Struct TabularCPD 

pub struct TabularCPD {
    pub node_idx: usize,
    pub n_values: usize,
    pub parent_card: Vec<usize>,
    pub parent_indices: Vec<usize>,
    pub table: Vec<Vec<f64>>,
    /* private fields */
}

Discrete CPD stored as a conditional probability table (CPT).

The table is indexed by the combined parent configuration. values[row][val] = P(X = val | parent_config = row).

Row indexing follows column-major (rightmost parent cycles fastest), matching pgmpy convention: row = Sum_i (parent_values[i] * stride[i]) where stride[i] = product of cardinalities of parents to the right.

Fields

node_idx: usize

Index of the node this CPD belongs to.

n_values: usize

Number of values (cardinality) of this node.

parent_card: Vec<usize>

Cardinalities of parent nodes.

parent_indices: Vec<usize>

Parent node indices.

table: Vec<Vec<f64>>

Table: table[row] = probability distribution over n_values states. Length = product(parent_card), each inner vec has length n_values.

Implementations

impl TabularCPD

pub fn new( node_idx: usize, n_values: usize, parent_indices: Vec<usize>, parent_card: Vec<usize>, values: Vec<Vec<f64>>, ) -> Result<Self, StatsError>

Create a new TabularCPD.

Arguments

• node_idx: Index of the node.
• n_values: Cardinality of this node.
• parent_indices: Indices of parent nodes.
• parent_card: Cardinalities of each parent (same order as parent_indices).
• values: Probability table. Each row is a probability distribution. If there are no parents, values should have exactly one row.

pub fn row_index(&self, parent_values: &[usize]) -> Result<usize, StatsError>

Compute the row index for a given parent configuration.

pub fn distribution( &self, parent_values: &[usize], ) -> Result<&[f64], StatsError>

Return the full conditional distribution P(X | parent_values).

Trait Implementations

impl CPD for TabularCPD

fn node(&self) -> usize

Name of the node this CPD belongs to.

fn prob(&self, value: usize, parent_values: &[usize]) -> f64

Evaluate the (unnormalised) probability (or density) of value given parent_values.

fn cardinality(&self) -> usize

Number of states (cardinality) for discrete nodes; 0 for continuous.

fn parent_indices(&self) -> &[usize]

Parent node indices.

fn is_continuous(&self) -> bool

Whether this CPD is continuous.

fn log_prob(&self, value: usize, parent_values: &[usize]) -> f64

Log-probability: default implementation wraps prob.

impl Clone for TabularCPD

fn clone(&self) -> TabularCPD

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TabularCPD

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

Formats the value using the given formatter.