Compact hypersurfaces in a unit sphere
Qing-Ming Cheng
Department of Mathematics, Faculty of Science and Engineering, Saga University, Saga 840-8502, Japan ([email protected]) Shichang Shu∗
Department of Mathematics, Weinan Teachers’ College, Weinan 714000, Shaanxi, People’s Republic of China ([email protected])
Young Jin Suh
Department of Mathematics, Kyungpook National University, Taegu 702-701, South Korea ([email protected])
(MS received 31 August 2004; accepted 9 February 2005)
We study curvature structures of compact hypersurfaces in the unit sphereSn+1(1) with two distinct principal curvatures. First of all, we prove that the Riemannian productS1(√
1−c2)×Sn−1(c) is the only compact hypersurface inSn+1(1) with two distinct principal curvatures, one of which is simple and satisfies
r >1−2
n, r=n−2
n−1 and S(n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2, wheren(n−1)ris the scalar curvature of hypersurfaces andc2= (n−2)/nr. This generalized the result of Cheng, where the scalar curvature is constant is assumed.
Secondly, we prove that the Riemannian productS1(√
1−c2)×Sn−1(c) is the only compact hypersurface with non-zero mean curvature inSn+1(1) with two distinct principal curvatures, one of which is simple and satisfies
r >1−2
n and S(n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2. This gives a partial answer for the problem proposed by Cheng.
1. Introduction
LetMbe ann-dimensional hypersurface in a unit sphereSn+1(1) of dimensionn+1.
It is well known that a compact minimal hypersurface with S = n in Sn+1(1) is isometric to a Clifford torus
S1
1
n
×Sn−1
n−1
n
,
whereS is the squared norm of the second fundamental form of the hypersurface (cf. [6,7,10]). Furthermore, Otsuki [12] investigated the converse problem. He proved
∗Present address: Department of Mathematics, Faculty of Science and Engineering, Saga Uni- versity, Saga 840-8502, Japan.
1129
c 2005 The Royal Society of Edinburgh
that Riemannian product Sk
k
n
×Sn−k
n−k
n
is the only compact minimal hypersurface with two distinct principal curvatures whose multiplicities are greater than 1. For compact minimal hypersurfaces with two distinct principal curvatures, one of which is simple, Otsuki constructed many examples that are not congruent to each other. However, they have the same topo- logical type asSn−1×S1 when they are compact. Cheng [3] investigated in detail compact minimal hypersurfaces with two distinct principal curvatures, one of which is simple, and proved that the Clifford torus
S1
1
n
×Sn−1
n−1
n
is the only compact minimal hypersurface inSn+1(1) with nS C(n), where C(n) is a constant depending only on n. As a generalization of the result of Cheng [3], Hasanis and Vlachos [9] deleted the condition of the above bounding ofS, that is, they proved that the Clifford torus
S1
1
n
×Sn−1
n−1
n
is the only compact minimal hypersurface in Sn+1(1) with two distinct principal curvatures, one of which is simple, ifSn.
Recently, Cheng [4] studied n-dimensional complete hypersurfaces in Sn+1(1) with constant scalar curvature and proved the following theorem.
Theorem 1.1(Cheng [4]). LetM be ann-dimensional complete hypersurface with constant scalar curvaturen(n−1)rinSn+1(1). IfM has only two distinct principal curvatures, one of which is simple, thenr >1−(2/n)holds andM is isometric to S1(√
1−c2)×Sn−1(c) ifr= (n−2)/(n−1) and S (n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2, wherec2= (n−2)/nr.
Further, he constructed infinitely many isometrically distinct compact hypersur- faces in Sn+1(1) which have constant scalar curvature but are not congruent to each other.
Alternatively, it is natural to ask whether the condition of constant scalar curva- ture in Cheng [4, theorem 1.1] is necessary. In this paper, we shall first of all solve the problem and prove the following theorem.
Theorem 1.2. Let M be ann-dimensional compact hypersurface in Sn+1(1) with two distinct principal curvatures, one of which is simple. Then,M is isometric to S1(√
1−c2)×Sn−1(c) if r >1−2
n, r=n−2
n−1 and S(n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2, wherec2= (n−2)/nr, andn(n−1)r denotes the scalar curvature.
Remark1.3. In our theorem above, we do not know whether the condition r = (n−2)/(n−1) is necessary, which was also assumed in theorem 1.1. From the examples constructed by Cheng [4], we know that the condition
S(n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2 is essential. It is well known that the Clifford torus
S1
1
n
×Sn−1
n−1
n
is a compact minimal hypersurface inSn+1(1) satisfyingr= (n−2)/(n−1).
On the other hand, Li [11] proved the following. Let M be an n-dimensional compact hypersurface with constant scalar curvaturen(n−1)r. Ifr1 and
S (n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2,
thenM is isometric to either the totally umbilical hypersurface or the Riemannian product
S1(
1−c2)×Sn−1(c), c2=n−2
nr n−2 n .
For any 0 < c < 1, by considering the standard immersions Sn−1(c)⊂Rn, S1(√
1−c2)⊂R2 and taking the Riemannian product immersion S1(√
1−c2)× Sn−1(c)→R2×Rn, we obtain a compact hypersurfaceS1(√
1−c2)×Sn−1(c) in Sn+1(1) with constant scalar curvaturen(n−1)r, wherer= (n−2)/nc2>1−2/n.
Hence, some of the Riemannian productsS1(√
1−c2)×Sn−1(c) do not appear in the results in [11].
From theorem 1.1 above (see Cheng [4]), it is natural and interesting to generalize Li’s results [11] to the caser >1−2/n. That is, it is interesting to prove the following problem.
Problem1.4 (Cheng [4]). LetM be ann-dimensional compact hypersurface with constant scalar curvaturen(n−1)rinSn+1(1). If
r >1−2
n and S(n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2, can we classify the hypersurfaceM?
In order to solve this problem, Cheng [5] proved the following theorem.
Theorem 1.5. Let M be an n-dimensional compact hypersurface with an infinite fundamental group inSn+1(1). If
r n−2
n−1 and S(n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2, then M is isometric to the Riemannian product S1(√
1−c2)×Sn−1(c), where n(n−1)ris the scalar curvature ofM andc2= (n−2)/nr.
In the following theorem, we solve the problem 1.4 partially.
Theorem 1.6. Let M be ann-dimensional compact hypersurface in Sn+1(1) with non-zero mean curvature and with two distinct principal curvatures, one of which is simple. Then,M is isometric toS1(√
1−c2)×Sn−1(c)if r >1−2
n and S(n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2, wherec2= (n−2)/nr andn(n−1)rdenotes the scalar curvature.
2. Preliminaries
Throughout this paper, all manifolds are assumed to be smooth and connected with- out boundary. Letϕ:M →Sn+1(1) be an isometric immersion of then-dimensional Riemannian manifoldM into a unit sphereSn+1(1). We choose a local orthonormal frame {e1, . . . ,en,en+1} and the dual co-frame {ω1, . . . , ωn, ωn+1} in such a way that{e1, . . . ,en} is a local orthonormal frame onM. Hence, we have
ωn+1= 0 onM.
From Cartan’s lemma, we have ωn+1,i=
n
j=1
hijωj, hij =hji. (2.1)
The mean curvature H and the second fundamental form α of M are defined, respectively, by
H = 1 n
n
i=1
hii, α=hijωi⊗ωjen+1.
When the mean curvature H of M is identically zero, we recall that M is by definition a minimal hypersurface.
From the structure equations ofM, the Gauss equation and Codazzi equation, respectively, are given by
Rijkl= (δikδjl−δilδjk) + (hikhjl−hilhjk), (2.2)
hijk=hikj. (2.3)
From (2.2), we have
n(n−1)r=n(n−1) +n2H2−S, (2.4) where n(n−1)r and S denote the scalar curvature and the squared norm of the second fundamental form ofM, respectively. We choose a local field of orthonormal framese1, . . . ,en onM such that, at the point that we consider,
hij =
λi ifi=j,
0 ifi=j, (2.5)
where theλi are the principal curvatures ofM.
Next, we consider the parallel hypersurfacesϕθ:M →Sn+1(1) ofM inSn+1(1) given by the map
ϕθ= cosθϕ+ sinθN, (2.6)
whereN is the Gauss map ofϕ:M →Sn+1(1). It is well known that, when cotθ is not a principal curvature ofϕ,ϕθis an isometric immersion if we endowM with the pullback metric gθ via ϕθ. We take cotθ = λi for any i = 1,2, . . . , n. Thus, we know that the shape operatorAθ of the immersionϕθ with respect to the unit normal vectorNθ= cosθN −sinθϕis given by
Aθ={cotθAN+I}{cotθI−AN}−1, (2.7) where AN is the shape operator with respect to the unit normal vector N and Iis the identity mapping. Hence, the principal curvatures ofϕθ:M →Sn+1(1) are given by
λi(θ) =cotθλi+ 1 cotθ−λi
. (2.8)
3. Proofs of theorem 1.2 and theorem 1.6
Before we prove our theorems, we shall review some known results which will be used in the proof.
In [5], Cheng studied compact hypersurfaces with infinite fundamental group in Sn+1(1) and proved the following theorem.
Theorem 3.1. LetM be ann-dimensional compact hypersurface with infinite fun- damental group inSn+1(1). If the sectional curvatures are non-negative, thenM is isometric to the Riemannian productS1(√
1−c2)×Sn−1(c).
The following theorem due to do Carmoet al. [8] will be used.
Theorem 3.2. LetM be ann-dimensional(n3)compact orientable hypersurface inSn+1(1)with two distinct principal curvatures, one of which is simple. Then,M is homeomorphic toS1(√
1−c2)×Sn−1(c).
Remark3.3. We note that, in the statement of theorem 3.2, do Carmo et al.
considered that M is an n-dimensional (n 4) conformal flat hypersurface in Sn+1(1). But, in their proof, they used only the condition that the hypersurface has two distinct principal curvatures, one of which is simple. Hence, their result is also true forn= 3.
Proof of theorem 1.2. Since M has two distinct principal curvatures, we denote them byλandµ. We then have
(n−1)λ+µ=nH, S= (n−1)λ2+µ2. (3.1) From (3.1) and the Gauss equation (2.4), we obtain
λµ= (n−1)(r−1)−(n−2)H2±(n−2)H
H2−(r−1).
Since
S(n−1)n(r−1) + 2
n−2 + n−2
n(r−1) + 2 and r= n−2 n−1,
we know thatH = 0. In fact, ifH = 0 holds at some point, we have, at the point, S=−n(n−1)(r−1). Hence,
−n(n−1)(r−1)(n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2,
−(n−1)(n(r−1) + 2) + 2(n−1)(n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2. Since the scalar curvature is continuous, we have
r > n−2
n−1 or r < n−2
n−1 onM.
Ifr >(n−2)/(n−1) holds, thenn(r−1) + 2>(n−2)/(n−1). Ifr <(n−2)/(n−1) holds, thenn(r−1) + 2<(n−2)/(n−1). We consider a function f(t) defined by
f(t) = (n−1)2
n−2 t+n−2
t −2(n−1).
We can prove that f(t) is an increasing function of t when t > (n−2)/(n−1).
Hence, we infer a contradiction because f
n−2 n−1
= 0.
Ift <(n−2)/(n−1), thenf(t) is a decreasing function. This is also impossible.
Hence, we haveH= 0.
Thus, we can assumeH >0 onM. Therefore,M is orientable. Since the principal curvaturesλandµare continuous, we have
λµ= (n−1)(r−1)−(n−2)H2+ (n−2)H
H2−(r−1) or
λµ= (n−1)(r−1)−(n−2)H2−(n−2)H
H2−(r−1) onM. From the Gauss equation (2.4), we have
λµ=2(n−1)
n (r−1)−(n−2)
n2 S+(n−2) n2
S+n(n−1)(r−1)
S−n(r−1) or
λµ=2(n−1)
n (r−1)−(n−2)
n2 S−(n−2) n2
S+n(n−1)(r−1)
S−n(r−1) onM.
First of all, we consider the case where λµ=2(n−1)
n (r−1)−(n−2)
n2 S+(n−2) n2
S+n(n−1)(r−1)
S−n(r−1) holds onM. In this case, we can proveλµ+ 10 onM. Indeed, ifλµ+ 1<0 at some point, then
(n−2) n2
S+n(n−1)(r−1)
S−n(r−1)<−1−2(n−1)
n (r−1) +(n−2) n2 S.
From
S(n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2
and a direct computation, we have a contradiction. Hence, we haveλµ+ 10 on M. From the Gauss equation (2.2), we infer that the sectional curvatures ofM are non-negative, sinceM has two distinct principal curvatures, one of which is simple.
From theorem 3.2, we find that M is homeomorphic to S1(√
1−c2)×Sn−1(c).
Thus, the fundamental group of M is infinite. From theorem 3.1, we know that ϕ(M) is isometric toS1(√
1−c2)×Sn−1(c).
Secondly, we consider the case where
λµ= 2(n−1)
n (r−1)−(n−2) n2 S
−(n−2) n2
S+n(n−1)(r−1)
S−n(r−1) (3.2) holds onM. From
S (n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2,
we find that λ = 0 and µ = 0. In fact, if λ = 0, we have n2H2 = S. From the Gauss equation (2.4) we obtainr= 1. Hence, from formula (3.2) forλµ, we have a contradiction. Ifµ= 0, we haven2H2= (n−1)S. From the Gauss equation (2.4), we have (n−2)S =n(n−1)(r−1). It is impossible because
S(n−1)n(r−1) + 2
n−2 + n−2
n(r−1) + 2, r >1−2 n. Next we shall prove thatλµ+ 10. If
2(n−1)
n (r−1)−(n−2)
n2 S+ 10,
then, from (3.2), it is obvious that the conditionλµ+ 10 holds. If 2(n−1)
n (r−1)−(n−2)
n2 S+ 1>0,
then we can prove that the condition thatλµ+ 10 also holds. In fact, from S (n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2, we have
(n−2)(n(r−1) + 2)S(n−1)n2(r−1)2+ 4n(n−1)(r−1) +n2, that is,
(n−2){4n(n−1)(r−1) + 2n2+ (n−2)2n(r−1)}S
{2n(n−1)(r−1) +n2}2+ (n−2)2n2(n−1)(r−1)2.
Hence,
n+ 2(n−1)(r−1)−n−2 n S
2
(n−2)2
n2 {n(n−1)(r−1) +S}{S−n(r−1)}. Since the condition
2(n−1)
n (r−1)−(n−2)
n2 S+ 1>0 holds, we obtain
2n(n−1)(r−1) +n2−(n−2)S (n−2)
S+n(n−1)(r−1)
S−n(r−1).
From
λµ=2(n−1)
n (r−1)−(n−2)
n2 S−(n−2) n2
S+n(n−1)(r−1)
S−n(r−1), we may infer thatλµ+ 10. Hence, we obtain (1/λµ) + 10.
Sinceλµ= 0, we know thatϕπ/2is an isometric immersion fromM intoSn+1(1).
Therefore, from (2.8), we know that the sectional curvatures Kπ/2 of ϕπ/2 with respect to the plane spanned byej anden forj = 1, . . . , n−1, satisfy
Kπ/2(ej,en) = 1
λµ+10, Kπ/2(ei,ej) = 1+ 1
λ2 >1 fori, j= 1,2, . . . , n−1.
It is obvious that the immersionϕπ/2has only two distinct principal curvatures and that one of them is simple. From theorem 3.2, we infer thatM is homeomorphic to S1(√
1−c2)×Sn−1(c). Thus, the fundamental group of M is infinite. From theorem 3.1, we know thatϕπ/2(M) is isometric toS1(√
1−c2)×Sn−1(c). Hence, the principal curvatures 1/λand 1/µofϕπ/2are constant. Thus, we see thatλandµ are constant andλµ+1 = 0. Hence, the sectional curvatures ofM are non-negative.
From theorem 3.1, we may infer thatϕ(M) is isometric toS1(√
1−c2)×Sn−1(c).
Proof of theorem 1.6. SinceM has two distinct principal curvatures, one of which is simple, from theorem 3.2 we know that M is homeomorphic toS1(√
1−c2)× Sn−1(c). Hence, the fundamental group ofM is infinite. Since the condition
S(n−1)n(r−1) + 2
n−2 + n−2 n(r−1) + 2 holds if and only if
n+ 2(n−1)(r−1)−n−2 n S
2
(n−2)2
n2 {n(n−1)(r−1) +S}{S−n(r−1)}, from the Gauss equationn(n−1)r=n(n−1) +n2H2−S, we conclude that
(S−S−)(S−S+)0, where
S± =n+ n3
2(n−1)H2± n−2 2(n−1)
n4H4+ 4n2(n−1)H2.
Since the mean curvature H = 0 holds on M, we have S+ > S− on M. Then, sinceSis a continuous function onM, we can infer that only one of the conditions SS+ onM andSS− onM is satisfied.
If the conditionSS− onM holds, from the result of [1], and the fundamental group is infinite, we know thatM is isometric toS1(√
1−c2)×Sn−1(c). If, on the other hand, the conditionSS+ onM holds, from the result of [2], we know that M is isometric toS1(√
1−c2)×Sn−1(c).
Acknowledgments
Q.-M.C. was partly supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science. S.S. was partly supported by the Nat- ural Science Foundation of China and the NSF of Shaanxi. Y.J.S. was partly sup- ported by Grant no. R14-2002-003-01001-0 from Korea Research Foundation, Korea 2005.
References
1 A. C. Asperti and E. A. Costa. Vanishing of homology groups, Ricci estimate for subman- ifolds and applications.K¯odai Math. J.24(2001), 313–328.
2 J. N. Barbosa, A. Brasil and I. Lazaro. Hypersurfaces of the Euclidean sphere with non- negative Ricci curvature.Arch. Math.81(2003), 335–341.
3 Q.-M. Cheng. The rigidity of Clifford torus S1(
1/n)×Sn−1(
(n−1)/n). Comment.
Math. Helv.71(1996), 60–69.
4 Q.-M. Cheng. Hypersurfaces in a unit sphereSn+1(1) with constant scalar curvature.J.
Lond. Math. Soc.64(2001), 755–768.
5 Q.-M. Cheng. Compact hypersurfaces in a unit sphere with infinite fundamental group.
Pac. J. Math.212(2003), 49–56.
6 Q.-M. Cheng and S. Ishikawa. A characterization of the Clifford torus.Proc. Am. Math.
Soc.127(1999), 819–828.
7 S. S. Chern, M. do Carmo and S. Kobayashi. Minimal submanifolds of a sphere with second fundamental form of constant length. InFunctional Analysis and Related Fields, pp. 59–75 (Springer, 1970).
8 M. do Carmo, M. Dajczer and F. Mercuri. Compact conformally flat hypersurfaces.Trans.
Am. Math. Soc.288(1985), 189–203.
9 T. Hasanis and T. Vlachos. A pinching theorem for minimal hypersurfaces in a sphere.
Arch. Math.75(2000), 469–471.
10 H. B. Lawson Jr. Local rigidity theorems for minimal hypersurfaces.Ann. Math.89(1969), 167–179.
11 H. Li. Hypersurfaces with constant scalar curvature in space forms.Math. Ann.305(1996), 665–672.
12 T. Otsuki. Minimal hypersurfaces in a Riemannian manifold of constant curvature.Am. J.
Math.92(1970), 145–173.
(Issued 16 December 2005)