I have 3 questions:

  1. Is a vector perpendicular to any tangent vector of any null geodesic also a null vector?

  2. How can we find hypersurface to a null geodesic?

  3. Suppose we have a null geodesic in 4d, the transverse metric to it is 2d (You can look it up online. It is 2 dimensional) Intuitively, I don't understand how is this possible.

  • $\begingroup$ I think that the answer to 3 should be that the hypersurface is indeed 3 dimensional but it has a null eigen direction and hence, it's contribution to the metric on the hypersurface is zero. Hence, the transverse metric is 2 dimensional but the transverse space is 3 dimensional $\endgroup$ May 7, 2018 at 6:15

1 Answer 1



Nope. For example, in Minkowski spacetime, consider the vector $(v^\mu)=(1,0,0,1)$, which is a null vector. The vector $ (u^\mu)=(0,1,0,0) $ is orthogonal to it, yet, it is not a null vector.


I don't understand this question.


Consider a null geodesic with tangent vector $u^\mu$ ($u^\mu u_\mu$=0). Let $\lambda$ be the parameter along the null geodesic. Let $\Sigma_p<T_pM$ be the orthogonal complement to $u^\mu$ at $p\in M$. Note that because $u^\mu$ is a null vector, it is orthogonal to itself, hence $u_p\in\Sigma_p$. Let us choose two additional vectors in $\Sigma_p$, $e^\mu_1$ and $e^\mu_2$. We can choose these vectors such that $e^\mu_Ae^\nu_Bg_{\mu\nu}=\delta_{AB}$ ($A,B=1,2$), and because $u^\mu$ is a normal vector, we have $u^\mu e^\nu_A g_{\mu\nu}=0$.

The line element at $p$ can be expressed as $$ ds^2(p)=g_{\mu\nu}(p)u^\mu u^\nu d\lambda^2+2g_{\mu\nu}(p)e^\mu_A u^\nu d\sigma^Ad\lambda+g_{\mu\nu}(p)e^\mu_A e^\nu_B d\sigma^A d\sigma^B, $$ where $\sigma^A$ are some coordinates for which at $p$, $e^\mu_A$ are the coordinate basis vectors.

Writing in the relations between the basis vectors gives $$ ds^2(p)=g_{\mu\nu}(p)e^\mu_A e^\nu_B d\sigma^A d\sigma^B\equiv \delta_{AB}d\sigma^A d\sigma^B, $$ since the other contractions are all vanishing.

We can then see that if you make an infinitesimal displacement $d\xi=(d\lambda,d\sigma^1,d\sigma^2)$ that is orthogonal to the null curve, the contribution from $d\lambda$ is zero, hence it doesn't matter. After all, $\lambda$ is a null parameter, along which there is vanishing arc length, and so the physical/geometric displacement corresponding to $d\xi$ depends only on $d\sigma^1$ and $d\sigma^2$, which is why the metric is effectively two dimensional.

  • $\begingroup$ In the 2nd question, I am asking how can we find an orthogonal hypersurface to a congruence of null geodesics? $\endgroup$ May 7, 2018 at 15:21

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.