Heavy ion collisions at RHIC are well described by the (nearly ideal) hydrodynamics for average events. In the present paper we study initial state fluctuations appearing on an event-by-event basis, and the propagation of perturbations induced by them. We found that (i) fluctuations of several lowest harmonics have comparable magnitudes, (ii) that at least all odd harmonics are correlated in phase, (iii) thus indicating the local nature of fluctuations. We argue that such local perturbation should be the source of the "Tiny Bang", a pulse of sound propagating from it. We identify its two fundamental scales as (i) the "sound horizon" (analogous to the absolute ruler in cosmic microwave background and galaxy distribution) and (ii) the "viscous horizon", separating damped and undamped harmonics. We then qualitatively describe how one can determine them from the data, and thus determine two fundamental parameters of the matter, the (average) speed of sound and viscosity. The rest of the paper explains how one can study mutual coherence of various harmonics. For that one should go beyond the two-particle correlations, to three (or more) particles. Mutual coherence is important for the picture of propagating sound wave.

The hydrodynamical description of the "Little Bang" in heavy ion collisions is surprisingly successful, mostly due to the very small viscosity of the Quark-Gluon plasma. In this paper we systematically study the propagation of small perturbations, also treated hydrodynamically. We start with a number of known techniques allowing for the analytic calculation of the propagation of small perturbations on top of the expanding fireball. The simplest approximation is the "geometric acoustics", which substitutes the wave equation by mechanical equations for the propagating "phonons". Next we turn to the case in which variables can be separated, where one can obtain not only the eikonal phases but also the amplitudes of the perturbation. Finally, we focus on the so called Gubser flow, a particular conformal analytic solution for the fireball expansion, on top of which one can derive closed equations for small perturbations. Perfect hydrodynamics allows all variables to be separated and all equations to be solved in terms of known special functions. We can thus collect the analytical expression for all the harmonics and reconstruct the complete Green function of the problem. In the viscous case the equations still allow for variable separation, but one of the equations has to be solved numerically. Summing all the harmonics we show real-time perturbation evolution, observing the viscosity-induced changes in the spectra and the correlation functions. The calculated angular shape of the correlation function is remarkably similar to the shape emerging from the experimental data, for sufficiently large viscosity. We predict a minimum at m ∼ 7 and maximum at m ∼ 9 harmonics, which also have some experimental evidence for it. We conclude that local "hot spots" in the initial state are the only visible origin of the observed correlations.

Hydrodynamical description of the "Little Bang" in heavy ion collisions is surprisingly successful: here we systematically study propagation of small perturbations treated hydrodynamically. Using analytic description of the expanding fireball known as the "Gubser flow", we proceed to linearized equations for perturbations. As all variables are separated and all equations solved (semi)analytically, we can collect all the harmonics and reconstruct the complete Green function of the problem, even in the viscous case. Applying it to the power spectrum we found acoustic minimum at the m = 7 and maximum at m = 9, which remarkably have some evidence for both in the data. We estimate effective viscosity and size of the perturbation from a fit to power spectrum. The shape of the two-point correlator is also reproduced remarcably well. At the end we argue that independent perturbations are local, and thus harmonics phases are correlated.