On p43 of Polchinski's book, it says that under the world-sheet translation $\sigma^a\rightarrow\sigma^a+\epsilon v^a$, $X^\mu\rightarrow X^\mu-\epsilon v^a\partial_a X^\mu$. And $$j^a=iv^b T_{ab},$$ $$T_{ab}=-\frac1{\alpha'}:(\partial_aX^\mu\partial_bX_\mu-\frac12\delta_{ab}\partial_c X^\mu\partial^cX_\mu):$$ Could someone please tell me where such expressions of $j^a$ and $T^{ab}$ come from? I assume that we start with the action $$S=\frac1{4\pi\alpha'}\int d^2\sigma\,(\partial_a X^\mu\partial_1 X_\mu+\partial_2 X^\mu\partial_2 X_\mu)=\frac1{2\pi\alpha'}\int d^2z\,\partial X^\mu\bar\partial X_\mu,$$ but I don't see how we can apply the definition of $T_{ab}$ from GR...

Further, on the next page, when expressed in complex coordinates, it says $$T(z)=-\frac1{\alpha'}:\partial X^\mu\partial X_\mu:,~~~~~\tilde T(z)=-\frac1{\alpha'}:\bar\partial X^\mu\bar\partial X_\mu:$$ How do you get here from the $T_{ab}$ above?


2 Answers 2


1) First, looking at $(2.3.4)$, you see that $j^a$ is the coefficient of $\partial_a\rho$. An application of this $(2.3.12), (2.3.13)$. To make connection with this formalism, it is preferable to choose the variations :

$X^\mu\rightarrow X^\mu-\epsilon \rho(\sigma) v^c\partial_c X^\mu$

From this, we get :

$\partial_a X^\mu\rightarrow \partial_aX^\mu-\epsilon \partial_a(\rho(\sigma) v^c\partial_c X^\mu)$

Starting with an euclidean metric $\delta_{ab}$ (so the determinant is $1$), we have : $S = \dfrac{1}{4 \pi \alpha'} \int d^2\sigma \,(\partial^aX^\mu)(\partial_aX_\mu)$

The variation is then :

$\delta S =-\dfrac{\epsilon}{2 \pi \alpha'}\int d^2\sigma \,(\partial^aX^\mu) \partial_a(\rho(\sigma) v^c\partial_c X_\mu) \\ =-\dfrac{\epsilon}{2 \pi \alpha'}\int d^2\sigma[(v^c\partial^aX^\mu \partial_c X_\mu) \partial_a(\rho(\sigma)) + \rho(\sigma) (\partial^dX^\mu \partial_d(v^c\partial_c X_\mu))] \\=-\dfrac{\epsilon}{2 \pi \alpha'}\int d^2\sigma[(v^c\partial^aX^\mu \partial_c X_\mu) \partial_a(\rho(\sigma)) + \rho(\sigma) ( \partial_c (\frac{1}{2}v^c\partial^d X^\mu \partial_d X_\mu)]$

Now, we perform an integration by parts on the second term (where we eliminate the total derivative which gives a surface integral), so finally we have :

$\delta S= \dfrac{\epsilon}{2 \pi \alpha'}\int d^2\sigma[v^c\partial^aX^\mu \partial_c X_\mu) \partial_a(\rho(\sigma)) - \partial_c(\rho(\sigma))(\frac{1}{2}v^c\partial^d X^\mu \partial_d X_\mu)] \\ =\dfrac{\epsilon}{2 \pi \alpha'}\int d^2\sigma [v^c(\partial^aX^\mu \partial_c X_\mu -\frac{1}{2}\delta^a_c \partial_d X^\mu \partial^d X_\mu))\partial_a(\rho(\sigma))] \\ = -\dfrac{\epsilon}{2 \pi}\int d^2\sigma \, (v^c\,T_c^a)\partial_a(\rho(\sigma))$

Comparing to $(2.3.4)$ gives : $j^a = i v^c\,T_c^a$, or $j_a= i v^cT_{ac}$

2) @Prahar has already indicated you how come the expression of $T_{ab}$ from the action, just remember that : $\dfrac{ \partial \sqrt{\gamma}}{\partial \gamma^{ab}} = - \frac{1}{2} \sqrt{\gamma} \gamma_{ab}$

3) For you last qestion, remember that, in a general metric $g$, we have : $T_{ab}=-\frac1{\alpha'}:(\partial_aX^\mu\partial_bX_\mu-\frac12 g_{ab}\partial_c X^\mu\partial^cX_\mu):$

Now, look at $(2.1.6$, and you see that the choosen metrics for $(a,b)= (z, \bar z)$ has $g_{zz}=g_{\bar z \bar z}=0$. The stress-energy tensor has zero trace (by symmetry, $2 g^{z\bar z} T_{z\bar z}=0$), so finally $T_{z\bar z}= T_{\bar z z}=0$.

The only non-zero components are then $T(z)=T_{zz}= -\frac1{\alpha'}:\partial_z X^\mu \partial_z X_\mu:$, and $\bar T(\bar z)=T_{\bar z\bar z}= -\frac1{\alpha'}:\partial_\bar z X^\mu \partial_\bar z X_\mu:$

  • $\begingroup$ Greatly appreciated!! $\endgroup$
    – user46348
    Sep 2, 2014 at 19:10

The action is $$ S = \frac{1}{2\pi \alpha'} \int d^2 \sigma \sqrt{\gamma} \gamma^{ab} \partial_a X^\mu \partial_b X_\mu $$ The definition of the stress tensor from GR is $$ T_{ab} = \lambda\frac{4\pi}{\sqrt{\gamma}} \frac{ \delta S}{ \delta \gamma^{ab}} $$ Usually $\lambda = 1$, but different books use different conventions. I do not remember what convention Polchinski uses for the definition of the stress tensor. You can use the formula above to determine the stress-tensor.

The current $j^a$ can be obtained from the Noether procedure. However, it can be related to the stress-tensor as follows:

The definition of the stress tensor above implies that if we perform a metric deformation $\gamma_{ab} \to \gamma_{ab} + \delta \gamma_{ab}$, the action transforms as $$ S \to S + \frac{1}{4\pi \lambda } \int d^2 \sigma \sqrt{\gamma} T^{ab} \delta \gamma_{ab} $$ Now, under a coordinate change $\delta \sigma^a = \epsilon v^a \implies \delta \gamma_{ab} = \nabla_a (\epsilon v_b ) + \nabla_b (\epsilon v_a )$. Thus, we note that a coordinate transformation can be "converted" to a metric transformation. Under such a metric transformation, the action transforms as $$ S \to S + \frac{1}{2 \pi \lambda } \int d^2 \sigma \sqrt{\gamma} T^{ab} \left[ \nabla_a \epsilon v_b + \epsilon \nabla_a v_b \right] $$ where we have used symmetry of the stress tensor.

But, we are being naive! What about the transformation of $X^\mu$?? The action will undergo an extra transformation due to that. However, since the above transformation is a symmetry for $\epsilon$ constant, the effect of the $X^\mu$ transformation will be to simply cancel the term proportional to $\epsilon$ above. Only the term depending on the derivative of $\epsilon$ survives. Thus, under the full transformation, we must have $$ S \to S + \frac{1}{2 \pi \lambda } \int d^2 \sigma \sqrt{\gamma} T^{ab} v_b \nabla_a \epsilon $$ The current is then propertional to $$ j^a \propto T^{ab} v_b $$ The proportionality constant is again a matter of convention. I'm not sure what conventions are followed in Polchinski.


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.