Zum Hauptinhalt springen
Artikel Öffentlich zugänglich

Invariants of the Axisymmetric Plasma Flows

  • EMAIL logo
Veröffentlicht/Copyright: 27. April 2018

Abstract

Infinite families of new functionally independent invariants are derived for the axisymmetric dynamics of viscous plasmas with zero electrical resistance. As a consequence, we find that, if two axisymmetric plasma states are dynamically connected, then their total number of magnetic rings must be equal (the same as for the total numbers of magnetic blobs) and the corresponding infinitely many new invariants must coincide.

1 Introduction

As is known [1], [2], the magnetohydrodynamic (MHD) equations with isotropic viscosity and zero electrical resistance have the form

(1)Vt+(V)V=1ρp+1ρμ(×B)×B+νΔV,
(2)Bt=×(V×B),
(3)V=0,  B=0,  ρt+Vρ=0.

For incompressible plasma flows with velocity V(x, t), (2) implies that the magnetic field B(x, t) is transported in time with the plasma flow (or “is frozen in the flow”). Here, ρ(x, t) is the plasma density, which according to the last equation of (3) is preserved along the plasma streaklines, μ is the magnetic permeability, p(x, t) is the pressure, ν is the kinematic viscosity, ▽ is the nabla operator, and Δ is the Laplace operator. We assume that during some finite time t, the vector functions V(x, t), B(x, t), and the scalar functions ρ(x, t), p(x, t) are of class C2. The special regimes of plasma relaxation are derived in the explicit form in [3]. Symmetry transforms for the ideal MHD equilibria are derived in [4]. Exact axisymmetric plasma equilibria are presented in [5] and the helically symmetric ones in [6].

The concrete form νΔV of the dissipative term in (1) can be altered for a more sophisticated formula, for example, for the one recently introduced in [7]. However, this would not have changed our results because they are based entirely on the MHD equations (2) and (3). Therefore, we use in (1) the classical viscosity term νΔV as in the Navier-Stokes equations.

As is known [1], [2], the frozenness of magnetic field B into the plasma flow leads to such invariants as the integer-valued Gauss linking number for any two closed magnetic field lines [8] and the discrete topological invariants of magnetic field knots [9].

Woltjer’s integral HD [10] of the magnetic helicity has the form

(4)HD=DABd3x,

where A is the vector potential of the magnetic field B=∇×A.

Woltjer’s magnetic helicity HD=∫DA·[∇×A]d3x [10], [11] has an analogue, which is the hydrodynamic helicity ℋD=∫DV·[∇×V]d3x. The well-known distinction between the two helicities is that the magnetic helicity HD (4) is not uniquely defined by the magnetic field B because HD depends on the gauge transformations of the vector potential A. In contradistinction to this, the hydrodynamic helicity ℋD is uniquely defined by the velocity field V because there are no gauge transformations of the velocity V.

The hydrodynamic helicity ℋD satisfies the equation

(5)dDdt=S[F(ρ)+12|V|2][×V]nds,

where the surface S=∂D is the boundary of the domain D⊂ℝ3, vector field n(x) is the outward unit vector field orthogonal to S, and ds is the area element of surface S. Equation (5) was first derived by Moffatt in [8], where it was assumed that the pressure p is a function of the density ρ only (the barotropic condition p=p(ρ)) and ∇F(ρ)=ρ−1p. In ([12], p. 108) Serre has shown that, in general (if [∇×Vn≠0), the helicity ℋD inside an invariant domain D (with the non-penetration condition V·n=0 on S) is not constant in time.

Helicity was studied by Kudryavtseva [13], [14] and by Enciso et al. [15]. The authors stated that for divergence-free vector fields the integral invariants (i.e. the integrals of certain densities “over compact three-dimensional manifold M without boundary” ([15], p. 2035) are often related to helicities.

We derive in this paper new integral invariants of the axisymmetric plasma flows with zero electrical resistance, which depend on the magnetic field B and in some cases on the plasma density ρ but do not depend on any gauge transformations of the vector potential A. Our key idea is to study the safety factor q [2], [16], [17] for the time-dependent axisymmetric magnetic fields B and to prove that the emerging function q(x, t) is conserved along the plasma streaklines.

In Section 2 we define the magnetic rings and magnetic blobs for the axisymmetric dynamics of plasma with zero electrical resistance and prove that the former are frozen into the plasma flows. In Section 3 we calculate the Woltjer’s helicity HD (4) for the axisymmetric magnetic rings and blobs and demonstrate how it is altered by the axisymmetric gauge transformations of the vector potential A. In Section 5 we prove that the safety factor q(x, t) (which is uniquely defined by the magnetic field B in the domains where all magnetic surfaces are tori or magnetic axes) is conserved along the plasma streaklines. In Sections 6 and 7 we prove that the axisymmetric plasma flows preserve infinitely many new invariants, which are either functional invariants or the integrals of the arbitrary C2-functions of the safety factor q(x, t) over the magnetic rings and magnetic blobs, for example, qn(x, t), n=1, 2, …. The functional invariants are the values of certain axisymmetric functions on the magnetic axes of the axisymmetric field B(x, t). In Section 7 we demonstrate that infinitely many new integral invariants with possibly one exception are functionally independent of the magnetic helicity.

2 Magnetic Rings and Blobs

Let A(r, z, t) be a z-axisymmetric vector potential

(6)A(r,z,t)=u(r,z,t)e^r+v(r,z,t)e^z+r1ψ(r,z,t)e^φ,

where êr, êz, êϕ are vectors of unit length along the axes of the cylindrical coordinates r≥0, z, φ. The z-axisymmetric gauge transformation of the vector potential A is

(7)A˜(r,z,t)=(u+f/r)e^r+(v+f/z)e^z+r1(ψ+χ(t))e^φ,

where u, v, f, ψ are C2-functions of r, z, and t, and χ(t) is a C1-function. The gauge transformation (7) is well defined in any (non-simply-connected) toroidal domain T3=D2×𝕊1 that lies in the space r>0. Here, D2 is a compact domain in the plane (r>0, z) and 𝕊1 is the circle corresponding to the anglular variable φ. The gauge transformation of function ψ(r, z, t) is

(8)ψ˜(r,z,t)=ψ(r,z,t)+χ(t).

The magnetic field B=∇×A has the form

(9)B(r,z,t)=r1ψ(r,z,t)/ze^r+r1ψ(r,z,t)/re^z+w(r,z,t)e^φ,

and is evidently gauge-invariant:

(10)B˜(r,z,t)=B(r,z,t),w˜(r,z,t)=w(r,z,t)=u/zv/r.

The gauge transformation (8) implies that the flux function ψ(r, z, t) is not uniquely defined by the magnetic field B(r, z, t) (9).

Equation (9) yields that the derivative of the flux function ψ(r, z, t) along the plasma streaklines is

(11)DψDt=ψt+ψrVr+ψzVz=ψt+r(BzVrBrVz).

Here, the initial poloidal components Br, Bz of the magnetic field B(r, z, t) and Vr, Vz for the plasma velocity V(r, z, t) can be choosen independently. For unsteady solutions, (11) yields that the generic flux function ψ˜(r,z,t) is not conserved along the plasma streaklines. Indeed, assuming that a special flux function ψ(r,z,t) is conserved (/Dt=0), we get that all functions ψ˜(r,z,t) (8) with dχ(t)/dt≠0 are not conserved because

(12)Dψ˜Dt=DψDt+dχ(t)dt0.

For any constant time t, the magnetic field B(r, z, t) (9) is tangent to the z-axisymmetric magnetic surfaces ψ(r, z, t)=C(t). The family of magnetic surfaces for the arbitrary function C(t) is invariant under the gauge transformations (8).

Dynamics along magnetic field lines for t=const is defined by the system

dxdτ=dxdτe^x+dydτe^y+dzdτe^z=drdτe^r+rdφdτe^φ+dzdτe^z=B(r,z,t),

which by virtue of (9) has the form

(13)drdτ=r1ψ(r,z,t)z,  dzdτ=r1ψ(r,z,t)r,
(14)dφdτ=r1w(r,z,t).

The system (13), (14) is evidently invariant under the gauge transformations (8), (10). After the time change dτ1/dτ=r−1, the system (13) for t=const becomes a τ1-autonomous Hamiltonian system.

Hamiltonian systems were applied to the “plasma-confining magnetic fields” by Kerst [18], see also pages 1062–1063 of review [19]. We derive the Hamiltonian system (13) in the poloidal coordinates (r, z) and not in the Cartesian ones (x, y) as in [18], [19]. The system (13) is derived above straight from the definition B=∇×A and without any assumption of “a uniform Bz” as in ([18], p. 253).

Boozer introduced in [20] for the Hamiltonian treatment of the plasma equilibria “generalised magnetic coordinates”, see also the review ([21], p. 1076). We do not use them because the standard cylindrical ones (r, z, φ) are sufficient for our study.

We will use the following classical properties of any autonomous Hamiltonian system in ℝ2 [22]:

  • (α)Its Hamiltonian function is a first integral (for system (13) it is the flux function ψ(r, z, t) with a fixed time t);

  • (β)All its non-degenerate equilibria are either saddles s or centers ck;

  • (γ)It preserves the area.

These properties imply that every closed trajectory Cψ(t) (ψ(r, z, t)=const) of system (13) for a fixed time t belongs to one of the two-dimensional sets Dk(t) in the poloidal plane (r, z) satisfying the following conditions:

  • Dk(t) is invariant with respect to the system (13);

  • the set Dk(t) is connected and closed;

  • all trajectories of system (13) in the set Dk(t) have finite Euclidean length and are either closed curves or separatrices of the equilibrium points in Dk(t);

  • a dense open subset of Dk(t) is filled with closed trajectories of system (13);

  • the set Dk(t) is the largest among the sets satisfying conditions 14 and having a non-empty intersection with Dk(t).

We will call each set Dk(t) satisfying conditions 1–5 a maximal set (the term “maximal” refers to the condition 5). Figure 1 shows the phase portrait of a concrete system (13) that has eight maximal sets D1, …, D8.

Figure 1: Poloidal sections of magnetic rings D3, D6, D7, D8 (pink) and magnetic blobs D1, D2, D4, D5 (blue) for the exact plasma equilibrium with the flux function ψ(r, z)=r2[0.005−zR−4((3−R2)R−1sinR−3cosR)], where R=r2+z2.$R = \sqrt {{r^2} + {z^2}} .$ Rotation of the sets D3, D6, D7, D8 around the axis z defines four magnetic rings and rotation of the sets D1, D2, D4, D5 around the axis z defines four magnetic blobs.
Figure 1:

Poloidal sections of magnetic rings D3, D6, D7, D8 (pink) and magnetic blobs D1, D2, D4, D5 (blue) for the exact plasma equilibrium with the flux function ψ(r, z)=r2[0.005−zR−4((3−R2)R−1sinR−3cosR)], where R=r2+z2. Rotation of the sets D3, D6, D7, D8 around the axis z defines four magnetic rings and rotation of the sets D1, D2, D4, D5 around the axis z defines four magnetic blobs.

The boundary ∂Dk(t) is a one-dimensional set 𝒞k(t) that is invariant with respect to the dynamical system (13). Therefore, 𝒞k(t) satisfies the equation ψ(r, z, t)=bk(t). System (13) preserves the poloidal area dS=2πr dr dz.

The closed trajectories Cψ(t) of system (13) (for t=const) after rotation around the axis of symmetry z define the two-dimensional tori Tψ(t)2=Cψ(t)×S13, which are invariant with respect to the system (13), (14) and therefore are called B(r, z, t)-invariant. Here, the circle 𝕊1 corresponds to the angular variable φ. The rotation of a maximal set Dk(t) around the z-axis defines a three-dimensional set Dk3(t)=Dk(t)×S13, which is closed and B(r, z, t)-invariant. Its 3D volume is

(15)VolDk3(t)=Dk3(t)rdrdzdφ=2πDk(t)rdrdz=ArmDk(t).

We call ArmDk(t) the poloidal area of Dk(t); it is equal to the 3D volume VolDk3(t). Rotation of the boundary curve 𝒞k(t)=∂Dk(t) around the axis z defines the boundary Dk3(t)=Ck(t)×S1.

2.1 Magnetic Rings and Blobs

If 𝒞k(t)=∂Dk(t) is a closed curve lying in the domain r>0, then the closed set Dk3(t)=Dk(t)×S1 is a maximal B(r, z, t)-invariant ring k3(t), which we call a magnetic ring. It is bounded by the torus Tk2(t)=Ck(t)×S1. Otherwise, a closed and bounded set Dk3(t) is a blob k3(t) (which we call a magnetic blob) containing a segment Sk(t) of the axis of symmetry z (r=0).

The gauge transformations (7), (8) are well defined in the magnetic rings k3(t) since they lie in the domain r>0 and are not defined in the magnetic blobs k3(t) because they contain the segments Sk(t) of the axis r=0.

Example 1:Figure 1 shows the phase portrait of the dynamical system (13) for the exact plasma equilibrium which has the flux function ψ(r, z)=r2[0.005−zR−4((3−R2) R−1sinR−3cos R)] and w(r, z)=ψ(r, z), where R=r2+z2. The plot in Figure 1 contains eight maximal sets Dk, which are densely filled with smooth, closed trajectories Cψ (ψ(r, z)=const) of system (13) encircling the centre equilibrium points ck, k=1, …, 8. The maximal sets D1, D2, D4, and D5 are bounded, respectively, by the pairs of separatrices of the saddle equilibria (a7, a8), (a3, a4), (a5, a6), and (a1, a2), correspondingly. The approximate values of the coordinate z for the points ak are indicated in Figure 1. Rotation of the sets D1, D2, D4, D5 around the axis of symmetry z defines four magnetic blobs k3. The maximal sets D3, D6, D7, D8 are bounded by the loop separatrices L1, L2, L3, L4 of the saddle equilibria s1, s2, s3, s4. (By definition, a loop separatrix of a dynamical system begins and ends at the same saddle equilibrium point.) Rotation of the sets D3, D6, D7, D8 around the axis z defines four magnetic rings j3 which are bounded by the tori Tj2=Lj×S1,j=1, 2, 3, 4.

3 Helicity in the Magnetic Rings and Blobs

Using formulas (6) and (9) and dv=rdrdzdφ, we find the following for Woltjer’s helicity (4) in the magnetic ring or blob Dk3(t):

(16)Hk(t)=Dk3(t)ABdv=2πDk(t)[ψrvψzu+ψ(uzvr)]drdz=2πDk(t)[(ψv)r(ψu)z+2ψw]drdz.

Using Green’s theorem and the equation ψ(r, z, t)=bk(t) on the boundary curve 𝒞k(t)=∂Dk(t), we get

Dk(t)[(ψv)r(ψu)z]drdz=Ck(t)ψ(vnrunz)ds=bk(t)Ck(t)(vnrunz)ds,

where ds is the element of the arc length of the curve 𝒞k(t), and the vector (nr, nz) is the outward unit normal vector to it. Applying Green’s theorem again, we find

bk(t)Ck(t)(vnrunz)ds=bk(t)Dk(t)(vruz)drdz=bk(t)Dk(t)wdrdz.

Hence we get

(17)Dk(t)[(ψv)r(ψu)z]drdz=bk(t)Dk(t)wdrdz.

Substituting formula (17) into (16), we derive for the magnetic helicity

(18)Hk(t)=Dk3(t)ABdv=2πDk(t)[2ψ(r,z,t)bk(t)]w(r,z,t)drdz.

As shown above, the gauge transformations (7), (8) are well defined in the magnetic rings k3(t). Formulas (8) and (10) imply b˜k(t)=bk(t)+χ(t),w˜(r,z,t)=w(r,z,t). Therefore, (18) yields for the magnetic helicity H˜k(t) after the gauge transformation (8), (10)

(19)H˜k(t)=Hk(t)+2πχ(t)Dk(t)w(r,z,t)drdz.

Formula (19) demonstrates that the magnetic helicity Hk(t) (16) is not invariant under the gauge transformations (8), (10) if Dk(t)w(r,z,t)drdz0. Indeed, the helicity Hk(t) (16) becomes an arbitrary function Fk(t)=H˜k(t) after the gauge transformation (8) with χ(t)=[Fk(t)Hk(t)](2πDk(t)w(r,z,t)drdz)1. The helicity H˜k(t) becomes a constant Ck after the gauge transformation (8) with the gauge function

(20)χk(t)=[CkHk(t)](2πDk(t)w(r,z,t)drdz)1.

4 Dynamics in the Poloidal Coordinates (r, z)

Suppose a z-axisymmetric magnetic field B(r, z, t1) at t=t1 has a closed magnetic field line L. Its dynamics in the poloidal coordinates (r, z) is described by system (13), which defines the corresponding closed trajectory P(L) that satisfies the equation ψ(r, z, t1)=C. The closed trajectory P(L) in the plane (r, z) belongs to a maximal set D(t1)⊂ℝ2, which is densely filled with closed trajectories Cψ(t1) of system (13), see Figure 1.

For the z-axisymmetric plasma flows, system (13) defines the so-called induced diffeomorphisms of the poloidal plane (r, z). The plasma flow by virtue of the equation divV=0 preserves the 3D volume dΩ=rdrdzdφ. For the z-axisymmetric flows, we get that the induced diffeomorphisms of the poloidal plane (r, z) preserve the poloidal area dS=2πrdrdz, the axis of symmetry z (r=0), and the invariant domain r>0. The frozenness of the magnetic field lines into the plasma flow [1] implies for the z-axisymmetric case that the trajectories of system (13) are frozen into the induced dynamics of plasma in the poloidal coordinates (r, z). Therefore, the phase portrait of dynamical system (13) is changed in time t by the induced diffeomorphisms of the poloidal plane (r, z). Let Mt1t be the induced diffeomorphism defined by the plasma dynamics from time t1 to time t>t1. Any closed trajectory Cψ(t1) of system (13) in the invariant domain r>0 is transformed by the diffeomorphism Mt1t into another closed trajectory Cψ(t) of system (13) in the same domain r>0. Since any maximal set Dk(t1) is densely filled with the closed trajectories Cψ(t1), it is transformed by the diffeomorphism Mt1t into the maximal set Dk(t). Using the frozenness of the magnetic field B(r, z, t) into the plasma flow, it is easy to prove that the image Mt1t(Dk(t1)) coincides with the maximal set Dk(t).

Because of the conservation of the area dS=2πrdrdz by system (13), we get the equality of the poloidal areas ArmDk(t1)=ArmDk(t)=2πDk(t)rdrdz. Since the induced diffeomorphisms Mt1t preserve the axis of symmetry z (r=0) and the domain r>0, we get that the total number Nr of magnetic rings k3(t)=Dk(t)×S1 and their 3D volumes (15) are constant in time. The same is true for the total number Ns of the magnetic blobs k3(t) and for their 3D volumes. The frozenness of the magnetic rings and blobs into the plasma flows with zero electrical resistance implies that no collapses or touchings between magnetic rings or blobs can occur during the axisymmetric dynamics of plasma.

5 Safety Factor for the Time-Dependent Axisymmetric Magnetic Fields

In this section, we prove the conservation along the plasma streaklines of the safety factor q for the general helical magnetic field lines on invariant tori 𝕋2 of a non-stationary axisymmetric magnetic field B(x, t). The safety factor q was studied previously in [2], [8], [16], [17], [23].

Let for a z-axisymmetric field B(x, t) a magnetic field line lie on a torus 𝕋2=C×𝕊1 and is closed and goes (say) m times a long way (along the circle 𝕊1) and n times a short way (along the closed curve C). In [17], the safety factor q for such a closed curve is defined as q=m/n. Other definitions of the pitch p=2πq for the vortex lines in the fluid equilibria are given in [8], [24].

We propose the following definition of the safety factor of the non-closed helical magnetic field lines on the time-dependent tori Tψ(t)2=Cψ(t)×S1. Let τ(Cψ(t)) be the period of the corresponding closed trajectory Cψ(t) of the system (13). The generalised safety factor q(r, z, t) of the helices is equal to the increment of the angle φ during one period τ(ψ(t)), divided by 2π:

(21)q(r,z,t)=12πCψ(t)dφdτdτ=12π0τ(Cψ(t))r1(τ)w[r(τ),z(τ),t]dτ.

Formula (21) defines the safety factor for both the non-closed, infinite helical trajectories of system (13), (14) on the torus Tψ(t)2 and for the closed ones. The safety factor q(r, z, t) (21) has the same value for all magnetic field lines on a given torus Tψ(t)2 because the circle integral in (21) does not depend on the starting point. If the number q(r, z, t) is rational (=m/n), then during n periods τ(Cψ(t)) the angle φ increases for n·2πm/n=2πm. Therefore, the corresponding curve on the torus Tψ(t)2=Cψ(t)×S1 is closed and makes m complete turns a long way and n turns a short way. Hence our definition (21) being applied to the closed magnetic field lines reduces to the definition given in [17] only for the closed ones. Such closed curves are called “torus knots Km,n” [9]. As is known, the rational number m/n is a topological invariant of the torus knot Km,n [9].

5.1 Conservation of the Safety Factor q(r, z, t)

Since all magnetic field lines are frozen into the plasma flow, the same is true for the magnetic field knots Km,n. Hence, during the plasma dynamics, the main characteristics of the knots m/n=q(r, z, t) is conserved because it is a topological invariant. Therefore, all rational values of the safety factor q(r, z, t) (21) are preserved along the plasma streaklines.

We have assumed that the vector fields V(x, t), B(x, t) are of class C2. This implies that the function q(r, z, t) (21) is at least continuous. The continuity of q(r, z, t) yields that tori Tψ(t)2 with rational values of q(r, z, t) are everywhere dense in each closed set Dk3(t). Any irrational value of q(r, z, t)=ξ is a limit of certain rational values q(r, z, t)=m/n: ξ=q(r, z, t)=limℓ→∞m/n. Therefore, the conservation during the plasma dynamics of the rational values q(r, z, t)=m/n and the continuiuty of the function q(r, z, t) (21) yield that all its irrational values q(r, z, t)=ξ also are conserved. Hence the safety factor q(r, z, t) and any functions of it F(q(r, z, t)) (e.g. qn(r, z, t) or cos(nq(r, z, t)) for any integer n≥1) are conserved along the plasma streaklines during the plasma dynamics. [1] The same is true for the plasma density ρ(x, t) as a result of the last equation of (3) and for any differentiable function f(q, ρ).

6 Functional Invariants of Plasma Flows

The boundary ∂Dk(t) is a one-dimensional set 𝒞k(t) that is invariant with respect to the dynamical system (13) and hence satisfies the equation ψ(r, z, t)=C(t). Therefore, for any flux function ψ(r, z, t) (t=const) of class C3, there exists in the interior of each maximal set Dk(t) at least one point ck(t)=(rk(t), zk(t)) of local maximum or minimum of ψ(r, z, t). The point ck(t) is an equilibrium point of system (13) and is the limit of the neigbouring invariant, small, closed curves Cψ(t), which are transported by the induced diffeomorphisms of the poloidal plane (r, z). Therefore, the points ck(t) also are trasported with the plasma flow. The same is true for the corresponding magnetic axes 𝕊k(t)=ck(t)×𝕊1. Let Ak(t) and Bk(t) be the limits of the safety factor q(r, z, t) at the point ck(t) and at the boundary 𝒞k(t)=∂Dk(t), respectively. The limits Ak(t) are finite if the local maximum or minimum ck(t) of function ψ(r, z, t) is non-degenerate. The limits Bk(t) can be infinite, see [25]. Since the safety factor q(r, z, t) is preserved along the plasma streaklines, we get that its limits Ak(t) and Bk(t) also are preserved. Therefore, the limits do not depend on time t and hence are the new invariants Ak and Bk of the viscous plasma flows. We call them the functional invariants.

Suppose that, for some t, the function ψ(r, z, t) has a non-degenerate local maximum or minimum at a point ck(t)=(rk(t), zk(t)). This means that ∂ψ(ck(t))/∂r=0, ∂ψ(ck(t))∂z=0, and the Hessian

(22)(ck(t))=2ψ(ck(t))r22ψ(ck(t))z2[2ψ(ck(t))rz]2

is positive, ℋ(ck(t))>0. Then the neighbouring curves ψ(r, z, t)=ψ=const are closed curves Cψ(t) encircling the point ck(t). The curves Cψ(t) are closed trajectories of system (13). Let τ(Cψ(t)) be their periods. In the limit ψ(r, z, t)→ψ(ck(t)), the closed curves Cψ(t) (ψ(r, z, t)=const) tend to the equilibrium point ck(t): r(τ)→rk(t) and z(τ)→zk(t) for all τ. Applying formula (21), we find

Ak(t)=lim(r,z)(rk(t),zk(t))q(r,z,t)=τ(ψ(ck(t)))w[rk(t),zk(t),t]2πrk(t),

where τ(ψ(ck(t)))=limτ(Cψ(t)) when ψ(t)→ψ(ck(t)).

As is shown above, the value Ak(t) does not depend on t: Ak(t)=Ak=const. Using the formula derived in [25] τ(ψ(ck))=2πrk/(ck) for the limit of the periods τ(ψ) at ψψ(ck(t)), we get

(23)Ak=w[rk(t),zk(t),t](ck(t)).

Formula (23) defines the explicit form of the new functional invariants Ak of the MHD equations (1)–(3). The Hessians ℋ(ck(t)) (22) and invariants Ak (23) are independent of the gauge transformations (8).

Remark 1: Formula (23) proves that for any time-dependent axisymmetric magnetic field B(x, t), the safety factor q(x, t) has a finite limit Ak at a magnetic axis 𝕊k(t)=ck(t)×𝕊1, which means at ψ(t)→ψ(ck(t)), provided that the non-degeneracy condition ℋ(ck(t))>0 is met. Moffatt mistakenly stated in [16] that for the plasma equilibria the limits Ak are always infinite and the limits Bk in case of blobs k3 are always zero. These two statements have led Moffatt to the erroneous conclusions about the structure of the magnetic field knots. [2]

7 Infinite Families of Integral Invariants

Let Dk3(t1) be a maximal ring or blob, and let dv=rdrdzdφ be the 3D volume element. For any n, m=0, 1, 2, 3, … the formulae

(24)Jkmn=Dk3(t)qn(x,t)ρm(x,t)dv=2πDk(t)qn(r,z,t)ρm(r,z,t)rdrdz

define the infinite family of invariants of the axisymmetric MHD equations (1)–(3). Indeed, the plasma flow preserves all functions qn(x, t)ρm(x, t)); it transforms each set Dk3(t1) into Dk3(t) for t>t1 and preserves the 3D volume. Hence the integrals (24) are preserved.

Let us consider also the functionals

(25)Ikn=Dk3(t)cos(nq(x,t))ρ(x,t)dv=2πDk(t)cos(nq(r,z,t))ρ(r,z,t)rdrdz,

which are defined for any integer n≥0 and form an infinite matrix of invariants of (1)–(3). The matrix Ikn (25) has N=Nr+Ns rows because k=1, 2, …, Nr+Ns and infinitely many colums because n=0, 1, 2, 3, ….

Formulas (24) for n=m=0 and (25) for n=0 imply that the volume and the total mass Mk of each magnetic ring k3(t) and each magnetic blob k3(t) are invariants of plasma dynamics.

Each invariant |Ikn| (25) is bounded by the total mass Mk of plasma inside the set Dk3(t). Indeed, because |cos(nq)|≤1, formulas (25) yield |Ikn|Dk3(t)|cos(nq(x,t))|ρ(x,t)dvDk3(t)ρ(x,t)dv=Mk.

The mutual functional independence of invariants Ikn follows from that for the functions cosnq for n=1, 2, 3, ….

The conservation of functions qn(x, t)ρm(x, t)) along the plasma streaklines is equivalent to the equation

[qn(x,t)ρm(x,t))]/t+i=13Vi(x,t)[qn(x,t)ρm(x,t))]/xi=0,

where Vi(x, t) are components of the plasma velocity. This equation, together with equation divV=0, defines an infinite family of conservation laws

[qn(x,t)ρm(x,t))]t+i=13[Vi(x,t)qn(x,t)ρm(x,t))]xi=0,

where n, m=0, 1, 2, 3, ….

7.1 Functional Independence of the New Integral Invariants from the Magnetic Helicity

Using formula (21) for the safety factor q(r, z, t), we present the conserved quantities Jk0n (24) in the form

(26)Jk0n=1(2π)n1Dk(t)[Cψ(t)r1wdτ]nrdrdz,

where Cψ(t) is the closed trajectory of system (12) in the time τ with the initial data (r, z) for t=const. Magnetic helicity Hk(t) (18) for the same magnetic ring or blob Dk3(t) is

(27)Hk(t)=2πDk(t)[2ψ(r,z,t)bk(t)]w(r,z,t)drdz,

where bk(t) is defined by the equation ψ(r, z, t)=bk(t), the boundary curve 𝒞k(t) of the domain Dk(t).

Let us show that the new invariants (26) are functionally independent of the helicity (27). Indeed, for any magnetic ring or blob Dk3(t), the integral invariants (26) as functions of the magnetic field B are mutually functionally independent because the power functions qn and qm for nm are independent. Therefore, if for a fixed k all integrals (26) with arbitrary n≥1 were functionally dependent on the helicity Hk(t) (27), then Jk0n would have been mutually functionally dependent on each other, but they are not. This proves that the integral invariants Jk0n for all n≥1 with possibly one exception are functionally independent of the magnetic helicity Hk (27).

The integrals Jkmn (24) for m≥1 and Ikn (25) depend on plasma density ρ and on the safety factor q. These integrals for the non-constant plasma density ρ(r, z, t) are functionally independent of the helicity Hk(t) by the same reasons as above and also because formula (27) does not contain ρ(r, z, t).

8 Conclusion

The following main results were obtained in this article:

  1. For the axisymmetric dynamics of viscous plasmas with zero electrical resistance, we have proved the existence of magnetic rings ℛk(t) and magnetic blobs ℬm(t) which are frozen into the plasma flows and therefore cannot intersect each other during the plasma dynamics.

  2. We constructed the family of functional invariants Ak which are presented by the explicit formula (23).

  3. We presented infinitely many integral invariants Jknm (24) and Ikn (25) and proved that they are functionally independent of the magnetic helicity (4), (27).

  4. The axisymmetric dynamics of plasmas with zero electrical resistance between two given states is possible only if the corresponding total numbers of magnetic rings Nr are equal (the same for the total numbers of magnetic blobs Ns) and all new invariants Ak, Jkmn, Ikn for them coincide.

Acknowledgements

The author thanks the referees for useful remarks.

References

[1] W. A. Newcomb, Ann. Phys. 3, 347 (1958).10.1016/0003-4916(58)90024-1Suche in Google Scholar

[2] D. Biskamp, Nonlinear Magnetohydrodynamics, Cambridge University Press, Cambridge 1997.Suche in Google Scholar

[3] O. I. Bogoyavlenskij, Phys. Lett. A 307, 281 (2003).10.1016/S0375-9601(02)01732-2Suche in Google Scholar

[4] O. I. Bogoyavlenskij, Phys. Rev. E 66, 5640 (2002).10.1103/PhysRevE.66.056410Suche in Google Scholar

[5] O. I. Bogoyavlenskij, Phys. Rev. Lett. 84, 1914 (2000).10.1103/PhysRevLett.84.1914Suche in Google Scholar

[6] O. I. Bogoyavlenskij, Phys. Rev. E 62, 8616 (2000).10.1103/PhysRevE.62.8616Suche in Google Scholar

[7] S. Grozdanov, D. M. Hofman, and N. Iqbal, Phys. Rev. D 95, 096003 (2017).10.1103/PhysRevD.95.096003Suche in Google Scholar

[8] H. K. Moffatt, J. Fluid Mech. 35, 117 (1969).10.1017/S0022112069000991Suche in Google Scholar

[9] R. H. Crowell and R. H. Fox, Introduction to Knot Theory, Ginn and Company, Boston 1963.Suche in Google Scholar

[10] L. Woltjer, Proc. Nat. Acad. Sci. 44, 489 (1958).10.1073/pnas.44.6.489Suche in Google Scholar

[11] M. A. Berger, Plasma Phys. Control. Fusion 41, B167 (1999).10.1088/0741-3335/41/12B/312Suche in Google Scholar

[12] D. Serre, Physica 13D, 105 (1984).10.1016/0167-2789(84)90273-2Suche in Google Scholar

[13] E. A. Kudryavtseva, Math. Notes 95, 877 (2014).10.1134/S0001434614050332Suche in Google Scholar

[14] E. A. Kudryavtseva, Math. Notes 99, 611 (2016).10.1134/S0001434616030366Suche in Google Scholar

[15] A. Enciso, D. Peralta-Salas, and F. Torres de Lizaur, PNAS 113 2035 (2016).10.1073/pnas.1516213113Suche in Google Scholar

[16] H. K. Moffatt, Magnetic Field Generation in Electrically Conducting Fluids, Cambridge University Press, Cambridge 1978.Suche in Google Scholar

[17] C. Mercier, Fusion Nucleaire 4, 213 (1964).10.1088/0029-5515/4/3/008Suche in Google Scholar

[18] D. W. Kerst, J. Nucl. Energy, Part C 4, 253 (1962).10.1088/0368-3281/4/4/303Suche in Google Scholar

[19] K. J. Whiteman, Rep. Prog. Phys. 40, 1033 (1977).10.1088/0034-4885/40/9/002Suche in Google Scholar

[20] A. H. Boozer, Phys. Fluids 26, 1286 (1983).10.1063/1.864166Suche in Google Scholar

[21] A. H. Boozer, Rev. Mod. Phys. 76, 1071 (2005).10.1103/RevModPhys.76.1071Suche in Google Scholar

[22] E. L. Ince, Ordinary Differential Equations, Dover Publications, New York 1956.Suche in Google Scholar

[23] J. R. Cary and R. G. Littlejohn, Ann. Phys. 151, 1 (1983).10.1016/0003-4916(83)90313-5Suche in Google Scholar

[24] W. M. Hicks, Phil. Trans. Roy. Soc. London A 192, 33 (1899).10.1098/rsta.1899.0002Suche in Google Scholar

[25] O. Bogoyavlenskij, Phys. Rev. E, 95, 043104, 14 (2017).10.1103/PhysRevE.95.043104Suche in Google Scholar PubMed

[26] O. Bogoyavlenskij, J. Math. Anal. Appl. 450, 21 (2017).10.1016/j.jmaa.2016.12.080Suche in Google Scholar

[27] O. Bogoyavlenskij, J. Math. Phys. 5, 013101-01 (2017).10.1063/1.4973802Suche in Google Scholar

Received: 2017-09-07
Accepted: 2018-04-02
Published Online: 2018-04-27
Published in Print: 2018-06-27

©2018 Walter de Gruyter GmbH, Berlin/Boston

Heruntergeladen am 17.4.2026 von https://www.degruyterbrill.com/document/doi/10.1515/zna-2017-0318/html
Button zum nach oben scrollen