1: \documentclass[usenatbib,a4paper]{mn2e}

2: 

3: \setlength{\topmargin}{-1.2cm}

4: 

5: %\usepackage{subfigure}

6: \usepackage{graphicx}

7: \usepackage{natbib}

8: \newif\ifAMStwofonts

9: %\AMStwofontstrue

10: \usepackage{dcolumn}

11: \usepackage{bm}

12: \usepackage{amsmath}    

13: \usepackage{amsfonts}

14: \usepackage{amssymb}

15: \setcounter{MaxMatrixCols}{30}

16: 

17: \sloppy

18: 

19: \newcommand{\ra}{\;\raise1.0pt\hbox{$'$}\hskip-6pt\partial\;}

20: \newcommand{\lo}{\;\overline{\raise1.0pt\hbox{$'$}\hskip-6pt\partial}\;}

21: \newcommand{\met}{\mbox{$\not\!\!E_{T}$}}

22: \newcommand{\phinu}{\phi^{\nu}}

23: \newcommand{\phiWB}{\phi^{\nu}_{\text{WB}}}

24: \newcommand{\postscript}[2]{\setlength{\epsfxsize}{#2\hsize}

25:    \centerline{\epsfbox{#1}}}

26: \newcommand{\mweak}{M_{\text{Weak}}}

27: \newcommand{\maux}{M_{\text{aux}}}

28: \newcommand{\msusy}{M_{\text{SUSY}}}

29: \newcommand{\mgut}{M_{\text{GUT}}}

30: \newcommand{\mplanck}{M_{\text{Pl}}}

31: \newcommand{\mstar}{M_{\ast}}

32: \newcommand{\md}{M_D}

33: \newcommand{\mbh}{M_{\text{BH}}}

34: \newcommand{\mbhmin}{M_{\text{BH}}^{\text{min}}}

35: \newcommand{\sbh}{S_{\text{BH}}}

36: \newcommand{\rbh}{R_{\text{BH}}}

37: \newcommand{\xmin}{x_{\text{min}}}

38: \newcommand{\ifb}{\text{fb}^{-1}}

39: \newcommand{\ev}{\text{eV}}

40: \newcommand{\mev}{\text{MeV}}

41: \newcommand{\gev}{\text{GeV}}

42: \newcommand{\tev}{\text{TeV}}

43: \newcommand{\pb}{\text{pb}}

44: \newcommand{\cm}{\text{cm}}

45: \newcommand{\km}{\text{km}}

46: \newcommand{\g}{\text{g}}

47: \newcommand{\s}{\text{s}}

48: \newcommand{\ns}{\text{ns}}

49: \newcommand{\yr}{\text{yr}}

50: \newcommand{\sr}{\text{sr}}

51: \newcommand{\cmwe}{\text{cmwe}}

52: \newcommand{\kmwe}{\text{kmwe}}

53: \newcommand{\xmax}{X_{\text{max}}}

54: \newcommand{\etal}{{\em et al.}}

55: \newcommand{\eg}{{\em e.g.}}

56: \newcommand{\ibid}{{\em ibid.}}

57: \newcommand{\Enu}{E_{\nu}}

58: \newcommand{\Tbh}{T_{\text{BH}}}

59: 

60: \def\lm{{lm}}

61: \def\elm{{E,lm}}

62: \def\blm{{B,lm}}

63: \def\Yo{{Y}}

64: \def\av#1{\langle #1\rangle}

65: \def\neg{N_{\mu^\pm}}

66: \def\bn{N-\neg}

67: \def\mah{M_{\rm A.H.}}

68: \def\d{\ d}

69: \def\sh{\sqrt{\hat s}}

70: \def\beqra{\begin{eqnarray}}

71: \def\eeqra{\end{eqnarray}}

72: \def\beq{\begin{equation}}

73: \def\eeq{\end{equation}}

74: \def\ds{\displaystyle}

75: \def\ts{\textstyle}

76: \def\ss{\scriptstyle}

77: \def\sss{\scriptscriptstyle}

78: \def\Vb{\bar{V}}

79: \def\phb{\bar{\phi}}

80: \def\rhb{\bar{\rho}}

81: \def\L{\Lambda}

82: \def\vp{\varphi}

83: \def\vt{\vartheta}

84: \def\T{\Theta}

85: \def\re#1{(\ref{#1})}

86: \def\D{\nabla^2}

87: \def\N{\nabla}

88: \def\G{\Gamma}

89: \def\p{\partial}

90: \def\half{\mbox{\small$\frac{1}{2}$}}

91: \def\de{\delta}

92: \def\rM{\rho_\chi}

93: \def\rr{\rho_{\mathrm rad}}

94: \def\rb{\rho_{\mathrm b}}

95: \def\a{\alpha}

96: \def\b{\beta}

97: \def\n{\nu}

98: \def\m{\mu}

99: \def\k{\kappa^2}

100: \def\ps{\psi_1}

101: \def\ph{\phi_1}

102: \def\pg{\varphi_g}

103: \def\rvp{\rho_\vp}

104: \def\mh{\mathcal{H}}

105: \def\mr{\mathcal{R}}

106: \def\mo{\mathcal{O}}

107: \def\mpp{\mathcal{P}}

108: \def\vn{\varphi_N}

109: \def\psn{\psi_{PN}}

110: \def\pn{\phi_N}

111: \def\phn{\phi_{PN}}

112: \def\ppn{\psi_{PPN}}

113: \def\fr{\frac}

114: \def\pr{\prime}

115: \def\r{\rho}

116: \def\tx{\tilde{x}}

117: 

118: % Journal and other miscellaneous abbreviations for references

119: \def \ajp#1#2#3{Am.\ J. Phys.\ {\bf#1}, #2 (#3)}

120: \def \arnps#1#2#3{Ann.\ Rev.\ Nucl.\ Part.\ Sci.\ {\bf#1}, #2 (#3)}

121: \def \ijmpa#1#2#3{Int.\ J.\ Mod.\ Phys.\ A {\bf#1}, #2 (#3)}

122: \def \nat#1#2#3{Nature {\bf#1}, #2 (#3)}

123: \def \pisma#1#2#3#4{Pis'ma Zh.\ Eksp.\ Teor.\ Fiz.\ {\bf#1}, #2 (#3) [JETP

124: Lett.\ {\bf#1}, #4 (#3)]}

125: \def \pl#1#2#3{Phys.\ Lett.\ {\bf#1}, #2 (#3)}

126: \def \pla#1#2#3{Phys.\ Lett.\ A {\bf#1}, #2 (#3)}

127: \def \plb#1#2#3{Phys.\ Lett.\ B {\bf#1}, #2 (#3)}

128: \def \ppmsj#1#2#3{Proc.\ Phys.\ Math.\ Soc.\ Japan {\bf#1}, #2 (#3)}

129: \def \prr#1#2#3{Phys.\ Rev.\ {\bf#1}, #2 (#3)}

130: \def \prrc#1#2#3{Phys.\ Rev.\ C {\bf#1}, #2 (#3)}

131: \def \prrd {Phys.\ Rev.\ D\ }%#1#2#3{Phys.\ Rev.\ D {\bf#1}, #2 (#3)}

132: \def \prrl{Phys.\ Rev.\ Lett.\ }%#1#2#3{Phys.\ Rev.\ Lett.\ {\bf#1}, #2 (#3)}

133: \def \prrp#1#2#3{Phys.\ Rept.\ {\bf#1}, #2 (#3)}

134: \def \ptp#1#2#3{Prog.\ Theor.\ Phys.\ {\bf#1}, #2 (#3)}

135: \def \rmp#1#2#3{Rev.\ Mod.\ Phys.\ {\bf#1}, #2 (#3)}

136: \def \rp#1{~~~~~\ldots\ldots{\rm rp~}{#1}~~~~~}

137: \def \sci#1#2#3{Science {\bf#1}, #2 (#3)}

138: \def\apj{Astrophys. J.\ }

139: \def\aj{Astron. J.\ }

140: \def\apjl{Astrophys. J. Lett.\ }

141: \def\apjs{Astrophys. J. Suppl.\ }

142: \def\mnras{Mon. Not. R. Astron. Soc.\ }

143: \def\pasp{Publ. Astron. Soc. Pac.\ }

144: \def\pasj{Publ. Astron. Soc. Japan\ }

145: \def\physrep{Phys. Rept.\ }

146: \def\araa{Annu. Rev. Astron. Astrophys.\ }

147: \def\aap{Astron. Astrophys.\ }

148: \def\aaps{Astron. Astrophys. Suppl.\ }

149: % End of journal references

150: 

151: %WMAP

152: \newcommand{\MAP}{\textsl{WMAP}}

153: \newcommand{\WMAP}{\textsl{WMAP}}

154: \newcommand{\refeqn}[1]{(\ref{#1})}

155: \newcommand{\reffig}[1]{Figure~\ref{#1}}

156: \newcommand{\reftbl}[1]{Table~\ref{#1}}

157: \newcommand{\refsec}[1]{Section~\ref{#1}}

158: \newcommand{\acos}{{\rm acos}}

159: \newcommand{\asin}{{\rm asin}}

160: \newcommand{\cof}{{\rm cof}}

161: \newcommand{\cov}{{\rm cov}}

162: \newcommand{\diag}{{\rm diag}}

163: \newcommand{\sinc}{{\rm sinc}}

164: \newcommand{\tr}{{\rm tr}}

165: \newcommand{\var}{{\rm var}}

166: 

167: \newcommand{\bi}{\begin{itemize}}

168: \newcommand{\ei}{\end{itemize}}

169: \newcommand{\be}{\begin{equation}}

170: \newcommand{\ee}{\end{equation}}

171: \newcommand{\bea}{\begin{eqnarray}}

172: \newcommand{\eea}{\end{eqnarray}}

173: \newcommand{\sm}{{\rm M}_\odot}

174: \newcommand{\C}{{\cal C}}

175: %\renewcommand{\el}{{\ell}}

176: \newcommand{\M}{{\bf M}}

177: \newcommand{\W}{{\bf W}}

178: \newcommand{\B}{{\bf B}}

179: \newcommand{\bft}{{\bf t}}

180: \newcommand{\NN}{{\cal N}}

181: \newcommand{\GG}{{\cal G}}

182: 

183: %\newcommand{\arcmin}{'}

184: \newcommand{\dt}{\tilde \delta}

185: \newcommand{\dD}{\delta_{\rm D}}

186: \newcommand{\del}{\nabla}

187: \newcommand{\knl}{k_{n\ell}}

188: \newcommand{\fg}{\bf}

189: \newcommand{\fore}{{\rm f}}

190: \newcommand{\bis}{B_{l_1 l_2 l_3}}

191: \newcommand{\bislm}{B_{l_1 l_2 l_3}^{m_1 m_2 m_3}}

192: \newcommand{\bisp}{B_{l_1' l_2' l_3'}}

193: \newcommand{\deld}{\delta^{\rm D}}

194: \newcommand{\bfn}{\hat{\bf n}}

195: \newcommand{\bfm}{\hat{\bf m}}

196: \newcommand{\bl}{\hat{\bf l}}

197: \newcommand{\bk}{\hat{\bf k}}

198: \newcommand{\tableskip}{\\[-6pt]}

199: \newcommand{\dop}{{\rm dop}}

200: \newcommand{\sz}{{\rm SZ}}

201: \newcommand{\firb}{{\rm FIRB}}

202: \newcommand{\cut}{{\rm cut}}

203: \newcommand{\cmb}{{\rm CMB}}

204: \newcommand{\kappasz}{{\rm \kappa-SZ}}

205: \newcommand{\sky}{{\rm sky}}

206: \newcommand{\tot}{{\rm tot}}

207: \newcommand{\noise}{{\rm noise}}

208: \newcommand{\isw}{{\rm ISW}}

209: \newcommand{\sw}{{\rm SW}}

210: \newcommand{\ov}{{\rm OV}}

211: \newcommand{\se}{{\rm S}}

212: \newcommand{\ri}{{\rm ri}}

213: \newcommand{\len}{{\rm len}}

214: \newcommand{\Ylm}[1]{Y_{\l_#1}^{m_#1}}

215: \newcommand{\Ylmn}{Y_{\l}^{m}}

216: \newcommand{\alm}[1]{a_{\l_#1 m_#1}}

217: \newcommand{\almn}{a_{\l m}}

218: \newcommand{\Dk}{\frac{d^3{\bf k}}{\left( 2\pi \right) ^3}}

219: \newcommand{\gas}{{\rm gas}}

220: \newcommand{ \ie }{{\it i.e.}}

221: \newcommand{\hatn}{{\bf \hat n}}

222: \newcommand{\hatm}{{\bf \hat m}}

223: \newcommand{\hatnp}{{\bf \hat n^\prime}}

224: \newcommand{\hatnpp}{{\bf \hat n^{\prime\prime}}}

225: \newcommand{\hatnone}{{\bf \hat n_1}}

226: \newcommand{\hatntwo}{{\bf \hat n_2}}

227: %end WMAP

228: 

229: \def\spose#1{\hbox to 0pt{#1\hss}}

230: \def\approxgt{\mathrel{\spose{\lower 3pt\hbox{$\sim$}}

231:         \raise 2.0pt\hbox{$>$}}}

232: 

233: \renewcommand{\textsf}[1]{{\small #1}}

234: 

235: %CMB Resource

236: \newcommand\eq{\begin{equation}}

237: \newcommand\en{\end{equation}}

238: %end CMB Resource

239: 

240: %All-Sky-Analysis

241: \def\edth{\;\raise1.0pt\hbox{$'$}\hskip-6pt\partial\;}

242: \def\baredth{\;\overline{\raise1.0pt\hbox{$'$}\hskip-6pt

243: \partial}\;}

244: %\def\bi#1{\bbox{#1}}

245: \def\bi#1{\hbox{\boldmath{$#1$}}}

246: \def\gsim{\raise2.90pt\hbox{$\scriptstyle

247: >$} \hspace{-6.4pt}

248: \lower.5pt\hbox{$\scriptscriptstyle

249: \sim$}\; }

250: \def\lsim{\raise2.90pt\hbox{$\scriptstyle

251: <$} \hspace{-6pt}\lower.5pt\hbox{$\scriptscriptstyle\sim$}\; }

252: %end All-Sky-Analysis

253: 

254: %\renewcommand{\thesubfigure}{} 

255: 

256: %\renewcommand{\includegraphics}[1]{{ }}

257: 

258: 

259: %%% Path to the directory containing the graphics and figures 

260: %\graphicspath{{/home/canopus/carbone/paper/TEB_lensing_fig_redsz/}}

261: 

262: \title[Lensed CMB maps from the Millennium Simulation]{Lensed CMB temperature and polarization maps from the Millennium Simulation}

263: \author[]{% 

264: Carmelita Carbone$^{1}$\footnotemark[1],

265: Carlo Baccigalupi$^{2}$\footnotemark[3],

266: Matthias Bartelmann$^{3}$\footnotemark[4],

267: \newauthor

268: Sabino Matarrese$^{4}$\footnotemark[5],

269: Volker Springel$^{5}$\footnotemark[2]

270: \\

271: \\ 

272: $^{1}$ Institut de Ci\`encies de l'Espai, CSIC/IEEC, Campus UAB, F. de Ci\`encies, Torre C5 par-2,  Barcelona 08193, Spain \\

273: $^{2}$ SISSA/ISAS, Astrophysics Sector, Via Beirut 4, I-34014, Trieste, Italy and \\

274: INFN, Sezione di Trieste, Via Valerio, 2, 34127, Trieste, Italy \\

275: $^{3}$ Institut f$\ddot{\rm u}$r Theoretische Astrophysik,

276: Universit$\ddot{\rm a}$t Heidelberg, Tiergartenstrasse 15, D-69121,

277: Heidelberg, Germany \\

278: $^{4}$ Dipartimento di Fisica `Galileo Galilei', Universit\`a di Padova and \\ 

279: INFN, Sezione di Padova, Via Marzolo 8, I-35131 Padova, Italy \\

280: $^{5}$ Max-Planck-Institute for Astrophysics, Karl-Schwarzschild-Str. 1, D-85741

281: Garching, Germany\\

282: }

283: 

284: \begin{document}

285: 

286: \maketitle

287: 

288: 

289: \begin{abstract}

290: We have constructed the first all-sky CMB temperature and polarization

291: lensed maps based on a high-resolution cosmological $N$-body

292: simulation, the Millennium Simulation (MS).  We have exploited the

293: lensing potential map obtained using a map-making procedure

294: \citep{Carbone_etal2008} which integrates along the

295: line-of-sight the MS dark matter distribution by stacking and

296: randomizing the simulation boxes up to $z=127$, and which

297: semi-analytically supplies the large-scale power in the angular

298: lensing potential that is not correctly sampled by the $N$-body

299: simulation. The lensed sky has been obtained by properly modifying the

300: latest version of the LensPix code \citep{Lewis05} 

301: to account for the MS structures.  We

302: have also produced all-sky lensed maps of the so-called $\psi_E$ and

303: $\psi_B$ potentials, which are directly related to the \emph{electric}

304: and \emph{magnetic} types of polarization.  The angular power spectra

305: of the simulated lensed temperature and polarization maps agree well

306: with semi-analytic estimates up to $l \leq 2500$, while on smaller

307: scales we find a slight excess of power which we interpret as being

308: due to non-linear clustering in the MS.  We also observe how

309: non-linear lensing power in the polarised CMB is transferred to large

310: angular scales by suitably misaligned modes in the CMB and the lensing

311: potential. This work is relevant in view of the future CMB probes, as

312: a way to analyse the lensed sky and disentangle the contribution from

313: primordial gravitational waves.

314: \end{abstract}

315: 

316: \begin{keywords}

317: gravitational lensing, cosmic microwave background, cosmology

318: \end{keywords}

319: 

320: 

321: \section{Introduction}

322: \renewcommand{\thefootnote}{\fnsymbol{footnote}}

323: \footnotetext[1]{E-mail: carbone@ieec.uab.es}

324: \footnotetext[5]{E-mail: bacci@sissa.it}

325: \footnotetext[3]{E-mail: mbartelmann@ita.uni-heidelberg.de}

326: \footnotetext[4]{E-mail: sabino.matarrese@pd.infn.it}

327: \footnotetext[2]{E-mail: volker@MPA-Garching.MPG.DE}

328: \renewcommand{\thefootnote}{\arabic{footnote}}

329: %\addtocounter{footnote}{-1}

330: 

331: The Cosmic Microwave Background (CMB) is characterized both by

332: \emph{primary} anisotropies, imprinted at the last scattering surface

333: at redshift $z\sim 1100$, and by \emph{secondary} anisotropies caused

334: along the way to us by density inhomogeneities and re-scattering off

335: electrons that are freed during the epoch of reionization, and heated

336: to high temperature when massive structures virialize.

337: 

338: On one hand, the primary CMB anisotropies give direct insight into the

339: structure of the very early Universe, and are one of the principal

340: pillars on which the standard cosmological $\Lambda$CDM model is

341: founded.  The temperature anisotropy power spectrum has now been

342: measured to very high precision \citep[e.g.][]{WMAP5} yielding tight

343: constraints on the basic parameters of the cosmological model.

344: 

345: On the other hand, one of the most important mechanisms that can

346: generate secondary anisotropies is the weak gravitational lensing of

347: the CMB, which arises from the distortions induced in the geodesics of

348: CMB photons by gradients in the gravitational matter potential

349: \citep{Matthias_rew,Lewis06}. The remapping of points produced by

350: lensing induces non-Gaussianities in the observed CMB sky, and also

351: changes the power spectra of the perturbations.

352: 

353: The CMB is also expected to be polarized at the $\sim10$\% level

354: principally because of Thomson scattering of photons off free

355: electrons during recombination.  Thomson scattering generates linear

356: polarization only, which is usually expressed in terms of the Stokes

357: parameters $Q$ and $U$.  They can in turn be decomposed into

358: coordinate independent $E$- and $B$-modes of polarization

359: \citep[e.g.][]{Zaldarriaga&Seljak97} with opposite parities, the

360: so-called ``electric'' and ``magnetic'' types of polarization.  To

361: linear order in perturbation theory, primordial scalar (density)

362: perturbations can only generate $E$-polarization, while primordial

363: vector and tensor (gravitational waves) perturbations can generate

364: both scalar $E$- and pseudoscalar $B$-polarization

365: \citep[e.g.]{Seljak&Zaldarriaga97}.  In particular, \emph{primary}

366: B-mode polarization represents the imprints left from the primordial

367: gravitational waves (GWs) on the CMB \citep{Hu et al. 1998, review1,

368:   review2}: if initial fluctuations are created very early,

369: e.g.~during inflation so that the vector growth is damped, primary

370: B-modes are produced only by tensor perturbations that, being damped

371: at last scattering by the horizon entering, produce the largest amount

372: of temperature quadrupole anisotropy and, consequently, by Thomson

373: scattering, the largest amount of polarization

374: \citep{Pritchard&Kamionkowski04}.  The primordial gravitational

375: radiation is thought to have been generated by quantum fluctuations of

376: the metric tensor during the inflationary era, with a strain amplitude

377: proportional to the square of the inflation energy scale.

378: Consequently, the indirect detection of this relic gravitational

379: background via the direct observation of the primary B-mode

380: polarization, as expected from future dedicated CMB missions by ESA

381: and NASA\footnote{See \textsf{lambda.gsfc.nasa.gov} for a complete

382:   list of operating and planned CMB experiments}, will shed light on

383: the physics of the very early Universe and will represent a powerful

384: magnifier on the inflationary era and the very first moments in the

385: existence of the Universe \citep{Kamionkowski et al. 1997,

386:   Zaldarriaga&Seljak97, Seljak&Zaldarriaga97}.

387: 

388: However, as for the CMB temperature, there are mechanisms also for the

389: polarization that can produce \emph{secondary} B-modes, with the

390: dominant one being again gravitational lensing, i.e. \emph{cosmic

391:   shear} (CS), which distorts the primary CMB pattern, in particular

392: converting E- into B-polarization \citep{Zaldarriaga&Seljak98}, even

393: in case of absence of primary B-modes.  Although comparable, B-modes

394: from primordial GWs exhibit their peak at multipoles $l\approx 100$,

395: corresponding to the degree scale, while, for lensed B-modes, the peak

396: is at $l\approx1000$, corresponding to the arcminute scale.

397: Nonetheless, if the energy scale of inflation is $V^{1/4}\le

398: 4\times10^{15}$GeV, the CS-induced curl represents a foreground for

399: the $l\approx50-100$ primordial GW-induced primary B-polarization

400: \citep{Cabella_Kamionkowski04}.  This could limit the extraction of

401: the gravity wave signal if not taken into account correctly

402: \citep[e.g.][]{seljak_hirata_2004}, even though forthcoming CMB probes

403: will have in principle the sensitivity and the instrumental

404: performance for the detection of the CMB anisotropies in total

405: intensity and polarization.

406: 

407: A precise knowledge of the lensing effects would also provide new

408: insights and constraints on the expansion history of the Universe, on

409: the process of cosmological structure formation

410: \citep{acquaviva_baccigalupi_2006,hu_etal_2006} and on the

411: cosmological parameter estimation \citep{Smith_etal2006,

412:   Smith_etal2006_bis}.  In particular, for a correct interpretation of

413: the data from the forthcoming Planck

414: satellite\footnote{www.rssd.esa.int/PLANCK}, it will be absolutely

415: essential to understand and model the CMB lensing, as the satellite

416: has the sensitivity for measuring the CMB lensing with good

417: accuracy. We note that a first detection of CMB lensing in data from

418: the Wilkinson Microwave Anisotropy Probe (WMAP\footnote{See

419:   \textsf{map.gsfc.nasa.gov}}) combined with complementary data has

420: already been claimed by \cite{smith_etal_2007} and \cite{Seljak0108},

421: and evidence for weak gravitational lensing of the CMB has been observed

422: at $>$3-sigma significance by \cite{ACBAR}.

423: 

424: From the arguments above it follows that the next generation CMB

425: experiments will require a detailed lensing reconstruction and an

426: accurate de-lensing methodology.  One can try to reconstruct the

427: gravitational lensing effects using the so-called quadratic and

428: maximum-likelihood estimators \citep{Hu&Okamoto2002,

429:   Hirata&Seljak2003}, which allow to reproduce at some level of

430: precision the lensing potential from the observed CMB itself, and to

431: invert the photon geodesic remapping induced by the cosmic shear.  Up

432: to now these methodologies have been applied only to limited patches

433: of sky \citep{Amblard et al 04} and/or under the hypothesis of a

434: Gaussian distribution of the lensing sources

435: \citep[e.g.][]{seljak_hirata_2004}.  However, this is only a first

436: order approximation, since the non-linear evolution of the cosmic

437: structures induces non-Gaussian features in the lensing potential,

438: features that in turn have an impact on the non-Gaussian statistic of

439: the CMB produced by the point remapping caused by the cosmic shear

440: itself on the CMB fluctuation pattern.

441: 

442: It is thus very interesting to test the performance of these

443: estimators on full-sky, lensed CMB maps which do include the effects

444: of the non-linear structure evolution at all the orders. This demands

445: detailed simulated lensed CMB maps.

446: 

447: The increasing availability of high-resolution $N$-body simulations in

448: large periodic volumes makes it possible to directly simulate the CMB

449: distortions caused by weak lensing using realistic cosmological

450: structure formation calculations. Our previous work

451: \citep{Carbone_etal2008} represents a first step into this direction.

452: Indeed, existing studies already give access to statistical properties

453: of the expected all-sky CMB lensing signal \citep[see e.g.][and

454:   references therein]{Lewis05}, but these studies are based on

455: `semi-analytic' calculations that use approximate parameterizations of

456: the non-linear evolution of the matter power spectrum.  On the other

457: hand, up to now $N$-body numerical simulations have been used to lens

458: the CMB only in limited patches of sky \citep{Das&Bode08}, or to

459: produce full-sky convergence maps confined to low redshifts $z\sim 1$

460: \citep{Fosalba07, arXiv:0807.3651}.

461: 

462: In \cite{Carbone_etal2008} we have developed and described a procedure

463: which gives access to the full statistics of the lensed CMB signal,

464: including non-linear and non-Gaussian effects on the full-sky. This

465: should allow improvements in the methods for separating the different

466: contributions to CMB anisotropies in the data, which would help

467: substantially to uncover all the cosmological information in the

468: forthcoming observations.

469: 

470: In this paper we apply the methodology developed in

471: \cite{Carbone_etal2008} to the construction of all-sky, lensed

472: simulated temperature and polarization maps.  In Section~2 we describe

473: the procedure utilized to lens, which is based on the simulated

474: lensing potential and deflection angle templates obtained in

475: \citet{Carbone_etal2008}.  In Section~3 we describe the resulting

476: simulated lensed maps.  In Section~4 and Section~5 we apply

477: statistical analyses to the obtained temperature and polarization

478: lensed maps showing consistency and possible differences with respect

479: to semi-analytical expectations. Finally in Section~6 we draw the

480: conclusions of this work and outline next steps and future

481: applications.

482: \begin{figure}

483: \includegraphics[width=.48\textwidth]{projpot_comparison_z127_l400_LS_tot_phiTcorr_seed2_deconv_new.eps}

484: \includegraphics[width=.48\textwidth]{projpot_comparison_6143_z127_l400_LS_tot_phiTcorr_seed2_deconv_ratio_new.eps}

485: \footnotesize

486: \caption{{\em Top panel:} The dotted orange line represents the simulated lensing

487:   potential power spectrum obtained via line-of-sight integration

488:   across the MS dark matter distribution up to $z=127$. The blue solid

489:   line is the same as the orange dotted one after reinstating the

490:   large scale power with the use of the LS-adding technique. The

491:   dot-dashed black line represents the power spectrum of the lensing

492:   potential obtained with the CAMB code stopping the line-of sight

493:   integration at $z=127$, and including an estimate of the

494:   non-linear contributions \citep{Halofit}.  The long-dashed light

495:   green line is the same as the dot-dashed black one with

496:   line-of-sight integration up to $z=1100$.  Finally, the dashed

497:   violet line represents the linear lensing potential power spectrum

498:   from the CAMB code in the linear approximation and integrating up to

499:   $z=127$.  {\em Bottom panel:} The ratios between the power spectra shown in the

500:   top panel and the CAMB linear lensing potential spectrum up to

501:   $z=127$. It is worth to note that there is no difference using

502:   $z=1100$ or $z=127$ for the semi-analytical expectations.}

503: \label{projpot_PS}

504: \end{figure}

505: 

506: \section{CMB lensing through the Millennium Simulation}

507: Weak lensing of the CMB deflects photons coming from an original

508: direction ${\bf \hat{n}}'$ on the last scattering surface to a

509: direction ${\bf \hat{n}}$ on the observed sky, so a lensed CMB field

510: is given by $\tilde{X}({\bf \hat{n}}) = X({\bf \hat{n}}')$ in terms of

511: the unlensed field $X=T, Q, U$ \citep[e.g.][]{Lewis05}.  The

512: displacement of the points is determined by the integral of the

513: gravitational potential along the line of sight to the last scattering

514: surface, as we review below.

515: 

516: In what follows we will consider only the \emph{small-angle

517:   scattering} limit, \ie~the case where the \emph{change} in the

518: comoving separation of CMB light-rays, owing to the deflection caused

519: by gravitational lensing from matter inhomogeneities, is small

520: compared to the comoving separation between the \emph{undeflected}

521: rays. In this case it is sufficient to calculate all the relevant

522: integrated quantities, \ie~the so-called \emph{lensing-potential} and

523: its angular gradient, the \emph{deflection-angle}, along the

524: undeflected rays. This small-angle scattering limit corresponds to the

525: Born approximation.

526: 

527: Adopting conformal time and comoving coordinates in a flat geometry

528: \citep{maber}, the integral for the projected lensing-potential due to

529: scalar perturbations with no anisotropic stress reads

530: \begin{align}

531: \label{lensingpotential}

532: \Psi({\bf \hat{n}})\equiv

533: -2\int_0^{r_*} \fr{r_*-r}{r_*r}\,\fr{\Phi(r{\bf\hat{n}};\eta_0-r )}{c^2}\,{\rm d}r\,,

534: \end{align}

535: where $r$ is the comoving distance, $r_*\simeq 10^4$ Mpc is its value

536: at the last-scattering surface, $\eta_0$ is the present conformal

537: time, and $\Phi$ is the physical peculiar gravitational potential

538: generated by density perturbations

539: \citep{Hu2000,Matthias_rew,Refregier,Lewis06}.

540: 

541: Actually, the lensing potential is formally divergent owing to the

542: $1/r$ term near $r=0$; nonetheless, this divergence affects the

543: lensing potential monopole only, which can be set to zero, since it

544: does not contribute to the deflection-angle. In this way the remaining

545: multipoles take a finite value and the lensing potential field is well

546: defined \citep{Lewis06}.

547: \begin{figure*}

548: \begin{center}

549: \resizebox{12.7cm}{!}{\includegraphics{T_diff_map_MS_b2_l400_z127_b2_seed2_noMonDip_Nside2048_lmax6143_phiTcorr_hist.ps}}

550: \resizebox{12.7cm}{!}{\includegraphics{P_diff_map_MS_b2_l400_z127_b2_seed2_noMonDip_Nside2048_lmax6143_phiTcorr.ps}}

551: \resizebox{12.7cm}{!}{\includegraphics{totmap_anglemodule_l400_z127_LS_tot_deconv_phiTcorr_seed2.ps}}

552: \footnotesize

553: \caption{{\em Top panel:} Difference $T$ map between the lensed and unlensed $T$

554:   fields obtained by supplying the LensPix code with the spherical

555:   harmonic coefficients extracted from the MS lensing potential map

556:   and by implementing in it the LS-adding technique, as described in

557:   the text. Units in $mK$. A histogram equalized color mapping has

558:   been used to increase the contrast.  {\em Middle panel:} Modulus of the

559:   polarization $\Delta P\equiv \sqrt{\Delta Q^2+\Delta U^2}$, where

560:   $\Delta Q$ and $\Delta U$ are the difference $Q$ and $U$ maps

561:   obtained using the same technique as for the $T$ difference map of

562:   the top panel. Units in $mK$.  {\em Bottom panel:} Map of the deflection-angle

563:   modulus obtained as angular gradient of the lensing-potential whose

564:   power spectrum is represented by the solid blue line in

565:   Fig.~\ref{projpot_PS}}

566: \label{TP_diff_maps}

567: \end{center}

568: \end{figure*}

569: The vector ${\bf \hat{n}}'$ is obtained from ${\bf \hat{n}}$ by moving

570: its end on the surface of a unit sphere by a distance

571: $|\nabla_{\bf\hat{n}}\Psi({\bf \hat{n}})|$ along a geodesic in the

572: direction of $\nabla_{\bf\hat{n}}\Psi({\bf \hat{n}})$, where

573: $[1/r]\nabla_{\hat{\bf n}}$ is the two dimensional (2D) transverse

574: derivative with respect to the line-of-sight pointing in the direction

575: ${\hat{\bf n}}\equiv(\vartheta,\varphi)$

576: \citep{Hu2000,Challinor02,Lewis05}.  We assume

577: $|\nabla_{\bf\hat{n}}\Psi({\bf \hat{n}})|$ to be constant between

578: ${\bf \hat{n}}$ and ${\bf \hat{n}}'$, consistent with the Born

579: approximation.

580: 

581: If the gravitational potential $\Phi$ is Gaussian, so is the lensing

582: potential.  However, the lensed CMB is non-Gaussian, as it is a second

583: order cosmological effect produced by cosmological perturbations onto

584: CMB anisotropies, yielding a finite correlation between different

585: scales and thus non-Gaussianity.  This is expected to be most

586: important on small scales, due to the non-linearity already present in

587: the underlying properties of lenses.

588: 

589: In order to generate full-sky $T$, $Q$, $U$ maps lensed by the matter

590: distribution of the Millennium Simulation, we have modified the

591: publicly available LensPix

592: code\footnote{http://cosmologist.info/lenspix/} (LP), which is

593: described in \cite{Lewis05}. In its original version, in fact, this

594: code lenses the primary CMB intensity and polarization fields via a

595: Gaussian realization, in the spherical harmonic domain, of the lensing

596: potential power spectrum as extracted from the publicly available Code

597: for Anisotropies in the Microwave Background

598: (CAMB\footnote{http://camb.info/}).  Our modification (hereafter

599: referred to as ``MS-modified-LP'') consists in forcing LP to deflect

600: the CMB photons using the fully non-linear and non-Gaussian lensing

601: potential realization obtained from the MS using the procedure

602: briefly summarized here, which was presented by

603: \cite{Carbone_etal2008}; we refer the reader to that paper for more

604: detail.

605: 

606: The MS is a high-resolution $N$-body simulation for a $\Lambda$CDM

607: cosmology consistent with the WMAP first year results

608: \citep{spergel2003}, carried out by the Virgo Consortium

609: \citep{Springel2005}. It uses about 10 billion collisionless particles

610: with mass $8.6\times 10^{8} h^{-1}{\rm M_\odot}$, in a cubic region

611: $500\,h^{-1}{\rm Mpc}$ on a side which evolves from redshift

612: $z_{*}=127$ to the present, with periodic boundary conditions.  Our

613: map-making procedure is based on ray-tracing of the CMB photons in the

614: Born approximation through the three-dimensional field of the MS

615: peculiar gravitational potential. In order to produce mock lensing

616: potential maps that cover the past light-cone over the full sky, we

617: stack the peculiar gravitational potential grids around the observer

618: (which is located at $z=0$), exploiting the pre-computed and stored

619: snapshots of the simulation. The spacing of the time outputs of the MS

620: simulation is such that it corresponds to an average distance of

621: $140\,h^{-1}{\rm Mpc}$ (comoving) on the past light-cone. We fully exploit this time resolution

622: which, at high accuracy, allows to avoid the adoption of time interpolation techniques, 

623: and use all the 63 outputs of the simulation along

624: our integration paths. In practice this means that the data

625: corresponding to a particular output time is utilized in a spherical

626: shell of average thickness $140\,h^{-1}{\rm Mpc}$.

627: Moreover, the total volume around the observer up

628: to $z_{*}$ is divided into spherical shells, each of thickness

629: $500h^{-1}{\rm Mpc}$. All the MS boxes falling into the same shell are

630: translated and rotated with the same random vectors generating a

631: homogeneous coordinate transformation throughout the shell, while

632: the randomization vectors change from shell to shell. The peculiar gravitational

633: potential at each point along a ray in direction $\bfn$ is

634: spatially interpolated from the pre-computed MS grid which possesses a spatial

635: resolution of about $195h^{-1}{\rm kpc}$. The deflection angle is

636: computed along the line of sight as well, by numerically evaluating

637: the gravitational potential gradient and interpolating at each point

638: along the line of sight \citep{Carbone_etal2008}. 

639: 

640: Being repeated on scales larger than the box size, the resulting weak

641: lensing distortion lacks large scale power, which manifests itself in

642: the lensing potential power spectrum as an evident loss of large scale

643: power with respect to semi-analytic expectations, which is most

644: noticeable for multipoles smaller than $l\simeq 400$. This has been

645: cured by augmenting large scale power (LS-adding) directly in the

646: angular domain, a procedure which we exploit here as well.  More

647: specifically, we have implemented the LS-adding technique directly

648: into the LensPix code as we now explain. We have again split

649: the spherical harmonics domain into two multipole ranges: $0\leq l

650: \leq 400$, where the MS fails in reproducing the correct lensing

651: potential power due to the limited box-size of the simulation, and $l

652: > 400$, where instead the power spectrum is reproduced correctly by

653: the Millennium Simulation (see Fig.~\ref{projpot_PS}).  On the latter

654: interval of multipoles, we have extracted the corresponding ensemble

655: $\Psi_{lm}^{\rm MS}$ of lensing-potential spherical harmonic

656: coefficients produced by the MS lens distribution.  We have modified

657: the LensPix code so that it reads and uses these MS harmonic

658: coefficients on the corresponding range of multipoles.  On the

659: interval $0\leq l \leq 400$, instead, we let LensPix generate its own

660: ensemble of spherical harmonic coefficients $\Psi_{lm}^{\rm LP}$,

661: which are a realization of a Gaussian random field characterised by

662: the CAMB semi-analytic lensing-potential power spectrum

663: \citep[including the estimate of the contribution from

664:   non-linearity][]{Halofit} inserted as input in the parameter file of

665: LP.

666: 

667: Since on low multipoles the effects of the non-Gaussianity from the

668: non-linear scales are negligible and the $\Psi_{lm}$ are independent,

669: every time that we run the MS-modified-LP, we generate a joined

670: ensemble of $\tilde{\Psi}_{lm}$, where

671: $\tilde{\Psi}_{lm}=\Psi_{lm}^{\rm LP}$ for $0\leq l \leq 400$ and

672: $\tilde{\Psi}_{lm}=\Psi_{lm}^{\rm MS}$ for $l > 400$.

673: 

674: This technique achieves two goals: firstly it reproduces correctly the

675: non-linear and non-Gaussian effects of the MS non-linear dark matter

676: distribution on multipoles $l > 400$, including at the same time the

677: contribution from the large scales at $l \le 400$, where the lensing

678: potential follows mostly the linear trend as shown from the

679: light-green dot-dashed line in Fig.~\ref{projpot_PS}. Secondly, it

680: allows to take correctly into account the cross-correlation between

681: the temperature and the lensing-potential \citep[e.g.][]{Lewis06} due

682: to the Integrated Sachs-Wolfe (ISW) effect on the low multipoles,

683: i.e. on the large scales \citep[this effect is instead negligible at

684:   $l \gtrsim 200$][]{Afshordi04}.  In fact, as it happens for the

685: semi-analytic contribution to the lensing-potential, also the primary

686: temperature and polarization fields are generated by LP as Gaussian

687: realizations, in the harmonic domain, of the corresponding power

688: spectra obtained by CAMB, which correctly includes the contribution to

689: the temperature from the ISW effect.

690: \begin{figure*}

691: \begin{center}

692: \resizebox{13cm}{!}{\includegraphics{psiE_diff_map_MS_b2_l400_z127_b2_seed2_noMonDip_Nside2048_lmax6143_phiTcorr.ps}}

693: \resizebox{13cm}{!}{\includegraphics{psiE_diff_map_test_MS_z127_b2_seed2_Nside2048_lmax6143.ps}}

694: \footnotesize

695: \caption{Difference maps between the $\psi_E$ lensed and unlensed

696:   fields, obtained in the MS-modified-LP case (upper panel) and in the

697:   unmodified-LP case (lower panel). Units in $mK$.}

698: \label{psiE_diff_maps}

699: \end{center}

700: \end{figure*}

701: \begin{figure*}

702: \begin{center}

703: \resizebox{13cm}{!}{\includegraphics{psiB_diff_map_MS_b2_l400_z127_b2_seed2_noMonDip_Nside2048_lmax6143_phiTcorr.ps}}

704: \resizebox{13cm}{!}{\includegraphics{psiB_diff_map_test_MS_z127_b2_seed2_Nside2048_lmax6143.ps}}

705: \footnotesize

706: \caption{Difference maps between the $\psi_B$ lensed and unlensed

707:   fields, obtained in the MS-modified-LP case (upper panel) and in the

708:   unmodified-LP case (lower panel). Units in $mK$.}

709: \label{psiB_diff_maps}

710: \end{center}

711: \end{figure*}

712: \begin{figure}

713: \includegraphics[width=.48\textwidth]{TEB_test_lensed_modes_6143_l400_Nol400_z127_b2_ratio_xlog_seed2_new.eps}

714: \includegraphics[width=.48\textwidth]{TEB_diff_test_lensed_modes_6143_l400_Nol400_z127_b2_ratio_xlog_seed2_new.eps}

715: \footnotesize

716: \caption{{\em Top panel:} The dot-dashed light green line represents

717:   the ratio between the lens-induced $B$ power spectrum, which is

718:   obtained using the multipole range $0\leq l_{lp}\leq 6143$ for the

719:   spherical harmonic coefficients of the lensing potential, and the

720:   spectrum obtained using only $0\leq l_{lp}\leq 400$. The solid blue

721:   line and the dashed red line represent these ratios in the $T$ and

722:   $E$ cases, respectively.  {\em Bottom panel:} The same as the upper

723:   panel when we calculate these ratios for the temperature and

724:   polarization power spectra extracted from the difference $T$,

725:   $Q$ and $U$ maps, in the $0\leq l_{lp}\leq 6143$ and

726:   $0\leq l_{lp}\leq 400$ cases, respectively.}

727: \label{Nol400}

728: \end{figure}

729: We emphasise that we always run CAMB using the same cosmological

730: parameters as the MS specified in Section 2, and, for consistency with

731: the MS map-making procedure, we have fixed the maximum redshift of the

732: line-of-sight integration for the CAMB lensing sources at $z_{\rm

733:   max}=127$. Indeed, the lensing power from even higher redshifts is

734: negligible for CMB lensing, as we show in Fig.~\ref{projpot_PS} where

735: the light-green long-dashed line overlaps the black dashed-dotted line

736: perfectly.

737: 

738: On multipoles $l > 400$, the cross-correlation between the

739: lensing-potential and the temperature is negligible, even if there

740: could be some residual contribution coming from the non-linear

741: Rees-Sciama effect. In this work we do not consider the

742: cross-correlation due to this second order effect.

743: 

744: To generate the lensed $T$, $Q$, $U$ fields from the MS-modified-LP

745: code, we adopt the interpolation

746: scheme described in Appendix E~4 of \cite{Lewis_E4}, using a high value of the multipole

747: $l_{\rm max}$ to maximize the accuracy.  This allows running the

748: simulation several times without excessive consumption of CPU time and

749: memory.  We work under the null hypothesis that tensor modes are

750: absent in the early Universe, so that the produced B-mode polarization

751: is due only to the power transfer from the primary scalar E-modes into

752: the lens-induced B-modes.  We choose $l_{\rm max}=6143$ and interp-factor $= 1.5$, 

753: effectively the same resolution as

754: HEALPix\footnote{http://healpix.jpl.nasa.gov/} with pixelization parameter

755: $N_{\rm side}=2048$, which corresponds to an angular resolution of

756: $\sim 1.72^\prime$ \citep{Healpix}, with $12 N_{\rm side}^2$ pixels in

757: total.

758: 

759: For comparison and testing, we also generate the corresponding

760: unmodified-LP and unlensed $T$, $Q$, $U$ maps, by choosing the same

761: positive seed in the LP parameter file (hereafter ``unmodified-LP''

762: stands for the results obtained by lensing the sky with the original

763: unmodified version of the LensPix code).

764: 

765: \section{Simulated lensed maps}

766: The subtraction of the unlensed CMB maps from the corresponding lensed

767: CMB maps allows highlighting the dark-matter distribution which causes

768: the gravitational deflection of the CMB photons. In the upper and

769: middle panels of Fig.~\ref{TP_diff_maps} we show the resulting

770: difference temperature and polarization maps for a particular seed

771: choice and for the MS-modified-LP case only, since the arcminute-scale

772: differences with respect to the unmodified-LP case are not visible by

773: the naked eye.  It is worth noting how the distribution of the

774: deflection-angle modulus (lower panel) reflects itself in the

775: distribution of the temperature difference map (upper panel) and in

776: the distribution of the map of the polarization modulus $\Delta

777: P=\sqrt{\Delta Q^2+\Delta U^2}$ (middle panel), which has been

778: obtained from the difference $Q$ and $U$ maps. 

779: Unfortunately, since the primary unlensed CMB is unknown, the difference maps cannot be 

780: directly observed. Anyway the simulated difference maps can help to physically and visually 

781: understand how large scale correlations are imprinted on the CMB due to the large scale modes in

782: the deflection field, and to catch effects that are not observable due to primary unlensed CMB.

783: 

784: The temperature and polarization difference maps have the peculiarity

785: of including all the information inferred from weak-lensing on the

786: primary unlensed CMB.  Nonetheless, it is well known that the lensed

787: $B$-modes of polarization are more sensitive to the non-linear

788: evolution of the cosmic structures than the $T$, $Q$ and $U$ modes

789: separately \citep{Lewis05}.  The difference between the unmodified-LP

790: and MS-modified-LP cases lies exactly at the non-linear level, since

791: the lensing-potential realizations differ at multipoles $l > 400$ in

792: the spherical harmonic space for the two cases.  Moreover, from

793: Fig.~\ref{projpot_PS}, we observe that the MS lensing potential shows

794: an excess of power on $l>2500$ with respect to the CAMB approximation,

795: and, as already pointed out, the non-linear dark matter evolution may

796: enhance the level of non-Gaussianity present in the lensed CMB maps.

797: 

798: Consequently, in order to verify if the different distributions of the

799: lensing potential in the MS-modified-LP and unmodified-LP cases could

800: visibly affect the lensed polarization distributions, we have

801: constructed the full-sky maps of the scalar $\psi_E$ and pseudo-scalar

802: $\psi_B$ potentials, which are related to the $Q$ and $U$ Stokes

803: parameters as follows \citep{E_B_decomposition}:

804: \begin{eqnarray}

805: Q+iU= \ra\ra(\psi_E+i\psi_B)\;,

806: \label{psiE}

807: \end{eqnarray}

808: \begin{eqnarray}

809: Q-iU= \lo\lo(\psi_E-i\psi_B)\;, 

810: \label{psiB}

811: \end{eqnarray}

812: where the spin-raising $\ra$ and spin-lowering $\lo$ operators on the

813: sphere are defined as \citep{Penrose}

814: \begin{eqnarray}

815: \ra = -\sin^s\theta\,\left[{\partial\over\partial\theta}+

816: i\csc\theta\,{\partial\over\partial\phi}\right]\sin^{-s}\theta \;,

817: \end{eqnarray}

818: \begin{eqnarray}

819: \lo=-\sin^{-s}\theta\left[{\partial\over\partial\theta}-i\csc\theta

820: {\partial\over\partial\phi}\right]\sin^s\theta \;,

821: \end{eqnarray}

822: and $s$ is the spin of the function to which the operator is applied.

823: The quantities $\psi_E$ and $\psi_B$ are directly related to the

824: electric and magnetic types of polarization, since their spherical

825: harmonic coefficients can be written in terms of the harmonic

826: coefficients of the $E$- and $B$-modes, respectively

827: \begin{eqnarray}

828: \psi_E=-\sum_\lm [(l-2)!/(l+2)!]^{1/2} a_\elm \Yo_\lm \;,

829: \label{psiE_lm}

830: \end{eqnarray}

831: \begin{eqnarray}

832: \psi_B=-\sum_\lm [(l-2)!/(l+2)!]^{1/2} a_\blm \Yo_\lm \;.

833: \label{psiB_lm}

834: \end{eqnarray}

835: 

836: The $\psi_E$ and $\psi_B$ potentials are very useful for real space

837: calculations and, exploiting the HEALPix routine SYNFAST

838: \citep{Healpix}, we have produced the corresponding maps as synthetic

839: realizations, using the $E$, $B$ spherical harmonic coefficients

840: extracted (via the HEALPix routine ANAFAST) from the lensed $Q$, $U$

841: simulated maps and multiplied by the prefactors of

842: Eqs.~(\ref{psiE_lm})-(\ref{psiB_lm}), respectively.  On high

843: multipoles these prefactors have the asymptotic form

844: $[(l-2)!/(l+2)!]^{1/2}\sim l^{-2}$, consequently their effect is to

845: suppress the small scale power with respect to the $E$- and $B$-mode

846: cases, and to make the large scale structure differences much more

847: evident.

848: 

849: In fact, looking at Figs.~\ref{psiE_diff_maps}-\ref{psiB_diff_maps},

850: we observe that the lensed $\psi_E$ and $\psi_B$ difference maps have

851: a \emph{degree}-scale distribution which differs in the MS-modified-LP

852: and unmodified-LP cases, even if the corresponding lensing-potential

853: maps differ on \emph{arcminute} scales ($l>400$).  This result is more

854: expected in the $B$-mode case since it is well-known that the

855: non-linear density field introduces $\sim 10$\% corrections to the

856: lens-induced $B$-mode power on all scales \citep{Lewis06}, so that

857: multipoles $l>400$ will affect the $\psi_B$-field realizations on

858: larger scales too. The same is instead less obvious in the $E$-mode

859: case.

860: 

861: In order to understand the reason of the different large-scale

862: distribution of these maps, we have also produced the lensed

863: temperature, $\psi_E$ and $\psi_B$ fields using a Gaussian

864: lensing-potential distribution which includes only the scales that

865: correspond to multipoles up to $l_{lp}=400$ in the spherical harmonic

866: domain (where the subscript ``lp'' stands for lensing-potential). For

867: comparison among the figures, we have used the same seed as for the

868: fully ($0\leq l_{lp} \leq 6143$) lensed maps, even if the results have

869: been tested for different seeds.  In Fig.~\ref{Nol400} we show the

870: ratios between the signals corresponding to the two cases, both for

871: the lensed- and difference-maps.

872: 

873: In particular, if we consider the lensed $T$- and $E$-mode power

874: spectra (top panel of Fig.~\ref{Nol400}) which include the contribution from the primary

875: CMB, we do not observe any substantial difference on the large scales

876: between the two cases $0\leq l_{lp}\leq 400$ and $0\leq l_{lp}\leq

877: 6143$, so that on large scales the lensed electric polarization

878: appears not to be much affected by the non-linear scales in the same

879: way as the lensed temperature is. The $B$ case is different since we

880: are working with the null hypothesis of vanishing primary magnetic

881: polarization, so that we are considering here only lens-induced

882: $B$-modes.

883: 

884: Nonetheless, if we consider the power spectra extracted from the

885: difference $T$ and $\psi_E$ maps (where we are subtracting the primary

886: CMB), and again take the ratios between the signals in the two cases

887: $0\leq l_{lp}\leq 400$ and $0\leq l_{lp}\leq 6143$ (lower panel of

888: Fig.~\ref{Nol400}), we see that on large scales the $E$-mode

889: polarization, cleaned of the primary signal, gets power from the

890: smaller scales. In particular we see that, in the $E$ and $B$ cases,

891: multipoles larger than $l_{lp}=400$ transfer about $100$\% of the

892: power to the low multipoles $l\lesssim 100$, i.e. on

893: scales large enough for the flat-sky approximation to be inadequate.  

894: This transfer of power towards low multipoles is possible only for peculiar alignment angles

895: between the unlensed $E$-mode and the lensing structures, i.e.  for

896: almost, but not quite aligned modes. This is analogous to the

897: long-wavelength beat mode one obtains from superposing two

898: oscillations with two almost equal frequencies.  We do not observe the

899: same effect when we consider the ratio between the signals extracted

900: from the temperature difference maps in the two cases $0\leq

901: l_{lp}\leq 400$ and $0\leq l_{lp}\leq 6143$, where the transfer of

902: power is less than $10$\% on $l\lesssim 100$.  As we move to larger

903: multipoles instead, the power transfer decreases and, for $ 400 \leq l

904: \leq 2000$, the lensed $E$ trend starts converging to the lensed

905: temperature one, while the power of the lens-induced $B$-modes goes on

906: increasing with the multipoles, still fed by the non-linear scales in

907: the lensing-potential at $l_{lp}>400$.

908: 

909: Therefore we conclude that in

910: Figs.~\ref{psiE_diff_maps}-\ref{psiB_diff_maps} the different

911: degree-scale distributions which characterize the MS-modified-LP and

912: unmodified-LP $\psi_E$ ($\psi_B$) difference map realizations (even

913: though the corresponding lensing-potential realizations differ only on

914: $l_{lp}>400$) represent the imprints of the power transfer from

915: smaller to larger scales induced by the lensing remapping onto the

916: lensed $E$ and $B$ signals cleaned of the primary CMB, in a way which

917: differs from what happens to the lensed temperature.

918: \begin{figure}

919: \begin{minipage}{1.0\linewidth}

920: \centering

921: \vspace{0.5cm}

922: \includegraphics[width=0.9\textwidth]{TT_MS_lensed_modes_6143_l400_z127_b2_seed2.eps}

923: \vspace{1.5cm}

924: \end{minipage}

925: \begin{minipage}{1.0\linewidth}

926: \centering

927: \includegraphics[width=0.83\textwidth]{TT_MS_lensed_modes_6143_l400_z127_b2_ratio_xlog_seed2.eps}

928: \end{minipage}

929: \vspace{1.5cm}

930: \footnotesize

931: \caption{{\em Top panel:} Temperature power spectra ($\mu K^2$) for

932:   the different cases described in the text.  {\em Bottom panel:}

933:   Temperature power spectrum ratios with respect to the linear lensed

934:   case.}

935: \label{TT_PS}

936: \end{figure}

937: \begin{figure}

938: \begin{minipage}{1.0\linewidth}

939: \centering

940: \vspace{0.5cm}

941: \includegraphics[width=0.9\textwidth]{EE_MS_lensed_modes_6143_l400_z127_b2_seed2.eps}

942: \vspace{1.3cm}

943: \end{minipage}

944: \begin{minipage}{1.0\linewidth}

945: \centering

946: \includegraphics[width=0.83\textwidth]{EE_MS_lensed_modes_6143_l400_z127_b2_ratio_xlog_seed2.eps}

947: \end{minipage}

948: \vspace{1.5cm}

949: \footnotesize

950: \caption{{\em Top panel:} $E$-mode power spectra ($\mu K^2$) for the

951:   different cases described in the text.  {\em Bottom panel:} $E$-mode

952:   power spectrum ratios with respect to the linear lensed case.}

953: \label{EE_PS}

954: \end{figure}

955: \begin{figure}

956: \begin{minipage}{1.0\linewidth}

957: \centering

958: \vspace{0.5cm}

959: \includegraphics[width=0.9\textwidth]{BB_MS_lensed_modes_6143_l400_z127_b2_seed2.eps}

960: \vspace{1.0cm}

961: \end{minipage}

962: \begin{minipage}{1.0\linewidth}

963: \centering

964: \includegraphics[width=0.83\textwidth]{BB_MS_lensed_modes_6143_l400_z127_b2_ratio_xlog_seed2.eps}

965: \end{minipage}

966: \vspace{2cm}

967: \footnotesize

968: \caption{{\em Top panel:} Lens-induced $B$-mode power spectra ($\mu

969:   K^2$) for the different cases described in the text.  {\em Bottom

970:     panel:} Lens-induced $B$-mode power spectrum ratios with respect

971:   to the linear lensed case.}

972: \label{BB_PS}

973: \end{figure}

974: \begin{figure}

975: \begin{minipage}{1.0\linewidth}

976: \centering

977: \vspace{0.5cm}

978: \includegraphics[width=0.9\textwidth]{TE_MS_lensed_modes_6143_l400_z127_b2_seed2.eps}

979: \vspace{3.2cm}

980: \end{minipage}

981: \begin{minipage}{1.0\linewidth}

982: \centering

983: \includegraphics[width=0.83\textwidth]{TE_MS_lensed_modes_6143_l400_z127_b2_diff_ratio_xlog_seed2.eps}

984: %{TE_MS_lensed_modes_6143_l400_z127_b2_ratio_xlog_seed2.eps}

985: \end{minipage}

986: \vspace{1.2cm}

987: \footnotesize

988: \caption{{\em Top panel:} $T$-$E$ cross-power spectra ($\mu K^2$) for

989:   the different cases described in the text.  {\em Bottom panel:}

990:   The fractional change in the lensed $T$-$E$ cross-power spectrum due to the non-linear matter 

991:   evolution for the different cases described in the text.}

992: \label{TE_PS}

993: \end{figure}

994: \begin{figure}

995: \includegraphics[width=.48\textwidth]{TEB_test_lensed_modes_6143_linearNOlinear_z127_b2_ratio_xlog_seed2_new.eps}

996: \includegraphics[width=.48\textwidth]{TEB_diff_test_lensed_modes_6143_linearNOlinear_z127_b2_ratio_xlog_seed2_new.eps}

997: \footnotesize

998: \caption{{\em Top panel:} The dot-dashed light green line represents

999:   the ratio between the lens-induced $B$ power spectrum, which is

1000:   obtained including the non-linear cosmic structure evolution, and

1001:   the spectrum obtained in the linear limit.  The solid blue line and

1002:   the dashed red line represent these ratios in the $T$ and $E$ cases,

1003:   respectively.  {\em Bottom panel:} The same as the upper panel when

1004:   we calculate these ratios for the temperature and polarization power

1005:   spectra extracted from the difference $T$, $Q$ and $U$

1006:   maps, in the linear and non-linear cases, respectively.}

1007: \label{linearNOlinear}

1008: \end{figure}

1009: 

1010: \section{Angular power spectra}

1011: In this Section we perform several quantitative analyses of the

1012: obtained results.  Our principal aim is to test the consistency and

1013: quantify the differences between the theoretical expectations and the

1014: findings from our simulation procedure.

1015: 

1016: As a first check, we have extracted from the corresponding maps the

1017: angular power spectra of the MS-modified-LP and unmodified-LP

1018: simulated lensed $T$-, $E$- and $B$-components, together with the $TE$

1019: lensed cross-correlation power spectrum. This is simply done with the

1020: use of ANAFAST, adopting the correct deconvolution rules for the

1021: HEALPix pixel window functions, and with

1022: Eqs.~(\ref{psiE_lm})-(\ref{psiB_lm}) relating the spherical harmonic

1023: coefficients of the $\psi_E$ and $\psi_B$ potentials to the

1024: corresponding coefficients of the $E$- and $B$-modes.

1025: 

1026: We have also checked that the MS-modified-LP and unmodified-LP

1027: simulated spectra extracted directly from the maps agree with the ones

1028: obtained as direct outputs of the MS-modified-LP and unmodified-LP

1029: codes, respectively.  Moreover, to test the accuracy, we have compared

1030: all the angular power spectra with the lensed CMB power spectra

1031: obtained using the all-sky correlation function technique implemented

1032: in CAMB \citep{Challinor&Lewis05}.  These angular power spectra are

1033: shown in the upper panels of Figs.~\ref{TT_PS}-\ref{TE_PS}

1034: respectively, together with the CAMB unlensed and the linearly lensed

1035: spectra, where the latter are obtained in the linear approximation of

1036: the cosmic structure growth. Moreover, in the lower panels of the same

1037: figures we show the ratios between the simulated lensed CMB angular

1038: spectra with respect to the corresponding linear signal.

1039: 

1040: In the $TT$, $EE$ and $TE$ cases, it is clearly visible that CMB

1041: lensing smears out the acoustic peaks by transferring power from

1042: larger to smaller scales.  Moreover, since we are working under the

1043: null hypothesis of vanishing intrinsic tensor modes, Fig.~\ref{BB_PS}

1044: represents the power spectrum of the lens-induced $B$-modes into which

1045: part of the primary $E$-modes has been converted as a result of the

1046: displacements and distortions induced from the gravitational

1047: deflection onto the electric-type polarization field.

1048: 

1049: As Figs.~\ref{TT_PS}, \ref{EE_PS} and \ref{TE_PS} show, for all the

1050: lensed $TT$, $EE$ and $TE$ angular spectra, and for all multipole

1051: orders up to $l\le 3500$, we observe a mostly perfect agreement

1052: between the MS-modified-LP simulated signals and the unmodified-LP and

1053: CAMB ones, where the non-linear structure evolution is

1054: semi-analytically taken into account.  On the very large multipoles

1055: $l>3000$, a small difference appears between the CAMB and

1056: unmodified-LP lensed $TT$, $EE$ and $TE$ spectra, probably due to a

1057: numerical effect deriving from the different computational machinery

1058: implemented in CAMB and LensPix respectively.  The non-linear effects

1059: start to be important at $l>2500$, according to the semi-analytical

1060: expectations, and, on multipoles $l\sim 4000$, which correspond to

1061: angular scales of few arcminutes, they grow up to $\sim$20\% for the

1062: lensed temperature and up to $\sim$10\% for the lensed electric

1063: polarization. Moreover, the excess of power present in the MS

1064: lensing-potential power spectrum at multipoles $l \ge 2500$ (see

1065: Fig.~\ref{projpot_PS}) manifests itself as a slight excess of power in

1066: the MS-modified-LP lensed $TT$ and $EE$ spectra at $l \ge 3500$.  In

1067: particular, at $l=4096$, the non-linearities present in the Millennium

1068: Simulation produce a $\sim$6\% excess in the lensed temperature and a

1069: $\sim$1.3\% excess in the lensed electric polarization, with respect

1070: to the unmodified-LP case.

1071: 

1072: On the other hand, the excess of power due to the MS non-linearities

1073: is much more evident in the angular power spectrum of the magnetic

1074: component (see Fig.~\ref{BB_PS}) where this effect starts already at

1075: $l\sim 2500$ and grows up to $\sim$14\% at $l=4096$, with respect to

1076: the unmodified-LP case.  Moreover, the non-linear effects, which in

1077: the lensing potential appear at $l \ge 400$, spread on all the scales

1078: of the lens-induced $B$-mode spectrum, being already of the order of

1079: $\sim$7.5\% on all the multipoles $l \leq 1000$, and growing up to

1080: more than $\sim$70\% at $l=4096$.  As we have previously noticed, this

1081: non-linear effect on the magnetic polarization is simply explained

1082: considering that the lens-induced $B$-modes are very sensitive to the

1083: non-linear evolution of the cosmic structures.

1084: 

1085: For what concerns the non-linear effects, as noticed already in

1086: Sec.~3, it is important to stress again that, when considering the

1087: power spectra extracted from the difference maps (in which the primary

1088: CMB has been subtracted), we notice a very similar large-scale

1089: behavior between the electric and magnetic polarization, which

1090: strongly differs from the temperature trend. In order to analyse this

1091: effect, we have produced the \emph{linearly}-lensed $T$, $Q$, $U$ maps

1092: and subtracted the unlensed field.  In Fig.~\ref{linearNOlinear} we

1093: show the two cases: the ratios of the lensed $T$, $E$, $B$ spectra

1094: with respect to the corresponding spectra extracted from the

1095: linearly-lensed maps, and the same ratios extracted from the

1096: corresponding lensed and linearly-lensed \emph{difference} maps. While

1097: in the first case (top panel) the transfer of power from small scales

1098: to large scales is only visible in the lens-induced $B$-mode spectrum,

1099: in the second case (bottom panel) this power transfer is also

1100: observable in the lensed $E$-mode spectrum, with the same amount of

1101: $\sim$7.5\% up to $l\lesssim 100$ as for the lens-induced $B$-modes.

1102: The same does not occur to the lensed temperature even after

1103: subtracting the unlensed CMB.  Finally, moving to higher multipoles,

1104: the $E$ and $T$ signals converge to the same trend, while the $B$

1105: signal keeps on growing because of the non-linear power.

1106: 

1107: \section{One-point statistics}

1108: 

1109: As a second analysis, we compare the one-point probability density

1110: distributions (PDFs) of the lensed $T$, $Q$, $U$ maps with the PDFs of

1111: Gaussian distributions randomly generated using the corresponding mean

1112: and standard deviation values of the lensed maps.  The PDFs of both

1113: the simulated MS-modified-LP and unmodified-LP $T$, $Q$, $U$ maps do

1114: not show evidence of non-Gaussian features (upper panel of

1115: Fig.~\ref{TT_pdf}), and are consistent with the unlensed temperature

1116: and polarization distributions within a few per cent. This occurs thanks to the high

1117: angular resolution of our maps (see e.g. \citet{cumulants}),

1118: and is understandable because the main effect of CMB lensing is to transfer

1119: power among different scales, without generating new power (and we

1120: assume that the primary unlensed CMB field has a Gaussian

1121: distribution). 

1122: 

1123: On the other hand, it is well-known that the remapping induced by

1124: lensing onto the $T$, $Q$, $U$ fields generates a non-Gaussian

1125: signature in the lensed sky which is optimally characterized by higher

1126: order statistics as the bispectrum and trispectrum \citep[see

1127:   e.g.][and references therein]{Lewis06}. This is mainly due to the

1128: fact that the lensed CMB can be considered as a function of two

1129: fields, the unlensed sky and the lensing potential, which in first

1130: approximation can be assumed to be Gaussian (unmodified-LP case), even

1131: if the non-linear evolution of large-scale structures produces

1132: non-Gaussian features in the distribution of the projected potential

1133: (MS-modified-LP case). In particular, the non-Gaussianity associated

1134: with the large-scale structure induces non-Gaussian contributions to the distribution 

1135: of the lensing potential such that its $n$-point correlator in Fourier space will

1136: be non-vanishing for some value of $n$, and this, in turn, has an impact on

1137: the connected part of the $n$-point correlators of the lensed CMB (see e.g. \cite{cumulants}).

1138: Anyway, there exist also other phenomena that can produce

1139: non-Gaussian effects in the lensed CMB, for instance the correlations between CMB lensing and the 

1140: Sunyaev-Zeldovich effect.

1141: 

1142: Here, as a first characterization of the non-Gaussianity strength

1143: produced by CMB lensing, we consider the PDFs of the \emph{difference}

1144: $T$, $Q$, $U$ maps. In this way, we subtract the unlensed Gaussian sky

1145: and isolate the non-Gaussian term which physically generates the

1146: non-Gaussian signatures in the lensed CMB.  The obtained PDF in the

1147: temperature case is showed in the lower panel of Fig.~\ref{TT_pdf}

1148: (the $Q$ and $U$ cases have a similar trend).

1149: 

1150: The PDFs of the difference maps are characterized by a kurtosis excess

1151: with respect to Gaussian distributions randomly generated with the

1152: same mean and standard deviation values.  More precisely, averaging

1153: over different realizations, the kurtosis excess is $\sim 2.41$ for

1154: the difference $T$-map, $\sim 2.14$ for the difference $Q$-map and

1155: $\sim 2.16$ for the difference $U$-map.  Actually we would expect that

1156: the excess of non-linearities in the dark-matter distribution of the

1157: Millennium Simulation should show up as an excess of non-Gaussianity

1158: in the MS-modified-LP lensed difference $T$, $Q$, $U$ maps, with

1159: respect to the unmodified-LP case.  However, the one-point statistic

1160: is probably unable to capture this effect, and Fig.~\ref{TT_pdf} shows

1161: in fact that there is not a significant difference between the

1162: unmodified-LP and MS-modified-LP cases.  In this respect, we believe

1163: that a more suitable estimator should be developed in order to detect

1164: the contribution from the non-linear structure evolution to the total

1165: non-Gaussian statistics of the lensed maps. We reserve this analysis

1166: for future work.

1167: 

1168: \begin{figure}

1169: \begin{minipage}{1.0\linewidth}

1170: \centering

1171: \vspace{0.7cm}

1172: \includegraphics[width=0.9\textwidth]{T_tot_comparison_pdf_1D3.eps}

1173: \vspace{2.5cm}

1174: \end{minipage}

1175: \begin{minipage}{1.0\linewidth}

1176: \centering

1177: \includegraphics[width=0.9\textwidth]{T_diff_comparison_pdf_1D3.eps}

1178: \end{minipage}

1179: \vspace{1cm}

1180: \footnotesize

1181: \caption{{\em Top panel:} The PDF of the temperature maps obtained in

1182:   the MS-modified-LP and unmodified-LP cases compared to the PDF of

1183:   Gaussian distributions with the corresponding mean value and

1184:   standard deviation, respectively. {\em Bottom panel:} The same as in

1185:   the upper panel for the PDFs of temperature difference maps.}

1186: \label{TT_pdf}

1187: \end{figure}

1188: 

1189: \section{Conclusions}

1190: 

1191: We have constructed the first all-sky CMB temperature and polarization

1192: lensed maps based on a high-resolution cosmological $N$-body

1193: simulation, the Millennium Simulation.

1194: 

1195: To this purpose we have exploited the lensing potential map obtained

1196: using the map-making procedure developed in \cite{Carbone_etal2008}

1197: which integrates along the line-of-sight the MS dark matter structures

1198: by stacking and randomizing the simulation boxes up to $z=127$.

1199: Specifically, we have modified the LensPix code \citep{Lewis05} by

1200: supplying it with the spherical harmonic coefficients extracted from

1201: the MS lensing potential map and by implementing directly in the code

1202: itself the large-scale structure adding technique

1203: \citep[see][]{Carbone_etal2008} which allows to reinstate the

1204: large-scale power in the angular lensing potential that is not

1205: correctly sampled by the $N$-body simulation.  In this way we also

1206: preserve the correct correlation between the lensing potential and the

1207: ISW effect on multipoles $l \leq 400$ in the simulated temperature

1208: anisotropies.

1209: 

1210: Using our modified version of the LensPix code, we have constructed

1211: lensed CMB simulated maps with $\sim 5$ million pixels and an angular

1212: resolution of $\sim 1.72'$, based on potential fields calculated on

1213: $2560^3$ mesh cells from the Millennium Simulation.

1214: 

1215: After subtraction of the unlensed maps, the corresponding lensed $T$,

1216: $Q$, $U$ simulated maps reflect clearly the same large scale structure

1217: which is present in the modulus of the angular gradient of the lensing

1218: potential map (see Fig.~\ref{TP_diff_maps}).

1219: 

1220: We have also constructed the maps of the scalar and pseudo-scalar

1221: potentials $\psi_E$ and $\psi_B$ which are directly related to the

1222: electric and magnetic types of polarization, respectively.  Their

1223: difference maps, shown in

1224: Figs.~\ref{psiE_diff_maps}-\ref{psiB_diff_maps}, present distinct

1225: degree-scale distributions in the MS-modified-LP and unmodified-LP

1226: cases (where the latter is obtained from the unmodified LensPix code

1227: as previously explained), owing to the power transfer from non-linear

1228: to large scales in the lensed $E$ and $B$ fields.

1229: 

1230: As a quantitative study of the simulated maps, we have performed power

1231: spectrum and one-point statistics analyses. We find that the lensed

1232: $TT$, $TE$, $EE$ and $BB$ power spectra, obtained using the MS dark

1233: matter distribution, mostly overlap with the corresponding

1234: semi-analytic expectations on a range of multipoles up to $l\sim 2500$

1235: (see Figs.~\ref{TT_PS}-\ref{BB_PS}).  This latter result points out

1236: the effectiveness of the Millennium Simulation in reproducing

1237: correctly the findings of the theoretical approach.  Furthermore, an

1238: excess of power is observable in the MS-modified-LP case on larger

1239: multipoles, in particular for the $BB$ spectrum. We believe that this

1240: excess originates from the accurate inclusion of non-linear power in

1241: the Millennium Simulation, which is present in the MS lensing

1242: potential power spectrum as well (Fig.~\ref{projpot_PS}).  This

1243: outcome should be taken into account in the various delensing

1244: approaches, since these non-linear effects can have an impact on the

1245: quality of the reconstructed unlensed sky, particularly in view of the

1246: detection of primary $B$-modes.

1247: 

1248: Finally, we have derived one-point statistics both of the simulated

1249: $T$, $Q$, $U$ lensed maps and of the corresponding difference maps, in

1250: the MS-modified-LP and unmodified-LP cases.  The comparison of the

1251: resulting PDFs with respect to Gaussian random distributions with the

1252: same mean and standard deviation does not show any statistical

1253: difference when the total lensed maps are considered. On the other

1254: hand, we find that the difference maps are characterized by a kurtosis

1255: excess.  This result represents the distinctive weak-lensing effect of

1256: inducing non-Gaussianity in the unlensed Gaussian CMB field.  This is

1257: simply explained if in first approximation we think of the lensed

1258: field as the product of two Gaussian fields, i.e. the lensing

1259: potential and the primary unlensed CMB. Actually, correctly speaking,

1260: the lensing potential map, derived by integration of the MS dark

1261: matter distribution, preserves an intrinsic degree of non-Gaussianity

1262: due to the non-linear evolution of the cosmic structures, which at

1263: some level contributes to the non-Gaussian statistics of the simulated

1264: lensed CMB.  Unfortunately the one-point statistic is not sufficient

1265: to disentangle the two contributions, i.e. the dominant effect coming

1266: from the product of two different fields, and the subdominant effect

1267: coming from the non-linear matter evolution. This is the reason why in

1268: Fig.~\ref{TT_pdf} there seems to be no difference between the

1269: MS-modified-LP and unmodified-LP cases.  We reserve a more accurate

1270: characterization of non-Gaussianity in CMB lensing statistics to a

1271: future work.

1272: 

1273: \section*{Acknowledgments}

1274: CC warmly thank L. Verde for precious suggestions, and E. Gaztanaga,

1275: P. Fosalba, S. Leach, M. Liguori for helpful discussions.  CB thanks

1276: B. Menard for helpful suggestions.  Some of the results in this paper

1277: have been derived using the Hierarchical Equal Area Latitude

1278: Pixelization of the sphere \citep{Healpix}.  CC is supported through a

1279: Beatriu de Pinos grant.

1280: 

1281: \begin{thebibliography}{}

1282: \bibitem[Acquaviva\ \&\ Baccigalupi(2006)]{acquaviva_baccigalupi_2006}

1283: Acquaviva V., Baccigalupi C., 2006, \prrd {\bf 74}, 103510.

1284: \bibitem[Amblard\ \etal(2004)]{Amblard et al 04}

1285: Amblard A., Vale C., White M., 2004, New\ Astron.\ {\bf 9}, 687.

1286: \bibitem[Afshordi(2004)]{Afshordi04}

1287: Afshordi N., 2004, \prrd {\bf 70}, 083536.

1288: \bibitem[Bartelmann\ \& Schneider(2001)]{Matthias_rew}

1289: Bartelmann M., Schneider P., 2001, \physrep {\bf 340}, 291.

1290: \bibitem[Bunn\ \etal(2003)]{E_B_decomposition}

1291: Bunn E. F., Zaldarriaga M., Tegmark M., De Oliveira-Costa A.,

1292: 2003, \prrd {\bf 67}, 023501.

1293: \bibitem[Cabella\ \&\ Kamionkowski(2005)]{Cabella_Kamionkowski04}

1294: Cabella P., Kamionkowski M., 2005, arXiv:astro-ph/0403392v2.

1295: \bibitem[Carbone\ \etal(2008)]{Carbone_etal2008}

1296: Carbone C., Springel V., Baccigalupi C., Bartelmann M., Matarrese S.,

1297: 2008, \mnras {\bf 388}, 1618. 

1298: \bibitem[Challinor\ \&\ Chon(2002)]{Challinor02}

1299: Challinor A., Chon G., 2002, \prrd {\bf 66}, 127301.

1300: \bibitem[Challinor\ \& Lewis(2005)]{Challinor&Lewis05}

1301: Challinor A., Lewis A., 2005, \prrd {\bf 71}, 103010.

1302: \bibitem[Das\ \&\ Bode(2008)]{Das&Bode08}

1303: Das S., Bode P., 2008, \apj {\bf 682}, 1.

1304: \bibitem[Fosalba\ \etal(2007)]{Fosalba07} 

1305: Fosalba, P., Gaztanaga, E., Castander, F., Manera, M., 2007, 

1306: arXiv:0711.1540.

1307: \bibitem[Gorski\ \etal(2005)]{Healpix}

1308: Gorski K.M. \etal, 2005, \apj {\bf 622}, 759.

1309: \bibitem[Hamimeche\ \&\ Lewis(2008)]{Lewis_E4}

1310: Hamimeche S., Lewis A., 2008, \prrd {\bf 77}, 103013.

1311: \bibitem[Hirata\ \etal(2008)]{Seljak0108}

1312: Hirata C.M., Ho S., Padmanabhan N., Seljak U., Bahcall N.,

1313: 2008, \prrd {\bf 78}, 043520.

1314: \bibitem[Hirata\ \&\ Seljak(2003)]{Hirata&Seljak2003}

1315: Hirata C.M., Seljak U., 2003, \prrd {\bf 68}, 083002.

1316: \bibitem[Hu(2000)]{Hu2000}

1317: Hu W., 2000, \prrd {\bf 62}, 043007-1.

1318: \bibitem[Hu\ \&\ Dodelson(2002)]{review2}

1319: Hu W., Dodelson S., 2002,

1320: Ann.\ Rev.\ Astron.\ Astrophys.\ \textbf{40}, 171.

1321: \bibitem[Hu\ \etal(2006)]{hu_etal_2006}

1322: Hu W., Huterer D., Smith K.M., 2006, \apjl {\bf 650}, L13.

1323: \bibitem[Hu\ \&\ Okamoto(2002)]{Hu&Okamoto2002}

1324: Okamoto T., Hu W., 2002,

1325: \prrd {\bf 66}, 063008.

1326: \bibitem[Hu\ \etal(1998)]{Hu et al. 1998}

1327: Hu W., Seljak U., White M., Zaldarriaga M., 1998,

1328: \prrd \textbf{57}, 3290.

1329: \bibitem[Kamionkowski\ \&\ Kosowsky(1999)]{review1} 

1330: Kamionkowski M., Kosowsky A., 1999,

1331: Ann.\ Rev.\ Nucl.\ Part.\ Sci. \textbf{49}, 77.

1332: \bibitem[Kamionkowski\ \etal(1997)]{Kamionkowski et al. 1997} 

1333: Kamionkowski M., Kosowsky A., Stebbins A., 1997,

1334: Phys.\ Rev.\ D \textbf{55}, 7368. 

1335: \bibitem[Kesden\ \etal(2002)]{cumulants}

1336: Kesden M., Cooray A., Kamionkowski M., 2002,

1337: \prrd {\bf 66}, 083007. 

1338: \bibitem[Komatsu\ \etal(2008)]{WMAP5}

1339: Komatsu E. \etal, 2008, arXiv:0803.0547v2.

1340: \bibitem[Lewis(2005)]{Lewis05}

1341: Lewis A., 2005, \prrd {\bf 71}, 083008.

1342: \bibitem[Lewis\ \&\ Challinor(2006)]{Lewis06}

1343: Lewis A., Challinor A., 2006, \physrep {\bf 429}, 1.

1344: \bibitem[Ma\ \&\ Bertschinger(1995)]{maber}

1345: Ma C.P., Bertschinger E., 1995, \apj {\bf 455}, 7.

1346: \bibitem[Newman\ \&\ Penrose(1966)]{Penrose}

1347: Newman E. T., Penrose R., 1966, 

1348: J.\ Math.\ Phys.\ {\bf 7}, 863.

1349: \bibitem[Pritchard\ \&\ Kamionkowski(2004)]{Pritchard&Kamionkowski04}

1350: Pritchard J., Kamionkowski M., 2005,

1351: Annals\ Phys.\ \textbf{318}, 2.

1352: \bibitem[Refregier(2003)]{Refregier}

1353: Refregier A., 2003, Annu.\ Rev.\ Astron.\ Astrophys.\ {\bf 41}, 645.

1354: \bibitem[Reichardt\ \etal(2008)]{ACBAR}

1355: Reichardt C. L. \etal, 2008, arXiv:0801.1491v2.

1356: \bibitem[Seljak\ \&\ Hirata(2004)]{seljak_hirata_2004}

1357: Seljak U., Hirata C.M., 2004, \prrd {\bf 69}, 043005.

1358: \bibitem[Seljak\ \&\ Zaldarriaga(1997)]{Seljak&Zaldarriaga97}

1359: Seljak U., Zaldarriaga M., 1997,

1360: Phys.\ Rev.\ Lett. \textbf{78}, 2054.

1361: \bibitem[Smith\ \etal(2003)]{Halofit}

1362: Smith R.E. \etal, The Virgo Consortium, 2003,

1363: \mnras {\bf 341}, 1311.

1364: \bibitem[Smith S.\ \etal(2006)]{Smith_etal2006_bis}

1365: Smith S., Challinor A., Rocha G., 2006,

1366: \prrd {\bf 73}, 023517.

1367: \bibitem[Smith K.\ \etal(2006)]{Smith_etal2006}

1368: Smith K. M., Hu W., Kaplinghat M., 2006, \prrd {\bf 74}, 123002.

1369: \bibitem[Smith\ \etal(2007)]{smith_etal_2007}

1370: Smith K.M., Zahn O., Dore O., 2007, \prrd {\bf 76}, 043510.

1371: \bibitem[Spergel\ \etal(2003)]{spergel2003}

1372: Spergel D.N. \etal (2003), \apjs {\bf 148}, 175.

1373: \bibitem[Springel\ \etal(2005)]{Springel2005}

1374: Springel V., Frenk C.S., White S.D.M., 2006, Nature {\bf 1137}, 440.

1375: \bibitem[Teyssier\ \etal(2008)]{arXiv:0807.3651}

1376: Teyssier R. \etal, 2008, arXiv:0807.3651.

1377: \bibitem[Zaldarriaga\ \&\ Seljak(1997)]{Zaldarriaga&Seljak97}

1378: Zaldarriaga M., Seljak U., 1997, \prrd {\bf 55}, 1830. 

1379: \bibitem[Zaldarriaga\ \&\ Seljak(1998)]{Zaldarriaga&Seljak98} 

1380: Zaldarriaga M., Seljak U., 1998,

1381: Phys.\ Rev.\ D \textbf{58}, 023003.

1382: \end{thebibliography}

1383: 

1384: \end{document}

1385: