You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm especially interested in the implementation of Distribution<num::Ratio> for Standard and the UniformSampler backend to use gen_range() on custom types that wrap num::Ratio
The text was updated successfully, but these errors were encountered:
Do you know how to actually achieve a uniform sample with a ratio? I think that's not easy, unless you just pick a specific denominator and then uniformly randomize the numerator. Is there a better way?
I agree with cuviper -- I'm not 100% sure uniform randomness inside a range for Ratio is something that can be easily quantified. Most people will have their own idea of how the algorithm should work, and mathematically it's not really well-founded since all definitions of probability rely on real numbers, which can't easily be approximated with ratios. In the absolute case, if you wanted to generate a random Ratio<BigInt>, I don't think that it'd be desirable that there's a chance that the numerator and denominator can be arbitrarily large.
The only solution that makes sense to me is to just randomly generate a float and then convert that to a ratio to whatever precision you want. Which, can easily be accomplished by doing… just that.
I'm especially interested in the implementation of
Distribution<num::Ratio> for Standard
and theUniformSampler
backend to usegen_range()
on custom types that wrapnum::Ratio
The text was updated successfully, but these errors were encountered: