Our solar system has so far held the record for the largest number of planets around a single star. The solar system Kepler 90 equals this record
Our solar system has so far held the record for the largest number of planets around a single star. The solar system Kepler 90 equals this record. Kepler 90, the sun of that system is a star similar to our sun, about 2,500 light years away. The system was discovered in follow-up observations of the most suitable stars for hosting solar systems discovered by the Kepler space telescope during its years of operation.
One of the discovered planets, Kepler-90i - a hot and rocky planet that orbits its star once every 14.4 days - was found using Google's machine learning. Machine learning is an artificial intelligence approach where the computers "learn". In this case, the computers learned to recognize the planets by finding changes in starlight when the planets pass between it and us.
"Just as we expected, there are exciting discoveries lurking in Kepler's archived data, waiting for the right instrument or the right technology to uncover them," said Paul Hertz, director of NASA's Astrophysics Division in Washington. "This finding shows that our data will serve as a treasure at the disposal of researchers for years to come.
Researchers Christopher Shallue and Andrew Vandenburg trained a computer to detect extrasolar planets in Kepler data even when changes in star brightness are small. The system is based on neural networks that were built inspired by the connection of neurons in the human brain. The neural network dug into the Kepler data and found signals that made it possible to identify the eighth planet in the Kepler 90 system, located in the constellation Draco.
Machine learning has previously been used in searches of the Kepler database, this follow-up study demonstrates that neural networks are a promising tool in detecting the faint signals of distant worlds.
Kepler-90i is about 30% larger than Earth, and is so close to its star that its average surface temperature is higher than 800 degrees Celsius, like on the planet Mercury. The system's outer planet, Kepler - 90h, orbits it at a similar distance to its star as Earth orbits the Sun.
"The star system of Kepler-90 is like a mini version of our solar system. It has small planets inside and large stars outside, but everything in it is much closer," said Vandenburg, an astronomer at the University of Texas at Austin.
Shala, a senior software engineer in Google's research team for artificial intelligence - Google AI, came up with the idea of applying a neural network to the Kepler data. He became interested in the discovery of planets outside the solar system after learning that this field of astronomy, like other branches of science, is increasingly flooded with data as the technologies for collecting data from space advance.
The project was born from Shala's initiative, when he dedicated the usual twenty percent of his time at Google to searching for planets in the Kepler data, and added Vandenberg to the project.
The database collected by Kepler in its four years of operation contains 35,000 signals from possible planets. Automated tests combined with a human eye make it possible to locate the most promising signals in the data. However, the weak signals are often missed. The two thought it would be interesting to find out if there were more planets lurking in the data.
They first trained the neural network to identify planets that had already been discovered, and after the network learned to recognize the pattern of planetary transits, the researchers used their model to look for faint signals in the 670 known solar systems with more than one planet. Their assumption was that multi-planet systems would be the best places to look for more planets.
"We had a lot of false results of discovering a planet where it wasn't, but also a real potential for discovering more planets," Vandenburg said. "It's like sifting through rocks to find jewelry. If you have a finer sieve you will catch more rocks but you will also be able to catch jewels and other treasures.”
Kepler-90i wasn't the only diamond the neural network spotted. In the Kepler-80 system, they found the sixth planet, Kepler 80g, with four of its planets circling the star in a resonant chain—where the planets are locked through their mutual attraction in a rhythmic orbital dance. The result is a very stable system.
The two want to use their neural network to look for planets around all the stars photographed by Kepler - over 150 stars, and look for signals in them that may have disappeared from the previous researchers.
Kepler produced the unprecedented database used to discover planets after observing one region of the sky for four years. Now the spacecraft is operating on a long mission, returning to each area every 80 days.
"These results illustrate the value of the Kepler mission," said Jessie Dotson, a Kepler mission scientist at NASA's Ames Research Center in Silicon Valley, California. "New ways of looking at the data, as we did in this study using proper learning promise to continue to yield significant progress in our understanding of planetary systems around other stars. I'm sure there are many more planets in the data waiting for people to discover them."
More of the topic in Hayadan:
- The Kepler space telescope found 219 more planets and revealed a clear gap between terrestrial planets and mini-Neptunes
- The Kepler data increase the likelihood of a widespread distribution of life in the universe
- Seven planets pop up on every hill - Kepler's discovery broke the record for planets in one solar system
4 תגובות
Although "Abby Blizovsky" doesn't sound like a code name for a Google translation project. In any case, there is room for improvement.
Father, the Google translation of the article contains entire sections that make no sense. Please hide copy more thoroughly.
Impressive and important
But you still can't trust the computer to translate properly.