1. Measurement of associated Z plus charm production in pp collisions at s= 8 TeV
CMS Collaboration
Juan Alcaraz Maestre, Alberto Escalante del Valle, Juan Pablo Fernández Ramos, Isabel Josa Mutuberría(CIEMAT).
CPAN VIII – Physics at LHC
Nov. 28, 2016
CMS-PAS-SMP (

2. Vector boson +HV jets tests PDF for s,c,b
Introduction
Measurement of σ(pp⇾Z+c) and its relative production w.r.t. Z+b both inclusively and differentially
Tests QCD predictions
Allows gluon (charm) PDF studies
Intrinsic Charm-quark component inside the proton enhances ↑pt(Z)
Background in searches (relative contributions of the different flavors to the background is important since Z+c can be misidentified as Z+b jet events)
Study the performance of c-tagging algorithms (new in CMS since 2015)
c-tagging = identification of jets coming from charm quarks 2

3. Analysis strategy: Z → l+l− with pT(l) > 20 GeV and |η(l)| < 2.1
SV-mass SMP
(Z+b)
Dilepton invariant mass: [71,111] GeV
Charm selection from std. CMS tools (b-tagging) not optimal
Take advantage of the relatively long life of charm hadrons
Identifify charm in final states:
Semileptonic: lepton in jet
D*+/- and D+/- exclusive decays in jet
pTjet >25 GeV & |jet|<2.5 3
Charm hadrons decay weakly (with decay lenghts > 100μm)
LO diagram for Z+c(c) production

4. Samples 4 DATA: 2012 8 TeV ( 19.7 fb-1 )
MC: DY+jets generated w. MADGRAPH5 + PYTHIA6 using PDF set CTEQ6L
Cross section normalized to σ(pp⇾Z+X) at NNLO calculated with FEWZ and
PDF MSTW2008NNLO
Main backgrounds:
DATA: TeV ( 19.7 fb-1 ) -
Contributions from ttbar, diboson, Z+light processes (from simulations except ttbar from data)
Missing transverse energy < 40 GeV (to reduce ttbar background)
Simulated samples corrected for differences between data and MC in describing the lepton trigger, reconstruction and identification efficiencies
Inclusion of a re-weighting factor to account for the presence of more than 1 interaction per pp crossing (add MB interactions with given profile and re-weight it).
Clasification: according to the flavour of the gen. level jet (gen jet)
Gen jet is b-flavoured if b-hadron
c-flavoured if c-hadron and no b-hadron
(light otherwise)

5. Semileptonic selection
The muon candidate is one of the constituents of the jet and of the tracks of a secondary vertex (SSV or IVF).
With moderate pT, pTμ<25 GeV, with pT(μ)/pTjet<0.6 & |ημ|<2.5
μ not-isolated from hadron activity, Icomb/pTμ>0.2
Semileptonic candidates:
5258 Z ⇾ μ+μ-
4145 Z ⇾ e+e-
Z+c: ~25% Z+b: ~65% Z+light: ~5% Others:~5% 5
The most reliable (demanding) Identification criteria

6. D± channel – Selection Jet with a 3 tracks vertex (SSV, IVF)
Search for D± → K∓π±π± resonant peak
Signal region :
|m(D±) − 1.87| < 0.05 GeV
Sideband region :
0.05 < |m(D±)−1.87| < 0.10 GeV
Non resonant background in the signal region subtracted from the neighboring sidebands
After sideband subtraction:
490 ± 48 D± (Z⇾μ+μ- )
375 ± 44 D± (Z⇾e+e- )
Z+c: ~60% Z+b: ~36% Z+light: <1% Others:<3% 6
If several candidates get the 1st candidate in the IVF collection in the jet with higher pt

7. D*± channel - Selection
Search for D*± → D0 πs± [D0 → K-π+(+c.c.)] decay chain
Loop over all tracks in the jet (instead of using SSV or IVF col.)
D0 → K-π+(+c.c.) ; Kaon: track with sign opposite to third track the πs (D*∓ → D0 πs∓) . Apply the following selection criteria :
◮ Lxy /σ(Lxy ) >3, D0 vertex prob.> 0.05
◮ pt (K) > 1.75, pt (π) > 0.75, pt (πs ) > 0.5
◮ |ΔR(D∗, jet)| < 0.5, |ΔR(D0, πs )| < 0.1.
◮ |m(D0) − 1.865| < 0.1 GeV, |Δm − 145| < 5 MeV
◮ Signal region :
1.97 < D∗ < 2.05 GeV
sidebands : adyacent bands
Non resonant background in the signal region subtracted from the neighboring sidebands
After sideband subtraction:
309 ± 22 D*±(Z⇾μ+μ-)
234 ± 22 D*±(Z⇾e+e-)
Z+c:~65% Z+b:~30% Z+light:<1% Others:<4% 7
If several candidates get the candidate with higher pt in the jet with higher pt

8. b/c separation (discriminants)
Vertex mass (for semileptonic mode)
JP (for D hadron modes): likelihood estimate of prob. of
jet tracks to come from primary vertex
To account for unidentified neutral decay products
The larger the IP of a track the more incompatible w.r.t. PV
JP discriminant from BTV
Template modeling:
- Z+c : data driven (W+charm)
- Z+b : from MC but corrected with data (ttbar)
- Z+light and Dibosons: from MC
- ttbar: Data driven 8
Z+b calibrated with data sample enriched in top (eμ) prior to signal extraction.
The calibration strategy validates the shape and vertex efficiencies in simulation.

9. Modeling for Z+c
Evaluated in a clean sample of c-jets from W+c production.
W e plus jets with similar selection to Z+c
Same identification of heavy flavor: μ in jet or D-hadron
OS–SS subtraction to remove symmetric backgrounds
After OS-SS subtraction the purity in W+c of the resulting sample is > 90% (semil. channel) and >98% (D-hadron chan.)
JHEP (W+c) 9

10. Modeling for Z+c
Templates after subtraction of remaining (little) background
Agreement in general distributions (pTjet , NSV )
SV-mass W+c MC and Z+c MC agree
SV-mass W+c MC and W+c data do not agree 10

11. Template modeling for Z+b-
Evaluated in a clean sample of b-jets from tt production.
Using a sample where the two W bosons from the top(antitop) quark decay leptonically into leptons of different flavour.
Identification of heavy flavor: muon inside the jet
Not enough statistics to validate the Z+b templates in the D-hadron channels 11

12. Signal extraction
Total # of observed Z+c/Z+b extrated from a χ2 minimization fit of the Z+c/Z+b templates to the experimental distributions of vertex mass and JP discriminants.
ni = Number of events in data
NiZ+c, NiZ+b = Number of Z+c, Z+b
Z+light, ttbar, Dibosons subtracted
Parameters to fit: μz+c & μz+b
μz+c & μz+b in the range
Z ⇾ ee
Z ⇾ μμ 12

13. Z ⇾ μμ D± D±
Z ⇾ ee
D*±
D*±
Z ⇾ μμ
Z ⇾ ee 13

14. Cross section determination14
MADGRAPH (Z+c) = 8.02  0.09 pb (statistical error only)
MADGRAPH (Z+b) = 4.96  0.03 pb (statistical error only)

15. Cross section determination
 (Z+c) 15

16. Cross section ratio
Only the semileptonic channel is used 16

17. Differential cross sections as a function of pTZ
Bins[GeV] : 0-30, and 17

18. Differential cross sections as a function of pTjet
Bins [GeV] : 25-40, and 18
There is some tension in the 1st pTjet bin but we can say nothing with available
Statistics.

19. Conclusions
Evaluated (Z+c) y ratio (Z+c)/(Z+b) inclusive and differential for two opposite sign leptons from the Z with pTlepton > 20 GeV and
|lepton| < 2.1 and pTjet >25 GeV & |jet|<2.5
(Z+c) = 8.6 ± 0.5 ± 0.7 pb
(Z+c)/(Z+b) = 2.0 ± 0.2 ± 0.2
In agreement with predictions from MadGraph5 and Madgraph renormalized to a FEWZ calculation.
CMS-PAS-SMP ( 19
The differential (Z + c) cross section measurement at high pZ could be slightly enhanced due to the existence of a charm component in the proton of nonperturbative origin. The most sensitive region, 60 GeV < pZ < 200 GeV, can be explained using PDFs that include only charm generated perturbatively. Future measurements with the new 13 TeV dataset at higher pZ > 200 GeV, where this enhancement is expected to be larger, should be able to exclude/confirm the existence of a intrinsic charm component inside the proton.

20. Back up

21. Systematic uncertainties
PU : uncertainties in the evaluation of the pileup
profile (assuming a different inelastic cross section )
PDFs (PDFNNPDF & PDFCT10) : reweighted DY MC
according to the new PDF's : NNPDF23_nnlo & PDFCT10
Jet Energy scale and Resolution (JES & JER) :
Change scale and resolution correction factors
by their uncertainties
Branching ratios of D⇾Xμ hadrons and fragmentation c⇾D (BRFRAG): We reweight the simulation to match the PDG on D-hadron channels. The syst comes from uncertainty on those reweighting factors For semileptonic channel the systematic comes from the difference between the BR from inclusive or exclusive individual contributions
Missing-et : Misestimations on the Missing transverse energy: modify the missing E_t by 10% of the unclustered missing E_t.
Cgluon-splitting, misestimation of the contribution to our background from gluon-splitting from charm : increase a factor 50 % the weight on events with 2c with
ΔR (jet,c) < 0.5
Bgluon-splitting, increase 50% the weight on events with 2b w. ΔR(jet,b) < 0.5
c-(b-) tagging efficiency. Repit analysis with efficiency increased by its error
Shape (semileptonic mode) : change Z+b template correction factor by its error
Lepton efficiencies :change efficiencies by their errors
Luminosity : 2.6 %

22. Preselection Preselection of Z ee and Z  plus jets Z ee: Zμμ :
Dielectron trigger thres-hold of 17 and 8 GeV
Electron Id Medium WP
pT(e)> 20 GeV, |(e)| < 2.1
Isolated: Icomb/pT <0.15
Energy scale corrected
MC corrected for differences in Electron Id and Iso efficiencies
Zμμ :
DiMuon trigger thres-hold of 17 and 8 GeV
Tight selection
pT(e)> 20 GeV, |(e)| < 2.1
Isolated: Icomb/pT <0.2
Momentum scale and resolution corrected (Rochester correction)
MC corrected for differences in Muon Trigger, Id and Iso
Jets:
Jets anti-kT, R=0.5
JEC corrected, both in data and MC
pT(jet) > 25 GeV, |(jet)| < 2.5

23. Previous measurements at hadron colliders:
- Z+c+X)/σ(ppbar-Z+b+X) and σ(ppbar-Z+c+X)/σ(ppbar-Z+jets+X), pT(jet) > 20 GeV
2) pT(D*) > 3 GeV [σ(ppbar-W+D*+X)/σ(ppbar-W+X) as well]
Evidence and σ( pp-Z+D0+X) andσ( pp-Z+D+X) in the forward region