lab_ex_6_1_DASK.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<img src=\"IMG/dask_horizontal.svg\"  width=\"30%\">\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Introduction to DASK\n",
    "Dask is a parallel computing library that scales the existing Python ecosystem. This tutorial will introduce Dask and parallel data analysis more generally.\n",
    "\n",
    "Dask can scale down to your laptop laptop and up to a cluster.\n",
    "\n",
    "\n",
    "*  **High level Interface:**  Dask provides high-level Array, Bag, and DataFrame\n",
    "   collections that mimic NumPy, lists, and Pandas but can operate in parallel on\n",
    "   datasets that don't fit into memory.  Dask's high-level collections are\n",
    "   alternatives to NumPy and Pandas for large datasets.\n",
    "\n",
    "*  **Low Level Interface:** Dask provides dynamic task schedulers that\n",
    "   execute task graphs in parallel.  These execution engines power the\n",
    "   high-level collections mentioned above but can also power custom,\n",
    "   user-defined workloads.  These schedulers are low-latency (around 1ms) and\n",
    "   work hard to run computations in a small memory footprint.  Dask's\n",
    "   schedulers are an alternative to direct use of `threading` or\n",
    "   `multiprocessing` libraries in complex cases or other task scheduling\n",
    "   systems like `Luigi` or `IPython parallel`.\n",
    "   \n",
    "## DASK Ressources \n",
    "*  [DASK Website](http://dask.org)\n",
    "*  [Documentation](https://dask.pydata.org/en/latest/)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#%run prep.py"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# High Level Interfaces: Parallelizing NumPy and Pandas"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Pandas\n",
    "Pandas Dataframes are a very popular tool that provide full speadsheet and in-memory database functionality for the ***Python*** ecco system. It has become one of tstandard tools in Data Science and is also widely used for pre-processing / data wrangling tasks in machine learning\n",
    "\n",
    "* the ***DASK*** API for ***Pandas*** is analog to the one for ***NumPy*** and will not be discussed here, see https://docs.dask.org/en/latest/dataframe.html for details.\n",
    "\n",
    "## NumPy\n",
    "As seen before, NumPy arrays are the dominat array and matrix data structure in the ***Python*** data science and machine learning ecco system.\n",
    "* refresh your NumPy skills with this extra [tutorial](Extra_Material/Numpy.ipynb)\n",
    "\n",
    "### Functionality of Dask arrays\n",
    "\n",
    "* Arithmetic and scalar mathematics: +, *, exp, log, ...\n",
    "* Reductions along axes: sum(), mean(), std(), sum(axis=0), ...\n",
    "* Tensor contractions / dot products / matrix multiply: tensordot\n",
    "* Axis reordering / transpose: transpose\n",
    "* Slicing: x[:100, 500:100:-2]\n",
    "* Fancy indexing along single axes with lists or NumPy arrays: x[:, [10, 1, 5]]\n",
    "* Array protocols like __array__ and __array_ufunc__\n",
    "* Some linear algebra: svd, qr, solve, solve_triangular, lstsq\n",
    "\n",
    "**Full Array API: https://docs.dask.org/en/latest/array-api.html**\n",
    "\n",
    "### Limitations\n",
    "However, Dask Array does not implement the entire NumPy interface. Users expecting this will be disappointed. Notably, Dask Array lacks the following features:\n",
    "\n",
    "* Much of np.linalg has not been implemented. This has been done by a number of excellent BLAS/LAPACK implementations, and is the focus of numerous ongoing academic research projects\n",
    "* Arrays with unknown shapes do not support all operations\n",
    "* Operations like sort which are notoriously difficult to do in parallel, and are of somewhat diminished value on very large data (you rarely actually need a full sort). Often we include parallel-friendly alternatives like topk\n",
    "* Dask Array doesn’t implement operations like tolist that would be very inefficient for larger datasets. Likewise, it is very inefficient to iterate over a Dask array with for loops\n",
    "  \n",
    "**DASK is still evolving ...**\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Just get started!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 41.2 ms, sys: 326 µs, total: 41.5 ms\n",
      "Wall time: 41.7 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "49999800.50526341"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#let's parallelize a simple array opperations\n",
    "import numpy as np\n",
    "np_A = np.random.rand(10000,10000) #create random NumPy array\n",
    "%time np_A.sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "#now in parallel \n",
    "import dask.array as da\n",
    "da_A = da.from_array(np_A, chunks=(5000)) #create DASk array from existing NumPy array\n",
    "res = da_A.sum() #tell dask to compute the sum of A"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<table>\n",
       "<tr>\n",
       "<td>\n",
       "<table>\n",
       "  <thead>\n",
       "    <tr><td> </td><th> Array </th><th> Chunk </th></tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr><th> Bytes </th><td> 8 B </td> <td> 8 B </td></tr>\n",
       "    <tr><th> Shape </th><td> () </td> <td> () </td></tr>\n",
       "    <tr><th> Count </th><td> 10 Tasks </td><td> 1 Chunks </td></tr>\n",
       "    <tr><th> Type </th><td> float64 </td><td> numpy.ndarray </td></tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</td>\n",
       "<td>\n",
       "\n",
       "</td>\n",
       "</tr>\n",
       "</table>"
      ],
      "text/plain": [
       "dask.array<sum-aggregate, shape=(), dtype=float64, chunksize=(), chunktype=numpy.ndarray>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Where is my result?\n",
    "In a nutshell: ***DASK*** basically works like ***TensorFlow***:\n",
    "* First, it builds an execution graph\n",
    "* Then, we need to execute it by calling ``compute()``"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 107 ms, sys: 6.51 ms, total: 114 ms\n",
      "Wall time: 37 ms\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "50000035.97348415"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#call compute\n",
    "%time res.compute()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Why is this not faster?\n",
    "Well, building the graph, parallelizing it and scheduling it's execution causes some overhead...\n",
    "* probably, our problem was to small :-)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### What is happening at ``compute()`` ?\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/svg+xml": [
       "<svg height=\"637pt\" viewBox=\"0.00 0.00 342.19 637.08\" width=\"342pt\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n",
       "<g class=\"graph\" id=\"graph0\" transform=\"scale(1 1) rotate(0) translate(4 633.076)\">\n",
       "<title>%3</title>\n",
       "<polygon fill=\"white\" points=\"-4,4 -4,-633.076 338.194,-633.076 338.194,4 -4,4\" stroke=\"none\"/>\n",
       "<!-- &#45;1582066233564522576 -->\n",
       "<g class=\"node\" id=\"node1\"><title>-1582066233564522576</title>\n",
       "<polygon fill=\"none\" points=\"192.097,-629.076 138.097,-629.076 138.097,-593.076 192.097,-593.076 192.097,-629.076\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"165.097\" y=\"-607.376\">()</text>\n",
       "</g>\n",
       "<!-- &#45;8633483478624562631 -->\n",
       "<g class=\"node\" id=\"node2\"><title>-8633483478624562631</title>\n",
       "<ellipse cx=\"165.097\" cy=\"-482.983\" fill=\"none\" rx=\"74.187\" ry=\"74.187\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"165.097\" y=\"-479.283\">sum-aggregate</text>\n",
       "</g>\n",
       "<!-- &#45;8633483478624562631&#45;&gt;&#45;1582066233564522576 -->\n",
       "<g class=\"edge\" id=\"edge1\"><title>-8633483478624562631-&gt;-1582066233564522576</title>\n",
       "<path d=\"M165.097,-557.332C165.097,-566.317 165.097,-575.015 165.097,-582.67\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"161.597,-582.717 165.097,-592.717 168.597,-582.717 161.597,-582.717\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 3693635772764919281 -->\n",
       "<g class=\"node\" id=\"node3\"><title>3693635772764919281</title>\n",
       "<polygon fill=\"none\" points=\"318.097,-372.889 256.097,-372.889 256.097,-336.889 318.097,-336.889 318.097,-372.889\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"287.097\" y=\"-351.189\">(1, 1)</text>\n",
       "</g>\n",
       "<!-- 3693635772764919281&#45;&gt;&#45;8633483478624562631 -->\n",
       "<g class=\"edge\" id=\"edge2\"><title>3693635772764919281-&gt;-8633483478624562631</title>\n",
       "<path d=\"M270.643,-372.895C258.408,-385.541 240.914,-403.622 223.678,-421.436\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"220.762,-419.417 216.324,-429.037 225.792,-424.284 220.762,-419.417\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 3693635772766084330 -->\n",
       "<g class=\"node\" id=\"node4\"><title>3693635772766084330</title>\n",
       "<polygon fill=\"none\" points=\"238.097,-372.889 176.097,-372.889 176.097,-336.889 238.097,-336.889 238.097,-372.889\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"207.097\" y=\"-351.189\">(1, 0)</text>\n",
       "</g>\n",
       "<!-- 3693635772766084330&#45;&gt;&#45;8633483478624562631 -->\n",
       "<g class=\"edge\" id=\"edge3\"><title>3693635772766084330-&gt;-8633483478624562631</title>\n",
       "<path d=\"M201.433,-372.895C198.704,-381.088 195.215,-391.561 191.483,-402.765\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"188.142,-401.721 188.302,-412.315 194.783,-403.933 188.142,-401.721\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 3693634511569085507 -->\n",
       "<g class=\"node\" id=\"node5\"><title>3693634511569085507</title>\n",
       "<polygon fill=\"none\" points=\"73.0969,-372.889 11.0969,-372.889 11.0969,-336.889 73.0969,-336.889 73.0969,-372.889\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"42.0969\" y=\"-351.189\">(0, 0)</text>\n",
       "</g>\n",
       "<!-- 3693634511569085507&#45;&gt;&#45;8633483478624562631 -->\n",
       "<g class=\"edge\" id=\"edge4\"><title>3693634511569085507-&gt;-8633483478624562631</title>\n",
       "<path d=\"M58.6852,-372.895C71.0786,-385.6 88.8224,-403.79 106.278,-421.684\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"104.236,-424.604 113.724,-429.318 109.247,-419.716 104.236,-424.604\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 3693634511570250556 -->\n",
       "<g class=\"node\" id=\"node6\"><title>3693634511570250556</title>\n",
       "<polygon fill=\"none\" points=\"154.097,-372.889 92.0969,-372.889 92.0969,-336.889 154.097,-336.889 154.097,-372.889\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"123.097\" y=\"-351.189\">(0, 1)</text>\n",
       "</g>\n",
       "<!-- 3693634511570250556&#45;&gt;&#45;8633483478624562631 -->\n",
       "<g class=\"edge\" id=\"edge5\"><title>3693634511570250556-&gt;-8633483478624562631</title>\n",
       "<path d=\"M128.761,-372.895C131.49,-381.088 134.979,-391.561 138.711,-402.765\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"135.411,-403.933 141.891,-412.315 142.052,-401.721 135.411,-403.933\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- &#45;5375119454392779938 -->\n",
       "<g class=\"node\" id=\"node7\"><title>-5375119454392779938</title>\n",
       "<ellipse cx=\"38.0969\" cy=\"-275.542\" fill=\"none\" rx=\"25.1964\" ry=\"25.1964\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"38.0969\" y=\"-271.842\">sum</text>\n",
       "</g>\n",
       "<!-- &#45;5375119454392779938&#45;&gt;3693634511569085507 -->\n",
       "<g class=\"edge\" id=\"edge6\"><title>-5375119454392779938-&gt;3693634511569085507</title>\n",
       "<path d=\"M39.3625,-301.015C39.7763,-309.016 40.2388,-317.959 40.664,-326.181\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"37.1823,-326.629 41.1942,-336.434 44.173,-326.267 37.1823,-326.629\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 4601426706834120754 -->\n",
       "<g class=\"node\" id=\"node8\"><title>4601426706834120754</title>\n",
       "<polygon fill=\"none\" points=\"67.0969,-214.194 5.09689,-214.194 5.09689,-178.194 67.0969,-178.194 67.0969,-214.194\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"36.0969\" y=\"-192.494\">(0, 0)</text>\n",
       "</g>\n",
       "<!-- 4601426706834120754&#45;&gt;&#45;5375119454392779938 -->\n",
       "<g class=\"edge\" id=\"edge7\"><title>4601426706834120754-&gt;-5375119454392779938</title>\n",
       "<path d=\"M36.5408,-214.36C36.736,-221.909 36.9726,-231.062 37.2018,-239.927\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"33.7072,-240.182 37.4646,-250.088 40.7048,-240.001 33.7072,-240.182\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 1346814030596148853 -->\n",
       "<g class=\"node\" id=\"node9\"><title>1346814030596148853</title>\n",
       "<ellipse cx=\"123.097\" cy=\"-275.542\" fill=\"none\" rx=\"25.1964\" ry=\"25.1964\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"123.097\" y=\"-271.842\">sum</text>\n",
       "</g>\n",
       "<!-- 1346814030596148853&#45;&gt;3693634511570250556 -->\n",
       "<g class=\"edge\" id=\"edge8\"><title>1346814030596148853-&gt;3693634511570250556</title>\n",
       "<path d=\"M123.097,-301.015C123.097,-309.016 123.097,-317.959 123.097,-326.181\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"119.597,-326.434 123.097,-336.434 126.597,-326.435 119.597,-326.434\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 4601426706832955705 -->\n",
       "<g class=\"node\" id=\"node10\"><title>4601426706832955705</title>\n",
       "<polygon fill=\"none\" points=\"154.097,-214.194 92.0969,-214.194 92.0969,-178.194 154.097,-178.194 154.097,-214.194\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"123.097\" y=\"-192.494\">(0, 1)</text>\n",
       "</g>\n",
       "<!-- 4601426706832955705&#45;&gt;1346814030596148853 -->\n",
       "<g class=\"edge\" id=\"edge9\"><title>4601426706832955705-&gt;1346814030596148853</title>\n",
       "<path d=\"M123.097,-214.36C123.097,-221.909 123.097,-231.062 123.097,-239.927\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"119.597,-240.088 123.097,-250.088 126.597,-240.088 119.597,-240.088\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 2139068713748063955 -->\n",
       "<g class=\"node\" id=\"node11\"><title>2139068713748063955</title>\n",
       "<ellipse cx=\"208.097\" cy=\"-275.542\" fill=\"none\" rx=\"25.1964\" ry=\"25.1964\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"208.097\" y=\"-271.842\">sum</text>\n",
       "</g>\n",
       "<!-- 2139068713748063955&#45;&gt;3693635772766084330 -->\n",
       "<g class=\"edge\" id=\"edge10\"><title>2139068713748063955-&gt;3693635772766084330</title>\n",
       "<path d=\"M207.78,-301.015C207.677,-309.016 207.561,-317.959 207.455,-326.181\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"203.952,-326.39 207.323,-336.434 210.952,-326.481 203.952,-326.39\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 4601425445637121931 -->\n",
       "<g class=\"node\" id=\"node12\"><title>4601425445637121931</title>\n",
       "<polygon fill=\"none\" points=\"241.097,-214.194 179.097,-214.194 179.097,-178.194 241.097,-178.194 241.097,-214.194\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"210.097\" y=\"-192.494\">(1, 0)</text>\n",
       "</g>\n",
       "<!-- 4601425445637121931&#45;&gt;2139068713748063955 -->\n",
       "<g class=\"edge\" id=\"edge11\"><title>4601425445637121931-&gt;2139068713748063955</title>\n",
       "<path d=\"M209.653,-214.36C209.458,-221.909 209.221,-231.062 208.992,-239.927\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"205.489,-240.001 208.729,-250.088 212.487,-240.182 205.489,-240.001\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 1094806852214141980 -->\n",
       "<g class=\"node\" id=\"node13\"><title>1094806852214141980</title>\n",
       "<ellipse cx=\"288.097\" cy=\"-275.542\" fill=\"none\" rx=\"25.1964\" ry=\"25.1964\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"288.097\" y=\"-271.842\">sum</text>\n",
       "</g>\n",
       "<!-- 1094806852214141980&#45;&gt;3693635772764919281 -->\n",
       "<g class=\"edge\" id=\"edge12\"><title>1094806852214141980-&gt;3693635772764919281</title>\n",
       "<path d=\"M287.78,-301.015C287.677,-309.016 287.561,-317.959 287.455,-326.181\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"283.952,-326.39 287.323,-336.434 290.952,-326.481 283.952,-326.39\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 4601425445638286980 -->\n",
       "<g class=\"node\" id=\"node14\"><title>4601425445638286980</title>\n",
       "<polygon fill=\"none\" points=\"325.097,-214.194 263.097,-214.194 263.097,-178.194 325.097,-178.194 325.097,-214.194\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"294.097\" y=\"-192.494\">(1, 1)</text>\n",
       "</g>\n",
       "<!-- 4601425445638286980&#45;&gt;1094806852214141980 -->\n",
       "<g class=\"edge\" id=\"edge13\"><title>4601425445638286980-&gt;1094806852214141980</title>\n",
       "<path d=\"M292.765,-214.36C292.18,-221.909 291.47,-231.062 290.782,-239.927\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"287.278,-239.847 289.994,-250.088 294.257,-240.389 287.278,-239.847\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 5827827906509150347 -->\n",
       "<g class=\"node\" id=\"node15\"><title>5827827906509150347</title>\n",
       "<ellipse cx=\"35.0969\" cy=\"-107.097\" fill=\"none\" rx=\"35.194\" ry=\"35.194\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"35.0969\" y=\"-103.397\">array</text>\n",
       "</g>\n",
       "<!-- 5827827906509150347&#45;&gt;4601426706834120754 -->\n",
       "<g class=\"edge\" id=\"edge14\"><title>5827827906509150347-&gt;4601426706834120754</title>\n",
       "<path d=\"M35.4921,-142.52C35.5882,-150.889 35.6897,-159.728 35.7815,-167.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"32.2846,-168.027 35.8993,-177.986 39.2842,-167.947 32.2846,-168.027\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 7653523461393938259 -->\n",
       "<g class=\"node\" id=\"node16\"><title>7653523461393938259</title>\n",
       "<polygon fill=\"none\" points=\"194.097,-36 140.097,-36 140.097,-0 194.097,-0 194.097,-36\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 7653523461393938259&#45;&gt;5827827906509150347 -->\n",
       "<g class=\"edge\" id=\"edge15\"><title>7653523461393938259-&gt;5827827906509150347</title>\n",
       "<path d=\"M140.015,-34.2079C122.345,-44.3971 98.9591,-58.3917 79.0969,-72 76.2193,-73.9715 73.2786,-76.0649 70.3496,-78.2065\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"68.1012,-75.5175 62.1886,-84.3091 72.2932,-81.1235 68.1012,-75.5175\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 4773542186639251572 -->\n",
       "<g class=\"node\" id=\"node17\"><title>4773542186639251572</title>\n",
       "<ellipse cx=\"123.097\" cy=\"-107.097\" fill=\"none\" rx=\"35.194\" ry=\"35.194\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"123.097\" y=\"-103.397\">array</text>\n",
       "</g>\n",
       "<!-- 7653523461393938259&#45;&gt;4773542186639251572 -->\n",
       "<g class=\"edge\" id=\"edge17\"><title>7653523461393938259-&gt;4773542186639251572</title>\n",
       "<path d=\"M158.404,-36.2075C154.026,-44.8732 148.522,-55.7678 143.172,-66.3595\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"139.926,-65.0217 138.541,-75.5256 146.174,-68.1782 139.926,-65.0217\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 783614570147352598 -->\n",
       "<g class=\"node\" id=\"node18\"><title>783614570147352598</title>\n",
       "<ellipse cx=\"211.097\" cy=\"-107.097\" fill=\"none\" rx=\"35.194\" ry=\"35.194\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"211.097\" y=\"-103.397\">array</text>\n",
       "</g>\n",
       "<!-- 7653523461393938259&#45;&gt;783614570147352598 -->\n",
       "<g class=\"edge\" id=\"edge19\"><title>7653523461393938259-&gt;783614570147352598</title>\n",
       "<path d=\"M175.79,-36.2075C180.168,-44.8732 185.671,-55.7678 191.022,-66.3595\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"188.02,-68.1782 195.653,-75.5256 194.268,-65.0217 188.02,-68.1782\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 2333316317557556173 -->\n",
       "<g class=\"node\" id=\"node19\"><title>2333316317557556173</title>\n",
       "<ellipse cx=\"299.097\" cy=\"-107.097\" fill=\"none\" rx=\"35.194\" ry=\"35.194\" stroke=\"black\"/>\n",
       "<text font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"299.097\" y=\"-103.397\">array</text>\n",
       "</g>\n",
       "<!-- 7653523461393938259&#45;&gt;2333316317557556173 -->\n",
       "<g class=\"edge\" id=\"edge21\"><title>7653523461393938259-&gt;2333316317557556173</title>\n",
       "<path d=\"M194.179,-34.2079C211.849,-44.3971 235.235,-58.3917 255.097,-72 257.974,-73.9715 260.915,-76.0649 263.844,-78.2065\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"261.901,-81.1235 272.005,-84.3091 266.093,-75.5175 261.901,-81.1235\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 4773542186639251572&#45;&gt;4601426706832955705 -->\n",
       "<g class=\"edge\" id=\"edge16\"><title>4773542186639251572-&gt;4601426706832955705</title>\n",
       "<path d=\"M123.097,-142.52C123.097,-150.889 123.097,-159.728 123.097,-167.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"119.597,-167.986 123.097,-177.986 126.597,-167.986 119.597,-167.986\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 783614570147352598&#45;&gt;4601425445637121931 -->\n",
       "<g class=\"edge\" id=\"edge18\"><title>783614570147352598-&gt;4601425445637121931</title>\n",
       "<path d=\"M210.702,-142.52C210.606,-150.889 210.504,-159.728 210.412,-167.725\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"206.91,-167.947 210.294,-177.986 213.909,-168.027 206.91,-167.947\" stroke=\"black\"/>\n",
       "</g>\n",
       "<!-- 2333316317557556173&#45;&gt;4601425445638286980 -->\n",
       "<g class=\"edge\" id=\"edge20\"><title>2333316317557556173-&gt;4601425445638286980</title>\n",
       "<path d=\"M297.121,-142.52C296.635,-150.984 296.121,-159.929 295.658,-167.998\" fill=\"none\" stroke=\"black\"/>\n",
       "<polygon fill=\"black\" points=\"292.164,-167.802 295.085,-177.986 299.152,-168.203 292.164,-167.802\" stroke=\"black\"/>\n",
       "</g>\n",
       "</g>\n",
       "</svg>"
      ],
      "text/plain": [
       "<IPython.core.display.SVG object>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#have look at the actual compute graph build by DASK\n",
    "res.visualize(filename='graph.svg')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ahh - MapReduce!\n",
    "* looks like ***DASK*** automatically implemented MapReduce\n",
    "* but, why did it use 4 parallel paths?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Block concept of DASK arrays\n",
    "* ***DASK*** splitts arrays into blocks \n",
    "* parallelization (here MapReduce) runs over Blocks\n",
    "* Blocks sizes are definde by the ``chunks`` \n",
    "\n",
    "```\n",
    "#recall array creation\n",
    "da_A = da.from_array(np_A, chunks=(5000))\n",
    "```\n",
    "\n",
    "<img SRC=\"IMG/array.png\">"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((5000, 5000), (5000, 5000))"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#get size of chucks\n",
    "da_A.chunks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(5000, 5000)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "da_A.chunksize"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#change chunk size\n",
    "#WARNING: can be very expensive - better think about the chunk size before ...\n",
    "B=da_A.rechunk(1000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((1000, 1000, 1000, 1000, 1000, 1000, 1000, 1000, 1000, 1000),\n",
       " (1000, 1000, 1000, 1000, 1000, 1000, 1000, 1000, 1000, 1000))"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "B.chunks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "res=B.sum()\n",
    "graph=res.visualize(filename='graph.svg')#saving graph to file"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Is this running in Parallel ?\n",
    "* open ternminal\n",
    "* start ``top``\n",
    "* execute ``compute()``"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "67.7 ms ± 623 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit #measure mean of 10 runs\n",
    "res.compute()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Memory Usage\n",
    "By processing each block at a time, ***DASK*** is not only able tu run in parallel, but also needs only tthe memory for the bock size...\n",
    "* allows easy processing of **larger than memory** data sets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 13.2 s, sys: 612 ms, total: 13.8 s\n",
      "Wall time: 13.8 s\n",
      "CPU times: user 255 ms, sys: 4 µs, total: 255 ms\n",
      "Wall time: 254 ms\n"
     ]
    }
   ],
   "source": [
    "#a larger example in NumPy\n",
    "%time x = np.random.normal(10, 0.1, size=(20000, 20000)) \n",
    "%time y = x.mean(axis=0) #compute mean along first axis\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 19.4 s, sys: 251 ms, total: 19.7 s\n",
      "Wall time: 1.12 s\n"
     ]
    }
   ],
   "source": [
    "#same in DASK\n",
    "x = da.random.normal(10, 0.1, size=(20000, 20000), chunks=(1000, 1000))\n",
    "y = x.mean(axis=0) \n",
    "%time res=y.compute() "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Distributed Execution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "import carme #import cluster env\n",
    "from dask.distributed import Client #import distributed client"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In our cluster enviroment, the ***DASK*** Cluster is automatically started for multi node jobs. See DASK documentation on how to setup a ***DASK*** cluster: http://distributed.dask.org/en/latest/setup.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "client = Client(MASTER-IP) #get new DASK cluster client"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tcp://192.168.152.21:33727\n",
      "tcp://192.168.152.21:38699\n",
      "tcp://192.168.152.26:33001\n",
      "tcp://192.168.152.26:34253\n"
     ]
    }
   ],
   "source": [
    "#see how many workers we have\n",
    "workers=client.scheduler_info()['workers']\n",
    "for w in workers:\n",
    "    print(w)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Submit work to the Cluster"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "#simple remote functions\n",
    "def square(x):\n",
    "    return x ** 2\n",
    "\n",
    "def neg(x):\n",
    "    return -x\n",
    "\n",
    "#using the map future, functions will be scheduled for execution - NO graph building and compute()\n",
    "A = client.map(square, range(1000))\n",
    "B = client.map(neg, A)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "total = client.submit(sum, B)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "done:  -332833500\n"
     ]
    }
   ],
   "source": [
    "\n",
    "res = total.result()\n",
    "print (\"done: \",res)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Arrays on the Cluster\n",
    "Starting a cluster client automatically cause the high level interfaces to use the cluster"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "np_A = np.random.rand(10000,10000) #create random NumPy array\n",
    "import dask.array as da\n",
    "cluster_A = da.from_array(np_A, chunks=(5000)) #create DASk array from existing NumPy array\n",
    "res = cluster_A.sum() #tell dask to compute the sum of A"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 13.1 ms, sys: 210 ms, total: 223 ms\n",
      "Wall time: 8.13 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "50003479.46830079"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "%time res.compute()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Using ***Scikit-Learn*** with ***Dask***\n",
    "Note: This is quit new and under heavy development"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import make_classification\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-1.06377997,  0.67640868,  1.06935647, -0.21758002,  0.46021477,\n",
       "        -0.39916689, -0.07918751,  1.20938491, -0.78531472, -0.17218611,\n",
       "        -1.08535744, -0.99311895,  0.30693511,  0.06405769, -1.0542328 ,\n",
       "        -0.52749607, -0.0741832 , -0.35562842,  1.05721416, -0.90259159],\n",
       "       [ 0.0708476 , -1.69528125,  2.44944917, -0.5304942 , -0.93296221,\n",
       "         2.86520354,  2.43572851, -1.61850016,  1.30071691,  0.34840246,\n",
       "         0.54493439,  0.22532411,  0.60556322, -0.19210097, -0.06802699,\n",
       "         0.9716812 , -1.79204799,  0.01708348, -0.37566904, -0.62323644],\n",
       "       [ 0.94028404, -0.49214582,  0.67795602, -0.22775445,  1.40175261,\n",
       "         1.23165333, -0.77746425,  0.01561602,  1.33171299,  1.08477266,\n",
       "        -0.97805157, -0.05012039,  0.94838552, -0.17342825, -0.47767184,\n",
       "         0.76089649,  1.00115812, -0.06946407,  1.35904607, -1.18958963],\n",
       "       [-0.29951677,  0.75988955,  0.18280267, -1.55023271,  0.33821802,\n",
       "         0.36324148, -2.10052547, -0.4380675 , -0.16639343, -0.34083531,\n",
       "         0.42435643,  1.17872434,  2.8314804 ,  0.14241375, -0.20281911,\n",
       "         2.40571546,  0.31330473,  0.40435568, -0.28754632, -2.8478034 ],\n",
       "       [-2.63062675,  0.23103376,  0.04246253,  0.47885055,  1.54674163,\n",
       "         1.6379556 , -1.53207229, -0.73444479,  0.46585484,  0.4738362 ,\n",
       "         0.98981401, -1.06119392, -0.88887952,  1.23840892, -0.57282854,\n",
       "        -1.27533949,  1.0030065 , -0.47712843,  0.09853558,  0.52780407]])"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#generate some random classification data\n",
    "X, y = make_classification(n_samples=1000, random_state=0)\n",
    "X[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# SKlearn Hyper-Parameter Search\n",
    "param_grid = {\"C\": [0.001, 0.01, 0.1, 0.5, 1.0, 2.0, 5.0, 10.0],\n",
    "              \"kernel\": ['rbf', 'poly', 'sigmoid'],\n",
    "              \"shrinking\": [True, False]}\n",
    "\n",
    "grid_search = GridSearchCV(SVC(gamma='auto', random_state=0, probability=True),\n",
    "                           param_grid=param_grid,\n",
    "                           return_train_score=False,\n",
    "                           iid=True,\n",
    "                           cv=3,\n",
    "                           n_jobs=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 396 ms, sys: 882 ms, total: 1.28 s\n",
      "Wall time: 4.97 s\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=3, error_score='raise-deprecating',\n",
       "       estimator=SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0,\n",
       "  decision_function_shape='ovr', degree=3, gamma='auto', kernel='rbf',\n",
       "  max_iter=-1, probability=True, random_state=0, shrinking=True, tol=0.001,\n",
       "  verbose=False),\n",
       "       fit_params=None, iid=True, n_jobs=-1,\n",
       "       param_grid={'C': [0.001, 0.01, 0.1, 0.5, 1.0, 2.0, 5.0, 10.0], 'kernel': ['rbf', 'poly', 'sigmoid'], 'shrinking': [True, False]},\n",
       "       pre_dispatch='2*n_jobs', refit=True, return_train_score=False,\n",
       "       scoring=None, verbose=0)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "%time grid_search.fit(X, y)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 4 µs, sys: 0 ns, total: 4 µs\n",
      "Wall time: 7.63 µs\n"
     ]
    }
   ],
   "source": [
    "#noe map this to the cluster\n",
    "from sklearn.externals import joblib\n",
    "\n",
    "%time\n",
    "with joblib.parallel_backend('dask'):\n",
    "    grid_search.fit(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>mean_fit_time</th>\n",
       "      <th>std_fit_time</th>\n",
       "      <th>mean_score_time</th>\n",
       "      <th>std_score_time</th>\n",
       "      <th>param_C</th>\n",
       "      <th>param_kernel</th>\n",
       "      <th>param_shrinking</th>\n",
       "      <th>params</th>\n",
       "      <th>split0_test_score</th>\n",
       "      <th>split1_test_score</th>\n",
       "      <th>split2_test_score</th>\n",
       "      <th>mean_test_score</th>\n",
       "      <th>std_test_score</th>\n",
       "      <th>rank_test_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.083925</td>\n",
       "      <td>0.006859</td>\n",
       "      <td>0.006478</td>\n",
       "      <td>0.000102</td>\n",
       "      <td>0.001</td>\n",
       "      <td>rbf</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.001, 'kernel': 'rbf', 'shrinking': True}</td>\n",
       "      <td>0.502994</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.502</td>\n",
       "      <td>0.000704</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.071765</td>\n",
       "      <td>0.002903</td>\n",
       "      <td>0.006412</td>\n",
       "      <td>0.000163</td>\n",
       "      <td>0.001</td>\n",
       "      <td>rbf</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.001, 'kernel': 'rbf', 'shrinking': False}</td>\n",
       "      <td>0.502994</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.502</td>\n",
       "      <td>0.000704</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.044910</td>\n",
       "      <td>0.003309</td>\n",
       "      <td>0.003728</td>\n",
       "      <td>0.000225</td>\n",
       "      <td>0.001</td>\n",
       "      <td>poly</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.001, 'kernel': 'poly', 'shrinking': True}</td>\n",
       "      <td>0.502994</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.502</td>\n",
       "      <td>0.000704</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.041414</td>\n",
       "      <td>0.001589</td>\n",
       "      <td>0.003569</td>\n",
       "      <td>0.000078</td>\n",
       "      <td>0.001</td>\n",
       "      <td>poly</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.001, 'kernel': 'poly', 'shrinking': Fa...</td>\n",
       "      <td>0.502994</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.502</td>\n",
       "      <td>0.000704</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.092081</td>\n",
       "      <td>0.003575</td>\n",
       "      <td>0.008514</td>\n",
       "      <td>0.000203</td>\n",
       "      <td>0.001</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.001, 'kernel': 'sigmoid', 'shrinking':...</td>\n",
       "      <td>0.502994</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.502</td>\n",
       "      <td>0.000704</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0.089387</td>\n",
       "      <td>0.002987</td>\n",
       "      <td>0.008993</td>\n",
       "      <td>0.000493</td>\n",
       "      <td>0.001</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.001, 'kernel': 'sigmoid', 'shrinking':...</td>\n",
       "      <td>0.502994</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.502</td>\n",
       "      <td>0.000704</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0.072298</td>\n",
       "      <td>0.001636</td>\n",
       "      <td>0.006508</td>\n",
       "      <td>0.000148</td>\n",
       "      <td>0.01</td>\n",
       "      <td>rbf</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.01, 'kernel': 'rbf', 'shrinking': True}</td>\n",
       "      <td>0.502994</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.502</td>\n",
       "      <td>0.000704</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0.073107</td>\n",
       "      <td>0.004890</td>\n",
       "      <td>0.006940</td>\n",
       "      <td>0.000452</td>\n",
       "      <td>0.01</td>\n",
       "      <td>rbf</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.01, 'kernel': 'rbf', 'shrinking': False}</td>\n",
       "      <td>0.502994</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.501502</td>\n",
       "      <td>0.502</td>\n",
       "      <td>0.000704</td>\n",
       "      <td>41</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0.043833</td>\n",
       "      <td>0.003328</td>\n",
       "      <td>0.003759</td>\n",
       "      <td>0.000152</td>\n",
       "      <td>0.01</td>\n",
       "      <td>poly</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.01, 'kernel': 'poly', 'shrinking': True}</td>\n",
       "      <td>0.508982</td>\n",
       "      <td>0.510511</td>\n",
       "      <td>0.504505</td>\n",
       "      <td>0.508</td>\n",
       "      <td>0.002547</td>\n",
       "      <td>39</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0.040512</td>\n",
       "      <td>0.001182</td>\n",
       "      <td>0.003584</td>\n",
       "      <td>0.000223</td>\n",
       "      <td>0.01</td>\n",
       "      <td>poly</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.01, 'kernel': 'poly', 'shrinking': False}</td>\n",
       "      <td>0.508982</td>\n",
       "      <td>0.510511</td>\n",
       "      <td>0.504505</td>\n",
       "      <td>0.508</td>\n",
       "      <td>0.002547</td>\n",
       "      <td>39</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>0.087808</td>\n",
       "      <td>0.000611</td>\n",
       "      <td>0.008022</td>\n",
       "      <td>0.000208</td>\n",
       "      <td>0.01</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.01, 'kernel': 'sigmoid', 'shrinking': ...</td>\n",
       "      <td>0.949102</td>\n",
       "      <td>0.924925</td>\n",
       "      <td>0.957958</td>\n",
       "      <td>0.944</td>\n",
       "      <td>0.013955</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>0.086106</td>\n",
       "      <td>0.002129</td>\n",
       "      <td>0.008053</td>\n",
       "      <td>0.000296</td>\n",
       "      <td>0.01</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.01, 'kernel': 'sigmoid', 'shrinking': ...</td>\n",
       "      <td>0.949102</td>\n",
       "      <td>0.924925</td>\n",
       "      <td>0.957958</td>\n",
       "      <td>0.944</td>\n",
       "      <td>0.013955</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>0.057442</td>\n",
       "      <td>0.001495</td>\n",
       "      <td>0.004915</td>\n",
       "      <td>0.000294</td>\n",
       "      <td>0.1</td>\n",
       "      <td>rbf</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.1, 'kernel': 'rbf', 'shrinking': True}</td>\n",
       "      <td>0.949102</td>\n",
       "      <td>0.933934</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.946</td>\n",
       "      <td>0.008854</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>0.058186</td>\n",
       "      <td>0.002424</td>\n",
       "      <td>0.004919</td>\n",
       "      <td>0.000311</td>\n",
       "      <td>0.1</td>\n",
       "      <td>rbf</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.1, 'kernel': 'rbf', 'shrinking': False}</td>\n",
       "      <td>0.949102</td>\n",
       "      <td>0.933934</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.946</td>\n",
       "      <td>0.008854</td>\n",
       "      <td>11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>0.037398</td>\n",
       "      <td>0.001929</td>\n",
       "      <td>0.003016</td>\n",
       "      <td>0.000173</td>\n",
       "      <td>0.1</td>\n",
       "      <td>poly</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.1, 'kernel': 'poly', 'shrinking': True}</td>\n",
       "      <td>0.871257</td>\n",
       "      <td>0.849850</td>\n",
       "      <td>0.918919</td>\n",
       "      <td>0.880</td>\n",
       "      <td>0.028855</td>\n",
       "      <td>37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>0.039730</td>\n",
       "      <td>0.002981</td>\n",
       "      <td>0.002995</td>\n",
       "      <td>0.000188</td>\n",
       "      <td>0.1</td>\n",
       "      <td>poly</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.1, 'kernel': 'poly', 'shrinking': False}</td>\n",
       "      <td>0.871257</td>\n",
       "      <td>0.849850</td>\n",
       "      <td>0.918919</td>\n",
       "      <td>0.880</td>\n",
       "      <td>0.028855</td>\n",
       "      <td>37</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>0.047046</td>\n",
       "      <td>0.002367</td>\n",
       "      <td>0.004072</td>\n",
       "      <td>0.000113</td>\n",
       "      <td>0.1</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.1, 'kernel': 'sigmoid', 'shrinking': T...</td>\n",
       "      <td>0.940120</td>\n",
       "      <td>0.930931</td>\n",
       "      <td>0.960961</td>\n",
       "      <td>0.944</td>\n",
       "      <td>0.012558</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>0.048430</td>\n",
       "      <td>0.001255</td>\n",
       "      <td>0.004129</td>\n",
       "      <td>0.000211</td>\n",
       "      <td>0.1</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.1, 'kernel': 'sigmoid', 'shrinking': F...</td>\n",
       "      <td>0.940120</td>\n",
       "      <td>0.930931</td>\n",
       "      <td>0.960961</td>\n",
       "      <td>0.944</td>\n",
       "      <td>0.012558</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>0.039987</td>\n",
       "      <td>0.001993</td>\n",
       "      <td>0.003071</td>\n",
       "      <td>0.000173</td>\n",
       "      <td>0.5</td>\n",
       "      <td>rbf</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.5, 'kernel': 'rbf', 'shrinking': True}</td>\n",
       "      <td>0.955090</td>\n",
       "      <td>0.927928</td>\n",
       "      <td>0.963964</td>\n",
       "      <td>0.949</td>\n",
       "      <td>0.015324</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>0.039662</td>\n",
       "      <td>0.000839</td>\n",
       "      <td>0.003157</td>\n",
       "      <td>0.000106</td>\n",
       "      <td>0.5</td>\n",
       "      <td>rbf</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.5, 'kernel': 'rbf', 'shrinking': False}</td>\n",
       "      <td>0.955090</td>\n",
       "      <td>0.927928</td>\n",
       "      <td>0.963964</td>\n",
       "      <td>0.949</td>\n",
       "      <td>0.015324</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>0.031550</td>\n",
       "      <td>0.000656</td>\n",
       "      <td>0.002340</td>\n",
       "      <td>0.000060</td>\n",
       "      <td>0.5</td>\n",
       "      <td>poly</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.5, 'kernel': 'poly', 'shrinking': True}</td>\n",
       "      <td>0.943114</td>\n",
       "      <td>0.930931</td>\n",
       "      <td>0.957958</td>\n",
       "      <td>0.944</td>\n",
       "      <td>0.011046</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>0.030683</td>\n",
       "      <td>0.001059</td>\n",
       "      <td>0.002204</td>\n",
       "      <td>0.000078</td>\n",
       "      <td>0.5</td>\n",
       "      <td>poly</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.5, 'kernel': 'poly', 'shrinking': False}</td>\n",
       "      <td>0.943114</td>\n",
       "      <td>0.930931</td>\n",
       "      <td>0.957958</td>\n",
       "      <td>0.944</td>\n",
       "      <td>0.011046</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>0.030276</td>\n",
       "      <td>0.002463</td>\n",
       "      <td>0.002496</td>\n",
       "      <td>0.000065</td>\n",
       "      <td>0.5</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 0.5, 'kernel': 'sigmoid', 'shrinking': T...</td>\n",
       "      <td>0.946108</td>\n",
       "      <td>0.930931</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.944</td>\n",
       "      <td>0.009916</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>0.030696</td>\n",
       "      <td>0.002513</td>\n",
       "      <td>0.002537</td>\n",
       "      <td>0.000090</td>\n",
       "      <td>0.5</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 0.5, 'kernel': 'sigmoid', 'shrinking': F...</td>\n",
       "      <td>0.946108</td>\n",
       "      <td>0.930931</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.944</td>\n",
       "      <td>0.009916</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>0.037658</td>\n",
       "      <td>0.000464</td>\n",
       "      <td>0.002740</td>\n",
       "      <td>0.000090</td>\n",
       "      <td>1</td>\n",
       "      <td>rbf</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 1.0, 'kernel': 'rbf', 'shrinking': True}</td>\n",
       "      <td>0.952096</td>\n",
       "      <td>0.939940</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.949</td>\n",
       "      <td>0.006507</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>0.039102</td>\n",
       "      <td>0.002538</td>\n",
       "      <td>0.002999</td>\n",
       "      <td>0.000165</td>\n",
       "      <td>1</td>\n",
       "      <td>rbf</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 1.0, 'kernel': 'rbf', 'shrinking': False}</td>\n",
       "      <td>0.952096</td>\n",
       "      <td>0.939940</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.949</td>\n",
       "      <td>0.006507</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>0.029608</td>\n",
       "      <td>0.000502</td>\n",
       "      <td>0.001972</td>\n",
       "      <td>0.000092</td>\n",
       "      <td>1</td>\n",
       "      <td>poly</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 1.0, 'kernel': 'poly', 'shrinking': True}</td>\n",
       "      <td>0.946108</td>\n",
       "      <td>0.927928</td>\n",
       "      <td>0.957958</td>\n",
       "      <td>0.944</td>\n",
       "      <td>0.012344</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>0.030518</td>\n",
       "      <td>0.001784</td>\n",
       "      <td>0.002116</td>\n",
       "      <td>0.000030</td>\n",
       "      <td>1</td>\n",
       "      <td>poly</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 1.0, 'kernel': 'poly', 'shrinking': False}</td>\n",
       "      <td>0.946108</td>\n",
       "      <td>0.927928</td>\n",
       "      <td>0.957958</td>\n",
       "      <td>0.944</td>\n",
       "      <td>0.012344</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>0.027252</td>\n",
       "      <td>0.001084</td>\n",
       "      <td>0.002077</td>\n",
       "      <td>0.000110</td>\n",
       "      <td>1</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 1.0, 'kernel': 'sigmoid', 'shrinking': T...</td>\n",
       "      <td>0.952096</td>\n",
       "      <td>0.924925</td>\n",
       "      <td>0.957958</td>\n",
       "      <td>0.945</td>\n",
       "      <td>0.014385</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>0.026960</td>\n",
       "      <td>0.001462</td>\n",
       "      <td>0.002009</td>\n",
       "      <td>0.000136</td>\n",
       "      <td>1</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 1.0, 'kernel': 'sigmoid', 'shrinking': F...</td>\n",
       "      <td>0.952096</td>\n",
       "      <td>0.924925</td>\n",
       "      <td>0.957958</td>\n",
       "      <td>0.945</td>\n",
       "      <td>0.014385</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>0.039095</td>\n",
       "      <td>0.002018</td>\n",
       "      <td>0.002659</td>\n",
       "      <td>0.000164</td>\n",
       "      <td>2</td>\n",
       "      <td>rbf</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 2.0, 'kernel': 'rbf', 'shrinking': True}</td>\n",
       "      <td>0.952096</td>\n",
       "      <td>0.939940</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.949</td>\n",
       "      <td>0.006507</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>0.037873</td>\n",
       "      <td>0.001429</td>\n",
       "      <td>0.002619</td>\n",
       "      <td>0.000115</td>\n",
       "      <td>2</td>\n",
       "      <td>rbf</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 2.0, 'kernel': 'rbf', 'shrinking': False}</td>\n",
       "      <td>0.952096</td>\n",
       "      <td>0.939940</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.949</td>\n",
       "      <td>0.006507</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>0.030128</td>\n",
       "      <td>0.001376</td>\n",
       "      <td>0.001756</td>\n",
       "      <td>0.000073</td>\n",
       "      <td>2</td>\n",
       "      <td>poly</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 2.0, 'kernel': 'poly', 'shrinking': True}</td>\n",
       "      <td>0.940120</td>\n",
       "      <td>0.930931</td>\n",
       "      <td>0.942943</td>\n",
       "      <td>0.938</td>\n",
       "      <td>0.005126</td>\n",
       "      <td>27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>0.031574</td>\n",
       "      <td>0.001522</td>\n",
       "      <td>0.001832</td>\n",
       "      <td>0.000064</td>\n",
       "      <td>2</td>\n",
       "      <td>poly</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 2.0, 'kernel': 'poly', 'shrinking': False}</td>\n",
       "      <td>0.940120</td>\n",
       "      <td>0.930931</td>\n",
       "      <td>0.942943</td>\n",
       "      <td>0.938</td>\n",
       "      <td>0.005126</td>\n",
       "      <td>27</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>0.024479</td>\n",
       "      <td>0.001015</td>\n",
       "      <td>0.001825</td>\n",
       "      <td>0.000060</td>\n",
       "      <td>2</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 2.0, 'kernel': 'sigmoid', 'shrinking': T...</td>\n",
       "      <td>0.952096</td>\n",
       "      <td>0.918919</td>\n",
       "      <td>0.948949</td>\n",
       "      <td>0.940</td>\n",
       "      <td>0.014951</td>\n",
       "      <td>25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>0.024857</td>\n",
       "      <td>0.001434</td>\n",
       "      <td>0.001848</td>\n",
       "      <td>0.000091</td>\n",
       "      <td>2</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 2.0, 'kernel': 'sigmoid', 'shrinking': F...</td>\n",
       "      <td>0.952096</td>\n",
       "      <td>0.918919</td>\n",
       "      <td>0.948949</td>\n",
       "      <td>0.940</td>\n",
       "      <td>0.014951</td>\n",
       "      <td>25</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>0.039802</td>\n",
       "      <td>0.001808</td>\n",
       "      <td>0.002815</td>\n",
       "      <td>0.000141</td>\n",
       "      <td>5</td>\n",
       "      <td>rbf</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 5.0, 'kernel': 'rbf', 'shrinking': True}</td>\n",
       "      <td>0.955090</td>\n",
       "      <td>0.936937</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.949</td>\n",
       "      <td>0.008524</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>0.040171</td>\n",
       "      <td>0.001478</td>\n",
       "      <td>0.002748</td>\n",
       "      <td>0.000162</td>\n",
       "      <td>5</td>\n",
       "      <td>rbf</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 5.0, 'kernel': 'rbf', 'shrinking': False}</td>\n",
       "      <td>0.955090</td>\n",
       "      <td>0.936937</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.949</td>\n",
       "      <td>0.008524</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>0.029475</td>\n",
       "      <td>0.000386</td>\n",
       "      <td>0.001659</td>\n",
       "      <td>0.000075</td>\n",
       "      <td>5</td>\n",
       "      <td>poly</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 5.0, 'kernel': 'poly', 'shrinking': True}</td>\n",
       "      <td>0.925150</td>\n",
       "      <td>0.921922</td>\n",
       "      <td>0.945946</td>\n",
       "      <td>0.931</td>\n",
       "      <td>0.010642</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>0.030795</td>\n",
       "      <td>0.000919</td>\n",
       "      <td>0.001737</td>\n",
       "      <td>0.000054</td>\n",
       "      <td>5</td>\n",
       "      <td>poly</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 5.0, 'kernel': 'poly', 'shrinking': False}</td>\n",
       "      <td>0.925150</td>\n",
       "      <td>0.921922</td>\n",
       "      <td>0.945946</td>\n",
       "      <td>0.931</td>\n",
       "      <td>0.010642</td>\n",
       "      <td>29</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>0.022427</td>\n",
       "      <td>0.001340</td>\n",
       "      <td>0.001594</td>\n",
       "      <td>0.000128</td>\n",
       "      <td>5</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 5.0, 'kernel': 'sigmoid', 'shrinking': T...</td>\n",
       "      <td>0.931138</td>\n",
       "      <td>0.909910</td>\n",
       "      <td>0.942943</td>\n",
       "      <td>0.928</td>\n",
       "      <td>0.013661</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>41</th>\n",
       "      <td>0.022493</td>\n",
       "      <td>0.001067</td>\n",
       "      <td>0.001581</td>\n",
       "      <td>0.000131</td>\n",
       "      <td>5</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 5.0, 'kernel': 'sigmoid', 'shrinking': F...</td>\n",
       "      <td>0.931138</td>\n",
       "      <td>0.909910</td>\n",
       "      <td>0.942943</td>\n",
       "      <td>0.928</td>\n",
       "      <td>0.013661</td>\n",
       "      <td>33</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>0.041968</td>\n",
       "      <td>0.001455</td>\n",
       "      <td>0.002683</td>\n",
       "      <td>0.000119</td>\n",
       "      <td>10</td>\n",
       "      <td>rbf</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 10.0, 'kernel': 'rbf', 'shrinking': True}</td>\n",
       "      <td>0.943114</td>\n",
       "      <td>0.945946</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.948</td>\n",
       "      <td>0.005048</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>0.040902</td>\n",
       "      <td>0.001419</td>\n",
       "      <td>0.002711</td>\n",
       "      <td>0.000160</td>\n",
       "      <td>10</td>\n",
       "      <td>rbf</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 10.0, 'kernel': 'rbf', 'shrinking': False}</td>\n",
       "      <td>0.943114</td>\n",
       "      <td>0.945946</td>\n",
       "      <td>0.954955</td>\n",
       "      <td>0.948</td>\n",
       "      <td>0.005048</td>\n",
       "      <td>9</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>0.029388</td>\n",
       "      <td>0.000919</td>\n",
       "      <td>0.001604</td>\n",
       "      <td>0.000077</td>\n",
       "      <td>10</td>\n",
       "      <td>poly</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 10.0, 'kernel': 'poly', 'shrinking': True}</td>\n",
       "      <td>0.919162</td>\n",
       "      <td>0.915916</td>\n",
       "      <td>0.945946</td>\n",
       "      <td>0.927</td>\n",
       "      <td>0.013452</td>\n",
       "      <td>35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>0.029630</td>\n",
       "      <td>0.000647</td>\n",
       "      <td>0.001551</td>\n",
       "      <td>0.000060</td>\n",
       "      <td>10</td>\n",
       "      <td>poly</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 10.0, 'kernel': 'poly', 'shrinking': False}</td>\n",
       "      <td>0.919162</td>\n",
       "      <td>0.915916</td>\n",
       "      <td>0.945946</td>\n",
       "      <td>0.927</td>\n",
       "      <td>0.013452</td>\n",
       "      <td>35</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>0.021693</td>\n",
       "      <td>0.002314</td>\n",
       "      <td>0.001403</td>\n",
       "      <td>0.000117</td>\n",
       "      <td>10</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>True</td>\n",
       "      <td>{'C': 10.0, 'kernel': 'sigmoid', 'shrinking': ...</td>\n",
       "      <td>0.937126</td>\n",
       "      <td>0.918919</td>\n",
       "      <td>0.930931</td>\n",
       "      <td>0.929</td>\n",
       "      <td>0.007559</td>\n",
       "      <td>31</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>0.022301</td>\n",
       "      <td>0.002395</td>\n",
       "      <td>0.001414</td>\n",
       "      <td>0.000124</td>\n",
       "      <td>10</td>\n",
       "      <td>sigmoid</td>\n",
       "      <td>False</td>\n",
       "      <td>{'C': 10.0, 'kernel': 'sigmoid', 'shrinking': ...</td>\n",
       "      <td>0.937126</td>\n",
       "      <td>0.918919</td>\n",
       "      <td>0.930931</td>\n",
       "      <td>0.929</td>\n",
       "      <td>0.007559</td>\n",
       "      <td>31</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    mean_fit_time  std_fit_time  mean_score_time  std_score_time param_C  \\\n",
       "0        0.083925      0.006859         0.006478        0.000102   0.001   \n",
       "1        0.071765      0.002903         0.006412        0.000163   0.001   \n",
       "2        0.044910      0.003309         0.003728        0.000225   0.001   \n",
       "3        0.041414      0.001589         0.003569        0.000078   0.001   \n",
       "4        0.092081      0.003575         0.008514        0.000203   0.001   \n",
       "5        0.089387      0.002987         0.008993        0.000493   0.001   \n",
       "6        0.072298      0.001636         0.006508        0.000148    0.01   \n",
       "7        0.073107      0.004890         0.006940        0.000452    0.01   \n",
       "8        0.043833      0.003328         0.003759        0.000152    0.01   \n",
       "9        0.040512      0.001182         0.003584        0.000223    0.01   \n",
       "10       0.087808      0.000611         0.008022        0.000208    0.01   \n",
       "11       0.086106      0.002129         0.008053        0.000296    0.01   \n",
       "12       0.057442      0.001495         0.004915        0.000294     0.1   \n",
       "13       0.058186      0.002424         0.004919        0.000311     0.1   \n",
       "14       0.037398      0.001929         0.003016        0.000173     0.1   \n",
       "15       0.039730      0.002981         0.002995        0.000188     0.1   \n",
       "16       0.047046      0.002367         0.004072        0.000113     0.1   \n",
       "17       0.048430      0.001255         0.004129        0.000211     0.1   \n",
       "18       0.039987      0.001993         0.003071        0.000173     0.5   \n",
       "19       0.039662      0.000839         0.003157        0.000106     0.5   \n",
       "20       0.031550      0.000656         0.002340        0.000060     0.5   \n",
       "21       0.030683      0.001059         0.002204        0.000078     0.5   \n",
       "22       0.030276      0.002463         0.002496        0.000065     0.5   \n",
       "23       0.030696      0.002513         0.002537        0.000090     0.5   \n",
       "24       0.037658      0.000464         0.002740        0.000090       1   \n",
       "25       0.039102      0.002538         0.002999        0.000165       1   \n",
       "26       0.029608      0.000502         0.001972        0.000092       1   \n",
       "27       0.030518      0.001784         0.002116        0.000030       1   \n",
       "28       0.027252      0.001084         0.002077        0.000110       1   \n",
       "29       0.026960      0.001462         0.002009        0.000136       1   \n",
       "30       0.039095      0.002018         0.002659        0.000164       2   \n",
       "31       0.037873      0.001429         0.002619        0.000115       2   \n",
       "32       0.030128      0.001376         0.001756        0.000073       2   \n",
       "33       0.031574      0.001522         0.001832        0.000064       2   \n",
       "34       0.024479      0.001015         0.001825        0.000060       2   \n",
       "35       0.024857      0.001434         0.001848        0.000091       2   \n",
       "36       0.039802      0.001808         0.002815        0.000141       5   \n",
       "37       0.040171      0.001478         0.002748        0.000162       5   \n",
       "38       0.029475      0.000386         0.001659        0.000075       5   \n",
       "39       0.030795      0.000919         0.001737        0.000054       5   \n",
       "40       0.022427      0.001340         0.001594        0.000128       5   \n",
       "41       0.022493      0.001067         0.001581        0.000131       5   \n",
       "42       0.041968      0.001455         0.002683        0.000119      10   \n",
       "43       0.040902      0.001419         0.002711        0.000160      10   \n",
       "44       0.029388      0.000919         0.001604        0.000077      10   \n",
       "45       0.029630      0.000647         0.001551        0.000060      10   \n",
       "46       0.021693      0.002314         0.001403        0.000117      10   \n",
       "47       0.022301      0.002395         0.001414        0.000124      10   \n",
       "\n",
       "   param_kernel param_shrinking  \\\n",
       "0           rbf            True   \n",
       "1           rbf           False   \n",
       "2          poly            True   \n",
       "3          poly           False   \n",
       "4       sigmoid            True   \n",
       "5       sigmoid           False   \n",
       "6           rbf            True   \n",
       "7           rbf           False   \n",
       "8          poly            True   \n",
       "9          poly           False   \n",
       "10      sigmoid            True   \n",
       "11      sigmoid           False   \n",
       "12          rbf            True   \n",
       "13          rbf           False   \n",
       "14         poly            True   \n",
       "15         poly           False   \n",
       "16      sigmoid            True   \n",
       "17      sigmoid           False   \n",
       "18          rbf            True   \n",
       "19          rbf           False   \n",
       "20         poly            True   \n",
       "21         poly           False   \n",
       "22      sigmoid            True   \n",
       "23      sigmoid           False   \n",
       "24          rbf            True   \n",
       "25          rbf           False   \n",
       "26         poly            True   \n",
       "27         poly           False   \n",
       "28      sigmoid            True   \n",
       "29      sigmoid           False   \n",
       "30          rbf            True   \n",
       "31          rbf           False   \n",
       "32         poly            True   \n",
       "33         poly           False   \n",
       "34      sigmoid            True   \n",
       "35      sigmoid           False   \n",
       "36          rbf            True   \n",
       "37          rbf           False   \n",
       "38         poly            True   \n",
       "39         poly           False   \n",
       "40      sigmoid            True   \n",
       "41      sigmoid           False   \n",
       "42          rbf            True   \n",
       "43          rbf           False   \n",
       "44         poly            True   \n",
       "45         poly           False   \n",
       "46      sigmoid            True   \n",
       "47      sigmoid           False   \n",
       "\n",
       "                                               params  split0_test_score  \\\n",
       "0    {'C': 0.001, 'kernel': 'rbf', 'shrinking': True}           0.502994   \n",
       "1   {'C': 0.001, 'kernel': 'rbf', 'shrinking': False}           0.502994   \n",
       "2   {'C': 0.001, 'kernel': 'poly', 'shrinking': True}           0.502994   \n",
       "3   {'C': 0.001, 'kernel': 'poly', 'shrinking': Fa...           0.502994   \n",
       "4   {'C': 0.001, 'kernel': 'sigmoid', 'shrinking':...           0.502994   \n",
       "5   {'C': 0.001, 'kernel': 'sigmoid', 'shrinking':...           0.502994   \n",
       "6     {'C': 0.01, 'kernel': 'rbf', 'shrinking': True}           0.502994   \n",
       "7    {'C': 0.01, 'kernel': 'rbf', 'shrinking': False}           0.502994   \n",
       "8    {'C': 0.01, 'kernel': 'poly', 'shrinking': True}           0.508982   \n",
       "9   {'C': 0.01, 'kernel': 'poly', 'shrinking': False}           0.508982   \n",
       "10  {'C': 0.01, 'kernel': 'sigmoid', 'shrinking': ...           0.949102   \n",
       "11  {'C': 0.01, 'kernel': 'sigmoid', 'shrinking': ...           0.949102   \n",
       "12     {'C': 0.1, 'kernel': 'rbf', 'shrinking': True}           0.949102   \n",
       "13    {'C': 0.1, 'kernel': 'rbf', 'shrinking': False}           0.949102   \n",
       "14    {'C': 0.1, 'kernel': 'poly', 'shrinking': True}           0.871257   \n",
       "15   {'C': 0.1, 'kernel': 'poly', 'shrinking': False}           0.871257   \n",
       "16  {'C': 0.1, 'kernel': 'sigmoid', 'shrinking': T...           0.940120   \n",
       "17  {'C': 0.1, 'kernel': 'sigmoid', 'shrinking': F...           0.940120   \n",
       "18     {'C': 0.5, 'kernel': 'rbf', 'shrinking': True}           0.955090   \n",
       "19    {'C': 0.5, 'kernel': 'rbf', 'shrinking': False}           0.955090   \n",
       "20    {'C': 0.5, 'kernel': 'poly', 'shrinking': True}           0.943114   \n",
       "21   {'C': 0.5, 'kernel': 'poly', 'shrinking': False}           0.943114   \n",
       "22  {'C': 0.5, 'kernel': 'sigmoid', 'shrinking': T...           0.946108   \n",
       "23  {'C': 0.5, 'kernel': 'sigmoid', 'shrinking': F...           0.946108   \n",
       "24     {'C': 1.0, 'kernel': 'rbf', 'shrinking': True}           0.952096   \n",
       "25    {'C': 1.0, 'kernel': 'rbf', 'shrinking': False}           0.952096   \n",
       "26    {'C': 1.0, 'kernel': 'poly', 'shrinking': True}           0.946108   \n",
       "27   {'C': 1.0, 'kernel': 'poly', 'shrinking': False}           0.946108   \n",
       "28  {'C': 1.0, 'kernel': 'sigmoid', 'shrinking': T...           0.952096   \n",
       "29  {'C': 1.0, 'kernel': 'sigmoid', 'shrinking': F...           0.952096   \n",
       "30     {'C': 2.0, 'kernel': 'rbf', 'shrinking': True}           0.952096   \n",
       "31    {'C': 2.0, 'kernel': 'rbf', 'shrinking': False}           0.952096   \n",
       "32    {'C': 2.0, 'kernel': 'poly', 'shrinking': True}           0.940120   \n",
       "33   {'C': 2.0, 'kernel': 'poly', 'shrinking': False}           0.940120   \n",
       "34  {'C': 2.0, 'kernel': 'sigmoid', 'shrinking': T...           0.952096   \n",
       "35  {'C': 2.0, 'kernel': 'sigmoid', 'shrinking': F...           0.952096   \n",
       "36     {'C': 5.0, 'kernel': 'rbf', 'shrinking': True}           0.955090   \n",
       "37    {'C': 5.0, 'kernel': 'rbf', 'shrinking': False}           0.955090   \n",
       "38    {'C': 5.0, 'kernel': 'poly', 'shrinking': True}           0.925150   \n",
       "39   {'C': 5.0, 'kernel': 'poly', 'shrinking': False}           0.925150   \n",
       "40  {'C': 5.0, 'kernel': 'sigmoid', 'shrinking': T...           0.931138   \n",
       "41  {'C': 5.0, 'kernel': 'sigmoid', 'shrinking': F...           0.931138   \n",
       "42    {'C': 10.0, 'kernel': 'rbf', 'shrinking': True}           0.943114   \n",
       "43   {'C': 10.0, 'kernel': 'rbf', 'shrinking': False}           0.943114   \n",
       "44   {'C': 10.0, 'kernel': 'poly', 'shrinking': True}           0.919162   \n",
       "45  {'C': 10.0, 'kernel': 'poly', 'shrinking': False}           0.919162   \n",
       "46  {'C': 10.0, 'kernel': 'sigmoid', 'shrinking': ...           0.937126   \n",
       "47  {'C': 10.0, 'kernel': 'sigmoid', 'shrinking': ...           0.937126   \n",
       "\n",
       "    split1_test_score  split2_test_score  mean_test_score  std_test_score  \\\n",
       "0            0.501502           0.501502            0.502        0.000704   \n",
       "1            0.501502           0.501502            0.502        0.000704   \n",
       "2            0.501502           0.501502            0.502        0.000704   \n",
       "3            0.501502           0.501502            0.502        0.000704   \n",
       "4            0.501502           0.501502            0.502        0.000704   \n",
       "5            0.501502           0.501502            0.502        0.000704   \n",
       "6            0.501502           0.501502            0.502        0.000704   \n",
       "7            0.501502           0.501502            0.502        0.000704   \n",
       "8            0.510511           0.504505            0.508        0.002547   \n",
       "9            0.510511           0.504505            0.508        0.002547   \n",
       "10           0.924925           0.957958            0.944        0.013955   \n",
       "11           0.924925           0.957958            0.944        0.013955   \n",
       "12           0.933934           0.954955            0.946        0.008854   \n",
       "13           0.933934           0.954955            0.946        0.008854   \n",
       "14           0.849850           0.918919            0.880        0.028855   \n",
       "15           0.849850           0.918919            0.880        0.028855   \n",
       "16           0.930931           0.960961            0.944        0.012558   \n",
       "17           0.930931           0.960961            0.944        0.012558   \n",
       "18           0.927928           0.963964            0.949        0.015324   \n",
       "19           0.927928           0.963964            0.949        0.015324   \n",
       "20           0.930931           0.957958            0.944        0.011046   \n",
       "21           0.930931           0.957958            0.944        0.011046   \n",
       "22           0.930931           0.954955            0.944        0.009916   \n",
       "23           0.930931           0.954955            0.944        0.009916   \n",
       "24           0.939940           0.954955            0.949        0.006507   \n",
       "25           0.939940           0.954955            0.949        0.006507   \n",
       "26           0.927928           0.957958            0.944        0.012344   \n",
       "27           0.927928           0.957958            0.944        0.012344   \n",
       "28           0.924925           0.957958            0.945        0.014385   \n",
       "29           0.924925           0.957958            0.945        0.014385   \n",
       "30           0.939940           0.954955            0.949        0.006507   \n",
       "31           0.939940           0.954955            0.949        0.006507   \n",
       "32           0.930931           0.942943            0.938        0.005126   \n",
       "33           0.930931           0.942943            0.938        0.005126   \n",
       "34           0.918919           0.948949            0.940        0.014951   \n",
       "35           0.918919           0.948949            0.940        0.014951   \n",
       "36           0.936937           0.954955            0.949        0.008524   \n",
       "37           0.936937           0.954955            0.949        0.008524   \n",
       "38           0.921922           0.945946            0.931        0.010642   \n",
       "39           0.921922           0.945946            0.931        0.010642   \n",
       "40           0.909910           0.942943            0.928        0.013661   \n",
       "41           0.909910           0.942943            0.928        0.013661   \n",
       "42           0.945946           0.954955            0.948        0.005048   \n",
       "43           0.945946           0.954955            0.948        0.005048   \n",
       "44           0.915916           0.945946            0.927        0.013452   \n",
       "45           0.915916           0.945946            0.927        0.013452   \n",
       "46           0.918919           0.930931            0.929        0.007559   \n",
       "47           0.918919           0.930931            0.929        0.007559   \n",
       "\n",
       "    rank_test_score  \n",
       "0                41  \n",
       "1                41  \n",
       "2                41  \n",
       "3                41  \n",
       "4                41  \n",
       "5                41  \n",
       "6                41  \n",
       "7                41  \n",
       "8                39  \n",
       "9                39  \n",
       "10               15  \n",
       "11               15  \n",
       "12               11  \n",
       "13               11  \n",
       "14               37  \n",
       "15               37  \n",
       "16               15  \n",
       "17               15  \n",
       "18                1  \n",
       "19                1  \n",
       "20               15  \n",
       "21               15  \n",
       "22               15  \n",
       "23               15  \n",
       "24                1  \n",
       "25                1  \n",
       "26               15  \n",
       "27               15  \n",
       "28               13  \n",
       "29               13  \n",
       "30                1  \n",
       "31                1  \n",
       "32               27  \n",
       "33               27  \n",
       "34               25  \n",
       "35               25  \n",
       "36                1  \n",
       "37                1  \n",
       "38               29  \n",
       "39               29  \n",
       "40               33  \n",
       "41               33  \n",
       "42                9  \n",
       "43                9  \n",
       "44               35  \n",
       "45               35  \n",
       "46               31  \n",
       "47               31  "
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#get results back\n",
    "pd.DataFrame(grid_search.cv_results_)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## ***DASK-ML*** Project\n",
    "Brand new approach to make ***DASK*** a full distributed ML framework.\n",
    "http://dask-ml.readthedocs.io/en/latest/joblib.html"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "#example: Clustering\n",
    "%matplotlib inline\n",
    "\n",
    "import dask_ml.datasets\n",
    "import dask_ml.cluster\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "#DASK setup\n",
    "import carme #import cluster env\n",
    "from dask.distributed import Client #import distributed client\n",
    "client = Client(carme.getDASK_Master()) #get new DASK cluster client"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dask.array<concatenate, shape=(10000000, 2), dtype=float64, chunksize=(1000000, 2)>"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#generate like in execise 2, but now with dask_mL\n",
    "X, y = dask_ml.datasets.make_blobs(n_samples=10000000,\n",
    "                                   chunks=1000000,\n",
    "                                   random_state=0,\n",
    "                                   centers=3)\n",
    "X = X.persist() #manual executions on X\n",
    "X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "KMeans(algorithm='full', copy_x=True, init='k-means||', init_max_iter=2000,\n",
       "    max_iter=300, n_clusters=3, n_jobs=1, oversampling_factor=10,\n",
       "    precompute_distances='auto', random_state=None, tol=0.0001)"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#call skleran k-means\n",
    "km = dask_ml.cluster.KMeans(n_clusters=3, init_max_iter=2000, oversampling_factor=10)\n",
    "km.fit(X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXYAAAD8CAYAAABjAo9vAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJzsvVmMXVeWpvetM9wp5okRjInBeaZIiZqoOZVZnVXVXVluVBvVcNmwDaPQL+2GDcNwPxj95Dc/uF8TBvqlGo2GG22juyprykE5KCWlRFEkxXkOBhnzPN17zzl7+WEdRjAYHIJkUEFR5wMIkbrnnrvvvRFr7732v/4lqkpGRkZGxouDt9EDyMjIyMhYX7LAnpGRkfGCkQX2jIyMjBeMLLBnZGRkvGBkgT0jIyPjBSML7BkZGRkvGFlgz8jIyHjByAJ7RkZGxgtGFtgzMjIyXjCCjXjR1tZW7evr24iXzsjIyPjWcvz48TFVbXvUdRsS2Pv6+vjiiy824qUzMjIyvrWIyI21XJelYjIyMjJeMLLAnpGRkfGCsS6BXUT+JxE5IyJfi8i/E5HCetw3IyMjI+PxeerALiJdwP8IHFXVA4AP/OnT3jcjIyMj48lYr1RMABRFJABKwO11um9GRkZGxmPy1IFdVW8B/yfQDwwC06r6d/deJyJ/LiJfiMgXo6OjT/uyGRkZGRkPYD1SMU3Aj4CtQCdQIyJ/du91qvpjVT2qqkfb2h4pw8zIyHgEqoqLh3DRJdTNbvRwMp4j1iMV833gmqqOqmoE/Efg2DrcNyMj4yFochuizyC+hFY/RrWy0UPKeE5Yj8DeD7whIiUREeBD4Nw63DcjI+Nh6BRQQLwW0Bh0YaNHlPGcsB459s+A/wB8CZxO7/njp71vRkbGwxF/MxCjyQh49SC1Gz2kjOeEdbEUUNV/Bfyr9bhXRkbG2hCvGfLvgy6C14BIuNFDynhO2BCvmIyMjPVBvBqgZqOHkfGckVkKZGRkZLxgZIE9IyMj4wUjC+wZGRkZLxhZYM/IyMh4wcgCe0ZGRsYLRhbYMzIyMl4wssCekZGR8YKRBfaMjIyMF4wssGdkZGS8YGSBPSMjI+MFIwvsGRkZGS8YWWDPyMjIeMHIAntGRkbGC0YW2DMyMjJeMDLb3oyMjCVcPArxGZACEh5CvNJGDynjCchW7BkZGQCoxhAft3+4STQ6u7EDynhissCekfEtR9Wt151AEyBM/8TrdN+Mb5ossGdkfItx0QW0/BNc5VeoLj7VvURCCA9ak2xRJNy7TqPM+KbJcuwZGd9S1M1BdBG8VkudxNeQcN9T3dMLelG/GxBEZH0GmvGNkwX2jIxvLQICljJxrNcGXCTbyH/byb7BjIxvKeLVQPgSUAWvAwm2bfSQMp4TshV7Rsa3GC/ohaB3TdeqKlAGQkSyX/0Xmezbzcj4DqDq0OpX4G6D5CH3JuLVbvSwMp4RWSomI+O7gM6CGwBpBY3RpH+jR5TxDMlW7BkbjropNLoKXgkJtpvsLmOdCQEPdAGoAoUNHk/GsyQL7BkbimoVrX4K6oMbQMn0088C8Upo+Cok18HrRIItGz2kjGdIFtgzNhaN7Y80ggq4+fW5rVYAReT5X5mqmwfxn+lYVSPQOfAakWALIv4ze62MjScL7BkbixTB74X4BkiIhE8v2XPxIERfAoqGB/CCvqe+53qjatpzja9CdAnEQ8NX8IKOZ/N60TlbreOjbhhy72QFSC8wWWDP2FBExMrYg+0W2CX39DeNz4LUAT5EZ1C/97kqunHJOESfg6uAG4dgL1CB5BI8o8COG7ddEaH9nYTs1//FJftmMzYcEQGpWccb1oCbAHzwap44qKtGaPVr0DHwt+Otw24CgPgcaA68WojOgdcBkoDXsz73vx/BdqieBFEItq9Zx+7iAUhug9eGBH3ZKv9bwroEdhFpBP5v4ACgwH+vqp+sx70zMh4Xyb2ERhdBEyTc9cT30fgmuH6QZoi+Rv0WxGtYflwraHwd8JGg9zF2G0VgBtRBuA/8DvCKSLB97WNTNQkjwZo8072gF/WagASk4ZHXA6ibhOqXILWmf/dK4LeveYxPiroJNBkGaXpmqakXnfVasf9r4G9U9U/Efrozd/5vGFXNVlMpIkUk99Kar7eKzAiryLz7M3TYr0iQerKstMfV6hfgJgFF3QySf3lt48vtRytVc1HMHcULOtc81qXXjk5DcgMQNDy6pgAoXt1jvkg1fV4N6uZRrfCsf8LUzaGV32LyzIs43sALnv1k8qLx1IFdROqBd4H/FkBVq5hQNuMbYHFuka9/c5756QV2vryNrp2bN3pI3ypMbvmFpW68TZB7eSlNIUEP6kbSPPiONEd953nOgrq0ADHo+GO8qANmgQSik6hX/1hVoKoVSPpB2oBFSK4+m9y81wx+K5qMgl+P+JvW/zXuRRdBQfxG1I2ZkocssD8u67Fi3waMAv9GRF4CjgP/QlVX6NZE5M+BPwfo7V2bt0XGo7l+doDF2TL1LXVcPH6Flq5mCqX8Rg/r24MbBTeKeB1oMgRuzFIjgEgeyR9LG1lUrLgnPQsQ8dBgO8SXTKYZHlj7a+oMaBXx2lA3iroJQECn7f5SROML4ObA34EXtN1zg8DURDoJRCvSI+rm0OiCjTHcY9cRP9GhtEgIuTdQtwBuCI3Ool7PfcazjngN4NehbtQO07+B1M+LyHoE9gB4GfjnqvqZiPxr4H8D/ve7L1LVHwM/Bjh69Kiuw+tmAJ4ITh0usTRBlo15XHxrHKRlO1hkddWrJoMQnbC/B3vwwh3pU3eCK4POI/7a8taA5awRNBmz15QcWv0YtGKPe2024VAD7nPU/xCR5claxIfc62hyA8gjd8k5NfoS3CKgaHXc3o/Oo8FWJNj32Ok6EQ/VKYjO2KST3Ea995+Zz4xIDnLH7PxASt+KOoTnkfXQgA0AA6r6Wfrv/4AF+oxvgL4DPTS2NVApV9n75i7yxRdjta6qaDKOS8ZwyRyqzyi757VDuBcQCPZb+uFe4vMg9ZZ2ic8vt6JL+u2PLqKVT21yWAPi1SL5tyHcj+TfRvDTFfwmwIdkFMinq/LECrjucw8v3I8X7lipcHHprkJqILoBWrZxR1fTtMYToPO2evbqMG3Es820ioSI15wF9afgqVfsqjokIjdFZLeqXgA+BLIuuN8Q+WKewx88RhrgW4LGZyG+bFI7FTTYAvnXkfsF3qdAREw58zD1jDSAGwY88Orukk/OgxQQr95y8VqFNQYj8RqWFDaqi7ZqdyP2YO4QxBctbx9uB3kMLUK4H6JT6d/3gk4AlXQr92TVpuJ3oUm/jc9rs0ku47lmvVQx/xz4t6ki5irw363TfTO+q8Q37LDSXcF+THNodAnJv/6ND0VyB9G4BMSIv2P5//tb0OQ2moyA35mmWB4P1aqtysNjCDMgNYjXgPpdQIJIHtUyrno+lW/ufmgaxAt60KW8tKDxOXDTELyyJlnk/RCvFvLvpRNXMbMj+BawLoFdVb8Cjq7HvTIyACvaSa4BC2Y1yzx4G6P4Ecnft5eoePWQ/yANeDUr8tfq5oAIpOGBBVLqpswAzUXgd0HuyNI9LL1iv55aPWWHugRo9XPIv//QXPndB6USHnr8N/yge65HVXDGN0JWeZrxXCK5Q2jSZuX2bh78WvC3P1Kvr27O1C1Si/jtz1zbf7+AZ141X9g/vJUBe8VY46ugnqU3klugu+6/6tf5VI0TpkoYhWeoKHcuAcpWD/AcWTFkrJ0ssGc8l4gESLBcYm+r21+iGqPhYbxg9epdtZKqS2IgRjmKBF24uB/ia+C3IMGeZ98WLrmWBuiiVWyyj/v6n0sdMGBySQlAcrj4lhUeeS1IsNMCa7AXouOWspFGNPoK9Xvx/NZ1H7pLpmHh35u8MdgJpT9GpLjur5PxbMmm4zWQxAmXT1zl8789weC14Y0ezncSjU6bZ7vUQvQVqsl9LloAjRCvFSiATqFuxjxSAOKraDzw7AfrNada9YnllfZduPg2rvKpBWp/l004+TdM7lj90v4bnUcTG6sXdCCFH5j9gE5BMgHVT01f/gSom8VVT+Ki82bnezfRCdBRkC6IL6PR5Sd6jYyNJVuxr4HBa8NcPzNAfUst5z69RH1zLTUN62ha9YJhK+czJq8Ldq+5JFy1av4suNR75W7pZgAsYmsRj/umIqQWvHqrlBQQfzMQm1Zc1VI6uvi0b++RSLATpQCUEb9nxWGjTTSf23vQm5A/hhfsSR+bAFFESqgsLJX0L5EMgi4ifjvqFoEKj+veoZqkjU2s6EqJkbuLq6QALgFvHisBzSSH30aywL4G4mqCH/rkCjlU50li9+gnPUPmZxaYGJykpqFEc0fTE90jiROq5Sr5Uh7PW9+Nm8aX0qbJtRAdR/3vr6nyUatfg7sFeKgbRfLHlh6T8CAafQUaQe6V++Z+71RK4qaXKi41ughuxvzZpQb8NlR3rKn9nosHwY2Atwkv2JyubgPQudSGYAHCQ3jhSldGER8J++7/HrVsMk7xbaIJdwJd6RMbwOtE3aCdEQRd6XPSYJwMQXwJ1QWTNT6R7DBOte2tNsm5mZVjzx1B3TTEVyH3IRKsk6NlimqSvn4hU9c8Q7LAvgY6tm5i+MYok8NTdO7ooLZp41brlcUKX/79KZI4IY4SDn+wn9aulse6R3mhwlc/P83iXJmm9gYOvrMPP1jHXzJXBXJAHpjDvL/XgI6DNAE+uIkVB6VLRT2PQCSXBu8YrfzMDid1zlQ2ucOWZtBZc2x82FDchK2spQjxVVxcv9zdyS3YPd00RCdxtf8D3l0+KlZcNWBKFm/zPQZdvuXT9c5qeHmRYBWlr2Ar8XA58GkZ3AwSbEG9BguKuaNPdLApkkeDbVawJB74B1Y9LsV/8Nj3XQvmy/M7cFPg1UHujXt2ZRnrRRbY10ChlOfVHx4miRPC3MY2Wl6cKxNFCS0djUyPzTA9PvvYgX305hgLc2VaOpoYvz3BzPgsTe2Nj37iGpFwJ1qdtBxzsGfth2/+9rR0XSHY+ZSKliSVIbamQfxmmvMO11bwoxVL53j1aDwMyYwddi7+lCXfmPAYSALRWbjbIMuNph2caiAeQL13l4qRxKtB/R2As92Ht/IA1N7zPekPKaQpppsQ3zJDMp1LXSk1lVSu/bOSYB/4W0ACRAppxawPJGjlYwv6wV6kcGx9VTFuAtwE4rWn5mpjJvPMWHeywL5GPM/Dy238WXNNQ4ma+iLjg5OICG2PGdQBcsUcLnYszC6iQJBb3x8D8epM3417LAWKF25D/VbAsVbP8AeOQfKm6ogvW0OL8A9BfMTvWCpVVy2bguYeDboNphnkjhlVLRBD3A+Uwd8D0aeWpgn3rpIoWv7bR7w6+7uWgTSwSwEKb6HJLZA6xO9Jx6LmQZ6MgGdpmDurWVvJv4GWfwb+ZqtSXfjPturF2eQQ7rJmGHdr6TXBvgNbjGgyjrpJxG9FvEbUzePKP7WqWq/FUjuV3wEFqPw96jchuf1P9T2s/FJygFqPVxy2q8t4FmSB/VtGmAs58uFB5ibnKNQUKNU9vhRtU08ru45uZ3J4im0vbaGuaf0NnWylZxOhurl0+12LeA/fGYi3fuXqXrjHrAjwV+X4XTxi7enUQbh9VQGSSB7yb5sPDAXrF5rMpCoXz1Im4RHwN60KfhJsQpOiTQpeA3grz0HEa1z1OWgyCJWPLLeNh+ZehsIHdwX3HCpF8ItAaB41wdu2uo5OovEQeL5dE2w1D5zod6BVNNiD+M3mcy4+Gp9Hc8eg+glUjoNXAxoAY0AMXr0Zibmpp/0K7nnfzWj4sp0V+DtW7VYy1o8ssD/HqCpJnOAH/oqVWC4fPvGhKdh2v3dPF717nv02WN0CWvkNkBpZ5Y89tt+Lld3Ppyvrx1vlPTANlFwGiuAVIb6CBjtW3VskNPMrQGlFg3ZwWy01UvghEu6/zwrZs9fMv5seEhbXtmvRKUsdeXWpgmca3Cz4d+Wgg31p4ZNan1g3Ae6GtdQTrOVe+BpUPkt3HAVLQ8XnUfbae/JaLA2S3La2fG7cUkJShNzrQGzFUn4nhLsf45NeG17QA8EzbAGYAWSB/bklSRLO/vYiowPjNLU3cODtPRue338idA6IU+/xMTSZeqzArlqxlabOg+Qh99YDPU9sAoiXHnfxDQtg0ozloh3i99njXt1dQa2AavwAa4BJNL4G0WU72M2/CTqBhLtWXOeiS2bc5dVCeBTxaiyfv0bE70AlD8kUUIB4GK18jIa7kWA3IoIXbEL9H9h7AZOGqrMziWQCqIHoNOi0pYmCbdiuyU8DfZimlgrLsk+/zT4j/x0kdzA9YJ4G8vYe7sFFV6yxh7QguYNrUhd9U1jzkxEsPdX2XI3tmyYL7M8pk8PTDPeP0dbVzNitCcZvT9LR9w10sHkIdljHYx3U4dWz5FyoCm4OF11Ggp61KSLcJOgc4m1Km1KMId7qRi3qJtDKZ0Bsq29vE0Qn7cCz8jfgt4LXbIEt9y4S7EEJbVXtNUPlI5QE/B4IX0JETMVR+QzwLH0gMyB9pmq5q7ZP3TzEF0BawE2hyRXEe7RHy0rVTzOa/wD8bojHQebNaiC+CP4mdClHvxysJNxuqSY3iroYKp9AfDpt+hGkuv4cBIfw/Ca08J417/DqTZIa7rKir2A3UnhrecfyAMWQumk73PaawQ2gcQMSrr1P67NG4wsQXbTdi7cZyb+20UPaMLLA/pziBz6CUi1XcarrK0d8AkZujnHus0sEoc/Bd/ZS37y2/pkiBci9DTqDRtfAXQPno27IVt+PmiRST3JNBgFF5P5SU40uATmQRquYDEupsqXWFCxSssfcOHcOFCW0wiBX/rWlPbxWiG9CsNNy6VpNlSttEO5Ir6k1h8W70ytLypEqy31S03G5STPxEkHCw6ay0QStngQ3iHqdqS/OIJR/ZR2ZuGMx0AZJP6of28SiE2jwClJ8F5EcLroM8ZW00rUCyYV0h9IKuZ1WByBF0HlUq5Yi8tPUVLDdiqXc9PIO4x5cPGjpHam11Tya1oX56Z/nrF9OMpj66YfgRlCNn719xHPKd/NdfwtobKtn58vbGLo+wrZDvbR0PnlO/WlxznHu00vUNBSJyhEXv7jC0d87vObnW2qkhEbnUp16Lg2wCY/+EQxMueJGbTV7z+GqamITRnQ6NQvrBS9vWvq4H5Iv00M6MXlduBMR33Tu8SVToVRPLkshgz3LKRSpMTlecsty7cXfQ7zV34NIEQ2P2OpamkHBRReQoM8KmdQHdWj0FZJ/195LfN3SS9WTy0VL1VPpzqBggT26bG364iGbnIIdEH+BxltMXhmfs9dLbkH1RKqOOZD2aP3Hdmgd/cb82SWHFv/LpcbZIsUVBWDLn6dLuyZVzV5AasCNodFFc4oMdqaTSQsSPGctLv0ua4qiAn7XdzaoQxbYn1tEhN693fTu7d7ooQC2KE1iR5I48v4Tyj6DrVA9bas+f8uafvE0GU4Lcl4233M3ZXlhLAhp5VOo/NwKkdyE5Vhzb9jKV+pAYitICvYhweYlaaLGN+wal9jzgt2gk+BvW0pJiAga7jMpoN8IWsVFVxF/0ypPdPHbLJVTPWOBWzxrfafOzgZwrOiE5CbswJTQ0iduDuuWNGXB2kW2S/BqIBkGFrCJMGS5qElZyqFLMS28Cu19R19aIVX1Yzt0dTNQ+SUE//S+n7NzMzY56CwabLWaAo3TtJMPmqRNSfaayuY57MEowS77rki+84qbLLBnPBLP8zjw9l4u/O4yhZo8u1/d8egn3e8+QR/qNQPmUrgmpAQSo24WRFjhXaILqQa7Lq12TSDYlc5Ci+AGLSBTBDeOeHd3SaoCIXiNaTD0wO9Bwi3Lt9cyVH8NrgLVOVQEqEXjr1GvDfCRcDdIfVpROWl5/WAHeFvSAq2jEH+VfgCduMoXJiXUaUubhK+A+ODX2DhZBH+rPe61Q3IJvE0gW0DKEB5Ggi1A3iaj+LKt3nOvWODWOO1yFEDQCJUyVv2rLPm7q7NDbQkRKeLim1D+OxtP8JKpZaLr9tkmwxBsty5Td76SdQrqlkq6DF4zkjv8RA2370ZE7CwlIwvsGWujub2RN//R2nqpVMtVzvz2AjMTc2w72EvP7mVZ5ePq1MVvR3k5Lc/vSvtu3nkwZyoU1wB6Kf13swW33GGIyml+vXFplb983x4rEtIZKHwIfjfi1S1ViAK2O9CKrcajAdBaJGxFq5+AXwZptoAeHrMUitdtDa6TfkthhLvxwnY0+D3UjUP1t7Yqj76C4FBaoXoZCt8zCWB8CYI7rec8qB634B5ssTJ/qV0Kqi4ZT18nh4Q7TWmUrsZd3J+u2IP0fGPOJrjC96wQqvoVuAHAR8PX0mrfBpBxSK4sv3ev0SbW3Cvr3rxa3axNIF6zHf5Gl1HiNI20He95S/N8y8gCe8a603/hNlOj0zS01HPx+FVaOpufqJAKbBVmudzVv+jW0f5N1L+Vtm4LQe0ATYKtEHai8W1L5fgr/dut3dsHQATk778KTQ9q1U1Ybl/NnAz106KjIrghqHwM8QAwAEEfhL+PhL1Lk5jdO7KjRqnB8v0TlqLxdwGzSNC+wjhMk0GUxSU1EG4aCe6a1KIvsbSMBWopfG952H4PSg50ASm0W5oGsdSSW0gN2jZZjj+5Bp61/cPVpemiIN0ZlWzVvlavn6chGQTm7AymehL1Gte1WO27RhbYM5aYn1kgiRJqm2oe6fjonLOg+4ht+XqnYtXNpEqVRnNR9OoQb89dV/Tg4hFruEENktv/QFmlmWyZ2sjSE4sg+aXcv1kjvIUmY3ZoKgXQBTsojU4BU+kOYRHyr0N0BXKH8O5xRFSN0WQc4lELol4PMGf54KAHmEzz2Xc9h1pwcyYxlHpUcmh0Clx8V1oktSO+50O2ybCD+yKhTSg6BVTA60b8/Wh0AXKdSLgHjc5aVaouQLD1mQRY8erQcK/tUrxWO/tIrmPKJuUbmUxeYLLA/pzinMMljiD8Zr6ioRsjnP3tBVTNzXLfG7uIqjFB6K8K8v3nBrhy6galuiKH3t2LeB4Xv7hCtVxlx5Gt5Aoh47cmmbg9yas/PLJqtV6tRFQWKpTqi/j+o2WcJrs7k7oqVk2b7XVA7tVVE4upOb6wFIIOonGIhAdX3dPsY2eAEKRgKZVkwg4r828sVayK13SPEqbWKlH9TYBaG77qCRub32j6+XtfK75iKQ6/C1iEmh+ZPLH6OTAF/hY0HkHd5+C3IcHedFXtWX7fbzCFjxvG+p5OQ/hyqtMXJHfkkZ/hHe5YG2vSDxSQoA/QdLdQMAVT7iAqAWg5LcR6Nj+DXrjDDoix8wzV8TQVs3XtZzAZ9yUL7M8h8zMLnPzoDOX5Cn0Hutl2sG/dX8M5tyJg37o0RE19iXwpz/D1EVSV0f4xirUFXnp/P8VaC3SL82Uuf3Wdxk31zE3Oc/3sAHElZnJkinwhx2c/+RLf9+je1cnc1Dz5mpWr5fnpeb782WmiSkTjpkZeem/ZMli1grqKtYZzw1Ys5G+1tINqaqPbjBSOWcETZezA8W6S9E8eqJq65B5UHVr90tIoiClAknHET4ug4mEk7FuWRLop8LetaBiyVCjkd6NhxQ5O/QMgxVQzftdBoJsH7lS7RggO8ZtwudctOMe37LDU77IuT9JgB5leJ3gh6Hi6oq/lTt9T8RpXpF8eB9vl7E8/ixitfmapIXyb1NLDzLVgFbsVeMr+qFbv8A6QfKdliutF9gk+h/Sfv0USJTS1N3D965ts3tq+FFifFuccF764wtDVYZo7m9j/5m6CMKChrZ7rp/tZnC8jnsfQtRFaO5uZHpvh1qVBdhyx9ILnCSKQRAlJkuD7HovViFwuJFfIUV2ski/mKdYWWJhdoFpe2QVo5OYYLnG0bG5m5OYYJ395llwhpGdXgbraMxZgklnIHTJNstTbajjpt9V6cgGNN4Pfzv3cAUWKaLA71ZQXkHBH2kf0FnitlnvXRUiG0kA+l8oJU9dBjVN5Imh83e5DHbjPcd5bVtnopiDYhRduRcRDwp12vZtGyz+1w0CvC8m/jnglJNyKVkfssDToXcrdU/3MCqLw0vFtMXVP9ZSlZbRi6hZvs01y0ZeAM0vdu7syacVSKVqxNI3U2muhaWn96l9zm9zOgbsO8TiS2219ZZPba7Z8UF00uambMzVK7tWnCsq2+8pC0nqQfYrPIWEuIKrGVMsRIh7ek+rG78PUyDS3Lg3S2tnM2M1xRnvG2by1na0HeiiUckSViIa2Br76xddUFiokUUKYXy5jzxfz7Du2mytf3aBlczN9+3soz1c49auzzIzP8tL7+5kcnmb89gQ1DTVs6lkpPyvWFokqEbOTc4wMjJMkCXWNtUwMfMKbf9iLLw2gV2yFrmLFRMF2C/J+HwSeFSrlDqFxP4qkFrfLY/TCXabFxrd0SzXtnpTcTnPcLeAV7FA0Hks77dWa7DA4hPh38tMVLOdbsoPG6EqqUW+C+Azqt61Qi2h0CZJxC9Lxx+j8vzVZZO4lM9gKW2y1LGKqlsWfYJrrZqDGdilu2N6npj7ruf0mX8RDvXfsENVbadWs1TOWuiFvq29vE8Q3bHLwupH8K6t+DrT6GZT/1lSQbhz18jahPYZdssZDoPM2QSYj6a6l7dFPzHjmZIH9OaR3bzfVSsT81DwH39lDvrh+XWYkTb8kUYIqS+kY3/fp3tm5dN2+N3dx4qenAKF4T468vbeN9t7lX+BcIcexH72KOrM+6N7VSbUcEeaDVTn09i1tJEnC7IQ1igjCgEJNnonbeZJoFr+QM4mfTqRVpC1IrgF1k0B6uJk7gkZnzOI2uYl6HVD6xyvK4pc8yF01fd815jTpKnh+aGqa+AYko/Y6WgXJ44XLB5/ib7FSfx0FcrZa13nTh+t9yum9nI3bLdqugxnbbbjrVvyUex3Jv2zXVk+lK/d5240U/xsovGnBltAOEL08eJvMfiC5kUoqW1CpgcJ7d00qC0ApLVKaSCewVqwT1dAKT5ol4n6bzLwWOySNb9uqO5nAJdeBPBIeeKDhmn1ABdDYdj2wtNPJ2HiywP4ckstQOw41AAAgAElEQVSH7Ht916MvfAIa2+rZ/tIWbl8ZpmdPJ63d999231mpF2uLfPmzUxz7o1cf6g/jed6SL5bneRRKD1KiCF3bN8N26OjbxMmPzrA4V2bLgVfJ1Vq+VvLvYC6Ld3YqPhTeSYtqaoCcpU+SUaAEyW00Oo3k37jPwJrAb7EG114tkubJxauFcA+a3MAG7qxK9O6xppJIdaNQ/gyCVqjeNvVG/o3V1afBHjQYhehXqfIkASbBNaYe54N3XVxIdxHjaQFpmFodzKcFTQEU/4np95MB0MB08FIArUddggYFs9f1d4E7nqpY9gCRpZBUIOi9v3Ip3G0To96wsYY7bbVe/ivIHQXm0Ojrhxppid+BhofswNPfl8kTnyOywP4dQ0TYemALWw8sV1gmcYJ4suIwdWZ8DhBGB8a4ef42Ih5v/ehVahvXr99rY1sDx370KnGUUKwpPPRakRx3uw5a384TtkoNurBKUmxVHn0NVMwv3Ws2iwFd5N4Gypbm2Q/l/w+0DOGbqCarr1FMHuisZZ81sN5qmnBikLpU+plDih/ickdg4T9D+a/tvl46afjLPuSSO2A+MtIF/j7QyVQvXwf591N7gBjwQBwk06ba8Xvt0Dc+AcH3IbqK5Fuh8H3M3Cxv0k2v1SYzN4yrfI7kDqzwppfwkKWJdN7SUUxjmn5NJ6U4/Tfm6qlTgId4DaZgifuBwN6TFECnUJd/bK/9jGdDFtg3kCRJmBmbxQ/9NbslPi5RNWLs1gSe59Ha3bwqNdJ/boArJ6+TK+Z46b391DbW0H9ugOtf93P5xFUW58vsOLKNQjHHrctD7D66vjatYS4kzIXMzywwPTpDqb5IY9uj87wS7kPR1AirDgmtKbNGZ1KDsTxa/Rzy37dA/QBXSETMt1zazGfGTazIE7uoHxb+n/TgMjCr28L3cPGwdWDCWbANXzJli5YRrwktvAF+C1bkMwD515HgbisGAb8J4lnMYqGY5vlDW3lrDC5CJV5ekfu9Nk7fT83Q7jgs6ko7X/FQrxGqX9h93ShaPbecBkqvkcAqgtXvNo28LkL+D0Bvg4RLXaU0PmvGXyposA/0ljldElsunxmggMZXIf/+fZ0iM75ZssC+QagqZ397kZH+MQD2vrmTzm0PKCpJiaOY859dYvz2BF27u9h+aMsjC4TO/vYCY7cmUVV693ax65XlwFwtV7n81TXqWuuYHZvj8olrHHx3L1dO3qC9r41SfZGzn1xgU08r5fkKpbqHr6ofxvTYDF9/fB51yp7Xd1CerxBVIjr6NiGex5d/f4o4inGJcuTDA4/sECUituoN92JVleluIxlNqynr09L81bnw5RWoj7lHOpCqHTauUJssmiY+GbNrpQ7wEC2jyVUsr12yPL80pS6KgIvtgFd8YAFyh5F7mnNrdNoOTQOx/Hv4QzyvBs29ZXn9ZNhy+vF58HcgpT9F1UvTMa3g7zYZpN9jh6Ur3l8VdWlHJmnAVt4xD0K80lIayxqLiOnIpdH0/vE1m1AktiYbVGwMVFI1j7X602TEJgeywL7RZIF9g4gqEWO3xmntaqa8UGHw6sgjA/tI/xjDN0Zp7mjixpmbtHY2PXR1myQJE0PTtHQ2EUcxY7cm2XWXQEI8CzQXPrvE1Mg07X2b2PP6Doq1eeanF3DOcfiDg9Q111LXVEPnjoeP72Gc/eQiQeAjnvDRv/8tjW0NhLmAoWuj7HxlK3EU09zRxPToDNPjs2tq/WdFRmlnJfKoVtIy+XnzYAn3pymYlbshjU7bYaRiLebCXWbfG+xHqYc7Pt56x1u9ageoumj5bQrmo+JGU0liYdkiQMtQ+du0Y1ET1PzXePetAk219n4v1kikJv1O6hCvDlc9bbl1AdwEmjtsK/zCn4C7bI6Q+Q8h2LVS+ujm0epvbVxaAR0DKS15zz/881xEK59gaqIbqHh4Qa8Zt7kJG7PfaxNWfM1y+OFhcP12DuE3rLJVztgYssC+QQS5gJqGEuODE7jEsf1wH6rKrUuDS52TevZ03XdFrmtscOD7Pu19rQxeGQYRtr20ZcXjYS6kd283V09ep3tXF8XaAhND0xx8dx/Xv75JEPpsPdhLrvD03eQ938MlDlFYnCvTvbOTsBDy9W/OMTU6zezEbNrfNaBljUFdq19Y+kRCyB3DWsAV0wrPChCh1U8h/72l4Kcap+ZZbbZKT67jFd6HcC8uHoXKT1ESNDyCF3Si/pY0xdIA/pb0kLAI3k7rwETF1DNuBly/dVICK3pygxac7xPYJTyYerXPm2Oj3FuBGwFl0BwQIRJZo+r4Ytrwuh7K/wnir3G5d/FypuYx2WFqXMYohK8gfrulZ9ysNda41+xs6UMtAy7tizplh9WA5F6xNnz4SNDNUm4ds3RQ3WbpI6n7Treje57IAvsG4XkeL72/n5GbY+TyOdp6WpganeHC55epbarl4vGr1DTW0LLZgly1EuEHHnUttcxOzLFlfw8NratXR/PT80yPz1HbUKK+pY49r+1cSnc0tq2+flNPK1sP9lGqKzA/s0i+mKOmvsT+Y9bIeHG+zJVT18nlQzq3dzxxJ6e9b+zk3KeXUOc49kevcuPsADPXhlFVG5949O7tpmtHBzUNj97K6x1nxKAbdBFNbiLBPgi2Q/TXZg0Q7EpTAzF3PGEs+Neb2oTErHSrJ+2heBC8AhBAdBLntaYt75ohqLUq0HDHspdMuJwzF68WlbdR2qxgxw3ZhPMAX3DxGpDCh/eXIgKE+7B+rDnwGmxMwQ5Y/LVNBjoB6gM50L9Fw39m6SmvBpXEPGZQxKtNg/ocWvk1d7zcNfcGIqW0YjR9fak3OaUbBXzEtxy8SH7FewVQFbT6OaqRHSYHK03WMjaWdQvsYkuOL4BbqvoP1+u+LzL5Yp6eXcuWtkkUA0KhlGduap64annROIo58bPTzE3NA3D4gwNLAf9uBi7d5md/8WvypRwtXc289sMjNLTWPzStkS/l2PP6DiaGJund101zRyO3Lg1y+atrlBpKLM6W0cQRRzGVhQpduzrpPzuAH1ggvrOaT+JkSZd+P+qb63j9D5YP7zZva2f4+giXTlzDJY4w59PS2bzGoL5gB4PJNTuY9PtA9i01gnBSnzaNWIBw2woTMBGB3GvpCjRMK1rHLC2T9IP0YRNBkDbDWLAVt5sz3/Nwn1VtxpdtcvC7Lb9cPWkOj0EnlP4kPYTtQYItFvyWyu5XToyrvW4Ujc9D5SvLked2pEZmgaWaCkdMIVP5m7SwqQTJKFr9HRpsR7w2CI9a4JeOJZtjTQZSb/Wtpocv/40dsAbblg5JRXzIvZYqcgpW5v+g7yD62lJVrgqLf4kr/qMHpJwyNoL1XLH/C+AckCXZnpDGTQ00b25kfGiSxrZ6mjebEdLiXJmFmUVaO5uZnZxj7Nb4qsC+OLfIlz89TWWxwsLsInNT8/Ts6rzvqv4OV05d58aZAaJKld593dQ0lCgvVLhw/AqNrfVMjkwzdG2Y/cf2UFmoMD02y8TQWRbnyiRxwsJcmUPv7GPs9gRnfnMeRdn7xq4VxUsPolRXpHefFWKN3hxny75umtrXWPWothq1TkkXzK/cX+405QVdqN+arnhXTxQihSUbABefSaV9vgVuqbNDS78E0dl0tZwG49zbeH49Lr6dPtYI1a8sNZbcTieCSWu/l/8wXSnfyXmXbfX+qLJ7N7pckaoV0I5l+1+/xQ42/QazHE5SCwRim2Sqo2juXXtdGhHfPk91E9Y6MEmtFaSU9nVtMW+aYNtSEDcF0WoDLtW00GvJA0esK1NyyYq1qp+h3nuIt3bzLtVq6sVTRsId908PZTwR6xLYRaQb+EPg/wD+5/W453eNaiVi8MoQTR0N7Dq6nWJtYUlXXqjJkyuGjA9N4mJHw31SKlE1Jl+0X7qLxy9T31zHla9u0LVzM3VNq5skTAxN8ot/9zFhLmB2fJbBKyP07O3iwFu7QUmrSD1au1sZH5xERNj75k7Of3qJxDkufXGVS8ev0be/hwufX6ZUX0Q84cLnV9YU2MHOAHYf3cHuo4/ZkUlqUrHLFPgdSLhntQGVLlowlRAzBEsDSTJkwcnbBIgFxuiEBaf895Bwm6UivHbQYTMh852lRMjj4rQ9n7vNUos6rdi9pC5VH95pau2hyW3QauqrPmw5bn+lJcBKIkx10gZMg5teSteI15T6zs+b33r5FxB9bZJGr9F07pXfWPNsv4CGhyH/HppM2GeWfweSmyCbgBlT+8iC5eX9zgdOOC66YH1hpYjmXsML2u2MIBm29xm+bOcVugg8RmCPz6dyyTxanYD8B5kB2DqxXp/i/wX8r8ADxdgi8ufAnwP09mbdUe7l3CcXGB+cAmDs1iStnc3kiiHtW9oIcyEvf/8Qk0NTFGry902t1DbW0L17M2c/vYjv+7T1tDA7NcfCzOJ9A/ulL6+iTpmbnuf2tWHe2N+D73tUKxG7X9vO1a9uUN9Sx743d1MtVwnCgNtXh7h1ZYjP/+YrSjUF2npb+N1PTtC4qZ7FuTIiQq7wDRyeSZ0pRKoXwKtFXZW747qLR8xgSzBteP5tzO72c1OYqINwjxmCiW/qEi2bph3fnkdktTpBB0gtWvkVIBAt2k7AlSE5bf1Ec2+khl7XTDUS7FsOUJqYF4uKBWRdAG1eKX3UyvLuwmuDYK8FbK/GbAjuutZSK3WQDKJeCIX3rNFHdNUsARZ+Yu9DG4GzkH8rXek7S7H47WZ5XLmapo42Q/UE6o+t0LkvfZbJuOn4Kdp4oq8haLdxFH+IVmrTNFMpbT7yGLh5oMYmJl1rc/OMtfDUn6KI/ENgRFWPi8j7D7pOVX8M/Bjg6NGja5N1PAeoaqrW8J9ZA19VZXJ4mqZNDURxzMlfnGbbS1tJ4oTKQoWtB7ZQrClQ3P7gHKbneex8eTtXT/XT3NFIVI5YnFmkrvlBLc2E7Yf7GLoxzNzkPIWaHOIJjW0NFGsLdG7rWHq/hVKei19e4e/+zUfMzy0wPTpN4Hvk8jnGBiY49qNXuXziKupgx5G++76/O+cDd08y1UrE7MQcxdrCY3ZYKpvhVLjfcuDJNQhSp0ads7y75K1K0o2mksgSuElbOWs5zTfvskDqzN0Rf6ulS4I9lrIId9nK3o1hapE2VG9bQArfAJ0Br9GqUIMu1EXAHCI5y5Uno6m/TATxcZBuWPwJ6jWj+Tfwgh7UTZhDoiYQ9iHBAaT4IzR3DHQReaCplo/tCsTsAcLDaWl/wayK3QCwBQhWrvSlHiq/gPCQ+djEI+YXk1xD9cjqn/HkNlAw3xo3BrodF52x/+93Q+4dhAWsbd9jqqeCnWbFrPMQ7HhgQ5SMx2c9pse3gD8SkT8ACkC9iPyFqv7ZOtx7Q0mShHOfXmSkf5zGtnoOvruXMLf+K1IRYcv+bq6e6ieqRNQ219Hc3sji3CJTIzNrvk+YC9iyt5swDCgvVtj35q6lgFktV3FOKZTyVCsR3bs2c/XUddq6W+no20S+mGfnK9s4/7tLnP/sEnXNdXz4X71NXVMdqkr/2QGKdUUmR6aoaSghnjA1Ms1b/8XracON/UTViGun+6ksVNiyv2epmvb6mX6uneoniRPa+9rY/lIfuWKOL396ioWZRUSEV35wiPqWtVbfhla+LhNWNON1om52WfXh5kE883aXggV1QvDarfhHgOBAepD6clqpGpjhmIg1qA53L72aeg0gJQvU5CG324Kc+Fa1iuWe1V2F+KJliYJdQJwaee2zQJhWgaLzUC2gXj0aXbZ7ejUQX4dgB+oWYfHfg5tC/S4o/jFyb/rGa4PgQNoQ+zASbEfjKvh7QAZB66H4B0uB+s5KXwDn1aUSzrQZeBykNsj3WTF7DdblKRk1dVDYC9F5mwCji4jXgvjLBVKqsXVgcuMQbMMLVkpsV9zab0ULH1rdQGo2ZtbBY4jfnNkTPAVPHdhV9V8C/xIgXbH/Ly9CUAeYHp1h+PoYLZ1NjA9OMjowQee29kc/8Qno299La1fLUhAd7h9DBLYeWv2L4Zzj9uUhrp+7iS8ePXu66NzRQXm+zLZDvXTuMFniHXnj6MAYZz6+gHNK184ORvrHiSpVGlrrEd+HREmqCWc+Ps/lE9fIF/LcujjIZ3/1Jd//s/cQEVo6m5kZn+PGmZt0bd9M565Oune0c+Dt5QB49eQNbl0eolDKcfKjMxz7o1fxA58bZwaob63n0pdXuHTiGmO3Jtl6qJeFmUVaNjcxPTbD2O2JNQf225dHuXLSo65+lp2v7KWmeYf1NlWX6rcxHblXa4HnzkryThCXYCloiISQ2vSqKi4eSBUtnXhpx3uRvKVz3IxNElJMlSPhCv8VonOWY/dqTWsevgxctdf0WrDS+0XwTPetbs4mnvgEENquQ31Y/GuonjODr2QIja+tCuw2Ae1ID4kXgYoF99w8uG4I9q6QIKqbRqvHLYUk9VZc5PfarkRqMSVQbCkjN2KeOX6H9U/NC+isyR91Ab2TsrJpYsW4NL5pOyhpguop1Gt6qDmYeQDl0jHOoeWP0pSVosXfx3voeUTGg8gSWg/B8z1Lq1Yi1Dn8YP180e9FRJbSFPuO7aZnT5cVMdWvtk29fXmIEz8/zc3ztwkLAcM3x6j+p88Z6R+jo28Tr/3+EXr3LqtErnx1nVJ9iSDnc/IXX9PW20rL5mbGb09QXqzQ3tuGSxyTI9NUFyNq6krkijkqC8tNMvYf201DWx17Xt9BHDlKtXl69/Ws2MEszpcp1uQp1BaYGpleSmE1tTcwcHmI8VuTdO/ajBd4TAxOEuYCJoamSJKE+gemjFYSVSMuHr9CQ+smFucbuHzK4/AHAfh1aJymP8RDgp5V7otQTXul5iyIJNeAPBJsRSS0w8DoOFCEpB/Nv790D5Gc5bCXvrCVwcoadgzYAaYA+Xfwgg5U3rK8ev5tiIch+iqVVeaWDiStqcYCECLMW/pIfDvo9OrtYPQ+uGQMKp+kZwklJP8OXv7V+15rXjDOCrqS4+Zt47eCOCC29nzJjI3PTYGXR5NOvPyrSLBsXqZaA16XFV/53alX/N3cqRl4/N6l6masByqBHTJHF8A/tubnZyyzroFdVT8CPlrPe24kDa317DyylaHrI/Qd3EJb9zezevA876EyxfmZRfzAp1hbAIHRm+PMTc7Rsa2dmYlZLn5xZUXVaqm+xPjgJEHoU9dST1yNmZ2YQzyPPa/t5MaZm4jncfiD/TR3NHLi519T31rH4Q8PLL1mrpB7ZIu+vv09nPzoDPNzi/TtX65Y3fvmbmoaa4jKVRLnOP3Ls3Tv6mT/W7sp1haoaSityULgzmfjBz7VckRUiQhytsq3PPLbaDKD+E2rgrqLzkN8GfDQ3FHzddEYqKJUzURM57CcdEPqe1IG1jbhoHOpf3xvukK3nZb4LQjpz02uEee3QuWX1vZOpyC5CC5IlTa/sxy43w3UmC9L/j0kWGm8tqSjrx43NU+w2Z6vc9ztgLkSAdSuk7rUh33STLtwaDIL0SdQPZM+fhjc8KoCKpEAyb+CqrtvKzwJelA3yBP1LvWKmFtmaLseKmt/bsYKshX7QxARevd2r1j9PozRgXEmh6dp7Wpac6BaC7euDDJ4ZZjmzU307e9h87ZN3L48iKri+R6Nm+rxfI/5yXmq5YiaxpoVv4y7X93O9a9vEkcJRz48yPTYDLMTc3T0baKxrYHuXZ2ImMXApp42Dr1rBSv3sxJI4oTbl4eoViM6t3essNutb6mjaXMjg5eHmZuYJapGhLmQXD5kx+GtdO3o4Kd/8Sv2vL6T5s2N3Lo8xHv/5M1VzbIfhh/4HHhnL1dOXKOpo4H2LW1c+OIKtY0lNm9rxwtXf+7W4Pos1kovsRyxli0IJgNQ/Rr1+8xfPDpnbeb8tlTDvkak3qpWXRWCDiS4/6GneCWzBkidINEA3NW0iUaXKVW8AoSbIH8Iya0+0NTkVpr2yUP15xDVmzoo/wZ3X+niW2mLwAZT2sRn0pSQZyqUYDteOgGaL33RmktHZ+0z8uusEXewbSmIm2XvIOBD0L0quD9N71KRBjT3Rnpgm1s6v8h4fLLAvk5Mjkxz8pdnKBTz3Lxwi9d+/8h9ZYaPYujGCKP9YzRvbqJzewezk3Nc+OwyNY01XD15nZr6Eu1b2njrj1/jld97iSD0QYSzn1yg/+wtOrZtoq27mWtf32Dztg4KpTz5Yp7dry5rxWsbaiBdBC7OLVKer1Bzl8/6w7xhrpy8zs3zt/BDn7GBCV794eGlwDw9OsPojXHat7Qx3D/K3PQJGlpr6dvfS11TLcXaIlv29TA+OMnNc7eYm5rn+pl++vb3rgju89PzqGKHtPdRIjW3N9L8wyNUFit8+pfH8XyP/vOWNurasbq0XTVObWdTl8P8BxD0WfOMZACCXWj1E5MtimeBVatpHn2NB3gSAEWrYvV3ca/x2NJlUoT8G1YJKg0QXUpXpyUggOQ8yBbwetKgbt+FS6atGEs9G6ME1smpWgIcJFOw+Au05p8iIladG52wCcfdBlfCy7+L5u44W6rlwZcG1gGkwT3YnxZUtUB0BpXQJJ+6aHLH6CzojO18Cn+46jt60t6lIh7k30oNx8LVB8YZa+ZbE9iTJOH6mZvMjs/RvbuT1s7n68R8ca6M5wl1zbVUBycoz1ceO7BPj81w5uMLlGoLDN8wO9+pkRmmx2do3tzEvO9TrVgAyxVyKwLw0R8c5ugPDnPx+BVunB3AC3xG+sdXBN57mZ2c48ufniKJHaW6Ai//4CVy+YerfqZHZ6hrriVXyFl+PE7wcnZ/c4tU4ihm6PooDc2LCHDyozO8+UdHrSDp1e2c/vU5+s8NsPVgL1dO9lPXVLeU5rp54RYXj19FgK2Helc0BLmX+ZlFhq6PUN9cR76UY25q4b7XCWU02GIHh5TBa0KC/WiYFjj5ragbsW5GaGrqNYEmE2tv7BzftoDpb7P8s06umhTUzVrBktSarFEER8Ekf369rfalaH43yaQd1vqtlktf/I9py7tN4G82zbubxFRAY5azj06iyXsgQdocRNMCLZ87tr0isjKgp3hBG+q9C7poh8/V3yFePepiiK6jTFnqqvKFWRm4GBb/Ku3p+tq6SYHtLCOzJnhant1p4Dpz+/IQ1071U54vc/pXZ1mcW9zoIa2gqb2BfCnPxNAkpfoSDa2P3zgjqljHmlJ9CXWOEz8/zeD1YaZGZrh6+jr1zbWPrOqcHrPA27SpgfnpeUudXB3i07/8grOfXiCqRkvXTgxOEldjcoWQqdEZxm6NM3JzjPmZlQFydnKOq6dvMDowTveeTuam5hkfnKRrR8eKw9OG1nq2H95KtRzT1FZP25ZWSnVFquUIlziSOOHmxUHmpxdoaK2noaUezxOi6rJX+I2zAzS21tPU3sj1MwPmnf4Abl26zeJsmSsnr3Hj7ACbt93rS+4sDSM1aYu8JvA2g9dugSjcA6IW1L1NqaVA3v5NjPirg7pqjCZjqcnWXYiHSWLcqufY8ypo9WNTy0Sfo3E/LrpmZmH5V5HafwalP4VwizWxEAEpWPFSfB6cs11AdAGqp8HfhRT/genXvSbL6WsOKp9ZIVZ03FQ1Og5eCfEfndYQr96cIL3WVBo6YjsD8YH00FQKEN8EHQFpMwWMTj7y3hnfLN+aFXtlsUqYDyjUmAthVI15nJKWZ02xpsCrPzxCea5Mqb74QDMsVSWqxgShv2ol3dBWT0NbHRNDk4SFHIrS1tlCoZinvqWOA2/veWQ+umd3F2c/uYDqPJu3bWJyaIrTvzpHc0cjQ9dGKdWV6NtvKod8TZ7LX10nqkaos3HV1JfwfOHo7x2mtrGGxfkyX/70FIJQrVR56f0DvP6Hr5DEyaodiYjQt7+Hvv09TAxPcepXZ5mcn2brQVPP3Dh7kxtf91OsLTA9OsPg9RFau5po7VoOoI1t9YzcHEc8oaGl7qErwZnxOfYe20VcSZifnlvRtk/dnAU4nbe0S+4YolMgxaVVuOe3ovkPsBL+BmuDl3vL8t9SWuVdourQ6peWD0fQ3Mt4aRci8TvRZBx01Aqc7loVq5bNMldjK3JyU2lP0gVboSe30Px7wKK13/MEcm8CMVr+DcS3LHjG/VgKpR7K/y+q70D+Fayv6nzatzRIX2McvAYkdxjwHmtFbWZgR9Px5S3AR1/Yv/NvQTwO7hz4B9MdwaNRjdD4Arh5JNiRpVmeMd+awN65vYOR/jEmhqfo2LppXXtvrhe5fPjQVIZzjrOfXmTkxii1jTW89P5+8sXlarswF3LkewepLFQJQp9TvzrL+OAkCHTt3LwiqFcrEUmcUCjlV/zSdvRtoq65lqgacfPcbT7/25NcO32Dl97fj+cLcbwsP8sX83RsbaNYU2B6fHapOnTgwi3yNQXe+IOXKc+XiWNH62ZrgjE7ObeUNlFVrp3pp//MTRo3NbD3zd1L77+5vZG3fvQqLnFL77FajgjzIaX6Et27NvPy/8/em8XIdadZfr+7x419y4jIfWWSTG4iJVL7UmurVN3T3QOP3Z7pNuCHaQN+MmD4wfaTAT8NYAOGAT800Bh7egZGjV3TbrdH06VSV0lVJZVU4r4kydz3yNj39W5+uMEgk5nJRaJqUeu8iEpm3rg3gvn9///zne+cb5/GAfLbRSKJILpf5+iFGQJRP7ZtH8iXP4iJuVHuXlwGx2Hm7OSe98ex1nsN0jiYqwjSeN+G9kG4ARe+B/7fi8t3H4QO2FkEKYljN1w54r3CLsgI2tl9P2G3PwbjUyAA0mDvNCD26Az7/nSsnXabm1LEnZDF6DUpRVdrbhgguwlOOGkg7EoXpUEE3z9zh7EwoXsZx3ItiQUxcoDP+5NBEMSevh0EeRhH8LjP32u+Ot1r7rCVPHEgtfMwHHPJtVzAi9P9FDzffvpJ1a/xxPidKezegM6L75zrm109WMxs2ya7kVlpQHgAACAASURBVKfb6jIwFn9sMPJvCtVCjcxajthghGKmTHY9z+ixvcVGkqT+tOhz3zhJtVBD01Usyya7kSMYd31Zrn94C8u0GJpOYXZNqsUGk6dGGZxM4gt6adZa5DbzDM+kaDdaLF9bJxjz96mN6dPjeHxaP4HJZ1hUclVWr2/gC3spbhe48pPr3P1sma2FNEMzKYZnUnskn/Vyg9Vr64STIQrpMjtLu/3TALBvSndoJkV2M09xt8zQkUFMw+Tah/OIoojqUbjwvbOoHpXxuVGeBMNHBgknQ5R6XH+tVH/gFOEGVEAbV+r3+QrcXqjuDtbcdOkJ8XDPI8duunRN58fu4uJkQYwieN7CNSUzcTofuZy7NIw7tC0gCJ7eAJDlFlbnNhgb9GPu7FJv2nQEUHoeMxGwLoO9javVH0VQJw7tD7jTnTkQQohy4sDveRj7BqS0F57o5/qwW+5zC17XyuCe1/xTwFXkrLiv/4Aj5dfYj9+Zwg6u1O2goIetuzvcvbSCrIjsLO9y/u2znzsQ4suEJEsIuLSSbdpI6qPvUVZkoqkIdy8t8+G//Rjd72H8+Ah6UEfVFDyxAFc+uEk0GSEUC3Dpx9d49Y8uEEmEUTQZxaNSK9YJxUNMnBijuFsiGA2wfnODaDJENBXh3LdPU9gp4Q97qZcbXHzvGsFogJ2VXW58dAezY9KquXriYxdmaFSayIqE7n+ACOs16h532vcFvbz0/ecxDRPVo7J8bQ3NoxKMufRTq94+UJHTqrdIr+XAsYmm9sYBdltd7n62jKxIrN3a5ML3zqL7dQR5Aoe2O+giP98fWT8MfZ8ZMbR3mnTPN9XcAuVUQBh2feAPgN2+6HLcjuhOhYqCOxwkSv1i69hNl8awGz3ddsLlte1c788DIFjQ/tBtVCpn3Wsq51xfGfMKSFEE7UWXIrF2eieJOgjmI4p6HafzEe7Ov4sjvIrw4ODVU8L1xNnsafeHD10oBGUap1twVUPKlFvgn/a1uld79g8Cjl1G0L4eXjoMv1OF/TCUshX8IS/egE4hXaLbMdB/Cwt7IOLn2EtH3J3tqTH8ER+NSuOR4RL3OG5JFjG7FunVDDNnJ2k1OliWjSiIyLLI2q0NMpt5Oo0OF753jpGjQ5z95knSKxk8Pg1ZlSllKr0FT8C2nf493dvlRpJh1m9vce2DeSRFolqoo+kaHr+Koslc/clNKrkqiqZy4XvPEYj4mX5ugvXbW8RHYvsyURvVJt22QTDq7y+0Dy7O0cEIG7e3KaRLeAMevEG9p6jJ9pOVuq0uP/3BR33PntTkAO/88+9w/ILrp14vN5BkkXAiRGG3RLPWdgu7oCAop57oc3HsCk7nF4DtNge11/u7Qccu4hhrriTRbrij+FYNjA/cQSLt5J5r2VYROu/ixhJ03N20Y7hDROob91/T2nS/Lo2CtYkgT+Io5xEEE1BcyaJxC9eX3YDuJ25T1X7fLfzyW25xFHzufYse13fdcXfshz9sE9fQLN5rAte+WGG3MmBcxh2o2u55su8XDghiz0Me8/PvtO3K/YEnu/y57/kfAr4ShX1wKsmNn9+mVWsRG472fcl/2+A4DkNTKYamUmze3ebi310Dx2HqzMQeCuNB2KaFL6DTqrZoVpsEwl6OXThCeiVDu9Hh6PkZbn10h9xWHt3nYXctz89++Alv/EcvMXp0mCPnXDWEZVoUJorktgoMTif2hVo0qk3ajQ6xoShnv3ESVVcx2gblXAXLcAjGAoSTYWKDUYq7JZrVFpquMT43uoc6sW2b3bUs+c0C6dUsiioTToY589YckrR3sY0mw1z43lnajTbBWABFVbj18R12V7MgCBR3ywyMxKkWauS3i7TrLXKbBa68f4OpU2NoukYkGUYQBQq7JTxeD4HIk/deHMcCTLfpCT3P9FxPZugqUpzOJ4AK9iYQdKWJ1rYbk9d5F1uZRhTv7/AF2i4fLYhgd9zMUc83QFAfGthRcCWIbXDAMe649r5iotfwVHoOlrOA7vL0nu+DeRmEGCgp7vnAC4IC6sv9ISOhZ7zl9D3pg/e5djEEgq9HxUhfqKi7L9IAR0GQgg9M6x6m4Zf5QiVHOeoGhiC4dslf41B8JQp7YjTe59+DMT/1UgPTMAnFg1+YkimkS6zPb+IP+5g6PX6o2uVxKO6WuPijqxiGxYW3z7J2c5NQPIAgCqzd3GB8buRA5YI36GXu1WPc+XQRyzA59uIRHNth5rnJ/ve89AeuSuXOr5bQ/R7iIxF217KMHr3P3+e23alYzefB7Fp89P/8Co9XZez4KJpX4+pPbrgmWJaD49gYXZPzbz/HkbOTWJZFKVNh9eYmxXQJVVfwhQ4+Sq/Pb7J8dY30Wg6j3eW5b56kkq3SqrXJbRfIrGaJDkVJjMbQdBV/2LenEV5Ml/EGdHJbRfJbBYb/dBBNV2lWW9i2g6opGF2j/7n6wz4uvHOOVq2FP+zrUzm2bfeoHWUf1+/YDXei0rjt7nqlGDhmv9jdaxq6gRn3uG4bpDCIASAOeN2hp+41HO38A4Uz4hYda93lzj2vHEgDCfIYDk23eEuTYK+68kF7B8dKuf4s8rTrJyOqbgaqFASO0LcHUF/oT34Koh9BPPHAMxZxOp+69y/GelpzaZ+h2eMoqsdBkFI41orbFJaih/raPAuI8oSrm+de0/trHIavRGEH+sVhZ2WX258sggPxkSin35j73MMTnVaHGz+bx+PT2MxUUDR5z8CMbdtUcq6tbjgR6r+Obdt7FRqOwyd/e5GtpV2wHd7dzHP6zROUMhUQIPLAzz4MQRCYOTNBYjTGpR9fZ3sxzc5ShvNvP9d/Zo/Xw4V3ztFudei2DERRIv5Ak7PbMbjzySK+oJf5TxaoFGpoXpVqrkZqcpnps5PIikwwFqCQLnHsxVk0XSMY86OoCnc/W2J3JQsORAfDHL1w5H7WqWVRzdeQVZlAxE85W8VxXIVQdj1LYauIN6jTqrVYvrJGMBbg5//uU2KpMKF4kDNvze2xXxg6kuT9v/oZRtcimgqR3cjxjT95Dcu0yWzk8egKL/3+C3sWWN3nwegYXPnpTQRg9vwM6/ObFLaKKB6Vc9862ae7HMdwp0yNu65hl3LSHUxSz7l9Ail6v9gJfle7bW2DqCLIUzj6O9D6oatHl0+AtYbTVUE9iSCoLj+vf8fdyQq+Qz3GBUF2/WkAxyrgdFcB2x0q6hkDiPIYjhgBLBwz53rHyFMI6unHKkoccwtwOX3HzvZ8ZEK9137I0OxzwnGsnvqIXsDIiS89Aenrgv5k+MoU9nvYXc3iD3nx+Dwu397u7pEUPg1Mw3I9zH0ejI5Jp9Xd8/dLV1bZvLONA0ycHGP8+DDznyxQ3CkxOJVi9oUpRNHVENfKDQRAVCRajTbjcyOEE0Fs22F0duix99JudHBsh9hglOxWntxWAW9Q7y8ggYif7/5nb1HYcc2+Yr3JXNu2WbqyyvwnC+DYFHertBpuZmktXyezkaOYqXD8xRlMw0LRZGKDkf57ZlkW20u7RAcjmF2TZq2Nx3svas5h/uMFsj2L4WMvHcHj15j/ZAFBEPAFfIydGGV0dohmrQWCgG3bNEt1l0rxqOws7RJJhsms5yjnKkSSYUaODOE4Djd+cYfVG5v8p//tH/OP/su3e6cFdV/ea7vpWgt4Ax4kSeLK+9fd92oo6urlV7P3TzhOx6ULpITbBLVzIA8jSAP7uF/Xr/0sOLM4SD3XRQ30f+IacBEF8yLYeZzuZderRTnmFvOnCY0Qo65/u7UB8iSCdF/meY+vFtQw7m79Sa8ZAnMVx7bd5uyXoCBxrN2eJUIErPWeC+Rv10T4P1R85Qp7bDDCwqUVGtUWgYgP5TEj8o+CN6AzMjvI9mIaVVf3FeCdXsFzHIedpTRev4f8ZpHYUITtxR2S43EiSfdoOvfyLD/4F38DDowdG8bj9/R3qrZts72UplVvk5o4WKPvD/uQZJHdtQzr89uYHYNytsLpN+aQZAnLtKiXmwQivv7u1DRM7vxqkfmPF5h+bpwPf/BLRo8No+kqH//tZwRjQWLDESzDYvaFGSRZJDYY3bMQSpLr617cLePYNsNH7r8HnVaH9GqG5NgAtVKdax/OEx4IcvSFKQIRP+1Gh9T4AN6AjuZVSU3E2V3NEk2FcWyHZq3JyOwgxd0yN39xG4/Pw85yhvG5Ef76f32XVrWN4lX5N//jD/mD/+I76AGd8ROjbpB2tYXHp2F0DC6+d43lK6v4Qj6mz07gOA75dIn8Tglf2IseeKCoCbpbSK2M29iUhkE5f2hDzx3B97vhEeYibnSeH9ST0P1lj6bp+coYN3AEFUGZfap/Z663+jF3ErYHx+n2dtnez9VsFKQxHNWN4xOk4S8pncjqTdyqPWXUk1v0fo0vF1+5wj56bBg9oGN0DAZGYk/lHPgwBEFg9vlpJk6MIinSvuZffCRKZjWHAwzNJBEl0Y3SM9x/4K53iovwQIiXf/8FvEEv3Y7hTqj2JIPbi2nuXlxG1RTSK1le/P65/qCP4zjkt4s0Kk1OvDzL1kIaxxEYnExQSJeoFut4gx5+/sNPqeZrrt3uWydQdZX5X95ldyXH7loWQRI49uIR/GEfgaifbwU8rN3YwOiYBMJ+YsNR/MGDTbdOvn6czGoWSZFIjrscZ3Yjx2fvXWP95ia7q24jNxQPYLYNauUGqkfDH/HjDbrPKEkSJ189zvEXZ2k3O+yuZfF4NQankmwvpZFkiUDETyFdIjWVZPjIIOu3NtE0hfRahss/ucnU6TEqhRoCrs2BoikkxuI4tsPJ149z4+e3qeVrhAbD5H+1RL1UZ2xuhMTofdrBnaq80LOv1Q5UcBwIK99TZGjg5BGUV3CEGHQ+dkf+xRigQPcKNl0EefZzD+C4TduPe3SO6k7N7vOWfzQEQUCQv9xsYZdf33K9aqQRd8H8Gr8V+MoVdlEU9/wiPws8rK2ulxtkN10HxtiQG0w8MBIDAUZmh8hvFxg/MbrHU90X8qL5NCTJHcbxPDBEVSvW0X0e/GEfhXQJo2P0C3t2M8+Nn82jqAqiJDJxapT8TolGpQkCqJrM5fdvcOeTRbxBHcuyuPbhTRwbFi6ukJoYYPLkGBt3t3nzn7xMYmyASr7G8NEU7UYH0zARRYF3/+J9pk6NceqNuT7Vcg+ZtSzp1QzRVBgEyG8X+Lf/09+S3y4QH4khiCKBiJ8jz09jGRbBWotTrx/HH/H3G5f3FqhWrUV8JMb06QkAyrkKpd0y9UoTx3YIRP2E4gG+9adv8C//+/8T27IZ6Z0yAhG/S/uIApFkmFatRbvRZnshjWEZjM+NcOabJ3n/X/8MWZEZmR2imC5TKzdZubZKo9x07YOPDD49xyxP9IIxHJdPdjqADdpZl/Zwqj0OfBbMdRzH7KlbPgd63uqCmMCxcjhWAUH049glHOM2oCGoc4fr7X9NEAS1b30gPKG1wNf49eArV9i/bHRaHS6/fx0ctyl58vVjpMbvD2VMnhqlWqiycXu7R3FMIwgC0VSkP0kaH4ruCW/uT2SmS8SGI26ARg+1Qq3XyAxQSBcJRPwcPT/NzkqG6dkJPD4P9VKDSCpMu9EmvZwhvbSLHtDxhXW2ltIcu3CE3/vPv8nMmQkK6RI3P7qDZZjUSw1OvznHnU8WkWWJRqXJ1sIO02cmaDfaSIpMu95m4dIywWiAtVtb+CN+5j9ZILeZJ79VJLfpOkiefOUoxUwZAYHZF6b7NJOrtLHJbRW4+Ys7yIrExp1tLrxzDrNrcuXvb6KoMooqM3l6jLFjI0iyxPixEf78X/wpdz5bopqvI8oi1UKNcDLIxfeuI4ki4WSQ3fUsF398FVGSSE0OIEoC5UyFWrmOKImE4gHSS2ka5SaBiJ+7l5aJDkXQfR667S6FdAnVoxJNhR/ZZHcbmSHAwnEUV/fuWLjF/VVXTdP+0NWY03ZplKeA4zjuMJOg9oZ3BBy7hGsN4Hcbld3PwJGACk7Xefrpzy8B7nv25RV1x3HIN5t0bYsBrw9VerTKrW0aSIKI8pjv+6rj68L+lLi3y40NRqkWatRLDRh3w6LL2Qq76zka1SbRVJitxTSDU8l+lmd8KHqg3XB4IMSL338eo2PgC3n30EeJsQE2F9L9ou4P+8is56iXGiyXVpFliZHZQcyuSW6rgO3Y1AoNWs0Ous/D0fPTvPjOOYKxAK1Gm0/+/SWWLi8TG4pSr7gnj27XxHYcbn10h1qpQaPaorBdRJLdPFXbclA0BVEUXJteQUAQwB/10Wl00AMeTr91glrRLab3Qqy77S7Xfz5PrVCn2zbwBnWCUXfKtNPsYBoWlmWha+4AVbdt8GBSxND0IMnxBG48oesAeenH1zj9+nFqxTq1Yp3NhR3CAyEkVSK9kuXo+Rm8QZ2d1QwzZyZ59R9foFFuuhOStuOmyAmCe7L54BbVYh3Hdjh2YYbhI7Ge7tt/oMdK3xTMyuJgufmqdh7sCoIUxhHjrlujoLlB0w/Bsau4CUHhPQEVrrnYFdcvRvAgqC8haK/03BUjCFIMxzF6Y/g+d0jqoXQhx+kC9q91zL7a6dAyDMIeD5r85ZSS1XKJa7u7iCJEdS+vjo4jHrIAz2ezLJcKyKLESyOjRPT7myfbcVgsFsjU64wEg0yGI8/Mavi3EV8X9qeEP+xDD+gsXV1F9SicfnOOWqnGL/76U1ZvbFLN1RicTuALuilGT5qTqvs8B3rcBGMBXvr+OdrNLoGIe83txV1iPZXK5t1tZs5NsnlnG0GA5MgA/pCParFBbCjKhe+dIxQPUsyU+X//tx+xcn2VTtNA93mIpSKcf/ssuY08F9+7ytDMIKqmcP1ntzj16nHSqxl+8cNPESXRbXQeGWJgNI6syixcWqGar5IcjzNyZLjXZN079LS7mqWarxFNRVi/vYXRNTE6BsF4AG9Ap9PuklnLMr/j9gpa9TaVXJXnvnGyT3/1J1a9Eq16C8uw+jTUwGiU9GqGnZUMlmkxMBpj8fIqpmEydmyY2HCEwckk3bZBo9KkXmowe34aj1ej1WhTLzeID0Vp1VsUdzcZGr2OW3gjOGICaLo8tRByTazsrCt/lFKuZ4ydwx3Nb+F0broPLYZAfXnP0BKAbW7soXL20DROBex0z5WxiGNtISrHcMM6XBtgQVBw5Dkw510HR+W5B66d7k1/OjjyCURlkmeFYqtJx7SI6vqe4l1sNfl4cwPbcfApKq+NjT91cXcch+Vikd1GjaHA/mLrOA7XdtMUWi1GgkHKrRYd00RX9p8Q2qbBcqlAzOuj0e2yWMxzYfj+4Fy2UWc+lyXi8XA9s0tI8xDzfjEN/28zvi7sTwlBFBAEAVVXkWSRO79aJLdV4FfvXmFoJkViMk61WMcyTY69OPNIu4Anhe7X+94sjuMQjPvJbOZpVZuMHR/lyvs30Lwa5VyFzFqegdE4vpDOK390vq/KufbTG+yuZhBFiU6ngaKrzL00y9qtTbrNDpIkk5pK0Kw2UVTFTTe6vs7w7BCDU0ma9Tbnvn0aQRBIjg3wH/83/4gP/+9fUs1VadVaVIu1/k79HkRZdO2AOwaBsI8Trx1D1RQCPYuBRrlJYmyA+FCM+U8W8Ie81Euu1/vgZHLPtVr1Fhd/dI12vc3tz5aJJUMkxuNEUxFOvXkcq2MiSAKZ9QK2aeMP+8htFhAEAY9X4+w399oLaLpKKB4kv1MExyE11sHodFH1ZG+UfwOkmBtDp8y5w0xiGIxrrhJEPY/gdNwpTmMeHNU157JvgjgC6kPKGHPNDaZGA2sLx5l7oLnq/texG70Q5zvY5up921z5GKIyg6hM4sgjuDa8D5wozPneIiCBOY8jjx+YR/q02K5V+Wx7GwEIezReHZtA7p0ms/UGkiAS9+pkGw3q3e5TF/Zso8HNbIaQx8ONzC5BzUP8gWK7U6uRazTYbdTZqlZ49RGLhySIyKJEo9ulZRqklL3NZsOyEBDwyAoCbUz7YN/8rwp+Z4I2fltgdAyatRZjR4eJpSLMf7JAfDiKP+Jn8/YWnXqXqTMTXHjnHENTzz4JRhAEjr4wQ7PSwOiaLFxc4tbHC1z76S3azQ6DM0k0n8rr//glhqaSWKaFaZhU83VatRYCArIocuHts4wcHQLbITmeID4SJb9VJBQL8q1/+jrRVJjBI4OEBoJ0Wl28fo/rX9Jzh1Q9KoGwn1Ovz+HxaWzd3dl3r4OTSUaODmFZNkfPz5AcGyCSDPeHizRdRZJERFlAlFyax7bsA43AaqUGZtcklAihqTLjcyP4Qm5INzZ02ia+cIChqQS+kI7q0YgNHX7cFkWR02/OcfqNObodgw9+cJ1f/PBDclvLgAGiv0e9WK4boSC6FIu5AZ2fQecj3OBrv6uJtxZdd0UxDOYdNyDjQUhxcMquCZYY5ME9lSD6QDnfi8izXUmmueQGWgix3vW6vc9f6Rd12ypgt38Mxl0cY5u18gaf7RS5lE4/k8KVrtXwKQpJv59Kt0vLuB/SEtE9dCyTbL1Opl7jo801Lqd3nup1DcsCQXB34IKAae+VSzaMLkOBIC8OjzIZjnByIHkoDaNILv0S8mhMRiIcje0NpEn4/MS8OrlGnaTf/5XercPXO/anhupRiaXC5LaLCAIMTSfpNLucePkIue0C06cnOfPWiX3SyGeJTqtLeCBEeCDER3/zGaPHB9ldyVHbaZAcTVDOVfnkbz9j+eoqqq4iIFApuva/0cEox1+eITWZAFx/9uxmjlKmjC/kY+PuNoGYj+MvzTI+N8Li5VUEQWBwJsW7f/k+lVyNM2/OMX12EkkWaVabdJqdvVrxHiRZ4ugLM/u+3u0YLF1eoV5pMjQ7iG3aJCcSvQGsiKu+eQi+kBdBEqgV66gehXajQ73UYGgqSSQV4ubP75Bbz9Ftd1F1lfHjQxx/8dF6clmR3bCRK6sEokOsLnRpd7N875//MRi3e8lKSTdU2Sm5gRfYIE4BLRxzFUGKuVF6ykl3xy4N9xqne5OfBPkojuADx0A4IARalJMgJ7GdCu4O3uf6rt+TPB5kO2xcBRRQZqg1F9msDuLRz7JVrTAUCDAUCO7/madAwutjq1qhZRr4VW0PBZL0B3htbILNSpmWaRDVfWxUyiS8PkZCTxYCPuDz9YvtgNdHTN9bbAf9AVZKRTqWyUw0Rr3b5T8sLhDRdc6mBvft3iO6vod+eRCqJHE0GqdlGgz6A/2Tx68LjuOwXatRajUZDgaJ6l/uwvKVKuyWZZFZz2F2TJITA5974vRhOI7TT3CSJImTrx2nnKsiKxKaV2P91ibCVJJv/9mbe16zWWu5w00eheEjg5/bZ+ZheAM6oiRSylZQNRnd7+XYhRkm5kbYXskwcWqMrYUdzK5FrVhCEAWOXThCp9Vl7Ogwsiox//FdREkkNTHAwsVlyvkamws7+CNedhbTFNNlpp9zFylBEHjvX/2UlWvreIM67//rnyHKInMvHyWznmNwOsnYsZEnvv/bnyywPr9JbCjK7nKGV/7w/CMDtMG1/H3hu89RLdY589YJ2vU29UqDrYW0O4imK/giXmqlBppH5dy3z+xJZjoMiqZg2w7LVzcwuhb1uo+3mgN4Az2TLcGPIIg46is4SrO3U6/jZqe6RcSdUL2AY1zqJRmdPmCKVeqbcz0S8hnXnleOAxPuyUE5Srnd5UZ2F0kQOJ0cJKBpuPy+DcggTlCyJ4k7HhyaCHzxxuBoKISuKnRMk7jXu68Yxr1eHBzWK+X+TtruLWiGZdE2TTyyjCwenOCkyTKvjo5jWBaqJO37noCm8Y2JKTqWiePAB+urxHSdXKPBSqnI8YEn85IHWC4WuZHNIIiwU6/x0vDor7V5uluv8dn2Frois1Yu8Y3Jafzql2dW+JUq7Ks3Nli7uYkoi+yuZXnh9w4Pcn5SWJbF/McL5LYLeAM6Z9464TYeHxhrP3p+/67Utm2ufXCLbruLZVh0Wl1mn5/+QvdyD96AzvPfOUM5V+HEK0ep5KtIisTgVJIr71+n3exiGxaS5P7DdRwbs2Mye26K53/vOX717mUiAyHXInc9j4NDrVBje2mHUCxEOBFk/pMFGpUm0cEItmWzeWcH1avSaXbZWkyzfHmN6ljtqd/j/E6Rax/colZqUC83SIwOYFlPdnx/2DDMcZy+VfP43Cibd7YZGIohKiKNSqNf2FuNNmbX3Kc4AtC9Gq/98QX+/q9+Rmw4yvCRQVr1Nr5gdI8tgCAICIIPR3vpARfF+01KQfQhaG/wRSHKCRzpu73XlPrP+dn2EqIoYtk2V3fTvD4+gaCedT3KcQgGXuZI12CrWmUmEiXh++K9HUEQGPA++jox3ct0JMp6pcJQIMigP0DLMPh4c4P1SplCs8GpZIpXRscIavcXu3uUnigIj+TmNVlGk2VqnY6raOotWY+Iwj0QO7UqEd2DR1bINup0LetLU/IchFq3iypJhD062Ua9dwr6urA/EUqZMsGYH9WjUsqUMQ0LVftihb1WrJPdzBMfivZSj3JPlPBjGhbtRptIMkyn2aFWPFjXbFkW3ZZLHzwNfeMP+/AGdWzL7k+DAjz3zVPsrmUJJ4JUczXGTo7i0VW6bdPlpQM6vqCXWrGOZdsoikRmNUez2sRx3IZnt91l/NgIDnDjF3c4dn7GTVXadVOf5l4+yvDsIMVMGaNjPNXJKL9VYPjIILtrWYrpMqfeOOFKH7vmvgzVx0EQhL5t8OTJUX70v3+A7tfRfBqKx6UNCukS1z+cx3EckhMDzL00u2+nduzCEZqVFqZh4vF5CEYPvw9BDCKIT+bz/nlxkNTScGwCkoIpCBj2vcnmcC+RycWJBJxIJPf97IOotNuslct4FZmpSBTpC258REHgVDLFyUSy/76mazUqnTbldgtHEKh2OtzO5XhxxP29KbVafLa9heHYnE0NHkoZ2Y7TPwkENI25eII7xTwRXWcqOGqjWwAAIABJREFU+vg4vgcxGAhwM5dFoE3C93g9/LNGyh9guVQk26gT1jyEtS9XlvqVKuzDRwa5/csFEARSkwkU9Ys/3r1rtOptLMNEfUKvd1VTGJpOsb2YBsHVqv/q766QHIszdty16O22u1z5yU3XZz3m58ybJ/ZZzB6GRqXBtQ/naTc6TJ4a7btO+oJeIokQazc3EQRoVtrMvXR0Txbr6TeOs3F7G0l2j8jtRgdRlhhpdYgORlE9CoIoUs6UEADbskmMDnD+7edAEFm7vkFhp0R8JLqHQul2DBYuLtOoNJg4OUZybODh2yacDLG1kCY1PsDY8VFsy+LSj68jAEdfnGF4+tE5p4chkozw1p+8Sno5QzgR7C92m3e38fg0dL+HzFqW6TMT+yZrfUEv57931t2ph7yPzK19GlQ7bSzbIezxHHrsN22b7WoV27EZDoYOLDiCIHA2NcjV3TQCAueH92e3Pgm6lsXHWxsIQNs0sYGjsWczpf3g83kUGdtx6JgmCAI47KFxrmd2kUQRj6RwZTdNyh/Y0xR1HIc7+RxLxSIhTePCyAgeWWEmFmM6Gv1cFMpMNEbI48GwbRJe369dwx7UNL45MUXTMAho2pfO8X+lCvvQVIpgNIBpmAQfk3D/OOyuZclvF4mPRDn52jFWb2wwcnRoz+74cZh9YZrhI4OU81XufLJAMBZk8coqgaifaCpCIV2iUXb15vmdIpVclfjw/aDoZrWJpMj7ChHA+u0tLNMikgixemOT1GSyr4MvpkuomnI/cq7W2lOsdL/ep48a1SY7K1lGjgzSrDQ5/dYck6fGKeyUuPbhLRrVFp/+f5dITSYYnEoweXqcdq1FfqtIdHCv6mRjfpPcZh5fyMf8xwuEB4L7dvOp8QSqpvR0+V4u/ugascEI3XaXncXdz13Y7137wSlggEDUT36r6J4svNqhi73Hqx34Pn9erJdLXM3sguMwGYlwOnnwc13P7LJRKQMC2Uajv6t9GEOBICm/Kyc9TBnyOBiWRdeySHh9iEKXWqfzyO+vtNtsVssEVI3RUPiJX3fA6+PC0AhhzUOp3WI0FGLuAT5cFgXapo0ggCTs7wbUuh0WCgUGfD6KrSarpVKfT/+8v9OCIJDwPd2J8FnjHq3068BXqrADBzojPi1K2Qq3PrqD7tfZXcuSHI/TrDZp1dqEE6F9xeNhdDsGmfUsoiiSmnC14YIguMlOjrsD7ra7rN7YYPHqKkbXRFblvhOl4zgsXFxmezGNIImcev34volVVVMwuybddhdRFJCk+zuAcDLcj5zTvCp64HBPEV/Qy4vvnKXd7OALevsDQapHQdM1Tr8xxyf//hKmafHpu1e4/csF/FE/oViAhYvLhGIBPD4NSZEwDAtJllBUGcexsQ/hzh+0G1BUhRs/n8cRBM59+9lTHBMnRt2FpNFhaCb1RMEr3Y6BZVp4vBr1coNmtUUw5t+b8/oYLJdLhDUPqiSxVi5zPJ44cMw922gQ93oRBZFcs4HjOIdLNB9R1BzHYa1SJluvM+gPMBZ2lUW241DvdtAkGa+iMBoMsVmtIAki4+Ewdwt5iq0m46HwHkqkY5p8vLkOCLQt07WmDj8Z/SEIAmPhMGPhMLlGg8VinpVSkdlYHEWSOJ0c5FomjWHbXBjeHzAjCiKCAKZtuXTMr1nB8lXAV66wPwsY7S6CKOILeWnVWqxc22Ds2DCmYbJ6feOxhf3WR3co7ZZxHIdKocbsuSliQ+4OfWA0RiQZZunKKp1mh5HZIbLrWd76k9f6pmHddrfvgd5utNm8s72vsI/NjboTldUmJ18/vocSiQ9Fef67ZyhlKugBD7LiFhTLstheSNOoNhmeGexbHeQ2C+ws7xIZjDB5chRJkvCFfSiqTDFdotsyqNNAUiSu/PQGwWiQU68fR1Jl5j9doFVtofs9HHl+ikquSjlXZfL0OLpf76cdHfTL6frIWGheje2lNBf/7hoer4eJuRHKWTefNfyIEJJHoVaqU8lXCUYDe5KkHodStsL1D29hmRbRVITirputKasyyZcn6EquWiTscYu8ZdssFgtucQxHGO4VxwHdy1KpiCwIhDXPoUfviXCYu/kc4NIFD09e3i3k2anVGAoEOBqLH/heWLbNB2srfLi+zmgwyHatik9Vieg6n21vsVwq0uh2eGtiirOpQWaiMVRJJNtocDuXJaBq/GR1hbFQuD9u3zZNl7bw+Sm3W4/d3R+Ejmny6fYWHlkm2ygiCgLHBxIENI3XxiYO/Tm/qnImmWKxWGC4N5H6NZ4OXxf2AxBOhPCF3GahP+rDE9D7zcaDvF4ehG3blLMVooOuz3lpt4ysyJx56ySmYSIrssuvdwwUVWFoMoTX7yExdp/rlFUZzatSyVWpFWuMH5CHqmoKcy8fPfQ+jK7Jyo11sB0S43FOvHKM7YU0C5dW8Hg1sht5Xv6DF2g3O9z5bIlAxMfajQ10v0at0GD+07vsrmSIDUV47q05fvqDj2g1OnRbXbwhH/Of3uXUa3PUC3XiIzEq+SqlTIUX3znXT5C6e3GJ7cU03qC3ryZ6EI7tYFk23qBOIOxD01XWb26Q38y7UXiOw+zz04wde3Rhzm7me8+lcuKVo9i2w8X3rrlZoo7DC793Zt9U7GFYv7WJ6lHx+DSWrqwQSoRIjMS5s7zF5tIasaEom9UK35ycwiMrbFYr3M5nCWkeLu1sE5zQCGgaxwcS+FUVw7YZCx2+OB2NxUn4fDgORPW9J4Jcs8HtfI6YrnMnlyPi0Un49vPDmUadtUqFoKpS6bRRRInL6W2qnQ4blQq241Bpt/m3t27wZ2fOMdbTmbdME0UU6VgGP19fYyYaZSsYQpNkAqrKRqXCpZ1tjsTivPiQPrxrWbRNA6+iHrpoGbaN5dj4FAXLtmmb5qHvu6vzrlLrdBgKBKm029zKZghrOtORKKr+m3Wy/F3D12ecA6B6VF747hkuvHOWF777HOe+dZLkxACjs4McPf9oyaIoigzNpCikS1Ty1X44hyAIKKrS/6WcOjWGKIuUcxWmz0zsC7c489YJmvU2zVqLzHqezEbuqZ5hezHtyjKHomQ38rSbHRrVJh6vRiDqxzJtjI6BZdpufKZHRRAFlq6scun961z+8XWufnCLT9+9yu1fLZEcHyA+HKFdbyNKApIsMXFyFFGR+qoWzaO6/jiSRL3cYHshTTQVIb2S4T/85d8z/8u7GN3704uSLDFxaoxGpUm7aTAwEsWyLMq5KrGhKMGon8x69sDny27mufKTGyxeWXFDOnq5qEtX12hWWzi23bcZblZbT/y+eYM6rXqbRqWJPxZgq1rh/cu3WKu7A1xhj45p23RMV5nStSwUUcKrqDjQV6zIoshkJMpsLI5HPrwZKwgCUd1LzLvXCz9br3Mrm6XaaqFKMgiwVavy7tIC768sUe20718DgaCq4VUVau0ODg6VToeI7iVdr7Jbq7LbqJFvNvlgdaU/QToaDGE78N7yMoZloYgiG5UyDaPLh2ur5OoNLMdhq1rd09RtdLt8sLbKB2urfLy57k6QHgCfojAVjpBrNgCYjh68Kap1Ory/ssTf3L3NUrHI+ytL/PDOLQDWq2V+tLz0uI/tQDiO85W3DjgMX3jHLgjCKPCvgBSuY9FfOI7zv3zR6/6mIckSvqA7HfZgs/FJMPv8NKmJBIIg9OmOh+EL+Xjp95/HsZ0DeV/LtKiX6gyMDaD7PGwtpA9UmRyGYDxAbrNAt93F4/Ogau6QVG6zQCFdIjEWQw/o6AFITsTJbRbIbeRYvbFBZiNPtVAjGAvgDWhs3N4mNhjB4/W4lrd+nUgiiDegc/LVYyxeWiE04Hq33IOsSCAI1Io1Nu/uMH16nN3VLHrAsyc3dvLEGPGhCCvX1mnV24weHyazlie/XcRxbKbP7je0alSb3PzFbbwBnexGnkalQSgWdOkex8Ef8aFoKsXdEpIkEniEfPFhTJ4aw+iaVPM18CpUOjbTz01S9jnUMBEbdRI+X1+DPNLjrLONOsOBIGGPTrXToWtZhD2HUzD30LUsTNtGl+V+YS+3W/xyexNFECm0WiwXi0yEQ6yWChRbbQqtFqVmi9cnJjAsi5juZTYWw6soXBhyXQ1vZHfxyDJz8QSr5RK2A6PhMLoiU+920RUFSRCQRIGwR8O0TPKtJn5FYzgQ5L2lRcqdJkHNQ8Po0DQMfL1nzjTqdEyThM9Ppl6n3G4zcIBuXuhJIY/EYsiidOB7YTsOn25vsloqUWg2CWkeZEGgY1p4ZIWWaWLYh+/0D4NhWVxMb5NrNBgKBDmbGvzC0s7fJTwLKsYE/mvHcS4LghAALgmC8GPHceafwbV/JyEIwp6QjcMgiuK+M1N+p8jy1TVWb2xQzpV7zdsBzn3nzFPdw9ix4Z76pMPgZBJJlghGA7z0+649sB64n5d64pVjGF2Dv/of/i9Gjg0j6yqFn97C6Jp0mgaJsThnv3mSuxeXiA/HiCSCBGNB/GEfRtek3ezQbna48YvbnP3mKSRJQvfrnHj1KAsXl4mmwiQnEzTKDfeE8BACkQBn3rpvc5sYjVPaLSPK0r58UwCza+I4Drpfp93sEE4Gye+W8Id8TJ2ZQPd5OP/2c9TLDbcJ1zWxervKZrXVbwwfhPXb22TWsty5vII+GaZjdKllKoSeG+JMMknM6xb1e0XCqyi8NT6JYVtokkymXudXO1s4QMLn48Xh0UObnuV2i19ubWJYluuF0tOCtwwTAYh6vcxEY5xOpRgJhviXVy9RbDXRRKnffPSpKlGvzquj45xJucobw7IotJpkGw3OD4/w9swsP11bQZEkAqpGQHMLdMs0sGybC0OjXN7dwaeo/NGx4zQNA3AotlosFouMhILUup1+8fbKCqZtk6nXSNerZOp1Yl7voc/5qBOLads0DYPxcIRSu8VOrcoro2MkfAF+vrFKQNP41uTTDfaZts3l9A43M7scjQ+wVCigSiLTkVh/cfqq4wsXdsdx0kC69+eaIAi3gWHgH2xh/7zotrvc+NltbMsiu5FlfG4Uy3KIDASYPPn4oagHIUkSwzP7JXaqR903vi8IAqqmMjAa486ni5QyFabPTfLqH54nNhQhv1nAH3Ej9USxydZCGn+oSiFdol1vo/s0fCEfxd0y7Xq772iZHBsgMRpnaCbFnU+X6DTbjBwd6nPwh0FRFRK904njONi2TafZod3o9O8jOT5AdqOApivYpoOiyICD2IsjvKdoufbhPDgO4aTLc5fSZSRF4uw3T+07TXU7Bpu3t9CiPtasBsqOiTodJV+ucjJygvFwZM+uz3EcNiplMo06Md1HyONhqVTAqyj4Vc2dMHxgp/swlopFJEFAU1QupbdJ+Hwk/QGiuo5PVck1GngUhaTPjyyKnIwneG9lGY8m45EVQh7XejbXqGM8MEmpSBIXhvdaPHx/9ij1bpewR+8XWr+qEdZ1Sq0Wp5IpXhwaYade43J6h1yryWQowkw0Rtzr43Y+17fVTfr9nEmmeG95kZBHZ7FUwCPLzMRih36mh0GVJKYjUZZKRY5E48wNDDATiyOLIq+NjyOL4lPLO29lM8zns6xXKtQ6XcqdNpbjsF2t8ebEJN4DbH+/animzVNBECaAs8CnB/zdnwN/DjA29uVmMf4uoFlr0aq3+4Hb24tplm+sc+Xvr+MLe6kU6uQ2i0ydcePqvkxTsXv41j97g+JOidhwjNnzMxitLqdfn2N31Y3GC8WDGB2TqVMTNKpNVq6tc+ylGXaWM7RbXTxeD9pDWnBBEBg5MsjG/BbBqJ/VGxt4vNqBi87DqOSrXPvpTbKbBUzDJDEWxxfyce7bpzjxyjGOnOtSK9W5/uE8scEoxd0y+a0io71m6/qtLeqlOoFogPRyBlESGZpKUSvW2VnZ3VfYJVlE1VUy2RLegBdZgI5g8/zrJzmdSO47yuebTS7vptFEiR8tLTITjboUgqLQNk18ivpI3bJXVlhut1gpFjFsm4+3Nvn25DQBTeP1sQmaRhddVvrXGPD58GkK5W6bsUCIbKPORqXM0Xj8sZOUQc2zZ6Qf3D7AS8OjVDodFFGk1umwWMgzFgrTNk0+3d4k5Q8yFAiiiBLVTgdPT4ud8PsZCoVIeH3UOh3KD3D+T4u5gQTj4QiyKOzZ3X+e6VDbcbi0s03N6BLXvezUaszEokxFIuSbTWqdzteF/WkgCIIf+CHwXzmOU3347x3H+QvgLwBeeOGFp3R6+GqhWqxx6cfXcWwb3a8z/dwEtz9dYGdpl9WbG/jDfganE8yen+Lct04fqM2vFmu0am2C8cCBAR2fB76glwu//zy7K1kcw8Ib1JFkiZHZIUZmh9hZ2eW9/+NDthd38IV9eEM6w9ODeLwe2o02saHogUZn3bbh+sjHAm7qVC/RqLhbptvuEhuMHGgCduezJZaurbM+v0mz1uIbf/IaADvLGULxAKF4sPfsAvVyw9We92IFLctiezntTuCKAkPTSaKpCNVCjXari/8hn/x6uUGj0mTm7CTi8g55n0MjrtI2Omw7bW5mM32qo/9cloWIgI1D17IJajqGbDEaDBHSPQwFgrRNk81KGY8sMxYK71kcjsRiZJp1dqpVnk+k6Jgm1U6HgKahShKqtFcJslmr8sLgCJokcSefQwBCmoemYR7qfZJrNKh1OsR9PoLafvpJkSSius4vtzbINhosl4p0bJugqvJPT57BxsFyHAzL5sP1VRRR5OXRMQKqhkeSWSoWCGka46HPb1EtCMIz803J1Ot0LJOWYdAwurw5OYXj2OSaDRRR6pmnffXxTAq74CbZ/hD4N47j/Ltncc3fJIq7JRavrOLxasy+MP3MCmf/+ukyoigQTkYp7paoFmp0WwamYRFNhVF1Fcd2XJ/3A4p6KVvh8vvuGL6qq5x/+7ln5mR55Owkmq5idi1Gjw7tUWoMTaX4xn/yCj/9wUeoHoVus0O33X2sBNQb1IkNRSnslJAViaHpJNuLae78aglBBLNrMT43yuBUYk9vopQrc+Pn84iSSK1U5+ZHtxmaTNJpdVE9CmPHRjhyboozb50gs55jYiDQN/7qtg38QS8nXztGtVAnNZXgxMtHSa9k8YV0Bqfue6rUyw0++9FVHNtGkiXOv32WaWGav747z/FQioCqsV2r7ivsca+XoOZSLn5VxXJsEAWmolEiuque+WBthY5pYdgWHdPi2MD9BrgiSbwwOEyza9A2DURBIKBpWLbNzWyGrVqV4UCQU73TQszjZaVc7LklQtLnJ6J7+6ZSDxf2TL3GL7c2kUURqSDw1sTUgelDLcOg0Gwy6A+gihIts8uZZKq/EFU7bT5YWyXh81Npt1kvV1BEkXrXNeZK+QMM/IanOu/Bsm1S/gDT0Ti5Ro1zqUF8qsJGpcJOrcYHayvMRGJMRaO/duveXyeehSpGAP4SuO04zv/8xW/pNwvLtLj5iztouko5W2H56honXz32TF8jEPVjdE2Ku2Uk2XVlrGQrbN7dxhvSqebrDIzEyG0WMM4Z+/xjyrkKiiITGghS3C3RrLWfWWFXVIXp0xOP/J7Jk2OEB0IUdks0Ks3HvrYkSZx6/Xi/cal6VFaur+MPe2k3u9z6+C6yIpFezfDS98/1Jzztro1jO7RbbYy2gaLIFHcrTJ4exxvQ2V5MM3N2kthgZF+TVdNVYsNRCttFAlE/kyfHCET86Kc9bC+mWbm+zvBMCt2vUy83cGzbpXPSJerlBoFUiJTXz1qpRNznO9BTRZNl3hifoGWaNLtdSp02Md3bz9o0LIuWYRD3+mgaBqXOftllQNN4fXyCWqdDQFMJahq79Rqr5RIJn4/VcomU30/KH2AukcCnqhi2xbnBQS6n02QbdSK6jl/d/xmU2m1USSKqe8k2GjQM48DCrslyvydg2w7nBoeZjNxfrDXJtd6ttNu0TIOApnI3nyPp8yMIkGk0Hvn530Oj20XsBWvUOh3WKyV0WWU8HH5mRTbh95P0Byg0GxyJDZDw+ehaFqulInfyeRpGl81KBdO2mUs8ue3v7xqexY79VeDPgBuCIFztfe2/cxzn3Wdw7V87bNsdhVc0xU0f6j691OpxiA1GeP7bp6lXGkSTYXxBL+e+c4bp5yb47EfX6La6DB8Z7FEV+wt7JBFi9fq6axmgK/iCX/7wxj1Pen/Uj2lYFNJFFE3BG3yywABRFPecPgZGY9z+ZJHcVp5AxE90MEIpU6Hd7PYLuzegc/zlWUq7ZYrpMi//4QUWLy2RXc8RjAUZGI0dOvgjiiInXz1GtVBD0ZT+ay9fW2N7IY2kSOS3Clx45xyBiA9REimkSwiSgC/k5bPtLffPqkpE8zA3kKBpGGiStIdOkUQRv6riV1US/r27Vo8sMxwMsdnzKz9xiH94UNMOpEkehiyKe7TgIU2nZZqEeqZSHdPEcpw+h5z0+VkqFsg26vhU9dDXkEWRl0dG2W3U0SSJQf/e3oMmy7w8MsZauYRXUfApKgHNQ7bZQACGg48O1riXbTqfzyIAJwaSLP3/7L3Jk1znueb3O/PJec7KrHnASAAECUIURYkSqSvdqfvedrhth8OO6JWjVw7/Hd564U2Hwys7wh1ht90Odfftdtyrq5HiCBIgAGIs1JjznOdkntmLk5VEAVVAgQAnqX7cAERVZlYi8X7f937P+zydNq4fetfYnvtM3upPQpUkfjC/gOv7UwsH27axPG8i7dRRRInuAYvsHxPCni/y18nly5eDDz/88Gt/3qOyc6/C7Q/vh2HVP37pme1kn4d2tcOn/3idIAjThC78+KUDFST99oDxcEwi9+J67IfhuR6f/e5zWrttoskoqy8vEfjBM/unPEqv2afX6HP/2ga+F2a5vvL2+WmffudehWu/usGwZzIajpk7WUYURRbPzKFFVWaWCs8cXvLBf7yC7wVokdDa+Uf/+fdRVIVqs8P//dlndAOX7y0tMLRtirH41DgrbLkYRFWFpVSG7X6PlK5zrlB84iWfHwT0J9OgR5Xa+UHAtVqV7X6P+WSK8wdc3D5KyzR5b2cL1/dZy2Sn9r1D26Y3HhNTFRKqxvVGjc1en3I8zslcnqZpEJFlSvGnm+Y5nsfvtjboWxZ+ELCWyZDSI5TjiQN9cIa2zQc725iOHS5uokhlMMAPAsqJBC8XS5iuS0xV+MH8Vyeo2LtQfXd7k954zKlcnreXV5iJH20a+duEIAgfBUFw+Wlfd2wpcABza2VKy0XEQ5JfvkqypQxv/NPXwh5xJnaoLDCZTRx5TP556bcGNLdb5OdytKodLNN6Jv+VRxl0htQ2GsRSURbOzFFaKT5mQgbh30Mym8BzPRRNYTQYEUvHnmshWzw7z/Xf38LoGsydKk9PQ1c6DWqBQz4a4R/u3+NccYZ1u0NCU1lMpdnodZmJxdnp9bjfbnMyl2Or1yUiy5zJHz44JgrC1FfmqARBeGEZAN5k49Udj7A9j4weObCI3mo1UCWZtK5wv9PmRDaHJsvYnsvVieY9reu0TJOZeIL1bieUZsoqru9zqVxm6SmeLEPbpm/ZU/8YQRBYTD0eY7jHzUYdy3OJqxpXalXiqkZzaLCayWC7Hhv9LnFF5cJX3BIRBYHLs3OcyRewXJeoqv7RK2OOC/shfB3ywsOIxCPIqkyv0UeLatMJ2K+DkTHm5h9uMxqOOfHqCjOLBWRVBkFgNBzhuf4TPenHpoUki4f6yttjmyt/fy38teXg+z5za+VDo/EePi29iPchno7x8ltnkXSFRmBxu9VkKZXG8V0UScT2fbYG/dCNURa5VJ4lqWls9LoYto3teyiSiCrJyKIztRA4Cu2Riek4SEJ48ZjW9QMvHeuGweZkIdnsdREFeNDtIiCQj0Z5Y37hsR18QtVomiZ+4KNK8vTP77bbSGKY3HOv00aRJIKJysX2XJZSGfqWRWs02lfY9yZiRSF0W1QliaiioEihC2V3NCIqK3y0u0NajxzYJxcFAT8IsD0PXZLQJIm0rlGIxShM7i3mkqmvJVhamFxMH6tijvnGcB2XK39/jUHHQBAFXnnnPNmZ/TujIAjYuVOhvtWkMJ9j/tTskU8XnufhOd6BxXT92gaD9pBYMsqN398iM5MmkYlz7oenqd6vU16doTB/8CDK3U/W2by5g6yEXjcHTd9aozAqMFvO0G8PGHbMI73mF8HOnQqff3AXARguRgjyMURBoGka/GB+ierQ4EGnzZl8gbVcjqZp4AcBMUXl8uwcW70eryVnsX2f+502cVVjNf30XFUIvV9+cecWjaFBe2yErQdB5EeLS+QfKWx7f417u/bacDjp40+Gnlz3MXng6XyBIAi43qgTU1U2e12WUmmCSSsIAjJ6hLlkksqgz+lcjoFjUxsOCQjQpARt0yQbjdIemby3vUV1GD7XQjLJa7NzzCaS/GhhiWv1Gk3D4NebD5BFkRPZHJbrPnYZebZQZOQ63Go0KScSGLaDJIokdZ2x5/Gg22Wj12U2kUSVJVbT2T+ZydCvmuPC/i3Dcz2u//4W19+9zYlXVyAIwlCLRwp7t9Hn1gd3iadj3PrgHtFk9MDx+0cxByM++eVnWKbF7FqJU5fX9lvF+uHFoyCGCZN74ZIHhVg8jDWy2Px8h8xMitFgzMaNbV7+8UuPfV00GSFVTNGqdJBkkfLK16dMeHB9k3Q+iSiJfHxvne8vvIwkCrRHI/LRKP/ZmbP0xmPutFq0TAPLcfm4ssu1eo1XSmXeeCgE4/RkOvKgxTQIAuqGgev7FCYxbOudDlv9HrokszsYYPseEUWib40fK+zFWJy1TJadwYCT2dAD5tNaderXoj8ia3R9H8/3Seo6EUUhret8Wq1wtx3a9XZGoVrne7PzZCIRXnnIemC71+P/u3+XP2xtoskKf3P6NH4QIAmhCsb2PfqWxf929Qp/deI0LxWKxBSVQixKa2TSG4+pDYekNO2xwh5VFH60uIwfwFqbt6dZAAAgAElEQVSQRRQEWqbJ6/MLoROmqvHrjQfcajY5UyjQMEzeWV752tuff4wcF/ZvGZu3dqjer+E7Llf/8TprF5dIXX68l76n1tFjOsOuieccrt7xPI9uvY8oCuEUp+WSyMZ59xcfsnlzh1OXV6c5risXFhl2w6Dp06+vHdoigfACd9g1SeXiYTqRIjEajBibNvn5g3eyoXPlSxhdEzWivtDUoqeRyiepb7UQRDiVz9MejYCA0/kCm70uV6phjGFqooK5UtmdSv0+qVYoPxTh9nCf2/V9NntdbM9jKZVmu9/jer0OQpgm9ObCIkldw/d9kCGiqnTGY+KqRv6AsGhREDhfnCGpaXTHFilN4wfzC4xdd2ovsMfQtnl3e5Ox40y3+oooMXIcRq7DaiaHLisTzXtoUHa33UKTJPLRGP/65jV+/eABuUiEs/kCv97Y4C/WTrA7GOL4Hu2RyY1GHV1S+H8+v4EoCJTiMda7bQaWxc5ggEConOmORoiiwLValQB4eaZEUtNZSqX4pFoBQWA+mSKlavhBwLs7W2z0O8xE47i+z9C28IIA+biwPzfHhf1bxtiwiKainPvRWbY+3+alH56huPC4hjozkyJTytCqdkgVE2RKh19i3frgLrv3akCo73Ydl927VQYdg0Qmxt0rD8iWMyQycaKJCN//60tPTPIJH/MO//iv3yWWijB7osQb//QyF98+z72rDzD6I1qVDrXNxoGOlJIkHep6+VVyahLK7fsB8yfLWGJAEIRyw3e3N0lqGlFFpW4MSaoaCVVlYNt4gYgiio9FuO1xs1nnbruNIopUh4PwwjSio8sKDWOI7XmcyOZ4Y36RnUGfM7k8l8pz5KLRQy/xdgZ9rlQqRBSFrX6Xn66sUYo//rVbvS6265GPxtgd9IirGpu9Hhk9guE6VAYDTNfmbD6P5/v8bmsDIQDb9/mwssPIdijEojRNk61BjzfnFzmVy9O3LO61W4wdj4CAV8olmoZJwzA4V5zh7YnnSts0qRoG2/0uf//gProso4jhovfh7i4/XVllKZ0hqem4gU9WjyCJIpdKs1SHA95eWmWj22Oz1+NnK2t/1ENDXyfHhf1bxsKpWVo7bcz+iJd/fI6FQ3rnsiLzyjvncCwHRVMOVc/4vk9tvUF+NovneBgDk/nTs9z79AELp2fRohrG4HFN75OKutEz+Oy3nyMpEvbYpVXpYPZN8nM50sXUZLJW5Mbvb5H8GuSYR0XVlH2WwQ+fFUrxBJ/Wqgwsm4yuo8kyF0uzfFav4vr+1HnxIFrmKPweSaZhGpzK5cORf8GiOGnFjF0XRZJIaWG7RJPDy8TDGFgWqiyR0nXqhkFvPKJuDNFleTIYFL4W2/P4uLLNVq+PT8A/OXkaRZJwAp+mYWC6YXhyzTD4tFbld5ubrGQyrGYyiIKAIsksJtJ4XsDl8jx/eeI01eGApmnw8kyJU9kc/+HebXb6faKKwpl8ns1ul+bIZDGZomGYdEYmy5ksiijRMkwW02kEwjDv7X4PTZYpPHIymU0meSlfpGuNOZ2XuTw7d+TovcPwfJ/ueIwiSfs0+57vc7vVpGEYLKbTz/083wWOC/u3jHg6xg/+5jKO7aJF1CcWWFEUnzj1aVsOmze3GfQNTGOMpissnp3n1GtrLJ9f5Oa7t+jUeqy+vPhMWbGe5xOJ6ww6Q4Ydk1QhSXyiXnEtF1mRUHUFsxfgu0dXjXyTrKQzxFUV2/UoxMJL1biq8sYR9NVrmQwfVysQjFnJZDidy5OPxnD9cCctCAKV4YCRYxNTFX63uUl1OECTJDJ6lPlkijOFwj4Xw7lkKswwNYZkdJ3r9Tqm4+AHAS+XSqxOJkNbIxPTcmiaJtlIhN9ubnChOMOZQpHr9TovFYtk9Qi/uH2Lya0Jd1pNgknItoKIJIn8txcvcipX4LcbD2iODO61O2QjOkvpDO8sr3GuUKScSDC0bf7Pm9cRhIBCNM6b84sokkhUUXF9j+/NzrHe6+D5Ppbn8dHuDgFweXae+eQXl+nyxHOmOx6jyfKRBrSeRBAEfFTZZb3TxvI8frS4NH2PdgcDbrWa4d1DrUpa159Zgvpd47iwv2A8z6O124EgIDebPVJ48qNIsvSlvg/CS0xJlpAVmbufrFO9XyeZjjMajjj/1tmpokXVlH0e6M9CIhPn1OU1RFlCPaNw6c8vTnvl86fKtHbbdGo9Fk7PHXky9SCeZu/7IvCDgLHroknSl06xX0ilSekR/MAnpemhtE5VaY/CrNBMJIImybhBwHa3iypLxBSNT6sV3lpc4la7SS4aJRsJJYkjx2Etm+Wny6uMXRdREPjlg/sU43EM26ZuGNOiBQKJiM5MIk5UVkLFjxO2YNK6hu16VIdDFFEkqipIQiR0nlRVREEkrmvMxhOcyRcZuw5b/R4N02DsunTGY84qCq+V56Y2Ce9ub9G3xhRjMe6127y9vMJbi0vUDYPSZJz/dKFA37L4zcY6hVicvmVRGfYRCAvwTDyOIkkoknRgQMeXYey63Gu3qAyHjF2X/3D3Nv/dq5fRZBnH9xAFEU2SCQIL1//j9yA8LuwvmDsfrbN9exeA0kqB8z88+7U9991PH7B5YxtZkXjlnfMMOwaxVBRVV/D9gPxs9oXo8wVB4MSrq6y8vPTY40XiEb73V6/S2GzieT722P5SPjabN7e5d3WDWDLChbfOPjbh6rke1shGi6rP9DN5k6g0URAwHYePq7t0RyOSmsYb84tPtNl9Eg/vOMMJzU0G9pgggB/ML1BOJLhQmMH3A1RJwnRtVEVCV2QML9yJ3241udtpTVs6P11ZI6Xr+EFAPhqdSBPhdP6LO5dXS2VaxpDGxK8lrUfIaBqe7/HO8iofV3bxg4BXy2VqhkHDMHh9bo6hbZPQVMaui+GGUXmaJDOwbEzHQZdkcpEor88t7Ot7x9XQUsB0HHRZpj8eca/TIQCSuj59b+OqGraAhkMAFFHg/f4WAgLleILvTxRGru/TGY/QJOkxW2EIF4KaERbrUjx+aGiHKknYnodh20QUGVmUGNg2miwzl0iy3QsXrPlkkoz+7WgNfpUcF/YXTGOrSbaURhAFGtvtp15CviiskcXmje2p3PDB9S1WLyxy7TefM+warFxYQFZkgiBg2DWQFem57ADg8CGu3btVbn1wF1ES2b4d4/JfXHym4jsahtml6WKSYdtg8+bOvmhCe2xz5R8+w+ybJDJxLr5z7tCBqIdpGEM+2N3B9sILwcHYpj0K5XcNw6Bpmswln5589TRMx8GwLWZiiVAOaBjMxBOcyIWugjv9PgPbYiUVpgbNxpMUYjE2el1iikZMUWiYBq7vTYMmXp9boD0yUSVpXxshE4nwX55/mb898xIb3S6ftxqTyLoBn9RqlBJJXN9HFgX+Zm4BTQ6//1arya1mA0kQOT/xafGCgJV0Gn0okdB04qqK/4jlyMWZMkPLZmjbvFyc4ZNaFUUSyUSi3GzUWZn0r0VB4GQ2T2dsUo4n+aiySz4SQxZFasZw+rjv72zTNAwQ4LXyHCldRxHF6QK73u1wtVpFFATWNY0fLy0faK8giSI/WVrh7+7dRpvcQ8QnnwlNlnlraRnH98NL8D8B1c1xYX/BlFeLbNzYBmDu5NGHhp6XsP0iMRqMGRtj8vNZ8nM53vxnl/G90Pc9CAI+/+Auu3eriKLAhbfOkp97cuqN67h8/t4dWrtt5k6WWXvl6TrjdrVLPB0jEo/Qrnawxw6R2NELuzCxpfVdH8/3kOT9/5A7tR7DrkF+Nktjp0W72j1SHuzVWo2oojC0bT5vNnh9do477dZ0V/mo50sQBPQmARJ7LZajEFEUdFnmbqvFyHM48ZBxlygIGI7NbzY3kAWBt5eWWZs4R84lkry3s83QslhKp9GkL/55yqL4xFaRJsukIzpeENAZjSZFTEAWBYJAmCYf7XE6l6cci/NZvcYv7txGFoVwYlUQGDp26B0D/N3d2yyn05wvlhAFgZSu85cnT+H5Plv9Hhu9LmM3dI38/kM6/5uNOndaLRDA8wMWUylutZoQBCyk0uH7YNs0TYNiPM7Qsvjlg/skVA1VEnljfpFMJELDMEho2iRRaojleUQPac8tZzL885fOM7TDkI2Hd/eCIHyp4I7vKseF/QWzenGZbDlDEISSxK+LUCVzngfXt8jPZ1m5EF76PdwGscc2lXs1cuUMYyMcKHpaYa9vNqltNMiWMmzc3CY3lyNTfPLPNbNc4MbvbmH0R2RLabQnWBC0Kh0G7SHZcnrqfaNHNc7+4BT3P90kV86yeHZ/zJuqKxAEtHbb3P7oHpZp8eqfXWB29clhD6okYTj2ZFReIKKoLKfTJFSVtVzusUGhW60mt5tNGqZBTFV5tTzLyWzuqVFtqiSxks6y1esTVVTWux0WUmlUSaI7HvEP6/cZWBYBAb/efEA5mWJgjfmkVsGwLNZ7Xda7HTqjET9cXDpyeygXiXIym+NXG+skFY1T+Rzt0RhJFLg088V76Hge2/0eO4M+d9otmqbBwLKxPIefrqyF/XdRZOy6FKIx1rtd5hJfjP6LgsDQdXjQ6XC+UKBumPRti3OFIiPH4Upll19tPOBMPs9MLMFWv88/OXmKXDRKEART7b4my8RUlaZhMHDs8G4hk2VgW6x3O2QiEeaSKT7c3QmLdSw6Hc6602pyp90iF4nySqn8RcpUNPaYAudPkePC/oIRRZFs6ZuRUyVziX3TnkEQYA5GSLKEHtWQVZlIXKfX7OPaLvn5I+SoTmpYONwucJRNa2mpSDQRwbFc0oXkoRegnVqXK/9wDUWVeXB9k9f/6hLRRGT6GIdNuqaLKc68cZL3/t3HLJ2Zp7Rc5Nb7dyku5J/o9vhqOZQvxhSFtXQW03N4Z2X1oYvI/dzttNEVmdbIpDUy0CSJuKIeqV0zsC3mU0mSmk7DGDJ2HVRJwvMDvMBHFUV8AlzfD/vr7RaqKHGlusvdTpuXiyWuVCucyOZYTB8+owBhUMYHO9v0LAvDsTmdyxORFWqGwV+snUSVpH2njau1KtuDHi3DZLPfJSIrxBQFx3dpmSZRVSUbiWA69tQLRxK/+P77nTbXajVapsF6t40giKR1DQGBO60W3fGYcjzBlUqFCzM+q5ksgiA8VnBlUeTN+UUqgwGCEO7y+5bFyP0iJ3Y+mSSuKtieRzYSRSBMhbreqJOPxqgZBhu9LqcO8Mv/U+a4sP8Rc/fKOluf7yBIYth2mc1y8e1z7N6roukqsyeeHmc2s1igU+3S3O2wdG7hQP+XgziK8+SwayDL0jS0YzQcTwv7kxAEgbm1MidfXaVT7eI6HuIjxesgjipf3KMQiXK33WJg26xmskiiiOUdzZ9/NpFgq99j7Ibj9lElLFSZSITL5Tl+vbGOiMgbC0vEVZWUqnOjXqc+sdFtmgaiIDzxZ+qMRmz2etSHA0aeSz4a5d52G0kQIRIu7LebTaLqF2EWtudxr9MipUc4mcvTsywCfKKywsulEnOJJLlolJlYnJuNBk3T4EJxZl9f/06rRTYSIa6q3GjUebU8S2xyMpFFkQBYzWYRBbhUnmPpCQtTRFFYnbSqMpEI9zsdYqrK2kOL7d5z+0HAlWqFO60m9zsdvj/3fBJJx/OoDgcICJQSiT+q4ajjwv4dp98ecPfKAxzLZv70LHNroQ+IbTls3dohu9d2ublNfjZLNBHhxCsrR358SZY48eoKQQD9Zp9Be/jCpkazpTT3RYFWpUMkrpPIPNsR+sSlFW69fxdrZHHhrTNfWiJ6GK+WZ8noESKygiAIJDWN2cTRfvaZeIKfLK0wdl0ykQiyKIYpSqMRZ/IFLs/OIQoCkijiBwFnCwWqxoDacMDAsrnfbbOQTPGg22EmHn+sPzx2HX6/vYksiNybKGnSus58Kkk5niAgwHAcNvs9HN9j5DqcyRf4w9Ym3dGY6/U6J7I5frK8zGIqTW04RJVkltLp6XMtpFJh3B9hUd1rQWWjUSqDPkPbxnQd1rtdgsBnJZPhZC7PjUadj6u7nCvOMJ9MHrlgpvUIl8qPL+x+EHCn3eJBp83ucMCZXIH+2OJup8WZXIGlJ1gHP4kr1Qq7gzCeeXmUmXroHJWvSxjxZTgu7N9hapsNfvtv3mP7doXySpFhxySZTZDIxJFkET2m028NcGyXzDN+aB/mzsfrNLZa9NsDrv7qBm//V2+ydG7huT/UsVSM1//6EuPhmFg6hqo9m0d2JKbzyjtfTot/FFRJ4lQ+z8lcDstzUSX5qf31h+lbY261miRVjdP5An/Y3sKaKHJ+uLBEStP4cHuLmjFkPpniR4vh7v1Ws8FCKsX35uapGwZN02A2sf+kZLkenu+Ti0VZSKYYTQKtvzc7Rz4ao2+NaRhbZOMxRq5LZzzCsG261pgLMyVy0RhB4HO72eKX6/fRZYWFVIrWyOSN+QXGrsNvtzYwbQfLcwmCgLVseB9zcaZEWtPY7fd5Y24htHR2HLKRKC0zjOB7rTTLerfNv/n8Ou8srVKaLIiW69KcqHvykeiRPkPb/R43G3UkUeBBt0MhGg+9dwjoWmNutRpcKJae6fMYGrWFl+ab3S5bvR5n84Uj3Wf4QcBn9Sob3R7FWIxL5dkDPfK/SY4L+3eYyr0aiqaQyicY9k2S+TCUAvbMts6xfWsXNaI8VzCGPbZxHZfGdpPAC7j14T0SucSR3CSfRiSmH9lyIAgCuvUeruORmUk9c3rSl0UQhEP104cxchw+qVZI6To1Y4jjh/7nxVic9sikaRqMHYfKcDD1XZ9PJnlzYYnVdJZ/d/c2N5oNFFE4sGjEVZViLMZWrwvAz1bXkESRf1y/z612k+V0mpHjEgSgSCKXZuemap093/abjTqaLLM76FNKJMjoJZqmSTDxUL/fbuH4AYZjk1L1aWFXJYmTuTwE8IfdbaKywnwqiSgI/HZzg+1+j/ud1sSHJ8IHu9v82eoJNEnive0tGqaJ4Vi8MlPm5SdsOPZ2xGPXRRZF8tEYy6kMggAxRSUgDBR/0OmylMqQegZ9uiAIFGNxfnH7c6KKQjqic7fdmiZPPYnWyOR+p8tMLEbVGFAdDlj4kqeGr4rjwv4dJlVMoqzXcB0P3/OYO1na1yaJJaP79N9flrWLy1Tu1RgNxpx5/QSiKD7RTfKrYvv2Lrc+uIsgimRLaV555/y38ii83u1wvVbjfqfDq+XydOrR8wM+q9eIKqF3yl6/3p0MTe39LIIArueFvuuKgio+XtglUWQtm2Oz30MSRG4063xWq3GtXsOcKH/OFYpcKs+S1sNWUMs0KMXikwtxge1+L7wSFwSGVqjpP5nNIwgCUUXF9jz641BV4wb7rSFsz+N2u8XpXJ6WaVKKhZ+7iKJyoTjDrzceUIjFWEql6FhjXN8LF+bxiJ41pjIYUDcMEro+1b7v4XgeH1d2aZgGS6k0K5lsOBFrDDlXnOG18izrnQ6fNWqMXBdB4Ev1x8/mC9xpNclGI/he8Jhm/zCEcIYWL/Dx/b3ff7s4LuzfYZbOzqNHNc68fpLcXJbkV5TNmson+dm/+AlXf3WDYWdIqpB6opvkV0V9q0Uim0CParSqHRzLeaKt8IvgWfuofcviaq1KRtfRZZnbzSYncjkW0yk2+92JQiUcOpqJxVnLZqkMhpzJF8hHQjmh4TjMJsKhpboxfOzC1g8Cbjbq/OrBAyRR4EJxhu1Bn6oxZD6VYrPbZbvX42KpTErT+biyw0avR2044HSugCyJvDJTZiGVYWBZyILIT1fXmEskSU92vZ2RSdM02e73KUSjPCqH8nwfL/ApxmKYtsONye5fEgQ0Seany6sokkTXGrOSTpOYWPVGFZWN6i4JVWMhmWKr13ussO8OBlSGA4qxOPc6bUqJBG8vrYRJTLKMIAgsZzKMPIfOaMyl2bkvFdCR0DRen5vnVqtJXFOnJ5KnkYtEODOxel7NZCgf8d7l6+S4sH+HEUWR8srTj44vAk1XufznF6dukt/ETrkwl+X2R/cxugapQjKM7PuK6Ftj3t/ZxvF8zuTzZCNR4qr61FDpIAgQApBFiblEkrVsLrwYHQ6QRZFTuTwNw6BvjclEIlwolrjwiKpzJhbnrtyiboSKGlWSGTnhEBBA0zS4025RiMX4tFbBD3wahoHn+7QNk6gs89rcHG8tLtMbj+hZFrosYzouludyp9Mnrqq8NjtLZzSiGIvvM+gC+HB3h4yuk45E8QOfyCOtqIiicCqX59NqhfvdFq+UZtnsdTmdK5CNREhoGtpkzH/vdUuCwA8Xlxg6FkM7lH8+PDS1hzT5bPmBjwCIhJfMkYfee1kUuVDcr+oKgoD1boemaTKfTD52L3EQJ3N51o4wm/AwgiBwJl94YtbtN81xYT/myAiC8JXvkJ/Ewpk5YukYnuOSKaW/UoOwa7Uqe15R//u1TzlfmKEYj/P9uflDi7vn+0RkmbVslvudNulIhJVMeLJJaRNbX8NAFAUykcPN0WKqOlXU1A2DX2+sIxCO8y+m03uhVpQTCZqmwefNOtlIaCKWzkT487WTU639yHHwfB9dlnF9jyvVCrlI6L+e1iJcKs8CoU1wZzQirqnEFBVRECjE4lSGQ0RgLhHnw90d+rbF2VyBciLBmXyBjB5BkyRykTBRSRLDCVc/CLhaq/B5o8lMPMaPFpdRJImIovBXJ06HxmSSSCn++G63lEiwPMpQGw45nQ8Xij2CIKA5MnE9n3w0uu/+oTIccLVaJa6pfLg74CdL6pH67s9S1L8rHBf2Y74zCILwQi5sH2bPZMrxPIqx+FQVIQji9M8EoBiLhROatnWg5Wt7ZPLezjau53N+Zoa/Pnl63wIQURR+vLRM37KIq+pTWweaLKNKEr/Z3CAXieIFATebDRbTaQqxGMvpDNv9ftjDHo1I6To9a4zsWDRNk6iikIlEKMZivDxTomYM+ednz3O9UWc+lcR2QwkkhMX/t5sPcHyfnUGfpKajSRKFaIx8NMb5mRkM26I6DHf6H1V2+Jm+hq4oFGIxFtMZtvthyMdCMpxKbpoGv1y/jxcE3GjWSag6l+fmpj/bQdr2seswclwSmnao9HC92+FqrQpAIRbjB/OL08I8dl0kUSCuahiOg+19NyyjvwqOC/sxf9Lcbbe4Xq8jCALZaIQfLiwhCmHf+kp1F1WSmE+m6NsWkiju8295mOuNOqookVA1rtdr0wL3MBFFmbYljoIgCCRVld54jE+wb6T/lVKZizMl7nc6bPZ79MZjuuMx+UiU6nDAVr/HXDxBY2Qyl0jwxqQA6orMnVYTWZI4mc2x2e3yoNuhMx4zn0xypVIhH4mS1HTSus7rc6EVwceVXSRBxJ9cAFuuy6vlWVYzWV4tlTlXKKJI0rTIOp5HZzxiPplm7LpUBn3gcGVWdzzi91ubuL5PLhLl+/MLB16I7g7607CSMJ3KnSqWyvEE690OddMgH41OrYb/FDku7Md8pwiCUL3wtF73UakZQ1L6F4XC8Tw0WSahafx4aQU/CNjsdelbFgvJ1KGFWZNkBmOLgLC//ryHe8/3Gbkur5RnedDpIIhw8pHLvd3hgM8aNTRJppRIcrk8ixMEpHSdO+0mA2vMfDLF7Vbo916Kh+2TlXSYnlQzDD6q7qIIIg+6HYQABCGYJj5FHpKTnszm6IxHXG80KEwe63q9RjmemCRCydSGA3YGA3KRCKVYnNVMlt3BgKSqPfVicqffRyCUINaNIQPLwnJderZFKRaftlRmE0k+rVbp22FAt/rQQhtRlGkLK6oof5QtlqNyXNiP+cZo17psXN8imoiwenHpqda7Q9vm/Z1tTNvmdD4faqmfk9lEkqu1Kn3LohiPPTbhKQrCkaLULhRnuFavYXsu5wqzz7XwOJ7HeztbtMdj4orKmwsLB+rod/p9YorCTKFI3TQ4U5jhaq1C3RiSj0YZOaH+WxCFfVK+vXbT0LYQAlBkieV0OHn55uISO4MeMUXlVDZ8f13f50G3E6YuJdMgghf4E6lk+Jh9y+K9nW00SebXG2E49lomx+VyqFhZfIrOO6npWG6LzmiEJIr0xmOu1CqoosS9dou3l1eJKgor6QwJVcP2PQrR2L7i7fqhNr81GnEim2X+gFPTV4XjedzrtHE8j+V0hsRzJkI9L8eF/ZhvBNtyuPbrG6i6SqfaRZJFTry6+sTvudtuYbku2WiEm80Gs4nkl5K5PcxqJktK08MWQPRok5AHEVGUadvieWmPRrRGI2ZicerDIfWhcaARWD4aZWfQx3RC06xsJMLl2XnkSZ/5aq3K7mBAOZZgMB7zi8ouCVXl5ZkymUiEjB7h81aTkeOQ1nXKE7njo/3v3UGfe5022UiEjj8irUQYux6XynPTBcdyQ0mmT0BnkofaGpmsZjLMHaHA7qly+taYuWSSmmGgihK5aJS6YWA6DlElVGMdlrq00e1yr9Mmret8XNklpelfeYENgoC77Ra/39rCcl3mk0lqxpB3lldf2Knyy3Bc2I/5RvBcD8/10WManusxMqynfo8iiniBj+2FMriHHQcfZU/61hqNWEgkpyPtj+J4Hn1rjB9AWtdfuBGU43lcrVWnu8jDnCQfRpUkgiDcUbtBgCbL9C2LT6sVvMDn4qQwr6QzRBWFseuSUDV+t7WB6bpkdJ035hZ4bXaOV4OAT6oVfnHnNgPLopRIMHI9fr66hh8EnMxmSesRBrbF0LamOvaH2bM5lgQRAYFL5dlpvx9CXX1CVclEImx0u8iiSFLXGNjWkRdKQRBYSKWAcBEQELjXboVDTKp6pEzUseswsC0UUcIPgqnPzfMSBAE7gz4Dy2IumZq+Fj8IuNGo88HuNgPLCtt1qSSm4+D6/nFhP+ZPj0hMZ/GleTZvbKHqKsvnnm4hfCKbw3RdBuNwKOVJY/570reYqrI76PPO8sqB0WvX6jW2+l0EBGrGgDcXlp7r53qUzV6PrX6fbAM8sj0AABpVSURBVCTCtVqNfDS673VYrsuNZgPbdTmdz4fRdpEIl2fn2B30OZHNUYzF+O3mBqbjIIsiH+xs8/O1EwiCMJULrnc7jBxnkp40pDMO9eniJEg7pqqoksTQtsMpUCCmKtN+ejhtevD7OZdIcrfd4h831kmqGtXhYFrYu+MR729vY3se5wpFLhZLrHc6VI0ha5ksM18yRzal67y9vIrpOCQ1jbZpcrVWRVcULpVniR9wUutZFlu9HndaTS7OhMNZL4KNXpdPqhUUUeJBr8tPl1fRZJnr9Rof7u5wv9OmFIvjeB51w+S12dkvHbH4ojgu7Md8Y5y4uMzi6dkjh3drssz3Zo/meTOe9JcTmsbIdbAOkb41TZOsHkUSRVoTn5Qv045xfZ/bzSZDx2Itk5sWPi/wkSYj7wI8NrZ+vV5jdzBAlWX+sL3NzyeeL/PJ5L6hIcf3UCQJWRSwPBc/CDBsG02S0GSZqKzg+gF9ywKBaSAFwHIqTXc04nariSZLZPQwTSup6fxwcSl8DyJRsodo6zVZJh+N8WqpTFoPA7eX0xliE9teURRIqzofVnYoxRMoksibC4vP3QaJKgpRRaE7HvNvb93E9jxUWUIWBN5c3L8AB0FAe2SGE6q+x9j1XtgQXXc8JiIrpHR9OgmsyTLb/T4rmQyu77E7GPCTpRVeLZcPlMN+3RwX9mO+Ub6qgadSYiJ9M4bkolEyh/xjW01nuN6oAbCWzSFMItv2AiceTVU6jLvtFnc6TaKyyh92tvjZylqo106laRgGLdPkRDb32C7SnARwDO0xDcNgaC+Q0vXp8b9ljphNxHmlVOb9nW1GbsArM2WuVCvsDPooosgP5hcnLoNlWqMRc4nEvlPBS4UixXicQixGbzymYRpcrVW5VJ5FIAzAOKgF8zCaJOH6AWPXRRSEactKFiVsb4woCDzodYmrKqIg8lFlh7eXn3xnchTWux3+sLXJ77c2WJho9humyVo2y8xDw02CIDCbSLI5MUU7cUR7gKOwkEyx3e9RNwxy0Sixibf+fCrJvXabXDTGyzNlXp+b/9Z4F72Qwi4Iwl8C/xMgAf9LEAT/44t43GOO+bJEJwNBluc9Ufp2IpcjH4viBwEZPYLluvx2c2O6K35jbuHQ/nwYbwcJVcV0HHRJIa6qNEwDx/fRCHe7P1xc2udn/jBn80X+rxufsT3os5hM83Fll58sr9A0DT7Y3SEqK2z2OvxkeYW/WDtJABi2zUeVHWZicXrjMevdDpfKsyylMywdoOARBAFdkrnXaVOIRMnqEarGkNvtFp83GkDAqVyes4WDE6sgLJS27zO0Lb5XmJu2Gs4XZ7haq2K5LkvJ9PQye/SCTOKuVcPHzkej3G21iasqpViMD3Z2+PnaiX0tj4szJWYTiXCxOuSC9Wn4QYD9iEVzLhrlneVVLM8lqenT3vm5wgzFaJyAgEI09q0p6vACCrsgCBLwPwM/B7aBDwRB+H+DILjxvI99zHcT3/fp1HrhEb2Y+sY+8IokHckn++Gjs+HY2H44hdoZjWiNzAML+3QCMoDT+TxrmSx1Y0jDNFhOp4k90q8+qKgHQRAuCLLMpfIsUVnhfrfNObPI0LaRBXF6/B+7YVERCC9XlYkkcOQ6JJ6iDHJ9n3e3txCBT2s1Sokh7yytcb/dmraM7nU608Juex5j1yWmKNMipskyrx4wDRpVFN6YRCzu9vv8ZvMBAfDjxeUnvqajsJfNWjMMRq5HUlc5nSuyms5iOPYkrvELJPFgi4KjYrku721v0bPG5KIxXp+bn55MYqpKjP3vs/hIQPi3iRexY38duBsEwX0AQRD+D+CfAceF/U+UWx/eY/dOhQBYPr/IiYvL3/RLOjJxVUOXZT6p7tAdW4fu/O60WmT0CLIocHdiX/tnK2u4E1+Woyxmm70uVyoVEAQ+2NlBIGAmkeC97S2+NzePJks0DIOUpu1bfDRZ5gcLi6x3OiRUjZXMk3X2jucxdh0uFEvMxBJIokA2EqFuDLlS3aFljlhMpbE9D8fz+N3WJmPPJa1qvDK5qHzasI/jeby7vcm1eo2EqrKdTjN7hGzYJzF2XUqJBLloFMNOMZtIIokCpuNwrjjzzB75T6M2HNK1xtML6JZpfmsL99N4EYV9Dth66PfbwPcf/SJBEP4l8C8BFhePnjt5zHeLIAiortfJzWbxXI/q/do3Vtj3+qLleJzyEZz+INwNr6Wz7PT7vDyTYaffZzWTeexiMROJUBn0ERFIRXQEQTjyCWGPnmURURRKiQRj20GWRM4UijRNAz8IeHt5ldFEo/6oDDOtR3ipoPD+7jYfVnY5k8txsVQ+cEHRZXlqkRsIYHke7+9u0xuP8fyAC4UZAkGgMhjg+B6W55LWNH67ucHW5ILw4d3rXlRdbTBgPpViZeIV80m1wshxqBtDupbFqXzhuS4So4rCTCxOc2QSURReKZWZicdpThQy690Or0385l8Eqizh++EpKiBAkb67GagvorAftJQ/5lgfBMG/Av4VwOXLl4/maH/Mdw5BECjMZaltNAgCWDg9+428jsakRx1TVDZ7XX6ytHJk7xBxsqPN6OGu1vMf/7henCmRUNVwGvMpO+bDmE+Gl311w2A1l8NyHVqmiSJJpDQdVZIem4R9mPudDr/d2ECRRG43G+RjsQOnLYWJt8xqJsvQtviosksxFsdyXFRRohiPh86MgoCiKHiez2avh+V6lBNxGqZBZzSanl7qxpAbjToZXedqrUpS0wmAiKxQHQ4RBEiqGi1z9FxFVxJFXp+bpzseo8kSMUXlaq3Kf7p3l1I8QSke56PKLn+2svaln+NhZmJxzhdnqBtD1krlQ1VC3wVeRGHfBh4WIc8Duy/gcY/5jnL2jVMUFvIIokBu9sW6MR6VseMgEQZQjxyHsXv0y7xyPMFWv0fdNJhLpvbZxu6hStI+P25vMsQD4Pg+iig+tR2TjUT56coaY9chqenYnsfAskhoGlFFYeQ4VI0hmiRRjicee7yBZTFyHfLRFAPLomkYxBSVm806qiRzvlictisEQSCmqrRHJoZtU/UHuBNbgN9ubvBauUw5kQgNxsplbrcauJ4fWgRP4vUgHALqjy0EwrjAIBjTMgyUyfvRHptk9Ainsrl9Sht/kp601euT0FSWJ341e3THI67WqsiixIXizFQqqUjSdEHZ6nVZ73aQRZHtQY+kpqLLCr3x+MBTzbMiCAIncjlO5F6couab4kUU9g+Ak4IgrAA7wH8N/Dcv4HGP+Y4iyRIzS99sCEEhFiOmqjSMISldO7A4H4Ymy/xoYenIBfp+p82Neh1ZElElmaEd9uYvl+ee2prZ02pDqHXf+/XehefQtvD9gDOFPEEAQQArmQwRReFELkt2O0rDMCglEszE4vxhZwtFFOmMRgBcfkj3/2m1wu1Wk9vtJlFZZjWTZy2bJaaEDpLexFxtOZ1hKZXmXrtNc2RyetJS2Rn0+Xh3B8f3cTyf2nDAwLb4+wf3qQ0HLKXSXCyWiakq5woFAqA6HLDb77PeDV0oT2fzWJ67z4MnCALe39lGFERGjsuntQo/OuDy1QsCQOBsPs+Hu7uYtoMX+PxqY52kpvHmwtITTzh/Sjx3YQ+CwBUE4b8H/iOh3PF/DYLg+nO/smOOeQ50OZQ7jiZOf8+6mxME4UhFYuw6XK/XyEaibPZ6VIZ93lxYojoc0DCNI6X4HIRh2zSMIXPJJGPX4w9b2+SiUUQBWmOTtxaXyUai/IuLr1AbDknrOvloDNfzSKoaIEz9WyAsnpXhkL5lkdZ0LM+jOhwwm0hguR6uv3+Aa7p75Yvd6+f1OglNR5VEmqbJD+cX+d32Jn3LIheJUjWG1IZDXioU+bu7XUrJJPYkLOTlmRk6oxG276PLCkN7v4WE4/skVAVXEA71UZ9NJKkMBzQMg789c4a4qvFZvTZ1hOyMRt/Zy84XzQvRsQdB8O+Bf/8iHuuYY14Uz3qZ+WUQJv85vocf+IiCOC2oX7Y1YHseH1V22er3uNNucTKbw/N9BrZFRo/QG1vTCdlHJ0bPFWe4Xq8hSxKXCmWCyaSrIAgsJJNcq1Xwg4CFZApVllnvdDBdh6V0hrHrHriYeb7Pp7Uq1xp1RAHWMjniqkZSDwM5JKBqGgxtC8tzsVyXzxp1lrMZMrrOnXYLEYGEptEZmZTiCRaSXxiNCYLAq6Uyn1QrCMD3DjFTUyWJH8wvTmcC6sYQPwjojccA+2yG/9QRguDrv8e8fPly8OGHH37tz3vMMV8F1cGAG60GMUkmqqo0RyaziSSnJpOsz0ptOOS9nS1ykSgPuh3Wsllu1Ots9/u4gc/fnjrD5Sc4SdqehyQItEcj3t/dZrffDy8GZ2bwfJ9PahUisspSKsV6t0MxFqdhGFycKR3oIlkdDvjD9hZZPcLVeo0z+QLfm53jVqvJRreD6/ssJJO0TJPrjQa6ItMyTUqxOBFFJa4pZPQIxVic5XSaqKIe6KXi+T6CIBzZRz2czh3QHoXPVfwT2K0LgvBREASXn/Z1x0vcMcc8A3uj/k3TZDaeoBiPh/3teJy6YeB4Hqdy+ecygdrbeQ5tm6iqklA18rEYq9lwCOppU5V7u+4r1V1Gtk3TNOjbFgEBP15a4b84ex4Aw3HY6HVpTOSVh3m7CAgEhGqhuUSC88XiNBVpIZWmOhywks1xYaaMLMnhZbCqIUlh4tKpbJ5LR/D4eVY3REEQHvPUOSbkuLAfc8wz0Ng36t/l7Ylr5L12m2v1KoIgknsoYu/LkNRC293NXg/HGHKr3aRjjiGAhKrva70EQYDhOMhi2Jv+tBrmgV4slZBEkbHn4hNaCgQIuIE/PUXEVZXvzc7Rs8YUovFD5aCFWIyT2Rw7/T4nc3mKsVAe6Qdgey6e77Pb75ONRHlrcQk3CPjN5gOK0RiOH0bkHfP1clzYjznmGTAdB1kQSOk6temo/17EXoToI1mcru+HMXOy/Ew70mI8zsC22Rn0KcXiBD68VCwyl0jui+d7d2uTe502SU3DDwLiarjr/nB3l8vlOd73fbpji2wkwnwySfYhXfnnjQa/23qAH8DPVtcOLeyiIHC+OMP54sz0/+UjUV4qFtnp91FEie1+j8pwQFrXeWtxmaVUOozbQ+DcEzxojvlqOC7sxxzzDBRjcTRZoWEYpB8a9Z9NJvm0WmFgWxSiMTRJZuQ4/H57E8NxyOkRXp+bf6bLXEEIJ/38ILyIzUWi+4r6eqfNf7x3h4Sq0xmN8fF5pRQOhNmuS0rX+fnaSX62euIx6abn+/zywT0Gtk0QwL+9dZP/4ftvHvmUIQgCp3N5Tufy/MOD+4gIaJJEwzRwfZ+XZ0rMJ1OIAoc6ax7z1XFc2I855hmIKgpvL68wdh2iyhdDMSvpDElVw/E9cpEwYq8yHGDadug9YgzpjsfP5Dq4kEzRGo1oGkPOFAqPWes2TCMM0JBFmqbBj+YXGTk2AK+Vv+hpHyTdFAUBx/MREJClsIe+F+T9rKykM1ytVkGApVR6ungd1fL4u4zn+9xpt+iOx6yk0/ushL9Jjgv7Mcc8I4eN+uceKWQRWcb1ffpWKE981uEZRZIODRbxfJ+4olKOJ+iOx6xlsry1vDItzA/vvG3Po2WaqJI0fY2CIPDnayf4T/fuIggCr8/NP1NRrw+HPOh1SOkRTmSyZPUIru8f2bbhj4XNXpfPmw3iqsYHuzu8s7z63Dm8L4Ljwn7MMV8RpXiCQjTGR5UdSokEj3Y5nD173CO4Jz6M6/u8t7NF0zDQZJmfra6xnM7sa9Ps4U2+tm2GF5ivlspTSeOZQpHZZArP90k9JWjjYUzH4b2dLSKKwu5ggCKKR8pyfR72dPyqJB8a4fdNMHZdFFEirqoYtoXjHzxc9XVzXNiPOeYFMrRthrZFahLI0BmPeG12HtOxudlo8P2Jd7lh26E9ruuQf8T7+2l0x+OJpWyCphn6tBxU1Pee51qtiiAIJFWNynCwT6t+lJDoR7E9l4DQ4tj2PEaO88yP8Sz4QcBHlV2qwwGiIPCD+cXHTkffFAupNNuDPnVjOAm6fjE5q8/LcWE/5pgXRN8a8+uNDXx8VFHizflFJFHE8Twcb39q/c6gj+25+/rvR+1Ja5OWTt+ycD2fyBN8ydujEb2xBYLA7qDP+ZmZQ7/2qCQ1ndlEkt1BH12WWUw9PtT0IjFsm+pwQDEWp2+N2eh1j1TYPd+naZqIgkA+Gv1KAl/iqso7y6vYnkfkiD78XwfHhf2YY14QndGIIAhC7xJziOk6XC7P8VmzTi4a4aWHZH9RRcHxfQaWhcgXxfooJDSNN+YW2Or3OJHNMvtQwtOjYdz+JPZOkUS64zGz8ecf5hEFgdfKs7xUKKKI4qFKHz8IEOC5i50my6iSRGc0Yuy5rKpHO2V8WqtOM1BPPyX673mQRfG5nSVfNMeF/ZhjXhBJTccPAhqGgSiIxFWNuKry0wNG3WcTSS7OeLRHIxaSM4dOfR5GMb5/hN7xPD6u7P7/7d3Nb1xnFcfx75kZ25lxPPHEsR137DhJCYg2JC1YIbTiRRRQgahs2IBASCzYgNRKVLz1T0ACFiChiA0SlRASICQkRIvEttC0TUElpSRRSOLExK8Zx+PMi+ewmImdgF/izLWf8Z3fZ+XrGc09emQdP3PuvedwY2GB0d49HB3YT8KMfE+W8fkCs4uLHB0YjKyEYWbr1rrPT0/z9tQk3Z2dnMgPN3VBsTOZ5ImRUS7fnGV3Z9d9fUOouTPeaKVQrdW4UihsWWJvRUrsIhHJpdN8ZPQghdJtcukMu9dJZgkzDuf2cjiidvUTt+a5fmuewe7dXJydZagnW7+fvtGCuFqrbXlDtDuKlQr/mLpBXzpDoVTiXzPTPLbKvNTNyHZ1cXRg/32//8480mvzBRw4snfn91jfDCV2kQjl0ukgt/wZhjsseW35ePm1xti+7ZJonLpaq7FUq5FMhKk7v3/oIfLZLAlad+j0VmmtwpCIPJChnh4O53IslCu8p28ffRH8c6nWalycneGd6SluV++988XdWSiXV+2dvivVwQceylOpLTHQ3c279+7b9LlrEXSdTSUS5Huyy5Oh2ol27CIxkEwkOL5/iOMRfua5yRtcmJ0hZQkm5uf58OhBzAx35+83Jrg0N0dHIsnJ4ZH/+5aS78mSf4AhI6VqldeujzNdXORQLsej/QMtc6fJTqIdu4isanKxSO+uNH2ZDHOl21Rr9TJPsVLh0twc/ZlukgnjwuzMhp9Vc+d2tbLhTvxqocBUcZG+TIYLM9PqDPmAtGMXkVU93Jvj7H8mwJ1DudxynT6VSNCRSDBfLlGsVBjJrv8kaGVpib9eu8pMsUhvOs0H8yNrtldIGNS8tvxPpN1KKFFRYheRVY325silMyzVavc0IOtKpTg5fIALszOMZDs40rd+DX2qWGSquMBgd8/yLNi1yjQje3qZK5WYWljg0f7B5e6ZsjlK7CKyprVaDuTSacbSG09FgnozM/d6CceBjsbDPIXS7Ua7gwRHBwbJdnWRSiR4vMlbI0U1dhHZYn3pNMcG99OVTHJsYJD+TL118avXxlmoVJkvlXjj+njgKONFO3YR2VK2/DDWvR0gy9Ulujs76iP2WqQrYlxoxy4iQTy2f4hb5TLFSpnjm3iqVDamHbuIBDHU08NA9xGATc2DlY0psYtIMEroW0OrKiISM0rsIiIxo8QuIhIzSuwiIjGjxC4iEjNK7CIiMdNUYjez75vZ22b2NzP7rZlt7bhyERHZULM79peBo+5+DHgH+G7zIYlIKOtNRpKdo6kHlNz9pbsOXwE+31w4IhKKu3N24jpXCjfpSCZ5YvgAe+5q1ys7R5Q19q8Cf1jrRTP7mpmdMbMzk5OTEZ5WRKJwq1zm8s2b9Ge6SWBcvI/JSNKaNtyxm9mfgNU69Lzg7r9rvOcFoAq8uNbnuPtp4DTA2NhY85NqRSRSnckkqWSC+XKZxWqFTGdn6JDkAW2Y2N39E+u9bmZfAU4BT7lHMFpcRILoSqX40PAI52dmOLBnDw//T5td2TmaqrGb2dPAt4GPunsxmpBEJJS96Qwn8pnQYUiTmq2x/xjoAV42s7Nm9tMIYhIRkSY0e1fMu6IKREREoqEnT0VEYkaJXUQkZpTYRURiRoldRCRmlNhFRGLGQjxTZGaTwL+34VT7gKltOM9OoLVYobVYobWo2ynrMOru/Ru9KUhi3y5mdsbdx0LH0Qq0Fiu0Fiu0FnVxWweVYkREYkaJXUQkZuKe2E+HDqCFaC1WaC1WaC3qYrUOsa6xi4i0o7jv2EVE2k7bJHYze97M3Mz2hY4llHYfPm5mT5vZP83svJl9J3Q8oZjZiJn92czOmdlbZvZs6JhCM7Okmb1hZr8PHUsU2iKxm9kI8EngcuhYAmvb4eNmlgR+AnwaeAT4gpk9EjaqYKrAN939vcBJ4OttvBZ3PAucCx1EVNoisQM/BL4FtPUFBXd/yd2rjcNXgOGQ8WyzE8B5d7/o7mXgl8DnAscUhLtfd/fXGz/PU09o+bBRhWNmw8BngZ+FjiUqsU/sZvYMMO7ub4aOpcWsO3w8hvLAlbuOr9LGyewOMzsIPA78JWwkQf2I+savFjqQqDQ1aKNVrDdwG/ge8KntjSicqIaPx5Ct8ru2/gZnZruBXwPPuXshdDwhmNkp4Ia7v2ZmHwsdT1RikdjXGrhtZu8DDgFvmhnUSw+vm9kJd5/YxhC3jYaPr+kqMHLX8TBwLVAswZlZB/Wk/qK7/yZ0PAE9CTxjZp8BdgFZM/uFu38pcFxNaav72M3sEjDm7juh2U/kGsPHf0B9+Phk6Hi2k5mlqF8wfgoYB14FvujubwUNLACr73J+Dsy4+3Oh42kVjR378+5+KnQszYp9jV3u0bbDxxsXjb8B/JH6xcJftWNSb3gS+DLw8cbfwdnGjlVioq127CIi7UA7dhGRmFFiFxGJGSV2EZGYUWIXEYkZJXYRkZhRYhcRiRkldhGRmFFiFxGJmf8CQtDIikf0sp4AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(X[::10000, 0], X[::10000, 1], marker='.', c=km.labels_[::10000],  cmap='viridis', alpha=0.25);\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Low Level Interfaces: Parallelizing Compute Graphs\n",
    "Building the Compute-Graph manually"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# simple example\n",
    "def inc(x):\n",
    "    return x + 1\n",
    "\n",
    "def double(x):\n",
    "    return x + 2\n",
    "\n",
    "def add(x, y):\n",
    "    return x + y\n",
    "\n",
    "data = [1, 2, 3, 4, 5]\n",
    "\n",
    "output = []\n",
    "for x in data:\n",
    "    a = inc(x)\n",
    "    b = double(x)\n",
    "    c = add(a, b)\n",
    "    output.append(c)\n",
    "\n",
    "total = sum(output)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "45"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### **Now, build a graph:**\n",
    "```dask.delayed()``` add functions and arguments to graph:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "import dask\n",
    "\n",
    "output = []\n",
    "for x in data:\n",
    "    a = dask.delayed(inc)(x)\n",
    "    b = dask.delayed(double)(x)\n",
    "    c = dask.delayed(add)(a, b)\n",
    "    output.append(c)\n",
    "\n",
    "total = dask.delayed(sum)(output)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/svg+xml": [
       "<svg height=\"464pt\" viewBox=\"0.00 0.00 741.45 463.69\" width=\"741pt\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\n",
       "<g class=\"graph\" id=\"graph0\" transform=\"scale(1 1) rotate(0) translate(4 459.6851)\">\n",
       "<title>%3</title>\n",
       "<polygon fill=\"#ffffff\" points=\"-4,4 -4,-459.6851 737.4469,-459.6851 737.4469,4 -4,4\" stroke=\"transparent\"/>\n",
       "<!-- 941160937453159389 -->\n",
       "<g class=\"node\" id=\"node1\">\n",
       "<title>941160937453159389</title>\n",
       "<polygon fill=\"none\" points=\"374.4469,-455.6851 320.4469,-455.6851 320.4469,-419.6851 374.4469,-419.6851 374.4469,-455.6851\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 1978824865916811505 -->\n",
       "<g class=\"node\" id=\"node2\">\n",
       "<title>1978824865916811505</title>\n",
       "<ellipse cx=\"347.4469\" cy=\"-357.6874\" fill=\"none\" rx=\"25.9954\" ry=\"25.9954\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"347.4469\" y=\"-353.9874\">sum</text>\n",
       "</g>\n",
       "<!-- 1978824865916811505&#45;&gt;941160937453159389 -->\n",
       "<g class=\"edge\" id=\"edge1\">\n",
       "<title>1978824865916811505-&gt;941160937453159389</title>\n",
       "<path d=\"M347.4469,-383.8817C347.4469,-392.0839 347.4469,-401.1688 347.4469,-409.502\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"343.947,-409.5324 347.4469,-419.5324 350.947,-409.5324 343.947,-409.5324\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 2484196008774290510 -->\n",
       "<g class=\"node\" id=\"node3\">\n",
       "<title>2484196008774290510</title>\n",
       "<polygon fill=\"none\" points=\"192.4469,-295.6897 138.4469,-295.6897 138.4469,-259.6897 192.4469,-259.6897 192.4469,-295.6897\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 2484196008774290510&#45;&gt;1978824865916811505 -->\n",
       "<g class=\"edge\" id=\"edge2\">\n",
       "<title>2484196008774290510-&gt;1978824865916811505</title>\n",
       "<path d=\"M192.6942,-289.6662C225.144,-303.9294 279.329,-327.7463 314.2106,-343.0785\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"312.94,-346.3431 323.5031,-347.163 315.7568,-339.9348 312.94,-346.3431\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 8600206936994836494 -->\n",
       "<g class=\"node\" id=\"node4\">\n",
       "<title>8600206936994836494</title>\n",
       "<polygon fill=\"none\" points=\"301.4469,-295.6897 247.4469,-295.6897 247.4469,-259.6897 301.4469,-259.6897 301.4469,-295.6897\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 8600206936994836494&#45;&gt;1978824865916811505 -->\n",
       "<g class=\"edge\" id=\"edge3\">\n",
       "<title>8600206936994836494-&gt;1978824865916811505</title>\n",
       "<path d=\"M291.0118,-295.8424C300.3768,-306.1052 312.2676,-319.1358 322.7171,-330.587\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"320.3463,-333.1813 329.6723,-338.2089 325.517,-328.4629 320.3463,-333.1813\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 3122473049151551376 -->\n",
       "<g class=\"node\" id=\"node5\">\n",
       "<title>3122473049151551376</title>\n",
       "<polygon fill=\"none\" points=\"374.4469,-295.6897 320.4469,-295.6897 320.4469,-259.6897 374.4469,-259.6897 374.4469,-295.6897\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 3122473049151551376&#45;&gt;1978824865916811505 -->\n",
       "<g class=\"edge\" id=\"edge4\">\n",
       "<title>3122473049151551376-&gt;1978824865916811505</title>\n",
       "<path d=\"M347.4469,-295.8424C347.4469,-303.3969 347.4469,-312.4511 347.4469,-321.2711\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"343.947,-321.4135 347.4469,-331.4136 350.947,-321.4136 343.947,-321.4135\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 5417255037287969848 -->\n",
       "<g class=\"node\" id=\"node6\">\n",
       "<title>5417255037287969848</title>\n",
       "<polygon fill=\"none\" points=\"468.4469,-295.6897 414.4469,-295.6897 414.4469,-259.6897 468.4469,-259.6897 468.4469,-295.6897\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 5417255037287969848&#45;&gt;1978824865916811505 -->\n",
       "<g class=\"edge\" id=\"edge5\">\n",
       "<title>5417255037287969848-&gt;1978824865916811505</title>\n",
       "<path d=\"M420.1169,-295.8424C406.7713,-307.2001 389.4425,-321.9476 375.0541,-334.1926\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"372.7542,-331.554 367.4071,-340.7006 377.291,-336.8849 372.7542,-331.554\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;2981091146645298244 -->\n",
       "<g class=\"node\" id=\"node7\">\n",
       "<title>-2981091146645298244</title>\n",
       "<polygon fill=\"none\" points=\"600.4469,-295.6897 546.4469,-295.6897 546.4469,-259.6897 600.4469,-259.6897 600.4469,-295.6897\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;2981091146645298244&#45;&gt;1978824865916811505 -->\n",
       "<g class=\"edge\" id=\"edge6\">\n",
       "<title>-2981091146645298244-&gt;1978824865916811505</title>\n",
       "<path d=\"M546.1861,-287.3393C504.9289,-301.9432 426.6802,-329.6411 381.6562,-345.5783\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"380.3001,-342.3454 372.0412,-348.9817 382.636,-348.9442 380.3001,-342.3454\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 7578463677693504372 -->\n",
       "<g class=\"node\" id=\"node8\">\n",
       "<title>7578463677693504372</title>\n",
       "<ellipse cx=\"633.4469\" cy=\"-200.2918\" fill=\"none\" rx=\"23.2963\" ry=\"23.2963\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"633.4469\" y=\"-196.5918\">add</text>\n",
       "</g>\n",
       "<!-- 7578463677693504372&#45;&gt;&#45;2981091146645298244 -->\n",
       "<g class=\"edge\" id=\"edge7\">\n",
       "<title>7578463677693504372-&gt;-2981091146645298244</title>\n",
       "<path d=\"M618.9226,-219.0277C611.3695,-228.7709 602.0142,-240.839 593.7563,-251.4913\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"590.785,-249.6116 587.4244,-259.6593 596.3174,-253.9004 590.785,-249.6116\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;7243694643610269344 -->\n",
       "<g class=\"node\" id=\"node9\">\n",
       "<title>-7243694643610269344</title>\n",
       "<polygon fill=\"none\" points=\"660.4469,-140.8939 606.4469,-140.8939 606.4469,-104.8939 660.4469,-104.8939 660.4469,-140.8939\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;7243694643610269344&#45;&gt;7578463677693504372 -->\n",
       "<g class=\"edge\" id=\"edge8\">\n",
       "<title>-7243694643610269344-&gt;7578463677693504372</title>\n",
       "<path d=\"M633.4469,-141.2361C633.4469,-148.883 633.4469,-158.0124 633.4469,-166.78\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"629.947,-166.7957 633.4469,-176.7958 636.947,-166.7958 629.947,-166.7957\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;1524552530009487147 -->\n",
       "<g class=\"node\" id=\"node10\">\n",
       "<title>-1524552530009487147</title>\n",
       "<polygon fill=\"none\" points=\"733.4469,-140.8939 679.4469,-140.8939 679.4469,-104.8939 733.4469,-104.8939 733.4469,-140.8939\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;1524552530009487147&#45;&gt;7578463677693504372 -->\n",
       "<g class=\"edge\" id=\"edge9\">\n",
       "<title>-1524552530009487147-&gt;7578463677693504372</title>\n",
       "<path d=\"M689.147,-141.2361C679.418,-151.5512 667.1449,-164.5637 656.5969,-175.7472\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"653.9378,-173.4655 649.6225,-183.1417 659.0301,-178.2684 653.9378,-173.4655\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 444331679719316009 -->\n",
       "<g class=\"node\" id=\"node11\">\n",
       "<title>444331679719316009</title>\n",
       "<ellipse cx=\"707.4469\" cy=\"-34.4469\" fill=\"none\" rx=\"21.3963\" ry=\"21.3963\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"707.4469\" y=\"-30.7469\">inc</text>\n",
       "</g>\n",
       "<!-- 444331679719316009&#45;&gt;&#45;1524552530009487147 -->\n",
       "<g class=\"edge\" id=\"edge10\">\n",
       "<title>444331679719316009-&gt;-1524552530009487147</title>\n",
       "<path d=\"M707.1998,-56.3102C707.0691,-67.8634 706.9073,-82.1795 706.7671,-94.5732\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"703.2643,-94.8131 706.6509,-104.852 710.2639,-94.8923 703.2643,-94.8131\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 5244261928555912704 -->\n",
       "<g class=\"node\" id=\"node12\">\n",
       "<title>5244261928555912704</title>\n",
       "<ellipse cx=\"633.4469\" cy=\"-34.4469\" fill=\"none\" rx=\"34.394\" ry=\"34.394\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"633.4469\" y=\"-30.7469\">double</text>\n",
       "</g>\n",
       "<!-- 5244261928555912704&#45;&gt;&#45;7243694643610269344 -->\n",
       "<g class=\"edge\" id=\"edge11\">\n",
       "<title>5244261928555912704-&gt;-7243694643610269344</title>\n",
       "<path d=\"M633.4469,-68.8961C633.4469,-77.395 633.4469,-86.401 633.4469,-94.5715\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"629.947,-94.7183 633.4469,-104.7184 636.947,-94.7184 629.947,-94.7183\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 4197206210790603160 -->\n",
       "<g class=\"node\" id=\"node13\">\n",
       "<title>4197206210790603160</title>\n",
       "<ellipse cx=\"485.4469\" cy=\"-200.2918\" fill=\"none\" rx=\"23.2963\" ry=\"23.2963\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"485.4469\" y=\"-196.5918\">add</text>\n",
       "</g>\n",
       "<!-- 4197206210790603160&#45;&gt;5417255037287969848 -->\n",
       "<g class=\"edge\" id=\"edge12\">\n",
       "<title>4197206210790603160-&gt;5417255037287969848</title>\n",
       "<path d=\"M473.6558,-221.033C468.4859,-230.1271 462.3495,-240.9212 456.8269,-250.6357\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"453.6711,-249.1049 451.7717,-259.5281 459.7565,-252.5644 453.6711,-249.1049\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;6996466992881930582 -->\n",
       "<g class=\"node\" id=\"node14\">\n",
       "<title>-6996466992881930582</title>\n",
       "<polygon fill=\"none\" points=\"512.4469,-140.8939 458.4469,-140.8939 458.4469,-104.8939 512.4469,-104.8939 512.4469,-140.8939\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;6996466992881930582&#45;&gt;4197206210790603160 -->\n",
       "<g class=\"edge\" id=\"edge13\">\n",
       "<title>-6996466992881930582-&gt;4197206210790603160</title>\n",
       "<path d=\"M485.4469,-141.2361C485.4469,-148.883 485.4469,-158.0124 485.4469,-166.78\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"481.947,-166.7957 485.4469,-176.7958 488.947,-166.7958 481.947,-166.7957\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 4832454587193447956 -->\n",
       "<g class=\"node\" id=\"node15\">\n",
       "<title>4832454587193447956</title>\n",
       "<polygon fill=\"none\" points=\"585.4469,-140.8939 531.4469,-140.8939 531.4469,-104.8939 585.4469,-104.8939 585.4469,-140.8939\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 4832454587193447956&#45;&gt;4197206210790603160 -->\n",
       "<g class=\"edge\" id=\"edge14\">\n",
       "<title>4832454587193447956-&gt;4197206210790603160</title>\n",
       "<path d=\"M541.147,-141.2361C531.418,-151.5512 519.1449,-164.5637 508.5969,-175.7472\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"505.9378,-173.4655 501.6225,-183.1417 511.0301,-178.2684 505.9378,-173.4655\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;2370443148372306804 -->\n",
       "<g class=\"node\" id=\"node16\">\n",
       "<title>-2370443148372306804</title>\n",
       "<ellipse cx=\"559.4469\" cy=\"-34.4469\" fill=\"none\" rx=\"21.3963\" ry=\"21.3963\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"559.4469\" y=\"-30.7469\">inc</text>\n",
       "</g>\n",
       "<!-- &#45;2370443148372306804&#45;&gt;4832454587193447956 -->\n",
       "<g class=\"edge\" id=\"edge15\">\n",
       "<title>-2370443148372306804-&gt;4832454587193447956</title>\n",
       "<path d=\"M559.1998,-56.3102C559.0691,-67.8634 558.9073,-82.1795 558.7671,-94.5732\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"555.2643,-94.8131 558.6509,-104.852 562.2639,-94.8923 555.2643,-94.8131\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;2777148357443324234 -->\n",
       "<g class=\"node\" id=\"node17\">\n",
       "<title>-2777148357443324234</title>\n",
       "<ellipse cx=\"485.4469\" cy=\"-34.4469\" fill=\"none\" rx=\"34.394\" ry=\"34.394\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"485.4469\" y=\"-30.7469\">double</text>\n",
       "</g>\n",
       "<!-- &#45;2777148357443324234&#45;&gt;&#45;6996466992881930582 -->\n",
       "<g class=\"edge\" id=\"edge16\">\n",
       "<title>-2777148357443324234-&gt;-6996466992881930582</title>\n",
       "<path d=\"M485.4469,-68.8961C485.4469,-77.395 485.4469,-86.401 485.4469,-94.5715\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"481.947,-94.7183 485.4469,-104.7184 488.947,-94.7184 481.947,-94.7183\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 4693600542186039410 -->\n",
       "<g class=\"node\" id=\"node18\">\n",
       "<title>4693600542186039410</title>\n",
       "<ellipse cx=\"251.4469\" cy=\"-200.2918\" fill=\"none\" rx=\"23.2963\" ry=\"23.2963\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"251.4469\" y=\"-196.5918\">add</text>\n",
       "</g>\n",
       "<!-- 4693600542186039410&#45;&gt;8600206936994836494 -->\n",
       "<g class=\"edge\" id=\"edge17\">\n",
       "<title>4693600542186039410-&gt;8600206936994836494</title>\n",
       "<path d=\"M258.2233,-223.0951C260.6996,-231.428 263.5295,-240.951 266.125,-249.6853\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"262.8254,-250.8691 269.029,-259.4578 269.5354,-248.875 262.8254,-250.8691\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 938054775501018026 -->\n",
       "<g class=\"node\" id=\"node19\">\n",
       "<title>938054775501018026</title>\n",
       "<polygon fill=\"none\" points=\"206.4469,-140.8939 152.4469,-140.8939 152.4469,-104.8939 206.4469,-104.8939 206.4469,-140.8939\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 938054775501018026&#45;&gt;4693600542186039410 -->\n",
       "<g class=\"edge\" id=\"edge18\">\n",
       "<title>938054775501018026-&gt;4693600542186039410</title>\n",
       "<path d=\"M196.5099,-141.2361C206.1056,-151.5512 218.2106,-164.5637 228.6141,-175.7472\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"226.1191,-178.2038 235.4929,-183.1417 231.2444,-173.436 226.1191,-178.2038\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;7100503105472462715 -->\n",
       "<g class=\"node\" id=\"node20\">\n",
       "<title>-7100503105472462715</title>\n",
       "<polygon fill=\"none\" points=\"278.4469,-140.8939 224.4469,-140.8939 224.4469,-104.8939 278.4469,-104.8939 278.4469,-140.8939\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;7100503105472462715&#45;&gt;4693600542186039410 -->\n",
       "<g class=\"edge\" id=\"edge19\">\n",
       "<title>-7100503105472462715-&gt;4693600542186039410</title>\n",
       "<path d=\"M251.4469,-141.2361C251.4469,-148.883 251.4469,-158.0124 251.4469,-166.78\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"247.947,-166.7957 251.4469,-176.7958 254.947,-166.7958 247.947,-166.7957\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 5570983629454099894 -->\n",
       "<g class=\"node\" id=\"node21\">\n",
       "<title>5570983629454099894</title>\n",
       "<ellipse cx=\"176.4469\" cy=\"-34.4469\" fill=\"none\" rx=\"21.3963\" ry=\"21.3963\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"176.4469\" y=\"-30.7469\">inc</text>\n",
       "</g>\n",
       "<!-- 5570983629454099894&#45;&gt;938054775501018026 -->\n",
       "<g class=\"edge\" id=\"edge20\">\n",
       "<title>5570983629454099894-&gt;938054775501018026</title>\n",
       "<path d=\"M177.1885,-56.3102C177.5804,-67.8634 178.066,-82.1795 178.4863,-94.5732\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"174.9979,-94.9765 178.835,-104.852 181.9939,-94.7391 174.9979,-94.9765\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;7356963797885617191 -->\n",
       "<g class=\"node\" id=\"node22\">\n",
       "<title>-7356963797885617191</title>\n",
       "<ellipse cx=\"250.4469\" cy=\"-34.4469\" fill=\"none\" rx=\"34.394\" ry=\"34.394\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"250.4469\" y=\"-30.7469\">double</text>\n",
       "</g>\n",
       "<!-- &#45;7356963797885617191&#45;&gt;&#45;7100503105472462715 -->\n",
       "<g class=\"edge\" id=\"edge21\">\n",
       "<title>-7356963797885617191-&gt;-7100503105472462715</title>\n",
       "<path d=\"M250.8364,-68.8961C250.9325,-77.395 251.0343,-86.401 251.1267,-94.5715\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"247.6285,-94.7586 251.2414,-104.7184 254.6281,-94.6794 247.6285,-94.7586\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;7301052512580335568 -->\n",
       "<g class=\"node\" id=\"node23\">\n",
       "<title>-7301052512580335568</title>\n",
       "<ellipse cx=\"347.4469\" cy=\"-200.2918\" fill=\"none\" rx=\"23.2963\" ry=\"23.2963\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"347.4469\" y=\"-196.5918\">add</text>\n",
       "</g>\n",
       "<!-- &#45;7301052512580335568&#45;&gt;3122473049151551376 -->\n",
       "<g class=\"edge\" id=\"edge22\">\n",
       "<title>-7301052512580335568-&gt;3122473049151551376</title>\n",
       "<path d=\"M347.4469,-223.9343C347.4469,-231.9363 347.4469,-240.9559 347.4469,-249.2885\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"343.947,-249.3475 347.4469,-259.3475 350.947,-249.3475 343.947,-249.3475\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 6874029846087986920 -->\n",
       "<g class=\"node\" id=\"node24\">\n",
       "<title>6874029846087986920</title>\n",
       "<polygon fill=\"none\" points=\"352.4469,-140.8939 298.4469,-140.8939 298.4469,-104.8939 352.4469,-104.8939 352.4469,-140.8939\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 6874029846087986920&#45;&gt;&#45;7301052512580335568 -->\n",
       "<g class=\"edge\" id=\"edge23\">\n",
       "<title>6874029846087986920-&gt;-7301052512580335568</title>\n",
       "<path d=\"M330.6606,-141.2361C332.9375,-149.2462 335.6767,-158.883 338.2757,-168.0267\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"334.9116,-168.9928 341.0125,-177.6547 341.6449,-167.0788 334.9116,-168.9928\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 458526445283514574 -->\n",
       "<g class=\"node\" id=\"node25\">\n",
       "<title>458526445283514574</title>\n",
       "<polygon fill=\"none\" points=\"425.4469,-140.8939 371.4469,-140.8939 371.4469,-104.8939 425.4469,-104.8939 425.4469,-140.8939\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 458526445283514574&#45;&gt;&#45;7301052512580335568 -->\n",
       "<g class=\"edge\" id=\"edge24\">\n",
       "<title>458526445283514574-&gt;-7301052512580335568</title>\n",
       "<path d=\"M386.3607,-141.2361C380.2394,-150.5258 372.6764,-162.0034 365.8506,-172.3624\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"362.9177,-170.4522 360.338,-180.7282 368.7628,-174.3038 362.9177,-170.4522\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 6643938299086937416 -->\n",
       "<g class=\"node\" id=\"node26\">\n",
       "<title>6643938299086937416</title>\n",
       "<ellipse cx=\"324.4469\" cy=\"-34.4469\" fill=\"none\" rx=\"21.3963\" ry=\"21.3963\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"324.4469\" y=\"-30.7469\">inc</text>\n",
       "</g>\n",
       "<!-- 6643938299086937416&#45;&gt;6874029846087986920 -->\n",
       "<g class=\"edge\" id=\"edge25\">\n",
       "<title>6643938299086937416-&gt;6874029846087986920</title>\n",
       "<path d=\"M324.6941,-56.3102C324.8248,-67.8634 324.9866,-82.1795 325.1267,-94.5732\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"321.63,-94.8923 325.243,-104.852 328.6296,-94.8131 321.63,-94.8923\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;6757755744607918926 -->\n",
       "<g class=\"node\" id=\"node27\">\n",
       "<title>-6757755744607918926</title>\n",
       "<ellipse cx=\"398.4469\" cy=\"-34.4469\" fill=\"none\" rx=\"34.394\" ry=\"34.394\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"398.4469\" y=\"-30.7469\">double</text>\n",
       "</g>\n",
       "<!-- &#45;6757755744607918926&#45;&gt;458526445283514574 -->\n",
       "<g class=\"edge\" id=\"edge26\">\n",
       "<title>-6757755744607918926-&gt;458526445283514574</title>\n",
       "<path d=\"M398.4469,-68.8961C398.4469,-77.395 398.4469,-86.401 398.4469,-94.5715\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"394.947,-94.7183 398.4469,-104.7184 401.947,-94.7184 394.947,-94.7183\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 1124306371525190194 -->\n",
       "<g class=\"node\" id=\"node28\">\n",
       "<title>1124306371525190194</title>\n",
       "<ellipse cx=\"107.4469\" cy=\"-200.2918\" fill=\"none\" rx=\"23.2963\" ry=\"23.2963\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"107.4469\" y=\"-196.5918\">add</text>\n",
       "</g>\n",
       "<!-- 1124306371525190194&#45;&gt;2484196008774290510 -->\n",
       "<g class=\"edge\" id=\"edge27\">\n",
       "<title>1124306371525190194-&gt;2484196008774290510</title>\n",
       "<path d=\"M121.4871,-219.0277C128.7884,-228.7709 137.8319,-240.839 145.8145,-251.4913\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"143.1377,-253.7558 151.9354,-259.6593 148.7394,-249.558 143.1377,-253.7558\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;477017968081640797 -->\n",
       "<g class=\"node\" id=\"node29\">\n",
       "<title>-477017968081640797</title>\n",
       "<polygon fill=\"none\" points=\"62.4469,-140.8939 8.4469,-140.8939 8.4469,-104.8939 62.4469,-104.8939 62.4469,-140.8939\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;477017968081640797&#45;&gt;1124306371525190194 -->\n",
       "<g class=\"edge\" id=\"edge28\">\n",
       "<title>-477017968081640797-&gt;1124306371525190194</title>\n",
       "<path d=\"M52.5099,-141.2361C62.1056,-151.5512 74.2106,-164.5637 84.6141,-175.7472\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"82.1191,-178.2038 91.4929,-183.1417 87.2444,-173.436 82.1191,-178.2038\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 3722305834063150444 -->\n",
       "<g class=\"node\" id=\"node30\">\n",
       "<title>3722305834063150444</title>\n",
       "<polygon fill=\"none\" points=\"134.4469,-140.8939 80.4469,-140.8939 80.4469,-104.8939 134.4469,-104.8939 134.4469,-140.8939\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 3722305834063150444&#45;&gt;1124306371525190194 -->\n",
       "<g class=\"edge\" id=\"edge29\">\n",
       "<title>3722305834063150444-&gt;1124306371525190194</title>\n",
       "<path d=\"M107.4469,-141.2361C107.4469,-148.883 107.4469,-158.0124 107.4469,-166.78\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"103.947,-166.7957 107.4469,-176.7958 110.947,-166.7958 103.947,-166.7957\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- 8495203888380480548 -->\n",
       "<g class=\"node\" id=\"node31\">\n",
       "<title>8495203888380480548</title>\n",
       "<ellipse cx=\"108.4469\" cy=\"-34.4469\" fill=\"none\" rx=\"21.3963\" ry=\"21.3963\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"108.4469\" y=\"-30.7469\">inc</text>\n",
       "</g>\n",
       "<!-- 8495203888380480548&#45;&gt;3722305834063150444 -->\n",
       "<g class=\"edge\" id=\"edge30\">\n",
       "<title>8495203888380480548-&gt;3722305834063150444</title>\n",
       "<path d=\"M108.1998,-56.3102C108.0691,-67.8634 107.9073,-82.1795 107.7671,-94.5732\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"104.2643,-94.8131 107.6509,-104.852 111.2639,-94.8923 104.2643,-94.8131\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "<!-- &#45;3138139714318105013 -->\n",
       "<g class=\"node\" id=\"node32\">\n",
       "<title>-3138139714318105013</title>\n",
       "<ellipse cx=\"34.4469\" cy=\"-34.4469\" fill=\"none\" rx=\"34.394\" ry=\"34.394\" stroke=\"#000000\"/>\n",
       "<text fill=\"#000000\" font-family=\"Times,serif\" font-size=\"14.00\" text-anchor=\"middle\" x=\"34.4469\" y=\"-30.7469\">double</text>\n",
       "</g>\n",
       "<!-- &#45;3138139714318105013&#45;&gt;&#45;477017968081640797 -->\n",
       "<g class=\"edge\" id=\"edge31\">\n",
       "<title>-3138139714318105013-&gt;-477017968081640797</title>\n",
       "<path d=\"M34.8364,-68.8961C34.9325,-77.395 35.0343,-86.401 35.1267,-94.5715\" fill=\"none\" stroke=\"#000000\"/>\n",
       "<polygon fill=\"#000000\" points=\"31.6285,-94.7586 35.2414,-104.7184 38.6281,-94.6794 31.6285,-94.7586\" stroke=\"#000000\"/>\n",
       "</g>\n",
       "</g>\n",
       "</svg>"
      ],
      "text/plain": [
       "<IPython.core.display.SVG object>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total.visualize(filename=\"graph2.svg\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "45"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#call copute to execute\n",
    "total.compute()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "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": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}