Mirko Berretti - Giuseppe Latino (University of Siena & INFN-Pisa)

1. Mirko Berretti -Giuseppe Latino (University of Siena & INFN-Pisa) Track and Jet reconstruction in T2 “Forward Physics at LHC with TOTEM” Penn State University University Park, PA 04-28-2008 1

2. Highlights • T2 Track Reconstruction Preliminary Studies • T2 Trackingalgorithm :Linear Least Square Method, • based on T1 Tracking Algorithm (Fabrizio Ferro) • Detailed studies on BP effects generating p-and 0at differenth • Topological Jet Algorithms: Cone-like & Kt-like • Algorithms developed at charged particle level on Pythia di-jet • events (5.3 < hP1,P2< 6.5) , parton PT > 3 GeV, • no beam remnants contribution • Fake-Jet incidence studies on Single Diffractive events • Studies extended at track level 2

3. T1 10.5 m T2 ~14 m Elastic Detectors (Roman Pots):position of p scattered elastically at small angles Active area up 1-1.5 mm from beam: 5-10 rad RP1 (RP2) RP3 147 m (180 m) 220 m TOTEM Detectors: Setup in CMS Detectors on both sides of IP5 Inelastic Telescopes: reconstruction of tracks and interaction vertex  T1:3.1 << 4.7 T2: 5.3 <  < 6.5 T1: 18 - 90 mrad T2: 3 - 10 mrad CMS h = - log(tg(/2)) HF 3

4. Beam Pipe Effect: Secondary Particles Effect of beam pipe (I) : pion hadronic interaction  lots of secondary particles cone section at η = 5.53 Each point = <Tracks> per event firing 1000 - with: E = 50 GeV 0 < f < 2p h- = as shown in x axis 1 cone section at η = 4.9 T2 4

5. Importance of Z@Rmin Distribution A cut on track Z@ Rmin allows to remove a big fraction of secondaries: given the track eta, the curves below can be used to decide wether the track is from the primary vertex or not 2 2 hRECO hRECO 4 The distribution of tracks Z@Rmin depends on track h and particles energy 5

6. Particle Energy Distributions in T1-T2 h Region (Important for Future Analyses Strategies) For example: potential losses of low energy tracks reconstruction in di-jet events are less important than in SD events or minimum bias events 6

7. Importance of Z@Rmin Distribution Using the curve at E = 50 GeV in previous slide almost all secondaries are removed requiring DZ=Z@Rmin<3s, c2<1, hREC in T2 1000 - E = 50 GeVfor each point Efficiency = mean number of tracks per event with c2 < 1, Z@Rmin < 3s, 5.2 < hrec< 6.6 7

8. hGEN inside critical zones hGEN outside critical zones hGEN outside critical zones hGEN inside critical zones DF & Dh Resolution with B.P. & M.F.: 1000 p-, E = 50 GeV , 0 < f < 2 Reconstructed <F> depends on track energy because of M.F. effect 8

9. sDh(h) sDf(h) (deg)  &  Resolution with B.P. & M.F.: 1000 p- E = 50 GeV , 0 < f < 2 “Intrinsic”  resolution: <F>rec depends on track energy because of M.F. effect Each point obtained from a gaussian fit on a histogram like the ones shown in previous slide 9

10. p0 Contribution Effect of beam pipe (II) :  (from 0) e.m.interaction  lots of secondary particles 10K 0 generated with: E = 50 GeV 5.2 <  < 6.5 , 0 <  < 2 h(track) h(track) Potential effect of secondaries from CMS HF calorimeter to be investigated h(p0) h(p0) 10

11. We want to study the possibility to develop a jet algorithm by using only charged particles reconstructed with T1/T2 • To do this, we have to exchange the (usual) PT information with the particle density information in the  plane Topological Jet Algorithms • First step: the algorithm has been developed/tested by using Pythia di-jets events where both partons are required to fall in the T2 region (5.3 <  < 6.5). 11

12. Construction of ordered list (in particle number) of seed avoiding to include the neighbouring cells of the major seeds • Build h-f grid • Fill the cells with the number of particles inside them multiplicity and shape cuts From the seeds start the search of stable cones (1st Jet list)‏ 2nd Jet List Merging - Splitting Mid Point Final Jet List We have developed a “cone” algorithm with some improvement, based on MidPoint algorithm Cone jet algorithm: implementation Jets ordered according to Ntrk 12

13. Summary of current Cone Jet Algorithm Parameters (to be optimized according to specific analysis needs) • Cone search-radius (R) 0.7 • Grid Size 0.4 • Percentage of particles shared between two jets for merging-splitting:50 % of the less populated jet If the number of shared particles is less than half multiplicity of the less populated jets, the two jets are split otherwise are merged. • Jet Charged Multiplicity cut : 2 • Jet Shape request: none 13

14. A “Kt-like” Jet Algorithm also Developed: • Features • The Kt algorithm is in the Cambridge/Aachen implementation • and it is based on “FastJet Algorithm” arXiv:hep-ph/0512210. • It does not use particle transverse momentum information, • only distance Rij: • The merging criteria for each couple of particles of our algorithm • (particles topologically clusterized to make a jet) gives similar results of • a “traditional” kt algorithm (that also uses transverse momentum • information). 14

15. Kt Jet Algorithm: Implementation Final Jet List multiplicity and shape cuts yes For each particle i, find the particle j with minimum Rij. no empty list Build a list of all the particles (proto-jets) Remove i from the jet list Rij > R update the jet list i is a Jet yes Merge i,j ina single jet no Jets ordered according to Ntrk Present algorithm setting: • R = 1 • No shape cut • Trk multiplicity = 2 15

16. Kt & Cone Jet Algorithms (Track and Particle Level Comparison) • First test with 1000 di-jet events (5.0 < hp1,p2< 6.5, PTp > 3 GeV) at track level • The configuration files shown below set the algorithm • working parameters. NOTE: they are not yet optimized in • order to maximize reconstruction efficiency and background rejection Kt configuration file module Ktjetfinder = Ktjf { double T2minEta= 5.1 double T2maxEta= 6.6 double T1minEta= 3.2 double T1maxEta= 4.7 double Range1m= 5.35 double Range1M= 6.15 double NumSigZ= 3.0 double Chicut=1.0 double GlobalRkt= 1.0 double Beamtreshdist= 1.0 double Jetradiusforshape= 0.5 double ThresholdShape= 0.0 double Shapepar=0.5 uint32 ThresholdMult= 2 } Cone configuration file module Conejetfinder = Conejf { double T2minEta= 5.0 double T2maxEta= 6.6 double T1minEta= 3.2 double T1maxEta= 4.7 double Chicut=1.0 double Range1m=5.35 double Range1M=6.15 double NumSigZ=3.0 double minetagrid=2.5 double maxetagrid=7.0 double minphigrid=0.0 double maxphigrid= 6.283185 double cellphi=0.4 double celleta =0.4 double SearchRadius =0.7 double ThresholdShape= 0.0 double Shapepar=0.5 uint32 ThresholdMult= 2 } Parameters for “track-quality” selection “Internal” working algorithm parameters Parameters for “jet-quality” selection 16

17. Distance Jet-Nearest Parton (Kt) 17 Particle Level Track Level

18. Di-jet Reconstruction Efficiency NOTE: efficiency ~ 50% (at PL) if no cut on T1/T2  acceptance Cone Kt Track Level Particle Level 18

19. Fake Jet Reconstruction Efficiency in SD Events Cone Kt Track Level Particle Level 19

20. Summary & Conclusions T2 Track Reconstruction (Preliminary Results) • Track reconstruction with typical efficiency well above 80% for charged  • First studies on track selection criteria performed • Secondary tracks from B.P. interaction efficiently removed • with standard cut criteria • Cone-like and Kt-like jet algorithms developed and tested • Both algorithms show similar behaviour with jet reconstruction • efficiency for di-jet events around 25-30% and fake rate in SD • events around 5% (preliminary results) • Possibility of improvements in jet selection (jet parameter optimization) • and in background rejection (for instance considering angular • correlation or combining energy information from Castor….) Jet Reconstruction 20

21. BACKUP SLIDES

22. T2Road : “histogramming method” • Build an integer matrix in R-phi plane • (at the moment cell size: [2.0 mm]x[3 deg]) • Increment the matrix cell value according to • the number of hits falling inside • Build the roads clusterizing the hits around the • “seed cells” (=cells with #hits > threshold) • at the moment road size: [6.0 mm]x[9 deg] T2Track : LeastLSQ on T2Road (copied & adapted from T1 Tracker) h, f R min Z at Rmin Reducedc2 1Road 1 Track Minimum distance Reconstructed Track-Z Axis (Z Axis =beam pipe Axis) Straight track in 3D are found looking for its projection in YZ and XZ plane Main Track Parameters B1

23. The Beam pipe Effect : Reconstructed Track h (expected interaction with the h=5.56b.p.cone section) Track hfor1000 p-with E=50 GeV fired at h=5.9 (left) and5.59 (right) B2

24. The Beam pipe Effect : Reconstructed Track c2 Track c2for1000 p- with E=50 GeV fired at h=5.9(left) and5.59 (right) (expected interaction with the h=5.56 b.p.cone section)‏ Suggested cut: Tracks belonging to primary particles have c2<1 B3

25. Detailed Z@ Rmin distribution studies inside “critical zones” Close to detector edge (lowh) h close to b.p. cone section Close to detector edge (highh) h outside critical zones B4

26. To use a reasonable grid size and search radius we have looked at the distribution of charged particles around the outgoing parton B5

27. Find the best cell size (all cuts activated)‏ Other features(CONE) Check the algorithm stability shifting the grid Angular resolution B6

28. Mid - Point For every pair of jets with distance < 2R , assign a new seed in the midpoint. This method can reduce “infrared-like” sensitivity of our algorithm due to missing s in the core of particle jetand “collinear-like” sensitivity due to the choice on grid size/position. Infrared-like sensitivity With Mid-Point we can merge two different jets when soft radiation/lost particle is present between them Collinear-like sensitivity Algorithm start from central most populated cell and it will probably find 3 jets or only 1 jet (if the jet are too close in the preclustering). Without Mid-Point, algorithm starts from the right-most cell and it will find only 2 jets. With Mid Point it will find 3 jets or 1 jet B7

29. We need some jet “quality” cuts Multiplicity (minimum number of particles forming a jet) >= 2 Mult. cut Shape # charged particles inside cone 0.4 Jet charged shape: # charged particles inside cone 0.7 > 0,5 Shape cut ? B8

30. hdistribution (all Jets) Cone Kt Track Level Particle Level B9

31. fdistribution (all Jets) Cone Kt B10 Track Level Particle Level

32. Dfdistribution for event with 2 Jet reconstructed Cone Kt B11 Track Level Particle Level

33. MinimumJ1-J2 fseparationfor events with 2 Jet reconstructed Cone Kt B12 Track Level Particle Level

34. Distance Jet-Nearest Parton (Cone) Track Level Particle Level B13

35. Minimum f Separation , Fake Jet in SD Events B14

36. f Separation , Fake Jet in SD Events B15

37. Track Multiplicity in Jet B16

38. Energy of Charged Particles Inside Jet in T2 B17