1: %\documentclass[12pt]{article}
2:
3:
4: \documentclass[aps,showpacs,preprint,superscriptaddress]{revtex4}
5: \usepackage{psfig}
6: \newcommand{\bea}{\begin{eqnarray}}
7: \newcommand{\beq}{\begin{equation}}
8: \newcommand{\eea}{\end{eqnarray}}
9: \newcommand{\eeq}{\end{equation}}
10:
11: %\textwidth=160mm
12: %\textheight=237mm
13:
14: %\setlength{\voffset}{-20mm}
15: %\oddsidemargin -5mm
16: %\evensidemargin -5mm
17:
18: \usepackage{epsf}
19: \usepackage[dvips]{graphicx}
20:
21: \begin{document}
22: \bibliographystyle{revtex}
23:
24: \begin{center}
25:
26: {\bfseries RANDOM MATRIX THEORY AND ANALYSIS OF NUCLEUS-NUCLEUS
27: COLLISION AT HIGH ENERGIES}
28:
29: \vskip 5mm
30:
31: E.\ I.\ Shahaliev $^{1,2 \dag}$, R.\ G.\ Nazmitdinov $^{3,4}$, A.\
32: A.\ Kuznetsov $^{1}$, M.\ K.\ Suleymanov $^{1}$, O.\ V. Teryaev
33: $^{4}$
34:
35: %\vskip 5mm
36:
37: {\small
38: (1) {\it High Energy Physics Laboratory, Joint Institute
39: for Nuclear Research, 141980, Dubna, Russia}
40: \\
41: (2) {\it Institute of Radiation Problems, 370143, Baku,
42: Azerbaijan}
43: \\
44: (3) {\it Departament de F{\'\i}sica, Universitat de les Illes
45: Balears, E-07122 Palma de Mallorca, Spain}
46: \\
47: (4) {\it Bogoliubov Laboratory of Theoretical Physics, Joint
48: Institute for Nuclear Research, 141980 Dubna, Russia}
49:
50: $\dag$ {\it E-mail: shah@sunhe.jinr.ru }}
51: \end{center}
52:
53:
54:
55:
56: \vskip 5mm
57:
58: \begin{center}
59: \begin{minipage}{150mm}
60: \centerline{\bf Abstract}
61:
62: We propose a novel method for analysis of experimental data
63: obtained at relativistic nucleus-nucleus collisions. The method,
64: based on the ideas of Random Matrix Theory, is applied to detect
65: systematic errors that occur at measurements of momentum
66: distributions of emitted particles. The unfolded momentum
67: distribution is well described by the Gaussian orthogonal ensemble
68: of random matrices, when the uncertainty in the momentum
69: distribution is maximal. The method is free from unwanted
70: background contributions.
71:
72:
73: \end{minipage}
74: \end{center}
75:
76: \vskip 10mm
77:
78:
79:
80: Relativistic heavy ion collisions are among major experimental
81: tools that allow to get insight into nuclear dynamics at high
82: excitation energies and large baryon densities. It is expected
83: that in central collisions, at energies that are and will be soon
84: available at SPS(CERN), RHIC(BNL) and LHC(CERN), the nuclear
85: density may exceed by tens times the density of stable nuclei. At
86: such extreme conditions one would expect that a final product of
87: heavy ion collisions could present a composite system that
88: consists of free nucleons, quarks and quark-gluon plasma.
89: However, identification of the quark-gluon plasma, for example, is
90: darken due to a multiplicity of secondary particles created at
91: these collisions. There is no a clear evidence of the quark
92: constituent as well. In fact, there are numerous additional
93: mechanisms of a particle creation that mask the presence of the
94: quark-gluon plasma (QGP). It appears that the QGP could be
95: manifested via the observation of indirect phenomena. The natural
96: question arises: how to identify a useful signal that would be
97: unambiguously associated with a certain physical process ?
98:
99:
100: The most popular methods of analysing data produced at
101: relativistic heavy ion collisions are the correlation
102: analysis~\cite{[1]}, the analysis of missing masses~\cite{[2]}
103: and effective mass spectra~\cite{[3]}, the interference method of
104: identical particles~\cite{[4]}. We recall that results obtained
105: within those methods are sensitive to assumptions made upon the
106: background of measurements and mechanisms included into a
107: corresponding model consideration. As was mentioned above, the
108: larger is the excitation energy, the larger is a number of various
109: mechanisms of the creation that should be taken into account.
110:
111:
112:
113: As an alternative approach, one could develop a method that should
114: be independent on the background contribution. For instance, there
115: are attempts to use the maximum entropy principle~\cite{2},
116: Fourier transform~\cite{22} and even by even analysis~\cite{[7]}.
117: Thus, a formulation of a criteria for a selection of meaningful
118: signals is indeed a topical objective of the relativistic heavy
119: ion collisions physics. The major aim of this paper is to suggest
120: a method that does not depend on the background information and
121: relies only upon the fundamental symmetries of the composite
122: system.
123:
124:
125:
126: Our approach is based on Random Matrix Theory \cite{M91} that was
127: originally introduced to explain the statistical fluctuations of
128: neutron resonances in compound nuclei \cite{P65} (see also
129: Ref.\onlinecite{Brody}). The theory assumes that the Hamiltonian
130: belongs to an ensemble of random matrices that are consistent with
131: the fundamental symmetries of the system. In particular, since the
132: nuclear interaction preserves time-reversal symmetries, the
133: relevant ensemble is the Gaussian Orthogonal Ensemble (GOE). When
134: the time-reversal symmetry is broken one can apply the Gaussian
135: Unitary Ensemble (GUE). The GOE and GUE correspond to ensembles of
136: real symmetric matrices and of Hermitian matrices, respectively.
137: Besides these general symmetry considerations, there is no need in
138: other properties of the system under consideration.
139:
140:
141:
142: Bohigas {\it et al} \cite{Boh} conjectured that RMT describes the
143: statistical fluctuations of a quantum systems whose classical
144: dynamics is chaotic. Quantum spectra of such systems manifest a
145: strong repulsion (anticrossing) between quantum levels, while in
146: non-chaotic (regular) systems crossings are a dominant feature of
147: spectra (see, e.g., \cite{H94}). In turn, the crossings are
148: observed when there is no mixing between states that are
149: characterized by different good quantum numbers, while the
150: anticrossings signal about a strong mixing due to a perturbation
151: brought about by either external or internal sources. Nowadays,
152: RMT has become a standard tool for analysing the fluctuations in
153: nuclei, quantum dots and many other systems (see for a review, for
154: example, Ref.\onlinecite{Gur}). The success of RMT is determined
155: by the study the statistical laws governing fluctuations having
156: very different origins. Regarding the relativistic heavy ion
157: collision data the study of fluctuation properties of the momentum
158: distribution of emitted particles could provide an information
159: about i)possible errors in measurements and ii)kinematical and
160: dynamical correlations of the composite system.
161:
162:
163:
164: Let us consider the discrete spectrum $\{E_i\}, i=1,...,N$ of a
165: d-dimensional quantum system (d is a number of degrees of
166: freedom). A separation of fluctuations of a quantum spectrum can
167: be based on the analysis of the density of states below some
168: threshold $E$ \beq \label{d} S(E)=\sum_{i=1}^N\delta(E-E_i) \, .
169: \eeq We can define a staircase function \beq \label{s}
170: N(E)=\int_{-\infty}^ES(E')dE'=\sum_{i=1}^N\theta (E-E'), \eeq
171: giving the number of points on the energy axis which are below or
172: equal to E. Here \beq \label{t} \theta (x) \,=\, \left \{
173: \begin{array}{l}
174: 0\qquad \,for \,\,\, x < 0 \\
175: 1\qquad \,for \,\,\, x > 1 \\
176: \end{array}
177: \right. \eeq We separate $N(E)$ in a smooth part $\zeta(E)$ and
178: the reminder that will define the fluctuating part $N_{\rm fl}(E)$
179: \beq \label{s1} N(E)=\zeta(E)+N_{\rm fl}(E) \eeq The smooth part
180: $\zeta(E)$ can be determined either from semiclassical arguments
181: or using a polynomial or spline interpolation for the staircase
182: function.
183:
184:
185:
186: To study fluctuations we have to get rid of the smooth part. The
187: usual procedure is to "unfold" the original spectrum $\{E_i\}$
188: through the mapping $E \rightarrow x$ \beq \label{m}
189: x_i=\zeta(E_i), \qquad i=1,...,N \eeq Now we can define spacings
190: $s_i=x_{i+1}-x_i$ between two adjacent points and collect them in
191: a histogram. The effect of mapping is that the sequence $\{x_i\}$
192: has on the average a constant mean spacing (or a constant
193: density), irrespective of the particular form of the function
194: $\zeta(E)$ \cite{boh2}. To characterize fluctuations one deals
195: with different correlation functions \cite{M91}. In this paper we
196: will use only a correlation function related to spacing
197: distribution between adjacent levels. Below, we follow a simple
198: heuristic argument due to Wigner \cite{Wi} that illustrates the
199: presence or absence of level repulsion in an energy spectrum.
200:
201:
202:
203: For a random sequence, the probability that the level will be in
204: the small interval $[x_0+s,x_0+s+ds]$ is independent of whether or
205: not there is a level at $x_0$. Given a level at $x_0$, let the
206: probability that the next level be in $[x_0+s,x_0+s+ds]$ be
207: $p(s)ds$. Then for $p(s)$, the nearest-neighbor spacing
208: distribution, we have \beq p(s)ds=p(1\in ds|0\in s)p(0\in s) \eeq
209: Here, $p(n\in s)$ is a probability that the interval of length $s$
210: contains $n$ levels and $p(n\in ds|m\in s)$ is the conditional
211: probability that the interval of length $ds$ contains $n$ levels,
212: when that of length $s$ contains $m$ levels. One has $p(0\in
213: s)=\int_s^{\infty}p(s')ds'$, the probability that the spacing is
214: larger than $s$. The term $p(1\in ds|0\in s)=\mu(s)ds$ $[\mu(s)$
215: is the density of spacings $s]$, depends explicitly on the
216: choices, 1 and 0, of the discrete variables $n,m$. As a result,
217: one obtains $p(s)=\mu(s)\int_s^{\infty}p(s')ds'$ which can be
218: solved to give \beq p(s)=\mu(s) \exp(-\int_0^s \mu(s')ds') \eeq
219: The function $p(s)$ and its first moment are normalized to unity,
220: \beq \int_0^s p(s)ds=1, \qquad \int_0^s sp(s)ds=1. \eeq For a
221: linear repulsion $\mu(s)=\pi s/2$ one obtains the Wigner surmise,
222: \beq p(s)=\frac{\pi}{2}s \exp(-\frac{\pi}{4}s^2), \qquad s\geq 0
223: \eeq For a constant value $\mu(s)=1$ one obtains the Poisson
224: distribution \beq p(s)=\exp^{-s}, \qquad s\geq 0 \eeq
225:
226:
227:
228:
229: As discussed above, when quantum numbers of levels are well
230: defined, one should expect for the spacings the Poisson type
231: distribution, while a Wigner type distribution occurs due to
232: either internal or external perturbations that destroy these
233: quantum numbers. In fact, one of the sources of external
234: perturbations can be attributed to the uncertainty in the
235: determination of the momentum distribution of emitted particles in
236: relativistic heavy ion collisions. We make a conjecture that the
237: above discussed ideas of the RMT are applicable to the momentum
238: distribution as well. We assume that the momentum distribution may
239: be associated with eigenstates (quantum levels) of a composite
240: system. The difference between energy and momentum is inessential
241: for pions (see below), while we assume that the proton mass should
242: not affect significantly the correlation function.
243:
244:
245: Another possibilities are the association of the momentum
246: distribution to the spectrum of scattering matrix, or density
247: matrix, which can equally be the object of statistical analysis.
248: Note also, that here we are dealing with the momentum distribution
249: in the target rest frame only, postponing its comparison to that
250: in the center of mass frame, which is more natural for description
251: of interaction. Therefore, we simply replace in
252: Eqs.(\ref{d})-(\ref{m}) the variable $E$ by the variable $|p|$ and
253: construct the corresponding correlation function $p(s)$.
254:
255:
256: To test the utility and the validity of the proposal we use the
257: experimental data that have been obtained from the 2-m propane
258: bubble chamber of LHE, JINR~\cite{[8],[9]}. The chamber, placed in
259: a magnetic field of 1.5 T, was exposed to beams of light
260: relativistic nuclei at the Dubna Synchrophasotron. Practically all
261: secondaries, emitted at a 4$\pi$ total solid angle, were detected
262: in the chamber. All negative particles, except those identified as
263: electrons, were considered as $\pi^-$-mesons. The contaminations
264: by misidentified electrons and negative strange particles do not
265: exceed 5$\%$ and 1$\%$, respectively. The average minimum momentum
266: for pion registration is about 70 MeV/c. The protons were selected
267: by a statistical method applied to all positive particles with a
268: momentum of $|p|>500$ MeV/c (we identified slow protons with
269: $|p|\le 700$ MeV/c by ionization in the chamber). In this
270: experiment, we had got 20407 ${}^{12}{CC}$ interactions at a
271: momentum of 4.2A GeV/c (for methodical details see~\cite{[9]})
272: contents 4226 events with more than ten tracks of charged
273: particles. Thus, it was known in advance the accuracy of
274: measurements for available range of the momentum distribution of
275: secondary particles. Consequently, our analysis has been done for
276: different range of values of the momentum distribution to
277: illuminate the degree of the accuracy.
278:
279:
280:
281: On Fig. 1 the dependence $dN/d{|p|}$ as a function of the measured
282: momentum (0.15-7.5 GeV/c) of the secondary particles is displayed
283: . The numerical data $N(p)$ were approximated by the polynomial
284: function of the sixth order and we obtain the distribution of
285: various spacings $s_i$ in 2636 events satisfying the condition of
286: $\chi^2$ per degree of freedom less than 1.0. Momenta are well
287: defined in the region 0.15-1.14 GeV/c (region I, Fig. 2a), where
288: the minimal value of the proton momentum is 0.15 GeV/c. The
289: intermediate region (region II, Fig. 2b) covers the values
290: 1.14-4.0 GeV/c. The region 4.0-7.5 GeV/c is the less reliable one
291: (Fig. 2c). The spacing probability nicely reproduces this tendency
292: depending on the region of the momentum distribution. The function
293: $p(s)$ has the Poisson distribution for the region I, where the
294: momentum distribution was defined with a high accuracy. The region
295: II corresponds to the intermediate situation, when the spacing
296: distribution lies between the Poisson and the Wigner
297: distributions. The less reliable region of the values has a Wigner
298: type distribution for the spacing probability (Fig. 2c). Indeed,
299: the distribution reflects a strong deviation from the regular
300: behavior, observed for the measurements with a high degree of the
301: accuracy.
302:
303:
304:
305: Summarizing, we propose a method to analyse data obtained at
306: relativistic heavy ion collisions. The method does not depend on
307: the background of the measurements and provides a reliable
308: information about correlations brought about by external or
309: internal perturbations. In particular, we demonstrate that the
310: method manifests the perturbations due to the uncertainty in the
311: determination of the momentum distribution of secondary emitted
312: particles.
313:
314:
315:
316: \section*{Acknowledgments}
317: This work was partly supported by Grant No.\ BFM2002-03241
318: from DGI (Spain). R. G. N. gratefully acknowledges support from
319: the Ram\'on y Cajal programme (Spain).
320:
321:
322:
323:
324: \begin{thebibliography}{99}
325: \bibitem{[1]}
326: E.\ A.\ De Wolf, I.\ M.\ Dremin, and W.\ Kittel,
327: Usp.\ Fiz.\ Nauk\ {\bf 163}, 3 (1993).
328: \bibitem{[2]}
329: E.\ Byckling, K.\ Kajantie,
330: {\it Particle Kinematics}
331: (John Wiley \& Sons, London, 1973).
332: \bibitem{[3]}
333: G.\ I.\ Kopylov,
334: {\it Bases of Resonans Kinematics}
335: (Nauka, Moscow, 1970).
336: \bibitem{[4]}
337: M. \ I.\ Podgoretsky, Particle and Nuclei,
338: {\bf 22}, 628 (1989).
339: \bibitem{2}
340: B.\ R.\ Frieden, {\it Picture Processing and Digital
341: Filtering}, {\bf 6}, 232 (Springer-Verlag, 1975).
342: \bibitem{22}
343: L.\ M.\ Soroko,
344: JINR Communication, 1-5030. Dubna (1970).
345:
346:
347: \bibitem{[7]}
348: Fu Jinghua and Liu Lianshou,
349: Phys.\ Rev.\ C\ {\bf 68}, 064904 (2003).
350: \bibitem{M91}
351: M.\ L.\ Mehta, {\it Random Matrices}
352: \,\, Second ed.(Academic, New York, 1991).
353: \bibitem{P65}
354: C.\ E.\ Porter,
355: {\it Statistical Theories of Spectra: Fluctuations}
356: (Academic, New York, 1965).
357: \bibitem{Brody}
358: T.\ A. Brody, J.\ Flores, J.\ B.\ French, P.\ A.\ Mello,
359: A.\ Pandy, and S.\ S.\ M. Wong,
360: Rev.\ Mod.\ Phys.\ {\bf 53}, 385 (1981).
361: \bibitem{Boh}
362: O.\ Bohigas, M.\ J.\ Giannoni, and C.\ Schmit,
363: Phys.\ Rev.\ Lett.\ {\bf 52}, 1 (1984).
364: \bibitem{H94}
365: W.\ D.\ Heiss, R.\ G.\ Nazmitdinov, and S.\ Radu,
366: Phys.\ Rev.\ Lett.\ {\bf 72}, 2351 (1994);
367: Phys.\ Rev.\ C\ {\bf 52}, R1179 (1995); {\it ibid.}
368: {\bf 52}, 3032 (1995).
369: \bibitem{Gur}
370: T.\ Cuhr, A.\ M\"uller-Groeling, and H.\ Weidenm\"uller,
371: Phys.\ Rep.\ {\bf 299}, 189 (1998).
372:
373:
374: \bibitem{boh2}
375: O.\ Bohigas, Lecture Notes in Physics {\bf 263}, 18
376: (Springer-Verlag, 1986).
377: \bibitem{Wi}
378: E.\ P.\ Wigner, Contribution to
379: {\it Conference on neutron physics by time-of-flight},
380: Gatlinburg, Tennessee 1956, reprinted in Ref.\onlinecite{P65}.
381:
382:
383: \bibitem{[8]}
384: The BBCDHSSTTU-BW Collaboration,
385: A. U. Abdurakhimov {\it et al.},
386: Phys.\ Lett.\ B\ {\bf 39}, 371 (1972).
387: \bibitem{[9]}
388: N. Akhababian et al., JINR Report No. 1-12114, 1979.
389: \end{thebibliography}
390:
391:
392: \newpage
393:
394: %% To insert figure (with the help of epsf.sty)
395: \begin{figure}[h]
396: \centerline{
397: \includegraphics[width=50mm,height=50mm]{fj1.eps}}
398: %\caption{Second figure inserted with the help of graphicx.sty}
399:
400: %\caption{}
401: {\bf Fig.1} $dN/d{|p|}$ as a function of the
402: measured momentum of the secondary particles.
403:
404: \end{figure}
405:
406: %% To insert figure (with the help of epsf.sty)
407: \begin{figure}[h]
408: \centerline{
409: \includegraphics[width=50mm,height=50mm]{fj1a.eps}}
410: % \caption{Second figure inserted with the help of graphicx.sty}
411: {\bf Fig.2 a}
412: \end{figure}
413:
414: %% To insert figure (with the help of epsf.sty)
415: \begin{figure}[h]
416: \centerline{
417: \includegraphics[width=50mm,height=50mm]{fj1b.eps}}
418: %\caption{Second figure inserted with the help of graphicx.sty}
419: {\bf Fig.2 b}
420: \end{figure}
421:
422: %% To insert figure (with the help of epsf.sty)
423: \begin{figure}[h]
424: \centerline{
425: \includegraphics[width=50mm,height=50mm]{fj1c.eps}}
426: %\caption{Second figure inserted with the help of graphicx.sty}
427:
428: {\bf Fig.2 c} Nearest-neighbor spacing momentum distribution
429: $p(s)$
430: for different regions of measured momenta:
431: a)$0.15<|p|<1.14$ GeV/c; b)$1.14<|p|<4.0$ GeV/c;
432: c)$4.0<|p|<7.5$ GeV/c.
433: The solid line is the Wigner-Dyson distribution and the
434: dashed line is the Poisson distribution.
435:
436:
437:
438: \end{figure}
439:
440: \end{document}
441: