PRIMUS Overview

Motivation: Redshifts are crucial for studying distant galaxies.  They provide an estimate of distance, allowing one to relate observed to intrinsic properties and to cut through the projection on the sky to map the cosmic web and large-scale structure.  Wide-field low-redshift surveys such as the SDSS have given us a detailed understanding of the multivariate properties of galaxies, including joint distributions of luminosity, color, morphology, and environment.  Now the question is how these galaxy properties evolve with time.  Deep imaging surveys have covered as much volume at z=1 as the SDSS has at z=0.1. Much of this imaging is panchromatic, with UV and mid-IR data providing critical information about star-formation and stellar mass.  The Spitzer SWIRE and GTO surveys and GALEX Deep Imaging Survey are imaging ~60 square degrees to depths appropriate for z=1 galaxies.  However, these surveys lack redshifts.

PRIMUS: The goal of the PRIsm MUlti-object Survey (PRIMUS) is to provide redshifts for a dense set of faint galaxies in the ~10 deg^2 of deep imaging accessible from Magellan that has panchromatic coverage. With a sample of >100,000 PRIMIUS redshifts, we will quantify the clustering and multivariate properties of galaxies at z=0.2-1 with similar precision as SDSS and 2dF, allowing detailed quantification of the evolution of the masses, luminosities, star formation rates, and clustering of galaxies with time.  As conventional spectroscopy of faint galaxies is observationally very expensive, broad-band photometric redshifts are often used.  However, they typically have statistical errors of 5% in redshift and the systematic errors are the limiting uncertainty in the quantification of galaxy properties and evolution.

With PRIMUS, we use Magellan/IMACS to obtain both a higher survey speed and higher spectroscopic resolution (R~30) than photometric redshift surveys. Instead of imaging multiple filters and wasting most of the detector on blank sky, the PRIMUS strategy is to acquire low resolution spectroscopy using a high-throughput prism behind an IMACS slit mask.  By using the detector pixels to multiplex spectrally, we cut the survey time by over an order of magnitude.

Relative to conventional spectroscopy, PRIMUS's low resolution is limited.  We measure redshifts to <0.5%, learn about the broad-band SEDs, and find the strongest emission lines (notably broad AGN lines).  We do not measure detailed line diagnostics.  However, redshift-space distortions smear out the cosmic web by ~500 km/s, such that redshifts more accurate than about 1% do not improve clustering analyses.  For our science goals, exchanging more precise redshifts for a faster, wider redshift machine to cover 10 times the area is an excellent trade.

We will use PRIMUS to quantify the multivariate distributions of optical light, stellar masses (from Spitzer NIR), and star formation rates (from UV and Spitzer mid-IR) of galaxies as a function of environment and across multiple redshift bins out to z=1.  The evolution of some of these properties have been detected, but it is critical to measure them as joint distributions, as we know that galaxies at low redshift are not a single monolithic group but instead have complicated differences as a function of color and mass.  We clearly must separate galaxies into at least this two-dimensional set when measuring evolution and connecting to theory.  We will also measure galaxy correlation functions, again across time and galaxy properties, so as to connect galaxies to their host parent halo masses.  Galaxy clustering also provides a critical test of galaxy formation models as it restricts the association of progenitors and descendants across redshift. 

All of these measurements are critical diagnostics of galaxy formation, but all require large, wide, uniform samples to control cosmic variance.  Cosmic variance is particularly troublesome for measuring clustering and environment; we know from SDSS that samples of 100,000 objects suffer from significant excursions and that measuring the subtle (10%) evolution of clustering requires very large data sets.  PRIMUS will be 4 times larger than other redshift surveys at z~0.5-1, and it will be the only faint galaxy redshift survey to compare in size to the low-redshift SDSS and 2dF surveys, which is an appropriate standard for considering the measurement of evolution. Moreover, PRIMUS will observe a significant fraction of the available Spitzer and deep GALEX imaging in the southern hemisphere, thereby supporting the best multiwavelength wide-field data for z~1 galaxies.