Join GitHub today
Fix two bugs in LightCurveEstimator, and improve speed considerably #1179
This PR fixes two bugs:
Furthermore, this PR leads to a huge speed-up by removing an uneccesary (and slow) loop over energy bins.
See also issue #1178 for more discussion.
referenced this pull request
Oct 20, 2017
I added another small fix: previously, the observed events were binned, bins outside of
It'd be good if someone could have a look at this and confirm that things make sense.
It would be great to have a documented analysis that shows fluxes computed by Gammapy LC are correct. For HESS e.g. one could compute runwise fluxes for the 4 Crab runs from the test data release and check that integral fluxes in one or two energy bands roughly match the known spectrum. If someone does testing with CTA, please focus on the PKS 2155 runs, which were agreed on as the target for first checks (see https://forge.in2p3.fr/projects/data-challenge-1-dc-1/wiki/Comparison_of_ctools_and_Gammapy)