1. A lecture on: Physics, Statistics, History & Sociology
Higgs Search at LEP
A lecture on:
Physics, Statistics, History & Sociology
Nikos Konstantinidis (UCL)

2. Overview The physics Statistics Results Lessons for the future Outlook
Production and decay of the SM Higgs at LEP
Event topologies and characteristics
Statistics
Results
Lessons for the future
Outlook
Nikos Konstantinidis
Higgs Search at LEP

3. SM Higgs production – decay @ LEP
Decays e_ Z e+ Z* H
Cross section ~0.1pb
Looking for near-threshold
production
( threshold = Ecm - MZ )
Nikos Konstantinidis
Higgs Search at LEP

4. HZ - Topologies (1) (2) BR50% BR15% (3) (4) BR5% BR8%
Nikos Konstantinidis
Higgs Search at LEP

5. bbll: the “golden” channel
Low rate, but excellent Higgs mass resolution (~3.5GeV) and almost no tails
MH calculated as the recoil to the dilepton system
Dominant bkg: ZZ
Small but nasty tails due to Final State Radiation (FSR) from the leptons
Dilepton mass lower, recoil mass higher…
Also: ZZ with Initial State Radiation (ISR)
Dilepton mass OK, but recoil mass high!
Nikos Konstantinidis
Higgs Search at LEP

6. bbnn
Second most important channel, but half of the event is missing, so Higgs mass resolution not good (~5GeV) and with tails
Main bkg
ZZ -> bbnn
ee -> ggbb : Difermion production with double Initial State Radiation
Nikos Konstantinidis
Higgs Search at LEP

7. 4jets More sensitive than all the other channels put together
Good MH resolution (~3.5GeV), but tails from
mismeasurements
overlapping jets
wrong jet pairings
Main bkg:
ZZ → bbqq (tails also from the Z width)
ee → bb → bbgg (double gluon radiation)
WW → cscs
Nikos Konstantinidis
Higgs Search at LEP

8. Reconstructed Higgs mass in 4jets
LEP : excellent energy resolution (~20MeV)
LEP detectors: good jet angular resolution (~2 degrees)
fix jet directions and rescale their energies/momenta
to satisfy energy-momentum conservation
(Ei,pi) → ai(Ei,pi)
SaiEi = Ecm and Saipi = 0
(4c-fit slightly more sophisticated)
In the end, because m12 and m34 are anti-correlated:
MH = m12 + m34 -mZ
Nikos Konstantinidis
Higgs Search at LEP

9. Putting it all together: confidence levels
Many channels and a lot of info to combine:
mass resolution of individual channels
BR’s and purity of individual channels
different centre-of-mass energies
other discriminating variables (e.g. b-tagging)
Combine all info to produce:
the confidence level for the signal+bkg hypothesis (CLs+b)
the confidence level for the bkg only hypothesis (CLb)
Nikos Konstantinidis
Higgs Search at LEP

10. How to construct CLs+b and CLb
Assume a signal (e.g. SM Higgs: m=115GeV)
Put all the properties of signal/bkg in one estimator e (e.g. LR: Q=L(s+b)/L(b))
Use a lot of bkg-only/sig.+bkg toy expts, and find the distribution of e
Find edata and
Separation between the two dist’ns
shows the sensitivity of the expt.
Nikos Konstantinidis
Higgs Search at LEP

11. LEP result
[ LEPHWG ’03 ]
Nikos Konstantinidis
Higgs Search at LEP

12. 1-CLb: bkg compatibility
ALEPH excess near 116 GeV
due to 2-3 “golden” 4jet events
“Nothing” observed by DLO
At mH=116GeV:
ALEPH: 3.0s (<5s)
LEP: s (<5s)
[ LEPHWG ’03 ]
Nikos Konstantinidis
Higgs Search at LEP

13. Other distributions [ LEPHWG ’03 ] Weights of candidates
Overall, numbers consistent with Higgs production at ~ GeV
Nikos Konstantinidis
Higgs Search at LEP

14. “The results of the experiment were inconclusive
so we had to use statistics…”
Nikos Konstantinidis
Higgs Search at LEP

15. Consistency checks Why only in ALEPH? Why only in 4-jets?
When we are talking about ~2 events anything is possible
ALEPH had the best single-experiment sensitivity
Why only in 4-jets?
Based on BRs, efficiencies etc you should expect
3 four-jet events before you see 1 bbnn event and
6 four-jet events before you see 1 bbll event
Better shown with stat. estimators 
Nikos Konstantinidis
Higgs Search at LEP

16. Estimators per experiment
Nikos Konstantinidis
Higgs Search at LEP

17. Estimators per channel
Nikos Konstantinidis
Higgs Search at LEP

18. Other systematic checks on…
behaviour as a function of CM energy
b-tagging
kinematical fits
etc…
All showed no inconsistencies/discrepancies
Nikos Konstantinidis
Higgs Search at LEP

19. Nikos Konstantinidis
Higgs Search at LEP

20. Candidate 54698/4881 Properties
(recorded on 14/06/2000)
Properties
Two clear b-jets (dec. length, inv. mass of tracks in vertex)
Event well-measured: Pmis in direction of jet with m from vertex
Planar event, as in threshold production of two heavy particles
b-jets: 55GeV & 59GeV, like in a decay almost at rest
Non-b jets: 43GeV & 49GeV, like in a Z decay almost at rest
Raw invariant mass of non-b jets 92.3GeV
non-b jets: leading parton effect, low multiplicity (q vs. gluon)
Impossible to be a WW, very unlikely to be bbgg, very unlikely a ZZ*→qqbb (and if it is we were very unlucky!)
One candidate is not a discovery, but if mH~116GeV, this event was the first Higgs ever observed!
Nikos Konstantinidis
Higgs Search at LEP

21. Lessons for the future
Event selections should be not only sensitive but also robust
Some “golden” candidates disappeared after “data reprocessing”  bad for LEP’s credibility.
Define analyses a priori; define systematic checks and consistency checks a priori
Hard to have a statistically meaningful result otherwise.
Improve analyses by inspection of candidates
But improvements to be used in new data only!
Nikos Konstantinidis
Higgs Search at LEP

22. Search window vs. mass resolution
What is the probability to see such a peak anywhere in the mass range that I am looking at?
MH=200GeV
Signal Bkg
Events / 0.5 GeV
CMS 10 fb-1
H  4l (l=e,)
m (4l)
Nikos Konstantinidis
Higgs Search at LEP

23. LHC: Channels for mH~115GeV
gg  H  
ttH  tt bb  bl bjj bb
VBF: qqH  qq H  b b
Different production/decay modes
Different backgrounds
 complementary!
*(K-factors  (NLO)/(LO)  2 not included)
total S/ B 
(10fb-1)
Nikos Konstantinidis
Higgs Search at LEP

24. The future is bright… Outlook The journey goes on
It has produced great excitement & there is more to come!
We should know the final answer by the end of this decade!
The future is bright…
Nikos Konstantinidis
Higgs Search at LEP