The soft/medium gamma-ray regime is the least astrophysically explored range across the electromagnetic spectrum. The sensitivity of current missions is orders of magnitudes worse than neighboring bands due to high instrumental and atmospheric backgrounds, low interaction cross sections and inherent difficulty of imaging at these energies. It remains an extremely interesting range harboring the positron annihilation line, signatures of stellar nucleosynthesis, emission from the most extreme environments and multi-messenger astrophysics.
The Compton Spectrometer and Imager (COSI) is a soft gamma-ray survey telescope (0.2-5 MeV) designed to probe the origins of Galactic positrons, uncover the sites of nucleosynthesis in the Galaxy, perform pioneering studies of gamma-ray polarization and find counterparts to multi-messenger sources. COSI’s compact Compton telescope combines improvements in sensitivity, spectral resolution, angular resolution, and sky coverage to facilitate groundbreaking science. Our group is strongly involved in the analysis and simulation pipeline development for COSI. In particular in the background simulation and modelling and the detector response approximation. We are also involved in the COSI science goals like dark matter search or the formation and evolution of chemical elements in our galaxy through observations of gamma-ray lines.
Background modelling
The observation of MeV gamma rays is dominated by background radiation, especially due to the activation of the detector materials induced by cosmic-ray interactions. Thus, background simulation and identification are crucial for the data analysis. Our group made a complete and very detailed simulation of 3 months of orbit in order to estimate the expected background COSI will measure. This background is used in the current COSI Data Challenge.
Response approximation
To compare our reconstructed data with scientific models, we need a precise yet lightweight characterization of the COSI detector response that can run efficiently within our extensive analysis pipeline. We address this by developing a neural-network approximation scheme in a dedicated relative coordinate space, coupling normalizing flows with a spherical harmonics expansion. Finally, we are also creating the necessary tools to support a new analysis framework around this approach.
Event Classification and Reconstruction
In order to improve the data quality we need to keep the data exposed to the reconstruction as clean as possible. To achieve this, an event classification prior to reconstruction is mandatory. Prior deep learning models developed for a different detector concept were showing promising results and will be adapted to COSI. Another angle to enhance the overall data quality is to improve the event reconstruction. Here, several Machine Learning driven approaches are to be tried and tested.
COSI and Dark matter searches
COSI is well-suited for indirect dark matter searches thanks to its excellent energy resolution in the MeV range. This makes it particularly sensitive to signals from various dark matter candidates, such as primordial black holes, which are expected to produce detectable gamma ray emissions in this energy band. Our group is studying how sensitive COSI could be to this potential sources of dark matter.
The Vela region and 26Al
Exhibiting peak sensitivity at 1.8 MeV, COSI is ideally equipped to investigate the characteristic decay radiation of the radioisotope aluminium-26. With a mean lifetime of approximately one million years, this isotope serves as a powerful tracer of ongoing nucleosynthesis (the formation of elements), shedding light on massive stars, Wolf–Rayet winds, ejection mechanisms and supernova explosions. A prime target is the Vela region, where COMPTEL first detected a pronounced excess of 1.8 MeV emission that remains unattributed to any known source. By analysing COSI’s initial observations, we aim to resolve this longstanding mystery and advance our understanding of aluminium-26 in our Galaxy.
A future MeV/GeV Compton/Pair Telescope
We are also working towards the realization of a future large Compton/pair telescope to fully cover the MeV sensitivity gap up to 100 MeV also for continuum souces, and to connect at superior angular resolution from 30 MeV to the accomplishments of the Fermi/LAT instrument in the GeV range. An example is the proposed ESA mission newASTROGAM with an energy range covering 200 keV to 3 GeV, with an added coded X-ray mask at 15-45 keV.
