Open OAGr opened 1 year ago
My guess is that this is going to come from calculating beta samples as:
sample_beta(a,b) = sample_gamma(a)/(sample_gamma(a) + sample_gamma(b))
and then getting a division by zero due to luck of the draw.
Added: Mmh, not finding this in stdlib, I was probably wrong.
beta(1, 0.2)