Primary vertex reconstruction with the SPD E Crescio

Primary vertex reconstruction with the SPD E. Crescio, M. Masera, F. Prino INFN e Università di Torino Offline week, reconstruction meeting – October 2 nd 2006 1

Vertexers Ali. ITSVertexer. Z: uses information from SPD c Goals: high efficiency, constrain the tracking c Before tracking Ali. ITSVertexer 3 D: extension of Ali. ITSVertexer. Z to x and y c New class presently under test c Before tracking (needs just SPD tracklets) Ali. ITSVertexer. Ions: uses information from SPD c measurement of Xv, Yv, Zv through fit procedures c Not working for low multiplicities Ali. Vertexer. Tracks: uses tracks c Goals: accuracy and good error determination c Measurement of Xv, Yv, Zv c After recostruction in the barrel 2

Events used in this study Event generation c Ali. Root v 4 -04 -Release c pp collisions (k. Py. Mb) c Vertex smearing on x, y (50 m) and z (5. 3 cm) FIRST SET: c 9800 events with beam centered in 0, 0 SECOND SET: c 9400 events with beam centered in (500 m, 0) ONGOING GENERATION: c events with 5 mm and 1 cm of beam offset Vertexer. Z and Vertexer 3 D performance studied in 6 bins of Ntracklets in SPD (Ali. Multiplicity: : Get. Number. Of. Tracklets) c Efficiency = ratio between events with reconstructed vertex and total number of events c Residual distributions = distribution of zmeasured-ztrue ü Estract resolution, pulls and bias 3

Ali. ITSVertexer. Z – the method Layer 2 Build “tracklets” from SPD Rec. Points c associate each point of the two central SPD modules of layer 1 to all the points of layer 2 within a window Δφ <0. 01 rad Layer 1 Beam axis Calculate Zi = Z of closest approach of tracklet and nominal beam axis. Fill histograms of Zi with 3 different bin sizes c 5 m (fine: for the actual measure) c 100 m (coarse: to single out the peak corresponding to the actual vertex) c 300 m (very coarse: used when the number of tracklets is very low) Define a z window (200 or 600 m wide) around the “peak” of the coarse histogram. c Window adjusted to be symmetric around the centroid of the peak Calculate on the “fine” histogram: c Zv = average of the Zi of the tracklets in the window c Dispersion parameter (= sqrt of the sample variance) 4

Number of contributors Get. NContributors() to check the “quality” of the vertex c>0 Vertex OK c 0 error in the vertex finding procedure c-1 no tracklets ü recoverable with iterative procedure c-2 no recpoints in SPD 5

Performance: efficiency Black = beam in nominal position Red = beam offset 500 m – assumed unknown Blue = beam offset 500 m – assuming to know the beam position Optimized with an iterative procedure to reduce the fraction of events with no tracklets (NContributors=-1) 6 c. Committed on June 1 st (Ali. ITSVertexer. Z Rev 1. 11)

Performance: resolution and bias Black = beam in nominal position Red = beam offset 500 m – assumed unknown Blue = beam offset 500 m – assuming to know the beam position Bias reduced (from ≈20 to ≈5 m) since Rev. 1. 14 c Committed in HEAD, but not in v 4 -04 -Release Good performance for beam offset of 500 m if the offset is known 7

Peformance: pulls From distributions of Δz/Error cerror on the zmeasured as given by Ali. ITSVertexer. Z c. RMS of pull distributions should be 1 for gaussian estimators cto be checked at higher multiplicities 8

Vertexer 3 D – the method Pairs of Rec. Points on layer 1 and layer 2 taken as candidate tracklets c selection cut: Δφ <0. 01 rad (i. e. straight lines) Tracklet pairs are combined and selected according to: c small DCA c Intersection close to beam axis c Intersection in the diamond region tuning of these cuts presently under study Use tracklets as “straight-line-tracks” and apply the same vertex finder algorithms used with ESD tracks (Ali. Vertexer. Tracks) c Background tracklets must be removed as much as possible before applying the algorithm Layer 2 Layer 1 Beam axis 9

Vertexer 3 D - efficiency Overall efficiency ≈ 65% c. Efficiency improvements under study 10

Vertexer 3 D – bias and resolution Bias not present c Hint that the problem with the Vertexer. Z is in the vertexing algorithm and not in SPD geometry Good resolution! 11

Vertexer 3 D and beam offset Black = beam in nominal position Red = beam offset 500 m – assumed unknown Blue = beam offset 500 m – assuming to know the beam position Large bias on X (the coordinate where the beam has the offset) c ≈ 1/2 of the beam offset c Due to Δφ< 0. 01 rad selection cut (select straight lines pointing to the beam axis given as input) 12

Correcting the 3 D bias (I) First (very naïve) idea: iterative procedure c. For each event use x, y vertex positions found in iteration i as nominal beam positions for iteration i+1 c. Works sufficiently well for high multiplicity (i. e. good resolution) c. Price to pay: loss of resolution ü due to events with worse vertex determination at the 1 st iteration? c. May be recovered using nominal positions averaged over several high multiplicity events 13

Correcting the 3 D bias (II) Second (very naïve) idea: enlarge Δφ cut c From trigonometry: 500 m of offset give a max. Δφ of 0. 011 rad. c Does not allow to completely cancel the bias ü no improvement when enlarging Δφ from 0. 03 to 0. 05 c Price to pay: loss of resolution c Main drawback: requires huge enlargement in case of larger offsets Third idea (presently under development): change the tracklet selection c use a DCA cut between pairs of tracklets crossing the beam pipe instead of the Δφ cut 14

Pile-up Expected interaction rate = 2× 105 Hz at a luminosity of 5× 1030 cm-2 s-2 c 1 interaction every 200 bunch crossings Foreseen SPD strobe duration is 200 ns c 8 bunch crossings (0. 04 interactions) ü All events in the strobe are overlapped even if not belonging to the same bunch-cross Caveat: high- multiplicity triggers will select piled-up events First check on Ali. ITSVertexer. Z in the case of pile-up c “Manual merging” of recpoints with an “ad hoc” macro Results: c Vertices with distances >600 μm: found the vertex of the event with higher multiplicity c Vertices with distances <600 μm: found an intermediate value of z c Under study: check if the vertexer can be used to “detect” the pile-up, searching for two peaks (possible in the case of well separated peaks) Study to be performed also on the Vertexer 3 D 15