When computing saturated log likelihood, only include active analysis…

threeML / hawc_hal

HAWC Accelerated Likelihood - python-only framework for HAWC data analysis

BSD 3-Clause "New" or "Revised" License

11 stars 22 forks source link

When computing saturated log likelihood, only include active analysis… #18

Closed sammarinelli closed 3 years ago

sammarinelli commented 5 years ago

… bins.

Map trees normally contain data for analysis bins with very marginal statistics that are not normally used in analysis. The user probably expects the saturated likelihood computed using only the bins active in the analysis rather than all of the bins.

(Also I changed what look like a couple of typos in a doc string and error message.)

sammarinelli commented 5 years ago

Apparently the Geminga-paper test fails on Linux but not Mac. I ran the test manually on CentOS 6.8 and can confirm it fails. The same test also fails for me (on CentOS) without my changes. Is this a known issue?

henrikef commented 5 years ago

This looks reasonable to me... Should we also change the loop in _compute_likelihood_biases to only run over the active bins? https://github.com/threeML/hawc_hal/blob/3ec5e1fd8c635133cda2f5a93049967264a9c809/hawc_hal/HAL.py#L181 ?

github-actions[bot] commented 3 years ago

This PR has become stale. Please check the status