0712.3498/ll.tex
1: 
2: 
3: \documentclass[twocolumn,eqsecnum,showpacs,showkeys,amsmath,amssymb,nofootinbib,superscriptaddress,floatfix]{revtex4}
4: %\documentclass[preprint,eqsecnum,showpacs,showkeys,amsmath,amssymb,nofootinbib,superscriptaddress,floatfix]{revtex4}
5: 
6: 
7: \usepackage{graphicx}
8: \usepackage{subfigure}
9: \usepackage{bm}
10: 
11: %\usepackage{showkeys}
12: 
13: \def\vec#1{\mathchoice{\mbox{\boldmath$\displaystyle#1$}}
14: {\mbox{\boldmath$\textstyle#1$}}
15: {\mbox{\boldmath$\scriptstyle#1$}}
16: {\mbox{\boldmath$\scriptscriptstyle#1$}}}
17: \makeatletter
18: \newcommand\erfc{\mathop{\operator@font erfc}\nolimits}
19: \def\slashchar#1{\setbox0=\hbox{$#1$}
20:    \dimen0=\wd0 \setbox1=\hbox{/} \dimen1=\wd1
21:    \ifdim\dimen0>\dimen1 \rlap{\hbox to \dimen0{\hfil/\hfil}} #1
22:    \else  \rlap{\hbox to \dimen1{\hfil$#1$\hfil}} / \fi}
23: \def\p{\slashchar{p}}
24: \def\w{\omega} 
25: 
26: \makeatother
27: 
28: \begin{document}
29: \title{Viscous evolution of the rapidity distribution of matter created 
30: in relativistic heavy-ion collisions\footnote{Supported by 
31: Polish Ministry of Science and Higher Education under
32: grant N202~034~32/0918}}
33: \author{Piotr Bo\.zek}
34: \email{Piotr.Bozek@ifj.edu.pl}
35: \affiliation{Institute of Physics, Rzesz\'ow University, PL-35959 Rzesz\'ow, Poland}
36: \affiliation{The H. Niewodnicza\'nski Institute of Nuclear Physics,
37: PL-31342 Krak\'ow, Poland}
38: %
39: %\date{10 December, ver 1.0}
40: 
41: 
42: \begin{abstract}
43: Longitudinal hydrodynamic expansion of
44:   the fluid created in relativistic 
45: heavy-collisions  is considered taking into account shear viscosity.
46:  Both a on-vanishing viscosity and a soft equation
47: of state  make particle distributions in  rapidity narrower. 
48: The presence of viscosity  has dramatic 
49: consequences on the value of the initial energy density. 
50: The reduction of the longitudinal work and dissipative processes 
51: due to the  shear viscosity, 
52: increase the total entropy and the particle multiplicity at central 
53: rapidities. The total energy in the collision, dominated by the 
54: longitudinal motion, is conserved. Viscous corrections make 
55: the longitudinal velocity of the fluid  to stay close 
56: to the Bjorken scaling $v_z = z/t$ through the evolution. 
57: At the freeze-out   viscous corrections are the strongest
58:  for non-central rapidities.
59: \end{abstract}
60: 
61: \pacs{25.75.-q, 25.75.Dw, 25.75.Ld}
62: 
63: 
64: \keywords{relativistic 
65: heavy-ion collisions, hydrodynamic model, viscosity, rapidity distributions}
66: 
67: \maketitle
68: 
69: 
70: 
71: \section{Introduction}
72: Properties of  hot and dense strongly interacting  matter can be 
73: studied in  ultrarelativistic nuclear collisions. 
74: The modeling of the evolution of the dense collective  phase is most 
75: commonly undertaken within the framework of  relativistic
76:  fluid dynamics \cite{Huovinen:2003fa,Kolb:2003dz,Nonaka:2007nn}. 
77: A clear evidence of a collective behavior 
78: of the system created in a collision is given by the observation 
79: of  transverse elliptic flow  of the produced particles. Collective 
80: flow arises naturally during a hydrodynamic evolution. In
81:  ultrarelativistic heavy-ion collisions the movement of the 
82: matter at the initial stage is dominated by the expansion in the 
83: longitudinal direction. %It can be described by the
84: % simple  Bjorken scaling solution with two 
85: %assumptions. A strong correlation of the momentum and space-time 
86: %rapidities and  boost invariance in the beam direction.
87:  Most 
88: of the experimental data at the Relativistic Heavy- Ion
89:  Collider (RHIC)
90:  are restricted to the central 
91: rapidity region. Therefore, hydrodynamic models often assume a 
92: simplified geometry of the fireball with a Bjorken boost invariant 
93: flow in the longitudinal direction and concentrate on the dynamics in the
94:  transverse directions
95: with  azimuthal symmetry for central collisions
96:  \cite{Baym:1983,Rischke:1996em,Kataja:1986}
97:  or azimuthally asymmetric geometry for
98:  collisions of nuclei at non-zero impact parameters 
99: \cite{Ollitrault:1992,Kolb:2000sd,Huovinen:2001cy,Teaney:2000cw}. Only 
100: few calculations consider a fully three-dimensional evolution of 
101: the fluid \cite{Nonaka:2000ek,Hirano:2001yi,Hirano:2002ds}. 
102:  Results of the hydrodynamic evolution
103:  are sensitive to the chosen equation of state. Values of the 
104: Hanbury Brown-Twist
105:  (HBT) radii observed at RHIC \cite{Adams:2004yc} 
106: seem to exclude equations of state
107:  with a strong first order phase transition, or even with  a more prominent
108:  soft point. The hydrodynamic 
109: evolution depends on  the initial temperature 
110: and its profile, on the chosen equation of state, 
111: and on the freeze-out temperature. 
112: Particles emitted
113:  at the hydrodynamic freeze-out can still rescatter, which modifies somewhat
114:  their spectra and elliptic flow
115:  \cite{Bass:2000ib,Teaney:2000cw,Hama:2005dz,Nonaka:2006yn}. 
116: 
117: Instead of a
118:  fully three-dimensional calculation only longitudinal expansion
119:  of the matter created in a collision 
120: can be considered 
121: \cite{Baym:1983,Kajantie:1983}. 
122: Recently the rapidity distributions of pions and 
123: kaons produced in $\sqrt{s} = 200$AGeV Au-Au collisions measured by the 
124: BRAHMS Collaboration \cite{Bearden:2004yx}
125:  have been analyzed in a 1+1 dimensional
126:  hydrodynamic model \cite{Satarov:2006iw}. 
127: The final rapidity distribution of
128:  mesons has been found to be sensitive to the initial energy density
129: distribution in rapidity and to the chosen equation of state.  
130: Experimental data indicate that the boost invariant Bjorken scaling 
131: solution is not realized at RHIC energies. Results of the 1+1 
132: dimensional longitudinal fluid dynamics for the meson rapidity 
133: distributions favor a soft equation of state and require a Gaussian 
134: initial energy density distribution in the space-time
135:  rapidity \cite{Satarov:2006iw}.
136: 
137: Analysis of the elliptic flow in the momentum distribution
138:  of  particles as a function of the initial eccentricity 
139: of the source points to the possibility of a noticeable effect of 
140: the shear viscosity of the fluid  \cite{Aguiar:2001ac,Drescher:2006ca}.
141:  Quantitative studies have to 
142: take into account many effects, initial conditions, the equation of state,
143:  the freeze-out temperature, possible final rescattering,
144:  therefore estimates of
145:  the viscosity coefficient are still under debate 
146: \cite{Song:2007fn,Lacey:2007na}. 
147: Theoretical estimates of the ratio of the shear viscosity coefficient
148:  $\eta$ to the entropy density
149: $s$ range from a conjectured lower bound $\eta/s=1/4\pi$ \cite{Kovtun:2004de} 
150: to $\eta\simeq s$
151:  \cite{Venugopalan:1992hy,Arnold:2000dr}.
152: 
153: The role of the shear viscosity in the dynamics is most
154: important during the early evolution of the system, when 
155: the velocity gradients are the largest. Gradients of the Bjorken
156:  flow give rise to corrections of the pressure tensor in the fluid.
157:  The transverse pressure increases, and the fluid expands faster in 
158: the transverse directions, this leads to stronger transverse flow and 
159: also to the  saturation of the elliptic flow 
160: \cite{Song:2007fn,Teaney:2003kp,Baier:2006sr,Baier:2006gy,Chaudhuri:2006jd,Romatschke:2007jx,Hirano:2005xf}. All recent calculations
161:  using viscous relativistic hydrodynamics assume boost invariant Bjorken
162:  flow in the longitudinal direction and study the transverse 
163: development of the fluid in azimuthally symmetric or asymmetric 
164: conditions. Longitudinal pressure is reduced, and hence so is the 
165: longitudinal flow of the fluid. The fluid cools slower, at least  until 
166: substantial transverse flow builds up. Weaker longitudinal expansion
167:  and entropy production due to dissipative evolution require an 
168: adjustment of the initial entropy (temperature). As a result, the 
169: lifetimes of the plasma in the viscous and ideal fluid evolutions are 
170: similar. Finally, let us note that viscous corrections to the distribution
171:  functions at the freeze-out modify the spectra, and the HBT radii
172:  \cite{Romatschke:2007jx,Muronga:2004sf}. 
173: %Such modifications of the particle emission at the freeze-out 
174: %can be as important as the modifications due to a different 
175: %freeze-out hypersurface and 
176: %flow from  viscous dynamics.
177: 
178: Enhanced transverse pressure and the resulting modification 
179: of the spectra at central rapidity, raise the question of possible 
180: modifications of the fluid dynamics in the longitudinal direction
181:  due to viscosity. Since the boost invariant scaling solution
182:  is not applicable at RHIC energies, quantitative description of the
183:  energy flow and entropy production in the fireball should take into
184:  account a fully three dimensional geometry. Such a task within the 
185: dissipative fluid dynamics has not been accomplished yet. In the following
186:  we consider a simpler problem of the evolution of a longitudinally 
187: expanding non-boost invariant fluid with viscosity. One expects that
188:  reduced longitudinal work in a viscous fluid generates narrower particle 
189: distributions in rapidity. The author is aware of only one, two-decades 
190: old work, considering this problem \cite{chu:1986}, where first-order viscous
191:  hydrodynamics has been applied.
192:  At lower energies no strong effect
193:  of viscosity on the longitudinal expansion has been observed \cite{chu:1986}.
194: 
195:  In
196:  the following we use BRAHMS data \cite{Bearden:2004yx}
197:  to constraint meson rapidity
198:  distributions after freeze-out. Solving viscous hydrodynamics in the 1+1
199:  longitudinal geometry, we find a significant reduction of the initial 
200: energy density when viscosity is taken into account. Also the longitudinal
201:  flow is modified, viscosity reduces the flow and counteracts the 
202: acceleration due rapidity gradients of the  pressure.
203: At the Large Hadron Collider (LHC) energies
204:  we find that the modification of the 
205: longitudinal dynamics
206: due to shear viscosity leads to an increase of the rapidity range where
207:  the Bjorken scaling flow applies.
208: 
209: \section{Longitudinal hydrodynamic equations with shear viscosity}
210: 
211: We consider a baryon-free fluid with non-zero shear viscosity. 
212: The energy-momentum tensor is the sum of the ideal fluid component 
213: and the  shear tensor $\pi^{\mu \nu}$
214: \begin{equation}
215: T^{\mu\nu}=(\epsilon+p)u^\mu u^\nu -g^{\mu \nu}+\pi^{\mu\nu} \ ,
216: \label{eq:emtensor}
217: \end{equation}
218: $\epsilon$ and $p$ are the local energy density and pressure of the fluid, and
219: \begin{eqnarray}
220: u^\mu &= & \gamma(1,0,0,v) \nonumber \\
221: & =& (\cosh Y ,0,0, \sinh Y) 
222: \end{eqnarray}
223: is the four-velocity of the fluid element ($\gamma=1/\sqrt{1-v^2}$),
224:  and $Y=\frac{1}{2}\ln \left(\frac{1+v}{1-v}\right)$ is its rapidity.
225: The energy density and the pressure are related by the 
226: equation of state.
227: The energy density $\epsilon(t,z)$
228:  and the longitudinal velocity component $v(t, z)$ 
229: are functions of time $t$ and the beam
230: axis coordinate $z$. Instead of the time  and the $z$ coordinate it  is
231: preferable to use the proper time $\tau=\sqrt{t^2-z^2}$
232:  and the space-time rapidity
233: \begin{equation}
234: \theta=\frac{1}{2}\ln\left( \frac{t+z}{t-z}\right) \ .
235: \end{equation}
236: Hydrodynamic equations $\partial_\mu T^{\mu \nu} =0 $ can be written as
237: \cite{Muronga:2006zw,Baier:2006um}
238: \begin{equation}(\epsilon+p) Du^\mu= 
239: \nabla^\mu p -\Delta^\mu_\nu \nabla_\alpha 
240: \pi^{\nu\alpha}+\pi^{\mu\nu}Du_\nu 
241: \label{eq:vh1}
242: \end{equation}
243: and
244: \begin{equation}
245: D\epsilon= -(\epsilon+p) \nabla_\mu u^\mu +\frac{1}{2} \pi^{\mu\nu} 
246:  < \nabla_\mu u_\nu > \ ,
247: \label{eq:vh2}
248: \end{equation}
249: where
250: \begin{equation}
251: D=u^\mu\partial_\mu=\cosh(Y-\theta)\partial_\tau
252: +\frac{\sinh(Y-\theta)}{\tau}\partial_\theta \ ,
253: \end{equation}
254: \begin{equation}
255: < \nabla_\mu u_\nu >=
256: \nabla_\mu u_\nu+\nabla_\nu u_\mu-
257: \frac{2}{3}\Delta_{\mu\nu}\nabla_\alpha u^\alpha \ ,
258: \end{equation}
259: \begin{eqnarray}
260: \nabla^\mu&=&\Delta^{\mu\nu}\partial_\nu\nonumber \\
261: & =& (-\sinh Y {\cal K},-\partial_x,-\partial_y,-\cosh Y{\cal K})
262: \end{eqnarray}
263: with 
264: $\Delta_{\mu\nu}=g_{\mu\nu}-u_\mu u_\nu$ and
265: \begin{equation}
266: {\cal K}=\sinh(Y-\theta)\partial_\tau+\frac{\cosh(Y-\theta)}
267: {\tau}\partial_\theta \ .
268: \end{equation}
269: The equations of the  second order viscous hydrodynamics
270:   (\ref{eq:vh1}) and (\ref{eq:vh2}) are supplemented with a
271: dynamic equation for the stress tensor 
272: \cite{Baier:2006sr,Muronga:2004sf,Muronga:2001zk,IS,Baier:2006um,Muronga:2003ta}
273: \begin{equation}
274: \tau_\pi \Delta^\mu_\alpha\Delta^\nu_\beta \pi^{\alpha\beta}+\pi^{\mu\nu}=
275: \eta < \nabla^\mu u^\nu >-2\tau_\pi \pi^{\alpha(\mu}\omega^{\nu)}_\alpha
276: \end{equation}
277: $\eta$ is the shear viscosity coefficient and $\tau_\pi$ is 
278: the relaxation time of the stress tensor.
279:  $\omega^{\mu\nu}=\Delta^{\mu\alpha}\Delta^{\nu\beta}
280: \left(\partial_\alpha u_\beta-\partial_\beta u_\alpha\right)$ 
281: is the vorticity of the fluid; it is zero for the longitudinal 
282: flow considered here. The relaxation time and the 
283: viscosity coefficient can be estimated from microscopic models,
284:  considering equilibration processes 
285: \cite{Kovtun:2004de,Venugopalan:1992hy,Arnold:2000dr,Muronga:2007qf}. 
286: In this work we 
287: take several values for the ratio of the viscosity coefficient 
288: to the entropy $\eta/s$, and we drop %(for numerical reasons)
289:  viscosity effects for temperatures
290: below $130$MeV, the relaxation time is \cite{Venugopalan:1992hy}
291: $\tau_\pi/\eta = 6/T s $, unless specified otherwise 
292: ($T$ is the local temperature).
293:  For short  relaxation times  the stress tensor 
294: relaxes fast and stays close to the Navier-Stokes
295:  value $\pi^{\mu\nu}=\eta<\nabla^\mu u^\nu>$. 
296: The dependence on the initial value of the stress tensor and 
297: on the relaxation time is discussed in Section \ref{sec:stressini}.
298: 
299: For a fluid expanding only in the longitudinal direction, 
300: with the energy density and the velocity constant in the transverse plane, 
301: the stress tensor can be written using one scalar function $\Pi$
302: \begin{equation}
303: \pi^{\mu\nu}=\left(\begin{array}{cccc} -\sinh^2 Y & 0 & 0 & -\sinh Y \cosh Y\\
304: 0 & \frac{1}{2} & 0 & 0 \\
305: 0 & 0 & \frac{1}{2} & 0 \\
306: -\sinh Y \cosh Y & 0 & 0 & -\cosh^2 Y
307: \end{array}\right)\Pi
308: \end{equation}
309: and viscous  hydrodynamic equations take the form
310: \begin{eqnarray}
311: (\epsilon+p)DY&=& -{\cal K}p+\Pi DY + {\cal K} \Pi \nonumber \\
312: D\epsilon&=& (\epsilon +p) {\cal K} Y -\Pi {\cal K} Y\nonumber \\
313: D\Pi&=& (\frac{4}{3}\eta {\cal K}Y-\Pi)/\tau_\pi  \ .
314: \label{eq:eqsolv}
315: \end{eqnarray}
316: The above equations are solved numerically in the
317:  $\tau$-$\theta$  plane, starting from
318:  some energy density $\epsilon(\tau_0,\theta)$ 
319: at the initial proper time of the evolution  $\tau_0=1.0$fm/c. 
320: For the initial fluid rapidity we always take the Bjorken flow
321: \begin{equation}
322: Y(\tau_0,\theta)=\theta \ .
323: \label{eq:flowini}
324: \end{equation}
325: Assuming boost invariance, i.e.
326: \begin{equation}
327: Y(\tau,\theta)=\theta \ , \ \ \epsilon(\tau,\theta)=\epsilon(\tau) \ , \ \
328: p(\tau,\theta)=p(\tau)
329: \end{equation}
330: hydrodynamic equations simplify to \cite{Muronga:2001zk,Baier:2006um}
331: \begin{eqnarray}
332: \frac{d\epsilon}{d\tau}&=& -\frac{\epsilon+p-\Pi}{\tau} \nonumber \\
333: \frac{d\Pi}{d\tau}&=& \frac{(4\eta)/(3\tau)-\Pi}{\tau_\pi} \ .
334: \label{eq:bjorkenvis}
335: \end{eqnarray}
336: 
337: \section{Equation of state}
338: \label{sec:eos}
339: 
340:  The equation of state determines the evolution of the fireball. 
341: In the following we take a parameterization of the equation 
342: of state proposed by Chojnacki and Florkowski \cite{Chojnacki:2007jc}. 
343: It is an interpolation of the lattice data at temperatures above 
344: $T_c = 170$MeV and of an equation of state of noninteracting hadrons
345:  at lower temperatures. The two limiting formulas are joined 
346: smoothly with only a slight softening of the equation of state around
347:  the critical temperature (Fig. \ref{fig:cs}). 
348: This minimally softened, realistic equation 
349: of state has shown itself suitable for the description of the transverse 
350: expansion of the fluid and of the build up of the elliptic flow and gives  
351: reasonable HBT radii
352:  \cite{Chojnacki:2007jc,Chojnacki:2007rq}.  
353: 
354: In this section we consider the hydrodynamic 
355: longitudinal expansion of an ideal fluid. Equations for the 
356: fluid rapidity and energy density are obtained from Eqs. (\ref{eq:eqsolv})
357:  setting 
358: $\Pi  = 0$. 
359: This problem has been discussed recently in the 
360: context of heavy-ion collisions at RHIC \cite{Satarov:2006iw}.
361:  For completeness
362:  we study the effect of the equation of state we use on the
363:  longitudinal expansion. For that purpose we compare the evolution
364:  using the equation of state of Chojnacki and Florkowski with an evolution
365: based on  a 
366: relativistic gas equation of state $p=\frac{1}{3}\epsilon$.
367: 
368: 
369: 
370: \begin{figure}
371: \includegraphics[width=.5\textwidth]{figcs2.eps}
372: \caption{Square of the velocity of sound as a function of the 
373: temperature for an equation of state interpolating between the
374:  hadron gas and the quark-gluon plasma expressions \cite{Chojnacki:2007jc}.}
375: \label{fig:cs}
376: \end{figure}
377: 
378: 
379: 
380: 
381: At the initial time
382: $\tau_0= 1.0$fm/c the energy density is
383: \begin{equation}
384: \epsilon(\tau_0,\theta)=\epsilon_0 \exp\left(-\theta^2/(2\sigma^2)\right) \ . 
385: \label{eq:eini}
386: \end{equation} 
387: The initial energy density $\epsilon_0$
388:  and the width of the 
389: initial rapidity distribution $\sigma$ are parameters adjusted to 
390: reproduce final meson rapidity distributions. 
391: The freeze-out takes place at the temperature $T_f = 165$MeV, this high 
392: freeze-out temperature is the same as the chemical freeze-out temperature
393:  \cite{BraunMunzinger:2001ip,Florkowski:2001fp}. The spectrum of 
394: particles emitted with four-momentum $q^\mu=(E,q)$ 
395:  is given by the Cooper-Frye formula \cite{Cooper:1974mv,Satarov:2006iw}
396: \begin{equation}
397: E\frac{d^3N}{d^3q}=\frac{d^3N}{dy d^2q_\perp}=\frac{1}{(2\pi)^3}\int
398: d\Sigma_\mu q^\mu f(q^\mu u_\mu) \ ,
399: \label{eq:cff}
400: \end{equation}
401: $f(E)=e^{-E/T_f}$ is the thermal distribution (Boltzmann 
402: distribution for simplicity), and $y$ denotes the rapidity of the 
403: emitted particle.
404:  The element of the hypersurface of 
405: constant freeze-out temperature is 
406: $d\Sigma^\mu = S\ (\tau^{'}(\theta)\sinh\theta +\tau(\theta)\cosh\theta,
407: 0,0,\tau^{'}(\theta)\cosh\theta -\tau(\theta)\sinh\theta)$, $S = \pi R_{Au}^2$ 
408: is the transverse area of the fireball in central collisions
409: and $\tau(\theta)$ is the line of constant temperature
410:  $T_f$ in the $\tau$-$\theta$ plane. Particle distributions
411:  in rapidity are obtained by integration over the  transverse momenta
412: $q_\perp$ in Eq. (\ref{eq:cff})
413: \begin{eqnarray}
414: \frac{dN}{dy}&=& \frac{S}{4\pi^2}\int_{-\theta_{max}}^{\theta_{max}} 
415: \left( \tau(\theta) \cosh(y-\theta)-\tau^{'}(\theta)\sinh(y-\theta)\right)
416: \nonumber \\ & & 
417: (2 m \xi +2 \xi^2 +m^2)\xi \nonumber \\ & & 
418: \exp\left( -\frac{m\cosh(y-Y_f(\theta))}{T_f}  \right) d\theta \ , 
419: \label{eq:dndyfree}
420: \end{eqnarray}
421: $m$ is the meson mass, $Y_f(\theta)=Y(\tau(\theta),\theta)$ is 
422: the fluid rapidity at the freeze-out hypersurface, and
423: \begin{equation}
424: \xi=\frac{T_f}{\cosh(y-Y_f(\theta))} \ .
425: \end{equation}
426: 
427: The above expression neglects the transverse expansion of the fluid 
428: at the freeze-out. This has only a small effect on rapidity distributions,
429:  the distribution should be narrower. Pions 
430: and kaons come to a large extend from secondary decays of resonances. The 
431: emission takes place in two stages, first an emission of a heavy
432:  resonance according to Eq. (\ref{eq:dndyfree}) and then the decay 
433: of the resonance into pions 
434: (kaons). The emission of resonances and their decay is also 
435: influenced by their transverse expansion. A hint on the 
436: spread in rapidity of the decay products of resonances is given
437:  by charge balance correlations \cite{Adams:2003kg,Bozek:2003qi,Cheng:2004zy}.
438:  Narrow charge balance 
439: functions indicate that  decay products of a resonance are only $0.5$ 
440: unit of rapidity away from the parent resonance, convoluting this 
441: distribution with the spread in rapidity of the emitted  resonances one obtains
442:  a distribution of half-width similar as for the emission of direct
443:  pions in Eq. (\ref{eq:dndyfree}).
444:  Since 75\% of pions come from resonances at
445: $T_f = 165$MeV \cite{Torrieri:2004zz},
446:  we  multiply the distribution from (\ref{eq:dndyfree}) 
447: by a factor 4 to account for all pions, direct
448: and from resonance decays (the factor is 1.7 for kaons).
449: \begin{figure}
450: \includegraphics[width=.5\textwidth]{dndyfree.eps}
451: \caption{(color online) Rapidity distribution 
452: of mesons calculated using a realistic equation of 
453: state (solid line) and using a relativistic gas
454:  equation of state for three different  initial
455:  conditions (dashed, dotted and dashed-dotted lines), Table I. 
456: Data are from the BRAHMS Collaboration \cite{Bearden:2004yx}.}
457: \label{fig:dndyfree}
458: \end{figure}
459: 
460: \begin{table}
461: \begin{tabular}{|l|c|c|c|} \hline 
462: & $\epsilon_0$ & $\sigma$& $\tau(0)$ \\
463: & GeV/fm$^3$ & & fm/c \\  \hline 
464: EOS \cite{Chojnacki:2007jc}& 16.9 & 1.05 & 14.8 \\
465: $p=\frac{1}{3}\epsilon$  case I & 71.5 & 1.05 & 14.8 \\
466: $p=\frac{1}{3}\epsilon$  case II & 102 & 0.8 & 15.7 \\
467: $p=\frac{1}{3}\epsilon$  case III & 50.8  & 1.5 & 13.7 
468:  \\
469: \hline
470: \end{tabular}
471: \caption{Parameters of the initial energy density distribution (\ref{eq:eini}),
472: for ideal fluid calculations, one 
473: using a realistic equation of state 
474: and three calculations using a relativistic gas equation of state. 
475: In the last column is shown the lifetime of the system until freeze-out.}
476: \label{table1}
477: \end{table}
478: \begin{figure}
479: \includegraphics[width=.5\textwidth]{flowfree.eps}
480: \caption{(color online) Difference between the  flow rapidity
481:  of the fluid and the Bjorken value, calculated using a realistic 
482: equation of state (solid line) and using a relativistic gas
483:  equation of state for three different  initial conditions 
484: (dashed, dotted and dashed-dotted lines), Table \ref{table1}.}
485: \label{fig:flowfree}
486: \end{figure}
487: 
488: The parameters $\epsilon_0$  and $\sigma$
489:  have been adjusted for the calculation using a realistic
490:  sound velocity (Fig. \ref{fig:cs}) to reproduce the 
491: width and normalization of the observed pion distribution. 
492: The resulting meson distributions are similar as the ones  
493: observed experimentally. 
494: When using a relativistic gas equation 
495: of state ($p=\frac{1}{3}\epsilon$) 
496: one always gets a meson distribution in rapidity that
497:  is too wide. We present three calculations with different 
498: initial widths $\sigma$ and with the initial energy densities
499:  $\epsilon_0$ adjusted to reproduce
500: $dN/dy$ for central rapidities only (Fig. \ref{fig:dndyfree}) . 
501: The final meson distribution is much wider
502:  than the initial energy density distribution in all cases. 
503: It is due to the
504: breaking of the Bjorken scaling of the longitudinal
505:  flow, $Y(\tau,\theta)>\theta$  (Fig. \ref{fig:flowfree}). 
506: Fast moving fluid elements emit
507: mesons in the far forward and backward rapidities. Concluding this section,
508:  we confirm the findings of Ref. \cite{Satarov:2006iw}. 
509: A hard equation of state never works; a narrow 
510: initial distribution of the energy density in rapidity leads to a strong 
511: acceleration of the longitudinal flow, wider initial distributions 
512:  are incompatible with  the narrow final meson distributions. Only 
513: by imposing a softened equation of state,
514:  experimental pion and kaon distributions
515:  can be approximately reproduced. In the following we take the realistic 
516: equation 
517: of state from Ref. \cite{Chojnacki:2007jc} and study 
518: the effect of non-negligible shear viscosity.
519: 
520: 
521: 
522: \section{Dissipative longitudinal expansion}
523: 
524: \begin{figure}
525: \includegraphics[width=.5\textwidth]{dndy.eps}
526: \caption{(color online) Rapidity distribution of mesons calculated 
527: using a realistic equation of state and viscosity $\eta/s = 0.2$
528:  (solid line) and using ideal fluid hydrodynamics (dashed line).
529:  The dotted line denotes the results of a viscous hydrodynamic evolution,
530:  but neglecting the viscous corrections to the particle emission at 
531: freeze-out (Eq. \ref{eq:dndydissi}). Data are from the BRAHMS Collaboration 
532: \cite{Bearden:2004yx}.}
533: \label{fig:dndy}
534: \end{figure}
535: 
536: Following the hydrodynamic evolution with viscosity requires the solution
537:  of the  coupled equations (\ref{eq:eqsolv}), for the fluid rapidity $Y$, 
538: the viscous correction $\Pi$,  and the energy density $\epsilon$. 
539: The initial conditions are given in Eqs. 
540: (\ref{eq:eini}) and (\ref{eq:flowini}). For 
541: the initial 
542: viscous corrections we take $\Pi (\tau_0,\theta)=p(\tau_0,\theta)$.
543:  This means that at the initial time the effective 
544: pressure is maximally anisotropic
545: in the fluid rest frame, with zero longitudinal pressure. 
546: We vary the 
547: value of the shear viscosity coefficient $\eta/s$,
548: and in each case we adjust the parameters of the initial energy
549:  density distribution (\ref{eq:eini}) to reproduce BRAHMS data for the pion 
550: distribution in rapidity. Viscous corrections modify the momentum 
551: distribution functions in the fluid. We assume the same modification 
552: for all the particle species in the fluid
553: \begin{equation}
554: f(q)+\delta f(q)=f(q)\left(1+\frac{q_\mu q_\nu \pi^{\mu\nu}}{2T^2 (\epsilon+p)}
555: \right) \ ,
556: \end{equation}
557: \begin{figure}
558: \includegraphics[width=.5\textwidth]{visc.eps}
559: \caption{Relative viscous corrections $\Pi/(\epsilon+p)$
560: at the freeze-out as a function of space-time rapidity (Eq. \ref{eq:visc}).}
561: \label{fig:visc}
562: \end{figure}
563: The correction to the particle distribution function is proportional
564:  to the ratio of the viscous correction $\Pi$  to the enthalpy
565: $\epsilon+p$. Viscous corrections to the 
566: distribution functions at the freeze-out modify the Cooper-Frye formula
567: \begin{equation}
568: \frac{dN_{visc}}{dy}=\frac{dN}{dy}+\frac{d\delta N}{dy} \ \ .
569: \label{eq:dndy}
570: \end{equation}
571: To the expression (\ref{eq:dndyfree}) one has to add
572: \begin{eqnarray}
573: \frac{d\delta N}{dy}& &= \frac{S}{4\pi^2}\int_{-\theta_{max}}^{\theta_{max}} 
574: \left( \tau(\theta) \cosh(y-\theta)-\tau^{'}(\theta)\sinh(y-\theta)\right)
575: \nonumber \\ & & 
576: \left[12 \xi^5 +5\xi^3m^2+12 \xi^4 m + \xi^2 m^3 -\sinh(y-Y_f(\theta)) 
577: \right. \nonumber \\ & &  \left.
578: (24 \xi^5 +12 \xi^3 m^2 +24\xi^4 m +4 \xi^2 m^3 +\xi m^4) \right]
579: \nonumber \\ & &
580: \frac{\Pi}{2T^2 (\epsilon+p)}
581: \exp\left( -\frac{m\cosh(y-Y_f(\theta))}{T_f}  \right) d\theta \ , 
582: \label{eq:dndydissi}
583: \end{eqnarray}
584: In Fig. \ref{fig:dndy}
585:  is shown the result of this procedure for the case
586: $\eta/s = 0.2$. Again the direct meson spectra are multiplied by a factor $4$
587:  for pions and $1.7$ for kaons, to account for the expected ratio of 
588:  all  mesons to
589: directly produced mesons  \cite{Torrieri:2004zz},
590:  at the chosen freeze-out temperature. Pion emission at 
591: the end of the viscous hydrodynamic evolution (solid line) is similar 
592: as observed experimentally. In Fig. \ref{fig:dndy}
593:  is also shown the meson distribution obtained using the equilibrium 
594: distribution at freeze-out (Eq. \ref{eq:dndyfree}) (dotted line). 
595: Deviations from the full result (Eq. \ref{eq:dndy}) is only noticeable 
596: at rapidities $3$ unit away from central rapidity. 
597: It can be understood as due to an  earlier freeze-out at large 
598: rapidities, which makes the relative viscous corrections
599: \begin{equation}
600: \frac{\Pi(\tau(\theta),\theta)}{\epsilon(\tau(\theta),\theta)
601: +p(\tau(\theta),\theta)}
602: \label{eq:visc}
603: \end{equation}
604: at freeze-out more important  (Fig. \ref{fig:visc}). 
605: Accordingly one expects the largest correction from viscosity 
606: to particle spectra, elliptic flow, and HBT radii at large rapidities.
607: 
608: \begin{figure}
609: \includegraphics[width=.5\textwidth]{flow.eps}
610: \caption{(color online)Difference between the  flow rapidity 
611: of the fluid and the Bjorken value, calculated for an evolution with 
612: shear viscosity coefficient $\eta/s = 0.2$ (solid lines), for an ideal
613:  fluid with a realistic equation of state (dashed lines) and using a
614:  relativistic gas equation of state (case I) (dashed-dotted lines).}
615: \label{fig:flow}
616: \end{figure}
617: \begin{table}
618: \begin{tabular}{|l|c|c|c|} \hline 
619: $\eta/s$ & $\epsilon_0$ & $\sigma$& $\tau(0)$ \\
620: & GeV/fm$^3$ & & fm/c \\  \hline 
621: 0 & 16.9 & 1.05 & 14.8 \\
622: 0.1 & 9.8 & 1.18 & 14.1 \\
623: 0.2 & 5.6 & 1.8 & 13.1 \\
624: 0.3 & 4.0 & 1.86 & 12.4 \\
625: \hline
626: \end{tabular}
627: \caption{Parameters of the initial energy density distribution (\ref{eq:eini}) 
628: for hydrodynamic calculations with several value of the shear viscosity
629:  coefficient. In the last column is show the lifetime of the system 
630: until freeze-out.}
631: \label{table2}
632: \end{table}
633: \begin{figure}
634: \includegraphics[width=.5\textwidth]{flowfreeze.eps}
635: \caption{(color online)Difference between the  flow rapidity 
636: of the fluid and the Bjorken value at the freeze-out hypersurface, 
637: calculated for an evolution with 
638: shear viscosity coefficient $\eta/s = 0.2$ (solid line), for an ideal
639:  fluid with a realistic equation of state (dashed line) and using a
640:  relativistic gas equation of state (case I) (dashed-dotted line).}
641: \label{fig:flowfreeze}
642: \end{figure}
643: \begin{figure}
644: \includegraphics[width=.5\textwidth]{eini.eps}
645: \caption{(color online) Initial energy density distribution for 
646: the ideal fluid hydrodynamic evolution with a realistic equation of
647:  state (dashed line), for  viscous hydrodynamic evolutions 
648: (solid lines), and for a relativistic gas equation of state (case I) 
649: (dashed-dotted line).}
650: \label{fig:eini}
651: \end{figure}
652: We have noticed (Section \ref{sec:eos}) that longitudinal gradients of the 
653: pressure 
654:  cause the 
655: acceleration of the motion of the fluid in the beam direction. 
656: On the other hand, 
657: viscosity reduces the longitudinal motion of the fluid, and as a 
658: consequence reduces its expansion in the longitudinal direction. 
659: Shear viscosity prevents large gradients of the velocity field to develop. 
660: One can compare the rapidity of the fluid with and without shear 
661: viscosity (Fig. \ref{fig:flow}). 
662: For $\eta/s = 0.2$,
663:  the flow stays close to the Bjorken one during a substantial part 
664: of the evolution. 
665: At the freeze-out hypersurface the flow is still 
666: Bjorken-like for $|\theta|<1.8$
667: for $\eta/s=0.2$ (Fig. \ref{fig:flowfreeze}). The acceleration from 
668: pressure gradients and the deceleration from viscosity approximately
669:  cancel for this choice of parameters.
670: Reduced longitudinal expansion with viscosity requires smaller
671:  initial energy densities to reproduce the final meson distributions.
672:  In Table \ref{table2}
673:  are listed the initial energy densities and widths of rapidity 
674: distributions adjusted to reproduce the observed meson distributions
675:  for several values of the shear viscosity coefficient. The shape of 
676: the initial energy density is extremely sensitive to the dynamics of the 
677: longitudinal expansion (Fig. \ref{fig:eini}). Obviously the value 
678: of the initial 
679: energy density, or in other words the cooling rate from the longitudinal 
680: motion, is an important ingredient in the modeling of the transverse 
681: expansion of the fluid. Let us also note, that 
682: when reduced initial energy densities are imposed, 
683: the lifetime of the system until freeze-out is only weakly dependent 
684: on the viscosity coefficient.
685: 
686: 
687: 
688: 
689: 
690: \section{Cooling and entropy production}
691: 
692: It is instructive to compare the cooling rate in our 
693: solution and in the boost invariant scaling solution. Finite
694:  extension in space-time rapidity makes the cooling rate faster.
695:  On the other hand, reduced velocity of sound and shear viscosity
696:  reduce the longitudinal work of the pressure, and slow down the cooling. 
697: In Fig. \ref{fig:cool} is compared the cooling  of the central 
698: region of a finite system to the 
699: cooling in the boost invariant case (Eq. \ref{eq:bjorkenvis}). 
700: To compare the solution in the 1+1 dimensional system to the boost-invariant 
701: one, the initial temperatures are fixed so 
702: as to give the same lifetimes of the two systems until freeze-out.
703: %In order to 
704: %realistically describe the cooling rate one has to take into account
705: % both the shear viscosity and the lack of the boost invariance of the 
706: %system.
707:  Boost invariant solutions underestimate the values of the initial
708:  temperature (energy density) and of the cooling rate 
709: for an ideal fluid evolution.
710: In a finite system additional cooling appears and the longitudinal flow 
711: is stronger than the Bjorken scaling flow. As the shear viscosity coefficient 
712: increases the velocity gradients in the dynamics are more and more constrained.
713:  The flow accelerates less. At $\eta/s=0.2$ the effects of the viscosity 
714: and space-time rapidity gradients of the pressure 
715: counterbalance each other and the flow is
716:  approximately Bjorken-like. With stronger viscosity $\eta/s=0.3$, 
717: the Bjorken flow 
718: is decelerated and the cooling is slower than for the boost-invariant solution.
719: It means that the gradients of the viscous correction $\Pi$ are larger 
720: than the gradients of the pressures itself and  applicability of second order
721:  viscous hydrodynamics is questionable.
722: 
723:  \begin{figure}
724: \includegraphics[width=.5\textwidth]{tempcool.eps}
725: \caption{(color online) Time dependence of the temperature at 
726: the center for a longitudinally expanding  ideal fluid fireball
727:  (dashed line) compared to 
728: the boost-invariant solution (dashed-dotted line). Same for the 
729: evolution with viscosities $\eta/s = 0.1,\ 0.2,\ 0.3$ (solid lines,
730:  increasing $\eta$ from top to bottom), 
731: and for the 
732: boost-invariant case with viscosity (dotted lines). For $\eta/s=0.2$ the
733:  solid and doted curves lie on top of each other.}
734: \label{fig:cool}
735: \end{figure}
736: Dissipative hydrodynamics conserves the total
737:  energy but produces entropy. The expression for the total energy of 
738: the system at proper time $\tau$ is
739: \begin{eqnarray}
740: E(\tau)&=& \tau S \int_{-\infty}^\infty d\theta \left[\epsilon 
741:  \cosh Y \cosh(Y-\theta)
742: \right. \nonumber \\
743: &+& \left. (p-\Pi)\sinh Y \sinh(Y-\theta)\right] \ .
744: \label{eq:etot}
745: \end{eqnarray}
746:  It is 
747: mainly composed of the kinetic energy of the longitudinal motion
748:  of the fluid. Small changes of the energy density at large rapidities
749:  can cause significant changes of the energy density for central rapidities. 
750: Softening of the equation of state and non-zero 
751: shear viscosity modify the dynamics 
752: at large rapidities, which leads to less cooling at $\theta= 0$, 
753: while the global
754:  energy of the fireball is unchanged. Dissipative processes
755:  driving the system locally towards equilibrium produce entropy 
756: \cite{IS,Muronga:2003ta,Muronga:2007qf,Elze:2001ss,Dumitru:2007qr}. 
757: The total entropy of the system
758: \begin{figure}
759: \includegraphics[width=.5\textwidth]{ent.eps}
760: \caption{(color online) Relative entropy production in the 
761: viscous hydrodynamic evolution (Eq. \ref{eq:stot}) (solid line), 
762: of the entropy density at central space-time rapidity (dashed line),
763:  and of the entropy 
764: from the boost-invariant Bjorken solution (Eq. \ref{eq:sbj})
765:  (dashed-dotted line). All calculation with $\eta/s=0.2$.}
766: \label{fig:ent}
767: \end{figure}
768: \begin{equation}
769: S(\tau)= \tau S \int_{-\infty}^\infty d\theta s \cosh(Y-\theta)
770: \label{eq:stot}
771: \end{equation}
772:  increases with time, if shear viscosity is active (Fig. \ref{fig:ent}). 
773: %Since the distribution of matter in space-time rapidity widens during 
774: %the hydrodynamic evolution, the entropy density for central rapidity
775: %$\tau s(\tau,\theta)$ increases slower than the 
776: %total entropy in the collision. 
777: The increase of the entropy at 
778: central rapidity $\tau s(\tau,\theta)$
779: is responsible for an increase of the particle 
780: multiplicity in the central rapidity region. Estimates of the 
781: entropy production \cite{Dumitru:2007qr} based on the 
782: boost-invariant solution (Eq. \ref{eq:bjorkenvis})
783: \begin{equation}
784: \frac{d(\tau s)}{d\tau}=\frac{\Pi}{T}
785: \label{eq:sbj}
786: \end{equation}
787: are very close to the 1+1 dimensional
788: dynamical result,
789:  it happens if the  shear viscosity is strong enough  to conserve the 
790:  Bjorken flow in the evolution.
791: 
792: 
793: 
794: 
795: \section{Role of the initial stress tensor}
796: \label{sec:stressini}
797: 
798: \begin{figure}
799: \includegraphics[width=.5\textwidth]{shear.eps}
800: \caption{(color online) Ratio of the dynamical shear viscosity correction 
801: $\Pi(\tau)$
802:  to the Navier-Stokes  value $\Pi_{NS}(\tau)$  at 
803: $\eta/s = 0.2$  for two initial condition for 
804: the viscous corrections $\Pi(\tau_0)=p(\tau_0)$ (solid line) and  for 
805:  $\Pi(\tau_0)=0$ (Eq. \ref{eq:pins}) (dashed-dotted line), and 
806: for a reduced relaxation time $\tau_\pi=1.5 \eta/T s$ (dashed  line).}
807: \label{fig:shear}
808: \end{figure}
809: 
810: Second order viscous hydrodynamics introduces a dynamical equation for the 
811: stress tensor. A crucial parameter is given by the relaxation time 
812: $\tau_\pi$. For small relaxation times the viscosity correction stays close 
813: to the Navier-Stokes value
814: \begin{equation}
815: \Pi_{NS}(\tau)=\frac{4 \eta}{3 \tau} \ .
816: \label{eq:pins}
817: \end{equation}
818: % which is given by the instantaneous velocity 
819: %gradients in the system. 
820: %For the longitudinal expansion it is
821: Such a behavior of  viscous corrections  has been 
822: confirmed in numerical simulations of the  transverse expansion in
823:  viscous hydrodynamics with small relaxation times 
824: \cite{Romatschke:2007jx,Song:2007fn},
825:  unlike in Ref. \cite{Bozek:2007di} where a large value of the relaxation
826:  time was postulated.
827: We check the effect of the initial value of $\Pi(\tau)$ on the evolution
828: by comparing two scenarios, $\Pi(\tau_0)=p(\tau_0)$ which corresponds to
829: an anisotropic momentum (pressure) at the initial time and $\Pi(\tau_0)=0$
830:  which corresponds to initially locally equilibrated distributions.
831: With the choice of the the relaxation time $\tau_\pi=6 \eta/T s$ 
832: the two initial conditions lead to different results. As before, 
833: for each choice of the initial conditions and parameters we readjust the
834:  initial energy density distribution (\ref{eq:eini}) to reproduce the final 
835: pion rapidity distribution. 
836: 
837: \begin{figure}
838: \includegraphics[width=.5\textwidth]{entini.eps}
839: \caption{(color online) Relative entropy increase in the dissipative evolution
840:  at 
841: $\eta/s = 0.2$  for two initial condition for 
842: the viscous corrections $\Pi(\tau_0)=p(\tau_0)$ (solid line) and for 
843:  $\Pi(\tau_0)=0$ (Eq. \ref{eq:pins}) (dashed-dotted line), and 
844: for a reduced relaxation time $\tau_\pi=1.5 \eta/T s$ (dashed line).}
845: \label{fig:entl}
846: \end{figure}
847: 
848: In Fig. \ref{fig:shear} is shown the ratio of
849:  the dynamical value of the viscous correction  $\Pi(\tau)$ to the 
850: Navier-Stokes value (\ref{eq:pins}). After several fm/c the 
851: stress tensor, that is set initially to 
852: zero (dashed-dotted line),
853:   relaxes to and overshoots the steady flow value (\ref{eq:pins}). 
854: For an initial stress tensor corresponding to  the anisotropic effective 
855: pressure (solid line) the 
856: viscous correction overshoots the Navier-Stokes value  almost immediately. 
857: Since the dissipative effects are the strongest at the early stages, 
858: the integrated entropy production is smaller for an evolution
859:  with initial zero stress tensor 
860: than for an evolution 
861: starting with nonzero stress tensor (Fig. \ref{fig:entl}). Even though,
862: the effect of the change of the initial viscosity correction can be 
863: counterbalanced by  a suitable change in the initial energy density
864:  distribution,  the amount of the entropy produced in the dynamics 
865: depends on the initial value  of the dissipative corrections.
866: We have also performed a calculation with a smaller relaxation time 
867: (dashed lines in Fig. \ref{fig:shear} and \ref{fig:entl}). In that 
868: case the initial value of $\Pi(\tau)$ is less important, as it rapidly 
869: relaxes to the the steady-flow value $\Pi_{NS}(\tau)$. The integrated 
870: dissipative effects are smaller.
871: 
872: 
873: 
874: 
875: \section{Expectations for LHC}
876: 
877: \begin{figure}
878: \includegraphics[width=.5\textwidth]{dndyl.eps}
879: \caption{(color online) Rapidity distribution of pions (upper curves) and kaons
880: (lower curves, $\times 3$) for LHC  plateau-like initial conditions
881: (Eq. \ref{eq:einilhc}), 
882:   calculated for an evolution with 
883: shear viscosity coefficient $\eta/s = 0.2$ (solid lines) and for an ideal
884:  fluid with a realistic equation of state (dashed lines).  The dotted line 
885: denotes the result of a viscous hydrodynamic evolution,
886:  but neglecting the viscous corrections to the particle emission at 
887: freeze-out (Eq. \ref{eq:dndydissi}).}
888: \label{fig:dndyl}
889: \end{figure}
890: 
891: In this section we present some simple estimates 
892: of the effects of the shear viscosity on the 
893: longitudinal expansion at LHC energies. To get a rough estimate
894: we set arbitrary the multiplicity of pions for central rapidity at 
895: twice the value observed for central collisions at RHIC.
896: The initial energy density distribution in space-time rapidity is modified,
897: it includes a plateau of width $\sigma_p$ \cite{Hirano:2001yi}
898: \begin{equation}
899: \epsilon(\tau_0,\theta)=\epsilon_0 
900: \exp\left(-(\theta-\sigma_p)^2\Theta(|\theta|-
901: \sigma_p)/2\sigma^2\right) \ .
902: \label{eq:einilhc}
903: \end{equation}
904: The width of the plateau is $\sigma_p=3.3$ and takes all the
905:  increase of the rapidity range when going from RHIC to LHC energies. The
906:  parameter $\sigma$ is the same as at RHIC energies, and the energy density
907:  $\epsilon_0$ is adjusted to reproduce the assumed final pion density 
908: $\frac{dN}{dy}|_{y=0}$. The meson distributions are shown in Fig. 
909: \ref{fig:dndyl}. For the viscous evolution the  plateau in the final
910:  meson distributions survives. 
911:  For an ideal fluid one gets $\epsilon_0=16.9$GeV/fm$^3$,  
912: while for a shear viscosity $\eta/s=0.2$ one needs $\epsilon_0=12.4$GeV/fm$^3$.
913: The difference between the initial energy densities in the viscous
914:  and ideal fluid evolutions is not as big as for RHIC energies. At LHC energies,
915: both the ideal and viscous fluid  evolutions have a Bjorken scaling form 
916: in several units of central rapidity (Fig \ref{fig:flowl}). The lack 
917: of space-time rapidity gradients in the distributions makes the 
918: evolution and cooling
919:  last longer, around $20$fm/c. Even at the freeze-out the flow is Bjorken-like
920: at central rapidities (Fig. \ref{fig:flowfreezel}). This observation
921:  justifies  the use at LHC energies of thermal and hydrodynamic models 
922: assuming boost invariance \cite{Kolb:2000sd,Kisiel:2005hn,Amelin:2006qe}. 
923: The hydrodynamic evolution can be 
924: restricted to the  1+2 dimensional boost invariant geometry 
925: to describe matter 
926: in the $4$-$5$ central units of rapidity. 
927: At the freeze-out hypersurface 
928:  dissipative corrections to the momentum distributions have almost 
929: disappeared (compare the solid and dotted lines in Fig. \ref{fig:dndyl}).
930: 
931: \begin{figure}
932: \includegraphics[width=.5\textwidth]{flowl.eps}
933: \caption{(color online)Difference between the  flow rapidity 
934: of the fluid and the Bjorken value at $\tau=10$fm/c,
935:  calculated for an evolution with 
936: shear viscosity coefficient $\eta/s = 0.2$ (solid line) and for an ideal
937:  fluid with a realistic equation of state (dashed line) for LHC 
938:  plateau-like initial conditions.}
939: \label{fig:flowl}
940: \end{figure}
941: 
942: \begin{figure}
943: \includegraphics[width=.5\textwidth]{flowfreezel.eps}
944: \caption{(color online) Difference between the  flow rapidity 
945: of the fluid and the Bjorken value at the freeze-out hypersurface,
946:  calculated for an evolution with 
947: shear viscosity coefficient $\eta/s = 0.2$ (solid line) and for an ideal
948:  fluid with a realistic equation of state (dashed line) for LHC 
949:  plateau-like initial conditions.}
950: \label{fig:flowfreezel}
951: \end{figure}
952: 
953: Since the cooling of the fluid due to the longitudinal expansion is slow 
954: (like in the Bjorken solution), a realistic modeling of the time 
955: scales and of the  freeze-out hypersurface must take into account 
956: the transverse expansion. The speed of the transverse expansion 
957: would determine the life-time of the system, while at RHIC energies
958:  longitudinal expansion (in space-time rapidity)  is also important.
959: We have also performed calculations using Gaussian initial energy 
960: distributions (\ref{eq:eini}) with rescaled width parameters
961:  $\sigma$ for the increased  LHC rapidity range.
962:  For a viscous fluid we find 
963: a broad region of rapidities where the 
964: Bjorken scaling flow survives through the 
965: evolution, $|\theta|<2.5$. For the ideal fluid the scaling is broken, but to
966: a significantly lesser extend than at RHIC energies.
967: 
968: \section{Conclusions}
969: 
970: The evolution of a fireball of dense and hot matter created in heavy-ion 
971: collisions can be modelled as a hydrodynamic expansion of a viscous fluid. 
972: We analyze the effects of the shear viscosity on the longitudinal expansion
973: of the matter. We solve numerically coupled evolution equations 
974: for the longitudinal flow, the energy density and the viscous 
975: corrections in a 1+1 
976: dimensional geometry, corresponding to a rapid expansion in the beam direction.
977: As a function of the space-time rapidity the distribution of  matter 
978: evolves slowly with proper time. The average density drops and the distribution 
979: gets wider. The last 
980: phenomenon takes place when the flow of the fluid 
981: gets stronger than the Bjorken one. 
982: At the freeze-out temperature the hydrodynamic stage finishes and
983: particles are emitted thermally according to the Cooper-Frye formula. 
984: Experimental measurements of the distribution of mesons in rapidity 
985: \cite{Bearden:2004yx} constrain the allowed distribution of the longitudinal 
986: velocities of the fluid elements. Correlation between space-time and 
987: momentum rapidities of the fluid, means that the space-time 
988: rapidity extension of the fluid must be limited and that its 
989: longitudinal flow cannot  deviate significantly from the Bjorken flow. 
990: Shear viscosity counteracts 
991: the gradients of the velocity field. As
992:  a consequence it slows down the longitudinal expansion. At the freeze-out 
993: the energy density distribution in space-time rapidity is narrower and the 
994: longitudinal flow gets less accelerated than for the ideal fluid hydrodynamics.
995: Fitting the initial energy density distribution
996:  to reproduce the final meson distributions, one observes
997:  a striking effect. With increasing shear 
998: viscosity coefficient the initial energy 
999: density of the fireball decreases significantly, 
1000: from $16.9$GeV/fm$^3$ for an ideal fluid to $5.6$GeV/fm$^3$ for  $\eta/s=0.2$.
1001: In the 1+1 dimensional longitudinal geometry, 
1002: this energy density corresponds to an average over the transverse plane.
1003:  Nevertheless, estimates of the maximal energy density reached
1004:  in heavy-ion collisions at RHIC energies must be strongly revised down if 
1005: shear viscosity is effective during the expansion of the fireball. 
1006: This dramatic reduction of the initial density
1007:  should also  be taken into account in hydrodynamic models
1008: dealing with transverse expansion only, both in 1+1 and 1+2 dimensions.
1009: 
1010:  Depending on 
1011: the balance of the acceleration of the flow from pressure gradients and 
1012: deceleration from viscosity, the flow gets faster or slower than 
1013: the Bjorken one. For some values of the parameters, effects of the
1014: shear viscosity and pressure gradients on the  longitudinal flow of the fluid
1015:  cancel, 
1016: i.e. the flow stays close to the Bjorken flow. This could be an argument 
1017: justifying models, which  combine transverse viscous expansion 
1018: with a Bjorken flow in the beam direction.
1019: When the initial conditions are adjusted to reproduce the final meson
1020:  distributions, we find that the freeze-out hypersurfaces are very similar, 
1021: irrespective of the value of the shear viscosity coefficient, obviously 
1022: the lifetime of the system  is not sensitive to viscous effects either
1023: (Table \ref{table2}). At the freeze-out the viscous corrections 
1024: (from the longitudinal flow) to the thermal
1025: distributions are small, except at large space-time rapidities.
1026: At LHC energies a substantial rapidity plateau, where Bjorken scaling applies, 
1027: is expected to appear. Shear viscosity helps to preserve 
1028: it in a wider rapidity interval 
1029: through
1030:  the evolution. 
1031: 
1032: 
1033: \section*{Acknowledgments}
1034: The author is grateful to Miko\l aj Chojnacki and Wojtek 
1035:  Florkowski for making available 
1036: the parameterization of the equation of state.
1037: \bibliography{hydr}
1038: 
1039: 
1040: \end{document}
1041: 
1042: 
1043: 
1044: 
1045: