{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "7f8bd806",
   "metadata": {},
   "source": [
    "# Training a simple Compocyte classifier with PBMC data."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "139caa13",
   "metadata": {},
   "source": [
    "In this tutorial, you will learn how to train a Compocyte classifier on any data. For simplicity, we will use published PBMC data. However, for your understanding we will go through the process of labelling these cells and fitting a hierarchy to the labels so that Compocyte has everything it needs to work."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8ff67182",
   "metadata": {},
   "source": [
    "## Preprocessing and analysis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "d4d98ffc",
   "metadata": {},
   "outputs": [],
   "source": [
    "import scanpy as sc\n",
    "import numpy as np\n",
    "\n",
    "# Load the 10x PBMC dataset\n",
    "adata = sc.datasets.pbmc3k()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "83f6aa1e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "AnnData object with n_obs × n_vars = 2700 × 32738\n",
       "    var: 'gene_ids'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "adata"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "62ecf547",
   "metadata": {},
   "source": [
    "We will make sure to save the count data to .raw, split off 33 % of cells for testing, then normalize, log-transform our data and subset to highly-variable genes. This will improve clustering results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f494f19f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X after preprocessing:  <Compressed Sparse Row sparse matrix of dtype 'float32'\n",
      "\twith 7 stored elements and shape (10, 10)>\n",
      "  Coords\tValues\n",
      "  (0, 2)\t1.111715316772461\n",
      "  (0, 4)\t1.111715316772461\n",
      "  (1, 2)\t1.4292607307434082\n",
      "  (2, 2)\t1.5663871765136719\n",
      "  (4, 8)\t1.7219784259796143\n",
      "  (7, 2)\t1.6449213027954102\n",
      "  (8, 2)\t1.4576793909072876\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "\n",
    "\n",
    "# Preprocess the data\n",
    "adata.raw = adata.copy()\n",
    "# Generate holdout data for testing.\n",
    "rng = np.random.default_rng(seed=0)\n",
    "test_adata = adata[rng.choice(adata.obs_names, size=900, replace=False), :].copy()\n",
    "if not os.path.exists(\"./exclude\"):\n",
    "    os.makedirs(\"./exclude\")\n",
    "    \n",
    "test_adata.write(\"./exclude/test_adata.h5ad\")\n",
    "adata = adata[~adata.obs_names.isin(test_adata.obs_names)].copy()\n",
    "# Normalize to 10,000 counts per cell\n",
    "sc.pp.normalize_total(adata, target_sum=1e4)\n",
    "# Log-transform the data\n",
    "sc.pp.log1p(adata)\n",
    "# Select the top 2000 highly variable genes for better signal-to-noise ratio upon clustering\n",
    "sc.pp.highly_variable_genes(adata, n_top_genes=2000)\n",
    "adata = adata[:, adata.var.highly_variable]\n",
    "print('X after preprocessing: ', adata.X[:10, :10])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "e545785a",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.14/site-packages/scanpy/preprocessing/_pca/__init__.py:359: ImplicitModificationWarning: Setting element `.obsm['X_pca']` of view, initializing view as actual.\n",
      "  adata.obsm[key_obsm] = x_pca\n",
      "/usr/local/lib/python3.14/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n"
     ]
    }
   ],
   "source": [
    "# Cluster cells using Leiden clustering after principal component analysis\n",
    "sc.tl.pca(adata, n_comps=50)\n",
    "sc.pp.neighbors(adata, n_neighbors=15, n_pcs=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "01cae0f9",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_17908/3412306151.py:1: FutureWarning: The `igraph` implementation of leiden clustering is *orders of magnitude faster*. Set the flavor argument to (and install if needed) 'igraph' to use it.\n",
      "In the future, the default backend for leiden will be igraph instead of leidenalg. To achieve the future defaults please pass: `flavor='igraph'` and `n_iterations=2`. `directed` must also be `False` to work with igraph’s implementation.\n",
      "  sc.tl.leiden(adata, resolution=0.8)\n"
     ]
    }
   ],
   "source": [
    "sc.tl.leiden(adata, resolution=0.8)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "859c1126",
   "metadata": {},
   "source": [
    "## Labelling cell types"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "9c2231ad",
   "metadata": {},
   "outputs": [],
   "source": [
    "sc.tl.umap(adata)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "a31b22cb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAlkAAAGvCAYAAABo28DeAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjEwLjgsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvwVt1zgAAAAlwSFlzAAAPYQAAD2EBqD+naQABAABJREFUeJzs3Xd4VNXWwOHfmZJJ7z2BBELvvXdURESxKyBce++9t6ui3mv7VGxXsYAFURRFEZAmvXdCCYH03tvU74+dTEmCihJKst7n4WFOmTPnhJKVvddeS3M4HA6EEEIIIcQJpTvVNyCEEEII0RxJkCWEEEII0QQkyBJCCCGEaAISZAkhhBBCNAEJsoQQQgghmoAEWUIIIYQQTUCCLCGEEEKIJiBBlhBCCCFEE5AgSwghhBCiCUiQJUQLNGvWLDRNIzU19bje9/TTT6Np2l86V9M0nn766eO/OSGEaCYkyBJCCCGEaAISZAkh/rLHH3+cqqqqU30bQghxRjCc6hsQQpw5DAYDBoP8tyGEEH+FjGQJIQD4+eefGT58OH5+fgQEBDBhwgR2797tcU5jOVk1NTXcc889REREEBAQwAUXXEB6enqjn5GRkcG1115LVFQUJpOJrl278tFHH3mcs3z5cjRN4+uvv+b5558nPj4eb29vxo4dy8GDB0/sQwshRBOSH0mFEHz22WdMnz6dcePG8dJLL1FZWcnMmTMZNmwYW7duJTEx8Zjvvf766/n888+ZPHkyQ4YM4bfffmPChAkNzsvJyWHQoEFomsbtt99OREQEP//8M9dddx2lpaXcfffdHufPmDEDnU7H/fffT0lJCS+//DJTpkxh/fr1J/jphRCiaUiQJUQLV15ezp133sn111/P+++/79w/ffp0OnbsyAsvvOCx39327dv5/PPPufXWW3n77bcBuO2225gyZQo7duzwOPexxx7DZrOxc+dOwsLCALj55pu56qqrePrpp7npppvw8fFxnl9dXc22bdvw8vICICQkhLvuuotdu3bRrVu3E/o1EEKIpiDThUK0cIsXL6a4uJirrrqK/Px85y+9Xs/AgQNZtmzZMd+7cOFCAO68806P/fVHpRwOB/PmzWPixIk4HA6Pzxk3bhwlJSVs2bLF4z3XXHONM8ACGD58OAApKSn/5HGFEOKkkZEsIVq4AwcOADBmzJhGjwcGBh7zvUeOHEGn05GUlOSxv2PHjh7beXl5FBcX8/777x9zVCw3N9dju3Xr1h7bISEhABQVFR3zfoQQ4nQiQZYQLZzdbgdUXlZ0dHSD4ydiNWHdZ0ydOpXp06c3ek6PHj08tvV6faPnORyOf3w/QghxMkiQJUQLVzcKFRkZyVlnnXVc701ISMBut3Po0CGP0avk5GSP8+pWHtpstuP+DCGEOFNJTpYQLdy4ceMIDAzkhRdewGKxNDiel5d3zPeOHz8egDfffNNj/+uvv+6xrdfrueSSS5g3bx67du06rs8QQogzlYxkCdHCBQYGMnPmTK6++mr69OnDlVdeSUREBEePHuWnn35i6NChvPXWW42+t1evXlx11VW88847lJSUMGTIEJYuXdpoPasZM2awbNkyBg4cyA033ECXLl0oLCxky5YtLFmyhMLCwqZ+VCGEOKkkyBJCMHnyZGJjY5kxYwavvPIKNTU1xMXFMXz4cK655po/fO9HH31EREQEs2fPZv78+YwZM4affvqJVq1aeZwXFRXFhg0bePbZZ/n222955513CAsLo2vXrrz00ktN+XhCCHFKaA7JIhVCCCGEOOEkJ0sIIYQQoglIkCWEEEII0QQkyBJCCCGEaAISZAkhhBBCNAEJsoQQQgghmoAEWUIIIYQQTeCU1Mmy2+1kZmYSEBCApmmn4haEEEIIcZwcDgdlZWXExsai08k4zZ85JUFWZmZmg0KFQgghhDgzpKWlER8ff6pv47R3SoKsgIAAQP0hBQYGnopbEEIIIcRxKi0tpVWrVs7v4+KPnZIgq26KMDAwUIIsIYQQ4gwjqT5/jUyoCiGEEEI0AQmyhBBCCCGagARZQgghhBBNQIIsIYQQQogmIEGWEEIIIUQTkCBLCCGEEKIJSJAlhBBCCNEEJMgSQgghhGgCEmQJIYQQQjQBCbKEEEIIIZqABFlCCCGEEE3glPQuFEIIcWZzWO2YM8vRB5kA0Pt7oemln50Q7iTIEkIIcVwcVjt57+/AfLQMNMABxmg/Im7ugc5bvq0IUUemC4UQQhwXS3aFCrAAHK59VbsLTt1NCXEakiBLCCHEcdGHeKsRrHoMod4n/2aEOI1JkCWEEOK46HwN6OsFVMY4f0xtgk7RHQlxepLJcyGEEH+ZOb2MwnkHsBVUe+y3FdcAYK+2UrYyHRwQMDwOna/xVNymEKcFCbKEEEL8JbYKC3kf7MRRY2twzKd7OACFX++neo/KzTJnlBNxbbeTeo9CnE4kyBJCCPGX2MvMDQIsYyt/Aka1wrerCrKseZXOY9bcSoRoySQnSwghxF9iiPTFu2MIAJqXjqDz2hB5Sy9ngAXgPzROJcVr4D8s7hTd6clXU2mhvKj6z08ULYqMZAkhhPhLNJ1G2PSuWAuq0Ad4NVoTy39QDN6dQ8HhwBDcMlYbpu0pZOHMHVgtdvqMS2DwRUmn+pbEaUJGsoQQQvxlmk7DGOH7h0VHDUGmFhNgAWxbmobVYgdg669HsNvsp/iOxOlCRrKEEEKIv2Hf2ix+n3sAh8Ph3BcQ7oNOL+MXQpEgSwghhDhODruD5XOSsdWOYOn0Ggndwhh6aftTfGfidCLhthBCCHGcNJ2G0aR3btttDvLTygmK8PE47+juAlZ9tZ/DO/JP9i2K04AEWUIIIcRfsPGnw8x66Hd+fHs75moro6d2wujtCrRs9XKx8tLK+PHtHexYls7CmTvITik52bcsTjGZLhRCCCH+REFmORsWHAagYmcBP729g8wDxegNGv6hJjQNRk7u5PGe4pxKHPbafC0HFGVXEN1WWg+1JDKSJYQQQvyJqjKLx3bmgWIAbFYHVaUWygpqWPvtIdZ8exBLbcHW1l1CCYnxAyAowofE7uGIlkVGsoQQQog/sHz2PnavykSn17DbHA2O26xqmrAgo5yCjHJqKi2MntoZk6+Ryx/tR2leNYHh3hi89A3eK5o3CbKEEEKIRqz/IYUdy9MwV6qRqcYCrLB4PwrSKzz2ZR1y5V4ZjHpCY/2a9kbFaUumC4UQQoh6clJL2LQw1RlgNSY42pcL7+5Nu76RHvtlWlDUkZEsIYQQop7ls5M9tv1DTXQaHENNuYWS/CpadQ6l89BYTD4Gxt3QjfBWqezfkENMUhCDJ7na6jgcDpLXZ1NdbqHzkBhMvsaT/SjiFJIgSwghhKinMLPcYzuhWxgDJ7ZtcJ7d7uDQllyCIny58vEBaDrN4/iGBYfZtDAVgAObcrns4X5Nds/i9CNBlhBCCOEmbW8hdrdZQoNJR/dR8QAU51aye2UGEa0D6DAgmhVzktnzeyYA3UbEMXJyR49r5Rx25Wflppaya2U6nYfEsvmXIxzdXUBi9zD6ndem6R9KnBISZAkhhGiRKkpqMPkaMBg9V/1lHSz22J5wSw/CYv2pqbTw5XMbnK10irIrSdtT6DwvdVcBCTvziW4ThLe/mhb08vH8Nrtizn4ObMolc7/6jJzDpUQmBtK6S9gJfjpxOpDEdyGEEC2Kw+Hgx7e3M+uh1cx6aDWFWZ6rAxN7hGMwqm+POr3GxoWpVBTXUJRd4QywAA5szCGxhyvJvaq0hp/e3sFXz2+gqsyMzWYnZWteg88vzPT8PHPVsZPrxZlNgiwhhBAtStahEo7sLACgptLK+u9TPI5HJgRy2aP90OlUXazM/cWs/yGF8PgAdHpXzlV0UhDDr2jPhFt70L5fJDarKvFQXlRD1qESdDoNnwCvBp9vNdvQNECDNj3DadtLViM2VzJdKIQQokWpLjd7bKdsz2PHsnRMvgaS12WhN+qIaO0PbjnsDocDg5eeyx/tx5JZe3HYoUP/KDRNI/twCQc25TrPNZr0hMX5Ya62MfHOnmxaeATfQC8Se4az8J3tWM2u0TC73YFOL+MdzZUEWUIIIVqUuI6h+AYbqSyubZXjgFVf7UfTwFFbbzR1RwF+wSb0Bg2/YBMDL1ArC4tzqshPUysPf5q5g5FXdmDrr0ed1/YPNTFgYhvmvbyZqjILrbuEct5tPdDrdTgcDhyePaQpyatq8ucVp46Ez0IIIVoUk4+Bs6Z1abDfUa+ge0VxDa26hDF2ehdStuWTkVxEcW6l87jd6mDZ58keleBj2wWzYs5+Z6/Do3sKydhXBICmaQy/soPHCNmACbKysDmTkSwhhBAtTmisf4NehLEdgp2r/ursXpnBgY3ZzuT0UVM74h9qorywpsE1DSYdOYdLPZLjAXyDTM7X3YbH0XVYLCW5VXj7G/H2k+KkzZmMZAkhhGhxaqqs9Dqntce+3me15pa3R3H+7T099ruv/ivKruSKx/rTaUhMg2taa+we0386vcbgi5IIj/f3OE/TNIKjfCXAagFkJEsIIUSLsvSTPexbm43BqKNNz3CKsitJ6BZGQvcwNE0joVsYPce2YseyNM8cKh0UpJfzv/t+JyTGl67DYzm6pxBztZWaCqvHZwy9rB0d+kfjG9hwdaFoOSTIEkII0WLUVFrYtzYbAKvFTl5aGTFJwfQc2wpNcyVLDbusPYk9wvn+ta3Ofe37RnFgYw4ARVmVVJVZqC6vTZ7XAAdoOo0O/aPoObpVgxY7ouWRIEsIIUSLYa6xOQMigPLCGg4U5lBeVM3F9/f1ODcw3NuZt6Vp0LZ3OAc35+KwqzfXVFqc50a0DmD8Td0JCPUGwGazo0eCrJZOcrKEEEK0GKW5Vc4Ay11Bejnrf0jBYnblX+UcLnUmxjscUFNh5axrOqPVfud02FEBmwZVZRaqyszY7Q5+fm8n7962nLkzNlFTZW34YaLFkJEsIYQQLUZEQgAh0b4UZVdi8NLhsDuwWR2Yq21sWphKYWY5NVU2fPyN9BjTCoOXDqvZjk6vEdUmkKAIX48gzWDQYbXYKS+sZtVX+xkwsa2zlU5uainzXtrk/KyLH+hDRKvAU/Tk4lSQIEsIIUSL4eVt4NKH+5GTWkpIlB8+/gbevXOFM3BK2Z7vfK3Ta1zyYD/S9xUS2z6Y8PgAAIZf0YF18w/hE+DlsZrQbneoxtBu05FF2aqultVs57dP9nHF4wNO1qOK04BMFwohhGhRvLwNtOoUin+ICb1RT6+xrVwH3UapygqrWPjOdtb/kMK+tVnO/d1HxXPD6yPpPyHR47pte0YQ0SqAsdM607prGD1Gx3scN3jJt9yWRv7EhRBCtGjdR3kGQzq9RkCYN3lHyigrrMFqtrNzeQZHduV7nBfTLhiTr5oQ8vIx0K5fFACdBscw5OIkqsothMX5YTDqCAjzZtyN3U7OA4nThkwXCiGEaNEsZht6g4bN6kCn17jisQEERfjw7h3LPc778a0dtOocwpjpXdj66xEcNgcT7+hJxv4ivHwM+AQY3c7dTnmRqgrfZXgso6d0OpmPJE4TEmQJIYRo0fb+nonNquYJ7TYH5moreqOOtr3CSdnmOXqVtreIn97e7mwSnXGgmOLsSux2B7tWZnLZw/3QdBoVJWbnewozKti25Citu4QRGut38h5MnHIyXSiEEKJFK8qp9Ng2mvQAjL+5Bz3HxhPROsDjeHWFqz5WWUEV9tq6WQXp5ZTmV4HDQVCkjzpBg+zDJaz+5iBzX9pIxoEi7DbP3oai+ZIgSwghRIvicDjYty6L7UvTMFdZCY119RbU9OAfoho67/49g+1L08k7WoZfsBdxHYMZNKktgyclodOrQqNhbn0JA8O9CQj1JmO/Gt1SH4Yzmd5aY2f+f7cy75UtHvW4RPMl04VCCCFalPU/pLD55yMAHNqay4TbelJRXENJbiV9xiVg8lW5VdsWpznfU1Fs5sonBjqbOsd3CmHey1vIPlQKQKch0Qye1A6Dlx7fQC80TRUwBZy1turkppaSsa+IxB7hJ+NxxSkkQZYQQogWJTulxPk661AJnzyymoBQb867pbsqNlqnXlecw9vz6DwkltQd+SRvyFZTg7V0eh1F2RVs/OkwJbmVtOsfhc1iJyYpiHZ9o9i7JpMNCw6rc3UaAeHeTfqM4vQgQZYQQogWpcOAaDL2Fzun8izVNgozK9j4Uypn/auL87zeZ7dm2Wf7nNu/fbaPrIMl7FuX7exfCKoptM1iZ/6rW3F33q09aFM7WtV/QhsCw33IPFhM254RhLlNUYrmS4IsIYQQLUqXobGExvox76XNHvu9ahPeM5KLWPrJXtBAr9ew1fYvxAF712R5vKfriFi6j4xn2ef7qK+qzOyx3XFgNB0HRp/AJxGnO0l8F0II0eJExAd4VGAPjPBhwAVtAVjx5X7KCqspK6jGZm/YTdo3yAuA4ChfBk5sS1icP3EdQjzOiU4Kon1tcVLRcslIlhBCiBZHb9Qx7oZubPzxMP4h3oya2tGZ1G4wuoIv30AvKks8R6T0Bo2r/z0YvyAT+tpzB01qS0TrAPLTy9i5PJ3slBJ2LEuj77mJJ+2ZxOlHRrKEEEK0SIndw5l0bx/iOoaQuqPAWe9q7PTOxLYPJrZ9MMOv6NAgAV7TNALDfZwBVt2+dn0jKciowFxlAwes/z4Fm9TEatFkJEsIIUSztfLL/RzYmEN0UhDjru+KwUvvcfyX93ZydE8hoIqJDru8PTmHS/EJMJLYPZx2fSKx/asLe1dnUpxbhcnXwMirOh7z8/yCTc7XPgFe6PUyltGSSZAlhBCiWco8UMTO5ekApO7IZ++arAbNoLPcyzmklJC+r9CZxH5oSx7rF6Tg5W1g9NWdiG4T9KefOeRiVai0qtRM3/EJJ/BpxJlIQmwhhBDNkt7gOWrlnuhep31/V3J6h/5RzqbOdcoLayjMrOCH17dht9v5fe4B5jy9jtXzDjb6mV7eBkZc0YGzr+2CucpKcb2WPaJlkZEsIYQQzVJUm0AGX5TEgU05xLQNouOgmAbnjJrckQ79ozCa9EQmBGKutrJzRQa5qaUe51lqbKRszWf7UlUFvij7KHEdgkns3njV9p/f3UnqzgJ0Oo1xN3ajba+IE/+A4rQnQZYQQohmq8+4BPqMO/a0naZpHuUXvLwNXPpQX6orLPz87k6yDpbUXqd1g/fuXpnBhgWHad8/it5nt6aipIbKUjP+wSZSdxYAYLc7OLAxR4KsFkqCLCGEEC1Cxv4iln22D02nMfZfnYluE4TdZsdud2AwuqYWNU3Dx9+Li+7tQ2lBFd6+Bkx+XtjtDtr3jyI9uZCQKF9nIJV3tAyjSc/qbw5gNdtJ7BFGcJSvc6owqk3gKXlecepJkCWEEKJFWD47mZI81W9w5Rf7GXpJO356ZwcWs40eo+JVuQY3mk7z6GX42yd7ObAxB/8QE/EdQ8g84EqaT9mW52wCnbqjgCse68/RPYX4h5joMECqvLdUEmQJIYRoEXR6zeP1+gUpWGpsAOxYlk5cx5AG03qWGhs2u5281DKS12cDUF5UQ3WFlcTuYWQdKqFdvyiCI31Iqy0F4RdsIiTaj/BWASfpycTpSoIsIYQQLcLY6Z1Z+eV+NE1j1JSOrPn2kMfxzIPFHkHWvnVZqoehgwYFSQMjfDxGvhwOB17eBkryqugyLMajUKlouSTIEkII0SJEJgRy6UP9nNujr+7Il89uoKbSiqaDNvVWCq777pAKsMD1OxDTLqhBvS1N0+gyLLapbl2coSTIEkII0exZLTZ+eW8X6fuKaNsrnLOu7Yp/sDfTXhhCRnIR3v5Gsg6VUFFaQ4f+KofKJ9CLivp9C406hlzcDp1Oa+xjhPAgQZYQQohmxWq2sfKr/RRkVNB9VBydBsVwYGMuR3ap1YAHNuXSYUA0iT3C8fI2kNgjnM8fX0tpQTUAhRkVDJqURLeRcSz/PNl53a4jYul9dmuPZHgh/ogEWUIIIZqV7b+lsXd1FgC/fVpGXIcQvHw8q797+bi+/SWvy3YGWACbfzlCRamZAee3wcvHgLnKit6go9uIeAmwxHGRIEsIIUSzYqm2OV877A5sFjtJvSPpNyGR1B35JPYIJ7Z9sPOc1J35Da6xb00Wgy5oy2WP9CMjuYjotkGExfmfjNsXzYgEWUIIIZqVnme1IvNAMQWZFfQYHU9wlBp9Ki+sJj+tnKKsSqLbBpHQNQyAqLZBHNqS53ENb38jJl8DfsEmgiNl9Er8PRJkCSGEaFZ8/L24+IG+HvvM1Vb2rVV1rmxWO3t+z3QGWQldw1jzjavhc2isH2df2xWDl5pirCozU1ZYTVicP3qDlGYQf50EWUIIIZo9o0lPYLg3pfkq9yos1s95LCDUG/8QE+VFNQD0HZ9AeLyaGsxPL+O7/27FXGUlpl0QF97TG71eAi3x10iQJYQQotnTNI1J9/Zh14p0/IJNdBsZz9G9BeSklNJ5SAwXP9CXg5tyCYnxJdGtXtb+DTmYq6wAZB0soTCjgojWrkruG348zNZfjxAS7ceE23rgF2Q66c8mTl8SZAkhhGgRAkK9GXxROwB2rUxnxZz9AGxamMq0F4bQ+5zWDd4T3sqV7G7yNRAQ6u3cLi2oYuOPhwHVJHrr4qMMu7R9Uz6COMNIkCWEEKLFSV6X7XxttzlI2ZrXoIo7QIf+0VjNdnKPlNJ9ZDze/kbnMYNRj06vYbepcvBe3vItVXiSiWUhhBAtTuvapPc60e2CGj0vO6WE1XMPsHtlJlsXH/U45hvoxdnXdiUmKYjOQ2MaHQkTLZuE3UIIIZoNq8XGjmXp2Cx2uo+Kx9vP2Oh5/Se0IWN/ETmppbTqHEr4MWpg7V6Zgbm27lbyumwGX5SEX5CJssJqaiqttOsbSbu+kU32POLMJkGWEEKIZmPFF/vZt0ZVe8/YX8yke3o3el7qznwykosBOLwtn8Pb8mnbO6LBecHRrhpZPoFemHwNJK/P5rdP9mK3O+g6PJZRUzqd+AcRzYIEWUIIIZqNgvTyRl/XZ7PaPbZrqiwe2ynb88hILqLDgCiGXtqO0rwq2vWL4usXNlGUVeE8b/fvmYy4sgM6KesgGiF/K4QQQjQbXYfHgqZedxsZd8zzKkvMHtsrv9xPxv4iAFbP3c/PM3ey47d0vpmxmdBYP0Zc1ZHMA0UeARaoelsSYIljkZEsIYQQzUbX4XHEdwrBZnEQ6lZwtL7S/CqPbavZzo7f0onrEMK+31MAL+exg5tzad0lDF+3GliapoK4vuMTT/QjiGZEgiwhhBDNSlDEn/ca7DIslv0bcqgsdY1oBUX4ADAs4GPWWydRZosCIK5DCACdh8RQUVxD3tEyOg6KJqm3JLyLP6Y5HA7Hyf7Q0tJSgoKCKCkpITAw8GR/vBBCCIHVYiPncCn7N2TjF+xN33EJ6I06LDPPxpx5gAPVwwjyyqfN87+c6ls9bcj37+MjI1lCCCGan/2LoOAQdLsEAqKoKK7hl/d3UZpfRb/zEuk+Kh6DUU9chxDnSFUd4+i7Mc69hl76BTDmCShMgV8eAbsNxj0PER1P0UOJM40EWUIIIZqXbXNg/i3q9Yb34bb1bFqYSnZKCQCrvtpPUp9I1s4/RPreQhJ7hDPiyg5oWm3GfKcJcH8yWGsgIBo+Pg+OrFbHKnLhppWn4KHEmUiCLCGEEM1D7j7I3QOHlrn2FR2G8hx0Bs25S9NrpO7Kd9bT2rUig4RuYR6NofFxG92qLmn8tRB/QoIsIYQQZ76j6+CTiWAzg38UaHpw2KDVIAiMo/8EO+VFNZTmV9H33MQGb9cKD0H6ETXyFdoGht8H+tpq8eOeh2+uU9cb//LJfS5xRpMgSwghxJnv4BIVYAGU58AlH4F3ECQOA50ebz8942/q7jzdYXfQfWQsaet3kqhbRcKyT2GVN1hrSzvo9DDiAfW67Sh48NDJfR7RLEiQJYQQ4szXZgSselWNNgXEQrux4BPsec7Ob+D31yE0Ee2CtxhxfggkX+86bnWrnVWcdjLuWjRzEmQJIYQ487UZAdcvhuxd0P6chgFWRQF8e6MKwnJ2QkginPUMRHVX2wAJQ1WCu284DLzpZD+BaIYkyBJCCHHmslTDhvfU7wNvhLi+Dc/J3Qcfj1cBVh2rWU0J/utH2PsDBLdW04IV+WAKAIOp4XWEOE4SZAkhhDhz/XQvbJutXh/4FW5Y6nm84BC8NwJsNa59sb1VYjuoEa9eU1TABeBXu8LQXAFFRyC0LRi9Xe/N2gGHV0Kb4RDTs0keSTQfEmQJIYQ4c2Vsdnu9SY1aRXZy7fvuZs8AS9PB1G/BN1QFYJ9dBCXpKsl99CPqnPJc+GAMlKRBWHu4fokKxvL2w//OBms16E1w0wqI7HxSHlOcmaR1uBBCiDNXq0Ge20fXeG5bKj23+12nAiyA31+D4iNqGnHFDDVVCLBzrgqwAAoOwK556nXmVhVggQrcMreeuOcQzZIEWUIIIc5cIx9QpRoAjH6QMMzz+Lkvgn80mAJh7FMw4T+uY75hrtdGPzDUTgtaqj2vUZKhfm8zHPxqm0L7hqtkeyH+gDSIFkIIcWYrPqqKkcb1hbAk1/7KQtj3I4S1g4Qhap+5UtXUCm4F4R1h8RNQlApD7lCJ7wD5B+HtAa5Eeb0Jel2lCpFWl6oRrNhe4B95Eh/y9CDfv4+P5GQJIYQ4swW3Vr/cVRXDm72hulhtXzYLOl8In14I6RsADS77GCb81/Ueux0cdpX87hcJ5artDrYa2DxL5WcNuR06nNPkjySaBwmyhBBCND/LnncFWAAHFkPrIbUBFoADkn+GrhepzYNL4evpKudK01zV492ZK9TvhYfhhzvAXA7jXoSEwU35JOIMJjlZQgghmh9rvSApYagaoYrq5trnnlP123NgLgO7xTPA0vSgM6r3D7xR7Vv4AKSuUtOG865rumcQZzwZyRJCCNG8WGsgZ5fbDg28/GqLj/4Ee+ZDcAIkjXad4tdIflVIIty1Xb3e+yN8dwtEd/cMwqw1Dd8nRC0JsoQQQpwcdjvUlDZsefNXOBxw6DfQG4+9qq8iHzZ9BGXZqmaW682w4iXI3ALdLoG+/2r43gvehMVPqsT4wBiw21T+FajE+rnTwW6F/T/D8PuhIg9qyjxzuoSoR4IsIcTfYsnOJvP+B7Dk5BBx550ETTz/VN+SOJ2V5cCs86DgIHSeCJd9Cim/qfIIXSe5yjAcy88Pwob31evh98PYJxqe8/klkLWt8ffn7lG/Nv4P7tzacGVgQDRc/H7D92VuUwVN7VbXPp0ebl37x/crBJKTJYT4m/LeeJPKTZuwpKWR+eCDWEtLT/UtidPZ1k9VgAWwdwEsfVoFRQvuhFnnq5GqP7LnB9frfT82PO5wQPZO13ZkF+h4PiSN9SxYai5vPBDb/iXMuRJWv+HaV5SqKrzn7XXtC4qHdmPVKsVZ56s2O0Icg4xkCSH+FnNammvD4SDjvvuw5ebhN2QIEffcjc7Ly3XYbKb8998xRETg0737KbhbcUqtfw/2/+ra1hkgZ49rO3uHWgnoE3Lsa7QZriqxA4S2aXhc06DPNNj8sUpWH3Yv9LhMHVv2IqStc51bUaBG1nL3qCnIrbOhqkAd2/+zysXqcqG6x/qrDC98W61CLKst7zD3X3DnlmPf969PwLY5EN8PLv1I5YaJFkOCLCGEk8Nmo3zVKvSBgfj26XPM86xFRVTv3u2xr3LV7wDUJCdTOGsWxlataP3RR3jFx5F28y1UrFHtTmJefJHgiyY12TOI08yOuWqqD1TwkzQG+l2rRpQOLgEcauXeHwVYoKYId89Xq//2L4K0DdBqgCo4CqpVzsTXVb6Vd6Bq7FwnaRSseAWoLS664E6V21VXkqG+shz1e+tBKkG++AigQc8rISDGFWCB6nMIkH9APU/8AIjvq/YdWQtr3lSv9/8Cmz525XmJFkGCLCGEU8Z991P2yy8ARD36CKHTpjV6njk1FUdVlXPb2Lo1lqNHXSc4HFiOHqXwk0+IvPsuZ4AFULZ0iQRZLUnxEddrh00V9IzoqCqzhyVBaSa0O9vzPQWHVOPnhCFqeg7UVKPdUnsdO2Rth9y98OPdqunzhW+rICi2l+e1fn4Y1s9ULXQsVaqXoc3ceB0sgJhe0ONy9do3FG5aqT4rsrPK4yrLViUd6u4lvh/88qjKF7Nb1CjdtYtUlfn5N3te22A6zi+eONNJTpYQwqlsyRLn64JPPuHoTTeT9cwzWPPyPM7z7tQJU8eOAOh8fYl79b8EXXIxuuBg0Ln+WzGEhaHz88PbbYrQb8DApn0IcXrpcK4aDQJAg/XvwAejoTgNNnwIP90PvzykVh4C5O6Dd4fBtzfAu8NVEAaQOBRCa1vm+Eep6678jwq47FZY+lzDzy7NUgEWQGUBGHyOfZ89r4KHUuGmFZ6rH32Coe1IFWBZqmHtWxDXp/ZeNEhZBuvedgVddiukrYdDS1VOV53WQ6DP9OP5yolmQEayhBBOfgMGOEedrBmZWDPUNzhzymESPpnlPE/n40PiF3Oo2rETr8QEjNHR+Dz/PDwP5atWUfT5bLzatCHkyiso+uprQqddjb2yCkNUJAGjRp2CJxOnxIK7VY6UXwT0ux42faj2V5fAxg9h+xy1vXmWCpo6jldFPi2Van9VIaRvgi4XqOnEm1aoPKnw9mqUKTQRSmpHUEvTVQmGs591fb7JH7wCVJFRUCNUxUcg+RfA7jqv3Vi46N0/f54VM2DN/6nXOiPQSLK+0Q/anQUpKz33Vxao5P/+1//554hT4u233+aVV14hOzubnj178n//938MGDDgH11TRrKEEE7x77xN5H33NdhvPnyY9DvuJOvJp7AWF2M+epSqbdvw6dUTY3S0x7n+w4fT6r13iXr4IdJuvY3sp54i84EHsZWWeARYp6A3vTiZSjJUgAWqplRlHnj5q+3AODVt6C59M1QVqWR0Xe3P/97BqulzHVMAtB6oAiyASz5S++rsnu95TVMATJmrktiH3q0CsB5X4BFgdTwPrvrqrz1TqVsuVt3Ilbt2Z8Nt69V06Kb/eR7LT4af7vNM+BeNstkdrD1UwPfbMlh7qACbven/r/jqq6+49957eeqpp9iyZQs9e/Zk3Lhx5Obm/qPrao5T8D+ddPEW4vTlsNk4Mm06VZs3qx06HTo/P+xlZa6TNA0cDnx69iTh88/QjMaG17FY2Ne9h3NbFxSEzsuLoMsuo3zJEmoOHSJ06lSiHnm4qR9JnAo15fBaFzVqBTDmCeh+marEbq1RBT4P/ApHVqvjml4FVHW9BeP7wyX/g5CExq9fZ/5tsO1z9brnZLho5h+fX5EP74+CkjTwDYcbflMrDLO2Q//rVEV3qxkKDqim0+5BXPZO+OwiFTQOvQsOLIHc2gUgQ+6Ec9ymLD86F442Ukvr+qUqj2vj/1S/xA7nNF4c9TTV1N+/f9mVxTML9pBVUu3cFxPkzVMTu3But5gT/nl1Bg4cSP/+/XnrrbcAsNvttGrVijvuuIOHH/77/0dJkCWEaMBeU0Plho3og4MwtmrFgeEjwNLIT+5A24ULMcbHUbpgAZrJm8DzxqPV5mWl3XIr5cuW/eFntV24EFPbRpbkizNf+mbY8J7KXxp+H+gNsO8n+HKyOu7l57nCzzsEqovU68iucOuahtesz2aFnV+rOlk9rlCfAeq6xyqXUFWk6ltFdVXTf6tfV/s1HVzzK/x8v6qlFRAL1/0Kwa08P89araYia8pg7jUqGX7EfSq5vqpYTXuWpMOix8BWowLN3H3QazKc9zKkrIBPL3Bd818LVc7ZGaApv3//siuLWz7f0mASVqv9febUPk0SaJnNZnx9ffnmm2+YNGmSc//06dMpLi7m+++//9vXlpwsIUQDOpMJ/+HDnNtR999Hzksvu5KT687z98cYHUXmffdTtngxANU7dzpHp2JeeJ7MBx+i5tAhrJm1Ccy1o2AAGI3o/KVuULMV3xfi61VRT9vgem2ucJVIaD8O4vrB8ucBDQb8hdyl9M0qiOk12bWvslAVCc3dDZ3Oh0nvwKpXVR2uYfeoGlg+ISqZ3W6DdW4jXw47rHjRVay0LBP2fO9ZdkFvAH3ttOeWT+Gg+nvP3Gtw5mh1Oh+unA1XfA6Ln1CjdX2vgawt8FyE5xQoeJaEaKFsdgfPLNjTWJYbDlSg9cyCPZzdJRq9TmvkrL8vPz8fm81GVFSUx/6oqCj27dv3j64tQZYQ4k+FTp9OyFVXkf3SyxTPnu3c79u/PzpfXyo3uL5xupdryP3vf6lYtQoAzcsLXUAAgeeOA72emgMH8enaherduzFERKBpJ/Y/TnGa6jxRFSe1VqmpuWt/VeUU6lb0db9E5WQZfVVDZnM5jHlc5Tm5W/UqLH1GvR50K5z7onq9bbZrCm/fj/BlKaTWJqEfWQu3uwV5lioVpLmrKga9ybU/slPjz7H3R1j0uNsOt/BgvyqDwq55ajUiQOZW13H3aUSdQdUJa+E2HC70mCKszwFklVSz4XAhg5PCTt6N/UOS+C6E+EushYWUzJvn2mE0Ejpd1dHyHer6JlFz+DAVW7aS++prVKxc5dzvMJuxFRRQNHsOfgMG4BUfT8GH/yP9llvJfenlk/Yc4hSL76eSw6+er+pJefl6lkwIS1J5WD/do1Yf7v0Bvrq64XV2fuP2eq7rtVZv7CDPbSSi6LDnMZM/DK5XHNTLD6bNh0G3wWWfQOIINfVnt3met+o/eCTQG31dr9uOVr/XrZL8I3armr5s4XLLjh1g/Z3zjkd4eDh6vZ6cnByP/Tk5OUTXW9hzvCTIEqIFsZWXY83P/0vnmtPSyHz4EdLvupujN9zIkX/9C0e16z+40GlX4zdI9YSrdh9St1rJnTGDgvffx5qbq6YH66nauYuy5cud28XffIPDam1wnmimQhIgafQft5gpd6vNVp7T8Hhrt3prrdxed79UJdHXSRrtqo818sGG1xn3vCovAaD3goE3qiKo574AMT3g5TbwWld4q7+qk1WnrkhqHZsFJn8NF72vpgpB5Yi1O8vzvOBEGHoPmGobYrcZAeEdPM9JWQGLn4KU5Q3vt5mKDPA+oecdDy8vL/r27cvSpUud++x2O0uXLmXw4MH/6NoyXShEC1G+ahXpd9yJo7qa8DvvIOLWWz2Oly5cSOmiRfgOGEDgOedw+MJJ2Csb/0lcFxBA+W/LsOblEfXww1iOHPE4rundvsk5HCQt+42M2++gevdudH5++PbpTdmyZdhqi5zay8sp/flngiZOxFpUBHY7hrAzZ0pANIHRj6oRLEslnPPvhsfHvwKxvdVKxd5uI13+EXD5pyrhPqw9jHsBJvxXBUj+EY1/1vn/hcG3qhITAW55OT/ep6YrAQoPqfysnleo7YlvqhWUKcsBh+qb2GGcOmazqDpg5kq4bBbk7Fa1wCI6gk+oakbd/TLoPQWie7iS9UGtcvzsIlUdf83/qdWP9avYN0MD2oQSE+RNdkl1o3lZGhAd5M2ANqFN8vn33nsv06dPp1+/fgwYMIDXX3+diooKrrnmmn90XQmyhGghCv73kXMkKn/mu4TfcoszD6p63z4y7lX1scoW/YolPeOYARaAvawMc1kZ5sOH0QUGgs01lWKMj8fUuRPVu3fjMJsJu+F6vGJiSJj9OdW7doHRSNqNN2EvKfG4pmYwUPL992Q+9jjY7UQ99iihU6ac6C+DOFO0HakqsDtsqs9gfXqDCmwa0/l89cvpT0Y/StJVSQXfEFWKoa79TUW9GkkBtVNHKStUOx+Dtxq1CozzDIS+uQ721q5IO7QUpn2vAqyN/4MFdwEOOLoGorup6vHucvepZwb1e+7eFhFk6XUaT03swi2fb0HDs8xr3Vj4UxO7nPCk9zpXXHEFeXl5PPnkk2RnZ9OrVy9++eWXBsnwx0umC4Vo5koXLiTj/gc8p+NsNg6dMw5zZiaVW7eS939vebynYv169BGun/o1n2O3I9H7+WGMd02dWNLTKZ7zBQ6zGX1UFBH33AOAztsb3379MB9K8QywdDo0H28wepE/cyZYrWC3U/D+B//wycUZT6drPMA6HtYatdrwxXh4f7Qqu1CSATOHwr+jYPkM+PwS1Rrnt3/Dr0+o95XnqcKodbz8VdV2gB9uh8IUdXzps55BkLlCJdzXObpW1d366Fz47Tk8wodqzx80AFV9PiRRvQ5OaDjd2Iyd2y2GmVP7EB3kGRRHB3k3WfkGd7fffjtHjhyhpqaG9evXM3DgP28BJiNZQjQTDocDW0EB+uBgNIP6p121cxcZ993vLJngP3Ys5UuXgt2OJS2No9OmY8nIcJVUqGWMjKT1hx9Qvup3DCHB+A0ZQvazz1L8tVuCsU6HqWNHwq67jpDJkzk4clSDe7Ll5GCvrELvVqbBp1cvNB8f1WBapwO7HUdVNdmPPYbNreCpsXXrE/jVES3WgV9Vqx6AzC2w+zs1fZezS+1b/iIe4w35yep3TcM1hoKaNvzmGijLUbWy6uTtU1OBdQVFq0tdI1EAIW1ViQb3BHw0tQCgfhHS9e+pVZNB8WoVZtLYY09xNlPndovh7C7RbDhcSG5ZNZEBaoqwqUawmpqMZAnRDDjMZtKuu44Dw4aTcv5ErAXqJ25LVqZHAOXdtYvH+6xFRQ0CLLy8iHzoQQyhoQRfeAH+I0agGQxEP/kk+pAQ13l2OzX795Pz8svk/Pt5fAc18lOfwcCRqVMpWbjQucvUtg1t5n5N9NNP49Ozp+tyFotHHa7wm2/6O18KITwF1Bv9CIgFk1sRTZ1RVXoHlfg+4CZY+7aqCh/RQRUYdbfoYVU/y93B2oRpc4UaKes9VW2bAuHi9yEwVpWrcHKoshV1bYYAynPh54egPBsyNql8rM8mqTIQxUfh00lqJC7197/3dTiD6HUag5PCuLBXHIOTws7YAAskyBKiWajctImKNar2jjk1lZIfFgDgP2IE3l1UYGXq2JHQqVPxrRsC1zS8EhM9rhN85ZW0X74MU5uGFdg1gwF9/WR0m42Sb+ZRtngxlVu24jdmNPpQt8RUq5WaffvIvPc+cv/7XwAcViteiYmEXHkFsS+/hN+wYRjiYnHUywH7q6sghfhD8f3gwndUgdDz/gPtz4KBN4F37eo+DdU4+s5tcM8elSe16FHVdid3r1ohaDj2dDmgcrKK0+D/+sLb/VU9rju3w/371QpFvbG2qrurwC9Z29UIl6Wq9j70oNM3vPbBpfDLI5CyTI3Ezf3XCfiiiJNFgiwhmgFDdDS4regzxsYCULF6NdXJavrD2KoV+sBAQq64XJ3kcFCze7fHdYq//JKU8ydStUflolRu2cKBESNJHjiI0l9/xVHjWbjRI+gym6n4bRm2wkI0X98GpRtKvv+BglmzSO7Xn+S+/Sia+w01Bw4S/cTjWLOyPR9I0zC1a/e3vx5CeOg9RSWoD7hBbRcfceVD2SyQ/DOEtlFTc5oOj2lCL38Yfm+9C9YbWfGPVCNOdZXbCw/BihlgdAvOvANV1fc6IYlqpOr5aPjuZlXOwn1kq+5z2p+tpiDruLchEqc9yckSohkwtW1L/P+9SenPv+DTuxeB484BoGT+986Vf+VLlmArK8NSL6DRvLxwmM3ObVthIek33Uy7FcvJffkVVesKyH7uOYyxcVjS0gDQh4YS9eQTZN51d4P7iX7sMYyxMaTdfgeOCvVNwVZWRu6Ml5znZD/9NNhs6Pz80AUGYi8uBsCrbVsi770Hn65dT8jXRrQQRamw4QM1NTfw5sZHheqEJoF/tJqaQ4PWbrWQguJVyYd1M9WKwCF3qAAptg98f1vtexyuvot+EdB1kloVWP9+6nMvOlpRAObaHMTtX6iipNXFruPhHeHCt6FVf9j+lWu/w6Gm1XUyRnImkD8lIZqJgDFjiHvlZUInu/q4efdw5YF4tWmDzt+fgNoAzEmvx6dPb4//tK15edTs24fO3/WTtd4/AJ3J5Ny2lZQ0GmAZW7fGf/Qo/AYPpuPaNUQ//RTG1q1Voru72uDPXlFB0IUXYoiNBYMBQ0QEuf/3FvuHDaf4u/nH/4UQLY/dDrMmqhY2ix6FFX/SQcAnGK5fDOe+BNMXQIdz1ArAnd/A/l9VjtYdm9Tol3dt/lb7s1Rj6Dph7eGmVaoy/I/3qtGruP6u43V5We4ytrhem8vcDmieVetBBW9xfVSl+ZI0136HTfVYFGcEGckSohkLu/56DBERWLNzCL70EjRNw15W5nGOzt+fxDlzyH39dQrefQ8AzWRCHx5OzDNPk/38Cziqq4h8QJWByLj7HixZWR61sdDpCL/1VvxGjcQ7KQldbckHzcuLkCuvJOfFGR6faYiJdk0RahrGqEhnA+nK9eud52U9+SRBE85D8/I60V8a0ZyYy6HkqGs7by+UZas+hMfqPRjcGgbd7Nr+5hpX6YVRj8Kohxq+Z/h9sPhJ1W9w2N1qynHp0+pYyjKYPFfV2PINrZfoXqvHZbDrG5X0njRWNYre/wtk72h47pHf1TRieAfXSkhNBxNe8yxeKk5r8iclRDOmaRrBkyZhLSrCcuQIOn9/vBISMMTHY01PB00j6iH1zSTirrswhIRQvXcfQRddhDEyEoBW77ztcc22i34huVdvj5WAvv36EXH7bce8D/8RIyhbvBiAgPHjsRUWuoIsh4PqlMONvk8zGmVaRPw570A1crT1c5WkHtUNXu+ugpm+18DE1//8GnUrBAEOLmk8yBp6p0qS13upQGr3fM/jVUWuivCNSRoDd2yG0kyI66eCpV3fHPv8vQug0wTXtsMOnc7782cRpw3N4ai/frvplZaWEhQURElJCYGBgX/+BiHE31aTcpgjkydjKy7G1LEjiV9+gb26moqVK/FKaodPN5X7ZE5Lw1ZQgHePHmhugY2tuBhbeTle8fGU/PQTWY8+pmpbWSxoXl4EXXIJUfffh87v2H3oHBYLZUuWoPP3x3/4cPLffY+811/3OCfgnHOw5GTj06Onupf8fCLuuhP/4cOb5OsimqG8/eATooqF7v/Ftf/xPDC4jYZW5MPX01TtqiF3QL9rYf4tsO8ndXzM4zDigWN/TkmGCnj8o+DLySooazMCJs1UU4phSX/9nj8eD0fWNH6s/Th1f3MuV+2FNJ1KmJ/8NYS3/+ufcQLJ9+/jIyNZQjRzZb/+iq02qbwmOZmUiRcQdtONBF96KQD5739A2ZIlquWN3U7geeOJe/VVACrWrSPtlltxVFURMnkyZcuWOVcYGtu0wbtdO8wpKVQn70fn60Pmgw9hr6km+skn8R861HkPmtFI4Pjxzm2fXq76WHVMnToS/+YbTfVlEC1BRG2j5YhOriArNMkzwAJY/QYcWa1eL3laVW3XDDD4DtVQut3YY3/G5lnw4z0qAX3c8zD1G5U3lbIc3ugJdgt0uVD1T/wz1hrIP+ja1nQqePPyh1GPQP/rwegNN6+G/+utjhWmqJyzS6QjwplAxuGFaOa8u3T22Lakp5P9xJPkvPwyB4YNI+/VV6nescM5/Ve68GfK16zBkplJ0ezZzoT1ojlzMLgVI7Xl56v6WOvXk3777eT8+3lq9u/HcuSoGu1y47B7Jup6tWqF5u1qnWGIjCT4kktP6HOLFmzMEzD4NpXzNLCRoraGer0MHXawm+Hw8oYBVnWJCqq+mgpZO2DtO7WJ5w5VtBTUSsbfnlMBFqhG0oWeTdMb9cujnv0R6xLah9wJQ25XAdbW2fD2AM/3+QT/+bXFcVm5ciUTJ04kNjYWTdOYP3/+CbmujGQJ0cz5jxhB0CWXUDJvnsf+0vnfYysqanC+5u1N2rXXoZlMBJx9tnO/ITaG2NdeJffll6n4fbVHAr2ttJSaI65vKrbCQufrw5dcSvXu3Wi+vrT5dh6mxESMcXEkzPqY0sWLMcbGYmrfHp3vnxR8FOKvqiqEzZ+qFXwZm1Udq64XuY4PvQuKDkNesqrQXlSbExic4Hkdqxm+uMo16pW2wXNlX0RH12t9vdGyilwIrXe9+opTG9+fscn1etGjruBN06vp0M4T//i6Zzq7TU2hlueoKdmEIX9ckuMEqKiooGfPnlx77bVcfPHFJ+y6EmQJ0QLEPPM0+vAwSuZ9iy0/H52vL4bICI8gy9SlCz59+1L82WcAOGpqqDl0CL9hwzDGxxN2zb/wSkjAml/gWZRU08BqxZaX59zlsFiwlZVR/vtqqmsLnjoqK8l+6mkSPpkFQPW+ZMqX/oYlMxOH2YwxPp7EuV97jJYJ8bcUH/UskZCzxzPIMvnDJR+q1yUZsOIltSpw1COe15l3nSvAAqjI8wyyBt2qfnc4VNX3OtE9IabhlHgDA2+Gw6vAVqMCwfLaUa32bmVW/CNd9bMcNqjMhwV3wZ1b//z6Z6I9P8AvD6nFAXUCY1W5jS4XNNnHjh8/nvFuKQ0nigRZQrQAmsFA1D33EHLllZQtWoT/6NFgtZL50MNU790LNhuG0FAi77qT0h9+wF6iqmHX7N1LDRB08cV4JSRgKy6mevt2j2vr/P0blIXQh4dx9Lrr8Onh+Y3GfPgwGffeR8jUKWQ/84xH30RLejqVGzY6C6kK8bdF94BWgyBtHfiEQvdGpqIPLlG5TYFxqviob2jDcw6v8NzufAHsme/a9o9Svx9YDLu/de0ffGvDPLDGtD8b7tunRtP8o1QJCb8IaDvSdc6EV9W0ZPpGFWCByuVqjvb8oBYkUG89XmmW2n/5p00aaDUFyckSooWoTt7PobPPIXfGS6RMOB80Dd8BA5z1rip+/x1LRgb6gIAG7y35/nuOXHMNtqpqTO1d7W6CLrsMna9vg/Nt+QVU79hJ0ezZmDp3VqNdmoY1N5fShQvJeeHFBo2pNZMJU4dTs2JKNDMGL/jXj6pY6J1bXNN6Dgfs+BrWvA1fXQ1p61Vw9Nu/G79OR7fyCX3/BQUHVc9DL3/wDYdDv9UerBcU1E0d7p4P/xsH398O5krY9DG83BbeHeaqCO8bCiEJKv8qcZgqR2GvrUG34G745Hw4uhbOeU4l8QfGQbdLVXX7mvJ//KU6bdhtagSr/tcSXPt+edj1tTlDyEiWEC1E0ZdfgNWqNqxW8t97H78BroRaXUAAhshIjHFxWNLTPd9ss1G5dh3ZTzxB608/pXTBAgzR0ZQvW441J+fYH+pwUJOc3CCgqq7tjaiZTJjat8N38GACzzmn0cbUQvwteqNqzuxu2fOw8pWG55qPEaxc+LZaKag3wPKXXEVB696z5Ck1tbfkGdf+tqOgyyQoz4N516t8qrR1aoRq9Ru1U34FsPABuOorVx241NXw6YXq/OAEuGYRbP5YHasuhsMrVcC49DlY9R+1/+cHoe1oVZneeIbnNB5Z4zlF2IADSjPUeW3OnLIuMpIlxBnGYbWS8eCD7B88hMyHH8FR156mpobKjRtVNXY3tvJy8t56G2t+gcd+706dCL7kYqKfe5aQq6+m9ccfYwgJIejii/Hp1w+vRho0V/z+O5WbN+M7YAB+Q4bgsFn/8F692rXzbBSt16tftasNHTU1xL/9DkETJ5L15FMcPGcc5at+/ztfFiH+3NF1bhsamAIhojOMerjx83U6VZ/qi6vUdF1jkn+GXPdG65p6n7XalbAOqs+he3L8gV9Vlfk6a992nV98BH68E4xuo8QhbRo+g8MOh5bC9i+P9cRnjvI/+GHt75x3mpAgS4gzTNnixZT+sABbUREl8+dT9ttvOMxmjk6bzpGrp3HovAlUbdvmPD/z/gfIf+stymsrrgNofn4EnH0WACGXXUb0Y4/i060rGfc/QNZDD1G1aRM6Xx867diOsW1bj8/PfvJJDk+6iJTzJhBy5VUepRjqePfqhd/IkZhTUpxNoP1GjiDpt6WE33mH8zx9aCj6oEByXnyRmr17sRw9SuajjzS4nhAnRLdLXK+7XwqPpMFt6yC07bHfs+wFVTm+sWkso49nP0NwFSINbqUS6Y1+Kgl+2D1w+ScqsKuzZz4sqi13Etza8zoHflUBntFX1f3qdH7tM1xEAxs+aDBafMapy287UeedJiTIEuIMo7k1aQbQeXlRc/gwVbUJ6Y6qKkp/dlW7rjlwoME1HBUVHDpnHMXffuexv2zJEufr6h070by88Onu2YPNVqhWJFpzc6neuYPgyy/zvLjBQPyr/1VV42tHrOwVFeCAQ6NGU7l2HeF33IFXuySMrVtTuXkzmlsvNq2Jl2qLFqzfNXDTSpj2A1z0/p+fX1X0xyMnlipoNQACYtR2ZGcY96Lr+KiH4bFM9ZmBMdB6cMOcorVvqdyuc1+EhGGg8/Kc+rNUqsr03/xLbfe/Hm5cASFugWHubrf8sDNUwhC1ihDtGCdoKh8tYUiTfHx5eTnbtm1jW+0PqIcPH2bbtm0cPXr0j9/4JyQnS4gzjP/o0YRdfx3lq37Hf+RI/EeOxFZWhj48HFu+Wn3k3UMFRg6zGQxuQYtb4IPdTtGXXxJ8sesnY0N0NJbUVOd2xZo1GELrrboyGJy5Xab27Qm5+mp0gUEUvPMO2O0YYmLInzmTmkOHXG+Ji6Ni5UoAKtetQ2fywnxQHU+/404SZ39OdlUV9ooKomQkSzSlv1Jaoc7mWVBTeuzjEZ1UsvodW6A8W+VS1ZTBF1eqps9tR8OFb6nyEKByjiwVDa9jNatp9Wtq2/qses3VeLqOe7AX2wsSBkNRimvfmrf+uFL96U6nV2Uavp6GCrTcR+ZqA69zZzRZvaxNmzYxevRo5/a9994LwPTp05k1a9bfvq4EWUKcYTRNI/L++4m8/37nPn1AAImzP6f0558xtW9PwNix5L3zDvlvv+NcPQigj4jAEBREzf79AA1W8yV+MYcDI0aCReWG5L7+BoYot+F5nY74N16n5MefMB86RPH8+Xh37YqjrMwZvFnT0ihOS1PnG41E3HE7pvbtSb/lVudl7FZXLpejshJDdDSJsz8/MV8gIf6O8lz4/GLI3QsDblQjS95BbidoaoTJUqk2/aPhul9do051U47vDVd1ugB2fg2xvVVJB1AjYwYfsFYBOnX9dmNV42h3FXk0MOxez+3Rj6pcLEftv++U39T1fc7gOnNdLlBlGhqtkzWjScs3jBo1iqZo5SxBlhDNhFdCAuE33wyANS+P/Df/r8E5tpwcbDk5+A4ejN+QwYROm+Zx3BASgikx0TnFaAgLw3z4sOsEux1bWRkVq1ZhLy+nZv9+dCYTuiDXNyPN24SjuraOj8WCw2Kleu9ewu++i7KfFuLdsyfhN1zPkWnTnSsTS+bNI+z660/kl0OI47P+XcjeqV6ve0eVbOgzXfUKzNgKva5SU3K7ajsntD+rXhCGKtlQXG96yWZWvzscaoTLqtpUEZIApgDY9Q3k7oFrfna1y6mrh+Vu6XOw5VMYfDt0uxiC4qH31bBlljruHeSZKH+m6nIBdJpw0iu+NxXN0RSh25+QLt5CNC1bWRn7hw4Ds7nR4/rgYDqsW+uxz15TQ86LL1K9ew9oGobICMJvv53Cjz6m9IcfnOf5jRxJxapVzpErXUgI9qIi0DQCzj2XkCuvIPvJpzBnZGDq2IGa3apcgy4oyFnkNPKBB6hYt05dB0DTaPf7KoxhYSf6SyHEX7P6TVj8hHqtM8BdOyDIrYr7/kVQkg42C3j5QmQXlbSuN8L5r0NAFLzaWU0X1gmMh9vWqwrz394EO9xWAXoHqb6IddqMVIFdt4shdx98dhGUZdJw6gzQGWHafFVr6+BSFcgNvUvlhzUx+f59fGQkS4hmqObAQTRNa7SsH4AxoTW2sjIcZjOG2sCm8OOPKf7yK3VCbe5W+YqVxM6YQeX69c5RJ98+ffAfNoycV14Bu10FWAAOB2WLFxN0wUSSFv2Cw+Eg5cJJzs+sC7AAyletwhAR7rohh4OjV08jaeFPJ+pLIMTxGXizqsN0cDF4B0Pq79DzCnVs7TuwqDZXMH4AXL8Y3uoP+WranR/vhqiungFWQJxqfWPwgp3feAZYAFHda1v21P4rPbxC/TIFqlGy+/aqJPnyXHi9u2c5CLtFFVOtKgQ0uPSjkxJgieMnqwuFaIaK58717C9Yy9C6FSFTphBy1WQODBvOgaHDyLj3PnJenEHNQVeiujM53mKh+IsvaPv9fCIffoiwG66n4vffqd69m4i773IVN61jtapq7kDJ/O8xu61sNLZq5XztP3IkfsM9CwqaU1KwFhZStmwZ5n+4okeI42bwgqSxUHBINZX+7ibI2KKOubfXSd8Ah5aBxa10g6XKc1QKYOzjrtY6qfVqvxm8YczjcPEHKjne3e+vul7r9GpV4qUfqelBd1V1Tdgdqq2POC3JSJYQzZC9qtK1UTsqZeralcRPP0Hn58eRq6c5g7DShQvVeSYTuoCABn0I8fJCFxSEMS6OvFdfUysWN27Eq23jtYX0AQHYKyvJe/NNZ+0ezdubpJ9+pGzpUvTBwfgNHowlIwPNZHLeh++QIRyZMhXz4cNo3t4kfPoJPj16NPoZQjSJ0gy3DQeU1Rb27TQB9rvKovDZJFUJ3mFXgdT4l9W/s70LVCX41oOh51Wu8wsOul5rBhVYfXyuGjFrPRiPKcHqEljzf6psQ99rIL5fbZ7S+fDD7bBtdsP7PpNXFTZzkpMlRDN08NzxzlIM+shIou6/j4LPPscrOproZ58h9z//oWTetw3eZ2rfvmFdLU0jcOJEj7ysOoETz0fT6fDp3ZvyFStx2KxEPfww+W+97QreAJ/evUn8Yk6D99ccPkzJjz/iFR+PMSGBo5OnOI+F33oLEXfe+Te/AkL8DdUl8MlEyNqualZ1HA+Fh1QCvM0CX01VpRoAglrBPbvUDxKLn4CDv6nGzv2uhbB6nQ4+u1hVZgcazbHy4HbcFAh373CtGNz0sZqarBOWpBphT3xD5YaBSt7/8R4VAE54VZV7OIHk+/fxkZEsIZoZh82G3teXugwOU1Jbsp56GkdVFTU7d6IPDSX60UcxhIRQvf8AlVu24Cgvx7tnD0KnTyfroYdxWNzyPxwOKtasafSzjDGxRN57DwAhV14JqAR690BN8/Ul/p23G32/qU0bIu9QFeBtxcWuWl+ahk+fvv/wKyHEcfIOUoU+q0tUI+mfH1D7d81TifDtzoJttaVGWg+CFS/D76+5yjrk7lblGMLrNTo/5zn4KlUFbMcKsLz8a3souh2vKYW9P0Kfq9V2pwnq84qPAJqa2iw4BKFtYETtvf5wB2RuVa+/vx1ukTZVp5LkZAnRjFSsW8fRG25wNmD2apdE7AsvqCm+Wo7qKnR+fviPGUPl2rU4yssxxETT+r33CDrvPFp/+gnevXuj+arl4H7DhuE/apTz/YbISOfr4m+/peDDDwHVI/HwZZeT3LOXOlj7k3z4jTdiCPnz2j364GASv/ySqEcfJeGzT/EfNvSffCmEOD6WKvj5IZhzOWRtgwK3Ed3qErWycOLrcMFbauRo5MOq4bSl0vM61oa5kER1hen1RoIjO6scMFABlnsza83tW/NP90Fxbd0533DoeB7oTXgEYyVuDd3d+4m6J8uLU0JGsoRoJnL/+yoFH3zgsc+al48xJoboJ58k77XXMERHE3777QAUfPyxc8TKmpWN+cgRfIKDyXnu384gzbtXT1p/+AEOux3/IYMBKPpmHtbcXABs+fnk/ue/eHXoiDUjneqdqs5QzYEDxLzwPJqXF/nvvkvJ/Pn4jx5F4HnnNWjT484rPo7QaVef0K+LEH/JqldVrSxQiepXz1erAqsKwRQE7w5VeVGXf6pqOP3vrIbX8PKHHV9B3l4Ydl9tPqQNFt4PR9ZC4nBI26CmE6fMVZ9jLofoHmokaumzqpG0byhs+p+6pq0Gls+Acc+rvLD1Mxt+rs7tW/n5r8L3t6npwvNfP9FfJXGcJMgSopkoW9xwhZE+IoKcGS9Rvnw5gRMnEvXoI2iahjk1lfKlrl5nmrc3WU8/g2/fvtjcEt8dVWoFlabTEXjeeQAUft4w8bbgnbcJvfZaj31Zjz2OPigIW3Gxet/Hsyj8fDaJX3yBT7eu2MrLyXrkUWoOHSL0mn8RctllDa4rxEnjXmXdUqmKhd69A35/A1a9ovbv+xFWvwHr31PV1eszl8PeH9SvulGnbXNg00fqeB5w3WJVbqEwBebfrIKhtPWqSOmub9T2uBdUw+i6wqbbPleFOduf0/i9b/4E2o6C355X9bqm/aBWJYpTTqYLhWgmfPv3b7jTZqNw1izMqakUffaZswG0tajIVaYBcFRXU7N3L0Wff46pQwd0gYHow8OJeuRhj8vZKyvRu9e3qlVzOJXsJ55EFxwM+trKzA6HM8Byslio2r4NgIIPP6Rs8WLMKSlkP/kURV9/TfmKFQhxSgy5Q/Ue1HSqhU1AtKrIHtnJ7SRNjTbVJb//ke1fwKtdGvYgrJtOrClXAVWdHV+BtVoVFl37luqHGJrkOp67F3pNVu186tM0+OZ6NYKWshyWPN3wHHFML774Iv379ycgIIDIyEgmTZpEcnLyCbm2BFlCNBNRTz2J5uPjsc/i3hIHwKb+U/fp2dM5MmVs5Vl/p3zpUuxVVcTOmIHfoEEex9LvvpvyRb8CoPP3c+63V1RgKynBXlzs0SsRwKtNG9VUGtAFBuI/tDbXyv08h4PsJ58i7aabyX399b/8zEKcMGFJauTq8Vw46ynX/u6XwtinoMN46HE5HrlQgfHQ7TIYfj+c84Ln9dLWg8MtP8ovEgbeAm1q68PF9PAs82Bzy58KiFWrBfv9y7Wv1UCwW+H6JSqfS2dUuVl+EXDR+5735fD8N3imsdltbMzeyMKUhWzM3ojN3rTPs2LFCm677TbWrVvH4sWLsVgsnHPOOVRUNNLM+zhJCYfTUI3Vhl7TMOglBhbHp3LjRnJff52qzVs89usCAvAfM5qqXbuxZmYSeN55xD7/b+w1NehMJo5ccy2VGzd6FBcNvuwyYp57lqpt26jcvBm/YcM4fPElzuBIFxRI648+wpKVTeYDD+CoUj3ZdIGB2EtLATDExNB+2W9Yi4qo2r4d706dMEarn8RtxcVk3Hc/NSkp2IqKcFSrqUlTl860/bZheQkhTrmiI/DhWVCRq2pcVRer/X4RjTd1djf2KZWDFZIA5/wbvPxUW5w3enqel3QWXPCGKj763S2w3a30SUAs3LAUPhjjquHVawpMekfljy15Ro3AXfKh+pwm0NTfv5ccWcKMDTPIqcxx7ovyjeLhAQ9zVkIjeXBNIC8vj8jISFasWMGIESP+0bXku/hp5osNR+n65CJ6PvMrqw78yT9aIerx7d+fxNmzCZ482WO/ZjJRtW07lkOHcFRVUTJvHhVr16IzmSj6+msq165tUL3dEBND1a7dpE69mtxX/kPqFVd6jD4ZIiIxJSaS8+yzzgALwKttWyLuu5eIu++m3RKVJ2YICSFg1ChngAVqNWHr/31I+2W/ETDWVUzRf5hnJXghThvBreH2DXDdEvW6jkeAVVcfy+3bq6ZX5R4OLVX5WV9cBVs+U7W2ul3q+RndL1H1sP6vDxxY5HmsLFP1UKwscO0rz61936Vwy2oVnH01FXbM/adPe9ItObKEe5ff6xFgAeRW5nLv8ntZcmTJSbmPktoWYKGhof/4WpL4fpqZ8fM+rHYHVrON15ccYHj7iFN9S+IMFPPkEwRfMJGjt92OvaAAW36+qj/lxqHTYSsvd64krC/ovPFUuAVfjupqfAcNpHLdejAYiLzvXixZWc6VhnWqk5NJ/GIOmnsxxj8R+9IMAs4ai2byJmDM6D9/gxAni7lS1aZK2wBp61SB0CvnQJ9patUggG+YK/AZeg9EdVZTeukb1UrEtmPg+1tc16zrU5i7V60GzNsHObvUKsNNH6vWPXX0RtdUoqaD6O5wzvOql6J3MIx6xHXuqv/C7tpR4Pk3Q9Jo8GuYQ3k6stltzNgwA0cjdcQcONDQeGnDS4xuNRq9Tt9k92G327n77rsZOnQo3bp1+8fXkyDrNBMZYKKkyuJ8LcTf5dOrF7jVx3Ln3asnGXfcqab19Hpn6x1DZCT6sDBCrrgCr8REMBjQBwdjKy7GEBND3KuvUvz1XMrXrqFm3z78hg7Fu2cPqrfvcF28qorDky4i4ZNZ6IOD/9K9agYDgePH/+NnFuKEW/QIbJ7l2rZWw2/PwdR54B+lyjvEdIetn6nVgFlbQXNAt0tUQNSvdtVtzk5Y947ntbd/oQqJ3rhCJdNrOni1s+c57c6BwbeqhPbEYarNTnw/dd3KAlj3NqT8BoNv96zRZbepHK4zxJbcLQ1GsNw5cJBdmc2W3C30j25kkc8Jctttt7Fr1y5+//3EFHGVIOs0897VfXl18X78vAw8NL7Tn56/5lA+T/+wG51Oo8Zip7DCzAPjOjJ1UNPMx4szS8hVV1Hw/vsA+A4dQtwrr+Awm8l44EFn3lTdFGDg+ecTO+NFNIPrvwWv+HjaLviB6n3JeHfrijU3j7w33gCHg6r1Gyj89DNaffAB9rJSKrdtI/+NNwGoSU6mZMGPhF499eQ+sBD/lMOhpvUMPpA4VFVUr887WLXfObpWlWq49hdVXuGdIYAdUpapabv+17veY61ueJ2qQjUaZqmEw6vAyxdC20GhW69Du1UFV4nD1HZNuSojsf1LKHFrpF6SDqMeVdXeCw7CsHtUftYZIq/yr6XH/NXz/o7bb7+dH3/8kZUrVxIfH//nb/gLjjvIqqqqYvPmzYSGhtKlSxePY9XV1Xz99ddMmzbthNxcS9Q2wp+3JvdxbqfmV2C22bHaHGxMLWRIUhjtowKcx+/9ajvZpZ7/eJ/6YTeX9InHx8s1pLpgeyYr9ucxqmME5/eIbfoHEaeFyHvvIfDccTgsFnx6uhJsG5vIM8bFeQRYdSzZORTPm4dx9WocDoez6TOAraiIgg8+IP6N19GHhpL/f285S0MYoqNO+PMI0eQW3AVbPlGvRz8Gg25VU4W2GghpC3F91CrDOZercyrz4aNz1e/uKmqnD9e9qxo+uxcM1Rk8R5mWPONaEegXoZpI161MNHp7Xnf+LaoOV325+1SNrOsWNTx2Bojw/WupMX/1vOPhcDi44447+O6771i+fDlt2rQ5Ydc+riBr//79nHPOORw9ehRN0xg2bBhffvklMTGq6FlJSQnXXHONBFl/Q3J2GasO5DGgTSg94oMBlQT/6Hc7cThAr4HNAV56HV/eOJA+CcdOyDMZdLWzPw7mbU1nd0YJn6w9gsMB87akExvsQ5/Wf97mRDQP3rU/DJnTM9AHBqAPDCT62WdIvfIq7CUl6Pz98Rs2jLAbrqdq+3Yq1q7Fb9AgfHr1wmGxcPS661yjXo3kWdlKS8m4/wF8B/Qn7j+vULroV3z79iHw7LNP5mMKcWLsXeD2+gcY+SDctw/ykmH5C6qIqN3q1muQhgFWdHfofx2UZMAvD+MsrxDdQ+VIDbhJTfMdXulZdBRUEv2Fb6tEeb8IOOsZt2MFcODXhves08OAG07I458qfSL7EOUbRW5lbqN5WRoaUb5R9Ins08i7/5nbbruNOXPm8P333xMQEEB2tqqDFhQUhE+9sjjH67iCrIceeohu3bqxadMmiouLnclhy5cvp3Xr1n9+AdHAobwyvtqQxqfrjlBtsWPUa3x361C6xQUxZ/1R56CBrfZ3s83OdbM2seKh0QR6G/nv5T158vtdmAw6wvxNlFVbuXZYIjpN4+zXVnAoz7POh8MBRwsqJchqYbKeeYbiL75E8/HBb8hg9AGBtJn3DcaYGLTa4qHV+/dzZOrVOCwW8ozv0OabuRjj4lwBFniMYgEYW7dWKxOB0h9/pNX77xH/xusn67GEOPESh7lGihJrV7r6hsKq/6igCGDhg6oX4a5voSxbVWqv03MyXDRT9RDc8aX6waTu3037s2Hsk+p1h3Fqym/ZC2AwufKpEodD76nql7uaMvjpXs9pR98IFVz1mXbGV3jX6/Q8POBh7l1+LxqaR6Cl1Y69PzTgoSZJep85U7UqGuXWoxXg448/5l//+tc/uvZxBVlr1qxhyZIlhIeHEx4ezoIFC7j11lsZPnw4y5Ytw8/P788vIpy2HCnkknfXenzfstgcbEsrpltcEF1iAtmZUdLgfUVVFg7mlNM9Poih7cJZet8oANKLKrnivXXc+cU22kcdbBBgAXSJCWRs58gG+0XzZSsro/iLLwFwVFU52+lUJyfT9jtXPaqaffucvQyxWKjesxfvjh3Rh4ZiKywEQPPzw+FWoM+a55kfYcnIwJKdTe4r/8FhsxF53714tWrVlI8nxIl1yf9g59dg8IauF7v229wWkdhqIK6v+mWpUjWrMjarRtDnv6bOmTtdteEBNSIV3x9y9sDzMRDRCaYvgJ8fhJraH2Li+sGohyFpTMN7Mleo+lx5+zz3V+apUbAzPMCqc1bCWbw66tVG62Q9NOChJquT1ZTlQo8ryKqqqsLglrOhaRozZ87k9ttvZ+TIkcyZM+cP3i3q+3FHVv2BAYJ8jAxvr5bcPjupK+0i/SmuMvPlhjQKKlz/yKd+uA6L3cG/J3Xjiv6tqbbYePTbnWQUq3pFB3LKPa6r12DhXSNIivCTIqctjDk93fOn6VqWzEwKZ89G0+kIvuQSfAcNwhAVhTUnB0NkJN5du1D05VeETptG/vvv47DZiHn2WSrXraN47lyMbdpgOXLEeT3N15fA8eNJv+ceKteuc35Gm6+/OqnPK8Q/YvBqOIoEqp/gvOtV4vnIB2HRYxAYq0anLvg/VTG+jsMByT+7tvVeENpWtcsByNwCX072nH4PiFbXakzWDs8AS2cEe+0PRDvnwqS3/96znobOSjiL0a1GsyV3C3mVeUT4RtAnsk+Tlm1oSscVZHXq1IlNmzbRubPnEtO33lJ/cS644IITd2ctwLD2EXy0OtW5fWmfeO4f15HoIJXoaDLouWFEW9YeKuDtZZ4rXCotKrn43z/t5Yr+rXlo3g5WHnDlBXjpdYT5G8kqqSE2yJt3pvShY3QAouUpnvOFR4ClmUw4rFaM8XHkPPdvAKq2bSf2pRm0/X4+1fuSMbVtw5Hp/8Jc25Yn9MYbKZ4zh8wHHyTygQcImTwZe2Ul2sCBFH/5JZq3t0p+Dw7GVlDo/Kz6tbmEOGNFd4fb1quFHW/0dK3sW/So+v3s52Doneq1pkHbkXCotgl7aQasrRcIHV0H/a5RU5A+oSqI27sAdn8HrQbBwBtd54a3V+dUFQIadJoAe+arY0YfFfiZ/JvqyU86vU7fpGUaTqbjCrIuuugivvjiC66++uoGx9566y3sdjvvvvvuCbu55q7KbCPA24DV7mD6oAQePq8zNruDvVmlRAV6E+rnBUD7KH+CfIzO+lnu7HYHD8/bwZajro7w3gYdH0zvx/D2EVRbbHgb9Q3esyerlHB/kzOgs9jszFx+iLTCSqYPSaTKYiM5u4xzukQRGVhvdYs4oxjjXKtJdYGBJP28EM3kTcrEic79Vdu3A2AtLKJyw3pqDhxwBlgAhR9+6Fw1mPuf/zgLlHolJdFuxQp0vj7oA1QQHzDuHMypqTg0jcgHH2jy5xPipDKXe5ZOqLPxA9Vk+ueH1DRhwlDoew1s/rj2BAeEJKpWOqCmHNe/q0bBAmLgw7GuyvG75qkpwM61/0b9wlXPwn0/QUxPMPq6gqzqYlXHa8jtTfXE4h+Q3oUnwEu/7OPrjWn0iA/ircl98DP9cexqttpJLajg4nfWUF6jvlkNbx/OJ9cM4LpPNrIsOY8Ak4HZNwx0rjRMza9g3pZ0Ks1WhrWP4KcdWfy0I4sqi1r22yUmkD1Zam5fA64a0IobRiTRJrxhntxts7fw084svAw6PpzWjxEdInhjyQFeW7IfAH+TgQqzFYcD4oJ9+PWeEc5n2pNZyrLkXPomhDCobdiJ+PKJJmLJzqZy40ZMnTpRtuhXzEePEDp1Kj49egCQ+9rrFLz3HgARd91JyNSpHBp3rjP/yhgfjyU9veGF9XpnbS3Nx4dOW119EvPffY+82gbP/qNH02rmOw3fL8SZ7vvbYOvnqnioQ/3wQYfxKgn9c7c8rvNfUyNYBQchqLUKlBbcBfvdphITh6vAqyTN8zN8w2HqNxDbu+Hn5+yGmUNc2xNeVYn0y56H1NVq1O2id8F04mcvmtv376Z23HWyUlNTWbx4MWazmZEjR56QsvNnsl0ZJcxcrqbyliXnMXv9EW4ckXTM8zOKKrnk3TVkl9Sgc5uO9/XSk1pQwbJk9ZNMWY2VeZvTnUFWYrgf953T0Xn+6I6RLNie6dwO8TOi08DuUIuF52xI4/ttmSy4YxhtI1zDyNklVfy0UzUWNVvtzN2czogOERwpdCUz1wV+ABnFVRwpqKRLbCBphZVc+u4aKs02dBrMvXkwff+glIQ4dSy5uRy++BJshYVoPj4kfvUl3h06OI/XHDyI/7Ch+I8aiWYw4t25ExWrVzsDLFBtdxrjO2AAVdu24aiuJuLOO537q5P3U/qra3l5xQmqmCzEaefCt2HIXWpV4I6vVaA16BaV/O7OFAg3rYKCAyonyxTQsNxDREe1QrG+ynz47GIVMJ31lEqyrxPVFca/DNvmqP1JZ6lCqDW1C6WKj6jgbtTDJ/a5xXE7riBr2bJlnH/++VTVNoM1GAx89NFHTJ3acqs6G+slkXvpdbywcC+/7s5mSLtwnruwG/raaMpud3DZe2vJLlFLde0OaBXiQ/uoAJ6+oCsB3kZC/bworE1w7/AnOVSPT+jM0wv24G8ycPdZHSip2s2uDNdy+wqzjY2phc4gy+FwcOeX2zyu0SVG/SRyzZA2LN2bS0mVhbGdIll5IA+LzUH7SH/aRqjRsP05ZVSabc5735FeIkHWaap61y5nwOSoqqJy/QZnkFWy4EcyH3oI7HYCzj2X2BkvcmTKVKq2b0fz8sJhNoPBgPVovSkRTcPUsSOxM15EHxSEw2JxThHmz5xJXm219zp+I/9Z93ohTmsRHVSR0t3fqhGt9merXoHD7lG5VcGtVU0rSxX0cUuxqV9BfsvnMPF1VQBV7wX5+9VqRVA5WIdXwFfT4N7dnu8beJP6BbDzG1eAVefkT1KJRhxXkPXEE09w9tlnM3PmTLy9vXn88cd58MEHW3SQ1TE6gCfP78LXm9R0YetQX55eoBruphYcZVDbMC7oqXJiSqosZBZ7VmdvE+7HtUPbEBPkw4r9eUzsEcOmI0V0jwtiysA/bo1z9eBERneKxM9kIMTXi0+uGcDbyw7y2dojWOzqH9jHq1OZvzWTIUlhTB2UwIbDrpGK7nFB3DyyLVabnWXJuYzqGMHFveMZ2TGCAzllHMqrYEi7MGdOV7+EUBLCfDlSUEmwr5HRHaUUxOnK1KUrOl9f7JWV4OWF7wBXEmnhZ58586vKfvmFigsvcOZkOcxmQqdNw5KXR9nPrikNU+fOtHr7LYyxbt0CvF25esVzXXWCDJGRRD3yMAFnNc1yayFOGz/cCXl71esf74ZJM1VwNfpJ+GY64IAdX6kRrnZjAT20Gug5XWirhqLDqjUPQPpmVcC0YD9UFat9lQWw8X9wcKmq6p5/QNXNGvO4Cu7i+3kWR9V00OHck/M1EH/ouIKsXbt2sWbNGmeF91deeYX33nuPgoICwsJabn7OtcPacO0wVYZ/xX7PukF1KW8/7cji8fk7MRl01FjtzuMrD+Sz7nAh0wcn8MEqV6Lx7sxSRneKZET7CBZszyTEz4uzu6g2JUUVZuZvy2DupnT2ZJViMmi8P60/IztEYNDrnAEWwL7sMgDWphTQLtLfI3frot5xvLZ4P2sOFbDpiEqcX56cx7pHxtI+KsCjfQ9AkK+RBXcMY1d6Ce2jAoiQBtanjMNmo3zFSnS+vvgNGtjgePHs2SrAAvwGDsC7o2uq2VE7El2nYs1aMBicyex+Q4fg1aYNNQf2Y8nIJOKOOwDI/c9/CbrwAvxHjmzwed5du2LJVNPXvv36SbNn0fJYquH90Sqhvb7NH7slwNdylmHQoPVg1/74vnD9YsjYAl9dDaWZYK1ShUjrm30pXP4ZdLkARj4Ei59Q+x12yN0DcY3kc4mT6riCrNLSUsLDw53bvr6++Pj4UFJS0qKDLFCjVI9+t5O0wkrGdIokObuMIUlhzj6Bj83fSXGlWh3YKdofg07HrkwV7Jitdlbub7jU/UBOGR+vPsy6FDX6dHHvOJ6a2JWL3llNakGl87waq4P//X6YLjGBvL8y5Zj3WFpt4YsbB/HTjixigr35YVsm323NaPAcpdUWj76H7gK9jQxpF97oMXHyZD7yCKU/qPYfEffcQ/hNN3ocr1i92vm6asdOj2MB542n5o0Dzu2izz4DgwGf/v0xhIbg0DRyX3kFQ0goMc8+hyXtKJkPqdyO0sWLSfp5IV71mqfGvvwSRXN6gU5HyFVXnshHFeL0deFbqnZWSRoUpzYeYB2LwQvGPq9a7SQMbng8rg9c/R28/SelDLZ+BuEd4eAS1z4v/8avKU664058X7RoEUFBQc5tu93O0qVL2bVrl3NfS6yX9fqS/fy0Q82jmww6tjxxtnNFXqXZipdb7lZ8iK9HJXcfo55zu0aTnFPm3Bfub+K87jH8d/F+575vt2ZwtLDSI8Cqkxjmi59Jj7/J4ExcH5gYgq+3gZX78xnUNpS8shqGv/QbbcL9eH9aP179dX+D61w1oDVRUrLhtFe+ZKnzddlvSxsEWf6jR1O9R01bB9RrFRF+003oAwKpWLOa8t+WqZ1WK1WbN4PdTtniJc7pxPQ77yR0ymTXmy0WrLl5DYIsnY8PYddde4KeTogzRGRnsFlUL0O7FbW2+y/mQnW/3JVTdSw+IX9+zZTl8M4Az319r1GJ9uKUO+4ga/r06Q323XST6y+KpmnYapd3tyRVZtczW2x2rLXNBnNLq7nonTXkltXg66Wnf0IIz17YjRcW7uXH2qDsrM6R/LRLvW4V4s0zF3Sjf5tQAryNjO0UxZK9rvYCm44U0TrUl6OFKtDSaXB+jxgiA0xc/8kmLugZw+K9OZRWWekYE8izF3bDZrPz6Pyd/Kc2qNqerlZEXj+8Dfd8tY262UU/k55HzuvU4Nme/2kPi3bnMLhtGC9c3N2ZyC9OHd8hg52Blt+ghj+xRtxxOz59euOorsa/XpCl6XSETp1C4LnjOLzrEqy5uWg+Pq5pRLtrOtteUUHgxRdT/N18LEeP4j9mDD49ezTZcwlxRpl9OZS6lTkJbw+DblWBj92mmkQHREPuPlVeoSQNul0CvaZAwhA4sBi2fKpWEA6/H+qv6C05yjEDrNg+oDdC2vqGxyqkCPDxmDlzJjNnziQ1NRWArl278uSTTzL+BKQ9HFeQZXf7z1d4umNse3ZllpBWWMU9Z7UnyNdIVkkVl85cQ0Ztsnul2cbUwYnEBvvw38t7MqhtGGarneTsUg7mqoTFtKJqbv58C59cO4DBSWHMnNqHR7/dydzN6h9yv4QQ3ru6L8v25aLTaQxOCiM5u4x/fbwRgDWHCpz39OnaI0zsGUt6USVfbfSsd+Rt1HNhrzhWHcjjm81qyrCixkZxhYVAbyOfrElly9Ei4oJ9nLliRwsrGZQUykW9PUcxxMkX9+qrlC5ciM7Xl8Bzzmn0HP+hQz22i774gsLZs/Hu0IGYf/8bQ3g4beZ/R/XOnehCQkm/9VZs+fkYIiLAaMReXk7044/hFRVF0s8LsZWWYgiRxuJCAOqHkaNrXNs+IXD5p2p0q981nufu/EaVcQDYNlutQCzLUq11bGbVkNo3FPpf7/m+sHYQEAtlmTRwxWeqpU5jQVb9lYZnGIfNRuWmzVjz8jBERODbr6+zkX1TiI+PZ8aMGbRv3x6Hw8Enn3zChRdeyNatW+nates/uvZxj2T9EbvdzsKFCzn//PNP5GXPCHHBPvx4x3CPfe8sO+QMsOqE+6sq7iaDnok9Yjn/rVWkFXomIpttduZtSWdwUhh2h4MJPWLo2SoIo17HxJ6x+HoZuLSfq+mu+4rB+n7ZlUWN1fMnIZNBx+1j2gFww/AklifnkV9u5uI+cbQO8+WnHVk89cPuxi7nVFptYUVyHm3C/egWF/SH54oTT+flRfCkSX/5fHN6BtnPPgcOB+aDhzC1b0/4LbegDwjAnJGBff8BWs2cia2iHJ8uXdDXKzKo6fUSYAnhTqdTFdn3fK+2xzyuAqw6dhts+khVcXcvp6DpVPPpinzPptOlboFUaZYq3RDTC25YqvogHlmtKsGDarGTvRMyt6lq8XUlH+okDjuRT3pSlf76KzkvvIg121U7zBAdTdSjjxzzB8p/aqJb9wuA559/npkzZ7Ju3brTI8g6ePAgH330EbNmzSIvLw+LpWH7l+Zke1oxn607QptwP24a0daj4bLFZuc/vyZzMKcco6FhMcdfdmfTu7X6ZrU1rcgjwDLoNKy1c3ddYwOx2x1c/b8NbDhciE6DEF8vPlx1mDeu7E2XWNc3wUBvo8cUIqhZfH+Tnv/9ngqo4C6/XP2DrrHa+WBlCvec3YGIABMxwT6UVFrQgGXJuaxNaXyo+cJesUzsEUuN1cZlM9eSnFOGXqfx4bR+jO4k5RxOazarx3/0DosVW3ExWU8/TdkviwDIA7x79iDh44+PcREhhIdLP1b9CX1C1apAd7/9G35/Vb2O6AS9r4bcvWq0KriVSlQPa69GuELaQL9rYfGTsOUzVYrBZga9Ca5bpKYdV77sunZNGXw9zTNIc7f1cxh4M5xhTZVLf/2VjLvublDjy5qTo/a/8XqTBVp1bDYbc+fOpaKigsGD//nigb8dZFVVVTF37lw+/PBDVq9ezfDhw3nyySe56KKL/vFNnc4qaqxc/b/1lFar5HKdpnHLKFeF9//9fpj3VqgVfl56jdEdI5xV3EEVJK3TKTqQAG8DZdVWdBq8cWUv9mWX0SrUl8v6xpNdWu0cpbI7oKDCTEGFmX//tIc5NwwCYMvRIq77ZCP2etP2mgZlNa48sfN7xPLd1gxn/8Nf9+Rwz9kd+HRtKjvT1dDyvC0ZzNuSgcnQMOcqIsDEK5f2xKDXcSSv3Jmkb7M7WJacK0HWac4rIYGI++6l6PPZmNq3x2/USA6dOx5bcbHHedXbd1C2bBlBEyacmhsV4kyi06s6VY3J3uF6nZcMN68Gfe233M2fqLpaAEY/6DgB9v4Iq9/wvIatBlJWqF6F7lXhDd5gLuOYcveo2lr+Z87/yw6bjZwXXmy8iKrDAZpGzgsvEjB2bJNMHe7cuZPBgwdTXV2Nv78/3333HV26dPnH1z3uIGvjxo18+OGHfPnllyQlJTFlyhTWrFnDO++8c0Ju6HSnShy42s4cLaxkR3oxM5cfIirQG/dWkGabgycndmVUxzzeW3GIthH+3DKqnfN4dJA33906hGX78ujdOph+iaFMcMspDvc30Sbcj8P5rpY3AF5uI2R7s0obBFgGncZto5NYtDuHfdllGHUaCWE+BPoYnEHWkCRVciPQ29jgGd2nFw0aGAw68spqeO7HPTw3qRtxwT7O+9I0GJIkJR3OBOE33ED4DTcAkPfW2w0CrDoeBUeFEH9Pn2m1CfBW6D3VFWAB5O1zvbZUwLq3Gr+G3gsiOsMvj+CxyrD/DVB8GHZ/p5pOX/Khqqv1y8OqRlbrwar34RmkctNmjynCBhwOrNnZVG7ajN/AAcc+72/q2LEj27Zto6SkhG+++Ybp06ezYsWKfxzXHFeD6B49elBaWsrkyZOZMmWKc67SaDSyffv2v3wzZ3qDyQfmbmfu5nRC/bz47NoBXP3RBmcrnMv6xrMhtZAjBZVc3CeOS/vE07t1CD5eehbtzia9qIpJvWIJ8//jQp7fbkln7aECckqr2ZddRmywD1abnSqLjV6tgtBpOjJLqhjdMZIPVqWQU+qqzzKwTSjrDxeiAWM6RbDyQD4Wm+cfc9fYAF66pCcdowN4YeFe1hzMJzmnvMF99GoVxLY0VxLlmofHEBvsQ2GFmSV7cmgb4Ue/RGmtcyqUr1yJvaKCgLPPRjMc389LpYsXk3GH6juoeXsTfMUV2PLz8R81kqB6+QlCiL+p+ChUFUFMT8/92Tth1vlQXXzs94YkwlVfwtLnIPkntc/gDUPugJEPq6Ctqkj1R6ybFszaASXpkDQGjE1Tiqepvn+X/PgTmfff/6fnxf7nPwSd3/Qj7WeddRZJSUm89957/+g6x/U/c3JyMldccQWjR49uEaNW7irNVmatSQXg6Qu68sC4jgT6GHE4oKjSNS+++lA+3982jNSCcqb9byPfbsmge1wQF/WJ49nadjtz1h/h13tGotdplFRZeGf5QcxWO7eOaseuzBJeXLiX/fUCntyyGvq0DmZXZimH8lwjW2sOFfDNzUPIL69h3aECYoK9eW2xWsXiAJYn52NrJI7enVnGA9/s4Oe7hvPUxK7MWn3Y2Q7IXXyIrzPICjAZCPBWf2VC/by4vH+rBueLE8NuNlO64Ec0bxOB48ej1Vvanff22+T/n/rp13/UKKy5udSkphJ2/XVE3Hrrn14/8Oyz4fXXqN69m4Bx5+LT7Z8ldwoh3BSnwfYvIbQNdL/Utb80C3Z8CcEJcNd22PUd/HR349coOgIVBVDoVmDaYFLJ9t9cA36R4OUL6RuhyyQYdDPE9FC/zkCGiIgTet4/Zbfbqak5juKyx3BcQVZKSgqzZs3illtuoaqqiquuuoopU6agac2/btL9c7ezcKcaytyeVsx7V/ejxmrjv78m0zbczxn4ZBZX8/j8nSSG+VFlUTlROzNKCPXzcl7rUF4FO9OLWZacx/LkXLbX5kTtSC9hd2YJ1ZbGS2VkFFc12OdwqLyocV2jGdc1GoD/++0g1K49cA+w3BPrAaw21+fUtdWp7+aRSbQN9yMlv4J/DUkkwNvIL7uy2J5ewnndYugeLysLm0LmffdTtngxANU7dhL1yMMex8tXrnS9/v13Z0uc/Df/j+CLLsJY2/rqjwSeey6B50p/MyFOKEs1fDxe1cQCqC5Ries2i9pfVNs+7bz/wIAbIGurKvFgqah3IQd8UjtiozeB0QcmvAqfXwoVuZ6nHl2r+hfG92vSR2tKvv36YoiOxpqT03helqZhiIrCt1/fhsf+oUceeYTx48fTunVrysrKmDNnDsuXL2fRokX/+NoNl7/9gbi4OB577DEOHjzIZ599RnZ2NkOHDsVqtTJr1iz2729YQby52F3bAgdgT2YpFpudmcsP8cGqwx4jSwAF5WZ6tw52bof7ezGpd5yziOegtqFc/+km3lh6wBlgAaQWVHj0NXSn1+DWke0I8VU5VAHeBjQgPsSbt5cd5JbPN7M7U13r7rHtne+b1CuGFy/uzu2j27HiwdHMu3kwSRF+tA3344WLuzvPaxXqS12sHOhjICnCjzeu7EWn6ABqrHaKKs1kFFexeE8ON3++hZnLD3Hl+2vJKVUlKg7mlnPu6yvp//wSvt/m2apHKI7jWHVbuWGD83XFRrfXGzZw9PobwOpa1OCV0Nr1RqMRzVsq9gtxylTkuQIsgMwt6vfKAleABbDuHfX7BW/CQ4fBr94IjeaW3G2rgQvegs7nQ+UxCo3+0dTjGUDT64l69JHajXoDN7XbUY8+0iRJ77m5uUybNo2OHTsyduxYNm7cyKJFizj77GMsajgOx5WT1ZiSkhJmz57NRx99xJYtW+jWrRs7duz4w/eciTlZ7688xAsLVbKiUa+BQ60O3JnpCpK89Dp0OrWC0M9kYFKvWCIDfTivezQJYX4cyCkjq6Sa6CBvznltpcf1NQ2emNAFs83O60v2423QUVzlSrCfe9Mg+rcJo7TaQm5pNan5FVz/6WaPa0QGmNjw2FkAJGeXUVZtoW9CiHOk0Wqzk1pQSWywN75erkHMz9cd4fH5qi1SbLA3v903Cm+j+ov82dpUnvhe1czSaZAQ5pmI//VNgxnQJpSbPtvEot2qMr3JoGPPs+eyLa2IFxfuI8DbwL8v6k5csM8/+BM4c1lycjl67bWYU1IIvvxyYp55+k/fk/n445R8o2riRNx1J+G33ILdbObAkKHYy9VUsv/YMQRNnIjfkCHkzHgJ8+HDhE6fTuC545rycYQQf8Ruh88vhpRlagRqytfQdpQanXklSQVbde4/CP61wVXOHtjwHtRUwK65jV+737UQ1AqWPgvewRDWFrJ3QddJMOndhhXjm0BTf/8+FXWymtI/rpMVFBTErbfeyq233sq2bdv46KOPTsR9nVLFlWbyy2toG+5PlcXGz7uy6RobxJJ7R3Lr7M3OfCn3AOu6oYncfXYHuj/9KwDmSgsfrznC/NuGkhDmB0D7qADaRwVgsztoF+nHwVxXsPLaZb2Y1CcOu93BzSOT2J1ZwpXvr6Os2srANqH0qq2tFehtJNDbyOqDbv9QaxVUmLHa7Bj0OjpGB3gcq7HamPzBejYfKSIq0MQ3Nw+hVagv4FnMNLO4mvIaKw4HfLI2ld8PuH5qsjvwCLC8DTq2pxUxoE0oJoPrpwsvgw6dBrfN3kp27UjX0z/s5oNpZ+5Q9j9RNGcO5kOHACj+6itCpkzGu0OHP3xPzHPPEThuHJrJhN8AtZLGYTZjr3B9/XW+fs7pvtgXnm+iuxdCHBedDqbMhfRNEBQHwbUjzZqmmkGnLHOdm7kFOtT+UBTVBSa+AUfXeQZZpiBXBfe9C+CBg6p1j954xtXB+isCzzmHgLFjT2rF96Z0Qiu+9+rVizfffPNEXvKk25FezJQP1lNWY6VrbCAlVRbSi1Qu1PMXdSM+xLdBUjpAakElep2GUa95rOTbfKSIxDBfbvpsMwUVZh47rxOjO0U1WO1nttv5bG0qz/24lxA/I/+b3p8VD4zm193ZmIx63l52kKV7cxmSFMbD4zsxsWcsn65N5VBeBQadht3h4KFzOzoLozocDkqrrQT5GCmuNDN/awaba/OuckpreG9lCv+e1A2A87pHs2B7Jg4g1M9IRY2Vp37Y7Wx47eelp8KtN2Odaqud5xfuY0znKB6b0JmyaguFFWYeGNcJTdOoNLtG4qrqvT+/Kp8HVjzA0dKjXN/jeq7qdNXx/lGdMQxhbqsvjUb0QX+ex6ZpGv7DPTsI6P39ibjnHvLefBNjZCThN/9Jc1khxKmhN0JCI4UsB9zkCrI0vSpCClBZCD/cAcVHYNi9MPAW2Pk1xA+AwFjY9D91Xl0l9yZaOXi60PT6JinTcCoc13ThmDFj/vyCmsbSpUv/8JzTebrwse92Mnv90UaPnds1mmcndeWFn/awcGcOZptn/tT3tw1la1oRT//gWqUXFWjC5nCQX6ZWIOo1+OH2Ydz0+WZn8OZt1LHx0bPo/dxiZ2K6BhjqBWx1njy/M9cOa4vVZqegwkyYrxc2HJgMeqw2Oy/+vJevNqZTXmNlQJtQUvLKndXe63gbdKx9ZCwhtQn5I176jaO199MjPoiyaqtz1Mq9WnxjFt09osHIGcDCnVk8+f0uAryNvDW5N11jXcHFjA0zmL13du2zaiy7fBlhPmHH/IwzmcNqJe+NN6nen0zIZZcRcNZZ/+x6dnuD1YZCiDPEti9UonrnC6B97f8FPz8M62eq13ovNVrlXfv/pd0Ou78Faw10vwwMXo1f9yQ5nb9/n46OayRr+fLlJCQkMGHCBIzGhkUsm4MOUQ2DhTpjOkcSGeBN34RQ5m/z7BVl0GlEBXpzXvcYnv9przM4cq9fBWBzwMS3fsfuUMFVsK8X/72sJ/7eBoJ8jBTU1ttyQKMBFqjGz9cOU+18ogLVTzRz1qQyZ8NRaqw2Due72uscq69htdVOblkNn687wqI92c4AC1QSe7Cv68932uBEFu7MYl+2q8Jw+yh/iistXDWgdaMBFsB53WM4r3vjq9yMOtf19ZoevXZmDgX/FZrBQOR9956460mAJcTpb/8iVQ+r60UQFA9fT1fFSTtNgIvf95zqq0uOB1W81O428q/TeZaBEGeU4wqyXnrpJT7++GPmzp3LlClTuPbaa+nWrVtT3dspMW1wAnqdxrzN6WxNKwZgRIdw7hzT3ll0MyW//lJbsNodzN2cRnSgNw6HShKvX4kdcLbRAai22Fl+/yiMeh1Pfr8bX5Mef5PqQfhHw4sVNVaP7deX7Of1JQcaPVerLRLsAIw6DZ1Oo8ZqZ0T7cPLLa/jv4oYrQjVUbhaoxtd3jm3PuV2juPGzzaQWVNI+0p+5Nw0m2NeLeZvTuXX2Zoa3j+CqAa0bXKu+xXtyePqH3fiZ+tK/XSFF9j1c2+1agr2D//S9QghxRtjzA3x9tXq9biaMehj2/6y2d32jgqaO49V2ynJIW+9674AbwFcKPDcXxxVkPfDAAzzwwAOsXbuWjz76iKFDh9KxY0euvfZaJk+e3CyGDjVNY+qgBKYOSmBjaiFmq52h7TzbE1zZvzVfb0qjosYzz2jB9kwyi6s9alHVuf/sDozuHMnSvbm8WhvYdIjyx9uo57ut6Xy27ojz3GuHJvDp2qPYHQ7C/LzIqzdVFxfi67G9PqXx0ar4EB8eOrcTYf5erE8pZGNqAWsOqXOrLbYG050ACWG++Hnp2ZOlRq30Oo1dGSVc+u4aqi12WoX48Nykbny+7gg/bMtkf67KT1u4M5vWob4NvlYNvg5ztztb+8QEXch310rCthCimcnY5HpdmQ+WejUOTW6j/xX1SjLEN49cJKH8rXmHwYMH88EHH5CVlcVtt93GRx99RGxsLKWlpX/+5jNI/8TQRoOGjtEBPD6hYcX73q1DnLWwQI0IAfgY9VzUN56usUHcMaYdL1/agwfGdXQ2ebZYPYOyIUkRBPt6YXfQIMAC1bpnyZ4cluzJweFwML57dINzLu/Xit8fGsPEnrEMSQrnnrM7UFjhqtN0uKCSke0juLxfvOcbHQ4Sw/3wMeqJ8PfipUt68NXGNGeB1LSiKqZ+uJ7//LrfGWDVySqpbnAf9Rn1mttrmfYSQjRDXS5UjZ8BYnvDwJtgxIPQaiCc9YwrgR1UBfek2nzntqPVtmg2/tHqwi1btrBixQr27t1Lt27dmlWe1r7sUmavO8qerFIiA0zcP64jCaG+LNmbS6C3gUv7xrP2UD6/7s5Bp9MY1i6MnekllFa5Apm60KnKYmPzkSLign3QNI3L+3m2o5nUO46VB/JYf7iQ8d2iaR/lT355w3L+kQEmZk7tw5Pf73YWR50yoBWTByVw/zkdeOu3A1TXBmzJ2aVUmq21qxv98DLoiAnyJjm7DDS4aURbdDqNly/tyXdbM5z5XxnF1RwpVD91mW124kN8GqwMbGykrktMIOd0ifrTr+v/XdWH5xfuIdDbyFMTW1ZrJiFECxHXF27fqIqPxvVV7XDGPAY81vBcgwmu/g5sVs8m0qJZOO4/0czMTGbNmsWsWbMoLS1l6tSprF+/vtn0MiyvsTJ73RHeWHqASrfg4lBeOe0i/Z2tdR46txNvXtXHefzx+Tv5dY9nq4O6vKwAk4HEUF9Kqy34eRl4c+kB9ueUERvszcKd2RSUm7l2WBvemqyuZ7c76BwTwN6sMo/rFVWasdocHtXn52/L5MtN6djsDgK9DVTXtlcZ1zWKS2euZU9WKSaDhsmgp7Q2Fywp3I8F2zPZmFrIixf3oGd8sLOtTpi/lzNZ32Z3cMV7az1GnHSAj0nvMVUa4e/FnqxSzn5tBbOvH0S7SP9jfn0HJ4Xx4x3Dj3lcCCGahaA49euvkgCrWTquP9XzzjuPZcuWcc455/DKK68wYcIEDIbm9RfjulkbWd/Iirz8cjOpBa4g6rd9OdwyKsm57WdyfR0MOo37zunAiPYR7Mos4XB+BRe+sxqjXsdFveP4amMa9b274hBX9G9Fm3A1xHwot2EtLovN4azMXsff20BFbVBUF0Sp66U4t2usDmqsrmN1bYC2p5cQ5p/Mfy7ryWtL9uNt0HPN0EQe+W4nezJLqbHayaw3BWgHfrxjGMnZ5Rj0UGm2cecX2wC1kvLj1Yd5/qLuCCGEEC3dcSXF/PLLL4SGhnL06FGeeeYZBgwYQJ8+fRr8OlM5HI5GGyXrNHj0vM6MaO/KzxrWzrPP1B1j2nNxnzj6JoTwzpQ+3DKqHV3jgriif2tmrzuKwwFmq50fjtHXz6DT8PVSS3otNjs6t9yuViGudjQHcsvpnxhCbLA3Q5PCeHpi1wZtnkAFXO75TyG+jU/lphdVMuHNVXy/LRODXqNTTCDf3TqUxfeMbPT8CH8ToHFut2jO6hzNqI6R+LsFmDFBzbtInhBCiOZtxowZaJrG3Xff/Y+vdVzDUE8++aSzD96ZzGZ38MyC3axPKWRCjxjurG2ovC+7DG+Dzlnd3NdLT6XZht2h+vG9Nbk3P+/KJsBk5Kx6+Uf+JgOvXt6rwWelFVZ6NH2usjRc0Rdg0vPiJT2cNa8enLfDmWhuMui4bVQSD3/nGsHamFrEQ+d2co6kDWkbxupDDdvsRAd6c/XgBDpFB9I3IZjnF+7jm81pmGvztnQarD6QT105rtnrjzJ1UGs+X3cUf28DD4zrwFcb0+kQFUB5jZV1KQXkldcw4c1VrHl4DMG+XgR6G/noX/35ZE0qieG+3DjCNbpXVF3EC+tfILcyl9t63caAGFk1I4QQ4o/Z7Q6yDhRTUVqDX6CJmPbBHgMPTWnjxo2899579OjR44Rc77iCrKeffvqEfOip9v22DD5dq0omJC8uI6ukihcv7sF/FiU7AyyDTvPIybrji620CffjixsGEX0cozXXfbLRo1SCv7ee8mp13cv6xrEjvZTknDK+25LBwDah5JbVsPpAnvP8Gqud7LIaAkx6ytzyoHakFztfH65Xt6vur2JuWQ2LducwqG0YfiYj1w5NZI5bNfv6+ev+JgN3fbHNuWqwbbgvb0/uQ/f4IHo+86vzvEqzjbTCKoJ9XZWHL+ody5hOUR7/EF7f8jq/pP6CpbQr67as5r5hoVw/POmk/WMRQghxZjm0NZdVXx2goti1+Msv2MTwK9qT1DuyST+7vLycKVOm8MEHH/Dvf//7hFzzuKYLQ0JCCA0NbfCrTZs2jBs3jsWLF5+Qm2pqZqvnaNIXG9LILav2yKsK9DbQJcaz7tfh/Ao+d6tnVV9qfgVv/XaAJXtynPvKqj0Lh9YFWAHeBnrGB5Oco5Lbl+7LZdQry5nw5u846o0W/rg90yPA0mlwUW9XQmWQj+dUoE5TKxtrrHY2HyliygfrqKix0ibc36OSe313jm1Hmlvl95T8Si54+3f255TSya2qu5+XniqLlUqzlXu+3Mrl763l+k83c/ZrKzyf1VyOpbQH1RlXU1PWgRd+Tua1JQ2Ln/6R3QW72VOw589PFEIIcUY7tDWXX97b5RFgAVQU1/DLe7s4tDX3GO88MW677TYmTJjAWf+w9Zm74xrJev311xvdX1xczObNmzn//PP55ptvmDjx9K7zcVGfOD5YleJMAPcy6DAZ9Dx+fmfMVjv5FTX0aR1CZKCJa7wTefTbnVhqh32iAk2NXjO/vIZL313j7PH3xhW9uLB3HI+e15m7vtyK3aHyoooqVYmHsmorTy/Y7XGNulG0gnIz47pGsXRvLla7g4N5niNV/3dVb87p6qqNlV3qmZxevxtPWY2Ng7ll9GwVwh1j2vPcjypo8TcZsNjs1Fjt+Bh1nNU5imqznVfdAiGHA37ckc0H0/txx5yt7M4sodJs4/L31hEd6O3x2YfyKjiQU0b72tZEHY1T+S4rxeNethxtmPN2LO9se4eZ21U/r9t73c5NPaUhshBCNEd2u4NVXzXeuaTO718foE3PiCaZDfnyyy/ZsmULGzduPKHXPa4ga/r06X94vFevXrz44ounfZBlMuj54fZhPLNgN6n5ldw4om3taJCRd6/uy/1zt/P+ShUcnN8jhnev7suXG9PoEhPI5IEJAGw9+v/t3Xd0VFXXwOHfnZbJpPdeIQQIhN57kaJUQUREsWDHhr29wmv7LK+9KwI2FBUEBJHeS4DQAoSE9N77ZDL1+2PITYYEBAELnmct15q5be6dgNmcs8/eFaxNKqRDkBvf7s0hIdNxReKD3x8ip0LPnOExDI31o1JvIqtMzy0LE+Q6U40Dap7OasZ0CuS7ZqsOY/3dWNdsRKyRQoJr4oMdtvWI8GLDiXNH+O9tSuPzWT25tX8ksQGu1BjMOKuV3LLI/geq3mRlyke7cFa37CE4PNaf1KIatqWW0Lyd+JnBnUTTqFql3shrqwuwWh2nVs+n9U6jFadWyK9Xpq0UQZYgCMIVqiC1ssUI1plqKxooSK0kJNbrkn52Tk4ODz74IOvXr0ervbSLty5p/YVx48ZdsnnMy83FScVrU7vI708V1/DD/lz83Z345XC+vD0xq4L3Z3RnRIemRPe8ynpmfLaXepNjkc4zvb0hlTnDY3DTqnHTqgnz1nFzvwi+2JnpcFzHYHd8XZ1QShKW01HMF7sy6R3pzd6Mck63HwTgqo4BfLD5FJV6I7cPjCbQQ0tuxRktG05TKyW5yGi1wURidgW3L9pHVb2Jp8a2Z1h7f7QqBYbT0V6F3kQFTcVUlQqJIe18aR/kylVvbnMIsMCet9YYMDqpFPxvWjz+p5P3rTawNjuhe7gnL0/uTPtmU7BVDVUcLjlMW8+2BLs6Bo4Anf06k1+XL78WBEEQrkyNpYgu1XEX4sCBAxQXFztUR7BYLGzbto3333+fhoYGlMqWAxDn45IGWQ0NDWg0mt8/8E9UpTdRoTcSebr+VGvqGsxM/3SPPNXXXM8Ib4b/bwsNJisvTe5El1BPUgtrfjfAgtbbxvRv4+sQZKkVcH2vMB787pDDcTabTa7XdeeQaIbE+GEwW9iSXMzrv50EYGtKCcPa+5Nc6Fi0tNGELsGsO1aEi5OKJ8e2592NqfJ05Utrknl34ykMZqtD0+rmLFYbm5JL+HhrukOullatYEbvcK6JD+ZobiVbU0qY1jOMsZ2D5GO8XTT8d2InPth8inBvHe/P6I6fW9NUa42xhhtW30BOTQ7OKme+GvsVsd6xDp//0sCX6ObfDQmJqe1EF3pBEIQrlctZUnH+6HEXYsSIERw9etRh26233kr79u154okn/nCABZc4yFqwYAFdu3a9lJe8KAeyypn1xT5qG8xc2z2k1RILYE9Yby3AUkiQVV5H+umcqPu+SaTOaMFFoyTa14X0M1b1zeofTnZZPdtTS3FSKXhvRjd538HsCuatPNaiT6HJCjUGk8O2CB9nssuagpqkvCqeGtuBr3ZnsqTZlGJqUS0pRS2LloI9ELqpbwSltUayy/XsyyiXGzM3qmmwB1Y1BjPD2/tzIr8aZ40SV62KI7lV8nHLDzbV9pKAt6Z1ZWznII7lV/HC6hNYrDa2ppTw3Z19CfZ0JtjDmXqThbhgd9Y+OBiPVpLtj5cdJ6fG/iz15nq2521vEWQ5KZ24scONrT6fIAiCcOUIivHExdPpnFOGrl72cg6XmpubG506dXLY5uLigo+PT4vtF+qCgqy5c+e2ur2qqorExERSUlLYtm3bRd3QpfT1nmxqTwcSyxLzePrqDvi62qPg8joj+ZX1xAa6seygY4FQL50KFyc1c4a15X/rT8rbGxPT64wWIrx13NgnnBdWn5D3t/VzY/6E1qe15i493KLUQqOi6gYeHBHDD/tzKKw2kFVWj0opYT491TeifQAVdUaeX3nMoeyCi5NKfr5GKoXEPUPbsOVkCdd9slueLnzl1+Szfk9qpcSTY9vz6bZ0fjyQC9iT4q02G1q1gqwyvXzs2M6BqJUKbDYbp4prsZy+IasNZi/eT7XBTJ9ob4qrDWSU6vFzc+Ln+wYQ4uns8JltPNvg4eRBVUMVCklBF78uCIIgCP9OCoXEoOtjWPtJ0lmPGTgt5h9XAuiCgqyDBw+2ut3d3Z2rrrqKZcuWERUVdUlu7FJo3kPP11VDalENtQYzidkV/GdFErUNFqJ8XZjczTEf6LYB0UzpEcqsLxIoqWka4WqeGxXuo+P63uH8mJjHiYJq/Nyc8Hc7e8Jc88bRzmqJepNNvmbnEA9GxQXipVMzb5V95Z/ZYuOeoW0Y2s6PPtE+VNWb7IVgm+U59Y7yZlNysXwdSbInlm84UdSi7+GZYgPcmNg1mAq9kWHt/Yn2daGoWSK70Wzl8TGxvNgsiARYc7SQNUcLublfBA+PbEekj47MMj1u2qbeiHvTmxYBlNQ0sPFEETf3i3S4jq+zL99c/Q3bc7cT5xtHN/9uCIIgCP9ebbr5M+auTi3qZLl6OTFw2uWvk9Xcli1bLsl1JJvtzHTmy6+6uhoPDw+qqqpwd3f//RP+IIvVxsKdGZwoqGZjcjGVepNDoNTIW6dmQIwve9LL6RXpxbzxccz/5TirjxQ4HHd9rzAsVhu+rk48OCIGZ42SBrOFZ5Yd5cdE+2jYvUPbcHXnIMK8dfJKO6vVRof/rJUrv/dv48P4+CCe+TkJqw1CvZz59cFBFFQZGPfudoynR59u6R/JvAlx8ufP+GwPu05XdldIsGrOQAxmKzabjW7hXqSV1HDNuzvk0atGPq5qymodpwrVSokuoZ4MjfUjo1TPsoO5RPq4UFJjQG+08Ow1Hcksq5OLtp4pxNOZnU8OR280k15Sx5e7M1m6P7fFcZIEz13TEYvVxui4QMJ9dOf4iQmCIAh/Z3/G7++/suL7pXZldXc+g1IhMXtQNF/uzuSn00FQaxFlud5Epd5EoLuWuGB3hv9vq8M0nEohsejWXgyMsfcrLKwykFOhp12AG04qpRz4ACzYkcGHW9LwddXw4939USokdqaVomz2B8RLp6GgyiBP/eVW1JNSVEOPCG/a+rtxvKAagB8P5MpBltFsZXd60+e4aVWMe38HTioFQR5a3r6+G9tTSxwCLGe1gn5tfHlnelceWHKQzSebKsmbLPY+jc17NWaU1tHWz4Vof1fSSmo5klvZalAK9kARQKdR0SnEg+m9w1sEWaM6BtApxJ3/nq7L9cm2NDbMHeJQKV4QBEEQmlMopEtepuGvckUHWY3igt1RSI5tZJzVSgwmCzbstZ22p5YCcDSvyuHcLqEezBkeIwdYa5MKuH/JQUwWGzP6hPPy5M70jvLm50P2UgONo1WltUaWJGSzJCFbbtYc4a0j0EPLU1e3J7eing+2pGGx2nB1UsrBUZSvixxkRfjoeG9jKuHeOkZ2DCDYw5m8SntCfFW9PQg0mKxklOp5evlRTp6xyrDeZGVTcjHTP93DsfxqefvZAieAUyV1LYqftqaNn+Nqze7hXjx7TQdeWn0CG/aRskdGxbLycFO+W2mtkcwyPV1FkCUIgiD8C1zR04XN7TxVSkJGOQoJtpws4WBOJf5uTtw3rC2nimv4ak92i3M8dWo2PTKUGoOJOd8eJK+ynso6I41NeSQJfntoMGFeOlYezmPRzkxONAt07hnSho+2psnvX5jUiZv6Rsjvv9ydyX9WNFV9d3dWYTRZ5bpVOo0CvdH+WqNU8MKkOHanlbEtpYRyveP0X3yoh8OKwOa8dWr5+MYAS6tWyE2oOf1erVDIKw7P5KRSODS6ViokPpjRnTGdAh2O25texs60Moa086VHhDeHciqZ/uluDCYrsQFu/HzfAJw1f3w5rCAIgvDX+St+f/+TXVDvwn+yPlHePDQyhum9wzmYUwnYGyjPX3UMjUpJ1Bl1tNy0Kn68uz/eLhpeW3uSo3lVlDcLsAAUksSot7Zxx5f7mdI9FE2zaukuGiU394/A63T5Ap1GSb9ob3n/koRsvtiR4fCZ1fVmOcAC5AALwGix8vn2DDx1GocAK9Ddid5R3rw5rSt3DY5GIdkDppjTSf86jZKb+0WiVtqnKxsjaoPJ6tCPcHK3UF6dGo9Oo8TFSSm3D+oU7IaTSoHRbKVbqId8vMVqY+n+pnIS8vcc7cPcq9rRI8L+rF3DPFn30GA+m9WFn+7tLwIsQRAE4V/jXzFduGhnBi+uPoGnTs37M7qh0yjRny7HYLXZ86jWPzyYaZ/slot13jk4Wl6d2BigNNe8UvqOU6W8uPoE9wyJ5qHvD2E0W3libHuCPJxZ/cAgEjLK6RrmKRdEPZJbyVPLjra4ZnNqpYSzumnFHkBaSS09IjwdjrtrSBtu6B2OVq3kqas78OTY9vZViMCmE0X8cqQAV62KbY8NI6+ynhs/3yuPSOk0Crlq+9aTxYyOCyBp3mgkCSRJwmi2cssXCTSY7aNzB3OrCPVylivM+7hoWLAjg8ExvnK/wjMV1BZw99bbyanJoeOpjpisJqI9o3m2z7N4aj1bPUcQBEEQrgRX/HShxWqj/XO/yjlPkT46Ood4sKrZykGNSkHC0yPQaVSsTSogzFtHt/CmpLvCKgOP/3SEwqp6gjyd8dCq6RTizstrHGtPDYv148Mbe2C2WnHTtizACfYRrP+tO9lq8dNGziqJ+tNFS5vnkmmUEk+O7SAnkjdyc1Lx1ew+dA3zlLdV6U0MfG2TXMl93viO3DIgim0pJXyzN8vev7FZ+6DGz/r1wcHENhvhmvLRTg5kVcrvP72pB4dyKrHabHy+PQOz1YaLRsnahwYT5m1fObgjtZQvdmYQ6eOC1v83vkr+osUzhruFs/ra1Wf9DgRBEIS/HzFdeGGu+JEspULCU6ehpMZecyOzTE9ms+KaSoXEhzO6yyveJnQNsR9XWkdOhZ5ekd4Eemj58rbeAGxOLuaxHw+TkFHOhPggVjYL1jafLGF/VjkWi40v92RxqriG2EA3XpzUmQB3LRV1Rp79OUku4Nn4+Y3vQz2dCfDQcqDZij+rzZ5vlVZci7eLBpvN5nAO2Cu33/v1AXY9NULeVlbX4NAq55cjBUztGcbgdn4MbufH9tSSFkGW1WYfLYsNdKO8toFv92XjrFbKgV63ME8W7Ehnb0YFLhql3Lewzmjhnq8PkFWmZ3SnQFYfKZDbDg2ND2n155Jdk43epEentgdmJ8pOoJAULaq+C4IgCMI/1RUfZNUYTER466jUG1vUjwL7SJeb1vFr2HmqlFsX7sNosdI93JOld/VDdboP4XMrkuRRqJVn1NECuGlBgsP77PJ6lNIxPr6ph30artm+W/pH4K5V8+6mU0gSjO0cxGfb0x3O16oUxAa4cSS3ijpjPa+vO8ljo9vxzoZU6pslrp/ZezDK14Vr4oPkWl/7syq4+6sDfD27DwAB7i0Lp7b1dyWnQs+Mz/awL7Pc4fvSaZTc1C+CuUsPA/bAqjH4cteqSMpvKjvRXLBze7q3v4MT5Sc4WX6Sknp7GYmhoUPlAOv9g+/zyZFPAJjbYy63drq1xb0JgiAIwj/NFR9kLdyZ6VALCsDdWe1Qgf3M1jRrjhZgtNgDmMTsSnIr6uV8Klen1r8yhWQflWotkKs8/VmeOg2vTonngy2niPDW0TfaB7PVxur7B+KqVVFQZWgRZOmclJTWNlW+tVhtvHtGgAVwVccAHv7+EL0ivZnRJxxJknj/hm6kl9TK1d8Pn074B6ioc5yunD0wimHt/bjxc8cgsZHeaGlxjreLhtemxpNdppcr1QNc2z2E5QfzCPVy5q7BbYjwiZf3pZSnUNFQQc+AnvK25aeWy69/PvWzCLIEQRCEK8IVv7owt1zv8N7NSclLk+IY3t4frVrBhC7BDI11LNXfI6IpHyvE05lAj6ZRn7endyXijKrlQe5ajs4bRecQD1qja7aibkqPUDY9MpSBMX7c/XUic749yH9/OU6D2UrfaB/uGhyNn5sG1enipeV1JgqqDHQJ9SDA3Yn/jOuI3uRYSgFg2cE8lh/M4+nlR9l4ooh9meXMWXLQoSDWlB6h8utekd6M72JvJxQf6sH9w2Ooa7Cc/YsEXlh9gkgfHRKglCSeHx/H8PYB3NQvktEdA/DSqRkTF8Cb07qS8uJYtj8+nAgfx1Wb7bzb0SeoD0pF03cS5xPX6utGB4sP8vaBt9mRt+Oc93c52Gw2yurLMFtbL20hCIIg/LPNmzcPSZIc/mvfvv0lufYVP5LVO8qbpc2msGoaLPzfryfZ+eTws55zbfdQnDVKvtyVhVajICmvip6R9pIE7QPdeebqDtz51YFm1zSz61QZn97ckyd/OsKGE8UO18uv0PPm+hR0GiW39I9Eq1ayObnpmL0Z5Yx6axudQzyY0Sec0hqjQ7FQpUJixZyB8vtqg5k3153E3Vktr4Zsbl9mOZ9tT+f0YBwKCV6/rgtTuoeSXlLLS6tPIEkSz43rwFvTushToY1tgM4ls0xP4nMj0WlUaE+XrMivrGfzyRKMFitrjxWx8nA+E7oEt3r+huNFPLnsKFq1gvdu6Ea3cC9eHfwqS08uRSkpmRY7zeH4jKoMbv/tdkxWE18kfcGiMYvoHtD9d+/zUjBZTdy/8X525u8k0j2ShWMW4uvs+6d8tiAIwr+V1Woh78QxaisrcPX0IqRDHArF5S3/ExcXx4YNG+T3KtWlCY+u+CBrfNdgVhzOlyu6g708wm/HCtmdVsaw9v4MaefX4rzkghq5jc3+jAp2PTkct9NByKi4QP7v2s48tfwoNpt9unHOkoOcfHEs03uFtwiySuuMvLsxFbAHGUvv6seQdn7sOFXqcNzRvCpeWXPCIcDSqhW8NLmzw3F3DY4mIaOcrSklnCnCW4feaJYDLLDnTWlV9j+gty/eT0apvaJ7WV0Dy+8dIB9XUGVwuFZjmYrmFeIjfXR46TRymYjG84zNPjDnjNHD5p75+ag8/fni6hP8dE9/nFXOzIqb1erxGVUZmKz2QNKGjZMVJ/+0IOtQ8SF25u8EILM6k1Vpq8RUpiAIwmWUuncXmxZ9Sm150+9HV29fht9yJzF9+l+2z1WpVAQGBv7+gRd63Ut+xb+Zz7dnsD+znEB3J7xdnFCrFFzXI5S7vz6AzQZf78nilwcG0j7QcSnqoWb5SzUNZuLnr6OtvysNZivD2/szq384zYtfWE+/GdkxgNFxAfx2rEjeV9csKX1/VgWfbEvnnqFtaBfoxnsbUx1yxhpHlRrdMSiarmGemC1WvtmbTaXeROdQD4cAK9rXhWk9wwj3duaB7w7y5W7H6vXRvi5IEhzPryKztKllTkZpHQezK3h+5TFUConnxnWkfaAbyYU1qBQSKqXEpE7BPDIqls0niymqNjAuPpjaBrNDiYru4Z4MivFle2opoV7OXNvdcUXhvsJ9fHn8S8LcwnBW95K3686jMGnPwJ5EeUSRUZWBv7M/w8KG/e45l4qfsx8qSYXZZv/5BbkG/WmfLQiC8G+TuncXK998ucX22vJSVr75MhPmPn3ZAq3U1FSCg4PRarX069ePV155hfDw8Iu+7hVdJ6u42kDvlzfK71UKiV8eGEhiViVPL28qBvrRjd0Z2zmI4moDB3Mq2Ztexhc7M3/3+o2tZpQKif9c05FrugThrFby3M9JLDvY1LPPU6emstm0XpcwD8rrjET6uPDalHieW5HExhPFaFQK+kX7oFUr2Z9VTmmtEV9XDUvu6MvS/Tl8tt1eIb59oBvpJXXy6NE707sysWsIz69IYvHuLId7vH94GzYll3AsvxqVApoVlGdIjB85lXrST/cq7BnhxXd39uXjrWm8sS5FPu77O/vSJ9qHBTsyeHH1cdRKBe9O7+bQUsdms1FU3YC3iwaNqilQ1Jv0DFs6DL3ZPro1MXQOKafiUSktjOujZ1BkB8Lcws75Pdeb60mvTCfcPRw3TetFTy+XLTlb+DXjV+L94rmxw41/6mcLgiD83Vyu399Wq4XP7rvdYQTrTG4+vsx+f8Elnzr89ddfqa2tJTY2loKCAubPn09eXh5JSUm4uV3c75wrZiSrUm/k1bXJ1BjMPDSyHW39XXFSKeWK5gBmq43NySVc1zOUT7alkVWmp32gGwNjfDmaW8nMBQlU1ZtQtCzw3qoGs5UbeofxyrXxvP5bMj1f3ICLRkmQh/MZ99YUYGmUEkdyq7DZIKe8nqve2sqHN/ZgTFwgj/54hC0pJQ59AktrjXyzN9uh+fOp4lpenBTHS2uSMVusnMivxmqz0amVxHsfFye5OXRjgKWUwMfViXkTO3LLwn3ysWarDZVSgf8Z5R2cTudevbsxFZsNjGYrH21NcwiyJElyWCDQSG/WywEWgMq5gI9nXcd1q67jlcRi3j7izDdXf0OMV8xZv2dnlTNxvi0T4v8MQ8OGMjRs6F/y2YIgCP8WeSeOnTPAAqgpKyXvxDHC4uLPedyFGjt2rPw6Pj6ePn36EBERwdKlS7n99tsv6tpXzOrC/6w4xpKEHH45UsCdX+4HwEOn5uVrO8sr8BSSfeWgr6sTvz00mA1zB7NizgASsyuZ9OEuqk6XWrC2MrYX6O7EwLa+BHtq0TSb0ov0ccFotvLhFnsj6DqjhVMlta3e4039Ilg1ZyCKZvlMtQ0Wnlp2lJzTrWrAHrzpmvVBjPTRcX2vMPk5pvcOY3tqGTUGM/UmKx9vS+fh7w+zYEcGC2/pyfD2/sSHePDgiBim9Qoj+Izgx2Kz921cfjCf/7s2nihfF9oFuPLCxE4ATOkeyqx+EcQFu/Pk2PZyJfkw76bgMdzbcYXl2fg6+3J7p9uRkAjQBVBjrGHOxjkU6+15a/Xmenbn7z6vawmCIAhXptrKit8/6AKOuxienp60a9eOU6dOXfS1rpiRrKLqpqTtwmavp/UMo1ekN5uSi+ka5iE3LtaqlbT1tw8DfpeQ7VBBXaUAT2c1pXX2oCs+1IOf7umP+nRwdTS3iq/3ZBHl58LsQdEoFRJB7lryz0gcb06rVvDY6FjctWpentyJ/6w4Jo9WqZUSXUIdR6Fu6hdBaU0DB7IreOGX4/Rv68vK+wZgAzqFeDDn28QWn5FcWEOIlzNf3NLLYftP9/Zn+id7yDojIf2nAzks3pVJ/zY+3NgnQm4KrVRIzD8dcDX3+pQuvL/5FIHuTjwwst1Zn/VMD/V4iHu73suLe16Ua2JJSNiwoVKo/rREdkEQBOHvydXT6/cPuoDjLkZtbS1paWncdNNNF32tK2Yk68GRMbg5qVAqJJ66uoO8fX9mObcv2sc3e7I4W/ZZ8159Csk+rVZlMDO6YwB3DYnm2zv6ygEWQOdQD16dGs/dQ9rIo0tfze7D9T3D6BDUdK1mA1b4uDjhfjpZfGisP7Zmawgr9EZuW7zf4Z6yyuroGeVNZpkeiw22p5Zy7Ue75KnPBlPrNa1Ka1r2RAzycGb5fQO4Y1A0kT720Si1UiKv0kBVvYlfkwqZuWAvI/63leTCapYkZNPzxQ1M/GAneZX18vc49eNdrD5aQHGt8bzKPTSnUWooN5TL7xUoeKbPM3x79bd08m0Z0AmCIAj/HiEd4nD1PneJHDcfX0I6XPrUkUcffZStW7eSmZnJrl27mDx5MkqlkhtuuOGir33FjGT1b+PLwf9chdlqk+s3ATz7cxLpp1fUPb3sKOvmDmlx7v3DY/DSaTicUyknrJssNgprDHxyc88WxwOU1jawObmY/EoDozsF0D7Qnbmj2pFbruejLWkgwcRuIby3MZUGs5WXJjWVYfhmbzZGc1OQVVXfstDl8PYBuJ7R7qfBbOX2RQnMHtQGxwY9dtF+LnQJ8+RUcQ2vrT2JVq3k2Ws64O+uxU2r4lBOBZll9WhVCu4d1pY316c4nF/TYGbVoQI+3paGxWqjtLaBdzak8NrULixJyKHOaA/sVh3OJ8zLmV6R3gxr71jItcFs4Z6vE9meWsKI9gG8P6ObvGLyzvg7OVR8iGpjNX4ufqRUpDAlZkqr3++fbX3Wek5VnGJM1BiiPKL+6tsRBEH4V1EolAy/5c5WVxc2GjbrzstSLys3N5cbbriBsrIy/Pz8GDhwIHv27MHPr2V5pwt1xQRZYC9/cLocFBV1RvIq69EbmwKY5tOIzSkVErP6R5JfqXdYFYgNjuVXEezhjJeLvYH0sfwqvtiRwfLEPBoX6n2+PZ0HR8bw0uqmGldeOjX3D4/hgRExxAV7EOXbVPm8cVquNf5uTtw2MIppvewr7sbHB7GqWY/EsjoTr65NBsBDq8JstdE7ypvZg6LpEuaJ2WJjyke75fwyvdHM57N6kVlax75M+1y2wWylQm/kgRExrDtWSHpJLcbT7YBMVgvNh/waVwq2C3CVt0kSp3PQ0vh4Zg+HBPg1RwvYdLrQ6tpjhWw4UcSYTkFYrBY+PPwhVcYq+8+irpAfUn6gWF/MiPARTGw7EYV07oHVtMo0VqatJNojmoltJ57z2AuxNmMtj217DIAlyUtYNXkVHk6tV+8XBEEQLo+YPv2ZMPfpFnWy3Hx8GTbr8tXJ+u677y7LdeEKC7IaJRdWc/0ne6iqN+HebDSo2mDm9kX7eGxMLO0D3alrMFNYbSDCW4dKqSDYU8e0nmEs3Z+DVqXABlzz7g5cnZTcM7QNUb4uPPz9YTmXqlFNg5nFuzIdiohW6E3MXLCXGoMZlUKiU4gHE7oEc9vAKG7oFU5JTQMrDuXj7+aEn5sTm5KLcXNS8clNPdiUXMyMz/YwNNaPzSdbFhxtVGUwM6K9Pwua5WA9+N1BOcACKDvdb9DfXeuwanFHaimr7h/I3KvakVpUw2/HCtEbLXICv1KS6N/WhyHt/KioM3JL/0jUSolDOVWsPJwvX/9QTqVDkOWudZxGbHyfWpnKzrydLZ5ha+5WtuZuJb8un/u63sfnRz/nq+NfEe0RzZtD36TaWE1CYQLtPNsxZ9McKhsqAbDarEyOmXzW7+ZCHCs7Jr+uaKigsK5QBFmCIAh/gZg+/WnTq8+fXvH9crkig6wVh/LlQKPa4DgVtzG5mJNFNXwzuw9TP95NSU0DfaK8+er2PmhUCl6bGk+3ME9+OZLPzjR7xffaBguv/5bS4nMatQ90I8JH57BCEKDm9GebrTYO5VRyKKeSDkHu9GvjQ7SfKxmldWSU1skV1fVGC3OWHCT39HV2nf78Rjf3C2dJQo5DE+rcinqMZisZpXWEeDm3qNr+2OhYAHanlToEh6nFtaw/XsT4LsHEBLgRE+DG//2aLO+32GykFtVyx5cHcHVSoW8wIykkFBK4O6uorjfjrFYytpNjhdwRHQJ45Kp2bE8tZXgHf/q3tc+x+zn74aJ2oc5kn7p1UjjRYG1qfH209Ci5Nbm8k/gOAOWGct5LfI81GWuoM9ehUWgwWpvyzVIqzv7zuFBjIsew9ORS9GY9Xfy6EO0RfcmuLQiCIFwYhUJ5ycs0/FWuqCCrtsHMuxtTScqrkre5OSl5+KpYFu7KIKfcHrwUVNYzb+UxSmrsv+T3ZpRzJLeSnpHeJOVV8VSzQqW/Z2QHf7LK9Kw/XmSvzG61kpRXfdbjE7MrqNAbSS5oOqb5CFjzUSiAMC9ncirq8XBWc3O/KEwWG0sScgB7VtbdQ9pw3Se7OZxTSYC7E0+N7cCxvCr0RgvPj+9I/zb2IKd5nlojX1fHacvxXYL4Zk8WNQ1mon1d5Fy22obTgarVhgUw1ZuZe1UMU3uEEezpzJnuHxHD/SMc6175OPvw2VWfsTJtJe2823Fdu+t4avtT/JL+CwpJwYToCagUKhSSAqvNHgyuz15Pndl+D0arkWiPaNKr0nFVuzIuetxZv+MzGcwGak21Z+07GOcbx6rJq8ivzaejT0fUytaT+gvrCpm3ex41DTU83ONhega2nq8nCIIgCHCFVXyf+/0hOacqxNOZcV2CmNglhI7B7uzLLOe2Rfvk0aUz/WdcB97ddAqVQqK01nGFXvNptkadgt0ZGuuPwWTh8x0Z8vbreoTyQ7OG1M21C3AlpcheQ8tNq6KuwYzVBjH+LqQWN7W7ifDRkVOuZ3yXYF6a1IkThTVE+brg6+qExWrj54N5VOqNTOoWwomCGmYu2Cuf+9TY9tw2MAqzxYbzGW1r3lx3ku/25VDXYCYmwI1PbupBwBmFR0tqGsit0KPTqJjw/o4Wz91oyR196dfGp9V958tms3G45DCeTp5EekQCsDx1Od8mf4tOpSOxuKlMhZPSiaXjltJgaSDAJQBvrfd5fcaxsmPctf4uqhqqmB47nWf6PvOH7/ehzQ+xMdveQcBb683W67f+4WsJgiD8E/1ZHVuuFFfUSFbzOlBV9SaeGttUyqFXpDeLbunFlI9bL3z5v3Up8uq5M5ktVt64rguHcys5mF1B70gfnr2mAwqFZF9JeJpGee7E7UAPrRxkNQ/2bugdwX9/OS6/n9I9lPuHt5WbMPeK9JafyWSxMqVHqHxshI/FIQjMKK1jwY4MonxdeHNdCh46NW9M7UK4j457hrZl5eF8imsaOJRTydylh/hmdl+He/Q7nSMGsPSufixLzMNstbIkIVsu0jo6LoB6k5mUohraBTSVrEivSuf1fa+jkBQ80esJwt3P3fdJkiS6+nd12DY5ZjKTYyaTVJrEjNUz5FIXz/V9jmjPC5/G+/bEt1Q12Ec2vzv5Hfd0vee8A7QzGSxNU7FGS8tSGYIgCILQ3BUVZN0zpA33fZuI0WLlgRFt5e0Pf3+QvenljI4LaNFHEOxBzY5TJXKQdeYxvaK8mdojlKnNgptGKoWEh7MKtdLeePrLM3oHNpraI5T2gW5sS2nZNqCuwcQ18UGsPlJAtJ8L03uFyQHWh1tO8cGmU3i5aCiubsBktfLU2PbcObgNAGHeOr6e3YdfjxaSlF/Fd/ty5PtqrKn14urjfHpzT25fvI/MsqZAdNepMto8vYarOjiWWmiUX1nP13uzsFhtdAvzJMJHR4cgd9YmFXDbov0oFRKf3dyD4e0DAHh6+9NyEnmtsZbFYxe3+l2cj06+nXh50MtsydlC78Def3g1YYhrU7NqLycvXNQu5zj63B7p8QiFtYVUG6t5qs9Tf/g6giAIwr/DFRVkjewYwP5nR2I0W/E5nW/00urjLD9oXw23cFcWfaK82JthL2Uwor0/z4+PI9xHR2J2BS+tPoFWreDVa+MprjFwNK8KhSRxbfeWwRXYR41eWnNCfv/FzswW02t+bk58PLM7YV46xr+/A6BFo+Y2/q7cP6Idr04xo1Up5GT94hoDr609CUCdsSmp/r2Np+QgC+wjXb0ivRn91jZ5W/MK9labDZvNxt6MpmKgYM8Fs1htrD1WyPrjRYztHOSwf+XhfPk6B3MqWXx7b6rrTbxyOkHeYrXxW1KRHGTVmpraCVUb7TlnpfWl/JL2C8GuwYyKHNXq99iaEn0Jg0IGnTP3ak/BHo6VHmNY+LCzJqvfEX8HkiSRV5PHzI4zcVKevXzG74nxiuHnST//4fMFQRCEf5crKsgCcDujhMDB7EqH93szKnhyTCyltUZ6RXkT7mPvwScBKYU11DSYeWnNcToGeeDpomFG73C5qvuZjuQ6Xrt5YNMr0otvZvdBrVQgSRLLD+ZSVG1PtG8eYGmUCtr4ulFtMGEwWZjw3h7SS+sY3t6f16fGo1EpMJ4RuDXec3OFVQZ6R3lxssjeSNrbVYOLRoWXTs3TV3dAkiRGxwWw5mhhq8+y+WQJtQ1mpvYIRZIkssrqSMxu6hEV4++Km5MKrUpJlK8LGaeT4ntHNU29Pd7rcZ7e8TQKFDzW6zHMVjO3rL2FrGr76N6w9GHc1/U+Yr1jW72HRp8c/oT3D72PRqHhjSFvMCx8WItjduXt4u4Nd2PDxhdJX7B8wnJWpa8iuyab62Ovp6NPRwDUCjX3dLnnnJ8nCIIgCJfDFZX43ppv92bz9BmrBdv4uZBWYg8SHhoRQ35VPb8dK2qxsg/grsHRDm16GqWV1DL27e0YLfYAKNhTS36lPWfH11XD0rv6Ee3XVMAzpaiGce/uwGixopDsTaidVApGdvBnTVIhaoWCa+ID5VE3gDendeGDzafIrzTQPdyTdoFumCxW7h3almBPZ8wWKw9+d4i1SYVYWvkx6jRKHhkVy+0D7RXMLVYb21JK8NCpScgo592NqSgkCa1aISf73zkomut7h/HOhlS5HpZCgp1PDCfo9ErC0toGFu89SrnlJFO6dGi192BaZRonyk7w1A7HaTVnlTPLJiwj1K310UGAnl/3pMFiD0h7BPRg0ZhFLY75+PDHfHDoA/n9de2u44eUHwDQKDRoVVo6+HTgzaFv4q4RyZmCIAiXgkh8vzBX3EjWmWb0CeejLafkGlauTioyS5tW8i3Zly2PMLXm+305PDAiBhcnx6/qVHGtHGABFFU1XcNosbL+eBEFVVnM6h9JlK8L7QLcWHp3PzYnF/POxlTA3iZn/fFibDb7OY0V2cGeU/XNnmw5GMws0/PNHY5J6ltOlrD6aAFnozdaeOGX4+gbzBRUGegV6cXk01Of3cO9uHtIG0wWKzHP/Cqf8/mOdD7dnk5Es9EyrVopJ8MDmKUKfiy6n1pTLb8UKlg4eqFDoLU9dzsPbHoAs82Mq9rVYRqx3lxPSkXKOYOsMLcwTlXau5+Hu7WePD8sbBgLkxaiN+uJ9oh2SEQ3Wo0YjUb2FuxlyYkl3NXlrrN+liAIgiBcLldMg+hzWXp3P27pH8FtAyJZ9/Bg/NyayhacK8ACqKw3ycnkAFarjR8P5JJRWkf06VY5Xjq1w0hSjcHMK78ms2hXJmPe3sqqw/lU6o10DfNkzvC2eOqapjRVyqapSIPJwnPjOjK+SzDvzejm0AbIbLVSUWfk9kX76PfKRmYv3kdxTct7jwlwpWOQ478u/rc+hW8Tsnl46WE+2JzqsE+tVDCxa3DT851+jKwyPR2D3HFSKWjj70pls1G+U5Wn5MDJarNytNRxpHBj9kbMNnteWa2plkd7PIqr2j6qF+wSTDf/bi3uu7kPR3zI9bHXM7vzbJ7s/aT9uzEbmLdrHjetuYm1GWuJ9Y5lxaQVfHrVp3x7zbfM7DgTTyfPFtfSqrQttp3N3oK9DPl+CIO+G8S23G2/f4IgCIJwRcjLy2PmzJn4+Pjg7OxM586d2b9//0Vf94qfLjzTW+tT5JGk1kT7uaCQJE4VN42+tPVzxcVJydNXd2DDiSI+226vi9W/jQ9PjGmPi0bF6Le3YjnHNxni6cyq+wfi7aLhUE4lH29JI9BDy8HsCg7n2ksMKCU49fLVSJLE//2azMdbm1rcfHFLTzafLGHRrkz5mhqVghBPLRml9hWDaqVEwtMjUCoV3PdNIttTW65kbB/oxtqHBmO12jBbbWhUCqxWGweyK1h9pEC+vr+bExV6o1xd/sY+4bw0uTN6k54HNj1AQmECNmy4qd34+pqvHRLPV6Wt4ukdTwP2UamfJ/5MVUMVqRWpxPnG/W7LmnpzPeuz1uOj9WFAyACgKU8LQCWpWDd1HX46e/POxccW80PKD7T1bMtD3R8ioTCBH1N+pL13e57t+ywapeacn9doysopciX5UNdQfp3y6++cIQiC8O/yZ/z+tlltNGRUYa0xonDT4BTlgXSW3OhLoaKigm7dujFs2DDuuece/Pz8SE1NpU2bNrRp0+b3L3AOV/x04ZlMltaLazYaHOPHvAlxLNqZwfKDeZTVGTlVYg+45iw5SKhXU4XzQzmVdAnzBCDaz5XUZoHZmfIq69mXWc7ouEC6hnny8U09AFi0M0MOskZ3CuRAVgUvrTlBbnnTakKLzUaPSG82nCh2uKbRbCWrWUkGpULCw1mDQiHx+Oj27MvchcHk+Lx9o33Yl1nO7MX7qW0w8/z4jtzcL5Jekd50D/cizFvHycIa9meUO4yUWU/H4j+k/MDewqbip68MeqXFyr7xbcbjrfUmszqT0ZGj0Sg1+On85KDoTF8kfcFHhz4ixDWE90e8z3M7n2N/kf1fEI/3epybOt4kt+MBMNvMcs2q9Kp03tj/BgBZ1Vl09u3M7Z1vZ1rstLP+LM6mcbTtzNeCIAjCn6M+qZTKVWlYqppSQJQeGjzHt8G5U+tdOy7Wq6++SlhYGAsXLpS3RUVFXZJrX3HThWW1DXyw+RRL9+XQ2iDdnYOjGdjWF63a8dEjfXSM7ODPg6fbwdwyIIoVcwYS5dtUV8lotjKhS9PUWvPXnUMdR2cUkn1kqZFOo2Tpvhx6vLCep5Ydke/tlgFR/HRPPxbM6sl7N3Tn3m8SOZhdSUltA6rTkfvsgVG4Oqm4f0RbuoZ5IDUL6K028HPVEOyh5Y3ruqBQSCzamcH493e0CLB6RXgxb0Ic72xIparehMVq46XVTSUolAqJ2wdG4aRSkF7WFNR0DfPkoZHtANAqHaffglwdyz40GhAygBs73HjWVjaNKg2VvH3gbQwWA2lVaXx65FM5wALYnW8vHntz3M3EecehUqjo4N0BvckeXKaWO45K2jj/gVmbzUaxvlj+WUS4R8j7vJy9zvs6giAIwsWrTyql7OsTDgEWgKXKSNnXJ6hPajk7cymsXLmSnj17ct111+Hv70+3bt347LPPLsm1r7iRrBs/30tyob2MQUGVgQdHOvbQ89Rp+Hp2H348kMujPxwGIDbAjdUPDGxRjBPgmWs68MCSg9QYzMyfEMeouEB6RnhT22CmwWzh8+3pXN05iJl9wlmWmCefZ7XBa9d2JsrXlYSMMpLyqlh9unzCkoQctpws4cvbehPl68KXu7PYkVrK8Pb+1BubKsFr1UqMZqscNvi7afn5voFkl+sZ9+52uZ6Ws0ZFbYOZijr7H8wfE1tv63PnEPuwZ/MkdlcnFUazFY2q6dmbJ/lrVAoW39obj9N5ZNe2u5aUihSOlh5lYtuJtPNqd7YfBXqTnud2PkdaZRozOsxwGF1KLErk0yOf4q31RqPQyM2i3TXu9A/uz678XQAMCR0CgK+zL6OjRnPswDFOlJ9g9rrZfDjiQ4fVi30C+zA9djo2m00u5gr20a4HNz1ISX0Jc3vMZVrsNKqN1Vz909VUGatwUjrx04SfOFHeFHAeL2uqwC8IgiBcXjarjcpVaec8pnJVOtqOPpd86jA9PZ2PPvqIuXPn8vTTT7Nv3z4eeOABNBoNs2bNuqhrX1FBlsFkkQMsaFnHqrmpPUIJdNeSU6Hn6k5BDgFWld7EA98d5EBWOZ7OGqb2DOGuwW2p0BuZ/uluSmoaGNEhgE+3pQPw8da0Vvv4bU4uZmlNLjZgX6ZjIdCCKgOvrj3JhK7BrDhkL5Xww4Fc7hkSzY+JeZgtVipOV51fsCODaT3D0GmUvLk+BSeVgkW39WbXqVL2ppez/ZQ9un9+5TEmdA0hPtSzRZNqSYIgD/so1PPjO5JTrmd/VgVldUa6v7CebY8Pw9vFnrs0Z3hbimsMZJXpuXNwNDsL11OsL2Zi24l4ab14rt9z5/PjYPHxxazLWgfAi3teZGDIQIJdgzFZTczZOIcak/1nNThkMMfLj9NgbsBJ6cR7w99jS84WfJx96BHQQ75eY70tgMqGSg4WH8RkbUrIb+PZhuE/DEclqXhjyBv0D+kPwIeHPiSzOhOAV/a+wqS2k1ictJgqo32atsHSwNuJbzMifATJ5fZCqyPCR5zXMwqCIAgXryGjqsUI1pksVQ00ZFShbeN5ST/barXSs2dPXn75ZQC6detGUlISH3/8sQiymtOqlUzoEszKw/koFS0rtX+XkM2HW9KI8NHxzvRuDIxpfSrr7Y0pbE0pAaC2oZ63N5ziQFYl3i4a9qTbg6XMsnT5+NJaI6sOO5ZSUIA8cnU2SoXEW+tOOmzr28aXJ8Z24PPt6bx4eipPqZDQaZTc+00iR/Oq5M/8fFZPSmuPyUGWWqlApZCYPyGODkHuGIwWlAqJA9kVjOoYQKcQ+5Smp07jMOVY22Bm4c4MHhllLxLq6qTizWldAViUtIj/bf8fACvSVrBswjIkSeLbE9/yQ8oPxHjFML//fJxVztQYa6gz1RHoEgiAxdrUC9KGTX5fVFckB1gA5YZySuvtz/Dp0U8ZFDqo1erw18Vex/qs9VQbqxkXPY7RkaNZeGwhpfWl+Dr7siFrA/Vmey7bW4lvyUFW8/wqrUqLUlIS5h7mcO1Q11Du7nI3PQJ60GBpYEDwgHP85ARBEIRLyVpzfv1gz/e4CxEUFETHjh0dtnXo0IGffvrpoq99RQVZAO9M78qs/pF4u2gc8qnK64w8tewoNiC7XM8LvxwntbiGouoGnhjT3qEv4ZkV1gESMuxJ640UkoRCibz6rlFjkHe29PrxXYLIrajHW6fBWa0go1nielyQG0Pa2ZPDb+4XSU65nhMFNczoE06Yt86hpENhtT2YeHhkO4prDORW1HPfsLbyVN9NfSOo1Bv576rjGIwWup1O0G/UJ9rHoS5X4yjWmQ6VHJJfn6o8hd6sp6y+jP9L+D9s2DhVeYoA5wBya3PZmrsVk9XEzR1v5rFej3Fz3M0klSaRWpnKjR1uJMw9DJvNxn0b75Ov6aZ2Y1DIIJLKkuRtjaNTazPW8mvGr3Tx78JtnW4jzieOdVPXUdlQKfck/GnCTySXJ9PBuwN3rb+L4nr74gAvp6acqoe6P0SdqY5ifTH3db0PpULJpLaTOFJyhC05W+js25kHuz8IQK/AXmf5yQmCIAiXi8Lt/FaBn+9xF2LAgAGcPOk44JGSkkJERMRZzjh//5oSDkXVBvq8vFF+7+mslms/qZUSSfNH46RSAvYWNTct2OuwWtBdq8JoseKiUeLipOLJsR3oGOTO3oxyXlx9nBqDGX83Jx4Z1Y4nfnKsGzWxazCTu4WgVEgMimlaYffgdwflqUKAaT1DeW1ql7M+w9L9OTy97CgqpcQHM7ozokPAOZ95yoe7OHC6NY6Hs5rDzzuODr285gQbTxQxoK0vz43riLqVnLR1met4fNvjWGwWhoQO4f0R75Nelc7En5saNp9ZcBRg/8z9rfYJ1Jv09Pm2j/y+T2Af+gf3Z2f+TtKr0hkTOYYnej9BWmUa1668FqvNHq6+Pvh1xkSNOefzZlVn8daBt1ApVDza81F5RE0QBEG4NC7X72+b1UbhqwnnnDJUejgR+ESvS56TtW/fPvr378/8+fOZNm0aCQkJ3HHHHXz66afceOONF3XtK24k62ycNUqH9y5OKjnIUikUKJrNnwV6aFk/155wvTutlFlfJMhJ5gaTlZcmxzOmk/0XeKSvC4NifEnKq0KtVLDlZDFdQj1Iyq/GYrVxbfcQ/nddF4dE7EaPjoolIaOcwmoDkT46HhvdHkBO3M4srSMhs5yeEV5E+7kyrWcYE7oEI0nIASHYq8+rlRIRPi4O108tbpqSq6q390bUqpvOe/rqDjzdSsug5uL94vHSelFaX0p5fTkGs4Foj2jmdJ3D0pNLifaMZk/BHodzglyC0Cha/9eGTq1jTOQY1mauRSkpOVh8kL2Fe1FJKr66+is6+XYC7I2lGwMsgCJ90TnvE+yrA98e9vbvHmez2dhftB+rzcqBogMEuQQxOWby754nCIIgXB6SQsJzfBvKvj5x1mM8x0dflnpZvXr1Yvny5Tz11FP897//JSoqirfffvuiAyz4FwVZ7lo1swdG8fmODDyc1bw0uRNf7c6iqMbA3KvatTqKA1BWZ8R4xpSgk8rx2GBPZ1RKiaGvbUZ/umzC7QMjeWx0e4eg5kxh3jp2P9WUYJ1SVMOkD3ZSUtPAnYOjWbQrk9oGMy4aJavuH0i0n6vD9X5NKuD1tcmkl+qRsDdrHtHBn/Fdgglw0zIk1k/OFfN11ZzzXs7mt8zf5Hypo2VHSSxKpH9If+7qcpfcrmbernn8lPoTKoWKAcEDeKzXY60GlY1eHfwqMzvO5GT5SV7Y8wJgr311pOSIHGT1DOjJ0LChbMnZQoxXDBPbTGxxnXpzPbvydhHkGiQ3hD4fL+19ie9Pfu+wbXf+bl4b8tp5X0MQBEG4tJw7+eIzs0MrdbKc8BwffdnqZAGMGzeOcePGXfLr/mumCxuV1xnRaZQOAYfeaOaLHRmYrTZuGxiFu7ap7c3B7Aomf7jL4RrH5o9u0cvwUE4lkz7YKb9XKiQSn7sKD2c1Z0otruHmBQnUGEzEBrqTUlRDzwgvnNVK1iTZk+UlCZr/ZF6bGs+0nk3J2t/syeKZn5POvLQsPtSDr2/vw7LEXIpqGgjzcmZfZgX92/hwXc+wVs/JKqtjy8kSuoR50vV0DtfWnK3M2TTHfk9IfHKVvfJ6TnUOszvP5ua4m+3nVmfhrnHHS3v+9aUqDBVM+2UahXWFuGncWHLNEodaVWCfXtSpdS3OtdqszPp1FodKDqGQFLw++PUWyfI/pPzAx4c+xmKz8FCPh5jUdhIAA78bSFVDlcOxrmpXds/Yfd73LgiC8G90JVZ8v5z+NSNZjVpL8H78xyP8csQ+4nMop5JFt/aW93UL9+KW/pFyu5n2gW5ygFZW24CLkwqtWklcsDsB7k5yL0Sr1caCHelc2y2USF8XDCYLc749SEJGGWarDb3RvtLuQJY9Z2rzyRJ6RzYFKDqNEmxQZ7SgVkrEBTv+YV64M+Ocz3kkt4pNycXcMiCKo7lVTHh/BzZg+cE8gj2dGdDW8V8ExdUGJn6wk0q9CaUkcW33EEK9dHSJafq+bNj45PAnHCk5AsAb+9/g6uir0Sg1eGu9cdO4tbiP5PJkNmdvpotfF3m1XyMvrRdLxy3laOlRYr1iCXBpmWN2vOw4Xx3/Cl9nX9Kq0sirzeOOzndgtpnlpHyrzcrmnM0OQdbu/N38d/d/5ffP7XyOBksD18deTze/bmzJ3eLwOS4qx6lWQRAE4a8hKaRLXqbhr/KvC7Ja0xjoAJwoqG6xf96EOPakl5FcWENyYQ0zPttDYbW9jpSXTs1Xt/ehU4gHP97dn1sX7SO3Qo/BZOXdjaf4ek82G+YO4bdjhWw4ce68olAvZwqqDORU1KNSKNCpFdQZLZgsNr7Ykcn/pjUlxXu7OkFJU1V2d2cV1fXmFtcD2JRc7FAHPadcz5mSC2uoPF2Xy2Kz8cMBe0HTgdk6aDbL6KJpCkaUCiW/ZfzG6/tfR5IkXhzwItdEXyPvL6gt4OZfb6beXI+ExGejPqNPUFPSO9gDrcGhgwE4XHIYV7UrbTztRVP1Jj1zNs1xaKkD9ppbZ+od2NvhfVply6J2b+5/k+tjr+f1Ia9z94a7OVB0QN7XN7hvi+ObO7PAqSAIgiD8niuurc6FMpgslNY29eirqjfx0urjcqsVq9VGcY3BoWH03oxyuWdghd7EnG8TuX9JImqlxPJ7+zOqY9OqtvI6IxmltfaRqVY4N2vvsym5hJyKevk+Cqqb7iuv0jEw0p6RF1Zdb8bD2TFmDva0B1nHCxynxga3a1rhmJBRzo2f7+G7hGy5WGnzUKKgQuK5vs/RK7AXd3S+g1cGvsKYyDF08O7AE72e4N2D72KxWTBbzXxy5BOHz8moypDrVtmwcaKsKaExpyaHzdmb5Wm7Wb/OYuaamUxeMZnlqcsBqDPVtQiwGp3ZPie7JtvhfZxPXItzXNT2AFGr0rJw9EKmxExBp9IR6xWLWqHm7vV3szF7Y4vzvjr+FT2+7sGoH0eRWnH25uKCIAiC0Ny/fiTLbLVhsTb9wjaYrHy2PYPeUT4MaOvDjM/2ciinUi75IEGL7niZZXoyy/SsP1ZEiJczac1GmCJ9dLy8OhmjxcK4+CCyyvRc3TmI63qGYrXZeHrZUbnxs1qpwEmtoOGMnoM6jZL7hrV12NYuwI1tqY59nGoMTSNZGqVCntaM8Xfjt2P2UbRAdycC3LU0mC2oFApmL94nr5yc3DWYsZ2DOJRTwYdb0pEkmNU/kmmxQx1a4rw+5HUArv/levTmpuCvsXZVo3i/eKI9okmvSken0lFSX0JuTS41xhpmrZ1FvbmeUNdQbom7hcTiRMAePC09uZTJMZPx0/lxa9ytLDy2EAlJDqzC3MLoHtCdn0/9LH/WmvQ1cq0rsCfSn8lf58+m7E0MDx+OJEnM6z+Pef3nsfjYYrnJ9N7Cvfw25Tf8df6AvRr8G/vfwGqzUlBXwMeHP+Z/Q//X4tqCIAiCcKZ/fZDl6qTiP+M68n9rkx0aKhvNVjYlF3MopxKAynoTE7oEs/54IfWm1kuNGsxWhwBrQpdgCqsMJJxuqVNVb2bb48Mcznn52s4Ylx5me2opJbUNOJ+xAtBdqyLhmZEtVgY+PqY9dUYz3yXkyEHfjN7hWGw20kvquG1glJx/9uDIGFycVBRW1XPLgCieW5HEt3uzifTRUdvQFIwYzFZGxQUyKi6Q63uFo5AkwrxbJp03yqnOkV/7aH14aeBLDvtdNa4suWYJC5MW8vGRj/ny+Jf8mvEr18ZcK49w5dbm8lOqY1XdlIoU0ivTifaMZm7PudwRfwdzN89lT+EeJCSya7LJrslGgQLr6bKv+XX5TP9lOmarGU+tJ8/3fZ5eAb3YV7QPpaTEYrNwrOwYj2x5hPkD5lNWX8bIiJGEuYVRYWiaLjZbzczfNZ+3h7+NWqFGKSlxUbtQY7SXw/DQODYCP5vEokS25GyhR0APhoQNOa9zBEEQhCvLv3K60Gp1HIu6ZUAUx+ePYVrPUHxdnZjaI5QxnQIJ8XR2aD+zN6OsRYAV7u181s+pM5oxW5uON1usvPHbSSa8v4N3NtinnfzdtPSJ9pEDpXqTBZ2m6ccye1B0q6UXNCoFr1wbz7qHB3Nzvwg0Sokl+3KID/Xk+7v6OVSnVysV3DO0DfMndsJitfLtXvvUWmaZnv5tfNCqFYR5O/PQyKZmzxE+LucMsADuiL8DCQm1Qs2zfZ/FW+vd4hidWufQQqekvoRQt1AUkv0ZJSSHxswARquRlWkr5fc/pvzInkJ7La7m04TWM+rqHys7xsmKk+wt2Ms7B9/h89Gfs37qemZ3ni0fY7aZeWbHM7x54E1mrplJVUMVcb5xDsHTtrxtrElfA4BKoeL94e/TO7A3Ploflp1axjM7nuFci3IzqzKZvW42C48t5P5N97OvcN/Zv0RBEAThivWvGsmqqjdxy8IEDudUMqlrCP+b1lQkVKGQWlRb7xbuxQsTO/H59nQC3LWU1Rnl1YONssvr6R3pRUKzFjWNNp4o5urOgbQLcMVgsjKlewhvnQ6ujuRW4e+uwWyFaF8XtGoFBpOVEE9nhsT6crKgmul9IriuR+vlFuzXqCS7XE9idqVcy+v1305yQ+/ws57jrlWjUSnk1kEezmqSXxh7Ht9eS7d2upXxbcajVqjxcLIHKTvzdnKg6ABDw4YS7xcPwKiIUfyY8iMNlgbi/eK5JvoaIt0jWZOxhiXJS1q99vGy4/LrYn3xBd+b2WpGISkIdAlkdORoPj3yaYs8rnJDOYuPLeazo5+1OF+laPqr0T2gO0PDhpJQmADAyrSVTGo76awteLKqs+TWQDZspFSkiHY9giAI/0L/qiBr6b4cDmZXArDsYB7Te4djtdloMFsZHOPb6uqx344VyjlX8aEeDGjrQ1aZntzTCeouGiXTe4VTVmekut5ESa1jSwCj2ca6h+3TRSsO5Tnse37FcYwWK85qJZ/d3JO6Bgt70svkchEHso+QmFXBK9fGt7ivtUmF3PvNAaw2HJLqA9y18utlibmsOVpAr0hv7hpiX7Hn767lsdHteGl1MmBvYn1DammLZtkmq4ljpccIcglqtbRCI1/npvP2F+7nng33YMPGl8e/ZNmEZYS7h9M9oDsrJ60krzaPeL941Ao1Xf270tazLbvyd5FVndXiursLdlNuKMdb682NHW5kc85m8mrziPWKJbUi1WEUSyEp6OjTkTGRY/jmxDd4ab14uMfD8v5qY7VDgNU4zdh4reY8nTzp5NOJsVGOgWdjEAn20Td3zdnrw/QI6EFbz7acqjyFv7M/I8JHnPXYs7GZTJjLylD5+SEpL7yIrCAIgvDX+1cFWZ66psKgkgS/HMnny932X/A39A7nlWs7tzgnr7Jefl2hN7JyzkD0RjMv/HKclKJaZvaN4PEfD8uNooe080OpkNh8shgPZzX3DWsjn3915yDWHStiW0oJMQGuJJ4O+OpNFt5cl8KXt/dmzdECh89fkpDDPUPaEu7jOHW3KbmIxlnPxppbkT46Pp7ZHYBD2RXMXXoYgA0nign31jG2cxAAIZ6O18pv9oxgHwW6e/3dJBQm4Kxy5tOrPqWrf9ezfKv2VYC783dztPSoHMw0WBpIr0on3N0+qhbsGkywa7DDea4aV7695lsOFR8iuTyZDw99iMVmfxZPJ095NWCoWyg/T/yZORvnkFCY0GJEykfrQwfvDqRXphPoEkiNsYas6iwi3CPIq81jwdEFDsfHesfyUI+H6OrXldUZq+WaWQpJQWVDJTvyd7A1ZyvDwodRoi8hrzaPHbk7iPWKRafWMbHNRGK9Y8/6fTTmoqVXpRPmFtZq/bBzMVdUkDXzJoxpaTh37Ur4ooUotNrfP1EQBEH4W/lXBVlTuoeSXa7nYHYlk7qF8MWOpoKea5MKWg2yHh7ZjkeWHsaGjblX2XOWJCR2p5WRWaYnKa9KDrAA2vi58p/xHdE3mPnPymPM/HwvvaK8+XhmD7RqJR/caA+CssrqGPnmVvncgzmVPLL0MPcNa8u644VyEr5Oo5SrxhdVG1h+MI9wbx0DY/xYuj/X4V6r6k1E+LiQU67ntsX7HfYV1zRNcw5v70+vSC/2ZVbQOcSDsZ0dGynn1OTIU2P15nrWZKw5a5BlspiY9essTlacRIECd4071cZqojyi6BHQ4yw/iSbuGncGhw5mcOhgZnWcxedJn5NXk8fMjjOpNdayNm8tbb3aklaZxt7CvQ7nqhVq4v3iKa8v54eUHxz2Pb7tcXbdsIv/7PyP/CyNxrcZT/9ge2HU69pdR4RbBAmFCQ4lKJLLkynSF/Hy3pcdgrqR4SOZ0m7K7z6XVqW9oFY/zdWsXYsxzV7nq/7QIep27cJt+PA/dC1BEAThr/OvCrIUColHRjWNQBzLr+L46eKjfaN9Whz/1e5MkvKq+eSmHvSJ9kansX9dO9NKyTxdJ6vBbMXHRYPBZCHaz5W2/i78Z0USwR7O/Hi6oOeWkyX8lJjLjX3sLWPyK+u5bdE+h+AM7IFWlzBP9j0zknc2pJJfVc/MPhF46NQczKrg1kX75KbWL0zqxOi4ALk0A0CnEPuU1tL9OZTXNU1btvVzZVK3pvIKWrWSpXf1o6rehIezusU0aYAuAF9nX7lnYQfv1ptIW21WcmtzOVlx0v4eK8PDhzO13VRiPGPkdjiJRYm8nfg27hp3nuv73FmnH51UTtzX9T7APjo2ZeUU8mrzUEpKeXujcdHjuDP+TqI8ohiwZECLaxktRkxWE9XGpuKyXk5evDr4VfoF93M49kT5CXbn78ZH60OZoQw3jRtXRVzFnE1zWoyaFddfeH7YhVKHNcupUyhQh4Sc/WBBEAThb+tfFWSd6blrOtI1zJMGk5WJ3RynspYl5vLcimMA/Hwojy2PDZWDrFBPxxWFFXoju54cQVpJLTd+bh9tObPftIum6aue9UWCQ6mHRn2j7Kvz3LRqnh3XkU+2pnHLwn34uGooqTFgbraY7lB2JeO7BMtBlpdOzQcz7KNkYV5N04FOKgXf3NHHoYfi0n05bE0tYXisP1N6hLa4D51ax+Ixi/kl/ReiPaIZEzXGYb/epOe+jfeRWJzI0NChhLmFkVNjL+cwOHQwXfwcFxA8vOVhyg32MhbW3VacVc6YrWYe7vEwkR6RLa79bfK3FNUVkVdrz2Gz2CxszN4or0icGjOV5/o9J58T7hZOUpm9j6Oz0pl6Sz0mq4nndjzHIz0f4fGtj2O2mpnXfx79gvthMBtYcWoFTionQl1D5RpZAA90e4BrY67Fx9lHnm4E++ils8qZe7vc2+L7utRcBw4g6JVX0Cck4DZyBNrYs09NCoIgCH9f/+ogS6GQmNi19VGCxoruYB+teu7nJPpE+XD7wCjaB7nTLdxTTqK32uy5W82rwluscF2PUFKKa+kT5c2ELsGsTSrkuZ+PtkiO99KpmdwthEdHN/0yrTGY+L+1ydhsUFBlcDheKUmMiw9iWHt/PJ01pJXUMrZTIO6nA6nreoZS02DmeH41U7qHOCTD70or5fGf7L0HVx8pINJXR4+IlqUXwt3Dubdr6wHF6ozV7C+yT0duytnE/w36P+pMdUR5RLW6is5gbrr/o6VHqWywf28FdQUsHb/U4dindzwtV11vDJjAXp6hUWNABVBjrHF4r1AowJ7WRUJhAq8NeY1t07c5fMYjWx9hW65929CwoQ77fJx98HH2ocZYQ6hrKO282tHVryt3dbkLb623vOrQZrOxPW87SknJgJCWI2kXy3PyJDwnT7rk1xUEQRD+PP/qIOtcpvYI5ft9ORRWG1AqJDacKGbDiWKKawxc3yuct6/vynUf76a4poH+bXyID/UgzNuZz3ekk1NeT6cQd+ZNiMPFqekrfnLZEbk/IIBKIXHn4GgeH9O+xeerlQpcNCq5WKizWkm9yYK7Vs3ns3rQO8o+vTkwxrfFykBJkrh9YFSrz7XumGP/xMM5Va0GWefi6eTp8D7MLUwu19Caef3n8creV3BVu2K0NgWYFQ32shc5NTm8m/guKoXKIZiy2CxoFBqHc8Ce6N5Ip9IR7BJMfl0+AO2928s9Cc+2qm9/YVO+WlFdEddEX8PGrI30DOwp9158Yc8L/JrxKwC5NbkYLAae7P2knMT+4p4XWZpiDxBv63Sbw2rGszFZTNiwoVG2bFIuCIIgXHkk27mqKl4m1dXVeHh4UFVVhbv72ZfC/9UMJguHciqZ/ukeh+2SBP93bWfGdwmmuLqBMG8dSoUkn5NXWU+Ylw6NSoHNZuP/1iazI7WUrDK9HDT1b+PDezd0w8fV6ayfvyutlA83pxHi6cx9w9uSWVpHXLD7Oc/5Pe9uTOXN9Sny+y9m9WJ4B/8Lvs4nhz9hf9F+roq4inpzPbvydzE4dDA3drix1eML6wqZ/st0ygxlSEioJBXh7uF09e/KifITcl0snUrn0KrnTHE+cbw2+DWCXINQK+wjd3m1eXyf/D0BLgFMj53OgaIDGK1GBgQPQJIk8mvzeW3faxgtRu7ofAe3r7tdrmM1JnKM3CaouRtX38iR0iMO22a0n8FTfZ4CYMTSEXJ+VrRHNCsmraC0vpS3DrxFvbme+7vdT5RHU6C7MWsjT2x/AqvNyosDXuTq6KvP96sWBEH42/in/P7+uxAjWafVGy1U6I1sPllMxyB3uoV7oVUr6RPlzbj4IH450lRawWaDH/bncn2vcCJ9Hb9CrVpJGz9X+f1vxwr5ZGu6/D7G35UgT2dentxJDpbMFiuvrzvJ8fxqru8Vxrh4e35Y/za+9G/TNEoV/jsV2M/HtJ5hLN2fQ25FPb2jvBkQ0zLh/3zc1eUu7uIuduXt4q4NdwGwK38X7bza0SuwFx8d/ojPjnxGmFsYH438iJ15OykzlAH2Ap02bKRVpZFWlYazqinHTW/W086rHSkVKa1+bq/AXoxbPg5XtSsfjPyAbv7dCHENYW7PuehNemzY6B3U2+Gc53c9z54Ce6CcU5MjB1hgT/JvzW2db+OxLY9hsjUd29haB6BvcF+5Kn3foL4AzN89ny05WwB7c+zlE5fLx7+x/w0aLPYVnu8efFcEWYIgCP8C//ogq7zOyIzP9pBcWIOTSkGD2YpCgq9n96F/G3uB0vdu6MbTV3dgzreJcm2rzqG/38PulyP5/HLYse5VemkdXcM8CWmWPP9tQrYciO08VUqXUM/fbWnzRwV6aNkwdwiltQ0EeTjLI3B/VHlDucP7LTlbOFh0kA8PfQhAelU6XyR9gdHSNOUnITk0cPbQeMi9DAH6BPahoLaAGlONPF2oRMk9Xe/ho8MfYcNGjamGN/a/wX/6/odY71je2PcGi48vRqPU8N7w9+QSDYDDCkOD2UDfoL7sKdiDu8ad8W3Gt/pcI8JHsG36Nr4+8TWfH/2cQJdA7oy/U94/v/98BgQPQKVQMTJiJACl+qaG3cV1xehNeqqN1ZQbyuUEeuCchUwFQRCEK8e/Pshauj+H5EL7CEXD6eV7VhvsTS+XR5EkSSLY05mFt/Tm671ZuGtV3NA7HKPZyvKDuagUCiZ1C3EIWH46kMsjP9iLgTqpFEgSGExWLFYbPxzIZWqPUPqcLhuRkNEUqFht9npXZ2+mc/G0aiWhXpcmiLsq4ip+Tv1ZrmH15fEvWxzjrnF3yLWyYePG9jeyNGUpzipnCvWFDsenVKQwJmoMxfpiDGYD4W7h9A3ui1allYuVAhwpOcLUVVO5r+t9LD6+GLCXbnhmxzNsnrZZPu6RHo/w6NZHMVlNPNn7SYaEDSGjKgN/nb9DJfczOamc6OjTkcVjFhPnGydvL6gtoMxQxpioMfKKRwB3p6bgyWqzMvqn0VQ2VNLZt7NDKYip7aae/QsVBEEQrhj/+iDL361lfpNGpWBorF+L7R46NfcNayu/v3/JQVYdtidcL9iRwevXxRMXbP+lfTSvSj6uwWzlsdHteP03+xSYJCGvBAR7UNVcW39X/imclE68POhlRvzQMsm8o3dHOvh0YHbn2WzN3cqegj1YbfZAdnXGatZcu4Z7NtzjMNIEsLdwr0Ph0b2Fe/kh9Qfc1K1XTl+VtgoJSQ5kzFazw/7eQb1brDCM8Yr53Wd7YNMD7MjbAcALA15gUttJ7MjbwQObHsBkNTEkdAhze84l2iOaF/a8wK78XfK5teZaOH0bR0uPykVaA3QBf6jNjiAIgvDP868PsiZ3C6GgysDB7Equ6uCPu7OadoFuDnlVZ5OQUSa/Pl5QzfRP9rDx0SH4u2mZ0DWY7/flUG+y0DvKmzsGRlPbYCExq4JJ3ULoENQ06tE93Ivtqfapprb+rmjVf22vuoyqDF7a85I88tPBp/VipI08nTwJdAmksK5pRCrKI4qFYxbKBUnHRo0luyab9w++D0BlQyWHSg4xLGwYpypPndd91ZhqWt3up/NjbNRYPj3yKRISj/d6/Lyud6aEggQWJC0gyCWIe+PvlQMssCeuT2o7iRWnVsg5XVtzt7I1dytTYqaw/NRyh2t19evKoZJD8ntvJ2+6+nXlsV6P8Uv6L3yR9AWR7pG8OfRNfJz/WF6cIAiC8PcmVhdehHkrj8nNnBv9dE9/ekR4AVBYZSCvUk/nEE80KkUrV7DXW6prMLPicD4lNQ3M6BOOv9tf26du1q+zSCxOBKCtZ1uHBO6zyavNY3nqcrRKLdGe0fQK7NWiZ9+mrE08us0+beemduPHCT+iUWoYtnSYfIy/sz/F9cX4OfvRYGmgxliDDZvc1LlRF78uHC6xT8cqJSXLJy7Hz9kPhaSQA7sLYTAbGPL9EHllo5vGzSHR/dGejzIrbhYLji7g7cS3W5zffCTNW+vNj+N+ZPiPLVvhDAweyI78puBtVsdZPNrr0Qu+X0EQhL/ClfL7+8/yrx/JuhjzJsTRMciN//5ygtoGM146NU/8dJiZfSK4ZUAUgR5aAj3OHjAV1xi44dM9pJXUMb5LMO9O79qixU1VvQl3rarF9supcRUcOBYSPZcQ1xDmdJtz1v1HS47yyLZHMFvNuKpd+fSqT+WG0a5qV2pN9kKuOrWOfdfuQ6vSYrPZqDfXc6zsGCX6Eubtnke9uZ5b4m4hqzpLvrbFZqGgrsChZMKFMlqNDqUjmgdYd8bfyay4WYC9JtaxsmOsz1ov71dKSjlXTK1Q8+KAF/F29sbP2Y+S+hKHzymuL3Y4XqPUUKwv5tWEV9Gb9TzU/aFzNp8WBEEQ/jlaH14Rztu0XuFsf3wYt/SPpEJv4lRxHfNWHSe9pPZ3z/0uIUdur7PqcD6Hc5vyuAwmC9d/spsu89cx8YOdcn2tP8OTvZ8k1DWUAF0Az/V97vdPOA8JhQlyrlStqZZqU1MeVjuvdvLrCPcItCp7YCpJEjq1jl6Bvbg6+mo2XbeJTddtoqiuiM05TYntYa5hPLH1CUb/OJqjJUfPeg8ZVRk8tPkhHt/2OMV6xx6E7hp3Huz+ICpJhURTQCshcXVUU7kFSZK4uePNKCX7lK5KoXJIxn+i1xMMCh2EUqFkbo+5Dp+hklTc3+1+XhjwArFesUS6R3Ki/ARzt8xlXdY6duTt4JGtj/z+lykIgiD8I4iRrEvAy0WDr6tjFW+L9fdnYZsn3asUEmuTCnjou4PEhXgwqmMAe0+vOjySW8WaIwVM63U51xw26erflV+n/HpJr9k3uC+aQ/ZyDP7O/g5Np18f8jqfHP4EpULJPV3ucTjPZrPJo3iuGldcNa4cLXUMpHJq7X0TK42VvLH/DRaPXUy5oZy0yjQ6eHfAVWPPr3t066Ny/a0aYw0fjfzI4TqzO89meux0+i1paiDdxa8LbTzbOBzX1b8rX439iqOlRwnQBfDwloflqcLmo2ljIsewLXcbW3K20Nm3M28Oe5Pk8mTWZ63H19mXnfk7yazOdFihWNVQhSAIgnBlEEHWJTKrfyQHsio4ll/NjX0iiAlwI7dCz8HsSrpHeDnUxWo0rWcYxTUNJOVV0a+ND/NX2aueZ5bp8XFxDNrONe34TxDnE8cP43/gRPkJegf2xkvrJe/z1/k7NHwGqDXWct/G+zhccpixUWN5aeBLcjAyqe0k3j/0fquf46J2Iac6hxvX3EhFQwUR7hF8c/U3eDh5UF7fVCqjsWH1mVw1rtze6XYWJC3ATeN21nypzn6d6ezXGYDXBr/Gttxt9A3uKxdCTalI4c51d1JmKJPb7hTWFXLfxvscpmPBXu7B08kTg9neukcQBEG4Mogg6xJx06pZeGtTpfGccj3XvLudaoMZD2c139/Zl6KaBmL8XQk+HXApFBIPjLCXEkgpclw5F+zpzJvTurDhRBH92/gyuF3LkhL/NNGe0UR7Rre67/Ojn/P18a9p49mG/w35H6szVsvJ97+k/8K1MdfSK7AXdaY6uvl347m+z/H6vtcxWAxEuUfh7eyNi9qFZ/o8w6bsTXJfxKzqLA4VH8JL6yXnXGkUGh7q/tBZ7/OhHg9xS9wtOKudcVKevYWR0WLkwc0Psjt/NwNDBjImYoy8b/GxxXKF+y+SvuDWuFsprS91CLAac9FuiL2BpLIkjpYeZU3GGkZFjpJbBgmCIAj/XCLIukz2ZpRTbbDnIFXVm5i1MIGi6gZcNEp+uLs/HYMdV2W0C3DjsdGxLEnIplOwBzf3i0CnUXFt99C/4vYvu2J9McnlycT5xFFvruedxHcAKCssY/HxxUR7NAVjEhIny0/ybuK7nKo8Ra2pFp1KJ5dSKKgr4L0R7xHhHgFAR5+OKCQFVpsVZ5UzbTzb8Mb+N+Qgy2Q10c2/W6v3ta9wH18e/5Jwt3Ae6P7AOZ9hfdZ6uczD1tyt9PymJ6MiR/HqoFfxc24Kit00bjirnSmoLZBLXQS6BLJw9EL8dH6sOLWCJSeXALAtdxu783czOHTwH/laBUEQhL8REWRdJt3DPdFplOiNFrQqBUXV9hGMOqOFDSeKWgRZAPcNa8udg6PZnVZGdrme9oFX5vLYr45/xev7XseGDS8nLz4c+aEcFIF9pGlc9DiOlR0jsSiRq6Ou5q0Db2G0NrXmab4S0GAxcKL8hBxkxfvFM6vjLDKrM7mj8x2EuoUS6BIoH++t9UajdJyOBfsU5ZyNc+Rr69Q67ut631mf48xq8TZs/Jb5G5PbTuaqyKtIr0pHQuLWTreyJ38Pc7c2JcLP7z+fUDd7AO2jbaqTJSGJulmCIAhXCBFkXSbRfq6snDOQfZnlxAS4Mnvxfir1JiTJXnz0bO78cj+bT5agkODt6d2Y0CX4T7zry6/B0sAb+9+QE8UrGirIqcnh+X7P8+2Jb4n2jGZW3CyOlx9neepye8Aj4RBgNWqcbtMoNLy27zWK6oqYFTeLl/a+xI8pPwKgVWl5ze81Huj2AApJQam+lNs63+aQbN6o3lzvELztyNtBfm0+U9tNbXXka2DIQB7u8TArT60krSpN3p5RlcEbG9/AYrPQ0bsjHx7+kN35ux3OLaorkl+PiBjBIz0eYX/RfkZGjCTOJw5BEAThn08EWZdRW39XuUXO8nsHsP54IZ1DPOnXpvWRihqDic0n7XWVrDZYc6TgiguyJCSUklIetVJKSjr6dCTMLYwifRGpFakkFieSUJggBzwnyk7Q1rOtXBneRe3C28Pepq1nW97Y/war01dToi/hjf1vMDx8OPsL98ufl1hkz+vSqXW/WwneT+fHrXG3svDYQtw17iSVJpFUmsSGrA2sm7oODycP1mauJbMqk3HR4wh1C+XWuFsZHDKYlWkrOV52nFGRo0guT5bLOhwvP97icyLdI/Fw8qC0vhRfZ3t/zFs63cItnW65uC9XEARB+FsRdbL+JFG+Ltw5uM1ZAywAVycV7QObqqT3jDz7iNc/lUap4bXBrxHqGkqkeySfXPUJEe4RLD25lA8Pfcj6rPU8uOlBh5WAnk6eDjlOBrOBngE98XTyxF3TNKUqIaGQFIyOHC1vuyriqgu6v7k957J/5n7GR4+Xt+nNesoN5SxPXc5jWx/jg0MfMOvXWeTV5DF11VQmr5zM9ye/p1dgL1zULmRXZ8vnOkmOifMSElnVWTy4+UEmr5hMTk1Oi3tIq0zj1YRX+fbEt3Iw+tXxrxj6/VBm/TqL0vrSC3omQRAE4a8hRrL+RiRJ4rs7+7L8YB6B7lrGdg76q2/pshgZMZKRESMdthXpm6bPjFYjK9JWANDZtzN3d7mbdw68I+dt3dv1Xqw2K3esu4PE4kQ0Cg3+On9mxc2SK8/3DeqL1WaVSyqcr4SCBFIrUxkeMZzfsn6jtL6UsVFjiXSP5OvjX8vHFdcXc9tvt5FfZ28QrjfrWy0r0WBrIEAXID9f4zQp2Ps37sjbwQ3tb5C3GcwGbvvtNrnEhMlq4uqoq+UctjJDGQuOLuCJ3k9c0HMJgiAIfz4RZP3NeOo03Drgj7eH+aea1m4av2b8Sl5tHk5KJ7nUQWl9KYuOLSKl0l5ENNglmDvj72R/4X65xIPRamRyzGSmt58uX69nYM8LvodN2Zt4cPODgL12108TfsJiteCns4+ijY0ay4q0FTRYGujm342DxQfP67rDQocxIHQAj219DIOlqU2RhEScTxwWq4XVGasxWoz0DuztUMMrvSodpUKJUlJittlXq7aWtC8IgiD8/YjpQuFvIcg1iDXXrmH3DbuZ2GaivL2groDksmT5fWOT6BDXEIcaVs1LPvxRjflbYC8xUaIvkQOswrpCFiQtABuEuYUxv/98BoYMPK/ratVahoYN5cWBL8oJ9xqFhuf6PofZaubFPS/yzI5nmL97Pm8eeFOe7nRVu3JtzLV4a73574D/EusVy1URVzG78+yLflZBEATh8pNsNtvv93+5xEQXb+FcbDYbs9fNJqEwQd4W6R6J2WrmqT5P4eHkQYhrCNnV2azNXEtn386MbzP+HFc8PweKDnDHujswWU2082rHt9d8i5PSCbPVzDXLrpGnBsFedf75fs+zPHU5/93zX3n7uOhx3NTxJp7d8SyplamAvWTE1uu3ApBdnc2hkkP4aH14bNtj1Bhr0Cg1GC1G+dgt07aQUZ2Bj9anRZkIQRCEv5L4/X1hxHSh8LcjSRKjI0fLQVa4WzjLJi7DZrNx62+3cqTkCK5qVxaOWcjTfZ7mUv07oUdAD5ZPXE5WdRY9AnrII2XVxmqHAAvsRUMf3vwwT/d5mg7eHThRfgKNQsO02Gl09OlIsGuwHGQ1riAECHcPJ9w9nE+PfEqN0V7lvzHAAnuS/xdJX3BL3C1IkkSFoQJPJ0+5f6MgCILwzyGCLOFv4VTFKd45+A7OKmce6/kY02Kn4a31Jrcml3FtxqFWqDlWeowjJUcAqDXVsjp9NakVqbyw5wWcVc68Pezts1ZyP18R7hFyUdNG3lpvhoUNY3POZgAUkoJyQzlbcrfg4eTBojGLOFh8kHC3cOot9aRXpjOv/zzeOvAW9eZ6HujWsnJ881pYLmoXnu/7PPP3zCe9Kp23E9/GZDWxJWcLx8qOEe8bz2ejPkOn1l3UswmCIAh/LjFdKPwtjF8+nszqTACGhQ3j3eHvtjimwlDBqB9Hycnj0R7RlNSXyCNCPQJ6sGjMostyfxarhaSyJDydPJm0YhJmqz0JfUzkGF4f8joAHx36iA8PfwjAoz0fZVbcLIdrlNWX8eq+V6k2VnN/t/spry8nqTSJ4eHD8XDy4Kofm8pNdPXryqGSQ/L7Vwe9ytXRV1+WZxMEQThf4vf3hRGJ78LfQmVDpfy6wlDR6jFeWi/i/eLl9+lV6Q51sjydPAGw2qwsTFrIMzuecUhmvxhKhZIufl2IcI9gXr95+Gh9aO/dnvu73S8f82Pqj/LrZanLMFvNbMvdRlJpEgCvJrzKrxm/sjNvJw9uepBBoYO4p+s9xHrHEugSiE7VNFLVGMQ1CnK9Mst5CIIgXMnEdKHwt/B4r8eZv3s+WpX2nI2Zm+dqdffvztN9nuadxHdwVbvyeG97Rffvkr/jzQNvAvYmzr9N+Q0v7aUr7Dqx7UQmtp3YYntH744U64sBiPGK4eHND7MldwsA8/rNcwgki/RFvJP4Dg92f1De1ryGlk6t44UBL7AzbyeDQwdf9DSoIAiC8OcTQZbwtzC+zXiuib4GCemcSd7TYqcR6R5JcX0xI8JH4Kxy5sORHzock1ebJ7+uN9dTbii/pEHW2fQP7i8HVaX6Ug4UH5D3bcjewP3d7udwyWG5XdDnRz/n6qirifGKAexTjK/sfQVntTP3dLmHnoE9mdR20mW/b0EQBOHyENOFwt+GQlKc1yq63kG9GRc9DmeVc6v7p7abKrfhGRM55pLU0Dofuwp2ya8PFB+gs29n+X2vwF509uvMvV3vdTjng0MfyK1zpsVOI+HGBHZM3/GHiqkKgiAIfy9iJEu44kR5RLF2ylqqjdUO5RMut35B/diSswWwJ65/fNXHrE5fjY+zDyPCRwBwQ/sb2Ja7TZ7y3Ji9kS05WxgePhyAnNocPj3yKe4ad+7vdj9uGrdWPkkQBEH4JxBBlnBF0ig1f2qABTCjwwx5KnNUxCh0ah3TYqe1uK8Z7Wc4FFptXrl+zsY5ctPoOlMdLw186c+5eUEQBOGSE0GWcMXKqc5hycklmC1mJsVMoqNPx8v+mf1D+v/uMcPDh3Nr3K3sLtjNoJBBDAgZIO8rqitq9bUgCILwzyOCLOGKdLTkKDevvVkuhbAybSVrpqzBW+v9F9+ZvaL93J5zW2w31NZyW/4AkvOPkty+gTvi73DYbzGbKcvNxt3XH62r6591u4IgCMIfJIIs4Yq0OWezQ62pOnMdhXWFf4sg62w2L/oEQ2I6kbgQb4mmzz195H0Ws4kfXniWvORjaF3dmD7/VXxCw//CuxUEQRB+j1hdKFyRugd0R6JppWI3/25yqYRLqaqhig8OfcCnRz5Fb9I77CvWF/PMjmd4cvuT5Nfmn+UKTWoryuXX+krHgqwlmRnkJR8DwFBbw4kdWy/B3QuCIAiXkxjJEq5IA0MGsmD0Ag4VH6KDTwf6BvVFpbj0f9wf3fooewr2ALAqbRV3dbmLcdHjAHh2x7PsLtgNQF5NHl9d/dU5r9Vv6g0UZZzCWF/P4Jm3Ouxz9/NH46zDWG8P5HzDI1q7hCAIgvA3InoXCgJgs9nOq0bXmYZ+P5QyQ5nDtreHvc2I8BFMXjGZU5WnAAh2Cea3qb8B9qm/XUu/oaIwn+5jJhDasZN8rtVqwWqxolKrW3xWUfopkndtwz8iig6Dhl3wvQqCIFws8fv7wojpQuEfb3vudh7d+igLji7gQv/NkFebx/jl4+n+VXfeP/j+BX/2jA4zWmxLr0wH4P5u96NValEr1DzU4yEA6ior2LHkKxJW/Ejq3l0se3U+DfqmaUaFQtlqgAUQEN2WITNvEwGWIAjCP4SYLhT+0QpqC3hw84OYrCZ+y7T3KLw25trzPn9h0kIyqzMB+OTIJ1zX7joCXALO+/w74+9keNhwnt7xNCfKT+Dn7MfYqLGAvVTDrhm7wAZqpZrj2zax9qO3sVmt8vkmQz0N+jqcdLoW1zYZDBRlpuEdHIrO3eO870kQBEH4exBBlvCPVmYow2Q1ye8L6wov6HwPp6bgxUnphFalveB7aOvVlm+u+Yac6hwCXQLRqZsCJrWiaVRq/y/L5QBLkiRsNhvdxo7H3devxTWN9Xq+ffZRynKz0bq5c8N/X8c7OOSC700QBEH464ggS/hH6+jTkVERo1iXtY4wt7ALGsUCmN15NlUNVWRVZ3FTx5scgq4LoVaoifY8d49Er6AQSrIyAAiO7ci1T81Do23qv1iWm82vH7yFqcFA3NCRlOVmA2CoqSZt/x68J0z5Q/cmCIIg/DVEkCX8oykkBf8b+j+qGqpw07ihkM4/zbA8P5f9q5Yx1COCPtc+hlrj9PsnXYRRdz2AZ2AQFpOJ3hOnOgRYAJsWfkJReioAiat/Ru2kxdRgAMA/qs1lvTdBEATh0hNBlnBF+CMjUD++9Bw1pSUAmBoMDJt1x++ccXGcdDoG3TDLYdvBtasoykgjbvBwh+2SUsn1818lbf9egmNiiejc9bLemyAIgnDpiSBL+FeymE3UlJXK7yuLCv70ezi2dSObFn4CwMmd25j81DxMBgNGQz3Db72bgKg2BIgRLEEQhH8sEWQJ/0pKlZreE6eS8PMPqLXO9Lxm0p9+D80DO7PJyI4lX9J+wGC6Xz3xT78XQRAE4dITQZbwrzXohll0HzsBtVbbIj/qz9Bp6EiStmyg9vSIWkFqMgWpyXiHhhMZ3+1Pvx9BEATh0hLFSIV/NRdPr78kwALw8A/k9rc/ZcTt9zhsr6+q/EvuRxAEQbi0RJAlCH8hlUZDp2GjiDg9chXeKZ62ffr/xXclCIIgXApiulAQLoHqkmLUWi3Obhfey0ulVjP1mRewmM0oVS3/SlqtFvb89B2lOVl0Hj6aqK49Whxjs9koSE3GSeeCT2j4H3oGQRAE4dISQZYgXKStX3/B/lXLUKk1jH/kKaK79XLYX5yZTlluNpFdulNXWYGhrpaQ2I4tGlIrVSqqigs5uXsHvmERRHe3X2fPsqXs/nEJAOmJ+5j93gJcvbwdzl33yXskbV4HksToux6g07CrLuMTC4IgCOdDBFmCcBGsVgsHfvkZsK8QPLj2F4cgKzvpMD++9Bw2qxWduwf66ioA4oaMZMy9Dzlcq0GvZ8lzj1FXWQHAuIeeILRDJ/at/FE+xmIyUV6Qx/Ftm3Dz8aXDwKEAHN+20X6AzcbxbZtEkCUIgvA3IIIsQTit4NRJNnz+IQqlklF33o9fRNTvnqNQKPEMDKSiIB8A7yDH/oIZhw7I/QobAyyAY1s3ENW9J0qlijY9+yBJEtUlRXKABVCQehKlWoO5oUHe5h/dlo2ffUB5fi4AtRXl9Bp/LYFt2pGfcgKAwLbt/uA3IAiCIFxKIvFdEE5b++HbFGekUXgqhfWfvX9e59RWlGOz2V8HtmnHoBtvddgfEd8N6XSrnzPzrX556/9Y8caLbP3qcwC8Q0IJbBMDgErjREyfAQRGt0V7Os9LoVTScdBwOcACKEw9CcDkJ59n8I23MuquB1pUlRcEQRD+GmIkSxBOaxxxOvP1uRz6bTWVhfZRrMK0FCoL8vANj5T3R8Z3Y8ZL/6MsN5uMQ4mc3LW1xTUyDh5g6M13oFCqGHvfI1QW5eMTGoGHfwAA18/7P77/zxMY6mrYsvhTAtrEUJSWiqRQ0H7gEAC0Lq70Eg2kBUEQ/lZEkCUIp426637Wf/o+CqWSEbffe17n6NybVhMqlCqcXFwd9teWl7H+0/cpz8shfuQY/CKjqS4pwjs4jILUZADc/fzZ/u0ikrZsQF9VidbNnYHTb6YkM42wuHi8Q8Iw1NXI1wyJ7cDQmbej8/TCO9hxelIQBEH4+5BstsbJjj9PdXU1Hh4eVFVV4e5+4UveBeHvwmqxsOO7LynJziR+xGhiejvWuNry5eccWP2z/H72ewvw8A/AZrVyav8ejm/bzKl9u1teWJJonIe89qn5bP92ESVZGSiUSrqOuobIrj2J6tr9cj6aIAhCC+L394URI1mCcBEUSiWDz8jDak7jrGt2rAq1kxMAkkJBTO/+DgGYg2b/9qkuKWL6f18j81AiGxZ8SOKvK0n8dSWu3j5MfuJ5/COjL8mzCIIgCJeWSHwXhPNgMjbQoNdf8Hm9Jk6h84jRhLSP45oHH0Pn4emwv/2AofJrSaFAqVLRafgo/KPaAOAZGERM7/5otM74hIZR32yFYm15GaveeuUPPY8gCIJw+YmRLEH4HekH97Hqzf/DYjIx7JY76DZm/Hmfq9Y4MerO+8+6v+uoqwlu1x6jXo93aDir3nqZ1D076TxiNOMeegI3b19UGg1g73Wo1jpjMtTL59eWl5N7Iony/Fza9uzbIogTBEEQ/jpiJEsQfseeZd9jNjZgs1nZ+f3XAFjMZjYs+Ihvn3mEIxvXXtT1/SOjCe3YiYO/riD3eBIN+jr2r1qG2WiUAyyAstzsFqsezcYGvp//FOs/fZ8lzz2Gydhw5uUFQRCEv4gYyRKE3+Hu609ByumVgL5+AGz7ZiGH160GoCAthdAOnfAODqW6pJj6mmr8o9q0aJujr6pk/WcfUFdVwYBpM4no3FXeV5aXw/7m+VmShEqjwWiopzgjDe/gUE7u3o75jCBKUijkwKuyqIDqkmJ8QsIu8TcgCIIg/BEiyBKE3zHy9ntxdnPHqNdjajDw0Z0z0VdVNh1gs2FqaCB1325+eev/sFosdBw0jLFzHnG4zrZvFsorCVe88RL3L/weSWEfTD66aZ1DZffQ9nHo3D355um5lOfloHV1o++U6S3uzWa1IkkSNpsN3/BIPPwDL/0XIAiCIPwhYrpQEH6H1tWVEbfdTUiHOFITdjkGWEBs/8F4+AeQtHk9VosFgOPbN2M2GuVjbDYbucnH5fcmQz0JK3+S3zeOkDVSKJUUpJygPC8HAENtDVaLheG33oVCqXQ41t0/UE6aNzUYLskzC4IgCBdPjGQJwkVw8fTi5K7tpO7dRZuefeTt3sGhDvlUq999naqiAodzdyxZTMqeHXj4BTD8trs5vn0zRWmpAGQfO4LZZELl5GQf4ZIkAqPbUltRLgdydpJ83aL0UxzZsJY+k667fA8sCIIgnDcRZAnCeYobMpzcE0nknzxOaIdORHbpzqZFnwA2rBYzlYX5jJx9L3WVFcSPHCufZ2owcHLXtlavWZyRRnFGGgqVihtfepN3Z021B1U2G/knj6N1daP3hKkEtWuPQqVmx3dfARLQWEfL5lC41NnV7bJ+B4IgCML5E0GWIJwnpUrN1WfkWSWuXUV9dTUAZqOR6B69cfP2dThG7aTFJzScstxsAFw8vbFaLRhqa7FZ7aNSWUcOUpKVgVdgCCVZ6fK5htoawuI64xcRxUd33oTFZORMvcZPoTAthYDotnQaftUlfWZBEAThjxNtdQThItSWl7H81fkUZ9oDI+/gUG5962N5f+GpFA6tW4OTzoXizDTMZhOlWZmYTUac3dwdiot6B4cy5ZkX+OKhO7GYTPJ2nYcn055/hUVz73H4bN/wKGL7DaTvtddf5qcUBEGwE7+/L4wYyRKEi+Dq7ePQFLo8PxeL2YRSpcZYr+fHl5+joa6u1XObB1gARkM97r5+TH5yHj+/9l95tWGDvg6vgCDCO8WTnXQEJIkeV09k6M2zL9+DCYIgCBdNBFmCcBH2r1pGzrEj8vvOI0aTe+IYDfo6/COjzhpgNafWOqPSaBhx+70AbFn8mRxgSQolw2+9G4VKxZRnXqA8NwcXbx+ReyUIgvAPIIIsQbgIB9askF97BATiFRjMjy8+C4DW1Q2v4FAq8nMdzlE7aUGSsFkt+ISGM+mx53D19pH3VxTkya9D2ncgfsRoABQKJb7hkZfxaQRBEIRLSdTJEoSL4BMaLr8OiGxDWmKC/N5QW0NFfi6hHTs5nGNqMGAy1BPasTOx/QeTfewIRRlprPvkXfYuX0qv8VMAUKrV9LhmsnzO3uVL2fXDNxjqav+EJxMEQRAulhjJEoSLMO7BJ9j/yzIkhZJe4yez6u1XWxzjGRBEYVqqQ0V3gNzjR8k8dAAAlcZJbpnTefho7vxwESonJ3lacP2n73NixxYA8lOSmfrMC5fxqQRBEIRLQYxkCcJF0Lq6MnD6zQREteHzB+4g98Qxh/3+kVEMumEWt7zxIYNn3uawr3lF+OY9CY9u+o39vyxzyLsqyc6UX2cfPUTBqZOX+EkEQRCES00EWYJwCWz4/APqq6swNxjsxUGBNj37cNOr76Hz8KQwLQV9VQUKZdPgscZZJ78OjIl1uN6RDb+RtGUDFrO9lEPXUVfL+2w2G8temSda6AiCIPzNielCQbgEmvcTVKnVTHjkGSLiuwKQvGsbq995rcU5zUeyXNw9HCq3m40N/PbR2xzZuJaakmK0rm5odK4Y9fZ8LENtDYfXraHn+Gsv41MJgiAIF0OMZAnCJdB81Z/ZaOT4ts0oFPbAq7FQ6ZkkRdNfP0NdHRMeeQZ3vwCU6qaehwUpydRWlFOak4Wbr4/D+Rmn87kEQRCEvycRZAnCJeAXEeXw3mjQy6/b9x+Mk86lxTmSsumvn0KppL66ktnvfU7viVPk7Uq1Wn7t6RfocH7zIqiCIAjC34+YLhSES6Df1BkUpZ8i++hhnN3dGXj9TfI+/8hobnvnU758/H7qKsrl7ZZm04U5x46Qc+wIdZWVeAeHMvqeh3Dx8CTj0AFS9u7EOyiE4bfdTVC79uz87it0Hh6inY4gCMLfnOhdKAiXkM1mQ5Ik6k/nTDnpdMT2G4RCpSI/JZnl/zdPPlatdcZkqHc4X+2kxdRgQKlS0a7fIE5s3wzYpyNnvf4+ABazGYVSiXQ6wV4QBOHPIn5/XxgxkiUIl1Bj4LPokXvQV1YCsGnhJyiUKsbe9zC9J11Hws8/oNI4MfTm2STv2kZJZjqG2hqUarW8YtBiNlOQkixftzQnC6vFgkKpRKkSf20FQRD+CcT/rQXhEstPSZYDrEZWi5m9P//ArNffp8tVV9Ogr8U7OJT4EaOxWa2UZGeiUmv44cVnqC0vQ6FU0mHQMPYu/x6rxULnYVc5rGAUBEEQ/v5EkCUIl5jayanV7R7+gRgN9az830sUpZ/COySM6fNfxdnNHf/IaABufOlNMg4fwD8iGp/QcDoOHk6Dvo6AqDZ/5iMIgiAIl4AIsgThEvOLiGL4rXdxeP2veIWE4OLuiVKlpu/UG8g8dICi9FMAlOflkJqwW24ADeDq7UPnYaPYvPgzEteswN3Pn+uefemvehRBEAThIoggSxAug25jxtNtzPgW2z38A5EkBTabFQCvwKAWx1SXFpO4ZoX9dUkxe5Z/z5h7Hrqs9ysIgiBceiLIEoQ/UUB0WyY88jTpiQmEdepCWFx8i2M0Wh1KjUYu8ZC8cyvdx06QpxQFQRCEfwZRjFQQ/mRte/Vl1F0P0GHAkFb3a11dGXLj7fJ7i8nEyd3b/6zbEwRBEC4REWQJwt9QbP+BaJtVdPePFInvgiAI/zRiulAQ/oZ07h5cP/9VTu7ejn9ENDF9+v/VtyQIgiBcIBFkCcLflG9YBL5hEX/1bQiCIAh/kJguFARBEARBuAxEkCUIgiAIgnAZiCBLEARBEAThMhBBliAIgiAIwmUggixBEARBEITLQARZgiAIgiAIl4EIsgRBEARBEC4DEWQJgiAIgiBcBiLIEgRBEARBuAxEkCUIgiAIgnAZiCBLEARBEAThMhBBliAIgiAIwmXwlzSIttlsAD/nUBEAAASESURBVFRXV/8VHy8IgiAIwh/Q+Hu78fe4cG5/SZBVU1MDQFhY2F/x8YIgCIIgXISamho8PDz+6tv425Nsf0E4arVayc/Px83NDUmS/uyPFwRBEAThD7DZbNTU1BAcHIxCITKOfs9fEmQJgiAIgiBc6UQYKgiCIAiCcBmIIEsQBEEQBOEyEEGWIAiCIAjCZSCCLEEQBEEQhMtABFmC8C80dOhQHnrooRbbFy1ahKenJwDz5s1DkiTGjBnT4rjXX38dSZIYOnRoi325ubloNBo6derU6mdLkiT/5+HhwYABA9i0aZO8f9u2bYwfP57g4GAkSeLnn3/+I48oCILwlxNBliAIZxUUFMTmzZvJzc112P7FF18QHh7e6jmLFi1i2rRpVFdXs3fv3laPWbhwIQUFBezcuRNfX1/GjRtHeno6AHV1dXTp0oUPPvjg0j6MIAjCn0wEWYIgnJW/vz+jRo1i8eLF8rZdu3ZRWlrKNddc0+J4m83GwoULuemmm5gxYwYLFixo9bqenp4EBgbSqVMnPvroI+rr61m/fj0AY8eO5cUXX2Ty5MmX56EEQRD+JCLIEgThnG677TYWLVokv//iiy+48cYb0Wg0LY7dvHkzer2ekSNHMnPmTL777jvq6urOeX1nZ2cAjEbjJb1vQRCEv5oIsgRBOKdx48ZRXV3Ntm3bqKurY+nSpdx2222tHrtgwQKmT5+OUqmkU6dOREdH88MPP5z12nq9nmeffRalUsmQIUMu1yMIgiD8Jf6S3oWCIPxzqNVqZs6cycKFC0lPT6ddu3bEx8e3OK6yspJly5axY8cOedvMmTNZsGABt9xyi8OxN9xwA0qlkvr6evz8/FiwYEGr1xQEQfgnE0GWIPwLubu7U1VV1WJ7ZWVlq01fb7vtNvr06UNSUtJZR7G+/fZbDAYDffr0kbfZbDasVispKSm0a9dO3v7WW28xcuRIPDw88PPzuwRPJAiC8PcjpgsF4V8oNjaWxMTEFtsTExMdgqFGcXFxxMXFkZSUxIwZM1q95oIFC3jkkUc4dOiQ/N/hw4cZNGgQX3zxhcOxgYGBtG3bVgRYgiBc0cRIliD8C91zzz28//77PPDAA8yePRsnJydWr17NkiVLWLVqVavnbNq0CZPJJNfRau7QoUMkJibyzTff0L59e4d9N9xwA//973958cUXUal+/385tbW1nDp1Sn6fkZHBoUOH8Pb2PmvZCEEQhL8jMZIlCP9C0dHRbNu2jeTkZEaOHEmfPn1YunQpP/zwQ6vFRwFcXFxaDbDAPorVsWPHFgEWwOTJkykuLmbNmjXndW/79++nW7dudOvWDYC5c+fSrVs3/vOf/5zfwwmCIPxNSDabzfZX34QgCIIgCMKVRoxkCYIgCIIgXAYiyBIEQRAEQbgMRJAlCIIgCIJwGYggSxAEQRAE4TIQQZYgCIIgCMJlIIIsQRAEQRCEy0AEWYIgCIIgCJeBCLIEQRAEQRAuAxFkCYIgCIIgXAYiyBIEQRAEQbgMRJAlCIIgCIJwGYggSxAEQRAE4TL4f4foIlBAFHrsAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sc.pl.umap(\n",
    "    adata,\n",
    "    color=[\"leiden\"],\n",
    "    size=30,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "4293f10f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 718x345 with 4 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sc.pl.dotplot(adata, var_names=['CD3D', 'CD4', 'CD8A', 'KLRB1', 'NCAM1', 'FCGR3A', 'CD19', 'CD38', 'CD14', 'VCAN', 'FCER1A', 'CLEC4C', 'HBB', 'ITGB3'], groupby='leiden')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "984ab4e6",
   "metadata": {},
   "source": [
    "The above is a very simplistic overview of some information that will help us classify the cells present in the dataset: spatial relationships in the gene space by dimensionality reduction with UMAP and gene expression on an aggregated per-cluster level in the dotplot. While one must be careful assigning meaning to spatial relationships on a UMAP during cell-type labelling, these two plots shall suffice for the purpose of cell-type labelling in this short tutorial without any claim to completeness."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "c957a2ee",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Map Leiden clusters to cell type labels based on the dotplot and known marker genes for each cell type\n",
    "adata.obs['label'] = adata.obs['leiden'].map(\n",
    "    {\n",
    "        '0': 'CD4 T cells',\n",
    "        '1': 'Classical monocytes',\n",
    "        '2': 'CD8 T cells', # probably includes both antigen-naive and antigen-experienced B cells\n",
    "        '3': 'B cells',\n",
    "        '4': 'Non-classical monocytes',\n",
    "        '5': 'ILCs', # in the sense of both NK cells and other ILCs\n",
    "        '6': 'mixed'\n",
    "    }\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "6b149190",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "np.False_"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "adata.obs['label'].isna().any()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d12d47ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Removed mixed cluster since it is not a well-defined cell type and would likely introduce noise into the classifier.\n",
    "adata = adata[adata.obs.label != 'mixed'].copy()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "db37c657",
   "metadata": {},
   "source": [
    "## Building a hierarchy"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "82c489f7",
   "metadata": {},
   "source": [
    "This is where it gets interesting. We have assigned, by cluster, one label per cell. This is the input data cell type classifiers usually receive. \n",
    "To harness the potential of Compocyte's structure, we need to explicitly define a hierarchy on which all above labels can be found. This will help the classifier weight relationships between different cell type labels and define branching points in the classification process that can be modified by exchange or extension down the road.\n",
    "\n",
    "There is one very important assumption that Compocyte works with that is important to keep in mind when building a hierarchy. For the labels at the bottom of the hierarchy, also called leaf nodes, all **prior labels of this branch must also be true**. I. e. a dendritic cell is also a myeloid cell and it is also a blood cell. A CD8 T cell is also a T cell and it is also a lymphoid cell. \n",
    "\n",
    "Violating this assumption will lead to performance issues should you try to build your hierarchy from a more developmental point of view. The point of the hierarchy is to group transcriptomically similar cells into shared classification  branches.\n",
    "\n",
    "A simple such hierarchy would be:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "a186f8c8",
   "metadata": {},
   "outputs": [],
   "source": [
    "from Compocyte.core.tools import make_graph_from_edges\n",
    "import networkx as nx\n",
    "\n",
    "hierarchy = {\n",
    "    'Blood': {\n",
    "        'Lymphoid': {\n",
    "            'T cells': {'CD4 T cells': {}, 'CD8 T cells': {}},\n",
    "            'B cells': {},\n",
    "            'ILCs': {}\n",
    "        },\n",
    "        'Myeloid': {\n",
    "            'Classical monocytes': {}, 'Non-classical monocytes': {},\n",
    "        },        \n",
    "    }\n",
    "}\n",
    "graph = nx.DiGraph()\n",
    "make_graph_from_edges(hierarchy, graph)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "c0ff4f80",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from networkx.drawing.nx_agraph import graphviz_layout\n",
    "\n",
    "# Plot the graph structure we gave to the classifier during training.\n",
    "pos = graphviz_layout(\n",
    "    graph,prog=\"dot\",\n",
    "    root='Blood',\n",
    "    args='-Gsplines=curved -Gnodesep=8 -Goverlap=scalexy -Gbeautify=false'\n",
    ")\n",
    "\n",
    "nx.draw(\n",
    "    graph, pos,\n",
    "    with_labels=True,\n",
    "    node_color=\"#9ecae1\",\n",
    "    node_size=1200,\n",
    "    edge_color=\"#888\",\n",
    "    width=1.5,\n",
    "    font_size=10,\n",
    "    font_weight=\"bold\",\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e2a70fa3",
   "metadata": {},
   "source": [
    "## Training the classifier"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0e15f72e",
   "metadata": {},
   "source": [
    "We have now defined cell type labels and a hierarchy that fits these labels and that can tell how we want our classifier structure to be set up. However there is still a small task to be completed before we can begin training.\n",
    "\n",
    "To provide training labels for training local classifiers at every branching point, we need to infer the intermediate labels of each cell in the hierarchy. This is done by using the `infer_levels` function takes in the hierarchy, the name of the column in `adata.obs` that contains the cell type labels, the name of the root node in the hierarchy, and the `adata` object itself. It adds the level annnotations as separate obs columns in the provided AnnData object and returns as `obs_names` the column names it used. These can then be passed to the classifier so it knows where to look."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "58f40b35",
   "metadata": {},
   "outputs": [],
   "source": [
    "from Compocyte.core.tools import infer_levels\n",
    "\n",
    "# Save level labels to adata.obs and receive the list of obs columns where they have been saved.\n",
    "_, obs_names = infer_levels(\n",
    "    hierarchy=hierarchy, \n",
    "    labels='label', \n",
    "    root_node='Blood', \n",
    "    adata=adata\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "a21512c9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['Level_0', 'Level_1', 'Level_2', 'Level_3']"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "obs_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "1ae2b96f",
   "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>leiden</th>\n",
       "      <th>label</th>\n",
       "      <th>Level_0</th>\n",
       "      <th>Level_1</th>\n",
       "      <th>Level_2</th>\n",
       "      <th>Level_3</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>index</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>AAACATTGAGCTAC-1</th>\n",
       "      <td>3</td>\n",
       "      <td>B cells</td>\n",
       "      <td>Blood</td>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>B cells</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AAACATTGATCAGC-1</th>\n",
       "      <td>0</td>\n",
       "      <td>CD4 T cells</td>\n",
       "      <td>Blood</td>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>T cells</td>\n",
       "      <td>CD4 T cells</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AAACCGTGCTTCCG-1</th>\n",
       "      <td>1</td>\n",
       "      <td>Classical monocytes</td>\n",
       "      <td>Blood</td>\n",
       "      <td>Myeloid</td>\n",
       "      <td>Classical monocytes</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AAACGCACTGGTAC-1</th>\n",
       "      <td>0</td>\n",
       "      <td>CD4 T cells</td>\n",
       "      <td>Blood</td>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>T cells</td>\n",
       "      <td>CD4 T cells</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AAACGCTGACCAGT-1</th>\n",
       "      <td>2</td>\n",
       "      <td>CD8 T cells</td>\n",
       "      <td>Blood</td>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>T cells</td>\n",
       "      <td>CD8 T cells</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TTTCGAACACCTGA-1</th>\n",
       "      <td>3</td>\n",
       "      <td>B cells</td>\n",
       "      <td>Blood</td>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>B cells</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TTTCGAACTCTCAT-1</th>\n",
       "      <td>1</td>\n",
       "      <td>Classical monocytes</td>\n",
       "      <td>Blood</td>\n",
       "      <td>Myeloid</td>\n",
       "      <td>Classical monocytes</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TTTCTACTGAGGCA-1</th>\n",
       "      <td>3</td>\n",
       "      <td>B cells</td>\n",
       "      <td>Blood</td>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>B cells</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TTTCTACTTCCTCG-1</th>\n",
       "      <td>3</td>\n",
       "      <td>B cells</td>\n",
       "      <td>Blood</td>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>B cells</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TTTGCATGAGAGGC-1</th>\n",
       "      <td>3</td>\n",
       "      <td>B cells</td>\n",
       "      <td>Blood</td>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>B cells</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1792 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                 leiden                label Level_0   Level_1  \\\n",
       "index                                                            \n",
       "AAACATTGAGCTAC-1      3              B cells   Blood  Lymphoid   \n",
       "AAACATTGATCAGC-1      0          CD4 T cells   Blood  Lymphoid   \n",
       "AAACCGTGCTTCCG-1      1  Classical monocytes   Blood   Myeloid   \n",
       "AAACGCACTGGTAC-1      0          CD4 T cells   Blood  Lymphoid   \n",
       "AAACGCTGACCAGT-1      2          CD8 T cells   Blood  Lymphoid   \n",
       "...                 ...                  ...     ...       ...   \n",
       "TTTCGAACACCTGA-1      3              B cells   Blood  Lymphoid   \n",
       "TTTCGAACTCTCAT-1      1  Classical monocytes   Blood   Myeloid   \n",
       "TTTCTACTGAGGCA-1      3              B cells   Blood  Lymphoid   \n",
       "TTTCTACTTCCTCG-1      3              B cells   Blood  Lymphoid   \n",
       "TTTGCATGAGAGGC-1      3              B cells   Blood  Lymphoid   \n",
       "\n",
       "                              Level_2      Level_3  \n",
       "index                                               \n",
       "AAACATTGAGCTAC-1              B cells               \n",
       "AAACATTGATCAGC-1              T cells  CD4 T cells  \n",
       "AAACCGTGCTTCCG-1  Classical monocytes               \n",
       "AAACGCACTGGTAC-1              T cells  CD4 T cells  \n",
       "AAACGCTGACCAGT-1              T cells  CD8 T cells  \n",
       "...                               ...          ...  \n",
       "TTTCGAACACCTGA-1              B cells               \n",
       "TTTCGAACTCTCAT-1  Classical monocytes               \n",
       "TTTCTACTGAGGCA-1              B cells               \n",
       "TTTCTACTTCCTCG-1              B cells               \n",
       "TTTGCATGAGAGGC-1              B cells               \n",
       "\n",
       "[1792 rows x 6 columns]"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "adata.obs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "326ea424",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.14/site-packages/sklearn/feature_selection/_univariate_selection.py:110: UserWarning: Features [  40  197  229  426  440  484  554  882  951  978 1084 1195 1341 1414\n",
      " 1446 1467 1639 1712 1864 1870 1942] are constant.\n",
      "  warnings.warn(\"Features %s are constant.\" % constant_features_idx, UserWarning)\n",
      "/usr/local/lib/python3.14/site-packages/sklearn/feature_selection/_univariate_selection.py:111: RuntimeWarning: invalid value encountered in divide\n",
      "  f = msb / msw\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training at Blood.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.14/site-packages/sklearn/feature_selection/_univariate_selection.py:110: UserWarning: Features [  24   40   67  127  151  168  186  197  207  216  229  233  243  277\n",
      "  283  313  321  339  412  426  440  455  473  484  516  549  554  560\n",
      "  568  661  681  707  726  794  851  882  901  902  937  940  951  961\n",
      "  969  978  990 1015 1019 1024 1054 1084 1123 1131 1150 1184 1195 1196\n",
      " 1206 1207 1219 1259 1289 1321 1341 1349 1397 1411 1414 1431 1446 1467\n",
      " 1491 1504 1524 1535 1607 1639 1652 1674 1677 1697 1699 1712 1713 1755\n",
      " 1765 1832 1864 1870 1873 1882 1899 1906 1915 1942] are constant.\n",
      "  warnings.warn(\"Features %s are constant.\" % constant_features_idx, UserWarning)\n",
      "/usr/local/lib/python3.14/site-packages/sklearn/feature_selection/_univariate_selection.py:111: RuntimeWarning: invalid value encountered in divide\n",
      "  f = msb / msw\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training at Lymphoid.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.14/site-packages/sklearn/feature_selection/_univariate_selection.py:110: UserWarning: Features [  11   12   22   24   40   47   52   67   81   90  107  115  127  151\n",
      "  168  186  189  197  207  216  226  229  231  233  237  239  243  267\n",
      "  277  280  283  304  313  315  317  318  321  325  328  339  347  354\n",
      "  362  364  398  399  412  423  426  437  440  455  473  484  494  515\n",
      "  516  529  532  549  554  560  562  568  573  587  591  629  661  681\n",
      "  690  707  719  722  726  744  746  749  767  781  788  794  805  807\n",
      "  843  851  870  882  885  889  896  901  902  924  937  940  949  951\n",
      "  961  969  974  978  982  990  999 1014 1015 1018 1019 1022 1024 1050\n",
      " 1053 1054 1059 1060 1076 1084 1100 1123 1131 1132 1150 1151 1152 1153\n",
      " 1154 1182 1184 1194 1195 1196 1198 1199 1206 1207 1214 1219 1223 1230\n",
      " 1240 1259 1273 1278 1280 1289 1301 1321 1341 1343 1344 1349 1375 1385\n",
      " 1386 1397 1410 1411 1414 1426 1431 1436 1446 1450 1461 1467 1474 1475\n",
      " 1491 1504 1520 1524 1535 1541 1542 1551 1561 1562 1570 1586 1589 1607\n",
      " 1639 1652 1669 1674 1677 1695 1697 1699 1712 1713 1734 1755 1765 1783\n",
      " 1799 1832 1837 1842 1843 1864 1870 1873 1876 1882 1899 1906 1912 1915\n",
      " 1918 1926 1933 1940 1942 1957 1978] are constant.\n",
      "  warnings.warn(\"Features %s are constant.\" % constant_features_idx, UserWarning)\n",
      "/usr/local/lib/python3.14/site-packages/sklearn/feature_selection/_univariate_selection.py:111: RuntimeWarning: invalid value encountered in divide\n",
      "  f = msb / msw\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training at T cells.\n",
      "Training at Myeloid.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.14/site-packages/sklearn/feature_selection/_univariate_selection.py:110: UserWarning: Features [   1    6   11   12   18   26   30   40   41   48   52   57   61   77\n",
      "   79   84   90   92   96  100  101  103  104  107  108  115  117  131\n",
      "  136  166  170  176  178  179  185  188  189  191  197  208  209  219\n",
      "  226  229  231  237  239  253  254  258  266  267  275  276  278  285\n",
      "  286  289  306  315  317  318  325  328  337  354  364  381  398  399\n",
      "  403  406  407  411  422  423  424  426  429  438  440  441  447  459\n",
      "  467  471  472  480  484  494  508  510  530  541  546  551  554  557\n",
      "  565  573  584  587  591  598  611  619  620  629  631  671  672  692\n",
      "  711  719  722  732  742  744  749  756  758  761  764  767  781  784\n",
      "  788  805  807  808  812  814  838  843  846  847  848  849  850  852\n",
      "  861  864  870  880  882  885  887  894  895  896  909  912  941  945\n",
      "  949  951  954  964  974  978  979  982  991 1002 1014 1018 1022 1029\n",
      " 1032 1056 1057 1059 1060 1063 1069 1074 1076 1084 1093 1100 1103 1108\n",
      " 1109 1115 1151 1153 1160 1168 1179 1186 1187 1190 1191 1194 1195 1204\n",
      " 1215 1221 1224 1228 1230 1234 1242 1243 1247 1248 1250 1273 1279 1280\n",
      " 1284 1287 1300 1301 1313 1331 1341 1344 1359 1360 1365 1372 1375 1376\n",
      " 1377 1385 1386 1388 1390 1392 1398 1405 1409 1410 1413 1414 1418 1421\n",
      " 1426 1430 1436 1439 1446 1452 1453 1455 1458 1461 1466 1467 1474 1475\n",
      " 1489 1501 1518 1520 1526 1533 1534 1542 1543 1547 1551 1556 1562 1567\n",
      " 1568 1570 1578 1579 1586 1587 1618 1623 1632 1637 1639 1642 1660 1669\n",
      " 1671 1672 1684 1689 1695 1701 1702 1703 1708 1712 1715 1720 1724 1734\n",
      " 1739 1740 1754 1759 1768 1775 1777 1783 1788 1798 1800 1801 1808 1813\n",
      " 1823 1825 1837 1849 1851 1861 1864 1866 1870 1871 1876 1883 1894 1897\n",
      " 1904 1908 1911 1912 1918 1924 1926 1933 1935 1938 1942 1955 1957 1971\n",
      " 1978 1982] are constant.\n",
      "  warnings.warn(\"Features %s are constant.\" % constant_features_idx, UserWarning)\n",
      "/usr/local/lib/python3.14/site-packages/sklearn/feature_selection/_univariate_selection.py:111: RuntimeWarning: invalid value encountered in divide\n",
      "  f = msb / msw\n"
     ]
    }
   ],
   "source": [
    "from Compocyte.core.hierarchical_classifier import HierarchicalClassifier\n",
    "from Compocyte.core.models.log_reg import LogisticRegression\n",
    "\n",
    "# Train the hierarchical classifier. This will train a separate classifier for each parent node in the hierarchy.\n",
    "classifier = HierarchicalClassifier(\n",
    "    save_path=\"./exclude/pbmc_classifier\", \n",
    "    adata=adata, \n",
    "    root_node='Blood', \n",
    "    dict_of_cell_relations=hierarchy, \n",
    "    obs_names=obs_names)\n",
    "# For training speed, set the classifier type for all nodes to logistic regression. \n",
    "# In practice, one would likely want to experiment with different classifier types for different nodes in the hierarchy.\n",
    "# The default is a 2-layer FCNN with 64 nodes each.\n",
    "for node in classifier.graph.nodes:\n",
    "    classifier.set_classifier_type(node, LogisticRegression)\n",
    "\n",
    "classifier.train_all_child_nodes()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "819d52a9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicting at Blood.\n",
      "Predicting at Lymphoid.\n",
      "Predicting at T cells.\n",
      "Predicting at Myeloid.\n"
     ]
    }
   ],
   "source": [
    "# Make sure the classifier has been trained by predicting on the test data. \n",
    "# This will add columns with predicted labels to adata.obs for each parent node in the hierarchy.\n",
    "classifier.load_adata(test_adata)\n",
    "classifier.predict_all_child_nodes('Blood')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "97286ade",
   "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>Level_1_pred</th>\n",
       "      <th>Level_2_pred</th>\n",
       "      <th>Level_3_pred</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>index</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>GCTCAAGAACCATG-1</th>\n",
       "      <td>Myeloid</td>\n",
       "      <td>Non-classical monocytes</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TATTTCCTGGTGTT-1</th>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>T cells</td>\n",
       "      <td>CD4 T cells</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TATAAGTGTGGTGT-1</th>\n",
       "      <td>Myeloid</td>\n",
       "      <td>Non-classical monocytes</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>AGCACTGATGCTTT-1</th>\n",
       "      <td>Myeloid</td>\n",
       "      <td>Classical monocytes</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>GAAACAGACATTCT-1</th>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>T cells</td>\n",
       "      <td>CD4 T cells</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CTTAGACTAAACGA-1</th>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>T cells</td>\n",
       "      <td>CD4 T cells</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CTAGAGACTTTGGG-1</th>\n",
       "      <td>Myeloid</td>\n",
       "      <td>Non-classical monocytes</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TCATCAACTGTTCT-1</th>\n",
       "      <td>Myeloid</td>\n",
       "      <td>Classical monocytes</td>\n",
       "      <td></td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>TAAGATTGTTGCTT-1</th>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>T cells</td>\n",
       "      <td>CD4 T cells</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>CTTGAACTACGCAT-1</th>\n",
       "      <td>Lymphoid</td>\n",
       "      <td>T cells</td>\n",
       "      <td>CD4 T cells</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>900 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Level_1_pred             Level_2_pred Level_3_pred\n",
       "index                                                              \n",
       "GCTCAAGAACCATG-1      Myeloid  Non-classical monocytes             \n",
       "TATTTCCTGGTGTT-1     Lymphoid                  T cells  CD4 T cells\n",
       "TATAAGTGTGGTGT-1      Myeloid  Non-classical monocytes             \n",
       "AGCACTGATGCTTT-1      Myeloid      Classical monocytes             \n",
       "GAAACAGACATTCT-1     Lymphoid                  T cells  CD4 T cells\n",
       "...                       ...                      ...          ...\n",
       "CTTAGACTAAACGA-1     Lymphoid                  T cells  CD4 T cells\n",
       "CTAGAGACTTTGGG-1      Myeloid  Non-classical monocytes             \n",
       "TCATCAACTGTTCT-1      Myeloid      Classical monocytes             \n",
       "TAAGATTGTTGCTT-1     Lymphoid                  T cells  CD4 T cells\n",
       "CTTGAACTACGCAT-1     Lymphoid                  T cells  CD4 T cells\n",
       "\n",
       "[900 rows x 3 columns]"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classifier.adata.obs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "230e62b3",
   "metadata": {},
   "outputs": [],
   "source": [
    "classifier.save()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b0788533",
   "metadata": {},
   "source": [
    "Congratulations. You have trained your own Compocyte classifier. Check out the other tutorials to see how it can be used!"
   ]
  }
 ],
 "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": "3.14.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}