Cosmic Ray composition at IceCube Observatory, using Graph Neural Network

IceCube Observatory, located within the Antarctic ice at South Pole, is a unique three-dimensional multi-component detector. Understanding the composition of cosmic-rays at IceCube Observatory holds the potential to shed light into their origin and acceleration mechanism, in the transition region between their galactic and extragalactic origin. The work details an integrated analysis utilizing both IceCube and its surface array (IceTop) to estimate elemental composition of cosmic-rays. The work developed a unifying approach of using physics observables with an air-shower physics inspired Graph Neural Network utilising the footprint of air-showers at the detector. It ensures a data-efficient estimate of cosmic-ray composition while benefiting from high as well as low-level information. Possible physics-implications of the results, consistency-checks and potential of adapting the work to the next-generation instrument called IceCube-Gen2 will also be discussed.