0806.0850/ms.tex
1: \def\mtx{$M$-$T_{\rm X}$}
2: \def\mlx{$M$-$L_{\rm X}$}
3: \def\tx{T_{\rm X}}
4: \def\tew{T_{\rm EW}}
5: \def\tsl{T_{\rm SL}}
6: \def\ix{i_{\rm X}}
7: \def\Rap{R_{\rm ap}}
8: 
9: \documentclass{emulateapj}
10: % \documentclass[onecolumn]{emulateapj}
11: % \documentclass[12pt,preprint]{aastex}
12: % \documentclass{aastex} \usepackage{natbib}
13: \bibliographystyle{apj}%Choose a bibliograhpic style
14: \begin{document}
15: \title{Substructure and Scatter in the Mass--Temperature Relations of
16:   Simulated Clusters}
17: 
18: \author{
19:   David A. Ventimiglia\altaffilmark{1},
20:   G. Mark Voit\altaffilmark{1},
21:   Megan Donahue\altaffilmark{1},
22:   S. Ameglio\altaffilmark{2}
23: }
24: 
25: % \slugcomment{$Revision: 1.57 $}
26: 
27: \altaffiltext{1}{Michigan State University, Physics \& Astronomy
28:   Dept., East Lansing, MI 48824-2320; ventimig@msu.edu,
29:   voit@pa.msu.edu, donahue@pa.msu.edu}
30: \altaffiltext{2}{Dipartimento di Astronomia dell'Universit\`a di
31:   Trieste, via Tiepolo 11, I-34131 Trieste, Italy; ameglio@ts.astro.it}
32: 
33: \begin{abstract}
34:   Galaxy clusters exhibit regular scaling relations among their bulk
35:   properties.  These relations establish vital links between halo mass
36:   and cluster observables.  Precision cosmology studies that depend on
37:   these links benefit from a better understanding of scatter in the
38:   mass-observable scaling relations.  Here we study the role of merger
39:   processes in introducing scatter into the \mtx\ relation, using a
40:   sample of 121 galaxy clusters simulated with radiative cooling and
41:   supernova feedback, along with three statistics previously proposed
42:   to measure X-ray surface brightness substructure.  These are the
43:   centroid variation ($w$), the axial ratio ($\eta$), and the power
44:   ratios ($P_{20}$ and $P_{30}$).  We find that in this set of simulated clusters,
45:   each substructure measure is correlated with a cluster's departures
46:   $\delta \ln \tx$ and $\delta \ln M$ from the mean \mtx\ relation,
47:   both for emission-weighted temperatures $\tew$ and for
48:   spectroscopic-like temperatures $\tsl$, in the sense that clusters
49:   with more substructure tend to be cooler at a given halo mass.  In
50:   all cases, a three-parameter fit to the \mtx\ relation that includes
51:   substructure information has less scatter than a two-parameter fit
52:   to the basic \mtx\ relation.
53: \end{abstract}
54: 
55: \keywords{galaxies: clusters: general, X-rays: galaxies: clusters}
56: 
57: \section{Introduction}
58: 
59: Clusters of galaxies play a critical role in our understanding of the
60: Universe and its history and are potentially powerful tools for
61: conducting precision cosmology.  For example, large cluster surveys
62: can discriminate between cosmological models with different
63: dark-energy equations of state by providing complementary observations
64: of the shape of the cluster mass function, evolution in the number
65: density of clusters with redshift, and bias in the spatial
66: distribution of clusters
67: \citep{1998ApJ...508..483W, 2002ApJ...577..569L,
68: 2003PhRvD..67h1304H,2004ApJ...613...41M,Voit:2005PhysRevD}.
69: However, this potential to put tight constraints on the properties of
70: dark energy will be realized only if we can accurately measure the
71: masses of clusters and can precisely characterize the scatter in our
72: mass measurements.
73: 
74: Scatter in X-ray cluster properties is directly related to
75: substructure in the intracluster medium.  If clusters were all
76: structurally similar, then there would be a one-to-one relationship
77: between halo mass and any given observable property.  Generally
78: speaking, deviations from a mean mass-observable relationship are
79: attributed to structural differences among clusters.  One kind of
80: structural difference is the presence or absence of a cool core, in
81: which the central cooling time is less than the Hubble time at the
82: cluster's redshift, and the prominence of a cool core is observed to
83: be a source of scatter in scaling relationships
84: \citep{1994MNRAS.267..779F,1998ApJ...504...27M,2002ApJ...576..601V,2004ApJ...613..811M}.
85: We also expect structural differences to arise from substructure in
86: the dark matter, galaxy, and gas distributions.  For instance, there
87: may be a spread in halo concentration at a given mass, variations in
88: the incidence of gas clumps, differences in the level of AGN feedback,
89: or various effects due to mergers.  All of these deviations can be
90: considered forms of substructure that produce scatter in the
91: mass-observable relations one would like to use for cosmological
92: purposes.  While it may ultimately be possible to constrain the amount
93: of scatter and its evolution with redshift using self-calibration
94: techniques \citep{2005PhRvD..72d3006L}, such constraints would be
95: improved by prior knowledge about the relationship between scatter and
96: substructure.
97: 
98: Traditionally, the most worrisome form of substructure has been that
99: due to the effects of merger events.  Clusters are often identified as
100: ``relaxed'' or ``unrelaxed'', with the former assumed to be nearly in
101: hydrostatic equilibrium and the latter suspected of being far from
102: equilibrium.  Cosmological simulations of clusters indicate that the
103: truth is somewhere in between.  The cluster population as a whole
104: appears to follow well-defined virial relations with log-normal
105: scatter around the mean, showing that clusters do not cleanly separate
106: into relaxed and unrelaxed systems (e.g.,\cite{2007astro.ph..2241E}).
107: Even the most relaxed-looking clusters are not quite in hydrostatic
108: equilibrium (e.g.,\cite{2006ApJ...650..128K}).  Instead of simply
109: being ``relaxed'' or ``unrelaxed,'' clusters occupy a continuum of
110: relaxation levels determined by their recent mass-accretion history.
111: 
112: Quantifying this continuum of relaxation offers opportunities for
113: reducing scatter in the mass-observable relations.  If mergers are
114: indeed responsible for much of the observed scatter around a given
115: scaling relation, then there may be correlations between a cluster's
116: morphology and its degree of deviation from the mean relation.  Once
117: one identifies a morphological parameter that correlates with the
118: degree of deviation, one can construct a new mass-observable relation
119: with less scatter by including the morphological parameter in the
120: relation.  Such an approach would be analogous to the improvement of
121: Type Ia supernovae as distance indicators by using light-curve stretch
122: as a second parameter to indicate the supernova's luminosity
123: \citep{1993ApJ...413L.105P,1996ApJ...473...88R}.
124: 
125: Here we investigate how merger-related substructure in simulated
126: clusters affects the relationship between a simulated cluster's mass
127: and the temperature of its intracluster medium, building upon
128: \cite{1995ApJ...452..522B} and \cite{2006ApJ...639...64O}.
129: \cite{1995ApJ...452..522B} quantified the morphologies and dynamical
130: states of observed clusters and found structure measures to be an
131: indicator of the dynamical state of a cluster.
132: \cite{2006ApJ...639...64O} also examined morphological measurements,
133: for both observed and simulated clusters, and found that simulations
134: without cooling showed no correlation between substructure and scaling
135: relation scatter.  In this work we examine substructure for simulated
136: clusters with radiative cooling and focus on the idea that merger
137: processes introduce intrinsic scatter into the \mtx\ relationship by
138: displacing clusters in the \mtx\ plane away from the mean X-ray
139: temperature $\langle{\tx}\rangle|_{_M}$ at a given mass $M$, either to
140: higher or lower average ICM temperature.  We then adopt a set of
141: statistics \citep{1995ApJ...452..522B,2006ApJ...639...64O} for
142: quantifying galaxy cluster substructure and merger activity in order
143: to investigate this hypothesis.  Section 2 discusses the \mtx\ scaling
144: relationship in our sample of simulated clusters and shows that
145: disrupted-looking clusters in this sample tend to be cooler at a given
146: cluster mass.  In Section 3 we attempt to quantify the relationship
147: between morphology and temperature using four different substructure
148: statistics and compare it to similar studies.  We then show that
149: substructure in these simulated clusters indeed correlates with
150: scatter in the \mtx\ relationship and assess the prospects for using
151: that correlation to reduce scatter in the \mtx\ plane.  Section 4
152: summarizes our results.
153: 
154: \section{Mass-Temperature Relation in Simulated Clusters}
155: 
156: This study is based on an analysis of 121 clusters simulated using the
157: cosmological hydrodynamics TREE+SPH code GADGET-2
158: \citep{2005MNRAS.364.1105S}, which were simulated in a standard
159: $\Lambda$ cold dark matter ($\Lambda$CDM) universe with matter density
160: $\Omega_M$ = 0.3, $h$ = 0.7, $\Omega_b$ = 0.04, and $\sigma_8$ = 0.8.
161: The simulation treats radiative cooling with an optically-thin gas of
162: primordial composition, includes a time-dependent UV background from a
163: population of quasars, and handles star formation and supernova
164: feedback using a two-phase fluid model with cold star-forming clouds
165: embedded in a hot medium.  All but four of the clusters are from the
166: simulation described in \cite{2004MNRAS.348.1078B}, who simulated a
167: box $192 \, h^{-1} \, {\rm Mpc}$ on a side, with $480^3$ dark matter
168: particles and an equal number of gas particles.  The present analysis
169: considers the 117 most massive clusters within this box at $z = 0$,
170: which all have $M_{200}$ greater than $5 \times 10^{13} \, h^{-1}
171: M_\odot$.  By convention, $M_{\Delta}$ refers to the mass contained in
172: a sphere which has a mean density of $\Delta$ times the critical
173: density $\rho_c$, and whose radius is denoted by $R_{\Delta}$
174: 
175: That cluster set covers the $\sim$1.5-5 keV temperature range, but the
176: $192 \, h^{-1} \, {\rm Mpc}$ box is too small to contain significantly
177: hotter clusters.  We therefore supplemented it with four clusters with
178: masses $> 10^{15} \, h^{-1} \, M_\odot$ and temperatures $> 5$ keV
179: drawn from a dark-matter-only simulation in a larger $479 \, h^{-1} \,
180: {\rm Mpc}$ box \citep{2006MNRAS.373..397S}.  The cosmology for this
181: simulation also was $\Lambda$CDM, but with $\sigma_8$ = 0.9.  These
182: were then re-simulated including hydrodynamics, radiative cooling, and
183: star formation, again with GADGET-2 and using the
184: zoomed-initial-conditions technique of \cite{1997MNRAS.290..411T},
185: with a fourfold increase in resolution.  This is comparable to the
186: resolution of the clusters in the smaller box.  Adding these four
187: massive clusters to our sample gives a total of 121 clusters with
188: $M_{200}$ in the interval $5 \times 10^{13} \, h^{-1} M_\odot$ to $2
189: \times 10^{15} \, h^{-1} M_\odot$.
190: 
191: We first need to specify our definitions for mass and temperature.  In
192: this paper, cluster mass refers to $M_{200}$.  For temperature, we use
193: two definitions.  The first is the emission-weighted temperature
194: \begin{equation}
195:   \tew = \frac { \int T [n^2 \Lambda(T) ] \, d^3 x } { \int n^2 \Lambda(T) \, d^3 x }  \; \; ,
196: \end{equation}
197: where $n$ is the electron number density and $\Lambda(T)$ is the usual
198: cooling function for intracluster plasma.  The second is the
199: spectroscopic-like temperature of \cite{2004astro.ph..4425M},
200: \begin{equation}
201:   \tsl = \frac { \int T [n^2 T^{-3/4}]  \, d^3 x } { \int n^2 T^{-3/4} \,  d^3 x }  \; \; ,
202: \end{equation}
203: where the power-law weighting function replacing $\Lambda(T)$ is
204: chosen so that $\tsl$ approximates as closely as possible the
205: temperature that would be determined from fitting a single-temperature
206: plasma emission model to the integrated spectrum of the intracluster
207: medium.  The presence of metals in the ICM of real clusters introduces
208: line emission that complicates the computation of $\tsl$ for clusters
209: $<$3 keV \citep{2006ApJ...640..710V}.  However, the simulated spectra
210: for the clusters in our sample are modelled with zero metallicity,
211: which eases this restriction in our analysis.
212: 
213: Figure \ref{fig:mass-temp} shows the mass-temperature relations based
214: on these definitions for our sample of simulated clusters.  The best
215: fits to the power-law form
216: \begin{equation}
217:   M = M_0 \left( \frac {\tx} {3 \, {\rm keV}} \right)^\alpha
218:   \label{eq:mass-temp}
219: \end{equation}
220: have the coefficients $M_0 \simeq 5.9 \times 10^{13}\ h^{-1}
221: M_{\odot}, \alpha \simeq 1.5$ for $\tx$ corresponding to $\tsl$ and
222: $M_0 \simeq 4.4 \times 10^{13}\ h^{-1} M_{\odot}, \alpha \simeq 1.4$
223: for $\tx$ corresponding to $\tew$.  As is generally the case for
224: simulated clusters, the power-law indices of the mass-temperature
225: relations found here are consistent with cluster self-similarity and
226: the virial theorem \citep{1986MNRAS.222..323K,1995MNRAS.275..720N}.
227: These relationships have scatter, which we characterize by the
228: standard deviation in log space $\sigma_{\ln M}$ about the best-fit
229: mass at fixed temperature $\tx$.  When relating $M$ to the
230: emission-weighted temperature $\tew$, we find $\sigma_{\ln M} =
231: 0.102$.  When relating cluster mass $M$ to the spectroscopic-like
232: temperature $\tsl$, the scatter is $\sigma_{\ln M} = 0.127$.  That the
233: scatter is larger for the spectroscopic-like temperature is not
234: surprising, given the sensitivity of $\tsl$ to the thermal complexity
235: of the ICM.
236: 
237: \begin{figure}[t]
238:   \centering
239:   \includegraphics[width=0.5\textwidth]{f1.eps}
240:   \caption{Mass-temperature (\mtx) relationships for the 121 clusters
241:     in our sample.  Spectroscopic-like temperature $\tsl$ is plotted
242:     in the top panel, while emission-weighted temperature $\tew$ is
243:     plotted in the bottom panel. The solid lines show the best-fit
244:     power-law relation $M = M_0 (\tx/3\ {\rm keV})^{\alpha}$ over the
245:     whole sample, while the dashed lines show the best-fit relation
246:     for systems with $\tx>$ 2 keV.  Open circles represent the
247:     clusters that have the greatest positive temperature offset from
248:     the mean relationship, and filled circles represent the clusters
249:     with the greatest negative temperature offset.}
250:   \label{fig:mass-temp}
251: \end{figure}
252: 
253: % \begin{equation}
254: %   \delta \ln \tx = \ln \left[ \frac {\tx} {3 \, {\rm keV}} \left( \frac {M} {M_0} \right)^{-1/\alpha} \right].
255: %   \label{eq:dev-temp}
256: % \end{equation}
257: 
258: Figure \ref{fig:mass-temp} also highlights two subsamples for each
259: definition of temperature, selected based on the clusters' deviations
260: in $\ln \tx$ space from the mean mass-temperature relation.  In each
261: panel, open circles represent the eight clusters that have the largest
262: positive deviations and are therefore ``hotter'' than other clusters
263: of the same mass, while filled circles represent the eight with the
264: largest negative deviation and are ``cooler'' than other clusters of
265: the same mass.  In general, these temperature estimates are well
266: correlated, so that hotter outliers in $\tew$ are also hotter outliers
267: in $\tsl$ and cooler outliers in $\tew$ are also cooler outliers in
268: $\tsl$.  Since Figure \ref{fig:mass-temp} distinguishes the most
269: extreme outliers for the two temperature estimates, this distinction
270: may define slightly different sets, though they still overlap.
271: 
272: Figure \ref{fig:cluster_gallery_extremes} presents a gallery of
273: surface brightness maps for two sets of eight clusters with the most
274: extreme offsets from the mean $M$-$\tsl$ relation.  The eight
275: unusually hot clusters are in the top panel, and the eight cooler
276: clusters are in the bottom panel.  In these plots the hotter clusters
277: appear more symmetric, and are seemingly ``more relaxed,'' and the
278: cooler clusters appear less symmetric and seemingly ``less relaxed.''
279: The gallery as a whole therefore suggests that relaxed clusters tend
280: to be hot for their mass and unrelaxed clusters tend to be cool for
281: their mass.
282: 
283: \begin{figure}[t]
284:   \centering
285:   \includegraphics[width=0.5\textwidth,trim=0mm 40mm 0mm
286:   40mm,clip]{f2.eps}
287:   \caption{Surface-brightness contour maps for sixteen of the clusters
288:     in our sample, overlaid with a circular aperture of radius
289:     $R_{500}$.  The top panel has eight maps representing the clusters
290:     that have the largest positive deviation $\delta \ln \tx$ from the
291:     mean \mtx\ relation, calculated using the spectroscopic-like
292:     temperature $\tsl$.  These are the objects represented by open
293:     circles in the top panel of Figure \ref{fig:mass-temp}.  The
294:     bottom panel has eight maps representing the clusters that have
295:     the largest negative deviation from the same \mtx\ relationship,
296:     represented by the filled circles in Figure \ref{fig:mass-temp}.
297:     Note that the ``hotter'' clusters in the top panel appear more
298:     symmetric, while the ``cooler'' clusters in the bottom panel
299:     appear more irregular.}
300:   \label{fig:cluster_gallery_extremes}
301: \end{figure}
302: 
303: At first glance, the result that disrupted-looking clusters in
304: cosmological simulations tend to be cooler than other clusters of the
305: same mass may seem counterintuitive, since one might expect that
306: mergers ought to produce shocks that raise the mean temperature of the
307: intracluster medium.  This finding has also been noted by
308: \cite{2001ApJ...546..100M} and \cite{2006ApJ...650..128K}.  Cluster
309: systems in the process of merging tend to be cool for their total halo
310: mass because much of the kinetic energy of the merger has not yet been
311: thermalized.
312: 
313: The idealized simulations of \cite{2007MNRAS.380..437P} illustrate
314: what may happen to the ICM temperature during a single merger.  Before
315: the cores of the two merging systems collide, the mean temperature is
316: cool for the overall halo mass because it is still approximately equal
317: to the pre-merger temperature of the two individual merging halos.
318: There is a brief upward spike in temperature when the cores of the
319: merging halos collide, after which the system is again cool for its
320: mass.  Then, as the remaining kinetic energy of the merger thermalizes
321: over a period of a few billion years, the temperature gradually rises
322: to its equilibrium value.  The merging system therefore spends a
323: considerably longer time at relatively cool values of mean temperature
324: for its halo mass than at relatively hot values.  Hence, such
325: simulations suggest a possible explanation for why more relaxed
326: systems would tend to lie on the hot side of the \mtx\ relation, while
327: disrupted systems would tend to lie on the cool side.  A caveat,
328: however, is that the current generation of hydrodynamic cluster
329: simulations tend to produce relaxed clusters whose temperature
330: profiles continue to rise to smaller radii than is observed in real
331: clusters \citep{2003MNRAS.342.1025T, 2007ApJ...668....1N}, potentially
332: enhancing average temperatures for such systems.  As a separate test
333: of this effect, we excise the core regions from our sample clusters,
334: calculate new substructure measures and new emission-weighted
335: temperatures for the core-excised clusters, and repeat our analysis.
336: 
337: \section{Quantifying Substructure}
338: 
339: The question we would like to address in this study is whether the
340: surface-brightness substructure evident in Figure
341: \ref{fig:cluster_gallery_extremes} is well enough correlated with
342: deviations from the mean mass-temperature relation to yield useful
343: corrections to that relation.  In order to answer that question, we
344: need to quantify the surface-brightness substructure in each cluster
345: image, so that we can determine the degree of correlation across the
346: entire sample.  \cite{2006ApJ...639...64O} explored the relationship
347: between cluster structure and X-ray scaling relations in both observed
348: and simulated clusters, and we adopt their suite of substructure
349: measures in this study.  These include centroid variation, axial
350: ratio, and the power ratios of \cite{1995ApJ...452..522B}.  In this
351: section we define and discuss those statistics and apply them to
352: surface-brightness maps made from three orthogonal projections of each
353: cluster.  Then we assess how well these statistics correlate with
354: offsets from the mean mass-temperature relation.
355: 
356: \subsection{Axial Ratio}
357: The axial ratio $\eta$ for a cluster surface-brightness map is a
358: measure of its elongation, which is of interest because it has been
359: found from simulations that the ICM is often highly elongated during
360: merger events \citep{1993ApJ...419L...9E,1994MNRAS.268..953P}.  It is
361: computed from the second moments of the surface brightness,
362: \begin{equation}
363:   \label{eq:aratio_moments}
364:   M_{ij}=\sum I_{\rm X} x_i x_j.
365: \end{equation}
366: The summation is conducted over the coordinates $(x_1, x_2)$ of the
367: pixels that lie within an aperture centered at the origin of the
368: coordinate system to which $(x_1, x_2)$ refer.  Following the work of
369: \cite{2006ApJ...639...64O}, we use an aperture of radius $R_{500}$
370: centered on the brightness peak.  We then compute $\eta$ from the
371: ratio of the non-zero elements that result from diagonalizing the
372: matrix $M$.  That is,
373: \begin{equation}
374:   \label{eq:diagonalize}
375:   D = U^T M U,
376: \end{equation}
377: where $U$ is a diagonalizing matrix for $M$, and  
378: \begin{equation}
379:   \label{eq:aratio}
380:   \eta = 
381:   \left\{ 
382:     \begin{array}{ll}
383:       \frac{D_{12}}{D_{21}}, & D_{12} \le D_{21} \\
384:       \frac{D_{21}}{D_{12}}, & D_{12} > D_{21}
385:     \end{array}
386:   \right\} \; \; .
387: \end{equation}
388: The axial ratio is therefore defined to be in the range $\eta \in [0,
389: 1]$, with $\eta = 1$ for a circular cluster.  Of course there are
390: other choices for the origin of the coordinate system, besides using
391: the brightness peak.  For instance, in order to avoid misplaced
392: apertures yielding artificially low axial ratios for nearly circular
393: distributions, one could adjust the position of the aperture to seek a
394: maximum in $\eta$.  Doing this, we sometimes find that $\eta \approx
395: 1$ even for non-circular clusters, as is evident in Figure
396: \ref{fig:badaxialratio}.  This figure depicts the surface-brightness
397: map of what appears to be a disturbed cluster, chosen from among those
398: in our sample that appear by eye to be the most unrelaxed. Yet, it
399: happens to have an axial ratio very close to 1 for an aperture placed
400: so as to maximize $\eta$.  This example demonstrates that, while the
401: axial ratio statistic may yield results consistent with a visual
402: interpretation of cluster substructure, it is also capable of
403: unexpected results for some clusters.
404: 
405: \begin{figure}[t]
406:   \centering
407:   \includegraphics[width=0.4\textwidth,trim=20mm 10mm 5mm 10mm,
408:   clip]{f3.eps}
409:   \caption{Surface-brightness contour plot of an asymmetric cluster
410:     which, for certain choices of aperture placement, yields an axial
411:     ratio close to 1.  The circle represents an aperture of
412:     $R_{500}$}.
413:   \label{fig:badaxialratio}
414: \end{figure}
415: 
416: \begin{figure}[t]
417:   \centering
418:   \includegraphics[width=0.5\textwidth,trim=0mm 10mm 0mm 10mm]{f4.eps}
419:   \caption{Surface of axial ratio $\eta$ as a two-dimensional function
420:     of the coordinates of the aperture center.  The axial ratio
421:     statistic appears to be ill-defined for this cluster.}
422:   \label{fig:badaxialratiosurface}
423: \end{figure}
424: 
425: \begin{figure}[t]
426:   \centering
427:   \includegraphics[width=0.5\textwidth,trim=0mm 10mm 0mm 10mm]{f5.eps}
428:   \caption{Abstract surface of axial ratio $\eta$ as a two-dimensional
429:     function of the coordinates of the aperture center.  This is a
430:     relaxed cluster, for which the axial ratio is better-defined.}
431:   \label{fig:goodaxialratiosurface}
432: \end{figure}
433: 
434: \begin{figure*}[t]
435:   \centering
436:   \includegraphics[width=1.0\textwidth,trim=5mm 0mm 0mm
437:   0mm,clip]{f6.eps}
438:   \caption{Relationship between a cluster's deviation $\delta \ln \tx$
439:     from the mean \mtx\ relationship and four measures of
440:     substructure: the centroid variation $w$, the axial ratio $\eta$,
441:     and the power ratios $P_{20}$ and $P_{30}$.  The lower panels are
442:     for spectroscopic-like temperature, and the upper ones are for
443:     emission-weighted temperature.  The light gray bands and vertical
444:     dashed lines show the extremes in substructure measurements, with
445:     80\% of our clusters falling in the central region between the
446:     bands.  The black filled circles correspond to clusters above 2
447:     keV, while the plus signs correspond to clusters below 2 keV.
448:     Finally, the solid lines indicate the best-fitting linear
449:     relationships between our substructure measures and temperature
450:     offset, for the above 2 keV sub-sample denoted by the black filled
451:     circles.}
452:   \label{fig:substruct_dev_temp}
453: \end{figure*}
454: 
455: To further illustrate this point, we have computed an axial ratio
456: value for this cluster for every possible choice of aperture
457: placement.  Apertures of radius $R_{500}$ were centered on each and
458: every pixel within the surface-brightness map, provided the aperture
459: so placed does not reach the edge of the map.  This procedure
460: generated an axial-ratio ``surface'' mapping all of the aperture
461: placements to a value of $\eta$.  Figure
462: \ref{fig:badaxialratiosurface} shows the axial-ratio surface for the
463: cluster in Figure \ref{fig:badaxialratio}.  For comparison purposes,
464: Figure \ref{fig:goodaxialratiosurface} presents an axial-ratio surface
465: map for a very symmetric, uniform, and apparently relaxed cluster, in
466: which the cluster's brightness peak reassuringly corresponds to the
467: aperture location that maximizes $\eta$.  In contrast, the presence of
468: two peaks in the axial-ratio surface for the asymmetric cluster shows
469: that $\eta$ can sometimes depend strongly on aperture placement.
470: Ideally, we would like to place the aperture on the ``center'' of this
471: cluster, but the center of an unrelaxed cluster can be difficult to
472: define, meaning that the axial ratio statistic may be likewise
473: ill-defined for such clusters.
474: 
475: \subsection{Power Ratio}
476: The power-ratio statistics
477: \citep{1995ApJ...452..522B,2006ApJ...639...64O} quantify substructure
478: by decomposing the surface-brightness image into a two-dimensional
479: multipole expansion, the terms of which are calculated from the
480: moments of the image, computed within an aperture of radius $\Rap$:
481: \begin{equation}
482:   \label{eq:a_m}
483:   a_m(\Rap) = \int_{R' \leq \Rap}\Sigma(\vec x')(R')^m \cos m \phi'\ d^2x'
484: \end{equation}
485: \begin{equation}
486:   \label{eq:b_m}
487:   b_m(\Rap) = \int_{R' \leq \Rap}\Sigma(\vec x')(R')^m \sin m \phi'\ d^2x'.
488: \end{equation}
489: The power in terms of order $m$ is then
490: \begin{equation}
491:   \label{eq:pm}
492:   P_m = \frac{(a_m^2 + b_m^2)}{2m^2\Rap^{2m}}.
493: \end{equation}
494: For $m = 0$, the power is given by
495: \begin{equation}
496:   \label{eq:p0}
497:   P_0 = [a_0 \ln(\Rap)]^2.
498: \end{equation}
499: The power ratios $P_{m0} \equiv P_m / P_0$ are then dimensionless
500: measures of substructure which have differing interpretations.  For
501: instance, $P_{10}$ quantifies the degree of balance about some origin
502: and can be used to find the image centroid, $P_{20}$ is related to the
503: ellipticity of the image, and $P_{30}$ is related to the triangularity
504: of the photon distribution.  As in the case of the axial ratio
505: computations, we set the aperture radius $\Rap$ equal to $R_{500}$.
506: The most appropriate place to center the aperture is at the set of
507: pixel coordinates that minimizes $P_{10}$, which we achieve using a
508: self-annealing algorithm.
509: 
510: \subsection{Centroid Variation}
511: The centroid variation statistic $w$ is a measure of the center shift,
512: or ``skewness'', of a two-dimensional photon distribution.  It is
513: measured for a cluster surface-brightness map in the following way.
514: For a set of surface-brightness levels one finds the centroids of the
515: corresponding isophotal contours and computes the variance in the
516: coordinates of those centroids, scaled to $R_{500}$.  Here we select
517: 10 isophotes evenly spaced in $\log I_{\rm X}$ between the minimum and
518: maximum of $I_{\rm X}$ within an aperture of radius $R_{500}$ centered
519: on the brightness peak, so as to adapt to the full dynamic range of
520: surface brightness for different clusters.  We employed this adaptive
521: scheme because using one set of isophotes for all clusters tended to
522: ignore important substructure in less massive clusters when they had
523: surface brightness substructure inside $R_{500}$ but outside of the
524: lowest isophote.
525: 
526: \subsection{Substructure and Scaling Relationships}
527: 
528: Using the quantitative measures of substructure described in the
529: previous section, we can test the significance of the relationship
530: between substructure and temperature offset hinted at in Figure
531: \ref{fig:cluster_gallery_extremes}.  We begin by treating four of our
532: substructure statistics---centroid variation $w$, axial ratio $\eta$,
533: and power ratios $P_{20}$ and $P_{30}$---as different imperfect
534: measurements of an intrinsic degree of substructure $S$.  Figure
535: \ref{fig:substruct_dev_temp} shows the relationship between
536: substructure and a cluster's deviation $\delta \ln \tx$ from the mean
537: \mtx\ relation for each substructure measure.  In each case we present
538: results for both the spectroscopic-like temperature $\tsl$ and the
539: emission-weighted temperature $\tew$.  Note that centroid variation
540: $w$ and the power ratios $P_{20}$ and $P_{30}$ have large dynamic
541: ranges, whereas the axial ratio $\eta$ is always of order unity.  We
542: therefore attempt to fit the relationships between $\delta \ln \tx$
543: and the different substructure measures with the following forms:
544: \begin{equation}
545:   \delta \ln \tx = \left\{
546:   \begin{array}{l}
547:     A w^{\alpha} \\
548:     B + \beta\eta \\
549:     C P_{20}^{\gamma} \\
550:     D P_{30}^{\lambda}
551:    \end{array} \right.
552: \end{equation}
553: To visually indicate where the bulk of our substructure measures lie,
554: Figure \ref{fig:substruct_dev_temp} has light gray bands covering the
555: extremes, so that 80\% of our sample clusters have substructure
556: measures lying between the extremes.  The power ratios in our study
557: generally span two decades (in units of $10^{-7}$), from $\sim$2---300
558: for $P_{20}$ and from $\sim$0.01---10 for $P_{30}$.  These ranges are
559: consistent with those of \cite{1995ApJ...452..522B},
560: \cite{2006ApJ...639...64O}, and \cite{2007arXiv0708.1518J}.  The
561: measurements of axial ratio in our sample, with 80\% of clusters
562: having $\eta$$\sim$0.4---0.95, cover a slightly wider range than do
563: the simulated clusters of \cite{2006ApJ...639...64O}.  Finally, our
564: measurements of centroid variation, with 80\% of clusters having
565: $w[R_{500}]$$\sim$0.01---0.1, are again similar to those of
566: \cite{2006ApJ...639...64O}.
567: 
568: As denoted in Figure \ref{fig:substruct_dev_temp} by black filled
569: circles, the systems with $\tx$ above 2 keV occupy a slightly narrower
570: range of substructure values than the systems below 2 keV, which are
571: denoted by plus signs.  For the axial ratio and the power ratios, the
572: variance is 15 to 25 percent larger among the low-temperature systems
573: when compared to the systems with $\tx >$ 2 keV.  For centroid
574: variation the variance among the low-temperature systems is
575: approximately the same as it is among the high-temperature systems.
576: However, it is not clear that there is a significant correlation
577: between substructure and mass, since the mean substructure values are
578: generally very similar between the low-temperature and
579: high-temperature subsamples.  The mean value of the power ratio
580: $P_{30}$ is significantly larger for the low-temperature subsample,
581: however this measure also has the weakest correlation with offsets
582: from the mean \mtx\ relation.
583: 
584: To test whether the low-mass clusters in our sample significantly
585: boost the overall scatter in the \mtx\ relation, we perform a cut at 2
586: keV and fit this relation both to the whole sample and to the
587: sub-sample above 2 keV.  Figure \ref{fig:resid_lowt} shows the
588: residuals in mass, actual minus predicted, where the predicted mass
589: derives only from the \mtx\ relation.  The plus signs indicate
590: clusters whose mass is predicted from an \mtx\ relation derived from
591: all 121 clusters.  The black filled circles indicate clusters that are
592: above 2 keV in X-ray temperature, with the mass estimated using the
593: sub-sample \mtx\ relation.  There is a negligible reduction in
594: scatter, from 0.127 to 0.124 for $\tsl$ and from 0.102 to 0.094 for
595: $\tew$, suggesting that at best only a modest improvement is found in
596: our sample if we remove the low-mass systems.  In order to test the
597: degree to which incorporating substructure measures adds to this
598: modest improvement, when we compare mass estimates derived using
599: substructure to those derived only from the \mtx\ relation, we focus
600: on clusters above 2 keV in the rest of our analysis.
601: 
602: Figure \ref{fig:substruct_dev_temp} shows that for our simulated
603: clusters, a greater amount of measured substructure tends to be
604: associated with ``cooler'' clusters while less substructure tends to
605: be associated with ``hotter'' clusters.  Also, the centroid variations
606: $w$ are more highly-correlated with $\delta \ln \tx$ than are the
607: other substructure parameters.  We interpret this to mean that the
608: centroid variation is a better predictor of the offset in the \mtx\
609: relationship than are the power ratios and the axial ratio, though all
610: four measures appear to be related to the temperature offset.  Again,
611: in this figure we denote systems above 2 keV by black filled circles,
612: and systems below 2 keV by plus signs.
613: 
614: \begin{figure}[t]
615:   \centering
616:   \includegraphics[width=0.5\textwidth,trim=5mm 0mm 0mm
617:   0mm,clip]{f7.eps}
618:   \caption{Comparison of mass offset $\delta \ln M$ between true mass
619:     and predicted mass, based on the $M(\tx)$ relation.  Plus signs
620:     indicate residuals for masses estimated from the $M$-$T_{\rm X}$
621:     relation derived from all 121 clusters, while black filled circles
622:     are for masses estimated from the $M$-$T_{\rm X}$ relation for
623:     clusters above 2 keV.  Upper panels are for spectroscopic-like
624:     temperature and lower panels are for emission-weighted
625:     temperature.  The standard deviations $\sigma$ in the residuals
626:     are given in each plot.}
627:   \label{fig:resid_lowt}
628: \end{figure}
629: 
630: \begin{figure*}[t]
631:   \centering
632:   \includegraphics[width=1.0\textwidth,trim=5mm 0mm 0mm
633:   0mm,clip]{f8.eps}
634:   \caption{Relationship between substructure and mass offset $\delta
635:     \ln M (\tx)$ from the mean \mtx\ relationship for the same
636:     substructure measures as in Figure \ref{fig:substruct_dev_temp}.
637:     The lower panels are for $\tx = \tsl$, and the upper panels are
638:     for $\tx = \tew$.  As in Figure \ref{fig:substruct_dev_temp} the
639:     solid lines indicate the best-fitting linear relationships to the
640:     above 2 keV sub-sample denoted by black filled circles, and the
641:     gray bands and vertical dashed lines mark the extremes in
642:     substructure between which 80\% of our clusters lie.  Also as in
643:     Figure \ref{fig:substruct_dev_temp}, the plus signs correspond to
644:     systems below 2 keV., while the black filled circles correspond to
645:     systems above 2 keV.}
646:   \label{fig:substruct_dev}
647: \end{figure*}
648: 
649: Correlations between substructure and $\delta \ln \tx$ can potentially
650: be exploited to improve on mass estimates of real clusters derived
651: from the mass-temperature relation.  Instead of computing the
652: temperature offset at fixed mass, we can determine a
653: substructure-dependent mass offset at fixed temperature and then apply
654: it as a correction to the predicted mass $M_{\rm pred}(\tx)$ one would
655: derive from the mean \mtx\ relation alone.  To assess the prospects
656: for such a correction, based on this sample of simulated clusters, we
657: first define the mass offset from the mean mass-temperature relation
658: to be
659: \begin{equation}
660:   \delta \ln M (\tx) = \ln \left[ \frac {M} {M_{\rm pred}(\tx)} \right] \; \; ,
661:   \label{eq:deltaM}
662: \end{equation}
663: where $M$ is the cluster's actual mass, and examine the correlations
664: between substructure measures and $\delta \ln M$.  Figure
665: \ref{fig:substruct_dev} shows the results.  These plots show mass
666: predictions from both the $M$-$\tsl$ relation and the $M$-$\tew$
667: relation.  Consistent with our analysis of $\delta \ln \tx$, the
668: centroid variation $w$ appears to be a more effective predictor of the
669: mass offset $\delta \ln M (\tx)$.  Nonetheless, all four measures of
670: substructure appear to be correlated with mass offset.
671: 
672: \begin{figure*}[t]
673:   \centering
674:   \includegraphics[width=1.0\textwidth,trim=5mm 0mm 0mm
675:   0mm,clip]{f9.eps}
676:   \caption{Comparison of mass offset $\delta \ln M$ between true mass
677:     and predicted mass, based on the $M(\tx)$ relation (open circles)
678:     and the $M(\tx,S)$ relation (filled circles).  Upper panels are
679:     for $\tsl$ and lower panels are for $\tew$.  The standard
680:     deviations $\sigma$ in the residuals are given in each plot.}
681:   \label{fig:resid2}
682: \end{figure*}
683: 
684: In order to incorporate a substructure correction into the
685: mass-temperature relation, we perform a multiple regression, fitting
686: our simulated clusters' mass, temperature, and substructure data to
687: the form,
688: \begin{equation}
689:   \log M_{\rm pred} (\tx,S) = \log M_0 + \alpha \log \tx + \beta S ,
690:   \label{eq:mass-temp-subst}
691: \end{equation}
692: where $S$ represents one of the following substructure measures:
693: $\eta$, $\log w$, $\log P_{20}$, or $\log P_{30}$.  This fit gives us
694: a substructure-corrected mass prediction $M_{\rm pred} (\tx, S)$ for
695: each substructure measure, and we can assess the effectiveness of that
696: correction by measuring the dispersion of the substructure-corrected
697: mass offset
698: \begin{equation}
699:   \delta \ln M (\tx,S) = \ln \left[ \frac {M} {M_{\rm pred}(\tx,S)} \right] \; \; ,
700:   \label{eq:deltaMTS}
701: \end{equation}
702: between the revised prediction and the true cluster mass.
703: 
704: Figure \ref{fig:resid2} shows the results of that test.  Open circles
705: in each panel indicate mass offsets $\delta \ln M(\tx)$ without
706: substructure corrections, which have a standard deviation
707: $\sigma_{M(T)}$.  Filled circles indicate mass offsets $\delta \ln
708: M(\tx,S)$ with substructure corrections, which have a standard
709: deviation $\sigma_{M(T,S)}$.  The upper set of panels shows results for
710: $\tsl$, and the lower set is for $\tew$.  In each case, incorporating
711: a substructure correction to the mass-temperature relation reduces the
712: scatter, yielding more accurate mass estimates.  The centroid
713: variation corrections are the most effective, reducing the scatter in
714: mass from 0.124 to 0.085 in the $M$-$\tsl$ relation and from 0.094 to
715: 0.072 in the $M$-$\tew$ relation, though admittedly this is again a
716: modest improvement.  Although non-negligible structure correlates
717: significantly with offsets in the \mtx\ plane, apparently it does so
718: with substantial scatter.  This scatter may be partly due to
719: projection effects, in which line-of-sight mergers are discounted by
720: the measures of substructure and may dilute their corrective power
721: \citep{2007arXiv0708.1518J}.
722: 
723: Lastly, Figure \ref{fig:resid3} shows the results for a similar
724: analysis to that of Figure \ref{fig:resid2}, except that in this case
725: we have excised a region of radius 0.15$R_{500}$ around the center of
726: each cluster and recomputed $\tew$.  We do this to test whether offset
727: in temperature, whose correlation with substructure is the basis of
728: our correction scheme, stems from a potentially unrealistic feature,
729: which is that the cores of many real clusters have temperature
730: profiles that decline at larger radii than occurs in simulated
731: clusters.  As in Figure \ref{fig:resid2}, we restrict our analysis to
732: clusters above 2 keV.  After doing this test, for $\tew$ excising the
733: core actually increases the scatter in \mtx\ from 0.094 to 0.106.  It
734: may be that by removing the bright central region, the average
735: temperature becomes more sensitive to structure outside the core.
736: Also, this figure shows that the effect of incorporating substructure
737: measurements into the mass-estimates is still present.  The scatter is
738: reduced to 0.075 for $w$, 0.093 for $\eta$, 0.090 for $P_{20}$ and
739: 0.094 for $P_{30}$.  Figure \ref{fig:resid3} summarizes the results of
740: this test, which support the conclusion that the reduction in scatter
741: we realize using substructure is a real effect and not an artifact of
742: known defects in the simulations.
743: 
744: \subsection{Comparisons with Other Substructure Studies}
745: 
746: \cite{2006ApJ...639...64O} examined the relationship between galaxy
747: cluster substructure and X-ray scaling relationships, including the
748: \mtx\ relation, using both a flux-limited sample of nearby clusters
749: and a sample of simulated clusters, and found a greater amount scatter
750: among the more relaxed clusters in their observed sample.  Contrasting
751: that result they also found a greater amount of scatter among the more
752: disrupted clusters in their simulation sample, though they
753: characterize the evidence for this second result to be weak.  Finally,
754: they see no evidence in either sample for more disrupted clusters to
755: be below the mean, and the more relaxed clusters to be above.  One
756: difference between our study and theirs is the presence of radiative
757: cooling and supernova feedback in the simulation that produced our
758: cluster sample.  Also, the focus of our work is different from theirs
759: in that we concentrate on the degree of correlation between the amount
760: of substructure and the size and direction of the offset from the mean
761: relation.  We do find significant evidence of this correlation, such
762: that relaxed clusters are hotter than expected given their mass.  We
763: also test, as best we can given our simulation sample, the hypothesis
764: that substructure can be used to improve mass estimates derived from
765: the ICM X-ray temperature.  It is possible that our detection of a
766: correlation between substructure and temperature offset arises from
767: the additional physics in our simulated clusters, since when radiative
768: cooling is included, cool lumps may be better preserved than in
769: simulations that don't include cooling.
770: 
771: \begin{figure*}[t]
772:   \centering
773:   \includegraphics[width=1.0\textwidth,trim=5mm 30mm 0mm
774:   0mm,clip]{f10.eps}
775:   \caption{Comparison of mass offset $\delta \ln M$ between true mass
776:     and predicted mass, based on the $M(\tx)$ relation (open circles)
777:     and the $M(\tx,S)$ relation (filled circles).  These results are
778:     for are for $\tew$, with the central 0.15 $R_{500}$ region removed
779:     both from the average temperature and from the substructure
780:     measures.  The standard deviations $\sigma$ in the residuals are
781:     given in each plot.}
782:   \label{fig:resid3}
783: \end{figure*}
784: 
785: Our results are in agreement with \cite{2006NewA...12...71V}, who
786: examined substructure in clusters simulated with cooling and feedback
787: and found that unrelaxed clusters, identified with a larger power
788: ratio $P_{30}$, have spectral-fit temperatures biased low relative to
789: the mass-weighted temperatures.  This trend aligns with our finding
790: that the spectroscopic-like temperature $\tsl$ is lower than the
791: best-fit temperature at fixed mass for clusters with larger power
792: ratios $P_{20}$ and $P_{30}$.  However, \cite{2006NewA...12...71V} did
793: not investigate the effectiveness of substructure measures in reducing
794: scatter in the mass-temperature relation.
795: 
796: Our results are also in agreement with some of the results of
797: \cite{2007arXiv0708.1518J}, who have recently investigated
798: correlations between substructure and offsets in mass predictions in
799: simulated clusters.  They found that measuring cluster structure is an
800: effective way to correct masses estimated using the assumption of
801: hydrostatic equilibrium, which tend to be underestimates.  Our
802: findings support these results, given that we find substructure can be
803: used to correct masses estimated directly from the \mtx\ relationship.
804: There also are differences between our findings and theirs.  They
805: report that the \mtx\ relation for their simulation sample shows no
806: dependence on structure, whereas the clusters in our sample exhibit
807: offsets that correlate with the degree of substructure.  One
808: possibility is that these differences stem from differences in the
809: simulations' feedback mechanisms.  Another possibility is that some of
810: the offset we observe derives from enhanced temperatures in
811: simulations with radiative cooling.  As we describe in section 3, we
812: perform a test in which we estimate $\tew$\ using projected
813: surface-brightness and temperature maps, in order to remove the core
814: regions from our analysis, but this may be less effective than
815: properly excising the cores in the simulations, as
816: \cite{2007arXiv0708.1518J} have done.
817: 
818: \cite{2006ApJ...650..128K} also looked at the relationship that
819: cluster structure has to the \mtx\ relation in simulated clusters, to
820: show that the sensitivity of mass proxies $Y_X$ and $Y_{SZ}$ to
821: substructure is not very strong.  They divided their sample into
822: unrelaxed and relaxed subsamples, based on the presence or absence of
823: multiple peaks in the surface-brightness maps of clusters, and found
824: the normalization of the \mtx\ relation to be biased to cooler
825: temperatures for the unrelaxed systems.  Other workers also have
826: looked at the relationship between the \mtx\ relation and
827: substructure, as reflected in the X-ray spectral properties.
828: \cite{2001ApJ...546..100M} have examined the ratio of X-ray
829: spectral-fit temperatures in hard and full bandpasses for an ensemble
830: of simulated clusters, and found it to be a way of quantifying the
831: dynamical state of a cluster.  We consider our approach of using
832: surface-brightness morphology information to be complementary to
833: theirs.  More recently, \cite{2007MNRAS.377..317K} performed an
834: interesting analysis on another large-volume simulation sample, using
835: as substructure metrics the centroid variation and measures of
836: concentration to report evolution in the luminosity-temperature
837: relationship.  Specifically, they report that the more irregular
838: clusters in their sample lie above the mean \mtx\ relation (i.e., they
839: are cooler than average), for the spectroscopic-like temperature
840: $\tsl$.
841: 
842: \section{Summary}
843: 
844: Using a sample of galaxy clusters simulated with cooling and feedback,
845: we investigated three substructure statistics and their correlations
846: with temperature and mass offsets from mean scaling relations in the
847: \mtx\ plane.  First, we showed that the substructure statistics $w$,
848: $\eta$, $P_{20}$ and $P_{30}$ all correlate significantly with $\delta
849: \ln \tx$, though with non-negligible scatter.  In all cases this
850: scatter is larger for $\delta \ln$ $\tsl$ than it is for $\delta \ln$
851: $\tew$.  Next, we considered the possibility that \mtx\ scatter is
852: driven by low-mass clusters.  We tested the degree to which scatter
853: can be reduced by filtering out these systems.  This consisted of
854: performing a cut at 2 keV, for which we saw that it yielded a modest
855: improvement in mass estimates.  To see whether incorporating
856: substructure could refine these mass estimates, we first showed that
857: $w$, $\eta$, $P_{20}$, and $P_{30}$ correlate significantly with the
858: difference $\delta \ln M$ between masses predicted from the mean
859: $M(\tx)$ relation and the true cluster masses, with non-negligible
860: scatter that again is less for $M(\tew)$ than it is for $M(\tsl)$.
861: Then we adopted a full three-parameter model, $M$-$\tx$-$S$, which
862: includes substructure information $S$ estimated using $w$, $\eta$,
863: $P_{20}$, and $P_{30}$. Scatter about the basic two-parameter
864: $M$-$\tew$ relation was 0.094.  Including substructure as a third
865: parameter reduced the scatter to 0.072 for centroid variation, 0.084
866: for axial ratio, 0.081 for $P_{20}$, and 0.084 for $P_{30}$.  Scatter
867: about the basic two-parameter $M$-$\tsl$ relation was 0.124, and
868: including substructure as a third parameter reduced the scatter to
869: 0.085 for centroid variation, 0.112 for axial ratio, 0.110 for
870: $P_{20}$, and 0.108 for $P_{30}$.  As one last test, and to increase
871: our confidence that our substructure measures are not relying on
872: potentially non-physical core structure in the simulations, we also
873: repeated the comparison of mass-estimates for $\tew$, with the core
874: regions of the clusters excised.  First, removing the core slightly
875: increased the scatter in \mtx\, possibly by making the average
876: temperature more sensitive to structure outside the core.  Second,
877: even with the cores removed the improvement in mass-estimates obtained
878: using substructure information remains.  Based on these results, it
879: appears that centroid variation is the best substructure statistic to
880: use when including a substructure correction in the $M$-$\tew$
881: relation.  However, the correlations we have found in this sample of
882: simulated clusters might not hold in samples of real clusters, because
883: relaxed clusters in the real universe tend to have cooler cores than
884: our simulated clusters do.
885: 
886: \acknowledgments The authors wish to thank Stefano Borgani for
887: contributing the simulation data on which this project was based and
888: for his helpful comments on the manuscript.  This work was supported
889: by NASA through grants NNG04GI89G and NNG05GD82G, through Chandra
890: theory grant TM8-9010X, and through Chandra archive grant
891: SAOAR5-6016X.
892: 
893: \begin{thebibliography}{33}
894: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
895: 
896: \bibitem[{{Borgani} {et~al.}(2004){Borgani}, {Murante}, {Springel}, {Diaferio},
897:   {Dolag}, {Moscardini}, {Tormen}, {Tornatore}, \&
898:   {Tozzi}}]{2004MNRAS.348.1078B}
899: {Borgani}, S., {Murante}, G., {Springel}, V., {Diaferio}, A., {Dolag}, K.,
900:   {Moscardini}, L., {Tormen}, G., {Tornatore}, L., \& {Tozzi}, P. 2004, \mnras,
901:   348, 1078
902: 
903: \bibitem[{{Buote} \& {Tsai}(1995)}]{1995ApJ...452..522B}
904: {Buote}, D.~A., \& {Tsai}, J.~C. 1995, \apj, 452, 522
905: 
906: \bibitem[{{Evrard} {et~al.}(2007){Evrard}, {Bialek}, {Busha}, {White}, {Habib},
907:   {Heitmann}, {Warren}, {Rasia}, {Tormen}, {Moscardini}, {Power}, {Jenkins},
908:   {Gao}, {Frenk}, {Springel}, {White}, \& {Diemand}}]{2007astro.ph..2241E}
909: {Evrard}, A.~E., {Bialek}, J., {Busha}, M., {White}, M., {Habib}, S.,
910:   {Heitmann}, K., {Warren}, M., {Rasia}, E., {Tormen}, G., {Moscardini}, L.,
911:   {Power}, C., {Jenkins}, A.~R., {Gao}, L., {Frenk}, C.~S., {Springel}, V.,
912:   {White}, S.~D.~M., \& {Diemand}, J. 2007, ArXiv Astrophysics e-prints
913: 
914: \bibitem[{{Evrard} {et~al.}(1993){Evrard}, {Mohr}, {Fabricant}, \&
915:   {Geller}}]{1993ApJ...419L...9E}
916: {Evrard}, A.~E., {Mohr}, J.~J., {Fabricant}, D.~G., \& {Geller}, M.~J. 1993,
917:   \apjl, 419, L9+
918: 
919: \bibitem[{{Fabian} {et~al.}(1994){Fabian}, {Crawford}, {Edge}, \&
920:   {Mushotzky}}]{1994MNRAS.267..779F}
921: {Fabian}, A.~C., {Crawford}, C.~S., {Edge}, A.~C., \& {Mushotzky}, R.~F. 1994,
922:   \mnras, 267, 779
923: 
924: \bibitem[{{Hu}(2003)}]{2003PhRvD..67h1304H}
925: {Hu}, W. 2003, \prd, 67, 081304
926: 
927: \bibitem[{{Jeltema} {et~al.}(2007){Jeltema}, {Hallman}, {Burns}, \&
928:   {Motl}}]{2007arXiv0708.1518J}
929: {Jeltema}, T.~E., {Hallman}, E.~J., {Burns}, J.~O., \& {Motl}, P.~M. 2007,
930:   ArXiv e-prints, 708
931: 
932: \bibitem[{{Kaiser}(1986)}]{1986MNRAS.222..323K}
933: {Kaiser}, N. 1986, \mnras, 222, 323
934: 
935: \bibitem[{{Kay} {et~al.}(2007){Kay}, {da Silva}, {Aghanim}, {Blanchard},
936:   {Liddle}, {Puget}, {Sadat}, \& {Thomas}}]{2007MNRAS.377..317K}
937: {Kay}, S.~T., {da Silva}, A.~C., {Aghanim}, N., {Blanchard}, A., {Liddle},
938:   A.~R., {Puget}, J.-L., {Sadat}, R., \& {Thomas}, P.~A. 2007, \mnras, 377, 317
939: 
940: \bibitem[{{Kravtsov} {et~al.}(2006){Kravtsov}, {Vikhlinin}, \&
941:   {Nagai}}]{2006ApJ...650..128K}
942: {Kravtsov}, A.~V., {Vikhlinin}, A., \& {Nagai}, D. 2006, \apj, 650, 128
943: 
944: \bibitem[{{Levine} {et~al.}(2002){Levine}, {Schulz}, \&
945:   {White}}]{2002ApJ...577..569L}
946: {Levine}, E.~S., {Schulz}, A.~E., \& {White}, M. 2002, \apj, 577, 569
947: 
948: \bibitem[{{Lima} \& {Hu}(2005)}]{2005PhRvD..72d3006L}
949: {Lima}, M., \& {Hu}, W. 2005, \prd, 72, 043006
950: 
951: \bibitem[{{Majumdar} \& {Mohr}(2004)}]{2004ApJ...613...41M}
952: {Majumdar}, S., \& {Mohr}, J.~J. 2004, \apj, 613, 41
953: 
954: \bibitem[{{Markevitch}(1998)}]{1998ApJ...504...27M}
955: {Markevitch}, M. 1998, \apj, 504, 27
956: 
957: \bibitem[{{Mathiesen} \& {Evrard}(2001)}]{2001ApJ...546..100M}
958: {Mathiesen}, B.~F., \& {Evrard}, A.~E. 2001, \apj, 546, 100
959: 
960: \bibitem[{{Mazzotta} {et~al.}(2004){Mazzotta}, {Rasia}, {Moscardini}, \&
961:   {Tormen}}]{2004astro.ph..4425M}
962: {Mazzotta}, P., {Rasia}, E., {Moscardini}, L., \& {Tormen}, G. 2004, ArXiv
963:   Astrophysics e-prints
964: 
965: \bibitem[{{McCarthy} {et~al.}(2004){McCarthy}, {Balogh}, {Babul}, {Poole}, \&
966:   {Horner}}]{2004ApJ...613..811M}
967: {McCarthy}, I.~G., {Balogh}, M.~L., {Babul}, A., {Poole}, G.~B., \& {Horner},
968:   D.~J. 2004, \apj, 613, 811
969: 
970: \bibitem[{{Nagai} {et~al.}(2007){Nagai}, {Kravtsov}, \&
971:   {Vikhlinin}}]{2007ApJ...668....1N}
972: {Nagai}, D., {Kravtsov}, A.~V., \& {Vikhlinin}, A. 2007, \apj, 668, 1
973: 
974: \bibitem[{{Navarro} {et~al.}(1995){Navarro}, {Frenk}, \&
975:   {White}}]{1995MNRAS.275..720N}
976: {Navarro}, J.~F., {Frenk}, C.~S., \& {White}, S.~D.~M. 1995, \mnras, 275, 720
977: 
978: \bibitem[{{O'Hara} {et~al.}(2006){O'Hara}, {Mohr}, {Bialek}, \&
979:   {Evrard}}]{2006ApJ...639...64O}
980: {O'Hara}, T.~B., {Mohr}, J.~J., {Bialek}, J.~J., \& {Evrard}, A.~E. 2006, \apj,
981:   639, 64
982: 
983: \bibitem[{{Pearce} {et~al.}(1994){Pearce}, {Thomas}, \&
984:   {Couchman}}]{1994MNRAS.268..953P}
985: {Pearce}, F.~R., {Thomas}, P.~A., \& {Couchman}, H.~M.~P. 1994, \mnras, 268,
986:   953
987: 
988: \bibitem[{{Phillips}(1993)}]{1993ApJ...413L.105P}
989: {Phillips}, M.~M. 1993, \apjl, 413, L105
990: 
991: \bibitem[{{Poole} {et~al.}(2007){Poole}, {Babul}, {McCarthy}, {Fardal},
992:   {Bildfell}, {Quinn}, \& {Mahdavi}}]{2007MNRAS.380..437P}
993: {Poole}, G.~B., {Babul}, A., {McCarthy}, I.~G., {Fardal}, M.~A., {Bildfell},
994:   C.~J., {Quinn}, T., \& {Mahdavi}, A. 2007, \mnras, 380, 437
995: 
996: \bibitem[{{Riess} {et~al.}(1996){Riess}, {Press}, \&
997:   {Kirshner}}]{1996ApJ...473...88R}
998: {Riess}, A.~G., {Press}, W.~H., \& {Kirshner}, R.~P. 1996, \apj, 473, 88
999: 
1000: \bibitem[{{Saro} {et~al.}(2006){Saro}, {Borgani}, {Tornatore}, {Dolag},
1001:   {Murante}, {Biviano}, {Calura}, \& {Charlot}}]{2006MNRAS.373..397S}
1002: {Saro}, A., {Borgani}, S., {Tornatore}, L., {Dolag}, K., {Murante}, G.,
1003:   {Biviano}, A., {Calura}, F., \& {Charlot}, S. 2006, \mnras, 373, 397
1004: 
1005: \bibitem[{{Springel}(2005)}]{2005MNRAS.364.1105S}
1006: {Springel}, V. 2005, \mnras, 364, 1105
1007: 
1008: \bibitem[{{Tormen}(1997)}]{1997MNRAS.290..411T}
1009: {Tormen}, G. 1997, \mnras, 290, 411
1010: 
1011: \bibitem[{{Tornatore} {et~al.}(2003){Tornatore}, {Borgani}, {Springel},
1012:   {Matteucci}, {Menci}, \& {Murante}}]{2003MNRAS.342.1025T}
1013: {Tornatore}, L., {Borgani}, S., {Springel}, V., {Matteucci}, F., {Menci}, N.,
1014:   \& {Murante}, G. 2003, \mnras, 342, 1025
1015: 
1016: \bibitem[{{Valdarnini}(2006)}]{2006NewA...12...71V}
1017: {Valdarnini}, R. 2006, New Astronomy, 12, 71
1018: 
1019: \bibitem[{{Vikhlinin}(2006)}]{2006ApJ...640..710V}
1020: {Vikhlinin}, A. 2006, \apj, 640, 710
1021: 
1022: \bibitem[{Voit(2005)}]{Voit:2005PhysRevD}
1023: Voit, G.~M. 2005, \prd, 77, 207
1024: 
1025: \bibitem[{{Voit} {et~al.}(2002){Voit}, {Bryan}, {Balogh}, \&
1026:   {Bower}}]{2002ApJ...576..601V}
1027: {Voit}, G.~M., {Bryan}, G.~L., {Balogh}, M.~L., \& {Bower}, R.~G. 2002, \apj,
1028:   576, 601
1029: 
1030: \bibitem[{{Wang} \& {Steinhardt}(1998)}]{1998ApJ...508..483W}
1031: {Wang}, L., \& {Steinhardt}, P.~J. 1998, \apj, 508, 483
1032: 
1033: \end{thebibliography}
1034: 
1035: \end{document}
1036: 
1037: % LocalWords:  ap XT Spearman's Pred VIR keV eq BT Ameglio al hbp eps Jeltema
1038: % LocalWords:  VALDARNINI Valdarnini EW SL Ventimiglia Voit Borgani ICM AGN Mpc
1039: % LocalWords:  PhysRevD pre thermalizes ij pred Dipartimento di Astronomia NNG
1040: % LocalWords:  dell'Universit Tiepolo arXiv ameglio