Struct QuantumSampler 

pub struct QuantumSampler { /* private fields */ }

Shot-based quantum sampler

Implementations

impl QuantumSampler

pub fn new(config: SamplingConfig) -> Self

Create new sampler with configuration

pub fn sample_state(&mut self, state: &Array1<Complex64>) -> Result<ShotResult>

Sample measurements from a quantum state

pub fn sample_state_with_noise( &mut self, state: &Array1<Complex64>, noise_model: &dyn NoiseModel, ) -> Result<ShotResult>

Sample measurements from a state with noise

pub fn sample_expectation( &mut self, state: &Array1<Complex64>, observable: &PauliOperatorSum, ) -> Result<ExpectationResult>

Sample expectation value of an observable

pub fn sample_multinomial( &mut self, probabilities: &[f64], n_samples: usize, ) -> Result<Vec<usize>>

Multinomial sampling: draw n_samples i.i.d. samples from a discrete probability distribution.

The result is a count vector of the same length as probabilities, where counts[i] is the number of times outcome i was drawn. Total counts sum to n_samples.

The implementation uses the inverse-CDF (Alias/sequential) method which is O(k · n_samples) in the number of outcomes k, but correct and deterministic given the sampler’s current RNG state.

pub fn importance_sampling( &mut self, target_probs: &[f64], proposal_probs: &[f64], f_values: &[f64], n_samples: usize, ) -> Result<ImportanceSamplingResult>

Importance sampling: estimate the expectation E_p[f] using samples drawn from a proposal distribution q.

Given a target distribution target_probs (unnormalised is fine) and a proposal distribution proposal_probs (must be strictly positive wherever target_probs is non-zero), this method:

1. Draws n_samples indices from proposal_probs.
2. Evaluates the function values f_values[i] at each outcome i.
3. Computes the self-normalised importance-sampling estimator Ê = Σ w̃_i f_i where w̃_i = w_i / Σ w_j and w_i = p_i / q_i.

Returns (estimate, effective_sample_size). The effective sample size (ESS) measures how many i.i.d. samples from target_probs the weighted sample is worth: ESS = (Σ w̃_i)² / Σ w̃_i².

Errors

Returns an error if the distributions have different lengths, if proposal_probs has a zero entry where target_probs is non-zero, or if either distribution is empty.

pub fn estimate_convergence( &mut self, state: &Array1<Complex64>, observable: &PauliOperatorSum, ) -> Result<ConvergenceResult>

Estimate convergence of sampling