hep-ph0108207/ANALYSIS/trigger.tex
1: \subsection{Trigger Selection\label{sec:trigger}}
2: The cross sections for SUSY production at the LHC for the points studied in
3: this work  range from $10^{-8}$--$10^{-12}$mb. 
4: For comparison, the cross section for e.g.\ $Z/W$ production is
5: $10^{-3}$mb, and the total interaction cross section 
6: (excepting elastic and diffractive
7: processes which do not give rise to hard jets, leptons, or neutrinos) 
8: is around 70mb, mostly consisting of small-angle QCD interactions. 
9: Background rejection is thus extremely essential. In an ideal world, all
10: data recorded by the experiment would be written to disk for later
11: scrutiny. With the extreme event rates and sizes (approximately 1
12: billion events per second and approximately 1MB per event\cite{atlastdr}) 
13: at the LHC,
14: such a strategy is entirely out of the question, moreover it is completely
15: uncalled for. The extreme rates are only
16: necessary to enable us to \emph{reach} the very small Higgs and New Physics
17: production cross sections. As already mentioned, the vast majority of events
18: are QCD processes with small momentum transfers. Relative to how much
19: statistics we need to study \emph{those} processes, we get a huge
20: over-abundance. We can then accept to limit the data flow by cutting away a
21: significant fraction of those events and writing only a very small subset of
22: them to disk, along with events which contain possible traces of low cross
23: section physics. To accomplish this reduction, an extensive trigger system is
24: being developed for ATLAS which 
25: will identify the (possibly) interesting
26: events and reject uninteresting ones. For SUSY,
27: typical signatures that can be triggered on 
28: include a lot of missing transverse energy (\ET) and
29: large (hard) jet and lepton multiplicities\footnote{Large
30: multiplicity = many particles.}.
31: The usefulness of the ``classical'' SUSY \ET\ signature 
32: will in general be decreased in \RV\ scenarios due to LSP decay, but
33: it remains of some discriminating power. On the other hand, the LSP decays to
34:  jets, leptons, and/or neutrinos, and so the signature is merely transformed,
35:  not lost. We can use this to define a set of triggers that will reject the
36: QCD background (which typically has small \ET\ and jet/lepton
37: multiplicities) but keep the majority of those events that might have been
38: caused by the production and decay of supersymmetric particles. 
39: 
40: \subsubsection{\LV-SUSY Triggers at Mid-Luminosity}
41: The acceptable rate of events that can be written to disk is about
42: 100Hz. The sum rate of \emph{all} the triggers which will be implemented in
43: ATLAS must therefore stay below this number. This includes triggers for Higgs
44: physics, $B$ physics, $Z/W$ physics, and all kinds of New Physics triggers,
45: of which \RV-SUSY is only one. A reasonable aim for the trigger rate for the
46: \RV-SUSY triggers is therefore about 1Hz. Since no detailed trigger rate
47: studies for mid-luminosity running are yet available, I here present some
48: crude estimates for the trigger rates for various trigger possibilities.
49: It should
50: be understood that there are quite significant uncertainties associated with 
51: these, as one might well imagine when already uncertain quantities are
52: extrapolated over orders of magnitudes in their domain. The jet rates,
53: especially, are not well under control, although interesting 
54:  work is in progress \cite{webber00,forshaw99,seymour00}.
55: It has not been possible here to perform a systematic study
56: of the effects of varying QCD parameters on the trigger rates, 
57: yet it is useful in the following to keep in mind that the multi-jet rates
58: can be uncertain by factors of 2 or more \cite{atlfast2.0}.
59: 
60: The only processes which have rates above the 1Hz
61: domain are QCD $2\to2$ processes and $Z/W$ production, the former with a
62: cross section around 70mb (but strongly peaked distributions), the latter
63: with a cross section around $10^{-3}$mb with more broad
64: distributions
65: and so more likely to be triggered on (By ``$Z/W$'' production is meant the
66: sum of $Z$, including photon interference, and $W$ production). 
67: In this crude study, we shall not focus on
68: where in the trigger system which triggers are placed (i.e.\ LVL1, LVL2, or
69: LVL3). This is a technical issue which requires more detailed knowledge of
70: the detector performance at mid-luminosity than is currently available in
71: \atlfast. Specifically, running at mid- and high luminosity means that there
72: can be several interactions recorded simultaneously by the detector (pile-up),
73: leading to some smearing of event distributions. Since the whole point is
74: that most events should lie \emph{outside} the trigger domain than inside it, 
75: more events will be smeared \emph{into} the trigger domain than out of it,
76: increasing the trigger rates when pile-up is at play. This is
77: currently only parametrized in \atlfast\ for high luminosity. We here adopt a
78: crude buest-guess approach, performing the simulation for low luminosity, i.e.\
79: without pile-up, and 
80: multiplying the resulting trigger rates by 5 instead of 3 to get from
81: $L=\tn{33}\scm$ to $L=3\ttn{33}\scm$. This number is based on 
82: the scaling exhibited
83: by the trigger rates presented in \cite[chp.11]{atlastdr} from
84: $L=\tn{33}\scm$ to $L=\tn{34}\scm$ where 
85: factors between 1 and 5 are found between triggers at equal thresholds 
86: after dividing out the factor 10 caused by the 
87: luminosity difference. 
88: The triggers for which this direct comparison was possible were
89: the inclusive electron, electron/photon, and \ET+2jets triggers.
90: Since we are here interested in going from $L=\tn{33}\scm$ to
91: $L=3\ttn{33}\scm$, a factor of 5/3 was deemed suitable.
92: These rates should then
93: to be understood as corresponding to LVL3 trigger rates. Suitable 
94: intermediate trigger levels will presumably be studied in more detail and
95: with full detector simulation by the ATLAS trigger/DAQ community in the near
96: future. 
97: 
98: Returning to the issue at hand, we now propose 11 
99: trigger menus for SUSY searches in \LV\ scenarios (bearing in mind the many
100: possible signatures in these scenarios).  
101:  Granting that the LSP might be the neutralino even in \LV\ scenarios, it is
102: tempting to define triggers based on the neutralino decay channels, $qq\nu$,
103: $qq\ell$, and $\ell\ell\nu$. This would work well at all of the SUSY points
104: analyzed here since all of them have a neutralino LSP, but one should keep in
105: mind that the lightest neutralino need not be the LSP when $R$ is
106: violated. Let us therefore go back to the terms in the superpotential that we
107: are after: $LLE$ and $LQD$, i.e.\ a coupling between three (s)leptons and a
108: coupling between two (s)quarks and one (s)lepton, where only one of the
109: particles coupling in each case will be a sparticle. 
110: From this very general standpoint, 
111: obvious trigger strategies suggest themselves. 
112: 
113: In the case of the first
114: coupling (LLE), 
115: we are searching for a number of leptons (depending on whether one
116: or two supersymmetric particles were produced and whether they decayed into
117: two- or three-body channels), possibly accompanied by
118: \ET. 
119: %\begin{figure}[ht!]
120: %\setlength{\extrarowheight}{0pt}
121: %\begin{center}
122: %\begin{tabular}{cccc}\toprule
123: %\multicolumn{4}{c}{\large\sc Trigger Rates}\\
124: %\cmidrule{1-4}
125: %\includegraphics*[scale=0.17]{PLOTS/tr_legend.eps}
126: %& \includegraphics*[scale=0.17]{PLOTS/tr_2mui.eps}
127: %& \includegraphics*[scale=0.17]{PLOTS/tr_2ei.eps} 
128: %&
129: %\includegraphics*[scale=0.17]{PLOTS/tr_eimu15i.eps}
130: %\\ \cmidrule{1-4}
131: % \includegraphics*[scale=0.17]{PLOTS/tr_memu40i.eps}
132: %& \includegraphics*[scale=0.17]{PLOTS/tr_mee40i.eps}
133: %& \includegraphics*[scale=0.17]{PLOTS/tr_1jmu40i.eps}
134: %& \includegraphics*[scale=0.17]{PLOTS/tr_1je40i.eps} 
135: %\\
136: %\cmidrule{1-4}
137: %\includegraphics*[scale=0.17]{PLOTS/tr_1jme175.eps}
138: %& \includegraphics*[scale=0.17]{PLOTS/tr_3jmu20i.eps}
139: %& \includegraphics*[scale=0.17]{PLOTS/tr_3je30i.eps}
140: %& \includegraphics*[scale=0.17]{PLOTS/tr_3jme125.eps}
141: %\\\bottomrule
142: %\end{tabular}
143: %\caption[\small Trigger rates]{The rate of events (in Hz) passing the various
144: %trigger $p_T$ thresholds  
145: %as functions of the $p_T$ thresholds. For the di-muon and di-electron
146: %triggers \emph{both} electrons/muons are required to pass the threshold.
147: % The reason the rates are not all strictly decreasing functions of their
148: %respective thresholds is due to the
149: %way electron and muon reconstruction efficiencies were taken into account in
150: %the analysis (see text). These and several other plots are collected
151: %in Appendix \ref{app:trigger} for reference together with further comments on
152: %the shapes and the numbers of events generated to make them. 
153: %\label{fig:triggerrates}}
154: %\end{center}
155: %\end{figure}
156: We do not here consider an all-neutrino signature since this would be
157: equivalent to the $R$-conserving SUSY \ET\ signature. 
158: We require: at least two isolated electrons and/or
159: muons or one isolated $e$ or $\mu$ accompanied by \ET. 
160: 
161: For the $LQD$ term, we expect jets accompanied by leptons or
162: \ET. 
163: The triggers proposed are: at least 3 jets accompanied by either \ET\ or
164: at least one isolated $\mu$ or $e$, or at least 1 jet, accompanied by
165: the same with higher thresholds. Each trigger was studied for several
166: possibilities and combinations of trigger thresholds. The complete 
167: scans around the selected threshold values are 
168: shown and commented in Appendix \ref{app:trigger}. 
169: \begin{table}[t]
170: \setlength{\extrarowheight}{2pt}
171: \begin{center}
172: \begin{tabular}{lcrrr}
173: \toprule
174: Trigger & \begin{tabular}{c}Background\\Rate\end{tabular} & 
175: \begin{tabular}{c}MSSM\\Efficiency\end{tabular} &
176: \begin{tabular}{c}LLE\\Efficiency\end{tabular}&\begin{tabular}{c}LQD\\Efficiency\end{tabular}\\
177: \cmidrule{1-5}
178: mu45I + mu45I   & 0.2 Hz & 1 --- 5 \% & 10 --- 40 \% & 1 --- 10 \% \\ % 0.22 
179: e45I  + e45I    & 0.1 Hz & 1 --- 5 \% & 1 --- 35 \% & 1 --- 10 \% \\ % 0.13 
180: mu15I + e15I    & 0.1 Hz & 2 --- 5 \% & 20 --- 60 \% & 2 --- 15 \% \\ % 0.09 
181: mu40I + me75    & 0.3 Hz & 10 --- 25 \% & 40 --- 75 \% & 10 --- 35 \% \\ % 0.25 
182: e40I  + me75    & 0.2 Hz & 10 --- 20 \% & 15 --- 70 \% & 10 --- 35 \% \\ % 0.22 
183: j100  + mu40I   & 0.5 Hz & 10 --- 20 \% & 45 --- 70 \% & 10 --- 40 \% \\ % 0.55 
184: j100  + e40I    & 0.5 Hz & 5 --- 15 \%  & 15 --- 65 \% & 10 --- 35 \% \\ % 0.61
185: j100  + me175   & 0.3 Hz & 50 --- 80 \% & 35 --- 80 \% & 25 --- 80 \% \\ % 0.68
186: 3j50  + mu20I   & 0.1 Hz & 5 --- 15 \% & 45 --- 60 \% & 12 --- 40 \% \\ % 0.20
187: 3j50  + e30I    & 0.1 Hz & 5 --- 10 \% & 15 --- 55 \% & 10 --- 35 \% \\ % 0.19 
188: 3j75  + me125   & 0.1 Hz & 30 --- 65 \% & 30 --- 70 \% & 40 --- 90 \% \\ % 0.43 
189: \cmidrule{1-5}
190: \begin{tabular}{l}Total Rate\\/Combined Efficiency\end{tabular}&   2.1 Hz &
191: 60 --- 90 \% & 90 --- 99.9 \% & 60 --- 96 \%  
192: \\
193: \bottomrule
194: \end{tabular}
195: \caption[\small Trigger rates and efficiencies for $L=3\times 10^{33}\mathrm{cm}^{-2}\mathrm{s}^{-1}$.]{Estimated trigger rates for background
196: processes and trigger efficiency ranges for the various MSSM points and \LV-SUSY
197:   scenarios studied for $L=3\times 10^{33}\mathrm{cm}^{-2}\mathrm{s}^{-1}$
198:   (summary of plots in appendix \ref{app:trigger}). 
199: The trigger symbols are described in the text. The total
200:   trigger rate is slightly smaller than the sum of the individual rates, 
201: since one event can fulfil several trigger criteria. One sees how much
202:   cleaner the signatures of the purely leptonic (LLE) coupling are 
203:   compared to the signatures involving quarks (LQD) where higher thresholds,
204:   due to the hadronic environment, mean smaller efficiencies.
205: \label{tab:triggerrates}}
206: \end{center}
207: \vspace*{-\tfcapsep}\end{table}
208: Here, event rates for the selected values of the trigger cuts are given in
209: table \ref{tab:triggerrates} where also the efficiencies with which these
210: triggers allow signal events to pass are listed. 
211: The nomenclature for each trigger item
212: is such that e.g.\ `mu20I' and `e40I' mean `a muon with $p_T>20\GeV$ which is
213: isolated', and `an isolated electron with $p_T > 40\GeV$', 
214: respectively. `3j100' and `me80' mean three reconstructed 
215: jets of (each) more than 100\GeV\
216: $p_T$ and more than 80\GeV missing transverse energy, respectively. 
217: To save space, the signal trigger efficiencies are given as ranges
218: (min---max), 
219: obtained by dividing the scenarios studied 
220: into three categories and scanning for the minimum
221: and maximum efficiencies within each category. These categories are: the 
222: MSSM (for reference), purely leptonic \LV\ (i.e.\ decays proceeding
223: via the $\lambda$ couplings (LLE)), 
224: and mixed quark-lepton $L$-violation (the $\lambda'$
225: couplings (LQD)). 
226: Scenarios where both types of couplings are non-zero generally
227: lie inbetween the two extremes, depending on the relative coupling
228: strengths. 
229: The efficiency ranges reached by combining all the triggers are shown in
230: the bottom of the table.
231: \subsubsection{Final Remarks}
232: Though these trigger proposals are designed explicitly with
233: \RV-SUSY in mind, they show a certain overlap with triggers proposed for more
234: conventional physics. The di-muon and 3 jets + lepton triggers, 
235: for example, have also been
236: proposed for various Higgs searches. The di-electron
237: trigger as well as 3 jets + electron are proposed to catch $t\bar{t}$
238: decays. Finally, the conventional SUSY searches also make use of
239: multi-lepton, jets + \ET, and multi-jet signatures \cite{bystricky96}. It is
240: therefore not unthinkable that the triggers here proposed can be incorporated
241: to some extent into the conventional trigger programme, possibly enabling a
242: relaxation of the trigger thresholds for the remaining objects.   
243: 
244: Finally, though no study of the mutual redundancy among these trigger objects
245: has been performed, one sees from table \ref{tab:triggerrates} 
246: that e.g.\ the di-electron trigger and the 3-jet
247: + electron trigger have low efficiencies, this mainly due to the low electron
248: reconstruction efficiency which is required to be sure one is not mistakenly
249: believing a jet to be an electron (see \cite[table 7-1]{atlastdr} for
250: estimated jet rejection factor as function of electron reconstruction
251: efficiency). As a suggestion for future analyses, it
252: might be worthwile to investigate what happens if one dispenses 
253: with these two triggers, replacing them with 
254: e.g.\ two-jet + leptons or two-jet + \ET\ triggers instead, yet caution in
255: such undertakings is advisable. The presence of 
256: electron triggers is, for example,
257: essential in catching scenarios where the first generation \LV\ couplings are
258: dominant. This is the reason why as little dependence on lepton flavour 
259: as possible has been strived for in the menus here proposed.
260: