Home Non-hyperbolic P-Invariant Closed Characteristics on Partially Symmetric Compact Convex Hypersurfaces
Article Open Access

Non-hyperbolic P-Invariant Closed Characteristics on Partially Symmetric Compact Convex Hypersurfaces

  • Hui Liu and Gaosheng Zhu EMAIL logo
Published/Copyright: February 14, 2018

Abstract

Let n2 be an integer, P=diag(-In-κ,Iκ,-In-κ,Iκ) for some integer κ[0,n], and let Σ2n be a partially symmetric compact convex hypersurface, i.e., xΣ implies PxΣ, and (r,R)-pinched. In this paper, we prove that when R/r<5/3 and 0κ[n-12], there exist at least E(n-2κ-12)+E(n-2κ-13) non-hyperbolic P-invariant closed characteristics on Σ. In addition, when R/r<3/2, [n+12]κn and Σ carries exactly nP-invariant closed characteristics, then there exist at least 2E(2κ-n-14)+E(n-κ-13) non-hyperbolic P-invariant closed characteristics on Σ, where the function E(a) is defined as E(a)=min{kka} for any a.

MSC 2010: 58E05; 37J45; 34C25

1 Introduction and Main Results

In this paper, we consider the number of non-hyperbolic P-invariant closed characteristics on partially symmetric hypersurfaces in 2n under a pinching condition. Let Σ be a C3 compact hypersurface in 2n, bounding a strictly convex compact set U with non-empty interior, where n2. We denote the set of all such hypersurfaces by (2n). Without loss of generality, we suppose that U contains the origin. We consider closed characteristics (τ,y) on Σ, which are solutions of the following problem:

(1.1) { y ˙ ( t ) = J N Σ ( y ( t ) ) , y ( t ) Σ for all  t , y ( τ ) = y ( 0 ) ,

where

J = ( 0 - I n I n 0 ) ,

I n is the identity matrix in n and NΣ(y) is the outward normal unit vector of Σ at y, normalized by the condition NΣ(y)y=1. Here ab denotes the standard inner product of a,b2n. A closed characteristic (τ,y) is prime if τ is the minimal period of y. Two closed characteristics (τ,x) and (σ,y) are geometrically distinct if x()y(). We denote by 𝒥(Σ) the set of all closed characteristics (τ,y) on Σ with τ being the minimal period of y. For any si,tiki, with i=1,2, we define (s1,t1)(s2,t2)=(s1,s2,t1,t2). Fixing an integer κ, with 0κn, let P=diag(-In-κ,Iκ,-In-κ,Iκ) and κ(2n)={Σ(2n)xΣPxΣ}. For Σκ(2n), let Σ(κ)={z2κ0zΣ}, where 0 is the origin in 2n-2κ. As in [1], a closed characteristic (τ,y) on Σκ(2n) is P-asymmetric if y()Py()=, it is P-symmetric if y()=Py(), with y=y1y2 and y10, and it is P-fixed if y()=Py() and y=0y2, where y12(n-κ), y22κ. We say that a closed characteristic (τ,y) is P-invariant if y()=Py(). Then a P-invariant closed characteristic is P-symmetric or P-fixed.

Let j:2n be the gauge function of Σ, i.e., j(λx)=λ for xΣ and λ0. Then jC3(2n{0},)C0(2n,) and Σ=j-1(1). Fix a constant α(1,+) and define the Hamiltonian Hα:2n[0,+) by

H α ( x ) := j ( x ) α .

Then HαC3(2n{0},)C0(2n,) is convex and Σ=Hα-1(1). It is well known that problem (1.1) is equivalent to the following energy problem of the Hamiltonian system:

(1.2) { y ˙ ( t ) = J H α ( y ( t ) ) , H α ( y ( t ) ) = 1 for all  t , y ( τ ) = y ( 0 ) .

Denote by 𝒥(Σ,α) the set of all solutions (τ,y) of problem (1.2), where τ is the minimal period of y. Note that elements in 𝒥(Σ) and 𝒥(Σ,α) are in one-to-one correspondence with each other. Let (τ,y)𝒥(Σ,α). We call the fundamental solution γy:[0,τ]Sp(2n), with γy(0)=I2n, of the linearized Hamiltonian system

z ˙ ( t ) = J H α ′′ ( y ( t ) ) z ( t ) for all  t

the associated symplectic path of (τ,y). The eigenvalues of γy(τ) are called Floquet multipliers of (τ,y). By [3, Proposition 1.6.13], the Floquet multipliers with their multiplicities and Krein type numbers of (τ,y)𝒥(Σ,α) do not depend on the particular choice of the Hamiltonian function in (1.2). As in [15, Chapter 15], for any symplectic matrix M, we define the elliptic height e(M) of M by the total algebraic multiplicity of all eigenvalues of M on the unit circle 𝕌 in the complex plane . And for any (τ,y)𝒥(Σ,α), we define e(τ,y)=e(γy(τ)) and call (τ,y)elliptic or hyperbolic if e(τ,y)=2n or e(τ,y)=2, respectively.

As in [3, Definition 5.1.6], a C3 hypersurface Σ bounding a compact convex set U containing 0 in its interior is (r,R)-pinched, with 0<rR, if

| y | 2 R - 2 1 2 ( H 2 ′′ ( x ) y , y ) | y | 2 r - 2 for all  x Σ .

For the existence, multiplicity and stability of closed characteristics on convex compact hypersurfaces in 2n we refer to [17, 23, 5, 6, 8, 4, 18, 10, 16, 22, 21] and the references therein. It is very interesting to consider closed characteristics on hypersurfaces with special symmetries. In [20, 12, 24], the multiplicity of closed characteristics with symmetries on convex compact hypersurfaces without pinching conditions was studied. For the stability problem of closed characteristics with symmetries, Hu and Sun [9] studied the index theory and stability of periodic solutions in Hamiltonian systems with symmetries. As an application they studied the stability of figure-eight orbit, due to Chenciner and Montgomery, in planar three-body problems with equal masses. In [11], Liu studied the ellipticity and non-hyperbolicity of symmetric closed characteristics on central symmetric compact convex hypersurfaces under a pinching condition. In [2], Dong and Long proved that there exists at least one P-invariant closed characteristic which possesses at least 2n-4κ Floquet multipliers on the unit circle of the complex plane. In [14], Liu and Zhang obtained two such closed characteristics under a pinching condition. In this paper, we continue to consider the non-hyperbolicity of P-invariant closed characteristic under some pinching conditions.

Theorem 1.1.

Assume ΣHκ(2n) and 0<r|x|R for all xΣ. Then the following hold:

  1. When R / r < 5 / 3 and 0 κ [ n - 1 2 ] , there exist at least E ( n - 2 κ - 1 2 ) + E ( n - 2 κ - 1 3 ) non-hyperbolic P -invariant closed characteristics on Σ.

  2. When R / r < 3 / 2 , [n+12]κn and Σ carries exactly nP-invariant closed characteristics, then there exist at least 2E(2κ-n-14)+E(n-κ-13) non-hyperbolic P-invariant closed characteristics on Σ.

Remark 1.2.

In Theorem 1.1, let κ=0. The P-symmetric closed characteristic is just central symmetric, the P-fixed closed characteristics vanish, and we have E(n-2κ-12)+E(n-2κ-13)=E(n-12)+E(n-13) for κ=0. Thus, Theorem 1.1 generalizes [11, Theorem 1.2].

In this paper, let ,,, and denote the sets of natural integers, integers, rational numbers, real numbers and complex numbers, respectively. Denote by ab and |a| the standard inner product and norm in 2n. We use coefficients for all homological modules. We define the functions [a]=max{kka} and E(a)=min{kka}.

2 A Variational Structure for P-Invariant Closed Characteristics

In the rest of this paper, we fix a Σκ(2n). In this section, we review the variational structure for P-invariant closed characteristics established in [12].

As in [12], we associate with U a convex function Ha. Consider the fixed period problem

(2.1) { x ˙ ( t ) = J H a ( x ( t ) ) , x ( 1 / 2 ) = P x ( 0 ) .

Then, by [12, Proposition 2.2], the nonzero solutions of (2.1) are in one-to-one correspondence with P-symmetric closed characteristics with period τ<a and P-fixed closed characteristics with period τ/2<a/2. Let

L κ 2 ( 0 , 1 / 2 ) = { u = u 1 u 2 L 2 ( ( 0 , 1 / 2 ) , 2 n ) | u 1 L 2 ( ( 0 , 1 / 2 ) , 2 n - 2 κ ) , u 2 L 2 ( ( 0 , 1 / 2 ) , 2 κ ) ,
u ( 1 / 2 ) = P u ( 0 ) , 0 1 / 2 u 2 ( t ) d t = 0 } .

Define a linear operator Πκ:Lκ2(0,1/2)Lκ2(0,1/2) by

( Π κ u ) ( t ) = x 1 ( t ) x 2 ( t ) ,

where

x 1 ( t ) = 0 t u 1 ( τ ) 𝑑 τ - 1 2 0 1 / 2 u 1 ( τ ) 𝑑 τ , x 2 ( t ) = 0 t u 2 ( τ ) 𝑑 τ - 2 0 1 / 2 𝑑 t 0 t u 2 ( τ ) 𝑑 τ

for any u=u1u2Lκ2(0,1/2).

The corresponding Clarke–Ekeland dual action functional is defined by

Ψ a ( u ) = 0 1 / 2 ( 1 2 J u Π κ u + G a ( - J u ) ) 𝑑 t ,

where Ga is the Fenchel transform of Ha defined by Ga(x)=sup{xy-Ha(y)y2n}. By [12, Proposition 2.6], Ψa is C1,1 on Lκ2(0,1/2) and satisfies the Palais–Smale condition. Suppose x is a solution of (2.1). Then u=x˙ is a critical point of Ψa. Conversely, suppose u is a critical point of Ψa. Then there exists a unique ξ2n such that Πκu-ξ is a solution of (2.1). In particular, the solutions of (2.1) are in one-to-one correspondence with the critical points of Ψa. Moreover, Ψa(u)<0 for every critical point u0 of Ψa.

Suppose u is a nonzero critical point of Ψa. Then the formal Hessian of Ψa at u is defined by

Q a ( v , v ) = 0 1 / 2 ( J v Π κ v + G a ′′ ( - J u ) J v J v ) 𝑑 t ,

which defines an orthogonal splitting Lκ2(0,1/2)=E-E0E+ of Lκ2(0,1/2) into negative, zero and positive subspaces. The index of u is defined by i(u)=dimE- and the nullity of u is defined by ν(u)=dimE0, cf. [12, Definition 2.10].

Note that we can identify Lκ2(0,1/2) with the space

{ u L 2 ( / , 2 n ) u | ( 0 , 1 / 2 ) L κ 2 ( 0 , 1 / 2 ) , u ( t + 1 / 2 ) = P u ( t ) } .

Then we have a natural S1-action on Lκ2(0,1/2), defined by θ*u(t)=u(θ+t) for all θS1/ and t. By [12, Lemma 2.8], Ψa is S1-invariant. Hence, if u is a critical point of Ψa, then the whole orbit S1u is formed by critical points of Ψa. Denote by crit(Ψa) the set of critical points of Ψa. Then crit(Ψa) is compact by the Palais–Smale condition.

Recall that for a principal U(1)-bundle EB, the Fadell–Rabinowitz index (cf. [7]) of E is defined to be sup{kc1(E)k-10}, where c1(E)H2(B,) is the first rational Chern class. For a U(1)-space, i.e., a topological space X with a U(1)-action, the Fadell–Rabinowitz index is defined to be the index of the bundle X×SX×U(1)S, where SCP is the universal U(1)-bundle.

As in [3, p. 199], we choose some α(1,2) and associate with U a convex function H(x)=j(x)α for all x2n. Consider the fixed period problem

{ x ˙ ( t ) = J H ( x ( t ) ) , x ( 1 / 2 ) = P x ( 0 ) .

The corresponding Clarke–Ekeland dual action functional on Lκ2(0,1/2) is defined by

Ψ ( u ) = 0 1 / 2 ( 1 2 J u Π κ u + H * ( - J u ) ) 𝑑 t for all  u L κ 2 ( 0 , 1 / 2 ) ,

where H* is the Fenchel transform of H.

For any ι, we define

Ψ ι - = { w L κ 2 ( 0 , 1 / 2 ) Ψ ( w ) < ι } .

As in [12, Section 2], we define

c i = inf { δ I ^ ( Ψ δ - ) i } .

where I^ is the Fadell–Rabinowitz index defined above. Then

c 1 c 2 c i c i + 1 < 0 .

By [12, Propositions 2.15 and 2.16], we have the following propositions.

Proposition 2.1.

Every ci is a critical value of Ψ. If ci=cj for some i<j, then there are infinitely many geometrically distinct P-invariant closed characteristics on Σ.

Proposition 2.2.

For every iN, there exists a critical point uα of Ψ, found in Proposition 2.1, such that

Ψ ( u α ) = c i , C S 1 , 2 i - 2 ( Ψ a , S 1 u ) 0 ,

where u is a critical point of Ψa corresponding to uα in the natural sense. In particular, we have i(u)2(i-1)i(u)+ν(u)-1.

3 Index Theory for P-Invariant Closed Characteristics

In this section, we review the index theory for P-invariant closed characteristics which was studied in [12, Section 3].

Note that if (τ,y) is P-symmetric, then ((2m-1)τ,y) is P-symmetric for any m. Thus, ((2m-1)τ,y) corresponds to a critical point of Ψa via [12, Propositions 2.2 and 2.6]; we denote it by u2m-1. If (τ2,y) is P-fixed, then (mτ/2,y) is P-fixed for any m. Thus, (mτ/2,y) corresponds to a critical point of Ψa; we denote it by um. Recall that the action of a closed characteristic (τ,y) is defined by (cf. [3, p. 190])

A ( τ , y ) = 1 2 0 τ ( J y y ˙ ) 𝑑 t .

Now we consider the linear Hamiltonian system

(3.1) { ξ ˙ ( t ) = J A ( t ) ξ , A ( t + τ 2 / 2 ) = P A ( t ) P ,

where A(t)=H2′′(y(t)) and τ2=A(τ,y). Denote by iPE(A,k) and νPE(A,k) the index and nullity of the k-th iteration of system (3.1) defined by Dong and Long (cf. [2, Definition 3.4]), respectively. Denote by iP,1(γAk,P) and νP,1(γAk,P) the P-index and P-nullity of the k-th iteration of system (3.1) defined by Dong and Long (cf. [1, Section 3]), respectively, where γA is the fundamental solution of (3.1), with γA(0)=I2n. Then we have the following theorem.

Theorem 3.1 (cf. [12, Theorem 3.2]).

  1. If u m is a nonzero critical point of Ψ a such that u corresponds to the P -fixed closed characteristic ( τ / 2 , y ) , then we have

    i ( u m ) = i P , 1 ( γ A m , P ) - κ , ν ( u m ) = ν P , 1 ( γ A m , P ) - 1 .

  2. If u 2 m - 1 is a nonzero critical point of Ψ a such that u corresponds to the P -symmetric closed characteristic ( τ , y ) , then we have

    i ( u 2 m - 1 ) = i P , 1 ( γ A 2 m - 1 , P ) - κ , ν ( u 2 m - 1 ) = ν P , 1 ( γ A 2 m - 1 , P ) - 1 .

Now we recall briefly the P-index theory for symplectic paths. All the details can be found in [15, 1, 12].

In this section, we assume that P is some matrix of pattern (-I2s-2t)I2t, where 0ts.

As usual, the symplectic group Sp(2n) is defined by

Sp ( 2 n ) = { M GL ( 2 n , ) M T J M = J } ,

whose topology is induced from that of 4n2. For τ>0, we are interested in the following paths in Sp(2n):

𝒫 τ ( 2 n ) = { γ C ( [ 0 , τ ] , Sp ( 2 n ) ) γ ( 0 ) = I 2 n } ,

which are equipped with the topology induced from that of Sp(2n). The following real function was introduced in [1]:

D P , ω ( M ) = ( - 1 ) n - 1 ω ¯ n det ( M - ω P ) for all  ω 𝕌 , M Sp ( 2 n ) ,

where 𝕌 is the unit circle in the complex plane. Thus, for any ω𝕌, the following codimension 1 hypersurface in Sp(2n) is defined in [1]:

Sp ( 2 n ) P , ω 0 = { M Sp ( 2 n ) D P , ω ( M ) = 0 } .

For any MSp(2n)P,ω0, we define a co-orientation of Sp(2n)P,ω0 at M by the positive direction ddtMetϵJ|t=0 of the path MetϵJ, with 0t1 and ϵ>0 being sufficiently small. Let

Sp ( 2 n ) P , ω * = Sp ( 2 n ) Sp ( 2 n ) P , ω 0 ,
𝒫 P , τ , ω * ( 2 n ) = { γ 𝒫 τ ( 2 n ) γ τ Sp ( 2 n ) P , ω * } ,
𝒫 P , τ , ω 0 ( 2 n ) = 𝒫 τ ( 2 n ) 𝒫 P , τ , ω * ( 2 n ) .

For any two continuous arcs ξ and η:[0,τ]Sp(2n), with ξ(τ)=η(0), it is defined, as usual,

η * ξ ( t ) = { ξ ( 2 t ) if  0 t τ / 2 , η ( 2 t - 1 ) if  τ / 2 t τ .

Given any two 2mk×2mk matrices of square block form

M k = ( A k B k C k D k ) ,

with k=1,2, as in [15], the -product of M1 and M2 is defined by the following 2(m1+m2)×2(m1+m2) matrix M1M2:

M 1 M 2 = ( A 1 0 B 1 0 0 A 2 0 B 2 C 1 0 D 1 0 0 C 2 0 D 2 ) .

Denote by Mk the k-fold -product MM. Note that the -product of any two symplectic matrices is symplectic. For any two paths γj𝒫τ(2nj), with j=0,1, let γ1γ2(t)=γ1(t)γ2(t) for all t[0,τ].

A special path ξn𝒫τ(2n) is defined by

ξ n ( t ) = ( 2 - t / τ 0 0 ( 2 - t / τ ) - 1 ) n for  0 t τ .

Definition 3.2 (cf. [1] and [12, Definition 3.3]).

For any ω𝕌 and MSp(2n), via [15, Definition 5.4.4], we define

ν P , ω ( M ) = dim ker ( M - ω P ) .

For any τ>0 and γ𝒫τ(2n), define

ν P , ω ( γ ) = ν ω ( γ P ) = ν P , ω ( γ ( τ ) ) .

If γ𝒫P,τ,ω*(2n), define

(3.2) i P , ω ( γ ) = [ Sp ( 2 n ) P , ω 0 : γ ξ n ] ,

where the right-hand side of (3.2) is the usual homotopy intersection number, and the orientation of γξn is its positive time direction under homotopy with fixed end points.

If γ𝒫P,τ,ω0(2n), we let (γ) be the set of all open neighborhoods of γ in 𝒫τ(2n), and define

i P , ω ( γ ) = sup U ( γ ) inf { i P , ω ( β ) β U 𝒫 P , τ , ω * ( 2 n ) } .

Then (iP,ω(γ),νP,ω(γ))×{0,1,,2n} is called the P-index function of γ at ω.

For any MSp(2n) and ω𝕌, the splitting numbersSM±(P,ω) of M at (P,ω) are defined by

S M ± ( P , ω ) = lim ϵ 0 + i P , ω exp ( ± - 1 ϵ ) ( γ ) - i P , ω ( γ )

for any path γ𝒫τ(2n) satisfying γ(τ)=M.

Let Ω0(M) be the path connected component containing M=γ(τ) of the set

Ω ( M ) = { N Sp ( 2 n ) σ ( N ) 𝕌 = σ ( M ) 𝕌  and  ν λ ( N ) = ν λ ( M )  for all  λ σ ( M ) 𝕌 } .

Here Ω0(M) is called the homotopy component of M in Sp(2n).

In [15], the following symplectic matrices were introduced as basic normal forms:

D ( λ ) = ( λ 0 0 λ - 1 ) , λ = ± 2 ,
N 1 ( λ , b ) = ( λ b 0 λ ) , λ = ± 1 , b = ± 1 , 0 ,
R ( θ ) = ( cos θ - sin θ sin θ cos θ ) , θ ( 0 , π ) ( π , 2 π ) ,
N 2 ( ω , B ) = ( R ( θ ) B 0 R ( θ ) ) , θ ( 0 , π ) ( π , 2 π ) ,

where B=(b1b2b3b4), with bi and b2b3.

Splitting numbers possess the following properties.

Lemma 3.3 (cf. [1, Proposition 3.8]).

Let (pω(MP),qω(MP)) denote the Krein type of MP at ω. For any MSp(2n) and ωU, the splitting numbers SM±(P,ω) are well defined and satisfy the following properties:

  1. S M ± ( P , ω ) = S M P ± ( ω ) , where the right-hand side is the splitting numbers given by [ 15 , Definition 9.1.4].

  2. S M + ( P , ω ) - S M - ( P , ω ) = p ω ( M P ) - q ω ( M P ) .

  3. S M ± ( P , ω ) = S N ± ( P , ω ) if N P Ω 0 ( M P ) .

  4. S M 1 M 2 ± ( P , ω ) = S M 1 ± ( P 1 , ω ) + S M 2 ± ( P 2 , ω ) for M j , P j Sp ( 2 n j ) , with n j { 1 , , n } , satisfying P = P 1 P 2 and n 1 + n 2 = n .

  5. S M ± ( P , ω ) = 0 if ω σ ( M P ) .

By [1, Proposition 3.10], we have the following Bott-type formula for the (P,ω)-index.

Lemma 3.4.

For any γPτ(2n), zU and mN, we have

i P m , z ( γ m , P ) = ω m = z i P , ω ( γ ) , ν P m , z ( γ m , P ) = ω m = z ν P , ω ( γ ) .

4 Proof of the Main Theorem

In this section, we give the proof of the main theorem.

Lemma 4.1.

Let ΣHκ(2n) be (r,R)-pinched. Suppose um is a nonzero critical point of Ψa such that u corresponds to a P-symmetric closed characteristic (τ,y) for m2N-1, or a P-fixed closed characteristic (τ/2,y) for mN and τ2=A(τ,y). Then we have

(4.1) i ( u m ) 2 n l , if  m τ 2 / 2 > l π R 2 ,
(4.2) i ( u m ) + ν ( u m ) 2 n ( l - 1 ) - 1 if  m τ 2 / 2 < ( l - 1 / 2 ) π r 2 ,

for some lN.

Proof.

If u corresponds to a P-symmetric closed characteristic (τ,y) and m2-1, our lemma is just [14, Proposition 3]. If u corresponds to a P-fixed closed characteristic (τ/2,y) for m, our proof is completely the same as that of [14, Proposition 3]. ∎

By [13, Theorems 1.1 and 1.2], we have the following lemma.

Lemma 4.2.

Assume ΣHκ(2n) and 0<r|x|R for all xΣ with R/r<2. Then the following hold:

  1. There exist at least n - κ geometrically distinct P -symmetric closed characteristics ( τ i , y i ) on Σ for 1in-κ, where τi is the minimal period of yi, and the actions A(τi,yi) satisfy

    (4.3) π r 2 A ( τ i , y i ) π R 2 .

  2. There exist at least κ geometrically distinct P-fixed closed characteristics (σi,zi) on Σ for 1iκ, where σi is the minimal period of zi, and the actions A(σi,zi) satisfy (4.3) for 1iκ.

Lemma 4.3.

Let {(τ1,y1),,(τn-κ,yn-κ)} be the P-symmetric closed characteristics found in Lemma 4.2. Then we have

(4.4) Ψ a ( u i ) = 0 , i ( u i ) 2 ( i - 1 ) i ( u i ) + ν ( u i ) - 1 ,

for 1in-κ, where ui is the unique critical point of Ψa corresponding to (τi,yi). Furthermore, if R/r<3/2, then for n-κ+1in, (4.4) also holds for a critical point ui of Ψa corresponding to a prime closed characteristic (τi,yi) which is P-symmetric or P-fixed.

Proof.

By the proof of Lemma 4.2 and Proposition 2.2, (4.4) holds for 1in-κ. For n-κ+1in, by Proposition 2.2, we can find a closed characteristic (τi,yi) which is P-symmetric or P-fixed and corresponds to a critical point ui of Ψa satisfying (4.4). In the following, we prove that (τi,yi) is prime for n-κ+1in when R/r<3/2.

Note that, by [3, Theorem V.1.4], for any closed characteristic (τ,y), we have A(τ,y)πr2. Then, by the pinching condition R/r<3/2, we obtain A(mτ,ym)mπr2>πR2 and A((2m-1)τ,y2m-1)/2(2m-1)πr2/2>πR2 for m2, which together with (4.1) give i(um)2n, where u is the unique critical point of Ψa corresponding to (τ,y). On the other hand, by (4.4), we have i(ui)2(i-1)2n-2 for in, thus we must have that (τi,yi) is prime. ∎

Corollary 4.4.

Assume ΣHκ(2n) and 0<r|x|R for all xΣ, with R/r<3/2. If Σ carries exactly nP-invariant closed characteristics, which we denote, as in Lemma 4.3, by {(τ1,y1),,(τn,yn)}, then for 1in-κ, (τi,yi) is prime P-symmetric, and for n-κ+1in, (τi,yi) is prime P-fixed. In addition, for all 1in, (4.3) holds.

Proof.

By Lemma 4.3, for 1in-κ, (τi,yi) is prime P-symmetric, and for n-κ+1in, (τi,yi) is prime P-fixed or P-symmetric. Since we have exactly nP-invariant closed characteristics by assumption, then, by Lemma 4.2 (ii), we know that for n-κ+1in, (τi,yi) is prime P-fixed and they are just the P-fixed ones {(σi,zi)1iκ}. Thus, for all 1in, (4.3) holds from Lemma 4.2. ∎

By [12, Proposition 2.13], we have that the fundamental solution γ=γy:[0,τ/2]Sp(2n), with γy(0)=I2n, of the linearized Hamiltonian system

γ y ˙ ( t ) = J H 2 ′′ ( y ( t ) ) γ y ( t ) for all  t

satisfies

(4.5) γ y ( τ / 2 ) P = Q y - 1 ( I 2 M ) Q y

for some symplectic matrices Qy and M (cf. also [15, Lemmas 15.2.3 and 15.2.4]). We call γ=γy:[0,τ/2]Sp(2n) the associated symplectic path of (τ,y).

Lemma 4.5.

Let (τ,y) (resp. (τ/2,y)) be a hyperbolic P-symmetric (resp. P-fixed) closed characteristic and let γγy:[0,τ/2]Sp(2n) be its associated symplectic path. Then the following hold:

  1. i P , - 1 ( γ ) = i P , 1 ( γ ) + 1 ,

  2. i P , 1 ( γ ) max ( κ , [ n 2 ] ) and i P , - 1 ( γ ) max ( n - κ , [ n 2 ] + 1 , κ + 1 ) ,

  3. i P m , 1 ( γ m , P ) = m ( i P , 1 ( γ ) + 1 ) - 1 , νPm,1(γm,P)=2 for all m.

Proof.

Since (τ,y) or (τ/2,y) is hyperbolic, e(τ,y)=2. Noticing that e(τ,y)=e(γ(τ))=e((γ(τ/2)P)2), we have σ(M)𝕌= in (4.5). Then (i) and (iii) follow from the definition of splitting numbers, [15, List 9.1.12] and Lemmas 3.33.4. By [1, Theorem 4.1], we have κiP,1(γ) and n-κiP,-1(γ), which, together with (i), gives (ii). ∎

Now we give the proof of the main theorem.

Let (τi,yi) be a hyperbolic P-invariant closed characteristic for some i[1,n]. By [12, Proposition 2.13], we have γyi(τi/2)P=Qyi-1(I2Mi)Qyi, where σ(Mi)𝕌=, and Qyi,Mi are symplectic matrices. We first consider two cases.

Case 1.

( τ i , y i ) is a hyperbolic P-symmetric closed characteristic and R/r<5/3.

By Theorem 3.1 (ii) and Lemma 4.5 (iii), we have

i ( u i 3 ) + ν ( u i 3 ) = i P , 1 ( γ i 3 ) - κ + ν P , 1 ( γ i 3 ) - 1
= 3 i P , 1 ( γ i ) + 2 - κ + 1
(4.6) = 3 i P , 1 ( γ i ) - κ + 3 .

By Lemma 4.3, we have that (4.3) holds for 1in-κ, which implies 3A(τi,yi)/2<(3-1/2)πr2. This, together with (4.2), implies

(4.7) i ( u i 3 ) + ν ( u i 3 ) 4 n - 1 .

Combining (4.6) with (4.7), we obtain

(4.8) i P , 1 ( γ i ) [ 4 n + κ - 4 3 ] .

By Theorem 3.1 (ii) and (4.4), we obtain

(4.9) i P , 1 ( γ i ) - κ = i ( u i ) = 2 ( i - 1 ) .

Now we proceed our proof according to the value of κ. Subcase 1.1.0κ[n-12].

From (4.8), (4.9) and Lemma 4.5 (ii), we have

[ n 2 ] i P , 1 ( γ i ) = 2 ( i - 1 ) + κ [ 4 n + κ - 4 3 ] ,

which implies

(4.10) E ( n - 2 κ + 3 4 ) i [ 2 n - κ + 1 3 ] .

On the other hand, note that iP,-1(γi)=iP,1(γi)+1, by Lemma 4.5 (i). Using this, together with (4.9) and Lemma 4.5 (ii), we get

n - κ i P , - 1 ( γ i ) = 2 ( i - 1 ) + κ + 1 .

This, together with (4.10), yields

(4.11) E ( n - 2 κ + 1 2 ) i [ 2 n - κ + 1 3 ] .

Comparing (4.10) with (4.11) and noticing that 0κ[n-12], we obtain

(4.12) E ( n - 2 κ + 1 2 ) i [ 2 n - κ + 1 3 ] .

Subcase 1.2. [ n + 1 2 ] κ n .

From (4.8), (4.9) and Lemma 4.5 (ii), we have

κ i P , 1 ( γ i ) = 2 ( i - 1 ) + κ [ 4 n + κ - 4 3 ] ,

which implies

(4.13) 1 i [ 2 n - κ + 1 3 ] .

On the other hand, note that iP,-1(γi)=iP,1(γi)+1, by Lemma 4.5 (i). Using this, together with (4.8), (4.9) and Lemma 4.5 (ii), we have

(4.14) [ n 2 ] + 1 i P , - 1 ( γ i ) = 2 ( i - 1 ) + κ + 1 [ 4 n + κ - 4 3 ] + 1 .

Comparing (4.13) with (4.14) and noticing that [n+12]κn, we obtain

1 i [ 2 n - κ + 1 3 ] .

In this subcase, if we impose the condition R/r<3/2, then by Lemma 4.3 we have that (4.3) holds for 1in-κ, which implies 3A(τi,yi)/23πr2/2>πR2. This, together with (4.1), implies

(4.15) i ( u i 3 ) 2 n .

By Theorem 3.1 (ii) and Lemma 4.5 (iii), we have

(4.16) i ( u i 3 ) = i P , 1 ( γ i 3 ) - κ = 3 i P , 1 ( γ i ) + 2 - κ .

Combining (4.9) with (4.15)–(4.16), we get

(4.17) i E ( n - κ + 2 3 ) .

Case 2.

( τ i , y i ) is a hyperbolic P-fixed closed characteristic and R/r<3/2.

By Lemma 4.5 (iii), we have

(4.18) i ( γ i 2 ) = i P 2 , 1 ( γ i 2 , P ) = 2 i P , 1 ( γ i ) + 1 ,
(4.19) i ( γ i 2 ) + ν ( γ i 2 ) - 1 = i P 2 , 1 ( γ i 2 , P ) + ν P 2 , 1 ( γ i 2 , P ) - 1 = 2 i P , 1 ( γ i ) + 2 .

On the other hand, by the pinching condition R/r<3/2, we have A(2τi,yi2)2πR2<3πr2, which, together with [19, Lemma 3.1] and [15, Theorem 15.1.1], gives

(4.20) i ( γ i 2 ) 3 n ,
(4.21) i ( γ i 2 ) + ν ( γ i 2 ) - 1 5 n - 1 .

By (4.18)–(4.21), we obtain

(4.22) E ( 3 n - 1 2 ) i P , 1 ( γ i ) [ 5 n - 3 2 ] .

Since (τi,yi) is hyperbolic, by Theorem 3.1 (i) and (4.4), we have

(4.23) i P , 1 ( γ i ) - κ = i ( u i ) = 2 ( i - 1 ) .

In the following, we proceed our proof according to the value of κ. Subcase 2.1.0κ[n-12].

By (4.22) and (4.23), we have

E ( 3 n - 1 2 ) i P , 1 ( γ i ) = 2 ( i - 1 ) + κ [ 5 n - 3 2 ] .

This implies

(4.24) E ( 3 n - 2 κ + 3 4 ) i [ 5 n - 2 κ + 1 4 ] ,

which, together with 0κ[n-12] and in, gives

(4.25) E ( 3 n - 2 κ + 3 4 ) i n .

Subcase 2.2. [ n + 1 2 ] κ n .

In this subcase, (4.24) also holds. Since [5n-2κ+14]n and E(3n-2κ+34)1 for [n+12]κn, we have

(4.26) E ( 3 n - 2 κ + 3 4 ) i [ 5 n - 2 κ + 1 4 ] .

Now we can prove Theorem 1.1.

Proof of Theorem 1.1.

(i) We have the condition 0κ[n-12] and Σ is (r,R)-pinched with R/r<5/3.

Since 2>5/3, by Lemma 4.3, we obtain n-κ geometrically distinct P-symmetric closed characteristics {(τi,yi)1in-κ} on Σ. From (4.12), there exist at most [2n-κ+13]-E(n-2κ+12)+1 hyperbolic ones in {(τi,yi)1in-κ}, hence there exist at least

(4.27) n - κ - [ 2 n - κ + 1 3 ] + E ( n - 2 κ + 1 2 ) - 1 = E ( n - 2 κ - 1 2 ) + E ( n - 2 κ - 1 3 )

non-hyperbolic P-symmetric closed characteristics on Σ. Since E(3n-2κ+34)n-κ+1, by (4.25), all of the closed characteristics {(τi,yi)n-κ+1in} may be hyperbolic. In summary, we obtain at least E(n-2κ-12)+E(n-2κ-13) non-hyperbolic P-invariant closed characteristics on Σ.

(ii) We have the condition [n+12]κn and Σ is (r,R)-pinched with R/r<3/2, which carries exactly nP-invariant closed characteristics.

By Corollary 4.4, we know that {(τi,yi)n-κ+1in} are P-fixed. Since E(3n-2κ+34)n-κ+1, by (4.26), there exist at most [5n-2κ+14]-E(3n-2κ+34)+1 hyperbolic ones in {(τi,yi)n-κ+1in}. Then there exist at least

(4.28) κ - [ 5 n - 2 κ + 1 4 ] + E ( 3 n - 2 κ + 3 4 ) - 1 = 2 E ( 2 κ - n - 1 4 )

non-hyperbolic P-fixed closed characteristics on Σ. By (4.17), at least E(n-κ+23)-1=E(n-κ-13) of the P-symmetric closed characteristics {(τi,yi)1in-κ} are non-hyperbolic. In summary, we obtain at least 2E(2κ-n-14)+E(n-κ-13) non-hyperbolic P-invariant closed characteristics on Σ. The proof is complete. ∎

Remark 4.6.

We prove identities (4.27) and (4.28). We have

n - κ - [ 2 n - κ + 1 3 ] + E ( n - 2 κ + 1 2 ) - 1 = - ( [ 2 n - κ + 1 3 ] - n + κ ) + E ( n - 2 κ - 1 2 )
= - [ - n + 2 κ + 1 3 ] + E ( n - 2 κ - 1 2 )
= E ( n - 2 κ - 1 3 ) + E ( n - 2 κ - 1 2 )

and

κ - [ 5 n - 2 κ + 1 4 ] + E ( 3 n - 2 κ + 3 4 ) - 1 = - [ n - 2 κ + 1 4 ] + ( E ( 3 n - 2 κ + 3 4 ) - 1 - n + κ )
= - [ n - 2 κ + 1 4 ] + E ( - n + 2 κ - 1 4 )
= 2 E ( 2 κ - n - 1 4 ) .


Communicated by Yiming Long


Award Identifier / Grant number: 11401555

Award Identifier / Grant number: 11771341

Award Identifier / Grant number: 11471240

Award Identifier / Grant number: 1608085QA01

Funding statement: The first author was partially supported by NSFC (Nos. 11401555, 11771341), Anhui Provincial Natural Science Foundation (No. 1608085QA01). The second author was partially supported by NSFC (No. 11471240).

References

[1] Y. Dong and Y. Long, Closed characteristic on partially symmetric compact convex hypersurfaces in 2n, J. Differential Equations 196 (2004), no. 1, 226–248. 10.1016/S0022-0396(03)00168-2Search in Google Scholar

[2] Y. Dong and Y. Long, Stable closed characteristics on partially symmetric convex hypersurfaces, J. Differential Equations 206 (2004), no. 2, 265–279. 10.1016/j.jde.2004.03.004Search in Google Scholar

[3] I. Ekeland, Convexity Methods in Hamiltonian Mechanics, Ergeb. Math. Grenzgeb. (3) 19, Springer, Berlin, 1990. 10.1007/978-3-642-74331-3Search in Google Scholar

[4] I. Ekeland and H. Hofer, Convex Hamiltonian energy surfaces and their periodic trajectories, Comm. Math. Phys. 113 (1987), no. 3, 419–469. 10.1007/BF01221255Search in Google Scholar

[5] I. Ekeland and J.-M. Lasry, On the number of periodic trajectories for a Hamiltonian flow on a convex energy surface, Ann. of Math. (2) 112 (1980), no. 2, 283–319. 10.2307/1971148Search in Google Scholar

[6] I. Ekeland and L. Lassoued, Multiplicité des trajectoires fermées de systèmes hamiltoniens convexes, Ann. Inst. H. Poincaré Anal. Non Linéaire 4 (1987), no. 4, 307–335. 10.1016/s0294-1449(16)30362-6Search in Google Scholar

[7] E. R. Fadell and P. H. Rabinowitz, Generalized cohomological index theories for Lie group actions with an application to bifurcation questions for Hamiltonian systems, Invent. Math. 45 (1978), no. 2, 139–174. 10.1007/BF01390270Search in Google Scholar

[8] M. Girardi, Multiple orbits for Hamiltonian systems on starshaped surfaces with symmetries, Ann. Inst. H. Poincaré Anal. Non Linéaire 1 (1984), no. 4, 285–294. 10.1016/s0294-1449(16)30423-1Search in Google Scholar

[9] X. Hu and S. Sun, Index and stability of symmetric periodic orbits in Hamiltonian systems with application to figure-eight orbit, Comm. Math. Phys. 290 (2009), no. 2, 737–777. 10.1007/s00220-009-0860-ySearch in Google Scholar

[10] C.-G. Liu, Y. Long and C. Zhu, Multiplicity of closed characteristics on symmetric convex hypersurfaces in 2n, Math. Ann. 323 (2002), no. 2, 201–215. 10.1007/s002089100257Search in Google Scholar

[11] H. Liu, Stability of symmetric closed characteristics on symmetric compact convex hypersurfaces in 2n under a pinching condition, Acta Math. Sin. (Engl. Ser.) 28 (2012), no. 5, 885–900. 10.1007/s10114-011-0494-9Search in Google Scholar

[12] H. Liu, Multiple P-invariant closed characteristics on partially symmetric compact convex hypersurfaces in 2n, Calc. Var. Partial Differential Equations 49 (2014), no. 3–4, 1121–1147. 10.1007/s00526-013-0614-8Search in Google Scholar

[13] H. Liu and D. Zhang, On the number of P-invariant closed characteristics on partially symmetric compact convex hypersurfaces in 2n, Sci. China Math. 58 (2015), no. 8, 1771–1778. 10.1007/s11425-014-4903-2Search in Google Scholar

[14] H. Liu and D. Zhang, Stable P-symmetric closed characteristics on partially symmetric compact convex hypersurfaces, Discrete Contin. Dyn. Syst. 36 (2016), no. 2, 877–893. 10.3934/dcds.2016.36.877Search in Google Scholar

[15] Y. Long, Index Theory for Symplectic Paths with Applications, Progr. Math. 207, Birkhäuser, Basel, 2002. 10.1007/978-3-0348-8175-3Search in Google Scholar

[16] Y. Long and C. Zhu, Closed characteristics on compact convex hypersurfaces in 2n, Ann. of Math. (2) 155 (2002), no. 2, 317–368. 10.2307/3062120Search in Google Scholar

[17] P. H. Rabinowitz, Periodic solutions of Hamiltonian systems, Comm. Pure Appl. Math. 31 (1978), no. 2, 157–184. 10.21236/ADA096642Search in Google Scholar

[18] A. Szulkin, Morse theory and existence of periodic solutions of convex Hamiltonian systems, Bull. Soc. Math. France 116 (1988), no. 2, 171–197. 10.24033/bsmf.2094Search in Google Scholar

[19] W. Wang, Stability of closed characteristics on compact hypersurfaces in 2n under a pinching condition, Adv. Nonlinear Stud. 10 (2010), no. 2, 263–272. 10.1515/ans-2010-0202Search in Google Scholar

[20] W. Wang, Closed trajectories on symmetric convex Hamiltonian energy surfaces, Discrete Contin. Dyn. Syst. 32 (2012), no. 2, 679–701. 10.3934/dcds.2012.32.679Search in Google Scholar

[21] W. Wang, Closed characteristics on compact convex hypersurfaces in 8, Adv. Math. 297 (2016), 93–148. 10.1016/j.aim.2016.03.044Search in Google Scholar

[22] W. Wang, X. Hu and Y. Long, Resonance identity, stability, and multiplicity of closed characteristics on compact convex hypersurfaces, Duke Math. J. 139 (2007), no. 3, 411–462. 10.1215/S0012-7094-07-13931-0Search in Google Scholar

[23] A. Weinstein, Periodic orbits for convex Hamiltonian systems, Ann. of Math. (2) 108 (1978), no. 3, 507–518. 10.2307/1971185Search in Google Scholar

[24] D. Zhang, P-cyclic symmetric closed characteristics on compact convex P-cyclic symmetric hypersurface in 2n, Discrete Contin. Dyn. Syst. 33 (2013), no. 2, 947–964. 10.3934/dcds.2013.33.947Search in Google Scholar

Received: 2017-09-23
Revised: 2018-01-24
Accepted: 2018-01-28
Published Online: 2018-02-14
Published in Print: 2018-11-01

© 2018 Walter de Gruyter GmbH, Berlin/Boston

This work is licensed under the Creative Commons Attribution 4.0 International License.

Downloaded on 23.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/ans-2017-6050/html?lang=en
Scroll to top button