JWST Detects a Dusty AGB-like Source Before the Type Ia-CSM Supernova 2026sqf

JWST Detects a Dusty AGB-like Source Before the Type Ia-CSM Supernova 2026sqf

Context. Supernova (SN) 2026sqf recently appeared in the nearby face-on spiral galaxy NGC 3310 and has shown signs of strong interaction with a circumstellar medium (CSM). Such intense interaction, rare among nearby SNe, offers a valuable opportunity to reveal details on the origin and nature of its progenitor system.

Aims. We present the results of early-time photometric and spectroscopic observations, along with our efforts to identify and characterize the potential progenitor system in pre-explosion space telescope imaging.

Methods. We analyzed the early-phase spectra and light curves (LCs) of SN 2026sqf. We also carried out photometry on the point source identified in pre-explosion JWST and HST images; from these fluxes we constructed and modeled the spectral energy distribution (SED) of the candidate progenitor system.

Results. The general shapes of the observed spectra, the strengths of the emission lines, and the LC evolution all suggest that SN 2026sqf belongs to the rare SN Ia-CSM subclass. If confirmed, this would be the closest known member of this class, at D ∼ 19 Mpc. We also identified the potential progenitor system of the event, the first such identification for this SN subclass. Our results are consistent with the expectation that SNe Ia-CSM emerge from a system consisting of an exploding white dwarf and an asymptotic giant branch (AGB) star undergoing a common-envelope phase.

Conclusions. We report the first candidate progenitor system for a thermonuclear supernova identified in JWST pre-explosion imaging, and the first evidence for a (probable) carbon-rich AGB donor to the exploding white dwarf. The CSM mass and dust content are consistent with expectations for an AGB environment, but the narrow-line width exceeds superwind expansion velocities, favoring an episodic ejection. Binary interaction shortly before the explosion is a natural explanation, though the channel and timescale remain uncertain. Late-time follow-up, especially with JWST, will test the identification and the conclusions of our early-phase analysis.

Improvements to AI Systems Based on This Paper:

1. Progenitor Identification from Pre-Explosion Multi-Wavelength Imaging

• Improvement: Train a deep-learning model to cross-correlate pre-explosion JWST and HST images across filters (e.g., F090W, F150W, F356W, F444W) to automatically detect point sources at the SN location, subtract host-galaxy contamination, and flag candidate progenitors.

• Enhanced capability: The AI can autonomously identify progenitor systems for any transient with pre-explosion space-based imaging, even in crowded or dusty environments, and output a confidence score for association.

2. Spectral Energy Distribution (SED) Modeling and Classification

• Improvement: Implement a Bayesian neural network or Gaussian process regressor that ingests multi-band photometry and outputs posterior distributions for stellar parameters (temperature, luminosity, mass-loss rate, dust mass, and extinction).

• Enhanced capability: The AI can instantly classify a progenitor as AGB-like, red supergiant, or binary-interaction candidate, and quantify the probability that the system is undergoing a common-envelope phase—without manual SED fitting.

3. Early-Time Spectral Line Diagnostics for SN Ia-CSM Classification

• Improvement: Develop a transformer-based spectral classifier that learns from line ratios (e.g., Hα, [O III], [N II], He I) and line widths to distinguish Type Ia-CSM from Type IIn, Ibn, and core-collapse SNe.

• Enhanced capability: The AI can provide real-time classification within hours of the first spectrum, flagging rare subclasses and triggering follow-up observations (e.g., JWST late-time imaging) automatically.

4. Ejection Mechanism Inference from Narrow-Line Widths

• Improvement: Train a physics-informed recurrent neural network on synthetic spectra from hydrodynamic simulations of episodic vs. continuous mass loss, using line-profile shapes and velocity shifts as inputs.

• Enhanced capability: The AI can predict whether the CSM was ejected in a single episodic burst (e.g., common-envelope merger) or via steady superwind, and estimate the ejection timescale relative to explosion—critical for testing binary evolution models.

5. Automated Late-Time Follow-Up Scheduling

• Improvement: Build a reinforcement-learning agent that uses early-phase data (LC slope, spectral evolution, progenitor SED) to decide optimal epochs and filters for JWST late-time observations, maximizing the chance of detecting the surviving companion or ejecta–CSM interaction.

• Enhanced capability: The AI can autonomously propose a follow-up campaign for any SN Ia-CSM candidate, balancing telescope time, target visibility, and scientific yield, and update its plan as new data arrive.

6. Dust Content and Mass-Loss Rate Estimation from IR Excess

• Improvement: Integrate a radiative-transfer emulator (e.g., a neural network trained on DUSTY or MCMax models) that maps JWST mid-IR photometry to dust mass, grain size, and temperature, with uncertainty propagation.

• Enhanced capability: The AI can quickly estimate the CSM dust content for any dusty transient, enabling comparisons across a large sample to identify outliers like SN 2026sqf and constrain progenitor metallicity and pulsation history.

7. Cross-Sample Rare-Class Discovery

• Improvement: Use a contrastive learning framework on a large transient database (e.g., ZTF, LSST) to embed light curves and spectra, then cluster to find new SN Ia-CSM candidates with similar progenitor signatures (e.g., pre-explosion IR excess, narrow Hα).

• Enhanced capability: The AI can proactively search for more members of this rare class, increasing sample size and enabling statistical studies of their progenitor channels and explosion mechanisms.

What the improved AI system can do overall:

It can autonomously detect, classify, and characterize rare interacting thermonuclear supernovae from raw imaging and spectral data, identify their progenitor systems with quantified confidence, infer the mass-loss history and binary interaction physics, and design optimal follow-up strategies—all in near-real-time, accelerating discovery and deepening physical understanding of these extreme events.

Supernova (SN) 2026sqf recently appeared in the nearby face-on spiral galaxy NGC 3310 and has shown signs of strong interaction with a circumstellar medium (CSM). Such intense interaction, rare among nearby SNe, offers a valuable opportunity to reveal details on the origin and nature of its progenitor system. We analyzed the early-phase spectra and light curves (LCs) of SN 2026sqf. The general shapes of the observed spectra, the strengths of the emission lines, and the LC evolution all suggest that SN 2026sqf belongs to the rare SN Ia-CSM subclass. If confirmed, this would be the closest known member of this class, at D 19 Mpc. We also report the first candidate progenitor system for a thermonuclear supernova identified in JWST pre-explosion imaging, and the first evidence for a (probable) carbon-rich AGB donor to the exploding white dwarf (WD). Our results are consistent with the expectation that SNe Ia-CSM emerge from a system consisting of a WD and an AGB star undergoing a common-envelope phase. The CSM mass and dust content are consistent with expectations for an AGB environment, but the narrow-line width exceeds superwind expansion velocities, favoring an episodic ejection. Binary interaction shortly before the explosion is a natural explanation, though the channel and timescale remain uncertain. Late-time follow-up, especially with JWST, will test the identification and the conclusions of our early-phase analysis.