# Examining the Radius Valley: a Machine Learning Approach

## TL;DR

This study confirms the existence of the radius valley in exoplanets using a large dataset and characterizes it as a negative power law slope, supporting photoevaporation theories over gas-poor formation models.

## Contribution

It provides a comprehensive analysis of the radius valley with a larger sample, confirming its power-law nature and quantifying its slope across multiple methods.

## Key findings

- Radius valley confirmed with larger sample
- Slope of the valley is approximately -0.319
- Results support photoevaporation as the main process

## Abstract

The ''radius valley" is a relative dearth of planets between two potential populations of exoplanets, super-Earths and mini-Neptunes. This feature appears in examining the distribution of planetary radii, but has only ever been characterized on small samples. The valley could be a result of photoevaporation, which has been predicted in numerous theoretical models, or a result of other processes. Here, we investigate the relationship between planetary radius and orbital period through 2-dimensional kernel density estimator and various clustering methods, using all known super-Earths ($R<4.0R_E$). With our larger sample, we confirm the radius valley and characterize it as a power law. Using a variety of methods, we find a range of slopes that are consistent with each other and distinctly negative. We average over these results and find the slope to be $m=-0.319^{+0.088}_{-0.116}$. We repeat our analysis on samples from previous studies. For all methods we use, the resulting line has a negative slope, which is consistent with models of photoevaporation and core-powered mass loss but inconsistent with planets forming in a gas-poor disk.