Research article Special Issues

Numerical spectral analysis of standing waves in quantum hydrodynamics with viscosity

  • §Current address: Department of Mathematics, University of Surrey, Guildford, United Kingdom
  • We study the spectrum of the linearization around standing wave profiles for two quantum hydrodynamics systems with linear and nonlinear viscosity. The essential spectrum for such profiles is stable; we investigate the point spectrum using an Evans function technique. For both systems we show numerically that there exists a real unstable eigenvalue, thus providing numerical evidence for spectral instability.

    Citation: Delyan Zhelyazov. Numerical spectral analysis of standing waves in quantum hydrodynamics with viscosity[J]. Mathematics in Engineering, 2024, 6(3): 407-424. doi: 10.3934/mine.2024017

    Related Papers:

    [1] Federica Di Michele, Bruno Rubino, Rosella Sampalmieri, Kateryna Stiepanova . Stationary solutions to a hybrid viscous hydrodynamic model with classical boundaries. Mathematics in Engineering, 2024, 6(5): 705-725. doi: 10.3934/mine.2024027
    [2] Hao Zheng . The Pauli problem and wave function lifting: reconstruction of quantum states from physical observables. Mathematics in Engineering, 2024, 6(4): 648-675. doi: 10.3934/mine.2024025
    [3] Roberto Feola, Felice Iandoli, Federico Murgante . Long-time stability of the quantum hydrodynamic system on irrational tori. Mathematics in Engineering, 2022, 4(3): 1-24. doi: 10.3934/mine.2022023
    [4] Riccardo Adami, Raffaele Carlone, Michele Correggi, Lorenzo Tentarelli . Stability of the standing waves of the concentrated NLSE in dimension two. Mathematics in Engineering, 2021, 3(2): 1-15. doi: 10.3934/mine.2021011
    [5] Dheeraj Varma, Manikandan Mathur, Thierry Dauxois . Instabilities in internal gravity waves. Mathematics in Engineering, 2023, 5(1): 1-34. doi: 10.3934/mine.2023016
    [6] Lars Eric Hientzsch . On the low Mach number limit for 2D Navier–Stokes–Korteweg systems. Mathematics in Engineering, 2023, 5(2): 1-26. doi: 10.3934/mine.2023023
    [7] Biagio Cassano, Lucrezia Cossetti, Luca Fanelli . Spectral enclosures for the damped elastic wave equation. Mathematics in Engineering, 2022, 4(6): 1-10. doi: 10.3934/mine.2022052
    [8] George Contopoulos . A Review of the “Third” integral. Mathematics in Engineering, 2020, 2(3): 472-511. doi: 10.3934/mine.2020022
    [9] Paola F. Antonietti, Ilario Mazzieri, Laura Melas, Roberto Paolucci, Alfio Quarteroni, Chiara Smerzini, Marco Stupazzini . Three-dimensional physics-based earthquake ground motion simulations for seismic risk assessment in densely populated urban areas. Mathematics in Engineering, 2021, 3(2): 1-31. doi: 10.3934/mine.2021012
    [10] Emilia Blåsten, Fedi Zouari, Moez Louati, Mohamed S. Ghidaoui . Blockage detection in networks: The area reconstruction method. Mathematics in Engineering, 2019, 1(4): 849-880. doi: 10.3934/mine.2019.4.849
  • We study the spectrum of the linearization around standing wave profiles for two quantum hydrodynamics systems with linear and nonlinear viscosity. The essential spectrum for such profiles is stable; we investigate the point spectrum using an Evans function technique. For both systems we show numerically that there exists a real unstable eigenvalue, thus providing numerical evidence for spectral instability.



    In this paper, we consider two quantum hydrodynamics (QHD) systems for which we investigate the spectrum of the linearized operator about standing wave profiles. The system with a linear viscosity term reads

    {ρt+mx=0,mt+(m2ρ+p(ρ))x=μmxx+k2ρ((ρ)xxρ)x, (1.1)

    where ρ0 is the density, m=ρu is the momentum, where u is the fluid velocity, p(ρ)=ργ with γ1 is the pressure, t0 and xR. Moreover, μ,k>0 are viscosity and dispersive coefficients, respectively. The function (ρ)xx/ρ is known as the Bohm potential, providing the model with a nonlinear third order dispersive term. These systems are used for instance in the theory of superfluidity [15] or to model semiconductor devices. In the case of nonlinear viscosity (see (3.1)) the term μmxx is replaced with μρ(mx/ρ)x. The latter describes the interaction between the superfluid and a normal fluid, or it can be interpreted as describing the interaction of the fluid with a background. Systems (1.1) and (3.1) can be viewed as mean-field limits of the de Broglie-Bohm pilot wave theory [4,5,6] with added viscosity terms of superfluid type.

    Standing waves for (1.1) are solutions of the form

    ρ(t,x)=P(x), m(t,x)=J(x),

    such that

    limx±P(x)=P±>0, limx±J(x)=J±.

    The first studies of models with dispersive terms were [12,22]; see also [11,13,20]. Moreover, existence of weak solutions for QHD systems has been considered in [1,2,3]. The spectral stability of traveling wave profiles for the p-system has been discussed in [14]. Concerning viscous QHD models, existence of traveling wave profiles for the system (1.1) was shown in [16] provided that the viscosity is sufficiently strong. The result was improved in [24] where global existence of dispersive shocks was proved without any restriction on the viscosity and dispersive coefficients and for arbitrary shock amplitude. A similar result about a related QHD system with nonlinear viscosity (see (3.1)) was obtained in [19]. Moreover, stability of traveling waves was established numerically in [17,18] in the case of non-monotone shocks, and analytically in [8,9] for small-amplitude profiles. These studies were complemented in [21] with a proof of local well-posedness and nonlinear decay of perturbations of subsonic constant states for (1.1).

    The aim of this paper is to study the spectrum of the linearization around standing waves for (1.1) and (3.2). Existence theory for such solutions was established in [24]. There, it was shown that the standing waves admited by the system (1.1) are pulses, i.e., P+=P and J+=J. Here below we recall the existence result [24, Theorem 4.1]. Let us denote by cs(ρ)=γργ1 the speed of sound and by U+=J+/P+ the velocity at the end state.

    Theorem 1.1 (existence of standing waves). There exists a non-constant standing wave solution of (1.1) with limx±[P(x),J(x)]=[P+,J+], if and only if 0<|U+|<cs(P+).

    This result shows, in particular, that there exist no standing waves with supersonic or sonic end states.

    The spectrum of the linearization around standing wave profiles for systems (1.1) and (3.2) consists of two parts: the essential spectrum and the point spectrum. It can be proved that the essential spectrum of such profiles is always stable (see Theorem 2.2 below). In the present paper, we carry out a numerical study of the point spectrum by using the Evans function method. We show that there exists an unstable real simple eigenvalue for both QHD systems with linear and nonlinear viscosity. Thereby, we provide numerical evidence for spectral instability of standing waves. This result is in contrast with the analytical and numerical results about spectral stability of traveling waves in [8,9,17,18]. A possible cause of the instability is the presence of a supersonic region along the standing wave profiles (see Section 4 and Conjecture 4.1 below).

    The paper is organized as follows. In Section 2 we consider the QHD system with linear viscosity (1.1). We introduce the equation solved by standing wave profiles and discuss their numerical computation. Then, we derive the linearization around such profiles, describe its essential spectrum, and recast the system in integrated variables. Section 2.6 contains the numerical result about point spectrum instability. In Section 3 we discuss the QHD system with nonlinear viscosity (see Eq (3.2) below) following the steps for the linear viscosity case. Finally, we conclude the paper with a discussion of the numerical results.

    Remark 1.2. The linearizations in Sections 2.3 and 3.2 can be obtained by setting s=0 in the corresponding linearizations in [16,18] which hold also for s=0. We provide them here for completeness.

    In this section we shall derive the equation satisfied by a standing wave profile for system (1.1). The Bohm potential can be rewritten in conservative form

    ρ((ρ)xxρ)x=12(ρ(lnρ)xx)x.

    Substituting the standing wave profiles P and J we get the following system of ODEs

    J=0, (2.1)
    (J2P+Pγ)=μJ+k22(P(lnP)). (2.2)

    From (2.1) it follows that the momentum is conserved along the profile

    J(x)=A, (2.3)

    where A=J+. Since J is constant we have J=0, hence the term coming from the viscosity in Eq (2.2) vanishes. Susbstituting Eq (2.3) in Eq (2.2) and proceeding as in [16, Section 2], we conclude that the density profile P solves the ODE

    P=2k2f(P)+P2P, (2.4)

    where f(P)=PγB+A2P, with B=(m2ρ+ργ)+.

    We shall compute numerically standing waves whose existence is guaranteed by Theorem 1.1. They correspond to homoclinic loops for the profile ODE (2.4). Since the steady-state for this dynamical system to which the profile converges is a saddle, the method to compute the profile from [17] cannot be used because it relies on one of the steady-states being stable. Instead, the profile will be computed by solving a nonlinear boundary value problem.

    To compute standing wave profiles we choose P+>0 and U+ that satisfy the condition of Theorem 1.1. Then, we choose a sufficiently large domain size L1>0 and integrate numerically (2.4) on the domain [L1,L1] subject to the boundary conditions P(L1)=P(L1)=P+, with the boundary value solver bvp4c in Matlab. As an initial guess we use the function

    P0(x)=P+cexp(x2) (2.5)

    for L1xL1, where c>0 is a parameter to be chosen so that the solver converges. From the qualitative analysis in [24, Section 4] we know that the profile has a single minimum, hence the initial guess has an appropriate shape. Using this initial guess the solver will not converge to the trivial constant solution.

    Because of the fact that the momentum is conserved along the profile we have u(x)=J+/P(x). Hence, the flow is supersonic in the region where the following inequality holds

    P(x)<((J+)2γ)1γ+1, (2.6)

    where the right-hand side is the density corresponding to sonic states. For our computations we use the parameter values

    P+=1, J+=0.8, γ=3/2, k=2, μ=0.1. (2.7)

    We have |U+|=0.8<1.22cs(P+), therefore the condition of Theorem 1.1 is satisfied. Moreover, we set L1=20 and c=0.8. For parameters (2.7) the right hand side of (2.6) is 0.71 and the flow is supersonic in the interval |x|<1.93 (see Figure 1). This type of reduced density pulse is known as dark soliton (see, for instance [7]).

    Figure 1.  Standing wave profile for parameters P+=1,J+=0.8,γ=3/2,k=2,μ=0.1. The solid line is the density profile, while the dashed line corresponds to sonic states. In the region where P(x) is below the dashed line the flow is supersonic.

    Let us denote by (ρ,m) the deviation from (P,J). We obtain the full linearized operator around the profile

    L[ρm]=[m((J2P2γPγ1)ρ)(2JPm)+μm+LVρ], (2.8)

    where

    LVρ=k22ρ2k2((P)(ρP)).

    The associated eigenvalue problem reads

    L[ρm]=λ[ρm]. (2.9)

    Since the profile is a pulse, the essential spectrum of (2.8) is obtained by considering the asymptotic operator at the end state

    L[ρm]=[mα+ρ+β+m+μm+k22ρ],

    where

    α+=(J+)2(P+)2γ(P+)γ1, β+=2J+P+.

    The associated eigenvalue problem is

    L[ρm]=λ[ρm].

    Let us rewrite it as a first order system

    V=M+V,

    where V=[ρ,m,u1,u2], with ρ=u1, u1=u2, and the limit matrix M+ is given by

    M+=[0010λ00000012β+λk22λk22k2(μλα+)0]. (2.10)

    The characteristic equation of M+ is det(νIdM+)=0. It reads

    ν4+2k2(α+λμ)ν22β+k2λν+2λ2k2=0. (2.11)

    Setting ν=iξ, ξR in (2.11) and dividing by 2/k2 we obtain the dispersion relation

    λ2+ξ(μξiβ+)λξ2(α+k2ξ22)=0. (2.12)

    Since the profile (P,J) is a pulse, the essential spectrum is given by the union of the two curves Σj={λ=λj(ξ)C:ξR}, j=1,2, where λj(ξ) are determined by the roots of (2.12). Let us now define stability of essential spectrum.

    Definition 2.1. The essential spectrum of L is stable, if it is contained in the closed left half-plane, that is

    σess(L){λC:Reλ0}.

    The essential spectum of L is unstable, if it intersects the unstable half-plane:

    σess(L){λC:Reλ>0}.

    Lemma 3 in [16] and Lemma 3.2 in [21] imply the following description of the stability of the essential spectrum.

    Theorem 2.2 (Stability of essential spectrum).

    (ⅰ) If |U+|cs(P+), then the essential spectrum of L is stable. Moreover, it holds that Reλ1,2<0, provided ξ0.

    (ⅱ) If |U+|>cs(P+), then the essential spectum of L is unstable.

    That is, the stability of the essential spectrum is related to the end state (P+,J+) being subsonic or sonic. Moreover, the existence condition in Theorem 1.1 is also related to stability of the essential spectrum. Indeeed, a standing wave profile exists if and only if the end state is subsonic with nonzero velocity. Hence, the essential spectrum of standing waves is always stable. Furthermore, Lemma 3 in [16] implies that we have consistent splitting, that is the limit matrix M+ has 2 eigenvalues with positive real parts and 2 eigenvalues with negative real parts provided λ is to the right of the curves Σj.

    Following [16, Section 4.2.1], we shall recast the eigenvalue problem (2.9) in terms of integrated variables

    ˆρ(x)=xρ(y)dy, ˆm(x)=xm(y)dy.

    This transformation removes the zero eigenvalue without further modifications of the spectrum. Expressing ρ and m in terms of ˆρ and ˆm and integrating from to x we get

    λˆρ=ˆm, (2.13)
    λˆm=f1ˆρ+f2ˆm+k22ˆρ2k2(P)(ˆρP), (2.14)

    where

    f1(x)=J(x)2P(x)2γP(x)γ1, f2(x)=2J(x)P(x).

    Let us rewrite the system (2.13)-(2.14) as ˆV=ˆM(x,λ)ˆV, where ˆV=[ˆρ,ˆm,ˆu1,ˆu2], ˆρ=ˆu1, ˆu1=ˆu2, and

    ˆM(x,λ)=[0010λ00000012λf2k22λk22λμk22f1k2P2P22PP]. (2.15)

    The limiting matrix of ˆM(x,λ) as x± is (2.10).

    To define the Evans function, let us consider the equation Y=ˆM(x,λ)Y with ˆM(x,λ) defined in (2.15). Since the profile (P,J) is a pulse, the limits of ˆM(x,λ) at ± coincide and are equal to the matrix M+ given by (2.10). We assume that M+ is hyperbolic. This is always true for λ to the right of the curves Σj. Denote by ν1,ν2 and by ν+1,ν+2 the eigenvalues of M+ with positive and negative real parts, respectively, and indicate with v±i the corresponding (normalized) eigenvectors. Let Yi be a solution of Y=ˆM(x,λ)Y, satisfying exp(νix)Y(x) tends to vi as x and exp(ν+ix)Y+(x) tends to v+i as x+. Then, the Evans function can be defined by

    E(λ)=det(Y1(0),Y2(0),Y+1(0),Y+2(0)).

    Consequently, λ is in the point spectrum of L if and only if E(λ)=0.

    To numerically compute the Evans function, we use the compound matrix method (see [14,17,18]). This method is used in order to get a stable numerical procedure, in spite of the fact that the system Y=ˆM(x,λ)Y is numerically stiff. Specifically, the compound matrix B(x,λ) is given by:

    B=[m11+m22m23m24m13m140m32m11+m33m34m120m14m42m43m11+m440m12m13m31m210m22+m33m34m24m410m21m43m22+m44m230m41m31m42m32m33+m44],

    where mjk are the entries of ˆM(x,λ) defined in (2.15). We integrate the equation ϕ=(B(x,λ)μ)ϕ numerically on a sufficiently large interval [L1,0], where μ is the unstable eigenvalue of B at with maximal (positive) real part. Denote the profile (P(x),J(x)) by ζ(x). Given a numerical approximation (ζk)Nk=1 of ζ(x) at points (xk)Nk=1 with L1=x1<x2<...<xN=L1, let ˜ζ(x) be the piecewise linear interpolant of (x1,ζ1),...,(xN,ζN). We obtain the matrix B(x,λ) using ˜ζ(x). Similarly we integrate the equation ϕ=(B(x,λ)μ+)ϕ on [0,L1] backwards, where this time μ+ is the stable eigenvalue of B at + with minimal (negative) real part. Then, the coefficients μ± compensate for the growth/decay at infinity. Finally, the Evans function can be constructed by means of linear combination of the components of the two solutions ϕ±=(ϕ±1,,ϕ±6) as follows:

    E(λ)=ϕ1ϕ+6ϕ2ϕ+5+ϕ3ϕ+4+ϕ4ϕ+3ϕ5ϕ+2+ϕ6ϕ+1|x=0.

    To compute the initial conditions we integrate the reduced Kato ODE

    dr±dλ=dP±dλr±, (2.16)

    where P± denotes the spectral projection of ˆB±=limx+ˆB(x,λ) corresponding to μ±. This choice of initial condition provides an analytic Evans function E(λ). To numerically integrate (2.16) we use the algorithm from [25], that is |r1±|=1 eigenvector as before (referring to maximal/minimal decay/growth rate of ˆB±) and for k>0,

    rk+1±=Pk±rk±.

    Then, provided E(λ) does not vanish on a closed contour Γ which does not intersect the essential spectrum of L, we can use the winding number

    12πiΓE(z)E(z)dz

    to count the number of zeros (taking into account multiplicity) inside the contour (see [23]).

    For our calculations we use the set of parameters (2.7). Recall that the associated profile is depicted in Figure 1. We compute the Evans function on a semi-circular contour with radius 20, center at λ=0 and vertical segment on the imaginary axis. Furthermore, we do not evaluate the Evans function at 0, but evaluate it up to ±i106. Along the contour we integrate the Kato ODE using 5104 points. Then, we compute E(λ) with the solver ode45 in Matlab with relative tolerance 106 and we set L1=40. Finally, we apply the symmetry of

    E(¯λ)=¯E(λ). (2.17)

    The Evans function E(λ) is plotted in Figure 2. Contrary to the case of traveling waves studied in [17,18], the winding number of the Evans function is (approximately) 1. This is a numerical evidence for the presence of an unstable real simple eigenvalue in the interval (0,20). Indeed, if ν would be a root of E(λ)=0 with nonzero imaginary part, then by (2.17) ¯ν would also be a root, and hence the winding number would be greater than 1. Therefore, there exists a real root with multiplicity 1.

    Figure 2.  The image of a semi-circular contour with radius 20 through the Evans function E(λ) for the QHD system with linear viscosity. The origin is marked in red.

    Let us compute E(λ) on the real line. Equation (2.17) implies that E(λ)R for λR. We initialize the Kato ODE at λ=20 and integrate it along the real line until λ=106 using 2104 points. The result is plotted in Figure 3. The Evans function has a real root λ00.0496.

    Figure 3.  The Evans function E(λ) on the real line for the QHD system with linear viscosity. It intersects the horizontal axis at λ00.0496.

    The instability possibly is related to the presence of a supersonic region along the profile as will be discussed in Section 4.

    In this section we consider the following quantum hydrodynamics system with nolinear viscosity

    {ρt+mx=0,mt+(m2ρ+p(ρ))x=μρ(mxρ)x+k2ρ((ρ)xxρ)x. (3.1)

    Let us rewrite system (3.1) in terms of (ρ,u) variables as

    {ρt+(ρu)x=0,ut+(u2)x2+(h(ρ))x=μ((ρu)xρ)x+k2((ρxx)ρ)x, (3.2)

    where the enthalpy h(ρ) is

    h(ρ)={lnρ,γ=1,γγ1ργ1,γ>1,

    see [10]. We are searching for standing wave profiles of the form

    ρ(t,x)=R(x), u(t,x)=U(x), (3.3)

    such that

    limx±R(x)=R+, limx±U(x)=U+.

    Substituting (3.3) into (3.2) we obtain

    (RU)=0, (3.4)
    12(U2)+h(R)=μ((RU)R)+k2((R)R). (3.5)

    Integrating Eq (3.4) up to ± we get

    U=AR, (3.6)

    where A=R+U+. Substituting (3.6) into (3.5) and integrating we obtain

    R=2k2f(R)+(R)22R, (3.7)

    where

    f(R)=Rh(R)+A22RRB,

    with

    B=12(U+)2+h(R+).

    Theorem 6.1 in [24] provides a characterization for the existence of homoclinic loops to (3.7). These homoclinic loops correspond to standing wave profiles for (3.2).

    We use the method from Section 2.2, that is we compute the profile numerically using a boundary value solver. As an initial guess we use a function of the form (2.5) with c=0.5,

    R0(x)=R+0.5exp(x2).

    We compute the profile for parameters

    R+=1, U+=0.9, γ=3/2, k=2, μ=0.08. (3.8)

    Since |U+|=0.9<1.22cs(R+), the condition for existence of profile of Theorem 6.1 in [24] is verified; see Figure 4. The flow along the profile is supersonic in the region where the following inequality holds:

    R(x)<((R+U+)2γ)1γ+1. (3.9)
    Figure 4.  Standing wave profile for parameters R+=1, U+=0.9, γ=3/2, k=2, μ=0.08. The solid line corresponds to R(x). In the region where R(x) is below the dashed line the flow is supersonic.

    For the parameter values (3.8) the right hand side of (3.9) is 0.78 and the flow is supersonic for |x|<2.29.

    Here we perform a linearization of (3.2) around a standing wave profile. Denoting by (ρ,u) the deviation from the profile (R,U), we obtain the following full linearized operator:

    L[ρu]:=[(Ru+Uρ)(Uu)(dhdR(R)ρ)+μ((R1(Ru+Uρ))(R2(RU)ρ))+k2LQρ],

    where

    LQρ=12(R1/2(R1/2ρ))12(R3/2(R1/2)ρ).

    The associated eigenvalue problem reads

    λ[ρu]=L[ρu]. (3.10)

    The asymptotic operator at is given by

    L[ρu]:=[U+ρR+uU+udhdR(R+)ρ+μ(u+U+R+ρ)+k22ρR].

    The eigenvalue problem associated to L is

    λ[ρu]=L[ρu].

    Let us rewrite it as a first-order system

    V=M+V,

    where

    M+:=[0010λR+0U+R+000012U+λk22R+λk22k2(R+dhdR(R+)(U+)2+μλ)0].

    The characteristic equation det(νIdM+)=0 is

    ν4+2k2((U+)2R+dhdR(R+)λμ)ν2+4U+k2λν+2λ2k2=0. (3.11)

    Setting ν=iξ, ξR, in (3.11) and dividing by 2/k2, we obtain the dispersion relation:

    λ2+(μξ2+2U+ξi)λ+(R+dhdR(R+)(U+)2)ξ2+k22ξ4=0.

    The stability condition for the essential spectrum is identical to Theorem 2.2. Therefore, standing wave profiles for the system (3.2) always have stable essential spectrum.

    Let us now consider the integrated variables

    ˆρ(x)=xρ(y)dy, ˆu(x)=xu(y)dy.

    Expressing ρ and u in terms of ˆρ and ˆu in (3.10) and integrating from to x we obtain the system in integrated variables:

    λˆρ=UˆρRˆu, (3.12)
    λˆu=f(x)ρUu+μ(R1(Rˆu+Uˆρ)R2(RU)ˆρ)+k22(R12(R12ˆρ)R32(R12)ˆρ), (3.13)

    with

    f(x)=dhdR(R(x)).

    Let us rewrite (3.12)-(3.13) as a first-order system

    ˆV=ˆM(x,λ)ˆV,

    where ˆV=[ˆρ,ˆu,ˆu1,ˆu2], with ˆρ=ˆu1, and ˆu1=ˆu2, and

    ˆM(x,λ):=[0010λR0UR00001ˆm4,1ˆm4,2ˆm4,3ˆm4,4], (3.14)

    with

    ˆm4,1=2λUk2,ˆm4,2=2λRk2,ˆm4,3=2k2(RfU2+μ((RU)R+λ))+RR(R)2R2,ˆm4,4=RR.

    In this section we compute the Evans function for the QHD system with nonlinear viscosity (3.2). We use the method from Section 2.6. We integrate the ODE ϕ=(B(x,λ)μ)ϕ on [L1,0] and ϕ=(B(x,λ)μ+)ϕ on [0,L1] backwards, where L1=40 as before, however here B(x,λ) is the compound matrix of ˆM(x,λ) defined in (3.14) and μ± is the stable (unstable) eigenvalue of B at + with minimal (maximal) real part.

    We evaluate the Evans function on the contour surrounding a semi-annular region with radii 20 and 106 with center at the origin and vertical segment on the imaginary axis. We integrate the reduced Kato ODE using 106 points, then we compute E(λ) with the solver ode15s in Matlab with relative tolerances 108 and 1010, the latter being used if E(λ) is evaluated wtih λ close to 0. The Evans function E(λ) is ploted in Figure 5, and in Figures 6 and 7 with enlargement of a region around the origin. Notice that the Evans function E(λ) is small in absolute value for λ0 along the contour, hence a higher accuracy is required for its numerical computation than the one used for the QHD system with linear viscosity (cf. Section 2.6). The winding number of the Evans function is (approximately) 1. As before, this is numerical evidence for the presence of a simple real eigenvalue in the interval (0,20).

    Figure 5.  The image of a semi-annular region with radii 20 and 106 through the Evans function E(λ) for the QHD system with nonlinear viscosity. The origin is marked in red.
    Figure 6.  Same as Figure 5 with enlargement of a region around the origin.
    Figure 7.  Same as Figures 5 and 6 with further enlargement of a region around the origin.

    Now, let us evaluate E(λ) on the real line. We initialize the Kato ODE at λ=20 and integrate it along the real line until λ=106 using 4105 points. The result is depicted in Figure 8. The Evans function E(λ) has a real root λ00.0026. Notice that both the viscosity coefficient and the unstable eigenvalue are smaller in the nonlinear viscosity case, compared to their counterparts in case with the linear viscosity (cf. the parameter set (2.7)), even though increasing the viscosity tends to stabilize the system. Therefore, our numerics suggests that the nonlinear viscosity has stabilizing effect, corroborating the analytical result in [9, Remark 4.7].

    Figure 8.  The Evans function E(λ) computed on a segment of the real line for the QHD system with nonlinear viscosity. It has a zero at λ00.0026.

    It was shown numerically in Sections 2.6 and 3.5 that the standing wave profiles considered for the QHD systems (1.1) and (3.2) with linear and nonlinear viscosity are spectrally unstable. The essential spectrum is stable, however there is a simple real unstable eigenvalue. This numerical result is in contrast with the numerical stability results for traveling wave profiles in [17,18] where it was shown that non-monotone traveling wave profiles are spectrally stable. We notice that in both cases there exists a supersonic region, where the velocity along the profile is larger than the speed of sound (see Figures 1 and 4). On the other hand, one can show numerically that the velocity along the profiles is subsonic everywhere for the cases considered in [17,18]. Therefore, we make the following:

    Conjecture 4.1. A standing or traveling wave profile for the QHD systems with linear or nonlinear viscosity (1.1) and (3.2) is spectrally stable if and only if the velocity is subsonic or sonic along the profile.

    In other words, the conjecture states that a profile (R,U) is spectrally stable if and only if

    |U(x)|cs(R(x)), xR.

    If this inequality holds, then taking limits as x± we obtain |U±|cs(R±), that is the end states are subsonic or sonic, in accordance with the necessary and sufficient condition in Theorem 2.2 for stability of the essential spectrum. Let us now compare the numerical results for the QHD systems with linear and nonlinear viscosity. The unstable eigenvalue is larger for the former system even though the viscosity coefficient is smaller for the latter (cf. the parameter sets (2.7) and (3.8)). Therefore, our numerical results suggest that the nonlinear viscosity has stabilizing effect.

    The authors declare they have not used Artificial Intelligence (AI) tools in the creation of this article.

    This research work was conducted in Gran Sasso Science Institute, University of L'Aquila and Universidad Nacional Autónoma de México.

    The author declares no conflict of interest.



    [1] P. Antonelli, P. Marcati, On the finite energy weak solutions to a system in quantum fluid dynamics, Commun. Math. Phys., 287 (2009), 657–686. https://doi.org/10.1007/s00220-008-0632-0 doi: 10.1007/s00220-008-0632-0
    [2] P. Antonelli, P. Marcati, H. Zheng, Genuine hydrodynamic analysis to the 1-D QHD system: existence, dispersion and stability, Commun. Math. Phys., 383 (2021), 2113–2161. https://doi.org/10.1007/s00220-021-03998-z doi: 10.1007/s00220-021-03998-z
    [3] P. Antonelli, P. Marcati, H. Zheng, An intrinsically hydrodynamic approach to multidimensional QHD systems, Arch. Ration. Mech. Anal., 247 (2023), 24. https://doi.org/10.1007/s00205-023-01856-x doi: 10.1007/s00205-023-01856-x
    [4] D. Bohm, A suggested interpretation of the quantum theory in terms of "hidden" variables. I, Phys. Rev., 85 (1952), 166–179. https://doi.org/10.1103/PhysRev.85.166 doi: 10.1103/PhysRev.85.166
    [5] D. Bohm, A suggested interpretation of the quantum theory in terms of "hidden" variables. II, Phys. Rev., 85 (1952), 180–193. https://doi.org/10.1103/PhysRev.85.180 doi: 10.1103/PhysRev.85.180
    [6] D. Bohm, B. J. Hiley, P. N. Kaloyerou, An ontological basis for the quantum theory, Phys. Rep., 144 (1987), 321–375. https://doi.org/10.1016/0370-1573(87)90024-X doi: 10.1016/0370-1573(87)90024-X
    [7] S. Burger, K. Bongs, S. Dettmer, W. Ertmer, K. Sengstock, A. Sanpera, et al., Dark solitons in Bose-Einstein condensates, Phys. Rev. Lett., 83 (1999), 5198–5201. https://doi.org/10.1103/PhysRevLett.83.5198 doi: 10.1103/PhysRevLett.83.5198
    [8] R. Folino, R. G. Plaza, D. Zhelyazov, Spectral stability of small-amplitude dispersive shocks in quantum hydrodynamics with viscosity, Commun. Pure Appl. Anal., 21 (2022), 4019–4040. https://doi.org/10.3934/cpaa.2022133 doi: 10.3934/cpaa.2022133
    [9] R. Folino, R. G. Plaza, D. Zhelyazov, Spectral stability of weak dispersive shock profiles for quantum hydrodynamics with nonlinear viscosity, J. Differ. Equations, 359 (2023), 330–364. https://doi.org/10.1016/j.jde.2023.02.038 doi: 10.1016/j.jde.2023.02.038
    [10] I. Gasser, Traveling wave solutions for a quantum hydrodynamic model, Appl. Math. Lett., 14 (2001), 279–283. https://doi.org/10.1016/S0893-9659(00)00149-X doi: 10.1016/S0893-9659(00)00149-X
    [11] A. V. Gurevich, A. P. Meshcherkin, Expanding self-similar discontinuities and shock waves in dispersive hydrodynamics, Sov. Phys. JETP, 60 (1984), 732–740.
    [12] A. V. Gurevich, L. P. Pitaevskii, Nonstationary structure of a collisionless shock wave, Sov. Phys. JETP, 38 (1974), 291–297.
    [13] M. A. Hoefer, M. J. Ablowitz, I. Coddington, E. A. Cornell, P. Engels, V. Schweikhard, Dispersive and classical shock waves in Bose-Einstein condensates and gas dynamics, Phys. Rev. A, 74 (2006), 023623. https://doi.org/10.1103/PhysRevA.74.023623 doi: 10.1103/PhysRevA.74.023623
    [14] J. Humpherys, On the shock wave spectrum for isentropic gas dynamics with capillarity, J. Differ. Equations, 246 (2009), 2938–2957. https://doi.org/10.1016/j.jde.2008.07.028 doi: 10.1016/j.jde.2008.07.028
    [15] I. M. Khalatnikov, An introduction to the theory of superfluidity, CRC Press, 2000. https://doi.org/10.1201/9780429502897
    [16] C. Lattanzio, P. Marcati, D. Zhelyazov, Dispersive shocks in quantum hydrodynamics with viscosity, Phys. D, 402 (2020), 132222. https://doi.org/10.1016/j.physd.2019.132222 doi: 10.1016/j.physd.2019.132222
    [17] C. Lattanzio, P. Marcati, D. Zhelyazov, Numerical investigations of dispersive shocks and spectral analysis for linearized quantum hydrodynamics, Appl. Math. Comput., 385 (2020), 125450. https://doi.org/10.1016/j.amc.2020.125450 doi: 10.1016/j.amc.2020.125450
    [18] C. Lattanzio, D. Zhelyazov, Spectral analysis of dispersive shocks for quantum hydrodynamics with nonlinear viscosity, Math. Mod. Meth. Appl. Sci., 31 (2021), 1719–1747. https://doi.org/10.1142/S0218202521500378 doi: 10.1142/S0218202521500378
    [19] C. Lattanzio, D. Zhelyazov, Traveling waves for quantum hydrodynamics with nonlinear viscosity, J. Math. Anal. Appl., 493 (2021), 124503. https://doi.org/10.1016/j.jmaa.2020.124503 doi: 10.1016/j.jmaa.2020.124503
    [20] S. Novikov, S. V. Manakov, L. P. Pitaevskii, V. E. Zakharov, Theory of solitons: the inverse scattering method, New York: Springer, 1984.
    [21] R. G. Plaza, D. Zhelyazov, Well-posedness and decay structure of a quantum hydrodynamics system with Bohm potential and linear viscosity, preprint, arXiv, 2023. https://doi.org/10.48550/arXiv.2309.00175
    [22] S. R. Z. Sagdeev, Kollektivnye protsessy i udarnye volny v razrezhennol plazme (Collective processes and shock waves in a tenuous plasma), Voprosy teorii plazmy (Problems of plasma theory), Vol. 5, Atomizdat, 1964.
    [23] B. Sandstede, Chapter 18–Stability of travelling waves, Handbook of dynamical systems, Elsevier, 2 (2002), 983–1055. https://doi.org/10.1016/S1874-575X(02)80039-X
    [24] D. Zhelyazov, Existence of standing and traveling waves in quantum hydrodynamics with viscosity, Z. Anal. Anwend., 42 (2023), 65–89. https://doi.org/10.4171/zaa/1723 doi: 10.4171/zaa/1723
    [25] K. Zumbrun, A local greedy algorithm and higher order extensions for global numerical continuation of analytically varying subspaces, Quart. Appl. Math., 68 (2010), 557–561. https://doi.org/10.1090/S0033-569X-2010-01209-1 doi: 10.1090/S0033-569X-2010-01209-1
  • This article has been cited by:

    1. Ramón G. Plaza, Delyan Zhelyazov, Well-posedness and decay structure of a quantum hydrodynamics system with Bohm potential and linear viscosity, 2024, 65, 0022-2488, 10.1063/5.0172774
  • Reader Comments
  • © 2024 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(1205) PDF downloads(150) Cited by(1)

Figures and Tables

Figures(8)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog