...
Hadronic flavor tagging
The aim of this project research program is to develop techniques to identify the flavor of hadronic jets, in particular: quark/gluon separation, tagging b-jets from gluon splitting, and the identification of highly boosted (bb) jets.
...
The method developed uses a Neural Network (NN) trained with tracking and calorimeter information based on the fact that merged (bb) jets are wider, and have larger calorimeter-cluster and track multiplicity than single b jets.
We plan to use this neural network to select control samples with different gluon splitting content to test data/Monte Carlo agreement, reject b-tag background events, and improve the estimation of b-tag backgrounds.
highly boosted (bb)-jets
Very recently it has been proposed that (low mass) Higgs production in association with W boson could be efficiently identified at the LHC if the Higgs is produced with high transverse momentum. Highly boosted Higgs decaying to b-quarks produce a signature of merged (bb)-jets, which can be distinguished from generic QCD jets.
The original paper proposes a method based on jet sub-structure to separate (bb) jets from QCD jets. We would like to investigate the use of a neural network technique, similar to the gbb NN developed to separate gluon->(bb) jets from b-jets, and study the kinematic differences of merged (bb) jets from gluon splitting and merged (bb) jets from Higgs.
Quark/gluon separation
We are working in the development of a neural network based algorithm that assigns jets a probability that they originated from quarks or gluons, based on Monte Carlo simulations.
...