Struct MultiplicativeBaselineConfig 

pub struct MultiplicativeBaselineConfig {
    pub b0_init: f64,
    pub b1_init: f64,
    pub b2_init: f64,
    pub fit_b0: bool,
    pub fit_b1: bool,
    pub fit_b2: bool,
    pub spatial_global: bool,
    pub b0_bounds: (f64, f64),
    pub b1_bounds: (f64, f64),
    pub b2_bounds: (f64, f64),
}

Bounded multiplicative polynomial baseline (issue #635): y(E) = (b0 + b1·z + b2·z²) · T_model(E) with z = ln(E/E_ref) and E_ref = √(E_min·E_max) of the fit grid.

A NEREIDS extension (see MultiplicativeBaselineModel for the SAMMY departure record): real VENUS sample/open-beam ratios sit a few % from unity with smooth energy dependence, and freeing the SAMMY Anorm together with temperature and density is degenerate on such data (IPTS-37432 finding A3: T → 4471 K with no warning). The bounded multiplicative form fitted jointly with temperature at fixed density is the production recipe (χ²/ν ≈ 2–8 across the 20-run campaign).

b0 is exactly degenerate with Anorm, so a free Anorm alongside ANY configured baseline — even a fully frozen one — is rejected (validate_multiplicative_baseline). A frozen-b0 + free-Anorm combination would be well-posed, but supporting it buys nothing (Anorm would just play b0’s role at a rescaled value); the sanctioned combination is the additive ABC(+DF) background with fit_anorm = false.

What prevents the A3-style runaway is SMOOTHNESS, not amplitude: a single global quadratic in ln E cannot produce a narrow local feature at a resonance whatever its coefficients, so the dips stay attributed to the physics model. The (configurable) coefficient boxes bound the baseline’s overall tilt/curvature PER ln-E UNIT, not the evaluated B(E) itself: the worst-case excursion from b0 grows with the grid’s log-width, |B − b0| ≤ |b1|·|z|_max + |b2|·z²_max (≈ ±0.23 beyond b0’s own ±0.1 box on a 1–30 eV grid at the default ±0.05 boxes; more on wider grids, where only the runtime B(E) > 0 guard applies). “A few % off unity” describes the typical FITTED values on campaign data and the hard b0 box at mid-grid — not a hard envelope on B(E) across the grid.

Fields

b0_init: f64

Initial mid-grid baseline value b0 (default 1.0).

b1_init: f64

Initial slope per ln-E unit b1 (default 0.0).

b2_init: f64

Initial curvature per ln-E² unit b2 (default 0.0).

fit_b0: bool

Whether b0 is free (default true). All-false = frozen baseline (the #633 fixed-parameter pattern; used by the spatial global mode).

fit_b1: bool

Whether b1 is free (default true).

fit_b2: bool

Whether b2 is free (default true).

spatial_global: bool

spatial_map_typed only: fit ONE baseline on the aggregated mean spectrum, then freeze it for every pixel (default true — per-pixel baselines on <500 counts/bin biased fitted temperatures by up to +150 K on IPTS-37432; the global mode removed 80 % of that). Ignored by the single-spectrum fitters.

b0_bounds: (f64, f64)

Optimizer box for b0 (default (0.9, 1.1) — “a few % off unity”).

b1_bounds: (f64, f64)

Optimizer box for b1 (default (-0.05, 0.05)).

b2_bounds: (f64, f64)

Optimizer box for b2 (default (-0.05, 0.05)).

Trait Implementations

impl Clone for MultiplicativeBaselineConfig

fn clone(&self) -> MultiplicativeBaselineConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for MultiplicativeBaselineConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for MultiplicativeBaselineConfig

fn default() -> Self

Returns the “default value” for a type.