We can define a divergence-less current from any Killing vector field $\xi^a$ by

$$ J^a[\xi] = \xi_b R^{a b}$$

I would like to show that

\begin{equation}\star J = \mathrm{d}\star\mathrm{d} \xi.\end{equation}

Or at least that

$$\int_\Sigma \star J = \int_\Sigma \mathrm{d}\star\mathrm{d} \xi.$$

This leads to the famous Komar integral.

Here are my attempts so far:



\begin{equation}\nabla_a \nabla_c \xi^a = R_{cb}\xi^b,\end{equation}

we can re-write the current as

$$ J^a[\xi] = \nabla_k \nabla_a \xi^k.$$

Then the Hodge dual is given by

$$ (\star J)_{abc} = \epsilon_{abcd} \nabla_k \nabla^d \xi^k.$$

RHS: Using the antisymmetry of $\xi$ it is easy to show that

$$\mathrm{d} \xi = 2 \nabla_d \xi_k.$$

Leading to

$$\star \mathrm{d} \xi = \epsilon_{abcd}\nabla^c \xi^d.$$


$$\mathrm{d} \star \mathrm{d} \xi = 3 \epsilon_{cd[ab}\nabla_{e]} \nabla^c \xi^d.$$

After that I am stuck. Could somebody please (!!) help me, I’ve spent a few hows on this already without success?


1 Answer 1


Let's compute this step by step.

Let $\xi^a$ be a Killing vector. We can lower the index to make it a one-form, so that we can apply the covariant derivative, $$(d\xi)_{ab} = 2 \nabla_{[a}\xi_{b]} = (\nabla_{a}\xi_{b}-\nabla_{b}\xi_{a}) = 2 \nabla_a\xi_b$$ where we used Killing's equation, $\nabla_a \xi_b+\nabla_b \xi_a=0$. Now we can apply the Hodge dual. In general, the Hodge dual of a $p-$form $\alpha$ is an $(n-p)-$form $\star \alpha$, with components given as follows: $$(\star\alpha)_{\mu_{1} \cdots \mu_{n-p}} = \frac{1}{p!} \alpha^{\nu_1\cdots \nu_p}\epsilon_{\nu_1 \cdots \nu_p \mu_1 \cdots \mu_{n-p}}$$ where $\epsilon$ is the Levi-Civita tensor, defined as

$$\epsilon_{\mu_1 \cdots \mu_n}=\sqrt{|\det{g}|}\, \widetilde{\epsilon}_{\mu_1 \cdots \mu_n}$$

where $\widetilde{\epsilon}$ is the Levi-Civita symbol, whose components are $\pm 1$. Then, applying this to the two-form $d\xi$ gives

$$(\star d\xi)_{ab} = \frac{1}{2}\epsilon_{abcd} (d\xi)^{cd} = \epsilon_{abcd} \nabla^c\xi^d$$

We can take an exterior derivative of this. In general, if $\alpha$ is a two-form, then

$$(d\alpha)_{abc} = 3 \nabla_{[a}\alpha_{bc]}$$

which in our case implies

$$(d\star d \xi)_{abc} = \frac{1}{4}\bigg[\nabla_a ((\star d\xi)_{bc})-\nabla_b ((\star d\xi)_{ac})+\nabla_c ((\star d\xi)_{ab})\bigg]=\\ =\frac{1}{4}\bigg[ \epsilon_{bcde} \nabla_a \nabla^d \xi^e -\epsilon_{acde} \nabla_b \nabla^d \xi^e+\epsilon_{abde} \nabla_c \nabla^d \xi^e\bigg]$$ where we used the fact that the Levi-Civita tensor is covariantly constant, see for example eq.s $(1.2.36-37)$ in these lecture notes.

Now we will compute the dual of this. According to the above definition,

$$(\star d\star d \xi)_f =\epsilon_{abcf} (d\star d \xi)^{abc}$$

Now we can use the following identity for $\epsilon$, which you can find on Wikipedia: $$\epsilon^{\mu_1 \cdots \mu_p \alpha_1 \cdots \alpha_{n-p}}\epsilon_{\mu_1 \cdots \mu_n \beta_1 \cdots \beta_{n-p}} = s\, p! (n-p)!\, \delta^{[\alpha_1}_{\beta_1} \delta^{\alpha_2}_{\beta_2} \cdots \delta^{\alpha_{n-p}]}_{\beta_{n-p}}$$ where $s$ is the sign of the determinant of the metric, in our case $s=-1$. Specialising to the case of interest for us, with $n=4$, $p=2$, we find

$$\epsilon^{abcd}\epsilon_{abef} = - 4\, \delta^{[c}_{e} \delta^{d]}_{f} = - 2\, (\delta^{c}_{e} \delta^{d}_{f}-\delta^{d}_{e} \delta^{c}_{f})$$

Therefore, we find

\begin{align*} (\star d\star d \xi)_f &=\frac{1}{4}\bigg[\epsilon_{abcf} \epsilon^{bcde} \nabla^a \nabla_d \xi_e -\epsilon_{abcf} \epsilon^{acde} \nabla^b \nabla_d \xi_e+\epsilon_{abcf} \epsilon^{abde} \nabla^c \nabla_d \xi_e\bigg]=\\ &=\frac{1}{4}\bigg[\epsilon_{bcfa} \epsilon^{bcde} \nabla^a \nabla_d \xi_e +\epsilon_{acbf} \epsilon^{acde} \nabla^b \nabla_d \xi_e+\epsilon_{abcf} \epsilon^{abde} \nabla^c \nabla_d \xi_e \bigg]= \\ &=-\frac{1}{2}\bigg[\nabla^a \nabla_f \xi_a - \nabla^a \nabla_a \xi_f + \nabla^b \nabla_b \xi_f-\nabla^b \nabla_f \xi_b+ \nabla^c \nabla_c \xi_f-\nabla^c \nabla_f \xi_c\bigg]=\\ &=-\frac{1}{2}\bigg[\nabla^a \nabla_a \xi_f-\nabla^a \nabla_f \xi_a\bigg] \end{align*}

Now we note the Ricci identity for Killing vectors, $$\nabla_a \nabla_b \xi^c = R^{c}_{\,\,bad} \xi^d$$ which implies $$\nabla^a \nabla_b \xi_a = R^{a}_{\,\,bad} \xi^d = R_{bd} \xi^d = J_b\\ \nabla_a \nabla^a \xi_c = R^{\,\,a}_{c\,\,ad} \xi^d = - R_{cd} \xi^d = -J_c$$

This means

$$(\star d\star d \xi)_f = J_f$$

or simply $\star d\star d \xi = J$. Now apply $\star$ on both sides, and note that $\star^2 = s (-1)^{p(n-p)}$ on a $p-$form in $n-$dimensional spacetime. So in this case $\star^2 = 1$ and we conclude that $\star J = d\star d \xi$.


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.