prophesy.sampling package¶

Sampling is the package to get a rough image of the parameter space by considering single points in the parameter space.

Sampler: How to sample?¶

The basic interface for sampling is the sampler interface. Besides sampling on given points, it supports uniform sampling as some samplers might perform better.

prophesy.sampling.sampler module¶

class Sampler¶

Bases: object

Base class for performing sampling of given set of points

abstract perform_sampling(samplepoints, ensure_welldefinedness=False)¶

Given some parameter instantiations, perform sampling on these instantiations.

Parameters: samplepoints – An iterable yielding parameter instantiations
Returns: A collection with the results of these samples
Return type: InstantiationResultDict

perform_uniform_sampling(parameters, region, samples_per_dimension)¶

Samples a uniform grid of points.

Given a list of intervals (i.e., the first and last point; for each dimension, in order) and the number of samples per dimension, a uniformly-spaced grid of points (the cartesian product) is sampled.

Parameters

parameters – Parameters together with their region.
samples_per_dimension – In how many points should the region be divided.

prophesy.sampling.sampler_ratfunc module¶

class RatFuncSampling(ratfunc, parameters)¶

Bases: prophesy.sampling.sampler.Sampler

Simple sampler based on pycarl rational function

perform_sampling(samplepoints)¶

Parameters: samplepoints – iterable of SamplePoint
Returns
Return type: InstantiationResultDict

Sampling via probabilistic model checking¶

One can also sample via probabilistic model checkers, see prophesy.modelcheckers.pmc module.

Sampling: Where to sample?¶

To determine sample points based on the already sampled information, we provide sample generators.

prophesy.sampling.sample_generator module¶

class SampleGenerator(sampler, parameters, region, samples=None)¶

Bases: object

Class to generate samples given a sampler. SampleGenerator uses the iteration interface to do so.

prophesy.sampling.sampling_linear module¶

class LinearRefinement(sampler, parameters, region, samples, threshold)¶

Bases: prophesy.sampling.sample_generator.SampleGenerator

Based on an initial set of samples, refines the samples by means of linear interpolation to approximate the threshold

prophesy.sampling.sampling_uniform module¶

class UniformSampleGenerator(sampler, parameters, region, samples, samples_per_dimension)¶

Bases: prophesy.sampling.sample_generator.SampleGenerator

Generates a uniform grid of samples

High-level convenience functions¶

prophesy.sampling.sampling module¶

Helper module for simplified sampling access.

refine_samples(sampler, parameters, region, samples, iterations, threshold)¶

Refine samples over several iterations.

Parameters

sampler – The sampler to use
parameters – The parameters of the problem
samples – The already known samples
iterations – The number of iterations the refinement generator should be called.
threshold – The threshold value we are most interested in

Returns

uniform_samples(sampler, parameters, region, samples_per_dim)¶

Generate a uniform grid of samples.

Parameters

sampler – The sampler to use
parameters – The parameters in our problem
samples_per_dim – The number of samples per dimension

Returns