DESI DR1: Precision Cosmology Data Release

• DESI DR1 is a comprehensive spectroscopic dataset featuring precise BAO distance measurements and extensive redshift catalogs for cosmological studies.
• The survey employs a multi-object spectrograph on the Mayall Telescope with 5000 fibers, achieving >99% redshift purity and robust calibration.
• Systematic analyses include advanced flux and wavelength calibration, covariance modeling, and model-agnostic reconstructions that underpin precise dark energy constraints.

The Dark Energy Spectroscopic Instrument (DESI) Data Release 1 (DR1) marks a transformative milestone in observational cosmology and spectroscopic survey science. DESI DR1, the inaugural public release from the DESI project, delivers unprecedentedly precise baryon acoustic oscillation (BAO) distance measurements, extensive spectroscopic data products, and robust systematic and cosmological validation. The dataset initiates a broad range of analyses targeting the late-time expansion history of the universe, tests of fundamental physics, dark energy model constraints, and the evolution of large-scale structure.

1. Survey Strategy, Data Products, and Processing

DESI operates on the 4-m Mayall Telescope and utilizes a multi-object spectrograph covering 3600–9800 Å with a resolution R=2000–5000. Its focal plane supports 5000 robotically positioned fibers, enabling simultaneous spectroscopy of a range of target classes including luminous red galaxies (LRGs), emission line galaxies (ELGs), quasars, and stars. Each spectrographic exposure is processed by a pipeline that performs instrumental calibration (bias, dark, flat, illumination), optimal 2D spectral extraction via forward-modeled PSF, precise wavelength calibration (including telluric line corrections), and rigorous flux calibration based on F-star standards and PHOENIX atmosphere models (Guy et al., 2022).

The DR1 release comprises:

• Wavelength- and flux-calibrated 1D spectra for over 2 million targets.
• Coadded spectra (weighted per exposure) on a fixed log-linear wavelength grid.
• Redshift catalogues with likelihood-based classifications via empirical eigenspectra and support from CNN-based approaches for quasars.
• Diagnostic matrices (resolution, PSF) and associated per-target metadata.

Automated quality assurance is integrated at every step, with real-time per-exposure monitoring and stringent nightly validation. Achieved spectroscopic redshift purities surpass 99% for all main target classes, and redshift success rates are 98–99% for galaxies and quasars in nominal signal-to-noise regimes.

2. BAO Distance Measurements and Covariance Structure

DESI DR1 provides BAO distance measurements across multiple tracers and redshift bins, yielding:

• Transverse ($D_M(z)$), radial ($D_H(z)$), and isotropic volume-averaged ($D_V(z)$) comoving distances, all in units of the sound horizon at drag epoch ($r_d$).
• Seven effective redshift bins spanning $0.31\lesssim z\lesssim2.33$, with per-bin fractional uncertainties $\sigma_{D_V}/D_V \lesssim 2\%$ at low/intermediate $z$.
• Extensive off-diagonal covariance information characterizing the correlations between distance measures and redshift bins, published as a full 13×13 matrix (Calderon et al., 2024).

The dataset forms the geometric backbone for precision cosmology and is used both directly in parametric fits (e.g., $w_0w_a$CDM) and as input for model-agnostic reconstruction frameworks.

3. Systematic Uncertainties in Target Selection and Spectroscopy

DR1 includes a comprehensive accounting of spectroscopic systematics, particularly for ELGs, which constitute the dominant sample for large-scale structure analyses (Yu et al., 2024). Main sources, definitions, and mitigations include:

• Redshift Completeness and Correction Weights:
  • $f_{\rm goodz}$ (success rate), $w_{\rm zfail}$ (compensating weight), $D_H(z)$0 (per-fiber residual correction).
  • Global $D_H(z)$1; remaining failures corrected by linear weights as functions of spectroscopic SNR and focal position.
• Catastrophic Redshift Errors:
  • 0.26% of secure redshifts are catastrophic ($D_H(z)$2 km/s), with explicit modeling distinguishing sky-residual confusion at $D_H(z)$3, double-object spectra, and other sources.
  • Dedicated simulation campaigns indicate that these rates shift two-point correlation functions and power spectra by $D_H(z)$4; pipeline marginalization absorbs almost all bias.
• Redshift Uncertainty:
  • For secure ELG redshifts, the pairwise velocity distribution is Lorentzian with width 8.5 km/s, substantially reducing the statistical broadening on small-scale clustering relative to eBOSS.
• Robustness and Future Prospects:
  • These corrections have minor impact on BAO and RSD fits at current precision, but anticipated sub-percent cosmology (e.g., primordial non-Gaussianity searches) and new target classes (LAEs) demand further refinements.

A dedicated software suite (https://github.com/Jiaxi-Yu/modelling_spectro_sys) provides practical recipes for systematic modeling in DESI-matched simulations.

4. Cosmological Inference: Dark Energy and Structure Formation

Parametric and Model-Agnostic Reconstructions

DESI DR1 BAO, combined with Planck CMB and various supernova datasets (PantheonPlus, Union3, DES-SN5YR), supports a spectrum of dark energy analyses:

• One-Parameter Mukherjee Model:
  • Parameterizes the DE equation of state via a single index $D_H(z)$5 ($D_H(z)$6: $D_H(z)$7CDM; $D_H(z)$8: quintessence; $D_H(z)$9: phantom), reducing parameter degeneracy relative to the two-parameter CPL model (Fikri et al., 2024).
  • With Planck+DESI:
  • $D_V(z)$0, $D_V(z)$1, $D_V(z)$2 km/s/Mpc.
  • Bayesian evidence (revised Jeffreys scale): $D_V(z)$3 for Phantom over $D_V(z)$4CDM, i.e., weak but consistent.
  • Addition of PantheonPlus brings the mean closer to $D_V(z)$5CDM ($D_V(z)$6), but tension with local $D_V(z)$7 increases; inclusion of SH0ES reduces $D_V(z)$8 tension to $D_V(z)$9, with a $r_d$0 pull toward mildly phantom DE.
  • Linear-perturbation-level differences with $r_d$1CDM are within $r_d$2 in $r_d$3 and power spectra, well below cosmic variance.
• Model-Agnostic Reconstructions (GP and Crossing Statistics):
  • Gaussian-Process Multi-Task Regression and Chebyshev Crossing Expansions recover $r_d$4 and $r_d$5 directly from distances, without assuming a parametric DE model (Dinda et al., 2024, Calderon et al., 2024).
  • Both approaches find hints of $r_d$6 (phantom behavior) at low $r_d$7, but $r_d$8CDM remains within $r_d$9 (GP) or is only excluded at $0.31\lesssim z\lesssim2.33$0 in some crossing-statistics runs.
  • Joint BAO+SN-only fits enable $0.31\lesssim z\lesssim2.33$1 determination at $0.31\lesssim z\lesssim2.33$2, decoupled from CMB physics.
  • A plausible implication is that, while data permit evolving or emergent DE, the deviation from $0.31\lesssim z\lesssim2.33$3CDM is not yet decisive.
• Physics-Focused Dark Energy Models:
  • Thawing quintessence, emergent (phase-transition-type), and mirage classes are tested explicitly (Lodha et al., 2024).
  • The mirage class, which interpolates between emergent and thawing behavior, achieves $0.31\lesssim z\lesssim2.33$4 relative to $0.31\lesssim z\lesssim2.33$5CDM, with only one extra parameter, nearly matching standard $0.31\lesssim z\lesssim2.33$6CDM ($0.31\lesssim z\lesssim2.33$7).
  • Bayes-factor analyses suggest moderate-to-strong evidence for mirage-like evolution over $0.31\lesssim z\lesssim2.33$8CDM when including all datasets.

DR1 BAO and Full-Shape (FS) Tension and Stability

• Several studies highlight that DESI DR1 BAO alone, using $0.31\lesssim z\lesssim2.33$9CDM with relaxed priors, does not confirm cosmic acceleration ($\sigma_{D_V}/D_V \lesssim 2\%$0 ruled out at $\sigma_{D_V}/D_V \lesssim 2\%$1) (Colgáin et al., 6 Apr 2025).
• Combined BAO+FS analyses yield matter densities and equation-of-state parameters entirely consistent with $\sigma_{D_V}/D_V \lesssim 2\%$2CDM, with no persistent evidence for $\sigma_{D_V}/D_V \lesssim 2\%$3.
• Outlier analysis in redshift bins attributes apparent evolution signals to statistical fluctuations, with leading roles transferring between bins in DR1 and DR2.
• Only the inclusion of SN data robustly restores late-time acceleration confirmation.

5. Large-Scale Structure, Lensing, and Small-Scale Probes

Quasar–CMB Lensing Tomography

• Cross-correlation of DESI DR1 quasars ($\sigma_{D_V}/D_V \lesssim 2\%$41.2 million, $\sigma_{D_V}/D_V \lesssim 2\%$5) with Planck CMB lensing (PR4, NPIPE) yields tomographic growth measurements at $\sigma_{D_V}/D_V \lesssim 2\%$6 (Belsunce et al., 27 Jun 2025).
• In three redshift bins ($\sigma_{D_V}/D_V \lesssim 2\%$7), joint fits incorporating BAO priors yield:
  • $\sigma_{D_V}/D_V \lesssim 2\%$8
  • $\sigma_{D_V}/D_V \lesssim 2\%$9
• Combined with ACT DR6 and Planck lensing:
  • $z$0 km/s/Mpc, independent of the sound horizon.
• These results are $z$1 higher than Planck $z$2CDM, but not in statistically significant tension.

3×2pt Lensing-Galaxy Validation

• DESI DR1 clustering is validated in joint analysis with overlapping KiDS, DES, and HSC lensing data (Emas et al., 7 Oct 2025).
• Scale cuts are chosen such that remaining biases in $z$3 and $z$4 from nonlinear modeling, bias, and baryons remain $z$5.
• Shear-ratio tests, using angle-averaged tangential shear ratios, are consistent across all surveys and improve constraints on source redshift distributions and intrinsic alignment amplitudes.

1D Lyα Power Spectrum

• FFT-based measurement of the 1D Lyα forest power spectrum from DR1 constitutes the most precise intermediate-resolution dataset to date (Ravoux et al., 14 May 2025).
• Advanced noise characterization (cross-exposure estimator), analytic covariance estimation, and rigorous systematics assessment yield statistical errors $z$6–$z$7 smaller than prior eBOSS, with systematic errors dominant at most $z$8.
• Agreement with QMLE and high-res data is within $z$9.
• Cosmological constraints: $w_0w_a$0, $w_0w_a$1, $w_0w_a$2 eV, $w_0w_a$3 keV for WDM, and $w_0w_a$4 eV, with improved IGM thermal history reconstruction.

6. Prospects, Limitations, and Future Directions

DESI DR1 establishes a precision baseline for both geometric and dynamical cosmological inference. However, several nontrivial caveats remain:

• The apparent DE dynamics in parametric fits often trace to statistical fluctuations or specific outlier bins; future releases with expanded statistics are required to substantiate or refute these hints.
• Full confirmation of late-time acceleration and discrimination between $w_0w_a$5CDM and dynamical DE still require combination with SN Ia and independent lensing data.
• Small residual systematics (sky subtraction for ELGs, redshift misidentifications, fiber throughput) are at or below the sub-percent level but could become relevant for future sub-percent level tests (e.g. $w_0w_a$6 constraints).
• The ongoing development of modeling, calibration, and emulation pipelines (e.g., for Lyα, clustering, CMB lensing) is critical for robust interpretation.

A plausible implication is that DESI DR1, while enabling a host of new cosmological analyses, also delineates the precision frontier for next-generation spectroscopic surveys. The methodologies, systematic control, and statistical tools developed will frame the analysis approaches for forthcoming DESI releases and next-generation extragalactic surveys.