We have studied the application of different classification algorithms in the analysis of simulated high energy physics data. Whereas Neural Network algorithms have become a standard tool for data ...analysis, the performance of other classifiers such as Support Vector Machines has not yet been tested in this environment. We chose two different problems to compare the performance of a Support Vector Machine and a Neural Net trained with back-propagation: tagging events of the type e+e- -> ccbar and the identification of muons produced in multihadronic e+e- annihilation events.
The influence of fluid distribution on packed bed thermocline storage is studied experimentally on a 107 kWh prototype-scale setup. Three distributors are placed at the top of the tank, comparing ...uniform, central and peripheral geometries. A reference case is first considered and consists in a charge between 100 °C and 140 °C with an interstitial fluid velocity of 1.25 mm·s−1. Local analysis reveals a short-term influence of the distributor at the top of the packed bed. Radial temperature is most homogeneous when using uniform distribution, then central, then peripheral. However, the packed bed quickly homogenises radial temperature: the differences observed between distributors are erased in less than 6% of the total tank height. As a consequence, the longitudinal temperature profiles are unaffected by the distributor and the same utilisation rate of 81% is obtained for the three geometries. The coupling between fluid distribution and velocity is then studied: five charges are performed at velocities varying between 0.5 mm·s−1 and 1.5 mm·s−1 for each distributor. The influence of velocity is visible over the range studied: the utilisation rate decreases by about 2% due to advection, in agreement with numerical simulations. Fluid distribution, on the other hand, has no influence on global storage performance, showing the robustness of packed bed thermocline systems.
•Exp. study of the influence of fluid distribution on packed bed thermocline storage•Three distributor geometries are compared: uniform, central and peripheral.•Local performance: radial temperature most homogeneous with uniform distributor•Global performance: storage utilisation rate unaffected by the distributor used•Utilisation rate decreases with fluid velocity over the range tested experimentally.
Anomalous quartic couplings between the electroweak gauge bosons may contribute to the vv gamma gamma and qq gamma gamma final states produced in e+e- collisions. This analysis uses the LEP2 OPAL ...data sample at centre-of-mass energies up to 209 GeV. Event selections identify vv gamma gamma and qq gamma gamma events in which the two photons are reconstructed within the detector acceptance. The cross-section for the process e+e- -> qq gamma gamma is measured. Averaging over all energies, the ratio of the observed e+e- -> qq gamma gamma cross-section to the Standard Model expectation is R(data/SM) = 0.92 +- 0.07 +- 0.04 where the errors represent the statistical and systematic uncertainties respectively. The vv gamma gamma and qq gamma gamma data are used to constrain possible anomalous W+W- gamma gamma and ZZ gamma gamma couplings. Combining with previous OPAL results from the W+W- gamma final state, the 95% confidence level limits on the anomalous coupling parameters aoz, acz, aow and acw are found to be: -0.007 GeV^-2 < aoz/Lambda^2 < 0.023 GeV^-2 -0.029 GeV^-2 < acz/Lambda^2 < 0.029 GeV^-2 -0.020 GeV^-2 < aow/Lambda^2 < 0.020 GeV^-2 -0.052 GeV^-2 < acw/Lambda^2 < 0.037 GeV^-2 where Lamdba is the energy scale of the new physics. Limits found when allowing two or more parameters to vary are also presented.
We present the first experimental results based on the jet boost algorithm, a technique to select unbiased samples of gluon jets in e+e- annihilations, i.e. gluon jets free of biases introduced by ...event selection or jet finding criteria. Our results are derived from hadronic Z0 decays observed with the OPAL detector at the LEP e+e- collider at CERN. First, we test the boost algorithm through studies with Herwig Monte Carlo events and find that it provides accurate measurements of the charged particle multiplicity distributions of unbiased gluon jets for jet energies larger than about 5 GeV, and of the jet particle energy spectra (fragmentation functions) for jet energies larger than about 14 GeV. Second, we apply the boost algorithm to our data to derive unbiased measurements of the gluon jet multiplicity distribution for energies between about 5 and 18 GeV, and of the gluon jet fragmentation function at 14 and 18 GeV. In conjunction with our earlier results at 40 GeV, we then test QCD calculations for the energy evolution of the distributions, specifically the mean and first two non-trivial normalized factorial moments of the multiplicity distribution, and the fragmentation function. The theoretical results are found to be in global agreement with the data, although the factorial moments are not well described for jet energies below about 14 GeV.