Open AccessArticle

Global Well-Posedness and Determining Nodes of Non-Autonomous Navier–Stokes Equations with Infinite Delay on Bounded Domains

Huanzhi Ge

^1,2

and

Feng Du

^2,*

School of Information and Mathematics, Yangtze University, Jingzhou 434023, China

School of Mathematics and Physics Science, Jingchu University of Technology, Jingmen 448000, China

Author to whom correspondence should be addressed.

Mathematics 2025, 13(2), 222; https://doi.org/10.3390/math13020222

Submission received: 22 December 2024 / Revised: 7 January 2025 / Accepted: 8 January 2025 / Published: 10 January 2025

(This article belongs to the Special Issue Advances in Study of Time-Delay Systems and Their Applications, 2nd Edition)

Download Versions Notes

Abstract

The asymptotic behavior of solutions to nonlinear partial differential equations is an important tool for studying their long-term behavior. However, when studying the asymptotic behavior of solutions to nonlinear partial differential equations with delay, the delay factor

u (t + θ)

in the delay term may lead to oscillations, hysteresis effects, and other phenomena in the solution, which increases the difficulty of studying the well-posedness and asymptotic behavior of the solution. This study investigates the global well-posedness and asymptotic behavior of solutions to the non-autonomous Navier–Stokes equations incorporating infinite delays. To establish global well-posedness, we first construct several suitable function spaces and then prove them using the Galekin approximation method. Then, by accurately estimating the number of determining nodes, we reveal the asymptotic behavior of the solution. The results indicate that the long-term behavior of a strong solution can be determined by its values at a finite number of nodes.

Keywords:

Navier–Stokes equations; infinite delay; global well-posedness; asymptotic behavior; determining nodes; long-time behavior

MSC:

35B40; 35B41; 37G35; 35Q35

1. Introduction

In this paper, we study the non-autonomous Navier–Stokes model with infinite delay in 2D bounded domains, which is given by the following equations:

\begin{matrix} \{\begin{matrix} \frac{\partial u}{\partial t} - ν Δ u + (u \cdot \nabla) u + \nabla p = f (t) + G (t, u (t + θ)) in (x, t) \in Ω \times (τ, T), \\ div u = 0 in (x, t) \in Ω \times (τ, T), \\ u = 0 on (x, t) \in \partial Ω \times (τ, T), \\ u (x, τ + θ) = ϕ (x, θ) x \in Ω θ \in (- \infty, 0], \end{matrix} \end{matrix}

(1)

where the unknown functions

u (x, t) = (u^{(1)} (x, t), u^{(2)} (x, t))

and

p = p (x, t)

represent the velocity field and the pressure of the fluid motion, respectively. The function

f (t) = (f^{(1)} (t), f^{(2)} (t))

is the external force, which is a given vector function that varies with time t. The parameter

ν > 0

denotes the viscosity coefficient. The function

u (t + θ)

represents the history of the state

u (\cdot)

at time t, where

θ \in (- \infty, 0]

. The function

G (\cdot, \cdot)

is a known continuous function, as required by the study.

In addition,

Ω \subset R^{2}

is an open set with smooth boundaries, and the function

ϕ (x, θ)

is the initial value of Equation (1) in the delay time

(- \infty, 0]

. These assumptions typically serve as the basis for ensuring the well-posedness of the solution to Equation (1) in

Ω

. They also imply that the boundary has well-defined tangent vectors at each point and that the shape of the boundary does not exhibit sharp points or discontinuous regions. The assumption of regularity in the initial data is to satisfy the requirements of adaptability, regularity, and long-term behavior of the solution to Equation (1) in

Ω

The Navier–Stokes equations describe the motion of fluids, involving multiple factors such as velocity, pressure, and viscosity. Introducing delay factors allows for a more realistic simulation of inertia and time-delay effects in fluid flow, including interactions at fluid–solid interfaces and the impact of temperature variations on fluid states. In many physical phenomena, the response of fluids is not immediate but involves a time delay, which leads to the influence of time delay on the motion state of fluids and indirectly affects the stability of the system. Therefore, studying the Navier–Stokes equations with delay provides insights into flow stability across varying parameters and elucidates how delays influence flow transitions and turbulence onset. Incorporating delay factors enhances model accuracy, improving the prediction and understanding of fluid behavior. The study of Navier–Stokes equations with delay is not limited to traditional fluid mechanics but also involves multiple fields such as control theory, nonlinear dynamics, and mathematical physics, promoting interdisciplinary research and communication.

Owing to their wide applications, the Navier–Stokes equations have been extensively studied by mathematicians and physicists. In particular, the well-posedness and regularity of solutions are the most significant concerns. Several papers have been dedicated to investigating the well-posedness of solutions, including references [1,2,3,4]. Foias and Temam, among others, introduced the concepts of definite modules, degrees of freedom, and determining nodes of the system in [5]. These concepts describe how the asymptotic behavior of the system can be determined by a finite number of quantities, revealing the internal structural properties of the attractor. The term “determining nodes” refers to the process of numerically solving partial differential equations, where the solutions at certain specific grid points or nodes are uniquely determined by specific boundary conditions, initial conditions, or solutions at other nodes. In other words, the values of these nodes do not depend on the values of other unknown nodes but are directly calculated based on given conditions, thereby reducing uncertainty in the solving process. This concept is further elaborated in references [6,7,8,9,10,11,12,13,14,15]. Kakizawa used the energy method to prove the asymptotic behavior of strong solutions to the initial-boundary value problem for general semilinear parabolic equations in [12]. Additional related research can be found in references [6,7,8,9,11,14,16,17].

The phenomenon of time delay is ubiquitous and closely related to real life. For example, when we aim to control or change the original state of a system by applying external forces, we must consider not only the current state of the system but also the impact of its previous states, i.e., the lag factor. Therefore, time delay effects are often incorporated into models when dealing with practical problems. Examples include the delayed Navier–Stokes model (see [16,18]) and wave equations with time delay (see [19]). Hernandez and Wu studied the well-posedness and global attractors of abstract Cauchy problems with state-dependent delay and provided examples of PDEs with state-dependent delay in [20], which offers important insights into the application of state-dependent delay in specific PDEs and the study of their well-posedness and asymptotic solutions. Other related results can be found in references [21,22]. Regarding the Navier–Stokes equations, García-Luengo, Marín-Rubio, and Real investigated the asymptotic behavior of the Navier–Stokes model with finite delay in [8,14], while Chueshov studied the finite-dimensional global attractors for parabolic nonlinear equations with state-dependent delay, using the Galerkin method to prove global well-posedness in [21]. For the study of infinite delay, Fu and Liu investigated the well-posedness of solutions to a class of second-order non-autonomous abstract time-delay functional differential equations with infinite delay and demonstrated the application of their conclusions through examples (see Section 6 in [23]), which can be referenced in [23,24,25]. However, these articles only studied the well-posedness of solutions for non-autonomous systems with infinite time delay. They did not address the asymptotic behavior and regularity of solutions, and the equations provided in the examples were not specifically applied to several specific differential equations. In this paper, we apply the above research conclusions to the Navier–Stokes equations. Based on the extensive foundational work on well-posedness established by Caraballo et al., we use the theory of determining nodes to characterize the asymptotic behavior of solutions to the non-autonomous Navier–Stokes equations with infinite delay by estimating their number. The most crucial step in studying the process and methods of non-autonomous Navier–Stokes equations with infinite delay is to handle the infinite time delay term

u (t + θ)

, which is particularly important for studying the well-posedness of nonlinear PDEs with time delay effects. In addition, dealing with the delay term is usually the most critical step in the research process of the well-posedness and asymptotic behavior of nonlinear PDE solutions with other delay effects.

The paper is arranged as follows: In Section 2, we first establish several basic function spaces and several operators, then introduce several relevant theorems and define the weak and strong solutions of Equation (1). In addition, to deal with the infinite delay term

u (t + θ)

, we refer to descent function space

C_{γ} (H)

. Unlike the finite delay term, we not only need to consider the influence of the system state at a certain time in the past, but also need to consider the comprehensive influence of the system state at the past time. In Section 3, we investigate the global well-posedness of the strong solutions for Equation (1) by the Galerkin method without considering the pressure term p. In Section 4, we commit our focus to proving Equation (1) has a finite number of determining nodes.

2. Preliminaries

In this section, we introduce several common conclusions that are essential applications that we used in this paper.

2.1. Basic Function Spaces and Conclusions

We introduce several basic function spaces, which are as follows:

\begin{matrix} V : = \{φ \in {(C_{0}^{\infty} (Ω))}^{2} : div φ = 0, φ = (φ^{(1)}, φ^{(2)})\}; \\ H : = closure of V in {(L^{2} (Ω))}^{2}, with norm {∥ \cdot ∥}_{H}; \\ V : = closure of V in {(H_{0}^{1} (Ω))}^{2}, with norm {∥ \cdot ∥}_{V}; \\ V^{'} : = dual space of V with norm {∥ \cdot ∥}_{V^{'}} . \end{matrix}

The definitions of norms

{∥ \cdot ∥}_{H}

and

{∥ \cdot ∥}_{V}

are as follows:

{∥ φ ∥}_{H} = {(\int_{Ω} {| φ |}^{2} d x)}^{\frac{1}{2}}, {∥ φ ∥}_{V} = {(\int_{Ω} ({| φ |}^{2} + {| \nabla φ |}^{2}) d x)}^{\frac{1}{2}};

for convenience, we use “

∥ \cdot ∥

” as “

{∥ \cdot ∥}_{H}

”. In addition, we denote the inner product in

L^{2} (Ω), H

(\cdot, \cdot)

and denote the inner product in V by

((\cdot, \cdot))

. In addition, we denote the dual product between V and

V^{'}

〈\cdot, \cdot〉

In order to be able to express Equation (1) as an abstract equation, we define several operators. Firstly, we define the Stokes operator A as follows:

\begin{matrix} 〈A u, φ〉 = ν ((u, φ)) = ν \sum_{i, j = 1}^{2} \int_{Ω} \frac{\partial u_{i}}{\partial x_{j}} \frac{\partial φ_{i}}{\partial x_{j}} d x, \end{matrix}

(2)

for any

u = (u^{(1)}, u^{(1)}), φ = (φ^{(1)}, φ^{(2)}) \in V

. Moreover,

D (A) : = V \cap {(H^{2} (Ω))}^{2}

, and the operator A is the linear continuous operator, which both form V to

V^{'}

and

D (A)

to H. Then, we denote the linear function

b (u, v, w) = \sum_{i, j = 1}^{2} \int_{Ω} u_{i} \frac{\partial v_{j}}{\partial x_{i}} w_{j} d x, \forall u, v, w \in V .

It is obvious that the linear function

b (\cdot, \cdot, \cdot)

is continuous trilinear on

V \times V \times V

and it satisfies

b (u, v, w) = - b (u, w, v), b (u, v, v) = 0, \forall u, v, w \in V .

In addition, we use the following lemma to provide estimators of the linear function

b (\cdot, \cdot, \cdot)

, which can be found in references [26,27].

Lemma 1.

There exists a positive constant

C_{Ω}

that depends only on Ω, such that

\begin{matrix} | b (u, v, w) | \leq C_{Ω} {∥ u ∥}^{\frac{1}{2}} {∥ A u ∥}^{\frac{1}{2}} {∥ v ∥}_{V} ∥ w ∥, \forall u \in D (A), v \in V, w \in H . \end{matrix}

(3)

\begin{matrix} | b (u, v, w) | \leq C_{Ω} {∥ A u ∥ ∥ v ∥}_{V} ∥ w ∥, \forall u \in D (A), v \in V, w \in H . \end{matrix}

(4)

u \in L^{\infty}, v \in V, w \in H

, then the linear function

b (\cdot, \cdot, \cdot)

satisfies

\begin{matrix} | b (u, v, w) | \leq {∥ u ∥}_{L^{\infty}} {∥ v ∥}_{V} ∥ w ∥ . \end{matrix}

(5)

u, v, w \in V

, it satisfies

\begin{matrix} | b (u, v, w) | \leq {∥ u ∥}^{\frac{1}{2}} {∥ \nabla u ∥}^{\frac{1}{2}} {∥ v ∥}^{\frac{1}{2}} {∥ \nabla v ∥}^{\frac{1}{2}} ∥ w ∥ . \end{matrix}

(6)

To handle the infinite delays

G (t, u (t + θ))

, we define the space

C_{γ} (H)

with a suitable constant

γ > 0

, as follows:

C_{γ} (H) = \{ζ \in C ((- \infty, 0); H) | \exists lim_{s \to - \infty} e^{γ s} ζ (s) \in H\},

which is a Banach space with the norm

{∥ ζ ∥}_{C_{γ} (H)} = sup_{s \in (- \infty, 0]} e^{γ s} ∥ ζ (s) ∥ .

The space

C_{γ} (H)

represents the space of descent functions, where the parameter

γ > 0

is typically used to describe the growth rate of the function in

C_{γ} (H)

. The purpose of introducing

C_{γ} (H)

is to ensure that the growth property of the function does not affect the long-term behavior of the system and also satisfies the descent conditions required by continuous functions with changing states due to time delay effects.

To establish the well-posedness of Equation (1), the continuous function

G : [τ, T] \times C_{γ} (H) \mapsto {(L^{2} (Ω))}^{2}

is assumed to satisfy the following conditions.

(I) For any

η

, the mapping

t \mapsto G (t, η)

is measurable;

(II)

G (t, 0) = (0, 0)

;

(III) There exists

L_{G} > 0

, such that

\begin{matrix} ∥ G (t, x_{1}) - G (t, x_{2}) ∥ \leq L_{G} {∥ x_{1} - x_{2} ∥}_{C_{γ} (H)}, \forall x_{1}, x_{2} \in C_{γ} (H) . \end{matrix}

(7)

When studying differential equations with infinite time delay, Lipschitz conditions also help to investigate the asymptotic behavior of solutions. For example, when studying the steady-state or periodic solutions of a system after long-term operation, the Lipschitz condition can ensure that the solutions do not experience explosive growth in time. In addition, the Lipschitz condition ensures understanding of the continuous dependence on the initial conditions, as infinite delay terms may lead to temporal irregularities in the solution. By ensuring that the mapping is Lipschitz, it can be proven that small initial perturbations do not cause significant changes in the solution, thereby ensuring its stability.

Based on the above conclusion, we define the weak and strong solution of Equation (1) as follows.

2.2. The Weak Solution and Strong Solution

Definition 1.

Assume

f \in L^{2} (τ, T; V^{'})

and

u (τ) = ϕ \in C_{γ} (H)

. A weak solution is any function

u \in C_{γ} (H)

of Equation (1) if it satisfies

(I)

u \in L^{\infty} (τ, T; H) \cap L^{2} (τ, T; V),

for all

T > τ

;

(II) For all

σ \in V

, such that

\begin{matrix} \frac{d}{d t} (u (t), σ) - ν 〈A u (t), σ〉 + b (u (t), u (t), σ) = 〈f (t), σ〉 + (G (t, u (t + θ)), σ), \end{matrix}

(8)

\begin{matrix} u (τ + θ) = ϕ (θ) \in C_{γ} (H), θ \in (- \infty, 0] \end{matrix}

(9)

holds in the distribution sense of

D^{'} (τ, T)

Assume

f \in L^{2} (τ, T; H)

; if

u \in C_{γ} (H)

is a weak solution of Equations (8) and () and

u \in L^{\infty} (τ, T; V) \cap L^{2} (τ, T; D (A))

for all

T > τ

, then

u \in C_{γ} (H)

is called a strong solution of Equations (8) and (9). Furthermore, the function

u^{'} \in L^{2} (τ, T; H)

Remark 1.

u \in C_{γ} (H)

is a weak solution to Equation (1), then in this case, from Equations (8) and (9), for any

τ \leq s \leq t

, the following energy equality holds

\begin{matrix} {∥ u (t) ∥}^{2} + 2 ν \int_{s}^{t} {∥ u (r) ∥}_{V}^{2} d r \\ = & {∥ u (τ) ∥}^{2} + 2 \int_{s}^{t} (〈f (r), u (r)〉 + (G (t, u (r + θ))), u (r)) d r . \end{matrix}

(10)

In addition, we introduce a useful inequality in the differential equation theory. This will play an important role in making a priori estimates.

Lemma 2.

(Gronwall Inequation) Assume

g, p, q \in C ([a, b]; R^{+})

and for all

t \in [a, b]

satisfy

g (t) \leq g (a) + \int_{a}^{t} (p (s) g (s) + q (s)) d s,

then any

t \in [a, b]

, and functions

g, p, q

satisfy

\begin{matrix} g (t) \leq e^{\int_{a}^{t} p (s) d s} (g (a) + \int_{a}^{t} q (s) d s), \forall t \in [a, b] . \end{matrix}

(11)

3. Global Well-Posedness

In this section, we study the global well-posedness of Equation (1). Therefore, we present the existence and uniqueness theorem for the weak solution of Equation (1). Therefore, we apply the treatment of the infinite delay term

u (t + θ)

to the proof of the global well-posedness of the non-autonomous Navier–Stokes equation, that is, adding the infinite delay term

u (t + θ)

to the proof of the global well-posedness of the non-autonomous Navier–Stokes equation.

3.1. The Existence of Solutions

Theorem 1.

Suppose

f \in L^{2} (τ, T; V^{'})

, and there exist constants

C = C (ν, λ_{1}, L_{G})

that satisfy

\frac{1}{ν} \int_{τ}^{t} e^{C (t - r)} {∥ f (r) ∥}_{V^{'}}^{2} d r < + \infty,

and the function

G (\cdot, \cdot)

satisfies the conditions (I)-(III). Then for all

T > τ

and

u (τ) = ϕ \in C_{γ} (H)

, the function

u \in C_{γ} (H) \cap L^{\infty} (τ, T; H) \cap L^{2} (τ, T; V)

is the weak solution of Equation (1).

Proof.

For the proof of Theorem 1, we use the Galerkin approximation method. The proof is divided into the following three steps.

Step 1:: Construct the approximate solution.

Taking a set of canonical orthonormal basis

\{ξ_{i}\} \subset V

on the Hilbert space H, denote

V_{k} = span [ξ_{1}, ξ_{2}, \dots, ξ_{k}], k = 1, 2, \dots;

from the Gram–Schmidt theorem, the projection of H onto

V_{k}

is defined by

P_{k} u = \sum_{i = 1}^{k} (u, ξ_{i}) ξ_{i} .

Since the Stokes operator A is compact operator on the V space, it satisfies

A ξ_{i} = λ_{i} ξ_{i}, (λ_{i} > 0, i = 1, 2 \dots, k),

where

λ_{i}

is the eigenvalue of Stokes operator A. Sort them in ascending order and renumber them as follows:

0 < λ_{1} \leq λ_{2} \leq \dots \leq λ_{n} \leq \dots,

and they satisfy

lim_{i \to \infty} λ_{i} = \infty

. We take the minimum eigenvalue

λ_{1}

of Stokes operator A as the first eigenvalue.

From this, we construct the approximate solution of Equation (1) from the space

V_{k}

as follows:

u_{k} (x, t) = \sum_{i = 1}^{k} a_{k i} (t) ξ_{i} (x),

where

a_{k i} (t) = (u_{k}, ξ_{i})

and all

t \in (τ, T)

satisfy

\begin{matrix} \frac{d}{d t} (u_{k} (t), ξ_{i}) + ν 〈A u_{k} (t), ξ_{i}〉 + b (u_{k}, u_{k}, ξ_{i}) = 〈f (t), ξ_{i}〉 + (G (t, u_{k} (t + θ)), ξ_{i}), \end{matrix}

(12)

and initial conditions

\begin{matrix} u_{k} (τ + θ) = P_{k} ϕ (θ), θ \in (- \infty, 0] . \end{matrix}

(13)

Therefore, Equations (12) and (13) satisfy the condition of existence and uniqueness of local solutions for the system of ordinary differential equation with infinite delay (e.g., [19]).

Step 2:: A priori estimate of the approximate solution.

Firstly, we estimate Equation (12), multiplied by

a_{k i} (t)

to Equation (12) and sum it to obtain the ODE:

\begin{matrix} \frac{1}{2} \frac{d}{d t} {∥ u_{k} (t) ∥}^{2} + ν ∥ u_{k} {(t) ∥}_{V}^{2} = 〈f (t), u_{k} (t)〉 + (G (t, u_{k} (t + θ)), u_{k} (t)), \end{matrix}

(14)

which implies

\begin{matrix} \frac{1}{2} \frac{d}{d t} ∥ u_{k} {(t) ∥}^{2} + ν ∥ u_{k} {(t) ∥}_{V}^{2} \leq {∥ f ∥}_{V^{'}} ∥ u_{k} {(t) ∥}_{V} + ∥ G (t, u_{k} (t + θ)) ∥ ∥ u_{k} {(t) ∥}_{V} . \end{matrix}

(15)

Using Young and Poincare inequations, we choose suitable constants

ε_{1} = ε_{2} = \frac{ν}{4}

, then we obtain

\begin{matrix} {∥ f ∥}_{V^{'}} ∥ u_{k} {(t) ∥}_{V} + ∥ G (t, u_{k} (t + θ)) ∥ ∥ u_{k} (t) ∥ \\ \leq & \frac{1}{2 ε_{1}} {∥ f (t) ∥}_{V^{'}}^{2} + \frac{ν}{4} ∥ u_{k} {(t) ∥}_{V}^{2} + \frac{L_{G}}{2 ε_{2}} {∥ u_{k} (t + θ) ∥}_{C_{γ} (H)}^{2}, \end{matrix}

(16)

where

∥ u_{k} {(t + θ) ∥}_{C_{γ} (H)} = sup_{s \in (- \infty, 0]} e^{γ s} ∥ u_{k} (t + s) ∥ \geq ∥ u_{k} (t) ∥, t \in [τ, T]

. Substituting (16) into Equation (15), for

τ \leq r \leq t \leq T

, we obtain

\begin{matrix} \frac{d}{d r} {∥ u_{k} (r) ∥}^{2} + ν λ_{1} ∥ u_{k} {(r) ∥}^{2} + \frac{ν}{2} ∥ u_{k} {(r) ∥}_{V}^{2} \leq \frac{4}{ν} {∥ f (r) ∥}_{V^{'}}^{2} + \frac{4 L_{G}}{ν} {∥ u_{k} (r + θ) ∥}_{C_{γ} (H)}^{2} . \end{matrix}

(17)

Multiplying by

e^{- ν λ_{1} (t - r)}

and then integrating it for r over

[τ, t]

, we obtain

\begin{matrix} ∥ u_{k} {(t) ∥}^{2} + \frac{ν}{2} \int_{τ}^{t} e^{- ν λ_{1} (t - r)} {∥ u_{k} (r) ∥}_{V}^{2} d r \\ \leq & e^{- ν λ_{1} (t - τ)} {∥ u_{k} (τ) ∥}^{2} + \int_{τ}^{t} e^{- ν λ_{1} (t - r)} (\frac{4}{ν} {∥ f (r) ∥}_{V^{'}}^{2} + \frac{4 L_{G}}{ν} {∥ u_{k} (r + θ) ∥}_{C_{γ} (H)}^{2}) d r . \end{matrix}

(18)

According to the definition of

C_{γ} (H)

, we obtain

\begin{matrix} ∥ u_{k} {(t + θ) ∥}_{C_{γ} (H)}^{2} & = {(sup_{s \in (- \infty, 0]} e^{γ s} ∥ u_{k} (t + s) ∥)}^{2}, \\ \leq max \{sup_{s \in (- \infty, τ - t]} e^{2 γ s} ∥ u_{k} {(t + s) ∥}^{2}, sup_{s \in (τ - t, 0]} e^{2 γ s} {∥ u_{k} (t + s) ∥}^{2}\} \\ = max \{J_{1} (t), J_{2} (t) + J_{3} (t)\}, \end{matrix}

(19)

where

\begin{matrix} J_{1} (t) = sup_{s \in (- \infty, τ - t]} e^{2 γ s} {∥ u_{k} (t + s) ∥}^{2}, \\ J_{2} (t) = sup_{s \in (τ - t, 0]} [e^{2 γ s - ν λ_{1} (t + s - τ)} {∥ u_{k} (τ) ∥}^{2}], \\ J_{3} (t) = sup_{s \in (τ - t, 0]} [e^{(2 γ - ν λ_{1}) s} \int_{τ}^{t + s} e^{- ν λ_{1} (t - r)} (\frac{4}{ν} {∥ f (r) ∥}_{V^{'}}^{2} + \frac{4 L_{G}}{ν} {∥ u_{k} (r + θ) ∥}_{C_{γ} (H)}^{2}) d r] . \end{matrix}

Now, let us estimate

J_{1} (t), J_{2} (t)

, and

J_{3} (t)

. On one hand, since

ν λ_{1} < 2 γ

\begin{matrix} J_{1} (t) & = sup_{s \in (- \infty, τ - t]} e^{2 γ s} {∥P_{k} ϕ (t + s - τ)∥}^{2} \leq sup_{s \in (- \infty, τ - t]} e^{2 γ s} {∥ ϕ (t + s - τ) ∥}^{2} \\ = sup_{s \in (- \infty, 0]} e^{2 γ [s - (t - τ)]} {∥ ϕ (s) ∥}^{2} = e^{2 γ (τ - t)} {∥ ϕ (s) ∥}_{C_{γ} (H)}^{2} \\ \leq e^{- ν λ_{1} (t - τ)} {∥ ϕ (s) ∥}_{C_{γ} (H)}^{2} . \end{matrix}

(20)

On the other hand, by direct computation, we have

\begin{matrix} J_{2} (t) \leq e^{- ν λ_{1} (t - τ)} ∥ u_{k} {(τ) ∥}^{2} \leq e^{- ν λ_{1} (t - τ)} {∥ ϕ (s) ∥}_{C_{γ} (H)}^{2}, \end{matrix}

(21)

\begin{matrix} J_{3} (t) \leq \int_{τ}^{t} e^{- ν λ_{1} (t - r)} (\frac{4}{ν} {∥ f (r) ∥}_{V^{'}}^{2} + \frac{4 L_{G}}{ν} {∥ u_{k} (r + θ) ∥}_{C_{γ} (H)}^{2}) d r . \end{matrix}

(22)

Then, Equations (19)–(22) give

\begin{matrix} ∥ u_{k} {(t + θ) ∥}_{C_{γ} (H)}^{2} \\ \leq & e^{- ν λ_{1} (t - τ)} {∥ ϕ (s) ∥}_{C_{γ} (H)}^{2} + \int_{τ}^{t} e^{- ν λ_{1} (t - r)} (\frac{4}{ν} {∥ f (r) ∥}_{V^{'}}^{2} + \frac{4 L_{G}}{ν} {∥ u_{k} (r + θ) ∥}_{C_{γ} (H)}^{2}) d r . \end{matrix}

(23)

By Lemma 2, this implies

\begin{matrix} ∥ u_{k} {(t + θ) ∥}_{C_{γ} (H)}^{2} \leq e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - τ)} {∥ ϕ (s) ∥}_{C_{γ} (H)}^{2} + \frac{4}{ν} \int_{τ}^{t} e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - r)} {∥ f (r) ∥}_{V^{'}}^{2} d r, \end{matrix}

(24)

combining Equations (18) and (24); it can be concluded that the

\{u_{k} (t)\}

is bounded in

L^{\infty} (τ, T; H)

. According to Equation (18), this means

\begin{matrix} \frac{ν}{2} e^{- ν λ_{1} (t - τ)} \int_{τ}^{t} {∥ u_{k} (r) ∥}_{V}^{2} d r \\ \leq & ∥ u_{k} {(τ) ∥}^{2} + \int_{τ}^{t} e^{- ν λ_{1} (t - r)} (\frac{4}{ν} {∥ f (r) ∥}_{V^{'}}^{2} + \frac{4 L_{G}}{ν} {∥ u_{k} (r + θ) ∥}_{C_{γ} (H)}^{2}) d r . \end{matrix}

(25)

By Equation (24), this implies that the

\{u_{k} (t)\}

is bounded in

L^{2} (τ, T; V)

. In addition, for any

ψ \in V

, we obtain

\begin{matrix} 〈u_{k}^{'} (t), ψ〉 + 〈A u_{k} (t), ψ〉 + b (u_{k}, u_{k}, ψ) = 〈f (t), ψ〉 + 〈G (t, u_{k} (t + θ)), ψ〉 . \end{matrix}

Using the Holder inequation, we obtain the following result:

\begin{matrix} |〈u_{k}^{'} (t), ψ〉| & \leq |〈A u_{k} (t), ψ〉| + |b (u, u, ψ)| + | 〈f (t), ψ〉 | + | 〈G (t, u_{k} (t + θ)), ψ〉 | \\ \leq {∥ ψ ∥}_{V} (ν ∥ u_{k} (t) ∥_{V} + C_{Ω} ∥ u_{k} (t) ∥ ∥ \nabla u_{k} (t) {∥ + ∥ f ∥}_{V^{'}} + ∥ G (t, u_{k} (t + θ)) ∥), \\ \leq {∥ ψ ∥}_{V} (ν ∥ u_{k} (t) ∥_{V} + C_{Ω} ∥ u_{k} (t) ∥ ∥ u_{k} (t) ∥_{V} + {∥ f ∥}_{V^{'}} + ∥ G (t, u_{k} (t + θ)) ∥) . \end{matrix}

(26)

Hence, it follows from Equations (7), (18), (24) and (25) that the sequence

\{u_{k}^{'} (t)\}

is bounded in

L^{2} (τ, T; V^{'})

Step 3:: Approximation of approximate solutions.

Now, based on the conclusions of the previous step, since the compactness theorem (see, e.g., [1,11,18]), for all

T > τ

, there exist subsequences, which we still denote by

\{u_{k}\}

, such that as

k \to \infty

, the following conditions satisfy:

\begin{matrix} u_{k} \overset{*}{⇀} u in L^{\infty} (τ - h, T; H), \end{matrix}

(27)

\begin{matrix} u_{k} ⇀ u in L^{2} (τ, T; V), \end{matrix}

(28)

\begin{matrix} u_{k}^{'} ⇀ u^{'} in L^{2} (τ, T; V^{'}), \end{matrix}

(29)

\begin{matrix} u_{k} \to u in L^{2} (τ, T; H), \end{matrix}

(30)

\begin{matrix} G (t, u_{k} (t + θ)) \overset{*}{⇀} η (t) in L^{\infty} (τ, T; {(L^{2} (Ω))}^{2}) . \end{matrix}

(31)

where

η (t) \in L^{\infty} (τ, T; {(L^{2} (Ω))}^{2})

is the limiting function of

G (t, u_{k} (t + θ))

. However, currently, we cannot yet obtain it:

\begin{matrix} G (t, u_{k} (t + θ)) \to G (t, u (t + θ)) \in L^{2} (τ, T; {(L^{2} (Ω))}^{2}) . \end{matrix}

(32)

Next, we will prove Equation (32) through conclusions (27)–(31).

From conclusions (27)–(31), we obtain

\begin{matrix} u_{k} (t) \to u (t), a . e . t \in (τ, T), in H, \end{matrix}

(33)

then for all

s, t \in [τ, T]

, it implies

\begin{matrix} u_{k} (t) - u_{k} (s) = \int_{s}^{t} {u_{k}}^{'} (r) d r, in V^{'}; \end{matrix}

(34)

therefore, the

\{u_{k}\}

is equicontinuous for all

[τ, T]

and k in H. By the embedding relation

V ↪ ↪ H ↪ ↪ V^{'}

, and the Ascoli–Arzela theorem, it yields that

\{u_{k}\}

is relatively compact in

V^{'}

, and for all

t \in [τ, T]

, which implies

\begin{matrix} u_{k} (t) \to u (t), in C ([τ, T]; V^{'}) . \end{matrix}

(35)

According to conclusions (27)–(31), for any sequence

\{t_{k}\} \subset [τ, T]

that satisfies

t_{k} \to t

\begin{matrix} u_{k} (t_{k}) ⇀ u (t), in H . \end{matrix}

(36)

Before proving Equation (32), we need to first prove

\begin{matrix} u_{k} (t) \to u (t), in C ([τ, T]; H) . \end{matrix}

(37)

Since for all

r \in [s, t] \subset [τ, T]

, the approximate solution

u_{k}

satisfies

\begin{matrix} ∥ u_{k} {(t) ∥}^{2} + 2 ν \int_{s}^{t} {∥ u_{k} (r) ∥}_{V}^{2} d r \\ = & ∥ u_{k} {(s) ∥}^{2} + 2 \int_{s}^{t} 〈f (r), u_{k} (r))〉 d r + 2 \int_{s}^{t} (G (r, u_{k} (r + θ), u_{k} (r))) d r, \end{matrix}

(38)

we obtain the following result by conclusions (27)–(31),

\begin{matrix} \int_{s}^{t} {∥ G (r, u_{k} (r + θ)) ∥}^{2} d r \leq C_{1} (t - s), \forall r \in [s, t] \subset [τ, T], \end{matrix}

(39)

where

C_{1}

is a positive constant. According to the conclusions of (27)–(31), we verify that the function

u \in C_{γ} (H) \cap L^{\infty} (τ, T; H) \cap L^{2} (τ, T; V),

is a weak solution of Equation (1), and it satisfies

\begin{matrix} \frac{d}{d t} (u (t), w) + ν 〈A u (t), w〉 + b (u (t), u (t), w) = 〈f (t), w〉 + (η (t), w) \forall w \in V, \end{matrix}

(40)

in addition, for any

r \in [s, t] \subset [τ, T]

, the energy estimates also satisfy

\begin{matrix} {∥ u (t) ∥}^{2} + 2 ν \int_{s}^{t} {∥ u (r) ∥}_{V}^{2} d r \\ = & {∥ u (τ) ∥}^{2} + 2 \int_{s}^{t} (〈f (r), u (r)〉 + (G (t, u (r + θ)), u (r)) d r . \end{matrix}

(41)

Due to the conclusions of (27)–(31) and Equation (39), we obtain

\begin{matrix} \int_{s}^{t} {∥ η (r) ∥}^{2} d r \leq \underset{k \to \infty}{lim inf} \int_{s}^{t} {∥ G (t, u (r + θ)) ∥}^{2} d r \leq C_{1} (t - s); \end{matrix}

(42)

therefore, consider the continuous functions

L_{k}, L : [τ, t] \to R

, which are defined by

\begin{matrix} L_{k} (t) = \frac{ν}{2} {∥ u_{k} (t) ∥}_{V}^{2} - \int_{s}^{t} 〈f (r), u_{k} (r)〉 d r - C_{1} t, \\ L (t) = \frac{ν}{2} {∥ u (t) ∥}_{V}^{2} - \int_{s}^{t} 〈f (r), u (r)〉 d r - C_{1} t . \end{matrix}

As shown in Equations (38) and (41), it is evident that the functions

L_{k} (t), L (t)

are non-increasing and continuous in

[τ, T]

. According to the conclusions of (27)–(30), we obtain

\begin{matrix} L_{k} (t) \to L (t) a . e . t \in [τ, T], \end{matrix}

(43)

from the conclusions of (27)–(30), it is obvious that

∥ u (t_{0}) ∥ \leq \underset{k \to \infty}{lim inf} ∥ u_{k} (t_{k}) ∥

. To prove

\begin{matrix} \underset{k \to \infty}{lim sup} ∥ u_{k} (t_{k}) ∥ \leq ∥ u (t_{0}) ∥, \end{matrix}

(44)

we assume

t_{0} > τ

and take the sequence

\{{\bar{t}}_{n}\} \subset \{t_{k}\}

that satisfies (43), which is increasing, and approach the value

t_{0}

from the left when

n \to \infty

. Since

L (\cdot)

is continuous, there exists

N_{ε}

, and when

n > N_{ε}

| L ({\bar{t}}_{n}) - L (t_{0}) | < ε .

Take

k > K (N_{ε})

such that

t_{k} > {\bar{t}}_{N_{ε}}

, by the sequence

L (\cdot), L_{k} (\cdot)

, is non-increasing and satisfies (43); we obtain

| L_{k} (t_{k}) - L (t_{0}) | \leq | L_{k} ({\bar{t}}_{N_{ε}}) - L ({\bar{t}}_{N_{ε}}) | + | L ({\bar{t}}_{N_{ε}}) - L (t_{0}) | < 2 ε .

At this time, the conclusions of (27)–(30) imply

\begin{matrix} \int_{τ}^{t_{k}} 〈f (r), u_{k} (r)〉 d r \to \int_{τ}^{t} 〈f (r), u (r)〉 d r, (k \to \infty) . \end{matrix}

(45)

Due to these and functions

L_{k}, L

, we have proven Equation (44).

According to the definition of

C_{γ} (H)

, we obtain

\begin{matrix} ∥ u_{k} {(t + θ) - u (t + θ) ∥}_{C_{γ} (H)} \\ = & sup_{s \in (- \infty, 0]} e^{γ s} ∥ u_{k} (t + s) - u (t + s) | = max \{Φ_{1} (s), Φ_{2} (s)\} \\ \leq & max \{e^{γ (τ - t)} ∥ P_{k} {ϕ - ϕ ∥}_{C_{γ} (H)}, sup_{s \in (τ - t, 0]} e^{γ s} ∥ u_{k} (t + s) - u (t + s) ∥\}, \end{matrix}

(46)

where

\begin{matrix} Φ_{1} (s) = sup_{s \in (- \infty, τ - t]} e^{γ s} ∥ P_{k} ϕ (t + s - τ) - ϕ (t + s - τ) ∥, \\ Φ_{2} (s) = sup_{s \in (τ - t, 0]} e^{γ s} ∥ u_{k} (t + s) - u (t + s) ∥, \end{matrix}

which implies that

\begin{matrix} lim_{k \to \infty} {∥ u_{k} (t + θ) - u (t + θ) ∥}_{C_{γ} (H)} = 0 . \end{matrix}

(47)

In addition, the convergence results are obtained by Equations (7), (37) and (47), and conclusions of (27)–(31); therefore, we prove Equation (32). Thus, we have proven Theorem 1. □

3.2. The Uniqueness of Solutions

Theorem 2.

If all the conditions in the Theorem 1 are satisfied, then the function

u \in C_{γ} (H) \cap L^{\infty} (τ, T; H) \cap L^{2} (τ, T; V)

is the unique weak solution of Equation (1).

Proof.

Set

{\bar{u}}_{1}

and

{\bar{u}}_{2}

be the weak solutions of Equation (1). Denote the function

U (t) = {\bar{u}}_{1} (t) - {\bar{u}}_{2} (t)

; for any

t > τ

, we have

\begin{matrix} \frac{1}{2} \frac{d}{d t} {∥ U (t) ∥}^{2} + ν {∥ \nabla U (t) ∥}^{2} + b (U, {\bar{u}}_{1}, U) = (G (t, {\bar{u}}_{1} (t + θ)) - G (t, {\bar{u}}_{2} (t + θ)), U (t)) . \end{matrix}

(48)

Furthermore, according to Lemma 1, and Equations (14) and (39), using Young, Holder, and Poincare inequations, we obtain

\begin{matrix} \frac{1}{2} \frac{d}{d t} {∥ U (t) ∥}^{2} + ν {∥ U (t) ∥}_{V}^{2} \\ \leq & C_{Ω} (\frac{{∥ U (t) ∥}^{2} {∥ {\bar{u}}_{1} (t) ∥}_{V}^{2}}{2 ε_{3}} + \frac{ε_{3}}{2} {∥ U (t) ∥}_{V}^{2}) + L_{G} {∥ U (t + θ) ∥}_{C_{γ} (H)} ∥ U (t) ∥ . \end{matrix}

(49)

Integrating Equation (49), we obtain

\begin{matrix} \frac{1}{2} {∥ U (t) ∥}^{2} - \frac{1}{2} {∥ U (τ) ∥}^{2} + ν \int_{τ}^{t} {∥ U (r) ∥}_{V}^{2} d r \\ \leq & C_{Ω} \int_{τ}^{t} (\frac{{∥ U (r) ∥}^{2} {∥ {\bar{u}}_{1} (r) ∥}_{V}^{2}}{2 ε_{3}} + \frac{ε_{3}}{2} {∥ U (r) ∥}_{V}^{2}) d r + \int_{τ}^{t} L_{G} {∥ U (r + θ) ∥}_{C_{γ} (H)} ∥ U (r) ∥ d r . \end{matrix}

(50)

Take

ε_{3} = \frac{2 ν}{C_{Ω}}

; then, according to the definition of

C_{γ} (H)

, we obtain

∥ U (r + θ) ∥ = sup_{s \in (- \infty, 0]} e^{γ s} ∥ U (r + s) ∥ \leq sup_{z \in (τ, r)} ∥ U (z) ∥, r \in [τ, t] .

Therefore, we have

\begin{matrix} sup_{z \in (τ, t)} {∥ U (z) ∥}^{2} \leq \int_{τ}^{t} (\frac{C_{Ω} {∥ {\bar{u}}_{1} (r) ∥}_{V}^{2}}{2 ν} + 2 L_{G}) sup_{z \in (τ, r)} {∥ U (z) ∥}^{2} d r, \end{matrix}

(51)

where

{∥ U (τ) ∥}^{2} = 0

. By Lemma 2, we conclude that

{∥ U (t) ∥}^{2} = 0

; therefore, we have proven Theorem 2. □

4. Estimate the Number of Determining Nodes

In this section, we prove that Equation (1) has a finite number of determining nodes.

4.1. Introduction to Relevant Lemmas

Let

{\bar{u}}_{1}

and

{\bar{u}}_{2}

be the two strong solutions of Equation (1) corresponding to the external force and torque

f_{1}

and

f_{2}

, respectively, and

f_{1}, f_{2} \in L^{2} (τ, T; H)

. The asymptotic strength of the external force and moment is described by its

L^{2}

norm, i.e.,

\begin{matrix} F : = \underset{t \to + \infty}{lim sup} {(\int_{Ω} {| f_{i} (x, t) |}^{2})}^{\frac{1}{2}} d x, i = 1, 2 . \end{matrix}

(52)

which can be found in references [12,28]. Therefore, we have the following definition.

Definition 2

([12,28,29]). Consider N measuring points

x_{i}

in the space

Ω \subseteq R^{2}

, where

i = 1, 2, \dots, N

. Denoted

Λ = \{x^{1}, x^{2}, \dots, x^{N}\}

. Suppose that

{\bar{u}}_{1}

and

{\bar{u}}_{2}

are two strong solutions of Equation (1). If by

\begin{matrix} lim_{t \to + \infty} max_{i = 1, 2, \dots, N} | {\bar{u}}_{1} (x^{i}, t) - {\bar{u}}_{2} (x^{i}, t) | = 0, \end{matrix}

(53)

we can obtain

\begin{matrix} lim_{t \to + \infty} \int_{Ω} {| {\bar{u}}_{1} (x, t) - {\bar{u}}_{2} (x, t) |}^{2} d x = 0, \end{matrix}

(54)

then the point set Λ is called the definite node set of Equation (1), and the point in Λ is the definite node of Equation (1).

In addition, in order to prove the main conclusions of this section, the following important lemmas (see [11,26,29,30,31]) should also be used.

Lemma 3.

Let Ω be covered by N identical squares. Remember the point set

Λ = \{x^{1}, x^{2}, \dots, x^{N}\} \subset Ω,

where the point

x^{i}

belongs to and only belongs to one of the squares, then for

\forall u \in D (A)

, there exists the normal number

C_{Ω}

, which is only dependent on Ω, such that

\begin{matrix} {∥ \nabla u ∥}^{2} \leq C_{Ω} N η^{2} (u) + \frac{C_{Ω}}{λ_{1} N} {∥ A u ∥}^{2}, \end{matrix}

(55)

where

η (u) = max_{i = 1, 2, \dots, N} | u (x^{i}) | .

Lemma 4.

P (t)

and

Q (t)

are real-valued functions on

[τ, + \infty)

and there exist

T > 0

and constant

C_{2}

, such that

\begin{matrix} \underset{t \to + \infty}{lim inf} \frac{1}{T} \int_{t}^{t + T} P (r) d r = C_{2} > 0, \underset{t \to + \infty}{lim sup} \frac{1}{T} \int_{t}^{t + T} P^{-} (r) d r < \infty, \end{matrix}

(56)

and

\begin{matrix} lim_{t \to + \infty} \frac{1}{T} \int_{t}^{t + T} Q^{+} (r) d r = 0, \end{matrix}

(57)

where

P^{-} (t) = max \{- P (t), 0\}, Q^{+} (t) = max \{Q (t), 0\}

. Suppose

ξ (t)

is the non-negative absolutely continuous function on

[τ, + \infty)

. If

ξ (t)

satisfies

\frac{d ξ (t)}{d t} + P (t) ξ (t) \leq Q (t),

then

lim_{t \to + \infty} ξ (t) = 0

Next, we prove that if any two strong solutions of Equation (1) have the same asymptotic behavior at a finite number of points in space, then these two solutions will have the same asymptotic behavior almost everywhere in the entire space.

4.2. Main Results

Theorem 3.

Let Ω be covered by N identical squares. The set Λ is defined in Lemma 3,

N > \frac{16 C_{Ω}^{3} C_{5}}{5 ν^{2} λ_{1}}

, where

C_{5} : = C_{4} {∥ ϕ (s) ∥}^{2} + C_{3} F^{2}

and each point

x^{i}

belongs to and only belongs to one of the squares. Let

{\bar{u}}_{1}

and

{\bar{u}}_{2}

be two strong solutions of Equation (1) corresponding to external forces

f_{1}

and

f_{2}

, respectively, and

f_{1}

and

f_{2}

have the same asymptotic strength, i.e.,

\begin{matrix} lim_{t \to + \infty} \int_{Ω} {| f_{i} (x, t) |}^{2} d x = 0 . \end{matrix}

(58)

Then, the point set Λ is the definite node set of Equation (1).

Proof.

Denote

U = {\bar{u}}_{1} - {\bar{u}}_{2}, f = f_{1} - f_{2}

and

G (t, U (t + θ)) = G_{1} (t, {\bar{u}}_{1} (t + θ)) - G_{2} (t, {\bar{u}}_{2} (t + θ)) .

Since

{\bar{u}}_{1}

and

{\bar{u}}_{2}

are the two solutions of Equation (1), in the sense of distribution, we obtain

\begin{matrix} \frac{\partial U}{\partial t} - ν Δ U + (U \cdot \nabla) {\bar{u}}_{1} + ({\bar{u}}_{2} \cdot \nabla) u = f (t, x) + G (t, U (t + θ)) . \end{matrix}

(59)

Next, we prove that under the condition of Theorem 1, if Equation (53) holds, then there is

\begin{matrix} lim_{t \to + \infty} {∥ U (t) ∥}^{2} = 0 . \end{matrix}

(60)

Firstly, the inner product of

A U

with Equation (59), respectively, is obtained with

\begin{matrix} \frac{1}{2} \frac{d}{d t} {∥ \nabla U (t) ∥}^{2} + ν {∥ A U ∥}^{2} \\ = & - b (U, {\bar{u}}_{1}, A U) - b ({\bar{u}}_{2}, U, A U) + (f, A U) + (G (t, U (t + θ)), A U), \end{matrix}

(61)

by the Lemma 1 and Young inequality. Take

ε_{3} = \frac{ν}{4}

, then we can obtain

\begin{matrix} | b (U, u_{1}, A U) | \leq C_{Ω} ∥ \nabla U ∥ ∥ \nabla {\bar{u}}_{1} ∥ ∥ A U ∥ \leq \frac{2 C_{Ω}^{2}}{ν} {∥ \nabla U ∥}^{2} ∥ \nabla {\bar{u}}_{1} ∥^{2} + \frac{ν}{8} {∥ A U ∥}^{2}, \end{matrix}

(62)

\begin{matrix} | b ({\bar{u}}_{2}, U, A U) | \leq C_{Ω} ∥ \nabla {\bar{u}}_{2} ∥ ∥ \nabla U ∥ ∥ A U ∥ \leq \frac{2 C_{Ω}^{2}}{ν} ∥ \nabla {\bar{u}}_{2} ∥^{2} {∥ \nabla U ∥}^{2} + \frac{ν}{8} {∥ A U ∥}^{2} . \end{matrix}

(63)

Similarly, it is obvious that

\begin{matrix} | (f, A u) | \leq \frac{ν}{16} {∥ A U ∥}^{2} + \frac{4}{ν} {∥ f ∥}^{2}, \end{matrix}

(64)

and

\begin{matrix} | (G (t, U (t + θ)), A U) | \leq \frac{ν}{16} {∥ A U ∥}^{2} + \frac{4 L_{G}^{2}}{ν} {∥ U (t + θ) ∥}_{C_{γ} (H)}^{2} . \end{matrix}

(65)

By Equation (24), substituting (62)–(65) into Equation (61), we obtain

\begin{matrix} \frac{d}{d t} {∥ \nabla U (t) ∥}^{2} + \frac{5 ν}{4} {∥ A u ∥}^{2} - \frac{4 C_{Ω}^{2}}{ν} {∥ \nabla U (t) ∥}^{2} (∥ \nabla {\bar{u}}_{1} ∥^{2} + {∥ \nabla {\bar{u}}_{2} ∥}^{2}) \\ \leq & \frac{8 L_{G}^{2}}{ν} e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - τ)} {∥ ϕ (s) ∥}_{C_{γ} (H)}^{2} + \frac{8}{ν} {∥ f ∥}^{2} + 2 {(\frac{4 L_{G}}{ν})}^{2} \int_{τ}^{t} e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - r)} {∥ f (r) ∥}_{V^{'}}^{2} d r . \end{matrix}

(66)

By Equation (55) in Lemma 3, we obtain

\begin{matrix} \frac{d}{d t} {∥ \nabla U (t) ∥}^{2} + [\frac{5 ν λ_{1} N}{4 C_{Ω}} - \frac{4 C_{Ω}^{2}}{ν} (∥ \nabla {\bar{u}}_{1} ∥^{2} + {∥ \nabla {\bar{u}}_{2} ∥}^{2})] {∥ \nabla U (t) ∥}^{2} \\ \leq & \frac{8 L_{G}^{2}}{ν} e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - τ)} {∥ ϕ (s) ∥}_{C_{γ} (H)}^{2} + \frac{8}{ν} {∥ f ∥}^{2} + 2 {(\frac{4 L_{G}}{ν})}^{2} \int_{τ}^{t} e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - r)} {∥ f (r) ∥}_{V^{'}}^{2} d r \\ + \frac{5 ν λ_{1} N^{2}}{4} η^{2} (U) . \end{matrix}

(67)

Secondly, we define

\begin{matrix} \{\begin{matrix} ξ (t) : = {∥ \nabla U (t) ∥}^{2}, \\ P (t) : = \frac{5 ν λ_{1} N}{4 C_{Ω}} - \frac{4 C_{Ω}^{2}}{ν} (∥ \nabla {\bar{u}}_{1} ∥^{2} + {∥ \nabla {\bar{u}}_{2} ∥}^{2}) . \\ Q (t) : = \frac{8 L_{G}^{2}}{ν} e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - τ)} {∥ ϕ (s) ∥}_{C_{γ} (H)}^{2} + \frac{8}{ν} {∥ f ∥}^{2} + \frac{5 ν λ_{1} N^{2}}{4} η^{2} (U) \\ + 2 {(\frac{4 L_{G}}{ν})}^{2} \int_{τ}^{t} e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - r)} {∥ f (r) ∥}_{V^{'}}^{2} d r . \end{matrix} \end{matrix}

Then, the inequality (67) can be expressed as

\begin{matrix} \frac{d ξ (t)}{d t} + P (t) ξ (t) \leq Q (t) . \end{matrix}

(68)

Next, we verify that

P (t)

and

Q (t)

satisfy the conditions of Lemma 4. There exist constants

C_{3}

and

C_{4}

such that

\begin{matrix} \underset{t \to + \infty}{lim sup} \frac{1}{T} \int_{t}^{t + T} {∥ \nabla {\bar{u}}_{i} (r) ∥}^{2} d r \leq C_{5}, i = 1, 2, \end{matrix}

(69)

where

C_{5} : = C_{4} {∥ ϕ (s) ∥}^{2} + C_{3} F^{2}

. Therefore, when

N > \frac{16 C_{Ω}^{3} C_{5}}{5 ν^{2} λ_{1}}

P (r)

and

P^{-} (r)

satisfy

\begin{matrix} \underset{t \to + \infty}{lim inf} \frac{1}{T} \int_{t}^{t + T} P (r) d r \geq & \frac{5 ν λ_{1} N}{4 C_{Ω}} - \frac{4 C_{Ω}^{2}}{ν} \underset{t \to + \infty}{lim sup} \frac{1}{T} \int_{t}^{t + T} {∥ \nabla {\bar{u}}_{i} (r) ∥}^{2} d r \\ \geq & \frac{5 ν λ_{1} N}{4 C_{Ω}} - \frac{4 C_{Ω}^{2} C_{5}}{ν} > 0, \end{matrix}

and

\begin{matrix} \underset{t \to + \infty}{lim sup} \frac{1}{T} \int_{t}^{t + T} P^{-} (r) d r \leq & \frac{5 ν λ_{1} N}{4 C_{Ω}} + \frac{4 C_{Ω}^{2}}{ν} \underset{t \to + \infty}{lim sup} \frac{1}{T} \int_{t}^{t + T} {∥ \nabla {\bar{u}}_{i} (r) ∥}^{2} d r \\ \leq & \frac{5 ν λ_{1} N}{4 C_{Ω}} + \frac{4 C_{Ω}^{2} C_{5}}{ν} < + \infty, \end{matrix}

they satisfy Equation (56).

By Lemma 4, this implies

lim_{t \to + \infty} η (U (t)) = 0 .

Combining them, then we have

\begin{matrix} lim_{t \to + \infty} \frac{1}{T} \int_{t}^{t + T} (\frac{5 ν λ_{1} N^{2}}{4} η^{2} (U (r)) + \frac{8}{ν} {∥ f (r) ∥}^{2}) d r = 0 . \end{matrix}

(70)

Since

ν^{2} λ_{1} > 4 L_{G}

, we obtain

\begin{matrix} lim_{t \to + \infty} \frac{1}{T} \int_{t}^{t + T} e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (r - τ)} {∥ ϕ (s) ∥}_{C_{γ} (H)}^{2} d r \\ = & lim_{t \to + \infty} \frac{1}{T (ν λ_{1} - \frac{4 L_{G}}{ν})} (e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t + T - τ)} - e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - τ)}) {∥ ϕ (s) ∥}_{C_{γ} (H)}^{2} \\ = & 0 . \end{matrix}

According to Equation (58), for all

ε > 0

there exists

T_{1} > τ

such that when

t > T_{1}

, it satisfies

\begin{matrix} \frac{1}{T (ν λ_{1} - \frac{4 L_{G}}{ν})} sup_{z \in [T_{1}, + \infty)} {∥ f (z) ∥}^{2} \int_{τ}^{t + T} [e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - r)} - e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t + T - r)}] d r < \frac{ε}{2}; \end{matrix}

(71)

therefore, for

\forall T_{1} > τ

, we have

\begin{matrix} lim_{t \to + \infty} (e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - r)} - e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t + T - r)}) = 0 . \end{matrix}

(72)

Hence, for the

ε

of Equation (71), there exists

T_{2}

such that when

t > T_{2}

, we have

\begin{matrix} \frac{1}{T (ν λ_{1} - \frac{4 L_{G}}{ν})} \int_{τ}^{T_{0}} {∥ f (r) ∥}^{2} d r [e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - T_{0})} - e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t + T - T_{0})}] < \frac{ε}{2} . \end{matrix}

(73)

Denote

T_{0} = max {T_{1}, T_{2}}

, then for

ε > 0

, when

t > T_{0}

, we have

\begin{matrix} \frac{1}{T} \int_{t}^{t + T} \int_{τ}^{r} e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (ρ - r)} {∥ f (r) ∥}^{2} d ρ d r \\ \leq & \frac{1}{T} \int_{τ}^{t + T} \int_{t}^{t + T} e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (ρ - r)} {∥ f (r) ∥}^{2} d ρ d r \\ = & \frac{1}{T (\frac{4 L_{G}}{ν} - ν λ_{1})} \int_{τ}^{t + T} {∥ f (r) ∥}^{2} \cdot [e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t + T - r)} - e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - r)}] d r \\ \leq & \frac{1}{T (ν λ_{1} - \frac{4 L_{G}}{ν})} \int_{τ}^{T_{0}} {∥ f (r) ∥}^{2} d r \cdot [e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - T_{0})} - e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t + T - T_{0})}] \\ + \frac{1}{T (ν λ_{1} - \frac{4 L_{G}}{ν})} sup_{z \in [T_{0}, + \infty)} {∥ f (z) ∥}^{2} \int_{T_{0}}^{t + T} [e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t - r)} - e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (t + T - r)}] d r \\ < & ε, \end{matrix}

(74)

which implies

\begin{matrix} lim_{t \to + \infty} \frac{1}{T} \int_{t}^{t + T} \int_{τ}^{ρ} e^{- (ν λ_{1} - \frac{4 L_{G}}{ν}) (ρ - r)} {∥ f (r) ∥}^{2} d ρ d r = 0 . \end{matrix}

(75)

By Equations (70) and (75), we can see that

Q^{+} (t)

satisfies Equation (57).

Finally, by Lemma 4, we obtain

lim_{t \to + \infty} ξ (t) = lim_{t \to + \infty} {∥ \nabla U (t) ∥}^{2} = 0 .

According to the Poincare inequation, we have proven Equation (60). According to Definition 2, Theorem 3 is proven. □

5. Conclusions

The problem addressed in this paper is the global well-posedness and asymptotic behavior of solutions to non autonomous Navier–Stokes equations with infinite time delay and node determination. This study establishes a mathematical framework based on the function space

C_{γ} (H)

and demonstrates the well-posedness of Equation (1) under the assumption that the function

G (\cdot, \cdot)

satisfies Lipschitz continuity with respect to time t. Furthermore, the long-term behavior of strong solutions is shown to be characterized by their values at a finite number of spatial nodes. The result of this paper is to apply the infinite delay term

u (t + θ)

to the non-autonomous Navier–Stokes equations and to study the well-posedness and asymptotic behavior of the Navier–Stokes equations with delay effects using relevant existing theoretical results. This result provides theoretical support and specific examples for studying nonlinear PDEs with time delay effects to a certain extent. It not only validates the relevant conclusions obtained from time delay differential equations but also reveals the research methods for nonlinear PDEs of other time delay effects, further verifying some examples given by time delay differential equations (see [23,24,25]).

However, this method also has some limitations, as the construction of

C_{γ} (H)

is only a function space established for handling the infinite delay term

u (t + θ)

, as the asymptotic behavior of function u becomes particularly important at

t \to \infty

and creates a convergence relationship between the infinite delay terms

u (t + θ)

and

u (t)

. In addition, the function

G (\cdot, \cdot)

itself must satisfy Lipschitz properties, otherwise it will greatly affect the well-posedness of the solution of Equation (1). For the physical meaning of the Navier–Stokes equation itself, we only consider the case where the Reynolds number is particularly small to study the global well-posedness and determining nodes of Equation (1). For other cases, due to the need to consider the pressure p term, the method used in this paper may not be applicable.

The research method presented in this paper has high universality and can be applied to various types of nonlinear PDEs. By adding an infinite delay term

u (t + θ)

to the existing conditions of a nonlinear PDE, it becomes a nonlinear PDE with infinite delay, which has a more profound impact on the well-posedness of solutions and other related issues. This provides theoretical support and technical means for solving other complex systems, especially for the study of differential equations with infinite delay. In the current research results, only the system of non-autonomous micro polar fluid flow has been studied for global well-posedness and asymptotic behavior of solutions by adding an infinite delay term

u (t + θ)

on the original basis (see [32]).

Due to the existence of time delay factors, the introduction and application of numerical methods need to consider the influence of historical states, which leads to a more complex implementation of the algorithm. In order to solve Navier–Stokes equations with time delay, it is generally necessary to discretize them. Furthermore, determining nodes are the discrete points determined during this process. In the process of analyzing problems, it is usually necessary to calculate analytical and numerical solutions at determined nodes and to strictly control the time step based on the introduced time delay factors in order to avoid the instability of numerical solutions and evaluate the accuracy of numerical methods, ensuring the stability and convergence of the algorithm. By comparing these solutions, we can assess the effectiveness of the selected nodes and discretization methods, thereby improving the efficiency and accuracy of solving Navier–Stokes equations with time delays, ensuring that researchers can better simulate and control fluid systems, and enhance the reliability and performance of engineering design.

Advancements in computational technology open new avenues for research on Navier–Stokes equations with time delay, particularly in exploring time-delay effects in high-dimensional spaces and complex geometries. This includes the behavior of fluids under complex boundary conditions or multiphase flow, particularly in applications such as biomedical engineering and environmental science. In recent years, machine learning has gained prominence in fluid dynamics, offering novel approaches to modeling and real-time control of fluid systems. Integrating time-delay factors with machine learning methods presents a promising direction for developing more accurate fluid models and real-time control strategies, advancing both theoretical understanding and practical applications. These methods can be used to learn fluid behavior from experimental data and predict the system’s response under different conditions.

In addition, for delay differential equations, there are various types of delay effects such as discrete delay and state-dependent delay. Future research will focus on investigating the global well-posedness and asymptotic behavior of solutions to differential equations with state-dependent delay

u (t - σ (t, u (t + θ)))

and generalize the relevant conclusions of abstract functional differential equations with state-dependent delay in reference [20] through several specific partial differential equations. The manuscript on the study of partial differential equations with state-dependent delay is currently in progress.

Author Contributions

Conceptualization, H.G.; methodology, H.G.; validation, H.G. and F.D.; writing—original draft, H.G.; writing—review and editing, H.G. and F.D. All authors have read and agreed to the published version of the manuscript.

Funding

This work is supported by the Research Project of Jingchu University of Technology (Grant Nos. HX20240049), and the NSF of Hubei Province (Grant No. 2022CFB527).

Data Availability Statement

Data sharing is not applicable to this article, as no datasets were generated or analyzed during the current study.

Conflicts of Interest

The authors declare no conflicts of interest.

References

Ferreira, L.C.F.; Precioso, J.C. Existence of solutions for the 3D-micropolar fluid system with initial data in Besov–Morrey spaces. Commun. Math. Sci. 2013, 64, 1699–1710. [Google Scholar] [CrossRef]
Galdi, P.; Rionero, S. A note on the existence and uniqueness of solusions of the micropolar fluid equations. Int. J. Eng. Sci. 1997, 15, 105–108. [Google Scholar] [CrossRef]
Lukaszewicz, G. Micropolar Fluids: Theory and Applications, Modeling and Simulation in Science; Birkhauser: Boston, MA, USA, 1999; pp. 177–215. [Google Scholar]
Lukaszewicz, G.; Tarasinska, A. On H¹-pullback attractors for nonautonomous micropolar fluid equations in a bounded domain. Nonlinear Anal.-TMA 2009, 71, 782–788. [Google Scholar] [CrossRef]
Foias, C.; Manley, O.; Rosa, R.; Temam, R. Navier-Stokes Equations and Turbulence; Encyclopedia of Mathematics and Its Application: Cambridge, UK, 2004; pp. 255–343. [Google Scholar]
Caraballo, T.; Lukaszewicz, G.; Real, J. Pullback attractors for non-autonomous 2D-Navier-Stokes equations in several unbounded domains. Dynam. Syst. 2006, 342, 263–268. [Google Scholar]
García-Luengo, J.; Marín-Rubio, P.; Real, J. H²-boundedness of the pullback attractors for non-autonomous 2D Navier-Stokes equations in bounded domains. Nonlinear Anal. 2011, 74, 4882–4887. [Google Scholar] [CrossRef]
García-Luengo, J.; Marín-Rubio, P.; Real, J. Pullback attractors in V for non-autonomous 2D-Navier-Stokes equations and their tempered behaviour. J. Differ. Equ. 2012, 252, 4333–4356. [Google Scholar] [CrossRef]
García-Luengo, J.; Marín-Rubio, P.; Real, J. Pullback attractors for 2D Navier-Stokes equations with delays and their regularity. Adv. Nonlinear Stud. 2013, 13, 331–357. [Google Scholar] [CrossRef]
Hoff, D.; Ziane, M. The global attractor and finite determining nodes for the Navier-Stokes equations of compressible flow with singular initial data. Indiana U. Math. J. 2003, 49, 843–890. [Google Scholar]
Jolly, M.S.; Sadigoy, T.; Titi, E.S. A determining form for the damped driven nonlinear Schröinger equation—Fourier modes case. J. Differ. Equ. 2015, 258, 2711–2744. [Google Scholar] [CrossRef]
Kakizawa, R. Determining nodes for semilinear parabolic equations. J. Math. Anal. Appl. 2011, 378, 375–386. [Google Scholar] [CrossRef]
Marín-Rubio, P.; Real, J. Attractors for 2D-Navier-Stokes equations with delays on several unbounded domains. Nonlinear Anal.-TMA 2007, 67, 2784–2799. [Google Scholar] [CrossRef]
Marín-Rubio, P.; Real, J. Pullback attractors for 2D-Navier-Stokes equations with delays in continuous and sub-linear operators. Discrete Cont. Dyn.-A 2010, 26, 989–1006. [Google Scholar] [CrossRef]
Szopa, P. Determining modes for 2D micropolar fluid flows. Math. Comput. Model. 2004, 42, 1079–1088. [Google Scholar] [CrossRef]
Marín-Rubio, P.; Real, J.; Valero, J. Pullback attractors for a two-dimensional Navier-Stokes model in an infinite delay case. Nonlinear Anal.-Theor. 2011, 74, 2012–2030. [Google Scholar] [CrossRef]
Ning, Z.; Shen, C.; Zhao, X.; Li, H.; Lin, C.; Zhang, Y. Nonlinear boundary stabilization of the wave equations with variable coefficients and time dependent delay. Appl. Math. Comput. 2014, 232, 511–520. [Google Scholar] [CrossRef]
Caraballo, T.; Real, J. Attractors for 2D-Navier-Stokes models with delays. J. Differ. Equ. 2004, 205, 271–297. [Google Scholar] [CrossRef]
Hale, J.K.; Verduyn Lunel, S.M. Introduction to Functional Differential Equations; Springer: New York, NY, USA, 1993; pp. 144–195. [Google Scholar]
Hernandez, E.; Fernandes, D.; Wu, J. Existence and uniqueness of solutions, well-posedness and global attractor for abstract differential equations with state-dependent delay. J. Differ. Equ. 2021, 302, 753–806. [Google Scholar] [CrossRef]
Chueshov, I. Finite-Dimensional global attractors for parabolic nonlinear equations with state-dependent dealy. Commun. Pur. Appl. Anal. 2015, 14, 1685–1704. [Google Scholar] [CrossRef]
Hernandez, E.; Pierri, M.; Wu, J. Existence and uniqueness of C^1+α-strict solutions for integro-differential equations with state-dependent delay. J. Differ. Equ. 2016, 261, 6856–6882. [Google Scholar] [CrossRef]
Fu, X.; Liu, X. Existence of periodic solutions for abstract neutral non-autonomous equations with infinite delay. J. Math. Anal. Appl. 2007, 325, 249–267. [Google Scholar] [CrossRef]
Chang, J. Solutions to non-autonomous integrodifferential equations with infinite delay. J. Math. Anal. Appl. 2007, 331, 137–151. [Google Scholar] [CrossRef]
Huang, A.; Huo, W.; Ma, J. Finite-Dimensionality and Determining Modes of the Global Attractor for 2D Boussinesq Equations with Fractional Laplacian. Adv. Nonlinear Stud. 2018, 18, 501–515. [Google Scholar] [CrossRef]
Lukaszewicz, G. Long time behavior of 2D micropolar fluid flows. Math. Comp. Model. Dyn. 2001, 34, 487–509. [Google Scholar] [CrossRef]
Lukaszewicz, G.; Sadowski, W. Uniform attractor for 2D magneto-micropolar fluid flow in several unbounded domains. Z. Angew. Math. Phys. 2004, 55, 247–257. [Google Scholar]
Jones, D. On the Number of Determining Nodes for the 2D Navier-Stokes Equations. J. Math. Anal. Appl. 1992, 168, 72–88. [Google Scholar] [CrossRef]
Foias, C.; Temam, R. Determination of the solutions of the Navier-Stokes equations by a set of nodal values. Math. Comput. 1984, 43, 117–133. [Google Scholar] [CrossRef]
Lu, Y. Uniform attractors for the closed process and applications to the reaction-diffusion equation with dynamical boundary condition. Nonlinear Anal.-Real. 2009, 71, 4012–4025. [Google Scholar] [CrossRef]
Sun, W.; Lai, C.; Liang, Y. Determining modes and determining nodes for the 3D non-autonomous regularized magnetohydrodynamics equations. Discrete Contin. Dyn. Syst. Ser. B 2023, 28, 3373–3392. [Google Scholar] [CrossRef]
Zhao, C.; Sun, W. Global well-posedness and pullback attractors for a two-dimensional non-autonomous micropolar fluid flows with infnite delays. Commun. Math. Sci. 2017, 15, 97–121. [Google Scholar] [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

Share and Cite

MDPI and ACS Style

Ge, H.; Du, F. Global Well-Posedness and Determining Nodes of Non-Autonomous Navier–Stokes Equations with Infinite Delay on Bounded Domains. Mathematics 2025, 13, 222. https://doi.org/10.3390/math13020222

AMA Style

Ge H, Du F. Global Well-Posedness and Determining Nodes of Non-Autonomous Navier–Stokes Equations with Infinite Delay on Bounded Domains. Mathematics. 2025; 13(2):222. https://doi.org/10.3390/math13020222

Chicago/Turabian Style

Ge, Huanzhi, and Feng Du. 2025. "Global Well-Posedness and Determining Nodes of Non-Autonomous Navier–Stokes Equations with Infinite Delay on Bounded Domains" Mathematics 13, no. 2: 222. https://doi.org/10.3390/math13020222

APA Style

Ge, H., & Du, F. (2025). Global Well-Posedness and Determining Nodes of Non-Autonomous Navier–Stokes Equations with Infinite Delay on Bounded Domains. Mathematics, 13(2), 222. https://doi.org/10.3390/math13020222

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Menu

Global Well-Posedness and Determining Nodes of Non-Autonomous Navier–Stokes Equations with Infinite Delay on Bounded Domains

Abstract

1. Introduction

2. Preliminaries

2.1. Basic Function Spaces and Conclusions

2.2. The Weak Solution and Strong Solution

3. Global Well-Posedness

3.1. The Existence of Solutions

3.2. The Uniqueness of Solutions

4. Estimate the Number of Determining Nodes

4.1. Introduction to Relevant Lemmas

4.2. Main Results

5. Conclusions

Author Contributions

Funding

Data Availability Statement

Conflicts of Interest

References

Share and Cite

Article Metrics

Article Access Statistics

Further Information

Guidelines

MDPI Initiatives

Follow MDPI