0

I am trying to find the Field equation for

$$S = \int \sqrt{-g}dx^4[f(\phi)R + h(\phi)g^{\mu \nu}\nabla_{\mu}\phi\nabla_{\nu}\phi - V(\phi)$$

but I could not take the variation of $$\delta(\sqrt{-g}h(\phi)g^{\mu \nu}\nabla_{\mu}\phi\nabla_{\nu}\phi)$$

I know that $\delta(\sqrt{-g}) = -\frac{1}{2}\sqrt{-g}g_{\mu\nu}\delta g^{\mu\nu}$. We can also take $\delta h = \frac{\partial h}{\partial \phi}\delta \phi$.

We can separate the expression into five terms.

  1. $$\delta(\sqrt{-g})hg^{\mu \nu}\nabla_{\mu}\phi\nabla_{\nu}\phi$$
  2. $$\sqrt{-g}\delta h g^{\mu \nu}\nabla_{\mu}\phi\nabla_{\nu}\phi$$
  3. $$\sqrt{-g}h \delta(g^{\mu \nu})\nabla_{\mu}\phi\nabla_{\nu}\phi$$
  4. $$\sqrt{-g}h g^{\mu \nu}\nabla_{\mu}\delta(\phi)\nabla_{\nu}\phi$$
  5. $$\sqrt{-g}h g^{\mu \nu}\nabla_{\mu}\phi\nabla_{\nu}\delta(\phi)$$

So I have found

  1. $$(-\frac{1}{2}\sqrt{-g}hg_{\mu\nu}\nabla_{\beta}\phi\nabla^{\beta}\phi)\delta g^{\mu\nu}$$
  2. $$(\sqrt{-g}\nabla_{\mu}\phi\nabla^{\mu}\phi\frac{\partial h}{\partial \phi})\delta \phi$$
  3. $$(\sqrt{-g}h\nabla_{\mu}\phi\nabla_{\nu}\phi)\delta(g^{\mu \nu})$$

From $(4)$ and $(5)$ I obtain,

$$(4)+(5) = 2\sqrt{-g}h\nabla^{\mu}\phi \nabla_{\mu}\delta(\phi)$$

and by taking,

$$\nabla_{\mu}(h\delta\phi\nabla^{\mu}\phi)=\nabla_{\mu}h\nabla^{\mu}\phi\delta \phi+h\nabla_{\mu}(\delta\phi)\nabla^{\mu}+h\square\phi \delta\phi$$

$(4)+(5)$ becomes,

$$(4)+(5)=-2\sqrt{-g}[\nabla_{\mu}h\nabla^{\mu}\phi + \square\phi h]\delta\phi$$

So In summary I have found

$$\delta(\sqrt{-g}h(\phi)g^{\mu \nu}\nabla_{\mu}\phi\nabla_{\nu}\phi) = \sqrt{-g}\Big[\delta g^{\mu\nu}\big[-\frac{hg_{\mu\nu}}{2}\nabla_{\beta}\phi\nabla^{\beta}\phi+ h\nabla_{\mu}\phi\nabla_{\nu}\phi\big] + \delta \phi \big[\nabla_{\mu}\phi\nabla^{\mu}\phi\frac{\partial h}{\partial \phi} -2[\nabla_{\mu}h\nabla^{\mu}\phi + \square\phi h] \big]\Big]$$

seVenVo1d
  • 3,112
  • 14
  • 32

2 Answers2

1

In your variation of (1) you shouldn't use the same indices, so write $g_{\alpha \beta} \delta g^{\alpha \beta}$ instead. Then for (4) and (5) you'll have to integrate by parts in order to get $\delta \phi$ terms. I'll make a guess that the field variations in $\phi$ will vanish on your boundary, leaving terms in your integral of the form $\delta \phi \nabla (...)$. You should have seen something similar for a standard scalar field theory with partial derivatives.

This should be enough to solve your problem.

Eletie
  • 3,213
  • You can look at the edit. Note that I am also taking a varition for $\delta g^{\mu\nu}$ and $h\equiv h(\phi)$ and not a constant. – seVenVo1d May 22 '22 at 11:15
  • 1
    (4) and (5) do not look correct. It'll give the covariant derivative of everything, $-2\nabla_{\mu}(\sqrt{-g} g^{\mu \nu} h \nabla_{\nu} \phi) \delta \phi$, and $\nabla_{\mu} h \neq 0$ like you have implied. So your (#) equation is incorrect. – Eletie May 22 '22 at 11:22
  • So both $(#)$ and $(*)$ are wrong ..? – seVenVo1d May 22 '22 at 11:25
  • 1
    Yes, I wrote the term above, which can be expanded as $-2\sqrt{-g}( h \square \phi + \nabla^{\mu} h \nabla_{\mu} \phi)$. I encourage you to repeat this calculation. – Eletie May 22 '22 at 11:35
  • Is it correct now ? – seVenVo1d May 22 '22 at 14:19
  • Yes I think so. Also some more simplifications can be made in the final answer using $\nabla_{\mu} h = h' \nabla_{\mu} \phi$, and you should see how this cancels the first term in the bracket multiplying $\delta \phi$. Hope this helps! You can also see that the correct limit is recovered in $ h \rightarrow \rm{constant}$. – Eletie May 22 '22 at 14:38
  • thanks for the help – seVenVo1d May 22 '22 at 15:46
1

For 1 do not repeat the same index more than two times. As Eletie says, you may rename the indices in one of the terms, then your expression will be meaningful. 4,5 are the same. The whole term is a summation. You can rename $\mu \to \nu$ in 4 to obtain 5 since the metric is symmetric. In order to procced from there the answers to these question will be useful.

Derivation of Klein-Gordon equation in General Relativity

Variation of a scalar field

EDIT

Ok. So the term is

$$h(\phi)\nabla^{\mu}\delta\phi\nabla_{\mu}\phi = \nabla^{\mu}(h(\phi)\delta\phi\nabla_{\mu}\phi) - \frac{dh}{d\phi}\nabla^{\mu}\phi\nabla_{\mu}\phi \delta\phi - h(\phi)\Box\phi\delta\phi$$

The total derivative vanishes due to the boundary conditions.

Noone
  • 2,538