Let $M^3$ be a closed, connected and oriented smooth $3$-manifold, and fix an integer $g \geq 1$. Is it true that for a generic set of Riemannian metrics on $M$ the set of closed, connected and orientable embedded minimal surfaces of genus $g$ in $M$ is compact? If not, is it true for a generic set of metrics of positive scalar curvature?

There is no hope that this holds for every metric, since $\Sigma_{\gamma} \times \mathbb{S}^1$ has a sequence of minimal embedded tori winding around $\mathbb{S}^1$ as many times as one wishes. Here $\Sigma_\gamma$ is a compact Riemannian surface of genus $\gamma \geq 1$ endowed with a metric of constant curvature equal to $0$, if $g=1$, or equal to $-1$ if $g \geq 2$ (see this paper, for instance).


The set of embedded minimal surfaces $\Sigma$ with $\textrm{genus}(\Sigma) \leq g_0$ and $\textrm{area}(\Sigma)\leq A_0$ is compact for a generic metric (this follows from White's version of the Choi--Schoen compactness theorem https://mathscinet.ams.org/mathscinet-getitem?mr=880951, see also Appendix A: https://mathscinet.ams.org/mathscinet-getitem?mr=1778099).

However, without the area bounds, this fails: https://mathscinet.ams.org/mathscinet-getitem?mr=1728019 (generic metrics without compactness of stable spheres), https://mathscinet.ams.org/mathscinet-getitem?mr=2026836 (same for higher genus), https://mathscinet.ams.org/mathscinet-getitem?mr=2103999 (positive scalar curvature). In the first two papers, the surfaces constructed are also stable.

Edit: The second part of the question (generic PSC metric) is trickier. I think that no counterexample is known but I would guess that one exists. The tricky part is that for $(M,g)$ a $3$-manifold with a PSC metric, the space of bounded Morse index surfaces is compact. So any counterexample would necessarily have unbounded Morse index (as in the Colding--De Lellis example above). This rules out minimization, which is the tool used to construct the other examples.

One reason to think that such surfaces exist comes from White's degree theory. See (2) in the introduction here https://mathscinet.ams.org/mathscinet-getitem?mr=978083 (obviously not exactly your question since it deals with disks not closed surfaces).

  • $\begingroup$ It is not clear what happens in the positive scalar curvature case. In the last paper you cite, the authors only construct one metric with special properties. So, the second question I posed above was not answered. $\endgroup$ – Eduardo Longa Nov 12 '20 at 18:32

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.