If I write (my understanding of) how to derive the equations of motion from the Polyakov action, I come up missing a term.

Beginning with the basic Polyakov action
\begin{equation} S = -\frac{T}{2}\int\sqrt{-h}h^{ab}g_{\mu\nu}\partial_aX^\mu\partial_bX^\nu\,d^2\xi \end{equation} the Lagrangian is \begin{equation} \mathcal{L}=-\frac{T}{2}\sqrt{-h}h^{ab}\partial_aX^\mu\partial_bX^\nu \end{equation} which after plugging into the Euler-Lagrange equation is \begin{equation} 0-\frac{\partial}{\partial \xi^b}\left[-\frac{T}{2} \sqrt{-h}h^{ab}\partial_aX^\nu\right]=0 \end{equation} which leads to (my result of) \begin{equation} \partial_b\left[ \sqrt{-h}h^{ab}\partial_a X^\nu\right] =0 \end{equation} The equation of motion should be \begin{equation} \Box X^\nu = \frac{1}{\sqrt{-h}}\partial_b\left[ \sqrt{-h}h^{ab}\partial_a X^\nu\right] =0 \end{equation} My question is that I do not see where the $1/\sqrt{-h}$ comes from. (This result is ok for the Polyakov action on its own, but I want to be able to add a mass term to the action which will make the $1/\sqrt{-h}$ matter then.)

I think that there is something fundamental that I am missing about deriving the equation of motion. I understand that the operator $\nabla_\mu X^\nu=\partial_\mu X^\nu +\Gamma^\nu_{\mu\lambda}X^\lambda$, but I am bothered by deriving the equations of motion when the Lagrangian density is $L(X^\nu,\partial_aX^\nu)$ instead of $L(X^\nu,\nabla_aX^\nu)$. If I derive the Euler-Lagrange equation of motion from varying an action containing $L(X^\nu,\partial_aX^\nu)$: \begin{eqnarray} \delta S &=& \int \delta L (X^\nu,\partial_a X^\nu)\sqrt{-g}d^nX \\ &=& \int \left[ \frac{\delta L}{\delta(\partial_aX^\nu)}\delta(\partial_aX^\nu) + \delta X^\nu\nabla_a\frac{\delta L}{\delta(\partial_aX^\nu)}\right]\sqrt{-g}d^nX + \int \left[ \frac{\delta L}{\delta X^\nu}-\nabla_a\frac{\delta L}{\delta (\partial_a X^\nu)}\right]\delta X^\nu\sqrt{-g}d^nX \end{eqnarray} can I still send the surface term (the first term above) to zero? Thanks for any insight.

  • $\begingroup$ Your result and the "right" result differ just by an overall factor. Just multiply both sides of equation $\Box X^{\mu}=0$ by $\sqrt{-h}$ and you are done. $\endgroup$ Feb 11, 2017 at 6:13
  • $\begingroup$ I can see that. The problem is that the $1/\sqrt{-h}$ term should come from solving the Euler-Lagrange equations, but I am finding that it does not (and I want to know what I am missing here). In the minimal theory (with no matter action) it won't make a difference, but I imagine that this will be important if there is a matter term added to the action. $\endgroup$
    – Bob
    Feb 11, 2017 at 7:09
  • $\begingroup$ The point is that both equations are equivalent. You obtain one or the other by multiplying or dividing by $\sqrt{-h}$ $\endgroup$ Jul 2, 2020 at 3:35

1 Answer 1


It's easier to compute the equations of motion by using variations of the inverse metric.

$$ \frac{\delta S}{\delta h^{ab}}=0 $$ You can use the distributive property of the variational operator and make use of the identity $$ \frac{1}{\sqrt{-h}}\frac{\delta\sqrt{-h}}{\delta h^{ab}}=-\frac{1}{2}h_{ab} $$

  • 2
    $\begingroup$ Judas503- That would give e.o.m for h_ab not X^mu $\endgroup$
    – Ravi Singh
    Feb 18, 2023 at 4:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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