Expected photon performance in the ATLAS experiment

We present the description and expected performance of the photon reconstruction and identification in the ATLAS experiment, with the software release used for the initial LHC data--taking at $\sqrt{s}=7TeV$. The latest approaches to the photon reconstruction and recovery from the electron collection are discussed, and their combined performance is reviewed. A new cut--based loose selection -- harmonized with the corresponding electron one and used for triggering purposes -- is presented; its expected efficiency, jet rejection and corresponding triggers rates are reviewed. The tight cut--based selections -- separately optimized for unconverted and converted photons -- are considered. Several isolation requirements based on calorimetric and tracking variables and targeting different signal efficiencies are discussed. The efficiencies of the selection cuts are discussed for the simulated benchmark prompt photons and $H\to\gamma\gamma$ signal events, as well as for high-$p_T$ photons from a graviton decay process $G\to\gamma\gamma$. Their rejection performances are estimated on background MonteCarlo jet samples, with the additional rejection factor obtained by adding the isolation requirements to the tight selection.