I was having quite some trouble trying to calculate the conjunction of two probabilities that are based on the same probabilities. Vastly simplified: What is the chance of when two dice are thrown, the higher of the two being above 3, and the lower of the two being below 4. Do you deal with anything like that? I imagine one would lift the sort, but it remains somewhat unclear how this actually plays out.
Could the choice to normalize (in the exhaustive version, with only types that are equatable) be made automatically by a runtime if the compiler can be taught that addition and multiplication are commutative?
I think this talk is a bit much for me. Lost me at 'Distribution of distributions'. I just want to learn how to write a Haskell program with aeson, mongodb, scotty and some State monad transformer stuff with logging and error handling
Let me help explain. A distribution of discrete set of outcomes can be be drawn as a tree of height 1 with the leaves labelled with each outcome and its weight. A distribution of distribution is can just be drawn as a tree of height 2.
Beautifully explained, Tikhon!
What an amazing talk! Thanks a lot Tikhon!
He spits out words faster than an auctioneer
I was having quite some trouble trying to calculate the conjunction of two probabilities that are based on the same probabilities. Vastly simplified: What is the chance of when two dice are thrown, the higher of the two being above 3, and the lower of the two being below 4.
Do you deal with anything like that?
I imagine one would lift the sort, but it remains somewhat unclear how this actually plays out.
Could the choice to normalize (in the exhaustive version, with only types that are equatable) be made automatically by a runtime if the compiler can be taught that addition and multiplication are commutative?
I think this talk is a bit much for me. Lost me at 'Distribution of distributions'. I just want to learn how to write a Haskell program with aeson, mongodb, scotty and some State monad transformer stuff with logging and error handling
Let me help explain. A distribution of discrete set of outcomes can be be drawn as a tree of height 1 with the leaves labelled with each outcome and its weight. A distribution of distribution is can just be drawn as a tree of height 2.