By using the variational method, some existence theorems
are obtained for periodic solutions of autonomous (q,p)-Laplacian system with impulsive effects.
1. Introduction
Let B={1,2,…,l}, C={1,2,…,k}, l,k∈ℕ.
In this paper, we consider the following system:
ddtΦq(u̇1(t))=∇u1F(u1(t),u2(t)),a.e.t∈[0,T],ddtΦp(u̇2(t))=∇u2F(u1(t),u2(t)),a.e.t∈[0,T],u1(0)-u1(T)=u̇1(0)-u̇1(T)=0,u2(0)-u2(T)=u̇2(0)-u̇2(T)=0,ΔΦq(u̇1(tj))=Φq(u̇1(tj+))-Φq(u̇1(tj-))=∇Ij(u1(tj)),j∈B,ΔΦp(u̇2(sm))=Φp(u̇2(sm+))-Φp(u̇2(sm-))=∇Km(u2(sm)),m∈C,
where p>1,q>1, T>0, u(t)=(u1(t),u2(t))=(u11(t),u12(t),…,u1N(t),u21(t),u22(t),…,u2N(t))τ, tj(j=1,2,…,l), and sm(m=1,2,…,k) are the instants where the impulses occur and 0=t0<t1<t2<⋯<tl<tl+1=T,0=s0<s1<s2<⋯<sk<sk+1=T, Ij:ℝN→ℝ(j∈B), and Km:ℝN→ℝ(m∈C) are continuously differentiableΦμ(z)=|z|μ-2z=(∑i=1Nzi2)(μ-2)/2(z1⋮zN),μ∈R,μ>1,∇Ij(x)=(∂Ij∂x1⋮∂Ij∂xN),∇Km(x)=(∂Km∂x1⋮∂Km∂xN),
and F:ℝN×ℝN→ℝ satisfies the following assumption.
F(x) is continuously differentiable in (x1,x2), and there exist a1,a2∈C(ℝ+,ℝ+) such that|F(x1,x2)|≤a1(|x1|)+a2(|x2|),|∇F(x1,x2)|≤a1(|x1|)+a2(|x2|),|Ij(x1)|≤a1(|x1|),|∇Ij(x1)|≤a1(|x1|),j∈B,|Km(x2)|≤a2(|x2|),|∇Km(x2)|≤a2(|x2|),m∈C,
for all x=(x1,x2)∈ℝN×ℝN.
When p=q=2, Ij≡0(j∈B), Km≡0(m∈C), and F(u1,u2)=F1(u1), system (1.1) reduces to the following autonomous second-order Hamiltonian system:ü1(t)=∇u1F1(u1(t)),a.e.t∈[0,T],u1(0)-u1(T)=u̇1(0)-u̇1(T)=0.
There have been lots of results about the existence of periodic solutions for system (1.4) and nonautonomous second order Hamiltonian systemü1(t)=∇u1F1(t,u1(t)),a.e.t∈[0,T],u1(0)-u1(T)=u̇1(0)-u̇1(T)=0,
(e.g., see [1–21]). Many solvability conditions have been given, for instance, coercive condition, subquadratic condition, superquadratic condition, convex condition, and so on.
When p=q=2, ∇Ij≢0(j∈B), Km≡0(m∈C), and F(u1,u2)=F1(u1), system (1.1) reduces to the following autonomous second-order Hamiltonian system with impulsive effects:ü1(t)=∇u1F1(u1(t)),a.e.t∈[0,T],u1(0)-u1(T)=u̇1(0)-u̇1(T)=0,u̇1(tj+)-u̇1(tj-)=∇Ij(u1(tj)).
Recently, many authors studied the existence of periodic solutions for impulsive differential equations by using variational methods, and lots of interesting results have been obtained. For example, see [22–28]. Especially, nonautonomous second-order Hamiltonian system with impulsive effects is considered in [25, 26] by using the least action principle and the saddle point theorem.
When Ij≡0(j∈B) and Km≡0(m∈C), system (1.1) reduces to the following system:
ddtΦq(u̇1(t))=∇u1F(u1(t),u2(t)),a.e.t∈[0,T],ddtΦp(u̇2(t))=∇u2F(u1(t),u2(t)),a.e.t∈[0,T],u1(0)-u1(T)=u̇1(0)-u̇1(T)=0,u2(0)-u2(T)=u̇2(0)-u̇2(T)=0.
In [29, 30], Paşca and Tang obtained some existence results for system (1.7) by using the least action principle and saddle point theorem. Motivated by [17, 22–30], in this paper, we are concerned with system (1.1) and also use the least action principle and saddle point theorem to study the existence of periodic solution. Our results still improve those in [17] even if system (1.1) reduces to system (1.4).
A function G:ℝN→ℝ is called to be (λ,μ)-quasiconcave if
G(λ(x+y))≥μ(G(x)+G(y)),
for some λ,μ>0 and x,y∈ℝN.
Next, we state our main results.
Theorem 1.1.
Let q′ and p′ be such that 1/q+1/q′=1 and 1/p+1/p′=1. Suppose F satisfies assumption (A) and the following conditions:
there exist
0<r1<(q′+1)q/q′TqΘ(q,q′),0<r2<(p′+1)p/p′TpΘ(p,p′),
such that
(∇x1F(x1,x2)-∇y1F(y1,y2),x1-y1)≥-r1|x1-y1|q,∀(x1,x2),(y1,y2)∈RN×RN,(∇x2F(x1,x2)-∇y2F(y1,y2),x2-y2)≥-r2|x2-y2|p,∀(x1,x2),(y1,y2)∈RN×RN,
where
Θ(q,q′)=∫01[sq′+1+(1-s)q′+1]q/q′ds,Θ(p,p′)=∫01[sp′+1+(1-s)p′+1]p/p′ds,
F(x)→+∞,as|x|→∞,wherex=(x1,x2),
there exists β∈ℝ such that
Ij(x)≥β,∀x∈RN,j∈B,Km(x)≥β,∀x∈RN,m∈C.
Then, system (1.1) has at least one solution in WT1,q×WT1,p, where WT1,s={u:[0,T]→ℝN∣u is absolutely continuous,u(0)=u(T) andu̇∈Ls(0,T;ℝN)},s∈ℝ.
Furthermore, if Ij≡0(j∈B), Km≡0(m∈C) and the following condition holds:
there exist δ>0, a∈[0,(q′+1)q/q′/qTqΘ(q,q′)) and b∈[0,(p′+1)p/p′/(pTpΘ(p,p′))) such that
-a|x1|q-b|x2|p≤F(x1,x2)≤0,∀|x1|≤δ,|x2|≤δ,
then system (1.7) has at least two nonzero solutions in WT1,q×WT1,p.
When p=q=2, F(x1,x2)=F1(x1), by Theorem 1.1, it is easy to get the following corollary.
Corollary 1.2.
Suppose F1 satisfies the following conditions:
F1(z) is continuously differentiable in z and there exists a1∈C(ℝ+,ℝ+) such that
|F1(z)|≤a1(|z|),|∇F1(z)|≤a1(|z|),|Ij(z)|≤a1(|z|),|∇Ij(z)|≤a1(|z|),j∈B,
for all z∈ℝN.
there exists 0<r<6/T2 such that
(∇zF1(z)-∇wF1(w),z-w)≥-r|z-w|2,∀z,w∈RN,
F1(z)→+∞,as|z|→∞,z∈ℝN;
there exists β∈ℝ such that
Ij(z)≥β,∀z∈RN,j∈B.
Then, system (1.6) has at least one solution in WT1,2. Furthermore, if Ij≡0(j∈B) and the following condition holds:
there exist δ>0 and a∈[0,(3/T2)) such that
-a|z|2≤F1(z)≤0,∀z∈RN,|z|≤δ,
then system (1.4) has at least two nonzero solutions in WT1,2.
For the Sobolev space W̃T1,2, one has the following sharp estimates (see in [3, Proposition 1.2]): ∫0T|u(t)|2dt≤T24π2∫0T|u̇(t)|2dt(Wirtinger′sinequality),‖u‖∞2≤T12∫0T|u̇(t)|2dt(Sobolev′sinequality).
By the above two inequalities, we can obtain the following better results than by Corollary 1.2.
Theorem 1.3.
Suppose F1 satisfies assumption (A)′, (F2)′, (I1)′ and
there exists 0<r<4π2/T2 such that (1.15) holds.
Then, system (1.6) has at least one solution in WT1,2. Furthermore, if Ij≡0(j∈B) and the following condition holds:
there exist δ>0 and a∈[0,(2π2)/T2) such that
-a|z|2≤F1(z)≤0,∀z∈RN,|z|≤δ,
then system (1.4) has at least two nonzero solutions in WT1,2.
Moreover, for system (1.6), we have the following additional result.
Theorem 1.4.
Suppose F1 satisfies assumption (A)′, (F1)′ ′ and the following conditions:
F1(z) is (λ,μ)-quasiconcave on ℝN,
F1(z)→-∞ as |z|→+∞, z∈ℝN,
there exist dj>0(j∈B) such that
|∇Ij(z)|≤dj,∀z∈RN,j∈B,
there exist bj>0,cj>0, γj∈ℝ, αj∈[0,2)(j∈B) such that
-bj|z|αj-cj≤Ij(z)≤γj,∀z∈RN,j∈B.
Then, system (1.6) has at least one solution in WT1,2.
Remark 1.5.
In [17], Yang considered the second-order Hamiltonian system with no impulsive effects, that is, system (1.4). When Ij≡0(j∈B), our Theorems 1.3 and 1.4 still improve those results in [17]. To be precise, the restriction of r is relaxed, and some unnecessary conditions in [17] are deleted. In [17], the restriction of r is 0<r<T/12, which is not right. In fact, from his proof, it is easy to see that it should be 0<r<12/T2. Obviously, our restriction 0<r<4π2/T2 is better. Moreover, in our Theorem 1.4, we delete such conditions (of in [17, Theorem 1]): ∇F1(0)=0, and there exist positive constants M,N such that
F1(z)≥-M|z|2-N,z∈RN.
Finally, it is remarkable that Theorems 1.3 and 1.4 are also different from those results in [1–16]. We can find an example satisfying our Theorem 1.3 but not satisfying the results in [1–21]. For example, let
F1(z)=π22T2(|z1|4+|z2|4+⋯+|zN|4)-π24T2|z|2,
where z=(z1,…,zN)τ. We can also find an example satisfying our Theorem 1.4 but not satisfying the results in [1–21]. For example, let
F1(z)=-r2|z|2,
where 12/T2<r<4π2/T2.
2. Variational Structure and Some Preliminaries
The norm in WT1,p is defined by‖u‖WT1,p=[∫0T|u(t)|pdt+∫0T|u̇(t)|pdt]1/p.
Set‖u‖p=(∫0T|u(t)|pdt)1/p,‖u‖∞=maxt∈[0,T]|u(t)|.
LetW̃T1,p={u∈WT1,p∣∫0Tu(t)dt=0}.
Obviously, WT1,p is a reflexive Banach space. It is easy to know that W̃T1,p is a subset of WT1,p and WT1,p=ℝN⊕W̃T1,p. In this paper, we will use the space W defined by
W=WT1,q×WT1,p,u(t)=(u1(t),u2(t)),
with the norm ∥(u1,u2)∥W=∥u1∥WT1,q+∥u2∥WT1,p. It is clear that W is a reflexive Banach space. Let W̃=W̃T1,q×W̃T1,p. Then, W=(W̃T1,q×W̃T1,p)⊕(ℝN×ℝN).
Lemma 2.1 (see [31] or [32]).
Each u∈WT1,p and each v∈WT1,q can be written as u(t)=u¯+ũ(t) and v(t)=v¯+ṽ(t) with
u¯=1T∫0Tu(t)dt,∫0Tũ(t)dt=0,v¯=1T∫0Tv(t)dt,∫0Tṽ(t)dt=0.
Then,
‖ũ‖∞≤(Tp′+1)1/p′(∫0T|u̇(s)|pds)1/p,‖ṽ‖∞≤(Tq′+1)1/q′(∫0T|v̇(s)|qds)1/q,∫0T|ũ(s)|pds≤TpΘ(p,p′)(p′+1)p/p′∫0T|u̇(s)|pds,∫0T|ṽ(s)|qds≤TqΘ(q,q′)(q′+1)q/q′∫0T|v̇(s)|qds,
where
Θ(p,p′)=∫01[sp′+1+(1-s)p′+1]p/p′ds,Θ(q,q′)=∫01[sq′+1+(1-s)q′+1]q/q′ds.
Note that if u∈WT1,p, then u is absolutely continuous. However, we cannot guarantee that u̇ is also continuous. Hence, it is possible that ΔΦp(u̇(t))=Φp(u̇(t+))-Φp(u̇(t-))≠0, which results in impulsive effects.
Following the idea in [22], one takes v1∈WT1,q and multiplies the two sides ofddt(|u̇1(t)|q-2u̇1(t))-∇x1F(u1(t),u2(t))=0,
by v1 and integrate from 0 to T, one obtains∫0T[ddt(|u̇1(t)|q-2u̇1(t))-∇x1F(u1(t),u2(t))]v1(t)dt=0.
Note that v1(t) is continuous. So, v1(tj-)=v1(tj+)=v1(tj). Combining u̇1(0)-u̇1(T)=0, one has
∫0T(dΦq(u̇1(t))dt,v1(t))dt=∑j=0l∫tjtj+1(d(Φq(u̇1(t)))dt,v1(t))dt=∑j=0l[(Φq(u̇1(tj+1-)),v1(tj+1-))-(Φq(u̇1(tj+)),v1(tj+))]dt-∑j=0l∫tjtj+1(Φq(u̇1(t)),v̇1(t))dt=(Φq(u̇1(T)),v1(T))-(Φq(u̇1(0)),v1(0))-∑j=1l(ΔΦq(u̇1(tj)),v1(tj))-∫0T(Φq(u̇1(t)),v̇1(t))dt=-∑j=1l(∇Ij(u1(tj)),v1(tj))-∫0T(Φq(u̇1(t)),v̇1(t))dt.
Combining with (2.10), one has∫0T(Φq(u̇1(t)),v̇1(t))dt+∑j=1l(∇Ij(u1(tj)),v1(tj))+∫0T(∇x1F(u1(t),u2(t)),v1(t))dt=0.
Similarly, one can get∫0T(Φp(u̇2(t)),v̇2(t))dt+∑m=1k(∇Km(u2(sm)),v2(sm))+∫0T(∇x2F(u1(t),u2(t)),v2(t))dt=0,
for all v2∈WT1,p. Considering the above equalities, one introduces the following concept of the weak solution for system (1.1).
Definition 2.2.
We say that a function u=(u1,u2)∈WT1,q×WT1,p is a weak solution of system (1.1) if∫0T(Φq(u̇1(t)),v̇1(t))dt+∑j=1l(∇Ij(u1(tj)),v1(tj))=-∫0T(∇x1F(u1(t),u2(t)),v1(t))dt,∫0T(Φp(u̇2(t)),v̇2(t))dt+∑m=1k(∇Km(u2(sm)),v2(sm))=-∫0T(∇x2F(u1(t),u2(t)),v2(t))dt
holds for any v=(v1,v2)∈WT1,q×WT1,p.
Define the functional φ:WT1,q×WT1,p→ℝ by
φ(u1,u2)=1q∫0T|u̇1(t)|qdt+1p∫0T|u̇2(t)|pdt+∫0TF(u1(t),u2(t))dt+∑j=1lIj(u1(tj))+∑m=1kKm(u2(sm))=ϕ(u1,u2)+ψ(u1,u2),
where (u1,u2)∈WT1,q×WT1,p,
ϕ(u1,u2)=1q∫0T|u̇1(t)|qdt+1p∫0T|u̇2(t)|pdt+∫0TF(u1(t),u2(t))dt,ψ(u1,u2)=∑j=1lIj(u1(tj))+∑m=1kKm(u2(sm)).
By assumption (A) and [33], we know that ϕ∈C1(WT1,q×WT1,p,ℝ). The continuity of Ij(j∈B) and Km(m∈C) implies that ψ∈C1(WT1,p×WT1,p,ℝ). So, φ∈C1(WT1,p,ℝ), and for all (v1,v2)∈WT1,q×WT1,p, we have
〈φ′(u1,u2),(v1,v2)〉=∫0T(Φq(u̇1(t)),v̇1(t))dt+∫0T(Φp(u̇2(t)),v̇2(t))dt+∫0T(∇x1F(u1(t),u2(t)),v1(t))dt+∫0T(∇x2F(u1(t),u2(t)),v2(t))dt+∑j=1l(∇Ij(u1(tj)),v1(tj))+∑m=1k(∇Km(u2(sm)),v2(sm)).
Definition 2.2 shows that the critical points of φ correspond to the weak solutions of system (1.1).
We will use the following lemma to seek the critical point of φ.
Lemma 2.3 (see [3, Theorem 1.1]).
If φ is weakly lower semicontinuous on a reflexive Banach space X and has a bounded minimizing sequence, then φ has a minimum on X.
Lemma 2.4 (see [34]).
Let φ be a C1 function on X=X1⊕X2 with φ(0)=0, satisfying (PS) condition, and assume that for some ρ>0,
φ(u)≥0,foru∈X1,‖u‖≤ρ,φ(u)≤0,foru∈X2,‖u‖≤ρ.
Assume also that φ is bounded below and infXφ<0, then φ has at least two nonzero critical points.
Lemma 2.5 (see [35, Theorem 4.6]).
Let X=X1⊕X2, where X is a real Banach space and X1≠{0} and is finite dimensional. Suppose that φ∈C1(X,ℝ) satisfies (PS)-condition and
there is a constant α and a bounded neighborhood D of 0 in X1 such that φ∣∂D≤α,
there is a constant β>α such that φ∣X2≥β.
Then, φ possesses a critical value c≥β. Moreover, c can be characterized as
c=infh∈Γmaxu∈D¯φ(h(u)),
where,
Γ={h∈C(D¯,X)∣h=idon∂D}.
3. Proof of TheoremsLemma 3.1.
Under assumption (A), φ is weakly lower semicontinuous on WT1,q×WT1,p.
Proof.
Let
ϕ1(u1,u2)=1q∫0T|u̇1(t)|qdt+1p∫0T|u̇2(t)|pdt,ϕ2(u1,u2)=∫0TF(u1(t),u2(t))dt.
Since
ϕ1(u1+v12,u2+v22)=1q∫0T|u̇1(t)+v̇1(t)2|qdt+1p∫0T|u̇2(t)+v̇2(t)2|pdt≤2q-1q∫0T12q|u̇1(t)|qdt+2q-1q∫0T12q|v̇1(t)|qdt+2p-1p∫0T12p|u̇2(t)|pdt+2p-1p∫0T12p|v̇2(t)|pdt≤12q∫0T|u̇1(t)|qdt+12q∫0T|v̇1(t)|qdt+12p∫0T|u̇2(t)|pdt+12p∫0T|v̇2(t)|pdt=ϕ1(u1,u2)+ϕ1(v1,v2)2,
then ϕ1 is convex. Moreover, by [33], we know that ϕ1 is continuous, and so, it is lower semicontinuous. Thus, it follows from [3, Theorem 1.2] that ϕ1 is weakly lower continuous. By assumption (A), it is easy to verify that ϕ2(u1,u2) is weakly continuous. We omit the details. Let
ψ1(u1)=∑j=1lIj(u1(tj)),ψ2(u2)=∑m=1kKm(u2(sm)).
Next, we show that ψ1 and ψ2 are weakly continuous on WT1,q and WT1,p, respectively. In fact, if
u1n⇀u1weaklyinWT1,p,asn⟶∞,
then by in [3, Proposition 1.2], we know that
u1n⟶u1stronglyinC(0,T;RN),asn⟶∞.
So, there exists M1>0 such that ∥u1∥∞≤M1 and ∥u1n∥∞≤M1, for all n∈ℕ. Thus, we have
|ψ1(u1n)-ψ1(u1)|=|∑j=1lIj(u1n(tj))-∑j=1lIj(u1(tj))|≤∑j=1l|Ij(u1n(tj))-Ij(u1(tj))|=∑j=1l|∫01(∇Ij(u1(tj)+s(u1n(tj)-u1(tj))),u1n(tj)-u1(tj))ds|≤‖u1n-u1‖∞∑j=1lmaxt∈[0,3M1]a1(t)⟶0.
Hence, ψ1 is weakly continuous on WT1,q. Similarly, we can prove that ψ2 is also weakly continuous on WT1,p. Thus, we complete the proof.
Proof of Theorem 1.1.
It follows from (F1) and (2.7) that
∫0T[F(u1(t),u2(t))-F(u1(t),u¯2)]=∫0T∫011s(∇Fx2(u1(t),u¯2+sũ2(t)),sũ2(t))dsdt=∫0T∫011s(∇Fx2(u1(t),u¯2+sũ2(t))-∇Fx2(u¯1,u¯2),sũ2(t))dsdt≥-r2p∫0T|ũ2(t)|pdt≥-r2TpΘ(p,p′)p(p′+1)p/p′∫0T|u̇2(t)|pdt,∀(u1,u2)∈W,∫0T[F(u1(t),u¯2)-F(u¯1,u¯2)]dt=∫0T∫011s(∇x1F(u¯1+sũ1(t),u¯2),sũ1(t))dsdt=∫0T∫011s(∇x1F(u¯1+sũ1(t),u¯2)-∇x1F(u¯1,u¯2),sũ1(t))dsdt≥-r1q∫0T|ũ1(t)|qdt≥-r1TqΘ(q,q′)q(q′+1)q/q′∫0T|u̇1(t)|qdt,∀(u1,u2)∈W.
Hence, by (I1), (3.7), and (3.8), we have
φ(u1,u2)=1q∫0T|u̇1(t)|qdt+1p∫0T|u̇2(t)|pdt+∫0T[F(u1(t),u2(t))-F(u1(t),u¯2)]dt+∫0T[F(u1(t),u¯2)-F(u¯1,u¯2)]dt+TF(u¯1,u¯2)+∑j=1lIj(u1(tj))+∑m=1kKm(u2(sm))≥(1p-r2TpΘ(p,p′)p(p′+1)p/p′)∫0T|u̇2(t)|pdt+(1q-r1TqΘ(q,q′)q(q′+1)q/q′)∫0T|u̇1(t)|qdt+TF(u¯1,u¯2)-(l+k)|β|.
Note that for u∈WT1,p,
‖u‖WT1,p⟶∞⟺(|u¯|p+∫0T|u̇(t)|pdt)1/p⟶∞,
and for v∈WT1,q,
‖v‖WT1,q⟶∞⟺(|v¯|q+∫0T|v̇(t)|qdt)1/q⟶∞.
So, (F2) and (3.9) imply that
φ(u1,u2)⟶+∞,as‖(u1,u2)‖W⟶∞.
Thus, by Lemma 2.3, we know that φ has at least one critical point which minimizes φ on W.
Furthermore, if Ij(u1(tj))≡0(j∈B) and Km(u2(sm))≡0(m∈C), then system (1.1) reduces to (1.7). When (F3) also holds, we will use Lemma 2.4 to obtain more critical points of φ. Let X=W, X2=ℝN×ℝN and X1=W̃=W̃T1,q×W̃T1,p.
By (3.9), we know that φ(u1,u2)→+∞ as ∥(u1,u2)∥W→∞. So, φ satisfies (PS) condition and is bounded below. Take ρ=δ/c1, where c1 is a positive constant such that ∥u1∥∞≤c1∥u1∥WT1,q≤c1∥u∥W and ∥u2∥∞≤c1∥u2∥WT1,p≤c1∥u∥W for all (u1,u2)∈W. It follows from (F3) and Lemma 2.1 that
φ(u1,u2)=1q∫0T|u̇1(t)|qdt+1p∫0T|u̇2(t)|pdt+∫0TF(u1(t),u2(t))dt≥1q∫0T|u̇1(t)|qdt+1p∫0T|u̇2(t)|pdt-a∫0T|u1(t)|qdt-b∫0T|u2(t)|pdt≥1q∫0T|u̇1(t)|qdt+1p∫0T|u̇2(t)|pdt-aTqΘ(q,q′)(q′+1)q/q′∫0T|u̇1(t)|qdt-bTpΘ(p,p′)(p′+1)p/p′∫0T|u̇2(t)|pdt,∀(u1,u2)∈X1.
Since a≤(q′+1)q/q′/(qTqΘ(q,q′)) and b≤(p′+1)p/p′/(pTpΘ(p,p′)), (3.13) implies that φ(u1,u2)≥0 for all (u1,u2)∈X1 with ∥u∥W≤ρ. By (F3), it is easy to obtain that φ(u1,u2)≤0, for all (u1,u2)∈X2 with ∥u∥W≤ρ.
If inf{φ(u1,u2):(u1,u2)∈W}=0, then from above, we have φ(u1,u2)=0 for all (u1,u2)∈X2 with ∥(u1,u2)∥W≤ρ. Hence, all (u1,u2)∈X2 with ∥(u1,u2)∥W≤ρ are minimal points of φ, which implies that φ has infinitely many critical points. If inf{φ(u1,u2):(u1,u2)∈W}<0, then by Lemma 2.4, φ has at least two nonzero critical points. Hence, system (1.7) has at least two nontrivial solutions in W. We complete our proof.
Proof of Theorem 1.3.
We only need to use (1.18) and (1.19) to replace (2.6) and (2.7) in the proof Theorem 1.1 with p=q=2, F(t,u1,u2)=F1(u1) and Km(u2)≡0(m∈C). It is easy. So, we omit it.
Lemma 3.2.
Under the assumptions of Theorem 1.4, the functional φ1 defined by
φ1(u1)=12∫0T|u̇1(t)|2dt+∫0TF1(u1(t))dt+∑j=1lIj(u1(tj))dt
satisfies (PS) condition.
Proof.
Suppose that {u1n} is a (PS) sequence for φ1; that is, there exists D1>0 such that
|φ(u1n)|≤D1,∀n∈N,φ′(u1n)⟶0,asn⟶∞.
Hence, for n large enough, we have ∥φ′(u1n)∥≤1. It follows from (F1)′′, (I2), and (1.18) that
‖ũ1n‖WT1,2≥〈φ1′(u1n),ũ1n〉=∫0T|u̇1n(t)|2dt+∫0T(∇x1F1(u1n(t)),ũ1n(t))dt+∑j=1l(∇Ij(u1n(tj)),ũ1n(tj))=∫0T|u̇1n(t)|2dt+∫0T(∇x1F1(u1n(t))-∇x1F1(u¯1n(t)),ũ1n(t))dt+∑j=1l(∇Ij(u1n(tj)),ũ1n(tj))≥∫0T|u̇1n(t)|2dt-rT24π2∫0T|u̇1n(t)|2dt-‖ũ1n‖∞∑j=1ldj≥[1-rT24π2]∫0T|u̇1n(t)|2dt-(T12)1/2(∫0T|u̇1n(t)|2dt)1/2∑j=1ldj,
for n large enough. By (1.18), we have
‖ũ1n‖WT1,2≤[T24π2+1]1/2(∫0T|u̇1n(t)|2dt)1/2,
and (3.16), (3.17), and r<4π2/T2 imply that there exists D2,D3>0 such that
∫0T|u̇1n(t)|2dt≤D2,‖ũ1n‖WT1,2≤D3.
It follows from (F4), (3.15), (I3), (1.18), and (3.18) that
-D1≤φ1(u1n)=12∫0T|u̇1n(t)|2dt+∫0TF1(u1n(t))dt+∑j=1lIj(u1(tj))≤12∫0T|u̇1n(t)|2dt+1μ∫0TF1(λu¯1n)dt-∫0TF1(-ũ1n(t))dt+∑j=1lγj=12∫0T|u̇1n(t)|2dt+TμF1(λu¯1n)-TF1(0)-∫0T[F1(-ũ1n(t))-F1(0)]dt+∑j=1lγj=12∫0T|u̇1n(t)|2dt+TμF1(λu¯1n)-TF1(0)+∑j=1lγj-∫0T∫011s(∇F1(-sũ1n(t))-∇F1(0),-sũ1n(t))dsdt≤12∫0T|u̇1n(t)|2dt+TμF1(λu¯1n)+r∫0T∫01s|ũ1n(t)|2dsdt-TF1(0)+∑j=1lγj≤12∫0T|u̇1n(t)|2dt+TμF1(λu¯1n)+r2∫0T|ũ1n(t)|2dt-TF1(0)+∑j=1lγj≤max{1,r}2‖ũ1n‖WT1,22+TμF1(λu¯1n)-TF1(0)+∑j=1lγj≤max{1,r}2D3q+TμF1(λu¯1n)-TF1(0)+∑j=1lγj,
for all n and (3.19) and (F5) imply that {u¯1n} is bounded. Combining (3.18), we know that {u1n} is a bounded sequence. Similar to the argument in [25], it is easy to obtain that φ satisfies (PS) condition.
Proof of Theorem 1.4.
From (I3) and (F5), it is easy to see that for x1∈ℝN,
φ1(x1)⟶-∞,as|x1|⟶∞.
For all u1∈W̃T1,2, by (1.18), (F1)′′ and (I3), we have
φ1(u1)=12∫0T|u̇1(t)|2dt+∫0TF1(u1(t))dt+∑j=1lIj(u1(tj))=12∫0T|u̇1(t)|2dt+∫0T[F1(u1(t))-F1(0)]dt+TF1(0)+∑j=1lIj(u1(tj))=12∫0T|u̇1(t)|2dt+∫0T∫01(∇F1x1(su1(t)),u1(t))dsdt+∑j=1lIj(u1(tj))+TF1(0)=12∫0T|u̇1(t)|2dt+∫0T∫011s(∇F1x1(su1(t))-∇F1x1(0),su1(t))dsdt+∑j=1lIj(u1(tj))+TF1(0)≥12∫0T|u̇1(t)|2dt-r12∫0T|u1(t)|2dt+TF1(0)-∑j=1lbj|u1(tj)|αj-∑j=1lcj≥12∫0T|u̇1(t)|2dt-r1T24π2∫0T|u̇1(t)|2dt+TF1(0)-∑j=1lbj‖u1‖∞αj-∑j=1lcj≥(12-r1T24π2)∫0T|u̇1(t)|2dt+TF1(0)-(T12)αj/2∑j=1lbj(∫0T|u̇1(t)|2dt)αj/2-∑j=1lcj.
Note that for all u1∈W̃T1,2, ∥u1∥WT1,2 is equivalent to ∥u̇1∥L2. Then, r1<4π2/T2, αj<2(j∈B) and (3.21) imply that
φ1(u1)⟶+∞,as‖u1‖WT1,2⟶∞,u1∈W̃T1,2.
It follows from (3.20) and (3.22) that φ1 satisfies (φ1) and (φ2) in Lemma 2.5. Combining with Lemma 3.2, Lemma 2.5 shows that φ1 has at least one critical point. Thus, we complete the proof.
4. ExamplesExample 4.1.
Let q=4, p=2, T=π, t1=1, and s1=2. Consider the following system:
ddtΦ4(u̇1(t))=∇u1F(u1(t),u2(t)),a.e.t∈[0,π],ddtΦ2(u̇2(t))=∇u2F(u1(t),u2(t)),a.e.t∈[0,π],u1(0)-u1(π)=u̇1(0)-u̇1(π)=0,u2(0)-u2(π)=u̇2(0)-u̇2(π)=0,ΔΦ4(u̇1(1))=Φq(u̇1(1+))-Φq(u̇1(1-))=∇I1(u1(1)),ΔΦ2(u̇2(2))=Φp(u̇2(2+))-Φp(u̇2(2-))=∇K1(u2(2)),
where F(x1,x2)=x114+x124+⋯+x1N4+(1/π2)(x214+x222+⋯+x2N2)-(1/2π2)|x2|2, x1=(x11,x12,…,x1N), x2=(x21,x22,…,x2N), I1(x)=e|x|2, K1(x)=e|x|2, x∈ℝN. It is easy to verify that all conditions of Theorem 1.1 hold so that system (4.1) has at least one weak solution. Moreover, if F(x1,x2)=(1/π2)(x214+x224+⋯+x2N4)-1/2π2|x2|2, x2=(x21,x22,…,x2N), I1(x)=0 and K1(x)=0, x∈ℝN, then system (4.1) has at least two nonzero solutions.
Example 4.2.
Let T=2, t1=1. Consider the following autonomous second-order Hamiltonian system with impulsive effects:
ü(t)=∇uF(u(t)),a.e.t∈[0,2],u(0)-u(2)=u̇(0)-u̇(2)=0,u̇(1+)-u̇(1-)=∇I1(u(1)),
where F(z)=z14+z22+⋯+zN2-1/2|z|2,I1(z)=e|z|2, z=(z1,…,zN)τ∈ℝN. It is easy to verify that all conditions of Theorem 1.3 hold so that system (4.2) has at least one weak solution. Moreover, if F(z)=z14+z24+⋯+zN4-1/2|z|2 and I1(z)=0, z∈ℝN, then system (4.2) has at least two nonzero solutions.
Example 4.3.
Let T=π, t1=2. Consider the following autonomous second-order Hamiltonian system with impulsive effects:
ü(t)=∇uF(u(t)),a.e.t∈[0,π],u(0)-u(π)=u̇(0)-u̇(π)=0,u̇(2+)-u̇(2-)=∇I1(u(2)),
where F(z)=-|z|2,I1(z)=2sinz1, z=(z1,…,zN)τ∈ℝN. It is easy to verify that all conditions of Theorem 1.4 hold so that system (4.3) has at least one weak solution.
BergerM. S.SchechterM.On the solvability of semilinear gradient operator equations197725297132050033610.1016/0001-8708(77)90001-9ZBL0354.47025MawhinJ.Semicoercive monotone variational problems1987733-4118130938142MawhinJ.WillemM.198974New York, NY, USASpringerxiv+277Applied Mathematical Sciences982267MawhinJ.WillemM.Critical points of convex perturbations of some indefinite quadratic forms and semilinear boundary value problems at resonance198636431453870864ZBL0678.35091TangC.Periodic solutions of non-autonomous second order systems with γ-quasisubadditive potential19951893671675131254610.1006/jmaa.1995.1044TangC. L.Periodic solutions of nonautonomous second order systems1996202465469TangC.-L.Periodic solutions for nonautonomous second order systems with sublinear nonlinearity19981261132633270147639610.1090/S0002-9939-98-04706-6ZBL0902.34036TangC.-L.WuX.-P.Periodic solutions for second order systems with not uniformly coercive potential20012592386397184206610.1006/jmaa.2000.7401ZBL0999.34039WillemM.Oscillations forces de systmes hamiltoniens1981Université de BesançonWuX.Saddle point characterization and multiplicity of periodic solutions of non-autonomous second-order systems2004587-889990710.1016/j.na.2004.05.0202086063ZBL1058.34053WuX.-P.TangC.-L.Periodic solutions of a class of non-autonomous second-order systems19992362227235170457910.1006/jmaa.1999.6408ZBL0971.34027ZhaoF.WuX.Periodic solutions for a class of non-autonomous second order systems20042962422434207517410.1016/j.jmaa.2004.01.041ZhaoF.WuX.Existence and multiplicity of periodic solution for non-autonomous second-order systems with linear nonlinearity200560232533510.1016/j.na.2004.08.0312101882ZBL1087.34022TangC.-L.WuX.-P.A note on periodic solutions of nonautonomous second-order systems200413251295130310.1090/S0002-9939-03-07185-52053333ZBL1055.34084TangX. H.MengQ.Solutions of a second-order Hamiltonian system with periodic boundary conditions20101153722373310.1016/j.nonrwa.2010.02.0022683825ZBL1207.34027WangZ.ZhangJ.Periodic solutions of a class of second order non-autonomous Hamiltonian systems201072124480448710.1016/j.na.2010.02.0232639196ZBL1206.34060YangR.Periodic solutions of some autonomous second order Hamiltonian systems2008281-25158243092110.1007/s12190-008-0075-yZBL1157.34324SchechterM.Periodic non-autonomous second-order dynamical systems20062232290302221493610.1016/j.jde.2005.02.022ZBL1099.34042SchechterM.Periodic solutions of second-order nonautonomous dynamical systems20069251042251801ZBL1140.34366MaJ.TangC.-L.Periodic solutions for some nonautonomous second-order systems2002275248249410.1016/S0022-247X(02)00636-41943760ZBL1024.34036ZhaoF.WuX.Saddle point reduction method for some non-autonomous second order systems20042912653665203907610.1016/j.jmaa.2003.11.019ZBL1053.34039NietoJ. J.O'ReganD.Variational approach to impulsive differential equations200910268069010.1016/j.nonrwa.2007.10.0222474254ZBL1167.34318TianY.GeW.Applications of variational methods to boundary-value problem for impulsive differential equations200851250952710.1017/S00130915060015322465922ZBL1163.34015ZhouJ.LiY.Existence and multiplicity of solutions for some Dirichlet problems with impulsive effects2009717-82856286510.1016/j.na.2009.01.1402532812ZBL1175.34035ZhouJ.LiY.Existence of solutions for a class of second-order Hamiltonian systems with impulsive effects2010723-41594160310.1016/j.na.2009.08.0412577560ZBL1193.34057DingW.QianD.Periodic solutions for sublinear systems via variational approach20101142603260910.1016/j.nonrwa.2009.09.0072661927ZBL1209.34046SunJ.sunjuntao2008@163.comChenH.math_chb@163.comNietoJ. J.juanjose.nieto.roig@usc.esInfinitely many solutions for second-order Hamiltonian system with impulsive effects2011541-254455510.1016/j.mcm.2011.02.044SunJ.ChenH.ZhouT.Multiplicity of solutions for a fourth-order impulsive differential equation via variational methods2010823446458273795510.1017/S0004972710001802ZBL1213.34048PaşcaD.TangC.-L.Some existence results on periodic solutions of nonautonomous second-order differential systems with (q,p)-Laplacian201023324625110.1016/j.aml.2009.10.0052565184PaşcaD.Periodic solutions of a class of nonautonomous second order differential systems with (q,p)-Laplacian20101758418502777774ZhangX.TangX.xhtang2003@gmail.comPeriodic solutions for an ordinary p-Laplacian system201115313691396TangX.ZhangX.Periodic solutions for second-order Hamiltonian systems with a p-Laplacian20106419311310.2478/v10062-010-0008-82735506TianY.GeW.Periodic solutions of non-autonomous second-order systems with a p-Laplacian200766119220310.1016/j.na.2005.11.0202271646BrezisH.NirenbergL.Remarks on finding critical points1991448-993996310.1002/cpa.31604408081127041ZBL0751.58006RabinowitzP. H.198665Providence, RI, USAAmerican Mathematical Societyviii+100CBMS-NSF Regional Conference Series in Applied Mathematics, Vol. 65845785