[EFIT] Expose and qualify GA-EFIT attached/SOL current model (`ICUTFP`) for VEST

[HongSik-Yun-Fusion]

Summary

GA-EFIT supports an attached / SOL-current model through the namelist parameter ICUTFP, which changes the spatial support of the reconstructed toroidal plasma current.

Current VAFT EFIT studies already examine whether the Grad–Shafranov source functions are forced to vanish at the LCFS through PCURBD / FCURBD, but this is a different modeling axis:

PCURBD / FCURBD
    control the source-function boundary condition at psi_N = 1

ICUTFP
    controls whether reconstructed current is allowed outside the LCFS

With ICUTFP = 0, plasma current is restricted to the closed-flux region. With the attached-current option enabled, GA-EFIT can permit a force-free current contribution outside the LCFS, subject to its flux-domain and limiter / vessel restrictions, while the pressure-gradient contribution remains confined to the closed-flux plasma.

This issue should first expose the upstream GA-EFIT option correctly in VAFT and then investigate whether this additional degree of freedom is relevant or meaningfully constrained in VEST equilibria.

The study plan below is an initial investigation plan, not a fixed protocol. It may be revised as the GA-EFIT implementation is verified and results from the first reconstruction tests become available.

Motivation

The profile-model study in #579 addresses whether the current density is forced to vanish at the LCFS through PCURBD and FCURBD. That does not answer the distinct question of whether reconstructed plasma current should be allowed on open field lines outside the LCFS.

For example,

PCURBD = 0
FCURBD = 0
ICUTFP = 0

may permit finite current approaching the LCFS from inside while still enforcing zero reconstructed plasma current outside the LCFS.

The GA-EFIT attached-current model changes the current-support domain itself. This may be relevant for VEST during phases such as:

• current ramp-up,
• ramp-down / termination,
• strong limiter interaction,
• vertically displaced or disruption-like evolution.

However, external magnetics may not uniquely identify such a current component. A better magnetic fit could instead result from an additional degree of freedom absorbing errors in the passive-current model, profile model, LCFS location, or diagnostic weighting.

The goal is therefore to qualify the model, not to adopt SOL current as a new routine default.

Upstream GA-EFIT semantics

The original GA-EFIT implementation uses ICUTFP to distinguish the standard closed-flux current model from an attached-current model.

Schematically:

ICUTFP = 0

0 <= psi_N <= 1
    -> p' and FF' current contributions

outside LCFS
    -> no reconstructed plasma current

For the attached-current branch:

ICUTFP > 0

closed-flux region
    -> p' and FF' contributions

extended flux region outside LCFS
    -> FF' contribution may remain

limiter / vessel mask
    -> restricts spatial support

Thus, outside the closed-flux region, the additional current is approximately force-free:

[
j_\phi = R p'(\psi) + \frac{FF'(\psi)}{\mu_0 R},
]

with the p' contribution removed while the FF' contribution may extend onto open field lines.

The exact semantics of ICUTFP, XPSIMIN, ALPHAFP, and which of these are intended as external controls should be verified directly against the GA-EFIT source and the VEST EFIT executable before defining the VAFT API.

Relationship to existing VAFT work

#71 — EFIT configuration

#71 exposed the main profile, boundary-condition, numerical, and constraint settings, including KPPCUR, KFFCUR, KPPFNC, KFFFNC, PCURBD, and FCURBD, but did not explicitly expose the attached / SOL-current controls.

This issue should extend the existing EFIT configuration mechanism rather than introduce a separate configuration path.

#579 — profile-model uncertainty

#579 studies hard-zero edge vs finite edge current through PCURBD / FCURBD.

The present issue instead studies current confined inside LCFS vs current support extended outside LCFS.

These effects should initially be kept separate.

#663 — constraint identifiability

#663 studies which VEST constraint families carry independent information. That question is directly relevant here because an SOL-current degree of freedom may reduce the fit residual without being independently constrained by the measurements.

#492 — halo / limiter-current diagnostics

#492 concerns measured limiter / halo-current events. That is distinct from GA-EFIT's reconstructed attached-current model. Comparison against limiter-current diagnostics may become useful later, but it should not be assumed that the two quantities are equivalent.

Initial implementation scope

Expose the required GA-EFIT attached-current control through the existing VAFT EFIT configuration system.

Possible API direction:

@dataclass(frozen=True)
class EFITProfileConfig:
    ...
    icutfp: int = 0
    xpsimin: float | None = None
    alphafp: float | None = None

The exact fields and placement should follow the verified GA-EFIT semantics rather than this preliminary sketch.

Initial requirements:

• preserve the current behavior as the default;
• write the relevant EFIT input fields explicitly;
• validate supported values;
• record the requested / resolved configuration in provenance;
• add regression tests for generated EFIT input.

If some quantities are internally derived by GA-EFIT rather than intended as user-facing parameters, they should not be exposed unnecessarily.

Initial physics study

Start with a minimal comparison intended to isolate the current-support effect.

Keep the existing profile model and reconstruction settings fixed, for example:

KPPCUR = 2
KFFCUR = 2
PCURBD = 1
FCURBD = 1
FWTBP = 0

and compare:

Model	ICUTFP	Description
S0	0	standard closed-flux current
S1	1	attached / SOL-current enabled

The exact configuration may be adjusted if inspection of the upstream implementation shows that additional parameters must be varied consistently.

Initially avoid scanning PCURBD / FCURBD at the same time so that the LCFS boundary-condition effect and the outside-LCFS current-support effect remain distinguishable.

Quantities to compare

At minimum, examine:

Current

[
I_{\rm closed} = \int_{\psi_N \le 1} j_\phi,dA
]

[
I_{\rm SOL} = \int_{\psi_N > 1} j_\phi,dA
]

and I_SOL / I_p.

Inspect the spatial distribution of j_phi outside the LCFS and, where practical, separate the p' and FF' contributions.

Magnetic fit

• total and diagnostic-family residuals,
• flux-loop residuals,
• B-pol probe residuals,
• plasma-current consistency.

Equilibrium

• LCFS,
• magnetic axis,
• plasma area / volume,
• li,
• beta_p,
• q95,
• relevant profile changes.

Numerical behavior

• convergence,
• iteration behavior,
• failed slices,
• slice-to-slice stability.

Discharge-phase dependence

The comparison should not be limited to a single equilibrium slice. Where practical, examine representative periods from:

1. current ramp-up,
2. quasi-stationary phase,
3. ramp-down / termination.

This can test whether the effect of the attached-current model becomes more important during transient or plasma-wall-interacting phases.

An improved fit with ICUTFP > 0 should not by itself be interpreted as evidence that physical SOL current has been identified.

Follow-up questions

Depending on the initial results, later work may investigate:

Interaction with edge boundary conditions

A useful follow-up could compare:

PCURBD	FCURBD	ICUTFP
1	1	0
0	0	0
1	1	1
0	0	1

This would distinguish finite edge current from current extending outside the LCFS.

Constraint sensitivity

If SOL-current freedom noticeably improves the magnetic fit, test whether the result is stable against changes in passive-current treatment, diagnostic-family weighting, and profile representation.

Limiter-current comparison

If suitable VEST limiter-current data are available, compare the reconstructed attached-current behavior with measured plasma-wall-current timing and trends. This should be treated as a separate validation step rather than assumed correspondence.

Initial shots

Where practical, reuse representative shots from #579 / #663 so that the SOL-current effect can be compared against existing profile-model studies.

Initial candidates include:

39915
41672

A discharge with stronger termination or plasma-wall interaction may be added if the first tests suggest that this regime is more informative.

Shot selection can be revised during the study.

Acceptance criteria

• Verify the GA-EFIT implementation and semantics of ICUTFP and related parameters.
• Expose the necessary controls through the existing VAFT EFIT configuration layer.
• Preserve the current standard model as the default.
• Add input-generation / configuration tests.
• Demonstrate a reproducible ICUTFP=0 vs attached-current comparison on representative VEST data.
• Quantify reconstructed current outside the LCFS and its effect on magnetic fit and equilibrium quantities.
• Document whether the added freedom appears negligible, useful, poorly constrained, or numerically problematic.
• Keep the routine EFIT default unchanged unless later evidence justifies a separate change.

The implementation and analysis details may be revised as the investigation progresses; the parameter choices and comparison matrix above are intended as a starting point rather than a fixed final protocol.

Out of scope

• immediately changing the production EFIT default;
• interpreting reconstructed attached current as measured halo current;
• disruption-event detection or full VDE reconstruction;
• vessel-force / halo-force calculations;
• broad simultaneous scans of unrelated profile, diagnostic, and numerical settings.

The initial question is narrower:

Does GA-EFIT's attached-current degree of freedom work correctly through VAFT, and does VEST provide enough information for it to materially affect or constrain the reconstructed equilibrium?