Hybrid computational methods for planetary ring dynamics: statistical data analysis and machine learning for predictive modeling of post-impact material
Carregando...
Data
Autores
Supervisor
Coorientador
Pós-graduação
Curso de graduação
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (UNESP)
Tipo
Relatório de pós-doc
Direito de acesso
Acesso restrito
Resumo
Since the discovery of Neptune’s arcs in the Adams ring, several models have been proposed to explain
these structures (e.g. Hubbard et al. (1986), Hedman et al. (2007b), and Lattari (2019)). However,
further efforts are still needed to sufficiently resolve the arcs’ production mechanisms to explain their
density, which is consistent with their brightness. Although collision models using N-body integrators
provide partial insight, they often overlook critical internal physical properties of the colliding bodies.
In contrast, Smoothed Particle Hydrodynamics (SPH) models offer a more suitable approach because
they account for mutual interactions and provide better computational resolution. Unfortunately, SPH
simulations are computationally expensive and require significant runtime.
This research project will use the SPH method, specifically via the miluphcuda code, to model
collisions between macroscopic bodies in planetary rings, aiming to generate realistic data to explain
the arcs’ structure. The code will be modified to improve usability, allowing a wider range of planetary
ring studies. These simulations will be combined with N-body simulations to study ring stability and
dynamics, integrating analytical models to estimate dust production rates and the lifespans of ring
objects before catastrophic collisions.
Building on previous results from hybrid methods that incorporate shock propagation, material
modification, and gravitational reaccumulation, this project will also develop a machine-learningbased surrogate model for N-body simulations. This model will provide fast and reliable predictions
for the masses and velocities resulting from collisions, as a function of the masses of the colliding
bodies, the impact velocity and the angle. By combining both simulations with machine learning, the
hybrid method significantly reduces computation time while maintaining a realistic representation of
all processes in the dynamics of the planetary rings.
Descrição
Palavras-chave
Idioma
Inglês
Citação
SIQUEIRA, Patrícia Buzzatto. Hybrid computational methods for planetary ring dynamics: statistical data analysis and machine learning for predictive modeling of post-impact material. 2026. Relatório científico (Pós-Doutorado) - Faculdade de Engenharia e Ciências, Universidade Estadual Paulista, Guaratinguetá, 2026.
