This is a condensed description with plots for the analysis HIG-12-002.
To improve the sensitivity of the search, selected diphoton events are subdivided into mutually exclusive classes tagged in three different channels
The background model is obtained by fitting polynomial to the observed diphoton mass distributions in the dijet-tag channel. In the lepton-tag channel the model is obtained by fitting the exponential which is obtained from MC due to extremely low statistic in data. In the inclusive channle we use two physical observables, diphoton mass and piT (piT is defined as the ratio of the diphoton pT and mass), and build a 2D maximum likelihood model. We explicitely account for the correlations between the observables in the background model. We fit the the observed diphoton mass by a power law function with a linear correlation term in the power. The higher order correlation terms can be neglected. The observed diphoton piT is modeled with a sum of an exponential and a gaussian function centered at zero. 2D compared to 1D treatment is 30% more sensitive in terms of cross section measurement.
Image | File links | Description |
---|---|---|
pdf png | Figure 1: Background model fit to the mgg distribution for the three event classes of the exclusive channels. We also show a simulated signal (m_{H}=120 GeV). The magnitude of the simulated signal is what would be expected if its cross section were equal to the FP expectation. (top) dijet-tag: observed diphoton mass is fit with a 2nd order polynomial, (midle) muon-tag: observed diphoton mass is fit with an exponential with shape derived from irreducible background MC , (bottom) electron-tag: observed diphoton mass is fit with an exponential with shape derived from a combination of ireducible background MC and a control sample defined by requiring one of the photons to be matched with a track. For the top plot the statistical uncertainty bands shown are computed from the data fit uncertainty on the background yield. For the two bottom plots the dominant uncertainty comming from the nor- malization to the data is not shown. Signal of FP Higgs at m=120 GeV (red line) is overlaid for reference. Individual plots are given for each event class of this inclusive channel. | |
pdf png | a) Dijet-tagged diphoton event class. Fit with a 2nd order polynomial | |
pdf png | b) Muon-tagged diphoton event class. Fit from MC, normalized to data. | |
pdf png | c) Electron-tagged diphoton event class. Fit from MC and CS, normalized to data. |
pdf png | Observed piT distribution in inclusive channel after dijet-tagged and lepton-tagged events have been removed. We also show a simulated signal (m_{H}=120 GeV). The magnitude of the simulated signal is normalized to data. Discriminating power of this observable's shape between background and signal is evident and motivates the use of this observable together with the diphoton mass to build a 2D likelihood model. Not shown on this plot, but the function used to model the signal piT distributions is a sum of a Gaussian and Bifurcated Gaussian. | |
pdf png | Bias observed if correaltion not included in the 2D background model of the inclusive channel. Bias is defined the mean of the pull distribution of signal strength relative to statistical error of pseudo-experiment. |
pdf png | Figure 2: Background model fit projected to the mgg distribution for the four event classes of the inclusive channel after the dijet-tagged and lepton-tagged events have been removed. In the power we use a term linear with piT to account for the correlation between the two observables. We also show a simulated signal (m_{H}=120 GeV). The magnitude of the simulated signal is what would be expected if its cross section were equal to the FP expectation. Individual plots are given for each event class of this inclusive channel. | |
pdf png | a) The best inclusive channel event class defined by diphoton EBEB HighR9. | |
pdf png | b) The inclusive channel event class defined by diphoton EBEB LowR9. | |
pdf png | c) The inclusive channel event class defined by diphoton NotEBEB HighR9. | |
pdf png | d) The inclusive channel event class defined by diphoton NotEBEB LowR9. |
pdf png | Fig 3. Background model fit to the piT distribution for the four event classes of the inclusive channel after the dijet-tagged and lepton-tagged events have been removed. We also show a simulated signal (mH=120 GeV). The magnitude of the simulated signal is normailzed to data to better ilustrated the difference in shape between the signal and backround. |
pdf png | Figure 4: Exclusion limit on the cross section of a FG Higgs boson decaying into two photons as a function of the boson mass relative to the FP cross section, where the theoretical uncertainties on the cross section have been included in the limit setting. The limit is calculated in the asymptotic CLs approximation. The expected limit is provided for the combined channels and the indiviudal channels (dijet-tag, lepton-tag and inclusive (with events removed) is also shown for reference, with the following paticipation in the combined limit (dijet 65%, lepton 17%, inclusive 17%). | |
pdf png | Figure 5: Observed local p-values for the three channels combined is shown in black line. P-value is also shown for individual channels: dijet-tagged class, lepton-tagged classes, and inclusive 2D classes. | |
pdf png | Figure 6: The best fit signal strength, in terms of the standard model Higgs boson cross section, for the combined fit to the five classes (vertical line) and for the individual contributing classes (points) for the hypothesis of a FP Higgs boson mass of 126 GeV. The band corresponds to +- 1sigma uncertainties on the overall value. The horizontal bars indicate +-sigma uncertainties on the values for individual classes. |