Astrophysicists trained a neural network on models of star cluster evolution to determine the mass of the most massive black hole that can form in a cluster based on its mass and size. Analysis of nearby globular clusters showed that the chance of finding a black hole heavier than 100 solar masses is extremely small—about 2%. However, some galaxy central clusters (such as NGC 5102 and NGC 5206) exhibit far more massive candidates. When predictions diverge from observations, it hints at alternative growth mechanisms—like gas accretion, not just black hole mergers.
Black holes come in two types: tiny ones — the remnants of supernova explosions, and giants in the cores of galaxies. But where are the intermediate ones, with hundreds and thousands of solar masses? The answer may lie in star swarms, where black holes collide like balls on a billiard table and, merging, bulk up. Back in the 1930s, Subrahmanyan Chandrasekhar calculated the limit beyond which a star inevitably collapses into a hole — that's why ordinary holes weigh just a few suns.
To figure out where the sticking happens faster, scientists simulated thousands of digital twins of clusters and trained a neural network to guess the mass of the main hole from the swarm's size and weight. Applying the model to real objects, they saw: in typical globular clusters, giants aren't born — the conditions aren't right. But a few central clusters of galaxies, like NGC 5102, promise holes over a hundred suns. The real surprise: a newborn hole often gets such a recoil kick from gravitational waves that it flies out of the swarm — like a billiard ball sent into a pocket. This explains why middleweights are so rare: many simply get launched into intergalactic void. Now astronomers know where to point their telescopes.
🎯 Merged black holes often get a kick from their own radiation and zoom away at hundreds of kilometers per second — like a ball flying off the billiard table.