units update?

[Axel]

On Mar 26, 3:48 am, Matthias <matthia... at psi.ch> wrote:
> Hi Axel,
>
> there are only roughly ten constants which are affected. I would
> suggest to keep the old CODATA 1998 constants and just to add the new
> CODATA 2006 constants together with an new input keyword in the GLOBAL
> section which allows to select the employed data set. The new

good point. one problem. init_physcon is called way before
the input file is read. i have not been able to find out yet
whether those factors are used in reading the input. if yes,
it would result in inconsistent behavior. it may be safer to
use a command line flag, which is processed before reading input
and then just re-initialize the values. any suggestions, comments?

> constants could become the default. In this way everybody can easily
> check the effect of such a change and the backward compatibility of
> released versions is achieved.
>
> Matthias
>
> On Mar 26, 7:12 am, Teodoro Laino <teodor... at gmail.com> wrote:
>
> > Hi Axel,
>
> > The patch is welcome as long as you keep only the fundamental units,
> > computing the
> > derived ones.. and anyway the changes should be *tiny* (hopefully
> > constants should not be changed
> > toooo much.. ;-) ).. just send the patch..
>

those things can add up fast, given a large enough problem.
i just had an "involuntary test" with CPMD and a semingly
minor change to the bohr/angstrom conversion in the QM/MM
interface code resulted in a >1kcal/mol change in the (total) energy.

cheers,
    axel.





> > teo
>
> > On 25 Mar 2008, at 22:55, Axel wrote:
>
> > > hi all,
>
> > > since there seems to be some "spring cleaning" (much appreciated, btw)
> > > going on
> > > in the cp2k cvs. one question about conversion factors and fundamental
> > > constants:
>
> > > does anybody object if i submit a patch to update the fundamental
> > > constants
> > > and conversion factors to CODATA 2006? since this will affect most if
> > > not all
> > > regtests it may be worth doing it at a time when everybody is happy
> > > with the
> > > code and there are no failing regtests. ...and of course it is one of
> > > the steps
> > > towards a release. ;-)
>
> > > cheers,
> > >     axel.