betaFeedback

WorksMathematics

On energy equality of the 3D anisotropic Navier-Stokes equations

Abstract

In the spirit of the recent work by Demmel and Wiedemann \cite{[DW]}, we investigate the validity of the energy balance law for weak solutions to the three-dimensional Navier-Stokes equations with only horizontal dissipation. We prove that famous Lions-Shinbrot type energy conservation criteria, originally established for the classical isotropic Navier-Stokes equations, continue to hold in this anisotropic setting. Furthermore, we present two distinct approaches that handle the torus case and the whole space case, respectively.

Introduction

In geophysical fluids, the Navier-Stokes equations feature anisotropic viscosities of the form −νΔh−ϵ∂332-\nu\Delta_h-\epsilon\partial_{33}^{2} (see [34]). Here, ϵ\epsilon is usually much smaller than ν\nu, and the Ekman boundary layers arise in rotating fluid when ϵ→0\epsilon\to0 (see [34]). In this paper, we are concerned with the following 3D Navier-Stokes equations subject solely to horizontal dissipation

{ut+u⋅∇u−Δhu+∇Π=0,div⁡u=0,u∣t=0=u0(x),(1)\begin{cases} u_t+u\cdot\nabla u-\Delta_hu+\nabla\Pi=0,\quad\operatorname{div}u=0,\\ u|_{t=0}=u_0(x), \tag*{(1)} \end{cases}

where uu represents velocity field and Π\Pi stands for the pressure, and Δhu=∂x1x12u+∂x2x22u\Delta_hu=\partial_{x_1x_1}^{2}u+\partial_{x_2x_2}^{2}u. The initial velocity u0u_0 satisfies div⁡u0=0\operatorname{div}u_0=0.

Very recently, Demmel and Wiedemann [18] proved the energy equality for weak solutions of the two-dimensional Navier-Stokes equations with only horizontal viscosity at the natural energy level, where weak solutions are lack of the regularity of ∂2u1\partial_2u_1 in the 2D anisotropic incompressible Navier-Stokes system. To this end, some new ingredients were provided in [18], where they split the energy equality into vertical component part and horizontal component part. In addition, the divergence-free condition is fully utilized and an anisotropic commutator estimate of the DiPerna-Lions theory for divergence-free field was established. In the spirit of Demmel and Wiedemann’s work [18], we study the validity of energy balance law

12∫Ω∣u(x,t)∣2 dx+∫0t∥∇ ⁣hu∥L22 ds=12∫Ω∣u0(x)∣2 dx(2)\frac{1}{2}\int_{\Omega} |u(x,t)|^{2} \, dx + \int_{0}^{t} \|\nabla_{\!h}u\|_{L^{2}}^{2} \, ds = \frac{1}{2}\int_{\Omega} |u_{0}(x)|^{2} \, dx \tag*{(2)}

for weak solutions of the tri-dimensional anisotropic Navier-Stokes equations (1), where Ω\Omega is either the whole space R3\mathbb{R}^{3} or the periodic domain T3\mathbb{T}^{3}. Before formulating our results, we recall the progress of energy equality

12∥u(T)∥L2(R3)2+∫0T∥∇u∥L2(R3)2 ds=12∥u0∥L2(R3)2(3)\frac{1}{2}\|u(T)\|_{L^{2}(\mathbb{R}^{3})}^{2}+\int_{0}^{T}\|\nabla u\|_{L^{2}(\mathbb{R}^{3})}^{2} \, ds=\frac{1}{2}\|u_{0}\|_{L^{2}(\mathbb{R}^{3})}^{2} \tag*{(3)}

for weak solutions of the tri-dimensional isotropic Navier-Stokes equations.

Energy conservation of weak solutions not only plays a crucial role in the study of weak-strong uniqueness, but also is closed to the Onsager conjecture. Onsager [33] asserted that the regularity threshold in Hölder spaces of the weak solutions for the validity of the kinetic energy conservation in the ideal fluid is 1/31/3. Since then, much effort has been devoted to identifying the minimal regularity assumptions that guarantee energy equality for weak solutions to viscous and inviscid flows, see, e.g. [2, 3, 4, 5, 6, 7, 8, 15, 16, 21, 22, 24, 30, 31, 35, 44] and references therein. Concerning the negative part of the Onsager’s conjecture, we refer the reader to [19, 20, 25, 27, 28]. In what follows, we restrict our attention to the energy equality and recall some known results in this direction below: A Leray-Hopf weak solution uu to the Navier-Stokes equations satisfies the energy equality (3) if one of the following conditions holds

  • Lions [31]: u∈L4(0,T;L4(R3))u\in L^{4}(0,T;L^{4}(\mathbb{R}^{3}));

  • Shinbrot [35]:

u∈Lp(0,T;Lq(R3)), with 2p+2q=1 and q≥4;(4)u\in L^{p}(0,T;L^{q}(\mathbb{R}^{3})),\ \text{with }\frac{2}{p}+\frac{2}{q}=1\ \text{and }q\geq4; \tag*{(4)}
  • Taniuchi [36], Beirao da Veiga-Yang [2]:

u∈Lp(0,T;Lq(R3)), with 1p+3q=1 and 3<q<4;(5)u\in L^{p}(0,T;L^{q}(\mathbb{R}^{3})),\ \text{with }\frac{1}{p}+\frac{3}{q}=1\ \text{and }3<q<4; \tag*{(5)}
  • Cheskidov-Constantin-Friedlander-Shvydkoy [15]: u∈L3(0,T;B3,∞13(R3))u\in L^{3}(0,T;B_{3,\infty}^{\frac{1}{3}}(\mathbb{R}^{3}));

  • Cheskidov-Luo [16]: u∈Lβ,∞(0,T;Bp,∞2p+2β−1(R3))u\in L^{\beta,\infty}(0,T;B_{p,\infty}^{\frac{2}{p}+\frac{2}{\beta}-1}(\mathbb{R}^{3})), with 2p+1β<1\frac{2}{p}+\frac{1}{\beta}<1 and 1≤β<p≤∞1\leq\beta<p\leq\infty;

  • Berselli-Chiodaroli [5], Beirao da Veiga-Yang [3], Zhang [44]:

∇u∈Lp(0,T;Lq(R3)), with 1p+3q=2 and 32<q<95, or 1p+65q=1 and q≥95.(6)\nabla u\in L^{p}(0,T;L^{q}(\mathbb{R}^{3})),\ \text{with }\frac{1}{p}+\frac{3}{q}=2\ \text{and }\frac{3}{2}<q<\frac{9}{5},\ \text{or }\frac{1}{p}+\frac{6}{5q}=1\ \text{and }q\geq\frac{9}{5}. \tag*{(6)}

Inspired by Demmel and Wiedemann’s work [18], a natural question arises whether the well-known properties of the isotropic Navier-Stokes equations continue to hold in this anisotropic setting. Compared with the isotropic counterpart, an extra difficulty stems from the lack of regularity for ∂3uh\partial_{3}u_{h} in the anisotropic Navier-Stokes equations (1), which further complicates the analysis of the anisotropic Navier-Stokes equations. For mathematical results on the anisotropic Navier-Stokes system, we refer the reader to [9, 10, 11, 12, 13, 23, 28, 39, 40, 45, 47] and the references therein. In the present work, we concentrate on the energy equality problem and aim to address whether classical Lions-Shinbrot type criteria for kinetic energy conservation remain valid for the three-dimensional anisotropic Navier-Stokes equations (1).

In what follows, we formulate our criteria for kinetic energy conservation of Leray-Hopf weak solutions to the anisotropic Navier-Stokes equations (1) with partial dissipation for the torus domain T3\mathbb{T}^{3}.

Theorem 1.1. The energy equality of Leray-Hopf weak solutions uu to the 3D anisotropic Navier-Stokes equations (1) is valid if one of the following three conditions is satisfied

  1. u∈L4(0,T;L4(T3))u \in L^{4}(0,T;L^{4}(\mathbb{T}^{3}));

  1. u∈Lp(0,T;Lq(T3))u \in L^{p}(0,T;L^{q}(\mathbb{T}^{3})), with 2p+2q=1\frac{2}{p}+\frac{2}{q}=1 and q≥4q \ge4;

  1. ∇u∈Lp(0,T;Lq(T3))\nabla u \in L^{p}(0,T;L^{q}(\mathbb{T}^{3})), with 1p+65q=1\frac{1}{p}+\frac{6}{5q}=1 and q≥95q \ge\frac{9}{5}.

Now we give some comments. The tri-dimensional anisotropic Navier-Stokes equations (1) are lack of vertical dissipation ∂x3x32u\partial_{x_{3}x_{3}}^{2}u, however, the famous Lions-Shinbrot’s energy conservation class originated from the standard Navier-Stokes equations is still valid for this system. In addition, energy conservation criteria in terms of gradient of velocity obtained in [3, 5, 44] are established for the tri-dimensional anisotropic Navier-Stokes equations (1). It should be pointed out that the proof of Theorem 1.1 is inspired by Demmel and Wiedemann’s recent work [18]. Compared with their deduction, the derivation of this theorem rests on the limiting case of Constantin-E-Titi type commutators shown in [32, 38] for the torus case. The starting point of Theorem 1.1 is different from that in [18], hence, we get the energy identity under the full components of velocity belonging to L4(0,T;L4(T3))L^{4}(0,T;L^{4}(\mathbb{T}^{3})) in one step rather than two steps in [18]. Roughly speaking, Demmel and Wiedemann proved that weak solutions satisfy the energy equality from u2∈L4(0,T;L4(R2))u_{2} \in L^{4}(0,T;L^{4}(\mathbb{R}^{2})) and u2u1∈L2(0,T;L2(R2))u_{2}u_{1} \in L^{2}(0,T;L^{2}(\mathbb{R}^{2})) in [18], where the delicate pressure decomposition and directionwise energy splitting are required. Besides, Demmel and Wiedemann’s anisotropic commutator estimates established in [18] allow us to get the same results for the whole space R3\mathbb{R}^{3}. Then, we state our second result as follows.

Theorem 1.2. The energy equality of Leray-Hopf weak solutions uu to the 3D anisotropic Navier-Stokes equations (1) is valid if one of the following three conditions is satisfied

  1. u∈L4(0,T;L4(R3))u \in L^{4}(0,T;L^{4}(\mathbb{R}^{3}));

  1. u∈Lp(0,T;Lq(R3))u \in L^{p}(0,T;L^{q}(\mathbb{R}^{3})), with 2p+2q=1\frac{2}{p}+\frac{2}{q}=1 and q≥4q \ge4;

  1. ∇u∈Lp(0,T;Lq(R3))\nabla u \in L^{p}(0,T;L^{q}(\mathbb{R}^{3})), with 1p+65q=1\frac{1}{p}+\frac{6}{5q}=1 and q≥95q \ge\frac{9}{5}.

It should be pointed out that the Lions’s energy conservation class u∈L4(0,T;L4(R2))u \in L^{4}(0,T;L^{4}(\mathbb{R}^{2})) in Theorem 1.2 is still valid for the two dimensional anisotropic Navier-Stokes equations via a slight modification. It is worth remarking that that Demmel and Wiedemann obtained energy conservation of weak solutions of the two-dimensional anisotropic Navier-Stokes equations without any extra integrability assumptions in [18].

The rest of this paper is organized as follows. In Section 2, we present the notations and some basic materials of mollifiers. Several critical lemmas for the proof of Theorem 1.1 and Theorem 1.2 are also given. Section 3 is devoted to our new criteria for energy conservation of weak solutions to the anisotropic Navier-Stokes equations for both the periodic domain T3\mathbb{T}^{3} and the whole space R3\mathbb{R}^{3}.

Notations and key auxiliary lemmas

Throughout this paper, we will use the summation convention on repeated indices. CC will denote positive absolute constants which may be different from line to line unless otherwise stated in this paper. For p∈[1,∞]p \in[1,\infty], the notation Lp(0,T;X)L^{p}(0,T;X) stands for the set of measurable functions ff on the interval (0,T)(0,T) with values in XX and ∥f∥X\|f\|_{X} belonging to Lp(0,T)L^{p}(0,T).

Mollifier kernel: Let ηε:Rn→R\eta_{\varepsilon}:\mathbb{R}^{n}\to\mathbb{R} be a standard mollifier, i.e. η(x)=C0e−11−∣x∣2\eta(x)=C_{0}e^{-\frac{1}{1-|x|^{2}}} for ∣x∣<1|x|<1 and η(x)=0\eta(x)=0 for ∣x∣≥1|x|\geq1, where C0C_{0} is a constant such that ∫Rnη(x) dx=1\int_{\mathbb{R}^{n}}\eta(x)\,dx=1. For ε>0\varepsilon>0, we define the rescaled mollifier ηε(x)=1εnη(xε)\eta_{\varepsilon}(x)=\frac{1}{\varepsilon^{n}}\eta\left(\frac{x}{\varepsilon}\right) and for any function f∈Lloc1(Rn)f\in L^{1}_{\mathrm{loc}}(\mathbb{R}^{n}), its mollified version is defined as

fε(x)=(f∗ηε)(x)=∫Rnf(x−y)ηε(y) dy,x∈Rn.f^{\varepsilon}(x)=(f*\eta_{\varepsilon})(x)=\int_{\mathbb{R}^{n}}f(x-y)\eta_{\varepsilon}(y)\,dy,\quad x\in\mathbb{R}^{n}.

Next, we collect some lemmas which will be used in the present paper.

Lemma 2.1. ([42]) Let p,q,p2,q2∈[1,+∞)p,q,p_{2},q_{2}\in[1,+\infty) and p1,q1∈[1,+∞]p_{1},q_{1}\in[1,+\infty] with 1p=1p1+1p2\frac{1}{p}=\frac{1}{p_{1}}+\frac{1}{p_{2}}, 1q=1q1+1q2\frac{1}{q}=\frac{1}{q_{1}}+\frac{1}{q_{2}}. Assume f∈Lp1(0,T;Lq1(Ω))f\in L^{p_{1}}(0,T;L^{q_{1}}(\Omega)) and g∈Lp2(0,T;Lq2(Ω))g\in L^{p_{2}}(0,T;L^{q_{2}}(\Omega)), then there holds

lim⁡ε→0∥(fg)ε−fεgε∥Lp(0,T;Lq(Ω))=0.(7)\lim_{\varepsilon\to0}\|(fg)^{\varepsilon}-f^{\varepsilon}g^{\varepsilon}\|_{L^{p}(0,T;L^{q}(\Omega))}=0. \tag*{(7)}

Lemma 2.2. ([32]) Suppose that f∈Lp(0,T;Lq(Tn))f\in L^{p}(0,T;L^{q}(\mathbb{T}^{n})). Then for any ε>0\varepsilon>0, there holds

∥∇fε∥Lp(0,T;Lq(Tn))≤Cε−1∥f∥Lp(0,T;Lq(Tn)),(8)\|\nabla f^{\varepsilon}\|_{L^{p}(0,T;L^{q}(\mathbb{T}^{n}))}\leq C\varepsilon^{-1}\|f\|_{L^{p}(0,T;L^{q}(\mathbb{T}^{n}))}, \tag*{(8)}

and, if p,q<∞p,q<\infty

lim sup⁡ε→0ε∥∇fε∥Lp(0,T;Lq(Tn))=0.(9)\limsup_{\varepsilon\to0}\varepsilon\|\nabla f^{\varepsilon}\|_{L^{p}(0,T;L^{q}(\mathbb{T}^{n}))}=0. \tag*{(9)}

Lemma 2.3. ([32, 38]) Let 1≤p,q,p1,p2,q1,q2≤∞1\leq p,q,p_{1},p_{2},q_{1},q_{2}\leq\infty with 1p=1p1+1p2\frac{1}{p}=\frac{1}{p_{1}}+\frac{1}{p_{2}} and 1q=1q1+1q2\frac{1}{q}=\frac{1}{q_{1}}+\frac{1}{q_{2}}. Assume that f∈Lp1(0,T;W1,q1(Tn))f\in L^{p_{1}}(0,T;W^{1,q_{1}}(\mathbb{T}^{n})) and g∈Lp2(0,T;Lq2(Tn))g\in L^{p_{2}}(0,T;L^{q_{2}}(\mathbb{T}^{n})). Then for any ε>0\varepsilon>0, there holds

∥(fg)ε−fεgε∥Lp(0,T;Lq(Tn))≤Cε∥f∥Lp1(0,T;W1,q1(Tn))∥g∥Lp2(0,T;Lq2(Tn)).(10)\|(fg)^{\varepsilon}-f^{\varepsilon}g^{\varepsilon}\|_{L^{p}(0,T;L^{q}(\mathbb{T}^{n}))}\leq C\varepsilon\|f\|_{L^{p_{1}}(0,T;W^{1,q_{1}}(\mathbb{T}^{n}))}\|g\|_{L^{p_{2}}(0,T;L^{q_{2}}(\mathbb{T}^{n}))}. \tag*{(10)}

Moreover, if p2,q2<∞p_{2},q_{2}<\infty, then

lim sup⁡ε→0ε−1∥(fg)ε−fεgε∥Lp(0,T;Lq(Tn))=0.(11)\limsup_{\varepsilon\to0}\varepsilon^{-1}\|(fg)^{\varepsilon}-f^{\varepsilon}g^{\varepsilon}\|_{L^{p}(0,T;L^{q}(\mathbb{T}^{n}))}=0. \tag*{(11)}

The next lemma was established in [18] for the two-dimensional setting R2\mathbb{R}^{2} with the special exponents pi=qi=2p_i=q_i=2 for i=1,2i=1,2. Now, we generalize the result by adopting space-time integral norms with distinct exponents and extending the spatial dimension from two to three.

Lemma 2.4. Assume that the exponents satisfy

1≤p1,q1≤∞,1≤p,q,pk,qk<∞,k=2,3,4,1 \le p_1,q_1 \le\infty,\quad1 \le p,q,p_k,q_k < \infty,\quad k=2,3,4,

and the scaling relations

1p=1p1+1p2=1p3+1p4and1q=1q1+1q2=1q3+1q4.\frac{1}{p}=\frac{1}{p_1}+\frac{1}{p_2}=\frac{1}{p_3}+\frac{1}{p_4}\quad\text{and}\quad\frac{1}{q}=\frac{1}{q_1}+\frac{1}{q_2}=\frac{1}{q_3}+\frac{1}{q_4}.

Suppose that

∂if∈Lp2(0,T;Lq2(R3)),andf∈Lp4(0,T;Lq4(R3)),\partial_i f \in L^{p_2}(0,T;L^{q_2}(\mathbb{R}^{3})),\quad\text{and}\quad f \in L^{p_4}(0,T;L^{q_4}(\mathbb{R}^{3})),

and the divergence-free vector field uu fulfills

u∈Lp1(0,T;Lq1(R3)),and∂iu∈Lp3(0,T;Lq3(R3)),u \in L^{p_1}(0,T;L^{q_1}(\mathbb{R}^{3})),\quad\text{and}\quad\partial_i u \in L^{p_3}(0,T;L^{q_3}(\mathbb{R}^{3})),

for all i=1,2,⋯ ,n−1i=1,2,\cdots,n-1. Then,

div⁡[(uf)ε−ufε]→0strongly in Lp(0,T;Lq(R3)),as ε→0.\operatorname{div}[(uf)^{\varepsilon}-uf^{\varepsilon}] \to0 \quad\text{strongly in } L^{p}(0,T;L^{q}(\mathbb{R}^{3})),\quad\text{as }\varepsilon\to0.

Proof. First, by direct computation, we deduce that

div⁡(uf)ε−ufε=∑i=1n−1∂i[(uif)ε−uifε]+∂n[(unf)ε−unfε]=∑i=1n−1[∂i(uif)ε−ui∂ifε]+[∂n(unf)ε−un∂nfε]=∑i=1n−1[(∂iuif)ε+(ui∂if)ε−ui∂ifε]+[∂n(unf)ε−un∂nfε]=R1ε+R2ε,(12)\begin{aligned} \operatorname{div}(uf)^{\varepsilon}-uf^{\varepsilon} &=\sum_{i=1}^{n-1}\partial_i[(u_i f)^{\varepsilon}-u_i f^{\varepsilon}]+\partial_n[(u_n f)^{\varepsilon}-u_n f^{\varepsilon}]\\ &=\sum_{i=1}^{n-1}[\partial_i(u_i f)^{\varepsilon}-u_i\partial_i f^{\varepsilon}]+[\partial_n(u_n f)^{\varepsilon}-u_n\partial_n f^{\varepsilon}]\\ &=\sum_{i=1}^{n-1}[(\partial_i u_i f)^{\varepsilon}+(u_i\partial_i f)^{\varepsilon}-u_i\partial_i f^{\varepsilon}]+[\partial_n(u_n f)^{\varepsilon}-u_n\partial_n f^{\varepsilon}]\\ &=R_1^{\varepsilon}+R_2^{\varepsilon}, \tag*{(12)} \end{aligned}

where the divergence-free condition div⁡u=0\operatorname{div}u=0 has been used.

For the first term R1εR_1^{\varepsilon}, it can be reformulated as

R1ε=∑i=1n−1[(∂iuif)ε+(ui∂if)ε−ui∂ifε]=∑i=1n−1(∂iuif)ε+∑i=1n−1[(ui∂if)ε−(ui∂if)+(ui∂if)−ui∂ifε].(13)\begin{aligned} R_1^{\varepsilon} &=\sum_{i=1}^{n-1}[(\partial_i u_i f)^{\varepsilon}+(u_i\partial_i f)^{\varepsilon}-u_i\partial_i f^{\varepsilon}]\\ &=\sum_{i=1}^{n-1}(\partial_i u_i f)^{\varepsilon}+\sum_{i=1}^{n-1}[(u_i\partial_i f)^{\varepsilon}-(u_i\partial_i f)+(u_i\partial_i f)-u_i\partial_i f^{\varepsilon}]. \tag*{(13)} \end{aligned}

Regarding the second term on the right-hand side of (2.7), it follows from the Hölder inequality that

∥ui∂if∥Lp(0,T;Lq(R3))≤∥ui∥Lp1(0,T;Lq1(R3))∥∂if∥Lp2(0,T;Lq2(R3)),(14)\|u_i\partial_i f\|_{L^{p}(0,T;L^{q}(\mathbb{R}^{3}))}\leq\|u_i\|_{L^{p_1}(0,T;L^{q_1}(\mathbb{R}^{3}))}\|\partial_i f\|_{L^{p_2}(0,T;L^{q_2}(\mathbb{R}^{3}))}, \tag*{(14)}

which together with the standard properties of the mollification yields

∥(ui∂if)ε−ui∂ifε∥Lp(0,T;Lq(R3))→0,as ε→0,(15)\left\|(u_i\partial_i f)^\varepsilon-u_i\partial_i f^\varepsilon\right\|_{L^p(0,T;L^q(\mathbb{R}^3))}\to0,\quad\text{as }\varepsilon\to0, \tag*{(15)}

and

∥(ui∂if)−ui∂ifε∥Lp(0,T;Lq(R3))≤∥ui∥Lp1(0,T;Lq1(R3))∥∂ifε−∂if∥Lp2(0,T;Lq2(R3))→0,as ε→0,(16)\begin{aligned} \left\|(u_i\partial_i f)-u_i\partial_i f^\varepsilon\right\|_{L^p(0,T;L^q(\mathbb{R}^3))} &\leq\|u_i\|_{L^{p_1}(0,T;L^{q_1}(\mathbb{R}^3))}\|\partial_i f^\varepsilon-\partial_i f\|_{L^{p_2}(0,T;L^{q_2}(\mathbb{R}^3))}\\ &\to0,\quad\text{as }\varepsilon\to0, \tag*{(16)} \end{aligned}

provided that 1≤p,q,p1,q1≤∞1\leq p,q,p_1,q_1\leq\infty, 1≤p2,q2<∞1\leq p_2,q_2<\infty satisfying 1p=1p1+1p2\frac{1}{p}=\frac{1}{p_1}+\frac{1}{p_2} and 1q=1q1+1q2\frac{1}{q}=\frac{1}{q_1}+\frac{1}{q_2}.

Consequently, in combination with (13), (15) and (16), we conclude that

R1ε→∑i=1n−1∂iuifstrongly in Lp(0,T;Lq(R3)),as ε→0.(17)R_1^\varepsilon\to\sum_{i=1}^{n-1}\partial_i u_i f\quad\text{strongly in }L^p(0,T;L^q(\mathbb{R}^3)),\quad\text{as }\varepsilon\to0. \tag*{(17)}

For the second term R2εR_2^\varepsilon, in light of the mean value theorem, we get

R2ε=∂n(unf)ε−un∂nfε=∫R3∂ynηε(y)(un(x−y)−un(x))f(x−y) dy=1ε∫R3∂znη(z)(un(x−εz)−un(x))f(x−εz) dz=−1ε∫R3∂znη(z)∫01εz⋅∇un(x−θεz) dθf(x−εz) dz=−∑i=1n∫R3∂znη(z)zi∫01∂iun(x−θεz) dθf(x−εz) dz.(18)\begin{aligned} R_2^\varepsilon&=\partial_n(u_n f)^\varepsilon-u_n\partial_n f^\varepsilon\\ &=\int_{\mathbb{R}^3}\partial_{y_n}\eta_\varepsilon(y)\bigl(u_n(x-y)-u_n(x)\bigr)f(x-y)\,dy\\ &=\frac{1}{\varepsilon}\int_{\mathbb{R}^3}\partial_{z_n}\eta(z)\bigl(u_n(x-\varepsilon z)-u_n(x)\bigr)f(x-\varepsilon z)\,dz\\ &=-\frac{1}{\varepsilon}\int_{\mathbb{R}^3}\partial_{z_n}\eta(z)\int_0^1\varepsilon z\cdot\nabla u_n(x-\theta\varepsilon z)\,d\theta f(x-\varepsilon z)\,dz\\ &=-\sum_{i=1}^{n}\int_{\mathbb{R}^3}\partial_{z_n}\eta(z)z_i\int_0^1\partial_i u_n(x-\theta\varepsilon z)\,d\theta f(x-\varepsilon z)\,dz. \tag*{(18)} \end{aligned}

On the one hand, by virtue of the Minkowski inequality and the Hölder inequality, we deduce that

∥∂n(unf)ε−un∂nfε∥Lp(0,T;Lq(R3))≤∑i=1n∫R3∫01∣∂znη(z)zi∣∥∂iun(x−θεz)∥Lp3(0,T;Lq3(R3))⋅∥f(x−εz)∥Lp4(0,T;Lq4(R3)) dθdz≤C∥∂iun∥Lp3(0,T;Lq3(R3))∥f∥Lp4(0,T;Lq4(R3)),(19)\begin{aligned} \left\|\partial_n(u_n f)^\varepsilon-u_n\partial_n f^\varepsilon\right\|_{L^p(0,T;L^q(\mathbb{R}^3))} &\leq\sum_{i=1}^{n}\int_{\mathbb{R}^3}\int_0^1\left|\partial_{z_n}\eta(z)z_i\right|\left\|\partial_i u_n(x-\theta\varepsilon z)\right\|_{L^{p_3}(0,T;L^{q_3}(\mathbb{R}^3))}\\ &\qquad\qquad\qquad\qquad\cdot\left\|f(x-\varepsilon z)\right\|_{L^{p_4}(0,T;L^{q_4}(\mathbb{R}^3))}\,d\theta dz\\ &\leq C\left\|\partial_i u_n\right\|_{L^{p_3}(0,T;L^{q_3}(\mathbb{R}^3))}\left\|f\right\|_{L^{p_4}(0,T;L^{q_4}(\mathbb{R}^3))}, \tag*{(19)} \end{aligned}

where 1p=1p3+1p4\frac{1}{p}=\frac{1}{p_3}+\frac{1}{p_4} and 1q=1q3+1q4\frac{1}{q}=\frac{1}{q_3}+\frac{1}{q_4}.

On the other hand, the continuity of the translations in Lebesgue space gives

∂iun(x−θεz)→∂iun(x)strongly in Lp3(0,T;Lq3(R3)),as ε→0,\partial_i u_n(x-\theta\varepsilon z)\to\partial_i u_n(x)\quad\text{strongly in }L^{p_3}(0,T;L^{q_3}(\mathbb{R}^3)),\quad\text{as }\varepsilon\to0,
f(x−θεz)→f(x)strongly in Lp4(0,T;Lq4(R3)),as ε→0,(20)f(x-\theta\varepsilon z)\to f(x)\quad\text{strongly in }L^{p_4}(0,T;L^{q_4}(\mathbb{R}^3)),\quad\text{as }\varepsilon\to0, \tag*{(20)}

provided that 1≤p3,q3,p4,q4<∞1\leq p_3,q_3,p_4,q_4<\infty.

Then combining (18),(19) and (20), we conclude from the dominated convergence theorem that

R2ε=∂n(unf)ε−un∂nfε→−∑i=1n(∫R3∂znη(z)zi dz)∂iun(x)f(x)strongly in Lp(0,T;Lq(R3)),as ε→0=∂nun(x)f(x).(21)\begin{aligned} R_2^\varepsilon&=\partial_n(u_n f)^\varepsilon-u_n\partial_n f^\varepsilon\\ &\to-\sum_{i=1}^{n}\left(\int_{\mathbb{R}^3}\partial_{z_n}\eta(z)z_i\,dz\right)\partial_i u_n(x)f(x)\quad\text{strongly in }L^p(0,T;L^q(\mathbb{R}^3)),\quad\text{as }\varepsilon\to0\\ &=\partial_n u_n(x)f(x). \tag*{(21)} \end{aligned}

Here in the last equality, we have used the integration by parts and the facts that ∑i=1n∂znzi=1\sum_{i=1}^{n}\partial_{z_n}z_i=1 and ∫R3η(z)dz=1\int_{\mathbb{R}^3}\eta(z)dz=1.

Thus, in combination with (2.11) and (2.15), and using the divergence-free condition, we obtain

R1ε+R2ε→(∑i=1n−1∂iui+∂nun)f=0,strongly in Lp(0,T;Lq(R3)), as ε→0.(22)R_1^\varepsilon+R_2^\varepsilon\to\left(\sum_{i=1}^{n-1}\partial_i u_i+\partial_n u_n\right)f=0,\quad\text{strongly in }L^p(0,T;L^q(\mathbb{R}^3)),\text{ as }\varepsilon\to0. \tag*{(22)}

Thus, the proof of this lemma is completed.

Finally, to end this section, we give the definition of weak solutions to the 3D incompressible anisotropic Navier-Stokes equations.

Definition 2.1. Let Ω=T3\Omega=\mathbb{T}^3 or R3\mathbb{R}^3. We call a vector field uu a weak solution to the 3D incompressible anisotropic Navier-Stokes equations (1.1) if

u∈L∞(0,T;L2(Ω)),∇hu∈L2(0,T;L2(Ω)),(23)u\in L^\infty(0,T;L^2(\Omega)),\qquad\nabla_hu\in L^2(0,T;L^2(\Omega)), \tag*{(23)}

it is weakly divergence-free, i.e.,

∫Ωu⋅∇ψ dx=0\int_{\Omega}u\cdot\nabla\psi\,dx=0

holds for all ψ∈C0∞(Ω)\psi\in C^\infty_0(\Omega) and almost every tt, and

∫0T∫Ω∂tϕ⋅u+∇ϕ:(u⊗u)−∇hϕ⋅∇hu dx dt=0(24)\int_0^T\int_{\Omega}\partial_t\phi\cdot u+\nabla\phi:(u\otimes u)-\nabla_h\phi\cdot\nabla_hu\,dx\,dt=0 \tag*{(24)}

holds for every divergence-free ϕ∈C0∞((0,T)×Ω)\phi\in C^\infty_0((0,T)\times\Omega).

Energy conservation of 3D anisotropic Navier-Stokes equations

In this section, we are devoted to the proof of the energy conservation of weak solutions to the 3D anisotropic Navier-Stokes equations (1.1), for both the periodic domain T3\mathbb{T}^3 and the whole space R3\mathbb{R}^3.

Torus case: T3\mathbb{T}^3

Proof of Theorem 1.1. Mollifying the equations (1.1) in spatial direction, we arrive at

∂tuε−Δhuε+(u⋅∇u)ε+∇Πε=0.(25)\partial_tu^\varepsilon-\Delta_hu^\varepsilon+(u\cdot\nabla u)^\varepsilon+\nabla\Pi^\varepsilon=0. \tag*{(25)}

Multiplying (3.1) by uεu^\varepsilon and integrating it with respect to xx and tt, we conclude by the divergence-free condition and integration by parts that

12∥uε(T)∥L2(T3)2−12∥uε(0)∥L2(T3)2+∫0T∥∇huε(s)∥L2(T3)2 dt=−∫0T∫T3(u⋅∇uh)ε⋅uhε dx dt−∫0T∫T3(u⋅∇u3)εu3ε dx dt=I+II.(26)\begin{aligned} \frac{1}{2}\lVert u^\varepsilon(T)\rVert_{L^2(\mathbb{T}^3)}^2-\frac{1}{2}\lVert u^\varepsilon(0)\rVert_{L^2(\mathbb{T}^3)}^2+\int_0^T\lVert\nabla_hu^\varepsilon(s)\rVert_{L^2(\mathbb{T}^3)}^2\,dt \\ &=-\int_0^T\int_{\mathbb{T}^3}(u\cdot\nabla u_h)^\varepsilon\cdot u_h^\varepsilon\,dx\,dt-\int_0^T\int_{\mathbb{T}^3}(u\cdot\nabla u_3)^\varepsilon u_3^\varepsilon\,dx\,dt \\ &=I+II. \tag*{(26)} \end{aligned}

According to the divergence-free condition again, we get

∫0T∫T3(uε⋅∇uhε)⋅uhε dxdt=0\int_{0}^{T}\int_{\mathbb{T}^{3}}(u^{\varepsilon}\cdot\nabla u_{h}^{\varepsilon})\cdot u_{h}^{\varepsilon}\,dxdt=0

As a result, we rewrite II as

−∫0T∫T3(u⋅∇uh)ε⋅uhε dxdt=∫0T∫T3[(uhε⋅∇huhε)−(uh⋅∇huh)ε]⋅uhε dxdt+∫0T∫T3[(u3ε∂3uhε)−(u3∂3uh)ε]⋅uhε dxdt=I1+I2.(27)\begin{aligned} -\int_{0}^{T}\int_{\mathbb{T}^{3}}(u\cdot\nabla u_{h})^{\varepsilon}\cdot u_{h}^{\varepsilon}\,dxdt &=\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{h}^{\varepsilon}\cdot\nabla_{h}u_{h}^{\varepsilon})-(u_{h}\cdot\nabla_{h}u_{h})^{\varepsilon}\right]\cdot u_{h}^{\varepsilon}\,dxdt\\ &\quad+\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{3}^{\varepsilon}\partial_{3}u_{h}^{\varepsilon})-(u_{3}\partial_{3}u_{h})^{\varepsilon}\right]\cdot u_{h}^{\varepsilon}\,dxdt\\ &=I_{1}+I_{2}. \tag*{(27)} \end{aligned}

For the last term on the right-hand side of (3.3), invoking the integration by parts and the divergence-free condition, we see that

∫0T∫T3[(u3ε∂3uhε)−(u3∂3uh)ε]⋅uhε dxdt=−∫0T∫T3[(∂3u3εuhε)−(∂3u3uh)ε]⋅uhε dxdt+∫0T∫T3∂3[(u3εuhε)−(u3uh)ε]⋅uhε dxdt=−∫0T∫T3[(∂3u3εuhε)−(∂3u3uh)ε]⋅uhε dxdt−∫0T∫T3[(u3εuhε)−(u3uh)ε]⋅∂3uhε dxdt=∫0T∫T3[(∂1u1εuhε+∂2u2εuhε)−(∂1u1uh+∂2u2uh)ε]⋅uhε dxdt−∫0T∫T3[(u3εuhε)−(u3uh)ε]⋅∂3uhε dxdt=∫0T∫T3[∂1u1εuhε−(∂1u1uh)ε]⋅uhε dxdt+∫0T∫T3[∂2u2εuhε−(∂2u2uh)ε]⋅uhε dxdt−∫0T∫T3[(u3εuhε)−(u3uh)ε]⋅∂3uhε dxdt=I21+I22+I23.(28)\begin{aligned} \int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{3}^{\varepsilon}\partial_{3}u_{h}^{\varepsilon})-(u_{3}\partial_{3}u_{h})^{\varepsilon}\right]\cdot u_{h}^{\varepsilon}\,dxdt &=-\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(\partial_{3}u_{3}^{\varepsilon}u_{h}^{\varepsilon})-(\partial_{3}u_{3}u_{h})^{\varepsilon}\right]\cdot u_{h}^{\varepsilon}\,dxdt\\ &\quad+\int_{0}^{T}\int_{\mathbb{T}^{3}}\partial_{3}\left[(u_{3}^{\varepsilon}u_{h}^{\varepsilon})-(u_{3}u_{h})^{\varepsilon}\right]\cdot u_{h}^{\varepsilon}\,dxdt\\ &=-\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(\partial_{3}u_{3}^{\varepsilon}u_{h}^{\varepsilon})-(\partial_{3}u_{3}u_{h})^{\varepsilon}\right]\cdot u_{h}^{\varepsilon}\,dxdt\\ &\quad-\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{3}^{\varepsilon}u_{h}^{\varepsilon})-(u_{3}u_{h})^{\varepsilon}\right]\cdot\partial_{3}u_{h}^{\varepsilon}\,dxdt\\ &=\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(\partial_{1}u_{1}^{\varepsilon}u_{h}^{\varepsilon}+\partial_{2}u_{2}^{\varepsilon}u_{h}^{\varepsilon})-(\partial_{1}u_{1}u_{h}+\partial_{2}u_{2}u_{h})^{\varepsilon}\right]\cdot u_{h}^{\varepsilon}\,dxdt\\ &\quad-\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{3}^{\varepsilon}u_{h}^{\varepsilon})-(u_{3}u_{h})^{\varepsilon}\right]\cdot\partial_{3}u_{h}^{\varepsilon}\,dxdt\\ &=\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[\partial_{1}u_{1}^{\varepsilon}u_{h}^{\varepsilon}-(\partial_{1}u_{1}u_{h})^{\varepsilon}\right]\cdot u_{h}^{\varepsilon}\,dxdt\\ &\quad+\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[\partial_{2}u_{2}^{\varepsilon}u_{h}^{\varepsilon}-(\partial_{2}u_{2}u_{h})^{\varepsilon}\right]\cdot u_{h}^{\varepsilon}\,dxdt\\ &\quad-\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{3}^{\varepsilon}u_{h}^{\varepsilon})-(u_{3}u_{h})^{\varepsilon}\right]\cdot\partial_{3}u_{h}^{\varepsilon}\,dxdt\\ &=I_{21}+I_{22}+I_{23}. \tag*{(28)} \end{aligned}

We assert that the following condition

∇hu∈Lp(0,T;Lq(T3)),u∈L2pp−1(0,T;L2qq−1(T3))(29)\nabla_{h}u\in L^{p}(0,T;L^{q}(\mathbb{T}^{3})),\quad u\in L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^{3})) \tag*{(29)}

implies the energy equality (1.2).

Actually, the incompressible condition allows us to deduce from (3.5) that

∇hu,∇u3∈Lp(0,T;Lq(T3)),u∈L2pp−1(0,T;L2qq−1(T3)).(30)\nabla_{h}u,\nabla u_{3}\in L^{p}(0,T;L^{q}(\mathbb{T}^{3})),\quad u\in L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^{3})). \tag*{(30)}

Making use of the Hölder inequality, we discover that

I21=∣∫0T∫T3[∂1u1εuhε−(∂1u1uh)ε]⋅uhε dxdt∣≤∥∂1u1εuhε−(∂1u1uh)ε∥L2pp+1(0,T;L2qq+1(T3))∥uhε∥L2pp−1(0,T;L2qq−1(T3)).(31)\begin{aligned} I_{21} &=\left|\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[\partial_{1}u_{1}^{\varepsilon}u_{h}^{\varepsilon}-(\partial_{1}u_{1}u_{h})^{\varepsilon}\right]\cdot u_{h}^{\varepsilon}\,dxdt\right|\\ &\leq\left\|\partial_{1}u_{1}^{\varepsilon}u_{h}^{\varepsilon}-(\partial_{1}u_{1}u_{h})^{\varepsilon}\right\|_{L^{\frac{2p}{p+1}}(0,T;L^{\frac{2q}{q+1}}(\mathbb{T}^{3}))}\left\|u_{h}^{\varepsilon}\right\|_{L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^{3}))}. \tag*{(31)} \end{aligned}

Owing to (3.6), we conclude by Lemma (2.1) that

lim⁡ε→0∥∂1u1εuhε−(∂1u1uh)ε∥L2pp+1(0,T;L2qq+1(T3))=0,(32)\lim_{\varepsilon\to0}\left\|\partial_{1}u_{1}^{\varepsilon}u_{h}^{\varepsilon}-(\partial_{1}u_{1}u_{h})^{\varepsilon}\right\|_{L^{\frac{2p}{p+1}}(0,T;L^{\frac{2q}{q+1}}(\mathbb{T}^{3}))}=0, \tag*{(32)}

which turns out that

lim⁡ε→0I21=0.(33)\lim_{\varepsilon\to0} I_{21}=0. \tag*{(33)}

Likewise,

lim⁡ε→0I22=0,(34)\lim_{\varepsilon\to0} I_{22}=0, \tag*{(34)}

and

lim⁡ε→0I1=0.(35)\lim_{\varepsilon\to0} I_1=0. \tag*{(35)}

Before going further, together with u3∈L∞(0,T;L2(T3))u_3 \in L^\infty(0,T;L^2(\mathbb{T}^3)), we conclude that ∇u3∈Lp(0,T;Lq(T3))\nabla u_3 \in L^p(0,T;L^q(\mathbb{T}^3)) implies u3∈Lp(0,T;W1,q(T3))u_3 \in L^p(0,T;W^{1,q}(\mathbb{T}^3)). Indeed, for 1≤q≤21 \leq q \leq2, we derive from the Hölder inequality that

∥u3∥Lp(0,T;Lq(T3))≤C∥u3∥Lp(0,T;L2(T3))≤C∥u3∥L∞(0,T;L2(T3)).\|u_3\|_{L^p(0,T;L^q(\mathbb{T}^3))} \leq C\|u_3\|_{L^p(0,T;L^2(\mathbb{T}^3))} \leq C\|u_3\|_{L^\infty(0,T;L^2(\mathbb{T}^3))}.

For q>2q>2, we conclude by the Gagliardo-Nirenberg inequality that

∥u3∥Lp(0,T;Lq(T3))≤C∥u3∥Lp(0,T;L2(T3))2q5q−6∥∇u3∥Lp(0,T;Lq(T3))3q−65q−6+C∥u3∥L∞(0,T;L2(T3))≤C∥u3∥L∞(0,T;L2(T3))2q5q−6∥∇u3∥Lp(0,T;Lq(T3))3q−65q−6+C∥u3∥L∞(0,T;L2(T3)).\begin{aligned} \|u_3\|_{L^p(0,T;L^q(\mathbb{T}^3))} &\leq C\|u_3\|_{L^p(0,T;L^2(\mathbb{T}^3))}^{\frac{2q}{5q-6}}\|\nabla u_3\|_{L^p(0,T;L^q(\mathbb{T}^3))}^{\frac{3q-6}{5q-6}}+C\|u_3\|_{L^\infty(0,T;L^2(\mathbb{T}^3))}\\ &\leq C\|u_3\|_{L^\infty(0,T;L^2(\mathbb{T}^3))}^{\frac{2q}{5q-6}}\|\nabla u_3\|_{L^p(0,T;L^q(\mathbb{T}^3))}^{\frac{3q-6}{5q-6}}+C\|u_3\|_{L^\infty(0,T;L^2(\mathbb{T}^3))}. \end{aligned}

Consequently, ∇u3∈Lp(0,T;Lq(T3))\nabla u_3 \in L^p(0,T;L^q(\mathbb{T}^3)) means that u3∈Lp(0,T;W1,q(T3))u_3 \in L^p(0,T;W^{1,q}(\mathbb{T}^3)). With this in hand, we continue to deal with I23I_{23}. In view of the Hölder inequality, we observe that

∣∫0T∫T3[(u3uh)ε−(u3εuhε)]∂3uhε dx dt∣≤C∥(u3uh)ε−(u3εuhε)∥L2pp+1(0,T;L2qq+1(T3))∥∇uhε∥L2pp−1(0,T;L2qq−1(T3)).(36)\left|\int_0^T\int_{\mathbb{T}^3}\left[(u_3u_h)^\varepsilon-(u_3^\varepsilon u_h^\varepsilon)\right]\partial_3u_h^\varepsilon\,dx\,dt\right| \leq C\|(u_3u_h)^\varepsilon-(u_3^\varepsilon u_h^\varepsilon)\|_{L^{\frac{2p}{p+1}}(0,T;L^{\frac{2q}{q+1}}(\mathbb{T}^3))}\|\nabla u_h^\varepsilon\|_{L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^3))}. \tag*{(36)}

On the one hand, taking advantage of u∈L2pp−1(0,T;L2qq−1(T3))u \in L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^3)) and Lemma 2.2, we notice that

∥∇uhε∥L2pp−1(0,T;L2qq−1(T3))≤Cε−1∥uh∥L2pp−1(0,T;L2qq−1(T3)),\|\nabla u_h^\varepsilon\|_{L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^3))} \leq C\varepsilon^{-1}\|u_h\|_{L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^3))},
lim sup⁡ε→0ε∥∇uhε∥L2pp−1(0,T;L2qq−1(T3))=0.(37)\limsup_{\varepsilon\to0}\varepsilon\|\nabla u_h^\varepsilon\|_{L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^3))}=0. \tag*{(37)}

On the other hand, it follows from the Lemma 2.3 and u3∈Lp(0,T;W1,q(T3))u_3 \in L^p(0,T;W^{1,q}(\mathbb{T}^3)) that

∥(u3uh)ε−(u3εuhε)∥L2pp+1(0,T;L2qq+1(T3))≤Cε∥u3∥Lp(0,T;W1,q(T3))∥uh∥L2pp−1(0,T;L2qq−1(T3)).(38)\|(u_3u_h)^\varepsilon-(u_3^\varepsilon u_h^\varepsilon)\|_{L^{\frac{2p}{p+1}}(0,T;L^{\frac{2q}{q+1}}(\mathbb{T}^3))} \leq C\varepsilon\|u_3\|_{L^p(0,T;W^{1,q}(\mathbb{T}^3))}\|u_h\|_{L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^3))}. \tag*{(38)}

Plugging this into (3.12), we observe that

∣∫0T∫T3[(u3uh)ε−(u3εuhε)]∂3uhε dx dt∣≤C∥(u3uh)ε−(u3εuhε)∥L2pp+1(0,T;L2qq+1(T3))∥∇uhε∥Lp(0,T;Lq(T3))≤ε∥∇uhε∥Lp(0,T;Lq(T3)),\begin{aligned} \left|\int_0^T\int_{\mathbb{T}^3}\left[(u_3u_h)^\varepsilon-(u_3^\varepsilon u_h^\varepsilon)\right]\partial_3u_h^\varepsilon\,dx\,dt\right| &\leq C\|(u_3u_h)^\varepsilon-(u_3^\varepsilon u_h^\varepsilon)\|_{L^{\frac{2p}{p+1}}(0,T;L^{\frac{2q}{q+1}}(\mathbb{T}^3))}\|\nabla u_h^\varepsilon\|_{L^p(0,T;L^q(\mathbb{T}^3))}\\ &\leq\varepsilon\|\nabla u_h^\varepsilon\|_{L^p(0,T;L^q(\mathbb{T}^3))}, \end{aligned}

which together with (3.13) leads to

lim⁡ε→0I23=0.(39)\lim_{\varepsilon\to0} I_{23}=0. \tag*{(39)}

Then in combination with (3.4), (3.9)–(3.11) and (3.15), we have

lim⁡ε→0I=0.(40)\lim_{\varepsilon\to0} I = 0. \tag*{(40)}

Next, we are ready to deal with the remainder term IIII. Firstly, thanks to

∫0T∫T3(uε⋅∇u3ε)u3ε dxdt=0,\int_{0}^{T}\int_{\mathbb{T}^{3}}(u^{\varepsilon}\cdot\nabla u_{3}^{\varepsilon})u_{3}^{\varepsilon}\,dxdt=0,

we reformulate IIII as

−∫0T∫T3(u⋅∇u3)u3ε dxdt=∫0T∫T3[(uhε⋅∇hu3ε)−(uh⋅∇hu3)ε]u3ε dxdt+∫0T∫T3[(u3ε∂3u3ε)−(u3∂3u3)ε]u3ε dxdt=∫0T∫T3[(uhε⋅∇hu3ε)−(uh⋅∇hu3)ε]u3ε dxdt−∫0T∫T3[(u3ε(∂1u1ε+∂2u2ε))−(u3∂1u1)ε−(u3∂2u2)ε]u3ε dxdt=∫0T∫T3[(uhε⋅∇hu3ε)−(uh⋅∇hu3)ε]u3ε dxdt−∫0T∫T3[(u3ε∂1u1ε)−(u3∂1u1)ε]u3ε dxdt−∫0T∫T3[(u3ε∂2u2ε)−(u3∂2u2)ε]u3ε dxdt=II1+II2+II3.(41)\begin{aligned} -\int_{0}^{T}\int_{\mathbb{T}^{3}}(u\cdot\nabla u_{3})u_{3}^{\varepsilon}\,dxdt &=\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{h}^{\varepsilon}\cdot\nabla_{h}u_{3}^{\varepsilon})-(u_{h}\cdot\nabla_{h}u_{3})^{\varepsilon}\right]u_{3}^{\varepsilon}\,dxdt \\ &\quad+\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{3}^{\varepsilon}\partial_{3}u_{3}^{\varepsilon})-(u_{3}\partial_{3}u_{3})^{\varepsilon}\right]u_{3}^{\varepsilon}\,dxdt \\ &=\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{h}^{\varepsilon}\cdot\nabla_{h}u_{3}^{\varepsilon})-(u_{h}\cdot\nabla_{h}u_{3})^{\varepsilon}\right]u_{3}^{\varepsilon}\,dxdt \\ &\quad-\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{3}^{\varepsilon}(\partial_{1}u_{1}^{\varepsilon}+\partial_{2}u_{2}^{\varepsilon}))-(u_{3}\partial_{1}u_{1})^{\varepsilon}-(u_{3}\partial_{2}u_{2})^{\varepsilon}\right]u_{3}^{\varepsilon}\,dxdt \\ &=\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{h}^{\varepsilon}\cdot\nabla_{h}u_{3}^{\varepsilon})-(u_{h}\cdot\nabla_{h}u_{3})^{\varepsilon}\right]u_{3}^{\varepsilon}\,dxdt \\ &\quad-\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{3}^{\varepsilon}\partial_{1}u_{1}^{\varepsilon})-(u_{3}\partial_{1}u_{1})^{\varepsilon}\right]u_{3}^{\varepsilon}\,dxdt \\ &\quad-\int_{0}^{T}\int_{\mathbb{T}^{3}}\left[(u_{3}^{\varepsilon}\partial_{2}u_{2}^{\varepsilon})-(u_{3}\partial_{2}u_{2})^{\varepsilon}\right]u_{3}^{\varepsilon}\,dxdt \\ &=II_{1}+II_{2}+II_{3}. \tag*{(41)} \end{aligned}

By virtue of the Hölder inequality,

∣II1∣≤∥(uhε⋅∇hu3ε)−(uh⋅∇hu3)ε∥L2pp+1(0,T;L2qq+1(T3))∥u3ε∥L2pp−1(0,T;L2qq−1(T3)).(42)|II_{1}|\leq\left\|(u_{h}^{\varepsilon}\cdot\nabla_{h}u_{3}^{\varepsilon})-(u_{h}\cdot\nabla_{h}u_{3})^{\varepsilon}\right\|_{L^{\frac{2p}{p+1}}(0,T;L^{\frac{2q}{q+1}}(\mathbb{T}^{3}))}\|u_{3}^{\varepsilon}\|_{L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^{3}))}. \tag*{(42)}

In view of (3.6), we deduce from Lemma (2.1) that

lim⁡ε→0∥(uhε⋅∇hu3ε)−(uh⋅∇hu3)ε∥L2pp+1(0,T;L2qq+1(T3))=0,(43)\lim_{\varepsilon\to0}\left\|(u_{h}^{\varepsilon}\cdot\nabla_{h}u_{3}^{\varepsilon})-(u_{h}\cdot\nabla_{h}u_{3})^{\varepsilon}\right\|_{L^{\frac{2p}{p+1}}(0,T;L^{\frac{2q}{q+1}}(\mathbb{T}^{3}))}=0, \tag*{(43)}

which implies that

lim⁡ε→0II1=0.(44)\lim_{\varepsilon\to0}II_{1}=0. \tag*{(44)}

By the same token, we also have

lim⁡ε→0II2=lim⁡ε→0II3=0.(45)\lim_{\varepsilon\to0}II_{2}=\lim_{\varepsilon\to0}II_{3}=0. \tag*{(45)}

As a consequence, we conclude that

lim⁡ε→0II=0.(46)\lim_{\varepsilon\to0}II=0. \tag*{(46)}

Hence, combining (3.16) and (3.22), we verified the assertment (3.5). Now, we are in a position to pass to the limit of ε\varepsilon in (3.2) to get the energy conservation (1.2).

(1) As the natural energy gives ∇hu∈L2(0,T;L2(T3))\nabla_{h}u\in L^{2}(0,T;L^{2}(\mathbb{T}^{3})), by choosing p=q=2p=q=2 in (3.5), we immediately prove that the condition u∈L4(0,T;L4(T3))u\in L^{4}(0,T;L^{4}(\mathbb{T}^{3})) yields the energy equality. It is worth remarking that the rest proof in Theorem 1.1 can be reduced to this special case.

(2) Next, we deal with the case (2) in Theorem 1.1 with q≥4q \ge4 and 2p+2q=1\frac{2}{p}+\frac{2}{q}=1. The Gagliardo-Nirenberg inequality guarantees that

∥u∥L4(0,T;L4(T3))≤C∥u∥L∞(0,T;L2(T3))q−42q−4∥u∥Lp(0,T;Lq(T3))q2q−4≤C.(47)\|u\|_{L^{4}(0,T;L^{4}(\mathbb{T}^{3}))} \le C\|u\|_{L^{\infty}(0,T;L^{2}(\mathbb{T}^{3}))}^{\frac{q-4}{2q-4}}\|u\|_{L^{p}(0,T;L^{q}(\mathbb{T}^{3}))}^{\frac{q}{2q-4}} \le C. \tag*{(47)}

From the result just proved, we obtain the energy equality via (1) in Theorem 1.1 with q≥4q \ge4.

(3) Now, we focus on the proof of (3). Indeed, note that u∈L2pp−1(0,T;W1,q(T3))u \in L^{\frac{2p}{p-1}}(0,T;W^{1,q}(\mathbb{T}^{3})), therefore, according to (3.5), it suffices to derive u∈L2pp−1(0,T;L2qq−1(T3))u \in L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^{3})) from (3.5). For q≥95q \ge\frac{9}{5}, by the Gagliardo-Nirenberg inequality, we get

∥u∥L2qq−1(T3)≤C∥u∥L2(T3)5q−95q−6∥∇u∥Lq(T3)35q−6.(48)\|u\|_{L^{\frac{2q}{q-1}}(\mathbb{T}^{3})} \le C\|u\|_{L^{2}(\mathbb{T}^{3})}^{\frac{5q-9}{5q-6}}\|\nabla u\|_{L^{q}(\mathbb{T}^{3})}^{\frac{3}{5q-6}}. \tag*{(48)}

Thanks to 1p+65q=1\frac{1}{p}+\frac{6}{5q}=1, we further infer that

∥u∥L2pp−1(0,T;L2qq−1(T3))≤C∥u∥L∞(0,T;L2(T3))5q−95q−6∥∇u∥Lp(0,T;Lq(T3))35q−6≤C.(49)\|u\|_{L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{T}^{3}))} \le C\|u\|_{L^{\infty}(0,T;L^{2}(\mathbb{T}^{3}))}^{\frac{5q-9}{5q-6}}\|\nabla u\|_{L^{p}(0,T;L^{q}(\mathbb{T}^{3}))}^{\frac{3}{5q-6}} \le C. \tag*{(49)}

In light of (3.5), we have proved case (3) in Theorem 1.1 for q≥95q \ge\frac{9}{5}.

Thus, we complete the proof of Theorem 1.1. □

Whole space case: R3\mathbb{R}^{3}

Proof of Theorem 1.2. Following the proof strategy of Theorem 1.1 and recalling (3.2), we obtain

12∥uε(T)∥L2(R3)2−12∥uε(0)∥L2(R3)2+∫0T∥∇huε(s)∥L2(R3)2 dt=−∫0T∫R3(u⋅∇uh)ε⋅uhε dxdt−∫0T∫R3(u⋅∇u3)εu3ε dxdt=I+II.(50)\frac{1}{2}\|u^{\varepsilon}(T)\|_{L^{2}(\mathbb{R}^{3})}^{2}-\frac{1}{2}\|u^{\varepsilon}(0)\|_{L^{2}(\mathbb{R}^{3})}^{2}+\int_{0}^{T}\|\nabla_{h}u^{\varepsilon}(s)\|_{L^{2}(\mathbb{R}^{3})}^{2}\,dt = -\int_{0}^{T}\int_{\mathbb{R}^{3}}(u\cdot\nabla u_{h})^{\varepsilon}\cdot u_{h}^{\varepsilon}\,dxdt-\int_{0}^{T}\int_{\mathbb{R}^{3}}(u\cdot\nabla u_{3})^{\varepsilon}u_{3}^{\varepsilon}\,dxdt = I+II. \tag*{(50)}

To deal with the first term, the incompressible condition helps us to get

∫0T∫R3(u⋅∇uh)ε⋅uhε dxdt=∫0T∫R3div⁡(uu1ε)⋅u1ε dxdt+∫0T∫R3div⁡(uu2ε)⋅u2ε dxdt=0,(51)\int_{0}^{T}\int_{\mathbb{R}^{3}}(u\cdot\nabla u_{h})^{\varepsilon}\cdot u_{h}^{\varepsilon}\,dxdt =\int_{0}^{T}\int_{\mathbb{R}^{3}}\operatorname{div}(uu_{1}^{\varepsilon})\cdot u_{1}^{\varepsilon}\,dxdt+\int_{0}^{T}\int_{\mathbb{R}^{3}}\operatorname{div}(uu_{2}^{\varepsilon})\cdot u_{2}^{\varepsilon}\,dxdt=0, \tag*{(51)}

and

−∫0T∫R3(u⋅∇uh)ε⋅uhε dxdt=−∫0T∫R3div⁡(uu1)ε⋅u1ε dxdt−∫0T∫R3div⁡(uu2)ε⋅u2ε dxdt.(52)-\int_{0}^{T}\int_{\mathbb{R}^{3}}(u\cdot\nabla u_{h})^{\varepsilon}\cdot u_{h}^{\varepsilon}\,dxdt =-\int_{0}^{T}\int_{\mathbb{R}^{3}}\operatorname{div}(uu_{1})^{\varepsilon}\cdot u_{1}^{\varepsilon}\,dxdt-\int_{0}^{T}\int_{\mathbb{R}^{3}}\operatorname{div}(uu_{2})^{\varepsilon}\cdot u_{2}^{\varepsilon}\,dxdt. \tag*{(52)}

In combination with (3.27) and (3.28), we can reformulate the first term as

I=−∫0T∫R3(u⋅∇uh)ε⋅uhε dxdt=∫0T∫R3div⁡[uu1ε−(uu1)ε]⋅u1ε dxdt+∫0T∫R3div⁡[uu2ε−(uu2)ε]⋅u2ε dxdt.(53)I=-\int_{0}^{T}\int_{\mathbb{R}^{3}}(u\cdot\nabla u_{h})^{\varepsilon}\cdot u_{h}^{\varepsilon}\,dxdt =\int_{0}^{T}\int_{\mathbb{R}^{3}}\operatorname{div}[uu_{1}^{\varepsilon}-(uu_{1})^{\varepsilon}]\cdot u_{1}^{\varepsilon}\,dxdt+\int_{0}^{T}\int_{\mathbb{R}^{3}}\operatorname{div}[uu_{2}^{\varepsilon}-(uu_{2})^{\varepsilon}]\cdot u_{2}^{\varepsilon}\,dxdt. \tag*{(53)}

By the same token, we also have

II=−∫0T∫T3(u⋅∇u3)εu3ε dxdt=∫0T∫R3div⁡[uu3ε−(uu3)ε]⋅u3ε dxdt.(54)II=-\int_0^T\int_{\mathbb{T}^3}(u\cdot\nabla u_3)^\varepsilon u_3^\varepsilon\,dxdt=\int_0^T\int_{\mathbb{R}^3}\operatorname{div}[uu_3^\varepsilon-(uu_3)^\varepsilon]\cdot u_3^\varepsilon\,dxdt. \tag*{(54)}

Before going further, we assert that

∇hu∈Lp(0,T;Lq(R3)),u∈L2pp−1(0,T;L2qq−1(R3))(55)\nabla_hu\in L^p(0,T;L^q(\mathbb{R}^3)),\quad u\in L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{R}^3)) \tag*{(55)}

implies the energy equality (1.2).

Next, we prove that the terms on the right-hand side of (50) vanish as ε→0\varepsilon\to0. Contrary to the strategy adopted for the torus domain, we employ the anisotropic version of Lions’ commutator estimates, instead of Constantin-E-Titi-type commutators, to handle these terms. Precisely, recall (55) and using the Hölder inequality, it gives

∣−∫0T∫R3(u⋅∇uh)ε⋅uhεdxdt∣≤∥div⁡[uu1ε−(uu1)ε]∥L2pp+1(0,T;L2qq+1(R3))∥u1ε∥L2pp−1(0,T;L2qq−1(R3))+∥div⁡[uu2ε−(uu2)ε]∥L2pp+1(0,T;L2qq+1(R3))∥u2ε∥L2pp−1(0,T;L2qq−1(R3)).(56)\left|-\int_0^T\int_{\mathbb{R}^3}(u\cdot\nabla u_h)^\varepsilon\cdot u_h^\varepsilon dxdt\right| \leq\|\operatorname{div}[uu_1^\varepsilon-(uu_1)^\varepsilon]\|_{L^{\frac{2p}{p+1}}(0,T;L^{\frac{2q}{q+1}}(\mathbb{R}^3))}\|u_1^\varepsilon\|_{L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{R}^3))} +\|\operatorname{div}[uu_2^\varepsilon-(uu_2)^\varepsilon]\|_{L^{\frac{2p}{p+1}}(0,T;L^{\frac{2q}{q+1}}(\mathbb{R}^3))}\|u_2^\varepsilon\|_{L^{\frac{2p}{p-1}}(0,T;L^{\frac{2q}{q-1}}(\mathbb{R}^3))}. \tag*{(56)}

Then, by virtue of lemma 2.4 with the choices u=uu=u, f=u1f=u_1 or u2u_2, p1=p4=2pp−1p_1=p_4=\frac{2p}{p-1}, q1=q4=2qq−1q_1=q_4=\frac{2q}{q-1}, p3=p2=pp_3=p_2=p, q3=q2=qq_3=q_2=q and p=2pp+1p=\frac{2p}{p+1}, q=2qq+1q=\frac{2q}{q+1}, we conclude

div⁡[uu1ε−(uu1)ε]→0,strongly in L2pp+1(0,T;L2qq+1(R3)), as ε→0;\operatorname{div}[uu_1^\varepsilon-(uu_1)^\varepsilon]\to0,\quad\text{strongly in }L^{\frac{2p}{p+1}}(0,T;L^{\frac{2q}{q+1}}(\mathbb{R}^3)),\text{ as }\varepsilon\to0;
div⁡[uu2ε−(uu2)ε]→0,strongly in L2pp+1(0,T;L2qq+1(R3)), as ε→0.(57)\operatorname{div}[uu_2^\varepsilon-(uu_2)^\varepsilon]\to0,\quad\text{strongly in }L^{\frac{2p}{p+1}}(0,T;L^{\frac{2q}{q+1}}(\mathbb{R}^3)),\text{ as }\varepsilon\to0. \tag*{(57)}

Then together with (56) and (57), we have

∣I∣=∣−∫0T∫R3(u⋅∇uh)ε⋅uhεdxdt∣→0,as ε→0.(58)|I|=\left|-\int_0^T\int_{\mathbb{R}^3}(u\cdot\nabla u_h)^\varepsilon\cdot u_h^\varepsilon dxdt\right|\to0,\quad\text{as }\varepsilon\to0. \tag*{(58)}

For term IIII, we note that the divergence-free condition together with (55) guarantees full regularity for u3u_3. Indeed, ∇hu∈Lp(0,T;Lq(R3))\nabla_hu\in L^p(0,T;L^q(\mathbb{R}^3)) implies ∇huh∈Lp(0,T;Lq(R3))\nabla_hu_h\in L^p(0,T;L^q(\mathbb{R}^3)). From the divergence-free constraint div⁡u=0\operatorname{div}u=0, we have ∂3u3=−∂h⋅uh\partial_3u_3=-\partial_h\cdot u_h, which further yields ∇u3∈Lp(0,T;Lq(R3))\nabla u_3\in L^p(0,T;L^q(\mathbb{R}^3)).

Based on this, using the integration by parts and the Hölder inequality, we get

∣II∣=∣∫0T∫R3div⁡[uu3ε−(uu3)ε]⋅u3εdxdt∣=∣−∫0T∫R3[uu3ε−uu3+uu3−(uu3)ε]⋅∇u3εdxdt∣≤C∥uu3ε−uu3+uu3−(uu3)ε∥Lpp−1(0,T;Lqq−1(R3))∥∇u3ε∥Lp(0,T;Lq(R3))(59)\begin{aligned} |II|=\left|\int_0^T\int_{\mathbb{R}^3}\operatorname{div}[uu_3^\varepsilon-(uu_3)^\varepsilon]\cdot u_3^\varepsilon dxdt\right| \\ &=\left|-\int_0^T\int_{\mathbb{R}^3}[uu_3^\varepsilon-uu_3+uu_3-(uu_3)^\varepsilon]\cdot\nabla u_3^\varepsilon dxdt\right| \\ &\leq C\|uu_3^\varepsilon-uu_3+uu_3-(uu_3)^\varepsilon\|_{L^{\frac{p}{p-1}}(0,T;L^{\frac{q}{q-1}}(\mathbb{R}^3))}\|\nabla u_3^\varepsilon\|_{L^p(0,T;L^q(\mathbb{R}^3))} \tag*{(59)} \end{aligned}

which together with the standard properties of the mollification yields

∣II∣=∣∫0T∫R3div⁡[uu3ε−(uu3)ε]⋅u3ε dxdt∣→0,as ε→0.(60)|II| = \left|\int_{0}^{T}\int_{\mathbb{R}^{3}} \operatorname{div}[u u_{3}^{\varepsilon} - (u u_{3})^{\varepsilon}] \cdot u_{3}^{\varepsilon} \, dxdt\right| \to0,\quad\text{as }\varepsilon\to0. \tag*{(60)}

Hence, combining (58) and (59), we conclude the proof of assertion (55). Since the proof below is analogous to that in the previous subsection, Theorem 1.2 follows by repeating the arguments therein, and we omit the details.

Acknowledgements

Ye was partially sponsored by the Training Program for Young Backbone Teachers in Higher Education Institutions of Henan Province (2024GGJS022). Wei was partially supported by the National Natural Science Foundation of China under grant (No. 12271433). Wang was sponsored by Natural Science Foundation of Henan Province (No. 232300421077) and supported by the National Natural Science Foundation of China under grant (No. 11971446 and No. 12071113).

Author Declarations statement

The authors have no conflicts to disclose.

Data Availability

This publication is supported by multiple datasets, which are openly available at locations cited in the reference section.

References

  1. [2]H. Beirao da Veiga and J. Yang, On the Shinbrot’s criteria for energy equality to Newtonian fluids: a simplified proof, and an extension of the range of application. Nonlinear Anal. 196 (2020), 111809, 4 pp.
  2. [3]H. Beirao da Veiga and J. Yang, The relations between Hölder continuity assumptions on the direction of vorticity and energy equality. J. Geom. Anal. 35 (2025), 15.DOI
  3. [4]L. C. Berselli, Energy conservation for weak solutions of incompressible fluid equations: The Hölder case and connections with Onsager’s conjecture. J. Differential Equations, 368 (2023), 350–375.
  4. [5]L. C. Berselli and E. Chiodaroli, On the energy equality for the 3D Navier-Stokes equations. Nonlinear Anal. 192 (2020), 111704, 24 pp.arxiv.org/abs/1807.02667
  5. [6]L. C. Berselli and E. Chiodaroli and R. Sannipoli, Energy conservation for 3D Euler and Navier-Stokes equations in a bounded domain: applications to Beltrami flows. J. Nonlinear Sci. 35 (2025), no. 1, Paper No. 10, 30 pp.
  6. [7]L. C. Berselli and S. Georgiadis, Three results on the Energy conservation for the 3D Euler equations. Nonlinear Differ. Equ. Appl. 31, 33 (2024), 1–14.
  7. [8]L. C. Berselli, A. Kaltenbach and M. Růžička, Energy conservation for weak solutions of incompressible Newtonian fluid equations in Hölder spaces with Dirichlet boundary conditions in the half-space. Math. Ann. 391 (2025), no. 4, 5911–5940.
  8. [9]Q. Bie, H. Fang, S. Wang and Y. Zhou, Stability and sharp decay for the 3D incompressible anisotropic Navier-Stokes equations with fractional horizontal dissipation. J. Differential Equations 463 (2026), Paper No. 114167, 38 pp.
  9. [10]C. Cao and Y. Guo, On the two-dimensional Navier-Stokes equations with horizontal viscosity. J. Math. Anal. Appl. 557 (2026), no. 2, Paper No. 130352, 24 pp.
  10. [11]C. Cao and J. Wu, Stability of the 3D Navier-Stokes equations with anisotropic dissipation. Nonlinearity. 38 (2025), no. 9, Paper No. 095012, 10 pp.DOI
  11. [12]C. Cao and J. Wu, Global regularity for the two-dimensional anisotropic Boussinesq equations with vertical dissipation. Arch. Ration. Mech. Anal. 208(2013), no. 3, 985–1004.DOI
  12. [13]J. Chemin, B. Desjardins, I. Gallagher and E. Grenier. Fluids with anisotropic viscosity. M2AN Math. Model. Numer. Anal. 34(2000), no. 2, 315–335.DOI
  13. [15]A. Cheskidov, P. Constantin, S. Friedlander and R. Shvydkoy, Energy conservation and Onsager’s conjecture for the Euler equations. Nonlinearity, 21 (2008), 1233–1252.
  14. [16]A. Cheskidov and X. Luo, Energy equality for the Navier-Stokes equations in weak-in-time Onsager spaces. Nonlinearity, 33 (2020), 1388–1403.arxiv.org/abs/1802.05785
  15. [18]J. Demmel and E. Wiedemann Energy rigidity and weak-strong uniqueness for the 2D anisotropic Navier-Stokes equations. arXiv.2608.19931.
  16. [19]C. De Lellis and L. J. Székelyhidi, The Euler equations as a differential inclusion. Ann. Math. 170 (2009), 1417–1436.
  17. [20]C. De Lellis and L. J. Székelyhidi, Dissipative continuous Euler flows. Invent. Math. 193 (2013), 377–407.DOI
  18. [21]E. Feireisl, P. Gwiazda, A. Świerczewska-Gwiazda and E. Wiedemann, Regularity and energy conservation for the compressible euler equations. Arch Ration Mech Anal. 223 (2017), 1375–1395.DOI
  19. [22]U. S. Fjordholm and E. Wiedemann, Statistical solutions and Onsager’s conjecture. Phys. D. 376-377 (2018), 259–265.DOI
  20. [23]M. Fujii, Large time behavior of solutions to the 3D anisotropic Navie–Stokes equation. Nonlinear Anal. Real World Appl. 71 (2023), Paper No. 103821, 33 pp.arxiv.org/abs/2108.11940
  21. [24]G. P. Galdi, An introduction to the Navier-Stokes initial-boundary value problem, in: Fundamental Directions in Mathematical Fluid Mechanics, in: Adv. Math. Fluid Mech., Birkhäuser, Basel, 2000, 1–70.DOI
  22. [25]V. Giri, H. Kwon and M. Novack. The L³-based strong Onsager theorem. Ann. of Math. (2) 204 (2026), no. 1, 265–421.
  23. [27]P. Isett, A proof of Onsager’s conjecture. Ann. of Math. 188 (2018), 871–963.arxiv.org/abs/1608.08301
  24. [28]R. Ji, L. Tian and J. Wu, 3D anisotropic Navier-Stokes equations in T² × ℝ : stability and large-time behaviour. Nonlinearity. 36 (2023), no. 6, 3219–3237.DOI
  25. [30]T. M. Leslie and R. Shvydkoy. Conditions implying energy equality for weak solutions of the Navier-Stokes equations. SIAM J. Math. Anal. 50 (2018), 870–890.arxiv.org/abs/1606.02363
  26. [31]J. L. Lions, Sur la régularité et l’unicité des solutions turbulentes des équations de Navier Stokes, Rend. Semin. Mat. Univ. Padova, 30 (1960) 16–23.
  27. [32]Q. Nguyen, P. Nguyen and B. Tang, Energy equalities for compressible Navier-Stokes equations. Nonlinearity 32 (2019), 4206–4231.DOI
  28. [33]L. Onsager, Statistical hydrodynamics. Nuovo Cim. (Suppl.) 6 (1949), 279–287.DOI
  29. [34]J. Pedlosky, Geophysical Fluid Dynamics. Springer-Verlag, Berlin, 1979.
  30. [35]M. Shinbrot, The energy equation for the Navier-Stokes system. SIAM J. Math. Anal. 5 (1974) 948–954.DOI
  31. [36]Y. Taniuchi, On generalized energy equality of the Navier-Stokes equations. Manuscripta Math. 94 (1997), 365–384.
  32. [38]Y. Wang and Y. Ye, A general sufficient criterion for energy conservation in the Navier-Stokes system. Math Meth Appl Sci. 46 (2023), 9268–9285.
  33. [39]W. Yang, Q. Jiu and J. Wu, The 3D incompressible Navier-Stokes equations with partial hyperdissipation. Math. Nachr. 292 (2019), no. 8, 1823–1836.
  34. [40]L. Xu and P. Zhang, Enhanced dissipation for the third component of 3D anisotropic Navier-Stokes equations. J. Differential Equations 335 (2022), 464–496.
  35. [42]Y. Ye, W. Wei and Y. Wang, Energy equality in the isentropic compressible Navier-Stokes equations allowing vacuum. J. Differential Equations 338 (2022), 551–571.
  36. [44]Z. Zhang, Remarks on the energy equality for the non-Newtonian fluids. J. Math. Anal. Appl. 480 (2019), no. 2, 123443, 9 pp.DOI
  37. [45]P. Zhang and W. Zhu, Continuous dependence on initial data for the solutions of 3-D anisotropic Navier-Stokes equations. J. Funct. Anal. 288 (2025), no. 1, Paper No. 110689, 36 pp.DOI
  38. [47]D. Zhou and J. Wu, H^1-uniqueness, stability and decay for anisotropic Navier-Stokes and Boussinesq equations. J. Differential Equations, 453 (2026), 113930.DOI

Paper details

vlnc:W_tjcnzvye
Version
v1 · latest
Published
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents