Enum FrailtySpec 

pub enum FrailtySpec {
    None,
    GaussianShift {
        sigma_fixed: Option<f64>,
    },
    HazardMultiplier {
        sigma_fixed: Option<f64>,
        loading: HazardLoading,
    },
}

Frailty modifier specification at the family level.

Two structurally different exact modifiers exist:

1. GaussianShift: additive Gaussian on the final transformation index. Exact for probit families — the existing sextic microcell kernel survives unchanged (just scale denested cell coefficients by 1/√(1+σ²)).

2. HazardMultiplier: lognormal multiplier on the loaded cumulative hazard. Exact for PH/cloglog families — row likelihoods are finite sums of K_{k,m}(μ, σ) kernel terms.

These are mathematically distinct families. Do not mix them.

Variants

None

No frailty modifier.

GaussianShift

Gaussian shift on the final scalar index: U ~ N(0, σ²) added to η. Exact for probit: E[Φ(η + U)] = Φ(η / √(1+σ²)). The existing sextic microcell kernel is preserved.

Fields

sigma_fixed: Option<f64>

Fixed σ, or None if learnable.

HazardMultiplier

Lognormal hazard multiplier: conditional hazard h(t|U) involves exp(U). Exact for PH/cloglog/survival via K_{k,m} kernel.

Fields

sigma_fixed: Option<f64>

Fixed σ, or None if learnable.

loading: HazardLoading

How the multiplier loads onto hazard components.

Implementations

impl FrailtySpec

pub fn is_active(&self) -> bool

Whether this spec requests an actual frailty modifier.

FrailtySpec::None is the sole “no frailty” value. Family/mode guards that only support the no-frailty case must reject on this predicate, or they misclassify every ordinary CLI fit as a frailty request.

pub fn validate(&self) -> Result<(), LognormalKernelError>

Validate the frailty scale domain independently of a model family.

pub fn resolve_fixed_gaussian_shift( &self, context: &str, ) -> Result<Self, LognormalKernelError>

Resolve the exact frailty subset supported by Gaussian-shift marginal-slope models.

These models admit either no frailty or a Gaussian shift with a fixed standard deviation. A learnable Gaussian-shift scale and the structurally different hazard-multiplier frailty require other fitting machinery and are rejected here at the type that owns those states.

pub fn validate_for_marginal_slope(&self) -> Result<(), String>

Validate that this frailty spec is compatible with score_warp/linkwiggle cubic marginal-slope families.

• GaussianShift is exact: the sextic microcell kernel is preserved (probit scaling by 1/τ, τ = √(1+σ²)).
• HazardMultiplier is exact only for PH/cloglog rowwise families. It is NOT finite-state exact with score_warp/linkwiggle cubic marginal-slope, because the multiplicative frailty breaks the polynomial kernel closure that the cubic cell derivatives require.

Returns an error if the combination is not exactly integrable.

pub fn validate_for_marginal_slope_typed( &self, ) -> Result<(), LognormalKernelError>

Typed variant of Self::validate_for_marginal_slope used internally; the String-returning entry point above is preserved as a one-line shim for external callers.

Trait Implementations

impl Clone for FrailtySpec

fn clone(&self) -> FrailtySpec

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for FrailtySpec

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

Formats the value using the given formatter.

impl Default for FrailtySpec

fn default() -> FrailtySpec

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for FrailtySpec

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for FrailtySpec

fn eq(&self, other: &FrailtySpec) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl Serialize for FrailtySpec

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for FrailtySpec