We have discovered a mini version of the solar system – the Kepler-90 star system, which has the highest number of planets in a single system ever recorded.
Our solar system is truly special, and there are two reasons for its uniqueness. First is Earth – the only planet known to humanity that can sustain life. Second, it is a star system with the most planets, featuring 8 celestial bodies orbiting our blazing sun.
However, it seems that the solar system has become a bit less special following NASA’s announcement during a recent press conference. The Kepler space telescope has discovered a new planet in the Kepler-90 star system, located 2,545 light-years away from Earth.
This is the star system with the most planets, featuring 8 celestial bodies orbiting the blazing sun.
This star system has been confirmed to have a total of 7 planets. With this new discovery, it officially matches the number of celestial bodies orbiting a star in our solar system.
The new planet – named Kepler-90i – is a scorching hot planet with a solid surface. It has an orbital period of 14.4 days around Kepler-90.
What’s remarkable is that Kepler has collaborated with Google’s AI system to discover this planet.
Specifically, this AI system can “self-learn” (machine learning). It analyzed information in Kepler’s data set while receiving signals from exoplanets and subsequently confirmed the signal of a new planet.
Kepler-90 star system.
“Just as expected, this discovery was hidden in Kepler’s data, waiting for us to have a powerful enough tool to unearth it” – quoted Paul Hertz, director of NASA’s Astrophysics Division at headquarters in Washington.
“This proves that our available data is a treasure for future generations to ‘dig up’ in the years to come.”
Discovery Inspired by the Human Brain
This discovery stemmed from researchers Christopher Shallue and Andrew Vanderburg training a computer to find exoplanets based on light data recorded by Kepler. These are very small changes in brightness when a planet passes in front of a star.
Inspired by the neurons in the human brain, the “neural networks” in AI sift through Kepler’s data, searching for tiny signal changes, ultimately uncovering the long-hidden eighth planet of Kepler-90 in the Draco constellation.
It is a yellow dwarf star similar to the sun, with a surface temperature slightly higher than the sun, around 5930K. Thus, its brightness is also slightly greater than that of the sun (with an absolute magnitude of about 4.54, brighter than the sun’s 4.83), but since this star system is 2,545 light-years away, its apparent magnitude is only 14, making it invisible to the naked eye.
Kepler 90 is a G-type main-sequence star located about 2,545 light-years (780 pc) from Earth in the constellation Draco. This star is notable for having a planetary system with the same number of observable planets as our solar system.
Thus, it can be said that Kepler 90 is a sibling of the solar system; however, it is a younger sibling, according to astronomical studies, the Kepler 90 star system formed 2.6 billion years later than the sun and is only 2 billion years old, while our solar system is 4.6 billion years old.
Our solar system has 8 planets orbiting around it, and so does Kepler 90. These planets are named in order of their proximity to the sun, designated as b, c, i, d, e, f, g, and h. The planet closest to the sun is called Kepler 90b, and the farthest is called Kepler 90h.
In fact, the use of machine learning technology to study Kepler’s data had been attempted before, but not with high expectations. However, this time, it has shown great potential in discovering the smallest signals in extremely distant galaxies that we could not reach.
The planet crossing the host star will create a very slight change in the light signal.
Kepler-90i is quite impressive in size, about 30% larger than Earth. However, the surface of the planet is extremely hot – estimated to be over 800 degrees Fahrenheit (over 420 degrees Celsius), comparable to Mercury in our solar system. The farthest planet of Kepler-90, Kepler-90h, is at a distance equivalent to that of Earth to the sun.
“The Kepler-90 star system is like a miniature version of the solar system. Smaller inner planets, larger outer planets, but with much smaller distances than ours” – Vanderburg, a NASA researcher and astronomer from the University of Texas, stated.
Shallue, a senior software engineer at Google, was very excited about discovering a new exoplanet after realizing that astronomy – like all other sciences – is always “drowned” in a sea of data collected from the universe.
“In my spare time, I tried to explore and realized that the Kepler mission has a massive data source” – Shallue shared. “At this point, machine learning technology is truly useful, as the data is too vast for humans to study on their own.”
Over the course of 4 years, Kepler’s data archive has collected 35,000 signals “potentially” indicating a new planet. Previously, the way to verify them was through automated software and… human eyes, but weak signals were often overlooked. For Shallue and Vanderburg, this data could conceal unprecedented discoveries.
How Did They Train the AI?
First, they taught the AI’s neural network how to confirm signals from exoplanets, based on 15,000 previously recognized signals.
In the initial tests, this system achieved an accuracy of up to 96%. The system continued to self-learn to achieve even higher accuracy. Then, experts directed it to search for weaker signals among the 670 star systems that already had many confirmed planets. This action was based on the hypothesis that star systems with more than one planet are the most suitable places to search for additional hidden planets.
Simulation of Kepler-90i.
“We found many ‘false’ planets, but at the same time, the chance of finding a real planet is also greater” – Vanderburg stated. “It’s like sifting through rocks to find gems; with a larger sieve, you will pick up more rocks, but you will also find more gems.”
Kepler-90i is not the only “gem” that AI has discovered. In the Kepler-80 star system, which already has 5 planets, it also found a sixth planet. This is Kepler-80g, a planet comparable in size to Earth. Together with neighboring planets, they form a resonant chain, stabilizing the entire system, similar to the “Solar System 2.0” TRAPPIST-1 discovered in February 2017.
According to Shallue and Vanderburg, the duo plans to apply this AI system to the entire Kepler dataset, which includes 150,000 stars. This massive number promises many more planets to be discovered, and clearly, the chance of encountering another civilization is also greater.
The research will be published in The Astronomical Journal soon.
