Multiprobe cosmology forecasts from halo-occupation-distribution-based forward modeling of galaxy and void statistics

arXiv:2504.08221 · astro-ph.CO · Submitted 2026-08-12 · 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 "Multiprobe cosmology forecasts from halo-occupation-distribution-based forward modeling of galaxy and void statistics".

Jocelyn: The paper was written by Andrés N. Salcedo, Alice Pisani and Nico Hamaus from University of Arizona and Aix-Marseille University and Princeton University and Universität München.

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

Summary: Vera: So, we've established the power of combining probes, but what specific measurements are they using to make these forecasts?

Jocelyn: They are analyzing four specific statistics that make up their data vector. These include the void size function, nv, and projected galaxy auto-correlation function, wp,gg.

Subrahmanyan: And then they add the projected void-galaxy cross-correlation function, wp,vg, and the void excess surface density profile from weak lensing, which is called vm.

Vera: It’s interesting that they are using projected statistics like wp,vg and also looking at real-space measurements like the void size function.Jocelyn: It seems they want to capture as much physical information as possible, not just the easy stuff.

Subrahmanyan: The combination of these four probes is designed to break down degeneracies that usually plague cosmological measurements.

Vera: It sounds like they are trying to get a very precise measurement of our universe's density and growth rate simultaneously.

Jocelyn: Subrahmanyan is right; we’re not just measuring one thing, we’re using the whole picture to constrain m and sigma eight.

Subrahmanyan: They found that using these four observables in their fiducial scenario yields constraints of.5 percent and.8 percent on m and sigma eight, which is quite tight for a survey like DESI-Y5.

Vera: That's incredibly competitive with the current state of the art, given how much complexity they are including in their modeling.

Jocelyn: The data vector is really doing the heavy lifting here, combining structure and anti-structure measurements to yield these impressive results.

Improvements: Vera: We've seen what they measure, but what does this method improve upon compared to just looking at galaxies?

Subrahmanyan: The biggest improvement is how it handles the galaxy-halo connection, which is often a major source of uncertainty in cosmology.

Jocelyn: They aren't just assuming a fixed relationship between matter and galaxies; they are modeling it dynamically using the HOD framework.

Vera: And this allows them to incorporate "galaxy assembly bias," where the probability of having a galaxy in certain environments depends on that environment's detailed history.

Subrahmanyian: Exactly, and since voids are in the lowest density regions, they are particularly sensitive to this bias.

Jocelyn: The paper shows that by marginalizing over these HOD parameters—like sigma M and M—they can achieve tight results.

Vera: It’s fascinating how the sensitivity analysis shows that variations in sigma M can fracture large voids into multiple smaller ones, which is a very specific physical process.

Subrahmanyian: That's the power of their approach; they are not just counting voids, they are understanding *why* those voids change size.

Jocelyn: And the results suggest that combining wp,gg and nv alone can give.9 percent and.1 percent constraints on m and sigma eight which is a huge improvement over using either observation by itself.

Vera: It seems like the combination of improving the modeling while also combining probes is what drives the most constraint power here at all scales of structure.

Conclusion: Vera: So, we've talked about how they do it and what the results are, but what does this mean for our understanding of the universe?

Subrahmanyan: The ability to constrain m and sigma eight so tightly is crucial because it helps us break down degeneracies that exist between those two parameters in traditional surveys.

Jocelyn: And given their results, the constraints are sharp enough that they can provide valuable cross-checks for other probes, like those coming from weak lensing or redshift-space distortions.

Vera: I remember them mentioning the possibility of "modified gravity" being tested through void kinematics—is that a big deal?

Subrahmanyan: Yes, since voids are such pristine regions, they are less likely to be screened by matter effects compared to clusters.

Jocelyn: That makes them excellent test cases for theories that deviate from standard.

Vera: The paper suggests that while these results are impressive for real-space measurements, future work in redshift space might yield even tighter constraints due to the Alcock-Paczynski effect.

Subrahmanyian: That's true; the full impact of Redshift Space Distortions, which is why we see a slight difference between our real-space and redshift-space expectations, can be accounted for with advanced AI modeling.

Jocelyn: It's encouraging to know that even without those effects accounted for, we are already seeing such tight constraints on the structure of the cosmos.

Final Wrap-up: Vera: We’ve really covered a lot today about "Multiprobe cosmology forecasts from halo-occupation-distribution-based forward modeling of galaxy and void statistics."

Jocelyn: It’s clear that by combining void statistics with the galaxy auto-correlation, this approach is proving extremely powerful.

Subrahmanyian: The ability to break those degeneracies between HOD parameters while simultaneously getting tight cosmological constraints is a major achievement in modeling.

Vera: It shows us how much we can learn about both the distribution of matter and the processes that form galaxies within it.

Jocelyn: I think this sets a very high bar for how we should approach future spectroscopic surveys, like DESI-Y5.

Subrahmanyian: We can't wait to see how these detailed HOD models are applied to upcoming real data.

Vera: It’s been a fascinating look into the sky today, and I think listeners will find this research very exciting too.

Andrés N. Salcedo, Alice Pisani, Nico Hamaus

University of Arizona · Aix-Marseille University · Princeton University · Universität München

astro-ph.CO

Submitted: 2026-08-12

Comments: 20 pages, 8 main figures, 3 tables, 1 appendix figure. Version 2, accepted by PRD

DOI: 10.1103/mydr-vtgc

Code: https://github.com/DESI-UR/VAST

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

Importance score: 69/100

The gist: The paper, "Multiprobe cosmology forecasts from halo-occupation-distribution-based forward modeling of galaxy and void statistics," investigates the potential of combining void statistics with galaxy

Key concepts

Halo-Occupation-Distribution (HOD)
The HOD framework is used to model the connection between galaxies and underlying matter. Instead of assuming a fixed relationship, it dynamically models how galaxies form within halos, which helps reduce uncertainty in cosmological measurements.
Void Size Function ($\text{n}_v$)
This statistic measures the distribution of sizes for cosmic voids (large empty regions). Analyzing this function allows researchers to understand the physical processes that change void sizes, providing a unique probe of the universe's structure.
Cosmological Degeneracies
These are situations where multiple cosmological parameters (like $\Omega_m$ and $\sigma_eight$) can produce similar observational results. Combining multiple, independent probes helps break these degeneracies, allowing for more precise measurements of the universe's properties.
Galaxy Auto-correlation ($\text{w}_{p,gg}$)
This measures how clustered galaxies are in space. By analyzing this projected statistic alongside void data and cross-correlations, researchers can gain a comprehensive picture of matter distribution across different cosmic structures.

Terminology

Summary

The paper, Multiprobe cosmology forecasts from halo-occupation-distribution-based forward modeling of galaxy and void statistics, investigates the potential of combining void statistics with galaxy clustering to achieve tight cosmological constraints using data from a simulated Dark Energy Spectroscopic Instrument (DESI) Year 5 survey.

Motivation and Methodology

The authors note that the large under-dense regions in the cosmological matter density field, known as cosmic voids, are powerful probes of cosmology but their potential is currently under-exploited. They argue that combining void and galaxy summary statistics is a particularly powerful probes of both cosmology and the galaxy-halo connection through self-calibration of the void-galaxy relation, as this combination breaks degeneracies in the galaxy-halo connection and cosmology relative to the case of galaxy clustering alone.

The study utilizes a grid of cosmological N-body simulations from the AbacusSummit suite. The methodology involves modeling galaxies within these simulated halos using a flexible Halo Occupation Distribution (HOD) model, which includes both central and satellite galaxy assembly bias. This extended HOD framework allows for joint modeling of void and galaxy statistics in order to self-calibrate the void-galaxy relation.

The primary observables (the data vector) analyzed are:

  1. Void size function (n v): The comoving space density of voids as a function of their radius R eff.

  2. Projected void-galaxy cross-correlation function (w p,vg).

  3. Projected galaxy auto-correlation function (w p,gg).

  4. Void excess surface density profile (vm) measured by weak lensing.

The voids are identified in the the HOD galaxy distribution using the V2 void finding algorithm, which is based on the Zobov algorithm. The effective radius of a void is defined as R eff = 3V over 4 pi 1/3, where V is the total volume of its constituent Voronoi cells.

Sensitivity Analysis to HOD and Cosmological Parameters

The authors conducted a detailed sensitivity analysis on the observables:

  • Void Size Function (n v): This statistic is sensitive to parameters governing the galaxy-halo connection. Increasing sigma M increases (decreases) the abundance of voids with R eff 50 h-1 Mpc relative to the fiducial parametrization, while decreasing (increasing) the abundance of voids R eff 50 h-1 Mpc. Similarly, decreasing M increases (decreases) the abundance of small voids and decreases (increases) the abundance of large voids. The central assembly bias parameter Q cen also influences this, as a negative Q cen causes halos regions that are underdense for their mass are less likely to host central galaxies that would otherwise fracture large voids into multiple smaller voids.

  • Void Two-Point Functions (w p,vg): The cross-correlation function shows sensitivity to sigma M and M. The behavior is linked to the change in galaxy bias (b g); for instance, decreasing (increasing) the parameter sigma M (M) has the effect of increasing (decreasing) the mean mass and bias of halos that host galaxies.

  • Void Lensing (vm): The authors found negligible sensitivity to HOD parameters.

** Cosmological Forecast Results (Fiducial Scenario)**

The main results stem from a fiducial scenario combining n v, w p,vg, and w p,gg in a DESI-like survey of 14000 deg squared.

"For our fiducial scenario combining the void size function nv, projected void-galaxy cross-correlation function wp,vg and the projected galaxy auto-correlation function wp,gg we forecast 1.5% and 0.8% constraints on m and sigma 8 from DESI-Y5 data."

The authors emphasize that this combination breaks the degeneracy between sigma 8 and m, as the void size function's response to m is unique, impacting small and large voids in opposite ways.

** Relative Contributions of Observables**

The paper explored scenarios where different parts of the data vector were omitted:

  • Using w p,gg alone resulted in significant degradation of constraints (e.g., 4.8% on m).

  • Using n v alone resulted in even more significant degradation (e.g., 5.9% on m).

The combination of w p,gg and n v showed a significant improvement in constraining power, yielding 1.9% and 1.1% constraints on m and sigma 8, respectively, with little degeneracy between the two. This combination of datavectors self-calibrates the void-galaxy connection.

** Void Lensing Forecast**

The authors also forecasted constraints using the void size function (n v) combined with void weak lensing (vm):

  • For Stage III surveys (1250 deg squared): 5.2% on m and 3.5% on sigma 8.

  • For Stage IV surveys: These constraints improve to 4.8% and 2.6% respectively.

  • With an optimistic overlap of 10,000 deg squared: We forecast constraints of 3.7% and 1.5% on m and sigma 8 respectively.

** Conclusion**

The authors conclude that the combination of void observables with galaxy clustering is highly competitive with existing probes. They note that "the covariance matrix for the void-galaxy cross-correlation is by far the most challenging to estimate of the three observables, and so it is encouraging that we can still obtain tight constraints from the combination of galaxy clustering, whose covariance is well studied in the literature, and the void size function."

Improvements for AI systems

The following improvements leverage advanced machine learning techniques to enhance the computational efficiency, predictive power, and analytical rigor of the methodology presented in this paper.

Improvement: Replace the current reliance on finite differences (which requires multiple targeted N-body simulations) with a Gaussian Process (GP) or Deep Surrogate Model. This model learns the mapping between the combined data vector (n v,,vg,,gg, vm) and the cosmological/HOD parameters (m, sigma 8, Q cen,) across a vast parameter space.

What the Improved AI System Can Do:

  • Accelerate Forecasting: Instead of running thousands of expensive simulations to calculate derivatives for Fisher forecasting, the AI can instantly predict the expected likelihood and Fisher information matrix (F) for any given set of parameters with high accuracy (e.g., 95% confidence).

  • Systematically Map Degeneracies: The system can identify and visualize all complex degeneracies (e.g., between M 1 and sigma 8, or Q cen and m) across the entire parameter space, not just around the fiducial point, allowing researchers to target specific regions for further high-fidelity simulation runs.

Abstract

The large under-dense regions in the cosmological matter density field, known as cosmic voids, are powerful probes of cosmology but their potential is currently under-exploited. Observationally, voids are identified within the large scale distribution of galaxies and are therefore sensitive to certain features of the galaxy-halo connection. This sensitivity makes the combination of void and galaxy summary statistics particularly powerful probes of both cosmology and the galaxy-halo connection through self-calibration of the void-galaxy relation. In particular the combination of void and galaxy summary statistics breaks degeneracies in the galaxy-halo connection and cosmology relative to the case of galaxy clustering alone. To demonstrate this we forecast cosmological constraints attainable from the combination of the void size function n v, projected void-galaxy cross-correlation function w p,vg and projected galaxy auto-correlation function w p,gg measured in Dark Energy Spectroscopic Instrument (DESI) Year 5 data. We use a grid of cosmological N-body simulations to model this datavector as a function of sigma 8, m and a flexible halo occupation distribution (HOD) model that includes central and satellite galaxy assembly bias. For our fiducial scenario combining n v, w p,vg and w p,gg we forecast 1.5% and 0.8% constraints on m and sigma 8 from DESI-Y5 data. We also forecast constraints from the combination of the void size function and stacked weak lensing signal of voids.

Sources

Related papers