medeq.DVASampler#

class medeq.DVASampler(d, seed=None)[source]#

Bases: object

Parameter sampler that targets the most uncertain regions while maximising the area covered.

MED samplers return values between [0, 1), which can then be upscaled to individual parameter ranges.

Parameters
dint

Sampling dimensionality - i.e. number of parameters.

seedint, optional

Seed for deterministic random sampling.

Examples

Simple, seeded sample generation:

import medeq
sampler = medeq.DVASampler(3, seed=123)
sampler.sample(5, None)

If you have a MED object, you can pass it as a second parameter to the sample method to specifically target high-uncertainty regions:

import medeq
parameters = medeq.create_parameters(...)
med = medeq.MED(parameters, ...)

sampler = medeq.DVASsampler(3)
sampler.sample(5, med)
__init__(d, seed=None)[source]#

Methods

__init__(d[, seed])

cost(x, med)

sample(n, med)

sample(n, med)[source]#
cost(x, med)[source]#