1. Possible study after SUSY discovery at the ATLAS experiment
March 27th, 2007, Tokyo Metropolitan Univ.
Naoko Kanaya (Kobe University)
Shoji Asai (University of Tokyo / ICEPP)

2. Introduction  SUSY discovery is highly expected at the
LHC experiments (if it exists).
We don’t know almost nothing about SUSY
so it should be searched by inclusive way
to have sensitivity over wide parameter space
as much as we can…
Background should be estimated using real data. This is very crucial task for SUSY discovery (see talk by Okawa-san)
Assume we see SUSY-like signature
at ATLAS.
What we can/should do next?
This is main topics of this talk.

3. Outline   Possible inclusive study of SUSY signature
SUSY discovery and inclusive measurement
Possible inclusive study of SUSY signature
Summary and Outlook 
Effective mass distribution
Effect of background
Studied observable sensitive to signal characteristic
Effect of background
All studied shown in this talk is done by using fast simulation

4. SUSY ATLAS
Several inclusive search are studied at the ATLAS experiment:
0-lepton mode
1-lepton mode
di-lepton mode (opposite/same signed leptons)
di-jet mode
We may see some excess in some of (all?) analyses…
0-lepton mode analysis
After background subtraction
L=1fb-1
L=1fb-1

5. Effective mass distribution
One of well-known inclusive studies : Effective mass (Meff)
Meff = MissingET +  | pTjet | +  |pTlepton|
Why Meff?
Since Meff does not only have good sensitivity for SUSY discovery
but also have good relation to SUSY mass scale, MSUSY=min(mg. mq )
~ ~
Meff -> Msusy -> sigma

6. Effect of background
Effective mass is very powerful observable, however it is not so
easy in real life due to background.
Effective Mass (1fb-1) : 0-lepton
Bkg correctly estimation
Over-estimation(30%)
Under-estimation(30%)
MSUSY~1TeV
Uncertainty of ~30% is expected due to contamination of SUSY signal to control sample (depends on mass scale).

7. Inclusive study after discovery

8. Possible study after discovery
Understanding detector and background estimation are very important, however we should think about next step after discovery. We believe that we must/can do some study on SUSY in an inclusive way before precise measurement (I.e. nominal luminosity).
Assume :
Background estimation is successfully done.
See excess or peak in MissingET and/or Meff distribution.
Meff gives information on SUSY mass scale.
Try to find other observable sensitive to SUSY production and/or decay pattern.
Following characteristics are studied in this talk.
Lepton richness
B-jet richness
Multi-jet likeness

9. Once we see excess … L=1fb-1
m0=500GeV, m1/2=400GeV
tan=10, >0, A=0
L=1fb-1
Assume we can estimate background level successfully.
Signal
(Msusy~1TeV)
Need different cut optimization
to know SUSY event topology
to reduce dilution from background
background
Applying additional cut for better S/N
After background
subtraction
Meff > 1.5TeV

10. Inclusive SUSY signature
(A) Light sneutrinos/sleptons
~ ~ ~ ~ ~
qL -> 1+ / 20 -> +l / l +l
>>Lepton rich event
Major production qq ~~
(B) Direct decay cc ~~
~ ~
qR -> 10+q
>> Lepton poor event
g->tt
~ ~ qg ~~
(C) Light Stop/Sbottom
~ ~ ~ ~ ~
g -> t+t -> 2++b -> 20+W / 1++Z gg ~~
>> Lepton/b-jet rich event
(D) gluino production/decay
~ ~ ~
g -> n+ / n0 + qq
>> Multi-jet like event

11. Lepton Richness <Rlep1/0> ( = N(=1)/N(=0) )
m1/2=300 GeV(Msusy~0.8TeV)
m1/2=500 GeV(Msusy~1.3TeV)
MSUSY~0.8TeV
MSUSY~1.3TeV
Meff > 1TeV
Meff >1.5TeV
~ ~
~ ~
Br (qL->1+ + q)~65%
Br (1+ -> l1+ + l )~38%
Br (qR->10 + q)~64%
Br (1+ -> 1+ +  )~63%
~ ~
~ ~
~ ~
Br (g -> t + t )~95%
With statistic errors of signal ONLY (L=30fb-1)

12. b-jet Richness <Rptb> ( = S |pT(b-jet)|/ S |pT(jet)| )
Atlfast btag :
eb=60%, Rc=10, Ru=100
b-jet Richness
<Rptb> ( = S |pT(b-jet)|/ S |pT(jet)| )
m1/2=300 GeV(Msusy~0.8TeV)
m1/2=500 GeV(Msusy~1.3TeV)
MSUSY~0.8TeV
MSUSY~1.3TeV
Meff > 1TeV
Meff >1.5TeV
Ambiguity of lepton source can be solved with help of b-jet information.
With statistic errors of signal ONLY (L=30fb-1)

13. Multi-jet likeness
<R12> ( = pT(2nd)/ pT(1st) in each hemisphere )
Hemisphere Algorithm(HA) : reconstructed objects are assigned to
each initially-produced particle (hemisphere)
Usage of HA in iclusive study
More information available in hemisphere than in event.
Objects coming from upstream of SUSY decay chain are assigned
to correct mother by good efficiency
multi-jet like : R12 ->1
g -> 1+ + qq
mono-jet like : R12 ->0
qR -> 10+q

14. Multi-jet likeness
<R12> ( = pT(2nd)/ pT(1st) in each hemisphere )
m1/2=300 GeV(Msusy~0.8TeV)
m1/2=500 GeV(Msusy~1.3TeV)
MSUSY~0.8TeV
MSUSY~1.3TeV
gluino-gluino production is dominant
With statistic errors of signal ONLY (L=30fb-1)

15. Lepton Richness + background
Of course it is not easy in real life due to background...
<Rlep1/0> ( = N(=1)/N(=0) )
m1/2=300 GeV(Msusy~0.8TeV)
m1/2=500 GeV(Msusy~1.3TeV)
MSUSY~0.8TeV
MSUSY~1.3TeV
+ cut
+ cut + background
S/B < 1
Not loose information on SUSY event with cut (Meff>1.5TeV)
With statistic errors of signal + bkg (L=30fb-1)

16. b-jet Richness , R12 + background...
Need to optimize cut for better S/B.
m1/2=500 GeV(Msusy~1.3TeV)
<Rptb> ( = S |pT(b-jet)|/ S |pT(jet)| )
<R12> ( = pT(2nd)|/ pT(1st) )
+ cut
+ cut + background
S/B < 1
Not loose information on SUSY event with cut (Meff>1.5TeV)
With statistic errors of signal + bkg (L=30fb-1)

17. Summary and Outlook ~~ qq ~~ cc ~~ qg ~~ gg
Meff is very good observable which is sensitive to SUSY mass scale.
Background easily smears the peak.
It is very important to estimate background using real data.
Search good variables sensitive to SUSY production and decay pattern as Meff has good relation to SUSY mass scale
Lepton richness, b-jet richness and multii-jet likeness are tested.
Effect of background should be suppressed by optimizing cut or subtracting background with help of MC information
Try to find better observable and think about how to remove/suppress
of background effect. qq ~~ cc ~~ qg ~~ gg ~~

18. Backup Slides

19. Lepton Richness (3) <Rlep2/0> ( = N(=2)/N(=0) ) m1/2=500 GeV
MSUSY~1.3TeV
MSUSY~0.8TeV
BR(UPL->W1SS+UP)~65%
BR(W1SS->SLEP+L)~38%
BR(UPL->W1SS+UP)~64%
BR(W1SS->STAU+NUT)~63%
BR(GLSS -> ST + T)~95%
Signal ONLY (L=10fb-1) with statistic errors

20. Lepton Richness + background
Of course it is not easy in real life due to background...
<Rlep1/0> ( = N(=1)/N(=0) ) with MET>400GeV
m1/2=500 GeV(Msusy~1.3TeV)
m1/2=300 GeV(Msusy~0.8TeV)
MSUSY~0.8TeV
MSUSY~1.3TeV
+ cut
+ cut + background
With statistic errors of signal + bkg (L=30fb-1) together including
systematic uncertainties of bkg estimation

21. b-jet Richness , R12 + background...
Need to opimize cut for better S/B.
m1/2=500 GeV(Msusy~1.3TeV) with MET>400 GeV
<Rptb> ( = S |pT(b-jet)|/ S |pT(jet)| )
<R12> ( = pT(2nd)|/ pT(1st) )
+ cut
+ cut + background
With statistic errors of signal + bkg (L=30fb-1) together including
systematic uncertainties of bkg estimation