110 likes | 201 Views
Energy scale calibration in MC (update). P.Gauzzi. Problems. EC A EC B. Overall 1.5% miscalibration in the energy scale Endcaps: three different behaviours: for central modules E clust > E kine for the other ones E clust < E kine Holes in the regions between the
E N D
Problems EC A EC B • Overall 1.5% miscalibration in the energy scale • Endcaps: three different behaviours: for central modules Eclust > Ekine for the other ones Eclust < Ekine • Holes in the regions between the long and short modules(x = ±163cm) • Y-behaviour of long mod. due to extra attenuation factor in CLUADJUST function (Matteo & Stefano) central modules (1, 2, 17, 18) long modules (3—12, 19—28) short modules (13—16, 29—32)
Old MC NEW MC OLD MC offset(MeV) slope offset(MeV) slope Barrel -2.84±0.03 0.9868±0.0002 -2.88±0.06 1.0060±0.0003 EC-Central -2.25±0.09 1.0500±0.0005 -2.0 ± 0.3 1.057 ± 0.001 EC-Long -3.31±0.04 0.9817±0.0002 -2.9 ± 0.1 1.0000±0.0005 EC-Short -4.3±0.2 0.935±0.001 -3.2 ± 0.4 0.966 ± 0.003 Almost the same offsets Barrel and EC-Long modules had slope = 1 EC-Central and EC-Short modules have the same slope
Old MC EC A EC B • The same three kinds of response • The same holes at ± 163 cm • Response was not decreasing with X • No Y-dependence central mod. (1, 2, 17, 18) long mod. (3—12, 19—28) short mod. (13—16, 29—32)
Three different responses • In the event generation and in the calorimeter reconstruction have been used different attenuation lengths Generation Reconstruction A 1(cm) 2(cm) A 1(cm) 2(cm) Barrel+ EC-Long 0.35 50 430 (pl.1, 2) 0.35 50 430 (pl.1, 2) 380 (pl.3) 380 (pl.3) 330 (pl.4,5) 330 (pl.4,5) EC_Central 0.35 50 430 (pl.1, 2) 0.35 50 330(all planes) 380 (pl.3) 330 (pl.4,5) EC 13, 29 0.35 50 330(all planes) 0.35 50 430 (pl.1, 2) 380 (pl.3) 330 (pl.4,5) EC other short mod. 0.35 50 330(all planes) 0.35 50 330(all planes)
Different 2 • For EC-Central we use thered curvein generation and the blue one in reconstruction the response is overestimated • For EC mod. 13 and 29 is the opposite:the response is underestimated the effect “short modules” is mainly due to mod. 13 and 29 that dominate the sample the holes at ± 163 cm are not holes; X= ± 163 cm is the position of the mod. 13 and 29 (a.u.) d (cm)
Y-dependence • Y-dependence of blue points due to the extra correction implemented in CLUADJUST Ycl (cm) • The correction is bigger for longer modules
Test • ~ 106 events (‘gg04’ production) • Redo the clustering with: 1) the same attenuation lengths of the generation 2) no extra correction along Y • Problem: when I dropped CELE and cluster banks and I tried to redo the clustering starting from MCEL, I got energies a factor of 5 bigger than before ! (already at MCEL level; it is not a clustering problem) • Divide all the cluster energies by 5 and see what happens
Test on Before After
Conclusions • The response is more uniform in the central part of the endcap, but decreases with increasing X and Y • The decreasing toward the module ends (Y coordinate) is bigger than in the old MC; to be investigated (this is the reason, I guess, of the extra correction by Matteo & Stefano) • With the help of Tommaso, we will compare data to MC looking for other possible differences to correct