6
$\begingroup$

I asked this question on MSE, but obtained no answer. Maybe this is the right place to post it.

Let $D$ be a properly embedded free boundary disk in the closed unit ball $\mathbb{B}^3$ of $\mathbb{R}^3$. This means that $D$ is a smooth disk embedded in this ball, $D \cap \partial \mathbb{B}^3 = \partial D$ and this intersection is orthogonal. By orthogonality here I mean this: if $N$ is a unit normal along $D$ (its Gauss map), then $\langle N(x), x \rangle = 0$ for all $x \in \partial D$.

Assume that $D$ is strictly convex, that is to say, the principal curvatures are positive at each point of $D$ with respect to the fixed unit normal $N: D \to \mathbb{S}^2$. Does it follow that $N$ is a diffeomorphism onto its image? Equivalently, is $N$ injective?

The motivation is the following: if $S$ is a closed and connected surface in $\mathbb{R}^3$ which is also convex, then $N : S \to \mathbb{S}^2$ is a local diffeomorphism, hence a covering map. Since $\mathbb{S}^2$ is simply connected, this implies that $N$ is a global diffeomorphism. What happens when the surface is a disk?

$\endgroup$
1
  • $\begingroup$ How about doubling the sphere (topologically), have you though about it? If you consider $\mathbb B^3\cup \mathbb B^3$, you get a surface $S=D\cup D$. By the free boundary hypothesis, N is continuous (maybe it can be made smooth) and the local diffeomorphism hypothesis carries over for every point outside $\partial D$. $\endgroup$
    – Llohann
    Jul 14, 2020 at 14:14

1 Answer 1

4
$\begingroup$

The answer is yes. To show this one can use the fact that any topological immersion (locally one-to-one continuous map) of an n-dimensional disk into a sphere of the same dimension is an embedding (globally one-to-one) for $n\geq 2$, provided only that the map is one-to-one on the boundary of the disk. A proof may be found in

Gauss map, topology, and convexity of hypersurfaces with nonvanishing curvature, Topology, 41 (2002) 107-117.

So it remains to show that $N$ is one-to-one on the boundary $\partial D$ of the disk $D$. To see this one can extend $D$ to a complete $\mathcal{C}^1$ convex surface by attaching to $\partial D$ all the rays which are orthogonal to $S^2$ from outside. These rays belong to a convex cone $C$ with apex at the center $o$ of $S^2$. Since $D$ has positive curvature, it follows from basic differential geometry that $\partial D$ has positive geodesic curvature in $S^2$, and hence is strictly convex, which in turn yields that $C$ is strictly convex. So $N$ will be one-to-one along $\partial D$, since $N$ is just the Gauss map of $C\setminus\{o\}$.

Incidentally, it is not necessary to assume that $D$ is convex or even embedded, but it is enough that it have positive curvature and satisfy the free boundary condition; see the following paper with Changwei Xiong

Nonnegatively curved hypersurfaces with free boundary on a sphere, Calc. Var. Partial Differential Equations, 58 (2019), Art. 94, 20 pp.

$\endgroup$
2
  • $\begingroup$ How does one know that $C$ is of class $C^1$? $\endgroup$ Nov 4, 2020 at 3:26
  • $\begingroup$ Because $\partial D$ is $C^1$. $\endgroup$ Nov 4, 2020 at 4:00

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.