Refactoring of the calibration and container module #78
Conversation
-update SPE fit parameters initialization following updates of the ctapipe extractors -improve the limits of generated charge histogram plots
and ChargesContainer following events_id
-container module for the data structure (to be extend) -makers module to make the computation of calibration quantities on the data structure
with n and p' fixed
at nominal voltage + update of user script
pre-computed one
the fitting process continues although the initial parameters computation failed
cleaning unit test for WaveformsMaker and ChargesMaker
Codecov ReportAttention:
Additional details and impacted files@@ Coverage Diff @@
## master #78 +/- ##
===========================================
+ Coverage 3.08% 33.18% +30.10%
===========================================
Files 36 47 +11
Lines 3571 3128 -443
===========================================
+ Hits 110 1038 +928
+ Misses 3461 2090 -1371
☔ View full report in Codecov by Sentry. |
Tool and Components framework
guillaumegrolleron
force-pushed
the
new_structure
branch
from
3683819
to
272b0c5
Compare
-> now we will use ctapipe HDF5TableWriter and Reader -creation of recursive containers
makers with the ctapipe.Tool framework
-implementation of the NectarCAMComponent, the base class for nectrachain components, - implementation of the ChargesComponent and WaveformsComponent
or waveforms using Tool and Component classes
|
Thanks a lot, @guillaumegrolleron , for all the work. |
|
While linting the code introduced in this PR via #74 , it seems some parts of @guillaumegrolleron 's refactoring of the code were not yet included in this PR. I am very sorry for this, it seems we merged this PR a bit too fast. Pinging @guillaumegrolleron and @vmarandon . |
* Replace notebooks by percent python scripts converted with jupytext * Clean examples directory * Lint code * Lint code * Lint code * Lint code * Lint DQM * Fix bug * Further linting after merging PR #78 * Disable some unit tests * Disable some unit tests * Try again, disable some unit tests * Try again, disable some unit tests --------- Co-authored-by: jlenain <[email protected]>
I have been working last couple of days on the refactoring of the calibration code. As discussed with @jlenain and @vmarandon, we would like to have one module (
makers) to make computations on raw data and give the calibrations coefficient (or whatever). Then the other modules should use themakersand all the computation part would be in the same place.In this PR, I tried to achieve that goal. The container module has been completely rebuilt. Now the
ChargesContainersandWaveformsContainersare only made to store data. This classes inherit from thectapipe.container. The computation part which was previously in thecontainermodule has been moved in themakersmodule into two classes, theWaveformsMakerwhich is basically looping through anEventSourceand stacking events informations and waveforms into aWaveformsContainers. TheChargesMakerdo the same but extracting the charge instead of the waveforms.The calibration part is now into the
makersmodule. The SPE fit classes have been rebuilt to speed up the process.I also started to write unit tests, but the coverage is not complete.