Struct MLCircuitOptimizer 

pub struct MLCircuitOptimizer {
    pub strategy: MLOptimizationStrategy,
    pub max_passes: usize,
    pub convergence_threshold: f64,
}

ML-driven circuit optimizer.

Uses a simple learned heuristic — prefer circuits with lower two-qubit gate count weighted by estimated depth — to select among candidate transformed circuits. The optimisation proceeds in passes; each pass tries commuting independent gates and removing adjacent inverse pairs.

Strategy

MLOptimizationStrategy	Scoring function
MinimizeGateCount	total gate count
MinimizeDepth	circuit depth (critical path)
BalancedOptimization	10 * two_q + total + depth

Fields

strategy: MLOptimizationStrategy

Optimisation strategy

max_passes: usize

Maximum number of optimisation passes

convergence_threshold: f64

Minimum score improvement to continue iterating (as a fraction)

Implementations

impl MLCircuitOptimizer

pub fn new(strategy: MLOptimizationStrategy) -> Self

Create a new optimizer with the given strategy and default pass limit.

pub fn with_passes(self, passes: usize) -> Self

Create a balanced optimizer with a custom pass limit.

pub fn optimize<const N: usize>( &self, circuit: &Circuit<N>, ) -> QuantRS2Result<Circuit<N>>

Optimise circuit and return the best circuit found together with its score.

The method uses graph-based topology analysis (SciRS2GraphTranspiler) to identify commuting gate pairs, then greedily reorders them to minimise the composite score under the chosen strategy.

pub fn score<const N: usize>(&self, circuit: &Circuit<N>) -> CircuitScore

Return the current score for a circuit without modifying it.