In this work, Wald & Zoupas developed a framework to define the "conserved quantities" in a diffeomorphism-invariant theory using the covariant phase space formalism.

Let $\xi^a$ be a vector field on the $n$-dimensional spacetime manifold $M$. Physical fields are collectively represented by $\phi$, which could be the metric $g_{ab}$ and the 4-potential $A_a$, etc.. $\xi^a$ will generate an infinitesimal diffeomorphism, which is a local symmetry of the theory, so the variation of the Lagrangian $n$-form $\boldsymbol L$ is

$$\mathscr L_\xi\boldsymbol L=\boldsymbol E\hat\delta\phi+d\boldsymbol\Theta(\phi,\mathscr L_\xi\phi)=d(i_\xi\boldsymbol L),$$

where $\hat\delta\phi=\mathscr L_\xi\phi$, $\boldsymbol E$ represent the equations of motion, $d$ means the exterior derivative, $\boldsymbol\Theta$ is a $(n-1)$-form called the presymplectic potential current, $i_\xi$ refers to the interior product.

The Noether current $(n-1)$-form $\boldsymbol j$ is given by

$$\boldsymbol j[\xi]=\boldsymbol\Theta(\phi,\mathscr L_\xi\phi)-i_\xi\boldsymbol L.\tag{9}$$

On shell, one can show that $d\boldsymbol j=0$ and locally, one can find a $(n-2)$-form $\boldsymbol Q[\xi]$ such that $\boldsymbol j[\xi]=d\boldsymbol Q[\xi]$. $\boldsymbol Q$ is the Noether charge $(n-2)$-form. In this reference, the general form of $\boldsymbol Q$ was given in Proposition 4.1, which is really complicated.

Now, in Wald & Zoupas, a different vector field $\eta^a$ is considered. Its effect on $\boldsymbol Q$ is given by Eq. (33),

$$\delta\boldsymbol Q[\xi]=\mathscr L_\eta\boldsymbol Q[\xi]-\boldsymbol Q[\mathscr L_\eta\xi].\tag{33}$$

However, I do not understand how this relation was obtained. It looks to me that the left-hand side is nothing but

$$\delta\boldsymbol Q[\xi]=\mathscr L_\eta(\boldsymbol Q[\xi]).$$

Clearly, this is not the option.

Please help me with this problem.


It seems Wald & Zoupas mean

$$(\delta\boldsymbol{Q})[\xi]~=~(\mathscr{L}_\eta\boldsymbol{Q})[\xi]~=~\mathscr{L}_\eta (\boldsymbol{Q}[\xi]) ~-~\boldsymbol{Q}[\mathscr{L}_\eta\xi].\tag{33}$$

  • $\begingroup$ Thanks! But $\delta\boldsymbol Q[\xi]$ should mean $\delta(\boldsymbol Q[\xi])=\mathscr L_\eta(\boldsymbol Q[\xi])$. This can be seen from Wald's earlier paper arxiv.org/abs/gr-qc/9307038. Eq. (14) is related to what I am asking for here. There, $\delta$ acts on $\boldsymbol j[\xi]$, not just on $\boldsymbol j$, otherwise, the right-hand side of Eq. (14) would have an extra term, something like $\boldsymbol j[\delta\xi]$. $\endgroup$ – Drake Marquis Jan 22 at 2:53
  • $\begingroup$ In most of Wald et al's writings the 2 operations $\delta$ and $\mathscr{L}_{\eta}$ are indeed unrelated. However, they make an exception around eq. (33). $\endgroup$ – Qmechanic Jan 23 at 13:53

I contacted R. Wald. He told me that the variation operation $\delta$ does not act on $\xi^a$. So

$$\delta\boldsymbol Q[\xi]\ne\mathscr L_\eta (\boldsymbol Q[\xi]).$$

But they wanted to use Lie derivative to carry out the actual computation, so they allowed $\delta$ acts on $\xi^a$, and then remove the action of $\delta$ on $\xi^a$ by subtracting $\boldsymbol Q[\mathscr L_\eta\xi]$ from $\mathscr L_\eta\boldsymbol Q[\xi]=\mathscr L_\eta(\boldsymbol Q[\xi])$. This is why Eq. (33) holds in Wald & Zoupas.


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.