Performance and Improvements of the ATLAS Jet Trigger System

At the harsh conditions of the LHC, with proton bunches colliding every 50ns and up to 40 pp interactions per bunch crossing, the ATLAS trigger system has to be flexible to maintaining an unbiased efficiency for a wide variety of physics studies while providing a fast rejection of non-interesting events. Jets are the most commonly produced objects at the LHC, essential for many physics measurements that range from precise QCD studies to searches for New Physics beyond the Standard Model, or even unexpected physics signals. The ATLAS jet trigger is the primary means of selecting events with high p_T jets and its good performance is fundamental to achieve the physics goals of ATLAS. The ATLAS trigger system is divided in three levels, the first one (L1) being hardware based, with a 2mu s latency, and the two following ones (called collectively High Level Triggers or HLT) being software based with larger processing times. It was designed to work in a Region of Interest (RoI) based approach, where the second level trigger (L2) was limited to verifying L1 signals in a narrow region of the detector around them. The RoI based strategy was not well-suited for multi-jet events since it lead to pathologies and efficiency losses. A re-design of the jet trigger to overcome this difficulty has happened in between 2011, with the implementation of the full calorimeter unpacking at EF, and 2012 with the introduction of a full scan at L2. It is now also possible to run a variety of jet algorithms, both at L2 and at EF, reducing even further possible biases for physics studies. The presentation describes the challenges of the ATLAS Jet Trigger system, its original limitations and the new developments that were implemented to overcome them. Performance results obtained with 2011 data are shown, including efficiency measurements with respect to the offline reconstructed jets, resolutions and jet energy scale studies.