Weighted average and uncertainties propagation in pyFAI

TODO

Distinction between the std and the sem

A word about notation

Five different histograms are calculated for an integration: sum_signal sum_variance sum_normalisation sum_normalisation_sqarred sum_count

Average

sum_signal/sum_normalisation

Uncertainties propagated from known variance

Often assuming Poisson statistics

Uncertainties calculated from the variance in a ring

Note the difference with schubert’s paper where wi are not squarred

Conclusion

Point to a notebook where the demonstration is performed that average/uncertainties scale well with normalization_factor