Hopf's problem on the rigidity and classification of constant-mean-curvature spheres beyond the classical setting: higher-dimensional space forms and higher genus

An immersed sphere of constant mean curvature (CMC) in a three-dimensional space form must be totally umbilic — Hopf's theorem, proved by Heinz Hopf in 1950 using the holomorphicity of a quadratic differential. Hopf posed a broader conjecture, recorded verbatim in Wente's 1986 counterexample paper: must an immersed closed hypersurface of constant mean curvature $H\ne 0$ in $\mathbb{R}^n$ be the standard sphere? Several branches of the surrounding question are settled. In Euclidean space the conjecture fails for immersed hyperspheres of dimension at least three (Hsiang, Hsiang–Teng–Yu, Hsiang–Huynh) and holds for embedded hypersurfaces in every dimension (Alexandrov); it fails for higher-genus closed CMC surfaces in $\mathbb{R}^3$ (Wente's immersed torus; Kapouleas' surfaces of arbitrary genus), with the embedded CMC tori of $S^3$ completely classified by Andrews–Li; and for non-constant ambient curvature the strongest analogue is now closed: Meeks–Mira–Pérez–Ros classified all immersed CMC spheres in an arbitrary homogeneous three-manifold, proving that two CMC spheres of the same mean curvature differ only by an ambient isometry. The surviving open core is twofold: (1) immersed CMC hyperspheres ($H\ne 0$) in the round sphere $S^{n+1}$ and hyperbolic space $H^{n+1}$ for $n\ge 3$, where not even the existence of a non-umbilic example is known; and (2) classification, up to ambient isometry, of CMC immersions of closed surfaces of genus at least two into the three-dimensional space forms $\mathbb{R}^3$, $S^3$, $H^3$. Both branches lie beyond the reach of the holomorphic-differential method that proves the classical theorem.

Background

A surface immersed in a Riemannian 3-manifold has constant mean curvature (CMC) when the average of its two principal curvatures is the same at every point; such surfaces model soap bubbles and are critical points of area under a volume constraint. The surface is called totally umbilic when the two principal curvatures coincide at every point, so that locally it is round; in a space form — that is, Euclidean $E^3$, the round sphere $S^3$, or hyperbolic space $H^3$, the three simply-connected 3-manifolds of constant sectional curvature — the totally umbilic surfaces are exactly the round spheres and planes. Hopf's classical device associates to an oriented surface $f:\Sigma\to M^3$ its Hopf differential $\Phi=\langle f_{xx}-f_{yy}-2i f_{xy},N\rangle\, dz^2$, the trace-free $(2,0)$-part of the second fundamental form. When the ambient is a space form and the mean curvature $H$ is constant, the Codazzi equations — the compatibility conditions between the first and second fundamental forms of an immersion — force $\Phi$ to be holomorphic. On a sphere $S^2$ every holomorphic quadratic differential vanishes identically (by the Riemann–Roch theorem, since the canonical bundle has negative degree), so every principal curvature equals $H$: an immersed CMC sphere in $E^3$, $S^3$, or $H^3$ is totally umbilic. Hopf proved this in 1950 (Über Flächen mit einer Relation zwischen den Hauptkrümmungen, Math. Nachr. 4, 232–249), and Chern's short expository note (Geometric Dynamics, Lecture Notes in Mathematics 1007, Springer 1983, pp. 104–108) presents this proof. Hopf also posed a broader conjecture, recorded verbatim in Wente's 1986 paper: 'Let $\Sigma$ be an immersion of an oriented, closed hypersurface with constant mean curvature $H\ne 0$ in $\mathbb{R}^n$. Must $\Sigma$ be the standard embedded $(n-1)$-sphere?' Two classical results delimit the conjecture in $\mathbb{R}^n$ (both recorded in Wente's introduction): Alexandrov's moving-planes reflection theorem proves it for embedded hypersurfaces in every dimension, and Hopf's theorem proves the immersed genus-0 case in dimension three, where the holomorphic-differential argument is available. The conjecture, and the rigidity phenomenon behind it, therefore fragments into precise sub-questions once any of the three hypotheses of the classical setting is relaxed. Relaxing dimension — passing from surfaces to hypersurfaces of dimension $n\ge 3$ — removes the holomorphic-differential mechanism entirely, and in Euclidean space equivariant differential geometry, the study of submanifolds invariant under a group action whose generic orbit has codimension two, yields infinitely many non-round immersed CMC hyperspheres. Relaxing topology — closed surfaces of genus $\ge 1$ — removes the Riemann–Roch obstruction, and immersions are governed instead by the elliptic Gauss–Codazzi system and, on tori, by integrable-systems methods such as the Dorfmeister–Pedit–Wu generalized Weierstrass representation, a loop-group factorization technique that constructs CMC immersions from holomorphic data. Relaxing ambient curvature has, however, already been carried to a complete answer in the mildest non-constant-curvature setting: for simply-connected homogeneous 3-manifolds — Riemannian 3-manifolds whose isometry group acts transitively, such as Berger spheres (the three-sphere with the metric rescaled along the great-circle fibers of the Hopf fibration), the Heisenberg group Nil$_3$, $\widetilde{SL_2(\mathbb{R})}$, $S^2\times\mathbb{R}$, $H^2\times\mathbb{R}$, and Sol$_3$ — Meeks–Mira–Pérez–Ros proved that two CMC spheres of the same mean curvature differ only by an ambient isometry and determined for which values of $H$ such spheres exist, a complete classification of immersed CMC spheres in arbitrary homogeneous three-manifolds. What survives is the non-euclidean space-form direction in hypersurface dimensions at least three — where, unlike the Euclidean setting, the embedded picture is not rigid even for constant mean curvature: the Clifford hypersurfaces $S^p(r)\times S^q(s)\subset S^{p+q+1}$, products of round spheres with $r^2+s^2=1$ whose principal curvatures $-s/r$ and $r/s$ are constant, are embedded compact CMC hypersurfaces of non-spherical topology — and the higher-genus direction in the three-dimensional space forms.

In both directions, progress has come from methods other than the holomorphic differential: equivariant constructions, gluing techniques, and integrable-systems representations.

Problem Statement

Determine the exact validity range of Hopf-type rigidity — total umbilicity of immersed constant-mean-curvature spheres — and classify the constant-mean-curvature immersions that appear where rigidity fails. Two branches of the classical problem remain open, and they are independent. (1) Hyperspheres in the non-euclidean space forms: in the round sphere $S^{n+1}$ and hyperbolic space $H^{n+1}$ with $n\ge 3$ (ambient dimension at least four), decide whether a non-umbilic immersed hypersphere of constant mean curvature $H\ne 0$ exists — a negative decision establishes Hopf-type rigidity there, while a positive decision exhibits a non-umbilic example — and determine the immersed CMC hyperspheres up to ambient isometry, the classification that a positive existence decision leaves open. In euclidean $\mathbb{R}^{n+1}$ the answer is known to be negative for rigidity — infinitely many pairwise non-congruent non-round immersed CMC hyperspheres exist by the equivariant constructions of Hsiang, Hsiang–Teng–Yu, and Hsiang–Huynh — and in the sphere the case $H=0$ also admits non-umbilic examples (embedded non-equatorial minimal hyperspheres in $S^4$ are known), so the open question, like Hopf's conjecture itself, concerns $H\ne 0$ and hypersurfaces of topological type $S^n$; for $n=2$ Hopf's theorem settles all three space forms positively. (2) Higher genus in the three-dimensional space forms: for closed orientable surfaces of genus $\ge 2$, determine the moduli space of their CMC immersions into $\mathbb{R}^3$, $S^3$, $H^3$ up to ambient isometry. Existence in every topological type is known (Kapouleas); the genus-one case is understood — immersed CMC tori in the space forms are described by the integrable-systems spectral-curve theory of Pinkall–Sterling and the Dorfmeister–Pedit–Wu representation, and the embedded CMC tori of $S^3$ are completely classified (Andrews–Li) — but nothing comparable is known for genus $\ge 2$. Both branches lie beyond the reach of the holomorphic-differential method, which is available only for surfaces ($n=2$) in space forms. A third direction — Hopf-type uniqueness of CMC spheres in homogeneous 3-manifolds beyond the three-sphere family — is no longer open: Meeks–Mira–Pérez–Ros classify immersed CMC spheres in every homogeneous three-manifold.

References

1. Shiing-shen Chern, 'On surfaces of constant mean curvature in a three-dimensional space of constant curvature', in Geometric Dynamics (Rio de Janeiro, 1981), Lecture Notes in Mathematics 1007, Springer, Berlin, 1983, pp. 104–108. DOI 10.1007/bfb0061413, https://doi.org/10.1007/bfb0061413
2. Heinz Hopf, 'Über Flächen mit einer Relation zwischen den Hauptkrümmungen', Mathematische Nachrichten 4 (1950), 232–249. DOI 10.1002/mana.3210040122, https://doi.org/10.1002/mana.3210040122
3. Wu-yi Hsiang, Zhen-huan Teng, Wen-ci Yu, 'Examples of constant mean curvature immersions of the 3-sphere into euclidean 4-space', Proceedings of the National Academy of Sciences USA 79 (1982), no. 12, 3931–3932. DOI 10.1073/pnas.79.12.3931, https://doi.org/10.1073/pnas.79.12.3931
4. Wu-yi Hsiang, 'Generalized rotational hypersurfaces of constant mean curvature in the Euclidean spaces. I', Journal of Differential Geometry 17 (1982), no. 2, 337–356. DOI 10.4310/jdg/1214436924, https://doi.org/10.4310/jdg/1214436924
5. Wu-Yi Hsiang, Hsueh-Ling Huynh, 'Generalized rotational hypersurfaces of constant mean curvature in the Euclidean spaces. II', Pacific Journal of Mathematics 130 (1987), no. 1, 75–95. DOI 10.2140/pjm.1987.130.75, https://doi.org/10.2140/pjm.1987.130.75
6. Henry C. Wente, 'Counterexample to a conjecture of H. Hopf', Pacific Journal of Mathematics 121 (1986), no. 1, 193–243. DOI 10.2140/pjm.1986.121.193, https://doi.org/10.2140/pjm.1986.121.193
7. Nicolaos Kapouleas, 'Compact constant mean curvature surfaces in Euclidean three-space', Journal of Differential Geometry 33 (1991), no. 3, 683–715. DOI 10.4310/jdg/1214446560, https://doi.org/10.4310/jdg/1214446560
8. U. Pinkall, I. Sterling, 'On the classification of constant mean curvature tori', Annals of Mathematics (2) 130 (1989), no. 2, 407–451. DOI 10.2307/1971425, https://doi.org/10.2307/1971425
9. Ben Andrews, Haizhong Li, 'Embedded constant mean curvature tori in the three-sphere', Journal of Differential Geometry 99 (2015), no. 2, 169–192. DOI 10.4310/jdg/1421415560, https://doi.org/10.4310/jdg/1421415560
10. William H. Meeks III, Pablo Mira, Joaquín Pérez, Antonio Ros, 'Constant mean curvature spheres in homogeneous three-spheres', Journal of Differential Geometry 120 (2022), no. 2, 211–239. DOI 10.4310/jdg/1645207520, https://arxiv.org/abs/1308.2612
11. William H. Meeks III, Pablo Mira, Joaquín Pérez, Antonio Ros, 'Constant mean curvature spheres in homogeneous three-manifolds', Inventiones mathematicae 224 (2021), 147–244. DOI 10.1007/s00222-020-01008-y, https://arxiv.org/abs/1706.09394
12. Francisco Torralbo, 'Rotationally invariant constant mean curvature surfaces in homogeneous 3-manifolds', Differential Geometry and its Applications 28 (2010), no. 5, 593–607. DOI 10.1016/j.difgeo.2010.04.007, https://doi.org/10.1016/j.difgeo.2010.04.007
13. Francisco Torralbo, 'Compact minimal surfaces in the Berger spheres', Annals of Global Analysis and Geometry 41 (2012), no. 4, 391–405. DOI 10.1007/s10455-011-9288-7, https://doi.org/10.1007/s10455-011-9288-7
14. Benoît Daniel, 'Isometric immersions into 3-dimensional homogeneous manifolds', Commentarii Mathematici Helvetici 82 (2007), no. 1, 87–131. DOI 10.4171/cmh/86, https://doi.org/10.4171/cmh/86
15. Manfredo do Carmo, Isabel Fernández, 'A Hopf theorem for open surfaces in product spaces $\mathbb{S}^2\times\mathbb{R}$ and $\mathbb{H}^2\times\mathbb{R}$', Forum Mathematicum 21 (2009), no. 6, 1031–1051. DOI 10.1515/forum.2009.047, https://doi.org/10.1515/forum.2009.047
16. José A. Gálvez, Pablo Mira, 'A Hopf theorem for non-constant mean curvature and a conjecture of A. D. Alexandrov', Mathematische Annalen 366 (2016), no. 3–4, 909–928. DOI 10.1007/s00208-015-1351-4, https://doi.org/10.1007/s00208-015-1351-4
17. Alexandre C. Gonçalves, Karen K. Uhlenbeck, 'Moduli space theory for constant mean curvature surfaces immersed in space-forms', Communications in Analysis and Geometry 15 (2007), no. 2, 299–305. DOI 10.4310/cag.2007.v15.n2.a4, https://doi.org/10.4310/cag.2007.v15.n2.a4
18. Manfredo do Carmo, M. Dajczer, 'Rotation hypersurfaces in spaces of constant curvature', Transactions of the American Mathematical Society 277 (1983), no. 2, 685–709. DOI 10.1090/s0002-9947-1983-0694383-x, https://doi.org/10.1090/S0002-9947-1983-0694383-X
19. J. Dorfmeister, F. Pedit, H. Wu, 'Weierstrass type representation of harmonic maps into symmetric spaces', Communications in Analysis and Geometry 6 (1998), no. 4, 633–668. DOI 10.4310/cag.1998.v6.n4.a1, https://doi.org/10.4310/cag.1998.v6.n4.a1
20. Adrian Butscher, 'Gluing constructions amongst constant mean curvature hypersurfaces in $S^{n+1}$', Annals of Global Analysis and Geometry 36 (2009), no. 3, 221–274. DOI 10.1007/s10455-009-9156-x, https://doi.org/10.1007/s10455-009-9156-x
21. José A. Gálvez, Pablo Mira, Marcos P. Tassi, 'A quasiconformal Hopf soap bubble theorem', Calculus of Variations and Partial Differential Equations 61 (2022), no. 4. DOI 10.1007/s00526-022-02222-7, https://doi.org/10.1007/s00526-022-02222-7
22. Tongrui Wang, Zhichao Wang, Xin Zhou, 'Equivariant min-max theory and the spherical Bernstein problem in $\mathbb{S}^4$', preprint (2026). arXiv:2602.03984, https://arxiv.org/abs/2602.03984