The Global Spherical Shell conjecture: completing Kodaira's classification of minimal class VII surfaces

Kodaira's 1964 classification of compact complex surfaces free from exceptional curves listed seven classes and left a question mark against exactly one: class VII, the minimal surfaces with first Betti number $b_1=1$ and geometric genus $p_g=0$. In the modern formulation, after the $\kappa=0$ members of Kodaira's class VII (the secondary Kodaira surfaces) were classified separately, class VII$_0$ denotes the minimal surfaces with $b_1=1$ and Kodaira dimension $\kappa=-\infty$. The cases $b_2=0$ (Hopf and Inoue surfaces; Bogomolov, with complete proofs by Li-Yau-Zheng and by Teleman) and $b_2=1$ (Teleman) are settled, and every known example with $b_2>0$ is a Kato surface, obtained from a ball by a tower of blow-ups and a regluing that leaves a global spherical shell (GSS) in the surface. The surviving open core is the Global Spherical Shell conjecture: every minimal class VII surface with $b_2>0$ is a Kato surface. For $b_2=2$ and $b_2=3$ the existence of a cycle of rational curves is known (Teleman), but the passage from a cycle to a global spherical shell has been proved only under auxiliary hypotheses such as the presence of foliations, and for $b_2\ge 4$ even the existence of a single holomorphic curve on a minimal class VII surface is open. The problem is to prove the conjecture, thereby filling the last entry of Kodaira's classification, or to refute it by exhibiting a minimal class VII surface with $b_2>0$ that is not a Kato surface.

Background

A compact complex surface is a two-dimensional compact complex manifold. The Enriques-Kodaira classification organizes such surfaces by classical invariants: the first Betti number $b_1$ (the rank of the first homology group), the geometric genus $p_g=\dim H^0(X,K_X)$ (the dimension of the space of holomorphic 2-forms, with $K_X$ the canonical line bundle), and the Kodaira dimension $\kappa$, which measures the growth of the spaces of pluricanonical forms $H^0(X,mK_X)$, $m\ge 1$, as $m$ grows ($\kappa=-\infty$ means all of these spaces vanish). A compact Kähler surface has even $b_1$, so every surface with $b_1=1$ is non-Kähler. A smooth rational curve of self-intersection $-1$ on a surface is an exceptional curve (of the first kind) and can be blown down, so classification is carried out first for minimal surfaces, those free from exceptional curves, and every surface is recovered from a unique minimal one by finitely many blow-ups.

In his 1964 PNAS note "On the structure of compact complex analytic surfaces, II", Kodaira proved (Theorem 11) that surfaces free from exceptional curves fall into seven classes determined by the parity of $b_1$, the vanishing of $p_g$, and the Chern class of the canonical bundle. Six classes received a structure description (algebraic surfaces; K3 surfaces, i.e. simply connected compact complex surfaces with trivial canonical bundle; complex tori; and elliptic surfaces of two types, i.e. surfaces admitting a holomorphic fibration over a curve whose fibers are elliptic curves), but for class VII - "the class of surfaces free from exceptional curves with $b_1=1$ and $p_g=0$" - the classification table carries a question mark. In the modern reading of the classification (Barth-Hulek-Peters-Van de Ven 2004), the minimal surfaces with $b_1=1$, $p_g=0$ and $\kappa=0$ were handled separately (they are the secondary Kodaira surfaces), and class VII now means: minimal surfaces with $b_1=1$ and $\kappa=-\infty$ (class VII$_0$); if moreover $b_2>0$ one writes $\mathrm{VII}_0^+$, where $b_2$ is the second Betti number.

Basic structure theory for these surfaces: a class VII$_0$ surface with $b_2=n$ has Euler number $e(X)=n$ and $K_X^2=-n$; its intersection form on $H^2(X,\mathbb{Z})$ modulo torsion is negative definite, in fact equivalent to the standard form $(-1)^{\oplus n}$; it carries at most $n$ irreducible curves; and every irreducible curve on it is rational or elliptic, the elliptic case characterizing the Enoki surfaces (Nakamura; Enoki). The known examples with $b_2=0$ are the Hopf surfaces (quotients of $\mathbb{C}^2\setminus\{0\}$ by a contraction, diffeomorphic to $S^1\times S^3$) and Inoue's three families, which contain no curves at all.

For $b_2>0$, every known example arises from Kato's construction, announced in his 1977 note and developed in the 1977 Kyoto symposium proceedings: take the unit ball in $\mathbb{C}^2$, perform a tower of $n=b_2$ blow-ups with each center on the preceding exceptional curve, remove a smaller ball, and reglue the two boundary 3-spheres by a biholomorphism. The image $\Sigma$ of the reglued shell is a global spherical shell (GSS): an open subset of the surface, biholomorphic to a neighborhood of the standard sphere $S^3\subset\mathbb{C}^2\setminus\{0\}$, whose complement is connected. A class VII$_0^+$ surface containing a GSS is called a Kato surface; it contains exactly $n$ rational curves, arranged around a cycle of rational curves (a reduced connected divisor $\Gamma$ of arithmetic genus 1: one nodal rational curve, two rational curves meeting in two points, or a ring of at least three rational curves). The known Kato surfaces comprise the Inoue-Hirzebruch, the Enoki (parabolic Inoue), and the intermediate Kato families, and Kato surfaces admit explicit moduli descriptions via blow-up towers and gluing maps.

The classification status of class VII is as follows. The case $b_2=0$ is a theorem: every minimal class VII surface with $b_2=0$ is a Hopf or an Inoue surface (Bogomolov 1976, with complete proofs by Li-Yau-Zheng 1990 via Hermitian metrics and Teleman 1994 via moduli of stable bundles, and a later group-theoretic proof by Buonerba-Bogomolov-Kurnosov 2017). The case $b_2=1$ was settled by Teleman (2005) using Donaldson theory (gauge theory: the study of moduli spaces of anti-self-dual connections and stable holomorphic bundles) on non-Kähler surfaces. Since every known class VII$_0^+$ surface is a Kato surface, the remaining question is the Global Spherical Shell (GSS) conjecture, attributed to Kato (1977): every minimal class VII surface with $b_2>0$ is a Kato surface. Following Teleman, the conjecture is attacked in two steps: (C1) every $\mathrm{VII}_0^+$ surface contains a cycle of rational curves; (C2) every $\mathrm{VII}_0^+$ surface containing a cycle of rational curves contains a GSS. Step (C2) is known under auxiliary hypotheses: existence of a global holomorphic vector field (Dloussky-Oeljeklaus-Toma 2000, 2001), existence of $b_2(X)$ rational curves (Dloussky-Oeljeklaus-Toma 2003), or existence of a non-zero twisted logarithmic 1-form (Dloussky 2024). Step (C1) is known for $b_2\le 3$ (Teleman 2010 for $b_2=2$; Teleman's 2017-2018 survey accounts for $b_2=3$, as credited in the recent literature). What remains open: the GSS conclusion for $b_2=2$ and $b_2=3$, where it is proved only under foliation hypotheses (Brunella 2011; Kurnosov-Spicer 2026), and for $b_2\ge 4$, where even the existence of a single curve is open. Since every Kato surface carries a singular holomorphic foliation (Dloussky-Oeljeklaus 1999), a minimal class VII surface with $b_2>0$ carrying no foliation would be a counterexample to the conjecture.

Problem Statement

Let $X$ be a minimal compact complex surface with first Betti number $b_1(X)=1$, Kodaira dimension $\kappa(X)=-\infty$, and second Betti number $b_2(X)>0$ (a surface of class $\mathrm{VII}_0^+$; equivalently, in Kodaira's original 1964 terminology, a surface free from exceptional curves with $b_1=1$ and $p_g=0$ that is not a secondary Kodaira surface, i.e. not one of the $\kappa=0$ elliptic members of that class). Prove that $X$ is a Kato surface, i.e. that $X$ contains a global spherical shell: an open subset $\Sigma$ together with a biholomorphism from a neighborhood of the unit sphere $S^3\subset\mathbb{C}^2\setminus\{0\}$ onto $\Sigma$ such that $X\setminus\Sigma$ is connected. Such a result completes Kodaira's classification of class VII surfaces by identifying every minimal surface with $b_1=1$ and $\kappa=-\infty$ with a member of the known explicit lists (Hopf and Inoue surfaces for $b_2=0$; Inoue-Hirzebruch, Enoki, and intermediate Kato surfaces for $b_2>0$, together with their blow-ups). The conjecture is equivalent to the conjunction of two statements: (C1) every minimal class VII surface with $b_2>0$ contains a cycle of rational curves (a reduced connected effective divisor of arithmetic genus 1: a nodal rational curve, two rational curves meeting in two points, or a ring of at least three rational curves); and (C2) every minimal class VII surface with $b_2>0$ that contains a cycle of rational curves contains a global spherical shell. Alternatively, refute the conjecture by constructing a minimal class VII surface with $b_2>0$ that is not a Kato surface. The instances $b_2=0$ and $b_2=1$ are already theorems, so the open content lies at $b_2\ge 2$: for $b_2\in\{2,3\}$ cycles are known to exist but the global spherical shell conclusion is open, and for $b_2\ge 4$ even the existence of a single holomorphic curve is open.

References

1. K. Kodaira, "On the structure of compact complex analytic surfaces, II", Proceedings of the National Academy of Sciences of the United States of America 51 (1964), no. 6, 1100-1104. DOI: 10.1073/pnas.51.6.1100, https://doi.org/10.1073/pnas.51.6.1100 (full text available via PubMed Central PMC300219)
2. M. Inoue, "On surfaces of class VII_0", Inventiones Mathematicae 24 (1974), no. 4, 269-310. DOI: 10.1007/BF01425563, https://doi.org/10.1007/BF01425563
3. F. A. Bogomolov, "Classification of surfaces of class VII_0 with b_2 = 0", Mathematics of the USSR-Izvestiya 10 (1976), no. 2, 255-269. DOI: 10.1070/IM1976v010n02ABEH001688, https://doi.org/10.1070/IM1976v010n02ABEH001688
4. J. Li, S.-T. Yau, F. Zheng, "A simple proof of Bogomolov's theorem on class VII_0 surfaces with b_2 = 0", Illinois Journal of Mathematics 34 (1990), no. 2, 217-220. DOI: 10.1215/ijm/1255988265, https://doi.org/10.1215/ijm/1255988265
5. A. Teleman, "Projectively flat surfaces and Bogomolov's theorem on class VII_0 surfaces", International Journal of Mathematics 5 (1994), no. 2, 253-264. DOI: 10.1142/S0129167X94000152, https://doi.org/10.1142/S0129167X94000152
6. M. Kato, "Compact complex manifolds containing 'global' spherical shells", Proceedings of the Japan Academy, Series A, Mathematical Sciences 53 (1977), no. 1, 15-16. DOI: 10.3792/pja/1195517920, https://doi.org/10.3792/pja/1195517920
7. M. Kato, "Compact complex manifolds containing 'global' spherical shells, I", in: Proceedings of the International Symposium on Algebraic Geometry (Kyoto, 1977), Kinokuniya Book Store, Tokyo (1978), 45-84. zbMATH: Zbl 0421.32010, https://zbmath.org/0421.32010
8. F. Buonerba, F. Bogomolov, N. Kurnosov, "Classifying VII_0 surfaces with b_2 = 0 via group theory", preprint (2017, revised 2019). arXiv:1709.00062, https://arxiv.org/abs/1709.00062
9. I. Nakamura, "On surfaces of class VII_0 with curves", Inventiones Mathematicae 78 (1984), no. 3, 393-443. DOI: 10.1007/BF01388444, https://doi.org/10.1007/BF01388444
10. I. Nakamura, "On surfaces of class VII_0 with curves, II", Tohoku Mathematical Journal (2) 42 (1990), no. 4, 475-516. DOI: 10.2748/tmj/1178227571, https://doi.org/10.2748/tmj/1178227571
11. I. Enoki, "Surfaces of class VII_0 with curves", Tohoku Mathematical Journal (2) 33 (1981), no. 4, 453-492. DOI: 10.2748/tmj/1178229349, https://doi.org/10.2748/tmj/1178229349
12. G. Dloussky, K. Oeljeklaus, "Vector fields and foliations on compact surfaces of class VII_0", Annales de l'Institut Fourier 49 (1999), no. 5, 1503-1545. DOI: 10.5802/aif.1728, https://doi.org/10.5802/aif.1728
13. G. Dloussky, K. Oeljeklaus, M. Toma, "Surfaces de la classe VII_0 admettant un champ de vecteurs", Commentarii Mathematici Helvetici 75 (2000), no. 2, 255-270. DOI: 10.1007/PL00000374, https://doi.org/10.1007/PL00000374
14. G. Dloussky, K. Oeljeklaus, M. Toma, "Surfaces de la classe VII_0 admettant un champ de vecteurs, II", Commentarii Mathematici Helvetici 76 (2001), no. 4, 640-664. DOI: 10.1007/s00014-001-8323-x, https://doi.org/10.1007/s00014-001-8323-x
15. G. Dloussky, K. Oeljeklaus, M. Toma, "Class VII_0 surfaces with b_2 curves", Tohoku Mathematical Journal (2) 55 (2003), no. 2, 283-309. DOI: 10.2748/tmj/1113246942, https://doi.org/10.2748/tmj/1113246942
16. A. Teleman, "Donaldson theory on non-Kählerian surfaces and class VII surfaces with b_2 = 1", Inventiones Mathematicae 162 (2005), no. 3, 493-521. DOI: 10.1007/s00222-005-0451-2, https://doi.org/10.1007/s00222-005-0451-2
17. A. Teleman, "Instantons and curves on class VII surfaces", Annals of Mathematics (2) 172 (2010), no. 3, 1749-1804. DOI: 10.4007/annals.2010.172.1749, https://doi.org/10.4007/annals.2010.172.1749
18. M. Brunella, "On a class of foliated non-Kählerian compact complex surfaces", Tohoku Mathematical Journal (2) 63 (2011), no. 3, 441-460. DOI: 10.2748/tmj/1318338951, https://doi.org/10.2748/tmj/1318338951
19. G. Dloussky, A. Teleman, "Infinite bubbling in non-Kählerian geometry", Mathematische Annalen 353 (2012), no. 4, 1283-1314. DOI: 10.1007/s00208-011-0713-9, https://doi.org/10.1007/s00208-011-0713-9
20. A. Teleman, "Instanton moduli spaces on non-Kählerian surfaces. Holomorphic models around the reduction loci", Journal of Geometry and Physics 91 (2015), 66-87. DOI: 10.1016/j.geomphys.2015.01.007, https://doi.org/10.1016/j.geomphys.2015.01.007
21. A. Teleman, "Towards the classification of class VII surfaces", in: Complex and Symplectic Geometry, Springer INdAM Series 21, Springer, Cham (2017), 249-262. DOI: 10.1007/978-3-319-62914-8_18, https://doi.org/10.1007/978-3-319-62914-8_18
22. A. Teleman, "Donaldson theory in non-Kählerian geometry", in: Modern Geometry: A Celebration of the Work of Simon Donaldson, Proceedings of Symposia in Pure Mathematics 99, American Mathematical Society (2018), 363-392. DOI: 10.1090/pspum/099/01746, https://doi.org/10.1090/pspum/099/01746
23. G. Dloussky, "Non Kählerian surfaces with a cycle of rational curves", Complex Manifolds 8 (2021), no. 1, 208-222. DOI: 10.1515/coma-2020-0114, https://doi.org/10.1515/coma-2020-0114
24. G. Dloussky, "On Classification of compact complex surfaces of class VII", preprint (2024, revised December 2025). arXiv:2403.20178, https://arxiv.org/abs/2403.20178
25. N. Kurnosov, C. Spicer, "Class VII surfaces with b_2 = 3 and two foliations are Kato", preprint (August 2026). arXiv:2608.29047, https://arxiv.org/abs/2608.29047
26. W. P. Barth, K. Hulek, C. A. M. Peters, A. Van de Ven, "Compact Complex Surfaces", 2nd edition, Ergebnisse der Mathematik und ihrer Grenzgebiete (3) 4, Springer, Berlin (2004). ISBN 9783540008323, DOI: 10.1007/978-3-642-57739-0, https://doi.org/10.1007/978-3-642-57739-0