Hierarchical mass quadrupole distribution likelihood fails when evaluating noise_log_likelihood for power law distribution
When trying to evaluate the noise_log_likelihood
of a MassQuadrupoleDistributionLikelihood
it sets all the distribution hyperparameters to zero. I expects that when this happens the associated distribution will return -∞ when the hyperparameter are zero. However, for the PowerLawDistribution
distribution it instead raises and error due to the maximum and minimum values both being set to zero. This can either be fixed in the MassQuadrupoleDistributionLikelihood.noise_log_likelihood()
method by wrapping the call to log_likelihood
in a try... except...
or in the PowerLawDistribution.log_pdf
it should explicitly return -∞ if maximum <= minimum
.