DESI DR2 reference mocks: clustering results from Uchuu-BGS and LRG

This paper presents high-fidelity mock galaxy catalogs for DESI's Bright Galaxy Survey (BGS) and Luminous Red Galaxy (LRG) samples, generated via subhalo abundance matching on the Uchuu simulation, which successfully reproduce observed redshift evolution, clustering statistics, and galaxy-halo connections with high accuracy.

The Gist

Imagine you are an architect trying to build a model of a massive, bustling city. But instead of bricks and mortar, your city is made of galaxies, and the "streets" are the invisible web of dark matter that holds the universe together.

This paper is essentially a quality control report for a giant, computer-generated model of the universe. The team is checking if their digital simulation matches the real thing, so they can use it to solve the biggest mysteries of cosmology.

Here is the breakdown of their work using simple analogies:

1. The Goal: Building a "Perfect" Digital Twin

The Dark Energy Spectroscopic Instrument (DESI) is a massive telescope survey currently mapping millions of galaxies. It's like a giant census of the universe. But to understand what the census data means, scientists need a "control group"—a perfect, known universe to compare it against.

The authors created this control group using a supercomputer simulation called Uchuu. Think of Uchuu as a massive, high-resolution video game world containing 2.1 trillion particles (representing dark matter). It's so detailed that it can see everything from giant galaxy clusters down to tiny dwarf galaxies.

2. The Method: The "Subhalo Abundance Matching" (SHAM) Recipe

The simulation has the dark matter, but it doesn't have the galaxies yet. The authors needed a recipe to turn invisible dark matter clumps into visible galaxies.

They used a technique called SHAM (Subhalo Abundance Matching).

• The Analogy: Imagine you have a bag of different-sized rocks (dark matter halos) and a bag of different-sized cars (galaxies).
• The Rule: The biggest rocks get the biggest, most powerful cars. The medium rocks get medium cars. The small rocks get small cars.
• The Twist: In the real world, it's not a perfect 1-to-1 match. Sometimes a big rock gets a slightly smaller car, or a medium rock gets a flashy sports car. The authors added a little bit of "randomness" (scatter) to their recipe to make it look like the real universe.

They applied this recipe to two specific types of "cars" (galaxies) that DESI is interested in:

1. LRGs (Luminous Red Galaxies): These are the "heavy trucks" of the universe—old, massive, and very bright.
2. BGS (Bright Galaxy Sample): These are the "sedans and SUVs"—bright, but a bit more varied and closer to us.

3. The Test: Does the Simulation Match Reality?

Once they built their digital universe, they had to see if it looked like the real one. They compared their simulation against the actual data collected by DESI over the first three years (DR2).

They checked three main things:

• The "Crowd Density" (Number Counts):

  • Question: If I look at a specific patch of sky, does the simulation have the same number of galaxies as the real telescope?
  • Result: Yes. For the brightest galaxies, the match was incredibly tight (within 5%). For the fainter ones, it was still very good (within 10%).

• The "Social Clustering" (Clustering Statistics):

  • Question: Do galaxies in the simulation hang out in groups and clusters the same way real galaxies do?
  • Result: Yes. The simulation reproduced the "social habits" of galaxies perfectly. Whether galaxies were close together or far apart, the simulation matched the real data almost exactly.

• The "Weight" of the Galaxy (Bias):

  • Question: Do the heaviest galaxies (the biggest trucks) sit in the biggest dark matter clumps?
  • Result: Yes. The simulation correctly predicted that the most massive galaxies live in the most massive dark matter neighborhoods.

4. The "Glitch" and the Fix

The authors found one small hiccup. When looking at the quadrupole (a specific way of measuring how galaxies are stretched out due to their motion), the simulation for the massive red galaxies (LRGs) was slightly off.

• The Metaphor: It's like the simulation got the location of the cars right, but the speedometers were slightly inaccurate.
• The Cause: The SHAM method is great at placing galaxies, but it doesn't perfectly simulate how fast they are moving. Since the "stretching" effect depends on speed, this caused a tiny mismatch.
• The Verdict: Despite this small speedometer issue, the overall map is still accurate enough to be incredibly useful.

5. Why Does This Matter?

Why spend so much time building and checking a fake universe?

1. Calibration: Just as a pilot trains in a flight simulator before flying a real plane, cosmologists need these "mock" universes to test their theories. If their math works on the simulation, they know it's ready for the real data.
2. Dark Energy: The ultimate goal of DESI is to figure out what Dark Energy is (the force pushing the universe apart). To do that, they need to measure the expansion of the universe with extreme precision. These simulations act as the "ruler" to ensure their measurements aren't skewed by errors.
3. Future Proofing: Now that they have proven their "recipe" works, they can generate millions of these fake universes to test every possible scenario for the rest of the DESI survey.

The Bottom Line

This paper is a success story. The authors built a digital twin of the universe using a supercomputer and a clever matching recipe. They proved that this digital twin looks, acts, and clusters almost exactly like the real universe we see through our telescopes.

This gives scientists a powerful new tool: a reliable, high-fidelity simulation that will help them unlock the secrets of dark energy and the history of our cosmos.

Technical Summary

Here is a detailed technical summary of the paper "DESI DR2 reference mocks: clustering results from Uchuu-BGS and LRG".

1. Problem Statement

The Dark Energy Spectroscopic Instrument (DESI) is a Stage IV survey aiming to map millions of galaxies to constrain the nature of dark energy. To maximize the scientific return of DESI Data Release 2 (DR2), specifically for the Bright Galaxy Sample (BGS) and Luminous Red Galaxy (LRG) tracers, robust simulated lightcones are required. These mocks must accurately reproduce:

• The redshift evolution of galaxy number density.
• Clustering statistics (Two-Point Correlation Function and Power Spectrum).
• Dependencies on baryonic properties (stellar mass for LRGs, absolute magnitude for BGS).
• Selection functions and incompleteness effects inherent to the survey.

Existing mocks often struggle to simultaneously match the high-fidelity clustering signals across different scales (from non-linear to BAO scales) while correctly modeling the dependence of clustering on galaxy properties. This work addresses the need for high-fidelity reference mocks tailored to DESI DR2 to validate cosmological analyses, quantify systematics, and optimize measurement techniques.

2. Methodology

The authors constructed mock catalogs using the Subhalo Abundance Matching (SHAM) technique applied to the Uchuu N-body simulation.

• Simulation Base (Uchuu):

  • A high-resolution simulation containing 2.1 trillion particles in a volume of $8 h^{-3} \text{Gpc}^3$.
  • Assumes a Planck base-$\Lambda$CDM cosmology ($h=0.6774, \Omega_m=0.3089$).
  • Resolves low-mass halos and subhalos, essential for modeling faint galaxies and satellites.
  • Uses the Rockstar halo finder and Consistent-Trees merger trees.

• SHAM Implementation:

  • BGS-BRIGHT ($r < 19.5$):
    • Target Property: Absolute $r$-band magnitude ($M_r$).
    • Luminosity Function: Derived from DESI Year 1 (Y1) North Galactic Cap data, $k+E$ corrected to $z=0.1$.
    • Scatter Model: The scatter ($\sigma$) in the magnitude-velocity relation is luminosity-dependent. $\sigma = 0.05$ for bright galaxies ($M_r > -20.5$) and $\sigma = 0.4$ for fainter galaxies.
    • Flux Limit: Apparent magnitudes are calculated by assigning $g-r$ colors (modeled as a double-Gaussian distribution) and applying $k+E$ corrections to enforce the $r < 19.5$ flux limit.
  • LRG ($0.4 < z < 1.1$):
    • Target Property: Stellar Mass ($M_*$).
    • Stellar Mass Function (SMF): Derived from DESI Y1 using CIGALE. The SMF is modeled as complete for high masses but incomplete for low masses; the authors implement a probabilistic downsampling to reproduce the observed incompleteness.
    • Scatter Model: The scatter ($\sigma$) is redshift and mass-dependent. Values were tuned in bins of redshift ($0.4<z<0.6, 0.6<z<0.8, 0.8<z<1.1$) and stellar mass cuts to match the observed Two-Point Correlation Function (TPCF).

• Lightcone Construction:

  • Cubic simulation boxes were assembled into spherical shells (width $\Delta z = 0.05$) to create full-sky lightcones.
  • The mocks were cut to match the DESI Y1 and Y3 survey footprints (Northern and Southern Galactic Caps).

• Analysis Tools:

  • Clustering: Measured using the Landy-Szalay estimator for the monopole and quadrupole of the redshift-space TPCF ($\xi(s, \mu)$) and the Power Spectrum ($P(k)$).
  • Weights: Applied Pairwise Inverse Probability (PIP) and FKP weights to the data; mocks were analyzed with consistent weighting schemes where applicable.
  • Bias: Linear bias ($b$) was measured by fitting the monopole TPCF in the linear regime ($10 < s < 40 h^{-1}\text{Mpc}$).

3. Key Contributions

1. First DR2 Reference Mocks: Provided the first set of high-fidelity Uchuu-based mocks specifically tuned to DESI DR2 BGS and LRG samples.
2. Refined SHAM Scattering: Demonstrated that a constant scatter is insufficient. The paper introduces and calibrates luminosity-dependent scatter for BGS and redshift/mass-dependent scatter for LRGs to match clustering data.
3. Comprehensive Validation: Validated the mocks against DESI data across a vast range of scales ($0.01$to$200 h^{-1}\text{Mpc}$), including BAO scales, non-linear clustering, and redshift-space distortions (RSD).
4. Halo Occupation Distribution (HOD): Presented detailed HOD measurements for both tracers, highlighting differences caused by the variable scatter implementation compared to previous constant-scatter models.

4. Key Results

• Clustering Agreement:
  • BGS-BRIGHT: The mocks reproduce the observed clustering with <5% agreement for separations $1 < r < 20 h^{-1}\text{Mpc}$and **<10$0.1 < r < 1 h^{-1}\text{Mpc}$.
  • LRG: Excellent agreement for the monopole ($<0.2\%$ deviation at $1 < s < 20 h^{-1}\text{Mpc}$).
  • Quadrupole Discrepancy: A noticeable discrepancy exists in the LRG quadrupole at $s > 1 h^{-1}\text{Mpc}$, attributed to stellar-mass incompleteness in the data and limitations in modeling galaxy velocities via SHAM.
• Dependence on Properties:
  • The mocks successfully reproduce the clustering dependence on absolute magnitude for BGS volume-limited samples and stellar mass for LRGs.
  • For the faintest BGS samples ($M_r < -20.0$), deviations increase, suggesting potential issues with $k+E$ corrections or the scatter model at the faint limit.
• Large-Scale Bias:
  • Measured bias factors as a function of magnitude (BGS) and stellar mass (LRG).
  • The mocks show a slightly higher bias than the data for the brightest/most massive galaxies, but the results are consistent within $1\sigma$ when considering cosmic variance.
  • Functional fits for bias ($b$) vs. magnitude/mass were derived and found to be compatible between data and mocks.
• Power Spectrum:
  • Excellent agreement in the monopole and quadrupole of the power spectrum for BGS ($M_r < -21.0$) and LRG ($0.6 < z < 0.8$), though window function differences complicate precise amplitude comparisons.

5. Significance

This work provides robust, high-fidelity tools essential for the remainder of the DESI survey.

• Cosmological Constraints: The mocks enable more accurate full-shape analyses of Baryon Acoustic Oscillations (BAO) and Redshift Space Distortions (RSD), leading to tighter constraints on dark energy parameters ($w_0, w_a$).
• Systematic Control: By accurately modeling the galaxy-halo connection and selection effects, these mocks allow researchers to quantify and mitigate systematic errors in future DESI data releases.
• Future Applications: The generated catalogs are critical for training covariance matrices, testing analysis pipelines, and studying large-scale structure features like cosmic voids.
• Methodological Advancement: The demonstration of variable scatter in SHAM improves the physical realism of mock catalogs, setting a new standard for generating reference simulations for next-generation spectroscopic surveys.

The authors conclude that the Uchuu-BGS and Uchuu-LRG mocks are sufficiently accurate to serve as the primary reference for DESI DR2 analyses, significantly enhancing the reliability of cosmological inferences drawn from the survey.