Gamma-ray images flares

A filtering and clustering system that should identify and extract the centroid of the event, the number of photons and other features must be developed. Working with both real and simulated gamma-ray sky maps is possible.

CTA

Cherenkov Telescope Array (CTA) is an initiative to build the next generation of ground-based gamma-ray astronomy made by dozens of Imaging Atmospheric Cherenkov Telescope (IACT) that provide an unprecedented sensitivity to detect transient events.

The project task concerns of an identification of short-time gamma-ray transients (from seconds to hundred seconds), analysing FITS file auto-generated by the simulation procedure given in the software package ctools. You can check major info in the user guide. ctools comprises a set of ftools-like binary executables and Python scripts with a command-line interface that allows the user to interact and perform step-by-step data analysis.

A typical FITS file generated by the simulation procedure is the one shown in the following figure.

Here indeed it is shown a pixelated and smoothed AGILE image of the sky captured by the cta instrument. A small blob can be seen the center of the image. It represents the gamma-ray event that must be detected.

A source model is defined as a collection of parameters in an XML file format. Changing the values of the RA and DEC parameters (the source coordinates) implies simulating source at a specific position within the map.

Quick tutorial

To handle astronomical units, coordinates, and angular separation one can rely on the astropy package:
Other useful library are:

• Angular separation
• Pixel to world coordinates

Additionally, the lxml package can be used to modify the XML model programmatically.

Environment configuration

• Download and install Anaconda
• After the installation install atropy:

  conda create -y -n cta_simulation python=3.7 astropy
  

• Activate your new environment

  source activate cta_simulation
  

• Install also ctools from the specified channels

  conda install -y -c conda-forge -c cta-observatory ctools
  

Notebooks

The repository contains the following notebooks

• build_dataset which has the main goal to build your dataset Please be careful: the following notebook takes a lot of time to cunstruct the all dataset, so I sudgest to you follow these two alternatives:
  • set a very low number of images to test the model
  • follow this drive link in order to access to the dataset I used to perform this task
• cta_project which contains the training of the model for classification and localization tasks