Presentation of my analysis on squark searches

Signal simulations (at 189 GeV) :

for the sbottom :
50 points in the plane Msbottom vs Mneutralino have been simulated using SUSYGEN. The ISR are fully described with the Kleis package, and it also includes final state radiation (FSR).

for the stop :
Since the stop should hadronise, a generator written by M. Besancon (Saclay) has been used to simulate the stop fragmentation. It includes soft and virtual QCD correction, and a simple treatment of the ISR is perfomed. 97 points in the plane Msbottom vs Mneutralino have been simulated.

Detector :
All these points have been passed through the all DELSIM + DELANA programmes.

Mass windows :

Due to topological aspects of the signals, the analysis has been separated into 3 mass analysis windows : Dm is the difference between the squark and the neutralino masses.
  • Dm > 30 GeV
  • 10GeV < Dm <= 30 GeV
  • DM <= 10 GeV (Marc Besancon)
  • Stop analysis : DM <= 5 GeV (Orsay)

    • Signal separation :

    A neural network method has been applied to separate the bagrounds from signal. The training of the neural network is perfomed in each analysis mass window. The Monte-Carlo (signals and backgrounds) used for the trainings are then skipped from the end of the analysis.

    The strucure of the neural network is :

  • 10 input nodes
  • 10 hidden nodes
  • 3 output nodes

    • The variables are : 

    	  charged energy              (Ech)
	  transverse energy           (Et)
	  visible mass                (Mvis)
	  forward (30)/visible energy (E30/Evis)
	  softness                    (Ejet1/Mjet1+Ejet2/Mjet2)
	  angle between the 2 jets    (thj12)
	  Pt jet1-jet2                (sqrt(Ptjet1**2+Ptjet2**2)
	  Acoplanarity                (acop)
	  Fox-Wolfram 1 and 3         (H10+H30)
	  Global Jet btagging         (Probz02)

    Neural network training :
    It is perfomed at the end of the event selection.


    Monte-Carlo Node 1 Node 2 Node 3
    signal 1 0 0
    qq-gamma/4-fermions 0 1 0
    gamma-gamma(QCD,QPM) 0 0 1

    The final cut :
    It is perfomed on the signal output (node 1). The optimisation of this cut is perfomed using the CUTOPTIM programme.

    Signal efficiency :

    The efficiency at the end of the analysis is calculated for each simulated point. An interpolation is perfomed to obtain the efficiency at each point of the plane Mneutralino vs Msquark with a bining of 1 GeV (for the moment).