1. Muon alignment with Cosmics: Real and Monte Carlo data S.Vecchi, S.Pozzi INFN Ferrara 37th Software Week CERN 15-19 June 2009

2. June 17th, 2009 37th LHCb Software week 2 Outline Reminder on the Alignment of real Cosmic data Reminder on the Alignment of real Cosmic data Analysis of Montecarlo Cosmics data Analysis of Montecarlo Cosmics data Conclusions Conclusions

3. June 17th, 2009 37th LHCb Software week 3 Real data sample Local alignment of Muon Half stations (Sep-Oct 2008 data) Local alignment of Muon Half stations (Sep-Oct 2008 data) — Trigger ECAL U Muon — Multiple BX acquisition ±7 TAE — Noisy channels masked on & off-line — Nnet-track finding (Xtalk hits) with at least 3 station hit — — 130k tracks useful for alignment

4. June 17th, 2009 37th LHCb Software week 4 Alignment results (translations x,y,Rz) Kalman-fit based alignment (iterative) Kalman-fit based alignment (iterative) — fixed 2 stations (M3&M5) as reference From S.Pozzi’s talk at the 3rd Detector Alignment Workshop 15/06/09 (link)link Compatible results if we fix two ≠ stations

5. June 17th, 2009 37th LHCb Software week 5 Alignment (translations & rotations) Kalman-fit based alignment (iterative) Kalman-fit based alignment (iterative) — Cosmics data have some sensitivity to the “z” position due to the larger track slopes — Significative up to 10mm misalignments found in z BUT inconsistent picture found if different set of stations are fixed — Rotations of the whole half-station around x/y have been tested — Significative up to 6mrad misalignments found in Ry BUT inconsistent picture found if different set of stations are fixed see S.Pozzi’s talk at the 52th LHCb week 26/05/09 (link)link … need a Monte Carlo to validate these results and better understand possible systematic effects (Cosmics highly non projective, geometric acceptance)

6. June 17th, 2009 37th LHCb Software week 6 Monte Carlo for Cosmics Simulation: based on Gen/ParticleGuns/Cosmics.cpp(h) Simulation: based on Gen/ParticleGuns/Cosmics.cpp(h) — Generates from a xz plane 6m on top of LHCb  ± in order to have a uniform coverage on a vertical xy plane. — Some minor changes applied to have the right proportions of “horizontal” cosmics (enlarge the sizeOfWorldVolume) Generated 3M events with  Rays on in MuonFilters Generated 3M events with  Rays on in MuonFilters — 1M evts on M2 / M3/ M4 surfaces (t o ~45-50 ns) Digitized with TAE ±4, Xtalk, Electronic Noise Digitized with TAE ±4, Xtalk, Electronic Noise NNet-based track reconstruction (like real data) No M1 NNet-based track reconstruction (like real data) No M1 Crucial point for the alignment: provide the same spatial & angular distributions of the Real data

7. June 17th, 2009 37th LHCb Software week 7 Some comparison Track requirements: Track requirements: — hit at least 3 stations / 4 (Nnet) — match the CALO acceptance Trigger efficiency? Real MC BigBox MC original #single hit/# hits -real -MCarlo

8. June 17th, 2009 37th LHCb Software week 8 Some comparison: vs x vs x similar distribution vs x similar distribution — depends on the initial distributions & on correlations Real MonteCarlo R1 R2 R3 R4

9. June 17th, 2009 37th LHCb Software week 9 Some comparison: vs y vs y similar distribution vs y similar distribution — Strong modulation due to chamber to chamber gaps & non projectivity — depends on the initial distributions & on correlations Real MonteCarlo R1 R2 R3 R4

10. June 17th, 2009 37th LHCb Software week 10 Residual distributions: hit - MChit Requirement: track passing through M3&M5&Calo Requirement: track passing through M3&M5&Calo — No major deviation from box distribution (Xtalk x) M4M3M5M2 …..but…… R1 R2 R3 R4

11. June 17th, 2009 37th LHCb Software week 11 vs X dependence vs X dependence Requirement: track passing through M3&M5&Calo Requirement: track passing through M3&M5&Calo Error bar =  res /sqrt(N) R1 R2 R3 R4 Non uniform distribution of the MChit of the selected tracks + large pad size + 1/4 pad sizepad size 1/4 (8) column effect?

12. June 17th, 2009 37th LHCb Software week 12 vs Y dependence vs Y dependence Requirement: track passing through M3&M5&Calo Requirement: track passing through M3&M5&Calo R1 R2 R3 R4 Non uniform distribution of the Mchit Due to the selection & GAPS beween chambers

13. June 17th, 2009 37th LHCb Software week 13 Mean Residual distributions: M3&M5 fixed(X) Sampling of the measurement + non uniformity of the MChit distribution within a pad Sampling of the measurement & of the track parameters Would be like this if we had a very high spatial resolution detector R1 R2 R3 R4

14. June 17th, 2009 37th LHCb Software week 14 Mean Residual distributions: M3&M5 fixed (Y) Sampling of the measurement + non uniformity of the MChit within a pad (GAPs) Sampling of the measurement & of the track parameters R1 R2 R3 R4

15. June 17th, 2009 37th LHCb Software week 15 Comparison with real cosmics (after Alignment) M3&M5 fixed Mcarlo X Mcarlo Y Real X Real Y NB: Mcarlo statistics ~0.56 Real Error bar =  res /sqrt(N) Very similar distributions 75k tracks130k tracks R1 R2 R3 R4

16. June 17th, 2009 37th LHCb Software week 16 Comparison with real cosmics (after Alignment) M2&M5 fixed Mcarlo X Mcarlo Y Real X Real Y Very similar distributions 60k tracks96k tracks R1 R2 R3 R4

17. June 17th, 2009 37th LHCb Software week 17 Alignment results (translations x,y) Kalman-fit based alignment (iterative) fixed 2 stations (M3&M5) Kalman-fit based alignment (iterative) fixed 2 stations (M3&M5) Compatible results if we fix two ≠ stations Adding translations in z or rotation x,y we still get controversial results by fixing different pairs of stations (sampling?) Adding translations in z or rotation x,y we still get controversial results by fixing different pairs of stations (sampling?) — Tz: some significative up to 8mm misalignments found in z — Rx, Ry: some significative up to 1mrad misalignments found in Ry 40k tracks

18. June 17th, 2009 37th LHCb Software week 18 Further steps: align single chambers? Mean residuals X (fit of all the measurement) Requirements: at least 3 stations hit (any) Non uniformity & large pad size effects shouldn’t be mis-interpreted ad misalignments of the chambers…… Inner regions should be OK. R1 R2 R3 R4

19. June 17th, 2009 37th LHCb Software week 19 Further steps: align single chambers? Mean residuals Y (fit of all the measurement) Requirements: at least 3 stations hit (any) geometric acceptance due to chambers @ different z (4 columns) ==> different gaps size (non projective muons) A NIGHTMARE..!!!! R1 R2 R3 R4

20. June 17th, 2009 37th LHCb Software week 20 Conclusion From the analysis of Real Cosmics (2008) 130k tracks From the analysis of Real Cosmics (2008) 130k tracks — Misalignments along X A-side (shearing + M5 out 7mm) incompatible with the survey measurements; Y and Rz compatible with 0 — If more d.o.f. are used (Rx, Ry, Tz) we get controversial results if we fix different pairs of stations as reference ===> Systematics? Sampling? Generated Montecarlo Cosmics 90k tracks Generated Montecarlo Cosmics 90k tracks — Same track selection ==> Similar angular & space distributions — The track selection makes the hit distribution non uniform (geometrical acceptance + gaps between chambers) — Border effects on ResX vs X M4 (large pad size) — Jumps in the ResY vs Y distribution due to sampling — Comparison with Real (aligned) Cosmics show a reasonable agreement — Alignment of half stations is compatible with the input null displacements (Tx, Ty, Rz) — If more d.o.f. are used (Rx, Ry, Tz) alalogies & differences (*) — The alignment of single chambers seems very difficult with cosmics (Y)