This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Hamiltonian Stationary Lagrangian Surfaces with Non-Negative Gaussian Curvature in Kähler-Einstein Surfaces

Patrik Coulibaly
e-mail: [email protected]
Abstract

In this paper, we give some simple conditions under which a Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein manifold must have a Euclidean factor or be a fiber bundle over a circle. We also characterize the Hamiltonian stationary Lagrangian surfaces whose Gaussian curvature is non-negative and whose mean curvature vector is in some LpL^{p} space when the ambient space is a simply connected complex space form.

1 Introduction

Let (M,g,J)(M,g,J) be a Kähler manifold of complex dimension nn. MM carries a natural symplectic structure given by the closed 22-form ω\omega which is defined by ω(X,Y)=g(JX,Y)\omega(X,Y)=g(JX,Y) for X,YTpMX,Y\in T_{p}M. We say that a Lagrangian submanifold LML\subset M is Hamiltonian stationary if it is a critical point of the volume functional under compactly supported Hamiltonian deformations, i.e. variations for which the variational vector field is of the form V=JfV=J\nabla f for some fCc(L)f\in C^{\infty}_{c}(L). In [14], Oh calculated the Euler-Lagrange equation of the variational problem and found that Hamiltonian stationary Lagrangian submanifolds are characterised by

δαH=0,\delta\alpha_{H}=0,

or equivalently by

divL(JH)=0,\text{div}_{L}(JH)=0,

where HH denotes the mean curvature vector of LL, which we define as the trace of its second fundamental form AA, i.e. H:=TrgAH\mathrel{\mathop{\mathchar 58\relax}}=Tr_{g}A, αH\alpha_{H} is the differential 11-form on LL defined by αH:=ιHω=g(JH,)\alpha_{H}\mathrel{\mathop{\mathchar 58\relax}}=\iota_{H}\omega=g(JH,\cdot) and δ\delta is the co-differential operator on LL induced by the metric gg.

By a theorem of Dazord (see, for example, Theorem 2.1 in [14]), in any Kähler manifold MM, the restriction of the Ricci form ricMric_{M} of MM to LL is given by dαHd\alpha_{H}. When MM is Kähler-Einstein, i.e. ricM=cωric_{M}=c\omega for some constant cc, then the differential 11-form αH\alpha_{H} is closed and thus defines a cohomology class in H(L)dR1H1(L;){}_{dR}^{1}(L)\cong H^{1}(L;\mathbb{R}) on any Lagrangian submanifold. Therefore, αH\alpha_{H} is both closed and co-closed, hence harmonic, on any Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein manifold.

In [2], Arsie proved that when MM is a Calabi-Yau manifold, then 1παH\frac{1}{\pi}\alpha_{H} represents an integral cohomology class of LL called the Maslov class. Therefore, we will refer to μ=1παH\mu=\frac{1}{\pi}\alpha_{H} as the Maslov form of LL.

Let NLNL denote the normal bundle of LL in MM and Γ(NL)\Gamma(NL) denote the collection of smooth sections of NLNL. Also, for any point xLx\in L and vector X¯TxM\overline{X}\in T_{x}M, let X¯\overline{X}^{\perp} denote the projection of X¯\overline{X} onto NxLN_{x}L and let ¯\overline{\nabla} denote the Levi-Civita connection on MM. Then, there is a connection \nabla^{\perp} in NLNL that is given by

XV:=(¯XV)\nabla^{\perp}_{X}V\mathrel{\mathop{\mathchar 58\relax}}=\left(\overline{\nabla}_{X}V\right)^{\perp}

for any normal vector field VΓ(NL)V\in\Gamma(NL) and tangent vector XTxLX\in T_{x}L. We say that a normal vector field VΓ(NL)V\in\Gamma(NL) is parallel if V0\nabla^{\perp}V\equiv 0.

For any point xLx\in L and vector X¯TxM\overline{X}\in T_{x}M, let X¯\overline{X}^{\top} denote the projection of X¯\overline{X} onto TxLT_{x}L. Then, since LL is Lagrangian and J=0\nabla J=0, we have that

XJV=(¯XJV)=(J¯XV)=J(¯XV)=JXV,\displaystyle\nabla_{X}JV=\left(\overline{\nabla}_{X}JV\right)^{\top}=\left(J\overline{\nabla}_{X}V\right)^{\top}=J\left(\overline{\nabla}_{X}V\right)^{\perp}=J\nabla^{\perp}_{X}V,

for any normal vector field VΓ(NL)V\in\Gamma(NL) and tangent vector XTxLX\in T_{x}L. Therefore, JVJV is parallel if and only if VV is parallel.

We say that LL has parallel second fundamental form if

(XA)(Y,Z)=XA(Y,Z)A(XY,Z)A(Y,XZ)\displaystyle(\nabla_{X}A)(Y,Z)=\nabla^{\perp}_{X}A(Y,Z)-A(\nabla_{X}Y,Z)-A(Y,\nabla_{X}Z)

vanishes for all X,Y,ZΓ(TL)X,Y,Z\in\Gamma(TL).

We present our results in two separate sections. In Section 2, we consider a complete, connected Hamiltonian stationary Lagrangian submanifold LL of arbitrary dimension nn inside a Kähler-Einstein manifold. We introduce a set of conditions, most of which consist of the non-negativity of the Ricci curvature of LL in the direction of JHJH and some pointwise or integral control over the absolute value of HH, that allows us to combine the Bochner formula for the harmonic 11-form αH\alpha_{H} and some Liouville-type theorems to deduce that HH must be parallel in the normal bundle of LL. The existence of a non-trivial global parallel vector field can restrict both the topology and the geometry of a manifold significantly. For example, if LL is simply connected, then it must be isometric to a Riemannian product of the form N×N\times\mathbb{R}. As for a purely topological consequence, if LL is not diffeomorphic to such a product, then it must admit a circle action whose orbits are not homologous to zero. In Section 3, we restrict our attention to the case when n=2n=2. We also strengthen our assumptions by requiring that our surface has non-negative Gaussian curvature which allows us to describe explicitly all complete, connected Hamiltonian stationary Lagrangian surfaces in 2\mathbb{C}^{2}, 2\mathbb{C}\mathbb{P}^{2} and in 2\mathbb{C}\mathbb{H}^{2} that has non-negative Gaussian curvature and whose mean curvature vector is in some LpL^{p} space.

The author would like to express his gratitude to Prof. Jingyi Chen for his invaluable suggestions and support, which were essential to the completion of this paper.

2 Hamiltonian Stationary Lagrangians in Kähler-Einstein Manifolds and the Bochner Method

Let 𝒜\mathcal{A} denote any of the following sets of assumptions:

  1. 1.

    RicL(JH,JH)0Ric_{L}(JH,JH)\geq 0 and |H|Lp\mathinner{\!\left\lvert H\right\rvert}\in L^{p} for some p(2,)p\in(2,\infty);

  2. 2.

    LL has non-negative Ricci curvature and |H|Lp\mathinner{\!\left\lvert H\right\rvert}\in L^{p} for some p(0,)p\in(0,\infty);

  3. 3.

    LL is oriented, RicL(JH,JH)0Ric_{L}(JH,JH)\geq 0 and |H|c:=infL|H|\mathinner{\!\left\lvert H\right\rvert}\rightarrow c\mathrel{\mathop{\mathchar 58\relax}}=\inf_{L}\mathinner{\!\left\lvert H\right\rvert} as r(x)r(x)\rightarrow\infty where r(x):=d(x0,x)r(x)\mathrel{\mathop{\mathchar 58\relax}}=d(x_{0},x) is the distance function on LL relative to a fixed point x0Lx_{0}\in L;

  4. 4.

    RicL(JH,JH)0Ric_{L}(JH,JH)\geq 0 and there exists a point x0Lx_{0}\in L, a non-decreasing function f:[0,)[0,)f\mathrel{\mathop{\mathchar 58\relax}}[0,\infty)\mapsto[0,\infty), constants C,R>0C,R>0 and p(2,)p\in(2,\infty) such that |H(x)|f(r(x))\mathinner{\!\left\lvert H(x)\right\rvert}\leq f(r(x)) for all xLx\in L and

    f(r)pVol(Br(x0))r2log(r)C\displaystyle\frac{f(r)^{p}Vol(B_{r}(x_{0}))}{r^{2}\log(r)}\leq C (1)

    whenever rRr\geq R. Here, Br(x0)B_{r}(x_{0}) denotes the geodesic ball in LL of radius rr around the point x0Lx_{0}\in L;

  5. 5.

    LL has conformal Maslov form, i.e. the vector field JHJH is conformal.

If at least one of the sets of assumptions labelled (1)–(5) is satisfied, we say that 𝒜\mathcal{A} is satisfied.

We can state the main result of this section as follows.

Theorem 2.1.

Let LL be a complete, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein manifold. If LL satisfies 𝒜\mathcal{A}, then

  • (a)

    HH is parallel and thus has constant length;

  • (b)

    RicL(JH,)Ric_{L}(JH,\cdot) vanishes identically, so if there exists a point xLx\in L such that RicL|xRic_{L}|_{x} is non-degenerate, then LL must be minimal;

  • (c)

    and the scalar curvature of LL must be constant along the integral curves of JHJH.

The growth bound (1) from condition (4) is satisfied, for example, when LL has quadratic volume growth and |H|\mathinner{\!\left\lvert H\right\rvert} does not grow faster than log(r)12p\log(r)^{\frac{1}{2p}} at infinity for some 1<p<1<p<\infty. In particular, it is satisfied when LL has quadratic volume growth and |H|L\mathinner{\!\left\lvert H\right\rvert}\in L^{\infty}. Therefore, we have the following corollary.

Corollary 2.2.

Let LL be a complete, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein manifold. If Ric(JH,JH)0Ric(JH,JH)\geq 0, |H|L\mathinner{\!\left\lvert H\right\rvert}\in L^{\infty} and LL has quadratic volume growth, then the conclusions of Theorem 2.1 hold; in particular, |H|\mathinner{\!\left\lvert H\right\rvert} must be constant.

Remark.

This phenomenon is related to the notion of parabolicity of a manifold. We say that a manifold is (strongly) parabolic if it does not admit a negative, non-constant subharmonic function, i.e. if f<0f<0 and Δf0\Delta f\geq 0, then it must be constant. It is easy to see that a parabolic manifold does not admit a non-constant subharmonic function that is bounded from above. A sufficient condition111As it is discussed in [6] after Corollary 7.7, when the Ricci curvature is non-negative, then this condition is also necessary. for the parabolicity of a manifold was given by Karp in [9], which implies, for example, that every complete, non-compact manifold with quadratic volume growth is parabolic.

The main restriction imposed on LL by the conclusion of the Theorem 2.1 is that JHJH is parallel since the existence of a non-trivial global parallel vector field restricts the topology of a manifold significantly. For example, the following result of Welsh [19], states that the existence of a complete non-trivial global parallel vector field forces the existence of a circle action whose orbits are not real homologous to zero. By a complete vector field, we mean a vector field whose integral curves are defined for all time.

Theorem 2.3 (Welsh [19]).

Suppose that MM is a Riemannian manifold that admits a non-zero complete parallel vector field. Then either MM is diffeomorphic to the product of a Euclidean space with some other manifold, or else there is a circle action on MM whose orbits are not real homologous to zero. Moreover, if MM is not diffeomorphic to the product of a Euclidean space with some other manifold and its first integral homology class is finitely generated, then MM is a fiber bundle over a circle with finite structure group.

Combining Theorem 2.1 and Theorem 2.3 gives us the following corollary.

Corollary 2.4.

Let LL be a complete, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein manifold. If LL is not minimal and it satisfies 𝒜\mathcal{A}, then LL is diffeomorphic to the product of a Euclidean space with some other manifold or there is a circle action on MM whose orbits are not real homologous to zero. Moreover, it satisfies the conclusion of Theorem 2.1; and if it is not diffeomorphic to the product of a Euclidean space with some other manifold and its first integral homology class is finitely generated, then MM is a fiber bundle over a circle with finite structure group.

Proof.

Suppose that LL is a complete, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein manifold that satisfies 𝒜\mathcal{A}. Then all the assumptions of Theorem 2.1 are satisfied thus all of its conclusions hold. In particular, JHJH is parallel so its integral curves are geodesics. Since LL is complete, all of its geodesics are defined for all tt\in\mathbb{R} and we see that JHJH is a complete vector field. Therefore, when LL is not minimal, JHJH is a non-zero complete parallel vector field on LL and we can apply Theorem 2.3 to finish the proof. ∎

When LL is not minimal, we can use Corollary 2.4 to establish the existence of a circle action on LL whose orbits are not real homologous to zero but only if LL is not diffeomorphic to a product of \mathbb{R} and some other manifold. It turns out that when LL is simply connected then the existence of such a splitting is guaranteed. Moreover, LL can be split in such a way isometrically.

Corollary 2.5.

Let LL be a complete, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein manifold. If LL is not minimal and it satisfies 𝒜\mathcal{A}, then its universal cover π:L~L\pi\mathrel{\mathop{\mathchar 58\relax}}\tilde{L}\to L equipped with the pull-back metric is isometric to N×N\times\mathbb{R} for some totally geodesic submanifold NN of L~\tilde{L}.

Proof.

Let LL be a complete, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein manifold that satisfies 𝒜\mathcal{A}. Suppose that LL is not minimal. Then, by Theorem 2.1, JHJH is a non-zero parallel vector field.

Let π:L~L\pi\mathrel{\mathop{\mathchar 58\relax}}\tilde{L}\to L denote the universal cover of LL which we equip with the pull-back metric. This makes π\pi into a Riemannian covering. We know that dαH=δαH=0d\alpha_{H}=\delta\alpha_{H}=0 and since π\pi is a local isometry, we must also have dα~H=δα~H=0d\tilde{\alpha}_{H}=\delta\tilde{\alpha}_{H}=0 for α~H=παH\tilde{\alpha}_{H}=\pi^{*}\alpha_{H}. Define JH~=(α~H)\widetilde{JH}=(\tilde{\alpha}_{H})^{\sharp}. Then JH=πJH~JH=\pi_{*}\widetilde{JH} so JH~\widetilde{JH} must also be a parallel vector field. Since L~\tilde{L} is simply connected and α~H\tilde{\alpha}_{H} is closed, there exists a smooth function fC(L~)f\in C^{\infty}(\tilde{L}) such that α~H=df\tilde{\alpha}_{H}=df or equivalently JH~=f\widetilde{JH}=\nabla f.

Since JH~=f\widetilde{JH}=\nabla f is parallel, by Lemma 2.3. in [16], ff is an affine function in the sense that fγ:f\circ\gamma\mathrel{\mathop{\mathchar 58\relax}}\mathbb{R}\to\mathbb{R} satisfies

fγ(λt1+(1λ)t2)=λfγ(t1)+(1λ)fγ(t2)f\circ\gamma(\lambda t_{1}+(1-\lambda)t_{2})=\lambda f\circ\gamma(t_{1})+(1-\lambda)f\circ\gamma(t_{2})

for all maximal unit speed geodesics γ\gamma in L~\widetilde{L}, λ(0,1)\lambda\in(0,1) and t1,t2t_{1},t_{2}\in\mathbb{R}. Also, since JH~=f\widetilde{JH}=\nabla f is non-zero, ff is a non-trivial affine function so, by a theorem of Innami [8], f1(0)f^{-1}(0) is a totally geodesic submanifold of L~\tilde{L} and f1(0)×f^{-1}(0)\times\mathbb{R} is isometric to L~\tilde{L}. ∎

In order to prove Theorem 2.1, we need the following lemma.

Lemma 2.6.

Let LL be a complete, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein manifold. If LL satisfies 𝒜\mathcal{A}, then |H|\mathinner{\!\left\lvert H\right\rvert} is constant.

Proof.

Let LL be a complete, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein manifold. Then αH=(JH)\alpha_{H}=(JH)^{\flat} is closed so we can apply the Bochner formula [15, p. 207] for |αH|2=|JH|2=|H|2\mathinner{\!\left\lvert\alpha_{H}\right\rvert}^{2}=\mathinner{\!\left\lvert JH\right\rvert}^{2}=\mathinner{\!\left\lvert H\right\rvert}^{2} to get

12ΔL|H|2\displaystyle\frac{1}{2}\Delta_{L}\mathinner{\!\left\lvert H\right\rvert}^{2} =JH,divLJH+RicL(JH,JH)+|JH|2\displaystyle=\langle JH,\nabla\text{div}_{L}JH\rangle+Ric_{L}(JH,JH)+\mathinner{\!\left\lvert\nabla JH\right\rvert}^{2}
=RicL(JH,JH)+|H|2\displaystyle=Ric_{L}(JH,JH)+\mathinner{\!\left\lvert\nabla^{\perp}H\right\rvert}^{2}

First, we observe that if 𝒜\mathcal{A} is one of the conditions (1)–(4) , then RicL(JH,JH)0Ric_{L}(JH,JH)\geq 0 and the function |H|2\mathinner{\!\left\lvert H\right\rvert}^{2} is clearly subharmonic. Since compact manifolds do not admit non-constant subharmonic functions, we may assume without the loss of generality that LL is non-compact when 𝒜\mathcal{A} is one of the conditions (1)–(4).

If |H|2\mathinner{\!\left\lvert H\right\rvert}^{2} is in LpL^{p} for some 1<p<1<p<\infty, then by a well-known result of Yau [21], |H|2\mathinner{\!\left\lvert H\right\rvert}^{2} is constant. Therefore, condition (1) implies that |H|\mathinner{\!\left\lvert H\right\rvert} is constant.

If LL has non-negative Ricci curvature, then by a result of Li and Schoen (Theorem 2.2. in [12]) it does not admit any non-negative LpL^{p} subharmonic function for all 0<p<0<p<\infty. Thus, condition (2) implies that |H|\mathinner{\!\left\lvert H\right\rvert} is constant.

Now, assume that condition (3) is satisfied. In [1], Alías, Caminha and do Nascimento prove that every non-negative subharmonic function that converges to 0 at infinity on a connected, oriented, complete and non-compact Riemannian manifold must be identically zero. Applying this maximum principle to the function f=|H|2cf=\mathinner{\!\left\lvert H\right\rvert}^{2}-c gives us that |H|2c\mathinner{\!\left\lvert H\right\rvert}^{2}\equiv c. Therefore, we conclude that condition (3) also implies that |H|\mathinner{\!\left\lvert H\right\rvert} is constant.

Next, assume that condition (4) is satisfied. Since ff is a non-negative and non-decreasing function,

1r2logrBr(x0)|H|p𝑑V\displaystyle\frac{1}{r^{2}\log r}\int_{B_{r}(x_{0})}\mathinner{\!\left\lvert H\right\rvert}^{p}dV 1r2logrBr(x0)f(r)p𝑑V\displaystyle\leq\frac{1}{r^{2}\log r}\int_{B_{r}(x_{0})}f(r)^{p}dV
=f(r)pVol(Br(x0))r2logr\displaystyle=\frac{f(r)^{p}Vol(B_{r}(x_{0}))}{r^{2}\log r}
C\displaystyle\leq C

whenever rRr\geq R. Therefore,

lim supr1r2logrBr(x0)(|H|2)p2𝑑V<\limsup_{r\to\infty}{\frac{1}{r^{2}\log r}\int_{B_{r}(x_{0})}\left(\mathinner{\!\left\lvert H\right\rvert}^{2}\right)^{\frac{p}{2}}dV}<\infty

However, in [9], Karp showed that every non-negative non-constant subharmonic function gg on a complete non-compact Riemannian manifold satisfies

lim supr1r2logrBr(x)gq𝑑V=\limsup_{r\to\infty}{\frac{1}{r^{2}\log r}\int_{B_{r}(x)}g^{q}dV}=\infty

for all q(1,)q\in(1,\infty) and center xx. Therefore, |H|2\mathinner{\!\left\lvert H\right\rvert}^{2} must be constant and we can conclude that condition (4) also implies that |H|\mathinner{\!\left\lvert H\right\rvert} is constant.

Finally, suppose that JHJH is conformal. Then, since JHJH is divergence-free,

JHg=2ndiv(JH)g=0.\mathcal{L}_{JH}\hskip 2.0ptg=\frac{2}{n}\text{div}(JH)g=0.

Therefore, the vector field JHJH is in fact Killing and the tensor JH,\langle\nabla JH,\boldsymbol{\cdot}\hskip 2.0pt\rangle is skew-symmetric. Since the dual 11-form αH\alpha_{H} is closed, we also know that JH,\langle\nabla JH,\boldsymbol{\cdot}\hskip 2.0pt\rangle is symmetric, and hence it must be zero. Therefore, JHJH is parallel which implies that it must also have constant length. We can conclude that if LL has conformal Maslov class, then |H|\mathinner{\!\left\lvert H\right\rvert} must be constant, which completes the proof. ∎

Now, we can prove Theorem 2.1.

Proof of Theorem 2.1.

Let LL be a complete, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein manifold that satisfies 𝒜\mathcal{A}. Then, as in the proof of Lemma 2.6, we have that

12ΔL|H|2=RicL(JH,JH)+|H|2.\displaystyle\frac{1}{2}\Delta_{L}\mathinner{\!\left\lvert H\right\rvert}^{2}=Ric_{L}(JH,JH)+\mathinner{\!\left\lvert\nabla^{\perp}H\right\rvert}^{2}. (2)

By Lemma 2.6, |H|\mathinner{\!\left\lvert H\right\rvert} is constant so the left-hand side of equation (2) vanishes identically. When 𝒜\mathcal{A} is one of the conditions (1)–(4), then we have two non-negative terms on the right-hand side so they must each vanish identically, i.e. we must have that RicL(JH,JH)0Ric_{L}(JH,JH)\equiv 0 and |H|20\mathinner{\!\left\lvert\nabla^{\perp}H\right\rvert}^{2}\equiv 0. When JHJH is conformal, then by the same argument that we used in the proof of Lemma 2.6, JHJH is parallel. So |H|2|JH|20\mathinner{\!\left\lvert\nabla^{\perp}H\right\rvert}^{2}\equiv\mathinner{\!\left\lvert\nabla JH\right\rvert}^{2}\equiv 0 which forces RicL(JH,JH)0Ric_{L}(JH,JH)\equiv 0. Therefore, we can conclude that HH is parallel and RicL(JH,JH)Ric_{L}(JH,JH) is identically zero whenever 𝒜\mathcal{A} is satisfied. This proves (a)(a).

Recalling the Weitzenböck formula [15, p. 211], we have

ΔαH=Trg(2αH)+RicL(JH,)\displaystyle\Delta\alpha_{H}=-Tr_{g}(\nabla^{2}\alpha_{H})+Ric_{L}(JH,\cdot)

where Δ\Delta is the Hodge-Laplacian acting on differential 11-forms. Since αH\alpha_{H} is both harmonic and parallel, we have that

ΔαH=Trg(2αH)=0\Delta\alpha_{H}=Tr_{g}(\nabla^{2}\alpha_{H})=0

and thus RicL(JH,)Ric_{L}(JH,\cdot) must also vanish identically. Let us also assume that there exists a point xLx\in L such that RicL|xRic_{L}|_{x} is non-degenerate. Since RicL|x(JH|x,JH|x)=0Ric_{L}|_{x}(JH|_{x},JH|_{x})=0, we must have that JH|x=0JH|_{x}=0. However, we know that JHJH has constant length so JHJH must vanish identically and thus LL is minimal. This proves (b)(b).

Let us also recall the contracted Bianchi identity (Proposition 7.18. [11])

12dSL=TrgRicL\displaystyle\frac{1}{2}dS_{L}=Tr_{g}\nabla Ric_{L} (3)

where SLS_{L} is the scalar curvature of LL. The trace is taken on the first and the third indices, i.e. given a local orthonormal frame E1,,EnE_{1},\dots,E_{n}, equation (3) reads as

12dSL\displaystyle\frac{1}{2}dS_{L} =(EiRicL)(,Ei).\displaystyle=(\nabla_{E_{i}}Ric_{L})(\cdot,E_{i}).

Therefore, plugging JHJH into equation (3) gives us that

12dSL(JH)\displaystyle\frac{1}{2}dS_{L}(JH) =(EiRicL)(JH,Ei)\displaystyle=(\nabla_{E_{i}}Ric_{L})(JH,E_{i})
=EiRicL(JH,Ei)RicL(EiJH,Ei)RicL(JH,EiEi)\displaystyle=E_{i}Ric_{L}(JH,E_{i})-Ric_{L}(\nabla_{E_{i}}JH,E_{i})-Ric_{L}(JH,\nabla_{E_{i}}E_{i})
=0.\displaystyle=0.

The first and the third terms vanish since RicL(JH,)0Ric_{L}(JH,\cdot)\equiv 0, while the second term is zero because JHJH is parallel. So we conclude that the scalar curvature must be constant along the integral curves of JHJH, which completes the proof of (c)(c). ∎

3 Hamiltonian Stationary Lagrangian Surfaces with Non-Negative Gaussian Curvature in Kähler-Einstein surfaces

Let Σ\Sigma denote a complete connected Hamiltonian stationary Lagrangian surface isometrically immersed in a Kähler-Einstein surface MM.

In order to obtain a characterization that is more explicit than the one given by Theorem 2.1, we adjust our sets of assumptions from before. Let 𝒜\mathcal{A}^{\prime} denote any of the following sets of assumptions:

  1. 1.

    Σ\Sigma has non-negative Gaussian curvature and |H|Lp\mathinner{\!\left\lvert H\right\rvert}\in L^{p} for some p(0,)p\in(0,\infty);

  2. 2.

    Σ\Sigma is oriented, it has non-negative Gaussian curvature and |H|c:=infΣ|H|2\mathinner{\!\left\lvert H\right\rvert}\rightarrow c\mathrel{\mathop{\mathchar 58\relax}}=\inf_{\Sigma}\mathinner{\!\left\lvert H\right\rvert}^{2} as r(x)r(x)\rightarrow\infty;

  3. 3.

    Σ\Sigma has non-negative Gaussian curvature and the growth condition (1) is satisfied;

  4. 4.

    Σ\Sigma has non-negative Gaussian curvature and conformal Maslov form.

If at least one of the sets of assumptions labelled (1)–(4) is satisfied, we say that 𝒜\mathcal{A}^{\prime} is satisfied.

We will treat the cases when Σ\Sigma is compact and when it is non-compact separately.

Theorem 3.1.

Let MM be a Kähler-Einstein surface and let Σ\Sigma be a closed connected Hamiltonian stationary Lagrangian surface in MM that satisfies 𝒜\mathcal{A}^{\prime}. If Σ\Sigma is orientable, then it is

  • a flat torus or

  • a minimal sphere.

If Σ\Sigma is not orientable, then it is

  • a flat Klein bottle or

  • a minimal projective plane.

In both cases, Σ\Sigma has parallel mean curvature.

Theorem 3.2.

Let MM be a Kähler-Einstein surface. If Σ\Sigma is a complete, connected non-compact Hamiltonian stationary Lagrangian surface in MM satisfying 𝒜\mathcal{A}^{\prime}, then it has parallel mean curvature and it is

  • isometric to 2\mathbb{R}^{2},

  • diffeomorphic to 2\mathbb{R}^{2} and is minimal or

  • it is flat and its fundamental group is isomorphic to \mathbb{Z}.

We start by proving the following lemma.

Lemma 3.3.

Let Σ\Sigma be a complete, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein surface MM. If Σ\Sigma satisfies 𝒜\mathcal{A}^{\prime} then it has parallel mean curvature and it is also flat or minimal.

Proof.

Let Σ\Sigma be a complete, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein surface MM. Since dim(Σ)=2{}_{\mathbb{R}}(\Sigma)=2, its curvature is entirely determined by its Gaussian curvature KK and, in particular,

RicΣ=KgΣ.Ric_{\Sigma}=Kg_{\Sigma}.

Suppose that Σ\Sigma satisfies 𝒜\mathcal{A}^{\prime}. It is easy to see that 𝒜\mathcal{A}^{\prime} is stronger than 𝒜\mathcal{A} so we can apply Theorem 2.1 which tells us that HH is parallel and that

K|H|2K|JH|2RicΣ(JH,JH)0.\displaystyle K\mathinner{\!\left\lvert H\right\rvert}^{2}\equiv K\mathinner{\!\left\lvert JH\right\rvert}^{2}\equiv Ric_{\Sigma}(JH,JH)\equiv 0.

Since HH is parallel, it has constant norm and therefore Σ\Sigma must be minimal or flat. ∎

Proof of Theorem 3.1.

Let Σ\Sigma be a closed, connected Hamiltonian stationary Lagrangian submanifold of a Kähler-Einstein surface MM. Also assume that Σ\Sigma satisfies 𝒜\mathcal{A}^{\prime}. Then, by Lemma 3.3, Σ\Sigma has parallel mean curvature and it must also be minimal or flat.

First, Suppose that Σ\Sigma is orientable. Then, by the Gauss-Bonnet theorem,

12πΣK𝑑A=χ(Σ)\displaystyle\frac{1}{2\pi}\int_{\Sigma}KdA=\chi(\Sigma) (4)

where χ(Σ)\chi(\Sigma) is the Euler characteristic of Σ\Sigma. Since χ(Σ)=22g\chi(\Sigma)=2-2g, where gg is the genus of Σ\Sigma, and KK is non-negative, we see that the genus must be 0 or 11. Therefore, Σ\Sigma is diffeomorphic either to a sphere or to a torus respectively. Equation (4) also tells us that Σ\Sigma is flat if and only if it has genus 11, i.e. it is a torus. So, if Σ\Sigma is not flat then it is not just minimal but it must also have genus 0 and thus it must be a minimal sphere. This completes the proof of the case when Σ\Sigma is orientable.

Now, suppose that Σ\Sigma is not orientable. In this case, χ(Σ)=2g^\chi(\Sigma)=2-\hat{g}, where g^\hat{g} is the non-orientable genus of Σ\Sigma which can be defined as the number of copies of P2\mathbb{R}P^{2} appearing when the surface is represented as a connected sum of projective planes. Also, equation (4) still holds if we interpret the left-hand side as an integral of a density. One can easily see this by passing to the orientable double cover equipped with the pull-back metric. So, similarly to the orientable case, we have that g^\hat{g} must be 11 or 22 and hence Σ\Sigma is diffeomorphic either to a real projective plane or to a Klein bottle respectively. We also see that Σ\Sigma is flat if and only if it is a Klein bottle. Therefore, when Σ\Sigma is not flat, then it is not just minimal but must also have non-orientable genus 11 and thus it is a minimal real projective plane. This completes the proof of the non-orientable case. ∎

Proof of Theorem 3.2.

It is known that the fundamental group of a non-compact surface is free (see, for example, [17, p. 142]). The first singular homology group of Σ\Sigma with coefficients in \mathbb{Z} is the abelianization of its fundamental group so H1(Σ;)H_{1}(\Sigma;\mathbb{Z}) is the free abelian group on the generator set of π1(Σ)\pi_{1}(\Sigma). Since H1(Σ;)H_{1}(\Sigma;\mathbb{Z}) is free, we know that H1(Σ;)=b1H_{1}(\Sigma;\mathbb{Z})=\mathbb{Z}^{b_{1}}, where b1b_{1} is the first Betti number of Σ\Sigma. Therefore, the cardinality of the generator set of π1(Σ)\pi_{1}(\Sigma) is equal to b1b_{1}.

First, assume that Σ\Sigma is orientable. Since K0K\geq 0, a result of Huber (Theorem 13. in [7]) tells us that Σ\Sigma is finitely connected, i.e. it is homeomorphic to a closed surface with finitely many punctures. Therefore, b1b_{1} must be finite. Moreover, since the top homology group of a non-compact manifold vanishes identically, we have that b2=0b_{2}=0 and thus χ(Σ)=1b1\chi(\Sigma)=1-b_{1}. Also, by Theorem 10. in [7],

12πΣKχ(Σ)\displaystyle\frac{1}{2\pi}\int_{\Sigma}K\leq\chi(\Sigma) (5)

so we have

b11.b_{1}\leq 1.

If Σ\Sigma is not orientable, then applying the same argument but to the orientable double cover of Σ\Sigma also yields b11b_{1}\leq 1.

Since the generator set of π1(Σ)\pi_{1}(\Sigma) has either 0 or 11 element, π1(Σ)\pi_{1}(\Sigma) is either trivial or isomorphic to \mathbb{Z}. If π1(Σ)=\pi_{1}(\Sigma)=\mathbb{Z}, then χ(Σ)=0\chi(\Sigma)=0 so, by (5), Σ\Sigma must be flat. If Σ\Sigma is simply connected, then it is diffeomorphic to 2\mathbb{R}^{2}. Finally, Lemma 3.3 tells us that Σ\Sigma has parallel mean curvature and it is also minimal or flat which finishes the proof. ∎

3.1 Hamiltonian Stationary Lagrangian Surfaces with Non-Negative Gaussian Curvature in Complex Space Forms

Let M(4c)M(4c) be a complete, connected complex space form of complex dimension 22 and constant holomorphic sectional curvature 4c4c. Let ΣM(4c)\Sigma\subset M(4c) be a Lagrangian submanifold.

We have the Wintgen-type inequality (Lemma 2.4. in [13]),

K+ρNc+|H|24\displaystyle K+\rho_{N}\leq c+\frac{\mathinner{\!\left\lvert H\right\rvert}^{2}}{4} (6)

where ρN0\rho_{N}\geq 0 is a normalized (partial) normal scalar curvature222A similar inequality can be obtained using the Chen-Ricci inequality presented, for example, in [5]..

First, we look at the case c=0c=0.

Theorem 3.4.

Let ΣM(0)\Sigma\subset M(0) be a complete, connected Hamiltonian stationary Lagrangian submanifold. If 𝒜\mathcal{A}^{\prime} is satisfied, then Σ\Sigma has parallel second fundamental form. Moreover, when the ambient manifold is 2\mathbb{C}^{2}, then Σ\Sigma is either

  • a Lagrangian plane,

  • a Riemannian product of a circle and a line (a Lagrangian cylinder),

  • or a Riemannian product of two circles (possibly of different radii).

Proof.

Let ΣM(0)\Sigma\subset M(0) be a complete, connected Hamiltonian stationary Lagrangian submanifold. Then, by Lemma 3.3, Σ\Sigma has parallel mean curvature and it must also be minimal or flat. Also, by (6),

0K|H|24\displaystyle 0\leq K\leq\frac{\mathinner{\!\left\lvert H\right\rvert}^{2}}{4} (7)

so we see that if Σ\Sigma is minimal, then it must also be flat. Therefore, we may assume, without a loss of generality, that Σ\Sigma is flat. By Theorem 2.6. in [10, p. 207], |A|=0\mathinner{\!\left\lvert\nabla A\right\rvert}=0 so we can conclude that Σ\Sigma has parallel second fundamental form.

For the rest of the proof, we assume that M(0)M(0) is 2\mathbb{C}^{2}. Let E1,E2E_{1},E_{2} be a local orthonormal frame on Σ\Sigma. Then E1,E2,JE1,JE2E_{1},E_{2},JE_{1},JE_{2} is a local orthonormal frame on 2\mathbb{C}^{2} and the components AijkA_{ij}^{k} of the second fundamental form AA of Σ\Sigma in 2\mathbb{C}^{2} are given by

A(Ei,Ej)=AijkJEk.A(E_{i},E_{j})=A_{ij}^{k}JE_{k}.

Let AkA^{k} denote the 2×22\times 2 matrix (Aijk)i,j(A_{ij}^{k})_{i,j}. Then, since Σ\Sigma is flat, by Lemma 2.5. in [10, p. 206], its second fundamental form commutes, i.e. AkAl=AlAkA^{k}A^{l}=A^{l}A^{k} for all k,l=1,2k,l=1,2. Therefore, by Theorem 2.9. in [10, p. 210], Σ\Sigma is congruent to one of the following standard Lagrangian submanifolds:

  1. 1.

    2={(z1,z2):Im(z1)=0 and Im(z2)=0}2\mathbb{R}^{2}=\{(z_{1},z_{2})\mathrel{\mathop{\mathchar 58\relax}}\text{Im}(z_{1})=0\text{ and Im}(z_{2})=0\}\subset\mathbb{C}^{2} (a Lagrangian plane),

  2. 2.

    S1(r)×={(z1,z2):|z1|2=r2 and Im(z2)=0}2S^{1}(r)\times\mathbb{R}=\{(z_{1},z_{2})\mathrel{\mathop{\mathchar 58\relax}}\mathinner{\!\left\lvert z_{1}\right\rvert}^{2}=r^{2}\text{ and Im}(z_{2})=0\}\subset\mathbb{C}^{2} for some r>0r>0 (a Lagrangian cylinder),

  3. 3.

    S1(r1)×S1(r2)={(z1,z2):|z1|2=r12 and |z2|2=r22}2S^{1}(r_{1})\times S^{1}(r_{2})=\{(z_{1},z_{2})\mathrel{\mathop{\mathchar 58\relax}}\mathinner{\!\left\lvert z_{1}\right\rvert}^{2}=r_{1}^{2}\text{ and }\mathinner{\!\left\lvert z_{2}\right\rvert}^{2}=r_{2}^{2}\}\subset\mathbb{C}^{2} for some r1,r2>0r_{1},r_{2}>0 (a product of two circles).

Before looking at the cases c>0c>0 and c<0c<0, we state some simple corollaries of Theorem 3.4.

Corollary 3.5.

Let Σ2\Sigma\subset\mathbb{C}^{2} be a complete, connected Hamiltonian stationary Lagrangian submanifold that has non-negative Gaussian curvature and |H|Lp\mathinner{\!\left\lvert H\right\rvert}\in L^{p} for some 0<p0<p\leq\infty. Then it is either a Lagrangian plane, a Lagrangian cylinder or a product of two circles. Moreover, it can only be a cylinder when p=p=\infty.

Proof.

Let Σ2\Sigma\subset\mathbb{C}^{2} be as stated in the corollary. When p(0,)p\in(0,\infty), then 𝒜\mathcal{A}^{\prime} is clearly satisfied. Since K0K\geq 0, we know by the Bishop-Gromov volume comparison theorem that Vol(Br)πr2Vol(B_{r})\leq\pi r^{2} and thus Σ\Sigma has quadratic volume growth. Therefore, as discussed in the previous section after Theorem 2.1, the growth condition (1) is satisfied whenever |H|L\mathinner{\!\left\lvert H\right\rvert}\in L^{\infty}. So 𝒜\mathcal{A}^{\prime} is satisfied when p=p=\infty as well and we can use Theorem 3.4 to conclude that Σ\Sigma must be a Lagrangian plane, a Lagrangian cylinder or a product of two circles for any 0<p0<p\leq\infty.

Finally, we note that when Σ\Sigma is non-compact, then it has infinite volume [21] so it must be minimal if it has a constant mean curvature that is in LpL^{p} for some p(0,)p\in(0,\infty). The standard Lagrangian cylinder in 2\mathbb{C}^{2} has constant mean curvature but it is neither compact nor minimal so it can only occur when p=p=\infty. ∎

We say that a complete non-compact submanifold LL is asymptoticaly minimal if its mean curvature vector HH converges to 0 at infinty, i.e. |H|0\mathinner{\!\left\lvert H\right\rvert}\rightarrow 0 as r(x)r(x)\rightarrow\infty.

Corollary 3.6.

The only complete, connected, oriented and asymptotically minimal Hamiltonian stationary Lagrangian submanifolds of 2\mathbb{C}^{2} with non-negative Gaussian curvature are Lagrangian planes.

Since a complete Kähler manifold of positive holomorphic sectional curvature is necessarily simply connected (see, for example, [18]), we may assume that M(4)M(4) is 2\mathbb{C}\mathbb{P}^{2} equipped with the standard Fubini-Study metric which has constant holomorphic sectional curvature 44. Let S5={z3:|z|=1}S^{5}=\{z\in\mathbb{C}^{3}\mathrel{\mathop{\mathchar 58\relax}}\mathinner{\!\left\lvert z\right\rvert}=1\} be the unit sphere in 3\mathbb{C}^{3} equipped with induced metric. Then the map Π:S52\Pi\mathrel{\mathop{\mathchar 58\relax}}S^{5}\to\mathbb{C}\mathbb{P}^{2} given by x[x]x\mapsto[x], which is usually referred to as the Hopf fibration, can be used to construct Lagrangian immersions into 2\mathbb{C}\mathbb{P}^{2}. For more details, see, for example, §3. in [3].

Theorem 3.7.

Let Σ2\Sigma\subset\mathbb{C}\mathbb{P}^{2} be a closed connected Hamiltonian stationary Lagrangian submanifold with non-negative Gaussian curvature. Then Σ\Sigma is

  • a totally geodesic 2\mathbb{R}\mathbb{P}^{2} or

  • flat and is locally congruent to the image of ΠL\Pi\circ L where Π:S52\Pi\mathrel{\mathop{\mathchar 58\relax}}S^{5}\to\mathbb{C}\mathbb{P}^{2} is the Hopf fibration and L:ΣS5L\mathrel{\mathop{\mathchar 58\relax}}\Sigma\to S^{5} is given by L(x,y)=(L1(x,y),L2(x,y),L3(x,y))L(x,y)=(L_{1}(x,y),L_{2}(x,y),L_{3}(x,y)) with

    L1(x,y)\displaystyle L_{1}(x,y) =aeixa1+a2,\displaystyle=\frac{ae^{-i\frac{x}{a}}}{\sqrt{1+a^{2}}},
    L2(x,y)\displaystyle L_{2}(x,y) =ei(ax+by)1+a2+b2sin(1+a2+b2y) and\displaystyle=\frac{e^{i(ax+by)}}{\sqrt{1+a^{2}+b^{2}}}\sin\left(\sqrt{1+a^{2}+b^{2}}y\right)\text{ and}
    L3(x,y)\displaystyle L_{3}(x,y) =ei(ax+by)1+a2(cos(1+a2+b2y)ib1+a2+b2sin(1+a2+b2y))\displaystyle=\frac{e^{i(ax+by)}}{\sqrt{1+a^{2}}}\left(\cos\left(\sqrt{1+a^{2}+b^{2}}y\right)-\frac{ib}{\sqrt{1+a^{2}+b^{2}}}\sin\left(\sqrt{1+a^{2}+b^{2}}y\right)\right)

    for some real constants a0a\neq 0 and bb.

Proof.

Let Σ2\Sigma\subset\mathbb{C}\mathbb{P}^{2} be a closed connected Hamiltonian stationary Lagrangian submanifold with non-negative Gaussian curvature. Then 𝒜\mathcal{A}^{\prime} is satisfied so by Theorem 3.1, Σ\Sigma has parallel mean curvature and it is a minimal sphere, a minimal real projective plane, a flat Klein bottle or a flat torus. If Σ\Sigma is a minimal sphere or a minimal real projective plane then, by Theorem 7. in [20], Σ\Sigma is immersed in such a way that its image is a totally geodesic 2\mathbb{R}\mathbb{P}^{2}. If Σ\Sigma is a flat Klein bottle or a flat torus then Σ\Sigma has parallel second fundamental form by Theorem 2.6. in [10, p. 207]. Therefore, the result follows from the classification of submanifolds with parallel second fundamental forms in 2\mathbb{C}\mathbb{P}^{2} given by Theorem 7.1. in [4].

Finally, we consider the case c<0c<0. Let 2\mathbb{C}\mathbb{H}^{2} denote the complex hyperbolic space of constant holomorphic sectional curvature 4-4, let 13\mathbb{C}^{3}_{1} denote 3\mathbb{C}^{3} equipped with the psuedo-Euclidean metric g=dz1dz¯1+dz2dz¯2+dz3dz¯3g=-dz_{1}d\bar{z}_{1}+dz_{2}d\bar{z}_{2}+dz_{3}d\bar{z}_{3} and set H15={z3:g(z,z)=1}H^{5}_{1}=\{z\in\mathbb{C}^{3}\mathrel{\mathop{\mathchar 58\relax}}g(z,z)=-1\}. Then the map Π:H152\Pi\mathrel{\mathop{\mathchar 58\relax}}H^{5}_{1}\to\mathbb{C}\mathbb{H}^{2} given by x[x]x\mapsto[x], which we will also refer to as the Hopf fibration, can be used to construct Lagrangian immersions into 2\mathbb{C}\mathbb{H}^{2}. For more details, see, for example, §3. in [3].

Theorem 3.8.

Let ΣM(4c)\Sigma\subset M(4c) be a complete, connected Hamiltonian stationary Lagrangian submanifold for some c<0c<0. If 𝒜\mathcal{A}^{\prime} is satisfied, then Σ\Sigma is flat and has parallel second fundamental form. Moreover, when the ambient manifold is 2\mathbb{C}\mathbb{H}^{2}, then Σ\Sigma is locally congruent to the image of ΠL\Pi\circ L where Π:H152\Pi\mathrel{\mathop{\mathchar 58\relax}}H^{5}_{1}\to\mathbb{C}\mathbb{H}^{2} is the Hopf fibration and L:ΣH15L\mathrel{\mathop{\mathchar 58\relax}}\Sigma\to H^{5}_{1} is given by L(x,y)=(L1(x,y),L2(x,y),L3(x,y))L(x,y)=(L_{1}(x,y),L_{2}(x,y),L_{3}(x,y)) where

  1. 1.

    L1(x,y)=ei(ax+by)1a2(cosh(1a2b2y)ib1a2b2sinh(1a2b2y))L_{1}(x,y)=\frac{e^{i(ax+by)}}{\sqrt{1-a^{2}}}\left(\cosh\left(\sqrt{1-a^{2}-b^{2}}y\right)-\frac{ib}{\sqrt{1-a^{2}-b^{2}}}\sinh\left(\sqrt{1-a^{2}-b^{2}}y\right)\right),
    L2(x,y)=ei(ax+by)1a2b2sinh(1a2b2y)L_{2}(x,y)=\frac{e^{i(ax+by)}}{\sqrt{1-a^{2}-b^{2}}}\sinh\left(\sqrt{1-a^{2}-b^{2}}y\right) and
    L3(x,y)=aeixa1a2L_{3}(x,y)=\frac{ae^{i\frac{x}{a}}}{\sqrt{1-a^{2}}} for some real constants aa and bb satisfying a0a\neq 0 and a2+b2<1a^{2}+b^{2}<1;

  2. 2.

    L1(x,y)=(ib+y)ei(1b2x+by)L_{1}(x,y)=\left(\frac{i}{b}+y\right)e^{i(\sqrt{1-b^{2}}x+by)},
    L2(x,y)=yei(1b2x+by)L_{2}(x,y)=ye^{i(\sqrt{1-b^{2}}x+by)} and
    L3(x,y)=1b2beix1b2L_{3}(x,y)=\frac{\sqrt{1-b^{2}}}{b}e^{i\frac{x}{\sqrt{1-b^{2}}}} for a real number 0<b2<10<b^{2}<1;

  3. 3.

    L1(x,y)=ei(ax+by)1a2(cos(a2+b21y)iba2+b21sin(a2+b21y))L_{1}(x,y)=\frac{e^{i(ax+by)}}{\sqrt{1-a^{2}}}\left(\cos\left(\sqrt{a^{2}+b^{2}-1}y\right)-\frac{ib}{\sqrt{a^{2}+b^{2}-1}}\sin\left(\sqrt{a^{2}+b^{2}-1}y\right)\right),
    L2(x,y)=ei(ax+by)a2+b21sin(a2+b21y)L_{2}(x,y)=\frac{e^{i(ax+by)}}{\sqrt{a^{2}+b^{2}-1}}\sin\left(\sqrt{a^{2}+b^{2}-1}y\right) and
    L3(x,y)=aeixa1a2L_{3}(x,y)=\frac{ae^{i\frac{x}{a}}}{\sqrt{1-a^{2}}} for some real constants aa and bb satisfying 0<a2<10<a^{2}<1 and a2+b2>1a^{2}+b^{2}>1;

  4. 4.

    L1(x,y)=aeixaa21L_{1}(x,y)=\frac{ae^{i\frac{x}{a}}}{\sqrt{a^{2}-1}},
    L2(x,y)=ei(ax+by)a2+b21sin(a2+b21y)L_{2}(x,y)=\frac{e^{i(ax+by)}}{\sqrt{a^{2}+b^{2}-1}}\sin\left(\sqrt{a^{2}+b^{2}-1}y\right) and
    L3(x,y)=ei(ax+by)a21(cos(a2+b21y)iba2+b21sin(a2+b21y))L_{3}(x,y)=\frac{e^{i(ax+by)}}{\sqrt{a^{2}-1}}\left(\cos\left(\sqrt{a^{2}+b^{2}-1}y\right)-\frac{ib}{\sqrt{a^{2}+b^{2}-1}}\sin\left(\sqrt{a^{2}+b^{2}-1}y\right)\right) for some real constants aa and bb satisfying a2>1a^{2}>1;

  5. 5.

    L1(x,y)=eix8b2(i+8b2(i+x)4by)L_{1}(x,y)=\frac{e^{ix}}{8b^{2}}\left(i+8b^{2}(i+x)-4by\right),
    L2(x,y)=eix8b2(i+8b2x4by)L_{2}(x,y)=\frac{e^{ix}}{8b^{2}}\left(i+8b^{2}x-4by\right) and
    L3(x,y)=ei(x+2by)2bL_{3}(x,y)=\frac{e^{i(x+2by)}}{2b} for a real number b0b\neq 0; or

  6. 6.

    L1(x,y)=eix(1+y22ix)L_{1}(x,y)=e^{ix}\left(1+\frac{y^{2}}{2}-ix\right),
    L2(x,y)=eixyL_{2}(x,y)=e^{ix}y and
    L3(x,y)=eix(y22ix)L_{3}(x,y)=e^{ix}\left(\frac{y^{2}}{2}-ix\right).

Proof.

Let ΣM(4c)\Sigma\subset M(4c) be a complete, connected Hamiltonian stationary Lagrangian submanifold for some c<0c<0. Suppose also that 𝒜\mathcal{A}^{\prime} is satisfied. Then by Lemma 3.3, Σ\Sigma has parallel mean curvature and it is also flat or minimal. However, since K0K\geq 0, it is clear from (6) that Σ\Sigma cannot be minimal. Therefore, Σ\Sigma must be flat and thus it has parallel second fundamental form by Theorem 2.6. in [10, p. 207]. When M(4c)M(4c) is 2\mathbb{C}\mathbb{H}^{2}, the result follows from the classification of submanifolds with parallel second fundamental forms in 2\mathbb{C}\mathbb{H}^{2} given by Theorem 7.2. in [4]. ∎

References

  • Alías et al. [2019] Luis José Alías, Antonio Caminha, and FY do Nascimento. A maximum principle at infinity with applications to geometric vector fields. Journal of Mathematical Analysis and Applications, 474(1):242–247, 2019.
  • Arsie [2000] Alessandro Arsie. Maslov class and minimality in Calabi–Yau manifolds. Journal of Geometry and Physics, 35(2-3):145–156, 2000.
  • Chen [1997] Bang-Yen Chen. Interaction of Legendre curves and Lagrangian submanifolds. Israel Journal of Mathematics, 99:69–108, 1997.
  • Chen et al. [2010] Bang-Yen Chen, Franki Dillen, and Joeri Van der Veken. Complete classification of parallel Lorentzian surfaces in Lorentzian complex space forms. International Journal of Mathematics, 21(05):665–686, 2010.
  • Deng [2009] Shangrong Deng. An improved Chen-Ricci inequality. Int. Electron. J. Geom, 2(2):39–45, 2009.
  • Grigor’Yan [1999] Alexander Grigor’Yan. Analytic and geometric background of recurrence and non-explosion of the Brownian motion on Riemannian manifolds. Bulletin of the American Mathematical Society, 36(2):135–249, 1999.
  • Huber [1958] Alfred Huber. On subharmonic functions and differential geometry in the large. Commentarii Mathematici Helvetici, 32(1):13–72, 1958.
  • Innami [1982] Nobuhiro Innami. Splitting theorems of Riemannian manifolds. Compositio Mathematica, 47(3):237–247, 1982.
  • Karp [1982] Leon Karp. Subharmonic functions on real and complex manifolds. Mathematische Zeitschrift, 179(4):535–554, 1982.
  • Kon and Yano [1985] Masahiro Kon and Kentaro Yano. Structures on manifolds, volume 3. World scientific, 1985.
  • Lee [2018] John M Lee. Introduction to Riemannian manifolds, volume 176. Springer, 2018.
  • Li and Schoen [1984] Peter Li and Richard Schoen. Lp\textit{L}^{p} and mean value properties of subharmonic functions on Riemannian manifolds. Acta Mathematica, 153(1):279–301, 1984.
  • Mihai [2014] Ion Mihai. On the generalized Wintgen inequality for Lagrangian submanifolds in complex space forms. Nonlinear Analysis: Theory, Methods & Applications, 95:714–720, 2014.
  • Oh [1993] Yong-Geun Oh. Volume minimization of Lagrangian submanifolds under Hamiltonian deformations. Mathematische Zeitschrift, 212(1):175–192, 1993.
  • Petersen [2006] Peter Petersen. Riemannian geometry, volume 171. Springer, 2006.
  • Sakai [1996] Takashi Sakai. On Riemannian manifolds admitting a function whose gradient is of constant norm. Kodai Mathematical Journal, 19(1):39–51, 1996.
  • Stillwell [2012] John Stillwell. Classical topology and combinatorial group theory, volume 72. Springer Science & Business Media, 2012.
  • Tsukamoto [1957] Yôtarô Tsukamoto. On Kählerian manifolds with positive holomorphic sectional curvature. Proceedings of the Japan Academy, 33(6):333–335, 1957.
  • Welsh [1986] David J Welsh. On the existence of complete parallel vector fields. Proceedings of the American Mathematical Society, 97(2):311–314, 1986.
  • Yau [1974] Shing-Tung Yau. Submanifolds with constant mean curvature. American Journal of Mathematics, 96(2):346–366, 1974.
  • Yau [1976] Shing-Tung Yau. Some function-theoretic properties of complete Riemannian manifold and their applications to geometry. Indiana University Mathematics Journal, 25(7):659–670, 1976.