04.06.2013 Views

Gravity and Strings

Gravity and Strings

Gravity and Strings

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

94 A perturbative introduction to general relativity<br />

to find, case by case, the corrections to the lowest-order coupled system, which we know<br />

is inconsistent. A weakness of the geometrical point of view is that there is no generalcovariant<br />

energy–momentum tensor of the gravitational field itself. As we have seen, there<br />

is a Lorentz-covariant energy–momentum tensor (or pseudotensor) of the gravitational field<br />

embedded in the Ricci tensor together with the wave operator, but it cannot be promoted<br />

to a general-covariant tensor, as can be understood from the equivalence principle. This<br />

obscures the physical interpretation of vacuum solutions of the geometrical theory, which<br />

are not strictly speaking vacuum solutions since the whole spacetime is filled by a nontrivial<br />

gravitational field that acts as a source for itself. We will come back to this point in<br />

Chapter 6.<br />

Where does this principle of general covariance come from? We started from a theory<br />

with an Abelian gauge symmetry 38 δ (0)<br />

ɛ hµν =−2∂(µɛν).Weargued that this symmetry was<br />

necessary in order to have a consistent theory of free massless spin-2 particles. Then we<br />

coupled this free theory to the conserved energy–momentum tensor of the matter fields,<br />

saw the need to introduce a self-coupling of the spin-2 field, <strong>and</strong> argued that the form of<br />

the coupling should be dictated by gauge invariance with respect to the corrected trans-<br />

formations δ (1)<br />

ɛ hµν =−2∂(µɛν) − χLɛhµν, which combined the Abelian gauge symmetry<br />

we started from <strong>and</strong> “localized” translations in such a way that the commutator of two δ (1)<br />

ɛ<br />

infinitesimal transformations gives another δ (1)<br />

ɛ transformation. This is the only possible<br />

extension of the Abelian δ (0)<br />

ɛ transformations [739, 933] <strong>and</strong> the algebra is the algebra of<br />

infinitesimal GCTs. 39 In fact, we can easily see how the full gauge transformation δ (1)<br />

ɛ hµν<br />

arises from the effect of a GCT on the metric gµν = ηµν + χhµν, just by substituting <strong>and</strong><br />

38 (0)<br />

Any two of these gauge transformations commute because δ ξ1 δξ2hµν = δ (0)<br />

ξ1+ξ2 hµν.<br />

39 As shown in [933], it is possible to have a self-coupled spin-2 theory with only “normal spin-2 gauge<br />

symmetry” (δ (0)<br />

ɛ ). For instance, we can add to the Fierz–Pauli Lagrangian a term proportional to some (for<br />

instance, the third) power of the linearized Ricci scalar<br />

∂ 2 h − ∂µ∂νh µν , (3.249)<br />

which is exactly invariant under δ (0)<br />

ɛ .Ofcourse, the resulting higher-derivative theory cannot have the same<br />

interpretation, since the r.h.s. of the equation of motion is not the gravitational energy–momentum tensor.<br />

Also, we can couple the linear theory to matter <strong>and</strong> obtain an interacting theory that is invariant under δ (0)<br />

ɛ :<br />

we just have to add to the free-matter Lagrangian <strong>and</strong> the Fierz–Pauli Lagrangian an interaction term of the<br />

form <br />

d d xh µν Jµν(ϕ), (3.250)<br />

where Jµν is any symmetric, identically conserved tensor built out of ϕ <strong>and</strong> its derivatives. This excludes<br />

the matter energy–momentum tensor, which is conserved only on-shell. Since Jµν is identically conserved,<br />

the modification introduced into the equations of motion for matter by the coupling to gravity is immaterial.<br />

Local Jµνs can be constructed from local four-index tensors with the symmetries Jµρνσ = J[µρ][νσ] =<br />

Jνσµρ, defining<br />

Jµν = ∂ ρ ∂ σ Jµρνσ. (3.251)<br />

These Jµνs are called Pauli terms in [943]. It is also possible to define identically conserved nonlocal Jµνs,<br />

for instance the non-local projection of the energy–momentum tensor Eq. (3.170). In all these cases, we see<br />

that the spin-2 field does not couple to the total energy–momentum tensor <strong>and</strong> the quantum theories are not<br />

consistent, according to [942].

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!