Study of Higgs bosons in the WW final state and development of a fast calorimeter simulation for the ATLAS experiment

Monte Carlo simulations of physics events, including the detailed simulation of the detector, are an indispensable tool in the development and implementation of analysis strategies for high energy physics experiments. Despite massive worldwide computing resources, it is expected that only the equivalent of a few percent of the recorded ATLAS data can be simulated with the full ATLAS detector simulation, while most analyses require significantly more Monte Carlo events than data events for signal and important background processes. In the course of this thesis the fast calorimeter simulation FastCaloSim was developed that is combined in the Atlfast-II detector simulation package with full simulations of the other detector components. Atlfast-II reduces the event simulation time by a factor of 10-20 without severely compromising the simulation quality of critical variables. During the last two years, Atlfast-II was validated by the ATLAS collaboration and used to simulate several hundred million events, many of which could never have been produced with the full detector simulation. It is expected that Atlfast-II will also play an important role in the ATLAS simulation strategy for the upcoming data taking era. The inclusive Higgs boson H->WW->lnu lnu decay is one of the most promising channels for an early discovery of a Higgs boson. However, due to the neutrinos from the W boson decays, a complete reconstruction of the final sta te is not possible. In the standard analyses for this channel, the presence of a signal can only be established with the help of Monte Carlo simulations for all background processes and hence the influence of systematic uncertainties from the use of Monte Carlo simulations limits the discovery potential severely. Especially the loop induced WW background process gg->WW, that was first analysed within this thesis, is very similar to the Higgs boson signal and therefore needs to be known well to avoid a misinterpretation of background events as signal. In order to avoid a strong dependency of a Higgs boson discovery on Monte Carlo predictions, two background normalisation methods were developed in this thesis. They allow to isolate and measure the two dominant processes in the analysis, the Higgs boson signal and the qq->WW background, exclusively from recorded data without using Monte Carlo information. The method of the isolation of the Higgs boson signal was applied to the current Monte Carlo predictions of signal and background processes. It was shown that a discovery of a Standard Model Higgs boson is possible in the H->WW channel in the mass range ~160-170 GeV with data corresponding to an integrated luminosity of 10 fb^(-1) even without the use of any Monte Carlo information. In the early phase of this PhD thesis it was analysed how well the nature of an observed particle as Higgs boson can be established after an initial discovery. This requires a measurement of the coupling strengths of the observed particle to other Standard Model particles, because it is this coupling strength that generates the particle masses in the Higgs theory. Due to limited observables at the LHC, only relative measurements of coupling parameters are possible without assumptions. However, using moderate theoretical assumptions, absolute measurements of the Higgs boson coupling strengths to the most important heavy Standard Model particles, W, Z and the top quark, are possible with accuracies of ~50% or better using data corresponding to an integrated luminosity of 30 fb^(-1).