Enum ResponseDegeneracyKind 

pub enum ResponseDegeneracyKind {
    BinomialAllZeros,
    BinomialAllOnes,
    PoissonAllZeros,
    NegativeBinomialAllZeros,
    GaussianNearConstant {
        sample_sd: f64,
        min_sd: f64,
    },
}

Classifier for a ResponseDegeneracy. Each variant carries the family- specific evidence the caller needs to format a useful message without having to re-derive the diagnostic.

Variants

BinomialAllZeros

Bernoulli / Binomial response with every observed value equal to 0.

BinomialAllOnes

Bernoulli / Binomial response with every observed value equal to 1.

PoissonAllZeros

Poisson response with no positive counts. The log-rate likelihood has its supremum at η = −∞, not at a finite fitted mode.

NegativeBinomialAllZeros

Negative-Binomial response with no positive counts. As for Poisson, the log-rate likelihood has no finite optimum or finite posterior moments.

GaussianNearConstant

Gaussian response that is effectively constant in f64 arithmetic (sample standard deviation at or below GAUSSIAN_MIN_SAMPLE_SD). The marginal REML log-likelihood −n/2·log σ² diverges to +∞ as the fitted scale σ → 0, so every outer evaluation rejects with a non-finite score. Carries the observed sample_sd and the min_sd threshold so the message can quote both verbatim (#332).

Fields

sample_sd: f64

The two-pass, mean-centred sample standard deviation of the response.

min_sd: f64

The rejection threshold (GAUSSIAN_MIN_SAMPLE_SD).

Trait Implementations

impl Clone for ResponseDegeneracyKind

fn clone(&self) -> ResponseDegeneracyKind

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ResponseDegeneracyKind

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

Formats the value using the given formatter.