Rodrigo

So I'm building a skipping CD emulator and as part of this, I'm setting up the probabilities for each time of event to happen, each one of which will have nested probabilities of something else happening beyond that.

When I've done stuff like this in the past I've used a RNG -> cascaded split objects, with the splits defining the probabilities of each section.

This works, but is a bit clunky, especially once I start nesting it all. The main issue is that although I can dynamically control the split points by setting the [split] objects arguments, I can't easily add more variables, without cascading more splits. (I guess I can just build like 20 cascaded splits, and then parse some kind of messages to use only the amount of splits as needed).

best-way-to-implement-a-probability-tree

Actually, as a fuller example of what I mean, here is my "glitchMetro" abstraction, that takes a tempo just like metro, but provides slightly irregular timing outputs based on measurements of CD player skipping inconsistencies.

So it's split across 7 probabilities, each with its own tempo multiplier.

Don't know if this is the best way either, but you could use [prob] and make each number have the same probability coming from every other possible state.

Ah [prob], that looks like it could do the trick, without nesting even.

I'll have a proper look at this, but perhaps prob + coll (to parse the states into a more useful symbol/named state) will do the trick. I just need to map out a clear state tree/diagram and build from there.

Here's a way to use expr in place of the split objects.

That's nice! Much tidier (and mutable) than what I had.

I'll have a more thorough look at this (and prob) once I get back from holiday.

Best way to implement a probability tree?