1: \section{Theory}
2:
3: Rich clusters of galaxies contain hot electron gas which scatters
4: the Cosmic Microwave Background photons, leading to a temperature
5: decrement at frequencies less than 217 GHz and a temperature increment at
6: higher frequencies (relativistic corrections lead to a slight modification
7: of this spectral signature). This
8: is called the thermal Sunyaev-Zeldovich effect
9: (see \citealp{holderc99} for a review).
10: There is also
11: a kinetic SZ effect due to a cluster's peculiar velocity; this Doppler
12: shift of the apparent CMB temperature
13: is independent of frequency and is very difficult to
14: detect.
15: The SZ effect may also be produced by the lobes of radio galaxies
16: \citep[see][]{yamadass99}
17: or by patchy early reionization around quasars
18: \citep[see][]{aghanimetal96, gruzinovh98, knoxsd98} and
19: in the Lyman $\alpha$ forest \citep{loeb96}.
20:
21:
22: For long-wavelength observations, the Sunyaev-Zeldovich effect in
23: galaxy clusters will produce a temperature shift of
24: \begin{equation}
25: \frac {\delta T}{T} = - 2 y =
26: - 2 \tau_{0} \frac {k T_{x}} {m_{e}c^{2}} \sim
27: -5 \times 10^{-5} h^{-\frac{1}{2}}
28: \end{equation}
29: where $T_{x}$ is the plasma temperature and $\tau_{0}$ is the optical
30: depth of the cluster plasma \citep{peebles93}. This is comparable to the
31: temperature fluctuations in the CMB observed by COBE and can therefore
32: mask the true CMB temperature anisotropies on scales of several arcminutes
33: corresponding to clusters. The SZ effect is a localized spectral distortion
34: and appears as a temperature anisotropy. One way to distinguish
35: it from the true primordial fluctuations in the CMB is to use several
36: different frequencies, including some in the Wien tail where the SZ effect
37: will increase the apparent temperature.
38:
39: The frequency dependence of SZ thermodynamic temperature fluctuations
40: is given by
41: \begin{equation}
42: j(x) = 2\frac{\exp(x)}{(\exp(x)-1)^2}
43: \left(x \frac{\exp(x) + 1}{\exp(x)-1} - 4\right)
44: (\cosh(x) - 1),
45: \end{equation}
46: where we have defined
47: \begin{equation}
48: x \equiv \frac{h \nu}{k_b T_{cmb}}.
49: \end{equation}
50: This allows a temperature decrement at a known frequency $\nu_0$ to
51: be extrapolated to a different frequency using
52: \begin{equation}
53: dT(\nu) = dT(\nu_0) \frac{ j(x)}{ j(x_0)}.
54: \end{equation}
55:
56:
57:
58:
59: \section{Simulations}
60:
61:
62: We use
63: the XBACS cluster catalog
64: \citep{ebelingetal96}
65: to predict the location on the high-Galactic latitude
66: sky ($|b|>30$) of the
67: 226 brightest SZ sources. Clusters at higher redshift will have less
68: flux, despite comparable temperature decrements, because they are decreasing
69: in angular size (XBACS only covers up to $z=0.2$).
70: The MAP satellite can
71: hope to detect about 10 SZ clusters at the $5\sigma$ level, and to barely
72: resolve
73: Coma and Virgo
74: \citep{refregiersh98}.
75: Once the minimum of the angular diameter distance is
76: neared, the SZ flux from clusters of a given X-ray
77: temperature becomes effectively independent of redshift.
78: Hence, for beams smaller
79: than 10$'$, as will be used in the future with Planck HFI and microwave
80: interferometers, clusters at all redshifts will have an impact.
81: Although Planck will also see nearby clusters
82: as brighter than more distant ones, its high sensitivity and improved
83: resolution present the prospect of detecting numerous
84: high-redshift clusters through
85: their SZ effect.
86:
87:
88: Previous simulations have ignored the known positions and characteristics
89: of nearby galaxy clusters
90: but offer an illustration of the expected number counts
91: of high-redshift clusters, which is strongly dependent on cosmology
92: \citep{aghanimetal97}.
93: For simulations at Planck resolution, we
94: will need to add a simulation of the clusters not included in XBACS using
95: the $N(S)$ predictions of the Press-Schechter method and adopted temperature
96: and density profiles for clusters of a given mass.
97: A complete
98: realistic
99: simulation of the SZ effect from galaxy clusters
100: on the full-sky requires three components:
101: \begin{itemize}
102: \item
103: XBACS clusters ($z < 0.2$)
104: \item
105: Low-galactic latitude clusters at $z < 0.2$. These must be simulated
106: since they are not present in the XBACS sample.
107: \item
108: Clusters at $z > 0.2$, which also must be simulated because full-sky
109: X-ray surveys to determine their actual locations and fluxes do not exist.
110: \end{itemize}
111:
112:
113: \citet{refregiersh98} used an
114: isothermal beta-model to fit
115: the X-ray data and predict the flux at 90 GHz due to the thermal SZ effect in
116: XBACS clusters.
117: There are two major sources of systematic errors for the predicted
118: 90 GHz fluxes, the
119: overall normalization, which depends on the gas fraction and
120: on the virialization state, and
121: the error in the X-ray temperatures in the XBACS catalog, which
122: were derived from ROSAT spectra.
123: Data from MAP will be used to improve the overall normalization, which
124: can also be tested against N-body simulations.
125: Pending more detailed analysis, we believe that our overall normalization
126: of the predicted 90 GHz fluxes is uncertain by a factor of 1.5, and that
127: the prediction for each cluster is also uncertain by a factor of 1.5. The
128: spectral dependence
129: of the SZ effect is well-understood and should not
130: be a significant source of error.
131:
132: One possible advantage of having the actual positions of these clusters
133: from the XBACS catalog is that their non-Poissonian clustering is included.
134: However, when we calculate the angular power spectrum at 90 GHz for the
135: XBACS clusters with positions randomized within $|b|>30$ we find no
136: noticeable difference in the $C_\ell$ (see Figure \ref{fig:xbacs_cl}).
137: There is non-Poissonian clustering noticeable to the human eye in the
138: XBACS catalog, and this randomized version has many fewer cases
139: of multiple clusters located near each other on the sky. This indicates
140: that even mildly non-Poissonian clustering of point sources
141: distinguishable by eye is
142: sometimes not sufficient to alter the $C_\ell$ spectrum from its
143: Poissonian behavior (constant). This is different for high-redshift
144: infrared galaxies because there are many of them per pixel and the
145: fluctuations themselves are therefore magnified by non-Poissonian clustering
146: rather than just redistributing the fluctuations in a more clustered way.
147:
148: \begin{figure}[htb]
149: \centerline{\psfig{file=xbacs_b30_combined.ps,width=3in}}
150: \caption{Predicted SZ angular power spectra.}
151: \mycaption{$\ell (\ell + 1) C_\ell$ in $\mu$K$^2$ versus $\ell$
152: for both the XBACS catalog with clusters at nominal positions and
153: with clusters at random positions. The two curves overlap, illustrating that
154: the visually distinguishable non-Poissonian clustering of the XBACS clusters
155: is insufficient to affect the angular power spectrum of their SZ contribution.
156: }
157: \label{fig:xbacs_cl}
158: \end{figure}
159:
160:
161:
162: We also explore whether filaments of hot gas connecting and extending from
163: clusters can have an impact on CMB anisotropy
164: \citep{persietal95, boughn98}.
165: We use a toy model for the filaments where XBACS clusters that are
166: within 50$h^{-1}$Mpc of each other are connected by a filament and clusters
167: located too far apart to have neighbors in the catalog are
168: assigned two filaments heading in random directions.
169: Each filament is assigned a total number of baryons equal to its
170: cluster endpoint (or the average of its endpoints). Hence if the gas were
171: equally hot in filaments and clusters the total SZ flux from a filament
172: would equal the average of that of its endpoints.
173: This reproduces the rough structure seen in the simulations of
174: \citet{ceno99}\footnote{
175: See http://www.astro.princeton.edu/\~{}cen/PROJECTS/p1/p1.html for
176: nice visualizations of their simulations.}.
177: The results of our simulation of the SZ effect from
178: filaments are shown in
179: Figure \ref{fig:sz_filaments}.
180: While these filaments look
181: great visually\footnote{and
182: contain the type of patterns that might get significant
183: media attention, especially from the Weekly World News},
184: they are nearly insignificant in comparison
185: to the XBACS clusters themselves in terms of SZ effect
186: because their
187: electron gas temperature is expected to be lower than that
188: of rich clusters by about a factor of 10.
189: Thus even if half of the baryons
190: are in filaments and groups, as predicted,
191: their impact on anisotropy is small. Detecting
192: these filaments by cross-correlating CMB and large-scale structure maps
193: has been discussed by \citet{refregiersh98}, and that method can be
194: tested using the WOMBAT Challenge simulations, which contain
195: a variety of models for the filaments as well as systematic
196: variations of the cluster fluxes.
197:
198:
199:
200: \begin{figure}[htb]
201: %\figurenum{3}
202: \centerline{\psfig{file=fil_compared_log_cl.ps,width=3in}}
203: \caption{Angular power spectra from SZ effect in filaments and clusters.}
204: \mycaption{
205: Log-log plot of $\ell (\ell + 1) C_\ell$ versus $\ell$
206: for SZ simulation based on XBACS only (lower curve) compared to
207: simulation including filaments as well (upper curve).
208: The high-$\ell$ behavior is Poissonian, and the
209: filaments have not significantly increased the total power. The filaments
210: lead to significant non-Poissonian clustering on large angular scales, but
211: this effect will be entirely swamped by large-angular CMB fluctuations and
212: Galactic foregrounds.
213: }
214: \label{fig:fil_cl}
215: \end{figure}
216:
217:
218: Combining the SZ effect from clusters and filaments leads to only
219: a small increase in the angular power spectrum, as shown in Figure
220: \ref{fig:fil_cl}.
221: Figure \ref{fig:sz}
222: shows all of the sources with a wider
223: temperature scale than the plot of filaments alone.
224: The catalog is incomplete
225: at $|b|<30$ and reveals large-scale structure at $z < 0.2$. However,
226: almost none of the filaments are visible, illustrating that the
227: expected reduction of temperature in the filaments is sufficient
228: to make their SZ effect negligible.
229: The filaments are not detectable by forthcoming CMB instruments,
230: but the XBACS sources are strong. The prediction of
231: \citet{refregiersh98} that MAP can detect about a dozen XBACS clusters
232: is consistent with the results of our simulations.
233:
234:
235:
236: %\begin{figure}[htb]
237: %\figurenum{3}
238: %\centerline{\psfig{file=sz_nofil.ps,width=6in}}
239: %\caption{
240: %Predicted temperature fluctuations at 90 GHz from the Sunyaev-Zeldovich
241: %effect in XBACS clusters. [RESOLUTION, T SCALE] }
242: %\label{fig:sz_nofil}
243: %\end{figure}
244:
245:
246: \cleardoublepage
247:
248: \begin{figure}[th]
249: %\figurenum{3}
250: \centerline{\psfig{file=sz_filaments.ps,width=6in,angle=90}}
251: \caption{
252: Predicted temperature fluctuations at 90 GHz from filaments extending from
253: and connecting XBACS clusters using the toy model described in the text.
254: }
255: \label{fig:sz_filaments}
256: \end{figure}
257:
258: \nopagebreak
259:
260: \begin{figure}[bh]
261: %\figurenum{3}
262: \centerline{\psfig{file=sz_total.ps,width=6in,angle=90}}
263: \caption{
264: Combined prediction of the SZ effect from XBACS clusters and a simulation
265: of the filamentary structure connecting them.
266: }
267: \label{fig:sz}
268: \end{figure}
269: