Bloque B2 - Algoritmos genéticos-Vacío.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "![Aeropython logo](./static/aeropython_name_mini.png)\n",
    "###### Carlos Dorado Cárdenas\n",
    "###### Siro Moreno Martín"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Simplifica tu vida con sistemas complejos y algoritmos genéticos"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Bloque B2 - Algoritmos genéticos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      ":0: FutureWarning: IPython widgets are experimental and may change in the future.\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import random as random\n",
    "from IPython.html.widgets import interact\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "class Individual (object):\n",
    "    \n",
    "    def __init__(self, genome):\n",
    "        \n",
    "        self.genome = genome        \n",
    "        self.traits = {}\n",
    "        self.performances = {}\n",
    "        self.fitness = 0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "El genoma es donde está codificada la información que sirve para construir una solución. Mientras se sea coherente, el genoma se puede definir con total libertad según se necesite: números reales, enteros, letras... En el mundo informático todo está codificado en estados binarios, así que haremos que nuestro genoma será una cadena de ceros y unos. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Para ello vamos a utilizar un diccionario genético que defina los genes de nuestro individuo y su longitud. En este caso, nuestro individuo sólo presenta dos características, que son las coordenadas de un punto, por lo que sólo habrá dos genes. Vamos a utilizar 10 dígitos, que nos permiten codificar 1024 estados diferentes."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "dict_genes = {\n",
    "    'x1' : 10,\n",
    "    'x2' : 10,\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Vamos a necesitar crear individuos aleatorios, para lo que utilizaremos una función \"Generar genoma\":"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "generate_random_binary_list = lambda n: [random.randint(0,1) for b in range(1,n+1)]\n",
    "\n",
    "def generate_genome (dict_genes):\n",
    "    \n",
    "    #We first calculate the total number of bits that the genome must contain\n",
    "    number_of_bits = sum([dict_genes[trait] for trait in dict_genes])\n",
    "    \n",
    "    #And we return a random genome of this length\n",
    "    return generate_random_binary_list(number_of_bits)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Cuando ya tengamos nuestros individuos, con su material genético, necesitaremos una función que nos descifre este código y lo traduzca a características reconocibles del mismo. En el mundo real sería como leer el \"gen del color de los ojos\" (en realidad son varios genes) para descubrir que el contenido de ese gen en concreto indica que el individuo tiene \"ojos azules\"."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def calculate_traits (individual, dict_genes, minimum, maximum):\n",
    "    #This function must decipher the genome and return the traits of the individual.\n",
    "    #Normally, the genome contains binary numerical values for the different traits.\n",
    "    \n",
    "    dict_traits = {}\n",
    "    index = 0\n",
    "    \n",
    "    for trait in dict_genes:\n",
    "        step = (maximum-minimum) / (int(dict_genes[trait]*'1',2))\n",
    "        dict_traits[trait] = minimum + step*int(''.join(str(bit) for bit in individual.genome[index : index+dict_genes[trait]]), 2)\n",
    "        index += dict_genes[trait]\n",
    "        \n",
    "    individual.traits = dict_traits"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "En el mundo natural, las características de un individuo determinarán en gran parte cómo de bien o de mal se desenvolverá en la vida, esto es, sus \"actuaciones\", \"desempeño\" o, en inglés, su \"performance\". Si tiene las piernas más largas, podrá correr más rápido; si tiene unos dientes afilados, podrá defenderse mejor... En nuestro ejemplo esto no es ni más ni menos que el eje vertical de nuestra función de coste."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from codigo import goldstein_price"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def calculate_performances (individual):\n",
    "    \n",
    "    dict_perfo = {\n",
    "        \n",
    "        'cost' : goldstein_price([individual.traits['x1'], individual.traits['x2']])\n",
    "\n",
    "    }\n",
    "    \n",
    "    individual.performances = dict_perfo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Según sean de buenas sus actuaciones, el individuo será más apto, estará mejor adaptado. En nuestro problema se trata de **definir qué es una solución buena**, y aquí se haya una de las armas más potentes del algoritmo genético. **El algoritmo genético nos obliga a definir qué es lo que estamos buscando en nuestro proceso de optimización mediante una función.**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Para simplificareste ejemplo, vamos a utilizar directamente el valor del coste:\n",
    "$$ Aptitud = - Coste $$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def calculate_fitness (individual):\n",
    "        \n",
    "    individual.fitness = - individual.performances['cost']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ahora ya podemos generar nuestras soluciones y tenemos medios para evaluarlas. Es hora de hacer que la selección natural haga su trabajo."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Primero necesitamos crear una población. Una manera de hacer esto es mediante una función de inmigración, que genere una sociedad formada por una determinada cantidad de individuos exóticos, generados de manera aleatoria:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def immigration (society, target_population, dict_genes, minimum, maximum):\n",
    "    \n",
    "    while len(society) < target_population:\n",
    "        \n",
    "        new_individual = Individual (generate_genome (dict_genes))\n",
    "        calculate_traits (new_individual, dict_genes, minimum, maximum)\n",
    "        calculate_performances (new_individual)\n",
    "        calculate_fitness (new_individual)\n",
    "        \n",
    "        society.append (new_individual)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ahora, los individuos de esta sociedad se reproducen entre ellos, generando nuevos individuos. Los individuos más aptos tienen más probabilidades de reproducirse que los menos aptos, con lo que las mejores soluciones generan más descendencia. Durante la concepción, los nuevos individuos sufren pequeñas mutaciones aleatorias. Después nacen y se incorporan a la sociedad."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#This function was taken from Eli Bendersky's website\n",
    "#It returns an index of a list called \"weights\", \n",
    "#where the content of each element in \"weights\" is the probability of this index to be returned.\n",
    "#For this function to be as fast as possible we need to pass it a list of weights in descending order.\n",
    "def weighted_choice_sub(weights):\n",
    "    \"\"\"\n",
    "    weights = [0.9, 0.05, 0.05]\n",
    "    N = 100000\n",
    "    lista = [weighted_choice_sub(weights) for ii in range(N)]\n",
    "    print( lista.count(0)/N, lista.count(1)/N, lista.count(2)/N)\n",
    "    \"\"\"\n",
    "    rnd = random.random() * sum(weights)\n",
    "    for i, w in enumerate(weights):\n",
    "        rnd -= w\n",
    "        if rnd < 0:\n",
    "            return i\n",
    "\n",
    "def crossover (society, reproduction_rate, mutation_rate, minimum, maximum):\n",
    "    \n",
    "    #First we create a list with the fitness values of every individual in the society\n",
    "    fitness_list = [individual.fitness for individual in society]\n",
    "    \n",
    "    #We sort the individuals in the society in descending order of fitness.   \n",
    "    society_sorted = [x for (y, x) in sorted(zip(fitness_list, society), key=lambda x: x[0], reverse=True)] \n",
    "    \n",
    "    #We then create a list of relative probabilities in descending order, \n",
    "    #so that the fittest individual in the society has N times more chances to reproduce than the least fit,\n",
    "    #where N is the number of individuals in the society.\n",
    "    probability = [i for i in reversed(range(1,len(society_sorted)+1))]\n",
    "    \n",
    "    #We create a list of weights with the probabilities of non-mutation and mutation\n",
    "    mutation = [1 - mutation_rate, mutation_rate]    \n",
    "    \n",
    "    #For every new individual to be created through reproduction:\n",
    "    for i in range (int(len(society) * reproduction_rate)):\n",
    "        \n",
    "        #We select two parents randomly, using the list of probabilities in \"probability\".\n",
    "        father, mother = society_sorted[weighted_choice_sub(probability)], society_sorted[weighted_choice_sub(probability)]\n",
    "        \n",
    "        #We randomly select two cutting points for the genome.\n",
    "        a, b = random.randrange(0, len(father.genome)), random.randrange(0, len(father.genome))\n",
    "        \n",
    "        #And we create the genome of the child putting together the genome slices of the parents in the cutting points.\n",
    "        child_genome = father.genome[0:min(a,b)]+mother.genome[min(a,b):max(a,b)]+father.genome[max(a,b):]\n",
    "        \n",
    "        #For every bit in the not-yet-born child, we generate a list containing \n",
    "        #1's in the positions where the genome must mutate (i.e. the bit must switch its value)\n",
    "        #and 0's in the positions where the genome must stay the same.\n",
    "        n = [weighted_choice_sub(mutation) for ii in range(len(child_genome))]\n",
    "        \n",
    "        #This line switches the bits of the genome of the child that must mutate.\n",
    "        mutant_child_genome = [abs(n[i] -  child_genome[i]) for i in range(len(child_genome))]\n",
    "        \n",
    "        #We finally append the newborn individual to the society\n",
    "        newborn = Individual(mutant_child_genome)\n",
    "        calculate_traits (newborn, dict_genes, minimum, maximum)\n",
    "        calculate_performances (newborn)\n",
    "        calculate_fitness (newborn)\n",
    "        society.append(newborn)        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "En el mundo natural, la escasez de recursos obliga a los individuos a competir. Una función de torneo escoge parejas de soluciones al azar, las compara y elimina las menos aptas."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def tournament(society, target_population):\n",
    "    \n",
    "    while len(society) > target_population:\n",
    "        \n",
    "        #index1, index2 = random.randrange(0, len(society)), random.randrange(0, len(society))\n",
    "        #if society[index1].fitness > society[index2].fitness:\n",
    "        #    society.pop(index2)\n",
    "        #else:\n",
    "        #    society.pop(index1)\n",
    "                \n",
    "        fitness_list = [individual.fitness for individual in society]\n",
    "        society.pop(fitness_list.index(min(fitness_list)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false,
    "scrolled": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ahora podemos crear una función que aplique el algoritmo genético a este problema concreto:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def genetic_algorithm (minimum = -2,\n",
    "                       maximum = 2,\n",
    "                       number_of_genes = 20,\n",
    "                       size_of_society = 50, \n",
    "                       reproduction_rate = 0.4,\n",
    "                       mutation_rate = 0.1, \n",
    "                       number_of_generations = 10,\n",
    "                       size_after_tournament = 40,\n",
    "                      ):\n",
    "\n",
    "    \n",
    "    dict_genes = {'x1' : number_of_genes,\n",
    "                  'x2' : number_of_genes,\n",
    "                 }\n",
    "    \n",
    "            \n",
    "    society = []\n",
    "    fitness_max = []\n",
    "    fitness_avg = []\n",
    "\n",
    "    for generation in range(number_of_generations):\n",
    "    \n",
    "        immigration (society, size_of_society, dict_genes, minimum, maximum)\n",
    "        fitness_max += list([max([individual.fitness for individual in society])])\n",
    "        fitness_avg += list([np.mean([individual.fitness for individual in society])])\n",
    "        crossover(society, reproduction_rate, mutation_rate, minimum, maximum)\n",
    "        tournament (society, size_after_tournament)\n",
    "    \n",
    "    fitness_list = [individual.fitness for individual in society]\n",
    "    winner = society[fitness_list.index(max(fitness_list))]\n",
    "    print(winner.traits)\n",
    "    \n",
    "        \n",
    "    plt.figure(figsize=(15,5))\n",
    "    plt.plot(fitness_max)\n",
    "    plt.title('$\\Phi_{max} = $' + str(round(fitness_max[-1], 5)), fontsize = 15)  \n",
    "    \n",
    "    plt.figure(figsize=(15,5))\n",
    "    plt.plot(fitness_avg)\n",
    "    plt.title('$\\overline{\\Phi}$', fontsize = 15)  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'x1': -0.006483084185680665, 'x2': -0.9444112247574088}\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA20AAAFGCAYAAAD5BncWAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3Xu8rvWc//HXu53SFh30I2qbQoYwUsmptNHJIckhkhwj\nIiFDYshhhn4ODZlyqHRQGTmNHGJHOxpqpDLpMCPVpIyh6aBUOuzP/HHfsSxrr9a617rX9z68no/H\neqx1X/fheq/7UWvt97q+1+dKVSFJkiRJGkyrtA4gSZIkSVo5S5skSZIkDTBLmyRJkiQNMEubJEmS\nJA0wS5skSZIkDTBLmyRJkiQNMEubJEmSJA0wS5skSZIkDTBLmyRJ00iyQZIbk6xIsvguHvu8JD9M\ncnWSm5NcnOQdSe42xWOfneTfk9yS5NIkb5p0/25JvpHkV0luSHJ2khfO9/cnSRp8ljZJ0rxIsiTJ\nYd2isiLJj5J8PElaZ5ujDwE3ADWDx64LnAq8EtgJOAp4B/DRiQ9K8kTgy8CZwDO7jzs4yX4THvZG\n4FrgDcDOwGnACUleP5dvRpI0fFI1k99BkiTNTJInA98F1q2q61rnmYskTwK+AvwDnfK2ZlXdNMvX\neD/wuqpaZ8K2bwN3r6ptJ2z7MPByYP2qui3JulV1zaTXOh54fFU9sOdvSpI0dDzSJkmaN0nuBTyy\ne3NpkjVb5pmLJIuAQ4H3AP87h5e6Bpi8PPJRwLJJ25YB6wCPB5hc2LrOA+4/hyySpCFkaZMkzVmS\nVbtHig4Ebu5uvg74cJID2yWbk9fQKVv/NNsnJlmUZHGSrYF9gU9OesjdgVsnbbvz9kOneenHA/8x\n2zySpOG2ausAkqSRcCSwRlXtlmRpd9s5VbW8O0Dj7lX1rob5ZiXJvYH3AntU1R09nJb3e2C17tcn\nAG+ddP8lwGMmbduq+3ndlWR6KrALnSWUkqQxYmmTJM1JknWBPYGVDcg4nc4wjb6Wtu5SzI9x16tI\nLqiqD3efM/H3YFXVHd2v/x74UVWd0mOcxwGLgcfS+b4PB/aecP8ngU8m2Qv4Ep3Cduf0yBWTXyzJ\nRnTK31er6tgeM0mShpSlTZI0VzfRObJ0z5Xcfw/gN/0OUVU30pnaOCPdI4Lfm7BpOfCUJA+nczTr\nSUnW7t5356j/tZNUVd3MNKrqvO6XP0xyNXBMkoOr6tLu9qPonNd2OPBpOu/fAXTOofv1pJzrAt8C\nLgP2mOn3J0kaHZY2SdKcVNUtSV4LvDXJoRPuSpL16Iy+H8QlfWcDW064fUP38yZ0zmX70RTPuRI4\nAnj1LPZzbvfzRsClAFW1Atg3yTuBDekUsk27jzvzzid2rwv3dTq/r59ZVbfMYr+SpBFhaZMkzVlV\nfS7JZcD7gNW7m/cB1gOeUVUX9DvDbJdHdo/MnTPF/T8Alk7a9jTgbd3Pl05+wl14YvfzZZPvqKrr\ngesBkuwD/GtV/Wf39qrAScCDgCdU1dWz3K8kaUR4nTZJ0ryasOxwnW4puXP7E+hcSPoi4EbgGXTO\n01oMPB94ZVXd3n3ss4H70ClgmwBvAV44YdtqwKVVddICfU8vo7Ok8Y/XaUvyku62javql91tp9AZ\n3X8hcAedwvZm4OSqetGE13sssA2dEf73AnYHtge2rqqfdR/zaWAvYD/gx5MinVNVk6dPSpJGlEfa\nJEnzIskS4G/pDOEo4PNJLgbeXJ2/EK5G55plVNVXuuVuUVV9PcnewBrADUkeDLykqp6TZFdgEZ3z\nv75OZwnhc4Dj6BS/hTT5r5yZ8HGnfwNeRmcp5O3AL+icqzZ55P9twG7Au+kMHvk+8MRJRyS37+7z\nY1Pk2Bi4ordvQ5I0bPp2pC3JTsA/0vlle0RVHdyXHUmShkaSrwAvrarfJTkV2IFO6fleVW3bfcx7\ngcuq6rNJDgZOvHOwR5L9gQur6luNvgVJkhZcXy6unWQR8Ak6J59vCuye5GH92JckaTgkWQVYu1vY\nFgO3dgdyPAn4XpJtug9dB7gwnYujPQ64Lsmjk9wfeDLwb0kekuR+Lb4PSZIWWr+WR24FXFJVlwMk\n+TydC4Je1Kf9SZIG30bAWd2vH8Gfxu3fTmfJ353XSDsa2BFYQuecrx2Be9NZWnk2naNzN1fVVxci\ntCRJrfVleWSS5wE7VtWrurdfDDy2qvad951JkiRJ0gjry/JI/vJkbUmSJElSD/q1PPIqOsta7rSE\nzgVJ/yiJxU6SJEnSWKuq3NVj+lXazgY2SbIR8CvgBXSuQfNnZhJQ8yfJQVV1UOsc48T3fOH5ni88\n3/OF53u+8HzPF57v+cLzPV94Mz2Q1ZfSVlW3J3k98G06I/+PrCqHkEiSJEnSLPXt4trda+h4HR1J\nkiRJmoN+DSLRYFreOsAYWt46wBha3jrAGFreOsAYWt46wBha3jrAGFreOsAYWt46gKbWl5H/M9px\nUp7TJkmSJGlczbQTeaRNkiRJkgaYpU2SJEmSBpilTZIkSZIGmKVNkiRJkgaYpU2SJEmSBpilTZIk\nSZIGmKVNkiRJkgaYpU2SJEmSBpilTZIkSZIGmKVNkiRJkgaYpU2SJEmSBpilTZIkSZIGmKVNkiRJ\nkgaYpU2SJEmSBpilTZIkSZIGmKVNkiRJkgaYpU2SJEmSBpilTZIkSZIGmKVNkiRJkgaYpU2SJEmS\nBpilTZIkSZIGmKVNkiRJkgaYpU2SJEmSBpilTZIkSZIGmKVNkiRJkgaYpU2SJEmSBpilTZIkSZIG\nmKVNkiRJkgaYpU2SJEmSBpilTZIkSZIGmKVNkiRJkgaYpU2SJEmSBpilTZIkSZIGmKVNkiRJkgaY\npU2SJEmSFljCkpk+1tImSZIkSQsoYRFw3Ewf35fSluSgJFcmObf7sVM/9iNJkiRJQ+its3nwqn0K\nUcBHq+qjfXp9SZIkSRo6CY8B3gRsAVwxk+f0c3lk+vjakiRJkjRUEtYEjgdeV8UvZ/q8fpa2fZP8\nNMmRSdbu434kSZIkaRgcAvxrFSfN5kmpqp72lmQZsP4Ud70DOBP4bff2+4D7VdUrJz2/gPdM2LS8\nqpb3FEaSJEmSBljCc+GUj8Pzj4Ebb+1ufndV3eUKxZ5L28zDZSPg5Kp65KTtNZOAkiRJkjTMEjYE\nfgI8q4qz/rR9Zp2oX9Mj7zfh5q7A+f3YjyRJkiQNsoRVgGOAQycWttno1/TIg5NsRmeK5GXA3n3a\njyRJkiQNsv2B1YAP9PoCfV8eudIduzxSkiRJ0ghL2Bw4BXhMFf/1l/c3XB4pSZIkSeMsYTFwArDf\nVIVtVq/lkTZJkiRJml8JhwNrVrHnyh8zs07Ur3PaJEmSJGksJewC7Ag8ej5ez9ImSZIkSfMk4X7A\np4DnVHH9fLym57RJkiRJ0jzojvc/GvhkFT+cr9e1tEmSJEnS/HgDcC/g/fP5og4ikSRJkqQ5SngU\ncCrw2CoundlzHPkvSZIkSX2XsAad8f5vnmlhm9Xre6RNkiRJknqXcCiwHvCiKmZcsBz5L0mSJEl9\nlvAMYGdgs9kUttmwtEmSJElSDxLuCxwB7FbFdf3aj+e0SZIkSdIsJQT4LHBkFT/o574sbZIkSZI0\ne6+jcx7be/q9IweRSJIkSdIsJDwCOA14fBWX9P46jvyXJEmSpHmVcHc64/3fOpfCNqt9eqRNkiRJ\nkmYm4RBgQzrDR+ZUphz5L0mSJEnzKGFH4Ln0cbz/VCxtkiRJknQXEv4fcBTw4iquWch9e06bJEmS\nJE2jO97/SOC4Kk5b6P17pE2SJEmSprc3sAHwvBY7dxCJJEmSJK1EwsOA7wNbV/Ef8/vajvyXJEmS\npJ4lrE5nvP+B813YZsPSJkmSJElTez9wOXBEyxCe0yZJkiRJkyRsB+zOAo/3n4pH2iRJkiRpgoR7\nA0cDL6/i6sZxHEQiSZIkSXfqjvf/EnBpFW/p775m1olcHilJkiRJf/JK4IF0lkYOBI+0SZIkSRKQ\n8NfAGcC2VVzY//058l+SJEmSZiRhNeB44F0LUdhmw9ImSZIkSfAe4L+BT7YOMpnntEmSJEkaawlL\ngZcyAOP9p+KRNkmSJEljK2Fd4FjgFVX8pnWeqTiIRJIkSdJY6o73/wJwVRVvXPj9O/JfkiRJkqbz\nUuChwJ6tg0zHI22SJEmSxk7Cg4EfAU+u4mdtMjjyX5IkSZL+QsLd6Iz3f2+rwjYbljZJkiRJ4+Zd\nwDXAJ1oHmYmeS1uS5ye5IMkdSTafdN/bk/w8ycVJdph7TEmSJEmau4RtgL2Alw/ieP+pzGUQyfnA\nrsCnJm5MsinwAmBTYAPg1CQPqaoVc9iXJEmSJM1JwtrAccCrqvh16zwz1fORtqq6uKr+c4q7dgFO\nrKrbqupy4BJgq173I0mSJElz1R3vfzjw9Sq+3jrPbPRj5P/9gTMn3L6SzhE3SZIkSWplD+BRwBat\ng8zWtKUtyTJg/SnuOrCqTp7FfoZiragkSZKk0ZOwMXAIsH0VN7fOM1vTlraq2r6H17wKWDLh9obd\nbX8hyUETbi6vquU97E+SJEmSppSwKp3x/h+o4ry2WbIUWDrr58314tpJTgPeUlU/6d7eFDiBznls\nGwCnAg+uSTvy4tqSJEmS+i3h3cATgZ2qGKjhiDPtRD2f05ZkV+DjwHrAN5KcW1VPq6oLk3wBuBC4\nHdhncmGTJEmSpH5LeALwWmDzQStsszHnI20979gjbZIkSZL6JOFewHnAm6r4l9Z5pjLTTmRpkyRJ\nkjRyEo4FbqriNa2zrEzfl0dKkiRJ0iBK2J3OjI2hG+8/FY+0SZIkSRoZCX8F/JjO4JFzWueZzkw7\n0SoLEUaSJEmS+i1hEXAc8OFBL2yzYWmTJEmSNCoOAG4DPtw6yHzynDZJkiRJQy9hK+ANwBbDPN5/\nKh5pkyRJkjTUEtYEjgf2qeLK1nnmm4NIJEmSJA21hKOAFVXs1TrLbDjyX5IkSdLIS3g+sA3w6NZZ\n+sXSJkmSJGkoJSwBPgE8s4obW+fpF89pkyRJkjR0uuP9jwU+VsWPW+fpJ0ubJEmSpGH0Fjp95uDW\nQfrN5ZGSJEmShkrCFsD+wJZV3NE6T795pE2SJEnS0Ei4B3ACsG8VV7TOsxAc+S9JkiRpaCR8Gli9\nipe2zjJXjvyXJEmSNFISdgWeygiP95+KpU2SJEnSwEu4P/BJ4NlV/K51noXkOW2SJEmSBlrCKsAx\nwD9V8aPWeRaapU2SJEnSoHsTsAbwD62DtOAgEkmSJEkDK2Ez4DvAVlVc3jjOvJppJ/JImyRJkqSB\nlLCYznj/N41aYZsNj7RJkiRJGkgJhwFrVbFH6yz94Mh/SZIkSUMrYWfgacBmrbO0ZmmTJEmSNFAS\n1gc+DTy/iutb52nNc9okSZIkDYzueP+jgc9UcUbjOAPB0iZJkiRpkOwLrAW8t3WQQeEgEkmSJEkD\nIeFvgO8Cj6viF63z9Jsj/yVJkiQNjYQ1gOOBt4xDYZsNj7RJkiRJai7h48B9gRdW0aakLDBH/kuS\nJEkaCglPB3YBNhuXwjYbljZJkiRJzSTcBzgC2L2Ka1vnGUSe0yZJkiSpiYQARwFHV3F66zyDytIm\nSZIkqZV96JzHdlDjHAPNQSSSJEmSFlzCw4HlwBOq+HnjOE048l+SJEnSQEpYHTgBOGBcC9tseKRN\nkiRJ0oJK+AiwEfC8cZ4W6ch/SZIkSQMnYQdgN+BR41zYZsPSJkmSJGlBJKwHfBbYs4prWucZFj2f\n05bk+UkuSHJHks0nbN8oyc1Jzu1+HDY/USVJkiQNq+54/yOA46v4Xus8w2QuR9rOB3YFPjXFfZdU\n1aPn8NqSJEmSRsurgQfQWRqpWei5tFXVxQCJs0QkSZIkrVzCQ4H3A9tUcWvrPMOmXyP/N+4ujVye\nZOs+7UOSJEnSgEtYjc54/3dWcXHrPMNo2iNtSZYB609x14FVdfJKnvYrYElVXds91+2rSR5eVTdM\n8foHTbi5vKqWzyy2JEmSpCHxPuAK4NOtg7SWZCmwdNbPm+t12pKcBuxfVefM5n6v0yZJkiSNtoSn\nAMcCm1Vxdes8g2amnWi+lkf+cUdJ1kuyqPv1A4FNgEvnaT+SJEmShkDCvYFjgFdY2OZmLiP/d03y\nS+BxwDeSfKt717bAT5OcC5wE7F1V1809qiRJkqRh0B3v/yngpCq+0zrPsJvz8sied+zySEmSJGkk\nJbwC2A94bBW3tM4zqGbaiSxtkiRJkuZNwibAD4GlVVzQOs8gW+hz2iRJkiSNuYS70Rnvf5CFbf5Y\n2iRJkiTNl4OA3wCHNc4xUqa9TpskSZIkzUTCtsDL6Yz3b3MO1ojySJskSZKkOUlYh8712F5ZxW9a\n5xk1DiKRJEmS1LPueP/PA/9TxRta5xkmM+1ELo+UJEmSNBd7ApsCL2ucY2R5pE2SJElSTxIeBJwJ\nPKWK81vnGTaO/JckSZLUN93x/scD77ew9ZelTZIkSVIv3glcBxzaOsio85w2SZIkSbOSsDXwamDz\nKla0zjPqPNImSZIkacYS1gKOA15dxX+3zjMOHEQiSZIkacYSPgf8rop9WmcZdo78lyRJkjSvEvYA\ntuh+aIF4pE2SJEnSXUrYGDgL2LGKc1vnGQWO/JckSZI0LxJWpXMe28EWtoVnaZMkSZJ0V94O3AIc\n0jrIOPKcNkmSJEkrlfA44HXAFo73b8MjbZIkSZKmlHAv4HjgtVVc1TrPuHIQiSRJkqQpJRwN3FrF\nq1tnGUWO/JckSZLUs4QXAI8HNm+dZdxZ2iRJkiT9mYQHAIcCT6/i963zjDvPaZMkSZL0RwmL6Iz3\n/0gVZ7fOI0ubJEmSpD/3VmAF8OHWQdTh8khJkiRJACQ8BngjsGUVd7TOow6PtEmSJEkiYU064/1f\nX8UvW+fRnzjyX5KkIZOwOXCP1jkkjZy96Yz3f0XrIOPCkf+SJI2YhAAfAF4MXNY4jqTRczXw2tYh\n9JcsbZIkDYHuNLfDgc2Azaq4unEkSdICsbRJkjTgElYHPgesAzy1ihsaR5IkLSAHkUiSNMC6gwFO\npvM7+xkWNkkaP5Y2SZIGVMK6wKnAFcALqvhD40iSpAYsbZIkDaCEDYDvdz9eVcXtjSNJkhqxtEmS\nNGASHgz8ADiuirdW0eb6PJKkgWBpkyRpgCQ8Cjgd+GAVB7fOI0lqz+mRkiQNiIStgS8Br6/ipNZ5\nJEmDwdImSdIASHg6cAywRxXfaZ1HkjQ4el4emeRDSS5K8tMkX06y1oT73p7k50kuTrLD/ESVJGk0\nJewOHAXsbGGTJE02l3PavgM8vKoeBfwn8HaAJJsCLwA2BXYCDkviuXOSJE0hYR/gQ8B2VZzZOo8k\nafD0XKaqallVrejePAvYsPv1LsCJVXVbVV0OXAJsNaeUkiSNmIQk/B3wZmCbKn7WOpMkaTDN1xGw\nVwDf7H59f+DKCfddCWwwT/uRJGnoJawCHAI8D9i6issaR5IkDbBpB5EkWQasP8VdB1bVyd3HvAO4\ntapOmOalvL6MJElAwt2AI4AHAdtWcV3jSJKkATdtaauq7ae7P8nLgKcDT52w+SpgyYTbG3a3TfX8\ngybcXF5Vy6fbnyRJwyxhDeCf6fz+3aGKmxpHkiQtoCRLgaWzfl5VbwfBkuwEfATYtqqunrB9U+AE\nOuexbQCcCjy4Ju0oSVVVetq5JElDJuFewNeAXwEvq+LWxpEkSY3NtBPN5TpthwKrAcuSAPyoqvap\nqguTfAG4ELgd2GdyYZMkaZwk3Af4FnAmsG8VK+7iKZIk/VHPR9rmvGOPtEmSxkDCA4BlwBeAd1V5\nnrckqWOmncjrp0mS1CcJDwXOAA6v4u8sbJKkXsxleaQkSVqJhC2Bk4G3VXFs6zySpOFlaZMkaZ4l\nPJnOlMi9qvha6zySpOHm8khJkuZRwrPpFLbdLGySpPlgaZMkaZ4kvAw4HHhaFcvbppEkjQqXR0qS\nNA8S3gzsBzy5iotb55EkjQ5LmyRJc5AQ4H3A84BtqriicSRJ0oixtEmS1KOERcAngMfQKWy/bRxJ\nkjSCLG2SJPUgYTXgWOC+wFOq+F3jSJKkEWVpkyRplhLuAXwR+AOdoSO3NI4kSRphTo+UJGkWEtYB\nvgP8GniehU2S1G+WNkmSZijhfsDpwFnAK6u4vXEkSdIYsLRJkjQDCQ8EzqBz4ez9q1jROJIkaUx4\nTpskSXch4ZHAt4C/r+Lw1nkkSePF0iZJ0jQSHg98Fdivis+3ziNJGj+WNkmSViJhR+A44CVVnNI6\njyRpPHlOmyRJU0jYjc512Ha1sEmSWrK0SZI0ScLewCHA9lX8a+s8kqTx5vJISZK6EgIcAOwFPKmK\nXzSOJEmSpU2SJPhjYfsQsCOwTRW/ahxJkiTA0iZJEgmrAp8GHgZsW8U1jSNJkvRHljZJ0lhLuDtw\nIrAY2K6K3zeOJEnSn3EQiSRpbCXcE/gGcCuws4VNkjSILG2SpLGUsB7wPeAS4EVV3No4kiRJU7K0\nSZLGTsIS4AfAMuA1VdzROJIkSStlaZMkjZWEh9ApbEdUcWAV1TqTJEnTcRCJJGlsJGwOfB14ZxVH\ntc4jSdJMWNokSWMhYVvgJGDvKr7SOo8kSTPl8khJ0shL2JlOYdvdwiZJGjaWNknSSEvYE/gM8Iwq\nvts6jyRJs+XySEnSyErYD9gfeHIVF7XOI0lSLyxtkqSRkxDgIGB3YJsq/qttIkmSemdpkySNlIRV\ngI8BWwNbV/GbxpEkSZoTS5skaWQk3A04GtgQWFrF9W0TSZI0d5Y2SdJISFgMfAEoYKcqbm4cSZKk\neeH0SEnS0EtYG/g2cC3wHAubJGmUWNokSUMt4b7AcuAc4KVV3NY2kSRJ88vSJkkaWgkbAWcAXwbe\nWMWKtokkSZp/PZe2JB9KclGSnyb5cpK1uts3SnJzknO7H4fNX1xJkjoSHg78APhYFe+tolpnkiSp\nH1LV2++4JNsD362qFUk+CFBVByTZCDi5qh55F8+vqkpPO5ckjbWExwL/AuxfxfGt80iS1IuZdqKe\nj7RV1bKqunMZyll0xitLktRXCdsBJwOvtLBJksbBfJ3T9grgmxNub9xdGrk8ydbztA9J0phLeC5w\nAvDcKr7ROo8kSQth2uu0JVkGrD/FXQdW1cndx7wDuLWqTuje9ytgSVVdm2Rz4KtJHl5VN0zx+gdN\nuLm8qpb38D1IksZAwl7Ae4EdqjivdR5JkmYryVJg6ayf1+s5bd2dvgx4FfDUqrplJY85Ddi/qs6Z\ntN1z2iRJM5LwVuC1dArbz1vnkSRpPsy0E017pO0udrAT8LfAthMLW5L1gGur6o4kDwQ2AS7tdT+S\npPGVEOCDwDOBrau4qnEkSZIWXM+lDTgUWA1YlgTgR1W1D7At8J4ktwErgL2r6ro5J5UkjZWERcAn\ngb8BnlTF/zaOJElSE3NaHjmnHbs8UpK0EgmrA58D1gGeXcWNjSNJkjTv+j7yX5KkfkhYk85I/1WA\nZ1jYJEnjztImSRoYCfcGTgWuAHar4g+NI0mS1JylTZI0EBI2AL4PnA68qoo7GkeSJGkgWNokSc0l\nbAKcARxbxduqaHPCtSRJA2gu0yMlSZqzhM2AbwLvruIzrfNIkjRoLG2SpGYStga+BLyuii+2ziNJ\n0iCytEmSmkh4OnA0sEcVyxrHkSRpYHlOmyRpwSW8CDgKeJaFTZKk6XmkTZK0oBJeBxwAPLWKC1rn\nkSRp0FnaJEkLIiHA3wEvAZ5UxWWNI0mSNBQsbZKkvktYBTgE2BbYuopfN44kSdLQsLRJkvoq4W7A\nkcADgaVVXNc4kiRJQ8XSJknqm4Q1gH+m8/tmhypuahxJkqSh4/RISVJfJKwFnALcAOxiYZMkqTeW\nNknSvEu4D3AacD6wZxW3NY4kSdLQsrRJkuZVwl8BPwC+DuxbxYrGkSRJGmqWNknSvEl4GJ3CdlgV\n76qiWmeSJGnYOYhEkjQvErYETgbeVsWxrfNIkjQqLG2SpDlLeArweWCvKr7WOo8kSaPE0iZJ6ln3\nGmzPAg4Hnl/F6Y0jSZI0cixtkjTmElYF1gbWBdaZ9DHVtonbVweuAHaq4pwFDy9J0hhIVZtzxJNU\nVaXJziVpxCQsolO8Zlu61gEWA9cD1wLXdD9P/Jhu240OG5EkqTcz7USWNkkaEAmrAGsx+9K1DrAm\nnYtYz7Z0XQvc4Fh+SZIW3kw7kcsjJWkeJQS4F7MvXet0n3cj0xesS1ey/XqLlyRJo8nSJkmTdIvX\nmqy8XE1XxNYCbmL64nXFFNuuBa6r4o6F+B4lSdLwsLRJGknd4rWY3oZrrA38gemXGF41xbY7i9dt\nC/E9SpKk8dD0nDaoi5rsXNIomziQ43Zmfl7XxG3XVXHrgieXJEljZSgGkUA9vMnOJY2yFcB1wLVV\n/KF1GEmSpJUZitLm9EhJkiRJ42qmnWiVhQgjSZIkSeqNpU2SJEmSBpilTZIkSZIGmKVNkiRJkgaY\npU2SJEmSBpilTZIkSZIGmKVNkiRJkgaYpU2SJEmSBljPpS3J+5L8NMl5Sb6bZMmE+96e5OdJLk6y\nw/xElSRJkqTxM5cjbf+/qh5VVZsBXwXeDZBkU+AFwKbATsBhSTyiNwCSLG2dYdz4ni883/OF53u+\n8HzPF57v+cLzPV94vueDq+cyVVU3TLi5JnB19+tdgBOr6raquhy4BNiq54SaT0tbBxhDS1sHGENL\nWwcYQ0tbBxhDS1sHGENLWwcYQ0tbBxhDS1sH0NRWncuTk/w9sCdwM38qZvcHzpzwsCuBDeayH0mS\nJEkaV9MeaUuyLMn5U3zsDFBV76iqBwCfBf5xmpeqecwsSZIkSWMjVXPvU0keAHyzqh6R5ACAqvpg\n975TgHdX1VmTnmORkyRJkjTWqip39Ziel0cm2aSqft69uQtwbvfrrwEnJPkonWWRmwD/1ks4SZIk\nSRp3czmn7QNJ/hq4A/gF8FqAqrowyReAC4HbgX1qPg7nSZIkSdIYmpflkZIkSZKk/mhy/bQkO3Uv\nvP3zJG9rkWGcJDkqyf8kOb91lnGRZEmS05JckORnSd7QOtOoS3L3JGclOS/JhUk+0DrTuEiyKMm5\nSU5unWXxbUa/AAAEOElEQVQcJLk8yb933/O/OP1A8y/J2km+mOSi7s+Xx7XONMqS/HX3v+87P673\n92j/JXl7998t5yc5IcnqrTONuiT7dd/vnyXZb9rHLvSRtiSLgP8AtgOuAn4M7F5VFy1okDGSZBvg\nRuDYqnpk6zzjIMn6wPpVdV6SNYGfAM/2v/P+SrK4qm5KsipwBvCWqjqjda5Rl+TNwBbAPavqWa3z\njLoklwFbVNU1rbOMiyTHAKdX1VHdny/3qKrrW+caB0lWofPvxa2q6pet84yqJBsB3wMeVlV/SPLP\ndIYMHtM02AhL8gjgROAxwG3AKcBrquoXUz2+xZG2rYBLquryqroN+DydQSbqk6r6AXBt6xzjpKp+\nXVXndb++EbiIzjUM1UdVdVP3y9WARYD/qO2zJBsCTweOABwwtXB8rxdIkrWAbarqKICqut3CtqC2\nA35hYeu739EpDou7f5hYTKcsq38eCpxVVbdU1R3A6cBzVvbgFqVtA2Di/3hefFsjrfvXq0cDZ03/\nSM1VklWSnAf8D3BaVV3YOtMYOAT4W2BF6yBjpIBTk5yd5FWtw4yBjYHfJvlsknOSfCbJ4tahxsgL\ngRNahxh13SP3HwGuAH4FXFdVp7ZNNfJ+BmyTZN3uz5RnABuu7MEtSpuTTzQ2uksjvwjs1z3ipj6q\nqhVVtRmdH3pPSrK0caSRluSZwG+q6lw88rOQnlhVjwaeBryuuwRe/bMqsDlwWFVtDvweOKBtpPGQ\nZDVgZ+Ck1llGXZIHAW8ENqKzMmjNJHs0DTXiqupi4GDgO8C36Fw+baV/AG1R2q4Clky4vYTO0TZp\npCS5G/Al4HNV9dXWecZJd+nSN4AtW2cZcU8AntU9x+pE4ClJjm2caeRV1X93P/8W+Aqd0w7UP1cC\nV1bVj7u3v0inxKn/ngb8pPvfuvprS+CHVfW/VXU78GU6P+PVR1V1VFVtWVXbAtfRmfsxpRal7Wxg\nkyQbdf+C8gI6F+SWRkaSAEcCF1bVP7bOMw6SrJdk7e7XawDb0/mrlfqkqg6sqiVVtTGdJUzfq6qX\ntM41ypIsTnLP7tf3AHYAnAzcR1X1a+CXSR7S3bQdcEHDSONkdzp/EFL/XQw8Lska3X/DbEfnmsvq\noyT36X5+ALAr0ywFnsvFtXtSVbcneT3wbTqDAo50ol5/JTkR2Ba4d5JfAu+qqs82jjXqngi8GPj3\nJHcWh7dX1SkNM426+wHHdCeNrQIcV1XfbZxp3Lj8vf/uC3yl828qVgWOr6rvtI00FvYFju/+sfkX\nwMsb5xl53T9KbAd43uYCqKqfdldKnE1nid45wKfbphoLX0xybzpDYPapqt+t7IFeXFuSJEmSBliT\ni2tLkiRJkmbG0iZJkiRJA8zSJkmSJEkDzNImSZIkSQPM0iZJkiRJA8zSJkmSJEkDzNImSZIkSQPM\n0iZJkiRJA+z/AAXqk45gHqSmAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x79b26a0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA4AAAAFICAYAAAABEJCnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBkBJkiRJGhD/D7JAr2cY/QmFAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x811d2e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "genetic_algorithm()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Vemos que alcanzamos el resultado con muchas menos evaluaciones de la función que el algoritmo heurístico más básico (350 frente a 100000 del ascenso de colinas con reinicios aleatorios). Hemos duplicado el rendimiento para esta función 2D, pero las mejoras de rendimiento son más espectaculares cuanto mayor es la dimensión del problema que se quiere optimizar."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##Ejercicio: \n",
    "\n",
    "Usando este algoritmo genético, minimiza la función de la \"huevera\" en el cuadrado x1, x2 ∈ [-512, 512].\n",
    "Esta función está llena de óptimos locales, cambia los parámetros del problema para ver su influencia en la convergencia.\n",
    "El óptimo absoluto vale -959.6407 y se encuentra en el punto (512, 404.2319)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from codigo import eggholder"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=imagenes/eggholder.png alt=\"Función de Huevera\" style=\"width: 500px;\"/>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def calculate_performances (individual):\n",
    "    \n",
    "    dict_perfo = {\n",
    "        \n",
    "        'cost' : eggholder([individual.traits['x1'], individual.traits['x2']])\n",
    "\n",
    "    }\n",
    "    \n",
    "    individual.performances = dict_perfo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def genetic_algorithm (minimum = -512,\n",
    "                       maximum = 512,\n",
    "                       number_of_genes = 15,\n",
    "                       size_of_society = 60, \n",
    "                       reproduction_rate = 0.5,\n",
    "                       mutation_rate = 0.2, \n",
    "                       number_of_generations = 100,\n",
    "                       size_after_tournament = 40,\n",
    "                      ):\n",
    "\n",
    "    \n",
    "    dict_genes = {'x1' : number_of_genes,\n",
    "                  'x2' : number_of_genes,\n",
    "                 }\n",
    "    \n",
    "            \n",
    "    society = []\n",
    "    fitness_max = []\n",
    "    fitness_avg = []\n",
    "\n",
    "    for generation in range(number_of_generations):\n",
    "    \n",
    "        immigration (society, size_of_society, dict_genes, minimum, maximum)\n",
    "        fitness_max += list([max([individual.fitness for individual in society])])\n",
    "        fitness_avg += list([np.mean([individual.fitness for individual in society])])\n",
    "        crossover(society, reproduction_rate, mutation_rate, minimum, maximum)\n",
    "        tournament (society, size_after_tournament)\n",
    "    \n",
    "    fitness_list = [individual.fitness for individual in society]\n",
    "    winner = society[fitness_list.index(max(fitness_list))]\n",
    "    print(winner.traits)\n",
    "    \n",
    "        \n",
    "    plt.figure(figsize=(15,5))\n",
    "    plt.plot(fitness_max)\n",
    "    plt.title('$\\Phi_{max} = $' + str(round(fitness_max[-1], 5)), fontsize = 15)  \n",
    "    \n",
    "    plt.figure(figsize=(15,5))\n",
    "    plt.plot(fitness_avg)\n",
    "    plt.title('$\\overline{\\Phi}$', fontsize = 15) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'x1': 487.97653958944284, 'x2': 442.93255131964804}\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA3EAAAFGCAYAAADElBZwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3X+4pWVd7/H3R8ZBRtERDBh+KHMJpKMmcgSxREcvIygD\ntROQVpg/Lk50kMxTgudcOdXJsqOcLA9aCkYdGSN/EHS0AG2skyYZSMg4CQYHBhl+BiJgzTDf88fz\nbFls9l5779lrrWetvd+v69rXXs+z7nWv75IHF5993899p6qQJEmSJE2Gx3VdgCRJkiRp/gxxkiRJ\nkjRBDHGSJEmSNEEMcZIkSZI0QQxxkiRJkjRBDHGSJEmSNEEMcZIkSZI0QQxxkiRJkjRBDHGSpImU\n5NVJ/inJd5P8S5K3zdDmPyb5YpK7kjyUZEuS/5rk8X36PSDJd5LsTLJqjhrm7D/JpravmX5eNK2/\nFUnOSnJ9+7luSXLOQj97kkOS/EHb5uEkf93vc0iSJsuKrguQJHUryUHA2cArgMOALwP/AJxZVdVl\nbbNJ8kPAp4CPAL8EHA28J8nOqnp/T9O9gCuA9wD3Ai8CNgD7AWfM0v3/AO4H9phHKfPp/+eBPXvL\nB34dOJzmf+defwS8vO1jC/B04Nm9Deb52dcBxwNfovmuH8t/jpKkXZMx/X6WJI1YkpcDnwP2qqp7\nu66nnyR/BTyhql7Wc+69wM8B+1XV9j6v/e/AL1TVU2d47qXAp4F304S5J1XVgwusbdb+2+dXAtuA\njVX1Cz3njwMuAX6gqrb06X/Oz54kUwE8ySdo/pm+YiGfQ5I0vpxOKUkiyZOB57WH65M8qct65uH5\nwOXTzl0OPBV48RyvvQd4zHTKJLsBvw/8GnD3Imqbsf8exwGrgY3Tzr8R+Fy/ANea87OP6wiqJGkw\nDHGStIy192C9F3gn8FB7+l7gvUne2V1lc3oC8O/Tzk0dP2t64yS7JVmV5CU00xw/NEOf/4kmfP2v\nhRYzz/6nnALcUlX/d9r5o4Drk3wgyX1JHkjyySRrprVb0GeXJC093hMnScvbecAeVXVSkvXtuauq\nalOSryR5QlX9aof1zeYG4Mhp545qf+81Q/sHgJXt4wuBX+l9MsneNPepvb6qHk6y0Hr69t/zPquA\nE4APzvD0GuANwFeBk4EnA79DM73z6J52C/3skqQlxhAnSctUkr2AnwH+8yxNvgC8FRhqiGunbr6f\nuWeHXFdV720ffwj4UJI3A5+kCTFTKzTunOG1RwOraBYe+VWaEHVaz/O/CXypqv5ylz7E3P1P+fG2\n3fSplNAseAJwYlX9K0CS24AvJHl5VU2tMLnQzy5JWmIMcZK0fD1IM4K05yzPPxG4Y9hFVNV3gDct\n8GXn09wb9kHgD2k+x1k097Rtm+E9vto+/GKSu4ALkrynqv4lyXNoFgV5aZLVbbuprQVWt2uEPDS9\nz/n2P63pKcD1VXXVDN3cA3xzKsC1/o5mquQ6YCrELeizS5KWHu+Jk6Rlqqq+S7P8/eun7YeWJE+j\nWYBjptGkzlXVzqo6A3gazYIs+9JsjQDw93O8/Or298Ht70Np7oX7Ek2Qugf4QPvcVppRwoWY3j8A\nSZ5Cs+z/TKNwAF9n5u/l0LNFwCI/uyRpCXAkTpKWsar630luBH4D2L09fTpNQPixqrpu2DXs4nRK\nAKrqPuC+tp/Tgb+rqm/M0c8Ptb9vbH//LbB+WpvjgXe0v6ePps1lev9TXkNz39xsIe4vgF9LsndV\nTa2O+VKagPnV6Y138bNLkpYA94mTJAHQLmzyeeCpbUCYOv+DwKtoRoq+A/wYzeIdq4CfBN5UVTva\ntq8G9qEJZIcC/4VmCuHUuZXAv1TVny2y1hcBx9CEmycDPwX8MPCSqvpaT7u/pFl+fzPwME3A+iXg\n0qp6XZ/+30AzbfF7+8Ql+dn23NqqumWh/bdt96mqI2Z5zz2BrwG30uxT92SaTcQ3V9WPLOSzJ9mD\n5p8TwNtppsxuaI//z1zTQyVJ482ROEla5pIcBPwyzeIcBXw8yRbgl9r9xlbSTDGkqj7dhr3dquov\nkpwG7AHcn+QQ4Ger6rVJXgPsRnPv1l/Q3NP1WuBPaILgYm0HTgLeRbOYx98APzTDyOGVNCs+Hgzs\nAL5Jc/9Yvy0Apkz/K2d6fhbUfzs99RXAf5v1zaruT/IK4PeAj9PcC3cxjyxaMmU+n31f4KJpn+Oi\n9vFa4ObZ6pAkjb85R+KSnAm8meZL68NV9f4kG9pzd7bN3llVn23bn02zYenDwFur6rIh1S5JGpEk\nnwZOrapvJ7kCOJbme+HzVfWyts2vAzdW1UeTvAfYOLXgR5K304wofbajjyBJ0pLR9/6DJM+lCWtH\n0vw19VVJnknzl7xzquoF7c9UgFtHs7fNOpob4s9N4uIpkjTB2v8fX90GuFXAv1fVTpr7tT6f5Ji2\n6VOBzWk2WTsauDfJC5LsD7wcuDLJYTNsXi1JkhZgrumUzwK+3K5gRpIv0EyHgUdPJ5lyIs1fXrcD\nNyW5gWb/GlfLkqTJdTCPrH74XJr75qCZPlg0My8A/gj4EeAgmvu1fgTYm2Yq5ldoRu8eqqqLR1G0\nJElLVd/plEmeBfw58GLgu8AVNF/Ed9PsqXNfe/z2qro3ye8Df19VH2tf/xHgs1X1yaF+CkmSJEla\nJvpOdayqLTQrY10GfJbmL6sPA+fS3Bh9OHAb8L5+3QykUkmSJEnS3KtTVtX5NEsqk+TdwM1VNbWg\nydRo26Xt4a0002imHNiee5QkBjtJkiRJy1pVzXSL2pzmDHFJ9qmqO5I8nWaj0hclWVNVt7VNXgNc\n2z6+BLgwyTnAATR7BF05yIKlYUqyoao2dF2HNBOvT40rr02NM69PjavFDGzNZ5+4TyTZm2ZfmtPb\n1ck+kORwmqmSNwKnAVTV5iQX0Wx6uqNt76ibJEmSJA3IfKZTvnSGcz/bp/27gXcvsi5JkiRJ0gzc\nw016tE1dFyD1sanrAqRZbOq6AKmPTV0XIA1a3y0GhvamSXlPnCRJkqTlajGZyJE4SZIkSZoghjhJ\nkiRJmiCGOEmSJEmaIIY4SZIkSZoghjhJkiRJmiCGOEmSJEmaIIY4SZIkSZogK7ouQJIkadgSngAc\n0nUdkpaVLVXsGEbHhjhJkrSkJQT4FPAs4MGOy5G0fBwD/OswOjbESZKkpe5UYD/g+6vY3nUxkrRY\nqarRv2lSVZWRv7EkSVpWEg4ArgZ+uIpruq5HkqYsJhO5sIkkSVqS2mmUHwI+aICTtJQ4nVKSJC1V\nrweeAfxE14VI0iA5nVKSJC05CfsB1wA/WsU/dl2PJE3ndEpJkqRWO43yXOA8A5ykpcjplJIkaak5\niWY7gdd1XYgkDYPTKSVJ0pKR8H3AtcCJVXy563okaTaLyUSGOEmStGQk/ClwcxW/3HUtktTPYjKR\n0yklSdKSkPBa4HDgDR2XIklDZYiTJEkTL2Fv4APAT1bxUNf1SNIwGeKGKOHngX26rkOSpGXgpcCf\nVvF3XRciScPmPXFDkvB44EHg3V3XIknSMvAQ8HtVPNh1IZI0Hy5sMoYSDgD+oYr9u65FkiRJ0nhx\ns+/xtAbY1nURkiRJkpYWQ9zw7IchTpIkSdKAGeKGZw1wW9dFSJIkSVpaDHHD40icJEmSpIEzxA3P\nfjgSJ0mSJGnADHHD48ImkiRJkgbOEDc8jsRJkiRJGjhD3PA4EidJkiRp4AxxQ5AQXNhEkiRJ0hAY\n4objKcD2Kh7ouhBJkiRJS4shbjgchZMkSZI0FIa44XBRE0mSJElDYYgbDhc1kSRJkjQUc4a4JGcm\nuTbJ15Kc2Z7bK8nlSb6R5LIkq3van53k+iRbkhw7zOLHmNMpJUmSJA1F3xCX5LnAm4EjgecDr0ry\nTOAs4PKqOgz4XHtMknXAycA64Djg3CTLcbRvDU6nlCRJkjQEcwWsZwFfrqrvVtXDwBeAnwBOAC5o\n21wAvLp9fCKwsaq2V9VNwA3AUQOvevw5EidJkiRpKOYKcV8DjmmnT64CfhQ4ENi3qm5v29wO7Ns+\n3h/Y2vP6rcABA6x3UriwiSRJkqShWNHvyarakuQ9wGXAA8BXgYentakk1a+bmU4m2dBzuKmqNs2n\n4AnhwiaSJEmSvifJemD9IPrqG+IAqup84Pz2jX+TZnTt9iT7VdW2JGuAO9rmtwIH9bz8wPbcTP1u\nWETd486ROEmSJEnf0w5abZo6TvKuXe1rPqtT7tP+fjrwWuBC4BLg1LbJqcDF7eNLgFOSrEyyFjgU\nuHJXi5tECSuB1cDdXdciSZIkaemZcyQO+ESSvYHtwOlVdV+S3wYuSvIm4CbgJICq2pzkImAzsKNt\n32+q5VK0D3Bn1aOnnUqSJEnSIKSLjJWkqiojf+MRSDgS+FAV/6HrWiRJkiSNp8VkouW4h9uwub2A\nJEmSpKExxA2ei5pIkiRJGhpD3OC5vYAkSZKkoTHEDZ7TKSVJkiQNjSFu8NbgdEpJkiRJQ2KIGzxH\n4iRJkiQNjSFu8FzYRJIkSdLQGOIGKCE4EidJkiRpiAxxg/UU4N+reLDrQiRJkiQtTYa4wXJ7AUmS\nJElDZYgbLKdSSpIkSRoqQ9xgub2AJEmSpKEyxA2WI3GSJEmShsoQN1huLyBJkiRpqAxxg+XCJpIk\nSZKGyhA3WE6nlCRJkjRUhrjBcmETSZIkSUNliBssR+IkSZIkDZUhbkASVgJPBu7quhZJkiRJS5ch\nbnD2Be6sYmfXhUiSJElaugxxg+NUSkmSJElDZ4gbHBc1kSRJkjR0hrjBcSROkiRJ0tAZ4gbHkThJ\nkiRJQ2eIGxxH4iRJkiQNnSFucPbDkThJkiRJQ2aIG5w1OBInSZIkacgMcYPjdEpJkiRJQ2eIG4CE\nYIiTJEmSNAKGuMFYDfxbFQ92XYgkSZKkpc0QNxguaiJJkiRpJAxxg+GiJpIkSZJGwhA3GN4PJ0mS\nJGkkDHGDsQanU0qSJEkaAUPcYDgSJ0mSJGkkDHGD4UicJEmSpJEwxA2GI3GSJEmSRmLOEJfk7CTX\nJbk2yYVJdk+yIcnWJFe3P8dPa399ki1Jjh1u+WPDLQYkSZIkjUSqavYnk4OBzwPPrqp/S/KnwGeA\ng4H7q+qcae3XARcCRwIHAFcAh1XVzmntqqoyuI/RrYS7gWdVcWfXtUiSJEkaf4vJRHONxH0b2A6s\nSrICWAXcOvW+M7Q/EdhYVdur6ibgBuCoXSlsUiTsDuwJ3N11LZIkSZKWvr4hrqruAd4H3Ax8C7i3\nqq5onz4jyTVJzkuyuj23P7C1p4utNCNyS9m+wB1V7JyzpSRJkiQtUt8Ql+SZwC/STJ/cH3hSktcD\nHwTWAofT3Av2vj7dzD5fc2lwURNJkiRJI7NijudfCHyxqu4GSPIp4Aer6mNTDZJ8BLi0PbwVOKjn\n9QfyyPTLR0myoedwU1VtWlDl48NFTSRJkiT1lWQ9sH4gfc2xsMnzgY/RLFTyXeCPgCuBT1bVtrbN\n24Ajq+p1PQubHMUjC5scUtPeZCktbJJwGvDCKt7SdS2SJEmSJsNiMlHfkbiquibJHwNfAXYCVwEf\nBj6S5HCaqZI3Aqe17TcnuQjYDOwATp8e4JYgp1NKkiRJGpm+I3FDe9OlNRL3IeCfqji361okSZIk\nTYZhbjGguTkSJ0mSJGlkDHGL58ImkiRJkkbGELd4a3AkTpIkSdKIGOIWISE4nVKSJEnSCBniFuep\nwENVPNR1IZIkSZKWB0Pc4jgKJ0mSJGmkDHGLswYXNZEkSZI0Qoa4xXEkTpIkSdJIGeIWx+0FJEmS\nJI2UIW5x3F5AkiRJ0kgZ4hbH6ZSSJEmSRsoQtzgubCJJkiRppAxxi+NInCRJkqSRMsQtjgubSJIk\nSRopQ9wuStgd2BO4p+taJEmSJC0fhrhdty9wexU7uy5EkiRJ0vJhiNt1bi8gSZIkaeRWdF3Arkj4\nLeCIjst4GnBrxzVIkiRJWmZSVaN/06SqKrv2WgLcC7wBeHCQde2Cf67ipo5rkCRJkjRhFpOJJnEk\n7kDgwSo+3XUhkiRJkjRqk3hP3HOB67ouQpIkSZK6MIkh7jnA17ouQpIkSZK6MIkhzpE4SZIkScvW\npIY4R+IkSZIkLUsTtTplwuOA+4H9q7hv8JVJkiRJ0vAtZnXKSRuJWwvcbYCTJEmStFxNWohzURNJ\nkiRJy9qkhTgXNZEkSZK0rE1aiHMkTpIkSdKyNmkhzpE4SZIkScvaxKxOmbCCZmXKp1XxwHAqkyRJ\nkqThWy6rUx4CfMsAJ0mSJGk5m6QQ5ybfkiRJkpa9SQpxz8H74SRJkiQtc5MU4hyJkyRJkrTsTVKI\nc3sBSZIkScveRKxOmbA7cB/wlCr+bXiVSZIkSdLwLYfVKQ8DbjTASZIkSVru5gxxSc5Ocl2Sa5Nc\nmGT3JHsluTzJN5JclmT1tPbXJ9mS5NgB1ekm35IkSZLEHCEuycHAW4Ajqup5wG7AKcBZwOVVdRjw\nufaYJOuAk4F1wHHAuUkGMdrnoiaSJEmSxNwjcd8GtgOrkqwAVgHfAk4ALmjbXAC8un18IrCxqrZX\n1U3ADcBRA6jT7QUkSZIkiTlCXFXdA7wPuJkmvN1bVZcD+1bV7W2z24F928f7A1t7utgKHDCAOh2J\nkyRJkiRgRb8nkzwT+EXgYJrVIf8syU/3tqmqStJvicsZn0uyoedwU1Vtmrkdq2iC4A39apUkSZKk\ncZVkPbB+EH31DXHAC4EvVtXd7Rt/CngxsC3JflW1Lcka4I62/a3AQT2vP7A99xhVtWGeNT4buL6K\n7fNsL0mSJEljpR202jR1nORdu9rXXPfEbQGOTrJHkgCvBDYDlwKntm1OBS5uH18CnJJkZZK1wKHA\nlbtaXMtNviVJkiSp1XckrqquSfLHwFeAncBVwB8CewIXJXkTcBNwUtt+c5KLaILeDuD0Wvxu4m4v\nIEmSJEmtLD5j7cKbLmB38oTPAH9QxZ8PuSxJkiRJGomFZKLpBrGH27C5vYAkSZIktcY6xCU8GXga\ncGPXtUiSJEnSOBjrEAesA75excNdFyJJkiRJ42DcQ5yLmkiSJElSj3EPcW4vIEmSJEk9xj3EORIn\nSZIkST0mIcQ5EidJkiRJrbENcQl7A08Ebum6FkmSJEkaF2Mb4mj3h6ti9LuRS5IkSdKYGvsQ13UR\nkiRJkjROxjnEeT+cJEmSJE0zziHO7QUkSZIkaZqxDHEJwe0FJEmSJOkxxjLEAfsAAbZ1XYgkSZIk\njZNxDXHPxZUpJUmSJOkxxjnEeT+cJEmSJE0zriHO7QUkSZIkaQbjGuIciZMkSZKkGYxdiGtXpnQk\nTpIkSZJmMHYhDjgQeKiKu7ouRJIkSZLGzTiGuEOA67suQpIkSZLG0TiGuKcD/6/rIiRJkiRpHI1j\niHsGhjhJkiRJmpEhTpIkSZImyDiGuKcDN3ddhCRJkiSNo3EMcY7ESZIkSdIsxirEJTwOR+IkSZIk\naVZjFeKA7wO+U8UDXRciSZIkSeNo3EKcUyklSZIkqQ9DnCRJkiRNEEOcJEmSJE2QcQtxLmoiSZIk\nSX2MW4hzJE6SJEmS+jDESZIkSdIEGbcQ53RKSZIkSepjbEJcwp7A7sBdXdciSZIkSeNqbEIczVTK\nm6uorguRJEmSpHHVN8Ql+f4kV/f83JfkzCQbkmztOX98z2vOTnJ9ki1Jjl1ALU6llCRJkqQ5rOj3\nZFX9M/ACgCSPA24FPgW8ETinqs7pbZ9kHXAysA44ALgiyWFVtXMetbioiSRJkiTNYSHTKV8J3FBV\ntwBpf6Y7EdhYVdur6ibgBuCoefZviJMkSZKkOSwkxJ0CbGwfF3BGkmuSnJdkdXt+f2Brz2u20ozI\nzYfTKSVJkiRpDvMKcUlWAj8O/Fl76oPAWuBw4DbgfX1ePt+FShyJkyRJkqQ59L0nrsfxwD9W1Z0A\nVXXH1BNJPgJc2h7eChzU87oD23OPkWRDz+EmKEOcJEmSpCUpyXpg/UD6qpp7oCzJx4HPVtUF7fGa\nqrqtffw24Miqel27sMmFNPfBHQBcARxS094kSVVVHjnm8cADwKoqdgzig0mSJEnSuJqeiRZizpG4\nJE+kWdTkLT2n35PkcJqpkjcCpwFU1eYkFwGbgR3A6dMD3CwOBLYZ4CRJkiSpv3mNxA38TR87Erce\n+I0qjhl5MZIkSZI0YosZiVvI6pTD5P1wkiRJkjQP4xLi3F5AkiRJkuZhXEKcI3GSJEmSNA+GOEmS\nJEmaIOMS4pxOKUmSJEnz0HmISwhNiHMkTpIkSZLm0HmIA/YBHqjiga4LkSRJkqRxNw4hzlE4SZIk\nSZqncQhxz8D74SRJkiRpXsYlxDkSJ0mSJEnzMA4hzumUkiRJkjRP4xDinE4pSZIkSfM0LiHOkThJ\nkiRJmodxCHFOp5QkSZKkeeo0xCU8CdgDuKvLOiRJkiRpUnQ9EvcM4OYqquM6JEmSJGkidB3inEop\nSZIkSQvQdYhzZUpJkiRJWoBxCHGOxEmSJEnSPBniJEmSJGmCdB3ivCdOkiRJkhag6xDnPXGSJEmS\ntACpGv3q/kkKaiXwAPDEKraPvAhJkiRJ6kiSqqrsymu7HIk7ANhmgJMkSZKk+esyxDmVUpIkSZIW\nqOsQ56ImkiRJkrQAXYY4V6aUJEmSpAXqeiTO6ZSSJEmStABdhzhH4iRJkiRpAZxOKUmSJEkTpOsQ\n53RKSZIkSVqALkPcQ1V8p8P3lyRJkqSJ02WIcyqlJEmSJC2QIU6SJEmSJkiXIc774SRJkiRpgRyJ\nkyRJkqQJYoiTJEmSpAnidEpJkiRJmiB9Q1yS709ydc/PfUnemmSvJJcn+UaSy5Ks7nnN2UmuT7Il\nybF9unckTpIkSZIWKFU1v4bJ44BbgaOAM4C7qup3krwDeGpVnZVkHXAhcCRwAHAFcFhV7ZzWV0E9\nror5vbkkSZIkLSFJqqqyK69dyHTKVwI3VNUtwAnABe35C4BXt49PBDZW1faqugm4gSb0PYYBTpIk\nSZIWbiEh7hRgY/t436q6vX18O7Bv+3h/YGvPa7bSjMhJkiRJkgZgxXwaJVkJ/DjwjunPVVU10yNn\nNeNzSTb0HG6qqk3zqUWSJEmSJk2S9cD6QfQ1rxAHHA/8Y1Xd2R7fnmS/qtqWZA1wR3v+VuCgntcd\n2J57jKrasAv1SpIkSdLEaQetNk0dJ3nXrvY13+mUP8UjUykBLgFObR+fClzcc/6UJCuTrAUOBa7c\n1eIkSZIkSY825+qUSZ5Isx3A2qq6vz23F3AR8HTgJuCkqrq3fe6dwBuBHcCZVfVXM/S5yyuxSJIk\nSdKkW0wmmvcWA4NkiJMkSZK0nI1qiwFJkiRJUscMcZIkSZI0QQxxkiRJkjRBDHGSJEmSNEEMcZIk\nSZI0QQxxkiRJkjRBDHGSJEmSNEEMcZIkSZI0QQxxkiRJkjRBDHGSJEmSNEEMcZIkSZI0QQxxkiRJ\nkjRBDHGSJEmSNEEMcZIkSZI0QQxxkiRJkjRBDHGSJEmSNEEMcZIkSZI0QQxxkiRJkjRBDHGSJEmS\nNEEMcZIkSZI0QQxxkiRJkjRBDHGSJEmSNEEMcZIkSZI0QQxxkiRJkjRBDHGSJEmSNEEMcZIkSZI0\nQQxxkiRJkjRBDHGSJEmSNEEMcZIkSZI0QQxxkiRJkjRBDHGSJEmSNEEMcZIkSZI0QQxxkiRJkjRB\nDHGSJEmSNEEMcZIkSZI0QQxxkiRJkjRBDHGSJEmSNEHmDHFJVif5RJKvJ9mc5OgkG5JsTXJ1+3N8\nT/uzk1yfZEuSY4dbviRJkiQtL/MZiXs/8JmqejbwA8DXgQLOqaoXtD+fBUiyDjgZWAccB5ybxNE+\nTYwk67uuQZqN16fGldemxpnXp5aivgEryVOAY6rqfICq2lFV9009PcNLTgQ2VtX2qroJuAE4aoD1\nSsO2vusCpD7Wd12ANIv1XRcg9bG+6wKkQZtrlGwtcGeSjya5KsmHk6xqnzsjyTVJzkuyuj23P7C1\n5/VbgQMGXLMkSZIkLVtzhbgVwBHAuVV1BPAAcBZwLk3AOxy4DXhfnz5qAHVKkiRJkoBUzZ6xkuwH\nfKmq1rbHLwHOqqpX9bQ5GLi0qp6X5CyAqvrt9rm/BN5VVV+e1q/BTpIkSdKyVlUz3aI2pxVzdLot\nyS1JDquqbwCvBK5Lsl9VbWubvQa4tn18CXBhknNoplEeClw5qGIlSZIkabnrG+JaZwAfS7IS+Cbw\nRuD3khxOM1XyRuA0gKranOQiYDOwAzi9+g31SZIkSZIWpO90SkmSJEnSeBnpHm5Jjms3Ab8+yTtG\n+d7SdEkOSvLXSa5L8rUkb23P75Xk8iTfSHJZz+qr0kgl2S3J1UkubY+9NjUWkqxO8okkX0+yOcmL\nvD41DpKc3X6vX5vkwiS7e22qC0nOT3J7kmt7zs16LbbX7vVtVjp2rv5HFuKS7AZ8gGYT8HXATyV5\n9qjeX5rBduBtVfUc4GjgF9pr8izg8qo6DPhceyx14Uya6elTUya8NjUu3g98pqqeDfwAsAWvT3Ws\nXWzvLcARVfU8YDfgFLw21Y2P0uSeXjNei0nWASfTZKTjgHOT9M1poxyJOwq4oapuqqrtwMdpNgeX\nOlFV26rqq+3j7wBfp1mQ5wTggrbZBcCru6lQy1mSA4EfBT4CTC0G5bWpziV5CnBMVZ0PUFU7quo+\nvD7VvW/T/IF2VZIVwCrgW3htqgNV9bfAv047Pdu1eCKwsaq2V9VNwA002WlWowxxBwC39By7EbjG\nRvvXuxcAXwb2rarb26duB/btqCwtb/8T+GVgZ885r02Ng7XAnUk+muSqJB9O8kS8PtWxqrqHZu/i\nm2nC271VdTlemxofs12L+9Nkoylz5qRRhjhXUNFYSvIk4JPAmVV1f+9z7eqqXrsaqSSvAu6oqqt5\nZBTuUbw21aEVwBHAuVV1BPAA06aneX2qC0meCfwicDDNfxQ/KclP97bx2tS4mMe12Pc6HWWIuxU4\nqOf4IB6dOKWRS/J4mgD3J1V1cXv69naje5KsAe7oqj4tWz8InJDkRmAj8Iokf4LXpsbDVmBrVf1D\ne/wJmlAck04bAAABZklEQVS3zetTHXsh8MWquruqdgCfAl6M16bGx2zf49Nz0oHtuVmNMsR9BTg0\nycHtnnMn02wOLnUiSYDzgM1V9bs9T10CnNo+PhW4ePprpWGqqndW1UFVtZbmpvzPV9XP4LWpMVBV\n24BbkhzWnnolcB1wKV6f6tYW4Ogke7Tf8a+kWRzKa1PjYrbv8UuAU5KsTLIWOBS4sl9HI90nLsnx\nwO/SrBZ0XlX91sjeXJomyUuAvwH+iUeGrM+m+ZfmIuDpwE3ASVV1bxc1SkleBry9qk5IshdemxoD\nSZ5Ps+jOSuCbwM/RfLd7fapTSX6F5j+OdwJXAW8G9sRrUyOWZCPwMuBpNPe//Srw58xyLSZ5J/BG\nYAfNLT5/1bd/N/uWJEmSpMkx0s2+JUmSJEmLY4iTJEmSpAliiJMkSZKkCWKIkyRJkqQJYoiTJEmS\npAliiJMkSZKkCWKIkyRJkqQJYoiTJEmSpAny/wFpf1vP/y+2DAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x83464e0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA3EAAAFICAYAAAD+nncAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x82be160>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "w = interact(genetic_algorithm,\n",
    "             minimum = (-512, 0, 1),\n",
    "             maximum = (0, 512, 1),\n",
    "             number_of_genes = (1, 50, 1),\n",
    "             size_of_society = (1, 100, 1), \n",
    "             reproduction_rate = (0, 1, 0.01), \n",
    "             mutation_rate = (0, 1, 0.01),\n",
    "             number_of_generations = (1, 200, 1),\n",
    "             size_after_tournament = (1, 100, 1)\n",
    "            )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "El tamaño del genoma, esto es, el número de bits utilizados para describir al individuo, determina la cantidad de soluciones posibles, por lo que abre la puerta a mejores optimizaciones. Un genoma mayor supone asimismo una mayor dificultad para converger, debido a que hay muchas más combinaciones posibles a probar."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "El tamaño de la población es, probablemente el factor más importante. El tiempo de cálculo aumenta de forma directamente proporcional al tamaño de la población, por lo que hay que tener cuidado al elegir el tamaño apropiado.\n",
    "\n",
    "Una población pequeña, aunque mejora los tiempos, puede ser víctima del llamado “efecto del fundador”, en los que una baja variabilidad genética en el origen y, en general, un tamaño poblacional pequeño, hacen que la “deriva genética” lleve a la población a presentar una serie de rasgos comunes que no necesariamente se corresponden con los de los individuos más aptos, que sí se observan fuera de ella."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Para evitar este problema, se pueden introducir altas tasas de inmigración y mutación. Si la población es pequeña, es muy sensible, por lo que la media de la función de aptitud evoluciona de manera muy turbulenta y puede, incluso, alejarse de la convergencia"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Por el contrario, una población grande es garantía de una gran variedad genética inicial. Estadísticamente, también es más probable que se encuentre antes la solución, aunque la convergencia sea más lenta. Además, el gran número estabiliza la evolución de la función de mérito media y hace que la población sea más robusta frente a la inmigración y la mutación."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Si te está gustando este taller:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<a href=\"https://twitter.com/share\" class=\"twitter-share-button\" data-url=\"https://github.com/AeroPython/Taller-Algoritmos-Geneticos-PyConEs16\" data-text=\"Aprendiendo Algoritmos Genéticos y Sistemas Complejos con\" data-via=\"AeroPython\" data-size=\"large\" data-hashtags=\"PyConEs\" data-dnt=\"true\">Tweet</a>\n",
       "<script>!function(d,s,id){var js,fjs=d.getElementsByTagName(s)[0],p=/^http:/.test(d.location)?'http':'https';if(!d.getElementById(id)){js=d.createElement(s);js.id=id;js.src=p+'://platform.twitter.com/widgets.js';fjs.parentNode.insertBefore(js,fjs);}}(document, 'script', 'twitter-wjs');</script>"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%%html\n",
    "<a href=\"https://twitter.com/share\" class=\"twitter-share-button\" data-url=\"https://github.com/AeroPython/Taller-Algoritmos-Geneticos-PyConEs16\" data-text=\"Aprendiendo Algoritmos Genéticos y Sistemas Complejos con\" data-via=\"AeroPython\" data-size=\"large\" data-hashtags=\"PyConEs\" data-dnt=\"true\">Tweet</a>\n",
    "<script>!function(d,s,id){var js,fjs=d.getElementsByTagName(s)[0],p=/^http:/.test(d.location)?'http':'https';if(!d.getElementById(id)){js=d.createElement(s);js.id=id;js.src=p+'://platform.twitter.com/widgets.js';fjs.parentNode.insertBefore(js,fjs);}}(document, 'script', 'twitter-wjs');</script>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Carlos Dorado, Aeropython, 7 de Octubre de 2016"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<link href='http://fonts.googleapis.com/css?family=Source+Sans+Pro|Josefin+Sans:400,700,400italic|Ubuntu+Condensed' rel='stylesheet' type='text/css'>\n",
       "\n",
       "\n",
       "<style>\n",
       "\n",
       "#notebook_panel { /* main background */\n",
       "    background: #f7f7f7;\n",
       "}\n",
       "\n",
       "div.cell { /* set cell width */\n",
       "    width: 900px;\n",
       "}\n",
       "\n",
       "div #notebook { /* centre the content */\n",
       "    background: #fff; /* white background for content */\n",
       "    width: 950px;\n",
       "    margin: auto;\n",
       "    padding-left: 0em;\n",
       "}\n",
       "\n",
       "#notebook li { /* More space between bullet points */\n",
       "    margin-top:0.7em;\n",
       "}\n",
       "\n",
       "/* draw border around running cells */\n",
       "div.cell.border-box-sizing.code_cell.running { \n",
       "    border: 1px solid #111;\n",
       "}\n",
       "\n",
       "/* Put a solid color box around each cell and its output, visually linking them*/\n",
       "div.cell.code_cell {\n",
       "    font-family: 'Source Sans Pro', sans-serif;\n",
       "    background-color: rgb(256,256,256);\n",
       "    font-size: 110%;\n",
       "    border-radius: 0px; \n",
       "    padding: 0.5em;\n",
       "    margin-left:1em;\n",
       "    margin-top: 1em;\n",
       "}\n",
       "\n",
       "div.text_cell_render{\n",
       "    font-family: 'Josefin Sans', serif;\n",
       "    line-height: 145%;\n",
       "    font-size: 125%;\n",
       "    font-weight: 500;\n",
       "    width:750px;\n",
       "    margin-left:auto;\n",
       "    margin-right:auto;\n",
       "}\n",
       "\n",
       "\n",
       "/* Formatting for header cells */\n",
       ".text_cell_render h1, .text_cell_render h2, .text_cell_render h3,\n",
       ".text_cell_render h4, .text_cell_render h5 {\n",
       "    font-family: 'Ubuntu Condensed', sans-serif;\n",
       "}\n",
       "\n",
       ".text_cell_render h1 {    /*Use this for Title*/\n",
       "    font-weight: 500;\n",
       "    font-size: 38pt;\n",
       "    line-height: 100%;\n",
       "    color: #ffbc29;\n",
       "    text-align: center;\n",
       "    margin-bottom: 0.1em;\n",
       "    margin-top: 0.3em;\n",
       "    display: block;\n",
       "}\n",
       "\n",
       ".text_cell_render h2 {    /*Use this for Subtitle*/\n",
       "    margin-top:16px;\n",
       "    font-size: 32pt;\n",
       "    font-weight: 500;\n",
       "    margin-bottom: 0.1em;\n",
       "    margin-top: 0.3em;\n",
       "    text-align: center;\n",
       "    font-style: regular;\n",
       "    color: #376a94;\n",
       "}\t\n",
       "\n",
       ".text_cell_render h3 {   /*Sections*/ \n",
       "    font-size: 30pt;\n",
       "    font-weight: 300;\n",
       "    text-align: left;\n",
       "    margin-bottom: 0.1em;\n",
       "    margin-top: 0.3em;\n",
       "    font-style: regular;\n",
       "    color:  #252525;\n",
       "}\n",
       "\n",
       ".text_cell_render h4 {    /*Subsections*/\n",
       "    font-size: 28pt;\n",
       "    font-weight: 200;\n",
       "    text-align: left;\n",
       "    margin-bottom: 0.1em;\n",
       "    margin-top: 0.3em;\n",
       "    font-style: regular;\n",
       "    color:  #376a94;\n",
       "}\n",
       "\n",
       ".text_cell_render h5 {    /*Subsubsections*/\n",
       "    font-size: 20pt;\n",
       "    font-weight: 300;\n",
       "    font-style: italic;\n",
       "    margin-bottom: .1em;\n",
       "    margin-top: 0.8em;\n",
       "    display: block;\n",
       "    color:  #ffbc29;\n",
       "}\n",
       "\n",
       ".text_cell_render h6 {    /*Author*/\n",
       "    font-family: 'Ubuntu Condensed', sans-serif;\n",
       "    font-weight: 100;\n",
       "    font-size: 14pt;\n",
       "    line-height: 100%;\n",
       "    color: #252525;\n",
       "    text-align: right;\n",
       "    margin-bottom: 1px;\n",
       "    margin-top: 3px;\n",
       "}\n",
       "\n",
       ".CodeMirror{\n",
       "        font-family: 'Duru Sans', sans-serif;\n",
       "        font-size: 100%;\n",
       "}\n",
       "\n",
       "</style>\n",
       "<script>\n",
       "    MathJax.Hub.Config({\n",
       "                        TeX: {\n",
       "                           extensions: [\"AMSmath.js\"],\n",
       "                           equationNumbers: { autoNumber: \"AMS\", useLabelIds: true}\n",
       "                           },\n",
       "                tex2jax: {\n",
       "                    inlineMath: [ ['$','$'], [\"\\\\(\",\"\\\\)\"] ],\n",
       "                    displayMath: [ ['$$','$$'], [\"\\\\[\",\"\\\\]\"] ]\n",
       "                },\n",
       "                displayAlign: 'center', // Change this to 'center' to center equations.\n",
       "                \"HTML-CSS\": {\n",
       "                    styles: {'.MathJax_Display': {\"margin\": 4}}\n",
       "                }\n",
       "        });\n",
       "</script>\n"
      ],
      "text/plain": [
       "<IPython.core.display.HTML object>"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Notebook style\n",
    "from IPython.core.display import HTML\n",
    "css_file = './static/style.css'\n",
    "HTML(open(css_file, \"r\").read())"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.4.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}