🔭 astrophysics

Union3.1: Self-consistent Measurements of Host Galaxy Properties for 2000 Type Ia Supernovae

This study presents Union3.1, a self-consistent analysis of host galaxy properties for approximately 2000 Type Ia supernovae using DESI Legacy Imaging Surveys data and the Prospector code, which refines cosmological standardization parameters and reveals a reduced but still significant tension with the cosmological constant when combined with BAO and CMB measurements.

Original authors: Taylor J. Hoyt, David Rubin, Greg Aldering, Saul Perlmutter, Andrei Cuceu, Ravi Gupta

Published 2026-01-28

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Taylor J. Hoyt, David Rubin, Greg Aldering, Saul Perlmutter, Andrei Cuceu, Ravi Gupta

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

The Big Picture: Measuring the Universe's Expansion

Imagine you are trying to measure how fast the universe is expanding. To do this, astronomers use "standard candles"—objects in space that we know how bright they should be. If they look dimmer than expected, they are far away; if they look brighter, they are closer.

Type Ia supernovae (exploding stars) are the best standard candles we have. However, they aren't perfectly identical. Sometimes, a supernova looks a bit brighter or dimmer than its "twin" just because of where it lives. This paper is about fixing the map we use to measure these stars so we can get a clearer picture of the universe's expansion.

The Problem: The "Neighborhood" Effect

Think of a supernova like a person moving to a new city. If you want to know how far they traveled, you need to know their starting point. But there's a catch: the "neighborhood" (the host galaxy) they live in changes how bright they appear.

The Old Map (Union3): In previous studies, astronomers tried to guess the "neighborhood" of these stars. For nearby stars, they assumed they lived in massive, heavy galaxies (like a bustling metropolis). For distant stars, they assumed smaller galaxies (like a quiet village).
The Mistake: It turned out this guessing game was wrong. Many nearby stars were actually discovered by wide-angle surveys that look everywhere, not just at big cities. So, the old map was assigning "metropolis" addresses to stars that actually lived in "suburbs." This created a systematic error, like a ruler that was slightly stretched on one end and shrunk on the other.

The Solution: A New, Uniform Address Book

The authors of this paper decided to stop guessing and start measuring. They created Union3.1, a new, unified database for about 2,000 supernovae.

One Set of Glasses: Instead of using different telescopes and different math for different groups of stars, they used a single, consistent set of tools (called Prospector) to analyze the light from the galaxies hosting these supernovae.
The "Legacy" Survey: They used a massive, high-quality sky survey (the DESI Legacy Imaging Surveys) to get a clear, uniform picture of every galaxy.
The Result: They found that the old map was indeed biased. The nearby stars were being overestimated in mass (thought to be in bigger galaxies than they were), while the distant stars in other studies had the opposite problem.

The Fix: Smoothing Out the Ruler

Once they corrected these "neighborhood" errors, something magical happened: the measurements from different studies suddenly agreed much better.

Before: Two major studies (Union3 and Pantheon+) disagreed on the distance to nearby stars by more than 0.03 magnitudes (a tiny but significant amount in astronomy).
After: With the new, consistent data, that disagreement shrank to just 0.01 magnitudes. It's like taking two different rulers that were slightly warped and straightening them out so they now measure the exact same length.

What This Tells Us About the Universe

With this corrected ruler, the authors recalculated the fundamental properties of our universe:

The Matter Content: They found that the amount of matter in the universe ( $\Omega_m$ ) is about 34.4%. This is a slight shift from previous estimates, but it's a more precise number.
The Mystery of Dark Energy: The universe is expanding faster and faster, driven by something called "Dark Energy." Scientists want to know if this energy is constant (like a cosmological constant) or if it changes over time.
- Using their new data combined with other cosmic measurements (like the Cosmic Microwave Background), they found evidence that Dark Energy might not be constant.
- The evidence against a constant Dark Energy is now about 3.4 sigma (a statistical way of saying "very likely, but not quite a slam dunk"). This is a slight decrease from previous claims of 3.8 sigma, but it comes from a much more consistent and reliable dataset.

The Takeaway

This paper is less about discovering a new star and more about calibrating the tools. By realizing that the "addresses" of the supernovae were wrong in previous maps, the authors fixed the ruler.

They didn't just fix the numbers; they showed that when you treat all the data with the same consistent rules, the universe starts to make more sense. The disagreements between different research teams largely vanished, leaving us with a cleaner, more reliable picture of how our universe is growing.

In short: They realized the old map had the wrong street names, redrew the map with a single, consistent style, and found that the universe's expansion story is now much clearer and more consistent than before.

Technical Summary of Union3.1: Self-consistent Measurements of Host Galaxy Properties for 2000 Type Ia Supernovae

Problem Statement
The determination of cosmological parameters using Type Ia supernovae (SNe Ia) relies on an empirical correction for the "mass step," a discontinuity in SN residuals correlated with host galaxy properties, most commonly global stellar mass. Historically, compilations like Union3 and Pantheon+ have suffered from systematic inconsistencies in host mass estimates. These inconsistencies arise from heterogeneous data sources, varying spectral energy distribution (SED) fitting codes, and differing assumptions regarding initial mass functions (IMF) and star formation histories (SFH). Furthermore, previous analyses often relied on literature values or default filler masses (e.g., 11.0 dex for low-redshift SNe) which introduced biases. Specifically, the Union3 compilation assumed a default high mass for low-redshift SNe to account for galaxy-targeted searches, a premise that failed to reflect the prevalence of untargeted, blind surveys (e.g., ASAS-SN, ZTF) in recent years. These discrepancies have been identified as a dominant source of tension between different SN cosmology analyses and impact the combined interpretation of SNe with Baryon Acoustic Oscillations (BAO) and the Cosmic Microwave Background (CMB).

Methodology
The authors present Union3.1, an updated compilation of approximately 2000 SNe Ia, featuring self-consistently derived host galaxy properties. The methodology involves:

Homogeneous Photometry: The study utilizes aperture-matched, homogeneously reduced optical–infrared photometry from the DESI Legacy Imaging Surveys (LS DR10). This includes $g, r, i, z$ bands and unWISE infrared data ( $W1, W2, W3, W4$ ).
UV Supplementation: For low-redshift hosts, the optical data is supplemented with GALEX FUV and NUV photometry from the GUVcat AIS catalog to better constrain recent star formation and break dust-metallicity-age degeneracies.
SED Fitting with Prospector: Global galaxy properties are inferred using the Prospector stellar population synthesis and SED-fitting code. The analysis adopts a single, consistent set of assumptions:
- A delayed exponential (delay- $\tau$ ) plus burst star formation history model.
- A consistent IMF and dust extinction priors.
- The dynesty nested sampler is used with increased sampling parameters (target $N_{eff}$ increased from 10,000 to 30,000) to ensure robust posterior distributions.
Host Identification: Host galaxies are identified by matching SN coordinates to the LS Tractor catalog. For cases where the LS catalog is incomplete (e.g., high-redshift GOODS fields or specific deep fields), the authors supplement the sample with published masses from other studies (e.g., Neill et al. 2012, Meyers et al. 2012, SNLS, DES, SDSS), ensuring these external sources agree with the homogeneous LSTP (Legacy Survey Tractor + Prospector) masses to within $\sim 0.2$ dex.
Database Updates: Beyond host masses, Union3.1 incorporates several corrections to the Union3 database:
- Correction of a double-application bug in Milky Way foreground reddening (Schlafly & Finkbeiner 2011 renormalization).
- Fixing a parsing error in SDSS-SN host masses that previously caused random swaps and underestimations.
- Identification and merging of four duplicate SN entries.
- Updates to ESSENCE sample metadata (redshifts and extinction) based on Narayan et al. (2016).
Cosmological Analysis: The updated data is processed through UNITY1.7, an improved version of the Bayesian hierarchical framework. This update includes a refined outlier model and an explicit fit parameter for the fiducial mass where the luminosity step occurs.

Key Results

Host Mass Discrepancies: The study finds that the host masses of $z < 0.10$ SNe in the original Union3 were, on average, overestimated relative to the rest of the sample due to the incorrect default mass assumption. Conversely, Pantheon+ low-redshift masses were underestimated relative to high-redshift ones.
Distance Modulus Agreement: After correcting for these systematic mass offsets, the average distance modulus estimated for low-redshift SNe in Union3 and Pantheon+, which were previously discrepant by $> 0.03$ mag, now agree to within 0.01 mag.
Standardization Parameters: The uncertainties on all standardization parameters in the UNITY analysis shrink to 0.6–0.9 $\times$ their previous sizes, indicating a more precise characterization of the SN population.
Cosmological Constraints:
- Flat $\Lambda$ CDM: Using SNe alone, the study derives $\Omega_m = 0.344^{+0.026}_{-0.025}$ , representing a $-0.3\sigma$ shift from the Union3 result.
- Flat $w_0$ – $w_a$ CDM: Combining SNe with DESI DR2 BAO and Planck CMB data, the study finds $w_0 = -0.719 \pm 0.084$ and $w_a = -0.95^{+0.29}_{-0.26}$ . This corresponds to 3.4 $\sigma$ evidence against a cosmological constant (down from 3.8 $\sigma$ in the DESI-DR2+Planck+Union3 analysis).
- Combined Probes: Updating the DESI-DR2+Planck+SN analysis with the new Pantheon+ corrections and the DES-SN5YR Dovekie recalibration yields 3.2 $\sigma$ (up from 2.8 $\sigma$ ) and 3.4 $\sigma$ (down from 4.2 $\sigma$ ) evidence against a cosmological constant in the $w_0$ – $w_a$ plane, depending on the specific combination of probes.

Significance and Claims
The paper claims that the primary contribution of Union3.1 is the establishment of a self-consistent system for host galaxy mass estimates across the full redshift range of current SN compilations. By eliminating the systematic offsets caused by heterogeneous mass estimation methods and incorrect default assumptions, the study significantly improves the consistency between independent SN analyses (Union vs. Pantheon+).

The authors emphasize that while the updated results show improved internal consistency and reduced uncertainties, disagreements over uncertainty estimation remain between different analysis groups. The study does not claim to resolve the Hubble Tension or definitively prove evolving dark energy but rather provides a more robust foundation for such inferences by removing a known, significant source of systematic bias (host mass estimation) from the SN cosmology pipeline. The results suggest that previous tensions in cosmological parameters derived from different SN compilations were partially driven by these uncorrected host mass systematics.