1: \documentstyle[12pt,epsfig]{article}
2: \oddsidemargin=0mm
3: \evensidemargin=0mm
4: \topmargin=-20mm
5: \textwidth=170mm
6: \textheight=240mm
7: \begin{document}
8: \baselineskip=20pt
9:
10: %%%%%%%%%%%%%%%%%%%%%%%% NEW DEFINITIONS
11: \def\la{\mathrel{\mathpalette\fun <}}
12: \def\ga{\mathrel{\mathpalette\fun >}}
13: \def\fun#1#2{\lower3.6pt\vbox{\baselineskip0pt\lineskip.9pt
14: \ialign{$\mathsurround=0pt#1\hfil##\hfil$\crcr#2\crcr\sim\crcr}}}
15: %%%%%%%%%%%%%%%%%%%%%%%% TITLE PAGE
16:
17: \begin{titlepage}
18: \begin{center}
19: {\Large \bf Extracting the jet azimuthal anisotropy from
20: higher order cumulants} \\
21:
22: \vspace{4mm}
23:
24: I.P.~Lokhtin$^\dag$,
25: L.I.~Sarycheva$^\ddag$ and
26: A.M.~Snigirev$^\S$ \\
27: M.V.Lomonosov Moscow State University, D.V.Skobeltsyn Institute of Nuclear Physics \\
28: 119992, Leninskie Gory, Moscow, Russia
29: \end{center}
30:
31: \begin{abstract}
32: We analyze the method for calculation of a coefficient of jet azimuthal
33: anisotropy without reconstruction of the nuclear reaction plane considering
34: the higher order correlators between the azimuthal position of jet axis and
35: the angles of particles not incorporated in the jet. The reliability of this
36: technique in the real physical situation under LHC conditions is illustrated.
37: \end{abstract}
38:
39: \bigskip
40:
41: \vspace{100mm}
42: \noindent
43: ---------------------------------------------\\
44: $^\dag$ E-mail: igor@lav01.sinp.msu.ru\\
45: $^\ddag$ E-mail: lis@alex.sinp.msu.ru\\
46: $^\S$ E-mail: snigirev@lav01.sinp.msu.ru\\
47: \end{titlepage}
48: \newpage
49:
50: \section {Introduction}
51:
52: Nowadays the strong interest is springing up to the investigations and measurements
53: of azimuthal correlations in ultrarelativistic heavy ion collisions
54: (see, for instance,~\cite{qm01} and references therein). One of the main
55: reasons is that the rescattering and energy loss of hard partons in the azimuthally
56: non-symmetric volume of dense quark-gluon matter, created initially in the nuclear
57: overlap zone in collisions with non-zero impact parameter, can result in the visible
58: azimuthal anisotropy of high-$p_T$ hadrons at RHIC~\cite{qm01,uzhi,wang00,gyul00}.
59:
60: Recent anisotropic flow data at RHIC~\cite{star,phenix,phobos} can be described well by
61: hydrodynamical models for semi-central collisions and $p_T$ up to $\sim 2$
62: GeV/c (the elliptic flow coefficient $v_2$ appears to be monotonously growing with
63: increasing $p_T$~\cite{kolb} in this case), while the majority of microscopical Monte-Carlo models
64: underestimate the flow effects (see however~\cite{zabrodin}). The saturation and
65: gradual decrease of $v_2$ at relatively large transverse momentum ($p_T \ga 2$ GeV/c),
66: predicted as a signature of strong partonic energy loss in a dense QCD plasma, seem now
67: to be supported by the preliminary
68: data extending up to $p_T \simeq 10$ GeV/c at RHIC. The interpolation between the
69: low-$p_T$ relativistic hydrodynamics region and the high-$p_T$ pQCD-computable region
70: was evaluated in~\cite{gyul00}.
71:
72: The initial gluon densities in Pb$-$Pb reactions at $\sqrt{s}_{NN}=5.5$~TeV
73: at the Large Hadron Collider (LHC) are expected to be significantly higher
74: than at RHIC, implying stronger partonic energy loss. Moreover, since at LHC
75: energies the inclusive cross section for hard jet production at
76: $E_T \sim 100$ GeV is large enough to study the impact parameter
77: dependence of such processes~\cite{lokhtin00}, one can hope to observe the
78: azimuthal anisotropy for hadronic jet itself~\cite{lokhtin01,Lokhtin:2002}.
79: In particular, CMS
80: experiment at LHC~\cite{cms94} will be able to provide both the jet reconstruction
81: and adequate measurement of impact parameter of nuclear collision using
82: calorimetric information~\cite{note00-060}. In the case of jets, the methodical
83: advantage of azimuthal observables is that one
84: needs to reconstruct only azimuthal position of the jet without measuring its total
85: energy. It can be done more easily and with high accuracy, while the
86: reconstruction of jet energy is a more ambiguous problem~\cite{note00-060}.
87: However the measurement of jet production as a function of azimuthal angle
88: requires event-by-event determination of the nuclear event plane based on
89: the anisotropic flow analysis.
90:
91: The methods for elliptic flow analysis can be generally divided in two categories:
92: two-particle methods suggested and summarized in works~\cite{voloshin,wang,ollit} and
93: multi-particle methods~\cite{ollitr,din}. In two-particle methods the contribution of
94: non-flow (non-geometric) correlations to the determination of the elliptic flow
95: coefficient $v_2$ is of the order of $1/\sqrt{N_0}$, where $N_0$ is the measured
96: multiplicity. In multi-particle methods this contribution goes down typically as
97: $1/N_0^{3/4}$, i.e., smaller by a factor of the order of $N_0^{1/4}$. Thus experimental
98: techniques based on higher order cumulant analysis should be able in many cases to
99: allow access to the smaller values of azimuthal particle anisotropy in comparison with
100: two-particle methods, due to automatic elimination of the major non-flow many-particle
101: correlations and the systematic errors originating from azimuthal asymmetry of the
102: detector acceptance.
103: This kind of analysis for particle flow has been already done by the
104: STAR Collaboration at RHIC~\cite{star02}.
105:
106: In our previous Letter~\cite{phl02} we proposed the method for measurement of
107: jet azimuthal anisotropy coefficients without direct reconstruction of the
108: event plane, and illustrated its reliability in a real experimental situation.
109: This technique is based on the calculation of correlations between the azimuthal
110: position of the jet axis and the angles of particles not incorporated in the jet, the
111: azimuthal distribution of jets being described by the elliptic form. To improve our
112: approach, in the given paper we extend our analysis~\cite{phl02}, considering the
113: cumulant expansion~\cite{ollitr} of multi-particle azimuthal correlations.
114:
115: \section{Correlators versus the jet elliptic anisotropy}
116:
117: Let us remind some features of our previous investigation~\cite{phl02}.
118: We start from the essence of techniques~\cite{voloshin, wang, ollit} for
119: measuring azimuthal elliptic anisotropy of particle distribution, which can be
120: written in the form
121: \begin{equation}
122: \label{phi_part}
123: \frac{dN}{d \varphi} = \frac{N_0}{2\pi}~ [1+2v_2\cos{2
124: (\varphi -\psi_R)}]~,~~~~
125: N_0 = \int\limits_{-\pi}^{\pi}d\varphi~\frac{dN}{d \varphi}~ .
126: \end{equation}
127: Knowing the nuclear reaction plane angle $\psi_R$ allows one to determine the
128: coefficient $v_2$ of azimuthal anisotropy of particle flow as an average (over
129: particles) cosine of $2 \varphi$:
130: \begin{equation}
131: \label{v2_part}
132: < \cos{2(\varphi-\psi_R)} >~=~
133: \frac{1}{N_0}~\int\limits_{-\pi}^{\pi}d\varphi~\cos{2(\varphi -\psi_R)}~
134: \frac{dN}{d \varphi}~=~v_2~.
135: \end{equation}
136: However in the case when there are no other correlations of particles except
137: those due to flow (or such other correlations can be neglected), the
138: coefficient of azimuthal anisotropy can be determined using two-particle
139: azimuthal correlator without the event plane angle $\psi_R$:
140: \begin{eqnarray}
141: \label{v2^2_part}
142: & & < \cos{2(\varphi_1-\varphi_2)} >~=~
143: \frac{1}{N_0^2}~\int\limits_{-\pi}^{\pi}d\varphi_1
144: ~\int\limits_{-\pi}^{\pi}d\varphi_2
145: ~\cos{2(\varphi_1 -\varphi_2)}~
146: \frac{d^2N}{d \varphi_1d\varphi_2}~\nonumber \\
147: & & =\frac{1}{N_0^2}~\int\limits_{-\pi}^{\pi}d\varphi_1
148: ~\int\limits_{-\pi}^{\pi}d\varphi_2
149: ~\cos{2((\varphi_1 -\psi_R) -(\varphi_2 -\psi_R))}~
150: \frac{dN}{d \varphi_1}~\frac{dN}{d\varphi_2}~=~v_2^2~.
151: \end{eqnarray}
152:
153: Let us consider now the event with high-p$_T$ jet (dijet) production, the
154: distribution of jets over azimuthal angle relatively to the reaction plane
155: being described well by the elliptic form~\cite{lokhtin01},
156: \begin{equation}
157: \label{phi_jet}
158: \frac{dN^{jet}}{d \varphi} = \frac{N^{jet}_0}{2\pi}~ [1+2v^{jet}_2\cos{2
159: (\varphi -\psi_R)}]~,~~~~
160: N^{jet}_0 = \int\limits_{-\pi}^{\pi}d\varphi~\frac{dN^{jet}}{d \varphi}~,
161: \end{equation}
162: where the coefficient of jet azimuthal anisotropy $v^{jet}_2$ is determined
163: as an average over all events cosine of $2 \varphi$,
164: \begin{equation}
165: \label{u2_jet}
166: \left< \cos{2(\varphi-\psi_R)} \right>_{event}~=~
167: \frac{1}{N^{jet}_0}~\int\limits_{-\pi}^{\pi}d\varphi~\cos{2(\varphi -\psi_R)}~
168: \frac{dN^{jet}}{d \varphi}~=~v^{jet}_2~.
169: \end{equation}
170: One can calculate the correlator between the azimuthal position of jet axis
171: $\varphi _{jet}$\footnote{The other possibility is to fix the azimuthal position of
172: a leading particle in the jet. In this case calculating azimuthal correlations can
173: provide the information about azimuthal anisotropy of high-$p_T$ particle spectrum.}
174: and the angles of particles, which are not incorporated in the jet(s). The value of
175: this correlator is related to the elliptic coefficients $v_2$ and $v^{jet}_2$ as
176: \begin{eqnarray}
177: \label{cor}
178: \left< < \cos{2(\varphi _{jet}-\varphi)} >\right>_{\rm event} & = &
179: \frac{1}{N^{jet}_0N_0}~\int\limits_{-\pi}^{\pi}d\varphi _{jet}~
180: \int\limits_{-\pi}^{\pi} d\varphi~
181: \cos{2(\varphi _{jet} -\varphi)}~ \frac{dN^{jet}}{d
182: \varphi_{jet}}~\frac{dN}{d\varphi}\nonumber \\
183: & = & \frac{1}{N^{jet}_0}~\int\limits_{-\pi}^{\pi}d\varphi _{jet}~
184: \cos{2(\varphi _{jet} -\psi _R)}~ \frac{dN^{jet}}{d
185: \varphi_{jet}}~v_2~=~v^{jet}_2~v_2~.
186: \end{eqnarray}
187:
188: Using Eq. (\ref{v2^2_part}) and intermediate result in Eq. (\ref{cor}) (after averaging
189: over particles $\cos{2(\varphi _{jet} -\varphi)}$ reduces to $v_2 \cos{2(
190: \varphi _{jet} -\psi _R)}$) we derive the formula for computing absolute value
191: of the coefficient of jet azimuthal anisotropy (without reconstruction of sign of
192: $v^{jet}_2$):
193: \begin{equation}
194: \label{u2_event}
195: v^{jet}_2 ~=~ \left< \frac{< \cos{2(\varphi _{jet}-\varphi)} >}
196: {\sqrt{< \cos{2(\varphi_1-\varphi_2)} >}}
197: \right>_{\rm event}~.
198: \end{equation}
199: This formula does not require the direct determination of reaction plane angle
200: $\psi_R$. The brackets $\left< ~~~~\right>$ represent the averaging over
201: particles (not incorporated in the jet) in a given event, while the brackets
202: $\left< ~~~~\right>_{event}$ are the averaging over events.
203:
204: The formula (\ref{u2_event}) can be generalized by introducing as weights the
205: particle momenta,
206: \begin{equation}
207: \label{u2(p)_event}
208: v^{jet}_{2(p)} ~=~ \left< \frac{<~p_T(\varphi) \cos{2(\varphi _{jet}-\varphi)}
209: >} {\sqrt{<~p_{T1}(\varphi_1)~p_{T2}(\varphi_2)~ \cos{2(\varphi_1-\varphi_2)}
210: >}}
211: \right>_{\rm event}~.
212: \end{equation}
213: In this case the brackets $\left< ~~~~\right>$ denote the averaging over angles
214: and transverse momenta of particles. The other modification of
215: (\ref{u2(p)_event}),
216: \begin{equation}
217: \label{u2(E)_event}
218: v^{jet}_{2(E)} ~=~ \left< \frac{<~E(\varphi) \cos{2(\varphi _{jet}-\varphi)}>}
219: {\sqrt{<~E_1(\varphi_1)~E_2(\varphi_2) \cos{2(\varphi_1-\varphi_2)}>}}
220: \right>_{\rm event}~,
221: \end{equation}
222: ($E_i (\varphi_i)$~being energy deposit in a calorimetric segment $i$
223: of position $\varphi_i$) allows one using calorimetric measurements
224: (\ref{u2(E)_event}) for the determination of jet azimuthal anisotropy.
225:
226: \section{Higher order correlators}
227:
228: The main advantage of the higher order cumulant analysis is in the fact that,
229: as argued in Ref.~\cite{ollitr}, if flow is larger than non-flow correlations,
230: the contribution of the latter to $v_2$ extracted from higher order correlators
231: is suppressed\footnote{This can be essential under data analysis
232: with not high enough multiplicity of particles in an event.} by powers of
233: particle multiplicity $N_0$ in an event.
234:
235: Thus, for example, the fourth order cumulant for elliptic particle flow is
236: defined as~\cite{ollitr}
237: \begin{eqnarray}
238: \label{cum}
239: c_2[4]~~\equiv~~< \cos{2(\varphi_1+\varphi_2 -\varphi_3-\varphi_4)} >~~~~~~~~~~~~~~~~
240: ~~~~~~~~~~~~~~~~\nonumber\\
241: -~< \cos{2(\varphi_1 -\varphi_3)} >~< \cos{2(\varphi_2
242: -\varphi_4)} >
243: -~< \cos{2(\varphi_1 -\varphi_4)} >~< \cos{2(\varphi_2
244: -\varphi_3)} > ~,
245: \end{eqnarray}
246: and in the case of existing only correlations with the reaction plane (i.e.
247: the factorization of multi-particle distributions is held as in Eq.~(\ref{v2^2_part}))
248: is equal to
249: \begin{equation}
250: \label{cum-v_2}
251: c_2[4]~=~-v_2^4.
252: \end{equation}
253:
254: If now one defines the coefficient $v_2$ of azimuthal anisotropy through
255: two-particle correlator
256: \begin{equation}
257: \label{v_2-cor}
258: v_2~=~ \sqrt{< \cos{2(\varphi_1-\varphi_2)}>},
259: \end{equation}
260: then the contribution of non-flow correlations, as argued in~\cite{ollitr},
261: is of order $1/\sqrt{N_0}$. While their contribution to $v_2$, extracted from
262: the fourth order correlator
263: \begin{equation}
264: \label{v_2-cor4}
265: v_2~=~ (-c_2[4])^{1/4},
266: \end{equation}
267: scales as $1/N_0^{3/4}$, i.e. it is suppressed by an extra factor of $1/N_0^{1/4}$.
268: Corresponding data analysis based on Eq.~(\ref{cum}) and result (\ref{cum-v_2})
269: has been already carried out at STAR~\cite{star02}.
270:
271: Now using the derivation of Eq.~(\ref{u2_event}) and the result
272: (\ref{cum-v_2}) it is straightforward to obtain the formula for calculation
273: (measurement) of coefficient of jet azimuthal anisotropy through the higher
274: order correlator, which is less sensitive to non-flow correlations:
275: \begin{eqnarray}
276: \label{v2(4)}
277: v_2^{jet}[4]~=\bigg\langle\frac{1}{(-c_2[4])^{3/4}}
278: [~-< \cos{2(\varphi_{jet}+\varphi_1 -\varphi_2
279: -\varphi_3)} >~\nonumber\\
280: ~~~~~~~~~+~< \cos{2(\varphi_{jet} -\varphi_2)} >~
281: ~< \cos{2(\varphi_1
282: -\varphi_3)} >~\nonumber\\
283: ~~~~~~~~~+~< \cos{2(\varphi_{jet} -\varphi_3)} >~~< \cos{2(\varphi_1
284: -\varphi_2)} >]\bigg\rangle_{\rm event}.
285: \end{eqnarray}
286: Here we stress once more that in the case of existing only correlations with
287: the reaction plane, Eq.~(\ref{v2(4)}) together with Eq.~(\ref{u2_event})
288: transforms into identity. The formula just derived can be, as in Sect.~2,
289: generalized for calorimetric measurements of energy flows:
290: \begin{eqnarray}
291: \label{v2E(4)}
292: & &v_{2(E)}^{jet}[4]~=\bigg\langle\frac{1}{(-c_{2(E)}[4])^{3/4}}[~-< E_{1}(\varphi_1)
293: E_{2}(\varphi_2)E_{3}(\varphi_3)
294: \cos{2(\varphi_{jet}+\varphi_1 -\varphi_2
295: -\varphi_3)} >\nonumber\\
296: & &~~~~~~~~~+~< E_{2}(\varphi_2)\cos{2(\varphi_{jet} -\varphi_2)} >~
297: ~< E_{1}(\varphi_1)E_{3}(\varphi_3)\cos{2(\varphi_1
298: -\varphi_3)} >\nonumber\\
299: & &~~~~~~~~~+~< E_{3}(\varphi_3)\cos{2(\varphi_{jet}
300: -\varphi_3)} >~~<E_{1}(\varphi_1) E_{2}(\varphi_2)\cos{2(\varphi_1
301: -\varphi_2)} >]~\bigg\rangle_{\rm event},
302: \end{eqnarray}
303: where
304: \begin{eqnarray}
305: \label{cumE}
306: & & c_{2(E)}[4]~=~<E_{1}(\varphi_1)E_{2}(\varphi_2)E_{3}(\varphi_3)
307: E_{4}(\varphi_4)
308: \cos{2(\varphi_1+\varphi_2 -\varphi_3-\varphi_4)} >\nonumber\\
309: & &-~<E_{1}(\varphi_1) E_{3}(\varphi_3)\cos{2(\varphi_1 -\varphi_3)} >~~
310: <E_{2}(\varphi_2)E_{4}(\varphi_4) \cos{2(\varphi_2 -\varphi_4)} >\nonumber\\
311: & &-~< E_{1}(\varphi_1)E_{4}(\varphi_4)\cos{2(\varphi_1 -\varphi_4)} >~~
312: <E_{2}(\varphi_2)E_{3}(\varphi_3) \cos{2(\varphi_2 -\varphi_3)} >.
313: \end{eqnarray}
314:
315: In the case when the azimuthal position of jet axis correlates not only with
316: the reaction plane, one can try to improve this technique using the multiple
317: correlators of another form: averaging over not all events but selecting some
318: their sub-events. For instance, one can consider sub-events $1$ and $2$, when jets
319: are produced with the rapidity $y>0$ and $y<0$. Then calculating correlator
320: \begin{eqnarray}
321: \label{c2jet(4)}
322: c_2^{jet}[4]~=\bigg\langle\frac{1}{\sqrt{<\cos{2(\varphi_1-\varphi_2)}>
323: <\cos{2(\phi_1-\phi_2)}>}}
324: [~< \cos{2(\varphi_{jet}-\varphi +\phi_{jet}-\phi)} >\nonumber\\
325: +< \cos{2(\varphi_{jet}-\varphi -\phi_{jet}+\phi)} >\nonumber\\
326: ~~~~~~~~~-~< \cos{2(\varphi_{jet} -\varphi)} >~~< \cos{2(\phi_{jet}
327: -\phi)} >]\bigg\rangle_{{\rm sub-event}~1,~2},
328: \end{eqnarray}
329: we obtain that, if there are flow particle correlations only and
330: the distribution of jets over azimuthal angles is described by the
331: elliptic form (\ref{phi_jet}) in every sub-event, it is equal to
332: \begin{equation}
333: \label{cum(jet)-v_2}
334: c_2^{jet}[4]~=~v_2^{jet}(y>0)~v_2^{jet}(y<0).
335: \end{equation}
336: In Eq.~(\ref{c2jet(4)}) the angles $\varphi$ are defined as the azimuthal angles
337: of particles and jets in sub-event with $y>0$, and $\phi$ --- in sub-event with
338: $y<0$.
339: Correspondingly the brackets $<~~~~>$ represent the averaging over particles in
340: sub-events 1, 2, while the brackets
341: $\bigg\langle~~~\bigg\rangle_{\rm sub-event~1,~2}$ are the averaging over these
342: sub-events. The generalization of Eq.
343: (\ref{c2jet(4)}) for calorimetric measurements of energy flow is obvious
344: (similar to Eqs.~(\ref{u2(E)_event}) and (\ref{v2E(4)})). We do not write also
345: this result specially as the examples of utilizing six- and other higher order
346: correlators.
347:
348: \newpage
349:
350: \section{Non-flow correlations}
351:
352: Here we discuss the influence of non-flow correlations\footnote{See also
353: Appendix of work~\cite{ollitr} and Ref.~\cite{kov}} on the $v_2^{jet}$
354: determination. There are various sources of such correlations, among which
355: minijet production~\cite{kov}, global momentum conservation~\cite{dan, bor02},
356: resonance decays (in which the decay products are correlated), final state
357: Coulomb, strong or quantum interactions~\cite{bor}. We restrict our
358: consideration to two-particle correlations only. It will be enough to
359: illustrate the advantage in using higher order cumulants. In this case
360: multi-particle distributions are not factorized again and instead
361: of Eq.~(\ref{v2^2_part}) we have
362: \begin{eqnarray}
363: \label{c2^2_part}
364: & & c_2[2]~\equiv~< \cos{2(\varphi_1-\varphi_2)} >~=~
365: \frac{1}{N_2}~\int\limits_{-\pi}^{\pi}d\varphi_1
366: ~\int\limits_{-\pi}^{\pi}d\varphi_2
367: ~\cos{2(\varphi_1 -\varphi_2)}~
368: [\frac{dN}{d \varphi_1}~\frac{dN}{d \varphi_2}~+~
369: \frac{dN_{cor}}{d \varphi_1d\varphi_2}]\nonumber \\
370: & &~~~~~~~~~~~~~~~~~~~ =~v_2^2~\frac{1~+~v_{cor}^{-}/v_2^2}{1~+~\Delta}~,
371: \end{eqnarray}
372: where
373: \begin{eqnarray}
374: \label{norma}
375: & & N_2~=~
376: ~\int\limits_{-\pi}^{\pi}d\varphi_1
377: ~\int\limits_{-\pi}^{\pi}d\varphi_2~
378: [\frac{dN}{d \varphi_1}~\frac{dN}{d \varphi_2}~+~
379: \frac{dN_{cor}}{d \varphi_1d\varphi_2}]~,\nonumber \\
380: & & \Delta~=~\frac{1}{N_0^2}
381: ~\int\limits_{-\pi}^{\pi}d\varphi_1
382: ~\int\limits_{-\pi}^{\pi}d\varphi_2
383: ~~ \frac{dN_{cor}}{d \varphi_1d\varphi_2}~,\nonumber \\
384: & & v_{cor}^{-}~=~\frac{1}{N_0^2}
385: ~\int\limits_{-\pi}^{\pi}d\varphi_1
386: ~\int\limits_{-\pi}^{\pi}d\varphi_2
387: ~\cos{2(\varphi_1 -\varphi_2)}~
388: ~~ \frac{dN_{cor}}{d \varphi_1d\varphi_2}~.
389: \end{eqnarray}
390: Eq.~(\ref{cor}) remains unchangeable and result (\ref{u2_event}) transforms
391: into
392: \begin{equation}
393: \label{v2_event}
394: v^{jet}_2[2] ~\equiv~ \left< \frac{< \cos{2(\varphi _{jet}-\varphi)} >}
395: {\sqrt{< \cos{2(\varphi_1-\varphi_2)} >}}
396: \right>_{event}~=~v_2^{jet}~
397: \sqrt{ \frac{1~+~\Delta}{1~+~v_{cor}^{-}/v_2^2}}~.
398: \end{equation}
399:
400: After some algebra the fourth order cumulant (\ref{cum}) reduces to
401: \begin{eqnarray}
402: \label{c_2[4]}
403: & & c_2[4] ~=~
404: ~v_2^4~\frac{1~+~4v_{cor}^{-}/v_2^2~+~
405: 2v_{cor}^{-~-}/v_2^4~+~2v_{cor}^{+}/v_2^2~
406: +~v_{cor}^{+~+}/v_2^4 }{1~+~6\Delta~+~3\Delta^2}\nonumber \\
407: & &
408: -2v_2^4~\left(\frac{1~+~v_{cor}^{-}/v_2^2}{1~+~\Delta}\right)^2,
409: \end{eqnarray}
410: where
411: \begin{eqnarray}
412: \label{v_cor+}
413: & & v_{cor}^{+}~=~\frac{1}{N_0^2}
414: ~\int\limits_{-\pi}^{\pi}d\varphi_1
415: ~\int\limits_{-\pi}^{\pi}d\varphi_2
416: ~\cos{2(\varphi_1-\psi_R +\varphi_2-\psi_R)}~
417: ~ \frac{dN_{cor}}{d \varphi_1d\varphi_2}~,\nonumber\\
418: & & v_{cor}^{+~+}~=~\frac{1}{N_0^4}
419: ~\int\limits_{-\pi}^{\pi}d\varphi_1
420: ~\int\limits_{-\pi}^{\pi}d\varphi_2
421: ~\int\limits_{-\pi}^{\pi}d\varphi_3
422: ~\int\limits_{-\pi}^{\pi}d\varphi_4
423: ~\cos{2(\varphi_1 +\varphi_2-\varphi_3 - \varphi_4)}~
424: ~ \frac{dN_{cor}}{d \varphi_1d\varphi_2}~\frac{dN_{cor}}{d \varphi_3d\varphi_4},
425: \nonumber\\
426: & & v_{cor}^{-~-}~=~\frac{1}{N_0^4}
427: ~\int\limits_{-\pi}^{\pi}d\varphi_1
428: ~\int\limits_{-\pi}^{\pi}d\varphi_2
429: ~\int\limits_{-\pi}^{\pi}d\varphi_3
430: ~\int\limits_{-\pi}^{\pi}d\varphi_4
431: ~\cos{2(\varphi_1 +\varphi_2-\varphi_3 - \varphi_4)}~
432: ~ \frac{dN_{cor}}{d \varphi_1d\varphi_3}~\frac{dN_{cor}}{d \varphi_2d\varphi_4}
433: \nonumber\\
434: & & =~(v_{cor}^{-})^2~-~\left(\frac{1}{N_0^2}
435: ~\int\limits_{-\pi}^{\pi}d\varphi_1
436: ~\int\limits_{-\pi}^{\pi}d\varphi_3
437: ~\sin{2(\varphi_1-\varphi_3)}~
438: ~ \frac{dN_{cor}}{d \varphi_1d\varphi_3}\right)^2~
439: \end{eqnarray}
440:
441: $=~(v_{cor}^{-})^2$~ if ~$\frac{dN_{cor}}{d \varphi_1d\varphi_2}$ ~is an even
442: function of angular difference $(\varphi_1-\varphi_2)$.
443:
444: \noindent
445: The numerator in Eq.~(\ref{v2(4)}) is rewritten in the following form:
446: \begin{eqnarray}
447: \label{Num}
448: & & Num~\equiv~
449: -< \cos{2(\varphi_{jet}+\varphi_1 -\varphi_2
450: -\varphi_3)} >~\nonumber \\
451: & & +< \cos{2(\varphi_{jet} -\varphi_2)} >< \cos{2(\varphi_1
452: -\varphi_3)} >
453: +< \cos{2(\varphi_{jet} -\varphi_3)} >< \cos{2(\varphi_1
454: -\varphi_2)} >\nonumber \\
455: & & =
456: ~-~v_2^3\cos{2(\varphi_{jet}-\psi_R)}~\frac{1~+~2v_{cor}^{-}/v_2^2~+~
457: v_{cor}^{+}/v_2^2~ }{1~+~3\Delta}\nonumber \\
458: & &
459: +2v_2^3~\cos{2(\varphi_{jet}-\psi_R)}~\frac{1~+~v_{cor}^{-}/v_2^2}{1~+~\Delta}~
460: +~SIN,
461: \end{eqnarray}
462: where the terms ~$SIN$~ are proportional to ~$\sin{2(\varphi_{jet} -\psi_R)}$~
463: and vanishing after averaging over $\varphi_{jet}$.
464:
465: At the first glance it is hard to see the advantage in using higher order
466: cumulants from Eqs.~(\ref{v2_event}), (\ref{c_2[4]}) and (\ref{Num}). However, it
467: is reasonable to suppose that the contribution of two-particle correlations to
468: the normalization factor $N_2$ is small, $\Delta \ll 1$, while their ``second
469: Fourier harmonic'' $v_{cor}^{-}$ can be of the order of $v_2^2$. Then {\it all}
470: direct two-particle correlations $v_{cor}^{-}$ are automatically {\it canceled
471: out}\footnote{This is one of the main advantage of the cumulant expansion.} from
472: Eqs.~(\ref{c_2[4]}) and (\ref{Num}) in first order in $\Delta$,
473: but are {\it survived} in Eq.~(\ref{v2_event}). The non-direct two-particle
474: correlations $v_{cor}^{+},~v_{cor}^{+~+}$ are survived. But they are suppressed in
475: the comparison with the direct correlations $v_{cor}^{-}$ (contributing to
476: $v_2^{jet}[2]$) due to the fact that $\frac{dN_{cor}}{d \varphi_1d\varphi_2}$
477: is an even function of angular difference
478: $(\varphi_1 - \varphi_2)$ only in most physically interesting
479: cases~\cite{kov}. Moreover, for the small-angle $\delta$-like correlations
480: ($\frac{dN_{cor}}{d \varphi_1d\varphi_2}~\sim~\frac{\exp(-(\varphi_1-
481: \varphi_2)^2/2\sigma^2)}{\sqrt{2\pi}\sigma},~ \sigma \rightarrow 0$) and for
482: the large-angle oscillating ones
483: ($\frac{dN_{cor}}{d \varphi_1d\varphi_2}~\sim~\cos2(\varphi_1-
484: \varphi_2)$) the non-direct correlations $v_{cor}^{+},~v_{cor}^{+~+}$ are
485: equal to zero. Then
486: \begin{eqnarray}
487: \label{v2(4)`}
488: v_2^{jet}[4]~=\bigg\langle\frac{Num}{(-c_2[4])^{3/4}}
489: \bigg\rangle_{event}~\simeq~ v_2^{jet}
490: \end{eqnarray}
491: in this case, while
492: \begin{equation}
493: \label{v2_event`}
494: v^{jet}_2[2] ~\simeq~v_2^{jet}~
495: \sqrt{ \frac{1}{1~+~v_{cor}^{-}/v_2^2}}~.
496: \end{equation}
497: Thus Eqs.~(\ref{v2(4)`}) and (\ref{v2_event`}) demonstrate the better accuracy of
498: higher order cumulants explicitly.
499:
500: \section{Discussion}
501: In order to illustrate the applicability of the method presented with regard for the
502: real physical situation, we consider the following model (see Ref.~\cite{phl02} for
503: details).
504:
505: \smallskip
506:
507: \centerline{\bf The model}
508:
509: The initial jet distributions in a nucleon-nucleon sub-collision at $\sqrt{s}=5.5$ TeV
510: have been generated using PYTHIA$\_5.7$~\cite{pythia}. We simulated the
511: rescattering and energy loss of jets in gluon-dominated plasma, created initially in
512: the nuclear overlap zone in Pb$-$Pb collisions at different impact parameters.
513: For details of this approach one can refer to Refs.~\cite{lokhtin00,lokhtin01}.
514: Essential for our consideration here is that
515: in non-central collisions the azimuthal distribution of jets is approximated well
516: by the elliptic form (\ref{phi_jet}). In the model the coefficient of jet azimuthal
517: anisotropy increases almost linearly with the growth of impact parameter $b$ and
518: becomes maximum at $b \sim 1.2 R_A$, where $R_A$ is the nucleus radius. After that
519: $v_2^{jet}$ drops rapidly with increasing $b$: this is the domain of impact parameter
520: values, where the effect of decreasing energy loss due to reducing effective transverse
521: size of the dense zone and initial energy density of the medium is crucial and not
522: compensated anymore by stronger non-symmetry of the volume.
523: The kinematical cuts on jet transverse energy and rapidity has been applied:
524: $E_T^{jet} > 100$ GeV and $|y^{jet}| < 1.5$. After this the dijet event is
525: superimposed on the Pb$-$Pb event containing anisotropic flow.
526:
527: Anisotropic flow was generated using the simple hydrodynamical Monte-Carlo
528: code~\cite{kruglov,phl02} giving hadron (charged and neutral pion, kaon
529: and proton) spectrum as a superposition of the thermal distribution and collective
530: flow. To be definite, we fixed the following ``freeze-out'' parameters: the
531: temperature $T_f = 140$ MeV, the collective longitudinal rapidity $Y_L^{max}=3$ and
532: the collective transverse rapidity $Y_T^{max}=1$. We set the Poisson multiplicity
533: distribution and took into account the impact parameter dependence of multiplicity in a
534: simple way, just suggesting that the mean multiplicity of particles is proportional to
535: the nuclear overlap function. We also suggested~\cite{phl02} that the spatial
536: ellipticity of the ``freeze-out'' region is directly related to the initial spatial
537: ellipticity of the nuclear overlap zone.
538: Such ``scaling'' allows one to avoid using additional parameters
539: and, at the same time, results in the elliptic anisotropy of particle and
540: energy flow due to the dependence of effective transverse size of the ``freeze-out''
541: region on the azimuthal angle of a ``hadronic liquid'' element.
542: Obtained in such a way azimuthal distribution of particles is described well by
543: the elliptic form (\ref{phi_part}) for the domain of reasonable impact parameter
544: values.
545:
546: To be specific, we consider the geometry of CMS detector~\cite{cms94} at LHC.
547: The central (``barrel'') part of the CMS calorimetric
548: system covers the pseudo-rapidity region $|\eta| < 1.5$, the segmentation
549: of electromagnetic and hadron calorimeters being $\Delta \eta \times \Delta
550: \phi = 0.0174 \times 0.0174$ and $\Delta \eta \times \Delta \phi = 0.0872 \times
551: 0.0872$ respectively~\cite{cms94}. In order to reproduce roughly the experimental
552: conditions (not including real detector effects, but just assuming
553: calorimeter hermeticity), we applied Eqs.~(\ref{u2(E)_event}) and (\ref{v2E(4)}) to the
554: energy deposition $E_i(\varphi _i)$ of generated particles, integrated over the
555: rapidity in $72$ segments (according to the number of segments in the hadron
556: calorimeter: $72~\times~0.0872~=~2\pi$; $i~=~1,...,72$) covering full azimuth.
557:
558: Note that in the CMS heavy ion physics program, the modified sliding window-type jet
559: finding algorithm has been developed to search for ``jet-like'' clusters
560: above the average energy, and to subtract the background from the underlying
561: event~\cite{note00-060}. Strictly speaking, after jet extraction the background energy
562: deposition in the calorimetric cells should be redefined and can appear to be not
563: exactly equal to the initially generated one. However we neglect this effect here. In a
564: real experimental situation, in order to avoid the influence of jet contribution on the
565: particle flow, one can consider jets and particles incorporated in the energy flow
566: analysis in different rapidity regions.
567:
568: \smallskip
569:
570: \centerline{\bf Numerical results}
571:
572: We have found~\cite{phl02} that the accuracy of $v^{jet}_2$
573: determination from Eq.~(\ref{u2(E)_event}) is close to $100 \%$ for semi-central
574: ($b~\la~R_A$) collision and becomes significantly
575: worse in very peripheral collision ($b \sim 2 R_A$), wherein decreasing
576: multiplicity and azimuthal anisotropy of the event results in relatively large
577: fluctuations of energy deposition in each segment.
578:
579: In the given paper we test the efficiency of the higher order correlator
580: (\ref{v2E(4)}) and have found, at first, that the results for $v_2^{jet}$
581: obtained from Eqs.~(\ref{u2(E)_event}) and (\ref{v2E(4)}) are practically the
582: same. This is explained by the fact that our simple Monte-Carlo event generator gives
583: the elliptic anisotropy of energy flow, correlated with the reaction plane, but
584: no correlations between energy deposition in the calorimeter
585: segments. We can introduce such correlations at the calorimetric level ``by
586: hand'', simply assuming that the probability of finding the energy $E_i$ in a
587: segment $i$ and the energy $E_j$ in a segment $j$ is proportional to
588: $E_i E_j(1~+~c_{ij})$, where the ``correlation strength'' $c_{ij}$ may be, for example,
589: proportional to $\delta_{ij}$ (the short-range $\delta$-like correlations) or
590: $\cos{2(\varphi_i-\varphi_j)}$ (the long-range oscillating correlations). In
591: this case we became convinced that the higher order cumulant (\ref{v2E(4)})
592: was almost independent of such correlations (as it was shown in Sect. 4),
593: while the result of calculation (\ref{u2(E)_event}) changed, closely
594: following the formula (\ref{v2_event`}) corrected by autocorrelation terms which
595: are non-vanishing in finite summation~\cite{ollitr}.
596:
597: We have also found at the calorimetric level that, taking into account the
598: effect of possible detector inefficiency (i.e. that the particles and jets are not
599: detected in a ``blind'' azimuthal sector of size $\alpha$),
600: the accuracy of $v_2^{jet}$ determination appears to be less than
601: $50 \%$ at $\alpha~\ga 30^{\circ}$ in our model calculation without correlations
602: and at $b \ge R_A$, whichever
603: algorithm (\ref{u2(E)_event}) or (\ref{v2E(4)}) we used.
604:
605: Fig.1 is presented to illustrate the improvement due to the fourth order cumulant
606: method in the determination of jet azimuthal anisotropy $v^{jet}_2$ depending on
607: the ratio $\bar v^-_{cor}/\bar v_2^2$, where $\bar v_2 $ is the coefficient of
608: elliptic azimuthal anisotropy of energy flow defined here as
609: \begin{equation}
610: \bar v_2 ~=~\frac{1}{2}~~ \frac{E_{max(i)} - E_{min(i)}}
611: {E_{max(i)} + E_{min(i)}}~,
612: \end{equation}
613: and $E_{max(i)}$ and $E_{min(i)}$ are the maximum and minimum energy deposit in a
614: segment respectively ($i~=~1,...,72$). The coefficient $\bar v^-_{cor}$ determines the
615: ``correlation strength'' at the calorimetric level, $c_{ij} = 72 \bar v^-_{cor}
616: \delta_{ij}$ for short-range correlations
617: \footnote{
618: Here one should note that the majority of sources of non-flow correlations mentioned
619: above is effective at small angles between particles. In our case these correlations
620: can be partially smoothed out after summing particle energies over the azimuthal angles
621: in a calorimeter segment of finite size ($\sim 5^{\circ}$). This can result in the
622: smaller value of the ratio $\bar v^-_{cor}/\bar v_2^2$ in comparison with the ratio
623: $v^-_{cor}/v_2^2$ (and, as consequence, in a somewhat less improvement due to the higher
624: order method at the calorimetric level (\ref{v2E(4)}) in comparison with the
625: particle level (\ref{v2(4)})). We still have no any adequate Monte-Carlo generator
626: for particle flow effects at LHC including the physical model for correlations. Thus
627: we can not estimate the real value of $\bar v^-_{cor}/\bar v_2^2$ (true benefit
628: from the higher order method) and just treat it here as a phenomenological parameter.}
629: (the similar result is obtained for long-range correlations with
630: $c_{ij} = 2 \bar v^-_{cor} \cos{2(\varphi_i-\varphi_j)}$).
631: The plots show the $b$-dependence of the ``theoretical'' value of $v^{jet}_2$
632: (calculated including collisional and radiative energy loss
633: when the reaction plane angle is known in each event), and the $v^{jet}_2$
634: determined by the methods (\ref{u2(E)_event}) and (\ref{v2E(4)})
635: for the three ratios $\bar v^-_{cor}/ \bar v_2^2 ~=~ 0,~ 0.01,~0.1$. We
636: used two values of the input parameter, the number of charged particles per unit
637: rapidity at $y=0$ in central Pb$-$Pb collisions: $dN^{\pm}/dy = 3000$ (Fig.1a) and
638: $6000$ (Fig.1b).
639:
640: One can see that improvement due to the fourth order cumulant method (result of
641: (\ref{v2E(4)}) is independent of $\bar v^-_{cor}/\bar v_2^2$ and
642: coincides with the result of (\ref{u2(E)_event}) for $\bar v^-_{cor}/\bar v_2^2=0$) is
643: pronounced for more peripheral collisions, smaller particle multiplicities and
644: larger ``correlation strengths''.
645:
646: \newpage
647:
648: \section{Conclusions}
649:
650: In the present paper we have analyzed the method for measurements of
651: jet azimuthal anisotropy coefficients without reconstruction of the event
652: plane considering the higher order correlators between the azimuthal position
653: of the jet axis and the angles of particles not incorporated in the jet.
654: The method is generalized by introducing as weights the particle momenta or
655: the energy deposit in the calorimeter segments. In the latter case, we have
656: illustrated its reliability in the real physical situation under LHC conditions.
657: Introducing in the model correlations between energy deposits in the
658: calorimeter segments does not practically change the accuracy of the method using
659: fourth order cumulant calculations (\ref{v2E(4)}), while the result obtained
660: with second order correlator (\ref{u2(E)_event}) is dependent significantly of
661: the ``strength'' of such correlations. The advantage of the higher order cumulant
662: analysis is pronounced for more peripheral collisions and smaller particle
663: multiplicities.
664:
665: To summarize, we believe that the present technique may be useful
666: investigating azimuthal anisotropy of jets and high-$p_T$
667: particles in heavy ion collisions at RHIC and LHC.
668:
669: {\it Acknowledgments}.
670: Discussions with A.I.~Demianov, S.V.~Petrushanko, V.V.~Uzhinskii, I.N.~Vardanian,
671: S.A.~Voloshin, U.~Wiedemann and E.E.~Zabrodin are
672: gratefully acknowledged.
673:
674: \begin{thebibliography}{99}
675: \bibitem{qm01} Proceedings of 15th International Conference on Ultrarelativistic
676: Nucleus-Nucleus Collisions ``Quark Matter'2001'' (New York, USA, 15-20 January, 2001),
677: Nucl.Phys. A {\bf 698} (2002); Proceedings of 16th International Conference on
678: Ultrarelativistic Nucleus-Nucleus Collisions ``Quark Matter'2002'' (Nantes, France
679: 18-24 June, 2002), in press.
680: \bibitem{uzhi} A.S. Galoyan and V.V. Uzhinskii, e-print hep-ph/0007122.
681: \bibitem{wang00} X.-N. Wang, Phys. Rev. C {\bf 63}, 054902 (2001).
682: \bibitem{gyul00} M. Gyulassy, I. Vitev, and X.-N. Wang, Phys. Rev. Lett. {\bf 86}, 2537
683: (2001).
684: \bibitem{star} K.H. Ackermann et al. (STAR Coll.), Phys. Rev. Lett. {\bf 86},
685: 402 (2001);
686: C. Adler et al. (STAR Coll.), Phys. Rev. Lett. {\bf 87}, 182302 (2001); Phys. Rev.
687: Lett. {\bf 90}, 032301 (2003);
688: R.J.M.Snellings et al. (STAR Coll.) Nucl. Phys. A {\bf 698}, 193 (2002).
689: \bibitem{phenix} R.A. Lacey et al. (PHENIX Coll.), Nucl. Phys. A {\bf 698}, 559 (2002).
690: \bibitem{phobos} I.C. Park et al. (PHOBOS Coll.), Nucl. Phys. A {\bf 698}, 564 (2002);
691: B.B. Back et al. (PHOBOS Coll.), Phys. Rev. Lett. {\bf 89}, 222301
692: (2002); e-print nucl-ex/0210036.
693: \bibitem{kolb} P.F. Kolb, J. Sollfrank, and U. Heinz, Phys. Rev. C {\bf 62},
694: 054909 (2000); P.F.Kolb et al., Phys. Lett. B {\bf 500}, 232 (2001);
695: Nucl. Phys. A {\bf 696}, 175 (2001).
696: \bibitem{zabrodin} E.E. Zabrodin, C. Fuchs, L.V. Bravina, and A. Faessler,
697: Phys. Lett. B {\bf 508}, 184 (2001).
698: \bibitem{lokhtin00} I.P. Lokhtin and A.M. Snigirev,
699: Eur. Phys. J. C {\bf 16}, 527 (2000).
700: \bibitem{lokhtin01} I.P. Lokhtin, S.V. Petrushanko, L.I. Sarycheva, and A.M.
701: Snigirev, e-print hep-ph/0112180; Phys. At. Nucl. {\bf 65}, 943 (2002).
702: \bibitem{Lokhtin:2002} I.P. Lokhtin, A.M. Snigirev, and I. Vitev, e-print
703: hep-ph/0212061.
704: \bibitem{cms94} CMS Collaboration, Technical Proposal, CERN/LHCC 94-38.
705: \bibitem{note00-060} G. Baur et al., ``Heavy Ion Physics Programme in CMS'',
706: CERN CMS Note 2000/060.
707: \bibitem{voloshin} S.A. Voloshin and Y. Zhang, Z. Phys. C {\bf 70}, 665 (1996);
708: A.M. Poskanzer and S.A.Voloshin, Phys. Rev. C {\bf 58}, 1671 (1998).
709: \bibitem{wang} S. Wang et al., Phys. Rev. C {\bf 44}, 1091 (1991).
710: \bibitem{ollit} J.-Y. Ollitrault, Phys. Rev. D {\bf 46}, 229 (1992);
711: Phys. Rev. D {\bf 48}, 1132 (1993).
712: \bibitem{ollitr} N. Borghini, P.M. Dinh, and J.-Y. Ollitrault, Phys. Rev. C {\bf 63},
713: 054906 (2001).
714: \bibitem{din} N. Borghini, P.M. Dinh, and J.-Y. Ollitrault, Phys. Rev. C {\bf 64},
715: 054901 (2001); e-print nucl-ex/0110016.
716: \bibitem{star02} C. Adler et al. (STAR Coll.), Phys. Rev. C {\bf 66}, 034904
717: (2002).
718: \bibitem{phl02}I.P. Lokhtin, L.I. Sarycheva, and A.M. Snigirev, Phys. Lett. B {\bf 537},
719: 261 (2002).
720: \bibitem{kov} Yu.V. Kovchegov, and K.L. Tuchin, Nucl. Phys. A {\bf 708},
721: 413 (2002); Nucl. Phys. A {\bf 717}, 249 (2002).
722: \bibitem{dan} P. Danielewicz et al., Phys. Rev. C {\bf 38}, 120 (1988).
723: \bibitem{bor02} N. Borghini et al., Phys. Rev. C {\bf 66}, 014901 (2002).
724: \bibitem{bor} N. Borghini, P.M. Dinh, and J.-Y. Ollitrault, Phys.
725: Lett. B {\bf 477}, 51 (2000); Phys. Rev. C {\bf 62}, 034902 (2000).
726: \bibitem{pythia} T. Sjostrand, Comp. Phys. Com.{\bf 82}, 74 (1994).
727: \bibitem{kruglov} N.A. Kruglov, I.P. Lokhtin, L.I. Sarycheva, and A.M. Snigirev, Z.
728: Phys. C {\bf 76}, 99 (1997).
729: \end{thebibliography}
730:
731: \newpage
732:
733: \begin{figure}[hbtp]
734: \begin{center}
735: \makebox{\epsfig{file=fig1.eps, height=200mm}}
736: \caption{The impact parameter dependence of ``theoretical'' value of $v^{jet}_2$
737: including collisional and radiative energy loss (solid curve), and $v^{jet}_2$
738: determined by the method (\ref{u2(E)_event}) for $\bar v^-_{cor}/\bar v_2^2 =0$ (dashed
739: curve), $0.01$ (dotted curve) and $0.1$ (dash-dotted curve). The result obtained using
740: the fourth order cumulant method (\ref{v2E(4)}) coincides with the dashed curve.
741: $dN^{\pm}/dy (y=0, b=0) = 3000$ (a) and $6000$ (b).}
742: \end{center}
743: \end{figure}
744:
745:
746: \end{document}
747:
748:
749:
750:
751:
752:
753: