We give several sufficient conditions under which the first-order nonlinear Hamiltonian system x'(t)=α(t)x(t)+f(t,y(t)),y'(t)=-g(t,x(t))-α(t)y(t) has no solution (x(t),y(t)) satisfying condition 0<∫-∞+∞[|x(t)|ν+(1+β(t))|y(t)|μ]dt<+∞, where μ,ν>1 and (1/μ)+(1/ν)=1, 0≤xf(t,x)≤β(t)|x|μ, xg(t,x)≤γ0(t)|x|ν, β(t),γ0(t)≥0, and α(t) are locally Lebesgue integrable real-valued functions defined on ℝ.
1. Introduction
In 1897, Poincaré [1] studied the existence of homoclinic solutions for Hamiltonian systems and realized that homoclinic solutions play a very important role in the study of the behavior of dynamical systems. Since then many methods have been developed to this study ([2–6]). Recently, the critical point theory has been successfully applied to establish the existence and multiplicity of homoclinic solutions for Hamiltonian systems; see [1, 7–20] and references therein.
Among the above-mentioned literature, there are two classes of Hamiltonian systems that have been widely investigated: one is the second-order Hamiltonian system
(1)(p(t)u′(t))′-L(t)u(t)+∇W(t,u(t))=0,
and the other is the first-order Hamiltonian system
(2)(x′(t)y′(t))=(0-II0)(Hx(t,x(t),y(t))Hy(t,x(t),y(t))).
By means of variational methods, in order to seek the homoclinic solutions for system (1), one usually defines a functional φ(u) on the Banach space
(3)E={∫u∈W1,2(ℝ,ℝn):E=+∫ℝ[a1(t)|u′(t)|2+a2(t)|u(t)|2]dt<+∞},
where a1,a2∈C(ℝ,(0,+∞)) associated with the coefficients p(t) and L(t) of system (1). And then one proves that φ possesses critical points on E which are homoclinic solutions of system (1). Thus, the nontrivial homoclinic solutions of system (1) which were studied in the existing work are actually a class of special solutions satisfying condition
(4)0<∫ℝ[(1+a1(t))|u′(t)|2+(1+a2(t))|u(t)|2]dt<+∞.
Similarly, the non-trivial homoclinic solutions of system (2) which were studied in the literature are also a class of special solutions satisfying condition
(5)0<∫ℝ[(1+b1(t))|x(t)|2+(1+b2(t))|y(t)|2]dt<+∞,
where b1,b2∈C(ℝ,(0,+∞)) associated with the potential H of system (2).
As mentioned earlier, the existence and multiplicity of homoclinic solutions for Hamiltonian systems have been studied extensively via critical point theory in recent years; various sufficient conditions for existence are established. However, as we know, there are no results on nonexistence of homoclinic solutions for Hamiltonian systems in the literature. For the simplest second-order Hamiltonian system,
(6)x′′(t)+q(t)x(t)=0
has no non-trivial homoclinic solutions as q(t)≡ constant, but when q(t)≢ constant, there seem to be no results on existence or non-existence of homoclinic solutions in the literature either.
In this paper, we consider the first-order nonlinear Hamiltonian system
(7)x′(t)=α(t)x(t)+f(t,y(t)),y′(t)=-g(t,x(t))-α(t)y(t),
where α(t) is locally Lebesgue integrable real-valued function defined on ℝ, f,g:ℝ2→ℝ. For every t∈ℝ, f(t,x) and g(t,x) are continuous on x in ℝ, and for every x∈ℝ, f(t,x) and g(t,x) are locally Lebesgue integrable real-valued functions on t.
For the sake of convenience, we give the following assumptions on f and g.
For any c≠0, meas{t∈ℝ:f(t,cexp(-∫0tα(s)ds))≠0}>0, and there exist a constant μ>1 and a locally Lebesgue integrable nonnegative function β(t) defined on ℝ such that
(8)0≤xf(t,x)≤β(t)|x|μ,∀(t,x)∈ℝ2.
g(t,0)=0 for t∈ℝ, and there exists a locally Lebesgue integrable nonnegative function γ0(t) defined on ℝ such that
(9)xg(t,x)≤γ0(t)|x|ν,∀(t,x)∈ℝ2,
where ν>1 and (1/μ)+(1/ν)=1.
Remark 1.
In case g(t,x)=γ(t)φ(x), where γ(t) is locally Lebesgue integrable real-valued function defined on ℝ, φ∈C(ℝ,ℝ), and satisfies that
(10)0≤xφ(x)≤|x|ν,∀(t,x)∈ℝ2,
then we can choose γ0(t)=γ+(t)=max{γ(t),0}.
Let u(t)=(x(t),y(t))⊤, H(t,x,y)=∫0xg(t,s)ds+α(t)xy+∫0yf(t,s)ds, and
(11)J=(01-10).
Then we can rewrite (7) as a standard first-order Hamiltonian system
(12)u′(t)=J∇H(t,u(t)).
There are two special forms of system (7) which have been dealt with extensively in the literature: one is the first-order linear Hamiltonian system
(13)x′(t)=α(t)x(t)+β(t)y(t),y′(t)=-γ(t)x(t)-α(t)y(t)
and the other is the first-order quasilinear Hamiltonian system
(14)x′(t)=α(t)x(t)+β(t)|y(t)|μ-2y(t),y′(t)=-γ(t)|x(t)|ν-2x(t)-α(t)y(t),
(see [21, 22] and the references therein), where μ,ν>1 and (1/μ)+(1/ν)=1, and β(t) and γ(t) are locally Lebesgue integrable real-valued functions defined on ℝ. In addition, the special forms of system (7) also contain many other well-known second-order differential equations such as the second-order linear differential equation
(15)(p(t)x′(t))′+q(t)x(t)=0,
the second-order half-linear differential equation
(16)[p(t)|x′(t)|r-2x′(t)]′+q(t)|x(t)|r-2x(t)=0,
and the second-order nonlinear differential equation
(17)[p(t)ϕ(x′(t))]′+h(t,x(t))=0,
where r>1, p(t) and q(t) are locally Lebesgue integrable real-valued functions defined on ℝ and p(t)>0, ϕ∈C(ℝ,ℝ), and h∈C(ℝ2,ℝ). Indeed, we can rewrite the above-mentioned second-order differential equations as the form of system (7). For example, let
(18)y(t)=p(t)|x′(t)|r-2x′(t).
Then (16) can be written as the form of (13):
(19)x′(t)=[p(t)]1/(1-r)|y(t)|(2-r)/(r-1)y(t),y′(t)=-q(t)|x(t)|r-2x(t),
where μ=r/(r-1), ν=r and α(t)=0, β(t)=[p(t)]1/(1-r), and γ(t)=q(t). If ϕ has an inverse ϕ-1, then let
(20)y(t)=p(t)ϕ(x′(t)).
Hence, (17) can be written as the form of (7):
(21)x′(t)=ϕ-1(y(t)p(t)),y′(t)=-h(t,x(t)),
where α(t)=0, f(t,x)=ϕ-1(x/p(t)), and g(t,x)=h(t,x).
In Sections 2 and 3, we will give some necessary conditions for existence of homoclinic solutions of systems (7) and (13), which satisfy conditions
(22)0<∫-∞+∞[|x(t)|ν+(1+β(t))|y(t)|μ]dt<+∞,0<∫-∞+∞[|x(t)|2+(1+β(t))|y(t)|2]dt<+∞,
respectively. These necessary conditions are actually Lyapunov-type inequalities, which generalize the classical Lyapunov inequality for system (6); see [21–25]. Taking advantage of these necessary conditions, we are able to establish some criteria for non-existence of homoclinic solutions of systems (7) and (13) in Section 4.
2. Lyapunov-Type Inequalities for System (7)
In this section, we will establish some Lyapunov-type inequalities for system (7). For the sake of convenience, we list some assumptions on α(t) and β(t) as follows:
Suppose that hypotheses (F), (G), (A0), (B0), and (B2) are satisfied. If system (7) has a solution (x(t),y(t)) satisfying
(25)0<∫-∞+∞[|x(t)|ν+(1+β(t))|y(t)|μ]dt<+∞,
then one has the following inequality:
(26)∫-∞+∞ζ(t)η(t)ζ(t)+η(t)γ0(t)dt≥1.
Proof.
Hypothesis (B2) implies that functions ζ(t) and η(t) are well defined on ℝ. Without loss of generality, we can assume that
(27)∫-∞+∞ζ(t)η(t)ζ(t)+η(t)γ0(t)dt<+∞.
It follows from (F), (25), and (B0) that
(28)liminft→-∞|x(t)|=liminft→+∞|x(t)|=0,(29)liminft→-∞|y(t)|=liminft→+∞|y(t)|=0,(30)∫-∞+∞f(τ,y(τ))y(τ)dτ≤∫-∞+∞β(τ)|y(τ)|μdτ<+∞.
Set A(t)={τ∈(-∞,t]:β(τ)>0} for t∈ℝ, and then it follows from (F) that
(31)[β(τ)]-ν/μ|f(τ,z)|ν=|z|-ν[β(τ)]-ν/μ|zf(τ,z)|ν≤|z|-ν[β(τ)]-ν/μ|β(τ)|z|μ|ν-1zf(τ,z)=zf(τ,z),τ∈A(t),t∈ℝ,z≠0.
Since f(τ,0)=0 for τ∈ℝ, it follows that
(32)[β(τ)]-ν/μ|f(τ,z)|ν≤zf(τ,z),τ∈A(t),t∈ℝ.
Hence, from (F), (23), (24), (30), (32), and the Hölder inequality, one has
(33)∫-∞t|f(τ,y(τ))|exp(∫τtα(s)ds)dτ=∫A(t)|f(τ,y(τ))|exp(∫τtα(s)ds)dτ≤[∫A(t)β(τ)exp(μ∫τtα(s)ds)dτ]1/μ×[∫A(t)[β(τ)]-ν/μ|f(τ,y(τ))|νdτ]1/ν≤[∫A(t)β(τ)exp(μ∫τtα(s)ds)dτ]1/μ×[∫A(t)f(τ,y(τ))y(τ)dτ]1/ν≤[∫-∞tβ(τ)exp(μ∫τtα(s)ds)dτ]1/μ×[∫-∞tf(τ,y(τ))y(τ)dτ]1/ν=[ζ(t)]1/ν[∫-∞tf(τ,y(τ))y(τ)dτ]1/ν<+∞,∀t∈ℝ,(34)∫t+∞|f(τ,y(τ))|exp(-∫tτα(s)ds)dτ≤[∫t+∞β(τ)exp(-μ∫tτα(s)ds)dτ]1/μ×[∫t+∞f(τ,y(τ))y(τ)dτ]1/ν=[η(t)]1/ν[∫t+∞f(τ,y(τ))y(τ)dτ]1/ν<+∞,∀t∈ℝ.
From (A0), (28), (33), (34), and the first equation of system (7), we have
(35)x(-∞)∶=limt→-∞x(t)=0=limt→+∞x(t)∶=x(+∞),(36)x(t)=∫-∞tf(τ,y(τ))exp(∫τtα(s)ds)dτ,∀t∈ℝ,(37)x(t)=-∫t+∞f(τ,y(τ))exp(-∫tτα(s)ds)dτ,∀t∈ℝ.
Combining (33) with (36), one has
(38)|x(t)|ν=|∫-∞tf(τ,y(τ))exp(∫τtα(s)ds)dτ|ν≤ζ(t)∫-∞tf(τ,y(τ))y(τ)dτ,∀t∈ℝ.
Similarly, it follows from (34) and (37) that
(39)|x(t)|ν=|∫t+∞f(τ,y(τ))exp(-∫τtα(s)ds)dτ|ν≤η(t)∫t+∞f(τ,y(τ))y(τ)dτ,∀t∈ℝ.
Hence, from (38) and (39), one has
(40)|x(t)|ν≤ζ(t)η(t)ζ(t)+η(t)∫-∞+∞f(τ,y(τ))y(τ)dτ,∀t∈ℝ.
Now, it follows from (27), (30), and (40) that
(41)∫-∞+∞γ0(t)|x(t)|νdt≤∫-∞+∞ζ(t)η(t)ζ(t)+η(t)γ0(t)dt∫-∞+∞f(t,y(t))y(t)dt<+∞.
By (29), we can choose two sequences {t-n}n=1∞ and {tn}n=1∞ such that
(42)-∞<⋯<t-3<t-2<t-1<t1<t2<t3<⋯<+∞,(43)limn→∞t-n=-∞,limn→∞tn=+∞,limn→∞y(t-n)=limn→∞y(tn)=0.
By (7), we obtain
(44)(x(t)y(t))′=f(t,y(t))y(t)-g(t,x(t))x(t).
Integrating the above equation from t-n to tn, we have
(45)∫t-ntng(t,x(t))x(t)dt=∫t-ntnf(t,y(t))y(t)dt+x(t-n)y(t-n)-x(tn)y(tn),n=1,2,3,….
Let n→∞ in the above equation, and using (30), (35), and (43) we obtain
(46)limn→∞∫t-ntng(t,x(t))x(t)dt=∫-∞+∞f(t,y(t))y(t)dt,
which, together with (41), implies that
(47)∫-∞+∞γ0(t)|x(t)|νdt≤∫-∞+∞ζ(t)η(t)ζ(t)+η(t)γ0(t)dt∫-∞+∞f(t,y(t))y(t)dt=[∫-∞+∞ζ(t)η(t)ζ(t)+η(t)γ0(t)dt]×limn→∞∫t-ntng(t,x(t))x(t)dt≤[∫-∞+∞ζ(t)η(t)ζ(t)+η(t)γ0(t)dt]×limn→∞∫t-ntnγ0(t)|x(t)|νdt=∫-∞+∞ζ(t)η(t)ζ(t)+η(t)γ0(t)dt∫-∞+∞γ0(t)|x(t)|νdt.
We claim that
(48)∫-∞+∞γ0(t)|x(t)|νdt>0.
If (48) is not true, then
(49)∫-∞+∞γ0(t)|x(t)|νdt=0.
From (F), (G), (46), and (49), we have
(50)0≤∫-∞+∞f(t,y(t))y(t)dt=limn→∞∫t-ntng(t,x(t))x(t)dt≤∫-∞+∞γ0(t)|x(t)|νdt=0,
which, together with (F), implies that
(51)f(t,y(t))y(t)=0,a.e.t∈ℝ.
Combining (36) with (51), we obtain that
(52)x(t)≡0,∀t∈ℝ,
which, together with (G) and the second equation of system (7), implies that
(53)y(t)=y(0)exp(-∫0tα(s)ds),∀t∈ℝ.
From (F), (51), and the above, one has
(54)y(t)≡0,∀t∈ℝ.
Both (52) and (54) contradict (25). Therefore, (48) holds. Hence, it follows from (47) and (48) that (26) holds.
Corollary 3.
Suppose that hypotheses (F), (G), (A1), (B0), and (B2) are satisfied. If system (7) has a solution (x(t),y(t)) satisfying (25), then one has the following inequalities:
(55)∫-∞+∞γ~(t)(∫-∞tβ~(τ)dτ∫t+∞β~(τ)dτ)ν/2μdt≥2exp(-ν2∫-∞+∞(|α(s)|+μ-1|ω(s)|)ds),(56)(∫-∞+∞β~(t)dt)1/μ(∫-∞+∞γ~+(t)dt)1/ν≥2exp(-12∫-∞+∞(|α(s)|+μ-1|ω(s)|)ds),(57)∫-∞+∞γ0(t)(∫-∞tβ(τ)dτ∫t+∞β(τ)dτ)ν/2μdt≥2exp(-ν2∫-∞+∞|α(s)|ds),(58)(∫-∞+∞β(t)dt)1/μ(∫-∞+∞γ0(t)dt)1/ν≥2exp(-12∫-∞+∞|α(s)|ds),
where ω∈L1(ℝ) is an arbitrary function and
(59)β~(t)=β(t)exp(∫t0tω(s)ds),γ~(t)=γ0(t)exp(-νμ∫t0tω(s)ds)
for some t0∈ℝ.
Proof.
(A1), (B0), and (B2) imply that (A0) and ∫-∞+∞β(s)ds<+∞. Since
(60)ζ(t)+η(t)≥2[ζ(t)η(t)]1/2,
then it follows from (23), (24), (26), (56), and (57) that
(61)1≤∫-∞+∞ζ(t)η(t)ζ(t)+η(t)γ0(t)dt≤12∫-∞+∞[ζ(t)η(t)]1/2γ0(t)dt=12∫-∞+∞γ0(t)[∫-∞tβ(τ)exp(μ∫τtα(s)ds)dτ×∫t+∞β(τ)×exp(-μ∫tτα(s)ds)dτ]ν/2μdt≤12[∫-∞+∞γ0(t)(∫-∞tβ(τ)dτ∫t+∞β(τ)dτ)ν/2μdt]×exp(ν2∫-∞+∞|α(s)|ds)≤12[∫-∞+∞γ~(t)(∫-∞tβ~(τ)dτ∫t+∞β~(τ)dτ)ν/2μdt]×exp(ν2∫-∞+∞(|α(s)|+μ-1|ω(s)|)ds),
which implies that (55) holds. Note that
(62)∫-∞tβ~(τ)dτ∫t+∞β~(τ)dτ≤14(∫-∞+∞β~(τ)dτ)2,
which, together with (55), yields that (56) holds. It follows from (55) and (56) that (57) and (58) hold.
In case hypothesis (B0) is replaced by (B1) in the proof of Theorem 2, then (40) is strict; that is,
(63)|x(t)|ν<ζ(t)η(t)ζ(t)+η(t)∫-∞+∞f(τ,y(τ))y(τ)dτ,∀t∈ℝ.
In fact, if (63) is not true, then there exists a t*∈ℝ such that
(64)|x(t*)|ν=ζ(t*)η(t*)ζ(t*)+η(t*)∫-∞+∞f(τ,y(τ))y(τ)dτ.
Hence, from (38), (39), and (64), we obtain
(65)|x(t*)|ν=ζ(t*)∫-∞t*f(τ,y(τ))y(τ)dτ,(66)|x(t*)|ν=η(t*)∫t*+∞f(τ,y(τ))y(τ)dτ.
It follows from (23), (38), and (65) that
(67)|∫-∞t*f(τ,y(τ))exp(∫τt*α(s)ds)dτ|ν=[∫-∞t*β(τ)exp(μ∫τt*α(s)ds)dτ]ν/μ×∫-∞t*f(τ,y(τ))y(τ)dτ,
which, together with the Hölder inequality, implies that there exists a constant c1 such that
(68)f(τ,y(τ))y(τ)=c1β(τ)exp(μ∫τt*α(s)ds),-∞<τ≤t*.
Similarly, it follows from (24), (39), (66), and the Hölder inequality that there exists a constant c2 such that
(69)f(τ,y(τ))y(τ)=c2β(τ)exp(-μ∫t*τα(s)ds),t*≤τ<+∞.
From (F), (68), and (69), one has that c1,c2≥0. If c1=c2=0, then f(τ,y(τ))y(τ)=0 for τ∈ℝ; it follows from (36) that x(t)=0 for t∈ℝ. Similar to the proof of (54), one has y(t)=0 for t∈ℝ, which contradicts (25). If c1+c2>0, then f(τ,y(τ))y(τ)>0 for τ∈(-∞,t*] or for τ∈[t*,+∞); it follows from (A0) and (36) that x(+∞)≠0, which contradicts (35). The above two cases show that (63) holds. Hence, in view of the proof of Theorem 2, we have the following theorem.
Theorem 4.
Suppose that hypotheses (F), (G), (A0), (B1), and (B2) are satisfied. If system (7) has a solution (x(t),y(t)) satisfying (25), then one has the following inequality:
(70)∫-∞+∞ζ(t)η(t)ζ(t)+η(t)γ0(t)dt>1,
where ζ(t) and η(t) are defined by (23) and (24), respectively.
Similar to the proof of Corollary 3, we can drive the following corollary from Theorem 4.
Corollary 5.
Suppose that hypotheses (F), (G), (A1), (B1), and (B2) are satisfied. If system (7) has a solution (x(t),y(t)) satisfying (25), then
(71)∫-∞+∞γ~(t)(∫-∞tβ~(τ)dτ∫t+∞β~(τ)dτ)ν/2μdt>2exp(-ν2∫-∞+∞(|α(s)|+μ-1|ω(s)|)ds),(∫-∞+∞β~(t)dt)1/μ(∫-∞+∞γ~(t)dt)1/ν>2exp(-12∫-∞+∞(|α(s)|+μ-1|ω(s)|)ds),∫-∞+∞γ0(t)(∫-∞tβ(τ)dτ∫t+∞β(τ)dτ)ν/2μdt>2exp(-ν2∫-∞+∞|α(s)|ds),(∫-∞+∞β(t)dt)1/μ(∫-∞+∞γ0(t)dt)1/ν>2exp(-12∫-∞+∞|α(s)|ds),
where β~(t) and γ~(t) are defined by (59).
Applying Theorem 4 and Corollary 5 to system (19) (i.e., (16)), we have immediately the following two corollaries.
Corollary 6.
Suppose that r>1, p(t)>0 for t∈ℝ and
(72)∫-∞+∞1[p(τ)]1/(r-1)dτ<+∞.
If (16) has a solution x(t) satisfying
(73)0<∫-∞+∞[|x(t)|r+[p(t)]1/(r-1)(1+p(t))|x′(t)|r]dt0<+∞,
then
(74)∫-∞+∞(∫-∞t[p(τ)]-1/(r-1)dτ)r-1(∫t+∞[p(τ)]-1/(r-1)dτ)r-1(∫-∞t[p(τ)]-1/(r-1)dτ)r-1+(∫t+∞[p(τ)]-1/(r-1)dτ)r-1×q+(t)dt>1,∫-∞+∞q+(t)(∫-∞t[p(τ)]-1/(r-1)dτ×∫t+∞[p(τ)]-1/(r-1)dτ)(r-1)/2dt>2.
Applying Theorem 4 to the second-order nonlinear differential equation (17) (i.e., system (21)), where α(t)=0, f(t,x)=ϕ-1(x/p(t)), and g(t,x)=h(t,x), we have the following corollary.
Corollary 7.
Suppose that r>1 and p(t)>0 for t∈ℝ, and that (72) and the following hypothesis are satisfied:
There exists a locally Lebesgue integrable nonnegative function γ0(t) defined on ℝ such that
(75)0≤xϕ-1(x)≤|x|r/(r-1),xh(t,x)≤γ0(t)|x|r,∀(t,x)∈ℝ2.
If (17) has a solution x(t) satisfying (73), then
(76)∫-∞+∞(∫-∞t[p(τ)]-1/(r-1)dτ)r-1(∫t+∞[p(τ)]-1/(r-1)dτ)r-1(∫-∞t[p(τ)]-1/(r-1)dτ)r-1+(∫t+∞[p(τ)]-1/(r-1)dτ)r-1×γ0(t)dt>1.
3. Lyapunov-Type Inequalities for System (13)
When μ=ν=2, assumption (B2) reduces to the following form:
∫-∞+∞β(τ)exp(-2∫0τα(s)ds)dτ<+∞.
Applying some results obtained in the last section to the first-order linear Hamiltonian system (13), we have immediately the following corollaries.
Corollary 8.
Suppose that hypotheses (A0), (B0), and (B2′) are satisfied. If system (13) has a solution (x(t),y(t)) satisfying
(77)0<∫-∞+∞[|x(t)|2+(1+β(t))|y(t)|2]dt<+∞,
then
(78)∫-∞+∞(+∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ)-1([∫-∞tβ(τ)exp(2∫τtα(s)ds)dτ]×[∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ])×(∫-∞tβ(τ)exp(2∫τtα(s)ds)dτ+∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ)-1)×γ+(t)dt≥1,∫-∞+∞γ+(t)(∫-∞tβ(τ)dτ∫t+∞β(τ)dτ)1/2dt≥2exp(-∫-∞+∞|α(s)|ds).
Corollary 9.
Suppose that hypotheses (A0), (B1), and (B2′) are satisfied. If system (13) has a solution (x(t),y(t)) satisfying (77), then(79)∫-∞+∞(+∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ)-1([∫-∞tβ(τ)exp(2∫τtα(s)ds)dτ]×[∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ])×(∫-∞tβ(τ)exp(2∫τtα(s)ds)dτ+∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ)-1)×γ+(t)dt>1,∫-∞+∞γ+(t)(∫-∞tβ(τ)dτ∫t+∞β(τ)dτ)1/2dt>2exp(-∫-∞+∞|α(s)|ds).
Corollary 10.
Suppose that p(t)>0 for t∈ℝ and that
(80)∫-∞+∞dτp(τ)<+∞.
If (15) has a solution x(t) satisfying
(81)0<∫-∞+∞[|x(t)|2+p(t)(1+p(t))|x′(t)|2]dt<+∞,
then
(82)∫-∞+∞q+(t)(∫-∞tdτp(τ)∫t+∞dτp(τ))dt>∫-∞+∞dτp(τ).
Corollary 11.
Suppose that p(t)>0 for t∈ℝ and that (80) and the following hypothesis are satisfied:
There exists a locally Lebesgue integrable nonnegative function γ0(t) defined on ℝ such that
(83)0≤xϕ-1(x)≤|x|2,xh(t,x)≤γ0(t)|x|2,∀(t,x)∈ℝ2.
If (17) has a solution x(t) satisfying (81), then
(84)∫-∞+∞γ0(t)(∫-∞tdτp(τ)∫t+∞dτp(τ))dt>∫-∞+∞dτp(τ).
4. Nonexistence of Homoclinic Solutions
Applying the results obtained in Sections 2 and 3, we can drive the following criteria for non-existence of homoclinic solutions of systems (7) and (13) immediately.
Corollary 12.
Suppose that hypotheses (F), (G), (A0), (B0), and (B2) are satisfied. If one of the conditions
(85)∫-∞+∞ζ(t)η(t)ζ(t)+η(t)γ0(t)dt<1,(86)∫-∞+∞γ0(t)(∫-∞tβ(τ)dτ∫t+∞β(τ)dτ)ν/2μdt<2exp(-ν2∫-∞+∞|α(s)|ds),(87)(∫-∞+∞β(t)dt)1/μ(∫-∞+∞γ0(t)dt)1/ν<2exp(-12∫-∞+∞|α(s)|ds)
holds, then system (7) has no solution (x(t),y(t)) satisfying
(88)0<∫-∞+∞[|x(t)|ν+(1+β(t))|y(t)|μ]dt<+∞.
Corollary 13.
Suppose that hypotheses (F), (G), (A0), (B1), and (B2) are satisfied. If one of the conditions
(89)∫-∞+∞ζ(t)η(t)ζ(t)+η(t)γ0(t)dt≤1,∫-∞+∞γ0(t)(∫-∞tβ(τ)dτ∫t+∞β(τ)dτ)ν/2μdt≤2exp(-ν2∫-∞+∞|α(s)|ds),(∫-∞+∞β(t)dt)1/μ(∫-∞+∞γ0(t)dt)1/ν≤2exp(-12∫-∞+∞|α(s)|ds)
holds, then system (7) has no solution (x(t),y(t)) satisfying (86).
Corollary 14.
Suppose that hypotheses (A0), (B0), and (B2′) are satisfied. If
(90)∫-∞+∞(+∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ)-1([∫-∞tβ(τ)exp(2∫τtα(s)ds)dτ]×[∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ])×(∫-∞tβ(τ)exp(2∫τtα(s)ds)dτ+∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ)-1)×γ0(t)dt<1
or
(91)∫-∞+∞γ+(t)(∫-∞tβ(τ)dτ∫t+∞β(τ)dτ)1/2dt<2exp(-∫-∞+∞|α(s)|ds)
holds, then system (13) has no solution (x(t),y(t)) satisfying
(92)0<∫-∞+∞[|x(t)|2+(1+β(t))|y(t)|2]dt<+∞.
Corollary 15.
Suppose that hypotheses (A0), (B1), and (B2′) are satisfied. If
(93)∫-∞+∞(+∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ)-1([∫-∞tβ(τ)exp(2∫τtα(s)ds)dτ]×[∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ])×(∫-∞tβ(τ)exp(2∫τtα(s)ds)dτ+∫t+∞β(τ)exp(-2∫tτα(s)ds)dτ)-1)×γ+(t)dt≤1
or
(94)∫-∞+∞γ+(t)(∫-∞tβ(τ)dτ∫t+∞β(τ)dτ)1/2dt≤2exp(-∫-∞+∞|α(s)|ds)
holds, then system (13) has no solution (x(t),y(t)) satisfying (92).
Corollary 16.
Suppose that p(t)>0 for t∈ℝ and that (80) holds. If
(95)∫-∞+∞q+(t)(∫-∞tdτp(τ)∫t+∞dτp(τ))dt≤∫-∞+∞dτp(τ),
then (15) has no solution x(t) satisfying (81).
Corollary 17.
Suppose that p(t)>0 for t∈ℝ and that (80) and (H2) are satisfied. If
(96)∫-∞+∞γ0(t)(∫-∞tdτp(τ)∫t+∞dτp(τ))dt≤∫-∞+∞dτp(τ),
then (17) has no solution x(t) satisfying (81).
Example 18.
Consider the second-order nonlinear differential equation
(97)[(1+t2)x′(t)]′+q(t)x(t)[1+sin2x(t)]=0,
where q(t) is locally Lebesgue integrable real-valued function defined on ℝ. In view of Corollary 16, if
(98)∫-∞+∞[π24-(arctant)2]q+(t)dt≤π2,
then (97) has no solution x(t) satisfying
(99)0<∫-∞+∞[|x(t)|2+(1+t2)2|x′(t)|2]dt<+∞.
Acknowledgments
This work is partially supported by the NNSF (no. 11201138) of China Hunan Provincial Natural Science Foundation (no. 11JJ2005), and the Scientific Research Fund of Hunan Provincial Education Department (12B034).
PoincaréH.1897–1899Paris, FranceGauthier-VillarsAmbrosettiA.Coti ZelatiV.Multiple homoclinic orbits for a class of conservative systems199389177194MR1229052ZBL0806.58018AmbrosettiA.RabinowitzP. H.Dual variational methods in critical point theory and applications197314349381MR0370183ZBL0273.49063ZelatiV. C.EkelandI.SeraE.A variational approach to homoclinic orbits in Hamiltonian systems1990288113316010.1007/BF01444526MR1070929ZBL0731.34050ZelatiV. C.RabinowitzP. H.Homoclinic orbits for second order Hamiltonian systems possessing superquadratic potentials19914469372710.2307/2939286MR1119200ZBL0744.34045SéréE.Existence of infinitely many homoclinic orbits in Hamiltonian systems19922091274210.1007/BF02570817MR1143210CaldiroliP.MontecchiariP.Homoclinic orbits for second order Hamiltonian systems with potential changing sign19941297129MR1280118ZBL0867.70012DingY.JeanjeanL.Homoclinic orbits for a nonperiodic Hamiltonian system2007237247349010.1016/j.jde.2007.03.005MR2330955ZBL1117.37032DingY.LeeC.Existence and exponential decay of homoclinics in a nonperiodic superquadratic Hamiltonian system200924672829284810.1016/j.jde.2008.12.013MR2503024ZBL1162.70014FeiG.The existence of homoclinic orbits for Hamiltonian systems with the potentials changing sign1996174403410MR1441653ZBL0871.58036IzydorekM.JanczewskaJ.Homoclinic solutions for a class of the second order Hamiltonian systems2005219237538910.1016/j.jde.2005.06.029MR2183265ZBL1080.37067LvY.TangC.-L.Existence of even homoclinic orbits for second-order Hamiltonian systems20076772189219810.1016/j.na.2006.08.043MR2331869ZBL1121.37048OmanaW.WillemM.Homoclinic orbits for a class of Hamiltonian systems19925511151120MR1171983ZBL0759.58018RabinowitzP. H.Homoclinic orbits for a class of Hamiltonian systems19901141-2333810.1017/S0308210500024240MR1051605ZBL0705.34054RabinowitzP. H.TanakaK.Some results on connecting orbits for a class of Hamiltonian systems1991206347349910.1007/BF02571356MR1095767ZBL0707.58022SzulkinA.ZouW.Homoclinic orbits for asymptotically linear Hamiltonian systems20011871254110.1006/jfan.2001.3798MR1867339ZBL0984.37072TanakaK.Homoclinic orbits in a first order superquadratic Hamiltonian system: convergence of subharmonic orbits199194231533910.1016/0022-0396(91)90095-QMR1137618ZBL0787.34041TangX. H.LinX.Homoclinic solutions for a class of second-order Hamiltonian systems2009354253954910.1016/j.jmaa.2008.12.052MR2515234ZBL1185.34056TangX. H.LinX.Existence of infinitely many homoclinic orbits in Hamiltonian systems201114151103111910.1017/S0308210509001346MR2838370ZBL1237.37044TangX. H.LinX.Infinitely many homoclinic orbits for Hamiltonian systems with indefinite sign subquadratic potentials201174176314632510.1016/j.na.2011.06.010MR2833414ZBL1225.37071TangX.-H.ZhangM.Lyapunov inequalities and stability for linear Hamiltonian systems2012252135838110.1016/j.jde.2011.08.002MR2852210ZBL1260.39025TangX. H.ZhangQ.-M.ZhangM.Lyapunov-type inequalities for the first-order nonlinear Hamiltonian systems20116293603361310.1016/j.camwa.2011.09.011MR2844910ZBL1236.34040EliasonS. B.A Lyapunov inequality for a certain second order non-linear differential equation19702461466MR0267191ZBL0201.11802EliasonS. B.Lyapunov type inequalities for certain second order functional differential equations197427180199MR035015210.1137/0127015ZBL0292.34077GuseinovG. Sh.KaymakçalanB.Lyapunov inequalities for discrete linear Hamiltonian systems2003456-913991416Advances in difference equations, IV10.1016/S0898-1221(03)00095-6MR2000605ZBL1055.39029