Future sky studies will be carried out by plans for a radio telescope that aims to observe millions of early galaxies in the universe. However, automated tools are needed to process this huge data flow. An algorithm developed by a team from the Institute of Astrophysics and Space Sciences (IA) at the Faculty of Science at the University of Lisbon in Portugal was designed to process this data and identify galaxies containing massive black holes at their centres.

Galaxies fill the picture of the deep universe as far as the eye can see. What processes determined their shapes, colors, and stellar populations? Astronomers believe that primordial black holes are the engine for the growth and transformation of galaxies and may explain the cosmic landscape we see today.

Groundbreaking discovery in identifying superluminal galaxies

In an article recently published in the journal Astronomy and Astrophysics, An international team led by Rodrigo Carvajal from the Institute of Astrophysics and Space Sciences (IA) and the Faculty of Science of the University of Lisbon (Ciências ULisboa) offers machine learning. A technique for detecting superluminal galaxies in the early universe.

These galaxies are believed to be dominated by the activity of the voracious black hole at their cores. According to the authors, this should be the first algorithm to predict when this activity emits an intense RF signal. Radio emission is often different from other galactic light, and it is sometimes difficult to relate them. This artificial intelligence technique will allow astronomers to be more efficient in their search for galaxies, called radio galaxies.

Developed in collaboration with data science technology solutions company Closer, the algorithm was trained on images of galaxies taken at various wavelengths of the electromagnetic spectrum. When tested with other images, it was able to predict four times more radio galaxies than traditional methods using explicit instructions. As machine learning develops its own algorithms, trying to understand its success could help shed light on the physical events that occurred in these galaxies 1.5 billion years after the Big Bang, when the universe was a tenth of its current age.

The importance of further research and analysis

“We should find more active galaxies in the sky because there are predictions that there should have been many more galaxies in the early history of the universe. According to current observations, we do not have such a number,” says Rodrigo Carvajal. More observations are needed to test whether the current understanding of how active galaxies evolve is correct or needs to be revised, this researcher said.

“It’s also important to analyze machine learning models and understand what’s going on inside them,” adds Carvajal. “Which functions are most relevant to decision-making? For example, we want to know whether the most important feature of a module claimed to be an active galaxy is the light emitted by the galaxy in the infrared, possibly indicating the rapid formation of new stars. In this way, we can develop a new law that distinguishes ordinary and active galaxies.” .”

Examining the role of radio radiation and star formation

The relative weight of the galaxy’s features in the computer’s decision can indicate what underlies intense activity, especially in the radio range. In an upcoming study, Carvajal examines the consequences of this apparent relationship between radio emission and star formation. “These models are mathematical tools that help us look in the right direction as the complexity of the data increases. This work can shed light on the processes that prevented the formation of new stars in the second half of the universe’s history,” explains Israel Matute of IA and Ciências ULisboa, second author of the paper. Seemingly missing galaxies from the early universe may be among the huge amounts of data that modern radio telescopes will produce in the coming years. Future studies of large regions of the sky will reveal billions of galaxies. An example of this is a survey of the entire southern celestial hemisphere using the ASKAP radio telescope in Australia. It is the Evolution Map of the Universe (EMU) that will map it. An AI-led team is currently working with data from the pilot project of this survey. Once perfected, these instruments will be critical in processing the astronomical amounts of data that the future Square Kilometer Observatory (SKAO) will produce. Portugal is a member of the consortium of this observatory, which is currently under construction.

“In the new era where astronomy will have access to large amounts of data, it is becoming increasingly important to develop advanced methods to process and analyze these data,” says José Afonso from IA and Ciências ULisboa and co-author of this paper. “As IA, we develop and apply these techniques to unravel the origins of galaxies and the supermassive black holes found in many of them.”

The idea for the collaboration between Closer and IA was suggested by one of the co-authors, Helena Cruz, who has a PhD in physics and is a data scientist at Closer. His involvement was crucial to analyze and address the impact of uncertainties and inconsistencies between different data sources from multiple telescopes and observing programs used to train the machine learning algorithm.

“I realized that astronomy was a field with great opportunities for the research and development of machine learning models, and it made sense for me to apply my professional skills in this field,” says Helena Cruz. “I shared my interest with Closer and both parties immediately demonstrated their desire to work together, I see this as a continuation of my work at the company.”

Joao Pires da Cruz, Closer’s co-founder, professor and researcher, adds: “Closer thrives on the knowledge of its employees; that is its capital.” “The more complex and scientifically challenging the projects in which our team members participate, the greater the company’s capital will be. We will have employees who can solve our customers’ problems, such as the problem of signals from distant galaxies.”