1. Converted photons efficiency
Edwige
LAPP 18/12/2014

2. Introduction
The fraction of reconstructed converted photons depends on
The detector material
The reconstruction efficiency
In this study: focus on conversions from Downstream tracks
Downstream tracking
Depends on Velo tracks multiplicity
LHCb-INT : Reconstruction efficiency of DD K0s mesons (P.Owen, T.Blake)
Tracking meeting 6/11/2014: Data driven reconstruction efficiency of K0S (Eluned Smith)
Material description:
LHCb-INT : Study of VELO material description with hadronic interactions
LHCb note : Estimation of the material budget of the LHCb detector (M.Needham, T. Ruf)
Problem of SPD multiplicity: are the secondary interactions not well modeled?

3. Downstream tracking LHCb-INT2013-051 :
Ratio of DD/LL Ks depends on nVELO tracks:
Due to Dwonstream tracks wrongly upgraded to Long
Downstream efficiency measured with KS with
reweighting according to nVeloTracks
 Data/MC~0.95 (raw) – Data/MC=1 (with reweighting)
!! KS vertexing is responsible for most
of Data/MC disagreement !!

4. K0S reconstruction efficiency
Tracking meeting 6/11/2014: Data driven reconstruction efficiency of K0S (Eluned Smith)
10<P<20, 2.0<  <3.2
20<P<40, 3.2<  <5.0
Ks vertexing: D/MC~
0.7

5. Material before magnet
LHCb note :
Uses DC06 description
Update in LHCb note
No big changes
Material up to z=270cm
Material up to z=270cm
DC06
July 2008

6. Selection All: acceptance cuts , SPD multiplicity 2.25<  <4.5
Remove most inner region for cleaner sample:
(abs(gamma_PX/gamma_PZ)> || abs(gamma_PY/gamma_PZ)>0.025)
Remove plane y=0 and outer edge to remove electron background:
abs(gamma_PY/gamma_PE)> &&
abs(gamma_PY/gamma_PE)<0.2 && abs(gamma_PX/gamma_PE)<0.2
nSPDhits<600
Conversions:
Use here only Downstream pairs
Standard ee reconstruction and selection
combDLLe>0, gamma_IPCHI2_OWNPV<200,
Mee<100MeV/c and Mee-0.04*ENDVERTEX_Z<20 mm
CALO photons:
isNotH>0.3 (no cut on isNotE as bad agreement data/MC)
No / separation: most photons come from pi0 so better keep also merged pi0s + better efficiency

7. DD vs LL: MC study Conversion in first part of Velo DD LL
true Z <500mm DD LL
Low : not enough hits to be reconstructed as Long, (especially if late conversion) => Downstream pairs
High : the number of Velo hits is less dependent on the z of conversion => still LL pairs at end of Velo
Conversion in 2nd part of Velo
500< true Z <900mm DD LL

8. DD vs LL Radiation length ( -  plane) (LHCb-2007-025)
Conversion after Velo
true Z >900mm DD LL
VELO to RICH interface
RICH

9. ee selection: LL MC 2011 LL pairs MC 2011 LL true gamma
Fraction of events rejected by this cut: 13% in MC - 17% in data
Background:
combinatorics including misID (e) and 0ee
~55% are rejected by this cut

10. ee LL invariant mass: data/MC
Distributions agree well
30% more pairs rejected by Mee vs Z_vertex cut in data
Background: combinatorics inc misID (e) and 0ee
55% are rejected by Mee vs Z_vertex cut
Remaining bkg in MC ~ 4%  5% in data
Removed by Meevs Z_vertex cut
-data
-MC

11. ee DD invariant mass: data/MC
Agreement less good than for LL: magnetic field uncertainties?
Fraction of events rejected: 10% in MC 12 % in data
Efficiency and background estimated with chi_c data
Rejected by Mee vs Z cut
-data
-MC

12. Efficiencies: ee DD
Efficiency of cut Mee-0.04*ENDVERTEX_Z<20 mm measured using c sample:
inefficiency MC Data
600<pT<1300 : /-0.3% /-0.3%
1300<pT<4000 : 2.2 +/-0.2% /-0.3%
Not passing cut
With cut
Data
600<pT<1300
Data
1300<pT<4000

13. Background in ee DD sample
Cannot rely on MC truth information to check photon real ID: check on c MC
In about 50% of the conversions 1 of the e+/- has TRUEID==0
Not good matching due to energy loss  worse Eres but good conversions
TRUEID!= 11 or 0  negligible (0.3%)
Gamma TRUEID==22
Gamma TRUEID==0
e+ TRUEID==11 , e- ==0
e+ and e- TRUEID==0
Check fraction of events in upper mass sidebands:
- only real photons for TRUEID==22
- mix of real photons and background for TRUEID==0
Background fraction = 13% of the TRUEID==0
3.40.5% of the full sample
No dependance with energy observed
Gamma TRUEID==22
Gamma TRUEID==0
M(ee) (MeV)

14. DD vs LL Ratio of number of DD to LL conversions in Data and MC
The agreement is pretty good except in 2 regions:
LL -> DD migration due to Velo inefficiency
badly reproduced in MC for late conversions? ?
Velo-> RICH interface
Slightly less material in MC?

15. Principle of the measurement
Measure the ratio of number ee to the number of CALO
 measure of conversion probability x reconstruction efficiency as a function of pT and 
(pT,)=N(ee )/N(CALO)
Compare data to MC using minimum bias events – 2011 data
MC: Event type Sim08c – Reco 14 stripping20r1
Data: Reco14 stripping 20r1
Trigger: Hlt1L0AnyRateLimitedDecision and use only L0DiMuon,L0Muon trigger and L0Hadron to avoid bias from CALO clusters (trigger independant of our signal)

16. Efficiencies: CALO Efficiency of isNotH cut (MC) Background fraction
Electrons
others 5% 3% pT pT
isNotH: data less peaked at 0 and 1
Matching region ~0.4 =>Photon inneficiency underestimated by ~20%
Mostafa: D/MC agreement for cut isNotH>0.3 for pT>2.5GeV
Seems to be ~10% less background in data but tail might be different
Data MC
MC: true gamma
isNotH distribution

17. pT distributions Data vs MC: harder photons in data pT ee pT CALO
Resolution: does not affect the distribution for ee but does for CALO
 better do some unfolding
pT ee
- Data
- MC
pT CALO
- Data
- MC
pT ee
- MC true pT
- MC rec pT
pT CALO
- MC true pT
- MC rec pT

18. Unfolding pT CALO true pT CALO +MC rec unfolded + MC rec raw
- MC true
CALO
+data unfolded
- Data rec raw
rec pT

19.  distributions Example for 1300<pT<1700  ee  CALO - Data
- MC
 ee
- Data
- MC
Material budget for z<930cm from DC06
N(ee)/N(CALO)
- Data
- MC

20.  distributions Example for 1700<pT<2300  CALO - Data - MC
 ee
- Data
- MC
N(ee)/N(CALO)
- Data
- MC
Material budget for z<930cm from DC06

21. Efficiency results: (pT,)=N(ee )/N(CALO)
For nVeloTracks <60 and nSPDHits<600
2.25 <  < 2.5
2.5 <  < 2.75 MC
Data
2.75 <  < 3.0
3.0 <  < 3.25

22. 3.25 <  < 3.5
3.5 <  < 3.75
3.75 <  < 4.0
4.0 <  < 4.25

23. Ratio Data/MC 2.25 <  < 2.5 2.5 <  < 2.75
2.75 <  < 3.0
3.25 <  < 3.5
3.5 <  < 3.75
3.0 <  < 3.25

24. Ratio Data/MC 3.75 <  < 4.0 4.25 <  < 4.5
4.0 <  < 4.25

25. Dependance with nVeloTracks
3.0 <  < 3.25
80<nVeloTracks<160
nVeloTracks<80

27. Summary To do: Corrections for efficiency selection
Background subtraction
Improve unfolding to recover true pT distribution
Study dependance of results with multiplicities (nVeloTracks, nSPDHits)
Preliminary results:
Data/MC ~0.9 at low and high 
 consistent with Downstream tracking efficiency ~0.95
Data/MC ~0.82 for 3.0 <  < 3.75
 lack of material in MC in this region?

28. 50<nVeloTracks<80 2.75 <  < 3.0 3.0 <  < 3.25
3.25 <  < 3.5
3.75 <  < 4.0
4.0 <  < 4.25
3.5 <  < 3.75