$\text{Exoformer}$: Accelerating Bayesian atmospheric retrievals with transformer neural networks

Abstract

Computationally expensive and time-consuming Bayesian atmospheric retrievals pose a significant bottleneck for the rapid analysis of high-quality exoplanetary spectra from present and next generation space telescopes, such as JWST and Ariel. As these missions demand more complex atmospheric models to fully characterize the spectral features they uncover, they will benefit from data-driven analysis techniques such as machine and deep learning. We introduce and detail a novel approach that uses a transformer-based neural network ($\text{Exoformer}$) to rapidly generate informative prior distributions for atmospheric transmission spectra of hot Jupiters. We demonstrate the effectiveness of $\text{Exoformer}$ using both simulated observations and real JWST data of WASP-39b and WASP-17b within the TauREx retrieval framework, leveraging the nested sampling algorithm. By replacing standard uniform priors with $\text{Exoformer}$-derived informative priors, our method accelerates nested-sampling retrievals by factor of 3-8 in the tested cases, while preserving the retrieved parameters and best-fit spectra. Crucially, we ensure that the retrieved parameters and the best-fit models remain consistent with results from classical methods. Furthermore, we confirm the statistical consistency of the two retrieval approaches by comparing their log-Bayesian evidence, obtaining absolute values of each Bayes factor $|\Delta \log Z|<5$, i.e., with no strong preference following common scales for either model. This hybrid approach significantly enhances the efficiency of atmospheric retrieval tools without compromising their accuracy, paving the way for more rapid analysis of complex exoplanetary spectra and enabling the integration of more realistic atmospheric models.

13 pages, 11 figures. Accepted for publication in Astronomy & Astrophysics

References in corpus (30)

• MultiNest: an efficient and robust Bayesian inference tool for cosmology and particle physics
• Machine learning and deep learning
• The James Webb Space Telescope
• Bayes in the sky: Bayesian inference and model selection in cosmology
• Prevalence of Earth-size planets orbiting Sun-like stars
• petitRADTRANS: a Python radiative transfer package for exoplanet characterization and retrieval
• The prior can generally only be understood in the context of the likelihood
• Identifying Exoplanets with Deep Learning: A Five Planet Resonant Chain around Kepler-80 and an Eighth Planet around Kepler-90
• Applications of Bayesian model selection to cosmological parameters
• PandExo: A Community Tool for Transiting Exoplanet Science with JWST & HST
• Early Release Science of the exoplanet WASP-39b with JWST NIRSpec PRISM
• Atmospheric Sulfur Photochemistry on Hot Jupiters
• Exoplanetary Atmospheres - Chemistry, Formation Conditions, and Habitability
• Nested sampling for physical scientists
• TiO and VO broad band absorption features in the optical spectrum of the atmosphere of the hot-Jupiter HD209458b
• Exploring Biases of Atmospheric Retrievals in Simulated JWST Transmission Spectra of Hot Jupiters
• An Updated Study of Potential Targets for Ariel
• Automatic Classification of Kepler Planetary Transit Candidates
• ExoGAN: Retrieving Exoplanetary Atmospheres Using Deep Convolutional Generative Adversarial Networks
• Neural posterior estimation for exoplanetary atmospheric retrieval
• Early Insights for Atmospheric Retrievals of Exoplanets using JWST Transit Spectroscopy
• Accurate Machine Learning Atmospheric Retrieval via a Neural Network Surrogate Model for Radiative Transfer
• Optimizing exoplanet atmosphere retrieval using unsupervised machine-learning classification
• Impact of planetary mass uncertainties on exoplanet atmospheric retrievals
• On the safe use of prior densities for Bayesian model selection
• JWST-TST DREAMS: A Precise Water Abundance for Hot Jupiter WASP-17b from the NIRISS SOSS Transmission Spectrum
• Understanding the atmospheric properties and chemical composition of the ultra-hot Jupiter HAT-P-7b: III. Changing ionisation and the emergence of an ionosphere
• Astroconformer: The Prospects of Analyzing Stellar Light Curves with Transformer-Based Deep Learning Models
• Analysis of the planetary mass uncertainties on the accuracy of atmospherical retrieval
• Toward a multidimensional analysis of transmission spectroscopy. Part III: Modelling 2D effects in retrievals with TauREx