sequince.steps.mutation.adaptive.policies.acceptance

Choose mutation budgets from observed acceptance probabilities.

Classes

AcceptanceRatePolicy

Target an expected number of accepted moves per particle.

Module Contents

class sequince.steps.mutation.adaptive.policies.acceptance.AcceptanceRatePolicy(target_steps, *, min_steps, max_steps, min_acceptance=1e-06)

Target an expected number of accepted moves per particle.

Propose ceil(target_steps / max(acceptance, min_acceptance)) for the next iteration. min_acceptance (default 1e-6) keeps the budget finite for zero or very small acceptance rates. Every returned budget is clipped to the required min_steps and max_steps bounds. A block with no attempted moves supplies no acceptance estimate, so its current budget is retained (or increased to one when zero) to allow a subsequent measurement.

Acceptance is estimated from the complete block’s accepted and attempted counts, including fractional expected counts from recycling algorithms.

Parameters:

• target_steps (float)

• min_steps (int)

• max_steps (int)

• min_acceptance (float)

min_steps

max_steps

target_steps

min_acceptance

initialise(*, initial_steps, context)

Parameters:

• initial_steps (int)

• context (sequince.steps.context.StepContext)

Return type:

int

update(*, current_steps, metrics, context)

Parameters:

• current_steps (int)

• metrics (sequince.steps.mutation.metrics.MutationMetricsOutput)

• context (sequince.steps.context.StepContext)

Return type:

int