0707.3232/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: \begin{document}
3: \title{Comparison between the Blue and the Red Galaxy Alignments detected 
4: in the Sloan Digital Sky Survey}
5: \author{Jounghun Lee}
6: \affil{Department of Physics and Astronomy, FPRD, Seoul National University, 
7: Seoul 151-747, Korea} 
8: \email{jounghun@astro.snu.ac.kr}
9: \author{Ue-Li Pen}
10: \affil{Canadian Institute for Theoretical Astrophysics, Toronto, ON M5S 3H8, 
11: Canada} 
12: \email{pen@cita.utoronto.ca}
13: 
14: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
15: \begin{abstract}
16: We measure the intrinsic alignments of the blue and the red galaxies 
17: separately by analyzing the spectroscopic data of the Sloan Digital Sky Survey 
18: Data Release 6 (SDSS DR6). For both samples of the red and the blue galaxies 
19: with axial ratios of $b/a\le 0.8$, we detect a $3\sigma$ signal of the 
20: ellipticity  correlation in the redshift range of $0\le z\le 0.4$ for $r$-band 
21: absolute (model) magnitude cut of $M_{r}\le -19.2$ (no $K$ correction). 
22: We note a difference in the strength and the distance scale for the red and 
23: the blue galaxy correlation $\eta_{2D}(r)$: For the bright blue galaxies, 
24: it behaves as a quadratic scaling of the linear density correlation of 
25: $\xi(r)$ as $\eta_{2D}(r)\propto\xi^{2}(r)$ with strong signal detected only 
26: at small distance bin of $r\le 3h^{-1}$Mpc. 
27: While for the bright red galaxies it follows a linear scaling as 
28: $\eta_{2D}(r)\propto\xi(r)$ with signals detected at larger distance out to 
29: $r\sim 6h^{-1}$Mpc. We also test whether the detected correlation signal is 
30: intrinsic or spurious by quantifying the systematic error and find that the 
31: effect of the systematic error on the ellipticity correlation is negligible. 
32: It is finally concluded that our results will be useful for the weak lensing 
33: measurements as well as the understanding of the large scale structure 
34: formation.
35: \end{abstract}
36: \keywords{cosmology:observations --- methods:statistical --- 
37: large-scale structure of universe}
38: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
39: 
40: It has been long suspected that the galaxies may have preferential directions 
41: in the orientations of their elliptical shapes (or spin axes) with being 
42: locally correlated between the neighbors, which is often called the galaxy 
43: intrinsic alignments. Since the galaxy intrinsic alignments could in principle 
44: contaminate weak lensing signals 
45: \citep[e.g.,][and references therein]{hir-etal07}, the measurement of its 
46: existence and strength from observations has been a prior interest especially 
47: of the weak lensing community.  
48:   
49: It has been claimed by \citet{man-etal06} that no signal of ellipticity 
50: correlation between the neighbor galaxies was detected in the spectroscopic 
51: data from the Sloan Digital Sky Survey \citep[SDSS,][]{yor-etal00}, although 
52: they mentioned that their results are statistically consistent with 
53: previous observational evidences for the existence of the galaxy intrinsic 
54: alignments \citep{bro-etal02,hey-etal04}. 
55: Extrapolating their results to redshift $z=1$ relevant for the weak lensing 
56: shear, they have drawn a tentative conclusion that the intrinsic ellipticity 
57: correlations between the neighbor galaxies would not cause any significant 
58: errors for the estimation of the power spectrum amplitude of the initial 
59: density field.
60: 
61: Very recently, however, a numerical analysis based on the high-resolution 
62: N-body simulation has demonstrated with high statistical power that strong 
63: ellipticity correlations between the dark matter halos exist at large 
64: distances out to $10h^{-1}$Mpc \citep{lee-etal07}. Therefore, it is necessary 
65: and important to reexamine observationally the existence and significance 
66: of the galaxy intrinsic alignments by analyzing the SDSS data.
67: 
68: The physical mechanism for the generation of the galaxy intrinsic alignments 
69: differ between the blue and the red galaxies. For the blue spiral galaxies 
70: which are usually rotationally supported, the intrinsic alignments refer to 
71: the correlations of the spin axes originated from the initial tidal field. 
72: The linear tidal torque theory explains that the intrinsic correlation 
73: should exist only locally between close galaxy pairs, whose functional 
74: form is predicted to be a quadratic scaling of the linear density 
75: correlation function \citep{pen-etal00}. 
76: 
77: While for the bright red galaxies which are not flattened by rotation, 
78: the intrinsic alignments refer to the correlations between the major principal 
79: axes of the galaxy's elliptical shapes which are determined in observations 
80: in terms of the anisotropy in their stellar distribution. 
81: It has yet to be fully understood what the origin of the anisotropic steller 
82: distribution of the red galaxies is and how it is related to the shapes of 
83: the host halos. But the standard model based on the cosmic web theory 
84: suggests that it may be due to the anisotropic accretion and infall of 
85: materials and gas into the host halo along the halo's major principal axes 
86: which are in turn elongated with the large scale filamentary matter 
87: distribution \citep{wes94,bon-etal96}. Given this anisotropic infall scenario, 
88: the red galaxies may have built-in memory of the large-scale filamentary 
89: structures and thus their ellipticities might be correlated on larger scales 
90: than the spin axes of the blue galaxies. 
91: 
92: Even for the blue galaxies, it was claimed that the growth of the 
93: non-Gaussianity in the density field would cause large-scale correlations 
94: of their spin axes \citep{hui-zha02}.  To account for the large-scale 
95: intrinsic correlations between the neighbor galaxies, we use the following 
96: formula: 
97: \begin{equation}
98: \label{eqn:3d}
99: \eta_{3D}(r) \approx \frac{1}{6}a^2_{\rm l}\frac{\xi^{2}(r;R)}{\xi^{2}(0;R)} 
100: + \varepsilon_{\rm nl}\frac{\xi(r;R)}{\xi(0;R)},
101: \end{equation}
102: where $r$ is the three dimensional separation distance of a galaxy pair 
103: and $\xi(r;R)$ is the two-point correlation function of the linear density 
104: field smoothed on the Lagrangian galactic scale $R$. 
105: Here, the two correlation parameters $a_{\rm l}$ and $\varepsilon_{\rm nl}$ 
106: represent the strength of the small-scale and the large-scale correlation, 
107: respectively. If $\varepsilon_{\rm nl}=0$ and $a_{\rm l}> 0$,  the galaxy 
108: intrinsic correlation will behave as a quadratic scaling of the density 
109: correlation function so that $\eta_{3D}(r)$ should rapidly diminishes 
110: to zero as $r$ increases, as the linear tidal torque theory predicts. 
111: In contrast, if $\varepsilon_{\rm nl} > 0$ and $a_{\rm l}=0$, the galaxy's 
112: intrinsic correlation will follow a linear scaling of $\xi(r)$ so that 
113: $\eta_{3D}(r)$ should be non-negligible even at large distance.
114: 
115: In fact, we have suggested this formula (eq.[\ref{eqn:3d}]) in our previous 
116: work \citep{lee-pen07} to account for the non-Gaussianity effect on the 
117: galaxy's spin-spin correlations which are tidally induced. 
118: Here we use the same analytic model to describe the intrinsic 
119: ellipticity-ellipticity correlations of the galaxies, assuming that since 
120: the anisotropic infall of materials that resulted in the ellipticities of 
121: the galaxies are also nonlinear manifestation of the tidal field and thus 
122: can be approached with the same analytic techniques.
123: Throughout this Letter, the galaxy intrinsic alignments refer to the 
124: intrinsic ellipticity-ellipticity alignments of the galaxies, unless 
125: otherwise stated. 
126:  
127: In practice, what can be measured is the intrinsic alignments not of the 
128: galaxy's three dimensional shapes but of its two dimensional shapes projected 
129: onto the plane of the sky. According to \citet{lee-pen01},  if the two 
130: dimensional projection effect is taken in to account properly, the 
131: correlation parameter has to be multiplied by a factor of $5/4$. Thus, the 
132: two dimensional projected intrinsic correlation function is modified 
133: from equation (\ref{eqn:3d}) simply as 
134: \begin{equation}
135: \label{eqn:2d}
136: \eta_{2D}(r) \approx 
137: \frac{25}{96}a^2_{\rm l}\frac{\xi^{2}(r;R)}{\xi^{2}(0;R)} + 
138: \frac{5}{4}\varepsilon_{\rm nl}\frac{\xi(r;R)}{\xi(0;R)}.
139: \end{equation}
140: Note here that although $\eta_{2D}(r)$ represents the intrinsic correlations  
141: of the galaxy's two dimensional projected ellipticities, the distance $r$ 
142: represents the three dimensional separation. It is also worth mentioning 
143: here that for the blue galaxies the intrinsic correlations of the galaxy's 
144: two dimensional projected ellipticities are in fact same as that of the 
145: galaxy's two dimensional projected spin axes since the spin axes are 
146: believed to be orthogonal to the galaxy's major axes.
147: 
148: To measure $\eta_{2D}(r)$ from the real universe, we use a dataset of galaxies 
149: at redshift $0\le z \le 0.4$ downloaded from the SDSS DR6 website 
150: (http://www.sdss.org/dr6/). A total of $640647$ galaxies are found to be in 
151: this redshift range. For each galaxy, we obtain information on right ascension 
152: ($\alpha$), declination ($\delta$), redshift ($z$), position angle ($p$), 
153: major-to-minor axis ratios ($b/a$), the $u-g$ and $g-r$ colors, and $r$-band 
154: model magnitude ($M_{r}$). Among $640647$ galaxies, the information on $p$ 
155: and $b/a$ are available only for $639727$ galaxies, from which we also select 
156: only those galaxies with $b/a\le 0.8$ since for the nearly face-on galaxies 
157: with $b/a > 0.8$ the measurements of the direction of their major axes may 
158: suffer from large uncertainty. A total of $434849$ galaxies are found to 
159: satisfy the criterion of $b/a\le 0.8$.
160: 
161: Using the empirical $u-r$ color separator suggested by 
162: \citet{str-etal01} for the SDSS data, we classify the galaxies 
163: into the red ($u-r>2.22$) and the blue ($u-r\le 2.22$) sample. 
164: Then, we select only bright galaxies by applying the absolute magnitude 
165: cut of $M_{r}\le -19.2$. 
166: Since the SDSS calibration uses the inverse hyperbolic sine magnitudes 
167: (asinh) \citep{lup-etal99}, we calculate $M_{r}$ as the asinh magnitudes 
168: using the information on the softening parameter given in the SDSS web site.
169: Finally, we end up having two SDSS samples which consist of a total of 
170: $87188$ blue and $283972$ red galaxies brighter than this magnitude cut 
171: of $M_{r}\le -19.2$. 
172: 
173: For each galaxy, we determine the direction of the major axis, $\hat{\bf e}$, 
174: in the equatorial coordinate system from the given information on $p$, 
175: $\alpha$ and $\delta$: Let us consider two galaxies in a given sample 
176: whose major axis directions are found as 
177: $\hat{\bf e}_{i}$ and $\hat{\bf e}_{j}$. First, we  calculate 
178: the unit projected separation vector as 
179: $\hat{\bf d}\equiv {\bf d}/\vert{\bf d}\vert$ with 
180: $\hat{\bf d}\equiv \hat{\bf r}_{i}-\hat{\bf r}_{j}$, where 
181: $\hat{\bf r}_{i}$ and $\hat{\bf r}_{j}$ represent the unit position 
182: vectors of the $i$-th and the $j$-th galaxy, respectively. The separation 
183: vectors projected onto the plane of the sky at the position of the two 
184: galaxies are given as $\hat{\bf d}_{i}\equiv \hat{\bf d} - 
185: (\hat{\bf d}\cdot\hat{\bf r}_{i})\hat{\bf r}_{i}$ and $\hat{\bf d}_{j}\equiv 
186: \hat{\bf d} - (\hat{\bf d}\cdot\hat{\bf r}_{j})\hat{\bf r}_{j}$, 
187: respectively. Then, the intrinsic correlations of the two dimensional 
188: projected ellipticities between the galaxies with separation of $r$ 
189: is calculated as	
190: \begin{equation}
191: \label{eqn:gam}
192: \eta_{2D}(r) \equiv \sum_{i,j}
193: \cos^{2}\left(\gamma_{i}-\gamma_{j}\right) - \frac{1}{2}, 
194: \qquad {\rm with} \qquad
195: \gamma_{i}\equiv \tan^{-1}\left[ \frac{(\hat{\bf r}_{i}\times 
196: \hat{\bf d}_{i})\cdot\hat{\bf e}_{i}}{\hat{\bf d}_{i}\cdot\hat{\bf e}_{i}}
197: \right].
198: \end{equation}
199: Here, the angle $\gamma_{i}$ represents the direction of $\hat{\bf e}_{i}$ 
200: projected onto the plane of sky at the position of ${\bf r}_{i}$. Note that 
201: for a galaxy pair with small distance, the quantity
202: $\Delta\gamma\equiv \gamma_{i}-\gamma_{j}$ is actually identical to the 
203: difference in the position angle $\Delta p$. To determine the three 
204: dimensional separation distance, $r$, of a galaxy pair,  we assume a flat 
205: $\Lambda$-dominated cosmology with matter density $\Omega_{m}=0.25$, 
206: vacuum energy density $\Omega_{\Lambda}=0.75$ and Hubble constant 
207: $H_{0}=100 h$km s$^{-1}$Mpc$^{-1}$ with $h=1$ \citep{bla-etal03}. 
208: 
209: Then, we measure $\eta_{2D}(r)$ for the selected red and the blue galaxies, 
210: separately. Then, we compare the observational results with the analytic model 
211: (\ref{eqn:2d}), determining the best-fit values of the two parameters 
212: $a_{\rm l}$ and $\varepsilon_{\rm nl}$ with the help of the 
213: $\chi^{2}$-minimization. In the fitting procedure, the values of 
214: the two parameters are confined to $0\le a_{\rm l}\le 0.6$ and 
215: $0\le \varepsilon_{\rm nl}\le 0.1$\citep{lee-pen01,lee-pen07}. 
216: The unmarginalized errors in the fitting parameters are calculated by the 
217: formula given in \citet{bev-rob96}: The error in the parameter 
218: $a_{\rm l}$ is found as 
219: $\sigma_{a}=\Delta a_{\rm l}\sqrt{2(\chi^{2}_{1}-2\chi^{2}_{\rm min} + 
220: \chi^{2}_{2})}$. Here $\Delta a_{\rm l}$ represents a fitting step-size, 
221: $\chi^{2}_{\rm min}$ is the minimum value of $\chi^{2}$, 
222: $\chi^{2}_{1} \equiv \chi^{2}(a_{\rm l0}+\Delta a_{\rm l})$
223: and $\chi^{2}_{1} \equiv \chi^{2}(a_{\rm l0}-\Delta a_{\rm l})$.
224: where $a_{\rm l0}$ is defined as 
225: $\chi^{2}_{\rm min}\equiv \chi^{2}(a_{\rm l0})$. 
226: The unmarginalized error $\sigma_{\varepsilon}$ in the parameter 
227: $\varepsilon_{\rm nl}$ is also calculated in a similar manner.
228: 
229: For the analytic model (eq.[\ref{eqn:2d}]), we use the $\Lambda$CDM 
230: power spectrum given by \citet{bar-etal86}. For the Lagrangian scale $R$ of 
231: each galaxy, we first calculate the luminosity $L$ of each galaxy from its 
232: absolute r-band magnitude and then convert it to the galaxy mass scale 
233: using the relation of $L\sim M^{0.88}$ \citep{val-ost06}. Then, the Lagrangian 
234: galactic scale $R$ is found as a top-hat radius that encloses the mean mass 
235: $\bar{M}$ averaged over the galaxies of each sample.
236: When we calculate the absolute magnitude of each galaxy, no $K$-correction is 
237: considered: If our goal were to investigate how $\eta_{2D}(r)$ changes in 
238: narrow absolute magnitude bin, then it would be necessary to determine 
239: $M_{r}$ accurately by considering the $K$-correction \citep{wak-etal06}. 
240: However, our goal here is rather to measure the average $\eta_{2D}(r)$ 
241: for those galaxies which are brighter than a single magnitude threshold 
242: in the redshift range of $0\le z \le 0.4$. Hence, a maximum $20\%$ error 
243: that may be caused by ignoring $K$-correction in the estimation of the 
244: galaxy absolute magnitude should not be an issue here. 
245: 
246: Fig. \ref{fig:cor} plots the observational results of $\eta_{2D}(r)$ from the 
247: red and the blue galaxy sample as solid dots in the top and the bottom panel, 
248: respectively. In each panel, the solid line represents the analytic model 
249: with the best-fit correlation parameters $a_{\rm l}$ and $\epsilon_{\rm nl}$.
250: The errors are calculated as standard deviation for the case of 
251: no correlation as  $\sigma_{\eta}=1/\sqrt{8n_{\rm p}}$where $n_{\rm p}$ is the 
252: number of galaxy pairs belonging to each bin \citep{pen-etal00}. As can be 
253: seen, clear signals of ellipticity correlation higher than $3\sigma_{\eta}$ 
254: are detected from both the red and the blue galaxy samples at the first 
255: bin ($r \le 3h^{-1}$Mpc). Note also that from the red galaxy sample the 
256: correlation signal is also detected at the second distance bin 
257: ($r \le 3h^{-1}$Mpc). 
258:  
259: Table \ref{tab:cor} lists the best-fit values of $a_{\rm l}$ and 
260: $\epsilon_{\rm nl}$. The results show a marked difference between the blue 
261: and the red galaxy intrinsic alignments: For the blue galaxies, the best-fit 
262: values of the correlation parameters are found to be $a_{\rm l}=0.20\pm 0.04$ 
263: and $\epsilon_{\rm nl}\approx0.$, which indicates that $\eta_{\rm 2D}$ of the 
264: blue galaxies follows a quadratic scaling of $\xi(r)$, as predicted  
265: by the linear tidal torque theory \citep{pen-etal00,lee-erd07}. 
266: In contrast, for the red galaxies, the best-fit values are 
267: $a_{\rm l}\approx 0$ and $\epsilon_{\rm nl}=(2.6\pm 0.5)\times 10^{-3}$, 
268: which implies that   $\eta_{2D}$ of the bright red galaxies follows a linear 
269: scaling of $\xi(r)$.  This observational result is consistent with the 
270: picture that the ellipticities of the blue and the red galaxies are induced 
271: by the initial tidal field and the anisotropic infall of materials, 
272: respectively. 
273: 
274: Before assuring ourselves that the detected signal is real, however, it is 
275: necessary to examine the effect of the systematic error in the measurement 
276: of the position angles of the SDSS galaxies because the systematic error 
277: could cause correlations of galaxy ellipticities that mimic intrinsic 
278: alignments. To quantify the systematic error, we perform two simple tests. 
279: First, we shuffle randomly the redshifts of the selected SDSS galaxies in 
280: each sample and remeasure $\eta_{2D}$ one hundred times. 
281: For each distance bin, we calculate the mean averaged over these $100$ random 
282: realizations and the standard deviation between realizations, assuming that 
283: these $100$ shuffling processes are mutually independent. 
284: If the systematic error had a dominant effect, then a correlation signal would 
285: not disappear even when the redshifts are shuffled. Fig \ref{fig:cor} plots 
286: the mean plus and minus one standard deviation as (green) dashed lines. 
287: As can be seen, when the redshifts of the selected galaxies are shuffled, 
288: the ellipticity correlations between the galaxies disappear, which suggests 
289: that the effect of the systematic error on the galaxy intrinsic alignments 
290: be negligible.
291: 
292: Second, we measure the ellipticity cross-correlations between the red and 
293: the blue galaxies, which is plotted as solid line and compared with the 
294: ellipticity correlation of the red (red dotted) and the blue (blue dashed) 
295: galaxies in Fig. \ref{fig:cross}. If the systematic error caused the 
296: correlation of galaxy ellipticities, then we would find non-negligible 
297: cross-correlations of the ellipticities between the red and the blue galaxies. 
298: As can be seen in Fig. \ref{fig:cross}, however, the cross-correlations 
299: are quite weak at the first distance bin, which reassures us that the 
300: detected ellipticity correlations of the red and the blue galaxies are 
301: not spurious intrinsic.
302: 
303: It is worth discussing the effect of the redshift distortion that we have 
304: ignored in our analysis. For the blue galaxies, its mean peculiar velocity 
305: dispersion has been observed to be quite small less than $150$ km/s on 
306: average \citep{dav-etal97}. Therefore it is well justified to ignore the 
307: redshift distortion effect. On the other hand, for the red galaxies 
308: which are observed to have higher peculiar velocity dispersion around 
309: $200$ km/s \citep{dav-etal03}, the redshift distortion effect may cause 
310: non-negligible degree of scatter in the measurement of the galaxy separation 
311: distance $r$.  Thus, we do not exclude a possibility that the 
312: characteristic distance scale for the red galaxy intrinsic alignments 
313: and the value of $\varepsilon_{\rm nl}$ could be smaller in real space 
314: than observed in redshift space. 
315: 
316: The key implication of our results for the weak lensing survey is also worth 
317: discussing here. According to our results, the degree of the galaxy intrinsic 
318: alignments is quite strong for the bright galaxies at redshifts $z\le 0.4$. 
319: Since the galaxy intrinsic correlations tend to decrease with redshift 
320: \citep{lee-pen07}, this implies that at $z=1$ relevant for the cosmic shear 
321: survey the intrinsic correlations may be larger than the detected signal of 
322: $a_{\rm l}=0.2$ and thus should not be completely negligible. 
323: Our results also suggest that when removing close galaxy pairs as intrinsic 
324: ellipticity contaminants from the cosmic shear analysis 
325: \citep{hey-hea03,tak-whi04}, a typical distance scale of a galaxy 
326: pair has to be determined with care, since the scaling of the intrinsic 
327: correlations depend on the intrinsic property of the galaxies (red or blue). 
328: 
329: We conclude that our first detection of a clear signal of the galaxy 
330: intrinsic correlations from the SDSS data will be useful for the weak 
331: lensing measurement as well as for understanding the evolution and formation 
332: of the red and the blue galaxies.
333: 
334: \acknowledgments
335: 
336: We thank an anonymous referee for his/her constructive report which helped 
337: us improve significantly the original manuscript. We also thank B. M\'enard 
338: for his help in downloading the SDSS data and R. Mandelbaum for useful 
339: comments.J.L. is grateful to the warm hospitality of the Canadian Institute 
340: for Theoretical Astrophysics (CITA) where this project was planned and 
341: carried out. J.L. acknowledges the financial support from the Korea 
342: Science and Engineering Foundation (KOSEF) grant funded by the Korean 
343: Government (MOST, NO. R01-2007-000-10246-0).
344: 
345: Funding for the SDSS and SDSS-II has been provided by the Alfred
346: P. Sloan Foundation, the Participating Institutions, the National
347: Science Foundation, the U.S. Department of Energy, the National
348: Aeronautics and Space Administration, the Japanese Monbukagakusho, the
349: Max Planck Society, and the Higher Education Funding Council for
350: England. The SDSS Web Site is http://www.sdss.org/. 
351: 
352: The SDSS is managed by the Astrophysical Research Consortium for the
353: Participating Institutions. The Participating Institutions are the
354: American Museum of Natural History, Astrophysical Institute Potsdam,
355: University of Basel, University of Cambridge, Case Western Reserve
356: University, University of Chicago, Drexel University, Fermilab, the
357: Institute for Advanced Study, the Japan Participation Group, Johns
358: Hopkins University, the Joint Institute for Nuclear Astrophysics, the
359: Kavli Institute for Particle Astrophysics and Cosmology, the Korean
360: Scientist Group, the Chinese Academy of Sciences (LAMOST), Los Alamos
361: National Laboratory, the Max-Planck-Institute for Astronomy (MPIA),
362: the Max-Planck-Institute for Astrophysics (MPA), New Mexico State
363: University, Ohio State University, University of Pittsburgh,
364: University of Portsmouth, Princeton University, the United States
365: Naval Observatory, and the University of Washington. 
366: 
367: 
368: \begin{thebibliography}{}
369: \bibitem[Bardeen et al.(1986)]{bar-etal86}
370: Bardeen, J. M., Bond, J. R., Kaiser, N., \& Szalay, A. S. 1986, \apj, 304, 15
371: \bibitem[Bevington \& Robinson(1996)]{bev-rob96}
372: Bevington, P. R., \& Robinson, D. K. 1996, Data Reduction and Error 
373: Analysis for the Physical Sciences (Boston : McGraw-Hill)
374: \bibitem[Blanton et al.(2003)]{bla-etal03}
375: Blanton, M. R., et al. 2003, \apj, 592, 819
376: \bibitem[Bond, Kofman, \& Pogosyan(1996)]{bon-etal96}
377: Bond, J., R., Kofman, L., \& Pogosyan, D. 1996, Nature, 380, 603
378: \bibitem[Brown et al.(2002)]{bro-etal02}
379: Brown, M. L., Taylor, A. N., Hambly, N. C., \& Dye, S. 2002, \mnras, 333, 501
380: \bibitem[Catelan et al. (2001)]{cat-etal01}
381: Catelan, P., Kamionkowski, M., \& Blandford, R. D. 2001, \mnras, 320, L7
382: \bibitem[Davis et al.(1997)]{dav-etal97}
383: Davis, M., Miller, A., \& White, S.D.M. 1997, \apj, 490, 63
384: \bibitem[Davis et al.(2003)]{dav-etal03}
385: Davis, A. N., Dragan, H., \& Krauss, L. M. 2003, \mnras, 344, 1029
386: \bibitem[Doroshkevich(1970)]{dor70}
387: Doroshkevich, A. G. 1970, Astrofizika, 6, 581
388: \bibitem[Heymans \& Heavens(2003)]{hey-hea03}
389: Heymans, C. \& Heavens, A. 2003, \mnras, 339, 711
390: \bibitem[Heymans et al.(2004)]{hey-etal04}
391: Heymans, C., Brown, M., Heavens, A., Meisenheimer, K., Taylor, A., \& 
392: Wolf, C. 2004, \mnras, 361, 160
393: \bibitem[Hirata \& Seljak(2004)]{hir-sel04}
394: Hirata, C. M. \& Seljak, U. 2004, \prd, 70, 063526
395: \bibitem[Hirata et al.(2004)]{hir-etal04}
396: Hirata, C. M., et al. 2004, \mnras, 353, 529
397: \bibitem[Hirata et al.(2007)]{hir-etal07}
398: Hirata, C. M., et al. 2007, preprint [astro-ph/0701671] 
399: \bibitem[Hui \& Zhang(2002)]{hui-zha02}
400: Hui, L. \& Zhang Z. 2002, preprint [astro-ph/0205512]
401: \bibitem[Jing(2002)]{jin02}
402: Jing, Y. 2002, \mnras, 335, 89
403: \bibitem[Lee \& Pen(2001)]{lee-pen01}
404: Lee, J. \& Pen, U. L. 2001, \apj, 555, 106
405: \bibitem[Lee \& Pen(2002)]{lee-pen02}
406: Lee, J. \& Pen, U. L.  2002, \apj, 567, 111
407: \bibitem[Lee \& Erdogdu(2007)]{lee-erd07}
408: Lee, J. \& Erdogdu, P. 2007, preprint [arXiv:0707.1611]
409: \bibitem[Lee \& Pen(2007)]{lee-pen07}
410: Lee, J. \& Pen, U. L.  2007, preprint [arXiv:0707.1690]
411: \bibitem[Lee et al.(2007)]{lee-etal07}
412: Lee, J. Springel, V., Pen, U. L., Lemson, G. 2007, preprint [arXiv:0709.1106]
413: \bibitem[Lupton et al.(1999)]{lup-etal99}
414: Lupton, R. H., Gunn, J. E., \& Szalay, A. S. 1999, \apj, 118, 1406
415: \bibitem[Mandelbaum et al.(2006)]{man-etal06}
416: Mandelbaum, R., Hirata, C. M., Ishak, M., Seljak, U., \& Brinkmann, J. 
417: 2006, \mnras, 367, 611
418: \bibitem[Pen et al.(2000)]{pen-etal00}
419: Pen, U. L., Lee, J., \& Seljak, U. 2000, 543, L107
420: \bibitem[Strateva et al.(2001)]{str-etal01}
421: Strateva, I., et al. 2001, \apj, 122, 1861
422: \bibitem[Takada \& White(2004)]{tak-whi04}
423: Takada, M., \& White, S. D. M. 2004, \apj, 601, L1
424: \bibitem[Vale \& Ostriker(2006)]{val-ost06}
425: Vale, A., Ostriker, J. P. 2006, \mnras, 371, 1173
426: \bibitem[York et al.(2000)]{yor-etal00}
427: York, D. G., et al. 2000, \mnras, 372, 537
428: \bibitem[Wake et al.(2006)]{wak-etal06}
429: Wake, D. A., et al. 2006, \mnras, 372, 537
430: \bibitem[West(1994)]{wes94}
431: West, M. J. 1994, \mnras, 268, 79
432: \bibitem[White(1984)]{whi84}
433: White, S. D. M. 1984, \apj, 286, 38
434: 
435: \end{thebibliography}
436: 
437: 
438: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
439: \clearpage
440: \begin{deluxetable}{cccccc}
441: \tablewidth{0pt}
442: \setlength{\tabcolsep}{5mm}
443: \tablehead{
444: Color & $z$ & $N_{g}$ & $M_{r}$ &  $a_{\rm l}\times 10^{1}$ & 
445: $\varepsilon_{\rm nl}\times 10^{3}$ \\ 
446: &&&(no $K$-correlation)&&} 
447: \tablecaption{The redshift range ($z$), a total number of galaxies 
448: ($N_{g}$), the range of the absolute magnitude ($M_{r}$), the 
449: best-fit-values of the linear and the nonlinear correlation parameters.}
450: \startdata
451: Red & $[0.0,0.4]$ & $283972$ & $\le -19.2$ & $0.0\pm 3.1$ & $2.6\pm 0.5$ \\ 
452: Blue & $[0.0,0.4]$ & $87188$ & $\le -19.2$ & $2.0\pm 0.4$ & $0.0\pm 1.6$ \\ 
453: \enddata
454: \label{tab:cor}
455: \end{deluxetable}
456: %\end{tabular}
457: %\end{table}
458: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
459: 
460: 
461: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
462: \clearpage
463: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
464:  \begin{figure}
465:   \begin{center}
466:    \plotone{f1.eps}
467: \caption{Correlations of the two dimensional projected ellipticities from 
468: the SDSS red and the blue galaxies (top and bottom, respectively) with the 
469: absolute magnitude cut of $M_{r}\le -19.2$ in the redshift range of 
470: $0\le z\le 0.4$.  In each panel, the solid dots represent the observed 
471: signals, while the solid line corresponds to the analytic model with the 
472: best-fit parameters  (eq.[\ref{eqn:2d}]) found through 
473: $\chi^{2}$-minimization. The dashed line corresponds to the correlation 
474: from the data with $z$-shuffled. For the observed signals, only those 
475: galaxies with major-to-minor axis ratio of $b/a\le 0.8$ are considered.}
476: \label{fig:cor}
477:  \end{center}
478: \end{figure}
479: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
480: \clearpage
481: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
482:  \begin{figure}
483:   \begin{center}
484:    \plotone{f2.eps}
485: \caption{Cross-correlations of the two dimensional projected ellipticities 
486: between the SDSS red and the blue galaxies (solid).}
487: \label{fig:cross}
488:  \end{center}
489: \end{figure}
490: 
491: \end{document}
492: