Comprehensive coverage

Artificial intelligence reveals a long-standing mistake in the study of black holes

A new study, led by the UK's University of Bath, has found that supermassive black holes need both merging galaxies and cold gas to grow. This discovery, obtained through machine learning, may change our understanding of galaxy evolution

Artificial intelligence reveals a long-standing mistake in the study of black holes. Credit: The Science website via DALEE
Artificial intelligence reveals a long-standing mistake in the study of black holes. Credit: The Science website via DALEE

Astronomers have long known that supermassive blacks, which reside at the centers of most large galaxies, grow over time. Until now, it was generally assumed that galaxy mergers were the main cause of this growth. However, a new study, published in the journal "Monthly Notices of the Royal Astronomical Society", shows that this is not the full picture.

The team, led by Dr Mathilde Aviart-McKenzie, used machine learning to classify galaxy mergers more accurately than ever before. An analysis of about 8,000 active black hole systems showed that mergers are directly related to the growth of black holes only in a specific type of galaxy: star-forming galaxies that contain significant amounts of cold gas.

Cold gas is a cloud of atoms and gas that is not heated by radiation. It is the raw material for the creation of new stars, as well as the food for black holes. The study found that when two galaxies merge, they may create a stream of cold gas towards the black hole at the center of the merged galaxy. This current causes the black hole to grow, while emitting tremendous radiation.

However, the study also found that merging galaxies is not enough on its own to cause a black hole to grow. A black hole will only grow if the merged galaxy contains significant amounts of cold gas.

These findings may change our understanding of galaxy evolution. It is possible that cold gas plays a more important role than we previously thought in the development of black holes and in the development of galaxies in general.

A convolutional neural network (CNN) was used in this study. CNNs are particularly effective neural networks for recognizing patterns in images.

The use of machine learning in this study opens a new window into the study of black holes. "Whereas in the past we were limited to manually analyzing a relatively small number of galaxies, now we can examine thousands of galaxies at the same time," explains Dr. Abiert-McKenzie. "This ability allows us to get a broader and more accurate picture of the relationship between galaxy mergers, the growth of black holes and the evolution of galaxies."

The training data included simulations of galaxy mergers constructed using a hydrodynamic code. These simulations included images and other data describing the process of merging galaxies, such as: gas distribution: density and temperature of gas inside the galaxies, star distribution: number, mass and type of stars inside the galaxies, morphology: shape and structure of the galaxies.

After training the network, the method was applied to real images of galaxies taken by telescopes. The network processed each image and outputted a result representing the likelihood of a galaxy merger.

The team used various accuracy metrics to evaluate network performance, such as:

  • accuracy: the percentage of images correctly classified as merging galaxies.
  • sensitivity: percentage of galaxy mergers detected by the network.
  • specificity: the percentage of images that were not falsely identified as merging galaxies.

The use of machine learning in this study presented a number of challenges, such as quantity of data: collecting many and accurate observational data on galaxy mergers is a complex and time-consuming process Data quality: images obtained by telescopic observations may be noisy and contain distortions, which makes it difficult to identify galaxy mergers. In addition, to improve the interpretation of results, it is important to understand the factors that influence the network's classification results and interpret them correctly.

The use of machine learning opens a new window for the study of black holes and galaxies:

  • Identifying new types of galaxy mergers: Machine learning can detect rare galaxy mergers that are difficult to detect through manual classification.
  • Exploring the relationship between galaxy mergers and nuclear activity: Machine learning allows researchers to examine the relationship between galaxy mergers and nuclear activity, such as the formation of new stars and black hole explosions.
  • Studying galaxy evolution over time: Machine learning can help researchers better understand the process of galaxy evolution over time, focusing on the effect of galaxy mergers and black hole growth.

This research is a collaboration between researchers from the University of Bath, Johns Hopkins University, the Max Planck Institute for Astrophysics and the Harvard-Smithsonian Center for Astrophysics.

More of the topic in Hayadan:

10 תגובות

  1. There are phrases here that seem scientific and technical but are really not clear. For example:
    1. "The training data included simulations of galaxy mergers built using a hydrodynamic code." What is hydrodynamic code? What does it mean for a model if it is not trained on real observations but on simulations?
    2. "Cold gas is a cloud of atoms and gas that is not heated by radiation." What is a cloud of atoms if not a gas?

  2. Sorry, your scoping experiment does not meet the minimum standards of an acceptable laboratory experiment. The measurement error possibilities in it due to temperature or simply instrumentation are too high.

    So here I will conduct an experiment here on the site on a different topic: not technology but psychology.

    In the experiment you brought, you made 60 revolutions of the circumference and got a deviation of about 3 mm. It seems that if you conduct the same experiment with 600 rounds you will get a deviation of about 3 centimeters and with 6000 rounds a deviation of about 30 centimeters.

    Which are already significant results by much more than 3 mm.

    In the video you brought, the entire duration of the experiment was a little over a minute.

    So here's the psychological experiment: spend a quarter of an hour, or maybe even an hour, doing the experiment with a much higher number of rounds to get a much more significant measurement result.

    As a scientist who dedicates his life to the subject, an extra hour won't change much I believe.

    You have all the necessary equipment and the nice girl will be happy, I believe, to take another video that you can upload to the rest of the world.

    And now a promo:

    Asbar will not conduct the improved scope experiment that I suggested would last an hour at most, but will continue to demand that the scientific community conduct it for him, and that someone address his strange claims (the universe moves at a speed of 12c.. relative to what Asbar? and why not 17c?).

    And the reasons, as I have already said, are not related to technology (although we must admit that the scope device he built is a beautiful technological device) but to psychology: Asbar wants to preserve his self-image as a groundbreaking thinker, and knows full well that a real experiment will shatter this image to pieces.

    So here we got a successful experiment after all. Although not the way the Magenon nerve wanted, but it can be used as raw material for researchers of megalomania and paranoia.

  3. Thanks to Avi Blizovsky, editor of the science.
    Thank you very much for answering me, and leaving the video.
    I will always fondly remember the freedom of expression in knowledge that you granted me.
    Thanks again

    A. Asbar

  4. Deleted because you flood the site with delusional comments, I gave you years and I only get complaints. People want to read the comments and come across your scrolls that invent new science and more boast of a Nobel nomination. Let surfers enjoy the site. I can also brag about absurd things like being nominated for the Israel Prize.
    my father

  5. My job is to warn unsuspecting readers of your wordplay and careless scope experiment, and to present you in the true light: a lying charlatan suffering from delusions of grandeur.

  6. It is advisable to include in the article a link or at least a full CITATION to the original article

  7. The mistakes are repeated over and over again until the truth was revealed that there is a creator for the world, it's really amazing how they try to shake off the truth, pride makes people think that they know everything and there is no supreme power that controls everything from the atom to the expanses of the galaxies without question and far, far above the human mind.

Leave a Reply

Email will not be published. Required fields are marked *

This site uses Akismat to prevent spam messages. Click here to learn how your response data is processed.