2021/04/26 by Kevin Hoffman, Hoffman, Kevin, Jae Yoon Sung +3
Chemistry · Engineering · #Advanced Thermodynamic Systems and Engines #Earth and Planetary Astrophysics (astro-ph.EP) #FOS: Computer and information sciences #FOS: Physical sciences #Instrumentation and Methods for Astrophysics (astro-ph.IM) #Machine Learning (cs.LG) #Molecular Spectroscopy and Structure #Spacecraft and Cryogenic Technologies
paper · pdf · doi:10.48550/arxiv.2104.12845
openalex publication_date 2021/04/26 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
In the current paradigm of planet formation research, it is believed that the\nfirst step to forming massive bodies (such as asteroids and planets) requires\nthat small interstellar dust grains floating through space collide with each\nother and grow to larger sizes. The initial formation of these pebbles is\ngoverned by an integro-differential equation known as the Smoluchowski\ncoagulation equation, to which analytical solutions are intractable for all but\nthe simplest possible scenarios. While brute-force methods of approximation\nhave been developed, they are computationally costly, currently making it\ninfeasible to simulate this process including other physical processes relevant\nto planet formation, and across the very large range of scales on which it\noccurs. In this paper, we take a machine learning approach to designing a\nsystem for a much faster approximation. We develop a multi-output random forest\nregression model trained on brute-force simulation data to approximate\ndistributions of dust particle sizes in protoplanetary disks at different\npoints in time. The performance of our random forest model is measured against\nthe existing brute-force models, which are the standard for realistic\nsimulations. Results indicate that the random forest model can generate highly\naccurate predictions relative to the brute-force simulation results, with an\nR2 of 0.97, and do so significantly faster than brute-force methods.\n