Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Is there a fundamental misunderstanding here or have I made an algebraic slip?

I have a Riemannian metric of the form $ds^2={du^2+dv^2\over 1-u^2-v^2}$ on an open disc and I want to prove that radial curve $(r(t), 0)$ is a geodesic.

I wrote the metric in polar form -- $ds^2={dr^2+r^2d\theta^2\over 1-r^2}$, so the coefficients of the first fundamental form are $E={1\over 1-r^2}, F=0, G={r^2\over 1-r^2}$.

So the geodesic equations become ${d\over dt} (E\dot{r})={1\over 2}{\partial \over \partial r} (E\dot{r}^2)$ and (another equation which works).

I can always reparametrise $r(t)$ so that it has unit speed.

And I get $2r\over (1-r^2)^2$= $r\over (1-r^2)^2$, which is clearly wrong! Why is there an extra factor of 2?

Thank you.

ADDED: Perhaps it would be helpful for me to point out that generally, for a Riemannian metric of the form $Edu^2+2Fdudv+Gdv^2$ and geodesic $g(t)=(a(t),b(t))$, the Euler Lagrange equations are

${d\over dt} (E\dot{a}+F\dot{b})={1\over 2} (E_u\dot{a}^2+2F_u\dot{a}\dot{b}+G_u\dot{b}^2)$ and

${d\over dt} (F\dot{a}+G\dot{b})={1\over 2} (E_v\dot{a}^2+2F_v\dot{a}\dot{b}+G_v\dot{b}^2)$

where $\dot{a}={d\over dt}a,\,\,\,E_u={\partial\over\partial u}E$ and so on.

EUREKA! Ooh, I think I know what the bug is, The geodesic has constant speed, but the inner product here is not simply the Euclidean one which is what I assumed when I set $\dot{r}=1$! I think it works fine now. :) -- I write it here because I can't post it as an answer...

share|cite|improve this question
Anyone, please? – Sup Apr 30 '12 at 15:39
I'm not familiar with this form of the Euler equations. Do you have a reference for them? – Jason DeVito Apr 30 '12 at 15:46
@JasonDeVito: thanks for commenting!… offers a bit of info. But generally, for a Riemannian metric of the form $Edu^2+2Fdudv+Gdv^2$ and a geodesic $g(t)=(a(t),b(t))$, the Euler Lagrange equations are ${d\over dt} (E\dot{a}+F\dot{b})={1\over 2} (E_u\dot{a}^2+2F_u\dot{a}\dot{b}+G_u\dot{b}^2)$ and ${d\over dt} (F\dot{a}+G\dot{b})={1\over 2} (E_v\dot{a}^2+2F_v\dot{a}\dot{b}+G_v\dot{b}^2)$ – Sup Apr 30 '12 at 15:55
You actually can post it as an answer. – F M Apr 30 '12 at 17:53
@FernandoMartin: But I have to wait a further 3 hours since it says that I have too few "reputation points" so I can't post an answer within 8 hours of my asking the question... – Sup Apr 30 '12 at 17:54

This is for @JasonDeVito:

It is actually easier to use the original ("Cartesian") metric. Noting that for a curve $g(t)=(a(t),b(t)),\,\,\,||\dot{g}(t)||^2={\dot{a}^2+\dot{b}^2\over1-a^2-b^2}$ instead of simply $\dot{a}^2+\dot{b}^2$, then plugging and chugging gives the answer.

share|cite|improve this answer
Ok, apparently I cannot accept the answer within 2 days... – Sup May 1 '12 at 0:00
I appreciate you typing this up (and I upvoted), but I'm still not seeing where you use $\| \dot{g}\|$ in your computations. In fact, I'm still having a lot of trouble getting my computations to work out. For example, in the $\frac{d}{dt}(E\dot{a})$ term, a $\ddot{a}$ appears. Where does it go? Do you get rid of it by differentiating the length = 1 equation? If you wouldn't mind, I'd really like to see a few more details (but please, feel free to ignore me as well!) – Jason DeVito May 1 '12 at 2:11
@JasonDeVito : No probs, note that in the original coordinates, that curve is just $a=u, b=0$. ${d\over dt} ({1\over 1-u^2}\dot{u})={1\over 2}({\partial \over \partial u}({1\over 1-u^2})\dot{u}^2)$ – Sup May 1 '12 at 6:50
@JasonDeVito : Also it is now ${\dot{u}^2\over 1-u^2}=1$ not simply $\dot{u}=1$ – Sup May 1 '12 at 6:56

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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