1: \documentstyle[times,pramana,epsfig,floats]{ias}
2:
3: \begin{document}
4: \newcommand{\stp}{\tilde t_1}
5: \def\chio{\textstyle\raise.4ex\hbox{$\textstyle\tilde\chi{^0_1}$}}
6:
7: \begin{titlepage}
8:
9: \thispagestyle{empty}
10: \def\thefootnote{\fnsymbol{footnote}} % symbols for footnotes
11:
12: \begin{center}
13: \mbox{ }
14:
15: \end{center}
16: \begin{flushright}
17: \vspace* {-2.0cm}
18: \Large
19: \mbox{\hspace{10.2cm} hep-ph/0609016} \\
20: \end{flushright}
21: \begin{center}
22: \vskip 1.0cm
23: {\Huge\bf
24: Small Visible Energy Scalar Top
25: }
26: \vspace{2mm}
27:
28: {\Huge\bf
29: Iterative Discriminant Analysis
30: }
31: \vskip 1cm
32: {\LARGE\bf A. Sopczak$^1$, A. Finch$^1$, A. Freitas$^2$, \\
33: C. Milst\'ene$^3$, M.~Schmitt$^4$
34: \smallskip
35:
36: \Large $^1$Lancaster University, UK; $^2$Zurich University, Switzerland; \\
37: $^3$Fermilab, USA; $^4$Northwestern University, USA}
38:
39: \vskip 2.5cm
40: \centerline{\Large \bf Abstract}
41: \end{center}
42:
43: \vskip 2.cm
44: \hspace*{-0.5cm}
45: \begin{picture}(0.001,0.001)(0,0)
46: \put(,0){
47: \begin{minipage}{\textwidth}
48: \Large
49: \renewcommand{\baselinestretch} {1.2}
50: Light scalar top quarks with a small mass difference with respect to the neutralino mass are
51: of particular cosmological interest. This study uses an Iterative Discriminant Analysis
52: method to optimize the expected selection efficiency at a International Linear Collider (ILC).
53: \renewcommand{\baselinestretch} {1.}
54:
55: \normalsize
56: \vspace{5.5cm}
57: \begin{center}
58: {\sl \large
59: Presented at the 2006 International Linear Collider Workshop - Bangalore,
60: India, \\
61: to be published in the proceedings.
62: \vspace{-6cm}
63: }
64: \end{center}
65: \end{minipage}
66: }
67: \end{picture}
68: \vfill
69:
70: \end{titlepage}
71:
72: \newpage
73: \thispagestyle{empty}
74: \mbox{ }
75: \newpage
76: \setcounter{page}{1}
77:
78:
79: \mark{{Linear Collider Workshop}{A. Sopczak, A. Finch, A. Freitas, C. Milst\'ene, M.~Schmitt}}
80: \title{\Large Small\,Visible\,Energy\,Scalar\,Top\,Iterative\,Discriminant\,Analysis}
81:
82: \author{A. Sopczak$^1$, A. Finch$^1$, A. Freitas$^2$, C. Milst\'ene$^3$, M.~Schmitt$^4$}
83: \address{$^1$Lancaster U., UK, $^2$Zurich U., Switzerland, $^3$Fermilab, USA, $^4$Northwestern U., USA}
84:
85: \abstract{
86: Light scalar top quarks with a small mass difference with respect to the neutralino mass are
87: of particular cosmological interest. This study uses an Iterative Discriminant Analysis
88: method to optimize the expected selection efficiency at a International Linear Collider (ILC).
89: }
90:
91: \maketitle
92: \vspace*{-7mm}
93: \section{Introduction}
94: \vspace*{-3.5mm}
95: The search for scalar top quarks and the determination of their parameters in the framework of
96: Supersymmetric models are important aspects of the Linear Collider physics programme.
97: The lightest neutralino with a small mass difference ($\Delta m$) to the scalar top quark
98: is a candidate for dark matter in the Universe~\cite{Balazs:2004bu}. The dark matter rate
99: could be measured at a Linear Collider~\cite{carena,sopczak}.
100: This study applies an Iterative Discriminant Analysis (IDA)~\cite{ida} to a scenario
101: involving a 122.5~GeV scalar top and a 107.2~GeV neutralino at $\sqrt{s}=260$~GeV which is near threshold.
102: This center-of-mass energy will be used for the scanning phase of the ILC operation and a luminosity of
103: 50~fb$^{-1}$ is assumed.
104: The event generation and detector simulation has been performed with unpolarised beams as for a sequential
105: cut-based analysis~\cite{carena}. In particular, a vertex detector concept of the Linear Collider Flavour
106: Identification (LCFI) collaboration~\cite{lcfi}, which studies pixel detectors for heavy quark flavour
107: identification, has been implemented in simulations\,for\,c-quark\,tagging in scalar top\,studies.
108:
109: \vspace*{-4.5mm}
110: \section{Preselection}
111: \vspace*{-3.5mm}
112: Only two c-quarks and missing energy (from undetected neutralinos) are expected from\,the reaction
113: $\rm e^+e^- \rightarrow \stp \bar{\tilde{t}}_1 \rightarrow c \chio \bar c \chio$.
114: After a simple event preselection, the expected remaining signal and background
115: events are given in Table~\ref{tab:pre}.
116: The requirement $0.1 < E_{\rm vis}/\sqrt{s} < 0.3 $ reduces the
117: $\rm \rm e^+e^- \to W^+W^-$, ZZ, $\rm q\bar{q}$ and
118: $\rm \gamma\gamma \to q\bar{q}$ backgrounds.
119: Remaining two-photon events are almost completely removed by the cut $p_{\rm t}({\rm event}) > 15$~GeV.
120: Requiring at least four but no more than 50 tracks removes mostly
121: very low multiplicity background.
122:
123:
124: \begin{table}[h]
125: \vspace*{-1mm}
126: \caption{Generated events, events used for the IDA training, events after the
127: preselection, $\sqrt{s} = 260$ GeV cross section, scaling factor, and expected number of events.}
128: \begin{tabular}{lrrcccc}
129: \hline
130: & Total & 50\% & After &$\sigma$ & Factor & Expected \\
131: Process & $\times 1000$ & training & preselection & (pb) & per 50~fb$^{-1}$& events \\
132: \hline
133: signal & 50 & 25 & 13113 & 0.0225 & 0.0450 & 590 \\
134: $\rm q \bar{q}$, $\rm q \neq t$
135: & 350 & 175 & 55 & 49.5 & 14.14 & 778 \\
136: $\rm W^+W^-$ & 180 & 90 & 49 & 16.9 & 9.300 & 456 \\
137: $\rm W e\nu$ & 210 & 105 & 914 & 1.73 & 0.824 & 753 \\
138: 2-photon & 1600 & 800 & 3 & 786 & 49.13 & 147 \\
139: $\rm ZZ$ & 30 & 15 & 13 & 5.05 & 18.33 & 238 \\
140: $\rm e e Z$ & 210 & 105 & 12 & 1.12 & 0.533 & 6.4
141: \end{tabular}
142: \label{tab:pre}
143: \vspace*{-0.7cm}
144: \end{table}
145: \clearpage
146: The distributions of visible energy, transverse momentum and charged track multiplicity
147: before preselection are shown in Fig.~\ref{fig:pre}.
148: After the preselection 52.5\% signal efficiency and 2379 background events remain per 50~fb$^{-1}$.
149:
150: \begin{figure}[htbp]
151: \vspace*{-0.6cm}
152: \begin{center}
153: \begin{minipage}{0.49\textwidth}
154: \epsfig{file=h2_eseen_a.eps,width=\textwidth}
155: \vspace*{-0.6cm}
156: \end{minipage}
157: \begin{minipage}{0.49\textwidth}
158: \epsfig{file=h1_pt.eps,width=\textwidth}
159: \vspace*{-0.6cm}
160: \end{minipage}
161: \begin{minipage}{0.49\textwidth}
162: \epsfig{file=h18_ntrack.eps,width=\textwidth}
163: \end{minipage}
164: \end{center}
165: \vspace*{-0.6cm}
166: \caption{Event distributions before preselection.}
167: \label{fig:pre}
168: \vspace*{-0.5cm}
169: \end{figure}
170:
171: \section{IDA Event Selection}
172: \vspace*{-2mm}
173: Figure~\ref{fig:mass} shows additional input variables used in the IDA:
174: the event invariant mass and the invariant mass of the two jets.
175: Further input variables are the c-quark tagging of the leading (most energetic)
176: and subleading jets (Fig.~\ref{fig:ctag}). The c-quark jet tagging has been performed
177: with a neutral network~\cite{kuhl} optimized for small $\Delta m$.
178:
179: The IDA has been applied in two steps in order to optimize the performance, as shown in Fig.~\ref{fig:ida}.
180: In the first step, a cut was applied on the IDA\_1 output variable such that 99.5\% of the
181: signal events remain.
182: This leads to 52.0\% signal efficiency and 490 background events per 50 fb$^{-1}$.
183: These remaining signal and background events have been passed to the second IDA step.
184: A cut on the IDA\_2 output variable determines the final selection efficiency and the
185: corresponding expected background.
186: The resulting performance is shown in Fig.~\ref{fig:perf}.
187:
188: \begin{figure}[htbp]
189: \begin{minipage}{0.49\textwidth}
190: \epsfig{file=h4_minv_a.eps,width=\textwidth}
191: \end{minipage}
192: \begin{minipage}{0.49\textwidth}
193: \epsfig{file=h16_minv2_a.eps,width=\textwidth}
194: \end{minipage}
195: \vspace*{-0.5cm}
196: \caption{Distributions of IDA inputs after preselection.}
197: \label{fig:mass}
198: \end{figure}
199:
200: \begin{figure}[htbp]
201: \begin{minipage}{0.49\textwidth}
202: \epsfig{file=ctag_leading.eps,width=\textwidth}
203: \end{minipage}
204: \begin{minipage}{0.49\textwidth}
205: \epsfig{file=ctag_subleading.eps,width=\textwidth}
206: \end{minipage}
207: \vspace*{-0.5cm}
208: \caption{Distributions of IDA inputs after preselection.}
209: \label{fig:ctag}
210: \end{figure}
211:
212: \begin{figure}[htbp]
213: \begin{minipage}{0.49\textwidth}
214: \epsfig{file=ida_1.eps,width=\textwidth}
215: \end{minipage}
216: \begin{minipage}{0.49\textwidth}
217: \epsfig{file=ida_2_lin.eps,width=\textwidth}
218: \end{minipage}
219: \vspace*{-0.5cm}
220: \caption{IDA output for steps 1 and 2.
221: In the first IDA step, a cut on IDA\_1 is applied at zero,
222: retaining 99.5\% of the simulated signal input events.}
223: \label{fig:ida}
224: \end{figure}
225:
226: \clearpage
227: \begin{figure}[htbp]
228: \vspace*{-0.8cm}
229: \begin{center}
230: \epsfig{file=h26_performance.eps,width=0.8\textwidth,height=0.36\textheight}
231: \end{center}
232: \vspace*{-0.5cm}
233: \caption{Expected background events as a function of the signal efficiency.}
234: \label{fig:perf}
235: \vspace*{-0.8cm}
236: \end{figure}
237:
238: \section{Conclusions}
239: \vspace*{-2mm}
240: The Iterative Discriminant Analysis leads to a good signal over background ratio
241: for the investigated Supersymmetric scenario with a scalar top mass of 122.5~GeV and a neutralino
242: mass of 107.2~GeV (Fig.~\ref{fig:perf}).
243: For 50\% signal efficiency (560 signal events) about 200 background events (mostly $\rm We\nu$)
244: are expected per 50~fb$^{-1}$.
245: The analysis is a step towards a precise scalar top mass determination.
246: The expected uncertainty in the light scalar top mass measurement dominates the
247: uncertainty in the dark matter prediction from the co-annihilation process~\cite{carena,sopczak}.
248: A further study will also include a systematic error analysis.
249: While this study has focused on $\sqrt{s}=260$~GeV, an IDA study for this small $\Delta m$ scenario
250: at $\sqrt s = 500$~GeV is in progress~\cite{susy06} and
251: a much improved determination of the dark matter prediction is expected~\cite{ayres}.
252: Further plans are to focus on the vertex detector design, including the implementation of a new
253: LCFI~\cite{lcfi} c-quark tagging.
254:
255: \vspace*{-4mm}
256: \section*{Acknowledgements}
257: \vspace*{-2mm}
258: AS would like to thank the LCFI Colleagues for fruitful discussions on the analysis and for comments
259: during the preparation of the presentation.
260:
261: \vspace*{-3.5mm}
262: \begin{thebibliography}{99}
263: \vspace*{-1mm}
264: \bibitem{Balazs:2004bu}
265: C.~Bal\'azs, M.~Carena and C.~E.~M.~Wagner, Phys.\ Rev.\ {\bf D 70} (2004) 015007.
266: \bibitem{carena} M.~Carena, A.~Finch, A.~Freitas, C.~Milst\'ene, H.~Nowak and
267: A.~Sopczak, Phys. Rev. {\bf D 72} (2005) 115008.
268: \bibitem{sopczak} A.~Sopczak, M.~Carena, A.~Finch, A.~Freitas, C.~Milstene and H.~Nowak,
269: SUSY'05 conference, hep-ph/0605225.
270: \bibitem{ida} J. Malmgren and K. Johansson, Nucl. Inst. Methods {\bf A~403} (1998) 481.
271: \bibitem{lcfi} LCFI Collaboration, http://hepwww.rl.ac.uk/lcfi
272: \bibitem{kuhl} T.~Kuhl, ``Hadronic branching ratio of a SM-like Higgs boson at a future linear collider'',
273: Proc. of the Int. Conference on Linear Colliders (LCWS 04), Paris, France, 19-24 April 2004.
274: \bibitem{susy06} A.\,Sopczak,\,A.\,Finch,\,A.\,Freitas,\,C.\,Milst\'ene\,and\,M.\,Schmitt,\,SUSY'06\,conference,\,Irvine,\,USA.
275: \bibitem{ayres} A. Freitas, C. Milst\'ene, M.~Schmitt and A. Sopczak, in preparation.
276: \vspace*{-3mm}
277: \end{thebibliography}
278: \end{document}
279: