Benchmarking Machine Learning Emulators of Stellar Evolution for Precision Asteroseismology
- 1Yale University
Abstract
Fast and accurate stellar evolution emulators—surrogate models that approximate expensive simulation outputs with machine learning (ML)—are powerful tools for modern stellar characterization, hierarchical inference, and population synthesis. We analyze the grid density required for reliable emulation by training ML algorithms on main-sequence models with masses M ∈ [0.7, 1.2] M ⊙ . This range is challenging to emulate due to rapidly varying evolutionary behavior caused by the radiative-to-convective core transition, as well as the requirement to match the part-per-thousand seismic precision that has been delivered for such stars from the NASA Kepler mission. Generating grids from analytical models, as well as MESA , YREC , MIST , and ASTEC , we compare linear interpolation, k-nearest neighbors, random forests, and neural networks (NNs) in interpolating the stellar observables T eff , L , Δ ν , and ν max . While NNs outperform other methods, sparse grids induce localized failures in the core-transition region, resulting in unstable derivatives, ensemble disagreement, and fragmented posterior distributions during inference. Performance gains from denser grids are nonuniform, suggesting that adaptive grid generation should be favored over uniform refinement. Finally, we show that NN ensembles allow for localized uncertainty propagation, more accurately reflecting emulator reliability across parameter space than global uncertainty estimates. As we consider only the 2D case of varying only stellar mass and age along the main sequence, these results represent a lower bound on the challenge in emulating stellar evolution simulations for precision asteroseismology.
Related
- Simulation-based inference of profile evolution for high-precision pulsar timing
- Astro2Method: Benchmarking LLMs on Machine Learning Method Selection in Astronomy
- Asteroseismology and interferometry of the F7V spectroscopic binary χ Draconis A in the TESS CVZ
- EPRV asteroseismology of the K dwarfs GJ 570 A and GJ 783 A with ESPRESSO. Empirical mode-amplitude scaling in tension with 3D convection simulations
- Accelerating Radiative Transfer in Planetary Atmospheres by Orders of Magnitude with Neural Network Emulators
BibTeX
@article{gluck2026benchmarking,
title = {{Benchmarking Machine Learning Emulators of Stellar Evolution for Precision Asteroseismology}},
author = {Naomi Gluck and Earl P. Bellinger and Yan Liang and Ebraheem Farag and Nicholas Saunders and Selim Kalici and Christopher J. Lindsay and Sarbani Basu},
journal = {The Astrophysical Journal},
year = {2026},
month = oct,
doi = {10.3847/1538-4357/aea14e},
url = {https://iopscience.iop.org/article/10.3847/1538-4357/aea14e},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Benchmarking Machine Learning Emulators of Stellar Evolution for Precision Asteroseismology
AU - Naomi Gluck
AU - Earl P. Bellinger
AU - Yan Liang
AU - Ebraheem Farag
AU - Nicholas Saunders
AU - Selim Kalici
AU - Christopher J. Lindsay
AU - Sarbani Basu
PY - 2026
DA - 2026/10/05
JO - The Astrophysical Journal
DO - 10.3847/1538-4357/aea14e
UR - https://iopscience.iop.org/article/10.3847/1538-4357/aea14e
AB - Fast and accurate stellar evolution emulators—surrogate models that approximate expensive simulation outputs with machine learning (ML)—are powerful tools for modern stellar characterization, hierarchical inference, and population synthesis. We analyze the grid density required for reliable emulation by training ML algorithms on main-sequence models with masses M ∈ [0.7, 1.2] M ⊙ . This range is challenging to emulate due to rapidly varying evolutionary behavior caused by the radiative-to-convective core transition, as well as the requirement to match the part-per-thousand seismic precision that has been delivered for such stars from the NASA Kepler mission. Generating grids from analytical models, as well as MESA , YREC , MIST , and ASTEC , we compare linear interpolation, k-nearest neighbors, random forests, and neural networks (NNs) in interpolating the stellar observables T eff , L , Δ ν , and ν max . While NNs outperform other methods, sparse grids induce localized failures in the core-transition region, resulting in unstable derivatives, ensemble disagreement, and fragmented posterior distributions during inference. Performance gains from denser grids are nonuniform, suggesting that adaptive grid generation should be favored over uniform refinement. Finally, we show that NN ensembles allow for localized uncertainty propagation, more accurately reflecting emulator reliability across parameter space than global uncertainty estimates. As we consider only the 2D case of varying only stellar mass and age along the main sequence, these results represent a lower bound on the challenge in emulating stellar evolution simulations for precision asteroseismology.
SN - 0004-637X
SN - 1538-4357
ER -