Struct Dirichlet 

pub struct Dirichlet { /* private fields */ }

Dirichlet distribution, conjugate prior for multinomial likelihood.

After observing counts c₁, ..., cₖ, the posterior is Dirichlet(α₁ + c₁, ..., αₖ + cₖ).

Implementations

impl Dirichlet

pub fn new(alpha: Vec<f64>) -> Result<Self>

Create a Dirichlet distribution with concentration parameters alpha.

All elements must be positive and the vector must have at least 2 elements.

pub fn symmetric(k: usize, alpha: f64) -> Result<Self>

Create a symmetric Dirichlet with k categories, each with concentration alpha.

pub fn alpha(&self) -> &[f64]

Concentration parameters.

pub fn update_multinomial(&self, counts: &[u64]) -> Self

Posterior after observing multinomial counts.

Panics

Panics if counts.len() != self.alpha.len().

pub fn mean(&self) -> Vec<f64>

Expected value (mean) of the Dirichlet: E[Xᵢ] = αᵢ / Σα.

pub fn variance(&self) -> Vec<f64>

Variance of each component: Var[Xᵢ] = αᵢ(Σα − αᵢ) / (Σα² (Σα + 1)).

pub fn ln_pdf(&self, x: &[f64]) -> Result<f64>

Log-PDF of the Dirichlet at point x.

Errors

Returns an error if x.len() != alpha.len() or if values don’t sum to approximately 1.

Trait Implementations

impl Clone for Dirichlet

fn clone(&self) -> Dirichlet

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for Dirichlet

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

Formats the value using the given formatter.