hep-ex0311019/analysis_chargino.tex
1: \subsection{Chargino searches}
2: \subsubsection{Non-degenerate scenarios}
3: \label{sub:CHASEL}
4: 
5: 
6: The search for charginos in the non-degenerate scenarios
7: covers the case when the mass difference  
8: \DM = \MXC{1}$-$ \MXN{1}\ is above 3 \GeVcc .
9: In order to take all possible signatures of chargino decays into account,
10: events were 
11: divided into four mutually exclusive topologies:
12: \begin{itemize}
13: \item{the \ll\ topology, with no more than five charged particles and no isolated photons;}
14: \item{the \jjl\ topology, with more than five charged particles, at least one isolated lepton and no 
15: isolated photons;}
16: \item{the \jjjj\ topology, with more than five charged particles and no isolated photons or leptons;}
17: \item{the \rad\ topology, with at least one isolated photon.}
18: \end{itemize}
19: The signal events selected in a given topology are mostly events from the corresponding decay
20: channel, but events from other channels may also contribute. For instance, for low \DM\
21: (and thus low visible energy) 
22: some events with hadronic decays are selected in the \ll\ topology, and some mixed decay  events 
23: with the isolated lepton unidentified enter into the {\it jets} topology. This migration effect tends to 
24: disappear as \DM\ increases. This effect was taken into account in the final efficiency and limit 
25: computations.
26: 
27: The signal events were simulated using 132 combinations of \XPM{1} and \XN{1} masses for nine 
28: chargino mass values 
29: (\MXC{1}~$\approx$~103, 102, 100, 98, 94, 85, 70, 50  and 45~\GeVcc) and with \DM\ ranging 
30: from 3~\GeVcc\ to 80~\GeVcc. 
31: A total of 264000 chargino events (2000 per mass combination) was generated. The kinematic 
32: observables (acoplanarity\footnote{To compute the  acoplanarity and acollinearity the particles 
33: were forced into two jets by the {\tt DURHAM} algorithm.}, \Evis, \Ptmis, etc.) of the signal events were studied 
34: in terms of their mean value and standard deviation, and six \DM\ regions were defined, each 
35: containing signal events with similar properties (see table  \ref{tab:REGION}). 
36: 
37: In each of these 24 windows (four topologies, six \DM\ regions), a likelihood ratio function 
38: ($\mathcal{L}_{R}$, see section~\ref{subsubsec:analysistech}) was defined.
39: The variables $\left\{ x_{i}\right\} $ used to build 
40: the $\mathcal{L}_{R}$ functions in the present analysis 
41: were \cite{char}: the visible energy (\Evis), visible mass (\Mvis), missing transverse momentum (\Ptmis),
42: polar angle of the missing momentum, number of charged particles, total number of particles, 
43: acoplanarity, acollinearity, ratio of electromagnetic energy to total detected energy, 
44: percentage of total energy 
45: within 30$^\circ$ of the beam axis, kinematic information concerning the isolated photons and leptons and 
46: the two most energetic charged particles, and finally the jet characteristics.\\ 
47: 
48: The generation of these 24 likelihood ratio functions was performed as follows:
49: 
50: \begin{itemize}
51: \item{The signal distributions of all the variables  $\left\{ x_{i}\right\} $  were built 
52: with signal events generated with parameter sets giving rise to charginos and 
53: neutralinos with masses in the 
54: corresponding \DM\ region. For each \DM\ region the events were classified according to the
55: above topological cuts. The background distributions were built with background events passing the same topological cuts.}
56: \item{Preselection cuts\cite{chapre}, different for each \DM\ region, were applied in order to reduce the 
57: %high cross-section backgrounds 
58: backgrounds with largest cross-section 
59: (two-photon interactions and Bhabha events) and to generate the pdf's.
60: The total background was reduced to 5\% of the one
61: passing the general event preselection (section \ref{sec:presel}).  
62: The pdf's were then generated as mentioned in section~\ref{subsubsec:analysistech}.} 
63: Figure~\ref{fig:DATAMC}(\ref{fig:DATAMCS1}) shows the distributions of some event variables
64: for the \jjjj, \ll\ and \rad\ topologies 
65: for the 2000 data and simulation with the TPC sector 6 on (off). 
66: \item{To reduce statistical fluctuations, a smoothing was performed 
67: by passing the 24 sets of pdf's for signal and background through a triangular filter~\cite{FILTER}.}
68: \item{In each window all the combinations of the pdf's were tested, starting from a minimal set of four variables. 
69: Every combination defined an $\mathcal{L}_{R}$  function 
70: and 
71: an $\mathcal{L}_{R_{CUT}}$,
72: as described in section~\ref{subsubsec:analysistech},
73: using 
74: the single channel formula~\cite{pdg96}. The parameters entering this computation were the 
75: number of expected background events and the window efficiency of the chargino selection,
76: defined  as the mean efficiency of the chargino-neutralino mass sets belonging to the investigated window~\footnote{The efficiency of one chargino-neutralino 
77: mass set is defined as the number of events satisfying $\mathcal{L}_{R}\! >\! \mathcal{L}_{R_{CUT}}$  divided by the total number of chargino 
78: events satisfying the topological cuts.}. Figures~\ref{fig:DATAMC}(d) and 
79: \ref{fig:DATAMCS1}(d) show the good agreement obtained between real and simulated 
80: events as a function of the likelihood ratio cut,
81: for $25 \leq \DM < 35$~\GeVcc\ in the \jjl\ topology.} 
82: \item{The combination of variables corresponding to the lowest excluded cross-section defined the $\mathcal{L}_{R}$ 
83: function and the $\mathcal{L}_{R_{CUT}}$ of each window.}
84: \end{itemize}
85: 
86: %Finally, in a given topology the selection inside one \DM\ region was
87: %defined as a combination (logical OR) of the criteria for several regions, 
88: %chosen to minimise the excluded cross-section expected in the absence of a signal \cite{char}.
89: 
90: %Finally, for each window, the six likelihood ratio functions from the corresponding topology 
91: %were combined by performing a logical 'or' of each possible subset (always including the likelihood 
92: %ratio function from the considered window). The combination minimising the expected cross-section limit was chosen. The detection efficiency 
93: %was computed only using the chargino events belonging to the investigated window.     
94:    
95: %This analysis was applied to the data collected by DELPHI in 1999 and 2000.
96: %For the data recorded in 2000 only the first 3 regions of \sqs\ defined in table~\ref{tab:ENERGY2000}
97: %were investigated.
98: 
99: %The search was divided in the three  {\mbox{$ E_{\mathrm{\small cm}}$}}  regions of investigation 
100: %defined in table \ref{tab:ENERGY2000}. These results were used separately 
101: %for the LSP mass limit computation
102: %but were summed in one sample for the chargino mass limit computation. 
103: 
104: \begin{table}[ht]
105:  \begin{center}
106:    \begin{tabular}{|c|c|}
107:  \hline
108: \multicolumn{2}{|c|}{\DM\ regions} \\
109:  \hline
110:  1   &  3$\leq$\DM $<$ 5~\GeVcc \\
111:  \hline
112:  2   &  5$\leq$\DM $<$ 10~\GeVcc \\
113:  \hline
114:  3   &  10$\leq$\DM $<$ 25~\GeVcc \\
115:  \hline
116:  4   &   25$\leq$\DM $<$ 35~\GeVcc \\
117:  \hline
118:  5   &  35$\leq$\DM $<$ 50~\GeVcc \\
119:  \hline
120:  6   &   50~\GeVcc$\leq$\DM  \\
121:  \hline
122: \end{tabular}
123: \caption[.]{
124: \label{tab:REGION}
125: Definition of the \DM\ (mass difference between the chargino and 
126: lightest neutralino) regions for the chargino search in non-degenerate scenarios.}
127: \end{center}
128: \end{table}
129: 
130: 
131: \subsubsection{Nearly mass-degenerate scenarios}
132: \label{sec:chadegsel}
133: 
134: 
135: The search for charginos in the nearly mass-degenerate scenarios uses several different
136: techniques, depending on the lifetime of the chargino, which in turn depends on the 
137: mass difference \DM\ between the chargino and the lightest neutralino (this is the
138: only relevant dependence, at least in the heavy slepton hypothesis).
139: When \DM\ is below the mass of the pion, the chargino lifetime is usually long enough
140: to let it pass through the entire detector before decaying. This range of \DM\ can be
141: covered by the search for long-lived heavy charged particles. 
142: For \DM\ of a  few hundred \MeVcc\, the chargino can decay inside the main tracking
143: devices. Therefore, a search for secondary vertices or kinks can be used to
144: cover this region. 
145: As the mass difference increases, the mean lifetime shortens until the position
146: of the $\tilde\chi_1^{\pm}$ decay can hardly be distinguished from the main event
147: vertex. In this case, the tagging of an energetic ISR photon
148: can help in exploring the \DM\ region between a few hundred \MeVcc\ and 3~\GeVcc.
149: The selection criteria
150: are similar to the ones used in the analysis of previous data,
151: which have been described in~\cite{isr}. 
152: %All data were finally combined to obtain the present results.
153: 
154: \vspace{0.35cm}
155: \noindent
156: \underline{Search for long lived charginos}
157: \vspace{0.35cm}
158: 
159: Long lived charginos can either be ``quasi-stable'' (decay outside  the tracking
160: system) or decay ``visibly'' inside  the  tracking devices.
161: 
162: The search for heavy stable charged particles is described in~\cite{GMSB-contribution}.
163: The method used to identify heavy stable particles relied on the ionisation loss 
164: measurements in the TPC and on the absence of signal in the DELPHI Cherenkov 
165: radiation detectors (RICH). 
166: Heavy stable charged particles crossing the detector would be seen in the tracking
167: system and have as distinctive signature the absence of  Cherenkov radiation and an
168: anomalous energy loss in the TPC.
169: Three different search windows were used in the search for heavy stable charginos:
170: \begin{itemize}
171: \item the charged particle had momentum above 15~\GeVc, and no photons in 
172: any of the two radiators of the RICH (liquid, refractive index $n=1.28$, and
173: gas, $n=1.0015$) were associated to the track;
174: \item the charged particle had momentum above 5~\GeVc, high ionisation loss in the TPC,
175:  and no signal in the gas RICH;
176: \item the charged particle had momentum above 15~\GeVc, a TPC ionisation loss not exceeding
177:  70\% of the expectation for a proton, and no signal in the gas RICH.
178: \end{itemize}
179: A fourth search window considered in \cite{GMSB-contribution} was not included,
180: in order to treat the two hemispheres of the event independently. 
181: 
182: If a heavy charged particle decays inside the central tracking devices 
183: (at a radius between 10~cm and 1~m) then both the incoming and the outgoing track
184: can be reconstructed, and the angle between the tracks can be calculated. 
185: Such a search for kinks was originally designed to search for long-lived staus
186: in the Gauge Mediated SUSY Breaking scenario~\cite{GMSB-contribution}. A similar
187: technique was applied to search for mass-degenerate charginos, with some specific
188: features needed because the visible decay products carry very little momentum in
189: the nearly mass-degenerate case.
190: Details of the selection criteria can be found in \cite{isr}. 
191: Here only a
192: brief and qualitative summary of the most important selection cuts is given.
193: 
194: A set of rather loose general requirements was applied in order
195: to suppress the low energy background (beam-gas, beam-wall, etc), two-photon,
196: \ee\ and hadronic events. 
197: For each event passing the preselection cuts, all the charged particles
198: were grouped in clusters according to their measured point closest to the
199: interaction vertex.
200: A cluster with only one track with momentum above 20~\GeVc\ was considered
201: as a possible chargino candidate if it was compatible with a particle coming
202: from the interaction point.
203: For each single track cluster fulfilling the above conditions, a search was
204: made for a second cluster possibly formed by the decay products of the \XP{1}\
205: and defining a secondary vertex or kink with the chargino candidate.
206: 
207: Reconstructed secondary vertices could also be the result of particles
208: interacting in the detector material, or having a particle
209: trajectory reconstructed in two separate track segments.
210: Additional requirements rejected these backgrounds in the events with an
211: acceptable secondary vertex~\cite{GMSB-contribution}.
212: Finally, for an event to be accepted, at least one charged particle had to be
213: found in each hemisphere (defined by the plane containing the beam spot
214: and perpendicular to the line connecting the beam spot to the kink).
215: 
216: The search for events with tracks at large impact parameter described
217: in~\cite{GMSB-contribution} was not possible in this case:
218: events with only two extremely soft charged particles with large 
219: impact parameter are difficult both to trigger on and to 
220: discriminate from machine related noise.
221: Such events were however considered if a
222: high \pt\ ISR photon was present, as explained in the next section.
223: 
224: \vspace{0.35cm}
225: \noindent
226: \underline{Search for charginos with ISR photons}
227: \vspace{0.35cm}
228: 
229: The visible particles resulting from the decay of
230: a chargino nearly mass-degenerate
231: with the LSP have typically little energy and momentum. 
232: The trigger efficiency is low for such events, and there is a very large background 
233: %They are hard to trigger on in the experiment, and in addition they are overwhelmed by the 
234: from two-photon events.
235: The ISR photon tag improves detectability and, if the transverse energy of the photon
236: is above a threshold which depends on the minimal polar angle acceptance of the 
237: experiment, it rejects most of the two-photon background.
238: 
239: After the preselection, which was summarised in \ref{sec:presel}, the following 
240: requirements were applied to the data and simulation samples.
241: 
242: \begin{itemize}
243: \item{There had to be at least two and at most six good charged particles passing 
244: the quality criteria (see section~\ref{sec:trksel}), and no more than ten 
245: tracks in total.}
246: \item{The transverse energy of the ISR photon candidate was required to be greater
247:  than $(E_T^{\gamma})^{\min} \simeq 0.03\cdot\rs$.}
248: \item{The mass recoiling against the photon had to be above 
249:  $2 \MXC{1}-\delta M$, where the term $\delta M$ takes into account 
250:  the energy resolution in the electromagnetic calorimeters.}
251: \item{The photon had to be isolated by at least $30^\circ$ with respect
252:  to any other charged or neutral particle in the event.}
253: \item{The sum of the energies of the particles with polar angles within $30^\circ$
254:  of the beam axis ($E_{30}$) was required to be less than 25\% of the total visible
255:  energy. If the photon itself was below $30^\circ$, it was the ratio
256:  $(E_{30}-E_{\gamma})/(\Evis-E_{\gamma})$ that was required to be below $0.25$.}
257: \item{If the ISR photon candidate was detected in the very forward 
258:  calorimeter STIC, it must not  be correlated with a signal in the
259:  scintillators placed in front of the STIC.}
260: \item{$(\Evis-E_{\gamma})/ \rs$ had to be below a kinematic threshold
261:  which depended on \DM\ and on \MXC{1}\ (and in any case below 6\%).}
262: \item{The ratio of the absolute value of the missing transverse momentum over the
263:  total transverse energy had to be above $0.40 / c$ if $\DM>300$~\MeVcc, and 
264:  above $0.75 / c$ for smaller $\DM$.}
265: \item{If $\DM>1$~\GeVcc~, at least two charged particles in the event had to be
266:  consistent with coming from the beam interaction region.}
267: \end{itemize}
268: 
269: Distributions of some of the variables used in the final selection are shown in
270: figure~\ref{fig:isrcomp} for data, simulated SM background, and simulated signal events. 
271: Although there is a certain overall qualitative agreement of the various distributions, there is 
272: already an excess of data. On the other hand, the two-photon generators used in the 
273: simulation lack the events which have small $\gamma\gamma$ invariant mass, and in some
274: cases (namely, $\gamma\gamma \to \ee $) no ISR generation is implemented at all. 
275: Moreover, background processes such as beam-gas interactions are not included in the 
276: simulation. As in previous publications~\cite{isr}, the most likely explanation of
277: such disagreement is therefore a deficit of simulated background events rather than an
278: excess of data from possible new physics. As no attempt will be made in the following
279: to account for the backgrounds missing in the simulation, the limits that
280: will be obtained are conservative.
281: 
282: 
283: 
284: 
285: 
286: 
287: 
288: 
289: 
290: 
291: 
292: 
293: 
294: 
295: %KHedit95
296: