...

Document

by user

on
Category: Documents
11

views

Report

Comments

Description

Transcript

Document
Status Report sul b
Alessia Tricomi
Universita’ and INFN Catania
Contributi di M. Chiorboli, F.Palla, R.Ranieri, G. Segneri
Outline




Studi HLT: canali di benchmark
Ricostruzione gluino e sbottom
Algoritmi
Piani a breve termine
HLT Studies

Inclusive b HLT
Low luminosity and High Luminosity
– Efficiencies
– Timing
– QCD studies
– Benchmark channels:
– WH bb (Riccardo) tra un paio di trasparenze…
– H hh bbbb (vedi presentazione di Livio)
– ttH bbbbjjjj (vedi presentazione di Livio)
Exclusive b HLT
Bs  m m
 Bs J/ 
 Bs Ds p (presentazione Andrei)
Per una review sul lavoro fatto per il DAQ TDR vedi presentazione di
Fabrizio al btau di luglio

WH

Study of the signal
W(μnμ)H115(bb)
– mH=115 GeV/c2
– σ=1.31x10-10 mb

Updates on studies of
background μjj
– regional seed generator
– muon isolation
Wjj HLT strategy

Require L1 Muons (Global Muon Trigger)
-

Reconstruction of Tracks around L1 μ
-



Isolation: ≤2 Tracks in ΔRμ,Tk<0.2, only 1 with pt>2 GeV/c
Reconstruction of Primary Vertex with L1 μ track
Require L1 Jets
Reconstruction of Tracks around L1 Jet
-
-

L1 μ Quality flag > 1
Refine L1 jets with RecTracks (if possible)
2 leading jets inside Tracker acceptance |η1,2|<2.4
Apply bTag (simple IP2d or IP3d counting)
Fixed mistagging rate u=0.10
Efficiency of b-jet tag
b
Central
HighLumi online 0.66
HighLumi offline 0.66
LowLumi online 0.72
LowLumi offline 0.72
Forward
0.52
0.58
0.58
0.62
Signal efficiency
bTag 3 dim applied
– 1 of the 2 leading jets
tagged
– 2 tracks with SIP3d>2.
– Signal efficiency strongly
dependent from L1 Jet
Etcor(1,2)
•Single μ pt>20 GeV/c
•2 jets with Etcor>20 GeV
Ratebkg=1.5±0.7 Hz
εWH=13.2% (350±20 evts/year)
WH - conclusions
Channel W(μn)H115( bb) can be triggered!!!
– online Tracker track reconstruction

(regional, partial)
– online bTag

simple counting algorithm
– Reduction of background rate from initial 1x108 Hz
to 1.5 Hz with 13.2% signal efficiency (~350
events/year)
– Work in progress to improve timing and reduce bkg
rate
Sbottom and Gluino decay chain
p

b
~
 20
~
g
~
10
Event final state:
•  2 high pt isolated leptons OS
~

~
b
p


•  2 high pt b jets
• missing Et
b
10 fb-1 of SUSY events at point B produced with the
“ISAPYTHIA” generator
Detector simulation: fast simulation (CMSJET 4.801)
SM bkg: tt, Z+jet, W+jet, ZZ, WW, ZW
• 2 SFOS isolated leptons, pT>15 GeV, |h|<2.4
•  2 b-jets, pT>20 GeV, |h|<2.4
The goal is to try to reconstruct the sbottom
and the gluino
First step: 20  l+l- 10
p

b
~
 20
~
g
~

~
b
~
10

~
 20
~
10
~


b
p
Z+jet
Etmiss > 150 GeV
ttbar
The invariant mass of the same flavour opposite
sign lepton pairs shows a sharp edge, due the
kinematical properties of the three-body decay.
M(e  e- )  M(μ μ  ) (GeV)
With a cut on the missing energy we can
suppress most of the SM backgound.

Sbottom reconstruction
p
b
~
g

~
10
• assuming M(10) known
•Selecting events “in edge”
~
 20
~

~
b
p


b
• Combining the 20 obtained
from the two leptons with the
most energetic b-jet in the event
E(ll) > 100 GeV
> 100
EtmissE(ll)
> 150
GeVGeV
> 150
Eb-jetE1tmiss
> 250
GeVGeV
Eb-jet 1 s>3
> 250 GeV
b-tag:
M ~ 0


1
p ~ 0  1 
2
 M
 


p  
 

10 fb-1
b-tag: s>3
SM
65 GeV  M      81 GeV
~
M( bL )  496.0 GeV
~
M( bR )  524.0 GeV
Generated values
M(~
χ 20 b)  499.4  6.6 GeV
σ  47.6
Result of the fit
Gluino reconstruction
p

b
~ 0
2
~g
~

~
b
~ 0
1
Next step: reconstruct gluinosbottom
reconstructed sbottom associatedto the bjet closest in angle to it

b
p
E(ll) > 100 GeV
Etmiss > 150 GeV
Eb-jet 1 > 250 GeV
b-tag: s>3
SM
10 fb-1
M(~
g )  595.1 GeV
Result of the fit
M(~20bb)  585.1  11.1 GeV
s  50.1
We achieve a peak
resolution better than 10%
(still at CMSJET level!)
b-tagging in fast simulation
b-tagging performed with FATSIM:
a jet is a b-jet if it contains at least
two tracks with significance sip> sbcut
c + udsg
b
c
udsg
b
c
uds - g
Susy events
Point B
s ip
sip

cont
>2
69%
42%
>3
61%
30%
>4
54%
25%
>5
47%
22%
b-tagging effect on the peak recostruction: sbottom
s ip  2.2
s ip  4.2
s ip  3.0
s ip  5.0
b-tagging effect on the peak recostruction: gluino
s ip  2.2
s ip  3.0
s ip  4.2
s ip  5.0
b-tagging effect on the peak recostruction
With low purity the mass peaks are shifted to fake values
With too high purity (low statistics) we cannot reconstruct the peaks at all
This is a very simple algorithm. We would need a full simulation to
properly evaluate this effect
Algoritmi: likelihood ratio
Main step of the method:
 For each track i in a jet the significance Si is evaluated
 The ratio of the significance probability distribution
functions for b and u-jets is computed: ri= fb(Si)/ fu(Si)
 A jet weight is constructed from the sum of logarithms of
the ratio: W=Slog ri
 By keeping jets above some value of W, the efficiency for
different jet samples can be obtained
 The rejection will have to be optimised for each specific
bkg under study
 Several track quality cuts applied
Likelihood ratio: distribuzione Wjet
ORCA 6.1.1
b
c
uds
W
Likelihood ratio: distribuzione Wjet
Algo: ORCA 6.1.1
Calibrazioni: ORCA 5.3.4
Non ottimizzate!!!
Mistag = 5%
b60%, b60%*
Mistag = 10%
b78%, b68%*
Mistag = 20%
b84%, b72%*
* Track
counting DAQTDR (Gabriele)
Piani futuri

ONLINE:
– Il lavoro sui canali di benchmark continua e si stanno
investigando strategie per ridurre ulteriormente la
banda necessaria e il timing

OFFLINE:
– Finalizzare l’algoritmo della Likelihood ratio
– Implementare l’algoritmo con i leptoni
– Esiste già un framework rilasciato da Gabriele in cui
vanno integrati i nuovi algoritmi
– Novembre riunione “tecnica” per definire un
pacchetto “finale” utilizzabile da qualunque utente
“non esperto”
Fly UP