As I was thinking about this other problem, I stumbled upon another issue (maybe easier to solve).

Suppose we have got a smooth manifold $M$, endowed with two different metrics ($g_1$,$g_2$). Then we have got two riemannian manifolds $M_1 := (M,g_1)$ and $M_2 := (M,g_2)$.

What is the Levi-Civita connection of $M_2$ when viewed in $M_1$? What is its meaning?

Example: $M_1 = (H^2,dx^2+dy^2)$ and $M_2 = (H^2,\frac{dx^2+dy^2}{y^2})$. What's the meaning of the connection of $M_2$ in the euclidean space?


This is more of a comment than an answer. I'm not entirely sure what kind of answer you're looking for / I don't necessarily have too much expertise in this area.

In general, I'm not sure if there is necessarily any meaningful relationship between the Levi-Civita connections induced by different metrics on the same smooth manifold. However, when the metrics are pointwise conformally related we have a neat, meaningful relationship.

Let $M_1=(M,g)$ be a smooth Riemannian manifold. Let $\lambda:M\to\mathbb{R}$ be a smooth function, define $\tilde{g}=e^{2\lambda}g$, and consider the pointwise conformally equivalent Riemannian manifold $M_2=(M,\tilde{g})$. Then we have $$\nabla^{\tilde{g}}_XY=\nabla^g_XY+(X\lambda)Y+(Y\lambda)X-g(X,Y)\mathrm{grad}(\lambda),$$ where $\nabla^g$ and $\nabla^{\tilde{g}}$ are the Levi-Civita connections of $M_1$ and $M_2$, respectively. This is Exercise 8.5 (has lots of hints) in do Carmo's Riemannian Geometry In terms of usefulness, we can use this formula to compute the Levi-Civita connection for a large class of smooth manifolds easily. I'm not sure what to say about it's "meaning", since that seems very problem specific / it depends on what the manifold and metrics represent.

  • $\begingroup$ Thank you for your answer! Yes, maybe you're right, I didn't quite explain thoroughly what I was looking for. Let's take my example with the euclidean and hyperbolic metrics, for instance, and say we take a straight line $c(t) = vt + x_0$. Then one can compute $\nabla^{H^2}_{c'(t)}c'(t)$, and what one finds out is that this vector field resembles very much a force that prevents the path to bend into a geodesic of the hyperbolic plane: if for example $v=(1,0)$, my calculation yields $\nabla^{H^2}_{v}v = (0,1/y)$, a field pointing upward (thus not making the path bend into a circumference). $\endgroup$ – Alessio Di Lorenzo Aug 5 '17 at 20:10
  • 1
    $\begingroup$ If, on the other hand, $\nabla^{H^2}_{c'(t)}c'(t) = 0$, there is no "force" trying to resist the "bending", and then the path is indeed a geodesic. Anyway, in this case your answer is very useful and I will try and put it to good use! $\endgroup$ – Alessio Di Lorenzo Aug 5 '17 at 20:12

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.