Advanced Techniques for Fast Simulation and Data Analysis in High Energy Physics (ASAP)
The discovery of the 125GeV Higgs boson by the LHC experiments has finally opened a new era in the exploration of the TeV scale. The physics programs of CMS and ATLAS aim far beyond the simple discovery, and vigorously pursue the full characterization of the newly discovered state and the full exploration of the TeV scale in search of new phenomena. A key lesson drawn from first two years of LHC running is that most probably first discoveries and then identification of new states/interactions will not be easy. On the one hand, model-independent searches in simple topologies such as single/multi lepton at high transverse momenta have not shown any hint of new physics so far. On the other, topologies with jets and/or missing transverse energies, much more challenging experimentally, do strongly depend on the underlying theoretical models so that efficiently identifying signal enhanced regions of the phase space is quite involved. In this context, multi-variate techniques have become more and more central in the analysis of data from hadron collider experiments, to maximally exploit the information available on the signal and on the backgrounds. Amongst the most advanced techniques and certainly the most powerful one from the theoretical point of view, the so called matrix element method stands out. The main goal of this proposal is to advance the use and the scope of the matrix-element method so to significantly extend the range of physics applications at the LHC to the search of new physics. First we aim at providing the experimental HEP community with complete and automatic simulation tools, such as MadWeight/MoMEMta and Delphes, that overcome the technical limitations of the method. Second we propose to test and apply the new tools to current analyses in signatures that involve final state leptons and b-jets. Finally, we explore new and original applications of the method to both model-dependent or model-independent searches of new physics at the LHC.
External collaborators: CMS collaboration.
Measurement of detector material with particles and application to the Tracker of the CMS experiment at the LHC
The amount and distribution of the material composing a particle detector that measures the trajectories of charged particles must be known with high accuracy for two main reasons: 1) avoid any bias in the measurements of the momentum of charged particles and 2) provide an accurate Monte Carlo simulation of the detector.
A novel method for measuring the material of a generic tracking apparatus has been developed. The method exploits the multiple scattering experienced by charged particles while they sail through the detector. The method relies on the precise position measurement of the crossing points provided by the tracking detectors. The method is completely general and can be applied to any experiment equipped with detectors with good enough space resolution.
The material of the CMS Silicon Strip Tracker has been measured with this technique to a precision at the level of 10%.
Search for new high-mass resonances decaying into di-muons with the CMS detector at the LHC
The CMS experiment is used to study the di-muon invariant mass spectrum. These spectra allow searches for high-mass unstable particles (resonances) to be performed in a yet unexplored high mass range.
High-mass resonances decaying into muon pairs are predicted in a number of models beyond the Standard Model of the fundamental interactions. Notable examples are heavy neutral gauge bosons predicted by grand unification theories, as well as gravitons arising in the Randall-Sundrum model of extra dimensions.
The first search for high mass resonances was published in JHEP by CMS using the data acquired in 2010. Updated results were produced using part of the 2011 dataset in Summer 2011. By combining di-electron and di-muon data, CMS excluded the existence of resonances predicted by a number of theoretical models with masses below about 2 TeV. These limits are the most stringent to date.
The UCL CP3 group contributed to these two early CMS publications by being one of the three teams of the CMS Collaboration that regularly analyzed new data, optimising the muon isolation criteria and conducting a full study of a mild excess observed in the low-mass region (at ~120 GeV) in both the di-electron and the di-muon channels.
Since 2012 the activity of the UCL-CP3 group is limited to the exploration of a matrix-element approach to this search. Preliminary results show that the exploitation of the full kinematical information of the di-muon events can give some sizable improvements over the classical one that uses just the di-muon invariant mass. In addition, the group develops a new algorithm for measuring the energy of TeV-muons (for details, please read the dedicated project). This algorithm is expected to bring improvements in both the di-muon and single muon+missing energy searches starting from 2014.
Simulation of the CMS silicon tracker
The Tracker Simulation group is responsible for the Geant-based simulation of the Pixel and Strip Tracker response, material budget and geometry description.
Members from CP3 are concentrating on various aspects of the validation with data. We also share the convenership of the group.
External collaborators: CMS tracker collaboration.
Validation of a fully automatic matrix element technique for CMS data analyses
The matrix element reweighting method attempts to compute the full likelihood of an observed event given a theoretical model. The method therefore measures the degree of compatibility of the event with the given model using as much information as available. MadWeight is a tool that fully automatize the computation of the event likelihood for any model implemented in MadGraph, by performing phase-space integration and providing a framework for taking into account the experimental resolution on the observed final state objects.
This project aims at validating the matrix element reweighting technique implemented in MadWeight on a number of benchmark searches. In some cases, the final goal is the efficient identification of background events. The final states that are being considered are: Zbb, single top, ttbar resonances and dimuon resonances.