Loading [MathJax]/jax/element/mml/optable/BasicLatin.js
Research article

On the variational principle and applications for a class of damped vibration systems with a small forcing term

  • Received: 04 May 2023 Revised: 16 June 2023 Accepted: 20 June 2023 Published: 12 July 2023
  • MSC : 34C25, 58E30, 58E50

  • This paper is dedicated to studying the existence of periodic solutions to a new class of forced damped vibration systems by the variational method. The advantage of this kind of system is that the coefficient of its second order term is a symmetric N×N matrix valued function rather than the identity matrix previously studied. The variational principle of this problem is obtained by using two methods: the direct method of the calculus of variations and the semi-inverse method. New existence conditions of periodic solutions are created through several auxiliary functions so that two existence theorems of periodic solutions of the forced damped vibration systems are obtained by using the least action principle and the saddle point theorem in the critical point theory. Our results improve and extend many previously known results.

    Citation: Shaomin Wang, Cunji Yang, Guozhi Cha. On the variational principle and applications for a class of damped vibration systems with a small forcing term[J]. AIMS Mathematics, 2023, 8(9): 22162-22177. doi: 10.3934/math.20231129

    Related Papers:

    [1] Adnan Khaliq, Mujeeb ur Rehman . Fixed point theorem combined with variational methods for a class of nonlinear impulsive fractional problems with derivative dependence. AIMS Mathematics, 2021, 6(2): 1943-1953. doi: 10.3934/math.2021118
    [2] Jie Xie, Xingyong Zhang, Cuiling Liu, Danyang Kang . Existence and multiplicity of solutions for a class of damped-like fractional differential system. AIMS Mathematics, 2020, 5(5): 4268-4284. doi: 10.3934/math.2020272
    [3] Song Wang, Xiao-Bao Shu, Linxin Shu . Existence of solutions to a class of damped random impulsive differential equations under Dirichlet boundary value conditions. AIMS Mathematics, 2022, 7(5): 7685-7705. doi: 10.3934/math.2022431
    [4] Feliz Minhós, Nuno Oliveira . On periodic Ambrosetti-Prodi-type problems. AIMS Mathematics, 2023, 8(6): 12986-12999. doi: 10.3934/math.2023654
    [5] Hai-yun Deng, Jue-liang Zhou, Yu-bo He . Existence of periodic solutions for a class of (ϕ1,ϕ2)-Laplacian discrete Hamiltonian systems. AIMS Mathematics, 2023, 8(5): 10579-10595. doi: 10.3934/math.2023537
    [6] Wei Guo, Jinfu Yang, Jiafeng Zhang . Existence results of nontrivial solutions for a new p(x)-biharmonic problem with weight function. AIMS Mathematics, 2022, 7(5): 8491-8509. doi: 10.3934/math.2022473
    [7] Radu Precup, Andrei Stan . Stationary Kirchhoff equations and systems with reaction terms. AIMS Mathematics, 2022, 7(8): 15258-15281. doi: 10.3934/math.2022836
    [8] Nasser. A. Saeed, Amal Ashour, Lei Hou, Jan Awrejcewicz, Faisal Z. Duraihem . Time-delayed control of a nonlinear self-excited structure driven by simultaneous primary and 1:1 internal resonance: analytical and numerical investigation. AIMS Mathematics, 2024, 9(10): 27627-27663. doi: 10.3934/math.20241342
    [9] A. Palanisamy, J. Alzabut, V. Muthulakshmi, S. S. Santra, K. Nonlaopon . Oscillation results for a fractional partial differential system with damping and forcing terms. AIMS Mathematics, 2023, 8(2): 4261-4279. doi: 10.3934/math.2023212
    [10] Tainian Zhang, Zhixue Luo . Optimal harvesting for a periodic competing system with size structure in a polluted environment. AIMS Mathematics, 2022, 7(8): 14696-14717. doi: 10.3934/math.2022808
  • This paper is dedicated to studying the existence of periodic solutions to a new class of forced damped vibration systems by the variational method. The advantage of this kind of system is that the coefficient of its second order term is a symmetric N×N matrix valued function rather than the identity matrix previously studied. The variational principle of this problem is obtained by using two methods: the direct method of the calculus of variations and the semi-inverse method. New existence conditions of periodic solutions are created through several auxiliary functions so that two existence theorems of periodic solutions of the forced damped vibration systems are obtained by using the least action principle and the saddle point theorem in the critical point theory. Our results improve and extend many previously known results.



    Consider the following forced damped vibration systems:

    {(A(t)˙u(t))'+A(t)q(t)˙u(t)=F(t,u(t))+f(t),u(0)u(T)=˙u(0)eQ(T)˙u(T)=0,   a.e. t[0,T]. (1.1)

    where T>0, qL1(0,T;R), Q(t)=t0q(s)ds, fL1(0,T;RN), A(t)=[aij(t)] is an invertible symmetric N×N matrix-valued function defined in [0,T] with aijC([0,T]) for all i,j=1,2,N and there exists a positive constant λ such that λ|ξ|2(A(t)ξ,ξ),  for all ξRN, a.e. t[0,T] and F:[0,T]×RNR satisfies the following assumption:

    (A) F(t,x) is measurable in t for every xRN and continuously differentiable in x for a.e. t[0,T], and there exist aC(R+,R+), bL1(0,T;R+) such that

    |F(t,x)|a(|x|)b(t), |F(t,x)|a(|x|)b(t) ,for  all xRNand  a.e. t[0,T].

    When q(t)0, A(t)=IN×N and f(t)0, the forced damped vibration systems [i.e., (1.1)] reduce to the following classical second order non-autonomous Hamiltonian systems:

    {¨u(t)=F(t,u(t)), u(0)u(T)=˙u(0)˙u(T)=0. (1.2)

    By the variational method, many existence results have been obtained under some suitable conditions in the last two decades. The readers may refer to [1,2,3,4,5,6,7,8,9,10,11,12] for more relevant results. In particular, Wang and Zhang [4] gave the following two existence theorems of periodic solutions of problem (1.2).

    Theorem A. Suppose that F satisfies assumption (A) and the following conditions:

    (H1) There exist constants C0>0, K10, K20, α[0,1) and a non-negative function hC([0, +); [0, +)) with the properties:

    (ⅰ) (i)h(s)h(t       st,s,t[0,+), 

    (ⅱ) h(s+t)C0(h(s+ h(t))        s,t[0,+), 

    (ⅲ) 0h(t)K1tα+K2       t[0,+), 

    (ⅳ) h(t)+ as  t+.

    Moreover, there exist fL1(0,T;R+) and gL1(0,T;R+) such that

    |F(t,x)|f(t)h(|x|)+g(t), for xRN,  a.e. t[0,T].

    (H2) There exists a non-negative function hC([0, +); [0, +)) which satisfies the conditions (ⅰ)–(ⅳ) and 1h2(|x|)T0F(t,x)dt+, as |x|+.

    Then problem (1.2) has at least one solution which minimizes the functional φ on H1T.

    Theorem B. Suppose that (H1) and assumption (A) hold. Assume that

    (H3) 1h2(|x|)T0F(t,x)dtas |x|+. Then problem (1.2) has at least one solution in H1T.

    With the increase in research, scholars began to study a more general form of Hamiltonian systems: damped vibration systems. In damped vibration, due to the need of the system overcoming resistance, the displacement and energy of vibration continuously reduced and their decreasing trend is correlated to factors such as the natural frequency and damping coefficient of the system. Therefore, the damped vibration system is greatly based on physics and can be one of the important mathematical models.

    Wu [13] studied the existence of periodic solutions of the following damped vibration systems:

    {¨u(t)+q(t)˙u(t)=A(t)u(t)+F(t,u(t)),u(0)u(T)=˙u(0)eQ(T)˙u(T)=0. (1.3)

    where q(t)˙u(t) is called damping term. The systems [i.e., (1.3)] are called damped vibration systems in physics. Wu put forward the variational principle of problem (1.3) for the first time and studied further the existence of periodic solution of problem (1.3). Subsequently, in case A(t)=0, Wang [14] studied the existence of periodic solution of the corresponding system.

    In addition, the vibration of a nonlinear vibration system under the action of a periodic dynamic force f(t) is called forced vibration. Take the spring oscillator model as an example. Suppose that the spring oscillator is subjected to both resistance γdxdt and dynamic force F0cosωt, and then Dynamic equation of the spring oscillator is

    md2xdt2=kxγdxdt+F0cosωt.

    This equation is a special forced damped vibration system. Another example is the famous Duffing oscillator model: Assume that it is subjected to both resistance c˙x and dynamic force fcosωt, and then the dynamic equation of the forced damped vibration of the Duffing oscillator is

    md2xdt2+c˙x+k(x+βx3)=fcosωt.

    The forced damped vibration systems [i.e., (1.1)] we have studied are more general than the two equations above. Therefore, the systems [i.e., (1.1)] are proved to not only have a very strong physical background but also be a more general class of new systems.

    Generally, a nonlinear vibration system is complex and it is difficult to get a strong solution to a differential equation. In recent years, the variational method has been used by many scholars to study the existence of solutions of differential equations, such as the classical second order non-autonomous Hamiltonian systems [i.e., (1.2)] (see [1,2,3,4,5,6,7,8,9,10,11,12]), the damped vibration systems [i.e., (1.3)] (See [13,14]) and the damped random impulsive differential equations under Dirichlet boundary value conditions (See [15,16,17]). The variational principle, including the Hamilton principle, is widely used in the nonlinear vibration theory. For the case of Hamiltonian-based frequency formulation for nonlinear oscillators (See [18]), its Hamilton principle is established by the semi-inverse method.

    Inspired by [4,13], we obtain a new class of forced damped vibration systems [i.e., (1.1)] and decide to study the existence of periodic solutions of this problem by the variational method. We explore, in depth, the existence of variational construction for problem (1.1) and study further the existence of periodic solutions of it under some solvability conditions by following the least action principle and the Saddle Point Theorem 4.7 in [3], and obtain two new existence theorems.

    Let us suppose H1T = u:[0,T]RN|u is absolutely continuous, u(0)=u(T), ˙uL2(0,TRN)} with the inner product

    u,v=T0(u(t),v(t))dt+T0(˙u(t),˙v(t))dt,for any  u,vH1T,

    where uH1T and || are the usual inner product and norm of RN. The corresponding norm is defined by

    u=(T0|u(t)|2dt+T0|˙u(t)|2dt)12,for uH1T.

    Then, H1T is obviously a Hilbert space.

    Set

    u0=(T0eQ(t)(A(t)˙u(t),˙u(t))dt+T0eQ(t)(u(t),u(t))dt)12, for uH1T.

    Obviously, the norm 0 is equivalent to the usual one on H1T. The proof is similar to the corresponding parts in [19].

    Let ˜H1T={uH1T|T0udt=0}, it is easy to know that ˜H1T is a subset of H1T, and H1T = RN˜H1T. It follows from Proposition 1.3 in [3] that

    T0|u(t)|2dtT24π2T0|˙u(t)|2dt,for every u˜H1T (Wirtingers inequality),

    and

    u2T12T0|˙u(t)|2dt,for every u˜H1T (Sobolevs inequality).

    Hence,

    u2(1+T24π2)T0|˙u(t)|2dt, for every u˜H1T. (2.1)

    Define the functional φ(u) on H1T by

    φ(u)=12T0eQ(t)(A(t)˙u(t),˙u(t))dt+T0eQ(t)F(t,u(t))dt+T0eQ(t)(f(t),u(t))dt. (2.2)

    We have the following facts.

    Theorem 2.1. The functional φ(u) is continuously differentiable and weak lower semi-continuous on H1T.

    Proof. Set L(t,x,y)=eQ(t)[12(A(t)y,y)+F(t,x)+(f(t),x)] for all x,yRN and t[0,T]. Then L(t,x,y) satisfies all assumptions of Theorem 1.4 in [3]. Hence, by Theorem 1.4 in [3], we know that the functional φ(u) is continuously differentiable on H1T and

    (φ(u),v)=T0eQ(t)(A(t)˙u(t),˙v(t))dt+T0eQ(t)(F(t,u(t)),v(t))dt+T0eQ(t)(f(t),v(t))dt,

    for all u,vH1T. Moreover, the proof for the weak lower semi-continuity of φ(u) is similar to the corresponding parts in [3, P12-13].

    Theorem 2.2. If uH1T is a solution of the Euler equation φ(u)=0, then u is a solution of problem (1.1).

    Proof. Since φ(u)=0, then

    0=(φ(u),v)=T0eQ(t)(A(t)˙u(t),˙v(t))dt+T0eQ(t)(F(t,u(t)),v(t))dt+T0eQ(t)(f(t),v(t))dt,

    for all u,vH1T. i.e.,

    T0eQ(t)(A(t)˙u(t),˙v(t))dt=T0eQ(t)(F(t,u(t))+f(t),v(t))dt,for all vH1T.

    By the Fundamental Lemma and Remarks in [3, P6-7], we know that eQ(t)A(t)˙u(t) has a weak-derivative, and

    (eQ(t)A(t)˙u(t))=eQ(t)[F(t,u(t))+f(t)],  a.e. t[0,T]. (2.3)
    eQ(t)A(t)˙u(t)=t0eQ(s)[F(s,u(s))+f(s)]ds+C,  a.e. t[0,T]. (2.4)
    T0eQ(t)[F(t,u(t))+f(t)]dt=0, (2.5)

    where C is a constant. We identify the equivalence class eQ(t)A(t)˙u(t) and its continuous representation

    t0eQ(s)[F(s,u(s))+f(s)]ds+C.

    Then, by (2.4), we have

    eQ(0)A(0)˙u(0)=0.

    i.e., ˙u(0)=0.

    By (2.4) and (2.5), one has

    eQ(T)A(T)˙u(T)=0.

    i.e., eQ(T)˙u(T)=0.

    By the existence of ˙u(t), we draw a conclusion similar to (2.5), that is

    T0˙u(t)dt=0.

    i.e., u(0)u(T)=0.

    Therefore, u satisfies the following periodic boundary condition

    ˙u(0)eQ(T)˙u(T)=u(0)u(T)=0.

    Moreover, by (2.3), u satisfies the following forced damped vibration equation

    (A(t)˙u(t))'+A(t)q(t)˙u(t)=F(t,u(t))+f(t),  a.e. t[0,T].

    Hence, u is a solution of problem (1.1). This completes the proof.

    From the proof of Theorems 2.1 and 2.2, it can be seen that the variational principle of problem (1.1) is indeed the φ(u)[i.e., (2.2)] we defined above.

    In fact, we can also directly derive the variational principle of problem (1.1) by using the semi-inverse method [18]. The derivation process is as follows.

    In case q(t)0, we can easily obtain the following variational principle:

    φ1(u)=T012(A(t)˙u(t),˙u(t))+F(t,u(t))+(f(t),u(t))dt.

    In order to obtain the variational principle of problem (1.1), we introduce an integrating factor g(t) which is an unknown function of time, and consider the following integral:

    φ2(u)=T0g(t)[12(A(t)˙u(t),˙u(t))+F(t,u(t))+(f(t),u(t))]+G(u,ut,utt,)} dt, (2.6)

    where G is an unknown function of u and/or its derivatives. The semi-inverse method is to identify such g and G that the stationary condition of Eq (2.6) satisfies problem (1.1). The Euler–Lagrange equation of Eq (2.6) reads

    g(t)(F(t,u(t))+f(t))(g(t)[(A(t)˙u(t)])'+δGδu=0, (2.7)

    where δGδu is called variational derivative [20,21,22] defined as

    δGδu=GutGut+2t2Gutt.

    We re-write Eq (2.7) in the form

    gg(A(t)˙u(t))+(A(t)˙u(t))'=(F(t,u(t))+f(t))+1gGu. (2.8)

    Comparing Eq (2.8) with problem (1.1), we set

    gg=q(t),Gu=0.

    Therefore, we have

    g=exp t0q(s)ds=eQ(t),G=0.

    Consequently, we obtain the needed variational principle for problem (1.1), which reads

    φ2(u)=T0eQ(t)[12(A(t)˙u(t),˙u(t))+F(t,u(t))+(f(t),u(t))]dt.

    Obviously, φ2(u)=φ(u).

    In convenience, we set

    d1=max

    Theorem 3.1. Let F(t, x) = {F_1}(t, x) + {F_2}(x) , suppose that {F_1}{\text{(}}t, x{\text{)}} and {F_2}{\text{(}}x{\text{)}} satisfy assumption (A) and the following conditions:

    {\text{(}}{H_1}{\text{)}} There exist constants {C_1} \gt 0, {K_1} \geqslant 0, \alpha \in (\frac{1}{2}, 1) and a non-negative function {h_1} \in C{\text{([0, }} + \infty {\text{); [0, }} + \infty {\text{))}} with the properties:

    ({\rm{i}})\ {h_1}{\text{(}}s{\text{)}} \leqslant {h_1}{\text{(}}t{\text{) }}\ \ \ \ \ \forall s \leqslant t, \; \; \; s, t \in {\text{[}}0, + \infty {\text{)}},

    ({\rm{ii}})\ {h_1}{\text{(}}s + t{\text{)}} \leqslant {C_1}{\text{(}}{h_1}{\text{(}}s{\text{) }} + {\text{ }}{h_1}{\text{(}}t{\text{)) }}\ \ \ \ \ \forall s, t \in {\text{[}}0, + \infty {\text{)}},

    ({\rm{iii}})\ \mathop {{\text{lim}}}\limits_{s \to + \infty } \frac{{{h_1}{\text{(}}s{\text{)}}}}{{{s^\alpha }}} \leqslant {K_1}.

    Moreover, there exist {r_1}, {r_2} \in {L^1}{\text{(}}0, T; {R^ + }{\text{)}} such that

    \left| {\nabla {F_1}{\text{(}}t, x{\text{)}}} \right| \leqslant {r_1}{\text{(}}t{\text{)}}{h_1}{\text{(}}\left| x \right|{\text{)}} + {r_2}{\text{(}}t{\text{)}} , \ for\ {K_2} \gt 0 , \ \ a.e.{\text{ }}t \in {\text{[}}0, T{\text{]}} .

    {\text{(}}{H_2}{\text{)}} There exist a constant {K_2} \gt 0 and two functions k \in {L^1}{\text{(0}}, T; {R^ + }{\text{)}} with \int_0^T {k(t)dt} \lt \frac{{3\lambda {d_2}}}{{{d_1}{K_2}T}} and {h_2} \in C{\text{([0, }} + \infty {\text{); [0, }} + \infty {\text{))}} which is non-decreasing, such that

    {\text{(}}\nabla {F_2}{\text{(}}x{\text{)}} - \nabla {F_2}{\text{(}}y{\text{)}}, x - y{\text{)}} \leqslant k{\text{(}}t{\text{)}}{h_2}{\text{(}}|x - y|{\text{)}} , \ for\ x, y \in {R^N} , \ \ a.e.{\text{ }}t \in {\text{[}}0, T{\text{]}} ,

    and

    \mathop {\lim \sup }\limits_{s \to + \infty } \frac{{{h_2}(s)}}{{{s^2}}} \leqslant {K_2} .

    {\text{(}}{H_3}{\text{)}} \mathop {\lim }\limits_{|x| \to + \infty } \frac{1}{{|x{|^{2\alpha }}}}\int_0^T {{e^{Q(t)}}F(t, x)dt} \gt \frac{{C_1^2d_1^2TK_1^2}}{{3\lambda {d_2}}}{(\int_0^T {{r_1}(t)dt})^2}.

    Then problem (1.1) has at least one solution which minimizes the functional \varphi {\text{(}}u{\text{)}} on H_T^1 .

    Proof. It is clear that

    \frac{1}{2}\int_0^T {{e^{Q(t)}}(A(t)\dot u(t), \dot u(t))dt} \geqslant \frac{1}{2}\int_0^T {\lambda {d_2}|\dot u(t){|^2}dt} = \frac{1}{2}\lambda {d_2}\left\| {\dot u} \right\|_2^2 . (3.1)

    It follows from condition {\text{(}}{H_1}{\text{)}} and Sobolev's inequality that

    \begin{array}{l}\left| {\int_0^T {{e^{Q(t)}}[{F_1}(t, u(t)) - {F_1}(t, \bar u)]dt} } \right|\\ = \left| {\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}\int_0^1 {\left( {\nabla {F_1}{\text{(}}t, \bar u + s\tilde u{\text{(}}t{\text{)}}, \tilde u{\text{(}}t{\text{)}}} \right)dsdt} } } \right| \\ \leqslant \int_0^T {\int_0^1 {{e^{Q{\text{(}}t{\text{)}}}}{r_1}{\text{(}}t{\text{)}}{h_1}{\text{(}}|\bar u + s\tilde u{\text{(}}t{\text{)}}|{\text{)}} \cdot |\tilde u{\text{(}}t{\text{)}}|dsdt} + \int_0^T {\int_0^1 {{e^{Q{\text{(}}t{\text{)}}}}{r_2}{\text{(}}t{\text{)}}|\tilde u{\text{(}}t{\text{)}}|dsdt} } }\\ \leqslant \int_0^T {\int_0^1 {{e^{Q(t)}}{r_1}(t){C_1}({h_1}(|\bar u|) + {h_1}(s|\tilde u(t)|))|\tilde u(t)|dsdt} + \int_0^T {{e^{Q(t)}}{r_2}(t)|\tilde u(t)|dt} } \\ \leqslant \int_0^T {{e^{Q(t)}}{r_1}(t){C_1}({h_1}(|\bar u|) + {h_1}(|\tilde u(t)|))|\tilde u(t)|dt} + {\left\| {\tilde u} \right\|_\infty }\int_0^T {{e^{Q(t)}}{r_2}(t)dt} \\ \leqslant {C_1}{d_1}({h_1}(|\bar u|) + {h_1}({\left\| {\tilde u} \right\|_\infty })){\left\| {\tilde u} \right\|_\infty }\int_0^T {{r_1}(t)dt} + {C_2}(\int_0^T {|\dot u(t){|^2}dt{)^{\frac{1}{2}}}} \\ = {C_1}{d_1}{h_1}(|\bar u|){\left\| {\tilde u} \right\|_\infty }\int_0^T {{r_1}(t)dt} + {C_1}{d_1}{h_1}({\left\| {\tilde u} \right\|_\infty }){\left\| {\tilde u} \right\|_\infty }\int_0^T {{r_1}(t)dt} + {C_2}(\int_0^T {|\dot u(t){|^2}dt{)^{\frac{1}{2}}}} \\ \leq C_1 d_1\left(\frac{3 \lambda d_2}{C_1 d_1 T}\|\tilde{u}\|_{\infty}^2+\frac{C_1 d_1 T}{3 \lambda d_2} h_1^2(|\bar{u}|)\left(\int_0^T r_1(t) d t\right)^2\right)+C_1 d_1\left(K_1\|\tilde{u}\|_{\infty}^\alpha+C_3\right)\|\tilde{u}\|_{\infty} \int_0^T r_1(t) d t\\ + {C_2}(\int_0^T {|\dot u{\text{(}}t{\text{)}}{|^2}dt{)^{\frac{1}{2}}}}\\ \leqslant \frac{{\lambda {d_2}}}{4}\left\| {\dot u} \right\|_2^2 + \frac{{C_1^2d_1^2T}}{{3\lambda {d_2}}}{{\text{(}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} {\text{)}}^2}h_1^2{\text{(}}|\bar u|{\text{)}} + {C_1}{d_1}{K_1}{(\frac{T}{{12}})^{\frac{{\alpha + 1}}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} \left\| {\dot u} \right\|_2^{\alpha + 1}\\ + {\text{(}}{C_1}{C_3}{d_1}{(\frac{T}{{12}})^{\frac{1}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} + {C_2}){\left\| {\dot u} \right\|_2}. \end{array} (3.2)

    It follows from the condition {\text{(}}{H_2}{\text{)}} and Sobolev's inequality that

    \begin{align*} \int_0^T {{e^{Q(t)}}[{F_2}(u(t)) - {F_2}(\bar u)]dt} = &\int_0^T {{e^{Q(t)}}\int_0^1 {(\nabla F{}_2(\bar u + s\tilde u(t)), \tilde u(t))dsdt} }\\ & = \int_0^T {{e^{Q(t)}}\int_0^1 {(\nabla F{}_2(\bar u + s\tilde u(t)) - \nabla {F_2}(\bar u), \tilde u(t))dsdt} } \\ & = \int_0^T {{e^{Q(t)}}\int_0^1 {\frac{1}{s}(\nabla F{}_2(\bar u + s\tilde u(t)) - \nabla {F_2}(\bar u), s\tilde u(t))dsdt} } \\ & \leq \int_0^T e^{Q(t)} \int_0^1 \frac{1}{s} k(t) h_2(|s \tilde{u}(t)|) d s d t \\ & \leqslant \int_0^T {{e^{Q(t)}}\int_0^1 {\frac{1}{s}k(t){h_2}(s{{\left\| {\tilde u} \right\|}_\infty })dsdt} } \\ & \leqslant \int_0^T {{e^{Q(t)}}\int_0^1 {\frac{1}{s}k{\text{(}}t{\text{)}}{s^2}{K_2}\left\| {\tilde u} \right\|_\infty ^2dsdt} }\\ &\leqslant {K_2}\int_0^T {{e^{Q(t)}}k(t)dt} \left\| {\tilde u} \right\|_\infty ^2 \\ & \leqslant \frac{{{d_1}{K_2}T}}{{12}}\int_0^T {k{\text{(}}t{\text{)}}dt} \left\| {\dot u} \right\|_2^2 , \end{align*} (3.3)

    for sufficiently large \left\| {\tilde u} \right\|_\infty ^{} . By Sobolev's inequality, we have

    \begin{align*} \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}f{\text{(}}t{\text{)}}, u{\text{(}}t{\text{))}}dt} = &\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}f{\text{(}}t{\text{)}}, \tilde u{\text{(}}t{\text{))}}dt} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}f{\text{(}}t{\text{)}}, \bar u)dt} \\ & \leqslant \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}|f{\text{(}}t{\text{)|}}dt} {\left\| {\tilde u} \right\|_\infty } + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}|f{\text{(}}t{\text{)|}}dt|} \bar u| \\ &\leqslant {C_4}{\left\| {\dot u} \right\|_2} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}|f{\text{(}}t{\text{)|}}dt|} \bar u| . \end{align*} (3.4)

    Thus, by (3.1)–(3.4), we obtain

    \begin{align*} \varphi {\text{(}}u{\text{)}} = &\frac{1}{2}\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}A{\text{(}}t{\text{)}}\dot u{\text{(}}t{\text{)}}, \dot u{\text{(}}t{\text{))}}dt} + \int_{\text{0}}^T {{e^{Q{\text{(}}t{\text{)}}}}F{\text{(}}t, u{\text{(}}t{\text{))}}dt} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}f{\text{(}}t{\text{)}}, u{\text{(}}t{\text{))}}dt}\\ & = \frac{1}{2}\int_0^T {{e^{Q(t)}}(A(t)\dot u(t), \dot u(t))dt} + \int_0^T {{e^{Q(t)}}[{F_1}(t, u(t)) - {F_1}(t, \bar u)]dt} \\ & + \int_0^T {{e^{Q(t)}}[{F_2}(u(t)) - {F_2}(\bar u)]dt} + \int_0^T {{e^{Q(t)}}F(t, \bar u)dt} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}f{\text{(}}t{\text{)}}, u{\text{(}}t{\text{))}}dt} \\ & \geqslant (\frac{{\lambda {d_2}}}{4} - \frac{{{d_1}{K_2}T}}{{12}}\int_0^T {k(t)dt} )\left\| {\dot u} \right\|_2^2 -{C}_{1}{d}_{1}{K}_{1}(\frac{T}{12}{)}^{\frac{\alpha +1}{2}}{\displaystyle {\int }_{0}^{T}{r}_{1}\text{(}t\text{)}dt}{\Vert \dot{u}\Vert }_{2}^{\alpha +1} \\ & - ({C_1}{C_3}{d_1}{(\frac{T}{{12}})^{\frac{1}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} + {C_2} + {C_4}){\left\| {\dot u} \right\|_2} + |\bar u{|^{2\alpha }}(\frac{1}{{|\bar u{|^{2\alpha }}}}\int_0^T {{e^{Q(t)}}F(t, \bar u)dt} \\ & - \frac{{C_1^2d_1^2T}}{{3\lambda {d_2}}}{(\int_0^T {{r_1}(t)dt} )^2}\frac{{h_1^2(|\bar u|)}}{{|\bar u{|^{2\alpha }}}} - \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}|f{\text{(}}t{\text{)|}}dt\frac{1}{{|\bar u{|^{2\alpha - 1}}}}} ) . \end{align*} (3.5)

    Since

    \|u\| \rightarrow+\infty \Leftrightarrow\left(|\bar{u}|^2+\|\dot{u}\|_2^2\right)^{\frac{1}{2}} \rightarrow+\infty \text {, } (3.6)

    and \frac{{\lambda {d_2}}}{4} - \frac{{{d_1}{K_2}T}}{{12}}\int_0^T {k(t)dt} \gt 0 as \int_0^T {k(t)dt} \lt \frac{{3\lambda {d_2}}}{{{d_1}{K_2}T}} , it follows from {\text{(}}{H_1}{\text{), }} {\text{(}}{H_3}{\text{), }} (3.5) and (3.6) that \varphi {\text{(}}u{\text{)}} \to + \infty as {F_2}{\text{(}}x{\text{)}} = {\text{0}} i.e., \varphi {\text{(}}u{\text{)}} is coercive. By Theorem 1.1 and Corollary 1.1 in [3] (i.e., the least action principle), we complete the proof of Theorem 3.1.

    Remark 1. The condition {\text{(}}{H_4}{\text{)}} in Theorem 3.1 is weaker than the condition h{\text{(}}t{\text{)}} \leqslant {K_1}{t^\alpha } + {K_2} in Theorem 1.1 in [4] (i.e., Theorem A in the present paper), so that Theorem 3.1 generalizes Theorem 1.1 in [4] even in the case of {F_2}{\text{(}}x{\text{)}} = {\text{0}} , q{\text{(}}t{\text{)}} \equiv 0 , A{\text{(}}t{\text{)}} = {I_{N \times N}} and f{\text{(}}t{\text{)}} \equiv 0 .

    Theorem 3.2. Let F(t, x) = {F_1}(t, x) + {F_2}(x) , and suppose that {F_1}{\text{(}}t, x{\text{)}} and {F_2}{\text{(}}x{\text{)}} satisfy assumption (A) and {\text{(}}{H_2}{\text{)}} . If the following conditions hold:

    {\text{(}}{H_4}{\text{)}} There exist constants {\text{(}}{H_5}{\text{)}} {K_1} \geqslant 0, \alpha \in {\text{[}}0, 1{\text{)}} and a non-negative function {h_1} \in C{\text{([0, }} + \infty {\text{); [0, }} + \infty {\text{))}} with the properties:

    ({\rm{i}})\ {h_1}{\text{(}}s{\text{)}} \leqslant {h_1}{\text{(}}t{\text{) }}\ \ \ \ \forall s \leqslant t, s, t \in {\text{[}}0, + \infty {\text{)}},

    ({\rm{ii}})\ {h_1}{\text{(}}s + t{\text{)}} \leqslant {C_1}{\text{(}}{h_1}{\text{(}}s{\text{) }} + {\text{ }}{h_1}{\text{(}}t{\text{)) }}\ \ \ \ \forall s, t \in {\text{[}}0, + \infty {\text{)}},

    ({\rm{iii}}) \ \mathop {{\text{lim}}}\limits_{s \to + \infty } \frac{{{h_1}{\text{(}}s{\text{)}}}}{{{s^\alpha }}} \leqslant {K_1}.

    Moreover, there exist \varphi {\text{(}}{u_n}{\text{)}} such that

    \left| {\nabla {F_1}{\text{(}}t, x{\text{)}}} \right| \leqslant {r_1}{\text{(}}t{\text{)}}{h_1}{\text{(}}\left| x \right|{\text{)}} + {r_2}{\text{(}}t{\text{)}} , \ for\ x \in {R^N} , \ a.e.{\text{ }}t \in {\text{[}}0, T{\text{]}} .

    {\text{(}}{H_5}{\text{)}}\ \mathop {{\text{lim}}}\limits_{|x| \to + \infty } \frac{1}{{h_1^2{\text{(}}|x|{\text{)}}}}\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}F{\text{(}}t, x{\text{)}} + |f{\text{(}}t{\text{)}}||x|{\text{)}}dt} = - \infty.

    Then problem (1.1) has at least one solution on H_T^1 .

    Proof. We will use the Saddle Point Theorem 4.7 in [3] to prove Theorem 3.2. First, we prove that the functional \varphi {\text{(}}u{\text{)}} satisfies the (PS) condition. Suppose that \{ {u_n}\} is a (PS) sequence for \varphi {\text{(}}u{\text{)}}, that is, \mathop {\lim }\limits_{n \to \infty } \varphi '({u_n}) = 0 and \varphi {\text{(}}{u_n}{\text{)}} is bounded. By a way similar to (3.2)–(3.4), we have

    \left| {\int_0^T {{e^{Q(t)}}\left( {\nabla {F_1}(t, {u_n}(t)), {{\tilde u}_n}(t)} \right)dt} } \right|
    \leqslant \frac{{\lambda {d_2}}}{4}\left\| {{{\dot u}_n}} \right\|_2^2 + {C_1}{d_1}{K_1}{{\text{(}}\frac{T}{{12}}{\text{)}}^{\frac{{\alpha + 1}}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} \left\| {{{\dot u}_n}} \right\|_2^{\alpha + 1} + {\text{(}}{C_1}{C_3}{d_1}{{\text{(}}\frac{T}{{12}}{\text{)}}^{\frac{1}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} + {C_2}{\text{)}}{\left\| {{{\dot u}_n}} \right\|_2} + \\ \frac{{C_1^2d_1^2T}}{{3\lambda {d_2}}}{{\text{(}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} {\text{)}}^2}h_1^2{\text{(}}|{\bar u_n}|{\text{), }}
    \int_0^T {{e^{Q(t)}}(\nabla F{}_2({u_n}(t)), {{\tilde u}_n}(t))dt = } \int_0^T {{e^{Q(t)}}(\nabla F{}_2({{\bar u}_n} + {{\tilde u}_n}(t)), {{\tilde u}_n}(t))dt}
    =\int_0^T e^{Q(t)}\left(\nabla F_2\left(\bar{u}_n+\tilde{u}_n(t)\right)-\nabla F_2\left(\bar{u}_n\right), \tilde{u}_n(t)\right) d t
    \leq \int_0^T e^{Q(t)} k(t) h_2\left(\left|\tilde{u}_n(t)\right|\right) d t
    \leq \int_0^T e^{Q(t)} k(t) h_2\left(\left\|\tilde{u}_n\right\|_{\infty}\right) d t
    \leqslant \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}k{\text{(}}t{\text{)(}}{K_2}\left\| {{{\tilde u}_n}} \right\|_\infty ^2 + {C_5}{\text{)}}dt}
    \leqslant {K_2}\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}k{\text{(}}t{\text{)}}dt} \left\| {{{\tilde u}_n}} \right\|_\infty ^2 + {C_6}
    \leqslant \frac{{{d_1}{K_2}T}}{{12}}\int_0^T {k{\text{(}}t{\text{)}}dt} \left\| {{{\dot u}_n}} \right\|_2^2 + {C_6}

    and

    \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}f{\text{(}}t{\text{)}}, {{\tilde u}_n}{\text{(}}t{\text{))}}dt} \leqslant \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}|f{\text{(}}t{\text{)|}}dt} {\left\| {{{\tilde u}_n}} \right\|_\infty } \leqslant {C_4}{\left\| {{{\dot u}_n}} \right\|_2} .

    Hence,

    \left\| {{{\tilde u}_n}} \right\| \geqslant {\text{(}}\varphi '{\text{(}}{u_n}{\text{)}}, {\tilde u_n}{\text{)}}
    = \int_0^T {{e^{Q(t)}}(A(t){{\dot u}_n}(t), {{\dot u}_n}(t))dt}
    + \int_{\text{0}}^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}\nabla {F_1}{\text{(}}t, {u_n}{\text{(}}t{\text{))}}, {{\tilde u}_n}{\text{(}}t{\text{))}}dt} + \int_{\text{0}}^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}\nabla {F_2}{\text{(}}{u_n}{\text{(}}t{\text{))}}, {{\tilde u}_n}{\text{(}}t{\text{))}}dt} + \int_{\text{0}}^T {{e^{Q(t)}}{\text{(}}f{\text{(}}t{\text{)}}, {{\tilde u}_n}{\text{(}}t{\text{))}}dt}
    \geqslant \lambda {d_2}\left\| {{{\dot u}_n}} \right\|_2^2 - \frac{{\lambda {d_2}}}{4}\left\| {{{\dot u}_n}} \right\|_2^2 - \frac{{C_1^2d_1^2T}}{{3\lambda {d_2}}}{{\text{(}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} {\text{)}}^2}h_1^2{\text{(}}|{\bar u_n}|{\text{)}} - {C_1}{d_1}{K_1}{(\frac{T}{{12}})^{\frac{{\alpha + 1}}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} \left\| {{{\dot u}_n}} \right\|_2^{\alpha + 1}
    - {\text{(}}{C_1}{C_3}{d_1}{(\frac{T}{{12}})^{\frac{1}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} + {C_2}{\text{)}}{\left\| {{{\dot u}_n}} \right\|_2} - \frac{{{d_1}{K_2}T}}{{12}}\int_0^T {k{\text{(}}t{\text{)}}dt} \left\| {{{\dot u}_n}} \right\|_2^2 - {C_4}{\left\| {{{\dot u}_n}} \right\|_2} - {C_6}
    = {\text{(}}\frac{{3\lambda {d_2}}}{4} - \frac{{{d_1}{K_2}T}}{{12}}\int_0^T {k{\text{(}}t{\text{)}}dt} {\text{)}}\left\| {{{\dot u}_n}} \right\|_2^2 - \frac{{C_1^2d_1^2T}}{{3\lambda {d_2}}}{{\text{(}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} {\text{)}}^2}h_1^2{\text{(}}|{\bar u_n}|{\text{)}}
    -{C}_{1}{d}_{1}{K}_{1}(\frac{T}{12}{)}^{\frac{\alpha +1}{2}}{\displaystyle {\int }_{0}^{T}{r}_{1}\text{(}t\text{)}dt}{\Vert {\dot{u}}_{n}\Vert }_{2}^{\alpha +1} - {\text{(}}{C_1}{C_3}{d_1}{(\frac{T}{{12}})^{\frac{1}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} + {C_2} + {C_4}{\text{)}}{\left\| {{{\dot u}_n}} \right\|_2} - {C_6}.

    It follows from (2.1) that

    \left\| {{{\tilde u}_n}} \right\| \leqslant {(1 + \frac{{{T^{\text{2}}}}}{{{\text{4}}{\pi ^{\text{2}}}}})^{\frac{1}{2}}}{(\int_{\text{0}}^T {{{\left| {{{\dot u}_n}(t)} \right|}^{\text{2}}}dt} )^{\frac{1}{2}}} .

    Thus,

    \left\| {{{\dot u}_n}} \right\|_2^{} \leqslant {C_7}{h_1}{\text{(}}|{\bar u_n}|{\text{)}} + {C_8} , \ {\rm{for}}\ {\rm{sufficiently}}\ {\rm{large}} \ n . (3.7)

    By (3.2)–(3.4) and {\text{(}}{H_5}{\text{)}} , we have

    \varphi {\text{(}}{u_n}{\text{)}} = \frac{1}{2}\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}A{\text{(}}t{\text{)}}{{\dot u}_n}{\text{(}}t{\text{)}}, {{\dot u}_n}{\text{(}}t{\text{))}}dt} + \int_{\text{0}}^T {{e^{Q{\text{(}}t{\text{)}}}}F{\text{(}}t, {u_n}{\text{(}}t{\text{))}}dt} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}f{\text{(}}t{\text{)}}, {u_n}{\text{(}}t{\text{))}}dt}
    = \frac{1}{2}\int_0^T {{e^{Q(t)}}(A(t){{\dot u}_n}(t), {{\dot u}_n}(t))dt} + \int_0^T {{e^{Q(t)}}[{F_1}(t, {u_n}(t)) - {F_1}(t, {{\bar u}_n})]dt}
    + \int_0^T {{e^{Q(t)}}[{F_2}({u_n}(t)) - {F_2}({{\bar u}_n})]dt} + \int_0^T {{e^{Q(t)}}F(t, {{\bar u}_n})dt} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}f{\text{(}}t{\text{)}}, {u_n}{\text{(}}t{\text{))}}dt}
    \leqslant (\frac{{{d_1}{a^N}}}{2} + \frac{{\lambda {d_2}}}{4} + \frac{{{d_1}{K_2}T}}{{12}}\int_0^T {k(t)dt} )\left\| {{{\dot u}_n}} \right\|_2^2 + {C_1}{d_1}{K_1}{(\frac{T}{{12}})^{\frac{{\alpha + 1}}{2}}}\int_0^T {{r_1}(t)dt} \left\| {{{\dot u}_n}} \right\|_2^{\alpha + 1}
    + {\text{(}}{C_1}{C_3}{d_1}{(\frac{T}{{12}})^{\frac{1}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} + {C_2} + {C_4}){\left\| {{{\dot u}_n}} \right\|_2} + \frac{{C_1^2d_1^2T}}{{3\lambda {d_2}}}{(\int_0^T {{r_1}(t)dt} )^2}h_1^2(|{\bar u_n}|)
    + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}|f{\text{(}}t{\text{)|}}dt\left| {{{\bar u}_n}} \right|} + \int_0^T {{e^{Q(t)}}F(t, {{\bar u}_n})dt}
    \leqslant {C_9} h_1^2{\text{(}}|{\bar u_n}|{\text{)}} + {C_{10}}h_1^{\alpha + 1}{\text{(}}|{\bar u_n}|{\text{)}} + {C_{11}}h_1^{}{\text{(}}|{\bar u_n}|{\text{)}} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}|f{\text{(}}t{\text{)|}}dt\left| {{{\bar u}_n}} \right|} + \int_0^T {{e^{Q(t)}}F(t, {{\bar u}_n})dt}
    = h_1^2{\text{(}}|{\bar u_n}|{\text{)(}}\frac{1}{{h_1^2{\text{(}}|{{\bar u}_n}|{\text{)}}}}\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}F{\text{(}}t, {{\bar u}_n}{\text{)}} + |f{\text{(}}t{\text{)}}||{{\bar u}_n}|{\text{)}}dt} + {C_{10}}\frac{1}{{h_1^{1 - \alpha }{\text{(}}|{{\bar u}_n}|{\text{)}}}} + {C_{11}}\frac{1}{{h_1^{}{\text{(}}|{{\bar u}_n}|{\text{)}}}} + {C_9}{\text{)}}
    \to - \infty \ {\rm{as}} \ \left| {{{\bar u}_n}} \right| \to + \infty ,

    which contradicts the boundedness of \varphi {\text{(}}{u_n}{\text{)}}. Therefore, {\text{\{ }}|{\bar u_n}|{\text{\} }} is bounded, and then {\text{\{ }}{u_n}{\text{\} }} is bounded by (3.7). We conclude that the (PS) condition is satisfied.

    Next, we only need to prove the following conditions:

    {\text{(}}{l_1}{\text{)}} \varphi {\text{(}}u{\text{)}} \to - \infty as \left| u \right| \to + \infty in {R^N} ;

    {\text{(}}{l_2}{\text{)}} \varphi {\text{(}}u{\text{)}} \to + \infty as \left\| u \right\| \to + \infty in \tilde H_T^1 .

    In fact, by {\text{(}}{H_5}{\text{)}} , we obtain

    \varphi {\text{(}}u{\text{)}} = \int_{\text{0}}^T {{e^{Q{\text{(}}t{\text{)}}}}F{\text{(}}t, u{\text{)}}dt} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}f{\text{(}}t{\text{)}}, u{\text{)}}dt}
    \leqslant \int_{\text{0}}^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}F{\text{(}}t, u{\text{)}} + |f{\text{(}}t{\text{)|}}|u|{\text{)}}dt} \to - \infty \ {\rm{as}} \ \left| u \right| \to + \infty \ {\rm{in}}\ {R^N} .

    Thus {\text{(}}{l_1}{\text{)}} is proved.

    For u \in \tilde H_T^1 , as we have argued in (3.2) and (3.3), we have

    \left| {\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{[}}{F_1}{\text{(}}t, \tilde u{\text{(}}t{\text{))}} - {F_1}{\text{(}}t, 0{\text{)]}}dt} } \right| = \left| {\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}\int_0^1 {{\text{(}}\nabla {F_1}{\text{(}}t, s\tilde u(t{\text{))}}, \tilde u(t{\text{))}}dsdt} } } \right|
    \leqslant \int_0^T {\int_0^1 {{e^{Q{\text{(}}t{\text{)}}}}{r_1}{\text{(}}t{\text{)}}{h_1}{\text{(}}|s\tilde u{\text{(}}t{\text{)}}|{\text{)}} \cdot |\tilde u{\text{(}}t{\text{)}}|dsdt} + \int_0^T {\int_0^1 {{e^{Q{\text{(}}t{\text{)}}}}{r_2}{\text{(}}t{\text{)}}|\tilde u{\text{(}}t{\text{)}}|dsdt} } }
    \leqslant {d_1}{h_1}{\text{(}}{\left\| {\tilde u} \right\|_\infty }{\text{)}}{\left\| {\tilde u} \right\|_\infty }\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} + {C_2}(\int_0^T {|\dot u{\text{(}}t{\text{)}}{|^2}dt{)^{\frac{1}{2}}}}
    \leqslant {d_1}({K_1}\left\| {\tilde u} \right\|_\infty ^\alpha + {C_3}{\text{)}}{\left\| {\tilde u} \right\|_\infty }\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} + {C_2}(\int_0^T {|\dot u{\text{(}}t{\text{)}}{|^2}dt{)^{\frac{1}{2}}}}\\ \leqslant {d_1}{K_1}{{\text{(}}\frac{T}{{12}}{\text{)}}^{\frac{{\alpha + 1}}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} \left\| {\dot u} \right\|_2^{\alpha + 1} + {\text{(}}{C_3}{d_1}{(\frac{T}{{12}})^{\frac{1}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} + {C_2}{\text{)}}{\left\| {\dot u} \right\|_2}

    and

    \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{[}}{F_2}{\text{(}}u{\text{(}}t{\text{))}} - {F_2}{\text{(}}0{\text{)]}}dt} \leqslant \frac{{{d_1}{K_2}T}}{{12}}\int_0^T {k{\text{(}}t{\text{)}}dt} \left\| {\dot u} \right\|_2^2 ,

    which implies that

    \begin{align*} \varphi {\text{(}}u{\text{)}} = &\frac{1}{2}\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}A{\text{(}}t{\text{)}}\dot u{\text{(}}t{\text{)}}, \dot u{\text{(}}t{\text{))}}dt} + \int_{\text{0}}^T {{e^{Q{\text{(}}t{\text{)}}}}F{\text{(}}t, u{\text{(}}t{\text{))}}dt} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}f{\text{(}}t{\text{)}}, u{\text{(}}t{\text{))}}dt}\\ & = \frac{1}{2}\int_0^T {{e^{Q(t)}}(A(t)\dot u(t), \dot u(t))dt} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{[}}{F_1}{\text{(}}t, u{\text{(}}t{\text{))}} - {F_1}{\text{(}}t, 0{\text{)]}}dt} \\ & + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{[}}{F_2}{\text{(}}u{\text{(}}t{\text{))}} - {F_2}{\text{(}}0{\text{)]}}dt} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}F{\text{(}}t, 0{\text{)}}dt} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}f{\text{(}}t{\text{)}}, u{\text{(}}t{\text{))}}dt} \\ & \geqslant (\frac{{\lambda {d_2}}}{4} - \frac{{{d_1}{K_2}T}}{{12}}\int_0^T {k(t)dt} )\left\| {\dot u} \right\|_2^2 - {d_1}{K_1}{(\frac{T}{{12}})^{\frac{{\alpha + 1}}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} \left\| {\dot u} \right\|_2^{\alpha + 1}\\ & - ({C_3}{d_1}{(\frac{T}{{12}})^{\frac{1}{2}}}\int_0^T {{r_1}{\text{(}}t{\text{)}}dt} + {C_2} + {C_4}){\left\| {\dot u} \right\|_2} + \int_0^T {{e^{Q{\text{(}}t{\text{)}}}}F{\text{(}}t, 0{\text{)}}dt} , \end{align*} (3.8)

    for all u \in \tilde H_T^1 .

    By Wirtinger's inequality, one has

    \left\| u \right\| \to + \infty \Leftrightarrow \;\;\;\left\| {\dot u} \right\|_2^{} \to + \infty , u \in \tilde H_T^1 .

    We know that \frac{{\lambda {d_2}}}{4} - \frac{{{d_1}{K_2}T}}{{12}}\int_0^T {k{\text{(}}t{\text{)}}dt} \gt 0 as \int_0^T {k{\text{(}}t{\text{)}}dt} \lt \frac{{3\lambda {d_2}}}{{{d_1}{K_2}T}} . It follows from (3.8) that \varphi (u) \to + \infty as \lambda \gt 0. in {F_1}{\text{(}}t, x{\text{)}} = {\text{(}}\frac{{{d_2}}}{{3{d_1}}}T - t{\text{)l}}{{\text{n}}^{\frac{{\text{3}}}{{\text{2}}}}}{\text{(}}1 + |x{|^2}{\text{)}} , that is, {F_2}{\text{(}}x{\text{)}} = \frac{{\lambda {d_2}}}{{{d_1}{K_2}{T^2}}}|x{|^2} is proved.

    By making use of the Saddle Point Theorem 4.7 in [3], we prove that problem (1.1) has at least one solution on {h_1}{\text{(|}}x|{\text{)}} = {\text{l}}{{\text{n}}^{\frac{{\text{1}}}{{\text{2}}}}}{\text{(}}1 + |x{|^2}{\text{)}} .

    Remark 2. The condition = {r_1}{\text{(}}t{\text{)}}{h_1}{\text{(|}}x{\text{|)}} + {r_2}{\text{(}}t{\text{)}} in Theorem 3.2 is weaker than the condition {\text{(}}\nabla {F_2}{\text{(}}x{\text{)}} - \nabla {F_2}{\text{(}}y{\text{), }}x - y{\text{)}} = \frac{{2\lambda {d_2}}}{{{d_1}{K_2}{T^2}}}|x - y{|^2} = k{\text{(}}t{\text{)}}{h_2}{\text{(}}|x - y|{\text{)}} in Theorem 1.2 in [4] (i.e., Theorem B in the present paper), so that Theorem 3.2 generalizes Theorem 1.2 in [4] even in the case of F_2(x)=0, q(t) \equiv 0, A(t)=I_{N \times N} \text { and } f(t) \equiv 0 \text {. }

    In this section, we give two examples to illustrate the feasibility and effectiveness of our main conclusions.

    Example 4.1. Let A{\text{(}}t{\text{)}} = {\left({\begin{array}{*{20}{c}} {\lambda + t} & 0 & 0 & \cdots & 0 \\ 0 & {\lambda + t} & 0 & \cdots & 0 \\ 0 & 0 & \ddots & \cdots & 0 \\ \vdots & \vdots & \vdots & \ddots & \vdots \\ 0 & 0 & 0 & \cdots & {\lambda + t} \end{array}} \right)_{N \times N}} , where \lambda \gt 0.

    Let {F_1}{\text{(}}t, x{\text{)}} = {\text{(}}\frac{{{d_2}}}{{3{d_1}}}T - t{\text{)l}}{{\text{n}}^{\frac{{\text{3}}}{{\text{2}}}}}{\text{(}}1 + |x{|^2}{\text{)}} , {F_2}{\text{(}}x{\text{)}} = \frac{{\lambda {d_2}}}{{{d_1}{K_2}{T^2}}}|x{|^2} , {h_1}{\text{(|}}x|{\text{)}} = {\text{l}}{{\text{n}}^{\frac{{\text{1}}}{{\text{2}}}}}{\text{(}}1 + |x{|^2}{\text{)}} .

    Then,

    {\text{l}}{{\text{n}}^{\frac{{\text{1}}}{{\text{2}}}}}{\text{[}}1 + {{\text{(}}s + t{\text{)}}^2}{\text{]}} \leqslant {\text{2(l}}{{\text{n}}^{\frac{{\text{1}}}{{\text{2}}}}}{\text{(}}1 + {s^2}{\text{)}} + {\text{l}}{{\text{n}}^{\frac{{\text{1}}}{{\text{2}}}}}{\text{(}}1 + {t^2}{\text{))}} , {\text{ }}\ \ \ \ \forall s, \;t \in {\text{[}}0, + \infty {\text{)}} ,

    i.e.,

    {h_1}{\text{(}}s + t{\text{)}} \leqslant {C_1}{\text{(}}{h_1}{\text{(}}s{\text{) }} + {\text{ }}{h_1}{\text{(}}t{\text{) ), }}\ \ \ \ \forall s, \;t \in {\text{[}}0, + \infty {\text{)}}

    and

    \mathop {{\text{lim}}}\limits_{s \to + \infty } \frac{{{h_1}{\text{(}}s{\text{)}}}}{{{s^\alpha }}} = \mathop {{\text{lim}}}\limits_{s \to + \infty } \frac{{{\text{l}}{{\text{n}}^{\frac{{\text{1}}}{{\text{2}}}}}{\text{(1}} + {s^2}{\text{)}}}}{{{s^\alpha }}} = 0 \leqslant {K_1} .

    Moreover, we obtained the following results through simple derivation:

    |\nabla {F_1}{\text{(}}t, x{\text{)|}} = \frac{3}{2}{\text{|}}\frac{{{d_2}}}{{3{d_1}}}T - t|{\text{l}}{{\text{n}}^{\frac{{\text{1}}}{{\text{2}}}}}{\text{(}}1 + |x{|^2}{\text{)}}\frac{{\text{1}}}{{{\text{1}} + |x{|^2}}} \cdot 2|x{\text{|}}

        \leqslant \frac{3}{2}{\text{|}}\frac{{{d_2}}}{{3{d_1}}}T - t|{\text{l}}{{\text{n}}^{\frac{{\text{1}}}{{\text{2}}}}}{\text{(}}1 + |x{|^2}{\text{)}}

        = {r_1}{\text{(}}t{\text{)}}{h_1}{\text{(|}}x{\text{|)}} + {r_2}{\text{(}}t{\text{)}},

    {\text{(}}\nabla {F_2}{\text{(}}x{\text{)}} - \nabla {F_2}{\text{(}}y{\text{), }}x - y{\text{)}} = \frac{{2\lambda {d_2}}}{{{d_1}{K_2}{T^2}}}|x - y{|^2} = k{\text{(}}t{\text{)}}{h_2}{\text{(}}|x - y|{\text{)}} , {\text{(}}\int_0^T {k{\text{(}}t{\text{)}}dt} \lt \frac{{3\lambda {d_2}}}{{{d_1}{K_2}T}}{\text{)}} ,

    \mathop {{\text{limsup}}}\limits_{s \to + \infty } \frac{{{h_2}{\text{(}}s{\text{)}}}}{{{s^2}}} = 1 \leqslant {K_2} ,

    \mathop {{\text{lim}}}\limits_{|x| \to + \infty } \frac{1}{{|x{|^{2\alpha }}}}\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}F{\text{(}}t, x{\text{)}}dt} \varphi {\text{(}}u{\text{)}}

       = \lim _{|x| \rightarrow \infty} \frac{1}{|x|^{2 \alpha}} \int_0^{\mathrm{T}} e^{Q(t)}\left(\frac{d_2}{3 d_1} T-t\right) \ln ^{\frac{3}{2}}\left(1+|x|^2\right) d t+\lim _{|x| \rightarrow \infty} \frac{1}{|x|^{2 \alpha}} \int_0^{\mathrm{T}} e^{Q(t)} \frac{\lambda d_2}{d_1 K_2 T^2}|x|^2 d t

       \geq \int_0^T e^{Q(t)}\left(\frac{d_2}{3 d_1} T-t\right) d t \cdot \lim _{|x| \rightarrow \infty} \frac{\ln ^{\frac{3}{2}}\left(1+|x|^2\right)}{|x|^{2 \alpha}}+\frac{\lambda d_2^2}{d_1 K_2 T} \lim _{|x| \rightarrow \infty}|x|^{2-2 \alpha}

       = +\infty \text {, }

    \lambda |x{|^{\text{2}}} \leqslant {\text{(}}\lambda + t{\text{)}}|x{|^2} = {\text{(}}A{\text{(}}t{\text{)}}x, x{\text{)}} , for all x \in {R^N} and a.e.{\text{ }}t \in {\text{[}}0, T{\text{]}} .

    It is easy to see that A{\text{(}}t{\text{)}} , {F_1}{\text{(}}t, x{\text{)}} , {F_2}{\text{(}}x{\text{)}} and {h_1}{\text{(|}}x|{\text{)}} satisfy all conditions of Theorem 3.1. Hence, problem (1.1) has at least one solution on H_T^1 .

    Example 4.2. Let A{\text{(}}t{\text{)}} = {\left({\begin{array}{*{20}{c}} {\lambda + t} & 0 & 0 & \cdots & 0 \\ 0 & {\lambda + t} & 0 & \cdots & 0 \\ 0 & 0 & \ddots & \cdots & 0 \\ \vdots & \vdots & \vdots & \ddots & \vdots \\ 0 & 0 & 0 & \cdots & {\lambda + t} \end{array}} \right)_{N \times N}} , where \lambda \gt 0.

    Let {F_1}{\text{(}}t, x{\text{)}} = {\text{(}}\frac{{{d_2}}}{{3{d_1}}}T - t{\text{)l}}{{\text{n}}^{\frac{{\text{3}}}{{\text{2}}}}}{\text{(}}1 + |x{|^2}{\text{)}} , {F_2}{\text{(}}x{\text{)}} = - \frac{{\lambda {d_2}}}{{{d_1}{K_2}{T^2}}}|x{|^2} , {h_1}{\text{(|}}x|{\text{)}} = {\text{l}}{{\text{n}}^{\frac{{\text{1}}}{{\text{2}}}}}{\text{(}}1 + |x{|^2}{\text{)}} .

    Then,

    \mathop {{\text{lim}}}\limits_{|x| \to + \infty } \frac{1}{{h_1^2{\text{(}}|x|{\text{)}}}}\int_0^T {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}F{\text{(}}t, x{\text{)}} + |f{\text{(}}t{\text{)}}||x|{\text{)}}dt}
    = \mathop {{\text{lim}}}\limits_{|x| \to \infty } {\text{l}}{{\text{n}}^{\frac{{\text{1}}}{{\text{2}}}}}{\text{(}}1 + |x{|^2}{\text{)}}\int_{\text{0}}^{\text{T}} {{e^{Q{\text{(}}t{\text{)}}}}{\text{(}}\frac{{{d_2}}}{{3{d_1}}}T - t{\text{)}}} dt - \mathop {{\text{lim}}}\limits_{|x| \to \infty } \frac{{|x{|^2}}}{{{\text{ln(}}1 + |x{|^2}{\text{)}}}}\int_{\text{0}}^{\text{T}} {{e^{Q{\text{(}}t{\text{)}}}}\frac{{\lambda {d_2}}}{{{d_1}{K_2}{T^2}}}} dt + \mathop {{\text{lim}}}\limits_{|x| \to \infty } \frac{{|x|}}{{{\text{ln(}}1 + |x{|^2}{\text{)}}}}\int_{\text{0}}^{\text{T}} {{e^{Q{\text{(}}t{\text{)}}}}|f{\text{(}}t{\text{)}}} |dt \\ \leqslant - \frac{{{d_2}{T^2}}}{6}\mathop {{\text{lim}}}\limits_{|x| \to \infty } {\text{l}}{{\text{n}}^{\frac{{\text{1}}}{{\text{2}}}}}{\text{(}}1 + |x{|^2}{\text{)}} - \frac{{\lambda d_2^2}}{{{d_1}{K_2}T}}\mathop {{\text{lim}}}\limits_{|x| \to \infty } \frac{{|x{|^2}}}{{{\text{ln(}}1 + |x{|^2}{\text{)}}}} + \int_{\text{0}}^{\text{T}} {{e^{Q{\text{(}}t{\text{)}}}}|f{\text{(}}t{\text{)}}} |dt \cdot \mathop {{\text{lim}}}\limits_{|x| \to \infty } \frac{{|x|}}{{{\text{ln(}}1 + |x{|^2}{\text{)}}}}
    = - \infty .

    The derivation of other conditions for Theorem 3.2 is the same as Example 4.1.

    It is clear that A{\text{(}}t{\text{)}} , {F_1}{\text{(}}t, x{\text{)}} , {F_2}{\text{(}}x{\text{)}} and {h_1}{\text{(|}}x|{\text{)}} satisfy all conditions of Theorem 3.2. Therefore, problem (1.1) has at least one solution on H_T^1 .

    In this paper, we study the existence of periodic solutions of the forced damped vibration systems [i.e., (1.1)] by using the variational method and the critical point theory.

    First, we provide an expression for functional \varphi {\text{(}}u{\text{)}} and further prove that the functional \varphi {\text{(}}u{\text{)}} is continuously differentiable and weak lower semi-continuous.

    Then, we prove that the critical point of \varphi {\text{(}}u{\text{)}} is a solution of problem (1.1) in the sense of weak-derivative. Moreover, we directly derive the variational principle of problem (1.1) via the semi-inverse method.

    Finally, it is proved that problem (1.1) has at least one solution under the given sufficient conditions through the least action principle and the saddle point theorem.

    In the future, we can continue to study this problem by looking for new sufficient conditions.

    The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

    This work was supported in part by the National Natural Science Foundation of China (No.11861005), the Joint Special Fund Project for Basic Research of Local Undergraduate Universities in Yunnan Province (No.202101BA070001-219) and Foundation of Dali University (No.KY2319101540).

    There is no conflict of interest.



    [1] X. P. Wu, C. L. Tang, Periodic solutions of a class of non-autonomous second order systems, J. Math. Anal. Appl., 236 (1999), 227–235. https://doi.org/10.1006/jmaa.1999.6408 doi: 10.1006/jmaa.1999.6408
    [2] P. H. Rabinowitz, On subharmonic solutions of Hamiltonian systems, Commun. Pure Appl. Math., 33 (1980), 609–633. https://doi.org/10.1002/cpa.3160330504 doi: 10.1002/cpa.3160330504
    [3] J. Mawhin, M. Willem, Critical point theory and Hamiltonian systems, New York: Springer, 1989. https://doi.org/10.1007/978-1-4757-2061-7
    [4] Z. Wang, J. Zhang, Periodic solutions of a class of second order non-autonomous Hamiltonian systems, Nonlinear Anal-Theor., 72 (2010), 4480–4487. https://doi.org/10.1016/j.na.2010.02.023 doi: 10.1016/j.na.2010.02.023
    [5] Z. Wang, J. Zhang, New existence results on periodic solutions of non-autonomous second order Hamiltonian systems, Appl. Math. Lett., 79 (2018), 43–50. https://doi.org/10.1016/j.aml.2017.11.016 doi: 10.1016/j.aml.2017.11.016
    [6] C. L. Tang, Periodic solutions for nonautonomous second order systems with sublinear nonlinearity, Proc. Amer. Math. Soc., 126 (1998), 3263–3270. https://doi.org/10.1090/S0002-9939-98-04706-6 doi: 10.1090/S0002-9939-98-04706-6
    [7] G. Bonanno, R. Livrea, M. Schechter, Some notes on a superlinear second order Hamiltonian system, manuscripta math., 154 (2017), 59–77. https://doi.org/10.1007/s00229-016-0903-6 doi: 10.1007/s00229-016-0903-6
    [8] J. Pipan, M. Schechter, Non-autonomous second order Hamiltonian systems, J. Differ. Equ., 257 (2014), 351–373. https://doi.org/10.1016/j.jde.2014.03.016 doi: 10.1016/j.jde.2014.03.016
    [9] N. Aizmahin, T. An, The existence of periodic solutions of non-autonomous second-order Hamiltonian systems, Nonlinear Anal-Theor., 74 (2011), 4862–4867. https://doi.org/10.1016/j.na.2011.04.060 doi: 10.1016/j.na.2011.04.060
    [10] J. Ma, C. L. Tang, Periodic solutions for some nonautonomous second order systems, J. Math. Anal. Appl., 275 (2002), 482–494. https://doi.org/10.1016/S0022-247X(02)00636-4 doi: 10.1016/S0022-247X(02)00636-4
    [11] C. Tang, Periodic solutions of non-autonomous second order systems, J. Math. Anal. Appl., 202 (1996), 465–469. https://doi.org/10.1006/jmaa.1996.0327 doi: 10.1006/jmaa.1996.0327
    [12] Z. Wang, J. Zhang, M. Chen, A unified approach to periodic solutions for a class of non-autonomous second order Hamiltonian systems, Nonlinear Anal.-Real, 58 (2021), 103218. https://doi.org/10.1016/j.nonrwa.2020.103218 doi: 10.1016/j.nonrwa.2020.103218
    [13] X. Wu, S. Chen, K. Teng, On variational methods for a class of damped vibration problems, Nonlinear Anal-Theor., 68 (2008), 1432–1441. https://doi.org/10.1016/j.na.2006.12.043 doi: 10.1016/j.na.2006.12.043
    [14] Z. Wang, J. Zhang, Existence of periodic solutions for a class of damped vibration problems, C. R. Math., 356 (2018), 597–612. https://doi.org/10.1016/j.crma.2018.04.014 doi: 10.1016/j.crma.2018.04.014
    [15] Q. B. Yin, Y. Guo, D. Wu, X. B. Shu, Existence and multiplicity of mild solutions for first-order Hamilton random impulsive differential equations with Dirichlet boundary conditions, Qual. Theory Dyn. Syst., 22 (2023), 47. https://doi.org/10.1007/s12346-023-00748-5 doi: 10.1007/s12346-023-00748-5
    [16] S. Wang, X. B. Shu, L. Shu, Existence of solutions to a class of damped random impulsive differential equations under Dirichlet boundary value conditions, AIMS Mathematics, 7 (2022), 7685–7705. https://doi.org/10.3934/math.2022431 doi: 10.3934/math.2022431
    [17] Y. Guo, X. B. Shu, Q. Yin, Existence of solutions for first-order Hamiltonian random impulsive differential equations with Dirichlet boundary conditions, Discrete Cont. Dyn.-B, 27 (2022), 4455–4471. https://doi.org/10.3934/dcdsb.2021236 doi: 10.3934/dcdsb.2021236
    [18] H. J. Ma, Simplified Hamiltonian-based frequency-amplitude formulation for nonlinear vibration systems, FACTA Univ.-Ser: Mech., 20 (2022), 445–455. https://doi.org/10.22190/FUME220420023M doi: 10.22190/FUME220420023M
    [19] F. Faraci, R. Livrea, Infinitely many periodic solutions for a second-order non-autonomous system, Nonlinear Anal-Theor., 54 (2003), 417–429. https://doi.org/10.1016/S0362-546X(03)00099-3 doi: 10.1016/S0362-546X(03)00099-3
    [20] J. H. He, Semi-inverse method of establishing generalized variational principles for fluid mechanics with emphasis on turbomachinery aerodynamics, Int. J. Turbo Jet Eng., 14 (1997), 23–28. https://doi.org/10.1515/TJJ.1997.14.1.23 doi: 10.1515/TJJ.1997.14.1.23
    [21] J. H. He, A classical variational model for micropolar elastodynamics, Int. J. Nonlinear Sci. Numer. Simulat., 1 (2000), 133–138. https://doi.org/10.1515/IJNSNS.2000.1.2.133 doi: 10.1515/IJNSNS.2000.1.2.133
    [22] J. H. He, Hamilton principle and generalized variational principles of linear thermopiezoelectricity, J. Appl. Mech., 68 (2001), 666–667. https://doi.org/10.1115/1.1352067 doi: 10.1115/1.1352067
  • Reader Comments
  • © 2023 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(1437) PDF downloads(95) Cited by(0)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog