function [best_fit, params] = pga(pop_size, generations, p_ranges, num_xovers, mut_prob)
  
  % This function is meant to implement a simple genetic algorithm I
  % intend to make it in such a way that it is easy for it to turn
  % it into a parrellel program.
  
  pop = makepop(pop_size, p_ranges);
  fits = zeros(pop_size,1);
  changed_members = ones(pop_size,1);
  fits = popfit(pop, fits, changed_members);
  changed_members = zeros(pop_size,1);
  best_fit = 0;
  params = 0;
  for gen = 1 : generations,
    [pop, fits] = selection(pop, fits);
    [changed_members, pop] = mutation(pop, changed_members, mut_prob, p_ranges);
    [changed_members, pop] = crosover(pop, changed_members, num_xovers);
    fits = popfit(pop, fits, changed_members);
    changed_members = zeros(pop_size,1);
    for i = 1 : pop_size,
      if fits(i) > best_fit
        best_fit = fits(i);
        params = pop(i, :);
        disp([best_fit, params])
      end
    end
  end
      
  
function [newly_changed_members, new_pop] = mutation(old_pop, prev_changed_members, ...
    mut_prob, p_ranges)
  [pop_size, num_params] = size(old_pop);
  mutate = rand(pop_size, num_params) < mut_prob;
  newly_changed_members = any(mutate,2) | prev_changed_members;
  new_pop = rand(pop_size, num_params);
  for i = 1 : pop_size,
    for j = 1 : num_params,
      if mutate(i, j)
        new_pop(i, j) = new_pop(i, j) * (p_ranges(j,2) - p_ranges(j,1)); 
        new_pop(i, j) = new_pop(i, j) + p_ranges(j,1);
      else
        new_pop(i, j) = old_pop(i, j);
      end
    end
  end
  
  
function [newly_changed_members, new_pop] = crosover(old_pop, prev_changed_members, num_xovers)
  [pop_size, num_params] = size(old_pop);
  newly_changed_members = prev_changed_members;
  for i = 1 : num_xovers,
    parent_a = round(rand * (pop_size - 1) + 1);
    parent_b = round(rand * (pop_size - 1) + 1);
    cut_point = round(rand * (num_params - 2)) + 1;
    child_a = [old_pop(parent_a,1:cut_point) old_pop(parent_b,cut_point+1:num_params)];
    child_b = [old_pop(parent_b,1:cut_point) old_pop(parent_a,cut_point+1:num_params)];
    old_pop(parent_a, :) = child_a;
    old_pop(parent_b, :) = child_b;
    newly_changed_members(parent_a) = 1;
    newly_changed_members(parent_b) = 1;
  end;  
  new_pop = old_pop;

function [new_pop, new_fits] = selection(old_pop, old_fits)
  [pop_size, num_params] = size(old_pop);
  total_fit = sum(old_fits);
  prob = old_fits / total_fit;
  prob = cumsum(prob);
  rand_nums = sort(rand(pop_size, 1));
  fitIn=1;
  newIn=1;
  while newIn <= pop_size
    if(rand_nums(newIn) < prob(fitIn))
      new_pop(newIn, :) = old_pop(fitIn, :);
      new_fits(newIn, :) = old_fits(fitIn, :);
      newIn = newIn + 1;
    else
      fitIn = fitIn + 1;
    end
  end
  
function new_pop = makepop(pop_size, p_ranges)
  % make an initial population randomly
  num_params = size(p_ranges,1);
  new_pop = rand(pop_size, num_params);
  for i = 1 : pop_size,
    for j = 1 : num_params,
      new_pop(i, j) = new_pop(i, j) * (p_ranges(j,2) - p_ranges(j,1)); 
      new_pop(i, j) = new_pop(i, j) + p_ranges(j,1);
    end
  end  
  
function new_fits = popfit(pop, old_fits, changed_members)
  % stub for calculating fitnesses of the population
  % this is the part that must become parrellel
  global count
  num_changed = sum(changed_members);
  [pop_size, num_params] = size(pop);
  target = 5 * ones(1,num_params);
  new_fits = old_fits;
  for i = 1 : pop_size,
    if changed_members(i),
      new_fits(i) = 1 / norm(target - pop(i, :));
      count = count + 1;
    end
  end