def test_cauchy_posterior_sampling(self):
"""
Test whether the posterior can successully be sampled from
"""
counts = np.array((10, 103))
biases = np.array((10.0, 30.0))
esimator = MultinomialBayesEstimator(counts=counts,
zetas=biases,
prior='cauchy',
location=0,
spread=100)
# Sample from the posterior
esimator.sample_posterior(nwalkers=4, nmoves=10)
def test_bias_initialization(self):
"""
Assess whether the biases are correctly read in and appropriately set to zero if no biases are applied.
"""
# Initialize the class with fake samples
counts = np.array((10, 103, 243, 82))
biases = np.array((0.0, 0.0, 0.0, 0.0))
esimator_biased = MultinomialBayesEstimator(counts=counts,
zetas=biases)
esimator_unbiased = MultinomialBayesEstimator(counts=counts)
assert np.all(
esimator_unbiased.free_energies == esimator_biased.free_energies)
def test_map_estimation(self):
"""
Test if the maximum aposteriori estimator works
"""
# Initialize the class with fake samples
counts = np.array((10, 103, 243, 82))
biases = np.array((10.0, 30.0, 43.0, 28.0))
esimator = MultinomialBayesEstimator(counts=counts,
zetas=biases,
prior='gaussian',
location=0,
spread=100)
map = esimator.map_estimator()
def test_map_estimation_stacked_inputs(self):
"""
Test if the maximum aposteriori estimator works when matrices are supplied for the counts and biases.
"""
# Initialize the class with fake samples
counts = np.array(((10, 103, 243, 82), (30, 11, 23, 72)))
biases = np.array(((10.0, 30.0, 43.0, 28.0), (4.0, 14.0, 36.0, 81.0)))
esimator = MultinomialBayesEstimator(counts=counts,
zetas=biases,
prior='gaussian',
location=0,
spread=100)
map = esimator.map_estimator()
def test_stacking_inputs(self):
"""
The class for Multinomial bayesian estimation should be able to handle histograms from simulations with
different applied biases, this is achieved with using matrices of histograms and biases.
"""
counts = np.array(((10, 103, 243, 82), (30, 11, 23, 72)))
biases = np.array(((10.0, 30.0, 43.0, 28.0), (4.0, 14.0, 36.0, 81.0)))
esimator = MultinomialBayesEstimator(counts=counts, zetas=biases)