Random variables

[Kevinpgalligan]

This would basically be a pretty shim over scipy. It would require a RandomVariable type, an Event type, probability function P that accepts Events, expected value function E that accepts random variables, and so on.

X ~ binom(n, p); P(X <= 3); E(X)

[Kevinpgalligan]

Don't need special syntax for assigning random variables, e.g. binom can be a function that returns a random variable and X is just a variable.

X = binom(n, p)

Note: > comparison operator would clash with current unit conversion operator.

[Kevinpgalligan]

Brainstorming:

• sample function that takes a random variable and returns a set. Also takes a set.
• mean function takes a random variable or set. E function is expectation, only takes random variables.

[Kevinpgalligan]

This is preeeetty much done, just have to add support for adding scalars to / multiply constants by RVs. I would like to support unions of events / joint probability / whatever, as well as adding random variables together, but that gets messy when RVs are dependent on each other. And should be easy enough to manually compute with the existing tools.

[Kevinpgalligan]

Actually, I don't see an immediate use for scalar arithmetic with RVs. Instead of E(X+1) you can just do E(X)+1, for example. So... passing on that for now.