-
Notifications
You must be signed in to change notification settings - Fork 0
/
thesis.lot
65 lines (65 loc) · 15.1 KB
/
thesis.lot
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
\addvspace {10\p@ }
\addvspace {10\p@ }
\contentsline {table}{\numberline {2.1}{\ignorespaces Higgs decay modes and branching fractions, for a Standard Model Higgs with mass of 125.09 GeV \cite {YellowReport4}\relax }}{19}{table.caption.16}%
\addvspace {10\p@ }
\addvspace {10\p@ }
\addvspace {10\p@ }
\contentsline {table}{\numberline {5.1}{\ignorespaces Summary of nominal signal samples\relax }}{63}{table.caption.41}%
\contentsline {table}{\numberline {5.2}{\ignorespaces Summary of alternative signal samples\relax }}{64}{table.caption.42}%
\contentsline {table}{\numberline {5.3}{\ignorespaces Cross sections times branching ratio values used to normalize each production mode. The values correspond to the state-of-the-art predictions and are taken from the CERN Yellow Report \cite {YellowReport4}.\relax }}{64}{table.caption.43}%
\contentsline {table}{\numberline {5.4}{\ignorespaces Summary of nominal background samples\relax }}{65}{table.caption.44}%
\contentsline {table}{\numberline {5.5}{\ignorespaces Parameters used in the Higgs Characterization model in order to allow for a CP-variant Higgs coupling only to the top quark. The HWW coupling is fixed to its SM value by imposing cos$\alpha \nobreakspace {}\kappa _{SM}$=1. In the set of samples above the line, $\kappa _{t}$ is fixed to 1 and $\alpha $ is varied, while in those below, $\kappa _{t}$ is set to values not equal to 1. Pure CP-odd samples with $\qopname \relax o{cos}\alpha $ strictly equal to 0 cannot be generated due to numerical precision concerns, and thus a value approaching it ($10^{-6}$) and a corresponding value for kSM ($10^{6}$) are used.\relax }}{66}{table.caption.45}%
\contentsline {table}{\numberline {5.6}{\ignorespaces NLO cross-sections for the $t\bar {t}H$, $tHjb$, $tWH$, and $ggF$ processes for the different CP-scenarios (see parameters in Table\nobreakspace {}\ref {tab:MCsamples_Parameters}). In the samples above the line, $\kappa _{t}$ is fixed to 1 and $\alpha $ is varied, while in the samples below the line, $\kappa _{t}$ is not equal to 1.\relax }}{66}{table.caption.46}%
\contentsline {table}{\numberline {5.7}{\ignorespaces Normalized NLO cross-sections for the $t\bar {t}H$, $tHjb$, $tWH$, and $ggF$ processes for the different CP-scenarios, scaled using the K-factors and the value of BR$(H\rightarrow \gamma \gamma )$. In the samples above the line, $\kappa _{t}$ is fixed to 1 and $\alpha $ is varied, while in the samples below the line, $\kappa _{t}$ is not equal to 1.\relax }}{67}{table.caption.47}%
\addvspace {10\p@ }
\contentsline {table}{\numberline {6.1}{\ignorespaces Parameterization of Higgs cross-section dependence on $\kappa $ coefficients, from \cite {PhysRevD.101.012002}\relax }}{75}{table.caption.52}%
\contentsline {table}{\numberline {6.2}{\ignorespaces Parameterization of Higgs branching ratio dependence on $\kappa $ coefficients, from \cite {PhysRevD.101.012002}\relax }}{76}{table.caption.53}%
\contentsline {table}{\numberline {6.3}{\ignorespaces Simplified template cross sections times diphoton branching ratio for each of the STXS 1.2 truth bins.\relax }}{79}{table.caption.55}%
\addvspace {10\p@ }
\contentsline {table}{\numberline {7.1}{\ignorespaces Category boundaries which optimize the Poisson number-counting rejection significance of the CP odd scenario in the 12 hadronic and 8 leptonic categories.\relax }}{101}{table.caption.74}%
\contentsline {table}{\numberline {7.2}{\ignorespaces Comparison of statistical uncertainty with key systematics and CP-Odd vs. SM separation in each category. PS indicates parton showering uncertainty, calculated by subtracting the yields from the Herwig and the Pythia Monte Carlo samples.\relax }}{102}{table.caption.77}%
\contentsline {table}{\numberline {7.3}{\ignorespaces Significance metrics for the full twenty-category CP BDT categorization, calculated using event yields in the signal $m_{\gamma \gamma }$ region $125\pm 2$ GeV.\relax }}{103}{table.caption.79}%
\addvspace {10\p@ }
\contentsline {table}{\numberline {8.1}{\ignorespaces Best-fit parameter values for the DSCB Gaussian core and exponential tails in each of the 20 analysis categories.\relax }}{106}{table.caption.81}%
\contentsline {table}{\numberline {8.2}{\ignorespaces Spurious signal test results in the 20 analysis categories.\relax }}{107}{table.caption.82}%
\contentsline {table}{\numberline {8.3}{\ignorespaces A summary of the theory uncertainties used in the likelihood model.\relax }}{109}{table.caption.83}%
\contentsline {table}{\numberline {8.4}{\ignorespaces A summary of the experimental uncertainties used in the likelihood model.\relax }}{110}{table.caption.84}%
\contentsline {table}{\numberline {8.5}{\ignorespaces Relative QCD renormalization and factorization scale ($\mu _R$, $\mu _F$) and PDF uncertainties on the Standard Model $ttH$ sample.\relax }}{114}{table.caption.85}%
\contentsline {table}{\numberline {8.6}{\ignorespaces Relative QCD renormalization and factorization scale ($\mu _R$, $\mu _F$) and PDF uncertainties on the Standard Model $tWH$ Madgraph sample.\relax }}{115}{table.caption.86}%
\contentsline {table}{\numberline {8.7}{\ignorespaces Relative QCD renormalization and factorization scale ($\mu _R$, $\mu _F$) and PDF uncertainties on the Standard Model $tHjb$ Madgraph sample.\relax }}{116}{table.caption.87}%
\contentsline {table}{\numberline {8.8}{\ignorespaces Relative effect [(Varied-Nominal)/Nominal] of the underlying event and parton showering (UEPS) theoretical uncertainties for $ttH$, $tHjb$, $tWH$ and $ggF$.\relax }}{117}{table.caption.88}%
\contentsline {table}{\numberline {8.9}{\ignorespaces Generator uncertainties on $ggF$ (aMCnloPy8 $ggF$ - PowhegPy8 $ggF$)/(PowhegPy8 $ggF$) and $ttH$ (PowhegPy8 $ttH$ - aMCnloPy8 $ttH$)/(aMCnloPy8 $ttH$) in each analysis category. \relax }}{118}{table.caption.89}%
\contentsline {table}{\numberline {8.10}{\ignorespaces Observed and expected $t\bar {t}H$ and $tH=tHjb+tWH$ yields per category, calculated in the smallest $m_{\gamma \gamma }$ window containing 90\% of the fitted signal. Expected yields assume $\kappa _{t}=1$.\relax }}{123}{table.caption.94}%
\addvspace {10\p@ }
\contentsline {table}{\numberline {9.1}{\ignorespaces List of training variables used for the multiclass and binary BDTs.\relax }}{126}{table.caption.102}%
\contentsline {table}{\numberline {9.2}{\ignorespaces For each category, values of the expected Higgs signal ($S$) and background ($B$) within the smallest mass window containing 90\% of signal events, as well as corresponding estimates of the signal purity $f = S/(S + B)$ and the expected significance $Z = \sqrt {2( (S+B) \qopname \relax o{log}(1 + S/B) - S)}$.\relax }}{130}{table.caption.105}%
\contentsline {table}{\numberline {9.3}{\ignorespaces The choice of background function and the size of spurious signal uncertainties in the mass range 120 GeV to 130 GeV. $S$ is the maximum fitted spurious signal, $\delta S$ is its associated uncertainty, and $S_{ref}$ is the expected size of Higgs signal events. The $\zeta $ is the maximum fitted spurious signal yield when expanded to accomodate $2\sigma $ statistical fluctuations of the background templates. The ``*" in the function name means the function decision is made using the Wald Test because there are fewer than 100 events in the sidebands. \relax }}{135}{table.caption.112}%
\contentsline {table}{\numberline {9.4}{\ignorespaces The choice of background function and the size of spurious signal uncertainties in the mass range 120 GeV to 130 GeV. $S$ is the maximum fitted spurious signal, $\delta S$ is its associated uncertainty, and $S_{ref}$ is the expected size of Higgs signal events. The $\zeta $ is the maximum fitted spurious signal yield when expanded to accomodate $2\sigma $ statistical fluctuations of the background templates. The ``*" in the function name means the function decision is made using the Wald Test because there are fewer than 100 events in the sidebands. \relax }}{136}{table.caption.113}%
\contentsline {table}{\numberline {9.5}{\ignorespaces The impact of groups of systematic uncertainties on the total error on the measured cross section times branching ratio ($\Delta \sigma $), given as a fraction of the total measured cross section ($\sigma $).\relax }}{138}{table.caption.115}%
\contentsline {table}{\numberline {9.6}{\ignorespaces Best-fit values and uncertainties for $\sigma \times Br_{\gamma \gamma }$ in each of the five major production modes. The total uncertainties are decomposed into statistical and systematic components. Expected values are also shown for the cross-section of each process.\relax }}{140}{table.caption.119}%
\contentsline {table}{\numberline {9.7}{\ignorespaces Best-fit values and uncertainties for the cross-section times \ensuremath {H\to \gamma \gamma }\xspace \ branching ratio $(\sigma _i \times Br_{\gamma \gamma })$ in each STXS region. The total uncertainties are decomposed into statistical and systematic components. SM predictions are also shown for each quantity.\relax }}{141}{table.caption.121}%
\addvspace {10\p@ }
\addvspace {8pt}
\contentsline {table}{\numberline {A.1}{\ignorespaces Comparison of KLFitter and top-reconstruction BDT.\relax }}{158}{table.caption.130}%
\addvspace {8pt}
\contentsline {table}{\numberline {B.1}{\ignorespaces Figures of merit for the fifteen-category CP BDT categorization. The right-hand column shows that an alternative setup using four-vector training variables in the CP BDT achieves similar sensitivity.\relax }}{170}{table.caption.146}%
\addvspace {8pt}
\addvspace {8pt}
\contentsline {table}{\numberline {D.1}{\ignorespaces The final background modelling decision and the size of spurious signal uncertainties. The reported number here is the base SS yield, without the bias uncertainty applied; the spurious signal with the bias is used in \ref {tab:comp_smooth_unsmooth1} and \ref {tab:comp_smooth_unsmooth2}. The functional form is chosen using a relaxed spurious signal test applied to the unsmoothed templates. \relax }}{214}{table.caption.190}%
\contentsline {table}{\numberline {D.2}{\ignorespaces The final background modelling decision and the size of spurious signal uncertainties. The reported number here is the base SS yield, without the bias uncertainty applied; the spurious signal with the bias is used in \ref {tab:comp_smooth_unsmooth1} and \ref {tab:comp_smooth_unsmooth2}. The functional form is chosen using a relaxed spurious signal test applied to the unsmoothed templates. \relax }}{215}{table.caption.191}%
\contentsline {table}{\numberline {D.3}{\ignorespaces Comparison of the SS test (function and systematic uncertainty assigned) with nominal un-smoothed templates and smoothed ones. The functional form is chosen using a relaxed spurious signal test applied to the unsmoothed templates. \relax }}{216}{table.caption.192}%
\contentsline {table}{\numberline {D.4}{\ignorespaces Comparison of the SS test (function and systematic uncertainty assigned) with nominal un-smoothed templates and smoothed ones. The functional form is chosen using a relaxed spurious signal test applied to the unsmoothed templates. \relax }}{217}{table.caption.193}%
\contentsline {table}{\numberline {D.5}{\ignorespaces The final background modelling decision and the size of spurious signal uncertainties. The reported number here is the base SS yield, without the bias uncertainty applied; the spurious signal with the bias is used in \ref {tab:comp_smooth_unsmooth1} and \ref {tab:comp_smooth_unsmooth2}. The functional form is chosen using a non-relaxed spurious signal test applied to the smoothed templates. \relax }}{218}{table.caption.194}%
\contentsline {table}{\numberline {D.6}{\ignorespaces The final background modelling decision and the size of spurious signal uncertainties. The reported number here is the base SS yield, without the bias uncertainty applied; the spurious signal with the bias is used in \ref {tab:comp_smooth_unsmooth1} and \ref {tab:comp_smooth_unsmooth2}. The functional form is chosen using a non-relaxed spurious signal test applied to the smoothed templates. \relax }}{219}{table.caption.195}%
\contentsline {table}{\numberline {D.7}{\ignorespaces Comparison of the SS test (function and systematic uncertainty assigned) with nominal un-smoothed templates and smoothed ones. The functional form is chosen using a non-relaxed spurious signal test applied to the smoothed templates. \relax }}{220}{table.caption.196}%
\contentsline {table}{\numberline {D.8}{\ignorespaces Comparison of the SS test (function and systematic uncertainty assigned) with nominal un-smoothed templates and smoothed ones. The functional form is chosen using a non-relaxed spurious signal test applied to the smoothed templates. \relax }}{221}{table.caption.197}%
\addvspace {8pt}
\contentsline {table}{\numberline {E.1}{\ignorespaces Spurious signal means and widths for the three test functional-form distributions for a range of different template statistics.\relax }}{231}{table.caption.206}%
\contentsline {table}{\numberline {E.2}{\ignorespaces Spurious signal means and widths for the three test functional-form distributions for a range of different template statistics.\relax }}{242}{table.caption.217}%
\contentsline {table}{\numberline {E.3}{\ignorespaces Spurious signal means and widths for the three test functional-form distributions for a range of different template statistics.\relax }}{251}{table.caption.225}%
\contentsline {table}{\numberline {E.4}{\ignorespaces Spurious signal means and widths for all choices of fit functional-form, using the "low" template with the ExpPoly2 generating functional form, for a range of different template statistics.\relax }}{252}{table.caption.226}%
\contentsline {table}{\numberline {E.5}{\ignorespaces Spurious signal means and widths for all choices of fit functional-form, using the "medium" template with the ExpPoly3 generating functional form and the "high" template with the ExpPoly3 generating functional form, for a range of different template statistics.\relax }}{253}{table.caption.227}%
\contentsline {table}{\numberline {E.6}{\ignorespaces The median spurious signal extracted from a distribution of 1000 toys (in this study, we use the median rather than the mean to be robust to potential outliers; however, distributions are approximately Gaussian so the two generally do not disagree) for a variety of feature-injection widths.\relax }}{260}{table.caption.233}%
\contentsline {table}{\numberline {E.7}{\ignorespaces Spurious signal means and widths for the three test functional-form distributions for a range of different template statistics, with a signal feature injection that is approximately 3 GeV wide and 1\% of the template integral.\relax }}{268}{table.caption.241}%
\contentsline {table}{\numberline {E.8}{\ignorespaces Spurious signal means and widths for the three test functional-form distributions for a range of different template statistics, with a signal feature injection that is approximately 3 GeV wide. The template statistics are fixed at one million events, and the feature size is varied.\relax }}{273}{table.caption.246}%
\contentsline {table}{\numberline {E.9}{\ignorespaces Spurious signal means and widths for the three test functional-form distributions for a range of different template statistics, with a signal feature injection that is approximately 3 GeV wide and 1\% of the template integral.\relax }}{281}{table.caption.254}%