The Hopf problem in generalized geometry: existence of an integrable generalized complex structure on the six-sphere $\mathbb{S}^6$

The six-dimensional sphere $\mathbb{S}^6$ is one of only two spheres admitting almost complex structures (Borel–Serre), carrying a canonical nearly Kähler one induced by octonion multiplication; yet whether it admits an integrable complex structure — the classical Hopf problem — remains open, and $H^2(\mathbb{S}^6,\mathbb{R})=0$ rules out symplectic structures. Etayo, Gómez-Nicolás and Santamaría posed the Hopf problem in generalized geometry: does $\mathbb{S}^6$ admit any integrable generalized complex structure on its generalized tangent bundle $\mathbb{T}\mathbb{S}^6=T\mathbb{S}^6\oplus T^*\mathbb{S}^6$? They proved that no such structure can be built as a pointwise spherical combination of the structures induced by the octonionic nearly Kähler data, and an unrefereed 2026 preprint by Ricciarini excludes constant types 1 and 2 via an Euler-class obstruction (constant type 0 being the classical $H^2=0$ obstruction to symplectic structures), together with gluing constructions from locally flat models. What survives untouched is the full existence question: the constant type-3 (complex-type) case is equivalent to the classical Hopf problem, and type-changing structures — the genuinely new possibility opened by generalized geometry — are subject to no known obstruction or construction on $\mathbb{S}^6$. The problem asks for either an explicit integrable structure of any admissible type configuration, or a proof that none exists.

Background

An almost complex structure on an even-dimensional manifold $M$ is a bundle endomorphism $J\colon TM\to TM$ with $J^2=-\mathrm{Id}$; it is integrable when it arises from holomorphic coordinates, which by the Newlander–Nirenberg theorem is equivalent to the vanishing of a certain tensor built from $J$, the Nijenhuis tensor. A question going back to Hopf in the 1940s (see the historical account of Agricola et al.) asks whether $\mathbb{S}^6$ admits an integrable almost complex structure — the classical Hopf problem. Borel and Serre proved in 1953 that, among all spheres, only $\mathbb{S}^2$ and $\mathbb{S}^6$ admit almost complex structures at all. The sphere $\mathbb{S}^6$ carries a canonical one, induced by multiplication of imaginary octonions (the octonions are the eight-dimensional normed division algebra, $\mathbb{S}^6$ sits inside its seven-dimensional imaginary part, and the exceptional Lie group $G_2$ is the automorphism group of the octonions); this structure makes $\mathbb{S}^6$ nearly Kähler — an almost Hermitian structure $(J,g)$ whose fundamental 2-form $\omega(\cdot,\cdot)=g(J\cdot,\cdot)$ has totally skew covariant derivative — but its Nijenhuis tensor is famously nonzero, and Blanchard, in a proof popularized by LeBrun, showed that no complex structure on $\mathbb{S}^6$ can be orthogonal with respect to the round metric. Since $H^2(\mathbb{S}^6,\mathbb{R})=0$, the sphere also admits no symplectic structure (a closed nondegenerate 2-form), so neither of the two classical integrable geometries is available on $\mathbb{S}^6$.

Generalized geometry, introduced by Hitchin and developed by Gualtieri, replaces $TM$ by the generalized tangent bundle $\mathbb{T}M=TM\oplus T^*M$, endowed with the canonical indefinite pairing $\mathcal{G}_0(X+\xi,Y+\eta)=\frac{1}{2}(\xi(Y)+\eta(X))$, which is symmetric of signature $(n,n)$ when $\dim M=2n$, and with the Dorfman bracket $\llbracket X+\xi,Y+\eta\rrbracket=[X,Y]+\mathcal{L}_X\eta-i_Yd\xi$, a natural bilinear bracket on sections of $\mathbb{T}M$ extending the Lie bracket of vector fields ($\mathcal{L}_X$ denotes the Lie derivative and $i_Y$ the contraction with $Y$). A generalized almost complex structure is an endomorphism $\mathcal{J}\colon\mathbb{T}M\to\mathbb{T}M$ with $\mathcal{J}^2=-\mathrm{Id}$; following the source authors, it is called strong when it is skew-adjoint for the pairing, $\mathcal{G}_0(\mathcal{J}u,v)=-\mathcal{G}_0(u,\mathcal{J}v)$ for all sections $u,v$ (equivalently, since $\mathcal{J}^2=-\mathrm{Id}$, $\mathcal{G}_0(\mathcal{J}u,\mathcal{J}v)=\mathcal{G}_0(u,v)$), and weak otherwise. It is integrable when its $+i$-eigenbundle $\mathbb{L}^{1,0}_{\mathcal{J}}=\{u-i\mathcal{J}u\colon u\in\Gamma(\mathbb{T}M)\}$ is closed under the Dorfman bracket, equivalently when the associated generalized Nijenhuis map vanishes identically; for isotropic subbundles, closure under the Dorfman bracket is equivalent to closure under its skew-symmetrization (the Courant bracket), so the two integrability notions agree for strong structures. Integrable generalized complex structures simultaneously generalize complex and symplectic structures: at each point the structure has a type $k$, with $0\le k\le n=\dim M/2$, given by the dimension of the intersection of the $+i$-eigenbundle with the complexified cotangent subspace, and on the regular locus the structure is locally the product of a complex $k$-dimensional and a symplectic $2(n-k)$-dimensional model. The local normal-form theory of Bailey and Gualtieri shows the type may jump along loci, but only by even amounts, giving type-changing integrable structures that are genuinely neither complex nor symplectic.

From an almost Hermitian triple $(J,g,\omega)$, Gualtieri's constructions induce generalized almost complex structures $\mathcal{J}_{\lambda,J}(X+\xi)=JX+\lambda J^*\xi$ for $\lambda\in\{1,-1\}$ (where $J^*$ is the dual action on covectors), the metric structure $\mathcal{J}_g$, and the 2-form structure $\mathcal{J}_\omega$ (the latter two built from $g$ and $\omega$ through the musical isomorphisms); Gualtieri proved that $\mathcal{J}_{-1,J}$ is integrable exactly when $J$ is integrable, and $\mathcal{J}_\omega$ exactly when $\omega$ is symplectic. Etayo, Gómez-Nicolás and Santamaría posed the Hopf problem in generalized geometry and proved that $\mathbb{S}^6$ admits no weak generalized complex structure expressible as a pointwise spherical combination $a\mathcal{J}_{1,J}+b\mathcal{J}_g+c\mathcal{J}_\omega$ with smooth functions $a,b,c$ satisfying $a^2+b^2+c^2=1$. Ricciarini subsequently excluded, in an unrefereed preprint, constant types 1 and 2 by an Euler-class argument, and structures glued from locally flat models over the stereographic cover, while observing that the constant type-3 case is exactly the classical Hopf problem. The existence question in its full generality — the subject of this problem — remains open.

Problem Statement

Does the six-dimensional sphere $\mathbb{S}^6$, with its standard smooth structure, admit a generalized complex structure — that is, a strong (pairing-compatible) generalized almost complex endomorphism $\mathcal{J}$ of $\mathbb{T}\mathbb{S}^6=T\mathbb{S}^6\oplus T^*\mathbb{S}^6$, satisfying $\mathcal{J}^2=-\mathrm{Id}$ and $\mathcal{G}_0(\mathcal{J}u,v)=-\mathcal{G}_0(u,\mathcal{J}v)$ for all sections $u,v$ (equivalently $\mathcal{G}_0(\mathcal{J}u,\mathcal{J}v)=\mathcal{G}_0(u,v)$), whose $+i$-eigenbundle $\mathbb{L}^{1,0}_{\mathcal{J}}=\{u-i\mathcal{J}u\}$ is closed under the Dorfman bracket (equivalently, whose generalized Nijenhuis map vanishes identically)?

(1) Existence: exhibit such a structure explicitly, of any type configuration, and prove its integrability on all of $\mathbb{S}^6$. Two subcases are qualitatively distinct: a constant type-3 (complex-type) structure, which amounts to a positive solution of the classical Hopf problem since such a structure is a $B$-field transform of an integrable complex structure; or a type-changing structure, which would be the first integrable structure on $\mathbb{S}^6$ that is neither complex nor symplectic.

(2) Non-existence: prove that no integrable strong generalized complex structure of any type configuration — constant type 3 or type-changing — exists on $\mathbb{S}^6$. Such a proof must in particular resolve the classical Hopf problem negatively, because a constant type-3 integrable structure on $\mathbb{S}^6$ is equivalent to an integrable complex structure on $\mathbb{S}^6$.

For scope: constant type 0 is excluded classically, since $H^2(\mathbb{S}^6,\mathbb{R})=0$ precludes symplectic structures, and an unrefereed 2026 preprint excludes constant types 1 and 2 by an Euler-class argument; neither exclusion touches the two branches above. Since a strong structure is in particular a weak generalized complex structure in the source authors' convention (dropping the $\mathcal{G}_0$-compatibility requirement), any exclusion covering the weak class would also cover the strong class.

References

1. Fernando Etayo, Pablo Gómez-Nicolás, Rafael Santamaría, "About generalized complex structures on $\mathbb{S}^6$", Mediterranean Journal of Mathematics 22 (2025), no. 5, article 138. DOI: 10.1007/s00009-025-02916-0; arXiv:2405.05681. https://arxiv.org/abs/2405.05681
2. Andrea Ricciarini, "On the integrability of generalized almost complex structures on $\mathbb{S}^6$", preprint (2026, unrefereed). arXiv:2606.03379. https://arxiv.org/abs/2606.03379
3. Marco Gualtieri, "Generalized complex geometry", Annals of Mathematics (2) 174 (2011), 75–123. DOI: 10.4007/annals.2011.174.1.3. https://doi.org/10.4007/annals.2011.174.1.3
4. Nigel Hitchin, "Generalized Calabi–Yau manifolds", The Quarterly Journal of Mathematics 54 (2003), 281–308. DOI: 10.1093/qmath/hag025. https://doi.org/10.1093/qmath/hag025
5. Armand Borel, Jean-Pierre Serre, "Groupes de Lie et puissances réduites de Steenrod", American Journal of Mathematics 75 (1953), 409–448. DOI: 10.2307/2372495. https://doi.org/10.2307/2372495
6. Claude LeBrun, "Orthogonal complex structures on $S^6$", Proceedings of the American Mathematical Society 101 (1987), 136–138. DOI: 10.1090/s0002-9939-1987-0897084-7. https://doi.org/10.1090/s0002-9939-1987-0897084-7
7. Ilka Agricola, Aleksandra Borówka, Thomas Friedrich, "$S^6$ and the geometry of nearly Kähler 6-manifolds", Differential Geometry and its Applications 57 (2018), 75–86. DOI: 10.1016/j.difgeo.2017.10.007. https://doi.org/10.1016/j.difgeo.2017.10.007
8. Ilka Agricola, Giovanni Bazzoni, Oliver Goertsches, Panagiotis Konstantis, Sönke Rollenske, "On the history of the Hopf problem", Differential Geometry and its Applications 57 (2018), 1–9. DOI: 10.1016/j.difgeo.2017.10.014. https://doi.org/10.1016/j.difgeo.2017.10.014
9. Gábor Etesi, "Complex structure on the six dimensional sphere from a spontaneous symmetry breaking", Journal of Mathematical Physics 56 (2015), no. 4, 043508. DOI: 10.1063/1.4918540. https://doi.org/10.1063/1.4918540
10. Daniel Guan, Na Li, Zhonghua Wang, "Some Remarks on Existence of a Complex Structure on the Compact Six Sphere", Axioms 13 (2024), no. 10, 719. DOI: 10.3390/axioms13100719. https://doi.org/10.3390/axioms13100719
11. Michael Bailey, Marco Gualtieri, "Local analytic geometry of generalized complex structures", Bulletin of the London Mathematical Society 49 (2017), no. 2, 307–319. DOI: 10.1112/blms.12029. https://doi.org/10.1112/blms.12029