Abstract
In this paper, we prove the local well-posedness of 3-D axi-symmetric Navier–Stokes system with initial data in the critical Lebesgue spaces. We also obtain the global well-posedness result with small initial data. Furthermore, with the initial swirl component of the velocity being sufficiently small in the almost critical spaces, we can still prove the global well-posedness of the system.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
In this paper, we investigate the well-posedness and long-time behavior of global solutions to 3D axisymmetric Navier–Stokes equations with a small swirl component. In general, 3-D Navier–Stokes system in \(\mathop {\mathbb R}\nolimits ^3\) reads
where \(u(t,x)=\left( u^1,u^2,u^3\right) \) stands for the velocity field and p the scalar pressure function of the fluid, which guarantees the divergence free condition of the velocity field. This system describes the motion of viscous incompressible fluid flows.
We recall that except the initial data with special structure, it is not known whether or not the system (1.1) has a unique global smooth solution with large smooth initial data. For instance, the system (1.1) is globally well-posed for data which is axisymmetric and without swirl component (that is the case when \(u^\theta =0\) in (1.3) below). In this case, Ladyzhenskaya [7] and independently Ukhovskii and Yudovich [10] proved the existence of weak solutions along with the uniqueness and regularities of such solution for (1.1). Leonardi et al. [8] gave a refined proof of the same result in [7, 10]. And even with a small swirl component, the authors [11] could also establish the global well-posedess of (1.1). In general, even the global wellposedness of (1.1) with axisymmetric initial data is still open.
On the other hand, in the seminal paper [9], Leray proved the global existence of finite energy weak solutions to (1.1). Yet the uniqueness and regularity to this weak solution are big open questions in the field of mathematical fluid mechanics. Furthermore, Leray emphasized two facts about Navier–Stokes system. Firstly, he pointed out that energy estimate method is very important to study Navier–Stokes system. The general energy inequality for (1.1)
is the cornerstone of the proof to the existence of global turbulent solution to (1.1) in [9]. The energy estimate relies (formally) on the fact that if v is a divergence free vector field, \( (v\cdot \nabla f|f)_{L^2} =0~\) and that \( (\nabla p|v)_{L^2}=0\). In the present work, we shall use the more general fact that for any divergence free vector field v and any function a, we have
This will lead to the \(L^p\) type energy estimate. Secondly Leray pointed out that the scaling invariance of (1.1), that is,
if (v, p) is a solution of (1.1) on \([0,T]\times \mathop {\mathbb R}\nolimits ^3\) associated with an initial data \(v_0\), then \((v_\lambda ,p_\lambda )\) is also a solution of (1.1) on \([0,\lambda ^{-2}T]\times \mathop {\mathbb R}\nolimits ^3\) associated with the initial data \(\lambda v_0(\lambda x),\) is another important fact in the study of Navier–Stokes system. The scaling property is also the foundation of the Kato theory which gives a general method to solve (locally or globally) the incompressible Navier–Stokes equation in critical spaces i.e. spaces with the norms of which are invariant under the scaling. In what follows, we shall use such scaling invariant space as \(L^\infty (]0,t[; L^1(\Omega )),\) where the norm \(L^1(\Omega )\) is given by (1.5).
In fact, Gally and \(\breve{S}\)verák [5] recently proved the global well-posedness of 3-D axisymmetric Navier–Stokes system without swirl and with initial data in the scaling invariant function spaces. We remark that the reason why one can prove the global well-posedness of (1.1) in this case is due to the \(\theta \) component of the vorticity, \(\omega ^\theta ,\) satisfies
The scaling invariant Lebesgue space for \(\frac{\omega ^\theta }{r}\) is \(L^\infty (]0,t[;L^1(\mathop {\mathbb R}\nolimits ^3))\). Motivated by [5], the purpose of this paper is to improve the norm for the initial data in [11] to be scaling invariant ones. We remark that the other motivation of this paper comes from [3] where the authors proved that one scaling invariant norm to one component of Navier–Stokes system controls the regularity of the solution. Yet we still do not know in general the global well-posedness of Navier–Stokes with one component being small in some scaling invariant space.
Now we restrict ourselves to the axisymmetric solutions of (1.1) with the following form
where \((r,\theta ,z)\) denotes the usual cylindrical coordinates in \(\mathop {\mathbb R}\nolimits ^3\) so that \(x=(r\cos \theta ,r\sin \theta ,z)\), and
Then in this case, we can reformulate (1.1) as
Let us denote \(\widetilde{u}\buildrel \mathrm{def}\over =u^r e_r+u^z e_z.\) Then it is easy to check that
so that the Biot–Savart law shows that \(u^r\) and \(u^z\) can be uniquely determined by \(\omega ^\theta \) (see Sect. 2.1). Hence we can write the System (1.3) as
Here and all in that follows, we always denote \(\mathrm{div}\,_* f\buildrel \mathrm{def}\over =\partial _r f^r+\partial _z f^z\) and abuse the notation \(\tilde{u}=(u^r,u^z).\)
As in [5], we shall equip the half-plane \(\Omega =\{(r,z)|r>0,z\in \mathop {\mathbb R}\nolimits \}\) with the 2D measure drdz, instead of the 3D measure rdrdz. For any \(p\in [1,\infty [\), we denote by \(L^p(\Omega )\) the space of measurable functions \(f{:}\,\Omega \rightarrow \mathop {\mathbb R}\nolimits \) which verifies
The space \(L^\infty (\Omega )\) can be defined with the usual modification. Sometimes, we shall also use the 3D Lebesgue measure rdrdz, and the corresponding Lebesgue spaces are then denoted by \(L^p(\mathop {\mathbb R}\nolimits ^3)\) or \(L^p\) with norm
Our main results state as follows.
Theorem 1.1
For any initial data \(\omega ^\theta _0\in L^1(\Omega )\) and \(u^\theta _0\in L^2(\Omega )\) satisfying \(r^{-\frac{3}{10}}u^\theta _0\in L^{\frac{20}{13}}(\Omega )\), there exists some \(T(\omega ^\theta _0,u^\theta _0)\) such that the equations (1.4) have a unique mild solution
Furthermore, the solution \((\omega ^\theta , u^\theta )\) verifies
-
for any \(p\in [1,\infty ],~q\in [2,\infty ]\) and \(\kappa \in [{20}/{13},\infty ],\) there holds
$$\begin{aligned} \begin{aligned}&L_p(T)\buildrel \mathrm{def}\over =\sup _{0\le t\le T}t^{1-\frac{1}{p}}\Vert \omega ^\theta (t)\Vert _{L^p(\Omega )}<\infty ,\quad M_q(T)\buildrel \mathrm{def}\over =\sup _{0\le t\le T}t^{\frac{1}{2}-\frac{1}{q}}\Vert u^\theta (t)\Vert _{L^q(\Omega )}<\infty ,\\&N_\kappa (T)\buildrel \mathrm{def}\over =\sup _{0\le t\le T}t^{\frac{13}{20}-\frac{1}{\kappa }}\Vert r^{-\frac{3}{10}}{u^\theta }(t)\Vert _{L^\kappa (\Omega )}<\infty . \end{aligned} \end{aligned}$$(1.7)Moreover, when \(p\in ]1,\infty ],~q\in ]2,\infty ]\) and \(\kappa \in ]{20}/{13},\infty ],\) we have
$$\begin{aligned} \lim _{t\rightarrow 0}\left( L_p(t)+M_q(t)+N_\kappa (t)\right) =0; \end{aligned}$$(1.8) -
if
$$\begin{aligned} \Vert \omega ^\theta _0\Vert _{L^1(\Omega )}+\Vert u^\theta _0\Vert _{L^2(\Omega )} +\Vert r^{-\frac{3}{10}}u^\theta _0\Vert _{L^{\frac{20}{13}}(\Omega )}\le c \end{aligned}$$(1.9)for some sufficiently small constant c, then \(T=\infty .\) And if \(\Vert u^\theta _0\Vert _{L^2(\Omega )}+\Vert r^{-\frac{3}{10}}u^\theta _0\Vert _{L^{\frac{20}{13}}(\Omega )}\) is small enough, then the lifespan \(T^\star \) of the solution depends only on \(\omega ^\theta _0\).
Remark 1.1
-
Let us remark that the norms \(\Vert \omega ^\theta _0\Vert _{L^1(\Omega )},~\Vert u^\theta _0\Vert _{L^2(\Omega )}\) and \(\Vert r^{-\frac{3}{10}}u^\theta _0\Vert _{L^{\frac{20}{13}}(\Omega )}\) are scaling invariant under the scaling transformation (1.2). Moreover, the method used here might be used to study axi-symmetric vortex ring for 3-D Navier–Stokes system with swirl (see the corresponding result of [4] for the case without swirl).
-
The reason for requiring \(r^{-\frac{3}{10}}u^\theta _0\in L^{\frac{20}{13}}(\Omega )\) is to handle the term \(\frac{\partial _z|u^\theta (s)|^2}{r}\) in \(\omega ^\theta \) equation of (3.1), so that the exponent \(\frac{3}{2}-\frac{1}{p}+\frac{1}{5}\) appearing in (3.7) is less than 1.
Theorem 1.2
Let \(\omega ^\theta _0\) and \(u^\theta _0\) satisfy \(\eta _0\buildrel \mathrm{def}\over =\frac{\omega ^\theta _0}{r}\in L^1, ~U_0\buildrel \mathrm{def}\over =\frac{u^\theta _0}{r}\in L^{\frac{3}{2}}\), \(ru^\theta _0\in L^{A}\bigcap L^\infty \) for some finite A. We assume that \(\Vert ru^\theta _0\Vert _{L^\infty (\Omega )}\) is sufficiently small, then the system (1.4) has a unique global solution which satisfies
Remark 1.2
-
The main difficulty in the proof of the above theorem is when \(\omega ^\theta _0\in L^p(\Omega )\) for \(p=1,\) the dissipative term, \(\frac{4(p-1)}{p^2}\Vert \nabla |\eta |^{\frac{p}{2}}\Vert _{L^2}^2,\) in (4.7) disappears. That is the reason why we divide the proof of Theorem 1.2 in the following two steps: we first get, by applying Theorem 1.1, that the system (1.4) has a unique local solution with \(\eta (t_0)\in L^{p_0}(\mathop {\mathbb R}\nolimits ^3)\) for some \(t_0>0\) and \(p_0>1;\) then in the second step, starting with initial data at \(t_0,\) we prove the global well-posedness of the system (1.4).
-
One may see (5.6), (5.17), (5.19) and (5.24) for the exact smallness condition for \(\Vert r u^\theta _0\Vert _{L^\infty }\). And the exact global estimate of \(\Vert \eta (t)\Vert _{L^1}\) and \(\Vert U(t)\Vert _{L^{\frac{3}{2}}}\) is given in (5.25).
-
It follows from Lemma 2.1 below and Hölder’s inequality
that the solutions constructed in Theorem 1.2 in fact satisfy
2 Preliminaries
2.1 Some elementary results
Lemma 2.1
(Proposition 1 of [2]) Let \((u^r,u^\theta , u^z)\) be a smooth enough solution of (1.3) on [0, T]. Then for any \(p\in [2,\infty ]\), we have
Lemma 2.2
(See Lemma 5.5 from [1] for instance) Let E be a Banach space, \(\mathfrak {B}(\cdot ,\cdot )\) a continuous bilinear map from \(E\times E\) to E, and \(\mathfrak {\alpha }\) a positive real number such that
Then for any a in the ball \(B(0,\mathfrak {\alpha })\) in E, there exists a unique x in \(B(0,2\mathfrak {\alpha })\) such that
Let us recall also some facts from Sect. 2 of [5]. We first recall the axisymmetric Biot–Savart law which determines \(\widetilde{u}=(u^r,u^z)\) in terms of \(\omega ^\theta ,\) namely
where
It follows from the Remark 2.2 of [5] that
Lemma 2.3
\(s^\alpha F(s)\) and \(s^\beta F'(s)\) are bounded on \(]0,\infty [\) for \(\alpha \in ]0,3/2]\) and \(\beta \in [1,5/2]\).
Lemma 2.4
(Proposition 2.3 of [5]) Let us denote \(\widetilde{u}\buildrel \mathrm{def}\over =(u^r,u^z).\) Then one has
-
(i)
Assume that \(1<p<2<q<\infty \) and \(\frac{1}{q}=\frac{1}{p}-\frac{1}{2}\). If \(\omega ^\theta \in L^p(\Omega )\), then \(\widetilde{u}\in L^q(\Omega )\) and
-
(ii)
If \(1\leqslant p<2<q\leqslant \infty \) and \(\omega ^\theta \in L^p(\Omega )\bigcap L^q(\Omega )\), then \(\widetilde{u}\in L^\infty (\Omega )\) and
$$\begin{aligned} \Vert \widetilde{u}\Vert _{L^\infty (\Omega )}\leqslant C\Vert \omega ^\theta \Vert _{L^p(\Omega )}^\sigma \Vert \omega ^\theta \Vert _{L^q(\Omega )}^{1-\sigma }, \quad \hbox {where}\quad \sigma =\frac{p(q-2)}{2(q-p)}\in ]0,1[. \end{aligned}$$(2.5)
Next we investigate the solution operator S(t) to the linearized system of (1.4), namely \(\omega ^\theta (t)=S(t)\omega _0\) verifies
Lemma 2.5
(Lemma 3.1, 3.2 of [5]) For any \(t>0\), one has
where the function \({H}:]0,+\infty [\rightarrow \mathop {\mathbb R}\nolimits \) is defined by
which is smooth on \(]0,\infty [\) and has the asymptotic expansions:
-
(i)
\({H}(t)=\frac{\pi ^{1/2}}{4t^{3/2}}+\mathcal {O}\left( \frac{1}{t^{5/2}}\right) ,\ {H}'(t)=-\frac{3\pi ^{1/2}}{8t^{5/2}}+\mathcal {O}\left( \frac{1}{t^{7/2}}\right) \), as \(t\rightarrow \infty \);
-
(ii)
\({H}(t)=1-\frac{3t}{4}+\mathcal {O}(t^2),\ {H}'(t)=-\frac{3}{4}+\mathcal {O}(t)\), as \(t\rightarrow 0\).
Corollary 2.1
\(t^\alpha H(t)\) and \(t^\beta H'(t)\) are bounded on \(]0,\infty [\) provided \(0\le \alpha \le \frac{3}{2}\), \(0\le \beta \le \frac{5}{2}\).
2.2 The estimate of \(\frac{u^r}{r}\) in terms of \(\frac{\omega ^\theta }{r}\)
In this subsection, we shall exploit the basic facts recalled in Sect. 2.1 to derive the estimate of \(\frac{u^r}{r}\) in terms of \(\frac{\omega ^\theta }{r}\), which will be used in Sect. 4 below. The main result states as follows:
Proposition 2.1
Let \(p\in ]1,3[\) and \(q\in \bigl ]\frac{3p}{3-p},\infty \bigr ].\) We assume that \(\eta \buildrel \mathrm{def}\over =\frac{\omega ^\theta }{r}\in L^p(\mathop {\mathbb R}\nolimits ^3)\cap L^{3p}(\mathop {\mathbb R}\nolimits ^3)\). Then we have
Proof
By virtue of (2.2) and (2.3), we write
We decompose the integral domain \(\Omega =I_1\bigcup I_2\) with
We first consider the case when \(q<\infty \). Let s be determined by \(\frac{1}{s}=\frac{1}{q}+\frac{1}{3}.\) Then due to \(q\in \bigl ]\frac{3p}{3-p},\infty \bigr [,\) we have \(s>p.\) Moreover, it follows from Lemma 2.3 that \(|F'(s)|\lesssim (\frac{1}{s})^{\frac{7}{6}}=\left( \frac{1}{s}\right) ^{1+\frac{1}{2}\left( \frac{1}{s}-\frac{1}{q}\right) }.\) Note that that \(\frac{\bar{r}}{r}\leqslant 2\) in \(I_1\) and \(\frac{3}{2}-\frac{1}{2}(\frac{1}{s}+\frac{1}{q})>0,\) we thus obtain
from which, and Hardy–Littlewood–Sobolev inequality, we infer
Note that in the region \(I_2,\) there holds \(\bar{r}\le 2|\bar{r}-r|\). Thus by using Lemma 2.3, we get
To proceed further, for any given \(R>0,\) we split \(I_2=I_{21}\cup I_{22}\) with
Then we get, by applying Young’s inequality, that
For the integral on \(I_{22}\), in the case \(q>3p\), by applying Young’s inequality, we get
While in the case \(\frac{3p}{3-p}<q\le 3p\), another use of Young’s inequality gives
As a result, it comes out
Taking \(R=\left( \frac{\Vert \eta \Vert _{L^p}}{\Vert \eta \Vert _{L^{3p}}}\right) ^{\frac{p}{2}}\) in the above inequality gives rise to
Due to \(\bigl \Vert r^{\frac{1}{q}}\frac{u^r}{r}\bigr \Vert _{L^q(\Omega )}=\bigl \Vert \frac{u^r}{r}\bigr \Vert _{L^q},\) (2.12) together with (2.13) ensures (2.9) for any \(q\in \bigl ]\frac{3p}{3-p},\infty \bigr [\).
The end-point case when \(q=\infty \) follows exactly along the same line. This completes the proof of Proposition 2.1. \(\square \)
2.3 The estimates of the solution operator S(t)
The goal of this subsection is to present the estimates of the solution operator S(t), which will be used in Sect. 3.
Proposition 2.2
Let S(t) be the solution operator given by (2.7). Then this family \(\left( S(t)\right) _{t\ge 0}\) are strongly continuous semigroups of bounded linear operators in \(L^m(\Omega )\) for any \(m\in [1,\infty [\). Moreover, for \(1\le p\le q\le \infty \), there holds
-
1.
For any \(\alpha ,~\beta \) satisfying \(\alpha +\beta \le 0,~a\ge -1\) and \(\beta \ge -1\), and any \(f=(f^r,f^z)\in L^p(\Omega )^2\), there holds
$$\begin{aligned} \Vert r^\alpha S(t)\mathrm{div}\,_* (r^{\beta }f)\Vert _{L^q(\Omega )}\le \frac{C}{t^{\frac{1}{2}-\frac{\alpha +\beta }{2}+\frac{1}{p}-\frac{1}{q}}}\Vert f\Vert _{L^p(\Omega )}. \end{aligned}$$(2.14)In particular, taking \(\alpha =\beta =0\), we have
$$\begin{aligned} \Vert S(t)\mathrm{div}\,_* f\Vert _{L^q(\Omega )}\le \frac{C}{t^{\frac{1}{2}+\frac{1}{p}-\frac{1}{q}}}\Vert f\Vert _{L^p(\Omega )}. \end{aligned}$$(2.15) -
2.
For any \(\alpha ,~\beta \) satisfying \(\alpha +\beta \le 1,~\alpha \ge -1\) and \(\beta \ge -1\), and any \(g\in L^p(\Omega )\), there holds
$$\begin{aligned} \Vert r^\alpha S(t) (r^{\beta -1} g)\Vert _{L^q(\Omega )} \le \frac{C}{t^{\frac{1}{2}-\frac{\alpha +\beta }{2}+\frac{1}{p}-\frac{1}{q}}}\Vert g\Vert _{L^p(\Omega )}. \end{aligned}$$(2.16)In particular, taking \(\alpha =0,~\beta =1\), and \(\alpha =\beta =0\), we have
$$\begin{aligned} \Vert S(t) g\Vert _{L^q(\Omega )} \le \frac{C}{t^{\frac{1}{p}-\frac{1}{q}}}\Vert g\Vert _{L^p(\Omega )},\quad \Vert S(t) \left( \frac{g}{r}\right) \Vert _{L^q(\Omega )} \le \frac{C}{t^{\frac{1}{2}+\frac{1}{p}-\frac{1}{q}}}\Vert g\Vert _{L^p(\Omega )}. \end{aligned}$$(2.17) -
3.
For any \(\delta \in \bigl [-\,1,\frac{1}{2}\bigr ],~m\in [1,\infty [\), and any g satisfying \(r^\delta g\in L^m(\Omega )\), we have
$$\begin{aligned} \Vert r^\delta S(t) g-r^\delta g\Vert _{L^m(\Omega )}\rightarrow 0,\quad \text{ as }\ t\rightarrow 0. \end{aligned}$$(2.18)
Proof
The boundedness of the semigroup \(\left( S(t)\right) _{t\ge 0}\) is shown in (2.17). Then in order to prove \(\left( S(t)\right) _{t\ge 0}\) is strongly continuous in \(L^m(\Omega )\) for any \(m\in [1,\infty [\), we only need to verify the continuity at the origin, which is a direct consequence of (2.18) (with \(\delta =0\)). Hence it remains to prove the estimates (2.14–2.17), which we handle term by term below.
-
1.
By integration by parts, we write
where
-
Let us first handle the term \(\bigl |A_r+\frac{r-\bar{r}}{2t}{H}\left( \frac{t}{r\bar{r}}\right) \bigr |\).
If \((\alpha ,\beta )\in \Omega _1\buildrel \mathrm{def}\over =\{(\alpha ,\beta )|0\le \frac{1}{2}-\frac{\alpha +\beta }{2}\le \frac{3}{2} ,~0\le \frac{1}{2}-\beta \le \frac{3}{2},~\beta \le \alpha \}\), we can divide the integral area into \(\{\bar{r}\ge \frac{r}{2}\}\) and \(\{\bar{r}< \frac{r}{2}\}\). When \(\bar{r}\ge \frac{r}{2}\), we can deduce from Corollary 2.1 that
And when \(\bar{r}<\frac{r}{2}\), there then holds \({r}<2|\bar{r}-r|\), another use of Corollary 2.1 gives
If \((\alpha ,\beta )\in \Omega _2\buildrel \mathrm{def}\over =\{(\alpha ,\beta )|0\le \frac{1}{2}-\frac{\alpha +\beta }{2}\le \frac{3}{2} ,~0\le \frac{1}{2}-\alpha \le \frac{3}{2},~\alpha \le \beta \}\), we divide the integral area in a different way as \(\{\bar{r}\le 2r\}\) and \(\{\bar{r}>2r\}\). Similar to the previous estimates, when \(\bar{r}\le 2r\), we have
And when \(\bar{r}> 2r\), there then holds \(\bar{r}<2|\bar{r}-r|\), thus we deduce
Thus combining the estimates (2.14–2.17), we conclude that whenever \((\alpha ,\beta )\in \Omega _1\bigcup \Omega _2\), i.e. \(\alpha ,~\beta \) satisfy \(\alpha +\beta \le 1\), \(\alpha \ge -1\) and \(\beta \ge -1\), there holds
-
For \(|A_z|\) term in the integrand (2.19). When \(\bar{r}>2r\) or \(\bar{r}\le \frac{r}{2}\), there then holds \(\bar{r}+r<3|\bar{r}-r|\). If in addition \(\alpha +\beta \ge -4\), \(\alpha \ge -1\) and \(\beta \ge -2\), there then exists a positive constant \(\gamma \) so that \(\max \bigl \{0,\frac{1}{2}-\alpha ,-\frac{1}{2}-\beta ,-\frac{1+\alpha +\beta }{2}\bigr \}\le \gamma \le \frac{3}{2}\). Then we deduce from Corollary 2.1 that
$$\begin{aligned}&\frac{\bar{r}^{\frac{1}{2}+\beta }}{r^{\frac{1}{2}-\alpha }} \exp \left( -\frac{\left( r-\bar{r}\right) ^2+\left( z-\bar{z}\right) ^2}{4t}\right) \left( |A_z|+|\frac{r-\bar{r}}{2t}{H}\left( \frac{t}{r\bar{r}}\right) |\right) \\&\quad \lesssim \frac{\bar{r}^{\frac{1}{2}+\beta }}{r^{\frac{1}{2}-\alpha }}\frac{|r-\bar{r}|+|z-\bar{z}|}{t}|\frac{r\bar{r}}{t}|^{\gamma } \left( \frac{5t}{\left( r-\bar{r}\right) ^2+\left( z-\bar{z}\right) ^2}\right) ^{\frac{1+2\gamma +\alpha +\beta }{2}}\\&\qquad \times \exp \left( -\frac{\left( r-\bar{r}\right) ^2+\left( z-\bar{z}\right) ^2}{5t}\right) \\&\quad \lesssim \frac{1}{t^{\frac{1}{2}-\frac{\alpha +\beta }{2}}}\exp \left( -\frac{\left( r-\bar{r}\right) ^2+\left( z-\bar{z}\right) ^2}{5t}\right) . \end{aligned}$$And when \(\frac{r}{2}\le \bar{r}\le 2r\), if in addition, \(-3\le \alpha +\beta \le 0\), we have
$$\begin{aligned}&\frac{\bar{r}^{\frac{1}{2}+\beta }}{r^{\frac{1}{2}-\alpha }} \exp \left( -\frac{\left( r-\bar{r}\right) ^2+\left( z-\bar{z}\right) ^2}{4t}\right) \left( |A_z|+|\frac{r-\bar{r}}{2t}{H}\left( \frac{t}{r\bar{r}}\right) |\right) \\&\quad \lesssim \frac{\bar{r}^{\frac{1}{2}+\beta }}{r^{\frac{1}{2}-\alpha }} \frac{|r-\bar{r}|+|z-\bar{z}|}{t}|\frac{r\bar{r}}{t}|^{-\frac{\alpha +\beta }{2}} \left( \frac{5t}{\left( r-\bar{r}\right) ^2+\left( z-\bar{z}\right) ^2}\right) ^{\frac{1}{2}}\\&\qquad \times \exp \left( -\frac{\left( r-\bar{r}\right) ^2+\left( z-\bar{z}\right) ^2}{5t}\right) \\&\quad \lesssim \frac{1}{t^{\frac{1}{2}-\frac{\alpha +\beta }{2}}}\cdot \exp \left( -\frac{\left( r-\bar{r}\right) ^2+\left( z-\bar{z}\right) ^2}{5t}\right) . \end{aligned}$$
Therefore as long as \(\alpha +\beta \le 0\), \(\alpha \ge -1\) and \(\beta \ge -2\), there holds
By combining (2.24) with (2.25), we achieve
provided \(\alpha +\beta \le 0,~a\ge -1\) and \(\beta \ge -1\). And then (2.14) follows from (2.19) and Young’s inequality in two space dimension.
-
2.
It follows from the proof of (2.24) that
$$\begin{aligned}&\frac{1}{r^{\frac{1}{2}-\alpha }\bar{r}^{\frac{1}{2}-\beta }} \bigl |{H}\left( \frac{t}{r\bar{r}}\right) \bigr | \exp \left( -\frac{(r-\bar{r})^2+(z-\bar{z})^2}{4t}\right) \\&\quad \lesssim \frac{1}{t^{\frac{1}{2}-\frac{\alpha +\beta }{2}}}\exp \left( -\frac{(r-\bar{r})^2+(z-\bar{z})^2}{5t}\right) , \end{aligned}$$whenever \(\alpha +\beta \le 1,~\alpha \ge -1\) and \(\beta \ge -1\). Then by virtue of (2.7), we get, by applying Young’s inequality, that there holds (2.16).
-
3.
In view of (2.7), we get, by using changes of variables that
$$\begin{aligned} \left( r^\delta S(t) g-r^\delta g\right) (r,z)=\frac{1}{4\pi }\int _{\Omega } \Psi (r,z,\rho ,\xi ,t)\cdot \exp \left( -\frac{\rho ^2+\xi ^2}{4}\right) \, d\rho d\xi , \end{aligned}$$(2.26)for all \((r,z)\in \Omega \), and where
$$\begin{aligned} \Psi (r,z,\rho ,\xi ,t)\buildrel \mathrm{def}\over =r^\delta \left( \frac{r+\sqrt{t}\rho }{r}\right) ^{\frac{1}{2}} {H}\left( \frac{t}{r\left( r+\sqrt{t}\rho \right) }\right) g\left( r+\sqrt{t}\rho ,z+\sqrt{t}\xi \right) -r^\delta g(r,z). \end{aligned}$$(2.27)
Notice that \(\frac{1}{2}-\delta \in [0,\frac{3}{2}]\), applying Corollary 2.1 gives
which implies for any given \((\rho ,\xi , t)\), we have
Moreover, noting that \(H(t)=1+\mathcal {O}(t),\) as \(t\rightarrow 0\), it is easy to observe that \(\Psi (r,z,\rho ,\xi ,t)\rightarrow 0\) as \(t\rightarrow 0\). Then Lebesgue dominated convergence theorem ensures that
from which and (2.26), another use of Lebesgue’s dominated convergence theorem gives
This completes the proof of the proposition. \(\square \)
3 Local existence of solutions to (1.1) in critical spaces
The purpose of this section is to investigate the local existence and uniqueness of the mild solutions to (1.4) in the spirit of [5, 6]. In view of (2.7), we rewrite the systems (1.4) as
Now we present the proof of Theorem 1.1.
Proof of Theorem 1.1
The main idea to prove Theorem 1.1 is to apply fixed point argument for the integral formulation (3.1). Toward this, for any \(T>0\), we introduce the functional space
where
For convenience, sometimes we may abuse the notation \(\Vert \omega ^\theta \Vert _{X_T}=\sup \limits _{0<t\le T}t^{\frac{1}{4}}\Vert \omega ^\theta \Vert _{L^{\frac{4}{3}}(\Omega )}\), \(\Vert u^\theta \Vert _{X_T}=\sup \limits _{0<t\le T}\left( t^{\frac{1}{4}}\Vert u^\theta (t)\Vert _{L^{4}(\Omega )}+t^{\frac{3}{20}}\Vert r^{-\frac{3}{10}}u^\theta (t)\Vert _{L^{2}(\Omega )}\right) \), and \(\omega ^\theta \in X_T\) (resp. \(u^\theta \in X_T\)) means that \(\Vert \omega ^\theta \Vert _{X_T}<\infty \) (resp. \(\Vert u^\theta \Vert _{X_T}<\infty \)).
-
The estimate of \(\omega ^\theta \) term
In view of (2.17), \(S(t)\omega ^\theta _0\in X_T\) for any \(T>0\), and there exists a universal constant \(C_1>0\) such that for any \(T>0\), we have
On the other hand, since \(L^1(\Omega )\bigcap L^{\frac{4}{3}}(\Omega )\) is dense in \(L^1(\Omega ).\) For any \(\varepsilon >0,\) there exists \(\widetilde{\omega }^\theta _0\in L^1(\Omega )\bigcap L^{\frac{4}{3}}(\Omega )\) satisfying \(\Vert \widetilde{\omega }^\theta _0-\omega ^\theta _0\Vert _{L^1(\Omega )}<\varepsilon .\) Then it follows from (2.17) that
which implies
Let us denote \(\widetilde{u}(\omega ^\theta )(t)\) be velocity field determined by the vorticity \(\omega ^\theta e_\theta \) via the axisymmetric Biot–Savart law (2.2). Given \(\left( \omega ^\theta _1,u^\theta _1\right) ,~\left( \omega ^\theta _2,u^\theta _2\right) \in X_T\), for any \(t\in [0,T],\) we define the mapping \(\mathcal {F}^\omega \) on \(X_T\times X_T\) by
Then for any \(p\in [1,\frac{10}{7}[,\) we deduce from (2.14) (with \(~\alpha =0,~\beta =-\frac{2}{5}\)), (2.15) and (2.4) that for any \(t\in ]0,T]\),
-
The estimate of \(u^\theta \) terms
Thanks to (2.16) (with \(\alpha =-\frac{3}{10},~\beta =\frac{13}{10}\)), by a similar derivation of (3.4), (3.5), we get
Given \(\left( \omega ^\theta _1,u^\theta _1\right) ,~\left( \omega ^\theta _2,u^\theta _2\right) \in X_T,\) for any \(t\in [0,T],\) we define the mapping \(\mathcal {F}^u\) on \(X_T\times X_T\) by
Then for any \(q\in [2,\infty [\), we deduce from (2.15), (2.17) and (2.4) that for any \(t\in ]0,T]\)
Along the same line, for any \(\kappa \in [\frac{20}{13},4[\), we deduce from (2.14), (2.16) (with \(\alpha =-\frac{3}{10},~\beta =\frac{3}{10}\)) and (2.4) that for any \(t\in ]0,T]\)
-
Fixed point argument
For \(\left( \omega ^\theta _1,u^\theta _1\right) ,~\left( \omega ^\theta _2,u^\theta _2\right) \in X_T\), we consider the following bilinear map
with \({\mathcal F}^\omega ,~{\mathcal F}^u\) given by (3.6) and (3.9) respectively.
By virtue of (3.5) and (3.8), for any \(\omega ^\theta _0 \in L^1(\Omega ),~u^\theta _0\in L^2(\Omega )\) with \(r^{-\frac{3}{10}}u^\theta _0\in L^{\frac{20}{13}}(\Omega )\), there exists a positive time T such that
where the constants \(A^\omega _{4/3},~A^u_4\) and \(B^u_2\) are determined by (3.7), (3.10) and (3.11) respectively. Then we deduce from Lemma 2.2 that (3.1) has a unique solution \((\omega ^\theta , u^\theta )\) in \(X_T\). Furthermore, by virtue of (3.5) and (3.8), for any \(\varepsilon >0,\) there exists \(T_\varepsilon >0\) so that
Then Lemma 2.2 ensures that
which implies that
Futhermore, for any \(T>0\), it follows from (3.4) and (3.8) that
provided that c in (1.9) satisfying \(c\le \frac{1}{4C_1\left( A^\omega _{4/3}+A^u_4+B^u_2\right) }.\) This together with Lemma 2.2 shows that (3.1) has a unique global solution in \(X_\infty .\)
-
Behavior near \(t=0\)
Let us now turn to the estimate (1.8). Let \((\omega ^\theta , u^\theta )\) be the unique solution of (3.1) on [0, T] obtained in the previous one step, we denote
Along the same line to the proof of (3.5), it is easy to observe that
While it follows from the proof of (3.7), (3.10) and (3.11) that
from which, (2.18) and (3.14), we infer
Whereas for any \(t_0>0,\) we get, by using a similar derivation of (3.7), that
Hence by virtue of the expression (3.1), we deduce that
Exactly along the same line, we can prove that
This together with (3.16) and (3.17) ensures that
For any \(q\in ]2,\infty ]\), by using (3.10), (3.14) and (3.15), we deduce that \(M_q(T)\) are bounded, and \(M_q(T)\rightarrow 0\) as \(T\rightarrow 0\).
For the estimate of \(N_\kappa (T)\) and \(L_p(T)\), we shall use a bootstrap argument. Indeed to estimate \(N_\kappa (T)\), we get, by a similar derivation of (3.11), that
where the exponents \(\kappa \in ]{20}/{13},\infty ],~\kappa _1\in ]{20}/{13},\infty ]\) satisfying
Meanwhile it follows from (3.1) and (3.19) that
Note that \(J_{p,q,\kappa }(T),~\Vert \omega ^\theta \Vert _{X_T},~N_2(T)\rightarrow 0\) as \(T\rightarrow 0\), by taking \(\kappa _1=2\) in (3.21), it follows from (3.20) that for any \(\kappa \in ]{20}/{13},4[\), we have
Next, taking \(\kappa _1=3\) in (3.21), we deduce from (3.20) that (3.22) holds for any \(\kappa \in [4,12[\). Along the same line, taking \(\kappa _1=10\) in (3.21) ensures (3.22) for any \(\kappa \in [12,\infty ]\). Hence we prove that (3.22) holds for any \(\kappa \in ]{20}/{13},\infty ]\).
To handle \(L_p(T)\), we get, by a similar derivation of (3.7), that but here we need to split the integral area in two parts,
where the exponents \(p,~p_1\in ]1,\infty ]\) satisfying
Then we deduce from (3.1) and (3.23) that
Note that \(J_{p,q,\kappa }(T),~L_{\frac{4}{3}}=\Vert \omega ^\theta \Vert _{X_T},~\Vert u^\theta \Vert _{X_T},~N_{2p}(T)\rightarrow 0\) as \(T\rightarrow 0\), by taking \(p_1=\frac{4}{3}\) in (3.25), we deduce from (3.24), that
for any \(p \in ]1,2[\). Next taking \(p_1=\frac{5}{3}\) in (3.25) ensures that (3.26) holds for any \(p\in [2,5[\). Similarly, taking \(p_1=\frac{5}{2}\) in (3.25) implies that (3.26) holds for any \(p\in [5,\infty ]\). Thus (3.26) holds for any \(p\in ]1,\infty ]\). This completes the proof of Theorem 1.1. \(\square \)
Let \(L_p(T),~M_q(T),~N_\kappa (T)\) be given by (1.7), for any \(s\in \mathop {\mathbb N}\nolimits ^+\), we shall denote
For later use, we state the following result.
Corollary 3.1
For any \(0\le \gamma \le \delta \le 1,~1\le p\le \infty ,~1\le q_2\le q_1\le \infty \), under the assumptions of Theorem 1.1, if we assume moreover that \(r^{-\gamma }u^\theta _0\in L^{q_2}(\Omega )\), then there holds
Proof
When \(q_1\in [1,10]\), in view of (3.1), we get, by applying Proposition 2.2 and then Lemma 2.4, that
which yields (3.28) for \(q_1\in [1,10]\).
When \(q_1\in ]10,\infty ]\), we get, by a similar derivation, that
This proves the estimate (3.28).
To handle the estimate (3.29), we first consider the case when \(\delta \in [0,\frac{1}{2}]\) and \(p\in [1,5]\). In this case, applying Proposition 2.2 to (3.1) and then using the estimate (3.28) gives rise to
where in the last step we use the fact that \(\Vert \omega ^\theta _0\Vert _{L^1(\Omega )}\le L_1,~\Vert u^\theta _0\Vert _{L^2(\Omega )}\le M_2.\)
On the other side, when \(p\in ]5,\infty ]\), we have
Combining (3.30) with (3.31) leads to the estimate (3.29) for \(\delta \in [0,\frac{1}{2}]\).
For the remaining case when \(\delta \in ]1/2,1]\), we first get, by a similar derivation of (3.30) and then using (3.28) that for \(p\in [1,5]\)
Finally when \(p\in ]5,\infty ]\), we deduce, by a similar derivation of (3.31), that
Combining (3.32) with (3.33) implies (3.29) for \(\delta \in ]\frac{1}{2},1]\). This completes the proof of the estimate (3.29) and hence the corollary. \(\square \)
4 Global a priori estimates of (1.4) with nearly critical initial data
The goal of this section is basically to prove that, as long as the initial data belongs to the almost critical spaces, the system has a unique global solution.
Let us introduce another two variables which are of great importance in our work, namely
And it is not difficult to deduce the equations for \(\eta \) and \(V^\varepsilon \) from (1.4) that
here and in all that follows, we always denote \(V^\varepsilon \) as V, if there is no ambiguity.
Proposition 4.1
Let \((u^r,u^\theta ,u^z)\) be a smooth enough solution of (1.3) on [0, T]. Let \(p\in ]1,\frac{21}{20}]\), \(\varepsilon =\frac{-9p^2+21p-4}{24p-2}\in \bigl [\frac{3251}{9280},\frac{4}{11}\bigr [\), and q be given by
We assume that the initial data \(\eta _0\in L^p\), \(V_0^\varepsilon =\frac{u^\theta }{r^{1-\varepsilon }}\in L^q\), \(ru^\theta _0\in L^\infty \bigcap L^{\frac{1}{p-1}},\) which satisfy
for some sufficiently small constant \(c_0\) which does not depend on the choice of p, and \(M_0\buildrel \mathrm{def}\over =\Vert V_0\Vert _{L^q}^q+\Vert \eta _0\Vert _{L^p}^p\). Then for any \(t\in [0,T]\), we have
Let us remark that both the index \(\frac{10(12p-1)}{9(p-1)(p+2)(p+3)}\) and \({\frac{2}{3(p-1)^2}}\) are close enough to \(\infty \) as long as p approaches 1, which corresponds to the case with initial data in the critical spaces.
The proof of the above proposition relies on the following lemmas:
Lemma 4.1
Under the assumptions of Proposition 4.1, for any \(t\in ]0,T],\) there holds
Lemma 4.2
Under the assumptions of Proposition 4.1, for any \(t\in ]0,T],\) there holds
Let us admit the above lemmas and continue our proof of Proposition 4.1.
Proof of Proposition 4.1
By virtue of Lemma 2.1, we get, by summarizing (4.6) for \(\varepsilon =\frac{-9p^2+21p-4}{24p-2}\) with (4.7), that
Let \(M_0\buildrel \mathrm{def}\over =\Vert V_0\Vert _{L^q}^q+\Vert \eta _0\Vert _{L^p}^p\), and let \(T'>0\) be determined by
If \(T'<T,\) then for \(t\le T',\) we deduce from the (4.8) that
which together with (4.4) ensures that for any t in \([0,T']\),
This in particular gives rise to
This contradicts with the definition of \(T'\) given by (4.9). As a result, it comes out \(T'=T\), and there holds (4.5) for any \(t\in [0,T].\) this completes the proof of the proposition. \(\square \)
Let us now turn to the proof of Lemmas 4.1 and 4.2.
Proof of Lemma 4.1
For any \(p\in ]1,\frac{21}{20}]\) and q given by (4.3), we get, by multiplying the second equation of (4.2) by \(|V|^{q-2}V\) and then integrating the resulting equality over \(\mathop {\mathbb R}\nolimits ^3,\) that
Using the fact that \(\partial _r(ru^r)+\partial _z(ru^z)=0\), which implies \(\int _{\mathop {\mathbb R}\nolimits ^3}(u^r\partial _r+u^z\partial _z)|V|^{q}\,dx=0\), and the homogeneous Dirichlet boundary condition for \(u^r\) on \(r=0,\) we deduce
which implies
Let us take \(q_1,q_2\) satisfying \( \frac{1}{q_1}+\frac{1}{q_2}=1,\) and temporarily assume that \(q_1\in \bigl ]\frac{3p}{3-p},\infty \bigr [,\) so that it follows from Lemma 2.1 and Sobolev embedding Theorem that
Take \(q_2=\vartheta +\sigma +\frac{2\sigma }{3p},\) with \(\vartheta >0\) and \(0<\sigma <1\) to be determined later, then we have
where in the last step, we use the interpolation inequality provided that
which will be verified later. Then we get, by applying Young’s inequality, that
provided that
and this can be satisfied by choosing
It is easy to verify that for any \(p\in \bigl ]1,\frac{21}{20}\bigr ]\), the corresponding \(\frac{2p\vartheta }{1-p\sigma }=\frac{2(p+2)(6p-5)}{3(4-3p)}\) is exactly in [2, 6], and \(q_2\) is exactly in \(\bigl ]1,\frac{3p}{4p-3}\bigr [\), thus the conjugate number \(q_1=\left( 1-\frac{1}{q_2}\right) ^{-1}>\frac{3p}{3-p}\). Thus all the above calculations go through.
With the indexes given by (4.16), by inserting the Estimate (4.15) into (4.12), we achieve (4.6). This completes the proof of Lemma 4.1. \(\square \)
Proof of Lemma 4.2
Analogue to the proof of Lemma 4.1, we get, by first multiplying the \(\eta \) equation of (4.2) by \(|\eta |^{p-2}\eta \) and then integrating the resulting equality over \(\mathop {\mathbb R}\nolimits ^3,\) that
Once again due to \(\partial _r(ru^r)+\partial _z(ru^z)=0\) and \(u^r|_{r=0}=0\), we get
It is easy to observe that
It follows from Sobolev embedding Theorem that
As a result, we obtain
To handle the term \(\Bigl \Vert \frac{|V|^{2-\frac{q}{2}}}{r^{2\varepsilon }}\Bigr \Vert _{L^{2p}},\) we split \(\frac{|V|^{2-\frac{q}{2}}}{r^{2\varepsilon }}\) as
with \(\alpha , \beta \) being determined by
It follows from Sobolev embedding Theorem that
It is easy to verify that \(\alpha +\frac{1}{r}=\frac{1}{2p}-\frac{(p-1)^2}{4p}\) provided selecting
which belongs to \(\bigl [\frac{3251}{9280},\frac{4}{11}\bigr [\) whenever \(p\in \bigl ]1,\frac{21}{20}\bigr ]\).
Moreover, under such choice of indexes, the term \(\bigl \Vert |V|^{\frac{q}{2}}\bigr \Vert _{L^2}^{2\alpha +\beta -\frac{3-p}{2p}}\) in (4.19) disappears. Then we get, by applying Hölder’s inequality, that
Inserting the above inequality into (4.18) gives rise to
Note that \(\frac{3(p-1)}{2p}+\left( \frac{3-p}{2p}+1-2\alpha \right) +2\alpha =2,\) by substituting the above inequality into (4.17) and using Young’s inequality, we achieve (4.7). This completes the proof of Lemma 4.2. \(\square \)
5 Global well-posedness with critical initial data
In what follows, we shall always denote \(U\buildrel \mathrm{def}\over =\frac{u^\theta }{r}\) and \(W\buildrel \mathrm{def}\over =r^{-\frac{7}{11}}u^\theta \).
Lemma 5.1
Under the assumption of Theorem 1.2, for any \(p\in [1,\frac{21}{20}]\), \(\varepsilon =\frac{-9p^2+21p-4}{24p-2}\), and \(q=\frac{3p}{2-\varepsilon }\), the local solutions constructed in Theorem 1.1 satisfy
Proof
Taking \(\delta =1-\frac{1}{p}\) in (3.29) yields (5.1). Likewise, taking \(\delta =\gamma =\frac{1}{3}\) and \(~q_1=q_2=\frac{3}{2}\) in (3.28) leads to (5.2).
On the other hand, for any \(p\in [1, {21}/{20}]\), due to the choice of \(\varepsilon \) and q, we have
Noting that \(\Vert V^\varepsilon \Vert _{L^q}=\Bigl \Vert \frac{u^\theta }{r^{1-\varepsilon -\frac{1}{q}}}\Bigr \Vert _{L^q(\Omega )}\) and \(\Vert r^{-\frac{7}{11}}u^\theta _0\Vert _{L^{\frac{11}{6}}}=\bigl \Vert r^{-\frac{1}{11}}u^\theta _0\bigr \Vert _{L^{\frac{11}{6}}(\Omega )}\), then applying (3.28) with \(q_1=q,~q_2=\frac{11}{6},\) \(\delta =1-\varepsilon -\frac{1}{q}\) and \(\gamma =\frac{1}{11}\) gives
where we have used Hölder’s inequality in the last step. \(\square \)
Proof of Theorem 1.2
Due to \(U_0=\frac{u^\theta _0}{r}\in L^{\frac{3}{2}},~ru^\theta _0\in L^\infty \), and
we deduce that both \(\Vert u^\theta _0\Vert _{L^2(\Omega )}=\Vert r^{-\frac{1}{2}}u^\theta _0\Vert _{L^{2}}\) and \(\Vert r^{-\frac{3}{10}}u^\theta _0\Vert _{L^{\frac{20}{13}}(\Omega )}=\Vert r^{-\frac{19}{20}}u^\theta _0\Vert _{L^{\frac{20}{13}}}\) are sufficiently small as long as \(\Vert ru^\theta _0\Vert _{L^\infty }\) is small enough. Then by Theorem 1.1, the equation (1.4) has a unique mild solution
and the lifespan \(T>0\) depends only on \(\omega _0^\theta \). We denote \(t_0\buildrel \mathrm{def}\over =\frac{T}{2}\). In the following, we will always abbreviate \(L_p(T)\) as \(L_p\), similar abbreviations for the remaining ones in (1.7), (3.27).
If \(ru^\theta (t_0)\in L^{A}\bigcap L^\infty \) and if there exists some \(p_0\in \left]1,\min \left( 1+\frac{1}{10A},\frac{21}{20}\right) \right[\), it follows from Proposition 4.1 that for \(\varepsilon _0\buildrel \mathrm{def}\over =\frac{-9p_0^2+21p_0-4}{24p_0-2}\) and \(q_0\buildrel \mathrm{def}\over =\frac{2(12p_0-1)}{3(p+3)},\) if there holds
where
Then the system (1.4) has a global solution. Moreover, in view of (4.5) and Lemma 5.1, for all \(t\in [t_0,\infty )\), there holds
By the choice of \(p_0\), \(\alpha (p_0)>10A\). Then we deduce from Lemma 2.1 and Lemma 5.1 that the smallness condition (5.4) holds provided that
In the following, we consider estimates in critical spaces. By a similar derivation as Lemma 4.1, that for \(q_{1,1}, q_{2,1}\) satisfying \(\frac{1}{q_{1,1}}+\frac{1}{q_{2,1}}=1\) and \(q_{1,1}>\frac{3p_0}{3-p_0}\), there holds
Take \(q_{2,1}=\vartheta _1+\sigma _1+\frac{2\sigma _1}{3},\) with \(\vartheta _1>0,~0<\sigma _1<1\) to be determined later, then we have
where in the last step, we used Galiardo–Nirenberg inequality provided that
which will be verified later. Then we get, by applying Young’s inequality, that
provided that
which in particular holds by taking
So that (5.8) holds and \(q_{1,1}=4>\frac{3p_0}{3-p_0}\). Hence all the above calculations go through.
By inserting the Estimate (5.9) into (5.7), with the indices given by (5.10), we obtain
Next, applying \(L^1\) energy estimate for the \(\eta \) equation in (4.2) yields
Noting that \( \frac{|W|^{\frac{13}{12}}}{r^{\frac{8}{11}}}=\bigl |\frac{|W|^{\frac{11}{6}}}{r^2}\bigr |^{\frac{1}{2}}\cdot \bigl |r^{\frac{18}{11}}W\bigr |^{\frac{1}{6}}\), so we have
Substituting (5.13) into (5.12), and using the fact that \(\int _{\mathop {\mathbb R}\nolimits ^3}(-\Delta \eta )\cdot \mathrm {sgn}\ \eta \,dx\le 0\), we achieve
Meanwhile, by taking \(p=p_0\) in (4.8), we have
Summarizing the estimates (5.11), (5.14) and (5.15) gives rise to
Then under the smallness conditions (5.6), and
we get, by a standard continued argument, as we did in the last step of the proof of Proposition 4.1, that
Recalling that \(1<p_0<\min \left( 1+\frac{1}{10A},\frac{21}{20}\right) \), we have \(\frac{11 p_0}{30(p_0-1)}>\frac{11}{3}A\). Then it follows from Lemmas 2.1 and 5.1 that the condition (5.17) is verified provided that
Finally we derive the \(L^{\frac{3}{2}}\) estimate for U. Indeed along the same line of the derivation of \(\Vert W(t)\Vert _{L^{\frac{11}{6}}},\) and using the indices given by (5.10), we infer
whereas by applying Hölder’s inequality, one has
Substituting (5.21) into (5.20), then using of Young’s inequality gives rise to
from which and (5.18), we deduce by a standard continued argument that
provided that the smallness conditions (5.6), (5.17), (5.19) and
hold. Yet it follows Lemma 5.1 that (5.23) can be satisfied as long as
Therefore under the smallness conditions (5.6), (5.17), (5.19) and (5.24), we get, by summing up (5.18) and (5.22) that for any \(0\le t<\infty \),
This completes the proof of Theorem 1.2. \(\square \)
References
Bahouri, H., Chemin, J.-Y., Danchin, R.: Fourier Analysis and Nonlinear Partial Differential Equations, Grundlehren der Mathematischen Wissenschaften. Springer, Berlin (2010)
Chae, D., Lee, J.: On the regularity of the axisymmetric solutions of the Navier–Stokes equations. Math. Z. 239, 645–671 (2002)
Chemin, J.-Y., Zhang, P.: On the critical one compoenent regularity for 3-D Navier–Stokes system. Ann. Sci. École Norm. Sup. (4) 49, 133–169 (2016)
Feng, H., S̆verák, V.: On the Cauchy problem for axi-symmetric vortex rings. Arch. Ration. Mech. Anal. 215, 89–123 (2015)
Gallay, T., S̆verák, V.: Remarks on the Cauchy problem for the axisymmetric Navier–Stokes equations. Conflu. Math. 7, 67–92 (2015)
Kato, T.: Strong \(L^p\)-solutions of the Navier–Stokes equation in \({\mathbb{R}}^m\), with applications to weak solutions. Math. Z. 187, 471–480 (1984)
Ladyz̆enskaja, O.A.: Unique global solvability of the three-dimensional Cauchy problem for the Navier–Stokes equations in the presence of axial symmetry. Nauc̆n. Sem. Leningrad. Otdel. Mat. Inst. Steklov. (LOMI) 7, 155–177 (1968). (in Russian)
Leonardi, S., Málek, J., Nec̆as, J., Pokorny, M.: On axially symmetric flows in \(\mathbb{R}^3\). Z. Anal. Anwend. 18, 639–649 (1999)
Leray, J.: Jean Sur le mouvement d’un liquide visqueux emplissant l’espace. Acta Math. 63, 193–248 (1934)
Ukhovskii, M.R., Iudovich, V.I.: Axially symmetric flows of ideal and viscous fluids filling the whole space. J. Appl. Math. Mech. 32, 52–61 (1968)
Zhang, P., Zhang, T.: Global axi-symmetric solutions to 3-D Navier–Stokes system. Int. Math. Res. Not. IMRN 3, 610–642 (2013)
Acknowledgements
P. Zhang is partially supported by NSF of China under Grant 11371347 and innovation grant from National Center for Mathematics and Interdisciplinary Sciences.
Author information
Authors and Affiliations
Corresponding author
Additional information
Communicated by F.H. Lin.
Rights and permissions
About this article
Cite this article
Liu, Y., Zhang, P. On the global well-posedness of 3-D axi-symmetric Navier–Stokes system with small swirl component. Calc. Var. 57, 17 (2018). https://doi.org/10.1007/s00526-017-1288-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s00526-017-1288-4