Skip to content

InnerLayer.SLAYER - TEST - Verify the two-fluid layer ODE reduces to the Fitzpatrick 2025 Eq. (100) width in the deep diffusive-resistive limit #411

Description

@d-burg

Motivation

While implementing the layer-overlap truncation (#298), the general two-fluid Riccati width
(riccati_del_s, the 2023 book's model — validated against its figs 6.2/6.3) and the closed-form
diffusive-resistive width of Fitzpatrick 2025 Eq. (100) (arXiv:2511.07666; the width the truncation
criterion is calibrated with) were compared on the DIII-D-like deck. They differ by a near-uniform
factor: Eq. (100) ≈ 0.47–0.59 × the ODE width across m = 2–10, surfaces spanning very different
parameters.

A near-constant ratio suggests a width-definition offset, not a physics disagreement: the ODE width
is defined through the dispersion relation (the equivalent resistive-sheet thickness, book Eq. 6.24,
extracted via the small-q limit of Eq. 6.41), while Eq. (100) is the spatial decay scale of the
layer solution (the |G|^(-1/4) p-space thickness of the paper's Eq. 97, converted to radius).
Those are inequivalent definitions of "width" for the same layer and should differ by a derivable
O(1) constant. But this is currently a hypothesis fit to one deck, and the ~20% drift in the ratio
is unexplained (model differences are plausible: four-field at Q = Q_e vs the paper's three-field
static-RMP response at Q_E + Q_e; compressibility/semi-collisional terms in one and not the other).

The check

In the deep diffusive-resistive corner (book regime map fig 6.1: Q̂*, P̂ → 0), either

  • extract a spatial-decay width from the general Riccati solve (the analogue of |G|^(-ν)), or
  • derive the book model's DR asymptote analytically,

and compare against Eq. (100) under a single width definition. The DIII-D 2/1 surface sits in that
corner (Q̂ ≈ 0.007, P̂ ≈ 0.009), so the existing slayer_dels_fitzpatrick harness fixture covers
the right parameter range for a regression pin once the constant is derived.

Acceptance

  • The ratio is explained by a derived constant (with the residual drift attributed), or a real model
    discrepancy is identified and its consequence for the truncation criterion stated.
  • The result is recorded where the two widths are documented (workflow docs page /
    references.md one-liner), and pinned in the harness if a clean identity exists.

Why it matters

The truncation criterion (#298) uses Eq. (100) because the paper's ε_c values and its fig 9
cross-validation are calibrated in that convention. On the DIII-D deck the factor ~2 between the two
widths is exactly the difference between truncating at ψ = 0.9971 (keep q ≤ 6) and ψ = 0.9892
(keep q ≤ 5), so knowing whether the offset is definitional or physical bounds the systematic
uncertainty of the domain choice. Not a blocker for #298.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions