Abstract
This text follows the lecture series given by the author in the CIME School “Complex non-Kähler geometry” (Cetraro, July 9–13, 2018) and is dedicated to the classification of non-Kählerian surfaces. In the first three sections we present the classical theory:
-
The Enriques Kodaira classification for surfaces and the classes of non-Kählerian surfaces,
-
Class VII surfaces and their general properties,
-
Kato surfaces: construction, classification and moduli.
In Sect. 3.4 we explain the main ideas and techniques used in the proofs of our results on the existence of cycles of curves on class VII surfaces with small b 2. Section 3.5 deals with criteria for the existence of smooth algebraic deformations of the singular surface obtained by contracting a cycle of rational curves in a minimal class VII surface. We included an Appendix in which we introduce several fundamental objects in non-Kählerian complex geometry (the Picard group of a compact complex manifold, the Gauduchon degree, the Kobayashi-Hitchin correspondence for line bundles, unitary flat line bundles), and we prove basic properties of these objects.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
D. Barlet, Convexité de l’espace des cycles. Bull. Soc. Math. France 106, 373–397 (1978)
W. Barth, K. Hulek, Ch. Peters, A. Van de Ven, Compact Complex Surfaces (Springer, Berlin, 2004)
N. Buchdahl, Hermitian-Einstein connections and stable vector bundles over compact complex surfaces. Math. Ann. 280, 625–648 (1988)
N. Buchdahl, Algebraic deformations of compact Kähler surfaces. Math. Z. 253, 453–459 (2006)
N. Buchdahl, Algebraic deformations of compact Kähler surfaces II. Math. Z. 258, 493–498 (2008)
G. Dloussky, Structure des surfaces de Kato. Mémoires de la Société Mathématique de France, tome 14, 1–120 (1984)
G. Dloussky, Sur la classification des germes d’applications holomorphes contractantes. Math. Ann. 289(4), 649–661
G. Dloussky, F. Kohler, Classification of singular germs of mappings and deformations of compact surfaces of class VII0. Ann. Polon. Math. LXX, 49–83 (1998)
G. Dloussky, A. Teleman, Infinite bubbling in non-Kählerian geometry. Math. Ann. 353(4), 1283–1314 (2012)
G. Dloussky, A. Teleman, Smooth deformations of singular contractions of class VII surfaces. https://arxiv.org/pdf/1803.07631.pdf
G. Dloussky, K. Oeljeklaus, M. Toma, Class VII0 surfaces with b 2 curves. Tohoku Math. J. 55, 283–309 (2003)
I. Enoki, Surfaces of class VII0 with curves. Tohoku Math. J. 33, 453–492 (1981)
Ph. Engel, A proof of Looijenga’s conjecture via integral-affine geometry. Ph.D. thesis, Columbia University, 2015. ISBN: 978-1321-69596-0. arXiv:1409.7676
Ch. Favre, Classification of 2-dimensional contracting rigid germs and Kato surfaces. I. J. Math. Pures Appl. 79(5), 475–514 (2000)
P. Gauduchon, Sur la 1-forme de torsion d’une variété hermitienne compacte. Math. Ann. 267, 495–518 (1984)
M. Gross, P. Hacking S. Keel, Mirror symmetry for log Calabi-Yau surfaces I. Publ. Math. IHES 122(1), 65–168 (2015)
M. Inoue, New surfaces with no meromorphic functions. Proc. Int. Congr. Math. Vancouver 1974, 423–426 (1976)
M. Kato, Compact complex manifolds containing “global” spherical shells. Proc. Jpn. Acad. 53(1), 15–16 (1977)
M. Kato, Compact complex manifolds containing “global” spherical shells. I, in Proceedings of the International Symposium on Algebraic Geometry (Kyoto Univ., Kyoto, 1977), Kinokuniya Book Store, Tokyo (1978), pp. 45–84
M. Kato, On a certain class of nonalgebraic non-Kähler compact complex manifolds, in Recent Progress of Algebraic Geometry in Japan. North-Holland Mathematics Studies, vol. 73 (North-Holland, Amsterdam, 1983), pp. 28–50
E. Looijenga, Rational surfaces with an anti-canonical cycle. Ann. Math. 114(2), 267–322 (1981)
M. Lübke, A. Teleman, The Kobayashi-Hitchin Correspondence (World Scientific Publishing Co Pte Ltd, Singapore, 1995), 250 pp.
M. Manetti, Normal degenerations of the complex projective plane. J. Reine Angew. Math. 419, 89–118 (1991)
I. Nakamura, On surfaces of class VII0 surfaces with curves. Invent. Math. 78, 393–443 (1984)
I. Nakamura, Towards classification of non-Kählerian surfaces. Sugaku Expositions 2(2), 209–229 (1989)
I. Nakamura, On surfaces of class VII0 surfaces with curves II. Tohoku Math. J. 42(4), 475–516 (1990)
K. Oeljeklaus, M. Toma, Logarithmic moduli spaces for surfaces of class VII. Math. Ann. 341(2), 323–345 (2008)
R. Plantiko, Stable bundles with torsion Chern classes on non-Kählerian elliptic surfaces. Manuscripta Math. 87, 527–543 (1995)
Y.T. Siu, Every K3-surface is Kähler. Invent. Math. 73, 139–150 (1983)
A. Teleman, Projectively flat surfaces and Bogomolov’s theorem on class VII0-surfaces. Int. J. Math. 05(02), 253–264 (1994)
A. Teleman, Donaldson Theory on non-Kählerian surfaces and class VII surfaces with b 2 = 1. Invent. Math. 162, 493–521 (2005)
A. Teleman, Instantons and holomorphic curves on class VII surfaces. Ann. Math. 172, 1749–1804 (2010)
A. Teleman, Introduction á la théorie de jauge, Cours Spécialisés, SMF (2012)
A. Teleman, Instanton moduli spaces on non-Kählerian surfaces. Holomorphic models around the reduction loci. J. Geom. Phys. 91, 66–87 (2015)
A. Teleman, Analytic cycles in flip passages and in instanton moduli spaces over non-Kählerian surfaces. Int. J. Math. 27(07), 1640009 (2016)
A. Teleman, A variation formula for the determinant line bundle. Compact subspaces of moduli spaces of stable bundles over class VII surfaces, in Geometry, Analysis and Probability: In Honor of Jean-Michel Bismut. Progress in Mathematics (Birkhäuser, Basel, 2017)
A. Teleman, Towards the Classification of Class VII Surfaces, in Complex and Symplectic Geometry. Springer INdAM Series, vol. 21 (Springer, Cham, 2017)
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, vol. 99 (2018), pp. 363–392
Acknowledgements
The author thanks Daniele Angella, Leandro Arosio, and Eleonora Di Nezza, the organizers of the “CIME School “Complex non-Kähler geometry”, for the invitation to give a lecture series, and to submit a written version of my lectures for publication in the proceedings of the meeting. The author is grateful to Georges Dloussky for his constant help, encouragement and collaboration, and for his useful suggestions and comments on the text.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix
Appendix
3.1.1 The Picard Group and the Gauduchon Degree
Let X be a connected, compact n-dimensional complex manifold. Recall that the Picard group Pic(X) of X is the group of isomorphism classes of holomorphic line bundles on X, and can be identified with \(H^1(X,{\mathcal O}^*_X)\). The exponential short exact sequence gives the cohomology exact sequence
where the Neron-Severi group NS(X) is defined by
A class \(c\in H^2(X,{\mathbb Z})\) belongs to NS(X) if and only if it can be represented by a closed real 2-form of type (1,1). Denoting by Pic0(X) the kernel of the Chern class morphism , we obtain a short exact sequence
and an identification
Remark 1
The image of \(2\pi i H^1(X,{\mathbb Z})\) in \(H^1(X,{\mathcal O}_X)\) is closed, in particular the quotient \({H^1(X,{\mathcal O}_X)}/{2\pi i H^1(X,{\mathbb Z})}\) has the structure of an Abelian connected complex Lie group.
Proof
The obvious embedding \(2\pi i H^1(X,{\mathbb Z})\hookrightarrow H^1(X,{\mathcal O}_X)\) factorizes as
The coefficients formula shows that \(2\pi i H^1(X,{\mathbb Z})\) is a lattice in \(H^1(X,i{\mathbb R})\), in particular it is closed in this real vector space. On the other hand, using de Rham and Dolbeault theorems, it is easy to prove that the \({\mathbb R}\)-linear morphism \(H^1(X,i{\mathbb R})\to H^1(X,{\mathcal O}_X)\) is injective. Therefore \(2\pi i H^1(X,{\mathbb Z})\) is closed in an \({\mathbb R}\)-linear subspace of \(H^1(X,{\mathcal O}_X)\), so it is closed in \(H^1(X,{\mathcal O}_X)\). \(\blacksquare \)
Using (3.8) it follows that
Remark 2
Pic(X) has a natural structure of an Abelian complex Lie group, and Pic0(X) is the connected component of its unit element.
For a class c ∈NS(X) we put \(\mathrm {Pic}^c(X):= \{[{\mathcal L}]\in \mathrm {Pic}(X)|\ c_1({\mathcal L})=c\}\), and
Recall that a Gauduchon metric on X is a Hermitian metric g on X whose Kähler form ω g satisfies \(dd^c \omega _g^{n-1}=0\). An important theorem of Gauduchon states that any conformal class of Hermitian metrics on X contains a Gauduchon metric (which is unique up to constant factor if n⩾2), so there is no obstruction to the existence of Gauduchon metrics.
The degree map associated with a Gauduchon metric g on X is the group morphism
defined by
where h is a Hermitian metric on \(\mathcal {L}\), and \(F_{A_h}\in iA^{1,1}(X)\) is the curvature of the Chern connection A h associated with the pair \(({\mathcal L},h)\). Changing h will modify the Chern form \(c_1({\mathcal L},h):= \frac {i}{2\pi } F_{A_h}\) by a dd c-exact form. Since g is Gauduchon, the right hand term of (3.10) is independent of h, so degg is well defined.
3.1.2 The Kobayashi-Hitchin Correspondence for Line Bundles
Let (X, g) be a connected, compact complex manifold endowed with a Gauduchon metric. We denote by \(\Lambda ^{p,q}_X\) the bundle of (p, q)-forms on X, and by \(\Lambda _g:\Lambda ^{p,q}_X\to \Lambda ^{p-1,q-1}_X\) the adjoint of the wedge product operator \(\omega _g\wedge \cdot : \Lambda ^{p-1,q-1}_X\to \Lambda ^{p,q}_X\). The same symbol Λg will be used for the induced operator A p, q(X) → A p−1, q−1(X) between spaces of global forms. Denoting by \(\mathrm {vol}_g=\frac {1}{n!}\omega _g^{n}\) the volume form on X, and using the identity Λg ω g = n, we obtain easily the identity
Definition 3
Let \({\mathcal L}\) be a holomorphic line bundle on X. A Hermitian metric h on \({\mathcal L}\) is called Hermitian-Einstein if the real function \(i\Lambda _g F_{A_h}\) is constant. If this is the case, this constant is called the Einstein constant of h and is denoted c h.
Applying (3.11) to \(i F_{A_h}=2\pi c_1({\mathcal L},h)\) we obtain the following formula
for the Einstein constant of a Hermitian-Einstein connection on \({\mathcal L}\). Therefore a Hermitian metric h on \({\mathcal L}\) is Hermitian-Einstein if and only if
Proposition 4
Let \({\mathcal L}\) be a holomorphic line bundle on X. Then \({\mathcal L}\) admits a Hermitian-Einstein metric h, which is unique up to constant factor.
Proof
Let h 0 be an arbitrary Hermitian metric on \({\mathcal L}\), and let \(u\in {\mathcal C}^\infty (X,{\mathbb R})\). The metric h = e u h 0 is Hermite-Einstein if and only if
i.e. if and only if u is a solution of the elliptic equation
The definition of \(\mathrm {deg}_g({\mathcal L})\) gives
so the result follows from Lemma 5 below (see [LT, Corollary 1.2.9]). \(\blacksquare \)
Lemma 5
Let (X, g) be a connected, compact complex manifold endowed with a Gauduchon metric. Denote by \( \underline {{\mathbb R}}\) the line of constant real functions on X. The operator \(P=i\Lambda _g\bar \partial \partial : A^0(X,{\mathbb R})\to A^0(X,{\mathbb R})\) has the following properties:
-
(1)
\(\ker (P)= \underline {{\mathbb R}}\).
-
(2)
\(\mathrm {im}(P)= \underline {{\mathbb R}}^{\bot }\) , where the symbol ⊥ stands for the orthogonal complement with respect to the L 2 -inner product.
Proof
-
(1)
Note first that P is a second order elliptic operator which vanishes on locally defined constant functions (has no 0-order terms in local coordinates). The maximum principle applies and shows that a function \(u\in \ker (P)\) is constant around any local maximum. Therefore the non-empty closed subset \(u^{-1}(u_{\max })\subset X\) where u attains its global maximum \(u_{\max }\) is open in X. Since X is connected, it follows \(u^{-1}(u_{\max })=X\), so u is constant.
-
(2)
Since g is Gauduchon it follows easily that \( \underline {{\mathbb R}}\subset \mathrm {im}(P)^\bot \). On the other hand, the symbol of P is self-adjoint, so the index of P vanishes. Taking into account (1) it follows that \(\dim (\mathrm {im}(P)^\bot )=1\), so the inclusion \( \underline {{\mathbb R}}\subset \mathrm {im}(P)^\bot \) is an equality.
\(\blacksquare \)
Proposition 4 has an important interpretation in terms of moduli spaces. In order to explain this interpretation, we will change the point of view and we will consider variable unitary connections on a fixed differentiable Hermitian line bundle. Let (L, h) be a differentiable Hermitian line bundle on X, and let \({\mathcal A}(L,h)\) be the set of unitary connection on (L, h). This set is an affine space over the linear space \(A^1(X,i{\mathbb R})\) of \(i{\mathbb R}\)-valued 1-forms on X, so it has a natural Fréchet topology. The gauge real group \({\mathcal C}^\infty (X,\mathrm {S}^1)\) acts on \({\mathcal A}(L,h)\) in the obvious way: denoting by
the linear connection associated with A, the gauge action on \({\mathcal A}(L,h)\) satisfies the identity:
The quotient \({\mathcal M}(L,h):= {\mathcal A}(L,h)/{\mathcal C}^\infty (X,\mathrm {S}^1)\), endowed with the quotient topology, is called the moduli space of unitary connections on (L, h). It is an infinite dimensional Hausdorff space [Te4].
Definition 6
Let (L, h) be a differentiable Hermitian line bundle on X with c 1(L) ∈NS(X). A unitary connection A on (L, h) is called Hermitian-Einstein if the curvature form \(F_{A_h}\) has type (1,1) and i Λg F A is constant.
The Hermite-Einstein condition is gauge invariant so, denoting by \({\mathcal A}^{\mathrm {HE}}(L,h)\subset {\mathcal A}(L,h)\) the subspace of Hermitian-Einstein on (L, h), we obtain a closed subspace
called the moduli space of Hermitian-Einstein on (L, h).
Let \(\bar \partial _A:A^0(L)\to A^{0,1}(L)\) be the second component of the first order differential operator d A : A 0(L) → A 1(L) = A 1, 0(L) ⊕ A 0, 1(L). The first condition in Definition 6 means that \(\bar \partial _A^2=0\), so \(\bar \partial _A\) defines a holomorphic structure \({\mathcal L}_A\) on L; the corresponding sheaf of holomorphic sections is given by
for open sets U ⊂ X. The assignment \(A\mapsto {\mathcal L}_A\) induces a well defined map
called the Kobayashi-Hitchin correspondence associated with (L, h). Proposition 4 can be reformulated as follows:
Corollary 7
Let (L, h) be a differentiable Hermitian line bundle on X with c 1(L) ∈NS(X). The Kobayashi-Hitchin correspondence
is a homeomorphism.
For a class c ∈NS(X) let (L c, h c) be a differentiable Hermitian line bundle of Chern class c. The classification theorem for differentiable S1-bundles shows that (L c, h c) is well-defined up to unitary isomorphism.
Remark 8
Corollary 7 gives a homeomorphism
The moduli space \({\mathcal M}^{\mathrm {HE}}\) has a natural group structure defined by tensor product of Hermite-Einstein connections and, with respect to this structure, KH g is an isomorphism of real Lie groups.
3.1.3 Moduli Spaces of Flat S1-Connections
Recall that a Hermitian line bundle (L, h) on a compact differentiable manifold X admits a unitary flat connection if and only if c 1(L) ∈Tors. This follows easily using the cohomology exact sequence associated with the short exact sequence of constant sheaves on X
Therefore the moduli space \({\mathcal M}_{\mathrm {fl}}\) of flat S1-connections on X decomposes as
where (L c, h c) is a Hermitian line bundle of Chern class c, and \({\mathcal M}_{\mathrm {fl}}(L_c,h_c)\) denotes the moduli space of flat unitary connections on (L c, h c). The classical classification theorem for flat connections gives an identification
given by the map hol which assigns to any flat connection its holonomy representation. Note that \({\mathcal M}_{\mathrm {fl}}\) has a natural Lie group structure, and fits in the short exact sequence
where \({\mathcal M}_{\mathrm {fl}}^0\) is the moduli space of flat unitary connections on the trivial Hermitian line bundle.
Remark 9
Let X be a compact differentiable manifold. The moduli space \({\mathcal M}_{\mathrm {fl}}^0\) is canonically isomorphic to the quotient \(H^1(X,i{\mathbb R})/H^1(X,2\pi i{\mathbb Z})\)-torsor, so it is a real torus of dimension b 1(X). In particular \({\mathcal M}_{\mathrm {fl}}\) is compact, and the connected component of its unit element is the torus \({\mathcal M}_{\mathrm {fl}}^0\).
Let now X be a compact complex manifold, and g be a Gauduchon metric on X. Any flat connection on a Hermitian line bundle on X is obviously Hermite-Einstein, so we get an obvious inclusion \({\mathcal M}_{\mathrm {fl}}\hookrightarrow {\mathcal M}^{\mathrm {HE}} \) which identifies \({\mathcal M}_{\mathrm {fl}}\) with a subgroup of \({\mathcal M}^{\mathrm {HE}}\).
Remark 10
The images
of \({\mathcal M}_{\mathrm {fl}}\) (\({\mathcal M}_{\mathrm {fl}}^0\)) in Pic(X) (respectively Pic0(X)) are independent of g; the first (respectively second) image coincides with the subgroup of isomorphism classes of (topologically trivial) holomorphic line bundles \({\mathcal L}\) on X admitting a Hermitian metric h with A h flat.
Corollary 11
Let X be a complex surface, and g be a Gauduchon metric on X. One has
In particular, the kernel is independent of the Gauduchon metric g, and is a compact Lie group of real dimension b 1(X).
Proof
The inclusion is obvious and holds for manifolds of arbitrary dimension. Conversely, let \([{\mathcal L}]\in \mathrm {Pic}^T(X)\) with \(\mathrm {deg}_g({\mathcal L})=0\). Remark 8 shows that \({\mathcal L}\simeq {\mathcal L}_A\) for a Hermite-Einstein connection \(A\in {\mathcal A}^{\mathrm {HE}}(L_c)\), where \(c:= c_1({\mathcal L})\in \mathrm {Tors}\). Since \(\mathrm {deg}_g({\mathcal L})=0\), formula (3.12) gives i Λg F A = 0. Taking into account that iF A is a (1,1)-form, it follows that \(\frac {i}{2\pi }F_A\) is an anti-selfdual 2-form on the compact, oriented Riemannian 4-manifold (X, g). Since this form is also closed, it follows that it is harmonic, so it coincides with the harmonic representative of the Chern class \(c_1^{\mathrm {DR}}(L)\) in de Rham cohomology. But this de Rham class vanishes, because c 1(L) ∈Tors. Therefore \(\frac {i}{2\pi }F_A=0\), which shows that A is flat, so \([A]\in {\mathcal M}_{\mathrm {fl}}\), and \([{\mathcal L}]\in KH_g({\mathcal M}_{\mathrm {fl}})=\mathrm {Pic}_{\mathrm {ufl}}(X)\) as claimed. The other claims follow from Remarks 9 and 10. \(\blacksquare \)
The real dimension of Pic0(X) is 2q(X), where \(q(X):= \dim (H^1(X,{\mathcal O}_X))\) is the irregularity of X. By Remark 9 the real dimension of \(\mathrm {Pic}_{\mathrm {ufl}}^0(X)\) is b 1(X). Taking into account Corollary 11 we obtain:
Corollary 12
Let X be a complex surface, and let q(X) be its irregularity.
-
(1)
One has 2q(X) − 1⩽b 1(X)⩽2q(X).
-
(2)
If b 1(X) is even, then b 1(X) = 2q(X), Picufl(X) = PicT(X) and the degree map associated with any Gauduchon metric on X is a topological invariant.
-
(3)
If b 1(X) is odd, then b 1(X) = 2q(X) − 1, Picufl(X) has real codimension 1 in PicT(X) and the degree map associated with any Gauduchon metric on X induces an isomorphism
The identity
is well known [BHPV]. Our proof uses gauge theoretical methods, and gives a geometric interpretation of this identity.
Rights and permissions
Copyright information
© 2019 Springer Nature Switzerland AG
About this chapter
Cite this chapter
Teleman, A. (2019). Non-Kählerian Compact Complex Surfaces. In: Angella, D., Arosio, L., Di Nezza, E. (eds) Complex Non-Kähler Geometry. Lecture Notes in Mathematics(), vol 2246. Springer, Cham. https://doi.org/10.1007/978-3-030-25883-2_3
Download citation
DOI: https://doi.org/10.1007/978-3-030-25883-2_3
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-25882-5
Online ISBN: 978-3-030-25883-2
eBook Packages: Mathematics and StatisticsMathematics and Statistics (R0)