charles-river-analytics/figaro

Why can't importance sampling for Normal distribution be consistent?

bjenkinsgit opened this issue · 0 comments

Given the following in Mathematica notation:
Probability[x < 15.2, x \[Distributed] NormalDistribution[16, 0.3]]

Always gives the same answer. Figaro does not. This at least IS called out in the documentation (pg. 40 in the book) that says this is based on a randomized algorithm. I was just wondering why Mathematica is able to give a consistent answer but Figaro is not. Is this for performance?