DESI: Dark Energy Spectroscopic Instrument

Updated 4 August 2025

Dark Energy Spectroscopic Instrument (DESI) is a wide-field, fiber-fed spectroscopic facility designed to map the universe’s large-scale structure using BAO and RSD measurements.
It employs 4,000–5,000 robotic fiber positioners and ten three-arm spectrographs covering 360–980 nm to capture high-quality spectral data from millions of celestial objects.
DESI’s survey strategy delivers sub-percent distance measurements and rigorous tests of dark energy, inflationary models, and neutrino masses through extensive 3D mapping.

The Dark Energy Spectroscopic Instrument (DESI) is a massively multiplexed, fiber-fed spectroscopic facility installed on the 4-meter Mayall Telescope at Kitt Peak National Observatory. Conceived to address fundamental questions in cosmology, DESI’s scientific mandate includes high-precision mapping of the Universe’s large-scale structure, with primary objectives encompassing measurements of baryon acoustic oscillations (BAO), redshift-space distortions (RSD), the summation of neutrino masses, and tests of primordial inflationary scenarios. With a focal plane populated by 4,000–5,000 independent robotic fiber positioners and a spectral system spanning 360–980 nm, DESI systematically targets emission-line galaxies (ELGs), luminous red galaxies (LRGs), and quasi-stellar objects (QSOs) across a contiguous survey footprint of up to 14,000–18,000 square degrees. Its data products—three-dimensional galaxy and Lyα forest maps—enable 1%–level measurements of the cosmic distance scale across 35 redshift bins, providing rigorous constraints on models of cosmic acceleration, dark energy, and fundamental physics (Levi et al., 2013).

1. Scientific Rationale and Core Objectives

DESI is explicitly designed as a Stage IV dark energy experiment. Its principal goals are to:

Probe the expansion history of the Universe using BAO as a standard ruler, yielding sub-percent distance scale accuracy.
Measure the gravitational growth rate through redshift-space distortions, enabling direct tests of general relativity on cosmic scales.
Provide constraints on the primordial power spectrum and signatures of inflation, including measurements of primordial non-Gaussianity.
Determine the sum of neutrino masses and infer the effective number of relativistic species ( $N_\nu$ ), with sensitivity sufficient to distinguish neutrino mass hierarchies.

Each of these objectives is encoded in explicit operational requirements, such as measuring $f \sigma_8$ to better than 2% per redshift bin, determining the BAO scale with aggregate precision of $\sim 0.17\%$ , and achieving redshift errors of less than $0.001(1 + z)$ for all key targets (Levi et al., 2013, Collaboration et al., 2016).

2. Instrumentation and Survey System

Telescope and Corrector

DESI is installed at the prime focus of the Mayall 4-m telescope, enhanced with a new six-lens optical corrector delivering a 3.2° field of view (∼8 deg² per pointing). Image quality targets (FWHM ≲ 0.4″ at zenith, bandpass 360–980 nm) are realized via fused silica and borosilicate optics, two forming an atmospheric dispersion compensator effective up to $z = 60^\circ$ (Abareshi et al., 2022, Miller et al., 2023).

Robotic Fiber Positioners and Focal Plane

A focal plane array comprised of 4,000–5,020 independently-actuated fiber positioners (arranged in ten “petals,” each servicing 500 fibers) can be reconfigured in less than two minutes, achieving $\leq$ 5–10 μm RMS placement (∼0.2″ on sky). Positioners operate via two-axis kinematics:

$x = R_1 \cos(\theta) + R_2 \cos(\theta+\phi) \ y = R_1 \sin(\theta) + R_2 \sin(\theta+\phi)$

where $\theta$ and $\phi$ are orthogonal rotation angles (Silber et al., 2022). The system includes 6 guide cameras, 4 wavefront cameras, and 123 fiducial point sources integrated with a fiber view metrology camera for closed-loop position correction.

Fiber and Spectrograph System

Light is routed via fusion-spliced fibers (minimized focal ratio degradation, $\sim$ 2% connection loss) to ten identical three-arm spectrographs, each with blue (360–555 nm; $R \gtrsim 1500$ ), red (555–656 nm; $R \gtrsim 3000$ ), and NIR (656–980 nm; $R \gtrsim 4000$ ) channels. Bench-mounted spectrographs, with anti-reflection coatings and stringent PSF stability ( $<1\%$ bias), ensure high throughput ( $>40\%$ instrument-only; peak efficiency $>50\%$ including atmosphere is projected in optimal conditions) (Collaboration et al., 2016, Levi et al., 2019, Abareshi et al., 2022).

3. Observational Strategy, Targeting, and Data Collection

DESI’s main survey is structured to optimize both statistical power and systematic control (Collaboration et al., 2016):

Target selection leverages optical (e.g., Legacy Surveys) and mid-infrared (WISE) photometry to identify:
- ELGs: star-forming galaxies ( $z \lesssim 1.7$ ) via [O II] emission.
- LRGs: high-mass, clustered galaxies ( $z \lesssim 1$ ).
- QSOs: $z \sim 1$ –$2$ as direct tracers, $z > 2$ for Ly $\alpha$ forest analysis.
Survey scheduling employs an “afternoon planning” pipeline for field selection, prioritizing tile completion, declination, and neighbor overlap. Tiles are observed with adaptive exposure control, guided by real-time transparency, seeing, and airmass metrics (Schlafly et al., 2023).
Bright time observations target a magnitude-limited (~10M objects, $z \sim 0.2$ ) bright galaxy sample (BGS) and stars for Milky Way science.

Spectroscopic data are acquired via 5000 fibers in parallel, with each field targeting up to 5000 objects per ∼20–25 minute exposure. Fiber repositioning proceeds in <2 minutes, overlapped with CCD readout ( $\sim$ 42s at 100 kHz pixel clock) (Collaboration et al., 2016, Honscheid et al., 2018).

4. Data Analysis, Calibration, and Products

The reduction pipeline implements:

Bias/dark correction, cosmic ray rejection, spectral extraction using “spectroperfectionism” with full-resolution matrix recovery (Collaboration et al., 18 Mar 2025).
Wavelength calibration via arc lamps and sky lines; flux calibration uses spectrophotometric standard stars.
Coaddition of exposures via inverse-variance weighting.
Automated classification and redshifting (e.g., redrock), with purity >99% for all target classes (Guy et al., 2022).

Data products are organized into tile-based and HEALPix-grouped catalogs, supporting two-point correlation and power spectrum analyses (BAO, RSD). Value-added catalogs (VACs) include emission-line fits, stellar parameters (for Milky Way science), and Ly $\alpha$ forest statistics. DR1 comprises $\sim$ 18.7M unique high-confidence redshifts (13.1M galaxies, 1.6M quasars, 4M stars), already exceeding pre-existing extragalactic samples by a factor of four (Collaboration et al., 18 Mar 2025).

5. Cosmological Measurements and Impact

DESI’s anticipated and validated measurement capabilities include:

BAO distance errors per $\Delta z = 0.2$ bin: 0.35–1.1%; aggregate $\sim$ 0.17% (Levi et al., 2013). Forecast BAO scale precision: 0.28% ( $z<1.1$ ), 0.39% ($1.1Collaboration et al., 2023).
RSD growth rate constraints: $f\sigma_8$ to better than 2% per bin.
Neutrino mass sum: projected 1 $\sigma$ error $\sim$ 0.02 eV.
Primordial non-Gaussianity and inflationary features via detailed shape and anisotropy of the 3D power spectrum.
Multi-tracer cross-checks using LRG, ELG, QSO, BGS, and Milky Way samples to optimize clustering measurements and control systematics (Collaboration et al., 2023).

Theoretically, growth and equation of state parameters are expressed as

$f = \frac{d\ln D}{d\ln a} \, , \quad w(z) = w_p + (a_p - a) w'$

where $D$ is the linear growth factor and $w_p$ , $w'$ parametrize the dark energy equation of state (Levi et al., 2013).

6. Technical Performance, Validation, and Operations

Operational validation has demonstrated:

Fiber placement accuracy $\leq$ 5–10 μm RMS, with iterative closed-loop corrections using FVC metrology (Silber et al., 2022).
Spectrograph throughput matching laboratory predictions, with median SNR performance exceeding requirements (e.g., SNR $>0.5\,/\sqrt{\rm \AA}$ for $z>2$ quasars with $0.28\times10^{-17}$ erg/s/cm²/Å at 380 nm in 4000s) (Abareshi et al., 2022).
Systematic control via uniform field depth, rigorous guiding and focus (tracking stability $<$ 0.03 arcsec), and low residual sky background (Honscheid et al., 2018, Abareshi et al., 2022).
Survey cadence exceeding expectations: as of early main operations, the dark survey progressed 7–14% ahead of its projected schedule (Schlafly et al., 2023).
Extensive commissioning and survey validation, including a “One-Percent Survey” of 140 deg² that validated redshift success rates, exposure strategies, and precision forecasts (Collaboration et al., 2023).

7. Significance and Broader Legacy

DESI establishes a new benchmark in wide-field, massively multiplexed spectroscopic cosmology, influencing future survey instrument design (e.g., modular petal architecture, fusion-spliced fiber systems). Its comprehensive, value-added, and publicly available datasets (including LSS catalogs, stellar parameter VACs, Lyα forest statistics) underpin precision cosmology—constraining dark energy, testing general relativity, probing neutrino mass, and tracing the history of cosmic structure formation (Collaboration et al., 18 Mar 2025). The data infrastructure facilitates ancillary studies in galaxy evolution, Milky Way structure, quasar physics, and the intergalactic medium.

By delivering high-density, wide-field, spectroscopic measurements, DESI uniquely enables precision constraints on the expansion history, the growth of structure, and parameters characterizing fundamental physics. Its operational paradigm—fast, accurate fiber reconfiguration with real-time operations integration—provides an archetype for next-generation cosmological experiments.