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

Final dynamics of systems of nonlinear parabolic equations on the circle

  • Received: 12 June 2021 Accepted: 18 August 2021 Published: 17 September 2021
  • MSC : 35B41, 35K57, 35K40, 35K90, 35K91

  • We consider the class of dissipative reaction-diffusion-convection systems on the circle and obtain conditions under which the final (at large times) phase dynamics of a system can be described by an ODE with Lipschitz vector field in RN. Precisely in this class, the first example of a parabolic problem of mathematical physics without the indicated property was recently constructed.

    Citation: Aleksandr V. Romanov. Final dynamics of systems of nonlinear parabolic equations on the circle[J]. AIMS Mathematics, 2021, 6(12): 13407-13422. doi: 10.3934/math.2021776

    Related Papers:

    [1] Kaihong Zhao . Attractor of a nonlinear hybrid reaction–diffusion model of neuroendocrine transdifferentiation of human prostate cancer cells with time-lags. AIMS Mathematics, 2023, 8(6): 14426-14448. doi: 10.3934/math.2023737
    [2] Ruonan Liu, Tomás Caraballo . Random dynamics for a stochastic nonlocal reaction-diffusion equation with an energy functional. AIMS Mathematics, 2024, 9(4): 8020-8042. doi: 10.3934/math.2024390
    [3] Cyril Dennis Enyi, Soh Edwin Mukiawa . Dynamics of a thermoelastic-laminated beam problem. AIMS Mathematics, 2020, 5(5): 5261-5286. doi: 10.3934/math.2020338
    [4] Dieunel DOR . On the modified of the one-dimensional Cahn-Hilliard equation with a source term. AIMS Mathematics, 2022, 7(8): 14672-14695. doi: 10.3934/math.2022807
    [5] Fengfei Jin, Baoqiang Yan . Existence and global behavior of the solution to a parabolic equation with nonlocal diffusion. AIMS Mathematics, 2021, 6(5): 5292-5315. doi: 10.3934/math.2021313
    [6] Xiaolei Dong, Yuming Qin . Finite fractal dimension of pullback attractors for a nonclassical diffusion equation. AIMS Mathematics, 2022, 7(5): 8064-8079. doi: 10.3934/math.2022449
    [7] Jianbo Yuan, Shixuan Zhang, Yongqin Xie, Jiangwei Zhang . Exponential attractors for nonclassical diffusion equations with arbitrary polynomial growth nonlinearity. AIMS Mathematics, 2021, 6(11): 11778-11795. doi: 10.3934/math.2021684
    [8] Yang Wang, Jihui Wu . Long-time dynamics of nonlinear MGT-Fourier system. AIMS Mathematics, 2024, 9(4): 9152-9163. doi: 10.3934/math.2024445
    [9] Mengyang Liang, Zhong Bo Fang, Su-Cheol Yi . Blow-up analysis for a reaction-diffusion equation with gradient absorption terms. AIMS Mathematics, 2021, 6(12): 13774-13796. doi: 10.3934/math.2021800
    [10] Narcisse Batangouna . A robust family of exponential attractors for a time semi-discretization of the Ginzburg-Landau equation. AIMS Mathematics, 2022, 7(1): 1399-1415. doi: 10.3934/math.2022082
  • We consider the class of dissipative reaction-diffusion-convection systems on the circle and obtain conditions under which the final (at large times) phase dynamics of a system can be described by an ODE with Lipschitz vector field in RN. Precisely in this class, the first example of a parabolic problem of mathematical physics without the indicated property was recently constructed.



    The problem of describing the final (at large times) dynamics of dissipative semilinear parabolic equations (SPE)

    tu=G(u)  ()

    (see [1]) with a Hilbert phase space X by an ordinary differential equation (ODE) in RN has been attracting researcher's attraction for a long time. In fact, it is required to separate finitely many "determining" degrees of freedom of an infinite-dimensional dynamical system. In this case, the key geometric object is the so-called (global) attractor [2,3,4], i.e., the connected compact invariant set AX that uniformly attracts bounded subsets X as t+.

    The required ODE can sometimes be implemented as an inertial form [3,4,5] obtained by restricting the initial equation to an inertial manifold, i.e, a finite-dimensional invariant C1-surface MX containing the attractor and exponentially attracting (with asymptotic phase) all trajectories of () as t+. The theory of inertial manifolds originally encountered systematic difficulties, and several alternative concepts of finite-dimensional reduction of SPE have therefore been developed starting from [6,7,8,9]. Following [8], we will say that the dynamics of () on the attractor (final dynamics) is finite-dimensional if there exists an ODE in RN with Lipschitz vector field, resolving flow {Θt}tR, and invariant compact set KRN such that the phase semiflows {Φt}t0 of equation () on A and {Θt}t0 are Lipschitz conjugate on K. The latter means that there is a Lipschitz embedding g:ARN,gA=K, such that gΦtu=Θtgu for uA and t0. The existence of the inertial manifold implies that the dynamics is finite-dimensional on the attractor and, in general, looks like a more attractive property. Indeed, in the first case, the inertial form provides an exponential asymptotics of any solution of the equation at large times, and in the second case, we have an ODE reproducing the original dynamics only on the attractor itself. Nevertheless, the fact that the dynamics is finite-dimensional on A means that the structure of limit regimes of SPE with infinitely many degrees of freedom is no more complicated than the structure of similar regimes of an ODE with Lipschitz vector field in RN.

    An alternative approach to the problem of finite-dimensional reduction of SPEs with a parameter was developed by J. Hale, A. N. Carvalho and many others (see, for example, [10] and references therein). For diffusion equations, this approach is based on the concept of "large diffusion".

    In this paper, we consider the problem of whether the final dynamics is finite-dimensional for one-dimensional systems of reaction-diffusion-convection equations

    tu=Dxxuu+f(x,u)xu+g(x,u), (1.1)

    where u=(u1,,um) and f and g are sufficiently smooth matrix and vector functions. We assume that xJ, where J is a circle of length 1. The matrix of diffusion coefficients D is assumed to be diagonal, D=diag{dj},dj>0. As the phase space we choose an appropriate space XC1(J,Rm) in the Hilbert semiscale {Xα}α0 generated by a linear positive definite operator uuDuxx in X=L2(J,Rm). We postulate that evolution Eq (1.1) is dissipative in X and there exists the attractor AX consisting of functions u=u(x),uC1(J,Rm). The algebraic structure of the "convection matrix" f=f(x,u),f={fij}, i,j¯1,m, on the convex hull coAX plays an important role. We will highlight the case of the scalar diffusion matrix D=dE, where d=const and E is the identity matrix.

    For scalar equations of the form (1.1), the fact that the dynamics is finite-dimensional on the attractor was established in [9]. The problem of finite-dimensional reduction (1.1) becomes much more complicated when passing to the systems. In the vector case, the final dynamics of systems (1.1) with scalar diffusion matrix D and spatially homogeneous nonlinearity f(u)xu+g(u) was studied in [11], and the second restriction seems to be technical. The existence of an inertial manifold was proved in [11] for the scalar equation (m=1), and for m>1, it was proved under the assumption that the function matrix f(u) is diagonal with a unique nonzero element in a convex neighborhood of the attractor. The results obtained in [11] are based on a non-local change of the phase variable u which "decreases" the dependence of the nonlinear part (1.1) on xu and allows using the well-known "spectral gap condition". Note that in the absence of convection (f0), the existence of inertial manifold is a classical fact [4].

    Generalizing and developing the approach in [9], we study whether the dynamics is finite-dimensional on the attractor, but we do not consider the problem of existence of an inertial manifold for systems of periodic Eq (1.1). At the same time, we here consider the case of nonscalar diffusion matrix D and spatially nonhomogeneous nonlinearity with f=f(x,u),g=g(x,u). We prove that the limit dynamics is finite-dimensional for wide classes of systems (1.1). Now, omitting the details related to the choice of phase space and dissipativity conditions, we formulate the main results of the paper are as follows:

    The phase dynamics on the attractor of system (1.1) is finite-dimensional if any of the following three conditions are satisfied.

    (A) The convection matrix f=diag on coA (Theorem 4.3).

    (B) The diffusion matrix D is scalar. For all (x,u)J×coA, the numerical matrices f(x,u(x)) have m distinct real eigenvalues and commute with each other (Theorem 4.5).

    (C) The diffusion matrix D is scalar. For all (x,u)J×coA, the matrices f(x,u) are symmetric and commute with each other (Theorem 4.6).

    In the case (A), we have Df=fD on coA. The assumptions that the matrices are commutative can conditionally be formulated as the consistency of convection with diffusion and the self-consistency of convection on the convex hull of the attractor. Usually, the attractor A of system (1.1) can be localized in a ball BX centered at zero. Since the embedding XC(J,Rm) is continuous, it is actually sufficient to verify the conditions on f=f(x,u) in assertions (A), (B), and (C) for xJ, uRm:|u|<r with an appropriate r>0.

    In the class of one-dimensional systems (1.1), was constructed [11, Theorem 1.2] the first example of semilinear parabolic equation of mathematical physics (actually, a system of eight equations with scalar diffusion) that does not demonstrate any finite-dimensional dynamics on the attractor. This class seems to be a good testing ground for understanding where the finite-dimensional final dynamics of semilinear parabolic equations terminates and the infinite-dimensional final dynamics begins.

    Our results can be generalized to systems on the circle of the form

    tu=Dxxu+f(x,u,xu) (1.2)

    with a smooth vector function f=(f1,,fm). Such systems with various boundary conditions can be reduced (see [11,12]) to the form (1.1) by the termwise differentiation and an appropriate change of the variable. The fact that the final dynamics is finite-dimensional for scalar Eq (1.2) was already proved in [9].

    We here do not consider the Dirichlet and Neumann boundary conditions for systems of the form (1.1) on (0,1), this can be studied in a subsequent publication. The existence of an inertial manifold is proved in a similar situation in [12] for systems of general form (1.2) with f=f(u,ux) and a scalar diffusion matrix. Surprisingly, the dynamics in the Dirichlet-Neumann cases looks simpler than in the periodic case. The original methods of recent works [13,14] allowed to establish the finite-dimensionality of the final dynamics and the existence of an inertial manifold for 3D periodic Ginzburg-Landau equations.

    So, in this paper, the possibility of finite-dimensional reduction systems of 1D periodic reaction-diffusion-convection Eq (1.1) is established in terms of the algebraic properties of the diffusion matrix D, the convection matrix f, and the relations between them.

    The paper is organized as follows. Section 2 contains necessary information about abstract SPEs and the conditions for their final dynamics to be finite-dimensional. In Section 3, it is shown how these conditions can be applied to parabolic systems (1.1). The main results are obtained in Section 4. In the short Section 5, we present several examples of system (1.1) which admit a finite-dimensional final dynamics. Finally, in Section 6, we discuss alternative approaches to the problem of finite-dimensional reduction of systems (1.1).

    First, we consider the abstract dissipative SPE

    tu=Au+F(u) (2.1)

    in a real separable Hilbert space X with scalar product (,) and the norm . We assume that the unbounded self-adjoint positive definite linear operator A with domain of definition D(A)X has a compact resolvent. We assume that Xα=D(Aα) with α0. Then uα=Aαu, X0=X, and X1=D(A). For arbitrary Banach spaces Y1 and Y2, we let BCν(Y1,Y2), νN0, denote the class of Cν-smooth mappings Y1Y2 that are bounded on balls. We assume that the nonlinear function F belongs to BC2(Xα,X) for some α[0,1) and equation (2.1) is dissipative, i.e., generates a resolving semiflow {Φt}t0 in the phase space Xα and there exists a absorbing ball Ba={uXα:uα<a} such that ΦtBrBa for any ball Br:uα<r for t>t(r). In this case, the semiflow {Φt} inherits [1] the C2-smoothness, and there exists the compact attractor ABa consisting of all bounded complete trajectories {u(t)}tRXα and uniformly attracting balls Xα as t+. In fact, AX1 due to the smoothing action of the parabolic equation [1, Section 3.5].

    The embeddings XσXα with α<σ<1 are dense and compact, and uαcuσ, c=c(α,σ), for uXσ. Moreover, the proof of Theorem 3.3.6 in [1] can be used to derive the estimate Φ1uσL(r)uα on the balls BrXα. This implies that FBCν(Xσ,X) if FBCν(Xα,X) and the Xα-dissipativity implies the Xσ-dissipativity. Thus, in all constructions related to SPE (2.1), one can replace the nonlinearity index α with any value σ(α,1). The linear operator A:Xϑ+1Xϑ is positive definite in Xϑ with ϑ>0. If FBC2(Xϑ+α,Xϑ), then one can consider (2.1) in the pair of spaces (Xϑ,Xϑ+α) instead of (X,Xα). In this case, the phase dynamics preserves all its properties listed above.

    We say that the phase dynamics of (2.1) is asymptotically finite-dimensional if there exists an inertial manifold, i.e., a smooth finite-dimensional invariant surface MXα containing the attractor and exponentially attracting (with asymptotic phase) all solutions u(t) at large times. Such a manifold is usually [3,4,5] a Lipschitz graph over the lowest modes of the operator A. The restriction of SPE (2.1) to M is an ODE in RN,N=dimM which completely describes the final dynamics of the original evolution system.

    A less rigorous approach to the problem of finite-dimensional limit dynamic of SPE was proposed in [8,9]. So the dynamics of (2.1) on the attractor is finite-dimensional if, for some ODE tx=h(x) in RN with hLip(RN,RN) and resolving flow {Θt}tR, there exists an invariant compact set KRN such that the dynamical systems {Φt} on A and {Θt} on K are Lipschitz conjugate for t0. The properties of the dynamics to be asymptotically finite-dimensional and to be finite-dimensional on the attractor have not yet been separated; there is a hypothesis [5] that they are equivalent.

    Here are two criteria [8] for the dynamics to be finite-dimensional on the attractor under the assumption that FBC2(Xα,X).

    (Fl) The phase semiflow on A can be extended to the Lipschitz flow:

    ΦtuΦtvαMuvαeκ|t|,tR,

    where M>0 and κ0 depend only on A.

    (GrF) The attractor is a Lipschitz graph over the lowest Fourier modes:

    PuPvαMuvα,M=M(A),

    for some finite-dimensional spectral projection PL(Xα) of the operator A and all u,vA.

    Property (GrF) was established for scalar Eq (1.2) in [15] independently of the results obtained in [8,9]. We shall further use other sufficient conditions for the dynamics to be finite-dimensional on the attractor, which were obtained in [9] (by misunderstanding, an important assumption that X is real was not mentioned in [9]). Assume that G(u)=F(u)Au is the vector field of (2.1), N=A×AXα×Xα is a compact set, and Y is a Banach space.

    Definition 2.1 ([9]). A continuous field Π:NY is said to be regular if, for any u,vA, the function Π(Φtu,Φtv):[0,+)Y belongs to the class C1 and its derivative tΠ(u,v) at zero is bounded uniformly with respect to (u,v)N.

    The smoothness of the semiflow {Φt} and the invariancy of the compact set AXα imply the regularity of the identical embedding NXα×Xα and hence the regularity of any field Π:NY that can be continued to a C1-mapping into the (Xα×Xα)-neighborhood of the set N. In this situation, tΠ(u,v)=DΠ(u,v)(G(u),G(v)), where D is Frˊechet differentiation. The regular fields Π:NY form a linear structure which is also multiplicative if Y is a Banach algebra. In the last case, if the elements of Π(u,v)Y are invertible, then the field Π1 is also regular, and tΠ1=Π1(tΠ)Π1 for (u,v)N. We start from the decomposition

    G(u)G(v)=(T0(u,v)T(u,v))(uv),(u,v)N, (2.2)

    of the vector field G(u) on A, where T0L(Xα) and TL(X1,X) are unbounded linear sectorial operators in X similar to normal ones. We write

    Γa={zC:Rez=a},Γ(a,ξ)={zC:aξReza+ξ}

    for a>ξ>0 and assume that, for some c>0, θ[0,1], the total spectrum

    ΣT=u,vAspecT(u,v)

    is localized in the domain

    Ω(c,θ)={x+iyC:|y|<cxθ},x>0. (2.3)

    Let β=α/2 for 0θα/2, and let β=(α+θ)/3 for α/2<θ1. Assume that the set CΣT contains strips Γ(ak,ξk),kN, with ak,ξk+ as k.

    Theorem 2.2 (see [9, Theorem 2.8]). Assume that

    T(u,v)=S1(u,v)H(u,v)S(u,v) (2.4)

    on N, where the unbounded linear sectorial operators H(u,v) are normal in X, the fields S,S1:NL(X) and T0:NL(Xα,X) are regular, and the field T0:NL(Xα) is bounded. In this case, if

    aβk=o(ξk),k, (2.5)

    then the dynamics of Eq (2.1) is finite-dimensional on the attractor.

    Now we consider the system of Eq (1.1) on J=R|modZ with u=(u1,...,um). We assume that the matrix function f=f(x,u) and the vector function g=g(x,u) belong to the smoothness class C on J×Rm and write system (1.1) in the abstract form (2.1) with X=L2(J,Rm), positive definite operator Au=uDuxx, and nonlinearity F:uf(x,u)xu+g(x,u). Assume that {Xα}α0 is the Hilbert semiscale generated by A and Hs=Hs(J) are generalized Sobolev L2-spaces (spaces of Bessel potentials [1,16]) of scalar functions on J with arbitrary s0. If s>1/2, then HsC(J) and Hs is a Banach algebra [16, Section 2.8.3]. The differentiation operator x belongs to L(Hs+1,Hs). As the phase space we choose Xα=H2α(J,Rm) with arbitrary α(3/4,1) which is fixed below.

    We will generalize the conclusions of [9, pages 991 and 992] about the smoothness of the nonlinear function F and the phase dynamics of (1.1) to the case m>1. We let the symbol denote linear continuous embeddings of function spaces and shall use necessary results obtained in [1,16]. For an arbitrary C-function z:J×RmR, the mapping ψ:uz(x,u) is a function of class BCν from Cs(J) in Cs(J) for all ν,sN. Since H2αC1(J), we have ψBCν(H2α,C1(J)). Embedding theorems imply that ψBCν(Hs(J,Rm),Hs(J)). As we see, FBC2(X1,X1/2) and FBC1(X3/2,X1). Moreover, XαC1(J,Rm)C(J,Rm)X, and hence FBC3(Xα,X). We also note that X3/2C2(J,Rm) and X2C3(J,Rm).

    In the case of finite functions f=f(u),g=g(u), the dissipativity of system (1.1) with phase space X1/2, and hence also with Xα, 3/4<α<1, was proved in [11, Theorem 3.1]. This result can easily be transferred to the case of functions f(x,u) and g(x,u) that are finite in u and can also be generalized in other directions. Anyway, we further assume that system (1.1) is dissipative in Xα and there exists the global attractor AXα. Using the above-listed properties of nonlinearity F and following the reasoning in [9, page 992], we formulate the following remark.

    Remark 3.1 (see [9, Remark 5.2]). The following assertions hold: (a) The attractor A is bounded in X2; (b) If Y is a Banach space, then each vector field Π:NY continuous in the (Xα×Xα)-metric can be continued to C1-mapping X1×X1Y regularly in the sense of Definition 2.1.

    Our goal is to apply Theorem 2.2 to system (1.1) and to prove that the final dynamics is finite-dimensional. Let

    G(u)=Au+F(u)=Dxxuu+f(x,u)xu+g(x,u) (3.1)

    be the vector field of system (1.1), and let N=A×AXα×Xα. The main idea, as in [9], is related to the change of variable in the linear differential expression with respect to xJ for the difference G(u)G(v) for a fixed (u,v)N, which allows one to eliminate the dependence on xh, h(x)=u(x)v(x). Along with the convection matrix f={fij} we consider the m×m function matrices

    gu={giuj},fuxu={ml=1filujxul},i,j¯1,m.

    We put

    B0(x;u,v)=E+10(fu(x,w(x))wx(x)+gu(x,w(x))dτ, (3.2)

    where E is the unit m×m matrix, and

    B(x;u,v)=10f(x,w(x))dτ (3.3)

    for u,vXα, w(x)=τu(x)+(1τ)v(x), xJ. The elements of the matrices B0 and B are continuous functions, and for u,vA, function of class C2 on J. If necessary, it is convenient to treat expressions (3.2) as Bochner integrals ranging in some function spaces. By Mm we denote the algebra of numerical m×m matrices with Euclidean norm, and by Y(J,Mm) we denote the linear spaces of such matrices with elements from some Banach space Y of scalar functions on J. Using the C1-smoothness of the mappings (u,v)fu(x,w)wx+gu(x,w) and (u,v)f(x,w),Xα×XαC(J,Mm) for a fixed τ[0,1] and differentiating the expression under the integral sign in (3.2) with respect to the parameter (u,v), we conclude that the mappings (u,v)B0(;u,v) and (u,v)B(;u,v) are of class C1(Xα×Xα,C(J,Mm)). By the integral mean-value theorem for nonlinear operators, we have

    G(u)G(v)=Ah+(10DF(τu+(1τ)v)dτ)h
    =Dhxx+B0(x;u,v)h+B(x;u,v)hxRh,

    where h=uv, u,vA, and τu+(1τ)vcoA. Here D is the Frˊechet differentiation. Proceeding as in [17], we apply the transformation h=Uη to the differential expression Rh, where the m×m matrix function U(x)=U(x;u,v), x[0,1], is a solution of the linear Cauchy problem

    Ux=12D1B(x)U,U(0)=E. (3.4)

    Similar problems are considered in [18, Chapters 3 and 5]. We often write B0, B, and U omitting the dependence on u and v and sometimes on x. Taking into account the fact that

    Uxx=12D1(Bx(x)U+B(x)Ux)=12D1Bx(x)U+14D1B(x)D1B(x)U,

    we have

    Rh=RUη=D(Uηxx+2Uxηx+Uxxη)+B0(x)Uη+B(x)(Uxη+Uηx)
    =DUηxxB(x)Uηx12Bx(x)Uη+14B(x)D1B(x)Uη+B0(x)Uη+B(x)Uηx
    +B(x)(12D1B(x)Uη)=DUηxx+(B0(x)12Bx(x)14B(x)D1B(x))Uη.

    Now we write a decomposition of the form (2.2) for the vector field (3.1) of evolution Eq (1.1) on the attractor A with linear components

    T0(u,v)h=ωh+(B0(x)12Bx(x)14B(x)D1B(x))h, (3.5)
    T(u,v)h=ωhDUxxU1h, (3.6)

    where the numerical parameter ω>0 will be chosen later.

    We slightly generalize the fact that linear problem (3.3) can be solved explicitly under the condition that the operators D1B(x) are commutative in xJ.

    Lemma 3.2. Let D1B(x)=CW(x)C1 with constant nondegenerate matrix C and matrix function WC(J,Mm), and let W(x1)W(x2)=W(x2)W(x1) for x1,x2J. Then U(x)=C¯U(x)C1 with

    ¯U(x)=exp(12x0W(ξ)dξ) (3.7)

    is the solution of the Cauchy problem (3.3) on [0,1].

    Proof. Under the conditions of the lemma, we have ¯U(x)W(x)=W(x)¯U(x), and hence ¯Ux=12W(x)¯U. Further, ¯U(0)=U(0)=E and

    Ux=C¯UxC1=C(12W¯U)C1
    =C(12C1D1BCC1UC)C1=12D1BU.

    Now we prove regularity in the sense of Definition 2.1 of some vector fields on the compact set NXα×Xα. If YY1 for the function spaces Y and Y1, then the regularity of the field Π:NY implies the regularity of Π:NY1.

    Lemma 3.3. The field of operators T0 on N is bounded ranging in L(Xα) and regularly ranging in L(Xα,X).

    Proof. Let α,α and α,0 be norms of operators in L(Xα) and L(Xα,X). We assume that T0h=Q(x;u,v)h in (3.4.1) with hcoAXα. By Remark 3.1.(a), the convex hull of the attractor is bounded in the norm of X2 which is equivalent to the norm H4(J,Rm), and hence the matrix functions B, B0, and BD1B are uniformly bounded with respect to (u,v)N in H3(J,Mm). Thus, the matrix functions Bx and Q are bounded on N in the norm H2(J,Mm) and T0 is the operator of multiplication of vector functions in Xα=H2α(J,Rm) by the matrix QH2α(J,Mm) with 2α(3/2,2). Since H2α(J) is a Banach algebra, we see that T0(u,v)L(Xα) and T0(u,v)α,αconst on N.

    Since H2α(J)C(J)L2(J), we have H2α(J,Mm)C(J,Mm)L2(J,Mm) and T0α,0cQ0,0, where Q0,0 is the norm of Q as an operator in L(X) and c=c(A). Therefore, the field of operators T0:NL(Xα,X) is regular if the field of matrix functions Q:NL2(J,Mm) is regular. The function uf(x,u),XαC(J,Mm), is of class C1. Since the mappings (u,v)B0(;u,v) and (u,v)B(;u,v) are of class C1(Xα×Xα,C(J,Mm)), it follows that their restrictions to N are regular. The regularity of the field BD1B:NC(J,Mm) follows from the regularity of the fields B and D1=const with the multiplicative structure of C(J,Mm) taken into account. Moreover, the fields of matrix functions B, B0, BD1B on N are regular with values in L2(J,Mm).

    Now we prove the regularity of the field ΠBx:NL2(J,Mm). Let Πτ(u,v)=(f(x,w))x with w=τu(x)+(1τ)v(x) for a fixed τ[0,1] and arbitrary u=u(x),v=v(x)X1. Then Π(u,v)=(B(x;u(x),v(x)))x is the result of integration of Πτ(u,v) over τ. The mapping uf(x,u) belongs at least to the class BC1(X1,X1/2), and hence ΠτC1(X1×X1,L2(J,Mm)). Differentiating the integral expression for Π(u,v) with respect to the parameter (u,v)X1×X1, we obtain ΠC1(X1×X1,L2(J,Mm)). It remains to verify that the fields Π:NL2(J,Mm) are continuous and to use Remark 3.1.(b). By [9, Lemma 1.1], the function uAu, AX, with Au=uDuxx is continuous in the Xα-metric; the same holds for the mappings uuxx and uux of the set coAXα into X for ucoAX1. In the relation (f(x,u))x=fx+fuux, the operators ufx(x,u), ufu(x,u)ux continuously act from Xα to C(J,Mm), and hence Πτ,ΠC(N,L2(J,Mm)). The proof of the lemma is complete.

    Everywhere below, IId in a Banach space. The matrix functions B(x) and U(x) in the Cauchy problem (3.3) can be treated as bounded linear operators in X. The following assertion is related to the smooth dependence of solutions of differential equations on a parameter.

    Lemma 3.4. The field of operators U:NL(X) is regular.

    Proof. We consider (3.3) for arbitrary u,vXα as the non-autonomous evolution problem

    xU=12D1B(x;(u,v))U,U(0)=I

    in the Banach algebra L(X) with identically zero sectorial linear part and the parameter (u,v)Xα×Xα. The function

    (x,U,(u,v))12D1B(x;(u,v))U

    ranging in L(X) is Lipschitz in x, linear in UL(X) and of class C1 with respect to the parameter (u,v). Under these conditions, by [1, Theorem 3.4.4], the mapping (u,v)U(x;(u,v)),Xα×XαL(X) is continuously differentiable, and hence the operator field U:NL(X) is regular.

    Now we formulate an important condition on the diffusion matrix D and the convection matrix f of system (1.1).

    Assumption 3.5. Df(x,u)=f(x,u)D for xJ,ucoA.

    For the scalar diffusion matrix D=dE, this assumption is satisfied automatically. In the case of m distinct diffusion coefficients dj, Assumption 3.5 holds under the condition that the matrix f is diagonal on coA, and in the case of s distinct diffusion coefficients, 1<s<m, it holds under the condition that the matrix f on coA inherits the block structure (with respect to the same dj) of the matrix D=diag{dj}.

    Lemma 3.6. If Assumption 3.5 holds, then

    T(u,v)=U(u,v)(ωIDxx)U1(u,v)

    for u,vA.

    Proof. Assumption 3.5 implies (for any xJ and u,vA) that DB(x)=B(x)D for the matrices B(x)=B(x;u,v) in (3.2.2). Thus, the matrices B(x) and D1B(x) inherit the block structure (with respect to the same dj) of the diffusion matrix D=diag{d1,,dm}. Therefore, the same also holds for the solutions U(x) of problem (3.3), and hence DU(x)=U(x)D,x[0,1], and the assertion of the lemma follows from (3.4.2).

    The conditions for the dynamics to be finite-dimensional on the attractor will depend on the structure of the diffusion matrix D and the nonlinear function f in (1.1). By Theorem 2.2, we need to prove that the operators T(u,v) in (3.4.2) are "uniformly and regularly" similar, like (2.4), to the normal operators in X and to establish the required sparseness (2.5) of the total spectrum ΣT.

    We note that B(0)=B(1), Bx(0)=Bx(1) for the matrix function B(x)=B(x;u,v) in (3.2.2) defined on J×N. The matrix function V(x)=U1(x),x[0,1], is a solution (see [18, Section 3.1.3]) of the Cauchy problem adjoint to (3.3):

    Vx=12VD1B(x),V(0)=E, (4.1)

    and we have

    η=Vh,ηx=Vxh+Vhx,Vx=12VD1B,
    Vx(0)=12V(0)D1B(0),Vx(1)=12V(1)D1B(1),
    η(0)=h(0),η(1)=V(1)h(1),
    ηx(0)=12D1B(0)h(0)+hx(0),ηx(1)=12V(1)D1B(1)h(1)+V(1)hx(1).

    So the periodic boundary conditions h(1)=h(0),hx(1)=hx(0) become

    η(1)=V(1)η(0),ηx(1)=V(1)ηx(0), (4.2)

    where V(1)E in general. Further, we use the notation U1=U1(u,v),V1=V1(u,v) with (u,v)N for the operators U(1),V(1)L(Rm)=Mm. Operators U1 and V1 are monodromy operators [18, Section 5.1] Cauchy problems (3.3) and (4.1) respectively.

    The following assertion plays the key role.

    Lemma 4.1. If system (1.1) is dissipative in Xα with α(3/4,1), then the phase dynamics on the attractor is finite-dimensional in each of the following two cases.

    (i) The diffusion matrix D is scalar and, for all u,vA, the monodromy operators V1(u,v) are similar to self-adjoint positive definite ones with a fixed similarity matrix C=C(A).

    (ii) Assumption 3.5 holds and, for all u,vA, the monodromy operators V1(u,v) are similar to diagonal positive definite ones with a fixed similarity matrix C=C(A).

    Proof. By the conditions of the lemma, we have V1=C1VC for self-adjoint positive definite operators V=V(u,v) in Rm. For fixed u,vA, we let φjRm and μj>0 denote orthonormal eigenvectors and eigenvalues of the operator V with j¯1,m. We assume that H0=H0(u,v)=ωIDxx,D=diag{dj}, with boundary conditions (4.2) on (0,1) for some ω>0. We also assume that

    χk,j(x)=e2πkixφj,xJ,kZ,j¯1,m.

    Since Vφj=μjφj, we have V1C1φj=μjC1φj and, for the functions ψk,j(x)=μxjC1χk,j, ψk,j(0)=C1φj, ψk,j(1)=V1C1φj,

    (ψk,j)x(0)=(lnμj+2πki)C1φj,(ψk,j)x(1)=(lnμj+2πki)V1C1φj.

    As we see, ψk,j are eigenfunctions of the operator H0 with eigenvalues

    λk,j=ωdj(lnμj+2πki)2=ω+dj(2πkilnμj)2, (4.3)

    where djd>0 in case (i). The operators V1(u,v) continuously depend on (u,v)N, and hence this also holds for their spectrum. By the compactness of NXα×Xα, we have 0<c1μjc2, j¯1,m, for some c1(A),c2(A). Thus, the values |lnμj| are uniformly bounded in j¯1,m and u,vA. We put

    S0(x)=CVx1=VxC

    for xJ and H=S0H0S10. Then S0ψk,j=χk,j and

    Hχk,j=S0H0ψk,j=λk,jS0ψk,j=λk,jχk,j.

    Since the system of functions {χk,j} is complete and orthonormal in X=L2(J,Rm), it follows that the operators H=H(u,v) are normal in X for u,vA. Let S=S0U1(x)=S0V(x). We use Lemma 3.6 to write decomposition (2.4) of the vector field (3.1) on the attractor A with

    T(u,v)=UH0U1=US10HS0U1=S1(u,v)H(u,v)S(u,v)

    and operators T0(u,v) of the form (3.4.1). We see that S1=U(x)S10.

    By Lemma 3.4, the operator field U on N ranging in the Banach algebra L(X) is regular, and hence, the field of inverse operators V:NL(X) is regular. Since V1=V(1) and V=CV1C1, it follows that the operator field V:NMm is regular and tV(u,v)c3 for the derivative tV at zero for all (u,v)N (see Definition 2.1). Here is the Euclidean norm of matrices. Let b=2max and \delta = 1-c_{1}/b , then \delta\in (0, 1) . Since the spectrum \sigma(b^{-1}\mathcal{V}-E)\subset (-\delta, 0) and \|b^{-1}\mathcal{V}-E\| < \delta , it follows that the matrix representation

    \begin{align} \ln \mathcal{V} = \ln(bE)+\ln (E+b^{-1} \mathcal{V}-E) = \ln(bE) +\mathop{\sum}\limits_{n = 1}\limits^{\infty}\dfrac{(-1)^{n-1}}{n}(b^{-1} \mathcal{V}-E)^{n} \end{align} (4.4)

    converges uniformly on \mathcal{N} . By [19, Section 5.8, Exercise 3], we have

    \partial_{t}(b^{-1} \mathcal{V}-E)^{n} = \mathop{\sum}\limits_{i = 1}\limits^{n}(b^{-1} \mathcal{V}-E)^{i-1}\partial_{t}(b^{-1} \mathcal{V})(b^{-1}\mathcal{V}-E)^{n-i},

    and therefore, \|\partial_{t}(b^{-1}\mathcal{V}-E)^{n}\| < n\delta^{n-1} . If we differentiate (4.4) with respect to t , we obtain a uniformly converging series with the estimate

    \|\partial_{t}\ln \mathcal{V}(u, v)\| < \mathop{\sum}\limits_{n = 1}\limits^{\infty}\delta^{n-1} = \dfrac{1}{1-\delta}\, , \quad (u, v)\in \mathcal{N}.

    So the operator field \ln \mathcal{V}:\mathcal{N}\rightarrow \mathbb{M}^{m} is regular. We have \mathcal{V}^{-x} = \mathrm{exp}(-x\ln \mathcal{V}) and the standard decomposition of the matrix exponent guarantees that the field \mathcal{V}^{-x}:\mathcal{N}\rightarrow \mathcal{L}(X) is regular, which implies the regularity of fields of the operators S_{0}, S, S^{-1}:\mathcal{N}\rightarrow \mathcal{L}(X) .

    Finally, by Lemma 3.3, the field of operators T_{0} in (3.4.1) is regular on \mathcal{N} with values in \mathcal{L}(X^{\alpha}, X) and bounded with values in \mathcal{L}(X^{\alpha}) .

    Let \Sigma_{H} = \Sigma_{T} be the total spectrum of the field of operators H(u, v) on \mathcal{N} . Using (4.3), we choose a parameter \omega > 0 that ensures the inclusion \Sigma_{H} \subset \Omega (c, \theta) of the form (2.3) with \theta = 1/2 and an appropriate c > 0 . As we can see, operators H(u, v) are sectorial. Since 3/4 < \alpha < 1 , we have \beta = (\alpha +\theta)/3 < 1/2 . Moreover, from (4.3) we derive that the set {\mathbb C}\backslash\Sigma_{H} contains vertical strips \Gamma(a_{k}, \xi_{k}) with

    a_{k} \sim 4\pi ^{2} k^{2} , \; \; \; \xi _{k} \sim 4\pi ^{2}k

    and hence a_{k}^{\beta} = o(\xi_{k}) as k\rightarrow \infty . Thus, all conditions of Theorem 2.2 are satisfied and the dynamics of system (1.1) is finite-dimensional on the attractor.

    Remark 4.2. For all u, v\in \mathcal{A} , the monodromy operators U_{1} = U_{1}(u, v) and V_{1} = V_{1}(u, v) = U_{1}^{-1} are self-adjoint and positive definite if any of the following two conditions is satisfied:

    (a) The matrices D^{-1}B(x_{1}) and D^{-1}B(x_{2}) are symmetric and commutative for x_{1}, x_{2}\in J ;

    (b) (D^{-1}B(x))^{\mathrm{t}} = D^{-1}B(1-x) for all x\in J , where (\cdot)^{\mathrm{t}} is the operation of transposition.

    Under conditions (a), Lemma 3.2 holds with C = E and the matrix

    U_{1} = \mathrm{exp}(-\frac {1} {2}\int\limits_0^{1}D^{-1}B(x)dx\, )

    is symmetric and positive definite. The sufficiency of condition (b) was proved in [20, Proposition 2.3].

    Theorem 4.3. Assume that system (1.1) is dissipative in X^{\alpha} with \alpha \in (3/4, \, 1) and the convection matrix f is diagonal on \mathrm{co}\, \mathcal{A} . Then the phase dynamics is finite-dimensional on the attractor.

    Proof. Under the conditions of the theorem, the matrices B(\cdot; u, v) from (3.2.2), and hence (see the proof of Lemma 3.6), also the matrices U(\cdot; u, v), V(\cdot; u, v) , are diagonal on \mathcal{A}\times \mathcal{A} . According to Remark 4.2.(a) matrices U(\cdot; u, v), V (\cdot; u, v) are positive definite. The monodromy operators V_{1}(u, v) are also positive definite and are diagonal, so we can refer to Lemma 4.1.(ii) with C = E .

    Lemma 4.4. Assume that system (1.1) is dissipative in X^{\alpha} with \alpha \in (3/4, \, 1) and D = dE . Then the phase dynamics is finite-dimensional on the attractor if, for (x, u)\in {J\times\rm co\, \mathcal{A}} ,

    \begin{align} D^{-1}f(x, u(x)) = CH(x, u(x))C^{-1}, \end{align} (4.5)

    where the symmetric matrix functions H(x; u) commute with each other for any (x, u)\in J\times \mathrm{co}\, \mathcal{A} and C is a constant nondegenerate matrix.

    Proof. From (3.2.2) we derive that D^{-1}B(x) = CW(x)C^{-1} , where

    W(x) = W(x;u, v) = \int_{0}^{1}H(x;w(x))d\tau

    for u, v\in \mathcal{A} , w(x) = \tau u(x)+(1-\tau)v(x) , x\in J . By Lemma 3.2, the monodromy operator V_{1}(u, v) = U_{1}^{-1}(u, v) , u, v\in \mathcal{A} , satisfies the relation V_{1} = C(\overline{U}(1))^{-1}C^{-1} with operator \overline{U} given in formula (3.5). In this case, \overline{U}_{x} = -\dfrac{1}{2}W(x)\overline{U} and the matrices W(x; u, v) are symmetric and commutative on J for all u, v\in \mathcal{A} . By Remark 4.2.(a), the operator \overline{U}(1) is self-adjoint and positive definite and the assertion of the theorem follows from Lemma 4.1.(i).

    We shall give two more arguments ensuring that the final dynamics is finite-dimensional.

    Theorem 4.5. Assume that system (1.1) is dissipative in X^{\alpha} with \alpha \in (3/4, \, 1) and D = dE . Then the phase dynamics is finite-dimensional on the attractor if the following two conditions are satisfied:

    (i) The numerical matrices f(x, u(x)) have m distinct real eigenvalues for each (x, u)\in J\times \mathrm{co}\, \mathcal{A} ;

    (ii) The matrices f(x, u) commute with each other for any (x, u)\in J\times \mathrm{co}\, \mathcal{A} .

    Proof. Condition (ii) and assumption D = dE imply that the matrices D^{-1}f(x, u(x)) commute with each other on J\times\mathrm{co}\, {\mathcal{A}} . It is known [19, Theorem 8.6.1] that two simple (similar to diagonal) commutative m\times m matrices have a common set of m of linearly independent eigenvectors. By condition (i), all eigenvalues of each numerical matrix f(x, u(x)) with (x, u)\in J\times \mathrm{co}\, \mathcal{A} are real and distinct, and hence there exists a unique (up to permutations and multiplications by -1 ) common (for all these matrices) normalized basis \mathcal{E} = (e_{1}, \dots, e_{m}) of their eigenvectors in \mathbb{R}^{m} . By C we denote the constant matrix of transition from the canonical basis in \mathbb{R}^{m} to the basis \mathcal{E} , and by H(x) we denote diagonal (symmetric) matrices of linear operators D^{-1}f(x, u(x))\in \mathcal{L}(\mathbb{R}^{m}) in this basis. We see that relation (4.5) is satisfied and it remains to apply Lemma 4.4.

    Theorem 4.6. Assume that system (1.1) is dissipative in X^{\alpha} with \alpha \in (3/4, \, 1) and D = dE . Then the phase dynamics is finite-dimensional on the attractor if the matrices f(x, u) are symmetric and commute with each other for any (x, u)\in J\times \mathrm{co}\, \mathcal{A} .

    Proof. The conditions of the theorem guarantee that the numerical matrices

    D^{-1}f(x, u(x)) commute with each other on J\times \mathrm{co}\, \mathcal{A} . As in the proof of Lemma 4.4, from formula (3.2.2) for B(x) , we derive that the matrices D^{-1}B(x_{1}) and D^{-1}B(x_{2}) are symmetric and commutative for arbitrary x_{1}, x_{2}\in J . By Remark 4.2.(a), the monodromy operators V_{1}(u, v) are positive definite for any u, v\in \mathcal{A} and the assertion of the theorem follows from Lemma 4.1.(i) with C = E .

    In contrast to Theorem 4.5, we here admit the multiplicity of eigenvalues of the numerical matrices f(x, u(x)) , but we assume that these matrices are symmetric.

    We consider several examples illustrating the above-described theory in terms of properties of the convection matrix f . Here we restrict ourselves to the case of scalar diffusion and assume that system (1.1) is dissipative in the phase space X^{\alpha} with \alpha\in (3/4, 1) . We assume that all the conditions assumed below on f = f(x, u) are valid for x\in J and u = u(x) , u\in\mathrm{co}\, \mathcal{A} .

    Proposition 5.1. Assume that D = dE and f(x, u) = f_{1}(x, u)Q with a scalar C^{\infty} function f_{1} and numerical m\times m matrix Q . Then, the dynamics on the attractor of system (1.1) is finite-dimensional if any of the following two conditions is satisfied:

    (i) The matrix Q has m distinct real eigenvalues and f_{1}(x, u(x))\neq 0 for x\in J and u\in\mathrm{co}\, \mathcal{A} ;

    (ii) The matrix Q is symmetric.

    Proof. The numerical matrices f = f_{1}(x, u(x))Q are commutative. In the case of (i), each of these matrices has distinct real eigenvalues \lambda_{j}f_{1}(x, u(x)) , where \lambda_{1}, \dots, \lambda_{m} are eigenvalues of Q , and Theorem 4.5 can be applied. In the case of (ii), the fact that the dynamics is finite-dimensional on the attractor is a direct consequence of Theorem 4.6.

    Remark 5.2. Condition (i) in Proposition 5.1 is satisfied on Q for upper-triangular and lower-triangular matrices with distinct elements on the diagonal. For m = 2 and Q = \{q_{jl}\} , this condition precisely means that (q_{11}-q_{22})^{2}+4q_{12}q_{21} > 0 .

    Example 5.3. The dynamics on the attractor of system (1.1) is finite-dimensional in the case of m = 2 , D = dE and f(x, u) = \{f_{jl}(x, u)\} with f_{11} = f_{22} and f_{12} = f_{21} . This is a consequence of Theorem 4.6 and the commutativity of numerical matrices of the form \left(\begin{array}{cc} {a} & {b} \\ {b} & {a} \end{array}\right) .

    Example 5.4. Assume that D = dE , the matrix f = P_{n}(Q) , where P_{n} is a polynomial of degree n\geq 0 with coefficients a_{i} = a_{i}(x, u) , 0\leq i \leq n , a_{i}\in C^{\infty}(J\times \mathbb{R}^{m}, \mathbb{R}) , and the numerical matrix Q is symmetric. Then the dynamics of the attractor of system (1.1) is finite-dimensional. This is also a consequence of Theorem 4.6.

    Proposition 5.5. Assume that D = dE and f = Q(x) , where Q is a C^{\infty} function matrix. Then the dynamics of system (1.1) is finite-dimensional on the attractor if Q^{\, \mathrm{t}}(x) = Q(1-x) for x\in J .

    Proof. Since f = Q(x) , the matrix B(x) in (3.2.2) satisfies the condition (D^{-1}B(x))^{\mathrm{t}} = D^{-1}B(1-x) for all x\in J and u, v\in \mathcal{A} . By Remark 4.2.(b), the monodromy operators V_{1}(u, v) are positive definite for all u, v\in \mathcal{A} , and we can apply Lemma 4.1.(i) with C = E .

    The above presentation is based on Theorem 2.2, which means the verification of regularity (in the sense of Definition 2.1) of operator vector fields on the attractor. Alternatively, one can obtain a finite-dimensional reduction of one-dimensional parabolic systems by using the technique [5, Sections 2.3, 2.4 and 3.3] closely related to the results obtained in [21] about the dichotomies of non-autonomous parabolic equations. Here we will discuss the X^{1/2} -dissipative systems of general form (1.2). In our short description (on the sketch level), we omit technical details and refer to the criteria for the final dynamics to be finite-dimensional, i.e., criteria (Fl) and (GrF) in Section 2. The results of [4, Section 3.6] about inverse uniqueness of solutions of SPE (2.1) allow one to conclude that the phase semiflow on the attractor \mathcal{A} expands to the continuous flow \{\Phi_{t}\}_{t\in \mathbb{R}} . If u_{1}, u_{2}\in\mathcal{A} and h(t) = \Phi_{t}u_{1}-\Phi_{t}u_{2} for t\in\mathbb{R} , then

    \begin{align} h_{t} = Dh_{xx}+B_{0}(t, x)h+B(t, x)h_{x}, \end{align} (6.1)
    B_{0}(t, x) = \int_{0}^{1}f_{u} (x, w, w_{x})d\tau, \quad \; B(t, x) = \int_{0}^{1}f_{u_{x}}(x, w, w_{x})d\tau,

    w = \tau\Phi_{t}u_{1}+(1-\tau)\Phi_{t}u_{2} , with matrix functions B_{0} and B sufficiently smooth in (t, x)\in \mathbb{R}\times J and bounded in (u_{1}, u_{2}) . We assume that the invertible change v(t, x) = S(t, x)h(t, x) with operators S, S^{-1}\in \mathcal{L}(X) depending on u_{1}, u_{2} allows one to reduce (6.1) to the equation

    v_{t} = -Hv+R(t)v,

    where H = H(u_{1}, u_{2})\in\mathcal{L}(X^{1}, X) are normal (or uniformly with respect to (u_{1}, u_{2}) similar to normal) sectorial operators and R = R(\cdot\, ; u_{1}, u_{2}):\mathbb{R}\rightarrow \mathcal{L}(X) is a continuous operator function. Assume that the norms of the operators S , S^{-1} , and R are uniformly bounded in the parameter (u_{1}, u_{2}) . In this case, if the spectrum \Sigma_{H} combined over u_{1}, u_{2}\in \mathcal{A} is "sufficiently rare", then using the technique give in [5], one can verify that the phase flow is Lipschitzian on the attractor and then apply criterion (Fl). In contrast to the preceding presentation, we here have to deal with second-order linear differential expression in t\in \mathbb{R} and not in x\in J . The assertions of Section 4 can be obtained in this way after the change v(t, x) = V^{-x}(t, 1)V(t, x)h(t, x) , where V(\cdot, x) is a solution of the Cauchy problem (4.1).

    Another possible approach to the problem of finite-dimension of the final dynamics is related to the verification of criterion (GrF). In [22], a scalar parabolic equation of the form (1.2) is considered in a rectangle with Dirichlet boundary condition. The authors present conditions under which the attractor is a Lipschitz graph over finitely many first modes of the Laplace operator. And they use the cone condition well-known in the literature [3,4,5]. In this connection, it seems to be very perspective to study the problem of finite-dimensional reduction of systems of Eq (1.2) on the two-dimensional torus \mathbb{T}^{2} .

    We show that for new wide classes of 1D periodic systems of reaction-diffusion-convection Eq (1.1) the final (for large values of time) phase dynamics is finite-dimensional. For this, we develop in a nontrivial way the general methodology of [9]. The difficulties associated with the transition from a scalar equation to a system of equations have been successfully overcome. Alternative approaches to the problem of finite-dimensional reduction for more general systems (1.2) are also discussed.

    The authors declare no conflict of interest.



    [1] D. Henry, Geometric Theory of Semilinear Parabolic Equations, Springer, 1981.
    [2] A. V. Babin, M. I. Vishik, Attractors of Evolution Equations, North-Holland, 1992.
    [3] J. C. Robinson, Infinite-Dimensional Dynamical Systems, Cambridge University Press, 2001.
    [4] R. Temam, Infinite-Dimensional Dynamical Systems in Mechanics and Physics, Springer, 1997.
    [5] S. Zelik, Inertial manifolds and finite-dimensional reduction for dissipative PDEs, P. Roy. Soc. Edinb. A, 144 (2014), 1245-1327. doi: 10.1017/S0308210513000073
    [6] A. Eden, C. Foias, B. Nicolaenko, R. Temam, Exponential Attractors for Dissipative Evolution Equations, Wiley, 1994.
    [7] J. C. Robinson, Global attractors: Topology and finite-dimensional dynamics, J. Dyn. Differ. Equ., 11 (1999), 557-581. doi: 10.1023/A:1021918004832
    [8] A. V. Romanov, Finite-dimensional limit dynamics of dissipative parabolic equations, Sb. Math., 191 (2000), 415-429. doi: 10.1070/SM2000v191n03ABEH000466
    [9] A. V. Romanov, Finite-dimensionality of dynamics on an attractor for non-linear parabolic equations, Izv. Math., 65 (2001), 977-1001. doi: 10.1070/IM2001v065n05ABEH000359
    [10] A. N. Carvalho, J. W. Cholewa, G. Lozada-Cruz, M. R. T. Primo, Reduction of infinite dimensional systems to finite dimensions: Compact convergence approach, SIAM J. Math. Anal., 45 (2013), 600-638. doi: 10.1137/10080734X
    [11] A. Kostianko, S. Zelik, Inertial manifolds for 1D reaction-diffusion-advection systems. Part II: Periodic boundary conditions, Comm. Pure Appl. Anal., 17 (2018), 285-317. doi: 10.3934/cpaa.2018017
    [12] A. Kostianko, S. Zelik, Inertial manifolds for 1D reaction-diffusion-advection systems. Part I: Dirichlet and Neumann boundary conditions, Comm. Pure Appl. Anal., 16 (2017), 2357-2376.
    [13] A. Kostianko, Bi-Lipschitz Man\acute{\mathrm{e}} projectors and finite-dimensional reduction for complex Ginzburg-Landau equation, Proc. Roy. Soc. A, 476 (2020), 20200144. doi: 10.1098/rspa.2020.0144
    [14] A. Kostianko, C. Sun, S. Zelik, Inertial manifolds for 3D complex Ginzburg-Landau equations with periodic boundary conditions, 2021, arXiv: 2106.10538.
    [15] I. Kukavica, Fourier parametrization of attractors for dissipative equations in one space dimension, J. Dyn. Differ. Equ., 15 (2003), 473-484. doi: 10.1023/B:JODY.0000009744.13730.01
    [16] H. Triebel, Theory of Function Spaces, Birkhauser Verlag, 1983.
    [17] D. A. Kamaev, Families of stable manifolds of invariant sets of systems of parabolic equations, Russ. Math. Surv., 47 (1992), 185-186. doi: 10.1070/RM1992v047n05ABEH000950
    [18] J. L. Daleckii, M. G. Krein, Stability of Solutions of Differential Equations in Banach Space, Amer. Math. Soc., 1974.
    [19] P. Lankaster, Theory of Matrices, Academic Press, 1969.
    [20] V. I. Yudovich, Periodic differential equations with self-adjoint monodromy operator, Sb. Math., 192 (2001), 455-478. doi: 10.1070/SM2001v192n03ABEH000554
    [21] A. Yu. Goritskii, V. V. Chepyzhov, The dichotomy property of solutions of quasilinear equations in problems on inertial manifolds, Sb. Math., 196 (2005), 485-511. doi: 10.1070/SM2005v196n04ABEH000889
    [22] M. Kwak, B. Lkhagvasuren, The cone property for a class of parabolic equations, J. Korean Soc. Ind. Appl. Math., 21 (2017), 81-87.
  • This article has been cited by:

    1. A. V. Romanov, Finite-Dimensional Reduction of Systems of Nonlinear Diffusion Equations, 2023, 113, 0001-4346, 267, 10.1134/S0001434623010297
    2. Александр Владимирович Романов, Alexander Vladimirovich Romanov, Конечномерная редукция систем нелинейных диффузионных уравнений, 2023, 113, 0025-567X, 265, 10.4213/mzm13616
  • Reader Comments
  • © 2021 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(1927) PDF downloads(73) Cited by(2)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog