gammapy / gammapy-data Goto Github PK

This lists some proposals for cleanup of gammapy data. There are some confusing/unreadable files that might be better to remove.

1. estimators/pks2155_hess_lc/pks2155_hess_lc.fits updated to have (at least) 2 bins in energy
2. Background files in tests/irf/ lack energy units. leads to UnitConversionError. There is in fact a TODO in /gammapy/makers/tests/test_map.py:520 to create a better example file.
3. Remove model.xml in tests/models/light_curve
4. Remove xml files in tests/models. Also remove the gc fits files in same location. These files are not used, and relevant files already exist
5. Inconsistency between column & header names in 1es0229_hess_spectrum tests/spectrum/flux_points. The README says these files were downloaded from the GADF, but the referenced GADF link does not exist any more. Remove files.
6. The stat_scan values in tests/spectrum/flux_points/binlike.fits are not compatible with the results. Remove? This is used in the docstring for FluxPoints.recompute_ul but can be replaced by the examples in estimators
7. The data available in tests/unbundled/tev_spectra can be useful. The are presently unused, but it may be useful to document this somewhere (in gammapy.astro maybe). Or, we can remove it...

This is separate from the issue of the catalogs we can provide (ie, all fermi catalogs or only the latest one, etc)

If this sounds okay, I can do the cleanup. Opinions @registerrier @QRemy @bkhelifi @MRegeard @Astro-Kirsty @adonath

Hi all,

Following the issue #38 and based on the latest discussions, I would like to make a proposal of licence for the sub-folders with `Gammapy' data only. For the other sub-folders, I will ask to the producer of data what they would like. I have several questions:

1. The proposal here concerns the subfolders: estimators, fermi-3fgl-crab (as these data are MapDataset), figures, gravitational_waves, joint-crab/spectra, . Do you confirm?

2. I have an issue with fermi-3fgl-gc, because it contains both Fermi data and Gammapy data. What do you? Ideally, one should 1/ create 2 sub-folders replacing this one or 2/ include a folder into it with only the Fermi-LAT data (then I can place the correct licence - and modifying the make.py). The folder fermi_3fhl will follow the same logic. @QRemy, what do you think?

3. I propose the licence CC BY-NC 4.0. Do you have any other objection or any other proposal?

4. Technically, I propose this implementation, and not only for the gammapy data, for all the gamma-data repository. In the concerned folder, I will

• add the licence term in full text into a file `LICENCE' (if the chosen licence is the one proposed just before for our data, the text will be https://creativecommons.org/licenses/by-nc/4.0/legalcode.txt)
• add in the readme file a licence section that will contain the summary information of the licence as given by this Common-Creative page. The needed information are these ones:
  
  And I propose respectively as entries: Gammapy data, Gammapy Coordination Committee, https://gammapy.org/, None, ??? No idea what to put as date... I might need to search for the history of each sub-folder. Do you agree with that?
  The output will look like:
  `Gammapy data © 2015 by Gammapy Coordination Committee is licensed under CC BY-NC 4.0'
  I will add a link towards the file containing the licence term

I ask for your feedback and after converging I will make the PR.

For the related documentation.

This repo is currently without licence, which is a bad practice.

This repo contains open data from experiments and we have copy them to our tests. One could add a licence that is a bit restrictive, because we are not the owners or the productors... Maybe the CC BY-NC-ND is the appropriate one (see https://howtofair.dk/how-to-fair/data-licences/#cc-licences or https://en.unesco.org/open-access/terms-use-ccbyncnd-en)

File GAMMAPY_DATA/magic/rad_max/data/20131004_05029747_DL3_CrabNebula-W0.40+035.fits contains the OBS_MODE keyword twice, once with POINTING, once with Wobble content resulting in astropy giving a list of values when accessing it via Table.meta

I wanted to make a quick snippet to share with the devs to illustrate a possible problem I found with LC estimation.

I realised that the only OGIP files that we have here are those in the joint-crab/spectra subdirectory.
These work fine for spectra estimation, but don't work with the LightCurveEstimator as they don't have a GTI HDU, as other OGIP files produced with more recent versions of Gammapy.

Should we add newer versions of OGIP files for these types of tests?