Die DPG-Frühjahrstagung in Bonn musste abgesagt werden! Lesen Sie mehr ...
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
T: Fachverband Teilchenphysik
T 60: Various topics in elementary particle physics
T 60.7: Vortrag
Mittwoch, 1. April 2020, 18:00–18:15, L-2.017
Software Compensation in a Highly Granular Calorimeter using Principal Component Analysis — •Jack Rolph1 and Erika Garutti1,2 for the CALICE-D collaboration — 1Institute for Experimental Physics, Hamburg University, Luruper Chaussee 149, D-22761 Hamburg, Germany — 2Deutsches Elektronen-Synchrotron, Notkestraße 85, D-22607 Hamburg, Germany
Hadronic calorimeters are insensitive to ’invisible energy’ (neutrons, binding energy). Large event-by-event fluctuations thereby worsen energy resolution. This effect may be corrected for by weighting events offline. This procedure is known as software compensation.
In this analysis, Principal Component Analysis (PCA) was used to study correlations between observables measurable by the CALICE Analogue Hadronic Calorimeter (AHCAL) steel prototype on both a cell-wise and calorimeter-wise basis using simulation. This was performed to assess the usefulness of these observables with respect to estimating the ’invisible energy’ content of hadronic showers.
A weighting method was devised using PCA projections to measure the differences between a purely cell-wise method, independent of total measured energy and the state of the art for simulated 10 to 80 GeV negative pion showers. Relative to the control, the method was found to improve compensation by a maximum of 8.0 ± 0.9% and degrade compensation at by a maximum of 9.9 ± 1.2% at energies above and below 35 GeV respectively. Most significantly, edge effects observed in the state of the art due to limited statistics were found to be strongly suppressed by the use of purely local compensation.