Compare commits

...

15 Commits

Author SHA1 Message Date
Sam Nolan
898750f8f9 Move selection logic to chooseCalculationMethod 2022-04-20 16:47:07 -04:00
Sam Nolan
65397b3fe0 Fix build errors in components 2022-04-20 14:52:23 -04:00
Sam Nolan
956b2fecef Merge branch 'develop' into pointwise-commutative-error 2022-04-20 14:38:07 -04:00
Sam Nolan
d27b777900 Remove Old code, restrict convolution to specific types 2022-04-20 14:36:19 -04:00
cd41459887 fix: tiny bug 2022-04-20 13:34:54 -04:00
468ecc05e3 feat: calculate multiplication using convolution 2022-04-20 13:29:33 -04:00
cfc1fd2239 tweak: minor code org 2022-04-20 13:25:13 -04:00
c8a505dcf0 tweak: comments 2022-04-20 13:22:52 -04:00
10f88b4a2a feat: Save progress before pushing to repository
Note: Not for use yet
2022-04-20 12:25:52 -04:00
6e834af7d7 Merge branch 'pointwise-commutative-error' of github.com:quantified-uncertainty/squiggle into pointwise-commutative-error 2022-04-20 11:32:52 -04:00
36b318dede feat: get 'convolutions' working for sum of distributions 2022-04-20 11:32:50 -04:00
Sam Nolan
8dab784d2c Merge branch 'pointwise-commutative-error' of github.com:quantified-uncertainty/squiggle into pointwise-commutative-error 2022-04-20 06:07:00 -04:00
8b042db2b8 fix: renamed Algebraic => Numeric
This is not doing algebraic manipulations, this is doing
numeric computations. Confusing difference.
2022-04-19 13:53:24 -04:00
Sam Nolan
da2af9cf10 Reduce code duplication with parameter 2022-04-19 07:48:39 +10:00
Sam Nolan
8deb796820 Fix non-commutative pointwise combinations 2022-04-18 16:42:11 +10:00
20 changed files with 281 additions and 1612 deletions

View File

@ -6,11 +6,12 @@ import {
errorValueToString,
squiggleExpression,
} from "@quri/squiggle-lang";
import type { samplingParams, exportEnv } from "@quri/squiggle-lang";
import type { samplingParams } from "@quri/squiggle-lang";
import { NumberShower } from "./NumberShower";
import { DistributionChart } from "./DistributionChart";
import { ErrorBox } from "./ErrorBox";
import useSize from "@react-hook/size";
type exportEnv = unknown;
const variableBox = {
Component: styled.div`

View File

@ -2,9 +2,10 @@ import * as React from "react";
import * as ReactDOM from "react-dom";
import { SquiggleChart } from "./SquiggleChart";
import { CodeEditor } from "./CodeEditor";
import type { exportEnv } from "@quri/squiggle-lang";
import styled from "styled-components";
type exportEnv = unknown;
export interface SquiggleEditorProps {
/** The input string for squiggle */
initialSquiggleString?: string;

View File

@ -67,7 +67,7 @@ describe("eval on distribution functions", () => {
testEval("lognormal(10,2) / lognormal(5,2)", "Ok(Lognormal(5,2.8284271247461903))")
testEval("lognormal(5, 2) / 2", "Ok(Lognormal(4.306852819440055,2))")
testEval("2 / lognormal(5, 2)", "Ok(Lognormal(-4.306852819440055,2))")
testEval("2 / normal(10, 2)", "Ok(Point Set Distribution)")
testEval("2 / normal(10, 2)", "Ok(Sample Set Distribution)")
testEval("normal(10, 2) / 2", "Ok(Normal(5,1))")
})
describe("truncate", () => {
@ -77,21 +77,21 @@ describe("eval on distribution functions", () => {
})
describe("exp", () => {
testEval("exp(normal(5,2))", "Ok(Point Set Distribution)")
testEval("exp(normal(5,2))", "Ok(Sample Set Distribution)")
})
describe("pow", () => {
testEval("pow(3, uniform(5,8))", "Ok(Point Set Distribution)")
testEval("pow(uniform(5,8), 3)", "Ok(Point Set Distribution)")
testEval("pow(3, uniform(5,8))", "Ok(Sample Set Distribution)")
testEval("pow(uniform(5,8), 3)", "Ok(Sample Set Distribution)")
testEval("pow(uniform(5,8), uniform(9, 10))", "Ok(Sample Set Distribution)")
})
describe("log", () => {
testEval("log(2, uniform(5,8))", "Ok(Point Set Distribution)")
testEval("log(normal(5,2), 3)", "Ok(Point Set Distribution)")
testEval("log(2, uniform(5,8))", "Ok(Sample Set Distribution)")
testEval("log(normal(5,2), 3)", "Ok(Sample Set Distribution)")
testEval("log(normal(5,2), normal(10,1))", "Ok(Sample Set Distribution)")
testEval("log(uniform(5,8))", "Ok(Point Set Distribution)")
testEval("log10(uniform(5,8))", "Ok(Point Set Distribution)")
testEval("log(uniform(5,8))", "Ok(Sample Set Distribution)")
testEval("log10(uniform(5,8))", "Ok(Sample Set Distribution)")
})
describe("dotLog", () => {

View File

@ -1,9 +1,4 @@
import * as _ from "lodash";
import type {
exportEnv,
exportDistribution,
} from "../rescript/ProgramEvaluator.gen";
export type { exportEnv, exportDistribution };
import {
genericDist,
samplingParams,
@ -48,7 +43,6 @@ import {
Constructors_pointwiseLogarithm,
Constructors_pointwisePower,
} from "../rescript/Distributions/DistributionOperation/DistributionOperation.gen";
import { pointSetDistFn } from "../rescript/OldInterpreter/DistPlus.bs";
export type { samplingParams, errorValue };
export let defaultSamplingInputs: samplingParams = {
@ -99,7 +93,7 @@ export type squiggleExpression =
export function run(
squiggleString: string,
samplingInputs?: samplingParams,
_environment?: exportEnv
_environment?: unknown
): result<squiggleExpression, errorValue> {
let si: samplingParams = samplingInputs
? samplingInputs

View File

@ -158,7 +158,7 @@ module AlgebraicCombination = {
let runConvolution = (
toPointSet: toPointSetFn,
arithmeticOperation: GenericDist_Types.Operation.arithmeticOperation,
arithmeticOperation: Operation.convolutionOperation,
t1: t,
t2: t,
) =>
@ -191,10 +191,23 @@ module AlgebraicCombination = {
| _ => 1000
}
let chooseConvolutionOrMonteCarlo = (t2: t, t1: t) =>
expectedConvolutionCost(t1) * expectedConvolutionCost(t2) > 10000
? #CalculateWithMonteCarlo
: #CalculateWithConvolution
type calculationMethod = MonteCarlo | Convolution(Operation.convolutionOperation)
let chooseConvolutionOrMonteCarlo = (
op: Operation.algebraicOperation,
t2: t,
t1: t,
): calculationMethod =>
switch op {
| #Divide
| #Power
| #Logarithm =>
MonteCarlo
| (#Add | #Subtract | #Multiply) as convOp =>
expectedConvolutionCost(t1) * expectedConvolutionCost(t2) > 10000
? MonteCarlo
: Convolution(convOp)
}
let run = (
t1: t,
@ -207,15 +220,10 @@ module AlgebraicCombination = {
| Some(Ok(symbolicDist)) => Ok(Symbolic(symbolicDist))
| Some(Error(e)) => Error(Other(e))
| None =>
switch chooseConvolutionOrMonteCarlo(t1, t2) {
| #CalculateWithMonteCarlo => runMonteCarlo(toSampleSetFn, arithmeticOperation, t1, t2)
| #CalculateWithConvolution =>
runConvolution(
toPointSetFn,
arithmeticOperation,
t1,
t2,
)->E.R2.fmap(r => DistributionTypes.PointSet(r))
switch chooseConvolutionOrMonteCarlo(arithmeticOperation, t1, t2) {
| MonteCarlo => runMonteCarlo(toSampleSetFn, arithmeticOperation, t1, t2)
| Convolution(convOp) =>
runConvolution(toPointSetFn, convOp, t1, t2)->E.R2.fmap(r => DistributionTypes.PointSet(r))
}
}
}

View File

@ -247,9 +247,10 @@ let downsampleEquallyOverX = (length, t): t =>
/* This simply creates multiple copies of the continuous distribution, scaled and shifted according to
each discrete data point, and then adds them all together. */
let combineAlgebraicallyWithDiscrete = (
op: Operation.algebraicOperation,
op: Operation.convolutionOperation,
t1: t,
t2: PointSetTypes.discreteShape,
discreteFirst: bool,
) => {
let t1s = t1 |> getShape
let t2s = t2.xyShape // TODO would like to use Discrete.getShape here, but current file structure doesn't allow for that
@ -262,15 +263,15 @@ let combineAlgebraicallyWithDiscrete = (
| #Stepwise => stepwiseToLinear(t1)
}
let combinedShape = AlgebraicShapeCombination.combineShapesContinuousDiscrete(
let combinedShape = NumericShapeCombination.combineShapesContinuousDiscrete(
op,
continuousAsLinear |> getShape,
t2s,
discreteFirst,
)
let combinedIntegralSum = switch op {
| #Multiply
| #Divide =>
| #Multiply =>
Common.combineIntegralSums((a, b) => Some(a *. b), t1.integralSumCache, t2.integralSumCache)
| _ => None
}
@ -280,7 +281,7 @@ let combineAlgebraicallyWithDiscrete = (
}
}
let combineAlgebraically = (op: Operation.algebraicOperation, t1: t, t2: t) => {
let combineAlgebraically = (op: Operation.convolutionOperation, t1: t, t2: t) => {
let s1 = t1 |> getShape
let s2 = t2 |> getShape
let t1n = s1 |> XYShape.T.length
@ -288,7 +289,7 @@ let combineAlgebraically = (op: Operation.algebraicOperation, t1: t, t2: t) => {
if t1n == 0 || t2n == 0 {
empty
} else {
let combinedShape = AlgebraicShapeCombination.combineShapesContinuousContinuous(op, s1, s2)
let combinedShape = NumericShapeCombination.combineShapesContinuousContinuous(op, s1, s2)
let combinedIntegralSum = Common.combineIntegralSums(
(a, b) => Some(a *. b),
t1.integralSumCache,

View File

@ -85,7 +85,7 @@ let updateIntegralCache = (integralCache, t: t): t => {
/* This multiples all of the data points together and creates a new discrete distribution from the results.
Data points at the same xs get added together. It may be a good idea to downsample t1 and t2 before and/or the result after. */
let combineAlgebraically = (op: Operation.algebraicOperation, t1: t, t2: t): t => {
let combineAlgebraically = (op: Operation.convolutionOperation, t1: t, t2: t): t => {
let t1s = t1 |> getShape
let t2s = t2 |> getShape
let t1n = t1s |> XYShape.T.length
@ -97,7 +97,7 @@ let combineAlgebraically = (op: Operation.algebraicOperation, t1: t, t2: t): t =
t2.integralSumCache,
)
let fn = Operation.Algebraic.toFn(op)
let fn = Operation.Convolution.toFn(op)
let xToYMap = E.FloatFloatMap.empty()
for i in 0 to t1n - 1 {

View File

@ -226,7 +226,7 @@ module T = Dist({
}
})
let combineAlgebraically = (op: Operation.algebraicOperation, t1: t, t2: t): t => {
let combineAlgebraically = (op: Operation.convolutionOperation, t1: t, t2: t): t => {
// Discrete convolution can cause a huge increase in the number of samples,
// so we'll first downsample.
@ -242,8 +242,18 @@ let combineAlgebraically = (op: Operation.algebraicOperation, t1: t, t2: t): t =
// continuous (*) continuous => continuous, but also
// discrete (*) continuous => continuous (and vice versa). We have to take care of all combos and then combine them:
let ccConvResult = Continuous.combineAlgebraically(op, t1.continuous, t2.continuous)
let dcConvResult = Continuous.combineAlgebraicallyWithDiscrete(op, t2.continuous, t1.discrete)
let cdConvResult = Continuous.combineAlgebraicallyWithDiscrete(op, t1.continuous, t2.discrete)
let dcConvResult = Continuous.combineAlgebraicallyWithDiscrete(
op,
t2.continuous,
t1.discrete,
true,
)
let cdConvResult = Continuous.combineAlgebraicallyWithDiscrete(
op,
t1.continuous,
t2.discrete,
false,
)
let continuousConvResult = Continuous.reduce(\"+.", [ccConvResult, dcConvResult, cdConvResult])
// ... finally, discrete (*) discrete => discrete, obviously:

View File

@ -96,7 +96,7 @@ let toDiscretePointMassesFromTriangulars = (
}
let combineShapesContinuousContinuous = (
op: Operation.algebraicOperation,
op: Operation.convolutionOperation,
s1: PointSetTypes.xyShape,
s2: PointSetTypes.xyShape,
): PointSetTypes.xyShape => {
@ -104,7 +104,6 @@ let combineShapesContinuousContinuous = (
// if we multiply the two distributions, we should probably use lognormal filters.
let t1m = toDiscretePointMassesFromTriangulars(s1)
let t2m = switch op {
| #Divide => toDiscretePointMassesFromTriangulars(~inverse=true, s2)
| _ => toDiscretePointMassesFromTriangulars(~inverse=false, s2)
}
@ -112,9 +111,6 @@ let combineShapesContinuousContinuous = (
| #Add => (m1, m2) => m1 +. m2
| #Subtract => (m1, m2) => m1 -. m2
| #Multiply => (m1, m2) => m1 *. m2
| #Divide => (m1, mInv2) => m1 *. mInv2
| #Power => (m1, mInv2) => m1 ** mInv2
| #Logarithm => (m1, m2) => log(m1) /. log(m2)
} // note: here, mInv2 = mean(1 / t2) ~= 1 / mean(t2)
// TODO: Variances are for exponentatiation or logarithms are almost totally made up and very likely very wrong.
@ -123,9 +119,6 @@ let combineShapesContinuousContinuous = (
| #Add => (v1, v2, _, _) => v1 +. v2
| #Subtract => (v1, v2, _, _) => v1 +. v2
| #Multiply => (v1, v2, m1, m2) => v1 *. v2 +. v1 *. m2 ** 2. +. v2 *. m1 ** 2.
| #Power => (v1, v2, m1, m2) => v1 *. v2 +. v1 *. m2 ** 2. +. v2 *. m1 ** 2.
| #Logarithm => (v1, v2, m1, m2) => v1 *. v2 +. v1 *. m2 ** 2. +. v2 *. m1 ** 2.
| #Divide => (v1, vInv2, m1, mInv2) => v1 *. vInv2 +. v1 *. mInv2 ** 2. +. vInv2 *. m1 ** 2.
}
// TODO: If operating on two positive-domain distributions, we should take that into account
@ -199,59 +192,28 @@ let toDiscretePointMassesFromDiscrete = (s: PointSetTypes.xyShape): pointMassesW
}
let combineShapesContinuousDiscrete = (
op: Operation.algebraicOperation,
op: Operation.convolutionOperation,
continuousShape: PointSetTypes.xyShape,
discreteShape: PointSetTypes.xyShape,
discreteFirst: bool,
): PointSetTypes.xyShape => {
let t1n = continuousShape |> XYShape.T.length
let t2n = discreteShape |> XYShape.T.length
// each x pair is added/subtracted
let fn = Operation.Algebraic.toFn(op)
let opFunc = Operation.Convolution.toFn(op)
let fn = discreteFirst ? (a, b) => opFunc(b, a) : opFunc
let outXYShapes: array<array<(float, float)>> = Belt.Array.makeUninitializedUnsafe(t2n)
let discretePoints = Belt.Array.zip(discreteShape.xs, discreteShape.ys)
let continuousPoints = Belt.Array.zip(continuousShape.xs, continuousShape.ys)
switch op {
let outXYShapes = switch op {
| #Add
| #Subtract =>
for j in 0 to t2n - 1 {
// creates a new continuous shape for each one of the discrete points, and collects them in outXYShapes.
let dxyShape: array<(float, float)> = Belt.Array.makeUninitializedUnsafe(t1n)
for i in 0 to t1n - 1 {
Belt.Array.set(
dxyShape,
i,
(
fn(continuousShape.xs[i], discreteShape.xs[j]),
continuousShape.ys[i] *. discreteShape.ys[j],
),
) |> ignore
()
}
Belt.Array.set(outXYShapes, j, dxyShape) |> ignore
()
}
| #Multiply
| #Power
| #Logarithm
| #Divide =>
for j in 0 to t2n - 1 {
// creates a new continuous shape for each one of the discrete points, and collects them in outXYShapes.
let dxyShape: array<(float, float)> = Belt.Array.makeUninitializedUnsafe(t1n)
for i in 0 to t1n - 1 {
Belt.Array.set(
dxyShape,
i,
(
fn(continuousShape.xs[i], discreteShape.xs[j]),
continuousShape.ys[i] *. discreteShape.ys[j] /. discreteShape.xs[j],
),
) |> ignore
()
}
Belt.Array.set(outXYShapes, j, dxyShape) |> ignore
()
}
discretePoints->E.A2.fmap(((dx, dy)) =>
continuousPoints->E.A2.fmap(((cx, cy)) => (fn(cx, dx), cy *. dy))
)
| #Multiply =>
discretePoints->E.A2.fmap(((dx, dy)) =>
continuousPoints->E.A2.fmap(((cx, cy)) => (fn(cx, dx), cy *. dy /. dx))
)
}
outXYShapes

View File

@ -0,0 +1,187 @@
// Types
let emptyXYShape: PointSetTypes.xyShape = {xs: [], ys: []}
exception LogicallyInconsistent(string)
// Helpers
let getArithmeticComplementOfDistributionForSubstraction = (
dist: PointSetTypes.xyShape,
): PointSetTypes.xyShape => {
let newXs = Belt.Array.map(dist.xs, x => -.x)
{xs: newXs, ys: dist.ys}
}
let getApproximatePdfOfContinuousDistributionAtPoint = (
dist: PointSetTypes.xyShape,
point: float,
): float => {
let closestFromBelowIndex = E.A.reducei(dist.xs, None, (accumulator, item, index) =>
item < point ? Some(index) : accumulator
) // This could be improved by taking advantage of the fact that these are ordered
let closestFromAboveIndexOption = Belt.Array.getIndexBy(dist.xs, item => item > point)
let weightedMean = (
point: float,
closestFromBelow: float,
closestFromAbove: float,
valueclosestFromBelow,
valueclosestFromAbove,
): float => {
let distance = closestFromAbove -. closestFromBelow
let w1 = (point -. closestFromBelow) /. distance
let w2 = (closestFromAbove -. point) /. distance
let result = w1 *. valueclosestFromAbove +. w2 *. valueclosestFromBelow
result
}
let result = switch (closestFromBelowIndex, closestFromAboveIndexOption) {
| (None, None) =>
raise(
LogicallyInconsistent(
"Logically inconsistent option in NumericShapeCombination2.res. Possibly caused by empty distribution",
), // to do: give an error type
) // all are smaller, and all are larger
| (None, Some(i)) => 0.0 // none are smaller, all are larger
| (Some(i), None) => 0.0 // all are smaller, none are larger
| (Some(i), Some(j)) => weightedMean(point, dist.xs[i], dist.xs[j], dist.ys[i], dist.ys[j]) // there is a lowerBound and an upperBound.
}
result
}
// Inner functions
let addContinuousContinuous = (
s1: PointSetTypes.xyShape,
s2: PointSetTypes.xyShape,
): PointSetTypes.xyShape => {
// Assumption: xyShapes are ordered on the x coordinate.
// Get some needed variables
let len1 = XYShape.T.length(s1)
let mins1xs = s1.xs[0] // Belt.Array.reduce(s1.xs, s1.xs[0], (a, b) => a < b ? a : b)
let maxs1xs = s1.xs[len1 - 1] // Belt.Array.reduce(s1.xs, s1.xs[0], (a, b) => a > b ? a : b)
let len2 = XYShape.T.length(s2)
let mins2xs = s2.xs[0] // Belt.Array.reduce(s1.xs, s1.xs[0], (a, b) => a < b ? a : b)
let maxs2xs = s2.xs[len1 - 1] // Belt.Array.reduce(s1.xs, s1.xs[0], (a, b) => a > b ? a : b)
let lowerBound = mins1xs +. mins2xs
let upperBound = maxs1xs +. maxs2xs
let numIntervals = 2 * Js.Math.max_int(len1, len2) // 5000
let epsilon = (upperBound -. lowerBound) /. Belt.Int.toFloat(numIntervals) // Js.Math.pow_float(~base=2.0, ~exp=-16.0)
let newXs: array<float> = Belt.Array.makeUninitializedUnsafe(numIntervals)
let newYs: array<float> = Belt.Array.makeUninitializedUnsafe(numIntervals)
let getApproximatePdfOfS1AtPoint = x => getApproximatePdfOfContinuousDistributionAtPoint(s1, x)
let getApproximatePdfOfS2AtPoint = x => getApproximatePdfOfContinuousDistributionAtPoint(s2, x)
let float = x => Belt.Int.toFloat(x)
// Compute the integral numerically.
// I wouldn't worry too much about the O(n^3). At 5000 samples, this takes on the order of 25 million operations
// The AMD Ryzen 7 processor in my computer can do around 300K million operations per second.
// src: https://wikiless.org/wiki/Instructions_per_second?lang=en#Thousand_instructions_per_second_(TIPS/kIPS)
for i in 0 to numIntervals - 1 {
// where are the x points in the resulting distribution
let z = lowerBound +. float(i) *. epsilon
newXs[i] = z
newYs[i] = 0.0
for j in 0 to numIntervals - 1 {
// how fine-grained do we want our approximation of the integral to be.
let x = lowerBound +. float(j) *. epsilon
let deltaYi = getApproximatePdfOfS1AtPoint(x) *. getApproximatePdfOfS2AtPoint(z -. x)
newYs[i] = newYs[i] +. deltaYi
}
}
// This could be improved by, for instance, choosing the location of the xs strategically
// for example, such that each of them is "equidistant" in a cdf, that is, such that the
// cdf increases by constant amounts from one point to another.
{xs: newXs, ys: newYs}
}
let multiplyContinuousContinuous = (
s1: PointSetTypes.xyShape,
s2: PointSetTypes.xyShape,
): PointSetTypes.xyShape => {
// Assumption: xyShapes are ordered on the x coordinate.
// Get some needed variables
let len1 = XYShape.T.length(s1)
let mins1xs = s1.xs[0] // Belt.Array.reduce(s1.xs, s1.xs[0], (a, b) => a < b ? a : b)
let maxs1xs = s1.xs[len1 - 1] // Belt.Array.reduce(s1.xs, s1.xs[0], (a, b) => a > b ? a : b)
let len2 = XYShape.T.length(s2)
let mins2xs = s2.xs[0] // Belt.Array.reduce(s1.xs, s1.xs[0], (a, b) => a < b ? a : b)
let maxs2xs = s2.xs[len1 - 1] // Belt.Array.reduce(s1.xs, s1.xs[0], (a, b) => a > b ? a : b)
let lowerBound = mins1xs *. mins2xs
let upperBound = maxs1xs *. maxs2xs
let numIntervals = 2 * Js.Math.max_int(len1, len2) // 5000
let epsilon = (upperBound -. lowerBound) /. Belt.Int.toFloat(numIntervals) // Js.Math.pow_float(~base=2.0, ~exp=-16.0)
let epsilonForIgnoreInIntegral = Js.Math.pow_float(~base=2.0, ~exp=-16.0)
let newXs: array<float> = Belt.Array.makeUninitializedUnsafe(numIntervals)
let newYs: array<float> = Belt.Array.makeUninitializedUnsafe(numIntervals)
let getApproximatePdfOfS1AtPoint = x => getApproximatePdfOfContinuousDistributionAtPoint(s1, x)
let getApproximatePdfOfS2AtPoint = x => getApproximatePdfOfContinuousDistributionAtPoint(s2, x)
let float = x => Belt.Int.toFloat(x)
// Compute the integral numerically.
// I wouldn't worry too much about the O(n^3). At 5000 samples, this takes on the order of 25 million operations
// The AMD Ryzen 7 processor in my computer can do around 300K million operations per second.
// src: https://wikiless.org/wiki/Instructions_per_second?lang=en#Thousand_instructions_per_second_(TIPS/kIPS)
for i in 0 to numIntervals - 1 {
// where are the x points in the resulting distribution
let z = lowerBound +. float(i) *. epsilon
newXs[i] = z
newYs[i] = 0.0
for j in 0 to numIntervals - 1 {
// how fine-grained do we want our approximation of the integral to be.
let x = lowerBound +. float(j) *. epsilon
let absX = Js.Math.abs_float(x)
if absX > epsilonForIgnoreInIntegral {
let deltaYi =
getApproximatePdfOfS1AtPoint(x) *. getApproximatePdfOfS2AtPoint(z /. x) *. (1.0 /. x)
newYs[i] = newYs[i] +. deltaYi
}
}
}
// This could be improved by, for instance, choosing the location of the xs strategically
// for example, such that each of them is "equidistant" in a cdf, that is, such that the
// cdf increases by constant amounts from one point to another.
{xs: newXs, ys: newYs}
}
// Main function
let combineShapesContinuousContinuous = (
op: Operation.algebraicOperation,
s1: PointSetTypes.xyShape,
s2: PointSetTypes.xyShape,
): PointSetTypes.xyShape => {
let result = switch op {
| #Add => addContinuousContinuous(s1, s2)
| #Subtract =>
addContinuousContinuous(s1, getArithmeticComplementOfDistributionForSubstraction(s2))
| #Multiply => emptyXYShape
| #Divide => emptyXYShape
| #Power => emptyXYShape
| #Logarithm => emptyXYShape
}
result
}
// Not sure I understand how to combine continuous and discrete distribution
let combineShapesContinuousDiscrete = (
op: Operation.algebraicOperation,
continuousShape: PointSetTypes.xyShape,
discreteShape: PointSetTypes.xyShape,
): PointSetTypes.xyShape => emptyXYShape
let combineShapesDiscreteContinuous = (
op: Operation.algebraicOperation,
discreteShape: PointSetTypes.xyShape,
continuousShape: PointSetTypes.xyShape,
): PointSetTypes.xyShape => emptyXYShape

View File

@ -34,14 +34,14 @@ let toMixed = mapToAll((
),
))
//TODO WARNING: The combineAlgebraicallyWithDiscrete will break for subtraction and division, like, discrete - continous
let combineAlgebraically = (op: Operation.algebraicOperation, t1: t, t2: t): t =>
let combineAlgebraically = (op: Operation.convolutionOperation, t1: t, t2: t): t =>
switch (t1, t2) {
| (Continuous(m1), Continuous(m2)) =>
Continuous.combineAlgebraically(op, m1, m2) |> Continuous.T.toPointSetDist
| (Continuous(m1), Discrete(m2))
| (Discrete(m2), Continuous(m1)) =>
Continuous.combineAlgebraicallyWithDiscrete(op, m1, m2) |> Continuous.T.toPointSetDist
| (Discrete(m1), Continuous(m2)) =>
Continuous.combineAlgebraicallyWithDiscrete(op, m2, m1, true) |> Continuous.T.toPointSetDist
| (Continuous(m1), Discrete(m2)) =>
Continuous.combineAlgebraicallyWithDiscrete(op, m1, m2, false) |> Continuous.T.toPointSetDist
| (Discrete(m1), Discrete(m2)) =>
Discrete.combineAlgebraically(op, m1, m2) |> Discrete.T.toPointSetDist
| (m1, m2) => Mixed.combineAlgebraically(op, toMixed(m1), toMixed(m2)) |> Mixed.T.toPointSetDist

View File

@ -1,24 +0,0 @@
open ASTTypes
let toString = ASTTypes.Node.toString
let envs = (samplingInputs, environment) => {
samplingInputs: samplingInputs,
environment: environment,
evaluateNode: ASTEvaluator.toLeaf,
}
let toLeaf = (samplingInputs, environment, node: node) =>
ASTEvaluator.toLeaf(envs(samplingInputs, environment), node)
let toPointSetDist = (samplingInputs, environment, node: node) =>
switch toLeaf(samplingInputs, environment, node) {
| Ok(#RenderedDist(pointSetDist)) => Ok(pointSetDist)
| Ok(_) => Error("Rendering failed.")
| Error(e) => Error(e)
}
let runFunction = (samplingInputs, environment, inputs, fn: ASTTypes.Function.t) => {
let params = envs(samplingInputs, environment)
ASTTypes.Function.run(params, inputs, fn)
}

View File

@ -1,257 +0,0 @@
open ASTTypes
type tResult = node => result<node, string>
/* Given two random variables A and B, this returns the distribution
of a new variable that is the result of the operation on A and B.
For instance, normal(0, 1) + normal(1, 1) -> normal(1, 2).
In general, this is implemented via convolution. */
module AlgebraicCombination = {
let tryAnalyticalSimplification = (operation, t1: node, t2: node) =>
switch (operation, t1, t2) {
| (operation, #SymbolicDist(d1), #SymbolicDist(d2)) =>
switch SymbolicDist.T.tryAnalyticalSimplification(d1, d2, operation) {
| #AnalyticalSolution(symbolicDist) => Ok(#SymbolicDist(symbolicDist))
| #Error(er) => Error(er)
| #NoSolution => Ok(#AlgebraicCombination(operation, t1, t2))
}
| _ => Ok(#AlgebraicCombination(operation, t1, t2))
}
let combinationByRendering = (evaluationParams, algebraicOp, t1: node, t2: node): result<
node,
string,
> =>
E.R.merge(
Node.ensureIsRenderedAndGetShape(evaluationParams, t1),
Node.ensureIsRenderedAndGetShape(evaluationParams, t2),
) |> E.R.fmap(((a, b)) => #RenderedDist(PointSetDist.combineAlgebraically(algebraicOp, a, b)))
let nodeScore: node => int = x =>
switch x {
| #SymbolicDist(#Float(_)) => 1
| #SymbolicDist(_) => 1000
| #RenderedDist(Discrete(m)) => m.xyShape |> XYShape.T.length
| #RenderedDist(Mixed(_)) => 1000
| #RenderedDist(Continuous(_)) => 1000
| _ => 1000
}
let choose = (t1: node, t2: node) =>
nodeScore(t1) * nodeScore(t2) > 10000 ? #Sampling : #Analytical
let combine = (evaluationParams, algebraicOp, t1: node, t2: node): result<node, string> =>
E.R.merge(
ASTTypes.SamplingDistribution.renderIfIsNotSamplingDistribution(evaluationParams, t1),
ASTTypes.SamplingDistribution.renderIfIsNotSamplingDistribution(evaluationParams, t2),
) |> E.R.bind(_, ((a, b)) =>
switch choose(a, b) {
| #Sampling =>
ASTTypes.SamplingDistribution.combineShapesUsingSampling(
evaluationParams,
algebraicOp,
a,
b,
)
| #Analytical => combinationByRendering(evaluationParams, algebraicOp, a, b)
}
)
let operationToLeaf = (
evaluationParams: evaluationParams,
algebraicOp: Operation.algebraicOperation,
t1: node,
t2: node,
): result<node, string> =>
algebraicOp
|> tryAnalyticalSimplification(_, t1, t2)
|> E.R.bind(_, x =>
switch x {
| #SymbolicDist(_) as t => Ok(t)
| _ => combine(evaluationParams, algebraicOp, t1, t2)
}
)
}
module PointwiseCombination = {
//TODO: This is crude and slow. It forces everything to be pointSetDist, even though much
//of the process could happen on symbolic distributions without a conversion to be a pointSetDist.
let pointwiseAdd = (evaluationParams: evaluationParams, t1: node, t2: node) =>
switch (Node.render(evaluationParams, t1), Node.render(evaluationParams, t2)) {
| (Ok(#RenderedDist(rs1)), Ok(#RenderedDist(rs2))) =>
Ok(
#RenderedDist(
PointSetDist.combinePointwise(
~integralSumCachesFn=(a, b) => Some(a +. b),
~integralCachesFn=(a, b) => Some(
Continuous.combinePointwise(~distributionType=#CDF, \"+.", a, b),
),
\"+.",
rs1,
rs2,
),
),
)
| (Error(e1), _) => Error(e1)
| (_, Error(e2)) => Error(e2)
| _ => Error("Pointwise combination: rendering failed.")
}
let pointwiseCombine = (fn, evaluationParams: evaluationParams, t1: node, t2: node) =>
switch // TODO: construct a function that we can easily sample from, to construct
// a RenderedDist. Use the xMin and xMax of the rendered pointSetDists to tell the sampling function where to look.
// TODO: This should work for symbolic distributions too!
(Node.render(evaluationParams, t1), Node.render(evaluationParams, t2)) {
| (Ok(#RenderedDist(rs1)), Ok(#RenderedDist(rs2))) =>
Ok(#RenderedDist(PointSetDist.combinePointwise(fn, rs1, rs2)))
| (Error(e1), _) => Error(e1)
| (_, Error(e2)) => Error(e2)
| _ => Error("Pointwise combination: rendering failed.")
}
let operationToLeaf = (
evaluationParams: evaluationParams,
pointwiseOp: Operation.pointwiseOperation,
t1: node,
t2: node,
) =>
switch pointwiseOp {
| #Add => pointwiseAdd(evaluationParams, t1, t2)
| #Multiply => pointwiseCombine(\"*.", evaluationParams, t1, t2)
| #Power => pointwiseCombine(\"**", evaluationParams, t1, t2)
}
}
module Truncate = {
type simplificationResult = [
| #Solution(ASTTypes.node)
| #Error(string)
| #NoSolution
]
let trySimplification = (leftCutoff, rightCutoff, t): simplificationResult =>
switch (leftCutoff, rightCutoff, t) {
| (None, None, t) => #Solution(t)
| (Some(lc), Some(rc), _) if lc > rc =>
#Error("Left truncation bound must be smaller than right truncation bound.")
| (lc, rc, #SymbolicDist(#Uniform(u))) =>
#Solution(#SymbolicDist(#Uniform(SymbolicDist.Uniform.truncate(lc, rc, u))))
| _ => #NoSolution
}
let truncateAsShape = (evaluationParams: evaluationParams, leftCutoff, rightCutoff, t) =>
switch // TODO: use named args for xMin/xMax in renderToShape; if we're lucky we can at least get the tail
// of a distribution we otherwise wouldn't get at all
Node.ensureIsRendered(evaluationParams, t) {
| Ok(#RenderedDist(rs)) =>
Ok(#RenderedDist(PointSetDist.T.truncate(leftCutoff, rightCutoff, rs)))
| Error(e) => Error(e)
| _ => Error("Could not truncate distribution.")
}
let operationToLeaf = (
evaluationParams,
leftCutoff: option<float>,
rightCutoff: option<float>,
t: node,
): result<node, string> =>
t
|> trySimplification(leftCutoff, rightCutoff)
|> (
x =>
switch x {
| #Solution(t) => Ok(t)
| #Error(e) => Error(e)
| #NoSolution => truncateAsShape(evaluationParams, leftCutoff, rightCutoff, t)
}
)
}
module Normalize = {
let rec operationToLeaf = (evaluationParams, t: node): result<node, string> =>
switch t {
| #RenderedDist(s) => Ok(#RenderedDist(PointSetDist.T.normalize(s)))
| #SymbolicDist(_) => Ok(t)
| _ => ASTTypes.Node.evaluateAndRetry(evaluationParams, operationToLeaf, t)
}
}
module FunctionCall = {
let _runHardcodedFunction = (name, evaluationParams, args) =>
TypeSystem.Function.Ts.findByNameAndRun(HardcodedFunctions.all, name, evaluationParams, args)
let _runLocalFunction = (name, evaluationParams: evaluationParams, args) =>
Environment.getFunction(evaluationParams.environment, name) |> E.R.bind(_, ((argNames, fn)) =>
ASTTypes.Function.run(evaluationParams, args, (argNames, fn))
)
let _runWithEvaluatedInputs = (
evaluationParams: ASTTypes.evaluationParams,
name,
args: array<ASTTypes.node>,
) =>
_runHardcodedFunction(name, evaluationParams, args) |> E.O.default(
_runLocalFunction(name, evaluationParams, args),
)
// TODO: This forces things to be floats
let run = (evaluationParams, name, args) =>
args
|> E.A.fmap(a => evaluationParams.evaluateNode(evaluationParams, a))
|> E.A.R.firstErrorOrOpen
|> E.R.bind(_, _runWithEvaluatedInputs(evaluationParams, name))
}
module Render = {
let rec operationToLeaf = (evaluationParams: evaluationParams, t: node): result<node, string> =>
switch t {
| #Function(_) => Error("Cannot render a function")
| #SymbolicDist(d) =>
Ok(
#RenderedDist(
SymbolicDist.T.toPointSetDist(evaluationParams.samplingInputs.pointSetDistLength, d),
),
)
| #RenderedDist(_) as t => Ok(t) // already a rendered pointSetDist, we're done here
| _ => ASTTypes.Node.evaluateAndRetry(evaluationParams, operationToLeaf, t)
}
}
/* This function recursively goes through the nodes of the parse tree,
replacing each Operation node and its subtree with a Data node.
Whenever possible, the replacement produces a new Symbolic Data node,
but most often it will produce a RenderedDist.
This function is used mainly to turn a parse tree into a single RenderedDist
that can then be displayed to the user. */
let rec toLeaf = (evaluationParams: ASTTypes.evaluationParams, node: node): result<node, string> =>
switch node {
// Leaf nodes just stay leaf nodes
| #SymbolicDist(_)
| #Function(_)
| #RenderedDist(_) =>
Ok(node)
| #Array(args) =>
args |> E.A.fmap(toLeaf(evaluationParams)) |> E.A.R.firstErrorOrOpen |> E.R.fmap(r => #Array(r))
// Operations nevaluationParamsd to be turned into leaves
| #AlgebraicCombination(algebraicOp, t1, t2) =>
AlgebraicCombination.operationToLeaf(evaluationParams, algebraicOp, t1, t2)
| #PointwiseCombination(pointwiseOp, t1, t2) =>
PointwiseCombination.operationToLeaf(evaluationParams, pointwiseOp, t1, t2)
| #Truncate(leftCutoff, rightCutoff, t) =>
Truncate.operationToLeaf(evaluationParams, leftCutoff, rightCutoff, t)
| #Normalize(t) => Normalize.operationToLeaf(evaluationParams, t)
| #Render(t) => Render.operationToLeaf(evaluationParams, t)
| #Hash(t) =>
t
|> E.A.fmap(((name: string, node: node)) =>
toLeaf(evaluationParams, node) |> E.R.fmap(r => (name, r))
)
|> E.A.R.firstErrorOrOpen
|> E.R.fmap(r => #Hash(r))
| #Symbol(r) =>
ASTTypes.Environment.get(evaluationParams.environment, r)
|> E.O.toResult("Undeclared variable " ++ r)
|> E.R.bind(_, toLeaf(evaluationParams))
| #FunctionCall(name, args) =>
FunctionCall.run(evaluationParams, name, args) |> E.R.bind(_, toLeaf(evaluationParams))
}

View File

@ -1,233 +0,0 @@
@genType
type rec hash = array<(string, node)>
and node = [
| #SymbolicDist(SymbolicDistTypes.symbolicDist)
| #RenderedDist(PointSetTypes.pointSetDist)
| #Symbol(string)
| #Hash(hash)
| #Array(array<node>)
| #Function(array<string>, node)
| #AlgebraicCombination(Operation.algebraicOperation, node, node)
| #PointwiseCombination(Operation.pointwiseOperation, node, node)
| #Normalize(node)
| #Render(node)
| #Truncate(option<float>, option<float>, node)
| #FunctionCall(string, array<node>)
]
type statement = [
| #Assignment(string, node)
| #Expression(node)
]
type program = array<statement>
type environment = Belt.Map.String.t<node>
type rec evaluationParams = {
samplingInputs: SamplingInputs.samplingInputs,
environment: environment,
evaluateNode: (evaluationParams, node) => Belt.Result.t<node, string>,
}
module Environment = {
type t = environment
module MS = Belt.Map.String
let fromArray = MS.fromArray
let empty: t = []->fromArray
let mergeKeepSecond = (a: t, b: t) =>
MS.merge(a, b, (_, a, b) =>
switch (a, b) {
| (_, Some(b)) => Some(b)
| (Some(a), _) => Some(a)
| _ => None
}
)
let update = (t, str, fn) => MS.update(t, str, fn)
let get = (t: t, str) => MS.get(t, str)
let getFunction = (t: t, str) =>
switch get(t, str) {
| Some(#Function(argNames, fn)) => Ok((argNames, fn))
| _ => Error("Function " ++ (str ++ " not found"))
}
}
module Node = {
let getFloat = (node: node) =>
node |> (
x =>
switch x {
| #RenderedDist(Discrete({xyShape: {xs: [x], ys: [1.0]}})) => Some(x)
| #SymbolicDist(#Float(x)) => Some(x)
| _ => None
}
)
let evaluate = (evaluationParams: evaluationParams) =>
evaluationParams.evaluateNode(evaluationParams)
let evaluateAndRetry = (evaluationParams, fn, node) =>
node |> evaluationParams.evaluateNode(evaluationParams) |> E.R.bind(_, fn(evaluationParams))
let rec toString: node => string = x =>
switch x {
| #SymbolicDist(d) => SymbolicDist.T.toString(d)
| #RenderedDist(_) => "[renderedShape]"
| #AlgebraicCombination(op, t1, t2) =>
Operation.Algebraic.format(op, toString(t1), toString(t2))
| #PointwiseCombination(op, t1, t2) =>
Operation.Pointwise.format(op, toString(t1), toString(t2))
| #Normalize(t) => "normalize(k" ++ (toString(t) ++ ")")
| #Truncate(lc, rc, t) => Operation.Truncate.toString(lc, rc, toString(t))
| #Render(t) => toString(t)
| #Symbol(t) => "Symbol: " ++ t
| #FunctionCall(name, args) =>
"[Function call: (" ++
(name ++
((args |> E.A.fmap(toString) |> Js.String.concatMany(_, ",")) ++ ")]"))
| #Function(args, internal) =>
"[Function: (" ++ ((args |> Js.String.concatMany(_, ",")) ++ (toString(internal) ++ ")]"))
| #Array(a) => "[" ++ ((a |> E.A.fmap(toString) |> Js.String.concatMany(_, ",")) ++ "]")
| #Hash(h) =>
"{" ++
((h
|> E.A.fmap(((name, value)) => name ++ (":" ++ toString(value)))
|> Js.String.concatMany(_, ",")) ++
"}")
}
let render = (evaluationParams: evaluationParams, r) => #Render(r) |> evaluate(evaluationParams)
let ensureIsRendered = (params, t) =>
switch t {
| #RenderedDist(_) => Ok(t)
| _ =>
switch render(params, t) {
| Ok(#RenderedDist(r)) => Ok(#RenderedDist(r))
| Ok(_) => Error("Did not render as requested")
| Error(e) => Error(e)
}
}
let ensureIsRenderedAndGetShape = (params, t) =>
switch ensureIsRendered(params, t) {
| Ok(#RenderedDist(r)) => Ok(r)
| Ok(_) => Error("Did not render as requested")
| Error(e) => Error(e)
}
let toPointSetDist = (item: node) =>
switch item {
| #RenderedDist(r) => Some(r)
| _ => None
}
let _toFloat = (t: PointSetTypes.pointSetDist) =>
switch t {
| Discrete({xyShape: {xs: [x], ys: [1.0]}}) => Some(#SymbolicDist(#Float(x)))
| _ => None
}
let toFloat = (item: node): result<node, string> =>
item |> toPointSetDist |> E.O.bind(_, _toFloat) |> E.O.toResult("Not valid shape")
}
module Function = {
type t = (array<string>, node)
let fromNode: node => option<t> = node =>
switch node {
| #Function(r) => Some(r)
| _ => None
}
let argumentNames = ((a, _): t) => a
let internals = ((_, b): t) => b
let run = (evaluationParams: evaluationParams, args: array<node>, t: t) =>
if E.A.length(args) == E.A.length(argumentNames(t)) {
let newEnvironment = Belt.Array.zip(argumentNames(t), args) |> Environment.fromArray
let newEvaluationParams: evaluationParams = {
samplingInputs: evaluationParams.samplingInputs,
environment: Environment.mergeKeepSecond(evaluationParams.environment, newEnvironment),
evaluateNode: evaluationParams.evaluateNode,
}
evaluationParams.evaluateNode(newEvaluationParams, internals(t))
} else {
Error("Wrong number of variables")
}
}
module SamplingDistribution = {
type t = [
| #SymbolicDist(SymbolicDistTypes.symbolicDist)
| #RenderedDist(PointSetTypes.pointSetDist)
]
let isSamplingDistribution: node => bool = x =>
switch x {
| #SymbolicDist(_) => true
| #RenderedDist(_) => true
| _ => false
}
let fromNode: node => result<t, string> = x =>
switch x {
| #SymbolicDist(n) => Ok(#SymbolicDist(n))
| #RenderedDist(n) => Ok(#RenderedDist(n))
| _ => Error("Not valid type")
}
let renderIfIsNotSamplingDistribution = (params, t): result<node, string> =>
!isSamplingDistribution(t)
? switch Node.render(params, t) {
| Ok(r) => Ok(r)
| Error(e) => Error(e)
}
: Ok(t)
let map = (~renderedDistFn, ~symbolicDistFn, node: node) =>
node |> (
x =>
switch x {
| #RenderedDist(r) => Some(renderedDistFn(r))
| #SymbolicDist(s) => Some(symbolicDistFn(s))
| _ => None
}
)
let sampleN = n =>
map(~renderedDistFn=PointSetDist.sampleNRendered(n), ~symbolicDistFn=SymbolicDist.T.sampleN(n))
let getCombinationSamples = (n, algebraicOp, t1: node, t2: node) =>
switch (sampleN(n, t1), sampleN(n, t2)) {
| (Some(a), Some(b)) =>
Some(
Belt.Array.zip(a, b) |> E.A.fmap(((a, b)) => Operation.Algebraic.toFn(algebraicOp, a, b)),
)
| _ => None
}
let combineShapesUsingSampling = (
evaluationParams: evaluationParams,
algebraicOp,
t1: node,
t2: node,
) => {
let i1 = renderIfIsNotSamplingDistribution(evaluationParams, t1)
let i2 = renderIfIsNotSamplingDistribution(evaluationParams, t2)
E.R.merge(i1, i2) |> E.R.bind(_, ((a, b)) => {
let samples =
getCombinationSamples(
evaluationParams.samplingInputs.sampleCount,
algebraicOp,
a,
b,
) |> E.O.toResult("Could not get samples")
let sampleSetDist = samples->E.R.bind(SampleSetDist.make)
let pointSetDist =
sampleSetDist->E.R.bind(r =>
SampleSetDist.toPointSetDist(~samplingInputs=evaluationParams.samplingInputs, ~samples=r)
)
pointSetDist |> E.R.fmap(r => #Normalize(#RenderedDist(r)))
})
}
}

View File

@ -1,87 +0,0 @@
open PointSetTypes
@genType
type t = PointSetTypes.distPlus
let pointSetDistIntegral = pointSetDist => PointSetDist.T.Integral.get(pointSetDist)
let make = (~pointSetDist, ~squiggleString, ()): t => {
let integral = pointSetDistIntegral(pointSetDist)
{pointSetDist: pointSetDist, integralCache: integral, squiggleString: squiggleString}
}
let update = (~pointSetDist=?, ~integralCache=?, ~squiggleString=?, t: t) => {
pointSetDist: E.O.default(t.pointSetDist, pointSetDist),
integralCache: E.O.default(t.integralCache, integralCache),
squiggleString: E.O.default(t.squiggleString, squiggleString),
}
let updateShape = (pointSetDist, t) => {
let integralCache = pointSetDistIntegral(pointSetDist)
update(~pointSetDist, ~integralCache, t)
}
let toPointSetDist = ({pointSetDist, _}: t) => pointSetDist
let pointSetDistFn = (fn, {pointSetDist}: t) => fn(pointSetDist)
module T = Distributions.Dist({
type t = PointSetTypes.distPlus
type integral = PointSetTypes.distPlus
let toPointSetDist = toPointSetDist
let toContinuous = pointSetDistFn(PointSetDist.T.toContinuous)
let toDiscrete = pointSetDistFn(PointSetDist.T.toDiscrete)
let normalize = (t: t): t => {
let normalizedShape = t |> toPointSetDist |> PointSetDist.T.normalize
t |> updateShape(normalizedShape)
}
let truncate = (leftCutoff, rightCutoff, t: t): t => {
let truncatedShape = t |> toPointSetDist |> PointSetDist.T.truncate(leftCutoff, rightCutoff)
t |> updateShape(truncatedShape)
}
let xToY = (f, t: t) => t |> toPointSetDist |> PointSetDist.T.xToY(f)
let minX = pointSetDistFn(PointSetDist.T.minX)
let maxX = pointSetDistFn(PointSetDist.T.maxX)
let toDiscreteProbabilityMassFraction = pointSetDistFn(
PointSetDist.T.toDiscreteProbabilityMassFraction,
)
// This bit is kind of awkward, could probably use rethinking.
let integral = (t: t) => updateShape(Continuous(t.integralCache), t)
let updateIntegralCache = (integralCache: option<PointSetTypes.continuousShape>, t) =>
update(~integralCache=E.O.default(t.integralCache, integralCache), t)
let downsample = (i, t): t => updateShape(t |> toPointSetDist |> PointSetDist.T.downsample(i), t)
// todo: adjust for limit, maybe?
let mapY = (
~integralSumCacheFn=previousIntegralSum => None,
~integralCacheFn=previousIntegralCache => None,
~fn,
{pointSetDist, _} as t: t,
): t => PointSetDist.T.mapY(~integralSumCacheFn, ~fn, pointSetDist) |> updateShape(_, t)
// get the total of everything
let integralEndY = (t: t) => {
PointSetDist.T.Integral.sum(toPointSetDist(t))
}
// TODO: Fix this below, obviously. Adjust for limits
let integralXtoY = (f, t: t) => {
PointSetDist.T.Integral.xToY(f, toPointSetDist(t))
}
// TODO: This part is broken when there is a limit, if this is supposed to be taken into account.
let integralYtoX = (f, t: t) => {
PointSetDist.T.Integral.yToX(f, toPointSetDist(t))
}
let mean = (t: t) => {
PointSetDist.T.mean(t.pointSetDist)
}
let variance = (t: t) => PointSetDist.T.variance(t.pointSetDist)
})

View File

@ -1,240 +0,0 @@
open TypeSystem
let wrongInputsError = (r: array<typedValue>) => {
let inputs = r |> E.A.fmap(TypedValue.toString) |> Js.String.concatMany(_, ",")
Js.log3("Inputs were", inputs, r)
Error("Wrong inputs. The inputs were:" ++ inputs)
}
let to_: (float, float) => result<node, string> = (low, high) =>
switch (low, high) {
| (low, high) if low <= 0.0 && low < high =>
Ok(#SymbolicDist(SymbolicDist.Normal.from90PercentCI(low, high)))
| (low, high) if low < high =>
Ok(#SymbolicDist(SymbolicDist.Lognormal.from90PercentCI(low, high)))
| (_, _) => Error("Low value must be less than high value.")
}
let makeSymbolicFromTwoFloats = (name, fn) =>
Function.T.make(
~name,
~outputType=#SamplingDistribution,
~inputTypes=[#Float, #Float],
~run=x =>
switch x {
| [#Float(a), #Float(b)] => fn(a, b) |> E.R.fmap(r => #SymbolicDist(r))
| e => wrongInputsError(e)
},
(),
)
let makeSymbolicFromOneFloat = (name, fn) =>
Function.T.make(
~name,
~outputType=#SamplingDistribution,
~inputTypes=[#Float],
~run=x =>
switch x {
| [#Float(a)] => fn(a) |> E.R.fmap(r => #SymbolicDist(r))
| e => wrongInputsError(e)
},
(),
)
let makeDistFloat = (name, fn) =>
Function.T.make(
~name,
~outputType=#SamplingDistribution,
~inputTypes=[#SamplingDistribution, #Float],
~run=x =>
switch x {
| [#SamplingDist(a), #Float(b)] => fn(a, b)
| [#RenderedDist(a), #Float(b)] => fn(#RenderedDist(a), b)
| e => wrongInputsError(e)
},
(),
)
let makeRenderedDistFloat = (name, fn) =>
Function.T.make(
~name,
~outputType=#RenderedDistribution,
~inputTypes=[#RenderedDistribution, #Float],
~shouldCoerceTypes=true,
~run=x =>
switch x {
| [#RenderedDist(a), #Float(b)] => fn(a, b)
| e => wrongInputsError(e)
},
(),
)
let makeDist = (name, fn) =>
Function.T.make(
~name,
~outputType=#SamplingDistribution,
~inputTypes=[#SamplingDistribution],
~run=x =>
switch x {
| [#SamplingDist(a)] => fn(a)
| [#RenderedDist(a)] => fn(#RenderedDist(a))
| e => wrongInputsError(e)
},
(),
)
let floatFromDist = (
distToFloatOp: Operation.distToFloatOperation,
t: TypeSystem.samplingDist,
): result<node, string> =>
switch t {
| #SymbolicDist(s) =>
SymbolicDist.T.operate(distToFloatOp, s) |> E.R.bind(_, v => Ok(#SymbolicDist(#Float(v))))
| #RenderedDist(rs) =>
PointSetDist.operate(distToFloatOp, rs) |> (v => Ok(#SymbolicDist(#Float(v))))
}
let verticalScaling = (scaleOp, rs, scaleBy) => {
// scaleBy has to be a single float, otherwise we'll return an error.
let fn = (secondary, main) => Operation.Scale.toFn(scaleOp, main, secondary)
let integralSumCacheFn = Operation.Scale.toIntegralSumCacheFn(scaleOp)
let integralCacheFn = Operation.Scale.toIntegralCacheFn(scaleOp)
Ok(
#RenderedDist(
PointSetDist.T.mapY(
~integralSumCacheFn=integralSumCacheFn(scaleBy),
~integralCacheFn=integralCacheFn(scaleBy),
~fn=fn(scaleBy),
rs,
),
),
)
}
module Multimodal = {
let getByNameResult = Hash.getByNameResult
let _paramsToDistsAndWeights = (r: array<typedValue>) =>
switch r {
| [#Hash(r)] =>
let dists =
getByNameResult(r, "dists")
->E.R.bind(TypeSystem.TypedValue.toArray)
->E.R.bind(r => r |> E.A.fmap(TypeSystem.TypedValue.toDist) |> E.A.R.firstErrorOrOpen)
let weights =
getByNameResult(r, "weights")
->E.R.bind(TypeSystem.TypedValue.toArray)
->E.R.bind(r => r |> E.A.fmap(TypeSystem.TypedValue.toFloat) |> E.A.R.firstErrorOrOpen)
E.R.merge(dists, weights)->E.R.bind(((a, b)) =>
E.A.length(b) > E.A.length(a)
? Error("Too many weights provided")
: Ok(
E.A.zipMaxLength(a, b) |> E.A.fmap(((a, b)) => (
a |> E.O.toExn(""),
b |> E.O.default(1.0),
)),
)
)
| _ => Error("Needs items")
}
let _runner: array<typedValue> => result<node, string> = r => {
let paramsToDistsAndWeights =
_paramsToDistsAndWeights(r) |> E.R.fmap(
E.A.fmap(((dist, weight)) =>
#FunctionCall("scaleMultiply", [dist, #SymbolicDist(#Float(weight))])
),
)
let pointwiseSum: result<node, string> =
paramsToDistsAndWeights->E.R.bind(E.R.errorIfCondition(E.A.isEmpty, "Needs one input"))
|> E.R.fmap(r =>
r
|> Js.Array.sliceFrom(1)
|> E.A.fold_left((acc, x) => #PointwiseCombination(#Add, acc, x), E.A.unsafe_get(r, 0))
)
pointwiseSum
}
let _function = Function.T.make(
~name="multimodal",
~outputType=#SamplingDistribution,
~inputTypes=[#Hash([("dists", #Array(#SamplingDistribution)), ("weights", #Array(#Float))])],
~run=_runner,
(),
)
}
let all = [
makeSymbolicFromTwoFloats("normal", SymbolicDist.Normal.make),
makeSymbolicFromTwoFloats("uniform", SymbolicDist.Uniform.make),
makeSymbolicFromTwoFloats("beta", SymbolicDist.Beta.make),
makeSymbolicFromTwoFloats("lognormal", SymbolicDist.Lognormal.make),
makeSymbolicFromTwoFloats("lognormalFromMeanAndStdDev", SymbolicDist.Lognormal.fromMeanAndStdev),
makeSymbolicFromOneFloat("exponential", SymbolicDist.Exponential.make),
Function.T.make(
~name="to",
~outputType=#SamplingDistribution,
~inputTypes=[#Float, #Float],
~run=x =>
switch x {
| [#Float(a), #Float(b)] => to_(a, b)
| e => wrongInputsError(e)
},
(),
),
Function.T.make(
~name="triangular",
~outputType=#SamplingDistribution,
~inputTypes=[#Float, #Float, #Float],
~run=x =>
switch x {
| [#Float(a), #Float(b), #Float(c)] =>
SymbolicDist.Triangular.make(a, b, c) |> E.R.fmap(r => #SymbolicDist(r))
| e => wrongInputsError(e)
},
(),
),
Function.T.make(
~name="log",
~outputType=#Float,
~inputTypes=[#Float],
~run=x =>
switch x {
| [#Float(a)] => Ok(#SymbolicDist(#Float(Js.Math.log(a))))
| e => wrongInputsError(e)
},
(),
),
makeDistFloat("pdf", (dist, float) => floatFromDist(#Pdf(float), dist)),
makeDistFloat("inv", (dist, float) => floatFromDist(#Inv(float), dist)),
makeDistFloat("cdf", (dist, float) => floatFromDist(#Cdf(float), dist)),
makeDist("mean", dist => floatFromDist(#Mean, dist)),
makeDist("sample", dist => floatFromDist(#Sample, dist)),
Function.T.make(
~name="render",
~outputType=#RenderedDistribution,
~inputTypes=[#RenderedDistribution],
~run=x =>
switch x {
| [#RenderedDist(c)] => Ok(#RenderedDist(c))
| e => wrongInputsError(e)
},
(),
),
Function.T.make(
~name="normalize",
~outputType=#SamplingDistribution,
~inputTypes=[#SamplingDistribution],
~run=x =>
switch x {
| [#SamplingDist(#SymbolicDist(c))] => Ok(#SymbolicDist(c))
| [#SamplingDist(#RenderedDist(c))] => Ok(#RenderedDist(PointSetDist.T.normalize(c)))
| e => wrongInputsError(e)
},
(),
),
makeRenderedDistFloat("scaleExp", (dist, float) => verticalScaling(#Power, dist, float)),
makeRenderedDistFloat("scaleMultiply", (dist, float) => verticalScaling(#Multiply, dist, float)),
makeRenderedDistFloat("scaleLog", (dist, float) => verticalScaling(#Logarithm, dist, float)),
Multimodal._function,
]

View File

@ -1,196 +0,0 @@
type node = ASTTypes.node
let getFloat = ASTTypes.Node.getFloat
type samplingDist = [
| #SymbolicDist(SymbolicDistTypes.symbolicDist)
| #RenderedDist(PointSetTypes.pointSetDist)
]
type rec hashType = array<(string, _type)>
and _type = [
| #Float
| #SamplingDistribution
| #RenderedDistribution
| #Array(_type)
| #Hash(hashType)
]
type rec hashTypedValue = array<(string, typedValue)>
and typedValue = [
| #Float(float)
| #RenderedDist(PointSetTypes.pointSetDist)
| #SamplingDist(samplingDist)
| #Array(array<typedValue>)
| #Hash(hashTypedValue)
]
type _function = {
name: string,
inputTypes: array<_type>,
outputType: _type,
run: array<typedValue> => result<node, string>,
shouldCoerceTypes: bool,
}
type functions = array<_function>
type inputNodes = array<node>
module TypedValue = {
let rec toString: typedValue => string = x =>
switch x {
| #SamplingDist(_) => "[sampling dist]"
| #RenderedDist(_) => "[rendered PointSetDist]"
| #Float(f) => "Float: " ++ Js.Float.toString(f)
| #Array(a) => "[" ++ ((a |> E.A.fmap(toString) |> Js.String.concatMany(_, ",")) ++ "]")
| #Hash(v) =>
"{" ++
((v
|> E.A.fmap(((name, value)) => name ++ (":" ++ toString(value)))
|> Js.String.concatMany(_, ",")) ++
"}")
}
let rec fromNode = (node: node): result<typedValue, string> =>
switch node {
| #SymbolicDist(#Float(r)) => Ok(#Float(r))
| #SymbolicDist(s) => Ok(#SamplingDist(#SymbolicDist(s)))
| #RenderedDist(s) => Ok(#RenderedDist(s))
| #Array(r) => r |> E.A.fmap(fromNode) |> E.A.R.firstErrorOrOpen |> E.R.fmap(r => #Array(r))
| #Hash(hash) =>
hash
|> E.A.fmap(((name, t)) => fromNode(t) |> E.R.fmap(r => (name, r)))
|> E.A.R.firstErrorOrOpen
|> E.R.fmap(r => #Hash(r))
| e => Error("Wrong type: " ++ ASTTypes.Node.toString(e))
}
// todo: Arrays and hashes
let rec fromNodeWithTypeCoercion = (evaluationParams, _type: _type, node) =>
switch (_type, node) {
| (#Float, _) =>
switch getFloat(node) {
| Some(a) => Ok(#Float(a))
| _ => Error("Type Error: Expected float.")
}
| (#SamplingDistribution, _) =>
ASTTypes.SamplingDistribution.renderIfIsNotSamplingDistribution(
evaluationParams,
node,
) |> E.R.bind(_, fromNode)
| (#RenderedDistribution, _) =>
ASTTypes.Node.render(evaluationParams, node) |> E.R.bind(_, fromNode)
| (#Array(_type), #Array(b)) =>
b
|> E.A.fmap(fromNodeWithTypeCoercion(evaluationParams, _type))
|> E.A.R.firstErrorOrOpen
|> E.R.fmap(r => #Array(r))
| (#Hash(named), #Hash(r)) =>
let keyValues =
named |> E.A.fmap(((name, intendedType)) => (name, intendedType, Hash.getByName(r, name)))
let typedHash =
keyValues
|> E.A.fmap(((name, intendedType, optionNode)) =>
switch optionNode {
| Some(node) =>
fromNodeWithTypeCoercion(evaluationParams, intendedType, node) |> E.R.fmap(node => (
name,
node,
))
| None => Error("Hash parameter not present in hash.")
}
)
|> E.A.R.firstErrorOrOpen
|> E.R.fmap(r => #Hash(r))
typedHash
| _ => Error("fromNodeWithTypeCoercion error, sorry.")
}
let toFloat: typedValue => result<float, string> = x =>
switch x {
| #Float(x) => Ok(x)
| _ => Error("Not a float")
}
let toArray: typedValue => result<array<'a>, string> = x =>
switch x {
| #Array(x) => Ok(x)
| _ => Error("Not an array")
}
let toNamed: typedValue => result<hashTypedValue, string> = x =>
switch x {
| #Hash(x) => Ok(x)
| _ => Error("Not a named item")
}
let toDist: typedValue => result<node, string> = x =>
switch x {
| #SamplingDist(#SymbolicDist(c)) => Ok(#SymbolicDist(c))
| #SamplingDist(#RenderedDist(c)) => Ok(#RenderedDist(c))
| #RenderedDist(c) => Ok(#RenderedDist(c))
| #Float(x) => Ok(#SymbolicDist(#Float(x)))
| x => Error("Cannot be converted into a distribution: " ++ toString(x))
}
}
module Function = {
type t = _function
type ts = functions
module T = {
let make = (~name, ~inputTypes, ~outputType, ~run, ~shouldCoerceTypes=true, _): t => {
name: name,
inputTypes: inputTypes,
outputType: outputType,
run: run,
shouldCoerceTypes: shouldCoerceTypes,
}
let _inputLengthCheck = (inputNodes: inputNodes, t: t) => {
let expectedLength = E.A.length(t.inputTypes)
let actualLength = E.A.length(inputNodes)
expectedLength == actualLength
? Ok(inputNodes)
: Error(
"Wrong number of inputs. Expected" ++
((expectedLength |> E.I.toString) ++
(". Got:" ++ (actualLength |> E.I.toString))),
)
}
let _coerceInputNodes = (evaluationParams, inputTypes, shouldCoerce, inputNodes) =>
Belt.Array.zip(inputTypes, inputNodes)
|> E.A.fmap(((def, input)) =>
shouldCoerce
? TypedValue.fromNodeWithTypeCoercion(evaluationParams, def, input)
: TypedValue.fromNode(input)
)
|> E.A.R.firstErrorOrOpen
let inputsToTypedValues = (
evaluationParams: ASTTypes.evaluationParams,
inputNodes: inputNodes,
t: t,
) =>
_inputLengthCheck(inputNodes, t)->E.R.bind(
_coerceInputNodes(evaluationParams, t.inputTypes, t.shouldCoerceTypes),
)
let run = (evaluationParams: ASTTypes.evaluationParams, inputNodes: inputNodes, t: t) =>
inputsToTypedValues(evaluationParams, inputNodes, t)->E.R.bind(t.run)
|> (
x =>
switch x {
| Ok(i) => Ok(i)
| Error(r) => Error("Function " ++ (t.name ++ (" error: " ++ r)))
}
)
}
module Ts = {
let findByName = (ts: ts, n: string) => ts |> Belt.Array.getBy(_, ({name}) => name == n)
let findByNameAndRun = (ts: ts, n: string, evaluationParams, inputTypes) =>
findByName(ts, n) |> E.O.fmap(T.run(evaluationParams, inputTypes))
}
}

View File

@ -1,290 +0,0 @@
module MathJsonToMathJsAdt = {
type rec arg =
| Symbol(string)
| Value(float)
| Fn(fn)
| Array(array<arg>)
| Blocks(array<arg>)
| Object(Js.Dict.t<arg>)
| Assignment(arg, arg)
| FunctionAssignment(fnAssignment)
and fn = {
name: string,
args: array<arg>,
}
and fnAssignment = {
name: string,
args: array<string>,
expression: arg,
}
let rec run = (j: Js.Json.t) => {
open Json.Decode
switch field("mathjs", string, j) {
| "FunctionNode" =>
let args = j |> field("args", array(run))
let name = j |> optional(field("fn", field("name", string)))
name |> E.O.fmap(name => Fn({name: name, args: args |> E.A.O.concatSomes}))
| "OperatorNode" =>
let args = j |> field("args", array(run))
Some(
Fn({
name: j |> field("fn", string),
args: args |> E.A.O.concatSomes,
}),
)
| "ConstantNode" => optional(field("value", Json.Decode.float), j) |> E.O.fmap(r => Value(r))
| "ParenthesisNode" => j |> field("content", run)
| "ObjectNode" =>
let properties = j |> field("properties", dict(run))
Js.Dict.entries(properties)
|> E.A.fmap(((key, value)) => value |> E.O.fmap(v => (key, v)))
|> E.A.O.concatSomes
|> Js.Dict.fromArray
|> (r => Some(Object(r)))
| "ArrayNode" =>
let items = field("items", array(run), j)
Some(Array(items |> E.A.O.concatSomes))
| "SymbolNode" => Some(Symbol(field("name", string, j)))
| "AssignmentNode" =>
let object_ = j |> field("object", run)
let value_ = j |> field("value", run)
switch (object_, value_) {
| (Some(o), Some(v)) => Some(Assignment(o, v))
| _ => None
}
| "BlockNode" =>
let block = r => r |> field("node", run)
let args = j |> field("blocks", array(block)) |> E.A.O.concatSomes
Some(Blocks(args))
| "FunctionAssignmentNode" =>
let name = j |> field("name", string)
let args = j |> field("params", array(field("name", string)))
let expression = j |> field("expr", run)
expression |> E.O.fmap(expression => FunctionAssignment({
name: name,
args: args,
expression: expression,
}))
| n =>
Js.log3("Couldn't parse mathjs node", j, n)
None
}
}
}
module MathAdtToDistDst = {
open MathJsonToMathJsAdt
let handleSymbol = sym => Ok(#Symbol(sym))
// TODO: This only works on the top level, which needs to be refactored. Also, I think functions don't need to be done like this anymore.
module MathAdtCleaner = {
let transformWithSymbol = (f: float, s: string) =>
switch s {
| "K" => Some(f *. 1000.)
| "M" => Some(f *. 1000000.)
| "B" => Some(f *. 1000000000.)
| "T" => Some(f *. 1000000000000.)
| _ => None
}
let rec run = x =>
switch x {
| Fn({name: "multiply", args: [Value(f), Symbol(s)]}) as doNothing =>
transformWithSymbol(f, s) |> E.O.fmap(r => Value(r)) |> E.O.default(doNothing)
| Fn({name: "unaryMinus", args: [Value(f)]}) => Value(-1.0 *. f)
| Fn({name, args}) => Fn({name: name, args: args |> E.A.fmap(run)})
| Array(args) => Array(args |> E.A.fmap(run))
| Symbol(s) => Symbol(s)
| Value(v) => Value(v)
| Blocks(args) => Blocks(args |> E.A.fmap(run))
| Assignment(a, b) => Assignment(a, run(b))
| FunctionAssignment(a) => FunctionAssignment(a)
| Object(v) =>
Object(
v
|> Js.Dict.entries
|> E.A.fmap(((key, value)) => (key, run(value)))
|> Js.Dict.fromArray,
)
}
}
let lognormal = (args, parseArgs, nodeParser) =>
switch args {
| [Object(o)] =>
let g = s =>
Js.Dict.get(o, s) |> E.O.toResult("Variable was empty") |> E.R.bind(_, nodeParser)
switch (g("mean"), g("stdev"), g("mu"), g("sigma")) {
| (Ok(mean), Ok(stdev), _, _) =>
Ok(#FunctionCall("lognormalFromMeanAndStdDev", [mean, stdev]))
| (_, _, Ok(mu), Ok(sigma)) => Ok(#FunctionCall("lognormal", [mu, sigma]))
| _ => Error("Lognormal distribution needs either mean and stdev or mu and sigma")
}
| _ => parseArgs() |> E.R.fmap((args: array<ASTTypes.node>) => #FunctionCall("lognormal", args))
}
// Error("Dotwise exponentiation needs two operands")
let operationParser = (name: string, args: result<array<ASTTypes.node>, string>): result<
ASTTypes.node,
string,
> => {
let toOkAlgebraic = r => Ok(#AlgebraicCombination(r))
let toOkPointwise = r => Ok(#PointwiseCombination(r))
let toOkTruncate = r => Ok(#Truncate(r))
args |> E.R.bind(_, args =>
switch (name, args) {
| ("add", [l, r]) => toOkAlgebraic((#Add, l, r))
| ("add", _) => Error("Addition needs two operands")
| ("unaryMinus", [l]) => toOkAlgebraic((#Multiply, #SymbolicDist(#Float(-1.0)), l))
| ("subtract", [l, r]) => toOkAlgebraic((#Subtract, l, r))
| ("subtract", _) => Error("Subtraction needs two operands")
| ("multiply", [l, r]) => toOkAlgebraic((#Multiply, l, r))
| ("multiply", _) => Error("Multiplication needs two operands")
| ("pow", [l, r]) => toOkAlgebraic((#Power, l, r))
| ("pow", _) => Error("Exponentiation needs two operands")
| ("dotMultiply", [l, r]) => toOkPointwise((#Multiply, l, r))
| ("dotMultiply", _) => Error("Dotwise multiplication needs two operands")
| ("dotPow", [l, r]) => toOkPointwise((#Power, l, r))
| ("dotPow", _) => Error("Dotwise exponentiation needs two operands")
| ("rightLogShift", [l, r]) => toOkPointwise((#Add, l, r))
| ("rightLogShift", _) => Error("Dotwise addition needs two operands")
| ("divide", [l, r]) => toOkAlgebraic((#Divide, l, r))
| ("divide", _) => Error("Division needs two operands")
| ("leftTruncate", [d, #SymbolicDist(#Float(lc))]) => toOkTruncate((Some(lc), None, d))
| ("leftTruncate", _) =>
Error("leftTruncate needs two arguments: the expression and the cutoff")
| ("rightTruncate", [d, #SymbolicDist(#Float(rc))]) => toOkTruncate((None, Some(rc), d))
| ("rightTruncate", _) =>
Error("rightTruncate needs two arguments: the expression and the cutoff")
| ("truncate", [d, #SymbolicDist(#Float(lc)), #SymbolicDist(#Float(rc))]) =>
toOkTruncate((Some(lc), Some(rc), d))
| ("truncate", _) => Error("truncate needs three arguments: the expression and both cutoffs")
| _ => Error("This type not currently supported")
}
)
}
let functionParser = (
nodeParser: MathJsonToMathJsAdt.arg => Belt.Result.t<ASTTypes.node, string>,
name: string,
args: array<MathJsonToMathJsAdt.arg>,
): result<ASTTypes.node, string> => {
let parseArray = ags => ags |> E.A.fmap(nodeParser) |> E.A.R.firstErrorOrOpen
let parseArgs = () => parseArray(args)
switch name {
| "lognormal" => lognormal(args, parseArgs, nodeParser)
| "multimodal"
| "add"
| "subtract"
| "multiply"
| "unaryMinus"
| "dotMultiply"
| "dotPow"
| "rightLogShift"
| "divide"
| "pow"
| "leftTruncate"
| "rightTruncate"
| "truncate" =>
operationParser(name, parseArgs())
| "mm" =>
let weights =
args
|> E.A.last
|> E.O.bind(_, x =>
switch x {
| Array(values) => Some(parseArray(values))
| _ => None
}
)
let possibleDists = E.O.isSome(weights)
? Belt.Array.slice(args, ~offset=0, ~len=E.A.length(args) - 1)
: args
let dists = parseArray(possibleDists)
switch (weights, dists) {
| (Some(Error(r)), _) => Error(r)
| (_, Error(r)) => Error(r)
| (None, Ok(dists)) =>
let hash: ASTTypes.node = #FunctionCall(
"multimodal",
[#Hash([("dists", #Array(dists)), ("weights", #Array([]))])],
)
Ok(hash)
| (Some(Ok(weights)), Ok(dists)) =>
let hash: ASTTypes.node = #FunctionCall(
"multimodal",
[#Hash([("dists", #Array(dists)), ("weights", #Array(weights))])],
)
Ok(hash)
}
| name => parseArgs() |> E.R.fmap((args: array<ASTTypes.node>) => #FunctionCall(name, args))
}
}
let rec nodeParser: MathJsonToMathJsAdt.arg => result<ASTTypes.node, string> = x =>
switch x {
| Value(f) => Ok(#SymbolicDist(#Float(f)))
| Symbol(sym) => Ok(#Symbol(sym))
| Fn({name, args}) => functionParser(nodeParser, name, args)
| _ => Error("This type not currently supported")
}
// | FunctionAssignment({name, args, expression}) => {
// let evaluatedExpression = run(expression);
// `Function(_ => Ok(evaluatedExpression));
// }
let rec topLevel = (r): result<ASTTypes.program, string> =>
switch r {
| FunctionAssignment({name, args, expression}) =>
switch nodeParser(expression) {
| Ok(r) => Ok([#Assignment(name, #Function(args, r))])
| Error(r) => Error(r)
}
| Value(_) as r => nodeParser(r) |> E.R.fmap(r => [#Expression(r)])
| Fn(_) as r => nodeParser(r) |> E.R.fmap(r => [#Expression(r)])
| Array(_) => Error("Array not valid as top level")
| Symbol(s) => handleSymbol(s) |> E.R.fmap(r => [#Expression(r)])
| Object(_) => Error("Object not valid as top level")
| Assignment(name, value) =>
switch name {
| Symbol(symbol) => nodeParser(value) |> E.R.fmap(r => [#Assignment(symbol, r)])
| _ => Error("Symbol not a string")
}
| Blocks(blocks) =>
blocks |> E.A.fmap(b => topLevel(b)) |> E.A.R.firstErrorOrOpen |> E.R.fmap(E.A.concatMany)
}
let run = (r): result<ASTTypes.program, string> => r |> MathAdtCleaner.run |> topLevel
}
/* The MathJs parser doesn't support '.+' syntax, but we want it because it
would make sense with '.*'. Our workaround is to change this to >>>, which is
logShift in mathJS. We don't expect to use logShift anytime soon, so this tradeoff
seems fine.
*/
let pointwiseToRightLogShift = Js.String.replaceByRe(%re("/\.\+/g"), ">>>")
let fromString2 = str => {
/* We feed the user-typed string into Mathjs.parseMath,
which returns a JSON with (hopefully) a single-element array.
This array element is the top-level node of a nested-object tree
representing the functions/arguments/values/etc. in the string.
The function MathJsonToMathJsAdt then recursively unpacks this JSON into a typed data structure we can use.
Inside of this function, MathAdtToDistDst is called whenever a distribution function is encountered.
*/
let mathJsToJson = str |> pointwiseToRightLogShift |> Mathjs.parseMath
let mathJsParse = E.R.bind(mathJsToJson, r =>
switch MathJsonToMathJsAdt.run(r) {
| Some(r) => Ok(r)
| None => Error("MathJsParse Error")
}
)
let value = E.R.bind(mathJsParse, MathAdtToDistDst.run)
value
}
let fromString = str => fromString2(str)

View File

@ -1,185 +0,0 @@
// TODO: This setup is more confusing than it should be, there's more work to do in cleanup here.
module Inputs = {
module SamplingInputs = {
type t = {
sampleCount: option<int>,
outputXYPoints: option<int>,
kernelWidth: option<float>,
pointDistLength: option<int>,
}
}
let defaultRecommendedLength = 100
let defaultShouldDownsample = true
type inputs = {
squiggleString: string,
samplingInputs: SamplingInputs.t,
environment: ASTTypes.environment,
}
let empty: SamplingInputs.t = {
sampleCount: None,
outputXYPoints: None,
kernelWidth: None,
pointDistLength: None,
}
let make = (
~samplingInputs=empty,
~squiggleString,
~environment=ASTTypes.Environment.empty,
(),
): inputs => {
samplingInputs: samplingInputs,
squiggleString: squiggleString,
environment: environment,
}
}
type exportDistribution = [
| #DistPlus(DistPlus.t)
| #Float(float)
| #Function(float => Belt.Result.t<DistPlus.t, string>)
]
type exportEnv = array<(string, ASTTypes.node)>
type exportType = {
environment: exportEnv,
exports: array<exportDistribution>,
}
module Internals = {
let addVariable = (
{samplingInputs, squiggleString, environment}: Inputs.inputs,
str,
node,
): Inputs.inputs => {
samplingInputs: samplingInputs,
squiggleString: squiggleString,
environment: ASTTypes.Environment.update(environment, str, _ => Some(node)),
}
type outputs = {
graph: ASTTypes.node,
pointSetDist: PointSetTypes.pointSetDist,
}
let makeOutputs = (graph, shape): outputs => {graph: graph, pointSetDist: shape}
let makeInputs = (inputs: Inputs.inputs): SamplingInputs.samplingInputs => {
sampleCount: inputs.samplingInputs.sampleCount |> E.O.default(10000),
outputXYPoints: inputs.samplingInputs.outputXYPoints |> E.O.default(10000),
kernelWidth: inputs.samplingInputs.kernelWidth,
pointSetDistLength: inputs.samplingInputs.pointDistLength |> E.O.default(10000),
}
let runNode = (inputs, node) => AST.toLeaf(makeInputs(inputs), inputs.environment, node)
let renderIfNeeded = (inputs: Inputs.inputs, node: ASTTypes.node): result<
ASTTypes.node,
string,
> =>
node |> (
x =>
switch x {
| #Normalize(_) as n
| #SymbolicDist(_) as n =>
#Render(n)
|> runNode(inputs)
|> (
x =>
switch x {
| Ok(#RenderedDist(_)) as r => r
| Error(r) => Error(r)
| _ => Error("Didn't render, but intended to")
}
)
| n => Ok(n)
}
)
let outputToDistPlus = (inputs: Inputs.inputs, pointSetDist: PointSetTypes.pointSetDist) =>
DistPlus.make(~pointSetDist, ~squiggleString=Some(inputs.squiggleString), ())
let rec returnDist = (
functionInfo: (array<string>, ASTTypes.node),
inputs: Inputs.inputs,
env: ASTTypes.environment,
) => {
(input: float) => {
let foo: Inputs.inputs = {...inputs, environment: env}
evaluateFunction(foo, functionInfo, [#SymbolicDist(#Float(input))]) |> E.R.bind(_, a =>
switch a {
| #DistPlus(d) => Ok(DistPlus.T.normalize(d))
| n =>
Js.log2("Error here", n)
Error("wrong type")
}
)
}
}
// TODO: Consider using ExpressionTypes.ExpressionTree.getFloat or similar in this function
and coersionToExportedTypes = (inputs, env: ASTTypes.environment, ex: ASTTypes.node): result<
exportDistribution,
string,
> =>
ex
|> renderIfNeeded(inputs)
|> E.R.bind(_, x =>
switch x {
| #RenderedDist(Discrete({xyShape: {xs: [x], ys: [1.0]}})) => Ok(#Float(x))
| #SymbolicDist(#Float(x)) => Ok(#Float(x))
| #RenderedDist(n) => Ok(#DistPlus(outputToDistPlus(inputs, n)))
| #Function(n) => Ok(#Function(returnDist(n, inputs, env)))
| n => Error("Didn't output a rendered distribution. Format:" ++ AST.toString(n))
}
)
and evaluateFunction = (inputs: Inputs.inputs, fn: (array<string>, ASTTypes.node), fnInputs) => {
let output = AST.runFunction(makeInputs(inputs), inputs.environment, fnInputs, fn)
output |> E.R.bind(_, coersionToExportedTypes(inputs, inputs.environment))
}
let runProgram = (inputs: Inputs.inputs, p: ASTTypes.program) => {
let ins = ref(inputs)
p
|> E.A.fmap(x =>
switch x {
| #Assignment(name, node) =>
ins := addVariable(ins.contents, name, node)
None
| #Expression(node) => Some(runNode(ins.contents, node))
}
)
|> E.A.O.concatSomes
|> E.A.R.firstErrorOrOpen
|> E.R.bind(_, d =>
d
|> E.A.fmap(x => coersionToExportedTypes(inputs, ins.contents.environment, x))
|> E.A.R.firstErrorOrOpen
)
|> E.R.fmap(ex => {
environment: Belt.Map.String.toArray(ins.contents.environment),
exports: ex,
})
}
let inputsToLeaf = (inputs: Inputs.inputs) =>
Parser.fromString(inputs.squiggleString) |> E.R.bind(_, g => runProgram(inputs, g))
}
@genType
let runAll: (string, Inputs.SamplingInputs.t, exportEnv) => result<exportType, string> = (
squiggleString,
samplingInputs,
environment,
) => {
let inputs = Inputs.make(
~samplingInputs,
~squiggleString,
~environment=Belt.Map.String.fromArray(environment),
(),
)
Internals.inputsToLeaf(inputs)
}

View File

@ -9,6 +9,13 @@ type algebraicOperation = [
| #Power
| #Logarithm
]
type convolutionOperation = [
| #Add
| #Multiply
| #Subtract
]
@genType
type pointwiseOperation = [#Add | #Multiply | #Power]
type scaleOperation = [#Multiply | #Power | #Logarithm | #Divide]
@ -20,6 +27,16 @@ type distToFloatOperation = [
| #Sample
]
module Convolution = {
type t = convolutionOperation
let toFn: (t, float, float) => float = x =>
switch x {
| #Add => \"+."
| #Subtract => \"-."
| #Multiply => \"*."
}
}
module Algebraic = {
type t = algebraicOperation
let toFn: (t, float, float) => float = x =>