opt_broadcast.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f23127ba9b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "from matplotlib import pyplot as plt\n",
    "import seaborn as sns\n",
    "import numpy as np\n",
    "\n",
    "np.set_printoptions(suppress=True, precision=3)\n",
    "\n",
    "sns.set(style='ticks', palette='Set2')\n",
    "sns.despine()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "c1, c2, c3, c4 = sns.mpl_palette('Set2', 4)\n",
    "red_queen_color = c1\n",
    "poisson_color = c2\n",
    "oracle_color = c3\n",
    "karimi_color = c4\n",
    "\n",
    "red_queen_fmt = 's-'\n",
    "oracle_fmt = 'd-'\n",
    "poisson_fmt = 'o-'\n",
    "karimi_fmt = '^-'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "%run -i redqueen/utils.py\n",
    "%run -i redqueen/opt_model.py\n",
    "%run -i redqueen/opt_runs.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from IPython.lib import deepreload"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "redqueen/opt_runs.py:174: UserWarning: Consider using `run_dynamic` instead of `run`.\n",
      "  follower_wall_intensities = wall_intensities\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "24.2522515977514"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sim_opts = SimOpts.std_poisson(world_seed=42, world_rate=1000.0)\n",
    "mgr = sim_opts.create_manager_with_opt(1)\n",
    "mgr.run()\n",
    "df = mgr.state.get_dataframe()\n",
    "u_int_opt(df, sim_opts=sim_opts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAwsAAAIUCAYAAABVUbBAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nBEEQBJEQMhYIgiAIgiAIgkjI/w/1D7G8CHHqBAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f71014918d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "rank_count = rank_of_src_in_df(df, 1)\n",
    "plt.step(rank_count.index, rank_count.mean(1), where='post')\n",
    "plt.xlabel('Time')\n",
    "plt.ylabel('r(t)')\n",
    "plt.tight_layout()\n",
    "format_axes(plt.gca())\n",
    "latexify(fig_width=3.16, largeFonts=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[0.25177123510846161, 1.0]"
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[time_in_top_k(df, src_id=x, K=10, sim_opts=sim_opts) for x in sorted(df.src_id.unique())]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "poisson_loss = calc_loss_poisson(df, u_const=u_int_opt(df, sim_opts=sim_opts), src_id=1, sim_opts=sim_opts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAANwAAACFCAYAAAAw25a+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fe76ece3828>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "calc_loss_opt(df, sim_opts).plot()\n",
    "plt.tight_layout()\n",
    "format_axes(plt.gca())\n",
    "latexify(columns=1, largeFonts=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "# Synthetic experiments"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/NL/stackexchange/work/prog/broadcast_ref/broadcast/opt/optimizer.py'"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Bopt.__file__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1.53 s, sys: 2.75 s, total: 4.28 s\n",
      "Wall time: 550 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "sim_opts = SimOpts.std_poisson(world_seed=1 + 42, world_rate=10.0).update({ 'end_time': 100.0 })\n",
    "tmp_kdd = worker_kdd((1, 567.0, 10, sim_opts, None, None), Ks=[1])\n",
    "tmp_poisson = worker_poisson((1, 567.0, sim_opts, None))\n",
    "# tmp_oracle = worker_oracle((1, 567.0, 956, sim_opts, None))\n",
    "# tmp_opt = worker_opt((1, sim_opts, None))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:Setting num_segments=10 for world-rate in worker_opt.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 2.86 s, sys: 3.23 s, total: 6.09 s\n",
      "Wall time: 1.38 s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "redqueen/opt_runs.py:82: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  intensity_df.pivot_table(values='intensity', index='sink_id', columns='segment')\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "sim_opts = SimOpts.std_piecewise_const(world_seed=1 + 42, \n",
    "                                       world_rates=np.arange(10) * 1.0,\n",
    "                                       world_change_times=np.arange(10) * 10.0\n",
    "                                      ).update({ 'end_time': 100.0 })\n",
    "tmp_kdd = worker_kdd((1, 567.0, 9, sim_opts, None, None))\n",
    "tmp_poisson = worker_poisson((1, 567.0, sim_opts, None))\n",
    "# tmp_oracle = worker_oracle((1, 567.0, 956, sim_opts, None))\n",
    "tmp_opt = worker_opt((1, sim_opts, None))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "({'avg_rank': 71.168424997482049,\n",
       "  'avg_rank_1': 71.168424997482049,\n",
       "  'avg_rank_10': 203.34746746143995,\n",
       "  'avg_rank_20': 208.60555404195742,\n",
       "  'avg_rank_3': 88.882515078230256,\n",
       "  'avg_rank_5': 138.87975916174446,\n",
       "  'avg_rank_k': 1,\n",
       "  'kdd_opt_1': array([ 35.553,  33.872,  41.65 ,  58.911,  61.168,  76.685,  78.884,\n",
       "          84.785,  95.492]),\n",
       "  'kdd_opt_10': array([ 385.379,   22.145,   22.145,   22.147,   22.151,   22.296,\n",
       "           22.42 ,   22.99 ,   25.327]),\n",
       "  'kdd_opt_20': array([ 388.172,   22.353,   22.353,   22.353,   22.353,   22.353,\n",
       "           22.353,   22.353,   22.353]),\n",
       "  'kdd_opt_3': array([ 163.163,   16.859,   23.384,   39.962,   42.334,   60.521,\n",
       "           63.277,   71.067,   86.432]),\n",
       "  'kdd_opt_5': array([ 306.464,   10.498,   13.798,   24.602,   26.293,   39.132,\n",
       "           41.16 ,   46.851,   58.203]),\n",
       "  'kdd_opt_iters_1': 284,\n",
       "  'kdd_opt_iters_10': 26815,\n",
       "  'kdd_opt_iters_20': 27311,\n",
       "  'kdd_opt_iters_3': 3988,\n",
       "  'kdd_opt_iters_5': 15615,\n",
       "  'r_2': 201.78194930382151,\n",
       "  'r_2_1': 201.78194930382151,\n",
       "  'r_2_10': 1330.3693220039302,\n",
       "  'r_2_20': 1441.6335398706383,\n",
       "  'r_2_3': 245.34967949082721,\n",
       "  'r_2_5': 576.60030859138988,\n",
       "  'r_2_k': 1,\n",
       "  's': 1.0,\n",
       "  'seed': 1,\n",
       "  'sim_opts': <__main__.SimOpts at 0x7fe7755d07f0>,\n",
       "  'top_1': 62.335189719601985,\n",
       "  'top_10': 96.902015904165268,\n",
       "  'top_20': 99.748343217878954,\n",
       "  'top_3': 88.712696911894994,\n",
       "  'top_5': 91.90263489170313,\n",
       "  'type': 'kdd',\n",
       "  'world_events': 430},\n",
       " {'avg_rank': 82.235780221212281,\n",
       "  'r_2': 267.71218956210504,\n",
       "  's': 1.0,\n",
       "  'seed': 1,\n",
       "  'sim_opts': <__main__.SimOpts at 0x7fe7755d07f0>,\n",
       "  'top_1': 61.408857339715361,\n",
       "  'top_10': 99.98856251826551,\n",
       "  'top_20': 99.98856251826551,\n",
       "  'top_3': 90.149622475671947,\n",
       "  'top_5': 96.651918261482578,\n",
       "  'type': 'Poisson',\n",
       "  'world_events': 430})"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp_kdd, tmp_poisson #, tmp_oracle, tmp_opt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "redqueen/opt_runs.py:174: UserWarning: Consider using `run_dynamic` instead of `run`.\n",
      "  follower_wall_intensities = wall_intensities\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 453 ms, sys: 6.67 ms, total: 460 ms\n",
      "Wall time: 456 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "sim_opts= SimOpts.std_hawkes(world_seed=1 + 42, world_lambda_0=1000.0, world_alpha=1.0, world_beta=10.0)\n",
    "mgr = sim_opts.create_manager_for_wall()\n",
    "mgr.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 0 ns, sys: 0 ns, total: 0 ns\n",
      "Wall time: 21.5 µs\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# sim_opts = SimOpts.std_hawkes(world_seed=1 + 42, world_lambda_0=1000.0, world_alpha=1.0, world_beta=10.0)\n",
    "# worker_kdd((1, 3, 100.0, 10, sim_opts, None, None))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'seed' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m/NL/stackexchange/work/prog/broadcast-online/opt_model.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mget_ipython\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mrun_cell_magic\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'time'\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m''\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m\"import itertools as I\\nsim_opts = SimOpts(other_sources=[Poisson2(src_id=2, seed=seed, rate=1000.0)], \\n                   end_time=100.0, \\n                   src_id=1,\\n                   q_vec=q_vec,\\n                   s=s,\\n                   sink_ids=[1001],\\n                   edge_list=[(1, 1001), (2, 1001)])\\n\\nnum_cpus = mp.cpu_count()\\nwith mp.Pool(num_cpus - 2) as pool:\\n    N = 10\\n    K = 10\\n    seeds = range(0, N)\\n    seed_capacity = [None] * N\\n    opt_top_k_perf = [None] * N\\n    for idx, opt_dict in enumerate(pool.map(worker_top_k_opt,\\n                                            [(s, K, SimOpts.std_poisson(world_seed=s, world_rate=1000.0), None)\\n                                            for s in seeds])):\\n        seed_capacity[idx] = opt_dict['capacity']\\n        opt_top_k_perf[idx] = opt_dict['perf']\\n    \\n    poisson_top_k_perf = [None] * N\\n    for idx, poisson_dict in enumerate(pool.map(worker_top_k_poisson,\\n                                                [(s, K, seed_capacity[idx], \\n                                                  SimOpts.std_poisson(world_seed=s, world_rate=1000.0), None)\\n                                                 for idx, s in enumerate(seeds)])):\\n        poisson_top_k_perf[idx] = poisson_dict['perf']\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;32m/home/utkarshu/miniconda3/envs/py35/lib/python3.5/site-packages/IPython/core/interactiveshell.py\u001b[0m in \u001b[0;36mrun_cell_magic\u001b[1;34m(self, magic_name, line, cell)\u001b[0m\n\u001b[0;32m   2118\u001b[0m             \u001b[0mmagic_arg_s\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mvar_expand\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mline\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mstack_depth\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2119\u001b[0m             \u001b[1;32mwith\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mbuiltin_trap\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2120\u001b[1;33m                 \u001b[0mresult\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mfn\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mmagic_arg_s\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mcell\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   2121\u001b[0m             \u001b[1;32mreturn\u001b[0m \u001b[0mresult\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2122\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<decorator-gen-60>\u001b[0m in \u001b[0;36mtime\u001b[1;34m(self, line, cell, local_ns)\u001b[0m\n",
      "\u001b[1;32m/home/utkarshu/miniconda3/envs/py35/lib/python3.5/site-packages/IPython/core/magic.py\u001b[0m in \u001b[0;36m<lambda>\u001b[1;34m(f, *a, **k)\u001b[0m\n\u001b[0;32m    191\u001b[0m     \u001b[1;31m# but it's overkill for just that one bit of state.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    192\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mmagic_deco\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0marg\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 193\u001b[1;33m         \u001b[0mcall\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mlambda\u001b[0m \u001b[0mf\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0ma\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mk\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mf\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0ma\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mk\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    194\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    195\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mcallable\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0marg\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m/home/utkarshu/miniconda3/envs/py35/lib/python3.5/site-packages/IPython/core/magics/execution.py\u001b[0m in \u001b[0;36mtime\u001b[1;34m(self, line, cell, local_ns)\u001b[0m\n\u001b[0;32m   1175\u001b[0m         \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1176\u001b[0m             \u001b[0mst\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mclock2\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1177\u001b[1;33m             \u001b[0mexec\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mcode\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mglob\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlocal_ns\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1178\u001b[0m             \u001b[0mend\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mclock2\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1179\u001b[0m             \u001b[0mout\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m/NL/stackexchange/work/prog/broadcast-online/opt_model.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n",
      "\u001b[1;31mNameError\u001b[0m: name 'seed' is not defined"
     ]
    }
   ],
   "source": [
    "# %%time\n",
    "# import itertools as I\n",
    "# sim_opts = SimOpts(other_sources=[Poisson2(src_id=2, seed=seed, rate=1000.0)], \n",
    "#                    end_time=100.0, \n",
    "#                    src_id=1,\n",
    "#                    q_vec=q_vec,\n",
    "#                    s=s,\n",
    "#                    sink_ids=[1001],\n",
    "#                    edge_list=[(1, 1001), (2, 1001)])\n",
    "\n",
    "# with mp.Pool() as pool:\n",
    "#     N = 10\n",
    "#     K = 10\n",
    "#     seeds = range(0, N)\n",
    "#     seed_capacity = [None] * N\n",
    "#     opt_top_k_perf = [None] * N\n",
    "#     for idx, opt_dict in enumerate(pool.map(worker_top_k_opt,\n",
    "#                                             [(s, K, SimOpts.std_poisson(world_seed=s, world_rate=1000.0), None)\n",
    "#                                             for s in seeds])):\n",
    "#         seed_capacity[idx] = opt_dict['capacity']\n",
    "#         opt_top_k_perf[idx] = opt_dict['perf']\n",
    "    \n",
    "#     poisson_top_k_perf = [None] * N\n",
    "#     for idx, poisson_dict in enumerate(pool.map(worker_top_k_poisson,\n",
    "#                                                 [(s, K, seed_capacity[idx], \n",
    "#                                                   SimOpts.std_poisson(world_seed=s, world_rate=1000.0), None)\n",
    "#                                                  for idx, s in enumerate(seeds)])):\n",
    "#         poisson_top_k_perf[idx] = poisson_dict['perf']"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Inference new"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 469,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import warnings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pickle\n",
    "import seqfile"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2018-03-31 02:47:47.509606\u001b[0m Checkpoint: active_procs = 220, procs = [('Opt', 38), ('Poisson', 60), ('Oracle', 61), ('kdd', 61)]\n",
      "*** \u001b[31m2018-03-31 02:47:47.557850\u001b[0m Checkpoint: active_procs = 230, procs = [('Opt', 32), ('Poisson', 64), ('Oracle', 67), ('kdd', 67)]\n",
      "*** \u001b[31m2018-03-31 02:47:47.635034\u001b[0m Checkpoint: active_procs = 240, procs = [('Opt', 27), ('Poisson', 69), ('Oracle', 72), ('kdd', 72)]\n",
      "*** \u001b[31m2018-03-31 02:47:47.990243\u001b[0m Checkpoint: active_procs = 250, procs = [('Opt', 21), ('Poisson', 73), ('Oracle', 78), ('kdd', 78)]\n",
      "*** \u001b[31m2018-03-31 02:47:48.365106\u001b[0m Checkpoint: active_procs = 260, procs = [('Opt', 15), ('Poisson', 77), ('Oracle', 84), ('kdd', 84)]\n",
      "*** \u001b[31m2018-03-31 02:47:48.387391\u001b[0m Checkpoint: active_procs = 260, procs = [('Opt', 14), ('Poisson', 76), ('Oracle', 85), ('kdd', 85)]\n",
      "*** \u001b[31m2018-03-31 02:47:48.399989\u001b[0m Checkpoint: active_procs = 270, procs = [('Opt', 10), ('Poisson', 80), ('Oracle', 90), ('kdd', 90)]\n",
      "*** \u001b[31m2018-03-31 02:47:48.745565\u001b[0m Checkpoint: active_procs = 270, procs = [('Opt', 0), ('Poisson', 75), ('Oracle', 100), ('kdd', 95)]\n",
      "*** \u001b[31m2018-03-31 02:47:49.527218\u001b[0m Checkpoint: active_procs = 260, procs = [('Opt', 0), ('Poisson', 72), ('Oracle', 100), ('kdd', 88)]\n",
      "*** \u001b[31m2018-03-31 02:47:50.093013\u001b[0m Checkpoint: active_procs = 250, procs = [('Opt', 0), ('Poisson', 68), ('Oracle', 100), ('kdd', 82)]\n",
      "*** \u001b[31m2018-03-31 02:47:50.698939\u001b[0m Checkpoint: active_procs = 240, procs = [('Opt', 0), ('Poisson', 65), ('Oracle', 100), ('kdd', 75)]\n",
      "*** \u001b[31m2018-03-31 02:47:51.219166\u001b[0m Checkpoint: active_procs = 230, procs = [('Opt', 0), ('Poisson', 61), ('Oracle', 100), ('kdd', 69)]\n",
      "*** \u001b[31m2018-03-31 02:47:51.847267\u001b[0m Checkpoint: active_procs = 220, procs = [('Opt', 0), ('Poisson', 58), ('Oracle', 100), ('kdd', 62)]\n",
      "*** \u001b[31m2018-03-31 02:47:52.826303\u001b[0m Checkpoint: active_procs = 210, procs = [('Opt', 0), ('Poisson', 54), ('Oracle', 100), ('kdd', 56)]\n",
      "*** \u001b[31m2018-03-31 02:48:23.233458\u001b[0m Checkpoint: active_procs = 200, procs = [('Opt', 0), ('Poisson', 51), ('Oracle', 97), ('kdd', 52)]\n",
      "*** \u001b[31m2018-03-31 02:48:28.850320\u001b[0m Checkpoint: active_procs = 190, procs = [('Opt', 0), ('Poisson', 48), ('Oracle', 94), ('kdd', 48)]\n",
      "*** \u001b[31m2018-03-31 02:48:32.897618\u001b[0m Checkpoint: active_procs = 180, procs = [('Opt', 0), ('Poisson', 44), ('Oracle', 90), ('kdd', 46)]\n",
      "*** \u001b[31m2018-03-31 02:48:43.728825\u001b[0m Checkpoint: active_procs = 170, procs = [('Opt', 0), ('Poisson', 41), ('Oracle', 87), ('kdd', 42)]\n",
      "*** \u001b[31m2018-03-31 02:48:47.694794\u001b[0m Checkpoint: active_procs = 160, procs = [('Opt', 0), ('Poisson', 38), ('Oracle', 83), ('kdd', 39)]\n",
      "*** \u001b[31m2018-03-31 02:48:48.716827\u001b[0m Checkpoint: active_procs = 150, procs = [('Opt', 0), ('Poisson', 34), ('Oracle', 81), ('kdd', 35)]\n",
      "*** \u001b[31m2018-03-31 02:48:53.343555\u001b[0m Checkpoint: active_procs = 140, procs = [('Opt', 0), ('Poisson', 31), ('Oracle', 77), ('kdd', 32)]\n",
      "*** \u001b[31m2018-03-31 02:48:54.997011\u001b[0m Checkpoint: active_procs = 130, procs = [('Opt', 0), ('Poisson', 28), ('Oracle', 74), ('kdd', 28)]\n",
      "*** \u001b[31m2018-03-31 02:48:56.854348\u001b[0m Checkpoint: active_procs = 120, procs = [('Opt', 0), ('Poisson', 24), ('Oracle', 71), ('kdd', 25)]\n",
      "*** \u001b[31m2018-03-31 02:48:58.663467\u001b[0m Checkpoint: active_procs = 110, procs = [('Opt', 0), ('Poisson', 21), ('Oracle', 67), ('kdd', 22)]\n",
      "*** \u001b[31m2018-03-31 02:49:01.793600\u001b[0m Checkpoint: active_procs = 100, procs = [('Opt', 0), ('Poisson', 18), ('Oracle', 64), ('kdd', 18)]\n",
      "*** \u001b[31m2018-03-31 02:49:10.017263\u001b[0m Checkpoint: active_procs = 90, procs = [('Opt', 0), ('Poisson', 14), ('Oracle', 61), ('kdd', 15)]\n",
      "*** \u001b[31m2018-03-31 02:49:13.994123\u001b[0m Checkpoint: active_procs = 80, procs = [('Opt', 0), ('Poisson', 11), ('Oracle', 57), ('kdd', 12)]\n",
      "*** \u001b[31m2018-03-31 02:49:19.550758\u001b[0m Checkpoint: active_procs = 70, procs = [('Opt', 0), ('Poisson', 8), ('Oracle', 54), ('kdd', 8)]\n",
      "*** \u001b[31m2018-03-31 02:49:33.321078\u001b[0m Checkpoint: active_procs = 60, procs = [('Opt', 0), ('Poisson', 4), ('Oracle', 51), ('kdd', 5)]\n",
      "*** \u001b[31m2018-03-31 02:49:59.017938\u001b[0m Checkpoint: active_procs = 50, procs = [('Opt', 0), ('Poisson', 1), ('Oracle', 47), ('kdd', 2)]\n",
      "*** \u001b[31m2018-03-31 02:50:24.198386\u001b[0m Checkpoint: active_procs = 40, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 40), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 02:51:00.152796\u001b[0m Checkpoint: active_procs = 30, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 30), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 02:51:23.567895\u001b[0m Checkpoint: active_procs = 20, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 20), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 02:51:45.468666\u001b[0m Checkpoint: active_procs = 10, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 10), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 02:52:30.300343\u001b[0m Checkpoint: active_procs = 0, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 0), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 02:52:32.939247\u001b[0m Checkpoint: active_procs = 290, procs = [('Opt', 1), ('Poisson', 92), ('Oracle', 99), ('kdd', 98)]\n",
      "*** \u001b[31m2018-03-31 02:52:33.019925\u001b[0m Checkpoint: active_procs = 290, procs = [('Opt', 0), ('Poisson', 92), ('Oracle', 100), ('kdd', 98)]\n",
      "*** \u001b[31m2018-03-31 02:52:33.736491\u001b[0m Checkpoint: active_procs = 280, procs = [('Opt', 0), ('Poisson', 86), ('Oracle', 100), ('kdd', 94)]\n",
      "*** \u001b[31m2018-03-31 02:52:34.978826\u001b[0m Checkpoint: active_procs = 270, procs = [('Opt', 0), ('Poisson', 80), ('Oracle', 100), ('kdd', 90)]\n",
      "*** \u001b[31m2018-03-31 02:52:36.284419\u001b[0m Checkpoint: active_procs = 260, procs = [('Opt', 0), ('Poisson', 75), ('Oracle', 100), ('kdd', 85)]\n",
      "*** \u001b[31m2018-03-31 02:52:36.951822\u001b[0m Checkpoint: active_procs = 250, procs = [('Opt', 0), ('Poisson', 72), ('Oracle', 100), ('kdd', 78)]\n",
      "*** \u001b[31m2018-03-31 02:52:38.099473\u001b[0m Checkpoint: active_procs = 240, procs = [('Opt', 0), ('Poisson', 67), ('Oracle', 100), ('kdd', 73)]\n",
      "*** \u001b[31m2018-03-31 02:52:39.350725\u001b[0m Checkpoint: active_procs = 230, procs = [('Opt', 0), ('Poisson', 64), ('Oracle', 100), ('kdd', 66)]\n",
      "*** \u001b[31m2018-03-31 02:52:40.922491\u001b[0m Checkpoint: active_procs = 220, procs = [('Opt', 0), ('Poisson', 59), ('Oracle', 100), ('kdd', 61)]\n",
      "*** \u001b[31m2018-03-31 02:52:45.420363\u001b[0m Checkpoint: active_procs = 210, procs = [('Opt', 0), ('Poisson', 55), ('Oracle', 100), ('kdd', 55)]\n",
      "*** \u001b[31m2018-03-31 02:53:30.169983\u001b[0m Checkpoint: active_procs = 200, procs = [('Opt', 0), ('Poisson', 51), ('Oracle', 97), ('kdd', 52)]\n",
      "*** \u001b[31m2018-03-31 02:53:33.413424\u001b[0m Checkpoint: active_procs = 190, procs = [('Opt', 0), ('Poisson', 49), ('Oracle', 92), ('kdd', 49)]\n",
      "*** \u001b[31m2018-03-31 02:53:36.284590\u001b[0m Checkpoint: active_procs = 180, procs = [('Opt', 0), ('Poisson', 44), ('Oracle', 90), ('kdd', 46)]\n",
      "*** \u001b[31m2018-03-31 02:53:45.473424\u001b[0m Checkpoint: active_procs = 170, procs = [('Opt', 0), ('Poisson', 41), ('Oracle', 87), ('kdd', 42)]\n",
      "*** \u001b[31m2018-03-31 02:53:52.199621\u001b[0m Checkpoint: active_procs = 160, procs = [('Opt', 0), ('Poisson', 38), ('Oracle', 83), ('kdd', 39)]\n",
      "*** \u001b[31m2018-03-31 02:53:57.491701\u001b[0m Checkpoint: active_procs = 150, procs = [('Opt', 0), ('Poisson', 35), ('Oracle', 79), ('kdd', 36)]\n",
      "*** \u001b[31m2018-03-31 02:54:02.323020\u001b[0m Checkpoint: active_procs = 140, procs = [('Opt', 0), ('Poisson', 31), ('Oracle', 77), ('kdd', 32)]\n",
      "*** \u001b[31m2018-03-31 02:54:11.555525\u001b[0m Checkpoint: active_procs = 130, procs = [('Opt', 0), ('Poisson', 28), ('Oracle', 74), ('kdd', 28)]\n",
      "*** \u001b[31m2018-03-31 02:54:19.698646\u001b[0m Checkpoint: active_procs = 120, procs = [('Opt', 0), ('Poisson', 25), ('Oracle', 69), ('kdd', 26)]\n",
      "*** \u001b[31m2018-03-31 02:54:23.439875\u001b[0m Checkpoint: active_procs = 110, procs = [('Opt', 0), ('Poisson', 22), ('Oracle', 65), ('kdd', 23)]\n",
      "*** \u001b[31m2018-03-31 02:54:24.856293\u001b[0m Checkpoint: active_procs = 100, procs = [('Opt', 0), ('Poisson', 19), ('Oracle', 61), ('kdd', 20)]\n",
      "*** \u001b[31m2018-03-31 02:54:28.243318\u001b[0m Checkpoint: active_procs = 90, procs = [('Opt', 0), ('Poisson', 14), ('Oracle', 60), ('kdd', 16)]\n",
      "*** \u001b[31m2018-03-31 02:54:35.863674\u001b[0m Checkpoint: active_procs = 80, procs = [('Opt', 0), ('Poisson', 11), ('Oracle', 57), ('kdd', 12)]\n",
      "*** \u001b[31m2018-03-31 02:54:52.474474\u001b[0m Checkpoint: active_procs = 70, procs = [('Opt', 0), ('Poisson', 8), ('Oracle', 54), ('kdd', 8)]\n",
      "*** \u001b[31m2018-03-31 02:55:20.394444\u001b[0m Checkpoint: active_procs = 60, procs = [('Opt', 0), ('Poisson', 4), ('Oracle', 51), ('kdd', 5)]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2018-03-31 02:55:56.185044\u001b[0m Checkpoint: active_procs = 50, procs = [('Opt', 0), ('Poisson', 1), ('Oracle', 47), ('kdd', 2)]\n",
      "*** \u001b[31m2018-03-31 02:56:37.604176\u001b[0m Checkpoint: active_procs = 40, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 40), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 02:57:00.032132\u001b[0m Checkpoint: active_procs = 30, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 30), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 02:57:45.678050\u001b[0m Checkpoint: active_procs = 20, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 20), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 02:58:21.721279\u001b[0m Checkpoint: active_procs = 10, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 10), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 02:59:52.747566\u001b[0m Checkpoint: active_procs = 0, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 0), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 02:59:53.776968\u001b[0m Checkpoint: active_procs = 210, procs = [('Opt', 44), ('Poisson', 54), ('Oracle', 56), ('kdd', 56)]\n",
      "*** \u001b[31m2018-03-31 02:59:53.785880\u001b[0m Checkpoint: active_procs = 210, procs = [('Opt', 42), ('Poisson', 54), ('Oracle', 57), ('kdd', 57)]\n",
      "*** \u001b[31m2018-03-31 02:59:53.856865\u001b[0m Checkpoint: active_procs = 220, procs = [('Opt', 36), ('Poisson', 58), ('Oracle', 63), ('kdd', 63)]\n",
      "*** \u001b[31m2018-03-31 02:59:53.871729\u001b[0m Checkpoint: active_procs = 230, procs = [('Opt', 31), ('Poisson', 63), ('Oracle', 68), ('kdd', 68)]\n",
      "*** \u001b[31m2018-03-31 02:59:53.920327\u001b[0m Checkpoint: active_procs = 260, procs = [('Opt', 15), ('Poisson', 77), ('Oracle', 84), ('kdd', 84)]\n",
      "*** \u001b[31m2018-03-31 02:59:53.951354\u001b[0m Checkpoint: active_procs = 270, procs = [('Opt', 9), ('Poisson', 81), ('Oracle', 90), ('kdd', 90)]\n",
      "*** \u001b[31m2018-03-31 02:59:53.971654\u001b[0m Checkpoint: active_procs = 270, procs = [('Opt', 8), ('Poisson', 80), ('Oracle', 91), ('kdd', 91)]\n",
      "*** \u001b[31m2018-03-31 02:59:55.044083\u001b[0m Checkpoint: active_procs = 270, procs = [('Opt', 0), ('Poisson', 74), ('Oracle', 100), ('kdd', 96)]\n",
      "*** \u001b[31m2018-03-31 02:59:56.145687\u001b[0m Checkpoint: active_procs = 260, procs = [('Opt', 0), ('Poisson', 72), ('Oracle', 100), ('kdd', 88)]\n",
      "*** \u001b[31m2018-03-31 02:59:56.752191\u001b[0m Checkpoint: active_procs = 250, procs = [('Opt', 0), ('Poisson', 68), ('Oracle', 100), ('kdd', 82)]\n",
      "*** \u001b[31m2018-03-31 02:59:57.159526\u001b[0m Checkpoint: active_procs = 240, procs = [('Opt', 0), ('Poisson', 67), ('Oracle', 100), ('kdd', 73)]\n",
      "*** \u001b[31m2018-03-31 02:59:57.414505\u001b[0m Checkpoint: active_procs = 230, procs = [('Opt', 0), ('Poisson', 64), ('Oracle', 100), ('kdd', 66)]\n",
      "*** \u001b[31m2018-03-31 02:59:57.855989\u001b[0m Checkpoint: active_procs = 220, procs = [('Opt', 0), ('Poisson', 60), ('Oracle', 100), ('kdd', 60)]\n",
      "*** \u001b[31m2018-03-31 02:59:58.468883\u001b[0m Checkpoint: active_procs = 210, procs = [('Opt', 0), ('Poisson', 55), ('Oracle', 100), ('kdd', 55)]\n",
      "*** \u001b[31m2018-03-31 03:00:22.868494\u001b[0m Checkpoint: active_procs = 200, procs = [('Opt', 0), ('Poisson', 51), ('Oracle', 97), ('kdd', 52)]\n",
      "*** \u001b[31m2018-03-31 03:00:26.961735\u001b[0m Checkpoint: active_procs = 190, procs = [('Opt', 0), ('Poisson', 48), ('Oracle', 93), ('kdd', 49)]\n",
      "*** \u001b[31m2018-03-31 03:00:29.098909\u001b[0m Checkpoint: active_procs = 180, procs = [('Opt', 0), ('Poisson', 44), ('Oracle', 91), ('kdd', 45)]\n",
      "*** \u001b[31m2018-03-31 03:00:31.417988\u001b[0m Checkpoint: active_procs = 170, procs = [('Opt', 0), ('Poisson', 41), ('Oracle', 87), ('kdd', 42)]\n",
      "*** \u001b[31m2018-03-31 03:00:32.947136\u001b[0m Checkpoint: active_procs = 160, procs = [('Opt', 0), ('Poisson', 38), ('Oracle', 83), ('kdd', 39)]\n",
      "*** \u001b[31m2018-03-31 03:00:34.116940\u001b[0m Checkpoint: active_procs = 150, procs = [('Opt', 0), ('Poisson', 35), ('Oracle', 79), ('kdd', 36)]\n",
      "*** \u001b[31m2018-03-31 03:00:35.214608\u001b[0m Checkpoint: active_procs = 140, procs = [('Opt', 0), ('Poisson', 31), ('Oracle', 77), ('kdd', 32)]\n",
      "*** \u001b[31m2018-03-31 03:00:35.911252\u001b[0m Checkpoint: active_procs = 130, procs = [('Opt', 0), ('Poisson', 28), ('Oracle', 74), ('kdd', 28)]\n",
      "*** \u001b[31m2018-03-31 03:00:38.845865\u001b[0m Checkpoint: active_procs = 120, procs = [('Opt', 0), ('Poisson', 24), ('Oracle', 71), ('kdd', 25)]\n",
      "*** \u001b[31m2018-03-31 03:00:40.852844\u001b[0m Checkpoint: active_procs = 110, procs = [('Opt', 0), ('Poisson', 21), ('Oracle', 67), ('kdd', 22)]\n",
      "*** \u001b[31m2018-03-31 03:00:45.959091\u001b[0m Checkpoint: active_procs = 100, procs = [('Opt', 0), ('Poisson', 18), ('Oracle', 64), ('kdd', 18)]\n",
      "*** \u001b[31m2018-03-31 03:00:47.864503\u001b[0m Checkpoint: active_procs = 90, procs = [('Opt', 0), ('Poisson', 14), ('Oracle', 60), ('kdd', 16)]\n",
      "*** \u001b[31m2018-03-31 03:00:52.198696\u001b[0m Checkpoint: active_procs = 80, procs = [('Opt', 0), ('Poisson', 11), ('Oracle', 57), ('kdd', 12)]\n",
      "*** \u001b[31m2018-03-31 03:00:57.698616\u001b[0m Checkpoint: active_procs = 70, procs = [('Opt', 0), ('Poisson', 8), ('Oracle', 54), ('kdd', 8)]\n",
      "*** \u001b[31m2018-03-31 03:01:05.634158\u001b[0m Checkpoint: active_procs = 60, procs = [('Opt', 0), ('Poisson', 4), ('Oracle', 51), ('kdd', 5)]\n",
      "*** \u001b[31m2018-03-31 03:01:10.181944\u001b[0m Checkpoint: active_procs = 50, procs = [('Opt', 0), ('Poisson', 1), ('Oracle', 47), ('kdd', 2)]\n",
      "*** \u001b[31m2018-03-31 03:01:26.472231\u001b[0m Checkpoint: active_procs = 40, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 40), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 03:01:40.467139\u001b[0m Checkpoint: active_procs = 30, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 30), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 03:01:54.836387\u001b[0m Checkpoint: active_procs = 20, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 20), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 03:02:07.012051\u001b[0m Checkpoint: active_procs = 10, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 10), ('kdd', 0)]\n",
      "*** \u001b[31m2018-03-31 03:02:36.822016\u001b[0m Checkpoint: active_procs = 0, procs = [('Opt', 0), ('Poisson', 0), ('Oracle', 0), ('kdd', 0)]\n",
      "CPU times: user 690 ms, sys: 1.03 s, total: 1.72 s\n",
      "Wall time: 14min 50s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "with warnings.catch_warnings():\n",
    "    warnings.simplefilter('ignore')\n",
    "    poisson_res_2 = run_inference_queue(opts=poisson_inf_opts)\n",
    "    hawkes_res_2 = run_inference_queue(opts=hawkes_inf_opts)\n",
    "    piecewise_res_2 = run_inference_queue(opts=piecewise_inf_opts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def strip_sim_opts_from_raw_results(d):\n",
    "    def rm_sim_opts(d_in):\n",
    "        return {k: v for k, v in d_in.items() if k != 'sim_opts'}\n",
    "    \n",
    "    return {k: (v if k != 'raw_results' else [rm_sim_opts(x) for x in v]) \n",
    "            for k, v in poisson_res._get_dict().items()}\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "tmp = strip_sim_opts_from_raw_results(poisson_res._get_dict())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['capacities', 'raw_results', 'df'])"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['world_events', 'capacity', 'seed', 'r_2', 'top_3', 'avg_rank', 'type', 's', 'num_events', 'top_5', 'top_1', 'top_20', 'top_10'])"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp['raw_results'][1].keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Wrote ./synthetic-6.pickle\n"
     ]
    }
   ],
   "source": [
    "with open(seqfile.findNextFile(prefix='synthetic-', suffix='.pickle'), 'wb') as pickle_file:\n",
    "    pickle.dump({'poisson': strip_sim_opts_from_raw_results(poisson_res._get_dict()), \n",
    "                 'hawkes': strip_sim_opts_from_raw_results(hawkes_res._get_dict()), \n",
    "                 'piecewise': strip_sim_opts_from_raw_results(piecewise_res._get_dict())}, pickle_file)\n",
    "    print('Wrote {}'.format(pickle_file.name))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 460,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "defined = False\n",
    "try:\n",
    "    poisson_res, hawkes_res, piecewise_res\n",
    "    defined = True\n",
    "except NameError:\n",
    "    pass\n",
    "\n",
    "confirmation = 'y'\n",
    "\n",
    "if defined:\n",
    "     confirmation = input('Some variables are already defined. Are you sure you want to clobber them? [Y/n]')\n",
    "    \n",
    "if confirmation.lower().startswith('y'):\n",
    "    with open('synthetic-6.pickle', 'rb') as pickle_file:\n",
    "        reloaded = pickle.load(pickle_file)\n",
    "\n",
    "    poisson_res = Options(**reloaded['poisson'])\n",
    "    hawkes_res = Options(**reloaded['hawkes'])\n",
    "    piecewise_res = Options(**reloaded['piecewise'])\n",
    "else:\n",
    "    print('Cloberring averted.')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# tmp = piecewise_res.df.pivot_table(index='s', columns='type', values='top_1', aggfunc=np.mean)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# tmp['capacities'] = [np.mean([y for x, y in piecewise_res.capacities[s]]) for s in tmp.index]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# tmp['Opt/Oracle'] = tmp['Opt'] / tmp['Oracle']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# poisson_res.raw_results[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# piecewise_res.capacities[piecewise_res.df.s[0]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# tmp = [x for x in raw_results if x['type'] == 'kdd']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# sim_opts = SimOpts.std_poisson(world_seed=tmp[idx]['seed'], world_rate=world_rate)\n",
    "# mgr = sim_opts.create_manager_for_piecewise_const(tmp[idx]['seed'], np.arange(num_segments) * (T / num_segments), \n",
    "#                                                   tmp[idx]['kdd_opt'] / (T / num_segments))\n",
    "# mgr.run()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# df = mgr.state.get_dataframe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(933, 5)"
      ]
     },
     "execution_count": 127,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# df[df.src_id == 1].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# world_changing_rates = np.array([ 6872.701,  9753.572,  8659.97 ,  7993.292,  5780.093,  5779.973, 5290.418,  9330.881,  8005.575,  8540.363])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# %%time\n",
    "# sim_opts = SimOpts.std_piecewise_const(world_seed=1, \n",
    "#                                        world_change_times=world_change_times,\n",
    "#                                        world_rates=world_changing_rates  / 10.0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2018-03-31 11:06:59.177817\u001b[0m Checkpoint: q_lo = 1.0, q_hi = 2.0, num_events = 62 \n",
      "*** \u001b[31m2018-03-31 11:07:06.825671\u001b[0m Checkpoint: q_lo = 2.0, q_hi = 4.0, num_events = 50 \n"
     ]
    }
   ],
   "source": [
    "# opt_oracle=find_opt_oracle(50.0, sim_opts, verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.23083736386486994"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# oracle_df = opt_oracle['df']\n",
    "# rb_mgr = sim_opts.create_manager_with_times(oracle_df.t[oracle_df.events == 1] + 1e-10)\n",
    "# rb_mgr.run_dynamic()\n",
    "# df = rb_mgr.state.get_dataframe()\n",
    "# time_in_top_k(rb_mgr.state.get_dataframe(), K=3, sim_opts=sim_opts) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# oracle_rank_calc = rank_of_src_in_df(df=df, src_id=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 166,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "50"
      ]
     },
     "execution_count": 166,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# (df.src_id == 1).sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 9.01 s, sys: 0 ns, total: 9.01 s\n",
      "Wall time: 9 s\n"
     ]
    }
   ],
   "source": [
    "# %%time\n",
    "# q_opt = find_opt_oracle_q(100, sim_opts, verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# q_opt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 8.73 s, sys: 0 ns, total: 8.73 s\n",
      "Wall time: 8.72 s\n"
     ]
    }
   ],
   "source": [
    "# %%time\n",
    "# oracle_df = get_oracle_df(sim_opts.update({'q': q_opt}))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "# sim_opts = SimOpts.std_poisson(world_seed=42, world_rate=1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 8.09 s, sys: 0 ns, total: 8.09 s\n",
      "Wall time: 8.08 s\n"
     ]
    }
   ],
   "source": [
    "# %%time\n",
    "# # sim_opts = sim_opt_gen(0)\n",
    "# wall_mgr = sim_opts.create_manager_for_wall()\n",
    "# wall_mgr.run_dynamic()\n",
    "# oracle_df = oracle_ranking(df=wall_mgr.state.get_dataframe(), \n",
    "#                            sim_opts=sim_opts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "@optioned(option_arg='opts')\n",
    "def rank_plot(res, field, file_name, title=None, with_std=True, save=False, log_y=True, legend=False, fig_width=None, end_time=None):\n",
    "    assert end_time is not None, \"end_time must be specified.\"\n",
    "    vals = res.df.pivot_table(index='q', columns='type', values=field, aggfunc=np.mean) / end_time\n",
    "    std  = res.df.pivot_table(index='q', columns='type', values=field, aggfunc=np.std) / end_time\n",
    "    \n",
    "    if title is None:\n",
    "        if field == 'avg_rank':\n",
    "            title = 'Avg. rank'\n",
    "        elif field == 'r_2':\n",
    "            title = r'$\\int r^2$dt'\n",
    "        else:\n",
    "            title = 'Unknown axis'\n",
    "            \n",
    "    if fig_width is None:\n",
    "        fig_width = 1.75\n",
    "        \n",
    "    latexify(fig_width=fig_width, largeFonts=False)\n",
    "    c1, c2, c3, c4 = sns.mpl_palette('Set2', 4)\n",
    "\n",
    "    X = [np.mean(res.capacities[s]) for s in vals.index]\n",
    "    plt.errorbar(X, vals.Opt, fmt='s-', yerr=std.Opt.values, label='RedQueen',  color=red_queen_color, markersize=3, linewidth=0.75)\n",
    "    # plt.errorbar(X, vals.Poisson, fmt='o-', yerr=std.Poisson.values, label='Poisson', color=poisson_color, markersize=3, linewidth=0.75)\n",
    "    plt.errorbar(X, vals.Oracle, fmt='d-', yerr=std.Oracle.values, label='Oracle', color=oracle_color, markersize=3, linewidth=0.75)\n",
    "    plt.errorbar(X, vals.kdd, fmt='^-', yerr=std.kdd.values, label='Karimi', color=karimi_color, markersize=3, linewidth=0.75)\n",
    "    \n",
    "    plt.xscale('log')\n",
    "    \n",
    "    if log_y:\n",
    "        plt.yscale('log')\n",
    "        plt.yticks([10 ** i for i in [-1, 0, 1, 2]])\n",
    "    \n",
    "    if legend:\n",
    "        plt.legend(loc=\"upper right\")\n",
    "        \n",
    "    # plt.ylabel(title)\n",
    "    plt.xlabel(r'Budget')\n",
    "    # plt.ylim(None, 105)\n",
    "    plt.xlim(4.5, None)\n",
    "    plt.tight_layout()\n",
    "    format_axes(plt.gca())\n",
    "    \n",
    "    if save:\n",
    "        plt.savefig(file_name, bbox_inches='tight', pad_inches=0)\n",
    "        \n",
    "    # plt.savefig('/NL/stackexchange/work/prog/papers/online-twitter-broadcast/nips2016/figures/capacity-poisson-oracle-opt.pdf')\n",
    "    # plt.savefig('/NL/stackexchange/work/prog/papers/online-twitter-broadcast/nips2016/figures/capacity-poisson-oracle-opt-kdd.pdf')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAH0AAABNCAYAAACL+muIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nmLMAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2399c07f28>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "rank_plot(hawkes_res_2, 'avg_rank', None, log_y=True, end_time=simulation_opts.T)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Legend"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### With Oracle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f238320b080>"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAH0AAABSCAYAAAB5etvGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAz2BgYEgVA6lAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2383204278>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2383204a58>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure()\n",
    "figlegend = plt.figure(figsize=(4 * 2.0, 0.25))\n",
    "ax = fig.add_subplot(111)\n",
    "lines = [\n",
    "    ax.plot(range(10), np.random.randn(10), red_queen_fmt, color=red_queen_color, markersize=3, linewidth=0.75)[0], \n",
    "    ax.plot(range(10), np.random.randn(10), oracle_fmt, color=oracle_color, markersize=3, linewidth=0.75)[0],\n",
    "    ax.plot(range(10), np.random.randn(10), karimi_fmt, color=karimi_color, markersize=3, linewidth=0.75)[0]\n",
    "]\n",
    "figlegend.legend(tuple(lines), (r'\\textsc{RedQueen}', 'Oracle', 'Karimi'), 'center', ncol=3)\n",
    "# figlegend.savefig('/NL/stackexchange/work/prog/papers/online-twitter-broadcast/wsdm2017/FIG/legend-with-oracle.pdf', \n",
    "#                   bbox_inches='tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### With Poisson"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7f238319cf98>"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAH0AAABSCAYAAAB5etvGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2399d926a0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2383270358>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "fig = plt.figure()\n",
    "figlegend = plt.figure(figsize=(4 * 2.0, 0.25))\n",
    "ax = fig.add_subplot(111)\n",
    "lines = [\n",
    "    ax.plot(range(10), np.random.randn(10), red_queen_fmt, color=red_queen_color, markersize=3, linewidth=0.75)[0], \n",
    "    # ax.plot(range(10), np.random.randn(10), poisson_fmt, color=poisson_color, markersize=3, linewidth=0.75)[0],\n",
    "    ax.plot(range(10), np.random.randn(10), karimi_fmt, color=karimi_color, markersize=3, linewidth=0.75)[0]\n",
    "]\n",
    "#figlegend.legend(tuple(lines), (r'\\textsc{RedQueen}', 'Poisson', 'Karimi'), 'center', ncol=3)\n",
    "figlegend.legend(tuple(lines), (r'\\textsc{RedQueen}', 'Karimi'), 'center', ncol=2)\n",
    "# figlegend.savefig('/NL/stackexchange/work/prog/papers/online-twitter-broadcast/wsdm2017/FIG/legend-with-poisson.pdf', \n",
    "#                   bbox_inches='tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>type</th>\n",
       "      <th>Opt</th>\n",
       "      <th>Oracle</th>\n",
       "      <th>Poisson</th>\n",
       "      <th>kdd</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>s</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1.000000e-08</th>\n",
       "      <td>0.004255</td>\n",
       "      <td>0.002257</td>\n",
       "      <td>0.377543</td>\n",
       "      <td>0.228007</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.000000e-07</th>\n",
       "      <td>0.013699</td>\n",
       "      <td>0.006435</td>\n",
       "      <td>0.708564</td>\n",
       "      <td>0.571351</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.000000e-06</th>\n",
       "      <td>0.058041</td>\n",
       "      <td>0.013762</td>\n",
       "      <td>1.468919</td>\n",
       "      <td>1.378015</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.000000e-05</th>\n",
       "      <td>0.229522</td>\n",
       "      <td>0.074677</td>\n",
       "      <td>4.076188</td>\n",
       "      <td>4.230126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.000000e-04</th>\n",
       "      <td>0.866249</td>\n",
       "      <td>0.287207</td>\n",
       "      <td>9.320475</td>\n",
       "      <td>16.104886</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.000000e-03</th>\n",
       "      <td>3.766542</td>\n",
       "      <td>0.898753</td>\n",
       "      <td>25.476946</td>\n",
       "      <td>281.915090</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.000000e-02</th>\n",
       "      <td>16.224040</td>\n",
       "      <td>4.240223</td>\n",
       "      <td>87.953591</td>\n",
       "      <td>1259.529470</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.000000e-01</th>\n",
       "      <td>55.644573</td>\n",
       "      <td>20.417019</td>\n",
       "      <td>232.210492</td>\n",
       "      <td>2758.889224</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.000000e+00</th>\n",
       "      <td>218.081889</td>\n",
       "      <td>93.121546</td>\n",
       "      <td>1806.676547</td>\n",
       "      <td>10281.882413</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.000000e+01</th>\n",
       "      <td>430.028635</td>\n",
       "      <td>315.280360</td>\n",
       "      <td>2342.076609</td>\n",
       "      <td>60039.330064</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "type                 Opt      Oracle      Poisson           kdd\n",
       "s                                                              \n",
       "1.000000e-08    0.004255    0.002257     0.377543      0.228007\n",
       "1.000000e-07    0.013699    0.006435     0.708564      0.571351\n",
       "1.000000e-06    0.058041    0.013762     1.468919      1.378015\n",
       "1.000000e-05    0.229522    0.074677     4.076188      4.230126\n",
       "1.000000e-04    0.866249    0.287207     9.320475     16.104886\n",
       "1.000000e-03    3.766542    0.898753    25.476946    281.915090\n",
       "1.000000e-02   16.224040    4.240223    87.953591   1259.529470\n",
       "1.000000e-01   55.644573   20.417019   232.210492   2758.889224\n",
       "1.000000e+00  218.081889   93.121546  1806.676547  10281.882413\n",
       "1.000000e+01  430.028635  315.280360  2342.076609  60039.330064"
      ]
     },
     "execution_count": 168,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# hawkes_res_higher_beta.df.pivot_table(index='s', columns='type', values='r_2', aggfunc=np.std)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'piecewise_res' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast-online/redqueen/opt_runs.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mrank_plot\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpiecewise_res\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'avg_rank'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mlegend\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mfig_width\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;36m8\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m: name 'piecewise_res' is not defined"
     ]
    }
   ],
   "source": [
    "# rank_plot(piecewise_res, 'avg_rank', None, legend=True, fig_width=8)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "@optioned(option_arg='opts')\n",
    "def top_k_plot(res, field, file_name, K=None, with_std=True, save=False, \n",
    "               legend=False, fig_width=None, largeFonts=False):\n",
    "    vals = res.df.pivot_table(index='q', columns='type', values=field, aggfunc='mean')\n",
    "    std  = res.df.pivot_table(index='q', columns='type', values=field, aggfunc=np.std)\n",
    "    \n",
    "    if K is None:\n",
    "        assert field.startswith('top_')\n",
    "        K = int(field[len('top_'):])\n",
    "\n",
    "    if fig_width is None:\n",
    "        fig_width = 1.75\n",
    "        \n",
    "    latexify(fig_width=fig_width, largeFonts=largeFonts)\n",
    "    c1, c2, c3, c4 = sns.mpl_palette('Set2', 4)\n",
    "\n",
    "    X = [np.mean(res.capacities[s]) for s in vals.index]\n",
    "    plt.errorbar(X, vals.Opt, fmt=red_queen_fmt, yerr=std.Opt.values, label=r'\\textsc{RedQueen}',  \n",
    "                 color=red_queen_color, markersize=3, linewidth=0.75)\n",
    "    # plt.errorbar(X, vals.Poisson, fmt=poisson_fmt, yerr=std.Poisson.values, label='Poisson', \n",
    "    #              color=poisson_color, markersize=3, linewidth=0.75)\n",
    "    plt.errorbar(X, vals.Oracle, fmt=oracle_fmt, yerr=std.Oracle.values, label='Oracle',\n",
    "                 color=oracle_color, markersize=3, linewidth=0.75)\n",
    "    plt.errorbar(X, vals.kdd, fmt=karimi_fmt, yerr=std.kdd.values, label='Karimi', \n",
    "                 color=karimi_color, markersize=3, linewidth=0.75)\n",
    "    plt.xscale('log')\n",
    "    if legend:\n",
    "        plt.legend(loc=\"upper left\")\n",
    "    # plt.ylabel(r'$\\int {{I}}(r(t) < {})dt$'.format(K))\n",
    "    plt.xlabel(r'Budget')\n",
    "    plt.ylim(-5, 105)\n",
    "    plt.yticks([0, 50, 100])\n",
    "    plt.xlim(4.5, None)\n",
    "    plt.tight_layout()\n",
    "    format_axes(plt.gca())\n",
    "    if save:\n",
    "        plt.savefig(file_name, bbox_inches='tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th>type</th>\n",
       "      <th>Opt</th>\n",
       "      <th>Oracle</th>\n",
       "      <th>Poisson</th>\n",
       "      <th>kdd</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>q</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1.000000e-04</th>\n",
       "      <td>93.345349</td>\n",
       "      <td>99.660187</td>\n",
       "      <td>47.966030</td>\n",
       "      <td>48.181728</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.668101e-03</th>\n",
       "      <td>78.889262</td>\n",
       "      <td>97.282474</td>\n",
       "      <td>43.693084</td>\n",
       "      <td>43.458128</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2.782559e-02</th>\n",
       "      <td>54.291271</td>\n",
       "      <td>83.206772</td>\n",
       "      <td>34.772634</td>\n",
       "      <td>34.490786</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4.641589e-01</th>\n",
       "      <td>31.186412</td>\n",
       "      <td>52.643343</td>\n",
       "      <td>23.482270</td>\n",
       "      <td>23.466532</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7.742637e+00</th>\n",
       "      <td>16.288967</td>\n",
       "      <td>30.891380</td>\n",
       "      <td>13.852873</td>\n",
       "      <td>13.682464</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.291550e+02</th>\n",
       "      <td>8.368097</td>\n",
       "      <td>17.117852</td>\n",
       "      <td>7.922070</td>\n",
       "      <td>7.704958</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2.154435e+03</th>\n",
       "      <td>4.195675</td>\n",
       "      <td>9.995241</td>\n",
       "      <td>4.388770</td>\n",
       "      <td>4.765749</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3.593814e+04</th>\n",
       "      <td>2.115764</td>\n",
       "      <td>4.792608</td>\n",
       "      <td>2.453178</td>\n",
       "      <td>2.600518</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5.994843e+05</th>\n",
       "      <td>1.184343</td>\n",
       "      <td>2.765323</td>\n",
       "      <td>1.318151</td>\n",
       "      <td>1.456269</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.000000e+07</th>\n",
       "      <td>0.670819</td>\n",
       "      <td>1.512837</td>\n",
       "      <td>0.633804</td>\n",
       "      <td>1.059776</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "type                Opt     Oracle    Poisson        kdd\n",
       "q                                                       \n",
       "1.000000e-04  93.345349  99.660187  47.966030  48.181728\n",
       "1.668101e-03  78.889262  97.282474  43.693084  43.458128\n",
       "2.782559e-02  54.291271  83.206772  34.772634  34.490786\n",
       "4.641589e-01  31.186412  52.643343  23.482270  23.466532\n",
       "7.742637e+00  16.288967  30.891380  13.852873  13.682464\n",
       "1.291550e+02   8.368097  17.117852   7.922070   7.704958\n",
       "2.154435e+03   4.195675   9.995241   4.388770   4.765749\n",
       "3.593814e+04   2.115764   4.792608   2.453178   2.600518\n",
       "5.994843e+05   1.184343   2.765323   1.318151   1.456269\n",
       "1.000000e+07   0.670819   1.512837   0.633804   1.059776"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "piecewise_res_2.df.pivot_table(index='q', columns='type', values='top_1')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWAAAADXCAYAAAAk2Ys1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f238326c7f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "top_k_plot(poisson_res_2, 'top_1', '/tmp/test.pdf', save=False, fig_width=5, largeFonts=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 250,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>avg_rank</th>\n",
       "      <th>num_events</th>\n",
       "      <th>r_2</th>\n",
       "      <th>s</th>\n",
       "      <th>seed</th>\n",
       "      <th>top_1</th>\n",
       "      <th>top_5</th>\n",
       "      <th>world_events</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>3775.626447</td>\n",
       "      <td>271.450000</td>\n",
       "      <td>930276.062968</td>\n",
       "      <td>1063771.421300</td>\n",
       "      <td>4.500000</td>\n",
       "      <td>16.792321</td>\n",
       "      <td>45.866765</td>\n",
       "      <td>998.400000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>6877.641685</td>\n",
       "      <td>318.522294</td>\n",
       "      <td>3746813.663833</td>\n",
       "      <td>2999066.878561</td>\n",
       "      <td>2.886751</td>\n",
       "      <td>17.185526</td>\n",
       "      <td>36.256616</td>\n",
       "      <td>24.631424</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>101.417892</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>286.276565</td>\n",
       "      <td>0.000100</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.525209</td>\n",
       "      <td>956.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>205.294872</td>\n",
       "      <td>18.750000</td>\n",
       "      <td>1026.565139</td>\n",
       "      <td>0.027826</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.726655</td>\n",
       "      <td>9.158226</td>\n",
       "      <td>980.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>937.210705</td>\n",
       "      <td>102.000000</td>\n",
       "      <td>17952.364853</td>\n",
       "      <td>68.448802</td>\n",
       "      <td>4.500000</td>\n",
       "      <td>8.554189</td>\n",
       "      <td>38.553385</td>\n",
       "      <td>993.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>4467.226670</td>\n",
       "      <td>494.750000</td>\n",
       "      <td>355991.808174</td>\n",
       "      <td>35938.136638</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>32.230900</td>\n",
       "      <td>85.744226</td>\n",
       "      <td>1022.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>51602.111467</td>\n",
       "      <td>963.000000</td>\n",
       "      <td>34894547.293472</td>\n",
       "      <td>10000000.000000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>49.635027</td>\n",
       "      <td>97.436874</td>\n",
       "      <td>1035.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           avg_rank  num_events              r_2                s        seed  \\\n",
       "count    100.000000  100.000000       100.000000       100.000000  100.000000   \n",
       "mean    3775.626447  271.450000    930276.062968   1063771.421300    4.500000   \n",
       "std     6877.641685  318.522294   3746813.663833   2999066.878561    2.886751   \n",
       "min      101.417892    0.000000       286.276565         0.000100    0.000000   \n",
       "25%      205.294872   18.750000      1026.565139         0.027826    2.000000   \n",
       "50%      937.210705  102.000000     17952.364853        68.448802    4.500000   \n",
       "75%     4467.226670  494.750000    355991.808174     35938.136638    7.000000   \n",
       "max    51602.111467  963.000000  34894547.293472  10000000.000000    9.000000   \n",
       "\n",
       "            top_1       top_5  world_events  \n",
       "count  100.000000  100.000000    100.000000  \n",
       "mean    16.792321   45.866765    998.400000  \n",
       "std     17.185526   36.256616     24.631424  \n",
       "min      0.000000    0.525209    956.000000  \n",
       "25%      1.726655    9.158226    980.000000  \n",
       "50%      8.554189   38.553385    993.500000  \n",
       "75%     32.230900   85.744226   1022.000000  \n",
       "max     49.635027   97.436874   1035.000000  "
      ]
     },
     "execution_count": 250,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp = poisson_res_2.df\n",
    "tmp[tmp.type == 'Poisson'].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>avg_rank</th>\n",
       "      <th>num_events</th>\n",
       "      <th>r_2</th>\n",
       "      <th>s</th>\n",
       "      <th>seed</th>\n",
       "      <th>top_1</th>\n",
       "      <th>top_5</th>\n",
       "      <th>world_events</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>4620.816542</td>\n",
       "      <td>301.930000</td>\n",
       "      <td>1435908.914914</td>\n",
       "      <td>1063771.421300</td>\n",
       "      <td>4.500000</td>\n",
       "      <td>16.515543</td>\n",
       "      <td>44.783209</td>\n",
       "      <td>1155.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>8514.776225</td>\n",
       "      <td>360.750124</td>\n",
       "      <td>5029173.581471</td>\n",
       "      <td>2999066.878561</td>\n",
       "      <td>2.886751</td>\n",
       "      <td>17.136917</td>\n",
       "      <td>36.462044</td>\n",
       "      <td>33.881105</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>94.909487</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>281.466563</td>\n",
       "      <td>0.000100</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.138730</td>\n",
       "      <td>1113.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>217.732407</td>\n",
       "      <td>20.500000</td>\n",
       "      <td>1139.069265</td>\n",
       "      <td>0.027826</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>1.706495</td>\n",
       "      <td>9.025254</td>\n",
       "      <td>1126.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>1071.553104</td>\n",
       "      <td>111.500000</td>\n",
       "      <td>23536.617241</td>\n",
       "      <td>68.448802</td>\n",
       "      <td>4.500000</td>\n",
       "      <td>8.259371</td>\n",
       "      <td>36.931077</td>\n",
       "      <td>1147.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>4980.140291</td>\n",
       "      <td>529.500000</td>\n",
       "      <td>416858.523927</td>\n",
       "      <td>35938.136638</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>30.852401</td>\n",
       "      <td>85.071333</td>\n",
       "      <td>1198.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>56634.343917</td>\n",
       "      <td>1087.000000</td>\n",
       "      <td>42792074.660449</td>\n",
       "      <td>10000000.000000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>51.374306</td>\n",
       "      <td>97.641337</td>\n",
       "      <td>1207.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           avg_rank   num_events              r_2                s  \\\n",
       "count    100.000000   100.000000       100.000000       100.000000   \n",
       "mean    4620.816542   301.930000   1435908.914914   1063771.421300   \n",
       "std     8514.776225   360.750124   5029173.581471   2999066.878561   \n",
       "min       94.909487     0.000000       281.466563         0.000100   \n",
       "25%      217.732407    20.500000      1139.069265         0.027826   \n",
       "50%     1071.553104   111.500000     23536.617241        68.448802   \n",
       "75%     4980.140291   529.500000    416858.523927     35938.136638   \n",
       "max    56634.343917  1087.000000  42792074.660449  10000000.000000   \n",
       "\n",
       "             seed       top_1       top_5  world_events  \n",
       "count  100.000000  100.000000  100.000000    100.000000  \n",
       "mean     4.500000   16.515543   44.783209   1155.500000  \n",
       "std      2.886751   17.136917   36.462044     33.881105  \n",
       "min      0.000000    0.000000    0.138730   1113.000000  \n",
       "25%      2.000000    1.706495    9.025254   1126.000000  \n",
       "50%      4.500000    8.259371   36.931077   1147.000000  \n",
       "75%      7.000000   30.852401   85.071333   1198.000000  \n",
       "max      9.000000   51.374306   97.641337   1207.000000  "
      ]
     },
     "execution_count": 251,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp = hawkes_res_2.df\n",
    "tmp[tmp.type == 'Poisson'].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 253,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>avg_rank</th>\n",
       "      <th>num_events</th>\n",
       "      <th>r_2</th>\n",
       "      <th>s</th>\n",
       "      <th>seed</th>\n",
       "      <th>top_1</th>\n",
       "      <th>top_5</th>\n",
       "      <th>world_events</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>100.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>2999.325583</td>\n",
       "      <td>220.010000</td>\n",
       "      <td>565300.529805</td>\n",
       "      <td>1063771.421300</td>\n",
       "      <td>4.500000</td>\n",
       "      <td>18.048286</td>\n",
       "      <td>48.686186</td>\n",
       "      <td>754.700000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>5323.113679</td>\n",
       "      <td>249.077023</td>\n",
       "      <td>2132068.274656</td>\n",
       "      <td>2999066.878561</td>\n",
       "      <td>2.886751</td>\n",
       "      <td>17.429374</td>\n",
       "      <td>35.935099</td>\n",
       "      <td>25.085913</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>93.517316</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>245.415447</td>\n",
       "      <td>0.000100</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.565336</td>\n",
       "      <td>715.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>184.210533</td>\n",
       "      <td>18.000000</td>\n",
       "      <td>865.050901</td>\n",
       "      <td>0.027826</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>2.726000</td>\n",
       "      <td>12.729771</td>\n",
       "      <td>736.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>763.677975</td>\n",
       "      <td>93.500000</td>\n",
       "      <td>13713.628282</td>\n",
       "      <td>68.448802</td>\n",
       "      <td>4.500000</td>\n",
       "      <td>11.058082</td>\n",
       "      <td>45.657781</td>\n",
       "      <td>749.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>3559.701735</td>\n",
       "      <td>412.500000</td>\n",
       "      <td>210573.253905</td>\n",
       "      <td>35938.136638</td>\n",
       "      <td>7.000000</td>\n",
       "      <td>34.647645</td>\n",
       "      <td>88.574192</td>\n",
       "      <td>776.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>39128.203550</td>\n",
       "      <td>743.000000</td>\n",
       "      <td>19743325.457668</td>\n",
       "      <td>10000000.000000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>50.042244</td>\n",
       "      <td>98.029689</td>\n",
       "      <td>799.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           avg_rank  num_events              r_2                s        seed  \\\n",
       "count    100.000000  100.000000       100.000000       100.000000  100.000000   \n",
       "mean    2999.325583  220.010000    565300.529805   1063771.421300    4.500000   \n",
       "std     5323.113679  249.077023   2132068.274656   2999066.878561    2.886751   \n",
       "min       93.517316    0.000000       245.415447         0.000100    0.000000   \n",
       "25%      184.210533   18.000000       865.050901         0.027826    2.000000   \n",
       "50%      763.677975   93.500000     13713.628282        68.448802    4.500000   \n",
       "75%     3559.701735  412.500000    210573.253905     35938.136638    7.000000   \n",
       "max    39128.203550  743.000000  19743325.457668  10000000.000000    9.000000   \n",
       "\n",
       "            top_1       top_5  world_events  \n",
       "count  100.000000  100.000000    100.000000  \n",
       "mean    18.048286   48.686186    754.700000  \n",
       "std     17.429374   35.935099     25.085913  \n",
       "min      0.000000    0.565336    715.000000  \n",
       "25%      2.726000   12.729771    736.000000  \n",
       "50%     11.058082   45.657781    749.000000  \n",
       "75%     34.647645   88.574192    776.000000  \n",
       "max     50.042244   98.029689    799.000000  "
      ]
     },
     "execution_count": 253,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp = piecewise_res_2.df\n",
    "tmp[tmp.type == 'Poisson'].describe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 249,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAI8AAABYCAYAAADMWKQEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n80iCRppHEjTSPJKgkeaRBM3/AzTWkP8+Wq9iAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc80888d198>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "top_k_plot(hawkes_res_2, 'top_1', None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 252,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAI8AAABYCAYAAADMWKQEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nkfiNNI/Eb/4P0EZ/2oh8pAAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7fc8398b79e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "top_k_plot(piecewise_res_2, 'top_1', '/tmp/test.pdf', save=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Plots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import os\n",
    "# base = '/NL/stackexchange/work/prog/papers/online-twitter-broadcast/nips2016/figures/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "poisson_res.raw_results[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1451,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 10.3 s, sys: 47.5 s, total: 57.7 s\n",
      "Wall time: 1min 55s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "plot_base = '/NL/stackexchange/work/prog/papers/online-twitter-broadcast/wsdm2017/FIG/'\n",
    "for name, res in [('poisson', poisson_res_2), ('hawkes', hawkes_res_2)]: # , ('piecewise', piecewise_res_2)]:\n",
    "    file_name = os.path.join(plot_base, 'avg-rank-{}-wall.pdf'.format(name))\n",
    "    rank_plot(res, 'avg_rank', file_name, save=True, end_time=simulation_opts.T)\n",
    "    plt.close()\n",
    "    \n",
    "    file_name = os.path.join(plot_base, 'r2-{}-wall.pdf'.format(name))\n",
    "    rank_plot(res, 'r_2', file_name, save=True, end_time=simulation_opts.T)\n",
    "    plt.close()\n",
    "    \n",
    "    file_name = os.path.join(plot_base, 'r2-{}-wall-linear.pdf'.format(name))\n",
    "    rank_plot(res, 'r_2', file_name, save=True, log_y=False, end_time=simulation_opts.T)\n",
    "    plt.close()\n",
    "    \n",
    "    file_name = os.path.join(plot_base, 'avg-rank-{}-wall-linear.pdf'.format(name))\n",
    "    rank_plot(res, 'avg_rank', file_name, save=True, log_y=False, end_time=simulation_opts.T)\n",
    "    plt.close()\n",
    "    \n",
    "    for k in [1]:\n",
    "        file_name = os.path.join(plot_base, 'top-{}-{}-wall.pdf'.format(k, name))\n",
    "        top_k_plot(res, 'top_{}'.format(k), file_name, save=True)\n",
    "        plt.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "avg_rank        4325.24\n",
       "num_events           10\n",
       "q                599484\n",
       "r_2              265292\n",
       "seed                  2\n",
       "top_1           2.33567\n",
       "type             Oracle\n",
       "world_events       1035\n",
       "Name: 392, dtype: object"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "poisson_res_2.df.iloc[392, :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10.0"
      ]
     },
     "execution_count": 68,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "simulation_opts.world_beta"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 0 ns, sys: 0 ns, total: 0 ns\n",
      "Wall time: 20 µs\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "def plot_intensity(sim_opts, name):\n",
    "    wall_mgr = sim_opts.create_manager_for_wall()\n",
    "    wall_mgr.run()\n",
    "    assert len(wall_mgr.sources) == 1\n",
    "    broadcaster = wall_mgr.sources[0]\n",
    "    X = np.linspace(start=0, stop=sim_opts.end_time, num=100)\n",
    "    Y = [broadcaster.get_rate(t) for t in X]\n",
    "    latexify(columns=1, largeFonts=False)\n",
    "    plt.plot(X, Y)\n",
    "    plt.xlabel('Time')\n",
    "    plt.ylabel(r'$\\lambda(t)$')\n",
    "    plt.title('Sample intensity of {} wall'.format(name))\n",
    "    plt.tight_layout()\n",
    "    format_axes(plt.gca())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "redqueen/opt_runs.py:174: UserWarning: Consider using `run_dynamic` instead of `run`.\n",
      "  follower_wall_intensities = wall_intensities\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 3.07 s, sys: 1.24 s, total: 4.31 s\n",
      "Wall time: 24.3 s\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPMAAACWCAYAAAALz77WAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQswCQYrw/wGQuwd0bXkhdQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2382f52128>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%%time\n",
    "plot_intensity(SimOpts.std_hawkes(world_seed=1, \n",
    "                                  world_beta=10, \n",
    "                                  world_alpha=1, \n",
    "                                  world_lambda_0=1000).update({'end_time': 2}),\n",
    "               'Hawkes')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "redqueen/opt_runs.py:174: UserWarning: Consider using `run_dynamic` instead of `run`.\n",
      "  follower_wall_intensities = wall_intensities\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPMAAACWCAYAAAALz77WAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJztnXlwHNed37/dPfdgpgHiIkgMSYmHJAKyubJkSUNL4iou\nG2LibK0qNrybKtf+Ya7WqS07ibOp/CO5wnXtJmUncexsJQ6S1R+bpARpnd21vRTkWBYtm2Nbti5y\nhiIlUjwaIIh7pgdz95E/ut+b7jkxwGAO8H2qVAKGPT0P0/19v/O95nRd18FgMLoevt0DYDAYzYGJ\nmcHYITAxMxg7BCZmBmOHwMTMYOwQmJgZjB0CE3ODSJKESCSCSCSCqampLZ0rFothZmam7ufFYrEt\nfU6zzzc9PV02bkmS8PTTT0OSJPqzLMsbOl8kEqn7PWwW8h1v5LvudpiYGyQSiUAURYTDYYTD4S2d\nKxQKQZKkuscFg8Gy1yRJ2vTNGQwGN/1+ItCJiQnb66FQCKOjowiFQvTnSuOuRCgUQiKRaHgsGz23\nJEkb/q67GUe7B9BtTE5OYmpqCt/85jdx6tQpaolisRhOnTpFrczk5CTOnDmDkydPIhqNYnJyEpFI\nBOfOnaOvPfXUUwAM65FIJCBJEiYnJ8s+c2ZmBmNjY/S80WgUoVAIFy5cQDgchiRJtvdbxxCNRiGK\nIsbGxhCLxeh5xsbG6PuDwSBmZmZw4cIFPPPMM5iensapU6fo2AAgkUggFApBlmXEYjHIslwm1mQy\niUgkQn8GYPt+gsEgnQyIuKyTwvT0NEKhEERRtP09ZLyxWIwePz09XfVc1utxJ8Esc4NIkoRTp07h\n+eefx7lz5yCKIgAgHo9DlmWMj48jFAphbGwMyWSS3oSAcdPt27fP9hpQvImJWKyQ81vPG4vF6Lkq\nvb/SsS+88ILtfNb3A4alTSaTiEajNhFMT09jbGwM4XAYU1NT9HMqWd1AIEA9lkAgYPu8eDyOp556\nCpIkYWxsjP4/FAoBMCaNmzdvIhwOl/091vETJicnK56r9HrcSTAxNwiJlyORCI4fP46XX34ZANDb\n24uXX36ZWiJZlm3/J8RiMWohybETExOQJIneiFbIMaXnJTd6pfeXHhuNRnHy5Enbv1nfT5icnCwT\nwMTEBGKxGCKRCBVQJXdVkiTMzs7Sc5KfS78fIlIA9P+SJCEQCGDfvn2Ympoq+3us47dS6VzVrke1\nce8kONab3TqI+3v69Ol2D6WMmZkZBIPBLecBGO2DiZnB2CEwN5vB2CEwMTMYO4SmipkkSgjT09OI\nRCJNbXpgMBiVaWqd+cyZM7j//vsBGEIOh8M0y8hgMLaXplpma/mAlGBmZmbKSgLvvPNOMz+WwWBg\nG2Nm0kAwNjZmK/ifPXuWiZnB2AaaWpqamppCPB7HM888AwA0frZ2+gCGoE+cONGsj2UwGGhTnZmJ\nmdHpaLqOvKrA43C2eygbhpWmGIwK/M31d/Bv3vhb5FWl3UPZMEzMDEYFljLryKgFrCu5dg9lwzAx\nMxgVUHQVAKBqWptHsnGYmBmMCqhmKknRmZgZjK6GWGSFWWYGo7shFpm4290AEzODUQHVFDOLmRmM\nLoe62SxmZjC6G+pmM8vMYHQ3qs4sM4OxIyBu9h0bM1s3J5AkCV/+8pcxPT3dzI9gMFqC2oXZ7G3b\nnAAAvv71r2/4qQYMRieh3Ol1ZuvmBKIo0kegWPdiPnv2LK5fv97Mj2Uwmo7CYuYi0WiUrmO2uton\nTpzAgQMHtutjGYymcMfXmSORCC5cuEAfKRKLxRCNRsseMsZgdDrdWGduasxsfUYRiZXHxsaa+REM\nxraj6To0mAsttO5JgLHSFINRgmqxxt1kmZmYGYwSrHHyHRszMxg7AatlLnRRnZmJmcEowepaM8vM\nYHQxVgGzmJnB6GJUZpkZjJ2BwrLZDMbOwFaaYpaZweheFFvMzLLZOwJN11Hoog4gRnNgMfMO5P9c\n+TWe/c0PoHVR3MTYOiybDfvmBITnnnuumR/RUubTCazl0sgo3fO8IcbWUVjMbGxOYF27HIvFmnn6\nlpPXDBFn1HybR8JoJaw3G/bNCSRJQjAYRCAQsB3TTZsT5FUjXs4ohTaPhNFKbG52F+VMti1mliQJ\nsixDkiRIkkRf76bNCUjyK6MyMd9J2No571TLbN2cIBwOIxQKIZlMNvMjWgp1sxXmZt9JdGudeds2\nJwCMDQqef/75Zn5ES6FuNrPMdxRWAd+xlnknoes68hqLme9EbEsgu8gyMzFXQdE16ObWMUzMdxa2\nppFNdIBdT67gv8TOIlXINXNYdWFirgJxsQFWmtppRFdvIZHPVP13WzvnJizz+dU5XFi9hWvJlU2N\nb7MwMVeBJL8AIMss845hKbOO78TO4u9vRqseo5Zks3Vdb+gzSBUkp7a22YiJuQrWnmyWANs5LGfX\nAQDJfLbqMUTMHAAdRo9+I5B7x2oQWgETcxWsF4KVpnYOa/k0gNoTNHGtXbxR7Gm0C4yKWW1twwkT\ncxXsMTOzzDuFtVwKQO0JWjUtsVswxdxg3EzdbGaZOwObm81i5h3DWs5IfGVqxLNkDbOHiLlORlvT\ndfz89hWkCsYEUbTMHSrmixcv4r333tvOsXQUdjebiXmnsJYz3exalpm42aaY661pvhS/jb/64A2c\nW7gKoIMt84svvoiXXnoJABCPx/HSSy/dEaJmpamdSdyMmbM1Qqeim+0EUD9mJmWutNJey1yznTOZ\nTOJzn/tc2evWhRNWYrEYEokEwuEwZFlGNBpFLBbD5ORk257TvJpLgQeHXrevoffZSlOqAk3XwHMs\nKul2Vk3LnNdUqJoGgS+/pkS8ng3GzClTxOSeIV1j+RavuKp5d5Lli7Ozs7b/h0Khisdb1zNHo1GM\nj48jGAxWFX8r+PaF1/DfL/284feVXohsi2fZTuTm+ir+84Wf1Gy46GRyqkKtJ1A9sama197Nbyxm\nJp1eBdObK5iibnWduaZlvnjxIl544QXIsozjx48DAD772c9WPf7kyZNUuMQ6y7LctidB6rqOpew6\nelR3w+8lLpJHcCKrFpBRCvA5XM0eYlcRXZ3HxfhtXJWX8cBA5Qm9k4mbVpmQVQvocZbfG6XZ7Hox\nc9EyEzETy9xBMfPRo0dx+vRpfPWrX8UjjzzSkCinp6cRDAYxNjaGqakp+norNycoaCoUXduUVSUX\nRnR5AHRG3Px319/Fn751pm2bzJGbs9WxYLNYLRFztcQmcbNpaapOzLxOLLPWXstcU8xWtzoUCuHo\n0aMAUDUBZl3PPD4+jkgkAkmSbA9bb+XmBGTGzKnKplvyRJcXQPsz2rqu4+e3r2I2FYdcqN69tJ0Q\nEbf6Jm0WJPkVdJoTdJWMNpksaQKsrmU2xNzumLmmmz06OopXXnkFsixDFEXE43Ekk0l8+tOfrni8\ndT1zJzxkncRHOozljGSm3Qjkxu0UMc+nZSri9UIOfQ0m9JoBuTmzWneW6khZao9fhBzPVo+Zyyxz\nvZjZdLNLYuaOymYDoMKVJAm9vb1le3p1MtZkR04tNCZm88als3ib3exL8dv0Z2IJWk2+Te5js6Bi\n9om4FF+oKmZFs4u5XlizXsUyd5SbbSUUCiGRSGznWJqOVcyNxs3kwnSKZbaJudCeiSXX5W426cse\n8fUCqL4aTtU1CBwPh1mKrBczpy2dX7rlwQkdVZoi/OIXv8DXvvY1vPLKK9s9nqZiF3NjYiRlBirm\nNvZnq7qGy4lF+vt6ixe9E/JtWtrXLNZyabgFB3aZIUpVy6xrcHA8HGYNulbMXNBU2umVNxOuhI5y\ns1988UVEIhGMj48jHA5XjZU7lZTNze5ey3wzuYqsWkC/24+VXIq6da2mmADr1pg5gz6XD16Hkdiq\ndk2NZhIOwgYss3U3kbyq2DoH85qReOU4rhnDr0tNyzw+Po5vfetb+OIXv4hgMIgf/ehHLRlUs9iK\nZaYxs6t25rMVvBdfAAB8bHAfALR8OxpCN8fMBU1FSsmh1+2F18xSV7snDDdbgIMXjN9rWGarwTBK\noUUx60BLn1VWt85MePTRRzE6OrrtA2omW4uZVTh5gTaKtNPNnkutAQA+smsvgDYmwEyr043dcGQy\n9jlc8FDLXHmCVnQNAsdZYubqgrSGPHlNLVvD3MrGkYaaja3i7gbSW3GzVQUuXqCzeDvd7JVcCgLH\nY9RvJG7W25UAI5a5xZ1NzSBr6ejzCrUnaFXX4OA3FjOXWuZS8eZauEHBjl45sFU328kLcAsOcODa\naplXsin0uX3wCE4IHN/GmLk9q4GaAbn+HsFhXtM6MTMnNBwzA+XWvpUhyR0j5kaTNnlVgUtwgOM4\neB2OtsXMBU2FXMii3+0Hx3HwO1wsZt4EVsvMcxzcZs99JaibvaGY2bgWTvPYVMl90rFudreRLjQW\nMyuaioW0seqroKlwmRfIK7jaZplXs8Y2N/0ePwCgx+luS8xsffB8N8bMOWqZjbDJ63BWt8zEzd6A\nZSYhD+nIIzE0ByOD3UovZkeLudHS1Gu33sfX3vwh5tMJ6mYDRtKkXdZwxdyzqt9dFHNaKbT8sSnW\nrGw3lqaIcKmYBWeNJZBm0wiNmavHvWRi7XMZYk5bEm1Aa/ML2/awdUmSEIvFMDU11bb1zGklT9sx\nNxIz387I0AHMrq9B1TW6O2PQ5UFeU9viXq6UWGa/w1iyl25xEsz6tyu61raVW5uFeBNuh3FNvQ4n\nskqhbAGOruumm70xy0wmeWKZiQHpcRpibuUOndv2sPVYLIZQKIRwOIyZmZlmfsyGyKsKFF2j3T4b\ncQ1Jm+RCxnhypUswLHPAnBCSbVitVGqZ/eZN0uokWFmWtssy2sSb8Fosswa9rOVSMx9J5OB5ugtJ\nLS8opeTBgaP9CETcZJ30RmLm8ytzWDWv81bYtoetT0xMIBgMIhKJ2JZAtgri7vSZItiIa0hcJipm\n0zITMcs1Nk7fLoqWuQdA8SZptdtfamG6LQmWLYmZPULlWjPxOAzLbEzmNd3sQg5+h4vuSkLuO+JB\n1fueVnMp/MXFn2JGutjQ31OJbY2ZI5EIfU4zoVWbE5AvNeB0w8HxG7LMJHmxkDG8C5IAI7Pudlpm\nxWzSL2WF7mFmtJX2mDfJep3suq7ruJJYhNak2Lq8ftpdcTN1swXiZrvM1+1/B7HCG+3NXlfy8Dtd\ndCdP4mb7qWWu7WbfShmLl0g4uBW27WHrkUgEU1NTmJ6etrnZrdqcgCYinC54BCdyG2j6KIqZuNnE\nMhsXZrOWud7jTeR8Bv/yl9/Da7feL/u3lew6+tw+WvP0b9AyX4ov4Bvnf4yfzn+wqTGXUpqV7YSM\n9rXkMp79zQ/o5FvK6/NX8OO5SwDKLXO1/mwiXGED2Wxd15FScvA73GWlqR6SAKvzPZGxD/u2vuHl\ntj1sPRwOIxwON/P0DZGyuDtuwVH35jMujPEecuGbYZm/f/08fnb7Cr509HHcHRyoeMytdAI5VcGH\nyWU8iXvo64qmIpHP4FBwiL7md2wsZl40J6To6jx+e889NY/dCCRGdnA8FF3rCDf73ZU5LGaSeD++\niGGvXQxpJY/pq7+BwPP45N576XJHImLqZpdYZiJcgasfM2fVAjRdNyyzea+klcZiZmI4dnu3LuYd\nW5qylgg8ghO5Ortj5FSl7KKVxsybEfO7q7OQC1l868JP8N7a7YrHEItfuuvlWi4NHcVMNmCNmWu7\n2YmCca4r8mJTylgkZg6YE1sniJn0BMiF8t1C31q+SSedgqaWu9lV2nRtbjZX280mNeYeh5uKmVwX\nEjPXqzPfNv+GIe/WN/24M8TscCCr1N4HrJKlc5Zksxt1swuailvpBIJOD1Rdw9SlcxVdbiLiRMn5\naSbbKmYSM9dxs8lTDrOqgpvJ1bpjvZWK0yckVoJYGBLbdULMfNt0UUu/NwD4xcI1+nOqkENWLYBD\ncftc6maXxswWN1uoI2Yyufc4PTQkI+cj4VCuTsy8kJHR5/I1tAtONXa8mEmmUYNeczlaJXGQ2ZbE\nzMkGM8hzqTg0XccDAyGM79qDlJKr+DnkZpRLLDPNZLuLYqYxcx0327rp36XEQs1jdV3Hf7rwE/zH\n869WrR9Ty0zE3KKlfb9euoFr8nLZ65qu0VCi1KNZyqzjirxEf08peWTNbaPI2mJvlWy21c3mzJVT\n1VZNxc3P7XV7qRdHIDFzLcucVQuI5zMY9jVnK64dL2biZgO1XUMijoBlH2VygRzmUshG3eyb64ZF\n3NezC71mhxDZIdIKcRMzasF28ckNOWhxwXwOJzjUt8yJfIa2FF6O1xZzRi1ALmSxkkvhrZXKDT7E\nMhfd7O23zIqm4i8vRfDih2+V/dtqLk2FV+ox/WrRsMoDpkezXsghqyr0PgCK4cr7iUWbx2Z1swHD\nQlcLUxLmtex1eWkCjEAtc417jkxGpfH+ZulaMb+fWMQfn5vGtWT5rA3Yxex2kMXoNcRsxjqj/j76\nGnGzAcO9bNTNvrlurEPe17MLfWZpKZ4rj++sloVY6bVcGr9avI5hbwCHLIkznuON9tI6palkIQvR\n5cEen4ir8lLNWqm1YeHHs+9VDEeomDdwkzaLlJKHBuNBBqWQWBMot8zRtVvgOQ7Hhw8BMMScUws2\nMd8VHMDh4BDOr87hldni1tFWNxswE35VvBVqmV1e2mBEIInKWgkw8jc0I/kFdLGY316+iYKm4lIV\nq5Oyitm0sLVaOomls4rZ6joFnB6klFxDyaSb66twcDz2+MTaltkySZAb88dzl6DqGj41erTsGVd+\np7tmaUrXdcj5LIIuD46Iw8hrKq4nV6oev5o1xsSBw/X1VVyt4NaSdbnFmHn7xUyuSbKQLfs8YtUA\nI6QgE5Cm67iVTmC3N4h+D2mxJJa5eD0Fjscf3nccvS4v/vb6O7i4Ng+gaJlJvOzghaqlKTIxiy4f\nDckIbsEBJy/UdLOLZak70M2+kliigvzQvOEW0pVrjGmlAB4cPIIDHrMft9YNSBJgoZ5e+pr1AgVc\nbujY+GZ6iqZiLhXHXn8vBJ6nTR+VLbNdzOuFHH42fwW9Li8eHjpQdnzQ6UGykKsa3+ZUBXlNRdDp\nwRHRKGtdqSBQArHMj+0+CAD462tvlU18pW52K+rM1u96pcQ6k+RXn8tnZquN8a7mUsipCvb6e6kr\nnchnUdBUuqk9Iejy4o+OPgaB4zF16RyWMutUuMTNdnB81e85USNmdvICXLyjZm5h4U51s2+nZXzj\n/P/D9z58G3lVwU1zK50FywxtJVXIwedwgeM46l7VsszE0o34RPBmrOmyzOTBBstT82kZiq5hf88u\nAMVVNaWWWTH3piIk8hm8tSwhpyl4cu89ZbEYAOxy+6FDp25eKST5FXB5aW37wyrhCFB8bMvHhw7g\n44P7cS25gr+I/dQ2+RELQy3zJnuzNV3HjeQq5lLxumvErd/LSknvMrFqh8VBAEVhzaXiAIA9vl5a\nHiITAclgW7krMIDfO/QQ0koe/+2912mpirjZAs9XTZzG8xn4HS44ecF2nTgYlt0t1LbMt9MynLyA\nXZYE51boIjEbbW9vLku4nlyhJZ5qYk4WctSKEDe7dgKMtH96IJpW1GaZiZjzG7PMJPkVMsVMHim7\nViJAIjwidrmQxaw5Ud0r7q547l2m+1h6g9NzmpZedHrQ5/ahz+XDh/Jy1dLcqmUxxx/c8yge6A/h\n/cQizkhRegxpSwxssTT1zoqEP3tnBqffOoOv/vL/UvHdSK7i57ev2MZotcylZbOFdBJ9Lh8GPYaL\nSrybObM9cq9fpJaZfE+eKuWfT+w+iOPDd2M2Fce7K8YjmWyWuaqbnaa7t1onficvgOM4uHhH1ZhZ\n03UsZpIY8gTAN2n3zq4R87J5QVJKDq/MGk3pDo6vWO5RdQ0pJUeTNR5Hfcu8blntQlZaWcVMusAq\nNSgAhmX47sWf0ZuumPwyYnC34IBXcJY9iZAIL2Qel8hncCuVAAcOI1Va/EipqtpKGzJGMpndHRxA\nspDFcrby8au5NDhwEN1eCByPf3r4IQCGd0GglnmLTSM3zEnurkA/VF3DhdU5AMALV3+Dv/rgDVsO\nxLrXmXXsOVXBWj6NYV+geF3MSfJW2pgc9vp7aU2elPhK3WwrDwwYO59K5kRajJn5ijFzTlWQUQvo\nNcVstczkZ5fgqPo9LWWTyGkK9vjFqmNqlK4R84rlYkbNZMVH+o3dKkt7c4nLTKwImZErfbHEhVov\n5ODmHTa3x5r9LHaBVbbM767M4q0ViZZFbqXj4GA8CoXQ6/aVudnEPbSKeS4dx5C3xzbbWyHjW60i\nTjJBEJe46GovVTx+NZdCrylkAPA53BA43lb3Jm613+ECB67su1zKJPG/Pnij7uofkrj6/UPGhHE5\nvoC0ksc1M0H3g5sXqHW2utlWy0wTR94gtYyJArHMcbh5B3a5/XALDggcT8MITw0xj5jXiUxg1M02\ns9mlXk0xXjYmft6ymycRs5t3IF9lAc0Ns5HnQKC/6pgaZds2J6j0+1YgrhLplBFdXtzXOwLAntkE\nioIjbhZ5T6llvpJYwpfPvYi3lyWjYd5cK3wyNI7fP/gQdlk6r+p1gZHnGM2ux6HrOuZScQx6AzZB\n9rq8SCsF5FQF//bNv8dfXo5Q93DIE4BbcODm+hrSSh57fL0VPwcA9RxKH1FKIK47udHvDphilssz\n2qquIZ7L2OI2nuPKSnF5VTVuWHOTQ6uYfzJ3Gc+9+UP87PYVfP/G+Zpx4mImCRcvIOTvw4hPxBV5\nCbG1eejQIXA8rspL1DrbE2DFicta0iGWOZHPQNFU3M7I2OMXwXMc3TNNN9co1xJzn9vISJfWmYe9\nQai6htfnr9iOJ/kK8h0DRVe7aJmN/1eKua+vG9figBmGNYNt25yg0u9bYSW7DrfgwIMD+wEYN+hu\ns5miNG4mN2HRMleuM59buAoNOn69dAOpQp4mTPb4RTyx57Dt2HqLLUgsPJtaQyKfQUrJY2+JIMks\n/t7aPG6lE3hz6SZNzgRdHoguLz3/3hruFxFevZiZ/P2hnj44OB4fJpdxI7mKXy/doNYikctAh04n\nCOvfay355DWFZmwNMRsTY05V8Hc33oVXcOGIOARV16grXYqu61jKrGPIGwDHcTgiDiGvqXjFtOaf\nvfsBAMAPb14AYN+SZyVXtMzEFR7191IxyfkMFjJJaLqOvf7i9259mHq1mBkwJrDdFi+KeClP33UM\nPocLf33tLSxZ7jMSLvVaxWyK2GqZyXdUyo3kKjhwGO3pK/u3zbJtmxNU+h3Y3HpmXdexnE2h3+3H\nI2ap5mjfCF02VlqeWifZXBIz0w6womVWNBXvmN1OF9fmkdMU24Uvpd5iC3JxF7PrNHNcGg+RC//G\n0g1jDLqGN5dvAjBmeNFZvDGsN2QpHocTPocLa9XcbGqZjTE7eQH7enZBWl/Fn78zg/9x6Rw+MJ9d\nReLu0oxq0OWxlXzyqkI9HOsqtLeXJWRVBU+MHMKJkSMAYGultJLIZ5DTFAyZSat7xGEAhji9ghOP\njxzCweAAbXJZL+TAcxz2+nuRVgq0EWg2ReLiPhpKJPJZ3KKZ7OL3bhNzhWy2FWuOgrjZosuL3zv4\nIPKaiv95OUK9jlI3GygXM7HUpZl/Tdcgra9hxBes6S00Sstj5s2sZ04rBeNZSx4/jvQO488e+h18\nYvdBBJ0euAVHmWVOlsTMRTe7+KVeii8grRRse2KTrp1KeMwmgLUqrq01sfXGoiHWUstMMtbnzaQP\nANrdJLo81PoDqOlmA6DPnaoUj8n5DASzU4xwRByCjuLN9/ptw20krnqZZTYnFhIG5C27lbotWdrI\nwocAgPDw3ThklomuVhHzovm3khVCpAYOAPf0DkPgeAx6AtBhhC2pQg49DjdtyyRx8+z6Gna5fXRT\nAK/ghJzPYi5NMtnF7454W0BtywwAu73FScBhadR5aHA/Pj54ANeSK5i6dM4ITSzdXwQnb3eziTEp\n7S2/nTaSX/ub6GID27g5QaXfNwtxsfrdxtY5/R4/jYmGvUEsZpO21UjJMstcHjP/xrSIv21xp2tZ\nZo7jcFegH7OpeEVBr1kSWyRDW2pdSeNIwWzoIBed5zj4HG7qMjp5AUPenqpjAQzx5TW1YlunnM8i\n4HTbHlj21L4x/PHYE/j6g5/BiDeIt5YlJPPZmpYZKFr5nFrqZitYzq7jcmIBh4NDGPQGILq8GPT0\n4Kq8VHF1GMltEDEHzHZTADhq5j/6qXBTWFfy6HG6MWBumbSSTUHOZyAXsrZOPdHlxVo+jV8tXoOD\n4xHyF0Vid7NrW8E9VstsETPHcfjCkYdxX+9unF+dw/TVN+nkbY+Z7Zb58ZHDEDgef3P9XVvcfMOM\nl/cHOljMp06dwp/8yZ8gGAxW/H2zkOTHgKe8uD7sDaCgqTaBUctM6swla1cVTcW7KxJ6XV6cDI2D\n3PL+GmIGgAcGjO2P3lmRkFHy+E70LN5elpBRCsiqCp2lySNBB0sESVo6AcMSEcskOr3gOY7eGCO+\nYFkLZykkOVdantJ1HXIhi6DlJgOMG/n+XXvh4AU8NnIIqq4hsvhh0TJ77JZZLCn55DWV3qwewQEd\noDujPDp8F33foeAg0koB8+nyZ3mXihkAjvWPwskLGN+1B0Cx7LaUXUdGMfIYZP+z5ew6dbFHLRNl\n0OVBWsljNZfGiT1HaCITgO3nemK2xsxkyyCCkxfwR0cfw4hPxOvzH+Bqctm2kR85Bii620PeAJ7c\ncwQruRRenbtMjyM5hf09zctkA11Smqq0rpdAmtRJ8wFQbpmdvIB+tx9Sag2aruODxBLSSgEPDIQQ\ncHloWajHUVvMx/oNMb+9PItXZt9DdO0Wfn77Kp1I7usboTP6iE+0ze4A6GILADgsDuFon2GNyA1B\nBFTPxQaql6fIYvxae0o9MnQ3nLyAV+cu04e4l1lmJ0ksZaFqmm3rYTI5vjp3GQGnmz6dEgAOBqu7\n2pXE/I/2349/9/HfodeW/F9aX4UOmJa5KPBi8stumY1xOTAxan8eWk8Dbvagt4des9JrZ7zfiacP\nHKNhQNCyqTznAAAKmElEQVTlsR1nXWVHOLlvHH6HGy9LUWqdbyRXwXOcbUJqBt0h5qzdzbZCbp4P\n5OLDyJOFHDgYrivhsDiEtJLHfDpBb+CxPsMakBJXLTcbMMoXdwX68X5iEa+ae0vNpeK0djzg9tMk\nSqUEVo+zePGPiEPUtSRx7Ki/DxwMq12P/ioZbfK76KouZr/ThU+P3oeEmQH2OZx0fS+BvD9RyND4\n2JoAAwAdOr5w+BGbxSNx85mbMfz52zN47VbRIi1mknALDttEI3A8eiy/k2tM2nX9TheGvUai6M2l\nm7iaMCaJ0R67ZQaAT+65l3pjhEbcbIHj6URTScwAcP+uPbTUJ5Z4P64SywwYC30eHNyHrKrgdlqG\npuuYTa1hxCtW7SPYLM092zZRuhG8lYPBAQgcj/dtnUNZ9Djdtja5w+IQfrl4DR8kFnEpfhs8x9Eb\n75N774Gma7QJpRYPDOzDteQKfeLFWj5NvYI+tw+j/l7MpuIVO3t4jsOgpwdZtYDd3qARix1+GPvM\nREiopw///uHf3dBOjZVqze+szOJ/f/AGALvlqsRn9n8En9h9COdX5tDv8Zc9EJyIIpnP0lZOcpMS\nUTwxcrjsO9vtDWLEG8R8RkY8n8H19VWouo4n99yDxWyS/t3V6HN7wYHDrNlB12Pu4fap0fvw/Rvn\n8e7qHNy8g7ZxAsAnhg9B14FPjd5Xdj5rAqxWBxhhxBfEfDpR5mYTOI7D7x74KP7DhVdtm0YA5TGz\n9ZwAMJ9OwCM4kdfUmqXHzdIdYs6l4BEcFbPNLsGBuwL9uCovI6MU4HU4kSxkbfEpUGzIf3d1DjfW\n13AwOEBvyqDLi39i1jjr8cBACN+79jYGPH58dNcoXr11GdG1WwAMMR8Rh/HLxes4bNmEz8qXjj4G\nHaA39HFzpRKhdLavBo2ZzYnu/Moc/uvF1+HgeDx94FhZnbwSfW5f1eNESzab1EmJJXls9yF4BScm\nQuWP+OU4Ds9+7CRUTUMin8E3z/8YL334Fq4nV1DQ1Lp7XTl4Ab1mQgsoWtZ/sPcevHbrfSQLWez1\n99om6j1+EZMHP1bxfLaY2VH/dn9gYB/mUoma4zzSO4wvHX28rN22NJtNKHaXJahXM7KBUKpROlbM\nZCublJLDgtmQXm1GPyIO4Yq8hCvyIu7r3Y20UkDIb7duQ54Agk4PXbd67wZc2UoMeHrwz44+jkFP\nD43fPjBdv16XD/f17sa9vbsrehGAPcmyFQJODxwcjxvrq4it3cLz70fg5AX8q498siktgm7BARcv\nQLa42SQm3OMX8Y/9H6n6XoHjIQg8Br0B/PP7n8R3omfxa7O2PryBjev6PX4qZr+lV+BkaAzTH75J\ncxwboceSN6nmOlt5aHA/HhrcX/e4Y/2jZa+V1pkJ1lZRMiHeUZZ5MZvEZcveVaWZYStHxGGckWK4\nHF9EyHQvAyXxL8dxOCwO0SaNe3srr0jaCB81LyQpvpAWwD63DxzHVRVyM+E5Dg8O7sMvF6/j29Gz\nAIAvHH64ab2+HMch6PJCzmfp/l+lu2lshBGfiD996DN4b+02LicW8PhIfY+h3+OnjSdWb+wJ870f\nrSCkapAEWDObM6pR2s5JCDo98DlcmE8n6L/tadKkbqVjxUx2u/iHoXHTfa3stgKWuDmxQBfz91SI\nOw+Lg3hz+SZcvIC7mnDTD3sDEMwlch7BUXG97HbyB0cexW8N7MOPZi/iYHCwzGXfKqLLg2vyCnWz\n3fzmbheB4zG+aw8tP9XDmlm3JrAEnseTexvbA9xrLgypl8luBs4qlpnjjBVw1+QVcOa/k3JbM+lg\nMRsz87GBUZogqoY1biblj6CrPDN92JwQDotDtvLBZnHwAkZ8Qcym4mUxeivgOA7H+kcrunzNIOD0\nQINOa9nNzr5WY6DC1sKbhec47O/pa9oGALWo5mYDhodyVV7GfEbGvp5dTVvDbKVjxXwlsQS34KjZ\no2xlfNceXJGX8BOzFFLJMo/6+/CFww/jYJUnS2yGvWb2us/dejFvNyQZRzZa2KxlbpRdFbYW3gr/\n+tin0HzplOOqkgADinEzsD0uNtChdeb1Qg63MzLuDgxsKGkBGL3BPMfRWCtQpbxzfPfBpiWhgGI9\nuXcHipmUyF6fvwIHx9Mml+2G5ByMXoGthy4Cx9ftqGsGI34RDo6vuKmE9bVmbkhgpSPFTFzsRiyo\n6PLit/qLT5ssTYBtFwfMlrzBbYiB2g1pCdWg48m999RMQjYTYpl9DldLRNgsjohD+Hb4czhUIb/T\ndZZZkiS6GQF5EuTMzEzDCy1I8utQlVptNZ6wZEqrWeZmc0QcwpeOPo4nm/Bwtk6DdIEFnG6cDI23\n7HOdvIARn7ihMlanIVRpNrE+gqZ0NV2zaKqYrcL97ne/S58EOT093dB5rspL4MA1nHE+Ig5hxBsE\nB9ga4LcTkoRqdSa7FRwI9GPYG8DnDz7Y8r/vX9z/JL509LGWfuZ2wnEcDgYGjA0Wtykka2pGIxwO\nQ5KMBf/JpJlVDgbpa8DGNicY9gYx4Ompu5i8FI7j8MX7jmM+lbCt5WVsDtHlxekHP9O2z95p/OF9\nn0BB02q2s26FbUtPBgIBaqVDoWIse+LECZw9e7bme79w5OFNf+6ov69uXzKD0Q68Dhe2c4pqqpgj\nkQji8TjC4TA+//nP0/h5cnKymR/DYDAqwOm1Hlq8TZw9exYnTpxo9ccyGDua7sn7MxiMmrSlA0yW\n5bpx8/Xr1xve+K8VdOK4OnFMQGeOqxPHBGxsXL29vTh27Fj1A/QO5bXXXmv3ECrSiePqxDHpemeO\nqxPHpOvNGVdbYuZayLKMaDQKWZYRDoe3vBngVpEkCZIkIRaLYWJiAqIodsz4nnvuOZw+fbqjvrPp\n6WmEQiGIoohQKNT2ccmyDEmSkEgk6LjaPSZyT4XD4bJrB2DT4+u4mHkrzSbbQSwWw/j4OMLhMGZm\nZjpmfLFYjP7cKWOanp6m4xgbG+uIcUWjUUQiEYiiCFmWO2JMtZqrtjK+jhNztWaTdjExMYFgMIhI\nJIKJiYmOGJ8kSQgGgwgEjHbHThgTYEwwsVgMMzMziMViHTGu8fFxSJKEZ599FqFQqCPGRCwwUH7t\ntjK+jlsCWa3ZpJ1EIhGEw2GEQqGOGJ8kSRBFkbprnTAmwLh24XAYiUQCL7zwQkeM6+WXX6ahyPT0\ndEeMyUrpeOLx+KbH13ExM4lPAbQ9/gMMIU9NTSEUClH3sRPGJ8syvvKVr+D06dMA0DFjIo1CY2Nj\nHTEuMh5RNFYqBYPBto9pamoK8XgczzzzDBKJhG08pb83Mr6OEzODwdgcHRczMxiMzcHEzGDsEJiY\nGYwdAhMzg7FD6LjSFGP7IFn5M2fO4Pjx47YuI1mWaQaa0Z0wMd9BkNIaaSUMhUK0MaHdJUDG1mFu\n9h1EJctLGhNmZmYQiUTw3HPPIRaL4Rvf+AZisRhtKYzFYohEIh3RYsuoDBMzgzZUjI+PIxQKYWxs\nDMlk0tYgQxZQhEKhhndbZbQGJuY7DFmWceHCBdoZBRRX8ZD/yEoj8n/A6FEnbaSMzoR1gDEYOwRm\nmRmMHQITM4OxQ2BiZjB2CEzMDMYOgYmZwdghMDEzGDsEJmYGY4fAxMxg7BD+P4hiysmOrd+KAAAA\nAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f23831decf8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_intensity(hawkes_inf_opts.sim_opts_gen(5), 'Hawkes')\n",
    "# plt.savefig(os.path.join(plot_base, 'hawkes_intensity.pdf'), bbox_inches='tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "redqueen/opt_runs.py:174: UserWarning: Consider using `run_dynamic` instead of `run`.\n",
      "  follower_wall_intensities = wall_intensities\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPMAAACWCAYAAAALz77WAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f238310a9e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_intensity(piecewise_inf_opts.sim_opts_gen(5), 'Piecewise constant')\n",
    "# plt.savefig(os.path.join(plot_base, 'piecewise_const_intenstiy.pdf'),  bbox_inches='tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Compare against Mohammed's method"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import broadcast.data as Bdata\n",
    "import broadcast.opt.optimizer as Bopt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'reload' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast-online/redqueen/opt_runs.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mreload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mBopt\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m: name 'reload' is not defined"
     ]
    }
   ],
   "source": [
    "reload(Bopt)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done within 5 iterations!\n"
     ]
    }
   ],
   "source": [
    "world_rate = 1.0\n",
    "num_segments = 10\n",
    "K = 1\n",
    "T = 1000.0\n",
    "\n",
    "rs = np.random.RandomState(42)\n",
    "low_high_ratio = 0.75\n",
    "world_rates = rs.uniform(low=world_rate * (1 - low_high_ratio),\n",
    "                         high=world_rate * (1 + low_high_ratio), \n",
    "                         size=num_segments)\n",
    "\n",
    "# wall_mgr = (SimOpts.std_poisson(world_rate=world_rate, world_seed=0)\n",
    "sim_opts = (SimOpts.std_piecewise_const(world_seed=1, world_rates=world_rates, \n",
    "                                        world_change_times=np.arange(num_segments) / num_segments * T)\n",
    "            .update({ 'end_time': T }))\n",
    "wall_mgr = sim_opts.create_manager_for_wall()\n",
    "wall_mgr.run_dynamic()\n",
    "\n",
    "wall_df = wall_mgr.state.get_dataframe()\n",
    "wall_intensities = wall_df.groupby((wall_df.t.values / T * num_segments).astype(int)).size() / (T / num_segments)\n",
    "# wall_intensities = np.asarray([1000] * num_segments)\n",
    "# wall_intensities = world_rates\n",
    "\n",
    "follower_wall_intensities = np.array([wall_intensities])\n",
    "follower_conn_prob = np.array([[1.0] * num_segments])\n",
    "follower_weights = np.array([1.0])\n",
    "\n",
    "def _util(x):\n",
    "    return Bopt.utils.weighted_top_k(x, follower_wall_intensities, follower_conn_prob, follower_weights, K)\n",
    "\n",
    "def _util_grad(x):\n",
    "    return Bopt.utils.weighted_top_k_grad(x, follower_wall_intensities, follower_conn_prob, follower_weights, K)\n",
    "\n",
    "x0 = np.ones(num_segments)\n",
    "budget = 100.0\n",
    "upper_bounds = np.array([1e6] * num_segments)\n",
    "threshold=0.005 # Relative tolerance for convergence\n",
    "\n",
    "kdd_opt = Bopt.optimize(util=_util, \n",
    "                        util_grad=_util_grad, \n",
    "                        budget=budget, \n",
    "                        upper_bounds=upper_bounds, \n",
    "                        threshold=threshold, \n",
    "                        x0=x0, verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([ 13.006,  11.875,  11.446,  10.409,   7.556,   7.694,   6.336,\n",
       "        10.98 ,  10.131,  10.566])"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "kdd_opt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9.0242120830208741"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Bopt.utils.weighted_top_k(kdd_opt, follower_wall_intensities, follower_conn_prob, follower_weights, K)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Find the `q`which provides a certain capacity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f238342c550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%run -i redqueen/utils.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 96.7 ms, sys: 130 ms, total: 227 ms\n",
      "Wall time: 53.9 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "sim_opts = SimOpts(**{ \n",
    "        'src_id': 1,\n",
    "        'end_time': 100.0,\n",
    "        's': np.asarray([1.0]),\n",
    "        'other_sources': [(Poisson, {'src_id': 2, 'seed': 1, 'rate': 10000.0})],\n",
    "        'q': 1.0,\n",
    "        'sink_ids': [1001],\n",
    "        'edge_list': [(1, 1001), (2, 1001)]\n",
    "    })\n",
    "# tmp = calc_q_capacity_iter(sim_opts)\n",
    "tmp3_poisson_1 = calc_q_capacity_iter(sim_opts, q=sim_opts.q)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7961.1676131321292"
      ]
     },
     "execution_count": 100,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp3_poisson_1.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 50 ms, sys: 33.3 ms, total: 83.3 ms\n",
      "Wall time: 1min 40s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "sim_opts = SimOpts(**{ \n",
    "        'src_id': 1,\n",
    "        'end_time': 100.0,\n",
    "        's': np.asarray([1.0]),\n",
    "        'other_sources': [(Poisson2, {'src_id': 2, 'seed': 1, 'rate': 10000.0})],\n",
    "        'q': 1.0,\n",
    "        'sink_ids': [1001],\n",
    "        'edge_list': [(1, 1001), (2, 1001)]\n",
    "    })\n",
    "# tmp = calc_q_capacity_iter(sim_opts)\n",
    "tmp3_poisson_2 = calc_s_capacity_iter(sim_opts, q=sim_opts.q)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7952.9203030817316"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp3_poisson_2.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 56.7 ms, sys: 33.3 ms, total: 90 ms\n",
      "Wall time: 1min 46s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "tmp2 = calc_s_capacity_iter(sim_opts, q=sim_opts.q, parallel=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%run -i redqueen/utils.py\n",
    "%run -i redqueen/opt_model.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 100 ms, sys: 107 ms, total: 207 ms\n",
      "Wall time: 1min 5s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "tmp4 = calc_s_capacity_iter(sim_opts, q=2.0, parallel=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2110.2250458304434"
      ]
     },
     "execution_count": 104,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp4.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "9469.9358011555705"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tmp4.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sim_opts = SimOpts(**{ \n",
    "        'src_id': 1,\n",
    "        'end_time': 100.0,\n",
    "        's': np.asarray([1.0]),\n",
    "        # 'other_sources': [Poisson2(src_id=2, seed=1, rate=10000.0)],\n",
    "        'other_sources': [(Poisson, {'src_id': 2, 'seed': 1, 'rate': 10000.0})],\n",
    "        'q': 1.0,\n",
    "        'sink_ids': [1001],\n",
    "        'edge_list': [(1, 1001), (2, 1001)]\n",
    "    })"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2016-05-04 21:59:44.864163\u001b[0m Checkpoint: Initial capacity = 769.0321811055888\n",
      "*** \u001b[31m2016-05-04 22:00:06.321337\u001b[0m Checkpoint: s_hi = 4096.0, s_lo = 2048.0\n",
      "*** \u001b[31m2016-05-04 22:00:07.986757\u001b[0m Checkpoint: new_capacity = 106.16027670602416, s = 3072.0\n",
      "*** \u001b[31m2016-05-04 22:00:09.666732\u001b[0m Checkpoint: new_capacity = 101.194622540465, s = 3584.0\n",
      "*** \u001b[31m2016-05-04 22:00:11.336612\u001b[0m Checkpoint: new_capacity = 99.42532014279948, s = 3840.0\n",
      "CPU times: user 210 ms, sys: 2.12 s, total: 2.33 s\n",
      "Wall time: 28.5 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "q_opt = sweep_q(lambda: SimOpts.std_poisson(world_seed=1, world_rate=100.0), capacity_cap=100.0, verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2018-03-29 14:32:24.093781\u001b[0m Checkpoint: q_init = 0.0005753494376300199\n",
      "*** \u001b[31m2018-03-29 14:32:24.288492\u001b[0m Checkpoint: Initial capacity = 40.60789581578206, target capacity = 50.0, q_init = 0.0005753494376300199\n",
      "*** \u001b[31m2018-03-29 14:32:24.475561\u001b[0m Checkpoint: q = 0.0002876747188150099, capacity = 44.185028335018124\n",
      "*** \u001b[31m2018-03-29 14:32:24.660886\u001b[0m Checkpoint: q = 0.00014383735940750496, capacity = 49.87523537642429\n",
      "CPU times: user 73.3 ms, sys: 213 ms, total: 287 ms\n",
      "Wall time: 603 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "q_opt = sweep_q(SimOpts.std_poisson(world_seed=1, world_rate=100.0), capacity_cap=50.0, verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3840.0"
      ]
     },
     "execution_count": 149,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "q_opt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2016-05-02 00:12:20.257822\u001b[0m Checkpoint: Initial capacity = 7967.162987045291\n",
      "*** \u001b[31m2016-05-02 00:44:22.309096\u001b[0m Checkpoint: s_hi = 0.0009765625, s_lo = 0.00048828125\n",
      "*** \u001b[31m2016-05-02 00:46:56.892996\u001b[0m Checkpoint: new_capacity = 47736.13035565209, s = 0.000732421875\n",
      "*** \u001b[31m2016-05-02 00:49:32.194914\u001b[0m Checkpoint: new_capacity = 49913.7414199194, s = 0.0006103515625\n",
      "*** \u001b[31m2016-05-02 00:52:06.251847\u001b[0m Checkpoint: new_capacity = 51218.96161754592, s = 0.00054931640625\n",
      "*** \u001b[31m2016-05-02 00:54:41.662719\u001b[0m Checkpoint: new_capacity = 50540.20193159806, s = 0.000579833984375\n",
      "*** \u001b[31m2016-05-02 00:57:16.650763\u001b[0m Checkpoint: new_capacity = 50224.656253556066, s = 0.0005950927734375\n",
      "CPU times: user 1.96 s, sys: 2.18 s, total: 4.14 s\n",
      "Wall time: 47min 44s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# This is with Poisson: Took ~ 47min, 44s\n",
    "s_opt = sweep_s(sim_opts, 50000.0, verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 233,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2016-05-01 22:57:44.982005\u001b[0m Checkpoint: Initial capacity = 7960.457586400027\n",
      "*** \u001b[31m2016-05-01 23:21:25.507846\u001b[0m Checkpoint: s_hi = 0.0009765625, s_lo = 0.00048828125\n",
      "*** \u001b[31m2016-05-01 23:23:19.044300\u001b[0m Checkpoint: new_capacity = 47686.2063074488, s = 0.000732421875\n",
      "*** \u001b[31m2016-05-01 23:25:12.821979\u001b[0m Checkpoint: new_capacity = 49871.05306762598, s = 0.0006103515625\n",
      "*** \u001b[31m2016-05-01 23:27:08.958602\u001b[0m Checkpoint: new_capacity = 51169.0749459018, s = 0.00054931640625\n",
      "*** \u001b[31m2016-05-01 23:29:03.197845\u001b[0m Checkpoint: new_capacity = 50501.14818392287, s = 0.000579833984375\n",
      "*** \u001b[31m2016-05-01 23:30:57.014591\u001b[0m Checkpoint: new_capacity = 50177.111303157675, s = 0.0005950927734375\n",
      "CPU times: user 1.58 s, sys: 2.26 s, total: 3.84 s\n",
      "Wall time: 35min 24s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "# This was with Poisson2: Took ~ 35min, 24s\n",
    "s_opt = sweep_s(sim_opts, 50000.0, verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.5033950805664062e-06"
      ]
     },
     "execution_count": 135,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s_opt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2.503393943698029e-06"
      ]
     },
     "execution_count": 132,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s_opt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "### Micro-benchmark: fastest poisson"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def foo(T):\n",
    "    rs = np.random.RandomState(42)\n",
    "    t, dt = 0, 0\n",
    "    n = 0\n",
    "    while t + dt < T:\n",
    "        t = t + dt\n",
    "        n += 1\n",
    "        dt = rs.exponential(1 / 1000.0)\n",
    "    return t, n\n",
    "\n",
    "def bar(T):\n",
    "    rs = np.random.RandomState(42)\n",
    "    n = rs.poisson(1000.0 * T)\n",
    "    last_time = sorted(rs.uniform(high=T, size=n))[-1]\n",
    "    return last_time, n\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 loop, best of 3: 394 ms per loop\n"
     ]
    }
   ],
   "source": [
    "%timeit tmp1 = foo(1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 loop, best of 3: 829 ms per loop\n"
     ]
    }
   ],
   "source": [
    "%timeit tmp2 = bar(1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 206,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((999.999833196945, 998932), (999.99831486095445, 999640))"
      ]
     },
     "execution_count": 206,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "foo(1000), bar(1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 245,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mIndexError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[1;32m/home/utkarshu/miniconda3/envs/py35/lib/python3.5/multiprocessing/pool.py\u001b[0m in \u001b[0;36mnext\u001b[1;34m(self, timeout)\u001b[0m\n\u001b[0;32m    679\u001b[0m             \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 680\u001b[1;33m                 \u001b[0mitem\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_items\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpopleft\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    681\u001b[0m             \u001b[1;32mexcept\u001b[0m \u001b[0mIndexError\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mIndexError\u001b[0m: pop from an empty deque",
      "\nDuring handling of the above exception, another exception occurred:\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m/NL/stackexchange/work/prog/broadcast-online/opt_model.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0mget_ipython\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mrun_cell_magic\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'time'\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m''\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'import itertools as I\\n\\ndef worker(param):\\n    return np.sum(range(param * 10000000))\\n\\ndef foo(params):\\n    start, stop = params\\n    total = 0\\n    with mp.Pool() as pool:\\n        for s in pool.imap_unordered(worker, range(start, stop)):\\n            total += s\\n    return start, stop, total\\n\\n# Does not work with \\'Pool\\' class because \"daemonic processes are not allowed to have children\"\\nwith mp.pool.ThreadPool() as pool:\\n    for start, stop, t in pool.imap_unordered(foo, zip(range(1, 10), range(100, 110))):\\n        print(\\'start, stop, t = \\', start, stop, t)\\n        \\n        \\n        '\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;32m/home/utkarshu/miniconda3/envs/py35/lib/python3.5/site-packages/IPython/core/interactiveshell.py\u001b[0m in \u001b[0;36mrun_cell_magic\u001b[1;34m(self, magic_name, line, cell)\u001b[0m\n\u001b[0;32m   2118\u001b[0m             \u001b[0mmagic_arg_s\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mvar_expand\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mline\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mstack_depth\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2119\u001b[0m             \u001b[1;32mwith\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mbuiltin_trap\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2120\u001b[1;33m                 \u001b[0mresult\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mfn\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mmagic_arg_s\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mcell\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   2121\u001b[0m             \u001b[1;32mreturn\u001b[0m \u001b[0mresult\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2122\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<decorator-gen-60>\u001b[0m in \u001b[0;36mtime\u001b[1;34m(self, line, cell, local_ns)\u001b[0m\n",
      "\u001b[1;32m/home/utkarshu/miniconda3/envs/py35/lib/python3.5/site-packages/IPython/core/magic.py\u001b[0m in \u001b[0;36m<lambda>\u001b[1;34m(f, *a, **k)\u001b[0m\n\u001b[0;32m    191\u001b[0m     \u001b[1;31m# but it's overkill for just that one bit of state.\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    192\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mmagic_deco\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0marg\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 193\u001b[1;33m         \u001b[0mcall\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mlambda\u001b[0m \u001b[0mf\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0ma\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mk\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mf\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0ma\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mk\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    194\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    195\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mcallable\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0marg\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m/home/utkarshu/miniconda3/envs/py35/lib/python3.5/site-packages/IPython/core/magics/execution.py\u001b[0m in \u001b[0;36mtime\u001b[1;34m(self, line, cell, local_ns)\u001b[0m\n\u001b[0;32m   1175\u001b[0m         \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1176\u001b[0m             \u001b[0mst\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mclock2\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1177\u001b[1;33m             \u001b[0mexec\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mcode\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mglob\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlocal_ns\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1178\u001b[0m             \u001b[0mend\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mclock2\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1179\u001b[0m             \u001b[0mout\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m/NL/stackexchange/work/prog/broadcast-online/opt_model.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n",
      "\u001b[1;32m/home/utkarshu/miniconda3/envs/py35/lib/python3.5/multiprocessing/pool.py\u001b[0m in \u001b[0;36mnext\u001b[1;34m(self, timeout)\u001b[0m\n\u001b[0;32m    682\u001b[0m                 \u001b[1;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_index\u001b[0m \u001b[1;33m==\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_length\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    683\u001b[0m                     \u001b[1;32mraise\u001b[0m \u001b[0mStopIteration\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 684\u001b[1;33m                 \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_cond\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtimeout\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    685\u001b[0m                 \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    686\u001b[0m                     \u001b[0mitem\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_items\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpopleft\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m/home/utkarshu/miniconda3/envs/py35/lib/python3.5/threading.py\u001b[0m in \u001b[0;36mwait\u001b[1;34m(self, timeout)\u001b[0m\n\u001b[0;32m    291\u001b[0m         \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m    \u001b[1;31m# restore state no matter what (e.g., KeyboardInterrupt)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    292\u001b[0m             \u001b[1;32mif\u001b[0m \u001b[0mtimeout\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 293\u001b[1;33m                 \u001b[0mwaiter\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0macquire\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    294\u001b[0m                 \u001b[0mgotit\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mTrue\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    295\u001b[0m             \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "# %%time\n",
    "# import itertools as I\n",
    "\n",
    "# def worker(param):\n",
    "#     return np.sum(range(param * 10000000))\n",
    "\n",
    "# def foo(params):\n",
    "#     start, stop = params\n",
    "#     total = 0\n",
    "#     with mp.Pool() as pool:\n",
    "#         for s in pool.imap_unordered(worker, range(start, stop)):\n",
    "#             total += s\n",
    "#     return start, stop, total\n",
    "\n",
    "# # Does not work with 'Pool' class because \"daemonic processes are not allowed to have children\"\n",
    "# with mp.pool.ThreadPool() as pool:\n",
    "#     for start, stop, t in pool.imap_unordered(foo, zip(range(1, 10), range(100, 110))):\n",
    "#         print('start, stop, t = ', start, stop, t)\n",
    "        \n",
    "        \n",
    "        "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Multiple followers"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Increase the number of followers and see the improvements wrought about by RedQueen.\n",
    "\n",
    "~~There should be a set of ~ 100 broadcasters and all the followers choose their broadcaster by preferential attachment.~~"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import decorated_options as Deco\n",
    "from opt_runs import make_piecewise_const"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 404,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The slowest run took 4.84 times longer than the fastest. This could mean that an intermediate result is being cached.\n",
      "1000 loops, best of 3: 641 µs per loop\n"
     ]
    }
   ],
   "source": [
    "%timeit make_piecewise_const(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 421,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.9997315530842155"
      ]
     },
     "execution_count": 421,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# tmp_segs = 1000\n",
    "# tmp_make_piecewise_const = make_piecewise_const(tmp_segs)\n",
    "# np.asarray(tmp_make_piecewise_const).sum() * 1/tmp_segs * np.pi"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# plt.plot(tmp_make_piecewise_const, 'o-')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# tmp_make_edge_list = make_edge_list(opts=mk_edge_list_opts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(5037, 1000), (5095, 1000), (5073, 1000), (5059, 1000), (5015, 1000)]"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# tmp_make_edge_list[0:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sim_opts = prepare_multiple_followers_sim_opts(num_followers=10, opts=multiple_follower_opts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(sim_opts.get_dict()['other_sources'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "opt_mgr = sim_opts.create_manager_with_opt(42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 5.64 s, sys: 127 ms, total: 5.76 s\n",
      "Wall time: 5.75 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<__main__.Manager at 0x7f26af3f5630>"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "opt_mgr.run_dynamic()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "opt_df = opt_mgr.get_state().get_dataframe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "wall_df = opt_df[opt_df.src_id != sim_opts.src_id]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "T = sim_opts.end_time\n",
    "num_segments = 24\n",
    "seg_idx = (wall_df.t.values / T * num_segments).astype(int)\n",
    "wall_intensities = wall_df.groupby(seg_idx).size() / (T / num_segments)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "intensity_df = (wall_df.groupby(['sink_id', pd.Series(seg_idx, name='segment')]).size() / (T / num_segments)).sort_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "redqueen/opt_runs.py:2: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  \n"
     ]
    }
   ],
   "source": [
    "world_intensities = \\\n",
    "    intensity_df.reset_index('segment', name='intensity').reset_index('sink_id').pivot_table(\n",
    "        values='intensity', index='sink_id', columns='segment').ix[sim_opts.sink_ids[::-1]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 333,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "791"
      ]
     },
     "execution_count": 333,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "opt_df[opt_df.src_id == 1].t.nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 334,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1.81 s, sys: 50 ms, total: 1.86 s\n",
      "Wall time: 1.86 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "poisson_results = worker_poisson((42, float(opt_df[opt_df.src_id == 1].t.nunique()), \n",
    "                                  sim_opts, None))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "poisson_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 336,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 2.13 s, sys: 23.3 ms, total: 2.16 s\n",
      "Wall time: 2.15 s\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "utils.py:726: VisibleDeprecationWarning: using a non-integer number instead of an integer will result in an error in the future\n",
      "  follower_weights)\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "kdd_results = worker_kdd((42, float(opt_df[opt_df.src_id == 1].t.nunique()), \n",
    "                          24, sim_opts, world_intensities, None))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "kdd_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 338,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 5.09 s, sys: 83.3 ms, total: 5.17 s\n",
      "Wall time: 5.17 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "opt_results = worker_opt((42, sim_opts, 24, None))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "opt_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import decorated_options as Deco\n",
    "import warnings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2018-03-31 11:22:08.199274\u001b[0m Checkpoint: active/total = 3798/1000, procs = [('Opt', 1800), ('Poisson', 999), ('kdd', 999)]\n",
      "*** \u001b[31m2018-03-31 11:23:41.881912\u001b[0m Checkpoint: active/total = 4798/2000, procs = [('Opt', 800), ('Poisson', 1999), ('kdd', 1999)]\n",
      "*** \u001b[31m2018-03-31 11:24:40.622052\u001b[0m Checkpoint: active/total = 5400/3000, procs = [('Opt', 0), ('Poisson', 2698), ('kdd', 2702)]\n",
      "*** \u001b[31m2018-03-31 11:27:05.225257\u001b[0m Checkpoint: active/total = 4400/4000, procs = [('Opt', 0), ('Poisson', 2198), ('kdd', 2202)]\n",
      "*** \u001b[31m2018-03-31 11:29:09.874190\u001b[0m Checkpoint: active/total = 3400/5000, procs = [('Opt', 0), ('Poisson', 1699), ('kdd', 1701)]\n",
      "*** \u001b[31m2018-03-31 11:30:57.175574\u001b[0m Checkpoint: active/total = 2400/6000, procs = [('Opt', 0), ('Poisson', 1197), ('kdd', 1203)]\n",
      "*** \u001b[31m2018-03-31 11:32:39.411699\u001b[0m Checkpoint: active/total = 1400/7000, procs = [('Opt', 0), ('Poisson', 695), ('kdd', 705)]\n",
      "*** \u001b[31m2018-03-31 11:34:29.079916\u001b[0m Checkpoint: active/total = 400/8000, procs = [('Opt', 0), ('Poisson', 186), ('kdd', 214)]\n",
      "CPU times: user 8.88 s, sys: 1.59 s, total: 10.5 s\n",
      "Wall time: 16min 10s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "new_multiple_follower_opts = \\\n",
    "    Deco.Options(repetitions=200, num_followers_list=list(range(1, 15)),\n",
    "                 num_segments=24,  # This matches with the number of segments the rate of users is divided in\n",
    "                 setup_opts=multiple_follower_opts)\n",
    "with warnings.catch_warnings():\n",
    "    warnings.simplefilter('ignore')\n",
    "    run_multiple = run_multiple_followers(opts=new_multiple_follower_opts)\n",
    "    run_multiple_df = run_multiple.df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [],
   "source": [
    "# list(filter(lambda x : x['type'] == 'kdd', run_multiple.raw_results))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>avg_rank</th>\n",
       "      <th>num_events</th>\n",
       "      <th>q</th>\n",
       "      <th>r_2</th>\n",
       "      <th>seed</th>\n",
       "      <th>top_1</th>\n",
       "      <th>world_events</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>type</th>\n",
       "      <th>num_followers</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"14\" valign=\"top\">Opt</th>\n",
       "      <th>1</th>\n",
       "      <td>739.178485</td>\n",
       "      <td>740.450</td>\n",
       "      <td>1.0</td>\n",
       "      <td>9929.478364</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>9.578593</td>\n",
       "      <td>9962.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>758.088181</td>\n",
       "      <td>756.820</td>\n",
       "      <td>4.0</td>\n",
       "      <td>10057.857737</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>9.658950</td>\n",
       "      <td>20048.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>767.843128</td>\n",
       "      <td>770.410</td>\n",
       "      <td>9.0</td>\n",
       "      <td>9974.780727</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>10.931668</td>\n",
       "      <td>30018.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>779.612632</td>\n",
       "      <td>781.875</td>\n",
       "      <td>16.0</td>\n",
       "      <td>9998.569799</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.606586</td>\n",
       "      <td>40150.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>791.228205</td>\n",
       "      <td>791.295</td>\n",
       "      <td>25.0</td>\n",
       "      <td>10031.004842</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>12.368835</td>\n",
       "      <td>50191.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>789.648420</td>\n",
       "      <td>790.660</td>\n",
       "      <td>36.0</td>\n",
       "      <td>10003.279903</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>12.276743</td>\n",
       "      <td>60074.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>791.488058</td>\n",
       "      <td>794.075</td>\n",
       "      <td>49.0</td>\n",
       "      <td>9990.471661</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>12.409365</td>\n",
       "      <td>70103.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>791.784455</td>\n",
       "      <td>793.260</td>\n",
       "      <td>64.0</td>\n",
       "      <td>9993.457307</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>12.427936</td>\n",
       "      <td>80086.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>792.629421</td>\n",
       "      <td>794.165</td>\n",
       "      <td>81.0</td>\n",
       "      <td>9980.121863</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>12.491009</td>\n",
       "      <td>90033.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>791.718174</td>\n",
       "      <td>792.205</td>\n",
       "      <td>100.0</td>\n",
       "      <td>9975.372412</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>12.378370</td>\n",
       "      <td>99949.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>791.995977</td>\n",
       "      <td>791.860</td>\n",
       "      <td>121.0</td>\n",
       "      <td>9979.692281</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>12.392811</td>\n",
       "      <td>109943.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>791.373179</td>\n",
       "      <td>790.870</td>\n",
       "      <td>144.0</td>\n",
       "      <td>9987.138560</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>12.327947</td>\n",
       "      <td>119938.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>792.049248</td>\n",
       "      <td>792.570</td>\n",
       "      <td>169.0</td>\n",
       "      <td>9981.199959</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>12.367304</td>\n",
       "      <td>129952.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>792.685469</td>\n",
       "      <td>791.885</td>\n",
       "      <td>196.0</td>\n",
       "      <td>9999.907473</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>12.355002</td>\n",
       "      <td>140109.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"14\" valign=\"top\">Poisson</th>\n",
       "      <th>1</th>\n",
       "      <td>1345.026473</td>\n",
       "      <td>736.375</td>\n",
       "      <td>1.0</td>\n",
       "      <td>45798.027055</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.171528</td>\n",
       "      <td>9962.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1326.388681</td>\n",
       "      <td>752.315</td>\n",
       "      <td>4.0</td>\n",
       "      <td>43662.814957</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>10.851750</td>\n",
       "      <td>20048.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1302.113664</td>\n",
       "      <td>766.075</td>\n",
       "      <td>9.0</td>\n",
       "      <td>40603.657258</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.020297</td>\n",
       "      <td>30018.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1288.157546</td>\n",
       "      <td>777.725</td>\n",
       "      <td>16.0</td>\n",
       "      <td>37538.261627</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.064927</td>\n",
       "      <td>40150.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1273.959990</td>\n",
       "      <td>787.125</td>\n",
       "      <td>25.0</td>\n",
       "      <td>33990.291850</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.188775</td>\n",
       "      <td>50191.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1271.609458</td>\n",
       "      <td>786.500</td>\n",
       "      <td>36.0</td>\n",
       "      <td>34124.939228</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.158069</td>\n",
       "      <td>60074.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1265.647739</td>\n",
       "      <td>789.905</td>\n",
       "      <td>49.0</td>\n",
       "      <td>33126.614017</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.178580</td>\n",
       "      <td>70103.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1267.239716</td>\n",
       "      <td>788.675</td>\n",
       "      <td>64.0</td>\n",
       "      <td>33344.388504</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.177068</td>\n",
       "      <td>80086.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1263.871247</td>\n",
       "      <td>790.025</td>\n",
       "      <td>81.0</td>\n",
       "      <td>32870.856437</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.210908</td>\n",
       "      <td>90033.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1265.641916</td>\n",
       "      <td>788.065</td>\n",
       "      <td>100.0</td>\n",
       "      <td>33249.746327</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.161414</td>\n",
       "      <td>99949.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1265.630654</td>\n",
       "      <td>787.480</td>\n",
       "      <td>121.0</td>\n",
       "      <td>33217.620402</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.179894</td>\n",
       "      <td>109943.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1267.182280</td>\n",
       "      <td>786.725</td>\n",
       "      <td>144.0</td>\n",
       "      <td>33584.733011</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.194257</td>\n",
       "      <td>119938.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>1265.304407</td>\n",
       "      <td>787.990</td>\n",
       "      <td>169.0</td>\n",
       "      <td>33261.640583</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.188989</td>\n",
       "      <td>129952.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>1267.933040</td>\n",
       "      <td>787.280</td>\n",
       "      <td>196.0</td>\n",
       "      <td>33505.811196</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.182357</td>\n",
       "      <td>140109.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"14\" valign=\"top\">kdd</th>\n",
       "      <th>1</th>\n",
       "      <td>1713.980032</td>\n",
       "      <td>737.215</td>\n",
       "      <td>1.0</td>\n",
       "      <td>94956.797307</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.774446</td>\n",
       "      <td>9962.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1693.758075</td>\n",
       "      <td>753.895</td>\n",
       "      <td>4.0</td>\n",
       "      <td>92692.473752</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.575802</td>\n",
       "      <td>20048.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1586.010842</td>\n",
       "      <td>767.060</td>\n",
       "      <td>9.0</td>\n",
       "      <td>75649.157586</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.843309</td>\n",
       "      <td>30018.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1445.380461</td>\n",
       "      <td>778.785</td>\n",
       "      <td>16.0</td>\n",
       "      <td>55062.565491</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.671590</td>\n",
       "      <td>40150.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>1302.912902</td>\n",
       "      <td>786.265</td>\n",
       "      <td>25.0</td>\n",
       "      <td>36691.824778</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.331690</td>\n",
       "      <td>50191.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>1312.500955</td>\n",
       "      <td>786.875</td>\n",
       "      <td>36.0</td>\n",
       "      <td>37962.514059</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.350671</td>\n",
       "      <td>60074.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>1292.959475</td>\n",
       "      <td>789.720</td>\n",
       "      <td>49.0</td>\n",
       "      <td>35586.180911</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.295018</td>\n",
       "      <td>70103.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>1298.751045</td>\n",
       "      <td>789.805</td>\n",
       "      <td>64.0</td>\n",
       "      <td>36541.692962</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.337545</td>\n",
       "      <td>80086.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>1284.186760</td>\n",
       "      <td>790.085</td>\n",
       "      <td>81.0</td>\n",
       "      <td>34753.283282</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.330288</td>\n",
       "      <td>90033.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>1298.987563</td>\n",
       "      <td>787.795</td>\n",
       "      <td>100.0</td>\n",
       "      <td>36304.993468</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.311214</td>\n",
       "      <td>99949.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>1306.968520</td>\n",
       "      <td>787.845</td>\n",
       "      <td>121.0</td>\n",
       "      <td>37142.450596</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.347942</td>\n",
       "      <td>109943.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>1314.007920</td>\n",
       "      <td>786.835</td>\n",
       "      <td>144.0</td>\n",
       "      <td>38126.408711</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.395800</td>\n",
       "      <td>119938.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>1308.146478</td>\n",
       "      <td>789.280</td>\n",
       "      <td>169.0</td>\n",
       "      <td>37467.461600</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.371786</td>\n",
       "      <td>129952.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>1306.878396</td>\n",
       "      <td>787.910</td>\n",
       "      <td>196.0</td>\n",
       "      <td>37371.122971</td>\n",
       "      <td>4234.0</td>\n",
       "      <td>11.381640</td>\n",
       "      <td>140109.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                          avg_rank  num_events      q           r_2    seed  \\\n",
       "type    num_followers                                                         \n",
       "Opt     1               739.178485     740.450    1.0   9929.478364  4234.0   \n",
       "        2               758.088181     756.820    4.0  10057.857737  4234.0   \n",
       "        3               767.843128     770.410    9.0   9974.780727  4234.0   \n",
       "        4               779.612632     781.875   16.0   9998.569799  4234.0   \n",
       "        5               791.228205     791.295   25.0  10031.004842  4234.0   \n",
       "        6               789.648420     790.660   36.0  10003.279903  4234.0   \n",
       "        7               791.488058     794.075   49.0   9990.471661  4234.0   \n",
       "        8               791.784455     793.260   64.0   9993.457307  4234.0   \n",
       "        9               792.629421     794.165   81.0   9980.121863  4234.0   \n",
       "        10              791.718174     792.205  100.0   9975.372412  4234.0   \n",
       "        11              791.995977     791.860  121.0   9979.692281  4234.0   \n",
       "        12              791.373179     790.870  144.0   9987.138560  4234.0   \n",
       "        13              792.049248     792.570  169.0   9981.199959  4234.0   \n",
       "        14              792.685469     791.885  196.0   9999.907473  4234.0   \n",
       "Poisson 1              1345.026473     736.375    1.0  45798.027055  4234.0   \n",
       "        2              1326.388681     752.315    4.0  43662.814957  4234.0   \n",
       "        3              1302.113664     766.075    9.0  40603.657258  4234.0   \n",
       "        4              1288.157546     777.725   16.0  37538.261627  4234.0   \n",
       "        5              1273.959990     787.125   25.0  33990.291850  4234.0   \n",
       "        6              1271.609458     786.500   36.0  34124.939228  4234.0   \n",
       "        7              1265.647739     789.905   49.0  33126.614017  4234.0   \n",
       "        8              1267.239716     788.675   64.0  33344.388504  4234.0   \n",
       "        9              1263.871247     790.025   81.0  32870.856437  4234.0   \n",
       "        10             1265.641916     788.065  100.0  33249.746327  4234.0   \n",
       "        11             1265.630654     787.480  121.0  33217.620402  4234.0   \n",
       "        12             1267.182280     786.725  144.0  33584.733011  4234.0   \n",
       "        13             1265.304407     787.990  169.0  33261.640583  4234.0   \n",
       "        14             1267.933040     787.280  196.0  33505.811196  4234.0   \n",
       "kdd     1              1713.980032     737.215    1.0  94956.797307  4234.0   \n",
       "        2              1693.758075     753.895    4.0  92692.473752  4234.0   \n",
       "        3              1586.010842     767.060    9.0  75649.157586  4234.0   \n",
       "        4              1445.380461     778.785   16.0  55062.565491  4234.0   \n",
       "        5              1302.912902     786.265   25.0  36691.824778  4234.0   \n",
       "        6              1312.500955     786.875   36.0  37962.514059  4234.0   \n",
       "        7              1292.959475     789.720   49.0  35586.180911  4234.0   \n",
       "        8              1298.751045     789.805   64.0  36541.692962  4234.0   \n",
       "        9              1284.186760     790.085   81.0  34753.283282  4234.0   \n",
       "        10             1298.987563     787.795  100.0  36304.993468  4234.0   \n",
       "        11             1306.968520     787.845  121.0  37142.450596  4234.0   \n",
       "        12             1314.007920     786.835  144.0  38126.408711  4234.0   \n",
       "        13             1308.146478     789.280  169.0  37467.461600  4234.0   \n",
       "        14             1306.878396     787.910  196.0  37371.122971  4234.0   \n",
       "\n",
       "                           top_1  world_events  \n",
       "type    num_followers                           \n",
       "Opt     1               9.578593        9962.0  \n",
       "        2               9.658950       20048.0  \n",
       "        3              10.931668       30018.0  \n",
       "        4              11.606586       40150.0  \n",
       "        5              12.368835       50191.0  \n",
       "        6              12.276743       60074.0  \n",
       "        7              12.409365       70103.0  \n",
       "        8              12.427936       80086.0  \n",
       "        9              12.491009       90033.0  \n",
       "        10             12.378370       99949.0  \n",
       "        11             12.392811      109943.0  \n",
       "        12             12.327947      119938.0  \n",
       "        13             12.367304      129952.0  \n",
       "        14             12.355002      140109.0  \n",
       "Poisson 1              11.171528        9962.0  \n",
       "        2              10.851750       20048.0  \n",
       "        3              11.020297       30018.0  \n",
       "        4              11.064927       40150.0  \n",
       "        5              11.188775       50191.0  \n",
       "        6              11.158069       60074.0  \n",
       "        7              11.178580       70103.0  \n",
       "        8              11.177068       80086.0  \n",
       "        9              11.210908       90033.0  \n",
       "        10             11.161414       99949.0  \n",
       "        11             11.179894      109943.0  \n",
       "        12             11.194257      119938.0  \n",
       "        13             11.188989      129952.0  \n",
       "        14             11.182357      140109.0  \n",
       "kdd     1              11.774446        9962.0  \n",
       "        2              11.575802       20048.0  \n",
       "        3              11.843309       30018.0  \n",
       "        4              11.671590       40150.0  \n",
       "        5              11.331690       50191.0  \n",
       "        6              11.350671       60074.0  \n",
       "        7              11.295018       70103.0  \n",
       "        8              11.337545       80086.0  \n",
       "        9              11.330288       90033.0  \n",
       "        10             11.311214       99949.0  \n",
       "        11             11.347942      109943.0  \n",
       "        12             11.395800      119938.0  \n",
       "        13             11.371786      129952.0  \n",
       "        14             11.381640      140109.0  "
      ]
     },
     "execution_count": 105,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "run_multiple_df.groupby(['type', 'num_followers']).mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def plot_multiple_followers_perf(run_multiple_df, measure='avg_rank', \n",
    "                                 yticks=None,\n",
    "                                 file_name=None):\n",
    "    run_multiple_df = run_multiple_df.sort_values(['type', 'num_followers'])\n",
    "    perf_data_grp = run_multiple_df.groupby(['type', 'num_followers'])\n",
    "    perf_mean = perf_data_grp.mean()\n",
    "    perf_std = perf_data_grp.std()\n",
    "    \n",
    "    # plt.figure(figsize=(8,6))\n",
    "    \n",
    "    if file_name is None:\n",
    "        latexify(fig_width=8, largeFonts=True)\n",
    "    else:\n",
    "        latexify(fig_width=1.75, largeFonts=False)\n",
    "    \n",
    "    num_followers = np.asarray(sorted(run_multiple_df.num_followers.unique().tolist()))\n",
    "    c1, c2, c3 = sns.color_palette('Set1', n_colors=3)\n",
    "    \n",
    "    plt.errorbar(num_followers + 0.0, perf_mean.loc['Opt'][measure].tolist(), fmt=red_queen_fmt,\n",
    "                 yerr=perf_std.loc['Opt'][measure].tolist(), label=r'\\textsc{RedQueen}',\n",
    "                 linewidth=0.75, markersize=3, color=red_queen_color)\n",
    "    \n",
    "    #plt.errorbar(num_followers - 0.1, perf_mean.loc['Poisson'][measure].tolist(), fmt=poisson_fmt,\n",
    "    #             yerr=perf_std.loc['Poisson'][measure].tolist(), label='Poisson',\n",
    "    #             linewidth=0.75, markersize=3, color=poisson_color)\n",
    "    \n",
    "    plt.errorbar(num_followers - 0.2, perf_mean.loc['kdd'][measure].tolist(), fmt=karimi_fmt,\n",
    "                 yerr=perf_std.loc['kdd'][measure].tolist(), label='Karimi',\n",
    "                 linewidth=0.75, markersize=3, color=karimi_color)\n",
    "    \n",
    "    if file_name is None:\n",
    "        plt.xlabel('Followers')\n",
    "\n",
    "        ylabel = 'Unknown'\n",
    "\n",
    "        if measure == 'avg_rank':\n",
    "            ylabel = 'Average position'\n",
    "            plt.legend(loc='upper right')\n",
    "        elif measure == 'top_1':\n",
    "            ylabel = r'$\\int I(r(t) < 1)\\,dt$'\n",
    "            plt.legend(loc='lower right')\n",
    "        else:\n",
    "            plt.legend()\n",
    "\n",
    "        plt.ylabel(ylabel)\n",
    "        \n",
    "    if yticks is not None:\n",
    "        plt.yticks(yticks)\n",
    "    \n",
    "    plt.xlim((0, num_followers.max() + 0.1))\n",
    "    plt.tight_layout()\n",
    "    format_axes(plt.gca())\n",
    "    \n",
    "    if file_name is not None:\n",
    "        plt.savefig(os.path.join(plot_base, file_name), bbox_inches='tight', pad_inches=0)\n",
    "    \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "redqueen/utils.py:20: DeprecationWarning: \n",
      ".ix is deprecated. Please use\n",
      ".loc for label based indexing or\n",
      ".iloc for positional indexing\n",
      "\n",
      "See the documentation here:\n",
      "http://pandas.pydata.org/pandas-docs/stable/indexing.html#ix-indexer-is-deprecated\n",
      "  \n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjgAAAFcCAYAAADF+MQnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAIABJREFUeJzs3XlwW+d9N/rvc7ASK1eJkghqlyySknfHgmxHcbwwzlo3\nNhXn9m39Nq7z3ttr9TZ2Z3pv606dzLwzr+z3nfTOnRtfpW+Spk1IO0ubNjJlO4mSSJBteZUISdYu\ngpJIihKJjcR6nvvHAUBSFDcRwAHA72eGA+DgAOdHEDz44jnPeR4hpZQgIiIiqiCK3gUQERER5RsD\nDhEREVUcBhwiIiKqOAw4REREVHEKHnD8fv+k2z6fDz6fD7t27Sr0pomIiGiRKmjA8fl82Llz56Tb\n3d3d8Hq9OHr06JTwQ0RERJQPxkI+udfrhcfjmXTb6/UCAAKBAFpbWwu5eSIiIlqkdOmDs3v3brzw\nwgt6bJqIiIgWAV0CzlNPPYXOzk6EQiE9Nk9EREQVrqgBx+/35/rdeDwedHV15e7bt28fOjs7i1kO\nERERVaiiBhyfz4dgMAgACIfDk/rnbN++HY2NjcUsh4iIiCpUQQNOd3c3enp60N3dDQDo6OhAIBDI\ntdy0t7cXcvNERES0SIlSmmxz37592L59u95lEBERUZnjSMZERERUcRhwiIiIqOIw4BAREVHFYcAh\nIiKiisOAQ0RERBWHAYeIiIgqDgMOERERVRwGHCIiIqo4DDhERERUcRhwiIiIqOIw4BAREVHFYcAh\nIiKiisOAQ0RERBWHAYeIiIgqDgMOERERVRwGHCIiIqo4DDhERERUcRhwiIjKnG/gjN4lEJUcBhwi\nojJ3kAGHaAoGHCIiIqo4DDhERPPAw0FE5YEBh4hoHng4iKg8MOAQUUliSwkRLQQDDhGVJLaUENFC\nMOAQFRlbJoiICo8Bh6jI2DJBi0EpBvlSrIkKhwGHiIjyrhSDfCnWRIXDgENERESTVEJrFwMOERER\nTVIJrV0MOERERFTy5tuqxIBDRESko0o4HFQM821VYsAhIu5giXRUCYeDShEDDhFxB0tEFceodwFE\nRDR3qlQxkhjDUCyKK7EIvn/iLQDA07//Ef7fe3ZAEfzeSgQw4BARlRRVSoSyASYewZVYFEOxKK7G\no0iqaShCoNpchTqrA/VW+6THvnT4V7AYjFjtrMNaVwNWO+tQZTTr9JsQ6YsBh4ioiKSUCCVjuJJp\ngRmKR3PXE2oaAgJusxX1VgfqrHasdNbi9vpm1FrtMCmGKc/3w5Pv5K4/d/ODGEslcTY8hNOhIfzq\nwnHE0ikss7mx1lWPNa56LLE6IYQo5q9MpAsGHCKiPJJSIpKM40o8iqGY1gKTvR5PpyAg4DJbUWe1\no85ih8deg1vqmlBnscNsmP8u+eV7n8BLh9/EN7Y8AACoMprQUrMMLTXLAGiHtC6NhnA6NITXev0Y\njEVgN5ozgacBqxy1N7RdolLHdzUR0TxIKRG9JsBkDyeNpZIQEHCYLKiz2lFvtWOZzY222uWos9ph\nNZiKXq8iFKywV2OFvRr3LVsHAAgnYjgTHoL/6kXs6e1BSlXhcVRjjasBa131qLXYZ3lWotLHgENE\nFUeVEkk1jUQ6hYSaRlLVLhPpNBJqatJ92v3a7aSq3X/tukk1DSmBk6FBAMBfvvVTfHr5RtRZ7Wio\ncmJTTSPqLPay6e/iNFtxc10Tbq5rAgCkVRWB6DDOhIbws7Mf4mo8imqzDWtc9VjrrIfHUQPjdQ6P\nEZUyBhwAoz0DsLUt1bsMIgKQlir+1/2dALQzg9o9LUim05kwkspc10LIdAQETIoBZoMBZsUAs2LM\n3M5cZpaZDQY4TZYJ92eWKwaYJlw3KgYoQuDp3/8ot43H195e8NeiWAyKglXOOqxy1uF+bAQADMdH\ncSY0hEND5/Gzcx9CQGClsxZrnVpfHpe5SueqiWbGgANgjAGHSDdSSgyMhXBspB/HRgYQTcYn3b+l\ndkUmlGhhw2wwwJQJJAo7yxZMjcWG2xuacXtDMwAgkU6hN3IVp8NDODh4FuFkDA1WB9a4GrDGWY8V\ndndJnKKuShVJVUVKTSMlVSTVNFJqGn/33i8BaKH55Xuf0LlKKgYGHCIqumBiDMdG+nF8uB8DY2E0\n2lzYVN2Ir667E25z1aSWkrWuBh0rnezaDr2LidlgxDr3EqxzLwGgBdOhWASnQ0P4ff8pXIiOwGww\nYrWjDr8M9ADQwsRf3/IwUmpaCx0ynQsfyUwAyd2nppGUaaRU9Zr1x5elpDqpJim1y4k5V2u9U2BU\nDDApBhiFMuXssxc/ehMSEi6TNXe6fb3VkTtz7Xpnq1H5YcAhKqLsB7c2KNtXFk0LRCyVxIngII6N\n9ON85CqcJgs2VTfiM55WLKmaetryYg4S5UIIgYYqJxqqnLh76WoAwFgqiXPhK0BgfL1Dl8/DqCgw\nCUMmdGjhwygUVBlMMJqs2jJhmBxMrvMYg1Bu+H/md/2nctefvfkBSCkRTsa1U/VjEfRGhvHBlQCu\nxMbHG6q12DIBKPtjR7W5qiRaqvJJSom4mkIkGUckGcd//XAvAG0/9Y0tD8AoFBgVBUahwJD5uxgy\nt7N/S0WIkht+gAGHqEgS6dSk2y8dfhOKEFhmc6PJXg2PvQbL7dWwVMApu2lVxdnwEI6N9ONk8DIM\nQmC9eynuWrISX15zKwwV9gFBmiqjCZtqGicte2zNbTpVM9m1oVkI7XR9l9mK1a76KeunpYrh+Gjm\nLLkIjg5fwpVYBCOJMahSwqwYMi0+k1uA7Eaz7h/0STWNaDKOSCqeCy3h5Pj1aGZ5Sh1vEbMYjHCY\nLHCaLJOe69jwJaSkirSqZlrcVKSzLW2529rP9VyvlS0blIzCMCk4TbxtVBQYJtzuOvMegPkdYiz/\nPSlRmfjF+SOTbj9384NIqWlcGg2hLzqMd4d6ceH8YSTUFKrNNjTZa+BxVKPJXoNqc5XuO82ZSClx\ncTSoHXYa6UcslcJqVx02VTfi4aYWjrOyyFRCC5xBKLnQshFT+2gm0qnc8ACXxyI4Ex7ClVgUkUwf\nstxQAZbJLUDz/V9QpcRoKoFIMpYLKNngMjG0xNMpCKEFCqOiwGGywG605EKL02TBMpsLDpMFDqMF\ndpNl2kNxb1w4nrv+xVU3z6veufw+6VxgmhqUUhOWpzN9qKYLT7PhXoeoCM6GhzAYC+M793wF//3I\nr3I7fqNigMdRA4+jJreulBIjiTH0RYfRFx3BwYGzGEmMwawYsNzuhsdegyZ7DZbZXLqeujscH9U6\nBg/343IsjBX2amyqbsQfr78bTrNVt7qIisFsMGK53Y3ldveU+6SUiKYSGMoc/ro4GsSRqxcwFI8i\nkRnsscaiTbfxWsAPQGuZuKdxLSLJOEZTiVzLhyIEbEazFkyyP0YLmuw1k5ZZFGPevgQVMqAqQkAR\nhky4Kuy4UAw4RAWWVNN49cz7eOqme+a0AxJCoMZiQ43Fhs21K3LLE+kULoyOIBAZwe/7T6F/NIS0\nVFFvtaMpE3qa7NUFCxejqQROjAzg2Eg/AtFhuM1V2FTdiC+s3Ix6q6OkW5iIikkIkQseq5x1U+6f\nOGFqNuAAwEMrNsFhsqDKaF40/fPm6kZCFwMOUYG91uuHd+ka1FhsC3oes8GI1c56rHaO9xdQpcSV\nWASB6AhOhgbxm0snEEnGUWUwoclejSZHDTz2aiypcs67Y2RSTeNMSOtHczp0GSbFgI3updjWuBZN\n9hrugIlukCIU1FrsU0aMXmpz6VRRZWLAISqgQGQY5yJX8PmVmwvy/MqEM1luq/fklo+mEuiLjqAv\nMozXh49jYCwEAYGlNqfWt8dejRX2GlQZx5uIVSlxITqCYyP9+HikH/F0Gmtd9bipuhGfbW7jqbNE\nBVAJ/ZVKFQMOUYGkVRWdp9/Fkxu3Fv3wjc1oxgb3EmzIjFkCaGeFDIyG0RcdxuGrF7En4EcsncLp\n0GUAwH/Z/2Pc17gOm2oasW2jF/ZrzqYgIionDDg0Z5zSYn5ev3AMt9c3o97q0LsUANpZIdlOkXdN\nWD5xUL2vrr9r6gOJiMoQB6OgORvrGdC7hLJxaTSI4yP92L58g96l0CKwdekavUsgKjkMOER5pkoV\nPz71Lr6y9s6y6Ij78r1PYIN7CefnKWNeBhyiKRhwiPLs1xdPoKVmGRp5RgQRkW4YcIjy6PJYGB8O\n9eHBppv0LoWIaFFjwCHKE1VK/Oj0u/jKujs41xIRkc64FybKk/39p7DGWYcV9mq9SyEiWvQYcIjy\n4Go8ircGz+Eznla9S6EC4xlL5Yt/u8WFAYdogaSU+NGpQ3h8zW26Tn5ZaUr1w4hnLJUv/u0WFwYc\nogV6a/Asltnc151Uj24cP4zKW6kGVJqbUvz7zbcmBhyiBQgmxvC7S6fw+ebCzDVFVK4YUMtbKf79\n5lsTAw7RAnSefhd/uPpWmA2c9YSIqJQw4BDdoPcu98JtrsI6d8O8HleKTb9EpB/uEwqDAYfoBkSS\nMbxx4Ri+tOrmeT+2FJt+iUg/3CcUBgMO0Q145cz7+OLKm2E1mPQuhYiIroMBp0SNcubuknX4ygWY\nFAM21TTqXQoREU2j4AHH7/dPut3V1YWuri7s2rWr0Jsua2MMOHNS7CA4mkrgl4EefHn1rUXdLhER\nzU9BA47P58POnTsn3fZ6vejo6EAgEIDP5yvk5mkRKHYQ/OnZD/BZTxuqjOaibpeIiOanoAHH6/XC\n4/Hkbk8MNR6PB4FAoJCbJ8qrY8P9SKppbKlboXcpREQ0i6IO3tHR0ZG7fvToUTzyyCPF3DzRDYul\nk/i38x/hz1u3611KQfA0VSKqNLp0Mvb7/WhpaUFr6/jEhPv27cO5c+f0KIdoVv967iM8uGITHCaL\n3qUUBE9TJaJKo0vA8fl8eO655yYt2759O1atWqVHOUQzOhUcRDAxhtvqPbOvTEREJaHoAaerqwtP\nPfUUALCTMZW8RDqFn579EDvW3gEhhN7lEBHRHBU04HR3d6Onpwfd3d0AtEDz4osv4oEHHsCdd95Z\nyE0T5cV/9PbgvmXr4DZX6V0KERHNQ0E7Gbe3t6O9vT132+v14tChQ4XcJFHenAtfwaXRIP7gBqZj\nICIifXEkY6LrSKlpvHLmfXxlHQ9NERGVIwYcout4LXAUdy9ZhVqLXe9SiIjoBjDgEF2jLzqMM+Eh\n3NO4Tu9SiIjoBjHgEE2Qlio6T72HJ9beAYWHpoiIyhYDDlU038CZea3/Zt9x3FLfhIYqZ4EqIiKi\nYmDAoYp2cB4Bp380BP/wJdy/fEMBKyIiomJgwKE5UWMppIbH9C6jYFQp8ePTh/DEujugCP5bEBGV\nu6JOtknlRybTiL5/EWP+QchIEkM//BC2W5bBur4OirVy3j77Lp7ATdWNaLS59S6FiIjygF9V6bqk\nKjH6UT+udh2BUmWCUmUCAAhFACkVI784juGfH8XYsctQE2mdq12YoVgE7w314qEVm/QuhYiI8qRy\nvoJTXkgpET9xBdF3L8C6vg61HZshTAbYtjTiSudh1O3YAgCw3boM6VAcsY+HMPwzPww2M6w31cOy\nphbCWD65WUqJH586hB1r74BBKZ+6iYhoZgw4lBPvHUHUF4BpuRM1f9g66yEog8sC+50rYL9zBVLD\nY4gdH0L03R4Yqq2o2tQAc7MbwlDaoeHAwGmsdNTB46jRuxQiIsqjRR9wpCqRDsb0LkNXyYEIIvvP\nQ3GY4f7sBhiclnk/h7GmCo6tHji2epC8HEXs4yFEDvTCuMQO6031MDe5tcNbJWQ4PgrfwBn8H5s/\nrXcpRESUZ4s+4Iz8+3GooQTCvvNwfMJT8i0O+ZQaiSGy/zxkWoVz+2oY62x5eV5Tgx2mBjvktmak\nBqIYO34Zkd+dh2m5E9ab6mFa5tR9ficpJX58+l08tuY2mBSDrrUQEVH+LfqAo44lAQBjRwaRODMC\nky0M2/13wFht1bmywklHE4geDCA1PAbHtpUwLy/MoHZCCJgaHTA1OiClRPJCGLGjlxH+9VmYPW5Y\nb6qHcYldl7DzzuVzWGJ1YrWzvujbJiKiwlv0Aadux5Zc51mpSsR+uBvhfXWQKRW2zUthWVdbMa06\naiKF6KELSJwPwrHVA+eq6qKFCyEEzE0umJtckKpEIhDE6IeXkBoahXlVNapuashbC9JsQokx/Obi\nCfwlD00RURGM9gzA1rZU7zIWnUUfcCYSioC5agRVX9qEdCSBsZ4BRH98BGaPC1VbGmGsqdK7xBsi\nUypGP+pH7Nhl2G5bBoe3WddDREIRsKyshmVlNWRaRfzcCCIHA0iH4rCsrYF1Y0NBW9A6T7+HR1ff\nCrOBb38iKrwxBhxdcA8/DYPDDMfdHtg/0YTEuRGEf3cOMpFG1ealsK6rK4tToaWUiB27jNH3L8Ha\n0oDar2wuudYoYVBgXVsL69payGQa8TPDCP/2LORYCpZ1tbDeVA+DY/6dnqfz/lAvnCYLNriX5O05\nr8Vva0RU6kp1P5XPuhhwZiGEgGV1DSyra5COJjDWM4irnUdganLBtqURxtrSa9WRUiJxdhiRt/pg\nWVmNmsdboZhL/08tTAZYN9bDurEeajyF+KmrCL1+GjKlwrqhDtYN9VBspht+/mgyjtf7juEv2u7P\nY9VTleq3tVLdoRFR8ZXqfiqfdZX+p14JMdjNcHyiCfa7ViDRG0Rk/3mosRSq2pbAuqG+JFp1EhfD\niBw4D2NtFaq/eBMMdrPeJd0QxWJEVesSVLUugTqWROzkFYz88mMIRcC6sR6W9XVQLPN7+75y5n18\nYeUWWI03HpLKWanu0IjyiUGeshhwboAQ431I0tEEYv5Mq85yJ2w3Nxats+xEqSujCO8/D2FU4Hpw\nXUWdBaZUmWDb0gjblkakIwnETgxh5F+PQViMSA2Nzuk5jly9AIMQaKlZVuBqqdzxA7K8MchTlv5N\nDmXOYDfDflcTar+6BZZ1tYgcDODqKz3a5JTJws/RlA7HEdx7EuHfnoNjqwfVn91YUeHmWgaHGfbb\nlqO2YzOsG+sgx1K4+moP1Hhq2seMpZL4ZW8PvrzmtiJWSuVqrGdA7xKmGC3BmoDSrYsIYAtO3ggh\nYGmuhqW5GupoEmP+QVx9pQemRgeqtjTC1GDP6/bUWArRt/uQvBSG3euBpbk6r89fDkY/6gcAqJEE\nhl/1w3HPSlhWTX0dfnb2A3zG0wqbsTwP1xGVaqtEqdZFBDDgFIRiM8F+5wrY7liOZF8I0bf7oEYT\nqGpdAuvGegjTjY+cK5NpRN+/hPipK7DfuQKO+1bqPiqwXiaOYaSOJRH61RnET12B475VUMzaazya\nSiCWTuLmuiadqyUiomLiIaoCEkLA7HGj+nMbUf2Fm6DGU7j6Sg9Cb55GcjA6r+eSqsTo4X5c7eqB\nUmVE7Ve2aB2bF2m4uZZSZUL15zbC7HFj+JUeJPpCiKdTGIpF0LH2dr3Lo2nwEAcRFQpbcIpEqTLB\nfvsK2G5bjuSFMKLvXoAaisPa2gDrxoZci8O1pJSIn7yC6KELsKyrQ21H24JagCqddWM9TE0uhN44\njV9FziLlSeEbb/0ML9/7hN6l0XXwEAcRFQoDTpFNnLJAjaUwdnQQw6/2wLjEDtuWRpiWOnLrJgJB\nRHy9MDU6UPOHrVCs/HPNhcFuRuAeJwZ/l8QfH12KPauv6l0SEREVGT8xdaRYjbDfthy2W5cheTGM\n0fcvIjUSg3VjPZIXwxg7Ogj3IxtgcOZvJN/FYDg+ij19fpxtiOKsK4bPn61F5EAv7Hc3ldxIzkRE\nVBgMOCVACAHzChfMK7RWnchbvUBawrqpgeFmntJSxQ9OvIUn1t2JZkctXjr8Ju799Kcx9mE/rr7S\nA/eD62CsL/44RUREVFz8OltiFKsRyf4IACDi69W5mvLzy/M92FK7As2O2twyIQRsty6Du309Qr86\njeg7fZCqLFpNqn9/0bZFiwPfU0SzY8ApQXU7tsDkcaFuxxa9Sykrx4b7cXE0iE8t33Dd+401Vah5\nrA0AMPwTP1LDY0WpS/p9RdkOLR58TxHNjgGHKkIoMYZ/O/8R/mj9XTOeOi8UAftdTXDevxrB7pMY\n/eASpCxeaw6VH7aWUL6V4nuqFGtaKAYcKnuqlPink2/jsTW3wW6aW58lU70dtY+3IT2awMjPjiId\nihe4ytJSqjuzUqyLrSXlje+puSnFmhb6t2PAobL3et9RrHM1YK2rYV6PEwYFzm0r4di2EiP/fhxj\nPQOLpjWnFHdmQOnWRbMrxSAB8D1Vzhb6t2PAobJ23jyKU6HLeKip5Yafw9ToQG3HZiSHRjHyb8eR\njibyWCHR4sAgQaWGAYfKVjQZxx73Jfyn9XdDWeCUFcKowLV9Nex3LMfIz48h9vFQnqokIiI9MODQ\nDdOzSVpKiR+efAcPhpbCZbbmrSZzkxs1j7ch0RvEyH98DDWWWmipRESkAwYcumF6Nkn/5uIJLLe5\nsS7umLQ8HzUpZgNcD65FVesSDP/Ej/gZTvVARFRuGHCo7PRGruLw1Qv47Mq2gm7HsroGNV9uxdjx\nIQT3noSaYGsOEVG5YMChshJLJfGjU4fwxxvuhkEU/u2rWI2ofmQDLGtqMfyKH4lAsODbJCKiheNc\nVFQ2pJT4l1OH8IinDTWW4s4nZV1fB/MKF4JvnELs5BU4710JYTIUtQYiIpo7Bhyas6q2pbpu3zdw\nBi6zFVvqVuiyfcVmQvUXbkLs2GVcfaUHzvvXwLzMqUstRFQaZFqFOpqEGk0gHU1CjSQy1xNQo0mo\nY0kkL4Vx5ceHIYwKhNkADK2D4c3TEGYDhMkAYVZy1xWzYcLyiZfKjKO001QMODRnNh0DzsXoCN4a\nPIudbZ/SrQZAm7izqmUJzB43Qm+cQrzBAYfXA2Hg0V6iSiGlhEymoUaSE8JKAmokmbsuUyoAbfoX\nxWaCYjfD4DBDsZthWu6CxW6CwWFGcjCK4Vd64PA2w7zCBZlII/Wvr0Pc/DBkIg2ZTGuXiTTkWArJ\nYHzysuz1pAqJqQORCuPEUJQJShOCkXJtUMqslxqJQY2lINMqoEqkUxbI4TEgLSFVbZlMS+1SzVym\n1cz92eXX3E5PWK7K8fumfU51/LaUMDgsSAdjefs7zjngRCIROByO2VckyrNEOoV/PvUOntzghVEp\njcNCBqcF1X/QgrHDA7ja1QPXA2thWmLXuywimoFUZa61Jdfikr0eSUAdSyKbIRSzAYrdnPkxweC0\nwNTo1IKM3TTnQ9SRg725y7odWyCMCgzGOAwNC99faEFMnRqIkipkIg01kUY6moAcnhyWUiMxyHAC\nwz/1Q3GYIQwK5IgHytt9gCIgDEK7VBTAICAUMX6pKJNvmxUoBgVQso+ZcJ8hs27uOa+9Pb4eBJAI\nBBE7dhnx3hFYmqsX/PrMGnBeffVVtLW1IRAI4KGHHgIAHDt2DCMjI9i6deuCCyCaTdeZ93D/8o1o\nqCqtgC2EgO3mRphXViP0+imYm92w39VUlG1nd2xIqZCpzA4tpWYuJ9zOXEdKRepyFPGzwzC4LJCq\ngtKIikQLI9Oq1hIxloIaSyI1NIro+xe1w0PRBNKDmyB/fBiA9j+rtbaYcuHFtNShLXOYoVSZtA/c\nPKrbsQVXOg+jbseWvD4voP0+wmwAzAZgHnnpSqf2esAgUPPFTQCA9Cv/BkP7o3mvcT4ivt7cZVEC\nzt13342DBw+is7MTe/bsgdvtxrZt2xAM8mwSKrxDg+egQOCOhpV6lzItY7UVNY+1YvS9ixh+pQep\nq2NIRxK5gJGOO5AOBDMBZGr4uDaYTAot08yNJSAAo6I1SRsVrfk5d9swvtyoQLGbkbw6ChlLY+z4\nZQhFQXpo4/hO32SAwWmGwWmB4rRMuq6YKzcGSVVCHUtq3+gn/CQHo4i8HdD6OygCMtQI5YNLmW+o\nyC2HIiZfVwQgMutkrme/rU6+jvHr1z6HwOTlGflsti9VUkrIeDoTVpJQYynt7zOmXcpYKrcse3gI\n0F5LYTVCqTJBJjJBZzQJy6pqKA4z5Mi/wbTj8zr+ZqWnkKFrIfJd16wBx+PxwOPxoK2tDS0tLQiH\nw+jp6UFLy43P/UM0F4NjYfzm0kn8xQL63WxduiaPFU1PCAH7HSsgrEYkXjuJkf84DqPbCmE0QI7W\nQTk/Mh4+MkFEqTJOCSa50JJdN0/fJqOZb2zpYAx1O7YgHfwZDI9rO301kYYaiSMdSiAdjiN5MYxY\neAjpcAIymdZ+vykhyAKDywzFUTohSEqpfQhmQ0t0QnAZG7+UqhYahSKgVJm0b+82E5QqE9RYEoin\nAQgY6myAKqEakhBVRq1PQaavAJLadVWVgJzQp0Bm+xZgfN2Jj8tdx/j1Kc8x4bEA1HgKaiiBy99/\nH4rVqN1/uRXoPDz+fsq9b665nQ27psx763ohOPMYGETeOrHKlDoeTmLJTOvKeGCRsUyAiadyh4Qg\nAMVs1P4vMoFFsRphcJhhbLBBsZqgVBmhWI3THh7Ktkwk+oJw3qN9KUor6nXXpco3Y8B59NFH0dzc\nDLfbjba2NrhcLjQ1NcHj8SAQCKCpqTjN8Yud6t8PpfUevcsoqqSaxj+deBt/tP4umA033hfeW6SA\nkzXWM5C77v7MBgBA+pWfwnBPR1HruNZM34wUswFKrQ3G2ulPvVcTaajhONLhONLhBJIXQ4h9HJ8U\nghSzAYojE3yyISjTInQjp9RnD8NpISUxpaUl+yPTmQ8wASgWYyawmHPBxVRbNSnEzBQasx+Q8bNX\n4fiEtn9Lf3AFhpvmN1N9Pl3pPIw0tDGZsn+/9CvdUB57JNPap05oCUxPbhHMXFejiWvWTU99TPo6\nQSATPsZD04RAblCQ7I9g5BfHteA40ApkWwUNWoBXqkyTwoppiR0ic12pMkFYDHk9M6hUWyZIHzN+\ncvzgBz+A06mdBvvd734X3d3dOHLkCACgubmZfXCKRPp9wCILOD87+yG8jWuwzObWu5R5qdQdrGI2\nQKmzwVg3UwhKQQ0nkA7FkY7EkewLIRaJIx2K5w4pKGbDhPBjRupyFNFDfbmwMrG/RG79CeFEsZnG\n+0xkA4tAaLR9AAAgAElEQVQxf2ewleLfb7qahBCASTs7BlWF276UmbNgrglQiQshIKnCsrEe1nW1\nUH/6GgyPf7ZwhRDN04wBJxtuAKC1tXVSoDl48GDhqqJF7aMrfRhNxbF1yWq9S6F5UMxGKHXGmUNQ\nPBOCwnHEzw9DxtJQUyos6+ug2MxA8BcwdHyO432UECEEYBRakLSOf2SEfn0aADD6wUVUbazXqzyi\nac35q08oFMLf/d3f4dVXX0VfXx/6+voKWRctUldjUXQHjuIra++c04ec3oMP0vwoFiOM9TZYVtcg\n2R8BACTOj8C83AVjtRVCURluykTdji0weVwl1dpFNNGcA87DDz+Mr33tawgGg+jq6uLhKcq7tFTx\ng5Nv4Yl1d8JqNM3pMXoOPkgLww9IIiqkefXe9Hg8+NrXvlaoWqiM+QbOLLhD77+fP4Jb6jzwOGry\nVBURES1WHF+e8uLgwJkFPf7o8CUMjIWwfdn6PFVERESL2ZxacPbu3YsDBw4gFArB7XZDSgkhBLZt\n25Yb3ZjoRgUTY/jF+cN4pu1T7H9BNIv0vk6g7wTU33ZB+aS+ww8QlbIZA87Ro0fh9/uxdetWPPzw\nw1PuDwQC2Lt3L5qbm7Fp06aCFUmVS5US/3TiLTy+5nbYjGa9yyEqCm2wwBgQGwVi0cxlBDI+Or4s\nri2XsSgcQ0Gku17THnzhJAAJ+d4bSPefA0xmCHs14HAD9mrtut0NOLRLsYBxpIjK2ayniT/22GPT\n3p8d5TgQCOS9MFocugNHsaF6Kda4Ku80U5lOQX3tu0Dfx0j/6/8N5fYHAYtt/Mds1Sazo7Ilk4lc\nENFCyShkLJK5PR5UZGwUSMa1qRyk1C5NFgirDbDaAYtdu7TaIOqWZZZp9wmLDZGfHENdh9YZW/1t\nF+R7b0Dc/iCUT3ZAJmJANAhERyCjQcjQEHDpNBAZgRwNAmltIEYYTRCZ0KMFoWwIygShOXbsr0RS\nTQPpFJBKAumkdj13O6UtS40vlxOXZ9dPXfu4ZG5d+5WrSP/kDe2+4GUgMoz0//csUL1kvAghAKFo\nP0r2+vilUJTx+7PLFeWax11zO7feNcsUBZZIP9T3BwChQJ7+QNtP/eS/Q2y6CxCGzHMp49uddGnI\nTCliuGbbmctrHyOUzLqZx1xz//Va7tXfdi24pXLGgOPxeOb0JHNdr9Tl4wWluTsVHMTZ8BD+S8t9\nepeSFzKVBPrPwhJ+G+mfvqnt5C6e0u48dwRySTMQH9U+BOOjQCKmfdhNZLZC5EJQVS4MiQnXx5dX\nldy3c6mqQCoOJBPazzXXZTKhfdCntEtrKAD1t8e1+wIfA1cvIf39v4FoXK3tDAEAmUsx8VJMuEvk\nbmpXxPjySetM2InO8NzW0CDU358EBCB7jwH9Z5H+/t9oH0bZv1k2qBjNmZAyOagIdz2wZEJosdq1\ndfNwCFb5ZAfSA+dz+yhhtgJmK1CzFDM9u0zGM0EoE4Yiw8DAufEglEppKxqN461AdncmFE0IQqbC\ntrTKdEp7f6SSmffNhJ+ktlxOup2YtH7VSD/UXx7U1hnsBcJXp4aJ7P9d9u+oKIDBpIVAgxEwGAGj\nSbs0mDL3GSfcztxvtmrr59bNXh9/rDAaEf35x6j9g1sAxQD1H76ubXssAsOfvZgpJzNdh1QBVQUg\ntUupTl5+7e3rrpe5zC6b5nFpwxiEsxaQKmTfCa2mvuPATXcBahJIao+TqgrIdOZ50+PPd73L3LZU\nQE1ntp8ef55rH5OrZ5xjKIL0K91A3wkAEvLD3wCFCDgzef311yuu/4388NcAJOQHv4K873H2Bymg\nSDKOn5z9AP9byyehlOnrLFNJ4NIZyL7jkBdPAek0RONqpE1LoHzuf4GwVI1/277101C8X5z5+aQK\nJOK5EKT9jEHGRyGjQeDqpUnLEB8d3zlkd9gmSyYgjYcgWG2AuQrG2ADkRTvUI78F+k4gvfd7UG57\nQAscSS2IyAnhI3c5JaAktG+n44WPX1cU7YPcZAZMFsBoAUxm7WfCdWFzAsY6pCwSYsNNgMkC2bNf\ne47gZYg/2JmZJkBesw05PncR5ORtSznh/usszy2W13/uzGXy7CnY1q7V5pV6d2+uJuUr/6f2gSbK\ns9VNmCzah3zmg366/zqZSgLRES0IRbRWIQz2jl9PJbQVDUbYhiXU330M2X9Ge0/95CWINTdPCR3Z\n27lgkg1TueImVKMYtKBhMgPG7I8p8/4xTVhm1t7rJkvmfm1Z/NIZVH1yM4TRDPXlv9Sec0KY0IVi\nzn0ZEbfcr+0TbhmfY0+IbABXgCJN7ZZ6R0Cs11oFxaUz4/upNn1HzR/pPIy6x7eM7zsnvE7zNWsf\nnJ07d6K5uXnScikl/H5/xQWc7BsPS5qh/vzbUO7/KkS1fnPQVCopJX548m38wapb4DRb9S5nzmQy\nkQk0H2uBRk1DLFsL0bQB4vaHIMzaePmpi4e1FhdM/bY9EyGUTDCpAlA3vnyu9UmpfXhkQtC1YciQ\nvAx56n3gqA+ABI76IM3XhBCTJdMqYc9dh8mSu09kbxuMefkCkHpHQCzThhcY3/HfD+HW7/8ubY5C\nLF+n1XTrp8drskw/QnMlEUYT4G7QfjBzEBrrfBvWdcsg339DW9j3McSWT04JHVrANUEYLbnWjkJ9\ngVSNVyEyQ01cL0zobT77hGKp1JpmDDgtLS144YUXrjuo3969e294o6Uq+4IaHv8ryMsBqHv/J8Ta\nWyBuewBCKY0ZkyvBry5+jCZ7NTZWl/YgfTIZBy6dhgycgLx0CpByPNDc8bB2aKCEiEy/DpgswISx\nhLIfI/FjdXDctwWqVLWd/m0PVNwOLd9KsaZSIYwmSKMLYvna8SB466chNtyhd2k5/PstbrMeoppu\nxOLrnVVVSUSDB8pjz0J+8CbUV1+E8qmvQCxpnv2BNKNz4SvwX72I/71tu96lTCGTceDiqUwLzWmt\nY9+ytRBNGyHu+ozWxF8BuNOnfON7ikpRafVQ1Ml08xkJxQBx+8OQa2+D+pt/gaj3QNz9+YJ3sqtU\nY6kEOk+/i6c33QulBPoxyERsPNBcOpMJNOsgPJsg7vpsxQQaIqLFiAEHs89nJKoboHxpJ+Sxg1B/\nsgvKPX8I4bmpSNVVBikl/uXUIXyueTNqdOrLIBNjwIUJgcZg0ALNylaITzC4EhFVEgacORJCQLR4\nIVe1Qe7rhDz2FsR9j2mdMWlW+/tPo9psQ1vt8qJtU8bHgAsntUDTf1br2Lh8HcTqzRBbv7iox/0g\nIqp0DDjzJGwuiEf+DPLMR1B/9j+g3NEOrL8972cETHfYrBxdiI7gncvnsLOtcGcyyFQS6m9+pA1W\n9c8vjJ8NtGIdxJqbIbxfYqAhIlpEZg04Bw8exObNm+FwOCYtf/XVV2cc5TjL7/ejtbV11mXlRqy5\nGUrTBsj9P4c8/jaUTz0B4czfLNizHTYrF/F0Cv9y6h38541eGOd5JppMxoHREBANAaMhyNFQ7rYc\nDWqjxGbHMDEYgN5j2vUrF6D8+f9TcoPgERFR8czpLKp//Md/xJ/+6Z/mlvX19cHlcs365D6fD88/\n/zzefPPNGZeVK2Gugrj/CciLp6D+8jsQLVshNt9XtgOBFULX6XfxwIqbUG91aOO0JMauH1pGQ5DR\nkBZaskxmCJsLsLkAuwuwubUxU2wubXmVc9JUB+MDQ93PcENEtMjN6VOgqakJfX19aGpqAqCNgTMx\n8EzH6/VOmcbhesvKnVi+Dspjz0G+swfqT/+HNkBgbaPeZRWFlCoQG0VtJAjZe1xrWckElsErF7A9\nMoIVRgvS2QdYqsZDSya4iNrl2qXNpQ1rf4MBkaeqElEpqqQuB4WWz9dqTgHn4YcfzrXiHDt2DC0t\nLXkroFIIgxFi6xcgr1yE+qsfQjRvgrijveJaEmRkGPLEu5DvvQ6Er0Ld/VfA0lW4JRaGhEFrXXHU\n4Kq7Hv9uAP7koSehWB2c9oKIFq1K6XJQDPl8reb8VbmlpQXHjh2Dz+ebdvA/AkTdcihf/gZgsUF9\ndZd2OnKZk9Eg1A9/jfRPXoL6q3/WJhAcDWl3jkVg+OKf49eb7oTi/RKUW+5Hat2t+J+RIXz+5gdh\nqnIy3BARlZjF0Ko05+aFrVu3YufOnfj6179ekEL27duHc+fOFeS5i00IRZsDZe0tUH/9LxDuBu0s\nnhIb2n8mcjQEefJ9yFPvAYoRYsPtUD739dxp8erlwLRzvPz07Ae4t3EdGm2z99MiKjWLYcdfyfj3\nm5vF0Ko0r+Mn3/72twtVB7Zv3459+/YV7Pn1IJy1UL7w55An3oX6k5egbP0CxOrNepc1LTkWgTz5\nnjYhIwCx/nYojzwNUeWYsu50/V0+GAoglk7h7qWri1IzzR13/HNTijt+/u3mrhT/fqSPgnYQ6e7u\nRk9PD7q7u9He3j7tskomhIDYeCdk8ybI370CefwtiO07IKqcepcGAJCxKOSpDyBPvguoKsT626C0\nfw3CNv/6rsSieL3vGHZuLp2Ze2kcd/zli387ovkraMBpb2+fEmKut2wxEFUOiIf/M2TvUag//wdt\n1t2bPqFL/xQZH4U8/SHkiXeBVBJi3a1QHnoSwu6+8eeUEj848Ra+uv5OWA0cUI/mhi0TRFQolXWK\nTxkQzS1QHlsLefAXUD9+B8r9T0C46gu+XZkYgzz9EeSJQ0AyDrH2VigP/BGEY+GDEz79+x/lrjfZ\n8zfYIVU+tkyUNwZUKmVzCjh79+7FgQMHEAqF4Ha7IaWEEALbtm3DQw89VOgaK44wWSDuewxy4DzU\n174Lsf52iFs+PWnQunyQiRjkmY+0lppYFGLtLdoYPc7avG6HJuNOnxYLBlQqZTMGnKNHj8Lv92Pr\n1q14+OGHp9wfCASwd+9eNDc3Y9OmTQUrslKJpSuhfPlZyPffgPqTF7XpHhqaFvScMhkHzh6BPHEI\nMhqCWHszlO0dRWklIg13+kRE+psx4Didzhnnm/J4PPB4PAgEAnkvbLEQBiPEnZ+BXHcb1F//CKJx\nFcQnPjeviSFlKjkeaiLDEKu3QNzzZSjVDQWsXKNKidXOOhiFgmdvfrDg2yMiIpqLGQPOxCkVIpHI\nlAk3r7ce3RhRsxTKo38B6d8P9dX/BuXexyCaNky7vkwlgfN+yI8PQYavQKxqg/B+CUpNcVsPPh4Z\nwDrXEpyPXCnqdomIiGYyax+cV199FW1tbQgEArn+NseOHcPIyAhHNM4zIQRE272Qq7dA/ubHkMff\nhrj3D3P3y3QKOH9Ua6kZGYRY2Qpx9+eg1C7Treb9/afw+ZVbcP4UAw4REZWOWQPO3XffjYMHD6Kz\nsxN79uyB2+3Gtm3bEAwGi1HfoiTsbojPfR3y1AdQv/d/AaMhpH/wt4C5CmJlC8Qd7VDqV+hdJiJC\nQTSV4IjFRERUcmYNONl+Nm1tbWhpaUE4HEZPTw87FReBWHcr8MvvaDdGBqE8852SmtfpbZsNdy/h\niMVExSZavXqXQFTyZjwvORKJ5K5nZxB3Op3YunXrlP45VBjilvsBCIhb7i+pcCOlxPvWKtxWz/5X\nRMWmtN6jdwlloxTDIGuam4XWNGPAOXLkCF5//fUZn2Dv3r04cuTIgoqg6Smf7ACaNkyZ80lvZ8JD\n8CSTMBs4ViTlTyXuZElfpRgGWdPcLLSmGT+dtm7dinA4jO9+97vo7e3NtSBkB/prbm5GR0fHtGdX\nUeXa338anxyN6l1GWeAH5NxV4k52seD7nErNrF+/nU4nvva1rxWjFioTY6kkLo9F4Ekl9S6lLJTi\nByQ/jCjfSvF9TovbnOcGOHbsWCHroDJy6PI53LVkpd5l0ALww4iIKt2cA86BAwfQ19dXyFqoTLwz\neB53NqzSuwwiIqJpzTngtLa2IhwOF7IWKgO9kauor3Kgah5TSRARERXbnE+B8fv9GBkZwXe+853c\nTOIzzVNFlWl//2nc27hW7zKIiIhmNOeA4/VqnRKz4+Fwgs3FJ5FOoS86jK8479C7FCIiohnNGHAm\nTrCZDTZZc52IkyrH+0MB3FbfXFIDDhIREV0PB/qjOXtr8CzuXrJK7zKIiIhmdUMD/WVxoL/Fo380\nBLvRDIfJqncpREREs+JAfzQn+/tP457GdXqXQURENCczHqKa7fAULQ4pNY1ToUFsrF6qdylERERz\nMmMLzt/8zd/gwIEDaGtrw9atW9HU1JS7jx2LF4/DVy9gS+0KKGXQuZhTEBARETBLwHn22Wfh8Xjg\n9/uxa9cuHD16FM3NzWhpaUEoFMLf//3fF6tO0pFv4Ay+uu4uvcuYE05BQEREwCwB5/HHHwegdTbO\nCofD8Pl86OrqKmxlVBKuxKJQhIIai03vUoiIiOZszgP9ZTmdTjz88MOTxsGhynVg4DS2LV2jdxlE\nRETzMue5qK517cB/VHlUqeLo8CW01S6fdd2tDEFERFRCbjjgVCp2Uh13dLgfN1U3wiBmf5t4GXCI\niKiEzPjJ9dJLL6Gvr69YtZQEdlIdd6B/5sNTDINERFSqZgw4Ho9n0qnhtHgEE2OIqyk0VDmnXYdh\nkIiIStWsUzU8//zzCIVCCIVCEEKgqakJzc3N8Hg8uR+Oh1N5Dg6cxdYlq/Uug4iI6IbMGHA8Hg9e\neOGFScsCgQD6+voQCoVw4MABBAIBbNu2DQ899FBBC6XikVLiwysBfGPLA3qXQkREdEPmfZp4ttVm\nor179+atINLfyeAg1jjrYVIMepdCRER0QxZ8FlU4HM5HHVRCft9/Gvc0rtW7DCIiohu24ICTHfiP\nKkM0mUAwMYbl9mq9SyEiIrphHAeHJnl78Cw+sWSV3mUQEREtCAMO5Ugp8e5QL25vaNa7FCIiogVh\nwKGcc5ErWG5zw2ow6V0KERHRgjDgUM5+di4mIqIKwYBDAIBYOon+0RBWOmr1LoWIiGjBGHAIAPDu\n5V7c0dAMIYTepRARES0YAw4BAN4ZPIe7Gjg1AxERVQYGHMKF6AiqLVWwm8x6l0JERJQXDDhlQLR6\nC/r8+/tP456l7FxMRESVgwGnDCit9xTsuZNqGmfDQ1jvXlKwbRARERUbA84i98FQALfWedi5mIiI\nKgoDziJ3cPAsti5l52IiIqosDDiL2OWxMKyKES5zld6lEBER5RUDziK2f+A0tnHkYiIiqkAMOItU\nWqr4eGQALTWNepdCRESUdww4i9SRqxfRWrMMiuBbgIiIKg8/3RYpX/9pbOPYN0REVKEYcBah4fgo\nVEjUWu16l0JERFQQDDiLkG/gDLxL1+hdBhERUcEw4CwyqpQ4cvUCttSu0LsUIiKigmHAWWSOj/Rj\nvXsJjIpB71KIiIgKhgFnkeHEmkREtBgw4Cwi4UQMo6kEltpcepdCRERUUAw4i8hbg2dx9xLOO0VE\nRJWPAWeRkFLi/aEAbqv36F0KERFRwTHgLBKnQ0NodtTAbDDqXQoREVHBMeAsEvsHTuOexnV6l0FE\nRFQUDDiLwFgqgaGxCDyOGr1LISIiKgoGnEXgncvncdeSlXqXQUREVDQMOLPwDZzRu4QFOzR4Hnc2\nrNK7DCIioqJhwJnFwTIPOL2Rq2iocqDKaNK7FCIioqIpeMDx+/2Tbnd3d8Pn82H37t2F3jQhM3Jx\nI0cuJiKixaWgAcfn82Hnzp2529mw4/V64XK5poQfyq9EOoW+6DDWOOv1LoWIiKioChpwvF4vPJ7x\ngeX27NkDl0ubJsDj8cDn8xVy84vee0O9uK2+GUIIvUshIiIqqqL2wQmHw3C73bnbIyMjxdx8xZhr\nx2dtaoZVhS2GiIioBLGTcRmaS8fn/tEgnCYrHCZrESoiIiIqLUUNOE6nE8FgEAAQCoVQXV2du2/f\nvn04d+5cMcupaPv7z2DbUnYuJiKixamoAeeRRx5BIBAAAAQCAXi93tx927dvx6pVq4pZTsVKqWmc\nCg1iY/VSvUshIiLSRUEDTnd3N3p6etDd3Q0AaG1tBaCdXeVyuXK3Kb8+unIBW2pXQGHnYiIiWqQK\nOrV0e3s72tvbJy3r6Ogo5CYJWifkP1p/l95lEBER6YadjCvMUCwCg6Kg2mLTuxQiIiLdMOBUGN/A\nGWxbukbvMoiIiHTFgFNBVKnCP3wJbbXL9S6FiIhIVww4FcQ/fAmbqhthEPyzEhHR4sZPwgpygGPf\nEBERAWDAqRjBxBgSagoNVQ69SyEiItIdA06FODhwBlvZuZiIiAgAA05FkFLiwyt9uKWuSe9SiIiI\nSgIDTgU4ERzEGlc9TIpB71KIiIhKAgNOBdjffwr3sHMxERFRDgNOmYsm4wgmYlhur559ZSIiokWC\nAafMvT14Dp9YskrvMoiIiEoKA04Zk1Li3aFe3NGwUu9SiIiISgoDThk7F7mCFTY3LIaCTgpPRERU\ndhhwytj+/tPY1sjOxURERNdiwClTsVQS/aMhrHTU6l0KERFRyWHAKVPvDp3HHQ0rIYTQuxQiIqKS\nw4BTpt4ePIe7GlbpXQYREVFJYsApQ/F0CrUWG+wms96lEBERlSQGnDIUTIxhW+M6vcsgIiIqWQw4\nZSStqnj69z/CSGIMLx1+U+9yiIiIShYHUClBiXQK/WMhXBoN4tJoCP2jIYSSMRgF8ygREdFcMODo\nKJpMoH8sOCnIxNJJmBUDltpcWFblRkt1I+5fvhFOkwVCCDz9+x/pXTYREVHJY8ApMCklgokxLcCM\naUFmYCyElKrCZjSj0ebCMpsbd9Q3o9HmQpVx5o7DL9/7BF46/Ca+seWBIv0GRERE5YcBJ09UqeJK\nLIpLoyFcGguifzSEoVgEqpRwm6uwzOZCo82N+5YtwdIqF0yKQe+SiYiIKhYDzjyl1DQGx8JakBkN\non8shOH4KBQhUGuxY1mmReaWuibUWx0wsN8MERFR0THgzCDb3+Xp3/8IN9c1IZqMwyAULK1yotHm\nwhpXPbyNa1BjtnFEYSIiohLCgDNH/2n9XXCYrHqXQURERHPA4ydzxHBDRERUPhhwZvDyvU9gg3sJ\nXr73Cb1LISIionlgwCEiIqKKw4BDRESkE9/AGb1LqFgMOERERDo5yIBTMAw4REREVHEYcIiIiKji\ncBwcIiIiHUwcTHahZ+sGAgHs2rULALBjxw643W74fD489dRTC3rerq4ueDweuN1u9PT0wOv1wuPx\nLOg5i4UBh4iIKM9eOf0eAtHhOa//0uE3Z7zfY6/B42tvn/5+jwfbtm1DKBSC1+sFoIWTrq4udHR0\nIBAI4Mknn8S3v/1tBINBdHZ24h/+4R8QCASwc+dOfPOb30QwGITf78+FoieffBIvvPBCLtC0trbi\nmWeewbe+9S0Eg0E8+eST+N73vgePx4Pnn38eHR0dcLlc025ntu23trbi+eefR3t7e+53WAgGHCIi\nojybKYxkZVtwAOAbWx7Iy3ZHRkbg9/vR1dUFp9OJjo4OAFoA8ng8aG1thc/nw3PPPZdb3tTUhNbW\n1txzdHV1wev1IhgMTmmt2bZtG15++WU899xzuecEgPb29txzTLed6Zb/2Z/9GV588UV873vfy1u4\nAdgHh4iISBeFGEy2uroara2teOGFF6bcFwwG4fP5cODAgWkPM3k8Hhw4cAAAEAqFrrtOOByesYbp\ntjPdcpfLhRdeeAHPP//8rL/ffLAFh4iIqALt3bsXTz/9NFwuFwDA7XbD6/XC6/UiEAgAwJSgEwgE\nsHnz5lyLS3a9UCiE1tZWdHd351qFnE5n7nFut3vS9ettZ6btZ1t3Ojs72YJDRERE43p7e3PBAQA6\nOjrQ1dUFn8+HQCCAQCAAv98Pn8+HXbt25QJMX19fbvnEPjjf/va30dnZie7ubgSDQTz66KNoaWlB\ne3s7AODpp59Gd3c3fD5fbpszbed6y/1+Pzo7O3P1VldX5+31EFJKmbdnW6B9+/Zh+/btepcxyUuH\n38zbsdF8KcWaiIho/nwDZ+BdukbvMuZk9+7d6OjowGuvvZZrxSllbMEhIiLSSbmEG0A7i+rTn/40\nTxMnIiKiyuH1enHo0CG9y5gztuAQERFRxWHAISIi0onq3693CRWLAYeIiEgn0u+bfaU58Pv9eOCB\nB7Br1y50d3dj165dk85umigQCOCZZ57Jy3ZLGfvgEBERlbnW1la0tLTgkUceQWtrK9rb23HnnXde\nt8+Mx+PBt771LR2qLC624JShrWXU656IiPThdrsnjYuTlR2TptKxBacMldNphUREdH3qb7uAvhNQ\nf9sF5ZP5HVcmFArB5XLB4/HkZgQPBAK58Wuycz9Ndxgry+v1Tnm8z+fD7t278dRTT8Hv98Pr9U6a\ny6pUMOAQERHlmbqvE3Kwd+aV+k4AkJDvvYH0wPkZVxVLmqFs3zHrdn0+H4LBIAKBAL7//e9j9+7d\naG1thdfrnRRMstMsHDhwANu2bctNn9DZ2Tnp9nSPf/HFF+H1euHxeNDZ2cmAQ0REtBjMJYyov+2C\nfO8NiNsfzFsLTjaMZB05ciQ3tUI2jEz09NNP4+WXX8aLL76Ib37zm1NuT/f4iXNPlSoGHCIiIh0o\nn+xAeuB83g9PTZQ9tJS93Lx586T7fT4fnnvuOQDaVAwej2fK7ZkeX8oYcIiIiMqc3+9HX18furu7\n0dbWlptB/LnnnsPu3btz62T7zRw9ehSBQABHjhzJPUd7e/ukFp729nZ4PJ4pj89O3pmdOPPo0aO5\nPj+lhJNtzoITWxIRUaGkX/lvMDz+V3qXUZF4mjgREZFORKt39pXohjDgEBER6URpvUfvEioWAw4R\nERFVHAYcIiIiqjgMOERERDoZ7RnIy/NMnGwzFAoBAJ555plJt2czn0k4y2HCTp4mTkREpJOxngHY\n2pYu+HkmTrbpcrkQCoWwY8eOSYP+zWY+k3CWw4SdbMEhIiKqMD09PfMKN8D8JuEshwk7GXCIiIgq\nRDAYxJNPPnndqRS6u7vR1dWFrq6u3O0nn3wS3d3ducH8XnzxRQDaCMdPPvlkbv6p7u5u+Hw+PP/8\n8wLJRlIAAAdOSURBVLlDXtl1SxUDDhERkU7SwVhen+/AgQN49tln8bd/+7eTlvv9/txs4NmA097e\njkAggPb2dnR0dMDj8eQm4fR6vQgGg/B6vWhvb8eePXtys4b39PRMWrdUsQ8OERFRnoV+fQbJwciM\n66ixFNRQApe//z4U68wfx6YlDrjuXzPrdh955BG0traira0N3d3duYkyW1tbEQgE4PP5JrXutLS0\nAMB1p1mYuJ7H4xn/3ebYaVlvDDiz2Lp09jcUERHRRHMJI1c6DyMNQLEaUbdjS163/+yzz+LRRx/N\nBZyJrTa7d+/OTaBZyXiIahZeBhwiIiqAuh1bYPK48hJushNo7tmzZ1ILyzPPPINQKASPx4NQKASf\nz4eWlpZJE2X6/f5Jz5Ft6clOqNnd3Y2DBw8iEAjgwIED8Pl8k9YtVUWfbDM7/XowGERHx+Qp4ktx\nsk0iIqJCudJ5OO+tN6QpaguOz+cDoDWR9fb2lnTyIyIiovJV1IBz4MCB3DG/5ubmXOAhIiJajKry\nMMgfXV9ROxlXV1cjGAwC0Hphj4yM5O7bt28fzp07V8xyiIiIdJWPUYzp+oragpM9NAVooyBWV1fn\n7tu+fTtWrVpVzHKIiIioQhU14Hg8HjzyyCPw+/1wOp0Vf4oaERER6aOoAcfv96Onpwetra0Ih8O5\n8/OJiIiI8qmoAae1tRVutxvd3d1TThEnIiIiypeij2TMVhsiIiIqNI5kTERERBWHAYeIiIgqDgMO\nERERVZySmk38448/1ruEsnDu3DmOGTRHfK3mhq/T3PG1mhu+TnPH12purn2dqqurccstt0y7fkkF\nnI0bN3KyzTngpKRz99d//df4kz/5E73LKHl8T80dX6vp+f1+tLa2AtBep1gsBpfLBb/fj6eeekrn\n6krLta/VwMAAAKC3txfPPfecnqWVlGtfp+z/3u7du2d9T5XEIaru7m74fD6cPHlS71JKXldXFwYG\nBrBr1y69Syl5Pp8P/f39epdR8vx+P2KxGLq7u/UupaRl91PZDyKazOfzYefOnbnbDQ0NAACv15sL\nOaS59rUym83wer3o6OhAIBDgPI0Z175O2XDj8/nm9BrpHnCyb3r+E8zO5/Pxn4Dy7uWXX0Z7ezsC\ngQD//6bh9/vh8Xjg9Xrh8Xj4Ol1H9rXJ2rNnD1wuFwBtFHvur8Zd+1pN3J97PB4EAgG9Sisp175O\n86V7wOE/wdzxn2Du/H4/vF6v3mWUvO7ubmzevBkA8NRTT+WagmmqF198EYD2f8jXaXbhcBhutzt3\ne+LkyjRZR0fH/9/eHSMl74RxHP+9B4Bwgb1AgiVNagomJQ21BVJaWWqhdtJg5+QCprFkcgFygRAO\nYC5gcgHfwmFfUUQZ4R/+4fupIrMymQy7efLsZh+7+e1isVC73a74jA7XNmN75QEOneDn6AQ/t6xa\nj83SNNXLy4uyLFMYhlWfzsHyPE/GGHU6nZXxCtilLMvkui4B9AbbjO2VBzjYHp1gM7I322m1Wva3\nxDqc9cqyVKPR0NnZmS4vL8me/kCj0bA3o7Is1Wq1Kj6jw5ckCQuMN9h2bK/8LSo6wfboBJvlea48\nz1UUhYqiWFmFj1WtVsvOcTebTaVpSjmVNaIo0mg0UrPZlDFGcRzzVtA3giDQfD6X9NYneejYLIoi\n+5tarrfEqm3H9sozOEEQ2KchOsH3PnYCfNbr9exNuizLis/msC0XF0tv12q5Hgdf6/V6dt0g/onj\nWPP53GYBlzeeJEnUbDZ5yHjn47VKkkTj8VjdbledTqfiszscH6/TtmP7n9fX19e9nuEPRFFkF81S\nZfxry1fmHMdRURSaTCYEhPi1KIrkOI7SNCUzuEEYhjLGqCgKxingf+AgAhwAAIBdqnyKCgAAYNcI\ncAAAQO0Q4AAAgNohwAGwU0mSqNvt6u7uTnEcKwxD9fv9L9uenp5+OgaA36p8HxwA9eL7vlzXVRAE\n374a7Pu+Hh8fPx0DwG+RwQGwV2VZsgcKgP8cGRwAe5Ukid2ca5s9r963Xf59c3Oj8/NzTSYT5Xmu\n6XSq29tbRVEkz/OU57kcx1EURRoMBmsLY7J3FHAcyOAA2IvpdKqrqytbimW5UZ7v+zLGbCzu+bFt\nWZZyHEee58n3fVsy4f7+Xg8PD/bz5+dnuztzr9fTYDDQbDaTJPtdAI4DAQ6AvQiCQNfX1zaoSNPU\nHhtjlKbpl/+7rq3rurbYXhzHNnDK81xlWSrLMlvLznVdSW/1tUajkWazmfr9PqU7gCNCgANgr5ZT\nQu+nm/I831j3al3bIAg0Ho9X6mdJ0snJiYwx8jxv7bTXsjjt09MT9duAI8IaHAA7lWWZFouFptOp\nnUqSpIuLCzstlWWZhsOhbZtlmSTZ43VtpX+ZGWOM2u22JGk4HCoMQ5vRef89nuetZIqolA4cD2pR\nAQCA2mGKCgAA1A4BDgAAqB0CHAAAUDsEOAAAoHYIcAAAQO0Q4AAAgNohwAEAALVDgAMAAGrnL1kD\nIyCR+hFKAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f237bb719b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_multiple_followers_perf(run_multiple_df=run_multiple_df, measure='top_1', \n",
    "                             # file_name='multiple-followers-top-1.pdf',\n",
    "                             yticks=[9, 10, 11, 12, 13])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjgAAAFcCAYAAADF+MQnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2374bf60b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_multiple_followers_perf(run_multiple_df=run_multiple_df, measure='avg_rank',\n",
    "                            # file_name='multiple-followers-average-rank.pdf',\n",
    "                            yticks=[500, 1000, 1500, 2000])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjgAAAFcCAYAAADF+MQnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nOAQcAADgOP8/5HHAXfFiTLcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2374b8bcf8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_multiple_followers_perf(run_multiple_df=run_multiple_df, measure='num_events')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "What is the expected number of broadcasters covered with broadcaster pool = n and follower pool = m?\n",
    "\n",
    " - Can do it via simulation\n",
    " - Can do it by exact combinatorics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Overlap between broadcasters"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Keep the number of followers fixed at 2, each with 50 broadcasters and one optimal broadcaster. Then increase the overlap among the broadcasters and look at the performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def plot_overlapping_perf(run_overlap_df, measure='avg_rank', \n",
    "                          yticks=None,\n",
    "                          file_name=None):\n",
    "    \n",
    "    run_multiple_df = run_overlap_df.sort_values(['type', 'overlap'])\n",
    "    perf_data_grp = run_overlap_df.groupby(['type', 'overlap'])\n",
    "    perf_mean = perf_data_grp.mean()\n",
    "    perf_std = perf_data_grp.std()\n",
    "    \n",
    "    # plt.figure(figsize=(8,6))\n",
    "    \n",
    "    if file_name is None:\n",
    "        latexify(fig_width=8, largeFonts=True)\n",
    "    else:\n",
    "        latexify(fig_width=1.75, largeFonts=False)\n",
    "    \n",
    "    overlap = np.asarray(sorted(run_multiple_df.overlap.unique().tolist()))\n",
    "    c1, c2, c3 = sns.color_palette('Set1', n_colors=3)\n",
    "    \n",
    "    plt.errorbar(overlap, perf_mean.loc['Opt'][measure].tolist(), fmt=red_queen_fmt,\n",
    "                 yerr=perf_std.loc['Opt'][measure].tolist(), label=r'\\textsc{RedQueen}',\n",
    "                 linewidth=0.75, markersize=3, color=red_queen_color)\n",
    "    \n",
    "    plt.errorbar(overlap + 0.005, perf_mean.loc['Poisson'][measure].tolist(), fmt=poisson_fmt,\n",
    "                 yerr=perf_std.loc['Poisson'][measure].tolist(), label='Poisson',\n",
    "                 linewidth=0.75, markersize=3, color=poisson_color)\n",
    "    \n",
    "    plt.errorbar(overlap + 0.01, perf_mean.loc['kdd'][measure].tolist(), fmt=karimi_fmt,\n",
    "                 yerr=perf_std.loc['kdd'][measure].tolist(), label='Karimi',\n",
    "                 linewidth=0.75, markersize=3, color=karimi_color)\n",
    "    \n",
    "    if file_name is None:\n",
    "        plt.xlabel('Overlap')\n",
    "\n",
    "        ylabel = 'Unknown'\n",
    "\n",
    "        if measure == 'avg_rank':\n",
    "            ylabel = 'Average position'\n",
    "            plt.legend(loc='upper right')\n",
    "        elif measure == 'top_1':\n",
    "            ylabel = r'$\\int I(r(t) < 1)\\,dt$'\n",
    "            plt.legend(loc='lower right')\n",
    "        else:\n",
    "            plt.legend()\n",
    "\n",
    "        plt.ylabel(ylabel)\n",
    "        \n",
    "    if yticks is not None:\n",
    "        plt.yticks(yticks)\n",
    "    \n",
    "    plt.xlim((0, overlap.max() + 0.02))\n",
    "    plt.tight_layout()\n",
    "    format_axes(plt.gca())\n",
    "    \n",
    "    if file_name is not None:\n",
    "        plt.savefig(os.path.join(plot_base, file_name), bbox_inches='tight', pad_inches=0)\n",
    "    \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "overlap_sim_opts = prepare_overlapping_followees_sim_opts(num_overlap=10, opts=overlap_opts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import warnings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2018-03-31 11:54:33.115177\u001b[0m Checkpoint: active/total = 6098/1000, procs = [('Opt', 4100), ('Poisson', 999), ('kdd', 999)]\n",
      "*** \u001b[31m2018-03-31 11:56:47.823915\u001b[0m Checkpoint: active/total = 7098/2000, procs = [('Opt', 3100), ('Poisson', 1999), ('kdd', 1999)]\n",
      "*** \u001b[31m2018-03-31 11:58:48.200162\u001b[0m Checkpoint: active/total = 8098/3000, procs = [('Opt', 2100), ('Poisson', 2999), ('kdd', 2999)]\n",
      "*** \u001b[31m2018-03-31 12:00:33.626718\u001b[0m Checkpoint: active/total = 9098/4000, procs = [('Opt', 1100), ('Poisson', 3999), ('kdd', 3999)]\n",
      "*** \u001b[31m2018-03-31 12:02:05.704835\u001b[0m Checkpoint: active/total = 10098/5000, procs = [('Opt', 100), ('Poisson', 4999), ('kdd', 4999)]\n",
      "*** \u001b[31m2018-03-31 12:12:44.444901\u001b[0m Checkpoint: active/total = 9300/6000, procs = [('Opt', 0), ('Poisson', 4639), ('kdd', 4661)]\n",
      "*** \u001b[31m2018-03-31 12:23:50.473119\u001b[0m Checkpoint: active/total = 8300/7000, procs = [('Opt', 0), ('Poisson', 4140), ('kdd', 4160)]\n",
      "*** \u001b[31m2018-03-31 12:38:23.870671\u001b[0m Checkpoint: active/total = 7300/8000, procs = [('Opt', 0), ('Poisson', 3640), ('kdd', 3660)]\n",
      "*** \u001b[31m2018-03-31 12:50:10.845854\u001b[0m Checkpoint: active/total = 6300/9000, procs = [('Opt', 0), ('Poisson', 3139), ('kdd', 3161)]\n",
      "*** \u001b[31m2018-03-31 13:02:27.261563\u001b[0m Checkpoint: active/total = 5300/10000, procs = [('Opt', 0), ('Poisson', 2636), ('kdd', 2664)]\n",
      "*** \u001b[31m2018-03-31 13:14:48.863063\u001b[0m Checkpoint: active/total = 4300/11000, procs = [('Opt', 0), ('Poisson', 2140), ('kdd', 2160)]\n",
      "*** \u001b[31m2018-03-31 13:26:36.024640\u001b[0m Checkpoint: active/total = 3300/12000, procs = [('Opt', 0), ('Poisson', 1638), ('kdd', 1662)]\n",
      "*** \u001b[31m2018-03-31 13:41:10.331471\u001b[0m Checkpoint: active/total = 2300/13000, procs = [('Opt', 0), ('Poisson', 1137), ('kdd', 1163)]\n",
      "*** \u001b[31m2018-03-31 13:53:33.866726\u001b[0m Checkpoint: active/total = 1300/14000, procs = [('Opt', 0), ('Poisson', 638), ('kdd', 662)]\n",
      "*** \u001b[31m2018-03-31 14:06:37.621251\u001b[0m Checkpoint: active/total = 300/15000, procs = [('Opt', 0), ('Poisson', 137), ('kdd', 163)]\n",
      "CPU times: user 45.6 s, sys: 5.45 s, total: 51.1 s\n",
      "Wall time: 2h 17min 55s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "num_broadcasters = 50\n",
    "overlapping_run_opts = Options(overlap_list=range(0, num_broadcasters + 1), num_segments=24, repetitions=100,\n",
    "                               setup_opts=overlap_opts.set(kind='PiecewiseConst', \n",
    "                                                           broadcasters_per_follower=num_broadcasters))\n",
    "with warnings.catch_warnings():\n",
    "    warnings.simplefilter('ignore')\n",
    "    overlapping_res = run_overlapping_followees(opts=overlapping_run_opts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "overlap_df = overlapping_res.df.copy()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def sim_opts_overlap(sim_opts):\n",
    "    followees = [set([s for s, t in sim_opts.edge_list if t == x]) for x in sim_opts.sink_ids]\n",
    "    assert len(followees) == 2\n",
    "    return (len(followees[0].intersection(followees[1])) - 1) / (len(followees[0]) - 1) # len(followees[0].union(followees[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "overlap_df['overlap'] = [sim_opts_overlap(x['sim_opts']) for x in overlapping_res.raw_results]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style>\n",
       "    .dataframe thead tr:only-child th {\n",
       "        text-align: right;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th>avg_rank</th>\n",
       "      <th>num_events</th>\n",
       "      <th>num_followers</th>\n",
       "      <th>q</th>\n",
       "      <th>r_2</th>\n",
       "      <th>seed</th>\n",
       "      <th>top_1</th>\n",
       "      <th>world_events</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>type</th>\n",
       "      <th>overlap</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th rowspan=\"30\" valign=\"top\">Opt</th>\n",
       "      <th>0.00</th>\n",
       "      <td>1254.033670</td>\n",
       "      <td>2505.85</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.517349e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>5.002597</td>\n",
       "      <td>100486.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.02</th>\n",
       "      <td>1252.434394</td>\n",
       "      <td>2505.76</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.506377e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.967638</td>\n",
       "      <td>99259.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.04</th>\n",
       "      <td>1251.395308</td>\n",
       "      <td>2503.70</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.504060e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.990877</td>\n",
       "      <td>98112.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.06</th>\n",
       "      <td>1251.574592</td>\n",
       "      <td>2501.54</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.506442e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.999216</td>\n",
       "      <td>97067.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.08</th>\n",
       "      <td>1249.800340</td>\n",
       "      <td>2495.19</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.497974e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>5.017915</td>\n",
       "      <td>95837.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.10</th>\n",
       "      <td>1253.093386</td>\n",
       "      <td>2502.30</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.512201e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.984509</td>\n",
       "      <td>95373.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.12</th>\n",
       "      <td>1251.403023</td>\n",
       "      <td>2501.88</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.506588e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.973182</td>\n",
       "      <td>94221.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.14</th>\n",
       "      <td>1252.533378</td>\n",
       "      <td>2500.30</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.511743e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.985901</td>\n",
       "      <td>93254.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.16</th>\n",
       "      <td>1252.773870</td>\n",
       "      <td>2500.28</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.515505e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.997922</td>\n",
       "      <td>92373.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.18</th>\n",
       "      <td>1251.549904</td>\n",
       "      <td>2497.90</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.510176e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.994623</td>\n",
       "      <td>91228.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.20</th>\n",
       "      <td>1252.595110</td>\n",
       "      <td>2500.88</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.515811e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.983969</td>\n",
       "      <td>90438.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.22</th>\n",
       "      <td>1250.555901</td>\n",
       "      <td>2497.15</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.506842e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.996334</td>\n",
       "      <td>89199.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.24</th>\n",
       "      <td>1197.047597</td>\n",
       "      <td>2390.36</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.508675e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>6.066430</td>\n",
       "      <td>88358.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.26</th>\n",
       "      <td>1252.587131</td>\n",
       "      <td>2497.95</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.521721e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.994079</td>\n",
       "      <td>87430.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.28</th>\n",
       "      <td>1252.831044</td>\n",
       "      <td>2498.45</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.519594e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.966719</td>\n",
       "      <td>86280.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.30</th>\n",
       "      <td>1252.439679</td>\n",
       "      <td>2500.10</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.519788e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.956781</td>\n",
       "      <td>85399.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.32</th>\n",
       "      <td>1249.529288</td>\n",
       "      <td>2494.15</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.508599e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.995153</td>\n",
       "      <td>84095.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.34</th>\n",
       "      <td>1249.185017</td>\n",
       "      <td>2493.20</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.507339e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.987309</td>\n",
       "      <td>83149.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.36</th>\n",
       "      <td>1250.304731</td>\n",
       "      <td>2496.62</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.512540e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.965397</td>\n",
       "      <td>82181.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.38</th>\n",
       "      <td>1250.670676</td>\n",
       "      <td>2494.14</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.514025e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.961991</td>\n",
       "      <td>81397.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.40</th>\n",
       "      <td>1249.355560</td>\n",
       "      <td>2495.83</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.513381e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.984861</td>\n",
       "      <td>80220.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.42</th>\n",
       "      <td>1249.400233</td>\n",
       "      <td>2494.74</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.514018e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>5.017429</td>\n",
       "      <td>79358.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.44</th>\n",
       "      <td>1250.563524</td>\n",
       "      <td>2495.70</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.517512e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.983700</td>\n",
       "      <td>78580.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.46</th>\n",
       "      <td>1250.040359</td>\n",
       "      <td>2490.32</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.516428e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.981656</td>\n",
       "      <td>77188.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.48</th>\n",
       "      <td>1248.995207</td>\n",
       "      <td>2490.50</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.520251e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>5.005936</td>\n",
       "      <td>76350.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.50</th>\n",
       "      <td>1247.394875</td>\n",
       "      <td>2496.60</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.509083e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.981496</td>\n",
       "      <td>75387.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.52</th>\n",
       "      <td>1247.685904</td>\n",
       "      <td>2490.55</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.511552e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>5.007174</td>\n",
       "      <td>74129.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.54</th>\n",
       "      <td>1247.599163</td>\n",
       "      <td>2494.05</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.511347e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.978266</td>\n",
       "      <td>73229.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.56</th>\n",
       "      <td>1245.254046</td>\n",
       "      <td>2487.60</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.501767e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>5.001678</td>\n",
       "      <td>71912.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.58</th>\n",
       "      <td>1248.972293</td>\n",
       "      <td>2492.49</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>2.519986e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>5.013705</td>\n",
       "      <td>71374.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th rowspan=\"30\" valign=\"top\">kdd</th>\n",
       "      <th>0.42</th>\n",
       "      <td>2010.865219</td>\n",
       "      <td>2492.58</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.280224e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.784000</td>\n",
       "      <td>79358.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.44</th>\n",
       "      <td>2018.378420</td>\n",
       "      <td>2491.91</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.333948e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.753638</td>\n",
       "      <td>78580.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.46</th>\n",
       "      <td>2016.470569</td>\n",
       "      <td>2489.03</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.311261e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.759382</td>\n",
       "      <td>77188.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.48</th>\n",
       "      <td>2110.000617</td>\n",
       "      <td>2488.99</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>9.667329e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.859819</td>\n",
       "      <td>76350.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.50</th>\n",
       "      <td>2028.462838</td>\n",
       "      <td>2494.30</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.485506e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.771912</td>\n",
       "      <td>75387.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.52</th>\n",
       "      <td>2018.178445</td>\n",
       "      <td>2487.06</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.322932e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.773000</td>\n",
       "      <td>74129.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.54</th>\n",
       "      <td>2014.426032</td>\n",
       "      <td>2491.42</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.298263e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.773278</td>\n",
       "      <td>73229.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.56</th>\n",
       "      <td>2025.516946</td>\n",
       "      <td>2484.19</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.465919e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.762195</td>\n",
       "      <td>71912.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.58</th>\n",
       "      <td>2022.999825</td>\n",
       "      <td>2490.81</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.397676e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.760543</td>\n",
       "      <td>71374.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.60</th>\n",
       "      <td>2017.724796</td>\n",
       "      <td>2486.85</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.359753e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.760193</td>\n",
       "      <td>70230.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.62</th>\n",
       "      <td>2012.950925</td>\n",
       "      <td>2485.06</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.307176e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.755066</td>\n",
       "      <td>69197.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.64</th>\n",
       "      <td>2028.635698</td>\n",
       "      <td>2493.78</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.524213e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.796645</td>\n",
       "      <td>68207.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.66</th>\n",
       "      <td>2013.842145</td>\n",
       "      <td>2484.67</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.306382e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.757653</td>\n",
       "      <td>67246.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.68</th>\n",
       "      <td>2033.710468</td>\n",
       "      <td>2492.51</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.538477e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.749325</td>\n",
       "      <td>66378.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.70</th>\n",
       "      <td>2052.698079</td>\n",
       "      <td>2486.60</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.845997e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.783153</td>\n",
       "      <td>65270.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.72</th>\n",
       "      <td>11412.333487</td>\n",
       "      <td>2385.32</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.111919e+07</td>\n",
       "      <td>500.5</td>\n",
       "      <td>9.179991</td>\n",
       "      <td>64290.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.74</th>\n",
       "      <td>2049.436901</td>\n",
       "      <td>2490.83</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.762680e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.763525</td>\n",
       "      <td>63358.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.76</th>\n",
       "      <td>2016.724485</td>\n",
       "      <td>2488.88</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.355655e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.760057</td>\n",
       "      <td>62286.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.78</th>\n",
       "      <td>2022.780612</td>\n",
       "      <td>2490.53</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.408137e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.756901</td>\n",
       "      <td>61428.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.80</th>\n",
       "      <td>2018.374091</td>\n",
       "      <td>2486.55</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.367210e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.738119</td>\n",
       "      <td>60195.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.82</th>\n",
       "      <td>2031.895086</td>\n",
       "      <td>2486.45</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.524525e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.764701</td>\n",
       "      <td>59348.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.84</th>\n",
       "      <td>2029.940693</td>\n",
       "      <td>2484.79</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.525097e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.777323</td>\n",
       "      <td>58228.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.86</th>\n",
       "      <td>2034.847593</td>\n",
       "      <td>2490.94</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.583152e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.743863</td>\n",
       "      <td>57457.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.88</th>\n",
       "      <td>2039.486948</td>\n",
       "      <td>2482.82</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.662782e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.764657</td>\n",
       "      <td>56263.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.90</th>\n",
       "      <td>2043.236620</td>\n",
       "      <td>2484.62</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.680866e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.760862</td>\n",
       "      <td>55306.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.92</th>\n",
       "      <td>2036.231877</td>\n",
       "      <td>2487.74</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.584850e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.753367</td>\n",
       "      <td>54337.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.94</th>\n",
       "      <td>2025.389852</td>\n",
       "      <td>2485.82</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.439125e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.768816</td>\n",
       "      <td>53223.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.96</th>\n",
       "      <td>2128.602724</td>\n",
       "      <td>2483.38</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>9.935613e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.854326</td>\n",
       "      <td>52338.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0.98</th>\n",
       "      <td>2036.798656</td>\n",
       "      <td>2481.39</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.590302e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.761638</td>\n",
       "      <td>51392.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1.00</th>\n",
       "      <td>2023.510243</td>\n",
       "      <td>2482.58</td>\n",
       "      <td>2.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>8.434099e+04</td>\n",
       "      <td>500.5</td>\n",
       "      <td>4.759098</td>\n",
       "      <td>50248.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>153 rows × 8 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                  avg_rank  num_events  num_followers    q           r_2  \\\n",
       "type overlap                                                               \n",
       "Opt  0.00      1254.033670     2505.85            2.0  1.0  2.517349e+04   \n",
       "     0.02      1252.434394     2505.76            2.0  1.0  2.506377e+04   \n",
       "     0.04      1251.395308     2503.70            2.0  1.0  2.504060e+04   \n",
       "     0.06      1251.574592     2501.54            2.0  1.0  2.506442e+04   \n",
       "     0.08      1249.800340     2495.19            2.0  1.0  2.497974e+04   \n",
       "     0.10      1253.093386     2502.30            2.0  1.0  2.512201e+04   \n",
       "     0.12      1251.403023     2501.88            2.0  1.0  2.506588e+04   \n",
       "     0.14      1252.533378     2500.30            2.0  1.0  2.511743e+04   \n",
       "     0.16      1252.773870     2500.28            2.0  1.0  2.515505e+04   \n",
       "     0.18      1251.549904     2497.90            2.0  1.0  2.510176e+04   \n",
       "     0.20      1252.595110     2500.88            2.0  1.0  2.515811e+04   \n",
       "     0.22      1250.555901     2497.15            2.0  1.0  2.506842e+04   \n",
       "     0.24      1197.047597     2390.36            2.0  1.0  2.508675e+04   \n",
       "     0.26      1252.587131     2497.95            2.0  1.0  2.521721e+04   \n",
       "     0.28      1252.831044     2498.45            2.0  1.0  2.519594e+04   \n",
       "     0.30      1252.439679     2500.10            2.0  1.0  2.519788e+04   \n",
       "     0.32      1249.529288     2494.15            2.0  1.0  2.508599e+04   \n",
       "     0.34      1249.185017     2493.20            2.0  1.0  2.507339e+04   \n",
       "     0.36      1250.304731     2496.62            2.0  1.0  2.512540e+04   \n",
       "     0.38      1250.670676     2494.14            2.0  1.0  2.514025e+04   \n",
       "     0.40      1249.355560     2495.83            2.0  1.0  2.513381e+04   \n",
       "     0.42      1249.400233     2494.74            2.0  1.0  2.514018e+04   \n",
       "     0.44      1250.563524     2495.70            2.0  1.0  2.517512e+04   \n",
       "     0.46      1250.040359     2490.32            2.0  1.0  2.516428e+04   \n",
       "     0.48      1248.995207     2490.50            2.0  1.0  2.520251e+04   \n",
       "     0.50      1247.394875     2496.60            2.0  1.0  2.509083e+04   \n",
       "     0.52      1247.685904     2490.55            2.0  1.0  2.511552e+04   \n",
       "     0.54      1247.599163     2494.05            2.0  1.0  2.511347e+04   \n",
       "     0.56      1245.254046     2487.60            2.0  1.0  2.501767e+04   \n",
       "     0.58      1248.972293     2492.49            2.0  1.0  2.519986e+04   \n",
       "...                    ...         ...            ...  ...           ...   \n",
       "kdd  0.42      2010.865219     2492.58            2.0  1.0  8.280224e+04   \n",
       "     0.44      2018.378420     2491.91            2.0  1.0  8.333948e+04   \n",
       "     0.46      2016.470569     2489.03            2.0  1.0  8.311261e+04   \n",
       "     0.48      2110.000617     2488.99            2.0  1.0  9.667329e+04   \n",
       "     0.50      2028.462838     2494.30            2.0  1.0  8.485506e+04   \n",
       "     0.52      2018.178445     2487.06            2.0  1.0  8.322932e+04   \n",
       "     0.54      2014.426032     2491.42            2.0  1.0  8.298263e+04   \n",
       "     0.56      2025.516946     2484.19            2.0  1.0  8.465919e+04   \n",
       "     0.58      2022.999825     2490.81            2.0  1.0  8.397676e+04   \n",
       "     0.60      2017.724796     2486.85            2.0  1.0  8.359753e+04   \n",
       "     0.62      2012.950925     2485.06            2.0  1.0  8.307176e+04   \n",
       "     0.64      2028.635698     2493.78            2.0  1.0  8.524213e+04   \n",
       "     0.66      2013.842145     2484.67            2.0  1.0  8.306382e+04   \n",
       "     0.68      2033.710468     2492.51            2.0  1.0  8.538477e+04   \n",
       "     0.70      2052.698079     2486.60            2.0  1.0  8.845997e+04   \n",
       "     0.72     11412.333487     2385.32            2.0  1.0  1.111919e+07   \n",
       "     0.74      2049.436901     2490.83            2.0  1.0  8.762680e+04   \n",
       "     0.76      2016.724485     2488.88            2.0  1.0  8.355655e+04   \n",
       "     0.78      2022.780612     2490.53            2.0  1.0  8.408137e+04   \n",
       "     0.80      2018.374091     2486.55            2.0  1.0  8.367210e+04   \n",
       "     0.82      2031.895086     2486.45            2.0  1.0  8.524525e+04   \n",
       "     0.84      2029.940693     2484.79            2.0  1.0  8.525097e+04   \n",
       "     0.86      2034.847593     2490.94            2.0  1.0  8.583152e+04   \n",
       "     0.88      2039.486948     2482.82            2.0  1.0  8.662782e+04   \n",
       "     0.90      2043.236620     2484.62            2.0  1.0  8.680866e+04   \n",
       "     0.92      2036.231877     2487.74            2.0  1.0  8.584850e+04   \n",
       "     0.94      2025.389852     2485.82            2.0  1.0  8.439125e+04   \n",
       "     0.96      2128.602724     2483.38            2.0  1.0  9.935613e+04   \n",
       "     0.98      2036.798656     2481.39            2.0  1.0  8.590302e+04   \n",
       "     1.00      2023.510243     2482.58            2.0  1.0  8.434099e+04   \n",
       "\n",
       "               seed     top_1  world_events  \n",
       "type overlap                                 \n",
       "Opt  0.00     500.5  5.002597      100486.0  \n",
       "     0.02     500.5  4.967638       99259.0  \n",
       "     0.04     500.5  4.990877       98112.0  \n",
       "     0.06     500.5  4.999216       97067.0  \n",
       "     0.08     500.5  5.017915       95837.0  \n",
       "     0.10     500.5  4.984509       95373.0  \n",
       "     0.12     500.5  4.973182       94221.0  \n",
       "     0.14     500.5  4.985901       93254.0  \n",
       "     0.16     500.5  4.997922       92373.0  \n",
       "     0.18     500.5  4.994623       91228.0  \n",
       "     0.20     500.5  4.983969       90438.0  \n",
       "     0.22     500.5  4.996334       89199.0  \n",
       "     0.24     500.5  6.066430       88358.0  \n",
       "     0.26     500.5  4.994079       87430.0  \n",
       "     0.28     500.5  4.966719       86280.0  \n",
       "     0.30     500.5  4.956781       85399.0  \n",
       "     0.32     500.5  4.995153       84095.0  \n",
       "     0.34     500.5  4.987309       83149.0  \n",
       "     0.36     500.5  4.965397       82181.0  \n",
       "     0.38     500.5  4.961991       81397.0  \n",
       "     0.40     500.5  4.984861       80220.0  \n",
       "     0.42     500.5  5.017429       79358.0  \n",
       "     0.44     500.5  4.983700       78580.0  \n",
       "     0.46     500.5  4.981656       77188.0  \n",
       "     0.48     500.5  5.005936       76350.0  \n",
       "     0.50     500.5  4.981496       75387.0  \n",
       "     0.52     500.5  5.007174       74129.0  \n",
       "     0.54     500.5  4.978266       73229.0  \n",
       "     0.56     500.5  5.001678       71912.0  \n",
       "     0.58     500.5  5.013705       71374.0  \n",
       "...             ...       ...           ...  \n",
       "kdd  0.42     500.5  4.784000       79358.0  \n",
       "     0.44     500.5  4.753638       78580.0  \n",
       "     0.46     500.5  4.759382       77188.0  \n",
       "     0.48     500.5  4.859819       76350.0  \n",
       "     0.50     500.5  4.771912       75387.0  \n",
       "     0.52     500.5  4.773000       74129.0  \n",
       "     0.54     500.5  4.773278       73229.0  \n",
       "     0.56     500.5  4.762195       71912.0  \n",
       "     0.58     500.5  4.760543       71374.0  \n",
       "     0.60     500.5  4.760193       70230.0  \n",
       "     0.62     500.5  4.755066       69197.0  \n",
       "     0.64     500.5  4.796645       68207.0  \n",
       "     0.66     500.5  4.757653       67246.0  \n",
       "     0.68     500.5  4.749325       66378.0  \n",
       "     0.70     500.5  4.783153       65270.0  \n",
       "     0.72     500.5  9.179991       64290.0  \n",
       "     0.74     500.5  4.763525       63358.0  \n",
       "     0.76     500.5  4.760057       62286.0  \n",
       "     0.78     500.5  4.756901       61428.0  \n",
       "     0.80     500.5  4.738119       60195.0  \n",
       "     0.82     500.5  4.764701       59348.0  \n",
       "     0.84     500.5  4.777323       58228.0  \n",
       "     0.86     500.5  4.743863       57457.0  \n",
       "     0.88     500.5  4.764657       56263.0  \n",
       "     0.90     500.5  4.760862       55306.0  \n",
       "     0.92     500.5  4.753367       54337.0  \n",
       "     0.94     500.5  4.768816       53223.0  \n",
       "     0.96     500.5  4.854326       52338.0  \n",
       "     0.98     500.5  4.761638       51392.0  \n",
       "     1.00     500.5  4.759098       50248.0  \n",
       "\n",
       "[153 rows x 8 columns]"
      ]
     },
     "execution_count": 128,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "overlap_df.groupby(['type', 'overlap']).mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjgAAAFcCAYAAADF+MQnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nXYcBDhEREXUdBjhERETUdf5/9iTenlJu6oEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2376dba780>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_overlapping_perf(overlap_df, measure='avg_rank')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAjgAAAFcCAYAAADF+MQnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQgghRMeRAEcIIYQQHef/AY04/e1mbTS1AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f2376dbaa20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_overlapping_perf(overlap_df, measure='top_1')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f226f21f2e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%run -i redqueen/utils.py\n",
    "%run -i redqueen/opt_model.py\n",
    "%run -i redqueen/opt_runs.py"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Testing out significance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sim_opts = SimOpts(src_id=1, \n",
    "                  s=np.asarray([1.0, 1.0]),\n",
    "                  q=1.0, \n",
    "                  end_time=100,\n",
    "                  other_sources=[('Poisson2', {'src_id': 1000, 'seed': 42, 'rate': 10.0}), \n",
    "                                 ('Poisson2', {'src_id': 1001, 'seed': 43, 'rate': 10.0})],\n",
    "                  sink_ids = [5001, 5002],\n",
    "                  edge_list=[(1, 5001), (1, 5002), (1000, 5001), (1001, 5002)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 383 ms, sys: 40 ms, total: 423 ms\n",
      "Wall time: 420 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "opt_pw_mgr = sim_opts.create_manager_with_significance(1, time_period=10.0, num_segments=24)\n",
    "opt_pw_mgr.run_dynamic()\n",
    "\n",
    "opt_mgr = sim_opts.create_manager_with_opt(1)\n",
    "opt_mgr.run_dynamic()\n",
    "\n",
    "opt_pw_df = opt_pw_mgr.state.get_dataframe()\n",
    "opt_df = opt_mgr.state.get_dataframe()\n",
    "\n",
    "poisson_mgr = sim_opts.create_manager_with_poisson(45, capacity=num_tweets_of(opt_df, sim_opts=sim_opts))\n",
    "poisson_mgr.run_dynamic()\n",
    "\n",
    "poisson_df = poisson_mgr.state.get_dataframe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "324.0 325.0 312.0\n",
      "(2592, 5) (2594, 5) (2568, 5)\n"
     ]
    }
   ],
   "source": [
    "print(num_tweets_of(opt_df, sim_opts=sim_opts), \n",
    "      num_tweets_of(opt_pw_df, sim_opts=sim_opts),\n",
    "      num_tweets_of(poisson_df, sim_opts=sim_opts))\n",
    "print(opt_df.shape, opt_pw_df.shape, poisson_df.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The optimal and optimal-picewise should have similar performance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "30.6505733464 29.6750640797 23.6733687659\n",
      "172.614283682 171.548425075 316.374108601\n"
     ]
    }
   ],
   "source": [
    "print(time_in_top_k(opt_df, K=1, sim_opts=sim_opts), \n",
    "      time_in_top_k(opt_pw_df, sim_opts=sim_opts, K=1), \n",
    "      time_in_top_k(poisson_df, sim_opts=sim_opts, K=1))\n",
    "print(average_rank(opt_df, sim_opts=sim_opts), \n",
    "      average_rank(opt_pw_df, sim_opts=sim_opts, K=1), \n",
    "      average_rank(poisson_df, sim_opts=sim_opts, K=1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Change importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sim_opts_unequal = sim_opts.update({'s': np.asarray([0.5, 1.5])})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 333 ms, sys: 13.3 ms, total: 347 ms\n",
      "Wall time: 337 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "opt_pw_mgr = sim_opts_unequal.create_manager_with_significance(41, time_period=10.0, num_segments=24)\n",
    "opt_pw_mgr.run_dynamic()\n",
    "\n",
    "opt_mgr = sim_opts_unequal.create_manager_with_opt(1)\n",
    "opt_mgr.run_dynamic()\n",
    "\n",
    "opt_pw_df = opt_pw_mgr.state.get_dataframe()\n",
    "opt_df = opt_mgr.state.get_dataframe()\n",
    "\n",
    "poisson_mgr = sim_opts_unequal.create_manager_with_poisson(4, capacity=num_tweets_of(opt_df, sim_opts=sim_opts))\n",
    "poisson_mgr.run_dynamic()\n",
    "\n",
    "poisson_df = poisson_mgr.state.get_dataframe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "325.0 323.0 336.0\n",
      "(2594, 5) (2590, 5) (2616, 5)\n"
     ]
    }
   ],
   "source": [
    "print(num_tweets_of(opt_df, sim_opts=sim_opts_unequal), \n",
    "      num_tweets_of(opt_pw_df, sim_opts=sim_opts_unequal), \n",
    "      num_tweets_of(poisson_df, sim_opts=sim_opts_unequal))\n",
    "print(opt_df.shape, opt_pw_df.shape, poisson_df.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "30.6124782772 30.2860675318 24.8904723685\n",
      "171.651501755 172.651153777 299.409321481\n"
     ]
    }
   ],
   "source": [
    "print(time_in_top_k(opt_df, K=1, sim_opts=sim_opts), \n",
    "      time_in_top_k(opt_pw_df, sim_opts=sim_opts, K=1), \n",
    "      time_in_top_k(poisson_df, sim_opts=sim_opts, K=1))\n",
    "print(average_rank(opt_df, sim_opts=sim_opts), \n",
    "      average_rank(opt_pw_df, sim_opts=sim_opts, K=1), \n",
    "      average_rank(poisson_df, sim_opts=sim_opts, K=1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Setting significance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "num_segs = 24\n",
    "sig = np.square(np.sin(np.arange(0, num_segs, step=1.0) * 4 * np.pi / num_segs))\n",
    "sig = (sig / sig.sum()).reshape((1, -1))\n",
    "sim_opts = SimOpts(src_id=1, \n",
    "                  s=1.0,\n",
    "                  q=1.0,\n",
    "                  end_time=100,\n",
    "                  other_sources=[('Poisson2', {'src_id': 1000, 'seed': 42, 'rate': 500.0})],\n",
    "                  sink_ids = [5001],\n",
    "                  edge_list=[(1, 5001), (1000, 5001)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4.82 s, sys: 133 ms, total: 4.95 s\n",
      "Wall time: 4.94 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "opt_sig_mgr = sim_opts.create_manager_with_significance(seed=10, time_period=sim_opts.end_time, \n",
    "                                                        significance=sig)\n",
    "opt_sig_mgr.run_dynamic()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df = opt_sig_mgr.state.get_dataframe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 850,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f1fce910518>]"
      ]
     },
     "execution_count": 850,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeUAAAEyCAYAAADEC8j1AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nnDJBEARBcAI5ZYIgCILghP8fkSJ9k1XN75cAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1fec916940>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(np.sqrt(sig.ravel()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 853,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAeAAAAEyCAYAAAAiIgOxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngAEAMIAABgDAgP8DpROsgzaRDmMAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1fce8644a8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(df.t[df.src_id == sim_opts.src_id].unique(), bins=num_segs);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## `s_init` heuristic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 287,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sim_opts = SimOpts.std_poisson(world_seed=1, world_rate=500)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 288,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2016-05-13 16:06:06.308683\u001b[0m Checkpoint: Initial capacity = 16.682815756349004, target capacity = 10\n",
      "*** \u001b[31m2016-05-13 16:06:06.515228\u001b[0m Checkpoint: s = 2.0, capcity = 14.461590479654564\n",
      "*** \u001b[31m2016-05-13 16:06:06.711105\u001b[0m Checkpoint: s = 4.0, capcity = 12.150333853228608\n",
      "*** \u001b[31m2016-05-13 16:06:06.903814\u001b[0m Checkpoint: s = 8.0, capcity = 10.439782962439807\n",
      "*** \u001b[31m2016-05-13 16:06:07.116908\u001b[0m Checkpoint: s = 16.0, capcity = 8.877513076274898\n",
      "*** \u001b[31m2016-05-13 16:06:07.118326\u001b[0m Checkpoint: s_hi = 16.0, s_lo = 8.0\n",
      "*** \u001b[31m2016-05-13 16:06:07.335871\u001b[0m Checkpoint: new_capacity = 9.554880390811231, s = 12.0\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "12.0"
      ]
     },
     "execution_count": 288,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sweep_s(lambda: sim_opts, 10, verbose=True, dynamic=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 289,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<__main__.Manager at 0x7fe76eb27780>"
      ]
     },
     "execution_count": 289,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "opt_mgr = sim_opts.update({ 's': 0.375 }).create_manager_with_opt(seed=1)\n",
    "opt_mgr.run_dynamic()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 290,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "opt_df = opt_mgr.state.get_dataframe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 291,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>event_id</th>\n",
       "      <th>sink_id</th>\n",
       "      <th>src_id</th>\n",
       "      <th>t</th>\n",
       "      <th>time_delta</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>100</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>104</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.002940</td>\n",
       "      <td>0.000070</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>132</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.035653</td>\n",
       "      <td>0.001081</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49</th>\n",
       "      <td>149</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.077264</td>\n",
       "      <td>0.005290</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>60</th>\n",
       "      <td>160</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.104315</td>\n",
       "      <td>0.001089</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>89</th>\n",
       "      <td>189</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.155652</td>\n",
       "      <td>0.002980</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>105</th>\n",
       "      <td>205</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.190397</td>\n",
       "      <td>0.001760</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>118</th>\n",
       "      <td>218</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.214218</td>\n",
       "      <td>0.002258</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>137</th>\n",
       "      <td>237</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.258508</td>\n",
       "      <td>0.001389</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>149</th>\n",
       "      <td>249</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.275824</td>\n",
       "      <td>0.005616</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>160</th>\n",
       "      <td>260</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.310070</td>\n",
       "      <td>0.007738</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>184</th>\n",
       "      <td>284</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.354171</td>\n",
       "      <td>0.005273</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>207</th>\n",
       "      <td>307</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.413325</td>\n",
       "      <td>0.002514</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>215</th>\n",
       "      <td>315</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.422783</td>\n",
       "      <td>0.001675</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>237</th>\n",
       "      <td>337</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.465730</td>\n",
       "      <td>0.000020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>257</th>\n",
       "      <td>357</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.521879</td>\n",
       "      <td>0.002528</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>268</th>\n",
       "      <td>368</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.538687</td>\n",
       "      <td>0.000060</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>335</th>\n",
       "      <td>435</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.655294</td>\n",
       "      <td>0.001156</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>360</th>\n",
       "      <td>460</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.706507</td>\n",
       "      <td>0.002990</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>386</th>\n",
       "      <td>486</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.753274</td>\n",
       "      <td>0.000518</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>396</th>\n",
       "      <td>496</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.782481</td>\n",
       "      <td>0.008607</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>422</th>\n",
       "      <td>522</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.824248</td>\n",
       "      <td>0.000238</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>434</th>\n",
       "      <td>534</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.847236</td>\n",
       "      <td>0.000407</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>453</th>\n",
       "      <td>553</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.884253</td>\n",
       "      <td>0.000947</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>472</th>\n",
       "      <td>572</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.909667</td>\n",
       "      <td>0.000031</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>511</th>\n",
       "      <td>611</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.978686</td>\n",
       "      <td>0.001927</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>514</th>\n",
       "      <td>614</td>\n",
       "      <td>1001</td>\n",
       "      <td>1</td>\n",
       "      <td>0.988569</td>\n",
       "      <td>0.005752</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     event_id  sink_id  src_id         t  time_delta\n",
       "0         100     1001       1  0.000000    0.000000\n",
       "4         104     1001       1  0.002940    0.000070\n",
       "32        132     1001       1  0.035653    0.001081\n",
       "49        149     1001       1  0.077264    0.005290\n",
       "60        160     1001       1  0.104315    0.001089\n",
       "89        189     1001       1  0.155652    0.002980\n",
       "105       205     1001       1  0.190397    0.001760\n",
       "118       218     1001       1  0.214218    0.002258\n",
       "137       237     1001       1  0.258508    0.001389\n",
       "149       249     1001       1  0.275824    0.005616\n",
       "160       260     1001       1  0.310070    0.007738\n",
       "184       284     1001       1  0.354171    0.005273\n",
       "207       307     1001       1  0.413325    0.002514\n",
       "215       315     1001       1  0.422783    0.001675\n",
       "237       337     1001       1  0.465730    0.000020\n",
       "257       357     1001       1  0.521879    0.002528\n",
       "268       368     1001       1  0.538687    0.000060\n",
       "335       435     1001       1  0.655294    0.001156\n",
       "360       460     1001       1  0.706507    0.002990\n",
       "386       486     1001       1  0.753274    0.000518\n",
       "396       496     1001       1  0.782481    0.008607\n",
       "422       522     1001       1  0.824248    0.000238\n",
       "434       534     1001       1  0.847236    0.000407\n",
       "453       553     1001       1  0.884253    0.000947\n",
       "472       572     1001       1  0.909667    0.000031\n",
       "511       611     1001       1  0.978686    0.001927\n",
       "514       614     1001       1  0.988569    0.005752"
      ]
     },
     "execution_count": 291,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "opt_df[opt_df.src_id == sim_opts.src_id]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 292,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "wall_mgr = sim_opts.create_manager_for_wall()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 293,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<__main__.Manager at 0x7fe76cf71898>"
      ]
     },
     "execution_count": 293,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "wall_mgr.run_dynamic()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 294,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df = wall_mgr.state.get_dataframe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 295,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "r_t = rank_of_src_in_df(df, src_id=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 296,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "495.0"
      ]
     },
     "execution_count": 296,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r_t.mean(1).iloc[-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 297,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(495, 5)"
      ]
     },
     "execution_count": 297,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 298,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "6.7826577479581491"
      ]
     },
     "execution_count": 298,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(4 * (r_t.iloc[-1].mean() ** 2) * (sim_opts.end_time) ** 2) / (np.pi * np.pi * (10 + 1) ** 4) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'avg_rank': 954.00681957353299,\n",
       " 'broadcaster_events': 49,\n",
       " 'num_events': 49,\n",
       " 'r0_num_events': 49,\n",
       " 'r_2': 13966.614080745609,\n",
       " 's': 1000.0,\n",
       " 'seed': 2,\n",
       " 'sim_opts': <__main__.SimOpts at 0x7fc80895c358>,\n",
       " 'top_1': 9.5668178314961985,\n",
       " 'top_10': 52.303469506298647,\n",
       " 'top_20': 90.47790271111036,\n",
       " 'top_3': 20.567177074149754,\n",
       " 'top_5': 30.583811352029109,\n",
       " 'type': 'Oracle',\n",
       " 'world_events': 1198}"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hawkes_res.raw_results[-1\n",
    "                      ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[]"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[x['best_avg_k'] for x in hawkes_res.raw_results if x['type'] == 'KDD']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Run for user 52885404"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "%run -i redqueen/read_real_data.py\n",
    "%run -i redqueen/utils.py\n",
    "%run -i redqueen/opt_model.py\n",
    "%run -i redqueen/opt_runs.py"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "base = '/NL/redqueen/work/opt-broadcast/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "user_id = 20712572"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# user_id = 35652801 s_opt = 3211150360.2065687\n",
    "# s_opt = 155655.86986992834 for user_id = 52885404\n",
    "user_raw = pickle.load(open(os.path.join(base, 'user-{}.pickle'.format(user_id)), 'rb'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "user_result = revert_to_sim_opts(user_raw)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['num_user_events', 'sim_opts_dict', 'user_id', 'user_event_times', 'scaled_period'])"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_raw.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sim_opts = user_result['sim_opts']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2018-03-29 15:41:10.277148\u001b[0m Checkpoint: q_init = 23432203434.13262\n",
      "*** \u001b[31m2018-03-29 15:42:03.589664\u001b[0m Checkpoint: Initial capacity = 28.884668978257146, target capacity = 30, q_init = 23432203434.13262\n",
      "*** \u001b[31m2018-03-29 15:42:56.022663\u001b[0m Checkpoint: q = 11716101717.06631, capacity = 35.03962393774085\n",
      "*** \u001b[31m2018-03-29 15:42:56.024860\u001b[0m Checkpoint: q_hi = 23432203434.13262, q_lo = 11716101717.06631\n",
      "*** \u001b[31m2018-03-29 15:43:49.334111\u001b[0m Checkpoint: new_capacity = 31.287997490988705, q = 17574152575.599464\n",
      "*** \u001b[31m2018-03-29 15:44:42.311992\u001b[0m Checkpoint: new_capacity = 29.89889301924032, q = 20503178004.866043\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "q_opt = sweep_q(sim_opts, user_raw['num_user_events'], verbose=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "20503178004.866043"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "q_opt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "opt_mgr = sim_opts.update({'q': q_opt}).create_manager_with_opt(seed=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 42.5 s, sys: 1.12 s, total: 43.6 s\n",
      "Wall time: 43.6 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<__main__.Manager at 0x7fe8340aba58>"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "opt_mgr.run_dynamic()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df_opt = opt_mgr.state.get_dataframe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(20712572, 20712572)"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_id, sim_opts.src_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "real_mgr = sim_opts.update({'q': q_opt}).create_manager_with_times(event_times=user_result['user_event_times'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 8.61 s, sys: 1.2 s, total: 9.81 s\n",
      "Wall time: 9.8 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<__main__.Manager at 0x7fe834104198>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "real_mgr.run_dynamic()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df_real = real_mgr.state.get_dataframe()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "32"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_opt.event_id[df_opt.src_id == user_id].nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "30"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_real.event_id[df_real.src_id == user_id].nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "ename": "IndexError",
     "evalue": "only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mIndexError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast-online/redqueen/opt_runs.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mkdd_perf\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mworker_kdd\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0muser_result\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'num_user_events'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0;36m1.0\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m60\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msim_opts\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mverbose\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mKs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast-online/redqueen/opt_runs.py\u001b[0m in \u001b[0;36mworker_kdd\u001b[0;34m(params, verbose, Ks)\u001b[0m\n\u001b[1;32m    232\u001b[0m             \u001b[0mx0\u001b[0m           \u001b[0;34m=\u001b[0m \u001b[0mx0\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    233\u001b[0m             \u001b[0mverbose\u001b[0m      \u001b[0;34m=\u001b[0m \u001b[0mverbose\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 234\u001b[0;31m             \u001b[0mwith_iter\u001b[0m    \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    235\u001b[0m         )\n\u001b[1;32m    236\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast_ref/broadcast/opt/optimizer.py\u001b[0m in \u001b[0;36moptimize\u001b[0;34m(util, util_grad, budget, upper_bounds, threshold, x0, verbose, with_iter)\u001b[0m\n\u001b[1;32m     99\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0mupper_bounds\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    100\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 101\u001b[0;31m     \u001b[0mopt_rates\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0moptimize_base\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mutil\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mutil_grad\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mproj\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mx0\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mthreshold\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mverbose\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mverbose\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mwith_iter\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mwith_iter\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    102\u001b[0m     \u001b[0;31m# delta = int(round(time.time() * 1000)) - start\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    103\u001b[0m     \u001b[0;31m# sys.stderr.write('Total time: %d' % delta)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast_ref/broadcast/opt/optimizer.py\u001b[0m in \u001b[0;36moptimize_base\u001b[0;34m(util, grad, proj, x0, threshold, gamma, c, verbose, with_iter)\u001b[0m\n\u001b[1;32m     58\u001b[0m         \u001b[0;31m# if i % 10 == 0:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     59\u001b[0m         \u001b[0;31m#     print(x)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 60\u001b[0;31m         \u001b[0mg\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mgrad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     61\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     62\u001b[0m         \u001b[0md\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mproj\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mg\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0;36m1000.\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mx\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast-online/redqueen/opt_runs.py\u001b[0m in \u001b[0;36m_util_grad\u001b[0;34m(x)\u001b[0m\n\u001b[1;32m    219\u001b[0m                                                       \u001b[0mfollower_wall_intensities\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    220\u001b[0m                                                       \u001b[0mfollower_conn_prob\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 221\u001b[0;31m                                                       follower_weights)\n\u001b[0m\u001b[1;32m    222\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    223\u001b[0m         \u001b[0;31m# Initial guess is close to Poisson solution\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mwork/prog/broadcast_ref/broadcast/opt/utils.pyx\u001b[0m in \u001b[0;36mbroadcast.opt.utils.weighted_top_one_grad (/home/utkarshu/.pyxbld/temp.linux-x86_64-3.6/pyrex/broadcast/opt/utils.c:6169)\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32mwork/prog/broadcast_ref/broadcast/opt/utils.pyx\u001b[0m in \u001b[0;36mbroadcast.opt.utils.gradient_top_one (/home/utkarshu/.pyxbld/temp.linux-x86_64-3.6/pyrex/broadcast/opt/utils.c:5454)\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;31mIndexError\u001b[0m: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices"
     ]
    }
   ],
   "source": [
    "kdd_perf = worker_kdd((1, user_result['num_user_events'] * 1.0, 60, sim_opts, None, None), verbose=True, Ks=[1])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "ename": "IndexError",
     "evalue": "only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mIndexError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast-online/redqueen/opt_runs.py\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0;31m# seed, capacity, num_segments, sim_opts, world_changing_rates, queue = params\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mkdd_perf\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mworker_kdd\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0muser_result\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'num_user_events'\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0;36m1.0\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;36m60\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0msim_opts\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mverbose\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mTrue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mKs\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      3\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mkdd_opt\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mkdd_perf\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m'kdd_opt_1'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast-online/redqueen/opt_runs.py\u001b[0m in \u001b[0;36mworker_kdd\u001b[0;34m(params, verbose, Ks)\u001b[0m\n\u001b[1;32m    232\u001b[0m             \u001b[0mx0\u001b[0m           \u001b[0;34m=\u001b[0m \u001b[0mx0\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    233\u001b[0m             \u001b[0mverbose\u001b[0m      \u001b[0;34m=\u001b[0m \u001b[0mverbose\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 234\u001b[0;31m             \u001b[0mwith_iter\u001b[0m    \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    235\u001b[0m         )\n\u001b[1;32m    236\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast_ref/broadcast/opt/optimizer.py\u001b[0m in \u001b[0;36moptimize\u001b[0;34m(util, util_grad, budget, upper_bounds, threshold, x0, verbose, with_iter)\u001b[0m\n\u001b[1;32m     99\u001b[0m             \u001b[0;32mreturn\u001b[0m \u001b[0mupper_bounds\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    100\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 101\u001b[0;31m     \u001b[0mopt_rates\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0moptimize_base\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mutil\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mutil_grad\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mproj\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mx0\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mthreshold\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mverbose\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mverbose\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mwith_iter\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mwith_iter\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    102\u001b[0m     \u001b[0;31m# delta = int(round(time.time() * 1000)) - start\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    103\u001b[0m     \u001b[0;31m# sys.stderr.write('Total time: %d' % delta)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast_ref/broadcast/opt/optimizer.py\u001b[0m in \u001b[0;36moptimize_base\u001b[0;34m(util, grad, proj, x0, threshold, gamma, c, verbose, with_iter)\u001b[0m\n\u001b[1;32m     58\u001b[0m         \u001b[0;31m# if i % 10 == 0:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     59\u001b[0m         \u001b[0;31m#     print(x)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 60\u001b[0;31m         \u001b[0mg\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mgrad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     61\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     62\u001b[0m         \u001b[0md\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mproj\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mg\u001b[0m \u001b[0;34m*\u001b[0m \u001b[0;36m1000.\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m-\u001b[0m \u001b[0mx\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m/NL/stackexchange/work/prog/broadcast-online/redqueen/opt_runs.py\u001b[0m in \u001b[0;36m_util_grad\u001b[0;34m(x)\u001b[0m\n\u001b[1;32m    219\u001b[0m                                                       \u001b[0mfollower_wall_intensities\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    220\u001b[0m                                                       \u001b[0mfollower_conn_prob\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 221\u001b[0;31m                                                       follower_weights)\n\u001b[0m\u001b[1;32m    222\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    223\u001b[0m         \u001b[0;31m# Initial guess is close to Poisson solution\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mwork/prog/broadcast_ref/broadcast/opt/utils.pyx\u001b[0m in \u001b[0;36mbroadcast.opt.utils.weighted_top_one_grad (/home/utkarshu/.pyxbld/temp.linux-x86_64-3.6/pyrex/broadcast/opt/utils.c:6169)\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;32mwork/prog/broadcast_ref/broadcast/opt/utils.pyx\u001b[0m in \u001b[0;36mbroadcast.opt.utils.gradient_top_one (/home/utkarshu/.pyxbld/temp.linux-x86_64-3.6/pyrex/broadcast/opt/utils.c:5454)\u001b[0;34m()\u001b[0m\n",
      "\u001b[0;31mIndexError\u001b[0m: only integers, slices (`:`), ellipsis (`...`), numpy.newaxis (`None`) and integer or boolean arrays are valid indices"
     ]
    }
   ],
   "source": [
    "# seed, capacity, num_segments, sim_opts, world_changing_rates, queue = params\n",
    "kdd_perf = worker_kdd((1, user_result['num_user_events'] * 1.0, 60, sim_opts, None, None), verbose=True, Ks=[1])\n",
    "\n",
    "kdd_opt = kdd_perf['kdd_opt_1']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "r_opt = rank_of_src_in_df(df_opt, user_id).mean(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "r_real = rank_of_src_in_df(df_real, user_id).mean(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1239,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(606525,)"
      ]
     },
     "execution_count": 1239,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r_opt.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1240,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "r_opt_sampled = pd.concat([r_opt[::200], r_opt[r_opt == 0]]).sort_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1241,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "32"
      ]
     },
     "execution_count": 1241,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.sum(r_opt == 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1242,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([  251.906,   441.235,   455.121,   886.766,  1413.65 ,  1429.583,\n",
       "        1700.907,  2508.858,  2509.022,  2706.411,  3645.805,  3648.546,\n",
       "        3993.621,  4269.935,  4305.957,  4306.243,  4309.272,  4313.168,\n",
       "        4325.23 ,  4662.313,  5412.7  ,  5415.166,  5790.916,  5791.006,\n",
       "        5799.317,  5809.817,  6078.444,  6099.087,  6207.644,  6854.717])"
      ]
     },
     "execution_count": 1242,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_real.t[df_real.src_id == user_id].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1243,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "t\n",
       "886.762619    775.066667\n",
       "886.766353      0.016667\n",
       "dtype: float64"
      ]
     },
     "execution_count": 1243,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r_real[(886.76 <= r_real.index) & (r_real.index <= 886.77)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1244,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Float64Index([ 251.90598865, 441.235439068, 455.120594385, 1413.64994026,\n",
       "              1700.90725806, 2508.85790024, 2509.02217742, 3645.80533154,\n",
       "              3993.62119176,  4269.9354092, 4305.95691458, 4306.24253286,\n",
       "              4309.27232676, 4313.16830944,  4325.2296147, 4662.31332139,\n",
       "              5412.69974612, 5415.16577061, 5790.91621864, 5791.00582437,\n",
       "              5799.31675627, 5809.81742832, 6207.64448925, 6854.71736858],\n",
       "             dtype='float64', name='t')"
      ]
     },
     "execution_count": 1244,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r_real[(r_real == 0)].index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1245,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "r_real_sampled = pd.concat([r_real[::200], r_real[r_real == 0]]).sort_index()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The average rank of user among his followers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1246,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f200f673e80>]"
      ]
     },
     "execution_count": 1246,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAegAAAFtCAYAAADMLflnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1f3d9f1908>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "latexify(fig_width=2.0, largeFonts=True)\n",
    "plt.figure(figsize=(8,6))\n",
    "plt.plot(60 * r_real_sampled.index / 10000, r_real_sampled.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1247,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "c1, c2, c3, c4 = sns.mpl_palette('Set2', 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1470,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAH0AAABQCAYAAAA0snrNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1ec4a0e8d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "latexify(fig_width=1.75, largeFonts=False)\n",
    "# plt.figure(figsize=(8,6))\n",
    "# plt.plot(60 * r_opt_sampled.index / 10000, r_opt_sampled.values, color=c1, linewidth=0.75)\n",
    "one_month = r_opt_sampled.index <= 5000.0\n",
    "ymin, ymax = 0, 3000\n",
    "\n",
    "plt.fill_between(60 * r_opt_sampled.index[one_month] / 10000, r_opt_sampled.values[one_month], color=c1,\n",
    "                 label=r'$\\bar{r}(t)$',\n",
    "                 alpha=0.5)\n",
    "\n",
    "tweet_times = df_opt[(df_opt.src_id == sim_opts.src_id)].t.unique()\n",
    "one_month_df = tweet_times < 5000.0\n",
    "num_tweets = np.sum(one_month_df)\n",
    "\n",
    "# plt.vlines(tweet_times[one_month_df] * 60 / 10000, ymin=0, ymax=1200, color=c2, linewidth=0.5)\n",
    "plt.step(tweet_times[one_month_df] * 60 / 10000.0, np.arange(num_tweets) / num_tweets * ymax, \n",
    "         label='$N(t)$',\n",
    "         color=c2, linewidth=1.0)\n",
    "\n",
    "# plt.legend(loc=\"upper left\")\n",
    "plt.legend(bbox_to_anchor=(0.5, 1.2), bbox_transform=plt.gca().transAxes)\n",
    "\n",
    "plt.yticks([0, 1500, 3000])\n",
    "# plt.yticks([])\n",
    "plt.xticks([3, 15, 27], ['01/06', '15/06', '31/06'])\n",
    "plt.tight_layout()\n",
    "format_axes(plt.gca())\n",
    "plt.gca().xaxis.set_ticks_position('none')\n",
    "# plt.savefig(os.path.join(plot_base, 'opt-data-counting.pdf'), bbox_inches='tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1465,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([], dtype=float64)"
      ]
     },
     "execution_count": 1465,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "tweet_times"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 898,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f1fc99e7668>"
      ]
     },
     "execution_count": 898,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAH0AAABNCAYAAACL+muIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nwlbnjmWdAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1fcee65358>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "latexify(fig_width=1.75, largeFonts=False)\n",
    "# plt.figure(figsize=(8,6))\n",
    "# plt.plot(60 * r_opt_sampled.index / 10000, r_opt_sampled.values, color=c1, linewidth=0.75)\n",
    "one_month = r_opt_sampled.index <= 5000.0\n",
    "plt.fill_between(60 * r_opt_sampled.index[one_month] / 10000, r_opt_sampled.values[one_month], color=c1, alpha=0.5)\n",
    "\n",
    "tweet_times = df_opt[(df_opt.src_id == user_id)].t.unique()\n",
    "one_month_df = tweet_times < 5000.0\n",
    "ymin, ymax = 0, 1200\n",
    "plt.vlines(tweet_times[one_month_df] * 60 / 10000, ymin=0, ymax=1200, color=c2, linewidth=0.5)\n",
    "plt.yticks([0, 600, 1200])\n",
    "plt.xticks([3, 15, 27], ['01/06', '15/06', '31/06'])\n",
    "plt.tight_layout()\n",
    "format_axes(plt.gca())\n",
    "# plt.savefig(os.path.join(plot_base, 'opt-data-ranks.pdf'), bbox_inches='tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 899,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "20"
      ]
     },
     "execution_count": 899,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.sum(one_month_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1469,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAH0AAABQCAYAAAA0snrNAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1ec4ddaef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "latexify(fig_width=1.75, largeFonts=False)\n",
    "# plt.figure(figsize=(8,6))\n",
    "# plt.plot(60 * r_opt_sampled.index / 10000, r_opt_sampled.values, color=c1, linewidth=0.75)\n",
    "one_month = r_real_sampled.index <= 5000.0\n",
    "plt.fill_between(60 * r_real_sampled.index[one_month] / 10000, r_real_sampled.values[one_month], \n",
    "                 label=r'$\\bar{r}(t)$', color=c1, alpha=0.5)\n",
    "\n",
    "tweet_times = df_real[(df_real.src_id == sim_opts.src_id)].t.unique()\n",
    "one_month_df = tweet_times < 5000.0\n",
    "num_tweets = np.sum(one_month_df)\n",
    "ymin, ymax = 0, 3000\n",
    "# plt.vlines(tweet_times[one_month_df] * 60 / 10000, ymin=0, ymax=1200, color=c2, linewidth=0.5)\n",
    "plt.step(tweet_times[one_month_df] * 60 / 10000.0, np.arange(num_tweets) / num_tweets * ymax, color=c2,\n",
    "         label=r'$N(t)$',\n",
    "         linewidth=1.0)\n",
    "\n",
    "plt.yticks([0, 1500, 3000])\n",
    "plt.xticks([3, 15, 27], ['01/06', '15/06', '31/06'])\n",
    "# leg = plt.legend(loc=\"upper left\")\n",
    "\n",
    "# Get the bounding box of the original legend\n",
    "# bb = leg.legendPatch.get_bbox().inverse_transformed(plt.gca().transAxes)\n",
    "# leg.legendPatch.\n",
    "plt.legend(bbox_to_anchor=(0.5, 1.2), bbox_transform=plt.gca().transAxes)\n",
    "\n",
    "plt.tight_layout()\n",
    "format_axes(plt.gca())\n",
    "plt.gca().xaxis.set_ticks_position('none')\n",
    "# plt.savefig(os.path.join(plot_base, 'real-data-counting.pdf'), bbox_inches='tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 902,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f1fc89819b0>"
      ]
     },
     "execution_count": 902,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAHkAAABNCAYAAACCEcvyAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1fc9111a90>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "latexify(fig_width=1.75, largeFonts=False)\n",
    "# plt.figure(figsize=(8,6))\n",
    "# plt.plot(60 * r_real_sampled.index / 10000, r_real_sampled.values)\n",
    "one_month = r_real_sampled.index <= 5000.0\n",
    "plt.fill_between(60 * r_real_sampled.index[one_month] / 10000, r_real_sampled.values[one_month], color=c1, alpha=0.5)\n",
    "\n",
    "tweet_times = df_real[(df_real.src_id == sim_opts.src_id)].t.unique()\n",
    "one_month_df = tweet_times < 5000.0\n",
    "plt.vlines(tweet_times[one_month_df] * 60 / 10000.0, ymin=0, ymax=3000, color=c2, linewidth=0.5)\n",
    "plt.tight_layout()\n",
    "plt.yticks([0, 1500, 3000]) # , [0, r'$10^3$', r'$3\\times 10^3$'])\n",
    "plt.xticks([3, 15, 27], ['01/06', '15/06', '31/06'])\n",
    "format_axes(plt.gca())\n",
    "# plt.savefig(os.path.join(plot_base, 'real-data-ranks.pdf'), bbox_inches='tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 903,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "20"
      ]
     },
     "execution_count": 903,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.sum(one_month_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 963,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:The SQLite files not found on /dev/shm, looking for them on local drives.\n",
      "WARNING:root:The HDF5 file not found on /dev/shm, looking for them on local drives.\n",
      "WARNING:root:Ignoring error during destruction: 'HDFSLoader' object has no attribute 'h5f'\n",
      "WARNING:root:Ignoring error during destruction: 'HDFSLoader' object has no attribute 'h5f'\n"
     ]
    }
   ],
   "source": [
    "hs = get_user_repository()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1363,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "expr_times = get_start_end_time()\n",
    "scaling_factor = (expr_time.end_time - expr_time.start_time) / scaled_period\n",
    "# num_segments = 6\n",
    "# real_segment_len =  (60 * 60 * 24 / 6)\n",
    "\n",
    "num_segments = 7\n",
    "real_segment_len = 60 * 60 * 24\n",
    "\n",
    "user_significance = find_significance(\n",
    "    user_id=sim_opts.src_id, \n",
    "    user_repository=hs,\n",
    "    num_segments=num_segments,\n",
    "    segment_length=real_segment_len,\n",
    "    return_tweet_times=True\n",
    ")\n",
    "\n",
    "sig_opts = Deco.Options(num_segments=num_segments, \n",
    "                        segment_length=real_segment_len / scaling_factor,\n",
    "                        start_time=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1364,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sig = user_significance.significance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1365,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(30, 7)"
      ]
     },
     "execution_count": 1365,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sig.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Distribution of significance among the followers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1366,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([ 13.,  25.,  22.,  31.,  17.,   9.,  24.]),\n",
       " array([ 0.   ,  0.857,  1.714,  2.571,  3.429,  4.286,  5.143,  6.   ]),\n",
       " <a list of 7 Patch objects>)"
      ]
     },
     "execution_count": 1366,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdsAAAEyCAYAAACyMXp+AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAER9JREFUeJzt3U9onHee5/GP1mG2D5FVhg3EkOpdMHiQ/+QaSj6nhBP6\n0o1EoA+C2GAfA86laXBIY8hFgfTNDWYW9dKQlOkcdsDp0qVP1kOuk6hM7+Lx0FVM4slmuiSloYfF\n1h6MazsdWyUr9XNJ5dfrVPXU4+JbZUpvPX/q0dT29vZ2AIBi/tO4BwCASSe2AFCY2AJAYWILAIWJ\nLQAUJrYAUNjQ2Lbb7VRVlcuXLw+WLS8vJ0larVa5yQBgQuwY26qqsra2lkajkV6vl1u3biV5ENlm\ns5l6vf5UhgSAg2xqtxe1+MlPfpLf/va3SZLV1dU0m82igwHApHhu2ApbW1v56KOPcuHChcGyjY2N\nVFWV9fX1nD9/frD8L3/5S959991cvHgxhw4dKjPxhFhZWcnS0tK4x9j3vE+7573aHe/T7nmvhrt3\n716uXr2ad955Jz/4wQ8eu96ut2zffPPN/OIXv8hLL700WLa8vJwzZ86k0WgkSX72s5/l448//p6j\nA8DB8uMf/zjvvffeYx/fccu20+lkamoqs7OzOXHiRH73u9/l8OHDqdVqaTabqdVq6fV6g/UvXryY\njz/+OL/5zW/y4osvju5VAMA+9OWXX+anP/1pLl68uON6O8Z2bW0tJ0+eTJJsbm7m5ZdfzvT0dE6f\nPp0k6ff7OXPmzGD9h7uOX3zxxW9tAQPAJBt26HTH2L7xxhv55JNP0mq1MjMzMzgpqt1uJ0mOHDmS\n2dnZEY0KAJNpx9g+//zzWVhY+M7y+fn5YgMBwKRxBSkAKExsAaAwsQWAwsQWAAoTWwAoTGwBoDCx\nBYDCxBYAChNbAChMbAGgMLEFgMLEFgAKE1sAKExsAaAwsQWAwsQWAAoTWwAoTGwBoDCxBYDCxBYA\nChNbAChMbAGgMLEFgMLEFgAKE1sAKExsAaCw58Y9AEyCe/fu5fbt2+MeY6yOHTuWQ4cOjXsM2JfE\nFkbg9u3beesf/yHTR18Y9yhjsfXFV/ngR2/m+PHj4x4F9iWxhRGZPvpCavWj4x4D2IccswWAwobG\ntt1up6qqXL58+TvLWq1W0eEAYBLsGNuqqrK2tpZGo5Fer5dbt26l0+lkamoqjUYjSXLr1q2nMigA\nHFQ7xrbRaOTdd99NkmxsbGR2djY3btzI9PR0kqRer2dtba38lABwgA09QWpraysfffRRLly4kCTZ\n3NxMrVYbPN7v98tNBwATYOgx2+np6Zw/fz4ffvhhut3u05gJACbKjrHtdDqDY7InTpxIu93OzMzM\nYGv2b7dyV1ZWCo4KAPvTsP7tGNu1tbVvhfWHP/xhXnvttfR6vSRJt9vN3NzcYP2lpaXvOy8AHDjD\n+rdjbN944430er20Wq3MzMyk2WxmdnY2yYMzlWdmZgb3AYBH2/EEqeeffz4LCwvfWf6oZQDAo7mC\nFAAUJrYAUJjYAkBhYgsAhYktABQmtgBQmNgCQGFiCwCFiS0AFCa2AFCY2AJAYWILAIWJLQAUJrYA\nUJjYAkBhYgsAhYktABQmtgBQmNgCQGFiCwCFiS0AFCa2AFCY2AJAYWILAIWJLQAUJrYAUJjYAkBh\nYgsAhYktABQmtgBQ2NDYtlqttFqtLC8vD5Y9vN1qtcpNBgATYsfYVlWVubm5LC4uptvtpqqqJA8i\n22w2U6/Xn8qQAHCQPbfTg91uN71eLwsLC6nX6+n1ekmSK1eupNlsPpUBAeCg2zG2i4uLg9udTiev\nv/56kmRjYyNVVWV9fT3nz58vOyEAHHA7xvahTqeTkydPZnZ2NkmysLCQJLl582aqqkqj0Sg34QFx\n79693L59e9xjjNWxY8dy6NChcY8BsO/sKrZVVeXSpUtJHhyvrdVqaTabqdVqg13LSbKyslJmygPg\n9u3beesf/yHTR18Y9yhjsfXFV/ngR2/m+PHj4x4F4KlbWVnJz3/+88c+PjS2rVYr586dS/IguvV6\nPadPn06S9Pv9nDlzZrDu0tJSfv3rX3/fmQ+s6aMvpFY/Ou4xAHjKlpaWdnx86NnI77//fl599dW8\n8sorSZJGo5GbN2+m3W7nyJEjg13LAMCj7bhl22g08umnn35n+fz8fLGBAGDSuIIUABQmtgBQmNgC\nQGG7+uoPADzKs36Ngbt37+5qPbEFYM+e9WsM/Pv/+pddrSe2AHwvz/I1Bv6jv7Wr9RyzBYDCxBYA\nChNbAChMbAGgMLEFgMLEFgAKE1sAKExsAaAwsQWAwsQWAAoTWwAoTGwBoDCxBYDC/NUfRuL+/fu5\nc+fOuMcYm2f5tQPDiS0j8ee7X+eXd3+f6a8/H/coY/HlP/0hL7789+MeA9inxJaReZb/puXWF1+N\newRgH3PMFgAKE1sAKExsAaAwsQWAwsQWAAoTWwAoTGwBoDCxBYDCxBYACht6BalWq5Uk+eMf/5i3\n3347SdJut3P48OF0u90sLi6WnRAADrgdt2yrqsrc3FwWFxfT7XZTVVU6nU6mpqbSaDSSJLdu3Xoq\ngwLAQbVjbB8GNknq9Xp6vV5u3LiR6enpwbK1tbXyUwLAAbbjbuS/3kXc6XTy+uuv57PPPkutVhss\n7/f75aYDgAmwqxOkOp1OTp48mdnZ2R3XW1lZGclQAHCQDOvfrmJbVVUuXbqUJJmZmRlszW5ubn5r\nK3dpaWmvcwLAgTWsf0Nj22q1cu7cuSQPovvaa6+l1+sleXBMd25ubgRjAsDkGno28vvvv59XX301\nr7zySpIMdiVXVZWZmZmhu5YB4Fm34wlSjUYjn3766XeWLywsFBsIACaNK0gBQGFiCwCFiS0AFCa2\nAFCY2AJAYWILAIWJLQAUJrYAUJjYAkBhYgsAhYktABQmtgBQmNgCQGFiCwCFiS0AFCa2AFCY2AJA\nYWILAIU9N+4BgIPv/v37uXPnzrjHGKtjx47l0KFD4x6DfUpsge/tz3e/zi/v/j7TX38+7lHGYuuL\nr/LBj97M8ePHxz0K+5TYAiMxffSF1OpHxz0G7EuO2QJAYWILAIWJLQAUJrYAUJjYAkBhYgsAhYkt\nABQmtgBQ2K5iu7y8/Mj7rVZr9BMBwIQZGttWq5XV1dXvLGs2m6nX68UGA4BJMfRyjYuLi2m3299a\nduXKlTSbzWJDAcAk2dMx242NjVRVlWvXro16HgCYOHuK7cLCQhqNRvr9fqqqGvVMADBRnji2f30M\nt1arpdfrjXwoAJgku4rt9vb24Ha9Xs/c3FySpN/v59SpU4PHVlZWRjweAOx/w/o3NLbtdjvr6+u5\nfv16kqTRaOTmzZtpt9s5cuRIZmdnB+suLS19z3EB4OAZ1r+hZyPPz89nfn7+O8sAgN1xBSkAKExs\nAaAwsQWAwsQWAAoTWwAoTGwBoDCxBYDCxBYAChNbAChMbAGgMLEFgMLEFgAKG/qHCPbil9f/R56v\nHS7x1PvW//nybvJfZ8Y9BgD7UJHY3v1vM9n8L7UST71v9f/zX8Y9AgD7lN3IAFCY2AJAYWILAIWJ\nLQAUJrYAUJjYAkBhYgsAhYktABQmtgBQmNgCQGFiCwCFiS0AFCa2AFCY2AJAYWILAIWJLQAUJrYA\nUNiuYru8vPyt++12O1VVpdVqFRkKACbJ0Ni2Wq2srq4O7nc6nUxNTaXRaCRJbt26VW46AJgAQ2O7\nuLiYer0+uH/jxo1MT08nSer1etbW1spNBwATYFe7kbe3twe3Nzc3U6vVBvf7/f7opwKACeIEKQAo\nbFexnZqaGtyemZkZbM3+7VbuysrKiMcDgP1vWP+eeDfy2bNn0+v1kiTdbjdzc3ODx5aWlvYyIwAc\naMP6NzS27XY76+vruX79epLkxIkTSZKqqjIzM5PZ2dkRjAkAk+u5YSvMz89nfn7+W8sWFhaKDQQA\nk8YJUgBQmNgCQGFiCwCFDT1mC8DO7t+/nzt37ox7jLF4Vl/3kxJbgO/pz3e/zi/v/j7TX38+7lGe\nui//6Q958eW/H/cY+57YAozA9NEXUqsfHfcYT93WF1+Ne4QDwTFbAChMbAGgMLEFgMLEFgAKE1sA\nKExsAaAwsQWAwsQWAAoTWwAoTGwBoDCxBYDCxBYAChNbAChMbAGgMLEFgMLEFgAKE1sAKExsAaAw\nsQWAwsQWAAoTWwAoTGwBoDCxBYDCxBYAChNbAChsT7FdXl5OkrRarZEOAwCTaE+xbbVaaTabqdfr\no54HACbOc3v5R1euXEmz2Rz1LAAwkfa0ZbuxsZGqqnLt2rVRzwMAE2dPsV1YWEij0Ui/309VVaOe\nCQAmyhPHttVqZXV1NUlSq9XS6/UGj62srIxuMgA4IIb174ljW6/XMzc3lyTp9/s5derU4LGlpaUn\nfToAOPCG9e+JT5BqNBppt9tJkiNHjmR2dnZvkwHAM2JPZyPPz8+Peg4AmFiuIAUAhYktABQmtgBQ\nmNgCQGFiCwCFiS0AFCa2AFCY2AJAYWILAIWJLQAUJrYAUJjYAkBhYgsAhYktABQmtgBQmNgCQGFi\nCwCFiS0AFCa2AFCY2AJAYWILAIWJLQAUJrYAUJjYAkBhYgsAhYktABQmtgBQmNgCQGFiCwCFiS0A\nFPbcXv5Ru93O4cOH0+12s7i4OOqZAGCiPPGWbafTydTUVBqNRpLk1q1bIx8KACbJE8f2xo0bmZ6e\nTpLU6/Wsra2NfCgAmCRPHNvNzc3UarXB/X6/P9KBAGDS7OmY7ePcu3cvSfLvf7iTv/vy+VE+9b73\nzb99ne3t7fxHf2vco4zFxr/8a/7v1p+9fq9/3KOMxbP8+p/l154kG//cS/L/+/c4TxzbmZmZwdbs\n327lXr16NUnyv//7/3zSpwWAA+vq1at57733Hvv4E8f27NmzWV9fT5J0u92cOXNm8Ng777yTJLl4\n8WIOHTr0pE/9TFlZWcnS0tK4x9j3vE+7573aHe/T7nmvhrt3716uXr066N/jTG1vb28/6ZNfv349\nL730Unq9XhYWFvY85LNseXk5b7/99rjHgGfStWvXcv78+XGPwTNkTxe1WFhYSKPR+FZo2+12qqpK\nq9Ua2XCTqtVqZXV1ddxj7HutViutVivLy8vjHmVfe/jZu3z58rhHORCqqkpVVeMeY197+Jnz83y4\nTqeTdrs99L0ayRWkfPf2ySwuLqZer497jH2tqqrMzc1lcXEx3W7XD8fHqKoqa2traTQa6fV6PnuM\nRKvVSrPZ9HNqF371q19lfn4+W1tbO37+RhJb371l1P46sPV6Pb1eb8wT7U+NRiPvvvtukmRjYyOz\ns7Njnmh/63Q6aTQa2cPRs2fKlStXsrq6OtiA4tHa7XZefvnlJMm5c+d2/PyNJLa+e8uoLS4uDg5T\ndDqdnDp1aswT7V9bW1u5du1aLly4MO5R9r2NjY1xj3AgbGxspKqqXLt2bdyj7GufffZZ+v1+Op3O\n0PfKHyJgX+t0Ojl58qQtth1MT0/n/Pnz+fDDD+0B2MHDrdokmZqaGvM0+9vD83L6/b5DOEPUarWc\nOHEiyYMt3ccZSWx3+u4tj2Y31u5UVZVLly6Ne4x9q9PpDI4TnThxYscP+7Ou2+1mdXU1rVYrf/rT\nnxzffoy/PoGzVqv5BW4HtVptcFz78OHD+fzzzx+77khie/bs2cF/SLfbzdzc3CiedmK12+2sr6/n\n+vXr4x5lX2u1Wjl37lyS+O36MdbW1r71i64TWh5vfn4+zWYzSfLNN9+MeZr9q16vD36G9/t9h3B2\nMD8/n263m+TB5+/06dOPXXdP37N9FN+9ZZSqqspbb72Vw4cPZ3NzMx988IGTNR7hm2++ySeffJLt\n7e10u117ARiJh3tIer3e4BdeHu369euDrdqdPn8jiy0A8GhOkAKAwsQWAAoTWwAoTGwBoDCxBYDC\nxBYAChNbACjs/wH/qKFtI11vBwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1f2e2dd470>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.hist(convert_to_bins([x for x in hs.get_user_tweets(sim_opts.src_id) if x < expr_times.start_time], \n",
    "                         start_time=expr_times.start_time, \n",
    "                         num_segments=num_segments,\n",
    "                         segment_length=real_segment_len), \n",
    "         bins=7)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1368,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdQAAAEZCAYAAAAuUdQGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAEdNJREFUeJzt3F1oXPeZ+PFHf4X8vcV6May2MXh8IxCyYxcaaIrkqxas\naRICxcaipS0CSbQuLXHAvQkkcgKFXFgG35U04xS1FNwxmCahwWNY6E00LYW0YGvELlG6WAN58e5W\nL9nSXIjZi+BZK9aM7PinGR3587kazRnbz08+M1+dM0fTUavVagEA3Jf/1+4BAGAnEFQASEBQASAB\nQQWABAQVABIQVABI4KHNHlCpVGJxcTGWl5djdHS0FTMBQOZseoT6yiuvRD6fj9XV1Zifn2/FTACQ\nOR3NPtihVCpFtVqNiYmJVs4EAJnT9JTvtWvXoqOjIyqVSszOzsbk5GR92z/+8Y+4fv169PX1RWdn\n55YPupGZmZkYGxtry7/dbg/y2iOs3/of3PU/yGuPaO/619bW4ubNm3Ho0KHYtWvXHdubHqFOT0/H\nnj17YmJiIgqFQuRyucjn8xER8dxzz8Xly5e3bnIA2IaOHTsWL7/88h33Nz1C7e3tjVwuFxER3d3d\ncf369XpQT548GZcvX45f//rX8cgjj2zByACwfXzwwQfxne98J06ePLnh9qZBzefzcfXq1YiIWFlZ\nicOHD9e33TrN+8gjj8S+fftSzUsGrK2txcLCQrvHSKa/v79tb1sA2dPo9aJpUHO5XHR3d0epVIrl\n5eUYGRnZkuHIloWFhXj2zdeia29fu0e5b6vv34zzT4/HwMBAu0cBMm7T30M9ceJERET9VC9ERHTt\n7Yve3N52jwGwbfikJABIQFABIAFBBYAEBBUAEhBUAEhAUAEgAUEFgAQEFQASEFQASEBQASABQQWA\nBAQVABIQVABIQFABIAFBBYAEBBUAEhBUAEhAUAEgAUEFgAQEFQASEFQASEBQASCBh9o9AADZsLa2\nFgsLC+0eI5n+/v7o7OxM9vcJKgB3ZWFhIZ5987Xo2tvX7lHu2+r7N+P80+MxMDCQ7O8UVADuWtfe\nvujN7W33GNuS91ABIAFBBYAEnPIFuAcuzKERQQW4By7MoRFBBbhHLsxhI95DBYAEBBUAEtg0qNPT\n0xERUSwWt3wYAMiqTYNaLBZjZGQkcrlcK+YBgEza9KKkn/70pzEyMtKKWQAgszY9Ql1eXo5yuRyF\nQqEV8wBAJm0a1BMnTsTQ0FAsLS1FuVxuxUwAkDlNg1osFuPq1asREdHb2xvVarW+bWZmZmsnA4Bt\nqFH/mgY1l8vF8PBwREQsLS3FoUOH6tvGxsYSjgcA2dCof00vShoaGopSqRQREXv27IkDBw6knwwA\ndoBNr/LN5/OtmAMAMs0nJQFAAoIKAAkIKgAkIKgAkICgAkACggoACQgqACQgqACQgKACQAKCCgAJ\nCCoAJLDpZ/lu5q9//Wv8/e9/TzFLW/X390dnZ2e7xwAgo+47qOeu/Wv8/8WuFLO0zer7N+P80+Mx\nMDDQ7lEAyKj7Dmr33n+Jf/rn3hSzAEBmeQ8VABIQVABIQFABIAFBBYAEBBUAEhBUAEhAUAEgAUEF\ngAQEFQASEFQASEBQASABQQWABAQVABIQVABIQFABIAFBBYAEBBUAEnio3QNk0draWiwsLLR7jGT6\n+/ujs7Oz3WMAZNpdB7VQKMTk5ORWzpIZCwsL8eybr0XX3r52j3LfVt+/GeefHo+BgYF2jwKQaXcV\n1HK5HOVyWVBv07W3L3pze9s9BgDbhPdQASCBTYNaqVRiaGgoarVaK+YBgEzaNKjLy8utmAMAMq1p\nUG8dnUZEdHR0tGQgAMiipkFdXFyMq1evRrFYjL/97W8xPz9f3zYzM7PlwwHAdtOof02Dms/nY2Rk\nJCIiPv7443XbxsbGEo0GANnRqH939Wszo6OjMTo6mnQgANhJ/NoMACQgqACQgKACQAKCCgAJCCoA\nJCCoAJCAoAJAAoIKAAkIKgAkIKgAkICgAkACggoACQgqACQgqACQgKACQAKCCgAJCCoAJCCoAJCA\noAJAAoIKAAkIKgAkIKgAkICgAkACggoACQgqACQgqACQgKACQAKCCgAJCCoAJCCoAJCAoAJAAoIK\nAAkIKgAkIKgAkMCmQS2VSlEul2NqaqoV8wBAJjUNarlcjtnZ2RgaGopqtRrz8/OtmgsAMuWhZhuH\nhoZiaGgoIiKWl5fjwIEDLRkKALKmaVAjIlZXV+M3v/lN/OAHP2jFPACQSZu+h9rV1RWTk5Nx8eLF\nqFar9ftnZma2dDAA2I4a9a9pUCuVSv1904MHD0apVKpvGxsbSzgeAGRDo/41Ders7GwsLS1FRMTK\nykrkcrn0kwHADtA0qN/61reiWq1GsViMnp6eGBkZadVcAJApTS9K2r17d5w4caJVswBAZvmkJABI\nQFABIAFBBYAEBBUAEhBUAEhAUAEgAUEFgAQEFQASEFQASEBQASABQQWABAQVABIQVABIQFABIAFB\nBYAEBBUAEhBUAEjgoXYPAFmztrYWCwsL7R4jmf7+/ujs7Gz3GJB5ggr3aGFhIZ5987Xo2tvX7lHu\n2+r7N+P80+MxMDDQ7lEg8wQVPoeuvX3Rm9vb7jGAbcR7qACQgKACQAJO+QL3xEVZsDFBBe6Ji7Jg\nY4IK3DMXZcGdvIcKAAkIKgAkIKgAkICgAkACggoACQgqACQgqACQgKACQAKbfrBDsViMiIgbN27E\nT37yky0fCACyqOkRarlcjuHh4RgdHY3FxcUol8utmgsAMqVpUG+PaC6Xi2q12pKhACBrmp7yHR0d\nrd+uVCrx1FNPbflAAJBFd3VRUqVSiUcffTQOHDhQv29mZmbLhgKA7apR/+4qqOVyOU6fPr3uvrGx\nsfufCgAyplH/Ng1qsViMiYmJiAgXJQFAA5te5Xvu3Lk4evRofPWrX23VTACQOU0vShoaGoo//vGP\nrZoFADLLJyUBQAKCCgAJCCoAJCCoAJCAoAJAAoIKAAkIKgAkIKgAkICgAkACggoACQgqACQgqACQ\ngKACQAKCCgAJCCoAJCCoAJCAoAJAAoIKAAkIKgAkIKgAkICgAkACggoACQgqACQgqACQgKACQAKC\nCgAJCCoAJCCoAJCAoAJAAoIKAAkIKgAkIKgAkMBdBXV6enqr5wCATNs0qMViMa5evdqKWQAgszYN\n6ujoaORyuVbMAgCZ5T1UAEhAUAEggc8d1JmZmZRzAEAmNOrfXQW1Vqvdcd/Y2Nj9TQQAGdSof5sG\ntVQqxdzcXFy6dCn5UACwUzy02QPy+Xzk8/lWzAIAmeWiJABIQFABIAFBBYAEBBUAEhBUAEhAUAEg\nAUEFgAQEFQASEFQASEBQASABQQWABAQVABIQVABIQFABIAFBBYAEBBUAEhBUAEhAUAEgAUEFgAQE\nFQASEFQASEBQASABQQWABAQVABIQVABIQFABIAFBBYAEBBUAEhBUAEhAUAEgAUEFgAQEFQASEFQA\nSEBQASCBhzZ7QKlUiu7u7lhcXIzR0dFWzAQAmdP0CLVSqURHR0cMDQ1FRMT8/HxLhgKArGka1Lfe\neiu6uroiIiKXy8Xs7GxLhgKArGl6yndlZSV6e3vrXy8tLdVvr62tRUTEf/7be/Hw+7u3aLzW+J+P\n/is+zH0YX/jCF+7q8R9++GH897//R3yytLrFk229e117hPVbv/U/qOt/kNceEfHBBx9ExP/177M6\narVardEfnpqaim9/+9tx4MCBKJfLMTs7G6dPn46IiOeeey4uX758L/MDQOYdO3YsXn755Tvub3qE\n2tPTUz8q/ezR6pkzZ+L48ePR19cXnZ2dice9OzMzMzE2NtaWf7vdHuS1R1i/9T+463+Q1x7R3vWv\nra3FzZs349ChQxtub3qEWqlUYm5uLk6cOBGFQiGOHDkSBw4c2LJhASCrml6UdPDgwYiIKJfL0dPT\n07KYTk1NxeDgYFy4cCEqlUocPXo0jh8/HvPz81EsFuvbGlldXY1Tp07FxMRES+ZN5ZlnnomRkZE4\ndepUDA4O1r8PV69ejUKhEIODg+0esaVKpVI8/vjjcfz48Th37lz9+7LR1eZHjx6Njz/+uA1Tbp1C\noRBnzpyJc+fOxfT0dNN9fid5UNddKpXi6NGjd7zOnj17NgYHB+PMmTNtmqx1mr0GZkJtmxocHKxV\nKpVarVarvfrqq7Xx8fH6tqmpqU3//Nzc3Lo/kwXj4+O11dXV2tzcXO3xxx+v1Wq1WrFYrBWLxVqt\nVqvf9yB55plnaoVCof719PR0bXZ29o7Hra6utnKsLXfjxo3aCy+8UP+6UCis+z5s5OzZs1s91pb7\nPOveSc6ePVs7evRorVQq1e974YUXaoODg22cqnU2ew3c7rbtJyXl8/n6r+ksLS3Vby8uLjY8f511\nBw8ejN27118x/Y1vfKN+u9b47PwDoVKpxPDw8IbbPvt9y7pqtRrXr1+P1dVPr6acmJiIXC7X8PGr\nq6tRKpVaNd6Wudd170T5fD5+97vfRcSnr3f79+9v80Sts9lr4Ha3bYP6xBNPxFtvvVX/OpfLRblc\njnK5XP+giampqbhw4UJMTU3VH3fq1Kk4c+ZMXLx4seUz369bV1DfrqurK06cOFH/+tKlSzE+Pl5f\n861T4I1ul8vlOHbsWFSr1RasYOssLi7GxYsXY2hoKP7yl7+sW9vPfvazHXc6/NY+/pWvfCXGx8ej\nVCrFyMhIRERMT09HoVCI8fHx+v/r7OxsLC4uxoULFzL9f91o3Zvt5+fOnVv3vMiqjo6OGB0drZ/i\nLJVKkc/n1z1mamoqCoXCHa8BO+F70Ow1sNE+ELFxC9ph2wZ1eHg45ubmYn5+Pg4fPhz5fD7efvvt\nWFxcjH379kWxWIyOjo6YmJiI7u7uuHTpUpRKpejo6IiXXnppRx7FrqysxOHDh+O1116LK1euRETE\n6OhodHR01G/39PTccf/JkyfXXaGdNW+//XZMTEzU1/PDH/5w3dq++93v1te9k1y+fDl+8YtfRHd3\ndzz77LNx7ty5WFxcjGq1GpOTkzE8PFw/Ks3n8/Xnw759+9o8+f3ZaN2b7edPPfXUuudFluVyufoB\nxPLy8rrnbqFQiI6OjpicnIyIT3/AvvWRsDvpe7CRRvvARi1ol20b1K6urjh48GA8//zzceTIkXji\niSfWndK6fv16LC8vx4ULF+qnh65du1Y/PbQTTxP19PTUfyrr7e2tX4Rz+6ng5eXl+u1arRa5XC5G\nRkYyfUr0yJEjd/zO8+1ru/VpXjtJoVCI1dXVGBoaivPnz8ef/vSnKJVKkcvlYnh4OAqFQly7dm3d\nh63sBI3WHdF4P+/u7t7weZFFt9Y4MjISr7766h2ne2/cuFG/b//+/XH9+vWIaPzasNPcvg+srKxE\nxMYtaJdtG9SIiCeffDJWV1dj9+7dcfDgwVheXo4vfelLEfHpztTT0xMTExP1I9L9+/fH4uJiRETM\nzc21c/T7UqvVNny/9Pb7br996ye1WzvY7bJ+tHLL7t2746WXXlp33+1r24nvL7/yyiv127eeA8Vi\nMW7cuBGTk5N3HJV3d3dHxKdRyrKN1h3ReD9v9LzIolvXijz55JMxOzsbw8PD614P9u/fHzdu3IiI\nT+N660zcTvoeRDR+DdxoH9ioBe3S+eKLL77Ytn99E319ffHJJ5/UL0SZm5uLsbGxePjhh+Oxxx6L\n119/Pcrlcrzzzjvx9a9/Pb785S/HlStXYm5uLj766KMol8tx6NChTB2tViqV+PnPfx7z8/Pr1l4q\nleKNN96IgYGB+Oijj+L111+PXbt2xWOPPRYrKytRKpWiWq3G3NxcrKys1C9S+eIXvxiPPvpom1f1\n+ZRKpfjVr34VH374YezZsyf6+/vr99++tnK5HJcuXcr0Wj/rnXfeiU8++STm5+fj3XffjStXrsT3\nv//96OnpiV/+8pdRrVZj165dMTs7G0eOHInu7u6Ym5uLK1euxNe+9rVM7fO3a7Tu7u7uhvt5o+dF\n1hQKhXjjjTfi3Xffje9973vx5z//Ob75zW/G888/H++9917cvHkzTp8+Hb///e9jYWEhVlZW4kc/\n+lHT14YsavQaGBEb7gM//vGP47e//W384Q9/qLfg4YcfbsvsTT/YAQC4O9v6lC8AZIWgAkACggoA\nCQgqACQgqACQgKACQAKCCgAJ/C/fp69qzggRbgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1f13be81d0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,4.5))\n",
    "plt.bar(range(num_segments), (user_significance.significance.sum(0)))\n",
    "plt.xticks(np.arange(0.5, num_segments + 0.5), ['Wed', 'Thur', 'Fri', 'Sat', 'Sun', 'Mon', 'Tue']);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is what the distribution of tweewts looked like without any significance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1281,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from datetime import datetime"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1282,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<function datetime.utcfromtimestamp>"
      ]
     },
     "execution_count": 1282,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "datetime.utcfromtimestamp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1288,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "days = [datetime.utcfromtimestamp(x).weekday() for y in user_significance.all_tweet_times for x in y]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1291,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdsAAAEVCAYAAAComE/wAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1f3e1e8cc0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,4.5))\n",
    "plt.bar(range(num_segments), np.bincount(days))\n",
    "plt.xticks(np.arange(0.5, num_segments + 0.5), ['Mon', 'Tue', 'Wed', 'Thur', 'Fri', 'Sat', 'Sun']);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1037,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([    0.   ,   341.478,   726.068,  1026.144,  1337.371,  1551.147,\n",
       "        1864.017,  2095.041,  2313.458,  2644.116,  2905.501,  3268.387,\n",
       "        3343.043,  3590.779,  3948.662,  4071.397,  4371.113,  4463.008,\n",
       "        4568.638,  4715.002,  5201.879,  5719.893,  6063.347,  6171.623,\n",
       "        6346.155,  6619.573,  6952.219,  7394.048,  7598.194,  8061.934,\n",
       "        8700.188,  9633.311])"
      ]
     },
     "execution_count": 1037,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_opt.t[df_opt.src_id == sim_opts.src_id].unique() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1038,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "161.29032258064518"
      ]
     },
     "execution_count": 1038,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sig_opts.segment_length"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1040,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time_period = 1129.0322580645163\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([0, 2, 4, 6, 1, 2, 4, 5, 0, 2, 4, 6, 6, 1, 3, 4, 6, 6, 0, 1, 4, 0, 2,\n",
       "       3, 4, 6, 1, 3, 5, 0, 4, 3])"
      ]
     },
     "execution_count": 1040,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "convert_to_bins(df_opt.t[df_opt.src_id == sim_opts.src_id].unique(), opts=sig_opts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1041,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "time_period = 1129.0322580645163\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(array([ 5.,  4.,  4.,  4.,  7.,  2.,  6.]),\n",
       " array([ 0.   ,  0.857,  1.714,  2.571,  3.429,  4.286,  5.143,  6.   ]),\n",
       " <a list of 7 Patch objects>)"
      ]
     },
     "execution_count": 1041,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdUAAAEVCAYAAAC2UX9DAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAEjyf61o4ymiG+3eAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1f3ace1908>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,4.5))\n",
    "plt.hist(convert_to_bins(df_opt.t[df_opt.src_id == sim_opts.src_id].unique(), opts=sig_opts), bins=num_segments)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The rank of the user in the feeds of his followers (in reality)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1042,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "422.95381261988513"
      ]
     },
     "execution_count": 1042,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "average_rank(df_opt, sim_opts=sim_opts) / sim_opts.end_time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 547,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1387.4009282506947"
      ]
     },
     "execution_count": 547,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "average_rank(df_real, sim_opts=sim_opts) / sim_opts.end_time"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 940,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(20712572, 20712572)"
      ]
     },
     "execution_count": 940,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "user_id, sim_opts.src_id"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1043,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "30"
      ]
     },
     "execution_count": 1043,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(sim_opts.sink_ids)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1369,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "*** \u001b[31m2016-08-05 16:22:23.631316\u001b[0m Checkpoint: Initial capacity = 31.08, target capacity = 30, s_init = 2999322039568.9756\n",
      "*** \u001b[31m2016-08-05 16:23:37.398815\u001b[0m Checkpoint: s = 5998644079137.951, capacity = 26.28\n",
      "*** \u001b[31m2016-08-05 16:23:37.399612\u001b[0m Checkpoint: s_hi = 5998644079137.951, s_lo = 2999322039568.9756\n",
      "*** \u001b[31m2016-08-05 16:24:54.201172\u001b[0m Checkpoint: new_capacity = 28.16, s = 4498983059353.463\n",
      "*** \u001b[31m2016-08-05 16:26:09.619781\u001b[0m Checkpoint: new_capacity = 28.88, s = 3749152549461.219\n",
      "*** \u001b[31m2016-08-05 16:27:23.752828\u001b[0m Checkpoint: new_capacity = 29.84, s = 3374237294515.0977\n",
      "CPU times: user 797 ms, sys: 51.8 s, total: 52.6 s\n",
      "Wall time: 6min 16s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "s_opt_sig = sweep_s_with_significance(\n",
    "    sim_opts=sim_opts.update({ 'q_vec': sig }), \n",
    "    capacity_cap=df_real.event_id[df_real.src_id == sim_opts.src_id].nunique(),\n",
    "    time_period=(7 * 24 * 60 * 60) / scaling_factor,\n",
    "    parallel=True,\n",
    "    verbose=True,\n",
    "    s_init=3374237294515.0977\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This is the value that `s` should have for running with significance."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1384,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3374237294515.0977"
      ]
     },
     "execution_count": 1384,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "s_opt_sig"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1370,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "sig_mgr = sim_opts.update({ 's': s_opt_sig, 'q_vec': sig }).create_manager_with_significance(\n",
    "    seed=878,\n",
    "    time_period=(7 * 24 * 60 * 60) / scaling_factor\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1371,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 54.9 s, sys: 2.1 s, total: 57 s\n",
      "Wall time: 56 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<opt_model.Manager at 0x7f1f40193748>"
      ]
     },
     "execution_count": 1371,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%%time\n",
    "sig_mgr.run_dynamic()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1372,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "df_sig = sig_mgr.state.get_dataframe()\n",
    "r_sig = rank_of_src_in_df(df_sig, sim_opts.src_id).mean(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1373,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 1, 2, 6, 0, 0, 2, 5, 6, 1, 1, 3, 5, 6, 1, 2, 3, 0, 1, 2, 4, 6, 2,\n",
       "       6, 1, 2, 4, 6, 4])"
      ]
     },
     "execution_count": 1373,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "convert_to_bins(df_sig.t[df_sig.src_id == sim_opts.src_id].unique(), opts=sig_opts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1374,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdYAAAEZCAYAAAAqpAQ7AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nif4H9X9/0JsM+3MAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1ed1fe66d8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,4.5))\n",
    "expr_time = get_start_end_time()\n",
    "\n",
    "\n",
    "plt.hist(convert_to_bins(df_sig.t[df_sig.src_id == sim_opts.src_id].unique(), opts=sig_opts),\n",
    "         bins=num_segments);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1375,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "r_sig_sampled = pd.concat([r_sig[::200], r_sig[r_sig == 0]]).sort_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1468,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAH0AAABNCAYAAACL+muIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nfUgkeh/yfwueIJmd5uBZAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1ec4b00400>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "latexify(fig_width=1.75, largeFonts=False)\n",
    "# latexify(fig_width=8, largeFonts=True)\n",
    "# plt.figure(figsize=(8,6))\n",
    "# plt.plot(60 * r_opt_sampled.index / 10000, r_opt_sampled.values, color=c1, linewidth=0.75)\n",
    "one_month = r_sig_sampled.index <= 5000.0\n",
    "ymin, ymax = 0, 3000\n",
    "\n",
    "plt.fill_between(60 * r_sig_sampled.index[one_month] / 10000, r_sig_sampled.values[one_month], color=c1,\n",
    "                 label=r'$\\bar{r}(t)$',\n",
    "                 alpha=0.5)\n",
    "\n",
    "tweet_times = df_sig[(df_sig.src_id == sim_opts.src_id)].t.unique()\n",
    "one_month_df = tweet_times < 5000.0\n",
    "num_tweets = np.sum(one_month_df)\n",
    "\n",
    "# plt.vlines([7, 14, 21, 28], ymin=0, ymax=3000, color=c3, linewidth=0.75)\n",
    "# plt.vlines(range(30), ymin=0, ymax=3000, color=c4, linewidth=0.5)\n",
    "\n",
    "# Marking out the weekends.\n",
    "plt.fill_between([3, 5], [ymax, ymax], color=c3, alpha=0.25)\n",
    "plt.fill_between([10, 12], [ymax, ymax], color=c3, alpha=0.25)\n",
    "plt.fill_between([17, 19], [ymax, ymax], color=c3, alpha=0.25)\n",
    "plt.fill_between([24, 26], [ymax, ymax], color=c3, alpha=0.25)\n",
    "\n",
    "plt.step(tweet_times[one_month_df] * 60 / 10000.0, np.arange(num_tweets) / num_tweets * ymax, \n",
    "         label='$N(t)$',\n",
    "         color=c2, linewidth=1.0)\n",
    "\n",
    "# plt.legend(loc=\"upper left\")\n",
    "# plt.legend(bbox_to_anchor=(0.5, 1.2), bbox_transform=plt.gca().transAxes)\n",
    "\n",
    "plt.yticks([0, 1500, 3000])\n",
    "plt.xticks([3, 15, 27], ['01/06', '15/06', '31/06'])\n",
    "plt.tight_layout()\n",
    "format_axes(plt.gca())\n",
    "plt.gca().xaxis.set_ticks_position('none')\n",
    "# plt.savefig(os.path.join(plot_base, 'sig-data-counting.pdf'), bbox_inches='tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1407,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAH0AAABNCAYAAACL+muIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAABdtJREFUeJztnTFs01gYx/++npjixCODbSExIFosMdxi06GAMGVjQY5Y\nTmordUVXEFtVIcb0BjbADLecSDaWKs5CBprHcDqdShrEgNSL3VuCBK6z+4aTLZoW200uoc77flNe\n6i/fV/3z3vP7vucXIQzDEARX/PC9AyAmD4nOISQ6h5DoHEKicwiJziGZRF9fXwcABEEAxhgajQaC\nIDjSJvJBquidTgeCIAAAnj59Cl3Xoes6qtUqnj17dqhN5IMfk/7oui5EUUSxWAQAHBwcAABEUUS3\n242/DKIownXdMYdK/F8k9nTP8xAEAbrdLlzXRalUQr/fRxAEUFUVxWLxUDui2Wzi5cuXYw+eGI7E\nnq7rOoIgQL/fhyAIsCwL29vbEAQB5XIZX758iduWZcV2CwsLaDab446dGBJhXLn3ZrOJhYWFcXw0\nMSK0ZOMQEp1DSHQOSbyRmzS//vZH5mt/+fmnMUYy3VBP55BT1dOHhUaIk0E9nUMSe3oQBGi329jd\n3UW5XEYYhmi32wiCALquA8ChtiiKEwmaGI1E0dvtNjRNg+u66Ha72Nrawv379xEEAarVKnzfx9ra\nWtxeWVmZVNzflbxPJ5nSsEEQYHZ2Ns6nT0vBJe/iDUvinF6r1SCKImZnZ2Hb9okKLnt7e+OOnRiS\nxJ6uaRoYY/A8D4uLiwBABZcpIFH0ixcvHnlPluX4daFQONQm8sHY1unsr3/w59/Z5sxpmi/zAK3T\nOYRE5xASnUNIdA4h0TmEROeQ1H3vnueh0+nANE1IkkQFlykgUfT379/DMAyUSiU4jnOkwMJrwSXv\nJA7vpmmiUCiAMYbFxcUjT7h83c5jwYVXUjNyjDEYhgFZluMCSxiGUFUVvu8fakdQ3v10kyg6Ywy2\nbUOWZVy5cuVET7iwv34fe/DEcKTW06MbtggquOQfWrJxCInOISQ6h0zFvve8cFr25FFP5xDq6aec\ncYwOqaJH+fdoOzTl3vNP6vDuOE58XBidLjUdpIpuGEb8mnLv08GJ5vToYQfKveebVNFbrRZ834dh\nGJR7nxJSRf+6Rj6Ya6fcez6hdTqHkOgcQqJzCInOISQ6h5DoHEKic8hIVbbBAgwVXPLBSD19sABD\n5IORRKeCSz4Z6ZD/zc1NrK6uIgxD1Go1LC8vA/iv4PLhwwdcuHDhWLu9vT2cO3fuRL6GscmL3Th8\nSZKEy5cvH28YjoDrumG9Xg8dxwmDIMhs9/r16xP7GsYmL3aTjnFsP+dBnF5mNjY2NsbthDGGGzdu\n4Pbt2ygWi6hUKnjx4gWuX7+OM2fOJNouLS2h1+vh4cOH6PV62N/fx9zcXKLN+vo6Xr16BVVVcffu\nXVy7dg2VSgWfPn061nbQh+d5YIzh48ePqb6i/29paQlnz57FkydPMDMzg/Pnz3/zetd18fbtW/R6\nPdi2jatXr6b6GMVukIms03Vdh2VZh+7wZVlGoVBItS2Xy7AsC6qqYm1tDaVSKdXmwYMH8dGm0VGm\nmqbhzp07mXxIkoS5uTlkHQR1XYeqqjBNE48fP4bv+4nXO44DQRCg63p8KGMWhrUbZGLJGcMw0Gq1\nwBiDpmmZ7UzTTGwfR5QvYIzBsizU6/WhfGxvb2NzcxONRiPV5+fPn3Hv3j34vv/NL1eEZVl48+YN\nTNNEp9MBYwy1Wi3+edMsdru7u7BtG47jwHEcVCqV1BgjJiJ6tLHy0qVLqNfrUBQlcy+KOOn1N2/e\nhG3bWF5eRrVazbTZY9DH/Pw8VldXsbOzk2orSdKh3UNJtFotPHr0CI1GAzs7O5AkCYqi4ODgAJ7n\nZbJ79+5dPPUoipLJb8RE9r1Xq1UwxrCysgLP89BqtbC/v49+v59piI+2bHmel3mnzq1bt+LcgWEY\nR56+TfPhum585HmSEADi83MVRckUX7vdhiAIkGUZ8/Pz2NragqZpKJVKcF33m58xaKcoCp4/fw5V\nVVNj/Bq6e+cQKrhwCInOISQ6h5DoHEKicwiJziEkOof8C4Ua+ayWQSXVAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f1edd4b6f60>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plt.figure(figsize=(8,4.5))\n",
    "latexify(fig_width=1.75, largeFonts=False)\n",
    "plt.bar(range(num_segments), (user_significance.raw_significance.sum(0))[[5, 6, 0, 1, 2, 3, 4]], color=c3, linewidth=0)\n",
    "plt.xticks(np.arange(0.5, num_segments + 0.5), ['M', 'T', 'W', 'Th', 'F', 'Sa', 'Su']);\n",
    "plt.yticks([0, 100, 200, 300, 400])\n",
    "plt.tight_layout()\n",
    "format_axes(plt.gca())\n",
    "plt.savefig(os.path.join(plot_base, 'sig-week-day.pdf'), bbox_inches='tight', pad_inches=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1378,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "389.45432613518233"
      ]
     },
     "execution_count": 1378,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "average_rank(df_opt[df_opt.t < 5000], end_time=5000, sim_opts=sim_opts) / sim_opts.end_time * 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1379,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "425.25265231631306"
      ]
     },
     "execution_count": 1379,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "average_rank(df_sig[df_sig.t < 5000], end_time=5000, sim_opts=sim_opts) / sim_opts.end_time * 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1380,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "698.0467206055821"
      ]
     },
     "execution_count": 1380,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "average_rank(df_real[df_real.t < 5000], end_time=5000, sim_opts=sim_opts) / sim_opts.end_time * 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1381,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "20"
      ]
     },
     "execution_count": 1381,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_opt[(df_opt.src_id == sim_opts.src_id) & (df_opt.t < 5000.0)].event_id.nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1382,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "20"
      ]
     },
     "execution_count": 1382,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_sig[(df_sig.src_id == sim_opts.src_id) & (df_sig.t < 5000.0)].event_id.nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1383,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "20"
      ]
     },
     "execution_count": 1383,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_real[(df_real.src_id == sim_opts.src_id) & (df_sig.t < 5000.0)].event_id.nunique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1415,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1431.0"
      ]
     },
     "execution_count": 1415,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r_sig[r_sig.index < 5000].max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1416,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1085.1666666666667"
      ]
     },
     "execution_count": 1416,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r_opt[r_opt.index < 5000].max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1418,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2597.9000000000001"
      ]
     },
     "execution_count": 1418,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "r_real[r_real.index < 5000].max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}