HushGP.cabal

@@ -63,7 +63,6 @@ library
                     , HushGP.GP
                     , HushGP.GP.PushArgs
                     , HushGP.GP.Variation
-                    , HushGP.GP.Downsample
                     , HushGP.Problems.IntegerRegression
 
 

src/HushGP/GP.hs

@@ -7,8 +7,6 @@ import Data.List (sort, uncons)
 import HushGP.GP.PushArgs
 import HushGP.Genome
 import HushGP.State
-import HushGP.GP.Variation
-import HushGP.GP.Downsample
 
 -- import Debug.Trace (trace, traceStack)
 
@@ -60,8 +58,7 @@ gpLoop' pushArgs generation evaluations population indexedTrainingData = do
           print "Incomplete Run, saving the best so far."
         | otherwise = gpLoop' pushArgs (succ generation)
             (evaluations + (populationSize pushArgs * length (fst $ trainingData pushArgs)) + (if generation `mod` downsampleParentsGens pushArgs == 0 then length parentReps * (length (fst indexedTrainingData) - length (fst $ trainingData pushArgs)) else 0) + (if bestIndPassesDownsample then length (fst indexedTrainingData) - length (fst $ trainingData pushArgs) else 0))
-            (if elitism pushArgs then bestInd : replicate (populationSize epsilonPushArgs - 1) (newIndividual epsilonPushArgs evaledPop) else replicate (populationSize epsilonPushArgs) (newIndividual epsilonPushArgs evaledPop))
-            (if enableDownsampling pushArgs && ((generation `mod` downsampleParentsGens pushArgs) == 0) then updateCaseDistances repEvaluatedPop indexedTrainingData indexedTrainingData (informedDownsamplingType pushArgs) (solutionErrorThreshold pushArgs / fromIntegral @Int @Double (length $ fst indexedTrainingData)) else indexedTrainingData)
+            (if elitism then bestInd : )
   nextAction
   where
     -- \| This will have downsampling added to it later.

src/HushGP/GP/Downsample.hs

@@ -1,7 +0,0 @@
-module HushGP.GP.Downsample where
-
-import HushGP.State
-import HushGP.Genome
-
-updateCaseDistances :: [Individual] -> ([[Gene]], [Gene]) -> ([[Gene]], [Gene]) -> String -> Double -> ([[Gene]], [Gene])
-updateCaseDistances evaledPop downsampleData trainingData informedDownsamplingType solutionThreshold = undefined