Processing math: 61%
Research article

Weighted composite asymmetric Huber estimation for partial functional linear models

  • Received: 01 December 2021 Revised: 19 January 2022 Accepted: 24 January 2022 Published: 16 February 2022
  • MSC : 62G05, 62G20

  • In this paper, we first investigate a new asymmetric Huber regression (AHR) estimation procedure to analyze skewed data with partial functional linear models. To automatically reflect distributional features as well as bound the influence of outliers effectively, we further propose a weighted composite asymmetric Huber regression (WCAHR) estimation procedure by combining the strength across multiple asymmetric Huber loss functions. The slope function and constant coefficients are estimated through minimizing the combined loss function and approximating the slope function with principal component analysis. The asymptotic properties of the proposed estimators are derived. To realize the WCAHR estimation, we also develop a practical algorithm based on pseudo data. Numerical results show that the proposed WCAHR estimators can well adapt to the different error distributions, and thus are more useful in practice. Two real data examples are presented to illustrate the applications of the proposed methods.

    Citation: Juxia Xiao, Ping Yu, Zhongzhan Zhang. Weighted composite asymmetric Huber estimation for partial functional linear models[J]. AIMS Mathematics, 2022, 7(5): 7657-7684. doi: 10.3934/math.2022430

    Related Papers:

    [1] Xintao Li, Rongrui Lin, Lianbing She . Periodic measures for a neural field lattice model with state dependent superlinear noise. Electronic Research Archive, 2024, 32(6): 4011-4024. doi: 10.3934/era.2024180
    [2] Shuang Wang, FanFan Chen, Chunlian Liu . The existence of periodic solutions for nonconservative superlinear second order ODEs: a rotation number and spiral analysis approach. Electronic Research Archive, 2025, 33(1): 50-67. doi: 10.3934/era.2025003
    [3] Lianbing She, Nan Liu, Xin Li, Renhai Wang . Three types of weak pullback attractors for lattice pseudo-parabolic equations driven by locally Lipschitz noise. Electronic Research Archive, 2021, 29(5): 3097-3119. doi: 10.3934/era.2021028
    [4] Nan Xiang, Aying Wan, Hongyan Lin . Diffusion-driven instability of both the equilibrium solution and the periodic solutions for the diffusive Sporns-Seelig model. Electronic Research Archive, 2022, 30(3): 813-829. doi: 10.3934/era.2022043
    [5] Zhili Zhang, Aying Wan, Hongyan Lin . Spatiotemporal patterns and multiple bifurcations of a reaction- diffusion model for hair follicle spacing. Electronic Research Archive, 2023, 31(4): 1922-1947. doi: 10.3934/era.2023099
    [6] Yixuan Wang, Xianjiu Huang . Ground states of Nehari-Pohožaev type for a quasilinear Schrödinger system with superlinear reaction. Electronic Research Archive, 2023, 31(4): 2071-2094. doi: 10.3934/era.2023106
    [7] Xiangwen Yin . A review of dynamics analysis of neural networks and applications in creation psychology. Electronic Research Archive, 2023, 31(5): 2595-2625. doi: 10.3934/era.2023132
    [8] Weiyu Li, Hongyan Wang . Dynamics of a three-molecule autocatalytic Schnakenberg model with cross-diffusion: Turing patterns of spatially homogeneous Hopf bifurcating periodic solutions. Electronic Research Archive, 2023, 31(7): 4139-4154. doi: 10.3934/era.2023211
    [9] Peng Yu, Shuping Tan, Jin Guo, Yong Song . Data-driven optimal controller design for sub-satellite deployment of tethered satellite system. Electronic Research Archive, 2024, 32(1): 505-522. doi: 10.3934/era.2024025
    [10] Peng Gao, Pengyu Chen . Blowup and MLUH stability of time-space fractional reaction-diffusion equations. Electronic Research Archive, 2022, 30(9): 3351-3361. doi: 10.3934/era.2022170
  • In this paper, we first investigate a new asymmetric Huber regression (AHR) estimation procedure to analyze skewed data with partial functional linear models. To automatically reflect distributional features as well as bound the influence of outliers effectively, we further propose a weighted composite asymmetric Huber regression (WCAHR) estimation procedure by combining the strength across multiple asymmetric Huber loss functions. The slope function and constant coefficients are estimated through minimizing the combined loss function and approximating the slope function with principal component analysis. The asymptotic properties of the proposed estimators are derived. To realize the WCAHR estimation, we also develop a practical algorithm based on pseudo data. Numerical results show that the proposed WCAHR estimators can well adapt to the different error distributions, and thus are more useful in practice. Two real data examples are presented to illustrate the applications of the proposed methods.



    This paper deals with periodic measures of the following reaction-diffusion lattice systems driven by superlinear noise defined on the integer set Zk :

    dui(t)+λ(t)ui(t)dtν(t)(u(i11,i2,,ik)(t)+ui1,i21,,ik(t)++ui1,i2,,ik1(t)2ku(i1,i2,,ik)(t)+u(i1+1,i2,,ik)(t)+u(i1,i2+1,,ik)(t)++u(i1,i2,,ik+1)(t))dt=fi(t,ui(t))dt+gi(t)dt+j=1(hi,j(t)+δi,jˆσi,j(t,ui(t)))dWj(t), (1.1)

    along with initial conditions:

    ui(0)=u0,i, (1.2)

    where i=(i1,i2,,ik)Zk, λ(t),ν(t) are continuous functions, λ(t)>0, (fi)iZk and (ˆσi,j)iZk,jN are two sequences of continuously differentiable nonlinearities with arbitrary and superlinear growth rate from R×RR, respectively, g=(gi)iZk and h=(hi,j)iZk,jN are two time-dependent random sequences, and δ=(δi,j)iZk,jN is a sequence of real numbers. The sequence of independent two-sided real-valued Wiener processes (Wj)jN is defined on a complete filtered probability space (Ω,F,{Ft}tR,P). Furthermore, we assume that system (1.1) is a time periodic system; more precisely, there exists T>0 such that the time-dependent functions λ,ν,fi,g,h,σi,j(iZk,jN) in (1.1) are all T-periodic in time.

    Lattice systems are gradually becoming a large and evolving interdisciplinary research field, due to wide range of applications in physics, biology and engineering such as pattern recognition, propagation of nerve pulses, electric circuits, and so on, see [1,2,3,4,5,6] and the references therein for more details. The well-posedness and the dynamics of these equations have been studied by many authors, [7,8,9,10] for deterministic systems and [11,12,13,14,15,16,17,18,19] for stochastic systems where the existence of random attractors and probability measures have been examined. Especially, the authors research the limiting behavior of periodic measures of lattice systems in [15].

    Nonlinear noise was proposed and studied for the first time in [19], the authors researches the long-term behavior of lattice systems driven by nonlinear noise in terms of random attractors and invariant measures. Before that, the research on noise was limited to additive noise and linear multiplicative noise, which can be transformed into a deterministic system. However, if the diffusion coefficients are nonlinear, then one cannot convert the stochastic system into a pathwise deterministic one, and thereby this problem cannot be studied under the frameworks of deterministic systems aforementioned. As an extension of [19], a class of reaction-diffusion lattice systems driven by superlinear noise, where the noise has a superlinear growth order q[2,p), is studied by taking advantage of the dissipativeness of the nonlinear drift function fi in (1.1) to control the superlinear noise in [20].

    In the paper, we will study the existence of periodic measures of reaction-diffusion lattice systems drive by superlinear noise. One of the main tasks in our analysis is to solve the superlinear noise terms. We remark that if the noise grows linearly, then the estimates we need can be obtained by applying the standard methods available in the literature. We adopt the ideas that take advantage of the nonlinear drift terms' the polynomical growth rate p (p2) to control the noise polynomical rate q[2,p). Furthermore, notice that l2 is an infinite-dimensional phase space and problem (1.1)–(1.2) is defined on the unbounded set Zk. The unboundedness of Zk as well as the infinite-dimensionalness of l2 introduce a major difficulty, because of the non-compactness of usual Sobolev embeddings on unbounded domains. We will employ the dissipativeness of the drift function in (1.1) as well as a cutoff technique to prove that the tails of solutions are uniformly small in L2(Ω,l2). Based upon this fact we obtain the tightness of distribution laws of solutions, and then the existence of periodic measures.

    In the next section, we discuss the well-poseness of solutions of (1.1) and (1.2). Section 3 is devoted to the uniform estimates of solutions including the uniform estimates on the tails of solutions. In Section 4, we show the existence of periodic measures of (1.1) and (1.2).

    In this section, we prove the existence and uniqueness of solutions to system (1.1) and (1.2). We first discuss the assumptions on the nonlinear drift and diffusion terms in (1.1).

    We begin with the following Banach space:

    lr={u=(ui)iZk:iZk|ui|r<+} with norm ur=(iZk|ui|r)1r,r1.

    The norm and inner product of l2 are denoted by (,) and , respectively. For the nonlinear drift function fiC1(R×R,R) in the equation we assume that for all sR and iZk,

    fi(t,s)sγ1|s|p+ϕ1,i, ϕ1={ϕ1,i}iZkl1, (2.1)
    |fi(t,s)|ϕ2,i|s|p1+ϕ3,i, ϕ2={ϕ2,i}iZkl, ϕ3={ϕ3,i}iZkl2, (2.2)
    |fi(t,s)|ϕ4,i|s|p2+ϕ5,i, ϕ4={ϕ4,i}iZkl, ϕ5={ϕ5,i}iZkl, (2.3)

    where p>2 and γ1>0 are constants. For the sequence of continuously differentiable diffusion functions ˆσ=(ˆσi,j)iZk,jN, we assume, for all sR and jN,

    |ˆσi,j(t,s)|φ1,i|s|q2+φ2,i, φ1={φ1,i}iZkl2ppq, φ2={φ2,i}iZkl2, (2.4)
    |ˆσi,j(t,s)|φ3,i|s|q21+φ4,i, φ3={φ3,i}iZklq, φ4={φ4,i}iZkl, (2.5)

    where q[2,p) is a constant. For processes g(t)=(gi(t))iZk and h(t)=(hi,j)iZk,jN are both continuous in tR, which implies that for all tR,

    g(t)2=iZk|gi(t)|2< and h(t)2=iZkjN|hi,j(t)|2<. (2.6)

    In addition, we assume δ=(δi,j)iZk,jN satisfies

    cδ:=jNiZk|δi,j|2<. (2.7)

    We will investigate the periodic measures of system (1.1)–(1.2) for which we assume that all given time-dependent functions are T-periodic in tR for some T>0; that is, for all tR,iZk and kN.

    λ(t+T)=λ(t),ν(t+T)=ν(t),h(t+T)=h(t),g(t+T)=g(t),f(t+T,)=f(t,),σ(t+T,)=σ(t,).

    If m:RR is a continuous T-periodic function, we denote

    ¯m=max0tTm(t),m_=min0tTm(t).

    We want to reformulate problem (1.1)–(1.2) as an abstract one in l2. Given 1jk,u=(ui)iZkl2 and i=(i1,i2,,ik)Zk. Let us define the operators from l2 to l2 by

    (Bju)i=u(i1,,ij+1,,ik)u(i1,,ij,,ik),(Bju)i=u(i1,,ij1,,ik)u(i1,,ij,,ik),(Aju)i=u(i1,,ij+1,,ik)+2u(i1,,ij,,ik)u(i1,,ij1,,ik),

    and

    (Aku)i=u(i11,i2,,ik)u(i1,i21,,ik)u(i1,i2,,ik1)+2ku(i1,i2,,ik)u(i1+1,i2,,ik)u(i1,i2+1,,ik)u(i1,i2,,ik+1).

    For all 1jk,u=(ui)iZkl2 and v=(vi)iZkl2 we see

    Bju2u,(Bju,v)=(u,Bjv),Aj=BjBj and Ak=kj=1Aj. (2.8)

    Again, define the operators f,σj:R×l2l2 by

    f(t,u)=(fi(t,ui))iZk and σj(t,u)=(δi,jˆσi,j(t,ui))iZk,tR,u=(ui)iZkl2.

    It follows from (2.3) that there exists θ(0,1) such that for p>2 and u,vl2,

    iZk|fi(t,ui)fi(t,vi)|2=iZk|fi(θui+(1θ)vi)|2|uivi|2iZk(|ϕ4,i||θui+(1θ)vi|p2+|ϕ5,i|)2|uivi|2iZk(22p4|ϕ4,i|2(|ui|2p4+|vi|2p4)+2|ϕ5,i|2)|uivi|2(22p4ϕ42l(u2p4+v2p4)+2ϕ52l)uv2. (2.9)

    This together with f(t,0)l2 by (2.2) yields f(t,u)l2 for all ul2, and thereby f:R×l2l2 is well-defined. In addition, we deduce from (2.9) that f:R×l2l2 is a locally Lipschitz continuous function, that is, for every nN, we can find a constant c1(n)>0 satisfying, for all u,vl2 with un and vn,

    f(u)f(v)c1(n)uv. (2.10)

    For q[2,p) and ul2, one can deduce from(2.4), (2.7) and Young's inequality that for all ϖ>0,

    ϖjNσj(t,u)2=ϖjNiZk|δi,jˆσi,j(t,ui)|22ϖjNiZk|δi,j|2(|φ1,i|2|ui|q+|φ2,i|2)2ϖcδiZk(|φ1,i|2|ui|q+|φ2,i|2)γ12iZk|ui|p+pqp(pγ12q)qpq(2ϖcδ)ppqiZk|φ1,i|2ppq+2ϖcδiZk|φ2,i|2γ12upp+pqp(pγ12q)qpq(2ϖcδ)ppqφ12ppq2ppq+2ϖcδφ22, (2.11)

    where γ1 is the same number as in (2.1). From (2.11) and l2lp for p>2, we find that σj(t,u)l2 for all ul2. Then σj:R×l2l2 is also well-defined. In addition, it yields from (2.5) and (2.7) that there exists η(0,1) such that for q[2,p) and u,vl2,

    jNiZk|δi,jˆσi,j(t,ui)δi,jˆσi,j(t,vi)|2=iZkjN|δi,j|2|ˆσi,j(t,ui)ˆσi,j(t,vi)|2=iZkjN|δi,j|2|ˆσi,j(ηui+(1η)vi)|2|uivi|2cδiZk(|φ3,i||ηui+(1η)vi|q21+|φ4,i|)2|uivi|2cδiZk(2q2|φ3,i|2(|ui|q2+|vi|q2)+2|φ4,i|2)|uivi|2cδiZk(2q2(4q|φ3,i|q+q2q|ui|q+q2q|vi|q)+2|φ4,i|2)|uivi|2cδ(2q1(φ3qq+uq+vq)+2φ42l)uv2. (2.12)

    This implies that σj:R×l2l2 is also locally Lipschitz continuous, more precisely, for every nN, one can find a constant c2(n)>0 satisfying, for all u,vl2 with un and vn,

    jNσj(u)2c22(n). (2.13)

    and

    jNσj(u)σj(v)2c22(n)uv2. (2.14)

    By above notations one is able to rewrite (1.1)–(1.2) as the following system in l2 for t>0 :

    du(t)+ν(t)Aku(t)dt+λ(t)u(t)dt=f(t,u(t))dt+g(t)dt+j=1(hj(t)+σj(t,u(t)))dWj(t), (2.15)

    with initial condition:

    u(0)=u0l2, (2.16)

    in the present article, the solutions of system (2.15)–(2.16) are interpreted in the following sense.

    Definition 2.1. Suppose u0L2(Ω,l2) is F0-measurable, a continuous l2-valued Ft-adapted stochastic process u is called a solution of equations (2.15) and (2.16) if uL2(Ω,C([0,T],l2))Lp(Ω,Lp(0,T;lp)) for all T>0, and the following equation holds for all t0 and almost all ωΩ:

    u(t)=u0+t0(ν(s)Aku(s)λ(s)u(s)+f(s,u(s))+g(s))ds+j=1t0(hj(s)+σj(s,u(s)))dWj(s) in l2. (2.17)

    Similar to Ref.[20], we can get (2.15) and (2.16) exist global solutions in the sense of Definition 2.1.

    In this section, we derive the uniform estimates of solutions of (2.15)–(2.16). These estimates will be used to establish the tightness of a set of probability distributions of u in l2.

    We assume that

    α(t)=λ(t)16k|ν(t)|>0. (3.1)

    Lemma 3.1. Let (2.1)–(2.7) and (3.1) hold. Then the solutions u(t,0,u0) of system (2.15) and (2.16) with initial data u0 at time 0 satisfy, for all t0,

    E(u(t,0,u0)2)+t0eα_(rt)E(u(r,0,u0)pp)drL1(E(u02)+j=1¯hj2+¯g2+φ12ppq2ppq+φ22+ϕ11), (3.2)

    where L1>0 is a positive constant which depends on α_,p,q,γ,cδ,t, but indepentent of u0.

    Proof. Applying Ito's formula to (2.15) we get

    d(u(t)2)+2ν(t)kj=1Bju(t)2dt+2λ(t)u(t)2dt=2(f(t,u(t)),u(t))dt+2(g(t),u(t))dt+j=1hj(t)+σ(t,u(t))2dt+2j=1u(t)(hj(t)+σj(t,u(t)))dWj(t).

    This implies

    ddtE(u(t)2)+2ν(t)kj=1E(Bju(t)2)+2λ(t)E(u(t)2)2E(f(t,u(t)),u(t))+2E(g(t),u(t))+2j=1E(hj(t)2)+2j=1E(σ(t,u(t))2). (3.3)

    For the second term on the left-hand side of (3.3), we have

    2|ν(t)|kj=1E(Bju(t)2)8k|ν(t)|E(u(t)2). (3.4)

    For the first term on the right-hand side of (3.3), we get from (2.1) that

    2E(f(t,u(t)),u(t))2γ1E(u(t)pp)+2ϕ11. (3.5)

    For the second term on the right-hand side of (3.3), we have

    2E(g(t),u(t))λ(t)E(u(t)2)+1λ(t)E(g(t)2). (3.6)

    For the last term on the right-hand side of (3.3), we infer from (2.11) with ω=2 that

    2j=1E(σj(t,u(t))2)γ12E(u(t)pp)+pqp(pγ12q)qpq(4cδ)ppqφ12ppq2ppq+4cδφ22. (3.7)

    By (3.3)–(3.7) we get

    ddtE(u(t)2)+α_E(u(t)2)+32γ1E(u(t)pp)E(j=12hj(t)2+1λ(t)g(t)2)+C1, (3.8)

    implies that

    ddtE(u(t)2)+α_E(u(t)2)+32γ1E(u(t)pp)2j=1¯hj2+1λ_¯g2+C1, (3.9)

    where C1=pqp(pγ12q)qpq(4cδ)ppqφ12ppq2ppq+4cδφ22+2ϕ11. Multiplying (3.9) by eα_t and integrating over (0,t) to obtain

    E(u(t,0,u0)2)+32γ1t0eα_(rt)E(u(r,0,u0)pp)dreα_tE(u02)+C2t0eα_(rt)dr, (3.10)

    where C2=2j=1¯hj2+1λ_¯g2+C1. This completes the proof.

    Lemma 3.2. Let (2.1)–(2.7), and (3.1) be satisfied. Then for compact subset K of l2, one can find a number N0=N0(K)N such that the solutions u(t,0,u0) of (2.15) and (2.16) satisfy, for all nN0 and t0,

    E(in|ui(t,0,u0)|2)+t0eα_(rt)E(in|ui(r,0,u0)|p)drε, (3.11)

    where u0K and i:=maxijk|ij|.

    Proof. Define a smooth function ξ:R[0,1] such that

    ξ(s)=0 for |s|1 and ξ(s)=1 for |s|2. (3.12)

    Denote by

    ξn=(ξ(in))iZk and ξnu=(ξ(in)ui)iZk,u=(ui)iZk,nN. (3.13)

    Similar notations will also be used for other terms. It follows from (2.15) that

    d(ξnu(t))+ν(t)ξnAku(t)dt+λ(t)ξnu(t)dt=ξnf(t,u(t))dt+ξng(t)dt+j=1(ξnhj(t)+ξnσj(t,u(t)))dWj(t). (3.14)

    By Ito's formula and (3.14) we have

    dξnu(t)2+2ν(t)(Ak(u(t)),ξ2nu(t))dt+2λ(t)ξnu(t)2dt=2(f(t,u(t)),ξ2nu(t))dt+2(g(t),ξ2nu(t))dt+j=1ξnhj(t)+ξnσj(t,u(t))2dt+2j=1(hj(t)+σj(t,u(t)),ξ2nu(t))dWj. (3.15)

    This yields

    ddtE(ξnu(t)2)+2ν(t)E(Ak(u(t)),ξ2nu(t))+2λ(t)E(ξnu(t)2)=2E(f(t,u(t)),ξ2nu(t))+2E(g(t),ξ2nu(t))+2j=1E(ξnhj(t)2)+2j=1E(ξnσj(t,u(t))2)dt. (3.16)

    For the second term on the left-hand side of (3.16), we have

    2ν(t)E(Ak(u(t)),ξ2nu(t))=2ν(t)kj=1E(Bju(t),Bj(ξ2nu(t)))=2ν(t)E(kj=1iZk(ui1,,ij+1,,ikui)×(ξ2((i1,,ij+1,,ik)n)u(i1,,ij+1,,ik)ξ2(in)ui))=2ν(t)E(kj=1iZkξ2(in)(ui1,,ij+1,,ikui)2)+2ν(t)E(kj=1iZk(ξ2((i1,,ij+1,,ik)n)ξ2(in))×(u(i1,,ij+1,,ik)ui)u(i1,,ij+1,,ik)). (3.17)

    We first deal with the first term on the right-hand side of (3.17). Notice that

    2|ν(t)|E(kj=1iZkξ2(in)(ui1,,ij+1,,ikui)2)=2|ν(t)|E(kj=1iZk|ξ(in)u(i1,,ij+1,,ik)ξ(in)ui|2)4|ν(t)|E(kj=1iZk|(ξ(in)ξ((i1,,ij+1,,ik)n))u(i1,,ij+1,,ik)|2)+4|ν(t)|E(kj=1iZk|ξ((i1,,ij+1,,ik)n)u(i1,,ij+1,,ik)ξ(in)ui|2). (3.18)

    By the definition of function ξ, there exists a constant C3>0 such that |ξ(s)|C3 for all sR. Then the first term on the right-hand side of (3.18) is bounded by

    4|ν(t)|E(kj=1iZk|(ξ(in)ξ((i1,,ij+1,,ik)n))u(i1,,ij+1,,ik)|2)=4|ν(t)|E(kj=1iZk|ξ(in)ξ((i1,,ij+1,,ik)n)|2|u(i1,,ij+1,,ik)|2)4C23n2|ν(t)|E(kj=1iZk|u(i1,,ij+1,,ik)|2)4C23kn2|ν(t)|E(u2). (3.19)

    By the definition of |Bju|i, the last term on the right-hand side of (3.18) is bounded by

    4|ν(t)|E(kj=1iZk|ξ((i1,,ij+1,,ik)n)u(i1,,ij+1,,ik)ξ(in)ui|2)4|ν(t)|E(kj=1Bj(ξnu(t))2)16k|ν(t)|E(ξnu(t)2). (3.20)

    Then we find from (3.18) to (3.20) that the first term on the right-hand side of (3.17) is bounded by

    2|ν(t)|E(kj=1iZkξ2(in)(u(i1,,ij+1,,ik)ui)2)16k|ν(t)|E(ξnu(t)2)+4C23kn2|ν(t)|E(u2). (3.21)

    In addition, we find that the last term on the right-hand side of (3.17) can be bounded by

    2|ν(t)E(kj=1iZk(ξ2((i1,,ij+1,,ik)n)ξ2(in))×(u(i1,,ij+1,,ik)ui)u(i1,,ij+1,,ik))|2|ν(t)|E(kj=1iZk|ξ2((i1,,ij+1,,ik)n)ξ2(in)|×|u(i1,,ij+1,,ik)ui||u(i1,,ij+1,,ik)|)4|ν(t)|E(kj=1iZk|ξ((i1,,ij+1,,ik)n)ξ(in)|×|u(i1,,ij+1,,ik)ui||u(i1,,ij+1,,ik)|)4C3n|ν(t)|E(kj=1iZk|u(i1,,ij+1,,ik)ui||u(i1,,ij+1,,ik)|)8kC3n|ν(t)|E(u2). (3.22)

    By (3.21), (3.22) and (3.17), we infer that the second term on the left-hand side of (3.16) satisfied

    2|ν(t)E(Ak(u(t)),ξ2nu(t))|C4|ν(t)|(1n+1n2)E(u2)+16k|ν(t)|E(ξnu(t)2), (3.23)

    where C4=4kC3(2+C3). For the first term on the right-hand side of (3.16), we find from (2.1) that

    2E(f(t,u(t)),ξ2nu(t))2γ1E(iZkξ2(in)|ui(t)|p)+2E(iZkξ2(in)|ϕ1,i|)2γ1E(iZkξ2(in)|ui(t)|p)+2in|ϕ1,i|. (3.24)

    For the second term on the right-hand side of (3.16), we infer from Young's inequality that

    2E(g,ξ2nu(t))λ_E(ξnu(t)2)+1λ_E(iZkξ2(in)|gi|2)λ_E(ξnu(t)2)+1λ_in|gi|2. (3.25)

    For the last term on the right-hand side (3.16), we infer from (2.4) and Young's inequality that

    2j=1E(ξnσj(t,u(t))2)=2j=1E(iZk|ξ(in)δi,jˆσi,j(t,ui(t))|2)4j=1E(iZkξ2(in)|δi,j|2(|φ1,i|2|ui(t)|q+|φ2,i|2))4cδE(iZkξ2(in)(|φ1,i|2|ui(t)|q+|φ2,i|2))γ1E(iZkξ2(in)|ui(t)|p)+pqp(pγ1q)qpq(4cδ)ppqiZkξ2(in)|φ1,i|2ppq+4cδiZkξ2(in)|φ2,i|2γ1E(iZkξ2(in)|ui(t)|p)+pqp(pγ1q)qpq(4cδ)ppqin|φ1,i|2ppq+4cδin|φ2,i|2. (3.26)

    Substituting (3.23)–(3.26) into (3.16) we get

    ddtE(ξnu(t)2)+α_E(ξnu(t)2)+γ1E(iZkξ2(in)|ui(t)|p)C4|ν|(1n+1n2)E(u2)+C5(in(¯|gi|2+|φ1,i|2ppq+|φ2,i|2+|ϕ1,i|)+inj=1¯|hi,j|2), (3.27)

    where C5=2+1λ_+pqp(pγ1q)qpq(4cδ)ppq+4cδ. One can multiply (3.27) by eα_t and integrate over (0,t) in order to obtain

    E(ξnu(t,0,u0)2)+γ1t0eα_(rt)E(iZkξ2(in)|ui(r,0,u0)|p)dreα_tE(ξnu02)+C4|ν|(1n+1n2)t0eα_(rt)E(u(r,0,u0)2)dr+C5α_(in(¯|gi|2+|φ1,i|2ppq+|φ2,i|2+|ϕ1,i|)+inj=1¯|hi,j|2). (3.28)

    Since \mathcal K is a compact subset of l^2 we infer from (3.1) that

    \begin{equation} \lim\limits_{n\to\infty}\sup\limits_{u_0\in\mathcal K}\sup\limits_{t\ge 0}e^{-\underline{\alpha}t}E(\|\xi_nu_0\|^2)\le \lim\limits_{n\to\infty}\sup\limits_{u_0\in\mathcal K}E(\sum\limits_{\|i\|\ge n}|u_{0,i}|^2) = 0. \end{equation} (3.29)

    By Lemma 3.1, we find that for all u_0\in\mathcal K and t\ge 0 , as n\to\infty ,

    \begin{align} \begin{split} &\Big(\frac{1}{n}+\frac{1}{n^2}\Big)\int_{0}^{t}e^{\underline{\alpha}(r-t)}E(\|u(r,0,u_0)\|^2)dr\\ &\le \frac{L_1}{\underline{\alpha}}\Big(\frac{1}{n}+\frac{1}{n^2}\Big)\Big(E(\|u_0\|^2)+\sum\limits_{j = 1}^{\infty}\overline{\|h_j\|}^2+\overline{\|g\|}^2+\|\varphi_1\|^{\frac{2p}{p-q}}_{\frac{2p}{p-q}}+\|\varphi_2\|^2+\|\phi_1\|_1\Big)\\ &\le \frac{L_1}{\underline{\alpha}}\Big(\frac{1}{n}+\frac{1}{n^2}\Big)\Big(C_{6}+\sum\limits_{j = 1}^{\infty}\overline{\|h_j\|}^2+\overline{\|g\|}^2+\|\varphi_1\|^{\frac{2p}{p-q}}_{\frac{2p}{p-q}}+\|\varphi_2\|^2+\|\phi_1\|_1\Big)\to0, \end{split} \end{align} (3.30)

    where L_1 is the same number of (3.1) and C_{6} > 0 is a constant depending only on u_0 .By \varphi_1\in l^{\frac{2p}{p-q}}, \varphi_2\in l^2, \phi_1\in l^1 , (2.6) and (3.1), we infer that

    \begin{equation} \sum\limits_{\|i\|\ge n}\Big(\overline{|g_i|}^2+|\varphi_{1,i}|^{\frac{2p}{p-q}}+|\varphi_{2,i}|^2+|\phi_{1,i}|\Big)+\sum\limits_{\|i\|\ge n}\sum\limits_{j = 1}^{\infty}\overline{|h_{i,j}|}^2\to 0\ \mathit{\text{as}}\ n\to\infty. \end{equation} (3.31)

    It follows from (3.28) to (3.31) that as n\to\infty ,

    \begin{equation} \sup\limits_{u_0\in\mathcal K}\sup\limits_{t\ge 0}\bigg(E(\|\xi_nu(t,0,u_0)\|^2)+\int_{0}^{t}e^{\underline{\alpha}(r-t)}E\Big(\sum\limits_{i\in\mathbb Z^k}\xi^2\Big(\frac{\|i\|}{n}\Big)|u_i(r,0,u_0)|^p\Big)dr\bigg)\to0. \end{equation} (3.32)

    Then for every \varepsilon > 0 we can find a number N_0 = N_0(\mathcal K)\in\mathbb N satisfying, for all n\ge N_0 and t\ge 0 ,

    \begin{align} \begin{split} &\bigg(E\Big(\sum\limits_{\|i\|\ge 2n}|u_i(t,0,u_0)|^2\Big)+\int_{0}^{t}e^{\underline{\alpha}(r-t)}E\Big(\sum\limits_{\|i\|\ge 2n}|u_i(t,0,u_0)|^p\Big)dr\bigg)\\ &\le \bigg(E\Big(\|\xi_nu(t,0,u_0)\|^2\Big)+\int_{0}^{t}e^{\underline{\alpha}(r-t)}E\Big(\sum\limits_{i\in\mathbb Z^k}\xi^2\Big(\frac{\|i\|}{n}\Big)|u_i(t,0,u_0)|^p\Big)dr\bigg)\le\varepsilon, \end{split} \end{align} (3.33)

    uniformly for u_0\in\mathcal K and t\ge 0 . This concludes the proof.

    In the sequel, we use \mathcal L(u(t, 0, u_0)) to denote the probability distribution of the solution u(t, 0, u_0) of (2.15)–(2.16) which has initial condition u_0 at initial time 0 . Then we have the following tightness of a family of distributions of solutions.

    Lemma 4.1. Suppose (2.1)–(2.7) and (3.1) hold. Then the family \{\mathcal L(u(t, 0, u_0)):t\ge 0\} of the distributions ofthe solutions of (2.15)–(2.16) is tight on l^2 .

    Proof. For simplicity, we will write the solution u (t, 0, u_0) as u(t) from now on. It follows from Lemma 3.1 that there exists a constant c_1 > 0 such that

    \begin{equation} E\left( {\left\| {u(t) } \right\|^2 } \right) \le c_1 ,\quad\text{for all} \quad t\geq0. \end{equation} (4.1)

    By Chebyshev's inequality, we get from (4.1) that for all t\geq0 ,

    P\left( {\left\| {u(t) } \right\|^2\ge R } \right) \le \frac{{c_1 }}{{R^2 }}\rightarrow 0\quad\text{as}\quad R\rightarrow \infty.

    Hence for every \epsilon > 0 , there exists R_1 = R_1(\epsilon) > 0 such that for all t\geq0 ,

    \begin{equation} P\left\{ {\left\| {u(t) } \right\|^2 \ge R_1 } \right\} \le \frac{1}{2}\epsilon. \end{equation} (4.2)

    By Lemma 3.2, we infer that for each \epsilon > 0 and m\in \mathbb{N} , there exists an integer n_m = n_m(\epsilon, m) such that for all t\geq 0 ,

    E\left( { \sum\limits_{\left| i \right| > n_m } {\left| {u_i \left( t \right)} \right|^2 } } \right) < \frac{\epsilon }{{2^{2m + 2} }},

    and hence for all t\geq 0 and m\in \mathbb{N} ,

    \begin{equation} P\left( {\left\{ { \sum\limits_{\left| i \right| > n_m } {\left| {u_i \left( r \right)} \right|^2 } \ge \frac{1}{{2^m }}} \right\}} \right) \le 2^m E\left( { \sum\limits_{\left| i \right| > n_m } {\left| {u_i \left( r \right)} \right|^2 } } \right) < \frac{\epsilon }{{2^{m + 2} }}. \end{equation} (4.3)

    It follows from (4.3) for all t\geq 0 ,

    P\left( {\mathop \cup \limits_{m = 1}^\infty \left\{ { \sum\limits_{\left| i \right| > n_m } {\left| {u_i \left( t \right)} \right|^2 } \ge \frac{1}{{2^m }}} \right\}} \right) \le \sum\limits_{m = 1}^\infty {\frac{\epsilon }{{2^{m + 2} }} \le \frac{1}{4}\epsilon ,}

    which shows that for all t\geq 0 ,

    \begin{equation} P\left( {\left\{ { \sum\limits_{\left| i \right| > n_m } {\left| {u_i \left( t \right)} \right|^2 } \le \frac{1}{{2^m }}\,\,\text{for all}\,\, m\in \mathbb{N}} \right\}} \right) > 1 -\frac{\epsilon}{2}. \end{equation} (4.4)

    Given \epsilon > 0 , set

    \begin{align} Y_{1,\epsilon } & = \left\{ {v \in l^2:\left\| {v } \right\| \le R_1 \left(\epsilon \right)} \right\}, \end{align} (4.5)
    \begin{align} Y_{2,\epsilon } & = \left\{ {v \in l^2: \sum\limits_{\left| i \right| > n_m } {\left| {v_i \left( r \right)} \right|^2 } \le \frac{1}{{2^m }}\,\,\text{for all}\,\, m\in \mathbb{N}} \right\}, \end{align} (4.6)

    and

    \begin{equation} Y_\epsilon = Y_{1,\epsilon} \cap Y_{2,\epsilon }. \end{equation} (4.7)

    By (4.2) and (4.4) we get, for all t\geq0 ,

    \begin{equation} P\left( {\left\{ {u(t) \in Y_\epsilon } \right\}} \right) > 1 - \epsilon . \end{equation} (4.8)

    Now, we show the precompactness of \left\{ {v:v \in Y_\epsilon } \right\} in l^2 . Given \kappa > 0 , choose an integer m_0 = m_0 \left(\kappa \right) \in \mathbb{N} such that 2^{m_0 } > \frac{8}{{\kappa ^2 }} . Then by (4.6) we obtain

    \begin{equation} \sum\limits_{\left| i \right| > n_{m_0 } } {\left| {v_i } \right|^2 } \le \frac{1}{{2^{m_0 } }} < \frac{{\kappa ^2 }}{8},\quad \forall v \in Y_\epsilon. \end{equation} (4.9)

    On the other hand, by (4.5) we see that the set \left\{ {(v_i)_{|i|\leq m_0}:v \in Y_\epsilon } \right\} is bounded in the finite-dimensional space R^{2m_0+1} and hence precompact. Consequently, \left\{ {v:v \in Y_\epsilon } \right\} has a finite open cover of balls with radius \frac{\kappa }{2} , which along with (4.9) implies that the set \left\{ {v:v \in Y_\epsilon } \right\} has a finite open cover of balls with radius \kappa in l^2 . Since \kappa > 0 is arbitrary, we find that the set \left\{ {v: v\in Y_\epsilon } \right\} is precompact in l^2 . This completes the proof.

    If \phi:l^2\to\mathbb R is a bounded Borel function, then for 0\le r\le t and u_0\in l^2 , we set

    (p_{r,t}\phi)(u_0) = E(\phi(u(t,r,u_0)))

    and

    p(r,u_0;t,\Gamma) = (p_{r,t}1_\Gamma)(u_0),

    where \Gamma\in\mathcal B(l^2) and 1_\Gamma is the characteristic function of \Gamma . The operators p_{s, t} with 0\le s\le t are called the transition operators for the solutions of (2.15)–(2.16). Recall that a probability measure \nu on l^2 is periodic for (2.15)–(2.16) if

    \int_{l^2}(p_{0,t+T}\phi)(u_0)d\nu(u_0) = \int_{l^2} (p_{0,t}\phi)(u_0)d\nu(u_0),\qquad\forall t\ge0.

    Lemma 4.2. [21]Let \varrho(\psi, \omega) be a scalar bounded measurable randomfunction of \psi , independent of \mathcal F_s . Let \varsigma be an \mathcal F_s -measurable random variable. Then

    E\left( {\varrho \left( {\varsigma ,\omega } \right)|\mathcal F_s } \right) = E\left( {\varrho \left( {\varsigma ,\omega } \right)} \right).

    The transition operators \{p_{r, t}\}_{0\le r\le t} have the following properties.

    Lemma 4.3. Assume that (2.1)–(2.7) and (3.1) hold. Then:

    (i) \{p_{r, t}\}_{0\le r\le t} is Feller; that is, for every bounded andcontinuous \phi: l^2\to\mathbb R , the function p_{r, t}\phi: l^2\to\mathbb R is also bounded and continuous for all 0\le r\le t.

    (ii) The family \{p_{r, t}\}_{0\le r\le t} is T-periodic; that is, for all 0\le r\le t ,

    p(r, u_0;t,\cdot) = p(r+T,u_0;t+T,\cdot),\qquad\forall u_0\in l^2.

    (iii) \{u(t, 0, u_0)\}_{t\ge 0} is a l^2 -valued Markov process.

    Finally, we present our main result on the existence of periodic measures for problem (2.15)–(2.16).

    Theorem 4.4. Assume that (2.1)–(2.7) and (3.1) hold. Then problem (2.15)–(2.16) has a periodic measure on l^2 .

    Proof. We apply Krylov-Bogolyubov's method to prove the existence of periodic measures of (2.15)–(2.16), define a probability measure \mu_n by

    \begin{equation} \mu_n = \frac{1}{n}\sum\limits_{l = 1}^{n}p(0,0;lT,\cdot). \end{equation} (4.10)

    By Lemma 4.1 we see the sequence \{\mu_n\}^\infty_{n = 1} is tight on l^2 , and hence there exists a probability measure \mu on l^2 such that, up to a subsequence,

    \begin{equation} \mu_n\to\mu,\qquad\text{as}\ n\to\infty. \end{equation} (4.11)

    By (4.10)–(4.11) and Lemma 4.3, we infer that for every t\ge0 and every bounded and continuous function \phi:l^2\to\mathbb R,

    \begin{align} \begin{split} &{\int_{l^2} {\left( {p_{0,t} \phi } \right)\left( u_0 \right)d\mu \left( u_0 \right)} } = \int_{l^2} {\left( {\int_{l^2} {\phi \left( y \right) p\left( {0, u_0 ;t,dy} \right)} } \right)} d\mu \left( u_0 \right) \\ & = \mathop {\lim }\limits_{n \to \infty } \frac{1}{n}\sum\limits_{l = 1}^n {\int_{l^2} {\left( {\int_{l^2} {\phi \left( y \right)p\left( {0,u_0 ;t,dy} \right)} } \right)} p\left( {0,0;lT,du_0 } \right)} \\ & = \mathop {\lim }\limits_{n \to \infty } \frac{1}{n}\sum\limits_{l = 1}^n {\int_{l^2} {\left( {\int_{l^2} {\phi \left( y \right)p\left( {kT,u_0 ;t + lT,dy} \right)} } \right)} p\left( {0,0;kT,du_0 } \right)} \\ & = \mathop {\lim }\limits_{n \to \infty } \frac{1}{n}\sum\limits_{l = 1}^n {\int_{l^2} {\phi \left( y \right)p\left( {0,0;t + lT,dy} \right)} } \\ & = \mathop {\lim }\limits_{n \to \infty } \frac{1}{n}\sum\limits_{l = 1}^n {\int_{l^2} {\phi \left( y \right)p\left( {0,0;t + lT + T,dy} \right)} } \\ & = \mathop {\lim }\limits_{n \to \infty } \frac{1}{n}\sum\limits_{k = 1}^n {\int_{l^2} {\left( {\int_{l^2} {\phi \left( y \right)p\left( {0,u_0 ;t + T,dy} \right)} } \right)} p\left( {0,0;lT,du_0 } \right)} \\ & = \int_{l^2} {\left( {\int_{l^2} {\phi \left( y \right)p\left( {0,u_0;t + T,dy} \right)} } \right)} d\mu \left( u_0\right)\\ & = {\int_{l^2} {\left( {p_{0,t + T} \phi } \right)\left( u_0 \right)d\mu \left( u_0 \right)} } , \end{split} \end{align} (4.12)

    which shows that \mu is a periodic measure of (2.15)–(2.16), as desired.

    The author declares there is no conflict of interest.



    [1] J. Ramsay, B. Silverman, Functional data analysis, 2 Eds., New York: Springer, 2005. http://dx.doi.org/10.1007/b98888
    [2] F. Yao, H. Müller, J. Wang, Functional linear regression analysis for longitudinal data, Ann. Statist., 33 (2005), 2873–2903. http://dx.doi.org/10.1214/009053605000000660 doi: 10.1214/009053605000000660
    [3] P. Hall, J. Horowitz, Methodology and convergence rates for functional linear regression, Ann. Statist., 35 (2007), 70–91. http://dx.doi.org/10.1214/009053606000000957 doi: 10.1214/009053606000000957
    [4] H. Shin, Partial functional linear regression, J. Stat. Plan. Infer., 139 (2009), 3405–3418. http://dx.doi.org/10.1016/j.jspi.2009.03.001 doi: 10.1016/j.jspi.2009.03.001
    [5] P. Hall, G. Hooker, Truncated linear models for functional data, J. Roy. Stat. Soc. Ser. B, 78 (2016), 637–653. http://dx.doi.org/10.1111/rssb.12125 doi: 10.1111/rssb.12125
    [6] K. Kato, Estimation in functional linear quantile regression, Ann. Statist., 40 (2012), 3108–3136. http://dx.doi.org/10.1214/12-AOS1066 doi: 10.1214/12-AOS1066
    [7] D. Kong, K. Xue, F. Yao, H. Zhang, Partially functional linear regression in high dimensions, Biometrika, 103 (2016), 147–159. http://dx.doi.org/10.1093/biomet/asv062 doi: 10.1093/biomet/asv062
    [8] P. Yu, Z. Zhang, J. Du, A test of linearity in partial functional linear regression, Metrika, 79 (2016), 953–969. http://dx.doi.org/10.1007/s00184-016-0584-x doi: 10.1007/s00184-016-0584-x
    [9] J. Zhou, Z. Chen, Q. Peng, Polynomial spline estimation for partial functional linear regression models, Comput. Stat., 31 (2016), 1107–1129. http://dx.doi.org/10.1007/s00180-015-0636-0 doi: 10.1007/s00180-015-0636-0
    [10] M. Imaizumi, K. Kato, PCA-based estimation for functional linear regression with functional responses, J. Multivariate Anal., 163 (2018), 15–36. http://dx.doi.org/10.1016/j.jmva.2017.10.001 doi: 10.1016/j.jmva.2017.10.001
    [11] Q. Tang, L. Cheng, Partial functional linear quantile regression, Sci. China Math., 57 (2014), 2589–2608. http://dx.doi.org/10.1007/s11425-014-4819-x doi: 10.1007/s11425-014-4819-x
    [12] P. Yu, Z. Zhu, Z. Zhang, Robust exponential squared loss-based estimation in semi-functional linear regression models, Comput. Stat., 34 (2019), 503–525. http://dx.doi.org/10.1007/s00180-018-0810-2 doi: 10.1007/s00180-018-0810-2
    [13] X. Cai, L. Xue, F. Lu, Robust estimation with a modified Huber's loss for partial functional linear models based on splines, J. Korean Stat. Soc., 49 (2020), 1214–1237. http://dx.doi.org/10.1007/s42952-020-00052-x doi: 10.1007/s42952-020-00052-x
    [14] Y. Lim, H. Oh, A data-adaptive principal component analysis: use of composite asymmetric Huber function, J. Comput. Graph. Stat., 25 (2015), 1230–1247. http://dx.doi.org/10.1080/10618600.2015.1067621 doi: 10.1080/10618600.2015.1067621
    [15] P. Huber, Robust estimation of a location parameter, Ann. Math. Statist., 35 (1964), 73–101. http://dx.doi.org/10.1214/aoms/1177703732 doi: 10.1214/aoms/1177703732
    [16] H. Zou, M. Yuan, Composite quantile regression and the oracle model selection theory, Ann. Statist., 36 (2008), 1108–1126. http://dx.doi.org/10.1214/07-AOS507 doi: 10.1214/07-AOS507
    [17] T. Cai, P. Hall, Prediction in functional linear regression, Ann. Statist., 34 (2006), 2159–2179. http://dx.doi.org/10.1214/009053606000000830 doi: 10.1214/009053606000000830
    [18] H. Ma, Y. Bai, Z. Zhu, Dynamic single-index model for functional data, Sci. China Math., 59 (2016), 2561–2584. http://dx.doi.org/10.1007/s11425-016-0051-3 doi: 10.1007/s11425-016-0051-3
    [19] Y. Lu, J. Du, Z. Sun, Functional partially linear quantile regression model, Metrika, 77 (2014), 317–332. http://dx.doi.org/10.1007/s00184-013-0439-7 doi: 10.1007/s00184-013-0439-7
    [20] H. Shin, M. Lee, On prediction rate in partial functional linear regression, J. Multivariate Anal., 103 (2012), 93–106. http://dx.doi.org/10.1016/j.jmva.2011.06.011 doi: 10.1016/j.jmva.2011.06.011
    [21] H. Zhu, R. Zhang, Z. Yu, H. Lian, Y. Liu, Estimation and testing for partially functional linear errors-in-variables models, J. Multivariate Anal., 170 (2019), 296–314. http://dx.doi.org/10.1016/j.jmva.2018.11.005 doi: 10.1016/j.jmva.2018.11.005
    [22] X. Cui, H. Lin, H. Lian. Partially functional linear regression in reproducing kernel Hilbert spaces, Comput. Stat. Data Anal., 150 (2020), 106978. http://dx.doi.org/10.1016/j.csda.2020.106978 doi: 10.1016/j.csda.2020.106978
    [23] D. Liebl, Modeling and forecasting electricity spot prices: a functional data perspective, Ann. Appl. Stat., 7 (2013), 1562–1592. http://dx.doi.org/10.1214/13-AOAS652 doi: 10.1214/13-AOAS652
    [24] M. Febrero-bande, M. Fuente, Statistical computing in functional data analysis: The R Package fda.usc, J. Stat. Softw., 51 (2012), 1–28. http://dx.doi.org/10.18637/jss.v051.i04 doi: 10.18637/jss.v051.i04
  • This article has been cited by:

    1. Xintao Li, Rongrui Lin, Lianbing She, Periodic measures for a neural field lattice model with state dependent superlinear noise, 2024, 32, 2688-1594, 4011, 10.3934/era.2024180
    2. Xintao Li, Lianbing She, Jingjing Yao, Periodic measures of fractional stochastic discrete wave equations with nonlinear noise, 2024, 57, 2391-4661, 10.1515/dema-2024-0078
    3. Hailang Bai, Mingkai Yuan, Dexin Li, Yunshun Wu, Weak and Wasserstein convergence of periodic measures of stochastic neural field lattice models with Heaviside ’s operators and locally Lipschitz Lévy noises, 2025, 143, 10075704, 108602, 10.1016/j.cnsns.2025.108602
  • Reader Comments
  • © 2022 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(2562) PDF downloads(104) Cited by(1)

Figures and Tables

Figures(4)  /  Tables(5)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog