- Research
- Open Access

# Periodic solutions of second order difference systems with even potential

- Huafeng Xiao
^{1, 2}Email author

**2014**:215

https://doi.org/10.1186/1687-1847-2014-215

© Xiao; licensee Springer. 2014

**Received: **24 March 2014

**Accepted: **11 July 2014

**Published: **4 August 2014

## Abstract

In this article, we study the multiplicity and minimality of periodic solutions to difference systems, which are globally superquadratic or subquadratic. A new technique is given to detect the minimal period of periodic solutions to autonomous systems. Some weaker conditions than a globally subquadratic condition are obtained to guarantee the existence of periodic solutions with prescribed minimal period to autonomous system.

## Keywords

- difference systems
- periodic solution
- minimal period
- superquadratic
- subquadratic

## 1 Introduction

On one hand, difference equations have been widely used to describe real-life situations in computer science, economics, neural network, ecology, cybernetics, *etc.* On the other hand, they are also natural consequences of the discretization of differential equations. So it is worthwhile to explore this topic. Many tools are used to study the existence of all kinds of solutions to discrete systems. A powerful tool is critical point theory, which firstly was introduced by Guo and Yu in 2003 to study the existence of periodic solutions to a difference system (*cf.* [1–3]). Since then, the study of discrete dynamic systems has got considerable development. We refer to boundary value problems (*cf.* [4, 5]), periodic solutions (*cf.* [6, 7]), homoclinic orbits (*cf.* [8, 9]), and heteroclinic orbits (*cf.* [10, 11]).

As is well known, the minimal periodic problem is an important but difficult problem. As far as the author knows, the study of solutions with a prescribed minimal period began in 2004. In that year, by estimating the energy of a variational functional, Yu *et al.* (*cf.* [12]) studied the existence of subharmonic solutions with a prescribed minimal period to a discrete forced pendulum equation. More recently, by making use of the Clark dual method, Bin (*cf.* [13]), Long (*cf.* [14]) and Long *et al.* (*cf.* [15]) studied the existence and multiplicity of periodic solutions with a prescribed minimal period to difference systems. Because of the lack of methods, results in this field are scarce.

It is well known that the Nehari manifold has been introduced by Nehari in 1960 (*cf.* [16, 17]) and developed by Szulkin and Weth in 2010 (*cf.* [18]). It has been used widely to study the existence of ground state solutions to ordinary differential systems, partial differential systems, and difference systems (*cf.* [19–21]). A ground state solution is a solution which possesses the minimal energy of all solutions. Since such a minimality can be used to prove the minimal periods of solutions, it has been used to study the existence of period solutions with prescribed minimal period to ordinary differential equations (*cf.* [22–25]).

Motivated by the above references, in this paper, one attempts to make use of a Nehari manifold to study the multiplicity and minimality of periodic solutions to difference systems. When the systems are globally superquadratic or subquadratic, by restricting our discussion to the Nehari manifold, firstly, we study the existence of multiple periodic solutions to nonautonomous systems; secondly, we study the existence of periodic solutions with a prescribed minimal period to autonomous systems. Also, some subquadratic conditions, which are weaker than the globally subquadratic condition, are obtained to guarantee the existence of periodic solutions with prescribed minimal period to autonomous system.

For convenience, we denote by ℕ, ℤ, ℝ the sets of all natural numbers, integers, and real numbers, respectively. For $a,b\in \mathbb{Z}$ with $a\le b$, define $Z[a,b]=\{a,a+1,\dots ,b\}$. For $m\in \mathbb{N}$, denote by ${\mathbb{R}}^{m}$ the Euclidean space with the usual inner product $(\cdot ,\cdot )$ and norm $|\cdot |$.

where ${x}_{n}\in {\mathbb{R}}^{m}$, △ is a difference operator defined by $\mathrm{\u25b3}{x}_{n}={x}_{n+1}-{x}_{n}$ and ${\mathrm{\u25b3}}^{2}{x}_{n}=\mathrm{\u25b3}(\mathrm{\u25b3}{x}_{n})$.

Assume that:

*z*,

*i.e.*,

where ${f}^{\prime}(t,\cdot )$ denotes the Hermite matrix of $F(t,\cdot )$;

**Remark 1** If *F* satisfies (F1), (F2), and (F3), without loss of generality, we can assume that $F(t,0)=0$ for all $t\in \mathbb{R}$. Otherwise, there exists a twice continuously differentiable function $g(t)$ such that $F(t,0)=g(t)$. Let $\stackrel{\u02c6}{F}(t,z)=F(t,z)-g(t)$. Then $\stackrel{\u02c6}{F}$ also satisfies (F1), (F2), and (F3) with *F* replaced by $\stackrel{\u02c6}{F}$. Similarly, if *F* satisfies (F1), (F2), and (F4), we can assume that $F(t,0)=0$ for all $t\in \mathbb{R}$.

For a nonautonomous system, we have the following two results.

**Theorem 1** *Suppose that* *F* *satisfies* (F1), (F2), *and* (F3). *Then* (1) *has at least* $m[(T-1)/2]-1$ *distinct pairs of different* *T*-*periodic solutions*.

**Theorem 2** *Suppose that* *F* *satisfies* (F1), (F2), *and* (F4). *Then* (1) *has at least* $m[(T-1)/2]-1$ *distinct pairs of different* *T*-*periodic solutions*.

Assume that *F* satisfies

*i.e.*,

where ${f}^{\prime}(\cdot )$ denotes the Hermite matrix of $F(\cdot )$;

**Remark 2** If *F* satisfies (F5) and (F6) (respectively, (F5) and (F7)), without loss of generality, we can assume that $F(0)=0$.

For an autonomous system, we have the following two results.

**Theorem 3** *Suppose that* *F* *satisfies* (F5) *and* (F6). *Then*, *for any integer* $P>1$, (2) *possesses at least a periodic solution with minimal period* *P*.

**Theorem 4** *Suppose that* *F* *satisfies* (F5) *and* (F7). *Then*, *for any integer* $P>1$, (2) *possesses at least a periodic solution with minimal period* *P*.

Now, one weakens the conditions (F5) and (F7). Assume that *F* satisfies the following conditions:

*i.e.*,

**Remark 3**Assume that

*F*satisfies (F8) and (F9). Then there exist some positive constants ${M}_{2}$ and ${M}_{3}$ such that, for all $x\in {\mathbb{R}}^{m}$,

**Theorem 5** *Suppose that* *F* *satisfies* (F8), (F9), *and* (F10). *Then*, *for any integer* $P>1$, (2)* possesses at least a periodic solution with minimal period* *P*.

The rest of this paper is divided into two parts. In Section 2, we study the multiplicity of periodic solutions to a nonautonomous system. In Section 3, firstly, we study the existence of periodic solutions with a prescribed minimal period to an autonomous system, which is globally superquadratic or subquadratic; secondly, some conditions, which are weaker than the globally subquadratic condition, are given to guarantee the existence of periodic solutions with prescribed minimal period to autonomous system.

## 2 Nonautonomous difference system

### 2.1 Preparations

It is easy to check that ${E}_{T}$ is linearly homeomorphic to ${\mathbb{R}}^{mT}$, which can also be identified with ${\mathbb{R}}^{mT}$.

where $x={({x}_{1}^{\tau},{x}_{2}^{\tau},\dots ,{x}_{T}^{\tau})}^{\tau}$, $A={P}^{-1}CP$, $C=diag{(B,B,\dots ,B)}_{mT\times mT}$. Here *τ* denotes the transposition of a vector and *B*, *P* are matrices defined in [2].

*J*can be viewed as a twice continuously differentiable functional defined on a finite-dimensional Hilbert space. Thus, for any $x,y,z\in {E}_{T}$, one has

It is easy to check that the critical points of *J* are *T*-periodic solutions of (1).

*B*can be given by

*A*has

*T*eigenvalues ${\lambda}_{0},{\lambda}_{1},\dots ,{\lambda}_{T-1}$, each of multiplicity

*m*. Clearly, 0 is an eigenvalue of

*A*and set $\eta ={({\eta}_{1},{\eta}_{2},\dots ,{\eta}_{m})}^{\tau}$ the eigenvector associated to 0. If

*T*is even, ${\lambda}_{T/2}=4$ is also an eigenvalue of

*A*and set ${\eta}^{\prime}={({\eta}_{1}^{\prime},{\eta}_{2}^{\prime},\dots ,{\eta}_{m}^{\prime})}^{\tau}$, the eigenvector associated to ${\lambda}_{T/2}$. For any $k\in Z[1,\lfloor (T-1)/2\rfloor ]$, where $\lfloor \cdot \rfloor $ denotes the upper integral function, set

*A*corresponding to eigenvalue ${\lambda}_{k}$. Denote by ${W}_{i}$ ($i=1,2,3,4$) the spaces defined as follows:

*T*is even. Also, we have

where *a*, *b*, ${a}_{k}$, ${b}_{k}$ ($k=1,2,\dots ,\lfloor (T-1)/2\rfloor $) are all constant vectors of ${\mathbb{R}}^{m}$.

*T*is even or odd, ${x}_{n}$ has a Fourier expansion

Problem (3) has $\lfloor (T-1)/2\rfloor $ eigenvalues, each of them of multiplicity *m*. The eigenvalues of (3) are ${\lambda}_{k}=4{sin}^{2}(k\pi /T)$, where $k=1,2,\dots ,\lfloor (T-1)/2\rfloor $. Obviously, the smallest eigenvalue is ${\lambda}_{1}$ and the largest eigenvalue is ${\lambda}_{\lfloor (T-1)/2\rfloor}$, which is denoted by ${\lambda}_{max}$.

Now we give a useful lemma. Since the proof is standard, we omit it.

**Lemma 1** *If* *x* *is a critical point of* *J* *restricted on* ${\tilde{E}}_{T}$, *then* *x* *is also a critical point of* *J* *on *${E}_{T}$.

At the end of this subsection, two useful lemmas are given.

Suppose that *H* be a real Banach space and *M* is a closed symmetric ${C}^{1}$-submanifold of *H* with $0\notin M$. Suppose that $\varphi \in {C}^{1}(M,\mathbb{R})$.

**Lemma 2** [26]

*Suppose that*

*ϕ*

*is even and bounded below*.

*Define*

*where* ${\mathrm{\Gamma}}_{j}:=\{A\subset M:A=-A,A\subset H\setminus \{0\},A\mathit{\text{is compact and}}\gamma (A)\ge j\}$. *If* ${\mathrm{\Gamma}}_{k}\ne \mathrm{\varnothing}$ *for some* $k\ge 1$ *and if* *ϕ* *satisfies the* (*PS*)_{
c
} *for all* $c={c}_{j}$, $j=1,2,\dots ,k$, *then* *ϕ* *has at least* *k* *distinct pairs of critical points*.

**Lemma 3** [18]

*If* *ϕ* *is bounded below and satisfies the* (*PS*) *condition*, *then* $c:={inf}_{M}\varphi $ *is attained and is a critical value of* *ϕ*.

### 2.2 Superquadratic case

In this subsection, we consider the existence of multiple periodic solutions of (1), where *F* satisfies (F1), (F2), and (F3). Arguing similarly to [25], we can prove the following lemma.

**Lemma 4**

*If*

*F*

*satisfies*(F1), (F2),

*and*(F3),

*then*$0<(1+\alpha )F(t,x)\le (f(t,x),x)$

*for all*$x\in {\mathbb{R}}^{m}\setminus \{0\}$.

*Also*,

*where* $\overline{M}={max}_{\{t\mid 0\le t\le T\}}{max}_{|x|=1}F(t,x)$ *and* $\underline{M}={min}_{\{t\mid 0\le t\le T\}}{min}_{|x|=1}F(t,x)$.

By a similar argument to reference [22], one can prove the following lemma.

**Lemma 5** ℳ *is* ${C}^{1}$-*manifold with dimension* $m\lfloor (T-1)/2\rfloor -1$. *If* ${x}_{0}$ *is a critical point of* *J* *restricted on* ℳ, *then* ${x}_{0}$ *is also a critical point of* *J* *restricted on* ${\tilde{E}}_{T}$.

By Lemmas 1 and 5, to search for periodic solutions of (1), we need find critical points of *J* restricted on ℳ.

**Lemma 6** *Fixing* $x\in {\tilde{E}}_{T}\setminus \{0\}$, *there exists a unique* ${t}_{x}>0$ *such that* ${t}_{x}x\in \mathcal{M}$.

*Proof* Fixing $x\in {\tilde{E}}_{T}\setminus \{0\}$, define ${\phi}_{x}(t):=J(tx)$ for $t\in (0,+\mathrm{\infty})$. Obviously, ${\phi}_{x}\in {C}^{2}$. It is easy to check that ${\phi}_{x}^{\prime}(t)=0$ if and only if $tx\in \mathcal{M}$.

Claim: There exists a unique ${t}_{x}>0$ satisfying (4).

Thus, there exists a ${t}_{3}\in ({t}_{1},{t}_{2})$ satisfying ${\phi}_{x}({t}_{3})={min}_{{t}_{1}\le {t}_{3}\le {t}_{2}}{\phi}_{x}(t)$. Consequently, ${\phi}_{x}^{\prime}({t}_{3})=0$ and ${\phi}_{x}^{\prime \prime}({t}_{3})\ge 0$. However, by a similar argument to (5), ${\phi}_{x}^{\prime \prime}({t}_{3})<0$, which is a contradiction. Thus ${t}_{x}$ is unique. □

Since ${\phi}_{x}^{\prime}({t}_{x})=0$ and ${\phi}_{x}^{\prime \prime}({t}_{x})<0$, then ${\phi}_{x}({t}_{x})={max}_{t\in (0,\mathrm{\infty})}{\phi}_{x}(t)$. Hence $J(tx)$ restricted on $(0,\mathrm{\infty})$ attains its maximum at ${t}_{x}$.

**Lemma 7** *J* *satisfies the* (*PS*) *condition on* ℳ.

*Proof*Assume that $\{{x}^{k}\}\subset \mathcal{M}$ is a (PS) sequence for

*J*. Then there exists a ${M}_{4}\ge 0$ such that $|J({x}^{k})|\le {M}_{4}$ for all $m\in \mathbb{N}$ and ${J}^{\prime}({x}^{k})\to 0$ as $k\to \mathrm{\infty}$. Set

*F*is continuous, there exists a ${M}_{5}>0$ such that

Thus $\underline{M}{C}_{1,1+\alpha}^{-(1+\alpha )}{\parallel {x}^{k}\parallel}^{1+\alpha}-1/2{\lambda}_{max}{\parallel {x}^{k}\parallel}^{2}\le {M}_{4}+T{M}_{5}$. Since $\alpha >1$, $\{\parallel {x}^{k}\parallel \}$ is bounded. Since ${\tilde{E}}_{T}$ is a finite-dimensional space, there exists a convergent subsequence of $\{{x}^{k}\}$. □

It follows from Lemma 6 that *g* is a bijection whose inverse ${g}^{-1}$ is given by ${g}^{-1}(x)=x/\parallel x\parallel $.

**Lemma 8** ${C}_{1}={inf}_{x\in \mathcal{M}}J(x)>0$.

*Proof*For any $x\in \mathcal{M}$, since $J(x)={sup}_{t\in (0,+\mathrm{\infty})}J(tx)={sup}_{t\in (0,+\mathrm{\infty})}J(tx/\parallel x\parallel )$, it follows that

To prove that ${C}_{1}>0$, one only need to show that ${inf}_{x\in {S}^{1}}{sup}_{t\in (0,+\mathrm{\infty})}J(tx)>0$.

*x*, such that

□

*Proof of Theorem 1* Because of (F2), ℳ is a closed symmetric manifold and $0\notin \mathcal{M}$. It follows from Lemma 5 that ℳ is a ${C}^{1}$ manifold with dimension $m\lfloor (T-1)/2\rfloor -1$. By Lemmas 7 and 8, *J* is bounded from below and satisfies the (PS) condition. It is easy to check that *J* is even. Then Lemma 2 implies that *J* has at least $m\lfloor (T-1)/2\rfloor -1$ distinct pairs of critical points. Thus (1) possesses at least $m\lfloor (T-1)/2\rfloor -1$ distinct pairs of periodic solutions. □

### 2.3 Subquadratic case

In this subsection, we consider the multiplicity of periodic solutions of (1), where *F* satisfies (F1), (F2), and (F4). In order to prove Theorem 2, we consider the functional ${J}_{1}=-J$ and the Nehari manifold ${\mathcal{M}}_{1}$ is defined as ${\mathcal{M}}_{1}:=\{x\in {\tilde{E}}_{T}\setminus \{0\}\mid \u3008{J}_{1}^{\prime}(x),x\u3009=0\}$. Since the technique of the proof of Theorem 2 is just the same as that of Theorem 1, where *J* and ℳ are replaced by ${J}_{1}$ and ${\mathcal{M}}_{1}$, we omit it here.

## 3 Autonomous difference equations

### 3.1 Variational framework

In this section, we consider the existence of periodic solutions with any prescribed minimal period of (2), that is, for any given positive integer *P*, we search for periodic solutions of (2) with minimal period *P*.

Then ${E}_{P}$ is a Hilbert space, which is homeomorphic to ${\mathbb{R}}^{mP}$. Denote by ${\parallel \cdot \parallel}_{P}$ the norm introduced by ${\u3008\cdot ,\cdot \u3009}_{P}$.

where #*D* is a matrix of order *mp*. Then *I* is twice continuously differentiable. The critical points of *I* are *P*-periodic solutions of (2).

has $\lfloor (P-1)/2\rfloor $ solutions, each of them of multiplicity *m*. The smallest eigenvalue is ${\overline{\lambda}}_{1}=4{sin}^{2}(\pi /P)$ and the largest eigenvalue is ${\overline{\lambda}}_{\lfloor (P-1)/2\rfloor}$, which is denoted by ${\overline{\lambda}}_{max}$.

**Lemma 9** *If* *x* *is a critical point of* *I* *restricted on* ${\tilde{E}}_{P}$, *then* *x* *is also a critical point of* *I* *on the whole space* ${E}_{P}$.

### 3.2 Superquadratic case

*F*satisfies (F5) and (F6). The main target of this subsection is to prove Theorem 3. A similar argument to Section 2.2, one can check the following facts:

- (i)$0<(1+\alpha )F(x)\le (f(x),x)$ for all $x\in {\mathbb{R}}^{m}\setminus \{0\}$. Also,$F(x)\le {\overline{M}}^{\prime}{|x|}^{\alpha +1},\phantom{\rule{1em}{0ex}}\text{when}|x|\le 1,\phantom{\rule{2em}{0ex}}F(x)\ge {\underline{M}}^{\prime}{|x|}^{\alpha +1}\phantom{\rule{1em}{0ex}}\text{when}|x|\ge 1,$

- (ii)
$\overline{\mathcal{M}}$ is a ${C}^{1}$ manifold;

- (iii)
critical points of

*I*restricted on $\overline{\mathcal{M}}$ are also critical points of*I*restricted on ${\tilde{E}}_{P}$; - (iv)
for any $x\in {\tilde{E}}_{P}\setminus \{0\}$, there exists a unique ${t}_{x}$ such that ${t}_{x}x\in \overline{\mathcal{M}}$ and $I({t}_{x}x)={max}_{t\in (0,\mathrm{\infty})}I(tx)$;

- (v)
*I*restricted on $\overline{\mathcal{M}}$ satisfies the (PS) condition; - (vi)
*I*restricted on $\overline{\mathcal{M}}$ is bounded from below and ${C}_{2}={inf}_{x\in \overline{\mathcal{M}}}I(x)>0$.

*Proof of Theorem 3* It follows from Lemma 3 that ${C}_{2}$ is a critical value. Denote by $\tilde{x}$ the critical point of *I* corresponding to ${C}_{2}$. Then $\tilde{x}$ is a *P*-periodic solution of (2). It is easy to check that $\tilde{x}$ is a nonconstant periodic solution.

Claim: $\tilde{x}$ has *P* as its minimal period.

This contradicts ${r}_{\tilde{y}}\tilde{y}\in \overline{\mathcal{M}}$. Hence $\tilde{x}$ has *P* as its minimal period. □

### 3.3 Subquadratic case (I)

*F*satisfies (F5) and (F7). A similar argument to Section 2.2, we can prove the following facts:

- (I)$0<(f(x),x)\le (1+{\beta}^{\prime})F(x)$ for all $x\in {\mathbb{R}}^{m}\setminus \{0\}$. Also,$F(x)\ge {\underline{M}}^{\prime}{|x|}^{{\beta}^{\prime}+1},\phantom{\rule{1em}{0ex}}\text{when}|x|\le 1,\phantom{\rule{2em}{0ex}}F(x)\le {\overline{M}}^{\prime}{|x|}^{{\beta}^{\prime}+1},\phantom{\rule{1em}{0ex}}\text{when}|x|\ge 1,$

- (II)
$\overline{\mathcal{M}}$ is a ${C}^{1}$ manifold;

- (III)
critical points of

*I*restricted on $\overline{\mathcal{M}}$ are critical points of*I*restricted on ${\tilde{E}}_{P}$; - (IV)
for any $x\in {\tilde{E}}_{P}\setminus \{0\}$, there exists a unique ${t}_{x}$ such that ${t}_{x}x\in \overline{\mathcal{M}}$ and ${I}_{1}({t}_{x}x)={min}_{t\in (0,\mathrm{\infty})}{I}_{1}(tx)$;

- (V)
*I*restricted on $\overline{\mathcal{M}}$ satisfies the (PS) condition.

**Lemma 10** *Assume that* *F* *satisfies* (F5) *and* (F7). *Then* *I* *restricted on* ${\tilde{E}}_{P}$ *is bounded from below*.

*Proof*It follows from Fact (I) that $F(x)\le {\overline{M}}^{\prime}{|x|}^{{\beta}^{\prime}+1}$ when $|x|\ge 1$. Since

*F*is continuous, there exists a ${M}_{6}>0$ such that $F(x)\le {M}_{6}$ when $|x|\le 1$. Consequently, $F(x)\le {\overline{M}}^{\prime}{|x|}^{{\beta}^{\prime}+1}+{M}_{6}$ for all $x\in {\mathbb{R}}^{m}$. Thus

Since ${\beta}^{\prime}\in (0,1)$, there exists a constant ${C}_{3}$ such that ${I}_{1}(x)\ge {C}_{3}$ for all $x\in {\tilde{E}}_{P}$. □

Because of Lemma 10, *I* restricted on $\overline{\mathcal{M}}$ is bounded from below. Denote by ${C}_{4}={inf}_{x\in \overline{\mathcal{M}}}I(x)$. A similar argument to Lemma 8, we can prove that ${C}_{4}<0$.

*Proof of Theorem 4* It follows from Lemma 3 that ${C}_{4}$ is a critical value. Denote by $\stackrel{\u02c6}{x}$ the critical point of *I* corresponding to ${C}_{4}$. Then $\stackrel{\u02c6}{x}$ is a nonconstant *P*-periodic solution of (2). By a similar discussion to the proof of Theorem 3, we can prove that $\stackrel{\u02c6}{x}$ has *P* as its minimal period. □

### 3.4 Subquadratic case (II)

In this subsection, we assume that *F* satisfies (F8), (F9), and (F10). By a similar argument to Lemma 10, we have the following lemma.

**Lemma 11** *Assume that* *F* *satisfies* (F8), (F9), *and* (F10). *Then* *I* *restricted on* ${\tilde{E}}_{P}$ *is bounded from below*.

**Lemma 12** ${C}_{5}={inf}_{x\in {\tilde{E}}_{P}}I(x)<0$.

*Proof*By (F10), $F(z)\ge {M}_{1}{|z|}^{{\gamma}^{\prime}}$ if $|x|<{G}_{2}$. Then, for any $x\in {\tilde{E}}_{P}$ with ${\parallel x\parallel}_{P}<{G}_{2}$, we have

Hence ${C}_{5}={inf}_{x\in {\tilde{E}}_{P}}I(x)<0$. □

**Lemma 13** *I* *restricted on* ${\tilde{E}}_{P}$ *satisfies the* (*PS*) *condition*.

*Proof*Assume that $\{{x}^{k}\}\subset {\tilde{E}}_{P}$ is a (PS) sequence for

*I*. Then there exists a ${M}_{7}\ge 0$ such that $|I({x}^{k})|\le {M}_{7}$ for all $k\in \mathbb{N}$ and ${I}^{\prime}({x}^{k})\to 0$ as $k\to \mathrm{\infty}$. Then

Thus $1/2{\overline{\lambda}}_{1}{\parallel {x}^{k}\parallel}^{2}-{M}_{2}{C}_{2,\gamma}^{\gamma}{\parallel {x}^{k}\parallel}^{\gamma}\le {M}_{3}P+{M}_{7}$. Since $\gamma <2$, it follows that $\{{x}_{k}\}$ is bounded. Since ${\tilde{E}}_{T}$ is a finite-dimensional space, there exists a convergent subsequence of $\{{x}^{k}\}$. □

*Proof of Theorem 5* Let $\{{x}^{k}\}\subset {\tilde{E}}_{P}$ be a minimal sequence of *I*, that is, $I({x}^{k})\to {C}_{5}$ as $k\to \mathrm{\infty}$. By the Ekeland variational principle, ${I}^{\prime}({x}^{k})\to 0$ as $k\to \mathrm{\infty}$. Since *I* satisfies the (PS) condition, ${C}_{5}$ is a critical point of *I* restricted on ${\tilde{E}}_{P}$. Denote by $\overline{x}$ the critical point of *I* corresponding to critical value ${C}_{5}$. Then $\overline{x}$ is a periodic solution of (2). It is easy to check that $\overline{x}$ is a nonconstant *P*-periodic solution. By a similar discussion to the proof of Theorem 3, we can prove that $\overline{x}$ has *P* as its minimal period. □

## Declarations

### Acknowledgements

This project is supported by NSFC (No. 11126063).

## Authors’ Affiliations

## References

- Guo ZM, Yu JS: Existence of periodic and subharmonic solutions for second-order superlinear difference equations.
*Sci. China Ser. A*2003, 46: 506-515. 10.1007/BF02884022MathSciNetView ArticleMATHGoogle Scholar - Guo ZM, Yu JS: Existence of periodic and subharmonic solutions to subquadratic second-order equations.
*J. Lond. Math. Soc.*2003, 68: 419-430. 10.1112/S0024610703004563MathSciNetView ArticleMATHGoogle Scholar - Guo ZM, Yu JS: Periodic and subharmonic solutions for superquadratic discrete Hamiltonian systems.
*Nonlinear Anal.*2003, 55: 969-983. 10.1016/j.na.2003.07.019MathSciNetView ArticleMATHGoogle Scholar - Tian Y, Du ZJ, Ge WG: Existence results for discrete Sturm-Liouville problem via variational methods.
*J. Differ. Equ. Appl.*2007, 13: 467-478. 10.1080/10236190601086451MathSciNetView ArticleMATHGoogle Scholar - Yu JS, Guo ZM: Boundary value problems of discrete generalized Emden-Foeler equation.
*Sci. China Ser. A*2006, 49: 1303-1314. 10.1007/s11425-006-1999-zMathSciNetView ArticleMATHGoogle Scholar - Mawhin J: Periodic solutions of second order nonlinear difference systems with
*ϕ*-Laplacian: a variational approach.*Nonlinear Anal.*2012, 75: 4672-4687. 10.1016/j.na.2011.11.018MathSciNetView ArticleMATHGoogle Scholar - Zhang H, Li ZX: Periodic solutions for a class of nonlinear discrete Hamiltonian systems via critical point theory.
*J. Differ. Equ. Appl.*2010, 16: 1381-1391. 10.1080/10236190902821689View ArticleMathSciNetMATHGoogle Scholar - Lin XY, Tang XH: Existence of infinitely many homoclinic orbits in discrete Hamiltonian systems.
*J. Math. Anal. Appl.*2011, 373: 59-72. 10.1016/j.jmaa.2010.06.008MathSciNetView ArticleMATHGoogle Scholar - Ma MJ, Guo ZM: Homoclinic orbits for second order self-adjoint difference equations.
*J. Math. Anal. Appl.*2006, 323: 513-521. 10.1016/j.jmaa.2005.10.049MathSciNetView ArticleMATHGoogle Scholar - Xiao HF, Yu JS: Heteroclinic orbits for a discrete pendulum equation.
*J. Differ. Equ. Appl.*2011, 17: 1267-1280. 10.1080/10236190903167991MathSciNetView ArticleMATHGoogle Scholar - Zhang H, Li ZX: Heteroclinic orbits and heteroclinic chains for a discrete Hamiltonian system.
*Sci. China Ser. A*2010, 53: 1555-1564. 10.1007/s11425-010-4009-4View ArticleMathSciNetMATHGoogle Scholar - Yu JS, Long YH, Guo ZM: Subharmonic solutions with prescribed minimal period of a discrete forced pendulum equation.
*J. Dyn. Differ. Equ.*2004, 16: 575-586. 10.1007/s10884-004-4292-2MathSciNetView ArticleMATHGoogle Scholar - Bin HH: Subharmonics with minimal periods for convex discrete Hamiltonian systems.
*Abstr. Appl. Anal.*2013., 2013: Article ID 508247Google Scholar - Long YH: Applications of Clark duality to periodic solutions with minimal period for discrete Hamiltonian systems.
*J. Math. Anal. Appl.*2008, 342: 726-741. 10.1016/j.jmaa.2007.10.075MathSciNetView ArticleMATHGoogle Scholar - Long YH, Guo ZM, Shi HP: Multiple periodic solutions with minimal period for second order discrete system.
*Acta Appl. Math.*2010, 110: 181-191. 10.1007/s10440-008-9396-yMathSciNetView ArticleMATHGoogle Scholar - Nehari Z: On a class of nonlinear second-order differential equations.
*Trans. Am. Math. Soc.*1960, 95: 101-123. 10.1090/S0002-9947-1960-0111898-8MathSciNetView ArticleMATHGoogle Scholar - Nehari Z: Characteristic values associated with a class of nonlinear second order differential equations.
*Acta Math.*1961, 105: 141-175. 10.1007/BF02559588MathSciNetView ArticleMATHGoogle Scholar - Szulkin A, Weth T: The method of Nehari manifold. In
*Handbook of Nonconvex Analysis and Applications*. International Press, Somerville; 2010.Google Scholar - Pankov A: Periodic nonlinear Schrödinger equation with application to photonic crystals.
*Milan J. Math.*2005, 73: 259-287. 10.1007/s00032-005-0047-8MathSciNetView ArticleMATHGoogle Scholar - Pankov A: On decay of solutions to nonlinear Schrödinger equations.
*Proc. Am. Math. Soc.*2008, 136: 2565-2570. 10.1090/S0002-9939-08-09484-7MathSciNetView ArticleMATHGoogle Scholar - Szulkin A, Weth T: Ground state solutions for some indefinite variational problems.
*J. Funct. Anal.*2009, 257: 3802-3822. 10.1016/j.jfa.2009.09.013MathSciNetView ArticleMATHGoogle Scholar - Ambrosetti A, Mancini G: Solutions of minimal period for a class of convex Hamiltonian systems.
*Math. Ann.*1981, 255: 405-421. 10.1007/BF01450713MathSciNetView ArticleMATHGoogle Scholar - Deng ST: Minimal periodic solutions of a class of Hamiltonian systems.
*Acta Math. Sin.*1984, 27: 664-675. (in Chinese)MATHGoogle Scholar - Magrone, P: Critical point methods for indefinite nonlinear Elliptic equations and Hamiltonian systems. PhD thesis, Universitá Degli Studi Di Roma ‘Tor Vergata’ (2001)Google Scholar
- Xiao YM: Periodic solutions with prescribed minimal period for the second order Hamiltonian systems with even potentials.
*Acta Math. Sin. Engl. Ser.*2010, 26: 825-830. 10.1007/s10114-009-8305-2MathSciNetView ArticleMATHGoogle Scholar - Szulkin A:Ljusternik-Schnirelmann theory on ${C}^{1}$-manifold.
*Ann. Inst. Henri Poincaré, Anal. Non Linéaire*1988, 5: 119-139.MathSciNetMATHGoogle Scholar

## Copyright

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.