Global well-posedness of one-dimensional compressible Navier-Stokes-Vlasov system

Due to the maximum principle for transport equations, (\ref{basiccnsv})_{1} holds. In addition, (\ref{basiccnsv})_{2}-(\ref{basiccnsv})_{3} can be derived after a direct computation. To show (\ref{basiccnsv})_{4}, multiplying (\ref{m1})_{1} and (\ref{m1})_{2} by \frac{\gamma}{\gamma-1}\rho^{\gamma-1} and u respectively, and integrating the resulted equations by parts over \mathbb{T}, one gets after adding them together that

Multiplying (\ref{m1})_{3} by \frac{1}{2}|v|^2, and integrating the resulted equation by parts over \mathbb{T}\times\mathbb{R}, we obtain

The combination of (Proof)-(Proof) gives rise to (\ref{basiccnsv})_{4}. The proof of Lemma 2.1 is completed.

We introduce a new effective velocity

where the operator \mathcal{I}:L^1(\mathbb{T})\rightarrow L^{\infty}(\mathbb{T}) is defined by

It can be verified verify for g\in L^1(\mathbb{T}) that

and for g\in W^{1,1}(\mathbb{T}) that

Integrating the equation (\ref{m1})_{3} over \mathbb{R} with respect to velocity variable v, we have

where n and w are defined by (\ref{nw}). Applying the operator \mathcal{I} to the equation (\ref{nj}), we deduce by (Proof) that

which together with (\ref{m1})_{1}, (Notations) and (Proof) leads to

Thus, by virtue of (Proof), we can re-write the momentum equation (\ref{m1})_{2} as

It follows from (Proof)-(Proof) and (Proof) that

which gives rise to

Furthermore, due to the equation (\ref{m1})_{1}, it holds

where the function \theta(\rho) is given by

Substituting (Proof) into (Proof), and re-writing the resulted equation along the particle path \mathcal{X}^{x,t}(s) for any (x,t)\in \mathbb{T}\times [0,T] defined as

we have

where F(x,t) is given by

It follows from (Notations), (2.1) and (Proof) for any (x,t)\in\mathbb{T}\times[0,T] that

where C_{2}>0 and C_{3}>0 are two constants given by

Since it follows

we get

Then, the upper bound of \rho uniformly in time can be derived by the arguments due to Zlotnik [zlotnik1]. For any (x,t)\in\mathbb{T}\times[0,T], if there is a time s\in[0,t] such that it holds

then by the continuity of \rho(\mathcal{X}^{x,t}(s),s) with respect to s and the fact \rho_{0}(\mathcal{X}^{x,t}(0))\leq C_{4}, we can find a time s_{*}\in[0,s) to have

Integrating (Proof) over [s_{*},s], we obtain by (Proof) and (Proof)-(?) that

which together with (\ref{basiccnsv})_{4} and the case 1\leq \rho(\mathcal{X}^{x,t}(s),s)\leq C_{4} leads to

By virtue of (\ref{basiccnsv})_{4}, (?) and the Gr{\rm{\ddot{o}}}nwall’s inequality, we have

Thus, we show the upper bound of \rho uniformly in time.

Finally, integrating (Proof) over [0,t], one deduces by (\ref{basiccnsv})_{4}, the upper bound in (2.2), (Proof) and (Proof) for any (x,t)\in\mathbb{T}\times[0,T] that

which together with the fact

yields the time-dependent lower bound of \rho. The proof of Lemma 2.2 is completed.

For any (x,v,t)\in \mathbb{T}\times\mathbb{R}\times[0,T], define the bi-characteristic curves (X^{x,v,t}(s),V^{x,v,t}(s)) by

It holds by (\ref{m1})_{3} that

which can be re-written along the bi-characteristic curves (X^{x,v,t}(s),V^{x,v,t}(s)) as

Thus, we have

By (2.2) and (Proof), we have (\ref{rhouh1cnsv})_{1}. Then we solve the equation (\ref{XV})_{3} over [0,s] to get

which shows for s=t that

For any (x,t)\in\mathbb{T}\times[0,T], defining

we have

Due to (Proof), it holds

Therefore, one deduces by (\ref{a1})_{2}, (2.1), (2.2) and (Proof) for any (x,t)\in\mathbb{T}\times[0,T] that

where we have used the fact

The combination of (Proof)-(Proof) gives rise to (\ref{rhouh1cnsv})_{2}.

Next, we are going to show (\ref{rhouh1cnsv})_{3}. Denote the second effective velocity as

It is easy to verify

derived from (\ref{m1})_{1}, and

so that the equation (Proof) can be re-written as

Multiplying (?) by p|U|^{p-2}U for any p\in[2,\infty), and integrating the resulted equation by parts over \mathbb{T}, we obtain by (2.1), (2.2) and (Proof) that

We deduce after dividing (?) by \big{(}\|(\rho^{\frac{1}{p}}U)(t)\|_{L^{p}}^{p}+\varepsilon^{p}\big{)}^{\frac{p-1}{p}} that

Integrating (?) over [0,t], making use of (2.1), (2.2), (Proof) and (Proof), and then taking the limit \varepsilon\rightarrow 0, we get

where C_{5}>0 is a sufficiently large constant independent of time T>0 and p\in [2,\infty). By (2.2) and (?), we have

Since the constant on the right-hand side of (?) is independent of p\in[2,\infty), one can take the limit p\rightarrow \infty in (?) to obtain

Choosing p=2 in (?), we deduce by (2.1), (2.2) and (?) that

Then, we divide the equation (\ref{m1})_{2} by \rho to get

By (\ref{nw}) and (\ref{rhouh1cnsv})_{1}-(\ref{rhouh1cnsv})_{2}, one has

It follows from (2.1), (2.2), (?), (?) and the Gagliardo-Nirenberg’s inequality that

Multiplying (?) by -u_{xx}, and integrating the resulted equation by parts over \mathbb{T}\times[0,t], one concludes from (?) and the Young’s inequality that

from which we infer

By (2.2) and (?)-(?), it also holds

One may get from (\ref{basiccnsv})_{3}, (Proof), (?), (?), (?) and the Sobolev embedding H^1(\mathbb{T})\hookrightarrow L^{\infty}(\mathbb{T}) that

The combination of (?) and (?)-(?) leads to (\ref{rhouh1cnsv})_{3}. The proof of Lemma 2.3 is completed.

Differentiating (\ref{m1})_{3} with respect to x and v respectively, we have

Thus, f_{x} and f_{v} can be re-written along the bi-characteristic curves (X^{x,v,t}(s),V^{x,v,t}(s)) defined by (Proof) for any (x,v,t)\in\mathbb{T}\times\mathbb{R}\times[0,T] as

which together with (2.2) and (2.3) leads to

and

where the constants \rho_{+}>0 and R_{T}>0 are given by (2.2) and (\ref{rhouh1cnsv})_{2} respectively. Substituting (Proof) into (Proof), and employing a Gr{\rm{\ddot{o}}}nwall type argument, we obtain

By (2.2), (2.3), (\ref{m33}) and (Proof), it holds

Then one deduces by (2.3) and (Proof)-(Proof) that

We turn to establish the H^2-norm estimate of \rho. Dividing (?) by \rho, and differentiating the resulted equation with respect to x, we have

Multiplying (?) by U_{x}, and integrating the resulted equation by parts over \mathbb{T}, one deduces by (2.2) and (2.3) that

which implies

Thereby, it follows from (2.2), (2.3), (?) and (?) that

To establish the H^2-estimate of u, we differentiate the equation (?) with respect to variable x to have

Multiplying (?) by -u_{xxx}, integrating the resulted equation by parts over \mathbb{T}\times[0,t], and making use of (2.1), (2.2), (2.3), (Proof) and (?), one may get after a complicated computation that

By (\ref{m1})_{1}, (?) and (?)-(?), it also holds

We are ready to establish the time-weighted estimates of u. We differentiate (?) with respect to t to get

Multiplying (?) by -tu_{txx}, integrating the resulted equation by parts over \mathbb{T}, and making use of the estimates (2.2), (Proof), (?) and (?)-(?), we have

Thus, one deduces by (?), (?), the Gr\rm{\ddot{o}}nwall’s inequality and the fact tu_{xt}|_{t=0}=0 that

By (2.2), (?)-(?) and (?), it also holds

Since it follows by (?) and (?) that

similarly to (Proof)-(Proof), one can show

The combination of (Proof)-(Proof), (?), (?)-(?) and (?)-(?) gives rise to (2.4). The proof of Lemma 2.4 is completed.

By (2.2), we have

where the time T_{0}>0 is given by Lemma 2.5.

We prove that the fluid density \rho is uniformly bounded from below for t\in(T_{0},T] in the case \beta=0, since the case \beta>0 can be dealt with in a similar way. It is easy to show

where m_{1}(t) is given by (\ref{m1m2})_{1}. By (\ref{nw}) and (?), the equation (\ref{m1})_{2} for \beta=0 can be re-written as

Applying the operator \mathcal{I} defined by (Proof) to (?), and re-writing the resulted equation along the particle path \mathcal{X}^{x,t}(s) given by (Proof) for any (x,t)\in\mathbb{T}\times[T_{0},T] and s\in[T_{0},t], we obtain by (Proof)-(Proof) and (Proof) that

where A(x,t) is given by

Due to (2.1), (2.2), (Proof) and (?), for any (x,t)\in\mathbb{T}\times[T_{0},T], it holds

By (2.1), (2.2), (Proof), (2.5), (?) and (?), for any t\in[T_{0},T] and s\in [T_{0},t], we have

where the constant C_{11}>0 is given by

Multiplying (?) by

we obtain

which together with (Proof) and (?)-(?) implies for any (x,t)\in\mathbb{T}\times [T_{0},T] that

The combination of (Proof) and (?) leads to (2.6). The proof of Lemma 2.6 is completed.

Multiplying (?) by U given by (?), integrating the resulted equation by parts over \mathbb{T}, and using (2.1), (2.2) and (2.6), we have

where the constant C_{14}>0 is given by

We divide (Proof) by (\|(\sqrt{\rho}U)(t)\|_{L^2}^2+\varepsilon^2)^{\frac{1}{2}} for \varepsilon>0, employ the Gr\rm{\ddot{o}}nwall’s inequality and then take the limit \varepsilon\rightarrow 0 to obtain

Due to (2.1), (2.2) and (?), it holds

By (2.1) and (Proof)-(Proof), we prove (\ref{rhoh1uniform})_{1}.

Moreover, one deduces by (\ref{a1})_{2}, (2.1), (2.2), (Proof), (Proof) and (2.6) that

where the initial energy E_{0} is given by (2.1). By (Proof) and (?), (\ref{rhoh1uniform})_{2} holds. The proof of Lemma 2.7 is completed.

(\ref{basicR})_{1}-(\ref{basicR})_{2} can be derived directly by (\ref{m1})_{1} and (\ref{m1})_{3}. Similarly to (Proof)-(Proof), we can obtain (\ref{basicR})_{3}. Then it is easy to verify

One deduces by (?), the Gagliardo-Nirenberg’s and Young’s inequalities that

For the second term on the right-hand side of (Proof), it holds

Substituting (Proof)-(Proof) into (Proof), we obtain

which together with (\ref{basicR})_{3} and the fact

gives rise to (\ref{basicR})_{4}. The proof of Lemma 2.8 is completed.

For any x_{0}\in \mathbb{R}, we claim that it holds

where \rho_{T}>0 is a constant independent of the choice of x_{0}, so the upper bound in (\ref{upper1})_{1} for any (x,t)\in \mathbb{R}\times [0,T] can be obtained immediately. Indeed, integrating the equation (\ref{m1})_{3} over (-\infty, x]\times\mathbb{R} for any x\in \mathbb{R}, we have

which together with (\ref{m1})_{1} leads to

Substituting (Proof) into momentum equation (\ref{m1})_{2}, we get

Similarly to (Proof)-(Proof), introduce the operator \mathcal{I}_{x_{0}}: L^{1}(x_{0},x_{0}+1)\rightarrow W^{1,1}(x_{0},x_{0}+1) as

which satisfies for g\in L^1(x_{0},x_{0}+1) that

and for g\in W^{1,1}(x_{0},x_{0}+1) that

Applying \mathcal{I}_{x_{0}} to the equation (Proof), one can show by (Proof)-(Proof) that

where \theta(\rho) and \mathcal{X}^{x,t}(s) for any (x,t)\in (x_{0},x_{0}+1)\times[0,T] are defined by (Proof) and (Proof) respectively, and H(s) is given by

For (x,t)\in (x_{0},x_{0}+1)\times[0,T] and s\in[0,t], by virtue of (Proof) and (\ref{basicR})_{4}, we have

which leads to

In addition, it is easy to verify

where C_{16}>0 is a constant depending only on \gamma and \widetilde{\rho}. With the help of (2.8), (Proof) and (Proof)-(Proof), we derive for any s\in [0,t] that

and

where C_{17}>0 and C_{18}>0 are two constants independent of time T>0. By (Proof), one has

Then, integrating the equality (Proof) over [0,t], we obtain by (2.8) and (Proof)-(Proof) for any (x,t)\in(x_{0},x_{0}+1)\times[0,T] that

which together with (\ref{basicR})_{3} and the Gr{\rm{\ddot{o}}}nwall’s inequality yields (Proof). By (?), (\ref{basicR})_{3} and (Proof), we have

Similarly, integrating (Proof) over [0,t], one deduce by (2.8), (Proof) and (Proof)-(Proof) for any (x,t)\in (x_{0},x_{0}+1)\times[0,T] that

The combination of (?) and (?) shows the lower bound of \rho. The proof of Lemma 2.9 is completed.

In accordance to (Proof) and (\ref{upper1})_{1}, (\ref{H1R})_{1} holds. Let \Sigma(x,t) be given by (\ref{Sigmadef}) for (x,t)\in\mathbb{R}\times[0,T]. Similarly to the proof of Lemma 2.3, one deduces by (\ref{a2})_{2}, (Proof), (\ref{basicR})_{4} and (\ref{upper1})_{1} that

which implies (\ref{H1R})_{2}. By (Notations) and (\ref{H1R})_{1}-(\ref{H1R})_{2}, we have

In addition, it follows from (2.8), (Proof) and the Young’s inequality for any p\in (1,\infty) that

and similarly,

Then, we turn to show (\ref{H1R})_{3}. Define the following effective velocity of Bresch-Desjardins type [bresch1, bresch2, haspot1]:

Making use of (?) and (?), we can re-write the equation (\ref{m1})_{2} as

Multiplying (?) by p|W|^{p-2}W for any p\in[2,\infty), and integrating the resulted equation by parts over \mathbb{R}, we derive by (\ref{upper1})_{1} and (Proof)-(Proof) that

where C_{T}>0 is a constant independent of p\in[2,\infty). Dividing (?) by (\|(\rho^{\frac{1}{p}}W)(t)\|_{L^{p}}^{p}+\varepsilon^{p})^{\frac{p-1}{p}}, integrating the resulted inequality over [0,t], and then taking the limit \varepsilon\rightarrow0, we have

Due to (\ref{basicR})_{4}, (\ref{upper1})_{1}, (?) and the fact

it follows

where \widetilde{C}_{T}>0 is a sufficiently large constant independent of p\in[2,\infty). One gets after taking the limit p\rightarrow\infty in (?) that

Choosing p=2 in (?), we have by (\ref{upper1})_{1} and (?) that

Similarly to (?)-(?), one can obtain

which together with (\ref{upper1})_{1}, (?) and (?) leads to

The proof of Lemma 2.10 is completed.

We establish the \mathcal{H}^1-estimate of f so as to obtain the H^1-estimate of the term nw-un on the right-hand side of the equation (?). Since the rest of the proof follows similarly to (\ref{fxfv})-(?) in Lemma 2.4, we omit the details here. Multiplying (\ref{fxfv})_{1} by f_{x}, and integrating the resulted equation by parts over \mathbb{R}\times\mathbb{R}, one deduces by (\ref{upper1})_{1} and (2.10) that

Similarly, one has

which together with (2.10), (Proof) and the Gr{\rm{\ddot{o}}}nwall’s inequality leads to

By (\ref{m33}), (\ref{upper1})_{1}, (2.10) and (Proof), it also holds

Thus, we obtain by (\ref{nw}), (2.10) and (Proof)-(Proof) that

Define the bi-characteristic curves (X_{i}(t),V_{i}(t)), i=1,2, for any (x,v,t)\in \mathbb{T}\times\mathbb{R}\times[0,T] as

Since it holds for i=1,2 that

we can express f_{i} along (X_{i}(t),V_{i}(t)) for any (x,v,t)\in \mathbb{T}\times\mathbb{R}\times[0,T] as

Due to (Proof), Proposition 1.2 in [majda1] and the fact

we have

which implies

We combine (Proof)-(Proof) together to show the following property about the coordinates transform (x,v)\mapsto (X_{i}(t),V_{i}(t)):

Inspired by [loeper1, han1], we estimate the quantity

For \Omega=\mathbb{T}, making use of (Proof), we can obtain after a complicated computation that

The right-hand side terms of (?) can be estimated as follows. First, one has

By (Proof), it holds

For the term I_{3}^1, we have

We estimate the term I_{4}^1 as

where in the last inequality we have used the following fact for i=1,2:

derived from (\ref{a1})_{2}, (1.1), (Proof), (2.7) and (Proof). By (Proof) and (?), it also holds

Substituting the above estimates of I^{1}_{i}, 1,...,5, into (?), and employing (1.1)-(1.1), we have

Next, we turn to estimate (\rho_{1}-\rho_{2},u_{1}-u_{2}). With loss of generalization, we prove the case \beta=0. It is easy to verify

Multiplying (?) by (\rho_{1}-\rho_{2}), integrating the resulted equation by parts over \mathbb{T}, and using (1.1)-(1.1), we have

In addition, we multiply (?) by (u_{1}-u_{2}) and integrate the resulted equation by parts over \mathbb{T} to obtain

One can show

For I_{2}^2, we have

where we have used (1.1)-(1.1), the Young’s inequality and the fact

Similarly, by (1.1) and (?), one has

With the help of (Proof), we obtain

and similarly,

By (?)-(?), the last term I_{4}^2 on the right-hand side of (?) can be estimated as follows:

To estimate I_{1}^3, we obtain by (1.1)-(1.1) and (?) that

Similarly, one can derive by (1.1)-(1.1), (?) and (?) that

and

It follows from (1.1)-(1.1), (?) and (3.1) in Lemma 3.2 below that

Making use of (\ref{r1}), (Proof) and (3.1) in Lemma 3.1 below, we can obtain

and similarly,

The combination of (?)-(?) leads to

Substituting the estimates of I_{i}^3, i=1,...,4, into (?), we obtain

Substituting the estimates of I_{i}^2, i=1,...,4, into (?), and adding the resulted inequality, (?) and (?) together, we have

Thus, one deduces by the Gr\rm{\ddot{o}}nwall’s inequality, (1.1), the fact Q(0)=0 and Lemma 3.2 below that

The proof of Proposition 3.1 is completed.

Step 1: Construction of approximate sequence. The iteration scheme for the approximate solutions (\rho^{n+1},u^{n+1},f^{n+1}), n\geq 0, is defined by solving the following equations:

with the initial data

Define the work space

where \rho_{0+}>0, \rho_{0-}>0 and E_{0}>0 are constants given by

Set (\rho^{0},u^{0},f^{0})=: (\rho_{0},u_{0},f_{0}). Suppose

for some time T_{0}>0. By the standard theory of linear O.D.E. systems [hsi1], for given u^{n}\in C([0,T_{0}];H^2(\mathbb{T}))\hookrightarrow C([0,T_{0}];C^{1}(\mathbb{T})), there is a unique particle path \mathcal{X}_{n}^{x,t}(s) for any (x,t)\in \mathbb{T}\times[0,T_{0}] defined by

Thus, the linear transport equation (\ref{m1n})_{1} can be solved by

Similarly, for given (\rho^{n},u^{n})\in C([0,T_{0}];H^2(\mathbb{T}))\hookrightarrow C([0,T_{0}];C^{1}(\mathbb{T})), we can define uniquely the bi-characteristic curves

for any (x,v,t)\in \mathbb{T}\times\mathbb{R}\times[0,T_{0}], so that the solution f^{n+1} for the linear transport equation (\ref{m1n})_{3} can be represented by

Finally, by the standard theorem of linear parabolic equations [jllions1, valli1], (\ref{m1n})_{2} admits a unique solution u^{n+1}\in C([0,T_{0}];H^2(\mathbb{T}))\cap L^2(0,T_{0};H^3(\mathbb{T})).

Step 2: Uniform estimates. We claim that there is a time T_{0}>0 such that for each n\geq 0, if (\rho^{n},u^{n},f^{n}) satisfies (Proof), then it is also true for (\rho^{n+1},u^{n+1},f^{n+1}). Indeed, by (Proof)-(Proof), one has

where N>0 denotes a a suitably large constant depending only on \rho_{0+},\rho_{0-} and R_{0}. In addition, the equation (\ref{m1n})_{1} implies

We multiply (?) by \sum_{i=0}^{2}\partial_{x}^{i}\rho^{n+1} and integrate the resulted equation by parts over \mathbb{T} to get

Thus, it follows from (Proof), (?) and the Gr{\rm{\ddot{o}}}nwall’s inequality that

We are ready to estimate u^{n+1}. By (\ref{m1n})_{2} and (Proof), it is easy to obtain

Then we differentiate (\ref{m1})_{2} with respect to x to have

Multiplying (\ref{unxxx}) by -(u^{n+1})_{xxx}, integrating the resulted equation by parts over \mathbb{T}, and making use of (Proof), we derive

which together with (?) gives rise to

To estimate f^{n+1}, we obtain by (Proof) that

It follows from (\ref{alocal})_{2}, (Proof), (Proof) and (?) for any (x,t)\in\mathbb{T}\times [0,T_{0}] that

from which we get

Similarly to (Proof)-(Proof), one has by a direct computation that

Choose

Then by the previous estimates (Proof)-(?), (?), (?) and (?)-(?), we conclude that for each n\geq 0, the strong solution (\rho^{n+1},u^{n+1},f^{n+1}) to the IVP (Proof)-(Proof) belongs to \mathcal{W}(T_{0}) as long as (\rho^{n},u^{n},f^{n}) satisfies (Proof). By repeating the procedure used above, we can obtain an approximate sequence (\rho^{n},u^{n},f^{n}) satisfying the uniform estimates in \mathcal{W}(T_{0}).

Step 3: Compactness and convergence. Let M_{0}>0 and T_{0}>0 be given by (?). We aim to prove the convergence of the approximate sequence (\rho^{n},u^{n},f^{n}) to the corresponding strong solution of the original IVP (?)-(?). Since (\rho^{n},u^{n},f^{n}) is uniformly bounded in \mathcal{W}(T_{0}), one can show after a tedious computation that

where N_{*}>0 is a constant depending only on \rho_{0+},\rho_{0-} and R_{0}. Choosing

we have

which implies that there is a limit (\rho,u,f) so that as n\rightarrow\infty, it holds

Due to the estimates of (\rho^{n},u^{n},f^{n}) uniformly in n\geq0, as n\rightarrow \infty, it holds up to a subsequence (still denoted by (\rho^{n},u^{n},f^{n})) that

Thus, we can pass to the limit n\rightarrow\infty for the equations (\ref{m1n}) in the sense of distributions and prove that (\rho,u,f) is indeed a strong solution to the IVP (\ref{m1})-(\ref{dn}) on \mathbb{T}\times\mathbb{R}\times[0,T_{*}].

In addition, applying the theory of renormalized solutions in [diperna1], we have \rho\in C([0,T_{*}];H^1(\mathbb{T})). By (\ref{m1})_{2}, (?)-(?) and the uniform estimate of (\rho^{n},u^{n},f^{n}), one can show u_{t}\in L^2(0,T_{*}; H^1(\mathbb{T})), and therefore u\in C([0,T_{*}];H^2(\mathbb{T})) follows from u\in L^2(0,T_{*};H^3(\mathbb{T}))\cap H^1(0,T_{*};H^1(\mathbb{T})). Moreover, with the help of (\rho,u) \in C([0,T_{*}];H^2(\mathbb{T}))\hookrightarrow C([0,T_{*}];C^{1}(\mathbb{T})), it is easy to verify the C^1-regularity of the bi-characteristic curves (X^{x,v,t}(s),V^{x,v,t}(s)) defined by (Proof) with respect to (x,v)\in\mathbb{T}\times\mathbb{R}, which together with (Proof) implies f\in C([0,T_{*}];C^{1}(\mathbb{T}\times\mathbb{R})). According to Proposition 3.1, the strong solution (\rho,u,f) is unique. The proof of Lemma 4.1 is completed.

Multiplying (?) by (u-m_{1}) and integrating the resulted equation by parts over \mathbb{T}, we obtain

where we have used the fact

We get after a direct computation that

Since it holds

and

multiplying the Vlasov equation (\ref{m1})_{3} by \frac{1}{2}|v-m_{2}|^2 and integrating the resulted equation by parts over \mathbb{T}\times\mathbb{R}, we have

By virtue of (\ref{basiccnsv})_{2}, (?) and (Proof), one can show

Adding (Proof), (Proof), (Proof) and (Proof) together, we can obtain (5.1).

One deduce from the Young’s inequality, (2.1) and (Proof) that

Due to (?) and (2.2), it holds

which together with (Proof) implies (5.2) directly.

We are going to estimate (5.2). It follows from (\ref{m1})_{1}, (Proof) and (?) that

Since it holds \mathcal{J}(\rho-\overline{\rho_{0}})(1)=\mathcal{J}(\rho-\overline{\rho_{0}})(0), we deduce by the equations (?) and (?) and integration by parts that

To estimate the first term on the right-hand side of (?), by (\ref{basiccnsv})_{1}, we infer

One can show

where we have used (\ref{rhocnsv})_{2} and (?). It also follows from (\ref{basiccnsv}), (2.2) and (Proof) that

Substituting (?)-(?), (?) into (?) and making use of (Proof), we gain (5.2).