Contents lists available at ScienceDirect

Journal of Mathematical Analysis and Applications www.elsevier.com/locate/jmaa

Vanishing theorems for L2 harmonic 1-forms on complete submanifolds in a Riemannian manifold Nguyen Thac Dung a , Keomkyo Seo b,∗ a

Department of Mathematics, Mechanics, and Informatics (MIM), Hanoi University of Sciences (HUS–VNU), Vietnam National University, 334 Nguyen Trai Str., Thanh Xuan, Hanoi, Vietnam b Department of Mathematics, Sookmyung Women’s University, Hyochangwongil 52, Yongsan-ku, Seoul, 140-742, Republic of Korea

a r t i c l e

i n f o

a b s t r a c t

Article history: Received 30 June 2014 Available online 31 October 2014 Submitted by H.R. Parks

Let M be an n-dimensional complete orientable noncompact hypersurface in a complete Riemannian manifold of nonnegative sectional curvature. For 2 ≤ n ≤ 6, we prove that if M satisﬁes the δ-stability inequality (0 < δ ≤ 1), then there is no nontrivial L2β harmonic 1-form on M for some constant β. We also provide suﬃcient conditions for complete hypersurfaces to satisfy the δ-stability inequality. Moreover, we prove a vanishing theorem for L2 harmonic 1-forms on M when M is an n-dimensional complete noncompact submanifold in a complete simply-connected Riemannian manifold N with sectional curvature KN satisfying that −k2 ≤ KN ≤ 0 for some constant k. © 2014 Elsevier Inc. All rights reserved.

Keywords: δ-Stability inequality L2 harmonic 1-form Traceless second fundamental form First eigenvalue

1. Introduction Let M n be an n-dimensional orientable minimal hypersurface in a Riemannian manifold N of nonnegative sectional curvature. We recall that a minimal hypersurface in a Riemannian manifold is called stable provided the second variation of the volume is nonnegative for any normal variation on a compact subset. More precisely, a minimal hypersurface M in a Riemannian manifold N is said to be called stable if for any f ∈ C0∞ (M )

|∇f |2 − |A|2 + Ric(ν, ν) f 2 dv ≥ 0,

(1.1)

M

where A is the second fundamental form, Ric is the Ricci curvature of N , ν is the unit normal vector of M , and dv is the volume form on M . * Corresponding author. E-mail addresses: [email protected] (N.T. Dung), [email protected] (K. Seo). URL: http://sookmyung.ac.kr/~kseo (K. Seo). http://dx.doi.org/10.1016/j.jmaa.2014.10.076 0022-247X/© 2014 Elsevier Inc. All rights reserved.

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1595

On the other hand, for a number 0 < δ ≤ 1, it is called δ-stable if any function f ∈ C0∞ (M ) satisﬁes

|∇f |2 − δ |A|2 + Ric(ν, ν) f 2 dv ≥ 0.

(1.2)

M

It is obvious that δ1 -stability implies δ2 -stability for 0 < δ2 < δ1 ≤ 1. In particular, if M is stable, then M is δ-stable for 0 < δ ≤ 1. There have been plenty of works on δ-stable complete minimal hypersurfaces in a Riemannian manifold. (See [3,8,10,19,27] and references therein for more details.) It is well-known that the only complete orientable stable minimal surface in R3 is a plane [4,7]. For 1/8 < δ, Kawai [15] showed that a δ-stable complete minimal surface in R3 should be a plane. Furthermore, do Carmo and Peng [5] proved that if a stable complete minimal hypersurface M in the Euclidean space with M |A|2 dv < ∞, then M is a hyperplane. Later, Shen and Zhu [26] proved that an n-dimensional stable complete minimal hypersurface M in the Euclidean space with M |A|n dv < ∞ is a hyperplane. Recently, Tam and Zhou proved that a complete n−2 n -stable minimal hypersurface whose second fundamental form satisﬁes some decay conditions in the Euclidean space is either a hyperplane or a catenoid. In case of complete orientable stable minimal hypersurfaces, several results on the nonexistence of L2 harmonic forms are well-known. Palmer [22] proved that if there exists a codimension one cycle on a complete minimal hypersurface M in the Euclidean space which does not separate M , then M is unstable by using the nonexistence of L2 harmonic 1-form. Thereafter, using Bochner’s vanishing technique, Miyaoka [20] showed that a complete orientable noncompact stable minimal hypersurface in a nonnegatively curved manifold has no nontrivial L2 harmonic 1-forms. In [32], Yun proved that if M ⊂ Rn+1 is a complete minimal hypersurface with suﬃciently small total scalar curvature M |A|n , then there is no nontrivial L2 harmonic 1-form on M . Yun’s result has been generalized into various ambient spaces [2,6,23–25]. For an n-dimensional complete orientable noncompact (not necessarily minimal) hypersurface M in a complete manifold N of nonnegative sectional curvature with 2 ≤ n ≤ 4, Kim and Yun [16] recently proved that if M satisﬁes the stability inequality (1.1), then there is no nontrivial L2 harmonic 1-form on M , which is an extension of a well-known fact in the case when M is a complete stable minimal hypersurface in N . In Section 2, motivated by this, we prove that if M is an n-dimensional complete noncompact (not necessarily minimal) hypersurface in a complete manifold N of nonnegative sectional curvature and M satisﬁes the δ-stability inequality (1.2) for a number 0 < δ ≤ 1, then there is no nontrivial L2β harmonic 1-form on M for some constant β. (See Theorem 2.6 for more details.) As a consequence, we extend Kim and Yun’s result into the case when n = 5, 6. In Section 3, we also provide suﬃcient condition for complete hypersurfaces to satisfy the δ-stability inequality in a Riemannian manifold. In Section 4, we deal with complete noncompact submanifold cases. For an n-dimensional complete noncompact submanifold M in a complete simply-connected Riemannian manifold N with sectional curvature KN satisfying that −k2 ≤ KN ≤ 0 for some constant k, it turns out that if the L2 norm φn of the traceless second fundamental form φ is suﬃciently small and the ﬁrst eigenvalue λ1 (M ) of the Laplacian is bigger than some constant depending only on k, n, and φn , then there is no nontrivial L2 harmonic 1-form on M . 2. Harmonic 1-forms on complete hypersurfaces of lower dimensions A complete manifold M is called non-parabolic if it has a positive Green function. Otherwise, M is called parabolic. We note that M is non-parabolic provided it has a non-constant positive superharmonic function on M . The following suﬃcient condition for parabolicity is well-known. Theorem. (See [11,12,14,28].) Let M be a complete manifold. If, for any point p ∈ M and a geodesic ball Bp (r) ⊂ M ,

1596

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

∞

r dr = ∞, Vol(Bp (r))

1

then M is parabolic. Using the above theorem, we see that if M is non-parabolic, then ∞

r dr < ∞, Vol(Bp (r))

(2.1)

1

and hence M has an inﬁnite volume. Deﬁnition 2.1. Let M n be an n-dimensional orientable hypersurface in a Riemannian manifold N . We say the δ-stability inequality holds on M for 0 < δ ≤ 1 if any f ∈ C0∞ (M ) satisﬁes

|∇f |2 − δ |A|2 + Ric(ν, ν) f 2 dv ≥ 0.

M

It turns out that a complete orientable noncompact hypersurface in a complete manifold with nonnegative sectional curvature has an inﬁnite volume if the δ-stability inequality holds for 0 < δ ≤ 1. Lemma 2.2. Let M n be a complete orientable noncompact hypersurface in a complete manifold N with nonnegative sectional curvature. If the δ-stability inequality holds on M for 0 < δ ≤ 1, then the volume of M is inﬁnite. Proof. If M is non-parabolic, then M has an inﬁnite volume by (2.1). We now assume that M is parabolic. Given 0 < δ ≤ 1, since the δ-stability inequality holds on M , we have

|∇ϕ| ≥ δ 2

M

Ric(ν, ν) + |A|2 ϕ2

M

for any f ∈ C0∞ (M ). Let q := δ(|A|2 + Ric(ν, ν)) and let D ⊂ M be any bounded domain with smooth boundary. Denote by λq1 (D) the ﬁrst eigenvalue of the Schrödinger operator Δ + q acting on functions vanishing on ∂D. The assumption that the δ-stability holds on M is equivalent to that λq1 (D) ≥ 0 for any bounded domain D ⊂ M . From the result in [7], it follows that there is a positive function u such that the equation Δu + qu = 0 on M . Since the sectional curvature of N is nonnegative, u is a positive superharmonic function on M . The parabolicity of M implies that u is constant. Hence |A| ≡ 0, which shows that M is totally geodesic in N . Thus M has nonnegative Ricci curvature, which gives the conclusion that M has an inﬁnite volume [31]. 2 Let M n be an n-dimensional orientable submanifold in an (n + p)-dimensional Riemannian manifold N . Fix a point x ∈ M and choose any local orthonormal frame {e1 , · · · , en+p } such that {e1 , · · · , en } is an orthonormal basis of the tangent space Tx M and {en+1 , · · · , en+p } is an orthonormal basis of the normal space Nx M . For each α ∈ {n + 1, · · · , n + p}, deﬁne a linear map Aα : Tx M → Tx M by n+p

¯ X Y, eα , Aα X, Y = ∇

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1597

¯ denotes the Levi-Civita connection on N . Then the (unwhere X, Y are tangent vector ﬁelds and ∇ normalized) mean curvature vector H is deﬁned by n+p

H=

(trace Aα )eα .

α=n+1

Deﬁne a linear map φα : Tx M → Tx M by φα X, Y = Aα X, Y − X, Y H, eα and a traceless bilinear map φ : Tx M × Tx M → Nx M by φ(X, Y ) =

n+p

φα X, Y eα .

α=n+1

This map φ is called the traceless second fundamental form of M . Denote by A the second fundamental form. Then |φ|2 = |A|2 −

H2 . n

Note that H2 ≤ n. |A|2 In particular, if p = 1, then φ=A−

H g, n

where g is the induced metric on M . Lemma 2.3. Let b :=

√ (n−2)2 n−1 √ . 2n( n−1+1)2

Then we have

2(n − 1)|H|2 − (n − 2)

√

n − 1|H|

n|A|2 − |H|2 ≥ −bn2 |A|2 .

(2.2)

Proof. If |A| = 0, then H = 0. Thus the inequality (2.2) is trivial. Now we assume that |A| > 0. The inequality (2.2) is equivalent to √ H2 2(n − 1) H 2 (n − 2) n − 1 |H| n − − ≤ b. n2 |A| |A|2 n2 |A|2 √ We deﬁne fn (t) on [0, n ] by √ (n − 2) n − 1 2(n − 1) 2 fn (t) = t n − t2 − t . n2 n2 Suppose that there is a positive constant B such that B ≥ max[0,√n ] fn (t). Then √ (n − 2) n − 1t n − t2 ≤ 2(n − 1)t2 + Bn2 ,

∀t ∈ [0,

√

n]

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1598

or equivalently, (n − 2)2 (n − 1)x(n − x) ≤ 4(n − 1)2 x2 + 4B(n − 1)n2 x + B 2 n4 ,

(2.3)

where x := t2 for all x ∈ [0, n]. A simple computation shows that the inequality (2.3) holds true if √ (n − 2)2 n − 1 √ B≥ = b, 2n( n − 1 + 1)2 which gives the conclusion. 2 In the following, we need the Ricci curvature estimate for submanifolds in a Riemannian manifold which was done by Leung [17]. Lemma 2.4. (See [17].) Let M be an n-dimensional submanifold in a Riemannian manifold N with sectional curvature KN satisfying that K ≤ KN where K is a constant. Then the Ricci curvature RicM of M satisﬁes RicM ≥ (n − 1)K +

√

n−1 2 1 2 |A| . 2(n − 1)|H| − (n − 2) n − 1|H| n|A|2 − |H|2 − n2 n

Using Lemma 2.3 and the above Ricci curvature estimate, one can obtain the following. Lemma 2.5. Let M n be a complete orientable noncompact hypersurface in N of nonnegative sectional curvature. Then √ RicM ≥ −

n−1 2 |A| . 2

(2.4)

Proof. By Lemma 2.3 and Lemma 2.4, we see RicM

√ (n − 2)2 n − 1 n−1 √ |A|2 + ≥− n 2n( n − 1 + 1)2 √ n−1 2 |A| , =− 2

which completes the proof. 2 Theorem 2.6. Let M n (2 ≤ n ≤ 6) be a complete orientable noncompact hypersurface in a complete manifold n−2 N with nonnegative sectional curvature. If the δ-stability inequality holds on M for some 2 √ ≤ δ ≤ 1, n−1 2β then there is no nontrivial L harmonic 1-form on M for any constant β satisfying 2δ 2δ n−2 n−2 √ 1− 1− √ <β< √ 1+ 1− √ . 2δ n − 1 2δ n − 1 n−1 n−1 Proof. We will prove Theorem 2.6 by using Fu–Yang’s arguments [10]. Let ω be an L2β harmonic 1-form on M . Then |ω|2β dv < ∞.

Δω = 0 and M

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1599

We shall use the same notation ω for a harmonic 1-form and its dual harmonic vector ﬁeld in an abuse of notation. Bochner’s formula implies Δ|ω|2 = 2 |∇ω|2 + RicM (ω, ω) . On the other hand, it follows that

2

Δ|ω|2 = 2 |ω|Δ|ω| + ∇|ω| . Thus we obtain

2

|ω|Δ|ω| − RicM (ω, ω) = |∇ω|2 − ∇|ω| . Since the Kato-type inequality

2 |∇ω|2 − ∇|ω| ≥

2 1

∇|ω| n−1

holds for any harmonic 1-form ω on an n-dimensional Riemannian manifold [29], |ω|Δ|ω| ≥

2 1

∇|ω| + RicM (ω, ω). n−1

(2.5)

Applying Lemma 2.5, we get

2 1

∇|ω| − |ω|Δ|ω| ≥ n−1

√

n−1 2 2 |A| |ω| . 2

Given any α > 0, we have

2

|ω|α Δ|ω|α = |ω|α α(α − 1)|ω|α−2 ∇|ω| + α|ω|α−1 Δ|ω|

2

α − 1

= ∇|ω|α + α|ω|2α−2 ω|Δ|ω α √

α − 1

1

n−1 2 2 α 2 2α−2 2 ≥ ∇|ω| ∇|ω| − |A| |ω| + α|ω| α n−1 2 √

n−2 n − 1 2 2α α 2

∇|ω| |A| |ω| . −α ≥ 1− (n − 1)α 2

(2.6)

Choose any nonnegative number q and a smooth function φ with compactly support in M . Multiplying both sides of the inequality (2.6) by |ω|2qα φ2 and integrating over M , we obtain

n−2 1− (n − 1)α

≤

|ω|

M

(2q+1)α 2

M

=α

√

−2 M

√ α

φ Δ|ω| + α

n−1 2

2 |ω|2qα φ2 ∇|ω|α n−1 2

|A|2 φ2 |ω|2(q+1)α M

|A|2 φ2 |ω|2(q+1)α − (2q + 1) M

φ|ω|(2q+1)α ∇φ, ∇|ω|α .

M

2 |ω|2qα ∇|ω|α φ2

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1600

Hence 2(q + 1) − √ ≤α

n−2 (n − 1)α

n−1 2

2 |ω|2qα ∇|ω|α φ2

M

φ|ω|(2q+1)α ∇φ, ∇|ω|α .

|ω|2(q+1)α |A|2 φ2 − 2 M

(2.7)

M

On the other hand, since M satisﬁes the δ-stability inequality and N has nonnegative sectional curvature, we have

|∇φ| ≥ δ 2

M

Ric(ν, ν) + |A| φ ≥ δ 2

|A|2 φ2 .

2

M

M

Replacing φ by |ω|(1+q)α φ in the above inequality gives

2 |ω|2qα ∇|ω|α φ2 +

|ω|2(q+1)α |A|2 φ2 ≤ (q + 1)2

δ M

M

+ 2(1 + q)

|ω|2(q+1)α |∇φ|2 M

|ω|(2q+1)α φ ∇φ, ∇|ω|α .

(2.8)

M

Combining (2.7) and (2.8), we obtain 2(q + 1) −

≤

α δ

√

2α δ

√

2 |ω|2qα ∇|ω|α φ2

M

n−1 (q + 1)2 2

+

n−2 (n − 1)α

M

n−1 (q + 1) − 2 2

2 |ω|2qα ∇|ω|α φ2 +

|ω|2(q+1)α |∇φ|2 M

|ω|(2q+1)α φ ∇φ, ∇|ω|α .

(2.9)

M

Given ε > 0, the Schwarz inequality implies

2α δ

√

|ω|(2q+1)α φ ∇φ, ∇|ω|α

n−1 (q + 1) − 2 2

M

√

α n−1

(q + 1)

2|ω|(2q+1)α φ|∇φ| ∇|ω|α ≤ 1 − δ 2 M

2 1 ≤ |D| ε |ω|2qα ∇|ω|α φ2 + |ω|2(q+1)α |∇φ|2 , ε M

M

where α D := 1 − δ

√

n−1 (q + 1). 2

(2.10)

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1601

From the inequalities (2.9) and (2.10), it follows that 2(q + 1) − √ ≤

n−2 − (n − 1)α

n − 1 α(q + 1)2 2 δ

2 |ω|2qα ∇|ω|α φ2

M

n−1α 2 δ

√

|ω|2(q+1)α |∇φ|2 M

+ |D|ε

2qα

∇|ω|

|ω|

|D| φ + ε

α 2 2

M

|ω|2(q+1)α |∇φ|2 , M

or equivalently, 2(q + 1) − √ ≤

n−2 − (n − 1)α

n − 1 α |D| + 2 δ ε

√

n − 1 α(q + 1)2 − |D|ε 2 δ

2 |ω|2qα ∇|ω|α φ2

M

|ω|2(q+1)α |∇φ|2 .

(2.11)

M

Now let β := (1 + q)α and choose the numbers α and q such that n−2 − 2(q + 1) − (n − 1)α

√

n − 1 α(q + 1)2 > 0. 2 δ

Therefore, for a suﬃciently small ε > 0, the inequality (2.11) implies that there is a constant C > 0 which depends on ε, δ, q, α such that

2 |ω|2qα ∇|ω|α φ2 ≤ C

M

|ω|2β |∇φ|2 ,

(2.12)

M

provided that n−2 − 2(q + 1) − (n − 1)α

√

n − 1 α(q + 1)2 > 0. 2 δ

(2.13)

Note that the inequality (2.13) is equivalent to n−2 − 2β − n−1

√ n − 1 β2 > 0, 2 δ

which is satisﬁed by the assumption 2δ n−2 n−2 2δ √ 1− 1− √ <β< √ 1+ 1− √ . n−1 n−1 2δ n − 1 2δ n − 1 Given R > 0 and a ﬁxed point p ∈ M , we take a test function φ(r) deﬁned on [0, ∞) such that φ ≥ 0, φ = 1 on [0, R] and φ = 0 in [2R, ∞) with |φ | ≤ R2 , where r(x) denotes the distance from p to x on M . Then the inequality (2.12) becomes M

2 4C |ω|2qα ∇|ω|α ≤ 2 R

|ω|2β . M

1602

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

Letting R → ∞, we conclude that |ω| is constant since ω is an L2β harmonic 1-form. However, since the volume of M is inﬁnite by Lemma 2.2, we obtain that ω ≡ 0, which completes the proof. 2 As a direct consequence of Theorem 2.6, if δ = 1, that is, M satisﬁes the stability inequality (1.1), we have the following. Corollary 2.7. For 2 ≤ n ≤ 6, let M n be a complete orientable noncompact hypersurface in a complete manifold N with nonnegative sectional curvature. Let α1 and b be the same constants as in Theorem 2.6. If the stability inequality (1.1) holds on M , then there is no nontrivial L2β harmonic 1-form on M for any constant β satisfying 2 2 n−2 n−2 √ 1− 1− √ <β< √ 1+ 1− √ . n−1 2 n−1 n−1 2 n−1 √

Moreover, if we take (n−1)2n n−1 < δ ≤ 1 in Theorem 2.6, we have the following vanishing theorem for L2 harmonic 1-forms on such hypersurfaces M . Corollary 2.8. For 2 ≤ n ≤ 6, let M n be a complete orientable noncompact hypersurface in a complete √ manifold N with nonnegative sectional curvature. If M satisﬁes δ-stability inequality for some (n−1)2n n−1 < δ ≤ 1, then there is no nontrivial L2 harmonic 1-form on M . Proof. Taking β = 1 in Theorem 2.6 and using our assumption that

√ (n−1) n−1 2n

< δ ≤ 1, we see

2δ n−2 n−2 2δ √ 1− 1− √ <1< √ 1+ 1− √ . n−1 2δ n − 1 n−1 2δ n − 1 Applying Theorem 2.6 gives the conclusion. 2 We remark that Corollary 2.7 and Corollary 2.8 can be regarded as extensions of the main result in [16]. For higher codimensional cases, we deﬁne the δ-super stability inequality as follows. Deﬁnition 2.9. Let M n be an n-dimensional orientable submanifold in the (n + p)-dimensional Euclidean space Rn+p . We say the δ-super stability inequality holds on M for 0 < δ ≤ 1 if any f ∈ C0∞ (M ) satisﬁes |∇f |2 − δ|A|2 f 2 dv ≥ 0. M

The deﬁnition of super stability in the Euclidean space was introduced by Q. Wang [30]. In particular, when p = 1 and δ = 1, the concept of super stability is the same as the usual deﬁnition of stability. Using the same arguments as in the proof of Theorem 2.6, we obtain the following. n+p Corollary 2.10. Let M n (2 ≤ n ≤ 6) be a complete . If the δ-super √ orientable noncompact submanifold in R (n−1) n−1 2 stability inequality holds on M for some < δ ≤ 1, then there is no nontrivial L harmonic 1-form 2n on M .

3. Suﬃcient conditions for complete hypersurfaces to satisfy the δ-stability inequality Given a complete noncompact Riemannian manifold M , we consider a compact domain Ω ⊂ M . Let λ1 (Ω) > 0 denote the ﬁrst eigenvalue of the Dirichlet boundary value problem

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1603

Δf + λf = 0 in Ω f =0 on ∂Ω

where Δ denotes the Laplace operator on M . By the domain monotonicity principle, the ﬁrst eigenvalue λ1 (M ) of a complete noncompact manifold M is deﬁned by λ1 (M ) = inf λ1 (Ω), Ω

where the inﬁmum is taken over all compact domains in M . In this section, we provide suﬃcient conditions for an n-dimensional complete hypersurface to satisfy the δ-stability inequality. More precisely, we prove that if the Ln or L∞ norm of the second fundamental form is suﬃciently small in terms for an n-dimensional complete hypersurface M in a Riemannian manifold with nonpositive sectional curvature, then the δ-stability inequality holds on M . Theorem 3.1. Let M be an n-dimensional complete hypersurface in a Riemannian manifold N with sectional curvature KN satisfying that KN ≤ K ≤ 0 where K ≤ 0 is a constant. Assume that for 0 < δ ≤ 1 |A|2 ≤

λ1 (M ) − nK. δ

Then the δ-stability inequality holds on M . Proof. The assumption on the sectional curvature on N implies that Ric(ν, ν) ≤ nK. Since for any f ∈ C0∞ (M ) λ1 (M ) ≤

|∇f |2 , f2 M

M

it follows from the assumption on |A|2 that

|∇f |2 − δ |A|2 + Ric(ν, ν) f 2 ≥

M

λ1 (M ) − δ|A|2 + nK f 2 ≥ 0.

M

Therefore the δ-stability inequality on M holds for any 0 < δ ≤ 1. 2 It is well known that the following Sobolev inequality holds on a complete submanifold in a complete simply-connected manifold with nonpositive sectional curvature. Lemma 3.2. (See [13].) Let M n (n ≥ 3) be an n-dimensional complete submanifold in a complete simplyconnected manifold with nonpositive sectional curvature. Then for any φ ∈ W01,2 (M ) we have |φ| M

2n n−2

n−2 n dv

≤ CS

2 |∇φ|2 + φ|H| dv,

(3.1)

M

where CS is the Sobolev constant which depends only on n ≥ 3. Using the above Sobolev inequality (3.1), one obtains another suﬃcient condition for a complete hypersurface in a complete simply connected Riemannian manifold to have the δ-stability inequality.

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1604

Theorem 3.3. Let M n (n ≥ 3) be an n-dimensional complete hypersurface in a complete simply connected Riemannian manifold N with nonpositive sectional curvature. Assume that for 0 < δ ≤ 1 |A|

n

n2

≤

1 , (n + δ)CS

M

where CS is a Sobolev constant in (3.1). Then the δ-stability inequality holds on M . Proof. From the Sobolev inequality (3.1) and Hölder inequality, it follows that for any f ∈ C0∞ (M ), |f |

2n n−2

n−2 n

≤ CS

M

2 |∇f |2 + f |H|

M

≤ CS

|∇f | + CS

|H|

2

M

n

n2

M

|f |

2n n−2

n−2 n .

M

Thus |f |

2n n−2

n−2 n

M

CS ≤ 1 − CS H2n

|∇f |2 , M

since H2n ≤ nA2n ≤

n 1 < (n + δ)CS CS

by our assumption. Therefore

1 − CS H2n |∇f | − δ |A| + Ric(ν, ν) f ≥ CS 2

2

|f |

2

M

2n n−2

n−2 n

−δ

M

≥

|A|2 f 2 M

1 − CS H2n − δA2n CS

|f |

2n n−2

n−2 n

M

≥

1 − nCS A2n − δA2n CS

2n

|f | n−2

n−2 n

M

≥

1 − (n + δ)A2n CS

2n

|f | n−2

n−2 n

M

≥ 0, which gives the conclusion. 2 4. Vanishing theorem for L2 harmonic 1-forms on complete noncompact submanifolds As mentioned in the introduction, there are several vanishing theorems for L2 harmonic forms on complete noncompact stable minimal hypersurfaces. Recall that ω is an L2 harmonic 1-form on M if it satisﬁes Δω = −(dδ + δd)ω = 0 and |ω|2 < ∞. M

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1605

Even without assuming stability of minimal hypersurfaces, if M ⊂ Rn+1 is a complete minimal hypersurface with suﬃciently small quantity of M |A|n , then we still have a vanishing theorem for L2 harmonic 1-forms on M . (See [21,25,32] for details.) The analogue of this result is also true for a complete minimal hypersurface in hyperbolic space [24]. Later, it turned out that these gap theorems hold for more general submanifolds. Given an n-dimensional complete noncompact submanifold in Euclidean space, Carron [1] proved that there exists a constant c(n) such that if |A| ≤ c(n), then all spaces of L2 harmonic forms are trivial. Fu and Li [9] showed that for a complete noncompact submanifold M n ⊂ RN there also exists a constant d(n) such that if the Ln norm of the traceless second fundamental form φ is less than d(n) then there is no nontrivial L2 harmonic 1-form on M . More generally, let M be an n-dimensional complete noncompact submanifold in a complete simplyconnected Riemannian manifold N with sectional curvature KN satisfying −k2 ≤ KN ≤ 0 for some constant k. In [2], Cavalcante, Mirandola and Vitório recently proved that if the Ln norm of the traceless second fundamental form φ is suﬃciently small and if the ﬁrst eigenvalue λ1 (M ) of the Laplacian satisﬁes λ1 (M ) >

(n − 1)2 2 k − inf |H|2 , n

then there is no nontrivial L2 harmonic 1-form on M . Note that the lower bound of λ1 (M ) depends on inf |H| in their result. In this section, we prove a similar vanishing theorem for L2 harmonic 1-forms on complete noncompact submanifolds under the same assumptions as in [2] except that the lower bound of λ1 (M ) depends on φn . More precisely, we prove Theorem 4.1. Let M n (n ≥ 3) be an n-dimensional complete noncompact submanifold in a complete simplyconnected Riemannian manifold N with sectional curvature KN satisfying that −k2 ≤ KN ≤ 0 where k is a constant. Assume that the traceless second fundamental form φ satisﬁes φn <

1 . n(n − 1)CS

In the case k = 0, assume further that the ﬁrst eigenvalue λ1 (M ) of M satisﬁes λ1 (M ) >

2n2 (n − 1)2 k2 , n3 − (n − 2)(n − 1) n(n − 1)CS φn − 2n(n − 1)CS φ2n

where CS is a Sobolev constant in (3.1). Then there is no nontrivial L2 harmonic 1-form on M . Proof. As in the proof of Theorem 2.6, it follows from the inequality (2.5) that for any harmonic 1-form ω |ω|Δ|ω| ≥

2 1

∇|ω| + RicM (ω, ω). n−1

Moreover, using Lemma 2.4 and the fact that |φ|2 = |A|2 −

|H|2 n ,

we see

n − 2 n−1 2 |H|2 2 |φ| . Ric ≥ (n − 1) −k − n(n − 1)|φ||H| − 2 2 n n n

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1606

Combining these two inequalities, we have

2 1

|H|2 2

|ω|Δ|ω| ≥ − k |ω|2 ∇|ω| + (n − 1) n−1 n2 n − 2 n−1 2 2 − n(n − 1)|φ||H||ω|2 − |φ| |ω| . n2 n

(4.1)

Consider a geodesic ball B(R) of radius R centered at x ∈ M . Choose a test function f satisfying 0 ≤ f ≤ 1, f ≡ 1 on B(R), f ≡ 0 on M \ B(2R), and |∇f | ≤ R1 . Multiplying both sides by a compactly supported function f 2 in BR ⊂ M and integrating over BR , we have

BR

2 1

∇|ω| f 2 + n−1

f 2 |ω|Δ|ω| ≥ BR

|H|2 2 (n − 1) − k |ω|2 f 2 n2

BR

n − 2 n(n − 1)|φ||H||ω|2 f 2 − n2

− BR

n−1 2 2 2 |φ| |ω| f . n

BR

Applying the divergence theorem, we have

f |ω| ∇f, ∇|ω| −

0 ≤ −2 BR

+

n − 2 n(n − 1) n2

n n−1

∇|ω| 2 f 2 + n − 1 n

BR

|φ||H||ω|2 f 2 + (n − 1)

BR

|φ|2 |ω|2 f 2 BR

k2 −

|H|2 |ω|2 f 2 . n2

BR

For any positive numbers α, β > 0, it follows from the Schwarz inequality

2 1

2

2 f |ω| ∇f, ∇|ω| ≤ α f ∇|ω| + |∇f |2 |ω|2 α BR

BR

(4.2)

BR

and

|φ||H||ω| f ≤ β 2 2

2 BR

1 |H| |ω| f + β 2

|φ|2 |ω|2 f 2 .

2 2

BR

BR

Combining these inequalities, we obtain 0≤

α− + +

n n−1

2 1 f 2 ∇|ω| + α

BR

|∇f |2 |ω|2 BR

β(n − 2) n−1 n(n − 1) − 2 2n n2 n − 2 n−1 n(n − 1) + 2βn2 n

BR

|H|2 |ω|2 f 2 BR

|φ|2 |ω|2 f 2 BR

+ k2 (n − 1)

|ω|2 f 2 .

(4.3)

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1607

Similarly, we have the following estimate:

|φ| |ω| f ≤ 2

|φ|

2 2

n

BR

n2

BR

2n |ω||f | n−2

n−2 n

BR

≤ CS

|φ|n

n2

BR

∇ |ω|f 2 +

BR

|φ|n

≤ CS

|H|2 |ω|2 f 2

BR

n2

(1 + α)

BR

2

1 f 2 ∇|ω| + 1 + |∇f |2 |ω|2 α

BR

|φ|n

+ CS

n2

BR

BR

|H|2 |ω|2 f 2 .

(4.4)

BR

In the case k = 0, we need an additional estimate. Using the monotonicity of the ﬁrst eigenvalue λ1 (BR ) of a ball BR , we observe that |∇f |2 λ1 (M ) ≤ λ1 (BR ) ≤ BR (4.5) f2 BR for any f ∈ C0∞ (M ). Putting |ω|f for f in the inequality (4.5) and using the Schwarz inequality (4.2) gives

2 1

f ∇|ω| + 1 + |∇f |2 |ω|2 . α

2

|ω| f ≤ (1 + α) 2 2

λ1 (M ) BR

BR

(4.6)

BR

Combining the inequalities (4.3), (4.4) and (4.6), we have A

2 f 2 ∇|ω| + B

BR

|H|2 |ω|2 f 2 ≤ C

BR

|∇f |2 |ω|2 ,

(4.7)

BR

where the constants A, B, C are deﬁned by 2 n2 k (n − 1) n − 2 n n−1 n −α− + CS (1 + α) A= |φ| n(n − 1) + n−1 λ1 (M ) 2βn2 n BR

B=

n − 1 n − 2 − n(n − 1)β − CS n2 2n2

|φ|n

n2

n−1 n − 2 n(n − 1) + 2βn2 n

BR

1 k2 (n − 1) + CS C= + α λ1 (M )

|φ|

n

n2

n−1 n − 2 n(n − 1) + 2βn2 n

1 1+ . α

BR

Using our assumption on φn and the following arithmetic–geometric mean inequality β + CS φ2n

1 ≥ 2 CS φn , β

(4.8)

√ we see that B > 0 for any β > 0. Take β = CS φn which makes equality in the inequality (4.8). By the assumptions on λ1 (M ) and φn , we can choose the number α > 0 small enough such that A > 0. Furthermore, it automatically follows that C > 0.

1608

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

Now letting R → ∞ in the inequality (4.7), we obtain |∇|ω|| ≡ 0 and |H||ω| ≡ 0. Since |∇|ω|| ≡ 0, we get |ω| ≡ constant. Suppose that |ω| is a nonzero constant. From the fact that |H||ω| ≡ 0, it follows that M is a minimal submanifold. However, since the volume of a complete minimal submanifold in a Riemannian manifold of nonpositive sectional curvature is inﬁnite, we have M |ω|2 = ∞. This is a contradiction to the assumption that ω is an L2 harmonic 1-form. Therefore ω ≡ 0, which completes the proof. 2 As a consequence, if the ambient space N is the Euclidean space, we obtain the following result. Corollary 4.2. Let M n (n ≥ 3) be an n-dimensional complete noncompact submanifold in the Euclidean space RN . If the traceless second fundamental form φ satisﬁes 1 , n(n − 1)CS

φn <

then there is no nontrivial L2 harmonic 1-form on M . It immediately follows from the above result that such M must have only one end. We remark that the upper bound of φn is less than the upper bound in Corollary 1.1 of [2], which is nonetheless a generalization of [9] and [18]. Moreover, when the ambient space N has a pinched nonpositive sectional curvature, we immediately obtain the following. Corollary 4.3. Let M n (n ≥ 3) be an n-dimensional complete noncompact submanifold in a complete simplyconnected Riemannian manifold N with sectional curvature KN satisfying that −k2 ≤ KN ≤ 0 for some constant k = 0. If φn <

1 n(n − 1)CS

and λ1 (M ) >

2n2 (n − 1)2 k2 , n3 − n2 + 3n − 4

then there is no nontrivial L2 harmonic 1-form on M . We remark that the upper bound of φn and the lower bound of λ1 (M ) depend only on the dimension of M and the curvature of the ambient space, which is diﬀerent from [2]. Acknowledgments The authors would like to thank the referee for the helpful comments and suggestions. The ﬁrst author was supported in part by NAFOSTED under grant number 101.02-2014.49. A part of this paper was written during a stay of the ﬁrst author at Vietnam Institute for Advanced Study in Mathematics (VIASM). He would like to express his sincerely thanks to staﬀs there for excellent working conditions and ﬁnancial support. The second author was supported in part by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT and Future Planning (NRF-2013R1A1A1A05006277).

N.T. Dung, K. Seo / J. Math. Anal. Appl. 423 (2015) 1594–1609

1609

References [1] G. Carron, L2 -cohomologie et inegalites de Sobolev, Math. Ann. 314 (4) (1999) 613–639. [2] M.P. Cavalcante, H. Mirandola, F. Vitório, L2 harmonic 1-forms on submanifolds with ﬁnite total curvature, J. Geom. Anal. 24 (1) (2014) 205–222. [3] T.H. Colding, W.P. Minicozzi II, The space of embedded minimal surfaces of ﬁxed genus in a 3-manifold. II. Multi-valued graphs in disks, Ann. of Math. 160 (2004) 69–92. [4] M. do Carmo, C.K. Peng, Stable complete minimal surfaces in R3 are planes, Bull. Amer. Math. Soc. (N.S.) 1 (1979) 903–906. [5] M. do Carmo, C.K. Peng, Stable complete minimal hypersurfaces, in: Proc. Beijing Symp. Diﬀerential Equations and Diﬀerential Geometry, vol. 3, 1980, pp. 1349–1358. [6] N.T. Dung, K. Seo, Stable minimal hypersurfaces in a Riemannian manifold with pinched negative sectional curvature, Ann. Global Anal. Geom. 41 (4) (2012) 447–460. [7] D. Fischer-Colbrie, R. Schoen, The structure of complete stable minimal surfaces in 3-manifolds of nonnegative scalar curvature, Comm. Pure Appl. Math. 33 (1980) 199–211. [8] H.P. Fu, The structure of δ-stable minimal hypersurface in Rn+1 , Hokkaido Math. J. 40 (2011) 103–110. [9] H.P. Fu, Z.Q. Li, L2 harmonic 1-forms on complete submanifolds in Euclidean space, Kodai Math. J. 32 (3) (2009) 432–441. [10] H.P. Fu, D.Y. Yang, Vanishing theorems on complete manifolds with weighted Poincaré inequality and applications, Nagoya Math. J. 206 (2012) 25–37. [11] A. Grigor’yan, On the existence of a Green function on a manifold, Uspekhi Mat. Nauk 38 (1) (1983) 161–162 (in Russian); Engl. transl. in Russian Math. Surveys 38 (1) (1983) 190–191. [12] A. Grigor’yan, On the existence of positive fundamental solution of the Laplace equation on Riemannian manifolds, Mat. Sb. 128 (3) (1985) 354–363 (in Russian); Engl. transl. in Math. USSR Sb. 56 (1987) 349–358. [13] D. Hoﬀman, J. Spruck, Sobolev and isoperimetric inequalities for Riemannian submanifolds, Comm. Pure Appl. Math. 27 (1974) 715–727. [14] L. Karp, Subharmonic functions, harmonic mappings and isometric immersions, in: S.T. Yau (Ed.), Seminar on Diﬀerential Geometry, in: Ann. of Math. Stud., vol. 102, Princeton University Press, Princeton, 1982. [15] S. Kawai, Operator Δ − aK on surfaces, Hokkaido Math. J. 17 (2) (1988) 147–150. [16] J.J. Kim, G. Yun, On the structure of complete hypersurfaces in a Riemannian manifold of nonnegative curvature and L2 harmonic forms, Arch. Math. (Basel) 100 (2013) 369–380. [17] P.F. Leung, An estimate on the Ricci curvature of a submanifold and some applications, Proc. Amer. Math. Soc. 114 (1992) 1051–1063. [18] Y.W. Li, X.W. Xu, J.R. Zhou, The complete hypersurfaces with zero scalar curvature in Rn+1 , Ann. Global Anal. Geom. 44 (4) (2013) 401–416, http://dx.doi.org/10.1007/s10455-013-9373-1. [19] W.H. Meeks III, J. Perez, A. Ros, Liouville-type properties for embedded minimal surfaces, Comm. Anal. Geom. 14 (4) (2006) 703–723. [20] R. Miyaoka, L2 harmonic 1-forms on a complete stable minimal hypersurface, Geom. Global Anal. (1993) 289–293. [21] L. Ni, Gap theorems for minimal submanifolds in Rn+1 , Comm. Anal. Geom. 9 (3) (2001) 641–656. [22] B. Palmer, Stability of minimal hypersurfaces, Comment. Math. Helv. 66 (1991) 185–188. [23] K. Seo, Minimal submanifolds with small total scalar curvature in Euclidean space, Kodai Math. J. 31 (1) (2008) 113–119. [24] K. Seo, Rigidity of minimal submanifolds in hyperbolic space, Arch. Math. (Basel) 94 (2) (2010) 173–181. [25] K. Seo, L2 harmonic 1-forms on minimal submanifolds in hyperbolic space, J. Math. Anal. Appl. 371 (2) (2010) 546–551. [26] Y. Shen, X. Zhu, On stable complete minimal hypersurfaces in Rn+1 , Amer. J. Math. 120 (1998) 103–116. [27] L.F. Tam, D. Zhou, Stability properties for the higher dimensional catenoid in Rn+1 , Proc. Amer. Math. Soc. 137 (10) (2009) 3451–3461. [28] N.T. Varopoulos, Potential theory and diﬀusion of Riemannian manifolds, in: Conference on Harmonic Analysis in Honor of Antoni Zygmund, vols. I, II, in: Wadsworth Math. Ser., Wadsworth, Belmont, CA, 1983, pp. 821–837. [29] X. Wang, On conformally compact Einstein manifolds, Math. Res. Lett. 8 (5–6) (2001) 671–688. [30] Q. Wang, On minimal submanifolds in an Euclidean space, Math. Nachr. 261/262 (2003) 176–180. [31] S.T. Yau, Some function-theoretic properties of complete Riemannian manifolds and their applications to geometry, Indiana Univ. Math. J. 25 (1976) 659–670. [32] G. Yun, Total scalar curvature and L2 harmonic 1-forms on a minimal hypersurface in Euclidean space, Geom. Dedicata 89 (2002) 135–141.