search

ehlimana / FOF_ILC_comparison

(will be added)
0 stars 0 forks source link

naming of statistical components #37

Open BeSeeTek opened 3 years ago

BeSeeTek commented 3 years ago

Await saschas' mail and change naming of pure statistical uncertainty components

BeSeeTek commented 2 years ago

i will have an talk withg kathy clauenberg about this topic today in the evening.

ehlimana commented 2 years ago

If statistically calculcated, then we call it Type A measurement uncertainty according to GUM . Type B covers other influences which might be known experimentally or from experience. So maybe, we can refer just to Type A Uncertainty, see: https://www.bipm.org/documents/20126/2071204/JCGM_100_2008_E.pdf/cb0ef43f-baa5-11cf-3f85-4dcd86f77bd6

4.2 Type A evaluation of standard uncertainty

BeSeeTek commented 2 years ago

thank you for linking the document. See the issue in the paper for detail informations

BeSeeTek commented 2 years ago

thank you for linking the document

we have an unclear situation with in fact at least tfour different "uncertainties types " here.

  1. In single Experiments for the components -->2* roots of variances for one sine approximation (components of the phase and magnitude) called "s" as type a GUM if i got it right
  2. With this components i did gaussian error propagation to calculate the uncertainty saven in the RAWTANSFERFUNCTION in the hdffile--> so this is no longer a root of variances what is called "s" but it isn't the uncertainty as well
  3. With this values you calculated with weighted means with an different (way smaller variance) this values and variances are used for the EN calculations right?--> we should avoid to call this "uncertainty" since like the En values show this does not cover al influences
  4. In the end we will have a look at our cmc's to estimate the actual measurement uncertainty

This question is regarded to make clear this differences in the 4 different situations to avoid confusion

ehlimana commented 2 years ago

We have referred to those as standard uncertainties containing only Type A information