MOEA-D.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pymoo"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Define the Traveling Salesman problem"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from motsp import MOTSP\n",
    "n_obj = 2\n",
    "problem = MOTSP(100, n_obj) # 100 cities, 2 objectives"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Initialize algorithms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pymoo.factory import get_sampling, get_crossover, get_mutation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Initialize NSGA2 algorithm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pymoo.algorithms.nsga2 import NSGA2\n",
    "nsga2 = NSGA2(\n",
    "    pop_size=100,\n",
    "    n_offsprings=100,\n",
    "    sampling=get_sampling(\"perm_random\"),\n",
    "    crossover=get_crossover(\"perm_erx\", prob=0.9),\n",
    "    mutation=get_mutation(\"perm_inv\"),\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Initialize MOEA-D algorithm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pymoo.algorithms.moead import MOEAD\n",
    "from pymoo.factory import get_visualization, get_reference_directions\n",
    "from pymoo.factory import get_sampling, get_crossover, get_mutation\n",
    "\n",
    "# First, divide our problem in subproblems:\n",
    "ref_dirs = get_reference_directions(\"energy\", n_obj, 100) # contains weights to define subproblems\n",
    "\n",
    "moead = MOEAD(\n",
    "    ref_dirs = ref_dirs,\n",
    "    n_neighbors=20, # new: n_neighbors is used in the algorithm for reproduction essentially\n",
    "    decomposition='tchebi', # new: it is the algorithm used to decompose the problem into subproblems\n",
    "                            #.     the decomposition is based on weights in ref_dirs\n",
    "    prob_neighbor_mating=.9, # new: the probability to take a solution of a neighbor subproblem in reproduction step\n",
    "    sampling=get_sampling(\"perm_random\"),\n",
    "    crossover=get_crossover(\"perm_erx\", prob=.9),\n",
    "    mutation=get_mutation(\"perm_inv\"),\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Run both algorithms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pymoo.optimize import minimize"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "res_nsga2 = minimize(problem, nsga2, termination=('n_eval', 10000))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "res_moead = minimize(problem, moead, termination=('n_eval', 10000))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Plot results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<pymoo.visualization.scatter.Scatter at 0x7f8bf0a8e790>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot = get_visualization('scatter')\n",
    "plot.add(res_nsga2.F, label='NSGA2')\n",
    "plot.add(res_moead.F, label='MOEA-D')\n",
    "plot.legend=True\n",
    "plot.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Questions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-warning\">\n",
    "    <b>Question 1:</b>\n",
    "\n",
    "What does ref_dirs refers to?\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "ref_dirs = get_reference_directions(\"energy\", n_obj, 100) # contains weights to define subproblems"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-danger\"><a href=\"#hint1\" data-toggle=\"collapse\">Hint (click to expand)</a><br>\n",
    "    <div id=\"hint1\" class=\"collapse\">\n",
    "        Get its shape, and then try to plot it with matplotlib.pyplot.scatter.\n",
    "    </div>\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# %load solutions/question1.py\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-warning\">\n",
    "    <b>Question 2:</b>\n",
    "    \n",
    "Do we need to specify pop_size when initializing MOEA-D algorithm? Why?\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# %load solutions/question2.py\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-warning\">\n",
    "    <b>Question 3:</b>\n",
    "    \n",
    "Compare MOEA-D with NSGA-II, given the same termination criterion n_eval = cst.\n",
    "    \n",
    "You can also try with <a href=\"http://pymoo.org/interface/termination.html\">other termination criterions</a>.\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# %load solutions/question3.py\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-warning\">\n",
    "    <b>Question 4:</b>\n",
    "    \n",
    "How the results of MOEA-D depend on prob_neihbor_mating parameter? Why?\n",
    "</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# %load solutions/question4.py\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "myenv_python_3.8",
   "language": "python",
   "name": "myenv_python_3.8"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}