[Nexus] Geometric information in NeXus

[Ray Osborn]

On 2002/11/26 11:32, "Johnson, G (Greg)" <G.Johnson at rl.ac.uk> wrote:

> Could we treat the geometric data similar to that from any other
> instrument?  Require an NXentry for it, using a NXinstrument called
> 'NXgeometry'|'NXdigitiser'|'NXsurface' (any suggestions?) and then
> store the surface itself in the NXdata element?

Hi Greg,
My instinct is that this should be stored in the NXsample group, since you
are describing properties of the sample rather than the instrument.

[As an aside, some have questioned why the NXsample group isn't also in the
NXinstrument group.  The reason is that we thought that it would be
relatively common for the instrument parameters to be the same in a number
of different NXentry groups, e.g., containing data from different samples or
measured with different sample environment parameters.  In those cases, the
whole NXinstrument group could be linked multiple times provided that it did
not also contain the NXsample group.]

It is perfectly acceptable to propose extensions to the NXsample definition
to accommodate all this extra information, although we could also consider
establishing a separate group class to be stored within the NXsample group.
This is really a matter of taste.

> But the NXdata element doesn't really seem to be the right place for
> the geometric info - at least, it would require entirely new fields
> to store them in (vertices, topology, precision, fiducial markings,
> etc).  Could NXdata be extended dramatically to incorporate spatial
> information like this?

The principle adopted since the beginning of NeXus is that the NXdata group
just contains the information required to plot the data.  This means that we
need to have some suitable plotting axes defined, but not anything else.  We
do, however, need to have a way to link data points with their respective
geometric parameters.  We would have to discuss how best to achieve that
when I know exactly what sort of information you are planning to put in.

> I've been trying to look at example files, and found Peter
> Peterson's TOF XML def'n very helpful (using XML Spy - are there
> other XML editors that are widely used?)  I can certainly craft
> something on the basis of that and submit it for substantial
> revision that would be helpful.  (Who can I burden with it???)

Well, this list is as good a place as any for the time being.  We will
shortly have an advisory committee to officially oversee this sort of thing.

Best regards,
Ray
-- 
Dr Ray Osborn                Tel: +1 (630) 252-9011
Materials Science Division   Fax: +1 (630) 252-7777
Argonne National Laboratory  E-mail: ROsborn at anl.gov
Argonne, IL 60439-4845