🔭 astrophysics

Cosmological inference with halo clustering reconstructed from the redshift-space galaxy distribution

This paper demonstrates that reconstructing halo centers from redshift-space galaxy distributions using the cylinder grouping method enables unbiased and more precise cosmological parameter inference within the Effective Field Theory framework, as reconstruction-induced systematics can be effectively modeled without introducing new parameters.

Original authors: Ryuichiro Hada, Teppei Okumura

Published 2026-04-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ryuichiro Hada, Teppei Okumura

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 you are trying to understand the structure of a massive, invisible city built out of dark matter. You can't see the buildings (the dark matter halos) directly, but you can see the lights (the galaxies) that sit on top of them.

The problem is that the lights are messy. Some lights are the main streetlamps (central galaxies), but many are just flickering neon signs attached to the same building (satellite galaxies). Worse, because the city is expanding and the lights are moving, some of these neon signs look stretched out or smeared in the direction of the expansion, like a long, blurry tail. This is called the "Finger-of-God" effect.

If you try to map the city using these messy lights, your map gets fuzzy, especially near the buildings, making it hard to measure how fast the city is growing or how gravity is pulling it together.

This paper is about a clever trick to clean up the map.

The Problem: The "Finger-of-God" Blur

In the real universe, galaxies aren't just floating alone. They live in groups. A big, heavy "host" galaxy sits in the center of a dark matter halo, and many smaller "satellite" galaxies orbit around it like moons.

When we look at these groups from Earth, the satellites are moving so fast around the center that their light gets smeared out along our line of sight. It looks like a long finger pointing at us. This smearing hides the true shape of the galaxy clusters and confuses our measurements of the universe's expansion and gravity.

The Solution: The "Cylinder Grouping" Trick

The authors propose a method called Cylinder Grouping (CG). Think of it like a smart cleaning robot for your data.

The Setup: Imagine you are looking at a crowd of people. You want to find the "leaders" (the central galaxies) and ignore the "followers" (the satellites).
The Cylinder: The robot draws an invisible, long, thin tube (a cylinder) around every person, aligned with your view.
The Sorting: It looks at who is inside that tube. If it sees a cluster of people, it assumes the most important-looking one is the leader and marks everyone else in that tube as a follower.
The Cleanup: It then removes all the followers from the list, leaving only the leaders.

By doing this, the robot effectively "pops" the satellites back into the center of their group. Instead of seeing a long, smeared finger of many galaxies, you now see a single, sharp point representing the center of the group.

The Test: Does it Work?

The authors didn't just guess; they built a giant virtual universe using supercomputers (simulations). They created a fake galaxy survey that looks just like the real one (specifically, what the DESI telescope will see).

They ran their "Cylinder Grouping" robot on this fake data and compared three things:

The Raw Galaxy Map: Messy, smeared, and full of satellite noise.
The True Halo Map: The "perfect" map where they knew exactly where the dark matter centers were (this is the gold standard, but we can't do this in real life).
The Reconstructed Map: The map after the robot cleaned it up.

The Result: The "Reconstructed Map" looked almost exactly like the "True Halo Map." The messy smearing was gone!

The Big Discovery: It's Not Just Clean, It's Smarter

Here is the most important part. Usually, when you change data (like cleaning it up), you introduce new errors or "systematics" that mess up your math. The authors were worried that their cleaning robot would create weird new patterns that would confuse their calculations.

They found that the robot's errors were very predictable.

The "Exclusion" Effect: The robot sometimes accidentally removes a real leader or keeps a follower. This creates a small "hole" in the data where pairs of galaxies shouldn't be. The authors realized this is just a smooth, simple mathematical effect that their standard equations could easily handle.
The "Masking" Effect: Because the robot looks in a specific direction (the cylinder), it slightly changes the overall brightness of the map in a way that depends on the angle. They found they could fix this by simply turning a "volume knob" (a scaling factor) for different angles.

Why This Matters for Cosmology

By using this cleaned-up map, the scientists could measure the Growth Rate of the Universe (how fast structures are forming) with much higher precision.

More Data: Because the "smearing" is gone, they can safely look at smaller, more detailed parts of the universe without getting confused.
Less Confusion: The measurements became much more stable. Even when they looked at very small scales, the results didn't wobble.
Better Precision: They reduced the uncertainty in their measurements by more than 20%.

The Bottom Line

Think of the universe as a noisy room where you are trying to hear a specific conversation.

Old Method: You try to listen to everyone, but the background noise (satellite galaxies) is so loud you can't hear the details.
New Method: You use a noise-canceling headset (Cylinder Grouping) that filters out the background chatter and focuses only on the main speakers.

The paper proves that this "headset" works perfectly. It doesn't just make the sound clearer; it actually lets you hear the conversation so well that you can learn new things about the room's acoustics (cosmology) that you couldn't hear before. This is a huge step forward for future telescopes like DESI, helping us understand dark energy and gravity with unprecedented accuracy.

1. Problem Statement

Cosmological parameter inference from spectroscopic surveys (like DESI) relies heavily on modeling Redshift-Space Distortions (RSD) in galaxy clustering. However, extracting information from small scales (quasi-nonlinear regime) is hindered by satellite galaxies within dark matter halos.

The Issue: Satellite galaxies possess high virial velocities, causing "Finger-of-God" (FoG) distortions that elongate structures along the line-of-sight (LOS). This suppresses clustering power on small scales and introduces complex non-linearities.
The Challenge: Standard Effective Field Theory (EFT) models attempt to parameterize these small-scale effects via counterterms and stochastic noise. However, as the analysis extends to higher wavenumbers ( $k_{max}$ ), the inference becomes increasingly sensitive to nuisance parameters, leading to degeneracies that degrade the precision of cosmological constraints (e.g., the growth rate $f\sigma_8$ ).
The Goal: To improve robustness and precision by reconstructing a tracer field that more closely resembles halo centers (where FoG effects are suppressed) directly from the observed galaxy distribution, and to determine if the systematic errors introduced by this reconstruction can be handled within standard EFT frameworks.

2. Methodology

A. Halo Reconstruction: Cylinder Grouping (CG)

The authors utilize the Cylinder Grouping (CG) method (Okumura et al., 2017) to reconstruct a halo-like field from galaxy mock catalogs.

Algorithm:
1. Ranking: Galaxies are ranked by a halo-mass proxy (local neighbor count within $10 h^{-1}$ Mpc).
2. Grouping: Starting from the highest-ranked galaxy (assumed central), the algorithm identifies neighbors within a cylinder aligned with the LOS (transverse radius $R=1.5 h^{-1}$ Mpc, LOS height $L=20 h^{-1}$ Mpc).
3. Removal: All galaxies within the cylinder are tagged as satellites and removed; only the central remains. This process iterates until all galaxies are classified.
Outcome: The resulting sample consists of "reconstructed centrals," which approximate the true halo center distribution, significantly reducing FoG smearing.

B. Simulation Setup

Data: Mock catalogs generated from the AbacusSummit N-body simulations at redshift $z=1.1$ .
Tracers: Three populations are analyzed:
1. Galaxies: Original DESI-like Luminous Red Galaxies (LRGs) populated via HOD.
2. Host Halos: True dark matter halos hosting the galaxies.
3. Reconstructed Halos: The output of the CG algorithm applied to the galaxy catalog.
Volume: 25 independent realizations of $(2 h^{-1} \text{Gpc})^3$ boxes.

C. Modeling and Inference Framework

Power Spectrum Modeling: The authors use velocileptors (based on Lagrangian Perturbation Theory) combined with EFT counterterms to model the redshift-space power spectrum multipoles ( $P_0, P_2$ ).
ShapeFit: They employ the ShapeFit parameter compression framework to constrain cosmological parameters ( $f\sigma_8$ , Alcock-Paczynski parameters $q_\parallel, q_\perp$ ) and broadband shape parameters ( $m$ ) relative to a fiducial cosmology.
Handling Reconstruction Systematics: The paper decomposes CG-induced errors into two components and integrates them into the EFT model:
1. Small-Scale "Exclusion" Effect: Imperfect grouping removes close pairs, creating a suppression similar to a hard-core exclusion window. The authors argue this manifests as smooth, high-derivative-like and stochastic contributions, which are absorbed by standard EFT counterterms and stochastic parameters ( $SN_0, SN_2$ ) without needing new parameters.
2. Large-Scale "Directional Masking" Effect: Imperfect identification creates an anisotropic bias. This is modeled as a multipole-dependent rescaling ( $R_\ell$ ) of the linear power spectrum amplitude. The authors determine $R_\ell$ directly from the data (ratios of reconstructed to original power spectra at low $k$ ) and apply it as a multiplicative correction to the deterministic part of the model.

3. Key Contributions

Systematic Integration: Demonstrates that reconstruction-induced systematics (exclusion and masking) can be consistently incorporated into standard EFT pipelines without introducing reconstruction-specific nuisance parameters.
Data-Driven Calibration: Proposes a method to calibrate large-scale anisotropic masking directly from observables ( $R^{obs}_\ell$ ), avoiding theoretical uncertainties in modeling the reconstruction bias.
Decoupling Degeneracies: Shows that reconstructing halo centers breaks the degeneracy between the growth rate ( $f\sigma_8$ ) and stochastic velocity parameters ( $SN_2$ ) that plagues standard galaxy analyses.

4. Key Results

A. Power Spectrum Behavior

The reconstructed-halo power spectrum closely matches the true host-halo spectrum, successfully eliminating the small-scale FoG suppression seen in the galaxy quadrupole.
Residual differences are smooth and broadband, confirming they can be modeled by standard EFT terms.

B. Cosmological Constraints

Unbiasedness: The reconstructed-halo sample yields unbiased constraints on $f\sigma_8$ , $q_\parallel$ , $q_\perp$ , and the shape parameter $m$ , consistent with the fiducial cosmology.
Precision Gain:
- $f\sigma_8$ : The reconstructed sample achieves a >20% reduction in uncertainty compared to the galaxy sample at $k_{max} \simeq 0.2 h \text{Mpc}^{-1}$ .
- Robustness: While galaxy-based constraints on $f\sigma_8$ show a systematic upward shift (bias) when extending $k_{max}$ beyond $0.2 h \text{Mpc}^{-1}$ , the reconstructed-halo constraints remain stable and unbiased.
- AP Parameters: Improvements of $\sim 10-20\%$ are observed for Alcock-Paczynski parameters, particularly $q_\parallel$ (LOS), which is most sensitive to FoG smearing.

C. Parameter Degeneracies

The reconstruction significantly reduces the degeneracy between the growth rate ( $f\sigma_8$ ) and the stochastic parameter $SN_2$ .
By removing satellite-driven velocity dispersion, the reconstructed sample decouples the velocity field from the stochastic noise, allowing for tighter constraints on the growth rate.

5. Significance and Conclusion

This work establishes halo reconstruction via cylinder grouping as a practical and powerful tool for next-generation RSD analyses (e.g., DESI, PFS).

Synergy with EFT: It proves that reconstruction does not require abandoning standard EFT modeling; rather, the reconstruction systematics map naturally onto existing EFT ingredients (counterterms and stochastic noise).
Future Prospects: The method enables the use of quasi-nonlinear scales ( $k > 0.2 h \text{Mpc}^{-1}$ ) with high confidence, offering a genuine path to reducing statistical errors on cosmological parameters by >20% without compromising robustness.
Next Steps: The authors suggest future work should optimize cylinder geometry for different tr redshifts, refine priors for the stochastic sector, and apply the method to real observational data to assess performance against fiber collisions and survey geometry effects.