A Missing Latent, Not a Missing Simulator: Radius-Augmented Inference for Real JWST Retrieval
Abstract
Amortized simulation-based inference (SBI), which is trained on radiative-transfer simulators, recovers exoplanet atmospheres accurately on synthetic James Webb Space Telescope (JWST) spectra but collapses when it comes to real reduced spectra. The flow posterior collapsed on real WASP-39b (importance-sampling effective sample size (ESS) = 1, best-fit \c{hi}2/N = 301), and such a failure is usually blamed on missing the forward-model physics, but ruling these levers out with nested sampling first (a temperature gradient, SO2 opacity, and a high-fidelity opacity set) leaves the fit unchanged, meaning the collapse is not from the simulator misspecification but instead from a missing latent, the planet radius. To fix this, we build MIRAGE, a radius-augmented flow-matching posterior calibrated against an independent nested-sampling reference with importance sampling and an optimal-transport map, which yields a physical and literature-consistent retrieval of real WASP-39b (with \c{hi}2/N from 301 to 0.06). This same method transfers unchanged across two instruments and three real JWST targets, including one spectrum self-reduced end-to-end from raw Mikulski Archive for Space Telescopes(MAST) data. The lesson is cross-domain, as a latent the simulator encodes but the inference omits can masquerade as misspecification.
Related
- Towards Retrieval Augmented Generation in High-Energy and Astroparticle Physics
- Refactoring the SIXTE simulator: Towards a more modular code base
- XuYi Interferometer Observation Simulator
- A simulator of the MAJIS instrument onboard the JUICE mission: description and application to operational and scientific cases
- The (missing) gamma-ray counterparts to Einstein Probe events. I: Search for keV-GeV counterparts of Einstein Probe fast X-ray transients
BibTeX
@misc{chakravertty2026missing,
title = {{A Missing Latent, Not a Missing Simulator: Radius-Augmented Inference for Real JWST Retrieval}},
author = {Angshuman Chakravertty and P V V Raj},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.02245},
eprint = {2610.02245},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.02245},
}
RIS
TY - UNPB
TI - A Missing Latent, Not a Missing Simulator: Radius-Augmented Inference for Real JWST Retrieval
AU - Angshuman Chakravertty
AU - P V V Raj
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2610.02245
UR - https://arxiv.org/abs/2610.02245
AB - Amortized simulation-based inference (SBI), which is trained on radiative-transfer simulators, recovers exoplanet atmospheres accurately on synthetic James Webb Space Telescope (JWST) spectra but collapses when it comes to real reduced spectra. The flow posterior collapsed on real WASP-39b (importance-sampling effective sample size (ESS) = 1, best-fit \c{hi}2/N = 301), and such a failure is usually blamed on missing the forward-model physics, but ruling these levers out with nested sampling first (a temperature gradient, SO2 opacity, and a high-fidelity opacity set) leaves the fit unchanged, meaning the collapse is not from the simulator misspecification but instead from a missing latent, the planet radius. To fix this, we build MIRAGE, a radius-augmented flow-matching posterior calibrated against an independent nested-sampling reference with importance sampling and an optimal-transport map, which yields a physical and literature-consistent retrieval of real WASP-39b (with \c{hi}2/N from 301 to 0.06). This same method transfers unchanged across two instruments and three real JWST targets, including one spectrum self-reduced end-to-end from raw Mikulski Archive for Space Telescopes(MAST) data. The lesson is cross-domain, as a latent the simulator encodes but the inference omits can masquerade as misspecification.
ER -