genetic-algorithm.h

/*
  MIT License

  Copyright (c) 2017 Coen Valk
  
  Permission is hereby granted, free of charge, to any person obtaining a copy
  of this software and associated documentation files (the "Software"), to deal
  in the Software without restriction, including without limitation the rights
  to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
  copies of the Software, and to permit persons to whom the Software is
  furnished to do so, subject to the following conditions:
  
  The above copyright notice and this permission notice shall be included in all
  copies or substantial portions of the Software.
  
  THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
  OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
  SOFTWARE.
*/

#ifndef __genetic_algorithm_h__
#define __genetic_algorithm_h__

#include <vector>
#include <iostream>
#include <cstdlib>
#include <limits.h>
#include <cmath>

template <class T>
class genetic_algorithm {
 public:

  // default constructor
  // CANNOT PERFORM EVOLUTION WITHOUT FUNCTIONS.
  // SIMPLY CREATING THE DEFAULT CONSTRUCTOR AND
  // RUNNING GENERATION WILL THROW ERROR.
  genetic_algorithm() {
    population_ = std::vector<T>();
    fitnesses_ = std::vector<T>();
    mutate_ = NULL;
    fitness_ = NULL;
    random_candidate_ = NULL;
    crossover_ = NULL;
    C_ = 0.5;
    M_ = 0.5;
    generation_ = 0;
  }
  
  /*
    constructor:
    needs population, a function to make a
    random candidate, crossover function,
    fitness function, and mutate function.
   */
 genetic_algorithm(int population,
		   float C,
		   float M,
		   T (*random_candidate)(),
		   std::vector<T> (*crossover)(const T& p1, const T& p2),
		   int (*fitness)(const T& candidate),
		   void (*mutate)(T& candidate)) :
  crossover_(crossover), fitness_(fitness), mutate_(mutate), generation_(0), M_(M), C_(C), random_candidate_(random_candidate) {
    population_ = std::vector<T>(population);
    fitnesses_ = std::vector<int>(population);

    if (random_candidate_ == NULL) {
      std::cerr << "ERROR: random canddiate function cannot be NULL!" << std::endl;
      throw 4;
    }
    
    for (int i = 0; i < population; i++) {
      population_[i] = (*random_candidate_)();
    }
    sort_pop();
  }

  void reset_generations() {
    if (random_candidate_ == NULL) {
      std::cerr << "ERROR: random candidate function cannot be NULL!" << std::endl;
      throw 4;
    }
    
    int p = population_.size();
    for (int i = 0; i < p; i++) {
      population_[i] = (*random_candidate_)();
    }
    generation_ = 0;
    sort_pop();
  }
  
  int get_generation_count() const { return generation_; }
  float get_crossover() const { return C_; }
  float get_mutation() const { return M_; }
  float avg_fitness() const {
    float avg = 0;
    for (int i = 0; i < fitnesses_.size(); i++) {
      avg += (float) fitnesses_[i];
    }
    avg /= fitnesses_.size();
    return avg;
  }

  void set_crossover(float new_C) { C_ = new_C; }
  void set_mutation(float new_M) { M_ = new_M; }
  void set_mutate_function(void (*mutate)(T& candidate)) { mutate_ = mutate; }
  void set_fitness_function(int (*fitness)(const T& candidate)) { fitness_ = fitness; }
  void set_crossover_function(std::vector<T> (*crossover)(const T& p1, const T& p2)) {
    crossover_ = crossover;
  }
  void set_random_function(T (*random_candidate)()) { random_candidate_ = random_candidate; }
  void set_pop(int P) {
    population_ = std::vector<T>(P);
    reset_generations();
  }
  
  // finds the best candidate
  const T& best_candidate() {
    return population_[0];
  }
  
  /*
    performs one generation. calculates fitness
    of all candidates, takes the top population,
    crosses some of them over, mutates some, and
    makes that the new population

    NOTE: THROWS ERROR IF ALL FUNCTIONS ARE NOT ACCOUNTED FOR!
  */
  void do_generation() {
    if (mutate_ == NULL) {
      std::cerr << "ERROR: mutation function cannot be NULL!!" << std::endl;
      throw 1;
    }
    if (crossover_ == NULL) {
      std::cerr << "ERROR: crossover function cannot be NULL!" << std::endl;
      throw 2;
    }
    
    if (population_.size() == 0) {
      std::cerr << "ERROR: population can't be zero!" << std::endl;
      throw 3;
    }
    
    int n = (int) ceil(sqrt(2 * (population_.size() + 1)) + 1);
    int x = 0; // index of population
    std::vector<T> new_population(population_.size());
    for (int i = 0; i < n; i++) {
      for(int j = i + 1; j < n; j++) {
	if (i < population_.size() && j < population_.size()) {
	  T one = population_[i];
	  T two = population_[j];
	  if ((float) rand() < C_ * (float) INT_MAX) {
	    // crossover!
	    std::vector<T> children = (*crossover_)(one, two);
	    one = children[0];
	    two = children[1];
	  }
	  if ((float) rand() < M_ * (float) INT_MAX) {
	    // mutate one!
	    (*mutate_)(one);
	  }
	  if ((float) rand() < M_ * (float) INT_MAX) {
	    // mutate two!
	    (*mutate_)(two);
	  }
	  if (x < new_population.size()) {
	    new_population[x] = one;
	    x++;
	  }
	  if (x < new_population.size()) {
	    new_population[x] = two;
	    x++;
	  }
	}
      }
    }
    population_ = new_population;
    sort_pop();
    generation_++;
  }

  
  
 private:
  int generation_;
  std::vector<T> population_;
  std::vector<int> fitnesses_;
  float C_;
  float M_;

  /*
    necessary functions for genetic algorithms
    must be implemented by the user and passed
    in with the constructor.
  */
  std::vector<T> (*crossover_)(const T& p1, const T& p2);
  int (*fitness_)(const T& candidate);
  void (*mutate_)(T& candidate);
  T (*random_candidate_)();
  
  void compute_fitnesses() {
    if (fitness_ == NULL) {
      std::cerr << "ERROR: fitness function cannot be NULL!" << std::endl;
      throw 5;
    }

    fitnesses_ = std::vector<int>(population_.size());
    
    for (int i = 0; i < population_.size(); i++) {
      fitnesses_[i] = (*fitness_)(population_[i]);
      // std::cout << "fitness " << i << ": " << fitnesses_[i] << std::endl;
    }
  }

  // swaps two elements in both population and fitnesses
  void swap(int index1, int index2) {
    T tmp = population_[index1];
    population_[index1] = population_[index2];
    population_[index2] = tmp;

    int TMP = fitnesses_[index1];
    fitnesses_[index1] = fitnesses_[index2];
    fitnesses_[index2] = TMP;
  }

  int partition(int lo, int hi) {
    int pivot = fitnesses_[hi];
    int i = lo - 1;
    for (int j = lo; j < hi; j++) {
      if (fitnesses_[j] > pivot) {
	i++;
	swap(i, j);
      }
    }
    if (fitnesses_[hi] > fitnesses_[i + 1]) {
      swap(i + 1, hi);
    }
    return i + 1;
  }

  void quicksort(int lo, int hi) {
    if (lo < hi) {
      int p = partition(lo, hi);
      quicksort(lo, p - 1);
      quicksort(p + 1, hi);
    }
  }
  
  /*
    private population sorting function:
    selection sort based on fitness function
    largest fitness first
  */
  void sort_pop() {
    compute_fitnesses();
    quicksort(0, population_.size() - 1);
  }  
};

#endif