Volume 2022, Issue 1 2034264
Research Article
Open Access

The Long-Time Behavior of 2D Nonautonomous g-Navier-Stokes Equations with Weak Dampness and Time Delay

Xiaoxia Wang

Corresponding Author

Xiaoxia Wang

College of Mathematics and Computer Science, Yan’an University, Yan’an, 716000 Shaanxi, China yau.edu.cn

Search for more papers by this author
Jinping Jiang

Jinping Jiang

College of Mathematics and Computer Science, Yan’an University, Yan’an, 716000 Shaanxi, China yau.edu.cn

Search for more papers by this author
First published: 25 July 2022
Citations: 2
Academic Editor: Baowei Feng

Abstract

In this paper, we discuss the long-time behavior of g-Navier-Stokes equations with weak dampnesss and time delay. The uniformly attracting sets of processes are obtained. On the basis of the method with asymptotic compactness, the existence of the uniform attractor for the equation is proved with the restriction of the forcing term belonging to translational compacted function space.

1. Introduction

The understanding of the behavior with dynamical systems was one of the most important problems of modern mathematical physics (see [117]). In the last decades, g-Navier-Stokes equations have received increasing attention due to their importance in the fluid motion. In [24], the existence of weak solution and strong solution for the 2D g-Navier-Stokes equation on some bounded domain was studied. The Hausdorff and fractal dimension of the global attractor about the 2D g-Navier-Stokes equation for the periodic and Dirichlet boundary conditions and the global attractor of the 2D g-Navier-Stokes equation on some unbounded domains were researched in [5]. In [610], the finite dimensional global attractor and the pullback attractor for g-Navier-Stokes equation were studied. Moreover, Anh et al. studied long-time behavior for 2D nonautonomous g-Navier-Stokes equations and the stability of solutions to stochastic 2D g-Navier-Stokes equation with finite delays in [11, 12]; Quyee researched the stationary solutions to 2D g-Navier-Stokes equation and pullback attractor for 2D g-Navier-Stokes equation with infinite delays in [13]. Recently, the random attractors for the 2D stochastic g-Navier-Stokes equation were researched in [14]. From these researches, we can see that the attractor of 2D g-Navier-Stokes equation is still important. We would like to use the theory of uniform attractors to study it. So, the present research is necessary and has a theoretical basis.

In this paper, we study the existence of the uniform attractor of the g-Navier-Stokes equation with weak dampness and time delay which have the following form:
(1)
where u(t, x) ∈ R2 and p(t, x) ∈ R denote the velocity and pressure, respectively. ν > 0 is the viscosity coefficient, αu denotes linear dampness, and α > 0 is positive constant. is the time-dependent external force term, h(t, ut) is another external force term with time delay. 0 < m0g = g(x1, x2) ≤ M0 and g = g(x1, x2) are suitable real-valued smooth functions; when g = 1, Equation (1) becomes the usual 2D Navier-Stokes equations.

This paper is organized as follows. In Section 2, we first introduce some notations and preliminary results for the g-Navier-Stokes equation. In Section 3, we prove existence of the uniform attractor of 2D g-Navier-Stokes equation with weak dampness and time delay on the bounded domains.

2. Preliminaries

We assume that the Poincare inequality holds on Ω, i.e., there exists λ1 > 0, such that
(2)

Let with the inner products (u, v) = ∫Ωu · νgdx and the norms |·| = (·,·)1/2, u, vL2(g). Let , which is endowed with the inner products and the norms ‖·‖ = ((·,·))1/2, where

Let D(Ω) be the space of function with the compact support contained in Ω, and let = {v ∈ (D(Ω))2:∇ · gv = 0 on Ω}; the closure of in L2(g) is Hg; the closure of in is Vg. Hg has the inner product and norm of L2(g), And Vg has the inner product and norm of .

It follows from (2) that
(3)

We define a g-Laplacian operator as follows: −Δgu = −(1/g)(∇·g∇)u = −Δu − (1/g)∇g · ∇u.

Using the g-Laplacian operator, we rewrite the first Equation (1) as follows:
(4)

From [2], we can define a g-orthogonal projection Pg : L2(g)⟶Hg and a g-Stokes operator Agu = −Pg((1/g)(∇·(gu))).

Applying the projection Pg into (4), we can obtain the following weak formulation of (1): let fVg and u0Hg, we find that
(5)
such that ∀vVg, ∀t > 0.
(6)
(7)
where bg : Vg × Vg × Vg⟶R is given by
(8)
and Ru = Pg[(1/g)(∇g · ∇)u], such that (Ru, v) = b(∇g/g, u, v), ∀u, vVg. Then, the weak formulation of (6) and (7) is equivalent to the functional equations
(9)
(10)
where Ag : VgVg is the g-Stokes operator defined by 〈Agu, v〉 = ((u, v)), ∀u, vVg.B(u) = B(u, u) = Pg(u · ∇)u is bilinear operator and B(u, v), w〉 = bg(u, v, w), ∀u, v, wVg, where B and R satisfy the following inequalities [2, 4]:
(11)

Let T > τ, . For every t ∈ (τ, T), we define ut(s) = u(t + s), s ∈ (−h, 0). For convenience, we denote .

Let satisfy the following assumptions:
  • (I)

    is measureable,

  • (II)

    t ∈ R, h(t, 0) = 0,

  • (III)

    Lg > 0, such that , there is

  • (IV)

    m0 ≥ 0, Cg > 0, ∀m ∈ [0, m0], τt, u, vC0([τr, t]; Hg), such that

(12)
t ∈ [τ, T], ∀u, vL2(τr, T; Hg), from (IV), we have
(13)

Definition 1. Let and satisfy the hypotheses (I)-(IV). For every τ ∈ R, a function uL2(τ, T; Vg)∩L(τ, T; Hg), ∀T > τ is called a weak solution of problem (1) if it fulfils

(14)

We can obtain the following theorem by the standard Faedo-Galerkin methods, where we let T > τ > 0. Other cases can be similarly proved.

Theorem 2. Let satisfies the assumptions (I)-(IV), there exists a unique solution

(15)

such that (6) and (7) holds.

Proof. We apply the Faedo-Galerkin methods. Since Vg is separable and is dense in Vg, there exists a sequence , which forms a complete orthonormal system in Hg and a basic for Vg. Let m be a positive integer, for each m, we define an approximate solution um of (6) as , which satisfies

(16)
for t ∈ [0, T], j = 1, ⋯, m and um(0) = u0m, where u0m is the orthegonal projection in Hg of u0 onto the space spanned by w1, ⋯, wm. Then, we can obtain
(17)

We can write the differential equations in the usual form

(18)
where αij, αijk, βij ∈ R.

Let ϕim(0) be the ith component of u0m. The nonlinear ordinary differential system (18) has a maximal solution defined on some interval [0, tm]. If tm < T, then |um(t)|⟶∞ as ttm. The following we will prove tm = T. We need several estimates to do.

We multiply (16) by ϕjm(t) and add these equations for j = 1, ⋯, m to obtain

(19)

Then, we have

(20)

so that

(21)

Let , then

(22)

By the Gronwall inequality, we have

(23)

Hence,

(24)
which implies that the sequence um remains in bounded set of L(0, T; Hg). From (22), we have
(25)

Then,

(26)

We intergrate (26) from 0 to T; we have

(27)

So, the um remains in a bounded set of L2(0, T; Vg).

Let denotes the function from R into Vg, which is equal to um on [0, T] and to 0 on the complement of this interval. The Fourier transform of is denoted by . Then, we will show that there exist a positive constant C and γ such that

(28)

Since the um remains in a bounded set of L2(0, T; Vg), the remains in a bounded set of Hγ(R; Vg, Hg). Since has two discontinuities at 0 and T, the distribute derivative of is given by

(29)
where δ0 and δT are the dirac distributions at 0 and T, and is the derivative of um on [0, T]. We obtain that
(30)
for j = 1, ⋯, m, where δ0 and δT are distributions at 0 and T, fm = f + hνAumbumνRumαum and on [0, T]. By the Fourier transform, we have
(31)
where and denoting the Fourier transforms of and , respectively. We multiply (31) by and add the resulting equations for j = 1, ⋯, m; we get
(32)

We obtain

(33)

So, fm(t) belongs to a bounded set in the space . For ∀m, we have Since um(0)| and |um(T)| are bounded, from (31), we obtain

(34)

Let γ < (1/4), we have

(35)
then
(36)

Since umL2(0, T; Vg), by the Parseval equality and by the Schwarz inequality and the Parseval equality, we obtain

(37)

So, umHγ(R; Vg, Hg), and um remains in a bounded set of L(0, T; Hg), L2(0, T; Vg) and Hγ(R; Vg, Hg). There exists an element uL2(0, T; Vg)∩L(0, T; Hg) and a subsequence such that in L2(0, T; Vg) weakly and in L(0, T; Vg) weak-star as m⟶∞. For any , we have strongly in .

For any support of wj, we have strongly in . Let ψ be a continuously differentiable function on [0, T] with ψ(T) = 0, we multiply (16) by ψ(t), then integrate by parts,

(38)

We have

(39)
where ∀vVg.

Finally, we prove that u satisfies (7). We multiply (6) by ψ and integrate

(40)

We compare (39) with (40) to obtain (u(0) − u0, v)ψ(0) = 0. Let ψ(0) = 1, then we have (u(0) − u0, v) = 0, ∀vVg. So, u(0) = u0.

Now, we will prove the solution of (6) and (7) is unique. We let u1 and u2 be the solutions of (9) and u = u1u2. We have

(41)
(42)

We take the scalar product of (41) with u(t), then

(43)

Therefore,

(44)

Then,

(45)

We have

(46)

Hence, |u(t)|2 = 0, ∀t ∈ [0, T]. So, u1 = u2.

From [15], we can define a family of two parametric maps {Uf(t, τ)} = {Uf(t, τ) | tτ, τ ∈ R} in Hg,

(47)

Here, is called the time symbol of the system. We have the following concepts and conclusions from [15].

Definition 3. For the given time symbol , a family of two-parametric maps {U(t, τ)} with tτ ≥ 0 is called a process in Hg, if

(48)

Now, we define translation operator in . .

(49)

We have

(50)

Denote Σ = {T(h)f(x, s) = f(x, s + h), ∀h ∈ R}, where T(·) is the positive invariant semigroups acting on Σ and satifying T(h)ΣΣ, ∀h ≥ 0 and

(51)

Let be a constant, obviously

(52)

Let E be the Banach space; we use to denote the set of all bounded sets on E and consider a family of processes {Uf(t, τ)} with fΣ, the parameter f is called the symbols of the process family {Uf(t, τ)}, Σ is called the symbol space, and we assume that Σ is a complete metric space.

Definition 4. A family of processes {Uf(t, τ)}, fΣ is called uniformly bounded (w.r.t.fΣ), if any , both

(53)

Definition 5. A set B0E is said to be uniformly absorbing for the family of processes {Uf(t, τ)}, fΣ}, if for any τR and each , there exists t0 = t0(τ, B) ≥ τ, such that for all tt0,

(54)

Definition 6. A set PE is said uniformly atttracting set of {Uf(t, τ)}, fΣ}, if for any τR, there is

(55)

A family of processes {Uf(t, τ)}, fΣ} is said to uniformly compact, if there exists a compacted uniformly absorbed set in {Uf(t, τ)}, fΣ}. A family of processes {Uf(t, τ)}, fΣ} is said to uniformly asymptotic compact, if there exists a compacted uniformly atttracting set in {Uf(t, τ)}, fΣ}.

Definition 7. A closed set is said to be the uniform attractor of the family of processes {Uf(t, τ)}, fΣ}, if

  • (1)

    is uniformly attractive

  • (2)

    is included in any uniformly attracting set of {Uf(t, τ)}, fΣ}, that is

Theorem 8. Let {fγ(θ): γΓ} ⊂ C = C([−r, 0]; X) be equicontinuous and for any ∀θ ∈ [−r, 0], {fγ(θ): γΓ} is quasicompact in X, then {fγ(θ): γΓ} is relatively compact in C([−r, 0]; X).

Lemma 9 (Uniform Gronwall lemma). Let g, h, y be local integrable function on (t0, ∞), y is also local integrable on (t0, ∞), and y(t) ≤ g(t)y(t) + h(t), ∀tt0., where r, a1, a2, a3 is positive constant. Then,

(56)

3. The Existence of Uniform Attractor for 2D g-Navier-Stokes Equations in Bounded Domain

First, we prove the existence of uniformly absorbing set in and ; we define u(·) = u(·; τ, (u0, ϕ), f), where f is translation compact function. That is,
(57)

The following we use to represent the translation compact function class.

Lemma 10. Let for any τt, , assume that (I)-(IV) hold, then there exist bounded absorbing sets of process family {Uf(t, τ): tτ} in .

Proof. Since is bounded, then there exists , such that

(58)

For any , we define u(·) = u(·; τ, (u0, ϕ)), then taking the inner product of (9) with u(t), we have

(59)

Let , then

(60)

Integrating both sides from τ to t, then

(61)

Then,

(62)
for
(63)

Let sτ = θ, then

(64)

Taking m ∈ (0, m0), such that m + σ + Cg − 2νλ1β < 1, then , so

(65)

Let tτ + h, ∀θ ∈ [−h, 0], then

(66)

Then,

(67)

Let , we will prove the existence of the uniformly absorbing bounded set in . First, we must prove the boundedness of .

Lemma 11. Given that , then there exist and constant , such that

(68)
where denotes any bounded set on the .

Proof. Taking the inner product of (9) with u(t),

(69)

Then,

(70)

Integrating on both sides in [t, t + 1], we have

(71)

Then,

(72)

When , that is , we have

(73)
where
(74)

Lemma 12. For any τt, . Assume that (I)-(IV) hold, then there exists uniformly bounded absorbing set of process family {Uf(t, τ): tτ} in .

Proof. Let , taking the inner product of (9) with Agu, we obtain

(75)
for
(76)

Then,

(77)

Applying Lemma 9,

(78)
where If taking , then
(79)

Let u(·) = u(·; ts, (u0, ϕ)), so . Then,

(80)

From [16], we have the following definition.

Definition 13. Let E be Banach space, if ∀ε > 0, there exists η > 0, such that

(81)

Then, is called normal function.

We will take the sets of all normal function classes in as . From [17], we can see that is the true subspace of . Therefore, the translation compact function must be a normal function.

Theorem 14. Suppose that nonlinear term h satisfies (I)-(III), f is translation compact function in , then process family {Uf(·, ·) | fΣ} exist uniform attractor , and .

Proof. Since B2 is bounded set in and uniform absorbed set of {Uf(·, ·) | fΣ}. For each τ ∈ R, we take a set

(82)
where j denotes any compact self-adjoint operator, then B3B2B1, and B3 is another uniform absorbing bounded set of {Uf(·, ·) | fΣ} in . Now, we will prove B3 is relatively compact in . From Theorem 8, we only need to prove B3 is equicontinuous and uniform bounded in . From the definition of B3, we can obtain it is uniformly bounded. Now, we will prove B3 is equicontinuous. For any θ1, θ2 ∈ [−r, 0], ϕB2, fΣ,
(83)

Let θ2 > θ1, and denote u(·) = u(·; τ, j(ϕ), f) as u(·), then

(84)

We estimate the items on the right end of the above formula, let θ1θ2,

(85)

Since

(86)
we let

When , and θ2 > θ1, then

(87)

Let

(88)

Then,

(89)

So,

(90)

And When θ1θ2, ∀ϕB2, fΣ, we have

(91)

Then, B3 is equicontinuous, and B3 is relatively compact in , so is compacted uniformly absorbing set of {Uf(·, ·) | fΣ} in , Let , since and the embedding mapping is continuous, so is compact in ; is also compacted uniformly absorbing set of {Uf(·, ·) | fΣ} in . Then, process family {Uf(·, ·) | fΣ} exists uniform attractor

Conflicts of Interest

This work does not have any conflicts of interest.

Acknowledgments

The author would like to thank the referees for the helpful suggestions. This work is supported by the Projects of Natural Science Basic Research Plan in Shaanxi Province of China (no. 2018JM1042).

    Data Availability

    The (data type) data used to support the findings of this study are available from the corresponding author upon request.

      The full text of this article hosted at iucr.org is unavailable due to technical difficulties.