astro-ph0701889/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: \begin{document}
3: \title{Void-Supercluster Alignments}
4: \author{Daeseong Park and Jounghun Lee}
5: \affil{Department of Physics and Astronomy, FPRD, Seoul National University, 
6: Seoul 151-747, Korea} 
7: \email{pds2001@astro.snu.ac.kr, jounghun@astro.snu.ac.kr}
8: \begin{abstract}
9: We investigate alignments between the spin axes of cosmic voids and the 
10: principal axes of nearby superclusters using the Millennium Run simulation 
11: of a $\Lambda$CDM cosmology. The concept of void spin was first introduced 
12: by Lee and Park in 2006 to quantify the tidal effect on voids from the 
13: surrounding matter distribution. Our numerical analysis reveals that the 
14: void spin axes are strongly aligned with the supercluster minor axes, but 
15: anti-aligned with the major axes, and have no correlations with the 
16: intermediate axes. We provide physical explanations to this numerical results 
17: on the basis of tidally induced correlations.  It is expected that our 
18: work will provide a new insight into the characterization of the cosmic web 
19: on the largest scale.
20: \end{abstract}
21: \keywords{cosmology:theory --- large-scale structure of universe}
22: 
23: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
24: 
25: \section{INTRODUCTION}
26: 
27: Voids stand for those regions in the universe which have extremely low 
28: density. There is still no consensus on how to define voids. In recent 
29: numerical approaches using N-body simulations, voids are often defined as 
30: regions empty of massive dark matter halos. In real observations, 
31: however, voids are usually defined as underdense regions with the mean density 
32: contrast close to $-0.9$ \citep[e.g.,][]{hoy-vog04}, given the limitation that we cannot 
33: observe very faint galaxies.  Adopting the latter definition of voids, one can say 
34: that voids are not empty of massive dark halos but have some structures in them. 
35: An interesting question is how the void structures evolve with time. 
36: 
37: The standard picture based on the cold dark matter paradigm explains that 
38: the voids originated from the local minima of the initial density fluctuation 
39: and have become more and more underdense via gravitational rarefaction in the 
40: subsequent evolution \citep{ick84}. Recent high-resolution simulations, 
41: however, have revealed that the evolution of the void structures is more 
42: complicated than naively predicted \citep[e.g.,][]{got-etal03,col-etal05}: 
43: Not all voids remain underdense but some fraction of them actually collapse 
44: into bound halos. Although the collapse of voids may be described 
45: qualitatively as the occurrence of the clouds-in-voids in the frame of the 
46: standard excursion set theory \citep{she-van04}, it has yet to be well 
47: understood under what conditions and how frequently the collapse of voids 
48: takes place.
49: 
50: A first suspicion goes to the tidal forces from the surrounding matter 
51: distribution \citep{sah-etal94}. As noted by \citet{sha-etal04, sha-etal06}, 
52: the voids should be severely disturbed by the tidal influence due to the 
53: low density. Considering this possibility, a plausible scenario is that when 
54: the tidal effect on a void wins over the gravitational rarefaction, the void 
55: will become squeezed out to collapse into a bound halo. A remaining question 
56: is how to quantify the tidal effect on voids and predict its consequences.
57: 
58: Very recently, \citet[][hereafter, LP06]{lee-par06} introduced a new concept, 
59: {\it the void spin angular momentum}, which is defined as 
60: \begin{equation}
61: \label{eqn:voidspin}
62: {\bf J} = \frac{1}{M_{V}}\sum_{\alpha}^{N_{V}}m_{\alpha}{\bf r}_{\alpha}
63: \times{\bf v}_{\alpha},
64: \end{equation}
65: where $N_{v}$ is the total number of halos belonging to a void, 
66: $m_{\alpha}$, ${\bf r}_{\alpha}$, and ${\bf v}_{\alpha}$ are the mass, 
67: the position, and the velocity of the $\alpha$-th halo belonging to a void 
68: and $M_{V}\equiv \sum_{\alpha}^{N_{V}}m_{\alpha}$. Here the positions and 
69: the velocities of the void halos are measured relative to the center 
70: of mass of the void halos. 
71: 
72: As mentioned by LP06, although the quantity defined in equation 
73: (\ref{eqn:voidspin}) is not a real spin angular momentum since voids  
74: are not bound system, the concept of void spin turned out to be a 
75: very effective measure of the tidal effect from the surrounding matter.
76: To show that the void spin angular momentum is tidally induced, LP06 
77: measured the spatial correlations between the spin axes of neighboring 
78: voids analyzing numerical data from high-resolution simulation, and  
79: found that the numerical results are in excellent agreement with the 
80: analytic predictions based on tidally generated correlations.
81: 
82: If the void spin angular momentum is really induced by the tidal 
83: effect from the surrounding matter, then there should exist correlations 
84: not only between the spin axes of neighboring voids but also between the 
85: spin axes of voids and the neighboring matter distribution. 
86: This cross correlation, if found to exist, should be a more direct 
87: indication of the tidal effect on voids from the surrounding matter.
88: 
89: Our goal here is to measure the cross-correlations between the void-spin 
90: axes and the axes of the neighboring superclusters using numerical data 
91: from high-resolution simulations and to explain it physically.
92: The superclusters are considered as the counterparts of voids since 
93: they are the largest bound structures, comparable to voids in size. 
94: 
95: The outline of this paper is summarized as follows.  In \S 2, the numerical 
96: data are analyzed and the signals of void-supercluster alignments are 
97: detected.  In \S 3, analytic model for the void-supercluster alignments 
98: is derived and compared with the numerical results. 
99: In \S 4, the implications and caveats of our work are discussed, 
100: and a final conclusion is drawn.
101: 
102: \section{NUMERICAL RESULTS}
103: 
104: \subsection{Measurements of Void Spins}
105: 
106: We analyzed the halo catalog extracted from the Millennium Run simulation 
107: \citep{spr-etal05}, which is now available at 
108: http://www.mpa-garching.mpg.de/millennium. 
109: The simulation is of a flat $\Lambda$CDM universe for which the values of 
110: the key parameters are given as $(\Omega_{m},\sigma_{8},h,n_{s}) = 
111: (0.25,0.9,0.73,1)$ where $\Omega_{m}$, $\sigma_{8}$, $h$ and $n_{s}$ represent 
112: the matter density parameter, the rms density fluctuation on the top-hat scale 
113: of $8h^{-1}$Mpc, the dimensionless Hubble parameter and the slope of the 
114: primordial power spectrum. 
115: 
116: For our analysis, we exclude those halos from the Millennium Run catalog 
117: which have too low particle numbers, as those low-particle number halos 
118: are just poorly sampled, heavily affected by noise (V. Springel in private 
119: communication). Technically, we set the cut-off particle number at $50$, 
120: a minimum particle number for defining a halo density profile. 
121: 
122: In accordance with the void-finder algorithm proposed by 
123: \citet[][hereafter, HV02]{hoy-vog02} to the Millennium Run halo catalog, 
124: we take the following steps to identify voids: First, we classify the halos 
125: according to the wall/field criterion, given as $l=\bar{d}_{3}+3\sigma/2$, 
126: where $\bar{d}_{3}$ is the average distance to the third nearest neighbor 
127: and $\sigma$ is its standard deviation. For the Millennium Run halos, the 
128: criterion distance is found to be $l=2.78h^{-1}$Mpc. Note that this value is 
129: slightly higher than the value $l=2.44h^{-1}$Mpc used in our previous work 
130: \citep{lee-par06}. It is because in our previous work the voids were 
131: identified not from the halo catalog but from the galaxy catalog.  
132: Second, we place the wall halos on cubic grid cells each of which has a 
133: linear size of $l$, counting the number of halos in each cell. Third, 
134: increasing the radii of empty spheres from the center of all empty grid 
135: cells till each of them include three wall halos on the surface, we 
136: determine the largest possible empty spheres. 
137: 
138: Finally, we detect voids by categorizing the overlapping empty spheres whose 
139: radii are greater than the minimum size threshold. The minimum void size are 
140: set at $6h^{-1}$Mpc which was found through statistical significance test in 
141: our previous work \citep{lee-par06}.  A total $20291$ voids are identified 
142: in the z=0 catalog, among which only $6430$ voids are found to have more than 
143: $30$ halos. The mean density contrast ($\bar{\delta}_{v}$) and the mean 
144: effective Eulerian radius ($\bar{R}_{E}$) of these $6430$ 
145: voids are found to be $\bar{\delta}_{v}=-0.9$ and 
146: $\bar{R}_{E}=12.22h^{-1}$Mpc, respectively. 
147: 
148: Considering only these large voids with more than $30$ halos, we calculate 
149: the spin angular momentum by equation (\ref{eqn:voidspin}). It is worth 
150: mentioning here that the direction of the void spin angular momentum might 
151: depend on the cut-off particle number for the halo selection. In our analysis, 
152: we include only those halos in the Millennium Run halo catalog which contain 
153: more than $50$ particles. One may suspect that a different cut-off particle 
154: number might yield different directions of the void spins.  
155: 
156: To examine how the void spin angular momentum depends on the cut-off 
157: particle number, we calculate a second angular momentum ${\bf J}^{\prime}$ 
158: for each of the 6430 voids, taking all halos inside the void with more than 
159: $30$ particles, and calculate the cosines of the angles, $\psi$, between 
160: ${\bf J}$ and ${\bf J}^{\prime}$ as
161: \begin{equation}
162: \cos\psi \equiv \frac{\vert {\bf J}\cdot{\bf J}^{\prime}\vert}
163: {\vert{\bf J}\vert\vert{\bf J}^{\prime}\vert}.
164: \end{equation}
165: Figure \ref{fig:psi} plots the probability distribution of $\cos\psi$ as 
166: histogram with the Poisson errors. The horizontal line corresponds to the 
167: case of no correlation. As can be seen, the two angular momentum vectors are 
168: very strongly aligned. It indicates that excluding those halos with lower 
169: particle number will not affect the void spin angular momentum significantly.
170: 
171: \subsection{Measurements of Supercluster Principal Axes}
172: 
173: A next task is to identify superclusters from the Millennium Run halo 
174: catalog. Following the common method \citep[e.g.,][]{wra-etal06}, 
175: we define a supercluster as a cluster of clusters, and identify superclusters 
176: with the help of the friends-of-friends algorithm (FOF). 
177: The cluster halos are selected as those halos whose mass exceeds a typical 
178: poor cluster mass, $M_{c}$. As in \citet{wra-etal06}, we set the value of 
179: $M_{c}$ at $1.75\times 10^{13}h^{-1}M_{\odot}$. 
180: 
181: Before applying the FOF algorithm to the Millennium Run halo catalog, one has 
182: to determine a linking length for the supercluster identification. To find an 
183: optimal linking length, $L$, we perform a statistical significance test by 
184: generate $100$ random Poisson samples of clusters which have the same number 
185: density in the same the box size as the Millennium halo catalog but having 
186: no clustering effect. The statistical significance of a supercluster can 
187: be estimated as $P(L) = 1 - N_{po}(L)/N_{sc}(L)$, where $N_{po}(L)$ and 
188: $N_{sc}(L)$ are the numbers of superclusters found in the random samples 
189: and in the Millennium Run simulation data at a linking length $L$, 
190: respectively.\citep{bas-etal06}.
191: 
192: Figure \ref{fig:lin} plots the statistical significance $P(L)$, which reveals 
193: that around $L=6 h^{-1}$ Mpc the statistical significance of finding a 
194: supercluster reaches $99\%$ confidence level.  Therefore, we set the linking 
195: length for the FOF algorithm at $L=6h^{-1}$Mpc which corresponds to the 
196: linkage parameter $b=0.35$. 
197: 
198: A total of $4014$ superclusters are found. Among them, a total of $382$ 
199: superclusters are found to consist of more than $5$ members. 
200: The mean mass of these $382$ superclusters are found to be 
201: $\bar{M}_{s}=4.2\times 10^{14}h^{-1}M_{\odot}$.  Using these superclusters, 
202: we measured the inertia momentum tensor of each supercluster. Rotating 
203: the system into the principal axis frame, we find the three eigenvectors 
204: of the inertia momentum tensor of each supercluster. 
205: 
206: \subsection{Alignments between Void Spin and Supercluster Principal Axes}
207: 
208: Now that the spin axes of voids and the principal axes of the superclusters 
209: are all determined, we calculate the squares of the cosines of the angles 
210: between the the void spin axes and the three axes of the superclusters as 
211: a function of the separation distance, $r$. 
212: Let $\hat{\bf y}^{\alpha},\hat{\bf y}^{\beta},\hat{\bf y}^{\gamma}$ 
213: be the three unit eigenvectors of the supercluster inertia momentum tensor, 
214: which represent the supercluster major, intermediate, and minor axes, 
215: respectively. Basically, we calculate the following three correlation 
216: functions using the selected void-supercluster pairs:
217: \begin{eqnarray}
218: \label{eqn:alpha}
219: \omega^{\alpha}(r) &\equiv& 
220: \langle\vert\hat{\bf J}\cdot\hat{\bf y}^{\alpha}\vert^{2}\rangle(r) 
221: - \frac{1}{3},\\
222: \label{eqn:beta}
223: \omega^{\beta}(r) &\equiv& 
224: \langle\vert\hat{\bf J}\cdot\hat{\bf y}^{\beta}\vert^{2}\rangle(r) 
225: - \frac{1}{3},\\
226: \label{eqn:gamma}
227: \omega^{\gamma}(r) &\equiv& 
228: \langle\vert\hat{\bf J}\cdot\hat{\bf y}^{\gamma}\vert^{2}\rangle(r)
229: - \frac{1}{3}.
230: \end{eqnarray}
231: If there is no alignment, then these cross-correlation function will be 
232: just zero.
233: 
234: To examine how these void-supercluster cross correlations depend on the 
235: linking length, we repeat the whole process using different values of $L$. 
236: The three cross-correlations functions between the void spin axes and the 
237: supercluster major ($\omega^{\alpha}$), intermediate ($\omega^{\beta}$), 
238: and minor axes ($\omega^{\gamma}$) for the fives different cases of the 
239: linking lengths are plotted in Figs. \ref{fig:varma}, \ref{fig:varin} and 
240: \ref{fig:varmi}.  As can be seen, for those values of $L$ that corresponds 
241: to higher than $90\%$ confidence levels (i.e., $L \le 7h^{-1}$Mpc), there are 
242: clear consistent anti-alignment and alignment signals between the void spin 
243: axes and the supercluster major and the minor axes, respectively, within a 
244: distance of $30h^{-1}$Mpc, and there is consistently no alignment signal 
245: with the supercluster intermediate axes.
246: 
247: We provide  physical explanations to this phenomena in \S 3 within the 
248: analytic framework proposed originally by \citet{lee-pen01}.
249: 
250: \section{ANALYTIC PREDICTION}
251: 
252: \subsection{Review of Spin-Shear and Direction-Shear Correlations}
253: 
254: According to the Lee-Pen formalism based on the linear tidal torque theory 
255: \citep{dor70,whi84}, the unit spin vector 
256: $\hat{\bf J}\equiv (\hat{J}_{i})$ 
257: of a bound halo is correlated with the unit traceless tidal shear tensor 
258: $\hat{\bf T}\equiv (\hat{T}_{ij})$ as  
259: \begin{equation}
260: \label{eqn:hspincorr}
261: \langle \hat{J}_{i}\hat{J}_{j}\vert\hat{\bf T}\rangle = 
262: \frac{1+a}{3}\delta_{ij} - a\hat{T}_{ik}({\bf x})\hat{T}_{kj}({\bf x}),
263: \end{equation}
264: where $a$ is the spin-shear correlation parameter in the range of 
265: $[0,3/5]$. It represents the strength of the correlation between $\hat{\bf J}$ 
266: and $\hat{\bf T}$: If $a=3/5$, the correlation is strongest. If $a=0$, 
267: there is no correlation. Here, the unit tidal tensor $\hat{\bf T}$ is 
268: intrinsic and local, defined at the halo position, ${\bf x}$.
269: 
270: \citet{lee-pen01} also suggested the following formula for the 
271: correlation between the direction vector to the nearest neighbor 
272: $\hat{\bf y}\equiv (\hat{y}_{i})$ with the unit traceless tensor 
273: $\hat{\bf T}$ as 
274: \begin{equation}
275: \label{eqn:hdencorr}
276: \langle \hat{y}_{i}\hat{y}_{j}\vert\hat{\bf T}\rangle = 
277: \frac{1-b}{3}\delta_{ij} + b\hat{T}_{ik}({\bf x})\hat{T}_{kj}({\bf x}), 
278: \end{equation}
279: Here the parameter, $b$ in the range of $[-1,1]$ represents the strength of 
280: the correlation between $\hat{\bf y}$ and $\hat{\bf T}$. 
281: Note the difference in the range between the two correlation 
282: parameters, $a$ and $b$. The maximum value of $b$ is $1$ while that of 
283: $a$ is $3/5$, less than unity. It reflects the fact that a perfect 
284: alignment is allowed between $\hat{\bf y}$ and $\hat{\bf T}$  
285: but not between $\hat{\bf J}$ and $\hat{\bf T}$. 
286: 
287: Anyway, a crucial implication of equations (\ref{eqn:hspincorr}) and 
288: (\ref{eqn:hdencorr}) is that the spin axes of halos should be closely 
289: correlated with the directional geometry of the nearby halo distribution 
290: since $\hat{\bf J}$ and $\hat{\bf y}$ are both correlated with the 
291: tidal field. In \S 3.2, we extrapolate the validity of 
292: equations (\ref{eqn:hspincorr}) and (\ref{eqn:hdencorr}) which hold good 
293: for halos to the voids and superclusters.
294: 
295: \subsection{Modeling Void-Supercluster Alignments}
296: 
297: LP06 have already extrapolated the validity of equation (\ref{eqn:hspincorr}) 
298: to unbound voids, assuming that the spin-shear correlation parameter $a$ has 
299: the maximum value of $3/5$ for the case of voids whose spin is defined  as 
300: (\ref{eqn:voidspin}):
301: \begin{equation}
302: \label{eqn:spincorr}
303: \langle \hat{J}_{i}\hat{J}_{j}\vert\hat{\bf T}\rangle = 
304: \frac{8}{15}\delta_{ij} - \frac{3}{5}
305: \hat{T}_{ik}({\bf x}_{v})\hat{T}_{kj}({\bf x}_{v}),
306: \end{equation}
307: where $\hat{\bf T}$ is now defined at the void center, ${\bf x}_{v}$. 
308: 
309: Now, we attempt to extrapolate the validity of equation (\ref{eqn:hdencorr}) 
310: to the alignments between the void spin axes and the supercluster principal 
311: axes. The superclusters are conspicuously elongated along local filaments 
312: where the dark matter are preferentially located. The nearest neighbors are 
313: most likely to be found in the direction along the supercluster major axes. 
314: But, the filaments are one dimensional structure, collapsed along the 
315: major and the intermediate principal axes of the local tidal tensors. 
316: Thus, the direction of local filaments (i.e., the supercluster major axes) 
317: are in fact anti-aligned with the tidal tensor major axes, and aligned 
318: with the tidal tensor minor axes.
319: 
320: Using the above logic, we assume the following:
321: \begin{itemize}
322: \item
323: For the direction of the supercluster major axis, $\hat{\bf y}^{\alpha}$, 
324: the direction-shear correlation parameter $b$ has the minimum 
325: value of $-1$:
326: \begin{equation}
327: \label{eqn:mdencorr}
328: \langle \hat{y}^{\alpha}_{i}\hat{y}^{\alpha}_{j}\vert\hat{\bf T}\rangle = 
329: \frac{2}{3}\delta_{ij} - \hat{T}_{ik}({\bf x}_{s})\hat{T}_{kj}({\bf x}_{s}). 
330: \end{equation}
331: \item
332: For the direction of the supercluster intermediate axis,$\hat{\bf y}^{\beta}$, 
333: the direction-shear correlation parameter $b$ has the value of $0$:
334: \begin{equation}
335: \label{eqn:idencorr}
336: \langle \hat{y}^{\beta}_{i}\hat{y}^{\beta}_{j}\vert\hat{\bf T}\rangle = 
337: \frac{1}{3}\delta_{ij}, 
338: \end{equation}
339: \item
340: For the direction of the supercluster minor axis, $\hat{\bf y}^{\gamma}$, 
341: the parameter $b$ has the maximum value of $1$:
342: \begin{equation}
343: \label{eqn:rdencorr}
344: \langle \hat{y}^{\gamma}_{i}\hat{y}^{\gamma}_{j}\vert\hat{\bf T}\rangle = 
345: \hat{T}_{ik}({\bf x}_{s})\hat{T}_{kj}({\bf x}_{s}), 
346: \end{equation}
347: \end{itemize}
348: Note that in equations (\ref{eqn:mdencorr})-(\ref{eqn:rdencorr}), the unit 
349: tidal tensor is defined at the supercluster center, ${\bf x}_{s}$.
350: 
351: Using equations (\ref{eqn:spincorr})-(\ref{eqn:rdencorr}), one can derive 
352: analytically the three correlation functions, $\omega^{\alpha}$, 
353: $\omega^{\beta}$ and $\omega^{\gamma}$, defined in \S 2.. 
354: In this derivation, the key part is to calculate the four point shear 
355: correlation, $\langle\hat{T}({\bf x}_{v})\hat{T}({\bf x}_{v})
356: \hat{T}({\bf x}_{s})\hat{T}({\bf x}_{s})\rangle$. 
357: \citet{lee-pen01} calculated this quantity for the case that the two unit 
358: tidal tensors are defined at the same halo position but smoothed on two 
359: different scales. What they found is that it is approximated as the density 
360: auto-correlation \citep[see Appendix I in][]{lee-pen01}.  
361: In our case, the two unit tidal tensors are not defined at the same 
362: position but at two different positions, the void and the supercluster 
363: centers. Therefore, the four-point shear correlation would not be approximated 
364: as the density autocorrelation. Instead, it may be approximated 
365: as the density two-point correlation, just like the void spin-spin 
366: correlation function \citep{lee-par06}.
367: 
368: Hence, using the same approximation used in \citep{lee-par06} but 
369: considering the fact that the maximum value of the correlation parameter 
370: $b$ is different from that of $a$, we find that the void-supercluster 
371: alignments can be approximated as 
372: \begin{eqnarray}
373: \label{eqn:aalpha}
374: \omega^{\alpha}(r) &\approx& 
375: -\frac{1}{10}\frac{\xi^{2}_{R}(r)}{\xi^{2}_{R}(0)},\\
376: \label{eqn:abeta}
377: \omega^{\beta}(r) &\approx& 0,\\
378: \label{eqn:agamma}
379: \omega^{\gamma}(r) &\approx& \frac{1}{10}\frac{\xi^{2}_{R}(r)}{\xi^{2}_{R}(0)},
380: \end{eqnarray}
381: where $r\equiv\vert{\bf x}_{v}-{\bf x}_{s}\vert$.
382: Here $\xi_{R}$ is the two point correlation function of the density field 
383: on the smoothing scale of $R$. For the void-supercluster alignments, the 
384: smoothing scale should be a minimum Lagrangian radius enclosing a 
385: void-supercluster pair. Since the separation distance $r$ cannot decrease 
386: below the sum of the supercluster radius and the void diameter in Lagrangian 
387: space, the smoothing scale may be written as
388: \begin{equation}
389: R = R_{s} + 2R_{v},
390: \end{equation}
391: where $R_{s}$ and $R_{v}$ represent the effective radii of a supercluster 
392: and a void, respectively. 
393: 
394: For the comparison with the numerical results, we relate the Lagrangian 
395: supercluster and void radii to the observables given in \S 2:
396: \begin{equation}
397: \bar{R}_{s} \equiv \left(\frac{3\bar{M}_s}{4\pi\bar{\rho}}\right)^{1/3}, 
398: \qquad \bar{R}_{v} \equiv (1 + \bar{\delta}_v)^{1/3}\bar{R}_{E},
399: \end{equation}
400: where $\bar{\rho}$ is the mean mass density of the universe.
401: 
402: Equations (\ref{eqn:aalpha}), (\ref{eqn:abeta}) and (\ref{eqn:agamma}) are 
403: plotted in Figs. \ref{fig:major}, \ref{fig:inter} and \ref{fig:minor}, 
404: respectively, where the numerical results (with $L = 6h^{-1}$Mpc) derived in 
405: \S 2.3 are also plotted as solid dots with Poissonian errors. For the analytic 
406: results, we use the transfer function of the initial power spectrum given by 
407: \citet{bar-etal86} with the cosmological parameters set at the values used in 
408: the Millennium Run simulations. And, the shape parameter, $\Gamma$ is 
409: approximated as $\Gamma=\Omega_{m}h$ (V. Springel in private communication). 
410: As can been obviously seen, the analytic approximations work very well. 
411: The excellent agreements between the analytical and the numerical results 
412: imply that the void-supercluster alignments are indeed generated by the tidal 
413: interactions between the voids and the superclusters.
414: 
415: \section{DISCUSSION AND CONCLUSION}
416: 
417: We have investigated correlations in spatial orientations between voids 
418: and their neighbor superclusters using the Millennium Run simulation of 
419: a concordance cosmology.  Adopting the concept of void spin proposed by 
420: \citet{lee-par06}, we have found for the first time that the void spin 
421: axes are very strongly correlated with the supercluster minor axes 
422: within the separation distance of $30h^{-1}$Mpc. Testing how the result 
423: depend on the choice of void and supercluster definition, we have found 
424: that our result is quite solid. 
425: 
426: To this numerical phenomena has a physical explanation been provided based 
427: on tidally generated correlations. Under the assumption that the 
428: neighboring superclusters are representative of the filamentary 
429: surrounding matter which wrap and exert tidal forces on the voids, 
430: we have derived an analytic formula for the alignments between the void 
431: spin axes and the supercluster minor axes. The analytic prediction has 
432: turned out to agree with the numerical result very well.
433: 
434: However, it is worth discussing a caveat which the success of our work is 
435: subject to. This caveat lies in the fact that there is no consensus on how 
436: to define voids and superclusters, unlike the case of halo-defining.
437: Using different algorithms could produce different results for correlations 
438: between voids and superclusters. Although we have shown here that the 
439: void-supercluster alignments do not strongly depend on the linking length 
440: of the FOF algorithm and the cut-off mass of the void halos, it will 
441: definitely necessary to compare between different methods for the void 
442: and supercluster identifications in the future.
443: 
444: Together with our previous work on the void spin-spin correlation 
445: \citep{lee-par06}, this new result supports the scenario that the voids 
446: originate from the initial regions where the tidal effect becomes maximum.
447: In addition, this new result on the void-supercluster alignments 
448: demonstrate more directly how the largest scale structure and voids are 
449: connected in a cosmic web through tidal influences.
450: 
451: A final conclusion is that our work will provide a new clue to describing the 
452: void-supercluster network in the cosmic web and leads us to have a deeper 
453: insight into the formation and evolution of the large scale structure of 
454: the universe.
455: \acknowledgments
456: 
457: The Millennium Run simulation used in this paper was carried out by the Virgo 
458: Supercomputing Consortium at the Computing Centre of the Max-Planck Society 
459: in Garching. We thank V. Springel and G. Lemson for plenty of helps. 
460: This work is supported by the research grant No. R01-2005-000-10610-0 from 
461: the Basic Research Program of the Korea Science and Engineering Foundation.
462: 
463: 
464: \begin{thebibliography}{100}
465: \bibitem[Bardeen et al.(1986)]{bar-etal86}
466: Bardeen, J.M., Bond, J.R.,Kaiser, N., \& Szalay, A.S. 1986, \apj, 304, 15
467: \bibitem[Basilakos et al.(2006)]{bas-etal06}
468: Basilakos, S., Plionis, M., Yepes, G., Gottlober, S., Turchaninov, V. 2005, 
469: \mnras, 365, 539
470: \bibitem[Colberg et al.(2005)]{col-etal05}
471: Colberg J. M., Sheth R. K., Diaferio A., Gao L., \& Yoshida N. 2005,
472: \mnras, 360, 216
473: \bibitem[Doroshkevich(1970)]{dor70}
474: Doroshkevich, A. G. 1970, astrofizika, 6, 581
475: \bibitem[El-Ad \& Piran(1997)]{ela-pir97}
476: El-Ad, H., \& Piran, T. 1997, \apj, 491, 421
477: \bibitem[Gottl\"{o}ber et al.(2003)]{got-etal03}
478: Gottl\"{o}ber, S., Lokas, E. L., Klypin, A., \& Hoffman, Y. 2003,
479: \mnras, 344, 715
480: \bibitem[Hoyle \& Vogeley(2002)]{hoy-vog02}
481: Hoyle, F., \& Vogeley, M. S. 2002, \apj, 566, 641
482: \bibitem[Hoyle \& Vogeley(2004)]{hoy-vog04}
483: Hoyle, F., \& Vogeley, M. S. 2004, \apj, 607, 751
484: \bibitem[Icke(1984)]{ick84}
485: Icke V. 1984, \mnras, 206, 
486: \bibitem[Lee \& Park(2006)]{lee-par06}
487: Lee, J. \& Park, D. 2006, \apj, 652, 1
488: \bibitem[Park \& Lee(2007)]{lee-par07}
489: Park, D. \& Lee, J. 2007, submitted to Phys. Rev. Lett.
490: \bibitem[Lee \& Pen(2001)]{lee-pen01}
491: Lee, J. \& Pen, U. L. 2001, \apj, 555, 106 
492: \bibitem[Sahni et al.(1994)]{sah-etal94}
493: Sahni, V., Sathyaprakah, B. S., \& Shandarin, S. F. 1994, \apj, 431, 20
494: \bibitem[Shandarin et al.(2004)]{sha-etal04}
495: Shandarin, S. F., Sheth, J., \& Sahni, V. 2004, \mnras, 353, 517
496: \bibitem[Shandarin et al.(2006)]{sha-etal06}
497: Shandarin, S., Feldman, H. A., Heitmann, K., \& Habib, S. 2006,
498: \mnras, 367, 1629
499: \bibitem[Sheth \& van de Weygaert(2004)]{she-van04}
500: Sheth, R. K., \& van de Weygaert, R. 2004, \mnras, 350, 517
501: \bibitem[Springel et al.(2005)]{spr-etal05}
502: Springel, V. et al. 2005, \nat , 435, 629
503: \bibitem[Wray et al.(2006)]{wra-etal06}
504: Wray, J. J., Bahcall, N. A., Bode, P., Boettiger, C., Hopkins, P. F. 2006, 
505: \apj, 652, 907
506: \bibitem[White(1984)]{whi84}
507: White, S. D. M. 1984, \apj, 286, 38
508: \end{thebibliography}
509: 
510: \clearpage
511: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
512:  \begin{figure}
513:   \begin{center}
514:    \plotone{f1.eps}
515: \caption{The probability density distribution of the cosines of the angles 
516: between ${\bf J}$ and ${\bf J}^{\prime}$.}
517: \label{fig:psi}
518:  \end{center}
519: \end{figure}
520: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
521: \clearpage
522: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
523: \begin{figure}
524:   \begin{center}
525:    \plotone{f2.eps}
526: \caption{The statistical significance of the linking length for the 
527: FOF algorithm to find superclusters in the Millennium Run halo catalog.}
528: \label{fig:lin}
529:  \end{center}
530: \end{figure}
531: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
532: \clearpage
533: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
534:  \begin{figure}
535:   \begin{center}
536:    \plotone{f3.eps}
537: \caption{Numerical result of the cross-correlation between the void spin 
538: axes and the major axes of the neighboring superclusters as a function of 
539: separation distance for the five different cases of the linking length, $L$.}
540: \label{fig:varma}
541:  \end{center}
542: \end{figure}
543: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
544: \clearpage
545: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
546:  \begin{figure}
547:   \begin{center}
548:    \plotone{f4.eps}
549: \caption{Same as Figure \ref{fig:major} but with the supercluster 
550: intermediate axes.}
551: \label{fig:varin}
552:  \end{center}
553: \end{figure}
554: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
555: \clearpage
556: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
557:  \begin{figure}
558:   \begin{center}
559:    \plotone{f5.eps}
560: \caption{Same as Figure \ref{fig:major} but with the supercluster 
561: minor axes.}
562: \label{fig:varmi}
563:  \end{center}
564: \end{figure}
565: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
566: \clearpage
567: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
568:  \begin{figure}
569:   \begin{center}
570:    \plotone{f6.eps}
571: \caption{Comparison of the analytic cross-correlation between the void spin 
572: axes and the major axes of the neighboring superclusters (solid line) 
573: with the numerical result from the Millennium Rum simulation (dots). The 
574: linking length for the numerical result is set at $6^{-1}$Mpc which 
575: corresponds to the $99\%$ confidence level for the supercluster identification 
576: using the FOF algorithm. The errors represent the standard deviation for the 
577: case of  no alignment. The horizontal line corresponds to the case of no 
578: alignment}
579: \label{fig:major}
580:  \end{center}
581: \end{figure}
582: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
583: 
584: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
585:  \begin{figure}
586:   \begin{center}
587:    \plotone{f7.eps}
588: \caption{Same as Figure \ref{fig:major} but with the supercluster 
589: intermediate axes.}
590: \label{fig:inter}
591:  \end{center}
592: \end{figure}
593: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
594: 
595: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
596:  \begin{figure}
597:   \begin{center}
598:    \plotone{f8.eps}
599: \caption{Same as Figure \ref{fig:major} but with the supercluster 
600: minor axes.}
601: \label{fig:minor}
602:  \end{center}
603: \end{figure}
604: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
605: \end{document}
606: