Ingeneral relativity, afluid solution is anexact solution of theEinstein field equation in which the gravitational field is produced entirely by the mass, momentum, and stress density of afluid.
Inastrophysics, fluid solutions are often employed asstellar models, since a perfect gas can be thought of as a special case of a perfect fluid. Incosmology, fluid solutions are often used ascosmological models.
Thestress–energy tensor of a relativistic fluid can be written in the form[1]
Here
The heat flux vector and viscous shear tensor aretransverse to the world lines, in the sense that
This means that they are effectively three-dimensional quantities, and since the viscous stress tensor issymmetric andtraceless, they have respectively three and fivelinearly independent components. Together with the density and pressure, this makes a total of 10 linearly independent components, which is the number of linearly independent components in a four-dimensional symmetric rank two tensor.
Several special cases of fluid solutions are noteworthy (here speed of lightc = 1 and the metric sign convention used is):
The last two are often used as cosmological models for (respectively)matter-dominated andradiation-dominated epochs. Notice that while in general it requires ten functions to specify a fluid, a perfect fluid requires only two, and dusts and radiation fluids each require only one function. It is much easier to find such solutions than it is to find a general fluid solution.
Among the perfect fluids other than dusts or radiation fluids, by far the most important special case is that of thestatic spherically symmetric perfect fluid solutions. These can always be matched to aSchwarzschild vacuum across a spherical surface, so they can be used asinterior solutions in a stellar model. In such models, the sphere where the fluid interior is matched to the vacuum exterior is the surface of the star, and the pressure must vanish in the limit as the radius approaches. However, the density can be nonzero in the limit from below, while of course it is zero in the limit from above. In recent years, several surprisingly simple schemes have been given for obtainingall these solutions.
The components of a tensor computed with respect to aframe field rather than the coordinate basis are often calledphysical components, because these are the components which can (in principle) be measured by an observer.
In the special case of aperfect fluid, anadapted frame
(the first is atimelike unitvector field, the last three arespacelike unit vector fields)can always be found in which the Einstein tensor takes the simple form
where is theenergy density and is thepressure of the fluid. Here, the timelike unit vector field is everywhere tangent to the world lines of observers who are comoving with the fluid elements, so the density and pressure just mentioned are those measured by comoving observers. These are the same quantities which appear in the general coordinate basis expression given in the preceding section; to see this, just put. From the form of the physical components, it is easy to see that theisotropy group of any perfect fluid is isomorphic to the three dimensional Lie group SO(3), the ordinary rotation group.
The fact that these results are exactly the same for curved spacetimes as for hydrodynamics in flatMinkowski spacetime is an expression of theequivalence principle.
Thecharacteristic polynomial of the Einstein tensor in a perfect fluid must have the form
where are again the density and pressure of the fluid as measured by observers comoving with the fluid elements. (Notice that these quantities canvary within the fluid.) Writing this out and applyingGröbner basis methods to simplify the resulting algebraic relations, we find that the coefficients of the characteristic must satisfy the following twoalgebraically independent (and invariant) conditions:
But according toNewton's identities, the traces of the powers of the Einstein tensor are related to these coefficients as follows:
so we can rewrite the above two quantities entirely in terms of the traces of the powers. These are obviously scalar invariants, and they must vanish identically in the case of a perfect fluid solution:
Notice that this assumes nothing about any possibleequation of state relating the pressure and density of the fluid; we assume only that we have one simple and one triple eigenvalue.
In the case of a dust solution (vanishing pressure), these conditions simplify considerably:
or
In tensor gymnastics notation, this can be written using theRicci scalar as:
In the case of a radiation fluid, the criteria become
or
In using these criteria, one must be careful to ensure that the largest eigenvalue belongs to atimelike eigenvector, since there areLorentzian manifolds, satisfying this eigenvalue criterion, in which the large eigenvalue belongs to aspacelike eigenvector, and these cannot represent radiation fluids.
The coefficients of the characteristic will often appear very complicated, and the traces are not much better; when looking for solutions it is almost always better to compute components of the Einstein tensor with respect to a suitably adapted frame and then to kill appropriate combinations of components directly. However, when no adapted frame is evident, these eigenvalue criteria can be sometimes be useful, especially when employed in conjunction with other considerations.
These criteria can often be useful for spot checking alleged perfect fluid solutions, in which case the coefficients of the characteristic are often much simpler than they would be for a simpler imperfect fluid.
Noteworthy individual dust solutions are listed in the article ondust solutions. Noteworthy perfect fluid solutions which feature positive pressure include various radiation fluid models from cosmology, including
In addition to the family of static spherically symmetric perfect fluids, noteworthy rotating fluid solutions include