1: \documentclass[preprint]{aastex}
2: %\input MAP_cosmo
3:
4: \newcommand{\refeqn}[1]{(\ref{#1})}
5: \newcommand{\reffig}[1]{Figure~\ref{#1}}
6: \newcommand{\reftbl}[1]{Table~\ref{#1}}
7: \newcommand{\refsec}[1]{Section~\ref{#1}}
8: \newcommand{\WMAP}{\textsl{WMAP}}
9: \newcommand{\TBD}{\textbf{TBD}}
10: \newcommand{\aside}[1]{}
11: \newcommand{\muK}{{\mu {\rm K}}}
12: \newcommand{\VxV}{{V$\times$V}}
13: \newcommand{\VxW}{{V$\times$W}}
14: \newcommand{\WxW}{{W$\times$W}}
15:
16: \shorttitle{\WMAP\ 5-year Angular Power Spectra}
17: \shortauthors{Nolta et al.}
18: \slugcomment{Accepted by ApJS}
19:
20: \begin{document}
21: \title{Five-Year \textsl{Wilkinson Microwave Anisotropy Probe}
22: (\WMAP\altaffilmark{1}) Observations:\\ Angular Power Spectra}
23: \author{
24: {M. R. Nolta} \altaffilmark{2},
25: {J. Dunkley} \altaffilmark{3,4,5},
26: {R. S. Hill} \altaffilmark{6},
27: {G. Hinshaw} \altaffilmark{7},
28: {E. Komatsu} \altaffilmark{8},
29: {D. Larson} \altaffilmark{9},
30: {L. Page} \altaffilmark{4},
31: {D. N. Spergel} \altaffilmark{3,10},
32: {C. L. Bennett} \altaffilmark{9},
33: {B. Gold} \altaffilmark{9},
34: {N. Jarosik} \altaffilmark{4},
35: {N. Odegard} \altaffilmark{6},
36: {J. L. Weiland} \altaffilmark{6},
37: {E. Wollack} \altaffilmark{7},
38: {M. Halpern} \altaffilmark{11},
39: {A. Kogut} \altaffilmark{7},
40: {M. Limon} \altaffilmark{12},
41: {S. S. Meyer} \altaffilmark{13},
42: {G. S. Tucker} \altaffilmark{14},
43: {E. L. Wright} \altaffilmark{15}}
44: \altaffiltext{1}{WMAP is the result of a partnership between Princeton
45: University and NASA's Goddard Space Flight Center. Scientific
46: guidance is provided by the WMAP Science Team.}
47: \altaffiltext{2}{{Canadian Institute for Theoretical Astrophysics, 60 St. George St, University of Toronto, Toronto, ON Canada M5S 3H8}}
48: \altaffiltext{3}{{Dept. of Astrophysical Sciences, Peyton Hall, Princeton University, Princeton, NJ 08544-1001}}
49: \altaffiltext{4}{{Dept. of Physics, Jadwin Hall, Princeton University, Princeton, NJ 08544-0708}}
50: \altaffiltext{5}{{Astrophysics, University of Oxford, Keble Road, Oxford, OX1 3RH, UK}}
51: \altaffiltext{6}{{Adnet Systems, Inc., 7515 Mission Dr., Suite A100, Lanham, Maryland 20706}}
52: \altaffiltext{7}{{Code 665, NASA/Goddard Space Flight Center, Greenbelt, MD 20771}}
53: \altaffiltext{8}{{Univ. of Texas, Austin, Dept. of Astronomy, 2511 Speedway, RLM 15.306, Austin, TX 78712}}
54: \altaffiltext{9}{{Dept. of Physics \& Astronomy, The Johns Hopkins University, 3400 N. Charles St., Baltimore, MD 21218-2686}}
55: \altaffiltext{10}{{Princeton Center for Theoretical Physics, Princeton University, Princeton, NJ 08544}}
56: \altaffiltext{11}{{Dept. of Physics and Astronomy, University of British Columbia, Vancouver, BC Canada V6T 1Z1}}
57: \altaffiltext{12}{{Columbia Astrophysics Laboratory, 550 W. 120th St., Mail Code 5247, New York, NY 10027-6902}}
58: \altaffiltext{13}{{Depts. of Astrophysics and Physics, KICP and EFI, University of Chicago, Chicago, IL 60637}}
59: \altaffiltext{14}{{Dept. of Physics, Brown University, 182 Hope St., Providence, RI 02912-1843}}
60: \altaffiltext{15}{{UCLA Physics \& Astronomy, PO Box 951547, Los Angeles, CA 90095-1547}}
61: \email{nolta@cita.utoronto.ca}
62:
63: % -----------------------------------------------------------------------------
64: \begin{abstract}
65: We present the temperature and polarization angular power spectra of the
66: cosmic microwave background (CMB) derived from the first 5 years of
67: {\WMAP} data.
68: The 5-year temperature (TT) spectrum is cosmic variance limited up to multipole
69: $\ell=530$,
70: and individual $\ell$-modes have $S/N>1$ for $\ell<920$.
71: The best fitting six-parameter $\Lambda$CDM model
72: has a reduced $\chi^2$ for $\ell=33-1000$ of
73: $\chi^2/\nu=1.06$, with a probability to exceed of 9.3\%.
74: There is now significantly improved data near the third peak which
75: leads to improved cosmological constraints.
76: The temperature-polarization correlation (TE) is seen with high significance.
77: After accounting for foreground emission,
78: the low-$\ell$ reionization feature in the EE power spectrum
79: is preferred by $\Delta\chi^2=19.6$ for optical depth
80: $\tau=0.089$ by the EE data alone,
81: and is now largely cosmic variance limited for $\ell=2-6$.
82: There is no evidence for cosmic signal in the BB, TB, or EB spectra after
83: accounting for foreground emission.
84: %In a simple analysis we find that,
85: We find that, when averaged over $\ell=2-6$,
86: $\ell(\ell+1)C^{BB}_{\ell}/(2\pi) < 0.15\,\muK^2$ (95\% CL).
87: \end{abstract}
88:
89: \keywords{cosmic microwave background, cosmological parameters,
90: cosmology: observations, early universe, large-scale structure of universe,
91: space vehicles: instruments}
92:
93: % -----------------------------------------------------------------------------
94: \section{Introduction}
95:
96: %\aside{about WMAP}
97: %The Wilkinson Microwave Anisotropy Probe ({\WMAP}) is a NASA satellite mission
98: %to map the microwave sky at five frequency bands between 22-94 GHz.
99: %
100: %\aside{previous spectrum papers}
101: %\cite{hinshaw/etal:2003} \cite{kogut/etal:2003}
102: %\cite{hinshaw/etal:2007} \cite{page/etal:2007}
103: %
104: %This is one of a series of papers describing the analysis of the {\WMAP}
105: %5-year data, the others of which describe:
106: %production of maps from the time-ordered data \citep{hinshaw/etal:prep};
107: %beam modeling and profiles \citep{hill/etal:prep};
108: %characterization and removal of galatic
109: %foreground emission \citep{gold/etal:prep};
110: %catalog of point sources \citep{wright/etal:prep};
111: %extraction of cosmological parameters from the data
112: %\citep{dunkley/etal:prep};
113: %and the physical implications \citep{komatsu/etal:prep}
114: %
115: %\aside{about this paper}
116: %
117: %This paper is organized as follows. In \refsec{sec:changes} we describe
118: %changes in the of 5-year analysis relative to the 3-year analysis.
119: %In \refsec{sec:ptsrc} we discuss the unresolved point source correction to
120: %the TT spectrum.
121: %%In \refsec{sec:ttlowl} we present the new Gibbs-sampling estimate
122: %%of the lowest TT multipoles.
123: %In \refsec{sec:discuss} we present the power spectra.
124: %We summarize our conclusions in \refsec{sec:summary}.
125:
126: The {\WMAP} satellite \citep{bennett/etal:2003} has measured the temperature and polarization
127: of the microwave sky at five frequencies from 23 to 94~GHz.
128: \cite{hinshaw/etal:prep}
129: presents our new, more sensitive temperature and polarization maps.
130: After removing a model of the foreground emission from
131: these maps \citep{gold/etal:prep}, we obtain our
132: best estimates of the temperature and polarization
133: angular power spectra of the cosmic microwave background (CMB).
134:
135: This paper presents our statistical analysis of these CMB temperature and polarization maps.
136: Our basic analysis approach is similar to the approach described in our
137: first year {\WMAP} temperature analysis \citep{hinshaw/etal:2003} and polarization analysis \citep{kogut/etal:2003}
138: and in the three year {\WMAP} temperature \citep{hinshaw/etal:2007} and polarization analysis \citep{page/etal:2007}.
139: While most of the {\WMAP} analysis pipeline has been unchanged from our 3-year analysis, there
140: have been a number of improvements that have reduced the systematic errors and increased the precision
141: of the derived power spectra.
142: \cite{hinshaw/etal:prep} describe the {\WMAP} data processing with an
143: emphasis on these changes.
144: \cite{hill/etal:prep} present our more complete analysis of the {\WMAP} beams
145: based on 5 years of Jupiter data and physical optics fits to both
146: the A- and B-side mirror distortions.
147: The increase in main beam solid angle leads to a revision in the beam function that impacts our computed power spectrum by raising the overall
148: amplitude for $\ell>200$ by roughly 2\%.
149: \cite{gold/etal:prep} introduce a new set of masks that are designed to remove regions of free-free emission that were a minor (but detectable) contaminant
150: in analyses using the previous Kp2 mask used in the 1- and 3-year analysis.
151: %The new 5-year temperature mask (KQ85) retains 81.7\% of the sky,
152: %while the 3-year mask (Kp2) retained 84.6\% of the sky.
153: \cite{wright/etal:prep} updates the
154: point source catalog presented in \cite{hinshaw/etal:2007} finding 67 additional sources.
155: \refsec{sec:changes} of this paper describes these changes and their implications for the measured power spectra.
156:
157: \refsec{sec:tt} presents the temperature angular power spectrum (TT).
158: {\WMAP} has made a cosmic variance
159: limited measurement of the angular power spectrum to $\ell = 530$ and we now report results
160: into the ``third peak'' region.
161: The {\WMAP} results, combined with recent ground-based measurements
162: of the TT angular power spectrum
163: \citep{readhead/etal:2004,jones/etal:2006,reichardt/etal:prep}, result in
164: accurate measurements well into the ``fifth peak'' region.
165: For {\WMAP}, point sources are the largest astrophysical contaminant to the
166: temperature power spectrum.
167: We present estimates for the point source contamination based on
168: multi-frequency data, source counts and estimates from the bispectrum.
169:
170: The polarization observations are decomposed into $E$ and $B$ mode components
171: \citep{kamionkowski/kosowsky/stebbins:1997,seljak/zaldarriaga:1997}.
172: Primordial scalar fluctuations generate only $E$ modes, while
173: tensor fluctuations generate both $E$ and $B$ modes.
174: With $T$, $E$ and $B$ maps, we compute the angular auto-power spectra of the
175: three fields, $TT$, $EE$ and $BB$,
176: and the angular cross-power spectra of these three fields, $TE$, $TB$ and $EB$.
177: If the CMB fluctuations are Gaussian random fields, then these six angular
178: power spectra encode {\it all} of the statistical information in the CMB.
179: Unless there is a preferred sense of rotation in the universe,
180: symmetry implies that the $TB$ and $EB$ power spectrum are zero.
181: In \refsec{sec:tetb} we present both the TE and TB temperature-polarization
182: cross power spectra.
183: The {\WMAP} measurements of the TE spectrum now clearly see multiple peaks.
184: \aside{note from lyman: is this new? the second peak is a bit dicey}
185: The large angle TE anti-correlation is a distinctive signature of superhorizon
186: fluctuations \citep{zaldarriaga/spergel:1997}.
187: \cite{komatsu/etal:prep} discuss how the TB measurements constrain
188: parity-violating interactions.
189: \refsec{sec:eebb} presents both the EE and BB polarization power spectra.
190: The EE power spectrum now shows a clear $\sim 5 \sigma$ signature of cosmic
191: reionization.
192: \cite{dunkley/etal:prep} show that the amplitude of the signal implies that the
193: cosmic reionization was an extended process. \cite{dunkley/etal:prep} and
194: \cite{komatsu/etal:prep} discuss the cosmological implications of the angular
195: power spectrum measurements.
196:
197: % -----------------------------------------------------------------------------
198: \section{Changes in the 5-year Analysis\label{sec:changes}}
199:
200: The methodology used for the 5-year power spectra analysis is similar to
201: as that used for the 3-year analysis.
202: In this section we list the significant changes and their
203: impact on the results:
204:
205: \begin{itemize}
206: \item \cite{hinshaw/etal:prep} describe the changes in
207: the map processing and the resultant reduction in the absolute calibration
208: uncertainty from 0.5\% to 0.2\%.
209:
210: % mask fsky
211: % ------------------------
212: % dr3_85p0_r10 0.816891
213: % dr3_p06_r9 0.732789
214: % dr2_kp2_r10 0.845731
215: % dr2_p06 0.735495
216: %
217:
218: \item The temperature mask used to compute the power spectrum has been updated,
219: removing slightly more sky near the galactic plane, and more high-latitude
220: point sources \citep{gold/etal:prep}.
221: %The temperature mask used to compute the power spectrum is a union of three
222: %separate masks (processing, galactic, and point source),
223: %all of which were updated for the 5-year analysis.
224: %\aside{brief description of the new processing mask}
225: The galactic mask used in the 3-year and 1-year releases (Kp2) was constructed
226: by selecting all pixels whose K-band emission exceeded a certain threshold.
227: This procedure worked well in identifying areas contaminated by synchrotron
228: emission; however, it missed a few small regions contaminated by free-free,
229: particularly around $\rho$~Oph, the Gum nebula, and the Orion/Eridanus Bubble.
230: For the 5-year analysis we have constructed a new galactic mask to
231: remove these contaminated areas.
232: %Instead of just masking out bright K-band pixels,
233: %we mask out the brightest pixels in both K- and Q-band.
234: %The Kp2 mask removed the brightest 15\% of the K-band pixels,
235: %while the new mask removes the brightest 15\% in
236: %K-band and the brightest 15\% in Q-band.
237: %\aside{effect on the power spectrum; compare to 5yr data + Kp2}
238: %The point source mask removes discs of radius 0.6\degr around a list of
239: %positions of known sources. The list of sources
240: %includes {\WMAP} and other radio source catalogs.
241:
242: \cite{wright/etal:prep} updated the {\WMAP} point source catalog, finding
243: 390 sources in the 5-year data, 67 more sources than in the 3-year catalog
244: \citep{hinshaw/etal:2007}.
245: Of these 67 new sources, 32 were previously unmasked, and therefore
246: added to the 3-year source mask to create the 5-year source
247: mask.\footnote{Six of the sources in the 5-year catalog were not added to
248: the mask; they were found in a late update to the catalog after the mask had
249: been finalized.}
250:
251: All told, the new 5-year temperature power spectrum mask (KQ85)
252: retains 81.7\% of the sky,
253: while the 3-year mask (Kp2) retained 84.6\% of the sky.
254:
255: \item
256: The 5-year polarization mask is the same as the 3-year P06 mask
257: described in \cite{page/etal:2007}, except that an additional
258: 0.27\% of the sky has been removed due to combining P06 with the new
259: processing mask \citep{hinshaw/etal:prep}.
260:
261: \item
262: In addition to masking, the maps are further cleaned
263: of galactic foreground emission using external templates.
264: The cleaning procedure is very similar to that of the 3-year analysis;
265: see \cite{gold/etal:prep} for details.
266: For the temperature map, three templates are used:
267: a synchrotron template (the {\WMAP} $\rm K-Ka$ difference map),
268: an H$\alpha$ template as a proxy for free-free \citep{finkbeiner:2003},
269: and a thermal dust template \citep{finkbeiner/davis/schlegel:1999}.
270: For the polarization maps, two templates are used (since free-free is
271: unpolarized): the polarized K-band map and a polarized dust template
272: constructed from the unpolarized dust template, a simple model of the galactic
273: magnetic field, and polarization directions deduced from starlight.
274:
275: \item
276: A great deal of work has gone into improving the determination of the
277: beam maps and window functions \citep{hill/etal:prep}.
278: %With 5 years of data we now have ten seasons of Jupiter observations,
279: %averaging 50 days each, enabling us to push the measured beam profiles out
280: %to larger radii.
281: %Past this radius the signal-to-noise ratio is too poor to measure the
282: %beam profile, and a physical
283: %optics model is used to extend the beam maps into the low-signal region.
284: %This was done in the 3-year analysis, but only for the A-side. In the 5-year
285: %analysis both the A- and B-side optics models are computed, and furthermore
286: %the smallest scale resolved has been reduced by a factor of two relative
287: %to the 3-year analysis.
288: The main beam solid angles are larger than the 3-year estimates
289: by $\approx1$-2\% in V- and W-band.
290: Increased solid angle (i.e., greater map smoothing)
291: reduces the value of the transfer function $b_l$,
292: raising the deconvolved CMB power spectra.
293: The ratio of the 3-year to 5-year transfer functions can be seen in Figure~13 of
294: \cite{hill/etal:prep}; the net effect is to raise the TT power spectrum
295: by $\approx2\%$ for $\ell>200$, which is within the 3-year beam
296: $1\sigma$ confidence limits.
297: The beam transfer function uncertainty is smaller than the 3-year
298: uncertainty by a factor of $\approx2$.
299: The window function uncertainty is now
300: $\approx 0.6\%$ in $\Delta C_l/C_l$ for $200<\ell<1000$.
301:
302: \end{itemize}
303:
304: % -----------------------------------------------------------------------------
305: \section{Temperature Spectrum\label{sec:tt}}
306:
307: The 5-year $\ell\le32$ spectrum is described in \cite{dunkley/etal:prep}.
308: At low-$\ell$ the likelihood function is no longer well approximated by a
309: Gaussian so that we explicitly sample the likelihood function to evaluate
310: the statistical distribution of each multipole.
311:
312: We construct the 5-year TT spectrum for $\ell>32$ in the same fashion as
313: the 3-year spectrum; we refer the reader to \cite{hinshaw/etal:2007}
314: for details, and only briefly summarize the process as follows:
315:
316: \begin{itemize}
317: \item We start with the single year V1,V2,W1--W4
318: resolution-10 maps,\footnote{12,582,912 pixels ($N_{\rm side} = 1024$).}
319: masked by the KQ85 mask, and further cleaned
320: via foreground template subtraction.
321: \item The pseudo-$C_l$ cross power spectra are computed for each pair
322: of maps.
323: Two weightings are used: flat weighting and inverse noise variance
324: ($N_{\rm obs}$) weighting.
325: \item The year/DA cross power spectra are combined by band,
326: forming the \VxV, \VxW, and {\WxW} spectra.
327: The auto power spectra are not included in the combination, eliminating the
328: need to subtract a noise bias.
329: \item A model of the unresolved point source contamination
330: with amplitude $A_{ps}=0.011\pm0.001\,\muK^2{\rm sr}$
331: is subtracted from the band-combined spectra.
332: See \refsec{sec:ptsrc} for more details.
333: \item The \VxV, \VxW, \& {\WxW} spectra are optimally combined $\ell$-by-$\ell$
334: to create the final CMB spectrum.
335: \end{itemize}
336:
337: As in the 3-year analysis, the diagonal elements of the $\hat C_l$
338: covariance matrix are calculated as
339: \begin{eqnarray}
340: (\Delta \hat C_l)^2 = \frac{2}{(2l+1)f^2_{\rm sky}(l)}(C_l+N_l)^2
341: \end{eqnarray}
342: where $C_l$ is the cosmic variance term and $N_l$ the noise term.
343: The value of $f_{\rm sky}(l)$, the effective sky fraction,
344: is calibrated from simulations:\footnote{
345: The Markov chains in \cite{dunkley/etal:prep} and \cite{komatsu/etal:prep}
346: were run with a version of the {\WMAP} likelihood code with older and
347: slightly larger values for $f_{\rm sky}$.
348: The change in $f_{\rm sky}$ increased the TT errors by on average 2\%.
349: Rerunning the $\Lambda$CDM chain with the new $f_{\rm sky}$ leads to parameter
350: shifts of at most $0.1\sigma$.}
351: \begin{equation}
352: f_{\rm sky}(\ell) = \cases{
353: 0.826 -0.091 (\ell/500)^2, & $\ell\le500$; \cr
354: 0.777 -0.127 (500/\ell), & $\ell>500$.
355: }
356: \end{equation}
357:
358: The 5-year TT spectrum is shown in \reffig{fig:tthighl}.
359: With the greater S/N of the 5-year data the third acoustic peak is
360: beginning to appear in the spectrum.
361: The spectrum is cosmic variance limited up to $\ell=530$,
362: and individual $\ell$-modes have $S/N>1$ for $\ell<920$.
363: In a fit to the best cosmological $\Lambda$CDM model,
364: the reduced $\chi^2$ for $\ell=33-1000$ is
365: $\chi^2/\nu=1.06$, with a probability to exceed of 9.3\%.
366:
367: \reffig{fig:ttunbinned} compares the unbinned 5-year TT spectrum
368: with the 3-year result. Aside from the small upward shift of the 5-year
369: spectrum relative to that of the 3-year, due to the new beam transfer function,
370: they are identical at low-$\ell$.
371: \reffig{fig:ttfreq} shows the unbinned TT
372: spectrum broken down into its frequency components (\VxV, \VxW, \WxW),
373: demonstrating that the signal is independent of frequency.
374:
375: How much has the determination of the 3rd acoustic peak improved
376: with the 5-year data?
377: Over the range $\ell=680-900$, which approximately spans the rise
378: and fall of the 3rd peak (from the bottom
379: of the 2nd trough to the point on the opposite side of the peak),
380: %the rise and fall of the peak is preferred by
381: %$\Delta\chi^2=7.6$ versus a constant mean spectrum.
382: the fiducial spectrum is preferred over a flat mean spectrum
383: by $\Delta\chi^2=7.6$.
384: For the 3-year data it was $\Delta\chi^2=3.6$.
385: With a few more years of data, {\WMAP} should detect the curvature of
386: 3rd peak to greater than $3\sigma$.
387:
388: In \reffig{fig:tthighlvsothers} we compare the
389: {\WMAP} 5-year TT power spectrum along with recent results from other
390: experiments \citep{readhead/etal:2004,jones/etal:2006,reichardt/etal:prep},
391: showing great consistency between the various measurements.
392: Several on-going and future ground-based CMB experiments plan on calibrating
393: themselves off their overlap with {\WMAP} at the highest-$\ell$'s;
394: improving {\WMAP}'s determination of the 3rd peak will have the added benefit
395: of improving their calibrations.
396:
397: \subsection{Unresolved Point Source Correction\label{sec:ptsrc}}
398:
399: A population of point sources, Poisson-distributed over the sky, contributes
400: an additional source of white noise to the measured TT power spectrum,
401: $C^{TT}_l \to C^{TT}_l + C^{ps}$.
402: Given a known source distribution $N(>S)$, the number of sources per steradian
403: with flux greater than $S$, the point-source induced signal is
404: \begin{eqnarray}
405: C^{ps} = g(\nu)^2 \int^{S_c}_0{dS\,\frac{dN}{dS}S^2} \qquad[\muK^2{\rm sr}]
406: \label{eqn:Cps}
407: \end{eqnarray}
408: where $S$ is the source flux, $S_c$ is the flux cutoff (above which sources
409: are masked and removed from the map), and $g(\nu)=(c^2/2k\nu^2)r(\nu)$
410: converts flux density to thermodynamic temperature, with
411: \begin{eqnarray}
412: r(\nu)=\frac{(e^x-1)^2}{x^2e^x},
413: \quad x\equiv h\nu/kT_{\rm CMB}
414: \end{eqnarray}
415: converting antenna to thermodynamic temperature.
416:
417: At the frequencies and flux densities relevant for {\WMAP}, source counts
418: are dominated by flat-spectrum radio sources, which have flux spectra
419: that are nearly constant with frequency
420: ($S\sim\nu^\alpha$ with $\alpha\approx0$).
421: \cite{wright/etal:prep} finds the average spectral index of sources
422: bright enough to be detected in the {\WMAP} 5-year data to be
423: $\langle\alpha\rangle=-0.09$, with an intrinsic
424: dispersion of $\sigma_\alpha = 0.176$.
425: Since a source with flux $S\sim\nu^\alpha$ has a thermodynamic temperature
426: $T\sim\nu^{\alpha-2}r(\nu)$,
427: we model the frequency dependence of $C^{ps}$ as
428: \begin{eqnarray}
429: C^{ps}({\nu_i,\nu_j}) = A_{ps} r(\nu_i)r(\nu_j)
430: \left(\frac{\nu_i\nu_j}{\nu^2_Q}\right)^{\alpha-2}
431: \label{eqn:srcmodel}
432: \end{eqnarray}
433: where $\nu_{i,j}$ are the frequencies of the two maps used to calculate
434: the TT spectrum, $A_{ps}$ is an unknown amplitude,
435: and $\nu_Q=40.7\,{\rm GHz}$ is the Q-band central frequency.
436:
437: In this section, we estimate the value of $A_{ps}$ needed to correct the TT
438: power spectrum, finding $A_{ps} = 0.011\pm0.001\,\muK^2{\rm sr}$,
439: and discuss incorporating its uncertainty into the likelihood function.
440:
441: \subsubsection{Estimating the correction}
442:
443: For a fixed beam size, flat-spectrum radio sources are much fainter in
444: the W-band temperature maps than in Q- or V-band,
445: allowing us to use the frequency dependence of the TT spectrum at high-$\ell$
446: to constrain the value of $A_{ps}$.
447: As in previous releases, the estimator we use is
448: \begin{eqnarray}
449: \hat A_{ps} &=& \frac{\sum_{l\alpha\beta} C^{\alpha}_l(\Sigma^{-1})^{\alpha\beta}_lh^\beta_l}{\sum_{l\alpha\beta} s^{\alpha}_l(\Sigma^{-1})^{\alpha\beta}_lh^{\beta}_l}
450: \label{eqn:Apsestimator} \\
451: h^{\gamma}_l & = & s^{\gamma}_l - \frac{\sum_{\alpha\beta}s^{\alpha}_l(\Sigma^{-1})^{\alpha\beta}_l}{\sum_{\alpha\beta}(\Sigma^{-1})^{\alpha\beta}_l}
452: \end{eqnarray}
453: where greek letters represent a pair of frequencies (e.g., VW),
454: $C^{\alpha}_l$ is the measured TT cross-power spectrum,
455: $\Sigma^{\alpha\beta}_l$ is the $\langle C^{\alpha}_lC^{\beta}_l\rangle$
456: covariance matrix including cosmic variance and detector noise,
457: and $s^{\alpha}_l = l(l+1)C^{ps}(\alpha)/2\pi$.
458: The inverse estimator variance ($[\delta\hat A_{ps}]^{-2}$)
459: is given by the denominator of \refeqn{eqn:Apsestimator}.
460: While $\Sigma^{\alpha\beta}_l$ does not include the off-diagonal
461: coupling due to the mask, the diagonal elements are renormalized to
462: account for the loss of sky coverage.
463:
464: Measured values for $A_{ps}$ are listed in \reftbl{tbl:Aps} for various
465: frequency combinations (QVW \& VW) and galactic masks (KQ85, KQ80, \& KQ75).
466: The QVW estimates are insensitive to the galactic mask; the VW estimate
467: increases somewhat as more of the sky is masked.
468: Both the QVW and VW estimates prefer the same value
469: ($\approx0.011\,\muK^2{\rm sr}$)
470: of $A_{ps}$ when the KQ75 mask is used.
471: While we restrict the data to $\ell=300-800$,
472: the QVW estimate is only a weak function of the chosen $\ell$-range;
473: \reffig{fig:ptsrc} shows $A_{ps}$ estimated in bins of width $\Delta\ell=100$.
474: We adopt $A_{ps}=0.011\pm0.001\,\muK^2{\rm sr}$ as our correction to the
475: final combined TT spectrum.
476: The consistency between $\ell$-bins and between QVW and VW seen
477: in \reffig{fig:ptsrc} is an important null test for the angular power spectrum.
478: $A_{ps}$ (VW) is proportional to the power in the (V-W) map in a given
479: $\ell$-range.
480: \reffig{fig:ttnull}
481: shows no evidence for any detectable residual signal in the VW maps after point source subtraction.
482:
483: Because radio sources can only have positive flux they introduce a positive
484: skewness to the maps, which can be detected in searches for non-Gaussianity.
485: \cite{komatsu/etal:prep} estimated the bispectrum induced by sources,
486: finding $b^{ps} = (4.3\pm1.3)\times10^{-5}\,\muK^3{\rm sr}^2$ at Q-band.
487: \aside{note from David: I am nervous that our point source selection at
488: K band slightly skews the maps toward a negative value that leads to an
489: underestimate of $b^{ps}$}
490: Is this consistent with the value of $A_{ps}$ measured from the power
491: spectrum?
492: Given a theoretical model for the source number counts $N(>S)$, one can
493: predict the measured values of $C^{ps}$ and $b^{ps}$.
494: Several models exist in the literature; we tested our results against two,
495: \citet[Tof98]{toffolatti/etal:1998}\footnote{In \cite{bennett/etal:2003}
496: we found that the Tof98 model needed to
497: be rescaled by a factor of 0.66 to match the {\WMAP} 1-year number counts;
498: \cite{wright/etal:prep} refined the rescaling factor to 0.64 to match the
499: {\WMAP} 5-year source counts.}
500: and \citet[deZ05]{dezotti/etal:2005}.
501: $C^{ps}$ is calculated via \refeqn{eqn:Cps}, and $b^{ps}$ from
502: \begin{eqnarray}
503: b_{\rm src} = g^3(\nu) \int_0^{S_c}{dS\,\frac{dN}{dS}S^3} .
504: \end{eqnarray}
505: where $g(\nu)$ and $S_c$ are defined in \refeqn{eqn:Cps}.
506: The comparison is complicated by the fact that $S_c$ is unknown.
507: We mask out not only the sources detected in {\WMAP} data, but
508: also undetected sources from external catalogues that are likely to
509: contribute contaminating flux.
510: However, a single value of $S_c$ predicts both $C^{ps}$ and $b^{ps}$,
511: so we can in principle tune $S_c$ to match one, and see if it agrees with the
512: other.
513: In \reftbl{tbl:ptsrc} we compare our measured values of $C^{ps}$
514: and $b^{ps}$ with the rescaled Tof98 and deZ05 predictions for
515: several values of $S_c$.
516: There is some tension between the measured values and the model predictions.
517: Given our measured value for $b_{ps}$ the models would prefer a smaller
518: value for $A_{ps}$, in the range 0.008-$0.010\,\muK^2{\rm sr}$.
519: For the Tof98 model, the $S_c\approx0.52$ predictions
520: are within $1\sigma$ of both $C^{ps}$ and $b^{ps}$.
521: However, the deZ05 model appears to be discrepant, and a single value for $S_c$
522: cannot match both $C^{ps}$ and $b^{ps}$.
523: %An effective flux cut of $S_c=0.5\,{\rm Jy}$ appears consistent with our
524: %measured results:
525: %the rescaled Tof98 model predicts $N(>S)=
526: % 57.5 & 10.4 & 4.9 \\
527: %The interpretation is a bit murky. If $A_{ps}=0.011$, the $dN/dS$ models
528: %suggest we should see $b_{ps}\approx9\times10^{-5}$, and yet the measured
529: %value of \aside{not the final number} $b_{ps}=(4\pm1)\times10^{-5}$
530: %indicates $A_{ps}\approx0.009$.
531:
532: Other groups have independently estimated the unresolved source contamination,
533: and their results are in general agreement with ours.
534: When the 3-year data was initially released the correction was
535: $A_{ps}=0.017\pm0.002\,\muK^2{\rm sr}$.
536: \cite{huffenberger/eriksen/hansen:2006}
537: reanalyzed the data and claimed $A_{ps}=0.011\pm0.001$,
538: noticing that $A_{ps}$ was sensitive to the choice of galaxy mask;
539: using the Kp0 mask instead of Kp2 reduced the value of $A_{ps}$.
540: Revisiting our original estimate for the 3-year analysis,
541: we reduced the correction to $0.014\pm0.003$ for the published papers.
542: In a subsequent paper, \cite{huffenberger/etal:2007prep}, the same group
543: corrected their original estimate after finding a small error, finding
544: $0.013\pm0.001$, consistent with our published result.
545:
546: % -----------------------------------------------------------------------------
547: \section{Temperature-Polarization Spectra\label{sec:tetb}}
548:
549: %There has been no significant change in the TE analysis pipeline.
550: %However, we have extended the pipeline so that it now also computes the TB
551: %spectrum at high-$\ell$.
552:
553: %\reffig{fig:tehighl} shows the high-$\ell$ {\WMAP} TE spectrum computed from
554: %the V and W band data.
555: %With five years of data, the negative valley at $\ell = -140$ and the peak at
556: %$\ell = 330$ are clearly seen in the data.
557: %This structure confirms that the fluctuations are primarily adiabatic and
558: %help constrain the presence of isocurvature modes.
559: %
560: %At low-$\ell$, the TE correlations are evaluated as part of the map-based
561: %analysis in the likelihood function and provide additional
562: %(albeit relatively weak) constraints on the reionization history.
563: %For example, if you zero out the $\ell<10$ multipoles of the fiducial TE
564: %\& EE spectra and calculate the change in $chi^2$, only 12\% of the change
565: %is due to TE.
566: %\cite{dunkley/etal:prep} discusses the TE and EE constraints on reionzation.
567:
568: %The TB spectrum is shown in \reffig{fig:tbhighl}, showing no evidence
569: %for cosmological signal. Over $\ell=24-450$ the reduced null $\chi^2$ is 0.97.
570: %The TB spectrum is a measure of parity violation;
571: %\cite{komatsu/etal:prep} describes how this spectrum is used to constrain the
572: %presence of any parity violating terms coupled to photons.
573:
574: % ----------------------------------------------------------------------------
575:
576: The standard model of adiabatic primordial density fluctuations
577: predicts a correlation between the temperature and polarization
578: fluctuations. The temperature traces
579: primarily the density, and E-mode polarization the velocity,
580: of the photon-baryon plasma at recombination.
581: The correlation was seen in earlier WMAP data by
582: \cite{kogut/etal:2003} and \cite{page/etal:2007}. The anti-correlation
583: near $\ell=30$ provides evidence that fluctuations exist on superhorizon
584: scales, as it is observed on an angular scale larger than the acoustic
585: horizon at decoupling \cite{spergel/zaldarriaga:1997}.
586:
587: No significant changes have been made in the five-year TE analysis.
588: We continue to use the method described in \cite{page/etal:2007} to
589: compute the TE power spectrum. The inputs are the
590: KaQV polarization maps \citep{gold/etal:prep}, and the
591: VW temperature maps. For high multipoles
592: $\ell>23$, the likelihood can be approximated
593: as a Gaussian, and we continue to use the ansatz given in
594: Appendix C of \cite{page/etal:2007} to compute the covariance matrix.
595: At low multipoles, $\ell \le 23$, the likelihood of the
596: polarization data is evaluated directly from the maps,
597: following Appendix D in \cite{page/etal:2007}.
598:
599: \reffig{fig:tehighl} shows the TE spectrum. At low-$\ell$ the
600: spectrum and error bars are approximated using the Gaussian form, although
601: these are not used for cosmological analysis. With five years of data the
602: anti-correlation at $\ell=140$ is clearly seen in the data, and the
603: correlation at $\ell=300$ is measured with higher accuracy.
604: The second anti-correlation at $\ell \sim 450$ is now better characterized,
605: and is consistent with predictions of the $\Lambda$CDM model.
606: The structure tests the consistency of the simple model, which fits
607: both the TT and TE spectra with only six parameters. The best-fit
608: $\Lambda$CDM model has $\chi^2=415$ for the TE component, with
609: 421 degrees of freedom, giving $\chi^2/\nu=0.99$.
610: The consistency confirms that the
611: fluctuations are predominantly adiabatic, and
612: constrains the amplitude of isocurvature modes.
613:
614: The signal at the lowest multipoles, evaluated using the
615: exact likelihood, is used to provide additional
616: constraints on the reionization history. Although small, the
617: measurement is consistent with the EE signal, and consistent with
618: the three-year WMAP observations \cite{page/etal:2007}.
619: \cite{dunkley/etal:prep} discuss constraints on reionization.
620:
621: No correlation is expected between the temperature and the
622: B-mode polarization. The TB spectrum is
623: therefore primarily used as a null test, and is shown in
624: \reffig{fig:tbhighl}. It is consistent with no signal, as expected;
625: over $\ell=24-450$ the reduced null $\chi^2$ is 0.97.
626: This measurement is used in \cite{komatsu/etal:prep}
627: to place constraints
628: on the presence of any parity violating terms coupled to
629: photons, that could produce a TB correlation.
630: We now include the TB spectrum at high $\ell$ as an optional module for the
631: likelihood code.
632:
633: % -----------------------------------------------------------------------------
634: \section{Polarization Spectra\label{sec:eebb}}
635:
636: %%The large-scale noise structure of the {\WMAP} polarization maps is
637: %%sufficiently complicated to render pseudo-$C_\ell$ based estimates of the
638: %%likelihood $L(d|C_\ell)$ too imprecise to be useful \citep{page/etal:2007}.
639: %%The pseudo-$C_\ell$ spectra were adequate for plotting purposes; however,
640: %%with the 5-year release they are now unsuitable even for plotting.
641: %%In particular, BB $\ell=3$ is poorly represented.
642: %%This mode is sensitive to long-wavelength baseline modes in the timestream
643: %%and thus poorly measured by {\WMAP}.
644: %
645: %
646: %The W band polarization maps are the least reliable WMAP products.
647: %Because of the anomalously high $\ell=7$ EE value computed by the
648: %pseudo-$C_\ell$ algorithm, we have avoided using the W-band maps in
649: %cosmological analysis and have resorted to using it as an additional check on
650: %various models.
651: %Figure~8 of \cite{hinshaw/etal:prep} shows that the $\ell=7$ value,
652: %while high, appears to be consistent with being in the tail of a properly
653: %computed likelihood distribution.
654: %The W-band $\ell=7$ problem may be a signature of poor statistics rather than
655: %a systematic.
656: %However, more data is needed to understand this potential anomaly.
657: %
658: %The low-$\ell$ EE power spectrum is shown in \reffig{fig:eelowl}.
659: %If we zero out the $\ell<10$ portion of the fiducial EE \& TE spectra
660: %the $\chi^2$ increases by 22.3, of which 2.7 is due to TE.
661: %Thus the reionization feature in the EE power spectrum
662: %is preferred by $\Delta\chi^2=19.6$.
663: %The $\ell=2$, 3, 4, \& 6 multipoles are cosmic variance limited, and
664: %the S/N ratio for the combined $\ell=2-7$ bandpower is 11.
665: %The low-$\ell$ BB power spectrum shows no evidence of cosmological signal,
666: %and we limit $\ell(\ell+1)C^{BB}_{\ell=2-6}/(2\pi) < 0.15$ (95\% CL),
667: %including cosmic variance.
668: %
669: %At high multipoles, {\WMAP} lacks the S/N to meaningfully detect
670: %the acoustic peaks in the EE spectrum, in the sense of reducing
671: %cosmological parameter uncertainties.
672: %However, there are hints of signal in the data consistent with the
673: %standard $\Lambda$CDM model.
674: %The 5-year high-$\ell$ EE spectrum is shown in \reffig{fig:eehighl},
675: %along with recent results from ground-based experiments
676: %\citep{leitch/etal:2005,montroy/etal:2006,sievers/etal:2007}.
677: %For $\ell=50-800$, $\chi^2=859.1$ assuming $C^{EE}_l=0$,
678: %and drops by $\chi^2$ by 8.4, or almost $3\sigma$, assuming the
679: %standard $\Lambda$CDM model.
680: %For the 3-year data the equivalent change in $\chi^2$ was 6.2.
681: %The high-$\ell$ BB spectrum is consistent with no signal,
682: %having a reduced $\chi^2$ of 1.02 over $\ell=50-800$.
683:
684: % -----------------------------------------------------------------------------
685:
686: Due to its thermal stability \citep{jarosik/etal:2007}
687: and well-characterized gain,
688: {\WMAP} can measure polarization signals even though the scan pattern was not
689: optimized for doing so.
690: %In the analysis, Stokes Q and U maps are generated from the
691: %detector signals as described in \cite{page/etal:2007}.
692: %The coordinate independent E-mode
693: %and B-mode power spectra, EE and BB respectively,
694: %are generated directly from the Q and U maps.
695: %
696: The polarization signal is manifested in the time ordered data (TOD) differently
697: from the temperature signal.
698: As a result, some of the low-$\ell$ polarization multipoles
699: are well sampled and other multipoles are poorly sampled and have large
700: statistical errors \citep{hinshaw/etal:2007,page/etal:2007}.
701: This is a rather different situation than from that of the temperature
702: spectrum, and the data must be analyzed with some care.
703:
704: %For example, when we analyze the $\ell=2$ temperature power spectrum,
705: %we use the likelihood
706: %function rather than Gaussian errors as the Gaussian approximation starts to
707: %break down. In effect we sample just four of the five possible $Y_{l=2,m}$
708: %multipoles,
709: %corresponding to $f_{\rm sky}$ being $\approx0.8$.
710: %For polarization, this effect is even more dramatic as our scan pattern
711: %significantly lower the effective number of multipoles measured at each $l$,
712: %particularly for EE $\ell=2$, 5, 7 and 9 and BB $\ell=3$
713: %\citep[the peaks seen in Figure 16 in]{page/etal:2007}.
714: %\reffig{fig:simplebb} and \reffig{fig:simpleee} demonstrate the importance of using the full likelihood
715: %description. The figures compares the to the conditional likelihoods.
716: %The conditional likelihoods are computed using the {\WMAP} likelihood code
717: %by varying the multipole in question, and keeping the rest of the spectrum fixed to the
718: %fiducial best-fit $\Lambda$CDM model. From the plots it is clear that the best estimate of
719: %the mean and the uncertainty are not captured with the pseudo-$C_l$ estimates. One must keep in mind,
720: %however, that the the likelihood distributions include "cosmic variance" whereas the
721: %pseudo-$C_l$ estimates do not.
722:
723: %%{\WMAP}'s scan pattern was not optimized for low-$\ell$ polarization
724: %%measurement, though it is sufficient for our purposes.
725: %Nevertheless, some multipoles are well sampled and other multipoles are poorly
726: %sampled and have large statistical errors
727: %\citep{hinshaw/etal:2007,page/etal:2007}.
728:
729: When we analyze the $\ell=2$ temperature power spectrum, we use the likelihood
730: function rather than Gaussian errors, as the Gaussian approximation starts to
731: break down with only $\approx4$ effective modes measured in the map
732: (the reduction is due to $f_{\rm sky}\approx0.7$).
733: For polarization, this effect is even more dramatic, as our scan pattern
734: significantly lowers the effective number of multipoles measured,
735: particularly for EE $\ell=2$, 5, 7 and 9 and BB $\ell=3$
736: \citep[the peaks seen in Figure 16 in][]{page/etal:2007}.
737: \reffig{fig:simplebb} demonstrates the importance of using the full likelihood
738: description.
739: The Figure shows both the pseudo-$C_l$ estimates of the
740: $\ell=2-7$ BB multipoles and the conditional likelihoods
741: computed using the {\WMAP} likelihood code by varying the multipole in
742: question, keeping the rest of the spectrum fixed to the
743: fiducial best-fit $\Lambda$CDM model.
744: %While the pseudo-$C_l$ estimate of BB $\ell=3$ implies a significant
745: %detection of power, the full likelihood code shows this to not be the case.
746: From the plots it is clear that the best estimates of
747: the mean and the uncertainty are not attained with the pseudo-$C_l$ estimates.
748:
749: We next consider the low-$\ell$ EE and BB power spectra in more detail.
750: The low-$\ell$ EE power spectrum is shown in \reffig{fig:eelowl}.
751: The uncertainties are obtained from the conditional likelihood
752: and include cosmic variance;
753: thus one cannot double the error flags to get the 95\% confidence limits.
754: If we zero out the $\ell<10$ portion of the fiducial EE \& TE spectra
755: the $\chi^2$ increases by 22.3, of which 2.7 is due to TE.
756: Thus the reionization feature in the EE power spectrum
757: is preferred by $\Delta\chi^2=19.6$.
758: The $\ell=2$, 3, 4, \& 6 multipoles are cosmic variance limited, and
759: the S/N ratio for the combined $\ell=2-7$ bandpower is 11.
760:
761: Considerable effort has gone into understanding the W-band $\ell=7$ EE signal.
762: Because of the apparent anomalously high $\ell=7$ EE value computed by the
763: pseudo-$C_\ell$ algorithm, we have avoided using the W-band maps in
764: cosmological analysis and use them only as an additional check on
765: various models. Figure~8 of \cite{hinshaw/etal:prep} shows that the $\ell=7$
766: value, while high, appears to be consistent with being in the tail of a properly
767: computed likelihood distribution.
768: The W-band $\ell=7$ problem may be a signature of
769: poor statistics rather than a systematic.
770: However, more data are needed to understand this potential anomaly.
771: The $\ell=3$ BB signal gives perhaps the clearest example of the importance
772: of using the full likelihood code.
773: While the pseudo-$C_l$ estimate implies a significant
774: detection of power, the full likelihood code shows this to not be the case.
775: The physical cause of the large uncertainty is that with our scan strategy
776: an $\ell=3$ BB signal resembles an offset in the data and thus is not well
777: separated from the baseline \citep{page/etal:2007,hinshaw/etal:prep}.
778:
779: We see no evidence for a B-mode signal at low $\ell$, limiting the possible
780: level to $\ell(\ell+1)C^{BB}_{\ell=2-6}/(2\pi) < 0.15~\mu$K$^2$ (95\% CL),
781: including cosmic variance. With $\tau= 0.1$ and $r=0.2$, a typical estimate for currently favored models of inflation, $\ell(\ell+1)C^{BB}_{\ell=2-6}/(2\pi)
782: \approx0.008~\mu$K$^2$.
783: Since a signal of $0.15~\mu$K$^2$ corresponds roughly to $r\approx20$,
784: one can see that {\WMAP}'s limit is not based on the BB data, but on the tensor
785: contribution to the
786: TT and EE spectra as discussed in \cite{komatsu/etal:prep}.
787:
788: For EE at $\ell>10$, there are hints of signal in the data consistent with the
789: standard $\Lambda$CDM model. However, the significance is not great enough to
790: contribute to knowledge of the cosmological parameters.
791: The 5-year high-$\ell$ EE spectrum is shown in \reffig{fig:eehighl},
792: along with recent results from ground-based experiments
793: \citep{leitch/etal:2005,montroy/etal:2006,sievers/etal:2007}.
794: For $\ell=50-800$, $\chi^2=859.1$ assuming $C^{EE}_l=0$,
795: and drops by 8.4, or almost $3\sigma$, assuming the
796: standard $\Lambda$CDM model.
797: %For the KaQVW combinations the significance is greater.
798: For the 3-year data the equivalent change in $\chi^2$ was 6.2.
799:
800: The high-$\ell$ BB spectrum is consistent with no signal,
801: having a reduced $\chi^2$ of 1.02 over $\ell=50-800$ for the QV data.
802: The lack of any signal in the low and high $\ell$ BB data is a necessary
803: check of the foreground subtraction. As seen in Page et al. (2007), foreground
804: emission produces E-modes and B-modes at similar levels; thus the absence of a
805: B-mode signal suggests that the level of contamination in the E-mode
806: signal is low.
807: This is quantified in \cite{dunkley/etal:prep}.
808:
809: % -----------------------------------------------------------------------------
810: \section{Summary and Conclusions\label{sec:summary}}
811:
812: We have presented the temperature and polarization angular power spectra of the
813: cosmic microwave background (CMB) derived from the first 5 years of
814: {\WMAP} data.
815: With greater integration time our determination of the third acoustic
816: peak in the TT spectrum has improved.
817: The low-$\ell$ reionization feature in the EE spectrum is now detected
818: at nearly $5\sigma$.
819: The TB, EB, \& BB spectra show no evidence for cosmological signal.
820: The spectra are in excellent agreement with the best
821: fit $\Lambda$CDM model.
822: Our knowledge of the power spectrum is improving
823: both due to more detailed analyses, better modeling and understanding of the
824: foreground emission, and more integration time.
825: %There is as yet no evidence that we will hit a systematic barrier.
826:
827: All of the 5-year {\WMAP} data products are being made available through the
828: Legacy Archive for Microwave Background Data Analysis
829: (LAMBDA\footnote{\url{http://lambda.gsfc.nasa.gov/}}),
830: NASA's CMB Thematic Data Center.
831: The temperature and polarization angular power spectra presented here are
832: available, as is the WMAP likelihood code which
833: incorporates our estimates of the Fisher matrix, point sources and
834: beam uncertainties.
835:
836: \acknowledgements
837:
838: The {\WMAP} mission is made possible by the support of the Science Mission
839: Directorate Office at NASA Headquarters.
840: This research was additionally supported by NASA grants NNG05GE76G,
841: NNX07AL75G S01, LTSA03-000-0090, ATPNNG04GK55G, and ADP03-0000-092.
842: EK acknowledges support from an Alfred P. Sloan Research Fellowship.
843: This research has made use of NASA's Astrophysics Data System Bibliographic
844: Services.
845: We acknowledge use of the CAMB, CMBFAST, CosmoMC, and HEALPix
846: \citep{gorski/etal:2005} software packages.
847:
848: \appendix
849:
850: \section{Likelihood treatment of source/beam uncertainties\label{sec:appsrclike}}
851:
852: In this section, we test the treatment of the unresolved source
853: correction and beam uncertainties in the {\WMAP} likelihood code,
854: and show that it produces the correct results for cosmological
855: parameters.
856:
857: We adopt the same likelihood treatment of the unresolved point source
858: correction uncertainty for the 5-year likelihood code as used in the 3-year code
859: \citep[Appendix A]{hinshaw/etal:2007}, updated for the 5-year value of $A_{ps}$.
860: Briefly, a correction to the logarithmic likelihood,
861: ${\cal L}\equiv -2\ln L= {\cal L}_0 + {\cal L}_1$
862: where ${\cal L}_0$ is the standard likelihood and ${\cal L}_1$ the
863: combined source \& beam correction,
864: is calculated assuming the $C_l$ are normally distributed, a reasonable
865: assumption at high $\ell$.
866:
867: \citet[Huf08]{huffenberger/etal:2007prep} disagreed with
868: the source \& beam likelihood module used in the 3-year analysis,
869: pointing out that the uncertainty in $n_s$ (the index of primordial scalar
870: perturbations) was unchanged even if the uncertainty in $A_{ps}$ was
871: increased by a factor of 100 (Fig. 2 in their paper).
872: They proposed an alternative approach,
873: integrating the beam/point source covariance
874: matrix into the cosmic variance/noise/mask covariance matrix and inverting the
875: result in order to compute ${\cal L}$ directly,
876: instead of calculating ${\cal L}_1$ as a separate correction.
877: Using this form of the likelihood, as $\delta A_{ps}$ was increased,
878: the uncertainty in $n_s$ increases (albeit modestly; $\delta n_s$ increased
879: by 38\% when $\delta A_{ps}\to100\times \delta A_{ps}$).
880:
881: However, while we agree that it is striking that the error in $n_s$
882: is seemingly unaffected by the uncertainty in $A_{ps}$, we have some concerns
883: regarding the Huf08 approach.
884: To quote Huf08, ``the errors on the source measurement do not make much
885: difference, as long as $[\delta A_{ps}] < 0.003\,[\muK^2{\rm sr}]$'', and their Fig.~2 implies
886: the same holds true when $\delta A_{ps}=0.003$. This value is significant,
887: because it is the uncertainty adopted for the 3-year {\WMAP} analysis.
888: When Huf08 adopted the same uncertainty,
889: they found the same absolute uncertainty in
890: $n_s$ as the {\WMAP} team, but their central value was shifted higher by 0.005.
891: This shift persisted as $\delta A_{ps}\to0$, and thus was seemingly not due to
892: the point source uncertainty. The conclusion we draw is that they
893: found the same value of $\delta n_s$ as the {\WMAP} 3-year analysis,
894: but their value of $n_s$ was biased high because of the way they treat the
895: beam uncertainties.
896: We believe the Huf08 value of $n_s$ would be in agreement with that found in
897: WMAP3, but that it is biased high due to their treatment of beam uncertainties.
898:
899: Huf08 quoted the value of ${\cal L}_1$ computed with their
900: alternative likelihood module for a particular CMB spectrum distributed with
901: the {\WMAP} 3-year likelihood code test program,
902: finding ${\cal L}_1=-2.64$, whereas the {\WMAP} value is ${\cal L}_1=-1.22$.
903: As a check, we numerically marginalize the ${\cal L}_0$ portion of the
904: likelihood over beam and point
905: source errors, to see if we can reproduce their value.
906: The desired integral is
907: \begin{eqnarray}
908: \exp(-{\cal L}_1/2)
909: &=& \frac{1}{L_0(d|C_l)}\int{dxd\vec{y}\,
910: e^{-(x^2 + \vec{y}^T\vec{y})/2}
911: L_0(d|C_l(x,\vec{y})) }
912: \end{eqnarray}
913: where
914: %$x$ parameterizes the point source uncertainty,
915: %$\vec{y}$ the beam uncertainties,
916: \begin{eqnarray}
917: C_l(x,\vec{y}) \equiv \left(C^{TT}_l + x \sigma^{\rm ptsrc}_l\right)
918: \left(1 + \sum_{i}y_i \sigma^{\rm beam}_l(i)\right),
919: \end{eqnarray}
920: is the theoretical model ($C^{TT}_l$) perturbed by point source and
921: beam errors.
922: With 10 dimensions to integrate over (nine beam modes and one point source
923: mode), normal grid-based quadrature is impractical,
924: so we turn to Monte Carlo integration instead:
925: \begin{eqnarray}
926: \exp(-{\cal L}^{MC}_l/2) &\approx& \frac{1}{N_{MC}}\sum_{i=1}^{N_{MC}}{e^{\ln L(d|C_l(x^{(i)},\vec{y}^{(i)}))-\ln L(d|C_l)}}
927: \end{eqnarray}
928: where $x^{(i)}$ and $y^{(i)}_j$ are independent unit-variance normal deviates.
929: With $N_{MC}=10^4$ points, we find ${\cal L}^{MC}_1=-1.29\pm0.04$,
930: consistent with the {\WMAP} result of $-1.22$, but not the
931: Huf08 result of $-2.64$.
932: %\aside{is there a factor of 2 issue here?
933: %our number is half the Huf08 number}
934:
935: As a further test of whether the our cosmological parameter
936: estimates fully capture the point source uncertainty,
937: we have run a Markov chain with a modified form of the point source likelihood
938: module, dubbed SRCMARG.
939: The point source correction is calculated via a simple numerical integration,
940: \begin{eqnarray}
941: \exp(-{\cal L}^{\rm ptsrc}_1/2) &=&
942: \int{d\alpha\,\frac{1}{\sqrt{2\pi}}e^{-\alpha^2/2} L_0(d|C_l+\alpha\sigma^{\rm ptsrc}_l)}
943: \label{eqn:srcmarg}\\ &\approx&
944: \frac{\Delta}{\sqrt{2\pi}}
945: \sum^{N}_{i=-N}w_i e^{-(i\Delta)^2/2} L_0(d|C_l+i\Delta\sigma^{\rm ptsrc}_l)
946: \label{eqn:srcquadrature}
947: \end{eqnarray}
948: with $N=25$, $\Delta=0.2$, and $w_i=1$ for except at the endpoints
949: where $w_{|N|}=1/2$ (the trapezoidal rule).
950: The resulting one-dimensional marginalized distribution for $\sigma_8$,
951: shown in the left panel of \reffig{fig:nssig8}, is
952: indistinguishable from our standard result.
953: We have also run a SRCMARG chain with the error
954: increased by a factor of $5$ (i.e., $\delta A_{ps}=0.005$).
955: In this case the uncertainty in $\sigma_8$ increases by 15\%.
956:
957: Likewise, we have run similar tests of the beam uncertainty, dubbed BEAMMARG.
958: The approach is the same as SRCMARG, but with
959: ``$C_l+\alpha\sigma^{\rm ptsrc}_l$'' in \refeqn{eqn:srcmarg}
960: replaced by ``$C_l(1+\alpha\sigma^{\rm beam}_l)$'',
961: where $\sigma^{\rm beam}_l$ is the noisiest beam eigenmode, shown in
962: Figure 12 of \cite{hill/etal:prep}.
963: The 1D marginalized distributions for $n_s$ are shown in the right panel of
964: \reffig{fig:nssig8}.
965: As with SRCMARG, the BEAMMARG result is indistinguishable from our
966: standard result.
967: Inflating the beam error by a factor of 20 results in a 14\% increase
968: in $\delta n_s$, along with a slight shift in $n_s$ away from unity.
969:
970: % -----------------------------------------------------------------------------
971:
972: %\bibliographystyle{wmap}
973: %\bibliography{apj-jour,wmap}
974:
975: \begin{thebibliography}{31}
976: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
977:
978: \bibitem[{{Bennett} et~al.(2003)}]{bennett/etal:2003}
979: {Bennett}, C.~L., et~al. 2003, \apj, 583, 1
980:
981: \bibitem[{{Bischoff} et~al.(2008)}]{bischoff/etal:2008}
982: {Bischoff}, C., et~al. 2008, ArXiv e-prints, 802
983:
984: \bibitem[{{de Zotti} et~al.(2005){de Zotti}, {Ricci}, {Mesa}, {Silva},
985: {Mazzotta}, {Toffolatti}, \& {Gonz{\'a}lez-Nuevo}}]{dezotti/etal:2005}
986: {de Zotti}, G., {Ricci}, R., {Mesa}, D., {Silva}, L., {Mazzotta}, P.,
987: {Toffolatti}, L., \& {Gonz{\'a}lez-Nuevo}, J. 2005, \aap, 431, 893
988:
989: \bibitem[{{Dunkley} et~al.(2008)}]{dunkley/etal:prep}
990: {Dunkley}, J., et~al. 2008, ArXiv e-prints, 803
991:
992: \bibitem[{{Finkbeiner}(2003)}]{finkbeiner:2003}
993: {Finkbeiner}, D.~P. 2003, \apjs, 146, 407, accepted (astro-ph/0301558)
994:
995: \bibitem[{Finkbeiner et~al.(1999)Finkbeiner, Davis, \&
996: Schlegel}]{finkbeiner/davis/schlegel:1999}
997: Finkbeiner, D.~P., Davis, M., \& Schlegel, D.~J. 1999, \apj, 524, 867
998:
999: \bibitem[{{Gold} et~al.(2008)}]{gold/etal:prep}
1000: {Gold}, B. et~al. 2008, \apjs
1001:
1002: \bibitem[{Gorski et~al.(2005)Gorski, Hivon, Banday, Wandelt, Hansen, Reinecke,
1003: \& Bartlemann}]{gorski/etal:2005}
1004: Gorski, K.~M., Hivon, E., Banday, A.~J., Wandelt, B.~D., Hansen, F.~K.,
1005: Reinecke, M., \& Bartlemann, M. 2005, \apj, 622, 759
1006:
1007: \bibitem[{{Hill} et~al.(2008)}]{hill/etal:prep}
1008: {Hill}, R. et~al. 2008, \apjs
1009:
1010: \bibitem[{{Hinshaw} et~al.(2003)}]{hinshaw/etal:2003}
1011: {Hinshaw}, G., et~al. 2003, \apjs, 148, 135
1012:
1013: \bibitem[{{Hinshaw} et~al.(2007)}]{hinshaw/etal:2007}
1014: ---. 2007, \apjs, 170, 288
1015:
1016: \bibitem[{{Hinshaw} et~al.(2008)}]{hinshaw/etal:prep}
1017: ---. 2008, ArXiv e-prints, 803
1018:
1019: \bibitem[{{Huffenberger} et~al.(2006){Huffenberger}, {Eriksen}, \&
1020: {Hansen}}]{huffenberger/eriksen/hansen:2006}
1021: {Huffenberger}, K.~M., {Eriksen}, H.~K., \& {Hansen}, F.~K. 2006, \apjl, 651,
1022: L81
1023:
1024: \bibitem[{{Huffenberger} et~al.(2007){Huffenberger}, {Eriksen}, {Hansen},
1025: {Banday}, \& {Gorski}}]{huffenberger/etal:2007prep}
1026: {Huffenberger}, K.~M., {Eriksen}, H.~K., {Hansen}, F.~K., {Banday}, A.~J., \&
1027: {Gorski}, K.~M. 2007, ArXiv e-prints, 710
1028:
1029: \bibitem[{{Jarosik} et~al.(2007)}]{jarosik/etal:2007}
1030: {Jarosik}, N., et~al. 2007, \apjs, 170, 263
1031:
1032: \bibitem[{{Jones} et~al.(2006)}]{jones/etal:2006}
1033: {Jones}, W.~C., et~al. 2006, \apj, 647, 823
1034:
1035: \bibitem[{{Kamionkowski} et~al.(1997){Kamionkowski}, {Kosowsky}, \&
1036: {Stebbins}}]{kamionkowski/kosowsky/stebbins:1997}
1037: {Kamionkowski}, M., {Kosowsky}, A., \& {Stebbins}, A. 1997, \prd, 55, 7368
1038:
1039: \bibitem[{{Kogut} et~al.(2003)}]{kogut/etal:2003}
1040: {Kogut}, A., et~al. 2003, \apjs, 148, 161
1041:
1042: \bibitem[{{Komatsu} et~al.(2008)}]{komatsu/etal:prep}
1043: {Komatsu}, E., et~al. 2008, ArXiv e-prints, 803
1044:
1045: \bibitem[{{Leitch} et~al.(2005){Leitch}, {Kovac}, {Halverson}, {Carlstrom},
1046: {Pryke}, \& {Smith}}]{leitch/etal:2005}
1047: {Leitch}, E.~M., {Kovac}, J.~M., {Halverson}, N.~W., {Carlstrom}, J.~E.,
1048: {Pryke}, C., \& {Smith}, M.~W.~E. 2005, \apj, 624, 10
1049:
1050: \bibitem[{{Montroy} et~al.(2006)}]{montroy/etal:2006}
1051: {Montroy}, T.~E., et~al. 2006, \apj, 647, 813
1052:
1053: \bibitem[{{P.~Ade} et~al.(2007)}]{ade/etal:2007}
1054: {P.~Ade}, et~al. 2007, ArXiv e-prints, 705
1055:
1056: \bibitem[{{Page} et~al.(2007)}]{page/etal:2007}
1057: {Page}, L., et~al. 2007, \apjs, 170, 335
1058:
1059: \bibitem[{{Readhead} et~al.(2004)}]{readhead/etal:2004}
1060: {Readhead}, A.~C.~S., et~al. 2004, \apj, 609, 498
1061:
1062: \bibitem[{{Reichardt} et~al.(2008)}]{reichardt/etal:prep}
1063: {Reichardt}, C.~L., et~al. 2008, ArXiv e-prints, 801
1064:
1065: \bibitem[{Seljak \& Zaldarriaga(1997)}]{seljak/zaldarriaga:1997}
1066: Seljak, U. \& Zaldarriaga, M. 1997, Phys. Rev. Lett., 78, 2054
1067:
1068: \bibitem[{{Sievers} et~al.(2007)}]{sievers/etal:2007}
1069: {Sievers}, J.~L., et~al. 2007, \apj, 660, 976
1070:
1071: \bibitem[{Spergel \&
1072: Zaldarriaga(1997{\natexlab{a}})}]{zaldarriaga/spergel:1997}
1073: Spergel, D.~N. \& Zaldarriaga, M. 1997{\natexlab{a}}, Phys. Rev. Lett., 79,
1074: 2180
1075:
1076: \bibitem[{Spergel \&
1077: Zaldarriaga(1997{\natexlab{b}})}]{spergel/zaldarriaga:1997}
1078: ---. 1997{\natexlab{b}}, \prl, 79, 2180
1079:
1080: \bibitem[{{Toffolatti} et~al.(1998){Toffolatti}, {Argueso Gomez}, {de Zotti},
1081: {Mazzei}, {Franceschini}, {Danese}, \& {Burigana}}]{toffolatti/etal:1998}
1082: {Toffolatti}, L., {Argueso Gomez}, F., {de Zotti}, G., {Mazzei}, P.,
1083: {Franceschini}, A., {Danese}, L., \& {Burigana}, C. 1998, \mnras, 297, 117
1084:
1085: \bibitem[{{Wright} et~al.(2008)}]{wright/etal:prep}
1086: {Wright}, E.~L. et~al. 2008, \apjs
1087:
1088: \end{thebibliography}
1089:
1090: % -----------------------------------------------------------------------------
1091: % tables
1092:
1093: \clearpage
1094:
1095: \begin{deluxetable}{cccc}
1096: \tablewidth{0pt}
1097: \tablecaption{Unresolved Point Source Contamination\label{tbl:Aps}}
1098: \tablehead{\colhead{Bands}&\colhead{Mask}
1099: &\colhead{$A_{ps}(\alpha=0)$ [$10^{-3}\muK^2{\rm sr}$]}
1100: &\colhead{$A_{ps}(\alpha=-0.09)$ [$10^{-3}\muK^2{\rm sr}$]}
1101: }
1102: \startdata
1103: QVW & KQ85 & $11.3 \pm 0.9$ & $11.2\pm0.9$ \\
1104: & KQ80 & $11.3 \pm 0.9$ & $11.2\pm0.9$ \\
1105: & KQ75 & $10.7 \pm 1.0$ & $10.6\pm1.0$ \\
1106: VW & KQ85 & $6.9\pm 3.4$ & $7.2\pm3.5$ \\
1107: & KQ80 & $9.1\pm 3.6$ & $9.5\pm3.8$ \\
1108: & KQ75 & $10.5\pm3.9$ & $11.1\pm4.1$ \\
1109: %\tableline
1110: %QVW & Kp0 & $11.3\pm0.9$ & $11.2\pm0.9$ \\
1111: % & xKQ85\tablenotemark{a} & $15.9 \pm 0.9$ & $15.7\pm 0.9$ \\
1112: %VW & Kp0 & $9.7 \pm 3.6$ & $10.1 \pm 3.8$ \\
1113: % & xKQ85\tablenotemark{a} & $11.0\pm 3.3$ & $11.5 \pm 3.5$ \\
1114: \enddata
1115: %\tablenotetext{a}{This mask only includes sources detected by {\WMAP},
1116: %and not the extra undetected sources in the standard KQ85 mask.}
1117: \tablecomments{All results are for $\ell=300-800$.}
1118: \end{deluxetable}
1119:
1120: %\begin{deluxetable}{ccccc}
1121: %\tablewidth{0pt}
1122: %\tablecaption{Unresolved Point Source Contamination\label{tbl:ptsrc}}
1123: %\tablehead{
1124: %&\colhead{$S_c$ [Jy]}
1125: %&\colhead{$N(>S_c)$ [${\rm sr}^{-1}$]}
1126: %&\colhead{$C^{ps}$ [$10^{-3}\muK^2{\rm sr}$]}
1127: %&\colhead{$b^{ps}$ [$10^{-5}\muK^3{\rm sr}^2$]}}
1128: %\startdata
1129: %%WMAP5 (QVW xKQ85)& $\approx0.75$ & $40\pm2$\tablenotemark{a} & $17.3\pm0.9$\tablenotemark{b} & \nodata \\
1130: %WMAP5 (KQ85)& \nodata & \nodata & $11.7\pm1.1$\tablenotemark{b} & $4.3\pm1.3$\tablenotemark{c} \\
1131: %\cite{toffolatti/etal:1998}$\times0.64$ & 1.0 & 20.6 & 18.0 & 16.5 \\
1132: % & 0.75 & 31.8 & 14.4 & 10.1 \\
1133: % & 0.50 & 57.5 & 10.4 & 4.9 \\
1134: %\cite{dezotti/etal:2005}& 1.0 & 22.4 & 16.4 & 16.7 \\
1135: %& 0.75 & 34.8 & 12.5 & 9.6 \\
1136: %& 0.50 & 61.5 & 8.3 & 4.3 \\
1137: %\enddata
1138: %\tablenotetext{a}{{\WMAP} detects 390 sources over 78\% of the sky
1139: %\citep{wright/etal:prep}.}
1140: %%with $\sin|b|>0.25$ ($|b|>14.5\degr$). The $S_c$ implied by this column
1141: %%is not necessarily equal to the implied $S_c$ of the $C^{ps}$ \& $b^{ps}$
1142: %%columns, because more sources are masked than just those in the WMAP5 catalog.
1143: %\tablenotetext{b}{By equation~\refeqn{eqn:srcmodel},
1144: %$C^{ps}(Q) = A_{ps}r(Q)^2 = 1.089\times A_{ps}$, where $A_{ps}$ is
1145: %taken from \reftbl{tbl:Aps} for QVW and KQ75 mask.}
1146: %\tablenotetext{c}{From \cite{komatsu/etal:prep}, using the Q-band map and
1147: %KQ75 mask.}
1148: %\tablecomments{All numbers are evaluated at Q-band (40.7 GHz).}
1149: %\end{deluxetable}
1150:
1151: \begin{deluxetable}{cccc}
1152: \tablewidth{0pt}
1153: \tablecaption{Unresolved Point Source Contamination\label{tbl:ptsrc}}
1154: \tablehead{
1155: &\colhead{$S_c$ [Jy]}
1156: &\colhead{$C^{ps}$ [$10^{-3}\muK^2{\rm sr}$]}
1157: &\colhead{$b^{ps}$ [$10^{-5}\muK^3{\rm sr}^2$]}}
1158: \startdata
1159: WMAP5 (KQ75)& \nodata & $11.7\pm1.1$\tablenotemark{a} & $4.3\pm1.3$\tablenotemark{b} \\
1160: \cite{toffolatti/etal:1998}$\times0.64$ & 0.6 & 12.1 & 6.8 \\
1161: & 0.5 & 10.4 & 4.9 \\
1162: %& 0.7 & 13.6 & 9.0 \\
1163: %& 0.58 & 11.7 & 6.4 \\
1164: %& 0.47 & 9.9 & 4.4 \\
1165: \cite{dezotti/etal:2005} & 0.7 & 11.7 & 8.4 \\
1166: & 0.5 & 8.3 & 4.3 \\
1167: \enddata
1168: %\tablenotetext{a}{{\WMAP} detects 390 sources over 78\% of the sky
1169: %\citep{wright/etal:prep}.}
1170: %with $\sin|b|>0.25$ ($|b|>14.5\degr$). The $S_c$ implied by this column
1171: %is not necessarily equal to the implied $S_c$ of the $C^{ps}$ \& $b^{ps}$
1172: %columns, because more sources are masked than just those in the WMAP5 catalog.
1173: \tablenotetext{a}{By equation~\refeqn{eqn:srcmodel},
1174: $C^{ps}(Q) = A_{ps}r(Q)^2 = 1.089\times A_{ps}$, where $A_{ps}$ is
1175: the QVW/KQ75 result from \reftbl{tbl:Aps}.}
1176: \tablenotetext{b}{From \cite{komatsu/etal:prep}, using the Q-band map and
1177: KQ75 mask.}
1178: \tablecomments{All numbers are evaluated at 40.7 GHz (Q band).}
1179: \end{deluxetable}
1180:
1181: \begin{deluxetable}{ccc}
1182: \tablewidth{0pt}
1183: \tablecaption{Beam/source likelihood treatment effect on parameters\label{tbl:srcnssig8}}
1184: \tablehead{\colhead{Likelihood treatment}
1185: &\colhead{$n_s$} & \colhead{$\sigma_8$}}
1186: \startdata
1187: %standard & $0.9635\pm 0.0144$ & $0.7964 \pm 0.0358$ \\
1188: %%standard (mn) & $0.9623\pm 0.0141$ & $0.7946 \pm 0.0342$ \\
1189: %SRCMARG & $0.9643\pm 0.0143$ & $0.7981\pm 0.0363$ \\
1190: %SRCMARG $\times5$ & $0.9654\pm 0.0149$ & $0.8034\pm 0.0418$ \\
1191: %%srcmarg (x) & $0.9654 \pm 0.0136$ & $0.7999\pm 0.0345$ \\
1192: %%srcmarg x5 (x)& $0.9658 \pm 0.0147$ & $0.7997\pm 0.0416$ \\
1193: %BEAMMARG & $0.9641\pm 0.0145$ & $0.7988\pm 0.0355$ \\
1194: %BEAMMARG $\times20$ & $0.9579\pm 0.0164$ & $ 0.7963\pm 0.0344$ \\
1195: %%beammarg x5 & $ 0.9606\pm 0.0149$ & $0.7986\pm 0.0338$ \\
1196: %%beammarg (x)& $0.9650\pm 0.0151$ & $0.7990\pm 0.0346$ \\
1197: %%beammarg x5 (x)& $0.9604\pm 0.0158$ & $0.7971\pm 0.0349$ \\
1198: standard & $0.964\pm 0.014$ & $0.796 \pm 0.036$ \\
1199: SRCMARG & $0.964\pm 0.014$ & $0.798\pm 0.036$ \\
1200: SRCMARG $\times5$ & $0.965\pm 0.015$ & $0.803\pm 0.042$ \\
1201: BEAMMARG & $0.964\pm 0.015$ & $0.799\pm 0.036$ \\
1202: BEAMMARG $\times20$ & $0.958\pm 0.016$ & $ 0.796\pm 0.034$ \\
1203: \enddata
1204: \tablecomments{
1205: One-dimensional marginalized values for $n_s$ and $\sigma_8$ for various treatments
1206: of the unresolved point source and beam uncertainty in the
1207: {\WMAP} likelihood code.
1208: See \refsec{sec:appsrclike} for descriptions of SRCMARG and BEAMMARG.
1209: Here ``$\times5$'' and ``$\times20$'' indicate the error has been increased
1210: by a factor of 5 and 20, respectively.
1211: }
1212: \end{deluxetable}
1213:
1214: % -----------------------------------------------------------------------------
1215: % figures
1216:
1217: \clearpage
1218:
1219: \begin{figure}
1220: %\plotone{figures/fig_tt_highl_wmap5.eps}
1221: \plotone{f1.eps}
1222: \caption{\label{fig:tthighl}
1223: The {\WMAP} 5-year temperature (TT) power spectrum.
1224: The red curve is the best-fit theory spectrum from the $\Lambda$CDM/{\WMAP}
1225: chain \citep[Table 2]{dunkley/etal:prep} based on {\WMAP} alone,
1226: with parameters
1227: $(
1228: \ensuremath{\Omega_bh^2},
1229: \ensuremath{\Omega_mh^2},
1230: \ensuremath{\Delta_{\cal R}^2},
1231: \ensuremath{n_s},
1232: \ensuremath{\tau},
1233: \ensuremath{H_0}
1234: ) = (
1235: \ensuremath{0.0227},
1236: \ensuremath{0.131},
1237: \ensuremath{2.41},
1238: \ensuremath{0.961},
1239: \ensuremath{0.089},
1240: \ensuremath{72.4}
1241: )$.
1242: The uncertainties include both cosmic variance, which dominates below
1243: $\ell=540$, and instrumental noise which dominates at higher multipoles.
1244: The uncertainties increase at large $\ell$ due to {\WMAP}'s
1245: finite resolution. The improved resolution of the third peak near $\ell=800$
1246: in combination with the simultaneous measurement of the rest of the spectrum
1247: leads to the improved results reported in this release.
1248: %{\WMAP} 5-year TT power spectrum.
1249: %The pink curve is the best-fit theory spectrum from the $\Lambda$CDM/{\WMAP}
1250: %chain \citep{dunkley/etal:prep},
1251: }
1252: \end{figure}
1253:
1254: %\begin{figure}
1255: %\plotone{figures/fig_tt_highl_wmap5_vs_wmap3.ps}
1256: %\caption{\label{fig:tthighlvs3yr}
1257: %{\WMAP} 5-year TT power spectrum compared with the previously released
1258: %{\WMAP} 3-year power spectrum \citep{hinshaw/etal:2007}.
1259: %\aside{bottom panel needs work; the dotted line is the 5-year spectrum
1260: %multiplied by 0.97; need plot of p2/p3 beam}
1261: %}
1262: %\end{figure}
1263:
1264: %\begin{figure}
1265: %\plotone{figures/fig_chisq_tt_highl_wmap5.eps}
1266: %\caption{\label{fig:ttchisq}
1267: %Reduced $\chi^2$ ($\chi^2/{\rm dof}$) of the 5-year TT spectrum
1268: %in bins of $\Delta\ell=100$ (black). %, versus the 3-year spectrum (red).
1269: %For $\ell=33-1000$, $\chi^2=1068$ with probability to exceed of 1.3\%.
1270: %For the 3-year spectrum, $\chi^2=1041$ over the same $\ell$-range,
1271: %with probability to exceed of 5.1\%.
1272: %The dotted lines are the 1, 2, and $3\sigma$ limits.
1273: %The gray curve is the average over 25 simulations.
1274: %% which have $1.05\pm0.01$.
1275: %}
1276: %\end{figure}
1277:
1278: \begin{figure}
1279: %\plotone{figures/fig_tt_highl_wmap5_vs_other_exp.eps}
1280: \plotone{f2.eps}
1281: \caption{\label{fig:tthighlvsothers}
1282: %{\WMAP} 5-year TT power spectrum compared with recent results from the
1283: %Acbar \citep[purple]{reichardt/etal:prep},
1284: %Boomerang \citep[green]{jones/etal:2006},
1285: %and CBI \citep[red]{readhead/etal:2004} experiments.
1286: %The pink curve is the best-fit $\Lambda$CDM model to the {\WMAP} data,
1287: %which agrees well with the other datasets when extrapolated to higher-$\ell$.
1288: The {\WMAP} 5-year TT power spectrum along with recent results from the
1289: ACBAR \citep[purple]{reichardt/etal:prep},
1290: Boomerang \citep[green]{jones/etal:2006},
1291: and CBI \citep[red]{readhead/etal:2004} experiments. The other experiments
1292: calibrate with {\WMAP} or {\WMAP}'s measurement of Jupiter (CBI).
1293: The red curve is the best-fit $\Lambda$CDM model to the {\WMAP} data,
1294: which agrees well with all data sets when extrapolated to higher-$\ell$.
1295: }
1296: \end{figure}
1297:
1298: %\begin{figure}
1299: %\plottwo{figures/fig_ptsrc_amplitude_lbins.eps}{figures/fig_tt_highl_v_minus_w.eps}
1300: %\caption{\label{fig:ptsrc}
1301: %\textit{Left:}
1302: %Unresolved point source contamination $A_{ps}$,
1303: %measured in bins of $\Delta\ell=100$.
1304: %\textit{Right:}
1305: %The TT $\rm V-W$ null spectrum, after correcting for unresolved point
1306: %source emission.
1307: %}
1308: %\end{figure}
1309:
1310: \begin{figure}
1311: %\plotone{figures/fig_ptsrc_amplitude_lbins.eps}
1312: \plotone{f3.eps}
1313: \caption{\label{fig:ptsrc}
1314: The unresolved point source contamination $A_{ps}$,
1315: measured in bins of $\Delta\ell=100$ evaluated at 40.7 GHz (Q-band).
1316: %For a source with flux scaling at $I_\mu\propto\nu^2$, $A_{ps}$
1317: %is independent of frequency.
1318: For a source population whose fluxes are independent of frequency
1319: $A_{ps}$ scales roughly as $\sim\nu^{-2}$ in the {\WMAP} data.
1320: The red data points are from the analysis of V and W bands
1321: alone and the blue points are from the analysis of Q, V, and W bands.
1322: The horizontal dashed green lines, at 0.010 and 0.012,
1323: show the $1\sigma$ bounds for our adopted value of $A_{ps}$.
1324: Note that the QVW amplitude is independent of $\ell$.
1325: }
1326: \end{figure}
1327:
1328: \begin{figure}
1329: %\plotone{figures/fig_tt_highl_v_minus_w.eps}
1330: \plotone{f4.eps}
1331: \caption{\label{fig:ttnull}
1332: The TT $\rm V-W$ null spectrum. After correcting for unresolved point
1333: source emission, the individual power spectra are subtracted
1334: in power spectrum space. The result is consistent
1335: with zero and thus there is no evidence of point source contamination.
1336: In these units, point source contamination would be evident as a horizontal
1337: offset from zero.
1338: At $\ell=500$, the TT power spectrum is $C^{\rm TT}_\ell\approx0.06$;
1339: thus the contamination is limited to roughly 3\% in power.
1340: }
1341: \end{figure}
1342:
1343: %\begin{figure}
1344: %\plotone{figures/fig_te_highl_wmap5.eps}
1345: %\caption{\label{fig:tehighl}
1346: %{\WMAP} 5-year TE power spectrum.
1347: %The green curve is the best-fit theory spectrum from the $\Lambda$CDM/{\WMAP}
1348: %chain \citep{dunkley/etal:prep} (parameters of which are given the caption
1349: %of \reffig{fig:tthighl}).
1350: %The reduced $\chi^2$ of the model is 0.97 for $\ell=24-450$.
1351: %Note that the vertical axis is $(\ell+1)C_\ell/(2\pi)$, and not
1352: %$\ell(\ell+1)C_\ell/(2\pi)$.
1353: %}
1354: %\end{figure}
1355:
1356: \begin{figure}
1357: %\plotone{figures/fig_te_highl_wmap5.eps}
1358: \plotone{f5.eps}
1359: \caption{\label{fig:tehighl}
1360: The {\WMAP} 5-year TE power spectrum.
1361: The green curve is the best-fit theory spectrum from the $\Lambda$CDM/{\WMAP}
1362: Markov chain \citep{dunkley/etal:prep}.
1363: For the TE component of the fit, $\chi^2=415$, and there are
1364: 427 multipoles and 6 parameters; thus the number of degrees of freedom is
1365: $\nu=421$, leading to $\chi^2/\nu=0.99$.
1366: The particle horizon size at decoupling corresponds to $l\approx100$.
1367: The clear anticorrelation between the primordial plasma density
1368: (corresponding approximately to T) and velocity (corresponding approximately
1369: to E) in causally disconnected regions of the sky indicates that the
1370: primordial perturbations must have been on a superhorizon scale.
1371: Note that the vertical axis is $(\ell+1)C_\ell/(2\pi)$, and not
1372: $\ell(\ell+1)C_\ell/(2\pi)$.
1373: }
1374: \end{figure}
1375:
1376: \begin{figure}
1377: \plotone{f6.eps} %{figures/fig_tb_highl_wmap5.eps}
1378: \caption{\label{fig:tbhighl}
1379: The {\WMAP} 5-year TB power spectrum, showing no evidence of cosmological
1380: signal.
1381: The null reduced $\chi^2$ for $\ell=24-450$ is 0.97.
1382: Note that the vertical axis is $(\ell+1)C_\ell/(2\pi)$, and not
1383: $\ell(\ell+1)C_\ell/(2\pi)$.
1384: }
1385: \end{figure}
1386:
1387: %\begin{figure}
1388: %\plottwo{figures/fig_bb_lowl_wmap5.eps}{figures/fig_bb_lowl_wmap5_zoomout.eps}
1389: %\caption{\label{fig:bblowl}
1390: %{\WMAP} 5-year BB power spectrum at low-$\ell$.
1391: %Note that the $y$-axis is $(\ell+1)C^{EE}_\ell/(2\pi)$,
1392: %not $\ell(\ell+1)C^{EE}_\ell/(2\pi)$.
1393: %}
1394: %\end{figure}
1395:
1396: \begin{figure}
1397: \plotone{f7.eps} %{figures/simple_EE.eps}
1398: \caption{\label{fig:simpleee}
1399: Conditional likelihoods for the $\ell=2-7$ EE multipole moments (black curves),
1400: computed using the {\WMAP} likelihood code by varying the multipole in
1401: question, with all other multipoles fixed to their
1402: fiducial values.
1403: For example, in the $\ell=4$ panel, the black curve is
1404: $f(x) \propto L(d|\dots,C^{EE}_3,C^{EE}_4=x,C^{EE}_5,\dots)$.
1405: For comparison, na\"ive pseudo-$C_l$ estimates are also shown
1406: with Gaussian errors (red curves).
1407: The pseudo-$C_\ell$ errors are noise only,
1408: while the conditional distributions include cosmic variance.
1409: }
1410: \end{figure}
1411:
1412: \begin{figure}
1413: \plotone{f8.eps} %{figures/simple_BB.eps}
1414: \caption{\label{fig:simplebb}
1415: Conditional likelihoods for the $\ell=2-7$ BB multipole moments (black curves),
1416: computed using the {\WMAP} likelihood code by varying the multipole in
1417: question, with all other multipoles fixed to their fiducial values.
1418: For comparison, na\"ive pseudo-$C_l$ estimates are also shown
1419: with Gaussian errors (red curves).
1420: The pseudo-$C_\ell$ errors are noise only,
1421: while the conditional distributions include cosmic variance.
1422: Note the large difference between the likelihood code and
1423: the pseduo-$C_\ell$ value for $\ell=3$; this mode is sensitive to the time-orderd
1424: data baseline and is extremely poorly measured by {\WMAP}, illustrating the
1425: complicated noise structure of the polarization data on large scales.
1426: }
1427: \end{figure}
1428:
1429: \begin{figure}
1430: %\plotone{figures/fig_ee_lowl_wmap5.eps}
1431: %\plotone{figures/simple_EE_spectrum.KaQV.eps}
1432: \plotone{f9.eps}
1433: \caption{\label{fig:eelowl}
1434: {\WMAP} 5-year EE power spectrum at low-$\ell$.
1435: The error bars are the 68\% CL of the conditional likelihood of
1436: each multipole, with the other multipoles fixed at their fiducial theory values;
1437: the diamonds mark the peak of the conditional likelihood distribution.
1438: The error bars include noise and cosmic variance;
1439: the point at $\ell=7$ is the 95\% CL upper limit.
1440: The pink curve is the fiducial best-fit $\Lambda$CDM
1441: model \citep{dunkley/etal:prep}.
1442: }
1443: \end{figure}
1444:
1445: \begin{figure}
1446: %\plotone{figures/fig_ee_highl_wmap5_vs_other_exp.eps}
1447: \plotone{f10.eps}
1448: \caption{\label{fig:eehighl}
1449: {\WMAP} 5-year EE power spectrum, compared with results from the
1450: Boomerang \citep[green]{montroy/etal:2006},
1451: CAPMAP \citep[orange]{bischoff/etal:2008},
1452: CBI \citep[red]{sievers/etal:2007},
1453: DASI \citep[blue]{leitch/etal:2005},
1454: and QUAD \citep[purple]{ade/etal:2007} experiments.
1455: The pink curve is the best-fit theory spectrum from the $\Lambda$CDM/{\WMAP}
1456: Markov chain \citep{dunkley/etal:prep}.
1457: Note that the $y$-axis is $C^{EE}_\ell$,
1458: not $\ell(\ell+1)C^{EE}_\ell/(2\pi)$.
1459: }
1460: \end{figure}
1461:
1462: \begin{figure}
1463: %\plottwo{figures/sigma8_srcmarg.eps}{figures/ns_beammargerrx20.eps}
1464: \plotone{f11.eps}
1465: \caption{\label{fig:nssig8}
1466: \textit{Left:}
1467: One-dimensional marginalized likelihood distributions of $\sigma_8$ for various treatments
1468: of the source uncertainty in the likelihood code:
1469: the standard likelihood function [black],
1470: the alternative treatment of the source uncertainty described in
1471: equation~\refeqn{eqn:srcquadrature} [blue],
1472: the alternative treatment, but with the unresolved point source error
1473: increased by
1474: $\times5$ [cyan].
1475: The agreement between black and blue curves shows that the standard treatment
1476: is producing the correct answer.
1477: \textit{Right:}
1478: 1D marginalized likelihood distributions of $n_s$ for various treatments
1479: of the beam uncertainties:
1480: the standard likelihood function [black],
1481: the alternative treatment of the beam uncertainty described in
1482: equation~\refeqn{eqn:srcquadrature} [red],
1483: the alternative treatment, but with the beam error increased by
1484: a factor of 20 [orange].
1485: The agreement between the black and red curves shows that the standard
1486: treatment is producing the correct answer.
1487: }
1488: \end{figure}
1489:
1490: \begin{figure}
1491: \plotone{f12.eps}
1492: \caption{\label{fig:ttunbinned}
1493: The unbinned {\WMAP} 5-year temperature (TT) power spectrum (black),
1494: compared with the {\WMAP} 3-year result (red). The slight upward shift
1495: of the 5-year spectrum relative to the 3-year spectrum is due to
1496: the change in the beam transfer function.
1497: The pink curve is the best-fit $\Lambda$CDM model to the {\WMAP}5 data.
1498: }
1499: \end{figure}
1500:
1501: \begin{figure}
1502: \plotone{f13.eps}
1503: \caption{\label{fig:ttfreq}
1504: The unbinned {\WMAP} 5-year temperature (TT) power spectrum as a function
1505: of frequency, divided by the best-fit $\Lambda$CDM model to the {\WMAP} data.
1506: }
1507: \end{figure}
1508:
1509: % -----------------------------------------------------------------------------
1510: \end{document}
1511: