Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

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

Lets say I have two (compact) manifolds $U$,$V$ and a diffeomorphism $\psi:U\rightarrow V $.

The shortest way between two points $a$ , $b \in V$ is given by a parametrisation $\gamma :W \rightarrow V $ which I found using Euler-Langrange-Equation. Now I want to have a parametrisation $\eta: W \rightarrow U$ with the same property, that means $\psi(\eta(t))=\gamma(t)$.

Moreover, I need more theory according to lengths on manifolds, I minimized the length using $L=\int \left \|\gamma '(t)) \right \|dt$ but I don't know if this is an elegent way of doing it. The problem which I want to solve is: what is the Length of to points $a,b\in U$ regarding to $V$, without transforming to $V$, finding a parametrisation and integrating over it.

share|cite|improve this question
The Euler-Lagrange equation for geodesics is easier to work with if you minimize the energy $\int \|\gamma'\|^2$ instead of the length $\int \|\gamma'\|$. The energy is better behaved, and its minimizers are geodesics parametrized with constant speed. – user53153 Feb 11 '13 at 0:20

When you are talking about lengths of curves at least $V$ is a Riemannian manifold to begin with: For any system of local coordinates $(v_i)_{1\leq i\leq n}$ on $V$ a metric tensor $v\mapsto\bigl(g_{ij}(v)\bigr)$ is given such that the length of a curve $$\gamma:\quad t\mapsto v(t)\in V\qquad(a\leq t\leq b)$$ computes to $$L(\gamma)=\int_a^b\sqrt{\sum_{i,j} g_{ij}\bigl(v(t)\bigr)\dot v_i(t)\dot v_j(t)}\ dt\ .$$ If a diffeomorphism $\psi:\ U\to V$ is given then any curve $$\gamma:\quad t\mapsto u(t)\in U\qquad(a\leq t\leq b)$$ has an image curve $$\gamma':=\psi(\gamma):\quad t\mapsto \psi\bigl(u(t)\bigr)\in V\qquad(a\leq t\leq b)\ .$$ You then wish to define the length of $\gamma$ as $\tilde L(\gamma):=L\bigl(\psi(\gamma)\bigr)$. It turns out that this idea can be implemented without reference to a particular $\gamma$. What is at stake here is the so-called pullback of the given Riemannian metric $(g_{ij})$ on $V$ to a Riemannian metric $u\mapsto \bigl(\tilde g_{kl}(u)\bigr)$ on $U$.

The $\tilde g_{kl}(u)$ can be computed from the $g_{ij}(v)$ and the Jacobian $\left[{\partial\psi_i\over\partial u_k}\right]_{ik}$ without eigenvalues or inverting a matrix. One just has to plug in $v(t)=\psi\bigl(u(t)\bigr)$ into $$\sum_{i,j} g_{ij}\bigl(v(t)\bigr)\dot v_i(t)\dot v_j(t)$$ and to apply the chain rule. It then becomes a matter of linear algebra.

share|cite|improve this answer

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.