Tyler O'Neal, Staff Editor ASTRONOMY

Dutch researcher's simulations reveal how black holes become supermassive

Leiden astronomy Master's student Arend Moerman has received an A+ for his thesis research on the simulation of chaotic interactions of three black holes. The simulations, which he carried out together with his Leiden and Oxford colleagues, show that lighter black holes tend to slingshot each other out into space, while heavier ones tend to merge. The research appeared in the academic journal Physical Review D.

Moerman spent a year investigating the dynamic interactions and collisions between three imaginary black holes. The interactions between three bodies such as stars, planets, or black holes cannot be predicted with an elegant formula. Moerman, therefore, used a supercomputer that calculates what happens for a short time and then uses the result as input for the prediction of the next period. Arend Moerman

Extended with the theory of relativity

The computer code is an extended version of the code used by first author Tjarda Boekholt in 2020 and 2018. The new, extended code also included Einstein's theory of relativity. This is important because the theory of relativity plays a major role, especially in the case of heavy objects such as black holes. 

The researchers varied the masses of the three interacting black holes. Starting with one solar mass, they increased the mass up to a billion times the mass of the sun. 

Tipping point

Around ten million solar masses, there appeared to be a tipping point. In the simulations, black holes that are lighter than about ten million solar masses mostly eject each other through a gravitational slingshot. Black holes that were heavier than this, started to merge. First, two black holes merge. The third follows later. The black holes merge because they lose kinetic energy and that is because they emit gravitational waves. 

"Arend's work has led to a new understanding of how black holes become supermassive," says Simon Portegies Zwart. "In the simulations, we see that heavy black holes no longer endlessly move around each other, but that, if they are heavy enough, they collide pretty much instantly." Black holes.

Following your interests 

Moerman was looking for a graduation subject at the interface of astronomy, mathematics, and computer science. This he found with his supervisor Portegies Zwart. "Programming appealed to me. In the beginning, it was not quite clear what I had to do," said Moerman. "I had to build the theory of relativity into an existing code, but how and what was not yet clear. But that was the fun part: it gave me the freedom to choose and simulate what I found interesting."

From stargazer to star chef

An A+ is wonderful, but Moerman takes most satisfaction from the more tangible result: "A grade is a grade, I am most proud of our research. Nevertheless, the A+ was celebrated extensively with friends. "It was a tough week," he confesses with a laugh. 

Moerman has meanwhile started a second graduation research project. That is about DESHIMA, a Dutch-Japanese spectroscope on-chip. After his master's he would like to pursue a Ph.D. I've seen interesting positions passing by and have even sent out an application. But even if that doesn't work out, there are options: "I'm now working as a sous-chef in a restaurant. If science doesn't work out, I'll switch to being a chef."

AI helping to quantify enzyme activity

Tyler O'Neal, Staff Editor ASTRONOMY

Without enzymes, an organism would not be able to survive. It is these biocatalysts that facilitate a whole range of chemical reactions, producing the building blocks of the cells. Enzymes are also used widely in biotechnology and in our households, where they are used in detergents, for example.

To describe metabolic processes facilitated by enzymes, scientists refer to what is known as the Michaelis-Menten equation. The equation describes the rate of an enzymatic reaction depending on the concentration of the substrate – which is transformed into the end products during the reaction. A central factor in this equation is the ‘Michaelis constant’, which characterizes the enzyme’s affinity for its substrate. Schematic presentation of the prediction process for Michaelis constants of enzymes using deep learning methods. (Image: HHU / Swastik Mishra) CREDIT HHU / Swastik Mishra

It takes a great deal of time and effort to measure this constant in a lab. As a result, experimental estimates of these constants exist for only a minority of enzymes. A team of researchers from the HHU Institute of Computational Cell Biology and Chalmers University of Technology in Stockholm has now chosen a different approach to predict the Michaelis constants from the structures of the substrates and enzymes using AI.

They applied their approach, based on deep learning methods, to 47 model organisms ranging from bacteria to plants and humans. Because this approach requires training data, the researchers used known data from almost 10,000 enzyme-substrate combinations. They tested the results using Michaelis constants that had not been used for the learning process.

Prof. Lercher had this to say about the quality of the results: “Using the independent test data, we were able to demonstrate that the process can predict Michaelis constants with an accuracy similar to the differences between experimental values from different laboratories. It is now possible for computers to estimate a new Michaelis constant in just a few seconds without the need for an experiment.”

The sudden availability of Michaelis constants for all enzymes of model organisms opens up new paths for metabolic supercomputer modeling, as highlighted by the journal PLOS Biology in an accompanying article.

Japanese built AI enables high-fidelity quantum supercomputing

Tyler O'Neal, Staff Editor ASTRONOMY

Researchers at SANKEN use machine learning classification to dramatically improve accuracy when reading the spin states of electrons on quantum dots, which may lead to more robust and practical quantum supercomputing

In Japan, researchers led by the Institute of Scientific and Industrial Research (SANKEN) at Osaka University have trained a deep neural network to correctly determine the output state of quantum bits, despite environmental noise. The team’s novel approach may allow quantum computers to become much more widely used.

Modern computers are based on binary logic, in which each bit is constrained to be either a 1 or a 0. But thanks to the weird rules of quantum mechanics, new experimental systems can achieve increased computing power by allowing quantum bits, also called qubits, to be in “superpositions” of 1 and 0. For example, the spins of electrons confined to tiny islands called quantum dots can be oriented both up and down simultaneously. However, when the final state of a bit is read out, it reverts to the classical behavior of being one orientation or the other. To make quantum supercomputing reliable enough for consumer use, new systems will need to be created that can accurately record the output of each qubit even if there is a lot of noise in the signal.

Now, a team of scientists led by SANKEN used a machine learning method called a deep neural network to discern the signal created by the spin orientation of electrons on quantum dots. “We developed a classifier based on a deep neural network to precisely measure a qubit state even with noisy signals,” co-author Takafumi Fujita explains.

In the experimental system, only electrons with a particular spin orientation can leave a quantum dot. When this happens, a temporary “blip” of increased voltage is created. The team trained the machine learning algorithm to pick out these signals from among the noise. The deep neural network they used had a convolutional neural network to identify the important signal features, combined with a recurrent neural network to monitor the time-series data.

“Our approach simplified the learning process for adapting to strong interference that could vary based on the situation,” senior author Akira Oiwa says. The team first tested the robustness of the classifier by adding simulated noise and drift. Then, they trained the algorithm to work with actual data from an array of quantum dots and achieved accuracy rates over 95%. The results of this research may allow for the high-fidelity measurement of large-scale arrays of qubits in future quantum supercomputers.