Is there a difference, conceptually speaking, between solving the geodesic equations using $\lambda$ as an arbitrary parameter vs substituting a coordinate from the metric in it's place?

For instance, if I have:

$\frac{d^2\theta}{d\lambda^2}=-\Gamma^{\theta}_{t\theta}\frac{dt}{d\lambda}\frac{d\theta}{d\lambda}-\Gamma^{\theta}_{p\theta}\frac{dp}{d\lambda}\frac{d\theta}{d\lambda}$, and I decide $\lambda=\theta$, I get $0=-\Gamma^{\theta}_{t\theta}\frac{dt}{d\theta}-\Gamma^{\theta}_{p\theta}\frac{dp}{d\theta}$, which I can multiply by $d\theta$ and then integrate for $t$ and $p$. But I know the answer must not be true for all geodesics, so I have to ask, what does using $\theta$ as $\lambda$ specify?

By the way, if it would help to explicity see what the metric and Christoffel symbols are, I could include those.

  • $\begingroup$ In general, the geodesic might have multiple places at which $\theta$ has a particular value. You seem to be assuming it’s single-valued and therefore $t$ and $p$ are well-defined functions of $\theta$. $\endgroup$
    – Ghoster
    Nov 30, 2022 at 21:51
  • $\begingroup$ I understood that there are 3 variables $t$, $p$ and $\theta$. In this case the geodesic equation should have much more than 2 terms at the right side. $\endgroup$ Nov 30, 2022 at 22:53
  • $\begingroup$ The metric is $-dt^2+dp^2+(5p^2+4t^2)d\theta^2$, so there are in fact only two Christoffel symbols with an upper index of $\theta$. So I believe the above geodesic equation is correct. There are no killing vectors other than $(5p^2+4t^2)d\theta$, so I don't see a way to solve for geodesics analytically. Do you think that some coordinate transformation could change that? $\endgroup$
    – user345249
    Dec 1, 2022 at 1:07
  • $\begingroup$ By the way, all this comes from the Wikipedia article for Ellis wormholes in the section entitled "Dynamic Ellis Wormhole." $\endgroup$
    – user345249
    Dec 1, 2022 at 1:10

1 Answer 1


When we use the Euler-Lagrange equations to find the stationary action of a path with a given parametrization in a manifold with a metric $g$: $$s_{ab} = \int _a^b \sqrt {\frac{\partial X^i}{\partial \lambda}\frac{\partial X^j}{\partial \lambda}g_{ij}(λ)}dλ$$

The resulting differential equation looks like the geodesic equation in the left side, with the connection defined as the metric one:

$$\frac{\partial u^p}{\partial \lambda} + \frac{1}{2}g^{pk}(\partial_i g_{kj} + \partial_j g_{ki} - \partial_k g_{ij})u^iu^j $$

Where $$u^a = \frac{\partial X^a}{\partial \lambda}$$

But the right side is zero only if $\lambda$ is an affine function of the arc length $s$.

That means: while it is possible to choose a parameter $\lambda = f(s)$, where $f$ is not an affine function, the geodesic in this case is not the shortest length, what (as far as I know) it should be in GR.

There a discussion about the issue on this answer.

  • $\begingroup$ Oh, so $\lambda=as+b$ (where $a$ and $b$ are constants)? $\endgroup$
    – user345249
    Dec 2, 2022 at 4:40
  • $\begingroup$ @user345249 yes. $\endgroup$ Dec 2, 2022 at 18:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.