TOI-7475 spectroscopic follow up - detection of a massive hot Jupiter on an eccentric orbit

arXiv:2608.09679 · astro-ph.EP · Submitted 2026-08-10 · Read on arXiv

Joanna Pietrucha, Jakub Jerzy Tokarek, Mateusz Witczak, Wojciech Dimitrov

Astronomical Observatory Institute, Faculty of Physics and Astronomy, Adam Mickiewicz University

astro-ph.EP

Submitted: 2026-08-10

Updated: 2026-08-11

Comments: 9 pages, 2 figures, 3 tables, submitted to Acta Astronomica

License: http://creativecommons.org/licenses/by-nc-nd/4.0/

Importance score: 35/100

The gist: We present the first spectroscopic follow-up for TOI-7475, and report the detection of a planetary-mass companion.

Terminology

Summary

We present the first spectroscopic follow-up for TOI-7475, and report the detection of a planetary-mass companion. Transits with a period of 3.25 days were observed by TESS. The dips in brightness are 0.006 mag and last approximately 4.5 hours. The host star is a spectral type F5 with a mass of approximately 1.48 M ⊙. We collected 24 high resolution echelle spectra using the PST1 telescope. Analysis of the TESS photometric data and our radial velocities was performed using the Wilson-Devinney method. We obtained a planetary mass of the companion of 12.0 ± 1.0 M jup. The small semi-major axis and short orbital period classify the planet as a Hot Jupiter. Its orbit is elliptical with an eccentricity of 0.16. The ellipticity is evident in the spectroscopic data and could indicates the young age of the system.

Improvements for AI systems

Improvements to AI Systems:

  1. Automated Exoplanet Validation Pipeline
  • Train an AI model to cross-correlate TESS transit light curves (depth, duration, period) with stellar parameters (spectral type, mass) to automatically flag candidate planetary-mass companions and classify them (e.g., Hot Jupiter) before spectroscopic follow-up.

  • Improved capability: Given raw photometric data and stellar catalog entries, the AI can predict the probability of a planetary companion, estimate its mass range, and recommend optimal observing windows for radial velocity confirmation.

  1. Orbital Eccentricity Inference from Sparse Radial Velocities
  • Develop a neural network that takes incomplete, noisy radial velocity time series (like the 24 spectra here) and directly outputs posterior distributions for eccentricity, semi-major axis, and mass, using physics-informed priors from Keplerian dynamics.

  • Improved capability: The AI can accurately recover eccentricity (e.g., 0.16) even with few epochs, reducing the need for dense sampling and enabling faster classification of young, eccentric systems.

  1. Age Estimation from Orbital Dynamics
  • Create a regression model that links eccentricity, orbital period, and host star mass to system age, trained on known young exoplanet populations.

  • Improved capability: Given a newly detected Hot Jupiter with moderate eccentricity, the AI can estimate the system’s age (e.g., young vs. old) and flag it for follow-up age-dating via stellar activity or lithium abundance.

  1. Automated Wilson-Devinney Parameter Fitting
  • Replace manual fitting with a deep learning surrogate that maps photometric + RV inputs directly to best-fit binary/planet parameters (mass ratio, inclination, limb darkening) in seconds, instead of iterative numerical methods.

  • Improved capability: The AI can process hundreds of candidate systems in real time, providing instant parameter estimates with uncertainty bounds, enabling rapid triage for detailed study.

  1. Transit Timing Variation (TTV) Anomaly Detector
  • Build a transformer-based model that ingests multi-sector TESS light curves and predicts if a single transiting planet’s timing variations indicate additional unseen companions or orbital precession (relevant for eccentric orbits).

  • Improved capability: The AI can flag systems like TOI-7475 for hidden companions or dynamical perturbations, guiding targeted follow-up observations.

What the improved AI system can do:

  • Automatically detect and characterize planetary-mass companions from combined photometric and spectroscopic data, including mass, eccentricity, and age estimates, with minimal human intervention.

  • Provide real-time probabilistic classifications (e.g., Hot Jupiter, young system) and prioritize targets for high-resolution spectroscopy, saving telescope time.

  • Generate synthetic training data for rare eccentric Hot Jupiters, improving robustness for future surveys like PLATO or Ariel.

Abstract

We present the first spectroscopic follow-up for TOI-7475, and report the detection of a planetary-mass companion. Transits with a period of 3.25 days were observed by TESS. The dips in brightness are 0.006 mag and last approximately 4.5 hours. The host star is a spectral type F5 with a mass of approximately 1.48 M. We collected 24 high resolution echelle spectra using the PST1 telescope. Analysis of the TESS photometric data and our radial velocities was performed using the Wilson-Devinney method. We obtained a planetary mass of the companion of 12.0 plus or minus 1.0 M jup. The small semi-major axis and short orbital period classify the planet as a Hot Jupiter. Its orbit is elliptical with an eccentricity of 0.16. The ellipticity is evident in the spectroscopic data and could indicates the young age of the system.

Related papers