In the lecture notes of David Tong on String Theory he defines the energy momentum tensor of the polyakov action as \begin{align*} T_{\alpha\beta}=-\frac{2}{T}\frac{1}{\sqrt{-g}}\frac{\delta S}{\delta g^{\alpha\beta}}.\tag{1} \end{align*} Since the equation of motion for the dynamical metric $g_{\alpha\beta}$ can be obtained by $$\frac{\delta S}{\delta g^{\alpha\beta}}=0\tag{2}$$ of course this energy momentum tensor vanishes on-shell. I'm a bit confused about this statement. In GR we define the energy momentum tensor exactly the same way, but there of course the EMT does not vanish, but its derivative $\nabla_\mu T^{\mu\nu}$ does. So why don't we conclude in GR that the EMT also vanishes on-shell? What am I getting wrong here?

  • $\begingroup$ The partial derivative $\partial_\mu T^{\mu\nu}$ does not vanish! The covariant derivative $\nabla_\mu T^{\mu\nu}$ vanishes due to energy-momentum conservation. $\endgroup$ Commented Jun 5, 2023 at 10:52
  • $\begingroup$ Einstein's equation is basically the statement that the stress tensor of GR does vanish. We only get a non-zero stress tensor for the worldsheet of a string because of gauge fixing. $\endgroup$ Commented Jun 5, 2023 at 11:03
  • $\begingroup$ But we get a vanishing EMT on the worldsheet of a string right? I'm also not sure if einstein's field equation imply that the EMT is zero in non vacuum cases. $\endgroup$
    – Aralian
    Commented Jun 5, 2023 at 11:09
  • 2
    $\begingroup$ In GR, we define the EM tensor as $T_{\alpha\beta} = - \frac{2}{T} \frac{1}{\sqrt{-g}} \frac{\delta S_{matter}}{\delta g^{\alpha\beta}}$. Notice that only $S_{matter}$ appears here, NOT the whole action $S$. DTs statement applies if you define $T$ using the whole action. $\endgroup$
    – Prahar
    Commented Jun 5, 2023 at 11:43

1 Answer 1


There are at least 2 differences:

  1. The metric $g_{\alpha\beta}$ in the Polyakov action $S$ of string theory is a worldsheet metric, not the metric of spacetime (=target space).

  2. The action $S$ in OP's eq. (1) in the context of GR is a matter action (and the $T_{\alpha\beta}$ is a matter SEM tensor); it does not include the gravitational sector. In other words, the corresponding Euler-Lagrange (EL) equations $\frac{\delta S}{\delta g_{\alpha\beta}}=0$ are incomplete; they are not the full equations of motion for $g_{\alpha\beta}$. We need to include the gravitational Einstein tensor to form EFE.


Your Answer

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