12.07.2015 Views

(ed.). Gravitational waves (IOP, 2001)(422s).

(ed.). Gravitational waves (IOP, 2001)(422s).

(ed.). Gravitational waves (IOP, 2001)(422s).

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

298 Elementary introduction to pre-big bang cosmologyto introduce the more conventional E-frame (16.21) in which the dilaton isminimally coupl<strong>ed</strong> to the metric, with a canonical kinetic term.In the two frames the field equations are different, and the perturbationequations are also different. This seems to raise a potential problem: which frameis to be us<strong>ed</strong> to evaluate the physical effects of the cosmological perturbations?The problem is only apparent, however, because physical observables (likethe perturbation spectrum) are the same in both frames. The reason is that thereis a compensation between the different perturbation equations and the differentbackground solution around which we expand. A general proof of this result canbe given by using the notion of canonical variable (see subsection 16.5.3). HereI will give only an explicit example for tensor perturbations in a d = 3, isotropicand spatially flat background.Let us start in the E-frame, with the background equations:R µν = 1 2 ∂ µφ∂ ν φ, (16.43)referring to the ‘tild<strong>ed</strong>’ variables of (16.42) (we will omit the ‘tilde’, forsimplicity, and will explicitly reinsert it at the end of the computation).Considering the transverse-traceless part of metric perturbations:δ (1) φ = 0, δ (1) g µν = h µν , δ (1) g µν =−h µν , ∇ ν h µ ν = 0 = h µ ν (16.44)(∇ µ denotes covariant differentiation with respect to the unperturb<strong>ed</strong> metric g,and the indices of h are also rais<strong>ed</strong> and lower<strong>ed</strong> with g). The perturbation of thebackground equations gives:δ (1) R ν µ = 0. (16.45)We can then work in the synchronous gauge, whereTo first order in h we getg 00 = 1, g 0i = 0, g ij =−a 2 δ ij ,h 00 = 0, h 0i = 0, g ij h ij = 0, ∂ j h j i = 0. (16.46)δ (1) Ɣ j 0i = 1 j 2ḣi , δ (1) Ɣ 0 ij =− 1 2 ḣij,δ (1) Ɣ k ij = 1 2 (∂ i h k j + ∂ j h k i − ∂ k h ij ). (16.47)The (0, 0) component of equation (16.45) is trivially satisfi<strong>ed</strong> (as well as theperturbation of the scalar field equation); the (i, j) components, by using theidentities (see for instance [54])g jk ḣ ik = ḣ j i + 2Hh j i ,g jk ḧ ik = ḧ j i + 2Ḣh j i + 4H ḣ j i + 4H 2 h j i , (16.48)giveδ (1) R j i =− 1 ()ḧ j i + 3H ḣ j i − ∇22a 2 h i j ≡− 1 2 £h i j = 0. (16.49)

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

Saved successfully!

Ooh no, something went wrong!