muon analysis r0 to dl1 specific function #556
Conversation
Codecov Report
@@ Coverage Diff @@
## master #556 +/- ##
==========================================
- Coverage 37.46% 37.43% -0.04%
==========================================
Files 82 82
Lines 7350 7352 +2
==========================================
- Hits 2754 2752 -2
- Misses 4596 4600 +4
Continue to review full report at Codecov.
|
There was a problem hiding this comment.
Look ok to me, thanks for this clean-up.
You say "in preparation of stage 1", but if we move to using directly the stage 1 tool (with all low-level calibs in ctapipe_io_lst), the whole muon analysis will have to be inside ctapipe, I think. On the other hand, that will not be the case if we just introduce a DL1 writer "à la ctapipe" in lstchain's r0_to_dl1, which is perhaps a faster path to have ctapipe-compatible LST1 DL1 files.
|
@moralejo I have not yet come around to doing it, because it is not needed until the rest works, but see cta-observatory/ctapipe#1353 |
|
Hi, thanks @vuillaut |
Yes, it works, I get the same muon file after processing a real data file. |
| # in calculation of muon ring time (peak sample): | ||
| min_pe_for_muon_t_calc = 10. | ||
|
|
||
| # FIXME: no need to read telescope characteristics like foclen for every event! |
There was a problem hiding this comment.
We can remove this FIXME comment, it is a pretty old comment in which I misinterpreted that this would imply some re-reading from disk.
|
|
||
| bad_pixels = event.mon.tel[telescope_id].calibration.unusable_pixels[0] | ||
| # Set to 0 unreliable pixels: | ||
| image = event.dl1.tel[telescope_id].image*(~bad_pixels) |
There was a problem hiding this comment.
Is there in ctapipe already some pixel interpolator? Would be interesting to test it. Should be better in case of isolated bad pixels (perhaps a bit dangerous when a whole pixel cluster is unreliable).
There was a problem hiding this comment.
I don't think there is, but yes, I think it is necessary.
There was a problem hiding this comment.
It should be easy enough to implement approaches like a mean of correct neighbors using the neighbor matrix.
|
Can we update this? I still think it would be good to separate the large imperative code in r0_to_dl1 |
step by step cleaning of the r0_to_dl1 step
In preparation for stage1: the final goal is to use directly the stage1 but a step by step approach and some code cleaning will make things easier I think and help us understand better the current status.