1. Contents
First section: pion and proton misidentification probabilities as Loose or Tight Muons. Measurements using Jet-triggered data (from run). Dependence on hadron pT, eta, production radius, event pileup, local track density and comparison with MC results.
Second section: performance of the RPC Muon identification. Mis-ID from pions and kaons measured on a single-muon trigger dataset (2012 run). Dependence on hadron pT and eta for several different ID algorithms including the standard Loose and Tight Muon and a few Tracker Muon definitions.

2. Muon ID performance: pion and proton misidentification probabilities
Approved at PPD General Meeting, 16th April 2014

3. Contents
Pion and Proton Misidentification probabilities as muons identified by the Loose or Tight Muon selection
These results complete (and supersede) the pion mis-ID results included in DP2013/024

4. Muon ID: definitions Loose Muon Tight Muon
Particle identified as a muon by the Particle-Flow event reconstruction
Discard muon candidates which are reconstructed only on the muon detectors, without hits reconstructed on the inner tracking system.
Tight Muon
Global muon track, including hits in the inner tracking system and in the muon detectors
At least one muon chamber hit used in the global track fit
Global track fit c2/ndf < 10
Muon Inner track extrapolation matched to segments in at least two muon stations (by the Tracker Muon algorithm)
Hits on more than 5 layers of the inner tracking system
At least one pixel hit.
Cuts on the impact parameters in the transverse and longitudinal planes w.r.t. the primary vertex of the event :|dxy| < 0.2 cm, |dz| < 0.5 cm
General description of the reconstruction and identification algorithms
in: JINST 7 (2012) P Here some different cuts introduced for the 2012 run.

5. K0s-> p+p- selection
Data from Jet Trigger selections
Clean Track and K0s selection
K0s Transverse flight length Lxy< 4 cm (decay within the beam pipe)
Track pT > 4 GeV (as for pT > 4 GeV the muon acceptance is good also in the barrel region)
One entry per selected track at the invariant mass of the pair (both tracks can be selected)
Signal fitted with a double Gaussian, background with 1st order polynomial

6. Pion Misidentification from K0s-> p+p- selection
K0s Transverse flight length Lxy< 4 cm (decay within the beam pipe)
Pion pT > 4 GeV, |h| < 2.4
p mis-ID (Loose m) : 2.16 ± 0.03 x 10-3
p mis-ID (Tight m) : 1.34 ± 0.02 x 10-3
Statistical errors only (from the fit)

7. L -> p p selection Data from Jet Trigger selections
Clean Track and L selection
Highest momentum track of the pair assumed to be a proton
L Transverse flight length Lxy< 4 cm (decay within the beam pipe)
Proton candidate track pT > 4 GeV (as for pT > 4 GeV the muon acceptance is good also in the barrel region)
Signal fitted with a double Gaussian, background with 1st order polynomial

8. Proton Misidentification from L -> p p selection
L Transverse flight length Lxy< 4 cm (decay within the beam pipe)
Proton pT > 4 GeV, |h| < 2.4
p mis-ID (Loose m) : 0.58 ± 0.05 x 10-3
p mis-ID (Tight m) : 0.16 ± 0.03 x 10-3
Statistical errors only (from the fit)

9. Mis-ID probability vs pion production Radius
Mis-ID probability as a function of the transverse distance from the beamspot to the production point of the pion track. This is determined by the K0s decay vertex (transverse flight length).
Loose Mis-ID is almost flat. Tight Mis-ID falls quickly due to the request of at least one pixel hit and the impact parameter cuts.
All the following results are obtained with a cut on the Transverse flight length of the decaying K0s, Lxy<4 cm (decays within the beam pipe), to be applicable to prompt hadrons from collisions.

10. Mis-ID probability vs proton production Radius
Mis-ID probability as a function of the transverse distance from the beamspot to the production point of the proton track. This is determined by the Lambda decay vertex (transverse flight length).
Loose Mis-ID is almost flat. Tight Mis-ID falls quickly due to the request of at least one pixel hit.
All the following results are obtained with a cut on the Transverse flight length of the decaying Lambda, Lxy<4 cm (decays within the beam pipe), to be applicable to prompt hadrons from collisions.

11. Mis-ID probability vs hadron production Radius
Mis-ID probability as a function of the transverse distance from the beamspot to the production point of the hadron track. This is determined by the K0s/Lambda decay vertex (transverse flight length).
Loose Mis-ID is almost flat. Tight Mis-ID falls quickly due to the request of at least one pixel hit and the impact parameter cuts.
All the following results are obtained with a cut on the Transverse flight length of the decaying K0s/Lambda, Lxy<4 cm (decays within the beam pipe), to be applicable to prompt hadrons from collisions.

12. Pion Mis-ID probability vs h
Good data-MC agreement
Dominant component: pion decays in flight

13. Proton Mis-ID probability vs h
Good data-MC agreement
Components: punch-through and random matching

14. Pion/Proton Mis-ID probability vs h
Good data-MC agreement
Dominant component: decays in flight (for pions), punch-through and random matching (for protons)

15. Pion Mis-ID probability vs pT
Good data-MC agreement
Dominant component: pion decays in flight

16. Proton Mis-ID probability vs pT
Good data-MC agreement
Components: punch-through and random matching

17. Pion/Proton Mis-ID probability vs pT
Good data-MC agreement
Dominant component: decays in flight (for pions), punch-through and random matching (for protons)

18. Pion Mis-ID probability: PileUp dependence
No significant dependence observed w.r.t. the number of reconstructed primary vertices. Good Data-MC agreement.

19. Proton Mis-ID probability: PileUp dependence
No significant dependence observed w.r.t. the number of reconstructed primary vertices. Good Data-MC agreement.

20. Pion/Proton Mis-ID probability: PileUp dependence
No significant dependence observed w.r.t. the number of reconstructed primary vertices. Good Data-MC agreement.

21. Pion Mis-ID probability vs Track Density
Pion Mis-ID probability as a function of the number of tracks contained in a cone with size DR=0.2 centered on the probed pion track.
No significant dependence is observed, consistent with the dominant source of mis-ID being decay in flight of the probed pion.

22. Proton Mis-ID probability vs Track Density
Proton Mis-ID probability as a function of the number of tracks contained in a cone with size DR=0.2 centered on the probed proton track.
Significant increase of the Mis-ID probability with the number of close tracks is observed from Data, consistent with the effect of random matching of the proton track with an independent close-by muon, in addition to the punch-through component.
This is expected from MC truth, although the MC sample has not enough statistics to show a significant trend.

23. RPC Muon ID performance and other pion and kaon Mis-Id measurements
Approved at PPD General Meeting, 19th February 2014

24. RPC Muon ID RPC Muon identification: RPC Muon working points:
Extrapolate the inner track to RPC detector layers
Match the extrapolated trajectory to RPC reconstructed hits by local position:
∆x ≤ 20cm or pull=∆x/σ(∆x) ≤ 4
RPC Muon working points:
RPCMuLoose:
Number of matched RPC layers ≥ 2
RPCMuMedium:
Number of matched RPC layers ≥ 2 Number of matched RPC stations ≥ 2
RPCMuTight: Number of matched RPC layers ≥ 3 Number of matched RPC stations ≥ 2

25. Tracker Muon ID: definitions
Tracker Muon ID (inside-out):
inner tracker tracks are extrapolated and matched to segments in DT or CSC chambers. A matching is found when the distance between the extrapolated track and a muon segment is less than 3 cm or the pull is less than 4, in the local best-measured position coordinate.
The matching of inner tracks and muon segments is arbitrated by choosing the best geometrical matching, to resolve ambiguities and remove duplicates (TMArbitrated).
TMOneStationLoose : at least one segment matching in local-x position within 3cm or pull≤3.
TMOneStationTight: at least one segment matching in both local position coordinates 
 within 3 cm or pull≤3.
General description of the reconstruction and identification algorithms in: JINST 7 (2012) P10002

26. RPC Muon Tag and Probe efficiencies
Efficiency of high pT prompt muon in |η|<1.6 (RPC coverage) with the RPCMuon definitions determined with the Tag-and-Probe method (example fit shown in the plots for passing and failing probes)
Data selected from single muon triggers
Tags : Tight muons matched to the trigger object
Probes : general tracks in the inner tracker with at least 10 valid hits, Impact parameter cuts in both the transverse and the longitudinal plane w.r.t. the primary vertex: |Dxy|<0.2 cm, |Dz|<0.5 cm
Fit with Voigtian(signal) + 2nd order polynomial (background)

27. RPC Muon efficiency vs pT and |η|
Efficiency of RPC Muon ID for general tracks with basic quality cuts
(number of valid tracker hits ≥ 10, impact parameter |dxy| < 0.2cm, |dz| < 0.5cm). Statistical errors only (from the fit)
Efficiency drops reflect the RPC detector layout:
Barrel : cracks between wheels
Endcap : only 3 RPC layers

28. K0s-> p+p- and f -> K+K- selections
Data from single muon triggers, (12fb-1). Independent dataset and selection w.r.t. the other Mis-ID results in the first section of this note
Veto cone(∆R=0.5) around the triggering muon to avoid residual biases
Clean track and decaying hadron selection: K0s ->pp, f->KK
K0s transverse flight length |Lxy| < 4cm 
 (decay within the beam pipe)
Track pT > 4GeV, |η| < 1.6
Other K mis-ID results, obtained from an independent selection of B->J/ψK are reported in a previous note (DP-2013/024)
Misidentified hadron candidate: a candidate hadron track that is also identified as a muon
Simultaneous fit of the mass distribution for track candidates with and without Muon (mis)ID

29. Pion Misidentification from K0s-> p+p- selection
Signal : Voigtian
Background : linear
Simultaneous fit to mass distributions for misID and non-misID pion probe tracks.

30. Kaon Misidentification from f -> K+K- selection
Signal : Voigtian
Background : 2nd order polynomial
Simultaneous fit to mass distributions for misID and non-misID kaon probe tracks.

31. π/K Mis-ID probabilities
Pion
Kaon
RPCMuLoose
0.364±0.073
0.579±0.045
RPCMuMedium
0.261±0.069
0.452±0.041
RPCMuTight
0.147±0.063
0.335±0.036
TMArbitrated
0.564±0.062
0.868±0.045
TMOneStationLoose
0.448±0.060
0.732±0.039
TMOneStationTight
0.370±0.057
0.574±0.036
Loose muon
0.171±0.050
0.405±0.030
Tight muon
0.092±0.046
0.238±0.025
Mis-ID probabilities from π± and K± for various muon identification criteria, comprising Tracker Muon and RPC Muon selections and the standard Loose and Tight Muon ID. Errors are statistical only.
The Loose and Tight mis-ID probabilities are in good agreement with the other independent results on pion mis-ID presented in the first section of this DP note, and with the results on kaon mis-ID obtained from B->J/ψK, reported in DP-2013/024

32. Pion Mis-ID probability vs pT and |η|
Muon mis-ID probability from pions for different selections: Tracker Muon Arbitrated, Loose Muon, Tight Muon and three RPC Muon selections as a function of pion transverse momentum (left) and pseudorapidity (right). Statistical errors only (from the fit)

33. Kaon Mis-ID probability vs pT and |η|
Muon mis-ID probability from Kaons for different selections: Tracker Muon Arbitrated, Loose Muon, Tight Muon and three RPC Muon selections as a function of kaon transverse momentum (left) and pseudorapidity (right). Statistical errors only (from the fit)