/go-genetic-math

102.41.01 A small experiment with genetic algorithms in Go.

Primary LanguageGo

Go - Genetic Math

A small experiment with genetic algorithms in Go.

Example usage: go run main.go genome.go --target 200

Example

The genomes

Each genome is a series of genomeSize genes. Each gene is a 4-bit opcode.

  • 0 - 9 is a single digit.
  • 10 (+) is addition.
  • 11 (-) is subtraction.
  • 12 (*) is multiplication.
  • 13 (/) is division.
  • 14 - 15 (X) is an invalid opcode and is ignored.

An example genome is 3+2*4/2.

Operations are resolved left to right. Duplicate numbers, operations, and invalid opcodes are ignored.

For example:

  • 3+2*4/2 = ((3+2) * 4) / 2 = 10
  • 32+*1X9 = 3 (ignored) + (ignored) 1 (ignored) (ignored) = 4

Flags

  • target float
    • The value the genomes are attempting to solve for. (default 200)
  • populationSize int
    • The number of genomes in a population. (default 100)
  • mutationRate float
    • The chance of a gene mutating when breeding. (default 0.01)
  • immortalChampion
    • If set, the fittest genome in a population will automatically survive to the next. (default true)
  • genomeSize int
    • The number of genes (operations) in a genome. (default 7)
  • crossover
    • If set, breeding a new genome will involve two parents with crossover. (default true)
  • colorOutput
    • If set, genomes will be printed colourized, with invalid genes printed in red. (default true)
  • diffOutput
    • If set, champions will only be printed if they're different to the previous generation's champion. (default true)