4
$\begingroup$

Let $\Sigma$ be a 2-sphere with a (smooth) Riemannian metric $g$ of nonnegative curvature. Let $\mathcal{C}$ be a simple closed smooth curve on $\Sigma$. Then $\mathcal{C}$ splits $\Sigma$ into two components $\Sigma^+$, $\Sigma^-$. Set $S(t):=\{p\in\Sigma^+\mid d(\mathcal{C},p)=t\}$ where $d$ is the distance function induced by $g$. If I rememeber correctly, the length $L(t)$ of $S(t)$ is well-defined for almost all $t\geq 0$.

Is it true that $$|L(t_1)-L(t_2)|\leq 2\pi|t_1-t_2|$$ whenever defined? (At least when $S(t_1),S(t_2)$ are non-empty.)

$\endgroup$

1 Answer 1

5
$\begingroup$

Cut out a small arc $AB$ of the equatorial circle of the round $S^2$ and replace it by minimizing geodesics connecting $A$ and $B$ to the north pole. Denote the resulting curve $\mathcal C$. Then the curves $S_t$ south of $\mathcal C$ vary in length wildly for small $t$, violating your bound.

$\endgroup$
3
  • $\begingroup$ Does it still have nonnegative curvature ? $\endgroup$ Jul 3, 2017 at 8:51
  • $\begingroup$ @Thomas: Only the surface was assumed to have nonnegative curvature. $\endgroup$
    – Pete
    Jul 3, 2017 at 10:50
  • $\begingroup$ @Mikhail: Thank you, of course! Wondering if it might hold true with more assumptions on $\mathcal{C}$. I'm interested in the case where $\mathcal{C}$ itself is the distance sphere of a point on the other hemisphere. $\endgroup$
    – Pete
    Jul 3, 2017 at 10:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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