1:
2: \documentclass[12pt,preprint]{aastex}
3: %\usepackage{psfig}
4: %\usepackage{graphicx}
5: %\textheight=23cm
6:
7: \begin{document}
8: \shorttitle{Minivoids in the Local Volume}
9:
10: \shortauthors{A. Tikhonov}
11:
12: %\break Astronomy Letters, 2007, Vol. 33, No. 8, pp. 499 -- 511.
13: %Pleiades Publishing, Inc., 2007.
14: \title{Voids in the SDSS Galaxy Survey}
15: \author{Anton V.\ Tikhonov }
16: \affil{St.Petersburg State University, Saint-Petersburg, Russia}
17: \email{ti@hotbox.ru, avt@gtn.ru}
18: %\begin{center}
19:
20: \begin{abstract}
21:
22: Using the method of searching for arbitrary shaped voids in the
23: distribution of volume-limited samples of galaxies from the DR5
24: SDSS survey, we have identified voids and investigated their
25: characteristics and the change in these characteristics with
26: decreasing $M_{lim}$ (from -19.7 to -21.2, $H_0 = 100$~km/s/Mpc)
27: --- the upper limit on the absolute magnitude of the galaxies
28: involved in the construction of voids. The total volume of the 50
29: largest voids increases with decreasing $M_{lim}$ with a break
30: near M* = -20.44 --- the characteristic value of the luminosity
31: function for SDSS galaxies. The mean density contrast in voids
32: increases with decreasing $M_{lim}$ also with a weak break near M*
33: The exponent of the dependence of the volume of a void on its rank
34: increases significantly with decreasing $M_{lim}$ starting from
35: $M_{lim} \sim -20.4$ in the characteristic range of volumes, which
36: reflects the tendency for greater clustering of brighter galaxies.
37: The averaged profile of the galaxy density contrast in voids has a
38: similar pattern almost at all $M_{lim}$. The galaxies mostly tend
39: to concentrate toward the void boundaries and to avoid the central
40: void regions; the density contrast profile is flat in the
41: intermediate range of distances from the void boundaries. The
42: axial ratios of the ellipsoids equivalent to the voids are, on
43: average, retained with changing $M_{lim}$ and correspond to
44: elongated and nonoblate void shapes, but some of the voids can
45: change their shape significantly. The directions of the greatest
46: void elongations change chaotically and are distributed randomly
47: at a given $M_{lim}$. The void centers show correlations
48: reflecting the correlations of the galaxy distribution on scales
49: $(35 - 70)h^{-1}$~Mpc. The galaxy distribution in the identified
50: voids is nonrandom --- groups and filaments can be identified. We
51: have compared the properties of the galaxies in voids (in our
52: case, the voids are determined by the galaxies with absolute
53: magnitudes $M_{abs} < M_{lim} = -20.44$, except for the isolated
54: galaxies) and galaxies in structures identified using the minimum
55: spanning tree. A bimodal color distribution of the galaxies in
56: voids has been obtained. A noticeable difference is observed in
57: the mean color indices and star formation rates per unit stellar
58: mass of the galaxies in dense regions (structures) -- as expected,
59: the galaxies in voids are, on average, bluer and have higher
60: $log(SFR/M_{star})$. These tendencies become stronger toward the
61: central void regions.
62: \end{abstract}
63:
64: Keywords: galaxies, voids, large-scale structure of Universe.
65: \vspace{1ex}
66:
67: \section{Introduction}
68:
69: The distribution of galaxies is a complex “cosmic network”. The
70: walls, filaments, and voids observed at the present epoch reflect
71: both linear and nonlinear evolution of clustering. The pattern of
72: the observed clustering out to $20-25h^{-1}$~Mpc can be described,
73: for example, by a power law with a complex dependence of the
74: exponent on the luminosity, color, and other properties of
75: galaxies with the subsequent transition to homogeneous
76: distribution, with the structures being traceable on scales
77: exceeding the scale of homogeneity (Tikhonov 2006a, 2006b). The
78: nature of such clustering depends on many small- and large-scale
79: factors such as the cosmological parameters, the environments of
80: galaxies and clusters, their formation history, the distribution
81: of dark matter, and the scenario according to which the luminous
82: and dark matter are related and evolve. The characteristics of
83: voids have long been considered as tests of cosmological models.
84: Regoes and Geller (1991) found that in their model for the
85: formation of structures, certain initial conditions lead to the
86: formation of a “cellular” structure with voids similar to those
87: observed in galaxy surveys. Voids are the forming components of
88: the large-scale structure. In recent years, various authors have
89: considered in detail both observational and theoretical aspects of
90: the existence and evolution of voids detected both in galaxy
91: catalogs and in the dark matter halo distributions obtained in the
92: $\Lambda CDM$ model calculations of the N-body problem by Hoyle
93: and Vogeley (2004), Gottlober et al. (2003), Shandarin et al.
94: (2004), Croton et al. (2004), Benson et al. (2003), Colberg et al.
95: (2005), and Patiri et al. (2006b). The void statistics are closely
96: related to the methods of calculating the galaxy clustering; for
97: example, VPF (Void Probability Function) provides information
98: about the high-order correlation functions (Croton et al. 2004).
99:
100: One important problem of the modern theory of the formation of
101: structures is that according the $\Lambda CDM$ model for the
102: evolution of dark matter structures including the $\Lambda$-term
103: attributable to the existence of “dark energy,” much matter must
104: be present in voids, while the expected number of galaxies in
105: voids is not observed (Peebles 2001).
106:
107: Sheth and van de Weygaert (2004) developed a model for the
108: distribution of void sizes and evolution in terms of the
109: hierarchical clustering scenario. Furlanetto and Piran (2006)
110: developed an analytical model that predicted the shape of the
111: distribution of void sizes. In particular, they found that because
112: of the so-called “bias” effect, the voids in the galaxy
113: distribution are considerably larger than those in the dark matter
114: distribution.
115:
116: The shapes of voids are of interest along with the spectrum of
117: their sizes. Based on numerical calculations, Ike (1984) concluded
118: that, in most cases, the voids between filaments must be nearly
119: spherical in shape. Plionis and Basilakos (2002) analyzed the
120: distribution of void sizes and shapes in the PSCz survey and
121: compared them with the artificial distributions obtained in terms
122: of various CDM models. Shandarin et al. (2004) found, in
123: particular, that their large voids defined as regions with a
124: density lower than a given value in the smoothed density field of
125: the dark matter distribution are essentially nonspherical.
126:
127: Patiri et al. (2006a) found that the distribution
128: of galaxies in voids in the distribution of 2dFGRS
129: galaxies differs significantly from a random one.
130:
131: It has been firmly established that, compared to the general
132: distribution, the galaxies in voids have bluer colors, lower
133: luminosities, and higher star formation rates. In addition, a
134: higher abundance of disk galaxies is observed among the galaxies
135: in voids (Peebles 2001; see also Patiri et al. 2006b). Rojas et
136: al. (2004, 2005) confirmed these tendencies by analyzing the
137: photometric and spectroscopic properties of the galaxies in voids.
138: Hogg et al. (2004) considered the dependence of the galaxy color
139: and luminosity distributions on the density contrast and found
140: that, on the one hand, the most luminous galaxies populate the
141: densest regions, while the blue galaxies are present mostly in
142: low-density regions, and, on the other hand, the mean parameters
143: of the distributions in absolute magnitude and color change only
144: slightly with over- density. Based on SDSS galaxies, Baldry et al.
145: (2004) showed that the galaxy color distribution is bimodal and is
146: described well by two Gaussians. They obtained a fitto the (u-r)
147: --- color–absolute magnitude relation and compared the luminosity
148: functions for red and blue galaxies. Hoyle et al. (2005) found
149: significant differences between the luminosity functions of the
150: galaxies in voids and dense regions.
151:
152: Patiri et al. (2006b), who analyzed the voids in the DR4 SDSS
153: survey, found no significant differences between the mean
154: parameters of the field and void galaxies. In this paper, we use a
155: different approach to selecting a “check sample” of galaxies for
156: comparison with the properties of the galaxies in voids --- we
157: selected the galaxies of the “check sample” in high density
158: structures located entirely outside the void boundaries.
159:
160: In this paper, we also analyze the variations of void
161: parameters with luminosity and perform a correlation
162: analysis of the distribution of void centers.
163:
164: \section{THE DATA}
165:
166: The spectroscopic redshifts are expected to be obtained for about
167: 106 galaxies and 105 quasars within the framework of the Sloan
168: Digital Sky Survey (SDSS) based on photometric data for a sky
169: region $10^4$ square degrees in area in the Northern Galactic
170: Hemisphere in five bands ($u$, $g$, $r$, $i$, $z$) with a limiting
171: magnitude of $r = 22.5$ (York et al. 2000; Stoughton et al. 2002)
172: once the program has been fully implemented. The photometric data
173: were used for a homogeneous selection of various classes of
174: objects to obtain their spectra. Two types of galaxies were chosen
175: for determining the redshifts from the list of objects classified
176: as extended ones: galaxies with a Petrosian magnitude $r < 17.77$
177: and a surface brightness exceeding 24 m/$\square$". formed the
178: Main Galaxy Sample (the number of objects in the final SDSS
179: version is $\sim900000$); the LRG (Luminous Red Galaxies) list
180: includes galaxies with very red colors and $r < 19.5$ (the number
181: of objects in the final SDSS version is $\sim100000$). In this
182: paper, we analyzed data from the fifth data release DR5 SDSS
183: (www.sdss.org, Adelman-McCarthy et al. 2006).
184:
185: When analyzing the DR5 data, we selected a rectangular part from
186: the region of spectroscopic sky coverage for the convenience of
187: allowance for the boundary conditions in determining the void
188: boundaries and for ensuring sample completeness. In the ($\lambda,
189: \eta $) coordinate system of the survey, the selected region is
190: $-48^\circ < \lambda < 48^\circ$, $6^\circ < \eta < 36.5^\circ$.
191:
192: The Main Galaxy Sample is an apparent-magnitude-limited survey,
193: which determines the method of constructing the volume-limited
194: sample to eliminate incompleteness in radial coordinate --— we set
195: the limit on the $r$-band absolute magnitude for the sample
196: galaxies equal to $M^0_{lim} = r_{lim}-25-5log(R_{max}
197: (1+z_{max}))-K(z)$,where $r_{lim} = 17.77$ was taken as the
198: limiting $r$-band magnitude, $K(z)$ is the K-correction, and
199: $R_{max}$ is the chosen far boundary in radial coordinate
200: corresponding to zmax.The u and r magnitudes used here were
201: corrected for extinction.
202:
203: To estimate the absolute magnitudes of the galaxies, we used a
204: mean Ê correction for SDSS galaxies in the form $K(z)=
205: 2.3537z^2+0.5735z-0.18437$ (Hickage et al. 2005; see also Blanton
206: et al. 2003). The metric distances were recalculated from the
207: redshifts with the Hubble parameter $H_0 =
208: 100$~km~s$^{-1}$Mpc$^{-1}$, $h=H/H_0$, where $H$ is the true value
209: of the Hubble constant, and the density parameters
210: $\Omega_{\Lambda} = 0.7$, $\Omega_0 = 0.3$ (see, e.g., Hogg 1999).
211:
212: \section{THE METHOD}
213:
214: The void construction algorithm presented here has already been
215: applied by Tikhonov (2006b) and Tikhonov and Karachentsev (2006)
216: and is basically similar to the algorithm described by El-Ad and
217: Piran (1997). The voids were constructed in the distribution of
218: “bright” galaxies with absolute magnitudes $M_{abs}$ (in the $r$
219: band) lower than a certain value of $M_{lim}$. Here, we searched
220: for voids containing a certain number of galaxies from the
221: volume-limited sample with $M_{abs} < M_{lim}$. For the
222: void-forming galaxies, we determined the mean distance to the
223: nearest neighbor $R_{n}$ and the standard deviation $\sigma_n$.If
224: there was no neighbor with $M_{abs} < M_{lim}$ in the sphere of
225: radius $R_{n}+\sigma_n$ around a particular galaxy of this sample,
226: then this galaxy was excluded from the list of galaxies involved
227: in the construction of voids. Thus, these excluded galaxies could
228: fall into voids. The mean distance from the isolated galaxies to
229: the nearest neighbor is considerably larger than that for isolated
230: pairs (when two galaxies lie in the sphere of radius
231: $R_{n}+\sigma_n$) --- it is close to the mean distance between the
232: galaxies for pairs. Thus, having eliminated the influence of
233: isolated galaxies, we obtain more “stable” voids.
234:
235: Next, we successively searched for galaxy-free seed spheres inside
236: the sample volume (first, the largest sphere is searched for) and
237: then expanded them by adding spheres whose centers are inside the
238: already fixed part of the void and whose radii $R_{sph}$ are not
239: smaller than the radius of the seed sphere multiplied by the
240: coefficient $k = 0.9$ ($R_{sph} > 0.9 \cdot R_{seed}$, where
241: $R_{seed}$ is the radius of the seed sphere). The voids are
242: assumed to be located entirely within the geometrical boundaries
243: of the sample.
244:
245: The voids constructed in this way (at $k = 0.9$) have arbitrary
246: shaped volumes. On the other hand, the voids are separated from
247: each other and fairly “thick” throughout the volume, which allows
248: them to be approximated by triaxial ellipsoids.
249:
250: There exist other methods of searching for voids that are more
251: commonly used in analyzing artificial dark matter distributions
252: (see, e.g., Shandarin et al. 2006). In this approach, a smoothed
253: (e.g., with a Gaussian filter) density field is constructed (the
254: parameters of the resulting structures depend on the smoothing
255: length) and a certain threshold local density that separates the
256: low-and high-density regions is specified. In this case, the voids
257: can be highly irregular in shape. A brief overview of the methods
258: and references can be found, for example, in Tikhonov (2006b) and
259: Patiri et al. (2006a).
260:
261: \section{THE DEPENDENCE OF VOID PROPERTIES
262: ON LUMINOSITY}
263:
264: To analyze how the properties of the voids vary over a wide
265: luminosity range of the galaxies forming them, we chose the
266: redshifts limits $z_{min} = 0.02$ and $z_{max} = 0.1$ (sample A).
267: The upper limit on the absolute magnitude of the galaxies in the
268: volume-limited sample with these boundaries is $M^0_{lim} =
269: -19.67$. For this sample, the number of galaxies is $N = 47892$,
270: the mean density is $\rho \approx 7 \cdot 10^{-3}h^{3}$,
271: $R_{n}+\sigma_n \approx 3.0h^{-1}$~Mpc. The range of limits
272: $M_{lim}$ on the absolute magnitudes of the galaxies involved in
273: constructing the voids in which we analyzed the dependence is from
274: $-19.7$ (40526 galaxies after the exclusion of isolated galaxies) to
275: $-21.2$ (1241 galaxies). We chose this range in such a way that
276: the sample was volume limited and that it contained a sufficient
277: number of galaxies. The resolution (the separation between the
278: grid points) was about 1.7 Mpc in all cases.
279:
280: In the same vein, we analyzed sample B with the redshift limits
281: $z_{min} \approx 0.080$ and $z_{max} = 0.115$ chosen in such a way
282: that the sample covered the same volume (about $6.82 \cdot 10^6
283: \cdot h^{-3}$)~Mpc$^{3}$) as sample A. In this case, $M^0_{lim} =
284: -20.00$, $R_{n}+\sigma_n \approx 3.4h^{-1}$~Mpc.
285:
286: \begin{figure}
287: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
288: \includegraphics[]{Fig1.eps}
289: } \figcaption{Total volume of the 50 largest voids as a function
290: of the limiting absolute magnitude $M_{lim}$ of the galaxies
291: involved in constructing the voids for samples A and B. Also shown
292: is the number of galaxies.}
293: \end{figure}
294:
295:
296: Figure 1 shows the total volume of the 50 largest voids in samples
297: A and B and the number of void- forming galaxies as a function of
298: $M_{lim}$. The increase in total volume occurs synchronously with
299: the decrease in $M_{lim}$ and in the number of galaxies with
300: $M_{abs}< M_{lim}$ (except for the isolated galaxies). A
301: significant break in the dependence is observed near $M_{lim} =
302: -20.5$, i.e., immediately after the characteristic value of $M^* =
303: -20.44$ of the luminosity function for SDSS galaxies followed by a
304: faster growth of the total volume. This is not just the result of
305: the corresponding decrease in the number of void-forming galaxies
306: (on the contrary, the decrease in the number of galaxies slows
307: down). The 50 largest voids occupy from about $21\%$ ($M_{lim} =
308: -19.7$) to $64\%$ ($M_{lim} = -21.1$) of the entire sample volume.
309:
310: \begin{figure}
311: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
312: \includegraphics[]{Fig2.eps}
313: } \figcaption{Mean overdensity of the galaxies in voids with
314: $R_{seed}>15$~Mpc with the distance $r$ from the void boundaries
315: normalized to $R_{eff}$ versus $M_{lim}$. }
316: \end{figure}
317:
318:
319: Figure 2 shows the dependence of the mean density contrast
320: (contrast profile) for the galaxies of sample A
321: $\delta\rho/<\rho_{VL}>$ (where $<\rho_{VL}>$ is the mean density
322: of the galaxies with $M_{abs}< -19.67$ in sample A) that fell into
323: a layer inside a particular void with the distance $r$ from the
324: void boundaries normalized to the effective void radius $R_{eff} =
325: (3 \cdot Vol/4\cdot\pi)^{-1/3}$ ($Vol$ stands for the void volume)
326: on the limiting absolute magnitude $M_{lim}$. We averaged the
327: density contrast profile inside all voids with $R_{seed}>15h^{-1}$
328: Mpc. The density contrast profile has common characteristic
329: features for different $M_{lim}$: the galaxies concentrate to the
330: void boundaries; there are virtually no galaxies in the central
331: regions ($r/R_{eff}> 0.7-0.8$); the density contrast profiles of
332: the galaxies in voids are flat up to $r/R_{eff}=0.3-0.2$ followed
333: by a significant increase in contrast. At $M_{lim} < -21.0$, the
334: galaxy density near the void boundaries (in the first bin) is
335: higher than the mean density ($\delta\rho/<\rho_{VL}> > 0$).
336:
337: \begin{figure}
338: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
339: \includegraphics[]{Fig3.eps}
340: } \figcaption{Mean overdensity $<\delta_v>$ of the galaxies in
341: voids (relative to the mean density of the sample galaxies with
342: $M_{abs} < M_1$) versus $M_{lim}$ for various $M_1$ (different
343: symbols are used for different values).}
344: \end{figure}
345:
346: Figure 3 shows the dependence of the mean density contrast for the
347: galaxies in voids $<\delta_v>$ with $R_{seed}>15h^{-1}$ Mpc on
348: $M_{lim}$ for four limits $M_1$ on the absolute magnitude of the
349: galaxies retained in the volume-limited sample (from which the
350: mean density in the formula for the density contrast $<\delta_v> =
351: \delta\rho/\rho_{M_1}$) was obtained). In each of the four cases,
352: the change in $M_{lim}$ began from $M_{lim} = M_1$. An indistinct
353: break in the dependence can be distinguished at $M_{lim} \approx
354: -20.6$. At fixed $M_{lim}$, the mean density contrast is lower
355: (the voids become “emptier”) at lower $M_1$. Convergence is
356: observed for all $M_1$ as $M_{lim}$ decreases. This indicates that
357: the most luminous galaxies form a stable skeleton of the structure
358: (the fraction of isolated galaxies decreases with decreasing
359: $M_1$). In all cases, the mean density contrast $<\delta_v> < 0$,
360: i.e., the identified voids are physically separated low-density
361: regions.
362:
363: After compiling the list of voids (assigning the three-dimensional
364: grid points to a particular void), we determined the void centers
365: and calculated the moments of inertia of the bodies formed by the
366: voids. We analyzed the void shapes based on the parameters of the
367: equivalent ellipsoids. We constructed a $3\times3$ matrix of the
368: moments of inertia $I_{ij}$ and used the condition
369: $det(I_{ij}-\lambda \cdot E)=0$, where $E$ is a $3\times3$ unit
370: matrix, to find its eigenvalues $\lambda_i$, which are equal to
371: the principal moments of inertia, from which the semiaxes of the
372: equivalent ellipsoid were determined. The eigenvectors of matrix
373: $I_{ij}$ give the directions of the semiaxes. The direction of the
374: greatest void elongation coincides with that of the largest
375: semiaxis of the equivalent ellipsoid.
376:
377: The void shapes were analyzed for the first 20 identified voids.
378: In general, the change in void configuration and shape with
379: decreasing $M_{lim}$ is indicative of an irregular change in the
380: configuration of the entire large-scale structure (in our
381: approach, the influence of isolated galaxies was eliminated). The
382: void centers are displaced significantly and the order of void
383: identification changes.
384:
385:
386: \begin{figure}
387: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
388: \includegraphics[]{Fig4.eps}
389: } \figcaption{Example of the correlation between the smallest-
390: to-largest (c/a)and medium-to-largest (b/a)axial ratios of the
391: equivalent ellipsoids of the first 20 identified voids. The slope
392: of the linear fit is $\phi = 1.0$ and $M_{lim} = -20.3$.}
393: \end{figure}
394:
395:
396: \begin{figure}
397: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
398: \includegraphics[]{Fig5.eps}
399: } \figcaption{Slope $\phi$ of the linear fit to the distribution
400: of the b/a and c/a ratios (in the approximation of a triaxial
401: ellipsoid) for the first 20 (at given $M_{lim}$)identified voids
402: versus $M_{lim}$.}
403: \end{figure}
404:
405: \begin{figure}
406: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
407: \includegraphics[]{Fig6.eps}
408: } \figcaption{Mean c/a and b/a ratios for the first 20 voids
409: versus $M_{lim}$. The distribution errors $\sigma$ are given for
410: the c/a ratios.}
411: \end{figure}
412:
413: The b/a and c/a axial ratios of the equivalent ellipsoid, where a,
414: b and c are the largest, medium, and smallest axes, respectively,
415: are correlated (see Fig.4 for $M_{lim}=-20.3$). Figure 5 shows the
416: mean slopes $\phi$ with the errors of the linear fits to the b/a
417: and c/a distributions for various $M_{lim}$. The dependence is
418: stable starting from $M_{lim} = -20.3$ and is approximately equal
419: to 1, i.e. the voids are predominantly in the shape of a slightly
420: elongated cucumber.
421:
422: The mean c/a and b/a show a small trend at $M_{lim}< -20.6$ (for
423: the c/a ratio, Fig.6 presents $\sigma$ of the distribution in the
424: form of an error) --- the rations decrease with increasing
425: luminosity and the voids, on average, become more elongated. At
426: the same time, the relation between c/a and b/a is retained in the
427: entire $M_{lim}$ range --- the void oblateness is small and
428: changes only slightly. For all $M_{lim}$, the cases where c/a <
429: 0.5 for any of the first 20 voids are rare (the void has a
430: significant elongation)—most of the voids have c/a > 0.6 (the
431: cases of “circular” voids with c/a > 0.9 are also rare).
432:
433: \begin{figure}
434: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
435: \includegraphics[]{Fig7.eps}
436: } \figcaption{Directions of the greatest elongations (the largest
437: axes a of the equivalent ellipsoids) of the first 20 voids
438: identified in sample À for $M_{lim}$. The ranges of the directions
439: corresponding to the geometrical boundaries of the sample are
440: shown.}
441: \end{figure}
442:
443: Figure 7 presents the directions of the greatest elongations for
444: the first 20 voids for the chosen $M_{lim}$ range. The void
445: orientations change significantly even for neighboring values of
446: $M_{lim}$ (although there are also cases where the void
447: orientation is retained in a certain $M_{lim}$ range, e.g., near
448: $\lambda = 25^o$, $\eta = -35^o$) and fill almost the entire area
449: of possible directions rather uniformly.
450:
451: We also considered the variation in the parameters of a single
452: void that, for most values of $M_{lim}$, is the first void
453: identified by the algorithm and its shape is determined only by
454: the geometry of the distribution and does not depend on the
455: boundaries of other voids. The volume and $R_{seed}$ of this void
456: increase synchronously with decreasing $M_{lim}$ (Fig. 8); the
457: dependencies have a break at $M_{lim}=-20.6$ followed by a faster
458: increase in $R_{seed}$ and the volume. The orientation of this
459: void changes significantly and chaotically (Fig. 9). The c/a and
460: b/a ratios increase irregularly with decreasing $M_{lim}$ (Fig.
461: 10) --- the void becomes “more circular” (except for
462: $M_{lim}=-21.2$ at which the surrounding structure apparently
463: changes greatly). The relation between c/a and b/a, along with the
464: direction of the greatest elongation, changes significantly and
465: irregularly.
466:
467: \begin{figure}
468: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
469: \includegraphics[]{Fig8.eps}
470: } \figcaption{Volume and $R_{seed}$ of one first identified void
471: versus $M_{lim}$.}
472: \end{figure}
473:
474: \begin{figure}
475: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
476: \includegraphics[]{Fig9.eps}
477: } \figcaption{Evolution of the direction of the elongation of one
478: first void with $M_{lim}$.}
479: \end{figure}
480:
481:
482: The dependence of the void volume ($Vol$) on the void rank
483: ($Rank$) at fixed $M_{lim}$ (the largest void has rank 1, the next
484: void has rank 2, etc.) may have a simple interpretation (Gaite and
485: Manrubia 2002). In particular, the break in this dependence
486: reflects the scale of the transition from a power-law galaxy
487: distribution to uniformity (Gaite 2005; Tikhonov 2006b). In this
488: paper, we analyzed the change in the slope $z$ of the $log(Vol)$
489: --- $log(Rank)$ relation with decreasing $M_{lim}$ in the chosen range
490: of volumes. We choose the range of volumes for each $M_{lim}$
491: between volume $V_1 \approx 10^4h^{-3}$Mpc$^3$ (the relation
492: exhibits a cutoff at the void ranks corresponding to volume V1,
493: which is probably determined by the constraint imposed on the
494: minimum $R_{seed}$ of the identified voids) and volume $V_2$
495: corresponding to the break in the $log(Vol)$ --- $log(Rank)$
496: relation, which is interpreted as the beginning of the transition
497: to uniformity in the galaxy distribution (Fig. 11). The power law
498: segment of the relation between $V_1$ and $V_2$ with exponent $z$
499: reflects a power-law galaxy distribution on small scales. The
500: exponent $z$ increases with decreasing $M_{lim}$ with a break near
501: $M^*$ (Fig. 12). Staring from $M_{lim} = -21.6$, the values of $z$
502: become larger than unity, which allows the formal fractal
503: dimension $D_z = 3/z$ of the galaxy distribution at given
504: $M_{lim}$ to be estimated (this dimension is considered as a
505: measure of the extent to which the sample volume is filled with
506: galaxies). The significant increase in $z$ with decreasing
507: $M_{lim}$ after the break reflects the well-known fact that more
508: luminous galaxies are clustered more strongly (see, e.g., Tikhonov
509: 2006a). The values of $z < 1$ and the nearly flat pattern of
510: variation in $z$ to $M_{lim} = -21.4$ need a further study and an
511: explanation. The effective radius of a void with volume $V_2$,
512: $R_{eff} = (3 \cdot V_{2}/4\cdot\pi)^{1/3}$, can be associated
513: with the scale of the transition to uniformity at given $M_{lim}$.
514: In the range of values under study, this scale ($R_{eff}$) lines
515: in the range $19-27 \cdot h^{-1}$~Mpc and increases irregularly
516: with luminosity.
517:
518: \begin{figure}
519: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
520: \includegraphics[]{Fig10.eps}
521: } \figcaption{c/a and b/a ratios of one first void versus
522: $M_{lim}$.}
523: \end{figure}
524:
525: \begin{figure}
526: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
527: \includegraphics[]{Fig11.eps}
528: } \figcaption{Void volume versus void rank at $M_{lim} = -20.3$.
529: The power-law segment in the volume interval $V_1 - V_2$.}
530: \end{figure}
531:
532: \begin{figure}
533: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
534: \includegraphics[]{Fig12.eps}
535: } \figcaption{Evolution of the exponent of the “void volume ---
536: rank" relation in the characteristic volume interval with
537: $M_{lim}$.}
538: \end{figure}
539:
540:
541: \section{THE DISTRIBUTION OF VOID CENTERS}
542:
543: In this section, we analyzed a sample with $z_{min} = 0.02$,
544: $z_{max} = 0.12$, $M^0_{lim} = -20.11$, in which 327 voids with
545: $R_{seed}>9h^{-1}$ Mpc were identified with a resolution of about
546: 1.5 Mpc. The void centers were defined as the centers of mass of
547: the set of grid points assigned to a given void. We analyzed the
548: distribution of void centers using an cumulative correlation gamma
549: function (conditional density) (Coleman and Pietronero 1992;
550: Tikhonov 2006a). Once the signal has been stabilized (on small
551: scales, the signal is absent, because the distances between the
552: void centers are larger than 18 Mpc), the void centers exhibit a
553: correlation on scales $35 - 70h^{-1}$~Mpc with exponent $\gamma_v
554: \sim 0.5$ (Fig. 13). The galaxies involved in determining the
555: voids ($M_{abs} < 20.11$, except for isolated galaxies with
556: $R_{n}+\sigma_n \approx 3.5$~Mpc) show a slightly weaker
557: correlation, $\gamma_g \sim 0.4$, on these scales. These scales
558: correspond to the range of scales in the galaxy distribution in
559: which the transition to uniformity occurs (Tikhonov 2006a). The
560: slightly larger exponent for the void centers is apparently
561: obtained due to the empty regions near the boundaries, where there
562: are no void centers (according to the definitions in the void
563: search algorithm) and due to the differences in void volumes.
564:
565: \begin{figure}
566: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
567: \includegraphics[]{Fig13.eps}
568: } \figcaption{(a) Correlation functions of the galaxies (filled
569: circles) and void centers (open circles). The right and left Y
570: axes show the amplitudes of the correlation functions of the void
571: centers and galaxies, respectively. (b) The distribution of void
572: centers and galaxies in a $\pm15$~Mpc layer along the Z-axis (in
573: $\lambda$, $\eta$ coordinates).}
574: \end{figure}
575:
576:
577:
578: \section{COMPARISON OF THE PROPERTIES
579: OF GALAXIES IN VOIDS AND STRUCTURES}
580:
581: To compare the properties of the galaxies in voids and structures,
582: we constructed a volume-limited sample with $z_{min} = 0.02$ and
583: $z_{max} = 0.12$ with a limit on the absolute magnitude $M^0_{lim}
584: = -20.11$. The voids were determined in the distribution of
585: galaxies with $M_{abs} < M_{lim} = -20.44$ in such a way that some
586: the galaxies from the volume-limited sample could fall into voids.
587: We then excluded isolated galaxies (without any neighbors at a
588: distance smaller than $R_{n}+\sigma_n \approx 4.2$~Mpc) from the
589: resulting list of void-forming galaxies. In this way, a certain
590: number of isolated galaxies with $M_{abs} < -20.44$ fall into
591: voids.
592:
593: We identified 235 voids with $R_{seed} > 10 h^{-1}$~Mpc. The voids
594: were divided into “large” ones with $R_{eff} = (3 \cdot Vol/4
595: \cdot\pi)^{1/3} > 20h^{-1}$~Mpc and “small” ones with $R_{eff} <
596: 20h^{-1}$ Mpc. $R_{seed}$ and $R_{eff}$ are correlated: $R_{eff} =
597: (1.26\pm0.02) \cdot R_{seed}$. However, the void with a smaller
598: $R_{seed}$ (i.e., identified later) may have a larger volume. The
599: galaxies that fell into voids were divided into “bright” galaxies
600: of the volume-limited sample with $M_{abs} < -20.11$, i.e., those
601: without any selection in the radial direction, and “faint” ones
602: with $M_{abs}> -20.11$. A total of 2480 “bright” and 6104 “faint”
603: galaxies fell into large voids. The mean overdensity of the
604: galaxies from the volume-limited sample with $M_{abs} < -20.11$
605: that fell into voids is $\delta\rho/<\rho_{VL}> = -0.78$,where
606: $<\rho_{VL}>$ is the mean density of the galaxies with $M_{abs} <
607: M_{lim} = -20.11$.
608:
609: The galaxy distribution in the identified voids shows the same
610: features as the galaxy distribution in the entire volume (as was
611: pointed out by Patiri et al., 2006b). Thus, for example, the
612: galaxy distribution in void ¹ 2 in order of identification in a
613: 20Mpc layer in Z (in $\lambda$, $\eta$ coordinates) is essentially
614: nonuniform (Fig. 14) --- groups of galaxies and a filament
615: crossing the void can be identified. The galaxies delineate the
616: void boundaries and avoid the central region.
617:
618: To compare the galaxy properties, we selected a check sample
619: containing galaxies in structures (dense regions). The galaxies in
620: structures, i.e., those that do not fall into voids, were
621: identified by constructing the minimum spanning tree. This tree
622: consists of knots and edges and is constructed by appending new
623: knots satisfying the condition for the distance to the already
624: constructed part of the tree being at a minimum (Barrow et al.
625: 1985). We constructed the minimum spanning tree in a sample with
626: the angular boundaries $-30^\circ < \lambda < 30^\circ$, $10^\circ
627: < \eta < 30^\circ$.
628:
629: Once the minimum spanning tree has been constructed from “bright”
630: galaxies with $M_{abs} < -20.11$, we identified the galaxies that
631: were connected by the edges of lengths no larger than $L_{max}$
632: determined from the assumed limit on the overdensity. The relation
633: between these parameters is defined by formula $\rho = 1/V =
634: 3/4\cdot\pi\cdot L_{max}^3$. As the lower limit, we chose the
635: overdensity $\delta\rho/<\rho_{VL}> = 2$ ($<\rho_{VL}> \approx 4.5
636: \cdot 10^3$~Mpc$^{-3}$), which corresponds to an edge of length
637: $L_{max}=2.6h^{-1}$~Mpc. The mean distance to the nearest neighbor
638: in this sample is $R_n \approx 2.0h^{-1}$~Mpc. When the edges
639: smaller than $L_{max}$ are excluded from the tree, the tree breaks
640: up into connected “islands” with $\delta\rho/<\rho_{VL}> > 2$. For
641: our analysis, we retained only those “islands” that contained more
642: than 40 “bright” galaxies. In the sample under consideration,
643: these structures contain a total of 3011 “bright” galaxies. The
644: “faint” galaxies with $M_{abs} > -20.11$ offset by less than
645: $L_{max}$ from the island galaxies were then appended to these
646: structures. The number of “faint” galaxies attributed to
647: structures turned out to be 3533.
648:
649:
650: \newpage
651: \begin{figure}
652: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
653: \includegraphics[]{Fig14.eps}
654: } \figcaption{Distribution of galaxies in a $\pm10h^{-1}$~Mpc
655: layer along the Z-axis relative to the center of void No. 2 in
656: order of identification. The asterisk indicates the void center.}
657: \end{figure}
658:
659:
660: The distribution of the “faint” galaxies that fell into voids and
661: structures is subject to selection (incompleteness) in the radial
662: direction, but we will consider their properties separately from
663: those of the galaxies from the volume-limited sample --- the
664: influence of incompleteness is averaged to some degree, because
665: the voids and structures are distributed quite uniformly over the
666: sample volume.
667:
668: Figure 15 presents the luminosity distribution for the galaxies in
669: voids and structures. On average, the galaxies in structures are
670: more luminous than those in voids --- an excess of “bright”
671: galaxies in structures and “faint” galaxies in voids is observed.
672: The abrupt cutoff of the histogram for the galaxies in voids near
673: $M_{abs} = -20.44$ results from the fact that only isolated
674: galaxies with $M_{abs} < -20.44$ fall into voids according to the
675: construction.
676:
677: \begin{figure}
678: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
679: \includegraphics[]{Fig15.eps}
680: } \figcaption{Distribution of galaxies in voids and structures in
681: absolute magnitudes (opposite hatching). $M_{VL}=-20.11$ is the
682: boundary of the volume-limited sample and $M_{lim}=-20.44$ is the
683: upper limit on the absolute magnitudes of the involved in
684: constructing the voids.}
685: \end{figure}
686:
687: Figure 16 presents the galaxy number distribution in large voids
688: as a function of the galaxy distance $r$ from the void boundary
689: normalized to the effective void radius $R_{eff}$ for “bright”
690: galaxies with $-20.44 < M_{abs} < -20.11$ and isolated galaxies
691: with $M_{abs} < -20.44$ that fell into voids. Both subsamples show
692: a similar increase in the number of galaxies toward the void
693: boundaries. This, in particular, suggests that our approach to
694: excluding isolated galaxies from the list of galaxies in the
695: distribution of which we identified voids is legitimate.
696:
697:
698:
699: \begin{figure}
700: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
701: \includegraphics[]{Fig16.eps}
702: } \figcaption{Galaxy number distribution as a function of the
703: distance from the void boundary normalized to the effective void
704: radius, $R_{eff}$, for all voids with $R_{eff} > 20h^{-1}$~Mpc.
705: Opposite hatching is used to denote the histograms for galaxies
706: with $-20.44<M_{abs}<-20.11$ and isolated galaxies in these voids
707: with $M_{abs}<-20.44$.}
708: \end{figure}
709:
710:
711: The galaxies in the identified voids exhibit a binomial (red and
712: blue) color distribution (Fig. 17). The mean color for “bright”
713: galaxies in large voids $<u-r>^v_{VL} = 2.22$,
714: $\sigma^v_{VL}=0.50$. For "faint" galaxies, $<u-r>^v_{dim} =
715: 1.94$, $\sigma^v_{dim}=0.60$, The systematic difference reflects
716: the fact that the “faint” galaxies are, on average, bluer. Similar
717: characteristics are observed for the galaxies in small voids. For
718: “bright” and “faint” galaxies in structures, $<u-r>^s_{VL} =
719: 2.47$, $\sigma^s_{VL}=0.50$ è $<u-r>^s_{dim} = 2.18$,
720: $\sigma^s_{dim}=0.57$. They are systematically redder than those
721: in voids. The histograms for the galaxies in voids and structures
722: differ significantly in shape --- the fraction of blue galaxies is
723: much larger in voids.
724:
725: \begin{figure}
726: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
727: \includegraphics[]{Fig17.eps}
728: } \figcaption{$u - r$ colors of galaxies in voids and structures
729: (the histograms with opposite hatching): (a) “bright” galaxies
730: with $M_{abs} < -20.11$, (b) “faint” galaxies with $M_{abs} >
731: -20.11$.}
732: \end{figure}
733:
734:
735: The data on the star formation rates per unit stellar mass for
736: SDSS galaxies ($log(SFR/M_{star}$), below referred to as SFR)
737: (Kauffmann 2003) were taken from the SDSS archive
738:
739: \begin{figure}
740: \centerline{%\psfig{figure=Fig1.eps,height=10cm,angle=0}
741: \includegraphics[]{Fig18.eps}
742: } \figcaption{Star formation rates per unit stellar mass
743: ($log(SFR/M_{star})$) for (a) “bright” and (b) “faint” galaxies in
744: voids and structures.}
745: \end{figure}
746:
747:
748: (http://www.mpagarching.mpg.de/SDSS/Dr4/). Figure 18 compares
749: $log(SFR/M_{star})$ for the galaxies in large voids and
750: structures. The mean value and dispersion of $log(SFR/M_{star})$
751: for large voids for “bright” galaxies are$<SFR>^v_{VL} = -10.26$,
752: $\sigma^v_{VL}=0.49$, respectively. For “faint” galaxies,
753: $<SFR>^v_{dim} = -10.09$, $\sigma^v_{dim}=0.49$. Again the
754: difference in the mean values stems from the fact that faint
755: galaxies have, on average, higher star formation rates than bright
756: galaxies. The “bright” and “faint” galaxies in structures have
757: $<SFR>^s_{VL} = -10.43$, $\sigma^s_{VL}=0.50$, $<SFR>^s_{dim} =
758: -10.25$, $\sigma^s_{dim}=0.49$, respectively. The star formation
759: rates in structures are systematically lower.
760:
761:
762: The red ($u-r > 2$) and blue ($u-r < 2$) galaxies show the
763: following characteristics for “bright” galaxies in large voids:
764: $<u-r>^v_{blue} = 1.68$, $\sigma^v_{blue}=0.21$; $<u-r>^v_{red} =
765: 2.52$, $\sigma^v_{red}=0.42$ and in structures: $<u-r>^s_{blue} =
766: 1.73$, $\sigma^s_{blue}=0.20$; $<u-r>^s_{red} = 2.63$,
767: $\sigma^s_{red}=0.37$. For “faint” galaxies in voids:
768: $<u-r>^v_{blue} = 1.56$, $\sigma^v_{blue}=0.26$; $<u-r>^v_{red} =
769: 2.46$, $\sigma^v_{red}=0.39$ and in structures: $<u-r>^s_{blue} =
770: 1.60$, $\sigma^s_{blue}=0.26$; $<u-r>^s_{red} = 2.51$,
771: $\sigma^s_{red}=0.41$.
772:
773: The differences for voids and structures are found to be
774: insignificant but larger than those obtained in a similar analysis
775: by Patiri et al. (2006b).
776:
777: For the galaxies in structures, the “($u-r$) color --- absolute
778: magnitude” (the trend with a slope of $slope_1 = -0.24$
779: --- brighter galaxies are, on average, redder),
780: “$log(SFR/M_{star})$ --- absolute magnitude ($slope_2 = 0.14$ ---
781: brighter galaxies have, on average, lower star formation rates),
782: and “color --- log(SFR/Mstar)” ($slope_3 = -0.59$ --- bluer
783: galaxies have, on average, higher star formation rates) relations
784: are less pronounced than those for the galaxies in voids, for
785: which $slope_1 = -0.27$, $slope_2 = 0.16$, $slope_3 = -0.66$,
786: respectively, although the differences are small.
787:
788: The mean luminosity of the galaxies in all the identified voids is
789: virtually independent of the galaxy distance from the void
790: boundary normalized to the effective void radius $R_{eff}$. The
791: “bright” and “faint” galaxies in voids show opposite weak trends:
792: on average, the galaxies with $M_{abs} < -20.11$ become brighter
793: and the galaxies with $M_{abs} > -20.11$ become fainter as one
794: goes from the boundaries to the central regions of the voids.
795:
796: The (“bright” and “faint”) galaxies in voids, on average, have
797: bluer colors (lower $u-r$) and higher star formation rates (the
798: slopes of the linear fits to the “(($u-r$) --- $r/R_{eff}$ è
799: $log(SFR/M_{star})$ --- $r/R_{eff}$” relations are -0.12 and 0.06,
800: respectively) as $r/R_{eff}$ increases (where $r$ is the distance
801: from the boundary of a given void). Thus, the differences in the
802: properties of the galaxies in voids and structures slightly
803: increase even further when less dense regions inside the
804: identified voids are considered.
805:
806: \section{CONCLUSIONS}
807:
808: In this paper, using an algorithm of identifying arbitrarily
809: shaped voids, we analyzed the change in some of the void
810: characteristics with luminosity of the galaxies (in the range of
811: limits $M_{lim}$ on the absolute magnitude -19.7 --- -21.2)
812: involved in the construction of voids and compared the properties
813: of the galaxies in voids (mean density contrast) and the galaxies
814: forming structures with an density contrast higher than 2.
815:
816: The evolution of a number of void characteristics with luminosity
817: shows a break near M*, in agreement with the change in exponent in
818: a correlation analysis of the distribution of SDSS galaxies
819: (Tikhonov 2006a). The structure of the galaxy distribution
820: apparently changes qualitatively at $M_{abs} < M^*$.
821:
822: As expected, the void volumes increase with decreasing $M_{lim}$
823: with a break near M* followed by a faster increase (in contrast,
824: the decrease in the number of galaxies that form voids slows
825: down), with the pattern being the same for two samples of equal
826: volume covering different regions of space.
827:
828: The mean density contrast in voids also increases with decreasing
829: $M_{lim}$ with a weak break near M*. The voids become “emptier” if
830: we reduce the limit on the absolute magnitude of the galaxies
831: under consideration at fixed $M_{lim}$.
832:
833: The galaxies inside voids concentrate to the void boundaries and
834: avoid the central regions. The density contrast profile is flat in
835: intermediate regions. These results agree with those obtained
836: previously from the galaxies (Patiri et al. 2006b) and the dark
837: matter haloes in voids (Gottlober et al. 2003). Our study also
838: confirms that the matter in voids is distributed irregularly and
839: has the same features as the general distribution obtained by the
840: above authors.
841:
842: In general, the mean characteristics of the void shapes are
843: retained with decreasing $M_{lim}$. We can only note a weak
844: tendency: at $M_{lim}$ < -20.6, the voids become, on average,
845: slightly more elongated. There is also a weak tendency for the
846: slope of the distributions of the medium-to-larger (b/a)
847: smaller-to-larger (c/a) axial ratios of the ellipsoid equivalent
848: to the void (these ratios are well correlated) to increase with
849: decreasing $M_{lim}$. The slopes of the linear fits to these
850: distributions are close to unity (0.8 -- 1.1) in the entire
851: $M_{lim}$ range, i.e., the voids are predominantly elongated and
852: nonoblate at all $M_{lim}$. At the same time, the individually
853: considered voids can change their shape with $M_{lim}$
854: significantly. The directions of the greatest void elongations are
855: distributed quite uniformly and change chaotically with $M_{lim}$,
856: which is indicative of an irregular change in the structure when
857: more luminous galaxies are considered.
858:
859: The exponent of the “void volume -- rank” ($log(Vol)$ --
860: $log(Rank)$) relation in the characteristic range of volumes
861: increases significantly with decreasing $M_{lim}$ starting from
862: $M_{lim} = -20.4$ (a break is again observed in the relation at a
863: value close to M*), which is a reflection of the tendency for more
864: luminous galaxies to cluster more strongly. The scales of the
865: beginning of the transition to uniformity (the break of the
866: power-law segment in the $Vol$ -- $Rank$ relation at volume $V_2$
867: -- see the text) determined from this relation increase with
868: decreasing $M_{lim}$ and agree with the results of a correlation
869: analysis of the galaxy distribution.
870:
871: The distribution of void centers shows a certain correlation and
872: reflects the correlations of the galaxy distribution on the
873: corresponding scales.
874:
875: The derived differences in the properties of the galaxies in voids
876: with a mean density contrast of -0.78 and the galaxies in
877: structures with an density contrast higher than 2 (identified
878: using the minimum spanning tree) agree qualitatively with the
879: existing views: the galaxies in voids are, on average, bluer and
880: have higher star formation rates per unit stellar mass. The last
881: two tendencies become stronger when galaxies located closer to the
882: central void regions are considered. However, quantitatively,
883: these differences are not large enough to conclude that the
884: formation histories of the galaxies in structures and voids differ
885: fundamentally.
886:
887: The need for a further study of the observed correlations and
888: multiparameter tendencies and for determining the mean galaxy
889: characteristics over wide ranges of density contrasts and
890: luminosities and for various definitions of the characteristic
891: galaxy environments is obvious. Our division into voids and
892: structures is to some extent arbitrary and consists mainly in
893: different density contrasts: there are also structures in the
894: voids determined in this paper.
895:
896: \section{ACKNOWLEDGMENTS}
897: This work was supported by grant No. MK-6899.2006.2 of the
898: President of Russian Federation.
899:
900:
901: \section{REFERENCES}
902:
903:
904: 1. J. K. Adelman-McCarthy, M.A.Agueros, S.S.Allam, et al.,
905: Astron.J.(in press).
906:
907: 2. I. Baldry, K. Glazebrook, and J. Brinkmann, Astrophys. J. 600,
908: 681 (2004).
909:
910: 3. J. Barrow, S. Bhavsar, and D. Sonoda, Mon. Not. R. Astron. Soc.
911: 216, 17 (1985).
912:
913: 4. A. J. Benson, F. Hoyle, F. Torres, and M. Vogeley, Mon. Not. R.
914: Astron. Soc. 340, 160 (2003).
915:
916: 5. M. Blanton, J. Brinkmann, and I. Csabai, Astron. J. 125, 2348
917: (2003).
918:
919: 6. J. Colberg, R. Sheth, A. Diaferio, et al., Mon. Not. R. Astron.
920: Soc. 360, 216 (2005).
921:
922: 7. P. H. Coleman and L. Pietronero, Phys. Rep. 213, 311 (1992).
923:
924: 8. D. J. Croton, M. Colles, E. Gaztanaga, et al., Mon. Not. R.
925: Astron. Soc. 352, 828 (2004); astro-ph/0401406.
926:
927: 9. H. El-Ad and T. Piran, Astrophys. J. 491, 421 (1997).
928:
929: 10. S. Furlanetto and T.Piran, Mon.Not. R.Astron. Soc. 366, 467
930: (2006).
931:
932: 11. S. Furlanetto and T.Piran, Mon.Not. R.Astron. Soc. 366, 467
933: (2006).
934:
935: 12. J. Gaite, Europ. Phys.J.B 47, 93 (2005).
936:
937: 13. J. Gaite and S.C.Manrubia, Mon. Not. R. Astron. Soc. 335, 977
938: (2002); astro-ph/0205188.
939:
940: 14. S.Gottlober, E.L.Locas, A.Klypin, and Y.Hoffman, Mon. Not. R.
941: Astron. Soc. 344, 715 (2003); astro-ph/0305393.
942:
943: 15. D.W. Hogg, astro-ph/9905116 (1999).
944:
945: 16. D. Hogg, M. Blanton, J. Brinchmann, et al., Astrophys. J. 601,
946: L29 (2004).
947:
948: 17. F. Hoyle and M. S. Vogeley, Astrophys. J. 607, 751 (2004);
949: astro-ph/0312533.
950:
951: 18. F. Hoyle, R. Rojas, M. Vogeley, and J. Brinkmann, Astrophys.
952: J. 620, 618 (2005).
953:
954: 19. B. V. Icke, Mon.Not.R.Astron. Soc. 206, 1 (1984).
955:
956: 20. G. Kauffmann, T. Heckman, S. White, et al., Mon. Not. R.
957: Astron. Soc. 341, 33 (2003).
958:
959: 21. S. G. Patiri, J. Betancort-Rijo, F. Prada, et al., Mon. Not.
960: R. Astron. Soc. 369, 335 (2006a); astro-ph/0506668.
961:
962: 22. S. Patiri, F. Prada, J. Holtzman, et al., Mon. Not. R. Astron.
963: Soc. 372, 1710 (2006á).
964:
965: 23. P. J. E. Peebles, Astrophys. J. 557, 495 (2001).
966:
967: 24. M. Plionis and S. Basilakos, Mon. Not. R. Astron. Soc. 330,
968: 399 (2002).
969:
970: 25. E. Regoes and M.Geller, Astrophys.J. 377,14 (1991).
971:
972: 26. R. Rojas, M. Vogeley, F. Hoyle, and J. Brinkmann, Astrophys.
973: J. 617, 50 (2004); astro-ph/0307274.
974:
975: 27. R. Rojas, M. Vogeley, F. Hoyle, and J. Brinkmann, Astrophys.
976: J. 624, 571 (2005).
977:
978: 28. W. Saunders, W. J. Sutherland, S. J. Maddox, et al., Mon. Not.
979: R. Astron. Soc. 317, 55 (2000).
980:
981: 29. S. Shandarin, J. Sheth, and V. Sahni, Mon. Not. R. Astron.
982: Soc. 353, 162 (2004).
983:
984: 30. S. Shandarin, H. A. Feldman, K. Heitmann, and S.Habib, Mon.
985: Not.R.Astron. Soc. 367, 1629 (2006).
986:
987: 31. R. Sheth and R. van de Weygaert, Mon. Not. R. Astron. Soc.
988: 350, 517 (2004).
989:
990: 32 C. Stoughton, R. Lupton, M. Bernardi, et al., Astron. J. 123,
991: 485 (2002).
992:
993: 33. A. V. Tikhonov, Pis’ma Astron. Zh. 32, 803 (2006a) [Astron.
994: Lett. 32, 721 (2006a)]; astro-ph/0610643.
995:
996: 34. A. V. Tikhonov, Pis’ma Astron. Zh. 33, 809 (2006b) [Astron.
997: Lett. 33, 727 (2006b)]; astro-ph/0610689.
998:
999: 35. A. V. Tikhonov and I. D. Karachentsev, Astrophys. J. 653, 969
1000: (2006).
1001:
1002:
1003: \end{document}
1004: