Decoding the jet of BL Lacertae using relativistic magneto-hydrodynamics

The paper investigates the origin of the highest recorded linearly polarised optical flare in the blazar BL Lacertae (BL Lac), which reached a measured optical polarisation degree of m opt = 47.6% on MJD 60261-60262.75. The authors use very-long-baseline interferometry (VLBI) observations at 43 GHz from the VLBA, taken during the Imaging X-ray Polarimetry Explorer (IXPE) campaign, to resolve the compact emission region and test scenarios for the flare's origin.

The VLBI observations, spanning MJD 60238 to 60351, reveal a jet structure modelled with circular Gaussian components labelled C (core), Q1, Q2, Q3, and a new component QN ejected in the last two epochs. The analysis shows that component Q2 exhibits a sweeping, helical motion, moving between position angles 175° and 185°, while moving away from the core. The total flux of the jet components increased during the flare peak, and the flare culminated with the ejection of the new component QN. The VLBI polarisation angle (ψ) of Q2 rotates smoothly with time, mirroring the optical and radio measurements, and the VLBI polarised flux density (P) increases between the first and second VLBI epochs and decreases afterwards.

Using the analytical approach by Hagen-Thorn et al. (2008), the authors estimate that for a bulk Lorentz factor Γ = 4.5, a viewing angle change of Δθ ≈ 15° is sufficient to reproduce the observed P changes. They note that Q2 moves a projected Δθ ≤ 10.7° in the sky, and that this interpretation is consistent with previous findings that the jet collimation decreases rapidly at the position of Q2.

To further explore the sweeping motion scenario, the authors use state-of-the-art 3D relativistic magneto-hydrodynamic (RMHD) simulations with the PLUTO code, ray-traced linearly polarised emission maps using RADMC-3D. The simulations reproduce the observed behaviour: the Stokes I flux density increases while the jet is pointing towards the line of sight and decreases while pointing away. The simulations reproduce the observations most optimally for a jet sweep of Δθ ∼ 7°, adhering to the upper limit imposed by the observations. The simulated ψ values show excellent agreement with the observed values, as demonstrated in Figure 2.

A key finding from the simulations is the exploration of the plasma parameter n, defined as the ratio between protons and positrons in the jet. The investigation shows that the optimal match between observations and simulations is achieved when protons outnumber positrons by a factor of n ≥ 100. Conversely, higher positron numbers fail to induce the observed ψ orientation; the simulated ψ remain constant regardless of the orientation of the jet, consistent with the dependence of the Faraday rotation coefficient on the plasma composition. The authors note that decreasing the number of protons around the time of the flare would increase the simulated P, matching the observations even more precisely, with the accelerated protons escaping the emission region on timescales of a few weeks, matching the cadence of the VLBI observations.

The paper also discusses the potential for BL Lac as a candidate for astrophysical neutrino emission, given the inferred hadronic component. The authors estimate the proton kinetic luminosity as L p ≈ 1 × 10 44 erg s-1, which is comfortably sub-Eddington (L p ≈ 5 × 10-3 L Edd). However, they find that the photomeson efficiency of the parsec-scale region is low, implying that detectable neutrino emission would require the hadronic component to persist in more compact regions closer to the central engine.

In conclusion, the authors state: "we have investigated the mechanism that caused the most prominent P optical flare ever in a blazar. We found that the helical swing of the approaching jet reproduces the observed increase in Stokes I flux density and P, as well as the smooth rotation of ψ, by comparing our VLBI observations to RMHD simulations as a first order approximation. Our results are further supported by analytical calculations, which indicate that a mild viewing angle change can reproduce the observed phenomenology. Ultimately, our work underscores the major impact of relativistic jet geometry on multi-wavelength variability, suggesting that many extreme blazar phenomena may be driven as much by geometric perspective as by intrinsic energetic changes."

Improvements to AI Systems:

1. Polarized Radiative Transfer Emulation for Jet Simulations

• Improvement: Integrate the RMHD + RADMC-3D pipeline (PLUTO for magneto-hydrodynamics, RADMC-3D for ray-traced polarized emission) into a surrogate AI model. Train a neural network to map jet geometry (e.g., viewing angle, sweep amplitude, plasma composition parameter n) directly to Stokes I, Q, U, and polarization angle ψ time series.

• Capability: The AI can predict multi-wavelength polarization variability from jet kinematics in real time, enabling rapid parameter estimation for new blazar flares without expensive 3D simulations.

2. Autonomous VLBI Component Tracking and Helical Motion Detection

• Improvement: Build a computer-vision and sequence-modeling system that automatically identifies and tracks circular Gaussian components (C, Q1, Q2, Q3, QN) in multi-epoch VLBI images. Use a transformer-based architecture to detect non-linear trajectories (e.g., sweeping, helical motion) and flag position-angle rotations.

• Capability: The AI can monitor live VLBI data streams, alerting astronomers to new component ejections or geometric swings that may precede high-polarization flares, enabling prompt multi-observatory follow-up.

3. Physics-Informed Inference for Plasma Composition

• Improvement: Develop a Bayesian neural network that incorporates the Faraday rotation coefficient’s dependence on proton-to-positron ratio (n). Train it on synthetic polarized spectra from the paper’s simulations to infer n from observed ψ and P evolution.

• Capability: The AI can estimate the hadronic content of blazar jets directly from polarization data, identifying candidates for neutrino emission (e.g., when n ≥ 100) and prioritizing them for IceCube follow-up.

4. Geometric Variability Decomposition for Multi-Wavelength Light Curves

• Improvement: Create a disentangled representation learning model that separates observed flux and polarization variability into (a) intrinsic energetic changes (e.g., particle acceleration) and (b) geometric effects (e.g., viewing angle swings). Use the paper’s analytical relation (Δθ ≈ 15° for Γ = 4.5) as a physical constraint in the loss function.

• Capability: The AI can decompose any blazar’s multi-wavelength light curve into geometric vs. intrinsic components, providing a diagnostic tool to classify extreme flares and predict whether they are line-of-sight alignment events or genuine energy injections.

5. Real-Time Sub-Eddington Luminosity and Neutrino Yield Estimator

• Improvement: Implement a regression model trained on the paper’s derived proton kinetic luminosity (L p ≈ 1 × 10 44 erg s−1) and photomeson efficiency calculations. Inputs: VLBI jet speed, component position, and inferred n. Outputs: L p, Eddington ratio, and expected neutrino flux.

• Capability: The AI can instantly assess the neutrino-production potential of any observed blazar flare, flagging high-probability events for neutrino observatories and optimizing target-of-opportunity scheduling.

6. Simulation-to-Observation Domain Adaptation

• Improvement: Use adversarial domain adaptation to bridge the gap between idealized RMHD simulation outputs and noisy real VLBI/optical observations. Train a discriminator to distinguish simulated from observed Stokes parameters, forcing the generator to produce realistic polarization maps.

• Capability: The AI can generate high-fidelity synthetic polarization images for any jet geometry, enabling robust testing of flare scenarios and improving the interpretability of future IXPE and VLBA campaigns.