What commonalities do we have?
Efficiencies and background subtractions
- Store the background subtraction and scaling in the output
- Background subtraction is a reduction step and should be stored in a reproducible fashion
- It is equally as important to store what has not been done to the data
- Don’t try and deconvolve resolution
- Store corrections!
- Masking, noise and monitors
- No defined q points should be expected
- The bin width should be inclded in the resolution
- Log binning might be a nice general agreement
- Neutron event mode helps with summing (although ill defined)
- Everything in metadata!!
- Everything should be defined as a value or an array
- The slit positions (and motor uncertainties) should be stored
Documentation: undocumented methods are useless
- a resource to store this in an agreed way
- link to publications (DOI)
- include docs in files
- a web resource could help users fix problems themselves (possibly including YouTube videos)
- in particular if reduction method and software is available, however specific metadata needs to be stored
Do the corrections if you can
- with error propagation
- Would a data reduction round robin be interesting?
- correct for “strange” experimental results if you can
Live data reduction is important for a “quick look”
What is reduction and what is analysis?
- These are reduction: