Machine learning applied to simulations of collisions between rotating, differentiated planets
Tóm tắt
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 classifying and 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 classification 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, while avoiding the cost of direct simulation. This work is based on a new set of 14,856 SPH simulations of pairwise collisions between rotating, differentiated bodies at all possible mutual orientations.
Từ khóa
Tài liệu tham khảo
Benner, P., Ohlberger, M., Patera, A., Rozza, G., Urban, K.: Model Reduction of Parametrized Systems (2017). https://doi.org/10.1007/978-3-319-58786-8
Benz, W., Slattery, W.L., Cameron, A.G.W.: The origin of the moon and the single-impact hypothesis. I. Icarus 66, 515–535 (1986). https://doi.org/10.1016/0019-1035(86)90088-6
Benz, W., Slattery, W.L., Cameron, A.G.W.: Collisional stripping of Mercury’s mantle. Icarus 74, 516–528 (1988). https://doi.org/10.1016/0019-1035(88)90118-2
Bonati, I., Lichtenberg, T., Bower, D.J., Timpe, M.L., Quanz, S.P.: Direct imaging of molten protoplanets in nearby young stellar associations. Astron. Astrophys. 621, 125 (2019). arXiv:1811.07411. https://doi.org/10.1051/0004-6361/201833158
Bonsor, A., Leinhardt, Z.M., Carter, P.J., Elliott, T., Walter, M.J., Stewart, S.T.: A collisional origin to Earth’s non-chondritic composition? Icarus 247, 291–300 (2015). arXiv:1410.3421. https://doi.org/10.1016/j.icarus.2014.10.019
Brundage, A.L.: Implementation of tillotson equation of state for hypervelocity impact of metals, geologic materials, and liquids. Proc. Eng. 58, 461–470 (2013). https://doi.org/10.1016/j.proeng.2013.05.053
Burger, C., Bazsó, Á., Schäfer, C.M.: Realistic collisional water transport during terrestrial planet formation: self-consistent modeling by an N-body–SPH hybrid code. arXiv e-prints (2019). arXiv:1910.14334
Cambioni, S., Asphaug, E., Emsenhuber, A., Gabriel, T.S.J., Furfaro, R., Schwartz, S.R.: Realistic on-the-fly outcomes of planetary collisions: machine learning applied to simulations of giant impacts. arXiv e-prints (2019). arXiv:1903.04507
Canup, R.M., Asphaug, E.: Origin of the Moon in a giant impact near the end of the Earth’s formation. Nature 412, 708–712 (2001)
Canup, R.M., Asphaug, E.: On an impact origin of Pluto-Charon. In: Mackwell, S., Stansbery, E. (eds.) Lunar and Planetary Science Conference, vol. 34 (2003)
Carter, P.J., Leinhardt, Z.M., Elliott, T., Walter, M.J., Stewart, S.T.: Compositional evolution during rocky protoplanet accretion. Astrophys. J. 813(1), 72 (2015). arXiv:1509.07504. https://doi.org/10.1088/0004-637X/813/1/72
Chambers, J.E.: A hybrid symplectic integrator that permits close encounters between massive bodies. Mon. Not. R. Astron. Soc. 304(4), 793–799 (1999). https://doi.org/10.1046/j.1365-8711.1999.02379.x
Chambers, J.E.: Late-stage planetary accretion including hit-and-run collisions and fragmentation. Icarus 224(1), 43–56 (2013). https://doi.org/10.1016/j.icarus.2013.02.015
Chandrasekhar, S.: Ellipsoidal Figures of Equilibrium (1969)
Chau, A., Reinhardt, C., Helled, R., Stadel, J.: Forming Mercury by giant impacts. Astrophys. J. 865, 35 (2018). arXiv:1808.02448. https://doi.org/10.3847/1538-4357/aad8b0
Chen, T., Guestrin, C.: XGBoost: a scalable tree boosting system. arXiv e-prints (2016). arXiv:1603.02754
Chen, T., Guestrin, C.: Xgboost: a scalable tree boosting system. In: Proceedings of the 22Nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. KDD ’16, pp. 785–794. ACM, New York (2016). https://doi.org/10.1145/2939672.2939785
Efron, B., Hastie, T., Johnstone, I., Tibshirani, R.: Least angle regression. Ann. Stat. 32(2), 407–499 (2004). arXiv:math/0406456. https://doi.org/10.1214/009053604000000067
Elser, S., Moore, B., Stadel, J., Morishima, R.: How common are Earth–Moon planetary systems? Icarus 214(2), 357–365 (2011). arXiv:1105.4616. https://doi.org/10.1016/j.icarus.2011.05.025
Emsenhuber, A., Cambioni, S., Asphaug, E., Gabriel, T.S.J., Schwartz, S.R., Furfaro, R.: Realistic on-the-fly outcomes of planetary collisions II: bringing machine learning to N-body simulations. arXiv e-prints (2020). arXiv:2001.00951
Friedman, J.H.: Greedy function approximation: a gradient boosting machine. Ann. Stat. 29(5), 1189–1232 (2001)
Genda, H., Fujita, T., Kobayashi, H., Tanaka, H., Suetsugu, R., Abe, Y.: Impact erosion model for gravity-dominated planetesimals. Icarus 294, 234–246 (2017). arXiv:1703.03053. https://doi.org/10.1016/j.icarus.2017.03.009
Ghanem, R.G., Spanos, P.D.: Stochastic Finite Elements: A Spectral Approach. Springer, Berlin (1991)
Goodfellow, I., Bengio, Y., Courville, A.: Deep Learning. MIT Press, Cambridge (2016). http://www.deeplearningbook.org
Hamano, K., Abe, Y.: Atmospheric loss and supply by an impact-induced vapor cloud: its dependence on atmospheric pressure on a planet. Earth Planets Space 62(7), 599–610 (2010). https://doi.org/10.5047/eps.2010.06.002
Hands, T.O., Dehnen, W., Gration, A., Stadel, J., Moore, B.: The fate of planetesimal discs in young open clusters: implications for 1I/’Oumuamua, the Kuiper belt, the Oort cloud and more. Mon. Not. R. Astron. Soc. 490(1), 21–36 (2019). arXiv:1901.02465. https://doi.org/10.1093/mnras/stz1069
Inaba, S., Wetherill, G.W., Ikoma, M.: Formation of gas giant planets: core accretion models with fragmentation and planetary envelope. Icarus 166(1), 46–62 (2003). https://doi.org/10.1016/j.icarus.2003.08.001
Kegerreis, J.A., Teodoro, L.F.A., Eke, V.R., Massey, R.J., Catling, D.C., Fryer, C.L., Korycansky, D.G., Warren, M.S., Zahnle, K.J.: Consequences of giant impacts on early Uranus for rotation, internal structure, debris, and atmospheric erosion. Astrophys. J. 861, 52 (2018). arXiv:1803.07083. https://doi.org/10.3847/1538-4357/aac725
Knabenhans, M., Stadel, J., Marelli, S., Potter, D., Teyssier, R., Legrand, L., Schneider, A., Sudret, B., Blot, L., Awan, S., Burigana, C., Carvalho, C.S., Kurki-Suonio, H., Sirri, G.: Euclid preparation: II. The EUCLIDEMULATOR—a tool to compute the cosmology dependence of the nonlinear matter power spectrum. Mon. Not. R. Astron. Soc. 484, 5509–5529 (2019). arXiv:1809.04695. https://doi.org/10.1093/mnras/stz197
Kobayashi, H., Tanaka, H.: Fragmentation model dependence of collision cascades. Icarus 206(2), 735–746 (2010). https://doi.org/10.1016/j.icarus.2009.10.004
Kokubo, E., Genda, H.: Formation of terrestrial planets from protoplanets under a realistic accretion condition. Astrophys. J. 714(1), 21–25 (2010). arXiv:1003.4384. https://doi.org/10.1088/2041-8205/714/1/L21
Le Gratiet, L., Marelli, S., Sudret, B.: Metamodel-based sensitivity analysis: polynomial chaos expansions and Gaussian processes. In: Ghanem, R., Higdon, D., Owhadi, H. (eds.) Handbook on Uncertainty Quantification, Chap. 8. Springer, Berlin (2016). https://doi.org/10.1007/978-3-319-11259-6_3
Leinhardt, Z.M., Richardson, D.C.: Planetesimals to protoplanets. I. Effect of fragmentation on terrestrial planet formation. Astrophys. J. 625(1), 427–440 (2005). https://doi.org/10.1086/429402
Leinhardt, Z.M., Stewart, S.T.: Collisions between gravity-dominated bodies. I. Outcome regimes and scaling laws. Astrophys. J. 745(1), 79 (2012). arXiv:1106.6084. https://doi.org/10.1088/0004-637X/745/1/79
Liu, S.-F., Hori, Y., Müller, S., Zheng, X., Helled, R., Lin, D., Isella, A.: The formation of Jupiter’s diluted core by a giant impact. Nature 572(7769), 355–357 (2019). https://doi.org/10.1038/s41586-019-1470-2
Lundberg, S.M., Erion, G.G., Lee, S.-I.: Consistent individualized feature attribution for tree ensembles. arXiv e-prints (2018). arXiv:1802.03888
Lundberg, S.M., Lee, S.-I.: A unified approach to interpreting model predictions. In: Proceedings of the 31st International Conference on Neural Information Processing Systems. NIPS ’17, pp. 4768–4777. Curran Associates, Red Hook (2017)
Marcus, R.A., Sasselov, D., Stewart, S.T., Hernquist, L.: Water/icy super-earths: giant impacts and maximum water content. Astrophys. J. 719(1), 45–49 (2010). arXiv:1007.3212. https://doi.org/10.1088/2041-8205/719/1/L45
Marelli, S., Lamas, C., Sudret, B., Konakli, K.: UQLab user manual—sensitivity analysis. Technical report, Chair of Risk, Safety & Uncertainty Quantification, ETH, Zurich, Zurich (2017). https://doi.org/10.13140/RG.2.1.1419.4404
Marelli, S., Sudret, B.: UQLab: a framework for uncertainty quantification in Matlab. In: Vulnerability, Uncertainty, and Risk (Proc. 2nd Int. Conf. on Vulnerability, Risk Analysis and Management (ICVRAM2014)), Liverpool, United Kingdom, pp. 2554–2563. American Society of Civil Engineers (2014). https://doi.org/10.1061/9780784413609.257
Marelli, S., Sudret, B.: UQLab user manual—polynomial chaos expansion. Technical report, Chair of Risk, Safety & Uncertainty Quantification, ETH Zurich, Zurich (2017). https://doi.org/10.13140/RG.2.1.3778.7366
McKay, M.D., Beckman, R.J., Conover, W.J.: A comparison of three methods for selecting values of input variables in the analysis of output from a computer code. Technometrics 21(2), 239–245 (1979)
Meier, T., Reinhardt, C., Stadel, J., Timpe, M.L.: (2020, in prep.)
Morishima, R., Stadel, J., Moore, B.: From planetesimals to terrestrial planets: N-body simulations including the effects of nebular gas and giant planets. Icarus 207, 517–535 (2010). arXiv:1007.0579. https://doi.org/10.1016/j.icarus.2009.11.038
Potter, D., Stadel, J., Teyssier, R.: PKDGRAV3: beyond trillion particle cosmological simulations for the next era of galaxy surveys. Comput. Astrophys. Cosmol. 4(1), 2 (2017). arXiv:1609.08621. https://doi.org/10.1186/s40668-017-0021-1
Quintana, E.V., Barclay, T., Borucki, W.J., Rowe, J.F., Chambers, J.E.: The frequency of giant impacts on Earth-like worlds. Astrophys. J. 821(2), 126 (2016). arXiv:1511.03663. https://doi.org/10.3847/0004-637X/821/2/126
Quintana, E.V., Lissauer, J.J.: VizieR Online Data Catalog: simulations of the late stage of planet formation (Quintana+, 2014). VizieR Online Data Catalog, 786–33 (2017)
Rasmussen, C.E., Williams, C.K.I.: Gaussian Processes for Machine Learning (Adaptive Computation and Machine Learning). MIT Press, Cambridge (2005)
Reinhardt, C., Chau, A., Stadel, J., Helled, R.: Bifurcation in the history of Uranus and Neptune: the role of giant impacts. arXiv e-prints (2019). arXiv:1907.09809
Reinhardt, C., Stadel, J.: Numerical aspects of giant impact simulations. Mon. Not. R. Astron. Soc. 467, 4252–4263 (2017). arXiv:1701.08296. https://doi.org/10.1093/mnras/stx322
Roth, A.E. (ed.): The Shapley Value: Essays in Honor of Lloyd S. Shapley. Cambridge University Press, Cambridge (1988). https://doi.org/10.1017/CBO9780511528446
Rumelhart, D.E., Hinton, G.E., Williams, R.J.: Learning representations by back-propagating errors. Nature 323(6088), 533–536 (1986). https://doi.org/10.1038/323533a0
Snyman, J.: Practical Mathematical Optimization: An Introduction to Basic Optimization Theory and Classical and New Gradient-Based Algorithms. Applied Optimization. Springer, Berlin (2005). https://books.google.ch/books?id=0tFmf_UKl7oC
Sobol’, I.M.: Sensitivity estimates for nonlinear mathematical models. Math. Model. Comput. Exp. 1, 407–414 (1993)
Stadel, J.G.: Cosmological N-body simulations and their analysis. Ph.D. thesis, University of Washington (2001)
Stewart, S.T., Leinhardt, Z.M.: Collisions between gravity-dominated bodies. II. The diversity of impact outcomes during the end stage of planet formation. Astrophys. J. 751(1), 32 (2012). arXiv:1109.4588. https://doi.org/10.1088/0004-637X/751/1/32
Tillotson, J.H.: Metallic equations of state for hypervelocity impact. Gen. At. 5, 0–141 (1962)
Torre, E., Marelli, S., Embrechts, P., Sudret, B.: Data-driven polynomial chaos expansion for machine learning regression. J. Comput. Phys. 388, 601–623 (2019). arXiv:1808.03216. https://doi.org/10.1016/j.jcp.2019.03.039
Valencia, D., Paracha, E., Jackson, A.P.: Can a machine learn the outcome of planetary collisions? arXiv e-prints (2019). arXiv:1902.04052
von Hoerner, S.: Die numerische Integration des n-Körper-Problemes für Sternhaufen. I. Z. Astrophys. 50, 184–214 (1960)
Wadsley, J.W., Stadel, J., Quinn, T.: Gasoline: a flexible, parallel implementation of TreeSPH. New Astron. 9, 137–158 (2004). arXiv:astro-ph/0303521. https://doi.org/10.1016/j.newast.2003.08.004
Walsh, K.J., Morbidelli, A., Raymond, S.N., O’Brien, D.P., Mandell, A.M.: A low mass for Mars from Jupiter’s early gas-driven migration. Nature 475(7355), 206–209 (2011). arXiv:1201.5177. https://doi.org/10.1038/nature10201
Wang, G.G.: Adaptive response surface method using inherited Latin hypercube design points. J. Mech. Des. 125(2), 210–220 (2003). https://doi.org/10.1115/1.1561044
Wilhelms, D.E., Squyres, S.W.: The martian hemispheric dichotomy may be due to a giant impact. Nature 309, 138–140 (1984). https://doi.org/10.1038/309138a0
Xiu, D., Karniadakis, G.E.: The Wiener-Askey polynomial chaos for stochastic differential equations. SIAM J. Sci. Comput. 24(2), 619–644 (2002). https://doi.org/10.1137/S1064827501387826