Bibliographic Citation | |
Full Text | ![]() |
---|---|
Title | A neural jet charge tagger for the measurement of the B/s0 anti-B/s0 oscillation frequency at CDF |
Creator/Author | Lecci, Claudia ; /Karlsruhe U., EKP |
Publication Date | 2005 Jul 01 |
OSTI Identifier | OSTI ID: 911837 |
Report Number(s) | FERMILAB-THESIS-2005-89 |
DOE Contract Number | AC02-76CH03000 |
Other Number(s) | TRN: US200801%%280 |
Resource Type | Thesis/Dissertation |
Research Org | Fermi National Accelerator Laboratory (FNAL), Batavia, IL |
Sponsoring Org | USDOE |
Subject | 46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND TECHNOLOGY; ALGORITHMS; ELECTRONS; FERMILAB COLLIDER DETECTOR; MUONS; NEURAL NETWORKS; OPTIMIZATION; OSCILLATIONS; PARTICLE IDENTIFICATION; PERFORMANCE; PROBABILITY; PRODUCTION; SIMULATION |
Related Subject | Instrumentation |
Description/Abstract | A Jet Charge Tagger algorithm for b-flavour tagging for the measurement of {Delta}m{sub s} at CDF has been presented. The tagger is based on a b-track probability variable and a b-jet probability variable, both obtained by combining the information available in b{bar b} events with a Neural Network. The tagging power measured on data is 0.917 {+-} 0.031% e+SVT sample; 0.938 {+-} 0.029% {mu}+SVT sample which is {approx}30% larger than the cut based Jet Charge Tagger employed for the B{sub s}{sup 0} mixing analysis presented by CDF at the Winter Conferences 2005. The improved power of the tagger is due to the selection of the b-jet with a Neural Network variable, which uses correlated jet variables in an optimal way. The development of the track and jet probability has profited from studies performed on simulated events, which allowed to understand better the features of b{bar b} events. For the first time in the CDF B group a Monte Carlo sample comprising flavour creation and additional b{bar b} production processes has been examined and compared to Run II data. It has been demonstrated that a Monte Carlo sample with only flavour creation b{bar b} production processes is not sufficient to describe b{bar b} data collected at CDF. The sample with additional processes introduced in this thesis is thus essential for tagging studies. Although the event description is satisfactory, the flavour information in the Monte Carlo sample differs with respect to data. This difference needs to be clarified by further studies. In addition, the track and the jet probabilities are the first official tools based on Neural Networks for B-Physics at CDF. They have proven that the simulation is understood to such an advanced level that Neural Networks can be employed. Further work is going on in this direction: a Soft Electron and a Soft Muon Tagger based on Neural Networks are under development as of now. Several possible tagger setups have been studied and the Jet Charge Tagger reached a high level of optimization. A further improvement of the tagging power can be achieved by combining the opposite side taggers in a single one, i.e. including particle identification in the track probability. A change of perspective might bring tagging at CDF to a higher performance: the traditional jet clustering could be abandoned in flavour of a track-based tag on the opposite side. This approach was successfully pursued in the DELPHI experiment with the BSAURUS project. A similar strategy is currently under investigation for the CDF experiment. The first studies on simulation are encouraging. The presented Jet Charge Tagger marks the advent of new flavour tagging techniques at CDF and it is going to greatly enhance the ongoing {Delta}m{sub s} analysis. |
Country of Publication | United States |
Language | English |
Format | Size: 174 pages |
System Entry Date | 2008 May 15 |
Document Discussions | |
Top |