I am trying to see when this type of metric is complete: Let $A$ be the set of $C^{1}$ paths in $U \in \mathbb{R}^{n}$. For any $x,y$ define $$\rho(x,y) = \inf_{\gamma \in A; \gamma(0) = x, \gamma(1) = y}\int_{0}^{1}V(\gamma(t))\|\gamma'(t)\|dt$$
We can assume that $U$ is convex, and that $V$ is as smooth as necessary.
What I've done so far:
- I was able to show that $\rho$ is a metric.
- Also, it seems that if one assumes $ c < V(x) < C $ for some $c,C > 0$ it can be shown to be strongly equivalent to a Euclidean norm.
However, I want to have some weaker constraints like $V > 0$. I also would appreciate any references for this.
Since you have put very little constraints on the subset $U$, there are so many different possibilities. I do not see how one can do anything other than list a few examples. I'll list two contrasting examples.