
> 
> A large part of the dragging effect is a renormalization of
> the ionic masses by an amount dM.  That means that
> temperature should be calculated as 3(N-1)kT = \sum (M+dM) v^2
> 
> A way to significantly improve dynamical properties and
> calculate the temperature more correctly is to evolve the ions using
> a mass of M-dM  (i.e. acceleration = F/(M-dM) ), and then to calculate
> temperature with a mass of M.
> (dM can be calculated by minimizing errors in Car-Parrinello
> forces w.r.t. ground state forces....this could also be facilitated
> in the code.)
> 
> It would be great if the code allowed you to input dM and have
> it do the rescaling, the calculation of temperature both with
> and without mass corrections (because away from the strongly-ionic
> limit the definition of temperature is less clear), and the
> calculation of the fictitious kinetic energy minus the
> dragging contribution.
> 
> I used to have all this implemented in FPMD, but somehow it never
> made its way into the official version.
> 
> Regards,
> 
> Paul

-- 
---------------------------------------------------------------------
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