In 2014, a team of scientists amazed the world with a simulation of the universe from its birth to the present. Having first confirmed that the cosmological model actually leads to the galaxy distribution that we see in space, the project went on to yield numerous discoveries — for instance about the properties of galaxies and the impact of supermassive black holes on cosmic structures. Still today, the project calculated on PRACE supercomputers inspires ever more new scientific approaches for investigating the origin of our universe.
It was the most detailed simulation of the universe to date, starting 12 million years after the big bang and tracing 13 billion years of cosmic evolution: Illustris. Aptly named, it illustrated the formation of the universe in a gargantuan cube with a side length of 350 million lightyears — from the generation of the first hydrogen and helium atomic nuclei to the aggregation of these gases and the creation of the first supermassive black holes, galaxies and stars that form the cosmic web that we see today, with interlaced gas filaments, clusters of galaxies and vast voids in between. In all, the simulation tracked the birth and evolution of more than 41 000 galaxies.
Illustris was developed by an international group of scientists, headed by Volker Springel at the Max Planck Institute for Astrophysics in Germany, using PRACE supercomputing resources. “For this to work, we needed the fastest supercomputers on earth”, says Springel. The simulation was built with more than 100 000 lines of program code. When it was published in 2014, it created ripples not only through the scientific world, but also beyond it. For instance, the German postal service even had an Illustris stamp made to honour the outstanding achievement.
Previous simulations of the universe had either been limited in resolution or forced to focus on only a small portion of the universe. In contrast, Illustris was the first simulation able to reproduce the universe on both large and small scales simultaneously. Most importantly, it generated a mixed population of spiral, elliptical and irregular galaxies as we see them in our universe – something previous simulations had failed to do because of numerical inaccuracies and incomplete physical models. Illustris also accurately simulated certain characteristics of galaxies, such as their rotation speed or their metal and hydrogen content.
Which one is the real universe, which one the simulation? This image is divided in the middle: the left half shows an image from the Hubble Space Telescope, the right half is a mock image from the universe as simulated by Illustris.
Image author: Illustris Simulation Team
The cosmological model confirmed
To develop the simulation, the scientists used hydrodynamic calculations, meaning models that describe the flow of fluids or gases, similar to methods used for the design of airplanes or wind turbines. However, for the simulation of the universe, the calculations had to deal with supersonic hydrogen and helium gas flows under the relentless pull of gravity, and with turbulences covering a vast range of length scales. The modelling of the physical effects affecting the baryonic particles — among them are the protons and neutrons which make up most of the mass of the matter in the universe — was especially refined compared to earlier simulations. “This is required to accurately model gas, stars and supermassive black holes, as well as their related energetic feedback”, explains Springel.
The accuracy of Illustris seemed to confirm a long-debated theory. The whole theoretical framework utilised for the simulation is based on the Lambda-CDM cosmological model, often referred to as the standard model of cosmology, including its postulated existence of dark matter. The fact that the outcome of the simulation is so similar to what astronomers actually observe in reality confirms the viability of Lambda CDM in an obvious and convincing manner.
Apart from this significant confirmation, the simulation led to numerous other new findings as well. “Within Illustris we can observe the history of tens of thousands of galaxies and thereby learn a lot about their evolution”, explains Springel, for instance concerning the role of supermassive black holes in galaxies. Such black holes are the main engine of quasars, which constitute the active centres of each galaxy. The quasars send out large quantities of energy, and it was previously unknown what processes this energy triggers and where it ends up. But thanks to Illustris, it became clear that the energy from the quasars has a tremendous effect on their host galaxies by strongly influencing their star formation. If the quasar energy bursts become too strong, star formation breaks down altogether. “In the simulation, we can uncover the specific conditions that trigger this development in a galaxy”, says Springel, “like another galaxy passing by very closely, prompting additional gas inflow to fuel the quasar.” Such observations from the simulation help scientists to develop concepts to track down such occurrences in the real universe.
A small sample of galaxies created by the Illustris simulation demonstrate the variety of formations.
Image author: Illustris Simulation Team
In addition, Illustris is public and therefore available to any scientists to address their wide-ranging research questions. Apart from the findings of the team that developed it, the simulation has further resulted in more than 200 additional scientific publications by scientists who took advantage of its public status. One example is an application to data collected by the Gaia satellite. This ESA probe is currently mapping our galaxy, the Milky Way, in unprecedented detail, including the age and the composition of stars and star debris. “Here, Illustris can be used to enrich the knowledge, since it reproduces the history of thousands of galaxies that are very similar to the Milky Way”, explains Springel. By comparing the observed data from Gaia with simulated galaxies of the same kind in Illustris, scientists get valuable clues for recounting the history of our home galaxy.
In a similar way, Illustris helps to expand the knowledge on extreme events, such as a galaxy absorbing a whole other galaxy — like the Milky Way did around 10 billion years ago, as data from Gaia demonstrated. Again, Illustris with its thousands of similar galaxies can be used to determine the likelihood and frequency of such an occurrence. The simulation thereby helps to validate the theoretical framework aiming to explain the observation.
A cubic region within Illustris with a side length of 35 million lightyears shows the evolution of matter and dark matter. The left image displays the dark matter density field, and the right image shows the temperature of the hydrogen and helium gas in the universe.
Image author: Illustris Simulation Team
The next generation
Recently, Illustris was taken one step further under the new project name IllustrisTNG. This sequel project is even more ambitious, consisting of several different universe simulations that vary in size, resolution, and complexity of the physics included. Among them is a smaller version to Illustris named TNG50 — in a cube with a side length of 165 million light years. Because it is smaller, TNG50 can simulate the universe at even higher resolution, which allows researchers to track the morphology of galaxies in finer detail — down to star populations of several hundred solar masses, in fact. Also, hydrogen and helium gas turbulence and shock waves are much better resolved in TNG50.
In contrast, the new TNG300 is a much larger simulation in a cube with a side length of almost 1 billion light years. This version allows observation of rare events and objects with great distance between them, such as very large galaxy clusters. The new set of simulations also includes more accurate and complete physics, most importantly the effects of baryonic particles on dark matter and the modelling of magnetic fields. Together, they depict the universe more accurately than ever — bringing us one step closer to the ultimate goal of understanding the origin of our universe and revealing its mysterious life story.
PRACE project title: LUCIDUS – Cosmological simulations of galaxy formation on a moving mesh
Research field: Universe Sciences
- PRACE Project Access – Call 4: 20 000 000 core hours on SuperMUC hosted by GCS at LRZ, Germany
- Preparatory Access: 200 000 core hours on Curie (thin nodes) hosted by GENCI at CEA, France
- Preparatory Access: 50 000 core hours on Hermit hosted by GCS at HLRS, Germany
See also our press release (2014) about this project.