a = normal(5,2) .+ uniform(8,30)
b = SampleSet.fromDist(a)
Sampling non-normalized pointset distributions is meaningless and easy to mess up. Arguably, these really aren't distributions, and should be appropriately named.
Right now we have errors that many mx() distributions and similar are close to being normalized but not quite there. Ideally, we should fix this separately, or, if we decide that they're "close enough", then we should sample from them here.
berekuk
commented
1 year ago
Sampling non-normalized pointset distributions is meaningless and easy to mess up. Arguably, these really aren't distributions, and should be appropriately named.
Right now we have errors that many mx() distributions and similar are close to being normalized but not quite there. Ideally, we should fix this separately, or, if we decide that they're "close enough", then we should sample from them here.