. . .
0.4.0.post1 (Mar 5th, 2021)
This is a post-release that takes care of project maintenance, so it doesn't change the performance of the code.
- Michele Peresano (@HealthyPear)
Pull-requests that contain changes belonging to multiple classes are repeated.
- Fix zenodo configuration file and add LICENSE file (:pr:`106`) @HealthyPear
- Update CHANGELOG (:pr:`108`) @HealthyPear
- Fix zenodo configuration file and add LICENSE file (:pr:`106`) @HealthyPear
- Prepare first upload to PyPI (:pr:`107`) @HealthyPear
0.4.0 (Feb 22th, 2021)
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
- upgrade to the API of
- 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
- calibration benchmarks need only
- Modeling and DL2 production
- fixed bugs and wrong behaviors
- Added missing features to get closer to
CTAMARS
- DL3
- Michele Peresano (@HealthyPear)
- Gaia Verna (@gaia-verna)
- Alice Donini (@adonini)
Pull-requests that contain changes belonging to multiple classes are repeated.
- Performance using Pyirf (:pr:`83`) @gaia-verna & @adonini
- Towards using Pyirf (:pr:`79`) @gaia-verna & @adonini
- Upgrade of DL2 production (:pr:`77`) @HealthyPear
- Upgrade calibration benchmarks (:pr:`59`) @HealthyPear
- Upgrade of data training (:pr:`58`) @HealthyPear
- Fix calibration benchmarking settings (:pr:`100`) @HealthyPear
- Fix plot of simulated signal and noise of 2nd pass image extraction (:pr:`99`) @HealthyPear
- Upgrade of DL2 production (:pr:`77`) @HealthyPear
- Upgrade of data training (:pr:`58`) @HealthyPear
- Fix zenodo configuration file and add LICENSE file (:pr:`106`) @HealthyPear
- Update documentation + general maintenance (:pr:`62`) @HealthyPear
- Use mamba to create virtual enviroment for the CI (:pr:`101`) @HealthyPear
- Upgrade all other notebooks and their docs version (:pr:`76`) @HealthyPear
- Upgrade calibration benchmarks (:pr:`59`) @HealthyPear
- Upgrade of data training (:pr:`58`) @HealthyPear
- Enable CI from GitHub actions (:pr:`84`) @HealthyPear
0.3.0 (Nov 9th, 2020)
- 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
- Michele Peresano (@HealthyPear)
- Thierry Stolarczyk (@tstolarczyk)
- Gaia Verna (@gaia-verna)
- Karl Kosack (@kosack)
- Thomas Vuillaume (@vuillaut)
- Add missing variables in write_dl2 (:pr:`66`) @HealthyPear
- Add missing dl1 parameters (:pr:`41`) @HealthyPear
- Updates on notebooks (:pr:`47`) @HealthyPear
- New plots for calibration benchmarking (:pr:`43`) @HealthyPear
- Double-pass image extractor (:pr:`48`) @HealthyPear
- Notebooks for low-level benchmarking (:pr:`42`) @HealthyPear
- Improved handling of sites, arrays and cameras for all Prod3b simtel productions (:pr:`33`) @HealthyPear
- Change gain selection (:pr:`35`) @HealthyPear
- Changes for adding Cameras beyond LSTCam and NectarCam (:pr:`29`) @tstolarczyk
- Update configuration files (:pr:`74`) @HealthyPear
- Update documentation for GRID support (:pr:`54`) @HealthyPear
- Rollback for GRID support (:pr:`52`) @HealthyPear
- Bugfix in Release Drafter workflow file (:pr:`71`) @HealthyPear
- Convert pointing values to float64 at reading time (:pr:`68`) @HealthyPear
- Rollback for GRID support (:pr:`52`) @HealthyPear
- Fix recording of DL1 image and record reconstruction cleaning mask (:pr:`46`) @gaia-verna
- consistent definition of angular separation to the source with config (:pr:`39`) @vuillaut
- Update write_dl1.py (:pr:`30`) @tstolarczyk
- Update benchmarks and documentation (:pr:`75`) @HealthyPear
- Bugfix in Release Drafter workflow file (:pr:`71`) @HealthyPear
- Add release drafter (:pr:`67`) @HealthyPear
- Add benchmark notebooks for medium and late stages (:pr:`55`) @HealthyPear
- Update documentation for GRID support (:pr:`54`) @HealthyPear
- Updated documentation (:pr:`50`) @HealthyPear
- Implementation of a first unit test (DL1) (:pr:`34`) @HealthyPear
- Updated documentation (Closes #23) (:pr:`32`) @HealthyPear
- Added Travis CI configuration file (:pr:`18`) @HealthyPear
- Update README.md (:pr:`28`) @tstolarczyk
- Added versioning to init.py and setup.py using the manual approach. (:pr:`20`) @HealthyPear
- Update README.md (:pr:`21`) @tstolarczyk
0.2.1 (Oct 28th, 2019)
- 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
- (:pr:`16`) Bugfix: Closes #15 (Michele Peresano)
0.2.0 (Oct 24th, 2019)
- 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
This list is incomplete. Small improvements and bug fixes are not listed here.
The complete list is found here.
- (:pr:`9`) Update image cleaning and make wavelet-based algorithms independent
- (:pr:`8`) Import CTA-MARS 1st pass cleaning from ctapipe
0.1.1 (Oct 1st, 2019)
- 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:
- ...
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)
- 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
- (:pr:`2`) Custom arrays, example configs and aux scripts (M.Peresano)