Rogue G

Not a rapper or a Star Wars reference. A more indicative title would have been "Rogue G values in the open exoplanet database". But I could not resist titling my post just "Rogue G", because Candelarios gonna Candelario. You understand.

Anyways, several months ago I was playing around with the exoplanet database when I noticed some of the data is inconsistent with with the known laws of physics. More specifically, Newton's gravitational constant (affectionately referred to in the physics/astronomy/rapper community as "big G") is off by orders of magnitude in a few star systems.

Before calling NASA and Cal Tech, I decided to take a closer look at the data. I saw that the orbital eccentricity of the planets with rogue G values was generally pretty high, and that their orbital periods were conspicuously round numbers. I also noticed that many of these systems were measured using the radial velocity method.

Gliese 317c stands out in particular, with an eccentricity of 0.81 and an orbital period of 10000 days. Gliese 317c's orbital period is exactly 10000 days like my name is Otis. I'm guessing this figure is meant as an upper bound rather than an attempt at pinpoint accuracy. Also, the high orbital eccentricity of Gliese 317c (0.81 is more what you'd expect of a comet orbit) likely makes this a very difficult orbit to measure via the radial velocity method.

Always take a closer look at the data before calling NASA.

But wait, there's more. While looking for clustering in the data, I found this:

Planets in the plot above are colored according to their discovery method. Most of the planets in the exoplanet database were discovered by the Kepler satellite via the transit method. These are colored purple.

In the plot we see all of the exoplanets mostly falling into one big central blob, with the notable exception of the solar system and one other system designated HR 8799, which diverge from the main blob. What makes our home system and HR 8799 stand out from the rest? Looks like the discovery method holds a clue. The only two systems measured through direct imaging are the solar system and HR 8799 (both colored red).

This suggests that our search for exoplanets may be heavily biased by discovery method, but in what way? How exactly do the solar system and HR 8799 differ from the rest of the planets? To get a further clue, I plotted densities of the orbital periods (and semi-major axis lengths) of all the planets, disaggregated by discovery method:

These plots show that the exoplanets discovered by Kepler or the radial velocity method have short periods (and short semi-major axes), whereas direct imaging was able to detect much longer period exoplanets in HR 8799 and our home system.

We are faced with two possible explanations: 1) The current exoplanet database is a fair representation of the total population of exoplanets in the universe, and so the plot above means that us and HR 8799 are very special and unique, long period planet having, star systems. Or 2) The current search for exoplanets is heavily biased by discovery method toward short period planets, and so we are detecting mostly only these short period planets, and missing a ton of long period planets.

Pretty sure that it's 2.

(See my post over at Kaggle for a more in depth analysis and the code behind these graphics: https://www.kaggle.com/benjams/rogue-g-values-and-other-space-oddities/notebook)