int regression start

HushGP.cabal

@@ -36,7 +36,9 @@ library
 
     -- Modules exported by the library.
     exposed-modules:  HushGP.Push
-                    , HushGP.GP
+                    , HushGP.TH
+                    , HushGP.Utility
+                    , HushGP.Genome
                     , HushGP.State
                     , HushGP.Instructions
                     , HushGP.Instructions.IntInstructions
@@ -58,10 +60,10 @@ library
                     , HushGP.PushTests.IntTests
                     , HushGP.PushTests.GenericTests
                     , HushGP.PushTests.UtilTests
+                    , HushGP.GP
                     , HushGP.GP.PushArgs
-                    , HushGP.TH
+                    , HushGP.Problems.IntegerRegression
-                    , HushGP.Utility
+
-                    , HushGP.Genome
 
     -- Modules included in this library but not exported.
     -- other-modules:

src/HushGP/GP/PushArgs.hs

@@ -4,6 +4,8 @@ import HushGP.State
 import Data.Map qualified as Map
 import HushGP.Instructions
 
+-- | The structure holding the arguments for the various aspects
+-- of the evolutionary run in Hush.
 data PushArgs = PushArgs
   {
     -- | For alternation, std deviation fo index when alternating.
@@ -42,8 +44,12 @@ data PushArgs = PushArgs
     downsampleParentsGens :: Int,
     -- | Whether or not to add the best individual to the next generation.
     elitism :: Bool,
-    -- User must provide their own error function. TODO: This
+    -- | User must provide their own error function.
-    -- errorFunction :: (PushArgs -> )
+    -- 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 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] -> [Double],
     -- | Type of informed downsampling. "solved", "elite", "soft".
     informedDownsamplingType :: String,
     -- | List of instructions to use in the evolutionary run.
@@ -76,10 +82,10 @@ data PushArgs = PushArgs
     stepLimit :: Int,
     -- | For tournament selection, amount of individuals in each tournament.
     tournamentSize :: Int,
-    -- Training data for the gp, must be provided.
+    -- | Training data for the gp, must be provided.
-    -- trainingData :: something
+    trainingData :: [[Gene]],
-    -- Testing data for the gp, must be provided if there is any.
+    -- | Testing data for the gp, must be provided if there is any.
-    -- testingData :: something
+    testingData :: [[Gene]],
     -- | Addition rate for UMAD (deletion rate derived from this).
     umadRate :: Float,
     -- | Genetic operators and probabilities for their use, should sum to one
@@ -87,6 +93,8 @@ data PushArgs = PushArgs
     variation :: Map.Map String Float
   }
 
+-- | The default values for which all runs of Hush derive
+-- their args from.
 defaultPushArgs :: PushArgs
 defaultPushArgs = PushArgs {
     alignmentDeviation = 2,
@@ -106,7 +114,7 @@ defaultPushArgs = PushArgs {
     downsampleParentRate = 0.01,
     downsampleParentsGens = 10,
     elitism = False,
-    -- errorFunction = something,
+    errorFunction = error "Must supply the error function yourself",
     informedDownsamplingType = "solved",
     instructionList = allInstructions,
     maxMotelyBatchSize = 10,
@@ -124,8 +132,8 @@ defaultPushArgs = PushArgs {
     ssxNotBmx = False,
     stepLimit = 1000,
     tournamentSize = 5,
-    -- testingData = [],
+    testingData = [],
-    -- trainingData = [],
+    trainingData = [],
     umadRate = 0.1,
     variation = Map.fromList [("umad", 1.0)]
   }

src/HushGP/Problems/IntegerRegression.hs

@@ -0,0 +1,41 @@
+module HushGP.Problems.IntegerRegression where
+
+import HushGP.State
+import HushGP.Instructions
+import Data.List.Split
+import HushGP.GP.PushArgs
+import HushGP.Genome
+import HushGP.Push
+import Data.Map qualified as Map
+
+-- | The target function for this run. The function the gp
+-- is trying to evolve.
+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])
+
+-- | 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]))
+
+-- | The instructions used to 
+runInstructions :: [Gene]
+runInstructions =
+  [
+    PlaceInput 0,
+    Close,
+    GeneInt 1,
+    GeneInt 0
+  ]
+  <> allIntInstructions
+
+-- |The error function for a single set of inputs and outputs.
+intErrorFunction :: PushArgs -> ([Gene], Gene) -> [Gene] -> [Double]
+intErrorFunction args (inputData, outputData) plushy =
+  head $ _int $ interpretExec loadedState
+  where
+    loadedState :: State
+    loadedState = (loadProgram (plushyToPush plushy) emptyState){_input = Map.fromList (zip [0..] inputData)}

src/HushGP/State.hs

@@ -21,9 +21,9 @@ data Gene
   | GeneVectorBool [Bool]
   | GeneVectorString [String]
   | GeneVectorChar [Char]
-  | -- | State -> State is the function itself. String stores the name of the function.
+    -- | State -> State is the function itself. String stores the name of the function.
-    StateFunc (State -> State, String)
+  | StateFunc (State -> State, String)
-  | PlaceInput String
+  | PlaceInput Int
   | Close
   | Open Int
   | Skip
@@ -160,7 +160,7 @@ data State = State
     _vectorString :: [[String]],
     _vectorChar :: [[Char]],
     _parameter :: [Gene],
-    _input :: Map.Map String Gene
+    _input :: Map.Map Int Gene
   }
   deriving (Show, Eq, Ord)