Much remains a mystery about the first billions of years in the history of the universe, the era in which space appeared from its dark ages with the dawn of the earliest stars and galaxies. Scientists have now developed the largest and most detailed computer model from this period to help shed light on how the children’s universe evolved. Named TEZANafter the Etruscan goddess of dawn, predictions of this new project for the primordial past will soon be tested with data from NASA’s recently launched space telescope James Webb (JWST) and other next-generation observatories.

Immediately after the Big Bang, about 13.8 billion years ago, the universe was filled with cosmic fog. The heat of creation was so great that electrons could not combine with protons and neutrons to form atoms, and the space was saturated with a dense soup of plasma – electrically charged (or ionized) particles that scattered rather than transmitted light. This cosmic fog rose briefly about 380,000 years later, during the so-called recombination era, when the universe cooled enough to allow atoms to freeze from plasma like clouds of optically transparent, electrically neutral hydrogen gas. Suddenly released, the light from the afterglow of the Big Bang flashes into the universe, which then fades back into darkness because the stars have not yet formed.

Darkness reigned for the next few hundred million years, until gravity began to collect matter in stars and galaxies. Even then, the darkness dissipates only gradually, as the intense ultraviolet radiation from the first glowing objects in the universe reionizes the surrounding neutral hydrogen, eventually burning the gaseous darkness. This “era of reionization” lasted more than half a billion years, but scientists know very little about its details. What they know for sure is that its end marks the cosmic moment when light from the entire electromagnetic spectrum – instead of the ordinary part that can pierce the veil of neutral hydrogen – began to travel freely through space. Simply put, this was when the universe finally became clear for research by curious astronomers seeking to know exactly how the cosmic dawn came about.

This does not mean that such research is easy. To see light from such ancient times, researchers must use the largest and most sensitive telescopes to search for objects that are as far away as possible. This is because the greater the distance of the object, the more time it takes for its light to reach the Earth – and the weaker that light will be.

Computational cosmic dawn

Another way to get an idea of ​​this bygone era is to simulate it on computers. The early stages of reionization are relatively easy to recreate because the universe was relatively dark and homogeneous at the time, said Aaron Smith, an astrophysicist at the Massachusetts Institute of Technology who helped develop THESAN. However, as primary matter is sorted into galaxies and stars, the complex interactions between gravity, light, gas, and dust are becoming increasingly difficult to model.

“Because light modeling is quite complex and computationally expensive, there are only a few cosmological simulations that focus on the study of this era,” said astrophysicist Rahul Kanan of the Harvard-Smithsonian Center for Astrophysics, which helped develop THESAN. “Each of these cosmological simulations has its advantages and disadvantages.”

TEZAN is designed to simulate the early universe to an unprecedented degree. Some cosmological simulations, such as Cosmic dawn (CoDa) simulations and Space reionization of computers (CROC) have modeled large volumes at relatively low resolutions, while others, such as Renaissance and SPHINX simulations are more detailed, but do not cover long distances. In contrast, THESAN “combines high resolution with high simulated volumes,” says Canan.

“There’s usually a trade-off between a detailed study of galaxy formation and space reionization, but THESAN manages to do both,” said astrophysicist John Wise of the Georgia Institute of Technology, who has not worked on THESAN.

The developers of THESAN built it on the back of an older series of simulations called Illustris-TNG, which have been shown to model exactly many of the properties and populations of developing galaxies. They then developed a new algorithm to model how light from stars and galaxies interacted and reonized the surrounding gas during the first billions of years of the universe – details that previous simulations have not successfully incorporated on a large scale. Finally, the THESAN team included a model of how space dust in the early universe may have affected galaxy formation.

“They combined two state-of-the-art models and added a little more – it looks really interesting,” said Risa Wexler, a cosmologist at Stanford University and director of the Cowley Institute for Particle Astrophysics and Cosmology, who did not. participate in THESAN.


THESAN can trace the birth and evolution of hundreds of thousands of galaxies in a cubic volume spanning more than 300 million light-years. Starting about 400,000 years after the Big Bang – before the first stars were thought to have appeared – the simulation extrapolated to the first billions of years of space history. To do all this, THESAN runs on one of the largest supercomputers in the world, SuperMUC-NG, which used nearly 60,000 computer processing cores to perform simulation calculations equivalent to 30 million CPU hours. (For perspective, the same computational feat would require 3,500 years of special crunching of numbers on a typical desktop computer.)

Depiction of the THESAN simulation, showing stars and galaxies in the early universe that interact and reionize the surrounding gas clouds to create the familiar cosmic structures we see today.

“One of the most exciting things about THESAN simulations for me is the increased resolution,” said astrophysicist Brian Welch of Johns Hopkins University, who has not worked on THESAN. “They seem to be able to connect the small-scale structures in galaxies that create ionizing photons with the larger intergalactic environment where these photons drive the reionization era. Simulations can then help determine how ionizing photons leave galaxies and thus how these galaxies drive reionization.

Using the Hubble Space Telescope, Welch and colleagues recently discovered the most distant single star ever discovered, called Earendel, which dates back to a time when the universe was only 900 million years old. Although THESAN cannot simulate individual stars like Earendel, “as this would require an excessive amount of computing power,” it can still shed light on the conditions in the galaxies in which Earendel and his compatriots form, he said.

Researchers say THESAN is already forecasting an early universe. For example, this suggests that the distance traveled by light has increased near the end of reionization more dramatically than previously thought – by a factor of 10 over several hundred million years – probably because dense gas pockets, which took longer to ionize, have been omitted from previous lower-resolution simulations.

One drawback of THESAN, however, is that it uses a relatively simplified model for cold solid gas in galaxies, Canan said. The THESAN team is currently working on a follow-up project called THESAN-ZOOMS to replace this model with “a much more complex one that takes into account many additional physical processes that affect the properties of this dense gas,” he said.

Another disadvantage of THESAN is that the volume it simulates is too small to correctly determine key details about how the early universe evolved, such as the size and number of pockets of ionized transparent gas, Canan said. Scientists currently plan to increase the simulation to 64 times the volume through a diverse set of optimization settings designed to improve its overall performance, he said.

Expectations versus reality

Whether any of these shortcomings actually make a significant difference to THESAN’s predictions may soon be revealed by new observations from JWST, which is designed to see the first stars and galaxies. Will the stars and galaxies that merge in THESAN’s virtual space reflect the populations of ancient objects, as seen from JWST’s optics? Researchers are eager to find out. Models of weak galaxies in the early universe are very sensitive to uncertainties in phenomena such as star formation, “which remain highly debated,” said Aaron Jung, a theoretical astrophysicist at NASA’s Space Flight Center who has not worked on THESAN. Simulations that can successfully model known galaxies “can provide different predictions in weak populations. [JWST] will discover these galaxies for the first time and provide limitations on the physics that drive the formation of these galaxies. “

By the end of this year, JWST will be able to gather enough data to test THESAN when it comes to many predictions about the properties of galaxies, Smith said. “We are already working with astronomers involved in JWST to interpret the data that will be available this year.

“My intuition tells me that JWST will match the statistics of bright galaxies modeled in CoDa, CROC and THESAN,” said Wise, who helped develop the Renaissance simulations. However, they do not have enough resolution to model low-mass and small galaxies where Renaissance and SPHINX will match better. Astrophysicists, he said, are likely to use a combination of the two types of simulations to interpret JWST observations of ancient galaxies.

No one expects TEZAN or any other simulation of the reionization era to fix everything completely. “Most, if not all, of the simulations made in this era lack some physics – although THESAN is quite high resolution, it is still low resolution compared to the physical processes that actually take place.” says Wexler. “Progress happens when data from observatories and insights from simulations work together. This interaction is exciting. “

Ultimately, “we will need more than JWST to confirm the full picture of cosmic evolution in the early universe,” Smith said. “Different tools covering a wide range of wavelengths are needed to understand the different aspects of this era. These include the Hydrogen Age of the Reionization Massif (HERA), the Square Kilometer (SKA), the Fred Young Submillimeter Telescope (FYST), the Universe Spectrophotometer, the Rheonization Age and Ice Explorer (SPHEREx), and the next of NASA, the Roman Space Telescope Nancy Grace. Ambitious computer models like THESAN can ultimately help scientists make sense of the flow of data that these projects will bring.

“THESAN aims to make predictions for as many of these observations as possible,” Smith said. “Data discrepancies are often just as exciting because it tells us that our models are missing, forcing us to rethink the basic physics of these complex processes.”

Previous articleData breaches by the UK Council underline the importance of in-depth training
Next articleGlobe joins Opensignal’s Global Rising Stars for the third year in a row