The first complete family tree of our home galaxy

Globular clusters (GC) formed when the Milky Way underwent a rapid assembly phase. The Milky Way is home to more than 150 such groups. Most of these clusters formed in the smaller galaxies which merged to form the galaxy we live in today.

For decades, the ancient ages of globular clusters have been viewed as “fossils” to reconstruct the earliest assembly histories of galaxies. However, it was only with the latest models and observations that it became possible to fulfill this promise.

An international team of researchers led by Dr Diederik Kruijssen at the Center for Astronomy at the University of Heidelberg (ZAH) and Dr Joel Pfeffer from John Moores University in Liverpool have now succeeded in deducing the fusion of the Milky Way. They succeeded in reconstructing the first complete family tree of our original galaxy by analyzing the properties of globular clusters orbiting the Milky Way with artificial intelligence.

Scientists achieved this by using new computer simulations of the formation of Milky Way-like galaxies.

Called E-MOSAICS, the simulations include a comprehensive model for the formation, evolution and destruction of globular clusters.

In the simulations, the researchers were able to relate the ages, chemical compositions and orbital motions of globular clusters to the properties of the progenitor galaxies in which they formed more than 10 billion years ago. By applying this knowledge to groups of globular clusters in the Milky Way, they determined the number of stars contained in these progenitor galaxies. They also had an idea when they merged with the Milky Way.

Dr Diederik Kruijssen from the Center for Astronomy at the University of Heidelberg (ZAH) said: “The main challenge in relating the properties of globular clusters to the fusion history of their host galaxy has always been that galaxy assembly is an extremely complicated process, in which the orbits of globular clusters are completely altered.

“To make sense of the complex system that remains today, we therefore decided to use artificial intelligence. We trained an artificial neural network on E-MOSAICS simulations to relate the properties of the globular cluster to the fusion history of host galaxies. We have tested the algorithm tens of thousands of times on the simulations. We were amazed at how accurately he was able to reconstruct the fusion histories of simulated galaxies, using only their populations of globular clusters.

After obtaining these results, the scientists set out to determine the fusion history of the Milky Way. To do this, they used groups of globular clusters that would each have formed in the same progenitor galaxy based on their orbital motion.

By applying the neural network to these groups of globular clusters, scientists could predict with great accuracy the stellar masses and fusion times of progenitor galaxies. He also revealed a hitherto unknown collision between the Milky Way and an enigmatic galaxy, named “Kraken”.

Kruijssen said, “The collision with the Kraken must have been the most significant merger the Milky Way has ever seen. Previously, it was believed that a collision with the Gaia-Enceladus-Sausage galaxy, which took place around 9 billion years ago, was the biggest collision event. However, the merger with Kraken took place 11 billion years ago, when the Milky Way was four times less massive. As a result, the collision with Kraken must have really transformed what the Milky Way looked like at the time.

Throughout its history, the Milky Way has cannibalized around five galaxies with over 100 million stars and around fifteen with at least 10 million stars. The most massive progenitor galaxies collided with the Milky Way between 6 and 11 billion years ago.

Scientists say the study could help boost future studies looking for the remains of these progenitor galaxies.

Journal reference:

1. JM Diederik Kruijssen et al. Kraken reveals itself – the story of the fusion of the reconstructed Milky Way with the E-MOSAICS simulations, Monthly Notices of the Royal Astronomical Society (2020). DOI: 10.1093 / mnras / staa2452