1: \documentclass[preprint2]{aastex}
2: \begin{document}
3:
4: \title{The differential energy distribution of
5: the universal density profile of dark halo}
6: \author{C. Hanyu and A. Habe}
7: \affil{Division of Physics, Graduate School of Science,
8: Hokkaido University, Sapporo 060-0810, Japan}
9: \email{chiaki@astro1.sci.hokudai.ac.jp}
10: \email{habe@astro1.sci.hokudai.ac.jp}
11:
12: \begin{abstract}
13: We study the differential energy distribution of dark matter halos, carrying out cosmological $N$-body simulation.
14: We give an analytical formula of the differential energy distribution of dark matter in the halos obtained by the numerical simulation.
15: From the analytical formula we reconstruct the density profile described by the Navarro,
16: Frenk, \& White (NFW) profile.
17: The NFW profile is consistent with the analytical formula of
18: our fractional mass distribution.
19: We find that a parameter in our analytical formula of
20: differential energy distribution which is related with
21: the slope of inner cusp of dark halo.
22: We obtain the distribution function for the NFW profile which has sharp
23: cut off at the high binding energy.
24: We discuss physical reason of form of the analytical formula.
25: \end{abstract}
26:
27: \keywords{Cosmology: dark matter; Galaxies: Formation, Halos, clusters}
28:
29: \section{Introduction}
30: It is interesting to understand how the density profiles of
31: galaxies and clusters of galaxies have formed.
32: Navarro, Frenk and White (1995, 1996, 1997; NFW) have shown
33: in their N-body simulations of Cold Dark Matter (CDM) in
34: the standard biased CDM and four power law spectra with indices
35: $n = 0,-0.5,$ and $-1$, open CDM ($\Omega_0 =0.1$) with
36: power-law spectra ($n=0$ and $-1$),
37: and $\Lambda$CDM cosmology that
38: density profiles of dark halos have an universal profile described as
39: \begin{eqnarray}
40: \rho(r) \propto \frac{1}
41: {\left(\frac{r}{r_s}\right) \left( 1 + \frac{r}{r_s} \right)^2}.
42: \end{eqnarray}
43: Several investigators have shown that
44: the formula provides a good fit to their numerical results
45: (Cole and Lacey 1996, Tormen, Bouchet \& White 1997,
46: Huss, Jain \& Steinmmetz 1999, Thomas {\it et al.} 1998).
47: However, some other simulations (Fukushige \& Makino 1997;
48: Moore et al. 1998; Okamoto \& Habe 1999, 2000) indicate
49: that their density profiles have
50: steeper inner cusp than the NFW profile.
51: Jing (2000) gives his numerical results that
52: steeper cusps of density profile are found in recent mergers and
53: in dark halos with substructures, by
54: a large set of high resolution cosmological simulations.
55:
56: Subramanian, Cen and Ostriker (2000) discussed
57: the general theoretical grounds of the density profiles.
58: They discussed that there is a possible connection between
59: slope of inner region of dark halo and
60: the formation epoch and proposed that
61: there is possible relation to cosmological parameters.
62:
63: However, physical reason of the NFW profile have not been explored,
64: although it is very important to understand
65: formation of galaxies and clusters of galaxies.
66:
67: We study the differential energy distribution of the NFW profile,
68: $d M/ d \varepsilon$, where $\varepsilon$ is the binding energy.
69: The differential energy distribution was studied to understand virialization process of $N$- body system (van Albada 1982).
70: Binney (1982) shows that $dM/d \varepsilon$ of
71: the elliptical galaxies is approximated by the Boltzmann factor, if
72: surface brightness of elliptical galaxies obeys the
73: de Vaucoureurs' $r^{1/4}$ law and their mass-to-light ratio is constant.
74: It is interesting to study phase space structure of virialized objects
75: with the NFW profile,
76: since the differential energy distribution may give insight of the relaxation process (Binney \& Tremain 1987).
77:
78: In this paper we calculate formation of clusters of galaxies
79: by $N$-body simulation and we study
80: the differential energy distribution of the clusters in our numerical results.
81: We find an analytical formula of the fractional mass distribution
82: that is defined by the differential distribution divided by mass of the object,
83: is fitted by an analytical formula.
84: We show that
85: the NFW profile is consistent with our analytical formula of
86: the fractional mass distribution.
87: Using the iteration method, we construct the density profile
88: from the analytical formula to show how
89: the slope of the cusp of the density changes with
90: parameters in our analytical formula.
91:
92: In the next section,
93: we illustrate our numerical method.
94: In the section 3, we present our numerical results.
95: We show the iteration method and give results by this method
96: in the section 4, and we summarize and
97: discuss our results in the section 5.
98:
99:
100: \section{Numerical simulation}
101: We use the Hofman and Ribak's (Hofman and Ribak 1991)
102: procedure to set initial conditions
103: in order to have a massive dark halo
104: near the centre of a simulation box.
105: The cosmology is SCDM model (e.g. Davis {\it et al}. 1985)
106: ($\Omega = 1$, $\sigma_8 = 0.67$ ,
107: $H_0 = 100h$ km s${}^{-1}$ Mpc${}^{-1}$,and $h = 0.5$).
108:
109: Numerical simulations are carried out using GRAPESPH code.
110: GRAPE is a special purpose hardware to calculate gravitation
111: between N-body particles (Sugimoto {\it et al}. 1990).
112: We combined Smoothed Particle Hydrodynamics
113: (SPH) (Monaghan, 1992) with GRAPE.
114: We select massive halos of which mass is as large as
115: that of cluster of galaxies
116: and calculate their density profile and the fractional mass distribution.
117:
118: Mass of a CDM particle and a SPH particle are
119: $5.89 \times 10^{11} M_{\odot}$ and
120: $3.10 \times 10^{10} M_{\odot}$, respectively.
121: Gravitational softening length is 100 kpc.
122: Both number of CDM and SPH particles are 29855, respectively.
123: Size of the simulation box is $80$ Mpc.
124:
125: \section{Numerical results}
126:
127: \subsection{Density distribution}
128: Before we show our numerical results, we summarize the NFW profile.
129: NFW proposed that the profile of the dark halo of cosmological object in their numerical results as
130: \begin{eqnarray}
131: \frac{\rho(r)}{\rho_{cr}} = \frac{\delta_c}{\left(\frac{r}{r_s}\right)
132: \left( 1 + \left( \frac{r}{r_s} \right)\right)^2}, \label{NFW}
133: \end{eqnarray}
134: where $r_s = r_{200}/c$, $c$ is a dimensionless parameter and $\rho_{cr} = (3 H_0^2)/(8 \pi G)$ is the critical density of the universe .
135: Since the mass of the halo is
136: $M(r_{200}) = 200 \times 4 \pi /3 \rho_{cr} r_{200}^3$,
137: there is a relation between $\delta_c$ and $c$ as
138: \begin{eqnarray}
139: \delta_c = \frac{200}{3} \frac{c^3}{[ \ln{(1+c) - c/(1+c)} ]},
140: \end{eqnarray}
141: where $M(r_{200})$ is the mass of which averaged density
142: inside $r_{200}$ is $200$ times $\rho_{cr}$.
143:
144: This density profile has a cusp in the inner region,
145: $\rho(r) \propto r^{-1}$.
146: $\rho(r) \propto r^{-3}$
147: in the outer region.
148: NFW97 showed that the parameter $c$ decreases with halo mass.
149:
150: \placetable{table1}
151:
152: Table 1 is physical values of our simulated clusters.
153: The units of mass, and
154: X-ray temperature are $10^{15} M_{\odot}$ and $10^8 $ K,
155: respectively.
156:
157: \placefigure{typical_density_profile_with_NFW}
158:
159: Figure \ref{typical_density_profile_with_NFW} shows a density profile
160: of our simulated typical rich cluster, CLc,
161: and the NFW profile with $c = 4.4$
162: which fits well the numerical result.
163: Density profiles of dark halos obtained by us agree well with
164: the NFW profile in the range
165: from the gravitational softening length to $r_{200}$.
166:
167: \subsection{The differential energy distribution}
168:
169: We introduce the differential energy distribution,
170: $d M/ d\varepsilon$ which gives
171: the mass of dark matter in the dark halo with binding energy
172: between $\varepsilon$ and $\varepsilon + d \varepsilon$,
173: where $\varepsilon$ is the specific binding energy,
174: \begin{eqnarray}
175: \varepsilon = \Psi(r) - \frac{1}{2} v^2,
176: \end{eqnarray}
177: and a relative potential, $\Psi \equiv - \Phi + \Phi_0 $.
178: $\Phi$ is gravitational potential and we choose $\Phi_0$ to
179: be such that a distribution function, $f$, is $f > 0$ for $\varepsilon > 0$
180: and $f = 0$ for $\varepsilon \le 0$.
181: In our analysis, $\varepsilon$ is normalized by $GM_{200}/r_{200}$.
182: And we also introduce the fractional mass distribution as
183: the differential energy distribution divided by
184: the total mass of the dark halo,
185: $N(\varepsilon) = d M / d \varepsilon / M$.
186:
187: \placefigure{fractional_mass_distribution1}
188:
189: Figure \ref{fractional_mass_distribution1} shows the fractional mass distribution of CLc.
190: In figure \ref{fractional_mass_distribution1},
191: we also show $N(\varepsilon)$ given by
192: \begin{eqnarray}
193: N(\varepsilon) = N_0 \left [ 1 - (1-q) \left( \frac{\varepsilon }
194: {\varepsilon_0} \right) \right]^{q/(1-q)},
195: \label{escort}
196: \end{eqnarray}
197: with $q = 0.667$ and $\varepsilon_0 = 1.47$.
198: Figure \ref{fractional_mass_distribution1} shows that
199: equation (\ref{escort}) agrees well with our numerical results in the range of
200: $0.5 < \varepsilon < 4$.
201: There is cut off near $\varepsilon \simeq 4$.
202: We find the fractional distribution $N(\varepsilon)$ can be fitted by
203: following formula,
204: for $q \simeq 0.6-0.7$ and
205: $\varepsilon_0 \simeq 1.2-1.5 GM_{200}/r_{200}$ for
206: rich clusters in our numerical results.
207:
208: \placetable{table2}
209:
210: In table 2 we give $N_0$, $q$,
211: and $\varepsilon_0$ of our numerical results for rich clusters.
212:
213: We have shown from our $N$- body simulation that the fractional mass distribution is also well fitted by the equation (\ref{escort}).
214: However, near $\varepsilon = 4$ in figure \ref{fractional_mass_distribution1} there are small number of $N$-body
215: particles.
216: We should confirm consistency between
217: $N(\varepsilon)$ given by equation (\ref{escort}) and the NFW profile.
218: We give a fractional mass distribution from
219: the NFW profile as follows for the comparison.
220:
221: We assume that phase-space distribution function $f({\bf x}, {\bf v})$ depends $\varepsilon$.
222: At a radius $r$, velocity of a dark matter particle of
223: the binding energy, $\varepsilon$, is $v = \sqrt{2(\Psi - \varepsilon)}$.
224: The density profile may be given as follows (Binney and Tremaine 1987)
225: \begin{eqnarray}
226: \rho (r) = 4 \pi \int_{\Psi(r_g)}^{\Psi(r)} f(\varepsilon)
227: [2 (\Psi - \varepsilon ) ]^{1/2} d \varepsilon, \label{density}
228: \end{eqnarray}
229: where $r_g$ is the edge of the dark halo.
230: From this equation, we may give $f(\varepsilon)$ as
231: \begin{eqnarray}
232: f(\varepsilon) = \frac{1}{\sqrt{8} \pi^2} \frac{d}{d \varepsilon}
233: \int_{\varepsilon_{min}}^{\varepsilon}
234: \frac{d \rho/ d \Psi}{[\varepsilon - \Psi ]^{1/2}} d \Psi, \label{fe}
235: \end{eqnarray}
236: where $\varepsilon_{min} = \Psi(r_g)$.
237:
238: Equation (\ref{density}) gives mass $M$ as
239: \begin{eqnarray}
240: M(r) = &&16 \pi^2 \int_0^{r} r^{2} d r \nonumber \\
241: && \times \int_0^{\Psi(r)} f(\varepsilon)
242: [2 (\Psi - \varepsilon ) ]^{1/2} d \varepsilon.
243: \label{mr}
244: \end{eqnarray}
245: From equation (\ref{mr}), the differential energy distribution is
246: \begin{eqnarray}
247: \frac{dM(\varepsilon)}{ d \varepsilon}
248: = f(\varepsilon) g(\varepsilon), \label{dmde}
249: \end{eqnarray}
250: where
251: \begin{eqnarray}
252: g(\varepsilon) = 16 \pi^2 \int_0^{r_m(\varepsilon)}
253: [2 (\Psi - \varepsilon) ]^{1/2} r^2 dr, \label{ge}
254: \end{eqnarray}
255: and $r_m(\varepsilon)$ is maximum radius
256: that can reached by a particle of the binding energy $\varepsilon$.
257:
258: If we assume the density profile is the NFW profile, we get $d M/ d \varepsilon $ from the equations (\ref{fe}), (\ref{dmde}),
259: and (\ref{ge}) for the NFW profile.
260:
261: \placefigure{fractional_mass_distribution2}
262:
263: Figure \ref{fractional_mass_distribution2} shows
264: the fractional mass distribution of NFW obtained in this way and
265: $N(\varepsilon)$ given by equation (\ref{escort}).
266: We show that $N(\varepsilon)$ given by equation (\ref{escort}) is
267: consistent with the NFW profile.
268: We note that the cut off at high binding energy seen in
269: figure \ref{fractional_mass_distribution1} is not artifact
270: due to limitation of our numerical resolutions.
271:
272: \placefigure{fe-e}
273:
274: Figure \ref{fe-e} shows the distribution function of the NFW profile
275: obtained by the above method.
276: Since lower binding energy particles evaporate from
277: the dark halo, their fraction become small.
278: This form of function is similar to the distribution function of King model,
279: $f_K(\varepsilon) \propto (e^{\varepsilon/\sigma^2} -1)$ (Binney 1982).
280: On the other hand, there is a peak and sharp cut off at the high binding energy.
281: This part corresponds to the central cusp.
282: Although the distribution function of NFW profile have cut off
283: at the high energy, the distribution function of King model
284: does not have such cut off.
285: This is an important difference between them.
286:
287: \section{The fractional mass distribution, density profile,
288: and the distribution function}
289:
290: We find that the NFW profile satisfies the fractional mass distribution
291: given by equation (\ref{escort}) with $q \simeq 0.6-0.7$.
292: We study how the density profile changes
293: when we change the parameter $q$ and $\varepsilon_0$
294: in equation (\ref{escort}).
295: In this study, we use an iteration method
296: as shown in the next subsections.
297:
298: \subsection{The iteration method}
299: Binney (1982, and see also Binney and Tremaine 1987) studied
300: the phase space structure of galaxies of
301: which surface brightness is the de Vaucouleurs' $r^{1/4}$ law.
302: We apply his method to our study of
303: the phase space structure of dark halo with the NFW profile.
304: We obtain the density profile and the phase space distribution
305: which are consistent with equation (\ref{escort}),
306: using the iteration method.
307: The procedure is as follows.
308:
309: We assume the dark halo is spherically symmetric.
310: From equation (\ref{escort}),
311: we assume $dM(\varepsilon)/d\varepsilon$
312: given by equation,
313: \begin{eqnarray}
314: \frac{dM (\varepsilon)}{d \varepsilon} = M_0 \left[
315: 1 - (1 - q) \left(\frac{\varepsilon}{\varepsilon_0}\right)
316: \right]^{q/(1-q)},
317: \label{escort_m}
318: \end{eqnarray}
319: where $M_0 = 1.0$.
320: We use a trial function of a density profile,
321: to calculate a gravitational potential
322: in $g_0(\varepsilon)$,
323: \begin{eqnarray}
324: \rho_0(x) = \frac{\rho_c}
325: {\left(1 +\left(\frac{x}{x_c}\right)^2 \right)^{3/2}},
326: \end{eqnarray}
327: where $x_c = r_c/r_{200} = 0.1$.
328: This trial function is similar to the $\beta$-model of
329: X-ray surface brightness of clusters of galaxies.
330: We obtain relative potential from this trial density function.
331: The relative potential is written as
332: \begin{eqnarray}
333: \Psi(x) &= &4 \pi G \left[
334: \frac{1}{x} \int_0^x \rho_0(x^{\prime} ) x^{\prime^2} d x^{\prime}
335: \right. \nonumber \\
336: &&\left. + \int_x^{x_{o}} \rho_0(x^{\prime}) x^{\prime} dx^{\prime}
337: \right] + \Phi_{x_{o}}, \label{Psi}
338: \end{eqnarray}
339: where $x = r/r_{200}$, and $x_o = 10$ is assumed.
340: $\Phi_{x_{o}}$ is chosen to
341: be $f > 0$ for $\varepsilon > 0$
342: and $f = 0$ for $\varepsilon \le 0$.
343: We obtain $g_0(\varepsilon)$ by equation (\ref{ge}) for $\rho_0$.
344: From equation (\ref{dmde}),
345: \begin{eqnarray}
346: f_0(\varepsilon) = \frac{dM(\varepsilon)/d \varepsilon}{g_0(\varepsilon)}.
347: \end{eqnarray}
348:
349: Next, we calculate new density as,
350: \begin{eqnarray}
351: &&\rho_{i+1}(x) = (1 - \alpha) \rho_i(x) \nonumber \\
352: &&+ \alpha \times 4 \pi \int_{\Psi_i > \varepsilon}
353: f_i(\varepsilon) \sqrt{2(\Psi_i - \varepsilon)} d \varepsilon,
354: \label{rho_i+1}
355: \end{eqnarray}
356: where $i$ is an iterative index and
357: $\alpha$ is an arbitrary constant of
358: convergence.
359: Here we assume $\alpha = 0.5$.
360: Next we again follow the step from equation (\ref{Psi}) to equation
361: (\ref{rho_i+1}) by using $\rho_1$ instead of $\rho_0$ and
362: we obtain $\rho_2$.
363: We continue until $\rho_i$ becomes converge.
364: In this way, we obtain $\rho$ and $f(\varepsilon)$ which are consistent with equation (\ref{escort_m}).
365:
366: \subsection{The density profile and the distribution function}
367:
368: Using the Binney's iteration method,
369: we reconstruct mass density profile
370: from equation (\ref{escort_m}).
371:
372: \placefigure{constructed_density1}
373:
374: We show our result for $q=0.67$ and $\varepsilon = 1.4$ in figure
375: \ref{constructed_density1}.
376: A solid line is given by the iteration method and
377: a dashed line is the NFW profile.
378: We confirm that $dM(\varepsilon)/d \varepsilon$ characterizes
379: well the NFW profile.
380:
381: \placefigure{q_density_relation}
382:
383: In figure \ref{q_density_relation},
384: we show density profiles for different $q$ but $\varepsilon = 1.4$.
385: Smaller $q$ (e.g. $q = 0.5$) results in shallower core
386: in the inner region.
387: On the other hand, larger $q$ ($q > 0.67 $)
388: makes a cusp steeper than the NFW profile, density profile
389: approaches $\rho \propto r^{-2}$
390: in the inner part, for $q \longrightarrow 1$.
391:
392: \placefigure{e_density_relation}
393:
394: For various values of $\varepsilon_0$,
395: the density profiles are similar to the NFW for $q = 0.6$
396: as shown in figure \ref{e_density_relation}.
397: Absolute value of the density depends on $\varepsilon_0$.
398: Therefore, the slope of the cusp depends on only $q$, not $\varepsilon_0$.
399:
400: \placefigure{q_dependence_f}
401:
402: Figure \ref{q_dependence_f} shows
403: the distribution function, $f$, obtained by the iteration method for
404: various values of $q$ .
405: These curves show the same dependence on $\varepsilon$ in $0 < \varepsilon <1$.
406: Peak values of $f$ are different each other.
407: We also show the Boltzmannian distribution for comparison
408: in figure \ref{q_dependence_f}.
409: For large $q$, peak value of $f(\varepsilon)$ and
410: the maximum binding energy of the distribution become large.
411: We have shown that the density profile with large $q$ have the steep cusp.
412: Therefore, sharp peak of $f(\varepsilon)$ corresponds to the steep cusp.
413:
414: \placefigure{e_dependence_f}
415:
416: Figure \ref{e_dependence_f} shows $f$ obtained by the iteration method
417: for same $q$ but various $\varepsilon_0$.
418: These curves do not show the same dependence on binding energy.
419: Height of peaks of these curves is constant.
420:
421:
422: \section{Summary and Discussion}
423: We analyze the universal density profile of dark halo proposed by NFW and its
424: differential energy distribution.
425: Our main results are summarized as follows.
426:
427: \begin{enumerate}
428: \item
429: We study
430: the fractional mass function $N(\varepsilon)$ for dark halo obtained
431: by our numerical simulation and find its analytical formula
432: which is the equation (\ref{escort}).
433: \item We show that the NFW profile is
434: given by the equation (\ref{escort_m}).
435: \item We show that the slope of the cusp in the density profile changes with a value of the parameter $q$ in the analytical formula.
436: \end{enumerate}
437:
438: We can regard that $N(\varepsilon)$ shows the statistical property of the NFW profile.
439: If the NFW profile is universal, $q = 0.6 - 0.7$ in equation (\ref{escort}).
440: Different $q$ makes slope of a cusp different.
441: Since $q$ plays an important role, we should make clear
442: what physical process determines $q$.
443: Recent high resolution numerical simulation
444: (Okamoto \& Habe 1999, 2000) shows the steeper cusp,
445: $\rho \propto r^{-1.5}$, than the the NFW profile.
446: This profile corresponds to $q = 0.75 - 0.8$.
447: Isothermal profile, $\rho \propto r^{-2}$, corresponds to $q = 1$.
448:
449: We study $f(\varepsilon)$ for the NFW profile.
450: The formula of this is not isothermal one nor
451: the King formula $f_K \propto e^{\varepsilon/\sigma^2} -1$.
452: $f(\varepsilon)$ for the NFW profile have the energy cut off
453: at the high end of $\varepsilon$.
454: We should study the reason why $f(\varepsilon)$ has such a form.
455: Lynden-Bell (1967) studied distribution function $f(\varepsilon)$ of a
456: virialized system.
457: Maximizing the Boltzmann entropy of the system,
458: resulting distribution is isothermal profile, $ \rho \propto r^{-2}$.
459: In this case the system has infinite extend and infinite mass.
460: This is not realistic for astronomical objects.
461: Cosmological simulations have shown that
462: galaxies and clusters of galaxies formed in these simulations have
463: more rapid radial decline than isothermal in the outer part.
464:
465: We note that the form of the distribution function
466: relates with the cusp profile.
467: For $q < 1$, there are the maximum binding energy
468: $\varepsilon_{{\rm max}}$ and the sharp peak in $f(\varepsilon)$
469: at $\varepsilon_{{\rm max}}$.
470: Since there is $\varepsilon_{{\rm max}}$,
471: phase space volume occupied by dark matter is limited
472: than the isothermal profile.
473: This may be the reason of
474: the cusp profile different from $\rho \propto r^{-2}$.
475: For $q < 1$, the distribution function $f(\varepsilon)$ changes with
476: $q$ values as shown in figure 8.
477:
478: We note that the form of equations
479: (\ref{escort}) and (\ref{escort_m}) are
480: similar to the Tsallis' escort distribution,
481: \begin{eqnarray}
482: \frac{P(E,T^{\prime})}{P(0,T^{\prime})} =
483: \left[ 1 - ( 1- q) \frac{E}{T^{\prime}} \right]^{q/(1-q)},
484: \end{eqnarray}
485: where $T^{\prime}$ is temperature parameter
486: and $q$ is entropic index (Tsallis, Mendes, \& Plastino, 1998).
487: Tsallis' non-extensive generalized statistics (Tsallis 1988) is
488: paid attention in the area of statistics of a multi-fractal system.
489: In non-extensive system
490: (long-range microscopy memory, long range forces,
491: fractral space time) the following generalized entropy has been proposed :
492: \begin{eqnarray}
493: S_q = k \frac{1 - \sum_i p_i^q}{q-1} \ \ \
494: \left( \sum_i = p_i = 1; \ \ \ q \in \Re \right),
495: \end{eqnarray}
496: where $k$ is a positive constant.
497: Optimization of $S_q$ yields, for the canonical ensemble,
498: \begin{eqnarray}
499: &&p_i = Z_q^{-1} [1 - (1-q) \varepsilon_i/T^{\prime} ]^{1/(1-q)}, \\
500: &&Z_q \equiv \sum_i [1 - (1-q) \varepsilon_i/T^{\prime} ]^{1/(1-q)}
501: \end{eqnarray}
502: and, when $q \longrightarrow 1$,
503: the Boltzmann-Gibbs result is recovered.
504: In this statistics an expected value of any physical variable is given by
505: the Tsallis' escort distribution :
506: \begin{eqnarray}
507: \left< A \right>_q = \frac{ \sum_i p_i^q A_i}{\sum_j p_j^q},
508: \end{eqnarray}
509: where $\{A_i\}$ are the eigenvalues of an arbitrary observable $A$.
510: The escort distribution, $P_i = p_i^q/\sum_j p_j^q$, is
511: similar to the form of equation (\ref{escort}) and (\ref{escort_m}).
512:
513: It is expected that long range interaction makes
514: the system non-extensive.
515: Our case may be the one of this cases.
516: Our results suggests that differential energy distribution of
517: collisionless particles is described by the escort distribution
518: of the Tsallis statistics.
519: This may indicate that energetic process of collisionless particles must be stochastic process.
520:
521: In the Tsallis' escort distribution function,
522: there is the maximum value of $\varepsilon$
523: for $p_i > 0$ for $0 < q < 1$.
524: This case is called superextensive.
525: The shallower cusp profile in the the NFW profile shows
526: the superextensive property of the dark matter distribution.
527:
528: We show that the NFW profile corresponds to $q = 0.6-0.7$.
529: Lavagno {\it et al.} (1998) have recently shown
530: that fraction of peculiar velocity of
531: cluster of galaxies (Bahcall \& Oh 1996) is well explained by
532: the Tsallis escort integral,
533: \begin{eqnarray}
534: P(> v) = \frac{\int_v^{v_{max}} (1 - (1-q)(v/v_0)^2)^{q/(1-q)}dv}
535: {\int_0^{v_{max}} (1 - (1-q)(v/v_0)^2)^{q/(1-q)}dv}.
536: \end{eqnarray}
537: They obtained $q=0.23$ to fit the fraction of peculiar velocity of cluster of galaxies and is smaller than in our case.
538: There is a conjecture that $q$ of system approaches unity
539: when the system proceeds relaxation (Tsallis 1999).
540: Our results in which $q = 0.6-0.7$
541: are consistent with this conjecture,
542: since dark matter distribution in a cluster of galaxies is
543: more relaxed system than
544: the large scale motion of clusters of galaxies.
545: The value of $q$ must be related with
546: the degree of relaxation of collisionless particles.
547: $q = 1$ corresponds to isothermal distribution in the Tsallis statistics.
548: The gravitational system like clusters of galaxies
549: with the universal profile may be non-extensive
550: because they formed recently.
551:
552:
553: We should study the reason why dark halo has the value of $q = 0.6-0.7$ in the hierarchical clustering scenarios.
554: It is interesting to study the differential energy distribution of
555: self-gravitational system formed in a circumstance without
556: hierarchical clustering to make clear mechanism what determines $q$ of
557: the gravitational system.
558:
559: \acknowledgments
560: We would like to thank Sumiyoshi Abe, Masayuki Fujimoto, and Seiichi Yachi for helpful comments and discussion.
561: %This research is supported in part by grants
562:
563: \begin{thebibliography}{99}
564: \bibitem{BO96} Bahcall, N. A., \& Oh., S. P., 1996, \apjl, 462, L49.
565:
566: \bibitem{Binney82} Binney, J. 1982, \mnras, 200, 951.
567:
568: \bibitem{BT87} Binney, J. \& Tremain S., 1987,
569: Galactic Dynamics (New Jersey : Princeton University Press).
570:
571: \bibitem{CL96} Cole, S. M., \& Lacey, C., 1996, \mnras, 281, 716.
572:
573: \bibitem{DEFW85} Davis, M., Efstathiou, G., Frenk, C.S., and White, S.D.M, 1985 \apj, 292, 371
574:
575: \bibitem{FM97} Fukushige, T., \& Makino, J., 1997, \apj, 484, 40.
576:
577: \bibitem{HR91} Hofman, Y. \& Ribak, E., 1991, \apjl, 380, L5.
578:
579: \bibitem{HJS99} Huss, A., Jain, B., \& Steinmetz, M, 1999,
580: \apj, 517, 64.
581:
582: \bibitem{Jing00} Jing., Y. P., 2000, \apj 535, 30.
583:
584: \bibitem{Lavagnoetal98} Lavagno, A., Kaniadakis, G., Rego-Monteiro, M.,
585: Quarati., P., \& Tsallis. C., 1998, Astro. Lett. and Communications, 35, 449.
586:
587: \bibitem{Lynden-Bell67} Lynden-Bell, D. 1967, \mnras 136, 101.
588:
589: \bibitem{Monaghan92} Monaghan, J.J., 1992, ARRA, 30, 543
590:
591: \bibitem{Mooreetal98} Moore, B., Governato, F,. Quinn, T., Stadel, J., \&
592: Lake, G., 1998, \apjl 499, L5.
593:
594: \bibitem{NFW95} Navarro, J. F., Frenk, C. S., \& White S. D. M. 1995,
595: \mnras, 275, 720.
596:
597: \bibitem{NFW96} Navarro, J. F., Frenk, C. S., \& White S. D. M. 1996,
598: \apj, 462, 563.
599:
600: \bibitem{NFW97} Navarro, J. F., Frenk, C. S., \& White S. D. M. 1997,
601: \apj, 490, 493.
602:
603: \bibitem{OH99} Okamoto, T., \& Habe, A. 1999, \apj, 516, 591.
604:
605: \bibitem{OH00} Okamoto, T., \& Habe, A. 2000, \pasj, 52, 457.
606:
607:
608: \bibitem{SCMIEU90} Sugimoto, D., Chikada, Y,. Makino, J., Ito, T., Ebisuzaki, T., \& Umemura M., 1990, \nat, 345,33
609:
610: \bibitem{SCO00} Subramanian, K., Cen, R., \& Ostriker, J. P., 2000, \apj, 538, 528,
611:
612: \bibitem{Tetal98} Thomas, P. A., {\it et al.}, 1998, \mnras, 296, 1061.
613:
614: \bibitem{TBW97} Tormen, G., Bouchet, F. R., \& White, S. D. M., 1997,
615: \mnras, 286, 865.
616:
617: \bibitem{Tsallis88} Tsallis, C., 1988, J. Stat. Phys. 52, 479.
618:
619: \bibitem{Tsallis99} Tsallis, C., 1999, Braz. J. Phys. 29, 1.
620:
621: \bibitem{TMP98} Tsallis, C., Mendes, R. S., Plastino, A. R., 1998, Physica A,
622: 261, 534
623:
624: \bibitem{vanAlbada} van Albada, T.S. 1982, \mnras, 201, 939.
625:
626: \end{thebibliography}
627:
628: \begin{figure}
629: \plotone{f1.eps}
630: \caption{Density profile of one of simulated clusters, CLc, and
631: the NFW profile with $c = 4.4$.
632: Solid line is our numerical result and dashed line is
633: the NFW profile.
634: Radial distance is normalized by $r_{200}$.
635: Our simulated clusters agree well with
636: the NFW formula between the gravitational softening length to $r_{200}$.}
637: \label{typical_density_profile_with_NFW}
638: \end{figure}
639:
640: \begin{figure}
641: \plotone{f2.eps}
642: \caption{Fractional mass distribution of our simulated rich cluster, CLc.
643: Solid line is our numerical result and dashed line is given by equation (\ref{escort}) with $q = 0.667$ and $\varepsilon_0 = 1.47$.}
644: \label{fractional_mass_distribution1}
645: \end{figure}
646:
647: \begin{figure}
648: \plotone{f3.eps}
649: \caption{Fractional mass distribution of dark halo.
650: Solid line is caliculated by equations (\ref{fe}), (\ref{dmde}),
651: and (\ref{ge}) and dashed line is given by equation (\ref{escort})
652: with $q = 0.66$ and $\varepsilon_0 = 1.4$.}
653: \label{fractional_mass_distribution2}
654: \end{figure}
655:
656: \begin{figure}
657: \plotone{f4.eps}
658: \caption{$f(\varepsilon)$ of the NFW profile.
659: Solid line is $f(\varepsilon)$ and dashed line is a fitted exponential
660: function.
661: }
662: \label{fe-e}
663: \end{figure}
664:
665: \begin{figure}
666: \plotone{f5.eps}
667:
668: \caption{Density profile constructed by the iteration method, assuming the equation (\ref{escort_m}).
669: Solid line is our result for $q = 0.67$ and $\varepsilon_0 = 1.4$
670: and dashed line is the NFW profile of $c = 6.59$.
671: Radial distance is normalized by $r_{200}$.}
672: \label{constructed_density1}
673: \end{figure}
674:
675: \begin{figure}
676: \plotone{f6.eps}
677: \caption{
678: Density profiles for various $q$.
679: Dot-dashed, dotted, short-dashed, dashed, and solid lines are for
680: $q = 0.25, 0.5, 0.67, 0.75, $ and $1.0$, respectively.
681: Density profile with smaller $q$ has a flat core.
682: On the other hand, one approaches $\rho \propto r^{-2}$
683: in $r < 0.1$ for $q \longrightarrow 1$.}
684: \label{q_density_relation}
685: \end{figure}
686:
687: \begin{figure}
688: \plotone{f7.eps}
689: \caption{
690: Density profiles for various $\varepsilon$.
691: Dotted, short-dashed, dashed, and solid lines are for
692: $\varepsilon_0 = 0.5, 1.0, 1.5, $ and $2.0$, respectively.
693: Density profiles are self-similar.
694: However, their normalizations are different.}
695: \label{e_density_relation}
696: \end{figure}
697:
698: \begin{figure}
699: \plotone{f8.eps}
700: \caption{We show $f(\varepsilon)$ for various $q$ ;
701: Dot-dashed, dotted, short-dashed, dashed, and solid lines are for
702: $q = 0.25, 0.5, 0.67, 0.75, $ and $1.0$, respectively.}
703: \label{q_dependence_f}
704: \end{figure}
705:
706: \begin{figure}
707: \plotone{f9.eps}
708: \caption{We show $f(\varepsilon)$ for various $\varepsilon_0$;
709: Dotted, short-dashed, dashed, and solid lines are for
710: $\varepsilon_0 = 0.5, 1.0, 1.5$, and $2.0$, respectively.}
711: \label{e_dependence_f}
712: \end{figure}
713:
714:
715: \begin{table}
716: \caption{The physical values of our simulated clusters}
717: \begin{tabular}{lccc}
718: \hline \hline
719: & Mass & $T$ & $c$ \\
720: &($10^{15}M_{\odot}$) & ($10^8$ K) \\ \hline
721: CLa & 3.05 & 0.953 & 2.50 \\
722: CLb & 2.80 & 0.890 & 2.72 \\
723: CLc ($z = 0$) & 3.36 & 1.29 & 4.41 \\
724: CLc ($z = 0.25$) & 1.60 & 0.796 & 2.31 \\
725: CLc ($z = 0.5$) & 0.831 & 0.540 & 2.67 \\
726: CLd & 3.71& 1.57 & 6.34 \\
727: CLe & 2.82& 1.19 & 4.65 \\
728: CLf & 2.31& 0.654 & 2.75 \\ \hline
729: \end{tabular}
730: \tablecomments{
731: Each value is calculated at redshift $z = 0$, except for CLc.
732: }
733: \label{table1}
734: \end{table}
735:
736: \begin{table}
737: \caption{$N_0$, $q$, and $\varepsilon_0$ of rich clusters in
738: our numerical results}
739: \begin{tabular}{lccc}
740: \hline \hline
741: & $N_0$ & $q$ & $\varepsilon_0$ \\ \hline
742: CLa & 0.309 & 0.612 & 1.27\\
743: CLb & 0.253 & 0.680 & 1.21\\
744: CLc ($z = 0$) & 0.190 & 0.667 & 1.47 \\
745: CLc ($z = 0.25$) & 0.230 & 0.606 & 1.05 \\
746: CLc ($z = 0.5$) & 0.305 & 0.600 & 0.672 \\
747: CLd & 0.276 & 0.699 & 1.26 \\
748: CLe & 0.117 & 0.503 & 2.03 \\
749: CLf & 0.258 & 0.580 & 1.41 \\ \hline
750: \end{tabular}
751:
752: \tablecomments{
753: Each value is calculated at redshift $z = 0$, except for CLc.
754: }
755: \label{table2}
756: \end{table}
757:
758: \end{document}
759: