Energy reconstruction for a hadronic calorimeter using neural networks

Often calorimeters have a non-compensating response ( e / h ≠ 1 ) . Non-compensation degrades both resolution and linearity figures. To improve on this, weighting techniques are frequently applied. These techniques normally use linear combination of the energy deposited in the calorimeter cells or longitudinal samples. For the hadronic calorimeter of ATLAS, Tilecal, the use of a neural network is proposed to perform the energy reconstruction for pions, taking into account both linearity and energy resolution. Experimental data from testbeam periods were used to perform neural reconstruction.