[Press Release]
Deep learning speeds up galactic calculations
A new way to simulate supernovae may help shed light on our cosmic...

Supernovae, exploding stars, play a critical role in the formation and evolution of galaxies. However, key aspects of them are notoriously difficult to simulate accurately in reasonably short amounts of time. For the first time, a team of researchers, including those from The University of Tokyo, apply deep learning to the problem of supernova simulation. Their approach can speed up the simulation of supernovae, and therefore of galaxy formation and evolution as well. These simulations include the evolution of the chemistry which led to life.

When you hear about deep learning, you might think of the latest app that sprung up this week to do something clever with images or generate humanlike text. Deep learning might be responsible for some behind-the-scenes aspects of such things, but it’s also used extensively in different fields of research. Recently, a team at a tech event called a hackathon applied deep learning to weather forecasting. It proved quite effective, and this got doctoral student Keiya Hirashima from the University of Tokyo’s Department of Astronomy thinking.

“Weather is a very complex phenomenon but ultimately it boils down to fluid dynamics calculations,” said Hirashima. “So, I wondered if we could modify deep learning models used for weather forecasting and apply them to another fluid system, but one that exists on a vastly larger scale and which we lack direct access to: my field of research, supernova explosions.”

Supernovae occur when suitably massive stars burn through most of their fuel and collapse in enormous explosions. They are so huge that they can, and do, influence large areas inside their host galaxies. If a supernova had happened a few hundred years ago within a few hundred light-years from Earth, you might not be reading this article right now. So, the better we understand supernovae, the better we can understand why galaxies are the way they are.

A more efficient simulation. During a supernova simulation, (left) shows the prediction by a current simulation method. (right) shows the prediction by 3D-MIM, which looks close enough to the that of the current leading method, but it takes far less time to execute, saving time, energy and costs for computing time. ©2023 Hirashima et al. CC-BY-ND
Divide and conquer. The upper images show a wide area of a galaxy being simulated. The time resolution is very low, in which each “step” of the simulation is around 100,000 years. The lower images show the specific area affected by a supernova explosion and have a finer time resolution where each step is under 10,000 years. These regions are combined with the more general simulation to improve the overall accuracy and efficiency of the simulation. ©2023 Hirashima et al., NASA/JPL-Caltech/ESO/R. Hunt/Hubble/L. Calçada CC-BY-ND

“The problem is the time it takes to calculate the way supernovae explode. Currently, many models of galaxies over long time spans simplify things by pretending supernovae explode in a perfectly spherical fashion, as this is relatively easy to calculate,” said Hirashima. “However, in reality, they are quite asymmetric. Some regions of the shell of material that forms the boundary of the explosion are more complex than others. We applied deep learning to help ascertain which parts of the explosion require more, or less, attention during a simulation to ensure the best accuracy, whilst also taking the least amount of time overall. This way of dividing a problem is called Hamiltonian splitting. Our new model, 3D-MIM, can reduce the number of computational steps in the calculation of 100,000 years of supernova evolution by 99%. So, I think we’ll really help reduce a bottleneck too.”

Of course, deep learning requires deep training. Hirashima and his team had to run hundreds of simulations taking millions of hours of computer time (supercomputers are highly parallel, so this length of time would be divided amongst the thousands of computing elements required). But their results proved it was worth it. They now hope to apply their methodology to other areas of astrophysics; for example, galactic evolution is also influenced by large star-forming regions. 3D-MIM models the deaths of stars, and perhaps soon it will be used to model their births as well. It could even find use beyond astrophysics altogether in other fields requiring high spatial and temporal resolutions, such as climate and earthquake simulations.

Papers

Keiya Hirashima, Kana Moriwaki, Michiko S. Fujii, Yutaka Hirai, Takayuki R. Saitoh, and Junichiro Makino, “3D-Spatiotemporal Forecasting the Expansion of Supernova Shells Using Deep Learning toward High-Resolution Galaxy Simulations,” Monthly Notices of the Royal Astronomical Society: October 23, 2023, doi:10.1093/mnras/stad2864.
Link (Publication)

Related links

[Press Release]
Deep learning speeds up galactic calculations
A new way to simulate supernovae may help shed light on our cosmic origins は
コメントを受け付けていません
News, UTOPS-News

Read more

[Press Release]
“Starquakes” could explain mystery signals
Fast radio bursts from distant neutron stars resemble earthquakes rather than solar flares

Fast radio bursts, or FRBs, are an astronomical mystery, with their exact cause and origins still unconfirmed. These intense bursts of radio energy are invisible to the human eye, but show up brightly on radio telescopes. Previous studies have noted broad similarities between the energy distribution of repeat FRBs, and that of earthquakes and solar flares. However, new research at the University of Tokyo has looked at the time and energy of FRBs and found distinct differences between FRBs and solar flares, but several notable similarities between FRBs and earthquakes. This supports the theory that FRBs are caused by “starquakes” on the surface of neutron stars. This discovery could help us better understand earthquakes, the behavior of high-density matter and aspects of nuclear physics.

Earthquake map. Data on earthquakes was taken from Japan’s Kanto region (including Tokyo and Narita) and Izumo in the Chugoku region (north of Hiroshima). Black dots represent the epicenters of earthquakes recorded between May 6, 2010, and December 31, 2012. ©2023 T. Totani & Y. Tsuzuki

The vastness of space holds many mysteries. While some people dream of boldly going where no one has gone before, there is a lot we can learn from the comfort of Earth. Thanks to technological advances, we can explore the surface of Mars, marvel at Saturn’s rings and pick up mysterious signals from deep space. Fast radio bursts are hugely powerful, bright bursts of energy which are visible on radio waves. First discovered in 2007, these bursts can travel billions of light years but typically last mere thousandths of a second. It has been estimated that as many as 10,000 FRBs may happen every day if we could observe the whole sky. While the sources of most bursts detected so far appear to emit a one-off event, there are about 50 FRB sources which emit bursts repeatedly.

Comparing FRBs and earthquakes. The researchers analyzed the time and energy distribution of FRB and earthquake events, and by plotting the aftershock likelihood as a function of time lag, they found that the two are very similar. ©2023 T. Totani & Y. Tsuzuki

The cause of FRBs is unknown, but some ideas have been put forward, including that they might even be alien in origin. However, the current prevailing theory is that at least some FRBs are emitted by neutron stars. These stars form when a supergiant star collapses, going from eight times the mass of our sun (on average) to a superdense core only 20-40 kilometers across. Magnetars are neutron stars with extremely strong magnetic fields, and these have been observed to emit FRBs.

“It was theoretically considered that the surface of a magnetar could be experiencing a starquake, an energy release similar to earthquakes on Earth,” said Professor Tomonori Totani from the Department of Astronomy at the Graduate School of Science. “Recent observational advances have led to the detection of thousands more FRBs, so we took the opportunity to compare the now large statistical data sets available for FRBs with data from earthquakes and solar flares, to explore possible similarities.”

So far, statistical analysis of FRBs has focused on the distribution of wait times between two successive bursts. However, Totani and co-author Yuya Tsuzuki, a graduate student in the same department, point out that calculating only the wait-time distribution does not take into account correlations that might exist across other bursts. So the team decided to calculate correlation across two-dimensional space, analyzing the time and emission energy of nearly 7,000 bursts from three different repeater FRB sources. They then applied the same method to examine the time-energy correlation of earthquakes (using data from Japan) and of solar flares (using records from the Hinode international mission to study the sun), and compared the results of all three phenomena.

Receiver of the Arecibo Telescope, Puerto Rico. FRB data was provided by the Five-hundred-meter Aperture Spherical Telescope (FAST) in China and the Arecibo Telescope in Puerto Rico, two of the largest single-dish telescopes in the world. Unfortunately, the Arecibo Telescope was damaged and subsequently decommissioned in 2020.

Totani and Tsuzuki were surprised that, in contrast to other studies, their analysis showed a striking similarity between FRBs and earthquake data, but a distinct difference between FRBs and solar flares. Totani explained: “The results show notable similarities between FRBs and earthquakes in the following ways: First, the probability of an aftershock occurring for a single event is 10-50%; second, the aftershock occurrence rate decreases with time, as a power of time; third, the aftershock rate is always constant even if the FRB-earthquake activity (mean rate) changes significantly; and fourth, there is no correlation between the energies of the main shock and its aftershock.”

This strongly suggests the existence of a solid crust on the surface of neutron stars, and that starquakes suddenly occurring on these crusts releases huge amounts of energy which we see as FRBs. The team intends to continue analyzing new data on FRBs, to verify that the similarities they have found are universal. “By studying starquakes on distant ultradense stars, which are completely different environments from Earth, we may gain new insights into earthquakes,” said Totani. “The interior of a neutron star is the densest place in the universe, comparable to that of the interior of an atomic nucleus. Starquakes in neutron stars have opened up the possibility of gaining new insights into very high-density matter and the fundamental laws of nuclear physics.”

Papers

Tomonori Totani and Yuya Tsuzuki, “Fast radio bursts trigger aftershocks resembling earthquakes, but not solar flares,” Monthly Notices of the Royal Astronomical Society: October 11, 2023, doi:10.1093/mnras/stad2532.
Link (Publication)

Related links

[Press Release]
“Starquakes” could explain mystery signals
Fast radio bursts from distant neutron stars resemble earthquakes rather than solar flares は
コメントを受け付けていません
News, UTOPS-News

Read more

[Press Release]
Why Does Asteroid Ryugu Look Different in Space and in the Laboratory?
~Space Weathering Hides the Signs of Water

Joint Press Release
Shogo Tachibana, Professor, Department of Earth and Planetary Science, Space and Planetary Science Organization

Moe Matsuoka, a researcher at the National Institute of Advanced Industrial Science and Technology (AIST) and Toru Kouyama, a team leader, National Institute of Advanced Industrial Science and Technology, in collaboration with Prof. Tomoki Nakamura, Tohoku University; Kana Amano, Japan Society for the Promotion of Science Research Fellow, Tohoku University; Dr. Takahito Osawa, Research Director, Japan Atomic Energy Agency; Prof. Shogo Tachibana, University of Tokyo; Prof. Hiroshi Naraoka and Associate Prof. Takashi Okazaki, Kyushu University, etc. and other researchers, conducted a direct comparison between the data obtained by the asteroid probe Hayabusa2 from the sky over the surface of the asteroid Ryugu and the data obtained by measuring samples collected and brought back (sample return) from Ryugu without exposing them to the Earth’s atmosphere.

As a result, we found that while the observed data from Ryugu’s surface and the sample return data agree well, there is a clear difference in the absorption of hydroxy groups (-OH), which is the key to determine the presence of water. In order to clarify the cause of this difference, experiments and data analysis of primitive meteorites similar to Ryugu, which are rich in hydrous silicates, revealed that Ryugu’s surface (about 1/100 mm) has been altered (space weathering) by exposure to cosmic rays and dust, resulting in partial loss of water.

This research result, which was only possible through a combination of remote observation from the spacecraft and analysis of collected samples, is one of the landmark results that demonstrate the importance of sample return in planetary exploration. The details of the study were published in Communications Earth & Environment on September 27, 2023 (JST).

For more information, please refer to the following

Graduate School of Science web: https://www.s.u-tokyo.ac.jp/ja/press/10047/
Publication URL: https://www.nature.com/articles/s43247-023-00991-3

[Press Release]
Why Does Asteroid Ryugu Look Different in Space and in the Laboratory?
~Space Weathering Hides the Signs of Water は
コメントを受け付けていません
News, UTOPS-News

Read more