AI Helps Astronomers Match X-Ray Sources to Visible Stars

By Space.Fan Editorial DeskSeptember 22, 2026 at 8:03 AM UTC

Researchers used a smart computer program to link over 100,000 mysterious X-ray objects with their partners in the visible night sky.

Image · Space.Fan

The Drop

Astronomers often look at the same part of the sky using different space telescopes. One telescope might see a bright X-ray source, while another sees a that glows with visible light. Matching these two images is hard because the telescopes see the sky in different ways. In the past, scientists mainly looked at where objects were located, but this can cause mistakes when objects are crowded together. To solve this, a team of researchers built a new artificial intelligence tool. This program looks beyond just location. It compares the colors, brightness, and distances of s found by the Gaia mission with thousands of objects detected by the Chandra X-ray Observatory. By training the computer to recognize patterns, it can correctly identify which s belong to the X-ray sources. The team successfully linked about 113,000 X-ray sources to their true partners. The program even identified about 7,000 cases where one X-ray source might be linked to multiple s, helping researchers decide which match is the most likely. By testing the tool on data from the Orion constellation, the scientists proved that their AI is very reliable, getting 95% of the matches right without needing to know the exact location of the objects.

Why It Matters

This new tool is a huge time-saver. Before this, astronomers had to check these matches by hand, which is slow and prone to human error. With over 100,000 confirmed matches, scientists now have a much clearer map of high-energy objects in our galaxy. This list helps researchers study how s grow, how they die, and how they behave throughout their lives. By knowing exactly which s are producing X-rays, scientists can better understand the physics behind powerful space events, such as s feeding or s releasing massive bursts of energy.

The Catch

While the AI is very good, it is not perfect. The tool works based on the information it was trained on. If a is very far away or too dim for the telescopes to see clearly, the AI might still make mistakes. There are also about 20,000 objects from the original list that could not be matched, and it is possible that some of these are just random alignments rather than true physical pairs.

Put That in Perspective

Scientists have spent years trying to organize the data from the Chandra and Gaia missions. This new catalog provides a solid foundation that others can use for future research. The team hopes that this method of using AI for matching can be applied to even more telescopes in the coming years.

Source September 22, 2026

V. Samuel Pérez-Díaz, Vinay L. Kashyap, Joshua D. Ingram, David Fouhey, Juan Rafael Martínez-Galarza, Pavlos Protopapas, Jeremy J. Drake, Dong-Woo Kim, Cecilia Garraffo

Center for Astrophysics Harvard & Smithsonian, Harvard John A. Paulson School of Engineering and Applied Sciences, Universidad del Rosario, NSF AI Institute for Artificial Intelligence and Fundamental Interactions, New York University, Carnegie Mellon University, New College of Florida, Lockheed Martin Solar and Astrophysics Laboratory·The Astrophysical Journal·10.3847/1538-4357/ae9f41