A small (and perhaps trivial :-() calculation is troubling me somewhat.

Consider the action of a right-handed Dirac fermion coupled to external gravity (a background gravitational field). The action can be written as

$$S = \int\sqrt{-\text{det}(g)}\, i\overline{\psi}_{R}\gamma^\mu\left(\partial_\mu + \frac{1}{2}\omega_{\mu}\right)\psi_{R} $$

where $\omega_\mu = {\omega_\mu}^{ab}\Sigma_{ab}$ and $\Sigma_{ab} = \frac{1}{4}[\gamma_a, \gamma_b]$. Here $\mu, \nu$ denote curved indices and $a, b$ denote tangent space (flat) indices. Now, the energy-momentum tensor is given by

$$T^{\mu\nu} = \frac{2}{\sqrt{-\text{det}(g)}} \frac{\delta S}{\delta g_{\mu\nu}}$$

To vary the action with respect to the metric, there are three pieces as I see it:

  1. Variation of the $\sqrt{-\text{det}(g)}$ piece. This is easy.
  2. Variation of the pieces with Lorentz indices: this is the part that I seem to have gotten a bit confused about.

Indeed $\gamma^\mu = \gamma^a e_a^\mu$ and so $\delta \gamma^\mu = \gamma^a \delta e^\mu_a$, and

$$\delta g_{\mu\nu} = \delta(\eta_{ab}e_{\mu}^a e_{\nu}^b) = \eta_{ab}[(\delta e^a_\mu)e_{\nu}^b + e^a_\mu \delta e_{\nu}^b]$$

But what happens on varying the operator $\gamma^\mu(\partial_\mu + \frac{1}{2}\omega_\mu)$ with respect to the metric?

One should ideally get

$$T^{\mu\nu} = \frac{i}{4}\overline{\psi}_{R}\gamma_\mu\overleftrightarrow{\nabla}_\nu\psi_R + (\mu \leftrightarrow \nu)$$

from this calculation. Here, $\nabla$ denotes the spinorial covariant derivative.


You can write

$\gamma^\mu(\partial_\mu + \frac{1}{2} \omega_\mu) = \gamma_\nu g^{\nu \mu}(\partial_\mu + \frac{1}{2} \omega_\mu) = \gamma_\nu g^{\mu \nu}(\partial_\mu + \frac{1}{2} \omega_\mu)$. (A)


$\gamma^\mu(\partial_\mu + \frac{1}{2} \omega_\mu) = \gamma_\mu g^{\mu \nu}(\partial_\nu + \frac{1}{2} \omega_\nu)$. (B)

If you add equations A and B you obtain

$2 \gamma^\mu(\partial_\mu + \frac{1}{2} \omega_\mu) = \gamma_\nu g^{\mu \nu}(\partial_\mu + \frac{1}{2} \omega_\mu)+\gamma_\mu g^{\mu \nu}(\partial_\nu + \frac{1}{2} \omega_\nu)$

and then differentiate by $g^{\mu \nu}$ to obtain the identity.

  • $\begingroup$ Thanks! That is very neat. But don’t you also generate $-g_{\mu\nu}i\overline{\psi}_R\nabla\psi_R$ in the functional differentiation wrt the background metric? $\endgroup$ – leastaction Jul 23 '17 at 22:18
  • $\begingroup$ The product $(\omega_\mu \psi_R)^a$ with spinor indices $a,b$ is equivalent to $\Omega_\mu^{ab} \psi_{Rb}$; this can also be transformed to with Greek letters denoted Lorentzian indices (with tetrads) to $\Omega_\mu^{\alpha \beta} \psi_{R \beta}$. Now use the same trick as showed above with representing contraction of indices with metric Tensor and derive with respect to it. $\endgroup$ – kryomaxim Jul 24 '17 at 14:10

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.