PAC in DESI. II. Galaxy-halo connection into the 10 6 M frontier

arXiv:2603.29331 · astro-ph.GA, astro-ph.CO · Submitted 2026-08-19 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Astrophysics Radio. Generated commentary on the latest astrophysics papers.

Vera: Next we'll be talking about the paper "PAC in DESI. II. Galaxy-halo connection into the 10 6 M frontier".

Jocelyn: The paper was written by the authors from Abastumani Astrophysical Observatory and Department of Physics, Kansas State University and Faculty of Natural Sciences and Medicine, Ilia State University and CIEMAT and University of Michigan and National Astronomical Observatories, Chinese Academy of Sciences.

Vera: Stay tuned as we take you through the paper and discuss its implications.

Jocelyn: We also have Subrahmanyan with us today — guest researcher.

Vera: Alright, let's get started.

Title: Vera: We're looking at a massive new paper today titled "PAC in DESI. II. Galaxy-halo connection into the six M frontier" by Kun Xu and a very large collaborative team.

Jocelyn: That six solar mass number in the title really jumps out at me.

Subrahmanyan: It represents an incredibly low mass scale for this kind of work, Vera.

Vera: It does, and it's essentially pushing our observational reach into the realm of the smallest possible dark matter structures.

Jocelyn: How do they even manage to target something that small with a survey like DESI?

Subrahmanyan: They aren't just looking at individual tiny dots; they're using the statistical power of the Photometric objects Around Cosmic webs method, or PAC.

Vera: Right, the PAC method is what allows them to bridge the gap between the spectroscopic data from DESI and the much deeper photometric data from DECaLS.

Jocelyn: So the title is telling us they've moved from general galaxy surveys into this ultra-low-mass regime.

Subrahmanyan: It's a transition from studying the big, bright galaxies to investigating the very building blocks of the cosmic web.

Vera: The authors are clearly aiming to test whether our current dark matter models hold up when we look at these tiny, faint scales.

Jocelyn: It sounds like they're trying to find the limit where the dark matter scaffolding actually supports star formation.

Subrahmanyan: That's a perfect way to put it, Jocelyn.

Vera: This paper is part of a series, which tells me they've been refining this specific mathematical approach for a while now.

Jocelyn: I'm curious to see if this second installment actually hits the targets they've set in the title.

Subrahmanyan: The scale they're targeting suggests they're looking for the very first signatures of galaxy assembly.

Vera: Let's move on to what they actually found once they applied this method to the data.

Summary: Vera: Now that we've seen the title, we need to talk about the actual results from "PAC in DESI. II. Galaxy-halo connection into the six M frontier."

Jocelyn: I saw something in the abstract about an upturn in efficiency, which seems counterintuitive.

Subrahmanyan: It's a fascinating result where the star-formation efficiency actually rises in these smaller haloes.

Vera: They found this clear upturn at around ten h-one M as they moved toward lower masses.

Jocelyn: So, instead of star formation dying out as the haloes get smaller, it actually gets more efficient for a while?

Subrahmanyan: That's exactly what the data is suggesting, which challenges some of our simpler assumptions.

Vera: They also discovered that central red galaxies are the ones that really dominate this low-mass regime.

Jocelyn: That's a huge detail because it tells us about the color and state of these dwarf galaxies.

Subrahmanyan: It points toward a specific history where these galaxies were active and then got shut down.

Vera: The paper proposes a hypothesis that star formation was actually much higher before reionization happened.

Jocelyn: Are you saying these galaxies formed their stars early and then just stopped?

Subrahmanyan: Yes, the UV background from reionization likely quenched them, leaving behind these red central dwarfs.

Vera: And the mass of these red dwarfs is actually much larger than what our current galaxy formation models usually predict.

Jocelyn: That's a significant discrepancy between what we see in the sky and what our simulations tell us.

Subrahmanyan: It means our current models might be missing a crucial piece of early-universe physics.

Vera: They even managed to set upper bounds on the smallest possible haloes, like eight point eight zero h-one M at the three-sigma level.

Jocelyn: So they're literally defining the floor of where dark matter haloes must exist.

Subrahmanyan: It's a very powerful way to constrain the physics of the early universe.

Vera: To understand how they got these precise numbers, we have to look at the methodology they used.

Improvements: Vera: The technical side of "PAC in DESI. II. Galaxy-halo connection into the six M frontier" is where the real heavy lifting happens.

Jocelyn: I was wondering how they handled the fact that you can't easily get a spectrum for every single tiny dwarf galaxy.

Subrahmanyan: That's the spectroscopic bottleneck that makes this work so impressive.

Vera: They bypassed it by using the PAC method to combine DESI's spectroscopy with the deep DECaLS photometric imaging.

Jocelyn: Did they have to deal with a massive amount of noise from foreground or background objects?

Subrahmanyan: They certainly did, but they used specific color cuts and masking to clean up the signal.

Vera: They also employed a Subhalo Abundance Matching framework, which they linked to the Jiutian N-body simulations.

Jocelyn: So the simulations provided the dark matter structure that they then populated with galaxies?

Subrahmanyan: Precisely, using those high-resolution simulations allowed them to model the connection across a huge mass range.

Vera: They even went a step further by testing for mass-dependent scatter and galaxy assembly bias.

Jocelyn: That sounds like they were trying to account for every possible way the model could be wrong.

Subrahmanyan: It's a very rigorous approach to ensure the upturn they found wasn't just a modeling artifact.

Vera: They even checked if using a different cosmology would change their conclusions.

Jocelyn: And it didn't seem to break the main results, did it?

Subrahmanyan: No, the core findings remained robust even under those different assumptions.

Vera: It's a very complete package of observations and modeling.

Jocelyn: We should probably wrap this up and talk about the bigger implications.

Conclusion: Vera: We've covered a lot of ground with "PAC in DESI. II. Galaxy-halo connection into the six M frontier."

Jocelyn: It really feels like this paper has given us a new map for the low-mass universe.

Subrahmanyan: It's more than a map, Jocelyn; it's a challenge to our fundamental understanding of how the first galaxies survived reionization.

Vera: The way they've linked the star-formation efficiency to these early quenching events is quite profound.

Jocelyn: It makes me wonder what the next generation of surveys like LSST will reveal about these red dwarfs.

Subrahmanyan: We're going to be able to see even deeper into that six solar mass frontier.

Vera: This work sets a high bar for how we combine photometric and spectroscopic datasets in the future.

Jocelyn: It's been a fascinating discussion, and I'm ready to see where this research leads next.

Subrahmanyan: The implications for dark matter models are going to be felt for a long time.

Vera: Thank you both for joining me to unpack this incredible paper.

Jocelyn: We'll be back next time to tackle a completely different part of the cosmos.

Subrahmanyan: I'm looking forward to it.

Vera: Goodbye for now, everyone.

Abastumani Astrophysical Observatory · Department of Physics, Kansas State University · Faculty of Natural Sciences and Medicine, Ilia State University · CIEMAT · University of Michigan · National Astronomical Observatories, Chinese Academy of Sciences

astro-ph.GA, astro-ph.CO

Submitted: 2026-08-19

Updated: 2026-08-21

Comments: 39 pages, 38+12 figures. Main results in Figure 34, 36 and 38. Published in MNRAS

Journal ref: Mon Not R Astron Soc (2026)

DOI: 10.1093/mnras/stag1487

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 68/100

The gist: The paper presents detailed comparative analyses using DESI data to investigate the galaxy-halo connection, focusing on constraints across different mass regimes.

Key concepts

PAC method
Photometric objects Around Cosmic webs (PAC) is a method used to bridge the gap between spectroscopic data from DESI and deeper photometric data from DECaLS. This allows researchers to study faint, low-mass structures that are difficult to observe otherwise.
Galaxy-halo connection
This concept examines how galaxies form and reside within dark matter structures (halos). The paper investigates this connection at ultra-low masses, aiming to test if current dark matter models accurately predict the smallest building blocks of the cosmic web.
Reionization
The process by which the universe became transparent. The hosts discuss how the UV background from reionization likely quenched star formation in early galaxies, leaving behind red central dwarfs that are key to understanding galaxy evolution.

Terminology

Summary

The paper presents detailed comparative analyses using DESI data to investigate the galaxy-halo connection, focusing on constraints across different mass regimes. The results are primarily visualized through three figures (Figure D5, Figure D6, and Figure D7), which compare model predictions derived from varying parameters—specifically M* spec, M* photo, 2 wp, and r p —while quantifying the goodness of fit using metrics such as chi/N and chi 2/N.

Figure D5 Analysis:

This figure compares results for cases with varying M* spec values, ranging from 10 11.0 M to 10 11.6 M. The analysis involves plotting metrics against the parameter 2 wp (h squared Mpc-2). For instance, when comparing models with M* spec values of 10 11.0 M, the corresponding chi/N values are shown for different 2 wp ranges (e.g., 0.85, 1.04, 1.37, 0.88), and the chi 2/N values follow a similar pattern (e.g., 0.46, 0.43, 0.42, 0.59). The figure also presents comparisons involving M* photo values across ranges such as [10 10.3, 10 11.1] M.

Figure D6 Analysis:

This section examines the relationship using different M* spec values, specifically comparing models with M* spec: 10 9.4 M, M* spec: 10 9.6 M, M* spec: 10 9.8 M, and M* spec: 10 10.0 M. The analysis here is conducted across varying values of 2 wp (h squared Mpc-2). The goodness-of-fit metrics are reported for multiple combinations, such as chi/N values of 0.64, 0.71, 0.44, 0.62 when paired with a specific 2 wp range of 10-2 to 10-3 h squared Mpc-2. Furthermore, the figure includes comparisons involving M* photo values ranging from 10 11.2 to 10 11.8 M.

Figure D7 Analysis:

The final figure analyzes the connection using varying values of r p (h-1 Mpc). The comparisons are shown for models with different M* spec values, including M* spec: 10 10.6 M, M* spec: 10 10.8 M, and M* spec: 10 11.2 M. The metrics chi/N and chi 2/N are reported for these cases across different ranges of r p. For example, when considering the relationship between M* spec and M* photo, the corresponding chi 2/N values are presented for specific ranges of r p, such as 0.66, 0.90, and 0.46 for one set of parameters.

In summary, the paper systematically presents model constraints by calculating and comparing chi/N and chi 2/N values across three distinct parameter spaces— 2 wp, r p, and varying M* spec —to constrain the galaxy-halo connection into the 10 6 M frontier using DESI data.

Improvements for AI systems

The core challenge presented by these figures is performing robust, multi-dimensional parameter inference (= M*, 2 wp, r p) across complex, non-linear likelihood landscapes, often requiring model selection (comparing M* spec vs. M* photo). My improvements focus on enhancing the AI's ability to navigate and interpret these high-dimensional astrophysical parameter spaces with quantifiable uncertainty.

Improvement: Replace standard Maximum Likelihood Estimation (MLE) or simple Markov Chain Monte Carlo (MCMC) methods with a specialized Hierarchical Bayesian Inference Engine architecture. This engine must treat the parameters not as independent inputs, but as nested levels of physical processes.

Technical Specificity:

  • Latent Variable Modeling: Implement Variational Autoencoders (VAEs) or Normalizing Flows within the inference loop to learn a compressed, low-dimensional latent space representation of the full parameter set. This addresses the computational cost and degeneracy inherent in high-dimensional scans.

  • Model Selection Integration: The H-BIE must natively integrate model comparison metrics (chi squared, AIC/BIC equivalents) as a loss function component, allowing it to dynamically weigh the relative evidence for competing physical models (e.g., comparing the fit quality of M* spec vs. M* photo) without requiring manual post-hoc analysis.

  • Uncertainty Quantification: The system must output full posterior probability distributions, not just point estimates, providing rigorously calculated credible intervals for all derived parameters (e.g., the range of M* given the observed chi 2/N contours).

What the Improved AI System Can Do:

The H-BIE can perform simultaneous, self-correcting inferences across multiple coupled astrophysical observables. It can determine not only what the best parameter set is, but also how confident it is in that determination relative to alternative physical models, significantly reducing systematic and statistical error propagation in complex analyses.

Abstract

Understanding dwarf galaxy formation is crucial for testing dark matter models and reionization physics. However, constructing stellar-mass complete spectroscopic samples at low masses is increasingly difficult, and the potential existence of a local void complicates studies in an average environment. The Photometric object Around Cosmic webs (PAC) method, which combines deep photometric and spectroscopic data to measure the excess surface density 2w(r) of photometric objects around spectroscopic tracers, offers a promising path forward. We model 349 2w(r) measurements from DESI Y1 BGS and DECaLS, reaching M*=10 6.4, M, using a stellar mass-halo mass relation (SHMR)-based subhalo abundance matching framework applied to two high-resolution N-body simulations from the Jiutian suite. The resulting SHMR is constrained down to M h 10 8.0,h-1 M, revealing a clear upturn at about10 10.0,h-1 M toward lower masses, indicating rising star-formation efficiency (SFE) in small haloes. This feature persists under extensions of the model that allow mass-dependent scatter, reionization-induced suppression of the halo occupation fraction, galaxy assembly bias, and alternative cosmologies. Combining with the results from Paper I, we find that central red galaxies dominate the low-mass regime. Our results motivate a hypothesis in which SFE is significantly higher than previously thought prior to reionization, enabling relatively massive galaxies to form in small haloes. These systems are subsequently quenched by the UV background, producing the central red dwarf galaxies observed. Finally, we obtain 3σ and 5σ upper mass bounds of 10 8.80,h-1 M and 10 10.24,h-1 M on the smallest haloes required to exist.

Sources

Related papers