Distribution Support & Mixed Sets #3115
Conversation
🦋 Changeset detectedLatest commit: 6b9e653 The changes in this PR will be included in the next version bump. This PR includes changesets to release 6 packages
Not sure what this means? Click here to learn what changesets are. Click here if you're a maintainer who wants to add another changeset to this PR |
|
The latest updates on your projects. Learn more about Vercel for Git ↗︎
1 Ignored Deployment
|
| @@ -232,6 +233,10 @@ export class DiscreteShape implements PointSet<DiscreteShape> { | |||
| (t) => t.shapeMap(XYShape.T.square).mean() | |||
| ); | |||
| } | |||
|
|
|||
| domain() { | |||
There was a problem hiding this comment.
todo: Change to "support"
OAGr
changed the title
| case "lesserThan": | ||
| return y < _threshold; | ||
| case "equals": | ||
| return Math.abs(y - _threshold) < Number.EPSILON; |
There was a problem hiding this comment.
There are probably subtle bugs here, or, at least, this boundary is arbitrary; Number.EPSILON is not the smallest representable number.
Not sure if this is important.
| return { | ||
| points: sortBy(points, ([x]) => x), | ||
| segments, | ||
| }; |
There was a problem hiding this comment.
General thoughts on this function, without going into details too much:
- there are a few things here that could be optimized (e.g. multiple calls to
comparisonIsSatisfied), but it's not very important - I'm kinda sad about it, because by now I did ~3 attempts to rewrite and clean up
XYShape; it's hard to do incrementally and hard to do in a single day; so the more code this file has, the harder it'll be to do - but I guess it's worth it and not a blocker for the PR
There was a problem hiding this comment.
so the more code this file has, the harder it'll be to do
I agree, but at the same time, I don't expect it to be too bad to convert this to any other XYShape format, especially with tests + LLMs.
There was a problem hiding this comment.
Yea, I wasn't very focused on optimization. I don't expect we'll call this too often, and we can optimize if it ever seems like we'll want to, later.
supportfunctions to PointSet distributions, to get their support. This is a Mixed Set.extractSubsetThatSatisfiesThresholdfunction to XYShape. This is kind of complicated. We only use part of it forsupport(), but the rest was easy, and could be useful for later.Most of this is done in squiggle-lang and isn't exposed to the function registry. However, two functions are -
PointSet.supportand subtractingMixedSetvalues.MixedSetvalues are just stored and provided as Dicts.