Directly imaging damped Lyman α galaxies at z > 2-I. Methodology and first results

We present the methodology for, and the first results from, a new imaging programme aimed at identifying and characterizing the host galaxies of damped Lyman alpha absorbers (DLAs) at z greater than or similar to 2. We target quasar sightlines with multiple optically thick HI absorbers and use the higher redshift system as a 'blocking filter' (via its Lyman-limit absorption) to eliminate all far-ultraviolet (FUV) emission from the quasar. This allows us to directly image the rest-frame FUV continuum emission of the lower redshift DLA, without any quasar contamination and with no bias towards large impact parameters. We introduce a formalism based on galaxy number counts and Bayesian statistics with which we quantify the probability that a candidate is the DLA host galaxy. This method will allow the identification of a bona fide sample of DLAs that are too faint to be spectroscopically confirmed. The same formalism can be adopted to the study of other quasar absorption-line systems (e. g. Mg II absorbers). We have applied this imaging technique to two quasi-stellar object sightlines. For the z similar to 2.69 DLA towards J073149+285449, a galaxy with impact parameter b = 1.54 arcsec = 11.89 h(72)(-1) kpc and an implied star formation rate (SFR) of similar to 5 h(72)(-2) M-circle dot yr(-1) is identified as the most reliable candidate. In the case of the z similar to 2.92 DLA towards J211444-005533, no likely host is found down to a 3 sigma SFR limit of 1.4 h(72)(-2) M-circle dot yr(-1). Studying the HI column density as a function of the impact parameter, including six DLAs with known hosts from the literature, we find evidence that the observed HI distribution is more extended than what is generally predicted from numerical simulation.