Boosting decision trees for the selection of main belt asteroids in planetary ephemerides: An alternative model

¹ Université Côte d’Azur, Observatoire de la Côte d’Azur, CNRS, Geoazur, 250 avenue A. Einstein, 06560 Valbonne, France
² CRAS, Sapienza University of Rome, via Eudossiana 18, 00184 Rome, Italy
³ LTE, Observatoire de Paris, Université PSL, Sorbonne Université, Univ. Lille, Laboratoire National de Métrologie et d’Essai, CNRS, 75014 Paris, France

^★ Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it. ; This email address is being protected from spambots. You need JavaScript enabled to view it.

Received: 11 May 2025
Accepted: 27 October 2025

Abstract

Context. One of the main bottlenecks in assessing the accuracy of the Mars orbit is the unknown value of the asteroids in the main asteroid belt. A modeling with 343 asteroids as point masses is used today, with the relative masses fit to observational data.

Aims. We propose an innovative method for reducing the number of asteroids implemented as point masses, which in turn reduces the number of parameters to be fit without a significant degradation of the postfit residuals.

Methods. We used boosting decision trees to obtain a ranking by relative importance of the 343 asteroids of the current main belt modeling.

Results. We were able to remove more than 100 of these asteroids without significantly degrading the postfit residuals and with a significant improvement of the uncertainties. Furthermore, we verified that the postfit masses found with the new modeling (INPOP25c) are consistent with respect to the mass estimation from an independent approach by means of the albedo properties of the asteroids.