🔭 astrophysics

The peculiar velocity correlation function of the Cosmicflows-4 catalog

This paper presents an analysis of the Cosmicflows-4 survey's peculiar velocity correlation function, addressing statistical uncertainties through improved estimators and weighting schemes to derive growth rate parameters of $f\sigma_8=0.384^{+0.116}_{-0.194}$ for the group dataset and $f\sigma_8=0.569^{+0.054}_{-0.06}$ for the local universe.

Original authors: Yuyu Wang, Hume A. Feldman, Richard Watkins, Xiaohu Yang

Published 2026-04-10

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yuyu Wang, Hume A. Feldman, Richard Watkins, Xiaohu Yang

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 not as a static stage, but as a giant, flowing river. Most galaxies are like leaves floating on this river, carried along by the overall expansion of space (the current). But some leaves are also drifting sideways due to local whirlpools and eddies created by massive clumps of matter. This sideways drift is called peculiar velocity.

This paper is like a team of cosmic detectives trying to map the strength of those whirlpools by measuring how fast these "leaves" (galaxies) are drifting relative to each other. They used the biggest, deepest map of these drifts ever created, called Cosmicflows-4 (CF4).

Here is the story of their investigation, broken down into simple concepts:

1. The New, Bigger Map

Previous maps (like Cosmicflows-3) were like looking at a neighborhood from a porch. You could see the houses nearby clearly, but the view got fuzzy and stopped quickly.

The Upgrade: The new CF4 map is like climbing a skyscraper. It sees three times more galaxies and reaches much deeper into the universe (from about 50 million light-years away to 100 million light-years away).
The Problem: Because the new data comes mostly from telescopes in the Northern Hemisphere (like the SDSS), the map looks lopsided. It's like trying to understand the weather of the whole Earth by only looking at the sky over New York and London, while ignoring the rest of the planet. This "lopsidedness" (anisotropy) makes the math tricky.

2. The "Drunk Sailor" Problem (Uncertainty)

Measuring how fast a galaxy is moving sideways is incredibly hard. You have to know exactly how far away it is.

The Analogy: Imagine trying to guess how fast a car is driving by looking at it through a foggy window. If you can't tell if the car is 100 meters away or 1,000 meters away, your speed estimate will be wild.
The Reality: For distant galaxies, the "fog" (measurement error) is thick. The further away the galaxy, the more uncertain its speed becomes. This creates a lot of "statistical noise" in the data, making it hard to see the true pattern of the universe's flow.

3. The Detective's Toolkit (Correlation Functions)

To cut through the noise, the scientists didn't look at single galaxies. Instead, they looked at pairs of galaxies.

The Analogy: If you watch two leaves drifting in a river, do they move together? If they are close to a big whirlpool, they might both get sucked in the same direction. This "togetherness" is called a correlation.
The Tools: They used two specific tools to measure this:
- Parallel Correlation: Do the galaxies move in the same direction as the line connecting them? (Like two cars driving down the same highway).
- Perpendicular Correlation: Do they move sideways relative to that line?
The Discovery: They found that the "Parallel" tool was the most stable and reliable, while the "Perpendicular" tool was too jittery and prone to errors.

4. The Weighting Dilemma

The team tried different ways to "weigh" the data to get a better answer.

The Analogy: Imagine a classroom vote. Should every student's opinion count equally? Or should you give more weight to the students who are sitting right in front of you (because you can hear them better) and less weight to the ones in the back (who are hard to hear)?
The Result:
- Giving more weight to nearby galaxies (the "front row") reduced the noise but made the map less representative of the whole universe.
- Giving equal weight to everyone (including the noisy, distant ones) kept the map fair but kept the noise high.
- The Decision: They decided to treat everyone equally (no special weighting) because the new map was so big that the "noise" from the distant galaxies was the biggest problem, and weighting them down would hide important deep-universe data.

5. The Final Verdict (Cosmology)

By analyzing these drifting galaxies, the team tried to measure a specific number called $f\sigma_8$ .

What is it? Think of it as the "growth rate" of the universe's structure. It tells us how fast gravity is pulling matter together to form giant clusters and filaments.
The Results:
- The "Local" View: When they looked only at the most precise, nearby data, they found a growth rate of 0.57. This is a very tight, confident number.
- The "Total" View: When they included all the data (including the noisy, distant parts), the number dropped to 0.38, but the uncertainty range got much wider.
The Takeaway: The results are consistent with what we expect from the standard model of the universe (based on the Planck satellite data), but the "fog" of measurement error in the new deep data makes the final number a bit fuzzy.

Summary

This paper is a triumph of scale. The team built the biggest map of galaxy movements ever, but the map was so deep and lopsided that the "fog" of measurement error became the main challenge. They proved that while the new data is noisy, the "Parallel Correlation" method is the best way to cut through it.

They found that the universe is growing and clumping together at a rate that matches our current theories, but to get a sharper, clearer picture, we need even deeper and less "foggy" surveys in the future (like the upcoming DESI project). Essentially, they took a step forward, but the fog is still thick enough that we need better flashlights to see the next step clearly.

1. Problem Statement

Peculiar velocity (PV) studies offer an independent probe of the large-scale structure of the universe and the growth of cosmic structure, distinct from galaxy distribution studies. However, these studies face significant challenges:

Measurement Precision: Accurate distance measurements are difficult to obtain, leading to large fractional errors (typically ~20%) in PV estimates. These errors are non-Gaussian due to the logarithmic relationship between distance and distance modulus.
Cosmic Variance & Observer Bias: Previous catalogs (e.g., Cosmicflows-3, CF3) suffered from "observer bias," where the measured correlation function varied depending on the observer's location relative to large-scale structures (like the Virgo Cluster).
Statistical Uncertainty: While newer catalogs like Cosmicflows-4 (CF4) offer greater depth ( $z \sim 0.1$ ) and sample size, they introduce new challenges. The CF4 catalog exhibits severe anisotropy (uneven distribution between northern and southern hemispheres) and large velocity uncertainties for distant galaxies. This combination leads to substantial statistical uncertainties in the velocity correlation function, potentially outweighing cosmic variance.
Estimator Stability: Different correlation function estimators (e.g., Gorski's $\psi_1, \psi_2$ vs. parallel/perpendicular $\Psi_\parallel, \Psi_\perp$ ) exhibit different stability and distribution properties (Gaussian vs. non-Gaussian cosmic variance).

2. Methodology

The authors employed a rigorous statistical framework to analyze the CF4 dataset:

Data Source: The study utilizes the Cosmicflows-4 (CF4) catalog, containing 55,877 galaxies grouped into 38,065 groups. This represents a significant expansion over CF3, extending the redshift depth from $z \sim 0.05$ to $z \sim 0.1$ and nearly tripling the sample size.
Velocity Estimators:
- Distance: Implemented the unbiased distance estimator from Watkins et al. (2023) to correct for non-Gaussian distance uncertainties.
- Velocity: Adopted the Davis & Scrimgeour (DS) velocity estimator, which incorporates Doppler shift and gravitational redshift corrections, yielding smaller velocity uncertainties compared to the standard Hubble estimator, especially for distant galaxies.
Correlation Estimators: The study focuses on the Parallel ( $\Psi_\parallel$ ) and Perpendicular ( $\Psi_\perp$ ) correlation functions. These are derived via least-squares methods and are preferred over Gorski's estimators because $\Psi_\parallel$ exhibits Gaussian-distributed cosmic variance, whereas $\Psi_\perp$ and $\psi_1$ show non-Gaussian distributions.
Mock Catalogs: To quantify uncertainties, the authors generated 1,000 mock catalogs from the Outer Rim Simulation (a dark-matter-only simulation).
- Observer Types: Simulated both "Random" (Rand) and "Local Group-like" (LG) observers.
- Anisotropy: Created "Mask" mocks that replicate the specific angular and radial anisotropy of the CF4 survey (deeper in the North, shallower in the South) and "Radial" mocks that are isotropic.
Uncertainty Decomposition: Total uncertainty ( $\delta_t$ ) was decomposed into Cosmic Variance ( $\delta_c$ , derived from the scatter of 100 mock catalogs) and Statistical Uncertainty ( $\delta_s$ , derived from perturbing velocity measurements based on their error bars).
Weighting Schemes: The authors tested three weighting schemes to mitigate statistical noise:
1. Standard FKP (Feldman-Kaiser-Peacock).
2. Revised FKP (incorporating velocity uncertainty).
3. Modified Kernel Density weighting.
- Finding: While revised FKP and kernel density weights reduced statistical uncertainty, they increased cosmic variance by effectively down-weighting distant galaxies (reducing effective survey depth). Consequently, the authors proceeded with uniform weights ( $w=1$ ) for the final analysis.
Cosmological Constraints: Used Markov Chain Monte Carlo (MCMC) methods to constrain the growth rate parameter $f\sigma_8$ by fitting the observed $\Psi_\parallel$ to linear theory predictions.

3. Key Contributions

First Analysis of CF4 Correlation Functions: This is the first comprehensive study of the velocity correlation function using the full CF4 dataset, leveraging its increased depth and sample size.
Characterization of Anisotropy Effects: The paper demonstrates that while the "observer effect" (bias due to observer location) becomes negligible in deep, large surveys like CF4, anisotropy (uneven sky coverage) significantly increases statistical uncertainty.
Estimator Validation: Confirmed that $\Psi_\parallel$ is the most robust estimator for CF4 data, exhibiting Gaussian cosmic variance, whereas $\Psi_\perp$ remains unstable and non-Gaussian.
Uncertainty Analysis: Provided a detailed breakdown showing that for CF4, statistical uncertainty (driven by distance/velocity errors) now dominates over cosmic variance, a reversal of trends seen in shallower catalogs like CF3.
Weighting Scheme Evaluation: Systematically tested weighting schemes, concluding that for current CF4 depth, uniform weighting is optimal to avoid artificially inflating cosmic variance by suppressing distant data.

4. Results

Cosmic Variance Distribution: $\Psi_\parallel$ and $\psi_2$ follow Gaussian distributions, while $\Psi_\perp$ and $\psi_1$ follow non-Gaussian (Wishart-like) distributions.
Observer Independence: The discrepancy between Rand and LG observers is negligible for $\Psi_\parallel$ in CF4, confirming that large sample sizes mitigate observer bias.
Anisotropy Impact: The anisotropic distribution of CF4 (deeper North, shallower South) leads to significantly higher total uncertainty in "Mask" mocks compared to isotropic "Radial" mocks, primarily driven by statistical uncertainty.
Growth Rate Constraints ( $f\sigma_8$ ):
- Total Uncertainty ( $f\sigma_8^T$ ): $0.384^{+0.116}_{-0.194}$ . This result is consistent with Planck 2018 results within $1\sigma$ but has large error bars due to the inclusion of cosmic variance and large velocity errors.
- Statistical Uncertainty Only ( $f\sigma_8^S$ ): $0.569^{+0.054}_{-0.06}$ . This "local" constraint is tighter and higher, but relies on ignoring cosmic variance, which may introduce bias if the local universe differs from the global average.
Goodness of Fit: The reduced chi-square ( $\chi^2_\nu$ ) values were high ($2.0$ for statistical, $2.2$ for total), attributed to bin-to-bin fluctuations, non-linear effects, and non-Gaussian cosmic variance components.

5. Significance

Validation of CF4 Utility: Despite the challenges of anisotropy and large velocity errors, the CF4 catalog successfully provides a robust measurement of the velocity correlation function, validating its use for cosmological constraints.
Transition in Error Budget: The study highlights a critical transition in peculiar velocity cosmology: as surveys get deeper, statistical uncertainty (from measurement errors) becomes the dominant limitation rather than cosmic variance. This shifts the focus for future improvements toward better distance indicators and velocity estimators.
Future Outlook: The results set a baseline for upcoming, deeper, and denser surveys (e.g., DESI Peculiar Velocity survey). The authors argue that while current weighting schemes have trade-offs, future surveys with reduced cosmic variance will allow for more sophisticated weighting to minimize statistical noise, leading to tighter constraints on $f\sigma_8$ and the nature of dark energy.
Methodological Benchmark: The paper provides a rigorous framework for handling anisotropic data and decomposing uncertainties, serving as a template for analyzing future large-scale peculiar velocity datasets.