Michael Punch

Hi @cosimoNigro , Couldn't the underlying data product table with just the $\tau$ as a function of both redshift and energy be used directly? [Optical depths from the model as...

Hi Christoph, Another suggestion I made is to store Edisp rather as `yy = e_true / e_reco`. - For the general case `e_true ~= e_reco`, so `yy = e_true /...

Okay, but does there need to be a long discussion on this. For me it's clear that `y = e_reco / e_true` is causing information loss, and seems evident that...

Another "data point", as you say, is this plot from CAT (http://arXiv.org/abs/astro-ph/9804133v1):  It is important that the flattening at the lower end is kept, for the definition of spectrometric...

Hi @jknodlseder , Information is lost because of clipping. At the threshold, only showers with upward fluctuations trigger the detector. These showers are reconstructed to be **at** the threshold energy,...

Well, either to invert (from `e_reco/e_true` to `e_true/e_reco`), or to store as log... or both! Maybe `log(e_reco/e_true)` is simplest to implement right now? Whatever works...