LOG#106. Basic Cosmology (I).

cosmologyLettersCMB_Timeline300_no_WMAP

The next thread is devoted to Cosmology. I will intend to be clear and simple about equations and principles of current Cosmology with a General Relativity background.

First of all…I will review the basic concepts of natural units I am going to use here. We will be using the following natural units:

\hbar=c=k_B=1

We will take the Planck mass to be given by

M_P=\sqrt{8\pi G_N}\approx 1\mbox{.}2\cdot 10^{19}GeV

The solar mass is M_\odot=2\cdot 10^{30}kg and the parsec is given by the value

1pc=3\mbox{.}26lyr=3\mbox{.}1\cdot 10^{16}m

Well, current Cosmology is based on General Relativity. Even if I have not reviewed this theory with detail in this blog, the nice thing is that most of Cosmology can be learned with only a very little knowledge of this fenomenal theory. The most important ideas are: metric field, geodesics, Einstein equations and no much more…

In fact, newtonian gravity is a good approximation in some particular cases! And we do know that even in this pre-relativistic theory

\mbox{Gravitational force}=\mbox{Matter/Mass density}

via the Poisson’s equation

\nabla^2\phi =4\pi G_N\rho

This idea, due to the equivalence principle, is generalized a little bit in the general relativistic framework

\mbox{Spacetime geometry}=\mbox{Matter content/Energy-momentum}

The spacetime geometry is determined by the metric tensor g_{\mu\nu}(x). The matter content is given by the stress-energy-momentum tensor T_{\mu\nu}. As we know one of these two elements, we can know, via Eisntein’s field equations the another. That is, given a metric tensor, we can tell how energy-momentum “moves” in space-time. Given the energy-momentum tensor, we can know what is the metric tensor in spacetime and we can guess how the spacetime bends… This is the origin of the famous motto: “Spacetime says matter how to move, energy-momentum says spacetime how to curve”! Remember that we have “deduced” the Einstein’s field equations in the previous post. Without a cosmological constant term, we get

G_{\mu\nu}=\kappa^2T_{\mu\nu}=8\pi G_NT_{\mu\nu}

Given a spacetime metric g_{\mu\nu}, we can calculate the (affine/Levi-Civita) connection

\Gamma^\sigma_{\;\;\mu\nu}=\dfrac{1}{2}g^{\sigma\rho}\left(\partial_\mu g_{\nu\rho}+\partial_\nu g_{\rho\mu}-\partial_\rho g_{\mu\nu}\right)

The Riemann tensor that measures the spacetime curvature is provided by the equation

R^\rho_{\;\; \sigma\mu\nu}=\partial_\mu \Gamma^\rho_{\;\;\mu\sigma}-\partial_\mu \Gamma^\rho_{\;\; \mu \sigma}+\Gamma^\rho_{\;\;\mu\lambda}\Gamma^\lambda_{\;\;\nu\sigma}-\Gamma^\rho_{\;\;\nu\lambda}\Gamma^\lambda_{\;\;\mu\sigma}

The Ricci tensor is defined to be the following “trace” of the Riemann tensor

R_{\mu\nu}=R^\lambda_{\;\;\mu\lambda \nu}

The Einstein tensor is related to the above tensors in the well-known manner

G_{\mu\nu}=R_{\mu\nu}-\dfrac{1}{2}Rg_{\mu\nu}

The Einstein’s equations can be derived from the Einstein-Hilbert action we learned in the previous post, using the action principle and the integral

\boxed{S_{EH}=\int d^4x \sqrt{-g} \left(\kappa^{-2}R+\mathcal{L}_M\right)}

The geodesic equation is the path of a freely falling particle. It gives a “condensation” of the Einstein’s equivalence principle too and it is also a generalization of Newton’s law of “no force”. That is, the geodesic equation is the feynmanity

\boxed{\dfrac{d^2 x^\mu}{d\tau^2}+\Gamma^\mu _{\rho\sigma}\dfrac{dx^\rho}{d\tau}\dfrac{dx^\sigma}{d\tau}=0}

Finally, an important concept in General Relativity is that of isometry. The symmetry of the “spacetime manifold” is provided by a Killing vector that preserves transformations (isometries) of that manifold. Mathematically speaking, the Killing vector fields satisfy certain equation called the Killing equation

\boxed{\xi_{\mu ; \nu}+\xi_{\nu ; \mu}=0}

Maximally symmetric spaces have n(n+1)/2 Killing vectors in n-dimensional (nD) spacetime. There are 3 main classes or types of 2D maximally symmetric that can be generalized to higher dimensions:

1. The euclidean plane E^2.

2. The pseudo-sphere H^2. This is a certain “hyperbolic” space.

3. The spehre S^2. This is a certain “elliptic” space.

The Friedmann-Robertson-Walker Cosmology

Current cosmological models are based in General Relativity AND  a simplification of the possible metrics due to the so-called Copernican (or cosmological) principle: the Universe is pretty much the same “everywhere” you are in the whole Universe! Remarkbly, the old “perfect” Copernican (cosmological) principle that states that the Universe is the same “everywhere” and “every time” is wrong. Phenomenologically, we have found that the Universe has evolved and it evolves, so the Universe was “different” when it was “young”. Therefore, the perfect cosmological principle is flawed. In fact, this experimental fact allows us to neglect some old theories like the “stationary state” and many other “crazy theories”.

What are the observational facts to keep the Copernican principle? It seems that:

1st. The distribution of matter (mainly galaxies, clusters,…) and radiation (the cosmic microwave background/CMB) in the observable Universe is homogenous and isotropic.

2nd. The Universe is NOT static. From Hubble’s pioneer works/observations, we do know that galaxies are receeding from us!

Therefore, these observations imply that our “local” Hubble volume during the Hubble time is similar to some spacetime with homogenous and isotropic spatial sections, i.e., it is a spacetime manifold M=\mathbb{R}\times \Sigma. Here, \mathbb{R} denotes the time “slice” and \Sigma represents a 3D maximally symmetric space.

The geometry of a locally isotropic and homogeneous Universe is represented by the so-called Friedmann-Robertson-Walker metric

\boxed{ds^2_{FRW}=-dt^2+a^2(t)\left[\dfrac{dr^2}{1-kr^2}+r^2\left(d\theta^2+\sin\theta^2d\phi^2\right)\right]}

Here, a(t) is the called the scale factor.  The parameter k determines the geometry type (plane, hyperbolic or elliptical/spherical):

1) If k=0, then the Universe is “flat”. The manifold is E^3.

2) If k=-1, then the Universe is “open”/hyperbolic. The manifold would be H^3.

3) If k=+1, then the Universe is “closed”/spherical or elliptical. The manifold is then S^3.

Remark: The ansatz of local homogeneity and istoropy only implies that the spatial metric is locally one of the above three spaces, i.e., E^3,H^3,S^3. It could be possible that these 3 spaces had different global (likely topological) properties beyond these two properties.

Kinematical features of a FRW Universe

The first property we are interested in Cosmology/Astrophysics is “distance”. Measuring distance in a expanding Universe like a FRW metric is “tricky”. There are several notions of “useful” distances. They can be measured by different methods/approaches and they provide something called sometimes “the cosmologidal distance ladder”:

1st. Comoving distance. It is a measure in which the distance is “taken” by a fixed coordinate.

2nd. Physical distance. It is essentially the comoving distance times the scale factor.

3rd. Luminosity distance. It uses the light emitted by some object to calculate its distance (provided the speed of light is taken constant, i.e., special relativity holds and we have a constant speed of light)

4th. Angular diameter distance. Another measure of distance using the notion of parallax and the “extension” of the physical object we measure somehow.

There is an important (complementary) idea in FRW Cosmology: the particle horizon. Consider a light-like particle with ds^2=0. Then,

dt=a(t)\dfrac{1}{\sqrt{1-kr^2}}

or

\dfrac{dr}{\sqrt{1-kr^2}}=\dfrac{dt}{a(t)}

The total comoving distance that light have traveled since a time t=0 is equal to

\boxed{\eta=\int_0^{r_H}\dfrac{dr}{\sqrt{1-kr^2}}=\int_0^t\dfrac{dt'}{a(t')}}

It shows that NO information could have propagated further and thus, there is a “comoving horizon” with every light-like particle! Here, this time is generally used as a “conformal time” as a convenient tiem variable for the particle. The physical distance to the particle horizon can be calculated

\boxed{d_H(t)=\int_0^{r_H}\sqrt{g_{rr}}dr=a(t)\int_0^t\dfrac{dt'}{a(t')}=a(t)\eta}

There are some important kinematical equations to be known

A) For the geodesic equation, the free falling particle, we have

\Gamma^0_{ij}=\dfrac{\dot{a}}{a}\overline{g}_{ij}

\Gamma^i_{0j}=\Gamma^i_{j0}=\dfrac{\dot{a}}{a}\delta_{ij}

\Gamma^i_{jk}=\overline{\Gamma}^i_{jk}

for the FRW metric and, moreover, the energy-momentum vector P^\mu=(E,\mathbf{p}) is defined by the usual invariant equation

P^\mu=\dfrac{dx^\mu}{d\lambda}

This definition defines, in fact, the proper “time” \lambda implicitely, since

\dfrac{d}{d\lambda}=\dfrac{dx^0}{d\lambda}\dfrac{d}{dx^0}=E\dfrac{d}{dt}

and the 0th component of the geodesic equation becomes

E\dfrac{dE}{dt}=-\Gamma^0_{ij}p^ip^j=-\delta_{ij}a\dot{a}p^ip^j

g_{\mu\nu}p^\mu p^\nu=-E^2+a^2\delta_{ij}p^ip^j=-m^2

EdE=a^2\vert \mathbf{p}\vert d\vert \mathbf{p}\vert

a^2 p\dfrac{dp}{dt}=-a\dot{a} p^2

\dfrac{1}{\vert \mathbf{p}\vert }\dfrac{d\vert \mathbf{p}\vert}{dt}+\dfrac{\dot{a}}{a}=0

Therefore we have deduced that \vert \mathbf{p}\vert \propto a^{-1}. This is, in fact, the socalled “redshift”.  The cosmological  redshift parameter is more generally defined through the equation

\boxed{\dfrac{a(t_0)}{a(t)}=1+z=\dfrac{\lambda_0}{\lambda}}

B) The Hubble’s law.

The luminosity distance measures the flux of light from a distant object of known luminosity (if it is not expanding). The flux and luminosity distance are bound into a single equation

\boxed{F=\dfrac{L}{4\pi d^2_L}}

If we use the comoving distance between a distant emitter and us, we get

\chi (a)=\int_t^{t_0}\dfrac{dt'}{a(t')}=\int_a^1\dfrac{da'}{a'^2 H(a')}

for a expanding Universe! That is, we have used the fact that luminosity itself goes through a comoving spherical shell of radius \chi (a). Moreover, it shows that

F=\dfrac{L (\chi)}{4 \pi \chi (a)^2 a_0^2}=\dfrac{L}{4\pi (\chi (a)/a)^2}

The luminosity distance in the expanding shell is

d_L=\dfrac{\chi (a)}{a}=\left(\dfrac{L}{4\pi F}\right)^{1/2}

and this is what we MEASURE in Astrophysics/Cosmology. Knowing a(t), we can express the luminosity distance in terms of the redshift. Taylor expansion provides something like this:

H_0d_L=z+\dfrac{1}{2}(1-q_0)z^2+\ldots

where higher order terms are sometimes referred as “statefinder parameters/variables”. In particular, we have

\boxed{H_0=\dfrac{\dot{a}_0}{a_0}}

and

\boxed{q_0=-\dfrac{a_0\ddot{a}_0}{\dot{a}_0^2}}

C) Angular diameter distance.

If we know that some object has a known length l, and it gives some angular “aperture” or separation \theta, the angular diameter distance is given by

\boxed{d_A=\dfrac{l}{\theta}}

The comoving size is defined as l/a, and the coming distance is again \chi (a). For “flat” space, we obtain that

\theta=\dfrac{l/a}{\chi (a)}

that is

d_A=a\chi (a)=\dfrac{\chi}{1+z}

In the case of “curved” spaces, we get

d_A=\dfrac{a}{H_0\sqrt{\vert \omega_k\vert}}\cdot\begin{cases}\sinh \left( \sqrt{\Omega_k}H_0\chi\right),\Omega_k>0\\ \sin \left( \sqrt{-\Omega_k}H_0\chi\right),\Omega_k<0\end{cases}

FRW dynamics

Gravity in General Relativity, a misnomer for the (locally) relativistic theory of gravitation, is described by a metric field, i.e., by a second range tensor (covariant tensor if we are purist with the nature of components). The metric field is related to the matter-energy-momentum content through the Einstein’s equations

G_{\mu\nu}=-\kappa^2 T_{\mu\ nu}

The left-handed side can be calculated for a FRW Universe as follows

R_{00}=-3\dfrac{\ddot{a}}{a}

R_{ij}=(a\ddot{a}+2\dot{a}^2+2k)\overline{g}_{ij}

R=6\left(\dfrac{\ddot{a}}{a}+\dfrac{\dot{a}^2}{a^2}+\dfrac{k}{a^2}\right)

The right-handed side is the energy-momentum of the Universe. In order to be fully consistent with the symmetries of the metric, the energy-momentum tensor MUST be diagonal and T_{11}=T_{22}=T_{33}=T. In fact, this type of tensor describes a perfect fluid with

T_{\mu\nu}=(\rho+p)U_\mu U_\nu+pg_{\mu\nu}

Here, \rho, p are functions of t (cosmological time) only. They are “state variables” somehow. Moreover, we have

U_\mu =(1,0,0,0)

for the fluid at rest in the comoving frame. The Friedmann equations are indeed the EFE for a FRW metric Universe

3\left(\dfrac{\dot{a}^2}{a^2}+\dfrac{k}{a^2}\right)=\kappa^2\rho for the 00th compoent as “constraint equation.

2\dfrac{\ddot{a}}{a}+\dfrac{\dot{a}^2}{a^2}+\dfrac{k}{a^2}=-\kappa^2p for the iith components.

Moreover, we also have

G_{\mu\nu}^{;\nu}=T_{\mu\nu}^{;\nu}=0

and this conservation law implies that

\dot{\rho}+3\dfrac{\dot{a}}{a}(\rho+p)=0

Therefore, we have got two independent equations for three unknowns (a, \rho, p). We need an additional equation. In fact, the equation of state for p=p(\rho) provides such an additional equation. It gives the “dynamics of matter”!

In summary, the basic equations for Cosmology in a FRW metric, via EFE, are the Friedmann’s equations (they are secretly the EFE for the FRW metric) supplemented with the energy-momentum conservations law and the equation of state for the pressure p=p(\rho):

1) \boxed{\dfrac{\dot{a}^2}{a^2}+\dfrac{k^2}{a^2}=\dfrac{\kappa^2}{3}\rho}

2) \boxed{\dot{\rho}+3\dfrac{\dot{a}}{a}(\rho+p)=0}

3) \boxed{p=p(\rho)}

There are many kinds of “matter-energy” content of our interest in Cosmology. Some of them can be described by a simple equation of state:

\boxed{p=\omega \rho}

Energy-momentum conservation implies that \rho\propto a^{-3(\omega +1)}. 3 special cases are used often:

1st. Radiation (relativistic “matter”). \omega=1/3 and thus, p=1/3\rho and \rho\propto a^{-4}

2nd. Dust (non-relativistic matter). \omega=0. Then, p=0 and \rho\propto a^{-3}

3rd. Vacuum energy (cosmological constant). \omega=-1. Then, p=-\rho and \rho=\mbox{constant}

Remark (I): Particle physics enters Cosmology here! Matter dynamics or matter fields ARE the matter content of the Universe.

Remark (II): Existence of a Big Bang (and a spacetime singularity). Using the Friedmann’s equation

\dfrac{\ddot{a}}{a}=-\dfrac{\kappa^2}{6}(\rho+3p)

if we have that (\rho+3p)>0, the so-called weak energy condition, then a=0 should have been reached at some finite time in the past! That is the “Big Bang” and EFE are “singular” there. There is no scape in the framework of GR. Thus, we need a quantum theory of gravity to solve this problem OR give up the FRW metric at the very early Universe by some other type of metric or structure.

Particles and the chemical equilibrium of the early Universe

Today, we have DIRECT evidence for the existence of a “thermal” equilibrium in the early Universe: the cosmic microwave background (CMB). The CMB is an isotropic, accurate and non-homogeneous (over certain scales) blackbody spectrum about T\approx 3K!

Then, we know that the early Universe was filled with a hot dieal gas in thermal equilibrium (a temperature T_e can be defined there) such as the energy density and pressure can be written in terms of this temperature. This temperature generates a distribution f(\mathbf{x},\mathbf{p}). The number of phase space elements in d^3xd^3p is

d^3xd^3p=\dfrac{d^3\mathbf{x}d^3\mathbf{p}}{(2\pi\hbar)^3}

and where the RHS is due to the uncertainty principle. Using homogeneity, we get that, indeed, f(x,p)=f(p), and where we can write the volume d^3x=dV. The energy density and the pressure are given by (natural units are used)

\rho_i=g_i\int \dfrac{d^3p}{(2\pi)^3}f_i(p)E(p)

p_i=g_i\int \dfrac{d^3p}{(2\pi)^3}f_i (p)\dfrac{p^2}{3E(p)}

When we are in the thermal equilibrium at temperature T, we have the Bose-Einstein/Fermi-Dirac distribution

f(p)=\dfrac{1}{e^{(E-\mu)/T}\pm 1}

and where the + is for the Fermi-Dirac distribution (particles) and the - is for the Bose-Einstein distribution (particles). The number density, the energy density and the pressure are the following integrals

\boxed{\mbox{Number density}=n=\dfrac{N}{V}=\dfrac{g}{2\pi^2}\int_m^\infty \dfrac{(E^2-m^2)^{1/2}}{e^{(E-\mu)/T}\pm 1}dE}

\boxed{\mbox{Density energy}=\rho=\dfrac{E}{V}=\dfrac{g}{2\pi^2}\int_m^\infty \dfrac{(E^2-m^2)^{1/2}E^2}{e^{(E-\mu)/T}\pm 1}dE}

\boxed{\mbox{Pressure}=p=\dfrac{g}{6\pi^2}\int_m^\infty \dfrac{(E^2-m^2)^{3/2}}{e^{(E-\mu)/T}\pm 1}dE}

And now, we find some special cases of matter-energy for the above variables:

1st. Relativistic, non-degenerate matter (e.g. the known neutrino species). It means that T>>m and T>>\mu. Thus,

n=\left(\dfrac{3}{4}\right)\dfrac{\zeta (3)}{\pi^2}gT^3

\rho=\left(\dfrac{7}{8}\right)\dfrac{\pi^2}{30}gT^4

p=\dfrac{1}{3}\rho

2nd. Non-relativistic matter with m>>T only. Then,

n=g\left(\dfrac{mT}{2\pi}\right)^{3/2}e^{-(m-\mu)/T}

\rho= mn+\dfrac{3}{2}p, and p=nT<<\rho

The total energy density is a very important quantity. In the thermal equilibrium, the energy density of non-relativistic species is exponentially smaller (suppressed) than that of the relativistic particles! In fact,

\rho_R=\dfrac{\pi^2}{30}g_\star T^4 for radiation with p_R=\dfrac{1}{3}\rho_R

and the effective degrees of freedom are

\displaystyle{\boxed{g_\star=\sum_{bosons}g_b+\dfrac{7}{8}\sum_{fermions}g_f}}

Remark: The factor 7/8 in the DOF and the variables above is due to the relation between the Bose-Einstein and the Fermi-Dirac integral in d=3 space dimensions. In general d, the factor would be

(1-\dfrac{1}{2^d})=\dfrac{2^d-1}{2^d}

Entropy conservation and the early Universe

The entropy in a comoving volume IS a conserved quantity IN THE THERMAL EQUILIBRIUM. Therefore, we have that

\dfrac{\partial p_i}{\partial T}=g_i\int \dfrac{d^3p}{(2\pi)^3}\dfrac{df}{dT}\dfrac{p^2}{3E(p)}=g_i\int \dfrac{4\pi pE dE}{(2\pi)^3}\dfrac{df}{dE}\left(-\dfrac{E}{T}\right)\dfrac{p^2}{3E}

and then

\dfrac{\partial p_i}{\partial T}=\dfrac{g_i}{2\pi^2}\int \left(-\dfrac{d}{dE}\left(f\dfrac{p^3E}{3T}\right)+f\dfrac{d}{dE}\left(\dfrac{p^3E}{3T}\right)\right)dE

or

\dfrac{\partial p_i}{\partial T}=\dfrac{1}{T}(\rho_i+p_i)

Now, since

\dfrac{\partial \rho}{\partial t}+3\dfrac{\dot{a}}{a}(\rho+p)=0

then

\dfrac{\partial}{\partial t}\left(a^3(\rho+p)\right)-a^3\dfrac{\partial p}{\partial t}=0

\dfrac{1}{a^3}\dfrac{\partial (a^3(\rho +p))}{\partial t}-\dfrac{\partial \rho}{\partial t}=0

if we multiply by T and use the chain rule for \rho, we obtain

\dfrac{1}{a^3}\dfrac{\partial}{\partial t}\left(\dfrac{a^3(\rho+p)}{T}\right)=0

but it means that a^3s=\mbox{constant}, where s is the entropy density defined by

\boxed{s\equiv \dfrac{\rho+p}{T}}

Well, the fact is that we know that the entropy or more precisely the entropy density is the early Universe is dominated by relativistic particles ( this is “common knowledge” in the Stantard Cosmological Model, also called \Lambda CDM). Thus,

\boxed{s=\dfrac{2\pi^2}{45}g_\star T^3}

It implies the evolution of temperature with the redshift in the following way:

T\propto g_\star^{-1/3}a^{-1}

Indeed, since we have that n\propto a^{-3}, s\propto a^{-3}, the yield variable

Y_i\equiv \dfrac{n_i}{s}

is a convenient quantity that represents the “abundance” of decoupled particles.

See you in my next cosmological post!

Advertisements

LOG#044. Hydrodynamics and SR(I).

Relativistic hydrodynamics is a branch of Relativity Theory that faces with fluids and/or molecules (“gases”) moving at relativistic speeds. Today, this area of Special Relativity has been covered with many applications. However, it has not been so since, not so long ago, the questions was:

Where could one encounter fluids or “gases” that would propagate with velocities close to the the speed of light?

It was thought that it seemed a question to be very far away from any realistic or practical use. At present time, relativistic hydrodynamics IS an importan part of Cosmology and the theory of processes going on in the sorrounding and ambient space of neutron stars (likely, of the quark stars as well), compact massive objects and black holes. When the relativistic fluid flows under strong gravitational fields existing in those extreme conditions at relativistic speeds, it drives to a big heating and X-ray emission, for instance. And then, a relativistic treatment of matter is inevitable there.

Caution note: I will use units with c=1 in this post, in general, without loss of generality.

Let me review a bit the non-relativistic hydrodynamics of ideal fluids and gases. Their dynamics is governed by the continuity equation (mass conservation) and the Euler equation:

\boxed{\mbox{Continuity equation:\;\;}\dfrac{\partial \rho}{\partial t}+\nabla \cdot(\rho \mathbf{v})=0}

\boxed{\mbox{Euler equation:\;\;}\rho \dfrac{d\mathbf{v}}{dt}+\nabla p=\mathbf{f}}

where the mass density and pressure of the fluid are respectively \rho and p. To complete the fluid equations, these equations need to be supplemented by an equation of state:

\boxed{\mbox{Equation of state:\;\;}p=p(\rho)}

The continuity equation expresses the fact that mass in an invariant in classical fluid theory. The Euler equation says how the changes of pressure and forces affect to the velocity of the fluid, and finally the equation of state encodes the type of fluid we have at macroscopic level from the microscopic degrees of freedom that fluid theory itself can not see.

By the other hand, it is worth mentioning that we can not write the continuity equation into a covariant form \partial_\mu j^\mu=0 in such “a naive” way, with j^\mu=\rho (x)v^\mu. Why? It is pretty easy: it is a characteristic property of special relativity that the mass density \rho (x) does not satisty such an equation, but we can derive a modified continuity equation that holds in SR. To build the right equations, we can proceed using an analogy with the electromagnetic field. Suppose we write an “stress-energy-momentum” tensor for an ideal fluid in the following way:

T_{\mu\nu}=\begin{pmatrix}\rho & 0 & 0& 0\\ 0 & p & 0 & 0\\ 0 & 0 & p & 0\\ 0 & 0 & 0 & p \end{pmatrix}

This tensor is written in the rest frame of a fluid. Note, then, that ideal fluids are characterized b the feature that their stress tensor T_{ij} contains no shear stresses ( off-diagonal terms) and they are thus “proportional” to the Kronecker delta tensor \delta_{\mu\nu}.

The generalization of the above tensor (an ideal fluid) to an arbitrary reference frame, in which the fluid element moves with some 4-velocity components u^\mu is given by the next natural generalization:

T_{\mu\nu}=(\rho +p)u_\mu u_\nu-\eta_{\mu \nu}p

where again, \rho (x) represents the density, p(x) is the local fluid pressure field as measured in the rest frame of the fluid element, and \eta_{\mu\nu} is the Minkovski metric. Therefore, the equations of motion can be found, in the absence of external forces, from the conservation laws:

\boxed{\mbox{Relativistic continuity equation:\;\;}\partial_\nu T^{\mu\nu}=0}

Inserting the above tensor for the fluid, we get

\partial_\nu \left[ \left( \rho +p\right)u^\mu u^\nu\right]-\partial^\mu p=0

We can find this equation to the non-relativistic Euler equation. The trick is easy: we firstly multiply by u^\mu and, after a short calculation, as we have that u^\mu u_\mu=-1, and (\partial_\nu u^\mu) u_\mu=0, it provides us with

\partial_\mu(\rho u^\mu)+p\partial_\mu u^\mu=0

This last equation shows, indeed, that mass current \rho u^\mu is not conserved itself. Recall that from the spatial part of our relativist fluid equations:

\partial_i \left[ \left( \rho +p\right)\mathbf{v} u^i\right]-\nabla p=0

If we define the so-called “convective” or “comoving” derivative of an arbitrary tensor field T as the derivative:

\dot{T}=\dfrac{DT}{d\lambda}=\partial_i Tu^i

we can rewrite the spatial part of the relativistic fluid dynamics as follows:

\boxed{\mbox{Relativistic Euler equation in 3d-space:\;}\dot{p}\mathbf{v}+(\rho+p)\dot{\mathbf{v}}+\nabla p=0}

We can check that it effectively corresponds to the classical Euler equation moving to the comoving frame where u^\mu=(1,\mathbf{0})^T, excepting for a pressure term equals to p/c^2 if we reintroduce units with the speed of light, and then it is the generalized mass-energy density conservation law from relativistic hydrodynamics!

Remark: In the case of electromagnetic radiation, we have a pressure term equals to p=\rho/3 due to the tracelessneess of the electromagnetic stress-energy-momentum tensor.

Returning to our complete relativistic equation, we observe that the time component of that equation has NOT turned out to be the relativistic version of the continuity equation, as we warned before. The latter rather has to be postulated separately, using additional insights from elementary particle physics! In this way, instead of a mass density conservation, we do know, e.g., that the baryon density n(x) does satisfy an equation of continuity (at least from current knowledge of fundamental physics):

\partial_\mu (n u^\mu)=0

and it merely says that baryon number is conserved under reasonable conditions ( of course, we do suspect at current time that baryon number conservation is not a good symmetry in the early Universe, but today it holds with good accuracy). Similarly, it can be said that for an electron gas, the baryon density has to be replaced by the so-called lepton density in the equation of continuity, where we could consider a gas with electrons, muons, tauons and their antiparticles. We could guess the neutrino density as well with suitable care. However, for phtoons and “mesons” there is NO continuity equation since tehy can be created and annihilated arbitrarily.

The relationship between n, p and \rho can be obtained from the equation of state and basic thermodynamics from the definition of pressure:

p=-\dfrac{\mbox{Energy per baryon}}{\mbox{volume per baryon}}=-\dfrac{(\rho/n)}{(1/n)}=n\dfrac{d\rho}{dn}-\rho

or

\int \dfrac{d\rho}{p(\rho)+\rho}=\int \dfrac{dn}{n}

Thus, with these equations, we can know the density n(\rho). The mass density \rho and the baryon density n differ by the density n\epsilon of the inner energy ( we have defined \epsilon as the specific inner energy or equivalently the inner energy per baryon):

\rho=n(1+\epsilon)

The inner energy is negative if energy is released at the formation of the state \rho, e.g. in the binding energy of a nucleus, and it is positive if energy has to be spent, e.g. if we make a compressional work onto the state.

Moreover, the specific entropy s=S/B or entropy per baryon and the temperature are defined by postulating the thermodynamical equilibrium and, with that integrating factor 1/T coming from elementary thermodynamics, we can write

ds=\dfrac{1}{T}\left(d\epsilon+pd\left(\dfrac{1}{n}\right)\right)

since we have an specific volume equal to v=1/n. Entropy is constant along a stream line of the ideal fluid, and it follows from

\dot{p}=\partial_\mu\left[(p+\rho)u^\mu\right]=\partial_\mu\left[(n+\epsilon n+p)u^\mu\right]=n\dot{\epsilon}+p\partial_\mu u^\mu+\dot{p}

if we divide by n

T\dot{s}=\dot{\epsilon}+p\left(\dfrac{1}{n}\right)

In conclusion, we deduce that the time conservation of the relativistic conservation law of fluid hydrodynamics tell us that in the case of an ideal fluid no energy is converted into heat, and entropy itself remains constant! Isn’t it amazing? Entropy rocks!

Final remark (I): For non-ideal fluids, the ansatz for the energy-momentum-stress tensor hast to be modified into this one

T_{\mu\nu}=(\rho+p)u_\mu u_\nu+(q_\mu u_\nu+q_\nu u_\mu)-\eta_{\mu\nu}p-\pi_{\mu\nu}

Final remark (II): Relativistic hydrodynamics can be generalized to charged fluids and plasmas, and in that case, it is called relativistic magnetohydrodynamics. One simply adds the stress-energy-momentum tensor of the electromagnetic field to the tensor of the fluid in such a way it is conserved as well, i.e., with zero divergence. Thus, we would get an extra Lorentz-like force into the relativistic generalization of the Euler equation!

Final remark (III): The measurement of thermodynamical quantities like pressure, entropy or temperature, and its treatment with classical thermodynamics suppose that the thermodynamical equilibrium state is reached. Please, note that if it were not the case, we should handle the problem with tools of non-equilibrium thermodynamics or some other type of statistical mechanics that could describe the out-of-the equilibrium states ( there are some suggestions of such a generalization of thermodynamics, and/or statistical mechanics, from the works of I. Prigogine, C.Tsallis, C.Beck and many other physicists).