Luminosity Determination Using the ATLAS Detector

We present the algorithms and results of the measurement of the luminosity in the ATLAS experiment during the first LHC run at energies of $\sqrt s = 900$ GeV and $\sqrt s = 7$ TeV. The LHC luminosity is determined in real time approximately once per second using a number of detectors and algorithms, each having different acceptances, systematic uncertainties and sensitivity to background. These results are displayed in the ATLAS control room and archived every two minutes; a single ``preferred'' measurement is reported to the LHC. During offline analysis, additional luminosity algorithms are studied and are compared to online results to further constrain systematic uncertainties on the measurement. Relative luminosities between detectors and methods agree to within a few per cent. Determination of the absolute luminosity using Monte Carlo calibrations is limited by a ~20\%\ systematic uncertainty from the modeling of diffractive components of the cross section. Method-specific systematic uncertainties due to the modeling of detector response and trigger are ~5%. Smaller uncertainties of 11% are obtained using an absolute calibration of the luminosity via beam separation scans and are dominated by the systematic uncertainty on the measurement of the LHC beam current. Monte Carlo based calibrations obtained using the PYTHIA MC09 (PHOJET) event generator differ from those obtained from the beam scans by 8% to 27% (26% to 35%).