charlie 0.6.0 → 0.7.0

data/test/test_cross.rb

@@ -11,7 +11,9 @@ class TestCrossover < Test::Unit::TestCase
   def test_singlechild
     $crs_mth.map{|c| SingleChild(c) }.each{|s|
       klass = TestClass(s)
-      assert_nothing_raised{ Population.new(klass,10).evolve_silent(10) }
+      r = nil
+      assert_nothing_raised{ r=Population.new(klass,10).evolve_silent(10) }
+      r.each{|e| assert_equal klass.new.size, e.size } # crossovers should preserver size
     }
   end
 
@@ -25,4 +27,24 @@ class TestCrossover < Test::Unit::TestCase
       assert_nothing_raised{ Population.new(klass,10).evolve_silent(10) }
     }
   end
+
+
+  def test_pcrossn
+    3.times{
+    $crs_mth.map{|c| PCrossN(c=>0.5,$crs_mth.at_rand=>0.3) }.each{|s|
+      klass = TestClass(s)
+      assert_nothing_raised{ Population.new(klass,10).evolve_silent(10) }
+    }
+    }
+    $crs_mth.map{|c| PCrossN(c=>0.9) }.each{|s|
+      klass = TestClass(s)
+      assert_nothing_raised{ Population.new(klass,10).evolve_silent(10) }
+    }
+    assert_raises(ArgumentError) {
+       PCrossN(UniformCrossover=>0.5,NullCrossover=>0.6) # => 1.1 > 1.0
+    }
+    s = PCrossN( Hash[*$crs_mth.map{|c| [c,1 / $crs_mth.size ] }.flatten] )
+    klass = TestClass(s)
+    assert_nothing_raised{ Population.new(klass,10).evolve_silent(10) }
+  end
 end

data/test/test_evolve.rb

@@ -46,7 +46,7 @@ class EvolveTest < Test::Unit::TestCase
 
   def test_until_pop  # test evolve_until_population
     start = Time.now
-    r,g = Population.new(RRTest,10).evolve_until_population(1000_000_000){|p| p.count{|x| x.fitness == 0 } > 3}
+    r,g = Population.new(RRTest,10).evolve_until_population(100_000_000){|p| p.count{|x| x.fitness == 0 } > 3}
     assert_not_nil g
     assert g >= 0 && g < 1000, "test_until_pop failed to converge within 1000 generations"
     assert( (Time.now - start < 1), "test_until_pop failed to converge within 1 second.")
@@ -58,6 +58,12 @@ class EvolveTest < Test::Unit::TestCase
     assert_nil g # did not converge
     assert_respond_to r, :[]
     assert_respond_to r[-1], :fitness
+    # test check_every
+    r,g = Population.new(RRTest,10).evolve_until(100,200){|b| b.fitness == 0 }
+    assert(g.nil? || g.zero?) # did not converge (except in the gen 0 check), because check_every > max_gens
+    r,g = Population.new(RRTest,10).evolve_until(1000,10){|b| b.fitness == 0 }
+    assert_not_nil g
+    assert( (g % 10).zero?)
   end
 
   # doesn't really belong here, and should be moved...eventually
@@ -84,4 +90,11 @@ class EvolveTest < Test::Unit::TestCase
     }
   end
 
+
+  def test_acc  # test accessors
+    p = Population.new(RRTest,10)
+    assert_equal p.max, p.population.max
+    assert_equal p.max, p.best
+    assert_equal p[6], p.population[6]
+  end
 end

data/test/test_mutator.rb

@@ -1,11 +1,18 @@
 require 't_common'
 
-ms = [:single_point,:single_point[],[:single_point],:n_point,:n_point[],:n_point[3],:expected_n,:expected_n[1],:probability,:probability[0.3]]
-pm = [:gaussian,:gaussian[],:gaussian[0.25],:uniform,:uniform[0.1],:flip,:replace[0.2,0.1,0.0,-0.1,-0.2]]
+module Enumerable
+ def uniq_by
+   h = {}; inject([]) {|a,x| h[yield(x)] ||= a << x}
+ end
+end
+
+ms = [:single_point,[:n_point],:n_point[3],:expected_n[],:expected_n[1],:probability,:probability[0.3]]
+pm = [:gaussian,:gaussian[0.25],:uniform,:uniform[0.1],:flip,:replace[0.2,0.1,0.0,-0.1,-0.2]]
 $mut_mth = [NullMutator]
 ms.each{|m| pm.each{|p| $mut_mth << ListMutator(m,p) } }
 
 
+
 class TestMutator < Test::Unit::TestCase
   def test_mutate
     $mut_mth.each{|m|
@@ -41,4 +48,23 @@ class TestMutator < Test::Unit::TestCase
     assert_nothing_raised{ p=Population.new(klass,10).evolve_silent(10) }
     assert p.all?{|s| s.genes.all?{|x| ('a'..'d')===x }}
   end
+
+  def test_pmutaten
+    3.times{
+    $mut_mth.map{|c| PMutateN(c=>0.5,$mut_mth.at_rand=>0.3) }.each{|s|
+      klass = TestClass(s)
+      assert_nothing_raised{ Population.new(klass,10).evolve_silent(10) }
+    }
+    }
+    $mut_mth.map{|c| PMutateN(c=>0.9) }.each{|s|
+      klass = TestClass(s)
+      assert_nothing_raised{ Population.new(klass,10).evolve_silent(10) }
+    }
+    assert_raises(ArgumentError) {
+       PMutateN(ListMutator(:single_point,:uniform)=>0.5,NullMutator=>0.6) # => 1.1 > 1.0
+    }
+    s = PMutateN( Hash[*$mut_mth.map{|c| [c,1 / $mut_mth.size ] }.flatten] )
+    klass = TestClass(s)
+    assert_nothing_raised{ Population.new(klass,10).evolve_silent(10) }
+  end
 end

data/test/test_permutation.rb

@@ -10,13 +10,21 @@ end
 
 class PermutationTestERO < PermutationTest
   use EdgeRecombinationCrossover
-  use PMutate(0.5,PermutationMutator)
+  use PMutate(0.5,TranspositionMutator)
 end
 
 class PermutationTestInvert < PermutationTest
   use InversionMutator, RandomSelection
 end
 
+class PermutationTestRotate < PermutationTest
+  use RotationMutator, NullCrossover
+end
+class PermutationTestInsert < PermutationTest
+  use InsertionMutator, NullCrossover
+end
+
+
 
 class PermTests < Test::Unit::TestCase
   def test_evolve
@@ -53,5 +61,19 @@ class PermTests < Test::Unit::TestCase
     p.each{|s| assert_equal s.genes.sort, (0...N).to_a }
   end
 
+  def test_rotation_mutator # test if rotation mutator works
+    p=nil
+    assert_nothing_raised{
+      p=Population.new(PermutationTestRotate,20).evolve_silent(20)
+    }
+    p.each{|s|  assert_equal s.genes.sort, (0...N).to_a }
+  end
 
+  def test_insertion_mutator # test if insertion mutator works
+    p=nil
+    assert_nothing_raised{
+      p=Population.new(PermutationTestInsert,20).evolve_silent(20)
+    }
+    p.each{|s|  assert_equal s.genes.sort, (0...N).to_a }
+  end
 end

data/test/test_sel.rb

@@ -22,10 +22,12 @@ class TestSelection < Test::Unit::TestCase
   end
 
   def test_elitism
-    $sel_mth.map{|s| Elitism(s,rand(4))  }.each{|s|
+    [0,1,5].each{|k|
+    $sel_mth.map{|s| Elitism(s,k)  }.each{|s|
       klass = TestClass(s)
       assert_nothing_raised{ Population.new(klass,10).evolve_silent(10) }
     }
+    }
   end
 
   def test_singlechild

data/test/test_tree.rb

@@ -1,6 +1,6 @@
 require 't_common'
 
-class TreeTest < TreeGenotype([1,:x,proc{rand}], [:+], [:+@])
+class TreeTest < TreeGenotype([1,:x,proc{rand}], [:+@], [:+])
   def fitness
     - (eval_genes(:x=>1) - 5).abs
   end
@@ -12,6 +12,26 @@ class BloatTest < TreeTest
   end
 end
 
+$tree_mutators = [
+   TreeReplaceMutator,
+   TreeReplaceMutator(),
+   TreeReplaceMutator(2,:full),
+   TreeReplaceMutator(0,:grow),
+   TreeReplaceMutator(0..3,:grow),
+   TreeReplaceMutator(3..3,:half),
+   TreeReplaceMutator(:same,:full),
+   TreeReplaceMutator(:same[-1,1],:full),
+   TreePruneMutator,
+   TreeChopMutator,
+   TreeRemoveNodeMutator,
+   TreeInsertNodeMutator,
+   TreeTerminalMutator,
+   TreeNumTerminalMutator,
+   TreeNumTerminalMutator(mutate=:gaussian[2.1]),
+   TreeNumTerminalMutator{|x|x},
+   TreeEvalMutator, NullMutator
+ ]
+
 # hard to test, also see examples/tree.rb
 class TreeTests < Test::Unit::TestCase
   def test_evolve
@@ -53,5 +73,47 @@ class TreeTests < Test::Unit::TestCase
   end
 
 
+  def test_init_full
+    p=Population.new(TreeGenotype([1],[],[:+],5,:full) )
+    p.population.each{|x|
+      assert_equal 5, x.depth, "#{x} has a depth < 5"
+      x.genes[1..-1].each{|st| assert_equal 4, GPTreeHelper.tree_depth(st), "#{st.inspect} has a depth < 4" }
+    }
+  end
+
+  def test_init_grow
+    p=Population.new(TreeGenotype([1],[],[:+],5,:grow), 100)
+    assert  p.population.any?{|x| x.depth < 5 } # with very high probability
+    assert !p.population.any?{|x| x.depth > 5 }
+    assert  p.population.all?{|x| x.depth > 0 }
+  end
+
+  def test_init_half
+    p=Population.new(TreeGenotype([1],[],[:+],5,:half), 100)
+    assert  p.population.any?{|x| x.depth < 5  } # with very high probability
+    assert  p.population.any?{|x| x.depth == 5 } # with very high probability
+    assert !p.population.all?{|x| x.depth == 5 } # with very high probability
+    assert !p.population.any?{|x| x.depth > 5  }
+    assert  p.population.all?{|x| x.depth > 0  }
+  end
+
+
+
+  def test_mutators
+    $tree_mutators.each{|m|
+      klass = Class.new(BloatTest) { use m }
+      assert_nothing_raised("#{m} failed") { Population.new(klass,10).evolve_silent(10) }
+    }
+    $tree_mutators.each{|m|
+      klass = Class.new(TreeTest) { use PMutate(0.5,m) }
+      assert_nothing_raised("#{m} failed") { Population.new(klass,10).evolve_silent(10) }
+    }
+
+    all_mut = Hash[*$tree_mutators.map{|x| [x,1.0 / $tree_mutators.size]}.flatten]
+    assert_nothing_raised {
+      klass = Class.new(TreeTest) { use PMutateN(all_mut) }
+      assert_nothing_raised("PMutateN failed") { Population.new(klass,30).evolve_silent(30) }
+    }
+  end
 
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: charlie
 version: !ruby/object:Gem::Version 
-  version: 0.6.0
+  version: 0.7.0
 platform: ruby
 authors: 
 - Sander Land
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2007-12-26 00:00:00 +01:00
+date: 2008-01-11 00:00:00 +01:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -21,7 +21,7 @@ dependencies:
       - !ruby/object:Gem::Version 
         version: 1.4.0
     version: 
-description: "== DESCRIPTION: Charlie is a library for genetic algorithms (GA) and genetic programming (GP).  == 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  == EXAMPLES: This example solves a TSP problem (also quiz #142): N=5 CITIES = (0...N).map{|i| (0...N).map{|j| [i,j] } }.inject{|a,b|a+b} class TSP < PermutationGenotype(CITIES.size) def fitness d=0 (genes + [genes[0]]).each_cons(2){|a,b|  a,b=CITIES[a],CITIES[b] d += Math.sqrt( (a[0]-b[0])**2 + (a[1]-b[1])**2 )  } -d # lower distance -> higher fitness. end use EdgeRecombinationCrossover, InversionMutator end Population.new(TSP,20).evolve_on_console(50)  This example finds a polynomial which approximates cos(x) class Cos < TreeGenotype([proc{3*rand-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)"
+description: "== DESCRIPTION: Charlie is a library for genetic algorithms (GA) and genetic programming (GP).  == 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  == INSTALL: * sudo gem install charlie  == EXAMPLES: This example solves a TSP problem (also quiz #142): N=5 CITIES = (0...N).map{|i| (0...N).map{|j| [i,j] } }.inject{|a,b|a+b} class TSP < PermutationGenotype(CITIES.size) def fitness d=0 (genes + [genes[0]]).each_cons(2){|a,b|  a,b=CITIES[a],CITIES[b] d += Math.sqrt( (a[0]-b[0])**2 + (a[1]-b[1])**2 )  } -d # lower distance -> higher fitness. end use EdgeRecombinationCrossover, InversionMutator end Population.new(TSP,20).evolve_on_console(50)  This example finds a polynomial which approximates cos(x) class Cos < TreeGenotype([proc{3*rand-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)"
 email: sander.land+ruby@gmail.com
 executables: []
 
@@ -32,6 +32,7 @@ extra_rdoc_files:
 - Manifest.txt
 - README.txt
 - TODO.txt
+- examples/EXAMPLES_README.txt
 - data/GENOTYPE
 - data/SELECTION
 - data/CROSSOVER
@@ -49,35 +50,26 @@ files:
 - data/MUTATION
 - data/SELECTION
 - data/template.html
-- examples/bit.rb
-- examples/function_opt_2peak.rb
-- examples/function_opt_sombero.rb
-- examples/gladiatorial_simple.rb
-- examples/gladiatorial_sunburn.rb
-- examples/gridwalk.rb
-- examples/money.rb
-- examples/output/flattened_sombero.html
-- examples/output/flattened_sombero2_.html
-- examples/output/fopt1_dblopt.html
-- examples/output/hill10.html
-- examples/output/hill2.csv
-- examples/output/hill2.html
-- examples/output/royalroad1_report.html
-- examples/output/royalroad2_report.html
-- examples/output/royalroadquick_report.html
+- examples/EXAMPLES_README.txt
+- examples/bitstring.rb
+- examples/coevolution.rb
+- examples/function_optimization.rb
+- examples/output/bitstring_royalroad.html
+- examples/output/function_optimization_sombrero.html
+- examples/output/function_optimization_twopeak.csv
+- examples/output/function_optimization_twopeak.html
+- examples/output/string_weasel.html
 - examples/output/tsp.html
-- examples/output/weasel1_report.html
-- examples/output/weasel2_report.html
-- examples/royalroad.rb
-- examples/royalroad2.rb
-- examples/simple_climb_hill2.rb
+- examples/permutation.rb
+- examples/string.rb
 - examples/tree.rb
 - examples/tsp.rb
-- examples/weasel.rb
 - lib/charlie.rb
+- lib/charlie/1.9fixes.rb
 - lib/charlie/crossover.rb
 - lib/charlie/etc/minireport.rb
 - lib/charlie/etc/monkey.rb
+- lib/charlie/gabenchmark.rb
 - lib/charlie/genotype.rb
 - lib/charlie/list/list_crossover.rb
 - lib/charlie/list/list_genotype.rb

data/examples/bit.rb

@@ -1,10 +0,0 @@
-require '../lib/charlie'
-
-
-class Test < BitStringGenotype(10)
-  def fitness
-    -(genes.map(&:to_s).join.to_i(2) - 512).abs # or just to_s.to_i(2)
-  end
-end
-Population.new(Test).evolve_on_console
-

data/examples/function_opt_2peak.rb

@@ -1,24 +0,0 @@
-require '../lib/charlie'
-
-class F <  FloatListGenotype(1,(0..1))
-  def fitness 
-    x=genes.first
-    3.1 / (1 + (40*x-4)**2)  + 2.0 / (1+(3*x-2.4)**4)  # two peaks, the left one being higher but very thin.
-  end
-
- # use TournamentSelection(4)
- # use NullCrossover
-end
-
-Population.benchmark(F,'output/fopt1_dblopt.html') {
-  selection BestOnlySelection, TournamentSelection(3), TournamentSelection(4), Elitism(RouletteSelection,1), Elitism(ScaledRouletteSelection(),1), RouletteSelection
-  crossover NullCrossover #, SinglePointCrossover #, UniformCrossover
-
-  mt = [:probability[0.5],:probability[0.75],:probability[1.0]].map{|s|
-         [:uniform[0.1],:uniform[0.2],:uniform[0.4],:uniform[0.8],:uniform[1.6],:gaussian[0.1],:gaussian[0.2],:gaussian[0.4],:gaussian[0.8],:gaussian[1.6]].map{|p| ListMutator(s,p) } }.flatten
-
-  mutator *mt
-  self.generations     = 10
-  self.population_size = 10 
-  self.repeat = 200
-} #if nil

data/examples/function_opt_sombero.rb

@@ -1,38 +0,0 @@
-require '../lib/charlie'
-
-class Sombero <  FloatListGenotype(2,(-4*Math::PI..4*Math::PI))
-  def fitness 
-    x,y=genes
-    1 + ( Math.cos(Math.sqrt(x*x+y*y)) / (1+0.1*(x*x+y*y)) )
-  end
-
- # use TournamentSelection(4)
- # use NullCrossover
-end
-
-#Population.new(Sombero,10).evolve_on_console(20)
-
-Population.benchmark(Sombero,'output/flattened_sombero.html') {
-  selection TruncationSelection(), BestOnlySelection, TournamentSelection(4), RouletteSelection, ScaledRouletteSelection(), Elitism(RouletteSelection,1)
-  crossover NullCrossover,UniformCrossover
-
-  mt = [:expected_n[1],:expected_n[2],:expected_n[3],:expected_n[5],:single_point,:n_point[2],:n_point[3]].map{|s|
-         [:uniform[0.1],:uniform[0.2],:uniform[0.3],:gaussian[0.1],:gaussian[0.2],:gaussian[0.3],:gaussian[0.8]].map{|p| ListMutator(s,p) } }.flatten
-
-  mutator *mt
-  self.generations     = 20
-  self.population_size = 10 
-}
-
-Population.benchmark(Sombero,'output/flattened_sombero2_.html') {
-  selection BestOnlySelection, TournamentSelection(3), TournamentSelection(4), Elitism(RouletteSelection,1), Elitism(ScaledRouletteSelection(),1)
-  crossover NullCrossover, SinglePointCrossover #, UniformCrossover
-
-  mt = [:expected_n[2],:expected_n[3],:expected_n[4],:expected_n[5]].map{|s|
-         [:uniform[0.2],:uniform[0.4],:uniform[0.8],:uniform[1.6],:gaussian[0.2],:gaussian[0.4],:gaussian[0.8],:gaussian[1.6]].map{|p| ListMutator(s,p) } }.flatten
-
-  mutator *mt
-  self.generations     = 15
-  self.population_size = 10
-  self.repeat = 50 
-}

data/examples/gladiatorial_simple.rb

@@ -1,17 +0,0 @@
-require '../lib/charlie'
-
-class TestFighter <  FloatListGenotype(2)
-  def fight(other)
-    genes.sum >= other.genes.sum
-  end
-  def to_s
-    genes.map{|f| '%02f' % f }.join(', ')
-  end
-  use GladiatorialSelection
-  use UniformCrossover
-  use ListMutator(:expected_n[1],:gaussian[0.25])
-end
-
-pop = Population.new(TestFighter,10).evolve_silent(200)
-puts pop
-

data/examples/gridwalk.rb

@@ -1,29 +0,0 @@
-require '../lib/charlie'
-
-DIRS = [[-1,0],[1,0],[0,-1],[0,1]]
-class Walk <  StringGenotype(25,DIRS)  # Walk with steps in 4 directions
-
-  def fitness 
-    grid = Array.new(5){Array.new(5)}
-    x,y=0,0
-    genes.each{|dx,dy|
-      grid[x][y] = :visited
-      nx,ny = x+dx,y+dy
-      x,y=nx,ny if (0..4)===nx && (0..4)===ny && grid[nx][ny].nil? # on grid and haven't been there before
-    }
-    grid[x][y] = :visited
-    grid.flatten.compact.size
-  end
-  def to_s; ''; end
-
-  use TournamentSelection(4)
-  #use ListMutator(:expected_n[2],:replace[*DIRS])
-end
-
-pop = nil
-begin
-  pop = Population.new(Walk,20).evolve_silent(100)
-  print '.'; $stdout.flush
-end while pop[-1].fitness != 25
-puts 'Found walk of length 25:', pop[-1].genes.inspect
-