{ "metadata": { "name": "Bayesian Coin Flip" }, "nbformat": 3, "nbformat_minor": 0, "worksheets": [ { "cells": [ { "cell_type": "heading", "level": 1, "metadata": {}, "source": [ "Bayesian Coin Flip" ] }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "Introduction" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "This notebook gives an introduction to Bayesian statistics, using mostly elementary probability theory and mathematics." ] }, { "cell_type": "code", "collapsed": false, "input": [ "from sympy import *\n", "from sympy.stats import *" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 41 }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "Coin Flipping" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Imagine we have a coin $c$ with two sides (heads $\\mathrm{H}$ and tails $\\mathrm{T}$). The coin may be fair, or it may biased, we don't know. We will say that $p(c=\\mathrm{H})$ is the probability of the coin toss facing heads, and $p(c=\\mathrm{T}) = 1 - p(c=\\mathrm{H})$ is the probability of tails. The value of $h := p(c=\\mathrm{H})$ itself is uncertain, and can take any value from 0 to 1. We represent this uncertainty by a probability density function $f(h)$. $f$ is a distribution of distributions: The value of $f(h)$ determines the bias of the coin, which itself determines the distribution of the coin tosses. The following image illustrates this distribution of distributions for a factory of fair coins. Any particular coin may have a bias (although coins with a large bias are rare), but most coins are fair within a margin of error." ] }, { "cell_type": "code", "collapsed": false, "input": [ "h = Symbol(\"h\")\n", "f = Beta(h, 10, 10)\n", "fig = plot(density(f)(h), (h, 0, 1), title=\"Probability density function of the probability of heads\", xlabel=\"$h$\", ylabel=\"$f(h)$\")" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEZCAYAAACEkhK6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XlYlOX6B/DvsGgOsg0gyoCiQLEoiKGouAyaC6hokia5\n4JLHLDU7xzplmljmsdzKNFIz/CluuaSogDvuSIaZu0Aii4gioCICAzy/P57jHIZFhm3eWe7PdXnJ\n8D7zzs07yz3PLmKMMRBCCCG1MBA6AEIIIdqBEgYhhBCVUMIghBCiEkoYhBBCVEIJgxBCiEooYRBC\nCFEJJYwaGBgY4O+//67XfR0dHXHs2LFqj50+fRqurq5KZY8fPw4AWLx4MaZOnVqvx6xNamoqDAwM\nUF5e3iTnB4AtW7Zg0KBBjXa+8PBw2NrawszMDHl5eY123tr85z//abLn4WV+++03ODg4wNTUFJcv\nX661fFxcHBwcHNQQWd3JZDJs2LChXvedOHEi5s+fX+NxU1NTpKamVilb+b3VlG7duoXOnTvDzMwM\nq1evrnK8IX9/XTXks6rOj6WWR1ETR0dHiMVimJqaonXr1pg0aRKePXum9jhEIhFEIlG1x3r37o2b\nN28qlX1h7ty5WL9+PQD1fMA3trFjx+LQoUOK2w15IcvlcvzrX//CsWPH8OTJE1haWjZWmEqq+9D9\n7LPPFM+DOs2ZMwc//vgjnj59Ci8vryrH1fnB0FAvew809L5Pnz6Fo6NjlbKV31sVv4w1tm+//Rb9\n+/fHkydPMGPGjCrHG/L3azKdShgikQgHDhzA06dPkZiYiIsXL2LRokVVypWWlgoQXf1o+7zK+sZ/\n//59FBUVwc3NrZEj0kyMMaSlpcHd3b3WckJjjDV5HHU5f01lRSJRk8V59+7dWp8rXaRTCaMiOzs7\nDB48GNeuXQPAv539+OOPcHFxwWuvvQYAWL9+PVxcXGBlZYXhw4cjKytL6RwHDx6Ek5MTbGxs8Mkn\nnyhefCkpKejXrx+sra1hY2ODcePG4fHjx0r3TUhIgIeHByQSCSZPnozi4mIAL29GCAsLw/jx4wEA\nffr0AQBYWFjAzMwMp06dgpWVFa5evaoo/+DBA5iYmODRo0dVzlVeXo45c+bAxsYGTk5OOHjwoNLx\nx48fY8qUKbCzs4O9vT3mz5+vqM1s3LgRvXr1wscffwyJRIIOHTogNjZWcd+NGzfCyckJZmZm6NCh\nA7Zu3ar4fe/evZXi9/LygpmZGX799Vd06tQJBw4cUJxHLpfD2tq6SvPL7du3FYnCwsICb7zxBu7e\nvVulxlWx2l9bzLm5uZg0aRKkUikkEglGjhyJwsJCBAQE4N69ezA1NYWZmRmysrKUngcAiIqKgoeH\nBywtLeHv71/lW+zy5cvh5eUFCwsLjBkzRvFcV8YYw6JFi+Do6AhbW1uEhobiyZMnKC4uhqmpKcrK\nyuDl5QUXF5cq9614PU1NTbFz507FsRUrVsDW1hZ2dnbYuHGj4vfFxcWYM2cO2rVrh9atW2P69Oko\nKiqqNraNGzfCz88PM2fOhIWFBdzc3JS+nctkMsybNw9+fn4wMTHBnTt3cO7cOXTt2hUWFhbo1q0b\nzp8/r3TO5ORk+Pr6wtzcHCNGjFBqVhw1ahTatGkDCwsL9O3bF9evX1e6b05ODgYOHAgzMzPIZDKk\npaUpjtVU06r43ho/fjzS0tIwbNgwmJqaYunSpRg6dGiV5iNPT0/s27ev2mtS0/Per18/xMXFYcaM\nGTAzM0NycnK1909NTUWvXr1gZmaGQYMGKb1P4+Pj0bNnT1haWqJz5844efKk4lhERATc3d1hZmYG\nJycnrFu3Tum8S5cuVbxvf/nlF6Vj0dHR8PDwgJmZGezt7bF8+fJqY6s3pkMcHR3Z0aNHGWOMpaWl\nMQ8PD/bFF18wxhgTiURs4MCBLC8vjxUVFbFjx44xa2trdunSJVZcXMxmzpzJ+vTpoziXSCRi/fr1\nY3l5eSwtLY29+uqr7Oeff2aMMZacnMyOHj3KSkpK2MOHD1mfPn3Y7NmzFfdt164d69SpE8vIyGC5\nubnMz8+PzZs3jzHG2IkTJ5i9vb1SzMeOHWOMMRYWFsbGjRvHGGMsNTWViUQiVlZWpij7/vvvs3//\n+9+K29999x0LCgqq9lqEh4czV1dXRQwymYwZGBgozjdixAj23nvvscLCQvbgwQPWrVs3tnbtWsYY\nYxEREczY2Jj9/PPPrLy8nIWHhzM7OzvGGGMFBQXMzMyM3b59mzHG2P3799m1a9cU9+vVq5fSNUxJ\nSVHc/vbbb9nbb7+tuL13717m6elZbfyV//47d+5UuR4ymYxt2LCh1pgZYywwMJCNGTOG5efnM7lc\nzk6dOsUYYywuLk7p+aj8PNy6dYuZmJiwo0ePstLSUvbtt98yZ2dnJpfLGWP8+fP19WVZWVksNzeX\nubm5sZ9++qnav2nDhg3M2dmZ3blzhxUUFLCRI0ey8ePH13i9Kqt8/MSJE8zIyIgtWLCAlZaWsujo\naCYWi1l+fj5jjLHZs2ez4cOHs7y8PPb06VM2bNgw9tlnn1V77oiICGZkZMS+++47Vlpaynbs2MHM\nzc1ZXl4eY4yxvn37snbt2rHr16+zsrIydv/+fWZhYcEiIyNZWVkZ27ZtG7O0tGS5ubmK8lKplF27\ndo09e/aMBQcHK67pi8crKChgJSUlbPbs2axz586KY6GhoczU1JSdPn2aFRcXsw8//LDG19XEiRNV\nem8xxtivv/7KfH19Fbf//PNPZmVlpXguK6rtea/42qtO3759mZOTE0tKSmLPnz9nMpmMffrpp4wx\nxjIyMpiVlRWLiYlhjDF25MgRZmVlxXJychhjjB08eJD9/fffjDHGTp48ycRiMUtMTGSMMRYTE8Ns\nbW0V1zUkJETperRu3ZqdOXOGMcZYfn6+4n6NRacSRrt27VjLli2ZhYUFa9euHfvggw9YUVERY4y/\nyE6cOKEoO3nyZKUP34KCAmZsbMzu3r2rKH/o0CHF8R9//JH179+/2sf97bffmLe3t+K2o6Oj4sOX\nMcaio6OZk5MTY+zlL+oFCxYo3lTVfUDGx8eztm3bKm6//vrrbOfOndXG5O/vrxTD4cOHFee7f/8+\na968OXv+/Lni+NatW5m/vz9jjL+ZnZ2dFceePXvGRCIRy87OZgUFBczCwoLt3r2bFRYWKj1mbQkj\nMzOTtWzZkj19+pQxxlhwcDBbunRptfFX/vtVSRg1xXzv3j1mYGCg+CCtqPLzwZjy8/Dll18qJbny\n8nImlUrZyZMnGWP8+duyZYvi+CeffMLee++9av+mfv36sfDwcMXtW7duMWNjY8XfVJ+E0aJFC6Vr\n0qpVK3bhwgVWXl7OTExMlMqfO3eOtW/fvtpzR0REKCVYxhjr1q0b27x5M2OMX+sFCxYojm3atEnp\nw5cxxnr06ME2btyoKF8xOV2/fp01a9aMlZeXV3nsvLw8JhKJ2JMnTxhjPGGEhIQojhcUFDBDQ0OW\nkZFR5TrUJWE8f/6cWVpasuTkZMYYY//617/YBx98UO31qO15l8lkii+Q1ZHJZOzrr79W3P7xxx/Z\n4MGDGWOMLVmyROmLAmOMDRo0iP3f//1ftecaMWIE+/777xljjE2aNEnput6+fVvperRt25atXbuW\nPX78uMbYGkKnmqREIhH27duHvLw8pKamYvXq1WjevLnieMWmoKysLLRr105x28TEBFZWVsjMzKy2\nfNu2bXHv3j0AQHZ2NsaMGQN7e3uYm5tj/PjxVZqFarpvQ/j6+qJFixaIi4vDzZs3kZKSgqCgoGrL\nZmVlVYnhhbt370Iul6NNmzawtLSEpaUl3nvvPTx8+FBRpnXr1oqfxWIxAKCgoAAmJibYsWMHfvrp\nJ9jZ2WHo0KG4deuWSvHb2dnBz88Pu3btQn5+PmJjYzF27Ng6XYOXqSnm9PR0SCQSmJub1/mc9+7d\nU7p2IpEIDg4OSq+Tio/bokULFBQUVHuuyq+5tm3borS0FNnZ2XWO6wUrKysYGPzvbSwWi1FQUICH\nDx+isLAQr7/+uuI5DggIQE5OTo3nkkqlSrfbtWun1Exb8fVU+bq8KF/xdV759SeXy5GTk4OysjJ8\n+umncHZ2hrm5Odq3bw8AithEIhHs7e0V9zUxMYFEImnwe+iVV17B6NGjsXnzZjDGsH37dqWmx4qy\nsrJqfd5r69Su6XVx9+5d7Ny5U/G8WFpa4uzZs7h//z4AICYmBt27d4eVlRUsLS0RHR2t+Hx52fsa\nAHbv3o3o6Gg4OjpCJpMhPj5elUujMiNVCz5//hzbtm3DlStXUFpaisLCQhgYGMDU1BS+vr4YNWqU\n0gtXE1V8gu3s7BRD8wDg2bNnePTokdKbJi0tTdGWnpaWpjg2d+5cGBoa4urVq7CwsMDevXsxc+ZM\npceq2OaalpYGOzu7esdaUWhoKCIjI2Fra4tRo0ahWbNm1ZZr06ZNlRhecHBwQPPmzfHo0aN6PWcD\nBw7EwIEDUVxcjM8//xxTp07FqVOnVLpvaGgoNmzYALlcjp49e6JNmzYq3c/ExAQAUFhYiJYtWwKA\n4g1WGwcHB+Tm5uLx48dVkkZtb3qpVIorV64objPGkJ6eXuXDVZXzVX7NpaWlwcjICLa2tir8FXVj\nbW2NFi1a4Pr16ypf44ofhgD/YBs+fLjidsW/TSqVYs+ePVXKBwQEKG5Xfv0ZGxvD2toakZGRiIqK\nwrFjx9CuXTvk5+dDIpEo+ghfXOMXCgoKkJubW+N7qKZrXt3vQ0NDMWHCBPj5+UEsFsPX17fa+9rZ\n2dXpea+Ltm3bYvz48VX6JgDe7xQcHIzIyEgMHz4choaGePPNNxXX5mXvawDw8fHB3r17UVZWhh9+\n+AGjR4+uUqYhVPq0OHr0KNavX4++ffti5cqV+OGHH7BhwwasX78ey5cvh6urK7777juVxo5ripCQ\nEERERODy5csoLi7G3Llz0b17d6WMvWzZMuTn5yM9PR2rVq3C22+/DeB/37TNzMyQmZmJpUuXKp2b\nMYY1a9YgMzMTubm5+PrrrzFmzJg6xWdjYwMDAwOkpKQo/X7cuHHYs2cPtmzZggkTJtR4/9GjR2PV\nqlXIzMxEXl4elixZojjWpk0bDBw4EP/85z/x9OlTlJeXIyUlRaUP/QcPHmDfvn149uwZjI2NYWJi\nAkNDw2rL2traVon/zTffRGJiIlatWvXS+CuzsbGBVCrF5s2bUVZWhl9++aXKuWvSpk0bBAQE4P33\n30d+fj7kcrnib7W1tcWjR4/w5MmTau87atQoHDx4EMePH4dcLsfy5cvxyiuvoGfPntWWZy8ZlRMS\nEoKVK1ciNTUVBQUFmDt3LsaMGaNy0q7uetbEwMAAU6dOxezZsxU1x8zMTBw+fLjG+zx48ACrVq2C\nXC7Hzp07cfPmTQQGBlb7twUGBuL27dvYtm0bSktLsWPHDty8eRNDhw5VlI2MjMSNGzdQWFiIL774\nAqNGjYJIJEJBQQGaN28OiUSCZ8+eYe7cuVViiY6OxtmzZ1FSUoL58+ejR48e1X5Ys5eM2KruevXo\n0QMikQhz5syp9f1T2/P+suf6ZcfHjRuH/fv34/DhwygrK0NRURHi4uKQmZmJkpISlJSUwNraGgYG\nBoiJiVF6zkaPHo2NGzcqruvChQsVx+RyObZs2YLHjx/D0NAQpqamNb4366vWV2pRUREcHR0xa9Ys\nODk5VTkuEong5eWFf/7znzAyUrnConaVv230798fX331FYKDg2FnZ4c7d+5g+/btSmWGDx+O119/\nHd7e3hg6dCgmT54MAFiwYAESExNhbm6OYcOGITg4WOn8IpEIY8eOxcCBA+Hk5AQXFxfMmzevxlgq\n/v7FMbFYjM8//xx+fn6wtLREQkICAP5tuUuXLjAwMECvXr1q/HunTp2KQYMGwcvLCz4+PlVi3LRp\nE0pKSuDu7g6JRIJRo0YpvrFXN4b8xe3y8nKsXLkSUqkUVlZWOH36NMLDw6u9X1hYGEJDQ2FpaYld\nu3YB4M0CI0eORGpqKkaOHFlj/NVdp/Xr12Pp0qWwtrbG9evX4efnV+21q+7+mzdvhrGxMVxdXWFr\na4tVq1YBAFxdXRESEoIOHTpAIpEgKytL6VyvvfYaIiMjMXPmTNjY2ODgwYPYv39/ja/1l42/nzx5\nMsaPH48+ffqgQ4cOEIvF+OGHH2r8eyurfD1rG+v/zTffwNnZGd27d4e5uTkGDBiA27dv11je19cX\nSUlJsLGxwfz587F7926l+S8VH0sikeDAgQNYvnw5rK2tsWzZMhw4cAASiURRdsKECZg4cSLatGmD\nkpISxTWfMGEC2rVrB6lUio4dOyo+xCs+ztixY7Fw4UJYWVnh0qVLiIyMrDaOyteg4s+fffYZFi1a\nBEtLS6xYsULx+wkTJuDKlSsYN25cjdfi1VdfrfV5r+35qilOe3t77Nu3D4sXL0arVq3Qtm1bLF++\nHIwxmJqaYtWqVRg9ejQkEgm2bdumVMsbPHgwZs+ejX79+uHVV19F//79lR4nMjIS7du3h7m5Odat\nW4ctW7a8NMa6ErHa0uRLJCYmwtPTU6MThS6aMmUKpFIpvvzyS6FDqZevvvoKSUlJ2LRpk9ChkP/a\nuHEjNmzYgNOnTwsdSpPbvHkz1q9fr3IzKvmfOjdgb926FbNnz8bGjRvRsmVL7Nixo14PXFZWBm9v\nbwwbNqza47NmzYKLiwu8vLxw6dKlej2GLkpNTcWePXswZcoUoUOpl9zcXPzyyy/4xz/+IXQoRA8V\nFhZizZo19PqrpzonDENDQ3zxxRdo1aoVVqxYoTSJqS6+//57uLu7V1uti46ORnJyMpKSkrBu3TpM\nnz69Xo+ha+bPn49OnTrhk08+URptoy3Wr1+Ptm3bIiAg4KXNaUT9dHUpi4oOHTqEVq1aoU2bNnjn\nnXeEDkcrqdQk5efnh27dusHHxweZmZmYPHkyrK2t6/2gGRkZmDhxIj7//HOsWLEC+/fvVzr+3nvv\nwd/fX9HJ7OrqipMnTzbJaBJCCCGqUanz4d///jdcXFxw/vx5JCcnY+TIkZBIJOjRowf8/f3RrVu3\nOj3oRx99hKVLl9Y4MiUzM1NprLG9vT0yMjKUEoZIJMKCBQsUt2UyGWQyWZ3iIIQQojqVEsaLyWFu\nbm6KkUIFBQX4/fff8ccff9QpYRw4cACtWrWCt7c34uLiaixXueJTXXU5LCxM5cclhBDSMLX2YRQX\nF1c7O7Rly5bw9/dX6l9QZYLIuXPnEBUVhfbt2yMkJATHjx+vMh5aKpUqTdzJyMholAkzhBBC6q/W\nhNG8eXPEx8dj69ateP78ebVl8vLysG7dOty9e7fWB1y8eDHS09MV8x769etXZXhlUFCQ4nfx8fGw\nsLCg/gtCCBGYSk1SpaWl6NChA1auXIkHDx6gqKgIcrkchoaGEIvFsLe3x9SpU+u1Vs+Lpqa1a9cC\nAKZNm4bAwEBER0fD2dkZJiYmiIiIqPN5CSGENC6VRkl99NFHGDt2LHx8fLBv3z6lmYdCacrNUQgh\nhFSlUg1j2LBh+Prrr1FUVITnz58jKSkJnTp1QseOHalvgRBC9ESdlwZZvnw5fHx8cO3aNVy9ehX3\n7t2Dvb09Zs6cqdjJTh2ohkEIIerVoLWkXti+fTvS09Px8ccfN0ZMKqGEQQgh6tUoG1g0a9YMrq6u\njXEqQgghGqpRahhCoBoGIYSol2ZvkUcIIURjUMIghBCiEkoYhBBCVEIJgxBCiEooYRBCCFEJJQxC\nCCEqoYRBCCFEJZQwCCGEqIQSBiGEEJVQwiCEEKISShiEEEJUQgmDEEKISihhEEIIUQklDEIIISpR\ne8IoKiqCr68vOnfuDHd3d3z22WdVysTFxcHc3Bze3t7w9vbGokWL1B0mIYSQSlTa07sxvfLKKzhx\n4gTEYjFKS0vRq1cvnDlzBr169VIq17dvX0RFRak7PEIaJDUVSEkBCgoAMzOgUyfA2lroqAhpHGpP\nGAAgFosBACUlJSgrK4NEIqlShjZHItqiuBj45RcgLg4QiQA7O6BZM6CwEAgP5z+PHAmMGAEYUCMw\n0WKCJIzy8nJ06dIFKSkpmD59Otzd3ZWOi0QinDt3Dl5eXpBKpVi2bFmVMgAQFham+Fkmk0EmkzVx\n5IQoi48HJk0CevUCPviA/18xKZSUALGxwI4dwHffARs2AC4uwsVLSEMIukXr48ePMWjQICxZskTp\nw/7p06cwNDSEWCxGTEwMPvzwQ9y+fVvpvrRFKxHaunXA/Pm8djFkyMvLlpUBa9bwfz/9BPj7qydG\nQhqToBVkc3NzDBkyBBcvXlT6vampqaLZKiAgAHK5HLm5uUKESEgVjAELFgArVwJnztSeLADA0BCY\nNYsni7ffBnbubPo4CWlsak8YOTk5yM/PBwA8f/4cR44cgbe3t1KZ7OxsRe0hISEBjLFq+zkIEcL8\n+cC5c7w5qq7NS/7+wJEjwEcfAVu3Nk18hDQVtfdhZGVlITQ0FOXl5SgvL8f48ePRv39/rF27FgAw\nbdo07Nq1C+Hh4TAyMoJYLMb27dvVHSYh1dqwAdi+HTh/HjA3r985vLyAEyeAvn2BVq2AN95o3BgJ\naSqC9mE0BPVhEHU7fBiYMAE4dQp49dWGny8uDhgzhtdWOnRo+PkIaWo0yI8QFWRlAd9/z/seGiNZ\nAIBMBsybBwwfzudtEKLpqIZBiArGjQMcHID//Kdxz8sY8MknfBTVihWNe25CGhslDEJqceoUTxg3\nbgAmJo1//pwcwN2dd4Z7eTX++QlpLNQkRchLlJYCM2YAy5Y1TbIA+NIhX33FJ/7RdyCiyShhEPIS\nP/4I2NgAo0Y17eO8+y5fYmTz5qZ9HEIagpqkCKlBdjbQsSNw8iRvMmpqCQm8A/zGDcDCoukfj5C6\nooRBSA0mTQKsrHhzlLr84x9AixZ8RBYhmoYSBiHVOH8eeOst4OZNwNRUfY+bkwN4ePA5H9QBTjQN\n9WEQUklZGe+AXrpUvckC4B3gX37JO9rp+xDRNJQwCKlk3TqeKEJChHn8d9/l/+/bJ8zjE1ITapIi\npIIXcyKOHeO75Qnl8GG+QOGVK7TpEtEc9FIkpII1a4D33hM2WQDAgAFA8+bAwYPCxkFIRVTDIOS/\nCguBtm2BCxcAJyeho+Gr4q5ZA5w+LXQkhHBUwyDkvyIjgZ49NSNZAHyUVmYmX82WEE1ACYMQ8BFJ\nq1YBH34odCT/Y2QEzJkDfPON0JEQwlHCIAR8QyPGgH79hI5E2cSJfGe/GzeEjoQQShiEAOAzq2fN\nAkQioSNRJhbzORlLlwodCSHU6U0I/v4b6NYNuHu36VakbYjcXMDZGfjrL8DeXuhoiD6jGgbRe6tX\nA5Mna2ayAACJBAgNpfWliPDUXsMoKipC3759UVxcjJKSEgwfPhz/qWYbs1mzZiEmJgZisRgbN26E\nt7e30nGqYZDGUFAAtGsHJCby/zVVWhrg7Q2kpNBKtkQ4aq9hvPLKKzhx4gT+/PNP/PXXXzhx4gTO\nnDmjVCY6OhrJyclISkrCunXrMH36dHWHSfTEpk18b21NThYAnx8yaRIf+kuIUIyEeFCxWAwAKCkp\nQVlZGSQSidLxqKgohIaGAgB8fX2Rn5+P7Oxs2NraKpULCwtT/CyTySCTyZo0bqJbysv5UNp164SO\nRDVBQcD77/OFETWtc57oB0ESRnl5Obp06YKUlBRMnz4d7pV2p8nMzISDg4Pitr29PTIyMl6aMAip\nqyNHgFdeAXr3FjoS1fTuDZSU8GG2PXoIHQ3RR4J0ehsYGODPP/9ERkYGTp06hbi4uCplKvdPiOgr\nFWlkmjqUtiYiEV/J9uefhY6E6CtBR0mZm5tjyJAhuHjxotLvpVIp0tPTFbczMjIglUrVHR7RYbdv\nAxcvAu+8I3QkdTNhArBnD/DkidCREH2k9oSRk5OD/Px8AMDz589x5MiRKiOggoKCsGnTJgBAfHw8\nLCwsqjRHEdIQP/wATJ3Km6S0SevWgL8/sGOH0JEQfaT2PoysrCyEhoaivLwc5eXlGD9+PPr374+1\na9cCAKZNm4bAwEBER0fD2dkZJiYmiIiIUHeYRIc9fgxs2cL3mtBG774LLFzIEx4h6kQzvYneiYgA\njh8HNm8WOpL6KSsDHB35XhmenkJHQ/QJzfQmemf1ar6on7YyNORzMjZsEDoSom+ohkH0yo0bwBtv\n8JnThoZCR1N/qamAjw+QkaF9/TBEe1ENg+iVLVuAkBDtThYAb5Lq0gX47TehIyH6hBIG0Rvl5Txh\njBsndCSNg+ZkEHWjhEH0xrlzfEVaLy+hI2kcw4fzJc9TUoSOhOgLShhEb0RG8tqFtszsrk3z5nym\n+s6dQkdC9AV1ehO9UFIC2NkBf/yh+SvT1kVCAjB+PHDzpu4kQqK5qIZB9EJMDODhoVvJAgC6dgVK\nS4FLl4SOhOgDShhEL7xojtI1IhEwZgywbZvQkRB9QE1SROfl5/OaRWoqYGkpdDSN7+pVIDCQ/30G\n9BWQNCF6eRGdt3s3n6yni8kCADp2BMzM+CgwQpoSJQyi87ZsAcaOFTqKphUSQs1SpOlRkxTRaenp\nQOfOQGambi+hkZLCd+G7dw8wEmQfTaIPqIZBdNq2bUBwsG4nCwBwcuLLhRw/LnQkRJdRwiA6TVdH\nR1WHmqVIU6MmKaKz/voLGDpUf0YP3bvHO8Dv3dP9GhURhh68jYi+etHZrQ/JAuAz2b28gNhYoSMh\nukpP3kpE35SXA9evA++8I3Qk6kWT+EhTUnvCSE9Ph7+/Pzw8PNCxY0esWrWqSpm4uDiYm5vD29sb\n3t7eWLRokbrDJFruwgXgzh2gUyehI1Gvt97iNYyCAqEjIbpI7QPwjI2NsXLlSnTu3BkFBQV4/fXX\nMWDAALi5uSmV69u3L6KiotQdHtERu3bxD099Y2UF9OoFREXpX+2KND211zBat26Nzp07AwBatmwJ\nNzc33Lt3r0o56tAm9cUYTxjBwUJHIgxqliJNRdApPqmpqbh06RJ8fX2Vfi8SiXDu3Dl4eXlBKpVi\n2bJlcHeK4DrDAAAazklEQVR3r3L/sLAwxc8ymQwymayJIyba4OJFPkqoY0ehIxHGiBHAjBlAbi4g\nkQgdDdElgg2rLSgogEwmw7x58zBixAilY0+fPoWhoSHEYjFiYmLw4Ycf4vbt20plaFgtqcm//81n\nO3/9tdCRCOejjwBvb2DCBKEjIbpEkFFScrkcwcHBGDduXJVkAQCmpqYQi8UAgICAAMjlcuTm5qo7\nTKKFXjRH6WP/RUXduwPbtwsdBdE1ak8YjDFMmTIF7u7umD17drVlsrOzFbWHhIQEMMYgobo1UcHl\ny/z//3aT6a3AQODMGb60OyGNRe19GGfPnkVkZCQ8PT3h7e0NAFi8eDHS0tIAANOmTcOuXbsQHh4O\nIyMjiMVibKevSkRFL2oX+r5dqakp4O8P7N/Pt3AlpDHQ0iBEZzAGuLry9aO6dhU6GuFt3sz3Atm7\nV+hIiK6ghEF0xtWrwJAhfO0ofa9hAP/baTAjg9c4CGkoWhqE6AxqjlJmYQH07AlERwsdCdEVlDCI\nzqDRUVUFB/NmKUIaAzVJEZ1w4wYwYACQlqY/q9Oq4uFDwNkZuH8faNFC6GiItqO3FtEJu3fzb9OU\nLJTZ2AA+PsChQ0JHQnQBvb2ITqDmqJpRsxRpLNQkRbReUhLQpw8fDWRoKHQ0micrC3B3581SzZsL\nHQ3RZlTDIFrv0CFg6lRKFjVp0wbw8ACOHRM6EqLtKGEQrbdpE9C3r9BRaLYJE4C4OKGjINqOmqSI\nVsvI4PtY378PGBsLHY3mSk0FunXjzVNUEyP1RTUMotX27gWGDqVkURtHR8Deni9ISEh9UcIgWm3P\nHmDkSKGj0A5vvgn89pvQURBtRk1SRGs9egR06ECT0lRFa22RhqIaBtFa+/cDb7xByUJVHh58WG1i\notCREG1FCYNoLWqOqhuRiJqlSMNQkxTRSgUFgJ0dXzvKwkLoaLTHhQvApEnA9etCR0K0EdUwiFaK\njQV69KBkUVdduwKPHwO3bgkdCdFGlDCIVtqzhzevkLoxMABGjKBmKVI/lDCI1ikpAWJigOHDhY5E\nO40cyRMuIXWl9oSRnp4Of39/eHh4oGPHjli1alW15WbNmgUXFxd4eXnh0qVLao6SaLLjx/liem3a\nCB2JdurTB0hJAdLThY6EaBu1JwxjY2OsXLkS165dQ3x8PNasWYMbN24olYmOjkZycjKSkpKwbt06\nTJ8+Xd1hEg1GzVENY2zMZ8fv3St0JETbqD1htG7dGp07dwYAtGzZEm5ubrh3755SmaioKISGhgIA\nfH19kZ+fj+zsbHWHSjRQWRmwbx8ljIYaORI4ckToKIi2MRLywVNTU3Hp0iX4+voq/T4zMxMODg6K\n2/b29sjIyICtra1SubCwMMXPMpkMMpmsKcMlGiAhAXBzA5ychI5Euw0YAISGAjk5gLW10NEQbSFY\nwigoKMBbb72F77//Hi1btqxyvPIcC1E1axlUTBhEP+zcCfj7Cx2F9hOL+Sz5/fv5vAxCVCHIKCm5\nXI7g4GCMGzcOI0aMqHJcKpUivUKPXEZGBqRSqTpDJBqIMeq/aEw065vUldoTBmMMU6ZMgbu7O2bP\nnl1tmaCgIGzatAkAEB8fDwsLiyrNUUT//PknYGQEdOokdCS6YcgQvqnS06dCR0K0hdqbpM6ePYvI\nyEh4enrC29sbALB48WKkpaUBAKZNm4bAwEBER0fD2dkZJiYmiIiIUHeYRAO9WDuKVlptHBYWgJ8f\nnzU/apTQ0RBtQGtJEa3h4QFs2AB07y50JLpj3TrgxAlg2zahIyHagBIG0Qq3bvHO7owMvrwFaRzZ\n2YCrK99TpHlzoaMhmo7eekQr/PYb76SlZNG4bG2Bjh2BY8eEjoRoA3r7Ea1Ae180nZEjabQUUQ01\nSRGNl54OeHsDWVl8WQvSuFJTgW7d+PU1NBQ6GqLJqIZBNN7evXztI0oWTcPREbC3B86eFToSouko\nYRCN99tv1BzV1N58k5Y8J7WjJimi0XJyAGdn3lzSooXQ0eiua9eAwEDePEXzXEhNqIZBNFpUFF8o\nj5JF03J358NqExOFjoRoMkoYRKNRc5R6iER89drjx4WOhGgyapIiGuvJE8DBgY+SMjMTOhrd9/vv\nwLhxwM2b1CxFqkc1DKKxDhwAgoIoWaiLjw9QVMT7MwipDiUMorF27QL69RM6Cv0hEgFvvcWvOyHV\noSYpopEKCgCpFLhzB5BIhI5Gf5w/D0ydCly9KnQkRBNRDYNopIMHgZ49KVmom68vkJ8P3LghdCRE\nE1HCIBpp507ePELUy8AACA4Gdu8WOhKiiahJimicZ88AOzvg778BKyuho9E/p08DM2fyHQ4JqYhq\nGETjREfzTZIoWQijZ0++T0ZSktCREE1DCYNoHGqOEpahIZ8sSc1SpDJKGESjFBYChw7xxfCIcIKD\naXgtqUqQhDF58mTY2tqiU6dO1R6Pi4uDubk5vL294e3tjUWLFqk5QiKUmBi+N4O1tdCR6Lc+fYC0\nND6smZAXBEkYkyZNQmxs7EvL9O3bF5cuXcKlS5cwb948NUVGhEbNUZrByAgYMYKapYgyQRJG7969\nYWlp+dIyNAJK/zx/DsTGUnOUpqBZ36QyI6EDqI5IJMK5c+fg5eUFqVSKZcuWwd3dvUq5sLAwxc8y\nmQwymUx9QZJGFxsLvP460KqV0JEQAPD35zPu794F2rUTOhqiCTQyYXTp0gXp6ekQi8WIiYnBiBEj\ncPv27SrlKiYMov2oOUqzGBsDPXoAv/4KfPyx0NEQTaCRo6RMTU0hFosBAAEBAZDL5cjNzRU4KtKU\nior4/Ava+0KzjBkDbN8udBREU2hkwsjOzlb0YSQkJIAxBgktKqTTDh0CvL0BW1uhIyEVyWTAvXtA\nNRV8oocEaZIKCQnByZMnkZOTAwcHByxcuBByuRwAMG3aNOzatQvh4eEwMjKCWCzGdvqKo/POnAFC\nQoSOglRmaAiMHg1s2wYsWCB0NERotJYUEdyzZ3wp89u3qcNbE8XHAxMn8hVsaSc+/aaRTVJEv+zf\nzztXKVloJl9foKSEFiMklDCIBtiyBXjnHaGjIDURiXjn97ZtQkdChEZNUkRQOTmAszOQng6Ymgod\nDanJlSvAkCFAairfM4PoJ3rqiaB27QICAihZaLpOnQAzM+DcOaEjIUKihEEERc1R2oOapQg1SRHB\n3L3LlwK5dw9o1kzoaEhtkpMBPz8gM5MvTkj0D9UwiGC2beNLgVCy0A7OznxNqePHhY6ECIUSBhHM\nli3A2LFCR0HqIiSEmqX0GTVJEUHQqBvtdO8e0LEj//+VV4SOhqgbvVWJIF50dlOy0C52dny/kkOH\nhI6ECIHerkTtyst5swY1R2mn3r2BDRuEjoIIgRIGUbuzZ/mY/hq2dCca7q23gNOngexsoSMh6kYJ\ng6jd4cN8MTuinVq25Pt9R0YKHQlRN+r0Jmr17Bng4AD89Rdgby90NKS+Tp4EZszgzyOtYKs/qIZB\n1GrnTqBnT0oW2q53b578ExOFjoSoEyUMolY//wxMnSp0FKShDAyA0FBg40ahIyHqRE1SRG2uXwfe\neANIS6OlJXTBnTtA1658qZDmzYWOhqgD1TCI2mzYwDu7KVnohvbtAU9PvgEW0Q9UwyBqUVzMO7vP\nnwecnISOhjSWTZuAX38FDhwQOhKiDmqvYUyePBm2trbo9JJB+LNmzYKLiwu8vLxw6dIlNUZHmsre\nvXzeBSUL3RIczOfVZGUJHQlRB7UnjEmTJiE2NrbG49HR0UhOTkZSUhLWrVuH6dOnqzE60lSos1s3\nmZjwfTL27BE6EqIOak8YvXv3hqWlZY3Ho6KiEBoaCgDw9fVFfn4+smlKqVb7+2/gzz/5ZC+ie0JD\ngRUr+JIvRLdpXPdjZmYmHBwcFLft7e2RkZEBW1vbKmXDwsIUP8tkMshkMjVESOrql1+AceNodVNd\n5esLWFgAsbFAYKDQ0ZCmpHEJA0CVzmxRDVNJKyYMoplKS4GICL4cCNFNIhGf9b1mDSUMXadxw2ql\nUinS09MVtzMyMiCVSgWMiDREdDTfpc3DQ+hISFMaMwZISODNj0R3aVzCCAoKwqZNmwAA8fHxsLCw\nqLY5imgH6uzWDy1a8Dk24eFCR0KaktrnYYSEhODkyZPIycmBra0tFi5cCLlcDgCYNm0aAGDGjBmI\njY2FiYkJIiIi0KVLl6qB0zwMjZeZyYfSpqfz0TREt/39N+/PSEvjCYToHpq4R5rM0qVASgrw009C\nR0LUZcgQvl/GpElCR0KaAiUM0iRKSgBHR76VJ22UpD+io4H584GLF2nZc12kcX0YRDds3Qp07EjJ\nQt8MHgzk5wMXLggdCWkKlDBIo2MMWLYM+PhjoSMh6mZgALz/Ph9iS3SPRs7DINotNpavSPvGG0JH\nQoQwaRJfM+zBA6BVK6GjIY2Jahik0S1dCsyZQ23Y+koiAUaO5MvZE91CCYM0qj/+AJKTgbffFjoS\nIqQPP+Qd4EVFQkdCGhMlDNKoli3jHxbGxkJHQoTk6QmYmfFlYYjuoGG1pNGkpgI+PnwCl5mZ0NEQ\nocXH85pmUhLQrJnQ0ZDGQDUM0mhWrgSmTKFkQbju3YFXXwU2bxY6EtJYqIZBGkVuLuDsDFy5AtBa\nkeSFU6eAyZOBmzdpL3ddQDUM0ih++gkICqJkQZT16cNfE9u3Cx0JaQxUwyANVlQEtG8PHDnCZ3cT\nUtHRo8DMmcDVq4ChodDRkIagGgZpsG3b+MY5lCxIdfr35zvy7d4tdCSkoaiGQRrk2TPAxQWIiuIj\npAipTnQ08OmnfG93A/qaqrXoqSMNsmIF0LcvJQvycgEBfGhtVJTQkZCGoBoGqbcHDwB3d741Z4cO\nQkdDNN3evcBXX9HS59qMahik3r76Chg7lpIFUU1QECCX88UpiXaiGgapl+RkPjHrxg3AxkboaIi2\nOHgQ+OQT3pdBy8doH6phkHr5/HPgo48oWZC6CQwE7Oxo215tJUjCiI2NhaurK1xcXPDNN99UOR4X\nFwdzc3N4e3vD29sbixYtEiBKUpOEBODMGWD2bKEjIdpGJOJLyHz1FZCTI3Q0pK7U3iRVVlaG1157\nDUePHoVUKkXXrl2xbds2uLm5KcrExcVhxYoViHrJkApqkhIGY4C/PzBuHPDuu0JHQ7TVzJlAeTnt\nzKdt1F7DSEhIgLOzMxwdHWFsbIwxY8Zg3759VcpRMtBM0dF8dNTEiUJHQrTZwoXAzp187TGiPdS+\nHFhmZiYcHBwUt+3t7XGh0o7xIpEI586dg5eXF6RSKZYtWwZ3d/cq5woLC1P8LJPJIJPJmipsAqCw\nEFi0CPjuO1pIjjSMRAIsWMD3Tjl2jIbZagu1v+1FKrwyunTpgvT0dIjFYsTExGDEiBG4fft2lXIV\nEwZpenPn8r2aBw4UOhKiC6ZNA8LDgT17gOBgoaMhqlB7k5RUKkV6erridnp6Ouzt7ZXKmJqaQiwW\nAwACAgIgl8uRm5ur1jiJslOneBPCqlVCR0J0hZERsHYt/yKSlyd0NEQVak8YPj4+SEpKQmpqKkpK\nSrBjxw4EBQUplcnOzlb0YSQkJIAxBolEou5QyX89e8b3NAgP500JhDQWPz9g0CBgxgyhIyGqUHuT\nlJGREVavXo1BgwahrKwMU6ZMgZubG9auXQsAmDZtGnbt2oXw8HAYGRlBLBZjOy2mL6jPPgN69uQz\ndQlpbEuWAN7ewK5dwFtvCR0NeRma6U1eKi6OD6G9cgWwtBQ6GqKr4uOBESP4DPDWrYWOhtSEZnqT\nGhUU8Kaon36iZEGaVvfufD/4f/yDz/UhmolqGKRGM2bwpLFxo9CREH1QUgJ06wbMmsW/qBDNQzUM\nUq2oKOD2bb6MAyHq0KwZsHkz35nv6lWhoyHVoYRBqkhM5Mt+LFpETVFEvTp1At55Bxg2jK8oQDQL\nJQyiJCMDGD6cD6Ht1k3oaIg+GjuW/3vzTaC4WOhoSEXUh0EUnj4Fevfmb9aPPxY6GqLPysuBt98G\nWrQA/u//aOkQTUEJgwAASkt5zcLeno+KojcoEVphIdCnD1825LPPhI6GAAJM3COahzG+t4VcDqxe\nTcmCaAaxGNi3jw+5dXGhSX2agBIGwQ8/ACdP8k2RaNtMokmkUiAmBhg8GHjyhIbbCo0Shh4rKwP+\n9S/g6FHgyBHA3FzoiAipqmNHvgT64MHAw4d8T3CqBQuDEoaeev6cd27n5fGahYWF0BERUrPXXuOv\n08GD+XDbpUsBAxrjqXZ0yfVQTg7Qvz8fgRIbS8mCaAeplC+zf+ECEBrK+9yIelHC0DMpKXzlWZmM\nz6pt3lzoiAhRnaUlcPgwkJ/PR/U9fix0RPqFEoaeYAzYupVvifnPfwKLF1OVnmgnsZjv0te9O+Dl\nxfs3iHrQPAw9cPcuMH06n8X98880g5vojpgYvsJtUBDwzTdAy5ZCR6Tb6DumDisr41uqvv460KsX\n8McflCyIbgkIAP76i+8K6ekJnDghdES6jWoYOogx/sb5+WcgMxNYt46PMiFEl+3fD3z7Le/nWLyY\nD8cljYsShg4pKwMOHgR+/ZWPJFm0CBg1ivoqiP4oKgJ+/JHPLWrenK+J1rOn0FHpDkoYOmDHjjgk\nJ8tw7Bivmn/8Md/u0kgPZ9nExcVBJpMJHYZG0Odr8ewZ3/hrxw4+Q7xXrzh88YUMrVoJHZnwGvK6\nEOS7Z2xsLFxdXeHi4oJvvvmm2jKzZs2Ci4sLvLy8cOnSJTVHqNnKy4HLl4HvvgN8fYHPP49DZiaf\nzHThAl9zRx+TBcDfDITT52thYgJ88AHfk37FCuD06TgMGQIMGsSbaJOT9Xcr2Ia8LtT+sVJWVoYZ\nM2bg6NGjkEql6Nq1K4KCguDm5qYoEx0djeTkZCQlJeHChQuYPn064uPj1R2qRmAMyMriO5DdusXf\nAKdPA35+QNu2vNnp9Gngyy+FjpQQzWNgAPTrx/fWmDMHOHQIOHAA2LWLbwnbujVfEdfTE3BzA6ys\nhI5Ys6k9YSQkJMDZ2RmOjo4AgDFjxmDfvn1KCSMqKgqhoaEAAF9fX+Tn5yM7Oxu2trbqDrfJlJXx\n/bIfP+b/Hj7kSx48fQqkpfHtUR894tXp1FRgyBDA0ZHvEbBqFZ/1+sLZs0L9FYRoj5Yt+VLpwcH8\ni1hKCl9u5NIl/kXs8WP+O1tboEsX/oWsZUvAxoYnFktLvt6ahQWfC6KXfYNMzXbu3Mneffddxe3N\nmzezGTNmKJUZOnQoO3v2rOJ2//792cWLF5XKAKB/9I/+0T/6V49/9aX2GoZIxWUmWaUGxsr3q3yc\nEEJI01J7pUoqlSI9PV1xOz09Hfb29i8tk5GRAWnFNhhCCCFqp/aE4ePjg6SkJKSmpqKkpAQ7duxA\nUFCQUpmgoCBs2rQJABAfHw8LCwud6r8ghBBtpPYmKSMjI6xevRqDBg1CWVkZpkyZAjc3N6xduxYA\nMG3aNAQGBiI6OhrOzs4wMTFBRESEusMkhBBSWb17P9QkJiaGvfbaa8zZ2ZktWbKk2jIzZ85kzs7O\nzNPTkyUmJqo5QvWp7VpERkYyT09P1qlTJ9azZ092+fJlAaJUD1VeF4wxlpCQwAwNDdnu3bvVGJ16\nqXItTpw4wTp37sw8PDxY37591RugGtV2LR4+fMgGDRrEvLy8mIeHB4uIiFB/kGowadIk1qpVK9ax\nY8cay9Tnc1OjE0ZpaSlzcnJid+7cYSUlJczLy4tdv35dqczBgwdZQEAAY4yx+Ph45uvrK0SoTU6V\na3Hu3DmWn5/PGONvHH2+Fi/K+fv7syFDhrBdu3YJEGnTU+Va5OXlMXd3d5aens4Y4x+aukiVa7Fg\nwQL26aefMsb4dZBIJEwulwsRbpM6deoUS0xMrDFh1PdzU6NHElecs2FsbKyYs1FRTXM2dI0q16JH\njx4w/+/G3L6+vsjIyBAi1CanyrUAgB9++AFvvfUWbGxsBIhSPVS5Flu3bkVwcLBicIm1tbUQoTY5\nVa5FmzZt8OTJEwDAkydPYGVlBSMdXBahd+/esLS0rPF4fT83NTphZGZmwsHBQXHb3t4emZmZtZbR\nxQ9KVa5FRRs2bEBgYKA6QlM7VV8X+/btw/Tp0wGoPpxb26hyLZKSkpCbmwt/f3/4+Phg8+bN6g5T\nLVS5FlOnTsW1a9dgZ2cHLy8vfP/99+oOUyPU93NTo1NrY83Z0AV1+ZtOnDiBX375BWd1dAq4Ktdi\n9uzZWLJkiWKRysqvEV2hyrWQy+VITEzEsWPHUFhYiB49eqB79+5wcXFRQ4Tqo8q1WLx4MTp37oy4\nuDikpKRgwIABuHz5MkxNTdUQoWapz+emRicMmrPxP6pcCwD466+/MHXqVMTGxr60SqrNVLkWf/zx\nB8aMGQMAyMnJQUxMDIyNjasM4dZ2qlwLBwcHWFtbo0WLFmjRogX69OmDy5cv61zCUOVanDt3Dp9/\n/jkAwMnJCe3bt8etW7fg4+Oj1liFVu/PzUbpYWkicrmcdejQgd25c4cVFxfX2ul9/vx5ne3oVeVa\n3L17lzk5ObHz588LFKV6qHItKpo4caLOjpJS5VrcuHGD9e/fn5WWlrJnz56xjh07smvXrgkUcdNR\n5Vp89NFHLCwsjDHG2P3795lUKmWPHj0SItwmd+fOHZU6vevyuanRNQyas/E/qlyLL7/8Enl5eYp2\ne2NjYyQkJAgZdpNQ5VroC1WuhaurKwYPHgxPT08YGBhg6tSpcHd3FzjyxqfKtZg7dy4mTZoELy8v\nlJeX49tvv4VEIhE48sYXEhKCkydPIicnBw4ODli4cCHkcjmAhn1uau0GSoQQQtRLo0dJEUII0RyU\nMAghhKiEEgYhhBCVUMIghBCiEkoYhBBCVEIJgxBCiEooYRBCiJ6IioqCn59fve9PCYMQQvSEi4sL\nunXrVu/7U8IghBA9cf78+Qatm0UJgxBC9ER8fDwyMzOxY8cObN26tc73p4RBCCF64ubNm5g8eTIG\nDBhQr3XmKGEQQogeKCgogEQigbW1NeLj49G5c+c6n4MSBiGE6IHff/8dPXr0AMBHS/Xs2ROJiYl1\nOgclDEII0QM3b96Ev78/AMDGxga///47PD0963QOWt6cEEKISqiGQQghRCWUMAghhKiEEgYhhBCV\nUMIghBCiEkoYhBBCVEIJgxBCiEooYRBCCFEJJQxCCCEq+X+z/+4/uocQVAAAAABJRU5ErkJggg==\n" } ], "prompt_number": 42 }, { "cell_type": "markdown", "metadata": {}, "source": [ "After sampling a specific coin from $f(h)$, we can then represent the coin toss with a discrete random variable." ] }, { "cell_type": "code", "collapsed": false, "input": [ "c = Symbol(\"c\")\n", "bias = sample(f)\n", "C = Coin(c, bias)\n", "print density(C)\n", "print [sample(C) for i in range(10)]" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "stream", "stream": "stdout", "text": [ "{H: 0.511884213785308, T: 0.488115786214692}\n", "[H, T, T, T, H, H, H, T, T, H]\n" ] } ], "prompt_number": 43 }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "Bayesian Inference" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Our goal is to understand the consequences of observing tosses of a coin with unknown bias $\\hat{h}$. Ultimately, we want to know the bias of the coin itself, but to answer that, we look at a much bigger question: What is the probability density function $f(h)$ of a coin factory that produces coins with this bias? Of course, any factory that assigns a non-zero probability mass to $\\hat{h}$ is a possible answer, but as we will see in the next section, some are more reasonable than others. Let's first assume that by some kind of trick we came up with a guess of such a *prior distribution* $f_n$ for our factory. Then the Bayesian inference process allows us to improve this guess, based on an observed coin toss $c_n$ (heads or tails). The resulting *posterior distribution* $f_{n+1}$ is the best estimate for the probability density function of our hypothetical factory, given the prior distribution and the observed data. It is derived by Bayes' theorem for conditional propabilities:\n", "\n", "$$\n", "f_{n+1}(\\hat{h}) := f_n(h=\\hat{h}|c=c_n) = \\frac{p(c=c_n|h=\\hat{h}) f_n(h=\\hat{h})}{p(c=c_n)}, \\hat{h}\\in [0, 1]\n", "$$\n", "\n", "The left hand side is the posterior distribution after one Bayesian update at step $n$ (there are as many steps as there are observed data points). It represents the updated probability of the coin bias being $\\hat{h}$ after observing $c_n$. The parameter $\\hat{h}$ varies over the whole range $[0,1]$, giving us a probability mass for each possible bias of the coin. On the right hand side, we have the *likelihood* $p(c=c_n|h=\\hat{h})$ of the observed data $c_n$, assuming that the bias of the coin is $\\hat{h}$, and the prior distribution for $h$. To get a proper probability density function, we have to make sure that the right hand side sums to $1$ over all possible values for $\\hat{h}$. For this, we divide by the *normalizing constant*, which is easily expressed by the sum over all possible numerators:\n", "\n", "$$\n", "p(c=c_n) = \\int_0^1 p(c=c_n|h) f_n(h) \\thinspace dh\n", "$$\n", "\n", "Don't panic. We will do these calculations several times for specific values of $f_n$ and $c_n$, which will give you plenty of time to review how the Bayesian update process works and what its effects are on the probability density function of the coin bias.\n" ] }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "Prior Distribution" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "To get started with Bayesian inference, we need a prior distribution of the probability function of our coin factory. If we had complete knowledge of the bias of the coin, this would be a simple Dirac delta function:" ] }, { "cell_type": "code", "collapsed": false, "input": [ "f_exact = DiracDelta(h-bias)\n", "# Note: DiracDelta is not a Sympy distribution, so we can not plot the density function or sample it.\n", "integrate(f_exact, (h, 0, 1))" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "pyout", "prompt_number": 44, "text": [ "1" ] } ], "prompt_number": 44 }, { "cell_type": "markdown", "metadata": {}, "source": [ "If we lack any information, we can assume that the coin comes from a uniform distribution:" ] }, { "cell_type": "code", "collapsed": false, "input": [ "f_uniform = Uniform(h, 0, 1)\n", "sample(f_uniform)\n", "# This craps out on me.\n", "# plot(density(f_uniform)(h), (h, 0, 1))" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "pyout", "prompt_number": 45, "text": [ "0.583601108813623" ] } ], "prompt_number": 45 }, { "cell_type": "markdown", "metadata": {}, "source": [ "Choosing the uniform distribution is not as random as it may appear. The uniform distribution has maximum entropy. In this sense it ensures that the prior does not include any hidden assumptions that are not justified by the model. However, sometimes it can be useful to choose other priors, and in general it may be a good idea to verify that the inference process is not overly sensitive to different reasonable choices for the prior. We will revisit this question at the very end." ] }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "Step 1: The Uniform Prior" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Because we assume no knowledge about the bias of the coin in question, we use a uniform prior:\n", "\n", "$$\n", "f_1(h) := 1\n", "$$" ] }, { "cell_type": "code", "collapsed": false, "input": [ "f1 = 1" ], "language": "python", "metadata": {}, "outputs": [], "prompt_number": 46 }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "Step 2: The First Coin Toss" ] }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "Heads" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We toss the coin and get heads:\n", "\n", "$$\n", "c_1 = H\n", "$$\n", "\n", "We now want to perform the first Bayesian update. For this, let's look at the individual components of Bayes Theorem. Clearly, the prior probability of any bias $\\hat{h}$ is $1$, because of the chosen uniform distribution:\n", "\n", "$$\n", "f_1(\\hat{h}) = 1 \\text{ (uniform prior)}\n", "$$\n", "\n", "The likelihood of observing heads, given the bias $\\hat{h}$, is simply $\\hat{h}$, because this is how we defined the bias. Note that the likelihood of tails would be $1-\\hat{h}$. These are the only two possible outcomes for the likelihood in this example.\n", "\n", "$$\n", "p(c=H|h=\\hat{h}) = \\hat{h} \\text{ (by definition)}\n", "$$\n", "\n", "Now we only need to calculate the normalization constant. Luckily, we already did most of the work above:\n", "\n", "$$\n", "\\int_0^1 p(c=H|h) \\cdot f_1(h) \\thinspace dh = \\int_0^1 h \\cdot 1 \\thinspace dh = \\left. \\frac{1}{2} h^2 \\right |_0^1 = \\frac{1}{2}\n", "$$\n", "\n", "Putting it all together, we get the posterior distribution:\n", "\n", "$$\n", "f_2(\\hat{h}) := f_1(h=\\hat{h}|c=H) = \\frac{\\hat{h} \\cdot 1}{\\frac{1}{2}} = 2\\hat{h}\n", "$$\n", "\n", "or simply: $f_2(h) = 2h$\n" ] }, { "cell_type": "code", "collapsed": false, "input": [ "f2 = h * f1 / integrate(h * f1, (h, 0, 1))\n", "plot(f2, (h, 0, 1), title=\"Posterior distribution after observing $c=(\\mathrm{H})$.\", xlabel=\"$h$\", ylabel=\"$f_2(h)$\")\n", "f2" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEcCAYAAAA2g5hwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nrkJggg==\n" }, { "output_type": "pyout", "prompt_number": 47, "text": [ "2*h" ] } ], "prompt_number": 47 }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can see that the posterior distribution favors heads-biased coins over tails-biased coins, in stark contrast to the uniform distribution. However, even extremely tails-biased coins are still possible, with one exception: Coins that exclusively show tails are now excluded ($f_2(0) = 0$). This makes sense, because we already observed heads once.\n" ] }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "Tails" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Because of the symmetry of the problem, we would get the same result if we had seen tails, but flipped horizontally (due to the transformation $p(c=\\mathrm{T}) = 1 - h$)." ] }, { "cell_type": "code", "collapsed": false, "input": [ "plot(f2.subs(h, 1-h), (h, 0, 1), title=\"Posterior distribution after observing $c=(\\mathrm{T})$.\", xlabel=\"$h$\", ylabel=\"$f_2(h)$\")\n", "f2.subs(h,1-h)" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEcCAYAAAA2g5hwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n9H5MuPgAAAAASUVORK5CYII=\n" }, { "output_type": "pyout", "prompt_number": 48, "text": [ "-2*h + 2" ] } ], "prompt_number": 48 }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "Step 3: The Second Coin Toss" ] }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "Heads again" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We toss the coin a second time, and we get heads again. Now we repeat the process. The previous posterior distribution now becomes the new prior, which requires recalculating the normalization constant. However, the likelihood is the same as in the previous step.\n", "\n", "$$\n", "\\int_0^1 p(c=H|h) \\cdot f_2(h) \\thinspace dh = \\int_0^1 h \\cdot 2\\hat{h} \\thinspace dh = \\left. \\frac{2}{3} h^3 \\right |_0^1 = \\frac{2}{3}\n", "$$\n", "\n", "Which yields:\n", "\n", "$$\n", "f_3(\\hat{h}) := f_2(h=\\hat{h}|c=H) = \\frac{\\hat{h} \\cdot 2\\hat{h}}{\\frac{2}{3}} = 3\\hat{h}^2\n", "$$\n", "\n", "or simply: $f_3(h) = 3h^2$\n" ] }, { "cell_type": "code", "collapsed": false, "input": [ "f3 = h * f2 / integrate(h * f2, (h, 0, 1))\n", "fig = plot(f3, (h, 0, 1), title=\"Posterior distribution after observing $c=(\\mathrm{HH})$.\", xlabel=\"$h$\", ylabel=\"$f_3(h)$\")\n", "f3" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYsAAAEcCAYAAAA2g5hwAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nUEoWIiJSKCULEREp1P8HcuNzLF0G6bcAAAAASUVORK5CYII=\n" }, { "output_type": "pyout", "prompt_number": 49, "text": [ "3*h**2" ] } ], "prompt_number": 49 }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can see that the distribution now favors heads-biased coins even more strongly. Considering that two heads in a row is a common event even for fair coins, this may seem strange. But recall that we used a uniform prior at the very beginning, and did not favor fair coins over strongly biased coins at all. Consequently, the Bayesian inference process choses the most favorable distribution based on the uniform prior and the observed data. Had we chosen a prior that preferred fair coins over biased ones (incorporating external knowledge about the world, coin tossing, and gravity), we would see a more moderate result here. We will see in the folowing sections that, as more data comes in, the choice of the prior becomes less dominant on the overall result." ] }, { "cell_type": "heading", "level": 3, "metadata": {}, "source": [ "... or Tails" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Had we observed tails in the second toss, we would get a very different distribution. The integrals are easy to solve, so we will just give the result here as computed with Sympy. Now that tails is observed, the distribution misses the strong leaning towards heads-biased coins, as it must vanish at $h=1$ (i.e. all-heads)." ] }, { "cell_type": "code", "collapsed": false, "input": [ "f3_ = (1-h) * f2 / integrate((1-h) * f2, (h, 0, 1))\n", "plot(f3_, (h, 0, 1), title=\"Posterior distribution after observing $c=(\\mathrm{HT})$.\", xlabel=\"$h$\", ylabel=\"$f_3(h)$\")\n", "f3_" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEcCAYAAADUX4MJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n" }, { "output_type": "pyout", "prompt_number": 50, "text": [ "6*h*(-h + 1)" ] } ], "prompt_number": 50 }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "Step 4: The Third Coin Toss" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We give the results for three coin tosses (apart from identities and symmetries):" ] }, { "cell_type": "code", "collapsed": false, "input": [ "f4 = h * f3 / integrate(h * f3, (h, 0, 1))\n", "fig = plot(f4, (h, 0, 1), title=\"Posterior distribution after observing $c=(\\mathrm{HHH})$.\", xlabel=\"$h$\", ylabel=\"$f_4(h)$\")\n", "f4" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEcCAYAAADUX4MJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nJdRendQAAAAASUVORK5CYII=\n" }, { "output_type": "pyout", "prompt_number": 51, "text": [ "4*h**3" ] } ], "prompt_number": 51 }, { "cell_type": "code", "collapsed": false, "input": [ "f4_ = (1-h) * f3 / integrate((1-h) * f3, (h, 0, 1))\n", "fig = plot(f4_, (h, 0, 1), title=\"Posterior distribution after observing $c=(\\mathrm{HHT})$.\", xlabel=\"$h$\", ylabel=\"$f_4(h)$\")\n", "f4_" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEcCAYAAADUX4MJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhEUkYQghhLCIJAwhhBAW+X8eMA84KSabzAAAAABJRU5ErkJggg==\n" }, { "output_type": "pyout", "prompt_number": 52, "text": [ "12*h**2*(-h + 1)" ] } ], "prompt_number": 52 }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "Arbitrary Number of Coin Tosses" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The final posterior distribution of a Bayesian update process does not depend on the order of the updates, or even if we do the updates incrementally one after another, or all at once. We can use this to calculate the posterior after any number of coin tosses $c_1, \\..., c_n$. Assume that the number of heads is $H$ and the number of tails is $T$, then, choosing the uniform prior and any order of the data, we have:\n", "\n", "$$\n", "f_1(\\hat{h}) = 1\n", "$$\n", "\n", "$$\n", "p(c_{1,\\...,n}|h=\\hat{h}) = h^H (1-h)^T\n", "$$\n", "\n", "$$\n", "\\int_0^1 h^H (1-h)^T \\cdot 1 \\thinspace dh = \\text{ ?}\n", "$$\n", "\n", "Unfortunately, we now have to calculate the integral, which is a bit tricky using only elementary methods (try it, if you are up for a challenge!). Luckily, Sympy can spit out an acceptable answer in a couple of minutes. Note that we simplify the expression, and also move the expression to the denominator, where it will end up anyway:" ] }, { "cell_type": "code", "collapsed": false, "input": [ "H = Symbol(\"H\", positive=True, integer=True)\n", "T = Symbol(\"T\", positive=True, integer=True)\n", "# Warning, slow!\n", "#expr = integrate(h**H*(1-h)**T, (h, 0, 1))\n", "expr = gamma(H + 1)*hyper((-T, H + 1), (H + 2,), exp_polar(2*I*pi))/gamma(H + 2)\n", "beta = expr.simplify()\n", "beta" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "pyout", "prompt_number": 53, "text": [ "gamma(H + 1)*gamma(T + 1)/gamma(H + T + 2)" ] } ], "prompt_number": 53 }, { "cell_type": "markdown", "metadata": {}, "source": [ "So, we end up with:\n", "\n", "$$\n", "\\int_0^1 h^H (1-h)^T \\cdot 1 \\thinspace dh = \\frac{1}{B(H+1,T+1)}\n", "$$\n", "\n", "with\n", "\n", "$$\n", "B(p, q) := \\frac{\\Gamma(p) \\Gamma(q)}{\\Gamma(p+q)}\n", "$$\n", "\n", "and consequently:\n", "\n", "$$\n", "f(h) = \\frac{1}{B(H+1,T+1)}h^H(1-h)^T\n", "$$\n", "\n", "Please say hello to the [Beta-Distribution](http://en.wikipedia.org/wiki/Beta_distribution)!\n", "\n", "Let's look at this function for a couple of examples:" ] }, { "cell_type": "code", "collapsed": false, "input": [ "f = (1/beta) * h**H * (1-h)**T\n", "fig = plot(f.subs({H:2,T:1}), (h, 0, 1), title=\"Posterior distribution after observing $c=(\\mathrm{HHT})$.\", xlabel=\"$h$\", ylabel=\"$f(h)$\")\n", "fig = plot(f.subs({H:10,T:1}), (h, 0, 1), title=\"Posterior distribution after observing $c=(H=10,T)$.\", xlabel=\"$h$\", ylabel=\"$f(h)$\")\n", "fig = plot(f.subs({H:10,T:10}), (h, 0, 1), title=\"Posterior distribution after observing $c=(H=10,T=10)$.\", xlabel=\"$h$\", ylabel=\"$f(h)$\")\n", "fig = plot(f.subs({H:100,T:50}), (h, 0, 1), title=\"Posterior distribution after observing $c=(H=100,T=50)$.\", xlabel=\"$h$\", ylabel=\"$f(h)$\")\n" ], "language": "python", "metadata": {}, "outputs": [ { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEcCAYAAADUX4MJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n" }, { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEcCAYAAADKlrO6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBEREVo6JgIjIyv0/j5Yob64Ex2IAAAAASUVORK5CYII=\n" }, { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYwAAAEcCAYAAADUX4MJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n" }, { "output_type": "display_data", "png": "iVBORw0KGgoAAAANSUhEUgAAAYgAAAEcCAYAAADdtCNzAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n" } ], "prompt_number": 58 }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "Prior Distribution, revisited" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "To illustrate what can go wrong if an unreasonable prior distribution is chosen, imagine we were to use the Dirac delta function $f(h) = \\delta(h_0 - h)$ as prior distribution. In this case, the integral in the denominator would collaps to $p(c|h=\\hat{h})$, which would cancel out with the likelihood, leaving the prior distribution unchanged in the posterior. We can understand this intuitively: The delta function represents an absolute prejudice in the value of $h$, and this belief is so strong that no factual data can change it. In this sense, the Bayesian inference process works as expected. To avoid such a situation, the prior should give at least some probability mass to any reasonable outcome, and let the data speak for itself. We could see above that after a coin toss facing heads, the posterior vanishes at $0$, which means that the Bayesian inference process excluded the possibility for an all-tails coin. Likewise, after a coin toss facing tails, all-heads coins are excluded. No harm was done by including those possibilities in the prior, and the inference process quickly adapted to the data and excluded impossible choices." ] }, { "cell_type": "heading", "level": 2, "metadata": {}, "source": [ "What just happened" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We saw that starting with a uniform prior, the Baysian inference process can give us a lot of information about the certainty of a random variable, even after just a couple of observations. The result can be improved incrementally when new data arrives, or in batches. For a coin flip, the result is a Beta distribution, where the parameters indicate the number of heads and tails in the data. The choice of the prior is significant, but as more data arrives, it fades into the background. A strong bias in the prior that is not justified by the model may be hard to overcome.\n", "\n", "A limitation of the process is that it assumes i.i.d. (independent and identically distributed) random variables. If there was some correlation in the data, or the bias of the coin changes over time, we would prefer to look at some other method (for example Markov Chain).\n", "\n", "Bayesian inference is one of the foundations of machine learning, where it has many applications, for example [testing emails for spam](http://en.wikipedia.org/wiki/Bayesian_spam_filtering).\n" ] }, { "cell_type": "code", "collapsed": false, "input": [], "language": "python", "metadata": {}, "outputs": [] } ], "metadata": {} } ] }