1: \chapter{Event Analysis}
2: \label{analysis}
3: \section{Introduction}
4: The technique for the selection of candidate signal events is a
5: two-step process. Firstly, using the ALPHA \cite{alpha} package which
6: facilitates access to the full event information, preliminary data
7: candidates are identified using some simple preselection. For those
8: events satisfying the preselection, variables are calculated that are
9: designed to be sensitive to the differences between signal and
10: background. These are written to HBOOK \cite{hbook} ntuples
11: (essentially tables of data in which rows correspond to successive
12: events and columns correspond to variables) forming a greatly reduced
13: data set. The second stage is the application of a selection,
14: consisting of a set of cuts on the variable values, to determine the
15: final signal candidates. This is the technique used by most searches
16: for physics beyond the Standard Model at LEP.
17: \par
18: Many, if not all, of the variables should pertain to the event as a
19: whole, such as the number of charged tracks, total energy and
20: invariant mass. In order to calculate such `global' variables the
21: relationships between the various reconstructed detector objects such
22: as tracks and calorimeter deposits should be known. This is the
23: purpose of the energy flow algorithm. But energy flow has only limited
24: abilities to make sense of processes that are happening away from the
25: interaction point. It ignores tracks with a \dzero\ greater than 2\cm\ or a
26: \zzero\ greater than 10\cm, and since JULIA does not reconstruct the
27: slepton decay vertex, it does not recognise the relationship between
28: the reconstructed slepton track (if there is one) and the resulting
29: lepton track. Thus energy flow will only yield correct results if the
30: slepton decay length is short enough for the resulting lepton to
31: satisfy the cuts on \dzero\ and \zzero, or long enough for the slepton to reach
32: the calorimeters before decay; and so it is of limited use in this
33: analysis.
34: \par
35: Even if energy flow did treat charged particle decay in the tracking
36: volume correctly, the signal would still not be well-defined by global
37: variables in the case of the \staustau\ channel (and the \selstau\ and
38: \smustau\ channels to a lesser extent) because of the larger number of
39: invisible particles (up to eight neutrinos on top of the two
40: gravitinos) and the larger range in the charged track multiplicity
41: (from multi-prong tau decays). The selection then relies primarily on
42: the presence of high-\dzero\ tracks: that being the primary feature of the
43: signal. Several other processes can generate high-\dzero\ tracks though,
44: such as nuclear interactions, ECAL splash-backs, photon conversions,
45: multiple scattering and cosmic rays. The frequency with which these
46: processes occur means that their rejection efficiency must be close to
47: 100\% if they are not to swamp any signal.
48: \par
49: After the event preselection is described in Section~\ref{presel},
50: this chapter goes on to detail the procedures that are used to
51: discriminate between high-\dzero\ tracks due to the signal process, and
52: those due to background. Sections~\ref{splashbacks} and \ref{NIs}
53: describe the identification of ECAL splash-backs and nuclear
54: interactions respectively. Then Section~\ref{fromslep} describes the
55: procedure that is used to confirm (or reject) that a high-\dzero\ track is
56: compatible with the hypothesis of originating from slepton decay. The
57: final cut-based selections are described in
58: Chapter~\ref{selections}. In order to facilitate the descriptions of
59: the procedures described in this chapter a terminology is used which
60: is explained below and in Figure~\ref{jargon}. The figure also
61: introduces some points of interest concerning tracks reconstructed
62: from particles that have not been produced at the IP. There are three
63: terms which are important to understand:
64: \par
65: \textbf{The post(pre)-\dzero\ trajectory of a particle.} This is the section of the particle's trajectory after it has passed (before it has reached) its point of closest approach to the beam-axis.
66: \par
67: \textbf{A post(pre)-\dzero\ track.} This is a track that has been formed by hits created on the post(pre)-\dzero\ portion of a particle's trajectory. Both this and the term above refer to truths about the event that are not known a priori from the reconstructed information.
68: \par
69: \textbf{The post(pre)-\dzero\ section of a track.} This is the section of a track which is post(pre)-\dzero\ according to the track direction (assigned by JULIA, and which may be incorrect). The track will only have assigned hits on its post-\dzero\ section. If it is a pre-\dzero\ track then its pre-\dzero\ section will probably follow closely the post-\dzero\ section of a post-\dzero\ track (see Figure~\ref{jargon}).
70: \par
71: Although this terminology might seem confusing, it is an unfortunate
72: reflection of the subtleties involved in the relationship between the
73: reconstructed and `truth' information of tracks not originating from
74: the IP.
75: \par
76: Many of the procedures \iforig{involving}{involve} the fitting of
77: tracks to vertices, either to test whether a group of tracks form a
78: new vertex, or whether they fit to an existing vertex. This is
79: performed using the ALEPH YTOP package described in \cite{ytop} and
80: \cite{lutz}. It provides a $\chi^2$ for each fit performed, which is
81: normalised to the number of degrees of freedom.
82:
83: \begin{figure}[htpb]
84: \setp{108mm}{70mm}{100mm}{1.4}
85: \centerline{
86: \framebox{\resizebox{10.3cm}{!}{\includegraphics*[\xa,\ya][\xb,\yb]{dali_trak.eps}}}
87: }
88: \caption{\lscap{The reconstructed track(s) of a particle that has been produced in the TPC, moving towards the beam-axis, through the ITC, then back out into the TPC and on to the ECAL. The \dzero-point is the point at which the track is closest to the beam-axis. Before it reaches the \dzero-point it is travelling along its pre-\dzero\ trajectory, afterwards, along its post-\dzero\ trajectory. Since the idea that particles are produced at the IP is a paradigm within JULIA, it assumes that the particles responsible for all tracks were moving away from the beam-axis and did not originate significantly before their \dzero-point. For this reason it will not form tracks from hits that would lie on both sides of the \dzero-point; and so the particle trajectory shown here is reconstructed as two tracks: one pre-\dzero\ track reconstructed from the pre-\dzero\ hits, and one post-\dzero\ track reconstructed from the post-\dzero\ hits. JULIA incorrectly assigns the directions of pre-\dzero\ tracks as being radially outwards, and so both their momentum 3-vector and charge have the wrong sign. Tracks that have no TPC hits will be referred to as ITC tracks, tracks that do have TPC hits will be referred to as TPC tracks and may or may not have ITC hits. Either type of track may or may not have VDET hits.}{An example of a high-\dzero\ track with labelling to indicate the meaning of certain terms.}}
89: \label{jargon}
90: \end{figure}
91: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
92: \section{Preselection}
93: \label{presel}
94: In order to speed up analysis, events were first subjected to some
95: very basic preselection aimed at fast rejection of events that were
96: clearly not signal candidates. The event was rejected if
97: \begin{itemize}
98: \item The scalar sum of the momenta of all tracks was less than \mbox{$4\times(E_{LEP}/188.6)\GeV$} (to reduce the number of two-photon events)
99: \item The total number of tracks was less than 3 or greater than 30 (to reduce both two-photon and $q\overline{q}$ events)
100: \item The total number of tracks was less than 4 and the two highest momentum tracks had momenta greater than half the beam energy and the angle between them was greater than $179^\circ$ (to reduce $e^+e^-$ and $\mu^+\mu^-$ events)
101: \item The total number of energy flow charged tracks was greater than 11 (to further reduce $q\overline{q}$ events)
102: \item The energy flow event energy was less than \mbox{$7\times(E_{LEP}/188.6)\GeV$} and the energy flow event momentum was less than \mbox{$1\times(E_{LEP}/188.6)\GeV$} (to further reduce two-photon events)
103: \end{itemize}
104:
105: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
106: \section{Rejection of ECAL splash-backs}
107: \label{splashbacks}
108: The electromagnetic shower created when a charged particle enters the
109: electromagnetic calorimeter typically propagates in the direction of
110: the incoming particle. Some of the electrons and positrons in the
111: shower can however, scatter back into the TPC where they can be
112: reconstructed as tracks. This is known as an `ECAL splash-back'.
113: \par
114: \begin{figure}[htb]
115: \centerline{
116: \resizebox{10cm}{!}{\includegraphics*[45mm,80mm][150mm,170mm]{tautau_189_BB_Y15279.1.eps}}
117: }
118: \caption{An example of an ECAL splash-back in a Monte Carlo $\tau\tau$ event at $189\GeV$. \ladd{Two electrons have escaped the ECAL to be reconstructed as high \dzero\ tracks in the TPC, one arcs through the ITC and passes out the other side to form a third track.}}
119: \label{splashbackpic}
120: \end{figure}
121: They are characterised by a high-\dzero\ track `entering' the
122: \iforig{TPC}{ECAL} (in fact the particle responsible is leaving but
123: JULIA assumes all tracks move radially away from the beam-axis: the
124: track is pre-\dzero) close to another track that has higher momentum,
125: with both forming a common vertex in the region of the ECAL (see
126: Figure~\ref{splashbackpic} for an example). The only circumstance
127: under which this is likely to occur in a signal event is when the
128: slepton decays inside the ECAL and the resulting lepton is ejected
129: back into the TPC. Since this analysis is only intended to be
130: sensitive to slepton decay in the tracking volume then the possibility
131: of rejecting signal events by confusing slepton decays with ECAL
132: splash-backs is not regarded a problem.
133: \par
134: The track that results from the splash-back will be referred to as the
135: output track, the track causing the splash-back will be referred to as
136: the input track. The conditions that must be satisfied by both a
137: candidate output and input track for the output track to be tagged as
138: a splash-back are as follows:\\
139: \begin{minipage}{\linewidth}
140: \begin{itemize}
141: \item The output track must have TPC hits and momentum, $p_{output}<\\5\times (E_{LEP}/188.6)\GeV$ and \dzero$>0.5\cm$ or \zzero$>2\cm$.
142: \item The input track must also have TPC hits and have $p_{input}>0.5\GeV$ and $p_{input}>p_{output}$.
143: \item The points at which the output and input tracks leave the TPC volume must be within 50\cm\ of each other.
144: \item The tracks must form a common vertex in the region
145: \begin{tabbing}
146: The tracks must for\= \kill
147: \> $165<r<230\cm$ and $|z|<300\cm$, \\
148: or \> $200<|z|<300\cm$ and $r<230\cm$
149: \end{tabbing}
150: (these roughly correspond to the vicinity of the ECAL barrel and endcaps respectively).
151: \end{itemize}
152: \end{minipage}
153: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
154: \section{Rejection of nuclear interactions}
155: \label{NIs}
156: Hadrons interacting with nuclei in the structure of the detector can
157: cause hadronic showers leading to multiple high-\dzero\ tracks. These are
158: characterised by many tracks originating from a common vertex which is
159: displaced from the IP (see Figure~\ref{NIpic} for an example). Like
160: splash-backs these have distinct characteristics that can be
161: tagged. Care must be taken though to avoid confusing a
162: \stautomultiprongtau decay vertex with a nuclear interaction.
163: \par
164: \begin{figure}[htb]
165: \centerline{
166: \resizebox{10cm}{!}{\includegraphics*[55mm,110mm][160mm,205mm]{tautau_189_NI_Y15279.1.eps}}
167: }
168: \caption{An example of a nuclear interaction in a Monte Carlo $\tau\tau$ event at $189\GeV$. \ladd{A pion from the decay of a tau has interacted with a nucleus in the inner TPC wall, giving a shower of high-\dzero\ tracks, mainly protons.}}
169: \label{NIpic}
170: \end{figure}
171: Clearly an algorithm that can find multiple tracks originating from a
172: common vertex is required. Initially this was done by testing all
173: track pairs for compatibility with a common vertex, and then
174: attempting to merge these 2-track vertices. This was very CPU
175: intensive however, and a method which used the fitting routine less
176: often was sought. The procedure settled upon relied on starting with a
177: single high-\dzero\ track, and trying to add further tracks to it to build
178: up a multi-track vertex. Once all the tracks that could be added had
179: been (which may be none at all), and if there were one or more other
180: high-\dzero\ tracks in the event which had not been included in the vertex,
181: then one of these was used in the same way as the first track to try
182: to build a new vertex. Unlike the first method tried, this method
183: required making a choice as to the order in which the tracks should be
184: selected so as to give the best chance of all tracks from a given
185: vertex actually being included. Using tracks in order of increasing
186: fractional momentum error was found to be successful. The exact
187: procedure used is shown in the flow diagram of Figure~\ref{NIflow}. In
188: this diagram `most reliable' means smallest fractional momentum
189: error. A track passes `basic cuts' if it has not been tagged as an
190: ECAL splash-back, has TPC hits, has a momentum smaller than half the
191: beam energy and greater than $0.1\GeV$, and a \dzero$>0.15\cm$ or a
192: \zzero$>3\cm$. A track is `hit-compatible' with a vertex if at least all
193: bar one of its hits lie on one side of the vertex, so that it is
194: compatible with the hypothesis that it was produced at that vertex.
195: \par
196: \begin{figure}[htbp]
197: \centerline{\iforig{\epsfig{angle=-90,file=flow_original.eps}}{\epsfig{file=flow.eps,width=\textwidth}}}
198: \label{NIflow}
199: \caption{Flow diagram showing the procedure used to find and tag nuclear interaction vertices.}
200:
201: \end{figure}
202: Once one multi-track vertex has been found with this procedure, then
203: if each new vertex after the first is less than $3\cm$ from, or shares
204: more than 50\% of its tracks with, a previously found vertex then an
205: attempt is made to merge the two. If this results in a new vertex with
206: a $\chi^2$ (given by YTOP) less than 70 then the merge is considered
207: successful, and the two vertices are replaced with the single merged
208: vertex.
209: \par
210: Once the vertices have been finalised, certain properties are
211: calculated for each, such as the summed momentum and charge of the
212: constituent tracks. Care was taken however, to account for a
213: potentially problematic feature of the reconstruction algorithm. Since
214: JULIA always assumes that particles move radially away from the beam
215: axis, a particle that is produced far enough from the beam axis and
216: moving back towards it can result in two reconstructed tracks (a pre-\dzero\
217: and a post-\dzero), as is shown in Figure~\ref{jargon}. In this case the
218: track corresponding to the portion of the particle trajectory before
219: it reaches its point of closest approach to the beam axis is assigned
220: the wrong direction. Thus its momentum 3-vector and charge have the
221: wrong sign, and this will corrupt the overall properties of any
222: multi-track vertex in which it is included. Also, since the total
223: number of tracks in the vertex is used to classify it, then a single
224: particle giving two tracks would corrupt this as well. To correct this,
225: any track whose TPC hits lie at a smaller radius than the vertex has
226: its charge and momentum 3-vector reversed. Then the \dzero, \zzero\ and
227: directions of these tracks are compared with any tracks which are part
228: of the same vertex and for which the vertex lies on the pre-\dzero\
229: portion. If the tracks differ in \dzero\ and \zzero\ by less than 30\% of their
230: respective average, and the angle between them at their \dzero-points is
231: greater than $170^\circ$, then both are assumed to have originated
232: from the same particle trajectory and the one with fewer TPC hits is
233: removed from the vertex. As an example, if both the tracks shown in
234: Figure~\ref{jargon} were found to originate from a nuclear
235: interaction, then the pre-\dzero\ track would be removed.
236: \par
237: Each vertex is then placed into one of 3 categories based on its
238: properties. A vertex is labelled as a \stautomultiprongtau vertex
239: if it satisfies all of the following conditions --
240: \begin{itemize}
241: \item its charge is $\pm1$
242: \item its largest two-track angle at the vertex is less than $20^\circ$
243: \item its momentum is $>4\times(E_{LEP}/188.6)\GeV$ and $<55\times(E_{LEP}/188.6)\GeV$
244: \item its kink-angle is between $5^\circ$ and $170^\circ$
245: \item its kink-angle is $>15^\circ$ or its $|\cos(\theta)|$ is less than 0.6
246: \item its number of tracks is less than 6
247: \item its distance from the IP is greater than $0.3\cm$
248: \item and its distance from the IP is greater than $1\cm$ or its $\chi^2$ is less than 3.
249: \end{itemize}
250: The charge of the vertex is the sum of the charges of all the
251: constituent tracks. The momentum is the vector sum of the momentum
252: 3-vectors of all the constituent tracks. The kink-angle is the angle
253: between the vertex momentum 3-vector and the straight line from the IP
254: to the vertex. The $|\cos(\theta)|$ of the vertex refers to the vertex
255: position, \ie\ $\iforig{|z_{vertex}/r_{vertex}|}{|z_{vertex}/\sqrt{r_{vertex}^2+z_{vertex}^2}\,|}$.
256: \par
257: Although this is a reasonably stringent set of cuts, background events
258: can \mbox{still} have vertices labelled as \stautomultiprongtau since a
259: charged particle emitting a bremsstrahlung photon which quickly
260: undergoes pair-conversion can create something very similar.
261: \par
262: If a vertex fails this selection then it is classified as a nuclear
263: interaction vertex if --
264: \begin{itemize}
265: \item its radius is greater than $4.8\cm$ (a nuclear-interaction is highly unlikely inside the beam-pipe, so the minimum radius is set at the beam-pipe radius of 5.3\cm, minus 5\mm\ leeway)
266: \item its largest two-track angle at the vertex is greater than $20^\circ$
267: \item and its number of tracks is greater than 2.
268: \end{itemize}
269: If it fails this also then it is classified as `unknown', and the
270: constituent tracks are still considered valid.
271:
272: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
273: \section{Confirmation of slepton-decay hypothesis}
274: \label{fromslep}
275: Although this analysis attempts to reject high-\dzero\ tracks from known
276: sources such as those discussed in Sections~\ref{splashbacks} and
277: \ref{NIs}, and these attempts are largely successful, they are not sufficient. The number of residual high-\dzero\ tracks is still too many. Hence a procedure was developed, not to test whether a high-\dzero\ track comes from a known background process, but to test whether it is compatible with coming from the signal process, \ie\ the decay of a heavy charged particle. The hits of a candidate track are examined to identify the region of its length over which the responsible particle could have originated, and this is then used to calculate what can be expected as the detector response to the parent slepton. In the absence of such a response the track can be rejected. For instance, if a track's hits indicate that the responsible particle must have been produced far out in the TPC, the parent slepton should have its own reconstructed track which forms a good kink-vertex with the high-\dzero\ track. In the absence of such a 2-track vertex the high-\dzero\ track can be rejected. This section describes the exact procedure in detail. It should be noted that only tracks that have TPC hits, a momentum greater than $1\GeV$, a \dzero\ greater than $0.1\mm$ and a $\chi^2_{BS}$ (the $\chi^2$ of the track-fit to the beam-spot per degree of freedom) greater than 5 are considered as candidate high-\dzero\ tracks.
278:
279: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
280: \subsection{A particle's point of origin}
281: \label{origin}
282: While a reconstructed track is a reliable indicator of a particle's
283: trajectory, it does not tell us exactly where that particle originated
284: or where it stopped/decayed. Nominally, the point of origin is
285: considered to be the \dzero-point. This is fine for tracks originating
286: from the IP, since the distance scale associated with the uncertainty
287: on the particle's exact point of origin will be at most the size of
288: the beam spot, which will be several orders of magnitude lower than
289: the track's radius of curvature. Thus the track is effectively
290: straight over the relevant range, and the uncertainty on its point of
291: origin does not translate into significant uncertainty on its initial
292: momentum 3-vector. For a track not originating from the IP no such
293: simple assumption can be made, and only analysis of its hits can yield
294: information as to its point of origin. This will take the form of a
295: range of the track length over which the particle could have
296: \iforig{orginated}{originated}.
297: \par
298: The following procedure for finding this range assumes that the track
299: in question is post-\dzero\ (\ie\ reconstructed from hits created as the
300: particle moves away from the beam axis), and so the track direction
301: and charge are correctly assigned by \mbox{JULIA}. Although the
302: following procedure is not sensible for a pre-\dzero\ track, it will be
303: explained later that such a track is in fact dealt with in the correct
304: way.
305: \par
306: The range over which the particle could have been produced is between
307: the \iforig{highest}{lowest}-radius point at which there is evidence
308: that the particle was following the track trajectory, and the
309: \iforig{lowest}{highest}-radius point at which there is evidence that
310: it was not. To find evidence that the particle was not following the
311: track trajectory, a scan is performed backwards along the track
312: length, starting from the point at which it achieves its highest
313: radius in the TPC, looking for missing hits. Nominally, a hit is
314: expected at every TPC pad radius that the track crosses, every ITC
315: wire radius that the track crosses while it is inside the ITC active
316: volume, and every intersection that the track makes with a VDET
317: wafer. There are several legitimate reasons why a track can be missing
318: a hit however. The TPC sector boundaries form dead regions of the
319: order of 1\cm\ wide, and in the ITC certain wires are not
320: active. Also, even if the particle has produced a hit it may not be
321: assigned to the respective track since the reconstruction algorithms
322: can simply fail to assign the hits correctly, or may only have one hit
323: to assign between two tracks (if they are closer than the minimum
324: two-track resolution). Thus if the track fulfils any of the conditions
325: given in Table~\ref{misshit} at a point where a hit is expected but
326: not observed, the lack of a hit is not considered abnormal.
327: \begin{table}
328: \renewcommand{\arraystretch}{1.3}
329: \centerline{
330: \begin{tabular}{|c|>{\scriptsize$\bullet$\hspace*{-2.5mm}}cp{12cm}|} \hline
331: \mr{4}{5ex}[-2ex]{TPC}& & The track is $<$\,1.5\cm\ in $r\phi$ from a sector boundary \\
332: & & The track is $<$\,1.5\cm\ in $r\phi$ and $<$\,4\cm\,(10\cm) in $z$ from another TPC (ITC) track on the same side (on either side) of the central membrane \\
333: & & The track is $<$\,2\cm\ in $r\phi$ and $<$\,4\cm\ in $z$ from an unassigned hit \\
334: & & The radius of the point in question is $<$\,3\mm\ greater than the \newline track's \dzero\ \\ \hline
335: \mr{4}{5ex}[-1ex]{ITC}& & The nearest wire has a hit \\
336: & & The nearest wire is dead \\
337: & & The track is $<$\,1\mm\ in $r\phi$ from the cell edge and the next-nearest wire has a hit\\
338: & & The point in question is at a radius $<$\,0.1\mm\ above the track's \dzero\ \\ \hline
339: \mr{2}{6ex}[0ex]{VDET}& & There is a hit within 0.5$^\circ$ in $\phi$ \\
340: & & The track is within one sigma of the wafer edge \\ \hline
341: \end{tabular}
342: }
343: \caption{Conditions under which the lack of a hit is not considered abnormal in each tracking subdetector.}
344: \label{misshit}
345: \end{table}
346: \par
347: If a point is encountered at which there is no hit and at which none
348: of the conditions of Table~\ref{misshit} are satisfied, then it is
349: assumed that the responsible particle was not following the track
350: trajectory at this point. Then if there are no hits at lower radii,
351: the range for the particle point of origin is set between this point
352: and the lowest-radius assigned hit. If there are one or more hits at lower
353: radii, then the track information appears to be inconsistent with a
354: genuine particle trajectory. In this case, if the point under
355: consideration is in the VDET the track is rejected as bad. If it is in
356: the ITC then the track is spared for now and the scan continues, but
357: if another such point is encountered the track is rejected. If the
358: point is in the TPC, then the location of the bad point is merely
359: recorded. These bad TPC points will be referred to again later.
360: \par
361: % I don't talk about the decision whether to continue the scan or not because I don't think it's important. The idea of the straight line from the track to the IP not crossing any tracking components will be introduced later.
362: If the scan reaches the track's \dzero-point without the track being
363: rejected or the particle range-of-origin being determined then the
364: scan continues along the pre-\dzero\ section of the track, as far as the
365: highest radius that the pre-\dzero\ section of the track reaches in the TPC
366: if necessary. On the pre-\dzero\ section the track will not have any
367: assigned hits and so Table~\ref{misshit} governs completely whether
368: the track is considered good at each point. If the scan reaches the
369: end of the pre-\dzero\ section without conclusion, then the track is
370: rejected (since it is probably a cosmic muon that has passed right
371: through the detector).
372:
373: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
374: \subsection{Where is the slepton?}
375: \label{mother}
376: Once a range for the particle's point of origin has been calculated,
377: it is possible to translate this into a range of possible trajectories
378: for the parent slepton, and to make the requirement that the event
379: information supports one of these trajectories and thus the hypothesis
380: that the candidate track originated from slepton decay. This section
381: describes the procedure that is applied to do this.
382: \par
383: There are four distinct scenarios for the reconstructed detector
384: response to the slepton depending on how far the slepton travels
385: before decay and in what direction. If the slepton achieves a high
386: enough radius before decay it can produce a reconstructed TPC track,
387: which should form a well-reconstructed vertex with the lepton
388: track. In the following discussion this will be referred to as a
389: TPC-vertex. Failing this, it can produce an ITC track which will form
390: a vertex with the lepton track that is not so well reconstructed in
391: $z$ due to the low $z$-resolution of the ITC. This will be referred to
392: as an ITC-vertex. Failing this, if the slepton at least passes through
393: some active tracking components before decay, then it should leave
394: some hits which point to the lepton track. These will be referred to
395: as pointing-hits. Finally there is the case that the slepton decays
396: before creating any hits, and so the only sign of its existence is the
397: \dzero\ of the lepton track. Examples of each case are shown in
398: Figure~\ref{mother2to5}. The range for the point of origin of the
399: particle responsible for the candidate high-\dzero\ track determines which
400: of these scenarios are possible, and which are not.
401: \setlength{\ljlen}{\topsep}
402: \setlength{\topsep}{0mm}
403: \setlength{\ljlenb}{7cm}
404: \renewcommand{\arraystretch}{0.5}
405: \begin{figure}[htbp]
406: \centerline{
407: \begin{tabular}{>{\PBS\centering}p{\ljlenb}>{\PBS\centering}p{\ljlenb}}
408: \setp{100mm}{131mm}{65mm}{1.2}
409: \resizebox{\ljlenb}{!}{\rotatebox{0}{\includegraphics*[\xa,\ya][\xb,\yb]{selstau-208-87-84-030-mother5-2.eps}}} &
410: \setp{148mm}{132mm}{55mm}{1.2}
411: \resizebox{\ljlenb}{!}{\rotatebox{0}{\includegraphics*[\xa,\ya][\xb,\yb]{selstau-208-87-84-030-mother4-2.eps}}} \\
412: a). & b). \\
413: & \\
414: \setp{100mm}{130mm}{75mm}{1.2}
415: \resizebox{\ljlenb}{!}{\rotatebox{0}{\includegraphics*[\xa,\ya][\xb,\yb]{selstau-208-87-84-030-mother3-2.eps}}} &
416: \setp{105mm}{55mm}{65mm}{1.2}
417: \resizebox{\ljlenb}{!}{\rotatebox{0}{\includegraphics*[\xa,\ya][\xb,\yb]{selstau-208-87-84-030-mother2-2.eps}}} \\
418: c). & d). \\
419: \end{tabular}
420: }
421: \caption{Examples of the typical reconstructed detector response to the slepton and lepton for decreasing slepton decay lengths. Shown are \ladd{(a)} a TPC-vertex, \ladd{(b)} an ITC-vertex, \ladd{(c)} pointing-hits and \ladd{(d)} a simple high-\dzero\ track from signal Monte Carlo data. \ladd{All tracks and hits other than those of the slepton-lepton pair shown have been removed for clarity.}}
422: \label{mother2to5}
423: \end{figure}
424: \setlength{\topsep}{\ljlen}
425: \par
426: The relationship between the point of origin range and the possible
427: scenarios is as follows. If, for at least part of the range,
428: \begin{itemize}
429: \item the track radius\iforig{}{ (\ie\ the distance of the track from the $z$-axis)} is above that of the 4th TPC pad row, then a TPC-vertex is possible.
430: \item the track radius is above that of the 4th ITC wire layer and the track is either inside the ITC or the straight line from the track to the IP enters the ITC at a radius greater than this (and thus the slepton, under the assumption that its trajectory was straight, crossed at least 4 ITC wires), then an ITC-vertex is possible.
431: \item the straight line from the track to the IP passes through some tracking components, then pointing-hits are possible.
432: \item the straight line from the track to the IP passes through no tracking components, then no slepton response at all is possible.
433: \end{itemize}
434: These possibilities are not mutually exclusive, they could all be
435: open. However, of those that are open, it may be possible to exclude
436: some based on likelihood. For instance, if the track radius is
437: significantly above that of the 4th TPC pad row for the entire range,
438: then anything other than a TPC-vertex is unlikely, and a TPC-vertex
439: can be demanded. Thus if any of these conditions, with $3\cm$ added to
440: the radius necessary for a TPC-vertex and $3\mm$ added to the radius
441: necessary for an ITC-vertex, are true over the entire range, then the
442: first of the respective possibilities becomes a demand. If the demand
443: is not satisfied, the track is rejected. If it is found that no single
444: response can be demanded in this way, then the track is tested for the
445: possible slepton responses in the order TPC-vertex, ITC-vertex,
446: pointing-hits. If the track does not satisfy any of the possible
447: responses (bearing in mind that if the case that the slepton did not
448: produce any hits has been identified as a possibility, then no
449: response is possible), then it is rejected.
450: \par
451: The procedure involved in looking for TPC-vertices, ITC-vertices and
452: pointing-hits will now be described.
453: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
454: \subsubsection{TPC-vertices}
455: The first step of testing a track for a TPC-vertex is to identify
456: candidates for the slepton track. These are required to have TPC hits
457: (by definition), a \dzero\ less than $2\cm$ and a \zzero\ less than
458: $2.5\cm$, and must not have been identified as tracks from a
459: nuclear-interaction or an ECAL splash-back. Its \dEdx\ (if \dEdx\
460: information is available for the track) must be no less than 10
461: standard deviations below that expected for a singly-charged particle
462: of mass $60\GeV$ (the approximate GMSB slepton mass limit), and the
463: track should show evidence that the particle responsible ceased to
464: follow the track trajectory before leaving the TPC (\ie\ be missing at
465: least one hit from its end without observable reason -- see
466: Table~\ref{misshit}). If there are no candidate slepton-tracks, there
467: can be no TPC-vertices.
468: \par
469: The high-\dzero\ track is tested for a good TPC-vertex with each
470: candidate slepton track. They must form a common vertex which lies on
471: the post-\dzero\ section of the slepton track and has a $\chi^2$ (from
472: YTOP) less than 30. The $z$-coordinates of the two tracks at the vertex
473: must differ by less than $7\cm$, the angle between them at the vertex
474: must be greater than $10^\circ$, and the last assigned hit of the
475: slepton track must not be at a radius more than $1\cm$ above that of
476: the vertex.
477: \par
478: The checks made on the high-\dzero\ track's hits are more complex,
479: since there are three possibilities for their locations. If the
480: decay-angle (defined here as the angle between the slepton and lepton
481: directions at the point of decay in the $r\phi$ plane) is less than
482: 90$^\circ$ then all the high-\dzero\ track's hits lie at radii greater
483: than the vertex radius, and both the track directions are correct and
484: so the decay-angle will be calculated correctly (see
485: Figure~\ref{mother2to5}.a). If the decay-angle is greater than
486: 90$^\circ$ and the high-\dzero\ track is post-\dzero, then again the
487: track directions are correct and the decay-angle will be calculated
488: correctly, but now the hits can lie both above and below the vertex
489: radius so no constraints can be made on their positions (see
490: Figure~\ref{tpcver}). However, if the decay-angle is greater than
491: 90$^\circ$ and the high-\dzero\ track is pre-\dzero\ (so all its hits
492: lie at a radius smaller than that of the vertex), then the decay-angle
493: will be calculated incorrectly to be less than 90$^\circ$. So if the
494: calculated decay-angle is less than 90$^\circ$, all the hits are
495: required to lie \textit{either} at radii above that of the vertex
496: minus 1\cm, \textit{or} at radii below that of the vertex plus
497: 1\cm. If the calculated decay angle is greater than 90$^\circ$ then no
498: constraints are made on the positions of the hits. Thus if a
499: high-\dzero\ track is found to have a TPC-vertex it can be tagged as
500: post-\dzero\ or pre-\dzero\ based on the positions of its hits
501: relative to the vertex.
502: \begin{figure}[btp]
503: \setr{130mm}{142mm}{150mm}{0.22}
504: \resizebox{\textwidth}{!}{\includegraphics*[\xa,\ya][\xb,\yb]{tpcver.eps}}
505: \caption{An example of a TPC-vertex with a decay angle greater than 90$^\circ$. \ladd{The lepton has formed a pre-\dzero\ and a post-\dzero\ track. The arrows show the track directions as assigned by JULIA.}}
506: \label{tpcver}
507: \end{figure}
508: \par
509: At this stage, the number of bad TPC-points for the track (see
510: Section~\ref{origin}) can be fixed. For pre-\dzero\ tracks it is set as the
511: number lying at least $6.5\cm$ (normal pad-row spacing plus $1\mm$)
512: below the radius of the vertex. For post-\dzero\ tracks it is the number
513: lying above the radius. The number of bad TPC points forms a track
514: variable which can be used as an indicator of the reliability of the
515: track, and is referred to in the selections described in the next
516: chapter.
517:
518:
519: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
520: \subsubsection{ITC-vertices}
521: As with TPC-vertices, the first step towards finding ITC-vertices is
522: to identify candidate slepton tracks. These are required to have no
523: TPC hits (by definition), a \dzero\ less than $1\cm$ (no constraint on \zzero\
524: due to the low ITC $z$-resolution) and a momentum greater than $1\GeV$.
525: \par
526: The high-\dzero\ track is then tested for a good ITC-vertex with each
527: candidate slepton track. Because of the low $z$-resolution of the ITC
528: a full vertex fit is not performed. It is merely required either that
529: the tracks' circular $r\phi$ projections cross in the $r\phi$ plane,
530: in which case the crossing point closest to the beam axis is taken as
531: the two-dimensional vertex, or that their distance of closest approach
532: in $r\phi$ is less than $5\mm$, in which case the vertex position is
533: the point of closest approach.
534: \par
535: The vertex radius is required to be no more than $5\mm$ below that of
536: the 4th ITC wire layer. The upper limit on the radius is determined by
537: a scan along the ITC track outwards into the TPC, looking for evidence
538: that the particle responsible was not following the track
539: trajectory. At successive pad-row radii a set of conditions similar to
540: those summarised in the TPC section of Table~\ref{misshit} are
541: evaluated, with the $z$ requirements relaxed\footnote{The separation
542: in $z$ for \iforig{proxmity}{proximity} to another track to be
543: considered a valid reason for missing a hit is set at 10\cm\ for both
544: ITC and TPC tracks, and no $z$ information at all is used in judging
545: the proximity of unassigned hits.}. The upper limit on the radius is
546: then set at the radius of the second point at which the track is
547: missing a hit without observable reason plus $5\mm$, unless the first
548: and second points are consecutive in which case the first is used. The
549: first point is in general not used since the track errors in the TPC
550: are much greater in the absence of assigned TPC hits, and so some
551: leeway is appropriate.
552: \par
553: The ITC track is required \iforig{to not}{not to} have any assigned
554: hits on wire-layers above the layer that is itself just above the
555: vertex. The constraints on the TPC hits of the high-\dzero\ track are
556: the same as for TPC-vertices, and so if it passes it is known whether
557: it is post-\dzero\ or pre-\dzero, and its number of bad TPC points is
558: fixed in the same way. If it is found to be post-\dzero\ with a
559: decay-angle less than 90$^\circ$ then it is further required to have
560: no ITC hits below the wire layer that itself is just below the vertex,
561: and no VDET hits at all.
562: \par
563: While TPC-vertices are very rare in background events, ITC-vertices
564: are less so because of the weaker constraints in the $z$ direction, and
565: because multiple scattering in the ITC outer wall and TPC inner wall
566: can cause the ITC and TPC hits of a particle to be reconstructed as
567: separate tracks, giving vertices very similar to those in signal. Some
568: rejection of these is attained by the requirement that the decay-angle
569: (corrected in the case of pre-\dzero\ tracks) is greater than
570: 15$^\circ$.
571:
572: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
573: \subsubsection{Pointing-hits}
574: The object here is to find a line of hits compatible with marking the
575: slepton's trajectory from the IP to the high-\dzero\ track. Although the
576: slepton's trajectory is helical, it should have sufficient momentum
577: that it can be considered straight over the distances involved. This
578: is done to keep matters simple, but a small curvature of the slepton's
579: path can be accommodated in the leeway used in the following
580: procedure.
581: \par
582: The two endpoints of the particle's point-of-origin range are used to
583: define the $\phi$-range in which the line of hits should lie. If the
584: track lies at sufficiently high $|\cos(\theta)|$ over the whole range,
585: then the straight line from the IP to the track at any point in the
586: range will not cross any tracking components and it is therefore
587: possible that the slepton did not produce any hits. The track is
588: accepted without further requirements in this case. If this is not the
589: case, the procedure continues to look for pointing-hits using
590: hypothesised slepton trajectories.
591: \par
592: Firstly it is determined if a slepton hit is required in the first
593: VDET layer. This is the case if the track radius is above the minimum
594: VDET layer-1 radius (VDET layers are not cylindrical and so do not sit
595: at a single radius), and the track $|\cos(\theta)|$ is below the
596: maximum VDET layer-1 $|\cos(\theta)|$ for the entire range. In this
597: case, unassigned layer-1 VDET hits are sought in the relevant $\phi$
598: range. If none are found the track is rejected. If one or more are
599: found, they dictate the possible trajectories for the parent
600: slepton. These are straight lines from the IP with $\phi$'s equal to
601: those of the VDET hits, and $\theta$'s such that they intersect the
602: track. If they also intersect the second VDET layer, an unassigned hit
603: is required here also, with a $\phi$ within $1.2^\circ$ of that of the
604: first VDET hit (\ie\ no more than $\sim2\mm$ from the slepton
605: trajectory given by the first hit). Trajectories for which a second
606: VDET hit is expected but not observed are rejected. If a slepton hit
607: in the VDET is only required for part of the track's point-of-origin
608: range, then VDET hits in the relevant $\phi$ range are still sought,
609: but the track is not rejected if none are found. For the part of the
610: range where slepton hits are not required (which may be all of the
611: range), hypothesis slepton trajectories are taken from the $\phi$'s of
612: the ITC wires in the first layer that lie in the relevant $\phi$
613: range. For each trajectory which crosses at least one ITC wire-layer,
614: ITC information is required to support the hypothesis that the slepton
615: followed this trajectory. This is tested by applying the conditions
616: summarised in the ITC section of Table~\ref{misshit} at each
617: wire-layer radius. If a slepton trajectory is found not to be valid in
618: this way, it is rejected. If the remaining number of slepton
619: trajectories is non-zero, the track is considered to have
620: pointing-hits.
621:
622: \subsubsection{A note on pre-\dzero\ tracks}
623: It was mentioned in Section~\ref{origin} that the procedure for
624: calculating the point-of-origin range assumes the track is post-\dzero,
625: and is nonsensical for pre-\dzero\ tracks. However, the ordering in
626: which the procedures are applied in the computer code is not exactly
627: as is implied in this chapter. In particular, the procedure for
628: finding TPC-vertices precludes the possibility that a TPC-vertex could
629: be found which is incompatible with the relevant track's point of
630: origin range. So the first thing that is actually done is that a
631: TPC-vertex is sought. If one is found then the track is accepted
632: without the particle point-of-origin range ever being
633: calculated. While this makes sense from a pragmatic viewpoint since it
634: speeds up the analysis, it also solves the problem of pre-\dzero\
635: tracks. Since a pre-\dzero\ TPC track can only be formed if the
636: particle point-of-origin is at a radius greater than that of the 4th
637: TPC pad row (the particle will need to create at least 4 TPC hits
638: before the \dzero-point), it is reasonably safe to say that the
639: slepton must also have crossed 4 TPC pad rows, and so also have a
640: reconstructed TPC track, and so the track should have a
641: TPC-vertex. The unconditional acceptance of tracks with TPC-vertices
642: therefore means that pre-\dzero\ tracks are treated correctly.
643:
644: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
645: \subsection{The track variable, \mother}
646: All tracks which are tested for compatibility with the hypothesis that
647: they originated from slepton decay using the procedures described in
648: this chapter, are assigned an integer value to indicate the level and
649: type of success. This value is the track variable, \mother.
650: \par
651: An \mother\ of 0 indicates that the track is not compatible with the
652: hypothesis, and such tracks are rejected as good high-\dzero\ tracks. A
653: track has an \mother\ of 5 if it has a TPC-vertex, an \mother\ of 4 if
654: it has an ITC-vertex, an \mother\ of 3 if it has pointing-hits, and an
655: \mother\ of 1 or 2 if the track is compatible with the hypothesis that
656: the slepton did not produce any hits. A value of 2 is
657: given in the case that the track has sufficiently small \dzero\ that it is
658: possible that the slepton decayed before reaching a high enough radius
659: to produce a hit. A value of 1 is assigned when the slepton did reach
660: a high enough radius, but had a $|\cos(\theta)|$ high enough that it
661: still managed not to pass through any tracking components before
662: decay. This distinction is drawn since an \mother\ of 1 is the most
663: unlikely of the values under the signal hypothesis, but this relative
664: probability is not mirrored in the background, where detector effects,
665: cosmic muons and beam-gas events mean that an \mother\ of 1 is not so
666: improbable.
667: \par
668: Examples of tracks with \mother's from 2 to 5 are shown in
669: Figure~\ref{mother2to5}, and an example of a track with an \mother\ of
670: 1 is shown in Figure~\ref{mother1}.
671:
672: \begin{figure}[htbp]
673: \setlength{\ljlenb}{7cm}
674: \centerline{
675: \renewcommand{\arraystretch}{1}
676: \begin{tabular}{>{\PBS\centering}p{\ljlenb}>{\PBS\centering}p{\ljlenb}}
677: %\setp{73mm}{90mm}{110mm}{1}
678: %\resizebox{\ljlenb}{!}{\includegraphics*[\xa,\ya][\xb,\yb]{selstau-208-103-84-080-mother1-xy-2.eps}} &
679: %\setp{95mm}{150mm}{150mm}{1}
680: %\resizebox{\ljlenb}{!}{\includegraphics*[\xa,\ya][\xb,\yb]{selstau-208-103-84-080-mother1-rz-2.eps}} \\
681: \setp{114mm}{117mm}{50mm}{1}
682: \resizebox{\ljlenb}{!}{\includegraphics*[\xa,\ya][\xb,\yb]{a1.eps}} &
683: \setp{74mm}{117mm}{80mm}{1}
684: \resizebox{\ljlenb}{!}{\includegraphics*[\xa,\ya][\xb,\yb]{a2.eps}} \\
685: a). & b). \\
686: & \\
687: \end{tabular}
688: }
689: \caption{An example of a track with $\mother=1$ from signal Monte Carlo\ladd{, viewed in (a) the $r\phi$ plane and (b) the $\rho z$ plane ($\rho$ is equal in magnitude to $r$, but with a sign that makes it negative for a certain half of the $\phi$ range which is chosen to make the event as clear as possible). Although the \dzero\ of the track shows that the slepton reached a radius above at least one VDET layer before decaying, it did so at such a high $|\cos(\theta)|$ that it did not produce hits.}}
690: \label{mother1}
691: \end{figure}
692:
693: % This is splet wrung.
694: