20.01.2015 Views

Feynman Path Integral Formulation

Feynman Path Integral Formulation

Feynman Path Integral Formulation

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.

18 1 Continuum <strong>Formulation</strong><br />

+ c 1 {R − 1 2 hα αR + h α β Rβ α − 1 8 Rhα αh β β + 1 4 Rhα β hβ α<br />

− h ν β hβ αR α ν + 1 2 hα αh ν β Rβ ν − 1 4 ∇ νh α β ∇ν h β α<br />

+ ∇ ν h α α∇ ν h β β − 1 2 ∇ β h α α∇ μ h β μ + 1 2 ∇α h ν β ∇ νh β α}] ,<br />

(1.94)<br />

up to total derivatives. Here ∇ μ denotes a covariant derivative with respect to the<br />

metric g μν .Forg μν = η μν the above expression coincides with the weak field<br />

Lagrangian contained in Eqs. (1.7) and (1.67), with a cosmological constant term<br />

added, as given in Eq. (1.55).<br />

To this expression one needs to add the gauge fixing and ghost contributions, as<br />

was done in Eq. (1.67). The background gauge fixing term used is<br />

− 1 2 C2 μ = − 1 2√ g(∇α h α μ − 1 2 ∇ μh α α)(∇ β h βμ − 1 2 ∇μ h β β ) , (1.95)<br />

with a corresponding ghost Lagrangian<br />

L ghost = √ g ¯η μ (∂ α ∂ α η μ − R μ αη α ) . (1.96)<br />

The integration over the h μν field can then be performed with the aid of the standard<br />

Gaussian integral formula<br />

∫<br />

ln [dh μν ] exp{− 1 2h · M(g) · h − N(g) · h}<br />

= 1 2 N(g) · M−1 (g) · N(g) − 1 2<br />

trlnM(g)+const. , (1.97)<br />

leading to an effective action for the g μν field. In practice one is only interested in<br />

the divergent part, which can be shown to be local. Specific details of the functional<br />

measure over metrics [dg μν ] are not deemed to be essential at this stage, as in perturbation<br />

theory one is only doing Gaussian integrals, with h μν ranging from −∞ to<br />

+∞. In particular when using dimensional regularization one uses the formal rule<br />

∫<br />

d d k =(2π) d δ (d) (0)=0 , (1.98)<br />

which leads to some technical simplifications but obscures the role of the measure.<br />

In the flat background field case g μν = η μν , the functional integration over the<br />

h μν fields would have been particularly simple, since then one would be using<br />

h μν (x)h αβ (x ′ ) → = G μναβ (x,x ′ ) , (1.99)<br />

with the graviton propagator G(k) given in Eq. (1.77). In practice, one can use the<br />

expected generally covariant structure of the one-loop divergent part<br />

ΔL g ∝ √ g ( α R 2 + β R μν R μν) , (1.100)<br />

with α and β some real parameters, as well as its weak field form, obtained from

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

Saved successfully!

Ooh no, something went wrong!