sequince.steps.resampling

Attributes

logger

Classes

ResamplingOutput	Fixed-structure resampling output with a static scheme label.
ResamplingStep	Base class for resampling steps with an optional ESS trigger.
MultinomialResampling	Draw ancestors with replacement according to the weights.
StratifiedResampling	Stratified resampling for reduced variance.
SystematicResampling	Systematic resampling for reduced variance.

Module Contents

sequince.steps.resampling.logger

class sequince.steps.resampling.ResamplingOutput

Bases: collections.abc.Mapping[str, Any]

Fixed-structure resampling output with a static scheme label.

ess: float | sequince.typing.Array

resampled: bool | sequince.typing.Array

ancestors: sequince.typing.Array

resampling_scheme: str

class sequince.steps.resampling.ResamplingStep(*, ess_threshold=0.5, always_resample=False, n_samples=None, resample_at_end=True)

Bases: sequince.steps.base.SMCStep

Base class for resampling steps with an optional ESS trigger.

Parameters:

• ess_threshold (float | None) – Resample when the effective sample size (ESS) ratio falls below this threshold. If None (default) resampling is never triggered based on ESS.

• always_resample (bool) – When True always resample regardless of ESS (overrides ess_threshold).

• n_samples (int | None) – Number of particles to draw when resampling. If None, draw the same number as are present in the input state.

• resample_at_end (bool) – If True (default) will always resample when the inverse temperature β reaches 1.0, regardless of ESS. If False, the final resampling may be skipped if always_resample is False and the ESS ratio is above ess_threshold.

resampling_scheme: ClassVar[str] = 'unknown'

apply(state, context, rng)

Parameters:

• state (sequince.state.ParticleState)

• context (sequince.steps.base.StepContext)

• rng (orng.RandomGenerator)

Return type:

tuple[sequince.state.ParticleState, sequince.steps.base.StepContext, ResamplingOutput]

initialize_functional_metadata(state, metadata)

Initialize fixed-shape resampling metadata.

Parameters:

• state (sequince.state.ParticleState)

• metadata (collections.abc.Mapping[str, Any])

Return type:

collections.abc.Mapping[str, Any]

apply_functional(state, context, random_state)

Resample while explicitly threading RNG state.

The initial compiled subset requires unconditional resampling. A traced ESS threshold needs a conditional RNG transition and is not supported yet.

Parameters:

• state (sequince.state.ParticleState)

• context (sequince.steps.base.StepContext)

• random_state (object)

Return type:

tuple[sequince.state.ParticleState, sequince.steps.base.StepContext, object, ResamplingOutput]

class sequince.steps.resampling.MultinomialResampling(*, ess_threshold=0.5, always_resample=False, n_samples=None, resample_at_end=True)

Bases: ResamplingStep

Draw ancestors with replacement according to the weights.

Parameters:

• ess_threshold (float | None)

• always_resample (bool)

• n_samples (int | None)

• resample_at_end (bool)

resampling_scheme = 'multinomial'

class sequince.steps.resampling.StratifiedResampling(*, ess_threshold=0.5, always_resample=False, n_samples=None, resample_at_end=True)

Bases: ResamplingStep

Stratified resampling for reduced variance.

Parameters:

• ess_threshold (float | None)

• always_resample (bool)

• n_samples (int | None)

• resample_at_end (bool)

resampling_scheme = 'stratified'

class sequince.steps.resampling.SystematicResampling(*, ess_threshold=0.5, always_resample=False, n_samples=None, resample_at_end=True)

Bases: ResamplingStep

Systematic resampling for reduced variance.

Parameters:

• ess_threshold (float | None)

• always_resample (bool)

• n_samples (int | None)

• resample_at_end (bool)

resampling_scheme = 'systematic'