0704.3553/v2.tex
1: \documentclass[twocolumn,showpacs,superscriptaddress,preprintnumbers,amsmath,amssymb]{revtex4}
2: \usepackage{graphicx}
3: \usepackage{dcolumn}
4: \usepackage{bm}
5: 
6: \def\simge{\mathrel{
7:      \rlap{\raise 0.511ex \hbox{$>$}}{\lower 0.511ex \hbox{$\sim$}}}}
8: \def\simle{\mathrel{
9:      \rlap{\raise 0.511ex \hbox{$<$}}{\lower 0.511ex \hbox{$\sim$}}}}
10: 
11: \begin{document}
12: \preprint{SPhT-T07/028}
13: \title{The centrality dependence of elliptic flow, the hydrodynamic
14:   limit, and the viscosity of hot QCD}
15: 
16: \author{Hans-Joachim Drescher}
17: \affiliation{
18: Frankfurt Institute for Advanced Studies (FIAS),
19: Johann Wolfgang Goethe-Universit\"at,
20: Max-von-Laue-Str.~1, 60438  Frankfurt am Main, Germany
21: }
22: \author{Adrian Dumitru}
23: \affiliation{
24: Institut f\"ur Theoretische Physik,
25: Johann Wolfgang Goethe-Universit\"at,
26: Max-von-Laue-Str. 1, 60438  Frankfurt am Main, Germany
27: }
28: \author{Cl\'ement Gombeaud}
29: \author{Jean-Yves Ollitrault}
30: \affiliation{Service de Physique Th\'eorique, CEA/DSM/SPhT,
31:   CNRS/MPPU/URA2306\\ CEA Saclay, F-91191 Gif-sur-Yvette Cedex.}
32: 
33: \date{\today}
34: \begin{abstract}
35: We show that the centrality and system-size dependence of elliptic
36: flow measured at RHIC are fully described by a simple model based on
37: eccentricity scaling and incomplete thermalization.  We argue that the
38: elliptic flow is at least 25\% below the (ideal) ``hydrodynamic
39: limit'', even for the most central Au-Au collisions. This lack of
40: perfect equilibration allows for estimates of the effective parton
41: cross section in the Quark-Gluon Plasma and of its viscosity to
42: entropy density ratio. We also show how the initial conditions affect the 
43: transport coefficients and thermodynamic quantities extracted from the
44: data, in particular the viscosity and the speed of sound.
45: \end{abstract}
46: \pacs{12.38.Mh,24.85.+p,25.75.Ld,25.75.-q}
47: \maketitle
48: 
49: When two ultrarelativistic nuclei collide at non-zero impact
50: parameter, their overlap area in the transverse plane has a short axis,
51: parallel to the impact parameter, and a long axis perpendicular to it. 
52: This almond shape of the initial profile is converted by the pressure gradient
53: into a momentum asymmetry,
54: so that more particles are emitted along the short 
55: axis~\cite{Ollitrault:1992bk}. 
56: The magnitude of this effect is characterized by elliptic flow,
57: defined as 
58: \begin{equation}
59: \label{defv2}
60: v_2\equiv\langle\cos 2(\varphi-\Phi_R)\rangle,
61: \end{equation}
62: where $\varphi$ is the azimuthal angle of an outgoing particle, $\Phi_R$ is
63: the azimuthal angle of the impact parameter,  and 
64: angular brackets denote an average over many particles and many 
65: events. The unexpected large magnitude of elliptic flow at
66: RHIC~\cite{Ackermann:2000tr} has generated a lot of
67: activity in recent years. 
68: 
69: Elliptic flow results from the interactions between the produced
70: particles, and can be used to probe local thermodynamic equilibrium. 
71: If the produced matter equilibrates, it behaves as an ideal fluid. 
72: Hydrodynamics predicts that at a given energy, $v_2$ 
73: scales like the eccentricity $\varepsilon$ of the 
74: almond~\cite{Ollitrault:1992bk,Sorge:1998mk}. 
75: It is independent of its transverse size $R$, as a
76: consequence of the scale invariance of ideal-fluid dynamics. 
77: If, on the other hand, equilibration is incomplete, then
78: eccentricity scaling is broken and $v_2/\varepsilon$ also depends on
79: the Knudsen number $K=\lambda/R$, where $\lambda$ is the length scale
80: over which a parton is deflected by a large angle.
81: 
82: Here, we show that the centrality dependence of $v_2/\varepsilon$, for
83: both Au+Au and Cu+Cu collisions, can be described by the following simple
84: formula~\cite{Bhalerao:2005mm}: 
85: \begin{equation}
86: \label{v2k}
87: \frac{v_2}{\varepsilon}=\frac{v_2^{\rm hydro}}{\varepsilon}\frac{1}{1+K/K_0}~.
88: \end{equation}
89: $v_2/\varepsilon$ is largest in the hydrodynamic limit $K\to 0$.  The
90: first order corrections to this limit, corresponding to viscous
91: effects, are linear in $K$. For large mean-free path, far from the
92: hydrodynamic limit, $v_2/\varepsilon\sim1/K$ vanishes like the number
93: of collisions per particle.  One expects the transition between these
94: two regimes to occur when $\lambda\simeq R$, hence that $K_0\simeq1$.
95: A recent transport calculation~\cite{Gombeaud:2007ub} in two spatial
96: dimensions indeed obtained $K_0\simeq 0.7$.
97: 
98: Elliptic flow develops gradually during the early stages of the collision.
99: Due to the strong longitudinal expansion, the thermodynamic 
100: properties of the medium depend on the time $\tau$, of course. The
101: average particle density, for instance, decreases like $1/\tau$ (if
102: their number is approximately conserved, see recent discussion
103: in~\cite{DMN}):
104: \begin{equation}  \label{density}
105: \rho(\tau) = \frac{1}{\tau S} \frac{dN}{dy},
106: \end{equation}
107: where $dN/dy$ denotes the total (charged + neutral) multiplicity per unit
108: rapidity, and $S$ is the transverse overlap area between the two nuclei.
109: The quantities that we shall extract from $v_2$ should
110: be intepreted as averages over the transverse area $S$, and over 
111: some time interval around $R/c_s$, which is the typical time scale for the
112: build-up of $v_2$ in hydrodynamics~\cite{Bhalerao:2005mm}. $c_s$ denotes
113: the velocity of sound. 
114: 
115: The Knudsen number $K$ is defined by evaluating the mean free path
116: $\lambda=1/\sigma\rho$ ($\sigma$ is a partonic cross section) at
117: $\tau=R/c_s$. Thus,
118: \begin{equation}
119: \label{knud}
120: \frac{1}{K} 
121: % = \sigma\rho R
122: =\frac{\sigma}{S}\frac{dN}{dy} \, c_s~.
123: \end{equation}
124: The purpose of this Letter is to show that the centrality and
125: system-size dependence of the data for $v_2$ at RHIC is described very
126: well by Eqs.~(\ref{v2k}) and~(\ref{knud}). This provides three
127: important pieces of information. First, such a fit allows us to
128: ``measure'' the Knudsen number corresponding to a given centrality,
129: which quantifies how close the dense matter produced in heavy-ion
130: collisions at RHIC is to perfect fluidity. Second, the extrapolation
131: to $K=0$ allows us to read off the limiting value for $v_2^{\rm
132: hydro}/\varepsilon$ extracted from the {\em data}; this is useful for
133: constraining the equation of state (EoS) of QCD via hydrodynamic
134: simulations, and we shall also see that it exhibits a rather
135: surprising dependence on the initial conditions. Finally, using 
136: Eq.~(\ref{knud}), 
137: %the
138: %above estimates for the relevant time-scale and for the speed of sound,
139: we can convert the Knudsen number into the typical
140: partonic cross section $\sigma$ (and viscosity) in the Quark-Gluon
141: Plasma (QGP). 
142: %It should be clear that the measured elliptic flow
143: %corresponds to an integral over the time-evolution of a heavy-ion
144: %collision. All of the above quantities should therefore be interpreted
145: %as averaged values, too. 
146: Since only the combination $K_0 \sigma c_s$ 
147: actually appears in Eq.~(\ref{v2k}), uncertainties in $K_0$ or $c_s$ 
148: then translate into corresponding uncertainties of $\sigma$. 
149: %We shall provide some estimates below. 
150: Unless mentioned otherwise, our standard choice is %$\tau=R/c_s$ with 
151: $c_s=1/\sqrt{3}\simeq 0.58$ (ideal quark-gluon plasma) and $K_0= 0.7$.  
152: Letting $K_0=1$ and $c_s^2=2/3$\footnote{Such a ``hard'' EoS can arise from
153:   repulsive long-range interactions among the
154:   partons such as classical fields. We thank V.~Koch for pointing this
155: out to us.} instead reduces the 
156: estimated $\sigma$ by a factor of two; on the other hand, taking
157: $K_0=0.5$ and $c_s^2=1/6$ increases $\sigma$ by the same factor.
158: 
159: 
160: 
161: For the elliptic flow, $v_2$, we use PHOBOS data for Au-Au
162: \cite{Back:2004mh} and Cu-Cu~\cite{Alver:2006wh} collisions. The same
163: analysis could be carried out using data from
164: PHENIX~\cite{Adare:2006ti} or STAR~\cite{Adams:2004bi}.  The initial
165: eccentricity $\varepsilon$ and the transverse density $(1/S)(dN/dy)$
166: are evaluated using a model of the collision.  Two such models will be
167: compared.  The remaining parameters $v_2^{\rm hydro}$ and $\sigma$ are
168: fit to the data. The first step is to plot $v_2/\varepsilon$ versus
169: $(1/S)(dN/dy)$~\cite{Voloshin:1999gs}.  Such plots have already been
170: obtained at SPS and RHIC~\cite{Alt:2003ab}, and they are puzzling:
171: while $v_2/\varepsilon$ increases with centrality, it shows no hint of
172: the {\em saturation} predicted by Eq.~(\ref{v2k}) for $K/K_0\simle 1$,
173: suggesting that the system is far from
174: equilibrium~\cite{Bhalerao:2005mm}.  On the other hand, the value of
175: $v_2$ for central Au-Au collisions at RHIC is about as high as
176: predicted by hydrodynamics, which is widely considered as key evidence
177: that a ``perfect liquid'' has been created at
178: RHIC~\cite{Tannenbaum:2006ch}.
179: 
180: It was understood only recently that the eccentricity of the overlap
181: zone has so far been underestimated, as the result of two effects.
182: The first effect is fluctuations in initial
183: conditions~\cite{Socolowski:2004hw}: the time scale of the
184: nucleus-nucleus collision at RHIC is so short that each nucleus
185: remains in a frozen configuration, with its nucleons distributed
186: according to the nuclear wave function.  Fluctuations in the nucleon
187: positions result in fluctuations of the overlap area. Their effect on
188: elliptic flow was first pointed out in Ref.~\cite{Miller:2003kd}. It
189: was later realized by the PHOBOS
190: collaboration~\cite{Alver:2006wh,Manly:2005zy} that the orientation of
191: the almond may also fluctuate, so that $\Phi_R$ in Eq.~(\ref{defv2})
192: is no longer the direction of impact parameter, but the minor axis of
193: the ellipse defined by the positions of the nucleons. These
194: fluctuations explain both the large magnitude of $v_2$ for small
195: systems, such as Cu-Cu collisions, as well as the non-zero magnitude
196: of $v_2$ in central collisions, where the eccentricity would otherwise
197: vanish. They have to be taken into account in order to observe the
198: expected saturation of $v_2/\varepsilon$ at high density mentioned
199: above.
200: 
201: The eccentricity is usually estimated from the distribution of
202: participant nucleons in the transverse plane (Glauber model). More
203: precisely, we assume here that the density distribution of produced
204: particles is given by a fixed 80\%:20\% superposition of participant
205: and binary-collision scaling, respectively~\cite{Kharzeev:2000ph}.
206: For Au-Au collisions, this simple model reproduces the centrality
207: dependence of the multiplicity reasonably well (we assume that charged
208: particles are 2/3 of the total multiplicity, and that
209: $dN/d\eta\simeq 0.8\, dN/dy$ at midrapidity), while it
210: underestimates it for central Cu-Cu collisions by about 10\%.
211: 
212: At high energies a second effect which increases the
213: eccentricity is perturbative gluon saturation, which determines the
214: $p_\perp$-integrated multiplicity from weak-coupling QCD without
215: additional models for soft particle production.  High-density QCD (the
216: ``Color-Glass Condensate'') predicts a different distribution of
217: produced gluons, $dN/d^2{\bf r}_\perp dy$, which gives a similar
218: centrality dependence of the multiplicity~\cite{Kharzeev:2000ph} but a
219: larger eccentricity~\cite{Hirano:2005xf,Drescher:2006pi}.  When
220: particle production is dominated by transverse momenta below the
221: saturation scale of the denser nucleus, then $dN/d^2{\bf r}_\perp
222: dy\sim {\rm min}(n^A_{\rm part}({\bf r}_\perp), n^B_{\rm part}({\bf
223: r}_\perp))$ traces the participant density of the more dilute
224: collision partner, rather than the average as in the Glauber
225: model~\cite{Drescher:2006pi}. Precise figures depend on how the
226: saturation scale is defined~\cite{Lappi:2006xc}. Naively, the larger
227: initial eccentricity predicted by the gluon saturation approach is
228: expected to require more dissipation in order to reproduce the same
229: experimentally measured $v_2$. Somewhat surprisingly, we shall find
230: that this expectation is incorrect, which underscores the non-trivial
231: role played by the initial conditions.
232: 
233: Both effects, fluctuations and gluon saturation, were recently
234: combined by Drescher and Nara~\cite{Drescher:2006ca}. In their
235: approach, the saturation momenta and the unintegrated gluon
236: distribution functions of the colliding nuclei are determined for each
237: configuration individually. The finite interaction range of the
238: nucleons is also taken into account. Upon convolution of the
239: projectile and target unintegrated gluon distribution functions and
240: averaging over configurations, the model leads to a very good
241: description of the multiplicity for both Au-Au as well as Cu-Cu
242: collisions over the entire available range of centralities.
243: 
244: Having determined the density distributions of produced particles from
245: either model as described above, we obtain the eccentricity
246: via~\cite{Miller:2003kd,Bhalerao:2006tp}
247: \begin{equation}
248: \varepsilon = \sqrt{\langle\varepsilon_{\rm
249:     part}^2\rangle}\quad,\quad
250: \varepsilon_{\rm part} = \frac{\sqrt{(\sigma_y^2-\sigma_x^2)^2 +
251:     4\sigma_{xy}^2}}
252: {\sigma_x^2+\sigma_y^2}~.
253: \end{equation}
254: $\sigma_x$, $\sigma_y$ are the respective root-mean-square widths of
255: the density distributions, and $\sigma_{xy}=\overline{xy}-\bar{x}
256: \bar{y}$ (a bar denotes a convolution with the density distribution
257: for a given configuration while brackets stand for averages over
258: configurations). The overlap area $S$ is defined by $S\equiv
259: 4\pi\sigma_x\sigma_y$~\cite{Gombeaud:2007ub}. We find it more
260: appropriate to define these moments via the number density
261: distribution $dN/d^2{\bf r}_\perp dy$ rather than the energy density
262: distribution $dE_\perp/d^2{\bf r}_\perp dy$. 
263: The reason is twofold: first, $v_2$ is extracted experimentally from
264: the azimuthal distribution of particle number, not transverse energy;
265: second, our CGC approach describes the centrality dependence of the
266: {\em measured} final-state multiplicity very well, which indicates
267: that the ratio of final-state particles to initial-state gluons
268: (including possible gluon multiplication processes~\cite{BMSS}) is
269: essentially constant.
270: 
271: 
272: 
273: 
274: \begin{figure}
275: \includegraphics*[width=\linewidth]{glauber}
276: \caption{Variation of the scaled elliptic flow with the density,
277:   assuming initial conditions from the Glauber model. 
278: The line is a 2-parameter fit using Eqs.~(\ref{v2k}) and
279:   (\ref{knud}). 
280: \label{fig:glauber}}
281: \end{figure}
282: 
283: 
284: 
285: \begin{figure}
286: \includegraphics*[width=\linewidth]{cgc}
287: \caption{Same as Fig.~\ref{fig:glauber}, using CGC initial conditions.
288: \label{fig:cgc}}
289: \end{figure}
290: 
291: Figs.~\ref{fig:glauber} and \ref{fig:cgc} display $v_2/\varepsilon$ as
292: a function of $(1/S)(dN/dy)$ for Au-Au and Cu-Cu collisions at various
293: centralities, within the Glauber and CGC approaches, respectively. For
294: both types of initial conditions, Cu-Cu and Au-Au collisions at the
295: same $(1/S)(dN/dy)$ give the same $v_2/\varepsilon$ within error bars.
296: Eccentricity fluctuations are crucial for this
297: agreement~\cite{Alver:2006wh}.  The figures also show that
298: Eqs.~(\ref{v2k}) and (\ref{knud}) provide a good fit to the data, for
299: both sets of initial conditions.  On the other hand, the values of the
300: fit parameters clearly depend on the initial conditions, which has
301: important consequences for the physics.
302: 
303: 
304: The first physical quantity extracted from the fit is the hydrodynamic
305: limit, $v_2^{\rm hydro}/\varepsilon$, obtained by extrapolating to
306: $(1/S)(dN/dy)\to \infty$.  The values are $v_2^{\rm hydro} /
307: \varepsilon = 0.30\pm 0.02$ with the Glauber parameterization, and
308: $v_2^{\rm hydro}/\varepsilon=0.22\pm 0.01$ with CGC initial
309: conditions. Comparing these numbers to the experimental data points
310: one observes that deviations from ideal hydrodynamics are as
311: large as 30\%, even for central Au-Au collisions. This is our first
312: important result.
313: 
314: So far, a quantitative extraction of the QCD EoS from
315: RHIC data via hydrodynamic analysis was hampered by the fact that
316: $v_2/\varepsilon$ had not been factorized into the perfect-fluid part
317: $v_2^{\rm hydro}/\varepsilon$ and the dissipative correction
318: $1/(1+K/K_0)$.  For example, Huovinen found~\cite{Huovinen:2005gy}
319: that an EoS with a rapid cross-over rather than a strong first-order
320: phase transition, as favored by lattice QCD~\cite{Bernard:2006nj},
321: overpredicted the flow data. This finding was rather puzzling, too, as
322: it was widely believed that the RHIC data fully saturates the
323: hydrodynamic limit. Our results suggest that ideal hydrodynamics {\em
324: should} in fact overpredict the measured flow. That is, that one should
325: not choose an EoS in perfect-fluid simulations that fits the
326: data. Rather, the EoS could be extracted by comparing ideal
327: hydrodynamics to $v_2^{\rm hydro}/\varepsilon$.
328: 
329: The next result is that CGC initial conditions, which predict a higher
330: initial eccentricity $\varepsilon$, naturally lead to a lower
331: hydrodynamic limit $v_2^{\rm hydro}/\varepsilon$.  Now, close to the
332: ideal-gas limit ($c_s=1/\sqrt{3}$), $v_2^{\rm hydro}/\varepsilon$
333: scales approximately like the sound velocity
334: $c_s$~\cite{Bhalerao:2005mm}.  This means that CGC initial conditions
335: imply a lower average speed of sound (softer equation of state) than
336: Glauber initial conditions, by a factor of $0.22/0.3\simeq 0.73$.
337: 
338: 
339: The second fit parameter is the partonic cross section $\sigma$. The
340: larger $\sigma$, the faster the saturation of $v_2/\varepsilon$ as a
341: function of $(1/S)(dN/dy)$. For our standard values of $K_0$ and $c_s$
342: we obtain $\sigma=4.3\pm 0.6$~mb for Glauber initial conditions and
343: $\sigma=5.5\pm 0.5$~mb for CGC initial conditions. These values are
344: significantly smaller than those found in previous transport
345: calculations~\cite{Molnar}, but match the findings of
346: ref.~\cite{XuGreiner}.
347: 
348: CGC initial conditions imply a larger value of $\sigma$ than Glauber
349: initial conditions, that is, a {\it lower} viscosity.  This can be
350: easily understood.  As already mentioned above, the CGC predicts a
351: larger eccentricity $\varepsilon$ than the Glauber model for
352: semi-central collisions of large nuclei (when there is a large
353: asymmetry in the local saturation scales of the collision partners,
354: along a path in impact-parameter direction away from the
355: origin~\cite{Drescher:2006pi}). However, for very peripheral
356: collisions or small nuclei, there is of course very little asymmetry
357: in the saturation scales, and the eccentricity approaches the same
358: value as in the Glauber model. This has been checked numerically in
359: fig.~7 of ref.~\cite{Drescher:2006ca}, and can also be clearly seen by
360: comparing our figures: while in Fig.~\ref{fig:cgc} $v_2/\varepsilon$
361: for semi-central Au+Au collisions is lower than in
362: Fig.~\ref{fig:glauber}, there is no visible difference for peripheral
363: Cu+Cu collisions. In all, with CGC initial conditions the scaled flow
364: grows less rapidly with the transverse density, which is the reason
365: for the larger elementary cross-section.
366: 
367: The dependence of $\sigma$ on the initial conditions is probably even
368: stronger than the numerical values above suggest, for the following
369: reason.  As alluded to above, our fit to the data really determines
370: the product $K_0\sigma c_s$, rather than $\sigma$ alone.  It appears
371: reasonable to assume that $K_0$ does not depend on the initial
372: conditions. However, for consistency, the speed of sound $c_s$
373: entering the Knudsen number should match the one underlying the
374: hydrodynamic limit $v_2^{\rm hydro} / \varepsilon$; hence, if CGC
375: initial conditions require a smaller $c_s$ by a factor $0.73$, the
376: elementary cross-section obtained above should be rescaled
377: accordingly. This leads to our final estimate $\sigma_{CGC}\simeq
378: 7.6\pm 0.7$~mb.
379: 
380: Our numerical results for $\sigma$ should be taken as rough estimates
381: rather than precise figures, because of the uncertainties related to
382: the precise values of $K_0$ and $c_s$. It is, however, tempting
383: to convert them into estimates of the shear viscosity 
384: $\eta$, which has been of great interest lately.  A
385: universal lower bound $\eta/s\ge 1/4\pi$ (where $s$ is the entropy
386: density) has been conjectured using a correspondence with black-hole
387: physics~\cite{Kovtun:2004de}, and it has been argued that the
388: viscosity of QCD might be close to the lower bound.  Extrapolations of
389: perturbative estimates to temperatures $T\simeq200$~MeV, on the other
390: hand, suggest that the viscosity of QCD could be much
391: larger~\cite{Huot:2006ys}.  On the microscopic side, $\eta$ is related
392: to the scattering cross-section $\sigma$. Following
393: Teaney~\cite{Teaney:2003kp}, the relation for a classical gas of
394: massless particles with isotropic differential cross sections 
395: (which applies, for example, to a Boltzmann-transport model)
396: is $\eta=1.264\, T/\sigma$~\cite{hardspheres}. On the other hand, the
397: entropy density of a classical ultrarelativistic cas is
398: $s=4n$, with $n$ the particle density, so that
399: \begin{equation}
400: \label{eta}
401: \frac{\eta}{s}=0.316\frac{T}{c\sigma n}=0.316\frac{\lambda T}{c}.
402: \end{equation}
403: The relevant particle density in Au-Au collisions at RHIC, which is
404: estimated at the time when $v_2$ develops~\cite{Bhalerao:2005mm}, is
405: 3.9~fm$^{-3}$ for both Glauber and CGC initial conditions, and
406: $T\simeq 200$~MeV. Our two estimates $\sigma=4.3$~mb (Glauber initial
407: conditions) and $\sigma=7.6$~mb (CGC initial conditions) thus
408: translate into $\lambda=0.60$~fm, $\eta/s=0.19$ and $\lambda=0.34$~fm,
409: $\eta/s=0.11$, respectively.  These values for $\eta/s$ agree with
410: those from ref.~\cite{lacey} if the mean-free path is scaled to our
411: result, and also with estimates of $\eta/s$ based on the observed
412: energy loss and elliptic flow of heavy quarks~\cite{phenixQ}, on
413: transverse momentum correlations~\cite{GavinAziz}, or
414: bounds on entropy production~\cite{DMN}. Hence,
415: for our best fit(s) $\eta/s$ is slightly larger than the conjectured
416: lower bound, but significantly smaller than extrapolations from
417: perturbative estimates. On the other hand, our lower value is close to
418: a recent lattice estimate~\cite{Meyer:2007ic} for SU(3) gluodynamics,
419: which gives $\eta/s=0.134\pm 0.033$ at $T=1.65\, T_c$.
420: 
421: A complementary approach to incorporate corrections from the
422: ideal-fluid limit is viscous relativistic hydrodynamics. A formulation
423: that is suitable for applications to high-energy heavy-ion collisions
424: has been developped in recent years~\cite{Muronga:2001gn}.  A first
425: calculation of elliptic flow~\cite{Romatschke:2007mq} shows that for 
426: Glauber initial
427: conditions and $\eta/s=0.16$, $v_2$ reaches about $70\%$ of the
428: ideal-fluid value for semi-central Au-Au collisions. It is interesting
429: to note that our simple estimates are in good agreement with this
430: finding.  Using Eq.~(\ref{eta}), $\eta/s=0.16$ corresponds to
431: $\sigma=5.1$~mb, for which Eqs.~(\ref{v2k}) and (\ref{knud}) give
432: $v_2/v_2^{\rm hydro}=0.68$.  The comparison to experimental data in
433: Ref.~\cite{Romatschke:2007mq}, however, appears to favor lower values
434: of $\eta/s$ because the EoS used there underpredicts $v_2^{\rm
435: hydro}/\varepsilon\simeq0.3$ required for Glauber initial
436: conditions. Alternatively, simulations could be performed with CGC
437: initial conditions which require only $v_2^{\rm
438: hydro}/\varepsilon\simeq0.22$.
439: 
440: In summary, we have shown that the centrality and system-size
441: dependence of the {\em measured} $v_2$ can be understood as follows:
442: $v_2$ scales like the initial eccentricity $\varepsilon$ (as predicted
443: by hydrodynamics), multiplied by a correction factor due to
444: off-equilibrium (i.e., viscous) effects. This correction involves the
445: multiplicity density in the overlap area, $(1/S)(dN/dy)$.  Two types
446: of initial conditions have been compared: a Glauber-type model, and a
447: Color-Glass Condensate approach. PHOBOS data can be described with
448: both. In particular, there is good agreement between Cu-Cu and Au-Au
449: data. The resulting estimates for thermodynamic quantities and
450: transport coefficients, on the other hand, depend significantly on the
451: initial conditions.
452: 
453: Color glass condensate-type initial conditions require {\em lower}
454: viscosity and a {\em softer} equation of state (smaller speed of
455: sound). The scaled flow extrapolated to vanishing mean-free path is
456: lower than for Glauber initial conditions by a factor of $\simeq
457: 0.22/0.3=0.73$; the effective speed of sound should also be lower by
458: about the same factor. Our estimates for the viscosity are
459: $\eta/s\simeq 0.19$ for Glauber initial conditions, and $\eta/s\simeq
460: 0.11$ for CGC initial conditions, but these numbers should be taken
461: only as rough estimates.
462: 
463: We have also shown that the data for the scaled flow indeed {\em
464: saturate} at high densities to a hydrodynamic limit. In central Au-Au
465: collisions at RHIC, $v_2$ reaches 70\% (resp.\ 75\%) of the
466: hydrodynamic limit for Glauber (CGC) initial conditions. The
467: corrections to ideal hydrodynamics are therefore significant, but
468: reasonably small compared to unity, implying that (viscous)
469: hydrodynamics should be a valid approach for understanding flow at
470: RHIC. Also, the asymptotic limit of $v_2/\varepsilon$ has been
471: isolated and could now be used to test realistic equations of state
472: from lattice-QCD with hydrodynamic simulations of heavy-ion
473: collisions.
474: 
475: 
476: \section*{Acknowledgments}
477: 
478: J.Y.O.\ thanks B.\ Alver, A.H.\ Mueller and D.\ Schiff for helpful
479: discussions.  H.J.D.\ is supported through BMBF grant 05 CU5RI1/3.
480: 
481: \begin{thebibliography}{99}
482: 
483: %\cite{Ollitrault:1992bk}
484: \bibitem{Ollitrault:1992bk}
485:   J.~Y.~Ollitrault,
486:   %``Anisotropy As A Signature Of Transverse Collective Flow,''
487:   Phys.\ Rev.\  D {\bf 46}, 229 (1992).
488:   %%CITATION = PHRVA,D46,229;%%
489: 
490: %\cite{Ackermann:2000tr}
491: \bibitem{Ackermann:2000tr}
492:   K.~H.~Ackermann {\it et al.}, % [STAR Collaboration],
493:   %``Elliptic flow in Au + Au collisions at s(N N)**(1/2) = 130-GeV,''
494:   Phys.\ Rev.\ Lett.\  {\bf 86}, 402 (2001).
495:   %%CITATION = NUCL-EX 0009011;%%
496: 
497: %\cite{Sorge:1998mk}
498: \bibitem{Sorge:1998mk}
499:   H.~Sorge,
500:   %``Highly sensitive centrality dependence of elliptic flow: A novel signature
501:   %of the phase transition in QCD,''
502:   Phys.\ Rev.\ Lett.\  {\bf 82}, 2048 (1999).
503:   %%CITATION = PRLTA,82,2048;%%
504: 
505: %\cite{Bhalerao:2005mm}
506: \bibitem{Bhalerao:2005mm}
507:   R.~S.~Bhalerao, J.~P.~Blaizot, N.~Borghini and J.~Y.~Ollitrault,
508:   %``Elliptic flow and incomplete equilibration at RHIC,''
509:   Phys.\ Lett.\  B {\bf 627}, 49 (2005).
510:   %%CITATION = PHLTA,B627,49;%%
511: 
512: %\cite{Gombeaud:2007ub}
513: \bibitem{Gombeaud:2007ub}
514:   C.~Gombeaud and J.~Y.~Ollitrault,
515:   %``A covariant transport approach to elliptic flow,''
516:   arXiv:nucl-th/0702075.
517:   %%CITATION = NUCL-TH/0702075;%%
518: 
519: \bibitem{DMN}
520: A.~Dumitru, E.~Molnar and Y.~Nara,
521: %``Entropy production in high-energy heavy-ion collisions and the
522: %correlation of shear viscosity and thermalization time,''
523: arXiv:0706.2203 [nucl-th].
524: %%CITATION = ARXIV:0706.2203;%%
525: 
526: %\cite{Back:2004mh}
527: \bibitem{Back:2004mh}
528:   B.~B.~Back {\it et al.}  [PHOBOS Collaboration],
529:   %``Centrality and pseudorapidity dependence of elliptic flow for charged
530:   %hadrons in Au + Au collisions at s(NN)**(1/2) = 200-GeV,''
531:   Phys.\ Rev.\  C {\bf 72}, 051901 (2005).
532:   %%CITATION = PHRVA,C72,051901;%%
533: 
534: %\cite{Alver:2006wh}
535: \bibitem{Alver:2006wh}
536:   B.~Alver {\it et al.}  [PHOBOS Collaboration],
537:   %``System size, energy, pseudorapidity, and centrality dependence of elliptic
538:   %flow,''
539:   arXiv:nucl-ex/0610037.
540:   %%CITATION = NUCL-EX/0610037;%%
541: 
542: %\cite{Adare:2006ti}
543: \bibitem{Adare:2006ti}
544:   A.~Adare {\it et al.}  [PHENIX Collaboration],
545:   %``Scaling properties of azimuthal anisotropy in Au + Au and Cu + Cu
546:   %collisions at s(NN)**(1/2) = 200-GeV,''
547:   arXiv:nucl-ex/0608033.
548:   %%CITATION = NUCL-EX/0608033;%%
549: 
550: %\cite{Adams:2004bi}
551: \bibitem{Adams:2004bi}
552:   J.~Adams {\it et al.}  [STAR Collaboration],
553:   %``Azimuthal anisotropy in Au + Au collisions at s(NN)**(1/2) = 200-GeV,''
554:   Phys.\ Rev.\  C {\bf 72}, 014904 (2005).
555:   %%CITATION = PHRVA,C72,014904;%%
556: 
557: %\cite{Voloshin:1999gs}
558: \bibitem{Voloshin:1999gs}
559:   S.~A.~Voloshin and A.~M.~Poskanzer,
560:   %``The physics of the centrality dependence of elliptic flow,''
561:   Phys.\ Lett.\  B {\bf 474}, 27 (2000).
562:   %%CITATION = PHLTA,B474,27;%%
563: 
564: %\cite{Alt:2003ab}
565: \bibitem{Alt:2003ab}
566:   C.~Alt {\it et al.}  [NA49 Collaboration],
567:   %``Directed and elliptic flow of charged pions and protons in Pb + Pb
568:   %collisions at 40-A-GeV and 158-A-GeV,''
569:   Phys.\ Rev.\  C {\bf 68}, 034903 (2003);
570:   %%CITATION = PHRVA,C68,034903;%%
571:   M.~M.~Aggarwal {\it et al.}  [WA98 Collaboration],
572:   %``Centrality and transverse momentum dependence of collective flow in
573:   %158-A-GeV Pb + Pb collisions measured via inclusive photons,''
574:   Nucl.\ Phys.\  A {\bf 762}, 129 (2005);
575:   %%CITATION = NUPHA,A762,129;%%
576:   S.~A.~Voloshin  [STAR Collaboration],
577:   %``Energy and system size dependence of charged particle elliptic flow and
578:   %v(2)/epsilon scaling,''
579:   arXiv:nucl-ex/0701038.
580:   %%CITATION = NUCL-EX/0701038;%%
581: 
582: %\cite{Tannenbaum:2006ch}
583: \bibitem{Tannenbaum:2006ch}
584:   E.~V.~Shuryak,
585:   %``What RHIC experiments and theory tell us about properties of  quark-gluon
586:   %plasma?,''
587:   Nucl.\ Phys.\  A {\bf 750}, 64 (2005);
588:   %[arXiv:hep-ph/0405066].
589:   %%CITATION = NUPHA,A750,64;%%
590:   M.~J.~Tannenbaum,
591:   %``Recent results in relativistic heavy ion collisions: From 'a new state  of
592:   %matter' to 'the perfect fluid',''
593:   Rept.\ Prog.\ Phys.\  {\bf 69}, 2005 (2006).
594:   %%CITATION = NUCL-EX 0603003;%%
595: 
596: %\cite{Socolowski:2004hw}
597: \bibitem{Socolowski:2004hw}
598:   O.~Socolowski, F.~Grassi, Y.~Hama and T.~Kodama,
599:   %``Fluctuations of the initial conditions and the continuous emission in
600:   %hydro description of two-pion interferometry,''
601:   Phys.\ Rev.\ Lett.\  {\bf 93}, 182301 (2004).
602:   %%CITATION = PRLTA,93,182301;%%
603: 
604: %\cite{Miller:2003kd}
605: \bibitem{Miller:2003kd}
606:   M.~Miller and R.~Snellings,
607:   %``Eccentricity fluctuations and its possible effect on elliptic flow
608:   %measurements,''
609:   arXiv:nucl-ex/0312008.
610:   %%CITATION = NUCL-EX/0312008;%%
611: 
612: %\cite{Manly:2005zy}
613: \bibitem{Manly:2005zy}
614:   S.~Manly {\it et al.}  [PHOBOS Collaboration],
615:   %``System size, energy and pseudorapidity dependence of directed and elliptic
616:   %flow at RHIC,''
617:   Nucl.\ Phys.\  A {\bf 774}, 523 (2006).
618:   %%CITATION = NUPHA,A774,523;%%
619: 
620: %\cite{Kharzeev:2000ph}
621: \bibitem{Kharzeev:2000ph}
622:   D.~Kharzeev and M.~Nardi,
623:   %``Hadron production in nuclear collisions at RHIC and high density QCD,''
624:   Phys.\ Lett.\  B {\bf 507}, 121 (2001);
625:   %%CITATION = PHLTA,B507,121;%%
626:   D.~Kharzeev and E.~Levin,
627:   %``Manifestations of high density QCD in the first RHIC data,''
628:   Phys.\ Lett.\  B {\bf 523}, 79 (2001).
629:   %[arXiv:nucl-th/0108006].
630:   %%CITATION = PHLTA,B523,79;%%
631: 
632: %\cite{Hirano:2005xf}
633: \bibitem{Hirano:2005xf}
634:   T.~Hirano, U.~W.~Heinz, D.~Kharzeev, R.~Lacey and Y.~Nara,
635:   %``Hadronic dissipative effects on elliptic flow in ultrarelativistic
636:   %heavy-ion collisions,''
637:   Phys.\ Lett.\  B {\bf 636}, 299 (2006);
638:   %%CITATION = PHLTA,B636,299;%%
639:   T.~Hirano,
640:   %``Relativistic Hydrodynamics at RHIC and LHC,''
641:   arXiv:0704.1699 [nucl-th].
642:   %%CITATION = ARXIV:0704.1699;%%
643: 
644: %\cite{Drescher:2006pi}
645: \bibitem{Drescher:2006pi}
646:   A.~Adil, H.~J.~Drescher, A.~Dumitru, A.~Hayashigaki and Y.~Nara,
647:   %``The eccentricity in heavy-ion collisions from color glass condensate
648:   %initial conditions,''
649:   Phys.\ Rev.\  C {\bf 74}, 044905 (2006).
650:   %%CITATION = PHRVA,C74,044905;%%
651: 
652: %\cite{Lappi:2006xc}
653: \bibitem{Lappi:2006xc}
654:   T.~Lappi and R.~Venugopalan,
655:   %``Universality of the saturation scale and the initial eccentricity in heavy
656:   %ion collisions,''
657:   Phys.\ Rev.\  C {\bf 74}, 054905 (2006).
658:   %%CITATION = PHRVA,C74,054905;%%
659: 
660: %\cite{Drescher:2006ca}
661: \bibitem{Drescher:2006ca}
662:   H.~J.~Drescher and Y.~Nara,
663:   %``Effects of fluctuations on the initial eccentricity from the color glass
664:   %condensate in heavy ion collisions,''
665:   Phys.\ Rev.\  C {\bf 75}, 034905 (2007).
666:   %%CITATION = PHRVA,C75,034905;%%
667: 
668: %\cite{Bhalerao:2006tp}
669: \bibitem{Bhalerao:2006tp}
670:   R.~S.~Bhalerao and J.~Y.~Ollitrault,
671:   %``Eccentricity fluctuations and elliptic flow at RHIC,''
672:   Phys.\ Lett.\  B {\bf 641}, 260 (2006).
673:   %%CITATION = PHLTA,B641,260;%%
674: 
675: \bibitem{BMSS}
676: R.~Baier, A.~H.~Mueller, D.~Schiff and D.~T.~Son,
677:   %``Does parton saturation at high density explain hadron multiplicities at
678:   %RHIC?,''
679:   Phys.\ Lett.\  B {\bf 539}, 46 (2002).
680:   %[arXiv:hep-ph/0204211].
681:   %%CITATION = PHLTA,B539,46;%%
682: 
683: 
684: %\cite{Molnar:2001ux}
685: \bibitem{Molnar}
686:   D.~Molnar and M.~Gyulassy,
687:   %``Saturation of elliptic flow at RHIC: Results from the covariant elastic
688:   %parton cascade model MPC,''
689:   Nucl.\ Phys.\  A {\bf 697}, 495 (2002)
690:   [Erratum-ibid.\  A {\bf 703}, 893 (2002)].
691:   %%CITATION = NUPHA,A697,495;%%
692: 
693: 
694: \bibitem{XuGreiner}
695:   Z.~Xu and C.~Greiner,
696:   Nucl.\ Phys.\  A {\bf 774}, 787 (2006).
697:   %[arXiv:hep-ph/0509324].
698:   %%CITATION = NUPHA,A774,787;%%
699: 
700: %\cite{Huovinen:2005gy}
701: \bibitem{Huovinen:2005gy}
702:   P.~Huovinen,
703:   %``Anisotropy of flow and the order of phase transition in relativistic heavy
704:   %ion collisions,''
705:   Nucl.\ Phys.\  A {\bf 761}, 296 (2005).
706:   %[arXiv:nucl-th/0505036].
707:   %%CITATION = NUPHA,A761,296;%%
708: 
709: %\cite{Bernard:2006nj}
710: \bibitem{Bernard:2006nj}
711:   C.~Bernard {\it et al.},
712:   %``QCD equation of state with 2+1 flavors of improved staggered quarks,''
713:   arXiv:hep-lat/0611031.
714:   %%CITATION = HEP-LAT/0611031;%%
715: 
716: %\cite{Kovtun:2004de}
717: \bibitem{Kovtun:2004de}
718:   P.~Kovtun, D.~T.~Son and A.~O.~Starinets,
719:   %``Viscosity in strongly interacting quantum field theories from black hole
720:   %physics,''
721:   Phys.\ Rev.\ Lett.\  {\bf 94}, 111601 (2005).
722:   %%CITATION = PRLTA,94,111601;%%
723: 
724: %\cite{Huot:2006ys}
725: \bibitem{Huot:2006ys}
726:   L.~P.~Csernai, J.~I.~Kapusta and L.~D.~McLerran,
727:   %``On the strongly-interacting low-viscosity matter created in relativistic
728:   %nuclear collisions,''
729:   Phys.\ Rev.\ Lett.\  {\bf 97}, 152303 (2006);
730:   %[arXiv:nucl-th/0604032].
731:   %%CITATION = PRLTA,97,152303;%%
732:   S.~C.~Huot, S.~Jeon and G.~D.~Moore,
733:   %``Shear viscosity in weakly coupled N = 4 super Yang-Mills theory  compared
734:   % to QCD,''
735:   hep-ph/0608062.
736:   %%CITATION = HEP-PH 0608062;%%
737: 
738: %\cite{Teaney:2003kp}
739: \bibitem{Teaney:2003kp}
740:   D.~Teaney,
741:   %``Effect of shear viscosity on spectra, elliptic flow, and Hanbury
742:   %Brown-Twiss radii,''
743:   Phys.\ Rev.\  C {\bf 68}, 034913 (2003).
744:   %%CITATION = PHRVA,C68,034913;%%
745: 
746: %\cite{hardspheres}
747: \bibitem{hardspheres}
748: A.~J.~Kox, S.~R.~de Groot, W.~A.~van Leeuwen, Physica {\bf A 84}, 155
749: (1976). 
750: 
751: \bibitem{lacey}
752:   R.~A.~Lacey {\it et al.},
753:   %``Has the QCD critical point been signaled by observations at RHIC?,''
754:   Phys.\ Rev.\ Lett.\  {\bf 98}, 092301 (2007).
755:   %[arXiv:nucl-ex/0609025].
756:   %%CITATION = PRLTA,98,092301;%%
757: 
758: \bibitem{phenixQ}
759: A.~Adare {\it et al.}  [PHENIX Collaboration],
760: Phys.\ Rev.\ Lett.\  {\bf 98}, 172301 (2007).
761: %%CITATION = PRLTA,98,172301;%%
762: 
763: \bibitem{GavinAziz}
764: S.~Gavin and M.~Abdel-Aziz,
765: Phys.\ Rev.\ Lett.\  {\bf 97}, 162302 (2006).
766: %%CITATION = PRLTA,97,162302;%%
767: 
768: %\cite{Meyer:2007ic}
769: \bibitem{Meyer:2007ic}
770:   H.~B.~Meyer,
771:   %``A calculation of the shear viscosity in SU(3) gluodynamics,''
772:   arXiv:0704.1801 [hep-lat].
773:   %%CITATION = ARXIV:0704.1801;%%
774: 
775: %\cite{Muronga:2001gn}
776: \bibitem{Muronga:2001gn}
777:   A.~Muronga,
778:   %``Causal Theories of Dissipative Relativistic Fluid Dynamics for Nuclear
779:   %Collisions,''
780:   Phys.\ Rev.\  C {\bf 69}, 034903 (2004);
781:   %[arXiv:nucl-th/0309055].
782:   %%CITATION = PHRVA,C69,034903;%%
783:   %``Relativistic Dynamics of Non-ideal Fluids: Viscous and heat-conducting
784:   %fluids I. General Aspects and 3+1 Formulation for Nuclear Collisions,''
785:   preprint arXiv:nucl-th/0611090;
786:   %%CITATION = NUCL-TH/0611090;%%
787:   U.~W.~Heinz, H.~Song and A.~K.~Chaudhuri,
788:   %``Dissipative hydrodynamics for viscous relativistic fluids,''
789:   Phys.\ Rev.\ C {\bf 73}, 034904 (2006);
790:   %%CITATION = NUCL-TH 0510014;%%
791:   R.~Baier, P.~Romatschke and U.~A.~Wiedemann,
792:   %``Dissipative hydrodynamics and heavy ion collisions,''
793:   Phys.\ Rev.\ C {\bf 73}, 064903 (2006).
794:   %%CITATION = HEP-PH 0602249;%%
795: 
796: %\cite{Romatschke:2007mq}
797: \bibitem{Romatschke:2007mq}
798:   P.~Romatschke and U.~Romatschke,
799:   %``How perfect is the RHIC fluid?,''
800:   arXiv:0706.1522 [nucl-th].
801:   %%CITATION = ARXIV:0706.1522;%%
802: 
803: \end{thebibliography}
804: \end{document}