search

nexusformat / definitions

Definitions of the NeXus Standard File Structure and Contents
https://manual.nexusformat.org/
Other
26 stars 56 forks source link

NXcansas - NXdata errors vs uncertainties #1185

Open woutdenolf opened 2 years ago

woutdenolf commented 2 years ago

Reported by @kif

/NXcanSAS/ENTRY/TRANSMISSION_SPECTRUM-group is an NXdata group. NXdata already handles errors on signals.

However NXcansas adds a required field and attribute to add errors on NXdata signal T: /NXcanSAS/ENTRY/TRANSMISSION_SPECTRUM/T@uncertainties-attribute /NXcanSAS/ENTRY/TRANSMISSION_SPECTRUM/Tdev-field

So NXcansas is not consistent with NXdata. I assume NXdata has changed and NXcansas did not follow.

woutdenolf commented 2 years ago

Same goes for /NXcanSAS/ENTRY/DATA-group

prjemian commented 2 years ago

@timsnow Will canSAS have a meeting at the SAS2022 conference this month? Perhaps y'all can discuss this and resolve?

prjemian commented 2 years ago

@kif: Thanks for pointing this out!

prjemian commented 2 years ago

@timsnow emailed:

Addressing initial comment of @woutdenolf , can’t the application definitions build upon the base classes, include new entries, and make them compulsory? I thought they could…

Correct.

The only thing I would immediately say about this is that the TRANSMISSION_SPECTRUM entry doesn’t play well with x-ray sources. Whilst the majority of x-ray measurements may be at a single energy, for ASAXS you would have spectral data associated with the dataset, which probably wouldn’t have uncertainties or a standard deviation value associated with the values acquired (they tend to be single point data).

True, but TRANSMISSION_SPECTRUM (describes intensity vs wavelength) was an addition from a pulsed neutron facility. When this is present, the canSAS decision was to require uncertainties to be described.

On I22 we’ve only been collecting more than one datapoint (usually 1,000) per x-ray frame for our I0 and It measurements but this is rare in the x-ray community which, in the current state of the definition, would likely preclude the x-ray community from being able to use the TRANSMISSION_SPECTRUM entry.

At this time (with the SAS2022 triennial conference concluded just last week), how best to gain consensus from the canSAS community about updating NXcanSAS with the current NXdata standard? I propose that either @woutdenolf or I (@prjemian) work up a branch and PR, then it be presented at some NeXus teleconference where canSAS attends. Can you gather a list of interested parties for that meeting?