Astronomers have used the biggest ever supercomputer simulations from the Big Bang to the present day to discover how the Universe evolved.
The FLAMINGO (Full-hydro Large-scale structure simulations with All-sky Mapping for the Interpretation of Next Generation Observations) supercomputer simulations calculate the Universe’s evolution by looking at all its components based on the laws of physics.
As the simulations progress, galaxy clusters and virtual galaxies emerge in greater detail.
The FLMAINGO team, comprised of researchers from Durham University, UK, Leiden University, the Netherlands, and Liverpool John Moores University, UK, hope that the supercomputer simulations will let researchers compare the virtual Universe with observations of the real thing being captured with high-powered telescopes.
This could help astronomers understand the accuracy of the standard model of cosmology, which is used to outline the Universe’s evolution.
The paper, ‘FLAMINGO: Calibrating large cosmological hydrodynamical simulations with Machine Learning,’ is published in the journal Monthly Notices of the Royal Astronomical Society.
Tiny particles must be taken into account to understand the Universe’s evolution
Previous supercomputer simulations used to compare observations of the Universe have focused on dark matter. This is because it is believed to be a key component of the structure of the cosmos.
Now, astronomers believe that the effect of ordinary matter, which makes up 16% of all matter of the Universe, as well as neutrinos and tiny particles, must also be taken into account to find out how the Universe evolved.
FLAMINGO Principal Investigator Professor Joop Schaye, of Leiden University, said: “Although the dark matter dominates gravity, the contribution of ordinary matter can no longer be neglected since that contribution could be similar to the deviations between the models and the observations.”
Following the standard model of physics, the FLAMINGO supercomputer simulations tracked the formation of the Universe’s structure in dark matter, ordinary matter, and neutrinos.
Over the past two years, the team ran the simulations at a powerful supercomputer in Durham using different resolutions. They also altered other factors like the galactic winds and the mass of the neutrinos.
The results revealed that ordinary matter and neutrinos must be included in the simulations in order to make an accurate prediction of the Universe’s evolution.
Questions around our current understanding of the evolution of the Universe
New telescopes like ESA’s Euclid space telescope, and the Dark Energy Survey Instrument, are collecting huge amounts of data on stars, quasars, and galaxies. These observations are posing questions about the current theories on the Universe’s evolution.
The FLAMINGO supercomputer simulations will play a key role in interpreting the data by comparing theoretical predictions with observation data.
FLAMINGO research collaborator Professor Carlos Frenk, Ogden Professor of Fundamental Physics, in the Institute for Computational Cosmology, Durham University, said: “Cosmology is at a crossroads.
“We have amazing new data from powerful telescopes some of which do not, at first sight, conform to our theoretical expectations. Either the standard model of cosmology is flawed or there are subtle biases in the observational data.
“Our super precise simulations of the Universe should be able to tell us the answer.”
New code was developed to make the supercomputer simulations possible
The simulations took more than 50 million processor hours on the Cosmology Machine supercomputer. This is hosted by the Institute for Computational Cosmology at Durham University, on behalf of the UK’s DiRAC High-Performance Computing facility.
The researchers developed a new code called SWIFT, to make the FLAMINGO supercomputer simulations happen. The new code distributes the computational work over thousands of Central Processing Units. This can be as many as 65,000 CPUs.
The data created by the simulations can help make more theoretical discoveries
The simulations have created a large amount of virtual data that can make more theoretical discoveries and to test new data analysis techniques such as Machine Learning.
The FLAMINGO project is part of the VIRGO consortium for cosmological supercomputer simulations.
The project received funding from the European Research Council, the UK’s Science and Technology Facilities Council, the Netherlands Organization for Scientific Research and the Swiss National Science Foundation.
