Transcript Z -> bb - D0

D
Z  bb
Physics and Calibration
•Small physics reminder
•Z  bb
vs
 + jet
•Selection w/o STT, Background rejection
•Calibration procedures
•A few plots on the Hit-Root-tuples
1
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
Higgs Search
If Mhiggs < 135 GeV promising channel:
pp  WH  l  b b
Signal to background depends crucially on b b mass
reconstruction resolution
Can
we go from 15% resolution to <10% ?
2
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
Higgs Search
Significance evolution with resolution at Mhiggs=120 GeV
bb mass Resolution:
15%
10%
Signal events/fb-1
Wbb
WZ
tt
single top
4
59
11
34
14
4
32
6
24
10
For 10 pb-1
3
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
Z  bb
vs
 + jet
 + jet : high statistics, allows for a tight
b-jet selection (b-tagging).
Systematics are not straightforward.
Z  bb: very low signal/noise but systematics
closer to physics processes (H or Top),
due to high pT . Resonance mass
independent of mult. interactions.
Complementarity !!
4
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
Z  bb selection
D0 will start without STT (and run this way about a year)
Can we select these events w/o STT?
CDF Run I
S/N below 10-3 implies very stringent selection.
On the total Run I lumi, only 45 events left, starting from about 120
000 Z  bb events
Trigger: central muon (pT> 7.5 GeV)  5.5 M evnts
Offline: request 2 tagged (0.7 cone) jets  5479 evnts
According to Pythia: expect 124 +/-14 evnts
Needs further topological cuts
5
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
3 ET and
12
 Signal and background are produced differently:
Z is produced by a time-like q-qbar anihilation,
Background is QCD induced. Color flow between initial
and final partons only in the bckgd.
Z is expected to have soft radiation between the jets,
Background will also have strong radiation between
IS and FS partons.
Use 2 kinematic variables to discriminate
 3 ET : sum of ET of the clusters outside the 2
leading jets
 12 : azimuthal angle difference between the 2 jets
6
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
QCD and E_Weak Samples
 Discriminating power of the 2 variables
3 ET and 12
on 2 samples of different origin
7
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
Z  bb Signal
 3 ET < 10 GeV 12
> 3 rad
 Now S/N=1/6 at the Z
mass peak
 Select/antiselect w.r.t.
the 2 var., tag
probability
 3.2  excess
 Background only
8
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
Z  bb Sample
3 ET
9

G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
Likelihood fit to the Dijet Sample
 Signal + Bckgd
 Background only
 Results:
MZ=90.0 +/- 2.4 GeV
Z = 9.4 +/- 3.5 GeV
NZ=91+/30(st.)+/19(sys.)
10
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
Neural Net ?
 With only 2 topological cuts, strong background
reduction.
 Neural net on further variables will improve purity, but
input variables must be decorrelated from calibration
variables.
 With 0.4 fb-1, if S/N better than 1/2
precision on absolute scale may be better than 1%
 With the STT, much purer sample, i.e. expect furter
improvement
11
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
How to calibrate???
 On detailed MC, compare b-jets with light-q jets.
Root-tuples available for b-jets (Hit),
on request for q-jets.
 Expect: wider b-jets (due to the large b-mass) ,
muons from semileptonic decays
(to be corrected).
study track multiplicity, and E/p to
improve the jet response using Track+Calo
 Devise a relative calibration b-jets vs q-jets from MC
 Check it on the Z  bb data sample selectionned with
a Neural Net, while awaiting for the STT
12
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
Back To Earth: Root-tuples
Hit Group Root-tuples :
follow the colour code:
Z-> bb , Z-> tau-tau , Z-> ee , Z-> mu-mu
8000 Events in each sample.
Includes min. bias events
To start: Compare Basic distributions
13
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7
D
Conclusions
•Hit Group Root-tuples are a good Tool to start
•A few variables could be added, others improved
•Need of Z -> qq to make real progress
•Selection of Z->bb will require Neural Net
•Calibration of b-jets will take time, but start now to
implement track+cell for jets
14
G.Bernardi, LPNHE-Paris
Calibration Meeting,
2000/12/7