Distinguishing Isolated Photons from Jets

We have developed isolation and shower-shape variables that are optimized for reducing the jet background for a high transverse-energy photon signal. To help understand the relative importance of each variable, we have computed the background rejection power as a function of signal efficiency for these variables as well as for some simple combinations of variables. We find that a combination of tracker plus ECAL information can give very significant background rejection power. By adding information from the HCAL, the rejection power can be improved. In addition we find that a very significant reduction in background can be achieved by adding photon shower shape information from the ECAL to the Neural Net inputs. About 1 in 200 jets contains a high E_T electromagnetic shower that is reconstructed by the CMS software and is therefore a potential background for photon detection. An additional jet rejection factor of 100 with 80% photon efficiency (or 400 with 50% efficiency) can be achieved by using the isolation variables developed here.