An international team of astronomers and computer scientists from India and France have developed a new technique to search for closely merging supermassive black holes.
The team has devised a numerical technique to automatically search for double nuclei galaxies which could help them predict what could happen when neighbouring galaxies like the Milky Way and the Andromeda collide.
According to the Department of Science and Technology who have supported this study, scientists need to study collisions between galaxies to predict uncertainties like whether the solar system would survive or be ripped apart from the violent mixing of stars and gas between the galaxies when they collide.
For this, they need to hunt for closely merging galaxies. However, only a handful of such galaxies are known, as they are very rare. During the merging process, the nuclei of the individual galaxies come closer resulting in the formation of double nuclei galaxies.
In order to hunt for more merging galaxies in the open sky, an international team of Astronomers and Computer Scientists from APPCAIR, BITS Pilani, Goa campus, Indian Institute of Astrophysics (IIA), Indian Institute of Science Education and Research, Allahabad University, and the Paris Observatory, France, have developed a new algorithm which has led to the discovery of thousands of double-nuclei galaxies.
“Out of these, 159 were confirmed to have pairs of accreting supermassive black holes or active galactic nuclei (AGN) as they are usually called,” the department stated.
For this study, the team crafted a specialised image processing technique called GOTHIC that can automatically detect galaxies that visually resemble the galaxy MRK 739, which is one of the earliest detected dual AGN. GOTHIC used sophisticated image processing techniques and data from one of the largest optical surveys, the Sloan Digital Sky Survey (SDSS) to detect such nuclei pairs.
“Earlier, there were approximately only 50 known double nuclei galaxies with good images. Since this is too small a sample size for any AI model to train properly, we had to resort to old-school image processing” said Snehanshu Saha, ML expert at APPCAIR, BITS Pilani.
Next, they applied GOTHIC to a random sample of 1 million galaxy images to filter out the rare double-nuclei galaxies.
“The process was tedious and it took many months to complete since terabytes of image data had to be downloaded. But the final result was totally worth it!” said Anwesh Bhattacharya, lead author and creator of GOTHIC.
