The DESI Experiment, a whitepaper for Snowmass 2013

TL;DR

DESI proposes a massively multiplexed, fiber-fed spectroscopic survey on the Mayall 4-m telescope to map tens of millions of galaxies and QSOs, enabling precise BAO and redshift-space distortion measurements across 0.5 < z < 3.5. By delivering high-density tracers over a wide footprint and leveraging Lyα forest data at z>2, DESI aims to constrain the expansion history, growth of structure, neutrino masses, and primordial physics with a DETF Stage-IV figure of merit approaching or exceeding a few hundred. The instrument design, target-selection strategy, and imaging requirements are tailored to maximize survey efficiency and minimize systematics, positioning DESI as a critical bridge between DES and LSST and yielding transformative cosmological insights. Overall, DESI is expected to deliver substantially improved dark energy constraints and neutrino physics opportunities, sustaining US leadership in cosmology through the next decade.

Abstract

The Dark Energy Spectroscopic Instrument (DESI) is a massively multiplexed fiber-fed spectrograph that will make the next major advance in dark energy in the timeframe 2018-2022. On the Mayall telescope, DESI will obtain spectra and redshifts for at least 18 million emission-line galaxies, 4 million luminous red galaxies and 3 million quasi-stellar objects, in order to: probe the effects of dark energy on the expansion history using baryon acoustic oscillations (BAO), measure the gravitational growth history through redshift-space distortions, measure the sum of neutrino masses, and investigate the signatures of primordial inflation. The resulting 3-D galaxy maps at z<2 and Lyman-alpha forest at z>2 will make 1%-level measurements of the distance scale in 35 redshift bins, thus providing unprecedented constraints on cosmological models.

Figures (4)

• Figure 1: The power of DESI is in both the precision and the wide range of redshifts it will cover, making it competitive even with the Euclid space-based mission. Shown are the fractional error on the BAO distance scale (isotropic dilation factor), as a function of redshift, per unit $ln(a)$ (in other words, the effect of any arbitrary redshift bin width $\Delta z$ is removed in this plot). Errors from the Ly$\alpha$ forest measurement, which dominate at $z>1.8$, are computed following McDonald & Eisenstein (2007), with a modest but significant additional contribution from cross-correlations with quasar density. We assume here an optimistic 50 million galaxies for EUCLID.
• Figure 2: The Dark Energy Spectroscopic Instrument is shown in the figure. A new corrector delivering a 8 deg$^2$ field-of-view is installed at the Mayall 4m telescope and brings light to an array of 5000 fibers. Each fiber tip can be moved independently to a random galaxy position by mechanical actuators under computer control. The fibers are brought to a system of spectrographs that can simultaneously observe the dispersed light from all 5000 fibers.
• Figure 3: Aitoff projection of the low-reddening ($E(B-V) < 0.094$) extragalactic footprint suitable for the DESI survey. The total footprint is 23,800 deg$^2$, where 11,800 deg$^2$ is the equatorial region. In this calculation the northern + equatorial sky visible from KPNO totals 18,000 deg$^2$. The imaginary lines are at Dec$=-28$ deg and Dec$=+28$ deg.
• Figure 4: Improvement in parameters over Planck plus BOSS BAO (projected finished Planck, including polarization). Yellow bars show the improvement with DESI BAO, including Ly$\alpha$ forest. Blue bars show the additional improvement when optimistic broadband power is used, with $k<0.2$ h/Mpc. Green additionally adds potential constraints from broadband (including small-scale 1D) Ly$\alpha$ forest power. Dark energy parameters are defined by the equation of state $w(z)=w_p + (a_p-a)w^\prime$, where $a_p$ is chosen to make the errors on $w_p$ and $w^\prime$ independent. $\omega_k \equiv \Omega_k h^2$ is the equivalent physical density of curvature. $\sum m_\nu$ is the sum of masses of neutrinos, in eV. $\alpha_s$ and $n_s$ parameterize the inflationary perturbation power spectrum, i.e., $P_{\rm inflation}(k) \propto k^{n_s +\frac{1}{2}\alpha_s \ln\left(k/k_\star\right)}$.