Latin Hypercubic sampling

class stats_arrays.LatinHypercubeRNG(params, seed=None, samples=10, **kwargs)

A random number generator that pre-calculates a sample space to draw from.

Inputs

• params : A Parameter array which gives parameters for distributions (one distribution per row).
• seed : An integer (or array of integers) to seed the NumPy random number generator.
• samples : An integer number of samples to construct for each distribution.

__init__(params, seed=None, samples=10, **kwargs)

x.__init__(…) initializes x; see help(type(x)) for signature

build_hypercube()

Build an array, of shape self.length rows by self.samples columns, which contains the sample space to be drawn from when doing Latin Hypercubic sampling.

Each row represents a different data point and distribution. The final sample space is self.hypercube. All distributions from uncertainty_choices are usable, and bounded distributions are also fine.

Builds

self.hypercube : Numpy array with dimensions self.length by self.samples.

next()

Draw directly from pre-computed sample space.