Physically-motivated priors in the local distance… — Plain-Language Explanation

Imagine the universe is a giant, expanding balloon. Scientists want to know exactly how fast it's inflating. This speed is called the Hubble Constant ( $H_0$ ).

For years, there has been a massive argument in the scientific community about this number. It's like two groups of people measuring the speed of a car:

Group A (The "Early Universe" team) looks at the very first light in the universe (the Cosmic Microwave Background) and calculates the speed should be about 67.
Group B (The "Local Universe" team) looks at nearby stars and exploding stars (Supernovae) and calculates the speed should be about 73.

The difference is small in numbers, but in science, it's a huge gap. It's a "5-sigma" tension, meaning there's only a 1 in 3.5 million chance this is just a fluke. Most scientists thought this meant our understanding of physics was broken and needed new laws of nature to fix it.

The "Ruler" Problem

This paper suggests the problem might not be the physics, but the ruler the Local team is using.

To measure the speed of the universe's expansion, astronomers use a "distance ladder."

The Bottom Rung: They measure the distance to nearby stars (Cepheids) using parallax (like how your thumb shifts when you look at it with one eye, then the other).
The Middle Rung: They use those stars to calibrate the brightness of nearby exploding stars (Supernovae).
The Top Rung: They use those calibrated explosions to measure how fast the universe is expanding far away.

The Hidden Bias: The "Flat" Assumption

The authors of this paper found a subtle but powerful mistake in how the "Local" team set up their math.

When calculating the distances, the team used a standard statistical assumption called a "flat prior." In everyday language, this is like assuming that in the universe, every distance is equally likely to be found.

The Analogy:
Imagine you are throwing darts at a giant, circular target that represents space.

If you assume a "flat prior" on distance, you are essentially saying, "I'm equally likely to hit a dart at 1 meter away as I am at 100 meters away."
But space isn't flat. As you go further out, the volume of space gets bigger and bigger (like the layers of an onion). There is way more space at 100 meters than at 1 meter.
Therefore, if you are looking for stars, you are statistically much more likely to find them far away than close by.

The paper argues that the "Local" team's math accidentally over-weighted the closer stars and under-weighted the farther ones. Because closer stars make the universe look like it's expanding faster to match the observations, this bias pushed their calculated speed up to 73.

The Fix: A "Physically Motivated" Ruler

The authors, Marcus Högås and Edvard Mörtsell, decided to fix the ruler. Instead of assuming every distance is equally likely, they applied a "physically motivated prior."

They told the math: "Remember, there is more space further out. We should expect to find more stars at larger distances."

They also made a conservative change to how they handled a tiny error in the satellite data (Gaia) used to measure star positions, letting the data speak for itself rather than forcing it to fit a specific guess.

The Result: The Tension Melts Away

When they ran the numbers with this new, more realistic ruler:

The calculated speed of the universe dropped from 73 down to 70.6.
The gap between the "Local" team and the "Early Universe" team shrank from a massive 5-sigma disagreement to a tiny 2-sigma difference.

In simple terms, the "5-sigma" crisis (which sounded like the universe was broken) turned out to be mostly a mathematical illusion caused by how they assumed distances were distributed.

The Takeaway

The paper concludes that the "Hubble Tension" might not require new, exotic physics. Instead, it highlights that statistical assumptions—the invisible rules we use to interpret data—can have a huge impact. By simply acknowledging that "there is more space further out," the conflict largely disappears.

It's a reminder that sometimes, when two measurements disagree, the answer isn't that the universe is weird; it's that our measuring tape was slightly bent.

1. Problem Statement

The paper addresses the Hubble tension, the persistent $5\sigma$ discrepancy between the locally measured Hubble constant ( $H_0$ ) and the value inferred from the Cosmic Microwave Background (CMB) under the standard $\Lambda$ CDM model.

Current Status: The SH0ES team's Cepheid–Type Ia Supernova (SN Ia) distance ladder yields $H_0 = 73.0 \pm 1.0$ km/s/Mpc, while Planck CMB data infers $H_0 = 67.4 \pm 0.5$ km/s/Mpc.
The Gap: While much research focuses on new physics or unmodeled systematics, the statistical assumptions underlying the inference—specifically the choice of priors in Bayesian analysis—have received less attention.
Specific Issue: The standard SH0ES analysis adopts flat priors on distance moduli ( $\pi(\mu) = \text{const}$ ). Since distance modulus $\mu \propto \ln D$ , a flat prior on $\mu$ implies an inverse prior on physical distance ( $\pi(D) \propto D^{-1}$ ). This statistically upweights smaller distances, systematically biasing the inferred $H_0$ toward higher values. Desmond et al. (2025) previously noted that a homogeneous, volume-limited population of distance indicators should instead follow a prior $\pi(D) \propto D^2$ .

2. Methodology

The authors perform a comprehensive Bayesian recalibration of the fourth-iteration SH0ES distance ladder using the following approach:

Dataset: They utilize the standard SH0ES dataset (42 calibrator SNe Ia in 37 host galaxies, anchored by the Milky Way (MW), Large Magellanic Cloud (LMC), and NGC 4258), combined with Gaia EDR3 parallaxes for MW Cepheids.
Joint Fit Framework: Unlike standard analyses that treat MW Cepheids as external constraints on the absolute magnitude, this study incorporates MW Cepheids directly into a single joint likelihood. This allows for a uniform application of priors across all distance indicators (MW, LMC, NGC 4258, and extragalactic hosts).
Likelihood Construction:
- A linear Gaussian component handles the Cepheid Period-Luminosity Relation (PLR) and SN Ia magnitude relations.
- A non-linear component specifically models the Gaia parallaxes for MW Cepheids, including a residual parallax zero-point offset ( $z_p$ ).
Prior Modifications (The "Phys-prior" Calibration):
1. Distance Modulus Prior: Replaces the flat prior ( $\pi(\mu) = \text{const}$ ) with a volume-weighted prior derived from the assumption of a homogeneous, volume-limited population. This corresponds to $\pi(D) \propto D^2$ , which in distance modulus space is $\ln \pi(\mu) = 0.6 \ln(10) \mu$ . This assigns greater weight to larger distances.
2. Parallax Offset Prior: Replaces the informative Gaussian prior on the Gaia residual parallax offset ( $z_p$ ) with a flat prior ( $\pi(z_p) = \text{const}$ ). This is a conservative choice reflecting the remaining uncertainty in Gaia EDR3 corrections, allowing the distance ladder data to constrain $z_p$ directly rather than relying on external assumptions.
Sampling: The posterior distribution is sampled using the emcee affine-invariant ensemble MCMC sampler.

3. Key Contributions

Unified Bayesian Framework: The first application of a single joint fit that includes MW Cepheids directly, enabling consistent prior application across the entire ladder.
Demonstration of Prior Sensitivity: The paper provides quantitative evidence that the choice of priors on distance moduli is not an "innocuous default" but a dominant factor in the inferred value of $H_0$ .
Conservative Parallax Treatment: By removing the informative Gaussian prior on the Gaia zero-point offset, the authors allow the data to self-consistently determine the offset, revealing a shift toward negative values consistent with independent literature.

4. Results

The application of physically motivated priors leads to the following results:

Shift in $H_0$ : The inferred Hubble constant shifts from the reference value of $H_0 = 72.7 \pm 1.0$ km/s/Mpc (SH0ES-ref) to $H_0 = 70.6 \pm 1.0$ km/s/Mpc (Phys-prior).
Reduction in Tension: The discrepancy with the Planck CMB value ( $67.4 \pm 0.5$ km/s/Mpc) is reduced from $5\sigma$ to $2\sigma$ .
Decomposition of the Shift:
- The volume-weighted distance prior ( $\pi(D) \propto D^2$ ) accounts for a downward shift of $-1.7$ km/s/Mpc.
- The flat prior on the parallax offset accounts for an additional $-0.2$ km/s/Mpc shift.
Distance Modulus Shifts: The new priors cause a coherent positive shift in all inferred distance moduli.
- SN Ia host galaxies: Mean increase of 0.066 mag ( $\approx 3.0\%$ increase in distance).
- MW Cepheids: Mean increase of 0.057 mag ( $\approx 2.6\%$ increase in distance).
- This is consistent with Hubble's Law ( $cz = H_0 D$ ): a 3% increase in distance implies a 3% decrease in $H_0$ .
Parallax Offset Constraint: The data, under the flat prior, constrains the Gaia residual offset to $z_p = -26 \pm 5$ $\mu$ as, which is consistent with independent estimates favoring negative offsets.
Robustness: The shift is driven primarily by the distance priors. Variations in priors for other parameters (e.g., PLR slopes, absolute magnitudes) have negligible effects ( $< 0.1$ km/s/Mpc).

5. Significance

Re-evaluation of Systematics: The paper argues that the Hubble tension may be significantly driven by statistical assumptions (priors) rather than new physics or unmodeled astrophysical systematics.
General Applicability: The findings are not limited to the Cepheid-SN Ia ladder. The authors note that any distance-ladder method (TRGB, Mira variables, SBF, Tully-Fisher, etc.) relies on calibrating distances as model parameters. Therefore, the choice of priors on these distances is likely a critical, yet often overlooked, factor in all local $H_0$ measurements.
Methodological Shift: The study suggests that future analyses should move away from "default" flat priors on distance moduli toward physically motivated priors that reflect the geometry of the universe (volume effects) and the completeness of the sample.

In conclusion, Högås and Mörtsell demonstrate that a rigorous, physically motivated treatment of priors in the local distance ladder naturally lowers the inferred Hubble constant, bringing it into much closer agreement with early-Universe CMB measurements and significantly alleviating the Hubble tension.

Physically-motivated priors in the local distance ladder significantly reduce the Hubble tension