On the clustering of submillimeter galaxies

We measure the angular two-point correlation function of submillimeter galaxies (SMGs) from 1.1mm imaging of the COSMOS field with the AzTEC camera and ASTE 10m telescope. These data yield one of the largest contiguous samples of SMGs to date, covering an area of 0.72deg² down to a 1.26mJybeam ^-1 (1σ) limit, including 189 (328) sources with S/N ≥3.5 (3). We can only set upper limits to the correlation length r ₀, modeling the correlation function as a power law with pre-assigned slope. Assuming existing redshift distributions, we derive 68.3% confidence level upper limits of r ₀ ≲ 6-8h ^-1Mpc at 3.7mJy and r ₀ ≲ 11-12 h ^-1Mpc at 4.2mJy. Although consistent with most previous estimates, these upper limits imply that the real r ₀ is likely smaller. This casts doubts on the robustness of claims that SMGs are characterized by significantly stronger spatial clustering (and thus larger mass) than differently selected galaxies at high redshift. Using Monte Carlo simulations we show that even strongly clustered distributions of galaxies can appear unclustered when sampled with limited sensitivity and coarse angular resolution common to current submillimeter surveys. The simulations, however, also show that unclustered distributions can appear strongly clustered under these circumstances. From the simulations, we predict that at our survey depth, a mapped area of 2deg² is needed to reconstruct the correlation function, assuming smaller beam sizes of future surveys (e.g., the Large Millimeter Telescope's 6″ beam size). At present, robust measures of the clustering strength of bright SMGs appear to be below the reach of most observations.