A Bayesian neural network predicts the dissolution of compact planetary systems
Princeton University(普林斯顿大学);University of Toronto at Scarborough(多伦多大学士嘉堡分校);University of Toronto(多伦多大学);DeepMind;Center for Astrophysics — Harvard & Smithsonian(哈佛-史密松天体物理中心);Stony Brook University(石溪大学);Flatiron Institute(弗莱特海伦研究所);Carnegie Mellon University(卡内基梅隆大学)
Miles Cranmer, Daniel Tamayo, Hanno Rein, Peter Battaglia, Samuel Hadden, Philip J. Armitage, Shirley Ho, David N. Spergel
Despite over three hundred years of effort, no solutions exist for predicting when a general planetary configuration will become unstable. We introduce a deep learning architecture to push forward this problem for compact systems. While current machine learning algorithms in this area rely on scientist-derived instability metrics, our new technique learns its own metrics from scratch, enabled by a novel internal structure inspired from dynamics theory. Our Bayesian neural network model can accurately predict not only if, but also when a compact planetary system with three or more planets will go unstable. Our model, trained directly from short N-body time series of raw orbital elements, is more than two orders of magnitude more accurate at predicting instability times than analytical estimators, while also reducing the bias of existing machine learning algorithms by nearly a factor of three. Despite being trained on compact resonant and near-resonant three-planet configurations, the model demonstrates robust generalization to both non-resonant and higher multiplicity configurations, in the latter case outperforming models fit to that specific set of integrations. The model computes instability estimates up to five orders of magnitude faster than a numerical integrator, and unlike previous efforts provides confidence intervals on its predictions. Our inference model is publicly available in the SPOCK package, with training code open-sourced.
Multi-layer dissolution exponential-family models for weighted signed networks
University College Dublin(都柏林大学学院);University of Trieste(的里雅斯特大学);Institute for Maternal and Child Health IRCCS Burlo Garofolo(Burlo Garofolo 儿童与母婴健康研究所);Rinn AI
Dynamic sparse graphs with overlapping communities
University of Cyprus(塞浦路斯大学);Imperial College London(伦敦帝国学院);Sapienza University of Rome(罗马第一大学);London School of Economics and Political Science(伦敦政治经济学院)
Towards Automated Satellite Conjunction Management with Bayesian Deep Learning
University of Oxford(牛津大学);University of Strathclyde(斯特拉斯克莱德大学);Technische Universität Darmstadt(达姆施塔特工业大学);IBM(国际商业机器公司);Imperial College London(伦敦帝国学院);European Space Agency(欧洲空间局);University of Surrey(萨里大学)
Understanding Exoplanet Habitability: A Bayesian ML Framework for Predicting Atmospheric Absorption Spectra
1 Department of Information Science, University at Albany, Albany, NY 12222, USA 2 Department of Physics, University at Albany, Albany, NY 12222, USA 3 NASA Goddard Institute for Space Studies, New York, NY 10025, USA 4 Theoretical Astrophysics, Department of Physics \& Astronomy, Uppsala University, 75105 Uppsala, Sweden
The HARPS search for southern extra-solar planets. XXXVIII. Bayesian re-analysis of three systems. New super-Earths, unconfirmed signals, and magnetic cycles
A Bayesian Periodogram Finds Evidence for Three Planets in 47 Ursae Majoris
A Bayesian analysis of 47 Ursae Majoris (47 UMa) radial velocity data confirms and refines the properties of two previously reported planets with periods of 1079 and 2325 days and finds evidence for an additional long period planet with a period of approximately 10000 days. The three planet model is found to be 10^5 times more probable than the next most probable model which is a two planet model. The nonlinear model fitting is accomplished with a new hybrid Markov chain Monte Carlo (HMCMC) algorithm which incorporates parallel tempering, simulated annealing and genetic crossover operations. E…
Convolutional neural networks as an alternative to Bayesian retrievals
Kapteyn Astronomical Institute, University of Groningen(格罗宁根大学卡普坦天文研究所);Netherlands Space Research Institute (SRON)(荷兰空间研究所(SRON));Centre for Exoplanet Science, University of Edinburgh(爱丁堡大学系外行星科学中心);School of GeoSciences, University of Edinburgh(爱丁堡大学地球科学学院)
CORALIE radial-velocity search for companions around evolved stars (CASCADES) V. Three planetary companions and achievable precision
Instituto de Astrofísica de Andalucía-CSIC(安达卢西亚天体物理研究所);Observatoire Astronomique de l’Université de Genève(日内瓦大学天文台);Instituut voor Sterrenkunde, KU Leuven(荷语鲁汶大学天文研究所);Space sciences, Technologies and Astrophysics Research(空间科学、技术与天体物理研究)
Instituto de Ciencias Físicas (CONICET / ECyT-UNSAM)(物理科学研究所(CONICET / 国家萨卡特卡斯大学));Universidad Católica de la Santísima Concepción(神圣感知天主教大学);Instituto Tecnológico de Buenos Aires (ITBA)(布宜诺斯艾利斯理工学院);Institute for Particle Physics and Astrophysics, ETH Zürich(苏黎世联邦理工学院粒子物理与天体物理研究所);Univ. Grenoble Alpes, CNRS, IPAG(格勒诺布尔阿尔卑斯大学,法国国家科学研究中心,IPAG);Observatoire de Genève, Département d’Astronomie, Université de Genève(日内瓦天文台,天文学系,日内瓦大学);SUPA, School of Physics and Astronomy, University of St Andrews(苏格兰物理与天文学联盟,圣安德鲁斯大学物理与天文学院)
Mass Production of 2023 KMTNet Microlensing Planets. III: Three Planets from the Subprime Field
Westlake University(西湖大学);Tsinghua University(清华大学);Korea Astronomy and Space Science Institute(韩国天文学与空间科学研究所);Astronomical Observatory, University of Warsaw(华沙大学天文台);Department of Earth and Space Science, Graduate School of Science, Osaka University(大阪大学理学研究科地球惑星科学专攻);Westlake Institute for Advanced Study(西湖高等研究院);University of Canterbury, School of Physical and Chemical Sciences(坎特伯雷大学物理与化学科学学院)