Merge pull request #8328 from imapp-pl/yul-phaser-gewep-algorithm

[yul-phaser] GEWEP algorithm
This commit is contained in:
chriseth 2020-03-12 14:54:33 +01:00 committed by GitHub
commit 47b72032c6
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
18 changed files with 1415 additions and 11 deletions

View File

@ -145,6 +145,8 @@ set(yul_phaser_sources
yulPhaser/Chromosome.cpp
yulPhaser/FitnessMetrics.cpp
yulPhaser/GeneticAlgorithms.cpp
yulPhaser/Mutations.cpp
yulPhaser/PairSelections.cpp
yulPhaser/Population.cpp
yulPhaser/Program.cpp
yulPhaser/Selections.cpp
@ -156,6 +158,8 @@ set(yul_phaser_sources
../tools/yulPhaser/Chromosome.cpp
../tools/yulPhaser/FitnessMetrics.cpp
../tools/yulPhaser/GeneticAlgorithms.cpp
../tools/yulPhaser/Mutations.cpp
../tools/yulPhaser/PairSelections.cpp
../tools/yulPhaser/Population.cpp
../tools/yulPhaser/Program.cpp
../tools/yulPhaser/Selections.cpp

View File

@ -24,6 +24,24 @@
using namespace std;
using namespace solidity;
using namespace solidity::yul;
using namespace solidity::phaser;
function<Mutation> phaser::test::wholeChromosomeReplacement(Chromosome _newChromosome)
{
return [_newChromosome = move(_newChromosome)](Chromosome const&) { return _newChromosome; };
}
function<Mutation> phaser::test::geneSubstitution(size_t _geneIndex, string _geneValue)
{
return [=](Chromosome const& _chromosome)
{
vector<string> newGenes = _chromosome.optimisationSteps();
assert(_geneIndex < newGenes.size());
newGenes[_geneIndex] = _geneValue;
return Chromosome(newGenes);
};
}
vector<size_t> phaser::test::chromosomeLengths(Population const& _population)
{
@ -44,6 +62,15 @@ map<string, size_t> phaser::test::enumerateOptmisationSteps()
return stepIndices;
}
size_t phaser::test::countDifferences(Chromosome const& _chromosome1, Chromosome const& _chromosome2)
{
size_t count = 0;
for (size_t i = 0; i < min(_chromosome1.length(), _chromosome2.length()); ++i)
count += static_cast<int>(_chromosome1.optimisationSteps()[i] != _chromosome2.optimisationSteps()[i]);
return count + abs(static_cast<int>(_chromosome1.length() - _chromosome2.length()));
}
string phaser::test::stripWhitespace(string const& input)
{
regex whitespaceRegex("\\s+");

View File

@ -30,9 +30,11 @@
#include <tools/yulPhaser/Chromosome.h>
#include <tools/yulPhaser/FitnessMetrics.h>
#include <tools/yulPhaser/Mutations.h>
#include <tools/yulPhaser/Population.h>
#include <cassert>
#include <functional>
#include <map>
#include <string>
#include <vector>
@ -52,11 +54,26 @@ public:
size_t evaluate(Chromosome const& _chromosome) const override { return _chromosome.length(); }
};
// MUTATIONS
/// Mutation that always replaces the whole chromosome with the one specified in the parameter.
std::function<Mutation> wholeChromosomeReplacement(Chromosome _newChromosome);
/// Mutation that always replaces the optimisation step at position @a _geneIndex with @a _geneValue.
///
/// The chromosome must be long enough for this position to exist.
std::function<Mutation> geneSubstitution(size_t _geneIndex, std::string _geneValue);
// CHROMOSOME AND POPULATION HELPERS
/// Returns a vector containing lengths of all chromosomes in the population (in the same order).
std::vector<size_t> chromosomeLengths(Population const& _population);
/// Returns the number of genes that differ between two chromosomes.
/// If the chromnosomes have different lengths, the positions that are present in only one of them
/// are counted as mismatches.
size_t countDifferences(Chromosome const& _chromosome1, Chromosome const& _chromosome2);
/// Assigns indices from 0 to N to all optimisation steps available in the OptimiserSuite.
/// This is a convenience helper to make it easier to test their distribution with tools made for
/// integers.

View File

@ -40,6 +40,23 @@ BOOST_AUTO_TEST_CASE(ChromosomeLengthMetric_evaluate_should_return_chromosome_le
BOOST_TEST(ChromosomeLengthMetric{}.evaluate(Chromosome("aaaaa")) == 5);
}
BOOST_AUTO_TEST_CASE(wholeChromosomeReplacement_should_replace_whole_chromosome_with_another)
{
function<Mutation> mutation = wholeChromosomeReplacement(Chromosome("aaa"));
BOOST_TEST(mutation(Chromosome("ccc")) == Chromosome("aaa"));
}
BOOST_AUTO_TEST_CASE(geneSubstitution_should_change_a_single_gene_at_a_given_index)
{
Chromosome chromosome("aaccff");
function<Mutation> mutation1 = geneSubstitution(0, chromosome.optimisationSteps()[5]);
BOOST_TEST(mutation1(chromosome) == Chromosome("faccff"));
function<Mutation> mutation2 = geneSubstitution(5, chromosome.optimisationSteps()[0]);
BOOST_TEST(mutation2(chromosome) == Chromosome("aaccfa"));
}
BOOST_AUTO_TEST_CASE(chromosomeLengths_should_return_lengths_of_all_chromosomes_in_a_population)
{
shared_ptr<FitnessMetric> fitnessMetric = make_shared<ChromosomeLengthMetric>();
@ -51,6 +68,34 @@ BOOST_AUTO_TEST_CASE(chromosomeLengths_should_return_lengths_of_all_chromosomes_
BOOST_TEST((chromosomeLengths(population2) == vector<size_t>{}));
}
BOOST_AUTO_TEST_CASE(countDifferences_should_return_zero_for_identical_chromosomes)
{
BOOST_TEST(countDifferences(Chromosome(), Chromosome()) == 0);
BOOST_TEST(countDifferences(Chromosome("a"), Chromosome("a")) == 0);
BOOST_TEST(countDifferences(Chromosome("afxT"), Chromosome("afxT")) == 0);
}
BOOST_AUTO_TEST_CASE(countDifferences_should_count_mismatched_positions_in_chromosomes_of_the_same_length)
{
BOOST_TEST(countDifferences(Chromosome("a"), Chromosome("f")) == 1);
BOOST_TEST(countDifferences(Chromosome("aa"), Chromosome("ac")) == 1);
BOOST_TEST(countDifferences(Chromosome("ac"), Chromosome("cc")) == 1);
BOOST_TEST(countDifferences(Chromosome("aa"), Chromosome("cc")) == 2);
BOOST_TEST(countDifferences(Chromosome("afxT"), Chromosome("Txfa")) == 4);
}
BOOST_AUTO_TEST_CASE(countDifferences_should_count_missing_characters_as_differences)
{
BOOST_TEST(countDifferences(Chromosome(""), Chromosome("a")) == 1);
BOOST_TEST(countDifferences(Chromosome("a"), Chromosome("")) == 1);
BOOST_TEST(countDifferences(Chromosome("aa"), Chromosome("")) == 2);
BOOST_TEST(countDifferences(Chromosome("aaa"), Chromosome("")) == 3);
BOOST_TEST(countDifferences(Chromosome("aa"), Chromosome("aaaa")) == 2);
BOOST_TEST(countDifferences(Chromosome("aa"), Chromosome("aacc")) == 2);
BOOST_TEST(countDifferences(Chromosome("aa"), Chromosome("cccc")) == 4);
}
BOOST_AUTO_TEST_CASE(enumerateOptimisationSteps_should_assing_indices_to_all_available_optimisation_steps)
{
map<string, char> stepsAndAbbreviations = OptimiserSuite::stepNameToAbbreviationMap();

View File

@ -29,6 +29,7 @@
#include <boost/test/unit_test.hpp>
#include <boost/test/tools/output_test_stream.hpp>
#include <algorithm>
#include <vector>
using namespace std;
@ -132,6 +133,107 @@ BOOST_FIXTURE_TEST_CASE(runNextRound_should_not_replace_any_chromosomes_if_whole
BOOST_TEST((chromosomeLengths(algorithm.population()) == vector<size_t>{3, 3, 3, 3, 5, 5, 5, 5}));
}
BOOST_AUTO_TEST_SUITE_END()
BOOST_AUTO_TEST_SUITE(GenerationalElitistWithExclusivePoolsTest)
BOOST_FIXTURE_TEST_CASE(runNextRound_should_preserve_elite_and_regenerate_rest_of_population, GeneticAlgorithmFixture)
{
auto population = Population::makeRandom(m_fitnessMetric, 6, 3, 3) + Population::makeRandom(m_fitnessMetric, 4, 5, 5);
GenerationalElitistWithExclusivePools::Options options = {
/* mutationPoolSize = */ 0.2,
/* crossoverPoolSize = */ 0.2,
/* randomisationChance = */ 0.0,
/* deletionVsAdditionChance = */ 1.0,
/* percentGenesToRandomise = */ 0.0,
/* percentGenesToAddOrDelete = */ 1.0,
};
GenerationalElitistWithExclusivePools algorithm(population, m_output, options);
assert((chromosomeLengths(algorithm.population()) == vector<size_t>{3, 3, 3, 3, 3, 3, 5, 5, 5, 5}));
algorithm.runNextRound();
BOOST_TEST((chromosomeLengths(algorithm.population()) == vector<size_t>{0, 0, 3, 3, 3, 3, 3, 3, 3, 3}));
}
BOOST_FIXTURE_TEST_CASE(runNextRound_should_not_replace_elite_with_worse_individuals, GeneticAlgorithmFixture)
{
auto population = Population::makeRandom(m_fitnessMetric, 6, 3, 3) + Population::makeRandom(m_fitnessMetric, 4, 5, 5);
GenerationalElitistWithExclusivePools::Options options = {
/* mutationPoolSize = */ 0.2,
/* crossoverPoolSize = */ 0.2,
/* randomisationChance = */ 0.0,
/* deletionVsAdditionChance = */ 0.0,
/* percentGenesToRandomise = */ 0.0,
/* percentGenesToAddOrDelete = */ 1.0,
};
GenerationalElitistWithExclusivePools algorithm(population, m_output, options);
assert(chromosomeLengths(algorithm.population()) == (vector<size_t>{3, 3, 3, 3, 3, 3, 5, 5, 5, 5}));
algorithm.runNextRound();
BOOST_TEST((chromosomeLengths(algorithm.population()) == vector<size_t>{3, 3, 3, 3, 3, 3, 3, 3, 7, 7}));
}
BOOST_FIXTURE_TEST_CASE(runNextRound_should_generate_individuals_in_the_crossover_pool_by_mutating_the_elite, GeneticAlgorithmFixture)
{
auto population = Population::makeRandom(m_fitnessMetric, 20, 5, 5);
GenerationalElitistWithExclusivePools::Options options = {
/* mutationPoolSize = */ 0.8,
/* crossoverPoolSize = */ 0.0,
/* randomisationChance = */ 0.5,
/* deletionVsAdditionChance = */ 0.5,
/* percentGenesToRandomise = */ 1.0,
/* percentGenesToAddOrDelete = */ 1.0,
};
GenerationalElitistWithExclusivePools algorithm(population, m_output, options);
SimulationRNG::reset(1);
algorithm.runNextRound();
BOOST_TEST((
chromosomeLengths(algorithm.population()) ==
vector<size_t>{0, 0, 0, 0, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 11, 11, 11}
));
}
BOOST_FIXTURE_TEST_CASE(runNextRound_should_generate_individuals_in_the_crossover_pool_by_crossing_over_the_elite, GeneticAlgorithmFixture)
{
auto population = (
Population(m_fitnessMetric, {Chromosome("aa"), Chromosome("ff")}) +
Population::makeRandom(m_fitnessMetric, 8, 6, 6)
);
GenerationalElitistWithExclusivePools::Options options = {
/* mutationPoolSize = */ 0.0,
/* crossoverPoolSize = */ 0.8,
/* randomisationChance = */ 0.0,
/* deletionVsAdditionChance = */ 0.0,
/* percentGenesToRandomise = */ 0.0,
/* percentGenesToAddOrDelete = */ 0.0,
};
GenerationalElitistWithExclusivePools algorithm(population, m_output, options);
assert((chromosomeLengths(algorithm.population()) == vector<size_t>{2, 2, 6, 6, 6, 6, 6, 6, 6, 6}));
SimulationRNG::reset(1);
algorithm.runNextRound();
vector<Individual> const& newIndividuals = algorithm.population().individuals();
BOOST_TEST((chromosomeLengths(algorithm.population()) == vector<size_t>{2, 2, 2, 2, 2, 2, 2, 2, 2, 2}));
for (auto& individual: newIndividuals)
BOOST_TEST((
individual.chromosome == Chromosome("aa") ||
individual.chromosome == Chromosome("af") ||
individual.chromosome == Chromosome("fa") ||
individual.chromosome == Chromosome("ff")
));
BOOST_TEST(any_of(newIndividuals.begin() + 2, newIndividuals.end(), [](auto& individual){
return individual.chromosome != Chromosome("aa") && individual.chromosome != Chromosome("ff");
}));
}
BOOST_AUTO_TEST_SUITE_END()
BOOST_AUTO_TEST_SUITE_END()
BOOST_AUTO_TEST_SUITE_END()

View File

@ -0,0 +1,394 @@
/*
This file is part of solidity.
solidity is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
solidity is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with solidity. If not, see <http://www.gnu.org/licenses/>.
*/
#include <test/yulPhaser/Common.h>
#include <tools/yulPhaser/Mutations.h>
#include <tools/yulPhaser/SimulationRNG.h>
#include <boost/test/unit_test.hpp>
#include <algorithm>
#include <vector>
using namespace std;
namespace solidity::phaser::test
{
BOOST_AUTO_TEST_SUITE(Phaser)
BOOST_AUTO_TEST_SUITE(MutationsTest)
BOOST_AUTO_TEST_SUITE(GeneRandomisationTest)
BOOST_AUTO_TEST_CASE(geneRandomisation_should_iterate_over_genes_and_replace_them_with_random_ones_with_given_probability)
{
Chromosome chromosome("fcCUnDvejs");
function<Mutation> mutation01 = geneRandomisation(0.1);
function<Mutation> mutation05 = geneRandomisation(0.5);
function<Mutation> mutation10 = geneRandomisation(1.0);
SimulationRNG::reset(1);
BOOST_TEST(countDifferences(mutation01(chromosome), chromosome), 2);
BOOST_TEST(countDifferences(mutation05(chromosome), chromosome), 5);
BOOST_TEST(countDifferences(mutation10(chromosome), chromosome), 7);
SimulationRNG::reset(2);
BOOST_TEST(countDifferences(mutation01(chromosome), chromosome), 1);
BOOST_TEST(countDifferences(mutation05(chromosome), chromosome), 3);
BOOST_TEST(countDifferences(mutation10(chromosome), chromosome), 9);
}
BOOST_AUTO_TEST_CASE(geneRandomisation_should_return_identical_chromosome_if_probability_is_zero)
{
Chromosome chromosome("fcCUnDvejsrmV");
function<Mutation> mutation = geneRandomisation(0.0);
BOOST_TEST(mutation(chromosome) == chromosome);
}
BOOST_AUTO_TEST_CASE(geneDeletion_should_iterate_over_genes_and_delete_them_with_given_probability)
{
Chromosome chromosome("fcCUnDvejs");
function<Mutation> mutation01 = geneDeletion(0.1);
function<Mutation> mutation05 = geneDeletion(0.5);
SimulationRNG::reset(1);
// fcCUnDvejs
BOOST_TEST(mutation01(chromosome) == Chromosome(stripWhitespace("fcCU Dvejs")));
BOOST_TEST(mutation05(chromosome) == Chromosome(stripWhitespace(" D ejs")));
SimulationRNG::reset(2);
BOOST_TEST(mutation01(chromosome) == Chromosome(stripWhitespace("fcUnDvejs")));
BOOST_TEST(mutation05(chromosome) == Chromosome(stripWhitespace(" Un s")));
}
BOOST_AUTO_TEST_CASE(geneDeletion_should_return_identical_chromosome_if_probability_is_zero)
{
Chromosome chromosome("fcCUnDvejsrmV");
function<Mutation> mutation = geneDeletion(0.0);
BOOST_TEST(mutation(chromosome) == chromosome);
}
BOOST_AUTO_TEST_CASE(geneDeletion_should_delete_all_genes_if_probability_is_one)
{
Chromosome chromosome("fcCUnDvejsrmV");
function<Mutation> mutation = geneDeletion(1.0);
BOOST_TEST(mutation(chromosome) == Chromosome(""));
}
BOOST_AUTO_TEST_CASE(geneAddition_should_iterate_over_gene_positions_and_insert_new_genes_with_given_probability)
{
Chromosome chromosome("fcCUnDvejs");
function<Mutation> mutation01 = geneAddition(0.1);
function<Mutation> mutation05 = geneAddition(0.5);
SimulationRNG::reset(1);
// f c C U n D v e j s
BOOST_TEST(mutation01(chromosome) == Chromosome(stripWhitespace(" f c C UC n D v e jx s"))); // 20% more
BOOST_TEST(mutation05(chromosome) == Chromosome(stripWhitespace("j f cu C U ne D v eI j sf"))); // 50% more
SimulationRNG::reset(2);
BOOST_TEST(mutation01(chromosome) == Chromosome(stripWhitespace(" f cu C U n D v e j s"))); // 10% more
BOOST_TEST(mutation05(chromosome) == Chromosome(stripWhitespace("L f ce Cv U n D v e jO s"))); // 40% more
}
BOOST_AUTO_TEST_CASE(geneAddition_should_be_able_to_insert_before_first_position)
{
SimulationRNG::reset(7);
Chromosome chromosome("fcCUnDvejs");
function<Mutation> mutation = geneAddition(0.1);
Chromosome mutatedChromosome = mutation(chromosome);
BOOST_TEST(mutatedChromosome.length() > chromosome.length());
vector<string> suffix(
mutatedChromosome.optimisationSteps().end() - chromosome.length(),
mutatedChromosome.optimisationSteps().end()
);
BOOST_TEST(suffix == chromosome.optimisationSteps());
}
BOOST_AUTO_TEST_CASE(geneAddition_should_be_able_to_insert_after_last_position)
{
SimulationRNG::reset(81);
Chromosome chromosome("fcCUnDvejs");
function<Mutation> mutation = geneAddition(0.1);
Chromosome mutatedChromosome = mutation(chromosome);
BOOST_TEST(mutatedChromosome.length() > chromosome.length());
vector<string> prefix(
mutatedChromosome.optimisationSteps().begin(),
mutatedChromosome.optimisationSteps().begin() + chromosome.length()
);
BOOST_TEST(prefix == chromosome.optimisationSteps());
}
BOOST_AUTO_TEST_CASE(geneAddition_should_return_identical_chromosome_if_probability_is_zero)
{
Chromosome chromosome("fcCUnDvejsrmV");
function<Mutation> mutation = geneAddition(0.0);
BOOST_TEST(mutation(chromosome) == chromosome);
}
BOOST_AUTO_TEST_CASE(geneAddition_should_insert_genes_at_all_positions_if_probability_is_one)
{
Chromosome chromosome("fcCUnDvejsrmV");
function<Mutation> mutation = geneAddition(1.0);
Chromosome mutatedChromosome = mutation(chromosome);
BOOST_TEST(mutatedChromosome.length() == chromosome.length() * 2 + 1);
vector<string> originalGenes;
for (size_t i = 0; i < mutatedChromosome.length() - 1; ++i)
if (i % 2 == 1)
originalGenes.push_back(mutatedChromosome.optimisationSteps()[i]);
BOOST_TEST(Chromosome(originalGenes) == chromosome);
}
BOOST_AUTO_TEST_CASE(alternativeMutations_should_choose_between_mutations_with_given_probability)
{
SimulationRNG::reset(1);
Chromosome chromosome("a");
function<Mutation> mutation = alternativeMutations(
0.8,
wholeChromosomeReplacement(Chromosome("c")),
wholeChromosomeReplacement(Chromosome("f"))
);
size_t cCount = 0;
size_t fCount = 0;
for (size_t i = 0; i < 10; ++i)
{
Chromosome mutatedChromosome = mutation(chromosome);
cCount += static_cast<int>(mutatedChromosome == Chromosome("c"));
fCount += static_cast<int>(mutatedChromosome == Chromosome("f"));
}
// This particular seed results in 7 "c"s out of 10 which looks plausible given the 80% chance.
BOOST_TEST(cCount == 7);
BOOST_TEST(fCount == 3);
}
BOOST_AUTO_TEST_CASE(alternativeMutations_should_always_choose_first_mutation_if_probability_is_one)
{
Chromosome chromosome("a");
function<Mutation> mutation = alternativeMutations(
1.0,
wholeChromosomeReplacement(Chromosome("c")),
wholeChromosomeReplacement(Chromosome("f"))
);
for (size_t i = 0; i < 10; ++i)
BOOST_TEST(mutation(chromosome) == Chromosome("c"));
}
BOOST_AUTO_TEST_CASE(alternativeMutations_should_always_choose_second_mutation_if_probability_is_zero)
{
Chromosome chromosome("a");
function<Mutation> mutation = alternativeMutations(
0.0,
wholeChromosomeReplacement(Chromosome("c")),
wholeChromosomeReplacement(Chromosome("f"))
);
for (size_t i = 0; i < 10; ++i)
BOOST_TEST(mutation(chromosome) == Chromosome("f"));
}
BOOST_AUTO_TEST_CASE(randomPointCrossover_should_swap_chromosome_parts_at_random_point)
{
function<Crossover> crossover = randomPointCrossover();
SimulationRNG::reset(1);
Chromosome result1 = crossover(Chromosome("aaaaaaaaaa"), Chromosome("cccccc"));
BOOST_TEST(result1 == Chromosome("aaaccc"));
SimulationRNG::reset(1);
Chromosome result2 = crossover(Chromosome("cccccc"), Chromosome("aaaaaaaaaa"));
BOOST_TEST(result2 == Chromosome("cccaaaaaaa"));
}
BOOST_AUTO_TEST_CASE(randomPointCrossover_should_only_consider_points_available_on_both_chromosomes)
{
SimulationRNG::reset(1);
function<Crossover> crossover = randomPointCrossover();
for (size_t i = 0; i < 30; ++i)
{
Chromosome result1 = crossover(Chromosome("aaa"), Chromosome("TTTTTTTTTTTTTTTTTTTT"));
Chromosome result2 = crossover(Chromosome("TTTTTTTTTTTTTTTTTTTT"), Chromosome("aaa"));
BOOST_TEST((
result1 == Chromosome("TTTTTTTTTTTTTTTTTTTT") ||
result1 == Chromosome("aTTTTTTTTTTTTTTTTTTT") ||
result1 == Chromosome("aaTTTTTTTTTTTTTTTTTT") ||
result1 == Chromosome("aaaTTTTTTTTTTTTTTTTT")
));
BOOST_TEST((
result2 == Chromosome("aaa") ||
result2 == Chromosome("Taa") ||
result2 == Chromosome("TTa") ||
result2 == Chromosome("TTT")
));
}
}
BOOST_AUTO_TEST_CASE(randomPointCrossover_should_never_split_at_position_zero_if_chromosomes_are_splittable)
{
SimulationRNG::reset(1);
function<Crossover> crossover = randomPointCrossover();
for (size_t i = 0; i < 30; ++i)
{
Chromosome result1 = crossover(Chromosome("aa"), Chromosome("TTTTTTTTTTTTTTTTTTTT"));
Chromosome result2 = crossover(Chromosome("TTTTTTTTTTTTTTTTTTTT"), Chromosome("aa"));
BOOST_TEST(result1 != Chromosome("TTTTTTTTTTTTTTTTTTTT"));
BOOST_TEST(result2 != Chromosome("aa"));
}
}
BOOST_AUTO_TEST_CASE(randomPointCrossover_should_never_split_at_position_zero_if_chromosomes_are_not_empty)
{
SimulationRNG::reset(1);
function<Crossover> crossover = randomPointCrossover();
for (size_t i = 0; i < 30; ++i)
{
Chromosome result1 = crossover(Chromosome("a"), Chromosome("T"));
Chromosome result2 = crossover(Chromosome("T"), Chromosome("a"));
BOOST_TEST(result1 == Chromosome("a"));
BOOST_TEST(result2 == Chromosome("T"));
}
}
BOOST_AUTO_TEST_CASE(randomPointCrossover_should_work_even_if_one_chromosome_is_unsplittable)
{
function<Crossover> crossover = randomPointCrossover();
SimulationRNG::reset(1);
BOOST_CHECK(crossover(Chromosome("ff"), Chromosome("a")) == Chromosome("f"));
BOOST_CHECK(crossover(Chromosome("a"), Chromosome("ff")) == Chromosome("af"));
}
BOOST_AUTO_TEST_CASE(randomPointCrossover_should_split_at_position_zero_only_if_at_least_one_chromosome_is_empty)
{
Chromosome empty("");
Chromosome unsplittable("a");
Chromosome splittable("aaaa");
function<Crossover> crossover = randomPointCrossover();
SimulationRNG::reset(1);
BOOST_CHECK(crossover(empty, empty) == empty);
BOOST_CHECK(crossover(unsplittable, empty) == empty);
BOOST_CHECK(crossover(empty, unsplittable) == unsplittable);
BOOST_CHECK(crossover(splittable, empty) == empty);
BOOST_CHECK(crossover(empty, splittable) == splittable);
}
BOOST_AUTO_TEST_CASE(fixedPointCrossover_should_swap_chromosome_parts_at_given_point)
{
Chromosome result1 = fixedPointCrossover(0.8)(Chromosome("aaaaaaaaaa"), Chromosome("cccccccccc"));
Chromosome result2 = fixedPointCrossover(0.8)(Chromosome("cccccccccc"), Chromosome("aaaaaaaaaa"));
BOOST_TEST(result1 == Chromosome("aaaaaaaacc"));
BOOST_TEST(result2 == Chromosome("ccccccccaa"));
}
BOOST_AUTO_TEST_CASE(fixedPointCrossover_should_determine_crossover_point_based_on_length_of_shorter_chromosome)
{
Chromosome result1 = fixedPointCrossover(0.4)(Chromosome("aaaaa"), Chromosome("cccccccccc"));
Chromosome result2 = fixedPointCrossover(0.4)(Chromosome("cccccccccc"), Chromosome("aaaaa"));
BOOST_TEST(result1 == Chromosome("aacccccccc"));
BOOST_TEST(result2 == Chromosome("ccaaa"));
}
BOOST_AUTO_TEST_CASE(fixedPointCrossover_should_round_split_point)
{
Chromosome result1 = fixedPointCrossover(0.49)(Chromosome("aaaaa"), Chromosome("ccccc"));
Chromosome result2 = fixedPointCrossover(0.49)(Chromosome("ccccc"), Chromosome("aaaaa"));
BOOST_TEST(result1 == Chromosome("aaccc"));
BOOST_TEST(result2 == Chromosome("ccaaa"));
Chromosome result3 = fixedPointCrossover(0.50)(Chromosome("aaaaa"), Chromosome("ccccc"));
Chromosome result4 = fixedPointCrossover(0.50)(Chromosome("ccccc"), Chromosome("aaaaa"));
BOOST_TEST(result3 == Chromosome("aaacc"));
BOOST_TEST(result4 == Chromosome("cccaa"));
Chromosome result5 = fixedPointCrossover(0.51)(Chromosome("aaaaa"), Chromosome("ccccc"));
Chromosome result6 = fixedPointCrossover(0.51)(Chromosome("ccccc"), Chromosome("aaaaa"));
BOOST_TEST(result5 == Chromosome("aaacc"));
BOOST_TEST(result6 == Chromosome("cccaa"));
}
BOOST_AUTO_TEST_CASE(fixedPointCrossover_should_split_at_position_zero_if_explicitly_requested)
{
Chromosome result1 = fixedPointCrossover(0.0)(Chromosome("aaaaa"), Chromosome("cccccccccc"));
Chromosome result2 = fixedPointCrossover(0.0)(Chromosome("cccccccccc"), Chromosome("aaaaa"));
BOOST_TEST(result1 == Chromosome("cccccccccc"));
BOOST_TEST(result2 == Chromosome("aaaaa"));
}
BOOST_AUTO_TEST_CASE(fixedPointCrossover_should_split_at_end_of_shorter_chromosome_if_crossover_point_is_after_last_position)
{
Chromosome result1 = fixedPointCrossover(1.0)(Chromosome("aaaaa"), Chromosome("cccccccccc"));
Chromosome result2 = fixedPointCrossover(1.0)(Chromosome("cccccccccc"), Chromosome("aaaaa"));
BOOST_TEST(result1 == Chromosome("aaaaaccccc"));
BOOST_TEST(result2 == Chromosome("ccccc"));
}
BOOST_AUTO_TEST_CASE(fixedPointCrossover_should_select_correct_split_point_for_unsplittable_chromosomes)
{
function<Crossover> crossover00 = fixedPointCrossover(0.0);
BOOST_CHECK(crossover00(Chromosome("fff"), Chromosome("a")) == Chromosome("a"));
BOOST_CHECK(crossover00(Chromosome("a"), Chromosome("fff")) == Chromosome("fff"));
BOOST_CHECK(crossover00(Chromosome("f"), Chromosome("a")) == Chromosome("a"));
function<Crossover> crossover10 = fixedPointCrossover(1.0);
BOOST_CHECK(crossover10(Chromosome("fff"), Chromosome("a")) == Chromosome("f"));
BOOST_CHECK(crossover10(Chromosome("a"), Chromosome("fff")) == Chromosome("aff"));
BOOST_CHECK(crossover10(Chromosome("f"), Chromosome("a")) == Chromosome("f"));
}
BOOST_AUTO_TEST_CASE(fixedPointCrossover_should_always_use_position_zero_as_split_point_when_chromosome_empty)
{
Chromosome empty("");
Chromosome unsplittable("f");
Chromosome splittable("aaaa");
function<Crossover> crossover00 = fixedPointCrossover(0.0);
BOOST_CHECK(crossover00(empty, empty) == empty);
BOOST_CHECK(crossover00(unsplittable, empty) == empty);
BOOST_CHECK(crossover00(empty, unsplittable) == unsplittable);
BOOST_CHECK(crossover00(splittable, empty) == empty);
BOOST_CHECK(crossover00(empty, splittable) == splittable);
function<Crossover> crossover10 = fixedPointCrossover(1.0);
BOOST_CHECK(crossover10(empty, empty) == empty);
BOOST_CHECK(crossover10(unsplittable, empty) == empty);
BOOST_CHECK(crossover10(empty, unsplittable) == unsplittable);
BOOST_CHECK(crossover10(splittable, empty) == empty);
BOOST_CHECK(crossover10(empty, splittable) == splittable);
}
BOOST_AUTO_TEST_SUITE_END()
BOOST_AUTO_TEST_SUITE_END()
BOOST_AUTO_TEST_SUITE_END()
}

View File

@ -0,0 +1,185 @@
/*
This file is part of solidity.
solidity is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
solidity is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with solidity. If not, see <http://www.gnu.org/licenses/>.
*/
#include <test/yulPhaser/Common.h>
#include <tools/yulPhaser/PairSelections.h>
#include <tools/yulPhaser/SimulationRNG.h>
#include <boost/test/unit_test.hpp>
#include <algorithm>
#include <tuple>
#include <vector>
using namespace std;
namespace solidity::phaser::test
{
BOOST_AUTO_TEST_SUITE(Phaser)
BOOST_AUTO_TEST_SUITE(PairSelectionsTest)
BOOST_AUTO_TEST_SUITE(RandomPairSelectionTest)
BOOST_AUTO_TEST_CASE(materialise_should_return_random_values_with_equal_probabilities)
{
constexpr int collectionSize = 10;
constexpr int selectionSize = 100;
constexpr double relativeTolerance = 0.1;
constexpr double expectedValue = (collectionSize - 1) / 2.0;
constexpr double variance = (collectionSize * collectionSize - 1) / 12.0;
SimulationRNG::reset(1);
vector<tuple<size_t, size_t>> pairs = RandomPairSelection(selectionSize).materialise(collectionSize);
vector<size_t> samples;
for (auto& [first, second]: pairs)
{
samples.push_back(first);
samples.push_back(second);
}
BOOST_TEST(abs(mean(samples) - expectedValue) < expectedValue * relativeTolerance);
BOOST_TEST(abs(meanSquaredError(samples, expectedValue) - variance) < variance * relativeTolerance);
}
BOOST_AUTO_TEST_CASE(materialise_should_return_only_values_that_can_be_used_as_collection_indices)
{
const size_t collectionSize = 200;
vector<tuple<size_t, size_t>> pairs = RandomPairSelection(0.5).materialise(collectionSize);
BOOST_TEST(pairs.size() == 100);
BOOST_TEST(all_of(pairs.begin(), pairs.end(), [&](auto const& pair){ return get<0>(pair) <= collectionSize; }));
BOOST_TEST(all_of(pairs.begin(), pairs.end(), [&](auto const& pair){ return get<1>(pair) <= collectionSize; }));
}
BOOST_AUTO_TEST_CASE(materialise_should_never_return_a_pair_of_identical_indices)
{
vector<tuple<size_t, size_t>> pairs = RandomPairSelection(0.5).materialise(100);
BOOST_TEST(pairs.size() == 50);
BOOST_TEST(all_of(pairs.begin(), pairs.end(), [](auto const& pair){ return get<0>(pair) != get<1>(pair); }));
}
BOOST_AUTO_TEST_CASE(materialise_should_return_number_of_pairs_thats_a_fraction_of_collection_size)
{
BOOST_TEST(RandomPairSelection(0.0).materialise(10).size() == 0);
BOOST_TEST(RandomPairSelection(0.3).materialise(10).size() == 3);
BOOST_TEST(RandomPairSelection(0.5).materialise(10).size() == 5);
BOOST_TEST(RandomPairSelection(0.7).materialise(10).size() == 7);
BOOST_TEST(RandomPairSelection(1.0).materialise(10).size() == 10);
}
BOOST_AUTO_TEST_CASE(materialise_should_support_number_of_pairs_bigger_than_collection_size)
{
BOOST_TEST(RandomPairSelection(2.0).materialise(5).size() == 10);
BOOST_TEST(RandomPairSelection(1.5).materialise(10).size() == 15);
BOOST_TEST(RandomPairSelection(10.0).materialise(10).size() == 100);
}
BOOST_AUTO_TEST_CASE(materialise_should_round_the_number_of_pairs_to_the_nearest_integer)
{
BOOST_TEST(RandomPairSelection(0.49).materialise(3).size() == 1);
BOOST_TEST(RandomPairSelection(0.50).materialise(3).size() == 2);
BOOST_TEST(RandomPairSelection(0.51).materialise(3).size() == 2);
BOOST_TEST(RandomPairSelection(1.51).materialise(3).size() == 5);
BOOST_TEST(RandomPairSelection(0.01).materialise(2).size() == 0);
BOOST_TEST(RandomPairSelection(0.01).materialise(3).size() == 0);
}
BOOST_AUTO_TEST_CASE(materialise_should_return_no_pairs_if_collection_is_empty)
{
BOOST_TEST(RandomPairSelection(0).materialise(0).empty());
BOOST_TEST(RandomPairSelection(0.5).materialise(0).empty());
BOOST_TEST(RandomPairSelection(1.0).materialise(0).empty());
BOOST_TEST(RandomPairSelection(2.0).materialise(0).empty());
}
BOOST_AUTO_TEST_CASE(materialise_should_return_no_pairs_if_collection_has_one_element)
{
BOOST_TEST(RandomPairSelection(0).materialise(1).empty());
BOOST_TEST(RandomPairSelection(0.5).materialise(1).empty());
BOOST_TEST(RandomPairSelection(1.0).materialise(1).empty());
BOOST_TEST(RandomPairSelection(2.0).materialise(1).empty());
}
BOOST_AUTO_TEST_SUITE_END()
BOOST_AUTO_TEST_SUITE(PairMosaicSelectionTest)
using IndexPairs = vector<tuple<size_t, size_t>>;
BOOST_AUTO_TEST_CASE(materialise)
{
BOOST_TEST(PairMosaicSelection({{1, 1}}, 0.5).materialise(4) == IndexPairs({{1, 1}, {1, 1}}));
BOOST_TEST(PairMosaicSelection({{1, 1}}, 1.0).materialise(4) == IndexPairs({{1, 1}, {1, 1}, {1, 1}, {1, 1}}));
BOOST_TEST(PairMosaicSelection({{1, 1}}, 2.0).materialise(4) == IndexPairs({{1, 1}, {1, 1}, {1, 1}, {1, 1}, {1, 1}, {1, 1}, {1, 1}, {1, 1}}));
BOOST_TEST(PairMosaicSelection({{1, 1}}, 1.0).materialise(2) == IndexPairs({{1, 1}, {1, 1}}));
IndexPairs pairs1{{0, 1}, {1, 0}};
BOOST_TEST(PairMosaicSelection(pairs1, 0.5).materialise(4) == IndexPairs({{0, 1}, {1, 0}}));
BOOST_TEST(PairMosaicSelection(pairs1, 1.0).materialise(4) == IndexPairs({{0, 1}, {1, 0}, {0, 1}, {1, 0}}));
BOOST_TEST(PairMosaicSelection(pairs1, 2.0).materialise(4) == IndexPairs({{0, 1}, {1, 0}, {0, 1}, {1, 0}, {0, 1}, {1, 0}, {0, 1}, {1, 0}}));
BOOST_TEST(PairMosaicSelection(pairs1, 1.0).materialise(2) == IndexPairs({{0, 1}, {1, 0}}));
IndexPairs pairs2{{3, 2}, {2, 3}, {1, 0}, {1, 1}};
BOOST_TEST(PairMosaicSelection(pairs2, 0.5).materialise(4) == IndexPairs({{3, 2}, {2, 3}}));
BOOST_TEST(PairMosaicSelection(pairs2, 1.0).materialise(4) == IndexPairs({{3, 2}, {2, 3}, {1, 0}, {1, 1}}));
BOOST_TEST(PairMosaicSelection(pairs2, 2.0).materialise(4) == IndexPairs({{3, 2}, {2, 3}, {1, 0}, {1, 1}, {3, 2}, {2, 3}, {1, 0}, {1, 1}}));
IndexPairs pairs3{{1, 0}, {1, 1}, {1, 0}, {1, 1}};
BOOST_TEST(PairMosaicSelection(pairs3, 1.0).materialise(2) == IndexPairs({{1, 0}, {1, 1}}));
}
BOOST_AUTO_TEST_CASE(materialise_should_round_indices)
{
IndexPairs pairs{{4, 4}, {3, 3}, {2, 2}, {1, 1}, {0, 0}};
BOOST_TEST(PairMosaicSelection(pairs, 0.49).materialise(5) == IndexPairs({{4, 4}, {3, 3}}));
BOOST_TEST(PairMosaicSelection(pairs, 0.50).materialise(5) == IndexPairs({{4, 4}, {3, 3}, {2, 2}}));
BOOST_TEST(PairMosaicSelection(pairs, 0.51).materialise(5) == IndexPairs({{4, 4}, {3, 3}, {2, 2}}));
}
BOOST_AUTO_TEST_CASE(materialise_should_return_no_pairs_if_collection_is_empty)
{
BOOST_TEST(PairMosaicSelection({{1, 1}}, 1.0).materialise(0).empty());
BOOST_TEST(PairMosaicSelection({{1, 1}, {3, 3}}, 2.0).materialise(0).empty());
BOOST_TEST(PairMosaicSelection({{5, 5}, {4, 4}, {3, 3}, {2, 2}}, 0.5).materialise(0).empty());
}
BOOST_AUTO_TEST_CASE(materialise_should_return_no_pairs_if_collection_has_one_element)
{
IndexPairs pairs{{4, 4}, {3, 3}, {2, 2}, {1, 1}, {0, 0}};
BOOST_TEST(PairMosaicSelection(pairs, 0.0).materialise(1).empty());
BOOST_TEST(PairMosaicSelection(pairs, 0.5).materialise(1).empty());
BOOST_TEST(PairMosaicSelection(pairs, 1.0).materialise(1).empty());
BOOST_TEST(PairMosaicSelection(pairs, 7.0).materialise(1).empty());
}
BOOST_AUTO_TEST_CASE(materialise_should_clamp_indices_at_collection_size)
{
IndexPairs pairs{{4, 4}, {3, 3}, {2, 2}, {1, 1}, {0, 0}};
BOOST_TEST(PairMosaicSelection(pairs, 1.0).materialise(4) == IndexPairs({{3, 3}, {3, 3}, {2, 2}, {1, 1}}));
BOOST_TEST(PairMosaicSelection(pairs, 2.0).materialise(3) == IndexPairs({{2, 2}, {2, 2}, {2, 2}, {1, 1}, {0, 0}, {2, 2}}));
}
BOOST_AUTO_TEST_SUITE_END()
BOOST_AUTO_TEST_SUITE_END()
BOOST_AUTO_TEST_SUITE_END()
}

View File

@ -18,6 +18,7 @@
#include <test/yulPhaser/Common.h>
#include <tools/yulPhaser/Chromosome.h>
#include <tools/yulPhaser/PairSelections.h>
#include <tools/yulPhaser/Population.h>
#include <tools/yulPhaser/Program.h>
#include <tools/yulPhaser/Selections.h>
@ -31,6 +32,7 @@
#include <boost/test/unit_test.hpp>
#include <cmath>
#include <optional>
#include <string>
#include <sstream>
@ -226,6 +228,70 @@ BOOST_FIXTURE_TEST_CASE(select_should_return_empty_population_if_selection_is_em
BOOST_TEST(population.select(selection).individuals().empty());
}
BOOST_FIXTURE_TEST_CASE(mutate_should_return_population_containing_individuals_indicated_by_selection_with_mutation_applied, PopulationFixture)
{
Population population(m_fitnessMetric, {Chromosome("aa"), Chromosome("cc"), Chromosome("gg"), Chromosome("hh")});
RangeSelection selection(0.25, 0.75);
assert(selection.materialise(population.individuals().size()) == (vector<size_t>{1, 2}));
Population expectedPopulation(m_fitnessMetric, {Chromosome("fc"), Chromosome("fg")});
BOOST_TEST(population.mutate(selection, geneSubstitution(0, BlockFlattener::name)) == expectedPopulation);
}
BOOST_FIXTURE_TEST_CASE(mutate_should_include_duplicates_if_selection_contains_duplicates, PopulationFixture)
{
Population population(m_fitnessMetric, {Chromosome("aa"), Chromosome("aa")});
RangeSelection selection(0.0, 1.0);
assert(selection.materialise(population.individuals().size()) == (vector<size_t>{0, 1}));
BOOST_TEST(
population.mutate(selection, geneSubstitution(0, BlockFlattener::name)) ==
Population(m_fitnessMetric, {Chromosome("fa"), Chromosome("fa")})
);
}
BOOST_FIXTURE_TEST_CASE(mutate_should_return_empty_population_if_selection_is_empty, PopulationFixture)
{
Population population(m_fitnessMetric, {Chromosome("aa"), Chromosome("cc")});
RangeSelection selection(0.0, 0.0);
assert(selection.materialise(population.individuals().size()).empty());
BOOST_TEST(population.mutate(selection, geneSubstitution(0, BlockFlattener::name)).individuals().empty());
}
BOOST_FIXTURE_TEST_CASE(crossover_should_return_population_containing_individuals_indicated_by_selection_with_crossover_applied, PopulationFixture)
{
Population population(m_fitnessMetric, {Chromosome("aa"), Chromosome("cc"), Chromosome("gg"), Chromosome("hh")});
PairMosaicSelection selection({{0, 1}, {2, 1}}, 1.0);
assert(selection.materialise(population.individuals().size()) == (vector<tuple<size_t, size_t>>{{0, 1}, {2, 1}, {0, 1}, {2, 1}}));
Population expectedPopulation(m_fitnessMetric, {Chromosome("ac"), Chromosome("ac"), Chromosome("gc"), Chromosome("gc")});
BOOST_TEST(population.crossover(selection, fixedPointCrossover(0.5)) == expectedPopulation);
}
BOOST_FIXTURE_TEST_CASE(crossover_should_include_duplicates_if_selection_contains_duplicates, PopulationFixture)
{
Population population(m_fitnessMetric, {Chromosome("aa"), Chromosome("aa")});
PairMosaicSelection selection({{0, 0}, {1, 1}}, 2.0);
assert(selection.materialise(population.individuals().size()) == (vector<tuple<size_t, size_t>>{{0, 0}, {1, 1}, {0, 0}, {1, 1}}));
BOOST_TEST(
population.crossover(selection, fixedPointCrossover(0.5)) ==
Population(m_fitnessMetric, {Chromosome("aa"), Chromosome("aa"), Chromosome("aa"), Chromosome("aa")})
);
}
BOOST_FIXTURE_TEST_CASE(crossover_should_return_empty_population_if_selection_is_empty, PopulationFixture)
{
Population population(m_fitnessMetric, {Chromosome("aa"), Chromosome("cc")});
PairMosaicSelection selection({}, 0.0);
assert(selection.materialise(population.individuals().size()).empty());
BOOST_TEST(population.crossover(selection, fixedPointCrossover(0.5)).individuals().empty());
}
BOOST_AUTO_TEST_SUITE_END()
BOOST_AUTO_TEST_SUITE_END()

View File

@ -23,6 +23,10 @@ add_executable(yul-phaser
yulPhaser/FitnessMetrics.cpp
yulPhaser/Chromosome.h
yulPhaser/Chromosome.cpp
yulPhaser/Mutations.h
yulPhaser/Mutations.cpp
yulPhaser/PairSelections.h
yulPhaser/PairSelections.cpp
yulPhaser/Selections.h
yulPhaser/Selections.cpp
yulPhaser/Program.h

View File

@ -16,7 +16,9 @@
*/
#include <tools/yulPhaser/GeneticAlgorithms.h>
#include <tools/yulPhaser/Mutations.h>
#include <tools/yulPhaser/Selections.h>
#include <tools/yulPhaser/PairSelections.h>
using namespace std;
using namespace solidity::phaser;
@ -48,3 +50,28 @@ void RandomAlgorithm::runNextRound()
m_options.maxChromosomeLength
);
}
void GenerationalElitistWithExclusivePools::runNextRound()
{
double elitePoolSize = 1.0 - (m_options.mutationPoolSize + m_options.crossoverPoolSize);
RangeSelection elite(0.0, elitePoolSize);
m_population =
m_population.select(elite) +
m_population.select(elite).mutate(
RandomSelection(m_options.mutationPoolSize / elitePoolSize),
alternativeMutations(
m_options.randomisationChance,
geneRandomisation(m_options.percentGenesToRandomise),
alternativeMutations(
m_options.deletionVsAdditionChance,
geneDeletion(m_options.percentGenesToAddOrDelete),
geneAddition(m_options.percentGenesToAddOrDelete)
)
)
) +
m_population.select(elite).crossover(
RandomPairSelection(m_options.crossoverPoolSize / elitePoolSize),
randomPointCrossover()
);
}

View File

@ -112,4 +112,57 @@ private:
Options m_options;
};
/**
* A generational, elitist genetic algorithm that replaces the population by mutating and crossing
* over chromosomes from the elite.
*
* The elite consists of individuals not included in the crossover and mutation pools.
* The crossover operator used is @a randomPointCrossover. The mutation operator is randomly chosen
* from three possibilities: @a geneRandomisation, @a geneDeletion or @a geneAddition (with
* configurable probabilities). Each mutation also has a parameter determining the chance of a gene
* being affected by it.
*/
class GenerationalElitistWithExclusivePools: public GeneticAlgorithm
{
public:
struct Options
{
double mutationPoolSize; ///< Percentage of population to regenerate using mutations in each round.
double crossoverPoolSize; ///< Percentage of population to regenerate using crossover in each round.
double randomisationChance; ///< The chance of choosing @a geneRandomisation as the mutation to perform
double deletionVsAdditionChance; ///< The chance of choosing @a geneDeletion as the mutation if randomisation was not chosen.
double percentGenesToRandomise; ///< The chance of any given gene being mutated in gene randomisation.
double percentGenesToAddOrDelete; ///< The chance of a gene being added (or deleted) in gene addition (or deletion).
bool isValid() const
{
return (
0 <= mutationPoolSize && mutationPoolSize <= 1.0 &&
0 <= crossoverPoolSize && crossoverPoolSize <= 1.0 &&
0 <= randomisationChance && randomisationChance <= 1.0 &&
0 <= deletionVsAdditionChance && deletionVsAdditionChance <= 1.0 &&
0 <= percentGenesToRandomise && percentGenesToRandomise <= 1.0 &&
0 <= percentGenesToAddOrDelete && percentGenesToAddOrDelete <= 1.0 &&
mutationPoolSize + crossoverPoolSize <= 1.0
);
}
};
GenerationalElitistWithExclusivePools(
Population _initialPopulation,
std::ostream& _outputStream,
Options const& _options
):
GeneticAlgorithm(_initialPopulation, _outputStream),
m_options(_options)
{
assert(_options.isValid());
}
void runNextRound() override;
private:
Options m_options;
};
}

View File

@ -0,0 +1,147 @@
/*
This file is part of solidity.
solidity is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
solidity is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with solidity. If not, see <http://www.gnu.org/licenses/>.
*/
#include <tools/yulPhaser/Mutations.h>
#include <tools/yulPhaser/SimulationRNG.h>
#include <libsolutil/CommonData.h>
#include <algorithm>
#include <cassert>
#include <cmath>
#include <string>
#include <vector>
using namespace std;
using namespace solidity;
using namespace solidity::phaser;
function<Mutation> phaser::geneRandomisation(double _chance)
{
return [=](Chromosome const& _chromosome)
{
vector<string> optimisationSteps;
for (auto const& step: _chromosome.optimisationSteps())
optimisationSteps.push_back(
SimulationRNG::bernoulliTrial(_chance) ?
Chromosome::randomOptimisationStep() :
step
);
return Chromosome(move(optimisationSteps));
};
}
function<Mutation> phaser::geneDeletion(double _chance)
{
return [=](Chromosome const& _chromosome)
{
vector<string> optimisationSteps;
for (auto const& step: _chromosome.optimisationSteps())
if (!SimulationRNG::bernoulliTrial(_chance))
optimisationSteps.push_back(step);
return Chromosome(move(optimisationSteps));
};
}
function<Mutation> phaser::geneAddition(double _chance)
{
return [=](Chromosome const& _chromosome)
{
vector<string> optimisationSteps;
if (SimulationRNG::bernoulliTrial(_chance))
optimisationSteps.push_back(Chromosome::randomOptimisationStep());
for (auto const& step: _chromosome.optimisationSteps())
{
optimisationSteps.push_back(step);
if (SimulationRNG::bernoulliTrial(_chance))
optimisationSteps.push_back(Chromosome::randomOptimisationStep());
}
return Chromosome(move(optimisationSteps));
};
}
function<Mutation> phaser::alternativeMutations(
double _firstMutationChance,
function<Mutation> _mutation1,
function<Mutation> _mutation2
)
{
return [=](Chromosome const& _chromosome)
{
if (SimulationRNG::bernoulliTrial(_firstMutationChance))
return _mutation1(_chromosome);
else
return _mutation2(_chromosome);
};
}
namespace
{
Chromosome buildChromosomesBySwappingParts(
Chromosome const& _chromosome1,
Chromosome const& _chromosome2,
size_t _crossoverPoint
)
{
assert(_crossoverPoint <= _chromosome1.length());
assert(_crossoverPoint <= _chromosome2.length());
auto begin1 = _chromosome1.optimisationSteps().begin();
auto begin2 = _chromosome2.optimisationSteps().begin();
return Chromosome(
vector<string>(begin1, begin1 + _crossoverPoint) +
vector<string>(begin2 + _crossoverPoint, _chromosome2.optimisationSteps().end())
);
}
}
function<Crossover> phaser::randomPointCrossover()
{
return [=](Chromosome const& _chromosome1, Chromosome const& _chromosome2)
{
size_t minLength = min(_chromosome1.length(), _chromosome2.length());
// Don't use position 0 (because this just swaps the values) unless it's the only choice.
size_t minPoint = (minLength > 0? 1 : 0);
assert(minPoint <= minLength);
size_t randomPoint = SimulationRNG::uniformInt(minPoint, minLength);
return buildChromosomesBySwappingParts(_chromosome1, _chromosome2, randomPoint);
};
}
function<Crossover> phaser::fixedPointCrossover(double _crossoverPoint)
{
assert(0.0 <= _crossoverPoint && _crossoverPoint <= 1.0);
return [=](Chromosome const& _chromosome1, Chromosome const& _chromosome2)
{
size_t minLength = min(_chromosome1.length(), _chromosome2.length());
size_t concretePoint = static_cast<size_t>(round(minLength * _crossoverPoint));
return buildChromosomesBySwappingParts(_chromosome1, _chromosome2, concretePoint);
};
}

View File

@ -0,0 +1,73 @@
/*
This file is part of solidity.
solidity is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
solidity is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with solidity. If not, see <http://www.gnu.org/licenses/>.
*/
/**
* Mutation and crossover operators for use in genetic algorithms.
*/
#pragma once
#include <tools/yulPhaser/Chromosome.h>
#include <functional>
#include <tuple>
namespace solidity::phaser
{
using Mutation = Chromosome(Chromosome const&);
using Crossover = Chromosome(Chromosome const&, Chromosome const&);
// MUTATIONS
/// Creates a mutation operator that iterates over all genes in a chromosome and with probability
/// @a _chance replaces a gene with a random one (which could also be the same as the original).
std::function<Mutation> geneRandomisation(double _chance);
/// Creates a mutation operator that iterates over all genes in a chromosome and with probability
/// @a _chance deletes it.
std::function<Mutation> geneDeletion(double _chance);
/// Creates a mutation operator that iterates over all positions in a chromosome (including spots
/// at the beginning and at the end of the sequence) and with probability @a _chance insert a new,
/// randomly chosen gene.
std::function<Mutation> geneAddition(double _chance);
/// Creates a mutation operator that always applies one of the mutations passed to it.
/// The probability that the chosen mutation is the first one is @a _firstMutationChance.
/// randomly chosen gene.
std::function<Mutation> alternativeMutations(
double _firstMutationChance,
std::function<Mutation> _mutation1,
std::function<Mutation> _mutation2
);
// CROSSOVER
/// Creates a crossover operator that randomly selects a number between 0 and 1 and uses it as the
/// position at which to perform perform @a fixedPointCrossover.
std::function<Crossover> randomPointCrossover();
/// Creates a crossover operator that always chooses a point that lies at @a _crossoverPoint
/// percent of the length of the shorter chromosome. Then creates a new chromosome by
/// splitting both inputs at the crossover point and stitching output from the first half or first
/// input and the second half of the second input.
///
/// Avoids selecting position 0 (since this just produces a chromosome identical to the second one)
/// unless there is no other choice (i.e. one of the chromosomes is empty).
std::function<Crossover> fixedPointCrossover(double _crossoverPoint);
}

View File

@ -0,0 +1,65 @@
/*
This file is part of solidity.
solidity is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
solidity is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with solidity. If not, see <http://www.gnu.org/licenses/>.
*/
#include <tools/yulPhaser/PairSelections.h>
#include <tools/yulPhaser/SimulationRNG.h>
#include <cmath>
using namespace std;
using namespace solidity::phaser;
vector<tuple<size_t, size_t>> RandomPairSelection::materialise(size_t _poolSize) const
{
if (_poolSize < 2)
return {};
size_t count = static_cast<size_t>(round(_poolSize * m_selectionSize));
vector<tuple<size_t, size_t>> selection;
for (size_t i = 0; i < count; ++i)
{
size_t index1 = SimulationRNG::uniformInt(0, _poolSize - 1);
size_t index2;
do
{
index2 = SimulationRNG::uniformInt(0, _poolSize - 1);
} while (index1 == index2);
selection.push_back({index1, index2});
}
return selection;
}
vector<tuple<size_t, size_t>> PairMosaicSelection::materialise(size_t _poolSize) const
{
if (_poolSize < 2)
return {};
size_t count = static_cast<size_t>(round(_poolSize * m_selectionSize));
vector<tuple<size_t, size_t>> selection;
for (size_t i = 0; i < count; ++i)
{
tuple<size_t, size_t> pair = m_pattern[i % m_pattern.size()];
selection.push_back({min(get<0>(pair), _poolSize - 1), min(get<1>(pair), _poolSize - 1)});
}
return selection;
}

View File

@ -0,0 +1,99 @@
/*
This file is part of solidity.
solidity is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
solidity is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with solidity. If not, see <http://www.gnu.org/licenses/>.
*/
/**
* Contains an abstract base class representing a selection of pairs of elements from a collection
* and its concrete implementations.
*/
#pragma once
#include <cassert>
#include <tuple>
#include <vector>
namespace solidity::phaser
{
/**
* Abstract base class for selections of pairs elements from a collection.
*
* An instance of this class represents a specific method of selecting a set of pairs of elements
* from containers of arbitrary sizes. The selected pairs always point at a subset of the elements
* from the container but may indicate the same element more than once. The pairs themselves can
* repeat too. The selection may or may not be fixed - it's up to a specific implementation whether
* subsequent calls for the same container produce the same indices or not.
*
* Derived classes are meant to override the @a materialise() method.
* This method is expected to produce pairs of selected elements given the size of the collection.
*/
class PairSelection
{
public:
PairSelection() = default;
PairSelection(PairSelection const&) = delete;
PairSelection& operator=(PairSelection const&) = delete;
virtual ~PairSelection() = default;
virtual std::vector<std::tuple<size_t, size_t>> materialise(size_t _poolSize) const = 0;
};
/**
* A selection that selects pairs of random elements from a container. The resulting set of pairs
* may contain the same pair more than once but does not contain pairs of duplicates. Always
* selects as many pairs as the size of the container multiplied by @a _selectionSize (unless the
* container is empty).
*/
class RandomPairSelection: public PairSelection
{
public:
explicit RandomPairSelection(double _selectionSize):
m_selectionSize(_selectionSize) {}
std::vector<std::tuple<size_t, size_t>> materialise(size_t _poolSize) const override;
private:
double m_selectionSize;
};
/**
* A selection that selects pairs of elements at specific, fixed positions indicated by a repeating
* "pattern". If the positions in the pattern exceed the size of the container, they are capped at
* the maximum available position. Always selects as many pairs as the size of the container
* multiplied by @a _selectionSize (unless the container is empty).
*
* E.g. if the pattern is {{0, 1}, {3, 9}} and collection size is 5, the selection will materialise
* into {{0, 1}, {3, 4}, {0, 1}, {3, 4}, {0, 1}}. If the size is 3, it will be
* {{0, 1}, {2, 2}, {0, 1}}.
*/
class PairMosaicSelection: public PairSelection
{
public:
explicit PairMosaicSelection(std::vector<std::tuple<size_t, size_t>> _pattern, double _selectionSize = 1.0):
m_pattern(move(_pattern)),
m_selectionSize(_selectionSize)
{
assert(m_pattern.size() > 0 || _selectionSize == 0.0);
}
std::vector<std::tuple<size_t, size_t>> materialise(size_t _poolSize) const override;
private:
std::vector<std::tuple<size_t, size_t>> m_pattern;
double m_selectionSize;
};
}

View File

@ -17,6 +17,7 @@
#include <tools/yulPhaser/Population.h>
#include <tools/yulPhaser/PairSelections.h>
#include <tools/yulPhaser/Selections.h>
#include <libsolutil/CommonData.h>
@ -93,6 +94,30 @@ Population Population::select(Selection const& _selection) const
return Population(m_fitnessMetric, selectedIndividuals);
}
Population Population::mutate(Selection const& _selection, function<Mutation> _mutation) const
{
vector<Individual> mutatedIndividuals;
for (size_t i: _selection.materialise(m_individuals.size()))
mutatedIndividuals.emplace_back(_mutation(m_individuals[i].chromosome), *m_fitnessMetric);
return Population(m_fitnessMetric, mutatedIndividuals);
}
Population Population::crossover(PairSelection const& _selection, function<Crossover> _crossover) const
{
vector<Individual> crossedIndividuals;
for (auto const& [i, j]: _selection.materialise(m_individuals.size()))
{
auto childChromosome = _crossover(
m_individuals[i].chromosome,
m_individuals[j].chromosome
);
crossedIndividuals.emplace_back(move(childChromosome), *m_fitnessMetric);
}
return Population(m_fitnessMetric, crossedIndividuals);
}
Population operator+(Population _a, Population _b)
{
// This operator is meant to be used only with populations sharing the same metric (and, to make

View File

@ -19,6 +19,7 @@
#include <tools/yulPhaser/Chromosome.h>
#include <tools/yulPhaser/FitnessMetrics.h>
#include <tools/yulPhaser/Mutations.h>
#include <tools/yulPhaser/SimulationRNG.h>
#include <optional>
@ -39,6 +40,7 @@ solidity::phaser::Population operator+(solidity::phaser::Population _a, solidity
namespace solidity::phaser
{
class PairSelection;
class Selection;
/**
@ -104,6 +106,8 @@ public:
);
Population select(Selection const& _selection) const;
Population mutate(Selection const& _selection, std::function<Mutation> _mutation) const;
Population crossover(PairSelection const& _selection, std::function<Crossover> _crossover) const;
friend Population (::operator+)(Population _a, Population _b);
std::shared_ptr<FitnessMetric const> fitnessMetric() const { return m_fitnessMetric; }

View File

@ -39,6 +39,39 @@ using namespace solidity::util;
namespace po = boost::program_options;
enum class Algorithm
{
Random,
GEWEP
};
istream& operator>>(istream& inputStream, Algorithm& algorithm)
{
string value;
inputStream >> value;
if (value == "random")
algorithm = Algorithm::Random;
else if (value == "GEWEP")
algorithm = Algorithm::GEWEP;
else
inputStream.setstate(ios_base::failbit);
return inputStream;
}
ostream& operator<<(ostream& outputStream, Algorithm algorithm)
{
if (algorithm == Algorithm::Random)
outputStream << "random";
else if (algorithm == Algorithm::GEWEP)
outputStream << "GEWEP";
else
outputStream.setstate(ios_base::failbit);
return outputStream;
}
namespace
{
@ -49,7 +82,7 @@ struct CommandLineParsingResult
};
void initializeRNG(po::variables_map const& arguments)
void initialiseRNG(po::variables_map const& arguments)
{
uint32_t seed;
if (arguments.count("seed") > 0)
@ -69,7 +102,7 @@ CharStream loadSource(string const& _sourcePath)
return CharStream(sourceCode, _sourcePath);
}
void runAlgorithm(string const& _sourcePath)
void runAlgorithm(string const& _sourcePath, Algorithm _algorithm)
{
constexpr size_t populationSize = 20;
constexpr size_t minChromosomeLength = 12;
@ -83,15 +116,41 @@ void runAlgorithm(string const& _sourcePath)
minChromosomeLength,
maxChromosomeLength
);
RandomAlgorithm(
population,
cout,
switch (_algorithm)
{
case Algorithm::Random:
{
/* elitePoolSize = */ 1.0 / populationSize,
/* minChromosomeLength = */ minChromosomeLength,
/* maxChromosomeLength = */ maxChromosomeLength,
RandomAlgorithm(
population,
cout,
{
/* elitePoolSize = */ 1.0 / populationSize,
/* minChromosomeLength = */ minChromosomeLength,
/* maxChromosomeLength = */ maxChromosomeLength,
}
).run();
break;
}
).run();
case Algorithm::GEWEP:
{
GenerationalElitistWithExclusivePools(
population,
cout,
{
/* mutationPoolSize = */ 0.25,
/* crossoverPoolSize = */ 0.25,
/* randomisationChance = */ 0.9,
/* deletionVsAdditionChance = */ 0.5,
/* percentGenesToRandomise = */ 1.0 / maxChromosomeLength,
/* percentGenesToAddOrDelete = */ 1.0 / maxChromosomeLength,
}
).run();
break;
}
}
}
CommandLineParsingResult parseCommandLine(int argc, char** argv)
@ -114,6 +173,11 @@ CommandLineParsingResult parseCommandLine(int argc, char** argv)
("help", "Show help message and exit.")
("input-file", po::value<string>()->required(), "Input file")
("seed", po::value<uint32_t>(), "Seed for the random number generator")
(
"algorithm",
po::value<Algorithm>()->default_value(Algorithm::GEWEP),
"Algorithm"
)
;
po::positional_options_description positionalDescription;
@ -156,11 +220,14 @@ int main(int argc, char** argv)
if (parsingResult.exitCode != 0)
return parsingResult.exitCode;
initializeRNG(parsingResult.arguments);
initialiseRNG(parsingResult.arguments);
try
{
runAlgorithm(parsingResult.arguments["input-file"].as<string>());
runAlgorithm(
parsingResult.arguments["input-file"].as<string>(),
parsingResult.arguments["algorithm"].as<Algorithm>()
);
}
catch (InvalidProgram const& _exception)
{