Struct InterventionOverlay 

pub struct InterventionOverlay {
    pub hard_set: Vec<Option<f64>>,
    pub shifts: Vec<f64>,
    pub stochastic: Vec<Option<StochasticPolicy>>,
    pub soft: Vec<Option<MechanismOverride>>,
    pub active: Vec<bool>,
}

Compact overlay describing how sampling differs from the observational plan.

The underlying CompiledCausalModel is never cloned; overlays are applied during ancestral sampling.

Fields

hard_set: Vec<Option<f64>>

Per-node hard sets (dense index → value), None = not hard-set.

shifts: Vec<f64>

Per-node additive shifts.

stochastic: Vec<Option<StochasticPolicy>>

Per-node stochastic policies.

soft: Vec<Option<MechanismOverride>>

Per-node soft mechanism overrides.

active: Vec<bool>

Optional temporal activation mask per node (true = active at current step).

Implementations

impl InterventionOverlay

pub fn observational(n_nodes: usize) -> Self

Empty overlay (observational) for n_nodes.

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

Reject a node carrying both a hard set and an additive shift.

A hard set replaces a node’s assignment outright, so an additive shift on the same node has nothing well-defined to add to: the mechanism it would shift is exactly the one the set discarded. Rather than pick a winner silently, both overlay construction and the overlay-accepting samplers refuse the pair and make the caller say which they meant.

Called for you by Self::from_interventions and Self::from_sequence_at; call it yourself if you populate the public fields directly.

Errors

ModelError::Unsupported naming the first offending dense node index.

pub fn is_empty(&self) -> bool

Whether any node is intervened.

pub fn from_interventions( model: &CompiledCausalModel, interventions: &[Intervention], ) -> Result<Self, ModelError>

Compile interventions against a model (simultaneous / single-step).

Errors

Unknown variables, invalid interventions, or a node both set and shifted (see Self::validate).

pub fn from_sequence_at( model: &CompiledCausalModel, seq: &InterventionSequence, t: i32, ) -> Result<Self, ModelError>

Overlay for a temporal sequence at discrete step t.

Errors

Invalid sequence, unknown variables, or a node both set and shifted at t (see Self::validate).

Trait Implementations

impl Clone for InterventionOverlay

fn clone(&self) -> InterventionOverlay

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for InterventionOverlay

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

Formats the value using the given formatter.

impl Default for InterventionOverlay

fn default() -> InterventionOverlay

Returns the “default value” for a type.