ChangelogΒΆ
0.5.0 (TBD)ΒΆ
β¦
0.4.0 (Feb 19th, 2021)ΒΆ
SummaryΒΆ
This release brings many improvements of which the most relevant are summarised here depending on their scope within the pipeline workflow.
Performance-wise, protopipe
caught up with the EventDisplay
and CTAMARS
historical pipelines starting from about 500 GeV onwards.
Below this threshold, even if compatible with requirements, the sensitivity
diverges. The cause seems to be a low-energy effect delimited to
the steps before model training.
All pipeline
upgrade to the API of
ctapipe 0.9.1
documentation also on
readthedocs
and link toZenodo
Continuous Integration is now performed on
GitHub
New benchmarks have been added
Reference analysis and benchmarks results have been updated
Data training
calibration benchmarks need only
ctapipe-stage1-process
write_dl1
has becomedata_training
DL1 parameters and (optionally) images are merged in a single file
DL1 parameters names as in
ctapipe
and they are in degrees (TelescopeFrame
)scale correction with the effective focal length
fixed bugs and wrong behaviors
Modeling and DL2 production
fixed bugs and wrong behaviors
Added missing features to get closer to
CTAMARS
DL3
ContributorsΒΆ
Michele Peresano (@HealthyPear)
Gaia Verna (@gaia-verna)
Alice Donini (@adonini)
Changes from previous releaseΒΆ
Pull-requests that contain changes belonging to multiple classes are repeated.
π General featuresΒΆ
Performance using Pyirf (#83) @gaia-verna & @adonini
Towards using Pyirf (#79) @gaia-verna & @adonini
Upgrade of DL2 production (#77) @HealthyPear
Upgrade calibration benchmarks (#59) @HealthyPear
Upgrade of data training (#58) @HealthyPear
π Bug FixesΒΆ
Fix calibration benchmarking settings (#100) @HealthyPear
Fix plot of simulated signal and noise of 2nd pass image extraction (#99) @HealthyPear
Upgrade of DL2 production (#77) @HealthyPear
Upgrade of data training (#58) @HealthyPear
π§° MaintenanceΒΆ
Use mamba to create virtual enviroment for the CI (#101) @HealthyPear
Upgrade all other notebooks and their docs version (#76) @HealthyPear
Upgrade calibration benchmarks (#59) @HealthyPear
Upgrade of data training (#58) @HealthyPear
Enable CI from GitHub actions (#84) @HealthyPear
Update documentation + general maintenance (#62) @HealthyPear
0.3.0 (Nov 9th, 2020)ΒΆ
SummaryΒΆ
early improvements related to the DL1 comparison against the CTAMARS pipeline
improvements to basic maintenance
a more consistent approach for full-scale analyses
bug fixes
ContributorsΒΆ
Michele Peresano (@HealthyPear)
Thierry Stolarczyk (@tstolarczyk)
Gaia Verna (@gaia-verna)
Karl Kosack (@kosack)
Thomas Vuillaume (@vuillaut)
Changes from previous releaseΒΆ
π General featuresΒΆ
Add missing variables in write_dl2 (#66) @HealthyPear
Add missing dl1 parameters (#41) @HealthyPear
Updates on notebooks (#47) @HealthyPear
New plots for calibration benchmarking (#43) @HealthyPear
Double-pass image extractor (#48) @HealthyPear
Notebooks for low-level benchmarking (#42) @HealthyPear
Improved handling of sites, arrays and cameras for all Prod3b simtel productions (#33) @HealthyPear
Change gain selection (#35) @HealthyPear
Changes for adding Cameras beyond LSTCam and NectarCam (#29) @tstolarczyk
π GRID supportΒΆ
Update configuration files (#74) @HealthyPear
Update documentation for GRID support (#54) @HealthyPear
Rollback for GRID support (#52) @HealthyPear
π Bug FixesΒΆ
Bugfix in Release Drafter workflow file (#71) @HealthyPear
Convert pointing values to float64 at reading time (#68) @HealthyPear
Rollback for GRID support (#52) @HealthyPear
Fix recording of DL1 image and record reconstruction cleaning mask (#46) @gaia-verna
consistent definition of angular separation to the source with config (#39) @vuillaut
Update write_dl1.py (#30) @tstolarczyk
π§° MaintenanceΒΆ
Update benchmarks and documentation (#75) @HealthyPear
Bugfix in Release Drafter workflow file (#71) @HealthyPear
Add release drafter (#67) @HealthyPear
Add benchmark notebooks for medium and late stages (#55) @HealthyPear
Update documentation for GRID support (#54) @HealthyPear
Updated documentation (#50) @HealthyPear
Implementation of a first unit test (DL1) (#34) @HealthyPear
Updated documentation (Closes #23) (#32) @HealthyPear
Added Travis CI configuration file (#18) @HealthyPear
Update README.md (#28) @tstolarczyk
Added versioning to init.py and setup.py using the manual approach. (#20) @HealthyPear
Update README.md (#21) @tstolarczyk
0.2.1 (Oct 28th, 2019)ΒΆ
SummaryΒΆ
Released Oct 28, 2019
1 contributor
1 pull requests
Description
The ctapipe-based cleaning algorithm for the biggest cluster was crashing in case of cleaned images with no surviving pixel clusters.
Contributors:
In alphabetical order by first name:
Michele Peresano
0.2.0 (Oct 24th, 2019)ΒΆ
SummaryΒΆ
Released Oct 24, 2019
3 contributor(s)
7 pull requests
Description
protopipe 0.2 now fully supports the stable release of ctapipe 0.7.0.
The main improvements involve the calibration process (high gain selected by default), the direction reconstruction and new camera-type labels.
Code based on pywi/pywi-cta libraries, relevant for wavelet-based image cleaning, has been removed in favor of ctapipe or made completely optional where needed. Wavelet cleaning is still optional but will need those two libraries to be additionally installed. Tailcut-based cleaning is now faster.
The README has been improved with installation, basic use, and developer instructions.
Dependencies are listed in protopipe_environment.yaml
and have been simplified.
The auxiliary scripts merge_tables.py
and merge.sh
have been added to allow merging of DL1 and DL2 HDF5 tables.
The mars_cleaning_1st_pass
method is now imported from _ctapipe_.
Novel code using the largest cluster of survived pixels
(number_of_islands
and largest_island
methods in the
event_preparer
module) has been hardcoded in _protopipe_ and will
disappear with the next release of _ctapipe_.
Model estimators now load the camera types directly from the analysis .yaml
configuration file.
Contributors:
In alphabetical order by first name:
Alice Donini
Michele Peresano
Thierry Stolarczyk
0.1.1 (Oct 1st, 2019)ΒΆ
SummaryΒΆ
Released Oct 1, 2019
X contributor(s)
X pull request(s)
Description
The write_dl1
and write_dl2
tools can now save an additional file
through the flag --save-images
when applied to a single run.
This file will contain the original and calibrated (after gain selection)
photoelectron images per event.
A new method save_fig
has been introduced in the utils
module,
so that model_diagnostic
can save images also in PNG format.
Additional docstrings and PEP8 formatting have been added throughout the code.
Contributors:
In alphabetical order by first name:
β¦
Pull RequestsΒΆ
The development of protopipe on GitHub started out directly in the master branch, so there are no pull request we can list here.
0.1.0 (Sep 23th, 2019)ΒΆ
SummaryΒΆ
Released Sep 23, 2019
6 contributor(s)
1 pull request(s)
Description
First version of protopipe to be publicly release on GitHub. This version is based on ctapipe 0.6.2 (conda package stable version). Its performance has been shown in a presentation at the CTAC meeting in Lugano 2019.
Contributors:
In alphabetical order by first name:
David Landriu
Julien Lefacheur
Karl Kosack
Michele Peresano
Thomas Vuillaume
Tino Michael