Machine learning applied to simulations of collisions between rotating, differentiated planets

Abstract

In the late stages of terrestrial planet formation, pairwise collisions between planetary-sized bodies act as the fundamental agent of planet growth. These collisions can lead to either growth or disruption of the bodies involved and are largely responsible for shaping the final characteristics of the planets. Despite their critical role in planet formation, an accurate treatment of collisions has yet to be realized. While semi-analytic methods have been proposed, they remain limited to a narrow set of post-impact properties and have only achieved relatively low accuracies. However, the rise of machine learning and access to increased computing power have enabled novel data-driven approaches. In this work, we show that data-driven emulation techniques are capable of predicting the outcome of collisions with high accuracy and are generalizable to any quantifiable post-impact quantity. In particular, we focus on the dataset requirements, training pipeline, and regression performance for four distinct data-driven techniques from machine learning (ensemble methods and neural networks) and uncertainty quantification (Gaussian processes and polynomial chaos expansion). We compare these methods to existing analytic and semi-analytic methods. Such data-driven emulators are poised to replace the methods currently used in N-body simulations. This work is based on a new set of 10,700 SPH simulations of pairwise collisions between rotating, differentiated bodies at all possible mutual orientations.

arXiv comment28 pages, 8 figures. Submitted to Computational Astrophysics and Cosmology

References in corpus (40)

• XGBoost: A Scalable Tree Boosting System
• Rejoinder to "Least angle regression" by Efron et al
• Discussion of "Least angle regression" by Efron et al
• Discussion of "Least angle regression" by Efron et al
• Least Angle Regression
• Discussion of "Least angle regression" by Efron et al
• Discussion of "Least angle regression" by Efron et al
• Discussion of "Least angle regression" by Efron et al
• Discussion of "Least angle regression" by Efron et al
• Discussion of "Least angle regression" by Efron et al
• Discussion of "Least angle regression" by Efron et al
• A Unified Approach to Interpreting Model Predictions
• A low mass for Mars from Jupiter's early gas-driven migration
• Gasoline: An adaptable implementation of TreeSPH
• Consistent Individualized Feature Attribution for Tree Ensembles
• Collisions between Gravity-Dominated Bodies: 2. The Diversity of Impact Outcomes during the End Stage of Planet Formation
• Collisions Between Gravity-Dominated Bodies: 1. Outcome Regimes and Scaling Laws
• Euclid preparation: II. The EuclidEmulator -- A tool to compute the cosmology dependence of the nonlinear matter power spectrum
• Data-driven polynomial chaos expansion for machine learning regression
• From planetesimals to terrestrial planets: N-body simulations including the effects of nebular gas and giant planets
• Formation of Terrestrial Planets from Protoplanets under a Realistic Accretion Condition
• Metamodel-based sensitivity analysis: Polynomial chaos expansions and Gaussian processes
• The Frequency of Giant impacts on Earth-like Worlds
• The Formation of Jupiter's Diluted Core by a Giant Impact
• Compositional evolution during rocky protoplanet accretion
• A Collisional Origin to Earth's Non-chondritic Composition?
• Water/Icy Super-Earths: Giant Impacts and Maximum Water Content
• Forming Mercury by Giant Impacts
• Consequences of Giant Impacts on Early Uranus for Rotation, Internal Structure, Debris, and Atmospheric Erosion
• Bifurcation in the history of Uranus and Neptune: the role of giant impacts
• How common are Earth-Moon planetary systems?
• The Allen Telescope Array Twenty-centimeter Survey - A 690-Square-Degree, 12-Epoch Radio Dataset - I: Catalog and Long-Duration Transient Statistics
• Numerical aspects of Giant Impact simulations
• The fate of planetesimal discs in young open clusters: implications for 1I/'Oumuamua, the Kuiper belt, the Oort cloud and more
• Realistic On-the-fly Outcomes of Planetary Collisions II: Bringing Machine Learning to N-body Simulations
• Realistic On-The-Fly Outcomes of Planetary Collisions: Machine Learning Applied to Simulations of Giant Impacts
• Direct imaging of molten protoplanets in nearby young stellar associations
• Realistic collisional water transport during terrestrial planet formation: Self-consistent modeling by an N-body--SPH hybrid code
• Impact Erosion Model for Gravity-Dominated Planetesimals
• Can a machine learn the outcome of planetary collisions?

Cited by in corpus (11)

• The terrestrial planet formation paradox inferred from high-resolution N-body simulations
• Forming Iron-rich Planets with Giant Impacts
• Collision Chains among the Terrestrial Planets. III. Formation of the Moon
• The Role of Giant Impacts in Planet Formation
• The EOS/Resolution Conspiracy: Convergence in Proto-Planetary Collision Simulations
• Collision Chains among the Terrestrial Planets. II. An Asymmetry between Earth and Venus
• Could Uranus and Neptune form by collisions of planetary embryos?
• Residual Neural Networks for the Prediction of Planetary Collision Outcomes
• A Systematic Survey of Moon-Forming Giant Impacts. II. Rotating bodies
• Protoplanet collisions: new scaling laws from SPH simulations
• GENGA II: GPU planetary N-body simulations with non-Newtonian forces and high number of particles