# Can a machine learn the outcome of planetary collisions?

## TL;DR

This paper explores the use of machine learning to predict the outcomes of planetary collisions, aiming to improve upon existing models used in planetary formation simulations.

## Contribution

It evaluates three machine learning methods for collision outcome prediction and demonstrates that gradient boosting regression trees perform best, with potential for guiding future simulations.

## Key findings

- Gradient boosting regression trees achieved the best prediction accuracy.
- Ensembling multiple algorithms slightly improved results.
- Gaussian processes identified key parameter regions for efficient data collection.

## Abstract

Planetary-scale collisions are common during the last stages of formation of solid planets, including the Solar system terrestrial planets. The problem of growing planets has been divided into studying the gravitational interaction of embryos relevant in million year timescales and treated with N-body codes and the collision between objects with a timescale of hours to days and treated with smoothed-particle hydrodynamics. These are now being coupled with simple parameterized models. We set out to investigate if machine learning techniques can offer a better solution by predicting the outcome of collisions which can then be used in N-body simulations. We considered three different supervised machine learning approaches: gradient boosting regression trees, nested models, and gaussian processes. We found that the former produced the best results, and that it was slightly surpassed by ensembling different algorithms. With gaussian processes, we found the regions of parameter space that may yield the most information to machine learning algorithms. Thus, we suggest SPH calculations to focus first on mass ratios above 0.5.