Introduction Formulation Method 2D 3D

Variational integration for ideal MHDand formation of current singularities

Yao Zhou

Princeton Plasma Physics Laboratory

Acknowledgements: H. Qin, A. Bhattacharjee, Y.-M. Huang, J. W. Burby, A. H. Reiman.

3rd Asia-Pacific Conference on Plasma PhysicsNov 6, 2019 Hefei, China

Introduction Formulation Method 2D 3D

Coronal heating by nano-flares

‘Conflicting’ features of solar corona:

very high temperature (T & 106 K);nearly perfect conductivity (S & 1012).

Parker’s conjecture1:

current singularities tend to form;then, magnetic reconnection.

Observational evidence exists2.

1E. N. Parker, Astrophys. J. 174, 499 (1972).2J. W. Cirtain et al., Nature 493, 501 (2013).

Introduction Formulation Method 2D 3D

The Parker problem

Ideal, mathematical abstraction3:

can genuine current singularities emergein 3D line-tied plasmas?

Practical relevance: tendency matters;

if not singular, how thin?

Line-tied geometry: no closed field lines;

unlike toroidal fusion plasmas.

Challenge: to preserve magnetic topology;

both analytically and numerically.

Still controversial today.

3E. N. Parker, Spontaneous Current Sheets in Magnetic Fields: With Applications to Stellar X-rays(Oxford University Press, New York, 1994).

Introduction Formulation Method 2D 3D

Highlights

Lagrangian labeling: the favorable description;

built-in frozen-in equation;magnetic topology automatically preserved.

Discretization: ideal MHD on a moving mesh;

no artificial reconnection, unlike Eulerian methods;variational, with discrete exterior calculus.

Current singularity formation in 2D: conclusively confirmed;

the Hahm–Kulsrud–Taylor (HKT) problem;analytical and numerical solutions agree.

Extension to 3D line-tied geometry: inconclusive;

smooth linear solution;smooth nonlinear solution for short systems;scaling suggests finite-length singularity.

Introduction Formulation Method 2D 3D

Ideal MHD in Lagrangian labeling

Dynamical variable: fluid configuration map x(x0, t). [xij.= ∂xi/∂x0j , J

.= det(xij).]

Advection equations are formally solved:

∂tρ+∇ · (ρv) = 0 ⇔ ρd3x = ρ0 d3x0 ⇔ ρ = ρ0/J, (1a)∂t(p/ρ

γ) + v · ∇(p/ργ) = 0 ⇔ p/ργ = p0/ργ0 ⇔ p = p0/Jγ , (1b)∂tB−∇× (v ×B) = 0 ⇔ Bi dSi = B0i dS0i ⇔ Bi = xijB0j/J. (1c)

Then, built into the momentum equation:

ρ0ẍi = B0j∂

∂x0j

(xikB0kJ

)− ∂J∂xij

∂

∂x0j

(p0Jγ

+xklxkmB0lB0m

2J2

). (2)

A field theory, with Lagrangian4:

L[x] =

∫ [1

2ρ0ẋ

2 − p0(γ − 1)Jγ−1 −

xijxikB0jB0k2J

]d3x0. (3)

4W. A. Newcomb, Nucl. Fusion 1962, Suppl. 2, 451 (1962)

Introduction Formulation Method 2D 3D

Current singularity formation: problem statement

Customary to study equilibria. (Dynamics can be complicated.)

Perfectly conducting plasma: magnetic topology preserved.

Eulerian equilibrium equation,

(∇×B)×B = ∇p, (4)

is underdetermined:

difficult to attach the topological constraint;existence of singularities does not imply formation.

Lagrangian equilibrium equation has the constraint built-in:

B0j∂

∂x0j

(xikB0kJ

)=

∂J

∂xij

∂

∂x0j

(p0Jγ

+xklxkmB0lB0m

2J2

). (5)

Goal: find solutions with current singularities to Eq. (5), given smooth initial andboundary conditions.

Introduction Formulation Method 2D 3D

Ideal MHD on a moving mesh

Eulerian ideal MHD simulations:

artificial reconnection5.

Discretization in Lagrangian labeling:

moving mesh simulates fluid motion;no artificial reconnection.

Can be done directly6, but we shall...5T. A. Gardiner and J. M. Stone, J. Comput. Phys. 205, 509 (2005).6I. J. D. Craig and A. D. Sneyd, Astrophys. J. 311, 451 (1986).

Introduction Formulation Method 2D 3D

Variational integration for ideal MHD

Discrete exterior calculus (DEC)7:

discrete manifold;discrete differential forms;discrete Stokes’ theoremguarantees ∇ ·B = 0, ∇ · j = 0.

Discretize the Lagrangian:

conservative many-body form8,

L(x, ẋ) =∑σ00

1

2M(σ00)ẋ

2 −W (x); (6)

discrete equilibrium, ∂W/∂x = 0.

Add friction for equilibration.

(Structure-preserving numerical methodsin plasma physics9: guiding center, PIC...)

7M. Desbrun et al., arXiv:math/0508341 (2005).8Y. Zhou et al., Phys. Plasmas 21, 102109 (2014).9P. J. Morrison, Phys. Plasmas 24, 055502 (2017); J. Xiao et al., Plasma Sci. Technol. 20, 110501 (2018).

Introduction Formulation Method 2D 3D

The Hahm–Kulsrud–Taylor (HKT) problem

Relevance to toroidal fusion plasmas:

resonant magnetic perturbations;3D equilibria with nested surfaces.

2D incompressible plasma, B0y = x0.

Mirrored sinusoidal boundarydisplacement10: ξ(±a, y) = ∓δ cos ky.Multiple known equilibrium solutions11:

some contain current singularities;magnetic topology not preserved.

10T. S. Hahm and R. M. Kulsrud, Phys. Fluids 28, 2412 (1985).11R. L. Dewar et al., Phys. Plasmas 20, 082103 (2013); J. Loizu et al., Phys. Plasmas 22, 022501 (2015).

Introduction Formulation Method 2D 3D

Topology-preserving equilibrium solutions

ξ: Lagrangian displacement.

Linear solution: unphysical,discontinuous displacement.Boundary-layer (BL) solution12:nonlinear, continuous displacement.Numerical solution: flux-preservingGrad–Shafranov (GS) solver13.

−π −π/2 0 π/2 πky

0.00

0.02

0.04

0.06

0.08

By(0

+,y

)

BL

GS

0.02

0.04

0.06

0.08

0.10

ξ(x

0,π

/k)

(a)

BL, outer

BL, inner

linear

GS

0.0 0.1 0.2 0.3 0.4 0.5x0

−0.10

−0.08

−0.06

−0.04

−0.02

0.00

ξ(x

0,0

)

(b)

Surface current I ′δ(y) = 2By(0+, y).

12Y. Zhou et al., Phys. Plasmas 26, 022103 (2019); M. N. Rosenbluth et al., Phys. Fluids 16, 1894 (1973).13Y.-M. Huang et al., Astrophys. J. 699, L144 (2009).

Introduction Formulation Method 2D 3D

Fully Lagrangian solution converges to GS solution

Normal (x) direction, at y0 = 0:

x ∼ x20, By = B0y/(∂x/∂x0) ∼ sgn(x);plasma infinitely compressed14.

10−4 10−3 10−2 10−1

x0

10−6

10−5

10−4

10−3

10−2

10−1

x(x

0,0

)

642

1282

2562

5122

GS

x ∼ x20

10−5 10−2

10−1

By(x, 0)

J = 1 requires tangential (y) direction:

∂y/∂y0 ∼ x−10 , non-differentiable;plasma infinitely stretched.

10−4 10−3 10−2 10−1

x0

100

101

102

∂y/∂

y 0| (x

0,0

)

642

1282

2562

5122

GS

∼ x−10

10−2 10−1

10−3

10−1

GS solver: limited applicability. Lagrangian method: more complex topology, 3D.

14Y. Zhou et al., Phys. Rev. E 93, 023205 (2016).

Introduction Formulation Method 2D 3D

Line-tied geometry: smooth linear solution

Parker’s original setup:

uniform B0 = ẑ;prescribed footpoint motion.

Our formula:

known ‘susceptible’ B0 in 2D;no-slip footpoints;focus: effect of line-tying.

Line-tied HKT: boundary perturbationξ(±a, y, z) = ∓δ cos ky sin(πz/L).Given L, linear solution is smooth15.

−0.2 −0.1 0.0 0.1 0.2x0

−0.04

−0.02

0.00

0.02

0.04

ξ x(x

0,0

,L/2

)

N = 32

N = 64

N = 128

2D

Current density jl scales linearly with L.

5 10 15 20 25 30L

1.4

1.6

1.8

2.0

2.2

2.4

j l

N = 64

N = 128

15Consistent with E. G. Zweibel and H.-S. Li, Astrophys. J. 312, 423 (1987).

Introduction Formulation Method 2D 3D

Nonlinear solutions: convergence and scaling

For short systems, nonlinear solutions are smooth and well-resolved16.jn ∼ (Ln − L)−1, suggests finite-length singularity:

stronger than previously proposed exponential scaling17;echoes existing analytical theory18.

64 96 128 160N

1.5

2.0

2.5

3.0

3.5

4.0

j n

L = 18

L = 16

L = 14

L = 12

L = 10

L = 8

L = 6

6 8 10 12 14 16 18L

0.2

0.3

0.4

0.5

0.6

0.7

0.8

j−1

n

N = 64

N = 96

N = 128

N = 160

j−1l

16Y. Zhou et al., Astrophys. J. 852, 3 (2018).17D. W. Longcope and H. R. Strauss, Astrophys. J. 437, 851 (1994).18C. S. Ng and A. Bhattacharjee, Phys. Plasmas 5, 4028 (1998).

Introduction Formulation Method 2D 3D

Longer systems: stronger shear tear up the mesh

Current density at footpoint and mid-plane

Reduced MHD equilibrium equation,

B · ∇jz = 0. (7)

Singularity threaded through all z19.

In-plane field (z = z0, B0z = 1, J = 1),

B⊥ =∂x⊥∂x0⊥

·B0⊥ +∂x⊥∂z0

. (8)

At the footpoints (x⊥ = x0⊥ at z = 0, L),

jz = j0z + ẑ · ∇⊥ ×∂x⊥∂z0

. (9)

Only the second term can be singular.

Strong shear: an inherent feature.19A. A. van Ballegooijen, Astrophys. J. 298, 421 (1985).

Summary and outlook

Lagrangian labeling: the favorable description.

Built-in frozen-in equation.Guaranteed invariance of magnetic topology.

Numerical method: ideal MHD on a moving mesh.

No artificial reconnection.Variational discretization; with discrete exterior calculus.Re-meshing? Eulerian?

Current singularity formation in 2D: conclusively confirmed.

Analytical and numerical solutions agree.Dynamics? Time scale?

3D line-tied geometry: inconclusive.

Smooth nonlinear solution for short systems.Scaling suggests finite-length singularity.Better discretization more robust against strong shear?Generalize RDR’s boundary-layer approach?

Discrete Lagrangians and advection equations

In Eulerian labeling (σk, t): fixed mesh,

L(v, ρ, p,B) =∑σ3

[ 〈ρ, σ3〉8

∑σ0≺σ3

〈v2, σ0〉 − 〈p, σ3〉

γ − 1 −∑σ2≺σ3

| ? σ2|2|σ2| 〈B, σ

2〉2]. (10)

In Lagrangian labeling (σk0 , t): moving mesh,

L(x, ẋ) =∑σ30

[ 〈ρ0, σ30〉8

∑σ00≺σ30

ẋ2 − 〈p0, σ30〉

(γ − 1)Jγ−1 −∑σ20≺σ30

| ? σ2|2|σ2| 〈B0, σ

20〉2]. (11)

Discrete advection equations:

〈ρ, σ3〉 = 〈ρ0, σ30〉, (12a)〈p, σ3〉|σ3|γ−1〈ρ, σ3〉γ =

〈p0, σ30〉|σ30 |γ−1〈ρ0, σ30〉γ

, (12b)

〈B, σ2〉 = 〈B0, σ20〉. (12c)

On the impossibility of current singulairities

An existing proof20 has an oversight.

P'

Q'

Q

12

z=L

z=0

Pa)

P'

Q'

Q

12

z=L

z=0

P

R'

R

b)

(a) Alleged contradiction:∫1+2

B(l) [1− b2(l) · b1(l)] dl = 0. (13)

(b) Accounting for the ‘hole’:∫1+2

B(l) [1− b2(l) · b1(l)] dl

=

∮∂Ω

B · dl =∫

Ω

j · dS. (14)

Magnetic shear keeps possibility alive.

20S. C. Cowley, D. W. Longcope, and R. N. Sudan, Phys. Rep. 283, 227 (1997).

General mechanism for current singularities in 2D

Coalescence instability of magnetic islands21:

different drive, more complex topology;same quadratic mapping as in HKT22.

Recipe: compress a sheared field.

21D. W. Longcope and H. R. Strauss, Phys. Fluids B 5, 2858 (1993).22Y. Zhou, Ph.D. thesis, Princeton University, arXiv:1708.08523 (2017).

Discontinuous 〈By〉 but continuous rotational transform

10−5 10−4 10−3 10−2 10−1 100

x0

10−610−510−410−310−210−1100

ι

initial

equilibrium

Rotational transform ι ∼ 〈B−1y 〉−1:stays invariant and continuous;contrary to a recent claim23.

0.000 0.001 0.002 0.003 0.004 0.005x

0.0

0.1

0.2

0.3

0.4

0.5

y 0.2 0.40.0

0.2

0.4

Equilibrium solution: flux surfaces

23J. Loizu and P. Helander, Phys. Plasmas 24, 040701 (2017).

Background and highlightsFormation of current singularities Variational integration for ideal MHD2D: the Hahm–Kulsrud–Taylor problem3D: extension to line-tied geometryAppendix

0.0: 0.1: 0.2: 0.3: 0.4: 0.5: 0.6: 0.7: 0.8: 0.9: 0.10: 0.11: 0.12: 0.13: 0.14: 0.15: 0.16: 0.17: 0.18: 0.19: 0.20: 0.21: 0.22: 0.23: 0.24: 0.25: 0.26: 0.27: 0.28: 0.29: 0.30: anm0: 0.EndLeft: 0.StepLeft: 0.PauseLeft: 0.PlayLeft: 0.PlayPauseLeft: 0.PauseRight: 0.PlayRight: 0.PlayPauseRight: 0.StepRight: 0.EndRight: 0.Minus: 0.Reset: 0.Plus: 1.0: 1.1: 1.2: 1.3: 1.4: 1.5: 1.6: 1.7: 1.8: 1.9: 1.10: anm1: 1.EndLeft: 1.StepLeft: 1.PauseLeft: 1.PlayLeft: 1.PlayPauseLeft: 1.PauseRight: 1.PlayRight: 1.PlayPauseRight: 1.StepRight: 1.EndRight: 1.Minus: 1.Reset: 1.Plus: