This question is inspired by question in reference.

Question : If $M$ is a simply connected closed Riemannian manifold of nonnegative sectional curvature, then there is a totally geodesic submanifold $S$ of codimension 1

Def : ${\rm conv}\ X$ is a smallest closed convex set containing $X$. And a subset $X$ is convex if for $x,\ y\in X$, any shortest geodesic between them is in $X$.

Construction of $S$ : Let $X_1=B(p,\delta)$ to be a closed ball, which is convex. If $p_1$ is not in $X_1$, then let $X_2={\rm conv}\ X_1\bigcup\{p_1\}$ s.t. $d_H(X_1,X_2) >0$ is small and ${\rm vol}\ X_2-{\rm vol}\ X_1 >0$ is small. Hence if we repeat this process, then $X_n$ goes to $ X_\infty$.

If $p_\infty$ is not in $X_\infty$, and $d_H(X_\infty,{\rm conv}\ X_\infty\bigcup\{p_\infty\})$ is large, then $\partial X_\infty$ is a desired one.

If not, we let $X_1=X_\infty$ and repeat these process. That is, ${\rm vol}\ X_\infty$ is supremum of volumes of convex sets containing $p$.

Example : If we starts this process at a point of negative sectional curvature in torus of revolution, then $X_\infty$ is closure of set of all points of negative sectional curvature.

Additional question : We have alvarezpaiva's answer. But I want to know in dimension $3$. Or I want to know why my argument fails in $\mathbb{C}P^2$.

Reference : Convex sets in Alexandrov spaces

  • 1
    $\begingroup$ As stated it is unclear (to me) what you are asking and what you are proving in the "proof". Also (to me) the proof does not make sense: you never explain why $X_2$ that satisfies your assumptions exists. What seems to be happening is that you propose some construction for a convex subset, and then ask whether the construction produces a totally geodesic submanifold of codimension $1$. $\endgroup$ Sep 1, 2018 at 17:25
  • $\begingroup$ "you never explain why $ X_2$ that satisfies your assumptions exists" In $\mathbb{C}P^2$, by considering following answer, this may not hold. This is my question. $\endgroup$ Sep 1, 2018 at 17:50
  • $\begingroup$ What I am saying is that the question does not logically make sense. What is ``small'' in the definition of $X_2$? What do you mean by the boundary of $X_\infty$? Why do you think the construction of $S$ yields a submanifold? If it is a submanifold, why do you expect it have codimension one? Why is it totally geodesic? $\endgroup$ Sep 1, 2018 at 18:54
  • 2
    $\begingroup$ The main result in arxiv.org/pdf/1703.09240.pdf gives an answer to your question (the way I understand it). Also in the introduction they mention that Tsukada found left-invariant metrics on 3–dimensional Lie groups without totally geodesic surfaces. $\endgroup$ Sep 1, 2018 at 19:32

2 Answers 2


There is no three-dimensional totally geodesic submanifold in $\mathbb{CP}^2$.


About you second question.

If $S$ is a totally geodesic submanifold, then for any geodesic in $S$ the tangent planes $\mathrm{T}S$ are invariant with respect to parallel translations and Jacobi fields with ends in $\mathrm{T}S$ have to be in $\mathrm{T}S$. If $\dim S>1$, then it gives a nontrivial equation on the curvature tensor along the geodesic.

So generic 3-dimensional manifold does not admit a geodesic sumbmanifold of dimension at least 2. In particular, any metric on a 3-manifold admits arbitrary small $C^\infty$ deformation such that the obtained Riemannian manifold has no totally geodesic surfaces, even locally.


Your Answer

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

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