Eduard De La Cruz Burelo

1
Observation of the doubly strange b-baryon ?-b

• Eduard De La Cruz Burelo
• CINVESTAV IPN Mexico
• On behalf of the D0 collaboration

• Outline
• Introduction
• ?b observation
• ?b search
• Summary

August 28th, 2008
WC Seminar at Fermilab
1
2
The ?-(sss) discovery (1964)
The Eightfold way SU3
3
What energies we will looking at?
Do we miss something here?
Z, W, top, Higgs?
Z,W,KK modes, etc.
TeV
few GeV
GeV
MeV
4
What energies we will look at?
B Physics
Bs Mixing, CKM, lifetimes, etc.
TeV
few GeV
GeV
MeV
5
B baryons at the Tevatron
J1/2, 1 b

• Unique to Tevatron (not produced in B factories)
• B baryons expected to be produced copiously at
  the Tevatron
• Only ?b was considered as observed back in 2001
  20 events.
• Interesting mass predictions using different
  models.
• However, very challenging.

?b(bud)
?b0(bud) ?b(buu) ?b-(bdd)
?b0(bus) ?b-(bds)
?b-(bss)
Not yet observed
6
When Tevatron Run II begun
Notation Quark content JP SU(3) (I,I3) S Mass
?b0 bud 1/2 3 (0,0) 0 5619.7?1.2?1.2 MeV
?b0 bsu 1/2 3 (1/2,1/2) -1 5.80 GeV
?b- bsd 1/2 3 (1/2,-1/2) -1 5.80 GeV
?b buu 1/2 6 (1,1) 0 5.82 GeV
?b0 bud 1/2 6 (1,0) 0 5.82 GeV
?b- bdd 1/2 6 (1,-1) 0 5.82 GeV
?b0 bsu 1/2 6 (1/2,1/2) -1 5.94 GeV
?b- bsd 1/2 6 (1/2,-1/2) -1 5.94 GeV
?b- bss 1/2 6 (0,0) -2 6.04 GeV
?b buu 3/2 6 (1,1) 0 5.84 GeV
?b0 bud 3/2 6 (1,0) 0 5.84 GeV
?b- bdd 3/2 6 (1,-1) 0 5.84 GeV
?b0 bus 3/2 6 (1/2,1/2) -1 5.94 GeV
?b- bds 3/2 6 (1/2,-1/2) -1 5.94 GeV
?b- bss 3/2 6 (0,0) -2 6.06 GeV
from hep-ph/9406359
7
?()-b in October 2006
CDF announced the observation of the ?bs with
1.1 fb-1 PRL 99, 202001 (2007)
8
Last year ?-b observation
Number of events 15.2 4.4 Mass 5.774
0.011(stat) GeV Width 0.037 0.008 GeV
We also measured
Signal Significance
PRL 99, 052001 (2007)
9
Last year ?-b observation
Also searched for in ?b-??c0?-
M(?b-) 5792.9 ? 2.5 (stat) ? 1.7 (syst) MeV/c2
Signal significance 7.8?
PRL 99, 052002 (2007)
10
During Tevatron Run II
Notation Quark content JP SU(3) (I,I3) S Mass
?b0 bud 1/2 3 (0,0) 0 5620.2 ? 1.6 MeV
?b0 bsu 1/2 3 (1/2,1/2) -1 5.80 GeV
?b- bsd 1/2 3 (1/2,-1/2) -1 5792.4 ? 3.0 MeV
?b buu 1/2 6 (1,1) 0 5807.8 ? 2.7 MeV
?b0 bud 1/2 6 (1,0) 0 5.82 GeV
?b- bdd 1/2 6 (1,-1) 0 5815.2 ? 2.0 MeV
?b0 bsu 1/2 6 (1/2,1/2) -1 5.94 GeV
?b- bsd 1/2 6 (1/2,-1/2) -1 5.94 GeV
?b- bss 1/2 6 (0,0) -2 6.04 GeV
?b buu 3/2 6 (1,1) 0 5829.0 ? 3.4 MeV
?b0 bud 3/2 6 (1,0) 0 5.84 GeV
?b- bdd 3/2 6 (1,-1) 0 5836.4 ? 2.8 MeV
?b0 bus 3/2 6 (1/2,1/2) -1 5.94 GeV
?b- bds 3/2 6 (1/2,-1/2) -1 5.94 GeV
?b- bss 3/2 6 (0,0) -2 6.06 GeV
11
B hadron observation status

• Mesons
• B, B0, Bs, Bc (before Tevatron RunII)
• B (before Tevatron RunII)
• Bd(Tevatron RunII)
• Bs (Tevatron RunII)
• Baryons
• ?b (before Tevatron RunII)
• ?b, and ?b(Tevatron RunII)
• ?-b (Tevatron RunII)

12
Data
In this analysis we use 1.3 fb-1 of data
collected by DØ detector (RunIIa data). Thanks
to the Fermilab Accelerator division for doing
wonderful work.
Muon and central tracker subdetectors are
particularly important in this analysis
13
How did we look for the ?-b?
?-b?J/??-
?
?-
p
?
?-b
?-
5 cm
5 cm
?-
0.7 mm
?-
14
Data reprocessing
When tracks are reconstructed, a maximum impact
parameter is required to increase the
reconstruction speed and lower the rate of fake
tracks.
?
?
p
?
?
But for particles like the ?b-, this requirement
could result in missing the ? and proton tracks
from the ? and ?- decays
?
?
15
Increase of reconstruction efficiency
D0
D0
D0
GeV
GeV
GeV
Opening up the IP cut (Before) ( After )
16
?b Reconstruction procedure

• Reconstruction procedure
• Reconstruct J/????-
• Reconstruct ??p?
• Reconstruct ??? ?
• Combine J/? ?
• Improve mass resolution by using an
  event-by-event mass difference correction
• The optimization
• ?b?J/?? decays in data
• J/? ?(fake from ?(p?-)? )
• Monte Carlo simulation of ?b-?J/??-

Final ?b selection cuts

• ??p? decays
• pT(p)gt0.7 GeV
• pT(?)gt0.3 GeV
• ?- ??? decays
• pT(?)gt0.2 GeV
• Transverse decay lengthgt0.5 cm
• Collinearitygt0.99
• ?-b particle
• Lifetime significancegt2. (Lifetime divided by its
  error)

17
Search for the ?-b(bss)

• bss quarks combination
• Mass is predicted to be 5.94 - 6.12 GeV
• M(?-b) gt M(?b)
• Lifetime is predicted to be 0.83lt?(?-b)lt1.67 ps

18
How do we look for it?
?
?-
p
?
?-b
?-
5 cm
?
?-
3 cm
Similar
K-
19
?-b vs ?-b differences
?-b(bds) ?-b(bss)
Decay ?b? J/?(??-) ??(???) ?b? J/?(??-) ??(?K?)
Mass 5.792 ? 0.003 GeV 5.94 - 6.12 GeV 1
Lifetime 1.42 ? 0.28 ps 0.84 1.69 ps. 2
Daughters ??(???) Mass 1321.71 ? 0.07 MeV c? 4.91 cm ??(?K?) Mass 1672.45 ? 0.29 MeV c? 2.461 cm

1 Phys. Rev. D 77, 014031 (2008)
arXiv0708.4027 hep-ph (2007). 2
arXivhep-ph/9705402
20
? reconstruction a challenge

• ??p? decays
• pT(p)gt0.7 GeV
• pT(?)gt0.3 GeV
• ?- ??? decays
• pT(?)gt0.2 GeV
• Transverse decay lengthgt0.5 cm
• Collinearitygt0.99

In this analysis for the ? reconstruction
D0
D0
21
Analysis strategy
Events are reprocessed to increase reconstruction
efficiency of long-lived particles.

• Select J/? candidates

Yield is optimized by using proper decay length
significance cuts.

• Select ??p?

Optimize yield by using multivariate techniques

• Reconstruction of ??? K

• Combine J/? (?K)

Keep blinded J/? ? combinations and optimize
on J/? (?K)
Improve mass resolution from 80 MeV to 34 MeV

• Event per event mass correction

Perform as many test as possible in different
background samples

• Fix selection criteria and then apply them to J/?
  ?

22
? optimization
First we select a proper decay length
significance cut to clean ? signal ( decay length
significance gt 10)
D0
?K will have huge combinatory background
23
? reconstruction

• Minimum selection cuts
• ?K vertex reconstructed
• Transverse decay length significancegt4
• Proper decay length uncertaintylt0.5 cm

D0
PDG mass value
Wrong-sign events ?K
Right-sign events (?K-)
24
Boosted Decision Trees (BDT)

• All variables are related to ?- or its decay
  products.
• We use a total of 20 variables.
• For training we use MC signal and background from
  wrong-sign events (J/?(?K)).

• Most important variables
• pT(K)
• pT(p)
• pT(?)
• ?- transverse decay length

25
?- after BDT selection
D0
Clean ?- signal
26
?-???- contamination

• This is a reflection contamination due to
  mistaken a pion as a kaon.
• It is easy to eliminate by requiring M(??)gt1.34.

D0
D0
27
?- after BDT selection
?????? decays removed
D0
Wrong-sign combination events
28
Final optimization

• We want to further reduce background (based on
  level we observe in the wrong-sign combinations.)
• We use ?- yield in MC signal verify that we
  maintain the highest possible signal efficiency.

29
Final optimization pT(B)gt6 GeV

• We compare MC signal vs wrong-sign background
  events pT distribution.

30
Final optimization ?(?)lt0.03 cm

• Similarly, MC signal is compare with uncertainty
  from wrong-sign events.

Uncertainty on ?
31
Wrong-sign combinations

• After optimization
• ??lt0.03 cm
• J/? and ? in the same hemisphere
• pT(J/??)gt6 GeV
• We define mass as
• Mass window for the search 5.6 - 7 GeV

After optimization, we look at wrong-sign
combination first
32
Other control sample
We analyze candidates in the sidebands of ?-
signal
D0
33
Other control sample
We analyze candidates in the sidebands of ?
signal
D0
34
Nothing where nothing should be
We check also high statistics MC samples
No excess is observed in any control samples
after selection criteria is applied to them.
35
Looking at right-sign combinations
Clear excess of events near 6.2 GeV
36
Mass measurement

• Fit
• Unbinned extended log-likelihood fit
• Gaussian signal, flat background
• Number of background/signal events are floating
  parameters

Number of signal events 17.8 4.9 Mean of the
Gaussian 6.165 0.010(stat) GeV Width of the
Gaussian fixed (MC) 0.034 GeV
37
Significance of the peak

• Two likelihood fits are performed
• Signal background hypothesis (LSB)
• Only background hypothesis (LB)
• We evaluate the significance
• Significance of the observed signal 5.4?

38
Consistency check Increase pT(B)
Significance gt6
39
Consistency check Look back plots
D0
D0
40
Consistency check lifetime
We compare to a MC sample with a lifetime of
1.54 ps (460 microns).
41
Alternative Cuts Based Analysis (CBA)
Variable BDT CBA
pT(?) (GeV) gt0.2 and input to BDT gt0.2
pT(p) (GeV) gt0.2 and input to BDT gt0.7
pT(K) (GeV) input to BDT gt0.3
?- collinearity input to BDT gt0.99
?- transverse decay length (cm) input to BDT gt0.5
Proper decay length uncertainty (cm) lt0.3 lt0.3
Variables selected based on relative importance
in BDT performance
42
Cut Based Analysis fit

• Fit
• Unbinned extended log-likelihood fit
• Gaussian signal, flat background
• Number of background/signal events are floating
  parameters

Number of signal events 15.7 5.3 Mean of the
Gaussian 6.177 0.015(stat) GeV Width of the
Gaussian fixed (MC) 0.034 GeV Signal
significance 3.9?
43
BDT or Cut Base Analysis

• After we remove duplicate events, we observe
  25.5 6.5 events.
• Significance 5.4?

44
Signal confirmed without BDT

• BDT vs CBA
• Consistent number of observed signal candidates
• Consistent mass
• Consistent reconstruction efficiencies
• BDT has better background rejection power.

45
One example Event display
46
One example Event display
47
Systematic uncertainties on the mass

• Fitting models
• Linear background instead of flat. Negligible.
• Varying Gaussian width between 28 40 MeV, 3 MeV
• Momentum scale correction
• Fit to the ?b mass peak in data, 4 MeV.
• Event selection
• Varying selection criteria and from the mass
  shift observed between the cut-based and BDT
  analysis, a 12 MeV variation is estimated .

48
Production rate
The systematic uncertainty includes contributions
from the signal yields as well as selection
efficiencies
49
Production rate
50
Summary
Number of signal events 17.8 4.9 (stat)
0.8(syst) Mass 6.165 0.010(stat)
0.013(syst) GeV Significance 5.4?
51
Summary
Consistent with expectations
52
Submitted to PRL
53
B baryons at the Tevatron
J1/2, 1 b

• B baryons If it decays to J/? it the Tevatron
  produces at reasonable rate, we will find it.
• Precision measurements will come with statistics.
• A legacy from the Tevatron

?b(bud)
?b0(bud) ?b(buu) ?b-(bdd)
?b0(bus) ?b-(bds)
?b-(bss)
54
Fermilabs latest discovery
b-baryon ?-b(bss)
55
Backup slides
56
Example Decision Trees

• A decision tree is a binary decision.
• Starting from the first node, between the
  variables the best split is selected.
• In the next nodes from the first split the
  operation is repeated.
• Splitting stops until you reach a given
  proportion of signal or background, or until no
  more splits can be done.
• We combine 100 DT by using the bagger technique

S/B
S1,B1
S2,B2
S12
S11,B11
S21,B21
B22
Splitting continue
57
Boosted Decision Trees (BDT)

• All variables are on ?- or its decay products.
• For training we use MC signal and background from
  wrong-sign events (J/?(?K)).

58
BDT output

• After training the BDT, it is applied to a
  validation sample
• Signal MC
• Background from wrong-sign events
• We select BDTgt0.
• BDT 1 Signal-like
• BDT -1 background-like.