betaFeedback

WorksMathematics

Entire curves in surface bundles over tori

Abstract

Cattalani recently initiated the study of Ahlfors currents in symplectic topology. We combine these with the symplectic pairs of Bande-Kotschick to show that some non-trivial surface bundles over tori have pseudoholomorphic entire curves. In particular the Kodaira-Thurston manifold has an entire curve for every JJ tamed by an invariant form.

Introduction

A nonconstant holomorphic map C→X\mathbb{C} \to X is called an entire curve and a complex manifold (X,J)(X,J) is Brody hyperbolic if it has no entire curves. This is equivalent to Kobayashi hyperbolicity for compact complex manifolds by Brody [Bro78]. Both notions are extensively studied in complex geometry [14 Kob98, 24, 3, 1].

These definitions of hyperbolicity extend to non-integrable structures: an almost complex manifold (X,J)(X,J) is called hyperbolic if every JJ-holomorphic map C→X\mathbb{C} \to X is constant. Otherwise, (X,J)(X,J) is non-hyperbolic if it contains a JJ-holomorphic entire curve. One can furthermore consider whether a fixed symplectic form ω\omega imposes either condition for all ω\omega-tame almost complex structure. This leads to the following definition:

Definition 1. Let (X,ω)(X,\omega) be a symplectic manifold. We say (X,ω)(X,\omega) is hyperbolic if, for every ω\omega-tame almost complex structure JJ, all JJ-holomorphic maps C→X\mathbb{C} \to X are constant. We say (X,ω)(X,\omega) is non-hyperbolic if, for every ω\omega-tame almost complex structure JJ, there exists a nonconstant JJ-holomorphic map C→X\mathbb{C} \to X.

The first result in this direction is due to Bangert [Ban98]. In contrast to the fact that tori never contain pseudoholomorphic rational curves, he proved the following:

Theorem 2 (Bangert [Ban98]). There exists an entire JJ-holomorphic curve C→T2n\mathbb{C} \to T^{2n} for every almost complex structure JJ tamed by the standard symplectic form ωstd\omega_{\mathrm{std}}.

Following Bangert’s result, Gromov [7] proposed development of a theory of symplectic (non-)hyperbolicity. His motivations for such an endeavor concern the study of symplectic structures on (T4,ω)(T^4,\omega); his overarching goal is to show that every symplectic structure ω\omega on T4T^4 is isomorphic to the standard symplectic structure, as is known for CP2\mathbb{CP}^2 from the study of rational curves1. If his program is viable, its machinery should surely lend valuable insights into the symplectic structures of broad classes of (aspherical) 4-manifolds, insights that rational curves miss.

Major steps toward Gromov’s vision were undertaken by Cattalani [Cat26], who initiated the study of Ahlfors currents in the symplectic category. With these objects he is able to generalize Bangert’s theorem by proving symplectic non-hyperbolicity for certain manifolds which are global products. We review his theorem in § 2 along with all relevant definitions.

Main theorem

We establish symplectic non-hyperbolicity of certain non-trivial bundles by utilizing Ahlfors currents and the theory of symplectic pairs, introduced by Bande-Kotschick [BK06].

Theorem 3. Let MM be a closed, connected 44-manifold. Suppose that a symplectic pair (ωF,ωB)(\omega_F,\omega_B) is compatible with a positive genus surface bundle

Σg→M→πT2,g≥1.\Sigma_g \to M \xrightarrow{\pi} T^2,\qquad g \ge1.

Then for the symplectic form2^2 ω=ωF+ωB\omega= \omega_F + \omega_B, the space (M,ω)(M,\omega) is non-hyperbolic.

We refer the reader to Definition 6 and Definition 7 in § 2 for the definition of a compatible symplectic pair. The general argument of the theorem is similar to that of Cattalani. We use the symplectic pair to construct expanding Lagrangians, and we show the existence of holomorphic disks with prescribed Lagrangian boundary. Then we show that these disks satisfy the required estimates to approximate Ahlfors currents.

Application to T2T^2-bundles over T2T^2

Geiges [6] studied the existence of symplectic structures within each case of the Sakamoto-Fukuhara and Ue [22 Ue90] geometric classification of T2T^2-bundles over T2T^2. See Appendix § 6 for an overview.

As an application of Theorem 3, we demonstrate symplectic non-hyperbolicity of invariant forms for new classes in Appendix § 6. In particular, Corollary 4 establishes case (b) and Corollary 5 completes cases (e),(f),(g),(h). Combined with Bangert’s seminal result for cases (a) and (c)2, this handles almost all classes of T2T^2-bundles over T2T^2. The only remaining case (d) is discussed in Conjecture 20.

Corollary 4. Let JJ be an almost complex structure on the Kodaira-Thurston manifold KK tamed by a left-invariant symplectic form. Then there exists an entire JJ-holomorphic curve C→K\mathbb{C} \to K.

Corollary 4 addresses the first part of [Cat26, Question 1]. In light of Theorem 3, it follows by showing that every left-invariant symplectic form on KK decomposes as a sum of a compatible symplectic pair. Similarly, we have:

Corollary 5. Let MM be a T2T^2-bundle over T2T^2 with geometric type Nil3×R\mathrm{Nil}_3 \times\mathbb{R} or Sol3×R\mathrm{Sol}_3 \times\mathbb{R}. Let JJ be an almost complex structure which is tamed by a left-invariant symplectic form on MM. Then there exists an entire JJ-holomorphic curve C→M\mathbb{C} \to M.

Organization of the paper

In section 2 we review relevant background on symplectic pairs and Ahlfors currents. In section 3 we construct the setting of the proofs; we use symplectic pairs to build Lagrangians and standard arguments to obtain pseudoholomorphic disks. We then prove Theorem 3 in section 4 using Ahlfors currents. Finally in section 5 we prove Corollary 4 and Corollary 5. In Appendix A we recall the classification of T2T^2-bundles over T2T^2.

Acknowledgements

I am very grateful to my advisor Tian-Jun Li for his support and guidance. This work was supported by a GAANN Fellowship.

Preliminaries

Symplectic Pairs

It is natural to consider relaxing the non-degeneracy condition on a symplectic form. But only one degenerate form is often not enough to give nontrivial geometric structures. However, we notice from Donaldson’s work [2] that a pair or triple of 2-forms instead may yield interesting geometry.

Symplectic pairs were introduced by Bande-Kotschick [BK06]. Analogous to the story for symplectic forms, versions of Darboux theorem, Moser stability [BGK04], neighborhood theorems [8], and contractibility of the space of compatible almost complex structures [9] also hold for symplectic pairs. We only consider them in four dimensions.

Definition 6. A symplectic pair on a 4-manifold MM is a pair of nontrivial closed 2-forms (ω1,ω2)(\omega_1,\omega_2) of constant and complementary ranks such that:

(i) ω1\omega_1 is a symplectic form on every leaf of ker⁡ω2\ker\omega_2.

(ii) ω2\omega_2 is a symplectic form on every leaf of ker⁡ω1\ker\omega_1.

The pair induces a symplectic form on MM given by ω+=ω1+ω2\omega^+ = \omega_1 + \omega_2 and ω−=ω1−ω2\omega^- = \omega_1 - \omega_2.

Manifolds with symplectic pairs are always symplectic, but they carry much stronger topological restrictions than general symplectic manifolds. For example, if a closed manifold M4M^4 admits a pair (ω1,ω2)(\omega_1,\omega_2), then (M,ω+)=(M,ω1+ω2)(M,\omega^+) = (M,\omega_1+\omega_2) is minimal [BG20]. Also if MM admits a symplectic pair then it has a symplectic structure for both choices of orientation.

Compatibility with a bundle structure

Definition 7. A symplectic pair (ωF,ωB)(\omega_F,\omega_B) on MM is compatible with a surface bundle

Σg→M→πΣh\Sigma_g \to M \xrightarrow{\pi} \Sigma_h

if the kernel foliation of ωB\omega_B is the fiber foliation of the surface bundle, i.e.

ker⁡ωB=ker⁡dπ\ker\omega_B = \ker d\pi

Remark 8. Note that the existence of a symplectic pair on a total space does not automatically imply the existence of a compatible symplectic pair for every bundle structure on that space. When compatible pairs do exist, they yield a flat connection and parallel transport preserves the symplectic form on the fibers.

In general, a surface bundle over a surface Σg→M→πΣh\Sigma_g \to M \xrightarrow{\pi} \Sigma_h has a compatible symplectic pair if and only if π\pi admits a flat area-preserving connection [BK06, Section 3.1]. However, when g≥3g \ge3, any Σg→M→πΣh\Sigma_g \to M \xrightarrow{\pi} \Sigma_h can be made flat by fiber summing with a trivial bundle [Bow11; KM05]. That is, M#Σg(Σg×Σk)→Σh+kM\#_{\Sigma_g}(\Sigma_g \times\Sigma_k) \to\Sigma_{h+k} can be made to support a compatible symplectic pair.

For sphere fibers, Bande-Ghiggini [BG20, Theorem 1.4] give a complete classification. They proved that S2→M→ΣhS^2 \to M \to\Sigma_h has a compatible symplectic pair for both the trivial and non-trivial bundles when h≥1h \ge1. When h=0h = 0, they proved that only the trivial bundle has a compatible symplectic pair.

In the case g=h=1g = h = 1, every oriented T2T^2-bundle over T2T^2 carries a symplectic pair [BK06] and furthermore, examples of compatible symplectic pairs also exists. [11] gives a characteristic class criteria for the existence of a fiberwise symplectic form coming from restriction of a symplectic form Ω\Omega on MM. Such a form can then be used to construct a compatible symplectic pair, but the symplectic pair does not necessarily sum to Ω\Omega. A natural question is then:

When does a prescribed symplectic form on a total space necessarily decompose as the sum of a compatible symplectic pair?

Hyperbolicity

As mentioned in the introduction, Cattalani [Cat26] initiated the study of Ahlfors currents in symplectic topology.

Definition 9. Let (X,J)(X,J) be an almost complex manifold. A 2-current T:Ω2(X)→RT:\Omega^2(X) \to\mathbb{R} is an Ahlfors current if there exists a sequence of JJ-holomorphic disks Di⊂XD_i \subset X such that

lim⁡i→∞Diarea⁡(Di)=Tandlim⁡i→∞length⁡(∂Di)area⁡(Di)=0.\lim_{i\to\infty}\frac{D_i}{\operatorname{area}(D_i)}=T \qquad\text{and} \qquad\lim_{i\to\infty}\frac{\operatorname{length}(\partial D_i)}{\operatorname{area}(D_i)}=0.

These objects are particularly useful because their existence is equivalent to the existence of entire pseudoholomorphic curves. The classical Ahlfors lemma [Ahl35] implies every entire curve yields an Ahlfors current, and Duval [4] established the opposite direction. Furthermore, such currents are much more tractable than directly dealing with entire curves.

Proposition 10. ([Ahl35], [4]) Let (X,J)(X,J) be a compact almost complex manifold. Then (X,J)(X,J) contains an Ahlfors current if and only if it contains a non-constant JJ-holomorphic map u:C→Xu:\mathbb{C} \to X.

Alongside the study of the moduli space and bubbling behavior of Ahlfors currents, Cattalani uses them to show the following existence result in certain global products:

Theorem 11 ([Cat26] Theorem 1.4). Let (X,ω)(X,\omega) be a closed connected symplectically aspherical manifold so that

(a) XX retracts onto a closed Lagrangian submanifold LXL_X and

(b) the homological Dehn function δX\delta_X of π1(X)\pi_1(X) grows at most quadratically

Then, there exists a nonconstant JJ-holomorphic map u:C→X×T2u:\mathbb{C} \to X \times T^2 for every almost complex structure tamed by ω⊕ωstd\omega\oplus\omega_{\mathrm{std}}.

Remark 12. Fuchs-Yang [5] have observed that the hypothesis of Cattalani’s theorem can be relaxed to just requiring the existence of a weakly exact Lagrangian. Similar arguments in this paper can also recover this version of Cattalani’s theorem.

Remark 13. Outside of Ahlfors currents, existence results such as [19] and [21] are modifications of Bangert’s arguments instead. The former used almost-Kähler cuts and the latter obtained a curvature criteria for symplectic non-hyperbolicity as a corollary of an almost-Kähler splitting theorem.

Lagrangian tori and pseudoholomorphic disks

In this section we construct the setting needed for the proof of Theorem 3. We first use symplectic pairs to construct the necessary Lagrangian tori and then we prove existence of the pseudoholomorphic disks that will eventually be used to estimate Ahlfors currents.

Let (M,ω)(M,\omega) be as in the assumptions of Theorem 3. Then

Σg⟶M→πT2,g≥1\Sigma_g \longrightarrow M \xrightarrow{\pi} T^2,\qquad g\ge1

where ω\omega is from a compatible symplectic pair (ωF,ωB)(\omega_F,\omega_B). Denote the kernel foliations by

V:=ker⁡ωBandH:=ker⁡ωF.\mathcal{V}:=\ker\omega_B\qquad\text{and}\qquad\mathcal{H}:=\ker\omega_F.

For each RR, take a finite covering TR2⟶T2T_R^2\longrightarrow T^2 where

TR2=R2/(nRZ)2,nR>2RT_R^2=\mathbb{R}^2/(n_R\mathbb{Z})^2,\qquad n_R>2R

and consider the pullback bundle πR:MR⟶TR2\pi_R:M_R\longrightarrow T_R^2. Note that MRM_R is aspherical. Denote again by ω\omega the pullback symplectic form on MRM_R.

The Lagrangian tori LRL_R with π2(MR,LR)≃Z\pi_2(M_R,L_R)\simeq\mathbb{Z}

Fix an essential embedded loop γ⊂Σg\gamma\subset\Sigma_g. The map π1(Σg)↪π1(MR)\pi_1(\Sigma_g)\hookrightarrow\pi_1(M_R) is injective because π2(T2)=0\pi_2(T^2)=0. Choose a disk in the universal cover R2\mathbb{R}^2 which embeds into TR2T_R^2. Denote the image of the disk in TR2T_R^2 by BRB_R. Horizontal parallel transport yields a product trivialization πR−1(BR(0))≅Σg×BR(0)\pi_R^{-1}(B_R(0))\cong\Sigma_g\times B_R(0). Set

SR:=∂BR(0)andLR:=γ×SR⊂MRS_R:=\partial B_R(0)\qquad\text{and}\qquad L_R:=\gamma\times S_R\subset M_R

and choose x0∈γx_0\in\gamma. The horizontal disk

DR,x0:={x0}×BR(0)D_{R,x_0}:=\{x_0\}\times B_R(0)

has boundary on LRL_R so we denote its relative homotopy class by βR∈π2(MR,LR)\beta_R\in\pi_2(M_R,L_R).

Lemma 14. The pair (LR,βR)(L_R,\beta_R) has the following properties.

(i) LRL_R is an embedded Lagrangian torus in (MR,ω)(M_R,\omega).

(ii) There are constants 0<c0≤C00<c_0\le C_0 independent of RR such that

c0R2≤ω(βR)≤C0R2.c_0R^2\le\omega(\beta_R)\le C_0R^2.

(iii) Any disk bounding LRL_R is generated by βR\beta_R, that is,

π2(MR,LR)≅Z⟨βR⟩\pi_2(M_R,L_R)\cong\mathbb{Z}\langle\beta_R\rangle

Proof. At each point of LRL_R, the tangent space is spanned by a vertical vector tangent to γ\gamma and a horizontal vector tangent to SRS_R. We know ωF\omega_F vanishes on horizontal vectors and ωB\omega_B vanishes on vertical vectors, hence ω∣LR=ωF∣LR+ωB∣LR=0\left.\omega\right|_{L_R}=\left.\omega_F\right|_{L_R}+\left.\omega_B\right|_{L_R}=0. Since the relative class is horizontal,

∫βRωF=0,∫βRω=∫βRωB=∫BR(0)η\int_{\beta_R}\omega_F=0,\qquad\int_{\beta_R}\omega=\int_{\beta_R}\omega_B=\int_{B_R(0)}\eta

where η\eta is the area form on T2T^2. Then (ii) follows from the area-growth estimate for disks with the Euclidean area. Now we show (iii). Since π2(MR)=0\pi_2(M_R)=0, from the long exact sequence for relative homotopy we know that π2(MR,LR)≃ker⁡{π1(LR)→i∗π1(MR)}\pi_2(M_R,L_R)\simeq\ker\{\pi_1(L_R)\xrightarrow{i_*}\pi_1(M_R)\}. Consider an arbitrary element a[γ]+b[SR]∈π1(LR)a[\gamma]+b[S_R]\in\pi_1(L_R). Since γ\gamma is essential and Σg\Sigma_g has no torsion, the image of [γ][\gamma] in π1(MR)\pi_1(M_R) has infinite order and cannot map to zero, as π1(Σg)→π1(MR)\pi_1(\Sigma_g)\to\pi_1(M_R) is injective. By construction we have i∗[SR]=0i_*[S_R]=0, which implies

ker⁡{π1(LR)→π1(MR)}≃Z⟨[SR]⟩\ker\{\pi_1(L_R)\to\pi_1(M_R)\}\simeq\mathbb{Z}\langle[S_R]\rangle

This implies that π2(MR,LR)=Z⟨βR⟩\pi_2(M_R,L_R)=\mathbb{Z}\langle\beta_R\rangle since ∂βR=[SR]\partial\beta_R=[S_R].

Existence of pseudoholomorphic disks

Now we consider the existence of pseudoholomorphic disks for the almost complex structures on the finite covers MRM_R. We adapt the argument in [Cat26, Lemma 2.26] to a local setting.

Choose a complex structure jRj_R on TR2T^2_R and a ωF\omega_F-compatible complex structure on V=ker⁡ωB\mathcal{V}=\ker\omega_B. Consider the horizontal lift of jRj_R to TMRT M_R. Let J0,RJ_{0,R} be the almost complex structure defined from the vertical complex structure and the horizontal lift of jRj_R. Then J0,RJ_{0,R} is ω\omega-compatible, preserves the splitting TMR=V⊕HT M_R=\mathcal{V}\oplus\mathcal{H}, and is a product almost complex structure on Σg×BR(0)\Sigma_g\times B_R(0). For any almost complex structure JJ on MRM_R, let M0,1(J,[D])\mathcal{M}_{0,1}(J,[D]) denote the space of JJ-holomorphic disks homologous to D∈π2(MR,LR)D\in\pi_2(M_R,L_R) with one marked point on the boundary. Let

ev⁡J:M0,1(J,[D])→L\operatorname{ev}_J:\mathcal{M}_{0,1}(J,[D])\to L

be the evaluation map taking a disk to its marked point.

Lemma 15. Every J0,RJ_{0,R}-holomorphic disk in the class βR\beta_R is Fredholm regular and the evaluation map ev⁡J0,R:M0,1(J0,R,βR)→LR\operatorname{ev}_{J_{0,R}}:\mathcal{M}_{0,1}(J_{0,R},\beta_R)\to L_R has

deg⁡(ev⁡J0,R)=1(mod2).\deg(\operatorname{ev}_{J_{0,R}})=1\pmod{2}.

Proof. Let uu be a J0,RJ_{0,R}-holomorphic disk in the class βR\beta_R. As J0,RJ_{0,R} preserves the splitting, we have

u∗ωF(∂s,∂t)=ωF(v,J0,R∣V(v))≥0(s,t)∈Du^*\omega_F(\partial_s,\partial_t)=\omega_F(v,J_{0,R}|_{\mathcal{V}}(v))\geq0\qquad(s,t)\in D

for any vertical vector v∈Vv\in\mathcal{V} along the image of uu. Note that

∫Du∗ωF=ωF(βR)=0\int_Du^*\omega_F=\omega_F(\beta_R)=0

because a representative disk is horizontal. This implies that v=0v=0 hence the image of uu lies in one horizontal leaf. In the trivialization over BR(0)B_R(0), each horizontal leaf is {x}×BR(0)\{x\}\times B_R(0). The boundary of uu lies in γ×SR\gamma\times S_R, so the constant point in the fiber satisfies x∈γx\in\gamma. Then u(D,∂D)u(D,\partial D) is one of the horizontal disks

DR,x={x}×BR(0),x∈γD_{R,x}=\{x\}\times B_R(0),\qquad x\in\gamma

up to reparametrization. Along a horizontal disk, the linearization of the Cauchy-Riemann operator is the direct sum Du=Du1⊕Du2D_u=D_{u_1}\oplus D_{u_2} for

  • (i) Du1D_{u_1} the standard linearized operator at a constant map u1:(D,∂D)→{x}u_1:(D,\partial D)\to\{x\}.

  • (ii) Du2D_{u_2} the standard linearized operator for u2:(D,∂D)→{TR2,SR}u_2:(D,\partial D)\to\{T_R^2,S_R\}.

Both operators are surjective, hence every disk in the class βR\beta_R is regular. After quotienting by reparametrization, the disk uu is determined by x∈γx\in\gamma and the boundary marked point is given by its image in SRS_R. Thus M0,1(J0,R,βR)≅γ×SR=LR\mathcal{M}_{0,1}(J_{0,R},\beta_R)\cong\gamma\times S_R=L_R. This shows that deg⁡(ev⁡J0,R)=1(mod2)\deg(\operatorname{ev}_{J_{0,R}})=1\pmod2 as the evaluation map is the identity under this identification.

Proposition 16. Let JRJ_R be any almost complex structure on MRM_R tamed by ω\omega. Then there is a nonconstant JRJ_R-holomorphic disk

uR:(D,∂D)→(MR,LR)u_R:(D,\partial D)\to(M_R,L_R)

representing βR\beta_R. Every such disk has fixed area

area⁡(uR)=∫DuR∗ω=ω(βR).\operatorname{area}(u_R)=\int_D u_R^*\omega=\omega(\beta_R).

Proof. We show disks in the class βR\beta_R are simple. In [15] and [16 Theorem A] Lazzarini shows that for any disk uu in the class βR\beta_R there exists finitely many simple disks {vi}i=1N\{v_i\}_{i=1}^N and {mi}i=1N∈N\{m_i\}_{i=1}^N\in\mathbb{N} with the same Lagrangian boundary such that

u(D)=⋃i=1Nvi(D)and[u]=∑i=1Nmi[vi]in H2(MR,LR;Z)u(D)=\bigcup_{i=1}^{N}v_i(D)\qquad\text{and}\qquad[u]=\sum_{i=1}^{N}m_i[v_i]\quad\text{in }H_2(M_R,L_R;\mathbb{Z})

Since [vi]∈π2(MR,LR)[v_i]\in\pi_2(M_R,L_R), we know from Lemma 14 that [vi]=kiβR∈π2(MR,LR)[v_i]=k_i\beta_R\in\pi_2(M_R,L_R). Note that the homological decomposition yields

ω(βR)=area⁡(u)=∑i=1Nmiarea⁡(vi)=∑i=1Nmikiω(βR)⇒1=∑i=1Nmiki\omega(\beta_R)=\operatorname{area}(u)=\sum_{i=1}^{N}m_i\operatorname{area}(v_i)=\sum_{i=1}^{N}m_ik_i\omega(\beta_R)\Rightarrow1=\sum_{i=1}^{N}m_ik_i

which implies that N=1N=1 and m1=k1=1m_1=k_1=1. Hence we obtain a simple disk vv such that [v]=βR[v]=\beta_R and area⁡(u)=area⁡(v)\operatorname{area}(u)=\operatorname{area}(v) and u(D)=v(D)u(D)=v(D). This proves that disks in βR\beta_R are simple. Disk bubbling cannot occur because every nonconstant disk component with boundary on LL has area at least ω(βR)\omega(\beta_R), so there is exactly one component. Sphere bubbling does not occur because π2(MR)=0\pi_2(M_R)=0. Thus the moduli spaces M0,1(J,βR)\mathcal{M}_{0,1}(J,\beta_R) are compact for a fixed tamed almost complex structure and for a compact one-parameter family.

Next, we first consider a special almost complex structure. Let J0,RJ_{0,R} be the ω\omega-compatible almost complex structure defined above which is a product almost complex structure on Σg×BR(0)\Sigma_g\times B_R(0). The evaluation map

ev⁡J0,R:M0,1(J0,R,βR)→LR\operatorname{ev}_{J_{0,R}}:\mathcal{M}_{0,1}(J_{0,R},\beta_R)\to L_R

has degree 1 mod 2 by Lemma 15. Consider a generic tame J1,RJ_{1,R} and choose a generic path of tame almost complex structures from J0,RJ_{0,R} to J1,RJ_{1,R}. The parameterized moduli space is a compact cobordism between the moduli spaces M0,1(J0,R,βR)\mathcal{M}_{0,1}(J_{0,R},\beta_R) and M0,1(J1,R,βR)\mathcal{M}_{0,1}(J_{1,R},\beta_R). This provides a bordism between ev⁡J0,R\operatorname{ev}_{J_{0,R}} and ev⁡J1,R\operatorname{ev}_{J_{1,R}}. Hence

deg⁡(ev⁡J1,R)=1 mod⁡2\deg(\operatorname{ev}_{J_{1,R}})=1\ \operatorname{mod} 2

So the moduli space for J1,RJ_{1,R} is nonempty. For an arbitrary ω\omega-tamed JRJ_R a Gromov compactness argument with approximating generic tame structures yields the desired result. Then for such a JRJ_R-holomorphic disk uRu_R representing βR\beta_R, the condition area⁡(uR)=ω(βR)\operatorname{area}(u_R)=\omega(\beta_R) is immediate.

Entire curves from Ahlfors currents

In this section, we prove Theorem 3.

For an arbitrary ω\omega-tamed JJ on MM, lift to JRJ_R on MRM_R. Choose a JRJ_R-holomorphic disk into MRM_R from Proposition 16. Let u~R:(D,∂D)→(M~,L~R)\widetilde{u}_R:(\mathbb{D},\partial\mathbb{D})\to(\widetilde{M},\widetilde{L}_R) denote the lift of such uRu_R to the universal cover, which is a bundle π~:M~→R2\widetilde{\pi}:\widetilde{M}\to\mathbb{R}^{2} and L~R=γ~×SR\widetilde{L}_R=\widetilde{\gamma}\times S_R for γ~\widetilde{\gamma} a lift of γ\gamma. We show that given a sequence of holomorphic disks in the universal cover M~\widetilde{M}, we can extract an Ahlfors current on the original space MM.

Let (X,g)(X,g) be a Riemannian manifold and dd the induced metric. Let S⊂XS\subset X be nonempty. Denote the distance function to SS

dS:X→[0,∞),dS(p):=inf⁡x∈S{d(p,x)}d_S:X\to[0,\infty),\qquad d_S(p):=\inf_{x\in S}\{d(p,x)\}

For a∈(0,∞]a\in(0,\infty] denote

Na(S):=dS−1([0,a))N_a(S):=d_S^{-1}([0,a))

the set of points of distance at most aa from SS. We use the following general version of Cattalani’s disk shrinking lemma.

Lemma 17 (c.f. [Cat26] Lemma 2.22). Let (D,g)(\mathbb{D},g) be a disk with a Riemannian metric gg. There exists an embedded disk D′⊂D\mathbb{D}'\subset\mathbb{D} with rectifiable boundary such that ∂D′⊂Na(∂D)\partial\mathbb{D}'\subset N_a(\partial\mathbb{D}) and

length⁡(∂D′)area⁡(D′)≤1aarea⁡(Na(∂D))area⁡(D−Na(∂D))\frac{\operatorname{length}(\partial\mathbb{D}')}{\operatorname{area}(\mathbb{D}')} \leq\frac{1}{a}\frac{\operatorname{area}(N_a(\partial\mathbb{D}))}{\operatorname{area}(\mathbb{D}-N_a(\partial\mathbb{D}))}

Proof. Fix a Riemannian disk (D,h)(\mathbb{D},h) and consider the rescaled metric h^=1a2h\widehat{h}=\frac{1}{a^{2}}h where 0<a<∞0<a<\infty. Notice that

length⁡h^=a−1length⁡h⇒N1h^(∂D)=Nah(∂D)andarea⁡h^=a−2area⁡h.\operatorname{length}_{\widehat{h}}=a^{-1}\operatorname{length}_{h}\Rightarrow N^{\widehat{h}}_{1}(\partial\mathbb{D})=N^{h}_{a}(\partial\mathbb{D})\quad\text{and}\quad\operatorname{area}_{\widehat{h}}=a^{-2}\operatorname{area}_{h}.

Then we apply [Cat26], Lemma 2.22 to (D,h^)(\mathbb{D},\widehat{h}) to obtain the inequality

alength⁡h(∂D′)area⁡h(D′)=length⁡h^(∂D′)area⁡h^(D′)≤area⁡h^(N1(∂D))area⁡h^(D−N1(∂D))=a−2area⁡h(Na(∂D))a−2area⁡h(D−Na(∂D))a\frac{\operatorname{length}_{h}(\partial\mathbb{D}')}{\operatorname{area}_{h}(\mathbb{D}')}=\frac{\operatorname{length}_{\widehat{h}}(\partial\mathbb{D}')}{\operatorname{area}_{\widehat{h}}(\mathbb{D}')} \leq\frac{\operatorname{area}_{\widehat{h}}(N_1(\partial\mathbb{D}))}{\operatorname{area}_{\widehat{h}}(\mathbb{D}-N_1(\partial\mathbb{D}))}=\frac{a^{-2}\operatorname{area}_{h}(N_a(\partial\mathbb{D}))}{a^{-2}\operatorname{area}_{h}(\mathbb{D}-N_a(\partial\mathbb{D}))}

which implies for (D,h)(\mathbb{D},h) that

length⁡h(∂D′)area⁡h(D′)≤1aarea⁡h(Na(∂D))area⁡h(D−Na(∂D))\frac{\operatorname{length}_{h}(\partial\mathbb{D}')}{\operatorname{area}_{h}(\mathbb{D}')} \leq\frac{1}{a}\frac{\operatorname{area}_{h}(N_a(\partial\mathbb{D}))}{\operatorname{area}_{h}(\mathbb{D}-N_a(\partial\mathbb{D}))}

By [Cat26, Remark 2.23] this lemma also applies to metrics induced by nonconstant pseudoholomorphic maps, e.g. hR=uR∗gJh_R=u_R^*g_J considered below.

Proposition 18. Assume the preceding setup. For R→∞R \to\infty, let uR:(D‾,∂D)→(M‾R,L‾R)u_R:(\overline{\mathbb{D}},\partial\mathbb{D})\to(\overline{M}_R,\overline{L}_R) be any holomorphic disk representing βR\beta_R, and let

u~R:(D‾,∂D)→M~\widetilde{u}_R:(\overline{\mathbb{D}},\partial\mathbb{D})\to\widetilde{M}

be their lifts to the universal cover. Then there is a constant CC and smoothly embedded subdisks DR′⊂int⁡D\mathbb{D}'_R\subset\operatorname{int}\mathbb{D} such that

length⁡(u~R∣∂DR′)area⁡(u~R∣DR′)≤CR→0.\frac{\operatorname{length}(\widetilde{u}_R|_{\partial\mathbb{D}'_R})}{\operatorname{area}(\widetilde{u}_R|_{\mathbb{D}'_R})}\leq\frac{C}{R}\to0.

After projection, a normalized subsequence converges to a nonzero Ahlfors current on MM.

Proof. The main steps are as follows.

  1. We first show that there are subdisks whose lower area bound grows quadratically.

  1. We show this lower area bound and Lemma 14 implies that the ratio

area⁡(NR(∂D))area⁡(D−NR(∂D))\frac{\operatorname{area}(N_R(\partial\mathbb{D}))}{\operatorname{area}(\mathbb{D}-N_R(\partial\mathbb{D}))}

is uniformly bounded, independent of RR.

  1. We apply Lemma 17, to get pseudoholomorphic disks with length/area ratio of order O(R−1)O(R^{-1}).

  1. We then project to MM and normalize to obtain an Ahlfors current.

(1) For any tamed JJ denote the induced metric gJg_J on MM and denote g0g_0 a flat metric on T2T^2. The induced tangent map of the bundle projection π:(M,gJ)→(T2,g0)\pi:(M,g_J)\to(T^2,g_0) has bounded operator norm since

∥dπp∥:=sup⁡v∈TpMv≠0∣dπp(v)∣∣v∣gJ,∥dπ∥:=sup⁡p∈M∥dπp∥<∞\lVert d\pi_p\rVert:=\sup_{\substack{v\in T_pM\\v\ne0}}\frac{|d\pi_p(v)|}{|v|_{g_J}},\qquad\lVert d\pi\rVert:=\sup_{p\in M}\lVert d\pi_p\rVert<\infty

and MM is compact. On each finite cover πR:MR→TR2\pi_R:M_R\to T_R^2, we use the lifted metrics of gJg_J. The supremum, denoted KK, bounds ∥dπR∥\lVert d\pi_R\rVert for all RR since MR→MM_R\to M is a local isometry. Recall that BRB_R is the embedded disk in TR2T_R^2 with SR=∂BRS_R=\partial B_R from § 3. Choose a smooth bump function χR:TR2→[0,1]\chi_R:T_R^2\to[0,1] with supp⁡(χR)⊂BR/2\operatorname{supp}(\chi_R)\subset B_{R/2} and

χR=1on BR/4.\chi_R=1\qquad\text{on }B_{R/4}.

Let μ\mu denote an area form on TR2T_R^2 and set

μR=χRμandαR=πR∗μR.\mu_R=\chi_R\mu\qquad\text{and}\qquad\alpha_R=\pi_R^*\mu_R.

αR\alpha_R is closed because dαR=dπR∗(μR)=πR∗(dμR)=0d\alpha_R=d\pi_R^*(\mu_R)=\pi_R^*(d\mu_R)=0 and the support supp⁡(αR)⊂MR\operatorname{supp}(\alpha_R)\subset M_R is disjoint from LR=γ×SRL_R=\gamma\times S_R, hence αR∣LR=0\alpha_R|_{L_R}=0. Since uRu_R and DR,x0D_{R,x_0} represent the same relative homotopy class βR\beta_R, we choose such a homotopy

H:[0,1]×D‾→MR,H([0,1]×∂D)⊂LR.H:[0,1]\times\overline{D}\to M_R,\qquad H([0,1]\times\partial D)\subset L_R.

Stokes theorem gives

∫D‾uR∗αR−∫DR,x0αR=∫[0,1]×DH∗dαR−∫[0,1]×∂DH∗αR=0\int_{\overline{D}}u_R^*\alpha_R-\int_{D_{R,x_0}}\alpha_R=\int_{[0,1]\times D}H^*d\alpha_R-\int_{[0,1]\times\partial D}H^*\alpha_R=0

Then it follows that

∫D‾uR∗αR=∫DR,x0αR=∫BRμR≥area⁡(BR/4)=πR216\int_{\overline{D}}u_R^*\alpha_R=\int_{D_{R,x_0}}\alpha_R=\int_{B_R}\mu_R\geq\operatorname{area}(B_{R/4})=\frac{\pi R^2}{16}

Now let VR=(πR∘uR)−1(BR/2)⊂DV_R=(\pi_R\circ u_R)^{-1}(B_{R/2})\subset D and note that uR∗αRu_R^*\alpha_R vanishes outside VRV_R. We have

∣uR∗αR(v,w)∣=∣uR∗πR∗μR(v,w)∣≤K2∣v∧w∣\left|u_R^*\alpha_R(v,w)\right|=\left|u_R^*\pi_R^*\mu_R(v,w)\right|\leq K^2|v\wedge w|

so that

πR216≤∫VRuR∗αR≤∫VR∣uR∗αR∣≤K2area⁡(uR∣VR)\frac{\pi R^2}{16}\leq\int_{V_R}u_R^*\alpha_R\leq\int_{V_R}|u_R^*\alpha_R|\leq K^2\operatorname{area}(u_R|_{V_R})

hence for a constant cc we have area⁡(uR∣VR)≥cR2\operatorname{area}(u_R|_{V_R})\geq cR^2.

(2) Let hR=uR∗gJh_R=u_R^*g_J be the pullback metric and dhRd_{h_R} the induced distance. For ρR=R/4K\rho_R=R/4K now consider NR:=NρR(∂D)N_R:=N_{\rho_R}(\partial D) a neighborhood of the domain disk boundary. VRV_R is disjoint from NRN_R. The preceding lower bound implies

area⁡(uR)−area⁡hR(NR)=area⁡hR(D‾∖NR)≥area⁡hR(VR)≥cR2\operatorname{area}(u_R)-\operatorname{area}_{h_R}(N_R)=\operatorname{area}_{h_R}(\overline{D}\setminus N_R)\geq\operatorname{area}_{h_R}(V_R)\geq cR^2

Using area⁡(uR)≤C0R2\operatorname{area}(u_R)\leq C_0R^2 from Lemma 14 we obtain

area⁡(uR)−area⁡hR(NR)≥δarea⁡(uR),δ=cC0\operatorname{area}(u_R)-\operatorname{area}_{h_R}(N_R)\geq\delta\operatorname{area}(u_R),\qquad\delta=\frac{c}{C_0}

For sufficiently large C0C_0, we have that 0<δ<10<\delta<1 and

area⁡hR(NR)area⁡(uR)−area⁡hR(NR)≤1−δδ.\frac{\operatorname{area}_{h_R}(N_R)}{\operatorname{area}(u_R)-\operatorname{area}_{h_R}(N_R)}\leq\frac{1-\delta}{\delta}.

(3) Now by Lemma 17 we obtain subdisks DR′⊂DD'_R\subset D with rectifiable boundary such that

length⁡(u~R∣∂DR′)area⁡(u~R∣DR′)≤CR→0.\frac{\operatorname{length}(\widetilde{u}_R|_{\partial D'_R})}{\operatorname{area}(\widetilde{u}_R|_{D'_R})}\leq\frac{C}{R}\to0.

Consider the projection of the disks to MM. A subsequence of the normalized currents of integration converges to an Ahlfors current by [23 Lemma 26.14]. □

Proof of Theorem 3. Fix an arbitrary ω\omega-tamed almost complex structure JJ on (M,ω)(M,\omega) and a sequence {Rk}→∞\{R_k\}\to\infty. For each RkR_k we construct a finite cover MRkM_{R_k}, Lagrangian torus LRk⊂MRkL_{R_k}\subset M_{R_k}, and primitive class βRk∈π2(MRk,LRk)\beta_{R_k}\in\pi_2(M_{R_k},L_{R_k}) as in Lemma 14. By Proposition 16 the lifted almost complex structure admits a pseudoholomorphic disk (D,∂D)→(MRk,LRk)(D,\partial D)\to(M_{R_k},L_{R_k}) in the class βRk\beta_{R_k}. Lift such disks to the universal cover. Thus Proposition 18 guarantees an Ahlfors current on (M,J)(M,J), and Duval’s quantitative Brody Lemma [4] gives a nonconstant JJ-holomorphic map C→M\mathbb{C} \to M. □

Left-invariant forms on T2T^2-bundles over T2T^2

In this section, we prove Corollary 4 and Corollary 5. To prove Corollary 4 it suffices to show the following:

Proposition 19. Every left-invariant symplectic form ω\omega on the Kodaira-Thurston manifold KK admits a decomposition into a symplectic pair ω=ωF+ωB\omega= \omega_F + \omega_B compatible with a bundle structure T2→K→πT2T^2 \to K \xrightarrow{\pi} T^2.

Proof. Let G=Heis⁡3(R)×RG = \operatorname{Heis}_3(\mathbb{R}) \times\mathbb{R} so that K=Γ\GK = \Gamma\backslash G and ω\omega is any left-invariant symplectic form. Following Geiges [6], consider fibrations π1,π2:K→T2\pi_1,\pi_2 : K \to T^2 induced by

(x,y,z,t)↦(x,t)and(x,y,z,t)↦(y,t)(x,y,z,t) \mapsto(x,t) \qquad\text{and} \qquad(x,y,z,t) \mapsto(y,t)

respectively. These correspond to the first and third descriptions of KK as a torus bundle in [6 Section 5(b)]. Let F1,F2F_1,F_2 denote torus fibers of these fibrations. Geiges showed that for every a∈H2(K;R)a \in H^2(K;\mathbb{R}) we have the following:

(1) If both a(F1)=0 and a(F2)=0 then a2=0.(1)\text{(1) If both } a(F_1)=0 \text{ and } a(F_2)=0 \text{ then } a^2=0. \tag*{(1)}
(2) a has a symplectic form representative if and only if a2≠0.(2)\text{(2) } a \text{ has a symplectic form representative if and only if } a^2 \ne0. \tag*{(2)}

Now let a=[ω]a=[\omega]. Taken together, (1) and (2) imply both fibrations cannot simultaneously have ω\omega vanishing on its fibers. It follows that ∫Fiω≠0\int_{F_i}\omega\ne0 for at least one i∈{1,2}i \in\{1,2\}, so choose such an index and set π=πi\pi=\pi_i. Its vertical distribution V=ker⁡dπ\mathcal{V}=\ker d\pi is invariant and ω∣V\omega|_{\mathcal{V}} is nondegenerate. Consequently, the complement H=Vω\mathcal{H}=\mathcal{V}^{\omega} is an invariant rank-two distribution, and there is a symplectic orthogonal splitting

TK=V⊕H.TK=\mathcal{V}\oplus\mathcal{H}.

Choose a translation-invariant area form σ\sigma on the base torus, and let β=π∗σ\beta=\pi^*\sigma. Notice π\pi is induced by a Lie group homomorphism, so the lift of β\beta to GG is left-invariant. The form β\beta is closed, ker⁡β=V\ker\beta=\mathcal{V}, and β∣H\beta|_{\mathcal{H}} is nondegenerate. Since ω∣H\omega|_{\mathcal{H}} and β∣H\beta|_{\mathcal{H}} are nonzero two-forms on a two-dimensional distribution, there is a unique smooth non-vanishing function c:K→Rc:K\to\mathbb{R} such that

ω∣H=cβ∣H.\omega|_{\mathcal{H}}=c\beta|_{\mathcal{H}}.

The lifts of ω\omega, β\beta, and H\mathcal{H} are all left-invariant, so the lift of cc is a left-invariant function on GG and is therefore constant. Thus cc is constant on KK. Define

ωB=cβ,ωF=ω−cβ.\omega_B=c\beta,\qquad\omega_F=\omega-c\beta.

Constancy of cc implies dωB=dωF=0d\omega_B=d\omega_F=0. By construction,

ωF∣H=0,ωB∣V=0,ωF∣V=ω∣V,ωB∣H=ω∣H.\omega_F|_{\mathcal{H}}=0,\qquad\omega_B|_{\mathcal{V}}=0,\qquad\omega_F|_{\mathcal{V}}=\omega|_{\mathcal{V}},\qquad\omega_B|_{\mathcal{H}}=\omega|_{\mathcal{H}}.

Moreover, for v∈Vv\in\mathcal{V} and h∈Hh\in\mathcal{H},

ω(v,h)=0,β(v,h)=0,\omega(v,h)=0,\qquad\beta(v,h)=0,

so both ωF\omega_F and ωB\omega_B have zero mixed terms. Since their nonzero restrictions are nondegenerate, it follows that

ker⁡ωF=H,ker⁡ωB=V.\ker\omega_F=\mathcal{H},\qquad\ker\omega_B=\mathcal{V}.

Both forms therefore have constant rank two. Their kernel distributions are integrable because the forms are closed, and each form restricts symplectically to the leaves of the kernel of the other. Thus (ωF,ωB)(\omega_F,\omega_B) forms a symplectic pair compatible with the bundle structure for KK, and their sum is the prescribed form ω\omega. □

We now restate and prove Corollary 5.

Corollary (Corollary 5). Let MM be a T2T^2-bundle over T2T^2 with geometric type Nil3×R\mathrm{Nil}^3 \times\mathbb{R} or Sol3×R\mathrm{Sol}^3 \times\mathbb{R}. Let JJ be an almost complex structure tamed by a left-invariant symplectic form on MM. Then there exists an entire JJ-holomorphic curve C→M\mathbb{C} \to M.

Proof. Consider T2T^2-bundles MM of geometric type Sol3×R\mathrm{Sol}^3 \times\mathbb{R}. These are the bundles cases (g) and (h) in Appendix § 6. In light of Theorem 3, we just need to show every left invariant symplectic form on MM can be decomposed with a symplectic pair compatible with π\pi. [BK06, Theorem 18, Case (g, h)] shows that symplectic pairs on MM exist. Furthermore by [6 Theorem 2] any prescribed symplectic form on MM always has ω\omega non-vanishing on its fibers. Therefore the construction in Proposition 19 can essentially be repeated verbatim.

The same applies to bundles with type Nil3×R\mathrm{Nil}^3 \times\mathbb{R} with b1=2b_1 = 2 [6 Theorem 1] and for the b1=3b_1 = 3 case this was Proposition 19. One can alternatively use the previous result Corollary 4 and note that cases (e) and (f) have a Kodaira-Thurston finite cover, so JJ-holomorphic entire curves descend via projection. The nonconstant property follows since the covering map is a local homeomorphism. □

Conjecture 20 (Case (d)). Let MM be a T2T^2-bundle over T2T^2 with geometric type Nil4\mathrm{Nil}^4. Let JJ be an almost complex structure which is tamed by a left-invariant symplectic form on MM. Then there exists a nonconstant entire JJ-holomorphic curve C→M\mathbb{C} \to M.

This is the only remaining case in the classification of T2T^2-bundles over T2T^2. In this setting, the torus fibration does not have symplectic torus fibers (see Appendix § 6). So although they admit pairs [BK06, Theorem 18 Case (d)], they are never compatible with the fibration. Nonetheless, we still expect symplectic non-hyperbolicity. One heuristic reason might be [10 Lemma 2.5, Conjecture 4.9], which shows the Lagrangian bundle structure is preserved by Luttinger surgery and predicts that all T2T^2-bundles over T2T^2 are related by such surgery. Since cases (a),(b), and (d) have Lagrangian bundle structure and are related under surgery, one might expect entire curves to be preserved if they avoid the Lagrangian fiber.

In general it would be interesting to study entire JJ-holomorphic curves with further qualitative control, such as those which either always avoid or pass through prescribed sets.

Classification of orientable T2T^2-bundles over T2T^2

The following is the classification of T2T^2-bundles over T2T^2 and their Thurston geometries, due to [22] and [Ue90] respectively. The matrices describe the monodromy corresponding to generators of π1(T2)\pi_1(T^2), and pairs of integers (m,n)(m,n) represent the Euler class. Geiges [6 Theorem 1] shows that outside of cases (b) and (d) the fibration of MM as a torus bundle is unique. [6 Theorem 2] shows that in cases (a) and (b) a given invariant symplectic form may or may not be compatible with a selected fibration. For case (d) the symplectic structure is never compatible with the fibration. For all other cases the symplectic structure is always compatible with the fibration.

b1b_1Monodromy & Euler classGeometric type
(a)4{I,I,(0,0)}=T4\{I,I,(0,0)\}=T^4R4\mathbb{R}^4
(b)3{I,I,(m,n)}\{I,I,(m,n)\}, (m,n)≠(0,0)(m,n)\ne(0,0)Nil⁡3×R\operatorname{Nil}^3\times\mathbb{R}
(c)2{(0−11−1),I,(0,0)}\left\{\begin{pmatrix}0&-1\\1&-1\end{pmatrix},I,(0,0)\right\}<br>{(0−11−1),I,(−1,0)}\left\{\begin{pmatrix}0&-1\\1&-1\end{pmatrix},I,(-1,0)\right\}<br>{(0−110),I,(0,0)}\left\{\begin{pmatrix}0&-1\\1&0\end{pmatrix},I,(0,0)\right\}<br>{(0−110),I,(−1,0)}\left\{\begin{pmatrix}0&-1\\1&0\end{pmatrix},I,(-1,0)\right\}<br>{(1−110),I,(0,0)}\left\{\begin{pmatrix}1&-1\\1&0\end{pmatrix},I,(0,0)\right\}<br>{−I,I,(0,0)}\{-I,I,(0,0)\}<br>{−I,I,(−1,0)}\{-I,I,(-1,0)\}R4\mathbb{R}^4
(d)2{(1λ01),I,(m,n)}\left\{\begin{pmatrix}1&\lambda\\0&1\end{pmatrix},I,(m,n)\right\}, λ≠0\lambda\ne0, n≠0n\ne0Nil⁡4\operatorname{Nil}^4
(e)2{(−1λ0−1),I,(m,n)}\left\{\begin{pmatrix}-1&\lambda\\0&-1\end{pmatrix},I,(m,n)\right\}, λ≠0\lambda\ne0Nil⁡3×R\operatorname{Nil}^3\times\mathbb{R}
(f)2{(1λ01),−I,(m,n)}\left\{\begin{pmatrix}1&\lambda\\0&1\end{pmatrix},-I,(m,n)\right\}, λ≠0\lambda\ne0Nil⁡3×R\operatorname{Nil}^3\times\mathbb{R}
(g)2{C,I,(m,n)}\{C,I,(m,n)\}, ∣tr⁡C∣≥3|\operatorname{tr}C|\ge3, C∈SL⁡2(Z)C\in\operatorname{SL}_2(\mathbb{Z})Sol⁡3×R\operatorname{Sol}^3\times\mathbb{R}
(h)2{C,−I,(m,n)}\{C,-I,(m,n)\}, tr⁡C≥3\operatorname{tr}C\ge3, C∈SL⁡2(Z)C\in\operatorname{SL}_2(\mathbb{Z})Sol⁡3×R\operatorname{Sol}^3\times\mathbb{R}

Table 1.

References

School of Mathematics, University of Minnesota, Minneapolis, MN, US

E-mail address: [email protected]

References

  1. [1]Jean-Pierre Demailly. “Recent results on the Kobayashi and Green-Griffiths-Lang conjectures”. In: Japanese Journal of Mathematics 15.1 (2020), pp. 1–120. DOI: 10.1007/s11537–019–1566–3.
  2. [2]S. K. Donaldson. Two-forms on four-manifolds and elliptic equations. 2006. arXiv: math/0607083 [math.DG].DOI
  3. [3]Simone Diverio and Erwan Rousseau. Hyperbolicity of Projective Hypersurfaces. Vol. 5. IMPA Monographs. Springer Cham, 2016. ISBN: 978-3-319-32314-5. DOI: 10 . 1007 / 978 – 3 – 319 – 32315 – 2.DOI
  4. [4]Julien Duval. “Sur le lemme de Brody”. In: Inventiones mathematicae (2008).DOI
  5. [5]Urs Fuchs and Yuchen Yang. “Holomorphic planes in M × T² from weakly exact Lagrangians in M”. In: (2026). arXiv: 2611.07799 [math.SG]. URL: https://arxiv.org/abs/2611.07799.
  6. [6]Hansjörg Geiges. “Symplectic structures on T²-bundles over T²”. In: Duke Mathematical Journal (1992).DOI
  7. [7]Misha Gromov. “Spaces and Questions”. In: Geometric and Functional Analysis Special Volume (2000). GAFA 2000 (GAFA2000) Special Volume, Part I, pp. 118–161.
  8. [8]Hai-Long Her. “On neighborhood theorems for symplectic pairs”. In: Journal of Geometry (2015).DOI
  9. [9]Hai-Long Her. “Almost complex structures for symplectic pairs”. In: Topology and its Applications (2018).DOI
  10. [10]Chung-I Ho and Tian-Jun Li. “Luttinger surgery and Kodaira dimension”. In: Asian Journal of Mathematics (2012).arxiv.org/abs/1108.0479
  11. [11]Peter J. Kahn. “Symplectic torus bundles and group extensions”. In: (2004). arXiv: math/0405109 [math.SG].DOI
  12. [14]Serge Lang. Introduction to complex hyperbolic spaces. Springer-Verlag, New York, 1987, pp. viii+271. ISBN: 0-387-96447-9. DOI: 10.1007/978–1–4757–1945–1. URL: https://doi-org.ezp2.lib.umn.edu/10.1007/978–1–4757–1945–1.DOI
  13. [15]Laurent Lazzarini. “Existence of a somewhere injective pseudo-holomorphic disc”. In: Geometric and Functional Analysis (2000).DOI
  14. [16]Laurent Lazzarini. “Relative frames on J-holomorphic curves”. In: Journal of Fixed Point Theory and Applications (2011).DOI
  15. [17]Tian-Jun Li and Ai-Ko Liu. “Symplectic structure on ruled surfaces and a generalized adjunction formula”. In: Mathematical Research Letters 2.4 (1995), pp. 453–471. DOI: 10.4310/MRL.1995.v2.n4.a6.DOI
  16. [18]François Lalonde and Dusa McDuff. “The classification of ruled symplectic 4-manifolds”. In: Mathematical Research Letters 3.6 (1996), pp. 769–778. DOI: 10.4310/MRL.1996.v3.n6.a5.
  17. [19]Tian-Jun Li and Weiwei Wu. “Note on a theorem of Bangert”. In: Acta Math. Sin. (Engl. Ser.) (2012).
  18. [20]Dusa McDuff. “The structure of rational and ruled symplectic 4-manifolds”. In: Journal of the American Mathematical Society 3.3 (1990), pp. 679–712.DOI
  19. [21]Anthony Nguyen, Shengzhen Ning, and Lauren Pusey-Nazzaro. “An almost Kähler Cheeger–Gromoll splitting theorem with applications”. In: (2026). arXiv: 2608.18477 [math.DG]. URL: https://arxiv.org/abs/2608.18477.
  20. [22]Koichi Sakamoto and Shinji Fukuhara. “Classification of T²-bundles over T²”. In: Tokyo Journal of Mathematics (1983).
  21. [23]Leon Simon. Lectures on Geometric Measure Theory. Vol. 3. Proceedings of the Centre for Mathematical Analysis, Australian National University. Australian National University, Centre for Mathematical Analysis, 1983.
  22. [24]Yum-Tong Siu. “Hyperbolicity in Complex Geometry”. In: The Legacy of Niels Henrik Abel: The Abel Bicentennial, Oslo, 2002. Ed. by Olav Arnfinn Laudal and Ragni Piene. Berlin, Heidelberg: Springer Berlin Heidelberg, 2004, pp. 543–566. ISBN: 978-3-642-18908-1. DOI: 10.1007/978-3-642-18908-1_16. URL: https://doi.org/10.1007/978-3-642-18908-1_16.
  23. [25]Clifford Henry Taubes. “The Seiberg–Witten and Gromov invariants”. In: Mathematical Research Letters 2.2 (1995), pp. 221–238.DOI

Paper details

vlnc:W_j6apdsaq
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 4

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents