Master thesis defence by Andreas Søgaard – University of Copenhagen

Forward this page to a friend Resize Print kalender-ikon Bookmark and Share

Discovery Center > Discovery calendar > Calendar 2015 > Master thesis defence ...

Master thesis defence by Andreas Søgaard

Title: Boosted Bosons and Wavelets

For the LHC Run 2 and beyond, experiments are pushing both the energy and the intensity frontier so the need for robust and efficient pile-up mitigation tools becomes ever more pressing. Several methods exist, relying on uniformity of pile-up, local
correlations of charged to neutral particles, and parton shower shapes, all in
y − φ space. Wavelets are presented as tools for pile-up removal, utilising
their ability to encode position and frequency information simultaneously. This
allows for the separation of individual hadron collision events by angular
scale and thus for subtracting of soft, diffuse/wide-angle contributions while
retaining the hard, small-angle components from the hard event. Wavelet methods
may utilise the same assumptions as existing methods, the difference being the
underlying, novel representation. Several wavelet methods are proposed and
their effect studied in simple toy simulation under conditions relevant for the
LHC Run 2. One full pile-up mitigation tool (‘wavelet analysis’) is optimised
and its impact on a few jet kinematic variables assessed in both toy and
official 13 TeV ATLAS MC. Finally, a mock search for new resonances in the
semi-leptonic WW channel is presented, focusing on the sensitivity improvements
achievable using the wavelet analysis. It is found that jet energy bias may be
removed and resolution improved by O(50%) for p⊥ = 300 GeV jets at ⟨μ⟩ = 40. Similarly, jet mass sensitivity for boosted boson jets may be improved by O(100%) under similar conditions. The latter has the effect of increasing the semi-leptonic diboson search
sensitivity at ⟨μ⟩ ≈ 25 by upwards of 10% for resonance masses
relevant for Run 2. Therefore, analyses at ATLAS—as well as e.g. CMS and
ALICE—may immediately benefit from employing wavelet-based methods, both for
searches and for other specialised tasks. The impact of using wavelet analyses
only increases with ⟨μ⟩, underlining their promise at current and
future hadron collider experiments.