2015-07-30-sampling-from-a-hierarchical-dirichlet-process.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "__Note__: This is best viewed on [NBViewer](http://nbviewer.ipython.org/github/tdhopper/stigler-diet/blob/master/content/articles/2015-07-30-sampling-from-a-hierarchical-dirichlet-process.ipynb). It is part of a series on [Dirichlet Processes and Nonparametric Bayes](https://github.com/tdhopper/notes-on-dirichlet-processes).\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Sampling from a Hierarchical Diriclet Process\n",
    "\n",
    "[As we saw earlier](http://stiglerdiet.com/blog/2015/Jul/28/dirichlet-distribution-and-dirichlet-process/) the Dirichlet process describes the _distribution_ of a random probability distribution. The Dirichlet process takes two parameters: a base distribution $H_0$ and a dispersion parameter $\\alpha$. A sample from the Dirichlet process is itself a probability distribution that _looks like_ $H_0$. On average, the larger $\\alpha$ is, the closer a sample from $\\text{DP}(\\alpha H_0)$ will be to $H_0$.\n",
    "\n",
    "Suppose we're feeling masochistic and want to input a distribution sampled from a Dirichlet process as base distribution to a new Dirichlet process. (It will turn out that there are good reasons for this!) Conceptually this makes sense. But can we construct such a thing in practice? Said another way, can we build a sampler that will draw samples from a probability distribution drawn from these nested Dirichlet processes? We might initially try construct a sample (a probability distribution) from the first Dirichlet process before feeding it into the second.\n",
    "\n",
    "But recall that fully constructing a sample (a probability distribution!) from a Dirichlet process would require drawing a countably infinite number of samples from $H_0$ and from the beta distribution to generate the weights. This would take forever, even with Hadoop!\n",
    "\n",
    "[Dan Roy, et al](http://danroy.org/papers/RoyManGooTen-ICMLNPB-2008.pdf) helpfully described a technique of using _stochastic memoization_ to construct a distribution sampled from a Dirichlet process in a just-in-time manner. This process provides us with the equivalent of the [Scipy `rvs`](http://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.rv_continuous.rvs.html) method for the sampled distribution.  Stochastic memoization is equivalent to the [Chinese restaurant process](http://www.cs.princeton.edu/courses/archive/fall07/cos597C/scribe/20070921.pdf): sometimes you get seated an an occupied table (i.e. sometimes you're given a sample you've seen before) and sometimes you're put at a new table (given a unique sample). \n",
    "\n",
    "Here is our memoization class again:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from numpy.random import choice \n",
    "from scipy.stats import beta\n",
    "\n",
    "class DirichletProcessSample():\n",
    "    def __init__(self, base_measure, alpha):\n",
    "        self.base_measure = base_measure\n",
    "        self.alpha = alpha\n",
    "        \n",
    "        self.cache = []\n",
    "        self.weights = []\n",
    "        self.total_stick_used = 0.\n",
    "\n",
    "    def __call__(self):\n",
    "        remaining = 1.0 - self.total_stick_used\n",
    "        i = DirichletProcessSample.roll_die(self.weights + [remaining])\n",
    "        if i is not None and i < len(self.weights) :\n",
    "            return self.cache[i]\n",
    "        else:\n",
    "            stick_piece = beta(1, self.alpha).rvs() * remaining\n",
    "            self.total_stick_used += stick_piece\n",
    "            self.weights.append(stick_piece)\n",
    "            new_value = self.base_measure()\n",
    "            self.cache.append(new_value)\n",
    "            return new_value\n",
    "        \n",
    "    @staticmethod \n",
    "    def roll_die(weights):\n",
    "        if weights:\n",
    "            return choice(range(len(weights)), p=weights)\n",
    "        else:\n",
    "            return None"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's illustrate again with a standard normal base measure. We can construct a function `base_measure` that generates samples from it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from scipy.stats import norm\n",
    "\n",
    "base_measure = lambda: norm().rvs() "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Because the normal distribution has continuous support, we can generate samples from it forever and we will never see the same sample twice (in theory). We can illustrate this by drawing from the distribution ten thousand times and seeing that we get ten thousand unique values."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of unique samples after 10000 draws: 10000\n"
     ]
    }
   ],
   "source": [
    "from pandas import Series\n",
    "\n",
    "ndraws = 10000\n",
    "print \"Number of unique samples after {} draws:\".format(ndraws), \n",
    "draws = Series([base_measure() for _ in range(ndraws)])\n",
    "print draws.unique().size"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "However, when we feed the base measure through the stochastic memoization procedure and then sample, we get many duplicate samples. The number of unique samples goes down as $\\alpha$ increases."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of unique samples after 10000 draws: 446\n"
     ]
    }
   ],
   "source": [
    "norm_dp = DirichletProcessSample(base_measure, alpha=100)\n",
    "\n",
    "print \"Number of unique samples after {} draws:\".format(ndraws), \n",
    "dp_draws = Series([norm_dp() for _ in range(ndraws)])\n",
    "print dp_draws.unique().size"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "At this point, we have a function `dp_draws` that returns samples from a probability distribution (specifically, a probability distribution sampled from $\\text{DP}(\\alpha H_0)$). We can use `dp_draws` as a base distribution for another Dirichlet process!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "norm_hdp = DirichletProcessSample(norm_dp, alpha=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "How do we interpret this? `norm_dp` is a sampler from a probability distribution that looks like the standard normal distribution. `norm_hdp` is a sampler from a probability distribution that \"looks like\" the distribution `norm_dp` samples from. \n",
    "\n",
    "Here is a histogram of samples drawn from `norm_dp`, our first sampler."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXsAAAEKCAYAAADzQPVvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nNzPrARd7M7MecLE3M+uB/w+vOVpp9+JCbAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112629290>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "pd.Series(norm_dp() for _ in range(10000)).hist()\n",
    "_=plt.title(\"Histogram of Samples from norm_dp\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And here is a histogram for samples drawn from `norm_hdp`, our second sampler."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXwAAAEKCAYAAAARnO4WAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nzDrCDd/MrCP+P04IQF/CeLpLAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x11137fed0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.Series(norm_hdp() for _ in range(10000)).hist()\n",
    "_=plt.title(\"Histogram of Samples from norm_hdp\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The second plot doesn't look very much like the first! The level to which a sample from a Dirichlet process approximates the base distribution is a function of the dispersion parameter $\\alpha$. Because I set $\\alpha=10$ (which is relatively small), the approximation is fairly course. In terms of memoization, a small $\\alpha$ value means the stochastic memoizer will more frequently reuse values already seen instead of drawing new ones.\n",
    "\n",
    "This nesting procedure, where a sample from one Dirichlet process is fed into another Dirichlet process as a base distribution, is more than just a curiousity. It is known as a [Hierarchical Dirichlet Process, and it plays an important role in the study of Bayesian Nonparametrics](http://www.cs.berkeley.edu/~jordan/papers/hdp.pdf) (more on this in a future post). \n",
    "\n",
    "Without the stochastic memoization framework, constructing a sampler for a hierarchical Dirichlet process is a daunting task. We want to be able to draw samples from a distribution drawn from the second level Dirichlet process. However, to be able to do that, we need to be able to draw samples from a distribution sampled from a _base distribution of the second-level Dirichlet process_: this base distribution is a _distribution drawn from the first-level Dirichlet process_.\n",
    "\n",
    "Though it appeared that we would need to be able to fully construct the first level sample (by drawing a countably infinite number of samples from the first-level base distribution). However, stochastic memoization allows us to only construct the first distribution just-in-time as it is needed at the second-level.\n",
    "\n",
    "We can define a Python class to encapsulate the Hierarchical Dirichlet Process as a base class of the Dirichlet process."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "class HierarchicalDirichletProcessSample(DirichletProcessSample):\n",
    "    def __init__(self, base_measure, alpha1, alpha2):\n",
    "        first_level_dp = DirichletProcessSample(base_measure, alpha1)\n",
    "        self.second_level_dp = DirichletProcessSample(first_level_dp, alpha2)\n",
    "\n",
    "    def __call__(self):\n",
    "        return self.second_level_dp()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Since the Hierarchical DP is a Dirichlet Process inside of Dirichlet process, we must provide it with both a first and second level $\\alpha$ value."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "norm_hdp = HierarchicalDirichletProcessSample(base_measure, alpha1=10, alpha2=20)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can sample directly from the probability distribution drawn from the Hierarchical Dirichlet Process."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 170,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaQAAAEKCAYAAABdWiGrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XmcJWV97/HPl01ls+FqBoatUQaBSDKIMC5XaWUJmAiY\nRIUo0mpMyAQl5poIJrkQvRnRxAXM1dz7AgWMQIgLLzAwLDrtLn0h02zjhJnERmZgBtlBjEL43T/q\nOUxx5pzu032Wqofzfb9e/Zp6av3VU1XnqXp+dc4oIjAzM6vaFlUHYGZmBm6QzMysJtwgmZlZLbhB\nMjOzWnCDZGZmteAGyczMaqEnDZKk2yS9thfrypWkN0m6S9Kjkn696nhakTQh6d09XucCSd+W9Iik\nv+3luucZzwWSPpKGXyNpdQ/XfZWkk9LwuKTv9HDdb5N0Ta/WN8u2npL0okFsq1N1O49aqcvn3Gzn\nXvk87dc2+mXWBknStKTDm8Y9I9iIeGlEfHuW9YymC+HZ+lT2d8DSiNghIm6uOpg2Iv310h8A90bE\njhHxZz1e93w8vY8R8Z2I2G+2BSSdJemLs6444g0RMet8HWxvs2shIr4UEb/R7bozVul51KvPuTro\n1XnaTun8fTT9bZB0paQjmuablvR4aZ4vSNpupnV30jj0+kNMPVzXppVKW/ZjvR1uW8CewKqqYqjQ\nXsCP2k2s6Lj09BxT0st1Nlbdh3V2pcIbxqrPo37crAEgaat5LJPDjfvzI2IH4NeA64CvSTq5ND2A\n30rzvAx4OfCXM61wvjv9jAOXWsLXp+FDJd0o6eHUKv5dmq1xZ/FQajGXpOv8L9PyGyVdKGnH0nrf\nIelOSfeV5mts5yxJX5b0RUkPAydLOkTSDyQ9KOluSZ+RtHVpfU9J+iNJa1LXwIclvTgt85CkS8vz\nN+1jy1glPQd4FNgSuFnSmjbLfyot97CkWyT9ahr/m5JWpvE/kXRmaZnGnch4mna/pFPSft6S9vMz\npfnHJX0v7fdDkn7UqK82Mb1L0ipJD0haLmnP2eJtWv4C4B3An6f6PLzNcVko6YoU/xpJv19ax1mS\n/jnN/0ja1iJJZ6Tt3ynpyBn24SBJ/5qWvRR4bmnamKS7SuUPSlqX5l0t6fWSjgbOAN6azsuVad4J\nSf9L0veAx4AXafMuT7WrazXdceuZT2Hla+ERSa9Q0924pFdJ+n9p3ZOSXlmaNpHO3e+m5a+R9N9m\nqKM/S9fDOknvaj6Gkj6nopvnMWBslnPyQkl/moZ3S+fn0lR+saT7S3W/TtKfpuN4t6TxNvFdQMXn\nURutPucOT8OSdLqktSo+n/5J0k5pWuO6fZekO4Hr0/h/lnRPOqbfknTALMdhD0lflXRv2sZnmuL5\nWxXX7n+k87gx/hnnqaT3qLjOH5F0u6SD0vhG/I3xx8+xfopKirg3Is4FzgI+1maeu4HlwEtnW9mM\nf8CPgcObxo0D32ma5/Vp+AfA29LwtsCSNLwX8BSwRWm5dwFrgFFgO+ArwEVp2gEUH/SvArYG/hb4\nZWk7Z6Xysan8XIpW+FCKhnYviieW00rbewr4GrB9Wv8vgG+m7e8I3A68o009tI21tO4XtVn2N4Ab\ngR1T+SXALmn4MOBX0/CBwAbguFQeTev9LLANcGSK+WvAC4CFwEbgtaXj8gRwGkUD+RbgIWAkTV8B\nvCsNH5f25yWpvv4C+N5s8bbYty8AHy6VWx2XbwN/n/bh14F7gdeV5v952rctgQuBaYpGYkvg94H/\naLPtbYA7S/v7O2nbH07Tx4C7Svvwk1K979k4XsCZ5WOZxk2kOPZP9bNVU/3NVtdPXxOlbXxxhmth\nnHRNATsDDwJvS9s+AXgA2KkU2xpgn1S/K4CPtqmjoynOqQMorseLKZ2rwAUp7lem8nOY+Zx8J3BF\nGv49YC1waeka+Vqp7p9Ix3dL4BjgZxR31bU6j+b5OXca8H2Ka3Br4B+Ai5uu2wuA5wHPKa1vuzT/\np4CVpXU3H4dtgZuBTzTWAbyqtJ5fAu+meMo+BVhfWlf5PH0zsA44OJVfDOyZhn+XTdfDWyhuvBa0\n2vememns3xZN41+Uxr+kuU6BPYDbgL9udwwioqMGaZqiYXiw9Pcz4NttDtS30snxgtl2AvgGcEqp\nvG+q6C2B/wl8qTTteRQfxuUGaWKW2P8E+Gqp/FTjgKfyjcCflcp/B3yqzbraxbpFad3tGqTXAf8G\nLGk+iC3m/TTwyaY627U0/T7gzaXyl0mNbjqJ1jet7wbg7S1O1Ksbw6m8RTque84x3i8AH2n6IJko\nlfcAngS2K41bBnyhNP81pWlvTOebUnmHVAc7ttj2a1vs7/do3SDtQ9F4Hw5s3bTMWaTGoumiPqvF\nuHKD1KquGzdjzQ3S09ug9bUwzqYG6STgh03r/j5wcimOD5Wm/RFwdZvj83lgWam8iM0bpAvmcE6+\nmKJxFPA5itxPo44vBP6kVPePN+3jRuDQup1Hafo0c/ucW9V0fHclfR6Uju/oDHU6kubZodVxAF5J\n0eBudv2lc2VNqbxtWtevtDhPrwHeO9PxLa1nJZtuAJ4+H1vM19i/5gbpuZQ+Y5vqdJriZuI5M8XQ\naQ7puIjYqfEHLKV9//e7KT6sf5S6Gn5zhnXvSnGH2/ATijvRBWnauqeDiPg5cH/T8uvKBUn7Svp6\neix+GPgboLkrY2Np+OctytvPI9YZRcQKioPxv4GNkv6PpB1SzEskrUiP5Q8BfziPmMuJwvVNy96Z\nYm+2F3COim6/B9lUtwtnirdD5eOyEHggIn5WGvcTYLdS+d7S8M+B+yKd0akMrY/LQlrv72YiYi3F\nDcpZFPt0iaRW9VJ21yzTW2174SzLdGIhRR3NtO4NpeHZztvyfjSvN5qmz3hORsS/U3xQLwZeA3wd\nuFvSvhQ3CN8qrer+iHiqVH58hjhbGdR5BHP/nBulyJk0rp9VFA1m+fOg3F28haSzUxfZwxSNGxQ9\nHY3tl/d3D+DOpvore/r4R8TjM+zb7sC/t1qBipTIytI+vJTNP3vmonEsHmiExqY6HY2IUyPiFzOt\nYL45pLbJ2IhYGxG/FxEvpOhP/LKk59E6YXg3xYFt2JPioG4A7qGozGKDxTqaK6t5nZ+jODH2iYjn\nU3RD9So52C7WjS3nbhIRn4mIl1N0newLNN4kuhi4HNg9IkYoHv27iXm3pvJeFLE3+wnwB+ULMCK2\ni4gfzhLvbIJnHpe7gZ0llS+WPWm6mZine2i9v60Di7gkIl6T5gk29Xe3OjdnGt8wU13/jGfeKOwy\nh/WuZ/P92IvNG8BO3ENR3w17tpuxZLZz8lsUXUFbR5Eb+BbFHfVOwNQ8YmxlkOdROzO9dPIT4Oim\n62fbiLinNE85/rcBx1J0YT0f2LvFNsrz3wXsqe5f5riLonfgGSTtBfxf4I+BnVMDfBsz7/Ns3gRs\njIh/m+8Kev4mh6S3S3phKj5MUclPAT9N/764NPslwPtTEnB7ikfwS9NdwVeAN0p6paRtKO5sZ6us\n7SkeER+XtB9FV8asIbcZbjZTrDNvQHp5uuvcmuIu8T+B/yrF/GBE/FLSoRT98rN9YM20D78i6X2S\ntpb0ZmA/4KoWy/wD8KFGYlXS89P8s8U707Y3K0fEXRTdTR+V9BxJv0aRa/jHue1iS98Hnizt728D\nh7QMsnh6fr2Kl1B+wTP3aQMwKm32Jt1s59tMdT0FnCBpK0kvp8hvNY5rq2uh7GpgX0knpuXfmtb9\n9TnE1nAZMC5pf0nbUuSyylqtZ7Zz8lvAqWx6OWMilb9TeiKZqyrPo/n4B2CZ0otAkl4o6dgZ5t+e\n4rx7QMWrz8uapjfv/w0UNxNnS9pW0nMlvWoecZ4HfEDSy1TYJ8W8HcUxvQ/YQtI7me2Fg80Jnv4O\n2akUaZYz5hHj07p5y67difcbwG2SHqVI3J0QEb9Ij5V/A3wvPSIeStG//UWKE/s/KD783gsQEben\n4Usp7o4epXgkbzzytYrhAxQXzyMUrf+lTfO0irl5erv9ahvrDOtu2DHF8wBFX+p9FC9pQNEt8GFJ\njwB/BfzTDPG1U57nBoo8wU+BjwC/ExEPbrZAxOUUTwiXpi6EWymO3Wzxttr2bHV4IsXT5d3AV4H/\nGRHfnGH+2cqNfXgC+G2Ku/P7KRKzX2mz7HOAj1LUyz0UXSWNi+ef07/3S7pxtu2Wpv2Q9nX9VxQN\nzoMUN1NfKsVdvhYekLSEZ35/6n7gt4D/QVH3H6B4fbbRFdIcW9vzNiKWU+SAvgncQZELnW3Z2c7J\nb1N8wDYapO9R5Hibv6Mzl8apsvNoDjGVnQNcAVyb6ukHFC9UtdvWRRTdruspnkR+wAz7m25030jx\ndPMTiiedt8wQV7vj/2WKc+1iis/Fr1K8HLOK4oWJH1DckL0U+G67eNp4SMUbgbdQvDzzuxFxwSzL\nzEjzv6FJKygeKW8E1kXEGyXtTHEC70XxYfaWiHgozXsGxV3NfwHvi4hr0/iDKZJ6zwWuiojTWmxn\ne4qLe5+IaJknGHYqXqt9d+qWMjPLSi+67E6jyNs0WrbTgesiYl+Ku7HTAVLX0FspchJHA58tdZF8\njuKDdBGwSOmdeklvTI+r21G8AXeLGyMzs2enrhokSbsDb6Dop2w0LsdSvP5J+rfxZavjgEsi4omI\nmKb4/sKS9KbTDhExmea7qLTMsRSPuOspuj9O6CbeIdDJY7aZWS3N+SctmnyK4u2rHUvjFkRE482z\njWx6DXIhRZ97wzqKt5Se4JlvyqxP44mI9wDv6TLGoRERF7LpZsDMLCvzfkKS9FsUP4a4kjZv/KQ3\nbnzHbmZms+rmCelVwLGS3kDxMsKOKn6ra6OkXSJiQ+qOa3xZbT3Fl70adqd4MlpP6ftGabjl9y0k\nuXEzM5ujiKjdD/m2Mu8npIj4UETsERF7U+R2vhkRJ1G8Cnlymu1kii/YkcafIGkbSXtTvC47GREb\ngEfS915E8bMpl9NGJz+BUfXfmWeeWXkMz4YYHafjrPtfDnHmpNscUlljz88GLlPxa7PTpHfnI2KV\npMvY9BMbS2NTbS1l0w8RXhXFdyeyNT09XXUIs8ohRnCcveY4eyuXOHPRkwYpIr5F+g2rKL68d0Sb\n+Zax+TeUiYibKH5V2MzMhlQO/wlUdsbHx6sOYVY5xAiOs9ccZ2/lEmcuuv6lhkGSFDnFa2ZWNUnE\ns/2lBmtvYmKi6hBmlUOM4Dh7zXH2Vi5x5sINkpmZ1YK77MzMnsXcZWdmZjZHbpD6IId+5RxiBMfZ\na46zt3KJMxdukMzMrBacQzIzexZzDsnMzGyO3CD1QQ79yjnECI6z1xxnb+USZy7cIJmZWS04h2Rm\n9iyWUw6pl//9hM2g+K+equOG3Mzqzl12fdC+Xzkq+ptLjPXiOHvLcfZWLnHmwg2SmZnVgnNIA1J0\n2VUVu9xlZzakcsoh+QnJzMxqwQ1SH+TQr5xDjOA4e81x9lYuceaiqwZJ0nMl3SBpStIqSR9N48+S\ntE7SyvR3TGmZMyStkbRa0lGl8QdLujVNO6ebuMzMLD9d55AkbRsRj0vaCvgu8AHgcODRiPhk07wH\nABcDhwC7AdcDiyIiJE0Cp0bEpKSrgHMjYnnT8s4hzW/rziGZDamhyiFFxONpcBtgS+DBVG5VAccB\nl0TEExExDawFlkjaFdghIibTfBcBx3cbm5mZ5aPrBknSFpKmgI3Aioi4PU16r6SbJZ0vaSSNWwis\nKy2+juJJqXn8+jQ+Szn0K+cQIzjOXnOcvZVLnLnoxRPSUxGxGNgdeK2kMeBzwN7AYuAe4BPdbsfM\nzJ7devbTQRHxsKR/AV4eERON8ZLOA65MxfXAHqXFdqd4Mlqfhsvj17fazvj4OKOjowCMjIywePFi\nxsbGgE13K3Utw0T6d9BlNotnbGys8vrotNwq/rqVXZ+uzzrFMzExwfT0NLnp6qUGSS8AnoyIhyQ9\nD7gG+Gvg9ojYkOZ5P3BIRPxe6aWGQ9n0UsM+6aWGG4D3AZPAv+CXGnq5db/UYDakhumlhl2Bb6Yc\n0g3AlRHxDeDjkm6RdDNwGPB+gIhYBVwGrAKuBpaWWpilwHnAGmBtc2OUk+Y7pzrKIUZwnL3mOHsr\nlzhz0VWXXUTcCrysxfh3zLDMMmBZi/E3AQd2E4+ZmeXLv2U3IO6yM7MqDFOXnZmZWU+4QeqDHPqV\nc4gRHGevOc7eyiXOXLhBMjOzWnAOaUCcQzKzKjiHZGZmNkdukPogh37lHGIEx9lrjrO3cokzF26Q\nzMysFpxDGhDnkAarqO9qDFtdW73llEPq2Y+rmtVPFQ1DFte9WS25y64PcuhXziFGyCfO5l9Wr6tc\n6tNxDic3SGZmVgvOIQ2Ic0iDVV19D19dW73llEPyE5KZmdWCG6Q+yKFfOYcYIZ84nUPqLcc5nNwg\nmZlZLTiHNCDOIQ2Wc0hmBeeQzMzM5sgNUh/k0K+cQ4yQT5zOIfWW4xxOXTVIkp4r6QZJU5JWSfpo\nGr+zpOsk3SHpWkkjpWXOkLRG0mpJR5XGHyzp1jTtnG7iMjOz/HSdQ5K0bUQ8Lmkr4LvAB4Bjgfsi\n4uOSPgjsFBGnSzoAuBg4BNgNuB5YFBEhaRI4NSImJV0FnBsRy5u25RzS/LY+dHkN55DMCkOVQ4qI\nx9PgNsCWwIMUDdKFafyFwPFp+Djgkoh4IiKmgbXAEkm7AjtExGSa76LSMmZmNgS6bpAkbSFpCtgI\nrIiI24EFEbExzbIRWJCGFwLrSouvo3hSah6/Po3PUg79yjnECPnE6RxSbznO4dT1r31HxFPAYknP\nB66R9Lqm6SGpZ30Y4+PjjI6OAjAyMsLixYsZGxsDNp0cVZcbmqdv+tAadLl1PDmUp6am5r2863vz\ncjf16fLm5TrWZ2N4enqa3PT0e0iS/gr4OfD7wFhEbEjdcSsiYj9JpwNExNlp/uXAmcCdaZ790/gT\ngcMi4pSm9TuHNL+tD11ewzkks8LQ5JAkvaDxBp2k5wFHAiuBK4CT02wnA5en4SuAEyRtI2lvYBEw\nGREbgEckLVHxSXJSaRkzMxsC3eaQdgW+mXJINwBXRsQ3gLOBIyXdAbw+lYmIVcBlwCrgamBp6ZFn\nKXAesAZY2/yGXU6au+7qKIcYIZ84nUPqLcc5nLrKIUXErcDLWox/ADiizTLLgGUtxt8EHNhNPGZm\nli//lt2AOIc0WM4hmRWGJodkZmbWK26Q+iCHfuUcYoR84nQOqbcc53Byg2RmZrXgHNKAOIc0WM4h\nmRWcQzIzM5sjN0h9kEO/cg4xQj5xOofUW45zOLlBMjOzWnAOaUCcQxos55DMCs4hmZmZzZEbpD7I\noV85hxghnzidQ+otxzmc3CCZmVktOIc0IM4hDZZzSGYF55DMzMzmyA1SH+TQr5xDjJBPnM4h9Zbj\nHE5ukMzMrBacQxoQ55AGyzkks4JzSGZmZnPkBqkPcuhXziFGyCdO55B6y3EOp64aJEl7SFoh6XZJ\nt0l6Xxp/lqR1klamv2NKy5whaY2k1ZKOKo0/WNKtado53cRlZmb56SqHJGkXYJeImJK0PXATcDzw\nFuDRiPhk0/wHABcDhwC7AdcDiyIiJE0Cp0bEpKSrgHMjYnnT8s4hzW/rQ5fXcA7JrDA0OaSI2BAR\nU2n4MeBHFA0NQKsKOA64JCKeiIhpYC2wRNKuwA4RMZnmu4iiYTMzsyHRsxySpFHgIOCHadR7Jd0s\n6XxJI2ncQmBdabF1FA1Y8/j1bGrYspNDv3IOMUI+cTqH1FuOczht1YuVpO66LwOnRcRjkj4HfDhN\n/gjwCeDdvdjW+Pg4o6OjAIyMjLB48WLGxsaATSdH1eWG5umbPrQGXW4dTw7lqampeS/v+t683E19\nurx5uY712Rienp4mN11/D0nS1sDXgasj4tMtpo8CV0bEgZJOB4iIs9O05cCZwJ3AiojYP40/ETgs\nIk5pWpdzSPPb+tDlNZxDMisMTQ5JxVV/PrCq3BilnFDDm4Bb0/AVwAmStpG0N7AImIyIDcAjkpak\ndZ4EXN5NbGZmlpduc0ivBt4OvK7pFe+PSbpF0s3AYcD7ASJiFXAZsAq4GlhaeuRZCpwHrAHWNr9h\nl5Pmrrs6yiFGyCdO55B6y3EOp65ySBHxXVo3alfPsMwyYFmL8TcBB3YTj5mZ5cu/ZTcgziENlnNI\nZoWhySGZmZn1ihukPsihXzmHGCGfOJ1D6i3HOZzcIJmZWS04hzQgziENlnNIZgXnkMzMzObIDVIf\n5NCvnEOMkE+cziH1luMcTm6QzMysFpxDGhDnkAbLOSSzgnNIZmZmc+QGqQ9y6FfOIUbIJ07nkHrL\ncQ4nN0hmZlYLziENiHNIg+UcklnBOSQzM7M5coPUBzn0K+cQI+QTp3NIveU4h5MbJDMzqwXnkAbE\nOaTBcg7JrOAckpmZ2Ry5QeqDHPqVc4gR8onTOaTecpzDqasGSdIeklZIul3SbZLel8bvLOk6SXdI\nulbSSGmZMyStkbRa0lGl8QdLujVNO6ebuMzMLD9d5ZAk7QLsEhFTkrYHbgKOB94J3BcRH5f0QWCn\niDhd0gHAxcAhwG7A9cCiiAhJk8CpETEp6Srg3IhY3rQ955Dmt/Why2s4h2RWGJocUkRsiIipNPwY\n8COKhuZY4MI024UUjRTAccAlEfFEREwDa4ElknYFdoiIyTTfRaVlzMxsCPQshyRpFDgIuAFYEBEb\n06SNwII0vBBYV1psHUUD1jx+fRqfpRz6lXOIEfKJ0zmk3nKcw2mrXqwkddd9BTgtIh4tuksKqTuu\nZ30Y4+PjjI6OAjAyMsLixYsZGxsDNp0cVZcbmqdv+tAadLl1PDmUp6am5r2863vzcjf16fLm5TrW\nZ2N4enqa3HT9PSRJWwNfB66OiE+ncauBsYjYkLrjVkTEfpJOB4iIs9N8y4EzgTvTPPun8ScCh0XE\nKU3bcg5pflsfuryGc0hmhaHJIam46s8HVjUao+QK4OQ0fDJweWn8CZK2kbQ3sAiYjIgNwCOSlqR1\nnlRaxszMhkC3OaRXA28HXidpZfo7GjgbOFLSHcDrU5mIWAVcBqwCrgaWlh55lgLnAWuAtc1v2OWk\nueuujnKIEfKJ0zmk3nKcw6mrHFJEfJf2jdoRbZZZBixrMf4m4MBu4jEzs3z5t+wGxDmkwXIOyaww\nNDkkMzOzXnGD1Ac59CvnECPkE6dzSL3lOIeTGyQzM6sF55AGxDmkwXIOyazgHJKZmdkcuUHqgxz6\nlXOIEfKJ0zmk3nKcw8kNkpmZ1YJzSAPiHNJgOYdkVnAOyczMbI7cIPVBDv3KOcQI+cTpHFJvOc7h\n5AbJzMxqwTmkAXEOabCcQzIrOIdkZmY2R26Q+iCHfuUcYoR84nQOqbcc53Byg2RmZrXgHNKAOIc0\nWM4hmRWcQzIzM5sjN0h9kEO/cg4xQj5xOofUW45zOHXVIEn6vKSNkm4tjTtL0jpJK9PfMaVpZ0ha\nI2m1pKNK4w+WdGuadk43MZmZWZ66yiFJeg3wGHBRRByYxp0JPBoRn2ya9wDgYuAQYDfgemBRRISk\nSeDUiJiUdBVwbkQsb7E955Dmt/Why2s4h2RWGJocUkR8B3iwxaRWO38ccElEPBER08BaYImkXYEd\nImIyzXcRcHw3cZmZWX76lUN6r6SbJZ0vaSSNWwisK82zjuJJqXn8+jQ+Wzn0K+cQI+QTp3NIveU4\nh9NWfVjn54APp+GPAJ8A3t2rlY+PjzM6OgrAyMgIixcvZmxsDNh0clRdbmievulDa9Dl1vHkUJ6a\nmpr38q7vzcvd1KfLm5frWJ+N4enpaXLT9feQJI0CVzZySO2mSTodICLOTtOWA2cCdwIrImL/NP5E\n4LCIOKXF+pxDmt/Why6v4RySWWFockitpJxQw5uAxht4VwAnSNpG0t7AImAyIjYAj0haouJT5CTg\n8l7HZWZm9dbta9+XAN8HXiLpLknvAj4m6RZJNwOHAe8HiIhVwGXAKuBqYGnpcWcpcB6wBljb6g27\nnDR33dVRDjFCPnE6h9RbjnM4dZVDiogTW4z+/AzzLwOWtRh/E7BZl5+ZmQ0P/5bdgDiHNFjOIZkV\nhjqHZGZmNh9ukPogh37lHGKEfOJ0Dqm3HOdwcoNkZma14BzSgDiHNFjOIZkVnEMyMzObIzdIfZBD\nv3IOMUI+cTqH1FuOczi5QTIzs1pwDmlAnEMaLOeQzAo55ZD68WvfVkPFB/Tg+cPZzDrlLrs+qGe/\ncjT9rWgxrtd/3atnXbYyUXUAHcmlPh3ncHKDZGZmteAc0oBUnUMatnyKc0hmhZxySH5CMjOzWnCD\n1Ad59CtPVB1AR/KoS3B99pbjHE5ukMzMrBacQxoQ55AGvGXnkMwA55DMzMzmzA1SH+TRrzxRdQAd\nyaMuwfXZW45zOHXVIEn6vKSNkm4tjdtZ0nWS7pB0raSR0rQzJK2RtFrSUaXxB0u6NU07p5uYzMws\nT13lkCS9BngMuCgiDkzjPg7cFxEfl/RBYKeIOF3SAcDFwCHAbsD1wKKICEmTwKkRMSnpKuDciFje\nYnvOIc1v6xVt2zkks6oNTQ4pIr4DPNg0+ljgwjR8IXB8Gj4OuCQinoiIaWAtsETSrsAOETGZ5ruo\ntIyZmQ2JfuSQFkTExjS8EViQhhcC60rzraN4Umoevz6Nz1Ye/coTVQfQkTzqElyfveU4h1Nff+07\ndcf1tP9ifHyc0dFRAEZGRli8eDFjY2PAppOj6nJD8/RNH1qDLjPL9H6Vizropj6npqbmvXzV9V2X\n87FX9eny5uU61mdjeHp6mtx0/T0kSaPAlaUc0mpgLCI2pO64FRGxn6TTASLi7DTfcuBM4M40z/5p\n/InAYRFxSottOYc0v61XtG3nkMyqNjQ5pDauAE5OwycDl5fGnyBpG0l7A4uAyYjYADwiaYmKT5GT\nSsuYmdmQ6Pa170uA7wMvkXSXpHcCZwNHSroDeH0qExGrgMuAVcDVwNLS485S4DxgDbC21Rt2Ocmj\nX3mi6gA6kkddguuztxzncOoqhxQRJ7aZdESb+ZcBy1qMvwk4sJtYzMwsb/4tuwFxDmnAW3YOyQxw\nDsnMzGzPzSAFAAAJHklEQVTO3CD1QR79yhNVB9CRPOoSXJ+95TiHkxskMzOrBeeQBsQ5pAFv2Tkk\nM8A5JDMzszlzg9QHefQrT1QdQEfyqEtwffaW4xxObpDMzKwWnEMaEOeQBrxl55DMAOeQzMzM5swN\nUh/k0a88UXUAHcmjLsH12VuOczi5QTIzs1pwDmlAnEMa8JadQzIDnEMyMzObMzdIfZBHv/JE1QF0\nJI+6BNdnbznO4eQGyczMasE5pAFxDmnAW3YOyQxwDsnMzGzO+tYgSZqWdIuklZIm07idJV0n6Q5J\n10oaKc1/hqQ1klZLOqpfcQ1CHv3KE1UH0JE86hJcn73lOIdTP5+QAhiLiIMi4tA07nTguojYF/hG\nKiPpAOCtwAHA0cBnJfnpzcxsiPQthyTpx8DLI+L+0rjVwGERsVHSLsBEROwn6QzgqYj4WJpvOXBW\nRPywaZ3OIc1v6xVt2zkks6o5h1QI4HpJN0p6Txq3ICI2puGNwII0vBBYV1p2HbBbH2MzM7Oa6WeD\n9OqIOAg4BvhjSa8pT0yPOjPdSmZ7m5lHv/JE1QF0JI+6BNdnbznO4bRVv1YcEfekf38q6WvAocBG\nSbtExAZJuwL3ptnXA3uUFt89jdvM+Pg4o6OjAIyMjLB48WLGxsaATSdH1eWG5umbPrQGXWaW6f0q\nF3XQTX1OTU3Ne/mq67su52Ov6tPlzct1rM/G8PT0NLnpSw5J0rbAlhHxqKTtgGuBvwaOAO6PiI9J\nOh0YiYjT00sNF1M0WrsB1wP7NCeMnEOa99Yr2rZzSGZVyymH1K8npAXA14oPBbYCvhQR10q6EbhM\n0ruBaeAtABGxStJlwCrgSWBpti2PWUXS9VYJX67WC33JIUXEjyNicfp7aUR8NI1/ICKOiIh9I+Ko\niHiotMyyiNgnIvaLiGv6Edeg5NGvPFF1AB3Joy6hPvUZs/yt6GCeuf71Xi7HPZc4c9G3HJKZ2bNV\nlU+j8Ox9IvVv2Q2Ic0gD3vIQ5pCGcZ+rUvX1PJf6zimH5F9DMDOzWnCD1Ad59CtPVB1AR/KoSyjX\np6RK/uYaZ53leNyte84hmfVclV2zZvlyDmlAqu5zHrbcQpX5FB/nZ7+qr2fnkMzMzPrIDVIf5NH/\nPVF1AB3Joy4hl/rMJU4f9+HkBsnMzGrBOaQBqbrPedhyC84hDXa7uV6X81X19ewckpmZWR+5QeqD\nPPq/JwaylXp/J6eXJirY5nxMVB1AR/K4hiCX+syFGyTrs25/uHO+PwZqZrlxDmlAqu5zdj7l2b7d\nKrftHNKAt+4ckpmZWT8N1U8H7bbb3jz22H/2fTtPPvkLttrqOX3fTncmKP9X4/U1gePspQlyiHNi\nYqL0X9G3V/V/A5FLfeZiqBqke+/dwJNP3g48r89b+j7wqlL588Bf9nmbZsOqqq5Z67WhyiFtvfXz\nePLJB+h/g9Ts74H3Moy5Be/zMGzb3zcb9LadQzIzM+ujWjVIko6WtFrSGkkfrDqe+ZuoOoAOTFQd\nQIcmqg6gQxNVB9ChiaoD6Ii/hzScatMgSdqSom/raOAA4ERJ+1cb1XxNVR1AB3KIERxnr+UR59RU\nHnHmUp+5qE2DBBwKrI2I6Yh4ArgUOK7imObpoaoD6EAOMYLj7LU84nzooTzizKU+c1GnBmk34K5S\neV0aZ2ZmQ6BOr333/ZUVCXbc8XeALfu6nccfX8m22970dPmXv/wx/9n/rz/N0XTVAXRouuoAOjRd\ndQAdmq46gI5MT09XHUKHpqsO4FmlNq99S3oFcFZEHJ3KZwBPRcTHSvPUI1gzs4zk8tp3nRqkrYB/\nAw4H7gYmgRMj4keVBmZmZgNRmy67iHhS0qnANRR9aue7MTIzGx61eUIyM7PhVqe37J5B0psl3S7p\nvyS9bIb5piXdImmlpMlBxpi232mclX7pV9LOkq6TdIekayWNtJmvkvrspH4knZum3yzpoEHF1hTD\njHFKGpP0cKq/lZIG/iOGkj4vaaOkW2eYpw51OWOcNanLPSStSNf4bZLe12a+SuuzkzjrUJ+zioha\n/gH7AftS/A9tL5thvh8DO9c5ToouyLXAKLA1xbfp9h9wnB8H/jwNfxA4uy712Un9AG8ArkrDS4Af\nVnCsO4lzDLhi0LE1xfAa4CDg1jbTK6/LDuOsQ13uAixOw9tT5LnreG52Emfl9TnbX22fkCJidUTc\n0eHslb1B0mGcdfjS77HAhWn4QuD4GeYddH12Uj9Pxx8RNwAjkhYMNsyOj2OlbzRFxHeAB2eYpQ51\n2UmcUH1dboiIqTT8GPAjYGHTbJXXZ4dxQs1/pry2DdIcBHC9pBslvafqYNqow5d+F0TExjS8EWh3\nwVRRn53UT6t5du9zXM06iTOAV6Wum6skHTCw6DpXh7rsRK3qUtIoxRPdDU2TalWfM8RZq/pspdK3\n7CRdR/Go2exDEXFlh6t5dUTcI+mFwHWSVqc7r57pQZwDeXNkhjj/4hnBRMQM3+nqe3220Gn9NN/d\nDfqNnE6296/AHhHxuKRjgMspunTrpuq67ERt6lLS9sCXgdPSE8hmszSVK6nPWeKsTX22U2mDFBFH\n9mAd96R/fyrpaxTdKj39AO1BnOuBPUrlPSjuonpqpjhT8niXiNggaVfg3jbr6Ht9ttBJ/TTPs3sa\nN0izxhkRj5aGr5b0WUk7R8QDA4qxE3Woy1nVpS4lbQ18BfjHiLi8xSy1qM/Z4qxLfc4kly67lv2e\nkraVtEMa3g44Cmj7ZtEAtOufvRFYJGlU0jbAW4ErBhcWpO2dnIZPprg7eoYK67OT+rkCeEeK7RXA\nQ6UuyEGZNU5JC6Ti/9WWdCjFVytqc8EndajLWdWhLtP2zwdWRcSn28xWeX12Emcd6nNWVb9V0e4P\neBNFv+zPgQ3A1Wn8QuBf0vCLKN50mgJuA86oY5ypfAzFmy9rK4pzZ+B64A7gWmCkTvXZqn6APwT+\nsDTP36fpNzPDm5dVxgn8caq7KYr/y/4VFcR4CcWvnfwynZvvqmldzhhnTeryvwNPpRhWpr9j6laf\nncRZh/qc7c9fjDUzs1rIpcvOzMye5dwgmZlZLbhBMjOzWnCDZGZmteAGyczMasENkpmZ1YIbJDMz\nqwU3SGZmVgv/H1K5zy49GGkDAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1116b9750>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.Series(norm_hdp() for _ in range(10000)).hist()\n",
    "_=plt.title(\"Histogram of samples from distribution drawn from Hierarchical DP\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`norm_hdp` is not equivalent to the Hierarchical Dirichlet Process; it samples from a _single distribution_ sampled from this HDP. Each time we instantiate the `norm_hdp` variable, we are getting a sampler for a unique distribution. Below we sample five times and get five different distributions."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaQAAAEKCAYAAABdWiGrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n7s1wQ3wCjDQAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1126956d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaQAAAEKCAYAAABdWiGrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10a840590>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaQAAAEKCAYAAABdWiGrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nC/8fDGLl0WUzPJkAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112939c90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaQAAAEKCAYAAABdWiGrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x10b3ac790>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaQAAAEKCAYAAABdWiGrAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/wFa3DXPGzw0ngAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1126ff750>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x112a2da50>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(5):\n",
    "    norm_hdp = HierarchicalDirichletProcessSample(base_measure, alpha1=10, alpha2=10)\n",
    "    _=pd.Series(norm_hdp() for _ in range(100)).hist()\n",
    "    _=plt.title(\"Histogram of samples from distribution drawn from Hierarchical DP\")\n",
    "    _=plt.figure()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In a later post, I will discuss how these tools are applied in the realm of Bayesian nonparametrics."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}