Artificial Intelligence Unravels the Mysteries of Cosmic Radio Bursts ⚡ экспресс
To boost the completeness of burst samples for repeating fast radio bursts (FRBs), an end-to-end machine learning model, EDEN, was created. It needs no dedispersion and operates directly on dynamic spectra. The model was applied to archival FAST telescope observations (L-band) of source FRB 20121102A during an extreme activity phase. The result is the largest catalog to date: 5927 individual bursts, tripling the previous count. The markedly improved completeness enabled a refined analysis of how the burst energy distribution evolves over time. It was shown that the bimodal energy distribution stays stable with time, indicating it is an inherent characteristic of the emission mechanism, rather than a consequence of co-evolution with the burst rate.
Fast radio bursts are millisecond-long cosmic signals. Previously, searching for them was like blindly tuning a radio dial. The EDEN neural network sees the entire frequency spread and recognizes the burst pattern like a musician reads notes. On source FRB 20121102A, it found 5,927 bursts—three times more than before.
Analysis revealed two clear energy classes: weak and powerful. This constancy points to an innate mechanism, likely in a neutron star with a strong magnetic field. Jocelyn Bell Burnell once discovered pulsars by poring over paper tape; EDEN discerned a hidden rhythm in radio data.
Each burst, in a millisecond, releases energy equal to the Sun's output over hundreds of years.
🎯 A single fast radio burst, in a thousandth of a second, emits as much energy as the Sun does in hundreds of years.
🎬 Fast radio bursts were once seriously considered signals from aliens—like in the novel 'Contact'.