An elementary definition of the stress-energy tensor $T_{\mu\nu}$ in terms of the Lagrangian for flat spacetime is $$T^{\mu\nu}=\frac{\partial \mathcal{L}}{\partial(\partial_\mu\phi)}(\partial^\nu\phi)-\eta^{\mu\nu}\mathcal{L}.$$ For translational symmetries of spacetime both $T^{\mu\nu}$ is conserved: $$\partial_\mu T^{\mu\nu}=0.$$

But $T_{\mu\nu}$ is not always automatically symmetric in the indices $\mu,\nu$. Therefore, one considers a modified stress-energy tensor $\Theta_{\mu\nu}$ defined as $$\Theta^{\mu\nu}:=T^{\mu\nu}+\partial_\kappa A^{\kappa\mu\nu}$$ where $A^{\kappa\mu\nu}$ is an arbitrary tensor antisymmetric in the indices $\kappa,\mu$. This too satifies $$\partial_\mu\Theta^{\mu\nu}=0.$$

  • What is the need of defining a symmetric stress-energy tensor? What happens if I continue to work with $T^{\mu\nu}$ instead of the modified tensor $\Theta^{\mu\nu}$?

  • It is also unclear how one would choose the tensor $A^{\kappa\mu\nu}$ uniquely to symmetrize $T^{\mu\nu}$ and find a unique $\Theta^{\mu\nu}$.

  • 2
    $\begingroup$ First, we don't need the symmetrimized tensor be unique, we can choose arbitrary second order tensor whose divergence is zero to symmetrimize energy-stress tensor. Second, the reason we want to symmetrimize energy-stress tensor is due to the consideration of conservation law of angular momentum. Please check the last chapter of Classical Mechanics by Goldstein. $\endgroup$
    – GK1202
    Commented Apr 10, 2020 at 5:53
  • 1
    $\begingroup$ @y255yan Answers should not belong to the comments. Consider expanding that into a proper answer. $\endgroup$
    – MannyC
    Commented Apr 10, 2020 at 7:16
  • 1
    $\begingroup$ Related: physics.stackexchange.com/q/68564/2451 and links therein. $\endgroup$
    – Qmechanic
    Commented Apr 10, 2020 at 9:08
  • 1
    $\begingroup$ See Belinfante–Rosenfeld stress–energy tensor in Wikipedia. A motivation is to agree with the stress energy tensor used in general relativity. $\endgroup$ Commented Jun 26, 2022 at 14:27

1 Answer 1


An asymmetrical energy-momentum tensor implies orbital angular momentum non-conservation. This is why a trick is used to symmetrize such EM tensors. My take on this is that instead the Lagrangian should be corrected.

  • 2
    $\begingroup$ I do not understand well: the (conserved) total angular momentum is obtained by the Noether theorem. It is always conserved even if the stress energy tensor is not symmetric. Could you expand your answer? $\endgroup$ Commented Jun 26, 2022 at 14:18
  • $\begingroup$ @ValterMoretti I added 'orbital'. You are right that the total angular momentum is conserved for a sufficently symmetric lagrangian. $\endgroup$
    – my2cts
    Commented Aug 1, 2022 at 11:38

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.