Comparison between full and fast simulations in top physics

All analyses performed in the top quark sector with a full simulation of the ATLAS detector show a good agreement with fast simulation results. The purpose of this note is to understand why detailed detector effects are not so crucial in these studies. A systematic comparison, based on a ``Rome'' Data Challenge fully simulated data sample and an ATLFAST simulated one, is done on all top event objects, electrons, muons, neutrinos, light and b-jets. A very good agreement on reconstructed top kinematics variables is shown. This assess the robustness of the results obtained with fast simulations in the top quark sector, and more particularly the detailed estimates of the related systematic uncertainties, which are the key points of most of these analyses.