Introduction
E N D
Presentation Transcript
Introduction • Miha Zgubič, summer student • Scintillating fibre tracker software • Analysis of performance of momentum reconstruction
What has been done? • Compare MC truth to reconstructed values (longitudinal and transverse momentum, pz&pt) – both PR and kalman • Call the width of Gaussian “resolution” Resolution plotted as a function of pz or pt. (histograms fitted separately for each MC momentum interval) noise, muonsand pions, kalman filter
Details • Lookup table between MC and recon side • Beam: • 10k spills at 200MeV • Emittance of 6.0 • Cut on reconstructed pz and pt at 500MeV • Kalman: • Algorithm 1: station 1 recon momentum used (better feel for what is going on) • Algorithm 2: recon momentum values averaged over the trackpoints (better resolutions results)
Results • Pattern recognition (mu plus, others similar) • Low statistics -> large error bars
Results • Noise on/off comparison, PR (mu plus)
Results • Kalman filter (mu plus, averaged)
Results • Kalman filter (mu plus, station 1)
Results • Kalman filter (mu plus, station 1)
Results • Compare kalman and pattern recognition (mu plus, averaged)
Results • Compare kalman and pattern recognition (mu plus, averaged, 400k spills, different emittance)
Results • Kalman filter (mu minus)
Results • Kalman filter (mu minus)
Conclusions • PR works fine • Noise has little impact on performance • Kalman as good as PR for pz– not better • Worse than PR for pt, and sometimes produces very large values of pt • Possibly a bug for negative particles?