An innovative artificial intelligence-based method that improves the accuracy of classifying black holes and neutron stars has been developed by a team of researchers at the University of Milano-Bicocca, led by Professor Davide Gerosa and supported by the European Research Council. The study, published in Physical Review Letters, challenges a decades-old assumption and paves the way for a more precise analysis of cosmic signals.
Gravitational-wave astronomy enables the observation of pairs of compact objects such as neutron stars and black holes. These waves are produced by the inspiral and merger of such binary systems. Traditional analyses assume a predetermined way to distinguish the two objects in each binary.
The new study overcomes this limitation by using an artificial intelligence technique called spectral clustering, which analyzes the entire dataset without imposing rigid labels beforehand. This method reduces uncertainties in the measurement of black hole spins — that is, the speed and direction of their rotation. Accurate spin measurements are crucial for understanding black hole formation and evolution. The novel approach significantly improves the precision of these measurements and enhances the reliability of distinguishing black holes from neutron stars.