TODO.md

• use fs2
• describe a population to population evolver
  • mutation + crossover is slightly prescriptive
• callbacks for islands/aggregation
• case class vs Ints/better primitives
• array vs vector as default Seq implementation
• add healing period for "invalid" (?) solutions
  • default value is none/identity
  • demonstrate with healing salesman (chromosome is sequence with duplicates, heals to unique sequence)
    • determinisitically with canonical sequence (slower?)
    • or recklessly with canonical set (faster?)
• describe base hierarchy
• describe combinatorics hierarchy
• doc primary methods and overrides by traits
• consider external operators so that value classes need not extend a trait
  • gut chromosome methods
• explore spontaneous generation and how it links to fixed/variable length homogenous chromosomes
• need drosophilia/helloworld
  • sodoku??
• support arbitrary types?
• support arbitary fitness?
• fitness min/maxer (implies ordered?)
• sample generator
• mutation/combination operators (provided as list?)
  • implemented as seq yields seq?
• selection operators
  • proportional to fitness?
  • select random parents?
  • elitism?
• supported chromosome types
  • fixed length homogenous (Seq[T])
    • cross over, spot mutation (requires () => T, or Chromosome => T)
  • variable length homogenous (Stack[T])
• termination conditions
  • time elapsed
  • generations elapsed
  • state fitness improvmeent
  • target fitness reached
  • and/or

Redesign

population evolver. does not specify mutation or crossover. input Seq[A] output Seq[A]. if A itself is also a Seq, can support by default mutation and popular crossover operations

population manager? used to control islands?

population evolver itself has termination conditions

better, earlier support for Array[Int/Double] as chromosomes

framework for working with Int over discrete collections/as chromosomes