1: \documentclass[12pt,preprint]{aastex}
2: \begin{document}
3: \title{Cluster-Supercluster Alignments}
4: \author{\sc Jounghun Lee\altaffilmark{1} and
5: August E. Evrard\altaffilmark{2,3,4}}
6: \altaffiltext{1}{Department of Physics and Astronomy, FPRD, Seoul National
7: University, Seoul 151-747 , Korea}
8: \altaffiltext{2}{Department of Astronomy and Michigan Center for Theoretical
9: Physics, University of Michigan, Ann Arbor, 500 Church St., MI 48109, USA}
10: \altaffiltext{3}{Department of Physics and Michigan Center for Theoretical
11: Physics, University of Michigan, Ann Arbor, 450 Church St.,MI 48109-1040, USA}
12: \altaffiltext{4}{Visiting Miller Professor, Physics Department,
13: University of California, Berkeley, CA94720, USA}
14: \begin{abstract}
15: We study correlations in spatial orientation between galaxy clusters and
16: their host superclusters using a Hubble Volume N-body realization of a
17: concordance cosmology and an analytic model for tidally-induced alignments.
18: We derive an analytic form for distributions of the alignment angle as
19: functions of halo mass ($M$), ellipticity ($\epsilon$), distance ($r$) and
20: velocity ($v$) and show that the model, after tuning of three parameters,
21: provides a good fit to the numerical results. The parameters indicate a
22: high degree of alignment along anisotropic, collapsed filaments.
23: The degree of alignment increases with $M$ and $\epsilon$ while it
24: decreases with $r$ and is independent of $v$. We note the possibility of
25: using the cluster-supercluster alignment effect as a cosmological probe to
26: constrain the slope of the initial power spectrum.
27: \end{abstract}
28: \keywords{cosmology:theory --- large-scale structure of universe}
29:
30: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
31: \section{INTRODUCTION}
32:
33: Superclusters are collections of galaxy groups and clusters that
34: represent the largest, gravitationally bound structures in the universe
35: \citep{sha30,kal-etal98}. If the dark energy is a cosmological
36: constant, then the collapse of these systems over the next few billion
37: years of the cosmic future will mark the end of hierarchical structure
38: formation in our universe \citep{nag-loeb03,busha05}. A conspicuous feature
39: of locally observed superclusters is the strong tendency of member clusters
40: to be elongated in their major axis orientations \citep{pli02,pli04},
41: which is in turn closely related to their filamentary shapes
42: \citep[e.g.,][]{bas03}. To describe the structure distribution
43: on the largest scale in the universe, it will be quite essential to
44: understand this effect of cluster-supercluster alignments from
45: first principles.
46:
47: The effect of structure-substructure alignment is in fact
48: observed on all different scales in the universe. On the subgalactic
49: scale the galaxy satellites are observed to be located preferentially
50: near the major axes of their host galaxies
51: \citep{val-etal78,kne-etal04,bra05,agu-bra06}. On the galactic scale
52: the major axes of cluster galaxies are observed to be aligned with
53: that of their host clusters \citep{pli-bas02,pli-etal03}. The cluster
54: galaxies are also observed to have a strong tendency toward radial
55: alignment \citep{per-kuh05}.
56:
57: Although this alignment effect has been shown to be a natural outcome
58: in the currently favored concordance $\Lambda$CDM cosmology
59: \citep{onu-tho00,lib-etal05,kan-etal05,lee-etal05,zen-etal05,kas-evr05,
60: bas-etal06}, its detailed origin remains a subject of debate between those
61: who emphasize the importance of anisotropic merging and those who stress
62: tidal interaction.
63:
64: The anisotropic merging scenario explains that the effect of
65: substructure-structure alignment is induced by the anisotropic
66: merging and infall of matter along filaments \citep{wes89}.
67: It was indeed shown by N-body simulations that the merging and infall
68: of matter to form bound halos indeed occur preferentially along filaments,
69: which provided supporting evidences for this scenario
70: \citep[e.g.,][]{wes-etal91,van-van93,dub98,fal-etal02,kne-etal04,zen-etal05}.
71:
72: The tidal interaction theory explains that the correlations between the
73: substructure angular momentum vectors and the principal axes of the
74: host tidal fields induce the alignment effect. \citet{lee-etal05} constructed
75: an analytic model for the effect of substructure alignment in the frame of the
76: tidal interaction theory, and showed that their analytic predictions are
77: in good agreement with the numerical results from N-body simulations.
78:
79: In fact, the above two theories are not mutually exclusive since the
80: anisotropic merging and infall itself is a manifestation of the primordial
81: tidal field \citep{bon-etal96}. The difference between the two scenarios,
82: however, lies in the question of whether the connection to filaments is a
83: major contribution or not.
84:
85: Very recently, \citet{atl-etal06} have quantified the influences
86: of both the tidal interaction and the anisotropic infall through the
87: analysis of data from recent high-resolution N-body simulations. What they
88: confirmed is the following: (i) For the majority of halos the alignment effect
89: is caused by the tidal field but not by the anisotropic infall; (ii) Only for
90: the cluster-size halos the alignment effect is dominantly due to the
91: anisotropic merging and infall of matter along filaments. In other words,
92: they made it clear that the filaments are important marker of local
93: orientations on the cluster halo scale.
94:
95: Now that the cluster-supercluster alignment turns out to be due to the
96: anisotropic merging along filaments, it is desirable to have a
97: theoretical frame work within which one can provide physical answers to the
98: remaining questions such as how the alignment effect depends on the
99: cluster properties such as mass, shape, and etc. Our goal here is to
100: construct such a theoretical framework by using both analytical
101: and numerical methods. Analytically we adopt the standard
102: cosmic web theory, and numerically we use the large Hubble volume
103: simulation data.
104:
105: The organization of this paper is as follows. In $\S 2$ we provide a
106: brief description of the Hubble volume simulation and summarize the
107: numerical results. In $\S 3$ we present an analytic model and compare
108: its predictions with the numerical results. In $\S 4$ we discuss our
109: results and draw final conclusions.
110:
111: \section{NUMERICAL RESULTS}
112:
113: For the numerical analysis, we use a mass-limited sample of cluster
114: halos extracted from the Hubble Volume simulation of a $\Lambda$CDM
115: universe \citep{evr-etal02}. The simulation models dark matter structure
116: resolved by particles of mass $m=2.25\times 10^{12}h^{-1}M_{\odot}$ in a
117: periodic cube of linear size $3000h^{-1}$Mpc, assuming
118: $\Omega_m = 0.3,\Omega_{\Lambda}=0.7$ and $\sigma_{8}=0.9$.
119: The $z=0$ catalog contains a total of $82973$ halos with
120: mass above a limiting value $M > 3\times 10^{14}h^{-1}M_{\odot}$,
121: % $345000$ clusters with mean mass $\bar{M}=5\times
122: % 10^{14}h^{-1}M_{\odot}$
123: with information on various properties such as center-of-mass position,
124: mass, inertia momentum tensor, and redshift. We refer the readers to
125: \citet{evr-etal02} and \citet{kas-evr05} for the details of the cluster
126: catalog, including the algorithm of cluster identification.
127:
128: The superclusters are identified in the catalog with the help of the
129: friends-of-friends algorithm with the linking length of $0.33\bar{l}$,
130: where $\bar{l} = 69h^{-1}$Mpc is the mean spacing of the mass-limited
131: sample. The total number and the mean mass of the identified superclusters
132: are $N_{s}=14007$ and $\bar{M}_{s}=1.26 \times 10^{15}h^{-1}M_{\odot}$,
133: respectively. This large number of superclusters allows us to study
134: the alignment effect with high statistical power.
135:
136: Figure~\ref{fig:sc3} shows orthogonal projections of the third
137: richest supercluster in the volume. It contains 12 halos above the
138: applied mass limit, and a total mass of $5.3 \times 10^{15}
139: h^{-1} \, {\rm M}_{\odot}$ associated with these halos. The spatial
140: distribution of the supercluster is highly elongated, much closer to
141: filamentary than spherical. In this example, the major axis orientations
142: of the halos, taken from \cite{kas-evr05}, are shown as whiskers in the plot.
143: The tendency for these halos to be aligned with their supercluster's principal
144: axis, although arguably visible in this plot, is a weak effect.
145: We therefore seek a statistical measure using the entire supercluster sample.
146:
147: For each supercluster, we measure its inertia momentum tensor,
148: ${\bf I}^{s} \equiv (I^{s}_{ij})$, as
149: \begin{equation}
150: I^{s}_{ij} = \frac{1}{M_{s}}\sum_{\alpha}M^{\alpha}_{c}x^{\alpha}_{c, i}
151: x^{\alpha}_{c,j},
152: \label{eqn:ine}
153: \end{equation}
154: where $M^{\alpha}_{c}$ and ${\bf x}^{\alpha}_{c} \equiv (x^{\alpha}_{c,i})$
155: represent the mass and the position of the $\alpha$-th member cluster,
156: respectively and $M_{s}$ is the total mass of the host supercluster.
157: Then, we compute the eigenvalues and eigenvectors through the diagonalization
158: of ${\bf I}^{s}$ and determine the major-axis direction as the direction of
159: the eigenvector corresponding to the largest eigenvalue.
160:
161: It is, however, worth mentioning here that for a supercluster which has
162: less than five clusters, the orientation of its major axis derived
163: using equation (\ref{eqn:ine}) must suffer from considerable inaccuracy.
164: The most idealistic way should be to derive the major axis of a supercluster
165: using all particles within it.
166:
167: Nevertheless, given that the major axes of the superclusters in
168: real observations cannot be determined by this idealistic way
169: since the positions of the dark matter particles are not measurable, the
170: advantage of our analysis based (eq.[\ref{eqn:ine}]) is its practicality.
171: That is, it can be readily repeated by observers based directly on cluster
172: catalogs.
173:
174: At any rate, to overcome the limitation of our analysis based on equation
175: (\ref{eqn:ine}, we construct a separate sample choosing only those
176: superclusters which have more than five clusters ($N_{c} > 5$ where $N_{c}$
177: is the number of clusters within the superclusters).
178: It is found that $217$ superclusters have more than five clusters
179: and total $1492$ clusters belong to those $217$ superclusters.
180:
181: First, we measure the probability distribution of the cosines of the angles,
182: $\cos\theta$, between the major axes of the superclusters and their member
183: clusters. Figure \ref{fig:dis} plots the result as solid dots with Poisson
184: errors. The upper panel corresponds to the case that all the $14,007$
185: superclusters are used, while the lower panel corresponds to the case
186: that only those superclusters with more than five clusters are used.
187: The dotted line in each panel corresponds to the case of no alignment.
188: As can be seen, the distribution, $p(\cos\theta)$, increases with
189: $\cos\theta$ in the both panels, revealing a clear signal of alignment
190: effect. Although the result of the lower panel shows less sharp increase,
191: suffering from large errors, the signal is robust at the $99\%$ confidence
192: level. It indicates that the cluster-supercluster alignment effect
193: is not a false signal originated from the inaccurate derivation of
194: the supercluster major axes but a real one. The mean value of $\cos\theta$
195: is found to be $0.54$ and $0.52$ in the upper and the lower panels,
196: respectively.
197:
198: Now that a robust signal of cluster-supercluster alignment effect is found,
199: we examine how the degree of the alignment depends on the cluster properties.
200: First, we examine how the average of $\cos\theta$ depends on the cluster
201: mass, $M_{c}$. Figure \ref{fig:mas} plots the result versus the rescaled
202: cluster mass, $\tilde{M}\equiv M_{c}/M_{s}$, as solid dots with errors
203: which are calculated as one standard deviation of $\cos\theta$ for the case
204: of no alignment. As can be seen in the upper panel, the degree of alignment
205: increases with $\tilde{M}$. A similar trend is also shown in the lower panel
206: although it suffers from the large errors.
207:
208: Second, we examine how the average of $\cos\theta$ depends on the separation
209: distance, $r$, from the supercluster center to the cluster center.
210: Figure \ref{fig:sep} plots the result versus the rescaled distance,
211: $\tilde{r}\equiv r/R_{s}$ as solid dots with errors.
212: As can be seen in the upper panel, the degree of the alignment decreases
213: with distance. That is, the closer a cluster is located to the supercluster
214: center,the more stronger the alignment effect is.
215:
216: Third, we examine how the average of $\cos\theta$ depends on the
217: cluster ellipticity, $\epsilon$. Here, we define the ellipticity of
218: a cluster as $\epsilon \equiv 1 - \sqrt{\varrho^{c}_{3}/\varrho^{c}_{1}}$
219: assuming a prolate cluster shape, where $\varrho^{c}_{1}$ and
220: $\varrho^{c}_{3}$ are the largest and the smallest eigenvalues of the
221: cluster inertia momentum tensor, respectively. Figure \ref{fig:ell}
222: plots the result versus the rescaled ellipticity, $\tilde{\epsilon}\equiv
223: \epsilon/\epsilon_{0}$ (where $\epsilon_{0}$ is
224: the maximum cluster ellipticity) as solid dots with errors, which reveals
225: that the degree of alignment increases with cluster ellipticity.
226: That is, the more elongated a cluster is, the more stronger the
227: alignment effect is.
228:
229: Fourth, we measure the average of $\cos\theta$ as a function of the
230: cluster velocity, $v$. Figure \ref{fig:vel} plots the result with solid
231: dots with errors, which reveals that the degree of alignment depends
232: very weakly on the cluster velocity.
233:
234: We provide physical explanations for these numerical results in $\S 3$.
235:
236: \section{PHYSICAL ANALYSIS}
237:
238: \subsection{\it Hypotheses}
239: To construct an analytic model for the cluster-supercluster
240: alignment effect, we assume the following.
241: \begin{enumerate}
242: \item
243: A supercluster forms through anisotropic merging of clusters along
244: filaments. In consequence, the major axis of a supercluster tends to
245: be in the direction of the dominant filament. A filament is defined as
246: one dimensional object collapsed along the major and intermediate
247: principal axes of the local tidal tensor \citep{zel70,pog-etal98}.
248: The direction of a filament thereby is aligned with the minor principal
249: axis of the tidal tensor. Therefore, the major axis of a supercluster
250: tends to be in the direction of the minor principal axis of the tidal tensor.
251: \item
252: Let ${\bf T}^{s}$ be the tidal tensor field smoothed on the supercluster mass
253: scale, and let also $\delta_{s}\equiv\Delta\rho/\bar{\rho}$ be the linear
254: density contrast of the supercluster where $\bar{\rho}$ is the mean mass
255: density. Let also $\lambda_1, \lambda_2,\lambda_3$ (with
256: $\lambda_1>\lambda_2 >\lambda_3$) be the three eigenvalues of ${\bf T}^{s}$.
257: The collapse condition for a supercluster is given as
258: \begin{equation}
259: \label{eqn:sup}
260: \delta_{s} = \lambda_{1} + \lambda_{2} + \lambda_{3} =1.3,\qquad
261: \lambda_1 > \lambda_2 > 0, \quad \lambda_3 < 0.
262: \end{equation}
263: Given that the supercluster passes the moment of turn-around but not
264: yet virialized, we expect its linear density contrast, $\delta_{s}$,
265: to be in the range $(1,1.68)$ where the values of $1$ and $1.68$
266: correspond to the linear densities at the moments of the turn-around
267: and the virialization, respectively \citep{eke-etal96}. Here, we
268: choose a fiducial value of $\delta_{s}=1.3$. The other condition,
269: $\lambda_1> \lambda_2 > 0, \lambda_3 < 0$ in equation (\ref{eqn:sup})
270: represents the collapse along filaments \citep{pog-etal98}.
271: \item
272: The cluster-supercluster alignment is a reflection of the anisotropic
273: spatial distribution of cluster galaxies in a filament-dominant
274: web-like cosmic structure. The correlation of the spatial positions of
275: galaxies with the local tidal field can be quantified by the following
276: quadratic equation which was first suggested by \citet{lee-kan06}.
277: \begin{equation}
278: \label{eqn:corr}
279: \langle {x}^{c}_{i}{x}^{c}_{j}|\hat{\bf T}^{s}\rangle =
280: \frac{1 - s}{3}\delta_{ij} + s\hat{T}^{s}_{ik}\hat{T}^{s}_{kj},
281: \end{equation}
282: where ${\bf x}^{c}\equiv (x^{c}_{i})$ and $\hat{\bf T}^{s}=
283: (\hat{T}^{s}_{kj})\equiv T^{s}_{ik}/\vert{\bf T}^{s}\vert$ are the
284: rescaled major axis of a galaxy cluster and the unit tidal shear
285: tensor smoothed on the supercluster mass scale, $M_{s}$. Here, the
286: parameter, $s \in [-1,1]$, represents the strength of the correlation
287: between ${\bf x}^{c}$ and ${\bf T}^{s}$. If $s=-1$, there is the
288: strongest correlation between ${\bf x}^{c}$ and $\hat{\bf T}^{s}$. If
289: $s=1$, there is the strongest anti-correlation between ${\bf x}^{c}$
290: and $\hat{\bf T}^{s}$. While if $s=0$, there is no correlation between them.
291: \item
292: The conditional probability distribution of ${\bf x}^{c}$ provided
293: that the local tidal field is given as ${\bf T}^{s}$ can be
294: approximated as Gaussian \citep{lee-kan06}:
295: \begin{equation}
296: \label{eqn:xdis}
297: P({\bf x}^{c}|\hat{\bf T}^{s}) = \frac{1}{[(2\pi)^3 {\rm det}(M)]^{1/2}}
298: \exp\left[-\frac{x^{c}_{i}(M^{-1})_{ij}x^{c}_{j}}{2}\right],
299: \end{equation}
300: where the covariance matrix
301: $M_{ij} \equiv \langle x^{c}_{i}x^{c}_{j}|\hat{\bf T}^{s}\rangle$ is
302: related to $\hat{\bf T}^{s}$ by equation (\ref{eqn:corr}).
303: \end{enumerate}
304:
305: It is worth mentioning here the difference of the cluster-supercluster
306: alignment from the galaxy-cluster alignment. For the former case, the
307: primordial tidal field induces the anisotropy in the spatial
308: distribution of galaxies along cosmic filaments, which results in the
309: alignment between the major axes of clusters and their superclusters.
310: For the latter case, the tidal field of a virialized cluster halo
311: induces the angular momentum of the cluster galaxies whose minor axes
312: tend to be aligned with the major axis of its host cluster
313: \citep{lee-etal05}. In other words, the alignments between the major
314: axes of cluster galaxies and their host clusters are related to the
315: generation of the angular momentum while the alignments between the
316: major axes of clusters and their host superclusters related to the
317: filamentary distribution of galaxies.
318:
319: \subsection{\it Analytic Expressions}
320:
321: Using the four hypotheses given in $\S 3.1$, we derive first $p(\cos\theta)$
322: analytically. According to the second hypothesis, it amounts to deriving the
323: probability density distribution of the cosines of the angles between the
324: major axes of clusters and the minor principal axes of the local tidal tensors.
325:
326: Let us express ${\bf x}^{c}$ in terms of the spherical polar coordinates
327: in the principal axis frame of ${\bf T}^{s}$ as ${\bf x}^{c} =
328: (x^{c}\sin\theta\cos\phi,x^{c}\sin\theta\sin\phi,x^{c}\cos\theta)$
329: where $x^{c}\equiv \vert{\bf x}^{c}\vert$ and $\theta$ and $\phi$ are the
330: polar and the azimuthal angles of ${\bf x}^{c}$, respectively. Then,
331: the polar angle, $\theta$, is nothing but the angle between ${\bf x}^{c}$
332: and the minor principal axis of ${\bf T}^{s}$. Now, the probability
333: density distribution of $\cos\theta$ can be derived by integrating
334: equation (\ref{eqn:xdis}) over $x^{c}$ and $\phi$ as $p(\cos\theta) =
335: \int_{0}^{2\pi}\int_{0}^{\infty}P(x^{c},\theta,\phi)x^{c 2}dx^{c}d\phi$,
336: which leads to \citep{lee-kan06}
337: \begin{eqnarray}
338: \label{eqn:theta_dis}
339: p(\cos\theta) &=& \frac{1}{2\pi}\prod_{i=1}^{3}
340: \left(1-s+3s\hat{\lambda}^{2}_{i}\right)^{-\frac{1}{2}}\times \nonumber \\
341: &&\int_{0}^{2\pi}
342: \left(\frac{\sin^{2}\theta\cos^{2}\phi}{1-s+3s\hat{\lambda}^{2}_{1}} +
343: \frac{\sin^{2}\theta\sin^{2}\phi}{1-s+3s\hat{\lambda}^{2}_{2}} +
344: \frac{\cos^{2}\theta}{1-s+3s\hat{\lambda}^{2}_{3}}\right)^{-\frac{3}{2}}
345: d\phi,
346: \end{eqnarray}
347: where $\{\hat{\lambda_i}\}_{i=1}^{3}$ are the unit eigenvalues of
348: $\hat{\bf T}$ related to $\{\lambda_i\}_{i=1}^{3}$ as
349: $\hat{\lambda_i} \equiv \lambda_{i}/
350: \left(\lambda^{2}_{1}+\lambda^{2}_{2}+\lambda^{2}_{3}\right)^{1/2}$.
351:
352: It was in fact \citet{lee-kan06} who first derived equation
353: (\ref{eqn:theta_dis}) as an analytic expression for the probability
354: distribution of the alignments between the positions of the galaxy
355: satellites in the major axis orientations of their host
356: galaxies. Here, we derive it as an analytic expression for the
357: probability distribution of the alignments between the major axes of
358: clusters and their host superclusters.
359: It is important to note a key difference between the two cases. For
360: the case of galaxy satellites, it is the tidal fields of the {\it
361: virialized} galactic halos that causes the alignment
362: effect. Therefore, all the eigenvalues, $\lambda_{1}, \lambda_{2},
363: \lambda_{3}$ in equation (\ref{eqn:theta_dis}) should be positive.
364: Whereas for the case of clusters in superclusters it is the local
365: filaments which collapse along only two principal axes of the
366: primordial local tidal tensors. Therefore, $\lambda_{3}$ in equation
367: (\ref{eqn:theta_dis}) has a negative value.
368:
369: The probability distribution, $p(\cos\theta)$
370: (eq.[\ref{eqn:theta_dis}]) is characterized by three independent
371: parameters, $s$, $\lambda_1$, and $\lambda_2$. Once the values of
372: $\lambda_{1}$ and $\lambda_{2}$ are determined, then the negative value of
373: $\lambda_{3}$ is automatically determined by equation (\ref{eqn:sup}).
374: Since the values of these three parameters depend on the properties of
375: individual superclusters as well as the local conditions of the
376: initial tidal fields, it may be quite difficult to determine them
377: analytically.
378:
379: Instead, we determine their average values by fitting equation
380: (\ref{eqn:theta_dis}) to the numerical results obtained in $\S 2$.
381: When the numerical result using all superclusters are fitted,
382: the best-fit values of the three parameters are found to be $s = -0.71$,
383: $\lambda_{1}=2.23$, and $\lambda_{2}=0.53$, which gives $\lambda_3=-1.46$.
384: When the numerical result using only those superclusters with $N_{c}>5$ are
385: fitted, it is found interestingly that the best-fit values of the parameters
386: are $s=-0.5$, $\lambda_{1}=2.23$, and $\lambda_{2}=0.53$. Note that the
387: two numerical cases yield the same best-fit values for $\lambda_{1}$ and
388: $\lambda_{2}$ although the best-fit values of the correlation parameters
389: are different as $s=-0.71$ and $s=-0.5$.
390:
391: Figure \ref{fig:dis} plots the analytic distributions with these best-fit
392: parameters (solid line) and compares it with the numerical data points.
393: In the upper panel, the analytic distribution with $s=-0.71$ is compared
394: with the numerical result obtained in $\S 2$ using all superclusters,
395: while in the lower panel the analytic distribution with $s=-0.5$ is compared
396: with the numerical result obtained using only those superclusters with more
397: than five clusters $(N_{c} > 5$). As can be seen, the analytic and the
398: numerical results are in good agreement with each other in the both panels.
399:
400: It is worth mentioning here that the best-fit values of the three
401: parameters are subject to our fiducial choice of $\delta_{s}=1.3$. As
402: mentioned in $\S 3.1$, there is no consensus on the critical linear
403: density of the superclusters unlike the case of clusters. Varying the
404: value of $\delta_{s}$ from $1.0$ to $1.68$, we have repeated the
405: fitting procedure, and we found that although the best-fit values of
406: $\lambda_{1}$ and $\lambda_{2}$ change by maximum $20\%$, the fitting
407: result itself does not sensitively change with the value of $\delta_{s}$.
408: Thus, it is concluded that our fiducial model is a stable choice.
409:
410: Now that we have the probability density distribution, $p(\cos\theta)$,
411: we would like to find an analytic expression for $\langle\cos\theta\rangle$
412: as a function of cluster mass, position, ellipticity, and velocity.
413: The dependence of the correlation parameter $s$ on the cluster mass $M$
414: may be obtained by considering the difference in mass between the cluster
415: and its host supercluster. Strictly speaking, equation (\ref{eqn:corr}) is
416: valid when the tidal tensor ${\bf T}^{s}$ and the position vector
417: ${\bf x}^{c}$ are smoothed on the same mass scale. In other words, the
418: correlation between ${\bf x}^{c}$ and ${\bf T}^{s}$ is expected to be highest
419: when the two smoothing mass scales are the same. In reality, however,
420: ${\bf T}^{s}$ is smoothed on the supercluster mass scale $M_{s}$ while
421: ${\bf x}^{c}$ is smoothed on the cluster mass scale $M_{c}$. The difference
422: between the two mass scales diminishes the correlation between ${\bf x}^{c}$
423: and ${\bf T}^{s}$.
424:
425: Let $s_{M_{0}}$ be the value of the correlation parameter when the
426: tidal field is smoothed on the same cluster mass scale, ${\bf T}^{c}$.
427: We expect $s_{M_0}=-1$. Given equation (\ref{eqn:corr}), we approximate
428: $s=s(\tilde{M})$ as
429: \begin{equation}
430: \label{eqn:sm}
431: s(\tilde{M}) \approx s_{M_0}\frac{\langle\hat{T}^{s}_{ik}
432: \hat{T}^{s}_{kj}\rangle}
433: {\langle\hat{T}^{c}_{ik}\hat{T}^{c}_{kj}\rangle}
434: \approx s_{M_0}\frac{\sigma^{2}_s}{\sigma^{2}_c}.
435: \end{equation}
436: Here, $\sigma_c$ and $\sigma_s$ represent the rms linear density fluctuations
437: smoothed on the mass scales of $M_c$ and $M_s$, respectively. In deriving
438: equation(\ref{eqn:sm}) we use the approximation of
439: $\langle\hat{T}^{c}_{ik}\hat{T}^{c}_{kj}\rangle \approx
440: \langle {T}^{c}_{ik}{T}^{c}_{kj}\rangle/\vert{\bf T}^{c}\vert^{2}$,
441: which was proved to be valid by \citet{lee-pen01}.
442:
443: Now that the functional form of $s(\tilde{M})$ is found, the average of
444: $\cos\theta$ as a function of $\tilde{M}$ can be calculated by equations
445: (\ref{eqn:theta_dis}) and (\ref{eqn:sm}) as
446: \begin{equation}
447: \label{eqn:mas}
448: \langle\cos\theta\rangle(\tilde{M})=\int^{\infty}_{0}\cos\theta
449: p[\cos\theta;s(\tilde{M})]d\!\cos\theta.
450: \end{equation}
451: Figure \ref{fig:mas} plots equation (\ref{eqn:mas}) with $\lambda_{1}=2.23$
452: and $\lambda_{2}=0.53$ (solid line), and compares it with the numerical
453: result (dots) obtained in $\S 2$. For the analytic distribution, the value
454: of $M_{s}$ is set to be the mean mass of the superclusters found in $\S 2$:
455: $1.26 \times 10^{15}h^{-1}M_{\odot}$ (upper panel);
456: $3.69 \times 10^{15}h^{-1}M_{\odot}$ (lower panel).
457: As can be seen, in the upper panel the analytic and the numerical results
458: agree with each other excellently. In the lower panel, although the numerical
459: result suffers from large errors, the analytic prediction is still quite
460: consistent with the numerical result.
461:
462: The dependence of the correlation parameter, $s$, on the distance, $r$,
463: between the centers of clusters and their host superclusters can be
464: obtained in a similar way. The correlation between ${\bf x}^{c}$ and
465: ${\bf T}^{s}$ in equation (\ref{eqn:corr}) becomes strongest when
466: $r=0$. In reality, however, $r$ always deviates from zero, which will
467: diminish the correlation strength.
468:
469: Let $s_{r0}$ be the value of the correlation parameter when $r=0$,
470: which is expected again to be $s_{r0} = -1$. With a similar
471: approximation made for equation (\ref{eqn:sm}), we find the following formula
472: for $s(r)$:
473: \begin{equation}
474: \label{eqn:sr}
475: s(r) \approx \frac{1}{2}s_{r0}\left[1 +
476: \frac{\langle\hat{T}^{c}_{ij}({\bf x}+{\bf r}){\hat T}^{c}_{ij}
477: ({\bf x})\rangle}
478: {\langle\hat{T}^{c}_{ij}({\bf x})\hat{T}^{c}_{ij}({\bf x})\rangle}\right]
479: \approx \frac{1}{2}s_{r0}\left[1 + \tilde{\xi}_{c}(r)\right],
480: \end{equation}
481: where $\tilde{\xi}_{c}(r)$ represents the two point density correlation
482: rescaled to satisfy $\tilde{\xi}_{c}(0)=1$. Since the distance $r$ is
483: a Eulerian quantity unlike the mass $M_{s}$, we use the Eulerian
484: filtering radius of $2h^{-1}$Mpc, the typical cluster size, to
485: convolve the correlation function $\tilde{\xi}_{c}$. Here, the factor
486: of $1/2$ comes from the average decreases of the correlation parameter
487: due to the mass difference between the clusters and their host
488: superclusters.
489:
490: Now that the functional form $s(r)$ is found, the average of
491: $\cos\theta$ as a function of $r$ can be calculated through
492: equations (\ref{eqn:theta_dis}) and (\ref{eqn:sr}) as
493: \begin{equation}
494: \label{eqn:sep}
495: \langle\cos\theta\rangle(r)=\int^{\infty}_{0}\cos\theta
496: p[\cos\theta;s(r)]d\!\cos\theta.
497: \end{equation}
498: Figure \ref{fig:sep} plots equation (\ref{eqn:sep}) (solid line)
499: as a function of the rescaled distance, $\tilde{r}\equiv r/R_{s}$, and
500: compares it with the numerical result (dots) obtained in $\S 2$.
501: For the analytic distribution, the value of $R_{s}$ is set to be the mean
502: Lagrangian radius of superclusters found in $\S 2$ using the relation
503: of $R_{s} = [3\bar{M}_{s}/(4\pi\bar{\rho})]^{1/3}$. As can be seen,
504: the analytic and the numerical results agree with each other
505: quite well.
506:
507: Regarding the dependence of $s$ on the cluster ellipticity,
508: $\epsilon$, although it is predicted qualitatively in our theoretical
509: model that the degree of the alignment increases with ellipticity, the
510: quantitative functional form of $s(\epsilon)$ is quite difficult to
511: determine analytically since the cluster ellipticity are sensitively
512: vulnerable to modifications caused by nonlinear merging and infall process.
513:
514: Instead of using analytic approach, numerical fitting is used to
515: determine the functional form of $s(\epsilon)$. Let $s_{\epsilon 0}$
516: represents the value of $s$ when the cluster ellipticity has the maximum
517: value, $\epsilon 0$. It is expected again that $s_{\epsilon 0}=-1$.
518: We find that the following formula gives a good fit to the numerical results
519: \begin{equation}
520: \label{eqn:se}
521: s(\tilde{\epsilon}) = s_{\epsilon 0}\tilde{\epsilon}^{1/2}.
522: \end{equation}
523: Since $\epsilon_{0}$ is defined as the maximum ellipticity,
524: $s(\tilde{\epsilon})$ has the extreme value of $-1$ at $\tilde{\epsilon}=1$.
525: Note that the value of $\epsilon_{0}$ is not fixed but sample-dependent.
526: Here, the Millennium Run data we use yields $\epsilon_{0}=0.7$.
527: But, a different sample could yield a different value of $\epsilon_{0}$.
528:
529: Now that the functional form of $s(\tilde{\epsilon})$ is found, the average
530: of $\cos\theta$ as a function of $\tilde{\epsilon}$ can be calculated
531: by equations (\ref{eqn:theta_dis}) and (\ref{eqn:se}) as
532: \begin{equation}
533: \label{eqn:ell}
534: \langle\cos\theta\rangle(\tilde{\epsilon})=\int^{\infty}_{0}
535: \cos\theta p[\cos\theta;s(\tilde{\epsilon})]d\!\cos\theta.
536: \end{equation}
537: The comparison between the analytic result (eq.[\ref{eqn:ell}]) and the
538: numerical data points shows good consistency, as seen in Fig. \ref{fig:ell}.
539:
540: Regarding the dependence of $s$ on the cluster velocity $v$, no
541: strong dependence is expected in our model since the primordial tidal
542: field is uncorrelated with the velocity field \citep{bar-etal86}.
543: Therefore, we model it as an uniform distribution as
544: $\langle\cos\theta\rangle(v) =\langle\cos\theta\rangle$. The average
545: value, $\langle\cos\theta\rangle$ is found to be $0.54$ when $s=-0.71$
546: (upper panel) while it is $0.52$ when $s=-0.5$ (lower panel).
547: As can be seen in Fig. \ref{fig:vel}, the analytic and the numerical
548: results are consistent with each other.
549:
550: \section{DISCUSSION AND CONCLUSION}
551:
552: In the context of the standard cosmic web picture of large-scale
553: structure, we have constructed a parametric model for the alignment of
554: cluster-sized halos with their host superclusters.
555: The underlying assumption is that cluster-supercluster alignment reflects
556: the spatial distribution of matter as it is organized along
557: filamentary structures by the primordial tidal field.
558: The parameters of the analytic model represent the dominance of
559: filaments and the spatial coherence of the initial tidal field.
560:
561: We show that the analytic model provides a good fit to orientation data
562: derived from mass-limited halo samples of a $\Lambda$CDM Hubble volume
563: simulation. After fitting the three free parameters using the overall
564: distribution of cluster-supercluster alignment angles, the model then
565: simultaneously matches the behavior of the mean alignments as a
566: function of relative mass, cluster position within the supercluster,
567: and cluster ellipticity. No trend with cluster velocity is predicted
568: or measured in the simulation.
569:
570: It is worth discussing a couple of simplified assumptions on which our
571: theoretical model is based. First,we have used the FOF algorithm to
572: identify superclusters in the N-body simulation data. Unlike the case
573: of virialized clusters, however, there is no established consensus on
574: how to define superclusters. Different supercluster-identification
575: algorithms could result in different multiplicity, mass, and shape of
576: superclusters which would in turn affect our results.
577:
578: Second, we have assumed that the filaments correspond to the
579: Lagrangian regions where only the largest and the second largest
580: eigenvalues of the local tidal tensor are positive. Although this
581: definition of a filament is consistent with the picture of the
582: Zel'dovich approximation, it is obviously an oversimplification of the
583: reality. A more realistic definition and treatment of cosmic filaments
584: will be necessary to refine the model.
585:
586: Another issue that we would like to discuss here is the possibility of
587: using the cluster-supercluster alignment effect as a cosmological
588: probe. We have shown that the phenomena of cluster-supercluster
589: alignments are closely related to the dominance of filaments, the
590: web-like distribution of galaxies on very large scales.
591: The dominance of filaments is in turn related to the spatial
592: correlations of the primordial tidal field, which depends sensitively on
593: the slope of the initial power spectrum on the supercluster
594: scale. Thus, by measuring the degree of cluster-supercluster alignment,
595: it might be possible to constrain the slope of the initial power
596: spectrum in a complementary way.
597:
598: Finally, we conclude that our model for the cluster-supercluster
599: alignments will provide a theoretical framework within which the
600: distribution of cosmic structures on the largest scales can be
601: physically understood and quantitatively described.
602:
603: \acknowledgments
604: We are grateful to the anonymous referee who helped us improve the original
605: manuscript. We are also grateful to the warm hospitality of Y.Suto and the
606: University of Tokyo, where this work was initiated. J.L. also thanks D. Park
607: for useful helps. J.L. is supported by the research grant No.
608: R01-2005-000-10610-0 from the Basic Research Program of the Korea Science and
609: Engineering Foundation. A.E.E. acknowledges support from the Miller Institute
610: for Basic Research in Science at the University of California, Berkeley,
611: from NSF ITR ACI-0121671 and from NASA ATP NAG5-13378.
612:
613: \begin{thebibliography}{}
614: \bibitem[Agustsson \& Brainerd(2006)]{agu-bra06}
615: Agustsson, I., \& Brainerd, T. G. 2006, \apj, 644, L25
616: \bibitem[Atlay et al.(2006)]{atl-etal06}
617: Atlay, G., Colberg, J. M., Croft, R. A. C. 2006, \mnras in press,
618: preprint [astro-ph/0605296]
619: \bibitem[Bardeen et al.(1986)]{bar-etal86}
620: Bardeen, J.M., Bond, J. R., Kaiser, N., Szalay, A.S. 1986, apj, 304, 15
621: \bibitem[Basilakos(2003)]{bas03}
622: Basilakos, S. 2006, \mnras, 344, 602
623: \bibitem[Basilakos et al.(2006)]{bas-etal06}
624: Basilakos, S., Plionis, M. et al. 2006, \mnras, 365, 539
625: \bibitem[Bond, Kofman, \& Pogosyan(1996)]{bon-etal96}
626: Bond, J., R., Kofman, L., \& Pogosyan, D. 1996, Nature, 380, 603
627: \bibitem[Brainerd(2005)]{bra05}
628: Brainerd, T. G. 2005, \apj, 628, L101
629: \bibitem[Busha et al.(2005)]{busha05}
630: Busha, M.T., Evrard, A.E., Adams, F.C.\& Wechsler, R.H. 2005, \mnras,
631: 363, 11
632: \bibitem[Dubinski(1998)]{dub98}
633: Dubinski, J. 1998, \apj, 502, 141
634: \bibitem[Eke et al.(1996)]{eke-etal96}
635: Eke, V., Cole, S., \& Frenk, C. 1996, \mnras, 283, 263
636: \bibitem[Evrard et al.(2002)]{evr-etal02}
637: Evrard, A. E, et al. 2002, \apj, 573, 7
638: \bibitem[Faltenbacher et al.(2002)]{fal-etal02}
639: Faltenbacher, A., Kerscher, A., Gottloeber, S., \& Mueller, M. 2002,
640: \aap, 395, 1
641: \bibitem[Fuller, West \& Bridges(1999)]{ful-etal99}
642: Fuller, T. M., West, M. J., \& Bridges, T. J. 1999, \apj, 519, 22,
643: \bibitem[Kalinkov et al.(1998)]{kal-etal98}
644: Kalinkov, M., Valtchanov, I., \& Kuneva, I. 1998, \apj, 506, 509
645: \bibitem[Kang et al.(2005)]{kan-etal05}
646: Kang, X., Mao, S., Gao, L., \& Jing, Y. P. 2005, \aap, 437, 383
647: \bibitem[Kasun \& Evrard(2005)]{kas-evr05}
648: Kasun, S. F., \& Evrard, A. E. 2005, \apj, 629, 781
649: \bibitem[Knebe et al.(2004)]{kne-etal04}
650: Knebe, A. et al. 2004, \apj, 603, 7
651: \bibitem[Kroupa et al.(2005)]{kro-etal05}
652: Kroupa, P., Theis, C., \& Boily, C. M. 2005, \aap, 431, 517
653: \bibitem[Lee \& Pen(2000)]{lee-pen00}
654: Lee, J. \& Pen, U. L. 2000, \apj, 532, L5
655: \bibitem[Lee \& Pen(2001)]{lee-pen01}
656: Lee, J. \& Pen, U. L. 2001, \apj, 555, 106
657: \bibitem[Lee et al.(2005)]{lee-etal05}
658: Lee, J., Kang, X., \& Jing, Y. 2005, \apj, 629, L5
659: \bibitem[Lee \& Kang(2006)]{lee-kan06}
660: Lee, J. \& Kang, X. 2006, \apj, 637, 561
661: \bibitem[Libeskind et al.(2005)]{lib-etal05}
662: Libeskind, N. I., Frenk, C. S., Cole, S., Helly, J. C., Jenkins, A.,
663: Navarro, J. F., \& Power, C. 2005, preprint (astro-ph/0503400)
664: \bibitem[MacGillivray et al.(1982)]{mac-etal82}
665: MacGillivray, H. T., Dodd, R. J., McNally, B. V., \& Corwin, H. G. 1982,
666: \mnras, 198, 605
667: \bibitem[Macfarland et al.(1998)]{mac-etal98}
668: Macfarland, T., Couchman, H. M. P., Pearce, F. R., \& Pichlmeier,
669: J. 1998, NewA, 3, 687
670: \bibitem[Nagamine \& Loeb(2003)]{nag-loeb03}
671: Nagamine, K., \& Loeb, A. 2003, NewA, 8, 439
672: \bibitem[Onuora \& Thomas(2000)]{onu-tho00}
673: Onuora, L. I., \& Thomas, P. A. 2000, \mnras, 319, 614
674: \bibitem[Pereira \& Kuhn(2005)]{per-kuh05}
675: Pereira, M. J., \& Kuhn, J. R. 2005, 627, L21
676: \bibitem[Plionis(1994)]{pli94}
677: Plionis, M. 1994, \apjs, 95, 401
678: \bibitem[Plionis(2002)]{pli02}
679: Plionis, M. 2002, in Modern Theoretical and Observational Cosmology,
680: (Kluwer Academic Publishers: Dordrecht), p.299
681: \bibitem[Plionis \& Basilakos(2002)]{pli-bas02}
682: Plionis, M., \& Basilakos, S. 2002, \mnras, 329, L47
683: \bibitem[Plionis et al.(2003)]{pli-etal03}
684: Plionis, M., Benoist, C., et al. 2003, \apj, 594, 153
685: \bibitem[Plionis(2004)]{pli04}
686: Plionis, M. 2004, Ourskirts of Galaxy Clusters, (Cambridge
687: Univ. Press: Cambridge), p. 19
688: (Kluwer Academic Publishers: Dordrecht), p.299
689: \bibitem[Pogosyan et al.(1998)]{pog-etal98}
690: Pogosyan, D., Bond, J. R., Kofman, L., \& Wadsley, J. 1998, Wide Field
691: Surveys in Cosmology, (Frontiers: Paris), p. 61
692: \bibitem[Shapley(1930)]{sha30}
693: Shapley, H. 1930, Harvard Obs. Bull., 874, 9
694: \bibitem[Springel et al.(2001)]{spr-etal01}
695: Springel, V., White, S. D. M., Tormen, G., Kauffmann, G. 2001, \mnras, 328, 726
696: \bibitem[Valtonen et al.(1978)]{val-etal78}
697: Valtonen, M., Teerikorpi, P., \& Argue, A. 1978, \aj, 83, 135
698: \bibitem[van Haarlem \& van de Weygaert(1993)]{van-van93}
699: van Haarlen, M., \& van de Weygaert, R. 1993, \apj, 418, 544
700: \bibitem[West(1989)]{wes89}
701: West, M. J. 1989, \apj, 347, 610
702: \bibitem[West(1991)]{wes-etal91}
703: West, M. J., Willumsen, C., \& Dekel, A. 1991, \apj, 369, 287
704: \bibitem[Zeldovich(1970)]{zel70}
705: Zel'dovich, Y. B. 1970, A\& A, 5, 84
706: \bibitem[Zentner et al.(2005)]{zen-etal05}
707: Zentner, A. R., Kravtsov, A. V., Gnedin, O. Y., \& Klypin, A. A. 2005,
708: 629, 219
709: \end{thebibliography}
710: \clearpage
711: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
712: \begin{figure}
713: \begin{center}
714: \plotone{f1.eps}
715: \caption{The spatial distribution of the third richest supercluster is
716: shown in orthogonal projections. Circles show halo locations, with
717: symbol size scaling as $M^{1/3}$, while lines through each halo show the
718: orientation of the major axis of its density field,
719: taken from \citep{kas-evr05}. The length of each line is
720: proportional to the halo's major-to-minor axis ratio.
721: \label{fig:sc3}}
722: \end{center}
723: \end{figure}
724: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
725:
726: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
727: \begin{figure}
728: \begin{center}
729: \plotone{f2.eps}
730: \caption{Probability density distributions of the cosines of the
731: angles between the major axes of clusters and their superclusters:
732: ({\it Upper}): the case that all $14007$ superclusters are used;
733: ({\it Lower}): the case that only $217$ superclusters with more than
734: five clusters are used. In each panel, the numerical result is represented
735: by dots with Poissonian errors while the analytic result (\ref{eqn:theta_dis})
736: corresponds to the solid curve. The dotted line corresponds to the case of
737: no alignment.
738: \label{fig:dis}}
739: \end{center}
740: \end{figure}
741: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
742:
743: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
744: \begin{figure}
745: \begin{center}
746: \plotone{f3.eps}
747: \caption{Average of the cosines of the angles as a function of the
748: cluster mass: ({\it Upper}): the case that all $14007$ superclusters
749: are used; ({\it Lower}): the case that only $217$ superclusters with
750: more than five clusters are used. In each panel, the dots and solid curves
751: represent the numerical and the analytic results, respectively. The errors
752: are calculated as one standard deviation of the cosines of the angles for
753: the case of no alignment. The dotted line corresponds to the case of no
754: alignment.
755: \label{fig:mas}}
756: \end{center}
757: \end{figure}
758: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
759:
760: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
761: \begin{figure}
762: \begin{center}
763: \plotone{f4.eps}
764: \caption{Average of the cosines of the angles as a function of the
765: distance from the supercluster center to the cluster center:
766: ({\it Upper}): the case that all $14007$ superclusters are used;
767: ({\it Lower}): the case that only $217$ superclusters with more than
768: five clusters are used. In each panel the dots and solid curves represent
769: the numerical and the analytic results, respectively. The dotted line
770: corresponds to the case of no alignment.
771: \label{fig:sep}}
772: \end{center}
773: \end{figure}
774: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
775:
776: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
777: \begin{figure}
778: \begin{center}
779: \plotone{f5.eps}
780: \caption{Average of the cosines of the angles as a function of the
781: cluster ellipticity: ({\it Upper}): the case that all $14007$
782: superclusters are used; ({\it Lower}): the case that only $217$ superclusters
783: with more than five clusters are used. In each panel, the dots and solid
784: curves represent the numerical and the analytic results, respectively.
785: The dotted line corresponds to the case of no alignment.
786: \label{fig:ell}}
787: \end{center}
788: \end{figure}
789: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
790:
791: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
792: \begin{figure}
793: \begin{center}
794: \plotone{f6.eps}
795: \caption{Average of the cosines of the angles as a function of the
796: cluster velocity: ({\it Upper}): the case that all $14007$ superclusters
797: are used; ({\it Lower}): the case that only $217$ superclusters with more than
798: five clusters are used. In each panel, the dots and solid curves represent the
799: numerical and the analytic results, respectively. The dotted line corresponds
800: to the case of no alignment.
801: \label{fig:vel}}
802: \end{center}
803: \end{figure}
804: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
805:
806: \end{document}
807: