1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass[apj]{emulateapj}
3: %\usepackage{epstopdf}
4: %\usepackage{graphicx}
5: %\usepackage{pstricks}
6: %\usepackage[T1]{fontenc}
7: %\usepackage{ae,aecompl}
8: %\bibliographystyle{apj}
9: %\RequirePackage{natbib}
10:
11: \begin{document}
12: \title{The $\beta$-Model Problem: The Incompatibility of X-ray and Sunyaev-Zeldovich Effect Model Fitting for Galaxy Clusters}
13: \author{Eric J. Hallman}
14: %\affil{Center for Astrophysics and Space Astronomy,
15: %University of Colorado, Boulder, CO 80309}
16: \author{Jack O. Burns}
17: \affil{Center for Astrophysics and Space Astronomy, Department of
18: Astrophysical and Planetary Sciences,
19: University of Colorado, Boulder, CO 80309}
20: \author{Patrick M. Motl}
21: \affil{Department of Physics and Astronomy,
22: Louisiana State University, Baton Rouge, LA 70803}
23: \author{Michael L. Norman}
24: \affil{Center for Astrophysics and Space Sciences, University of
25: California-San Diego, 9500 Gilman Drive, La Jolla, CA 92093}
26: \email{hallman@casa.colorado.edu}
27: \slugcomment{Accepted to the Astrophysical Journal}
28: \begin{abstract}
29: We have analyzed a large sample of numerically simulated
30: clusters to demonstrate the adverse effects resulting from use of X-ray fitted $\beta$-model
31: parameters with Sunyaev-Zeldovich effect (SZE) data. There is a
32: fundamental incompatibility between $\beta$ fits to X-ray surface
33: brightness profiles and those done with SZE profiles. Since
34: observational SZE radial profiles are in short supply, the X-ray
35: parameters are often used in SZE analysis. We show that this leads to
36: biased estimates of the integrated Compton y-parameter inside
37: $r_{500}$ calculated from
38: clusters. We suggest a simple correction of the method, using a
39: non-isothermal $\beta$-model modified by a universal temperature
40: profile, which brings these calculated quantities into closer
41: agreement with the true values.
42: \end{abstract}
43: \keywords{galaxies:clusters:general--cosmology:observations--hydrodynamics--methods:numerical--cosmology:cosmic microwave background}
44: \section{Introduction}
45: The hot gas in clusters of galaxies is responsible for inverse Compton
46: scattering cosmic microwave background (CMB) photons as they travel through the
47: intracluster medium (ICM). This results in a spectral distortion of
48: the CMB at the location of clusters on the sky, referred to as the \citet{sz} effect (SZE). This distortion is characterized by a low
49: frequency ($<$218 GHz) decrement, and higher frequency ($>$218 GHz) increment in the CMB
50: intensity \citep{carl}. The X-ray emission in clusters consists of thermal
51: bremsstrahlung and line emission from the same highly ionized plasma that scatters
52: the CMB.
53:
54: High resolution X-ray or SZE observations of clusters coupled with assumptions
55: about the gas distribution lead to estimates of the gas mass in the
56: cluster dark matter potential well. The electron number density is
57: often assumed to fit a $\beta$ model \citep{caval},
58: \begin{equation}
59: n_{e}(r) = n_{e0}\left(1 + \left(\frac{r}{r_{c}}\right)^{2}\right)^{-3\beta/2}.
60: \end{equation}
61: In the above relation, $n_{e0}$ is the central density normalization
62: and $r_c$ indicates the fitted parameter referred to as the core radius.
63: Fitting an observed X-ray or SZE profile to these projected $\beta
64: $ model X-ray surface brightness and SZE $y$ parameter distributions
65: results in a description of the density distribution, which can be
66: integrated to obtain the gas mass. The difference in dependence on gas
67: density and temperature of X-ray emissivity and the SZE $y$ parameter
68: makes the combination of these two methods of observation potentially very
69: powerful. Because of this difference, the observability of clusters
70: via each method is affected differently by the impact of physics in cluster cores including
71: radiative cooling and feedback mechanisms, as well as the transient
72: boosting of surface brightness and spectral temperature generated
73: during merging events \citep{roettmass,motl_sf}. These two methods not only select a
74: different sample of clusters, but combined SZE/X-ray observations of
75: individual clusters allow one to extract the density and temperature
76: of the gas without relying on X-ray spectral temperatures.
77: \subsection{Isothermal Beta Models}
78: Under the assumption that the gas in clusters is isothermal, one can
79: fit an isothermal $\beta$-model to the data in order to deduce the
80: density profile. To generate the projected X-ray surface
81: brightness profile, we integrate the $\beta$-model density distribution
82: \begin{equation}
83: S_X = \frac{1}{4 \pi (1+z)^4} \int n_e(r) n_H(r) \Lambda(T) dl,
84: \end{equation}
85: where in the bremsstrahlung limit, $\Lambda(T) \propto T^{1/2}$, and
86: in a fixed X-ray band is more weakly dependent on temperature \citep{mohr99}.
87: This integration results in
88: \begin{equation}
89: S_X(b) = S_{X0} \left(1 + \left(\frac{b}{r_c}\right)^2 \right) ^
90: {\frac{1}{2} - 3\beta},
91: \end{equation}
92: where $S_{X0}$ is the fitted central X-ray surface brightness
93: of the model, and $b$ indicates the projected radius.
94: Similarly for the SZE, a $\beta$-model density profile can be integrated
95: \begin{equation}
96: y = \int \sigma_T n_e(r) \frac{k_b T}{m_e c^2} dl,
97: \end{equation}
98: which results in a
99: projected radial distribution of the Compton $y$ parameter
100: \begin{equation}
101: y(b) = y_0 \left(1 + \left(\frac{b}{r_c}\right)^2\right)^{\frac{1}{2}
102: - \frac{3\beta}{2}}.
103: \end{equation}
104: For very hot clusters (T $>$ 10 keV), relativistic corrections must be included to
105: the SZE integral \citep{itoh}.
106: While these fits result in a description of the cluster density
107: profile, we must be aware that that description is only approximate,
108: since it has been shown both in simulations and observations \citep{loken,utp_obs} that
109: many clusters show a radial dependence of temperature. That means that
110: when using isothermal models (the above equations) to fit the data, we should expect error
111: to be introduced in the derived quantities.
112: \subsection{SZE/X-ray Derived Quantities}
113: Recent studies have used the values of the
114: $\beta$-model parameters determined from the X-ray surface brightness
115: profiles combined with SZE cluster observations to determine the value of the Hubble constant ($H_0$) \citep{reese,bonamente} and the cluster gas fraction \citep{joy,laroque}. While joint fits
116: of X-ray and SZE interferometric data are used to determine the $\beta$-model
117: parameters in these studies, the X-ray data drives the fit, since the
118: SZE data currently lacks the resolution (and interferometric U-V plane
119: coverage) to constrain the parameters well \citep{laroque}.
120:
121: An additional calculation can be performed using combined SZE/X-ray
122: data from clusters. It is expected both from analytic arguments and numerical simulations
123: that the integrated SZE signal, as a measure of the total ICM pressure in clusters, should be an excellent
124: proxy for cluster total mass \citep{dasilva,motl05,nagai,kravtsov06}. This
125: presupposes that one can determine accurately the value of the
126: integrated SZE signal to some mass-scaled cluster radius, as well as
127: perform an accurate calibration of this relationship
128: observationally. In \citet{motl05}, we showed that the value of
129: $Y_{500}$, the integrated Compton $y$ parameter inside a radius where
130: $\delta \rho/\rho$ = 500 (with respect to the critical density) accurately measures the cluster total mass
131: inside that same radius. In order to accurately calibrate this
132: relationship, one must measure the $Y$-$M$ relationship to high
133: precision at low redshift, where clusters have combined SZE/X-ray
134: observations to use. Then the measured relation, scaled for
135: redshift, can be used to determine masses of SZE-selected clusters,
136: which should be indentifiable to high redshifts, in order to constrain
137: cosmology.
138:
139: Since SZE radial profiles are of relatively poor
140: quality so far, a
141: direct determination of the compatibility of the X-ray $\beta$-model
142: parameters with the SZE profiles in clusters can not currently be done
143: observationally. Here we compare the values of $\beta$-model
144: parameters for a large number of simulated clusters when fitting the
145: X-ray surface brightness profile, the SZE radial Compton $y$ profile,
146: and jointly fitting both X-ray and SZE profiles to a common
147: $\beta$-model. We show that the use of X-ray parameter values leads to
148: biased estimates of $Y_{500}$, as well as the integrated gas mass
149: $M_{500,gas}$.
150:
151: We discuss our numerical simulations in Section 2, results of the
152: analysis in Section 3, the consequences of the use of X-ray
153: $\beta$-model parameters for estimating the $Y_{500}$-$M_{500}$
154: relationship for clusters in Section 4, and discussion and conclusions
155: in Section 5.
156: \section{Numerical Simulations}
157: Our simulations use the hybrid Eulerian adaptive mesh refinement hydro/N-body code
158: \textit{Enzo} (\citet{enzo}; http://cosmos.ucsd.edu/enzo)
159: to evolve both the dark matter and baryonic fluid in the clusters,
160: utilizing the piecewise parabolic method (PPM) for the
161: hydrodynamics. With up to seven levels of refinement
162: in high density regions, we attain spatial resolution up to
163: $\sim \; 16 \; h^{-1}$ kpc in the clusters. We assume a concordance
164: $\Lambda$CDM cosmological model with the following parameters:
165: $\Omega_{\mathrm{b}} = 0.026$, $\Omega_{\mathrm{m}} = 0.3$,
166: $\Omega_{\Lambda} = 0.7$, $\mathrm{h} = 0.7$, and $\sigma_{8} = 0.9$.
167: Refinement of high density regions is performed as described in \citet{motl04}.
168:
169: We have constructed a catalog of AMR refined clusters identified in the simulation
170: volume as described in \citet{loken}. The catalog of clusters used in
171: this study includes the effects of radiative cooling, models the loss of low
172: entropy gas to stars, adds a moderate
173: amount of supernova feedback due to Type II supernovae in the zones
174: where stars form, and is identified as the SFF (Star Formation with Feedback) catalog in \citet{hall06}. The catalog includes clusters with total mass
175: (baryons + dark matter) greater than $10^{14} M_{\odot}$ out to z=2 in
176: the simulation. This catalog
177: includes roughly 100 such clusters at z=0, and has 20 redshift
178: intervals of output, corresponding to a total of roughly 1500 clusters
179: in all redshift bins combined.
180:
181: For parts of the analysis, we have cleaned the cluster sample
182: as described in \citet{hall06}, removing cool core clusters and
183: obviously disturbed clusters. This is done primarily to more closely mimic
184: observational studies of cluster properties. We examine by eye all cluster projections and
185: remove any with obvious double peaks in the X-ray or SZE
186: surface brightness images, have disturbed morphology within R =
187: $1h^{-1}$ Mpc, exhibit edges consistent with shocked gas, or those
188: that have cool cores. The cool core clusters are identified as those
189: with a projected emission-weighted temperature profile which declines at small
190: radius, or those with strongly peaked X-ray emission. The cool core
191: clusters are eliminated because as we have shown in \citet{hall06},
192: they lead to strong biases and increased scatter in estimates of
193: cluster physical properties when standard observational methods are
194: used. We show in our previous work that eliminating obviously
195: disturbed clusters has a minor, but measurable effect on
196: observationally derived quantities. Indeed the simple assumptions
197: that are typically made in the observational derivation (e.g., spherical
198: symmetry, hydrostatic equilibrium) clearly do not apply
199: to such clusters. We also use two orthogonal
200: projected images for each cluster. The final cleaned sample contains 493
201: cluster projected images from a series of evolutionary epochs from z=0
202: to z=2.
203: \section{Results}
204: \subsection{Isothermal $\beta$-model Fits}
205: \begin{figure*}
206: \begin{center}
207: \includegraphics[width=0.9\textwidth]{./f1.eps}
208: \vspace{-5mm}
209: \end{center}
210: \caption{Upper panels: Left panel shows comparison of fitted $\beta$
211: values for clusters in our numerical sample at z=0 out to $r_{2500}$ using
212: the X-ray and SZE images. Right panel shows comparison of fitted value
213: of $r_{c}$ for same clusters. Middle panels: Same as above, except
214: values are for profiles fitted out to $r{500}$. Lower panels: Same as
215: above, but for profiles fitted out to $r_{200}$.}
216: \label{6panel}
217: \end{figure*}
218: We have generated images of our simulated cluster catalog by
219: projecting the physical quantities on the grid to get the value of the
220: SZE Compton $y$ parameter and the X-ray surface brightness. For the
221: X-ray, we have used a simple bremsstrahlung emissivity. It is a
222: simpler calculation than using a model X-ray emissivity, and we will
223: show later that the values of the $\beta$-model parameters are nearly
224: identical irrespective of which emissivity calculation we use.
225:
226: From these images, we have created radial profiles in annular bins,
227: which we subsequently fit to $\beta$-model profiles. Each
228: cluster, for both the SZE and X-ray profile, has a set of
229: $\beta$-model parameters that describe it, though there is some
230: degeneracy in the parameters in each case. The three $\beta$-model
231: parameters, namely the normalization ($S_{X0}$ or $y_0$), the value of
232: the core radius ($r_{core}$), and the power law index $\beta$ are left
233: as free parameters in the fit. We have fit to each of three limiting
234: outer radii in both the SZE and X-ray case, $r_{2500}$, $r_{500}$, and
235: $r_{200}$. In each case, the subscript indicates the average
236: overdensity with respect to critical inside that radius. This radius
237: is calculated from the simulation data using the overdensity of the
238: dark matter. We have used the full sample for this part of the analysis,
239: but have excluded the cool core region (typically $\approx$100kpc) of the profiles from the
240: fitting procedure.
241:
242: We have found that there are significant differences in the values of
243: the fitted model parameters depending on whether the X-ray or SZE
244: profiles are used. Figure \ref{6panel} shows
245: the comparison of the values of $\beta$ and $r_{core}$ plotted against
246: one another for individual z=0 clusters in our catalog for the three
247: limiting outer radii used. It is clear that there is a large amount of
248: scatter in the relationship for the $\beta$ values, and also at
249: $r_{500}$ and $r_{200}$, a definite discrepancy. Fitting SZE profiles results in a
250: consistently higher value of $\beta$ than does fitting X-ray
251: profiles. While there is scatter in the compared values of
252: $r_{core}$, there is general agreement within the errors.
253: However, since there is a degeneracy between $r_{core}$
254: and $\beta$ in the fitting, some scatter is expected. Since the
255: values of $\beta$ and $r_{core}$ are used directly in the equation for the
256: density profile, inconsistent values for these parameters lead to
257: different deduced density profiles, and discrepant values
258: for the cluster gas mass.
259:
260: \begin{figure}
261: \begin{center}
262: \includegraphics[width=0.5\textwidth]{./f2.eps}
263: \end{center}
264: \vspace{-5mm}
265: \caption{Upper: Ratio of SZE $\beta$ fitted to
266: radial profiles from projections of full SFF simulated sample of
267: clusters to $\beta$ fitted to X-ray radial profiles for the same
268: clusters. Median values and 1$\sigma$ error bars for the ratio
269: $\beta_{SZE}/\beta_{X-ray}$ in z=0 clusters fitted to each of three radii $r_{2500}$, $r_{500}$,
270: and $r_{200}$ using the isothermal $\beta$-model. Lower: Median values and 1$\sigma$ error bars for the
271: ratio $r_{core,SZE}/r_{core,X-ray}$ in z=0 clusters for the three radii $r_{2500}$, $r_{500}$,
272: and $r_{200}$.}
273: \label{comp_rad}
274: \end{figure}
275: Next, looking at Figure \ref{comp_rad}, we show the 1$\sigma$ scatter
276: in the ratio of the SZE deduced parameters and the X-ray parameters
277: for each of our three fiducial radii for all clusters in our catalog. While there is not a
278: statistically significant bias in $r_{core}$, one clearly exists in $\beta$ at the larger radii. To answer the question of why the isothermal $\beta$-model fit for
279: X-ray and SZE observations give a different set of model parameters in
280: each case, consider the simple argument in the next section.
281: \subsection{The Problem with the Isothermal Model}
282: The X-ray and SZE surface brightness functions resulting from the isothermal
283: $\beta$-model density profile integration have distributions
284: at $r>r_c$ which depend on radius as
285: \begin{equation}
286: S_X(r) \propto r^{1-6\beta}
287: \end{equation}
288: and
289: \begin{equation}
290: y(r) \propto r^{1-3\beta}.
291: \end{equation}
292:
293: We have fit the
294: temperature profile of our simulated clusters to a universal
295: temperature profile (UTP) of the form
296: \begin{equation}
297: T(r) = \langle T \rangle_{500}T_{0}\left(1 + \left(\frac{r}{\alpha r_{500}}\right)^{2}\right)^{-\delta},
298: \end{equation}
299: where $T_{500}$ indicates the average spectral
300: (in this study, emission-weighted) temperature inside $r_{500}$. $T_0$, $\alpha$, and
301: $\delta$ are dimensionless fitted parameters to the spherically averaged (from the
302: three-dimensional simulated data) temperature profiles of all clusters at
303: each redshift in the simulations used in this study. The mean values
304: of these parameters for clusters at select redshifts from our
305: simulations are shown in Table
306: \ref{par_tab}. We have used the redshift-specific mean value for
307: each cluster in the analysis.
308: \begin{table}
309: \caption{Mean parameters for Universal Temperature Profile}
310: \begin{tabular}{cccccc}
311: \hline
312: \hline
313: Parameter & $z=0$ & $z=0.1$ & $z=0.25$ & $z=0.5$ & $z=1.0$ \\
314: \hline
315: $T_0$ & 1.25$\pm$0.06\tablenotemark{a} & 1.27$\pm$0.06 & 1.31$\pm$0.08 & 1.37$\pm$0.09 &
316: 1.37$\pm$0.14 \\
317: $\delta$ & 0.51$\pm$0.21 & 0.51$\pm$0.19& 0.42$\pm$0.13& 0.43$\pm$0.12
318: & 0.53$\pm$0.22\\
319: $\alpha$ & 1.17$\pm$0.46 & 1.12$\pm$0.40& 0.91$\pm$0.34& 0.80$\pm$0.28
320: & 0.90$\pm$0.45 \\
321: \hline
322: \end{tabular}
323: \label{par_tab}
324: \tablenotetext{a}{Error bars indicate 1$\sigma$ dispersion.}
325: \end{table}
326:
327: If the cluster's true gas temperature declines with radius with the
328: above described dependence
329: \begin{equation}
330: T(r) \propto r^{-2\delta},
331: \end{equation}
332: then the cluster observable profiles have dependence at $r>r_c$
333: \begin{equation}
334: S_X(r) \propto r^{1-6\beta-\delta}
335: \end{equation}
336: and
337: \begin{equation}
338: y(r) \propto r^{1-3\beta-2\delta}.
339: \end{equation}
340: Our simulations show a typical value of $\delta=0.5$ (see Table \ref{par_tab}).
341: If we assume the true cluster density profile is a $\beta$-model
342: modified in this way by a UTP for the temperature profile (and
343: $\delta$=0.5), and that the true value for $\beta$ set by the cluster
344: density profile is $\beta$=0.8, then the radial dependence of X-ray
345: surface brightness and SZE surface brightness are
346: \begin{equation}
347: S_X(r) \propto r^{-4.3}
348: \end{equation}
349: and
350: \begin{equation}
351: y(r) \propto r^{-2.4}.
352: \end{equation}
353:
354: Finally, if we then fit an isothermal $\beta$-model to cluster
355: profiles with the above dependence, setting powers equal for the X-ray
356: \begin{equation}
357: 1-6\beta = -4.3
358: \end{equation}
359: we would get $\beta = 0.88$,
360: and for the SZE, the isothermal fit would give us
361: \begin{equation}
362: 1-3\beta = -2.4
363: \end{equation}
364: or $\beta = 1.13$.
365:
366: While the declining temperature profile has a relatively
367: small effect on the $\beta$ value extracted from an isothermal fit to
368: the X-ray surface brightness (+10\%), there is a larger effect on the
369: value of $\beta$ in the SZE case (+41\%). This is not surprising given
370: the difference in temperature dependence between Equations 2 and
371: 4. Table \ref{betas} shows the
372: median values of fitted $\beta$ in our cluster sample out to
373: $r_{200}$, with 1$\sigma$ scatter. Indeed the variation between the
374: X-ray and SZE fits is consistent with the simple analysis shown
375: above. Also, one line of the table shows the result of fitting
376: the true density profiles extracted from the simulated data. Though
377: the differences between the true value and the SZE and X-ray fitted
378: values are bigger than expected from a simple analysis, the true value
379: is indeed smaller than that fitted from the emission profiles. There
380: are additional sources of bias which may be introduced due to
381: clumpiness in the ICM and deviations from spherical symmetry which may
382: contribute to the larger difference in fitted parameters
383: \citep{sulkanen, nagai00}. There is
384: also the degeneracy in $r_c$ and $\beta$ in any model fitting to these
385: profiles which can contribute to differences. We also show in the last
386: line of the table the result of fitting the isothermal $\beta$-model
387: to the profile of X-ray surface brightness resulting from a projection
388: of the Raymond-Smith \citep{rs} model emissivity. It is clear that the
389: statistical properties of the fitted parameters are nearly
390: indistinguishable from those fitted to the simple bremsstrahlung model.
391:
392: It is therefore clear that the effect of a declining temperature
393: profile in real clusters is to alter the fitted values of
394: $\beta$ in this simple model, such that the SZE isothermal $\beta$ is larger
395: than both the X-ray, and the ``true'' value associated with the
396: density profile of the cluster. Indeed a similar analysis by
397: \citet{ameglio} has shown results consistent with ours in this regard.
398: We also find no significant trend in the SZE/X-ray parameter ratios
399: with redshift.
400: \begin{table}
401: \caption{Median Parameter Values from Profile Fitting to $r_{200}$}
402: \begin{center}
403: \begin{tabular}{ccccccc}
404: Method & $\beta$ & +1$\sigma$ & -1$\sigma$ & $r_{core}$(kpc) & +1$\sigma$ & -1$\sigma$\\
405: \hline
406: Iso X-ray & 0.84 & 1.02 & 0.70 & 168 & 330 & 99\\
407: Iso SZE & 1.05 & 1.27 & 0.88 & 196 & 340 & 130\\
408: U-$\beta$ X-ray & 0.81 & 0.97 & 0.67 & 165 & 320 & 96\\
409: U-$\beta$ SZE & 0.82 & 0.97 & 0.69 & 160 & 260 & 103 \\
410: Density & 0.70 & 0.87 & 0.55 & 137 & 242 & 73 \\
411: Iso Raymond-Smith & 0.82 & 1.03 & 0.67 & 168 & 318 & 96 \\
412: \hline
413: \end{tabular}
414: \label{betas}
415: \end{center}
416: \end{table}
417: \subsection{Fitting to a UTP Modified Model}
418: Since the variation in model parameters between X-ray and SZE fitting
419: appears to be due to the failure to account for the radial dependence of
420: temperature in the intracluster medium, it makes sense to use a model
421: that includes this radial dependence for fitting the surface
422: brightness. Our non-isothermal $\beta$-model for the surface
423: brightness is created by integrating the expression for the X-ray
424: surface brightness and Compton $y$ parameter using the standard
425: $\beta$-model for the density, and the UTP for the
426: temperature. We refer to this model as the U-$\beta$. Integrating Equations 2 and 4 with the substitution of
427: the $\beta$-model density profile (Eq. 1) and the UTP (Eq. 8) results
428: in the fitting relations for the X-ray and SZE surface brightness
429: \begin{equation}
430: S_X(b) = S_{X0} \left( 1 + \left(\frac{b}{r_c} \right)^2
431: \right)^{-3\beta} \left(1+\left(\frac{b}{\alpha r_{500}} \right)^2
432: \right)^{-\frac{\delta}{2}} I_X(b)
433: \end{equation}
434: and
435: \begin{equation}
436: y(b) = y_{0} \left( 1 + \left(\frac{b}{r_c}\right)^2
437: \right)^{-\frac{3\beta}{2}} \left(1+\left(\frac{b}{\alpha r_{500}} \right)^2
438: \right)^{-\delta} I_{SZ}(b),
439: \end{equation}
440: respectively, where $I_X(b)$ and $I_{SZ}(b)$ are line integrals described in the Appendix.
441:
442: As described in Section 3.2, the parameters $\alpha$ and $\delta$ are
443: fixed in the fitting to the surface brightness distributions, and
444: result from fitting of the spherically averaged (from the
445: three-dimensional simulated data) temperature profiles from
446: the simulated clusters. The average values of the parameters in each
447: redshift bin are used
448: for fitting the surface brightness profiles of a cluster in that same
449: redshift bin. When fitting the cluster profiles to these relations, we find that
450: there is now consistency in the values of $\beta$ and $r_c$ between
451: X-ray and SZE fitting, shown in Figure \ref{comp_rad_utp}. While there is still some scatter, the offset is
452: removed. Table \ref{betas} shows the results for all clusters in the
453: sample when fitting to an isothermal $\beta$-model or a non-isothermal
454: (the U-$\beta$) model. The value of $\beta$ is virtually identical in SZE and
455: X-ray fitting when using the U-$\beta$. Additionally, they result in similar
456: distributions of $r_c$ values.
457:
458: Note that this model has the same number of free parameters (three) used to
459: fit the surface brightness as a standard $\beta$-model, the others are
460: fixed from simulations. We should note the well-known caveat about
461: $\beta$-model fitting, that the values of $r_{core}$ and $\beta$ are
462: somewhat degenerate in this fitting.
463: \begin{figure}
464: \begin{center}
465: \includegraphics[width=0.5\textwidth]{./f3.eps}
466: \vspace{-5mm}
467: \end{center}
468: \caption{Upper: Ratio of SZE $\beta$ fitted to
469: radial profiles from projections of full SFF simulated sample of
470: clusters to $\beta$ fitted to X-ray radial profiles for the same
471: clusters. Median values and 1$\sigma$ error bars for the ratio
472: $\beta_{SZE}/\beta_{X-ray}$ in z=0 clusters when fitted to each of three radii $r_{2500}$, $r_{500}$,
473: and $r_{200}$ when fitting with a U-$\beta$ model. Lower: Median values and 1$\sigma$ error bars for the
474: ratio $r_{core,SZE}/r_{core,X-ray}$ in z=0 clusters for the three radii $r_{2500}$, $r_{500}$,
475: and $r_{200}$.}
476: \label{comp_rad_utp}
477: \end{figure}
478: \section{Consequences of the $\beta$ Incompatibility: Example Calculation of $Y_{500}$ vs $M$}
479: Using the isothermal $\beta$-model parameters from a fit dominated by the X-ray data to do
480: SZE or combined X-ray/SZE analysis introduces an additional error or
481: bias to various derived quantities. Here, we characterize
482: the nature of these errors on the measured $Y_{500}$ vs $M_{500,gas}$
483: relation for clusters. As described in the introduction, it has been
484: shown that the integrated Compton $y$ parameter inside $r_{500}$ is an
485: excellent proxy for total mass
486: \citep{dasilva,motl05,nagai,kravtsov06}. We examine whether one can accurately
487: determine the true value of $Y_{500}$ (and $M_{500,gas}$) from a
488: cluster observation using standard observational techniques.
489:
490: Generating a value for $Y_{500}$ using a $\beta$-model fit
491: dominated by the X-ray emission, one must extrapolate to $r_{500}$
492: using the model for the SZE emission. Currently existing cluster SZE
493: profiles from interferometric instruments do not constrain individual
494: $\beta$-model parameters well. Since we have shown that the
495: isothermal $\beta$-model paramters for SZE and X-ray cluster profiles are
496: different, there is an error introduced. Additionally, there are
497: errors introduced in the calculation of the cluster gas mass, $M_{500,gas}$, since in
498: any individual cluster, the $\beta$-model for the density deduced from
499: the observations is only approximately correct.
500: \subsection{Estimation of Y and M}
501: We have used the simulated cluster sample described above (SFF) to determine the
502: systematic errors introduced in this procedure. We have analyzed the
503: simulated clusters such that our synthetic observations are analogous to
504: unbiased, high signal-to-noise observations of real clusters. The simulated
505: observations are idealized, since no instrumental effects
506: or foreground/background source removal are simulated. We have performed our calculations on projected SZE and
507: X-ray images generated from each simulated cluster, assuming the
508: X-ray profiles could be determined to $r_{500}$.
509:
510: Using this method, and assuming that the values of $\beta$ and $r_c$
511: are defined by fitting of the X-ray surface brightness profile, we have characterized errors by taking the model parameters from
512: the X-ray fitting and calculated the estimated values of $Y_{500}$
513: and $M_{500,gas}$ from each cluster. We then compare the value to the true
514: value taken from the simulation grid.
515:
516: We have used the isothermal
517: model fits and also the fits to the U-$\beta$ models. In
518: each case, $Y_{500}$ is determined by extrapolating the model for
519: $y(r)$ using the parameters $\beta$ and $r_c$ from the X-ray fitting,
520: and the value of $y_0$ generated by fitting the SZE profile with
521: $\beta$ and $r_c$ fixed. $M_{500,gas}$ is estimated by integrating the
522: $\beta$ model for density out to an overdensity of 500. This method
523: effectively measures $r_{500}$ from the fitted gas density profile in order to
524: get the integrated gas mass. For the
525: U-$\beta$ cases, the values of $\alpha$ and $\delta$ are fixed at the mean
526: values for the whole sample of clusters at each redshift, recognizing
527: that these values would be provided by simulations, and would not be
528: left as free parameters in the observational analysis.
529: \subsection{Comparison of Isothermal and U-$\beta$ Methods}
530: \begin{figure}
531: \includegraphics[width=0.5\textwidth]{./f4.eps}
532: \caption{Upper: Integrated Compton $y$ parameter inside $r_{500}$ from
533: simulated clusters at z=0 epoch plotted against integrated mass inside the same
534: radius. Stars are for the true values on the simulation grid,
535: red diamonds represent values deduced by fitting to isothermal
536: $\beta$-models, blue triangles are for fitting to a U-$\beta$ model. Lower: Log deviation of isothermal and U-$\beta$ points
537: from the best-fit scaling relation for the true values.}
538: \label{yvalues}
539: \end{figure}
540: On first inspection of Figure \ref{yvalues}, it is not clear that
541: either the isothermal or U-$\beta$ generates a more accurate
542: result. The plot shows values from all clusters from the z=0
543: simulation epoch. The true values of $Y_{500}$ and $M_{500,gas}$ are plotted as
544: stars, the isothermal model values as diamonds, and the U-$\beta$ model
545: values as triangles.
546:
547: The plot of deviation of these values from the best fit to the
548: true values shows some difference between these
549: two methods. However, it is important to remember that there are
550: deviations from both the true integrated SZE, and the true gas mass,
551: so these points deviate in both dimensions of the plot. Since errors
552: in both are correlated, due to the dependence on the determination of
553: $r_{500}$ from the $\beta$-model, this should reduce the apparent
554: separation of points from the best fit, even when the values of both
555: parameters are in error.
556: \begin{figure}
557: \includegraphics[width=0.5\textwidth]{./f5.eps}
558: \caption{\small Upper: Plot of the ratio of estimated $Y_{500}$ to its true value
559: (y-axis) versus the ratio of estimated $M_{500,gas}$ to its true value
560: for 493 simulated cluster projections for isothermal $\beta$-model fits.
561: Lower: Ratio of estimated $Y_{500}$ to its true value
562: versus the ratio of estimated $M_{500,gas}$ to its true value for U-$\beta$-model fits. Lines indicate location of perfectly accurate estimation
563: of each quantity.}
564: \label{yvsm}
565: \end{figure}
566:
567: This important point is illustrated in Figure \ref{yvsm}, which shows
568: the the ratios of estimated to true values of $Y_{500}$ and $M_{500,gas}$
569: for each cluster. In both the isothermal (upper plot) and U-$\beta$
570: (lower plot) methods, there is a clear trend in over and
571: underestimation in both values. These figures also illustrate nicely
572: the improvement in values one gets with the U-$\beta$ method. The U-$\beta$
573: points are clustered more strongly around the correct values than are
574: the isothermal estimates. This effect is illustrated in Figure \ref{shift}, where we
575: have plotted the true values of $y$ and $M$, with lines indicating the
576: corresponding estimates of those values for the isothermal and U-$\beta$
577: methods. There are typically overestimates in both estimated $Y_{500}$
578: and $M_{500,gas}$, and the use of the U-$\beta$ typically brings the estimated
579: values back toward the true.
580:
581: The outliers in these plots are also interesting. The clusters which
582: give strong overestimates of $Y$ and $M_{gas}$ typically result from
583: line-of-sight overlap of multiple structures. This enhances the
584: integrated SZE signal in projection, but overestimates the true value
585: for the main cluster. The clusters with low estimated values of $Y$
586: and/or $M_{gas}$ typically have poor quality $\beta$-model fits (as
587: measured by a $\chi^2$ statistic), and appear to have
588: systematically low fitted values for the core radius. Some of these
589: clusters appear to have weak cool cores, not meeting the cool core
590: criteria, and thus were not excluded from the sample. It is interesting
591: that the clusters with mass underestimates do not all generate
592: underestimates of $Y$. This effect warrants further study.
593:
594: \begin{figure}
595: \includegraphics[width=0.5\textwidth]{./f6.eps}
596: \caption{Plot of eight clusters on the $Y_{500}$-$M_{500,gas}$
597: relation chosen to span a wide mass range, but otherwise randomly. Stars indicate the true values from the simulation, red
598: diamonds the isothermal $\beta$-model estimates, and blue triangles
599: the U-$\beta$ model estimates. Lines link points
600: associated with the same cluster.}
601: \label{shift}
602: \end{figure}
603: Table \ref{y500_tab} also shows the
604: result of this analysis. When using isothermal models, the median
605: estimated value of $Y_{500}$ is 23\% larger than the true value (the
606: mean is a 28\% overestimate). Additionally, when estimating the gas
607: mass, the median estimated value is a 16\% overestimate. In previous
608: work \citep{hall06}, we have shown that the magnitude of the
609: overestimate via X-ray $\beta$-model methods is lower than this, but
610: in that case we assumed one could correctly calculate the value of
611: $r_{500}$ with no error. In either case, this overestimate is
612: consistent with our previous work, and with that of others
613: \citep{mohr99, math99}. The overestimate in mass results from
614: substructure, merging and other physical processes not described by a
615: simple model. In the current analysis, the value of
616: $r_{500}$ is deduced directly from the $\beta$-model. In contrast, the
617: U-$\beta$ method gets closer to the correct value for $Y_{500}$, with a
618: median value 11\% higher than the true value. For the mass, the U-$\beta$
619: method gets an estimate close to the true mass, a median overestimate of only 4\%.
620:
621: Additionally, using the U-$\beta$ method reduces the scatter in $Y_{500}$
622: values, as shown in Figure \ref{yhist}. The distribution of values is
623: more sharply peaked, with a smaller high end tail in addition to a
624: reduced bias compared to the isothermal method. In mass, the result
625: is a reduced bias in the median values with a slight improvement in
626: the scatter as shown in Figure \ref{mhist}.
627: \begin{table}
628: \caption{Ratio of Estimated to True Values of Simulated Cluster Properties to $r_{500}$}
629: \begin{center}
630: \begin{tabular}{ccc}
631: & Isothermal & U-$\beta$ \\
632: \hline
633: $\langle y_{500,est}/y_{500,true} \rangle$ & 1.28 & 1.13 \\
634: $(y_{500,est}/y_{500,true}), median$& 1.23 & 1.11\\
635: 1$\sigma$ upper & 1.50 & 1.24\\
636: 1$\sigma$ lower & 1.08 & 1.03\\
637: $\langle M_{500,gas,est}/M_{500,gas,true} \rangle$ & 1.17 & 1.05 \\
638: $(M_{500,gas,est}/M_{500,gas,true}), median$& 1.16 & 1.04\\
639: 1$\sigma$ upper & 1.31 & 1.18\\
640: 1$\sigma$ lower & 1.06 & 0.94\\
641: \hline
642: \end{tabular}
643: \label{y500_tab}
644: \end{center}
645: \end{table}
646:
647: \begin{figure}
648: \includegraphics[width=0.5\textwidth]{./f7.eps}
649: \caption{Histogram of 493 simulated cluster projections of ratio of estimated
650: $Y_{500}$ to true value for isothermal (black lines) and U-$\beta$ (red
651: lines) methods. Dotted lines indicate median values for each distribution.}
652: \label{yhist}
653: \end{figure}
654: \begin{figure}
655: \includegraphics[width=0.5\textwidth]{./f8.eps}
656: \caption{Histogram of 493 simulated cluster projections of ratio of estimated
657: $M_{500,gas}$ to true value for isothermal (black lines) and U-$\beta$ (red
658: lines) methods. Dotted lines indicate median values for each distribution.}
659: \label{mhist}
660: \end{figure}
661: \section{Conclusions}
662: There is an inconsistency between X-ray and SZE fitted
663: model parameters that leads to a bias in deduced values of $Y_{500}$ and
664: $M_{500,gas}$ when using isothermal $\beta$-models. Using our U-$\beta$
665: model reduces the bias and scatter, resulting in a more precise and
666: accurate determination of the $y$-$M$ scaling relation for
667: clusters.
668:
669: X-ray and SZE radial profiles from the same galaxy cluster can
670: not be fit well by identical isothermal $\beta$-models. The stronger
671: dependence of the SZE emission on temperature leads to problems when
672: the cluster temperature declines with radius. In contrast, the strong
673: dependence of the X-ray emission on density minimizes the error
674: introduced by variations in cluster temperature. We show that fitting
675: either the X-ray or SZE profiles to a modified, non-isothermal
676: $\beta$-model corrected by the inclusion of a univeral temperature
677: profile for clusters results in better consistency with the real
678: values of $Y_{500}$ and $M_{500,gas}$.
679:
680: We expect this inconsistency should affect measurements of the Hubble
681: constant at some level, indeed it appears to result in a bias in
682: $H_0$, as shown by \citet{ameglio}. There will
683: certainly be an effect on gas fraction determinations, though it will
684: also depend on any bias inherent in X-ray hydrostatic total mass
685: estimates. While current precision in SZE/X-ray derived quantities is
686: not high enough to reveal effects at the 10-20\% level of the type
687: described here, we expect it will be higher with larger samples and new
688: instruments available in the near term. When contemplating clusters as
689: precision cosmological tools, effects as this level must be considered.
690:
691: While some have used double $\beta$-models \citep{laroque} to fit cluster radial
692: profiles with some success, a model like the U-$\beta$ is preferred, since
693: it is physically motivated by the observed ICM properties. This work
694: illustrates a simple
695: modification of the isothermal $\beta$-model to more realistically
696: account for cluster physics, and thus more accurately and precisely
697: measure cluster properties.
698: \acknowledgments
699: The simulations presented in this work were conducted at the
700: National Center for Supercomputing Applications at the University of
701: Illinois, Urbana-Champaign through computer allocation grant
702: AST010014N. We also acknowledge the support of the NSF through grant AST-0407368.
703: \appendix
704: \section{Derivation of UTP Surface Brightness Model}
705: To derive a surface brightness model resulting from the UTP modified
706: $\beta$-model, we take the standard $\beta$-model for density,
707: \begin{equation}
708: n_{e}(r) = n_{e0}\left(1 + \left(\frac{r}{r_{c}}\right)^{2}\right)^{-3\beta/2},
709: \end{equation}
710: and the UTP model for temperature,
711: \begin{equation}
712: T(r) = \langle T \rangle_{500}T_{0}\left(1 + \left(\frac{r}{\alpha r_{500}}\right)^{2}\right)^{-\delta},
713: \end{equation}
714: and substitute them into the integral of the X-ray surface brightness,
715: \begin{equation}
716: S_X(b) = \frac{1}{4\pi(1+z)^4}\int n_e(r) n_H(r) \Lambda(T(r)) dl.
717: \end{equation}
718: In the simple bremsstrahlung case, $\Lambda(T)$ = $\Lambda_0
719: T^{1/2}$. In that case,
720: \begin{equation}
721: S_X(b) = S_{X0} \left( 1 + \left(\frac{b}{r_c} \right)^2
722: \right)^{-3\beta} \left(1+\left(\frac{b}{\alpha r_{500}} \right)^2
723: \right)^{-\frac{\delta}{2}} I_X(b),
724: \end{equation}
725: where
726: \begin{equation}
727: S_{X0} = \frac{2{n_{e0}}^2 \Lambda_0(\langle T \rangle_{500}
728: T_0)^{1/2}}{4\pi\kappa(1+z)^4},
729: \end{equation}
730: where $\kappa = \mu_H/\mu_e$, the ratio of the mean molecular weights of hydrogen and electrons. For $r_c < \alpha r_{500}$,
731: \begin{equation}
732: I_X(b) = \frac{r_c}{2} \left( 1 + \left(\frac{b}{r_c}\right)^2 \right)^{1/2}
733: B\left(\frac{1}{2},3\beta + \frac{\delta}{2} - \frac{1}{2} \right)
734: F_{2,1} \left(\frac{\delta}{2}, \frac{1}{2};3\beta +
735: \frac{\delta}{2},1-\frac{{r_c}^2 + b^2}{\alpha^2 {r_{500}}^2 + b^2}
736: \right),
737: \end{equation}
738: where $F_{2,1}$ is Gauss' hypergeometric function and $B$ is the Beta
739: function defined by
740: \begin{equation}
741: B(x,y) = \frac{\Gamma(x) \Gamma(y)}{\Gamma(x+y)};x,y > 0.
742: \end{equation}
743:
744: For $r_c > \alpha
745: r_{500}$,
746: \begin{equation}
747: I_X(b) = \frac{\alpha r_{500}}{2} \left( 1 + \left(\frac{b}{\alpha r_{500}}\right)^2 \right)^{1/2}
748: B\left(\frac{1}{2},3\beta + \frac{\delta}{2} - \frac{1}{2} \right)
749: F_{2,1} \left(3\beta, \frac{1}{2};3\beta +
750: \frac{\delta}{2},1-\frac{\alpha^2 {r_{500}}^2 + b^2}{{r_c}^2 + b^2}
751: \right).
752: \end{equation}
753:
754: Similarly for the SZE, where
755: \begin{equation}
756: y = \int \sigma_T n_e(r) \frac{k_b T}{m_e c^2} dl,
757: \end{equation}
758: the substitution results in
759: \begin{equation}
760: y(b) = y_{0} \left( 1 + \left(\frac{b}{r_c}\right)^2
761: \right)^{-\frac{3\beta}{2}} \left(1+\left(\frac{b}{\alpha r_{500}} \right)^2
762: \right)^{-\delta} I_{SZ}(b),
763: \end{equation}
764: and
765: \begin{equation}
766: y_0 = 2\sigma_T n_{e0} \frac{k_b \langle T \rangle_{500} T_0}{m_e c^2}.
767: \end{equation}
768: For the two cases of $r_c < \alpha r_{500}$ and $r_c > \alpha
769: r_{500}$ respectively,
770: \begin{equation}
771: I_{SZ}(b) = \frac{r_c}{2} \left( 1 + \left(\frac{b}{r_c}\right)^2 \right)^{1/2}
772: B\left(\frac{1}{2},\frac{3\beta}{2} + \delta - \frac{1}{2} \right)
773: F_{2,1} \left(\delta, \frac{1}{2};\frac{3\beta}{2} +
774: \delta,1-\frac{{r_c}^2 + b^2}{\alpha^2 {r_{500}}^2 + b^2}\right),
775: \end{equation}
776: and
777: \begin{equation}
778: I_{SZ}(b) = \frac{\alpha r_{500}}{2} \left( 1 + \left(\frac{b}{\alpha r_{500}}\right)^2 \right)^{1/2}
779: B\left(\frac{1}{2},\frac{3\beta}{2} + \delta - \frac{1}{2} \right)
780: F_{2,1} \left(\frac{3\beta}{2}, \frac{1}{2};\frac{3\beta}{2} +
781: \delta,1-\frac{\alpha^2 {r_{500}}^2 + b^2}{{r_c}^2 + b^2}
782: \right).
783: \end{equation}
784:
785: Figure \ref{ival} shows the values for each of the line integrals as a function
786: of projected radius normalized to $r_{500}$ for typical model
787: parameters. We have created a calculator for the line integral in the
788: Interactive Data Language (IDL), which is available on-line at http://solo.colorado.edu/~hallman/UTP/Ib\_calc.tar.gz.
789: \begin{figure}
790: \begin{center}
791: \includegraphics[width=0.6\textwidth]{./f9.eps}
792: \end{center}
793: \caption{Value of line integrals $I_X$ and $I_{SZ}$as a function of radius normalized to
794: $r_{500}$. This calculation uses values near the mean of the
795: distribution for all clusters for the $\beta$-model paramters, and
796: $r_{500}$=1.0$h^{-1}$Mpc. $r_{core}$=160kpc, $\beta$=0.82,
797: $\alpha$=1.15, $\delta$=0.5, therefore this is the case where
798: $r_{core} < r_{500}$. Solid line is for X-ray integral,
799: dotted is for SZE.}
800: \label{ival}
801: \end{figure}
802:
803: %\input{ms.bbl}
804:
805:
806:
807: \begin{thebibliography}{22}
808: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
809:
810: \bibitem[{{Ameglio} {et~al.}(2006){Ameglio}, {Borgani}, {Diaferio}, \&
811: {Dolag}}]{ameglio}
812: {Ameglio}, S., {Borgani}, S., {Diaferio}, A., \& {Dolag}, K. 2006, \mnras, 369,
813: 1459
814:
815: \bibitem[{{Bonamente} {et~al.}(2006){Bonamente}, {Joy}, {LaRoque}, {Carlstrom},
816: {Reese}, \& {Dawson}}]{bonamente}
817: {Bonamente}, M., {Joy}, M.~K., {LaRoque}, S.~J., {Carlstrom}, J.~E., {Reese},
818: E.~D., \& {Dawson}, K.~S. 2006, \apj, 647, 25
819:
820: \bibitem[{{Brickhouse} {et~al.}(1995){Brickhouse}, {Raymond}, \& {Smith}}]{rs}
821: {Brickhouse}, N.~S., {Raymond}, J.~C., \& {Smith}, B.~W. 1995, \apjs, 97, 551
822:
823: \bibitem[{{Carlstrom} {et~al.}(2002){Carlstrom}, {Holder}, \& {Reese}}]{carl}
824: {Carlstrom}, J.~E., {Holder}, G.~P., \& {Reese}, E.~D. 2002, \araa, 40, 643
825:
826: \bibitem[{{Cavaliere} \& {Fusco-Femiano}(1978)}]{caval}
827: {Cavaliere}, A. \& {Fusco-Femiano}, R. 1978, \aap, 70, 677
828:
829: \bibitem[{{da Silva} {et~al.}(2000){da Silva}, {Barbosa}, {Liddle}, \&
830: {Thomas}}]{dasilva}
831: {da Silva}, A.~C., {Barbosa}, D., {Liddle}, A.~R., \& {Thomas}, P.~A. 2000,
832: \mnras, 317, 37
833:
834: \bibitem[{{Hallman} {et~al.}(2006){Hallman}, {Motl}, {Burns}, \&
835: {Norman}}]{hall06}
836: {Hallman}, E.~J., {Motl}, P.~M., {Burns}, J.~O., \& {Norman}, M.~L. 2006, \apj,
837: 648, 852
838:
839: \bibitem[{{Itoh} \& {Nozawa}(2004)}]{itoh}
840: {Itoh}, N. \& {Nozawa}, S. 2004, \aap, 417, 827
841:
842: \bibitem[{{Joy} {et~al.}(2001){Joy}, {LaRoque}, {Grego}, {Carlstrom}, {Dawson},
843: {Ebeling}, {Holzapfel}, {Nagai}, \& {Reese}}]{joy}
844: {Joy}, M., {LaRoque}, S., {Grego}, L., {Carlstrom}, J.~E., {Dawson}, K.,
845: {Ebeling}, H., {Holzapfel}, W.~L., {Nagai}, D., \& {Reese}, E.~D. 2001,
846: \apjl, 551, L1
847:
848: \bibitem[{{Kravtsov} {et~al.}(2006){Kravtsov}, {Vikhlinin}, \&
849: {Nagai}}]{kravtsov06}
850: {Kravtsov}, A.~V., {Vikhlinin}, A., \& {Nagai}, D. 2006, \apj, 650, 128
851:
852: \bibitem[{{LaRoque} {et~al.}(2006){LaRoque}, {Bonamente}, {Carlstrom}, {Joy},
853: {Nagai}, {Reese}, \& {Dawson}}]{laroque}
854: {LaRoque}, S.~J., {Bonamente}, M., {Carlstrom}, J.~E., {Joy}, M.~K., {Nagai},
855: D., {Reese}, E.~D., \& {Dawson}, K.~S. 2006, \apj, 652, 917
856:
857: \bibitem[{{Loken} {et~al.}(2002){Loken}, {Norman}, {Nelson}, {Burns}, {Bryan},
858: \& {Motl}}]{loken}
859: {Loken}, C., {Norman}, M.~L., {Nelson}, E., {Burns}, J., {Bryan}, G.~L., \&
860: {Motl}, P. 2002, \apj, 579, 571
861:
862: \bibitem[{{Mathiesen} {et~al.}(1999){Mathiesen}, {Evrard}, \& {Mohr}}]{math99}
863: {Mathiesen}, B., {Evrard}, A.~E., \& {Mohr}, J.~J. 1999, \apjl, 520, L21
864:
865: \bibitem[{{Mohr} {et~al.}(1999){Mohr}, {Mathiesen}, \& {Evrard}}]{mohr99}
866: {Mohr}, J.~J., {Mathiesen}, B., \& {Evrard}, A.~E. 1999, \apj, 517, 627
867:
868: \bibitem[{{Motl} \& {Burns}(2005)}]{motl_sf}
869: {Motl}, P.~M. \& {Burns}, J.~M. 2005, in X-Ray and Radio Connections (eds. L.O.
870: Sjouwerman and K.K Dyer) Published electronically by NRAO,
871: http://www.aoc.nrao.edu/events/xraydio Held 3-6 February 2004 in Santa Fe,
872: New Mexico, USA, (E8.03) 8 pages
873:
874: \bibitem[{{Motl} {et~al.}(2004){Motl}, {Burns}, {Loken}, {Norman}, \&
875: {Bryan}}]{motl04}
876: {Motl}, P.~M., {Burns}, J.~O., {Loken}, C., {Norman}, M.~L., \& {Bryan}, G.
877: 2004, \apj, 606, 635
878:
879: \bibitem[{{Motl} {et~al.}(2005){Motl}, {Hallman}, {Burns}, \&
880: {Norman}}]{motl05}
881: {Motl}, P.~M., {Hallman}, E.~J., {Burns}, J.~O., \& {Norman}, M.~L. 2005,
882: \apjl, 623, L63
883:
884: \bibitem[{{Nagai}(2006)}]{nagai}
885: {Nagai}, D. 2006, \apj, 650, 538
886:
887: \bibitem[{{Nagai} {et~al.}(2000){Nagai}, {Sulkanen}, \& {Evrard}}]{nagai00}
888: {Nagai}, D., {Sulkanen}, M.~E., \& {Evrard}, A.~E. 2000, \mnras, 316, 120
889:
890: \bibitem[{{O'Shea} {et~al.}(2005){O'Shea}, {Bryan}, {Bordner}, {Norman},
891: {Abel}, {Harkness}, \& {Kritsuk}}]{enzo}
892: {O'Shea}, B.~W., {Bryan}, G., {Bordner}, J., {Norman}, M.~L., {Abel}, T.,
893: {Harkness}, R., \& {Kritsuk}, A. 2005, {in Adaptive Mesh Refinement: Theory
894: and Applications} (Berlin: Springer), 341
895:
896: \bibitem[{{Reese} {et~al.}(2002){Reese}, {Carlstrom}, {Joy}, {Mohr}, {Grego},
897: \& {Holzapfel}}]{reese}
898: {Reese}, E.~D., {Carlstrom}, J.~E., {Joy}, M., {Mohr}, J.~J., {Grego}, L., \&
899: {Holzapfel}, W.~L. 2002, \apj, 581, 53
900:
901: \bibitem[{{Roettiger} {et~al.}(1996){Roettiger}, {Burns}, \&
902: {Loken}}]{roettmass}
903: {Roettiger}, K., {Burns}, J.~O., \& {Loken}, C. 1996, \apj, 473, 651
904:
905: \bibitem[{{Sulkanen}(1999)}]{sulkanen}
906: {Sulkanen}, M.~E. 1999, \apj, 522, 59
907:
908: \bibitem[{{Sunyaev} \& {Zeldovich}(1972)}]{sz}
909: {Sunyaev}, R.~A. \& {Zeldovich}, Y.~B. 1972, Comments on Astrophysics and Space
910: Physics, 4, 173
911:
912: \bibitem[{{Vikhlinin} {et~al.}(2005){Vikhlinin}, {Markevitch}, {Murray},
913: {Jones}, {Forman}, \& {Van Speybroeck}}]{utp_obs}
914: {Vikhlinin}, A., {Markevitch}, M., {Murray}, S.~S., {Jones}, C., {Forman}, W.,
915: \& {Van Speybroeck}, L. 2005, \apj, 628, 655
916:
917: \end{thebibliography}
918:
919:
920: \end{document}
921:
922: