1:
2: \documentclass[final]{aipproc}
3:
4: \layoutstyle{8x11single}
5:
6:
7:
8:
9: \def\lsim{\,\lower2truept\hbox{${< \atop\hbox{\raise4truept\hbox{$\sim$}}}$}\,}
10: \def\gsim{\,\lower2truept\hbox{${> \atop\hbox{\raise4truept\hbox{$\sim$}}}$}\,}
11:
12:
13: \SetInternalRegister\hbadness{8000} % pseudo latin isn't breaking very well :-)
14:
15:
16: % We need something to show special stuff for ARLO only.
17: %
18: \newcommand\doingARLO[2][]{%
19: \ifx\mmref\undefined #1\else #2\fi
20: }
21:
22:
23:
24:
25:
26: \begin{document}
27:
28: \title
29: [Polarized Galactic radio emission at degree scales]
30: {A Multifrequency Analysis of the Polarized Diffuse Galactic Radio
31: Emission at Degree Scales}
32:
33: \classification{43.35.Ei, 78.60.Mq}
34: \keywords{Document processing, Class file writing, \LaTeXe{}}
35:
36: \author{C. Burigana}{
37: address={Istituto TeSRE/CNR, via Gobetti 101, I-40129 Bologna, Italy},
38: email={burigana@tesre.bo.cnr.it},
39: thanks={This work was commissioned by the AIP}
40: }
41:
42: \iftrue
43: \author{L. La Porta}{
44: address={Istituto TeSRE/CNR, via Gobetti 101, I-40129 Bologna, Italy},
45: email={laporta@tesre.bo.cnr.it},
46: }
47:
48: \fi
49:
50: % \copyrightholder{Acoustical Scociety of America}
51: \copyrightyear {2001}
52:
53:
54:
55: \begin{abstract}
56: The polarized diffuse Galactic radio emission, mainly synchrotron emission,
57: is expected to be one of the most relevant source of astrophysical contamination
58: at low and moderate multipoles in cosmic microwave background
59: polarization anisotropy experiments
60: at frequencies $\nu \le 50 \div 100$~GHz.
61: We present here preliminary results based on a recent analysis of the
62: Leiden surveys covering about 50\% of the sky at low
63: as well as at middle and high Galactic latitudes.
64: By implementing specific interpolation methods to
65: deal with these data, which show a large variation of the sampling across
66: the sky, we produce maps of the polarized diffuse Galactic synchrotron
67: component at frequencies between 408 and 1411~MHz with pixel sizes
68: larger or equal to $\simeq 0.92^\circ$. We derive the angular power spectrum
69: %(in terms of antenna temperature), $C_\ell^{ant}$,
70: of this component for the whole covered region and for three patches
71: in the sky significantly oversampled with respect to the average and
72: at different Galactic latitudes. We find multipole spectral
73: indices typically ranging between $\sim -3$ and $\sim -1 \div -1.5$,
74: according to the considered frequency and sky region.
75: At $\nu \ge 610$~MHz, the frequency spectral indices observed in the considered
76: sky regions are about $-3.5$, compatible with an intrinsic frequency
77: spectral index of about $-5.8$ and a depolarization due to Faraday
78: rotation with a rotation measure RM of about $15$~rad/m$^2$.
79: This implies that the observed angular power spectrum of the polarized
80: signal is about 85\% or 20\% of the intrinsic one
81: at 1411~MHz or 820~MHz respectively.
82: \end{abstract}
83:
84: \date{\today}
85:
86: \maketitle
87:
88: \section{Introduction}
89:
90:
91: The polarized diffuse Galactic synchrotron emission,
92: whose study provides important insights into the properties of
93: the Galactic magnetic field and the interstellar ionized matter,
94: is expected to be one of the most relevant source of astrophysical contamination
95: at low and moderate multipoles in
96: cosmic microwave background (CMB) polarization anisotropy experiments
97: at frequencies $\nu \le 50 \div 100$~GHz. At frequencies about 1~GHz, the Galactic
98: synchrotron emission dominates over the Bremsstrahlung emission
99: which, although expected to be weakly polarized, significantly increases
100: with the frequency, $\nu$, in comparison with the synchrotron one.
101:
102: While ongoing and future experiments with
103: high sensitivity and resolution are expected to cover large sky areas
104: at different Galactic latitudes and frequencies (see, e.g., these Proceedings),
105: the Leiden surveys [1], covering about 50\% of the sky, can be used
106: to derive the angular power spectrum of the polarized diffuse Galactic radio
107: emission at frequencies between 408 and 1411~MHz.
108: We have implemented specific interpolation methods
109: to project surveys with large variations of the
110: sampling across the sky into maps;
111: we work here with pixel sizes larger or equal to $\simeq 0.92^\circ$,
112: appropriate to the case of the Leiden surveys.
113: Some sky areas, both at low and middle/high Galactic latitudes,
114: show a much better sky sampling than the average
115: and are particularly suitable for an analysis in terms of angular
116: power spectrum. Their extents, few tens of degrees both in Galactic latitude
117: $b$ and longitude $l$, together with the beamwidths
118: (HPBW from $2.3^\circ$ to $0.6^\circ$ for $\nu$ from 408 to 1411~MHz)
119: and the measure sensitivities (of about 100~mK at 610~GHz and
120: about 1.5 (3) times better (worst) at the highest (lowest) frequency)
121: imply that, in principle, we can study
122: the angular power spectrum, $C_\ell^{ant}$
123: (here in terms of antenna temperature),
124: in a multipole range from $\ell \sim$~few~tens, where boundary effects
125: begin to be neglibible, to about $\ell \sim 100$, where, even for the
126: better sampled regions, the noise power is close to the signal power.
127: In addition, the survey full coverage analysis allows to recover
128: the $C_\ell^{ant}$s at $\ell$ between $\sim 5$ and few tens.
129:
130: As a by-product of this work, these maps of polarized diffuse Galactic
131: emission, once properly
132: rescaled in frequency,
133: can be used as inputs for simulation activities in current and
134: future microwave polarization anisotropy experiments.
135:
136:
137: \section{Map production and consistency tests}
138:
139: The main problem in the analysis of the Leiden
140: surveys is due to their poor sampling across the sky; it is necessary
141: to project them into maps with pixel size of about $3.7^\circ$
142: (i.e. $n_{side}=16$; we use here the HEALPix scheme [2]
143: in which the number of pixel in the sky is $12 n_{side}^2$),
144: to find about one observation into each pixel of the whole
145: observed sky region.
146: For the same reason, a smoothing of the whole data with a
147: window function with comparable size can not be properly applied.
148: On the other hand, for some sky regions the sampling of the surveys
149: is significantly much better, by a factor $\simeq 4$. We identified
150: three regions (see [3]):
151: patch 1 [($110^\circ \le l \le 160^\circ$, $0^\circ \le b \le 20^\circ$)];
152: patch 2 [($5^\circ \le l \le 80^\circ$, $b \ge 50^\circ$) together with
153: ($0^\circ \le l \le 5^\circ$, $b \ge 60^\circ$) and
154: ($335^\circ \le l \le 360^\circ$, $b \ge 60^\circ$)];
155: patch 3 [($10^\circ \le l \le 80^\circ$, $b \ge 70^\circ$)].
156:
157: Therefore, we chose to derive the angular power spectrum
158: at $\ell$ between $\simeq 5$ and $\simeq 50$,
159: as representative of the ``whole'' sky,
160: by working with the survey full coverage,
161: and at $\ell$ between $\simeq 30$ and $\simeq 100$ by working
162: with the above patches, which allow to analyze both low and
163: middle/high Galactic latitudes.
164: We anticipate that in these patches the polarized signal is typically higher
165: than the average and it also shows relevant intensity
166: variations on scales of some degrees. We then expect to find in the patches
167: an angular power spectrum relatively higher than the spectrum obtained from the
168: survey full coverage analysis in the common multipole range.
169:
170: By simply averaging the survey observations in a given pixel,
171: we can produce maps of polarized signal $P$ (and also of Stokes parameters
172: $Q$ and $U$) that can be analyzed in terms of $C_\ell^{ant}$s
173: of the polarized signal (and $E$ and $B$ modes).
174: On the other hand, we verified that the results produced in this way,
175: although in rather good agreement with those derived from the improved method
176: described below both in terms of maps and of $C_\ell^{ant}$s,
177: are affected by discontinuity effects on
178: scales of the order of the pixel size.
179: This tends to add power in the recovered $C_\ell^{ant}$s,
180: particularly at multipoles close to that corresponding to
181: the pixel size [$\ell \sim 180/(\theta_{\rm pix}/{\rm deg})$],
182: because of its ``euristic'' similarity with point source confusion noise.
183:
184: To overcome this problem, we implemented a specific ``interpolation''
185: method and decided to produce maps at resolution of about $0.92^\circ$
186: (in order to smooth the discontinuities discussed above on scales
187: of about $2^\circ$), which, of course, provide reliable information
188: only on scales larger than $\simeq 2^\circ$, i.e. only up to $\ell \simeq 100$.
189: We assign to each sky pixel the average of the signals
190: falling close to the pixel centre, properly weighted according to
191: a certain power of the distance from the pixel. Different powers
192: have been tested (from $\sim 0.5$ to $\sim 2$): clearly, the higher
193: the power the higher the map contrast. On the other hand,
194: we verified that the results depend very weakly on the adopted power.
195: The algorithm searches, pixel by pixel, for a suitable
196: number of observations to use in the weighted average, according to the
197: following basic recipes:
198: (i) to have enough observations (typically more than 3, if compatible
199: with the following criteria); (ii) to use only observations quite close to the
200: considered pixel centre (typically, less than few degrees, if compatible with
201: the other criteria); (iii) to obtain the convergency of the result
202: (i.e. minimize its fractional variation) with the variation of the
203: number of observations and of the circle around the pixel centre
204: that contain them.
205: By applying our algorithm with different resolutions (we consider here
206: $n_{side}$ from 16 to 64), we can test its dependence on the details
207: of its practical
208: application\footnote{We tested also a mix of a simple average
209: of the signals in the pixel - whenever possible - and of
210: this ``interpolation'' scheme. We found preferable to apply everywhere
211: the ``interpolation'' scheme, in order to reduce better the
212: discontinuity effects.}.
213: Clearly, each interpolation method may alter the power
214: at small scales; in principle, it operates as a kind of filter or
215: regularization of the map and may decrease the power at small scales.
216: Then, a reasonable agreement with the angular power spectrum
217: derived, whenever possible, from other data with a better sampling
218: across the sky (at the same $\nu$ and in same sky region)
219: is crucial to probe the validity of the code recipes: in particular,
220: for the reason discussed above, we require that it does not
221: produce an underestimation of the power spectrum.
222:
223: We obtained maps for $P$, $Q$,
224: and $U$ at all frequencies and with
225: different resolutions ($16 \le n_{side} \le 64$) by using
226: the different methods discussed above
227: \footnote{We verified also, pixel by pixel in the maps, that
228: $P \simeq (Q^2+U^2)^{0.5}$ with an accuracy significantly better
229: than the rms noise in each pixel.}.
230:
231: From each map,
232: by using the HEALPix package (properly the ``anafast'' code)
233: we derive the angular power spectrum of the polarized signal
234: for the survey full coverage and for the three considered patches
235: (we renormalized the $C_\ell^{ant}$s to the case of whole sky coverage).
236: Preliminary results at 1411~MHz
237: as well as tests of the applicability of this method to derive
238: the $C_\ell^{ant}$s on relatively small patches have been reported in [3]
239: (in our case the limited patch extent is a minor
240: problem, because of the larger dimension of the considered sky areas).
241:
242: In the next section we present some of our results and show how
243: both the maps and the angular power spectra derived from them pass
244: all the consistency tests discussed above.
245:
246: \section{Results}
247:
248: In Fig.~1 we show one of the maps of polarized signal produced
249: with the code described in the previous
250: section\footnote{A color figure may be requested via e-mail to the authors.}.
251:
252: In Fig.~2 we compare the angular power spectra of the patch 1 at 610~MHz
253: obtained by working at different
254: resolutions: note the good agreement in the common multipole range. The
255: same holds at all frequencies for the different sky regions considered here
256: and also by extending the comparison to the lowest resolution maps
257: ($n_{side}=16$), a crucial test in the case of the
258: survey full coverage analysis.
259:
260: The angular power spectrum derived from each map is, of course, mainly
261: given by the sum of the astrophysical diffuse component relevant here,
262: the synchrotron, and of the instrumental noise; in particular, the latter
263: dominates at large multipoles, as it is evident also from the flattening
264: we find there.
265: We fit the recovered $C_\ell^{ant}$s as sum of two components,
266: represented by a set of parameters:
267: (i) the synchrotron component,
268: smoothed with the beam (assumed to be perfectly symmetric and Gaussian),
269: is approximated
270: as $k \ell^\alpha {\rm exp}[-(\sigma_b \ell)^2]$, $\sigma_b$ being
271: the 1$\sigma$ beamwidth (of course, the window function is not crucial
272: in this context at the highest frequencies);
273: (ii) the noise contribution is approximated
274: as a flat, white noise, component, $c_{wn}$. All the parameters of the fit
275: depend on $\nu$ and on the sky region. In each case, we find the best
276: fit parameters at $\ell$ between $\sim 30$ and $\sim 100 \div 200$;
277: we separately repeat the fit at $\ell$ between $\sim 5$ and $\sim 30$
278: in the case of the survey full coverage analysis.
279:
280:
281:
282: \begin{figure}
283: \caption{Map of polarized signal obtained at
284: $n_{side}=64$ ($\simeq 0.92^\circ$ of pixel size).}
285: \includegraphics[height=.5\textheight,angle=90]{S_nested64u3_610_mK.ps}
286: \end{figure}
287:
288: \begin{figure}
289: \caption{Comparison between the angular power spectra obtained by projecting
290: the Leiden surveys into maps at different resolution,
291: $n_{side}=32$ (i.e., pixel size $\simeq 1.8^\circ$,
292: dotted line) and 64 (i.e., pixel size $\simeq 0.92^\circ$, solid line).
293: %Note the good
294: %agreement between the two angular power spectra up to
295: %$\sim$ the maximum multipoles reachable by analyzing
296: %the lowest resolution map.
297: }
298: \includegraphics[height=.7\textheight,width=.32\textwidth,angle=90]{ps_S1_32e64_610.ps}
299: \end{figure}
300:
301: Of course, it is extremely important to check that the level of the noise
302: power derived from the fit is similar to that derived on the basis of
303: the rms noise per pixel in the map.
304: Given rms noise maps, we generate simulated maps of white noise
305: and extract their $C_\ell^{ant}$s.
306: This is shown in Fig.~3.
307: We quote the rms noise per pixel by using three different methods:
308: (i) the rms noise calculated by using the rms noise of each observation
309: and propagating the error according to the
310: mathematical rules applied to produce the maps (solid lines);
311: (ii) the same as in the case (i), except for pixels in which survey
312: observations fall, in which case we apply the standard weighted error on the
313: standard weigthed average (dot-dashed line in the left panel);
314: (iii) the standard weighted error on the
315: standard weigthed average for pixels in which survey
316: observations fall, and the average of these errors for the other pixels
317: (three dots-dashes). Of course, for the patches, and particularly for the patch 3,
318: the evaluations (ii) and (iii) provide practically the same results, given the
319: good sampling across the sky; thus, we only report the evaluation (iii).
320: Note that, for the survey full coverage analysis,
321: the noise spectrum evaluated according
322: to (iii) is of the order of that derived from the signal map
323: at $\ell \sim 50$:
324: therefore, at $\ell$ larger than $\sim 50$
325: no reliable information can be derived.
326: For the patches, on the contrary, the noise spectrum is below
327: that derived from the signal map, independently of the different
328: noise evaluations, up to $\ell \sim 100$. This confirms that,
329: not only from the point of view of the sampling across the sky, but also
330: from the point of view of the sensitivity, the patches can be used to
331: extract the synchrotron $C_\ell^{ant}$s up to $\ell \sim 100$.
332: We note that this is not valid, in principle, for possible analyses of
333: patches in other sky regions less sampled
334: and of lower polarized signal intensity,
335: where we expect a signal
336: to noise ratio close to that derived here for the survey full coverage
337: analysis,
338: i.e. a reliable information can be obtained only up to $\ell \sim 50$.
339:
340: From the survey full coverage analysis,
341: we derive reliable information also at $5 \le \ell \le 30$
342: (of course, only nearly full sky observation can provide reliable information
343: at very low multipoles). In this case, both the noise and possible
344: effects introduced by the poor sampling or by the adopted ``interpolation''
345: technique are clearly negligible.
346: We find a flattening of the spectrum with respect to that at higher multipoles;
347: the slope (dotted line of left panel of Fig.~3)
348: at $\nu = 1411$~MHz, crucial for the extrapolation to higher
349: frequencies, is close to $-2.5$. This implies an essentially
350: flat (slope $\sim -0.25$) behaviour of
351: $\delta T_\ell = \sqrt{\ell(2\ell+1)C_\ell/4\pi}$ at $5 \le \ell \le 30$
352: ($\delta T_\ell$ is a quantity particularly
353: relevant in the comparison with the CMB temperature and polarization anisotropy
354: observations).
355:
356: \begin{figure}
357: \caption{Angular power spectra derived from the map (thin solid lines) and
358: synchrotron (thick solid lines) and synchrotron plus noise (thick
359: dot-dashed lines) best fit spectra at multipoles larger than $\sim 30$
360: compared with the those derived from simulated maps of pure white noise
361: with different estimates of the rms noise per pixel (roughly horizontal lines).
362: In the case of the survey full coverage analysis we report
363: also the $C_\ell^{ant}$s
364: obtained at lower multipoles (again, thin solid line)
365: together with the best fit spectrum (dotted line) dominated
366: by the synchrotron component (see also the text).}
367: \includegraphics[height=.25\textheight,width=1.\textwidth]{ps_SeS3_64u3_1411.ps}
368: \end{figure}
369:
370:
371: We focus now on the angular power spectrum
372: of the synchtrotron component at $\ell$ between $\simeq 30$ and
373: $\simeq 100 \div 200$.
374: Our results are summarized in Fig.~4.
375: We find slopes varying from about $-3$ to about $-1 \div -1.5$,
376: mainly depending on the frequency, although, as expected,
377: different sky regions show different slopes even at the same frequency.
378:
379: As already recognized (see [3]), we find
380: a particularly impressive agreement
381: between the result obtained for the patch 1 and
382: the angular power spectrum (long dashes in Fig.~4) obtained
383: at 1411~MHz in a smaller region (the region 2 in eq. (20)
384: of [3]) inside the patch 1 by exploiting
385: the data from [4] (which have
386: better sampling across the sky, sensitivity and resolution and
387: in which only the absolute calibration takes
388: advantage of the Leiden surveys).
389: This strongly supports the validity of our code to project
390: the Leiden surveys into HEALPix maps or, at least, indicates
391: that it does not introduce significant errors in
392: the evaluations of the $C_\ell^{ant}$s
393: even at $\ell \sim 100$ and in the worst case of
394: $\nu = 1411$~MHz (here the ratio between the beamwidth
395: and the typical angular distance among adjacent survey pointings
396: is minimal and the survey sky coverage is not the best
397: -- it occurs at 610~MHz).
398:
399: \begin{figure}
400: \caption{Synchrotron component of the angular power spectrum derived
401: for the entire map and the three patches at the different frequencies:
402: 408 (solid lines), 465 (dotted lines), 610 (dashed lines), 820
403: (dot-dashed lines) and 1411~MHz (three dots-dashes). In the case
404: of the patch 1, note the good agreement (within a factor $\sim 2$)
405: with the spectrum (long dashes) obtained from a smaller patch
406: inside the patch 1
407: by exploiting the survey [4] at 1411~MHz (see also the text).}
408: \includegraphics[height=.7\textheight,width=.5\textwidth,angle=90]{BSallpssync.ps}
409: \end{figure}
410:
411:
412: \section{Discussion and Conclusions}
413:
414: We presented here preliminary results based on a recent analysis of the
415: Leiden surveys for about 50\% of the sky and
416: three patches at low
417: as well as at middle/high Galactic latitudes
418: significantly better sampled than the average.
419: By implementing specific interpolation methods,
420: we produce maps of the polarized diffuse Galactic synchrotron
421: component at frequencies between 408 and 1411~MHz with pixel sizes
422: larger or equal to $\simeq 0.92^\circ$.
423:
424: %As a by-product, these maps can constitute, when properly
425: %rescaled in frequency, inputs for simulation activities in current and
426: %future microwave polarization anisotropy experiments.
427:
428: We derive the angular power spectrum
429: of the polarized diffuse Galactic synchrotron
430: emission for the whole covered region and for the three patches.
431: The multipole spectral indices typically range between $\sim -3$ and
432: $\sim -1 \div -1.5$,
433: depending on the frequency and on the sky region.
434:
435: Together with the large sky coverage, the multifrequency
436: Leiden surveys offer the opportunity
437: to study the dependence of the Galactic synchrotron polarized signal on $\nu$.
438: Clearly, depolarization effects and possible transitions
439: from essentially optically thin regimes to significantly self-absorbed
440: ones may ``obscure'' the intrinsic synchrotron spectral behaviour.
441: We find large variations of the spectral index observed in the different sky
442: pixels, probably due to real variations, but also to
443: the limited sensitivity and, perhaps, to differential depolarization
444: effects related to the $\nu$-dependent beamwidth or,
445: finally, to possible systematic and not well understood effects in the data.
446: We try to circumvent, at least in part,
447: these difficulties by exploiting the
448: ``statistical'' information contained in the $C_\ell^{ant}$s.
449:
450: In Fig.~5 we show the angular power spectrum as a function of $\nu$
451: at the multipole $\ell =50$ (a middle, representative
452: value considered in this study); we find very similar results
453: also at $\ell = 30$ and 100.
454: The observed frequency spectral indices
455: at $\ell =50$ are $\sim$ $-3.2$ (survey full coverage analysis),
456: $-3.3$ (patch 1), $-3.6$ (patch 2) and $-3.9$ (patch 3).
457: At the highest frequencies (820, 1411~MHz),
458: that are crucial for the extrapolation
459: at frequencies of $20\div100$~GHz relevant for CMB anisotropy
460: polarization radiometric experiments
461: and where beamwidth depolarization effects are expected to be
462: quite similar (HPBW varying only of $\sim 50$\% between 820 and 1411~MHz),
463: the observed slope can be easily explained by assuming an
464: intrinsic slope of $-5.8$ (as in the case of a slope $-0.9$ in terms of flux)
465: and a depolarization due to Faraday rotation with a rotation
466: measure RM of about $15$~rad/m$^2$.
467: This results is in quite good agreement with that previously obtained
468: from the analysis of a sky region partially overlapped with the patch 1,
469: as well as clearly compatible with the upper limits on RM
470: found in the Leiden surveys (see [5]).
471: This implies that the observed angular power spectrum of the polarized
472: signal is about 85\% or 20\% of the intrinsic one
473: (strictly, in absence of Faraday rotation depolarization)
474: at 1411~MHz or 820~MHz respectively.
475:
476:
477:
478: \begin{figure}
479: \caption{Angular power spectrum at multipole $\ell = 50$ as function
480: of $\nu$ for the different sky regions:
481: the survey full coverage (solid lines), the patch 1 (dotted lines),
482: the patch 2 (dashed lines) and the patch 3 (dot-dashed lines).
483: The straight longer lines represent a simple power law dependence of the
484: $C_\ell^{ant}$s (at $\ell = 50$) on $\nu$ corresponding to
485: an ``intrinsic'' slope of $-5.8$,
486: rescaled to the angular power spectra
487: derived here at 1411~MHz. The straight shorter lines
488: represent the dependence of the $C_\ell^{ant}$s
489: on $\nu$ between 820 and 1411~MHz
490: for a Faraday rotatation depolarization
491: with RM~=~15~rad~/~m$^2$ applied to the above ``intrinsic'' slope
492: (in the case of the survey full coverage analysis, note
493: the practically perfect agreement with the observed spectrum
494: between 820 and 1411~MHz).
495: Note also the further flattening
496: of the spectrum at the lowest frequencies in the case of the two
497: patches at relatively high Galactic latitudes (see also the text).}
498: \includegraphics[height=.7\textheight,width=.45\textwidth,angle=90]{BS_l50_pssync.ps}
499: \end{figure}
500:
501: \begin{theacknowledgments}
502: It is a pleasure to thank C.~Baccigalupi, G.~De~Zotti and R.~Fanti
503: for many constructive comments.
504: We acknowledge also D.~Maino, M.~Maris, R.~Paladini and F.~Perrotta
505: for the fruitful collaboration on this topic.
506: We are grateful to T.A.T. Spoelstra for his kind clearifications and to
507: L.~Chiappetti and P.~Platania for useful explanations
508: on the database of the Leiden surveys, provided us in a more
509: user-friendly form. The HEALPix package is acknowledged.
510: \end{theacknowledgments}
511:
512:
513: \section{References}
514:
515: 1. Brouw W.N., Spoelstra T.A.T., A\&AS, {\bf 26}, 129 (1976)
516:
517: \noindent
518: 2. G\`orski K.M., Hivon E., Wandelt B.D. 1998,
519: ``Analysis Issues for Large CMB Data Sets'',
520: in {\it MPA/ESO
521: Conference on Evolution of Large-Scale
522: Structure: from Recombination to
523: Garching}, edited by A.J. Banday, R.K. Sheth, L. Da Costa, 37, astro-ph/9812350
524:
525: \noindent
526: 3. Baccigalupi C., Burigana C., Perrotta F., et al., A\&A, {\bf 372}, 8 (2001)
527:
528: \noindent
529: 4. Uyaniker B., Furst E., Reich W., Reich P., Wielebinski R.,
530: A\&AS, {\bf 138}, 31 (1999)
531:
532: \noindent
533: 5. Spoelstra T.A.T., A\&A, {\bf 135}, 238 (1984)
534:
535: % choose bibtex style depending on layout style and options used in
536: % sample:
537:
538: \doingARLO[\bibliographystyle{aipproc}]
539: {\ifthenelse{\equal{\AIPcitestyleselect}{num}}
540: {\bibliographystyle{arlonum}}
541: {\bibliographystyle{arlobib}}
542: }
543: \bibliography{sample}
544:
545: \end{document}
546:
547: