Tools¶

Resample particles¶

pocomc.tools.systematic_resample(size: ndarray, weights: ndarray, random_state: int = None)¶

Resample a new set of points from the weighted set of inputs such that they all have equal weight.

Parameters:

size (int) – Number of samples to draw.
weights (~numpy.ndarray with shape (nsamples,)) – Corresponding weight of each sample.
random_state (int, optional) – Random seed.

Returns:

indeces – Indices of the resampled array.

Return type:

~numpy.ndarray with shape (nsamples,)

Examples

>>> x = np.array([[1., 1.], [2., 2.], [3., 3.], [4., 4.]])
>>> w = np.array([0.6, 0.2, 0.15, 0.05])
>>> systematic_resample(4, w)
array([0, 0, 0, 2])

Notes

Implements the systematic resampling method.

pocomc.tools.compute_ess(logw: ndarray)¶

Compute effective sample size (per centage).

Parameters:: logw (np.ndarray) – Log-weights.
Returns:: ess – Effective sample size divided by actual number of particles (between 0 and 1)
Return type:: float

pocomc.tools.effective_sample_size(weights)¶

Compute effective sample size (ESS).

pocomc.tools.unique_sample_size(weights, k=None)¶

Compute unique sample size (ESS).

Parameters:

Returns:

uss – Unique sample size.

Return type:

float

class pocomc.tools.ProgressBar(show: bool = True, initial=0)¶

Progress bar class.

Parameters:: show (bool) – Whether or not to show a progress bar. Default is True.

update_stats(info)¶

Update shown stats.

class pocomc.tools.FunctionWrapper(f, args, kwargs)¶

Make the log-likelihood or log-prior function pickleable when args or kwargs are also included.

Parameters: