This browser-side companion uses a simplified Green’s-function model in the poloidal cross-section, but it now includes a boundary-fitted Grad–Shafranov-style inverse step. The active ITER preset now uses the public DINA-IMAS rectangular CS/PF winding-pack model, together with a plasma current found either from a boundary-fitted GS profile solve or from simpler reference models. Start from the full preset configuration, then turn families on and off to see how each contribution bends the flux.
ITER uses the imported public CS/PF rectangle set. SPARC and DIII-D are temporarily disabled until their coil packs are curated to the same standard.
The analytic ITER-like mode uses the Cerfon/Freidberg single-null family with ITER-like positive-triangularity shaping: the full analytic flux is drawn directly, while the plasma-current source is rebuilt from the analytic particular solution Δ*ψ ∝ A + (1-A)x² and fed through the same Green's-function lattice used elsewhere on the page. The Cerfon-seeded ITER mode instead fits both the plasma R p'(ψ) / F F'(ψ)/R source amplitudes and the active external coil-family scales so the target boundary becomes as close as possible to an isoflux contour while matching the chosen plasma current. The boundary-fitted mode only adjusts the plasma profile amplitudes for a fixed external-coil mix; the free-boundary mode keeps the older moving-mask Picard iteration, and the superposition mode keeps the original prescribed source for decomposition-only viewing.
Controls how hollow or peaked the pressure and current source functions are through (1 - ψN)α.
0 biases the plasma current toward the pressure-like R p'(ψ) family, while 1 biases it toward the toroidal-field F F'(ψ)/R family. In the boundary-fitted mode this acts as a prior, while the solver still adjusts both amplitudes to flatten the target boundary.
Awaiting solve.
Awaiting solve.
Awaiting solve.
Awaiting solve.
Awaiting solve.
Awaiting solve.
Awaiting solve.
Awaiting solve.
∇⊥2ψ = -C jφ
together with precomputed Green’s-function response tables for the fixed toroidal-filament source locations. The coil families and lattice-based plasma sources now use the axisymmetric elliptic-integral ring-current kernel, while the older softened 2D log kernel is retained only for the dense grid-sourced iterative fallback. External coils are represented as finite coil packs, each approximated by a small grid of toroidal-current filaments projected into the poloidal plane. In the iterative mode the plasma update uses a Grad–Shafranov-inspired source family jφ ∼ (1-ψN)α[(1-\lambda)R p'(ψ) + \lambda F F'(ψ)/R] before renormalizing to the chosen total plasma current.
In the analytic ITER-like mode the target LCFS and interior contour family come directly from a positive-triangularity single-null Cerfon/Freidberg equilibrium using ITER-like shaping parameters, while the plasma-current source is rebuilt from its particular solution and then passed through the same Green's-function model used for the coil families. In the Cerfon-seeded ITER mode those same profile families are used as a D-shaped seed and fitted together with the active external coil-family amplitudes. In the boundary-fitted mode the plasma families alone are adjusted so the prescribed plasma boundary becomes as close as possible to an isoflux contour while satisfying the chosen plasma current.
The full analytic ITER-like single-null equilibrium is the comparison target in this mode. Its interior and exterior flux contours are drawn directly from the analytic solution, the dashed green curve marks the analytic separatrix, and the black contour shows the current total-field reconstruction from the plasma source plus the selected coil families.
The lineout along Z = 0 shows how the active total field and the individual families contribute to the vacuum-plus-plasma flux balance across the machine.
The total-field trace measures ψ(θ) - ⟨ψ⟩ around the target boundary. A perfectly self-consistent equilibrium would leave this curve flat.
The curve uses the closed flux contours from the current iterate and reports a normalized geometric safety-factor shape q/qedge. In this reduced model it is the profile shape, not the absolute normalization, that is trustworthy.