Compare commits

...

1 Commits

Author SHA1 Message Date
Sam Nolan
59595d2e4b Partial function parsing code 2022-05-20 00:34:30 +00:00
6 changed files with 256 additions and 95 deletions

View File

@ -6,13 +6,63 @@ type genericDist =
type asAlgebraicCombinationStrategy = AsDefault | AsSymbolic | AsMonteCarlo | AsConvolution
type expressionType =
| ArrayType
| ArrayStringType
| BoolType
| CallType
| DistributionType
| LambdaType
| NumberType
| RecordType
| StringType
| SymbolType
type argumentError =
| WrongTypeError(expressionType, expressionType)
| IncorrectNumberOfArgumentsError(int, int)
| MustBeFinite
| MustBePositive
| OtherArgumentError(string)
module ArgumentError = {
type t = argumentError
let expressionTypeToString = (eType: expressionType): string =>
switch eType {
| ArrayType => "array"
| ArrayStringType => "arraystring"
| BoolType => "boolean"
| CallType => "call"
| DistributionType => "distribution"
| LambdaType => "lambda"
| NumberType => "number"
| RecordType => "record"
| StringType => "string"
| SymbolType => "symbol"
}
let toString = (err: t) : string =>
switch err {
| WrongTypeError(expected, actual) =>
`Argument has wrong type. Expected ${expressionTypeToString(expected)} but got ${expressionTypeToString(actual)}`
| IncorrectNumberOfArgumentsError(expected, actual) => `Expected ${Belt.Int.toString(expected)} arguments but got ${Belt.Int.toString(actual)}`
| MustBeFinite => "Argument must be finite"
| MustBePositive => "Argument must be positive"
| OtherArgumentError(msg) => msg
}
}
@genType
type error =
| NotYetImplemented
| Unreachable
| DistributionVerticalShiftIsInvalid
| SampleSetError(SampleSetDist.sampleSetError)
| ArgumentError(string)
| ArgumentError(ArgumentError.t)
| OperationError(Operation.Error.t)
| PointSetConversionError(SampleSetDist.pointsetConversionError)
| SparklineError(PointSetTypes.sparklineError) // This type of error is for when we find a sparkline of a discrete distribution. This should probably at some point be actually implemented
@ -33,7 +83,7 @@ module Error = {
| NotYetImplemented => "Function Not Yet Implemented"
| Unreachable => "Unreachable"
| DistributionVerticalShiftIsInvalid => "Distribution Vertical Shift is Invalid"
| ArgumentError(s) => `Argument Error ${s}`
| ArgumentError(s) => ArgumentError.toString(s)
| LogarithmOfDistributionError(s) => `Logarithm of input error: ${s}`
| SampleSetError(TooFewSamples) => "Too Few Samples"
| SampleSetError(NonNumericInput(err)) => `Found a non-number in input: ${err}`
@ -51,6 +101,7 @@ module Error = {
let sampleErrorToDistErr = (err: SampleSetDist.sampleSetError): error => SampleSetError(err)
}
@genType
module DistributionOperation = {
@genType

View File

@ -5,10 +5,8 @@ let normal95confidencePoint = 1.6448536269514722
module Normal = {
type t = normal
let make = (mean: float, stdev: float): result<symbolicDist, string> =>
stdev > 0.0
? Ok(#Normal({mean: mean, stdev: stdev}))
: Error("Standard deviation of normal distribution must be larger than 0")
let make = (mean: SafeFloat.finite, stdev: SafeFloat.positive): symbolicDist =>
#Normal({mean: SafeFloat.Finite.toFloat(mean), stdev: SafeFloat.Positive.toFloat(stdev)})
let pdf = (x, t: t) => Jstat.Normal.pdf(x, t.mean, t.stdev)
let cdf = (x, t: t) => Jstat.Normal.cdf(x, t.mean, t.stdev)
@ -68,14 +66,10 @@ module Normal = {
module Exponential = {
type t = exponential
let make = (rate: float): result<symbolicDist, string> =>
rate > 0.0
? Ok(
let make = (rate: SafeFloat.positive): symbolicDist =>
#Exponential({
rate: rate,
}),
)
: Error("Exponential distributions rate must be larger than 0.")
rate: SafeFloat.Positive.toFloat(rate),
})
let pdf = (x, t: t) => Jstat.Exponential.pdf(x, t.rate)
let cdf = (x, t: t) => Jstat.Exponential.cdf(x, t.rate)
let inv = (p, t: t) => Jstat.Exponential.inv(p, t.rate)
@ -86,10 +80,8 @@ module Exponential = {
module Cauchy = {
type t = cauchy
let make = (local, scale): result<symbolicDist, string> =>
scale > 0.0
? Ok(#Cauchy({local: local, scale: scale}))
: Error("Cauchy distribution scale parameter must larger than 0.")
let make = (local: SafeFloat.finite, scale: SafeFloat.positive): symbolicDist =>
#Cauchy({local: SafeFloat.Finite.toFloat(local), scale: SafeFloat.Positive.toFloat(scale)})
let pdf = (x, t: t) => Jstat.Cauchy.pdf(x, t.local, t.scale)
let cdf = (x, t: t) => Jstat.Cauchy.cdf(x, t.local, t.scale)
let inv = (p, t: t) => Jstat.Cauchy.inv(p, t.local, t.scale)
@ -100,10 +92,12 @@ module Cauchy = {
module Triangular = {
type t = triangular
let make = (low, medium, high): result<symbolicDist, string> =>
low < medium && medium < high
? Ok(#Triangular({low: low, medium: medium, high: high}))
: Error("Triangular values must be increasing order.")
let make = (low: SafeFloat.finite, medium: SafeFloat.finite, high: SafeFloat.finite): result<symbolicDist, DistributionTypes.argumentError> =>{
let (l, m, h) = (SafeFloat.Finite.toFloat(low), SafeFloat.Finite.toFloat(medium), SafeFloat.Finite.toFloat(high))
l < m && m < h
? Ok(#Triangular({low: l, medium: m, high: h}))
: Error(OtherArgumentError("Triangular values must be increasing order."))
}
let pdf = (x, t: t) => Jstat.Triangular.pdf(x, t.low, t.high, t.medium) // not obvious in jstat docs that high comes before medium?
let cdf = (x, t: t) => Jstat.Triangular.cdf(x, t.low, t.high, t.medium)
let inv = (p, t: t) => Jstat.Triangular.inv(p, t.low, t.high, t.medium)
@ -114,10 +108,8 @@ module Triangular = {
module Beta = {
type t = beta
let make = (alpha, beta) =>
alpha > 0.0 && beta > 0.0
? Ok(#Beta({alpha: alpha, beta: beta}))
: Error("Beta distribution parameters must be positive")
let make = (alpha: SafeFloat.positive, beta: SafeFloat.positive) =>
#Beta({alpha: SafeFloat.Positive.toFloat(alpha), beta: SafeFloat.Positive.toFloat(beta)})
let pdf = (x, t: t) => Jstat.Beta.pdf(x, t.alpha, t.beta)
let cdf = (x, t: t) => Jstat.Beta.cdf(x, t.alpha, t.beta)
let inv = (p, t: t) => Jstat.Beta.inv(p, t.alpha, t.beta)
@ -128,10 +120,8 @@ module Beta = {
module Lognormal = {
type t = lognormal
let make = (mu, sigma) =>
sigma > 0.0
? Ok(#Lognormal({mu: mu, sigma: sigma}))
: Error("Lognormal standard deviation must be larger than 0")
let make = (mu: SafeFloat.finite, sigma: SafeFloat.positive) =>
#Lognormal({mu: SafeFloat.Finite.toFloat(mu), sigma: SafeFloat.Positive.toFloat(sigma)})
let pdf = (x, t: t) => Jstat.Lognormal.pdf(x, t.mu, t.sigma)
let cdf = (x, t: t) => Jstat.Lognormal.cdf(x, t.mu, t.sigma)
let inv = (p, t: t) => Jstat.Lognormal.inv(p, t.mu, t.sigma)
@ -199,8 +189,16 @@ module Lognormal = {
module Uniform = {
type t = uniform
let make = (low, high) =>
high > low ? Ok(#Uniform({low: low, high: high})) : Error("High must be larger than low")
let make = (low: SafeFloat.finite, high: SafeFloat.finite) => {
let l = SafeFloat.Finite.toFloat(low)
let h = SafeFloat.Finite.toFloat(high)
if h > l {
Ok(#Uniform({low: l, high: h}))
}
else {
Error(DistributionTypes.OtherArgumentError("High must be larger than low"))
}
}
let pdf = (x, t: t) => Jstat.Uniform.pdf(x, t.low, t.high)
let cdf = (x, t: t) => Jstat.Uniform.cdf(x, t.low, t.high)
@ -218,16 +216,8 @@ module Uniform = {
module Gamma = {
type t = gamma
let make = (shape: float, scale: float) => {
if shape > 0. {
if scale > 0. {
Ok(#Gamma({shape: shape, scale: scale}))
} else {
Error("scale must be larger than 0")
}
} else {
Error("shape must be larger than 0")
}
let make = (shape: SafeFloat.positive, scale: SafeFloat.positive) => {
#Gamma({shape: SafeFloat.Positive.toFloat(shape), scale: SafeFloat.Positive.toFloat(scale)})
}
let pdf = (x: float, t: t) => Jstat.Gamma.pdf(x, t.shape, t.scale)
let cdf = (x: float, t: t) => Jstat.Gamma.cdf(x, t.shape, t.scale)
@ -255,12 +245,16 @@ module Float = {
}
module From90thPercentile = {
let make = (low, high) =>
switch (low, high) {
| (low, high) if low <= 0.0 && low < high => Ok(Normal.from90PercentCI(low, high))
| (low, high) if low < high => Ok(Lognormal.from90PercentCI(low, high))
| (_, _) => Error("Low value must be less than high value.")
let make = (low: SafeFloat.positive, high: SafeFloat.positive) : result<SymbolicDistTypes.symbolicDist, DistributionTypes.argumentError> => {
let l = SafeFloat.Positive.toFloat(low)
let h = SafeFloat.Positive.toFloat(high)
if l < h {
Ok(Lognormal.from90PercentCI(l, h))
}
else {
Error(OtherArgumentError("Low value must be less than high value."))
}
}
}
module T = {

View File

@ -0,0 +1,103 @@
type expressionValue = ReducerInterface_ExpressionValue.expressionValue
type error = DistributionTypes.error
type argumentError = DistributionTypes.ArgumentError.t
let expressionValueToType = (value: expressionValue): DistributionTypes.expressionType =>
switch value {
| EvArray(_) => ArrayType
| EvArrayString(_) => ArrayStringType
| EvBool(_) => BoolType
| EvCall(_) => CallType
| EvDistribution(_) => DistributionType
| EvLambda(_) => LambdaType
| EvNumber(_) => NumberType
| EvRecord(_) => RecordType
| EvString(_) => StringType
| EvSymbol(_) => SymbolType
}
module Primitive = {
let distribution = (argument: expressionValue): result<DistributionTypes.genericDist, argumentError> =>
switch argument {
| EvDistribution(dist) => Ok(dist)
| _ => Error(WrongTypeError(DistributionType, expressionValueToType(argument)))
}
let finite = (argument: expressionValue): result<SafeFloat.finite, argumentError> =>
switch argument {
| EvNumber(num) =>
switch SafeFloat.Finite.make(num) {
| Some(safeNum) => Ok(safeNum)
| None => Error(MustBeFinite)
}
| _ => Error(WrongTypeError(NumberType, expressionValueToType(argument)))
}
let positive = (argument: expressionValue): result<SafeFloat.positive, argumentError> =>
switch argument {
| EvNumber(num) =>
switch SafeFloat.Positive.make(num) {
| Some(safeNum) => Ok(safeNum)
| None => Error(MustBePositive)
}
| _ => Error(WrongTypeError(NumberType, expressionValueToType(argument)))
}
}
module Functions = {
let function1 = (
f: 'a => 'b,
parseArg1: expressionValue => result<'a, argumentError>,
args: array<expressionValue>,
): result<'b, argumentError> =>
switch args {
| [arg1] => E.R.fmap(f, parseArg1(arg1))
| _ => Error(IncorrectNumberOfArgumentsError(1, E.A.length(args)))
}
let function2Bind = (
f: ('a, 'b) => result<'c, argumentError>,
parseArg1: expressionValue => result<'a, argumentError>,
parseArg2: expressionValue => result<'b, argumentError>,
args: array<expressionValue>,
): result<'c, argumentError> =>
switch args {
| [arg1, arg2] => E.R.merge(parseArg1(arg1), parseArg2(arg2)) -> E.R.bind(((a, b)) => f(a, b))
| _ => Error(IncorrectNumberOfArgumentsError(2, E.A.length(args)))
}
let function2 = (
f: ('a, 'b) => 'c,
parseArg1: expressionValue => result<'a, argumentError>,
parseArg2: expressionValue => result<'b, argumentError>,
args: array<expressionValue>,
): result<'c, argumentError> =>
function2Bind((a, b) => Ok(f(a, b)), parseArg1, parseArg2, args)
let function3Bind = (
f: ('a, 'b, 'c) => result<'d, argumentError>,
parseArg1: expressionValue => result<'a, argumentError>,
parseArg2: expressionValue => result<'b, argumentError>,
parseArg3: expressionValue => result<'c, argumentError>,
args: array<expressionValue>,
): result<'d, argumentError> =>
switch args {
| [arg1, arg2, arg3] => E.R.merge3(parseArg1(arg1), parseArg2(arg2), parseArg3(arg3)) -> E.R.bind(((a, b, c)) => f(a, b, c))
| _ => Error(IncorrectNumberOfArgumentsError(2, E.A.length(args)))
}
}
type function = Function(string, array<expressionValue> => result<SymbolicDistTypes.symbolicDist, argumentError>)
let allFunctions: array<function> =
[ Function("exponential", Functions.function1( SymbolicDist.Exponential.make, Primitive.positive))
, Function("normal", Functions.function2( SymbolicDist.Normal.make, Primitive.finite, Primitive.positive,))
, Function("uniform", Functions.function2Bind( SymbolicDist.Uniform.make, Primitive.finite, Primitive.finite,))
, Function("beta", Functions.function2( SymbolicDist.Beta.make, Primitive.positive, Primitive.positive,))
, Function("lognormal", Functions.function2( SymbolicDist.Lognormal.make, Primitive.finite, Primitive.positive,))
, Function("cauchy", Functions.function2( SymbolicDist.Cauchy.make, Primitive.finite, Primitive.positive,))
, Function("gamma", Functions.function2( SymbolicDist.Gamma.make, Primitive.positive, Primitive.positive,))
, Function("to", Functions.function2Bind( SymbolicDist.From90thPercentile.make, Primitive.positive, Primitive.positive,))
, Function("triangular", Functions.function3Bind( SymbolicDist.Triangular.make, Primitive.finite, Primitive.finite, Primitive.finite,))
]

View File

@ -1,5 +1,6 @@
module ExpressionValue = ReducerInterface_ExpressionValue
type expressionValue = ReducerInterface_ExpressionValue.expressionValue
type argumentError = DistributionTypes.argumentError
module Helpers = {
let arithmeticMap = r =>
@ -80,25 +81,25 @@ module Helpers = {
)->DistributionOperation.run(~env)
}
let parseNumber = (args: expressionValue): Belt.Result.t<float, string> =>
let parseNumber = (args: expressionValue): Belt.Result.t<float, argumentError> =>
switch args {
| EvNumber(x) => Ok(x)
| _ => Error("Not a number")
| _ => Error(OtherArgumentError("Not a number"))
}
let parseNumberArray = (ags: array<expressionValue>): Belt.Result.t<array<float>, string> =>
let parseNumberArray = (ags: array<expressionValue>): Belt.Result.t<array<float>, argumentError> =>
E.A.fmap(parseNumber, ags) |> E.A.R.firstErrorOrOpen
let parseDist = (args: expressionValue): Belt.Result.t<DistributionTypes.genericDist, string> =>
let parseDist = (args: expressionValue): Belt.Result.t<DistributionTypes.genericDist, argumentError> =>
switch args {
| EvDistribution(x) => Ok(x)
| EvNumber(x) => Ok(GenericDist.fromFloat(x))
| _ => Error("Not a distribution")
| _ => Error(OtherArgumentError("Not a distribution"))
}
let parseDistributionArray = (ags: array<expressionValue>): Belt.Result.t<
array<DistributionTypes.genericDist>,
string,
argumentError,
> => E.A.fmap(parseDist, ags) |> E.A.R.firstErrorOrOpen
let mixtureWithGivenWeights = (
@ -109,7 +110,7 @@ module Helpers = {
E.A.length(distributions) == E.A.length(weights)
? Mixture(Belt.Array.zip(distributions, weights))->DistributionOperation.run(~env)
: GenDistError(
ArgumentError("Error, mixture call has different number of distributions and weights"),
ArgumentError(OtherArgumentError("Error, mixture call has different number of distributions and weights"))
)
let mixtureWithDefaultWeights = (
@ -125,7 +126,7 @@ module Helpers = {
args: array<expressionValue>,
~env: DistributionOperation.env,
): DistributionOperation.outputType => {
let error = (err: string): DistributionOperation.outputType =>
let error = (err: DistributionTypes.argumentError): DistributionOperation.outputType =>
err->DistributionTypes.ArgumentError->GenDistError
switch args {
| [EvArray(distributions)] =>
@ -138,7 +139,7 @@ module Helpers = {
| (Ok(distrs), Ok(wghts)) => mixtureWithGivenWeights(distrs, wghts, ~env)
| (Error(err), Ok(_)) => error(err)
| (Ok(_), Error(err)) => error(err)
| (Error(err1), Error(err2)) => error(`${err1}|${err2}`)
| (Error(err1), Error(err2)) => error(err1)
}
| _ =>
switch E.A.last(args) {
@ -158,36 +159,19 @@ module Helpers = {
| Ok(distributions) => mixtureWithDefaultWeights(distributions, ~env)
| Error(err) => error(err)
}
| _ => error("Last argument of mx must be array or distribution")
| _ => error(OtherArgumentError("Last argument of mx must be array or distribution"))
}
}
}
}
module SymbolicConstructors = {
let oneFloat = name =>
switch name {
| "exponential" => Ok(SymbolicDist.Exponential.make)
| _ => Error("Unreachable state")
}
let twoFloat = name =>
switch name {
| "normal" => Ok(SymbolicDist.Normal.make)
| "uniform" => Ok(SymbolicDist.Uniform.make)
| "beta" => Ok(SymbolicDist.Beta.make)
| "lognormal" => Ok(SymbolicDist.Lognormal.make)
| "cauchy" => Ok(SymbolicDist.Cauchy.make)
| "gamma" => Ok(SymbolicDist.Gamma.make)
| "to" => Ok(SymbolicDist.From90thPercentile.make)
| _ => Error("Unreachable state")
}
let threeFloat = name =>
switch name {
| "triangular" => Ok(SymbolicDist.Triangular.make)
| _ => Error("Unreachable state")
}
let checkSymbolicConstructors = (call: ExpressionValue.functionCall) : option<result<SymbolicDistTypes.symbolicDist, argumentError>> => {
let (fnName, args) = call
let function = E.A.find((ReducerInterface_FunctionParser.Function(name, argsParser)) => name == fnName, ReducerInterface_FunctionParser.allFunctions)
E.O.fmap((ReducerInterface_FunctionParser.Function(_, argsParser)) => argsParser(args), function)
}
let symbolicResultToOutput = (
symbolicResult: result<SymbolicDistTypes.symbolicDist, string>,
@ -204,23 +188,8 @@ let dispatchToGenericOutput = (
): option<DistributionOperation.outputType> => {
let (fnName, args) = call
switch (fnName, args) {
| ("exponential" as fnName, [EvNumber(f)]) =>
SymbolicConstructors.oneFloat(fnName)
->E.R.bind(r => r(f))
->SymbolicConstructors.symbolicResultToOutput
| ("delta", [EvNumber(f)]) =>
SymbolicDist.Float.makeSafe(f)->SymbolicConstructors.symbolicResultToOutput
| (
("normal" | "uniform" | "beta" | "lognormal" | "cauchy" | "gamma" | "to") as fnName,
[EvNumber(f1), EvNumber(f2)],
) =>
SymbolicConstructors.twoFloat(fnName)
->E.R.bind(r => r(f1, f2))
->SymbolicConstructors.symbolicResultToOutput
| ("triangular" as fnName, [EvNumber(f1), EvNumber(f2), EvNumber(f3)]) =>
SymbolicConstructors.threeFloat(fnName)
->E.R.bind(r => r(f1, f2, f3))
->SymbolicConstructors.symbolicResultToOutput
| ("sample", [EvDistribution(dist)]) => Helpers.toFloatFn(#Sample, dist, ~env)
| ("mean", [EvDistribution(dist)]) => Helpers.toFloatFn(#Mean, dist, ~env)
| ("integralSum", [EvDistribution(dist)]) => Helpers.toFloatFn(#IntegralSum, dist, ~env)
@ -323,7 +292,7 @@ let dispatchToGenericOutput = (
a,
~env,
)->Some
| _ => None
| _ => checkSymbolicConstructors(call)
}
}

View File

@ -253,6 +253,13 @@ module R = {
| (_, Error(e)) => Error(e)
| (Ok(a), Ok(b)) => Ok((a, b))
}
let merge3 = (a, b, c) =>
switch (a, b, c) {
| (Error(e), _, _) => Error(e)
| (_, Error(e), _) => Error(e)
| (_, _, Error(e)) => Error(e)
| (Ok(a), Ok(b), Ok(c)) => Ok((a, b, c))
}
let toOption = (e: Belt.Result.t<'a, 'b>) =>
switch e {
| Ok(r) => Some(r)
@ -531,10 +538,13 @@ module A = {
let keepMap = Belt.Array.keepMap
let slice = Belt.Array.slice
let init = Array.init
let filter = (fn, xs) => Belt.Array.keep(xs, fn)
let reduce = Belt.Array.reduce
let reducei = Belt.Array.reduceWithIndex
let isEmpty = r => length(r) < 1
let stableSortBy = Belt.SortArray.stableSortBy
let find = (f: 'a => bool, xs: array<'a>) => first(filter(f, xs))
let toRanges = (a: array<'a>) =>
switch a |> Belt.Array.length {
| 0

View File

@ -0,0 +1,34 @@
type finite = Finite(float)
module Finite = {
type t = finite
let valid = Js.Float.isFinite
let make = (x: float) : option<t> =>
if valid(x) {
Some(Finite(x))
}
else {
None
}
let toFloat = (x: t) =>
switch x {
| Finite(inner) => inner
}
}
type positive = Positive(float)
module Positive = {
type t = positive
let valid = (x: float) => Finite.valid(x) && x > 0.
let make = (x: float) : option<t> =>
if valid(x) {
Some(Positive(x))
}
else {
None
}
let toFloat = (x: t) =>
switch x {
| Positive(inner) => inner
}
}