Loading [MathJax]/jax/output/SVG/jax.js

Quasineutral limit for the compressible two-fluid Euler–Maxwell equations for well-prepared initial data

  • Received: 01 April 2020 Revised: 01 April 2020
  • Primary: 35L60, 35B40; Secondary: 35C20

  • In this paper, we study the quasi-neutral limit for the compressible two-fluid Euler–Maxwell equations for well-prepared initial data. Precisely, we proved the solution of the three-dimensional compressible two-fluid Euler–Maxwell equations converges locally in time to that of the compressible Euler equation as ε tends to zero. This proof is based on the formal asymptotic expansions, the iteration techniques, the vector analysis formulas and the Sobolev energy estimates.

    Citation: Min Li, Xueke Pu, Shu Wang. Quasineutral limit for the compressible two-fluid Euler–Maxwell equations for well-prepared initial data[J]. Electronic Research Archive, 2020, 28(2): 879-895. doi: 10.3934/era.2020046

    Related Papers:

    [1] Min Li, Xueke Pu, Shu Wang . Quasineutral limit for the compressible two-fluid Euler–Maxwell equations for well-prepared initial data. Electronic Research Archive, 2020, 28(2): 879-895. doi: 10.3934/era.2020046
    [2] Fei Shi . Incompressible limit of Euler equations with damping. Electronic Research Archive, 2022, 30(1): 126-139. doi: 10.3934/era.2022007
    [3] Min Li . Long-wavelength limit for the Green–Naghdi equations. Electronic Research Archive, 2022, 30(7): 2700-2718. doi: 10.3934/era.2022138
    [4] Jie Zhang, Gaoli Huang, Fan Wu . Energy equality in the isentropic compressible Navier-Stokes-Maxwell equations. Electronic Research Archive, 2023, 31(10): 6412-6424. doi: 10.3934/era.2023324
    [5] Xiaoxia Wang, Jinping Jiang . The uniform asymptotic behavior of solutions for 2D g-Navier-Stokes equations with nonlinear dampness and its dimensions. Electronic Research Archive, 2023, 31(7): 3963-3979. doi: 10.3934/era.2023201
    [6] Natália Bebiano, João da Providência, Wei-Ru Xu . Approximations for the von Neumann and Rényi entropies of graphs with circulant type Laplacians. Electronic Research Archive, 2022, 30(5): 1864-1880. doi: 10.3934/era.2022094
    [7] Long Fan, Cheng-Jie Liu, Lizhi Ruan . Local well-posedness of solutions to the boundary layer equations for compressible two-fluid flow. Electronic Research Archive, 2021, 29(6): 4009-4050. doi: 10.3934/era.2021070
    [8] Li-Bin Liu, Ying Liang, Jian Zhang, Xiaobing Bao . A robust adaptive grid method for singularly perturbed Burger-Huxley equations. Electronic Research Archive, 2020, 28(4): 1439-1457. doi: 10.3934/era.2020076
    [9] Haibo Cui, Junpei Gao, Lei Yao . Asymptotic behavior of the one-dimensional compressible micropolar fluid model. Electronic Research Archive, 2021, 29(2): 2063-2075. doi: 10.3934/era.2020105
    [10] Xincai Zhu, Yajie Zhu . Existence and limit behavior of constraint minimizers for elliptic equations with two nonlocal terms. Electronic Research Archive, 2024, 32(8): 4991-5009. doi: 10.3934/era.2024230
  • In this paper, we study the quasi-neutral limit for the compressible two-fluid Euler–Maxwell equations for well-prepared initial data. Precisely, we proved the solution of the three-dimensional compressible two-fluid Euler–Maxwell equations converges locally in time to that of the compressible Euler equation as ε tends to zero. This proof is based on the formal asymptotic expansions, the iteration techniques, the vector analysis formulas and the Sobolev energy estimates.



    The purpose of this present paper is to investigate the quasi-neutral limit for the two-fluid Euler–Maxwell equations consisting of a set of nonlinear conservation laws for densities and momentums coupled to the Maxwell equations in time t>0 and space R3, which describes the transport of electrons of charge qe=1 and ions of charge qi=1 without viscosity in a magnetized plasma [3]

    {nνt+div(nνuν)=0,ν=e,i,mν[(nνuν)t+div(nνuνuν)]+Pν(nν)=qνnν(E+uνטB),t˜B+×E=0,ϵ0tEμ10טB=(qiniui+qeneue),div˜B=0,ϵ0divE=qini+qene,(1a)(1b)(1c)(1d)

    where ni, ui (respectively, ne, ue) denote the density and velocity of the ions (respectively, electrons), and E, ˜B are the electric field and the magnetic field. The coefficients ϵ0,μ0,c are the vacuum permittivity, vacuum permeability and light speed with ε0μ0c2=1, and the parameter γ=1/(ϵ120c) is usually chosen to be inversely proportional to the light speed c. As in classical fluid dynamics, the pressure functions Pν(nν)=a2νnbνν (ν=i,e) are supposed to be smooth and strictly increasing with aν>0,bν1. Moreover, qνnν(E+uνטB), qiniui+qeneue, qini+qene stand for the Lorentz force, the current density and the free charges for the particle, respectively. We introduce the Debye length λ2=ϵ0KBTen0, where the physical parameters are the mean density of the plasma n0>0, the Boltzmann constant KB>0 and the temperature of the electron Te>0. The scaled Debye length is denoted by ε2=ϵ0.

    With these parameters, system (1) under study can be scaled to the following form

    {tnν+(nνuν)=0,mν(tuν+uνuν)+hν(nν)=qν(E+γuν×B),γtB+×E=0,γε2tE×B=γ(niuineue),divB=0,ε2divE=nine,(2a)(2b)(2c)(2d)

    where B=˜Bγ.

    Usually, the dimensionless parameters γ,ε are small compared with the size of the other quantities as for the physical situation. In quasi-neutral plasma, the Debye length is small compared with the typical length L of the plasma. By taking the limit λL0 formally, we can derive an equilibrium between the positive and negative charges. In the non-relativistic limit, we regard γ as the singular perturbation parameter and let γ0. For such scales, the plasma can be considered as the compressible Euler–Poisson system. Furthermore, γ=ε20 is the combined non-relativistic and quasi-neutral limit, which leads to incompressible (one-fluid) or compressible (two-fluid) Euler equation.

    When taking ni,ui=0, system (2) is reduced to the unipolar Euler–Maxwell equations, and there have been many interesting results for the well-posedness and asymptotic analysis [11,12,13,20]. To list a few, Peng and Wang [13] studied the convergence of Euler–Maxwell equations in three-dimensional case to the e-MHD equations under well prepared initial data in the quasi-neutral limit. This result was then generalized in the quantum counterpart recently [8] and later in [16] for the general initial data. Moreover, in [11], the authors justified rigorously the convergence of Euler–Maxwell equations to compressible Euler–Poisson equations in time intervals independent of γ by an analysis of asymptotic expansions up to first order for general initial data and up to any order for well-prepared initial data. The combined non-relativistic and quasi-neutral limit can be found in [12]. For the formation of boundary layers, the interested readers can refer to [1,2,15,17,19] and the references therein for example.

    Recently, the two-fluid Euler/Navier-Stokes equations with electromagnetic field become more and more interesting as well as important in fluid dynamics. The local smooth solution was established in [5] since Euler–Maxwell equations are symmetrizable hyperbolic for nν in the sense of Friedrichs. The global existence and large time behavior were obtained in [3,10]. For the asymptotic limits with small parameters, there have been many mathematical investigations for Euler–Maxwell equations, see [14,21,22] for example. With boundary effects, one can see [7,4] and references therein. However, the rigorous study of the quasi-neutral limit for two-fluid Euler–Maxwell equations is also open [14]. The goal of this paper is to consider this problem. For convenience, we assume γ=1, which can be chosen independently of the Debye length.

    Different from the unipolar case, the formal quasi-neutral limit for the two-fluid Euler–Maxwell equations is the compressible type. Here, we are going to establish the quasi-neutral limit for the two-fluid system (2) under well-prepared initial data, which means the compatibility conditions (9) are satisfied. Based on the asymptotic expansion and the iteration techniques, we proved rigorously the main result stated in Theorem 2.3. Formally, setting ε=0, we obtain the compressible limit system (5). Note, too, that the displacement current and the charge separation are neglected in (5), which is essential different from the Euler–Maxwell equations which are symmetrizable and hyperbolic. Moreover, the singularities in the coupling electromagnetic field can also not be cancelled by a symmetrizer of hyperbolic systems [6], which leads the straightforward energy method invalid. We solve these difficulties by introducing the general vorticity and vector analysis formulas (see (51) in Lemma 3.3). In the mean time, long-time existence for smooth solutions of the two-fluid Euler–Maxwell equations as ε0 is also obtained provided that the smooth solution of one-fluid Euler equations exists. Indeed, if the initial data are not well-prepared, we cannot obtain the uniform energy estimates because of u(0)iu(0)eO(ε). Therefore, extending the result to the general case is not so obvious since we not only need to obtain the initial layer corrections but also to construct a new energy method, and thus is also open.

    The paper is organized as follows. In Sect. 2, we perform the formal asymptotic analysis and give the main result stated in Theorem 2.3. In Sect. 3, we justify rigorously the uniform (in ε) energy estimates for the error system (22) by the iteration techniques. Finally, we complete the proof of Theorem 2.3 by taking the limit of the sequences satisfying Cauchy's criterion in Sect. 4.

    Before proceeding, let us introduce the notations and lemmas which will be frequently used throughout this paper. We denote by Hs(R3) the standard Sobolev's space in the whole space R3, and denote by Hs the norm of the Banach space Hs(R3). In addition, we denote α as the multi-index, and α=α1x1α2x2α3x3, |α|=α1+α2+α3. Moreover, C is the general constant independent of the Debye length ε.

    In the following, we state the basic Moser-type calculus inequalities which will be used widely in the error estimates.

    Lemma 1.1. Let α be any multi-index with |α|=k, k1 and p(1,). Then there holds

    α(fg)LpCfLp1αgLp2+CαfLp3gLp4,
    [α,f]gLpCfLp1α1gLp2+CαfLp3gLp4, (3)

    where f,gS, the Schwartz class, and p2,p3(1,) such that

    1p=1p1+1p2=1p3+1p4.

    We make the following ansatz for ν=i,e in terms of the Debye length ε to the initial value problem (2)

    (nεν,uεν,Eε,Bε)=k0ε2k(n(k)ν,u(k)ν,E(k),B(k)). (4)

    Plugging the formal expansion (4) into system (2), we can obtain the following results.

    (ⅰ)The leading term (n(0)i,u(0)i,n(0)e,u(0)e,E(0),B(0)) satisfies

    {tn(0)+n(0)divu(0)ν+u(0)νn(0)=0,mν(tu(0)ν+u(0)νu(0)ν)+hν(n(0))=qν(E(0)+u(0)ν×B(0)),×B(0)=n(0)(u(0)iu(0)e),tB(0)+×E(0)=0,divB(0)=0.(5a)(5b)(5c)(5d)

    The local existence of smooth solutions cannot be obtained directly by the result of [5] since the displacement current and the charge separation are neglected in the limit system (5). In order to overcome the difficulty, we introduce the general vorticity ω(0)ν=×(u(0)ν+qνmνA(0)), where we have used divB(0)=0, which implies there exists some magnetic potential A(0) such that ×A(0)=B(0). Hence, we substitute the following equation for (5b)

    tω(0)ν+×(u(0)ν×ω(0)ν)=0, (6)

    namely,

    tω(0)ν+ω(0)νdivu(0)ν+u(0)νω(0)νω(0)νu(0)ν=0. (7)

    Taking the inner product of (7) with ω(0)ν, we derive

    12ddtω(0)ν2L2Cu(0)νH3ω(0)ν2L2. (8)

    We supplement the above limit system (5) with the initial data

    n(0)i0=n(0)e0=n(0)0,ω(0)ν0=0,×B(0)0=n(0)0(u(0)i0u(0)e0). (9)

    In view of (8), we get ω(0)ν=0 (ν=i,e) are naturally preserved for all time. Moreover, it follows from (5c) and ω(0)ν=×u(0)ν+qνmνB(0) that

    0=×(ω(0)iω(0)e)=×(×(u(0)iu(0)e))+mi+memimen(0)(u(0)iu(0)e). (10)

    Taking the inner product of (10) with u(0)iu(0)e, it holds

    ×(u(0)iu(0)e)2L2+mi+memimen(0)(u(0)iu(0)e)2L2=0,

    which implies u(0)i=u(0)e.

    Therefore, for well-prepared initial data (9), solutions (n(0)ν,u(0)ν,E(0),B(0)) (ν=i,e) to system (5) are the smooth solutions of the following compressible Euler equation

    {tn(0)+n(0)divu(0)+u(0)n(0)=0,tu(0)+u(0)u(0)+h0(n(0))=0,E(0)=ψ,(11a)(11b)(11c)

    where

    ψ=mimemi+me(hi(n(0))mihe(n(0))me), (12)

    and

    h0(n(0))=hi(n(0))+he(n(0))mi+me. (13)

    Theorem 2.1. Let ˜s3, and (n(0)0,u(0)0)H˜s be any given initial data satisfying n(0)0>0. Then there exists some 0<T+, the maximal time of existence, such that the initial value problem (11) has a unique solution such that, for any T0<T,

    (n(0),u(0))L(0,T0;H˜s+3),E(0)L(0,T0;H˜s+2).

    (ⅱ)For any j1, provided that we have proved the profiles

    (n(k)ν,u(k)ν,E(k),B(k))0kj1

    are smooth as much as we want in previous steps, we can get the following linear system satisfied by (n(j)ν,u(j)ν,E(j),B(j)) (ν=i,e)

    {tn(j)ν+n(0)divu(j)ν+n(j)νdivu(0)+u(0)n(j)ν+u(j)νn(0)=f1ν,mν(tu(j)ν+u(0)u(j)ν+u(j)νu(0))+(hν(n(0))n(j)ν)=qν(E(j)+u(0)×B(j))+f2ν,×B(j)=tE(j1)+n(0)(u(j)iu(j)e)+u(0)(n(j)in(j)e)+f3,tB(j)+×E(j)=0,n(j)in(j)e=divE(j1),(14a)(14b)(14c)(14d)(14e)

    where

    {f1ν=j1k=1(n(k)νdivu(jk)ν+u(k)νn(jk)ν),(15a)f2ν=mνj1k=1(u(k)νu(jk)ν)(hj1ν(n(k)ν)kj1)+qνj1k=1(u(k)ν×B(jk)),(15b)f3=j1k=1n(k)ν(u(jk)iu(jk)e).(15c)

    Here, we have used the following relation

    hν(n(0)+j1ε2jn(j)ν)=hν(n(0))+hν(n(0))j1(ε2jn(j)ν)+j2ε2jhj1ν((n(k)ν)kj1), (16)

    where

    hj1ν((n(k)ν)kj1)=1j!ddε2jhν(n(0)+j1ε2jn(j)ν)|ε=0hν(n(0))n(j)ν. (17)

    Obviously, the profile (f1ν,f2ν,f3) only depends on the known terms by the previous steps.

    Letting (n(j),u(j))=(n(j)i,u(j)i)(n(j)e,u(j)e), we rewrite the induction system (14) as the following form

    {tω(j)+ω(j)divu(0)+u(0)ω(j)ω(j)u(0)=×(f2if2e),(18a)×ω(j)=divu(j)Δu(j)+(1mi+1me)(n(0)u(j)+divE(j1)u(0)+tE(j1)+f3),(18b)

    where ωj=×u(j)+(1mi+1me)B(j). Existence of solutions to system (18) has been derived by [8]. Moreover, we need the following compatibility conditions

    n(j)i,0n(j)e,0=divE(j1)0,   divB(j)0=0. (19)

    Letting (n(j)+,u(j)+)=(n(j)i,u(j)i)+(n(j)e,u(j)e), we can rewrite the induction system (14) as

    {tn(j)++n(0)divu(j)++n(j)+divu(0)+u(0)n(j)++u(j)+n(0)=f1i+f1e,(20a)(mi+me)(tu(j)++u(0)u(j)++u(j)+u(0))+((hi(n(0))+he(n(0)))n(j)+)=(mime)(tu(j)+u(0)u(j)+u(j)u(0))+((hi(n(0))he(n(0)))divE(j1))+f2i+f2e.(20b)

    Then (n(j)+,u(j)+) (j1) are the solutions to the linear nonhomogeneous compressible Euler equations [5]. Finally, combining the definition (n(j),u(j))=(n(j)i,u(j)i)(n(j)e,u(j)e) and the equation (14e), we derive the following result

    Theorem 2.2. Let ˜s3, and (n(j)ν,0,u(j)ν,0,E(j)0,B(j)0) (ν=i,e) be any given initial data satisfying (19) with n(j)ν,0>0. Then there exists some T0>0, such that the initial value problem (14) has a unique solution that satisfies

    (n(j)ν,u(j)ν,E(j),B(j))L(0,T0;H˜s+3×H˜s+3×H˜s+2×H˜s+2).

    Take the ansatz for ν=i,e in terms of the Debye length ε

    {nεν=n(0)+mk=1ε2kn(k)ν+ε2mΨν=¯nν+ε2mΨν,(21a)uεν=u(0)+mk=1ε2ku(k)ν+ε2mUν=¯uν+ε2mUν,(21b)Eε=E(0)+mk=1ε2kE(k)+ε2mF=ˉE+ε2mF,(21c)Bε=mk=1ε2kB(k)+ε2mG=ˉB+ε2mG,(21d)

    where (nεi,uεi,nεe,uεe,Eε,Bε) is the exact solution to system (2), (n(0),u(0),E(0)) is the solution to the limit system (11), (n(j)i,u(j)i,n(j)e,u(j)e,E(j),B(j)) is the solution to the linear system (14) and (Ψi,Ui,Ψe,Ue,F,G) is the remainder term. By careful computation, we derive the following system satisfied by the remainder term

    {tΨν+uενΨν+nενdivUν=Ψνdiv¯uνUν¯nνε2ν1,(22a)mν(tUν+uενUν)+hν(nεν)ΨνqνFqν(uεν×G)=mνUν¯uν+qνUν×ˉBhν(ˉnν)Ψνˉnνε2ν2,(22b)ε2tF×G+nεiUinεeUe+Ψi¯uiΨe¯ue=ε23,(22c)tG+×F=0,divG=0,(22d)ε2divFΨi+Ψe=0,(22e)

    where the profile (ν1,ν2,3) is O(1), and depends only on the known and sufficiently smooth functions.

    Let Wν=(Ψν,Uν). We can rewrite the remainder system (22) in the form

    {tWν+3i=1Ai(nεν,uεν)xiWν+J=ν1ε2ν0,(23a)ε2tF×G+nεiUinεeUe+Ψi¯uiΨe¯ue=ε23,(23b)ε2divFΨi+Ψe=0,(23c)tG+×F=0,divG=0,(23d)(Wν,F,G)|t=0=(Wν0,F0,G0),(23e)

    where

    Ai(nεν,uεν)=(uενnενeihν(nεν)mνeiuενI), (24)
    J=(0qνFqν(uεν×G)mν), (25)
    ν1=(Ψνdiv¯uνUν¯nνUν¯uν+qνmνUν×ˉB1mν(hν(ˉnν)Ψνˉnν)), (26)

    and

    ν0=(ν1,ν2).

    Moreover, Ai(nεν,uεν) can be symmetrized by the following symmetric and positive matrix

    A0(nεν)=(hν(nεν)mν00nενI), (27)

    where I is the 3×3 identity matrix.

    Thanks to the symmetrizable structure of system (22), we obtain the standard existence theory of local smooth solutions [5]. Based on this, we will prove the main result stated in the following

    Theorem 2.3. Let 2m be any integer with 2m>4. Assume that the initial data (nεν,0,uεν,0,Eε0,Bε0) (ν=i,e) satisfy the compatibility conditions (9) and

    (nεν,0n(0)02mj=1εjn(j)ν,0,uεν,0u(0)02mj=1εju(j)ν,0,Eε0E(0)02mj=1εjE(j)0,   Bε02mj=1εjB(j)0)H4Cε2m4,

    where C is some positive constant independent of ε. Then there exist ε0>0 and solution (nεν,uεν,Eε,Bε) of system (2) with initial data (nεν,0,uεν,0,Eε0,Bε0) on [0,Tε) with lim infε0TεT, the maximal existence time of solutions to the limit system (11). In particular, for every T0<T, 0<ε<ε0, there holds

    supt[0,T0](nενn(0)2mj=1εjn(j)ν,uενu(0)2mj=1εju(j)ν,EεE(0)2mj=1εjE(j),Bε2mj=1εjB(j))(t)H3Cε2m4.

    To state the main theorem, we introduce the set Sεe, of function in L(0,T0;H3) that satisfies (Wν,F,G)(x,0)=(Wν0,F0,G0) and

    (Wν,εF,G)H3˜Cε3,

    where ˜C is a constant independent of ε, which will be determined later. Moreover, we define the weighted norm as

    |||(Wν,εF,G)|||ε,s=s|β|=0ε|β|β(Wν,εF,G)L2, (28)

    for s0. Our next goal is to prove system (23) has a smooth solution (Wν,F,G)Sεe for appropriate ˜C and ε, which implies the desired estimates stated in Theorem 2.3. As in [9,18], we consider the nonlinear remainder system coupled with Maxwell equations by the following iteration

    (Wν,0,F0,G0)=(Wν0,F0,G0),

    where Wν,0=(Ψν,0,Uν,0), and

    (Wν,p+1,Fp+1,Gp+1)=Φ(Wν,p,Fp,Gp),

    where Φ maps vector (Wν,p,Fp,Gp) into solution (Wν,p+1,Fp+1,Gp+1) of the following linear system

    {tWν,p+1+3i=1˜Ai(nε,pν,uε,pν)xiWν,p+1+˜Jν,p+1=˜ν,p+1,(29a)ε2tFp+1×Gp+1+nε,piUi,p+1nε,peUe,p+1+¯uiΨi,p+1¯ueΨe,p+1=ε23,(29b)tGp+1+×Fp+1=0,divGp+1=0,(29c)ε2divFp+1Ψi,p+1+Ψe,p+1=0,(29d)(Wν,p+1,Fp+1,Gp+1)|t=0=(Wν0,F0,G0),(29e)

    where

    (nε,pν,uε,pν)=(ˉnν,ˉuν)+ε2m(Ψν,p,Uν,p), (30)
    ˜Jν,p+1=(0qνFp+1qν(uε,pν×Gp+1)mν), (31)
    ˜Ai(nε,pν,uε,pν)=(uε,pνnε,pνeihν(nε,pν)mνeiuε,pνI), (32)

    and

    ˜ν,p+1=(Ψν,p+1div¯uνUν,p+1¯nνUν,p+1¯uν+qνmνUν,p+1×ˉB1mν(hν(ˉnν)Ψν,p+1ˉnν))ε2ν0. (33)

    Denote

    Ep,ε(Wν,p,Gp,εFp)H4. (34)

    Proposition 1. Let 2m be any integer with 2m>4, and

    (nε,pν,uε,pν,Wν,p+1,Gp+1,Fp+1) (ν=i,e)

    be the solutions of the iteration equations (29). Assume that

    |||(Wν0,εF0,G0)|||ε,4C,

    where C is a generic constant. Then for all t[0,T0], there exists a positive constant ε0 such that, for all p1,

    |||(Wν,p,εFp,Gp)(t)|||ε,4C.

    Further, by the definitions (28) and (34) that

    Ep,εCε4,

    for any 0<ε<ε0.

    The following section is devoted to the proof of Proposition 1.

    Obviously, since

    (Wν,0,F0,G0)=(Wν0,F0,G0),

    we can obtain

    E0,εCε4. (35)

    Now we assume there exists a sufficiently small ε such that

    Ep,εCε4. (36)

    Hence, we need to prove that

    Ep+1,εCε4. (37)

    Recalling the expansion (21), we immediately obtain that there exists positive constant ε0 such that for any 0<ε<ε0, nε,pν=ˉnν+ε2mΨν,p is bounded from above and below, namely,

    n(0)2nε,pν3n(0)2. (38)

    Here, we need the condition 2m>4. Similar arguments applying to (nε,pν,uε,pν)Hk with 0k4 yields

    Lemma 3.1. For any 2m>4 and sufficiently small ε, it holds that

    (nε,pν,uε,pν)HkC(1+ε2mEp,ε). (39)

    Proof. By the expansion (21) and Sobolev imbedding that

    (nε,pν,uε,pν)HkC+Cε2m(Ψν,p,Uν,p)HkC(1+ε2mEp,ε). (40)

    Analogously, we get

    Lemma 3.2. Under the same assumptions as in Lemma 3.1 hold. Then we obtain

    tnε,pνLC+Cε4mE2p,ε. (41)

    In the following, we first give the L2-estimates.

    Lemma 3.3. For any t[0,T0], there exists a sufficiently small ε>0 such that

    ddt(Wi,p+1,We,p+1,εFp+1,Gp+1)2L2C(1+ε4mE2p,ε)(Wi,p+1,We,p+1,εFp+1,Gp+1)2L2+Cε2. (42)

    Proof. Applying the operator mνˆA0 to system (29) and taking the inner product with (Wν,p+1,Fp+1,Gp+1), we obtain

    mνν=i,eTWν,p+1,Wν,p+1+ddt(εFp+1,Gp+1)2L2+nε,piUi,p+1Fp+1nε,peUe,p+1Fp+1ν=i,eqνnε,pνFp+1Uν,p+1=¯uiΨi,p+1Fp+1+¯ueΨe,p+1Fp+1+ν=i,eqνnε,pνuε,pν×Gp+1Uν,p+1+ν=i,eˆA0˜ν,p+1Wν,p+1ε2ˆA03Fp+15i=1Ii0, (43)

    where

    ˆA0=(hν(nε,pν)mν00nε,pνI), (44)

    and the abbreviated operator T is defined by

    TWν,p+1,Wν,p+1=12ddtˆA0Wν,p+12L212(t,)(ˆA0,ˆA)|Wν,p+1|2, (45)

    where ˆA=ˆA0[(˜A1,˜A2,˜A3)(nε,pν,uε,pν)]. Here, we have used the vector analysis formula

    ×Gp+1Fp+1+×Fp+1Gp+1=div(Fp+1×Gp+1)=0. (46)

    Thanks to Sobolev embedding H2L, (36), (38) and (41), we derive

    tˆA0L(1+hν(nε,pν))tnε,pνLC+Cε4mE2p,ε, (47)

    and

    divˆALC(1+hν(nε,pν))(nε,pν,divuε,pν)L+C(1+hν(nε,pν))nε,pνuε,pνLC+Cε4m(Ψν,p,Uν,p)2H3C+Cε4mE2p,ε. (48)

    Hence,

    12(t,)(ˆA0,ˆA)|Wν,p+1|2C(1+ε4mE2p,ε)Wν,p+12L2. (49)

    Inserting this into (43), and using (46) and qi=1,qe=1, we have

    ddt(ˆA0Wν,p+1,εFp+1,Gp+1)2L25i=1Ii0+C(1+ε4mE2p,ε)Wν,p+12L2. (50)

    From (29d), the first and second term on the right hand side of (43) can be accordingly decomposed into

    I10+I20=(Ψi,p+1Ψe,p+1)¯uiFp+1(¯ui¯ue)Ψe,p+1Fp+1=ε2divFp+1Fp+1¯ui(¯ui¯ue)Ψe,p+1Fp+1.

    By the vector analysis formulation

    divff=div(ff)12(|f|2)×f×f, (51)

    where f is a vector function, I10+I20 can be further decomposed by

    I10+I20=ε2(div(Fp+1Fp+1)12(|Fp+1|2)×Fp+1×Fp+1)¯ui(¯ui¯ue)Ψe,p+1Fp+1ε2×Fp+1×Fp+1¯ui+C(Ψe,p+1,εFp+1)2L2,

    thanks to (¯ui¯ue)O(ε2) since u(0)i=u(0)e. To deal with the term ε2×Fp+1×Fp+1¯ui, we apply (29b)×Gp+1(29c)×ε2Fp+1 to derive

    ε2t(Fp+1×Gp+1)×Gp+1×Gp+1ε2×Fp+1×Fp+1  +(nε,piUi,p+1nε,peUe,p+1)×Gp+1+(¯uiΨi,p+1¯ueΨe,p+1)×Gp+1  +ε23×Gp+1=0. (52)

    Thus, we have

    ε2×Fp+1×Fp+1¯ui=ε2ddt(Fp+1×Gp+1)¯uiε2(Fp+1×Gp+1)t¯uidiv(Gp+1Gp+1)ˉu+12(|Gp+1|2)ˉu+(nε,piUi,p+1nε,peUe,p+1)×Gp+1¯ui+(¯uiΨi,p+1¯ueΨe,p+1)×Gp+1¯ui+ε23×Gp+1¯uiε2ddt(Fp+1×Gp+1)¯ui+Cε4+C(Wν,p+1,Gp+1,εFp+1)2L2,

    thanks to divGp+1=0 and (38). Recalling ˆA0 is positively definite, we have

    ν=i.eˆA0Wν,p+12L2CWν,p+12L2.

    In fact, since ε is sufficiently small, we get

    (εFp+1,Gp+1)2L2ε2(Fp+1×Gp+1)¯uiC(εFp+1,Gp+1)2L2.

    For the last three terms I30I50, by Young's inequality and Hölder inequality, we get

    I30I50Cε2+C(1+ε2mEp,ε)(εFp+1,Gp+1,Wν,p+1)2L2.

    Putting the above estimates together, the proof of Lemma 3.3 is then complete.

    Lemma 3.4. Let 1k4 be an integer, α be a multi-index with |α|=k, then we have

    ddt(αWi,p+1,αWe,p+1)2L2ν=i,eqνnε,pναFp+1αUν,p+1Cε4+C(1+ε4mE2p,ε)(Gp+1,Wν,p+1)2Hk. (53)

    Proof. First, applying the operator α to (29a), we derive

    tαWν,p+1+3i=1˜Ai(nε,pν,uε,pν)xiαWν,p+1+α˜Jν,p+1=α˜ν,p+1+H1, (54)

    where

    H1=3i=1[α,˜Ai(nε,pν,uε,pν)]xiWν,p+1. (55)

    Then taking inner product of (54) with mνˆA0αWν,p+1, we derive

    mνTαWν,p+1,αWν,p+1qνnε,pναFp+1αUν,p+1=mνˆA0α˜ν,p+1αWν,p+1+qνnε,pνα(uε,pν×Gp+1)αUν,p+1+mνˆA0H1αWν,p+1=3i=1Iiε. (56)

    Again, the abbreviated operator T is defined by

    TWν,p+1,Wν,p+1=12ddtˆA0αWν,p+12L212(t,)(ˆA0,˜A)|αWν,p+1|2,

    where ˜A=ˆA0[(˜A1,˜A2,˜A3)(nε,pν,uε,pν)]. For the first term on the left side hand of (56), we can employ arguments similar to those used in the estimate of (45) to obtain

    mνTαWν,p+1,αWν,p+1C(1+ε4mE2p,ε)Wν,p+12Hk. (57)

    On the other hand, using Young's inequality, Hölder inequality, Lemma 1.1, together with (38), we get

    I1εCWν,p+12Hk+Cε4,

    and

    I2εC(1+ε2mEp,ε)(Gp+1,Wν,p+1)2Hk.

    For the commutator term I3ε, by Lemma 1.1, we have,

    I3εCˆA0LH1L2αWν,p+1L2C(˜AiLWν,p+1Hk1+˜AiHkWν,p+1L)αWν,p+1L2C(1+ε2mEp,ε)Wν,p+12Hk.

    Finally, putting all the above estimates together, we completes the proof of Lemma 3.4, thanks to (38).

    Lemma 3.5. Under the same conditions in Lemma 3.4, we have the following estimate

    ddt(εαFp+1,αGp+1)2L2+ν=i,eqνnε,pναUν,p+1αFp+1C(Gp+1,Wν,p+1,εFp+1)2Hk+Cε2(1+ε4mE2p,ε)Wν,p+12Hk1+Cε2. (58)

    Proof. An application of the operator α to (29b) and (29c) leads to

    {ε2tαFp+1α×Gp+1+nε,piαUi,p+1nε,peαUe,p+1+¯uiαΨi,p+1¯ueαΨe,p+1=ε2α3+H2+H3,(59a)tαGp+1+α×Fp+1=0,αdivGp+1=0,(59b)

    where

    H2=[α,nε,pi]Ui,p+1[α,nε,pe]Ue,p+1, (60)

    and

    H3=[α,¯ui]Ψi,p+1[α,¯ue]Ψe,p+1. (61)

    Multiplying (62) by (αFp+1,αGp+1), it holds

    ddt(εαFp+1,αGp+1)2L2+nε,piαUi,p+1αFp+1nε,peαUe,p+1αFp+1=¯uiαΨi,p+1αFp+1+¯ueαΨe,p+1αFp+1+H2αFp+1+H3αFp+1ε2α3αFp+15i=1Jiε. (62)

    Here, the vector analysis formula

    div(f×g)=(×f)g(×g)f (63)

    has been used again, for any vector functions f and g.

    We proceed to control the five terms on the right hand side of (62). For J1ε and J2ε, we can employ arguments similar to those used in the estimates of I10 and I20 to obtain

    J1ε+J2εε2ddt(αFp+1×αGp+1)¯ui+C(Gp+1,Wν,p+1,εFp+1)2Hk+Cε4.

    The usual estimate (3) on commutator leads to

    J3εCεFp+1Hk1εH2L2CεFp+1Hk1ε(nε,pνLUν,p+1Hk1+Uν,p+1Lnε,pνHk)CεFp+12Hk+Cε2(1+ε4mE2p,ε)Wν,p+12Hk1.

    Similarly, the forth term J4ε can be bounded by

    J4εCεFp+12Hk+Cε2Wν,p+12Hk1.

    It follows from Young's inequality and Hölder inequality, the last term J5ε can be estimated as

    J5εCε2+CεFp+12Hk.

    Adding these above estimates together, the proof of Lemma 3.5 is then complete.

    Proof. Combining Lemmas 3.4 and 3.5, we have

    ddt(αWp+1,εαFp+1,αGp+1)2L2Cε2+C(1+ε4mE2p,ε)(Gp+1,Wν,p+1,εFp+1)2Hk+Cε2(1+ε4mE2p,ε)Wν,p+12Hk1, (64)

    where 1|α|=k4.

    Recalling Lemma 3.3 and the weighted energy norm (28), we obtain, for t[0,T0],

    |||(Wp+1,εFp+1,Gp+1)(t)|||2ε,4C|||(Wp+1,εFp+1,Gp+1)(0)|||2ε,4+Cε2+Ct0(1+ε4mE2p,ε)|||(Wp+1,εFp+1,Gp+1)(τ)|||2ε,4. (65)

    From (36), there exists ε1>0 such that for any 0<ε<ε1, we have

    ε2mEp,ε<1, (66)

    thanks to the assumption 2m>4. Using Gronwall inequality, (65) and (66), we infer that

    supt[0,T0]|||(Wν,p+1,εFp+1,Gp+1)(t)|||2ε,4˜C, (67)

    where ˜C=e2T0(C0+1)T0, and C0 is a constant dependent on the initial data. A straightforward calculation implies

    Ep+1,εCε4|||(Wν,p+1,εFp+1,Gp+1)(t)|||ε,4˜Cε4. (68)

    This completes the proof of Proposition 1.

    Proof. Set

    (Wν,p,εFp,Gp)=(Wν,p+1,εFp+1,Gp+1)(Wν,p,εFp,Gp),

    where

    Wν,p=(Ψν,p,Uν,p)=(Ψν,p+1,Uν,p+1)(Ψν,p,Uν,p).

    By careful computation, we obtain the system satisfied by (Wν,p,εFp,Gp)

    {tWν,p+3i=1˜Ai(nε,pν,uε,pν)xiWν,p+˜Jν,p+1˜Jν,p˜ν,p+1+˜ν,p=3i=1(˜Ai(nε,pν,uε,pν)˜Ai(nε,p1ν,uε,p1ν))xiWν,p,(69a)ε2tFp×Gp+nε,piUi,pnε,peUe,p+¯uiΨi,p¯ueΨe,p=(nε,pinε,p1i)Ui,p+(nε,penε,p1e)Ue,p,(69b)tGp+×Fp=0,divGp=0,(69c)ε2divFpΨi,p+Ψe,p=0,(69d)(Wν,p,εFp,Gp)|t=0=0.(69e)

    Based on the similar arguments of Lemmas 3.3-3.5 and (64)-(68), we deduce

    supt[0,T0]|||(Wν,p,εFp,Gp)(t)|||ε,3c|||(Wν,p1,εFp1,Gp1)(t)|||ε,3,

    where 0<c<1, which depends on the bound C in Proposition 1, thanks to (69e), 2m>4 and (nε,pν,uε,pν)=(ˉnν,ˉuν)+ε2m(Ψν,p,Uν,p). This implies

    (Ψi,p,Ui,p,Ψe,p,Ue,p,εFp,Gp)

    is a Cauchy sequence, and hence there exists

    (Ψi,Ui,Ψe,Ue,εF,G)C([0,T0];H3)

    such that, as p, we can obtain the convergence of the whole sequence

    (Wi,p,We,p,Gp,εFp)p1

    to (Wi,We,G,εF), as well as

    supt[0,T0](Wi,We,εF,G)(t)H3Cε3,

    for any 0<ε<ε0. Indeed, in a similar manner to [9], we infer that (Wi,We,εF,G)Ci([0,T0];H3i) for i=0,1. Passing the limit p in the system (29), we obtain system (23) admits a classical solution (Wi,We,F,G) that satisfies

    supt[0,T0](Wi,We,F,G)(t)H3Cε4.

    With the aid of the expansion (21), (nεi,uεi,nεe,uεe,Eε,Bε) converges strongly to (n(0),u(0),n(0),u(0),E(0),0) in C(0,T0;H3), for any integer 2m>4.

    The proof of Theorem 2.3 is then complete.

    We would like to thank the editor and the anonymous referees for the valuable comments and corrections which greatly improved our original manuscript.



    [1] Quasineutral limit of the Euler-Poisson system for ions in a domain with boundaries. Indiana Univ. Math. J. (2013) 62: 359-402.
    [2] D. Gérard-Varet, D. Han-Kwan and F. Rousset, Quasineutral limit of the Euler-Poisson system for ions in a domain with boundaries Ⅱ, J. Éc. Polytech. Math., 1 (2014), 343–386. doi: 10.5802/jep.13
    [3] Global solutions of the Euler–Maxwell two-fluid system in 3D. Ann. of Math. (2) (2016) 183: 377-498.
    [4] Quasineutral limit of the two-fluid Euler-Poisson system in a bounded domain of R3. J. Math. Anal. Appl. (2019) 469: 169-187.
    [5] The Cauchy problem for quasi-linear symmetric hyperbolic systems. Arch. Rational Mech. Anal. (1975) 58: 181-205.
    [6] Singular limits of quasilinear hyperbolic systems with large parameters and the incompressible limit of compressible fluids. Comm. Pure Appl. Math. (1981) 34: 481-524.
    [7] The asymptotic behavior and the quasineutral limit for the bipolar Euler-Poisson system with boundary effects and a vacuum. Chin. Ann. Math. Ser. B (2013) 34: 529-540.
    [8] Quasineutral limit for the compressible quantum Navier-Stokes-Maxwell equations. Commun. Math. Sci. (2018) 16: 363-391.
    [9] A. Majda, Compressible Fluid Flow and Systems of Conservation Laws in Several Space Variables, Applied Mathematical Sciences, 53. Springer-Verlag, New York, 1984. doi: 10.1007/978-1-4612-1116-7
    [10] Y. J. Peng, Global existence and long-time behavior of smooth solutions of two-fluid Euler-Maxwell equations, Ann. Inst. H. Poincaré Anal. Non Lińeaire, 29 (2012), 737–759. doi: 10.1016/j.anihpc.2012.04.002
    [11] Convergence of compressible Euler-Maxwell equations to compressible Euler-Poisson equations. Chin. Ann. Math. Ser. B (2007) 28: 583-602.
    [12] Convergence of compressible Euler-Maxwell equations to incompressible Euler equations. Comm. Partial Differential Equations (2008) 33: 349-476.
    [13] Rigorous derivation of incompressible e-MHD equations from compressible Euler-Maxwell equations. SIAM J. Math. Anal. (2008) 40: 540-565.
    [14] Asymptotic expansions in two-fluid compressible Euler-Maxwell equations with small parameters. Discrete Contin. Dyn. Syst. (2009) 23: 415-433.
    [15] Boundary layers and quasi-neutral limit in steady state Euler-Poisson equations for potential flows. Nonlinearity (2004) 17: 835-849.
    [16] Asymptotic behaviors for the full compressible quantum Navier-Stokes-Maxwell equations with general initial data. Discrete Contin. Dyn. Syst. B (2019) 24: 5149-5181.
    [17] A boundary layer problem for an asymptotic preserving scheme in the quasi-neutral limit for the Euler-Poisson system. SIAM J. Appl. Math. (2010) 70: 1761-1787.
    [18] Quasineutral limit of Euler-Poisson system with and without viscosity. Comm. Partial Differential Equations (2004) 29: 419-456.
    [19] Incompressible limit of isentropic Navier-Stokes equations with Navier-slip boundary. Kinet. Relat. Models (2018) 11: 469-490.
    [20] Global classical solutions to the compressible Euler-Maxwell equations. SIAM J. Math. Anal. (2011) 43: 2688-2718.
    [21] The non-relativistic limit of Euler-Maxwell equations for two-fluid plasma. Nonlinear Anal. (2010) 72: 1829-1840.
    [22] Convergence of the Euler-Maxwell two-fluid system to compressible Euler equations. J. Math. Anal. Appl. (2014) 417: 889-903.
  • This article has been cited by:

    1. Fei Shi, Incompressible limit of Euler equations with damping, 2021, 30, 2688-1594, 126, 10.3934/era.2022007
  • Reader Comments
  • © 2020 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(2382) PDF downloads(120) Cited by(1)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog