Last active
June 21, 2023 15:53
-
-
Save ilmonteux/c8e00f21e1ef99dbb8832cdc7410a8b8 to your computer and use it in GitHub Desktop.
Convolutional Neural Networks for jet classification - barebone pipeline
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
{ | |
"cells": [ | |
{ | |
"cell_type": "markdown", | |
"metadata": { | |
"toc": true | |
}, | |
"source": [ | |
"<h1>Table of Contents<span class=\"tocSkip\"></span></h1>\n", | |
"<div class=\"toc\"><ul class=\"toc-item\"><li><span><a href=\"#Generate-jet-images\" data-toc-modified-id=\"Generate-jet-images-1\">Generate jet images</a></span><ul class=\"toc-item\"><li><span><a href=\"#plot-some-jets\" data-toc-modified-id=\"plot-some-jets-1.1\">plot some jets</a></span></li></ul></li><li><span><a href=\"#Training-neural-networks\" data-toc-modified-id=\"Training-neural-networks-2\">Training neural networks</a></span><ul class=\"toc-item\"><li><span><a href=\"#run-on-Colab\" data-toc-modified-id=\"run-on-Colab-2.1\">run on Colab</a></span></li><li><span><a href=\"#analyze\" data-toc-modified-id=\"analyze-2.2\">analyze</a></span><ul class=\"toc-item\"><li><span><a href=\"#Check-effect-of-pixel-normalization\" data-toc-modified-id=\"Check-effect-of-pixel-normalization-2.2.1\">Check effect of pixel normalization</a></span></li><li><span><a href=\"#Compare-to-high-pT-jets\" data-toc-modified-id=\"Compare-to-high-pT-jets-2.2.2\">Compare to high pT jets</a></span></li></ul></li></ul></li></ul></div>" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"See details and walkthrough for this notebook at https://ilmonteux.github.io/2018/10/15/jet-tagging-cnn.html" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 1, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"import sys, os\n", | |
"import matplotlib.pyplot as plt\n", | |
"import matplotlib as mpl\n", | |
"import numpy as np\n", | |
"\n", | |
"from pythia_to_images import *" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"# Generate jet images" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"In my setup, the notebook directory has two symbolic links, `madgraph` and `pythia8219`, that point to the Madgraph and Pythia installation directories." | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 2, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"cmap = plt.get_cmap('gray_r')\n", | |
"nevents = 50000" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 4, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"outdir = 'images_out/'\n", | |
"if not os.path.isdir(outdir): os.system('mkdir {}'.format(outdir))\n", | |
"cwd = os.getcwd()" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 4, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"write_mg_cards([500,700], nevents=nevents)" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"Generate tt jets, run through pythia and save corresponding jet images. If anything does not work, run same Madgraph commands through the command line" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 7, | |
"metadata": { | |
"scrolled": false | |
}, | |
"outputs": [], | |
"source": [ | |
"! cd madgraph; bin/mg5_aMC $cwd/generate_tt.mg5" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": null, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"lhe_file_name = 'madgraph/jets_tt/Events/run_01_decayed_1/unweighted_events.lhe.gz'\n", | |
"leading_jet_images, all_jet_images, jetpep = run_pythia_get_images(lhe_file_name, PTRANGE=[500,700], PTRANGE2=[450,700], plot_first_few=False)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": null, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"np.savez_compressed(outdir+'tt_leading_jet.npz', leading_jet_images)\n", | |
"np.savez_compressed(outdir+'tt_all_jets.npz', all_jet_images)\n", | |
"np.savez_compressed(outdir+'tt_jetpep.npz', jetpep)" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"Generate QCD jets, run through pythia and save corresponding jet images" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": null, | |
"metadata": { | |
"scrolled": true | |
}, | |
"outputs": [], | |
"source": [ | |
"! cd madgraph; bin/mg5_aMC $cwd/generate_qcd.mg5" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": null, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"lhe_file_name = 'madgraph/jets_qcd/Events/run_01/unweighted_events.lhe.gz'\n", | |
"leading_jet_images, all_jet_images, jetpep = run_pythia_get_images(lhe_file_name, PTRANGE=[500,700], PTRANGE2=[450,700], plot_first_few=False)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": null, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"np.savez_compressed(outdir+'qcd_leading_jet.npz', leading_jet_images)\n", | |
"np.savez_compressed(outdir+'qcd_all_jets.npz', all_jet_images)\n", | |
"np.savez_compressed(outdir+'qcd_jetpep.npz', jetpep)" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"## plot some jets" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 5, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"leading_jet_images = np.load(outdir+'tt_leading_jet.npz')['arr_0']\n", | |
"all_jet_images = np.load(outdir+'tt_all_jets.npz')['arr_0']\n", | |
"jetpep = np.load(outdir+'tt_jetpep.npz')['arr_0']" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 6, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"leading_jet_images0 = np.load(outdir+'qcd_leading_jet.npz')['arr_0']\n", | |
"all_jet_images0 = np.load(outdir+'qcd_all_jets.npz')['arr_0']\n", | |
"jetpep0 = np.load(outdir+'qcd_jetpep.npz')['arr_0']" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"Pad leading jet images so they all have the same pixels." | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 7, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"std_jet_images0 = map(pad_image, leading_jet_images0)\n", | |
"std_jet_images = map(pad_image, leading_jet_images)" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"Plot a few events, first only the leading jet and then the whole event" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 8, | |
"metadata": { | |
"scrolled": false | |
}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 864x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
}, | |
{ | |
"data": { | |
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAqAAAADUCAYAAABDEJE7AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzt3X+4XVV95/H3JxB+CQo0lxggMVSjI9IafW6RSnUo+AOo0+CM+OAgRss0+hQEqq0CnXlq69ChrUpprbTXggmWCjwqJUNDAVHGWgVMMEBCtAa4hUjMj/JDEEVDvvPHXhfOvff82Pfsc+5Z55zP63n2k3P2Xmfvde5NPll7r7XXVkRgZmZmZjZb5vS6AmZmZmY2XNwANTMzM7NZ5QaomZmZmc0qN0DNzMzMbFa5AWpmZmZms8oNUDMzMzObVW6AmpmZmdmscgPUzMzMzGaVG6AZk7RR0nG9roeZmdXnnDZrjxugJUl6r6R7JT0t6YeSPiPpRVPK/HdJayU9JWmrpBsl/VraNi7pJ5KelPS4pG9K+oCkhr+DiHhVRNxWsn7jkt5U6Ut2kaTbJP00/WyekvS9KdtfKemrkp6QtFnS22u27S3pckn/nn5+35F0UovjnSbpDkk/lrQ9vf4dSSpR15sk/XGd9cvS737PmXx3M+utbubjMOV0KnOapE0pW++X9Iayn22wrxnntDN6MLgBWoKkDwN/Cvw+8CLgGGAxcLOkuanMh4C/AP4EmA8sAj4DLKvZ1X+JiAOAlwAXAx8FLp+db5GFsyNi/7S8YmJlCovrgRuAg4EVwN9LenkqsifwMPCfKX7+/wu4VtLiegdJv69LgT8HXkzx+/gAcCywV4l6rgTOqBOCZwBXRcSuEvswM+tHdXMaQNKbKf4vfB9wAPBG4IEyn52qYk6vxBnd/yLCS5MFeCHwFPDOKev3B7YDyykaRU8BpzbZzzjwpinrjgZ2A0eV+QxwKPAlYAfwIHBOWv/5tJ+fpHp8pOR3Ox34JnAN8EOKRt5JXfo53gb8jwbbjkr1Vs26m4GPN9nfPcB/q7P+RcCP622rU7bRz3Nf4AngjTVlDwJ+Cry6138nvXjxUn6pl4/AK1MmPQ5sBH5zymfGgQuA+4DHgM8B+zTY/3M53ShTGtWjRN1nLaPT8RrmdNr+TeDMdj47pWypnHZGD/biK6CtvR7YB/hy7cqIeAq4EXgL8KupzHUz2XFE3AlsAd7Qqmzqqv+/wN3AYcAJwHmS3hoRZwAPUVxh3T8i/ix95jOSPtNkt78EvIbiH/hCirPRvylTd0k3pKEE9ZYbGnzs/0jaKelfp4yZqtfdIoqGab1jzwdeTvEfx1S/CuxNcUW1Wf2b/Tx/AlwLvKfmI+8EvhsRdzfbr5nlZWo+ApdQ/Nu/GTgE+CBwlaSpV+xOB94KvJQib/5ns+M0y5R69YiIP+tmRqc6dSynJe0BjAIjaZjUFkmflrRvq8/W0TKnndFDoNct4NwX4N3ADxtsu5gixE5vVKam7DhTroCm9bcDf9DqM8DrgIembL8A+Fyz/beo0z8Bf1Lz/hAgKAL5Noozzo3p9e9V/Dm+jqLLZm+Kq8ZPAi9N2+ZSdON8JL1+C/Az4KY6+5kLfAX427K/L4qz9scprjy8seTP89cozrD3Te//FfjdXv999OLFy8yXKVn6BoqriXNqtn8B+NiU8h+oeX8ycH+zfbfKlKn1KFnvRhm9Tw9y+tB07LXAAmBeysWLWn22znFa5rQzevAXXwFtbScwr8Gg5gUUXQP/0aRMK4cBj5Yo9xLg0NozWOBCinEz7fol4Is17w8BnoqIv4qI4yga1ysi4riI+ESF4xARd0TEkxHxTESsogiLk9O2nwOnAL9B8R/DhynObrfU7iOdEX+eonF6doNDTftdRMTrI+LAtG3i73zTn2dEfIPid7tM0i8CvwL8Q5WfgZll4VDg4YjYXbPu3ymyuNbDU7Yf2mK/s5nRP53tnKZoGAL8VURsjYidwKd4PsebfXaqMjntjB5wvlOstW8BzwD/laJRBICkFwAnUXTLfIti7MkpTA6LpiT9CkXofaNE8YeBByNiSYPtUfa46dgHUnTp7KhZ/Q6KYQUTXkUxBqre52+k8dCBf4mIpnepU9T3ua73iLiH4iajif1/E1hV814UN2zNB05OjdZ6Jn5fyyi6rRpp9fMEuJKii+cVwM0Rsa3ZFzKzbNXm4yPAQklzahqhi4B/m/KZhTWvF6XPNVMmU0rndMmMhlnK6Yh4TNIWyn+HSRk/RZmcdkYPul5fgu2HhaJreBtwIkUX8GJgDcU/+hekMh9KZU4B9kvlTgL+LG0f5/kuoBcCbwPuB65sctzaz+wBrKO4c37f9P4o4FfS9tspzoLLfqc3ALvS/vakuPq4HTiypsyGDv38DqQYS7VPOtbpFAPQX1FT5pfT9v2A36PoVtq7ZvvfpO+4/wx+X++guFlsDrCU4maC48r8PFOZxRRXW7fQ5AYzL1685L3U5iPFHdb3A+ennD6Oorv4P9WUHwfuBQ6nmJnjX6jpCp+y73GKLvgymVI6p8tkdCo3mzn9x8C3Ka7EHpR+Lh8v89k6x2ua087owV96XoF+WYAzgQ0UVzqDYrzNoVPKnE4xPubHFF3J/wS8Pm0bp+jCeJJi3Mq3gLOAPZocc5zpd8F/Ie37sRRmEw3UZRQD3B8njQOiaLT9TYN9/w7FFcZ/THVaO1HXtP0wijPKTvzsRlJoPZnqdzvw5ill/jx9p4mbu15Ws+0l6Wf+07R9Yjm9yTFPB+4Enqa4gnAHxfROe5X5edaUuS1t27vqz8GLFy+9WabmI8VVw/+Xsvg+4O1Tyo/z/F3wj6es3K/Bvp/L6VaZUqcebWd0KjPbOT2XYnrBx9N3/EuKRmfLzzY4ZtOcdkYP9qL0C7QZkPRbwB8Bx0bEQ108zkPAuyPi613Y92XAv0XEJQ22vxU4MSJ+t9PHNjPLmaRxiimFvlKibFdyulVGpzLOaetbHgPahoi4QtLPKaZo6koDVNIIxVnleDf2TzG4vdlURQ3HFZmZWddzulVGg3Pa+pgboG2KiM93a9/p5qRbKO427NYV1qOA7zbaGBGf6tJxzcz63izkdNOMBue09Td3wZuZmZnZrPI8oGZmZmY2q9wANTMzM7NZNatjQOfNmxeLFy+ezUOaWTI+Ps7OnTsbTQzdlKQyY3VuiogT29m/5cmZbdY7VTIbSuV2TzO7UgNU0onApRQTxP5dRFzcrPzixYtZu3ZtlUOaWZtGR0crfX7OnOYdJrt3755X6QA2K2aS285ss96pmtnQPLd7ndltd8FL2gP4a4qn/RwJvEvSkZ2qmJnlRVLTxfLn3DYbLjlndpUxoEcDmyPigYj4GXA1xVMezGzASGLOnDlNF+sLzm2zIdEqt3utSg0OAx6ueb8lrTOzAeQroAPBuW02RKpktqR9JN0p6W5JGyX9UVp/hKQ7JH1f0jWS9krr907vN6fti5vtv0oDtF7tpw14lbRC0lpJa3fs2FHhcGbWS1WugDYJspWSHpS0Pi1L03pJ+ssUZPdIeu0sfMVh0DK3ndlmg6PiFdBngOMj4tXAUuBESccAfwpcEhFLgMeAM1P5M4HHIuJlwCWpXOO6tfmdoDhzXljz/nDgkamFImIsIkYjYnRkZKTC4cyslypeAW0UZAC/HxFL07I+rTsJWJKWFcBlXfhKw6hlbjuzzQZHlSugUXgqvZ2blgCOB76Y1q8CTkmvl6X3pO0nqMmBqjRAvw0sSZdi9wJOA1ZX2J+ZZarqGNAmQdbIMuDK9LnbgQMlLejIlxluzm2zIVFiDOi8id6OtKyos489JK0HtlM8evZ+4PGI2JWK1A7jeW6IT9r+BPALjerXdgM07fxs4CZgE3BtRGxsd39mlreqNyFNDbKIuCNtuih1s18iae+0zmMVu8C5bTZcWmT2zonejrSMTf18RDwbEUspekuOBl5Z5zATFxNKDc2cUGke0IhYA6ypsg8z6w8lumzmSaqdNHKsNtAi4llgqaQDgeskHQVcAPwQ2AsYAz4K/DEzDDIrz7ltNjw6dYNoRDwu6TbgGIoeqT3TCW3tMJ6JIT5bJO0JvAh4tNE+e38fvpllr2QXfMuzaSiCDLgNODEitqZu9meAz1GcYUPJMeZmZlZf1WmYJI2kCwZI2hd4E0XPydeAd6Riy4Hr0+vV6T1p+1cjouGFAzdAzayUKjchNQiy706M60wD1U8BNqSPrAbek+6GPwZ4IiK2duu7mZkNoopT5y0AvibpHorx47dExA0UPVUfkrSZYozn5an85cAvpPUfAs5vtvNZfRa8mfWvihMXLwBWqXgSzxyKsYc3SPqqpBGKLvf1wAdS+TXAycBm4GngfVUObmY2jKrkdkTcA7ymzvoHeL63qnb9T4FTy+7fDVAza2kGZ8x1NQmy4xuUD+Cstg9oZjbkquZ2t7kBamal5PDoNjMzKy/n3HYD1MxKyflM2szMpss5t90ANbOWJu6mNDOz/pB7brsBamal5BxkZmY2Xc657QaomZWSc1eOmZlNl3NuuwFqZi3l3pVjZmaT5Z7bboCaWSk5n0mbmdl0Oee2G6BmVkrOZ9JmZjZdzrndds0kLZT0NUmbJG2UdG4nK2ZmeanyKE7Lg3PbbLjknNlVroDuAj4cEXdJOgBYJ+mWiLivQ3Uzs0zkPpbISnNumw2J3HO77QZoRGwFtqbXT0raBBwGOMjMBlDOQWblOLfNhkvOud2RMaCSFlM85/mOOttWACsAFi1a1InDmVkP5NBlY53TKLed2WaDI+fcrtw0lrQ/8CXgvIj40dTtETEWEaMRMToyMlL1cGbWAxNdOc0W6x/NctuZbTYYWuV2r1WqgaS5FCF2VUR8uTNVMrMcVbkJSdI+ku6UdHe6+eWP0vojJN0h6fuSrpG0V1q/d3q/OW1f3PUvOCSc22bDI+ebkKrcBS/gcmBTRHyqc1UysxxVvAL6DHB8RLwaWAqcKOkY4E+BSyJiCfAYcGYqfybwWES8DLgklbOKnNtmw2VQr4AeC5wBHC9pfVpO7lC9zCwjra5+tjqbjsJT6e3ctARwPPDFtH4VcEp6vSy9J20/QTmcsvc/57bZkMh96rwqd8F/A+j9NzCzWVH1jFnSHsA64GXAXwP3A49HxK5UZAvFHdmkPx8GiIhdkp4AfgHYWakSQ865bTZcquS2pIXAlcCLgd3AWERcKuljwG8DO1LRCyNiTfrMBRQ9WM8C50TETY327ychmVkpJYJsnqS1Ne/HImJs4k1EPAsslXQgcB3wyjr7iPRnvUZS1FlnZmYNVLxwUHfe4LTtkoj4RG1hSUcCpwGvAg4FviLp5Sn7p3ED1MxaKtllszMiRlsViojHJd0GHAMcKGnPdBX0cOCRVGwLsBDYImlP4EXAo+3W38xs2FTtam8yb3Ajy4CrI+IZ4EFJm4GjgW/VK9z7Uahm1heq3IQkaSRd+UTSvsCbgE3A14B3pGLLgevT69XpPWn7VyPCV0DNzGagRWbPk7S2ZlnRaD915g0+W9I9kq6QdFBa99zQqaR2WNU0vgJqZqVUHLS+AFiVxoHOAa6NiBsk3QdcLel/A9+huEOb9Ofn0xn0oxTdOmZmNgMtcrtUr9XUeYMlXQZ8nGJY1MeBTwK/xQyHTrkBamYtVX2mcETcQ3H2PHX9AxRdNFPX/xQ4te0DmpkNuU48C77evMERsa1m+2eBG9LbiaFTE2qHVU3jLngzKyX3KT3MzGyyKpndaN5gSQtqir0d2JBerwZOSw8SOQJYAtzZaP++AmpmpeQwcbGZmZVXMbcn5g2+V9L6tO5C4F2SllJ0r48D7weIiI2SrgXuo7iD/qxGd8CDG6BmVkInunLMzGz2dGDoVKN5g9c0+cxFwEVl9u8GqJmV4m52M7P+knNuuwFqZqX4CqiZWX/JObfdADWzUnI+kzYzs+lyzu3KDdA0r99a4AcR8bbqVTKz3HgM6OBwZpsNh9xzuxNXQM+leKLJCzuwLzPLVM5n0jYjzmyzIZFzbldqGks6HPgN4O86Ux0zy1WVR3FaHpzZZsMl58yuegX0L4CPAAd0oC5mlqncu3KsNGe22ZDIPbfbrpmktwHbI2Jdi3IrJh50v2PHjnYPZ2Y95ich9TdnttnwyTmzqzSNjwV+U9I4cDVwvKS/n1ooIsYiYjQiRkdGRioczsx6yV3wfc+ZbTZkcs7stmsQERdExOERsRg4DfhqRLy7YzUzs2y0uvqZw9m0NefMNhsuuWd275vAZtYXqlwBlbRQ0tckbZK0UdK5af3HJP1A0vq0nFzzmQskbZb0PUlv7fLXMzMbODlfAe3IRPQRcRtwWyf2ZWZ5qnjGvAv4cETcJekAYJ2kW9K2SyLiE1OOdSTFVbpXAYcCX5H08oh4tkolrODM7m9PP/10qXL77bdfl2tiucvhSmcjfhKSmbVU9W7KiNgKbE2vn5S0CTisyUeWAVdHxDPAg5I2A0cD32q7EmZmQ2Rg74I3s+HSqZuQJC0GXgPckVadLekeSVdIOiitOwx4uOZjW2jeYDUzsyly7oLvfQ3MrC+UGNA+b2L6nrSsqLOP/YEvAedFxI+Ay4CXAksprpB+cqJonSpEV76YmdmAyvkmJHfBm1lLJbtydkbEaJN9zKVofF4VEV8GiIhtNds/C9yQ3m4BFtZ8/HDgkTaqbmY2lNwFb2YDocqUHioKXA5siohP1axfUFPs7cCG9Ho1cJqkvSUdASwB7uzoFzIzG3C+Ampmfa/imfSxwBnAvZLWp3UXAu+StJSie30ceD9ARGyUdC1wH8Ud9Gf5Dngzs5mpktuSFgJXAi8GdgNjEXGppIOBa4DFFLn9zoh4LF1ouBQ4GXgaeG9E3NVo/26AmllLVc+YI+Ib1B/XuabJZy4CLmr7oGZmQ6wDVzobTZ/3XuDWiLhY0vnA+cBHgZMoequWAK+jGOP/ukY7dwPUzErJeSyRmZlN16Xp85YBx6ViqyjmFP5oWn9lRARwu6QDJS1I+5nGDVAzK8UNUDOz/tIit+dJWlvzfiwixuoVnDJ93vyJRmVEbJV0SCrWaPo8N0DNrD25DFo360fPPltu+PL9999fqtyNN95Yqty5555bqly/2L17d6lyPlkulMjtpjOX1Oxn0vR5TfY5o+nz3AA1s1Ic6mZm/aVqbtebPg/YNtG1nmYy2Z7Wz2j6vEo1S/37X5T0XUmbJP1qlf2ZWb6qTMNk+XBumw2PKpndaPo8imnylqfXy4Hra9a/R4VjgCcajf+E6ldALwX+OSLeIWkvYL+K+zOzDOU+obHNiHPbbAh0ILcbTZ93MXCtpDOBh4BT07Y1FFMwbaaYhul9zXbedgNU0guBN1Lcjk9E/Az4Wbv7M7O8+Spn/3Numw2XLk2fB3BCnfIBnFV2/1Waxr8I7AA+J+k7kv5O0gsq7M/MMjZnzpymi/UF57bZEMk5s6vUYE/gtcBlEfEa4McUk5FOImmFpLWS1u7YsaPC4cysVya6cnIOMyulZW47s80GQ6vc7rUqNdgCbImIO9L7L1IE2yQRMRYRoxExOjIyUuFwZtZLvglpILTMbWe22eDIObPbboBGxA+BhyW9Iq06geK5zWY2gHwFtP85t82GS86ZXfUu+A8CV6U7KR+gxR1PZta/cjhjto5wbpsNiZxzu1IDNCLWAy1n0Tez/uZpmAaHc9tsOOSe234SUkY+/elPlyp36qmnti4EzJ8/v0p1zCapciYtaSFwJfBiYDfFM4cvlXQwcA2wGBgH3hkRj6UJkC+lmFPuaeC9EXFXpS9g1iNl/+0sXLiwdSHgvPPOK1XunHPOKVUO8r5SNuHnP/95qXJ77LFHqXJ77jn4TaCcf6/5No3NLCsVx4DuAj4cEa8EjgHOknQkxR3Yt0bEEuBWnr8j+yRgSVpWAJd14zuZmQ2yQR4DamZDoGpXTnoc29b0+klJm4DDgGXAcanYKuA24KNp/ZVpYuPb0+MjFzR7rJuZmT3PXfBmNhA61ZUjaTHwGuAOYP5EozIitko6JBU7DHi45mNb0jo3QM3MSsq5C94NUDMrpcSZ9DxJa2vej0XEWG0BSfsDXwLOi4gfNQnHehuibF3NzKxUbveMG6Bm1lLJiYt3RkTDu6slzaVofF4VEV9Oq7dNdK1LWgBsT+u3ALV3ZBwOPNJe7c3Mhk8uE843km/T2MyyUuUmpHRX++XApoj4VM2m1cDy9Ho5cH3N+veocAzwhMd/mpnNjG9CMrO+V/FM+ljgDOBeSevTuguBi4FrJZ0JPARMzDG2hmIKps0U0zB5snQzsxnK+QqoG6Bm1lIH7oL/BvXHdULxOMip5QM4q+0DmpkNOd8Fb2YDIecgMzOz6XLObTdAM/LBD36wVLkzzjijyzUxmy7nrhyznJVtBOy7776lyhUdBMOn7M+x7JOQhkHFJ9hdAbwN2B4RR6V1HwN+G9iRil0YEWvStguAM4FngXMi4qZm+3cD1Mxayr0rx8zMJutAbq8EPk3xGOVal0TEJ6Yc60jgNOBVwKHAVyS9PCKebbTzSjWT9LuSNkraIOkLkvapsj8zy9fElB6NFusPzm2z4VElsyPi68CjJQ+1DLg6Ip6JiAcpbiA9utkH2m6ASjoMOAcYTZdm96Bo/ZrZAKr4LHjLgHPbbLh0KbPPlnSPpCskHZTWNXp6XeO6VakBRRf+vpL2BPbDE0WbDSxfAR0Yzm2zIdEis+dJWluzrCixy8uAlwJLKR6N/MmJQ9Up23SwcttjQCPiB5I+QTF330+AmyPi5qnl0hdaAbBo0aJ2D2dmPeQxoIOhTG47s80GQ4ncbvr0unoiYlvN/j8L3JDezvjpdVW64A+i6PM/gmLA6QskvbtOZcciYjQiRkdGRto9nJn1mLvg+1+Z3HZmmw2OTmd2emTyhLcDG9Lr1cBpkvaWdASwBLiz2b6q3AX/JuDBiNiRKvVl4PXA31fYp5llyt3sA8G5bTZEKk7D9AXgOIqu+i3AHwLHSVpK0b0+DrwfICI2SroWuA/YBZzV7A54qNYAfQg4RtJ+FF05JwBrK+zPzDLlLviB4dw2GxIdeILdu+qsvrxJ+YuAi8ruv8oY0DskfRG4i6K1+x1grN39mVnefAW0/zm3zYZLzrldaSL6iPhDikuy1gFln26xe/fuLtfEbDpfAR0Mzm3rV3Pnzu11FfpOzrntJyGZWUueasnMrL/kntv5No3NLCtV74JPkxZvl7ShZt3HJP1A0vq0nFyz7QJJmyV9T9Jbu/S1zMwGVs4zl/gKqJmV0oHAWkkXnytsZmaT5dDQbCTfmplZNlo9BSmH5wqbmdnzcn96nRugZlZKFyei78hzhc3MbLKcu+B7XwMz6wslzqZ7+lxhMzObLOcroB4DamYtlZzQuKfPFTYzs+fl/gCRfGtmZlnpxniiTj5X2MzMJvMVUDPre1XPpLv9XGEzM5ss5yugboD2oZz/QtngqnrG3O3nCpuZ2WQ5XOlsxA1QM2sp97FEZmY2We653bJmDZ5ecrCkWyR9P/15ULN9mFn/6+I0TNZhzm0zg/6fhmklcOKUdecDt0bEEuDW9N7MBlQnJqK3WbUS57bZUMs9s1s2QBs8vWQZsCq9XgWc0uF6mVlmfAW0fzi3zQzyvgLa7hjQ+RGxFSAitko6pIN1MrMM5XDGbJU4t82GTM653fUmsKQVE09G2bFjR7cPZ2ZdMDGYPeezaesMZ7bZYGiV2yU+X3osuQp/KWlzerTya1vtv93/NbZNTCCd/tzeqGBEjEXEaESMjoyMtHk4M+u13McTWUulctuZbTY4Kmb2SsqPJT+J4oEhS4AVFI9ZbqrdBuhqYHl6vRy4vs39mFmf8BXQvufcNhsyVTJ7hmPJlwFXRuF24MApT7qbpuUY0AZPL7kYuFbSmcBDwKktv4mZ9a3c55OzyZzbZtal3G40lvww4OGaclvSuq2NdtSyAdrg6SUAJ5Srq5kNAnez9w/ntplBy9yeJ2ltzfuxiBhr91B11kWzD/hJSGZWiq+Ampn1lxa5vTMiRme4y22SFqSrn7VjybcAC2vKHQ480rRuMzywmQ0p34RkZtZfupDZjcaSrwbek+6GPwZ4YqKrvhFfATWzljwG1Mysv1TN7RmOJV8DnAxsBp4G3tdq/26AmlkpVa9ySroCeBuwPSKOSusOBq4BFgPjwDsj4jEVB7uUItCeBt4bEXdVqoCZ2ZCpktszGUseEQGcNZP9+5KGmZXSgWmYVtLFOeXMzGyynKfO630NzCx7nXgSUrfnlDMzs+dVfRJSt7kL3sxK6dKNRh2bU87MzCbL+QZRN0DNrJQSZ8w9nVPOzMwmy+FKZyNugJpZSyWn7ejpnHJmZva83KfIy7dpbGZZ6dKz4Ds2p5yZmU3mMaBm1vc6MA1TV+eUMzOzyXK+AuoGqJm11ImJ6Ls9p5yZmT0v9weItKyZpCskbZe0oWbdn0v6rqR7JF0n6cDuVtPMeq1LXfDWBc5tM4O8u+DL1GAl0yePvgU4KiJ+Gfg34IIO18vMMuNnwfeVlTi3zYZezpndsgFab/LoiLg5Inalt7dT3KFqZgOqExPR2+xxbptZ7hPRd6IGvwXc2GijpBWS1kpau2PHjg4czsx6wVdAB0rD3HZmmw2OnDO7UgNU0h8Au4CrGpWJiLGIGI2I0ZGRkSqHM7Me8hXQwdAqt53ZZoMj58xu+y54ScuBtwEnpDtWzWyA5XDGbNU4t82GS8653VYDVNKJwEeB/xwRT3e2SmaWm9yn87DWnNtmwyX33G7ZAG0wefQFwN7ALal1fXtEfKCL9TSzHss5yGwy57aZQd653bIB2mDy6Mu7UBczy1jOXTk2mXPbzCDv3PaTkMyspdy7cszMbLJO5LakceBJ4FlgV0SMSjoYuAZYDIwD74yIx2a6b/+PYmaleBomM7P+0qHM/vWIWBoRo+n9+cCtEbEEuDW9nzFfATWzUnwF1Mysv3Qpt5dRjDEHWAXcRnGD44z4fxQza6nV1U9fATUzy0uHMjuAmyWtk7QirZsfEVsB0p+HtFM/XwE1s1JyHktkZmbTtcjteZLW1rwfi4ixKWWOjYhHJB1CMYPGdztVNzdAzayUDnXhL5dhAAAGtklEQVTl/HpE7Kx5PzGW6GJJ56f3M+7KMTOz6Vrk9s6acZ11RcQj6c/tkq4Djga2SVoQEVslLQC2t1W3dj5kZsOli13wyyjGEJH+PKUjFTYzG3JVM1vSCyQdMPEaeAuwAVgNLE/FlgPXt1M/XwE1s1I6cAV0YixRAH+bunomjSVK3TxmZtYBFXN7PnBdaqzuCfxDRPyzpG8D10o6E3gIOLWdnbsBamallDhjbjWeqGtjiczMbLoqN4hGxAPAq+us/w/ghArVAtwANbMSSk5o3HQ8UTfHEpmZ2WS5P0Ak35qZWVaqjCfq9lgiMzObLuep81o2QCVdIWm7pA11tv2epJA0rzvVM7NczJkzp+nSwnzgG5LuBu4E/iki/hm4GHizpO8Db07vrSLntplB89zutTJd8CuBTwNX1q6UtJDiP4yHOl8tM8tJ1a6cbo8lsmlW4tw2G2p93wUfEV8HHq2z6RLgIxR3tprZgPOTkPqHc9vMIO8u+LZuQpL0m8APIuLuHL6EmXVfzmfS1ppz22z45JzbM26AStoP+AOKmwjKlF8BrABYtGjRTA9nZplwo6V/zSS3ndlmgyPn3G6nafxS4Ajg7vRs58OBuyS9uF7hiBiLiNGIGB0ZGWm/pmbWMxNjiXIe0G5Nlc5tZ7bZYGiV27024yugEXEv8NzTSlKYjU55vrOZDZicz6StOee22XDKObfLTMP0BeBbwCskbUmPXjKzIeMroP3DuW1m0OfTMEXEu1psX9yx2phZlnKfzsMmc26bWe657UdxmlkpOXflmJnZdDnnthugZtZS7mfSZmY2We657QaomZWS85m0mZlNl3NuuwFqZqXkfCZtZmbT5ZzbboCaWSk5n0mbmdl0Oee2G6Bm1lLuY4nMzGyy3HM735qZWVaqzgMq6URJ35O0WdL5s1BlM7OhlnNmuwFqZqVIarq0+OwewF8DJwFHAu+SdOQsVNvMbGjlnNnugjezljrQlXM0sDkiHkj7uxpYBtzXgeqZmdkUFXO765ntK6BmVkqVK6DAYcDDNe+3pHVmZtYlOWf2rF4BXbdu3U5J/z5l9Txg52zWo4v8XfLk71J4SbsHXbdu3U1z5syZ16LYPpLW1rwfi4ix9Lpe2kW79bHZ4czuK/4ueepJZkOp3O5pZs9qAzQiRqauk7Q2IkZnsx7d4u+SJ3+X6iLixIq72AIsrHl/OPBIxX1alzmz+4e/S556+V0q5nbXM9td8GY2G74NLJF0hKS9gNOA1T2uk5mZ1df1zPZNSGbWdRGxS9LZwE3AHsAVEbGxx9UyM7M6ZiOzc2iAjrUu0jf8XfLk75KBiFgDrOl1Payyvv07WIe/S578XTLQ7cxWhO8DMDMzM7PZ4zGgZmZmZjaretoAHaRH80kal3SvpPVTpjXInqQrJG2XtKFm3cGSbpH0/fTnQb2sY1kNvsvHJP0g/W7WSzq5l3UsQ9JCSV+TtEnSRknnpvV9+XuxweDMzoMzO0/O7ZnpWQN0QB/N9+sRsbQPp49YCUydruF84NaIWALcmt73g5VM/y4Al6TfzdI0riV3u4APR8QrgWOAs9K/j379vVifc2ZnZSXO7Bw5t2egl1dAn3vMU0T8DJh4zJPNsoj4OvDolNXLgFXp9SrglFmtVJsafJe+ExFbI+Ku9PpJYBPFUyj68vdiA8GZnQlndp6c2zPTywbooD2aL4CbJa2TtKLXlemA+RGxFYp/VMAhPa5PVWdLuid19/RV94ekxcBrgDsYvN+L9Q9ndt4GLRv6NrPBuV1GLxugg/ZovmMj4rUU3VNnSXpjrytkz7kMeCmwFNgKfLK31SlP0v7Al4DzIuJHva6PDTVnts2Wvs1scG6X1csG6EA9mi8iHkl/bgeuo+iu6mfbJC0ASH9u73F92hYR2yLi2YjYDXyWPvndSJpLEWJXRcSX0+qB+b1Y33Fm521gsqFfMxuc2zPRywbowDyaT9ILJB0w8Rp4C7Ch+aeytxpYnl4vB67vYV0qmfiHn7ydPvjdSBJwObApIj5Vs2lgfi/Wd5zZeRuYbOjHzAbn9kz1dCL6NLXCX/D8Y54u6lllKpD0ixRn0FA8Xeof+um7SPoCcBwwD9gG/CHwj8C1wCLgIeDUiMh+oHiD73IcRVdOAOPA+yfG4+RK0q8B/wLcC+xOqy+kGE/Ud78XGwzO7Dw4s/Pk3J4ZPwnJzMzMzGaVn4RkZmZmZrPKDVAzMzMzm1VugJqZmZnZrHID1MzMzMxmlRugZmZmZjar3AA1MzMzs1nlBqiZmZmZzSo3QM3MzMxsVv1/mDwjG53rRvkAAAAASUVORK5CYII=\n", | |
"text/plain": [ | |
"<Figure size 864x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
}, | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 864x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
}, | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 864x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
}, | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 864x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"logscale = !True\n", | |
"logscale = dict(norm=mpl.colors.LogNorm()) if logscale else {}\n", | |
"\n", | |
"for idx in range(5):\n", | |
" fig, axes = plt.subplots(1,2, figsize=(12,3))\n", | |
" for iax, ax in enumerate(axes):\n", | |
" im = ax.imshow([std_jet_images0, std_jet_images][iax][idx], cmap=cmap, **logscale)\n", | |
" plt.colorbar(im, ax=ax)\n", | |
"\n", | |
" ax.set(title='{} jet: $p_T=${:.0f} GeV'.format(['QCD','top'][iax], [jetpep0,jetpep][iax][idx][0][0]))" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"Normalizing the histogram values in 0-255 range: only differences are seen if I use log scale, in which case a bunch of soft energy deposits disappear. The classifier should not care about smaller changes (do not fit the noise!)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 8, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"all_jet_images0 = map(normalize, all_jet_images0)\n", | |
"std_jet_images0 = map(normalize, std_jet_images0)\n", | |
"\n", | |
"all_jet_images = map(normalize, all_jet_images)\n", | |
"std_jet_images = map(normalize, std_jet_images)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 11, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 864x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
}, | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 864x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
}, | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 864x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"logscale = !True\n", | |
"logscale = dict(norm=mpl.colors.LogNorm()) if logscale else {}\n", | |
"\n", | |
"for idx in range(3):\n", | |
" fig, axes = plt.subplots(1,2, figsize=(12,3)) \n", | |
" for iax, ax in enumerate(axes):\n", | |
" im = ax.imshow([std_jet_images0, std_jet_images][iax][idx], cmap=cmap, **logscale)\n", | |
" plt.colorbar(im, ax=ax)\n", | |
"\n", | |
" ax.set(title='{} jet: $p_T=${:.0f} GeV'.format(['QCD','top'][iax], [jetpep0,jetpep][iax][idx][0][0]))" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"Plotting whole events in the $\\eta-\\phi$ plane. Also draw a circle around each jet, with opacity set by jet pT" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 19, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 720x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
}, | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 720x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
}, | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 720x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"logscale = !True\n", | |
"logscale = dict(norm=mpl.colors.LogNorm()) if logscale else {}\n", | |
"\n", | |
"for idx in range(3):\n", | |
" fig, axes = plt.subplots(1,2, figsize=(10,3)) \n", | |
" for iax, ax in enumerate(axes):\n", | |
" im = ax.pcolor(etaedges, phiedges, sum([all_jet_images0, all_jet_images][iax][idx]), cmap=cmap, **logscale)\n", | |
" plt.colorbar(im, ax=ax)\n", | |
" jets = [jetpep0[idx], jetpep[idx]][iax]\n", | |
" for j in jets:\n", | |
" ax.add_artist(plt.Circle((j[1],j[2]),1, color='r', fill=False, alpha=j[0]/max([jj[0] for jj in jets])))\n", | |
" ax.scatter(j[1],j[2], s=4, c='r', alpha=j[0]/max([jj[0] for jj in jets]))\n", | |
" ax.set(title='{} event'.format(['QCD','ttbar'][iax]), xlabel='$\\eta$', ylabel='$\\phi$', yticks=np.linspace(-np.pi, np.pi,5), \n", | |
" yticklabels=['$-\\pi$','$-\\pi/2$',0,'$\\pi/2$','$\\pi$'])\n", | |
" for h in [-np.pi, np.pi]: \n", | |
" ax.axhline(h, ls='--', lw=1, c='gray')" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"Compare average images:" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 117, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 1440x180 with 8 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"fig, axes = plt.subplots(1,4,figsize=(20,2.5))\n", | |
"ims = [0]*len(axes)\n", | |
"ims[0]=axes[0].imshow(np.average(std_jet_images0, axis=0), vmax=150,cmap=cmap)\n", | |
"ims[1]=axes[1].imshow(np.average(std_jet_images0, axis=0), vmax=150, norm=mpl.colors.LogNorm(),cmap=cmap)\n", | |
"ims[2]=axes[2].imshow(np.average(std_jet_images, axis=0), vmax=150,cmap=cmap)\n", | |
"ims[3]=axes[3].imshow(np.average(std_jet_images, axis=0), vmax=150, norm=mpl.colors.LogNorm(),cmap=cmap)\n", | |
"\n", | |
"for iax, ax in enumerate(axes):\n", | |
" plt.colorbar(ims[iax], ax=ax)\n", | |
" ax.set_axis_off(); \n", | |
" ax.set(title='averaged {} jet image'.format(['QCD','top'][iax//2]))" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 118, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 1584x216 with 8 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"fig, axes = plt.subplots(1,4,figsize=(22,3))\n", | |
"ims = [0]*len(axes)\n", | |
"ims[0]=axes[0].imshow(np.var(std_jet_images0, axis=0), vmax=12000,cmap=cmap)\n", | |
"ims[1]=axes[1].imshow(np.var(std_jet_images0, axis=0), vmax=12000, norm=mpl.colors.LogNorm(),cmap=cmap)\n", | |
"ims[2]=axes[2].imshow(np.var(std_jet_images, axis=0), vmax=12000,cmap=cmap)\n", | |
"ims[3]=axes[3].imshow(np.var(std_jet_images, axis=0), vmax=12000, norm=mpl.colors.LogNorm(),cmap=cmap)\n", | |
"\n", | |
"for iax, ax in enumerate(axes):\n", | |
" plt.colorbar(ims[iax], ax=ax)\n", | |
" ax.set_axis_off(); \n", | |
" ax.set(title='variance of {} jet image'.format(['QCD','top'][iax//2]))" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 121, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAl0AAADiCAYAAAB0gXKiAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJztvXnYZVV55v17qphEkAIKrLkKKApBMZKuFofulnwiQWNCp1vbIVFQE2J3NJpoHHMZPxO/jkmbmG5NDCZ8HYfW1k40FRtbTSttBjEUDhhGCwRqAqqKWVQEVv+x9wtn32efs855z7Sqzv27rveqWmfvs/faa+91n7Wf51nPipQSxhhjjDFmsiyZdQWMMcYYY+YBD7qMMcYYY6aAB13GGGOMMVPAgy5jjDHGmCngQZcxxhhjzBTwoMsYY4wxZgp40FUgEbEhIlJEHDTruoyL+no2LvK7PxcRXxh3nSZFRFwVEWfNuh7GTJtxaFdEXBARfzfOeo2DUesVEZ+LiPPHWadJERH/MiKum3U9DkQ86NpPqQXg2xFxf0TcGhF/FBFHyT6bIuJTEbE3Iu6OiCsj4tciYmmHON5X/90WEZ+NiOfM6pp6kVL6WErpnEkcOyJuioizx3nMlNITU0qXjvOYxhwIaH87EF8we5FSem5K6c/HfdyIOCsidozzmCmlv00pnTLOY5oKD7oGoDRBiIg3AO8Bfh04CngasAH4QkQcXO9zEvA1YDtwekrpKOCFwGbgyI7DLUspHQH8GPBF4NMRccF0riRPaW3fSVS4D5liKbn/TIMSrr90nSihjeaKlNJ+/Qe8BbgBuBe4GvjZ+vNDgbuAJ3XsexzwfeD4uvx84Jv1fv8APLlj35uANwNXAj8EDup1rnr/pcB7gb3Ad4HXAAk4qN5+FPBnwG5gJ/DbwNKO7/6n+rs3Ar/c+V253scB9wH/Tj4/ArgdOL8ufxT4n33abUPbOYA3ArcBS3p874lUg7M76v3eVn/+VOCrdVvuBt4PHNLxvQRs7GiLDwN7gJuB31g4H3AB8PfAH9Tn+O36s7+TY/0H4Dv1vfgt4KT6/PcAn5Rzt95n4CPAw/UzcR/wpvrzp9X73QV8Czir41iXAu+u6/j9hWuSNroJOLv+/zuBT9X3417g28Am4K31/doOnNPx3VcA19T73gj8khz7TXX77gJ+Qdr10Po5uqW+Nx8EHjPrPuq/nn1w3rSrq7/Vz2qqy/cBT+dRDfgvwN3AtcCzB+kjwFnAjvr6bwU+0qPtf7HjGFcDP97vntTbLqCpQ88ALq/reDnwjI5tlyI6UX/2Cx3HWtC5u+rreEb9+XY6tLzjmejq28Bj6+M/3NGGq6gMKgvXso9KE4+pj7WhbvNX1cf7Skv7nAXskGfq16meqe/Vz8Pjgc/VbfU3wNEd+3+qbv+7ga8AT+zYdizw11RafTnV89TZrk/g0d+Y65Dfuv39b+YVGPkCKuvNwkP2ovqBWFlvuxh4d8e+vwz8r/r/P14/2GdSCcf59YN1aMdD9k1gLfUPV+Zcr6bqpGuAo+uHsFO4PgP8Sd1Jjgf+kVos6u9eW5/rGODL9Bauc4EHe2z7c+Bj9f9vBV7Rp902tJ0DOLH+/NSW7xxJJbxvAA6ry2fW2/4Z1WDloPrY1wCv7/hu5+Dgw8Bf1d/fAFwPvKredkF9fa+tj/UY2gddW6gGoE+k+mH533Xdj6rvw/lD3OezO469mkqknlff5+fU5ePq7ZdSCdUT6/od3NJOjxyTatD1A+An6/0/TPXD9nbgYCrx/27Hd3+KagAZwLOA+3n0B+Hc+r4+ETic6kess13fV7fLMXXb/jXwH2fdR/1n7WrrG3V5g+7Poxrwq3UfeRHVj/fCoKFfHzmr/u57qAYqXS8ddVvsBP55fYyNwPoB2ukCah2qr/VO4GVU/foldfnYevuliE7QPeh6kGoAuZRq4HEL8IG63udQDWaOqPfv2beRAVL92euBy+p7emh9/z4ubf7h+p62tVHjmPV9u4xqoLWa6vn7OnBGffwvAb/Zsf8r63oeWtf9mx3bPlH/HQ6cRjXIXGjXx9blV9Tt9uNUA/onah3317+ZV2DsF1SJzXn1/88GbuzY9vfAy+v//zHwW/Ld64BndTxkrxziXF+i+cZ1dv1gH1Q/qD/sfLjrTvrlju++umPbOfQedP08cGuP+vwO8IX6/z8Czu1T9w1t56AaTCXgmS3feQnwjQHvw+uBT3eUE5W4La3b4rSObb8EXFr//wLgFjnWBXQPup7ZUb4CeHNH+b3A+4a4z50/Am9G3o6Bz/PoIO5S4F2Za3/kmFSDri92bPtpqrfRBUvBkfX1LOtxrM8Ar6v/fzEdg6i6PRfaNah+IE7q2P50OgZ0/iv7jwNcuzrqNsigaxcQHZ/9I/CyHsfs7CNnAQ8Ah/W59s8v7D/kPbmARwcHLwP+Ufb9KnBB/f8unaB70PWdjm2n1+3w+I7P9gFPyfVt2gdd19C0Dq6k+k1YeClOwIl9rrtxzPq+/VxH+S+AP+4ovxb4TI9jLavPdxSV/v8IOKVj+yOWLqqB7t/K9/+EjgHd/v633/tyI+LlwK9RPUhQudmW1///EvCYiDiTykLwFODT9bb1wPkR8dqOwx1C9ZazwPYhzrVK9u/8/3qqN53dEbHw2ZKOffS7N7dda81eYHlEHJRSelC2raRy2UHVYVf2OU4vVtf/3tGybS2VubqLiNgE/D5VzNjhVJ37ipZdl1O1c+c13txxXpB278FtHf//fkt5Rf3/Qe5zJ+uBF0bET3d8djDVG/ww9etX170ppYc6ylA9S3dFxHOB36RyQS6hastv1/usArb2qMdx9b5XdDxjQSVypkDmULuGYWeqf3E7jrsKINNHAPaklH7Q59j9dKxfO3Wyiu5rHVbHVBdIKelnR7C4vr2eKj734Y7PHqIaRA9av1x92+pKRCylcq2+sK77Qh2WU3kuDqL/83ZmRNzV8dlBVFb9A4L9etAVEeuBDwHPBr6aUnooIr5J9UCSUno4Ij5J9WZ2G/DZlNK99de3U5nv393nFI90+ty5qNxuazq+u7bj/9up3haXtwyUFr7buf+6PnX6an2sf0Plp1+o32OB51LFR0HlIvi3wP/f51ht/CyV6bhtuvB2qrZs44+BbwAvSSndGxGvB17Qst9eqjed9VQuDaiud2fHPkm/NAK5+6zn2k5l6frFPsccZ/0eISIOpXqDfDnwVymlH0XEZxjsGdtLJXxPTCl1tqUpkDnVrka9epQXWB0R0THwWgdsGaCP9DvmAtup3JMNBminTnZRaVgn64D/NUQ9BiXXt9vOs53K2vn3uiEiNoy5fspLgfOoLKY3UVm47qRqxz1UbtU1VGEl0P28/Z+UUnGz6MdFsTMqBuSxVA/OHoCIeAXwJNnnv1GZLH+u/v8CHwJeHRFn1rNLHhsRPxURR9JO7lyfBF4XEasjYhmVmwqAlNJu4AvAeyPicRGxJCJOiohndXz3VyJiTUQcTRUA2UpK6W7g/wX+S0ScGxEH153oU1Sd82P1rr8JPCMifi8iVtR13hgRH63r1yAiHh8Rr6m/99aU0sO6D/BZYEVEvD4iDo2II+s3cajcZPcA90XEE4B/36P+D9XX++76++up3iw/2uuaRyR3n2+jigVb4KPAT0fET9apNQ6rp2Sv6Try+DmEKgZiD/Bg/UbfmSrjk8ArIuLUiDgceMfChvp+fQj4g4g4HqB+Fn9yCvU2wzN32lWj/W0PlSXkRNnv+Pq4B0fEC4FTgUvI95FB+FPgjRHxz+r221jr0CD3ZIFLgE0R8dKIOCgiXkQVn/TZIeuSZYC+fRtwbDRTBn2QSmPX1/sfFxHnjbtuPTiSaqC+j8pC9/8tbKj1/y+Bd0bE4fVvxcs7vvtZqnZ9WX3vD46Ifx4Rp06p7hNnvx50pZSuporf+SrVg3c6VexD5z5fo/KHr6KaabHw+VaqIOb3U43Ct1H52Rd7rg9RidOVVBafS6hG9AtupJdTCcbV9fn+B4+6/z5EFWfwLargxL/MXPfvAm+jms1yL1Vg9uFUsRLfq/e5gcrvvwG4KiLupnpD3Fp/Z4G7IuJ7VOb55wEvTCld3OO891IFlv80lcvjO8BP1JvfSPWGc299Pf+9zyW8luqe3Aj8HdUPSus5R2WA+/wfgd+IiLsi4o0ppe1Ub2lvoxLf7VSzdibeV+r2/RWqH7I7qdpzS8f2zwH/mcrVuY3qWYRK4KD6sdwGXBYR91BZO51rp0DmVbvo7m/3U8/yqz97Wr3f14CTqV4k3w28IKW0L9dHBiGl9Kn6mP+NSq8+QxWkn70nHcfYRzWD9A1Ug4s3Ac9PKe0dpi5D0LNvp5SuBT4O3Fi34SrgD6na5QsRcS9VEPyZrUcePx+mcrXupHpmLpPtr6Gyft1K5Tb8OLWG1ff3HODFVNbEW3l0UsQBQTTd5mZc1G9gH0wpqQl6Eud6JZX165kppVsmfT5TBvXb3z9RzVprc/0YMzTT1C5jIuI9wIqU0vmzrss02K8tXSUREY+JiOfVpubVVG66T+e+Nw5qy9TbqPK8mAOYiPjZiDikduW8B/hrD7jMKMxSu8z8ERFPiIgn167dp1LlC5ub582DrvERVNamO6lM9NfQEXMzaVJKH0kpfWJa5zMz45eo3J43ULl/WmPnSiAiLo6I2yPin3ps/7molqa6MiL+ISJ+bNp1NMCMtcvMHUdSuaG/R+Umfi9V3saimJR+2b1ojJkIEfGvqHKSfTil1BWQHBHPAK5JKd1Zu7TemVKaVtyJMcb0ZFL6tV+njDDGlEtK6Svx6PT0tu3/0FFcyJ5tjDEzZ1L6ZfeiMaYEXkXHDD1jjNmPGFi/irN0RZXNditVRuLn99t3+fLlacOGDVOplzH7M1dcccXelNJxw3wnInKxB1dRrSu5wEUppYuGrVtE/ASVaP2LYb9bIoNqmPXLmME4kPSruEEX8DqqQM7H5XbcsGEDW7duze1mzNwTEYtanmXJkt7G8IcffvgHKaXNi64UEBFPpkpW+dw699GBwEAaZv0yZjAOJP0qyr1YZ/3+KaqLMMbMmIjo+TeGY6+jmsX0spTS9bn99wesYcaUQ4n6VZql631UmX17LWdBRFwIXAiwbl1umS9jzGKJiL5vig899FDPbfX3Pw6cRbVA+w6q/E8HA6SUPkiVluBY4I9qEXxw1DfPAuirYdYvY6ZDqfpVzKArIp4P3J5SuiIizuq1X+1zvQhg8+bNzndhzAQZ5Y0wpdRrcfSF7b8A/MKiT1AYg2iY9Wu2aIqkcVg8TLmUqF/FDLqAZwI/ExHPAw4DHhcRH00p/fyM62XM3NLvTdF0YQ0zpiBK1K9iapRSemtKaU1KaQPVYpdfslgZM1smGRNxoGENM6YsStSvkixdZsy0rTagD9ukze2jHt/ugNmRi4kw88U0Vi+Z9DlUPx5++OG+++v2SfSHYTU5t3/u+PNCqfpV5KArpXQpcOmMq2HM3FOiaO0PWMOMmT0l6leRgy5jTBnM61uyMWb/p0T98qBrjOTMwqOagcdhetdj5MrKqA+xmuuHbaMci6m/fmfUt6MDxSVaqnnejIdh+1rb/rlnPee+G1aPcsfLueJyepM730EHNX8y29IOLF26tO8xc3qQq3NOQ7Wc68PD1md/oVT98qDLGNOT/VVwjTGmRP3yoMsY05MS3xSNMWYQStQvD7qMMa3Memq1McYsllL1y4OuPowaX6T+/lF99bo9F3+wmAdu2OnKitZB4xuUH/3oR41yLmZC31y0nGujtvrrZ7nlIYZ9e9qfY7xKfFM0gzFsOoRcX9O+2oYeI1eHBx54oO/+qh/33Xdfo3zIIYf0LefqnOubDz74YN/yYYcd1nXMnIbpNWkdDz/88EZZ74PWQY+nZW0TJaehBx98cN/vjzsmdpyUVJcFPOgyxvRkfxogGmNMJyXqlwddxphWSp39Y4wxOUrVr7kedOXM6TlTuY6i1VSux8u5K9VsrMcb1p3Y5ibTfdR0rHX+3ve+1yir6VzROub2V9O6muuHncKdM7233YOc+f4xj3lM32Nqm+ZcqjkhyH1/mpQoWqYi5wbXvqXP/ve///1GWe+1br///vu7zpHTC9Uw7Sv33ntvo/zDH/6wUVbX2O233973/IceemijfM899zTK2rfVXXncccf1rZ9qwWMf+1gUvWatk16T3hfVQL3GZcuW9T2+1lHbXDXzcY97XKOs+qP3Xb+f08NZWptK1K+5HnQZY/pTonneGGMGoUT98qDLGNNKqeZ5Y4zJUap+edBljOlJiW+KxhgzCCXq11wPujTmQWMk7rjjjr7f1+0aj6DH0/iBPXv2NMoaH6DbjzjiiEb51ltvbZTVt3/TTTd11Vn991q+++67G+Xly5c3ynqNO3fubJSPP/74RlmvWcsaD7By5cpGWTuNxjvcddddjfKKFSvoh14v5KedaxyItrO2kU751mvUmAhts9z5pkmJb4qmQuMbtZzTN43Z0v1Vf9riIW+55Za+x7jzzjsbZdUsjSndt29fo7x79+6+dda+c+211zbK2nduu+22Rlnjo9asWdMoawyZ9t22dAqqMRrnpnXOpYg45phj+tZZfxeOOuqoRvnoo49ulI888shGWTVXz6+/S7kUE8PGuE6SEvVrrgddxpj+lPimaIwxg1CifpU3DDTGFMFCTESvvwG+f3FE3B4R/9Rje0TEf46IbRFxZUT8+Ngvwhgzl5SqX3Nt6Ro2HcGwq7nv3bu3UVYz7bHHHtsoq+k950rT8+v1tJl11SV68803N8pqLlf3ge6v59Qp1+vWrWuU1X2pqAtEXW2rV6/uW99t27Y1yuoOHcS9ePrpp/eto04J1/usbdB2zk4GyXI9K0Y0z/9X4P3Ah3tsfy5wcv13JvDH9b9mEQyb0kZddfocaqoA1SfoDhfQ/nbjjTc2yupuVHei1lHrpPrx9a9/vVFWfVPXmYZDaH0VdeWpBrf1VdX5jRs3NsraBqrTGiKh51T3obpEc2kzVO/Ujay/S7mwmVwKjFlSon7Z0mWM6cnC+mVtfzlSSl8B+gVGngd8OFVcBiyLiJV99jfGmIEpUb/m2tJljOnNFKZcrwa2d5R31J/tbt/dGGMGo1T9mutBl5p11QybWzxZ0UzAagbWmTM6M0dnoezatatR3rBhQ6OsM3XUXaDXA3mXqZqKf/CDHzTKOhNGzfWawVndAblszbnZRdom2qke//jHN8rapnqPoNucry4TPYe2gZ5DyWW8VzeOzmaaJZk3wuURsbWjfFFK6aJhDt/yWTm+if0M1St1C+nMwtwKFvrctrnJ1Z2n7n3VWNUHdS9qX7jhhhsaZdWT3IoU6t4cFj1eTh/bPtMQCNUPvSbVPL0P2uYazqC6r+fX3wl1h+aeg9wqH7lFzqdJifo114MuY0x/Mm+Ke1NKm0c4/A5gbUd5DbCrx77GGDMUJeqXY7qMMa30i4cY01TsLcDL61lATwPuTinZtWiMGZlS9cuWLmNMT0aJiYiIjwNnUZnxdwC/CRwMkFL6IHAJ8DxgG3A/8IoRq2uMMY9Qon7N9aArF8+kvmv11ev+ivq2dXqxTmdWX71mOtfzq+9fMx/r8QF27NjRKK9atapR1mnhGtOgbaJxZhpHovEEWie9Jo0P0DbTNtfszppBWuunMSXQHeuiMVp6Tt2ubZLLwKzf1+dQn5v9NaNzSuklme0J+OVFn2DOyb2t554bjYfS1AS5dA3Qnd1cYzRVkzQu7Oqrr26UNQZL4xv1mjUlzagxXIrGdw6Cap6WtV31PuQy0mussMZgqZ7p8VTj9R5q/Yatby4ObpqUqF9zPegyxvRmjGZ4Y4yZKqXqlwddxpielLh2mTHGDEKJ+jXXgy51ZeUW8lRXmE69VXN+Ll2Cmt7VzKvbdTp1LiO9pnNoq5MeU03hmtZCM8KrO1HdfXoNmr1dTdWa8kHro/dAM+ZrSgi9R23T3rXdNGO0mvf1uVDzuW7Xa9A2V3dBSe7FEt8UTYU+t1rOhQZo31jMChf6LKseqCs917fUNaYuT63TsO7E3DUOS1ub6DWoi1LroJqmIRKqD3rfNGRDr0nbUFPkDJsaKfe7qZo+S0rUr3JaxxhTFFNILmiMMROhVP3yoMsY05MS3xSNMWYQStQvD7qMMT0p8U3RGGMGoUT9mutBV25qbW67Lp+g8Q26XWMsdBR+3XXXNcrqe9d4LJ2erfEFmlICumMu9Jgnnnhio6xtkLuG3DIdGvOlx9NlgHTFe51Cnot70zQcbWg76TVrzJaeIxeDpXXWGC+NkdAp4Lr/tCjVPG8qcjFXmm4ll/5An1Ptm22xPtq/NYZL0xOoXmgdNe61bdmuURg1hktpixHVuDUta5upxul9VI3TZYRUn7TPauqhnB7p8fSe6D3LxRbOilL1q5gaRcTaiPhyRFwTEVdFxOtmXSdj5p0JZ3Q+oLCGGVMWJepXSZauB4E3pJS+HhFHAldExBdTSlfnvmiMmQwlvikWjDXMmIIoUb+KGXTVaxbtrv9/b0RcA6wGJiZYuYzz6jZS87ya39UMq6Z2Rafyrlmzpu/xtD46tVjNzppOoe0Yuaz2eo16Dq2jtpG6B9T0rCkk1q5dSz9y7sphXX3Qfd+1o6qLUt2+ao7Xsrp01e08bEb7aWKL1uBMW8O0L+m90ucul8pE+5a60bUMcM899/Q9htZJwwdyIRm7djXXDz7ttNMaZXVfqitv0q4uPV8b2p/VJal1VF1XfdJVRHJouILWOZfhXuuvz1HJGlFi3YoZdHUSERuAM4CvtWy7ELgQupeAMMaMj1JjIvYHemmY9cuY6VCqfhVXo4g4AvgL4PUppXt0e0rpopTS5pTS5uOOO276FTRmjigxJqJ0+mmY9cuY6VGifhVl6YqIg6nE6mMppb+cwvkaZTW3q6tNTd+aGVjNxPp9NYWrmVlN71ofnYWiZl917bW5qVTodQFoNWXrzBo1PWsbaBup+2/Pnj2Nsmag1zbKZcDPZYPPzbaEbheH1knbQOugsx11ZYGci2PcWbLHSYlviiUzTQ3T/q3PoT5XWta+q31F9UtnS0O3O1H7n7rz1TV2/fXXN8rbt29vlLXvad8YZHbyOFFXXNuMUNVILavGalZ+DclQDcy5cLWOqlcbN27s+/3cyiz6nOlsxpJeyErUr2IGXVHdqT8Drkkp/f6s62PMvFOqeb5UrGHGlEOp+lVSjZ4JvAz4fyLim/Xf82ZdKWPmmRLN8wVjDTOmIErUr2IsXSmlvwOs5MYURIlviqViDTOmLErUr2IGXbMgF4OVi63JZSrXGAndX7er711jujR+QGMsNF5KYwmgO2O8xg/kUkCoP1/bSKd45+qs16xxbvpGojEi2qm0/hovofe47RwaM7Fz585GWePiNKZBp9FrLJ7eF63zrDLQK7N+IzT9Uf3IlTWGS/tOLj6zDdWP3AoR+mxrihhNc6P60ZYGpxPtixo/NSqqn4PUQWOitN01rlY1UH83tA6qL3o81Uytj+qR3jNNE6T3sNSY1FL1a64HXcaY/pT4pmiMMYNQon550GWM6UmJb4rGGDMIJerXXA+6cq4rNZOqWTWX/VjNtnr8ffv2NcqaqkBTSKj5X6dL51yDkJ82rnXQY+g51BS9adOmRjnnQs2lV9DpzzrFfOXKlY1yboHctu1qvs8tHKzmed2u16htrCsVqDuzJPP8KG+KEXEu8IfAUuBPU0q/I9vXAX8OLKv3eUtK6ZLF13i+0HuT0yvVH3VDKbmVGqBb49Q1ptt1BYpcOoObb765bx01pGLc7sTFkKuD6rZqpmqs/k6o21dXNlFNVM3X/VWDNSRD77uWc6lKZsU4Zi9OQsPKs70ZY4phyZIlPf/6ERFLgQ8AzwVOA14SEafJbr8BfDKldAbwYuCPJnAJxpg5ZbH6BZPTMA+6jDE9GWHK9VOBbSmlG1NKDwCfAM6TfRKwYBo5CtiFMcaMiRFTRkxEw+bavWiM6c0A5vnlEbG1o3xRSumi+v+rgc704juAM+X77wS+EBGvBR4LnD1ajY0xpmJE/YIJaZgHXR3kVlfX6dB6QzW2R6f2Llu2rFFW377GAuj+ej6Nl1LfetsDp3Efuk9bSoVOcikmbrjhhkb5hBNOaJRzKSI0Tk7jTjTGS99Y9B4OsjSSxlRpnMj69esb5dzUfI250DrrfdaYspJm3GTeCPemlDb3+mrLZxqs9hLgv6aU3hsRTwc+EhFPSik93PJdI+SWMdPnTPtGLr5T9Uf7JnTHaOkxNeZT+7/qh8Yb5dIraJyanm9UtE1VYwdBNVNjprSdtQ31d0X1Jve7oHG6ejzVH31ONIZVf0NyKTFmyQj6BRPSMA+6jDE9GUFAdwCdSZjW0G16fxVwLkBK6asRcRiwHLh9sSc1xpgFRhwATkTDyhmSGmOKol88xAAxEZcDJ0fECRFxCFWQ6RbZ5xbg2fW5TgUOA/ZgjDEjMqJ+wYQ0zJauDvRGqJk2Z2rWbOyaGkCzKetUXZ0erVOFc5mF1dR//PHHo2id1aWg16zuOM22rtmK1VS9ffv2RllN6Wp612tUl4beI3VPqGn8lFNOaZSPPfZYFD2Gmt/1Pqi7UM3t6q5Ul4q2kaL3dZYs9k0xpfRgRLwG+DzVVOqLU0pXRcS7gK0ppS3AG4APRcSvUpntL0il5MvYD1HXWs71rmXtG7feemujrPoCsG7dukZ59+7djbJmS1f9yIVoqDtRHw/Vq3GzGHfihg0bGuWbbrqpUVZ9yLl9tayaq5qpmq5l1R/9HdLt+run9yy3fZaMUpdJaZgHXcaYnowoWpcAl8hn7+j4/9VUi0QbY8zYGXUAOAkN86DLGNNKqWuXGWNMjlL1a64HXerG0bKagTVDvLq+dNZabjaQmuLVTKumeTWl60LMOitmkIz06jpTl6i6/9TdePvtzXjBNWvW9K2TXpOeXxfEVpeHHk/N/2p6V9dem7tAr+H0009vlHOzE3XGlbaZtnkuo7M+d7OkJFeBaaJ6pX1HZyLrc5xzY6ubqc0tri5I7UsaUrFq1apGWfVk2Ozq27Zt66rTrFE3rIYj6IxuvU/qLlT3X262Yu6+59ybw86Kz9VmOYPNAAAgAElEQVRnlnpWon6Vo+7GmOIo8U3RGGMGoUT98qDLGNPKONYuM8aYWVCqfnnQZYzpSYlvisYYMwgl6tdcD7rUl60xCxrzoL51nRmq8Qht2c870Wzsur/WT6dba/yV1ufkk0/uOueNN9441DHUn69TujWGSmM8NP5Jp0/npkNrXJq2mbaRxoxpfIW2IXTHiWmcmWZ01li+FStWNMr6dqVtqnXW50pjc2ZJiW+KpiK3uoTGiGrf1ecwF7ujx4Pu+B2NY9W+o9s1rlVX8dB4oFwMl/YdjemcBrkVK3KreujvkP4uaJtoDJfqkf6O6flUg7XN9L5rffR6SqJE/ZrrQZcxpjelmueNMSZHqfrlQZcxpiclmueNMWYQStSvuR50qZlUzbC6YKyOmvWGqplVXVm5xVx1ey6zcC6dg7oSoduloKZmnd6sGaFzri81Rev+6h7U46urTVNGqPtR3RE6ZV1dgW3T5PU+aBvpdm1nrYO6fXIpJLSccydMkxLfFE07wy6AnVuoOPd96A4fyKXZ0XNo+ICuoqHhAZoBX/t7Wx1njbazptHQFBKqgaqR2sbaZtrGur+2uYZk6P6qX6pvuRU9ZkmJ+jXXgy5jTH9KfFM0xphBKFG/POgyxrRSakyEMcbkKFW/POgyxvSkxDdFY4wZhBL1y4OuDnTKtE711Ruovm/1devUW03PoKNwjX/SZX50GaGcr//+++9H0WPoNWn6BI0j0zg1jX/S+AJNIaExXbt27WqUNf5h/fr1jbLGcGj8k9ZXlxHR+kF33JfGsmg76zn1mvQ+6H1uW06lk5KEosQ3RVOheqN6pei91L6i8VDa1zV+C7o1TpewUU3Tc6omakoI3V/7TokxXIqmzdDfGY1b05grjWvV+6BtoMdXvdLj5Zb10ePpPdB7XJJmlFSXBTzoMsa0Uqp53hhjcpSqXx50GWN6UpLVzRhjhqFE/ZrrQVcuA7xOlVVTes6Mq/trSgg1E6spXdM36NRe3X7zzTf33R+63X3Kjh07GmV1523atKlR1mzI6i7Q7eoSUfefmtq1TXU6spbV1K7uDj0+wOrVqxtldSfqMdQFmptire5HbYOSplgrJb4pmopcqhN9TnMpJXR/7XttbvFcBnlN8aKaqa6pk046qVHWEAtNQTFtTjnllEb5uuuu69onl7Fd3Yd6357+9Kc3ytpGqh+KaqD+rqgbWn+n9HdFr6dtZYJO2tLyzIoS9WuuB13GmN5ERJFvisYYk6NU/fKgyxjTkxLfFI0xZhBK1K+iBl0RcS7wh8BS4E9TSr8zzfOra0xN52oW1huqZlo1ravZVV1vOirPzQTUTMCKmvKh2/R84okn9j3m2rVrG2V1veVcpLpdXW26XU3fumi3btdrVNO3mvaPO+44FHX75map6n3MzdBS83xutlBJ7sZR3hQH6c8R8e+AdwIJ+FZK6aWLPmEBzFLD9F7lZvbpc639QJ9D1UPo7k/q/tuzZ0+jnFshQsMZtI46807rPGna3ImK9v877rijUc7NRtSVRFQD1R2pmqfPQW7VAN1f3ci52YwHqn7V3x+7hg016IqIQ4ClKaXvZ3cekohYCnwAeA6wA7g8IraklK4e97mMMXlGmf0zSH+OiJOBtwLPTCndGRHd+TzGjDXMmPlg1NmLk9KwgWsUEa8DdgPbIuKaiHjNsBeR4anAtpTSjSmlB4BPAOeN+RzGmCFYsmRJz78Mg/TnXwQ+kFK6EyCldDsTxBpmzHwxgn7BhDQse+aIeF9EvBx4HXBqSmk18K+A0yLitwap+YCsBrZ3lHfUn2l9LoyIrRGxVU3XxpjxshCM2vYHLF/oi/XfhR1fHaQ/bwI2RcTfR8RltSl/EtdQjIZZv4yZHiPoF0xIwwZxL/4f4AxgOfAPEXEPcCXwbeDVEfHelNI4HOttzteuubEppYuAiwA2b97cf+5sBo0PysULaVn9xRo/pL599b2rb1197xpDoVOFNT5C46V0e1sd9Jr0GHoNWmeNAdMs1itWrGiUjzzyyEZZUzhoG+o1aDyExsXl0nxoGbpjGLQN9D5oCgmNsdD7lIvp0vPnpoRPiwHM83tTSpt7fb3lM72wg4CTgbOANcDfRsSTxqQnnRSjYePUL9UfLWtsTS7eUleH0L7dtsKF6se+ffsaZU0/oHFmGpOl/VP1RlNIzJq2NBq5WFu9xpzGamyv6oPG5eZWvND6qaZqDJrG7eWOX0rw+oj6BRPSsOygK6X0aeDTEfE04FepzPM/BjwZOAa4NCKOSCltzB0rww6gM2p7DbCrx77GmCkwQiDqIP15B3BZSulHwHcj4joqAbt8sSdtwxpmzHwyYiD9RDRsmCHpLwMfBd5L9db4JODbKaWnAKcNcZxeXA6cHBEn1MGuLwa2jOG4xphFMkJMxCD9+TPATwBExHIqU/2NTA5rmDFzxIgxXRPRsIFnL6aUvhMRZ1JF8j+Fyjz/pnpbf3vqYMd/sA5s/TzV9MyLU0pXjXrcfugoWM2+ap5Xc7xOd1ZXmB5Pzbrq6suZmdWsq8dTs7O69qDbVaYpINQ9qMdU94C2gWbB1zrv3r27Uc65//QeqAtY21wX9FZTuboOodtdqLE2ek3aJuoi1QVuFf1+zk00SxZbl179OSLeBWxNKW2pt50TEVcDDwG/nlLa1/uoo3GgaZjeG3Vzt9SvUVYtyIVXqB5Bt8tR+7v2v1x6Fg0PuPLKK/vur2l+VJ9yi4CPSlv2dXXXabvpD34ubY5q9KpVqxrl3ILU+jujmqr3MBc+MezKBrNkFC2dlIYNlTKiFqb/Wf+NnZTSJcAlkzi2MWY4Rp1y3dafU0rv6Ph/An6t/psK1jBj5oNR9Qsmo2FFJUc1xpRFKUGxxhgzLCXqlwddxpielOTqNMaYYShRv+Z60KVT+dUXrukIdH/1dWs8US5mQo+n2zU+Ydu2bY2yxg5oPJRO34buJWt02Q6NR9I3BY130hgpjQnTODVtE43J0HiC3Pf1Hmg8gU63bovx0HPoNWnMgpY1DYaeQ9swV+dShGIc5nkzOTRGS/u26oc+V7r/Mccc0yhrLJHGLkK35ul3tH9r/JLur+gyZKpxumzQtJ/XNj1p091OtA2e8YxnNMq63Jv+TmgMluqPxvpqm2mb6/eHjQ0slVL1a64HXcaY/pQyADTGmGEpUb886DLG9KTEN0VjjBmEEvVrrgdd6lbKmVXVDaTmezX76lRhzc6s06lzq8HrVOGcGVnPB90uBc02rKbxW265pVFW96OmU9A20ZQOOiVcz6/3QN0Tek3a5mo61zeddevWoeh90DrmsvhrHfSa1E2t16Tug1LM9x3LZZgC0R+UXPiC3svcc6au/LZnQd172n81XEDDD3R/1TTV3Pvuu69PjUdH++Y4zqfttnLlykZZNSvn9tU21e36XGiang0bNjTKqtmqsXqPcmV97mZFqfo114MuY0x/SnxTNMaYQShRvzzoMsb0pETRMsaYQShRv+Z60KVmXDWnq3tPzbrqZlJyi7eqO0DdVDqzUE3f6vp7whOe0Cir+xS6Tc06O0jN+bkFqnUGZW7x6JxrTc+n7gjNaK3H0/Pr7Kq2RXvVrat11jpqpn9tZzXXq7tShSCX9V+f02lRqnneVGhfzT1HuedO9UrdVm2ZxtV1rv1VOeWUUxrla6+9tu851HV19NFHN8ran0fNhq7XMw5Wr17dKGvIhc6wzmlqbiChGef1+/o7pM+BhlPo/rpdj+/wiP7M9aDLGNOfEt8UjTFmEErULw+6jDE9KfFN0RhjBqFE/fKgyxjTSqnJBY0xJkep+jXXgy4dBasvXGN1dDqzxhNoTIQeX2MsNJYoF2Omxz/ttNMaZU190BafoOkMNPuxxq3pNas/X9GYjlyG6htuuKFR1hgznbKt9dN4K42f0hgxbVPojkHQOmvsnsY46H3W1B56X/V4+lxo3MosKfFN0VToc6vPpfYF1SuNl9Lv6/6qFdAdl6pl7e9XXXVV3/1Vg3ft2tW3Tor2NdXMHOOIR1JN0thcTVujeqD3TTVQUz6oJmrc7caNGxtlvSfaZjpQ0evR7RpHp8/RLClRv8ppHWNMcZT4pmiMMYNQon550GWM6UmJb4rGGDMIJerXXA+69IaoGVXNpupuVNO4uvN0/9z39XxqhtbFptWdqFO8d+7ciaLme3UxbN++vVFWV5ma67/73e82yuoq0+nPakrX+mgb5BawVventpG6C/R80N1uuZUD1MWhddLnSNtE66B1LEUoSo2JMBXDusJy2d7VLaTPsa5GAbBjx45GWV3jul31IxfCodnbt23b1lWHTkZNGTEO1H2n4Qq5VESqyepu1BAO1Yv169c3yrmVCnKpjHLuQn0OS7gHUK5+zfWgyxjTnxJFyxhjBqFE/fKgyxjTSqnJBY0xJkep+lXeMNAYUwxLlizp+ZcjIs6NiOsiYltEvKXPfi+IiBQRm8daeWPMXDOKfsFkNMyWrg50VJxbPkHTEeR84xpjpb519dVrLNBNN93UKGs8hD5IGk/Vhl6zxlBoPILGdeTSK2h8Qm5KucYHaIyHHk/31+Pr9GmNd4Dua9AYCo250Jgsve86xTs3JVzLel/1OZsmi31TjIilwAeA5wA7gMsjYktK6WrZ70jgV4CvjVjVuSfXF/ReavoXjUnV+Ku251DjH/ft29coa3yPpjvQdArf+MY3GuWcPii5pdnGTVucm6ZgUA3LpdHRdtft2qaqH6qRGpOqMaiKLoPWtpxcJ6p/Jbn0RrF0TUrDymkdY0xRLASiLvJN8anAtpTSjSmlB4BPAOe17PdbwO8CP2jZZowxi2JE/YIJaZgHXcaYnizERbT9AcsjYmvH34UdX10NdE6F3VF/1nnsM4C1KaXPTvxCjDFzxwj6BRPSsLl2L+ZMj2rmVTNqLqOzfl/N+fp9NY2re1PLulq9ugfapvpqugN1WWqKCM2ar+46NUXnUlIouSnjJ598cqOspnu95jVr1jTK6hJpy6qtbz36HS3nrinnhs7dd3VnzpLMG+HelFKvGIa2zvWIbygilgB/AFyw6MrNOXpvtKz6pttz7kh9bnfv3t1VB3Wlq0Zpf1N3pJ5T9UT1QI+v28dNLmRE3aNtqAtSfweWLVvWKGtYyO23394oq0ZrHVWDNY2H6o22uWqs1lfvQe65miUj6BdMSMPmetBljOnNiHludgCdazqtATrXdDkSeBJwaT04WAFsiYifSSltXexJjTEGxpKnayIa5kGXMaYnIwSiXg6cHBEnADuBFwMvXdiYUrobWN5xnkuBN3rAZYwZFyOmjJiIhs31oCu3IHUuE6+a1u+6665GOZeJOJe5V11vq1c33MnZWS9t9deRvx5DTcm6IPXy5csbZXU/apsoaprWa1LT9LXXXtsob9q0qe/5tE3Vldc2A0vN61rW2T9qrs+5ddQFoouO52aVzZLFvimmlB6MiNcAnweWAhenlK6KiHcBW1NKW8ZYzblE7426jXLPpeqF7q+hBG3PQs4VdsoppzTK119/fd8664L3X/nKV7J16GTc7kd1Jw6CtrP+DujsQdUodeNqiIW6F1XT9PwaUqKaq22kIRv6G6G/KyXPXhylLpPSsLkedBlj+jPKm2JK6RLgEvnsHT32PWvRJzLGmBZGTY46CQ3zoMsY00qpa5cZY0yOUvXLgy5jTE9KXEbDGGMGoUT9mutBl/rrNR5AYx5y8QHqW9f4olx6BM0srLEAOpVX91ffu6aUgO6M83rNGte2YsWKRlnjjdSfr1n2cykbdDr1rl27GmW9R9qmGm+l8VgaM9J2D7XOGjOhKRw0BkOzcGvcWy5jtJZ1/1lS4puiqdAfFH2ONZZHn7NcLKH27bZs79o/NcZzz549fY+hGnfLLbc0yqqZGtOpKW/GnUIil5ajLeZL41L1mjUOVcsaB6fXpCuXqCaqXmkd9XyaxkOfCz2+bi8pRYRSon4VUaOI+L2IuDYiroyIT0fEsvy3jDGTZAwZnecGa5gxZVGqfpWinF8EnpRSejJwPfDWGdfHGEM2o7N5FGuYMYVRon4V4V5MKX2ho3gZ8IJpnDfn5lFTeW4xVTXnq1lXy2rOzy2UrAtmqxtLsyOruxHgO9/5TqOsI35Ne6EuC3UHqClbTc06pVyv+e67726U1bSurju9R2eccUajrG2oLmJ1l0K3+VzLel/1GrQDq0tE99fnQN3GuVQi08QWrcGYhYYtJp1BJzl9yy2IDd2alFusXV1Z27Zta5RVT1TjVA9yi0vr8zts38ppwyCoPqjmHXPMMY2yapiGO6gm633KZbzXNsulRtL9c/qVC9uZJiXqV3k1glcCn+u1MSIuXFgrSeMFjDHjo99boi1dfempYdYvY6ZDqfo1NUtXRPwNVZp85e0ppb+q93k78CDwsV7HSSldBFwEsHnz5nIj+Iw5ACjxTXFWjEPDrF/GTI8S9Wtqg66U0tn9tkfE+cDzgWenkqdDGDNH2KL1KNYwY/YvStSvImK6IuJc4M3As1JK9+f2HxfqG9f4H41fUtQXrjFUuSndej71fWv6A/WVq66r777NfaH+d30T0JgqjQHTc2qMlV6DpojQ+KjDDjusUdZlRDTORK9RY0o0JkTTL2hKDOhOW6HtnFt2I7fck7aJbtfnpJS3s1KTC5bILDRM9SQXM6r3UuOjcvFQbUt8af9STdOULtp3dEkc7c+aHmHdunWNssag5pYF07437rg46I59y6UGWrNmTaOsdVYN1TbWNtQ2Vs3V8+WWLcvF6en2UlLelKpfRQy6gPcDhwJfrB+Ay1JKr55tlYwxJYpWoVjDjCmMEvWriEFXSmnjrOtgjOmmRPN8iVjDjCmPEvWriEHXrNAboq4yNfOqOV/dRrpdpwarqV1dZ2rqVjOtugvUPaCj+rapurlz6jHVHK+mZ53Sra42dR+qaVq3b9++vVHW6dTqbtA2Uvej1l+/D90uSXUT631Vl4Zek5a1DuriPProo7vqVAKlmudNheqT9k3VN3VL6XOuoQPqNlI3FuTTFWgaG9XEnHtP+46GTOjxcozqThwETYuhIR25FDMnn3xyo5zTC/1d0ZVIhs1Ir+5R1WB9TtTFWopmlKpfcz3oMsb0p8Q3RWOMGYQS9cuDLmNMT0p8UzTGmEEoUb/metCVyzas5nc1z7fN5umHutL0+zpTULO5K7nZkWrqh3zWfTVF6/7HHXdco6zuSL0mNYWvXbu2Ud67d2+jrDNr1GWi16j3aPfu3Y2yuhvVlA7dsxd1to+6OLTN1D2Qm6WqddA2z2WIniYlvimadnILFWs5t3KCuu61r7ehri11Rel2DU/IZWvPaXLb7ORJovoI3TqvdVKNU83URbw1XCEX4qH6kXMP6v66akduJRZ1n5aULaVE/SpH3Y0xRVFqTIQxxuQoVb886DLG9KRE0TLGmEEoUb/Kq5ExphhGWbssIs6NiOsiYltEvKVl+69FxNURcWVE/O+IWD+RizDGzCWjrr04CQ2ba0uXxt6oL15jbTRVgKK+e72xmsJB4xM03kkz0ut06ZUrVzbK6uu/+eabu+qo/nk9Zy6eSWMy9Bp1WrnGP2lciMZ8aYyGfl+3a9zJhg0bGmWdTq1tBt1xGfpcaBtoG+p91LcrTUmhx9NYm1JiIkYxz0fEUuADwHOAHcDlEbElpXR1x27fADanlO6PiH8P/C7wohGrPTfoc6P6knuu9bnUbPCaOkC/D90pXnLf0f6u8Uwaf6TXoHFnw8ZwqZ5ofXOo3qlGQ3eMlV6DHmPjxmaKN70m1TiN8dI21rQ4Wp9cxnnVN/1d1Hugv5P6XM2KUd2Lk9IwW7qMMT0Z4U3xqcC2lNKNKaUHgE8A53XukFL6cseSOZcBazDGmDExoqVrIho215YuY0x/RnhTXA10mkF2AGf22f9VwOcWezJjjFFGjOmaiIbN9aBLzb7DTo3V0bKaXdW1pe4ANU2rWVhdcZpCQk376rZSMzR0u0h1erKa+9evb7qo1T23bdu2RjmX/TiX9VrLOoVc20BN8Tt27GiUN23a1CiraR262/mEE05olLVd26aJd6LPgd53FQI135cyzXmAN8LlEbG1o3xRSumiha+37N/qN42Inwc2A89aVEVNK+qKz+md6oX21TZ9POWUUxrlb37zm42yuprUBaqaqn1R07no9hNPPLFRVr25/fbbG+VRA6u1TfT40K2xqkG5jPTqYtUUDtpmqh/aZuqeVBdrbhHwQVY6KZER9QsmpGFzPegyxvQn8yO1N6W0uce2HUBnwOAaYJfuFBFnA28HnpVS+qFuN8aYxTKCfsGENMyDLmNMT0awDFwOnBwRJwA7gRcDL+3cISLOAP4EODel1G0yMMaYERjRsjkRDfOgyxjTyjBTq5WU0oMR8Rrg88BS4OKU0lUR8S5ga0ppC/B7wBHAp+rz3JJS+pnx1N4YM8+Mol8wOQ2b60GX+qbVl63TiXPLvyi63IvGH+mSNep715gvjdHILXHT9sDp9GI9py7Lo22gywBpTIXGM+g1rFq1qu/5NB5K31T0eBqjpfdU4x3a3nw0ZkLbTdNo6DFzKR90f70HJS8DNMqbYkrpEuAS+ewdHf8/e/E1M/qc6nOjfUm359KvqJ60pYzQONRTTz21UdY0NxqjqXqidcjFO2l8kx5f+6bGfObSamiKi3vvvbdRVj2Dbl3OpVzQNDuqYbq/xt5pH9VlhvQa9b7qc5FLdZSLZS4pIekYYvjGrmHlqLsxpjhKCeo3xphhKVG/POgyxrRS6tplxhiTo1T9mutBl5qS1X2opmu9gbnUATrFWk3dirqZ1Gyr9VW3Vy41AeTN6bpdzfs5U7LWSdtQy+oe1GtQ07fWN7d/Lkt32zHV7aLHzJW1jfScun+JwrBAiW+KpiKXmkRXStDnWtO/6HOqffmkk07qqoNqnLrfNKRC66DPvror9Ro1RUNuxQq9JtUDPV/OZavX25Y+Qa9J70POPafXpO7E3Kod6j7UcAc9v6b10Pqry1dpczuXQon6NdeDLmNMf0oeEBpjTD9K1C8PuowxrZRqnjfGmByl6tdcD7rU7Ko3SM2+Ss6sqt/PmXlz2xU1bQ+yULK6IHLn0NlCuTbLzeTTNtPtueOru0C3D1uftu/kzO25OuZcFMPe51lSct3mndyzr+RmoeVm3baFR2hIhM720/CBXHiAXoN+XxeH1uPlNFnrq22g588tTq+uPsiHVOgxcgti58IRcvqUm8Wq9dE2HFYDStKMkuqywFwPuowx/SnxTdEYYwahRP3yoMsY05MS3xSNMWYQStQvD7qMMa2UGhNhjDE5StWvuR505UbB4x4lDxsblIvRysVHDfKdXFzHsOQe8mHbdNj4qVx92q5Pj5E75qgxWSW+ffVif6qr6U8utjD3XLfFS+XiXjXdgJ4jF8+k6RA0JmvY+Cb9vl5TLoZ0EL3UGKlcbK8y7O+AkvtdyN3Xaf8uTpIS6zrXgy5jTH9KfFM0xphBKFG/POgyxrRSqnneGGNylKpfHnRNkFHNtLNwWw3rWpv0+cZtHp5GGx1IzNO1zhvDutUXQ861NWw6lWGznw/rShv1fIPUYdqUOPCYFrNu+zY86DLGtFLqm6IxxuQoVb886DLG9KTEN0VjjBmEEvWrqGFgRLwxIlJELJ91XYwxlWui15/pxhpmTDmUqF/FWLoiYi3wHOCWWdfF9KbENwczOXy/B8ca1s2wMZuTjvEcNSbM7F+UeP9Kel39A+BNwHgjtY0xi2IhJqK0N8WCsYYZUwil6lcRyhkRPwPsTCl9a4B9L4yIrRGxdc+ePVOonTHzyyiiFRHnRsR1EbEtIt7Ssv3QiPjv9favRcSGCVzCVBhUw6xfxkyPUQddk9CwqbkXI+JvgBUtm94OvA04Z5DjpJQuAi4C2Lx5s98ojZkgizXPR8RS4ANU7rYdwOURsSWldHXHbq8C7kwpbYyIFwPvAV40YpUnxjg0zPplzPQYxb04KQ2b2qArpXR22+cRcTpwAvCtuoHWAF+PiKemlG6dVv2MMU1GnHL9VGBbSunG+lifAM4DOgXrPOCd9f//B/D+iIg07mRwY8IaZsz+wxhSRkxEw2YeSJ9S+jZw/EI5Im4CNqeU9s6sUsYYYKQ3xdXA9o7yDuDMXvuklB6MiLuBY4H9qu9bw4wpkxED6SeiYTMfdI3CFVdcsTcibp51PYDl7Gc/FBPEbdGklPZYP+wXrrjiis8vWbKkX+qDwyJia0f5otp9BtCmdvr2N8g+ByzWr2JxezxKKW0xbf2CCWlYcYOulNKGIfY9boJVGZiI2JpS2jzrepSA26LJ/tweKaVzR/j6DmBtR3kNsKvHPjsi4iDgKOCOEc5ZBINqmPWrTNwej7I/t8WI+gUT0rAiZi8aYw44LgdOjogTIuIQ4MXAFtlnC3B+/f8XAF8qNZ7LGDN3TETDirN0GWP2f+r4htcAnweWAhenlK6KiHcBW1NKW4A/Az4SEduo3g5fPLsaG2PMo0xKw8IvlqMTEReKL3hucVs0cXuY0vEz2sTt8Shui/HjQZcxxhhjzBRwTJcxxhhjzBTwoGtMRMTvRcS1EXFlRHw6IpbNuk7TJrdkwjwREWsj4ssRcU1EXBURr5t1nYzphfXL+tWJ9Wty2L04JiLiHKqZCw9GxHsAUkpvnnG1pka9ZML1dCyZALxElkyYGyJiJbAypfT1iDgSuAL41/PaHqZsrF/Wr06sX5PDlq4xkVL6Qkrpwbp4GVVOj3nikSUTUkoPAAtLJswlKaXdKaWv1/+/F7iGKnuxMcVh/bJ+dWL9mhwedE2GVwKfm3UlpkzbkgnupEC98vwZwNdmWxNjBsL6Zf16BOvXeHGeriGIiL8BVrRsentK6a/qfd4OPAh8bJp1K4C5XtKlFxFxBPAXwOtTSvfMuj5mfrF+9cX61YL1a/x40DUEKaWz+22PiPOB5wPPnsPM2oMsmb6wxYUAAAH+SURBVDBXRMTBVIL1sZTSX866Pma+sX71xfolWL8mgwPpx0REnAv8PvCslNKeWddn2tTrTl0PPBvYSRWI+tKU0lUzrdiMiGp5+z8H7kgpvX7W9TGmH9Yv61cn1q/J4UHXmKiXATgU2Fd/dFlK6dUzrNLUiYjnAe/j0SUT3j3jKs2MiPgXwN8C3wYerj9+W0rpktnVyph2rF/Wr06sX5PDgy5jjDHGmCng2YvGGGOMMVPAgy5jjDHGmCngQZcxxhhjzBTwoMsYY4wxZgp40GWMMcYYMwU86DLGGGOMmQIedBljjDHGTAEvA2TGQkS8GTgJOB74MeADKaX/NNtaGWPMYFjDzDSwpcuMi9OpMlr/W+Ac4OWzrY4xxgyFNcxMHFu6zLh4MvCzKaWHIuIh4I5ZV8gYY4bAGmYmji1dZmTq1eiXp5RuqD96MtWaXcYYUzzWMDMtPOgy4+AU4JqO8lOAb82oLsYYMyzWMDMVPOgy4+B04MqOsgXLGLM/YQ0zUyFSSrOugznAiIhtwOkppe/Pui7GGDMs1jAzKWzpMmMlIpYBD1isjDH7I9YwM0ls6TLGGGOMmQK2dBljjDHGTAEPuowxxhhjpoAHXcYYY4wxU8CDLmOMMcaYKeBBlzHGGGPMFPCgyxhjjDFmCnjQZYwxxhgzBTzoMsYYY4yZAv8X7Rw9myTHtzMAAAAASUVORK5CYII=\n", | |
"text/plain": [ | |
"<Figure size 720x216 with 4 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"fig, axes = plt.subplots(1,2,figsize=(10,3))\n", | |
"ims = [0]*len(axes)\n", | |
"ims[0]=axes[0].pcolor(etaedges, phiedges, np.average(map(sum,all_jet_images0), axis=0), vmax=1.2,cmap=cmap)\n", | |
"ims[1]=axes[1].pcolor(etaedges, phiedges, np.average(map(sum,all_jet_images), axis=0), vmax=1.2,cmap=cmap)\n", | |
"\n", | |
"for iax, ax in enumerate(axes):\n", | |
" plt.colorbar(ims[iax], ax=ax)\n", | |
" ax.set(xlabel='$\\eta$',ylabel='$\\phi$') \n", | |
" ax.set(title='averaged {} calorimeter image'.format(['QCD','ttbar'][iax]))" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"# Training neural networks\n", | |
"\n", | |
"We will here train the neural networks for classifying between QCD jets and top quark jets. The trainig itself was done on a [Colab notebook](https://colab.research.google.com/) with TPU support to train the networks faster than on my laptop." | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"## run on Colab" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 9, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"name": "stderr", | |
"output_type": "stream", | |
"text": [ | |
"Using TensorFlow backend.\n" | |
] | |
}, | |
{ | |
"name": "stdout", | |
"output_type": "stream", | |
"text": [ | |
"We have 35379 QCD jets and 36708 top jets\n", | |
"((72087, 16, 22), (72087,))\n", | |
"((72087, 16, 22, 1), (72087, 2))\n", | |
"We will train+validate on 50000 images, leaving 22087 for cross-validation\n" | |
] | |
} | |
], | |
"source": [ | |
"import keras\n", | |
"data0 = np.load(outdir+'qcd_leading_jet.npz')['arr_0']\n", | |
"data1 = np.load(outdir+'tt_leading_jet.npz')['arr_0']\n", | |
"\n", | |
"print('We have {} QCD jets and {} top jets'.format(len(data0), len(data1)))\n", | |
"\n", | |
"x_data = np.concatenate((data0, data1))\n", | |
"# pad and normalize images\n", | |
"x_data = map(pad_image, x_data)\n", | |
"x_data = map(normalize, x_data)\n", | |
"\n", | |
"y_data = np.array([0]*len(data0)+[1]*len(data1))\n", | |
"\n", | |
"np.random.seed(0) # for reproducibility\n", | |
"x_data, y_data = np.random.permutation(np.array([x_data, y_data]).T).T\n", | |
"\n", | |
"# the data coming out of previous commands is a list of 2D arrays. We want a 3D np array (n_events, xpixels, ypixels)\n", | |
"x_data = np.stack(x_data)\n", | |
"\n", | |
"print(x_data.shape, y_data.shape)\n", | |
"\n", | |
"# reshape for tensorflow: x_data.shape + (1,) = shortcut for (x_data.shape[0], 16, 22, 1)\n", | |
"x_data = x_data.reshape(x_data.shape + (1,)).astype('float32')\n", | |
"x_data /= 255.\n", | |
"\n", | |
"y_data = keras.utils.to_categorical(y_data, 2)\n", | |
"\n", | |
"print(x_data.shape, y_data.shape)\n", | |
"\n", | |
"n_train = 50000\n", | |
"(x_train, x_test) = x_data[:n_train], x_data[n_train:]\n", | |
"(y_train, y_test) = y_data[:n_train], y_data[n_train:]\n", | |
"\n", | |
"print('We will train+validate on {0} images, leaving {1} for cross-validation'.format(n_train,len(x_data)-n_train))" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": null, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"from keras.models import Sequential\n", | |
"from keras.layers import Dense, Flatten, Dropout, Activation, Conv2D, MaxPooling2D" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": null, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"%%time\n", | |
"model0 = Sequential()\n", | |
"model0.add(Flatten(input_shape=(16, 22, 1))) # Images are a 3D matrix, we have to flatten them to be 1D\n", | |
"model0.add(Dense(2, kernel_initializer='normal', activation='softmax'))\n", | |
"\n", | |
"# Compile model\n", | |
"model0.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n", | |
"\n", | |
"history_logi = model0.fit(x_train, y_train, validation_split=0.2, epochs=40, batch_size=100, shuffle=True, verbose=0)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": null, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"%%time\n", | |
"model1 = Sequential()\n", | |
"model1.add(Flatten(input_shape=(16, 22, 1))) # Images are a 3D matrix, we have to flatten them to be 1D\n", | |
"model1.add(Dense(100, kernel_initializer='normal', activation='tanh'))\n", | |
"model1.add(Dropout(0.5)) # drop a unit with 50% probability.\n", | |
"\n", | |
"model1.add(Dense(100, kernel_initializer='orthogonal',activation='tanh'))\n", | |
"model1.add(Dropout(0.5)) # drop a unit with 50% probability.\n", | |
"\n", | |
"model1.add(Dense(100, kernel_initializer='orthogonal',activation='tanh'))\n", | |
"# model.add(Activation('sigmoid'))\n", | |
"model1.add(Dense(2, kernel_initializer='normal', activation='softmax')) # last layer, this has a softmax to do the classification\n", | |
"\n", | |
"model1.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n", | |
"\n", | |
"history_mlp = model1.fit(x_train, y_train, validation_split=0.2, epochs=40, batch_size=100, shuffle=True, verbose=1)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": null, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"%%time\n", | |
"model_cnn = Sequential()\n", | |
"model_cnn.add(Conv2D(32, (3, 3), input_shape=(16, 22, 1), activation='relu'))\n", | |
"model_cnn.add(Conv2D(32, (3, 3), activation='relu'))\n", | |
"model_cnn.add(MaxPooling2D(pool_size=(2, 2)))\n", | |
"model_cnn.add(Dropout(0.25))\n", | |
"\n", | |
"model_cnn.add(Conv2D(64, (3, 3), padding='same', activation='relu'))\n", | |
"model_cnn.add(Conv2D(64, (3, 3), padding='same', activation='relu'))\n", | |
"model_cnn.add(MaxPooling2D(pool_size=(2, 2)))\n", | |
"model_cnn.add(Dropout(0.25))\n", | |
"\n", | |
"model_cnn.add(Flatten())\n", | |
"model_cnn.add(Dense(300, activation='relu'))\n", | |
"model_cnn.add(Dropout(0.5))\n", | |
"model_cnn.add(Dense(2, activation='softmax'))\n", | |
"\n", | |
"# Compile model\n", | |
"model_cnn.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])\n", | |
"history_cnn = model_cnn.fit(x_train, y_train, validation_split=0.2, epochs=40, batch_size=100, shuffle=True, verbose=1)" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"Save NN models and training histories:" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": null, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"model_dir='trained_models_low_pt/'\n", | |
"if not os.path.isdir(model_dir): os.system('mkdir '+model_dir)\n", | |
"model0.save(model_dir+'logi.h5')\n", | |
"model1.save(model_dir+'mlp.h5')\n", | |
"model_cnn.save(model_dir+'cnn.h5')\n", | |
"np.savez(model_dir+'training_histories.npz', [ history.history for history in [history_logi, history_mlp, history_cnn ]])\n" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"## analyze\n", | |
"\n", | |
"Download the trained models from Colab, load them and analyze training/ROC curves (mkae sure you have loaded and reshaped the train/test datasets above)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": null, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"model_dir='trained_models_low_pt/'\n", | |
"\n", | |
"history_logi, history_mlp, history_cnn = np.load(model_dir+'training_histories.npz')['arr_0']\n", | |
"model0 = keras.models.load_model(model_dir+'logi.h5')\n", | |
"model1 = keras.models.load_model(model_dir+'mlp.h5')\n", | |
"model_cnn = keras.models.load_model(model_dir+'cnn.h5')" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 11, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"predictions0 = model0.predict(x_test)\n", | |
"predictions1 = model1.predict(x_test)\n", | |
"predictions_cnn = model_cnn.predict(x_test)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 23, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 720x288 with 2 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"fig, axes = plt.subplots(1,2,figsize=(10,4))\n", | |
"\n", | |
"nn=20\n", | |
"axes[1].scatter(nn, 1.*sum(np.argmax(predictions0, axis=1) == np.argmax(y_test, axis=1))/ len(y_test),c='b')\n", | |
"axes[1].scatter(nn, 1.*sum(np.argmax(predictions1, axis=1) == np.argmax(y_test, axis=1))/ len(y_test),c='y')\n", | |
"axes[1].scatter(nn, 1.*sum(np.argmax(predictions_cnn, axis=1) == np.argmax(y_test, axis=1))/ len(y_test),c='g')\n", | |
"for i, history in enumerate([history_logi, history_mlp, history_cnn]):\n", | |
" axes[1].plot(history['acc'][:nn], c=plt.get_cmap(\"tab10\")(i), label='train')\n", | |
" axes[1].plot(history['val_acc'][:nn], c=plt.get_cmap(\"tab10\")(i), ls ='--', label='validation')\n", | |
" axes[0].plot(history['loss'][:nn], c=plt.get_cmap(\"tab10\")(i), label=['logistic','small MLP', 'CNN', 'CNN v2'][i])\n", | |
" axes[0].plot(history['val_loss'][:nn], c=plt.get_cmap(\"tab10\")(i), ls='--')\n", | |
"for iax in range(2):\n", | |
" axes[iax].set(xlabel='epoch', ylabel=['loss','accuracy'][iax])\n", | |
"axes[0].legend()\n", | |
"axes[1].legend(['train','validation']);\n", | |
"fig.savefig('training_history.png')" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 24, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"from sklearn.metrics import roc_curve\n", | |
"fpr0, tpr0, thresholds = roc_curve(y_test.ravel(), predictions0.ravel())\n", | |
"fpr1, tpr1, thresholds = roc_curve(y_test.ravel(), predictions1.ravel())\n", | |
"fpr_cnn, tpr_cnn, thresholds = roc_curve(y_test.ravel(), predictions_cnn.ravel())\n", | |
"\n", | |
"from sklearn.metrics import auc\n", | |
"auc0 = auc(fpr0, tpr0)\n", | |
"auc1 = auc(fpr1, tpr1)\n", | |
"auc2 = auc(fpr_cnn, tpr_cnn)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 25, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 432x288 with 1 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"plt.plot([0, 1], [0, 1], 'k--')\n", | |
"plt.plot(fpr0, tpr0, label='Logistic regression (area = {:.3f})'.format(auc0))\n", | |
"plt.plot(fpr1, tpr1, label='Multilayer perceptron (area = {:.3f})'.format(auc1))\n", | |
"plt.plot(fpr_cnn, tpr_cnn, label='Convolutional NN (area = {:.3f})'.format(auc2))\n", | |
"plt.gca().set(xlabel='False positive rate', ylabel='True positive rate', title='ROC curve', xlim=(-0.01,1.01), ylim=(-0.01,1.01))\n", | |
"plt.grid(True, which=\"both\")\n", | |
"plt.legend(loc='lower right');\n", | |
"plt.savefig('ROC_curve.png')" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 26, | |
"metadata": { | |
"scrolled": true | |
}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 432x288 with 1 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"np.seterr(divide='ignore', invalid='ignore') # disable warning for 1/0 divisions\n", | |
"plt.plot(thresholds, 1/thresholds, 'k--')\n", | |
"plt.plot(tpr0, 1/fpr0, label='Logistic regression (area = {:.3f})'.format(auc0))\n", | |
"plt.plot(tpr1, 1/fpr1, label='Multilayer perceptron (area = {:.3f})'.format(auc1))\n", | |
"plt.plot(tpr_cnn, 1/fpr_cnn, label='Convolutional NN (area = {:.3f})'.format(auc2))\n", | |
"plt.gca().set(ylabel='1/$\\epsilon_B$ - 1/False positive rate', xlabel='$\\epsilon_S$ - True positive rate', title='Tagging efficiency vs. background rejection', xlim=(-0.01,1.01), ylim=(1,5*10**3), yscale='log')\n", | |
"plt.grid(True, which=\"both\")\n", | |
"plt.legend(loc='upper right');\n", | |
"plt.savefig('ROC_curve_bg_rej.png')" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"Check most certain and most uncertain NN predictions. Also check when the NN is most certain of its failed predictions" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 139, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "iVBORw0KGgoAAAANSUhEUgAABYoAAADECAYAAAAxghU6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzt3Xm4JFV9//HPZ1hFUMABIuu4AIpGkaC4oUMUBMSgoOhAEBQENUT5iQsuKCquiRoVog4RB0QGUUQxokJ8mIBLIiNxARFFHGDYBkR2UIHv749zGmp6eru93NPV/X49Tz1zu9bTdW99pvtbVaccEQIAAAAAAAAATK85pRsAAAAAAAAAACiLQjEAAAAAAAAATDkKxQAAAAAAAAAw5SgUAwAAAAAAAMCUo1AMAAAAAAAAAFOOQjEAAAAAAAAATDkKxQAAAACAkbN9sO37bd9p+4ml24PJY3ub/Pd1v+1DS7cH44P8wajZXmT7HtvLS7dlEBSKAQAAAKAGcqHjV7bvtn2D7c/ZXn8Gyy+z/cIhtqef9f0kItaNiMsq6/l/+f3cZvsk22t12OYLbP8m74PzbW9VmbZWXv72vL63VKatafvruc1he36Lde9g+4JcSLrR9pvz+C3zuOoQto/K03fJv5dbbf/R9lm2N2ta9wttX2z7LtvX2N4vj9/G9rds32T7Ftvft71th/dv2x/L2/mj7Y/bdod532376rxPTrf9iMr0zfK2b7G93Pbrm5b/+9zm221fafuwyrSO79n2v9r+ne078u/r1U3rXs32cbavy/P8X+Nv2farbF+e/x5W2D652u7KPJfl/fl72ztLUkT8NiLWlXRhu32I/pA/ku1DbV+RM+B7tjetTFs//62uyMOxlWkb216c/95vs/0j2ztVps+3/UBTxhyUp61l+4u2r6ocK3s0tWu/fDzcYfvXtl9amVYtkDeG+U3Lv9n2H/LxdJntbTrsg9nI6wOa2nu3U+b+XdP618zrX14Zt7Nb5/W+efqTnXL2ZtvRoe1b277X9qmVcbu4Q+5FxMGS9mi1vjqhUAwAAAAAY86pKPkxSW+T9EhJz5S0laTzbK9Zsm2DsP0iSUdLeoGkeZIeK+n9beadK+kbko6RtKGkpZK+WpnlWElbK+2XXSS93fbulek/lPSPkm5os+7vSfqCpEdJerykcyUpIq7OxaV1cxHybyU9IOnMvPivJb0oItaXtKmk30n6XGXd20k6TdK7lX5320v6WZ68vqSzJW0raRNJP5X0rdZ7S5J0mKSXSnqqpKdI2kvS4W3mfbWkAyU9J7frYZI+W5l+qqQ/5O2+WNKHbe+S27yGpLPy/nikpFdK+qTtp/byniXdJekledmDJH3a9rMr098v6dmSniXpEbmd9+ZpP5L0nIh4pNLfw+qSjmssaHtXpWPhNZLWk/Q8SVe22QcYAvJHsv18SR+WtLdS/vxB0uLKLJ+StE5ezzMkHWj7NXnaupIukvR3edmTJX3H9rqV5a+r5kxEnJzHry7pGknPV9r3x0g6w/a83K7NlI7ltygdS2+TdJrtjSvr/knTupdU3tehkg5RyoB1lTLl5iHsr77zOiK+0pS5b1Q6xi9u2szbJK2ojoiIC5uW3UvSnUr5Lkl/lXRGfs+dnKD0O6vqlnuTISIYGBgYGBgYGBgYGBgYxnRQ+vJ/p6T9msavq/Ql+bX59SJJx1Wmz5e0PP/8ZaXi5j15XW9X+qIfSsXH6yRdL+moyvIzWl8P7+NgST9sGneapA9XXr9A0g1tlj9M0o8rrx+et/+E/PpaSbtVpn9Q0ukt1rNc0vymcR+W9OUefx/vk3R+m2lrSfqIpF83vccP9rjuDfPv5FFtpv9Y0mGV14dI+p82835d0tsqr5+tVIxdJ//thKSNKtMXNvaBUvE4JK1TmX6RpAW9vOcW85zd+NuStEH+m3lcD/tjXUmnSDqnaR8c0mW5JZIO7ed4Y1hlX5I/adq/Sjqh8nrT3P7H5dc3S3p6Zfq7JF3YoT23S/q75vfW4+/kl5L2zT/vJGlF0/SbJD2r3fuuzDdHqQj9gh63O+t5naedL+l9TeMeI+kypSt42+47SV+S9KUW4x8vKdos8yqlYvKxkk5tM0/L3Jvp73IcB64oBgAAAIDx9mxJaytdnfWgiLhT0ncl7dptBRFxoKSrJb0k0pVWH69M3kXpyq7dJB3dy+3c7dZn+5e29+/tbUmSniTpF5XXv5C0ie1HdZs3Iu6S9HtJT7K9gVLhpnldT+qxHc+UdIvtH+fbxr9te8s2875a6YrABzl1T3GrUiHkrZKq+/eZeZ5f2b7e9qm2N2yz7ucpFV7+2GZ6q/3V7j06D9XXayn9rl0ZV53+ZEmKiBuVrpZ8jVM3Ec9SuvLvhw/O3Pk9qzLfwyQ9XdKledTfSrpP0svzLee/tf1PTcs81/Ztku6QtK+kf8vjV5O0o6SNnLoAWG77+LwNjAb5k7Q6nqR8zDSNa/xcnfbQBHt7SWtKuqIyemOnLm/+YPtTth/eZtlNJG2jh46npZIus/0P+Vh9qaQ/KxWTG56Wu1r4re1jbK+ex2+ehyc7dYnzB9vvt92uVjjreZ27q3ie0gmjqs8qFePvadNW2V5H0svVlNedOHVz8wFJR7WZ3lPu1RmFYgAAAAAYb3Ml3RwR97WYdn2ePoj3R8RdEfErpauvFvS7ooh4SkScNoNF1pV0W+V14+f1epi3Mf96eZq06rparaeVzZW6SHizpC216m3lklL/l0pX2369Oj5S9xTrK/0u3iPpN03rPlCp4Lm1Vu0CorHuzZVud35L87SKVvtrXbtlP8XflXSo7Xm2HynpHXn8OhFxh1IXD8fYXtv2Drl961SWXyzpvUpFpwslvTsirunxPVd9XqkI9P38enOlW+i3Uboq8OWSjs1dSjTW/cNIXU9sLulfJC3LkzaRtEZeZmelbjyelreP0SB/knMk7Wf7KfnExHuVr7rP07+nVOhez/bjJb1WKx9Pkh4sRH5Z6X03tvcbpb/lR0v6e6UuKj7ZYtk1JH1F0skR8RtJioj7lYqopykdq6dJOjwXZiXpAqWC9cZKx/gCpS4bpHR8SalI/7dKRfsFat8tQ4m8frXSldl/aIyw/TJJq0fEWW3a2bCv0pXe/91lvqoPSvpiNeuqZpB7tUWhGAAAAADG282S5lauAqt6tNr0JzkD1S/EVyld6TVb7lS6tb2h8fMdPczbmP+OPE1adV2t1tPKPZLOioiLIuJe5T50c4G16iBJZ+arKVcREbcoXb32rcrv6x6lW59/m5f7sKQ9q8vZ3kipT+R/j4hVCtQVrfbXnZHveW5yklKxd4nS1Yfn5/GNBz8doFSovUapn82vNKbZfoJSf6KvVrry8UlKfYi+uMf33Hhf/6JUpNqv0sbGFYAfiIh7IuKXkk5X0z7J675WqQB3etOyn42I6yPiZqWC2irLYmjIH0kR8QOlbmfOVGrnsjxf43h6k9Lf5++U+hlfXJkm6cGr67+t1F3MRyrrviEifh0RD+SC6NuVToZUl52jVGD+i6QjKuNfqHRV63ylY/X5kv4jX7WsiLgyIv6Q1/0rpatlG+tuHE8fj4hbI2KZUr/k7Y6nEnm90h0c+Urrj0v65zZtrDpI0ilt8nEVeZ+9UKm/6Y465V7dUSgGAAAAgPH2E6UrxfapjsxfmPeQ9IM86i6tfAXb3zStp92X5S0qP2+p1F/oIOubiUuVHszW8FRJN7bpemGlefP7f5ykSyPiT0pXNzav61L15pda+f00fn7wSt1c5HmFut/GvLrS1XuNIkjzuleSb8M+V9LZEfGhLututb9avsdcGHpfRMyLiM3zfNfmQRFxVUTsFREbRcROSg/x+2le/MmSLo+I7+f1XC7pO0p/b728Z9l+f55/t4i4vTJv45b4Xv9+Vlf6PSv/npfPYFkMjvxpbDDihIjYOiI2VioYry7pkjztlog4ICL+JiKepFRvaxxPsr2WpG8qHX/tHkD54Ka0cvZY0heVrqjfNyL+Wpl3e0kXRMTSfKxeJOl/lQqe3dZ9uVLhudd9Oat5bbvxIM7qHRxbK/VvfaHtG5S6RHl07sZmXmXZLZSK581dVnQyP6/76rzut0ra13bzQ/QaVsm9SUChGAAAAADGWL49+f2SPmt7d9tr5C/EX1Mqmn05z/pzSXva3tD230g6smlVNyo9pb7ZMbbXsf0kSa/RQ0+m73d9M3GKpENsb5cLpu9ReohVK2cp9aW5r+21lW79/mXjFuy8rvfY3iBfEfu66rpsr5WXk6Q1c5cLjYLJlyS9zPb2+fbuY5QeAHVrZfsvk3SrHroyt7HefWxva3tOvjL4k5L+L19x1lj3a2w/NveZ+Q5J/5mXfYRSlww/ioije9xfb7G9me1NlfrRbLm/8u/tcU62y+36QEQ8kKc/Md8mv6btf1S6/bxxu/v/Sdra9t/n5R8naS/lPkW7vWfb75S0v6Rdm4tIEfF75a4s8u/kiZJeWdknB+R+QJ37J/2QHipGNvbnP9veOP/NHNlYFsNH/iQ5L56c/y63VHr446dz0VP5WHuUUz/Beyg9zO24PG0NpWLnPZJe3TgGK+ueX/mb30LSR5WuSm74nKQnKvXJ3Nwn70WSdm5cQWz7aUrdsvwyv97DqV/jxp0CxzTWHRF3K+3vt+cs2FwpN9sdT7OW11njDo7qlcaXKJ1c2D4Phyr9LWyvla9OP1DpYXq/r64w7+O1la6+bvxe18qTFyoVsxvr/rzSCbIX5Xm7Zf1kiDF4oh4DAwMDAwMDAwMDAwND50Gp38hLlIoNNyrdIrxBZfraSl/6b1cqEvw/VZ6+LmlvpQdA3ap0pdQ8pSvJDlO6iu8GSW/vd3153KWSDmjT/oOViq/N49+S38/tSkXAtSrTVlqf0lVyv8n7YImkeZVpayl1t3B7Xt9bmrazLL/f6lBd/g1KV/v9Sen28C2alv++pA+2aP8/K/VpfFfeh6dL2qppnvdLuikPX2783pQKIZGXvbMybNlmH1rptutb8vBxSa5Mv1PSzvnnbZSuGLxb6Vb55v1xZG7PXUoPqduxafp++e+tcXv9xyTN6eU95/f056b39K7K9M2UupS4U9KVSn2qNqZ9KG/vrvzvQkmPqkxfQ9K/57+7GyR9RtLaTW1fIunQ0sfsJA2a8vyRtH5uR+Nv/iOSVms6Xq7Lx9vPJb2oMu35+b3e3XRM7Fxpw7V5+jVKfZivl6dtlZe9t2nZai4eofRgvDvy8XRUZdq/5vd3V572AUlrVKY/Qun4vSNv+72qZEq/+yu/HiSv186/2xd0+bucX/27qIz/jaRDWoxv/N1Vh2Vt1n2spFMrr3vJ+pbtqdPg/EYAAAAAAFMkXxX4B6WiQasHVQ17ewcqFZf+IulZEXHZqLeJ6WJ7a6UrLNeU9MaIWFS2RWiH/MGksf1Fpe6JVkTE40u3p18UigEAAABgCs12oQYAGsgfYDzRRzEAAAAAAAAATDmuKAYAAAAAAACAKccVxQAAAAAAAAAw5SgUQ7aX2X5h/vldtv8j/zzPdthefQjbuNT2/EHXA6BebB9r+9QO05fYPjT/fIDtcyvTwvbADwGw/V3bBw26HgCQVv6s1Meyn7d9zLDbBAAAAAwDheLCuhVR8jwPFnJHLSI+HBGHjmC9T4qIJd3mG2ZxGkBrveROCRHxlYjYbQTr3SMiTu5l3mEVp4FpN64500r1hFUvBvmsFBGvj4gP9tiuWfv8B4zSTI+xYS3bZZ332r5gmOvtcdsvtH2n7Qc4voHRIntW2jbZg55RKAYAAMDUccJnYaBHE3by4oiIeF7jhe0jbC+1/Wfbi2ayopwlH7P9xzx83LZbzRsR/xUR60q6erDmz6h9B9u+PxeJGsPxTfMcl0+W79A0/lDbSyqvl3OXKGbbhGfPhrbPsn2X7ats79/rimzvZ/vHtu+uHqetkD2YCT4cd5AD6W22f5kP3C/a3iTfxnyH7f+yvUFl/n/IXSzcms8WPbEy7R22r83LXW77BbZ3l/QuSa/MB80vWrThy5K2lPTtPM/be9jWMtvvtP1r23+y/SXba/f4ntteAWR737zuJ+fXz8zBdKvtX3Q6cL1y9xZzbB9t+/f5w9QZtjfMszbOrt2a3++zemk3MCnGIXfysk+zfXFe9qu2T7d9XGX63rZ/bvv2fCzvnsc/xvZ/5+XOkzR3Bu/9YNs/bDPtubavsb1Lfv0E2+fZviW/t/06rHelKwJsv9b2ZTkfv297qzy+kT+/yPvmlb22HaiTMcqZla7gt72okTO25zt9KTjK9grb19t+TWXeh9n+hNOXqtts/9D2w/K0tp9Pcvs/ZPtHku6W9GVJO0s63pUvMLY/nTPndts/s71zZR0PflbyQ3dCHWT7ats32353h33/4HvMr/fKWXprbvNT8viWn/8AjMx1ko6TdFIfyx4m6aWSnirpKZL2knT48Jo2FD+JiHUrwxGNCbYt6UBJt0iiqy5gdp0g6S+SNpF0gKTP2X5Sj8veIunfJH10RG0bBrKnhigUd7evpF0lbSPpJZK+q/TlZ67S/nuTJNneRtJiSUdK2kjSOUof7te0va2kIyQ9PSLWk/QiScsi4nuSPizpq/mgeWrzxiPiQKWzPi/J83y807Yqix6Qt/O43Pb3DLIT8pezj0l6YURcYnszSd9R+kC1oaS3SjrT9kY9rO5NSh+mni9pU0l/UgpISWqcXVs/v9+fDNJuoKaK5k7Okm8qFVA2lPS13KbG9GdIOkXS2yStr3TcLsuTT5P0s9zWD2oI/+nbflF+n/tGxPm2Hy7pvLytjSUtkPTvvXyosv1SpX25j9I+uzCvW5Wz+0/N++arg7YdGGNFc6ZHfyPpkZI2k3SIpBMqBex/lfR3kp6tlFNvl/RAj59PDlQq7Kwn6WClHDii6QvMRZK2z+s4TdLX3Pmk+3MlbSvpBZLeWy2mt+N09cxJSgWlR0n6gqSzba/V6vNft/UBpdjewPZ/2r4pn4T9T9ubN832ONs/zSd2vuWHLhKZ0cUnoxIR34iIb0r6Yx+LHyTpExGxPCKulfQJpWypi12Usv9ISfvbXqNwe4Ce1D178neafSUdExF3RsQPJZ2t9Dmlq3yV8BlKJ7rqiOwZUxSKu/tsRNyY/9O/UNL/RsT/RcSfJZ0l6Wl5vldK+k5EnBcRf1X6AvMwpS8w90taS9J2tteIiGUR8fsB2tRpWw3HR8Q1EXGLpA8pFVL6daRSQWh+RFyRx/2jpHMi4pyIeCAizpO0VNKePazvcEnvzh+m/izpWEkvN/0SAw2lc+eZktaQ9G8R8deI+LpS0aThEEkn5e0+EBHXRsRvbG8p6elKH3b+HBEXSPr2QHtCeoWkhZL2jIif5nF7KRWjvhQR90XExZLOlPTyHtZ3uKSPRMRlEXGfUjFre+erioEpUjpnevFXSR/IOXSOpDslbevUXcRrJb0558/9EfHj3PZePp8siohLc378tdWGI+LUiPhjnucT+X1u26Gt74+IeyLiF5J+oXRlYTevk/SFiPjf/B5OlvRnpQwG6mSOpC9J2krpSvh7JB3fNM+rlY7bTSXdJ+kzkjTIxSe2988FnnbDlkN6f908Sem4b/hFHlcXB0n6lqQzJK0uaY+yzQF6Vvfs2UbS/RHx28q4uuXHIMieMUWhuLsbKz/f0+L1uvnnTSVd1ZgQEQ9IukbSZrm4eqRSQXSF0y3cmw7QprbbqsxzTeXnq/IycrqttNE/zAE9bu9tkk6IiOWVcVtJekU1EJWupnl0D+vbStJZleUuU/qyuUmP7QEmXenc2VTStRERlXFXVX7eQlKrYtCmkv4UEXe1Ws725yv5864e23KkpDMi4leVcVtJ2qkpfw5Quvqwm60kfbqy3C2SrJXzE5gGpXOmF3/MJ3Qa7s7tmitpbbXOoV4+n1zTYrmVOHV5cVm+AulWpSubO3Wlc0OLdnazlaSjmtq6hfJnNqAu8kmVMyPi7oi4Q+kilec3zfbliLgkf0Y4RtJ+tlfTABefRMRpEbF+h2G2+uJcV9Jtlde3SVo331Y9Lp7ZVMh6prTSFY2n5ZNt3xC3gKMmJiB7mrND+fV6PS5fB2RPDVEoHp7rlD7wS3qwv5UtJF0rPRgmz83zhFI3Dso/d9M8T8dtZVtUft4yL6OI2CMe6h/mKz1sW5J2k/Qe2/tWxl2jFLrVQHx4RPTSP841kvZoWnbtfFVTL/sDQDKq3Lle0mZNX3CqZ8avUerWptVyG+T/+FdZLiJeX8mfD3dpQ8MrJL3U9pFN2//vpgxZNyLe0MP6rpF0eNOyD4uIH/fYHmDajPLzzd2S1qm87uVkjyTdLOletc6hXj6fNLdtpddO/RG/Q9J+kjaIiPWVvrgNu+hzjaQPNbV1nYhY3KadwFiyvY7tLzj1GX670jNH1s/FmIbmi1jWUDr5MsjFJ+PiTkmPqLx+hKQ7m064j0TzSXjbO1deX1qZ9X+asuZ/8viXK+Xp9/Prr0jaq3p7fo/tOLey3Vc69dveeD3o3WVASxOQPc3Zofz6jllsQ1/InslGoXh4zpD0YqeHuKwh6Sil2wd/bHtb239vey2lg+EepStopXQFzzx3fur2jZIe28u2KvP8k+3N84H2LkmD9LV5qaTdlfoG/Ic87lRJL7H9Itur2V7b6cEzzX0CtfJ5SR/yQw+Q2sj23nnaTZIe0MrvF0Bro8qdnyjdmvUm26vb3kfSMyrTvyjpNXm7c2xvZvsJEXGV0pn49zv1X/pcpb5PB3GdUp+fb7L9xjzuPyVtY/tA22vk4enuoU9Qpfx5p3N/xrYfafsVlenNeQtMu1F+vvm5Up90qzk9AK/5KqCW8lXNJ0n6pO1N8/LPyu3o5/NJ83G/nlIG3iRpddvv1apf5IbhREmvt72Tk4fbfrHtxpVE5BHq4iilrll2iohH6KFnjlRPrjRfxPJXpZM+fV98YvuASkGg1TBbXU9cqpW7m3lqHjdyzSfhI+LCyutebl8/SCnfrrF9g1Kf9GtKetUM27FbZbtfjYiTK68H/SwItFP37Pmt0ueMrSvjZi0/BkH2TDYKxUMSEZcr3b7wWaXgeYnSA0j+otSv3Ufz+BuUHr7UuO36a/nfP9q+uM3qP6J0Re+ttt/aZVsNp0k6V9KVeThOA4jU595ekk60vUdEXCNp7/w+blIK2rept7+pTyt10n6u7Tsk/Y+knfJ27la6ZeRHrtyaAGBVo8qdvPw+Sg9i+ZNSH6XfqEz/qaTXSPqU0lV2/62HrjjcX+l4vkXS+5Qeejfo+7xaqVj8DtuHRrq1bDelDxLX5ff3sfyeu63rrDzv6fnKg0u0cn9Yx0o6OefPfoO2Hai7EX++eXNeX6P7mG/OoGlvlfQrpf7Tb1E6ruf0+fnk00rPSviT7c8oXd3yXaUvcFcpFcG7dlcxUxGxVKmf4uOVsvYKrfwArJU+/w17+0Cf1sgnYBrD6konV+6RdGu+SOV9LZb7R9vb2V5H0gckfT0i7tcAF59ExFcqBYFWQ89dT+QT42tLWk3SapX31otTJL0lnzjfVKl4tajXbZeSi1nzlT4HbZ+Hpyo9jK/TLeBrtvgbAEZt4rInUncY35D0gXyy+DlKn2G+3MvyjXYr9e87J7+HsX8gHNlTAxHBMGGDpGWSXli6HU1tulrS80q3g4GBob9B6QvPcaXb0WfbL5D06tLtYGBgYFAqKL23dDsYGPoZ8neMaBqOU+pXe4nSbdS/VXpwbEhaPS+3ROnEx08l3a70oNu5lfXupHTS+RalEzzfkbRlZdlDh/w+Vlmn0oni5vd2bJ62ZX5vW7ZZnyV9PLf/lvyzK9PvlLRzi305K9/XlE5A/bDF+PcoPci0efwWSndVPEHSoZKWVKYtb7efGBhGNUx49myodKL8LqWayf6VaTsrdWPTbn0Ht9gviyrTyR6GvgbnnY4JYnuZUgD9V+m2SKlrCaXQ2zZm76ESAIbI9iJJyyPiPaXbMhP56oHfSloQEReWbg+A6ZWvfPmRpOMjoqerhQAMn+1zJT1L0tKI2GWWt/0CSWcq3ZGxZ0ScP5vbB1AO2YO64FJtjJTtp0s6T9JnKRKX4/RwsX+X9BelM3O9PsgQqC3bGyvdyv1tST8s3JypRf4AD7pB0s+UvqhhxMgetBMRuxXc9g8krV9q+xg9sgftkD0YpWFmD30UT6CImDcuVxNHxEWROoZ/e+m2TBrbJ9leYfuSpvG7277c9hW2j86j91Hqj+l1kv5hlZUBXUTEwXW7mjgiVkTEIyLigOD2maEif4CZi4i5EfGiSM9jQB/IHgAlkD0ASiiVPRSKgfpaJGn36gjbq0k6Qalj+O0kLbC9naTN9dCDeO4XAAxmkcgfALNvkcgeALNvkcgeALNvkQpkz6x2PTF37tyYN2/ebG4SmFU/+9nPbo6IjXqd33a7qxwvVXrKesPCiFhYnSEiLrA9r2m5Z0i6IiKuzOs/XenJqcuVguPnmsITRGQPJt0Qs0cif4aG7MGkI3vGF/mDSUb2jC+yB5NuGmo+AxWKbe8u6dOSVpP0HxHx0U7zz5s3T0uXLh1kk8BYs31VH8usMi4i7o2IHftowmZ66CySlMJiJ0mfkXS87Rcr9ddaezPJH7IHk25Y2SORP92QPcBDyJ7Zw/cu4CFkz+whe4CVTUPNp+9CceVy511zwy6yfXZE/HqQBgHTpk1o9L26FuMiIu6S9Jp+VzpuyB9gcB2+MPW9ytarm5z8IXuAwZE9M0f2AIMje2aO7AGGo241n0EuR37wcueI+IukxuXOAGZgzpw5qwyS5tpeWhkO63F1yyVtUXm9uaTrht3mMUD+AANqlT3kT1dkDzAgsqcvZA8wILKnL2QPMAR1y55Bup5od7kzgB7ZboREs5v7vA3hIklb236MpGslvUrS/gM0cVyRP8AAOmSPRP50QvYAAyB7+kb2AAMge/pG9gADqmPNZ5Arilte7rzKTPZhjQr5TTfdNMDmgMnU5uxSV7YXS/qJpG1tL7d9SETcJ+kISd+XdJmkMyLi0pE1vpyu+UP2AJ11uLKmqynOH7IHGBDZ0xe+dwEDInv6QvYAQ1C37BnkiuKeLnfOT+1bKEk77rhj351wAJMMqCn1AAAayUlEQVSow9mlubarTwFo9QTMBa0WjIhzJJ0zvFaOpa75Q/YA7XW5sob8aY/sAQZA9vSN713AAMievpE9wIDqWPMZpFA8DbdaACM35NsQpgX5AwxoBLdgTgOyBxgQ2dMXsgcYENnTF7IHGIK61Xz6LhRHxH22G5c7rybppAm81QIYuV5vO8BDyB9gcGTPzJE9wODInpkje4DBkT0zR/YAw1G3/BnkiuJpuNUCGKlBbkOYduQP0L9Bb8GcZmQP0D+yp39kD9A/sqd/ZA8wmDrWfAYqFAMYXN1uQwAwGbgFE0AJZA+AEsgeAKXUreZDoRgoqMvZbQAYCbIHQAlkD4ASyB4ApdQxfygUA4XVLTQATAayB0AJZA+AEsgeAKXULX8oFAOF2W41emz7qwEwGdpkj0T+ABghsgdACWQPgFLqVvOhUAwU1OE2hLHtrwZA/XW5BYr8ATASZA+AEsgeAKXUseZDoRgorG63IQCYDGQPgBLIHgAlkD0ASqlb/lAoBgqrW2gAmAxkD4ASyB4AJZA9AEqpW/5QKAYK6nAbwtj2VwOg/rrcgkn+ABgJsgdACWQPgFLqWPOhUAwUVrf+agBMBvrqA1AC2QOgBLIHQCl1q/lQKAYK6nJ2GwBGguwBUALZA6AEsgdAKXXMHwrFQGF1Cw0Ak4HsAVAC2QOgBLIHQCl1yx8KxUBhdQsNAJOB7AFQAtkDoASyB0ApdcsfCsVAQXW8DQFA/ZE9AEogewCUQPYAKKWO+UOhGCjMdqvRY/sETACToU32SOQPgBEiewCUQPYAKKVuNR8KxUBhdXsCJoDJwNO/AZRA9gAogewBUErdaj5TWShevHhxx+kLFiyYpZZg2tXxNgQA9Uf2ACiB7AFQAtkDoJQ65s9UFoqBcVK30AAwGcgeACWQPQBKIHsAlFK3/KFQDBRUx7NLAOqP7AFQAtkDoASyB0ApdcwfCsVAYXULDQCTgewBUALZA6AEsgdAKXXLHwrFQGF1Cw0Ak4HsAVAC2QOgBLIHQCl1yx8KxUBBHW5DmGt7aeX1wohYOEvNAjDhutwCRf4AGAmyB0AJZA+AUupY86FQDBTWJjRujogdZ7stAKZHhy9M5A+AkSF7AJRA9gAopW41HwrFQEF17NgcQP2RPQBKIHsAlED2ACiljvlDoRgorG6hAWAykD0ASiB7AJRA9gAopW75M5WF4gULFpRuAvAg26WbAGAKkT0ASiB7AJRA9gAopW75M5WFYmBc1PE2BAD1R/YAKIHsAVAC2QOglDrmD4VioLC6hQaAyUD2ACiB7AFQAtkDoJS65Q+FYqCwuoUGgMlA9gAogewBUALZA6CUuuUPhWKgoDrehgCg/sgeACWQPQBKIHsAlFLH/KFQDBTWJjTm2l5aeb0wIhbOUpMATIEOH1jIHwAjQ/YAKIHsAVBK3Wo+FIqBgjqcXbo5Inac7fYAmA5dzmyTPwBGguwBUALZA6CUOtZ8KBQDhdXtNgQAk4HsAVAC2QOgBLIHQCl1yx8KxX1YtmxZ13nmzZs38nYMwze+8Y2O0/fZZ59Zasn0qltoAJgMZA+wqsWLF3edZ8GCBbPQkslF9tRHt+884/J9Z8mSJR2nz58/v+s6nv70p3ecftFFF3Wc3u07ldR9f+2www5d14H+kT0ASqlb/gxUKLa9TNIdku6XdN+4XjYNjKs6dmw+LsgfoH9kT//IHqB/ZE//yB6gf2RP/8geYDB1zJ9hXFG8S0TcPIT1AFPJdukm1Bn5A/SJ7BkI2QP0iewZCNkD9InsGQjZAwygbvlD1xNAQXU8uwSg/sgeACWQPQBKIHsAlFLH/Bm0tSHpXNs/s33YMBoETJs5c+asMqAn5A8wgFbZQ/70hOwBBkD29I3sAQZA9vSN7AEGVLfsGfSK4udExHW2N5Z0nu3fRMQF1RlymBwmSVtuueWAmwMmz7iHxBjrmD9kD9AZ2dM3sgcYANnTN753AQMge/pG9gADqlv+DNTaiLgu/7tC0lmSntFinoURsWNE7LjRRhsNsjlg4jRuQ6jT2aVx0S1/yB6gvXbZQ/50R/YA/SN7+sf3LqB/ZE//yB5gMHWs+fTdOtsPt71e42dJu0m6ZFgNA6ZF3UJjHJA/wOD4wjRzZA8wOLJn5sgeYHBkz8yRPcBw1C17Bul6YhNJZ+Wn960u6bSI+N5QWgVMiTp2bD4myB9gAGRP38geYABkT9/IHmAAZE/fyB5gQHXMn74LxRFxpaSnDrEttTFv3rzSTehJDvSOut0assMOO3RdR132x7iqW2iMg2nOH2BYyJ6ZI3sm34IFC7rOc+KJJ3acvuuuu3ZdxzR/diJ7Zq5U9ozD3+myZcu6zjN//vyBt/O1r31toOXf8IY3dJ3n+c9/fsfpBx98cNd17Lnnnr02CU3Inpnjc8/oLVmyZKDlh5F/GL265c+gD7MDMKC6hQaAyUD2ACiB7AFQAtkDoJS65U+9WgtMmNIdm9t+rO0v2v76rG0UQHHj8FAX8geYPmQPgBLIHgCl1LHmQ6EYKKzf0LB9ku0Vti9pGr+77cttX2H76E7riIgrI+KQAZoPoKYG+cJE/gDoF9kDoASyB0Apdcseup4ACuulL+k2Fkk6XtIplXWtJukESbtKWi7pIttnS1pN0kealn9tRKzod+MA6m2A7JHIHwB9InsAlED2ACilbjUfCsVAQR2egDnX9tLK64URsbA6Q0RcYHte03LPkHRFfvCAbJ8uae+I+IikvYbWcAC11uXpu+QPgJEgewCUQPYAKKWONR8KxUBhbULj5ojYsY/VbSbpmsrr5ZJ2ajez7UdJ+pCkp9l+Zw4XAFOgwxcm8gfAyJA9AEogewCUUreaD4VioKAuZ7f7WmWLcdFu5oj4o6TXD7MBAMbfCLJHIn8AdEH2ACiB7AFQSh1rPhSKR+Scc87pOH2fffbpuo577713oDacf/75XeeZP3/+QNvA4Pq9DaGN5ZK2qLzeXNJ1AzQPNdOt/6PDDz+86zo+//nPD6s5GGOD3ILZBvmDgXTLrzPPPLPj9F4+Ww3Drrvu2nH64sWLu65jwYIFHafPmzdvJk2qFbJnulx88cUdp2+44YYdp/dyLCxZsqTj9EMO6f4Mn6OOOqrj9De+8Y0dp994441dt4GyyB6Mo1122aXj9Ii29T/USN1qPhSKgcKGfBvCRZK2tv0YSddKepWk/QdoHoAJNYJbMMkfAF2RPQBKIHsAlFK3ms/Q778A0LvGbQjNQ4/LLpb0E0nb2l5u+5CIuE/SEZK+L+kySWdExKUjewMAaqld9pA/AEaJ7AFQAtkDoJQ61ny4ohgobIAnYLa8bzUizpHUue8TAFNvwKd/kz8A+kL2ACiB7AFQSt1qPhSKgYI6dGze720IANBVl4cqkD8ARoLsAVAC2QOglDrWfCgUA4WN4Am8ANAV2QOgBLIHQAlkD4BS6pY/FIqBwob8BEwA6MkInv4NAF2RPQBKIHsAlFK3mg+FYqAg27LdatLY3oYAoP46ZI9E/gAYEbIHQAlkD4BS6ljzoVA8Itttt13H6UcfffTI2zB//vyRbwODq9ttCBhvEVG6CagJsgfjplt+LVu2rOP0iy++uOs2dthhh5k0qaV58+Z1nP7Od75z4G1MMrJnugzjmOum23ee3//+9yNvA8Yf2YNxNOh3t9n67IPB1C1/KBQDhdUtNABMBrIHQAlkD4ASyB4ApdQtfygUAwV1eALm2PZXA6D+ujz9m/wBMBJkD4ASyB4ApdSx5kOhGCisTWiMbX81ACZDhy9M5A+AkSF7AJRA9gAopW41HwrFQEFdzm4DwEiQPQBKIHsAlED2ACiljvlDoRgorG6hAWAykD0ASiB7AJRA9gAopW75Q6EYKKxu/dUAmAz01QegBLIHQAlkD4BS6lbzoVAMFNThNoSx7a8GQP11uQWK/AEwEmQPgBLIHgCl1LHmQ6EYKKxutyEAmAxkD4ASyB4AJZA9AEqpW/5QKB6RefPmdZx+7LHHdl3HiSee2HH6hRde2HH6qaee2nUbDzzwQNd5MFq2SzcBwBQie1A3v/71rztOv+222wbexg477DDwOtAZ2YM66vZ3GxGz1BL0i+zBJOJzSz3ULX8oFAMF1fEJmADqj+wBUALZA6AEsgdAKXXMHwrFQGF169gcwGTgoS4ASiB7AJRA9gAopW41HwrFQEF17NgcQP3xUBcAJZA9AEogewCUUseaD4VioLC63YYAYDKQPQBKIHsAlED2ACilbvlDoRgorG6hAWAykD0ASiB7AJRA9gAopW75Q6EYKKiOHZsDqD+yB0AJZA+AEsgeAKXUMX8oFAOF1S00AEwGsgdACWQPgBLIHgCl1C1/KBSPsde97nUDTT/llFOG2RyMQB3PLgGoP7IHdXTFFVd0nP74xz++6zp22GGHYTUHfSB7UFeHHXZYx+nLli3ruo558+YNpzGYMbIHQCl1zJ+urbV9ku0Vti+pjNvQ9nm2f5f/3WC0zQQm15w5c1YZJM21vbQydP50OqHIH2B0WmUP+ZOQPcDokD3tkT3A6JA97ZE9wGjVLXt6KWsvkrR707ijJf0gIraW9IP8GsAMNc4utQiNmyNix8qwsHRbC1kk8gcYunbZQ/48aJHIHmDoyJ6uFonsAYaO7OlqkcgeYCTqWPPpWiiOiAsk3dI0em9JJ+efT5b00iG3C5gatlcZkJA/wOi0yh7yJyF7gNEhe9oje4DRIXvaI3uA0apb9vTbR/EmEXG9JEXE9bY3HmKbgKlSt/5qxgD5AwwB2TNjZA8wBGTPjJE9wBCQPTNG9gBDUrf8GfnD7HI/G4dJ0pZbbjnqzQG1UseOzeuC7AHaI3tGh+wB2iN7Rov8AVoje0aL7AHaq2P+9NvaG20/WpLyvyvazRgRCxt9bmy00UZ9bg6YXG36q0F7PeUP2QN01qGvPrRG9gBDQPbMGN+7gCEge2aM7AGGpG7Z02/rzpZ0UP75IEnfGk5zgOlTt9AYA+QPMAR8YZoxsgcYArJnxsgeYAjInhkje4AhqVv2dO16wvZiSfMlzbW9XNL7JH1U0hm2D5F0taRXjLKRwKSq420Is4n8AUaD7OmM7AFGg+zpjOwBRoPs6YzsAUanjvnTtVAcEQvaTHrBkNsCTKW6hcZsIn+A0SF72iN7xtOb3vSm0k3AEJA97ZE94+sLX/hC6SZgQGRPe2QPMFp1y5+RP8wOQHt1PLsEoP7IHgAlkD0ASiB7AJRSx/yhUAwU1iY05tpeWnm9MCIWzlKTAEyBDh9YyB8AI0P2ACiB7AFQSt1qPhSKgcLahMbNEbHjbLcFwPTo8IWJ/AEwMmQPgBLIHgCl1K3mQ6EYKMi2bJduBoApQ/YAKIHsAVAC2QOglDrmD4VioLC69VcDYDKQPQBKIHsAlED2ACilbvlDoRgorG6hAWAykD0ASiB7AJRA9gAopW75Q6EYKKiOT8AEUH9kD4ASyB4AJZA9AEqpY/5QKAYKq1toAJgMZA+AEsgeACWQPQBKqVv+UCgGCqrj2SUA9Uf2ACiB7AFQAtkDoJQ65g+FYqCwuoUGgMlA9gAogewBUALZA6CUuuUPhWKgsLqFBoDJQPYAKIHsAVAC2QOglLrlD4VioKA63oYAoP7IHgAlkD0ASiB7AJRSx/yhUAwUVrfQADAZyB4AJZA9AEogewCUUrf8oVAMFFTHs0sA6o/sAVAC2QOgBLIHQCl1zB8KxUBhtks3AcAUInsAlED2ACiB7AFQSt3yh0IxUFjJs0u2XyrpxZI2lnRCRJxbrDEAZlXpM9vkDzCdyB4AJZA9AEqpW82nXtc/AxOmcRtC89DjsifZXmH7kqbxu9u+3PYVto/utI6I+GZEvE7SwZJe2e/7AFAv7bKH/AEwSmQPgBLIHgCl1LHmwxXFQGEDnF1aJOl4Sac0RtheTdIJknaVtFzSRbbPlrSapI80Lf/aiFiRf35PXg7AlBjwzPYikT8A+kD2ACiB7AFQSt1qPhSKgcLahMZc20srrxdGxMLqDBFxge15Tcs9Q9IVEXGlJNk+XdLeEfERSXs1b8Sps5yPSvpuRFzc95sAUDsdPrCQPwBGhuwBUALZA6CUutV8KBQDBXV4AubNEbFjH6vcTNI1ldfLJe3UYf5/lvRCSY+0/fiI+Hwf2wRQM12evkv+ABgJsgdACWQPgFLqWPOhUAwUNuSOzVs9TjPazRwRn5H0mWE2AEA9jOChCuQPgK7IHgAlkD0ASqlbzYdCMVBQh7NLXW9DaGO5pC0qrzeXdN0ATQQwgbpcWUP+ABgJsgdACWQPgFLqWPOhUAwUNuTbEC6StLXtx0i6VtKrJO0/QPMATKgR3IJJ/gDoiuwBUALZA6CUutV8hn7/BYCZmTNnzipDL2wvlvQTSdvaXm77kIi4T9IRkr4v6TJJZ0TEpSNrPIDaapU95A+AUSN7AJRA9gAopW7ZwxXFQEG2lR5CuYpenoC5oNWCEXGOpHOG10oAk6ZD9kjkD4ARIXsAlED2ACiljjUfCsVAYUO+DQEAejKCWzABoCuyB0AJZA+AUupW86FQDBQ2gifwAkBXZA+AEsgeACWQPQBKqVv+UCgGChrBEzABoKsRPf0bADoiewCUQPYAKKWONR8KxUBhdbsNAcBk4BZMACWQPQBKIHsAlFK3mg+FYqCgLme3AWAkyB4AJZA9AEogewCUUsf8oVAMFFa32xAATAZuwQRQAtkDoASyB0Apdav5UCgGCqvbbQgAJgO3YAIogewBUALZA6CUutV8KBQDBdXxNgQA9Uf2ACiB7AFQAtkDoJQ65k/X1to+yfYK25dUxh1r+1rbP8/DnqNtJjC55syZs8qAhPwBRqdV9pA/CdkDjA7Z0x7ZA4wO2dMe2QOMVt2yp5fWLZK0e4vxn4qI7fNwznCbBUyHxtmlOoXGLFsk8gcYunbZQ/48aJHIHmDoyJ6uFonsAYaO7OlqkcgeYCTqWPPp2vVERFxge97omwJMJ9utRo9tx+azifwBRqdN9kjkD9kDjBDZ0x7ZA4wO2dMe2QOMVt1qPoP0UXyE7VdLWirpqIj405DaBEyVunVsPibIH2BAPNSlL2QPMCCypy9kDzAgsqcvZA8wBHWr+fR7vfPnJD1O0vaSrpf0iXYz2j7M9lLbS2+66aY+NwdMpjrehjAGesofsgdoj1sw+0L2AAMie/rC9y5gQGRPX8geYAjqWPPpq3URcWNE3B8RD0g6UdIzOsy7MCJ2jIgdN9poo37bCUysuoVGab3mD9kDdMYXppkhe4DhIHtmhu9dwHCQPTND9gDDU7fs6at1th9defkySZe0mxdAe3U8u1Qa+QMMjitrZo7sAQZH9swc2QMMjuyZObIHGI461ny69lFse7Gk+UodLS+X9D5J821vLykkLZN0+AjbCEy0NiExth2bzybyBxidDh9Qpj5/yB5gdMie9sgeYHTInvbIHmC06lbzcUTM3sbsmyRdVRk1V9LNs9aA/tHO4Zrkdm4VET3fb2P7e3k7zW6OiN1nuG20QfaMHO0crpLZI5E/Q9Mie6TJ/jssgXYO10zbSfaMKT77jBztHC6yZ0KQPSNHO4eLmk8Ls1ooXmXj9tJxfcpfFe0cLtqJ0uryu6Wdw0U7MQ7q8vulncNFO1FaXX63tHO4aCdKq8vvlnYOF+2st/HuGAMAAAAAAAAAMHIUigEAAAAAAABgypUuFI9FR809oJ3DRTtRWl1+t7RzuGgnxkFdfr+0c7hoJ0qry++Wdg4X7URpdfnd0s7hop01VrSPYgAAAAAAAABAeaWvKAYAAAAAAAAAFEahGAAAAAAAAACmXLFCse3dbV9u+wrbR5dqRze2l9n+le2f215auj0Ntk+yvcL2JZVxG9o+z/bv8r8blGxjblOrdh5r+9q8T39ue8+Sbcxt2sL2+bYvs32p7Tfn8WO3TzEYsmcwZM9wkT3Tg+wZDNkzXGTP9CB7Bkf+DLWNZM8UIX8GQ/YMtY1kzwwUKRTbXk3SCZL2kLSdpAW2tyvRlh7tEhHbR8SOpRtSsUjS7k3jjpb0g4jYWtIP8uvSFmnVdkrSp/I+3T4izpnlNrVyn6SjIuKJkp4p6Z/y3+Q47lP0iewZikUie4aJ7JkCZM9QLBLZM0xkzxQge4ZmkcifYSF7pgT5MxSLRPYMC9kzA6WuKH6GpCsi4sqI+Iuk0yXtXagttRQRF0i6pWn03pJOzj+fLOmls9qoFtq0c+xExPURcXH++Q5Jl0naTGO4TzEQsmdAZM9wkT1Tg+wZENkzXGTP1CB7hoD8GR6yZ6qQPwMie4aH7JmZUoXizSRdU3m9PI8bRyHpXNs/s31Y6cZ0sUlEXC+lA0HSxoXb08kRtn+Zb1MYq8v7bc+T9DRJ/6t67VN0R/aMRp2OE7IHJZA9o1Gn44TsQQlkz+jU6VgZy/wheyYe+TMadTpWyJ6aKlUodotxMeut6M1zImIHpVsm/sn280o3aAJ8TtLjJG0v6XpJnyjbnIfYXlfSmZKOjIjbS7cHQ0f2TDeyB6WQPdON7EEpZA/GMn/InqlA/kw3sqfGShWKl0vaovJ6c0nXFWpLRxFxXf53haSzlG6hGFc32n60JOV/VxRuT0sRcWNE3B8RD0g6UWOyT22voRQaX4mIb+TRtdin6BnZMxq1OE7IHhRE9oxGLY4TsgcFkT2jU4tjZRzzh+yZGuTPaNTiWCF76q1UofgiSVvbfoztNSW9StLZhdrSlu2H216v8bOk3SRd0nmpos6WdFD++SBJ3yrYlrYaB2L2Mo3BPrVtSV+UdFlEfLIyqRb7FD0je0ajFscJ2YOCyJ7RqMVxQvagILJndGpxrIxb/pA9U4X8GY1aHCtkT705oszV/7b3lPRvklaTdFJEfKhIQzqw/VilM0qStLqk08alnbYXS5ovaa6kGyW9T9I3JZ0haUtJV0t6RUQU7VS8TTvnK92CEJKWSTq80S9MKbafK+lCSb+S9EAe/S6lfmvGap9iMGTPYMie4SJ7pgfZMxiyZ7jInulB9gyO/Bkesme6kD+DIXuGh+yZmWKFYgAAAAAAAADAeCjV9QQAAAAAAAAAYExQKAYAAAAAAACAKUehGAAAAAAAAACmHIViAAAAAAAAAJhyFIoBAAAAAAAAYMpRKAYAAAAAAACAKUehGAAAAAAAAACm3P8Hbw1EDT5nxIQAAAAASUVORK5CYII=\n", | |
"text/plain": [ | |
"<Figure size 1440x180 with 10 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"logscale = True\n", | |
"logscale = dict(norm=mpl.colors.LogNorm()) if logscale else {}\n", | |
"\n", | |
"fig, axes = plt.subplots(1,5,figsize=(20,2.5))\n", | |
"for i in range(2):\n", | |
" im = axes[i].imshow(x_test[predictions_cnn.argmin(axis=0)[i],:,:,0], vmax=1, cmap=cmap, **logscale)\n", | |
" plt.colorbar(im, ax=axes[i])\n", | |
" axes[i].set(title='most {}-like jet'.format(['top','qcd'][i]))\n", | |
"axes[2].imshow(x_test[abs(predictions_cnn-0.5).argmin(axis=0)[0],:,:,0], vmax=1, cmap=cmap, **logscale)\n", | |
"plt.colorbar(im, ax=axes[2])\n", | |
"axes[2].set(title='most uncertain jet');\n", | |
"\n", | |
"for iax, i in enumerate((predictions_cnn - y_test).argmin(axis=0)):\n", | |
" im = axes[iax+3].imshow(x_test[i,:,:,0], vmax=1, cmap=cmap, **logscale)\n", | |
" plt.colorbar(im, ax=axes[iax+3])\n", | |
" axes[iax+3].set_title('Output: {}\\nLabel = {} --FAIL--'.format(predictions_cnn[i], y_test[i]))\n", | |
"\n", | |
"fig.tight_layout()\n", | |
"fig.savefig('cnn_jet_sample.png', dpi=150)" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"What does the model think of an empty image?" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 34, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"text/plain": [ | |
"[array([[0.7229311 , 0.27706897]], dtype=float32),\n", | |
" array([[0.6891926 , 0.31080747]], dtype=float32),\n", | |
" array([[0.9867889, 0.013211 ]], dtype=float32)]" | |
] | |
}, | |
"execution_count": 34, | |
"metadata": {}, | |
"output_type": "execute_result" | |
} | |
], | |
"source": [ | |
"empty = np.zeros(x_test[0].shape)\n", | |
"[model.predict(np.array([empty])) for model in [model0, model1, model_cnn]]" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"Not great! Looks a lot like a QCD jet! Because the only difference from a sharp QCD jet is in one or a few pixels! Bad CNN! \n", | |
"\n", | |
"In real world applications, one would need to make sure they are passing images with at least one hot pixel. For example, see the image brightness:" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 55, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAD8CAYAAAB+UHOxAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAFmdJREFUeJzt3XuQVeWZ7/HvY0NkdHRQRGPRJA0ZTo7RijcKQS2iYbxOKhhLR6lzJmhZ6VMJTiXnRB00VqiJycRJyiGxMlrFKHNwEgMcLyWVMnGIYgyVUS6KYVQc8Rb7gNDDxctYJhGf+aNfTANN927Y3XvT6/up6tprvftdaz+LovZvr3e9e+3ITCRJ1XNQowuQJDWGASBJFWUASFJFGQCSVFEGgCRVlAEgSRVlAEhSRRkAklRRBoAkVdSwRhfQm6OOOirb2toaXYYkHVBWr179H5k5uq9+TR0AbW1trFq1qtFlSNIBJSJeraWfQ0CSVFEGgCRVlAEgSRXV1NcAJGlf/f73v6ejo4N333230aUMmBEjRtDa2srw4cP3aXsDQNKQ1NHRwWGHHUZbWxsR0ehy6i4z2bJlCx0dHYwbN26f9uEQkKQh6d1332XUqFFD8s0fICIYNWrUfp3hGACShqyh+ua/0/4enwEgSRXlNQBJlTB36b/XdX//+5z/1mefjo4OZs2axbPPPsuOHTu48MILueWWWzj44INZsWIF11xzDZs2bSIiOPPMM7n11ltZvHgx1157La2trbz99tuMHz+eOXPmcPrpp9e1fjAAGqoR/yElDY7M5OKLL+aLX/wiDzzwADt27KC9vZ3rrruOG264gUsvvZSFCxcyZcoUMpN7772Xt956C4DLLruMH/zgBwAsW7aMiy++mGXLlnHcccfVtUYDQJIGwCOPPMKIESO48sorAWhpaWHu3Ll89KMfpaWlhZkzZzJlyhSgayz/kksu6XE/Z599Nu3t7cybN4+5c+fWtUavAUjSAHjmmWc49dRTd2k7/PDDaWtrY82aNXs815tTTjmFdevW1btEA0CSBkJm9jhLJzP3aV8DwQCQpAFw/PHH73E34zfffJNNmzZx6qmnsnr16pr39dRTT9V9/B8MAEkaENOmTeOdd97hrrvuAmDHjh189atf5eqrr+aaa65hwYIFPPHEEx/0/+EPf8jrr7++x35+8YtfMG/ePL7whS/UvUYvAkuqhMGeJRcR3H///cyaNYubbrqJzs5OLrvsMr72ta8BsHDhQq655ho2b97MQQcdxNSpU7n44osBWLRoEcuXL+edd95h3Lhx3HvvvQNyBmAASNIAGTt2LEuWLAHgV7/6FTNmzGD16tWceuqpTJkyhV/+8pd7bHPFFVdwxRVXDEp9NQ0BRcTIiLgnItZFxHMRMSUijoyIpRHxQnk8ovSNiLg1ItZHxK8j4pRu+5lZ+r8QETMH6qAkqdmcfvrpvPrqq/2a/TPQar0G8H3gZ5n534ETgeeA2cDDmTkBeLisA1wATCh/7cDtABFxJDAHOA2YBMzZGRqSpMHXZwBExOHAVOBOgMz8XWZuB6YDC0q3BcBFZXk6cFd2eRwYGRHHAucBSzNza2ZuA5YC59f1aCRJNavlDGA80An8U0Q8FRF3RMShwDGZuRGgPB5d+o8BXuu2fUdp21v7LiKiPSJWRcSqzs7Ofh+QJKk2tQTAMOAU4PbMPBn4T/4w3NOTnu5Pmr2079qQOS8zJ2bmxNGjR9dQniRpX9QSAB1AR2bunLB6D12BsKkM7VAeN3frP7bb9q3Ahl7aJUkN0Oc00Mx8PSJei4iPZ+bzwDTg2fI3E7i5PD5QNlkCXB0RC+m64PtGZm6MiIeAv+124fdc4Pr6Ho4k7cWyb9d3f2f3/va1fft27r77br70pS/V93XrqNZZQH8F/Cgifg2cBPwtXW/850TEC8A5ZR3gQeAlYD3wj8CXADJzK3ATsLL8faO0SdKQs337dm677bZGl9Grmr4IlplrgIk9PDWth74JzNrLfuYD8/tToCQdiGbPns2LL77ISSedxDnnnAPAT3/6UyKCG2+8kcsuu4xHH32Ur3/964waNYrnn3+eqVOnctttt3HQQYNzlx7vBSRJA+Dmm2/mYx/7GGvWrGHy5MmsWbOGp59+mp///Odce+21bNy4EYAVK1Zwyy23sHbtWl588UXuu+++QavRAJCkAbZ8+XJmzJhBS0sLxxxzDJ/61KdYuXIlAJMmTWL8+PG0tLQwY8YMli9fPmh1GQCSNMB6u5//7r8Z0NNvCAwUA0CSBsBhhx32wW/8Tp06lUWLFrFjxw46Ozt57LHHmDRpEtA1BPTyyy/z/vvvs2jRIs4888xBq9G7gUqqhj6mbdbbqFGjOOOMMzjhhBO44IIL+OQnP8mJJ55IRPCd73yHD3/4w6xbt44pU6Ywe/Zs1q5dy9SpU/nc5z43aDUaAJI0QO6+++5d1r/73e/u0eeQQw5h0aJFg1XSLhwCkqSK8gxAkhrkrLPO4qyzzmrY63sGIGnI6m32zVCwv8dnAEgakkaMGMGWLVuGbAhkJlu2bGHEiBH7vA+HgCQNSa2trXR0dDCUf1dkxIgRtLa27vP2BoCkIWn48OGMGzeu0WU0NYeAJKmiDABJqigDQJIqygCQpIoyACSpogwASaooA0CSKsoAkKSKMgAkqaIMAEmqKANAkiqqpgCIiFciYm1ErImIVaXtyIhYGhEvlMcjSntExK0RsT4ifh0Rp3Tbz8zS/4WImDkwhyRJqkV/zgDOzsyTMnNiWZ8NPJyZE4CHyzrABcCE8tcO3A5dgQHMAU4DJgFzdoaGJGnw7c8Q0HRgQVleAFzUrf2u7PI4MDIijgXOA5Zm5tbM3AYsBc7fj9eXJO2HWgMggX+JiNUR0V7ajsnMjQDl8ejSPgZ4rdu2HaVtb+2SpAao9fcAzsjMDRFxNLA0Itb10jd6aMte2nfduCtg2gE+8pGP1FieJKm/ajoDyMwN5XEzcD9dY/ibytAO5XFz6d4BjO22eSuwoZf23V9rXmZOzMyJo0eP7t/RSJJq1mcARMShEXHYzmXgXODfgCXAzpk8M4EHyvIS4PNlNtBk4I0yRPQQcG5EHFEu/p5b2iRJDVDLENAxwP0RsbP/3Zn5s4hYCSyOiKuA3wCXlv4PAhcC64F3gCsBMnNrRNwErCz9vpGZW+t2JJKkfukzADLzJeDEHtq3ANN6aE9g1l72NR+Y3/8yJUn15jeBJamiDABJqigDQJIqygCQpIoyACSpogwASaooA0CSKsoAkKSKMgAkqaIMAEmqKANAkirKAJCkijIAJKmiDABJqigDQJIqygCQpIoyACSpogwASaooA0CSKsoAkKSKMgAkqaIMAEmqKANAkirKAJCkiqo5ACKiJSKeioiflPVxEfFERLwQEYsi4kOl/eCyvr4839ZtH9eX9ucj4rx6H4wkqXb9OQP4MvBct/W/A+Zm5gRgG3BVab8K2JaZfwrMLf2IiE8AlwPHA+cDt0VEy/6VL0naVzUFQES0An8O3FHWA/g0cE/psgC4qCxPL+uU56eV/tOBhZn528x8GVgPTKrHQUiS+q/WM4DvAdcB75f1UcD2zHyvrHcAY8ryGOA1gPL8G6X/B+09bCNJGmR9BkBEfAbYnJmruzf30DX7eK63bbq/XntErIqIVZ2dnX2VJ0naR7WcAZwBfDYiXgEW0jX08z1gZEQMK31agQ1luQMYC1Ce/xNga/f2Hrb5QGbOy8yJmTlx9OjR/T4gSVJt+gyAzLw+M1szs42ui7iPZOb/AJYBl5RuM4EHyvKSsk55/pHMzNJ+eZklNA6YAKyo25FIkvplWN9d9uqvgYUR8U3gKeDO0n4n8M8RsZ6uT/6XA2TmMxGxGHgWeA+YlZk79uP1JUn7oV8BkJmPAo+W5ZfoYRZPZr4LXLqX7b8FfKu/RUqS6s9vAktSRRkAklRRBoAkVZQBIEkVZQBIUkUZAJJUUQaAJFWUASBJFWUASFJFGQCSVFEGgCRVlAEgSRVlAEhSRRkAklRRBoAkVZQBIEkVZQBIUkUZAJJUUQaAJFWUASBJFWUASFJFGQCSVFEGgCRVVJ8BEBEjImJFRDwdEc9ExN+U9nER8UREvBARiyLiQ6X94LK+vjzf1m1f15f25yPivIE6KElS32o5A/gt8OnMPBE4CTg/IiYDfwfMzcwJwDbgqtL/KmBbZv4pMLf0IyI+AVwOHA+cD9wWES31PBhJUu36DIDs8nZZHV7+Evg0cE9pXwBcVJanl3XK89MiIkr7wsz8bWa+DKwHJtXlKCRJ/VbTNYCIaImINcBmYCnwIrA9M98rXTqAMWV5DPAaQHn+DWBU9/YetpEkDbKaAiAzd2TmSUArXZ/aj+upW3mMvTy3t/ZdRER7RKyKiFWdnZ21lCdJ2gfD+tM5M7dHxKPAZGBkRAwrn/JbgQ2lWwcwFuiIiGHAnwBbu7Xv1H2b7q8xD5gHMHHixD0CQr1Y9u367/Ps6+u/T0lNoZZZQKMjYmRZ/iPgz4DngGXAJaXbTOCBsrykrFOefyQzs7RfXmYJjQMmACvqdSCSpP6p5QzgWGBBmbFzELA4M38SEc8CCyPim8BTwJ2l/53AP0fEero++V8OkJnPRMRi4FngPWBWZu6o7+FIkmrVZwBk5q+Bk3tof4keZvFk5rvApXvZ17eAb/W/TElSvflNYEmqKANAkiqqX7OAVEH1nlnkrCKpaXgGIEkVZQBIUkUZAJJUUQaAJFWUASBJFWUASFJFGQCSVFEGgCRVlAEgSRVlAEhSRRkAklRRBoAkVZQBIEkVZQBIUkUZAJJUUQaAJFWUASBJFWUASFJFGQCSVFEGgCRVlAEgSRU1rK8OETEWuAv4MPA+MC8zvx8RRwKLgDbgFeAvMnNbRATwfeBC4B3gisx8suxrJnBj2fU3M3NBfQ+n2v71pS113+eU8aPqvk9JzaGWM4D3gK9m5nHAZGBWRHwCmA08nJkTgIfLOsAFwITy1w7cDlACYw5wGjAJmBMRR9TxWCRJ/dBnAGTmxp2f4DPzLeA5YAwwHdj5CX4BcFFZng7clV0eB0ZGxLHAecDSzNyamduApcD5dT0aSVLN+nUNICLagJOBJ4BjMnMjdIUEcHTpNgZ4rdtmHaVtb+2SpAaoOQAi4o+Be4GvZOabvXXtoS17ad/9ddojYlVErOrs7Ky1PElSP9UUABExnK43/x9l5n2leVMZ2qE8bi7tHcDYbpu3Aht6ad9FZs7LzImZOXH06NH9ORZJUj/0GQBlVs+dwHOZ+ffdnloCzCzLM4EHurV/PrpMBt4oQ0QPAedGxBHl4u+5pU2S1AB9TgMFzgD+ElgbEWtK2w3AzcDiiLgK+A1waXnuQbqmgK6naxrolQCZuTUibgJWln7fyMytdTkKHTiWfbu++zv7+vruT6qQPgMgM5fT8/g9wLQe+icway/7mg/M70+BkqSB4TeBJamiDABJqigDQJIqygCQpIoyACSpogwASaooA0CSKsoAkKSKMgAkqaIMAEmqKANAkirKAJCkijIAJKmiDABJqigDQJIqygCQpIoyACSpogwASaqoWn4TWANk8m/mNboESRXmGYAkVZRnAOrVv760pa77mzJ+VF33J2nfeQYgSRVlAEhSRRkAklRRfV4DiIj5wGeAzZl5Qmk7ElgEtAGvAH+RmdsiIoDvAxcC7wBXZOaTZZuZwI1lt9/MzAX1PRRV0rJv13+fZ19f/31KTaiWM4D/C5y/W9ts4OHMnAA8XNYBLgAmlL924Hb4IDDmAKcBk4A5EXHE/hYvSdp3fQZAZj4GbN2teTqw8xP8AuCibu13ZZfHgZERcSxwHrA0M7dm5jZgKXuGiiRpEO3rNYBjMnMjQHk8urSPAV7r1q+jtO2tfQ8R0R4RqyJiVWdn5z6WJ0nqS70vAkcPbdlL+56NmfMyc2JmThw9enRdi5Mk/cG+BsCmMrRDedxc2juAsd36tQIbemmXJDXIvgbAEmBmWZ4JPNCt/fPRZTLwRhkiegg4NyKOKBd/zy1tkqQGqWUa6I+Bs4CjIqKDrtk8NwOLI+Iq4DfApaX7g3RNAV1P1zTQKwEyc2tE3ASsLP2+kZm7X1iWJA2iPgMgM2fs5alpPfRNYNZe9jMfmN+v6iRJA8ZvAktSRRkAklRRBoAkVZQBIEkVZQBIUkUZAJJUUQaAJFWUvwks7a7evzHg7wuoSRkAGlT+yLzUPBwCkqSKMgAkqaIMAEmqKANAkirKAJCkijIAJKmiDABJqigDQJIqyi+C6YBW7y+WwQB8ucxvFqtJeQYgSRVlAEhSRRkAklRRBoAkVZQXgaXdNP0dS72orDoZ9DOAiDg/Ip6PiPURMXuwX1+S1GVQzwAiogX4B+AcoANYGRFLMvPZwaxDGkxNf0ahyhrsIaBJwPrMfAkgIhYC0wEDQKpR3QOFOg8pgcNKB4jBDoAxwGvd1juA0wa5BkndDMiX6QYiVOrJgAIGPwCih7bcpUNEO9BeVt+OiOf34/WOAv5jP7YfaM1eHzR/jc1eHzR/jc1eH9S9xhvqt6s/aKZ/x4/W0mmwA6ADGNttvRXY0L1DZs4D5tXjxSJiVWZOrMe+BkKz1wfNX2Oz1wfNX2Oz1wfWOFAGexbQSmBCRIyLiA8BlwNLBrkGSRKDfAaQme9FxNXAQ0ALMD8znxnMGiRJXQb9i2CZ+SDw4CC9XF2GkgZQs9cHzV9js9cHzV9js9cH1jggIjP77iVJGnK8F5AkVdSQDIBmv91ERMyPiM0R8W+NrqUnETE2IpZFxHMR8UxEfLnRNe0uIkZExIqIeLrU+DeNrqknEdESEU9FxE8aXUtPIuKViFgbEWsiYlWj6+lJRIyMiHsiYl35Pzml0TXtFBEfL/92O//ejIivNLquWg25IaByu4l/p9vtJoAZzXS7iYiYCrwN3JWZJzS6nt1FxLHAsZn5ZEQcBqwGLmqyf8MADs3MtyNiOLAc+HJmPt7g0nYREf8HmAgcnpmfaXQ9u4uIV4CJmdks89f3EBELgF9m5h1l9uAhmbm90XXtrrz3/H/gtMx8tdH11GIongF8cLuJzPwdsPN2E00jMx8Dtja6jr3JzI2Z+WRZfgt4jq5vcTeN7PJ2WR1e/prq00xEtAJ/DtzR6FoOVBFxODAVuBMgM3/XjG/+xTTgxQPlzR+GZgD0dLuJpnrzOpBERBtwMvBEYyvZUxleWQNsBpZmZrPV+D3gOuD9RhfSiwT+JSJWl2/hN5vxQCfwT2Uo7Y6IOLTRRe3F5cCPG11EfwzFAOjzdhOqTUT8MXAv8JXMfLPR9ewuM3dk5kl0faN8UkQ0zXBaRHwG2JyZqxtdSx/OyMxTgAuAWWV4spkMA04Bbs/Mk4H/BJrxut6HgM8C/6/RtfTHUAyAPm83ob6VcfV7gR9l5n2Nrqc3ZUjgUeD8BpfS3RnAZ8sY+0Lg0xHxw8aWtKfM3FAeNwP30zWE2kw6gI5uZ3f30BUIzeYC4MnM3NToQvpjKAaAt5vYT+UC653Ac5n5942upycRMToiRpblPwL+DFjX2Kr+IDOvz8zWzGyj6//gI5n5Pxtc1i4i4tBykZ8yrHIu0FQz0zLzdeC1iPh4aZpGc94+fgYH2PAPDMGfhDwQbjcRET8GzgKOiogOYE5m3tnYqnZxBvCXwNoyxg5wQ/kWd7M4FlhQZl4cBCzOzKacatnEjgHu78p7hgF3Z+bPGltSj/4K+FH5QPcScGWD69lFRBxC16zD/9XoWvpryE0DlSTVZigOAUmSamAASFJFGQCSVFEGgCRVlAEgSRVlAEhSRRkAklRRBoAkVdR/AfSinFET9ErNAAAAAElFTkSuQmCC\n", | |
"text/plain": [ | |
"<Figure size 432x288 with 1 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"plt.hist(map(np.sum, [ x for x,y in zip(x_test, y_test)[:] if np.argmax(y)==0] ), bins=np.arange(0,8,0.5), alpha=0.5, label='QCD')\n", | |
"plt.hist(map(np.sum, [ x for x,y in zip(x_test, y_test)[:] if np.argmax(y)==1] ), bins=np.arange(0,8,0.5), alpha=0.5, label='top')\n", | |
"plt.legend();" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"### Check effect of pixel normalization\n", | |
"\n", | |
"Trained models without normalizing the data. Any difference? Reload data but skip normalization step" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 299, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"name": "stdout", | |
"output_type": "stream", | |
"text": [ | |
"We have 35379 QCD jets and 36708 top jets\n", | |
"((72087, 16, 22), (72087,))\n", | |
"((72087, 16, 22, 1), (72087, 2))\n", | |
"We will train+validate on 50000 images, leaving 22087 for cross-validation\n" | |
] | |
} | |
], | |
"source": [ | |
"data0 = np.load(outdir+'qcd_leading_jet.npz')['arr_0']\n", | |
"data1 = np.load(outdir+'tt_leading_jet.npz')['arr_0']\n", | |
"\n", | |
"print('We have {} QCD jets and {} top jets'.format(len(data0), len(data1)))\n", | |
"\n", | |
"x_data = np.concatenate((data0, data1))\n", | |
"# pad images\n", | |
"x_data = map(pad_image, x_data)\n", | |
"# x_data = map(normalize, x_data)\n", | |
"\n", | |
"y_data = np.array([0]*len(data0)+[1]*len(data1))\n", | |
"\n", | |
"np.random.seed(0) # for reproducibility\n", | |
"x_data, y_data = np.random.permutation(np.array([x_data, y_data]).T).T\n", | |
"\n", | |
"# the data coming out of previous commands is a list of 2D arrays. We want a 3D np array (n_events, xpixels, ypixels)\n", | |
"x_data = np.stack(x_data)\n", | |
"\n", | |
"print(x_data.shape, y_data.shape)\n", | |
"\n", | |
"# reshape for tensorflow: x_data.shape + (1,) = shortcut for (x_data.shape[0], 16, 22, 1)\n", | |
"x_data = x_data.reshape(x_data.shape + (1,)).astype('float32')/255.\n", | |
"\n", | |
"y_data = keras.utils.to_categorical(y_data, 2)\n", | |
"\n", | |
"print(x_data.shape, y_data.shape)\n", | |
"\n", | |
"n_train = 50000\n", | |
"(x_train_no, x_test_no) = x_data[:n_train], x_data[n_train:]\n", | |
"(y_train_no, y_test_no) = y_data[:n_train], y_data[n_train:]\n", | |
"\n", | |
"print('We will train+validate on {0} images, leaving {1} for cross-validation'.format(n_train,len(x_data)-n_train))" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"Training happened via Colab... load saved CNN model" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 300, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"model_dir='trained_models_low_pt/'\n", | |
"model_cnn_no = keras.models.load_model(model_dir+'cnn_no_norm.h5')" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 301, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"predictions_cnn_no = model_cnn_no.predict(x_test_no)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 302, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "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\n", | |
"text/plain": [ | |
"<Figure size 432x288 with 1 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"fpr_cnn_no, tpr_cnn_no, thresholds = roc_curve(y_test_no.ravel(), predictions_cnn_no.ravel())\n", | |
"\n", | |
"np.seterr(divide='ignore', invalid='ignore') # disable warning for 1/0 divisions\n", | |
"plt.plot(tpr_cnn, 1/fpr_cnn, label='Convolutional NN - discretized pixels')\n", | |
"plt.plot(tpr_cnn_no, 1/fpr_cnn_no, label='Convolutional NN - undiscretized')\n", | |
"plt.gca().set(ylabel='1/$\\epsilon_B$ - 1/False positive rate', xlabel='$\\epsilon_S$ - True positive rate', title='Effect of discretizing+normalizing pixels', xlim=(-0.01,1.01), ylim=(1,5*10**3), yscale='log')\n", | |
"plt.grid(True, which=\"both\")\n", | |
"plt.legend(loc='upper right');" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"### Compare to high pT jets\n", | |
"\n", | |
"We have also generated jets at higher momentum, in the 800-900 GeV range. In this case the jets will are more boosted, so that all the top decay products typically fall within the $\\Delta R=1$ jet cone. We can train the models on these images and re-train the models, and compare." | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 313, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"name": "stdout", | |
"output_type": "stream", | |
"text": [ | |
"We have 24395 QCD jets and 24561 top jets\n", | |
"((48956, 16, 22), (48956,))\n", | |
"((48956, 16, 22, 1), (48956, 2))\n", | |
"We will train+validate on 40000 images, leaving 8956 for cross-validation\n" | |
] | |
} | |
], | |
"source": [ | |
"import keras\n", | |
"data0 = np.load(outdir+'qcd_high_leading_jet.npz')['arr_0']\n", | |
"data1 = np.load(outdir+'tt_high_leading_jet.npz')['arr_0']\n", | |
"\n", | |
"print('We have {} QCD jets and {} top jets'.format(len(data0), len(data1)))\n", | |
"\n", | |
"x_data = np.concatenate((data0, data1))\n", | |
"# pad and normalize images\n", | |
"x_data = map(pad_image, x_data)\n", | |
"# x_data = map(normalize, x_data)\n", | |
"\n", | |
"y_data = np.array([0]*len(data0)+[1]*len(data1))\n", | |
"\n", | |
"np.random.seed(0) # for reproducibility\n", | |
"x_data, y_data = np.random.permutation(np.array([x_data, y_data]).T).T\n", | |
"\n", | |
"# the data coming out of previous commands is a list of 2D arrays. We want a 3D np array (n_events, xpixels, ypixels)\n", | |
"x_data = np.stack(x_data)\n", | |
"\n", | |
"print(x_data.shape, y_data.shape)\n", | |
"\n", | |
"# reshape for tensorflow\n", | |
"x_data = x_data.reshape(x_data.shape[0], 16, 22, 1)\n", | |
"x_data = x_data.astype('float32')\n", | |
"x_data /= 255.\n", | |
"\n", | |
"y_data = keras.utils.to_categorical(y_data, 2)\n", | |
"\n", | |
"print(x_data.shape, y_data.shape)\n", | |
"\n", | |
"n_train = 40000 # less events in this dataset!\n", | |
"(x_trainH, x_testH) = x_data[:n_train], x_data[n_train:]\n", | |
"(y_trainH, y_testH) = y_data[:n_train], y_data[n_train:]\n", | |
"\n", | |
"print('We will train+validate on {0} images, leaving {1} for cross-validation'.format(n_train,len(x_data)-n_train))" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 314, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"model_dir='trained_models_high_pt/'\n", | |
"\n", | |
"model1H = keras.models.load_model(model_dir+'mlp.h5')\n", | |
"model_cnnH = keras.models.load_model(model_dir+'cnn.h5')" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 315, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"predictions1H = model1H.predict(x_testH)\n", | |
"predictions_cnnH = model_cnnH.predict(x_testH)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 316, | |
"metadata": {}, | |
"outputs": [], | |
"source": [ | |
"from sklearn.metrics import roc_curve\n", | |
"fpr1H, tpr1H, thresholds = roc_curve(y_testH.ravel(), predictions1H.ravel())\n", | |
"fpr_cnnH, tpr_cnnH, thresholds = roc_curve(y_testH.ravel(), predictions_cnnH.ravel())\n", | |
"\n", | |
"from sklearn.metrics import auc\n", | |
"auc1H = auc(fpr1H, tpr1H)\n", | |
"auc2H = auc(fpr_cnnH, tpr_cnnH)" | |
] | |
}, | |
{ | |
"cell_type": "code", | |
"execution_count": 320, | |
"metadata": {}, | |
"outputs": [ | |
{ | |
"data": { | |
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAY0AAAEcCAYAAAA7neg3AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvIxREBQAAIABJREFUeJzsnXdYVMf3h9+hdyzYKXZQOihYULFrYu9dY9SYGFN+30RJtUdNTExM1HRrjMYUo8ZERcEWFVFRY8Hee196m98fF1aQtgtL9b7Pc5/dW2bm3FnY2TtzzucIKSUqKioqKiq6YFTSBqioqKiolB3UQUNFRUVFRWfUQUNFRUVFRWfUQUNFRUVFRWfUQUNFRUVFRWfUQUNFRUVFRWfUQUNFRUVFRWfUQUNFRUVFRWfUQUNFBRBCLBVCzCxpO56mtNql8uyiDhoqKioqKjqjDhoqKioqKjqjDhoqZRIhxFAhxL9CiDVCiJtCiCtCiK56lPcVQhwSQmiEEGsAi6fO1xRC/CaEuCOEuCCEeC3TuYtCiHeEECeEEA+EEEuEEBZ6lH1LCHFUCPEo3f7MZQtrV151Owkhfk8ve08I8VV+daqoPI06aKiUVTwBX+A3wAn4Avhal4JCCDNgHbACqASsBfpmOm8EbACOALWA9sAbQojOmaoZCnQG6gENgff1KDsA6ALUAbyAUQa0K7e6jYGNwCWgdnr51TrWqaKiRR00VMoqnsB8KeUvUspkYDngLISYKIQIT//FfDz9/VtPlW0GmAKfSymTpZS/AgcynW8KVJFSTpdSJkkpzwPfAYMyXfOVlPKKlPI+MAsYrEfZBVLK6+llNwA+BrQrt7oDgJrA21LKWCllgpRyt451qqhoMSlpA1RUCogn8EGm/apAjJTyS+BLIcQ3wHIp5Z4cytYErsmseQEuZXrvAtQUQjzMdMwY2JVp/8pTZWvqUfZmpvdxmcoawq7c6nYCLkkpU8iKLnWqqGhRBw2VMocQogLKl+CdTIf7AX9n2ncHTuRSxQ2glhBCZPqCdgbOpb+/AlyQUjbIwwynTO+dget6lM0NQ9iVG1dQnsRMnho4ClOnyjOIOj2lUhbxBFKBIUIIEyHE88ArwNRM11SQUj7IpfxeIAV4Lb18H5TpmwwigMdCiMlCCEshhLEQwkMI0TTTNROEEI5CiErAu8AaPcrmhiHsyo0IlEFpjhDCWghhIYRoWcg6VZ5B1EFDpSziCfwENAceANOAXlLKEwBCiFo8+eWfDSllEtAHZZH4ATAQ+D3T+VSgO8p6wAXgLvA9YJ+pmlXAFuB8+jZTj7JFaVdudWeUrQ9cBq4CAwtTp8qziVDTvaqUNYQQi4HTUsr5uZzvDHSRUr5ZRO1fBMZIKUOLon4VldKM+qShUhbxBE7mcT6v9QwVFZVCoC6Eq5RFPIBTuZ2UUn5WjLaoqDxTqNNTKioqKio6o05PqaioqKjoTLmYnhJCdAe6W1lZjXV0dOTx48fcvHmTOnXqYGpqmuVai4TbmKTGE2PtAoDx/fvI2DhuVq5FVSuhve5B6gNiU2NxNHMszlsxOGlpaRgZqb8N1H5Q+yADtR8UMvfD6dOn70opq+hUUEpZbjZ/f38ppZTLli2TgDx79qzMxh8vS/mZu3b3+pQpcr9PUzl22YEsl3207yPZYlWL7OXLGGFhYSVtQqlA7Qe1DzJQ+0Ehcz8AkVLH79ly8aTxNGZmZlhZWSGEyPmC+Aew6W3l/eVjWCfFMPrPEG5eCAIjpUu8bkVi/uABl6M/wMzYPNe27J7ripW/v6FvQUVFRaVUUi4HjUGDBlG9enXq1q2b/WRNX4jeBMfWAmAhjDExNqHSmSs8vrxRO2g4piZQJUVy9+Q6bEytc2wnVaMh5fYtddBQUVF5ZiiXg0aeBIxVtnQqAvM/+5Lpj9+H0cvAuRkAcclxBK0OwsPBg+Vdl+dY1fnefZCpacVhtYqKikqpoFwOGrt372b69Om4urpSo0aN/AvkMItlZWqFX1U/ktOScy8mBKSmFsJSldJIcnIyV69eJSEhoaRNMSj29vacPJlXTOSzwbPcDxYWFjg6OmZzENKHcjFoZHhP1apVi/DwcLZu3UpYWBjbtm3D0TF/76eEuDgADh0+zOPzT74oHjx8QIpMITw8PMdylWJjSRNwLpfzpYGYmJhc7X+W0KcfbGxsqFatGrVq1cp9XawMkpqairGxcUmbUeI8q/0gpeTRo0ccOXKEmJiYAn83lItBQ0q5AdjQpEmTscHBwVy/rmjVBQQE0LBhw3zL74z6Dx7CoRh7ruBAD59a+LtUZOXmley/uZ/WbVpjJLK76F1c/DVG1tb4BAcb+I4MR3h4OMGl2L7iQp9+OHnyJI6OjuVqwADQaDTY2tqWtBklzrPcD7a2tsTExNCkSZMCfzeUi0HjaTJ8j9PSdFtvaFDNBh7C7rN32ZFwiftxyfi7VCRVKlNPVzRXcLFzyV7Q2Bgp1TWN8kh5GzBUVMAwf9flMsIlY9BI1XG9YXCAMwBLXwigjoM1Ml1apV/DfgDa/ewNCVAXwlVUVJ4hyuWgYW1tjYODQ5FHfQojY2SauhCuYniEEAwfPly7n5KSQpUqVejWrVu+ZW1sbAC4ePEiq1at0h4/dOgQr732GgBLly7l1VdfNbDVZYOFCxcSl76OaWjCw8N1+owKS2BgID4+Pjg7O1OlShV8fHzw8fHh4sWLRd52uRw0nn/+edauXUujRo0KVP7w5Ye8uSaKlfuU9Mwf/rOZezGJ2S80NoI0VfBRxfBYW1vz33//ER8fD8DWrVupVauWXnU8PWj4+fmxYMECg9qZGykpT6ciLxy6zhrowuLFi3MdNAzZTlGyf/9+oqKimD59OgMHDiQqKoqoqChq165d5G2Xy0GjwNw+wZAaN/GS0SSc34vptfsAXLgexpmD2+BKRJZNJGpIu3+LuM2rSIrYmO18gbarByE1dzdflWeHrl278tdffwHw888/M3jwYO25qVOnMm/ePO2+h4dHtl+ZISEh7Nq1Cx8fH+bPn8+uXbty/BW8YcMGAgMD8fX1pUOHDty6dYu0tDQaNGjAnTtKGva0tDTq16/P3bt3uXPnDn379qVp06Y0bdqUPXv2aG0aN24cnTp1YsSIEVnaCA8Pp3Xr1vTu3ZvGjRszfvx47Zrjli1baN68OX5+fvTv35+YmBgAateuzfTp0wkKCmLt2rWcPXuWDh064O3tjZ+fH+fOKanTP/nkE5o2bYqXlxdTpkwBlAHTzc2NkSNH4uXlRb9+/YiLi2PBggXcuHGDtm3b0rZtW0B5Mvvwww8JDAxk7969bNu2DV9fXzw9PRk9ejSJiYlae6ZMmYKfnx+enp6cOpWrOj8A9+/fp1evXnh5edGsWTOOHj0KgKenJw8fPkRKSeXKlVm+XIkDGz58OKGhWfN65dVvJUW5WAh/2uX25MmT/Pjjj7zxxhs6/TqzexSNH8DGNxkDjMk4kQiBaY70MN5Ls7C/ISxrOaM7FUm8Ysml12cgjCQN+9zICCgvFGfrjeaqU8/CV4TqcpuBPv1gb2+PRqMBYO6Wc5y6FWNQW9yq2TC5U718r+vevTtz586lTZs2REVFMWjQIMLCwtBoNCQmJmJqaqq1My0tjZiYGO2+RqPhww8/ZMGCBaxdq6gf7Nixg5SUFDQaDQkJCSQlJaHRaPD29mbr1q0IIVi2bBkzZ87ko48+on///vzwww9MmDCBbdu24e7ujrm5OaNHj+all16iefPmXLlyhd69exMZGUliYiIHDhxg8+bNWFpaam0BiIuLIyIigoiICJydnenTpw8//fQTrVq1Ytq0afzxxx9YW1szf/58Zs+eTUhICFJKhBD8/fffALRt25b/+7//o3v37iQkJJCWlsa6des4ceIE27ZtQ0rJwIED+eeff3B0dCQ6Opovv/ySr776ildeeYX58+fz2muvMW/ePDZs2EDlypXRaDTExsZSr149QkNDSUhIwNfXl/Xr19OgQQPGjRvH/PnzmTBhAlJKbGxs2LFjB9999x2zZ8/mq6++yvKZxcXFafv4nXfeoXHjxqxYsYIdO3YwbNgw9uzZQ0BAAFu3bsXZ2RkXFxe2b99O7969+ffff/n444916rdevXoBZPkcdSUhIYHw8HDV5ZZMLrexsbFERkbi6upKkyZN8q8grTX4+kJS9i+H5H/f5i/pi6tnH7xqVcC5khUZSyXVu8dQ4ewVNHuP8PCfPch+K8DGqhA3AvzUl/pO1ahvIDdZ1eVWQV+X2wyXTFMzU4P79Juamerk8tm8eXOuXr3Kxo0b6datG1ZWVpiYmGBra4u5uTnm5ubaeoyMjLCxsdHu29raZrk+45qMfQsLC8zMzLC1teXixYuMGTOGGzdukJSURJ06dbC1teXll1+mZ8+ehISEsHr1asaOHYutrS07duzgzJkzWjszngzMzc3p1asXVatWzXYvVlZWBAQE4OXlBcCwYcM4ePAgFStWJDo6mi5dugCQlJRE8+bNsbW1RQjBiBEjsLW1RaPRcPPmTYYMGaK9P1ACecPCwmjdurXWlmvXruHm5oaTkxMdO3YE4IUXXmDBggXaejP3lbGxMcOGDcPY2Jjz589Tt25d/Pz8ABgzZgwLFy4kJCQEIQRDhgzB1taWli1bsmnTpmyfY+Y+j4iI4LfffsPW1pZu3brx8ssvk5aWRrt27YiMjOTOnTu8+uqrfPvttzx+/BgHB4dswci59VvGelfmz1FXLCws8PX1VV1uM5PhVqbzY5yRkVY+5GmS96RwyyyF1w5UhgPw3QhPOjauBiidZ+MPSQnL4J89UDcY7OwKbriaEKvUMaW7e4m236NHD9566y3Cw8O5d++e9riJiUmWv+/CRK9PnDiR//u//6NHjx6Eh4czdepUAJycnKhWrRrbt29n//79/PTTT4Dyf7V3714sLS2z1WVtnbNOG2R39xRCIKWkY8eO/PzzzzmWyagvNw9GKSXvvPMOL730UpbjFy9ezLG9nLCwsND+MMjVUzIdc3NFvNTY2DjfdZuc6hJC0Lp1axYuXMjly5eZNWsWf/zxB7/++iutWrXKsR5d76O4KJdrGvrGaeSFg6UDgc7OfNxPGeljEtX1BpXiY/To0Xz44Yd4enpmOV67dm0OHToEKF5RFy5cyFY24xd6fjx69Eg7jbts2bIs58aMGcOwYcMYMGCA9ou1U6dOWaZloqKidLqXiIgILly4QFpaGmvWrCEoKIhmzZqxZ88ezp49CyjTMadPn85W1s7ODkdHR9atWwdAYmIicXFxdO7cmR9//FH7tHPt2jVu374NwOXLl9m7dy+grAkFBQUByhpGbv3i5ubGxYsXtfasWLGCNm3a6HR/T9O6dWvtQBseHo6DgwN2dnY4OTlx9+5dzpw5Q926dQkKCmLevHm5Dho59VtJog4a+VDRoiKPUm4QULsSAAnJalyGSvHh6OjI66+/nu143759uX//Pj4+PixevDhH5QMvLy9MTEzw9vZm/vz5ubYxdepU+vfvT6tWrXBwcMhyrkePHsTExPDCCy9ojy1YsIDIyEi8vLxo3LgxX3/9tU730rx5c0JCQvDw8KBOnTr07t2bKlWqsHTpUgYPHqxdMM5tgXnFihUsWLAALy8vWrRowc2bN+nUqRNDhgyhefPmeHp60q9fP+2A0KhRI5YtW4aXlxf379/n5ZdfBmDUqFF07dpVuxCeGQsLC5YsWUL//v3x9PTEyMiI8ePH63R/TzN16lRtP4WEhGQZkAMDA7WfWatWrbh27Vqug0FO/Vai6Jp4oyxsGUmY9uzZI52cnGRkZKQ+OUlypPOvnaXHUg956PoZ6TJ5oxy3/EC2a+4tXSpPuLrJlEePCtdYWpqUU+ykDJtduHoyoSacUdCnH06cOFF0hpQgjx8/1rvMgQMHZFBQUKHbDgsLk88//3yh69GVCxcuSHd39xzPFaQfSoqi6LeMv++CJmEql08aLVq0YPny5fgbIM/FiMaK66CluTJ/WcHSrNB1qqiUBebMmUPfvn2ZPXt2SZuiUoool4OGIalmrSx6rzyxEoeqZzhy9SFz/j7FrjN3sl17vkdPbn/6WeEb3bsQ5nso26EVha9PRaUAhISEcOnSJYPMoQcHB7Nx40YDWKUbtWvX5r///iu29oqK4u43XSiXg8aRI0d4/fXXOXjwYKHrcrRxxM7MjvXn1kOlDZy/G8s3O88xb3O09hqbdu2o0L8/pKURu39/wRsTAjpMhUbdoU5riL0LV/YV+h5UVFRUDEW5HDQ0Gg1Hjx7lwYMHha7LtZIrewbvoVvdblS0MeLABy1o61qV+OQncgNmTk7UmDEdc1fXQrdH0JvQa5GyWVUqfH0qKioqBqRcDhqG9J7KwNTYlJuxNwlaHcTd5GhO34ohWVW4VVFRecYoF8F9T8uInDhxAlD8x83MDLNw7Z3iTZx9HP88+geRegOoyLawHViYPAm0qXD/HkaxcQaT7WiWmMCDGzeJLkR9qoyIQkFlRMoTqamp5fK+9OVZ74fCyoiUuJusIbcMl9v9+/dLQP71118FdErLmej70dJjqYf0XOolXSb/KZt/FCq/3XFOe/7SmLHyXLduMv7kSRl/8qRMunGjcA1+2kjKVYOlvHE063b/gs5VqC63CmXN5RaQw4YN0+4nJydLBwcHndwvra2tpZSK2+lPP/2kPR4eHi4nTpwopZRyyZIlcsKECQa2umwwe/ZsGRsbWyR159WvXbt2lQ8ePMizfJs2beSBA9nd+nUhICBAent7SycnJ+ng4CC9vb2lt7e3vHDhQpbrCutyWy6eNJ7GxsYGV1dX7Aoj6ZEDdezrYGNqQ0xyDH38q7DthIb9F+4xtnVdAIwszEk8c5YLvdKDb4yNabhnN8YVKhSsQRMLiP5L2Z5mwgGokn8qW5WySWZpdEtLy0JJo2foNfn5+RU4ullfUlJSMDEx3NeLIfN6L168mDFjxmBllV0nrijzh2/atKlI6s1gf7oTztKlS4mMjMwmpmgoyuWaRkaUqqHD7U2NTJngMwGAlt7XqVXBktCTt9l95i4A1d7/gFpfLqDWlwuoOGQIpKaSFhtb8AYHr4aBK7Nurd9WziU8KuztqJRyVGn0simNfv36dbp06UKDBg2YNGmS9njt2rW5e1f5rpgxYwZubm507NiRwYMHZ/ks165dS0BAAA0bNmTXrl3Z6i9pufRy+aRRlATVCmLugbn8d+8/Xmg5mrd/PcqBi/cJauCAabWqmKaraqY9NsCcaZWG2Z8mTLKLxKkUIX+HwM1jhq2zuid0nZPvZYMGDWL69Ol069aNo0ePMnr06By/RHJjzpw5zJs3T+vnn9sv3aCgIPbt24cQgu+//56PP/6YTz/9lGHDhvHTTz/xxhtvEBoaire3Nw4ODgwZMoQ333yToKAgLl++TOfOnTl58iQABw8eZPfu3TmKGUZERHDixAlcXFzo0qULv//+O8HBwcycOZPQ0FCsra2ZO3cun332GR9++CGgyHrs3r0bUKQ3QkJC6N27t1YafcuWLZw5c4aIiAiklPTo0YOdO3fi7OxMdHQ0P/zwAy1btmT06NEsWrSIt956i08//ZSwsDCtZEpsbCweHh5Mnz6dhIQEGjRowLZt22jYsCEjRoxg8eLFvPHGGwA4ODhw6NAhFi1axLx58/j++++z3WdUVBSHDx/G3NwcV1dXJk6ciJOTk/Z8ZGQkv/32G4cPHyYlJQU/P78sgcgpKSlERESwadMmpk2bli3HRm592a9fv/z/KAxAuXzSOHPmDGPHjs2xswtLbfva1LCuQVJqEv2bKH8Ii8PPMeXPsh9IpFK68PLy4uLFi/z8888899xzRdbO1atX6dy5M56ennzyySccP34cUMQSMxIE/fjjj1r9qdDQUF599VV8fHzo0aMHjx8/1i4s9+jRI8cBAyAgIIC6detibGzM4MGD2b17N/v27ePEiRO0bNkSHx8fli1bxqVLl7RlBg4cCChu9NeuXdPqLllYWGBlZcWWLVvYsmULvr6++Pn5cerUKa1su5OTEy1btgQUSfGMwedpjI2N6du3LwDR0dHUqVNHqws1cuRIdu7cqb22T58+APj7++eaWrV9+/bY29tjYWFB48aNs9wPKHLuPXv2xNLSEltbW7p3757lvC5t5NSXxUW5fNJITEzk7NmzPHz4sEjqNzEy4armKgCzenuwOPwchy7n3Nbd77/H2CY9x0Gnjlg+pVZaYA58D+fDofkEMCtEDg+VvNHhiaAoUaXRy540esY1uV1niDZKUi69XD5pFEWcRmZS0lI4dPsQN2JuMDTQhcY17LLFbJjVdsHI2ppHv/7G/aVLuffdd9z7LvujrN5UcAaLCvDfrxA2E64UIgJdpdSjSqMrlEVp9NwICgpiw4YNJCQkEBMTo1230oeSlEtXB40CMMhtEAAn7inxIGYmRtx8nMDnoad5nKDk27Dy98f1YCRux47iduwo5o0aIfNJ2qITVRpCyCUYlT4/LdUAw/KMKo3+hLImjZ4bTZs2pUePHnh7e9OnTx+aNGmCvb29XnWUqFy6rr65ZWHLiNOIjo6WQBYfdUNy6t4p6bHUQ3564FMppZTf7DgrG7y7SbpM3ij/PpZzbMb5vv3kpXHjDGfEpX2KjPqZ0DwvU+M0FMpanEZRoEqjK5QGaXSNRiOllDI2Nlb6+/vLgwcP6ly2sH2pSqPngLW1Nb6+vjnmKjYErpUUjaklx5dw4t4JxrWux8bXlMfD8SsPcu5O9lzjwtgYUlKzHS8026ZDqppNUMXwqNLoRce4cePw8fHBz8+Pvn37anOSlwXK5aBRq1YtPvvsMzp06FBkbbzo8SIAe68rc6b1qtjQxb06AF9uO8ONR/FZC5iYIFMNOGhUdFFeb0TBnZwf51VUCoMqjV50rFq1iqioKE6dOsU777yjV9mSlksvl4NGcTDSfSQAt+JuAWBsJPiwe2MqWZuxLuo6vx+6luV65UnDAGsaGdhWh4GKNwv5eGOoqKioGIpyOWhcv36doUOH8ssvvxRZGxUtKlLJohKpaU+eHmpWsGTfO+0BCDt1O2sBYyNSY2JIOHVKuyXfeuqagvLwUv7XqKioqBiAUhunIYRoBLwOOADbpJSLdS2blpbG9evXefz4cZHZB0q8xi+nf6FPgz64O7gDiieVmbERJ288psNnOwDo6lGdAdbWxO3d90SXChBmZjTcvw+jXIKh8sXUQnldMwxejQSHBoW6HxUVFZX8KNYnDSHEj0KI20KI/5463kUIES2EOCuECAGQUp6UUo4HBgBN9GmnqF1uM+jTQInc/OPsH1mOT+/pTrBrVVyr2fI4PpmV+y6xImAAxjPnarWp7Pv1RSYlkVaIoCzqBEOg4kaoalGpqKgUB8U9PbUU6JL5gBDCGFgIdAUaA4OFEI3Tz/UAdgPb9GmkuAaNsZ5jAUhMTcxyfFCAMwuH+rFwqB+jWtbG2Ejw7QkNszXVOe/WFLuOHbFwdVMuLsx6hLEJ1Femw3hwEYpRtEylaLl58yaDBg2iXr16NG7cmOeeey7HoLeiILOwXm4sXbqU69eva/fHjBmjzWNjKJYuXcqrr76a43EjIyOOHj2qPZZZsLF27dpaWRCAX3/9lVGjRmWrJzw8PEcBR0MTGBiIj48Pzs7OVKlSBR8fH3x8fHKVCCntFOv0lJRypxCi9lOHA4CzUsrzAEKI1UBP4ISUcj2wXgjxF7AqpzqFEOOAcQDVqlUjPDxcm+b11KlTRZ6AqIJxBQ5fPkx4cs7tNAbmBZnyf+HJhEXf4cSVu8xtbYXlpYvYAYfnzCG2ELpCFR6cwgfgtxc5c2Qv1xyz6tioSZgUylISJpkuvDdkyBC+++47AI4ePcqFCxeoUaNGgevVNfmQlJKYmJgschhP88MPP1CnTh1sbRWJnIzgQUP2W0JCAklJSdnqTEhIoFatWkybNo2lS5cCyg/EmJgYNBoNUkoOHDhAREQEjRo1Ij4+nuTkZG09Gf0QFxdHSkpKkX/WGRp4P/30E4cOHeLTTz/VniuJv7Myl4QJqA38l2m/H/B9pv3hwFdAMLAA+AaYoEvdGcF9jx49ki1atJAbN24saPyLzrRY1UJ6LPWQjxPzDhiKT0qR//slStYJUWxKjY+XJ1zd5K3PPy+cAakpUp7eogT6hc3OdloN7lMoS8F927Ztk61atcrxXFpamnzrrbeku7u79PDwkKtXr5ZSKvfXpk0b2bdvX+nq6iqHDBki09LS5KZNm2T//v2llEpQW1hYmOzWrZuUUspVq1ZJDw8P6e7uLidNmqRtw8XFRd65cydbgNwnn3wip0yZIteuXSutra1lw4YNpbe3t4yLi8uSPCi3eq2treW7774rvby8ZGBgoLx586aUUsr169fLgIAA6ePjI9u3b689nltCoyVLlsiXX35Zuru7y1OnTkkppXR3d9cmG3JxcZELFiyQQ4YMkVJKuXbtWjly5Eht+YzgvsxBcvfu3ZM9e/aUnp6eMjAwUB45ckRKKaWHh4d88OCBTEtLk5UqVZLLli2TUko5bNgwuXXr1ix2hYWFyVatWslevXrJRo0ayZdeekmmpqZmsbs0JL4qD0mYclLaklLKcCC8IBXa2dkxa9YsgoODC2GWbrzg8QJfHPqCh4kPsTWzzfU6C1NjLEyNSJNw8NJ9/F0qARB/6HDhDDAyhvrp8SgXd8OeBdBkNJjbFK5eFQDmRszl1H3DxsG4VXJjcsDkXM//999/WaSyM/P7778TFRXFkSNHuHv3Lk2bNqV169YAHD58mOPHj1OzZk1atmzJnj176NixIy+99BKx6Xld1qxZw8CBA7l+/TqTJ0/m4MGDVKxYkU6dOrFu3Tp69eqVr/39+vXjq6++Yt68eTRpknW5Ma96Y2NjadasGbNmzWLSpEl89913vP/++7lKs+eFkZERkyZN4qOPPsqmlwUwYMAAFi1apNWQyo8pU6bg6+vLunXr2L59OyNGjCAqKkrbjy4uLtStW5ddu3YxYsQI9u3bx+LF2X1zSlKyvLgoDYPGVcAp074jcD2Xa3Pk6RzhUHzTMvFxShDfJ1s/oU+lPnleWzFRidPou3gv01pYEABozp4tvJ0FVimbAAAgAElEQVRSEmBZE6uLu+DiLv67HsvdKs0BdXoqg4JOTyUlJZFqyKDM9DrzmpbIbVoGYPv27fTu3Zu4uDisrKxo0aIFO3fuxNbWFn9/f+zt7YmNjcXd3Z2TJ0/i7e1N+/bt+eWXX+jevTsbN27kgw8+YOfOnbRs2RILCwvi4+Pp27cvoaGhtG/fXjs9FRMTQ1pamtaOxMREEhMT0Wg0pKamEhsbm2XKJzY2Ns96zczMaNOmDRqNhsaNGxMWFoZGoyE6Opp3332XW7dukZSUhIuLCxqNJs/pqaSkJLp3786MGTM4duxYtump+Ph4Jk6cyPTp0+nYsWO+01M7d+5kxYoVaDQamjZtyt27d7l69SpNmjQhNDQUJycnXnjhBZYsWUJ0dDT29vZIKbPYFhcXh7+/P1WqVCEuLo7evXuzfft2OnfunO/nWpwUdnqqNAwaB4AGQog6wDVgEDBEnwqklBuADU2aNBkbHByMRqPBxcWFWbNmaUXKioqWaS35dsW3hGnCWNBnQZ7XBqWmcV0c47dDV5nybwKrW7aj4oVTtGqSt3OYMDXFKI/5ZQCCT8Ct/+DrIDwauYJHMKAs9hXHE1dpR59+OHnypHau/oOgD4rQqpzx9/dn48aNWhsyY2pqioWFhfacqakplpaWWFlZYWVlpT1uYWGBqakptra2DBs2jIULF1KxYkUCAgKoWbNmlvMZ15uZmWFra4sQAhsbG2261oxrpJSYm5tja2uLsbEx1tbW2nMZ+3nVa2pqqk3BbGNjgxACW1tbQkJCskmz29raZimbmYzjFStW5O2332bhwoUYGRlhY2OTxf6xY8cyf/58fHx8stik0WiwtbXFysoKExOTLGUyrhFCYGdnp1XRvXnzJrNmzeLvv//mn3/+ITg4OJtdmevLsDOjv57ui5LEwsICX1/fAn836Ow9JRSGCSE+TN93FkIE6NOYEOJnYC/gKoS4KoR4UUqZArwKbAZOAr9IKY/rU+/TmJiY8ODBgyKP0wAlBWwd+zoAxCXH5W2XsREf9fFgfJt6AOy9GkPa9eucbtI07y2wWf6BgEKAcfrAsivvR3uV0k27du1ITEzULoIDHDhwgB07dtC6dWvWrFlDamoqd+7cYefOnQQE5P1vGBwczKFDh1i2bJk2qVFgYCA7duzg7t27pKam8vPPP2eTAK9WrRq3b9/m3r17JCYmZpGuyE12XZd6nyYvafb8GDVqFKGhodq0tJkxNTXlzTff5PPPP8+3ntatW2vzhYSHh+Pg4ICdnR1OTk7cvXuXM2fOULduXYKCgpg3bx6tWrXKsZ6SlCwvLvR50lgEpAHtgOmABvgNaKprBVLKwbkc3wQUOOv609NTGdMJ0dHRxTI14yk8ucAFFmxeQFPrphiLvBPTB1hIbJqYs0N04LJtNdwqGeFbNeePwuTadSz37mX/ls2kODnleI0WKQkGHsUnc7iYp+lKO2XJewoUGfCQkBA++ugjLCwscHZ2Zs6cOfj6+rJjxw48PT0RQjBt2jSsra2zeQIlJSWRkJCg3e/UqROrVq3i66+/RqPRaPNit2nTBiklnTp1ol27dtrpnQzvqUmTJtG0aVNq165NvXr1tNNTAwcOZNy4cVhaWhIaGqqdnsqrXnjiLZTZo2ny5Mn069ePGjVq0LRpU+30UX7TUxnHx40bx+TJk7NMT2XYP2DAAGbMmJHv9NT//vc/XnnlFTw8PLC0tGTRokXa6/38/LRl/P39uXbtGt7e3tnsiouLIyAggLfeeovjx4/TsmVLOnTooL2uvExPCaljnIAQ4pCU0k8IcVhK6Zt+7IiU0lvvVouIJk2ayMjISEW+18iIDz74gOnTpxd5uzuv7mTCtgkAvBf4njbfRn4kJKfi9sE/uNe004odVrQ2Y2igszYTlyY0lKuvTgQjI8zq1smxHtsOHaiansOY5T2VjH4DlkPjnur0VDr6Tk81atSoaA0qATKmZZ51iqofwsPDs+RkL61k/H1n/p8QQhyUUuoURK3Pk0ZyeiCeTG+kCsqTR6lDCIGJiQnJycUjGd7asTWrn1/NoL8GMWv/LHo36I25cT5rECgeVe3dqrLt1G2OX38ylVbF1pzO6YOIpa8v9j175ho5Hn/oEDHhO54MGv4vKIPGxd3QuGeh701FRUUlM/oMGguAP4CqQohZKPEVxb9KqCMdOnTA27v4HoLcHdxxtHHkasxVjt45ioeDB5Ym+WtKfT+yCalpytNe9C0Nzy/Yzcy/TmgHDZPKlak5N/c81VdemUDM9u0kREdj4eoK7r3gT1uI+BZa/c8wN6eiopIvwcHBz8RTvc7TUwBCCDegPUpsxTYp5cmiMkwfMq1pjF25ciWgzGHb2BRvrMKZhDMsuPXEg6qtbVt6VeyFkdBdrWXUP4o//cetLalqlX85y7Aw7Nb8wuOhQ4hPX5zzPDqDyvcjOeI1lStmDYq9H0oj+vw92NvbU79+/SK2qPhJTU3V5vl+lnnW++Hs2bM8evQoy/9E27ZtdZ6e0mdNY66UcnJ+x0qSjDUNKBlX0+S0ZP448wf3E+6zMGohACu6rsCnqo/OdXy1/QzztpymroM1298Kzr/N27c527oNRjY2T9RyU5Mx4Ta1P3qFHfg/E79+8kNd01DXNDJ41vuhsGsa+ggWdszhWFc9yhcrw4cP55VXXinWNk2NTBngOoDx3uOZ12aeYsffw5m4fWK+7rgZTGir/MI9fzeWf8/lLRoHYFKlCg6vvIJd167YBAdjExyMqbMzCffNSLh8F+M7d0iNiS34TamoqKhkIt8nDSHEy8ArQF3gXKZTtsC/UsqhRWeebuQ0PTVw4EB8fX0JCQkpEZuS0pL49cGv7I3Zqz3Wq2Iv6prXpY55zl5QGYRfSWbp8SQA5rWxxMFSPzFiy907sVv5s3Y/xcGBezNn6FVHeUOdnlKnZTJ41vuhsNNTuggM2qOIDP4MuGTaKukqcFVcW4ZgoZRS1qpVSytYVpLcibsjB2wYID2Wemi3wRsHZ9kGbhgofz/9e5Zy7/9xTLpM3ihX7L2od5upjx/KRy9UlQ8ntZeH+/SRJzw85e2FC7VbfAkL8pUEZUmwUEopb9y4IQcOHCjr1q0rGzVqJLt27Sqjo6MLVWeGUF9+ZAgW5sWSJUvktWvXtPsvvviiPH78eKHsy6mN3AQLhRBaUUEpswsW9unTR3suN8HCvAQEu3btKh88eJCnfZlFGvUlICBAent7SycnJ+ng4CC9vb2lt7e39h6KkiIXLJRSPgIeoeS5qAg0ACxAcW2VUu7Ub5wrHkxMTEhKSippM3CwdGBNtzXEJcex+eJmNl/cnE2icc+1PUzdO5XeDZ5k9ZvQtj4r9l3i253nGdbMRa82jWzssHNOgMoPSbRug/nxE9xd8KX2fOLJUzh+mbfkiUrJIaWkd+/ejBw5ktWrVwMQFRXFrVu3aNiwYQlbp7B06VI8PDyoWbMmAN9//32xtu/o6MisWbNYs2ZNjucjIyM5fvw47u7uBap/06YCxxrrxP79+wGlHyMjI/nqq6+KtD1Doo+MyBhgJ4rcx7T016lFY1bhMTY2LrY4DV2wMrWid4PefN3xa77ukHULqhVEmkxj+fHl7Ly6k5ikGKrbK6lcL9+P41G8nvchBLh1A2NT4lu3wu34f9rNvFEjNNu3E92kabbt/oqVRXDnKvoSFhaGqakp48eP1x7z8fGhVatWSCl5++238fDwwNPTU/ulmbGo2a9fP9zc3Bg6dChSSv7++28GDBigrSc8PJzu3ZWcKz///DOenp54eHgweXJ2f5aLFy/i4eGh3Z83bx5Tp07l119/JTIykqFDh+Lj40N8fDzBwcFkOKHkVq+NjQ3vvfce3t7eNGvWjFu3bgGwYcMGAgMD8fX1pUOHDtrjedGtWzeOHz9OdHR0juffeustPvroo3zruX79Ol26dKFBgwZMmjRJezxzIqoZM2bg5uZGx44dGTx4MPPmzdNet3btWgICAmjYsCG7du3KVn94eDitW7emd+/eNG7cmPHjxxd5criiRp84jddRJEP2SSnbprvfTisaswpP27Ztc5WXLm0MaDiA3dd280nkJ9pjQbWCaOHtyb9HXOizaA/b/hesf8W3T2BZ+wYi0/xtlYkTidu/L9ulD9b+yq1Zs7BuFoh5AzXXeAY3P/qIxJOGlUY3b+RG9XffzfW8Ko1efNLoUVFRHD58GHNzc1xdXZk4cSJOmeR6IiMj+e233zh8+DApKSn4+fll+WxSUlKIiIhg06ZNTJs2TZtwKTPlTS5dn0EjQUqZIIRACGEupTwlhHAtMsv0ICdp9F69emFjY1MmdJcEgg9qfkB8WjwbH24kNi2W3dd2o2S6ncW5O7HMWLmVVo6mOtdZN9YMZ6DG+Z8JD8+U7c1IQPPm2a6vEHEA85MnOfHhFDQDB2Q7L83NkZb5ByuWVgqqPZWclEyKgaXRjZKS89QfUqXRi0caPSEhgdatW2NkZERycjINGzbk5MmTVKhQQdsHoaGhdOnShZQUJa1B586ds/RBly5d0Gg0uLq6cv78+Rz1qEqbXHpxSqNfFUJUANYBW4UQD9Az70VRIZ+SRgflEb9ly5aYmur+RVsaGMlIACZsm8DOqzuxbfQeyY/d+fFUd55r1VWbvClfgoNh5iac7+7AOXh9vpfLNm04/9zzcPgwFoezJ4YSZmbU3xGOScWK+txOqaGg0ui2U6cUoVU5o0qjF480uoWFRRY5dHNz82x9YG5unkXe3MzMLEsfVKxYEVtbWxITE0lLSysTcunFIo0uFPW816SUD6WUU1HkQ34A8n+WLSHeeecdWrRoUdJmFJh3At5hWKNhAJjaHcey1ir6Lt5LvXc38f66Y7pV4tJSed2a/xefEIKaH39M9WnTsm32ffogk5K4NXNWQW9HRQ9UaXTdMZQ0em4EBQWxYcMGEhISiImJ4a+//tK7jvIml67ToJHukrUu0/4OKeV6KWXJuyflgpmZGfHx8SVtRoFxtHVkcsBkjow4AoCx1WUqNPoAK9fJrD3/NVce6/CQ1+495TXiOwidBpq8FxgtPT2oOHBAts0hPUgyJiyMW3M/JmbHjkLdm0reCCH4448/2Lp1K/Xq1cPd3Z2pU6dSs2ZNevfujZeXF97e3rRr146PP/6Y6tWr51mfsbEx3bp1Y+vWrXTr1g2AGjVqMHv2bNq2bYu3tzd+fn707JlV4NLU1JQPP/yQwMBAunXrhpubm/bcqFGjGD9+vHYhPANd6n2aqVOn0r9/f1q1aoWDg4NefWVmZsZrr73G7ds555t58cUXtVNLBaFp06b06NEDb29v+vTpQ5MmTbC3t9erjubNmxMSEoKHhwd16tShd+/e+RcqzejqmwssBJrqen1JbJnjNNq3by/r1auXr89yWeDio4tyzv45cszmMVniPYZvGi6va67nWfbM8jelnFldyil2UobNKbANd5cskSd9/eSJRo3luZ69ClxPSVHW4jSKAl3jNMo7+vaDRqORUkoZGxsr/f395cGDB3UuGxYWJp9//nm92itqChunoU+ocVtgrxDinBDiqBDimBDiaFEMZIbA3Ny8TD9pZMbFzoXJAZP5rtN3/NNrNwk3lF8qh28fptNvnfg08lPuJ9zPsexVpx7wdroHyaU9cCUC7p3L8dq8qDxqFG6HDmLboQNpGg1xhw9n25Kvl4olLhUVgzJu3Dh8fHzw8/Ojb9+++Pn5lbRJJYo+goU5RphJKS8Z1KICkJOMyLx589i1axd//vlnyRpXBCw/kcj2y0lYunyLidVF7fH/Vf8ftc1rZ7k2Qyqg+b+jME96oD2+P2AR8Va19G7bbvkKLP/9N8dzaRYW3H/3HSVORAeksTFplXRc2C8kqoyIKp+RwbPeD8WmclsWyKxyO2vWLJKSkpg2rdSGkhSKAxfv0//rvQwNdCJKvsNlzWUA6leoT0hACIE1AoFMXkP3L8D9c3Dp3yc5xN+9AWZWerWb+ugR8UezL8Rrtmzm4dpf9b6Pmp98jH16sFlRoq/3lJubmzZ7YnnhWVd3zeBZ7gcpJadOnSq2zH1lipYtW5ZrSXBfpwoAbD5+m8j3/+Lbo9/y5eEvOfvwLGO2jAHgFe9XcEh24EHCAypWqgOV6igeVVcj4cIO2Pgm9FoERrr/6jK2t8emVXbvD0tvL6wCAiFNt5iGlHv3uf3xx1x/exJWTZpgWqNG/oWKCQsLC+7du0flypXL3cCh8uwipeTevXtYWFgUqp5yO2gkJydz7949KlWqVC7/8U2MjWhetzJ7z99j1f7LjAscx1jPsUTdiWLLxS2sPLmSRUcWATB9zXRq2dRilPsoutXths3AlTDHCY6uBs0NGPGnzlNKuWFsZ4d99256lYk7cICYsDDOtm1H9alTsQ5SXISNK1TE2Ma6UPYUBkdHR65evZqjG2dZJiEhodBfGOWBZ7kfLCwscHR0LFQd+qxpCGAoUFdKOV0I4QxUl1JGFMoCA5J5emrChAksWrSIhw8f6u0iV1Y4eOkBfRcr6wurxgTSov4Td8Xjd49z/tF5dhzdwb/x/6JJfuJTv7D9QuwfXIGN6XnF/UaBnxITUse+DnZmdsVif2pMLA9WLOfOF9nFExtGHsDYgBkHSyIpV2lD7QMFtR8UimN6ahGQBrQDpgMa4DcUPapSh5mZGQDx8fHldtDwd6lIZ/dqbD5+iyHf7+eLQT709FEWt90d3HF3cMf2ii3zes/jdtxtBm0cxJ34O0zYNkGpoGa6f//Nf2DTP9p6nW2dcXdw5+PWHxep/cY21ji8/DIW7u6k3FO8v+5+vZjkS5e52K8/9f75u0jbV1FR0R99Bo1AKaWfEOIwgJTygRDCrIjsKjTm5uYA5cbtNje+Gd6EbSdv8eKySF5fHYVnLXvqVsn+C72qVVW2D9jOsTvHeJT06MmJ9a/D46swcBV7bkdyL+Eef1/4m8uayxy8eZDFHRdTyaISNqY2WJgUzSO9TbrgHoBdp45E+zch6eJF0uLiMLLSb6FeRUWlaNFn0EgWQhgDEkAIUQXlyaPEyUmwMGMdIywsjEuXStwruEgxBuraG3H+URrtPt3Bks5W2vvPT5TM0aYN9W/9wLHIgwQ6KB5XbtXcmH9rPrfjb9N3fV/ttW9Ue4N6FvWK8lYAsOrZA9s/1xMx/3MSWhpGCqag4mzlCbUPFNR+UChwP+gaBYiynrEeuArMAqKB/rqWL44tc0T43LlzJSB3796tT7BkmabF7G3SZfJG6TJ5o/ZYvpHQVyKVaPHfx2c7FXopVK45tUb2W99PG4V+7uE5A1udnZh9++UJVzd5wtVNnu/bT57vP0DG7N1XqDr1iQgvr6h9oKD2g0KRR4RLKX8CJgGzgRtALynlWv2HqeLB2dmZmTNnFtpToCyx+c0n0zxNZ2XX9c8RR3+oVA+OrILdWYXd2ju3Z4DrANZ2X0vfBsoTR891PfFc5kmf9X2YGzHXYLZnxjowgFrzP8O6TWuMK1ci4ehRLo8aRey//5Jw+nSRtKmioqIb+mTuexOIkVIulFJ+JaU8WYR2FZrq1avz3nvv4eKiX6rUsoyNuQnHpnYC4I4mkUXheSeg0dJltvIaOgXiH+R4yQfNPuCTNp/gV9WPtk5tOfPgDCtPrqTtL20J2RVC1O0oQ9yCFruuXXH+5hucv/kG61atALg8+kUu9OjJ/eUrDNqWioqK7uijPWUHbBZC7BJCTBBCVCsqowxBamoqly9f5uHDhyVtSrFia2HK5jeUJ46P/4nm08iE/As17AxBbyrvY3JWCzU2MqZL7S4s67qMBe0WsKbbGixNLHmc+Ji/zv/F8L+H47nMk11Xd2VMZxoMp0ULcflpJVXeUFyEb330EZdGjCS1GBPXqKioKOgzPTVNSukOTABqAjuEEDrOgRQ/MTExuLi46K3PXx5wrW7LuglKoNyxu6kcupzz00MWaqWnsFwYANfzf2poXLkxEUMjODj8IJ+3fTKt9cq2V/Ba7kXvP3sTn2IYzzVhaoqVvz8O41/CZaXylBEXEcHppgHcX7GSx1u2GKQdFRWV/NHnSSOD28BN4B5Q1bDmGA6rdFfNmJiYErakZPBxqsAXg3wAGPGDDvGX9TuAY3rIzbdtYP+3oOMTQ3vn9hwbeYzV3VbT1qktAGcfniXgpwACfgpg3JZx3Ii5UaD7eBqrJk1wPXQQq+bNALg1axbXXnudc12fIz4qirSkUpviRUWlXKDPmsbLQohwYBvgAIyVUnoVlWGFxdTUFDMzs2LNvVvayAj0i0lM4eSNx3lfbGoJY0Khyxxl/++34efBerXnXtmdBe0WEDE0gld8XsG1oivxKfHsvbGXTr91Ivp+dEFuIxtGVla4LFlCg73/4rJyBSY1apB04QIXBw0m2subtMREg7SjoqKSHX2eNFyAN6SU7lLKKVLKE0VllKHILSXls8R4LyXIsesXu3Qr0OxlGBeuvD9dsIhsSxNLXvZ+mV97/MqxkceoX0GRGe+3oR+eyzxZd3ZdPjXohknFilg1aUL97duo9fl8RPrTZbS3D7F79xqkDRUVlazos6YRIqU0rItMEVOhQgUePNBhPr8cE1jjiYJtx890TNNa0xcadFben9teaBv+6PkHnwd/Tk3rmgB8sOcDPJd5cv7h+ULXDUogp12XLjTcsxvzxo0AuPzCaK6HvGPwRXkVlWedfAcNIcTu9FeNEOJxpk0jhMhnzqNkmTZtGqNGjSppM0oUIQTb/9cGgDO3Y6gd8hdHrujgUdb0ReV1RW/40h/C50JCwT/u9i7t2dxvM4s7LNYe6/lnT2bum2mwL3YjS0vq/PYbNWbOAODRunUkX71qkLpVVFQUykUSppwy9+mTqa08k9EP9xPS+L/wJ95Ms4MsqWGT928Ghzt78Tg+J8ux3S1XkmJauAQ2Ukq2Pt7KhocbtMcamDfg1WqvYiQK4puRHYt9+7BfuozUChW4O2M6MYmJz/zfg/o/oaD2g0KRZ+4TQsyVUk7O71hJklkaPTw8HDc3N65du4a/v38JW1ZyPC0DvTj8HHP/OQXAqRldsDDVIQFTWir80BGuHYSei8B3qEFsi0mKYfTm0Zy8/yROdLz3eAa6DsTB0iGPkvmTFhdHtN+Tz/3R8OE0e+/dQtVZ1lElwRXUflAoqDS6Pj/rOuZwrKse5YudTz/9lKCgIHVeOxMvBz8RHHT74B+idJmqMjKGQauU94cMF/diY2bDL91/4eCwg9pjXx/5mra/tGXd2XWF+tyMrKxwPRhJhQEDALBfsYJrb71N0uXLhbZbReVZRpc1jZeFEMcAVyHE0UzbBeBo0ZtYcBwcHEhISCAuLq6kTSlVnJrRhc7uSkB/r4V7dCtkWx0sK8KV/ZCsQ5S5HpgZm3Fs5DH2DdnHQNeBgLJY7rXciw3nNuRTOneMrK2pMX0aNWbNBODxxo2c69TZIDarqDyr6PKksQrojqJw2z3T5i+lHFaEthWaKlWqAHD37t0StqR0YWFqzDfDnzyJfr9LRy+musHK69LnDG4TgLWpNe83e59lXZZRw1rJGf7u7ndp90u7QnlaVejbl9ufz9fuR/s3Ie7gwTxKqKio5Ea+g4aU8pGU8qKUcrCU8lKm7X5xGFgYHByUeXF10MiZjRODAJj510nWH7mef4He3yqv1w7CLyOKzC6/an5s6beFeW3mYWZkxp34O/T8syd91/flbvxdHifp78UlLSyovy0UYWpKWmwsl4YO49H69UVgvYpK+aagLreasuBymzFo3Llzp4QtKZ141LLnfx0bAvDaz4c5eCmfmBYTM3g1/Rf6iT8hqWin/TrX7szB4QfpUa8HAKcfnKbtL21p+XNLPJd5surkKr3qM61VC7djR6kxW1H1vT5pMo83b1HXOVRU9ECXJ42g9FdbKaVd+mabsV/0JhYcNzc3Vq1ahZdXqVU7KXEmtm/A857KVFDfxf/mX8ChPgSOV95/4Q3XDhWhdQqzgmZxaPgh3mryFu8EvENVK0XybHbEbNqsaaP3gnmF3r2waatoZF17/XXOdeqsalapqOiIPtpT/YUQtunv3xdC/C6E8C060wpPpUqVGDx4MDVr1ixpU0o1C4f6UcHKFICdp3V4Kgt+R3mNvQ3ftQXNrSK0TsHUyJSR7iMZ0mgI2/pv0wYJ3k+4j9dyL+YdmEdcsu5PPo6LFlL3r42YpifpujR8eJHYraJS3tDH5fYDKaVGCBEEdAaWAV8XjVmGY9euXRw/frykzSj1LBmlKNyO+DGCXgv3kJSSR/p3ywow9RF4DVL2P20In9SHu2eKwVKFoFpBWVx1l51YRuCqQIZtGkZMUv7KxkIIzOvVo94/ir5WwpGjqtChiooO6DNopKa/Pg8sllL+CZgZ3iTD0q9fPxYsWFDSZpR6fJ0rMqOXBwBRVx7S8P2/+e1gPhIcvRZBJ8Wdldg78FUT2PEJ/PsVHPsV0vIYeAxAZlfdOvZ1ADhy5wjNf27OF4e+0GnaSpiYYN1ayQx44933kEVss4pKWUefQeOaEOIbYACwSQhhrmf5EqFatWrcuGGYXA7lneHNXDg1owuu1RSZkP+tPYLP9C2kpuXy5WtkDC0mZn3qCJsJW96D316E6RXhws4it9va1Jr1vdZzePhh6tkrwYvfH/ser+Ve7LuxL9/yNaZMAeDxX39xytsHmZqaTwkVlWcXfb70BwCbgS5SyodAJeDtIrHKgDg5OXHlypWSNqPMYGFqzOY3W7NmnJLk6GFcMvXe3ZT/U0efb+DdGxByBSYceHJ8WXdY2KwILX6CiZEJ63qtY/eg3VSzUoIXx24Zi+cyT64m5W6/aa1a1AtNT0KZnMzDtWvVgUNFJRf0kUaPA84BnYUQrwJVpZSlPs+ms7OzOmgUgMC6lTn30XM4VbIElKeO2iF/8SguOfdCZlZgYQdVGsKUhzA6/c/jzkm4GlkMVivYm9sT2pqAWY0AACAASURBVD+U+cFPAvrm3piL5zJPPJd5MmjjIJYfX57VdMdaOP/4AwA3p07jlLsHtz+bz6MNG1QZGhWVTOjjPfU68BNKiteqwEohxMSiMiy9zV5CiO+EEH8KIToVpA4nJyfu3bunSokUAGMjwa5J7fh2+BPhv86f6zjdJAQ4B0L/dK2q79vDqoFw5zTE66B3ZQA6uHTg2MhjfNz6Y5pZN6OSRSUqW1Tm+L3jfBL5CZ7LPIm6HaUdFKxbtMBl1Sosvb0BuPftt1x/exKnGjUm/sgRUp7x3CwqKqDf9NSLQKCU8kMp5YdAM2Csvg0KIX4UQtwWQvz31PEuQohoIcRZIUQIgJRynZRyLDAKGKhvWwCDBw9m586dmJqaFqS4CtDJvToX5zwPwM3HCXwRqoeXlHsv8OgLwhhO/wMLm8JcFwidBhd2QXJ8/nUUkq51ujLUYSg7Bu4gfGA4v3T7RXtu+N/D8Vruxfpz60lIScDKz5faa1bTMPIA9Tb/o73u4sBBnGnegtvz5hW5vSoqpRl9Bg3BEw8q0t+LArS5FOiSpWIhjIGFKKq5jYHBQojGmS55P/283tSpU4dWrVqpg4YBmN3HE4D5oadp9bEeGf36/QjvXoMBy6F+B+XY7s9gWTeYVb0ILM2bRpUbcXTEURa0XUAVS0Wf7L3d79H0p6Z4LvNk8s7JXEi+iZmLC41OncTp++9xePVVAO59/wNn2gRzc/oM0mJji912FZWSRp98Gv8HjAT+QBksegJLpZSf692oELWBjVJKj/T95sBUKWXn9P306DHmpG9bpZShudQ1DhgHUK1aNf/Vq1cDTxKMpKSksHPnTpycnGjQoIG+ppZ5DJ1w5tCtFBYcVuIZ2jmbMKKxud51mCTHYB17Cd8oJb/FKdfXuFmjvcFszIm8+uFm8k3+1fxLuCYcSdb/h3drvEsNMyVi3uTKFSp+/gVGmQaLZBcX7v/v/8Cs1Hufq8mH0lH7QaHIkzABCCH8gKD03V1SysP6GppeT22yDhr9ULyyxqTvDwcCgdMoA9UBIEpKmWcw4dNJmIKDg0lNTcXKyoo333yTOXPm5FW8XFIUCWd2nbnD8B8iAKjz/+ydd3hO5/vAP+d9syOJkZ0Qggyb2JsSsUcoUmqUmqVaqy3V8qVFaX9t1aiiRm2lVo0QI0bslYgQZCGLLJnv+/z+OAS1kkjyJpzPdZ3rynvOeZ5zn/s6J/d5nucelqYs6e9BZZs8VPNLjIL5ck1vvo6XXXgLiJzq4WHmQzZe28jii4tJykgCwFjPmA2dNlDeojwgF3i6M2UKSft9EY/Sj5jUq4fIysL6888w9vBAkvIyCC9YlOJDMooeZAq8CJMkSUZAS6AV0AJo+WhffvCiN0wIIX4WQngIIYa/zmC8DLVaTcWKFbl27dobiqjwmGaVrTg3tS01HC24GZtC2x8Ps+l1LrkvwtwezOSveP5nAw907+Vmom/CgKoDONb3GEs9lwKQmpVK562d8VjlwVdHv0IyNsZh/nxcz57BuFYtjOt68PDUKVLPneN2v/5cda9CSLPmaBISdHw3Cgr5T26mpzYAScDqR7v6AqWEEL1yfdEcTk8JIb7LYX+vrBE+ZcoUIiMjWb58eW5FLfYU9FD8fHQWP52Vp6uczFVMaWiEvirnX9mSNpMWh3tm/9ZK+oSX7cqDktW4Xzr/Upu9iR6up13HP8mf0w/lUayVnhVdSnWhlkmtJycJgX5oKEYnTmBy5Gj27tgZ09E8quuia5RpGRlFDzJ5nZ5CCJGjDbiQk3057Ks8cPmp33pAKFABOTXJBaBqbvv18PAQjzl48GD23xMmTBCGhoYiKytLvGs8rYeC4n5KuujyyxHhNGmHaPydr7gZk5y7DjIeChHwuxDTzJ/f9k3LFxnzQw+pmani+5Pfi2orqolqK6qJpmubijvJd547T5OaKkJ7eItAVzcR6OomQtp6itTg4De+/ptSGM9CcUDRg8zTegBOixz+n82N99Q5SZKyQ3slSWoA5LBW6BMkSVoLHEcuHxshSdJHQogsYDRyxHkQsEEIkW9ZBl1cXEhPT1eC/AqIkiYGbBvdlAGNnIh8kErLH/wYuvI0aZk5jKrWN4Z6Q+R0JFOiYfAeqPuRfOzoj3Bu9avbFxJGekZMqj+JfT33UcuqFg/SH9B2U1t2hO545jyVkREVNm/CevznAGSGhXGzS1eut/Uk5pdf0ablb7lcBYXCJDfTU0GAK/C4Yk055H/wWuT1B50VrXjd9FRSUhIpKSlYW1ujUhX5dFn5SmEPxQ9HZLL8cgYCcCwhMa2xca6mq57GKtqfqoFzAIi09yLUeQAaPZM89VUQerj08BJLYuRqhnVN6tKgRAPcjN2eO6/E5i0YXL2K/lMfLbHTvkZjZ5ev8rwOZVpGRtGDTIF7T0mS5PSq40KI2znqqAB5kffUu46u9NB2/iFCouUU5T4NyjGtcxUM9fLgHXVkHvhOf/I7j15WBaWH24m3+e7kd/hHyYNuS2NLxtYZS7dK3Z47NysmhvBhw0kLDARAz94O03r1sPr8c/StrfNdtv+ivBMyih5kCtx7SjxbH/y5LW9iFx7Lli1jyZIluhbjnWHvuOZ82cENA7WKv06GUeXrPfywJ5hr95Jy11Gzz+HLKFDpyb+nl4a7l/Jf4DziZO7EoraLONDrAL1dexObGstU/6lU/7M6x6OOP5O3Ss/KigpbNmM5ciSSiQnahEQStv3D9eYtCHJzJ+LTcYR9/DEpJwOUKSyFIkuu4jSKKq+bngKYPHkyMTEx/PHHHzqSUjfoeiguhOBQRBYrrjwpp1reXEWVMmq6VdLHQJ2zqStJm0Wt819hkXg1e98tp96EleuJVv36wLrC0kN8VjwL7i0gOis6e18783Z4lfRCT9J77nxj3wOYHDyIOiEBKfPZZJBCkoib8hUaB4d8kU3Xz0JRQdGDTKEE9xV1XjU99eWXXzJ37lySk5MxNMx9FHNxpagMxTM1Wvyvx7LQ7wYnb8Zn75/bswa96pbNeUenl8PhuZAY+ez+Qf+CU6OXNitsPdx4cIPfzv/GsahjJGfK03SeTp7Mbj4bPdXzxuMxD0+dImHnTh6sW5+9z7iuB5KkotyyP5DeIB1OUXkWdI2iB5kCn54q7tSsWZOsrCyCgoJ0Lco7ib5aRUtXa9YPa8SNWR2Y0lGOBJ+w6SLlJ++kw/8dISH1FWnXH1N3EHwWCAN3QeupYFxa3r/cCzQ5aF9IVCxZkXkt53G0z1Em1JXLzuy9vZfaq2oz+chkrsS92DnQpF497L75BverQdh+8w16traknj7Dw1OnuFq9BkkHDhbmbSgoPEduIsIlSZL6SZL09aPf5SRJql9wouUvderUAeDUqVOvOVOhoFGrJIY0c+b81235uLkzZUwNCLyTSM1v9/LrgRDSs3Lgqlu+CTQfD5NugtOjzDYzLCE+tGCFzyVqlZoPq37Iuf7nGFBlAAA7Q3fSZ0cfqv9Znbab2jI7YDaJGYnPtS3VpzeV/Q7iduUy6pIlAYgYOZJgj7o8PHtOKRSloBNy4z21ENm9trUQwl2SpFLAXiFEvYIUMCfkZE1DCIG3tzedOnVi8ODBOpK08Cku87fbb2SwOeTJSKFVWT0+cDdALwfuugbp8TQ6/hEScn3vQPdxRNu0fOacoqSHe5n32Juwl8upl3mofVLnpbxBeZqbNcdS3xInAydU0rPfdHrh4ZSZOeu5/tLqepAwZMhrr1uUdKBLFD3IFIbL7VkhRB1Jks4JIWo/2ndBCFEzr0LnN69zuU1PT3+n1jOgeM3fpmZo+PtcJD/7hnA3MY3Spgb89kEdGjqXyVkHWz6Gi4/WAup/DHUGgG01oOjqQQjBFP8p/HPjn+eOuZZypaFdQ6pbVadV2VYYqA3QpqWR7HeItMBA0q5cIcX/SXytUY0a2E3/FiO352NFoOjqoLBR9CCT1zWNl6/IPU/mo7oX4tFFrODRp10x4V0zGMUNYwM1Pg3K0bd+WaZsvcyak2H0WXKCz9q6MLpVJVSvG3X0WAKp9yFkLwQskTeQqwcKi4K/gTwgSRIzm87kf03+x72H90hIT+BA2AE2hWwi+H4wwfeDs89t4diCgVUHUterHeZe7QBIv3GD0M5dQKsl7eJFbnbrjuOCXzF7r2BTzSu8u+RmIfxn5Foa1pIkzQSOAs+PlYsw4eHhtG3blj179uhaFIVXIEkSM7tXZ+PwRpQvY8L8fddw/nIXH688TcT915Tt/WAjTA6DPn892bdxAC0PdYMFDeFeYMEKn0ckScLW1BbX0q6MqDUC316+nO13lh3ddzC2zljMDcw5FHGIQXsGUf3P6nhu8mTb9W0YODvjHngF96tBWE8YD0DEqNEEubkTv3oNyYcO6fjOFN42cltPww14/AnjK4S4+qrzC5vXTU+lpqZSqlQpRo0axbx583QgYeFT3IfiaZkavt52mQ2nn029XrNsSdpXs2VYc+dX1664fVyuTZ7+nzTlbb6Fpp8WgMQFR2BcIEsvLeVA2AE04ski+HfNvqOTcycAHmzaxJ0pU59rW2b4MAJr1KBl69aFJm9Rpbi/E/lFXqencrOm0Qv4VwiRJEnSVKA28D8hxNm8iZx/5GQh/DHjx4/n/v3770yQ39uy6CeE4NQ9DftvZxKbKohPk59bK2OJ/lUMqGH16pnW5KQkXBP8cA5diVr7JNAw1ciGBAt37tm04H7pOgV6D/lJkiaJFbEruJb2pE5MvzL9qGNaB31JHyklBfX9+5ivXIV+mJwuTmtgQEqnTiAEqc2aIkzylseruPO2vBNvSmEshF8UQtSQJKkp8rTUPOBLIUSDPMqc7+Qk99TcuXOZOHEiUVFR2BVywjhd8LZ+VaVlapi5M4hVJ+QMNi42JfCu40inmvY4lDR+7vxn9BATDNtGQ0YKRP8nXsK9i1zLvAhW3nsR4UnhzDs9D98w3+x9vV174+Pmg3NJZwCyYmMJadYc/vOul2jdGrvp36JnaVmoMuuat/WdyC2FEdz3eDzcEVgkhNiGXPuiWOHp6QnA3r17dSyJwptgpK9mRrdq+E9uzSetK3E77iHf7b5Kk+8P0Hb+IQYuD2D3pTtkaV7gq2HlCkP2wchjMO0BDD0ADh7ysaB/4NuSsGUYhOyXS9IW4awJZc3K8lOrnzja5yhDqw9FX6XP+uD1dN3WlXEHx5GhyUDP0hK3wCtE//QjLgEnKf3I5Tz5wAFCmjYjyM2dyIkTST6a60oHCu8guRlp7AAigTaAB5AKBBQnl1sArVaLt7c3gwYNokuXLoUsYeHzrnxVZWq0HLgazcnQeP65EElssjwFZaCnomwpY1rZZjKx93sY6L3iOyk2BDYNenFCxA+3gXPLApE9P9EKLSfunODPK39yLOoYAJ2cO9HArgHm4ea0bvVkTSNuxQrur1pNZuSTlCzGHh44rVqJ9BaXEHhX3onXURhrGiaAF3BJCBEiSZIdUF0IUWQ+2ZXU6M/zruohKS2T3ZfuMm9fMPcS5XK0xvqyS++gJuVxLPWK+fzMNAg7BvE3Yf+3zy6iDzsCdjorHZMrNgRvYMH5BcSnPcn11dapLeM8xlHW7Em+L6HRkLB9O3cmfyHv0NOjst/Bt3ba6l19J/5LgRuN4kBujEZCQgLJyck45FMG0aKK8oJAYlomP2/2w++eAdcf1fhwtjKlUw17+jUoh7W50as7uLAO/h72/P7+f0O5RnLlwSLMw8yHbAjewP+d+T+yyALAVN+UjhU6MqHeBIz05PvXpqURXOtJXXbTZs2w+9//0Lcp+FofhYnyTsgUmNGQJCmJRwF9/z2EXLHPPHei5j+58Z4C0Gg0eHt706RJEyZMmFCIkhY+iqeITHJyMqampoQmaPGPyuJAWFb2sfLmKvq6GeBa+tXFnRwidmKYHku58C3P7M/Qt+Bc7e9INSnaHyDJyckkGiTil+TH8eTjAJiqTKlqXBVPC09s9G0gIwOzjRsxOXI0u52QJDTW1qQ2akhqy5YIo9cY2SKO8k7IKKnRyd1Iw8fHhz179hAVFfVWR4orX1Uy/9VDSnoWvlej+ePoTS6EPwCgpqMFo1pVopWbNfrq18zpx92A675yZcHku/K+iu9B9V5Q6T0wsYQiti7wtA7kOieHmHNqDuFJchlahxIO9HbtjbeLN2Z6JUjav5/kg34k/P33M/1Y9PTGZvJk1MX0H6/yTsgUSmp0SZJKSZJUX5Kk5o+33ItaNOjfvz/x8fHs3LlT16Io6ABTQz261LRn26gm+E9ujUNJYy5EJPDxqjNU/mo3Hyw9wbmw+y/voExFaPAxjA8Gz/+Bvinc8IWtw+GHyjC9FKzoBEE7Cu+mcoEkSbQs25JdPXaxqfMmvMp7EZkcyfwz82mytgk/nfs/jN9rhf13s3C/GoRbUCClPvgAgIRNm7lWtx43e/Yi6aCSqv1dIzcL4UOAsYAjcB5oCBwXQhSZENPcjDQ0Gg3lypXDw8ODf/55Plnc24LyVSWTEz3cjE3h6PVYNp4O52LEk8VvnwblmNmt2qsjzwFS4iBkD0Sdh6izEPFUGv7KnvD+KtDX3dTO63SQrklnY/BGZp+anb2vqUNTJtabSAWLCoA8Qon/4w9ifl2AeFSS1qRuXcw7dsCie3dUxWDqSnknZApjpDEWqAfcFkK0Qo4Ij8mlnEUGtVpN//792b17N/Hx8a9voPDWU8HSlP4NnfhndFMOT2hFh+q2APx1MoyOPx9l85kI0jJfUcPCtAzU8oEOc2DIfhgfAo1Gg0pfTqI401ZOa1JEMVQb0q9KP871P8eEuhMoZ1aOo5FH6bK1Cx22dGBX6C6ytFmUGTIEt/PncFqzGpMGDXh4+jR3v51OcK3aRH3xJdqHr8kPplCsyY3RSBNCpAFIkmT4KO+Ua8GIVTh88sknXL58mdKlS+taFIUiRrkyJvz2gQehszowoZ0rgXcS+XzjBdym/suYtec4eDUajfY1o/QS1tBuJnwZBQ2GA0KuMPiNBfzhCXcuFMq95BY9lR4fVv2QnT12srjNYupY1yE8KZxJRyZRZ3UdBu8ZzPYb2zGuUwenP1fgciqAku+/D0DC338TXMeDO1O/JuXECdJv3uRtWjdVyF1q9AhJkkoCW4F9kiTdB6IKRqzC4W13t1V4c1QqiVGtKjG4SQX+PhfJgav3+OdCFP9ckB/9hs6laeNuQ486jpQ2fUmCBD0DaD8b6g6GQ7PldY7wk7C4OZSpBM6twL0z2NcGI507Iz5DY4fGNHZoTLomndWBqwm4G8CxqGOcunuKL49+yXfNvqNjhY7YTf8W22lfEz50KA9Pn+HBxo082Lgxux+H//s/zNt56vBOFPKLPHlPSZLUArBATmCY8brzC5rcutw+TWJiIvPnz6d169Y0b15s1/VfiuJeKJOfeohP07L3ViZ3UwS3ErU8SJffoQrmKjzL61PHWo2h3qvXP0yTb+N0ewPWMUef2R9Xug7BrmPIMCyVL7I+TX7pIE2bxs4HOzn78CyJmkQMJUP6lelHTZOa2es+6shI1HHxmPruxyBYTqr4sEkTknt0R5iavrEMb4LyTsgUmMutJEnlhBBhby5iwZOXiHCNRoOLiwu2trb4+799uXeURT+ZgtKDEIKdl+6w/lQ4R0Jis/fXdLRgcNMKdKxuh96r3He1Wjlp4s3DsOfLJ/ubfgYNR0IJq3yTNb91kKnNZMXlFay9upaY1BjsTe2xNbWlo3NHvCt7o1bJcS/poaHc9vkAzQPZtdmsvRcWXbpg1qpVvsmSG5R3QqYgF8K3Pv5DkqTNeZKuCKNWqxk7dizHjh3j8OHDuhZHoZghSRKdatiz6qMGXP62Hf/rVg3vOo5ciEhg7LrzuH/9L90W+LPo0A0SUjOf70ClAtvq0GiUnDyx159QwhaOzocfKsEsR1jtDbeK3geNvkqfoTWGsrPHTr5s8CVJmUmcjT7LjBMzaLquKVP9p3Ip5hKGzs5U9j+K9YTxqMzNSdr9LxEjRhLk5k7yoUPKmkcxIycjjadrgmf/XRTJa+6phw8f4uzsTJUqVThw4EABSlj4KF9VMoWth7RMDXuu3OXUrXgOXYshPD4VfbVECxdrJnm5UtnG7OWNhYBzq+DCenhwGxLCnxxzqAv1PgKnJlDKKVcyFYYOMjQZ7Lq5i+8DviclMwWAzs6dmdZ4GoZqOYg2Izyc0K7dEI+8rAwrV8aiW1dK9emDqhCmrpR3QqYga4SLl/z91mBiYsLkyZMZN24chw4dokWLFroWSaGYY6SvpmstB7rWkp0t/K/HMuffq+wPusf+oHvoqSR61XWkcw17GlUs82wMiCRBnQ/lDSD8FBz7WU7bHnla3gDsaspTWC5eYGRRJGqAGKgN6FapG90qdeNWwi3mnp7L9tDtbA/dTh3rOkxrNA3nss64nT1Dxu3bxPy6gMTt24me+wPRc3/A0NUVy9GjMG/bVte3ovAScmI0akiSlIica8r40d9QhHJP5QfDhg3jwYMHVK1aVdeiKLyFNKlkybbRTQmNSeavk2EE3klkbUA4awPCqeFowcDG5fGqZouJwQteybL1oPcq+e/U+3BioeyFdefCs4kUP9oHZesXzg3lgPIW5fm19a9svb6VzSGbORt9lq7butKtUjc+qf0J1k5OOMydg/2smcStWEHMvPmkBwcT+ckYxLwfsOjYUde3oPACcjI91Rg4IYR4QTWbooWSGv15FD3IFEU9JKZl8seRmyz3v0limpxAsX750vRr5ETH6naoVa8ZOSRHQ+A2uLQJwk/I+6yrQs0+UG8IGDyb/l3XOrh+/zo/nPkB/0h5faZbpW6MqT0GK5Mni/1pgYHcfL83ZGWh7+iI06qV6OdzhU1d66GoUJAL4R8CpyVJWidJ0kBJkmzzLmbR58SJE3Tu3JnU1FRdi6LwlmNupM+4ti6c/9qTOd41qOZgTsCteMasPUelr3bx1d+XOHwthswXVR8EOXiw/lD4aA8MOwyO9WVPrH1TYZYdLGkJD4tOtoNKpSqxqM0i1nVch1d5L7Ze30rbTW0ZuX8ku2/uRgiBUZUqVNwl54PLjIjgeqvWhHbvQVpgoI6lV3jMa42GEGK4EKIO8A1QClghSdJxSZJmPUpa+Op80sWMtLQ0duzYwQ8//KBrURTeEVQqiffrlWXHJ8048cV7jH2vMg4ljVlzMowPlwXgNlX2wFoXEPbi8rUgr28M2QdfRECXX+R9UedgTgX4qTpsG1VkytZWtazK3BZzWdtxLT0q9yAwLpCJhyfSZ2cffMN80SvriFtQIPbz5HcwPSiImz28CRs2jMw7d3QsvUJeg/uMgVZAe6BRToc1BU1+TU/16tWLHTt2cOHCBVxcXPJRwsJHGYrLFEc9JKVl4hsUjV9wNIdDYolPycDMUA8nSxM6VLejeWUrqjlYvLixViNPWx3/5dnyta2mQJOxcpR6ESFLm8WWkC3MOTWHdE06pY1K82GVDxlYdSBqlZq0oCBudu+RfX7JPr2x+fJLVAZ5u4fi+CwUBDqp3CdJ0iAhxPI8d5BPvElE+IuIi4tj0KBBODk58dNPP6FWF9/BlBL9KlPc9ZClFZyN1uAfmcXdFC33HsrvbUULFaWNJTys9Whgp35xJl4hcA+ah030kexdGfoWpJiW5a7te9yzLRqJqtO16fgn+/P3fbl+h62+LTVNatLarDUmKmOMAgIosXUb6vtyyvqkrl146OkJuXw/i/uzkF/opAiTJElhQohyee4gn8nPhfCVK1cyYMAAVq1aRb9+/fJJwsJH+aqSedv0cDM2hXWnwljuf4uMLHnKSq2SaF7Zkpau1rSvbou12bNpyg/77qV5iVsQEwQXN0CGXPqWErZQ431oMAwsHAv5Tp5HK7RsDtnMvNPzSMlMwdzAnDG1x9ClUheMVIbELviN2AULss8v5dMXy1Gj0CtTJkf9v23PQl4pyHKvF192CHARQhSZsnf5aTSEEGzcuJEePXqgp5ebvI5FC+UFkXmb9ZCcnsXf5yK5GP6Av89FkvUo+24NRwtau1nT3MWKWo4lOXz40LM6SE+GbSMhaDs8do40Li17XzUaVSQMyPno80w4PIG7KXJ1RPfS7nhX9qaHc1eSt2zjwaZNpF2+jKSvT9mlSzFt8HqX47f5WcgNBWk07gHtgP+WMZOAY0II+1xLW0AUlMvt3bt3MTc3x8TE5PUnFzGUF0TmXdFDRpaWs2H3ORISw6YzEdxLTM8+5lVej8k9m1Le8gVR16GH4MRvcO3fJ/sajZbXP0pYF4LkL0crtGwJ2cKxqGPsu70ve3/3St0ZWn0opc7fJGLsp4jUVIyqVcP+++8wrFTppf29K8/C6yjIiPAdQAkhxPn/HpAkyS8XMhZLEhISqF27Np07d2bJkiW6FkdB4ZUY6Klo6FyGhs5lGO/pyp2ENPYFylHo/4bE8u8PftQsW5JeHo54VrV5MoXl3ELetBoI2QeH58LxX+Wt9RRoMg7UuhlxqyQVPV160tOlJ5maTNYEreFK3BX+vv43u27uYkDVAXywfSMJ474g7dIlQjt1RjIwwMzTE8uRIzF0rqATud9WcuJy+5EQ4uhLjvnkv0hFCwsLCwYNGsTvv//O77//rmtxFBRyjCRJ2Jc0ZkDj8qz6qAGzmxkzqEl5LoQ/YMrWy9Sf6UvXX4/y0/5rpKTLwYWo1ODqBUN9oeN8OZnigf/BjDKw5n2IPCNn5tUR+mp9BlYbyNwWc9ncZTM1rWqy5OISWvh2Y9dXLbDf8Tcle/UESSJxxw5CO3Tg9oCBiKwsncn8tpGbyn3vLNOnT8fLy4uRI0dy8OBBXYujoJAnbExVTOtclWv/a8+ygXVp427DhYgEftofQs1v9/LhsgAOX3uqgnO9j2DYEejwAxiYyfXPf28tGxDf6ZCo25gJl1Iu/NHuD1a2X0lj+8YsurCItic/ZFF7FTYndJEYVAAAIABJREFUD1Jh2zb07O14ePIkN7v3IOPWLZ3K+7ZQfFd4CxE9PT3WrVtH48aN8fb2JiAggEqvmDNVUCjKGOipaO1mQ2s3G4QQnLl9n7/PRbL78l0+XBaAq40ZLVyteL+uI5WszeSo8/pDITEKjv0KAUvgyDw4+iO4dZLrort46SxhYm3r2ixuu5jDEYdZHbiaLSFb2BW6i4b2DZm2YwPahX8S9/tSbni1x3rSJKhQXidyvi0oI40cYmFhwfbt2+nUqRM2Nja6FkdBIV+QJIm65Uszs3t1jk5qxZSO7sQkp7PkcCht5h9m1JqzHLoWI0eim9uD1yyYGgOD90LF1nLm3bV9YL67PI0Vd0Nn99LcsTlLPJewruM6yhiXwS/cj/c2vce2tuZYLJwPQPTs2ViNn0DUpEloM3RedLRYkteIcFshxN0CkOeNKMyEhYmJiWRlZVG6dOkCu0Z+oHiKyCh6yJ0OQmOSWXr0Jn+dlIt2mhqoaVTRkm617Wlf7alkiimxcHAm3DwCcSGPWkvQ5hs5aaKh7oLoguKC+OnsTxyLOoaERG+rdrzvD5l/70LSatEvWxbbb6ZRokkTncmoSwrSe+pF7ALq5LFtsUcIQadOnUhLS2P//v2Ym78V2eEVFLJxtirBrO7V+bRNZY7fiOPY9Tj+vXKX/UH3sDILpEM1W95zt8HDqRSmnX6UG0Wchh2fymlL9k+Tt7ofyV5ZZRuCWeGO0N3LuLOozSIC4wLZFLKJ9dc2s94F3L+pxsg7zlgv3Er4R0Nw+OVnpX5HLsir0dB9tRcdIkkSEyZMoEePHnTo0IGdO3diYfGSHEAKCsUYazOj7GJSM7tXY8+Ve2w8E8760+H8efw2JgZqOla3o3e9stQu54F6+FHQZMGRH+D6fjj9h7wBlHaGBiPkBXZV4aTmkSSJqpZVqWpZlcHVBrM6cDWbgzczumQgnp9UZOCfd4j8ZAyx7u6UGfKRUsMjB+TVaLzzvqedO3dm7dq19O3bl9atW7Nnzx4sLS11LZaCQoGhp1bRsYYdHWvYkZKeRcDNeHZcvMM/F6LYeCYCEwM19cqXZsx7lfBoORlaToaMFAg/CaF+cGYF7J4gb471oPr7UL0nmBTOFG9Zs7J80eAL6j2sx12bu/x2/jf8Pk7nMz8L6pwOIurz8cSvXInNpMmY1CmyVa11Tp4WwoUQv+W3IMWRnj17sm3bNgIDAxkwYICuxVFQKDRMDfVo5WbNvPdrEvBVG37qXYvmla04fSse74XHaffjYVaduE1shp68YN52Ooy/Dp4zwaIcRJySjcecCvCHJ0Q9FztcYOhJevSr0o/DfQ4zquFnrOpekkGfqjnYzpa0Cxe57ePD7YGD0Dx4UGgyFScUl9s3pEOHDuzduxd7+yKTTUVBoVCxMNanW20HutV2IOFhJnP2XGXXpTtM3XqZqVsvU79CaQY3KU8bdxv0Go+GxqMfRZ7vhe2fyiORJS3A0hXq9IdKbcDavcDl1lPpMbjaYPpX6c+ma5tYYbmCdZXVfLPVELsTJ7jepi3O27ai7+BQ4LIUJ4qsy60kSc6SJP0hSdImXcvyOpo1a0bFihURQjBixAi2bduma5EUFHSChYk+M7tX58yUtqwd2pD+DZ0IjUlm+Oqz1J25n2+3X+FeYtqjyPP2MD4YRp6AWv3gYSzsnQK/NYRvS8HB7wql8qC+Sp++bn3Z2WMnw9/7kskDVaxqpSIrJZmQjh2I/vEn3iQb+NtGro2GJEnukiS1ffS3cS7bLpMkKVqSpMv/2e8lSVKwJEnXJUmaDCCECBVCfJRb+XRJUlISZ86coXv37sydO1d50BTeWVQqiUYVyzCjWzWOf/EeCz+oQ8MKZVh1/DZNvj/A4BWn2HAqnITUTHlU0W0BTAyFMeeg8RiwcodD38vTVzvHQ3xogcusp9LDx92Hfb32U3bYaOYMsyRVm0Hc4sUENWxAkp9fgctQHMjLSGMhUE2SpH+AlZIkTc9F2xWA19M7HpWLXYBcBbAK0FeSpCp5kEvnmJub4+fnh7e3NxMnTsTb25sHyryowjuOvlpF++p2LOrvwb7PWjCwcXmuRCUwcfNF6szYx7j15zl4NRqNVsgeVp4zYOQx+PAfsK8Np36Hn2vD/Cpw40CBl621MLRgRK0RLBq9l4AVoznjboCUkETE8BFc6euN9uHDAr1+USfXwX2SJP0ihPhEkqSlQoghkiT9KoQYnYv25YEdQohqj343Ar4RQrR79PsLACHEd49+bxJC9HxFfx8DHwPY2Nh4rFu3DtBtda7HtTiWLFlC5cqV+e23315cUa0QUKqUySh6KFo60ArBjQdaDoRncvaehnQNmOhBWTMVbZ30qWOjRvXonTFLDKFc2GasYo8DkGZoRYRjZ6LsvdCqc1/OJ7d6SNIkcTR0M90XBmDxUJBmrE9qx05kvNdWZ6lT8oNCq9wnSdIp4CTgAnwODBdCjMpF+/I8azR6Al5CiCGPfvcHGgDTgJlAW2DpYyPyKgozIjwn+Pv7k5KSgqenJxqNBkmSUKkKdxmpKOihKKDooejqIC1Tw8Gr0fwVEMbpW/dJzdRQ2tSA1m7WtK1iQ/PKVhgbqCEtES5tgBMLIe46GFpAo5HQ7HNQ6+f4ennVQ3JGMssWfIzb5nOUjYWU5rWpPX8J6iJiiHNLoUWECyHqSZLkCHgAvQCn3PbxH15kqoUQIg4Y/oZ965QmT6Un+OGHH9i7dy/Lly+nXLkiUyFXQUHnGOmraV/djvbV7cjUaPn7XCQHgqLZdekOmx7FfzSoUJrx7VypWm+IHGUevBsOzAC/7+DUUjnmo0YveTqrgChhUIIx4/7i5PvHOTFpDA0Pn+NCs4bETu5Pm/cnoJKKrF9RvvJGNcLzdMFcTk/lsM/OQGcHB4ehq1evBorWUBxg9+7d/PLLL6hUKkaPHk27du0KZcqqqOlBVyh6KH46yNAIAuM0HAjL4mKsBgArY4kqZdQ0tNOjUkkJh9gjlA3filmyvFCeamTLzQofEG3T/KX95oceNELDjVMbqbXuCGYPtQS6mmLkMwpTm+JT8Cmv01MIIXK0Acf/89sMqJ3T9k+1Kw9cfuq3HhAKVAAMgAtA1dz2K4TAw8NDPObgwYOiqHHjxg3RrFkzAQhPT09x48aNAr9mUdSDLlD0ULx1cC8xVfx++IYY+ucp4T51t3CatENU+nKnGLw8QJwMjRMi9roQfnOEmGEtxDRzIZa0EuLYAiHibgih1T7TV37qITMuXhz81EcEurqJi+5u4uywfiIrMTHf+i9IntYDcFrk8P9sbsZThgCSJM1/ZGySgFxFhkuStBY4DrhKkhQhSdJHQogsYDSwBwgCNgghruSm3+KCs7MzBw8e5OeffyYgIIDY2Fhdi6SgUCywNjNiSDNnlnxYl4Cv2rDApw5965fjeGgc7y8+jteqSP4p+QHaT69A/Y8h6R7s+UL2uvqlDuyaUCAxH3qlS9HyxzXorf2Ne7aGGPmd5kqThkRs/ivfr1VUyPH0lCRJ54B2wGXARgghJEk6L4SoVZAC5oTiMD31Xx4+fIiJiQkAq1evxsXFhfr16+f7dYq6HgoLRQ9vpw7SswR+EVnsu51JbKrA2kSio7M+TezUlE4OoeSDCzhG7MAgMwGNyoD7pWpx2bYnWLnmvyzadK6cXUfDtScolQLXatmi5zMcU/OiWX+nMKanvIBrwE+PtpHAxZy2L4ytqE9PvYjk5GRRuXJlAQgvLy9x+fLlfO2/uOihoFH08HbrIEujFesDwkSLOQeE06Qdosn3vuLwtegnJ1zdJcTCpvLU1TRzIVb1ECL0cIHIcv3OFbFxZHsR6OomTtZyF/8sniQ0WVkFcq03ocCmpyRJaiRJkiSE+FcI4SKE+BRYD1QCilXEdlHE1NSUS5cuMW/ePE6cOEGNGjUYNmwYMTExr2+soKAAgFol8X69shz4vCW/fVAHtUqi/x8B9F58nDUnb5Pm7AnDj8DYi4SV7SFn3f2zE/z+HuyeBA/C8k2WirZV6LlgFyz+DpWeHpXmb+NYp+Y8uHAm366hS147PSVJ0iKgPvIo41/gX1HEqvYVx+mpF5GQkMDKlSvZs2cPy5cvx8rKCiHEG3lZFUc9FASKHt4tHaRlCXbezOTMvSyikgVmBtC6rD6tyuqhl/UQc2M9yoZvpcKttc+0i7Rvz63yfcg0KJkvcmgz0rm96zfq7r2GnhYeVHIkc8gotCXzp/83ocCD+yRJckNO9dEOsAAOIhsRfyGEJi9C5zdFLbgvryQkJGBhYYEQgs6dO1OrVi0+++yzPJWWLc56yE8UPbybOhBCcPR6LAv9bnDsRhwAzhYq+jRxoX01O8qWNJTTtF/bI2fdvfcoLV7NvtBiEpTOHxfa45d2cfaX6bQ+nEBqCX3sJ07GpldfnWWKgLwH9+XYe0oIcVUI8aMQwgtoDRxFDu47mXtxFV7F4yqAqampmJqaMnPmTCpUqMC0adOIi4vTsXQKCsUHSZJoVtmKv4Y2ZP9nLRjzXmXSNIJZu67SbM5BRq27gG9KBVKbT4ER/tD/bzlZ4oW18HMtWNYezr+5J1Sj6h0YtOAA57/sSpI6k/tfzyCocSMyo6Ly4S4Ll7wWYUoVQuwCzubUOinkHhMTE9avX8+FCxdo06YN06dPp1y5chw6dEjXoikoFDsqWZfgs7YuzGpqgt/4lgxuUoFDwTF89Odpas/Yy7j157lhXh9GnYChB+Rsu2HHYOsIWNQUbh97o+ub6JvQ98PvUW1YxJXKhmgfJBDc3ouYRYvQJCbm010WPG8UES5JUpgQQuc5Md6WNY3XcfPmTbZu3crw4cMxNjbmzJkzlChRAlfXl7sPvo16yAuKHhQdPOZpPWRoBNfuazh1V4N/VBZaAR0r6FPfTo+yZiokbSaOEdtxur0ePU0asWXqEV62Owklq76RDImaRHZfW0bLbdeocUugMTYmuWdP0po0zo9bzBEFtqYhSdLFlx0CXIQQuU8zWUC8LWsaOaVBgwYEBATQrFkzRo4cSY8ePTAwMHjmnHdBDzlB0YOig8e8TA/3EtOYviOQnRfvAOBuZ86gJuXpWccRVUo0+H4L59fIJzu3gvf/BCOLPMuh0WpYfHExR/5ZyLB9YBOTRSmfvthMnVooax0FuaZhA3wIdH7Bpkyw65B9+/Yxb948oqKi6Nu3L+XKlWP58uW6FktBoVhiY27EAp86+E9uzTedqyCEYOKmi/T5/QSXE40QXRfAhFBoMAJCD8KCBhCRdzdatUrNyFoj+ezjZXz7sQXHq6i5/9dabvV6n9RLl1/fgY7IidHYAZQQQtz+z3YL8CtQ6RReibm5OZ999hnXrl1j9+7dNGjQAFNTUwBiYmL4999/0WiKhGObgkKxwaGkMQObVGDXmGbM6FqVy5EJdPrlKL2XnCA4yQDafw9esyH1ASxtDX94QtiJPF+vnm091vXYzL8D3VndSsXDoCvc6tWLe999h9Bq8/HO8ofXGg0hxEdCiKMvOeaT/yIp5BaVSoWXlxfbtm3j/fffB2DFihW0b9+ePn368MUXX3D16lUdS6mgULxQqST6NyrP0Umt+aqDO1ciE+j0yxG+2x1EhOuHMO6KXMsj+iosawcrOsHdS3m6lo2pDas6rkavf0+Gj1QRWcue+D9XEj5seJHzsCr01OgFwbuyEJ4bMjIyOH78ODt37uTMmTNotVqqVavGjz/+iJ5ersuoFHve9ecBFB08Jq96iE/Tsu5qBgF3NagkqGujpkdlAxwM0ygXthmnsE0AxFg2JNj1E7L0c38NIQTbHmzDN2E/nx20oOFJOcliok9fUpu/PN17Xii0yn1FmXdtITwn+Pn54ebmxurVq7l9+za//PILAN9++y1VqlShY8eO2YkT32aU50HRwWPeVA+hMcmsPH6bdafCSM/SUtepFF1rOdDFWcL86P/g4nowtYJui6DSe3kqCbvu6jpmnpxJy3ulGRlQCi4HY/vNNEr16ZNnuf9LoVXuUyh+2NraMn78+OzfqampLFmyhKioKExNTenSpQu9e/fGy8sLQ8Mi4wynoFAkcbYqwTddqjKshTN/nQxj9+W7TNl6me8M1PSqO5bhXm2wPTQJ1niDqTV0+QVcvXJ1jT5ufahYsiLj/MYx1P4ei9KduPfd96BSUbJXL51Gkr8b9QkVnsHY2JiwsDB8fX3x8fFhz549dOvWjd9+k8ujpKWlkZqaqmMpFRSKNnYWxnzu6cq+cc3ZMKwRzV2sWHXiNg23mvKJ7UpiWswCBKztDau6w42DkIuZnXq29VjdfjUlTcswuv0d0kqacPfradz9+uuCu6kcoBiNdxS1Wk3r1q1ZsmQJd+/eZffu3fTt2xeAjRs3YmlpSY8ePfjzzz+V1CUKCq9AkiTqVyjNwn4eHJ3UiuEtKuJ7I4VmB5xZ33g71BsCEadhVTdY3iFXbrrlLcqztuNaXCs3ZOCARGKauPFg4ybuTp+uM8+qt2JNQ1kIfzl50UNISAg7d+7E39+f2NhYVCoV1apV4/vvv8fY2LiAJC1YlOdB0cFjCkMP0Q+1/H4xnZAHWipaqPjQJYvmSTspG/4PepoU7ti25lZ5H9KNrHLUn1ZoWRKzhMCUy8zzdcbxdAhpHh4kfDQYVHn79lcWwlEWwl/Em+hBCMGZM2fYtm0bQUFBbNoke4dMmDABrVaLl5cXzZo1w8jIKB8lLhiU50HRwWMKSw8arWDx4RssPhRKUlomQ5s7M6aBBaYnfoSAJSCpoeEIaD0V9F//DqVmpTJg9wCC4gJZeKUeZbafwKxtG+xmzUJtZpZr+Qo8y63Cu4ckSdStW5cZM2ZkGwyAW7du8euvv+Lp6Unp0qXp2LEjGzZs0KGkCgpFD7VKYmTLShya0JLutR1ZfCiUlgsD8as4AUafgard4fiv8HsriA15bX/Gesas8FqBh21dRlQ9RVTneiTt28/N7j0QGRmFcEcyitFQyDUbN24kPj6enTt3MmTIEK5fv87Zs2cBOT5kzJgxbNmyRVkLUVAASpoYMO/9mvw1tAHmRnoMXH6Kgdvjud3qZ+jwA0QHwq914eCs1y6Um+ibsLjtYt5zasOn1c4R1r8VmRER3Jk+vZDuRjEaCnnE1NSUDh068PPPPxMcHMyMGTMACA4OZtmyZXh7e2NlZUWtWrUYN24cwcHBOpZYQUG3NK5oya6xzRjXxoUToXG0nX+YKVENSRh4CCq0gEOzYUkLiA99ZT+GakPmtphLY/vGjHc8QnTbWiRs2kz8ylWFch+K0VDIF/T19QGoXr069+/fx9/fn+nTp1OmTBkWLVpEbGwsAP7+/kycOJHdu3eTlJSkS5EVFAodQz01Y9tUxvfzlnSv7cBfJ8NovOwuf1acj2jyGdy7An+0g8hXe1jpq/T5v1b/R0vHloytfYnkynbcmzWLpAMHC/weFKOhkO/o6+vTuHFjpkyZgq+vL/fv36dhw4YAnDt3jp9++okOHTpQqlQp6taty9ixY0lJSdGx1AoKhYdDSWNm96zBv582p6q9BdN2BON5qSUhHTeBJh2WtgHf6aDJfGkfRnpG/NDyBxqXa87ILjEIY0PuzpiBtoBjrBSjoVDgGBkZoVarARg9ejQPHjxg3759TJ48GTMzM7Zs2ZLtyjt16lT69evHokWLuHTpEtoimOVTQSG/cLEx46+hDfixd02S07Pw2pzKgip/oa3cDo7Mg99bw92Xp0k3VBsyu/lsTMxLs7KdAVl37hDx6acFKvNb4XKrxGm8nOKgB61Wi+qRr/lvv/2Gr68v8fFyojZTU1NatWrF559/DsjR6nlx8S0OeihoFB3IFFU9JGUI/rqazvEoDeVKSMx2PEa98N9RaTMJdR5ApEMnhEr9wraRGZH8dPcnJm7WUi34IQkfDSatXr1XXk+J00CJ03gRxVEPQghCQ0Px9/fH398fGxsbpk+fjhACW1tbLCwsqF+/fvZWq1at1xqS4qiH/EbRgUxR18OOi1FM3XqZlHQN45tYMCTqG1QRJ6FMJWj3Hbh4vrDdlpAtfHfwa1bO16C2sqSSry+q/1TyfBolTkPhrUGSJCpWrMiHH37I4sWLmf7InTAzM5Nx48ZRrVo1/Pz8GDt2LI0aNWLKlCmAPApZvnw5ly9fJisrS5e3oKCQZzrVsOef0U1pW8WGWYfv0zp+Ejdb/gwZKfBXLzgy/4Wuud0rdae1W0e2NZDQxMQSOe6zApFPyXKrUGwwMDBg8uTJ2b8jIyM5deoUzs7OAFy4cIHBgwcD8jpK9erVqV27NiNHjtSJvAoKeaVsaRMWfFCHnsHRfLXlEp77rJjcci2DQseh8v0WEiPl6oHqJ//CJUliWqNpeMdcxPnBHar7+pIWHIyRq2u+yqaMNBSKLQ4ODnTr1o0aNWoAUK9ePa5evcqqVasYOXIkJUqUYMOGDdnrIzt37qRq1ar069ePefPmceDAAe7fv6/LW1BQeCWtXK3ZMaYZTStZMsM3iiH635Pl0gFOLZVLzabEPnO+ib4J05tM57c2WWjVKqLnzM13mZSRhsJbg0qlwtXVFVdXV/r16wfI6yNCCA4fPoyJiQnOzs74+fmxZs2a7HbXrl2jcuXKBAQEcOvWLapVq4aLi8s7WeFQoehR2tSAZQPrsdz/FjN2BtLNbgQbW9bF2G86LGwMPuvBvnb2+fVs69GwVieOuP9DC39/kg4exKxVq3yTRxlpKLzVSJKU7ZnVqlUrtm/fTkREBNHR0ezZs4fZs2dnT28tX76c3r17U7VqVUxNTalZsyb9+/dHo9EAkJSUxNvkOKJQfJAkicFNK/Br3zoE3U2my9m6xPfcAmkJsKw9RJ1/5vwvG3zJ9s7WAMT8/Eu+yqIYDYV3EisrKzw9PZk4cWJ2DMmPP/7IuXPnWLVqFZ9++imOjo7cvHkz+/gHH3yAhYUFjRs3ZtiwYfzyyy/4+/vr8jYU3jE61rBj2cB6hN9/SMftcNt7F6jUcp2Op9KPmBuYM7r5JDY1kUgPCiJx9+58k0EZfysoPMLIyIhatWpRq1atFx738fGhXLlyXLp0iY0bN7JkyRIaN26cbTgGDBiASqXC3d0dNzc33N3dqVChgjLNpZCvtHCxYv3HjRiwPIAOa2NY1uQHGpz6FBa3hKEHwLISAG3KtWFm8xL09E8ifsWfmLdvny/XV55mBYUc0qdPH/r06QPIayV37twhISEh+3hcXBxnz55lxYoVz7RZu3YtADNnzqR8+fK4ubnh5uaGqalpocqv8PZQs2xJdo5pxsg1Z+njp2F2o995//IIWNEBeq+GsvXRV+vj6dKJYMf1uF66RFZsLHqWlm987bciuE+JCH85ih5kClMPycnJhIWFERYWhpWVFR4eHiQnJ9O1a9dn0qJYW1vzwQcf0KVLFzIyMggICMDR0RF7e3sMXhGU9SZyKc/C26WHdI3gj0vpBNzVMLrsLT55MAc9TRonGywkw7A08VnxrDo9jVkrsnjYvDlJPn2z2+Y1Ijzbu+Rt2Dw8PMRjDh48KBQUPTymKOghLS1NXLlyRWzatEnMmDFDfPDBB2Lr1q1CCCHOnz8vAAEISZKEk5OTaNOmTbbciYmJIigoSKSnp+f5+kVBB0WBt00PWRqtGL/hvHCatEOsW79KiGnmQvxaX4j0FCGEED+e/lHsb+QmLterK7QaTXa7p/UAnBY5/D+rTE8pKBQShoaGVKlShSpVqjx3zNXVlZMnTxISEvLM9thz6/Dhw3Tq1AmVSoWTkxOVK1emUqVKfPbZZ1SsWDE7S7Ay5fXuoVZJfO9dg7uJaUw6C1L5z3n/7jzYPAT6rGFkrZHMaLEBn7/vk7h7NxYdO77R9RSjoaBQBDAyMsrOpfUiateuzcqVKwkJCeH69euEhISwZs0aRowYAcCqVasYMWIE1tbWODs7U6FCBSpUqMDnn39O6dKlefjwYbYBUnj7UKskVgyqz+x/rzLxMDhad6Bx8E44+ycGHgOp2HMAD/b8BNOnY9ayJao3+LhQjIaCQjHA3t6e/v37P7NPPLUe2bBhQ2bOnMnNmzcJDQ3lxIkTbNiwgQkTJgDyIvzs2bNxcnKiQoUK2YZl/Pjx6Ovrk56ejoGBAZIkFep9KeQfapXEF+3d0FNJ9PfrywWzE5TYOxXKNsDLrSuTO/zMpE2JxC1bjtUno/N8HcVoKCgUU57+B/8iV+GsrKxsd982bdpw69YttFotN2/eZOvWraSlpWXn8ho6dCibN2+mXLlyODk5Ua5cOVxdXbNT0sfHx2NmZpZdoVGhaCJJEhO93LiXmE7381+wy+Qb9Be3wHbAP4gmdYnffwb12rWK0VBQUHiep+NDWrVqhSRJz6QET01NzTY8Xbp0wdLSktu3bxMWFsbZs2cpU6ZMttHw9vbm8OHD2NvbZxuV+vXr8+mjgj+3bt2iTJkymJmZFd4NKryU6V2r0utOIt5x09liOA29Vd1p32Y8Jyueov2ZeDLv3Mlz34rRUFB4R3lcLRGgZ8+e9OzZ85njmZlPSo2OGjWKZs2aERYWxu3btzl58iSJiYnZRqNNmzbcuHGDUqVKZY9W2rVrl51h+Pjx41haWuLg4ICJiUkh3N27jamhHr8PqMug5Vq846awTW8y7Y/9wdhKatqfySL50GGwtclT34rRUFBQeCFPT0W9yKg8vaYye/Zsbty4kW1Ubt68SXBwcPZ5LVu2JCMjA4BSpUrh6OjI4MGD+fTTT9FqtSxfvhwHBwccHR1xdHTEwsJCWV95QxxKGvPn4Pp0/iWDH1UfMy5hCaaO1YG4N6ojrhgNBQWFPPH0P3Vvb++XnieEYPfu3URERBAZGUlERAQRERHZI47Y2FiGDBnyTBtTU1NmzZrFmDFjePDgAXPmzMk2KA4ODtjb22NtbZ2dF0zhxdhZGLOwnweDl2vpqdqOZdpdQJ/0hHigfJ76VIyGgoJCgaJSqWhZsHdIAAALu0lEQVTduvVLj5cpU4Zbt25lG5PHxqVq1aoAhIeHM3fu3OeqMS5dupSPPvqIwMBAJk6ciJ2dHXZ2dtjb22NnZ0fjxo2xsrIq0HsrDtQrX5oVg+vx8dIJjEz9lkRjfcTtYKhVJ0/9KUZDQUFBp6jVapycnHBycnrh8erVq5OWlkZ0dHT2SCUqKoqmTZsCkJiYSGRkJKdPnyY6Ojp72mzPnj14enqybds2hgwZkm1QJEli7969jBw5EkdHR+Lj40lMTMTW1va1teaLKx5OpRnbuyPL/77AYNN9GJ87DV37vr7hC1CMhoKCQpFHrVZnjyTq1n02RVLDhg05d+4cILsZR0dHc+fOHSpXrgzIFR69vb25c+cOUVFR3Lp1i3379mUX6lqzZg1jxowB5PUWOzs7bG1tWbNmDba2tpw6dYrLly9jY2OTvVlbWxdIfrCCxKuaLRfv9kNv9z6yUlOQtHkL9lSMhoKCwluDnp4e9vb22NvbZ++rW7fuM4bGz8+PZs2aZa/JeHp68scff3Dnzp3s7e7du9lrLps2bWLOnDnPXSs5ORlTU1MWLlzIwYMHs42JjY0Ntra2dO7cGUmS0Gg0RWbtZUyLOvxdUoVVuAajtOg89aEYDQUFhXeOp/+JPy4R/DK++eYbhg8fzr1797K3uLi47DxfsbGxXLhwgXv37mWnyi9ZsmR2/XkfHx927dr1zEjFxcWF77//HpDdkTMzM7GyssLKyopSpUoVmJEx0tcjupSaWqFaMlPv56mPIms0JEkyBX4DMgA/IcSa1zRRUFBQyHeMjY2zc3m9iKlTpzJ16lSA7LWXp+usdO3aFTs7u2yDExwcTGxsbPbxCRMmPFMBUqVS0bJlS3x9fQGYPHkyCQkJ2UbFysqKSpUqZY+ecjuS0StlBUSREhaZ4zbPtM9TqzwiSdIyoBMQLYSo9tR+L+D/ADWwVAjxPdAD2CSE2C5J0npAMRoKCgpFGiMjI8qVK/fMPh8fH3x8fF7aZunSpURERBATE5O9Pe31dfbsWc6dO0dcXFz2In/Hjh3ZsWMHAE5OTiQnJz9jVJ4OrFy3bh0WFhZYWlpiZWWFiZ0zEMXdlOIxPbUC+BVY+XiHJElqYAHQFogATkmS9A/gCFx6dJqSnlNBQeGt5HElx5exd+9eQB5R3L9/n5iYmGdGFqNHjyYqKirb4Ny6dYvw8PDsNj4+Ps8EYvq0dqMhkJaWkid5C71ynyRJ5YEdj0cakiQ1Ar4R4v/bu/9Yq+s6juPPlyEopBASEw2kGpaOtlBX9MOi4apRg1xUtDG1uWhqOVPacrbJsFYGrubSFJVJWmKxNCBTm2kYCUiiCPgjMkMozYxYQD9gvvrj87l6vJxz7+feuN/zg/dj++5+f53P933eO+d87vf7+X4/H38kL1+ad90O7LS9UtJS27MalDcHmJMX3wY8ledHAX+r95pDTOQhiTxEDrpEHpLaPJxgu+ihllZo0zgeeK5meTvwbuBq4HuSPgasaPRi24uARd3XS1rv0uELO1jkIYk8RA66RB6S/uahFSqNeh3M2PYe4HNVBxNCCKGxw5odAOnMYmzN8puAPzcplhBCCD1ohUrjYWCCpDdLGgzMApYfhHIPuGR1iIo8JJGHyEGXyEPSrzxU2hAu6TZgCqkB5gXgcts3SZoGfJd0y+1i29+oLKgQQgjFKr97KoQQQvtqhctTIYQQ2kTbVxqSPirpKUlbJX21zvYhkm7P29fm50Q6TkEeLpa0RdJGSfdJqt8PdRvrLQc1+82UZEkdedtlSR4kfTp/HjZL+lHVMVah4DsxTtL9kjbk78W0ZsQ5kCQtlvRXSZsabJekq3OONkrqfZAN2207kdpA/gC8BRgMPAac3G2f84Hr8vws4PZmx92kPHwIGJrnz+u0PJTkIO93FLAKWAOc1uy4m/RZmABsAN6Ql0c3O+4m5WERcF6ePxl4ttlxD0AePgCcAmxqsH0a8AvSow+TgbW9ldnuZxrvArbafsb2f4GlwIxu+8wAluT5ZcBUdd7gw73mwfb9tvfmxTWkW5s7SclnAeAK4NvAv6sMrkIlefg8cI3tnQC2+9cJUWsryYOBo/P8cDrwVn/bq4C/97DLDOAHTtYAIySN6anMdq806j1NfnyjfWzvB3YBx1QSXXVK8lDrXNJ/F52k1xxImgSMtb2yysAqVvJZOBE4UdJqSWtyh6GdpiQP84DZkrYDdwFfqia0ltLX346WeCL8/1H3afJ+7NPuit+jpNnAacAHBzSi6vWYA0mHAd8BzqkqoCYp+SwMIl2imkI643xQ0kTb/xjg2KpUkofPAjfbvir3gXdLzsPLAx9ey+jz72O7n2mUPE3+yj6SBpFOQ3s6XWtHRU/VSzoDuAyYbvs/FcVWld5ycBQwEXhA0rOk67fLO7AxvPQ78TPb+2z/kdTJ54SK4qtKSR7OBX4MYPsh4AjSM2SHkj73yNHulUbJ0+TLgbPz/EzgV84tQB2k1zzkSzPXkyqMTryG3WMObO+yPcr2eNvjSe06022vb064A6bkO3En6cYIJI0iXa56ptIoB15JHrYBUwEknUSqNF6sNMrmWw6cle+imgzssv2Xnl7Q1penbO+X9EXgHl59mnyzpPnAetvLgZtIp51bSWcYdbtYb2eFeVgAvB74Sb4PYJvt6U0L+iArzEHHK8zDPcCHJW0hjVXzFdsvNS/qg68wD5cAN0j6MumSzDmd9g9lbS8cue3mcuBwANvXkdpypgFbgb0UdBIbT4SHEEIo1u6Xp0IIIVQoKo0QQgjFotIIIYRQLCqNEEIIxaLSCCGEUCwqjRBCCMWi0gghhFAsKo3QsSQdI+nRPD0vaUfN8uBmx1ePpN/mvyMknV9vW0VxHHD8ECAe7guHCEnzgN22F9bZJtJ3oWU6qsuDha20PXEAj9HwfVdx/NCe4kwjtJw84lrXGcHa3EPtwT7GeElPSLoWeAQ4vXZ0M0lzc0WDpNmS1uV4rpf0ugblPSlpSR4BbZmkoXnbxZI25emivG6YpJ9Leiyv/0xevzsX+S3grfmYC2q3Sbqy9ixA0jxJl5TEWud9j5V0p6TfKY3iN6eH4/eah3AIaPbIUjHF1H0Cfg8ce5DLnAfMrVkeD7wMTK5Z3lSzfW5+zUnACuDwvP5a4Kw65Y8n9V/0vry8OJdxKvA4MIzU99dmYBLwSeCGmtcPz39314un27ZJwK9r1m8BxpXE2v1953Uj898jgU2k8Wa656MoDzF1/tTWHRaGjnUX8LikH9q+KLc/XEXqXG8EcJntHQfhOH9yGq2sJ1NJP/wP544ejwQa9RL8nO3Vef5W4EJgH3CH7T0Akn4KnA7cDSyUdCXpMtCDpUHb3iBptKTjgDcCO21vyx30lcTa/X1fKOnMPD+W1E36891e05c8hA4WlUZoKZLeSxoYZozTSIsA7wdetD2/wWsuIA1hCjDNdumwnXtq5vfz2su1R3QVDyyxfWlBed0bCE39QW6w/bSkU0k9jH5T0r2N3l8Dy0hd/R9LGsq0L7G+8r4lTQHOAN5je6+kB3j1vdfqSx5CB4s2jdBqPgU87dS1tSQdDTwEDJF0i9LIg69h+xrb78xTf8d5fgEYne+4GgJ8PK+/D5gpaTSApJGSTmhQxjilEeAgjQr3G2AV8AlJQyUNA84kjZR3HLDX9q3AQuCUbmX9kzRwVCNLSd38zyRVIH2Ntctw0pnKXklvJw1OVe/4/Sk7dKCoNEKruQ34gqSNpIGSJtj+l+3LSINpXTAQB7W9D5gPrAVWAk/m9VuArwH35ph+CYxpUMwTwNl5v5HA920/AtwMrMtl32h7A/AOYJ2kR0mjKX69WzwvAatzI/mCOvFuJv2o73AeNKePsXa5GxiU97+ClPMDjt/PskMHiltuQ8vLd/rsJzUk32F7RZNDOkDcohoOFdGmEVqe7XjILIQWEWcaIYQQikWbRgghhGJRaYQQQigWlUYIIYRiUWmEEEIoFpVGCCGEYlFphBBCKBaVRgghhGL/A/93JYhshLLNAAAAAElFTkSuQmCC\n", | |
"text/plain": [ | |
"<Figure size 432x288 with 1 Axes>" | |
] | |
}, | |
"metadata": { | |
"needs_background": "light" | |
}, | |
"output_type": "display_data" | |
} | |
], | |
"source": [ | |
"np.seterr(divide='ignore', invalid='ignore') # disable warning for 1/0 divisions\n", | |
"plt.plot(thresholds, 1/thresholds, 'k--')\n", | |
"plt.plot(tpr1, 1/fpr1, label='Multilayer perceptron low pT')\n", | |
"plt.plot(tpr1H, 1/fpr1H, label='Multilayer perceptron high pT')\n", | |
"plt.plot(tpr_cnn, 1/fpr_cnn, label='Convolutional NN low pT ')\n", | |
"plt.plot(tpr_cnnH, 1/fpr_cnnH, label='Convolutional NN high pT')\n", | |
"plt.gca().set(ylabel='1/$\\epsilon_B$ - 1/False positive rate', xlabel='$\\epsilon_S$ - True positive rate', title='$p_T$ dependence', xlim=(-0.01,1.01), ylim=(1,5*10**3), yscale='log')\n", | |
"plt.grid(True, which=\"both\")\n", | |
"plt.legend();" | |
] | |
}, | |
{ | |
"cell_type": "markdown", | |
"metadata": {}, | |
"source": [ | |
"We see that high-pT jets are slightly better classified, but not by a big margin" | |
] | |
} | |
], | |
"metadata": { | |
"kernelspec": { | |
"display_name": "Python [conda env:tensorflow]", | |
"language": "python", | |
"name": "conda-env-tensorflow-py" | |
}, | |
"language_info": { | |
"codemirror_mode": { | |
"name": "ipython", | |
"version": 2 | |
}, | |
"file_extension": ".py", | |
"mimetype": "text/x-python", | |
"name": "python", | |
"nbconvert_exporter": "python", | |
"pygments_lexer": "ipython2", | |
"version": "2.7.15" | |
}, | |
"toc": { | |
"base_numbering": 1, | |
"nav_menu": {}, | |
"number_sections": false, | |
"sideBar": true, | |
"skip_h1_title": false, | |
"title_cell": "Table of Contents", | |
"title_sidebar": "Contents", | |
"toc_cell": true, | |
"toc_position": {}, | |
"toc_section_display": true, | |
"toc_window_display": false | |
} | |
}, | |
"nbformat": 4, | |
"nbformat_minor": 2 | |
} |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
import sys, os | |
import matplotlib.pyplot as plt | |
import numpy as np | |
# find pythia path so you can import it | |
pythia_path = 'pythia8219/' | |
cfg = open(pythia_path+"examples/Makefile.inc") | |
lib = pythia_path+"lib" | |
for line in cfg: | |
if line.startswith("PREFIX_LIB="): lib = line[11:-1]; break | |
sys.path.insert(0, lib) | |
import pythia8 | |
def write_mg_cards(PTRANGE, delta=10, nevents=200000, extra=''): | |
""" | |
Write two Madgraph cards for production of ttbar and jj events. | |
Apply generator-level cut on the particle momenta given by PTRANGE, | |
allowing for a tolerance delta. | |
If argument extra is specified, append it to output directories | |
""" | |
with open('generate_tt.mg5','w') as f: | |
f.write(""" | |
generate p p > t t~ | |
output jets_tt{3} | |
launch | |
madspin=none | |
done | |
set nevents {2} | |
set pt_min_pdg {{ 6: {0} }} | |
set pt_max_pdg {{ 6: {1} }} | |
decay t > w+ b, w+ > j j | |
decay t~ > w- b~, w- > j j | |
done | |
""".format(PTRANGE[0]-delta, PTRANGE[1]+delta, nevents, extra)) | |
with open('generate_qcd.mg5','w') as f: | |
f.write(""" | |
generate p p > j j | |
output jets_qcd{3} | |
launch | |
done | |
set nevents {2} | |
set ptj {0} | |
set ptjmax {1} | |
done | |
""".format(PTRANGE[0]-delta, PTRANGE[1]+delta, nevents, extra)) | |
def extend_jet_phi(phi, jet_phi): | |
""" | |
If a jet center is close to either 0 or 2*pi, its constituents could be on the other side | |
of the periodicity line. This takes care of this problem by remapping phi to be either | |
above 2*pi or below zero. | |
""" | |
if abs(jet_phi + np.pi)<1.: # phi close to -pi | |
return phi-2*np.pi*int(abs(phi-np.pi) <1-abs(jet_phi + np.pi)) | |
elif abs(jet_phi - np.pi)<1.: # phi close to pi | |
return phi+2*np.pi*int(abs(-phi-np.pi) < 1-abs(jet_phi - np.pi)) | |
else: | |
return phi | |
def make_image_leading_jet(leading_jet, leading_jet_constituents): | |
""" | |
Jet and constituents are passed as pythia vec4 objects. | |
Restricts image grid to within a DeltaR=1.2 range around jet center. | |
Returns pT-weighted histogram, and tuple with histogram grid. | |
""" | |
jet_phi = leading_jet.phi() | |
jet_eta = leading_jet.eta() | |
### redefine grid to only be Delta R=1 around jet center | |
yedges = [phi for phi in phiedges if abs(phi-jet_phi)<=1.2+(phiedges[1]-phiedges[0])] | |
xedges = [eta for eta in etaedges if abs(eta-jet_eta)<=1.2+(etaedges[1]-etaedges[0])] | |
jet_constituents = np.array([ [c.pT(), c.eta(), extend_jet_phi(c.phi(), jet_phi) ] for c in leading_jet_constituents ]) | |
histo, xedges, yedges = np.histogram2d(jet_constituents[:,1], jet_constituents[:,2], bins=(xedges,yedges), weights=jet_constituents[:,0]) | |
### transpose to have eta=x, phi=y | |
return histo.T, (xedges, yedges) | |
def make_image_event(all_jets, all_constituents): | |
""" | |
Jets are passed as pythia vec4 objects | |
Returns list of pT-weighted histogram, and tuple with histogram grids, covering full (eta,phi) ranges | |
""" | |
out=[] | |
for i in range(len(all_jets)): | |
jet_phi = all_jets[i].phi() | |
jet_constituents = np.array([ [c.pT(), c.eta(), extend_jet_phi(c.phi(), jet_phi) ] for c in all_constituents[i] ]) | |
histo, xedges, yedges = np.histogram2d(jet_constituents[:,1], jet_constituents[:,2],bins=(etaedges,phiedges),weights=jet_constituents[:,0]) | |
### append to output (transpose to have eta=x, phi=y) | |
out.append(histo.T) | |
return out, (xedges, yedges) | |
def run_pythia_get_images(lhe_file_name, PTRANGE = [500., 700.], PTRANGE2=None, nevents=10**6, plot_first_few=True): | |
""" | |
Take an LHE file, run pythia on it, outputs images. | |
For each event, cluster jets, check if the two highest pT jets are in PTRANGE and PTRANGE2, | |
and make 2D histograms of the leading jet and of the whole event. | |
If plot_first_few=True, plot both images for first 5 events | |
""" | |
# unzip LHE file if it is zipped | |
if lhe_file_name.endswith('gz') and not os.path.isfile(lhe_file_name.split('.gz')[0]): | |
os.system('gunzip < {} > {}'.format(lhe_file_name, lhe_file_name.split('.gz')[0])) | |
lhe_file_name = lhe_file_name.split('.gz')[0] | |
if not os.path.isfile(lhe_file_name): raise Exception('no LHE file') | |
PTRANGE2 = PTRANGE if PTRANGE2 is None else PTRANGE2 | |
pythia = pythia8.Pythia() | |
### read LHE input file | |
pythia.readString("Beams:frameType = 4") | |
pythia.readString("Beams:LHEF = "+lhe_file_name) | |
pythia.init() | |
### define jet parameters: antikt, R, pT_min, Eta_max | |
slowJet = pythia8.SlowJet(-1, 1.0, 20, 2.5) | |
# outputs: lists of leading jet and full detector images, and (pT, eta, phi) of each jet | |
leading_jet_images, all_jet_images = [], [] | |
jetpep=[] | |
iplot=0 | |
### Begin event loop. Generate event. Skip if error or file ended. Print counter | |
for iEvent in range(0,nevents): | |
if not pythia.next(): continue | |
print('{}\r'.format(iEvent//10*10)), | |
### Cluster jets. List first few jets. Excludes neutrinos by default | |
slowJet.analyze(pythia.event) | |
njets = len([j for j in range(0,slowJet.sizeJet()) if slowJet.p(j).pT()> PTCUT]) | |
jetpep.append([[slowJet.p(j).pT(), slowJet.p(j).eta(), slowJet.p(j).phi()] for j in range(0, njets)]) | |
jet_list = [ slowJet.p(j) for j in range(0, njets)] # if PTRANGE2[0]<=slowJet.p(j).pT()<=PTRANGE2[1]] | |
jet_constituent_list=[ [ pythia.event[c].p() for c in slowJet.constituents(j)] for j in range(0, njets)] | |
### at least two high-pT large R jets in the right range | |
if njets<2: continue | |
if not (PTRANGE[0] < jetpep[iEvent][0][0] < PTRANGE[1] and PTRANGE2[0] < jetpep[iEvent][1][0] < PTRANGE2[1]): continue | |
hh, (xx, yy) = make_image_leading_jet(jet_list[0], jet_constituent_list[0]) | |
hh1, _ = make_image_event(jet_list, jet_constituent_list) | |
leading_jet_images.append(hh) | |
all_jet_images.append(hh1) | |
if plot_first_few and iplot<5: | |
fig, (ax1, ax2) = plt.subplots(1,2, figsize=(10,4)) | |
ax1.pcolor(xx,yy, hh, cmap=cmap) | |
ax2.pcolor(etaedges, phiedges, sum(hh1), cmap=cmap) | |
for j in jet_list: | |
ax2.add_artist(plt.Circle((j.eta(),j.phi()),1, color='r', fill=False) ) | |
for h in [-np.pi, np.pi]: | |
ax2.axhline(h, ls='--', lw=1, c='gray') | |
iplot+=1 | |
return leading_jet_images, all_jet_images, np.array(jetpep) | |
def pad_image(image, max_size = (16,22)): | |
""" | |
Simply pad an image with zeros up to max_size. | |
""" | |
size = np.shape(image) | |
px, py = (max_size[0]-size[0]), (max_size[1]-size[1]) | |
image = np.pad(image, (map(int,((np.floor(px/2.), np.ceil(px/2.)))), map(int,(np.floor(py/2.), np.ceil(py/2.)))), 'constant') | |
return image | |
def normalize(histo, multi=255): | |
""" | |
Normalize picture in [0,multi] range, with integer steps. E.g. multi=255 for 256 steps. | |
""" | |
return (histo/np.max(histo)*multi).astype(int) | |
######################################################## | |
############# Run everything ############# | |
############# one run by default ############# | |
######################################################## | |
etaedges = np.arange(-3,3+0.01,0.12) | |
phiedges = np.arange(-np.pi*4/3,np.pi*4/3+0.01,np.pi/18.) | |
cmap = plt.get_cmap('gray_r') | |
PTCUT = 50. | |
nevents = 50000 | |
if __name__ == "__main__": | |
outdir = 'images_out/' | |
if not os.path.isdir(outdir): os.system('mkdir {}'.format(outdir)) | |
cwd = os.getcwd() | |
### write madgraph cards, run madgraph -- PTRANGE=500-700 | |
write_mg_cards([500,700], nevents=nevents) | |
os.system('cd madgraph; bin/mg5_aMC {}'.format(os.path.join(cwd, 'generate_tt.mg5'))) | |
os.system('cd madgraph; bin/mg5_aMC {}'.format(os.path.join(cwd, 'generate_qcd.mg5'))) | |
### run pythia | |
lhe_file_name = 'madgraph/jets_tt/Events/run_01_decayed_1/unweighted_events.lhe.gz' | |
leading_jet_images, all_jet_images, jetpep = run_pythia_get_images(lhe_file_name, PTRANGE=[500,700], PTRANGE2=[450,700], plot_first_few=False) | |
np.savez_compressed(outdir+'tt_leading_jet.npz', leading_jet_images) | |
np.savez_compressed(outdir+'tt_all_jets.npz', all_jet_images) | |
np.savez_compressed(outdir+'tt_jetpep.npz', jetpep) | |
lhe_file_name = 'madgraph/jets_qcd/Events/run_01/unweighted_events.lhe.gz' | |
leading_jet_images, all_jet_images, jetpep = run_pythia_get_images(lhe_file_name, PTRANGE=[500,700], PTRANGE2=[450,700], plot_first_few=False) | |
np.savez_compressed(outdir+'qcd_leading_jet.npz', leading_jet_images) | |
np.savez_compressed(outdir+'qcd_all_jets.npz', all_jet_images) | |
np.savez_compressed(outdir+'qcd_jetpep.npz', jetpep) | |
# one can just do another run at another pT range by copying the commands above, e.g. write_mg_cards([800,900], nevents=nevents) | |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Simple pipeline for jet classification at the Large Hadron Collider. Make images of QCD and top quark jets, train a convolutional neural network to distinguish them.
You can find a detailed introduction and step-by-step notebook walkthrough at https://ilmonteux.github.io/2018/10/15/jet-tagging-cnn.html.