In 2013, the European Space Agency launched the Gaia spacecraft. As the successor to the Hipparcos mission, this space observatory has spent the past three and a half years collecting information about space. Before its mission comes to an end next year (although it could be extended), the information obtained will be used to build the largest and most accurate 3D astronomical map.
During the space survey, Gaia also discovered several interesting things. For example, after exploring the Gaia catalog with a specially designed artificial neural network, a European team of researchers recently found six new hyperspeed stars in the Milky Way. And one of these stars is moving so fast that it could eventually leave our galaxy.
Their research, titled An Artificial Neural Network to Discover Hypervelocity Stars: Candidates in Gaia DR1 / TGAS, was recently published in Monthly Notices of the Royal Astronomical Society. It was also presented at the end of last month at the "European Week of Astronomy and Space Sciences", which was held from 26 to 30 June in Prague, Czech Republic.
Artistic concept of the Gaia telescope against the background of the Milky Way, Credit: ESA / ATG medialab; background: ESO / S. Brunier
Hypervelocity stars are rare and fascinating objects. While all the stars in the Milky Way are in constant motion in orbits around the center of our galaxy, some of them accelerate to speeds of hundreds of kilometers per second. In the past, astronomers have concluded that these fast-moving stars are the result of the passage of stars close to each other or a supernova of a stellar companion.
And just over a decade ago, astronomers became aware of a new class of high-speed stars believed to have been accelerated from past interactions with the supermassive black hole Sagittarius A *, which is in the center of the galaxy. These stars are very important for studying the general structure of the Milky Way, as they indicate the specific types of events and forces that have shaped their history.
As Elena Maria Rossi of Leiden University in the Netherlands, who is one of the authors of the article, explained in an ESA press release:
“These are stars that have traveled enormous distances through the Galaxy, but can be traced to its core - an area so dense and obscured by interstellar gas and dust that it is usually very difficult to observe - so that they provide us with important information about the gravitational field of the Milky Way from the center to its outskirts."
Promotional video:
Artistic concept of stars racing through the galaxy. Credit: European Space Agency
Finding such stars is not an easy task, in large part because their speed makes them extremely difficult to detect in the huge and crowded disk of the Milky Way. As a result, scientists have relied on young massive stars (from 2.5 to 4 solar masses) in the old stellar population of the Galaxy. Basically, their young age and large mass are indicators that they cannot arise there.
Combined with measurements of their past velocities and trajectories, this method confirmed the presence of hyperspeed stars in the past. However, so far, scientists have found only 20 hyperspeed stars, and they were all young and massive. Scientists believe that many more stars of other ages and masses are moving through the Milky Way at such high speeds, but they could not be detected before.
To tackle this problem, a European team of researchers led by Tomasso Marchetti of Leiden University in the Netherlands began to consider how to use Gaia's vast dataset to optimize searches for hyperspeed stars. After testing various methods, they adapted a neutron network for these purposes, that is, they used a machine learning algorithm to search the Gaia stellar census data during the collection process.
Beginning in the first half of 2016, the team began developing and training this program in preparation for the first Gaia data release, which happened a few months later, on September 14, 2016. Leiden University graduate student Tomasso Marchetti described the process as follows:
“In the end, we decided to use an artificial neutron network - software designed to simulate how our brains work. With the right "training", it can learn to recognize certain objects or patterns in a huge dataset. In our case, we trained her to look for hyperspeed stars in a star catalog, like the one compiled by Gaia.
Artistic representation of a hyperspeed star that was discovered by the ESO's Very Large Telescope. Credit: European Southern Observatory
In addition to a map with the locations of over a billion stars, the first data included a small catalog of the distances and movements of two million stars. This catalog, titled "Tycho-Gaia Astrometric Solution (TGAS)", combined data from the Gaia and Hipparcos missions.
On the day the catalog was published, Marchetti and his team ran an algorithm on two million stars within the TGAS, which led to interesting findings. "In just one hour, the artificial brain has already reduced the dataset to 20,000 potential high-speed stars, down to 1%," Rossi said. "Further selection, including only distance and movement with a certain measurement accuracy, brought in as many as 80 candidate stars."
The team then studied these 80 stars in more detail and compared their movement information with data from other catalogs. Combined with additional observations, they eventually found six stars that appear to be moving faster than 360 km / s. One of them even has a speed of over 500 km / s, which means that it is no longer bound by the gravity of our Milky Way and will eventually leave it.
But perhaps the most important aspect of this find is that these stars are not particularly massive, like the previous 20, whose mass was comparable to that of our Sun. In addition, the 5 slower stars are likely to become the focus of the entire study, as scientists want to determine what is holding them back. One possible explanation is that dark matter in the galaxy is responsible for this interaction.
Gaia's first sky map. Credit: ESA / Gaia / DPAC, A. Moitinho & M. Barros (CENTRA - University of Lisbon), on behalf of DPAC
Since TGAS was in many ways only a small fraction of Gaia's vast and valuable data, this study demonstrates the kinds of research discoveries that will be possible. With data on billions of stars, astronomers will certainly discover new and amazing knowledge about the dynamics of our Milky Way and the forces that created it.
To this end, Marchetti and his team are refining their software to handle the much larger dataset, which is expected to be released in April 2018. This catalog will include distances and motions for over a billion stars, as well as speeds for a specific subset. From it, the team learn that fast-moving stars in the Milky Way are much more common than previously thought.
And be sure to watch this video showing the trajectories of these newly discovered fast moving stars from ESA: