A semi-analytical surrogate model for giant planet evolution: bypassing ordinary differential equation solvers with localised thermodynamics, softplus asymptotes, and B-spline photometry
Abstract
Context. Evolutionary models translate the observable luminosity, temperature, and colours of giant planets and brown dwarfs into mass and age. Generating their cooling tracks normally requires integrating the internal energy over time with an ordinary differential equation (ODE) solver coupled to pre-computed atmospheric grids, which becomes numerically stiff at sharp transitions such as cloud condensation and the onset of electron degeneracy, and is fragile inside Bayesian retrievals. Aims. We aim to generate continuous cooling tracks and photometric light curves directly from discrete atmospheric grids, without an ODE solver. Methods. We mapped the grids into a logarithmic thermodynamic space and extracted localised surrogate models with Gaussian-weighted regressions at fixed planetary parameters. We fitted the entropy and cooling rate against the internal temperature using bounded piecewise softplus functions to capture structural and cooling-rate transitions; the radius was fitted on the same temperature axis; and band photometry was represented with fixed-knot cubic B-splines. The age followed from numerical integration of these analytic functions, and uncertainties were propagated from the residual scatter of each fit. Results. The surrogate, CoolTrack, reproduces the transition into electron degeneracy and the L-to-T spectral-type transition in the colour-magnitude diagram, converges on Solar System benchmarks, and evaluates a full evolutionary track in milliseconds on a standard desktop CPU. Conclusions. By removing the forward-modelling bottleneck, CoolTrack is suitable for direct use in Bayesian retrieval pipelines, where the age, mass, and formation entropy of a planet can be inferred jointly with its atmospheric properties.
Related
- Photometry and analysis of comet 220P/McNaught during the 2026 outbursts
- A new method to cleanly separate hot and cool populations using just Gaia and 2MASS photometry
- A Uniform Census of Variable Stars in M101 Based on HST/ACS Photometry
- Hunting for Compact Object Binaries from eRASS1 Optical Counterparts through ZTF Time-domain Photometry and Multi-wavelength Surveys
- Hunting for Compact Object Binaries from eRASS1 Optical Counterparts through ZTF Time-domain Photometry and Multiwavelength Surveys
BibTeX
@misc{wilkinson2026semianalytical,
title = {{A semi-analytical surrogate model for giant planet evolution: bypassing ordinary differential equation solvers with localised thermodynamics, softplus asymptotes, and B-spline photometry}},
author = {Christian Wilkinson and Jonas Wehrung-Montpezat and Benjamin Charnay and Mathilde Mâlin and Lukas Delaye and Anne-Marie Lagrange and Vito Squicciarini and Johan Mazoyer and Stéphane Mazevet and Baptiste Perrier},
howpublished = {arXiv},
year = {2026},
month = aug,
doi = {10.48550/arxiv.2608.29421},
eprint = {2608.29421},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.29421},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - A semi-analytical surrogate model for giant planet evolution: bypassing ordinary differential equation solvers with localised thermodynamics, softplus asymptotes, and B-spline photometry
AU - Christian Wilkinson
AU - Jonas Wehrung-Montpezat
AU - Benjamin Charnay
AU - Mathilde Mâlin
AU - Lukas Delaye
AU - Anne-Marie Lagrange
AU - Vito Squicciarini
AU - Johan Mazoyer
AU - Stéphane Mazevet
AU - Baptiste Perrier
PY - 2026
DA - 2026/08/29
PB - arXiv
DO - 10.48550/arxiv.2608.29421
UR - https://arxiv.org/abs/2608.29421
AB - Context. Evolutionary models translate the observable luminosity, temperature, and colours of giant planets and brown dwarfs into mass and age. Generating their cooling tracks normally requires integrating the internal energy over time with an ordinary differential equation (ODE) solver coupled to pre-computed atmospheric grids, which becomes numerically stiff at sharp transitions such as cloud condensation and the onset of electron degeneracy, and is fragile inside Bayesian retrievals. Aims. We aim to generate continuous cooling tracks and photometric light curves directly from discrete atmospheric grids, without an ODE solver. Methods. We mapped the grids into a logarithmic thermodynamic space and extracted localised surrogate models with Gaussian-weighted regressions at fixed planetary parameters. We fitted the entropy and cooling rate against the internal temperature using bounded piecewise softplus functions to capture structural and cooling-rate transitions; the radius was fitted on the same temperature axis; and band photometry was represented with fixed-knot cubic B-splines. The age followed from numerical integration of these analytic functions, and uncertainties were propagated from the residual scatter of each fit. Results. The surrogate, CoolTrack, reproduces the transition into electron degeneracy and the L-to-T spectral-type transition in the colour-magnitude diagram, converges on Solar System benchmarks, and evaluates a full evolutionary track in milliseconds on a standard desktop CPU. Conclusions. By removing the forward-modelling bottleneck, CoolTrack is suitable for direct use in Bayesian retrieval pipelines, where the age, mass, and formation entropy of a planet can be inferred jointly with its atmospheric properties.
SN - 0004-6361
SN - 1432-0746
ER -