Matthew Tamsett An Introduction to my Recent Work for the University of Oklahoma 1
Overview This talk aims to: Give an overview of my experience as a physicist. Demonstrate my scientific and technical abilities and my enthusiasm for the subject. Highlight my key achievements. Illustrate how I would contribute to your group. Table of Contents Academic background. ATLAS. My work with the Trigger. My work on SUSY and top. My work on exotics. My research interests. 2
Matthew Tamsett Education MSci Physics, University College London 2002-2006, 2:1 Honours. Final Year Project: Use of Artificial Neural Networks within the ATLAS Electron Trigger. Ph.D. in Particle Physics, Royal Holloway, University of London 2006-today. Thesis Title: Triggering on Leptons for Beyond the Standard Model Searches ATLAS. Supervisor: Dr A. De Santo. Advisor: Dr P. Teixeira-Dias. Expected submission date: Next week! Viva: 24th September. STFC funded. Conference and Plenary Presentations International Conference on Particles and Nuclei (PANIC), Eilat Isreal, Nov 2008 (poster presented). ATLAS Trigger Workshop, Beatenberg, Feb 2009. Institute of Physics HEP Particle Physics 2009, Oxford, April 2009. ATLAS UK Annual Physics Meeting, Oxford, Jan 2009, Jan 2008. Other Presentations SUSY plenary, SUSY sub-group, SUSY-exotics UK, e/γ. e/γ trigger, exotics plenary, exotics sub-group (30+ talks total). 3
Matthew Tamsett II Publications Expected performance of the ATLAS Experiment, CERN-OPEN-2008-20, December 2009. Named contributor to: Reconstruction and Identification of Electrons. The ATLAS Trigger for Early Running. Physics Performance Studies and Strategy of the Electron and Photon Trigger Selection. Dilepton Resonances at High Mass. Electron Trigger Efficiency Determination from Data and its Application in a Supersymmetric Environment. ATL-PHYS-INT-2010-025, March 2010. Lead author. Triggering on High-Mass Dielectron Resonances. ATL-COM-PHYS-2009-641, December 2009. Lead author. Trigger Egamma 2009 Collision Studies, ATL-COM-DAQ-2010-032, April 2010. Miscellanea Teaching: 3rd year undergraduate Particle Physics discussion class 2007, 2009. Outreach: CERN International Masterclass 2008, RHUL Masterclass 2007, 2010. Bursaries and Grants: STFC funded Ph.D, IOP student conference bursary. 4
The LHC and ATLAS 5
The Large Hadron Collider LHCb ATLAS ALICE CMS LHC LHC CoM energy Luminosity (cm-2s-1) Bunch crossing Overlaid events Stored energy Design Values 14 TeV Low: 2x10 33 High:10 34 25 ns 23 @ 10 34cm-2s-1 362 MJ/beam SPS 27 km long, superconducting particle accelerator. Beams of counter rotating protons collide at four interaction points around the LHC ring. Four main experiments: ATLAS, ALICE, CMS and LHCb. Began operation on 20th November 2009 at 450 GeV per beam. 7 TeV collisions began on 20th March 2010 and have been ongoing since then. Peak Luminosity ~ 4 x 1030cm-2s-1 Currently the highest energy collider experiment in the world! 6
The ATLAS Detector Magnetic field Tracker ATLAS 2T solenoid + toroid (0.5 T barrel 1 T endcap) Si pixels, strips + TRT pt /pt 5x10-4pt + 0.01 EM calorimeter Pb +Lar σe/e 10%/ E + 0.007 Hadronic calorimeter Fe+scint, Cu +Lar (10 ) σe/e 50%/ E + 0.03 GeV Muon System σpt/p T 2% @ 50GeV to Trigger 10% @ 1TeV (ID+MS) Hardware Level 1 + Region of Interest-based HLT A Torodial LHC ApparatuS. One of two general purpose experiments at the LHC. Comprised of several highly granular, hermetic subdetectors over η < 4.9. Since LHC collisions began, ATLAS has collected ~1 pb-1 of data at s = 7 TeV. Very high data collection efficiency ~99%. 7
The Inner detector My work has primarily focused on electrons which are reconstructed using a combination of ID tracks and calorimeter clusters. The inner detector (ID) covers the η region < 2.5 and is composed from: The pixels; the semiconducting silicon micro-strip layers (SCT); and the transition radiation tracker (TRT). Measures charged particle trajectories using discrete high resolution positional measurements. Readout channels: 80.4M (pixels) + 6.3M(SCT) + 0.3M (TRT) Resolution: σpt / PT = 0.05% PT + 1% 8
The Calorimeters The calorimeter system covers the η range < 4.9 and is comprised from: The Electro-Magnetic Calorimeter (ECAL) and the Hadronic Calorimeter (HCAL). Calorimeters measure the energy of incident particles by causing them to shower in layers of dense mediums and then measuring this energy in interleaved scintillating sampling regions. An average electron will deposit the vast majority of its energy in the ECAL. The ECAL is ~26 X0 deep and has 170k readout channels, its resolution is: σe / E = 10.1% / E + 0.2% 9
The Trigger System 10
The Trigger Maximum raw event rate seen in ATLAS at design luminosity ~ 40 MHz (from the 25 ns bunch crossing). Trigger system needs to reduce this to the 100200Hz rate able to be written to tape. Rejection factor to be achieved at maximum luminosity ~ 105. Interesting cross sections often at least ~106 times smaller than total cross section. Trigger must remain efficient for rare signal processes. In one second at design luminosity & energy: 40,000,000 bunch crossings ~1,000 W events ~ 500 Z events ~ 10 top events ~ 9 SUSY events? ~ 0.1 Higgs events? 200 events written out Must ensure the correct 200 events are written out. 11
The Trigger Sequential three tier trigger system. Custom hardware Level 1 (L1). Hardware based. Calorimeter and muons only. Latency 2.5 μs 40 MHz 75 khz. Software High Level Trigger (HLT), comprised of L2 + EF. L2: 500 farm nodes Only utilises Regions of Interest (RoIs) seeded by L1. Fast, software based reconstruction. 100 khz 3.5 khz. EF: 1,600 farm nodes Seeded by L2. Potential to reconstruct the entire event. Offline algorithms. 3.5 khz 200 Hz. 12
The EM L1 Trigger L1 electron trigger applies very simple hardware cuts in electromagnetic and hadronic calorimeters. Uses analogue sum of calorimeter towers with coarse granularity. x = 0.1 x 0.1 Energy signals are converted in ET measurements by calibrated look up tables. EM candidates are identified by 2 x 2 trigger tower sliding window which searches for an ET maxima exceeding the trigger threshold within a 1 x 2 or 2 x 1 region. Jet backgrounds are primarily rejected by the ET cut. Further rejection can be obtained by requiring isolation criteria on the amount of energy surrounding the central cluster. 13
The Electron HLT Trigger The electron HLT is seeded by the L1 EM RoI. Guided algorithms then are run in seeded and stepwise manner to test electron specific hypotheses. The L2 trigger: extracts the calorimeter clusters using the full calorimeter granularity in an RoI 0.075 x 0.175 in η x φ and closely matched tracks. Backgrounds are rejected by fast hypothesis tests based on the shower shape and track quality. Region of Interest mechanism means only 1 4% of detector information is needed. The EF is seeded by a L2 pass, it then reuses the offline algorithms to reconstruct electrons and applies ET and isem hypotheses. 14
Commissioning the Trigger with 900 GeV data The first LHC collisions were recorded in late 2009. Between the 6th and the 15th of December around ~9 μb-1 of data was delivered to and recorded by ATLAS. During this time I was involved in the commissioning of the electron trigger. Approximately 730 electrons were observed in 903,309 MBTS triggered events. 15
Commissioning the Trigger with 900 GeV data During this time the L1 trigger was active, with the most inclusive minimum bias trigger: MBTS_1. The L1 EM threshold EM3 was run online in a rejection mode, with the HLT e5_nocut triggers active in pass through modes. Within the dataset 189 electrons were present in events which pass L1_EM3. The inclusiveness of MBTS_1 allows for the unbiased extraction of the L1_EM3 efficiency. For electrons with ET > 7 GeV, I measured the efficiency to be 93 ± 5%* This agrees well with the MC prediction of 95 ± 2% The small bump in efficiency below 3 GeV in MC is due to secondary energy clusters. * using Bayesian error calculation. 16
Commissioning the Trigger with 900 GeV data I further studied the online shower shapes and resolutions and compared these to MC. B G K E A FJ H I D C In general good agreement was seen between the data and MC, and the trigger itself behaved remarkably well. I helped to identify resolution issues associated with large energy deposits in the 0th and 3rd layers of the ECAL, as well as φ resolution asymmetries at L1 between e+ and e-. ATL-COM-DAQ-2010-032 17
The Tag and Probe method A large portion of my Ph.D. work involved the methodologies for the extraction of trigger efficiencies from data. One such method is the famous Tag and Probe method. The Tag and Probe method. Sample defined by: Z e+e- reconstructed (from offline e+e-) + 1 e trigger signature satisfied (tag). Trigger efficiency determined by counting in how many cases the second e± satisfies the trigger requirements (probe). The T&P method provides a good, clean sample of electrons by utilising the 'standard candle' Z ee process. Minimizes reliance on Monte Carlo. Exactly the same technique can be used in the Z μμ channel. 18
The Tag and Probe method The T&P method determines efficiencies well across all studied variables in the Z e+e- sample. Global T&P efficiencies agree with the MC result to within 0.2%. 100 pb-1 at 7 TeV: ~77,000 Z e+e- global trigger efficiency to < 0.01%. Or differential in 100 bins to around 0.1% in each bin. ATL-PHYS-INT-2010-025 I worked extensively with the TDT and contributed to ARA tutorials and HN. 19
My work on Supersymmetry and tt 20
SUSY motivation The trigger is a tool to study physics. The primary focus of my work was on leptons in Beyond the Standard Model theories, in particular: Supersymmetry (SUSY). SUSY proposes the introduction of a new fundamental symmetry involving the transformation of bosonic fields in fermionic fields and vis-versa. Such a theory requires at least a doubling of the observed particle spectrum as well as the extension of the Higgs sector. The most general form of the Minimal Supersymmetric Standard Model (MSSM) introduces fast proton decay, therefore most SUSY models include R-parity conservation. The sparticle content of the MSSM. 21
SUSY motivation By doing this SUSY offers elegant explanations of: Cold dark matter, as required by astronomical observations. The anomalous magnetic moment of the muon: 3.2 σ. As well as a solution to the infamous hierarchy problem, a possible mechanism for 3-force unification, and links to theories of quantum gravity and string theory. 22
SUSY at the LHC SUSY could be observed at the LHC via the creation and cascade decays of sparticles. Typically this results in final states rich in hadronic activity, with sizeable missing ET and one or more leptons. ATLAS limits already competitive with Tevatron results! Leptonic modes provide an extra handle for triggering and for the suppression of SM backgrounds. It is on leptonic SUSY modes my work has focused. Particularly trileptons and tt backgrounds. Numerous presentations given, e.g.: http://indico.cern.ch/contributiondisplay.py?contribid=7&confid=66331 23
Top quark processes I ve also spent a significant amount of time working on Top physics. Top signals are a common background to many leptonic SUSY signals. As well as a fascinating probe of the SM in their own right. Like SUSY, top signatures typically contain significant hadronic activity, missing ET and leptons. The similarity of top events to SUSY means that top quark decays represent and excellent benchmark process for the evaluation of ATLAS performance in SUSY-like environments. 24
Electron trigger systematics (according to the CSC book) I focused on electron identification in SUSY events as this is the most interesting area to study isolation. SUSY multilepton search systematics from CSC note 10fb-1 at 14TeV Uncertainties scaled from EW boson cross section measurement CSC note. Uses tag and probe to determine the Global trigger efficiency (0D) ie not differential in ET and. Top uses more sophisticated differential efficiency correction in ET and (2D). Systematics O(1%) from 100pb-1 Tag and Probe data. Gray histogram: with no trigger Black histogram: after trigger. Points: after trigger with 2D T&P corrections 25
Electron trigger efficiency in busy events Sample Z ee ttbar SU3 SU4 Efficiency (%) 97.1 0.1 92.9 ± 0.1 87.7 0.4 90.6 ± 0.3 Systematic differences between samples of up to 10% Efficiency as determined in CSC notes is not representative of SUSY events. Main factors that can affect trigger efficiency: 1. Fakes and non-prompt electrons: fakes are not electrons, non-prompt electrons typically originate from jets. 2. Kinematics: Trigger efficiency differs over kinematic (e.g. ) spectra. If different samples have different distributions then the trigger efficiency integrated over this will show differences. 3. Event topology: If an event has more jets, these will interfere with the isolation and identification of electrons. 26
Fake electrons tt tt Fake and non-prompt electrons exhibit noticeably lower efficiencies. These account for ~5% of the efficiency difference observed previously. I have worked on a method to estimate the contribution of fake and non-prompt electrons in SUSY and tt samples based on the jet and b-jet composition of the sample: http://indico.cern.ch/contributiondisplay.pycontribid=10&confid=85723 To do this I re-wrote the MC truth classifier tool in ARA. 27
Looking only at truth matched electrons Sample Z ee ttbar SU3 SU4 Efficiency (%) 97.1 0.1 93.9 ± 0.1 92.0 0.3 92.6 ± 0.2 Systematic differences between samples reduced to 5% 1-D and Global efficiencies in Z ee samples are not the same as SUSY or tt samples. 28
2-D efficiencies In order to compare efficiency easily in more than 1-D we switch to a parametrisation method. This method is similar to that which will be used to evaluate trigger efficiency with data. Take Z ee, work out 2-D (ET vs ) efficiency then apply to sample and compare to MC. tt SU3 Differences up observed of up to 4% 2-D efficiencies in Z ee samples are not the same as SUSY or tt samples. This means the CSC method will wrongly determine the trigger efficiency by 4%. 29
Isolation Z ee SU4 The remaining discrepancy can be described by event topology, specifically: isolation. SUSY and tt are much 'busier' environments than Z ee. Proximity of energetic objects leads to a decrease in trigger efficiency. 30
Isolation ET and are not the only variables that affect trigger efficiency. tt There is a turn-on in isolation as well. 2-D parametrisation in ET and is not accounting for this. 31
Extensions of Tag and Probe to busy events Z ee efficiency is computed in 3-D (ET vs vs etcone40) then applied to other samples and compared to MC. tt Agreement to within 1%! Without extension to 3-D trigger efficiency measurements from data are not valid in busy environments. I was the first person to show this conclusively. ATL-PHYS-INT-2010-025 This was presented in numerous SUSY, top, e/gamma and trigger meetings and earned me an invitation to talk at the trigger workshop in Beatenburg in 2009: tt http://indico.cern.ch/conferenceotherviews.py? view=standard&confid=44626 32
Scans of msugra grid space SUSY benchmark points are useful for establishing baselines for the performance of ATLAS, however this is not necessarily how nature may chose to manifest herself. Avoids too much model dependence. I ve studied scans of msugra grid space, contributing to the multi-lepton search mode strategies as well as evaluating leptonic production mechanisms and the ATLAS performance. In addition these studies helped to determine the trigger strategy for leptons in SUSY events. The studies of the large number of grid points involved the extensive use of complex scripting and analysis automation. 33
Evolution of some example variables 34
Trigger efficiency Each point in these grids was produced using ATLFAST-II, which lacks any trigger simulation. However it is still desirable to know how the trigger will perform in these environments. The parametrisations developed by me and introduced previously allows an estimate of trigger efficiency to be made. Single electron trigger. Single muon trigger. 35
Trigger efficiency Each point in these grids was produced using ATLFAST-II, which lacks any trigger simulation. However it is still desirable to know how the trigger will perform in these environments. The parametrisations developed by me and introduced previously allows an estimate of trigger efficiency to be made. Single electron trigger. Single muon trigger. 36
Trilepton events Similarly, this method allows for the evaluation of event level trigger efficiencies. single objects triggers: trileptons single and double object triggers: trileptons Efficiencies or above 90% are observed for all grid points. This can further be enhanced via the use of orthogonal trigger streams. 37
My work on Neutral BSM Resonances 38
Neutral BSM resonances High mass di-electron resonances are predicted by several extensions to the standard model. Very high energetic electrons. Fit plateau efficiency from low mass Z resonance and extrapolate results to high ET. CERN-OPEN-2008-20 and ATL-COM-PHYS-2009-641 This work earned my an invitation to present a poster at the PANIC 08 conference. 39
My Research Interests 40
My research interests My primary research interests are focused on new physics, top quarks and the trigger. Top Physics Properties of the top quark. Leptonic (including tau) performance and the measurement of object performance from data. Measurements in events of complicated topology. Novel signal extraction mechanisms included multi-variant analysis techniques and derived event variables. Supersymmetry Early data SUSY searches. The model dependence of the SUSY decay phenomenology and how this impacts specific search channels. Triggers Establishment of the performance of the trigger system and measurement of efficiency from data. Rejection strategies for high luminosity. Miscellanea BSM physics including: Models involving multi-dimensional theories of gravity, and connections with cosmological constraints. Multi-variant analysis techniques in general. Jets and QCD. B-jets. Computing methods at ATLAS. I m very keen to expand into new areas and would welcome the opportunity to work on something outside of this scope. 41