1
$\begingroup$

For a given scalar field $\phi$, the stress energy tensor is \begin{equation} T_{\mu\nu}=\partial_{\mu}\phi\partial_{\nu}\phi - \frac{1}{2} g_{\mu\nu}(\partial_{\alpha}\phi\partial^{\alpha}\phi). \end{equation} How can I get a second order differential equation for $\phi$, assuming that the metric $g_{\mu\nu}$ satisfies Einstein's equation? I tried to use the identity $\nabla^{\mu}G_{\mu\nu}=0$, getting the following equation \begin{equation} 0=\nabla^{\mu}T_{\mu\nu}=g^{\mu\rho}\nabla_{\rho}T_{\mu\nu}=g^{\mu\rho}(\partial_{\rho}T_{\mu\nu}-\Gamma^{\alpha}_{\rho\mu}T_{\alpha\nu}-\Gamma^{\alpha}_{\nu\rho}T_{\mu\alpha}). \end{equation} But turns out that this is not a second order differential equation. How can I get such equation?

$\endgroup$
  • 1
    $\begingroup$ Are you sure this is the stress tensor you want to use? Usually the stress tensor for a scalar field is quadratic in the field. I would have expected the first term in your stress tensor to be $$\partial_\mu \phi \partial_\nu\phi,$$ instead of $$\partial_\mu \partial_\nu \phi.$$ In that case, you could get the field equation by just requiring that the stress tensor is divergence-free. $\endgroup$ – asperanz Dec 30 '17 at 22:05
  • $\begingroup$ You're right, it's $\partial_{\mu} \phi \partial_{\nu} \phi$. That's what I've tried to do without any success. $\endgroup$ – Elias Dec 31 '17 at 18:31
1
$\begingroup$

Note that what you did does not provide a second order differential equation for $\phi$, since $T_{\mu\nu}$ is already second order in the derivatives.

The stress-energy tensor of a matter sector is given by: $$ T_{\mu\nu}=\frac{1}{\sqrt{g}}\frac{\delta S_m}{\delta g^{\mu\nu}} $$

so we need to find an action $S_m=\int d^4x\,\sqrt{g}\,\mathcal{L}[\phi,\partial_\mu\phi]$ such that $$ T_{\mu\nu}=\partial_{\mu}\phi\partial_{\nu}\phi - \frac{1}{2} g_{\mu\nu}(\partial_{\alpha}\phi\partial^{\alpha}\phi) $$ and since $\phi$ is scalar, and $T_{\mu\nu}$ is second order in the derivatives, we can easily conclude that: $$ \mathcal{L}=g^{\mu\nu}\partial_{\mu}\phi\partial_{\nu}\phi $$

Now, you calculate the equations of motion for $\phi$:

$$ \frac{\delta S_m}{\delta \phi}=0 $$ that is: $$ \partial_{\mu}(\sqrt{g}\,\frac{\partial\mathcal{L}}{\partial(\partial_\mu\phi)})=0 $$ a second order differential equation for $\phi$, for a given $g_{\mu\nu}$ determined.

$\endgroup$
  • $\begingroup$ I'm not familiar with the Lagrangian formulation of RG. I've been told that one way to obtain the equation is by means of the Bianchi identities $\nabla_{[\mu}R_{\nu\gamma\delta]}=0$ and Einstein's equation. $\endgroup$ – Elias Dec 30 '17 at 2:08
  • $\begingroup$ I simply use the definition of $T_{\mu\nu}$. There is no mention to the lagragian of $g_{\mu\nu} $. Your proposal, of using $\nabla_{[\mu}R_{\nu\gamma]\delta\lambda}=0$ will not work. You are going to find the conservation of $T_{\mu\nu}$ again. See this in the part of 'Features'/Comservation of energy and momentum $\endgroup$ – Nogueira Dec 30 '17 at 2:22
  • $\begingroup$ After the OP's edit, $T_{\mu \nu}$ only contains first derivatives of the scalar field, which means that requiring its divergence to be zero will result in a second-order differential equation in $\phi$. $\endgroup$ – Michael Seifert Jan 10 '18 at 19:21
  • $\begingroup$ @MichaelSeifert yes, you are right. $\endgroup$ – Nogueira Jan 10 '18 at 20:11
1
$\begingroup$

HINT: It'll be easiest to just write the stress-energy tensor in terms of the covariant derivative (note that these are equivalent for scalar fields): $$ T_{\mu \nu} = \nabla_\mu \phi \nabla_\nu \phi - \frac{1}{2} g_{\mu \nu} \nabla^\rho \phi \nabla_\rho \phi $$ Now take the divergence of $T_{\mu \nu}$ without writing down any Christoffel symbols; leave all your derivatives in terms of $\nabla_\mu$, rather than $\partial_\mu$. Use the facts that $\nabla_\mu$ obeys the product rule, and that by definition $\nabla_\mu g_{\rho \sigma} = 0$.

BTW, the assumption that $g_{\mu \nu}$ satisfies Einstein's equations is completely unnecessary for this derivation.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.