Responsabile: Mario Pasquato

Descrizione: At the lowest level, the X-ray observations collected by XMM-Newton and similar telescopes, can be represented as photon lists. Each photon has individual properties, such as energy, arrival time, angle, etc. While currently most of scientific analysis is based on aggregating photons, e.g. into light curves or images, a purely unbinned approach is the most promising in the long term. To this end photons can be treated as points in a suitable N-dimensional space, where each dimension is one of their recorded properties. In this thesis you will apply machine learning methods suitable for point cloud analysis to photons with the goal of performing scientifically valuable tasks such as source-background separation and classification. You will have the opportunity to learn state-of-the art techniques for analyzing point cloud data, which also have applications besides photons throughout astronomy and beyond, and to be put in touch with one of the inventors of the Deep Sets algorithm (Prof. Ravanbakhsh from McGill University).

Durata: 9-12 mesi

Laurea: Magistrale

Prerequisiti: Nessuno