Short Overview on calibration Jet calibration as it is now What we correct and how Latest news on default calibration from F.Paige step to final comparison of existing schemes Future steps May 4th, 2006 Hadron Calibration - Munich C.Roda
What we mean for calibration… Reconstructed Jet (with cone, or KT) MC particle Jet (with cone, or KT) Reconstruction We apply the calibration algorithm to take the scale of the Reconstructed jet from the EM scale to the to the MC jet scale (compensation). Parton Jet invertire : a dx rico a sx gen May 4th, 2006 Hadron Calibration - Munich C.Roda
EM Scale Difference from 1 is due to: e/h non compensation Energy lost in dead material B field effect Jet clustering effects A.Gupta May 4th, 2006 Hadron Calibration - Munich C.Roda
Size of contribution from various effects Results from Atlfast: comparison of Erec/Emc P.Francavilla I.Vivarelli May 4th, 2006 Hadron Calibration - Munich C.Roda
Size of contribution of various effects Results from Atlfast P.Francavilla I.Vivarelli May 4th, 2006 Hadron Calibration - Munich C.Roda
Present flow of Jet Calibration EM Scale, noise suppresion EM Scale, more noise suppresion CaloCells CaloTowers / TopoClusters Cone, Kt algorithm Calibration scheme Calibrated Jets to MC energy Uncalibrated Jets In situ calibration, underlying, pileup, out of cone … Calibrated Jets May 4th, 2006 Hadron Calibration - Munich C.Roda
Use global jet scale, See A.Dotti presentation, (Pisa groups) Calibration scheme Energy resolution minimization (with linearity constraint) to obtain weights W(): W(Jet Energy) x Sample energy W(Cell Energy density) x Cell energy W(Cell Energy, Jet Energy) x Cell energy All weights are also function of eta. Different functions are assumed to describe the weight energy dependence. Simple and fast but less performant, useable at trigger level ? (A.Gupta) Default in Athena, used in all physics analysis up to now and for ETMiss. Indicated with H1 (F.Paige) Use global jet scale, See A.Dotti presentation, (Pisa groups) May 4th, 2006 Hadron Calibration - Munich C.Roda
F.Paige/H1 Method - defaults Linearity,Resolution F.Paige/H1 Method - defaults ||<0.7 May 4th, 2006 Hadron Calibration - Munich C.Roda
ETfinal = ETcal x F(, Etcal) for = 0.5 bins 0 < || < 0.5 Etrec/ETMc ETfinal = ETcal x F(, Etcal) for = 0.5 bins This is all done calculated on dijet samples for different je algorithms. May 4th, 2006 Hadron Calibration - Munich C.Roda
Correction functions F(Et,eta) Given a sufficient fine binning in eta and Et it is clear that linearity is restored. No improve on the resolution is espected. The real test is to see if linearity is restored on a different sample. Check on SU3 sample: quark rather than gluon jets and Herwig instead of Pythia. May 4th, 2006 Hadron Calibration - Munich C.Roda
Before applying correction function After applying correction function May 4th, 2006 Hadron Calibration - Munich C.Roda
SU3 Cone R = 0.4 CaloTowerJets May 4th, 2006 Hadron Calibration - Munich C.Roda
SU3 Cone R = 0.4 CaloTowerJets May 4th, 2006 Hadron Calibration - Munich C.Roda
Comparison of different methods Pisa method F.Paige/H1 method A.Gupta The three calibration methods are in Athena. All parameters obtained on Rome samples. More results on Pisa calibration will be shown by Andrea in the next presentation. The framework to do the comparison in a consistent manner is in place. Final results on comparison by Barcellona. May 4th, 2006 Hadron Calibration - Munich C.Roda
Next steps Need to better understand the calibration method for the low energy (ET<30 GeV) jets. Understand degradation of energy resolution with respect to TDR Noise treatment: noise is suppressed using different tools. However a detailed study of noise suppression algorithms and performance is still missing. In the official production we have requested a few dijet events without noise. This can be used as the reference events to finalize this study. Weak point of all chi**2 minimization: the method to obtain the weights is not authomatic and need human iterations. Ambreesh has started to implement in Athena the framework to have this done in more authomatic way however the framework needs test from users. Detailed comparison of three schemes with standard macros. Robusteness of the method: apply to real data, CTB pions – see I.Vivarelli presentation for very preliminary results. Assess performance of calbration when pile-up is used Start assessing integration between jet and had calibration. May 4th, 2006 Hadron Calibration - Munich C.Roda
May 4th, 2006 Hadron Calibration - Munich C.Roda
May 4th, 2006 Hadron Calibration - Munich C.Roda
Comparing different schemes CaloTowerNoise EME 3 sigma asym EMB CaloTopoCluster Plots show how much negative enegy is left in jets after noise subtraction algorithm is applied in each calorimeter region. TileBarrel TileExB HadLar FW CaloTopoCluster < 3 sigma sym < CaloTowerNoise Check what are the conclustion on Ambreesh note Gap May 4th, 2006 Hadron Calibration - Munich C.Roda