search

VERITAS-Observatory / gammapy-tools

Repository with tools for gammapy analysis
GNU General Public License v3.0
3 stars 0 forks source link

ON counts discrepancy with ED #51

Closed samanthalwong closed 2 weeks ago

samanthalwong commented 3 months ago

For point-like validation sources, ON counts are approximately +/- 1 count/run different from ED (as far as I've seen, there's always fewer counts in gammapy). It would be nice to know which events these are and why they're being thrown out in gammapy (i.e., is it the SafeMaskMaker, is it a rounding/binning issue, etc.).

Since the DL3 files are produced in ED, the initial event list, which corresponds to the event list after gamma/hadron cuts from anasum, should be identical. However, gammapy does not track the event numbers (IDs) through analysis and only tracks the integer number of events in each spatial/energy bin.

samanthalwong commented 3 months ago

See this presentation for more details & numbers!

steob92 commented 3 months ago

On slide 9, are you using the latest DL3 files? There was a point-like vs full enclosure issue that was resolved by using the latest DL3 files.

samanthalwong commented 3 months ago

No, those are the old ones (so that particlular issue is resolved as far as I know), but the counts/significance stuff still is a problem with the updated files