Interacting Supernovae: a Radio and X-ray Strategy to Constrain the Structure of the Circumstellar Medium

The interaction of supernova (SN) ejecta with the dense circumstellar medium (CSM) converts shock kinetic energy into radiation across multiple wavebands. We investigate the dependence of the X-ray and radio emission on the CSM geometry, considering spherical, hourglass, and disk shapes for the CSM. We find that the spectral and light-curve properties, both in X-ray and radio, significantly differ for spherical and non-spherical CSM structures. For a non-spherical CSM, the radio light curve flattens out near the peak frequency, due to efficient free-free absorption by the unshocked CSM. Moreover, the early rise of the radio light curve is shallower when the CSM density along the observer line of sight is larger than that in other directions. If the CSM density is lower along the observer line of sight, the radio light curve flattens near its peak, and the reverse-shock component is negligible in X-rays. Building on these features, we provide a method to constrain the CSM structure based on the rising part the radio light curve in the proximity of its peak; we show that the decay part of the radio light curve, after its peak, carries insight on whether the CSM density profile is wind-like or not. We further adopt the X-ray signal to corroborate the information extracted from radio. We test our strategy on SN 1993j and SN 2023ixf. For both SNe, we find that an asymmetric CSM is in excellent agreement with radio and X-ray observations and provides a viable alternative to non-wind scenarios suggested in the literature. Our findings highlight the crucial insight provided by radio and X-ray signals into the mass-loss history of the SN progenitor.

Improvements to AI Systems:

1. Multi-Waveband Synthetic Observational Model Integration

• Enhance AI astrophysics simulators to incorporate geometry-dependent (spherical, hourglass, disk) CSM radiative transfer for both X-ray and radio, enabling the generation of synthetic light curves and spectra that match observed SN data.

• The improved system can automatically fit multi-band observations (radio + X-ray) to infer 3D CSM geometry and density profiles, reducing degeneracies present in single-band analyses.

2. Physics-Informed Light-Curve Classification

• Train AI classifiers to distinguish spherical vs. non-spherical CSM based on specific radio light-curve features (e.g., flattening near peak frequency, shallow early rise, decay slope behavior).

• The improved system can flag candidate SNe with asymmetric CSM from survey data (e.g., ZTF, LSST) in real time, prioritizing them for follow-up X-ray/radio observations.

3. Reverse-Shock X-ray Component Predictor

• Implement a neural network that predicts the relative contribution of reverse-shock vs. forward-shock X-ray emission based on CSM geometry and observer line-of-sight density.

• The improved system can estimate whether X-ray observations will be dominated by reverse-shock emission, aiding in the design of observation campaigns and interpretation of faint X-ray signals.

4. Wind-Profile Discriminator from Radio Decay

• Develop an AI module that analyzes the post-peak radio decay slope to infer whether the CSM density profile follows a wind-like (r−2) or non-wind (e.g., plateau, clumpy) structure, independent of spectral fitting.

• The improved system can automatically test alternative mass-loss histories for SN progenitors, providing a fast, model-agnostic diagnostic for stellar evolution models.

5. Joint Radio-X-ray Constraint Optimizer

• Build a Bayesian optimization framework that jointly fits radio and X-ray data to constrain CSM geometry, density, and shock microphysics, using the paper’s method as a prior.

• The improved system can produce posterior distributions for CSM asymmetry parameters (e.g., axis ratio, opening angle) for individual SNe, enabling population-level studies of progenitor mass-loss geometry.

6. Early-Warning Asymmetry Detector for Transient Surveys

• Integrate the paper’s finding that the early radio rise is shallower when line-of-sight CSM density is higher into an AI-based anomaly detector.

• The improved system can identify SNe with likely asymmetric CSM within days of the radio detection, triggering rapid multi-wavelength follow-up to capture the critical early-phase signal.

7. Simulation-to-Observation Transfer Learning

• Use the paper’s synthetic light-curve libraries to pre-train a deep learning model that maps observed radio/X-ray light curves to CSM geometry parameters, then fine-tune on real SNe (e.g., SN 1993j, SN 2023ixf).

• The improved system can generalize to unseen SNe, providing rapid, automated CSM structure inference without requiring expensive radiative transfer simulations for each event.