Physics · PlasmaLean handbook

Plasma models and equilibrium theory

Ideal MHD, static and weak equilibria, boundary conditions, and model reductions.

Lean 4.34.1Mathlib 4.34.19 tracked tasks
Implementation status · 2026-10-14Exact finite-pressure equilibria (toroidal Solov'ev and a nonaxisymmetric helical one) now satisfy the kernel force-balance law, with all-time full-orbit confinement theorems, filament Maxwell identities, and Suydam/Newcomb stability criteria with exact certificates. All are idealized symmetric models; no coil-produced stellarator is certified and no all-mode MHD stability is proved. Compiled declarations, build results and assumptions → Chapter notes retain their original planning context; the task progress fields report subsequent work.

← Illustrated introduction: The kernel

Read 00-model-contracts.md before using these specifications. References refer to 09-literature.md. All proposed Lean declarations and solver bridges remain to be implemented.

Contents

State, scales, and units

Keep mass density ρ, number density nₛ, pressure p, temperature Tₛ, magnetic field B, current J, and fluid velocity u distinct. For a Maxwellian isotropic mixture use a specified equation of state

$$p=\sum_s n_s k_BT_s,\qquad \rho=\sum_s m_s n_s.$$

If temperature is supplied as energy in keV, write pₛ = nₛ Θₛ with Θₛ = k_B Tₛ and perform the energy-to-joule conversion. Do not silently insert keV values into a kelvin formula.

Introduce reference scales L₀, B₀, ρ₀ and derive the nondimensional equations. Track at least

$$v_A=\frac{B_0}{\sqrt{\mu_0\rho_0}},\quad \beta=\frac{2\mu_0p_0}{B_0^2},\quad \epsilon_\rho=\frac{\rho_i}{L_0},\quad \rho_i=\frac{\sqrt{2m_i\Theta_i}}{|q_i|B_0},$$

and collision, resistive, thermal, orbit, and operating timescales. The thermal-speed convention is explicitly v_th = √(2Θ/m). A model-validity interface should record the ordering parameters rather than treating “small” as a theorem.

For realistic case studies, import a documented experimental or reactor operating profile together with its geometry; use W7-X transport data or the HELIAS engineering study as entry points. Neither provides a universal minimum field, coil spacing, or material lifetime.

Ideal MHD

For a smooth isotropic adiabatic fluid without gravity, heating, or resistivity:

$$\partial_t\rho+\nabla\cdot(\rho u)=0,$$
$$\rho(\partial_tu+u\cdot\nabla u)=-\nabla p+J\times B,$$
$$\partial_tB=\nabla\times(u\times B),\quad \nabla\cdot B=0,\quad J=\mu_0^{-1}\nabla\times B,$$
$$(\partial_t+u\cdot\nabla)p+\Gamma p\,\nabla\cdot u=0,\qquad \Gamma>1.$$

Equivalently, D_t(pρ^{-Γ})=0 where ρ>0. Define solutions with a declared time interval, regularity, initial data, and boundary conditions. The source model Freidberg, Ideal MHD supplies the physical assumptions and variants.

First prove:

  1. div B remains zero because div curl=0, in the chosen classical solution class.
  2. Mass is conserved on a fixed impermeable boundary u·n=0.
  3. For the appropriate closed conducting boundary, total energy
$$ E=\int_\Omega\left(\tfrac12\rho|u|^2+\frac{p}{\Gamma-1} +\frac{|B|^2}{2\mu_0}\right)dx$$

is conserved; retain boundary fluxes in the general identity.

  1. Smooth material magnetic flux is conserved under the ideal induction equation, with a proved flow map.
  2. Positive initial density stays positive along a sufficiently regular flow for the existence interval.

Conservation identities alone do not establish global smooth existence or nonlinear stability in a strong norm.

Static equilibria and boundary conditions

For u=0, isotropic static equilibrium means

$$J\times B=\nabla p,\qquad \nabla\cdot B=0,\qquad \nabla\times B=\mu_0J.$$

Specify whether profiles, total currents, fluxes, helicity, or boundary shape are prescribed. Density is not determined by force balance alone; it enters stability through inertia and the operating model.

Fixed boundary. Require an appropriate magnetic boundary, often B·n=0. Initially analyze perturbations with zero boundary trace; later extend to the chosen conducting-wall condition.

Free plasma–vacuum interface. Include a vacuum region, its magnetic field, and coil/external wall boundary data. For an ideal magnetic interface B·n=0 on both sides with no additional imposed traction, require continuity of total pressure:

$$\left[p+\frac{|B|^2}{2\mu_0}\right]=0.$$

Use normal-field continuity, tangential jump/current conditions, and the full stress formulation for more general interfaces. An equilibrium computed with a fixed plasma boundary does not supply a free-boundary stability certificate.

Vacuum. J=0 gives curl B=0, in addition to div B=0. On a simply connected region with sufficient regularity, a scalar-potential theorem gives B=∇Φ and ΔΦ=0. On a toroidal region retain harmonic/circulation components.

Force-free. Constant pressure permits J∥B where B≠0. A useful exact counterexample to “zero pressure implies vacuum” uses B₀,L₀>0 on a Cartesian domain periodic in z with period 2πL₀:

$$B=B_0(\cos(z/L_0),\sin(z/L_0),0),\quad \nabla\times B=-B/L_0,\quad J=-B/(\mu_0L_0),\quad p=0.$$

This model has nonzero current and zero Lorentz force, with explicit field and length scales preserving SI dimensions. It is a kernel sanity witness, not a stellarator reactor.

Axisymmetry. With cylindrical coordinates and a selected poloidal flux convention,

$$B=\frac1R\nabla\psi\times e_\phi+\frac{F(\psi)}R e_\phi,\quad p=p(\psi),$$
$$\Delta^*\psi=-\mu_0R^2p'(\psi)-F(\psi)F'(\psi),\qquad \Delta^*=R\partial_R(R^{-1}\partial_R)+\partial_Z^2.$$

Prove this reduction on R>0 and track the sign/flux convention. Use it for verified manufactured equilibria and a simpler PDE existence milestone. Axisymmetric examples provide infrastructure and sanity checks, while the practical target remains a three-dimensional stellarator.

Weak and stepped-pressure equilibria

Define the static stress

$$\mathsf S=\left(p+\frac{|B|^2}{2\mu_0}\right)\mathsf I -\frac{B\otimes B}{\mu_0}.$$

For smooth divergence-free fields, div S=0 is equivalent to force balance. For piecewise smooth fields formulate div S=0 distributionally, div B=0 distributionally, and the resulting interface jumps. This prevents multiplication of undefined distributional currents by discontinuous fields without a specified interpretation.

For a multi-region relaxed-MHD model:

  • decompose the toroidal domain into regions separated by magnetic interfaces;
  • use constant pressure within each relaxed region;
  • impose curl B = λᵢ B within each region;
  • impose interface force balance, flux and helicity/current constraints;
  • permit islands and chaotic field lines within the regions.

The SPEC/MRxMHD paper supplies a candidate-generating formulation. Prove the appropriate existence and stability statements for that model; do not silently transfer them to smooth continuously nested ideal-MHD equilibrium.

Enciso–Luque–Peralta-Salas prove piecewise smooth nonconstant-pressure equilibria under nondegenerate toroidal-domain hypotheses. Extract their actual theorem, including the Beltrami/KAM assumptions. This is a substantial existing theorem to formalize, rather than an assertion that all nonaxisymmetric equilibria are unknown.

Smooth nonaxisymmetric equilibrium with a nonconstant pressure profile and a complete prescribed surface foliation raises small-divisor and resonance questions. VMEC/DESC representations impose nested surfaces; a converged finite representation alone is not an existence proof for a classical continuum solution. Research should target a specified class with constructive constants, allow weaker/stepped-pressure alternatives where appropriate, and account for obstructions. Constantin–Drivas–Ginsberg construct equilibria with a small additional force; that force must remain in their conclusion.

Beyond ideal MHD

Keep separate interfaces for:

  • resistive Ohm law E+u×B=η_e J, where η_e is electrical resistivity, and induction with magnetic diffusivity η_e/μ₀;
  • generalized Ohm law with Hall, electron-pressure, and inertia terms;
  • anisotropic stress P=p_⊥ I+(p_∥-p_⊥)bb, b=B/|B|;
  • viscosity, heat conduction, sources, and losses;
  • equilibrium flows and resistive walls.

In an isothermal/constant-resistivity incompressible toy model with closed boundaries, magnetic plus kinetic energy decreases through resistive and viscous dissipation. In a thermodynamically closed resistive model, Joule heating transfers electromagnetic energy into internal energy rather than destroying total energy. State which balance is being proved.

Static no-flow self-adjoint MHD stability machinery does not automatically apply to a flowing equilibrium or a dissipative nonself-adjoint operator.

Kinetic reference and reduction contracts

A multispecies nonrelativistic kinetic model is

$$\partial_tf_s+v\cdot\nabla_xf_s +\frac{q_s}{m_s}(E+v\times B)\cdot\nabla_vf_s =C_s(f)+S_s,$$

with Maxwell equations and explicitly chosen collision/source operators. Specify decay or velocity boundaries so that moments and integration by parts are legitimate. Define ρ_c=Σqₛ∫fₛdv and J=Σqₛ∫vfₛdv.

Prove moment identities under collision conservation hypotheses. Closure of the pressure tensor and heat flux is an additional assumption or reduction theorem, not a consequence of taking moments alone.

A useful reduction contract has the form

$$\|{\cal O}(z_{\rm full}(t))-{\cal O}(z_{\rm reduced}(t))\| \le \varepsilon_{\rm model}(t)$$

for specified initial data, observables, parameters, and times. Plasma stability in a reduced model transfers only through such a contract and a sufficient robustness margin. Proving useful kinetic-to-fluid and orbit-reduction error bounds is part of the ambitious roadmap.

Theorem backlog

IDTargetDependency / statusProgress so far
E1Force balance implies tangency; force-free/vacuum distinctionElementary existing theoryLean verified — tangency; force-free slab; exact toroidal (Solov'ev) and helical finite-pressure equilibria satisfy the kernel ForceBalance law with actual Ampère current.
E2MHD conservation identities on smooth closed domainsCalculus, divergence theorem, ODE flowPlanning only — no Lean code written
E3Grad–Shafranov equivalence and explicit examplesCylindrical calculusPartially implemented — an exact Solov'ev solution is verified directly against J×B=∇p (Δ*ψ=3R² checked by hand only); general Grad–Shafranov equivalence pending. A helical (JFKO-type) exact equilibrium is also verified.
E4Distributional stress and magnetic interface jump conditionsWeak derivatives and tracesPlanning only — no Lean code written
E5Validated equilibrium in a specified simpler PDE classContraction/inverse bounds, 05Planning only — no Lean code written
E6Formalize a sourced piecewise smooth nonaxisymmetric existence theoremBeltrami fields, KAM, analytic estimatesPlanning only — no Lean code written
E7Certified finite-pressure free-boundary equilibrium for external coilsCoupled interior/vacuum fields and interfacePlanning only — no Lean code written
E8Constructive smooth/weak equilibrium branches over useful uncertainty setsResearch target with specified admissible classPlanning only — no Lean code written
E9Quantified kinetic/fluid or resistive/ideal model comparisonResearch target; singular limits and time horizonPlanning only — no Lean code written