Merge pull request #995 from quantified-uncertainty/sampleSet-truncate

Added truncate for SampleSet distribution

packages/squiggle-lang/__tests__/ReducerInterface/ReducerInterface_Distribution_test.res

@@ -74,6 +74,7 @@ describe("eval on distribution functions", () => {
     testEval("truncateLeft(normal(5,2), 3)", "Ok(Point Set Distribution)")
     testEval("truncateRight(normal(5,2), 3)", "Ok(Point Set Distribution)")
     testEval("truncate(normal(5,2), 3, 8)", "Ok(Point Set Distribution)")
+    testEval("truncate(normal(5,2) |> SampleSet.fromDist, 3, 8)", "Ok(Sample Set Distribution)")
     testEval("isNormalized(truncate(normal(5,2), 3, 8))", "Ok(true)")
   })
 

packages/squiggle-lang/src/rescript/Distributions/GenericDist.res

@@ -242,11 +242,19 @@ module Truncate = {
       switch trySymbolicSimplification(leftCutoff, rightCutoff, t) {
       | Some(r) => Ok(r)
       | None =>
-        toPointSetFn(t)->E.R2.fmap(t => {
-          DistributionTypes.PointSet(
-            PointSetDist.T.truncate(leftCutoff, rightCutoff, t)->PointSetDist.T.normalize,
-          )
-        })
+        switch t {
+        | SampleSet(t) =>
+          switch SampleSetDist.truncate(t, ~leftCutoff, ~rightCutoff) {
+          | Ok(r) => Ok(SampleSet(r))
+          | Error(err) => Error(DistributionTypes.SampleSetError(err))
+          }
+        | _ =>
+          toPointSetFn(t)->E.R2.fmap(t => {
+            DistributionTypes.PointSet(
+              PointSetDist.T.truncate(leftCutoff, rightCutoff, t)->PointSetDist.T.normalize,
+            )
+          })
+        }
       }
     }
   }

packages/squiggle-lang/src/rescript/Distributions/SampleSetDist/SampleSetDist.res

@@ -131,3 +131,12 @@ let max = t => T.get(t)->E.A.Floats.max
 let stdev = t => T.get(t)->E.A.Floats.stdev
 let variance = t => T.get(t)->E.A.Floats.variance
 let percentile = (t, f) => T.get(t)->E.A.Floats.percentile(f)
+
+let truncateLeft = (t, f) => T.get(t)->E.A2.filter(x => x >= f)->T.make
+let truncateRight = (t, f) => T.get(t)->E.A2.filter(x => x <= f)->T.make
+
+let truncate = (t, ~leftCutoff: option<float>, ~rightCutoff: option<float>) => {
+  let withTruncatedLeft = t => leftCutoff |> E.O.dimap(left => truncateLeft(t, left), _ => Ok(t))
+  let withTruncatedRight = t => rightCutoff |> E.O.dimap(left => truncateRight(t, left), _ => Ok(t))
+  t->withTruncatedLeft |> E.R2.bind(withTruncatedRight)
+}

packages/website/docs/Api/Dist.mdx

@@ -290,12 +290,29 @@ quantile: (distribution, number) => number
 quantile(normal(5, 2), 0.5);
 ```
 
-### truncateLeft
+### truncate
 
-Truncates the left side of a distribution. Returns either a pointSet distribution or a symbolic distribution.
+Truncates both the left side and the right side of a distribution.
 
 ```
-truncateLeft: (distribution, l => number) => distribution
+truncate: (distribution, left: number, right: number) => distribution
+```
+
+<Admonition type="note" title="Implementation Details">
+  <p>
+    Sample set distributions are truncated by filtering samples, but point set
+    distributions are truncated using direct geometric manipulation. Uniform
+    distributions are truncated symbolically. Symbolic but non-uniform
+    distributions get converted to Point Set distributions.
+  </p>
+</Admonition>
+
+### truncateLeft
+
+Truncates the left side of a distribution.
+
+```
+truncateLeft: (distribution, left: number) => distribution
 ```
 
 **Examples**
@@ -306,10 +323,10 @@ truncateLeft(normal(5, 2), 3);
 
 ### truncateRight
 
-Truncates the right side of a distribution. Returns either a pointSet distribution or a symbolic distribution.
+Truncates the right side of a distribution.
 
 ```
-truncateRight: (distribution, r => number) => distribution
+truncateRight: (distribution, right: number) => distribution
 ```
 
 **Examples**
@@ -388,7 +405,7 @@ The only functions that do not return normalized distributions are the pointwise
 
 ### normalize
 
-Normalize a distribution. This means scaling it appropriately so that it's cumulative sum is equal to 1. This only impacts Pointset distributions, because those are the only ones that can be non-normlized.
+Normalize a distribution. This means scaling it appropriately so that it's cumulative sum is equal to 1. This only impacts Point Set distributions, because those are the only ones that can be non-normlized.
 
 ```
 normalize: (distribution) => distribution