def setUp(self):
self.cov = [[1, 0, 0], [0, 4, 0], [0, 0, 9]]
self.sigma = [1, 2, 3]
self.mean_1 = [10, 11, 12]
self.mean_2 = [20, 21, 22]
self.likelihood = AnalyticalMultidimensionalBimodalCovariantGaussian(
mean_1=self.mean_1, mean_2=self.mean_2, cov=self.cov
)
for i, search_parameter_key in enumerate(result.search_parameter_keys):
logger.info(search_parameter_key)
logger.info('Expected posterior standard deviation: ' +
str(likelihood.sigma[i]))
logger.info('Sampled posterior standard deviation: ' +
str(sampled_std[i]))
# BIMODAL distribution
label = "multidim_gaussian_bimodal"
dim = len(cov[0])
mean_1 = 4 * np.sqrt(np.diag(cov))
mean_2 = -4 * np.sqrt(np.diag(cov))
likelihood = AnalyticalMultidimensionalBimodalCovariantGaussian(
mean_1, mean_2, cov)
priors = bilby.core.prior.PriorDict()
priors.update({
"x{0}".format(i): bilby.core.prior.Uniform(-5, 5, "x{0}".format(i))
for i in range(dim)
})
result = bilby.run_sampler(likelihood=likelihood,
priors=priors,
sampler="dynesty",
outdir=outdir,
label=label,
check_point_plot=True,
resume=True)
result.plot_corner(
parameters={"x{0}".format(i): mean_1[i]
def test_log_likelihood(self):
likelihood = AnalyticalMultidimensionalBimodalCovariantGaussian(
mean_1=[0], mean_2=[0], cov=[1]
)
self.assertEqual(-np.log(2 * np.pi) / 2, likelihood.log_likelihood())