DESI DR2 BAO Distances: Precision Cosmology

• DESI DR2 BAO distances are precise cosmic measurements derived from over 14 million galaxy, quasar, and Lyman-α forest redshifts across a broad redshift range.
• They employ advanced methodologies including density field reconstruction and spline-based template fitting to achieve sub-percent precision and rigorously control systematic errors.
• Results constrain crucial cosmological parameters (e.g., H0 and Ωm), address tensions with CMB data, and offer new insights into dark energy dynamics and spatial curvature.

Baryon acoustic oscillation (BAO) distances derived from the Dark Energy Spectroscopic Instrument (DESI) Data Release 2 (DR2) are quantitative measurements of the cosmic distance scale as inferred from large volumes of spectroscopic galaxy and quasar data. These distances are critical for mapping the expansion history of the Universe, constraining cosmological parameters, and probing the nature of dark energy, curvature, and fundamental relations between cosmological observables. DESI DR2 incorporates galaxies spanning a broad redshift range (0.1 < z < 2.1), with BAO information also extracted from Lyman-α forest correlations at z ≳ 2. DESI DR2 delivers sub-percent precision BAO distance constraints across multiple tracers, with robust control of systematic and statistical errors.

1. Survey Design, Sample Characteristics, and Pipeline Validation

DESI DR2 BAO analyses employ more than 14 million spectroscopically confirmed galaxy and quasar redshifts, encompassing four main tracers:

• Bright Galaxy Sample (BGS) at low redshift (0.1 < z < 0.5)
• Luminous Red Galaxies (LRG, 0.4 < z < 1.1)
• Emission Line Galaxies (ELG, 0.8 < z < 1.6)
• Quasars (QSO, 0.8 < z < 2.1)
• Lyman-α forest (z ≳ 2.1), measured in quasars’ absorption spectra

Catalogs are constructed with strict completeness criteria (e.g., BGS at 50% completeness, LRG at 57.9%), random catalogs at 2500 deg⁻² density, and extensive systematic weighting for imaging, fiber assignment, and FKP correction: $w_\mathrm{FKP} = [1 + n(z) C P_0]^{-1}$ where $n(z)$ is redshift-dependent density, $C$ is completeness, and $P_0$ a fixed power spectrum amplitude.

The DESI spectrograph and "Guadalupe" pipeline enable automated fiber assignment and high-throughput redshift confirmation, using quality diagnostics such as ZWARN and DELTACHI2 for robust data validation. Critical pre-unblinding and post-unblinding validations include:

• Extensive use of AbacusSummit and EZmocks for covariance estimation and algorithm bias checks
• Mock data realizations tailored to each sample's selection and clustering
• Systematic error propagation via the RascalC framework and robust splits by sky region or photometric regime

2. BAO Signal Measurement and Modelling Framework

The BAO signal is identified in the two-point correlation function $\xi(s,\mu)$ or its Fourier counterpart, modeled using the Landy–Szalay estimator: $$\xi(s,\mu) = \frac{DD(s,\mu) - 2 DR(s,\mu) + RR(s,\mu)}{RR(s,\mu)}$$ Multipole moments ($\ell = 0,2,4$) are projected to analyze isotropic (monopole, $\xi_0$) and anisotropic (quadrupole, $\xi_2$) BAO scales.

Density field reconstruction (following Eisenstein et al. 2007) is used to partially undo the nonlinear degradation of BAO, implemented with both RecIso and RecSym conventions, the latter shown to minimize nonlinear BAO shifts in DESI analyses (Chen et al., 2024, Andrade et al., 18 Mar 2025). A template-fitting approach (akin to Xu et al. 2012) describes the monopole BAO as: $$\xi_\ell(s) = B^2 \, \xi_\ell^\mathrm{BAO}(s, \alpha) + A_\ell(s)$$ with scaling parameter $n(z)$0 relating observed and fiducial BAO positions. Spline-based broadband marginalization absorbs smooth deviations and further suppresses systematics. The exponential damping model is applied to nonlinear BAO broadening: $n(z)$1 where $n(z)$2 is derived from theoretical models and mock calibrations.

BAO dilation parameters are defined as

$n(z)$3

with isotropic and anisotropic combinations $n(z)$4, $n(z)$5 (Chen et al., 2024).

3. Precision Distance Measurements

DESI DR2 measures BAO distances with unprecedented precision:

• Galaxy BAO (BGS, LRG, ELG, QSO): isotropic scale $n(z)$60.24% (compared to DR1’s 0.52%) (Andrade et al., 18 Mar 2025)
• Lyman-α BAO (zₑff = 2.33): statistical precision of 1.1% (radial), 1.3% (transverse), combined to 0.65% overall, with systematic error of 0.3% per direction (Collaboration et al., 18 Mar 2025)
• Key high-z ratios: $n(z)$7, $n(z)$8

The multi-tracer approach guarantees robust cross-checks and tightly controlled combined error budgets. Improvements over previous surveys (e.g., SDSS eBOSS, DR1) exceed 40% in combined distance precision, driven by increased volume and redshift leverage. Systematic and theory/modeling errors are restricted to 0.1% (isotropic) and 0.2% (anisotropic) (Chen et al., 2024).

4. Cosmological Parameter Inference and Dataset Tensions

DESI DR2 BAO distances, when interpreted within the flat $n(z)$9CDM framework, yield:

• $C$0
• Calibrated Hubble constant $C$1 km s⁻¹ Mpc⁻¹ (using BBN prior on $C$2)

A persistent, modest tension of $C$3 exists between DESI-inferred distances and Planck CMB-inferred values, with BAO distances $C$41.5% shorter than CMB predictions (Collaboration et al., 18 Mar 2025, Chen et al., 1 May 2025). This discrepancy also appears when combining with different SNe datasets, and is mirrored by inconsistencies in $C$5 and the distance calibration factor $C$6.

Allowing for evolving dark energy ($C$7–$C$8) improves fits, with best-fit solutions in $C$9, $P_0$0 quadrant and $P_0$1 preference for dynamical dark energy models (growing to $P_0$2 when including select SNe samples) (Collaboration et al., 18 Mar 2025, Gu et al., 8 Apr 2025). Some analyses, however, caution that these results are contingent on treatment of SNe-BAO calibration offsets, possible systematics, and specific parameterizations (Afroz et al., 23 Apr 2025, Efstathiou, 5 May 2025). Indeed, adjusting for a phenomenological redshift-dependent distance duality offset ($P_0$3) can suppress this evolving dark energy preference and bring inferred $P_0$4, $P_0$5 near $P_0$6 (Afroz et al., 23 Apr 2025).

Allowing for spatial curvature (as parameterized by $P_0$7 or $P_0$8) relieves the tension, slightly modifies BAO-calibrated distances, and relaxes the sum of neutrino masses constraint to $P_0$9 eV (Chen et al., 1 May 2025). This suggests that curvature, not just dark energy dynamics, can play a role in reconciling BAO and CMB scales.

5. Model-Independent Analyses and Fundamental Relations

DESI DR2 BAO distances have become the gold standard for model-independent tests of fundamental cosmological relations:

• The Cosmic Distance Duality Relation (CDDR), $\xi(s,\mu)$0, is tested at high statistical precision across redshifts by combining DESI BAO $\xi(s,\mu)$1 (angular diameter distances) with smooth SNe $\xi(s,\mu)$2 reconstructions (Wang et al., 15 Jun 2025, Zhang et al., 22 Jun 2025, Kanodia et al., 15 Jul 2025).
• Fully model-independent approaches—implementing two-point diagnostics, Gaussian Process regression, compressed control points, and analytic marginalization over SNe absolute magnitude $\xi(s,\mu)$3 and $\xi(s,\mu)$4—show no significant deviation from $\xi(s,\mu)$5 at low and intermediate redshift. Deviations above $\xi(s,\mu)$6 may arise at $\xi(s,\mu)$7, but their significance is tempered by the number of comparisons and the need for confirmation.
• The treatment of the standard ruler $\xi(s,\mu)$8 exerts a non-trivial influence. Marginalization about $\xi(s,\mu)$9 or use of differing calibrations (e.g., $$\xi(s,\mu) = \frac{DD(s,\mu) - 2 DR(s,\mu) + RR(s,\mu)}{RR(s,\mu)}$$0 Mpc vs. $$\xi(s,\mu) = \frac{DD(s,\mu) - 2 DR(s,\mu) + RR(s,\mu)}{RR(s,\mu)}$$1 Mpc) shifts $$\xi(s,\mu) = \frac{DD(s,\mu) - 2 DR(s,\mu) + RR(s,\mu)}{RR(s,\mu)}$$2 results. The degeneracy between $$\xi(s,\mu) = \frac{DD(s,\mu) - 2 DR(s,\mu) + RR(s,\mu)}{RR(s,\mu)}$$3 and $$\xi(s,\mu) = \frac{DD(s,\mu) - 2 DR(s,\mu) + RR(s,\mu)}{RR(s,\mu)}$$4 can be broken by including calibration-free maser distances, yielding $$\xi(s,\mu) = \frac{DD(s,\mu) - 2 DR(s,\mu) + RR(s,\mu)}{RR(s,\mu)}$$5 Mpc and $$\xi(s,\mu) = \frac{DD(s,\mu) - 2 DR(s,\mu) + RR(s,\mu)}{RR(s,\mu)}$$6 (Kanodia et al., 15 Jul 2025).
• Parametric and non-parametric approaches (e.g., shape function or statefinder reconstructions, correlation priors) consistently support mild dynamical dark energy signatures when combining DESI DR2 with SNe and CMB, with signal significances up to $$\xi(s,\mu) = \frac{DD(s,\mu) - 2 DR(s,\mu) + RR(s,\mu)}{RR(s,\mu)}$$7 but systematic offsets between datasets remain a key concern (Gu et al., 8 Apr 2025, Mukherjee et al., 25 May 2025).
• No evidence for cosmic opacity or Etherington relation violation is found within current precision (Zhang et al., 22 Jun 2025, Zheng et al., 23 Jul 2025).

6. Systematic Error Budget, Robustness Checks, and Future Prospects

DESI DR2 BAO distances are supported by:

• Validation on extensive suites of mock catalogs incorporating non-linear structure growth, realistic observational effects, and a blend of N-body and semi-empirical approaches (e.g., AbacusSummit, CoLoRe–QL for Lyman-α, Saclay mocks).
• Control of systematics is achieved at the sub-0.2% level, with shape, selection, weighting, and modeling choices all tested for biases below the statistical limit (Andrade et al., 18 Mar 2025, Chen et al., 2024, Casas et al., 18 Mar 2025).
• Statistical methodology (choice of data vector, basis for likelihood compression, treatment of parameter priors and degeneracies) is under active scrutiny; some analyses point to the importance of Bayesian evidence, pivot-parameterization, and care in interpreting significance metrics (Efstathiou, 5 May 2025).

Looking forward, DESI-5YR, Euclid, and LSST surveys will extend and refine these results, enabling discrimination between systematics, curvature, and dynamical dark energy scenarios (Collaboration et al., 18 Mar 2025, Mukherjee et al., 25 May 2025). Advanced, model-agnostic analyses—potentially including gravitational wave sirens and further improved calibration-free geometrical probes—are forecast to further tighten constraints on distance relations and expansion rate anomalies, especially at the characteristic redshifts where current data reveal $$\xi(s,\mu) = \frac{DD(s,\mu) - 2 DR(s,\mu) + RR(s,\mu)}{RR(s,\mu)}$$8–$$\xi(s,\mu) = \frac{DD(s,\mu) - 2 DR(s,\mu) + RR(s,\mu)}{RR(s,\mu)}$$9 hints of new physics (Mukherjee et al., 25 May 2025).

---

Table: Key Quantitative Results from DESI DR2 BAO Distances

Quantity	Value / Precision	Data Source
$\ell = 0,2,4$0 (isotropic BAO, typical precision)	0.24%	(Andrade et al., 18 Mar 2025), combined sample
$\ell = 0,2,4$1 (Hubble distance, Lyα)	$\ell = 0,2,4$2 (stat) $\ell = 0,2,4$3 (sys)	(Collaboration et al., 18 Mar 2025)
$\ell = 0,2,4$4 (transverse dist., Lyα)	$\ell = 0,2,4$5 (stat) $\ell = 0,2,4$6 (sys)	(Collaboration et al., 18 Mar 2025)
Hubble constant $\ell = 0,2,4$7 (BAO+BBN)	$\ell = 0,2,4$8 km s⁻¹ Mpc⁻¹	(Collaboration et al., 18 Mar 2025)
Sum of neutrino masses ($\ell = 0,2,4$9CDM)	$\xi_0$0 eV (95% CL)	(Collaboration et al., 18 Mar 2025)
Dark energy eqn. of state at $\xi_0$1	$\xi_0$2	(Efstathiou, 5 May 2025), BAO+CMB
Tension DESI BAO vs. CMB (distance scale)	$\xi_0$32.3$\xi_0$4	(Collaboration et al., 18 Mar 2025, Chen et al., 1 May 2025)
CDDR deviation (low z)	None significant (<1$\xi_0$5); small $\xi_0$6 at $\xi_0$7	(Wang et al., 15 Jun 2025, Zhang et al., 22 Jun 2025)
Calibration parameters (BAO, maser, SNIa)	$\xi_0$8 Mpc, $\xi_0$9	(Kanodia et al., 15 Jul 2025)

---

7. Synthesis and Outlook

DESI DR2 BAO distance measurements establish a new standard for the precision cosmology of the expansion history. They provide statistically powerful, systematically robust constraints on the cosmic distance–redshift relation, increasingly tight bounds on dark energy dynamics and curvature, and a uniquely stringent arena for testing model-independent fundamental relations such as the Cosmic Distance Duality Relation. While hints of late-time expansion anomalies and evolving dark energy exist—especially in joint analyses with SNe and CMB—careful consideration of systematic effects, dataset consistency, and calibration degeneracies remains essential. The next decade, leveraging further DESI expansions and multi-messenger cosmology, promises to either resolve these tensions or reveal deeper departures from the standard cosmological model.