For all $p \in \mathbb{R} \smallsetminus \lbrace 0,1 \rbrace$, let $\displaystyle M(p)=\frac{1}{p^{2}(p-1)^{2}}$. Then, let $g_{p} \, : \, (u,v) \, \longmapsto \, uM(p)v$.

I am not sure about the following assertions :

  • Let $M=\mathbb{R} \smallsetminus \lbrace 0,1 \rbrace$. $\big( g_{p} \big)_{p \in M}$ defines a Riemannian metric on $M$. Hence, $M$ equipped with this metric is a Riemannian manifold.
  • $M$ cannot be geodesically complete for this metric because, according to Hopf-Rinow theorem, $(M,\vert \cdot \vert)$ would be a complete metric space (which it is not).

I think both are true but I'd feel better if someone could validate or invalidate these assertions.


You have modified the standard metric by a conformal factor function which is always positive so this certainly defines a Riemannian metric.

On the other hand, the metric resembles a hyperbolic metric on the y-axis near the poles of the conformal factor, so it is actually complete "near" these points. Whether or not it is complete "at infinity" I leave to you as an exercise (hint: integrate).

  • $\begingroup$ Thank you for the answer but what about the Hopf-Rinow theorem ? If $M$ equipped with this metric was complete, then $M$ as a metric space would be complete as well, right? $\endgroup$ – jibounet Jul 31 '14 at 12:45
  • $\begingroup$ Whether or not it is incomplete depends on whether you run off to infinity in finite time. This is consistent with Hopf-Rinow. $\endgroup$ – Mikhail Katz Jul 31 '14 at 12:51
  • $\begingroup$ I have written the differential equation for geodesics and solved it. I might be mistaken but the geodesics but not all the geodesics are defined on $\mathbb{R}$. Therefore, $M$ would be incomplete. $\endgroup$ – jibounet Jul 31 '14 at 13:00
  • $\begingroup$ Here you don't even need the geodesic equation. Integrate the conformal factor to show that you run off to infinity in finite time. $\endgroup$ – Mikhail Katz Jul 31 '14 at 14:04
  • 1
    $\begingroup$ Ok sorry, I misunderstood "arclength". If the curve is parametrized by arclength, its length is equal to the length of the interval on which it is defined (say $t \, \mapsto \, \tilde{\gamma}$ (with $a \leq t \leq b$) is the curve parametrized by arclength) then $L(\tilde{\gamma}) = b-a$. $\endgroup$ – jibounet Jul 31 '14 at 16:15

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.