Prsentation PowerPoint

1
Search for Heavy Resonances Decaying into Zj
S. Kermiche, E. Nagy
CPPM

• Physics interest
• Data sample
• QCD Background
• SM background
• Model Independent Search for M and G
• Comparison with models

2
Physics Interest

• Signal is clear very little QCD background
• Not yet explored
• Excited quarks can produce such a signal

 gauge  interaction Z1j peak in MZj, pTZ, pTj
 contact  interaction Z2j peak in MZj
Data
MC
3
Data sample

• 2EMhightpt Pass2 skim 2 loose electrons pt gt 15
  GeV
• TMBTree produced by p16.05.01 (thanks to Oana
  Boeriu)
• duplicated events removed
• 4 407 181 events selected

pre-V12 EM_MD12_CEM10
2EM_HI EM_HI_SH 2EM_2MD7
EM15_2JT15 EM_MX 2EM_2MD12
EM_HI_METSH 2EM_MD10_CEM10 V12 and V13
E1_2L15_SH15 E1_2L20 E3_2SH8
E2_2L8_T8L8 E1_SHT20 E3_5_11_SH5_T4L5
3J15_2J25_PVZ E1_SH30 E2_2L15_SH15
E3_SHT15_2J_HT50 3JT15_PVZ
E1_SH15_2J20_M10 E1_L20_M25
Out of 836 triggers 22 selected triggered 99.8
of skimmed events
Trigger efficiency determined using µ triggers
(15437 evts) is compatible with 100 for pte1 gt
30,
pte2 gt 20 GeV ? No correction will be applied,
only in the error it will be accounted for
4
Trigger efficiency
Pte2 15 GeV
17 GeV
21 GeV
Pte1?
29 GeV
37 GeV
Pte1 13 GeV
41 GeV
Pte2?
27 GeV
31 GeV
39 GeV
49 GeV
57 GeV
5
Preselection

• Bad runs, bad LBNs removed
• Triggers (see before) selected
• Calorimeter quality required
• EmptyCrate, CoherentNoise, RingOfFire,
  NoonNoise
• gt 2  loose   scone  electrons
• of pTgt15 GeV pointing to Vtx ( 1  or
   2 )
•  loose  electron id 10, 11
• emfrac gt 0.9
• iso lt 0.15
• HMx8 lt 25 if
  ? gt 1.1
• HMX7 lt 15 if
  ? lt 1.1

181 110 events preselected
6
QCD background

• Very little
• Can be reduced by track-match and HMx cut
• Simulation veto on track-match and invert HMx
• HMX7 gt 10 in CC
• HMX8 gt 20 in EC

Track-match veto only
Track-match veto and inversion of HMx
7
SM background Z nj

• There are several generators to do the job
• Pythia 2?1 process (one could also add ff?Zj)
• Alpgen separate generations of 0, 1, 2, jets
• have to be combined by weighting
  with xsection
• resolving double counting by MLM
  prescription
• CKKW separate generation of 0, 1, 2, jets
• have to be combined by weighting
  with xsection
• but no care is needed to avoid
  double counting
• Sherpa

Since official production is available at that
moment only for Pythia and Alpgen, comparison of
data and MC will be reported here only for those
generators
Comparison will be shown for Pythia, Alpgen MLM
and for some variants of the MLM prescriptions
8
Realization of the MLM prescription for Znj
sample (courtesy of M. Begel )

• Filter events to obtain "skimmed" Zkj samples
• Combine the skimmed samples by weight wk sk/Nk
• sk Alpgen xsection
• Nk number of unweighted events before
  skimming

A skimmed event has to meet the following three
criteria i/ Nj gt Np Nj number of jets
reconstructed using secondary partons
(i.e. not the ones originaly generated by
Alpgen) Np number of original partons
(generated by Alpgen) ii/ Each Alpgen parton
must match uniquely with a parton-jet (no
two partons can match the same parton-jet)
?R lt ?Rlim 0.4 of PTj gt PTcut 10 GeV
(One has to tune ?Rlim and PTcut it has
been taken from M. Begel) iii/ Either there
remains no unmatched reconstructed jets
( exclusive ) or Np n where n is the
number of jets looked for ( inclusive  sample)
The skimming code for TMBTrees can be found
on /work/pagnol-clued0/enagy/qstar/skimMC It is
an adoptation of the code of M. Begel for
TopTrees
9
All events were fixed with Pass 2
stored on TMBTrees with d0correct of p16.05.01
smeared jets and electrons à la
U. Blumenschein
10

• Selection of objects for comparison
• Data vs MC
• Is identical for Data and MC
• Is aiming at the least QCD background
• gt 2 tight electrons
• id 10,11
• emfrac gt 0.9
• iso lt 0.15
• is in fiducial (not in ?- nor in f-crack)
• HMX7 lt 10 in CC, HMX8 lt 20 in EC
• trackmatch
• pTe1 gt 30 GeV, pTe2 gt 25 GeV
• gt 1 good jet
• pT gt 10 GeV
• corrJCCB type L1 confirmation
• standard jet-ID
• no match to
  good em-object
• SpTch/pTj gt0.05
• NZ 1

11
Comparison is done for the shapes only -- data
and MC will be normalized at small MZj since we
are looking for excess at high MZj
Shapes are compared for testing Mee ?
smearing pTZ, pTe1 ? smearing and QCD matrix
element pTj1, pTj2, Nj ? QCD m.e. and
fragmentation MZj1, pTZj1 ? signal for q
Comparison is shown for Pythia Alpgen
Z0,1,2j skimmed and weighted à la MLM (MLM)
Z0,1,2j skimmed but weighted with Nevt
of skimmed events (MLMmod) Z0,1j
skimmed and Z2j unskimmed events (Z2jnoskim)
Z0j skimmed Z1j unskimmed events
(Z1jnoskim) Z0,1,2j unskimmed
events (Zallnoskim)
12
Mee distributions
Pythia
?2/Ndf124/43
MLM
98/39
81/41
77/39
Z2jnoskim
MLMmod
126/40
Z1jnoskim
Zallnoskim
130/42
13
pTe1 distributions
MLM
91/34
Pythia
X2/Ndf124/38
Z2jnoskim
331/41
422/34
MLMmod
37/35
Z1jnoskim
Zallnoskim
97/41
14
pTZ distributions
MLM
217/27
X2/Ndf218/30
Pythia
46/33
33/27
Z2jnoskim
MLMmod
Zallnoskim
42/33
82/22
Z1jnoskim
15
pTj1 distributions
Pythia
X2/Ndf228/34
139/27
MLM
108/27
Z2jnoskim
MLMmod
97/37
100/24
Z1jnoskim
Zallnoskim
63/37
16
pTj2 distributions
49/17
X2/Ndf51/15
Pythia
MLM
234/17
Z2jnoskim
MLMmod
29/18
31/13
Z1jnoskim
Zallnoskim
9/18
17
Nj distributions pTjgt20 GeV
Data
--
MC
142/5
MLM
Pythia
X2/Ndf151/5
294/6
Z2jnoskim
356/5
MLMmod
Zallnoskim
112/6
30/4
Z1jnoskim
18
pTZj1 distributions
X2/Ndf57/17
Pythia
31/19
MLM
MLMmod
33/19
28/23
Z2jnoskim
28/17
Z1jnoskim
Zallnoskim
23/23
19
MZj1 distributions
X2/Ndf86/37
Pythia
71/37
MLM
124/37
MLMmod
90/39
Z2jnoskim
Z1jnoskim
Zallnoskim
81/40
68/36
20
Conclusions for the comparison
Mee some excess in data for lower end of the Z ?
QCD bg
Good
Fair
Bad
21
Model independent search for M and G (will be
done next)

• Determination of the ratio of efficiencies
  Data/MC
• Generation of the q?Zq vs M and G
• Determination of the luminosity for the triggers
  selected
• Determination of the excess/deficit of data vs
  SM
• Draw s(95) limit

Comparison with q production models to exclude
masses widths
Target Winter conferences 2005