Polarimetric Black-Hole Imaging

Polarimetric black-hole imaging adds the orientation and coherence of the electric field to total intensity. That information is sensitive to magnetic geometry, but only after polarized synchrotron transfer, relativistic transport, Faraday effects, and instrumental leakage are modeled together. The central result is therefore conditional: calibrated linear polarization can discriminate among magnetized-flow models, but polarization ticks are not a direct map of magnetic field lines.

Helpful background. Black-Hole Geometry and Mechanics supplies null transport and screen geometry; Relativistic Matter and Compact Objects supplies magnetized plasma and synchrotron emission; Black-Hole Astrophysics supplies accretion-state and jet context; and Computational Relativity supplies polarized ray tracing and synthetic-observation checks.

Complex linear polarization

Choose the usual astronomical sky convention: position angle increases from north toward east, $Q\gt0$ denotes an electric vector along north–south, and $U\gt0$ denotes one at $45^\circ$ east of north. The linear Stokes parameters are then combined as

$$P=Q+iU, \qquad m_L=\frac{|P|}{I}, \qquad \chi=\frac12\arg P\pmod{\pi},$$

where $m_L$ is fractional linear polarization and $\chi$ is the electric-vector position angle (EVPA). The factor $1/2$ appears because a linear-polarization axis is unchanged by a $180^\circ$ rotation. A convention for positive Stokes $V$, sky position angle, circular-feed handedness, and Fourier phase must be declared before comparing software or papers.

Optically thin synchrotron emission from an isotropic electron distribution has EVPA perpendicular to the projected magnetic field in the local plasma frame. This familiar rule is not a complete image interpretation. Optical depth can rotate the relation by $90^\circ$; aberration and parallel transport change projected orientation; integration through unresolved zones causes beam depolarization; and a nonthermal or anisotropic electron distribution changes emissivity. The screen pattern reflects an emission-weighted, transported field, not a field-line tracing experiment.

Faraday rotation supplies another angle,

$$\chi(\lambda)=\chi_0+{\rm RM}\,\lambda^2, \qquad P(\lambda)=P_0\exp\!\left(2i\,{\rm RM}\,\lambda^2\right).$$

These expressions describe a single external Faraday screen, with $\lambda$ the observed wavelength and RM the corresponding observed rotation measure. For one screen at cosmological redshift $z$, ${\rm RM}_{\rm obs}={\rm RM}_{\rm rest}/(1+z)^2$. Internal rotation can also depolarize emission, while Faraday conversion transfers power between linear and circular polarization. A narrow frequency range can leave intrinsic EVPA and rotation measure degenerate. A multi-frequency fit must account for $n\pi$ angle wrapping rather than treating displayed EVPAs as ordinary directed vectors.

What an interferometer measures

In circular-feed notation, parallel-hand and cross-hand correlations sample different combinations of $I,Q,U,V$. A compact matrix form of the measurement equation is

$$\mathbf C^{\rm obs}_{ab} =\mathbf J_a\,\mathbf C_{ab}\,\mathbf J_b^\dagger +\mathbf N_{ab},$$

where $\mathbf C_{ab}$ is the source coherency visibility and $\mathbf J_a$ contains station gains, feed rotation, bandpass, relative R/L phase and amplitude, and complex leakage terms called D-terms. Leakage mixes the much brighter Stokes $I$ signal into cross-hand polarization. A small coherent leakage pattern can therefore resemble ordered source polarization.

Absolute EVPA calibration fixes a cross-hand phase zero point; relative R/L gain affects inferred circular polarization. Network and total-intensity calibration remain part of the chain. Station and scan flags, polarization signal-to-noise selections, coherent averaging, band combination, leakage-quality cuts, masked low-intensity regions, and time-averaging windows must be recorded. Without them, two images with the same color scale need not represent the same sample.

A full-Stokes likelihood should compare measured correlation products or calibrated Stokes visibilities with a forward model while retaining gain and leakage nuisance parameters. Priors on D-term magnitude, gain variability, smoothness, total flux, and polarization fraction affect posterior credible intervals. Thermal scatter is statistical uncertainty; leakage-model error, EVPA zero point, scattering, unresolved structure, variable Faraday depth, electron physics, and reconstruction choice are systematic uncertainty or model dependence.

The DOI-bound M87* release

The reproducibility anchor is EHT data product 2023-D01-01, “First M87 EHT Results: Calibrated Polarimetric Data,” DOI 10.25739/q46m-m857. It is the first public release of the calibrated 2017 M87* polarimetric visibility data. A reproduction must preserve its feed and Stokes conventions, stations, scans, bands, flags, leakage solutions, and EVPA calibration rather than substituting values from a later epoch.

The 2021 polarization analysis reconstructed Stokes $I,Q,U$ with independent imaging methods and compared image-domain and visibility-domain polarization summaries. Coherent, spatially resolved linear polarization appeared on parts of the ring, while intermethod scatter and low-intensity masking bounded detailed claims Event Horizon Telescope Collaboration 2021, §§ 2–7. Its calibration and cross-pipeline checks are integral to the result, not a detachable technical preface.

The companion physical-model analysis passed polarized GRMHD images through the same observable constraints. Within its declared simulation and electron-temperature library, models with dynamically important magnetic fields—often magnetically arrested disks—more often satisfied total-intensity and polarization conditions than weaker-flux alternatives Event Horizon Telescope Collaboration 2021, §§ 3–5. This is a conditional library comparison, not a theorem that every viable M87* flow is MAD.

Results and limitations through 2026-08-18

The evidence cutoff for mutable observational claims is 2026-08-18.

Evidence layer	Result at the cutoff	Scope and limitation
Analytic mechanism	Synchrotron emissivity, optical depth, parallel transport, and Faraday rotation can each organize or scramble EVPA.	The same projected pattern can arise from different three-dimensional fields and transfer depths.
Synthetic observations	Injected polarization passed through measured array sampling and D-terms can test leakage recovery and algorithm bias.	A successful injection only covers its chosen source, leakage prior, noise, and calibration domain.
Individual object: M87* 2017	Release 2023-D01-01 supports resolved linear polarization recovered by independent pipelines.	Detailed EVPAs depend on leakage, EVPA zero point, low-intensity masking, and time averaging.
Individual object: M87* 2017–2021	Later epochs retain a ring but show materially changing linear-polarization fraction and EVPA helicity Event Horizon Telescope Collaboration 2025, §§ 4–6.	The change may reflect the flow, an external Faraday screen, or both; the DOI-bound teaching release here remains the 2017 product.
Individual object: Sagittarius A*	Product 2024-D02-01, DOI 10.25739/k39c-mh27, provides a second source’s polarized data; ring-scale polarization is time variable Event Horizon Telescope Collaboration 2024, §§ 3–7.	Rapid variability, scattering, and different mass scale prevent a simple one-to-one comparison with M87*.
Population and null evidence	Two selected, exceptionally large apparent horizons do not establish a population distribution; unpolarized and leakage-dominated injections are required null tests.	Selection is highly incomplete, and nondetection can reflect sensitivity or depolarization rather than absent magnetic order.

The credible competing interpretations are ordered fields in a magnetically arrested flow, turbulent weaker-flux flows, optical-depth or Faraday scrambling, relativistic projection, leakage artifacts, and genuine temporal change. Current data discriminate among declared forward models, yet the unique three-dimensional field topology remains unresolved.

The interpretation could change after a new polarimetric data release, revised leakage or EVPA calibration, higher-frequency imaging, a robust circular-polarization constraint, or a polarized-transfer model that changes which magnetic states pass the visibility-domain tests.

Injection and recovery protocol

A defensible reproduction begins with 2023-D01-01 and at least two independent imaging or Bayesian pipelines. First fit total intensity and retain its gain uncertainty. Then inject known $Q,U,V$ structures into the exact sampled baselines, apply station-specific complex D-terms and R/L gain offsets, add thermal noise, and hide the input from the reconstruction operator. Repeat with zero intrinsic polarization, with a uniform external rotation measure, and with internal Faraday depolarization. Compare recovered integrated $P$, closure-like polarimetric ratios, EVPA helicity, and image-domain features before interpreting topology.

The analysis must report software versions, random seeds, likelihood, priors, selections, and interval construction. This page specifies that reproducible experiment but performs no new fit. In particular, it does not claim that the later 2018–2021 results can be regenerated from the 2017 DOI-bound release.

Common pitfalls

Rotating every tick by $90^\circ$ is not a magnetic-field map. That shortcut assumes optically thin synchrotron emission and neglects relativistic transport and Faraday effects.

Low polarized intensity does not imply a weak field. Unresolved cancellation, optical depth, internal rotation, or leakage uncertainty can suppress $|P|$.

Agreement after shared D-term calibration is correlated. Independent imaging codes can inherit the same leakage solution and EVPA zero point; injection tests must vary those shared nuisance parameters.

Exercises

An external screen has ${\rm RM}=1.0\times10^5\ {\rm rad\,m^{-2}}$. What EVPA change occurs between wavelengths $1.3$ mm and $1.0$ mm? What phase change occurs in the complex polarization $P$?

Solution

The EVPA difference is

$$\Delta\chi={\rm RM}(\lambda_1^2-\lambda_2^2) =(10^5)(1.69-1.00)\times10^{-6} =0.069\ {\rm rad}\simeq3.95^\circ.$$

Because $P=|P|e^{2i\chi}$, its complex phase changes by $2\Delta\chi=0.138$ rad, or about $7.91^\circ$. Confusing these two angles produces a factor-of-two error. The calculation assumes one external, uniform screen; internal Faraday rotation or unresolved RM structure can also change $|P|$, so a two-frequency angle difference alone does not establish that model.

References

• Event Horizon Telescope Collaboration. “EHT Sagittarius A* Polarized Data.” Data product 2024-D02-01 (2024). DOI.
• Event Horizon Telescope Collaboration. “First M87 EHT Results: Calibrated Polarimetric Data.” Data product 2023-D01-01 (2023). DOI.
• Event Horizon Telescope Collaboration. “First M87 Event Horizon Telescope Results. VII. Polarization of the Ring.” The Astrophysical Journal Letters 910 (2021): L12. DOI.
• Event Horizon Telescope Collaboration. “First M87 Event Horizon Telescope Results. VIII. Magnetic Field Structure Near the Event Horizon.” The Astrophysical Journal Letters 910 (2021): L13. DOI.
• Event Horizon Telescope Collaboration. “First Sagittarius A* Event Horizon Telescope Results. VII. Polarization of the Ring.” The Astrophysical Journal Letters 964 (2024): L25. DOI.
• Event Horizon Telescope Collaboration. “Horizon-Scale Variability of M87* from 2017–2021 EHT Observations.” Astronomy & Astrophysics 704 (2025): A91. DOI.