middle of converting to PushData

HushGP.cabal

@@ -64,6 +64,7 @@ library
                     , HushGP.GP.PushArgs
                     , HushGP.GP.Variation
                     , HushGP.GP.Downsample
+                    , HushGP.GP.PushData
                     , HushGP.Problems.IntegerRegression
 
 

src/HushGP/GP.hs

@@ -9,6 +9,7 @@ import HushGP.Genome
 import HushGP.State
 import HushGP.GP.Variation
 import HushGP.GP.Downsample
+import HushGP.GP.PushData
 import HushGP.Utility
 
 -- import Debug.Trace (trace, traceStack)
@@ -22,7 +23,7 @@ generatePopulation pushArgs = do
 
 -- | Evaluates a population of plushies with the error function passed in via PushArgs and sorts them.
 -- TODO: Need to make this runnable in parallel too.
-evaluatePopulation :: PushArgs -> ([[Gene]], [Gene], [Int]) -> [Individual] -> [Individual]
+evaluatePopulation :: PushArgs -> [PushData] -> [Individual] -> [Individual]
 evaluatePopulation pushArgs passedTrainingData population = sort $ zipWith updateIndividual (map (errorFunction pushArgs pushArgs passedTrainingData . plushy) population) population
 
 -- | A helper function used in evaluatePopulation. Takes a [Double] as the error scores and an individual.

src/HushGP/GP/PushArgs.hs

@@ -1,8 +1,9 @@
 module HushGP.GP.PushArgs where
 
 import HushGP.State
-import Data.Map qualified as Map
 import HushGP.Instructions
+import HushGP.GP.PushData
+import Data.Map qualified as Map
 
 -- | The structure holding the arguments for the various aspects
 -- of the evolutionary run in Hush.
@@ -46,10 +47,10 @@ data PushArgs = PushArgs
     elitism :: Bool,
     -- | User must provide their own error function.
     -- Arg 1: PushArgs for the current set of arguments.
-    -- Arg 2: ([[Gene]], [Gene]) is the input data. Input is the first index and output is the second index.
+    -- Arg 2: [PushData] is the input data.
     -- Arg 3: [Gene] is the plushy representation of a program.
     -- Returns the error list for a given set of inputs of type [Double].
-    errorFunction :: PushArgs -> ([[Gene]], [Gene], [Int]) -> [Gene] -> [Double],
+    errorFunction :: PushArgs -> [PushData] -> [Gene] -> [Double],
     -- | Type of informed downsampling. "solved", "elite", "soft".
     informedDownsamplingType :: String,
     -- | List of instructions to use in the evolutionary run.
@@ -83,9 +84,9 @@ data PushArgs = PushArgs
     -- | For tournament selection, amount of individuals in each tournament.
     tournamentSize :: Int,
     -- | Training data for the gp, must be provided.
-    trainingData :: ([[Gene]], [Gene]),
+    trainingData :: [PushData],
     -- | Testing data for the gp, must be provided if there is any.
-    testingData :: ([[Gene]], [Gene]),
+    testingData :: [PushData],
     -- | Addition rate for UMAD (deletion rate derived from this).
     umadRate :: Float,
     -- | Genetic operators and probabilities for their use, should sum to one
@@ -134,8 +135,8 @@ defaultPushArgs = PushArgs {
     ssxNotBmx = False,
     stepLimit = 1000,
     tournamentSize = 5,
-    testingData = ([], []),
-    trainingData = ([], []),
+    testingData = error "Must supply the testingData yourself",
+    trainingData = error "Must supply the trainingData yourself",
     umadRate = 0.1,
     variation = Map.fromList [("umad", 1.0)],
     epsilons = Nothing

src/HushGP/GP/PushData.hs

@@ -0,0 +1,14 @@
+module HushGP.GP.PushData where
+
+import HushGP.State
+
+data PushData = PushData {
+  inputData :: [Gene],
+  outputData :: Gene,
+  downsampleIndex :: Maybe Int,
+  caseDistances :: Maybe [Double]
+}
+
+-- |Utility function: Sets the index of the passed training data.
+makeIndexedTrainingData :: [PushData] -> [PushData]
+makeIndexedTrainingData oldData = zipWith (\dat idx -> dat{downsampleIndex = Just idx}) oldData [0..]

src/HushGP/Problems/IntegerRegression.hs

@@ -6,6 +6,7 @@ import Data.Map qualified as Map
 import HushGP.State
 import HushGP.Instructions
 import HushGP.GP.PushArgs
+import HushGP.GP.PushData
 import HushGP.Genome
 import HushGP.Push
 import HushGP.Instructions.Utility
@@ -25,12 +26,23 @@ targetFunction :: Integer -> Integer
 targetFunction x = (x * x * x) + (2 * x)
 
 -- | The training data for the model.
-trainData :: ([[Gene]], [Gene])
-trainData = (chunksOf 1 $ map GeneInt [-10..10], map (GeneInt . targetFunction) [-10..11])
+trainData :: [PushData]
+trainData = map (\num -> PushData {
+      inputData = [GeneInt num],
+      outputData = (GeneInt . targetFunction) num,
+      downsampleIndex = Nothing,
+      caseDistances = Nothing})
+    [-10..10]
 
 -- | The testing data for the model.
-testData :: ([[Gene]], [Gene])
-testData = (chunksOf 1 $ map GeneInt $ [-20..(-11)] <> [11..21], map (GeneInt . targetFunction) ([-20..(-11)] <> [11..21]))
+testData :: [PushData]
+-- testData = (chunksOf 1 $ map GeneInt $ [-20..(-11)] <> [11..21], map (GeneInt . targetFunction) ([-20..(-11)] <> [11..21]))
+testData = map (\num -> PushData {
+      inputData = [GeneInt num],
+      outputData = (GeneInt . targetFunction) num,
+      downsampleIndex = Nothing,
+      caseDistances = Nothing})
+    [-20..(-11)] <> [11..21]
 
 -- | The instructions used in the evolutionary run.
 runInstructions :: [Gene]
@@ -57,9 +69,9 @@ loadState plushy vals =
   (loadProgram (plushyToPush plushy) emptyState){_input = Map.fromList (zip [0..] vals)}
 
 -- | The error function for a single set of inputs and outputs.
-intErrorFunction :: PushArgs -> ([[Gene]], [Gene], [Int]) -> [Gene] -> [Double]
-intErrorFunction _args (inputData, outputData, _) plushy =
-  map abs $ zipWith (-) (map ((fromIntegral @Integer @Double . (errorHead . _int) . interpretExec) . loadState plushy)  inputData) (map (fromIntegral @Integer @Double . extractGeneInt) outputData)
+intErrorFunction :: PushArgs -> [PushData] -> [Gene] -> [Double]
+intErrorFunction _args pushData plushy =
+  map abs $ zipWith (-) (map ((fromIntegral @Integer @Double . (errorHead . _int) . interpretExec) . loadState plushy) pushData) (map (fromIntegral @Integer @Double . extractGeneInt) outputData)
 
 intPushArgs :: PushArgs
 intPushArgs = defaultPushArgs

src/HushGP/Utility.hs

@@ -30,7 +30,3 @@ thrd (_, _, x) = x
 -- |Utility function: Converts a tuple to a triple with a passed value.
 tupleToTriple :: (a, b) -> c -> (a, b, c)
 tupleToTriple (x, y) z = (x, y, z)
-
--- |Utility function: Converts the training data passed in to an indexed representation
-makeIndexedTrainingData :: ([[Gene]], [Gene]) -> ([[Gene]], [Gene], [Int])
-makeIndexedTrainingData (inputs, outputs) = (inputs, outputs, [0..(length inputs)])