🔭 astrophysics

Revisiting GW170817 at milliarcsecond scale: high-precision constraints on jet geometry and $H_0$

This paper presents a robust Bayesian analysis of all available VLBI data for GW170817 using new hydrodynamical models to precisely constrain the jet viewing angle and derive a Hubble constant of $65.5\pm4.4$ km s $^{-1}$ Mpc $^{-1}$ , a value that aligns more closely with early-Universe Planck measurements than with late-Universe SH0ES results.

Original authors: Kelly Gourdji, Adam T. Deller, Chris Flynn, Taya Govreen-Segal, Cullan Howlett, Kunal P. Mooley, Ehud Nakar

Published 2026-05-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Kelly Gourdji, Adam T. Deller, Chris Flynn, Taya Govreen-Segal, Cullan Howlett, Kunal P. Mooley, Ehud Nakar

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 the universe as a giant, expanding balloon. For decades, scientists have been arguing about exactly how fast that balloon is inflating. This rate is called the Hubble Constant ( $H_0$ ). The problem is, when you measure it using the "early universe" (like the afterglow of the Big Bang), you get one number. When you measure it using the "late universe" (like exploding stars nearby), you get a different, slightly faster number. This disagreement is known as the "Hubble Tension."

In 2017, astronomers had a breakthrough. They detected a cosmic crash: two neutron stars smashing together. This event, named GW170817, was special because it sent out both gravitational waves (ripples in space-time) and light (a bright flash). This made it a "Standard Siren"—a cosmic lighthouse that could help measure the universe's expansion rate.

However, there was a catch. To measure how far away the lighthouse is, you need to know exactly how it's tilted relative to us. If you look at a spinning top from the side, it looks different than if you look at it from the top. If you guess the tilt wrong, you guess the distance wrong, and your calculation of the universe's expansion rate is off.

The "Blind Spot" Problem
For GW170817, the gravitational wave detectors couldn't tell us the tilt very well because the event was viewed almost head-on (like looking straight down the barrel of a gun). It was hard to distinguish the "left-right" spin from the "up-down" spin.

To fix this, scientists needed to look at the "afterglow"—the fading light left over from the crash. As the debris from the crash zooms away, it creates a jet of particles. If we can watch this jet move across the sky over time, we can figure out the angle.

The Paper's New Approach: A High-Precision Map
This paper, written by Kelly Gourdji and colleagues, is like upgrading from a blurry street map to a high-definition satellite image.

The Old Way: Previous studies tried to guess the jet's shape by looking at how bright the light was over time (the "light curve"). But this is like trying to guess the shape of a car just by listening to its engine noise. It's tricky because the engine noise depends on both the car's shape and how fast it's going.
The New Way: The authors used a technique called VLBI (Very Long Baseline Interferometry). Imagine linking radio telescopes across the entire Earth to create one giant telescope the size of the planet. This gives them a view so sharp they can see details the size of a milliarcsecond (which is like seeing a coin on the Moon from Earth).

They didn't just look at the "center" of the light blob; they analyzed the entire shape and movement of the radio waves directly, using a sophisticated computer model that simulates how a jet of gas behaves when it crashes into space. They combined data from four different time points:

Day 0: The exact moment of the crash (found by the Hubble Space Telescope).
Day 75 & 230: Radio observations from a network of telescopes.
Day 207: A high-resolution radio observation that acted as a crucial "checkpoint."

The Results: A Sharper Angle
By fitting their complex 3D models to this ultra-sharp data, they determined the viewing angle with much higher precision than before.

The Angle: They found the jet is tilted about 18 to 20 degrees away from our line of sight. (Previous estimates were a bit fuzzier, ranging from 14 to 28 degrees).
The Distance: Because they knew the angle better, they could calculate the distance more accurately. They found the crash happened about 44 million light-years away.
The Expansion Rate ( $H_0$ ): Using this new, precise distance, they calculated the Hubble Constant to be 65.5 km/s/Mpc.

Why This Matters
This result is a "tie-breaker" in the cosmic argument.

The "Early Universe" (Planck satellite) says the rate is 67.4.
The "Late Universe" (SH0ES team) says the rate is 73.0.
This new measurement says 65.5.

The authors note that their result is very close to the "Early Universe" number (within a tiny margin of error) but significantly lower than the "Late Universe" number. This suggests that perhaps the "Late Universe" measurements might be slightly off, or that our understanding of the universe's expansion needs a rethink.

The Caveats
The authors are careful to note that their result depends on a few things:

The Jet Model: They assume the jet behaves exactly like their computer simulations. If the real jet is weird or has a hidden energy source (like a leftover remnant of the crash glowing later than expected), the angle calculation could shift.
Local Motion: Galaxies don't just move with the expansion of the universe; they also have their own "peculiar velocities" (drifting due to local gravity). The authors had to guess and average out these local drifts, which adds some uncertainty.

In Summary
This paper is like taking a blurry photo of a spinning top and using a super-lens to see exactly how it's tilted. By doing so, they measured the distance to the crash more accurately and found that the universe is expanding at a rate that aligns more closely with the "Big Bang" measurements than the "nearby star" measurements. It doesn't solve the mystery of the Hubble Tension completely, but it provides a very strong, independent piece of evidence that leans toward the "slower" expansion rate.

Technical Summary: Revisiting GW170817 at Milliarcsecond Scale

Problem Statement
The detection of the binary neutron star merger GW170817 provided the first opportunity for a "standard siren" measurement of the Hubble constant ( $H_0$ ). However, the precision of this measurement is fundamentally limited by the strong covariance between the luminosity distance ( $D_L$ ) and the inclination angle ( $\iota$ ) of the binary system. While gravitational wave (GW) data alone cannot precisely constrain the inclination for face-on events like GW170817, the relativistic jet launched by the merger offers an electromagnetic handle on this geometry. Previous attempts to constrain the viewing angle ( $\theta_v$ ) using the afterglow light-curve have been limited by geometric degeneracies between the viewing angle and the jet opening angle ( $\theta_{cp}$ ). Conversely, Very Long Baseline Interferometry (VLBI) observations of the jet's superluminal proper motion provide orthogonal constraints but have historically relied on fitting discrete, semi-analytic jet models or limited subsets of the parameter space, often neglecting full spatial information (such as source size) or systematic uncertainties in a rigorous manner.

Methodology
This study presents a comprehensive Bayesian visibility-plane model-fitting framework designed to extract robust constraints on the viewing geometry of GW170817 and, subsequently, $H_0$ . The methodology integrates the following components:

Data Integration: The analysis incorporates all four available milliarcsecond-scale astrometric datasets:
- Two High Sensitivity Array (HSA) VLBI epochs (Days 75 and 230 post-merger) from Mooley et al. (2018).
- One Global-VLBI epoch (Day 207) from Ghirlanda et al. (2019), which the authors independently reduced to ensure consistency with the HSA calibration.
- A Hubble Space Telescope (HST) measurement of the kilonova position (Day 8), serving as the merger anchor (Day 0).
Hydrodynamical Modeling: Instead of relying on discrete, pre-computed models, the authors employ a scaling formalism applied to a continuum of new hydrodynamical jet afterglow models based on 2-D relativistic simulations (Govreen-Segal & Nakar 2023). These models cover four jet energy distributions (top-hat and power-laws with indices $b=3, 4, 6$ ). The scaling allows the models to be adjusted for arbitrary viewing angles ( $\theta_v$ ), opening angles ( $\theta_{cp}$ ), and observer times, utilizing the invariance of the models under specific parameter ratios.
Bayesian Inference: The framework operates directly on the observed radio interferometer visibilities using the GPU-accelerated software GALARIO and the nested sampler Dynesty (via Bilby). This approach avoids the information loss associated with image-plane deconvolution. The model includes nuisance parameters to rigorously handle systematic uncertainties, including astrometric offsets ( $\delta_{ra}, \delta_{dec}$ ) and flux scale uncertainties ( $\delta_F$ ) for each epoch.
Cosmological Constraints: To measure $H_0$ , the framework incorporates GW likelihoods (via a kernel density estimator of inclination and distance) and peculiar velocity corrections. The authors consider 28 combinations of galaxy group catalogs and peculiar velocity reconstruction maps, employing Bayesian Model Averaging (BMA) to marginalize over these systematic uncertainties.

Key Contributions

Continuum Parameter Space: The study moves beyond fitting discrete models to exploring a continuous range of jet geometries ( $\theta_v$ and $\theta_{cp}$ ), thereby avoiding biases introduced by selecting specific jet structures.
Visibility-Plane Fitting: By fitting directly to visibilities rather than image-plane centroids, the method utilizes the full spatial information of the data, including source size constraints, which is critical for breaking degeneracies.
Systematic Uncertainty Handling: The inclusion of nuisance parameters for astrometric and flux calibration uncertainties, alongside the BMA approach for peculiar velocity corrections, provides a more robust error budget than previous studies that often assumed fixed corrections.
Independent Global-VLBI Reduction: The authors performed an independent reduction of the Global-VLBI data to ensure frame consistency with the HSA data, identifying and correcting for astrometric offsets relative to previous reductions.

Results

Viewing Angle Constraints:
- When assuming a fixed luminosity distance of $D_L = 40.7$ Mpc (based on surface brightness fluctuation measurements), the analysis yields a viewing angle of $\theta_v = 18.3^\circ - 20.3^\circ$ (68% credible interval) when incorporating light-curve informed priors on the jet geometry.
- When marginalizing over $D_L$ and peculiar velocity corrections, the preferred viewing angle shifts slightly to $\theta_v = 16.8^\circ - 19.2^\circ$ .
Luminosity Distance: The direct fit for distance yields $D_L = 44.0 \pm 1.6$ Mpc. This is approximately $1\sigma$ larger than the $40.7 \pm 2.4$ Mpc value derived from the surface brightness fluctuation method, highlighting the sensitivity of viewing angle constraints to the assumed distance.
Hubble Constant ( $H_0$ ): The Bayesian Model Averaged measurement across 28 peculiar velocity correction scenarios is $H_0 = 65.5 \pm 4.4$ $H_{0} = 65.5 \pm 4.4$ km s $^{-1}$ $^{- 1}$ Mpc $^{-1}$ $^{- 1}$ .
- This value is within $0.5\sigma$ of the early-Universe Planck Collaboration value ( $67.4 \pm 0.5$ km s $^{-1}$ Mpc $^{-1}$ ).
- It is $1.7\sigma$ lower than the late-Universe SH0ES measurement ( $73.04 \pm 1.04$ km s $^{-1}$ Mpc $^{-1}$ ).

Significance and Claims
The paper claims that this independent late-Universe measurement of $H_0$ is more consistent with the early-Universe Planck result than with the SH0ES result. The authors suggest that, if taken at face value, this finding could challenge the necessity of non-standard cosmology models to resolve the Hubble tension. However, the authors remain modest regarding this claim, explicitly noting several caveats:

Model Mispecification: The late-time evolution of the afterglow light-curve ( $\gtrsim 350$ days) is not well-fit by standard synchrotron models, and the source of excess emission remains unknown. While the authors argue that their analysis (focused on earlier epochs) is dominated by geometry rather than model inaccuracies, they acknowledge that unmodeled physics could bias the results.
Jet Alignment: The analysis assumes the jet is launched along the orbital angular momentum axis. If a significant misalignment exists, it could non-linearly bias the $H_0$ estimate.
Peculiar Velocity Dominance: The authors identify peculiar velocity corrections as the dominant source of uncertainty in the $H_0$ measurement (contributing ~5% fractional uncertainty), rather than the viewing angle constraint itself.

The study concludes that while high-precision VLBI observations of jet afterglows remain a uniquely powerful tool for constraining inclination angles in the near term, achieving a definitive resolution to the Hubble tension will require a larger sample of bright sirens and improved modeling of jet physics and peculiar velocities.

More like this