Hi all,
I’m pleased to announce the second public release of charlie, a
genetic algorithms library for Ruby.
This release includes a TreeGenotype, making the library useful for
genetic programming. This is still an early version with very limited
mutation options, and I would really appreciate suggestions on how to
extend these.
It also includes several performance improvements, changes for ruby
1.9, convergence checks and improved mutator/crossover operators for
permutations.
Ruby 1.9 Compatibility
This release is mostly compatible with Ruby 1.9.
However, there is a bug in Ruby (#16493) that makes a lot of tests,
and some benchmark examples fail.
To avoid this bug, don’t have two or more results of a function that
returns a module in the same array (e.g. don’t have something like
“crossover PCross(0.5,SinglePointCrossover),
PCross(0.5,UniformCrossover)” in your benchmarks).
FEATURES:
- Quickly develop GAs by combining several parts (genotype, selection,
crossover, mutation) provided by the library. - Sensible defaults are provided with any genotype, so often you only
need to define a fitness function. - Easily replace any of the parts by your own code.
- Test different strategies in GA, and generate reports comparing
them. Example report:
http://charlie.rubyforge.org/example_report.html
EXAMPLE
This example finds a polynomial which approximates cos(x)
class Cos < TreeGenotype([proc{3rand-1.5},:x], [:+,:,:-], [:-@])
def fitness
-[0,0.33,0.66,1].map{|x| (eval_genes(:x=>x) - Math.cos(x)).abs }.max
end
use TournamentSelection(4)
end
Population.new(Cos).evolve_on_console(500)
Several other simple examples are included in the gem/tarball.
INSTALLATION:
- sudo gem install charlie
Links
LICENSE:
MIT license.