|View source on GitHub|
Stratified resampler for sequential Monte Carlo.
tfp.experimental.mcmc.resample_stratified( log_probs, event_size, sample_shape, seed=None, name=None )
The value returned from this algorithm is similar to sampling with
expanded_sample_shape = tf.concat([[event_size], sample_shape]), axis=-1) tfd.Categorical(logits=log_probs).sample(expanded_sample_shape)`
but with values sorted along the first axis. It can be considered to be
sampling events made up of a length-
event_size vector of draws from
Categorical distribution. However, although the elements of
this event have the appropriate marginal distribution, they are not
independent of each other. Instead they are drawn using a low variance
stratified sampling method suitable for use with Sequential Monte
The sortedness is an unintended side effect of the algorithm that is
harmless in the context of simple SMC algorithms.
This function is based on Algorithm #1 in the paper [Maskell et al. (2006)].
||A tensor-valued batch of discrete log probability distributions.|
||the dimension of the vector considered a single draw.|
||a tensor of samples.|
: S. Maskell, B. Alun-Jones and M. Macleod. A Single Instruction Multiple Data Particle Filter. In 2006 IEEE Nonlinear Statistical Signal Processing Workshop. http://people.ds.cam.ac.uk/fanf2/hermes/doc/antiforgery/stats.pdf