Re: [h5md-user] Finalization of units module

[Felix Höfling]

Am 19.12.2013, 14:57 Uhr, schrieb Pierre de Buyl
<address@hidden>:

> On Thu, Dec 19, 2013 at 04:03:08AM +0100, Peter Colberg wrote:
> > Since there is considerable interest in publishing a finalized version
> > of the units module along with the first release of H5MD, I familiarized
> > with the udunits2 library, and tested the specification of the units
> > module.
> >
> > Please see the test results in examples/h5md.lua of Lua Units [1]:
> >
> > http://git.colberg.org/lua-units/
> >
> > The grammar is fine, i.e. it is a subset of the udunits2 grammar.
> >
> > There are a number of symbols that do not match a unit in the udunits2
> > database. There is one symbol ("h") that corresponds to a different
> > unit than in the udunits2 database ("hour").
> >
> > Some issues can be resolved by using the name of the unit (e.g.
> > "degree" instead of "deg"). Other issues can not be resolved due to
> > not having a corresponding unit in the udunits2 database. I suggest to
> > drop the latter units, since it is possible but painful to maintain a
> > custom database for the H5MD units module.
>
> From your file, I see that some units/constants are commented out.
>
> Would renaming "deg" to "degree" and removing dyne, dalton, poise, h  
> solve the
> problems?
>
> P

The collision between hour and Planck's constant is really unfortunate. I
copied the latter from Mosaic, but I'm not aware of an immediate
application in semi-classical MD simulations. So way may drop it. BTW, wth
sufficient knowledge about physical dimensions, the two symbols can not
really be confused---but this can't be expected from automated parsing as
in udunits.

Having "erg" but "dyne" out is a bit inconsistent, I would blame udunits
for this. Dalton and Poise are frequently encountered in scientific
literature, to specify polymer weights and fluid viscosities,
respectively. Anyway, for version 1.0 we may blindly follow udunits and
drop them too if it helps.

For the "degree" issue: I found "degC" in the XML file. We may equally use
the longer form or use the extended ASCII character °, both are listed in
udunits2.

Felix