Presentation on theme: "Fast b tagging at L2 B tagging meeting XX-XX-04 Sascha Caron (NIKHEF) … using the Silicon Track Trigger (STT) Methods to do a very fast b tagging Some."— Presentation transcript:
Fast b tagging at L2 B tagging meeting XX-XX-04 Sascha Caron (NIKHEF) … using the Silicon Track Trigger (STT) Methods to do a very fast b tagging Some first results and a proposal for a L2 algorithm
Sascha Caron (NIKHEF) 2 Silicon Track Trigger The STT is the
Sascha Caron (NIKHEF) 3 STT status hardware installed, working and STT info is implemented in v13 test trigger list STT simulator agrees well with hardware STT output correlates well with D0 reconstruction Time to think about using the STT improved L2 track information for b tagging
Sascha Caron (NIKHEF) 4Trigsim STT information for Monte Carlo events gained using a modified trigsim t04.05.00 (with p17.01.00 STT routines) and trigsimcert to get the root tuple Events used are the usual sets for b tagging Certification: QCD: pt>40 GeV file with 10000 events (id 8791) Z-> bb: 10000 events (id 8845) Z-> cc: 1000 events (id Top-> all jets Z-> bb with 12 multiple interactions : 1000 events
Sascha Caron (NIKHEF) 5 Comparison STT simulation code and modified trigsim/trigsimcert Agreement in the used variables (chi2, impact parameter significance)
Sascha Caron (NIKHEF) 7 ELIP method Pdf derived using Kernel estimation Each sample point is smeared by pulling 10000 sample points out of a Gaussian with a width optimized as a function of S Probability for a track to come from the vertex is just the integral of the pdf from S to inf.
Sascha Caron (NIKHEF) 8 LM (likelihood method) Derive probability density function of the track significance S for signal (Z) and background (QCD) events Use all good tracks with a scaled chi2<5 Idea : Store R(S)=pdf Signal (S)/ pdf background (S) in a lookup table for S (256 entries) Loop over all good tracks i and derive the product of R Derive a Likelihood as a descriminant:
Sascha Caron (NIKHEF) 9 LM: Kernel estimation pdf using kernel estimation method with Gaussians
Sascha Caron (NIKHEF) 10 MULM method Significance is heavily dependent on the goodness of the track fit Goodness the track fit given by scaled chi2 (less pt dependent) Idea: Include chi2 information in discriminator by using 2d p(S,chi2) pdfs This degrades tracks with large chi2 while still Using the full information provided by the STT.
Sascha Caron (NIKHEF) 12 Summary High energy ep collisions at HERA are a unique testing ground for the SM Various searches for new physics are performed (model independent and dedicated) Some interesting events are found … and HERA 2 has just started !