resample

ssapy.utils.resample(particles, ln_weights, obs_times=None, pod=False)[source][source]

Resample particles to achieve more uniform weights. This produces the same number of particles as are input (i.n. dimension matches “particles”) From “Bayesian Multiple Target Tracking” (2nd ed.) p. 100

Parameters:
  • particles (numpy.ndarray) – 2D array of particles. Each row is an nD particles

  • ln_weights (numpy.ndarray) – 1D array of particle log-weights. Length should match number of particles

  • obs_times ([type], optional) – Time at which observation occured. This is only used if pod=True, defaults to None

  • pod (bool, optional) – Specify whether this is the special case of regularizing orbit parameters, defaults to False

Returns:

Resampled particles and weights (same number as input)

Return type:

(numpy.ndarray, numpy.ndarray)