Computation of annular capacity by Hamiltonian Floer theory of non-contractible periodic trajectories
Morimichi Kawasaki, Ryuma Orita

TL;DR
None
Contribution
None
Abstract
The first author introduced a relative symplectic capacity for a symplectic manifold and its subset which measures the existence of non-contractible periodic trajectories of Hamiltonian isotopies on the product of with the annulus . In the present paper, we give an exact computation of the capacity of the -torus relative to a Lagrangian submanifold which implies the existence of non-contractible Hamiltonian periodic trajectories on . Moreover, we give a lower bound on the number of such trajectories.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Computation of annular capacity by Hamiltonian Floer theory of non-contractible periodic trajectories
Morimichi Kawasaki
Center for Geometry and Physics, Institute for Basic Science (IBS), Pohang 790-784, Republic of Korea
and
Ryuma Orita
Graduate School of Mathematical Sciences, the University of Tokyo, Tokyo 153-0041, Japan
[email protected] https://sites.google.com/site/oritaryuma/
Abstract.
The first author [Ka] introduced a relative symplectic capacity for a symplectic manifold and its subset which measures the existence of non-contractible periodic trajectories of Hamiltonian isotopies on the product of with the annulus . In the present paper, we give an exact computation of the capacity of the -torus relative to a Lagrangian submanifold which implies the existence of non-contractible Hamiltonian periodic trajectories on . Moreover, we give a lower bound on the number of such trajectories.
2010 Mathematics Subject Classification:
53D40, 37J10, 37J45
Contents
1. Introduction
Let be a closed connected symplectic manifold and a compact subset. For let denote the annulus , and for we define its subset . For we put
[TABLE]
We consider the product symplectic manifold \bigl{(}A_{R}\times N,(dp_{0}\wedge dq_{0})\oplus\omega_{N}\bigr{)} and the free homotopy class , where and is the free homotopy class of trivial loops in . Our main result is Theorem 1.1. We consider (N,\omega_{N})=(\mathbb{T}^{2n},\omega_{\mathrm{std}})=\bigl{(}(\mathbb{R}/2\mathbb{Z}\times\mathbb{R}/\mathbb{Z})^{n},\omega_{\mathrm{std}}\bigr{)} and stands for where is the standard symplectic form on .
Theorem 1.1**.**
Let and be real numbers such that . For any smooth Hamiltonian with compact support and any such that
[TABLE]
there exists a Hamiltonian periodic trajectory in the homotopy class with action , where is the action functional defined in Section 2. Moreover, if is non-degenerate, then the number of such ’s is at least
[TABLE]
The result in Theorem 1.1 is sharp in the sense that for any there exists a Hamiltonian with without periodic trajectory in (see the proof of Theorem 1.2 in Subsection 4.2).
To obtain Theorem 1.1, we will prove the non-zeroness of the homomorphism which is defined in Subsection 2.5. To prove the non-zeroness, we use Poźniak’s theorem (Theorem 3.2) several times. Biran, Polterovich and Salamon [BPS], Niche [Ni] and Xue [Xu] also used Poźniak’s theorem to get an upper bound of the Biran–Polterovich–Salamon capacity. In their papers, the homomorphism is an isomorphism. However in our case, is not an isomorphism. Thus we need more sophisticated arguments.
We define a capacity introduced by the first author in [Ka] in terms of the Biran–Polterovich–Salamon capacity [BPS, Subsection 3.2] (see also [Ni, We, Xu]). Let be an open symplectic manifold and a compact subset. For and we define the Biran–Polterovich–Salamon capacity by
[TABLE]
where is the set of time-dependent Hamiltonian functions with compact support such that
[TABLE]
and is the set of periodic trajectories of the Hamiltonian isotopy associated to representing . For , , and , we then define a relative symplectic capacity by
[TABLE]
By using the capacity , we can rewrite Theorem 1.1 as Theorem 1.2 (see Subsection 4.2 for details).
Theorem 1.2**.**
For any , , and , we have
[TABLE]
After the first draft of this paper was completed, Ishiguro [Is] pointed out the following proposition.
Proposition 1.3** ([Is, Proposition 5.1]).**
For any , , , and , we have
[TABLE]
On the other hand, the first author essentially showed Theorem 1.4 in [Ka].
Theorem 1.4** ([Ka, Theorem 1.2]).**
For any and we have
[TABLE]
Therefore Theorem 1.2 improves Theorem 1.4. Moreover, the first author proposed the following conjecture.
Conjecture 1.5** ([Ka, Conjecture 3.1]).**
Let be a stably non-displaceable compact subset of a closed symplectic manifold . Show that the equality
[TABLE]
holds for any and .
Theorem 1.2 proves Conjecture 1.5 for .
The paper is organized as follows. In Section 2, we introduce the Floer homology and the symplectic homology for non-contractible trajectories which are the main tools to prove our main theorems. In Section 3, we calculate the dimensions of the Floer homology and the symplectic homology to prove our main theorems. In Section 4, we prove our main theorems (Theorems 1.1 and 1.2).
2. Symplectic homology
In this section, we define the Floer homology for non-contractible periodic trajectories (see [BPS, Section 4] for details).
Let be a compact symplectic manifold with convex boundary (i.e., there exists a Liouville vector field defined on an open neighborhood of in and pointing outward along ) and denote . Although the product of compact symplectic manifolds with convex boundary need not have convex boundary, we can still define the Floer homology of the product according to [FS, Products, Section 3]. In Section 3, we will consider the Floer homology of the product which has no convex boundary, where .
2.1. Action functional
For a free homotopy class denote by the space of free loops representing . In addition, we assume that our manifold is symplectically -atoroidal, i.e., for any free loop in , that is for considered as the map from the two-torus,
[TABLE]
hold where is the first Chern class.
Let be a Hamiltonian with compact support. Let denote for . The Hamiltonian vector field associated to is defined by
[TABLE]
The Hamiltonian isotopy associated to is defined by
[TABLE]
and its time-one map is referred to as the Hamiltonian diffeomorphism of . Let be the set of one-periodic trajectories of representing . A one-periodic trajectory is called non-degenerate if it satisfies \det\bigl{(}d\varphi_{H}(x(0))-\mathrm{id}\bigr{)}\neq 0.
Fix a reference loop . We define the action functional by
[TABLE]
where and is a path in between and considered as a map from the annulus to . Since our manifold is symplectically -atoroidal, the functional is well-defined as a real-valued function. Note that is equal to the set of critical points of .
We define the action spectrum of by
[TABLE]
Let and be real numbers such that . Suppose that the Hamiltonian satisfies and that it is regular, i.e., every one-periodic trajectory is non-degenerate. We define where .
2.2. Filtered Floer chain complex
We define the chain group of our Floer chain complex to be the -vector space
[TABLE]
where
[TABLE]
Let be a time-dependent smooth family of -compatible almost complex structures on such that is convex and independent of near the boundary . Consider the Floer differential equation
[TABLE]
Here we note that
[TABLE]
for all . For a smooth solution to (1) we define the energy by the formula
[TABLE]
Then we have the following lemma.
Lemma 2.1** ([Sa]).**
Let be a smooth solution to (1) with finite energy.
- (i)
There exist periodic solutions such that
[TABLE]
where both limits are uniform in the -variable. 2. (ii)
The energy identity holds:
[TABLE]
We call a family of almost complex structures regular if the linearized operator for (1) is surjective for any finite-energy solution of (1) in the homotopy class . We denote by the space of regular families of almost complex structures. This subspace is generic in (see [FHS]). For any and any pair the space
[TABLE]
is a smooth manifold whose dimension near such a solution is given by the difference of the Conley–Zehnder indices (see [SZ]) of and relative to . The subspace of solutions of relative index 1 is denoted by . For the quotient is a finite set for any pair . We define the boundary operator by
[TABLE]
for where denotes the modulo 2 counting.
Theorem 2.2** ([Fl]).**
If is regular, then the operator is well-defined and satisfies .
The energy identity (ii) implies that is invariant under the boundary operator . Thus we get an operator on the quotient .
Definition 2.3**.**
The filtered Floer homology group is defined to be
[TABLE]
Theorem 2.4** ([Fl, Sa, SZ]).**
If are two regular almost complex structures, then there exists a natural isomorphism
[TABLE]
We refer to as the Floer homology associated to .
2.3. Continuation
We define the set
[TABLE]
Proposition 2.5** ([BPS, Remark 4.4.1]).**
Every Hamiltonian has a neighborhood such that the Floer homology groups , for any regular and any regular almost complex structure , are naturally isomorphic.
According to Proposition 2.5, one can define the Floer homology whether is regular or not.
Definition 2.6**.**
For we define , where is any regular Hamiltonian sufficiently close to .
Remark 2.7* ([BPS, Remark 4.4.2]).*
We can define the filtered Floer homology if the action interval does not contain zero, i.e, either or .
2.4. Monotone homotopies
We introduce a bidirected partial order on by
[TABLE]
Then there exists a homotopy from to such that . We call such a homotopy of Hamiltonians monotone. Let be a nontrivial free homotopy class and such that . It follows from the energy identity
[TABLE]
that the Floer chain map , defined in terms of the solutions of the equation
[TABLE]
preserves the subcomplexes and . Hence every monotone homotopy induces a natural homomorphism
[TABLE]
whenever satisfy (see [BPS, Subsection 4.5]). The homomorphism is called the monotone homomorphism from to . We call a monotone homotopy action-regular if takes values in a connected component of .
Lemma 2.8** ([FH, CFH]).**
The monotone homomorphism is independent of the choice of the monotone homotopy used to define it and
[TABLE]
whenever satisfy .
Lemma 2.9** ([Vi]).**
The monotone homomorphism associated to an action-regular monotone homotopy is an isomorphism.
Given and two Hamiltonians satisfying , we obtain the following commutative diagram, whose rows are the short exact sequences for and for .
[TABLE]
where and denote the natural inclusion and projection, respectively. The associated long exact sequences induce the following commutative diagram.
[TABLE]
2.5. Symplectic homology
In this subsection, we consider a homology introduced in [FH, CFH, Ci]. We refer to [BPS, Subsection 4.8] for details. Let be a nontrivial free homotopy class and such that . As mentioned in Subsection 2.4, there is a natural homomorphism
[TABLE]
whenever satisfy . These homomorphisms define an inverse system of Floer homology groups over \bigl{(}\mathcal{H}^{a,b}(M;\alpha),\preceq\bigr{)}. We denote the symplectic homology of in the homotopy class for the action interval by
[TABLE]
Fix a compact subset and a constant . We define the set
[TABLE]
This defines a directed system of Floer homology groups over \bigl{(}\mathcal{H}_{c}^{a,b}(M,A;\alpha),\preceq\bigr{)}. We denote the relative symplectic homology of the pair at the level in the homotopy class for the action interval by
[TABLE]
Proposition 2.10** ([BPS, Proposition 4.8.2]).**
Let be a nontrivial homotopy class and suppose that . Then for any there exists a unique homomorphism
[TABLE]
such that for any two Hamiltonians with the following diagram commutes.
[TABLE]
Here and are the canonical homomorphisms. In particular, since for any , the following diagram commutes.
[TABLE]
Remark 2.11*.*
As we noted in Remark 2.7, we can still define the filtered symplectic homology for and the conclusion of Proposition 2.10 still holds if the action interval does not contain zero.
3. Computation
3.1. Morse–Bott theory in Floer homology
We refer to [BPS, Subsection 5.2] for details.
Definition 3.1** ([BPS]).**
A subset is called a Morse–Bott manifold of periodic trajectories for if the set is a compact submanifold of and T_{x_{0}}C_{0}=\mathop{\mathrm{Ker}}\nolimits\bigl{(}d\varphi_{H}^{1}(x(0))-\mathrm{id}\bigr{)} for any .
Theorem 3.2** ([BPS, Theorem 5.2.2]).**
Let , , and . Assume that the set is a connected Morse–Bott manifold of periodic trajectories for . Let be a Riemannian metric on and a Morse–Smale function. Then the Floer homology \mathrm{HF}_{\ast}^{[a,b)}(H;\alpha)=H_{\ast}\bigl{(}\mathrm{CF}_{\ast}^{[a,b)}(H;\alpha),\partial_{\ast}\bigr{)} coincides with the Morse homology \mathrm{HM}_{\ast}(C_{0},f,g)=H_{\ast}\bigl{(}\mathrm{CM}_{\ast}(C_{0},f,g),\partial_{\ast}\bigr{)}. Namely, we have
[TABLE]
The original version of Theorem 3.2 can be found in [Po, Theorem 3.4.11].
Remark 3.3*.*
The grading of the Floer homology groups is well-defined up to an additive constant. More precisely, with a suitable choice of this grading, is isomorphic to , and hence to if is orientable. In the present paper, we choose this grading for simplicity.
3.2. Dimensions of symplectic homology groups
Let be a closed connected symplectic manifold and a compact subset. Let denote the annulus , and for we define its subset . For we put
[TABLE]
We consider the product symplectic manifold \bigl{(}A_{R}\times N,(dp_{0}\wedge dq_{0})\oplus\omega_{N}\bigr{)} and the free homotopy class , where .
In this subsection, we give an explicit computation of symplectic homology groups in the case that (N,\omega_{N})=(\mathbb{T}^{2n},\omega_{\mathrm{std}})=\bigl{(}(\mathbb{R}/2\mathbb{Z}\times\mathbb{R}/\mathbb{Z})^{n},\omega_{\mathrm{std}}\bigr{)} and . In the following theorem, let [math] denote the trivial homotopy class .
Theorem 3.4**.**
Let and be real numbers such that . Then for any and any , we have
[TABLE]
and
[TABLE]
For , we have the following result. For the sake of brevity, we put
[TABLE]
Theorem 3.5**.**
Let and be real numbers such that , and . Then for any and , we have
[TABLE]
and
[TABLE]
Theorem 3.6**.**
Let and be real numbers such that and let . The homomorphism
[TABLE]
is non-zero if and only if and . Moreover, in this case, the homomorphism is surjective and
[TABLE]
where is the -th Betti number of .
We shall denote a point in by
[TABLE]
where and . Then we have
[TABLE]
We put and .
3.3. Proof of Theorem 3.4
Proof.
The proof of [BPS, Theorem 5.1.1] carries over almost literally. Fix a positive real number and choose a smooth family of real functions , defined for , with the following properties (cf. [BPS, Subsection 5.4]).
- (i)
for all and . 2. (ii)
For any
[TABLE] 3. (iii)
For all and we have . 4. (iv)
If , then
[TABLE]
and for all . 5. (v)
If , then
[TABLE]
for and for . 6. (vi)
For any the only critical point of with is .
Now choose a family of Hamiltonians with compact support so that for
[TABLE]
and for
[TABLE]
We note that every contractible trajectory in is constant. By (vi), for the set of contractible periodic trajectories is denoted by
[TABLE]
and for
[TABLE]
Lemma 3.7** ([BPS, Lemma 5.3.1]).**
For every the set is a Morse–Bott manifold of periodic trajectories for if and only if .
By (ii) and Lemma 3.7, the set is a Morse–Bott manifold of periodic trajectories for . Moreover, for and . For every , the action of is
[TABLE]
By Theorem 3.2, for every we have
[TABLE]
By [BPS, Proposition 4.5.1], the monotone homomorphism
[TABLE]
is an isomorphism whenever and either or . By [BPS, Lemma 4.7.1 (ii)], the homomorphism
[TABLE]
is an isomorphism for any such that . Therefore,
[TABLE]
By [BPS, Lemma 4.7.1 (i)], the homomorphism
[TABLE]
is an isomorphism for any if , and for any with if . Hence we obtain
[TABLE]
Thus the proof of Theorem 3.4 is complete. ∎
3.4. Proof of Theorem 3.5
Proof.
We assume, without loss of generality, that . Fix a positive real number and choose a smooth family of real functions , defined for , with the following properties (cf. [BPS, Subsection 5.5]).
- (i)
for all and . 2. (ii)
For any
[TABLE] 3. (iii)
For all and we have . 4. (iv)
If , then
[TABLE]
for all , and
[TABLE] 5. (v)
If , then
[TABLE]
for , for , and
[TABLE] 6. (vi)
For any such that there exist real numbers such that
[TABLE]
and for any . 7. (vii)
For any there exist real numbers such that
[TABLE]
and for any . 8. (viii)
For any the only possible points with must satisfy and the number of such is finite.
\begin{overpic}[width=227.62204pt,clip]{1+4.png} \put(115.0,155.0){\line(-1,0){2.0}} \put(116.0,104.0){} \put(116.0,150.0){} \put(191.0,104.0){} \put(25.0,104.0){} \put(230.0,111.0){} \put(180.0,170.0){} \put(16.0,46.0){} \end{overpic}
Now choose a family of Hamiltonians with compact support so that for
[TABLE]
and for
[TABLE]
Here we note that the condition ensures that the Hamiltonians , are well-defined.
\begin{overpic}[width=170.71652pt,clip]{1.png} \put(88.0,63.0){} \put(162.0,45.0){} \put(140.0,133.0){} \end{overpic}
\begin{overpic}[width=170.71652pt,clip]{4.png} \put(78.0,76.0){} \put(164.0,45.0){} \put(140.0,133.0){} \end{overpic}
Then for the corresponding Hamiltonian vector field is of the form
[TABLE]
and for
[TABLE]
For let us now consider the nontrivial free homotopy class
[TABLE]
Then for the set of periodic trajectories in is denoted by
[TABLE]
where
[TABLE]
and for
[TABLE]
where
[TABLE]
and
[TABLE]
Note that there are no periodic trajectories representing if and . Given with , we denote
[TABLE]
Given with , we denote
[TABLE]
Given with , we denote
[TABLE]
Then and are diffeomorphic to , and is diffeomorphic to . In summary, we have
[TABLE]
where the union runs over such that .
Lemma 3.8** ([BPS, Lemma 5.3.2]).**
The sets , and are Morse–Bott manifolds of periodic trajectories for if and only if .
We claim that the value of the action functional on is negative. In fact, since we have for any periodic trajectory in , the action of such a periodic trajectory is , and this is negative by (viii). On the other hand, by (vi), (vii) and Lemma 3.8, and are Morse–Bott manifolds of periodic trajectories for and the values of the action functional on these critical manifolds are
[TABLE]
and
[TABLE]
respectively. Fix a real number . We prove Theorem 3.5 in four steps.
Step 1**.**
If , then .
\begin{overpic}[width=227.62204pt,clip]{1g.png} \put(116.0,29.0){} \put(191.0,29.0){} \put(24.0,29.0){} \put(230.0,36.0){} \put(50.0,29.0){} \put(43.0,29.0){} \put(38.0,19.0){} \put(8.0,120.0){tangent at } \put(190.0,110.0){tangent at } \end{overpic}
For the only families of periodic trajectories are and . Since both and converge to as , the values of the action functional on and are both bigger than for sufficiently large. Hence
[TABLE]
for sufficiently large. Here since is independent of (see [BPS, Proposition 4.5.1]) and every -small Hamiltonian has only contractible periodic trajectories, the last equation holds. Now Step 1 follows from [BPS, Lemma 4.7.1 (ii)].
Step 2**.**
If , then . Moreover, the homomorphism
[TABLE]
is an isomorphism whenever .
By (vi), for any the number (resp. ) is the maximum point (resp. the minimum point) of the function defined by
[TABLE]
If is sufficiently large so that , then and hence
[TABLE]
Since , we have
[TABLE]
Hence, by Theorem 3.2, and, by [BPS, Proposition 4.5.1], the monotone homomorphism
[TABLE]
is an isomorphism whenever for and and Step 2 follows from [BPS, Lemma 4.7.1 (ii)].
Step 3**.**
If , then .
\begin{overpic}[width=227.62204pt,clip]{4g.png} \put(113.0,155.0){\line(1,0){2.0}} \put(116.0,104.0){} \put(116.0,150.0){} \put(191.0,104.0){} \put(25.0,104.0){} \put(230.0,111.0){} \put(145.0,104.0){} \put(140.0,180.0){tangent at } \put(-5.0,26.0){tangent at } \put(208.0,125.0){} \put(208.0,70.0){} \end{overpic}
In Figure 4, the lines are tangent to at , where is such that (if such exists). Since , we can ignore all trajectories which belong to . Since both and converge to 0 as , the values of the action functional on and are both less than for sufficiently large. Hence for sufficiently large. Now Step 3 follows from [BPS, Lemma 4.7.1 (i)].
Step 4**.**
If , then . Moreover, the homomorphism
[TABLE]
is an isomorphism for .
As we may ignore trajectories which belong to . By (vii), for any the number (resp. ) is the maximum point (resp. the minimum point) of the function defined by
[TABLE]
Then, by (ii), and hence
[TABLE]
If , then and hence
[TABLE]
Applying Theorem 3.2, for . By [BPS, Proposition 4.5.1], the monotone homomorphism
[TABLE]
is an isomorphism for . Thus Step 4 follows from [BPS, Lemma 4.7.1 (i)]. The proof of Theorem 3.5 is complete. ∎
3.5. Proof of Theorem 3.6
Proof.
If the homomorphism
[TABLE]
is non-zero, then Theorem 3.4, Step 1 and Step 3 in the proof of Theorem 3.5 imply that .
Fix . For simplicity, we assume that . The proof for the case follows the same path as in the case . We choose a sufficiently large so that for all the homomorphisms
[TABLE]
and
[TABLE]
are isomorphisms. By Proposition 2.10, we have the following commutative diagram.
[TABLE]
Hence it is enough to show that the homomorphism is non-zero and surjective.
We put and , where is the evaluation map given by . We define a Morse function by
[TABLE]
for and a Morse function by
[TABLE]
for . Let be a maximal point of with respect to the coordinate and a minimal point with respect to . Then we have .
Let be tubular neighborhoods of of radii , respectively. We extend the functions to functions on by making constant in the direction normal to . Let be smaller tubular neighborhoods of of radii . Moreover, let be -functions satisfying
[TABLE]
and we define bump functions by \rho_{\pm T}(x)=\hat{\rho}_{\pm T}\bigl{(}d(x,C_{\pm T})\bigr{)} where is the distance induced from a metric on .
Let be a symplectic manifold. For two Hamiltonians with compact support, we define a juxtaposition of and by
[TABLE]
where is a smooth function with the properties that for all , , , and vanishes to infinite order at [math] and at (see [Us, Proof of Proposition 3.1] for details). Since vanishes to infinite order at [math] and , is one-periodic in time, that is, that .
For a real number we consider the two juxtapositions and . By [BPS, Proof of Theorem B], we may assume, without loss of generality, that and are one-periodic in time. Now we choose neighborhoods and as in Proposition 2.5. We choose a sufficiently small so that , and . We then have
[TABLE]
Moreover, we note that . These perturbations enable us to compute the Floer homology groups via Poźniak’s theorem (Theorem 3.2) and reveal their behavior when the action interval varies.
Claim 1**.**
For all we have
[TABLE]
Proof.
By Step 2, we have
[TABLE]
By Lemma 3.8, is a Morse–Bott manifold of periodic trajectories for . By Theorem 3.2, the Floer homology coincides with the Morse homology \mathrm{HM}_{2n+1-k}\bigl{(}\mathcal{P}(r_{T};\hat{\alpha}_{\ell}),F_{T}\bigr{)}. Therefore we have
[TABLE]
Thus for any ,
[TABLE]
We choose a positive real number such that
[TABLE]
Since the minimum value of is , we obtain
[TABLE]
\begin{overpic}[width=227.62204pt,clip]{2.png} \put(107.0,84.0){} \put(229.0,69.0){} \put(185.0,176.0){} \end{overpic}
We deform small enough so that every periodic trajectory lying in has action less than (i.e., the tangent line at of slope does not take values greater than at ), where is a small open neighborhood of . We then obtain a Hamiltonian with compact support so that
[TABLE]
[TABLE]
and the Hamiltonian isotopy associated to does not admit periodic trajectories of action greater than and less than .
\begin{overpic}[width=227.62204pt,clip]{3.png} \put(229.0,69.0){} \put(185.0,176.0){} \end{overpic}
We may assume, without loss of generality, that our monotone homotopy contains , and . By (2) in Subsection 2.4, we have the following commutative diagram, whose rows are the short exact sequences for , for and for .
[TABLE]
Here we temporarily omitted in the notation. By Definition 2.6 and Lemma 2.8, note that
[TABLE]
Claim 2**.**
For all we have
[TABLE]
Proof.
By the choice of , we obtain
[TABLE]
Let be a critical point of the restriction . Then each and take values in . We claim that for any we have . Actually, if for some (we may assume, without loss of generality, that ), then we have
[TABLE]
This contradicts the fact that . Hence is homeomorphic to the product of a small open -cell and the -torus . Theorem 3.2 and Claim 1 show that the Floer homology coincides with the Morse homology \mathrm{HM}_{2n+1-k}\bigl{(}\mathcal{P}^{[b,\infty)}(H_{T};\hat{\alpha}_{\ell}),F_{T}\bigr{)} even if the action interval varies. Hence we have
[TABLE]
Claim 3**.**
The homomorphism
[TABLE]
is an isomorphism.
Proof.
We can choose a monotone homotopy defining the map so that does not allow new periodic trajectories of action greater than for all , i.e., is action-regular. Hence Lemma 2.9 shows that is an isomorphism. ∎
Claim 4**.**
The homomorphism
[TABLE]
is an isomorphism.
Proof.
Since the Hamiltonian isotopy associated to does not admit periodic trajectories of action greater than and less than , we have . The exactness of the second row in (6) shows Claim 4. ∎
Claim 5**.**
The homomorphism
[TABLE]
is an isomorphism.
Proof.
A similar observation in the proof of Claim 3 shows Claim 5. ∎
Actually, Step 4 directly shows that . By Claim 3–5 and (7), we deduce that the homomorphism
[TABLE]
is non-zero and surjective if and only if so is the homomorphism
[TABLE]
Therefore, the exactness of the first row in (6) implies that it is suffices, for proving the non-zeroness of , to show that the homomorphism
[TABLE]
is not surjective if .
Claim 6**.**
For all we have
[TABLE]
Proof.
By the choice of , the set is the product . Theorem 3.2 and Claim 1 show that the Floer homology coincides with the Morse homology \mathrm{HM}_{2n+1-k}\bigl{(}\mathcal{P}^{[a,b)}(H_{T};\hat{\alpha}_{\ell}),F_{T}\bigr{)}. Hence we have
[TABLE]
By Poincaré(–Lefschetz) duality, we have and
[TABLE]
Therefore,
[TABLE]
Since
[TABLE]
we conclude that
[TABLE]
We note that and for any . ∎
Thus Claim 6 shows that the homomorphism
[TABLE]
is not surjective if and .
On the other hand, if and , by the fundamental homomorphism theorem and the exactness of the first row in (6), then we have the isomorphism
[TABLE]
Since is induced by the inclusion, we obtain
[TABLE]
Hence the homomorphism
[TABLE]
is surjective if and . Thus the proof of Theorem 3.6 is complete. ∎
4. Annular capacity
To prove Theorem 1.1, we introduce a homological “annular” capacity. This capacity is defined in terms of the homological Biran–Polterovich–Salamon capacity [BPS].
4.1. Homological relative capacity
Let be an open symplectic manifold and a compact subset. For any nontrivial free homotopy class and we define the set
[TABLE]
where is a homomorphism defined in Subsection 2.5.
Definition 4.1** ([BPS, Subsection 4.9]).**
For and we define the homological Biran–Polterovich–Salamon capacity by
[TABLE]
Here we use the convention that and . Let be a closed connected symplectic manifold and a compact subset. Let denote the annulus , and for define its subset . For put
[TABLE]
Definition 4.2**.**
For , , and , we define a homological relative capacity by
[TABLE]
Here is considered as the symplectic manifold equipped with the product symplectic form where .
4.2. Proof of Theorem 1.1
In the case that and , we can compute the capacity directly due to Theorem 3.6.
Proposition 4.3**.**
For any , , and ,
[TABLE]
Proof.
Let . By Theorem 3.6, the homomorphism
[TABLE]
is non-zero if and only if . Hence we have
[TABLE]
for any such that . Therefore, for any we have
[TABLE]
Proposition 4.4 relates the capacities and .
Proposition 4.4** ([BPS, Proposition 4.9.1]).**
Let and be real numbers such that . Let be an integer and a real number. If then every Hamiltonian with compact support on with has a one-periodic trajectory in the homotopy class with the action . In particular,
[TABLE]
Proof of Theorem 1.2.
By Proposition 4.3 and Proposition 4.4, it is enough to show that
[TABLE]
If and , then every Hamiltonian with compact support has a contractible periodic trajectory whose action is zero and hence . Thus we assume that either and , or . We set . For any choose a smooth function with compact support satisfying
[TABLE]
[TABLE]
and
[TABLE]
Now consider the Hamiltonian with compact support given by . Then every periodic trajectory x\colon t\mapsto x(t)=\bigl{(}p(t),q(t)\bigr{)} of in the homotopy class satisfies
[TABLE]
where is such that . Moreover, the action of is . If , then we have and hence
[TABLE]
Here we note that (resp. ) if and only if (resp. ). The above observation contradicts the fact that . Hence there is no one-periodic trajectory of length . We assume that . Then . If , then
[TABLE]
If , then
[TABLE]
Thus there is no periodic trajectory in whose action is at least .
As a conclusion, we obtain that
[TABLE]
for any . It means that
[TABLE]
Finally we prove Theorem 1.1.
Proof of Theorem 1.1.
By [BPS, Proof of Theorem B], we may assume, without loss of generality, that is one-periodic in time. If and , then every Hamiltonian with compact support has infinitely many contractible periodic trajectories whose actions are zero, and hence Theorem 1.1 is proved. Thus we assume that either and , or . According to Theorem 1.2, we have
[TABLE]
It implies that for all there exists such that . Moreover, by Theorem 3.6, if is non-degenerate, then the number of such ’s is at least
[TABLE]
Acknowledgement
The authors would like to express their sincere gratitude to their advisor Professor Takashi Tsuboi for his practical advice and to the referee for very important remarks. They also thank Satoshi Sugiyama for giving them the trigger to study the present topic. They thank Hiroyuki Ishiguro for pointing out many careless mistakes in a previous draft, too. The second author thanks Natsumi Magome for drawing beautiful 3D graphics. This work was supported by IBS-R003-D1 (the first author), JSPS KAKENHI Grant Numbers 25-6631 (the first author), 26-7057 (the second author) and the Program for Leading Graduate Schools, MEXT, Japan. The authors were supported by the Grant-in-Aid for JSPS fellows.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[BPS] P. Biran, L. Polterovich and D. Salamon, Propagation in Hamiltonian dynamics and relative symplectic homology , Duke Math. J. 119 (2003), no. 1, 65–118.
- 2[Ci] K. Cieliebak, Handle attaching in symplectic homology and the chord conjecture , J. Eur. Math. Soc. (JEMS) 4 (2002), 115–142.
- 3[CFH] K. Cieliebak, A. Floer and H. Hofer, Symplectic homology, II. A general construction , Math. Zeit. 218 (1995), 103–122.
- 4[Fl] A. Floer, Symplectic fixed points and holomorphic spheres , Comm. Math. Phys. 120 (1989), 575–611.
- 5[FH] A. Floer and H. Hofer, Symplectic homology, I. Open sets in ℂ n superscript ℂ 𝑛 \mathbb{C}^{n} , Math. Zeit. 215 (1994), 37–88.
- 6[FHS] A. Floer, H. Hofer and D. Salamon, Transversality in elliptic Morse theory for the symplectic action , Duke Math. J. 80 (1995), 251–292.
- 7[FS] U. Frauenfelder and F. Schlenk, Hamiltonian dynamics on convex symplectic manifolds , Israel J. Math. 159 (2007), 1–56.
- 8[Is] H. Ishiguro, Non-contractible orbits for Hamiltonian functions on Riemann surfaces , Ar Xiv e-print 1612.07062 (2016).
