GARLIC Marcel Reinhard LLR Ecole polytechnique Marcel Reinhard

GARLIC Marcel Reinhard LLR – Ecole polytechnique Marcel Reinhard LCWS ‘ 08, Chicago LLR - Ecole Polytechnique

The GARLIC algorithm • • • Based on REPLIC Seed search via 2 -dim energy projection in first 7 X 0 Clustering based on neighbour criterion Several iterations from front to back Originally designed for pointing photons, works for all angles • Rejection via simple criteria (#hits, minimum energy, seed criteria, . . . ) • + Computation of cluster variables (Eccentricity, width, direction, energy deposit in different regions, . . . ) • Correction for guard ring and module gaps Marcel Reinhard LLR - Ecole Polytechnique 2

6 Ge. V e- shower Marcel Reinhard LLR - Ecole Polytechnique 3

Gap correction • • • Introducing „Ghost hits“ in a gap between to adjacent hits Linear energy interpolation angle independent Sensible to position in the shower Expecting reasonable improvement Marcel Reinhard LLR - Ecole Polytechnique 4

Gap correction: performance Marcel Reinhard LLR - Ecole Polytechnique 5

Dirty events Marcel Reinhard LLR - Ecole Polytechnique 6

Additional Event cleaning • Use energy deposit in shower core to reject pions • Core = 2 x 2 pixel All +cherenkov +E_core/E_tot +1 cluster only Cherenkov: on/off All +E_core/E_tot &1 cluster only Marcel Reinhard LLR - Ecole Polytechnique 7

Linearity unclustered 2006 TB data MC Simulation Cluster values are fitted with a parabolic function to correct for non-linearity effects The quadratic turns out to be small Marcel Reinhard LLR - Ecole Polytechnique 8

Residuals to linearity unclustered 2006 TB data MC Simulation Linearity in all cases better than 1% Marcel Reinhard LLR - Ecole Polytechnique 9

Resolution unclustered 2006 TB data Marcel Reinhard LLR - Ecole Polytechnique MC Simulation 10