PS
Fields psSize :: Int How big the population should be psInputs :: Int Number of inputs psOutputs :: Int Number of outputs psParams :: Parameters Parameters for large species sparse :: Maybe Int If Just n, will be sparse with n connections. Otherwise fully connected. psStrategy :: Maybe PhaseParams

Instances

Show PopSettings

newPop :: Int -> PopSettings -> Population Source

Generates a starter population

Training

trainOnce :: Applicative f => TrainMethod f -> Population -> f Population Source

newtype TrainMethod f Source

Describes how to handle each species or genome when processing fitness, allowing the addition of additional effects

Constructors

TrainMethod
Fields tmGen :: forall t. Traversable t => t Genome -> f (t Double) How to process each `Genome` of a species into a score.

TrainMethods

pureTrain :: GenScorer a -> TrainMethod Identity Source

Train method without any additional effects

winTrain :: GenScorer a -> TrainMethod ((,) (First Genome)) Source

Train method that possibly returns a solution

Convenience

trainN :: (Applicative f, Monad f) => TrainMethod f -> Int -> Population -> f Population Source

Train the population n times. Values less than 1 return the original.

trainUntil :: Int -> GenScorer a -> Population -> (Population, Maybe (Genome, Int)) Source

Train until the provided goal is reached, or the max number of generations (first parameter) is reached. Possibly also returns a solution and the number of generations elapsed.

trainPure :: GenScorer a -> Population -> Population Source

Train without effects

Statistics

speciesCount :: Population -> Int Source

Gets the number of species

Debugging

validatePopulation :: Population -> Maybe [String] Source

Validate a population, possibly returning a list of errors