Astronomers Use AI to Discover Rare Quasar Lenses

• Astronomers have identified seven new quasar lenses, effectively doubling the known count of these rare cosmic phenomena.
• Researchers utilized a neural network trained on synthetic data to scan through 800,000 quasar spectra.
• These discoveries allow for the measurement of galaxy mass in environments hosting active supermassive black holes.

The Challenge of Quasar Lensing

• Quasars are the brightest objects in the universe, often creating too much glare to observe their host galaxies clearly.
• Gravitational lensing occurs when a foreground object warps the light of a more distant background galaxy.
• If a quasar host acts as a lens, its gravity distorts the background object's light, providing a way to weigh the foreground galaxy regardless of the quasar's brightness.
• These alignments are exceptionally rare, requiring precise, coincidental positioning along the line of sight.

Discovery Methodology

• Ohio State graduate student Everett McArthur and his team processed data from the Dark Energy Spectroscopic Instrument (DESI).
• Because natural examples are scarce, the team trained a neural network using "mock lenses" created by blending real quasar and background galaxy spectra.
• The AI reduced the 800,000-candidate dataset to 200 high-probability leads, which were then manually verified by researchers to confirm the seven candidates.

Scientific Implications

• The confirmed lenses are located five to six billion light-years away.
• These pairs provide a crucial testing ground for understanding how supermassive black holes and their host galaxies grow in tandem.
• Future verification, likely using telescopes like Hubble, will be necessary to confirm the findings.
• The success of this AI-based approach demonstrates a scalable method for uncovering rare cosmic anomalies in increasingly large astronomical surveys.