Can Kepler Data and Machine Learning Discover Habitable Exoplanets?

[gdarie]

hello everyone anyone knows if this research is made? imagine you have available the data from the kepler telescope to search exoplanet, create a training set of data based on the concept of find exo planet , calculate their orbit and try to design the virtual exo solar sistem , understand the timing of impact between planet to finally identify the exo solar system with exo planet right for life and the all solar system it has been more quite (without planet impacts) for longer time. in that solar system is possible that life is present. does make sense?

 

[mfb]

Your post is a bit hard to read.

The Kepler collaboration has scripts to search for exoplanets automatically. They are verified manually as scripts can't do everything. Sometimes dedicated measurements from other telescopes are used to help constrain some of the parameters of the system.

The conditions for habitability are not well-known.

 

[rootone]

Kepler is designed to detect planets in other solar systems.
It has no ability to discover if these planets could have life on them,
The result so far is lots of planets, most of them like Jupiter or Neptune, not places where life similar to Earth could exist.
It has made clear though that solar systems similar to our one do exist.
It would be easy to miss a small possibly habitable planet in a system dominated by large gas giants.

 

[gdarie]

with kepler data you might know the composition of the planet , the position in the solar system, and the atmosphere and with the data you can create the orbits. Well, I've read that Trappist and Kepler b are good candidates. Apologize for my English! Do you know any resercher is using machine learning to study the life/history of solar system found by Kepler. Maybe understand if there would have been collision in the past between planets, or there would be any collision in future. How long the system is been quite. Does make sense?

 

[mfb]

Kepler only gives you the orbital period and the size of the planet relative to the star. In a few cases transit timing variation gives a mass estimate.
Kepler data doesn't tell you anything about atmospheres.

Well, I've read that Trappist and Kepler b are good candidates.

There are no planets with this name. There is a TRAPPIST-1 system, which has very interesting planets. There are many Kepler-xxx systems, where xxx is just a number. The planets in these systems are then the star name followed by b,c,d,... As an example, there is a star Kepler-90 with a planet Kepler-90 b.

Maybe understand if there would have been collision in the past between planets, or there would be any collision in future. How long the system is been quite.

I don't understand that question.
We only see the present state of the system, and Kepler needs developed planets to see anything - planetary collisions in the near past are unlikely.

 

[rootone]

The planets which have been identified by Kepler have orbits around their host star.
That's how Kepler works, variation of light seen periodically from a star indicates that something is orbiting it.
At present we can only say that a planet exists and not much more other than it's orbital period
I don't see your point about machine learning other than it is a useful tool for data analysis.

 

[gdarie]

Machine learning can help to understand the data, with a special algorithm it's possible to create the orbits of the planets discovered by Kepler. Do you think that knowing the rotation, the orbits , the mass and the distance of the planets from the star, is it possible to know if there has been / there will be impacts? To understand how quite it has been the solar system. With machine learning someone found stars there are faster then the escape speed needed to go out of milky way. To allow life beside all conditions I think the solar system should be relatively quite.

 

[gdarie]

I know more or less how Kepler works. I also saw the website where users manually can indicate if there is a shrink in lights.
There is an enormous amount of data that only machine learning algorithms can manage. I've read that some university are start using ML in astrophysics , do you know any involved in this research. Apologize for my English!

 

[rootone]

Yes it's hard to imagine how life might survive in a gravitationally chaotic region where planets are crashing into each other and sometimes being kicked out of their solar system.
Most of the solar systems discovered so far are not so unstable, but many of the planets are not good candidates for life for other reasons.
The main reason being the planet is either too close to the host star or too far away for liquid water to exist.

 

[gdarie]

I agree rootone, that's for the moment! ML needs time to improve and a lot of training data set. Soon there will be other system like Trappist-1, I think it's a good timing now, waiting for JWST, which in few years will come tons of data that need to be studied. A ML well trained can read the data in real-time and give you results. I know a guy well know on ML community , which is a genius in ML and python development, he should be in contact with NASA and do his Magic . Does make sense?

 

[rootone]

I am certain that NASA tries hard to recruit the best software engineers that they can,
and amongst those there will be a few of them who are specialist in such fields as AI and automated data analysis.