Artificial neural networks for quasar selection and photometric redshift determination

7 pages, 7 figures; Context. Baryonic acoustic oscillations (BAO) and their effects on the matter power spectrum can be studied using the Lyman-α absorption signature of the matter density field along quasar (QSO) lines of sight. A measurement sufficiently accurate to provide useful cosmological constraints requires the observation of ~105 quasars in the redshift range 2.2 < z < 3.5 over ~8000 deg2. Such a survey is planned by the Baryon Oscillation Spectroscopic Survey (BOSS) project of the Sloan Digital Sky Survey (SDSS-III). Aims: We assess one of the challenges for this project, that of building from five-band imaging data a list of targets that contains the largest number of quasars in the required redshift range. In practice, we perform a stellar rejection of more than two orders of magnitude with a selection efficiency for quasars better than 50% to magnitudes as bright as g ~ 22. Methods: To obtain an appropriate target list and estimate quasar redshifts, we develop artificial neural networks (ANNs) with a multilayer perceptron architecture. The input variables are photometric measurements, i.e., the object magnitudes and their errors in the five bands (ugriz) of the SDSS photometry. The ANN developed for target selection provides a continuous output variable between 0 for non-quasar point-like objects to 1 for quasars. A second ANN estimates the QSO redshift z using the photometric information. Results: For target selection, we achieve a non-quasar point-like object rejection of 99.6% and 98.5% for a quasar efficiency of, respectively, 50% and 85%, comparable to the performances of traditional methods. The photometric redshift precision is on the order of 0.1 over the region relevant to BAO studies. These statistical methods, developed in the context of the BOSS project, can easily be extended to any quasar selection and/or determination of their photometric redshift.