🔭 astrophysics

Measuring the Black Hole and Accretion Parameters of Sagittarius A* from EHT Observations using a Semi-Analytic Model

This paper presents a Bayesian hierarchical analysis of Event Horizon Telescope snapshots using a semi-analytic model to infer Sagittarius A*'s accretion and black hole parameters, revealing that while spin and magnetic fields remain unconstrained by 2017 data, the source exhibits a nearly face-on inclination with emission concentrated near the horizon and above the equatorial plane.

Original authors: Braden J. Marazzo-Nowicki, Paul Tiede, Dominic O. Chang, Daniel C. M. Palumbo, Michael D. Johnson

Published 2026-01-26

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Braden J. Marazzo-Nowicki, Paul Tiede, Dominic O. Chang, Daniel C. M. Palumbo, Michael D. Johnson

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine trying to take a clear photograph of a tiny, flickering firefly that is spinning wildly in a dark room, but you only have a camera with a broken lens and a very slow shutter speed. That is essentially the challenge astronomers faced when trying to image Sagittarius A* (Sgr A*), the supermassive black hole at the center of our galaxy, using the Event Horizon Telescope (EHT).

This paper describes a new way to solve that puzzle. Instead of trying to force a single, perfect picture out of messy, changing data, the authors used a "snapshot" approach combined with a clever statistical trick to figure out the black hole's true nature.

Here is a breakdown of what they did and what they found, using simple analogies.

The Problem: The "Flickering Firefly"

Sgr A* is not a static object like a mountain; it is a swirling storm of hot gas and magnetic fields that changes its shape every few minutes.

The Old Way: Usually, radio telescopes take pictures by watching the Earth rotate, which moves the telescopes around the source to fill in the gaps (like taking many photos from different angles to build a 3D model). But because Sgr A* changes so fast, by the time the Earth rotates enough to fill in the picture, the "firefly" has already moved and changed shape. The old method blurs the image.
The New Way: The authors treated the observation like a series of quick, 2-minute "snapshots." They realized that even though the data in each snapshot is sparse (like a puzzle with missing pieces), they could fit a specific mathematical model to each one individually.

The Tool: The "Dual-Cone" Model

To make sense of the blurry data, they used a semi-analytic model called a "dual-cone."

The Analogy: Imagine the black hole is a spinning top. Instead of simulating every single particle of gas (which would take a supercomputer years to calculate), the authors imagined the glowing gas is trapped inside two thin, hollow ice-cream cones pointing up and down along the black hole's spin axis.
Why it works: This "cone" shape is a simplified version of what complex physics simulations show. It's light enough to run quickly on a computer but detailed enough to capture the main features: a dark hole in the middle (the shadow) and a bright ring of light around it.

The Method: The "Group Average"

The authors took 161 of these 2-minute snapshots from a single day in 2017.

Individual Fits: They fitted the "cone" model to each snapshot separately. Because the data was sparse, some snapshots gave weird answers (like the black hole spinning one way, then the other).
The Stack: They then used a Bayesian hierarchical model. Think of this as asking 161 different people to guess the weight of a cat. Some guesses are wild, but if you stack all their answers together, you can find the average weight (the true structure) and see how much the guesses varied (the flickering).
- The Result: This allowed them to separate the stable features (what the black hole always looks like) from the variable features (how the gas moves and changes).

What They Found

Even though the data was tricky, this method revealed several clear facts about Sgr A*:

We are looking almost straight down: The black hole is spinning like a top, and we are looking at it from almost directly above its "North Pole." The angle is nearly face-on (about 9 degrees off from straight down).
The light is very close to the edge: The brightest part of the glowing gas is right next to the black hole's event horizon (the point of no return).
The gas is moving slowly: The material swirling around isn't racing at the speed of light; it's moving at a more moderate pace.
The "Spin" is a mystery: While they could tell us where we are looking and how the gas is arranged, they could not definitively say which way the black hole is spinning (clockwise or counter-clockwise). The data was too noisy to pin that down.

The "Truth Test"

To make sure their method wasn't broken, they tested it on fake data.

They created a computer simulation of a black hole with a known spin, angle, and shape.
They ran their "snapshot stacking" method on this fake data.
The Result: The method successfully recovered the known angle and shape, proving the tool works. However, just like with the real black hole, it struggled to pinpoint the exact spin direction, confirming that this is a hard problem for current telescopes, not a mistake in their math.

The Bottom Line

The authors successfully separated the "steady" structure of our galaxy's central black hole from its "flickering" behavior. They confirmed that we are looking at it from a nearly top-down view and that the glowing gas is concentrated right near the edge. However, they also learned that with current technology, the exact direction of the black hole's spin remains a mystery, likely because the "flickering" of the gas hides the subtle clues needed to solve it.

What they did NOT do:

They did not use this to predict future black hole behavior.
They did not apply this to other types of stars or planets.
They did not claim to have solved the spin mystery; they explicitly stated it remains unconstrained.

This paper is essentially a new, smarter way to take a blurry, moving picture and extract the most reliable facts possible from it, while honestly admitting what the picture still can't tell us.

Technical Summary: Measuring the Black Hole and Accretion Parameters of Sagittarius A from EHT Observations using a Semi-Analytic Model*

Problem Statement
The Event Horizon Telescope (EHT) Collaboration produced the first image of the apparent shadow of Sagittarius A* (Sgr A*), the supermassive black hole at the center of the Milky Way. However, Sgr A* exhibits significant structural variability on timescales shorter than the duration of an observation (minutes to hours). This rapid intraday variability prevents the use of traditional Earth-rotation aperture synthesis to improve $(u, v)$ -plane sampling and accurately reconstruct the source structure. Consequently, standard imaging techniques struggle to separate time-variable emission features from stable signatures of strong gravity and the accretion environment. Previous analyses using simple geometric models constrained the ring diameter but left other morphological parameters, such as brightness asymmetries and physical accretion conditions, poorly constrained. Furthermore, while General Relativistic Magnetohydrodynamics (GRMHD) simulations are essential for interpretation, their computational demands make them unsuitable for direct Bayesian inference on sparse visibility data.

Methodology
To address these challenges, the authors employ a "snapshot modeling" approach combined with a Bayesian hierarchical framework. The core methodology involves:

Semi-Analytic Emission Model: The study utilizes a lightweight, robust, phenomenological "dual-cone" emission model. This model assumes synchrotron-emitting, magnetized plasma constrained to flow within two thin, oppositely oriented cones aligned with the black hole's spin axis. The model is axisymmetric and ray-traced in a Kerr geometry. It includes parameters for black hole spin ( $a_*$ ), observer inclination ( $\theta_o$ ), cone opening angle ( $\theta_s$ ), projected spin position angle (p.a.), characteristic emission radius ( $R_{peak}$ ), emissivity profile exponents ( $p_1, p_2$ ), fluid velocity ( $\beta_v$ ), and magnetic field configuration ( $\iota, \eta, \chi$ ). The mass-to-distance ratio ( $\theta_g$ ) is fixed at $5.03\,\mu\text{as}$ based on prior constraints.
Snapshot Inference: The EHT data from April 7, 2017, are split into 120-second "snapshots." For each snapshot, the authors perform Bayesian inference on complex visibilities using the Comrade.jl package. They jointly infer the physical model parameters and complex station gains (amplitude and phase) to account for instrumental errors.
Sampling Strategy: Due to the high dimensionality and potential multimodality of the posterior distributions, the authors use Parallel Tempering Markov Chain Monte Carlo (PTMCMC) sampling via Pigeons.jl. This method allows the sampler to traverse low-probability regions between peaks, which standard MCMC methods (like Metropolis-Hastings) often fail to do in this context.
Hierarchical Stacking: To extract persistent source structure and quantify temporal variability, the authors construct a global Bayesian hierarchical model. They stack the posterior distributions from all 161 snapshots to form a global posterior. This model estimates the average source structure (mean $\mu_p$ ) and the variability distribution (standard deviation $\sigma_p$ ) for each parameter, separating average structure uncertainty from the impacts of temporal variability and model misspecification.

Key Contributions

Novel Inference Pipeline: The paper introduces a pipeline that directly fits a physics-based semi-analytic model to sparse EHT visibility data on a snapshot-by-snapshot basis, avoiding the need for full GRMHD simulations during the inference loop.
Hierarchical Variability Separation: By stacking snapshot posteriors, the method successfully disentangles the time-averaged, persistent source structure from the rapid temporal variability inherent to Sgr A*.
Validation on Synthetic Data: The pipeline is validated using synthetic EHT observations generated from a GRMHD simulation with known ground-truth parameters. This test confirms that the inference pipeline does not introduce significant biases in spin, inclination, or position angle measurements, although spin remains unconstrained even in the synthetic test.

Results
Applying this method to the 2017 Sgr A* data yields the following constraints (separating average structure uncertainty $\pm$ from temporal variability/model misspecification $\pm_v$ ):

Observer Inclination: The analysis strongly favors a nearly face-on observer inclination of $\theta_o = 9.2^\circ \pm 3.6^\circ \pm_v 11.6^\circ$ . This result is consistent across snapshots and aligns with previous studies suggesting low inclination angles.
Emission Radius: The emission peak is constrained to be near the horizon, $R_{peak} = 4.9 \pm 0.1 \pm_v 0.5 \, GM/c^2$ .
Spin Position Angle: The projected spin position angle is constrained to be near-vertical, $p.a. = 7.3^\circ \pm 7.08^\circ \pm_v 43.5^\circ$ (counterclockwise from vertical).
Emission Geometry: The dominant emission is found to be $43.4^\circ \pm 2.0^\circ \pm_v 5.9^\circ$ above the equatorial plane, suggesting significant non-equatorial emission (potentially from a thick disk or jet).
Spin and Magnetic Fields: Black hole spin ( $a_*$ ) and magnetic field parameters ( $\eta, \iota$ ) remain largely unconstrained. The spin posterior is nearly uniform across the prior range, indicating that 2017 EHT observations alone cannot constrain spin without further assumptions or data.
Plasma Speed: The analysis favors slow-moving plasma flows relative to the innermost stable circular orbit (ISCO), with $\beta_v = 0.33 \pm 0.22 \pm_v 2.1$ .

Significance and Limitations
The paper claims to provide an independent measurement of the observer inclination relative to the black hole spin axis, supporting the hypothesis of a nearly face-on view of Sgr A*. The results suggest that the accretion flow is dominated by emission near the horizon and above the equatorial plane.

However, the authors are modest regarding the constraints on black hole spin and magnetic fields, attributing the lack of constraints to the sparsity of the data, the degeneracies between parameters, and the limitations of the axisymmetric dual-cone model in capturing non-axisymmetric, time-variable turbulence. The authors explicitly state that the rapid variability of Sgr A* and the model's inability to fully capture complex fluid dynamics lead to significant uncertainties. They note that while the method successfully reproduces parameters from GRMHD simulations in synthetic tests, the real data analysis is subject to model misspecification. Future work is identified as necessary to incorporate optical depth effects, multi-component emission (e.g., jets), and polarized light to reduce biases and improve constraints on spin and magnetic field configurations.