Gammapy

Gammapy is an open-source Python package for gamma-ray astronomy built on Numpy and Astropy. It can be found at gammpy.org. It starts from High Level data preferentially, but not restricted to, in the open DL3 format. This format describes collections of reconstructed particles plus the Instrument Response Functions necessary to derive physical quantities from them. Gammapy produces higher level, publication-ready products such as sky maps and spectra

Implementation in ESFRI projects

Gammapy is a candidate to the science tools for CTA. It has been used in its last CTA Data Challenge to gauge the capacities of the instrument indifferent fields. It can also be used to analyze high level data from existing gamma ray observatories as H.E.S.S., MAGIC, FERMI and FACT. Most importantly, it allows to combine data from different observatories in a single analysis.

Advantages

Gammapy is open source, and therefore naturally adapts itself to the needs of the community. Although it is adapted to gamma ray astronomy, it could be extended to other particle-based observatories, as many of the functionalities required are common.

The latest contribution on gammapy can be found in:

• Gammapy - A prototype for the CTA science tools, Proc. 35th ICRC, Busan, South Korea, PoS(ICRC2017)766 arXiv:1709.01751
• [open-dl3] Data formats for gamma-ray astronomy: https://github.com/open-gamma-ray-astro/gamma-astro-data-formats

Another activity concerns the deployment of parallel filesystems on storage bricks created with platforms powered by low power SoCs (ARMv7, ARMv8 and x86). Several CPU have been considered and tested, such as the Intel N3700 (Braswell), the Intel Xeon-D, the NVIDIA Tegra x1 and as soon as they will become available the next generation Intel low power SoCs, namely Apollo Lake.

So far, the most promising are the Xeon-D. We installed a test based on XeonD-1540 and on the BeeGFS filesystem developed by the Fraunhofer (Fraunhofer-Gesellschaft application-oriented research organization engaged in several fields)

Concerning data movements, INFN-CNAF production services are based on StoRM (http://italiangrid.github.io/storm/index.html) that implements the SRM protocol and on gridftp servers to realize the actual data transfers. To simplify the usage of these services and to reduce the learning curve for new users CNAF developed a data transfer application called dataclient which is currently under performance evaluation. It is a wrapper around the gridftp clients that hides the complexity of the auth/authZ infrastructure based on X509 proxy certificates.  We are benchmarking in different testbeds as well as production environments such as the computing models for the Km3, Cuore and Cupid experiments. It has proved to be scalable for a throughput of a few TB per day.

Contact persons & website

Contact : Jose Luis Contreras, UCM      

Website : https://gammapy.org