[hhnijljo] Composing Bernoulli trials

A Bernoulli trial is the flipping of a (possibly biased) coin.  A Bernoulli trial can have only 2 possible outcomes: success or failure.

A Bernoulli trial can be constructed out of smaller Bernoulli trials.  Combine small results with a boolean function.  Examples: "at least one success", "flawless victory".

Things composed of smaller similar things are generally intriguing, but I don't know where this might be useful.

What is the distribution of the number of small trials necessary to sample a composite Bernoulli trial?  Allow short-circuit boolean function evaluation.

Generalizing further yields a Bayesian network.