demo_customize_your_own_function.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# ACML2020 Tutorial: Recent Advances in Bayesian Optimization\n",
    "# http://vu-nguyen.org/BayesOptTutorial_ACML20\n",
    "\n",
    "\n",
    "# Dr Vu Nguyen vu@ieee.org\n",
    "\n",
    "# Let define our own black-box function to be optimized with MiniBO"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from mini_bo import BayesOpt,visualization\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from tabulate import tabulate\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Customize your own function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "class YourFunction:\n",
    "    def __init__(self):\n",
    "        self.bounds = np.asarray([         # define the search range for each variable\n",
    "                                [-5,10], # variable 1: min=-5 ; max=10\n",
    "                                  [0,15] # variable 2: min=0; max=15\n",
    "                                 ])\n",
    "            \n",
    "        self.input_dim = self.bounds.shape[0] # this is 2\n",
    "\n",
    "        # do we want to maximize the function or minimize ?\n",
    "        self.ismax=-1  # set -1 if we want to minimize\n",
    "        \n",
    "        self.name='MiniBO'         # set the name of your function\n",
    "        \n",
    "    # evaluate y=f(X)\n",
    "    def evaluate_single_fx(self,X): # this is actually a Branin function\n",
    "        X = np.reshape(X,self.input_dim)\n",
    "        x1,x2=X[0],X[1]\n",
    "     \n",
    "        a,b=1,5.1/(4*np.pi**2)\n",
    "        c,r,s=5/np.pi,6,10\n",
    "        t=1/(8*np.pi)\n",
    "        y=a*(x2-b*x1*x1+c*x1-r)**2+s*(1-t)*np.cos(x1)+s    \n",
    "      \n",
    "        return y*self.ismax # return the (-1) * fx for minimization problem\n",
    "    \n",
    "    def func(self,X):\n",
    "        X=np.asarray(X)        \n",
    "       \n",
    "        if len(X.shape)==1: # 1 data point\n",
    "            fx=self.evaluate_single_fx(X)\n",
    "        else: # multiple data points\n",
    "            fx=np.apply_along_axis( self.evaluate_single_fx,1,X)\n",
    "            \n",
    "        return fx  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Init the black-box function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "myfunction=YourFunction()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Specify Bayesian optimization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Lenovo\\AppData\\Local\\conda\\conda\\envs\\py36\\lib\\site-packages\\sklearn\\utils\\validation.py:595: DataConversionWarning: Data with input dtype int32 was converted to float64 by MinMaxScaler.\n",
      "  warnings.warn(msg, DataConversionWarning)\n"
     ]
    }
   ],
   "source": [
    "acq_name='ei'\n",
    "bo=BayesOpt(myfunction.func,myfunction.bounds,acq_name=acq_name,verbose=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Initialize BO using 3*dim number of observations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "bo.init(n_init_points=3*myfunction.input_dim)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Run for 10*dim iterations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "estimated lengthscale [0.0353949]\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "     0  [-0.862  5.967]          -18.501                -18.501\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "     1  [-1.818  7.607]          -10.586                -10.586\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "     2  [-2.858 10.062]           -3.158                 -3.158\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "     3  [-4.881 10.904]          -46.923                 -3.158\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "     4  [-1.63   9.679]           -9.986                 -3.158\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "     5  [4.21  7.399]          -39.126                 -3.158\n",
      "estimated lengthscale [0.0417672]\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "     6  [-2.428 12.466]           -6.132                 -3.158\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "     7  [-2.241 15.   ]          -26.926                 -3.158\n",
      "  Iter  Selected x      Output y=f(x)    Best Observed Value\n",
      "------  ------------  ---------------  ---------------------\n",
      "     8  [10.  0.]             -10.961                 -3.158\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "     9  [7.471 0.   ]          -15.332                 -3.158\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    10  [9.046 2.372]           -1.117                 -1.117\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    11  [6.881 3.052]          -21.499                 -1.117\n",
      "estimated lengthscale [0.04193778]\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "    12  [10.     4.575]           -4.414                 -1.117\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "    13  [10.     7.691]          -23.925                 -1.117\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "    14  [10.    12.868]          -99.259                 -1.117\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    15  [8.283 5.527]          -20.801                 -1.117\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    16  [3.591 0.   ]           -5.156                 -1.117\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "    17  [10.     2.538]           -2.159                 -1.117\n",
      "estimated lengthscale [0.04712622]\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    18  [2.647 1.937]           -2.121                 -1.117\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    19  [1.763 0.207]          -19.649                 -1.117\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    20  [4.007 1.537]           -3.798                 -1.117\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "    21  [ 4.678 15.   ]         -195.118                 -1.117\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    22  [3.139 4.024]           -3.449                 -1.117\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    23  [1.131 3.555]           -14.74                 -1.117\n",
      "estimated lengthscale [0.0706129]\n",
      "  Iter  Selected x      Output y=f(x)    Best Observed Value\n",
      "------  ------------  ---------------  ---------------------\n",
      "    24  [-5. 15.]             -17.508                 -1.117\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    25  [3.221 2.883]           -0.876                 -0.876\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    26  [9.278 3.498]            -1.81                 -0.876\n",
      "  Iter  Selected x       Output y=f(x)    Best Observed Value\n",
      "------  -------------  ---------------  ---------------------\n",
      "    27  [7.257 9.254]          -79.407                 -0.876\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "    28  [-3.188  8.578]           -14.91                 -0.876\n",
      "  Iter  Selected x         Output y=f(x)    Best Observed Value\n",
      "------  ---------------  ---------------  ---------------------\n",
      "    29  [-3.679 13.355]           -1.814                 -0.876\n",
      "==================================================================================\n",
      "Best found x      Best found f(x)\n",
      "--------------  -----------------\n",
      "[3.221 2.883]              -0.876\n"
     ]
    }
   ],
   "source": [
    "NN=15*myfunction.input_dim\n",
    "for index in range(0,NN):\n",
    "    bo.select_next_point()\n",
    "\n",
    "    print(tabulate([[ index,np.round(bo.X_ori[-1],3), np.round(bo.Y_ori[-1],3), np.round(bo.Y_ori.max(),3)]], \\\n",
    "               headers=['Iter','Selected x', 'Output y=f(x)', 'Best Observed Value']))\n",
    "\n",
    "print(\"==================================================================================\")\n",
    "idxMax=np.argmax(bo.Y_ori)\n",
    "print(tabulate([[ np.round(bo.X_ori[idxMax],3), np.round(bo.Y_ori[idxMax],3)]], \\\n",
    "           headers=['Best found x', 'Best found f(x)']))   "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Plot the performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Performance')"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAADmCAYAAAAk/am7AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+j8jraAAAgAElEQVR4nO3de5xVVf3/8dd7uAwggngBYQBxQMkyMxot0szyin1/oqWW2c1+RaZm2bcypdRKft++punXTJO+mVmYaXnNS2plVmKK5DWkhLjjDIhyERiGmc/vj70HDsOZOWdmzuGcM7yfj8d5cPbe6+z9YcOcz6y19lpLEYGZmVlHqkodgJmZlT8nCzMzy8nJwszMcnKyMDOznJwszMwsJycLMzPLycnCKoakT0mKjNdaSc9KOldS7wJdYxdJP5fUkF7j6kKc16zSFeQHzGwHOxVYAgxK3/8AGApcXIBznwOcDnwa+CewvADnNKt4ThZWiZ6JiJfT9w9JGgd8iW4kC0nVEdEIHAAsi4ibCxBn5nnNKpqboawneArYVdJQSW+TdI+k1yRtkPRXSe/JLCzpJklLJE2U9LikDcDlkgL4FDAqo6nryPQz4yXdKen19LxPSDq+zXkvTT9zoKTfSVoH3JYeC0mXSfpPSQslvSHpvjTmoZJuk7Ra0mJJF7Q5716SbpD0T0nr0zK3SKpp5/r7pedel17rYklVWc55XXquxvTPn0uqziiT817azsPJwnqCfYFmYBzwOLA78FngQ8CrwCOS3tHmM4OBW4FfApOAW4CJwO+AV9L3E4HZkkYAfwHeBpwLnAa8DtwnaVKWeO4G/gScCFyVsf/jwPuBs4EvAO8BbgbuBJ5L470f+K6kEzI+tzuwEbgQOB74KrAf8FdJ/bJc/07gD8BJwF3At4BPth6UNCS9Tx8Gvg+cAHwN6AP0TctM6MS9tJ1BRPjlV0W8SH7rD2A8SRPqEOBzJIniLuD3wBygb8ZneqX77srYd1N6nslZrvELYEGbfVcAm4Fxbc47F5idse/S9LxfzHLeIOkD6Z2x7/vp/m9k7OsNNAA/7eA+9AJGpZ89Ocv1z2xT/nngoYztb6f37O0dXCOve+nXzvNyzcIq0UtAE7AKuA6YQdIx/V7gdqBFUu/0CSkBjwBHtDnHZuC3eV7vCOCJ2NpPQkQ0k9RKDpY0qE35O9s5z8MRsbnN3wOS2kzreTcDL5Mkgy0kfT598mtdGvui9ND4LNe5r832C8DojO1jgaci4u/ZgpTUn87dS9sJuIPbKtHJJE9DrQUWRsTGtP2+F/DN9LUdSVUR0ZJuNqRf+PnYHcj2xfoKyRfoEGBNxv72nqB6rc32pg72b2lekvQF4BqSmshX0/JVwBOZ5TKsarPd2KbcHsCz7cQIyd+3M/fSdgJOFlaJXsj8LT/1OtAC/JCkH2A7bb7cOjM3/ypg7yz7907P0/bLudDz/n8E+H1E/GfrDkn7duN8K4GaDo539l7aTsDJwnqEiHhD0p9JOqFnF/jL7E/AlySNiYgFAJJ6kXQQ/z0i1hbwWtkMYNuaC8CZ3TjfQ8A3JL0tIrarYRT5XlqFcrKwnuTLwGPA7yT9hKQ5aE9gAtArIr7exfNeRdK5/rCkS0i+uM8G9gc+0N2g8/AgcIGki4AnSZ6oOqUb57sK+CjJk02XkXSA7wlMBs5Kk1+x7qVVKCcL6zEiYrakQ4BLSNr4BwMrgNnAj7px3mWSDgf+G7geqAaeAT4QEQ92O/Dcvg3sBpxP0vfwJ+A4YH5XThYRr0s6DLgM+DpJH0Y9yeO2m9IyRbmXVrkU4WVVzcysY3501szMcnKyMDOznJwszMwsJycLMzPLycnCzMxy6pGPzu65554xZsyYUodhZlZRnn766ZURsVe2YxWTLNK1A/6HZM6a/42I77ZXdsyYMcyaNWuHxWZm1hNIWtjesYpohkqnVvghyboDbwZOl/Tm0kZlZrbzqIhkARwKvBwR8yNiE8miNZNLHJOZ2U6jUpJFDbA4Y3sJHc+aaWZWdPUz6pk5ZiaPVj3KzDEzqZ9R3+kyhThHvmW6o1KShbLs22aeEklTJM2SNGvFihU7KCzbWZXLF0BPu045xZLP8blT5tK4sBECGhc2MnfK3G3K5SpTiHPkW6a7KmJuKEkTgUsj4rh0+0KAiPivbOXr6urCHdxWLK0/mC3rt87crWox+uuj2f3Y3Vn10CoWfXcR0RhZjwMFKdPTrlNOsWQ93lfscdIe9B/dn6ZXm2i4tYGWDVlmb+8F/UYna01tXLQxWcC2nTK5judzjo7KVO9TzcQFE7N8ODtJT0dEXdZjFZIsepOsX3wUsBR4CvhoRLyYrbyThRXT4zWPs2nZptwFzUpNcGTLkfkX7yBZVEQzVLou8bkkaxXPAW5rL1GYFUtLYwsLvrWgw0QxaGLb5bi3P16oMj3tOuUUS0dqv1vL/tP3p8+efbIe7zuiL++c907eOe+d9B3Rt8MyuY53t0z16OoO/y6dURHJAiAi7o+I/SNibERMK3U8tnN57dHXeOqgp1hw6YJ2y1TvU82ExydQvU/2H9DW44Uq09OuU06xdHR89AWjGfHZEYy7ehxVA7b9Cq0aUMXYy8fSv7Y//Wv7M/bysR2WyXW8u2Vqp9Vm/Xt0RcUkCyuNntQh2aUyo2Yy+4jZPPu+Z9nwzw0MeNMARn9jdIc/mLXTanP+4BaiTE+7TjnFks85hp0xjPHTxyeJRUkiGT99PMPOGJZ3mUKcI98y3VURfRad5T6LwsjWkVs1oGqb/4S5yhTiHKW+DgC9YMylYxj91dFUVVdRP6Oe+VPn07iokerR1dROq93mBzPX8UKV6WnXKadY8jlHT1PxHdyd5WRRGDPHzEwexWtDfcXAtw0EYN2z64hN2/8fai2T63g+5yiH6/Qd0Zd3L333dvvNehInC+uSR6sebTOaZSfWyadKzCpRxT8NZaXR3pMUffbuw4S/TWDC3ybQZ+/sT4S0lsl1vFBlin2dQj5VYlaJnCysXSO/NHK7fVUDqhh3xTgGHTqIQYcOYtwV2Z8IaS2T63ihyhT7OoV8qsSsEjlZWLsaFyX9FVW7VJX90x6V9FSJWSVyn4VltXnNZmaOnEnz2mbe8fQ72HXCrqUOycyKzH0W1mmv/PQVmtc2M/iIwU4UZuZkYduL5mDJNUuA7P0WZrbzcbKw7ay8dyUb52+k37792PPEPUsdjpmVAScL286Sq5NaRc15NahXtqVEzGxn42Rh21j797Ws/tNqeu3ai+GfHl7qcMysTOSdLCRNkvRbSf+QNCrd9xlJRxUvPNvRWmsVw//vcHoP6l3iaMysXOSVLCSdAdwG/AvYF2gd5toL+FpxQrMdrXF5Iw2/bICqpAnKzKxVvjWLrwGfjYjzgc0Z+58ADi54VFYSy65fRjQFe07ek/779i91OGZWRvJNFvsBM7PsXwd0vKyUVYTmDc0su34ZACPP9+OyZratfJPFMmD/LPuPAOYVLhwrlYZbGmha2cTACQMZfPjgUodjZmUm32QxHbhG0mHp9ihJnwQuB67vbhCSTpX0oqQWSXVtjl0o6WVJcyUd191r2fYigsVXLQaSWoXkx2XNbFt5Pe4SEZdLGgw8DPQD/gg0AldExA8LEMcLwAeBGzJ3Snoz8BHgLcAI4BFJ+0dEcwGuaanXfv8a619cT9/hfRl62tBSh2NmZSjvZyMjYqqkacCbSWok/4iIdYUIIiLmANl+o50M3BoRjcC/Jb0MHEr2/hPrpC3LRqar4Q06bBBVfT30xsy216lvhohYHxGzIuLJQiWKHGqAxRnbS9J925E0RdIsSbNWrFixA0KrbK1rTWcum7rqvlXUz6gvYVRmVq7yqllIuqej4xFxYh7neATYO8uhqRFxd3sfy3a5dmKYTtK3Ql1dXc+bd73A5k+dT8v6lm32tWxoYf7U+V67wcy2k28z1KtttvsAbwNGAXfkc4KIOLoTcbVakl6j1UiSJ7Osm1oXNsp3v5nt3PLt4D4z235JVwJrCxrRtu4BbpH0fZIO7v2AJ4t4vZ1G9ejqbZqgMvebmbXV3d7MG4CzuxuEpJMlLQEmAvdJ+h1ARLxIMs3IP4AHgXP8JFRh7Dtt3+0a+bzWtJm1p7szxY0vRBARcSdwZzvHpgHTCnEd26rf6H5J708VEEmNonZarfsrzCyrfDu4r2m7CxgOTAJuLHRQVnytU3uMvnA0tZe5NmFmHcu3ZvHWNtstwArgfJwsKs6m+k2s+PUKqIIRU0aUOhwzqwD5dnC/r9iB2I6z/CfLiaZgjxP3SJqjzMxy8HDdnUw0B8tuSJqgas72mhVmlp92axa5BuJlymdQnpWHV+9/lcZFjfQb248hxwwpdThmViE6aoZqOxDPeoDWju0RZ41AVZ5d1szy026yaG8gnlWuDfM2sOrBVahaDD9zeKnDMbMK4j6LnciyG5ZBwNAPD6XPHn1yf8DMLJX3oDxJ7wNOB0YDfTOPRcT7CxyXFVjzxmaW37gccMe2mXVeXjULSZ8CHgB2BY4kGWMxBJhAMhWHlbkVt69g86ubGThhILseumupwzGzCpNvM9RXgHMj4nSgCbgwIt4O/ALYEetaWDdt6dj+/Agvm2pmnZZvsqgFHknfNwID0/fXAp8qcExWYGufWcuamWvoNbgXw0733E9m1nn5JotXSZqgAJYCB6bv9wD6FzooK6zWWsXen9ybXrv0KnE0ZlaJ8u3g/jNwLPA8yZTh10g6BjgKeLhIsVkBbF69mfpfJEuljvi854Eys67psGYhaVz69lzgl+n7/wK+R1KruA34TNGis26pn1HPE/s+Qcv6FlQt1j3t7iUz65pcNYt/SnoM+DHwG4CIaAH+u9iBWffUz6hn7pS5W9bZjsZg7pS5AF6zwsw6LVefxX+QPCZ7I7Bc0rWSDi5+WNZd86fO35IoWrWsb2H+1PklisjMKlmHySIi7o+IU4Ea4DvAe4GnJc2W9HlJg3dEkNZ5jYu2X1+7o/1mZh3J62moiFgZEd+PiLcC7wZmAd8Flkm6ubtBSPqepJckPSfpTkm7ZRy7UNLLkuZKOq6719pZVI+szr5/dPb9ZmYd6fTcUBHxN+BzwBRgI3BGAeJ4GDgwIg4C/glcCCDpzcBHgLcAxwPXSfKzn3nY4+Q9tttXNaCK2mleQtXMOq9TyULSOEn/D1hMMnr7CeCU7gYREQ9FxOZ08wlgZPp+MnBrRDRGxL+Bl4FDu3u9nUHjwqS5qfeQ3iCo3qea8dPHu3PbzLok5zgLSf2B04BPA4cDC4AfAT+NiKVFiOnTwK/S9zUkyaPVknRftjinkNR2GD16dBHCqhxNq5pYdf8qqIJDXjyE6uFuejKz7ukwWUiaDnyYZJbZu4BjI+L3XbmQpEeAvbMcmhoRd6dlpgKbgRmtH8tSPrKdPyKmA9MB6urqspbZWaz49QqiKRhy9BAnCjMriFw1i3cC3wR+HhGvdedCEXF0R8clfZLkUd2jIqL1y34JMCqj2EhgWXfi2BnUz0hGbA89Y2iJIzGznqLDZBERb9sRQUg6HrgAeG9ErM84dA9wi6TvAyOA/YAnd0RMlWrjoo2sfmw1Vf2q2OuDe5U6HDPrIfJe/KjIrgWqgYfT6bOfiIizIuJFSbeRrJmxGTgnIppLGGfZa/hlAwB7nLgHvQeVyz+vmVW6svg2iYhxHRybBkzbgeFUtNYmKD/1ZGaF5DW4e5B1z6/jjeffoPfuvdn9+N1LHY6Z9SBOFj1Ia61ir1P3oqqv/2nNrHDabYaSlPdghYhYVJhwrKuiJWi4JemvcBOUmRVaR30WC2hnTEMWnoKjxFb/ZTWNixupHl3N4MM8v6OZFVZHyeKQjPf7A5eTjNyeme6bSDJH1AXFCc06Y0vH9keHoapsYxnNzLqu3WQREU+3vk/HOZwfEb/OKPIHSXOBL7J1FT0rgZZNLay4fQXggXhmVhz59oIeCjyXZf9zwDsKF451xaoHVrH5tc3sctAuDDxwYKnDMbMeKN9ksQA4O8v+s4GFBYvGusRjK8ys2PIdlHc+cGc6LUfrLLDvBMYAHyxCXJanzWs28+q9r4Jg6OlugjKz4sh3pbwHSeZlugMYBAxO3+8fEQ8ULzzLZcUdK2jZ2MLgIwbTb1S/UodjZj1U3tN9RMQS4KIixmJd0DDDYyvMrPjyThaSBgAHA0NpUyOJiDsKHJfloXF5I6/94TXUV+x1imeYNbPiyStZSDqa5PHY7Rd2TgbueVBeAdXPqGf+1Pk0LkoG2dVOq92u5lA/o55/feFf0ALqJVbdv8q1CzMrmnxrFv8D3AdcFBFefKiI6mfUM3fKXFrWtwDJWtovnfkSK+9dya4TdgVg7ey1rLxjJdGUDLBv2dDC3ClzATdHmVlxaOuidB0Ukt4ADoqIecUPqfvq6upi1qxZpQ6j09Y9u46/v+fvNK/t2pId1ftUM3HBxAJHZWY7C0lPR0RdtmP51iz+CowHKiJZlLvMZqa+I/oy+D2DWf/iet54/o0OPzfqq8kKs4u/tzjr8cZFjQWP1cwM8k8WPwKukDQCeB5oyjwYEbMLHVhP1baZadPSTay4NZmqo/fuvYlNQfO67WsW1ftUM/bysQA03NZA48LtE0P16OoiRm5mO7N8R3D/GngTMJ1kIsFZGa+nuhuEpO9Iek7SM5IeSpMSSlwj6eX0+ITuXqvU5k+dvyVRZOqzVx/evfzd7P+j/akasO0/S9WAKmqn1W7Zrp1Wm7OMmVkh5Vuz2LeoUcD3IuKbAJLOAy4GzgImkQwG3I9kxPj16Z8Vq72moqaVTVT1rdrSQd3R01D5lDEzK6S8kkVEFHX+p4hYk7G5C1vX0ZgM3BxJL/wTknaTNDwilhcznmKqHl2dswlp2BnDcn7x51PGzKxQ8h1n0eH8T4UYlCdpGvAJYDXwvnR3DZDZm7sk3VexyaJ2Wi1zPj5nm2Wl3IRkZuUu32aoX7ezv/UrL+egPEmPAHtnOTQ1Iu6OiKnAVEkXAucClwDZVvHJ+qyvpCnAFIDRo/NeEXaHG/rRocz59BzYBAg3IZlZRci3GWqb3lRJvYG3A98DpuZ5jqPzjOkWkgGAl5DUJEZlHBsJZB0UGBHTSTrgqaury3c52B1u82ubYRP02rUXh68+HMmr2plZ+cv3aahtRMTmiHiKZGLB67obhKT9MjZPBF5K398DfCJ9KupdwOpK7q8A2LhoIwD99unnRGFmFSPviQTb8TowtgBxfFfSeKCFZDGls9L99wMnAC8D64EzC3Ctkmrt3PaYCDOrJPl2cLcd3yBgOHAB8PfuBhERH2pnfwDndPf85WRLzWK0154ws8qRb81iFknHctt2kyfoAb/t70it4yxcszCzStLVQXktwIqI2FjgeHo81yzMrBKVxaC8ncmWmsU+rlmYWeXI+2koSQdJulnSLElPSfqZpLcWM7ieaONC1yzMrPLklSwknQjMJhnz8ADwIDAamC3p/xQvvJ6lpbGFTcs3QRX0HdG31OGYmeUt3z6Ly4BpEXFJ5k5J306P3VvowHqixqVpE1RNNVW9uzTExcysJPL9xtof+HmW/T8nWRTJ8tDaue0nocys0uSbLBqAd2TZ/w6gvnDh9Gytndv99nF/hZlVlnyboX4M3CBpHPA4yZiLw4GvkMwPZXlw57aZVarO9FmsA/4T+E66bxnJZH/XFCGuHskD8sysUuU7ziKAq4CrJO2a7ltbzMB6Ig/IM7NK1WGfhaTZkoZkbJ8OtDhRdI1rFmZWqXJ1cB8M9MnYvgHwKj1dEBHbTE9uZlZJOvuwvxdg6KKmV5toWd9Cr8G96D2ouzPDm5ntWB4ZtoNseWzW/RVmVoHy+RX3A5JWp++rgOMkbTO2IiLuKHhkPYwH5JlZJcsnWfykzfYP22wH0Ksw4fRcrlmYWSXrMFlEhJupCsSd22ZWycoqGUj6iqSQtGe6LUnXSHpZ0nNZlnetGF5728wqWdkkC0mjgGOARRm7JwH7pa8pwPUlCK0gPCDPzCpZ2SQLkhHiXyPpA2k1Gbg5Ek8Au0kaXpLouskD8syskpVFskgXV1oaEc+2OVQDLM7YXpLuy3aOKekqfrNWrFhRpEi7pqWxhU2vbIJe0He4Fz0ys8qzw0aHSXoE2DvLoanARcCx2T6WZV9k2UdETAemA9TV1WUtUyqNS9JaxUgvemRmlSmvZCHpD8AHI+L1NvsHAXdFxPtznSMijm7n3G8F9gWelQQwkmS51kNJahKjMoqPJJnttqJ4anIzq3T5/pp7JJCt/aQf8J7uBBARz0fE0IgYExFjSBLEhIh4BbgH+ET6VNS7gNURsbw71ysFD8gzs0rXYc2izaOqB0lalbHdCzgOWFqMwFL3AycALwPrgTOLeK2i8YA8M6t0uZqhZpH0EQTwUJbjG4AvFDKgtHbR+j6Acwp5/lJwzcLMKl2uZLEvSSfzfOBQIPMxo01AQ0Q0Fym2HsNrb5tZpcs13cfC9K0f4ekGd3CbWaXLKwlIOk3SsRnbF0taIul3lTpIbkeJiK0D8ka5GcrMKlO+NYZLW9+knd4XAdeQrKJ3ZeHD6jmaVjbRsrGF3rv19qJHZlax8v322geYm74/mWRsxeWSHgJ+V5TIeghP82FmPUG+NYuNwK7p+6OAR9L3qzP2WxaemtzMeoJ8axZ/Bq6U9BegDjgl3b8/287dZG20dm67ZmFmlSzfmsW5JI/KngKcFRGtU25Mws1QHfKAPDPrCfKqWUTEEuD/ZNn/pYJH1MN4QJ6Z9QR5j5+Q1E/SKZIukLRbum+spN2LF17lc83CzHqCfGedHUfSqT0Q2A24HXgd+Hy6/ZliBVjptgzIcwe3WbetWbOGhoYGmpqaSh1KRenTpw9Dhw5l0KBBXT5Hvh3cV5PMDfV5kiTR6h7gp12+eg/XvKGZpoYm1Fv03duLHpl1x5o1a6ivr6empob+/fuTLmlgOUQEGzZsYOnSZM7XriaMfJPFu4F3RURzm3+gRcCILl15J5C56JF6+T+2WXc0NDRQU1PDgAEDSh1KRZHEgAEDqKmpYdmyZV1OFp2Z86lPln2jScZaWBYekGdWOE1NTfTv37/UYVSs/v37d6v5Lt9k8RDw5YztSFfJ+xZwX5ev3sNtGZDnzm2zgnDTU9d1997l2wz1ZeCPkuaSrI73K2AcUA+c1q0IerAtA/L2cc3CzCpbvuMslkk6GDgdmEBSI5kOzIiIDUWMr6L5sVkz6yny7rOIiA0RcWNEnBsRZ0fE/xYqUUi6VNJSSc+krxMyjl0o6WVJcyUdV4jr7SgekGdmHbnpppuQtOXVq1cvampqOO2005g7d+42ZZubm7n++us59NBD2XXXXRk4cCCHHHII1113Hc3NxV+DLt9xFtUR0Zi+rwGmAAOAeyPisQLFclVEXNHmum8GPgK8heSpq0ck7V8pq/O5ZmFm+bj99tsZOXIkzc3NzJs3j+985zscddRRvPjiiwwePJimpiZOOukkHn74Yc4991wuu+wyJPHggw9y/vnnc//993PXXXfRu3fxlkHo8MySxgN3AG+S9BxwBvAwMAhoAc6XdEpE3FWk+CYDt6aJ6t+SXiZZ3nVmka5XMBHhmoWZ5eXggw9m3LhxABx22GGMGDGCY445hscff5xJkyYxbdq0LQlh8uTJWz53zDHHcMQRR3DSSScxbdo0LrnkkqLFmKsZ6gpgOXAi8AJwP8nEgYOBIcANwNcLFMu5kp6TdKOkIem+Grad1XZJuq/sNTU0EY1B791703ugFz0yK0f1M+qZOWYmj1Y9yswxM6mfUV/qkICtA+eamppobGzk6quv5oQTTtgmUbSaPHkykyZN4uqrr6axsbFoMeVKFu8CvhIR9wFnk4yruC4iWiKiBfgB8KZ8LiTpEUkvZHlNBq4HxgIHkySn1tX3sj3rFe2cf4qkWZJmrVixIp+QisqPzZqVt/oZ9cydMpfGhY0Q0LiwkblT5pYkYTQ3N7N582YaGxuZM2cOF110EUOHDuXII4/k6aefZvXq1Zx44ontfv7EE0/k9ddfZ/bs2UWLMdevvHsAywAiYq2kN4BVGcdfI8/FjyLi6HzKSfox8Nt0cwkwKuPwyNZ4spx/OskTWtTV1WVNKDuSB+SZ7RiP6tGCnatlfQtzPjaHOR+b0+nPHhlHdvm6b3rTtr9zjxgxgt/+9rcMGjSIxYuTxpUxY8a0+/nWY4sXL2bixIldjqMj+bSPtP3iLfgXsaThEbE83TyZpMkLkrmnbpH0fZIO7v2AJwt9/WJwzcLM8nXnnXcycuRIIoJly5Zx7bXXcsIJJ/DYY48RkfsrN58y3ZVPsviFpNaGsH7AjyWtT7cL9Wvz5ek4jgAWAJ8DiIgXJd0G/APYDJxTaU9CuWZhVlxd/Y1+5piZSRNUG9X7VDNxQXF+O2/PgQceuKWDG+DYY49l1KhRXHrppZx33nkALFiwoN3PL1y4EIBRo0a1W6a7cvVZ/Iyk2efV9PULkg7n1u1lwM3dDSIiPh4Rb42IgyLixIxaBhExLSLGRsT4iHigu9faUTw1uVl5q51WS9WAbb8CqwZUUTuttkQRbdW/f39qa2t57rnnqKurY9CgQdxzzz3tlr/nnnsYPHgwEyZMKFpMHdYsIuLMol25h/Njs2blbdgZwwCYP3U+jYsaqR5dTe202i37S2n9+vXMmzePt7zlLVRXV3Peeedx2WWXcffdd2/3RNTdd9/NAw88wMUXX0x1dfG+b/xMZ5F4QJ5Z+Rt2xrCySA7PPPMMK1euJCJYvnw51157LatWreILX/gCABdffDGzZs3itNNO45xzzmHSpElbBuX94Ac/4Pjjj+cb3/hGUWN0siiC5g3NNK1oQn286JGZ5Xbqqadueb/XXntx4IEH8uCDD3LccckMR3369OHee+/lhhtu4KabbuKGG24A4IADDuDKK6/krLPOKurobQDtiF70Ha2uri5mzZpVsuuv/+d6nhz/JP327ce75r+rZHGY9SRz5szhgAMOKHUYFS3XPZT0dETUZTvWmcWPLE/u3DaznsbJogj82KyZ9TROFkXgAXlm1tM4WRSBaxZm1tM4WRSBaxZm1tM4WaTymao43zKvP/Y6AC99+qWymYQXNQwAAAhlSURBVPLYrCfoiU9v7ijdvXceZ8HWqYpb1rcAW6cqhq2jPDtThs3JeTct27RdGTPrmj59+rBhwwYGDBhQ6lAq0oYNG+jTp0+XP+9xFrQ/oVjVLlUM/fBQABp+1UDLGy1dKlOKicnMepo1a9ZQX19PTU0N/fv3R8q23I21FRFs2LCBpUuXMmzYsC0LK2XT0TgL1yzY2iHdVssbLbxy4ysdfjafMu2d38zy1/olt2zZMpqamkocTWXp06dPzkSRi5MFyVNL2WoWvXfvzdjLxwIw72vz2Lxqc5fK+Kkos8IYNGhQt77wrOucLEimKs7sj4BkquL9rtlvS19DVb+qLpcphymPzcy6w09DkXQ+j58+nup9qkFJH8P46eO36ZQuVBkzs0rkDm4zMwM8kaCZmXWTk4WZmeXkZGFmZjn1yD4LSSuAhV38+J7AygKGU2yVFG8lxQqVFW8lxQqVFW8lxQrdi3efiNgr24EemSy6Q9Ks9jp4ylElxVtJsUJlxVtJsUJlxVtJsULx4nUzlJmZ5eRkYWZmOTlZbG96qQPopEqKt5JihcqKt5JihcqKt5JihSLF6z4LMzPLyTULMzPLyckig6TjJc2V9LKkr5c6nlwkLZD0vKRnJJXV/CaSbpTUIOmFjH27S3pY0r/SP4eUMsZM7cR7qaSl6f19RtIJpYyxlaRRkv4oaY6kFyV9Md1fdve3g1jL9d72k/SkpGfTeL+V7t9X0t/Se/srSX3LONabJP07494eXJDruRkqIakX8E/gGGAJ8BRwekT8o6SBdUDSAqAuIsruGXBJRwDrgJsj4sB03+XAqoj4bpqMh0TEBaWMs1U78V4KrIuIK0oZW1uShgPDI2K2pF2Bp4GTgE9RZve3g1hPozzvrYBdImKdpD7AX4AvAl8G7oiIWyX9CHg2Iq4v01jPAn4bEb8u5PVcs9jqUODliJgfEZuAW4HJJY6pYkXEY8CqNrsnAz9L3/+M5EujLLQTb1mKiOURMTt9vxaYA9RQhve3g1jLUiTWpZt90lcA7wdav3zL5d62F2tROFlsVQMsztheQhn/p04F8JCkpyVNKXUweRgWEcsh+RIBhpY4nnycK+m5tJmq5M06bUkaA7wd+Btlfn/bxAplem8l9ZL0DNAAPAzMA16PiNaVzcrmu6FtrBHRem+npff2KkkFWX3NyWKrbAv6lnsb3WERMQGYBJyTNqVY4VwPjAUOBpYDV5Y2nG1JGgj8BvhSRKwpdTwdyRJr2d7biGiOiIOBkSQtDgdkK7Zjo8qubaySDgQuBN4EHALsDhSkKdLJYqslwKiM7ZHAshLFkpeIWJb+2QDcSfIfu5zVp23YrW3ZDSWOp0MRUZ/+MLYAP6aM7m/aRv0bYEZE3JHuLsv7my3Wcr63rSLideBR4F3AbpJaVxYtu++GjFiPT5v+IiIagZ9SoHvrZLHVU8B+6VMPfYGPAPeUOKZ2Sdol7TBE0i7AscALHX+q5O4BPpm+/yRwdwljyan1izd1MmVyf9OOzZ8AcyLi+xmHyu7+thdrGd/bvSTtlr7vDxxN0s/yR+CUtFi53Ntssb6U8QuDSPpWCnJv/TRUhvTxvauBXsCNETGtxCG1S1ItSW0CkrXUbymneCX9EjiSZAbMeuAS4C7gNmA0sAg4NSLKolO5nXiPJGkmCWAB8LnWPoFSknQ48GfgeaB1wfeLSPoCyur+dhDr6ZTnvT2IpAO7F8kv07dFxLfTn7dbSZp1/g58LP3NvWQ6iPUPwF4kTevPAGdldIR3/XpOFmZmlouboczMLCcnCzMzy8nJwszMcnKyMDOznJwszMwsJycLszKTzshaFuMOzFr50VnbqUm6CdgzIv4j8/0OuvYY4N/AIRExK2P/QKA6Il7dEXGY5aN37iJm1hnptBDN0cXfxNIBVN0eRGVWSG6GMmPL2hWfBD4gKdLXkemxGkm3Snotfd0nab/Mz0p6QdKnJM0DGoFdlCym9ef0M6sk/U5S5qR0/07/fCq93qOZ58s4f5Wkb0paLKlRyYJXkzOOj0k//yElix6tl/QPSccU6XbZTsjJwixxBclUGY8Aw9PX45IGkMwLtBF4LzCRZJbUR9JjrfYFPgqcCrwtLb8LyfQxh5JMHbIauFdbV1lrneDt+PR6H2wnti8CXyWZPfStJNO83JFlBbRpwDXp9Z8Cbk2btMy6zc1QZiRNP5I2AI0R8UrrfkkfI5lj58zWZiVJnyOZ0fU/SBIMQF/g4xFRn3Ha32ReQ9KZwBqSJPEXYEV66NXMa2bxFeCKiLgl3b44nY7+K8DHMspdFRH3pte6CPgEyfxLf8njFph1yDULs469g6TWsFbSOknrSGoIQ0jWY2i1pE2iQNJYSbdImidpDckEhVUkE/3lRdIgYATw1zaH/gK8uc2+5zLet06hXVYLIFnlcs3CrGNVJDN3fiTLscwZXd/IcvxeYCnwufTPzcA/SGohnZWts7ztvqYtByIimaHavxBaYThZmG21iWS650yzSabTXpkuMJMXSXuQrLB2TkT8Md03gW1/5jalf7a95hYRsUbSMuBw4A8Zhw4nSTxmO4R/6zDbagFwoKTxkvZMV3ibQdJ8dLek96aLYx0h6crMJ6KyeA1YCXxW0jhJ7wV+RFK7aNUAbACOkzRM0uB2zvU94CuSTpe0v6RvA++hjJYitZ7PycJsqx+TrIo2i6Tz+bCIWA8cAcwHbgdeIllwZghJQsgqXS70w8BBJCuV/RD4Jsljta1lNgPnAZ8h6WNob/W1a0gSxuXpuU4GPhQRz3Tx72nWaR7BbWZmOblmYWZmOTlZmJlZTk4WZmaWk5OFmZnl5GRhZmY5OVmYmVlOThZmZpaTk4WZmeXkZGFmZjn9f5m1Gq+i3CGRAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig=plt.figure(figsize=(6, 3))\n",
    "myYbest=[bo.Y_ori[:idx+1].max() for idx,val in enumerate(bo.Y_ori)]\n",
    "plt.plot(range(len(myYbest)),myYbest,linewidth=2,color='m',linestyle='-',marker='o',label='BO')\n",
    "\n",
    "\n",
    "plt.xlabel('Iteration',fontsize=14)\n",
    "plt.ylabel('Best Found Value',fontsize=14)\n",
    "plt.legend(prop={'size': 16})\n",
    "plt.title('Performance',fontsize=16)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Run Random Search Baseline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# number of points\n",
    "n_point=len(bo.Y_ori) \n",
    "\n",
    "# uniformly select the input\n",
    "X_rand = np.random.uniform(myfunction.bounds[:, 0], myfunction.bounds[:, 1],size=(n_point, myfunction.input_dim))\n",
    "\n",
    "# evaluate the black-box function\n",
    "Y_rand=myfunction.func(X_rand)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Comparing Bayes Opt with Random Search"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Performance')"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig=plt.figure(figsize=(6, 3))\n",
    "\n",
    "myYbest_BO=[bo.Y_ori[:idx+1].max() for idx,val in enumerate(bo.Y_ori)]\n",
    "plt.plot(range(len(myYbest_BO)),myYbest_BO,linewidth=2,color='m',linestyle='-',marker='o',label='BO')\n",
    "\n",
    "myYbest_Rand=[Y_rand[:idx+1].max() for idx,val in enumerate(Y_rand)]\n",
    "plt.plot(range(len(myYbest_Rand)),myYbest_Rand,linewidth=2,color='b',linestyle=':',marker='s',label='Rand')\n",
    "\n",
    "plt.ylim([-7,0])\n",
    "plt.xlabel('Iteration',fontsize=14)\n",
    "plt.ylabel('Best Found Value',fontsize=14)\n",
    "plt.legend(prop={'size': 16})\n",
    "plt.title('Performance',fontsize=16)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Most of the time, BayesOpt will outperform Random Search"
   ]
  }
 ],
 "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}