 Research
 Open Access
 Published:
At least three solutions for the Hamiltonian system and reduction method
Journal of Inequalities and Applications volume 2013, Article number: 91 (2013)
Abstract
We investigate the multiplicity of solutions for the Hamiltonian system with some asymptotically linear conditions. We get a theorem which shows the existence of at least three 2πperiodic solutions for the asymptotically linear Hamiltonian system. We obtain this result by the variational reduction method which reduces the infinite dimensional problem to the finite dimensional one. We also use the critical point theory and the variational method.
MSC:35A15, 37K05.
1 Introduction and statement of the main result
Let G(t,z(t)) be a {C}^{2} function defined on {R}^{1}\times {R}^{2n} which is 2πperiodic with respect to the first variable t. In this paper we investigate the number of 2πperiodic solutions of the following Hamiltonian system:
where p,q\in {R}^{n}, z=(p,q). Let J be the standard symplectic structure on {R}^{2n}, i.e.,
where {I}_{n} is the n\times n identity matrix. Then (1.1) can be rewritten as
where \dot{z}=\frac{dz}{dt} and {G}_{z} is the gradient of G. We assume that G\in {C}^{2}({R}^{1}\times {R}^{2n},{R}^{1}) satisfies the following asymptotically linear conditions:
(G1) G(t,z(t))=o({z}^{2}) as z\to 0, G(t,\theta )=0, {G}_{z}(t,\theta )=\theta, where \theta =(0,\dots ,0).
(G2) There exist constants α, β (without loss of generality, we may assume \alpha ,\beta \notin Z) such that
(G3) Let {j}_{1} be an integer within [\alpha ,\beta ] such that
(G4) {lim}_{z\to \mathrm{\infty}}\frac{{G}_{z}(t,z)\cdot z}{{z}^{2}} exists and there exists {j}_{2}={j}_{1}+1 which satisfies
(G5) G is 2πperiodic with respect to t.
We are looking for the weak solutions of (1.1) with conditions (G1)(G5). The 2πperiodic weak solution z=(p,q)\in E of (1.1) satisfies
i.e.,
where E is introduced in Section 2. By Lemma 2.1 in Section 2, the weak solutions of (1.1) coincide with the critical points of the functional
Several authors [1–4] considered the multiplicity of solutions for the Hamiltonian system. Chang proved in [1] that if G\in {C}^{2}({R}^{1}\times {R}^{2n},{R}^{1}) satisfies conditions (G2), (G5) and the following additional conditions:
(G3)′ Let {j}_{0}, {j}_{0}+1, … , and {j}_{1} be all integers within [\alpha ,\beta ] (without loss of generality, we may assume \alpha ,\beta \notin Z) such that {j}_{0}1<\alpha <{j}_{0}<{j}_{1}<\beta <{j}_{1}+1={j}_{2}. Suppose that there exist \gamma >0 and \tau >0 such that {j}_{1}<\gamma <\beta and
(G4)′ {G}_{z}(t,\theta )=\theta and j\in [{j}_{0},{j}_{1})\cap Z such that
then (1.1) has at least two nontrivial 2πperiodic weak solutions. Jung and Choi proved in [2] that if G satisfies the following conditions:
(G1) G:{R}^{2n}\to R is {C}^{1} with G(\theta )=0.
(G2) There exists h\in N such that
(G3) There exists m\in N such that
or
(G4) There exists an integer Γ such that \mathrm{\Gamma}\le \frac{{G}^{\mathrm{\prime}}(z)\cdot z}{{z}^{2}}\le \mathrm{\Gamma}+1,
then (1.1) has at least m weak solutions, which are geometrically distinct and nonconstant.
Our main result is the following:
Theorem 1.1 Assume that G satisfies conditions (G1)(G5). Then system (1.1) has at least three 2πperiodic solutions.
Theorem 1.1 will be proved by the finite dimensional reduction method, the critical point theory and the variational method for the perturbed operator {A}_{\u03f5}. The finite dimensional reduction method combined with the critical point theory and the variational method reduces the critical point results of the functional I(z) on the infinite dimensional space to those of the corresponding functional \tilde{I}(v) on the finite dimensional subspace.
The outline of this paper is organized as follows. In Section 2, we introduce the Hilbert normed space E, show that the corresponding functional I(z) of (1.1) is in {C}^{1}(E,R), Fréchet differentiable and prove the reduction lemma for the perturbed operator {A}_{\u03f5}. In Section 3, we show that the reduced functional \tilde{I}(v) satisfies {(P.S.)}_{c} condition and v=0 is the strict local point of minimum of \tilde{I}(v) and prove Theorem 1.1 by the shape of graph of the reduced functional.
2 The perturbed operator {A}_{\u03f5}
Let {L}^{2}([0,2\pi ],{R}^{2n}) denote the set of 2ntuples of the square integrable 2πperiodic functions and choose z\in {L}^{2}([0,2\pi ],{R}^{2n}). Then it has a Fourier expansion z(t)={\sum}_{k=\mathrm{\infty}}^{k=+\mathrm{\infty}}{a}_{k}{e}^{ikt}, with {a}_{k}=\frac{1}{2\pi}{\int}_{0}^{2\pi}z(t){e}^{ikt}\phantom{\rule{0.2em}{0ex}}dt\in {C}^{2n}, {a}_{k}=\overline{{a}_{k}} and {\sum}_{k\in Z}{{a}_{k}}^{2}<\mathrm{\infty}. Let
with the domain
where ϵ is a positive small number. Then A is a selfadjoint operator. Let \{{M}_{\lambda}\} be the spectral resolution of A, and let
Let
For each u\in {L}^{2}([0,2\pi ],{R}^{2n}), we have the composition
where {u}_{0}\in {L}_{0}, {u}_{+}\in {L}_{+}, {u}_{}\in {L}_{}. According to A, there exists a small number \u03f5>0 such that \u03f5\notin \sigma (A). Let us define the space E as follows:
with the scalar product
and the norm
The space E endowed with this norm is a real Hilbert space continuously embedded in {L}^{2}([0,2\pi ],{R}^{2n}). The scalar product in {L}^{2} naturally extends as the duality pairing between E and {E}^{\mathrm{\prime}}={W}^{\frac{1}{2},2}([0,2\pi ],{R}^{2n}). We note that the operator {(\u03f5+A)}^{1} is a compact linear operator from {L}^{2}([0,2\pi ],{R}^{2n}) to E such that
Let
Let
Then E={E}_{0}\oplus {E}_{+}\oplus {E}_{} and for z\in E, z has the decomposition z={z}_{0}+{z}_{+}+{z}_{}\in E, where
Thus we have
and that {E}_{0}, {E}_{+}, {E}_{} are isomorphic to {L}_{0}, {L}_{+}, {L}_{}, respectively. Let us define the functional f(u) on {L}^{2} as follows:
where {M}_{+}={\int}_{0}^{\mathrm{\infty}}d{M}_{\lambda}, {M}_{}={\int}_{\mathrm{\infty}}^{0}d{M}_{\lambda} and {\psi}_{\u03f5}(z)=\psi (z)+\frac{\u03f5}{2}{\parallel z(t)\parallel}_{{L}^{2}}^{2}, \psi (z)={\int}_{0}^{2\pi}G(t,z(t))\phantom{\rule{0.2em}{0ex}}dt. Let
By G\in {C}^{2} and (G2), \psi (z)={\int}_{0}^{2\pi}G(t,z(t))\in {C}^{2}(E,{R}^{1}). Let
The system (1.1) is equal to
The Euler equation of the functional f(u) is the system
Thus z={z}_{0}+{z}_{+}+{z}_{} is a solution of (2.2) if and only if u={u}_{0}+{u}_{+}+{u}_{} is a critical point of f. System (2.3)(2.5) is reduced to
By (G2),
By (G2), there exists a \gamma >\beta +\u03f5 such that
We note that
Let us set
Now we will prove a reduction lemma which reduces the problem on the infinite dimensional space E to that of the finite dimensional subspace.
Let {z}_{0}\in {E}_{0} be fixed and consider the function h:{E}_{}\times {E}_{+}\to R defined by
The function h has continuous partial Fréchet derivatives {D}_{1}h and {D}_{2}h with respect to its first and second variables given by
for {y}_{1}\in {E}_{} and {y}_{2}\in {E}_{+}, i=1,2. Let v={z}_{0}.
Lemma 2.1 Assume that G satisfies the conditions (G1)(G5).

(i)
For given v\in {E}_{0}, there exists a unique {z}_{}+{z}_{+}\in {C}^{1}({E}_{0},{E}_{}\oplus {E}_{+}) satisfying the equation
{A}_{\u03f5}({z}_{}+{z}_{+})=({P}_{}+{P}_{+}){F}_{\u03f5}(v+{z}_{}+{z}_{+}).(2.11) 
(ii)
There exists {m}_{1}<0 such that if {z}_{} and {y}_{} are in {E}_{} and {z}_{+}\in {E}_{+}, then
({D}_{1}h({z}_{},{z}_{+}){D}_{1}h({y}_{},{z}_{+}))({z}_{}{y}_{})\le {m}_{1}{\parallel {z}_{}{y}_{}\parallel}^{2}.(2.12) 
(iii)
There exists {m}_{2}>0 such that if {z}_{+} and {y}_{+} are in {E}_{+} and {z}_{}\in {E}_{}, then
({D}_{2}h({z}_{},{z}_{+}){D}_{2}h({z}_{},{y}_{+}))({z}_{+}{y}_{+})\ge {m}_{2}{\parallel {z}_{+}{y}_{+}\parallel}^{2}.(2.13) 
(iv)
For given v\in {E}_{0}, if we put the unique solution {z}_{}(v)+{z}_{+}(v) of (2.11) as {z}_{}(v)+{z}_{+}(v)=\theta (v), then \theta (v) is continuous on {E}_{0} and satisfies a uniform Lipschitz condition in {E}_{0} with respect to {L}^{2} norm (also norm {\parallel \cdot \parallel}_{E}) and {{A}_{\u03f5}}^{\frac{1}{2}}{z}_{}(v)\in {C}^{1}({E}_{0},{E}_{}\oplus {E}_{+}), {{A}_{\u03f5}}^{\frac{1}{2}}{z}_{+}(v)\in {C}^{1}({E}_{0},{E}_{}\oplus {E}_{+}). Moreover,
DI(v+\theta (v))(w)=0\phantom{\rule{1em}{0ex}}\mathit{\text{for all}}\phantom{\rule{1em}{0ex}}w\in {E}_{}\oplus {E}_{+}. 
(v)
If \tilde{I}:{E}_{0}\to R is defined by
\tilde{I}(v)=I(v+\theta (v))=I(v+{z}_{}(v)+{z}_{+}(v)),
then \tilde{I} has a continuous Fréchet derivative D\tilde{I} with respect to v, and

(vi)
v\in {E}_{0} is a critical point of \tilde{I} if and only if v+\theta (v)=v+{z}_{}(v)+{z}_{+}(v) is a critical point of I.
Proof (i) Let \delta =\frac{\alpha +\beta}{2}+\u03f5. If {F}_{\u03f5}^{\delta}(\psi )={F}_{\u03f5}(\psi )\delta, then equation (2.11) is equivalent to the equation
The operator {({A}_{\u03f5}\delta )}^{1}({P}_{}+{P}_{+}) is a selfadjoint, compact and linear map from ({P}_{}+{P}_{+}){L}_{2} into itself and its norm is {(min\{{j}_{2}\delta ,{j}_{1}1\delta \})}^{1}. We note that
We claim that the righthand side of (2.14) is a Lipschitz mapping of ({P}_{}+{P}_{+}){L}_{2} into itself with a Lipschitz constant r<1. In fact, let v be a fixed element in {E}_{0} and w=v+{z}_{}+{z}_{+}, y=v+{w}_{}+{w}_{+} be any elements in E. Then we have
Since the operator norm of {{A}_{\u03f5}\delta }^{\frac{1}{2}}({P}_{}+{P}_{+}) is less than or equal to \frac{1}{\sqrt{min\{{j}_{2}\delta ,{j}_{1}1\delta \}}+\u03f5} and
we have
since min\{{j}_{2}\delta ,{j}_{1}1\delta \}+\u03f5>max\{\alpha \delta ,\beta \delta \}+\u03f5. Therefore, by the implicit function theorem, for given v\in {E}_{0}, there exists a unique solution {z}_{}(v)+{z}_{+}(v)\in {E}_{}\oplus {E}_{+} which satisfies (2.15).

(ii)
For all {z}_{}\in {E}_{},
{\parallel {z}_{}\parallel}_{E}^{2}\le ({j}_{1}1){\parallel {w}_{1}\parallel}_{{L}^{2}}^{2}.(2.16)
For all {z}_{+}\in {E}_{+},
If v\in {E}_{0}, {z}_{} and {y}_{} are in {E}_{}, {z}_{+}\in {E}_{+} and z=v+{z}_{}+{z}_{+}, then
Since ({G}_{z}^{\u03f5}({\xi}_{2}){G}_{z}^{\u03f5}({\xi}_{1}))({\xi}_{2}{\xi}_{1})>(\alpha +\u03f5){({\xi}_{2}{\xi}_{1})}^{2} and (2.15) holds, we see that if {z}_{} and {y}_{} are in {E}_{} and {z}_{+}\in {E}_{+}, then
where {m}_{1}=1\frac{\alpha}{{j}_{1}1}<0.

(iii)
Similarly, using the fact that ({G}_{z}^{\u03f5}({\xi}_{2}){G}_{z}^{\u03f5}({\xi}_{1}))({\xi}_{2}{\xi}_{1})<(\beta +\u03f5){({\xi}_{2}{\xi}_{1})}^{2} and (2.17) holds, we see that if {z}_{+} and {y}_{+} are in {E}_{+} and {z}_{}\in {E}_{}, then
({D}_{2}h({z}_{},{z}_{+}){D}_{2}h({z}_{},{y}_{+}))({z}_{+}{y}_{+})\ge {m}_{2}{\parallel {z}_{+}{y}_{+}\parallel}^{2},
where {m}_{2}=1\frac{\beta}{{j}_{1}+1}>0.

(iv)
If \theta (v) denotes the unique ({z}_{}+{z}_{+})(v)\in {E}_{}\oplus {E}_{+} which solves (2.11), then \theta \in {C}^{1}({E}_{0},E). In fact, if v,{v}^{\mathrm{\prime}}\in {E}_{0}, and {p}_{1}=\theta (v), {p}_{2}=\theta ({v}^{\mathrm{\prime}}), then we have
\begin{array}{rcl}{\parallel {p}_{1}{p}_{2}\parallel}_{E}& =& {\parallel {({A}_{\u03f5})}^{1}({P}_{}+{P}_{+})[{F}_{\u03f5}(v+{p}_{1}){F}_{\u03f5}({v}^{\mathrm{\prime}}+{p}_{2})]\parallel}_{E}\\ \le & C{\parallel (v+{p}_{1})({v}^{\mathrm{\prime}}+{p}_{2})\parallel}_{E}\\ \le & C{\parallel (v{v}^{\mathrm{\prime}})({p}_{1}{p}_{2})\parallel}_{E}.\end{array}
Thus we have
Thus θ is continuous. Since {F}_{\u03f5}\in {C}^{1}(E,E), \theta \in {C}^{1}({E}_{0},E). Since dim{L}_{0} is finite and all topologies on {L}_{0} are equivalent, we have
Let v\in {E}_{0}. If q\in {E}_{}\oplus {E}_{+}, then from (2.11) we have
Since {\int}_{0}^{2\pi}{A}_{\u03f5}v\cdot q=0, we have
for all q\in {E}_{}\oplus {E}_{+}.

(v)
Since the functional I has a continuous Fréchet derivative DI, \tilde{I} has a continuous Fréchet derivative D\tilde{I} with respect to v.

(vi)
Suppose that there exists v\in {E}_{0} such that D\tilde{I}(v)=0. From D\tilde{I}(v)(h)=DI(v+\theta (v))(h) for all v,h\in {E}_{0}, DI(v+\theta (v))(h)=0 for all h\in {E}_{0}. Since DI(v+\theta (v))(w) for all w\in {E}_{}\oplus {E}_{+}, it follows that DI(v+\theta (v))=0. Thus v+\theta (v) is a solution of (1.1). Conversely if u is a solution of (1.1) and v={P}_{0}u, then D\tilde{I}(v)=0. □
3 Proof of Theorem 1.1
Lemma 3.1 Assume that G satisfies the conditions (G1)(G5). Then \tilde{I}(v) is bounded below and satisfies (P.S.) condition.
Proof Let v\in {E}_{0}. By the finite dimensional reduction,
where \theta (v)={\theta}_{}(v)+{\theta}_{+}(v), v\in {E}_{0}, {\theta}_{}(v)\in {E}_{}, {\theta}_{+}(v)\in {E}_{+}, {G}^{\u03f5}(t,v(t)+\theta (v(t)))=G(t,v(t)+\theta (v(t)))+\u03f5{(v(t)+\theta (v(t)))}^{2}. Let w=v+{\theta}_{}(v). Then we have
Moreover, we have
By (G4), we have chosen a number γ such that {j}_{1}<\gamma <{d}_{z}^{2}G(t,\mathrm{\infty})<\beta. Thus we have
Thus \tilde{I}(v) is bounded from below and satisfies (P.S.) condition. □
Lemma 3.2 Assume that G satisfies conditions (G1)(G5). Then v=0 is a strict local point of minimum of \tilde{I}(v) with \tilde{I}(0)=0.
Proof
where
Thus we have
Thus v=0 is a strict local point of minimum of \tilde{I}(v). Since \theta (0)=0, \tilde{I}(0)=0. □
Proof of Theorem 1.1 By Lemma 2.1(v), \tilde{I}(v) is continuous and Fréchet differentiable in {E}_{0}. By Lemma 3.1, \tilde{I}(v) is bounded above, satisfies the (P.S.) condition and \tilde{I}(v)\to \mathrm{\infty} as {\parallel v\parallel}_{E}\to \mathrm{\infty}. By Lemma 3.2, v=0 is a strict local point of minimum of \tilde{I}(v) with a critical value \tilde{I}(0)=0. We note that {max}_{v\in {E}_{0}}\tilde{I}(v)>0 is another critical value of \tilde{I}. By the shape of the graph of the functional \tilde{I} on the onedimensional subspace {E}_{0}, there exists the third critical point of \tilde{I}(v). Thus (1.1) has at least three solutions, one of which is a trivial solution u=v+\theta (v)=0+0=0. □
References
Chang KC: Infinite Dimensional Morse Theory and Multiple Solution Problems. Birkhäuser, Boston; 1993.
Jung T, Choi QH: On the number of the periodic solutions of the nonlinear Hamiltonian system. Nonlinear Anal. TMA 2009, 71(12):e1100e1108. 10.1016/j.na.2009.01.095
Jung T, Choi QH: Periodic solutions for the nonlinear Hamiltonian systems. Korean J. Math. 2009, 17(3):331–340.
Rabinowitz PH CBMS Regional Conf. Ser. Math. 65. In Minimax Methods in Critical Point Theory with Applications to Differential Equations. Am. Math. Soc., Providence; 1986.
Acknowledgements
The authors appreciate very much the referees for their kind corrections. This work (Tacksun Jung) was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (KRF20110026920).
Author information
Authors and Affiliations
Corresponding author
Additional information
Competing interests
The authors did not provide this information.
Authors’ contributions
The authors did not provide this information.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 2.0 International License (https://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
About this article
Cite this article
Jung, T., Choi, QH. At least three solutions for the Hamiltonian system and reduction method. J Inequal Appl 2013, 91 (2013). https://doi.org/10.1186/1029242X201391
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/1029242X201391
Keywords
 Hamiltonian system
 asymptotical linearity
 variational reduction method
 critical point theory
 variational method
 {(P.S.)}_{c} condition