Download presentation
Presentation is loading. Please wait.
Published byRosa Baker Modified over 9 years ago
1
Wahid Bhimji on behalf of the Edinburgh group
2
Apologies for the mostly text-based slides and for any misrememberings
3
2011 substructure (CA 1.2) WH data analysis: Chiara Debenedetti, Brendan O’Brien, Wahid, Robert Harrington Performed sensitivity determination; Comparison to data and Some small attempt at optimisation Extensive cross validation with UCL atlas-phys-higgs-bb-boosted@cern.ch list atlas-phys-higgs-bb-boosted@cern.ch Weekly meetings with more groups inc Z->ll and nunu All groups found little sensitivity gain from substructure Edinburgh effort largely stopped as Chiara moved onto HCP matters and Robert to H->4l CA fat/sub jet b-tagging scale factors 2011 and 2012 Ongoing (Brendan’s talk)
4
Using Split-filtered CA 1.2 Jets ala: Phys. Rev. Lett. 100, 242001 (2008) ATL-PHYS-PUB-2009-088 Using the tools provided in GroomedJetsD3PDGroomedJetsD3PD Initially added these to private version of SMD3PDs – but had to make them privately So (Robert) branched this into a HSG5D3PD and requested in automatic production :HSG5D3PD Also had a pT cut and lepton filter Now these branches are also in production SMD3PDs I haven’t checked all the branches that we used before are there (e.g, btagging / calibration) Nor what split-filtering parameters are used (see later)
5
not within R=1.2 of fat-jet
6
See BoostedH2bb. Good agreement between UCL and Edinburgh.BoostedH2bb Sensitivity similar anti-Kt vs Substructure Different events selected – but that is also because of things like jet veto used
7
Defaults set in BDRSFilteringTool.cxxBDRSFilteringTool.cxx declareProperty( "NSubjets", m_nsubjets = 3); declareProperty("maxR", m_maxR = 1.5); declareProperty("ktycut2", m_ktycut2 = 0.09); Tried varying from 1.2 to 0.6 declareProperty("massFraction",m_massfrac = 0.67); Tried varying from 0.57 to 0.76 m j1 / m j < mu declareProperty("minRfilt", m_minRfilt = 0.3); we set minSplitR = 0 on making D3PD declareProperty("minSplitR", m_minSplitR = 0.3)
8
Doesn’t make any difference really to either S/B or resolution Could also look to change Fat-jet R or filtering radius Remaking D3PDs each time is a pain – needed a better approach really… WH tt
10
Staff: Small fractions of (0.2) Wahid Bhimji and Victoria Martin and Andy Buckley Recruiting new (mostly physics) post-doc – but not here yet and maybe not H->bb. Students Working towards thesis : Brendan (1.0) Most active (based at CERN): Ben, Chiara (~1.0) : May/will work on unboosted analyses too. Starting: Tim Bristow (~0.5)
11
1. Revive 2011 results with modern limit fit (Chiara) 2. 2012 update to match inclusive (post-)HCP analysis: WH (Brendan, Chiara, Wahid) + Z->nunuH (Ben): 1. Perhaps towards a (LHCP?) public plot comparing sensitivity 2. Easier now as we work CERN code that has many of the common required corrections and Ben has added substructure code to that. 3. Obviously mH different so control regions need to be checked and flavour fits done – quite a lot of work. 3. Systematics: e.g. Btagging (Brendan) gamma+jet balance calibration (Chiara) 4. Using substructure variables with HSG5 MVA (Tim) 5. Optimisation of splitting filtering or other substructure technique on MC level or reclustering in D3PD 6. Confirming if at higher energies or luminosities (with higher pT) substructure has an advantage.
12
Developed some experience and tools for 2011 substructure analysis Interested in reviving this on 2012 data – finding a way to make it work. But manpower constraints and conflicting demands (from HSG5 and outside) So its somewhat depends on the enthusiasm of other groups here!
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.