Astronomers Identify Extremely Bright Quasar Lenses in New Findings

An international team of scientists has turned machine learning loose on one of astronomy’s most elusive targets: rare quasars that behave as strong gravitational lenses. In a new study, researchers report seven fresh candidate systems identified from the Dark Energy Spectroscopic Instrument (DESI) survey—doubling the number of known quasar lenses found by earlier efforts.

Quasars are intensely bright galactic cores powered by supermassive black holes. Their luminosity can outshine the galaxy hosting them, making it difficult to measure the surrounding structure precisely. Gravitational lensing offers a workaround: the quasar’s mass bends and magnifies light from nearby sources, producing telltale distortions that reveal information about both the lens and the environment around it.

Because strong lensing by quasars is uncommon, the team faced a key problem—there are not enough real examples to train a traditional detection pipeline. Instead, they built a training strategy using a mixture of genuine quasar spectra and background galaxy spectra, then generated mock lens systems. A neural network learned the subtle spectral signatures that distinguish “normal” quasars from those with lensing-related features.

The researchers analyzed a catalog of 800,000 quasar candidates from DESI DR1. After applying the model, the list was narrowed to 200 objects, which were then hand-reviewed before selecting seven final candidates. The systems lie at least 5–6 billion light-years from Earth, meaning the observations probe a distant cosmic epoch while also helping astronomers refine models of quasar and galaxy growth.

The work underscores a broader scientific motivation: quasars may represent “missing links” in the early universe, connecting black hole formation to the evolution of galaxies. By studying the correlation between galaxies and their central black holes, researchers hope to better understand why galaxies—and the black holes inside them—develop along particular pathways, including why some black holes appear dormant.

Future confirmation will rely on powerful space-based observatories such as the Hubble Space Telescope. Once deeper follow-up validates these candidates, the same AI framework could be extended to hunt for other rare spectral anomalies across massive survey datasets.

The study was published July 22 in The Astrophysical Journal, and it was supported by the U.S. Department of Energy and the EU Horizon 2020 program—an early sign that “viral” breakthroughs in astronomy may increasingly come from machine learning applied to big data.

Keywords

Quasar lenses, gravitational lensing, machine learning, neural networks, DESI, strong lenses, supermassive black holes, spectral analysis, early universe