def prior_eval(self):
""" Returns the prior evaluator class initialized with a set of
pre-defined distributions for each parameters.
"""
parameters, values = zip(*self.parameters.items())
prior_dists = []
for param, val in zip(parameters, values):
if param in ["mass1", "mass2"]:
dist = distributions.Uniform(**{param: (6, 50)})
elif param in ["inclination", "dec"]:
dist = distributions.SinAngle(**{param: None})
elif param in ["polarization", "ra", "coa_phase"]:
dist = distributions.Uniform(**{param: (0, 2 * 3.1415)})
elif param in ["distance"]:
dist = distributions.UniformRadius(distance=(val - 100,
val + 300))
elif param in [
"spin1x", "spin1y", "spin1z", "spin2x", "spin2y", "spin2z"
]:
dist = distributions.Uniform(**{param: (-0.1, 0.1)})
elif param in ["tc"]:
dist = distributions.Uniform(tc=(val - 0.2, val + 0.2))
else:
raise KeyError("Do not recognize parameter %s" % param)
prior_dists.append(dist)
return distributions.JointDistribution(parameters, *prior_dists)
def full(self, fd_waveform, fd_waveform_generator, zdhp_psd, request):
marg_prior = [distributions.Uniform(distance=(50, 5000))]
model = self.TEST_CLASS(
['tc'], fd_waveform, fd_waveform_generator, self.fmin,
psds={ifo: zdhp_psd for ifo in self.ifos},
distance_marginalization=True, marg_prior=marg_prior)
return self.CALL_CLASS(model, self.DEFAULT_CALLSTAT,
return_all_stats=False)
def simple(self, random_data, fd_waveform_generator):
marg_prior = [distributions.Uniform(distance=(50, 5000))]
data = {ifo: random_data for ifo in self.ifos}
model = self.TEST_CLASS([], data, fd_waveform_generator,
f_lower=self.fmin,
distance_marginalization=True,
marg_prior=marg_prior)
return self.CALL_CLASS(model, self.DEFAULT_CALLSTAT)
def test_likelihood_evaluator_init(self):
# data args
seglen = 4
sample_rate = 2048
N = seglen * sample_rate / 2 + 1
fmin = 30.
# setup waveform generator and signal parameters
m1, m2, s1z, s2z, tsig = 38.6, 29.3, 0., 0., 3.1
ra, dec, pol, dist = 1.37, -1.26, 2.76, 3 * 500.
gen = generator.FDomainDetFrameGenerator(generator.FDomainCBCGenerator,
0.,
variable_args=["tc"],
detectors=["H1", "L1"],
delta_f=1. / seglen,
f_lower=fmin,
approximant="TaylorF2",
mass1=m1,
mass2=m2,
spin1z=s1z,
spin2z=s2z,
ra=ra,
dec=dec,
polarization=pol,
distance=dist)
signal = gen.generate(tc=tsig)
# get PSDs
psd = pypsd.aLIGOZeroDetHighPower(N, 1. / seglen, 20.)
psds = {"H1": psd, "L1": psd}
# get a prior evaluator
uniform_prior = distributions.Uniform(tc=(tsig - 0.2, tsig + 0.2))
prior_eval = inference.prior.PriorEvaluator(["tc"], uniform_prior)
# setup likelihood evaluator
likelihood_eval = inference.GaussianLikelihood(gen,
signal,
fmin,
psds=psds,
prior=prior_eval,
return_meta=False)
def model(self, fd_waveform, fd_waveform_generator, prior_eval, zdhp_psd,
request):
eval_class = models.models[request.param]
if request.param == "marginalized_gaussian_noise":
# Test that the sampler works for time_marginalization
model = eval_class(
fd_waveform_generator.variable_args,
fd_waveform,
fd_waveform_generator,
self.fmin,
psds={ifo: zdhp_psd
for ifo in self.ifos},
prior=prior_eval,
time_marginalization=True,
marg_prior=[distributions.Uniform(time=(0, 10000))])
else:
model = eval_class(fd_waveform_generator.variable_args,
fd_waveform,
fd_waveform_generator,
self.fmin,
psds={ifo: zdhp_psd
for ifo in self.ifos},
prior=prior_eval)
return models.CallModel(model, "logposterior", return_all_stats=False)
import matplotlib.pyplot as plt
from pycbc import distributions
# Create a mass distribution object that is uniform between 0.5 and 1.5
# solar masses.
mass1_distribution = distributions.Uniform(mass1=(0.5, 1.5))
# Take 100000 random variable samples from this uniform mass distribution.
mass1_samples = mass1_distribution.rvs(size=1000000)
# Draw another distribution that is Gaussian between 0.5 and 1.5 solar masses
# with a mean of 1.2 solar masses and a standard deviation of 0.15 solar
# masses. Gaussian takes the variance as an input so square the standard
# deviation.
variance = 0.15 * 0.15
mass2_gaussian = distributions.Gaussian(mass2=(0.5, 1.5),
mass2_mean=1.2,
mass2_var=variance)
# Take 100000 random variable samples from this gaussian mass distribution.
mass2_samples = mass2_gaussian.rvs(size=1000000)
# We can make pairs of distributions together, instead of apart.
two_mass_distributions = distributions.Uniform(mass3=(1.6, 3.0),
mass4=(1.6, 3.0))
two_mass_samples = two_mass_distributions.rvs(size=1000000)
# Choose 50 bins for the histogram subplots.
n_bins = 50
# Plot histograms of samples in subplots
fig, axes = plt.subplots(nrows=2, ncols=2)