Abstract
In this paper, we study the multiplicity of periodic solutions for the second order Hamiltonian systems \(\ddot{u}+\nabla F(t,u)=0\) with the boundary condition \(u(0)-u(T)=\dot{u}(0)-\dot{u}(T)=0\), where the potential F is either subquadratic k(t)-concave or subquadratic \(\mu (t)\)-convex. Based on the reduction method and a three-critical-point theorem due to Brezis and Nirenberg, we obtain the multiplicity results, which complement and sharply improve some related results in the literature.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction and Main Results
Consider the second order Hamiltonian systems
where \(T>0\) and \(F:[0,T]\times {\mathbb {R}}^N\rightarrow {\mathbb {R}}\) satisfies the following assumption:
-
(A)
F(t, x) is measurable in t for every \(x\in {\mathbb {R}}^N\) and continuously differentiable in x for a.e. \(t\in [0,T]\), and there exist \(a\in C({\mathbb {R}}^+,{\mathbb {R}}^+)\), \(b\in L^1(0,T;{\mathbb {R}}^+)\) such that
$$\begin{aligned} |F(t,x)|\le a(|x|)b(t),\ \ \ \ \ \ |\nabla F(t,x)|\le a(|x|)b(t) \end{aligned}$$for all \(x\in {\mathbb {R}}^N\) and a.e. \(t\in [0,T]\).
With the aid of variational methods, the existence and multiplicity of periodic solutions of problem (1.1) have been extensively investigated in the literature during the past two decades, see [3,4,5, 7,8,9,10,11,12,13,14,15,16, 18,19,26] and references therein. Many solvability conditions are obtained, such as: the convexity conditions (see [14, 17, 19] and the references therein); the subquadratic conditions (including the sublinear nonlinearity case, see [18, 19] the references therein); the superquadratic conditions (see [6, 26] and the references therein) and the asymptotically linear conditions (see [20, 26, 29] and the references therein).
Lazer, Landesman and Meyers in [8] consider a minimax theorem for a \(C^2\) functional \(\varphi \) defined on a Hilbert space H. They prove the existence of a critical point characterized by \(\displaystyle {\varphi (u_0)=\max _{v\in V} \min _{w\in W} \varphi (v+w)}\), where V, W are subspaces of H, V is finite-dimensional and \(H=V\bigoplus W\). This result has been generalized by Castro and Lazer [4], supposing a weaker condition that \(\varphi (v)\rightarrow -\infty \) as \(\Vert v\Vert \rightarrow \infty \), \(v\in V\). Manasevich [12] extend the result to infinite-dimensional cases by using a global inversion theorem. See also Bates and Ekeland [2] and Manasevich [13] for related results. Recently, by means of the reduction method, a perturbation argument and the least action principle, Tang and Wu [19] greatly improve the result of Lazer et al. [8] in three aspects: requiring the spaces being reflexive Banach space instead of Hilbert space; requiring the functionals being \(C^1\) instead of \(C^2\); and using much weaker convexity of the functionals. As a main application, they successively studied the existence of periodic solutions of problem (1.1) with subquadratic convex potential, with subquadratic \(\mu (t)\)-convex potential and with subquadratic k(t)-concave potential, which unifies and significantly improves the recent results in [14, 24, 28]. Here we say that the potential F is \(\mu (t)\)-convex, if there exists \(\mu \in L^1(0,T;{\mathbb {R}})\) such that \(F(t,x)-\frac{1}{2}\mu (t)|x|^2\) in convex in x for a.e. \(t\in [0,T]\). It is worth pointing out that the reduction technique, which has been motivated in earlier papers by Amann [1] and Thews [21] to discuss the existence and multiplicity of solutions for nonlinear equations, is rather powerful. For its applications on semilinear elliptic equations, we refer the readers to [24].
In the present paper, based on the reduction method and some abstract critical point theorem, i.e., the three-critical-point theorem proposed by Brezis and Nirenberg, we shall study the multiplicity of periodic solutions of (1.1) with subquadratic k(t)-concave or subquadratic \(\mu (t)\)-convex potential, which complements the results mentioned above and improves the corresponding results in Wu [24] and Zhao and Wu [27].
1.1 The Subquadratic k(t)-Concave Case
For the subquadratic k(t)-concave case, we make the following hypotheses:
- (\(A_1\)):
-
There exists \(k\in L^1(0,T;{\mathbb {R}}^+)\) with \(\int _0^Tk(t)dt<12/T\) such that \(-F(t,x)+\frac{1}{2}k(t)|x|^2\) is convex in x for a.e. \(t\in [0,T]\).
- \((A_2)\):
-
\({\int _0^TF(t,x)dt\rightarrow +\infty }\) as \(|x|\rightarrow +\infty \), \(x\in {\mathbb {R}}^N\).
- \((A_3)\):
-
There exists \(\delta >0\) such that \(F(t,x)-F(t,0)\le 0\) for all \(|x|\le \delta \hbox { and a.e. }t\in [0,T]\).
- \((A_4)\):
-
There exists \(k\in L^1(0,T;{\mathbb {R}}^+)\) with \(k(t)\le \omega ^2\) for a.e. \(t\in [0,T]\) and
$$\begin{aligned} \text {meas}\left\{ t\in [0,T]:k(t)<\omega ^2\right\} >0, \end{aligned}$$such that \(-F(t,x)+\frac{1}{2}k(t)|x|^2\) is convex in x for a.e. \(t\in [0,T]\), where \(\omega =2\pi \backslash T\).
Theorem 1.1
Assume that the potential F satisfies assumptions (A) and \((A_1)\)-\((A_3)\). Then problem (1.1) possesses at least three distinct solutions.
Remark 1.1
Theorem 1.1 generalizes [27, Theorem 1], where the existence of one nonzero solution is obtained under the assumptions (A), \((A_2)\) and
- \((A_1')\):
-
There exists \(k\in L^1(0,T;{\mathbb {R}}^+)\) with \(0<\int _0^Tk(t)dt<12/T\) such that \(|\nabla F(t,x_1)-\nabla F(t,x_2)|\le k(t)|x_1-x_2| \) for all \(x_1\), \(x_2\in {\mathbb {R}}^N\) and a.e. \(t\in [0,T]\).
- \((A_3')\):
-
For a.e. \(t\in [0,T]\), \(\nabla F(t,0)=0\), and there exists \(\delta >0\) such that \(F(t,x)-F(t,0)<0 \) for all \(x\in {\mathbb {R}}^N\) with \(0<|x|\le \delta \) and for a.e. \(t\in [0,T]\).
Since, by condition \((A_1')\), we have
for all \(x,y\in {\mathbb {R}}^N\) and a.e. \(t\in [0,T]\), it follows that \(-F(t,x)+\frac{1}{2}k(t)|x|^2\) is convex in x for a.e. \(t\in [0,T]\). Hence \((A_1')\) implies \((A_1)\). In addition, the condition \(\nabla F(t,0)=0\) for a.e. \(t\in [0,T]\) (see assumption \((A_3')\)) is deleted and our conclusion is better. There are functions F satisfying Theorem 1.1 but not satisfying the results in [14, 24, 27, 28]. For example, let
where \(\alpha \in L^\infty (0,T;{\mathbb {R}})\) with
and \(\alpha ^\pm (t)\)=max\(\left\{ \pm \alpha (t),0\right\} \). Then F satisfies Theorem 1.1 with \(k=\alpha ^+\). But it does not satisfy the corresponding results in [14, 24, 27, 28], for \(\int _0^T|\alpha (t)|dt>12/T\) and F is not convex in x for \(t\in [0,T]\) with \(\alpha ^-(t)>0\).
Theorem 1.2
The conclusion of Theorem 1.1 holds if we replace (\(A_1\)) by (\(A_4\)).
Remark 1.2
Theorem 1.2 complements [19, Theorem 5.1], where the existence of one solution was obtained under the assumptions (A), (\(A_2\)) and \((A_4)\). There are functions F satisfying Theorem 1.2 but not satisfying Theorem 1.1. For example, let
Then \(\alpha (t)=\frac{4}{3T}\omega ^2\left( \frac{3T}{4}-t\right) \) for \(t\in [0,T]\),
and
for \(x\in {\mathbb {R}}^N\) and a.e. \(t\in [0,T]\). It is easy to check that F satisfies the conditions of Theorem 1.2 with \(k=\alpha ^+\). However, since
it does not satisfy condition \((A_1)\) and hence not satisfy Theorem 1.1.
1.2 The Subquadratic \(\mu (t)\)-Convex Case
For the subquadratic \(\mu (t)\)-convex case, we make the following hypotheses:
- \((A_5)\):
-
There exists \(\mu \in L^1(0,T;{\mathbb {R}})\) with \(\int _0^T\mu (t)dt>0\) such that \(F(t,x)-\frac{1}{2}\mu (t)|x|^2\) is convex in x for a.e. \(t\in [0,T]\).
- \((A_6)\):
-
There exist \(\alpha \in L^1(0,T;{\mathbb {R}}^+)\) with \(\int _0^T\alpha (t)dt<12/T\) and \(\gamma \in L^1(0,T;{\mathbb {R}}^+)\) such that
$$\begin{aligned} F(t,x)\le \frac{1}{2}\alpha (t)|x|^2+\gamma (t),\ \ \ \ \ \ \ \ \ \forall x\in {\mathbb {R}}^N \text { and a.e. }t\in [0,T]. \end{aligned}$$(1.2) - \((A_7)\):
-
There exist \(\delta >0\) and an integer \(k\ge 1\) such that
$$\begin{aligned} \frac{1}{2}k^2\omega ^2|x|^2\le F(t,x)-F(t,0) \le \frac{1}{2}(k+1)^2\omega ^2|x|^2 \end{aligned}$$for all \(|x|\le \delta \) and a.e. \(t\in [0,T]\).
- \((A_8)\):
-
There exists \(\gamma \in L^1(0,T;{\mathbb {R}}^+)\) such that \(F(t,x)\le \frac{1}{2}\omega ^2|x|^2+\gamma (t) \) for all \(x\in {\mathbb {R}}^N\) and a.e. \(t\in [0,T]\), and \(\hbox {meas}\left\{ t\in [0,T]:F(t,x)-\frac{1}{2}\omega ^2|x|^2\rightarrow -\infty \hbox { as} |x|\rightarrow \infty \right\} >0. \)
- \((A_9)\):
-
There exists \(\alpha \in L^\infty (0,T;{\mathbb {R}}^+)\) with \(\alpha (t)\le \omega ^2\) for a.e. \(t\in [0,T]\) and
$$\begin{aligned} \text {meas}\left\{ t\in [0,T]:\alpha (t)<\omega ^2\right\} >0, \end{aligned}$$such that \( \displaystyle {\limsup _{|x|\rightarrow \infty }|x|^{-2}F(t,x)}\le \frac{1}{2}\alpha (t)\) \(\text {uniformly for a.e. }t\in [0,T]\).
Theorem 1.3
Assume that assumptions (A) and \((A_5)\)-\((A_7)\) are satisfied. Then problem (1.1) possesses at least three distinct solutions.
Remark 1.3
Theorem 1.3 extends [24, Theorem 2.2]; there Wu consider problem (1.1) under the hypothesis (A), \((A_5)\) and
- \((A_6')\):
-
There exist f, \(g\in L^1(0,T;{\mathbb {R}}^+)\) with \(\int _0^Tf(t)dt<12/T\) such that \(|\nabla F(t,x)|\le f(t)|x|+g(t)\) for all \(x\in {\mathbb {R}}^N \text { and a.e. }t\in [0,T]\).
- \((A_7')\):
-
There exist \(\delta >0\) and an integer \(k\ge 1\) such that \(\frac{1}{2}k^2\omega ^2|x|^2\le F(t,x) \le \frac{1}{2}(k+1)^2\omega ^2|x|^2 \) for all \(|x|\le \delta \) and a.e. \(t\in [0,T]\).
We point out that \((A_6')\) is stronger than \((A_6)\). This is because by (A) and (\(A_6'\)), we deduce that
which is just (1.2) with
where \(|\cdot |_1\) denotes the usual norm of \(L^1(0,T)\). Moreover, \((A_7')\) implies that \(F(t,0)=0\) for a.e. \(t\in [0,T]\), which gives \((A_7)\) directly. Hence Theorem 1.3 implies [24, Theorem2.2]. There are functions F satisfying Theorem 1.3 but not satisfying [24, Theorem 2.2]. For example, let
where \(b\in L^\infty (0,T;{\mathbb {R}})\) with \(k^2\omega ^2\le b(t)\le (k+1)^2\omega ^2\) \((k\ge 1)\) for a.e. \(t\in [0,T]\), and \(\mu \in L^\infty (0,T;{\mathbb {R}})\) with \(\mu (t)\le \omega ^2\) for a.e. \(t\in [0,T]\), \(\frac{6}{T}\le \int _0^T\mu ^-(t)dt<\int _0^T\mu ^+(t)dt\le \frac{12}{T}, \) and \(\mu ^\pm =max\left\{ \pm \mu (t),0\right\} \). Then one has
where \(p(t)=3(b(t)-\mu (t))/4\in L^\infty (0,T;{\mathbb {R}}^+)\) and \(c_1=2(k+1)^2\omega ^2\), and hence
which is just (1.2) with
Thus F satisfies all the conditions of Theorem 1.3. But it does not satisfy the corresponding conditions of [24, Theorem 2.2], since \(\int _0^T|\mu (t)|dt>12/T\) and (\(A_6'\)) does not hold.
Theorem 1.4
The conclusion of Theorem 1.3 holds if we replace \((A_6)\) by \((A_8)\).
Remark 1.4
There are functions F satisfying Theorem 1.4 but not satisfying Theorem 1.3 and [24, Theorem 2.2]. For example, let
where \(b\in L^\infty (0,T;{\mathbb {R}})\) with \(\omega ^2\le b\le 4\omega ^2\) for a.e. \(t\in [0,T]\), and \(\mu \in L^\infty (0,T;{\mathbb {R}})\) with \(\mu (t)\le \omega ^2\) for a.e. \(t\in [0,T]\), \(\int _0^T\mu (t)dt>0\), meas\(\left\{ t\in [0,T]:\mu ^+(t)<\omega ^2\right\} >0\) and \(b(t)-\mu (t)\le 4(\omega ^2-\mu ^+(t))\) for a.e. \(t\in [0,T]\). Hence we obtain
where \(p(t)=3(b(t)-\mu (t))/4\in L^\infty (0,T;{\mathbb {R}}^+)\) and \(c_1=8\omega ^2\). Then one has
for all \(x\in {\mathbb {R}}^N\) and a.e. \(t\in [0,T]\). Consequently, F satisfies all the assumptions of Theorem 1.4. But in the case \(\int _0^T\mu ^+(t)dt>12/T\), it does not satisfy the conditions of Theorem 1.3 and [24, Theorem 2.2].
Theorem 1.5
The conclusion of Theorem 1.3 holds if we replace \((A_6)\) by \((A_9)\).
Remark 1.5
Theorem 1.5 is a complement of Theorem 4.2 in [19], where the existence of one solution of (1.1) was proved under conditions (A), \((A_5)\) and \((A_9)\). There are functions F satisfying Theorem 1.5 but not satisfying the results mentioned above. For example, let
where \(b\in L^\infty (0,T;{\mathbb {R}}^+)\) with \(k^2\omega ^2\le b(t)\le (k+1)^2\omega ^2\) \((k\ge 2)\) for a.e. \(t\in [0,T]\), and \(\mu \in L^\infty (0,T;{\mathbb {R}})\) with \(\mu (t)\le \omega ^2\) for a.e. \(t\in [0,T]\), \(\int _0^T\mu (t)dt>0\) and
Then F satisfies Theorem 1.5 with \(\alpha =\mu ^+(t)\). However, in the case
and \(\int _0^T\mu ^+(t)dt> 12/T\), it does not satisfy the conditions of Theorems 1.3, 1.4 and [24, Theorem 2.2].
The paper is organized as follows. Section 2 is devoted to some related preliminaries. In Sect. 3, we show that the reduction functional \(\psi \) is coercive and satisfies the (PS) condition. Then we apply the three-critical-point theorem to prove the theorems.
2 Preliminaries
Under assumption (A), the energy functional associated to problem (1.1) given by
is continuously differentiable and weakly lower semi-continuous on \(H_T^1\), where
is a Hilbert space with the norm defined by
Furthermore, we have
for all \(u,v\in H_T^1\). It is well known that the weak solutions of problem (1.1) correspond to the critical points of \(\varphi \) (see [14]).
In view of [14, Proposition 1.3], for \(u\in \widetilde{H}_T^1=\left\{ u\in H_T^1:\int _0^Tu(t)dt=0\right\} \), we have
and
which implies that
for all \(u\in H_T^1\) and some \(C>0\), where \(\Vert u\Vert _\infty =\max _{t\in [0,T]}|u(t)|\).
Lemma 2.1
If \(\alpha \in L^\infty (0,T;{\mathbb {R}}^+)\) such that \(\alpha (t)\le \omega ^2\) for a.e. \(t\in [0,T]\) and
then there exists \(a<1\) such that
Proof
We prove this assertion by contradiction. In fact, if not, there exists a sequence \((u_n)\subset \widetilde{H}_T^1\) such that
which implies that \(u_n\ne 0\) for all n. By the homogeneity of the above inequality, we may assume that \(\int _0^T|\dot{u}_n|^2dt=1\) and
It follows from the weak compactness of the unit ball of \(\widetilde{H}_T^1\) that there exists a subsequence, still denoted by \((u_n)\), such that
Combining this with (2.3), we obtain
and then,
Hence we obtain
and
which yields that \(u=a\cos \omega t+b\sin \omega t\), a, \(b\in {\mathbb {R}}^N\), \(u\ne 0\) and \(u=0\) on a positive measure subset. This contradicts the fact that \(u=a\cos \omega t+b\sin \omega t\) only has finite zeros if \(u\ne 0\). \(\square \)
To end this section, we state the reduction method developed by Tang and Wu [19, Lemma 2.1] and the three-critical-point theorem proposed by Brezis and Nirenberg (see [3, Theorem4]).
Proposition 2.1
(see [19]) Suppose that V is a reflexive Banach space and W is a Banach space, \(\varphi \in C^1(V\times W,{\mathbb {R}})\). Assume that there exists \(\mu >0\) such that
for all \(v_1\), \(v_2\in V\) and \(w\in W\). Then there exists a map \(\theta \in C(W,V)\) such that \(\theta (w)\) is the unique maxmum of \(\varphi (\cdot ,w)\) for all \(w\in W\) and the functional \(\psi \), given by
is continuously differentiable and
Moreover, \((\theta (w),w)\) is a critical point of \(\varphi \) if and only if w is a critical point of \(\psi \).
Remark 2.1
When V and W are Hilbert spaces, Proposition 2.1 is a corollary of Amann [1, Theorem 2.3].
Proposition 2.2
(Brezis and Nirenberg [3]) Let X be a Banach space with a direct sum decomposition \(X=X_1\bigoplus X_2\) with dim\(X_2<+\infty \) and let \(\psi \) be a \(C^1\)-functional on X with \(\psi (0)=0\), satisfying the (PS) condition. Assume that, for some \(\delta _0>0\),
and
Assume also that \(\psi \) is bounded from below and \(\inf _{u\in X}\psi (u)<0\). Then \(\psi \) has at least two nonzero critical points.
3 Proofs of the Theorems
First, we consider the subquadratic k(t)-concave case. Let \(\varphi \) be the functional introduced in (2.1) and decompose the Hilbert space \(H_T^1\) as \(H_T^1=V\bigoplus W\) with
Proof of Theorem 1.1
The proof will be divided into several steps.
-
(1)
By condition (\(A_1\)), since
$$\begin{aligned} (\nabla (-F(t,x))-\nabla (-F(t,y)),x-y)\ge -k(t)|x-y|^2,\ \ \ \ \ \ \forall x,y\in {\mathbb {R}}^N, \end{aligned}$$we obtain, for \(v_1\), \(v_2\in \widetilde{H}_T^1\) and \(x\in {\mathbb {R}}^N\),
$$\begin{aligned}&\langle D_1\varphi (v_1,x)-D_1\varphi (v_2,x),v_1-v_2\rangle \\&= -\int _0^T|\dot{v}_1-\dot{v}_2|^2dt+\int _0^T(\nabla F(t,v_1+x)-\nabla F(t,v_2+x),v_1-v_2)dt\\&\le -\int _0^T|\dot{v}_1-\dot{v}_2|^2dt+\int _0^Tk(t)|v_1-v_2|^2dt\\&\le -\int _0^T|\dot{v}_1-\dot{v}_2|^2dt+\int _0^Tk(t)dt\cdot \Vert v_1-v_2\Vert _\infty ^2\\&\le -\left( 1-\frac{T}{12}\int _0^Tk(t)dt\right) |\dot{v}_1-\dot{v}_2|_2^2, \end{aligned}$$that is, \(\varphi \) satisfies (2.4) with \(\mu =1-(T/12)\int _0^Tk(t)dt\). Applying Proposition 2.1, we obtain a continuous map \(\theta :{\mathbb {R}}^N\rightarrow \widetilde{H}_T^1\) and a \(C^1\)-functional \(\psi :{\mathbb {R}}^N\rightarrow {\mathbb {R}}\) such that
$$\begin{aligned} \psi (x)=\varphi (\theta (x)+x)=\sup _{v\in \widetilde{H}_T^1}\varphi (v+x). \end{aligned}$$It suffices to find two nonzero critical points of \(\psi \).
-
(2)
It follows from the definition of \(\psi \) and condition (\(A_2\)) that
$$\begin{aligned} \psi (x)\ge \varphi (x)=\int _0^T[F(t,x)-F(t,0)]dt\rightarrow +\infty \ \ \ \ \ \ \ \ \text {as }|x|\rightarrow \infty . \end{aligned}$$Hence \(\psi \) is coercive on \({\mathbb {R}}^N\), and then \(\psi \) is bounded from below and satisfies the (PS) condition, i.e., \((u_n)\subset W\) possesses a convergent subsequence whenever \(\left\{ \psi (u_n)\right\} \) is bounded and \(\psi '(u_n)\rightarrow 0\) as \(n\rightarrow \infty \).
-
(3)
Set \(W_1=\left\{ 0\right\} \) and \(W_2={\mathbb {R}}^N\). The continuity of \(\theta \) implies that there exists \(\delta _0>0\) small enough such that
$$\begin{aligned} \Vert \theta (x)+x\Vert _\infty \le C\Vert \theta (x)+x\Vert \le \delta \end{aligned}$$for all \(x\in W_2\) with \(|x|\le \delta _0\). Thus, using (\(A_3\)), we obtain
$$\begin{aligned} \psi (x)=\varphi (\theta (x)+x)\le \int _0^T[F(t,\theta (x)+x)-F(t,0)]dt\le 0 \end{aligned}$$(3.1)for all \(x\in W_2\) with \(|x|\le \delta _0\). On the other hand, noting
$$\begin{aligned} \psi (0)\ge \varphi (0)=0, \end{aligned}$$we have
$$\begin{aligned} \psi (w)\ge 0,\ \ \ \ \ \ \ \ \forall w\in W_1 \text { with }\Vert w\Vert \le \delta _0. \end{aligned}$$This, jointly with (3.1), shows that \(\varphi \) has a local linking at 0 with respect to \((W_1,W_2)\).
-
(4)
If \(\displaystyle {\inf _{x\in {\mathbb {R}}^N}\psi (x)=0}\), then all \(x\in W_2\) with \(|x|\le \delta _0\) are minimums of \(\psi \), which implies that \(\psi \) has infinitely many critical points. If \(\displaystyle {\inf _{x\in {\mathbb {R}}^N}\psi (x)<0}\), Proposition 2.2 implies that \(\psi \) has at least two nontrivial critical points. In any case, we can find two nonzero critical points \(x_1\), \(x_2\) of \(\psi \). Consequently, \(u_1=\theta (x_1)+x_1\) and \(u_2=\theta (x_2)+x_2\) are two nontrivial solutions of problem (1.1). This completes the proof.
\(\square \)
Proof of Theorem 1.2
The proof is identical to that of Theorem 1.1 except that now (2.4) follows from condition \((A_4)\). Indeed, using Lemma 2.1, we obtain
for all \(v_1\), \(v_2\in \widetilde{H}_T^1\) and \(x\in {\mathbb {R}}^N\). Therefore, (2.4) holds with \(\mu =1-a\). \(\square \)
Next we consider the subquadratic \(\mu (t)\)-convex case. Let \(\Phi =-\varphi \) and decompose the Hilbert space \(H_T^1\) as \(H_T^1=V\bigoplus W\) with
Proof of Theorem 1.3
We divide the proof into several steps.
-
(1)
It follows from \((A_5)\) that
$$\begin{aligned} (\nabla F(t,x)-\nabla F(t,y),x-y)\ge \mu (t)|x-y|^2,\ \ \ \ \ \ \forall x,y\in {\mathbb {R}}^N, \end{aligned}$$and hence,
$$\begin{aligned} \langle D_1\Phi (x_1,w)-D_1\Phi (x_2,w),x_1-x_2\rangle&=-\int _0^T(\nabla F(t,x_1+w)-\nabla F(t,x_2+w),x_1-x_2)dt\nonumber \\&\le -\int _0^T\mu (t)|x_1-x_2|^2dt\nonumber \\&\le -{\frac{1}{T}}\int _0^T\mu (t)dt\Vert x_1-x_2\Vert ^2, \end{aligned}$$(3.2)for all \(x_1\), \(x_2\in {\mathbb {R}}^N\) and \(w\in \widetilde{H}_T^1\). Therefore, (2.4) holds with \(\mu =T^{-1}\int _0^T\mu (t)dt\). In view of Proposition 2.1, we obtain a continuous map \(\theta :\widetilde{H}_T^1\rightarrow {\mathbb {R}}^N\) and a \(C^1\)-functional \(\psi :\widetilde{H}_T^1\rightarrow {\mathbb {R}}\) such that
$$\begin{aligned} \psi (w)=\Phi (\theta (w)+w)=\sup _{x\in {\mathbb {R}}^N}\Phi (x+w). \end{aligned}$$It suffices to find two nonzero critical points of \(\psi \).
-
(2)
Since the reduced functional \(\psi \) is defined on an infinite dimensional subspace, it is difficult to verify the coerciveness and (PS) condition for \(\psi \). To overcome this difficulty, we need to study the properties of \(\theta \). We claim that the functional \(\theta \) obtained above is bounded, i.e., it maps bounded sets into bounded sets. Indeed, note that \(D_1\Phi (\theta (w),w)=0\). Setting \(x_1=\theta (w)\) and \(x_2=0\) in (3.2), we obtain
$$\begin{aligned} {\frac{1}{T}}\int _0^T\mu (t)dt\Vert \theta (w)\Vert ^2&\le -\langle D_1\Phi (\theta (w),w)-D_1\Phi (0,w),\theta (w)\rangle \\&=\langle D_1\Phi (0,w),\theta (w)\rangle \\&=-\int _0^T(\nabla F(t,w),\theta (w))dt\\&\le \max _{s\in [0,C\Vert w\Vert ]}a(s)\int _0^Tb(t)dt\cdot C\Vert \theta (w)\Vert \end{aligned}$$by assumption (A) and (2.2). Hence the desired result follows from the last inequality. Now letting \((w_n)\subset \widetilde{H}_T^1\) be a bounded sequence such that \(\psi '(w_n)\rightarrow 0\) as \(n\rightarrow \infty \), we show that \((w_n)\) has a convergent subsequence. Without loss of generality, we can assume that
$$\begin{aligned}&w_n\rightharpoonup w\ \ \ \ \ \ \ \text {in }\widetilde{H}_T^1,\\&w_n\rightarrow w\ \ \ \ \ \ \ \text {in } C(0,T;{\mathbb {R}}^N). \end{aligned}$$Since \((w_n)\) is bounded, then \(\left\{ \theta (w_n)\right\} \) and hence \(\left\{ \theta (w_n)+w_n\right\} \) are also bounded in \(H_T^1\). Hence, taking
$$\begin{aligned} a_1=\max _{s\in [0,C(\Vert \theta (w_n)+w_n\Vert )]}a(s),\ \ \ \ \ a_2=\max _{s\in [0,C(\Vert \theta (w)+w\Vert )]}a(s), \end{aligned}$$we have
$$\begin{aligned}&|\dot{w}_n-\dot{w}|_2^2\\&\quad =\langle \psi '(w_n)-\psi '(w),w_n-w\rangle +\int _0^T(\nabla F(t,\theta (w_n)+w_n)-\nabla F(t,\theta (w)+w),w_n-w)dt\\&\quad \le \Vert \psi '(w_n)\Vert \Vert w_n-w\Vert -\langle \psi '(w),w_n-w\rangle +\left( a_1\int _0^Tb(t)dt+ a_2\int _0^Tb(t)dt\right) \Vert w_n-w\Vert _\infty \\&\quad \rightarrow 0 \end{aligned}$$because of assumption (A) and (2.2). The equivalence of the \(L^2\)-norm for \(\dot{w}\) and the \(H_T^1\)-norm on \(\widetilde{H}_T^1\) implies that
$$\begin{aligned} w_n\rightarrow w\ \ \ \ \ \ \text {in }\widetilde{H}_T^1. \end{aligned}$$Hence the (PS) condition holds. Moreover, it follows from \((A_6)\) and Sobolev’s inequality that
$$\begin{aligned} \Phi (w)&=\frac{1}{2}\int _0^T|\dot{w}|^2dt-\int _0^T[F(t,w)-F(t,0)]dt\\&\ge \frac{1}{2}\int _0^T|\dot{w}|^2dt-\frac{1}{2}\int _0^T\alpha (t)|w|^2dt-\int _0^T\gamma (t)dt+\int _0^TF(t,0)dt\\&\ge \frac{1}{2}\int _0^T|\dot{w}|^2dt-\frac{1}{2}\int _0^T\alpha (t)dt\cdot \Vert w\Vert _\infty ^2-\int _0^T\gamma (t)dt+\int _0^TF(t,0)dt\\&\ge \frac{1}{2}\left( 1-\frac{T}{12}\int _0^T\alpha (t)dt\right) |\dot{w}|_2^2-\int _0^T\gamma (t)dt+\int _0^TF(t,0)dt \end{aligned}$$for all \(w\in \widetilde{H}_T^1\), which implies that
$$\begin{aligned} \Phi (w)\rightarrow +\infty \ \ \ \ \ \ \ \text {as }\Vert w\Vert \rightarrow \infty ,\ \ w\in \widetilde{H}_T^1. \end{aligned}$$(3.3)This, jointly with the definition of \(\psi \), shows that
$$\begin{aligned} \psi (w)\rightarrow +\infty \ \ \ \ \ \ \ \ \ \text {as}\ \ \Vert w\Vert \rightarrow \infty ,\ \ w\in \widetilde{H}_T^1. \end{aligned}$$So \(\psi \) is bounded from below.
-
(3)
Letting
$$\begin{aligned} W_2=\left\{ \sum _{j=1}^k(a_j\cos j\omega t+b_j \sin j\omega t)\left| a_j,b_j\in {\mathbb {R}}^N, j=1,2,\cdots ,k\right. \right\} \end{aligned}$$and \(W_1=({\mathbb {R}}^N\bigoplus W_2)^\bot \). Then \(\widetilde{H}_T^1=W_1\bigoplus W_2\), and the reduction functional \(\psi \) has a local linking at 0 with respect to \((W_1,W_2)\). In fact, condition \((A_7)\) and (2.2) imply that
$$\begin{aligned} \psi (w)&=\sup _{x\in {\mathbb {R}}^N}\Phi (x+w)\\&\ge \Phi (w)\\&=\frac{1}{2}\int _0^T|\dot{w}|^2dt-\int _0^T[F(t,w)-F(t,0)]dt\\&\ge \frac{1}{2}\int _0^T|\dot{w}|^2dt-\frac{(k+1)^2\omega ^2}{2}\int _0^T|w|^2dt\\&\ge 0 \end{aligned}$$for all \(w\in W_1\) with \(\Vert w\Vert \le \delta /C\). Furthermore, by the continuity of \(\theta \), the map \(w\longmapsto \theta (w)+w\) from \(\widetilde{H}_T^1\) to \(H_T^1\) is continuous. Hence, using (2.2), there exists a constant \(\delta _0\in (0,\delta /C)\) such that
$$\begin{aligned} \Vert \theta (w)+w\Vert _\infty \le C\Vert \theta (w)+w\Vert \le \delta \end{aligned}$$for \(w\in \widetilde{H}_T^1\) with \(\Vert w\Vert \le \delta _0\), so that, using \((A_7)\), we obtain
$$\begin{aligned} \psi (w)&=\Phi (\theta (w)+w)\nonumber \\&= \frac{1}{2}\int _0^T|\dot{w}|^2dt-\int _0^T[F(t,\theta (w)+w)-F(t,0)]dt\nonumber \\&\le \frac{1}{2}\int _0^T|\dot{w}|^2dt-\frac{1}{2}k^2\omega ^2\int _0^T|\theta (w)+w|^2dt\nonumber \\&\le 0 \end{aligned}$$(3.4)for \(w\in W_2\) with \(\Vert w\Vert \le \delta _0\). Thus, \(\psi \) has a local linking at 0 with respect to \((W_1,W_2)\).
-
(4)
It follows from (3.4) that \(\inf _{w\in \widetilde{H}_T^1}\psi (w)\le 0\). If \(\inf _{w\in \widetilde{H}_T^1}\psi (w)=0\), all \(w\in W_2\) with \(\Vert w\Vert \le \delta _0\) are minimums of \(\psi \), which implies that \(\psi \) has infinitely many critical points. If \(\inf _{w\in \widetilde{H}_T^1}\psi (w)<0\), Proposition 2.2 implies that \(\psi \) has at least two nontrivial critical points. In any case, the functional \(\psi \) possesses two nonzero critical points, denoted by \(w_1\), \(w_2\). Consequently, \(u_1=\theta (w_1)+w_1\) and \(u_2=\theta (w_2)+w_2\) are two nonzero solutions of problem (1.1). This completes the proof.
\(\square \)
Proof of Theorem 1.4
The proof is similar to that of Theorem 1.3 with the exception that now (3.3) follows from condition \((A_8)\). We verify this assertion by contradiction. If not, there exist \(c_2\in {\mathbb {R}}\) and a sequence \((u_n)\subset \widetilde{H}_T^1\) such that \(\Vert u_n\Vert \rightarrow \infty \) as \(n\rightarrow \infty \) and
Set
Then \(H_T^1=H_1\bigoplus H_1^\bot \), and for \(u_n\in \widetilde{H}_T^1\), we can rewrite
where \(a_n\), \(b_n\in {\mathbb {R}}^N\) and \(w_n\in H_1^\bot \). It is obvious that
Hence, using \((A_8)\), we have
which implies that \((w_n)\) is bounded. Taking \(v_n=u_n/\Vert u_n\Vert \), then \(\Vert v_n\Vert =1\), and hence \((a_n)\) and \((b_n)\) are bounded. Up to a subsequence, we can assume that
for some a, \(b\in {\mathbb {R}}\). By the boundedness of \((w_n)\), one has \(w_n/\Vert u_n\Vert \rightarrow 0\). Hence
and \(|a|+|b|\ne 0\), which yields that
by Sobolev’s inequality. Therefore,
for a.e. \(t\in [0,T]\), since \(a\cos \omega t+b\sin \omega t\) only has finite zeros. Now set
It follows from Fatou’s lemma (see [25]) that
a contradiction with (3.5). Thus
\(\square \)
Proof of Theorem 1.5
It follows from (A) and \((A_9)\) that, for \(\varepsilon \in (0,1-a)\), there exists \(M_\varepsilon >0\) such that
for all \(x\in {\mathbb {R}}^N\) and a.e. \(t\in [0,T]\). Hence, we obtain, by Lemma 2.1 and Wirtinger’s inequality,
for all \(w\in \widetilde{H}_T^1\). The equivalence of the norm \(|\dot{w}|_2\) and the \(H_T^1\)-norm on \(\widetilde{H}_T^1\) shows that
The remainder of the proof is the same as that of Theorem 1.3. Consequently, Theorem 1.5 holds. \(\square \)
Data Availability
Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.
References
Amann, H.: Saddle points and multiple solutions of differential equations. Math. Z. 169, 127–166 (1979)
Bates, P.W., Ekeland, I.: A saddle-point theorem. In: Differential Equations. Proc. Eighth Fall Conf., Oklahoma State Univ., Stillwater, Okla, pp. 123–126. Academic Press, New York (1980)
Brezis, H., Nirenberg, L.: Remarks on finding critical points. Comm. Pure Appl. Math. 44, 939–963 (1991)
Castro, A., Lazer, A.C.: Applications of a max-min principle. Rev. Colombiana Mat. 10, 141–149 (1976)
Chen, G., Zhan, G.: Infinitely many small negative energy periodic solutions for second order Hamiltonian systems without spectrum 0. Houston J. Math. 43, 847–860 (2017)
Fei, G.: On periodic solutions of superquadratic Hamiltonian systems. Electron. J. Differ. Equ. 8, 12 (2002)
Izydorek, M., Janczewska, J., Waterstraat, N.: The equivariant spectral flow and bifurcation of periodic solutions of Hamiltonian systems. Nonlinear Anal. 211(17), 112475 (2021)
Lazer, A.C., Landesman, E.M., Meyers, D.R.: On saddle point problems in the calculus of variations, the Ritz algorithm, and monotone convergence. J. Math. Anal. Appl. 52, 594–614 (1975)
Liu, Y., Guo, F.: Multiplicity of periodic solutions for a class of second-order perturbed Hamiltonian systems. J. Math. Anal. Appl. 491(14), 124386 (2020)
Liu, G.: Periodic solutions of second order Hamiltonian systems with nonlinearity of general linear growth. Electron. J. Qual. Theory Differ. Equ. 27, 19 (2021)
Long, Y.M.: Nonlinear oscillations for classical Hamiltonian systems with bi-even subquadratic potentials. Nonlinear Anal. 24, 1665–1671 (1995)
Manasevich, R.F.: A min max theorem. J. Math. Anal. Appl. 90, 64–71 (1982)
Manasevich, R.F.: A nonvariational version of a max-min principle. Nonlinear Anal. 7, 565–570 (1983)
Mawhin, J., Willem, M.: Critical Point Theory and Hamiltonian Systems. Appl. Math. Sci., vol. 74. Springer-Verlag, New York (1989)
Shen, T., Liu, W.: Infinitely many rotating periodic solutions for suplinear second-order impulsive Hamiltonian systems. Appl. Math. Lett. 88, 164–170 (2019)
Strzelecki, D.: Periodic solutions of symmetric Hamiltonian systems. Arch. Ration. Mech. Anal. 237, 921–950 (2020)
Tang, C.-L.: Existence and multiplicity of periodic solutions for nonautonomous second order systems. Nonlinear Anal. 32, 299–304 (1998)
Tang, C.-L.: Periodic solutions for nonautonomous second order systems with sublinear nonlinearity. Proc. Amer. Math. Soc. 126, 3263–3270 (1998)
Tang, C.-L., Wu, X.-P.: Some critical point theorems and their applications to periodic solution for second order Hamiltonian systems. J. Differ. Equ. 248, 660–692 (2010)
Tang, X.H., Jiang, J.: Existence and multiplicity of periodic solutions for a class of second-order Hamiltonian systems. Comput. Math. Appl. 59, 3646–3655 (2010)
Thews, K.: A reduction method for some nonlinear Dirichlet problems. Nonliear Anal. 3, 795–813 (1979)
Wang, Z., Zhang, J.: New existence results on periodic solutions of non-autonomous second order Hamiltonian systems. Appl. Math. Lett. 79, 43–50 (2018)
Wang, Z., Zhang, J., Chen, M.: A unified approach to periodic solutions for a class of non-autonomous second order Hamiltonian systems. Nonlinear Anal. Real World Appl. 58, 12, 103218 (2021)
Wu, X.: Saddle point characterization and multiplicity of periodic solutions of non-autonomous second-order systems. Nonlinear Anal. 58, 899–907 (2004)
Yosida, K.: Functional Analysis, 6th edn. Springer-Verlag, Berlin (1980)
Zhang, Q., Liu, C.: Infinitely many periodic solutions for second order Hamiltonian systems. J. Differ. Equ. 251, 816–833 (2011)
Zhao, F., Wu, X.: Existence and multiplicity of nonzero periodic solution with saddle point character for some nonautonomous second order systems. J. Math. Anal. Appl. 308, 588–595 (2005)
Zhao, F., Wu, X.: Existence and multiplicity of periodic solution for non-autonomous second-order systems with linear nonlinearity. Nonlinear Anal. 60, 325–335 (2005)
Zou, W., Li, S.: Infinitely many solutions for Hamiltonian systems. J. Differ. Equ. 186, 141–164 (2002)
Funding
The work is supported by the National Natural Science Foundation of China (11971393) and the Science and Technology Research Program of Chongqing Municipal Education Commission (No. KJQN202200523).
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Ye, Y., Liu, S. Notes on Multiple Periodic Solutions for Second Order Hamiltonian Systems. Qual. Theory Dyn. Syst. 21, 141 (2022). https://doi.org/10.1007/s12346-022-00673-z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s12346-022-00673-z