LinearizationImportanceDistribution

class probnum.filtsmooth.particle.LinearizationImportanceDistribution(dynamics_model, linearization_strategy)

Bases: probnum.filtsmooth.particle.ImportanceDistribution

Local linearisation importance distribution.

Methods Summary

from_ekf(dynamics_model[, ...])

Create an importance distribution that uses the EKF for proposals.

from_ukf(dynamics_model[, spread, priorpar, ...])

Create an importance distribution that uses the UKF for proposals.

generate_importance_rv(particle, data, t[, ...])

Generate an importance distribution.

log_correction_factor(proposal_state, ...)

Logarithmic correction factor.

process_initrv_with_data(initrv, data, t[, ...])

Process the initial random variable based on data.

Methods Documentation

classmethod from_ekf(dynamics_model, forward_implementation='classic', backward_implementation='classic')[source]

Create an importance distribution that uses the EKF for proposals.

classmethod from_ukf(dynamics_model, spread=0.0001, priorpar=2.0, special_scale=0.0)[source]

Create an importance distribution that uses the UKF for proposals.

generate_importance_rv(particle, data, t, dt=None, measurement_model=None)[source]

Generate an importance distribution.

log_correction_factor(proposal_state, importance_rv, dynamics_rv, old_weight)

Logarithmic correction factor.

Return type

float

process_initrv_with_data(initrv, data, t, measurement_model=None)[source]

Process the initial random variable based on data.