Researchers have created an artificial intelligence system to analyze millions of Type Ia supernovae using imaging data alone. The approach could sharpen estimates of cosmic distances and probe the nature of dark energy. It is designed for upcoming large-scale surveys by the Vera C. Rubin Observatory.
Scientists at the Institute of Cosmos Sciences of the University of Barcelona developed the CIGaRS framework, which models supernovae, host galaxies, dust effects and cosmic expansion together. Published in Nature Astronomy, the method uses simulation-based inference and neural networks to process photometric observations.
Raúl Jiménez said the technique allows all parameters to vary simultaneously and helps identify unknown systematics. Lead author Konstantin Karchev noted that it avoids selection biases while extracting full information from large datasets.
Type Ia supernovae serve as standard candles for distance measurements. The new system achieves redshift accuracy comparable to spectroscopy without spectra, a key advantage since only a small fraction of future detections will receive spectroscopic follow-up.
The Vera C. Rubin Observatory in Chile is set to begin its decade-long survey soon and will detect vast numbers of supernovae. Researchers estimate the framework could tighten cosmological constraints by up to a factor of four compared with current methods.