1: \documentclass[doublecol]{epl2}
2: % or \documentclass[page-classic]{epl2} for one column style
3: \newcommand{\bea}{\begin{eqnarray}}
4: \newcommand{\eea}{\end{eqnarray}}
5: \newcommand{\be}{\begin{equation}}
6: \newcommand{\ee}{\end{equation}}
7: \newcommand{\bef}{\begin{figure}}
8: \newcommand{\eef}{\end{figure}}
9:
10:
11: \def\eg{{e.g.}}
12: \def\ie{{i.e.}}
13: \def\spose#1{\hbox to 0pt{#1\hss}}
14: \def\ltapprox{\mathrel{\spose{\lower 3pt\hbox{$\mathchar"218$}}
15: \raise 2.0pt\hbox{$\mathchar"13C$}}}
16: \def\gtapprox{\mathrel{\spose{\lower 3pt\hbox{$\mathchar"218$}}
17: \raise 2.0pt\hbox{$\mathchar"13E$}}}
18: \def\inapprox{\mathrel{\spose{\lower 3pt\hbox{$\mathchar"218$}}
19: \raise 2.0pt\hbox{$\mathchar"232$}}}
20:
21: \title{Absence of self-averaging and of homogeneity in the
22: large scale galaxy distribution}
23: \shorttitle{Inhomogeneity of galaxy distribution}
24:
25: \author{Francesco Sylos Labini\inst{1,2}
26: \and Nikolay L. Vasilyev \inst{3}
27: \and Luciano Pietronero \inst{2,4}
28: \and Yurij V. Baryshev \inst{3}
29: }
30: \shortauthor{F. Sylos Labini \etal}
31:
32: \institute{
33: \inst{1} Museo Storico della Fisica e Centro Studi e
34: Ricerche Enrico Fermi, - Piazzale del Viminale 1,
35: 00184 Rome, Italy \\
36: \inst{2} Istituto dei Sistemi
37: Complessi CNR, - Via dei Taurini 19, 00185 Rome, Italy \\
38: \inst{3} Institute of Astronomy, St.Petersburg State
39: University - Staryj Peterhoff, 198504, St.Petersburg, Russia\\
40: \inst{4} Dipartimento di Fisica, Universit\`a di Roma
41: ``Sapienza'' - P.le Aldo Moro 2, 00185, Rome Italy
42: }
43: \pacs{98.80.-k}{Cosmology}
44: \pacs{05.40.-a}{Fluctuations phenomena in random processes}
45: \pacs{02.50.-r}{02.50.-r Probability theory, stochastic processes,
46: and statistics}
47:
48: \abstract{ The properties of the galaxy distribution at large scales
49: are usually studied using statistics which are assumed to be
50: self-averaging inside a given sample. We present a new analysis
51: able to quantitatively map galaxy large scale structures while
52: testing for the stability of average statistical quantities in
53: different sample regions. We find that the newest samples of the
54: Sloan Digital Sky Survey provide unambiguous evidence that galaxy
55: structures correspond to large amplitude density fluctuations at all
56: scales limited only by sample sizes. The two-point correlations
57: properties are self-averaging up to approximately $30$ Mpc/h and are
58: characterized by a fractal dimension $D=2.1\pm0.1$. Then at all
59: larger scales probed density fluctuations are too large in amplitude
60: and too extended in space to be self-averaging inside the considered
61: volumes. These inhomogeneities are compatible with a continuation
62: of fractal correlations but incompatible with: (i) a homogeneity
63: scale smaller than 100 Mpc/h, (ii) predictions of standard
64: theoretical models, (iii) mock galaxy catalogs generated from
65: cosmological Nbody simuations.}
66:
67: \begin{document}
68:
69: \maketitle
70:
71:
72: \section{Introduction}
73:
74: Understanding the large scale structure of the universe as mapped by
75: galaxy distribution represents one of the cornerstones of modern
76: cosmology. It provides the basic test for theories of structure
77: formation in the universe. A primary question in the statistical
78: analysis of three-dimensional galaxy catalogs (where, in addition to
79: the angular coordinates, the redshift is measured and through
80: Hubble's law \cite{pee80} the distance of each object) concerns the
81: determination of a scale where the distribution becomes
82: homogeneous. Such a scale $\lambda_0$ can be defined to be the one
83: beyond which counts of galaxies in three dimensional spherical volumes
84: of radius $r$ grow as $ r^3$ \cite{book}.
85:
86: %A related analysis concerns
87: %the determination of characteristic correlations at different spatial
88: %scales.
89:
90:
91: A decade ago, by measuring the conditional density, i.e., the local
92: galaxy density seen by a galaxy in a spherical volume of radius $r$
93: around itself \cite{pie87,book}, some of us found that galaxy
94: correlations are power-law with an exponent $\gamma \approx 1$ up to
95: the sample sizes, i.e., $\sim 30$ Mpc/h
96: \footnote{We use $H_0=100h$ km/sec/Mpc, with $0.4\le h \le 0.7$, for
97: the Hubble's constant}, corresponding to a fractal dimension
98: $D=3-\gamma\approx 2$ \cite{cp92,slmp98,jmsl99}. These results were
99: in contrast with the analysis of the same samples by, e.g.,
100: \cite{dp83,park,benoist}, who found $\lambda_0 \approx 10$ Mpc/h and
101: $\gamma=1.8$. The reason for these differences lies in the a-priori
102: assumption of homogeneity, inside a given sample, of the standard
103: statistical analysis \cite{pie87,slmp98,book}.
104:
105:
106: At larger scales, with weaker statistical significance, there was an
107: evidence compatible with the fact that power-law correlations in the
108: conditional density extend up to $r \sim 100$ Mpc/h or more
109: \cite{slmp98,esp}. These results generated a debate in the field
110: \cite{wu,davis,pmsl96} because even though galaxy structures were
111: found in many different catalogs to extend to scales of the order of
112: hundreds of megaparsecs, the characteristic length scale $\lambda_0$
113: statistically describing their correlations was determined to be a few
114: megaparsecs \cite{dp83,park,benoist}. While for some this was a
115: paradox \cite{pie87,cp92,slmp98}, for others
116: \cite{dp83,park,davis,benoist,wu} the explanation was that large scale
117: structures have small amplitude relative to the average density.
118: However this interpretation is problematic as in the range of scales
119: where the conditional density shows power-law correlations the sample
120: density is not well defined while density fluctuations have large
121: amplitude \cite{pie87,slmp98,book}. The determination of the
122: crossover scale $\lambda_0$, where the conditional density from a
123: power-law turns to a constant, allowing a meaningful determination of
124: the average density, has been thus an important task of galaxy
125: correlations studies in the last decade \cite{bt05}.
126:
127:
128: Two new galaxy catalogs, the Sloan Digital Sky Survey (SDSS ---
129: \cite{york}) and the Two degree Field Galaxy Redshift Survey
130: \cite{colless03}, have recently provided great advances in the mapping
131: of the local universe both for the number of objects measured in
132: continuously growing volumes and for the determination of several
133: parameters for each of them. Several studies
134: \cite{2df_paper,dr4_paper,hogg,tikonov} of different samples of these
135: surveys confirmed the small scale correlations measured by
136: \cite{slmp98,jmsl99}. In addition it has been claimed that a slow
137: crossover toward homogeneity occurs \cite{hogg} with the average
138: conditional density in spheres at $\sim 20$ Mpc/h having twice the
139: amplitude of the asymptotic density reached at $r>70$ Mpc/h
140: \cite{tikonov}. It was however noticed that galaxy structures could
141: bias the determination of correlations in these samples introducing
142: uncontrolled systematic effects \cite{joyce05,dr4_paper,2df_paper}.
143: Recently in the Two degree Field Galaxy Redshift Survey it has been
144: found that \cite{paper_2df_prl,paper_2df_aea} galaxy distribution
145: characterized by large amplitude fluctuations with a large spatial
146: extension, whose size is only limited by the sample's boundaries. In
147: addition at scales $r < 40$ Mpc/h, it has been observed a well defined
148: and statistically stable power-law behavior of the average number of
149: galaxies in spheres in agreement with previous determinations.
150:
151:
152:
153: A different and complementary method to characterize structures is
154: provided by galaxy counts as a function of the radial distance from
155: us or of the apparent luminosity \cite{gsl00}. These show large
156: fluctuations around the average behavior both in redshift
157: \cite{kerscher98} and angular surveys \cite{picard91,bd97}. There
158: have been controversies as to whether these are due to real clustering
159: or to incompleteness of the catalogs \cite{wu,gsl00}. Recent results
160: support the conclusion that the local galaxy distribution is
161: characterized by large scale structures with significant correlations
162: on scales $r > 50$ Mpc/h \cite{busswell03,frith03}.
163:
164:
165:
166: In this paper we use a new method which is able to establish, in a
167: given sample, the link between the small scales $r<30$ Mpc/h
168: correlations and the large scales $r >30$ Mpc/h fluctuations in galaxy
169: counts and which clarifies how the latter influence the determination
170: of the former. Using it we can test whether sample means, variances
171: and correlations are well defined, i.e., whether they are
172: statistically stable in different sub-volumes of the given sample. By
173: applying this method to the data of the SDSS project \cite{paperdr6}
174: we detect large density fluctuations of spatial extension limited by
175: the samples' sizes. We show that these introduce systematic biases in
176: the determination of large scale correlations.
177:
178:
179:
180: \section{The Data}
181: \label{samples_sdss}
182:
183: The SDSS \cite{york} is currently the largest spectroscopic survey of
184: extragalactic. Here we consider
185: %the data from the latest public data
186: %release of
187: the main galaxy (MG) sample (DR6) \cite{paperdr6}
188: containing redshifts for about 800,000 galaxies which %. The spectroscopic
189: %survey
190: covers an area of 7425 square degrees on the celestial sphere.
191: %We have used the following criteria t
192: To query the DR6 database we constrain the flags indicating the type
193: of object so that we select only the galaxies from the MG sample. We
194: then consider galaxies in the redshift range $10^{-4} \leq z \leq
195: 0.3$. The redshift confidence parameter is constrained to be $z_{conf}
196: \ge 0.35$ with flags indicating no significant redshift determination
197: errors. In addition we apply the filtering condition $m_r < 17.77$,
198: using Petrosian apparent magnitudes in the $r$ filter which are
199: corrected for galactic absorption, and thus taking into account the
200: target magnitude limit for the MG sample in the DR6
201: \cite{strauss2002}. In this way we have selected 479,417 objects. We
202: considered also more stringent limits in apparent magnitude, to test
203: whether a possible incompleteness of the survey at bright and/or faint
204: apparent magnitudes could generate a fake signal.
205: To this aim we used
206: we have
207: %selected galaxies with Petrosian magnitude with extinction corrections
208: $14.5 \le m_r \le 17.5$
209: %. In this way, by using the same redshift
210: %constraints as for the previous case,
211: and we selected 370,893 objects, i.e., about $25 \%$ less than with
212: less conservative constraints. We have considered a rectangular
213: angular fields, with uniform coverage, in the SDSS internal angular
214: coordinates $(\eta,\lambda)$ limited by $-6^{\circ} \le \eta \le
215: 36^\circ$ and $-48^\circ \le \lambda \le 32.5^\circ$.
216:
217: %: We do not use
218: %corrections for the redshift completeness mask or for fiber collision
219: %effects. Completeness varies most near the current survey edges which
220: %are excluded in our samples. Fiber collisions in general do not
221: %present a problem for measurements of large scale galaxy correlations
222: %\cite{strauss2002}.
223: %
224: To construct volume limited (VL) samples that are unbiased for the
225: selection effect related to the cuts in the apparent magnitude, we
226: have applied a standard procedure \cite{zehavietal02}. Firstly we
227: compute metric distances $R(z)$ using the standard cosmological
228: parameters $\Omega_M=0.3$ and $\Omega_\Lambda=0.7$.
229: %(We have checked
230: %that our results do not depend sensibly on the choice of cosmological
231: %parameters as the redshif is small).
232: %: this is expected since the redshift involved in these
233: %analysis is limited at $z \le 0.3$ and relativistic redshift-distance
234: %corrections are generally linear in the redshift).
235: Secondly the
236: galaxy absolute magnitude is determined to be $ M_r = m_r - 5 \cdot
237: \log_{10}\left[R(z) \cdot (1+z)\right] - K_r(z) - 25\;, $ where
238: $K_r(z)$ is the K-correction. We determine the $K_r(z)$ term from the
239: NYU VACG data \cite{nyu_vacg}. Finally, we have considered two
240: different VL samples (named VL1 and VL2) defined by two chosen limits
241: in absolute magnitude and metric distance: for VL1 $R\in [100,300]$
242: Mpc/h, $M\in [-22,-20]$ and for VL2 $R\in[200,600]$ Mpc/h,
243: $M\in[-23,-21.5]$. The number of galaxies is about $4\cdot 10^4$ in
244: VL1 and $3 \cdot 10^4$ in VL2. Different cuts in absolute magnitude
245: do not introduce substantial differences in the results presented in
246: this paper. When more conservative and stringent limits in apparent
247: magnitude are applied we find that in the samples with the same limits
248: in distance, there are up to three times less galaxies. However the
249: main results presented in this paper are affected only in the fact
250: that statistics is less robust.
251:
252: The Millennium project \cite{springel05} has performed several
253: cosmological simulations of standard theoretical models. Amount of
254: dark matter and cosmological parameters are given in agreement with
255: standard models. The dark matter simulations have about $10^{10}$
256: particles. From these galaxies are identified according to
257: semi-analytics models of galaxy formation \cite{cronton06}. We have
258: cut a sample with exactly the same geometry as the SDSS VL1 sample and
259: a sample close to the geometrical parameters of the SDSS VL2 applying
260: the same absolute magnitude limits in $r$-filter as for the SDSS data.
261: In the SDSS we use a redshift space analysis while in mock catalogs a
262: real space one. The difference between the real and redshift space
263: analysis is relevant for very small scales, i.e., $r<5$ Mpc/h and thus
264: does not influence results on scales of the order of $100$ Mpc/h
265: \cite{bt05,paper_2df_prl,paper_2df_aea}.
266:
267: \section{Statistical methods}
268:
269:
270:
271: Statistical properties are determined by making averages over the
272: whole sample volume~\cite{book}. In doing so one implicitly assumes
273: that a certain quantity measured in different regions of the sample
274: is statistically stable, i.e., that fluctuations in different
275: sub-regions are described by the same probability density function
276: (PDF). However it may happen that measurements in different
277: sub-regions show systematic differences, which
278: depend, for instance, on the spatial position of the specific
279: sub-regions. In this case the considered statistic is not
280: statistically stationary in space, the PDF systematically differs in
281: different sub-regions and its whole-sample average value is
282: not a meaningful descriptor \cite{book}.
283:
284:
285:
286: In general such systematic differences may be related to two different
287: possibilities: (i) that the underlying distribution is not
288: translationally and/or rotationally invariant; (ii) that the volumes
289: considered are not large enough for fluctuations to be self-averaging
290: \cite{Aharony}.
291: On very general grounds, we expect the galaxy distribution to satisfy
292: the condition of {\it statistical} stationarity in space to avoid
293: special points or directions \cite{pee80,book}. Hence the question we
294: face in a finite volume analysis concerns whether it is large enough
295: to obtain statistically stable results. Note that stationary
296: stochastic distributions satisfy the condition of spatial statistical
297: isotropy and homogeneity also when they have zero average density in
298: the infinite volume limit \cite{book}. This condition is called the
299: Conditional Cosmological Principle \cite{book} to differentiate it
300: from the stronger Cosmological Principle which requires exact
301: homogeneity and deterministic rotational and translation invariance
302: \cite{pee80,book}.
303:
304:
305: For the case of galaxy surveys there is an intrinsic preferred
306: direction which is set by the radial position from the observer, i.e.,
307: the Earth. It is thus necessary to show that statistical quantities
308: do not depend on the radial distance from us. To this aim, in a given
309: sample, a simple approach is to determine the number $N(r;R)$ of
310: galaxies in spheres of radius $r$, centered on a galaxy whose distance
311: from the origin is $R$: we call it the scale-length (SL) analysis. As
312: we discuss below, this is found to be very efficient in mapping large
313: scale structures which appear as large fluctuations of $N(r;R)$. For
314: instance by studying it in various angular slices of the SDSS samples
315: we identify a giant filament covering, in the largest contiguous
316: angular area of the survey, more than 400 Mpc/h at $R\sim 500$ Mpc/h.
317: In different sky directions the SL analysis reveals a variety of
318: structures, showing that large density fluctuations are quite typical.
319:
320:
321: Averaged over the whole sample the quantity $N(r;R)$ gives an estimate
322: of the average conditional number of galaxies in spheres of radius
323: $r$. An estimator making the weakest a-priori-assumptions about the
324: properties of the distribution outside the sample volume is
325: \cite{pie87,slmp98,book}
326: \be
327: \label{eq1}
328: \overline{N(r)} = \frac{1}{M(r)} \sum_{i=1}^{M(r)} N_i(r) \;, \ee
329: where $N_i(r)$ is the number of galaxies seen by the $i^{th}$
330: center-point and the number of centers $M(r)$ varies with $r$ because
331: only those galaxies for which the sphere is fully included in the
332: sample volume are considered as centers \cite{book}. Even in this case,
333: there is an intrinsic selection effect related to the geometry of the
334: samples, which are portions of spheres: when $r$ is large only a part
335: of the sample is explored by the volume average. Hence for large
336: sphere radii $M(r)$ decreases and the location of the galaxies
337: contributing to the average in Eq.\ref{eq1} is mostly at radial
338: distance $\sim [R_{min}+r$, $R_{max}-r]$ from the radial boundaries of
339: the sample at $[R_{min}$, $R_{max}]$.
340:
341:
342:
343: When Eq.\ref{eq1} scales as $ \overline{N(r)} \sim r^D$ and $D=3$ the
344: distribution is homogeneous, while for $D<3$ it is fractal
345: \cite{gsl00,book}. Furthermore fluctuations
346: $
347: \delta^2(r) = [\overline{N(r)^2} - \overline{N(r)}^2] / \overline{N(r)}^2
348: $
349: are small for a homogeneous distribution with any kind of
350: small-amplitude correlations ($\delta^2(r) \ll 1$) and large for a
351: fractal one ($\delta^2(r) \sim 1$) \cite{gsl00,book}. To study
352: fluctuations we determine the PDF of $N_i(r)$, which is expected to
353: converge to a Gaussian when $r \gg \lambda_0$ \cite{book}.
354:
355: \section{Results}
356: Let us now consider the VL1 sample. Here the SL analysis (Fig.\ref{figure2})
357: detects large density fluctuations without a clear radial-distance
358: dependent trend. Correspondingly the PDF has a regular shape
359: characterized by a peak with a long $N$ tail and it is sufficiently
360: statistically stable in different non-overlapping sub-samples of equal
361: volume. This occurs except for the largest sphere radii, i.e., for
362: $r>$30 Mpc/h, for which the number of independent centers becomes too
363: small.
364:
365: \begin{figure}
366: \onefigure[scale=0.32]{Fig1.eps}
367: \caption{{\it Left panels}: From top to bottom the SL analysis for the
368: SDSS sample VL1, with $r=10,20,40$ Mpc/h.
369: %The small dots correspond
370: % to an uncorrelated distribution (Poisson) in the same volume and
371: % with same number of points.
372: {\it Right panels}: Probability density
373: function of $N(r;R)$ in two non-overlapping sub-samples with equal
374: volume (each half of the sample volume) at small and large $R$.
375: While for $r=10,20$ Mpc/h the PDF (nearby sub-sample solid line and
376: faraway sub-sample dotted line) is reasonably statistically stable,
377: for $r=40$ Mpc/h there is a clear difference. The dashed line
378: corresponds to the Poisson distribution: a Gaussian function gives
379: very good fits for all $r$. }
380: \label{figure2}
381: \end{figure}
382:
383:
384: In the deeper VL2 sample we find instead a systematic trend of $N(r;R)
385: $ as a function of $R$ (Fig.\ref{figure3}).
386: \begin{figure}
387: \onefigure[scale=0.32]{Fig2.eps}
388: \caption{
389: The same of Fig.\ref{figure2} for the VL2 sample
390: with $r=20,40,100$ Mpc/h. The fact that the signal for SDSS data
391: becomes smoother when $r$ increases, but still with a systematic
392: radial distance-trend, is due to the fact that many spheres overlap
393: when $r$ growths. While for $r=20$ Mpc/h the PDF
394: is reasonably statistically stable, for $r=40,100$ Mpc/h
395: there is a clear difference. }
396: \label{figure3}
397: \end{figure}
398: Particularly $N(r;R)$, for $R>300$ Mpc/h, grows without any clear
399: saturation, for sphere radii up to $r \sim 100$ Mpc/h. The PDF
400: in two non-overlapping sub-samples of equal volume is found to differ
401: systematically for $r>30$ Mpc/h, and its average value moves as a
402: function of $R$. This shows that fluctuations are not self-averaging
403: at those scales. This behavior is due to the large scale structures
404: at scales $R> 300$ Mpc/h. However at smaller scales, i.e., $r<30$
405: Mpc/h, the PDF in different sub-samples is reasonably statistically
406: stable and similar to the one found in VL1. This shows that, at those
407: scales, fluctuations are self-averaging because the volume average can
408: explore different regions of the sample. Thus these results show
409: that, at the largest scales probed, there are large density
410: fluctuations which are not self-averaging because of the limited
411: sample volume. These determine relative variations larger than unity
412: in the estimation of the average density in spheres of radius $r=100$
413: Mpc/h. We thus conclude that the homogeneity scale must be $\lambda_0
414: > 100$ Mpc/h, the largest sphere radius we considered.
415:
416:
417: Previous analyses of smaller galaxy catalogs, e.g.,
418: \cite{cp92,slmp98,jmsl99,dr4_paper,2df_paper,tikonov,
419: zehavietal02,norberg02,hogg}, considered
420: sample averaged statistics without quantitatively testing whether a
421: significant bias could affect the results (but see
422: \cite{dr4_paper,2df_paper}). For instance the estimator of the most
423: commonly used statistics, the two-point correlation function
424: \cite{tk69}, can be written as \cite{book}
425: \be
426: \label{xi}
427: \xi(r) +1 = \frac{\overline{N(r,\Delta r)}} {V(r,\Delta r)} \cdot
428: \frac{V} {N} \;.
429: \ee
430: The first ratio in the r.h.s. of Eq.\ref{xi} is
431: the average conditional density, i.e., the number of galaxies in shells
432: of thickness $\Delta r$ averaged over the whole sample, divided by the
433: volume $V(r,\Delta r)$ of the shell. The second ratio in the r.h.s. of
434: Eq.\ref{xi} is the average density estimated in a sample containing $N$
435: galaxies and with volume $V$. When measuring this function we
436: implicitly assume, in a given sample, that: (i) fluctuations are
437: self-averaging in different sub-volumes \cite{book} (ii) the linear
438: dimension of the sample volume is $V^{1/3} \gg \lambda_0$
439: \cite{pie87,book}, i.e., the distribution has reached homogeneity
440: inside the sample volume. When the latter condition is not verified
441: the $\xi(r)$ analysis is biased by
442: systematic finite size effects even if
443: fluctuations are self-averaging
444: \cite{pie87,book}.
445: To show how non self-averaging fluctuations inside a given sample
446: bias the $\xi(r)$ analysis, we consider the estimator
447: \be
448: \label{xi2}
449: \xi(r;R,\Delta R) +1 = \frac{\overline{N(r,\Delta r)}} {V(r, \Delta
450: r)} \cdot \frac{V(r^*)}{\overline{N(r^*;R,\Delta R)}} \,,
451: \ee where
452: the second ratio on the r.h.s. is the density of points
453: in spheres of radius $r^*$ averaged over the centers
454: lying in a shell of thickness $\Delta R$
455: around the radial distance $R$. If the distribution is homogeneous,
456: i.e., $r^*>\lambda_0$, and statistically stationary, Eq.\ref{xi2}
457: should be statistically independent on the range of radial distances
458: $(R,\Delta R)$ considered. For instance we consider, in the VL2 sample, $\Delta
459: R=40$ Mpc/h and $R = 240$ Mpc/h or $R=520$ Mpc/h, with $r^* > 50$
460: Mpc/h. We thus find large variations in the amplitude of $\xi(r)$
461: (Fig.\ref{figure4}). This is simply an artifact generated by the
462: large density fluctuations on scales of the order of the sample sizes. The
463: results that the estimator Eq.\ref{xi} (or others based on
464: pair-counting \cite{book,cdm_theo}) has nearly the same amplitude in
465: different samples,
466: e.g.,
467: \cite{dp83,park,benoist,zehavietal02,norberg02},
468: despite the large fluctuations of $N(r;R)$, are
469: simply explained by the fact that $\xi(r)$ is a ratio between the
470: local conditional density and the sample average density:
471: both vary in the same way when the radial distance is
472: changed and thus the amplitude is nearly constant.
473:
474: \begin{figure}
475: \onefigure[scale=0.32]{Fig3.eps}
476: \caption{{\it Upper Panel}: The sample average conditional number of
477: galaxies (Eq.\ref{eq1}) for the SDSS VL1 and VL2 samples up to
478: $r<30$ Mpc/h. The best fit slope (soldi line) gives $D= 2.1 \pm
479: 0.1$. The difference in amplitude between the two samples is simply
480: ascribed to a luminosity selection effect \cite{book}, as VL1
481: contains fainter galaxies than VL2. The amplitude of the mock VL1
482: and VL2 samples has been rescaled by the same arbitrary factor for
483: seek of clarity. {\it Bottom Panel}: Standard two-point correlation
484: function in the VL2 sample estimated by Eq.\ref{xi2}: the sample
485: average density is computed in spheres of radius $r^*$ and
486: considering all center-points lying in a bin of thickness $\Delta
487: R=40$ Mpc/h centered at different radial distance $R$. The case
488: $r^*=80$ Mpc/h and $R=520$ Mpc/h gives the average over the whole
489: sample, i.e., Eq.\ref{xi}, and it coincides with the estimation of
490: \cite{zehavietal02} in a similar sample. The flat tail of $\xi(r)$,
491: for $r^*=40,80$ Mpc/h and $R=300$ Mpc/h, reflects the
492: inhomogeneities with strong correlations at large scales. The stars
493: and triangles correspond to the behavior of the mock VL1 and VL2
494: samples: in this case the amplitude is statistically stable and thus
495: meaningful. The small amplitude difference in this case is ascribed
496: to the different selection in luminosity \cite{springel05}. }
497: \label{figure4}
498: \end{figure}
499:
500:
501:
502:
503:
504:
505: %
506:
507: On the other hand Eq.\ref{eq1}, {\it averaged over the volumes where
508: the PDF has a statistically stable shape}, shows in both considered
509: samples a power-law behavior for $r<$ 30 Mpc/h corresponding to a
510: fractal dimension $D= 2.1 \pm 0.1$ in agreement with
511: \cite{slmp98,jmsl99,hogg,2df_paper,dr4_paper} (Fig.\ref{figure4}).
512: Due to the non self-averaging nature of fluctuations at larger scales,
513: i.e., due to limited volumes, we are not able to determine
514: correlations for $r>30$ Mpc/h.
515:
516: %
517:
518: \section{Discussion}
519: According to standard models of cosmological structure formation,
520: gravitational clustering gives rise to non-linear perturbations from
521: homogeneous initial conditions in the early universe \cite{pee80}. If
522: the initial amplitude of fluctuations is normalized to the
523: anisotropies of the Cosmic Microwave Background Radiation (CMBR)
524: \cite{spergel}, then the homogeneity scale is about $\lambda_0^m = 10$
525: Mpc/h \cite{joyce05}, i.e., twice the value at which $\xi(r)=1$
526: \cite{pee80,book}.
527:
528: Indeed in mock galaxy catalogs generated from N-body simulations of
529: standard cosmological models \cite{springel05,cronton06}, $N(r;R)$
530: does not show, for $r >\lambda_0^m$, large fluctuations or systematic
531: trends as a function of $R$ (Fig.\ref{figure5}).
532: \begin{figure}
533: \onefigure[scale=0.32]{Fig4.eps}
534: \caption{ The same of fig.\ref{figure2} for mock sample VL2, with
535: $r=20,40,100$ Mpc/h. Although for $r=20$ Mpc/h fluctuations are
536: still important, they rapidly become small for larger radii. The
537: PDF is statistically stable for all $r$ and $R$. The solid line
538: corresponds to a Gaussian fit. }
539: \label{figure5}
540: \end{figure}
541: Because in these
542: artificial catalogs fluctuations are small and self-averaging,
543: whole-sample averaged statistics are meaningful at all scales. From
544: the $\overline{N(r)}$ analysis we find, differently from the real
545: galaxy data, that $\overline{N(r)} \sim r^{D}$ with $D=1.1 \pm 0.1$
546: for $r<\lambda_0^m$ and $D=3$ for $r>\lambda_0^m$ (Fig.\ref{figure4}).
547: Correspondingly the PDF rapidly converges to a Gaussian for
548: $r>\lambda_0^m$. The $\xi(r)$ function, estimated by Eq.\ref{xi} or
549: Eq.\ref{xi2}, has a statistically stable amplitude. The residual small
550: amplitude difference between the mock VL1 and VL2 samples is ascribed
551: to a different selection in luminosity \cite{springel05}. The
552: amplitude of $\xi(r)$ is equal to unity at $r \approx 6 $ Mpc/h in
553: agreement with \cite{springel05,cronton06}. In addition its shape
554: reasonably agrees with the standard estimation of $\xi(r)$ from galaxy
555: catalogs
556: \cite{dp83,park,benoist,zehavietal02,norberg02}.
557: However, as discussed above, the latter estimation is biased by
558: systematic effects making the agreement fortuitous.
559:
560:
561:
562:
563: %
564:
565: Standard theoretical models predict that, for $r>\lambda_0^m$, the
566: small fluctuations in the early universe are linearly amplified by
567: gravitational clustering. Therefore for $r>\lambda_0^{m}$ the shape of
568: the theoretical $\xi^m(r)$ must be the same as the initial one
569: \cite{pee80}. This is characterized by a length-scale $r_c$, where
570: $\xi^m(r_c)=0$, which is fixed by the physics of the early universe
571: and estimated from CMBR anisotropies to be $r_c \approx$ 100 Mpc/h
572: \cite{spergel,cdm_theo}. For $r>r_c$, $\xi^m(r)$ becomes negative,
573: corresponding to super-homogeneous correlations characterized by the
574: most rapid possible decay of fluctuations \cite{glass,torquato}. This
575: theoretical framework applies to the whole mass distribution, where
576: dark matter is supposed to provide with the main contribution.
577: Galaxies would form on the largest peaks of the density
578: field. Standard models of galaxy formation describe this physical
579: phenomenon as a selection mechanism \cite{kaiser}. This leaves
580: unperturbed the scale $r_c$ and slightly changes $\lambda_0^m$
581: \cite{bias,cdm_theo}. We find $\lambda_0 > 100 $ Mpc/h $\ge
582: r_c \gg \lambda_0^{m}$. This raises a fundamental inconsistency for
583: the relation between galaxy structures and CMBR anisotropies as no
584: physical mechanism is known, which by sampling a super-homogeneous
585: density field transforms it into a strongly inhomogeneous one
586: \cite{bias,book,cdm_theo}.
587: %
588:
589:
590:
591:
592: \section{Conclusion}
593:
594: In summary, by applying the SL analysis to the newest SDSS galaxy
595: samples, we measure large density fluctuations of spatial extension
596: limited by sample sizes. At scales $r<30$ Mpc/h we detect
597: statistically stable fractal correlations with $D=2.1 \pm 0.1$. On
598: larger scales, $r>30$ Mpc/h, we find that the galaxy distribution is
599: strongly inhomogeneous and fluctuations are not self-averaging in the
600: samples considered. This situation is compatible with fractal
601: power-law correlations extending to such length-scales but
602: incompatible with homogeneity at $\lambda_0 \le 100$ Mpc/h. Indeed,
603: in a portion of a fractal, large structures are expected to be present
604: at any scale, fluctuations being self-averaging only if the sample
605: volume is large enough \cite{book}.
606: %
607: These results have important consequences on the theoretical
608: interpretation of the large scale universe, where models, normalized
609: to CMBR anisotropies, predict there is not enough time to form
610: structures with relative density fluctuations larger than unity on
611: scales larger than $\lambda_0^m \approx 10$Mpc/h
612: \cite{pee80,springel05}. This length scale is more than ten times
613: smaller than our lower limit to $\lambda_0$. Indeed the latter is of
614: the order of the scale $r_c$ where theoretical model predict matter
615: distribution to have negative correlations, a situation which is in
616: contrast with the results from the data analyzed here. Thus the large
617: scale inhomogeneities detected in the SDSS samples are incompatible
618: with the predictions of standard theoretical models relating the early
619: universe physics, with CMBR normalization, to structures in the
620: present universe. Moreover we found that for $r<\lambda_0^m$, mock
621: galaxy catalogs have different correlations from real galaxy data,
622: i.e., $D = 1.1 \pm 0.1$ instead of $D=2.1 \pm 0.1$. Thus structures
623: generated by N-body simulations are intrinsically different from
624: observed ones.
625:
626:
627:
628:
629:
630: Recently three dimensional maps of dark matter distribution from
631: weak lensing observations have been published \cite{massey}.
632: Dark matter is observed to trace the same structures as
633: galaxies. Thus the whole matter
634: distribution would be inhomogeneous on scales larger than $100$ Mpc/h.
635: This has a great impact on the whole
636: theoretical framework for the physical understanding of the large
637: scale universe. For instance it may imply a new type of evolution
638: scenario within an open Friedmann model \cite{pwa} or new types of
639: spatial averaging of the Einstein equations \cite{buchert,wiltshire}
640: which relate the observed inhomogeneities to the apparent acceleration
641: measured from supernovae observations \cite{perl}.
642:
643:
644:
645: The determination of statistical properties
646: of those very large structures, which we detected but could not analyze
647: in detail, should be
648: possible when sample volumes become large enough so that the
649: corresponding fluctuations will be self-averaging on scales larger
650: than the ones studied here. The application of the SL analysis to the
651: forthcoming galaxy redshift surveys, like the complete SDSS
652: \cite{york}, thus represents an important task.
653:
654:
655:
656:
657:
658:
659:
660:
661: \acknowledgments We warmly thank M. Joyce for his careful reading of
662: the manuscript and the very detailed comments and suggestions. We
663: thank A. Gabrielli, R. Durrer, M. Lopez Correidoira for useful remarks
664: and discussions. We acknowledge the use of the Sloan Digital Sky
665: Survey data \cite{paperdr6} and of the Millennium run semi-analytic
666: galaxy catalog \cite{cronton06}
667:
668: \begin{thebibliography}{0}
669:
670:
671: \bibitem{pee80}
672: Peebles P. J. E., {\it The Large-Scale Structure of the
673: Universe} (Princeton University Press, 1980)
674:
675: %2
676: \bibitem{book}
677: Gabrielli A., Sylos Labini F., Joyce M., Pietronero L., {\it
678: Statistical Physics for Cosmic Structures}
679: (Springer, 2004)
680:
681:
682: %3
683: \bibitem{pie87} Pietronero L.,
684: %The fractal structure of the
685: % universe: Correlations of galaxies and clusters and the average mass
686: % density,
687: {\it Physica A}, {\bf 144}, 257 (1987)
688:
689: %4
690: \bibitem{slmp98} Sylos
691: Labini, F., Montuori, M., Pietronero, L.,
692: %Scale invariance of galaxy clustering,
693: %{\it Physics reports}, {\bf 293}, 61-226 (1998)
694: {\it Phys. Rep.}, {\bf 293}, 61 (1998)
695:
696: %5
697: \bibitem{cp92} Coleman P., Pietronero L.,
698: % The fractal structure of the universe,
699: %{\it Physics Reports} , {\bf 213}, 311-389 (1992)
700: {\it Phys.Rep.} , {\bf 213}, 311 (1992)
701:
702: %6
703: \bibitem{jmsl99} Joyce, M., Montuori, M., Sylos Labini, F.,
704: %Fractal correlations in the The CfA2-south galaxy redshift survey,
705: %{\it Astrophysical Journal}, {\bf 514}, L5-L9 (1999)
706: %{\it Astrophys.J.},
707: {\it Ap.J.},{\bf 514}, L5 (1999)
708:
709: %7
710: \bibitem{dp83} Davis, M., Peebles, P.J.E.,
711: %A survey of galaxy redshifts. V - The two-point
712: %position and velocity correlations,
713: %{\it Astrophysical Journal}, {\bf 267}, 465-482 (1983)
714: %{\it Astrophys.J.},
715: {\it Ap.J.}, {\bf 267}, 465 (1983)
716:
717: \bibitem{park} Park, C.,
718: et al.,
719: %Vogeley, M. S., Geller, M. J., Huchra, J. P.,
720: %Power spectrum, correlation function, and tests for luminosity bias
721: %in the CfA redshift survey,
722: % {\it The Astrophysical Journal}, {\bf 431}, 569-585 (1994)
723: % {\it Astrophys.J.},
724: {\it Ap.J.}, {\bf 431}, 569 (1994)
725:
726: \bibitem{benoist}
727: Benoist, C., et al.,
728: %Maurogordato, S., da Costa, L. N., Cappi, A., Schaeffer, R.,
729: %Biasing in the Galaxy Distribution,
730: %{\it Astrophysical Journal}, {\bf 472}, 452-459 (1996)
731: %{\it Astrophys.J.},
732: {\it Ap.J.}, {\bf 472}, 452 (1996)
733:
734:
735: \bibitem{esp} Joyce, M.,
736: et al.,
737: %Sylos Labini, F.,
738: %Montuori, M., Pietronero L.,
739: %Comment on the paper 'The ESO Slice
740: %Project galaxy redshift survey V. Evidence for a D=3 sample dimensionality',
741: %{\it Astronomy and Astrophysics},
742: %{\it Astron.Astrophys.},
743: {\it A\&A}, {\bf 344}, 387 (1999)
744:
745: %8
746: \bibitem{wu}
747: Wu, K.K., Lahav, O., Rees, M.,
748: %The large-scale
749: %smoothness of the Universe,
750: {\it Nature},
751: {\bf 397}, 225 (1999)
752:
753: %9
754: \bibitem{davis} Davis, M.,
755: %Is the Universe Homogeneous on Large Scales?,
756: Proc. of the conference {\it Critical Dialogues in Cosmology}
757: ed. N. Turok, p.12
758: (World Scientific, 1997)
759:
760:
761:
762: %10
763: \bibitem{pmsl96} Pietronero L., Montuori M., Sylos Labini F.,
764: %On the Fractal Structure of the Visible Universe,
765: %Proc. of the conference {\it Critical Dialogues in Cosmology},
766: %ed. N. Turok, p.24 (World Scientific, 1997)
767: {\it ibidem} pg.24
768:
769:
770:
771: \bibitem{bt05}
772: Baryshev, Yu., Teerikorpi P.,
773: %Fractal Approach to Large-Scale Galaxy Distribution
774: %{\it Bulletin Special Astrophysical Observatory},
775: {\it Bull.Spec.Astrophys.Obs.},
776: {\bf 59}, 92-154 (2006)
777:
778:
779:
780:
781: %11
782: \bibitem{york} York, D., et al.,
783: % The Sloan Digital Sky Survey: Technical Summary,
784: %{\it Astronomical Journal}, {\bf 120}, 1579-1587 (2000)
785: {\it Astronom.J.}, {\bf 120}, 1579 (2000)
786:
787:
788: %12
789: \bibitem{colless03}
790: Colless M., et al.,
791: %The 2dF Galaxy Redshift Survey: Spectra and Redshifts,
792: % {\it Monthly Notices of the Royal
793: % Astronomical Society}, {\bf 328}, 1039-1063 (2001)
794: %{\it Mon.Not.R.Astron.Soc.},
795: {\it MNRAS}, {\bf 328}, 1039 (2001)
796:
797:
798: %13
799: \bibitem{hogg}
800: Hogg, D.W., et al.,
801: %Eisenstein, D.J., Blanton M.R., Bahcall N.A, Brinkmann,
802: %J., Gunn J.E., Schneider D.P. ,
803: %Cosmic Homogeneity Demonstrated with Luminous Red Galaxies,
804: %{\it Astrophysical Journal},
805: %{\it Astrophys.J.},
806: {\it Ap.J.},
807: {\bf 624}, 54 (2005)
808:
809: \bibitem{tikonov} Tikhonov, A.V.,
810: %Correlation Properties of Galaxies from
811: %the Main Galaxy Sample of the SDSS Survey,
812: %{\it Astronomical Letters}, {\bf 32}, 721-726(2006)
813: {\it Astron.Lett.}, {\bf 32}, 72 (2006)
814:
815:
816:
817: %15
818: \bibitem{2df_paper} Vasilyev, N.L.,
819: Baryshev, Yu. V., Sylos Labini, F.,
820: % Large-scale correlation properties galaxy correlations
821: %in the 2dF redshift survey,
822: {\it A\&A}, {\bf 447}, 431 (2006)
823:
824:
825: \bibitem%[Sylos Labini et al., 2009a]
826: {paper_2df_prl} Sylos Labini, F.,
827: Vasilyev, N.L., Baryshev, Yu. V.,
828: {\it Europhys.Lett}, {\bf 85}, 29002-p1 (2009)
829:
830:
831: \bibitem%[Sylos Labini et al., 2009b]
832: {paper_2df_aea} Sylos Labini, F.,
833: Vasilyev, N.L., Baryshev, Yu. V.,
834: {\it Astron.Astrophys.}, {\bf 496}, 7 (2009)
835:
836:
837:
838:
839: %16
840: \bibitem{dr4_paper} Sylos Labini, F., Vasilyev, N.L. Baryshev, Yu.V.,
841: %Power law correlations in galaxy distribution and finite volume
842: % effects from the Sloan Digital Sky Survey Data Release Four,
843: %{\it Astronomy and Astrophysics}, {\bf 465}, 23-33 (2007)
844: %{\it Astron.Astrophys.},
845: {\it A\&A} {\bf 465}, 23 (2007)
846:
847:
848: %14
849: \bibitem{joyce05} Joyce, M., et al.,
850: %Sylos Labini, F., Gabrielli, A.,
851: %Montuori, M., Pietronero, L.,
852: %Basic properties of galaxy
853: %clustering in the light of recent results from the Sloan Digital Sky Survey,
854: {\it A\&A},
855: %{\it Astron.Astrophys.},
856: {\bf 443}, 11 (2005)
857:
858:
859:
860: \bibitem{gsl00}
861: Gabrielli, A., Sylos Labini, F.,
862: %Fluctuations in
863: % galaxy counts: a new test for homogeneity versus fractality,
864: {\it Ep.L.} , {\bf 54},
865: 286 (2001)
866:
867: \bibitem{kerscher98}
868: Kerscher, M., et al.,
869: %Schmalzing, J., Buchert, T. and
870: % Wagner, H.,
871: % Fluctuations in the IRAS 1.2 Jy Catalogue,
872: %{\it Astron.Astrophys.}, \
873: {\it A\&A}, {\bf 333}, 1 (1998)
874:
875: \bibitem{picard91}
876: Picard A.,
877: %Inhomogeneities in the universe on scales of (125/h Mpc)$^3$,
878: {\it A.J.}, {\bf 102}, 445-453 (1991)
879:
880: \bibitem{bd97}
881: Bertin E. and Dennefeld M.,
882: %Galaxy evolution at low redshift? I. Optical counts,
883: {\it A.\&A.}, {\bf 317}, 43
884: (1997)
885:
886: \bibitem{busswell03}
887: Busswell, G.S. , et al.,
888: % The Local Hole in the Galaxy
889: % Distribution: New Optical Evidence,
890: {\it MNRAS},
891: %{\it Mon.Not.R.Astron.Soc.},
892: {\bf 354}, 991 (2004)
893:
894: \bibitem{frith03}
895: Frith, W.J. et al.,
896: %G.S. Busswell, R. Fong, N. Metcalfe, T. Shanks,
897: %The Local Hole in the Galaxy Distribution: Evidence from 2MASS,
898: % {\it Monthly Notices of the Royal
899: % Astronomical Society},
900: %{\it Mon.Not.R.Astron.Soc.}
901: {\it MNRAS},
902: {\bf 345}, 1049 (2003)
903:
904:
905:
906: \bibitem{paperdr6} Adelman-McCarthy, J. K. , et al.,
907: % The Sixth Data Release of the Sloan Digital Sky Survey,
908: % {\it Astrophys.J.Suppl.},
909: {\it Ap.J.Suppl.},
910: {\bf 75}, 297, (2008)
911:
912: \bibitem{strauss2002} Strauss, M.A., et al.,
913: % Spectroscopic Target
914: % Selection in the Sloan Digital Sky Survey: The Main Galaxy Sample,
915: {\it A.J. },
916: {\bf 124}, 1810 (2002)
917:
918:
919: \bibitem{nyu_vacg} Blanton, M.R., Roweis, S., {\it Astron.J.}, {\bf
920: 133}, 734 (2007)
921:
922:
923:
924: \bibitem{zehavietal02}
925: Zehavi, I., et al.,
926: %Galaxy Clustering in Early Sloan Digital Sky Survey Redshift Data,
927: {\it Ap.J.}, {\bf 571}, 172 (2002)
928:
929: %\bibitem{norberg01} Norberg E., et al.,
930: %The
931: % 2dF Galaxy Redshift Survey: luminosity dependence of galaxy
932: % clustering,
933: %{\it Mon.Not.R.Astron.Soc.}
934: % {\it Monthly Notices of the Royal
935: % Astronomical Society},
936: %{\bf 328}, 64 (2001)
937:
938: \bibitem{norberg02}
939: Norberg E., et al.,
940: %The 2dF Galaxy Redshift Survey: the dependence of galaxy clustering on
941: %luminosity and spectral type,
942: %{\it Mon.Not.R.Astron.Soc.}
943: % {\it Monthly Notices of the Royal
944: % Astronomical Society},
945: {\it MNRAS}, {\bf 332}, 827, (2002)
946:
947: \bibitem {tk69} Totsuji, H., Kihara, T.,
948: %The Correlation Function for
949: % the Distribution of Galaxies,
950: %{\it Publ.Astron.Soc.Jap.},
951: {\it PASJ}, {\bf 21}, 221 (1969)
952:
953:
954:
955: \bibitem{cdm_theo}
956: Sylos Labini, F., Vasilyev, N.L.,
957: %Extension and estimation of correlations in Cold Dark Matter models,
958: %{\it Astron.Astrophys.},
959: {\it A\&A}, {\bf 477} 381-395 (2008)
960:
961:
962: \bibitem{spergel}
963: Spergel, D.N., et al.,
964: %Three-Year Wilkinson Microwave Anisotropy Probe (WMAP)
965: %Observations: Implications for Cosmology,
966: %{\it Astrophysical Journal Supplement},
967: {\it Ap.J.Suppl.}, {\bf 170}, 377 (2007)
968:
969:
970: \bibitem{springel05} Springel, V., et al.,
971: %Simulations of the
972: % formation, evolution and clustering of galaxies and quasars,
973: {\it Nature}, {\bf 435}, 62 (2005)
974:
975:
976: \bibitem{cronton06} Croton, D.J., et al.,
977: %The many lives of
978: %active galactic nuclei: cooling flows, black holes and the
979: %luminosities and colours of galaxies,
980: %{\it Mon.Not.R.Astron.Soc.}
981: %{\it Monthly Notices Royal Astronomical Society},
982: {\it MNARS}, {\bf 365}, 11 (2006)
983:
984:
985: \bibitem{Aharony} Aharony, A., Harris, B., {\it Phys. Rev. Lett}, {\bf
986: 77}, 3700 (1996)
987:
988:
989: \bibitem {glass} Gabrielli, A., Joyce, M., Sylos Labini, F.,
990: % The
991: % Glass-like Universe: Real-space correlation properties of standard
992: % cosmological models,
993: {\it Phys.Rev.}, {\bf D65},
994: 083523 (2002)
995:
996: \bibitem{torquato} Torquato, S., Stillinger, F.H.,
997: %Local density
998: %fluctuations, hyperuniformity, and order metrics,
999: {\it Phys.Rev.}, {\bf E68}, 041113 (2003)
1000:
1001: \bibitem{bias}
1002: Durrer, R., et al.,
1003: %Gabrielli, A., Joyce, M., Sylos Labini, F.,
1004: %Bias and the power spectrum beyond the
1005: %turn-over,
1006: % {\it Astrophys.J.},
1007: {\it Ap.J.}, {\bf 585}, L1 (2003)
1008:
1009: \bibitem{kaiser}
1010: Kaiser, N.,
1011: %On the spatial correlations of Abell clusters,
1012: {\it Ap.J.} {\bf 284}, L9 (1984)
1013:
1014:
1015: \bibitem{massey}
1016: Massey, R., et al.,
1017: %Dark matter maps reveal cosmic scaffolding,
1018: {\it Nature}, {\bf 445}, 286 (2007)
1019:
1020:
1021:
1022: \bibitem{pwa}
1023: Joyce, M., et al.
1024: %Anderson, P.W., Montuori, M., Pietronero, L., Sylos Labini, F.,
1025: % Fractal Cosmology in an Open Universe,
1026: {\it EpL}, {\bf 50}, 416 (2000)
1027:
1028:
1029: \bibitem{buchert} Buchert, T,
1030: % On Average Properties of Inhomogeneous
1031: % Fluids in General Relativity: Dust Cosmologies,
1032: {\it Gen.Rel.Grav.}, {\bf 32}, 105 (2000)
1033:
1034:
1035: \bibitem{wiltshire}
1036: Wiltshire, D.L.,
1037: %Exact solution to the averaging problem in cosmology,
1038: {\it Phys.Rev.Lett.},
1039: {\bf 99}, 251101 (2007)
1040:
1041: \bibitem{perl}
1042: Perlmutter, S., et al.,
1043: % Measurements of Omega and Lambda from 42 High-Redshift Supernovae,
1044: %{\it Astrophys.J.},
1045: {\it Ap.J.}, {\bf 517}, 565 (1999)
1046:
1047:
1048:
1049:
1050: %\bibitem{nyu_vacg}
1051: %http://ssds.physics.nyu.edu/vagc-dr6/vagc0/%kcorret.non.petro.z0.00.fits
1052:
1053: %\bibitem{cronton06web}
1054:
1055: \end{thebibliography}
1056:
1057: \end{document}
1058:
1059: