def setUp(self):
self.date = str(datetime.date.today())
self.folder = os.path.join("tests", "test06_%s" % self.date)
self.number = 30
self.custom_command = "run -N %d -p test.param -o %s" % (self.number, self.folder)
self.cosmo, self.data, self.command_line, _ = initialise(self.custom_command)
sampler.run(self.cosmo, self.data, self.command_line)
def test_block_behaviour(self):
"""
Are the mcmc arguments well grouped by block?
"""
# The block separation must have selected three blocks, of size
# respectively 2, 1, and 1 (the two test nuisance likelihoods are
# sharing a nuisance parameter)
self.assertEqual(self.data.block_parameters, [2, 3, 4], "The block selection went wrong")
# By default, the over-sampling should be set to 1
self.assertEqual(self.data.over_sampling, [1, 1, 1], "The default over sampling is messed up")
# For the sake of the test, set them to something else
self.data.over_sampling = [1, 3, 5]
self.data.assign_over_sampling_indices()
self.assertEqual(self.data.over_sampling_indices, [0, 1, 2, 2, 2, 3, 3, 3, 3, 3])
# Run the sampler
sampler.run(self.cosmo, self.data, self.command_line)
# Verify that the output is well behaved
output_path = os.path.join(self.folder, "%s_%d__1.txt" % (self.date, self.number))
with open(output_path, "r") as output_file:
chain = np.loadtxt(output_file)
for index in range(2, 4):
self.assertLessEqual(
len(np.unique(chain[:, index])), float(self.number) / 10 * 2, "%s" % np.unique(chain[:, index])
)
self.assertLessEqual(len(np.unique(chain[:, 3])), float(self.number) / 10 * 3)
self.assertLessEqual(len(np.unique(chain[:, 4])), float(self.number) / 10 * 5)
def setUp(self):
self.date = str(datetime.date.today())
self.folder = os.path.join('tests', 'test06_%s' % self.date)
self.number = 30
self.custom_command = ('run -N %d -p test.param -o %s' %
(self.number, self.folder))
self.cosmo, self.data, self.command_line, _ = initialise(
self.custom_command)
sampler.run(self.cosmo, self.data, self.command_line)
def test_cosmological_arguments(self):
"""
Are the cosmological arguments well set?
"""
# After the initialisation, the dictionnary mcmc_parameters should
# contain five Parameter instances
self.assertEqual(len(self.data.mcmc_parameters), 5, "mcmc_parameters badly defined")
# Run the sampler
sampler.run(self.cosmo, self.data, self.command_line)
self.assertEqual(
self.data.mcmc_parameters["omega_b"]["current"] * self.data.mcmc_parameters["omega_b"]["scale"],
self.data.cosmo_arguments["omega_b"],
"the cosmo_arguments dict was not updated properly",
)
def test_cosmological_arguments(self):
"""
Are the cosmological arguments well set?
"""
# After the initialisation, the dictionnary mcmc_parameters should
# contain five Parameter instances
self.assertEqual(len(self.data.mcmc_parameters), 5,
"mcmc_parameters badly defined")
# Run the sampler
sampler.run(self.cosmo, self.data, self.command_line)
self.assertEqual(
self.data.mcmc_parameters['omega_b']['current'] *
self.data.mcmc_parameters['omega_b']['scale'],
self.data.cosmo_arguments['omega_b'],
"the cosmo_arguments dict was not updated properly")
def test_block_behaviour(self):
"""
Are the mcmc arguments well grouped by block?
"""
# The block separation must have selected three blocks, of size
# respectively 2, 1, and 1 (the two test nuisance likelihoods are
# sharing a nuisance parameter)
self.assertEqual(
self.data.block_parameters,
[2, 3, 4],
"The block selection went wrong")
# By default, the over-sampling should be set to 1
self.assertEqual(
self.data.over_sampling,
[1, 1, 1],
"The default over sampling is messed up")
# For the sake of the test, set them to something else
self.data.over_sampling = [1, 3, 5]
self.data.assign_over_sampling_indices()
self.assertEqual(
self.data.over_sampling_indices,
[0, 1, 2, 2, 2, 3, 3, 3, 3, 3])
# Run the sampler
sampler.run(self.cosmo, self.data, self.command_line)
# Verify that the output is well behaved
output_path = os.path.join(
self.folder, '%s_%d__1.txt' % (self.date, self.number))
with open(output_path, 'r') as output_file:
chain = np.loadtxt(output_file)
for index in range(2, 4):
self.assertLessEqual(
len(np.unique(chain[:, index])),
float(self.number)/10*2,
"%s" % np.unique(chain[:, index]))
self.assertLessEqual(
len(np.unique(chain[:, 3])),
float(self.number)/10*3)
self.assertLessEqual(
len(np.unique(chain[:, 4])),
float(self.number)/10*5)
def test_behaviour(self):
"""Check Nested Sampling global behaviour"""
self.assertTrue(os.path.exists(os.path.join(self.folder, "NS")))
sampler.run(self.cosmo, self.data, self.command_line)
def test_behaviour(self):
"""Check PolyChord global behaviour"""
self.assertTrue(os.path.exists(
os.path.join(self.folder, 'PC')))
sampler.run(self.cosmo, self.data, self.command_line)
def test_behaviour(self):
"""Check Nested Sampling global behaviour"""
self.assertTrue(os.path.exists(os.path.join(self.folder, 'NS')))
sampler.run(self.cosmo, self.data, self.command_line)