1. REPORT ON THE LYON WORKSHOP
BLIND SUSY ANALYSIS
REPORT ON THE LYON WORKSHOP
OUTLINE
Introduction
Updates
First public results
Conclusions and Outlook 1
S. Muanza, CPPM Marseille 1

2. Introduction - Reminders -
The « SUSY Blind Analysis » is a SUSY challenge that was coocked
~2 years ago in the SUSY GDR and proposed to SUSY hunters
communities (LH 2007)
The aim was to test our methods and strategies to:
assess a SUSY signal at the LHC
measure some observables (end-points, thresholds, s, BR ratios,…)
constrain or determine the underlying model parameters
all of this in a BLIND ANALYSIS 2 2

3. Introduction - Cookbook -
A SUSY point was chosen in a model to be determined
All the SUSY & Higgs processes were produced simultaneously and mixed
with the SM background (MC truth infos removed)
CoM energy: 14 TeV, L=100 pb-1
Individual SM background samples were also made available
Format: simple root-tuples available at CCIN2P3 in HPSS, also made
available elsewhere using SRB
Website: 3 3

4. Introduction - Lyon Workshop -
Very few people dared to try & solve this SUSY challenge
The Lyon workshop was proposed to revive this project within the GDR
and to make myself available to answer any technical questions while some
analyzers were actually « getting their hands dirty » with the challenge
Place: IPN Lyon
Date: march 2009
Low attendance:
2 post-docs from Lyon => very informal workshop
3 people based at CERN expressed interest but finally did not have
time to try the challenge at the time of the workshop 4 4

5. Updates - Technicalities -
At CCIN2P3 the Linux platforms have evolved:
at the time I produced the samples:
ROOT v5.14, SL v3, gcc v3.2.3, 32-bits
currently:
ROOT v5.14, SL v4, gcc v3.4.6, 64-bits platform (supports 32-bits)
h2root does not work anymore!
solution connect to ccalisl432.in2p3.fr :
works interactively
no batch support for 32-bits code 5 5

6. Updates - Revealing a Clue -
Here’s the main experimental trick I introduced to differentiate the
pseudo-data sample from the SM background samples:
PsD1: was produced using the option KEYLUM=2 (high lumi)
BKGD: was produced using the option KEYLUM=1 (low lumi)
effects:
I chose not to change the trigger nor the jets thresholds
the only effects are different resolutions
can find it in the code (tarball on the webpage):
atlfast /atlfast/atlfast.F
FUNCTIONS: RESHAD, RESEL, RESPHO 6 6

7. Updates - Revealing a Clue -
This is propagate to the mET calculation:
SUBROUTINE: MAKMIS
Example: Jet Resolutions
|h|<3.0:
|h|>3.0:
for DR=0.4, Epile-up=7.5 GeV 7 7

8. Updates - New Analysis Code -
I updated the proposed analysis code:
a bug fix on a Z(qq)+jets cross section (thanks to Patrice Verdier)
slightly different access to the sub-processes root-tuples:
each sub-process in its own sub-dir
the analysis code automatically makes TChain w/ all the root-tuples
in there
it does not depend on a specific root-tuple name anymore
the user will have to copy the sub-dir tree in his/her working area
the code is set to analyze the skimmed (mET > 150 GeV) root-tuples
the code can perform further skimmings 8 8

9. Updates - New Analysis Code: Outputs -
Arbitrary Cuts:
Trigger OR
mET > 250 GeV
Njets > 3 9 9

10. First Public Results - P. Verdier -10 10

11. Conclusions & Outlook
Apologies I did not find the time to update the website (end of this week)
I encourage all of you to try this SUSY challenge
Especially to measure some features of the SUSY+Higgs excess
And to hand it over to SUSY fitters (Sfitter, Fittino, OSET methods,…)
I’ll « open the box » at the time of the Les Houches 2009 meeting
And give the bottle champagne to the author of the
closest answer! 11 11