🔭 astrophysics

Increasing the sensitivity of full-shape galaxy clustering measurements in configuration-space with three-point statistics

Using AbacusSummit simulations and HOD models for Roman ELG and DESI LRG samples, this study demonstrates that incorporating line-of-sight dependent three-point correlation functions on small scales (<80 Mpc/h) significantly enhances cosmological constraints, tightening the sigma_8 measurement by a factor of five compared to two-point statistics alone.

Original authors: Zachery Brown, Lado Samushia

Published 2026-06-19

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Zachery Brown, Lado Samushia

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 the universe as a giant, cosmic city built by gravity. For decades, astronomers have tried to understand how this city was built and how it's growing by looking at how the "buildings" (galaxies) are arranged.

Traditionally, scientists have looked at pairs of galaxies. If you stand on one building and look at another, you can measure the distance between them. This is like looking at a map and counting how many houses are 100 yards apart, 200 yards apart, and so on. This method, called the "two-point" measurement, has been very successful, especially for finding the "acoustic scale"—a specific distance imprinted by the Big Bang that acts like a cosmic ruler.

However, the authors of this paper argue that looking at pairs is like trying to understand a complex story by only reading sentences with two words. To get the full picture, you need to look at triplets.

Here is a simple breakdown of what they did and what they found:

1. The New Tool: Looking at Triangles

Instead of just measuring the distance between two galaxies, the researchers started measuring the shape formed by three galaxies.

The Analogy: Imagine you are trying to guess the layout of a park. If you only know the distance between two trees, you know very little. But if you know the distances between three trees, you know the shape of the triangle they form. Is it a skinny, stretched-out triangle? Or a fat, equilateral one?
The Twist: The universe isn't static; it's expanding, and we are looking at it from a specific angle (our line of sight). The authors developed a new way to measure these triangles that pays close attention to how they are tilted relative to us. They asked: "Is this triangle pointing straight at us, lying flat across our view, or somewhere in between?"

2. The Challenge: The "Noise" of Galaxy Formation

There is a problem with studying galaxies. Galaxies don't just float randomly; they live inside invisible clouds of dark matter called "halos." The way galaxies form inside these halos is messy and complicated (like how different architects build different houses on the same plot of land).

The Analogy: If you try to measure the shape of a city, but you don't know if the houses are built by careful architects or chaotic squatters, your map will be blurry.
The Solution: The researchers used powerful computer simulations (like a video game engine for the universe) to create millions of fake universes. They then used a statistical method to "average out" the messy details of how galaxies form, allowing them to see the underlying cosmic structure more clearly.

3. The Discovery: A Massive Leap in Precision

The team tested their new method on two types of galaxy samples:

Roman-like galaxies: Bright, star-forming galaxies expected to be seen by the future Nancy Grace Roman Space Telescope (at a high redshift, meaning far away in time).
DESI-like galaxies: Luminous Red Galaxies observed by the current Dark Energy Spectroscopic Instrument (closer to us).

The Results:
When they added their new "triangle" measurements to the old "pair" measurements, the results were dramatic.

The "Five-Fold" Improvement: For the key parameter that tells us how clumpy the universe is (called $\sigma_8$ ), adding the triangle data made their measurements five times more precise than using pairs alone.
Why it works: It wasn't because they found one specific "magic triangle" that held all the answers. Instead, it was like listening to a choir. One singer (the pair measurement) is good, but when you add the harmony of many other voices (the thousands of different triangle shapes and orientations), the song becomes incredibly clear.
The Importance of Tilt: They found that knowing the orientation of the triangles (whether they are pointing at us or sideways) was crucial. If they just averaged everything out (ignoring the tilt), they lost about half the benefit.

4. The Bottom Line

The paper claims that by looking at the shapes of galaxy triplets and paying attention to how those shapes are tilted relative to Earth, we can learn much more about the universe's history and composition than we ever could by just looking at pairs.

They emphasize that this isn't just a small tweak; it's a fundamental upgrade. It allows us to squeeze significantly more information out of the same amount of data, helping us understand the invisible forces (like dark energy and dark matter) that are shaping our cosmic city.

In short: They took a blurry, two-dimensional map of the universe and turned it into a sharp, 3D hologram by counting triangles instead of just lines, revealing a much clearer picture of how the cosmos is built.

Technical Summary: Increasing the Sensitivity of Full-Shape Galaxy Clustering Measurements in Configuration-Space with Three-Point Statistics

Problem and Motivation
Standard cosmological analyses of large-scale structure (LSS) rely heavily on two-point statistics (2PCF), particularly the baryon acoustic oscillation (BAO) feature and full-shape modeling on linear and quasi-linear scales. While robust, these approaches discard non-Gaussian information generated by late-time gravitational evolution, redshift-space distortions, and galaxy bias. The three-point correlation function (3PCF) captures these higher-order correlations, offering a natural extension to 2PCF for probing non-linear structure. However, extracting cosmological information from non-linear scales remains challenging due to the breakdown of perturbation theory and the computational complexity of modeling high-dimensional data vectors. Furthermore, uncertainties in the galaxy–halo connection (bias) often obscure cosmological constraints on small scales. This work investigates whether a compressed, line-of-sight (LOS)-dependent 3PCF estimator can significantly enhance cosmological constraints on scales smaller than the BAO feature ( $\lesssim 80 \, h^{-1}\text{Mpc}$ ) for future surveys like the Nancy Grace Roman Space Telescope (Roman) and the Dark Energy Spectroscopic Instrument (DESI).

Methodology
The authors utilize the AbacusSummit suite of N-body simulations, specifically snapshots at $z=1.1$ and $z=0.8$ , populated with galaxies using Halo Occupation Distribution (HOD) models.

Simulations: The fiducial cosmology ( $c000$ ) is based on Planck 2018. A set of offset simulations ( $c100, c102, \dots$ ) varies cosmological parameters ( $h, \omega_{cdm}, n_s, \alpha_s, \sigma_8, N_{ur}, w_0$ ), while additional simulations ( $c901, \dots$ ) vary HOD parameters ( $\log M_{cut}, \sigma, \log M_1, \alpha$ ).
Galaxy Samples: Two representative samples are modeled: Roman-like Emission Line Galaxies (ELGs) at $z \approx 1.1$ and DESI-like Luminous Red Galaxies (LRGs) at $z \approx 0.8$ .
Statistics:
- 2PCF: Computed using Corrfunc, restricted to scales $2.5 < s < 82.5 \, h^{-1}\text{Mpc}$ to avoid the BAO feature. Monopole ( $\ell=0$ ) and quadrupole ( $\ell=2$ ) multipoles are used.
- 3PCF: Measured using TriCo, a fast configuration-space triangle-counting code. The authors introduce a novel LOS-dependent decomposition. Triangles are binned by two side lengths ( $s_1, s_2$ ) and the cosines of the angles these sides make with the LOS ( $\mu_1, \mu_2$ ).
- Compression: To manage dimensionality, $\mu$ is binned into three categories: parallel ( $\parallel$ ), transverse ( $\perp$ ), and mixed. This yields three distinct 3PCF components ( $\bar{\zeta}_{\parallel}, \bar{\zeta}_{\text{mix}}, \bar{\zeta}_{\perp}$ ) in addition to the standard 2PCF multipoles.
Modeling: A linear response model is constructed around the fiducial parameters using the offset simulations to estimate the Jacobian of the data vector. HOD parameters are treated as nuisance parameters and marginalized over.
Covariance: The covariance matrix is estimated from 1000 independent small-box simulations ( $500 \, h^{-1}\text{Mpc}$ ), rescaled to the full box volume ( $2 \, h^{-1}\text{Gpc}$ ). A fixed effective covariance is used for forecasts to avoid parameter-dependent likelihood complications, incorporating a Hartlap correction for the finite number of mocks.

Key Contributions

Novel Estimator: The development of a compressed, LOS-dependent 3PCF basis that preserves anisotropic information relevant to redshift-space clustering while maintaining a manageable data vector size (440 elements for the full set).
Simulation-Based Forecasting: The application of a simulation-based framework that explicitly marginalizes over HOD uncertainties, allowing for a robust quantification of the relative information gain of the 3PCF without relying on absolute bias calibration.
Validation: A rigorous validation of the linear response model using additional AbacusSummit cosmologies, confirming its adequacy for forecast-level comparisons near the fiducial point while acknowledging limitations on the smallest scales.

Results
The forecasts, performed using Markov Chain Monte Carlo (MCMC) on the Roman ELG and DESI LRG samples, yield the following findings:

Significant Constraint Improvement: Incorporating the LOS-dependent 3PCF alongside the 2PCF monopole and quadrupole substantially tightens cosmological constraints. For the Roman ELG sample, the constraint on $\sigma_8$ improves by a factor of $\sim 5$ compared to the 2PCF-only case. Similar, though slightly smaller, improvements (factors of 2–4) are observed for other cosmological parameters ( $h, \omega_{cdm}, n_s, w_0$ ) and HOD parameters.
Cumulative Information Gain: The improvement is not driven by a single scale or specific triangle configuration. Cumulative $\chi^2$ diagnostics show that information is distributed broadly across the triangle configurations, indicating that the gain arises from the cumulative information content of the non-linear clustering signal.
Importance of LOS Dependence: Restricting the analysis to the monopole of the 3PCF (LOS-averaged) captures only part of the available information. The full LOS-dependent measurement provides an additional factor of $\sim 2\text{--}3$ improvement over the monopole alone, demonstrating that the orientation of triangle configurations relative to the LOS carries critical cosmological information.
Robustness to Bias: The gains persist even after marginalizing over HOD parameters, suggesting that the 3PCF helps break degeneracies between cosmological parameters and galaxy bias.

Significance and Claims
The authors claim that this work demonstrates the viability of using small-scale, LOS-dependent 3PCF measurements to significantly enhance the sensitivity of full-shape galaxy clustering analyses. They emphasize that the method provides a "genuine higher-order clustering effect" rather than a response to a localized feature.

The paper explicitly states that these results are relative forecasts based on an idealized periodic-box volume and do not predict the exact constraining power of specific Roman or DESI survey samples, which would require accounting for survey geometry, fiber collisions, and other systematics. However, the authors argue that the relative improvement (e.g., the factor of 5 gain on $\sigma_8$ ) is a robust indicator of the value of including 3PCF information in future full-shape analyses. The work serves as a step toward simulation-based analyses that combine two- and three-point information while marginalizing over the galaxy–halo connection, offering a promising path to utilizing non-linear scales that are currently difficult to model.

1. The New Tool: Looking at Triangles

2. The Challenge: The "Noise" of Galaxy Formation

3. The Discovery: A Massive Leap in Precision

4. The Bottom Line

More like this