{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "## Chapter 18 - Metric Predicted Variable with Multiple Metric Predictors" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "- [18.1 - Multiple Linear Regression](#18.1---Multiple-Linear-Regression)\n", " - [18.1.4 - Redundant predictors](#18.1.4---Redundant-predictors)\n", "- [18.2 - Multiplicative Interaction of Metric Predictors](#18.2---Multiplicative-Interaction-of-Metric-Predictors)" ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "# %load ../../standard_import.txt\n", "import pandas as pd\n", "import numpy as np\n", "import pymc as pm\n", "import arviz as az\n", "import matplotlib.pyplot as plt\n", "import seaborn as sns\n", "\n", "from matplotlib import gridspec\n", "from IPython.display import Image\n", "\n", "%matplotlib inline\n", "plt.style.use('seaborn-white')\n", "\n", "color = '#87ceeb'\n", "f_dict = {'size':16}" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### 18.1 - Multiple Linear Regression\n", "#### Data" ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "\n", "RangeIndex: 50 entries, 0 to 49\n", "Data columns (total 8 columns):\n", " # Column Non-Null Count Dtype \n", "--- ------ -------------- ----- \n", " 0 State 50 non-null object \n", " 1 Spend 50 non-null float64\n", " 2 StuTeaRat 50 non-null float64\n", " 3 Salary 50 non-null float64\n", " 4 PrcntTake 50 non-null int64 \n", " 5 SATV 50 non-null int64 \n", " 6 SATM 50 non-null int64 \n", " 7 SATT 50 non-null int64 \n", "dtypes: float64(3), int64(4), object(1)\n", "memory usage: 3.2+ KB\n" ] } ], "source": [ "df = pd.read_csv('data/Guber1999data.csv')\n", "df.info()" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
StateSpendStuTeaRatSalaryPrcntTakeSATVSATMSATT
0Alabama4.40517.231.14484915381029
" ], "text/plain": [ " State Spend StuTeaRat Salary PrcntTake SATV SATM SATT\n", "0 Alabama 4.405 17.2 31.144 8 491 538 1029\n", "1 Alaska 8.963 17.6 47.951 47 445 489 934\n", "2 Arizona 4.778 19.3 32.175 27 448 496 944\n", "3 Arkansas 4.459 17.1 28.934 6 482 523 1005\n", "4 California 4.992 24.0 41.078 45 417 485 902" ] }, "execution_count": 3, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df.head()" ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [], "source": [ "X = df[['Spend', 'PrcntTake']]\n", "y = df['SATT']\n", "\n", "meanx = X.mean().to_numpy()\n", "scalex = X.std().to_numpy()\n", "zX = ((X-meanx)/scalex).to_numpy()\n", "\n", "meany = y.mean()\n", "scaley = y.std()\n", "zy = ((y-meany)/scaley).to_numpy()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "#### Model (Kruschke, 2015)" ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "data": { "image/png": "iVBORw0KGgoAAAANSUhEUgAAAUYAAAFFCAYAAACQQ0ReAAAAAXNSR0IArs4c6QAAAdVpVFh0WE1MOmNvbS5hZG9iZS54bXAAAAAAADx4OnhtcG1ldGEgeG1sbnM6eD0iYWRvYmU6bnM6bWV0YS8iIHg6eG1wdGs9IlhNUCBDb3JlIDUuNC4wIj4KICAgPHJkZjpSREYgeG1sbnM6cmRmPSJodHRwOi8vd3d3LnczLm9yZy8xOTk5LzAyLzIyLXJkZi1zeW50YXgtbnMjIj4KICAgICAgPHJkZjpEZXNjcmlwdGlvbiByZGY6YWJvdXQ9IiIKICAgICAgICAgICAgeG1sbnM6dGlmZj0iaHR0cDovL25zLmFkb2JlLmNvbS90aWZmLzEuMC8iPgogICAgICAgICA8dGlmZjpDb21wcmVzc2lvbj41PC90aWZmOkNvbXByZXNzaW9uPgogICAgICAgICA8dGlmZjpQaG90b21ldHJpY0ludGVycHJldGF0aW9uPjI8L3RpZmY6UGhvdG9tZXRyaWNJbnRlcnByZXRhdGlvbj4KICAgICAgICAgPHRpZmY6T3JpZW50YXRpb24+MTwvdGlmZjpPcmllbnRhdGlvbj4KICAgICAgPC9yZGY6RGVzY3JpcHRpb24+CiAgIDwvcmRmOlJERj4KPC94OnhtcG1ldGE+CrDjMt0AAEAASURBVHgB7F0HYFRV1j6Q3kkCBBICCb333kE6KDZALIi67tpd+/6WtbuWtaxt3V27giCKDRABpffee0soAdJ7A/7vu5MXJmGSTEmZmdyrw0xeue+9c98999Tv1LmIJrppCmgKaApoChRToG7xL/1DU0BTQFNAU0BRQDNG/SJoCmgKaAqUooBmjKUIov/UFNAU0BTQjFG/A5oCmgKaAqUooBljKYLoPzUFNAU0BTRj1O+ApoCmgKZAKQpoxliKIPpPTQFNAU0BzRir+R1g0OgFhI7q6NFqJryDl7uAgeO46VY7KOBZOx6zZp+yELPqPD6cVnmFFyUt57z4eNaRQJ+64lG3jtStU0c8sUTxWzfnoADH63wRMyzEj7Tc83Lhgkg9fw/x9MCY4TY9MXYcP93cjwKaMVbhmHJy5WNS7T2TKyfS8iU9r1AKCi9InTomQb0OWKW3V12p7+8p7Rr4SoMgb/EGw9RTrQoHpYKuMWRScP6CxKUWyJGkHEnKPi/5hedx1iXlygtjFOLjIS3q+0qzel7iizHUi1oFhHWx3XV0SmDVjBgZ4pGkfNl2KkNyCyq+BplkRLC39GkaICG+nphoFZ+jj6hcClCyT8oqkA3x2fjOR+eXmGFZVwryrit9YgKkYSAWNUiSurkHBTRjrIJxzCu4ICuPZcgpSIkXLhZNljoXxQOqsgfmGr8pmdBmdf4Cf1PJNjVKIwOaBUtkiAcmWsUT0zhPfztGAarLOxOyZc+ZHMmDgGhQnpKgR12MXdFKdQEDh0MxbpeuVxdj3D7SRzpE+CvpUbPHS7Rx1V+aMVbyyOVh1izenyaJ2RATi5iil4dIkI+XNAj0kJbhPhIG1ZnM8FhSgcSl50lyVqFkFWA2Fh3vUbeu9GvmLzFhvsr2WMm3qLsrRQFKiptPZMvBc9lSiIWKjWuSr6eHNAzygLrsK1FQmbmgnU7Pl2PJeZKA72zYi80ZZOsGPtItKkD8oFrr5toU0IyxEsevAExx2eE0OZluYopUj/28PaRTI39pWd9HvMpQtcgU1xzNkjOZ+aLMWbgnTsxhLUIkMthLwCd1qyIKcIHanZAjO05nw7ZougiGTJqF+krPaH84ySwTnxL/hvhMLG65yqF2scgy3DUyQNo38tHSfhWNV3V1qxljJVGajpb1cVnKYE+pg0wxCLbC4S0DJcTPyyqHyqb4LDmQmANjf9EEhVo9uk09CfXz0DbHShon825owTieki9rj6eDuZn2MFqgb7MgiYaESK9zRS0eTpr1OD8z32QOoVrdv0WgxNTzVt7ris7X+52TApaXQ+e8V6e9K0oPB87lytGU3GJVLABeyxGtgq1miny4ntEB0rq+HyRL06PmQ1VbcSRDMjBrTdPOaUngkjeWnFMomyD15cImzEbnyYDYYGlqJVPkOWSg/WKCJMDbxEQvwJa8DvblxOxCMTND8lDdXIgCmjFWwmAlwpO581Q2pA7TVPD1qiNDmgchTtHDKknR/BbIHKNCvCEhmlhhGibvlpM5klM0ec2P1b/tpwDjSVdh0cnMN4VP0XTRBfbBxkGexY4Wa3vneHVsFAC123QGX4MVh9MlA7GPekGzlorOdZxmjA6OByfYmmOZynmCaDaoTyK9o4PgYIFtsGJNzOLVBzcPlgYBnGUmRnscxv5jKQWwP+ppZpFgNm6k2WM1pLr03CL9GYtQczi6WoZ7l2kHrugS7SIQ0xjmA/XbNEbZBRdlE0wrOYbhsqIO9H6nooBmjA4MByfYmuNQdTHBVNA2JlgrqMJNQih12N8xGeoASJzB8GQbzHFzXLqk4jpaPbOfrjyTbOtwcr7yLp8vigIIxyLWPcqvTEeLtVfsB9tkgwBvvAumUYpHuNbx5AI4dfSCZi0NneU4B6avszxCzd3HsVITrAEmWOfGfshecZyswdDLukK1Y8gIGyfx6mNZkp1f5Dqtucd26StnIUhxB8we+UVmDz8EaA9EgLavYdh14OmoIAyIDSi5oCEMSKvUDhC1hk51fAbX0I3X9GWpQm89mVnsQWZYR39OsEpgisazNYdqFwXvZt0iCSQFBv0jWgIxyGPz93m4oVfD7JENtz8lfAZvd48KVNED+FkpLRALWicsjsZ7UIgE6/WINsiBLVM316GAZox2jBXDPBjDlm04RKBCd4nyh7PFExPOjg7LOaUvUgRD/b2hA5om1o6T2ZIFqVErZ+UQrYxd8QjNSULgvZGNFBniJdEwe1gTllNGlxY3t0QOdSOkdxoOtDMZBXIKH61RWySXU27UjNGOYUkGsMCptLziCUa7UizCNsoK4OYlLkByKCgokPz8fDl/vnx1uLCwUB3HYzlpOzf2V4Hi7IcSyBaoZ4YHnNusbayUy3vIyUHaWx7uH30ZH2v7cNXjmN2yTanQpidQ8YqIAPCpIEvFGDN+l1dpmGNagPHimJGmDOEJQlSCYSPeAqkxywiWtJGI7JvjlZuLcDC8G7wPbivvfmy8hD68FAU0uk4pglT0J2MWN0GFNsW+1RUvMK6+SN+ryEaVmpIie/bskRR8RzdtKh06dBAvLzpXSjZOwI0bN0pWeob4BfhLr969kYXhI4eTvOVEag6YcV0g9RTAo3pevANMkGUle7D8FyfrubNn5dChQ3L06DEJRN+dunRREzksLEzq168vHh5FAZSWu3DprbvP5hYxJlPwfafIQMUUyxPwORZbNm+WxMRE8fXzkwH9B+Db9zI6kLb79++X+OPH1XHt23eQ8PrhyhG3IyFLmVuoXRyFxNo+wkO9M5d1UsaGzMxMiUO/+/ftkxwwxpYtW0p0dLSkpqVJkyZNJCgoqIwz9WZHKKAlRhupx1zZZMQtXiyCGWCIRiAM9xWp0CH16iGUx1NefvFF+fR/Hwtf+NKNE2z1qlXy0AMPyJo1a6R1mzbi7Q01Gq1ntJ/4M1cNjWlszJLJhZ3T2nb69Gm5689/keXLlmNytZD8gkJ58dnn5M7b75BcSJB13TjvkPbgQ2dzpKAoDzoYGUnNQ8uX8ElXLlxkPhyvZ596Wi1qluh96tQpeeLRR+UTHBeF48kUSc+OsDUSKckwg+xC6mE2bI00xVjTKB1++P4H8v677yp3esOICFkwf75Mu+lmWb1ypZIerelHH2M7BTRjtIFmDM/ZfIISgOnNpoG9C17+itQxXoLSWIOGDcTHx0fOnEmQ83jpSzdOsLf++aZkgWleP3mykuLqFHFceqkbw25lxMmdyyyQxMwLsFtZN8vmzJ4t/v5+8qc/3yl9+/WTyVMmy9PPPSvde/aQgIBAMPbyZKfSd+o6f5M6WzBm2UY8IezB3ZswGNu6V79R48ZqzJKSkyQvN++yB8/OzpZ33npbjhw+LBOvvkaioqJKLDLMk/eB55utAO8N87IzCN9jRaOUuGL5Mrnm2utl/FVXyogRI+TvWMy69egOjaOjlhatoKG9h1j3dtjbu5uddyAxT73UBAxgLnTbhr4KZsrax1y3br106dJV0lJTEaxdcnJkZWXJNzNmmCSNjp2kYUTDEhOM1+iJ8J0AMEhYBnH1OgggTr/kAKrgJs6dS5QESI2puLZhm6JaNmTIUPGzoB5W0J3L7CYTOpGaD2nRtIA0CPCSiEDrs1sOHDggTWOaQdr3kHOJ50pIaZTo5s6dKznZWeLrC9CJXj0VEzUnTtNQbwARI7axKAL1MDA6U3Np2zU/yvLvlOQUZROOj48XMmC2ugiQvWLESGneorl4empLmGXKOb5VM0YraciJtRd4fYbgQQZFmKnyHC7mXZMZnU1IkI6dO0taWrrk5xEI1dRoy/puzhzp1q27smf17ttH/DDRSjdKprHhvuJThNKTDrUsDnaropC80oeX+LsfpETaqD549z2hZGowx/Hjx4t/QECJY93lD/LCDcczJQeQRcxKItl6NQ1UKOnWPuPGDRtk1KhREhAYqBYWmjvYSL/NmzYpSZtmkQ5YzMJhp7VkkujZxK/YeXYe5+88naWC9Su6hzZt20hkZJR8/dWXsgl2ZzrN2IYNGyb1GzSo6HS93wEKaMZoBfEoa2xFmIy5OtYV4Tm2AMnS6RIcEiJRkZHA8Duv7FX0LHKibdywEVKbn1KhCwsLpF+/AeJjgTHyVqm6U6022o7TOQjfqRhkYtyE8XL99ZNk9epV8vY/3ypmjnQmuKsaTXPDOdiDjfCcJogJZUkCa8sQkPklJyZJ2/btTSaQhDPFEmPC6QTY+VZJv7595RicI7379C5T8g5F4H9kiA+ua2KqiTSDILKhohTPBg0byt333SvhYaHy+quvyRrYn+mZ9vbxdmtHmfFu1+S3ZoxWUJ/ZJsxXzi9Sx5hCFgWcRBZFsrZxxe/UpbM0i4lRL3UCpEcyRjpF1q9bJyMhlezYsQN2xQbSvGVzix5rXotX7NiIUo9JNSRa+L5zUBUrEBspyTz48EOwLU6RP/5YophjItRrQ3K09jlc5TjaXumgInAwG2vp9ORihm9rWxKYYiiYEj2/wcEhcuYsbcPnleRGs8d1kyfJPtgB6bwqbzHj9XhtghWTNdIMsguhQykACDHdXdl31BtRCY88/jjswP7y8kuvyAaYYxgSpFvVUkAzxgroyxfX5AE2U8cY/2bDBOMl9u3dKy1atECoTrRiemcgcWSkp8vsmd/A0TJJSYwboLb17NFD/S7vtmLDvUwgFUXT6sDZbKiLpiqEpc+jREopgwyQDqCHHnlERo8ZIwt/XSCLf1uk9pU+xx3+jgfoBnPLiYpOe3DrBiYnmfVsEWr4+vXSuWtX8YXDrHGjRsKFhLSc+/1cGTBwgDSGY2bdmrVK3Y1uFl2uzY8Fs5qFoS5MEchEOmyfZzNZaMuysZFqs6G201l29733iRccRp9+/LHEx8W5wxA59TNoxljB8DBekJkLRqhHY0BM1fO1Xh1j9wzOJVOisTwE6rS/f4CcPHVS5nz7rQweOkQaIQxjLxhnIoz7/QYgVq6UGk0JgXZISphG6wFHjOENJ7z+dhW8fLn8oSTStevUPfBchqA8+tjjShLavHlTCcbIiWhcy5UlSUYPmBC5TUyHKOrtYA+2NYf94MED0qxZM/FCyFTDxo3k3LlzKsaUrLZjp06IwoFUCjtjr549y13MSEvSlujeCoquSKXeDZt1Gph36VHj8cv+WIrohTPGcMuo0aOkb//+iEE9Cht1WvF28x+8hiuPm/mz1PRvzRjLGQFqzhtVvOAFFbXoCXW0G1FYsPrb0nbu3CmtW7cujkkkI6T6zAnXrl079b16xUowRD/p2LFj8XG8BhnVmtWrZdvWrXIQHlL+zdYgkDVkkHZW5O08ngKPOcoRlpY/duHa8fFxxbYxnhscEqxCdCIaRZSQcuKOx6k4ym3btklyUrLLTjJT9ADKS4A6lBbbN7Id2CMnO0ctIkbQOyVGLjKrVqyQsePHKmfM9u2gE8J4+nIxg1RZVsvIyJAzMJ1chG25JdCXAoriURn0HY9g/TxAlJk3Xnsrxjsl+dIYkOExCD+ySdRlnm/j3JMnTsBOneWy42Y8hzN82zbDneGOq/EeEjLyhSC0hvG+KVQhFcxt5T1wBadK9NuCXyUS8W3Fkwy/Y2Nj5JprrlFOlkxkuTCwOwbbGI5hvur/vmSJUuE6gGHOnjULTC6+WMXqrrydJkcMpcZNALTNLQVWsBfZNr/i+rRpcoJSfV+zarV4enkqbyudPmyURt9/7z1p27admoBffP5ZmZKJOsFJ/2Ewtyl6wKQ0s7xEc4TM2BI9QAl/2bKlysvMx+R4MJ6xHqT96XfcrmyOPGbJosXYh/owzZpibC85xMxJw3fg7JmzchwOGob3dASTDmP4TpHUuA8ZOcnI32Y/Rjt06KDs2L5ddu/eraRUjhuzb/bu2SsDwIQZ6F26cfyOHD4iiUmJJTSL0sfpv62jgOXRtO5ctz6KubWbUF84V8Xn1FU2xW7wCNtivOfLvGzpUrzcZ1XGSXAQyqJixR84aKD0gPpVLzRUDiMweCWyUeoj+NsbUgfDQ0aOHl2smi3G5JtywxQlvTSC1LIJHmzatvz9/aHSe8Lb6a3qzLAM6xlk5ZzLOi9RsH8awAh0FgwcPEg+/+QzhUydC4mzLtIYX0QGTgeog0YsHNPOcnKyTdeB2rgazHPK1KlSDxk7rtLIXDYhmJsgG3RTEcShG9RXW1RoSuTLly6TVcgsCQCND7ZoKV26dpFWkPiffvbvWNBiVUQB7Y9nz56Rrti3bu1aFfIUAYZVOlyH7wDty4OGDC6W9No08IYKXSBpOcifByMnc2QpjBCYaNjOIHWTwdwHkGa4GSmJ3p5ekgtG3LdfXxl/5ZUSHh6ujjP/5+DBgxhLDwnFO2WMqfl+/ds2CmjGaIFenGC0UaXj5S1WxyKI2Qfrkg0ZIg0RbkEvMD/mbfyECcV/tmrVSvgpq51GzKEfJignHL+VRABmZ7Re8HaeRfhHKsu1AkprHYBzx7QNkWBMMspMf33k4WJJldILW+nJy22nT51Wtk8yTT5jAcKG8pHpQWnJlmdmXzXVkgDLFo+6OwVwuFAVqo9g7sbBtqHnMDNp1JjR6mP+HJ0Rf2q0Bogh5Biaj6Oxr/Q3bcrz582TPmBqRotC6E4DOIey8nJUjSDmvkeHFYo/TDSUbIcPHy4eIy/lrVMa5JiVNw60fx47ckRaIGif19TNMQpoVdoC/VJzCuTQuUvhOcytbVWf6lj1k4svuWFUz4b9iNIC1W2jeWMytW14SZJlbZidiG00CjwZ6juP5+SyxBS5j3bHC0XOHV4P7FEBJpQ3GXmeszSiZK9BuQKq0qSOF5xdfYCPaW3qX1U9B51qnTt3KpYWjet0jYRKjeqRbMx93wYYu2Qwdi7K5mPG/fy7onHYj+sw6oFaiG6OU+DSDHO8L7fogSr06mMM5i7EBEMNF1CoT7Oam2B9EEB86uRJlRJ2Fip5WzhrDGAJg+DtkJpI6QjWSbXpSFKeCgWxNo+aJ1FVpOc7GzbRkydOIpOjg1LtjWs48zcZyxYE4LOGC2MEBSp0m4Y+EuRVMbhHVT8X7bnd6LUuFWkQCAcM0wX9vU220Cw4YFhpkhUhbW2UKOnBDg0Lvyyiwda+9PEmCmjGaPYm0Au97STStSAxGio0U/Dq+1ufW2vWXaX8nHrjjYpZrYVnul279krtLg1XxjzrDmEXxQ+GESrLZBSUnmyB1KcK+Zd77patm7fIokW/yW233eEyKhltq0eTL5WurefrJe3BGG2xLVbKYJXqhJL34cOHpEXzFirywHw3bZnR/oUS5otUxaLYxqOJ+XIynTVizI+s+PdRqNAREY3gFAqs+GB9hFUU0DbGIjJR1jqFCXYokRPMtJEezW5QeWpygjFl7/rrr1dhOmSIllSqt9/4pyQAsWfSfU/BvuYr9D0QyHY9qtQNjA0sDg+p6I0gRiTDiqi6laVyV9RHde+n6WBN3CUnmTdspP1iAOwLabEmGwPBz0KKa9WyFSIPLg/loXPn808/lzvue1DCQltKImJliYpGJKBgOGKIpIRHKbfRy03mux0e7E5wpIVZcMqU24HeWSYFNGMsIk0WwlzWQ8pisgQbJxgLGzFjwRlaafXZ/J4IkhqHWMWnwz2lMB2AtunMdKmLSngFchBSCFVtazN1Skuj5tdxtt80e6w8ml4EQAvDB1TodhH+ynZng4+s0h+Lqi1DpAry8+Tq666VQABQlG4nYR45dvSI+NYplAiUbs1GJkxWPhDWsagRob1vszrKPFLecxyBpMj4SG8vb2nRqqWKVCh9Hf23fRTQjBF0o+F++ZF0VXzdCPNojwLqYdBNiZuYhjgygjowdk15COH9pdrCAG02bme8IldwSlrMXGF8IKU7qrm03SkHClZ3IkHzOO5jX+yfhZkMMIc8SBo81gNZMsHBwcWSGxFc8nl97OO5vA73lw7NGAAJcdGBCwDTRSA4+mVFvHp4DtY28ahIBFFP4xr/0H66/VSOnKWkBS80GwPeWd+Znl0ipntCwma4EoE5OK7EoyT9ST+OQQ6gvPIxNmxceBgCRWmZ40V6k8Y8jtu4YLB0QZ2ibfTWcxvHmrnSxtgThYfbmanC8/leJBcFavPvACAZlV7kCF/HWjRHk0zmAHrX98P5x0QClkcoS3AkbBwzc/gO8B51qzwK1HrGSKmDIS7JWSbDPTMlIoK84eklpFhdWb9pozz28CMyAeEZqxCEffLkCUgmHvLiSy/IMMSa8WX/5aefZMbXM+TEiXhkQPjJmHFj5NZbp6uYxXfeegs5txtUgG5mRrrcdPMtgKvarNQrrvYmANR8uW7SdSpubsni3xH3eAbI0dHy7ocfqBefAb5E/l6J7BgyTjLUwMAgee3NN6R3r14l3gZKuD2a+MOBdB5o0UhFwwRefSxdRrYKggRSsXpWojMn/QOPJHEoXbv/nCnchbfph+ceQC90kYQ/dcoNKquIEGtHjhwG88qTW6dPkz/fdZcEAhSCmUT/+fe/EUi9EzS6ID0B2PuXu+9WqX6bEEv65P89iUD3cBVn2rJlK+kMAJCVy1dIOEJ1jqG/jIxMadO2nVx99USZN+8XVTLCz89fnnzmaRVuQ5vtzK+/lo8//kSyMzMwZsTQFLn33nvk6muvvYyyvaMDFbbmadSiJqM/nJijqhe2qS9lls3gu8fr6Fb5FHAOPbHyn8uqHil17DmTI3EAMmXmCBtzWQfEBhWr0ERtzs3Nke+++x4AoVfIM888Iz5+PvLaq68r0NfVcIr88403lDrz/AsvCuG9fvn5Z3n0kUeRVpekanMcOnhAGgA1hwjPIyFJUIJZv34dPMFt5KWXX0a2SRv5ZuZM+emnX+TGm2+U2/90p5wAA541c5Zkgin+58N/A8l5uYqH/OzLL+X+Bx+UvPxcee+ddy1mp0Qh6LsVQBMMMxuDiFccySry2pqe01X/JXMhlBhTNfOKvBTekBApKTPVzpCuOGa/LlggMTEx8iJo3BKq5qeffKqCswkU/Pijj4GZHZa77rlL7gG017Fjx+RPt90uJ5BZlAVJMhPM7HhcvFx51VVy/aTrlUR4OuG0pAE89m9PPaWY24H9e/EevIY0zk7yHMaeMB7vF43J7l27APjwiXTp3EXeBwN+/oXnhfGPH37wocTFxV1Gfkq53SL9JRyLFxdnetcp7Z9IpTOGT61bdVKg1kqM9EBT6tgNxmh4AX0QwD0IE4yBtqXbnX++E0zrJqUKLUU2yyqEYVAN/vg//1Wq8Dvv/UsaA2uRjNHHx1u++PxzqHNpRQHSdeXdD96XpkgdY6OqFQSJ76lnnlLnJKekKqnmiSeekNGQNqlq/QBkaDJUqnU9IBVGRkXK0OHDVdhHZNTV8tOPP4LxIv0LqrWlxkmWCpUsDtUML8KqnwkbFmsqDwLTDyoK/rZ0nrNvywT69cojGUq6ogmCIZ2dkN0Sgdzx0qYCOiQeefwxxZBIx2exqBHJ/Ie5P6gc569nfCPtOrRTamh/pNrddss0+e3XhdIosrEiw5/+9Ce56Zab1Zi/8dpratsbb7+FkKm2KribZQeYQnnPfffB8REmc7/7ToGB8L2g9Hj9pElgoNeorBgG8a8DmMd8MGtK/JYaQ65aN4BqjhcyIw9lKzC26+MylfOP9cXBO3WrJgrUSsZI9JXTsE1twEtnoDtTuuoVHSRhjAe08AIy88FQWwKR2mccQvtRFCYS7VlGI0yUkdfMCUlGyHQt88bAbdqr2FS+MuxenMg8lo1qkpEzTRBUwk3d+5e7kPmSBFtjrqQjvzqycWTxMeqkUv8MaR4kv8HeeAY53ww/OpdZqCStXoBNczXmSPWZqX6/H0ovivUD7eBsaQbHRetwy7nQMbGxxfSsD48tbXG0+bFSoocHoNvCQ4ttc21QeIx23t2790Bdhv6KVjpmlHY8jhvHxmis+GiMGb9pv2SLiY1RDpEnIJmeAJIS7ZnZWdlK8uSzlNVa1fdR5TMIJUdHIJzushYo5P3rBEkksnhKM/+y+tHbHaPApRF2rB+XOJvvYz7Ukv0IpF15OEOyYJg3grg7wNnSDLWhjRxjax+I0h2dL3SYGI11VWhTqhdaT00iS4ZxMkVjEhnnlfX93N+fVYyR4Ri3TJsmXwIklVKLeQaMpXOZ3jcYzJGgBaYIR5F4mA2WQ+JKBUgqFwhXaLQDn4MzadGBdElh6iNGjUyxIRax3ixsZUHC53MZC0vpZ6SThUg3dJgYxxhjGIrccIPx0VFi/GYfZY2X0Yf5dRb++qv8/emnFTMcO2asvINKf9ded53F0B3z8/i7O9I8m4bSiWTaw5CklXAOHoaGw4ym8hhr6b703/ZRoNZIjLTT5IIRrjuWpeIVacNRTBEvH1PqOhR5M20lY7v27WThgoVC4Fl/epyhJv0MNdejrgeCbhte5jW2tX8evxPI3tFwxjz7/HOq1gc93awbQ5Wtoka725AWQbAx0sHEsq91JAmOpl/3pUnXxv4SE+4jrHZoSUquqO+q3k/GTaa483QuFrNsoHED/o3qM4SycHigh7YILLYF23IvgwCqsfSP32UtQGY5ZowumD9vvoogGDRsiBTmmzzVtvRZ+thFCxfCDFKoHGRNoqPh0S6QD997vxg2rvTxpf/u2ywQDPCCHIeNkVUpuaCvgynkFBbvTizLikgDSo+G5lL6fP23YxRwO8ZIGYieWApw/FYyEf7ZBkP2YYRDEDGZtikauH2wJHfGS9YKdh2vMjgDJQZKfObSAlUyfrjtftSAZoDt7dNvk+7duykjPlP3pk27VYKKwmmoZpc+31ztYv9KqjR7y82v0aJ5c1kPT+l77/xLAhAmRFtVenqa8oBTfOCx5v2VfiWI2jKiZbCsOJqJVEGG/NRRk20DHBiUnntHI/avyITA+yQpWBelDJKU7r5S/qbwaho305hRKiJW4fZTWcjgMYU0cSGjFBUNEIaK0jRJT9LFaMY48psI5t8AOf31V/+hbLkMwzl67CiCsVsDLaerWohKjznDp1R/ZmOkxo0mkqJt5uPQvUdP2bJlC6IJXpb2sGNu27INsabHWddASaq8Dx5v/l4Y98pv0r5/bLD4IdXxUCJTVBlhIHIMxc9OIT6Vi3kL5O8zF5yXN8ZLgYCYd6R/20WBS2+OXac730msz3I6o1ASYENMgfOBdim891JYx6Q2Gwb7QBSUGhgTCC+gp3qpynqSxoDg6typs4Sh9gdfZrYumDwZYEw+Pr7K6P4BvI5zZs2WtevWSgNIiTdNu0UZ3qmKdeveXQryCortkzyfVQAp9Rk2S+L5cUL6mBWm6oHwkebNWyhHzEv/+AecOZ/JihUrMAnqyDUIGo4Acs8fv/+hbJu9kE8de65FcX+8Runm511XRrUJVjbGowgFyYEEQtj/NCCULz6YARMCYilh3gzyqasCiyNR04ZhS9XBHHkrp5F1lIAJT49zFhwPOVjACiEhqkUMCxknPu+vE0weLWGHqwhfsVvXbor2RsxgA0CCdVKV/ADCgWf98KN/y49wwjD9ke8HQ2juhROFIB1RUU2kK7zJBNYwGBezgvr06VuCxp1gd2Z9Z+MahJJjjRjGNt52x+349lGe8VUrV0l7FNT6+/PPyrtvv4O89yyFz9kd9aEZi1pWI8OjWl0/wEMBgyQSt5GLGghG9KddqFHtjRnMvOt6ftBQMF5NIFH6gVnq5hgF6sA+gtfSfdoqSEVHk3JQiJ6vVcnGzAgfSBLNMLG6K2zF2vkCMSh6A1LP0mBnVPWWLdCK+bvXdgpHCEzV04ilI5YeSivOOioxahgzZiFRde4bjQqJ4I7gkbWuETVo84lMVSM7F4uGEdRemhCDmwfDPmm7rbx0P7X9b7eTGM0HlOoyJxHVC2/Y/MIDPVTwcwikxdo4uQzaNAzykgnt6sneM7lIGcxFKA8cMVgeFRqPBSZpnFcd3yqGD2PmgQGiI4yV9WjuiAyp3ZOdKZ39Y4JUps8uhJixNCwXtQv4ULSh7Vi3yqOA2zFGqlgqvxkTi+G+AVANo+v5SHOEdNDJoNslCjB9jh/WOT4IqDJm/1CFpdODTo7qWjyorisHEH5wiGjvbQRVvjUKWBELU0/5S2PGRW04PvRUH0ZJ39Oww9IGywQFQErocJ5LpHLol9up0jTg0w5XXZPaIeo78cmUROqAQVUHU6Ith+NGKVE3+ymg3337aVf6TLdjjKUfUP+tKaApoClgKwXcTpXOhrc3B0ALbuZTsnVcHTqeMZjMAqF31fDKOtRhBScz0Dod1QvNg+QrOEXvLkUBjpMXwn+I7sMwIt0co4DbMUYCBcyePVtywCCdvVH1oeXcCANylvuNRM73JFQmZKYGQ46quu3bu08eeeghSUJFPWdvBvN2tjEjnFr3bt3kocceldjYWGcno9Pfn9up0qyPwlxYSiHO3s6iTGYm4KuaRDcpjoVzhnuuDxSYdqgtY+QAV/U9MYd469YtKrWyqq/lSP9kiix3GwwAEOZTmweQO9JvZZzrUddTOnftrMqnVkZ/tb0Pt2OMrjSgC+bPR2raUnnsiccVAkx1qK2uRB9nu9cE4GcSF3MMcp9HjBpZItjb2e5V349jFNDxK47Rz6GziaYTBxzAbdu2WZ1D69AF9cl2U4BQYUsWLUIapq/07NXTqSR8ux9Kn1gmBTRjLJM0Vb+DkFUDBg1UaNK5OblVf0F9BbspQCDb5cuWyYCBAyUECDxaureblC5xomaMNTxMrBu9FxiArPJXFoBpDd9irb88UYwoLUZGRsmQYUOVt77WE8XNCaAZYw0PcPcePSQctUU2bdiIEgpaaqzh4bB4+b1798qKZcul34D+1eKlt3gTemO1UkAzxmol9+UXo+e3b7/+sn3bVoW4o+MvL6dRTW5hdMPiRYulY+dO0q9/f21brMnBqMZra8ZYjcQu61K9UNPl6NFjqFi3QzthyiJSDW3fvGmTrEXBs/6wLRLUVrfaQQHNGJ1gnJvFNINRfwC801tR5lOr004wJOoWWK9n6e+/Q4UeABDi7sV1xJ3l/vR9VB0FNGOsOtra1LPhhDlz9oxOjbOJclV38CrU8d67Z68MGjxYO1yqjsxO2bNmjE4yLHTCsHjWxvUbtBPGCcaEjrBlS5dKP4RTse53dWUBOcGj61sABTRjdJLXgGUO+qDs6k7Uj8nMzNQgGDU4LnSA/b54sTDTZTCkRT9UFdStdlFAM0YnGm86YQ4iz3vXzp3aCVOD45KSkiJLly6TgbD7NmvWTKPV1OBY1NSlNWOsKcpbuC5RUQbC+8kUwVwUhtet+ilAoAhKi/mokDVk2DDx057o6h8EJ7iiZoxOMAjmt2BywuyWc4DgMiCuzPfr31VLgVOnTinb4qBBg6QBKjE6G7xY1T697t2ggGaMBiWc5JslNYOCgrQTpgbGgymZi39bJCEhITIUqX9aWqyBQXCSS2rG6CQDYdwGAUf7IsNi+1adCWPQpLq+D8O+u3I5U/8GSFA59Z6r6370dWqOApox1hzty7wynTD7DxyQPbt2aSdMmVSq3B1G6l8z2Hl13GLl0tYVe9OM0QlHLbZ5c+WE2aqcMDoTpjqGaBcWoRWAFesPoAh/HZ5THSR36mtoxuikw0MnzJ5duyVRO2GqfISY+vc7YMV69OwppLu3t3eVX1NfwLkpoBmjk45P9549lOSycaPOhKnqIdqwbr1s3rxZ+iPLRTtcqprartG/ZoxOOk5U5/r27wcnzDYFR+akt+nyt8V4xaV//AEVeoB06dJFp/65/IhWzgNoxlg5dKySXuiEOQAnzO7du7UTpgoozNQ/litg5b+BTP3TwdxVQGXX7FIzRicetxYtWwIctZ9s3bJV8jS6d6WPVBZqjy9j6h9U6FagtTOVQ630h9Ud2kQBzRhtIlf1H6ycMLt3SVJSss6EqUTyU1r8A1iLyclJMmQI6rhoabESqev6XWnG6ORjSDgyluzcsGG9hiOrxLFKPJeoUv8GDhgokU2iNFBEJdLWHbrSjNHJRzEgIEAVYdJOmMobKOags+ofaqDKsOHDy7QtUqrkR7faRwHNGF1gzHv37gMnzH6FJk0vqm6OUSA+Lk6WwelCoIjQ8LAya0SfO3dO+NFgHo7R2xXP1oyxikeNSNBJCNIm6GlqSqpkZ2dbrB+dmZEpeXl5FiWUlq1aSq/evWUr8qdrmxOGElsOINjIoM6eOSupqanqb0vMKi0tTZjaV15TQBGAFWvYoAGAIsqHFZv59dfy6cefSEZGRnld6n1uSAFPN3wmp3kkMsVvZ82WWd98I/HxcRLdJFqGjxghI0ePUgCojFWkJ7SwoFA+/t//UBBroHTt1tViLB2dMDNnzICzIEUCgb5TB2pgbWhkSjO/niHffTtHzp47I82Ry3zFyFGKho0bN1ZB8B4eHsLslTdff0OmTb9VmFLJbZbagf37ZTk80VNvurHM1D8yYzLQ6bffLt5e3kDw1tUBLdHSnbd5PIfmzg9Yk882f948+frLr6RTp47SFEjQCQlnZMO6tTIHk3znju3q1ph+tgITdeHChXL95EkSGhpq8ZbDw8NRse4P8fTylBYtWlhknhZPdPGNX335pSyYv0B6AI4tolEjORF/UtauWSVzZn8rB/btR+U+L4GuK4t++022bN4iV06cqGDbLD02pcnZWKSCAoNA68lC+62lRnPF/r37ZNOmjUJpXacIWqKSe2/TEmMVju/aNWvkxZdekg5gjCymRKlm2R9LZe7336tSqYxPvADphPvuu+++Mpkib5EYjcyE2bltR7kOgyp8nBrpeiuY3Rtv/VOaF0mBNEX8vniJ/AAabtm6RVaDxrA/KBo+/eyzCkuxrBulKWLl8hVy1333okZ02XVcyEDXrl0rGxEJMGTo0DIZaFnX0dtdnwKaMVbhGI4dO07atGtbLN2RAVKN5ufYsWMI3N4i6enpkIZ6SJu2l44r65Z69+kjv/z4k+zft08xgNogyUyeMkWaNGlSrBrT/HDlxKvUh1lB24FAVAAJjw6qmOaxZQZpqxrRS1AjGnVcmFGkJM0yCE1JkvGNPXCcL4qU6Vb7KFAH9hQdj+BC4/7aK/8Q/8AAmT59ugZTtWHcVgCA9r8f/Uce+OuD0q179+LFylIXtC/ecP0kefixR6UnmSgWNN1qFwW0V9rFxtuUCbNbeWf1mmbd4JmAIpZKf0iL7dt3qJDRHT50WArBHKm+a6ZoHY3d7SjNGF1sRFkTpi480usBlUWvt27lU4CLB1P/GLvIGtG+fr7ln4C969atk65A2vHxrfjYCjvTB7gkBTRjdLFhC0ahpn6sCQPbGh0RupVPgXTENi5bulQhojOMpyKgCDJSOl26dOsmvpoxlk9cN96rGaMLDm4vOGEOwAGzHzF5OhOm7AFkEPjvS5aoBWTw0CFlpv6xB9oVyRTTEEB+8uRJ6dIVEqN2vJRNXDffoxmjCw5wmzZtpCscCPRqa3W67AE8c+aM/AEQ2gEAimiEYPDyakSzQmBSUpKSLq+7/nppiJrStSWIvmwK1t49OlzHRceeTpg5380RqoqMcdSTuORAUgJcsmixildkjeiKClytWrlKglEyNax+uIweO7Zc6bLklfRf7kgBzRhddFR79OqJDJpvZf369TKufn09kUuN49GjRxU6N2MeQ+rVK7X38j9v/9Mdl2/UW2otBbQq7aJDH6KcMKgJs227dsKUGkNmrvwOoIioqCiVuaKdKKUIpP+skAKaMVZIIuc9wHDCEBiBmR26mSiwd89eBRTRDwWuKlKhNc00BSxRQDNGS1RxkW3t2rWTzvCeEjxBO2FMg1aIBWIxQGhJF9bLqQ1pky7yurrUbWrG6FLDdfnN0gmza9cuyQQ8l5EJwyJPxu/Lz3CvLQxXMn/WTZs2yToAQAwACK2/ruPiXoNdjU9jN2NkgnXBeUC/V+PN6ktdToEePXvKBXhgN23cqIBucxD0/d2cOYpRXn60+21ZDLixlJQUxRwVUATDc4hr2RW4loB0001TwB4K2MUYyRBTcwplY3yWZOSeB3SWPZfW5zhCAUNKqgeP6xWjRoI5XlDByR9+8IGC5MoG6nVtaIvhZCHGYnJyMkwKm+XIIdSIRjlU7XCpDaNfdc9oV7jOmmMZcjqjQHILLsoZfPePCZBwfy/x9KgdqNJVNxzW98x0QOVwQbZGf6QI0gHz9JNPyepVq6R+/QYl1Evre3WtI7k4MAd6MeIVSYueKP9w4y03S1vYXpn6x/06vtO1xtRZ7tYuxngiLR9qtOkR0iAxLjmYLj2jAyU21Ee8PWuOOXIiGJKUQWAj28G8Rggni6tPmIMHDso7b74pCWcS5AxqoWRmpOPZTU+tHDHGHwYh3PA7G7ZU1oJhHZx33/mX9OjZS4Yi9Y/SIpFx/IDd6KzB7+W9q244VJX+SOXRr/Q+e+a7XYyx+CnrYEXGZCw4X0fWQYpMzC6Ubo39xc+rLhhP8VHV9iMnOwdpXYkqrs/L00sCg4MgPdVXBZJOIf+VUgUnTWhYmEJldmXmyNow026bLs/9/e9FjpdLZC6rqNalI9zjV3x8fIkwpc0oRcAPpcW2bdrKS6++Uiw9OtsTZ6RnSHJKsmLqhDYjOC7fS+1Ft26kqDHRfEKbOuv7BAYGFtOPBdNodyaAMfPdmdFE8JWKAETMr2yXjdHoIBD1RwJ8PRQMFlikHDybK78fSpdzmQXAs7tgHFZt37m5OfLt7Nky6brr5J9v/FNOnjihwAFYUOnvTz8tTzz2mKxcsVKyMjMvkyyr7SYr8ULDr7hC7kFJhAaoeGfO5OmpNZeQK/GSTtVVHNTo0lUByWSIiP7mu+9I+w4dbJoM1flw6ZDwP/34Y7nphqnywfsfCDEglWmkOm/Cha+VnZWtbOm33nKLvPziywrV3ngXyBjfePVVmYKc988//1xOnTpt83xwSGL0RCG2oS2DZeuJbDkJ9boQvDApq1CWgDl2aewnLev7iTfsjtUlPYahYFRUVBMlFd755zulU+fO6mWjx7YT8PWm3Xqr24ED3DB1qpw7e06+wAtAZBiqEQUF+VgQqn9hqu55Fnf8eAnGSOlg0OAh8uzzzyrQCPPForrvraLrsVwDtZnIqEi5edot0g0wZ7pZT4EGDRsogSA4OASVIacBhHhg8SIYExMDdPsQBUo85YYbhPG+tjaHJEZOwgAvDxnSIlg6RVKFhv4M3Tq/EJh28dmy6gjUhZzz1So97ty5A6VJY9QLx3i+7xG6QpvbvZCsIiIiSkhWthLLWY+/59575cqrroI6hgJPRSaM/DJqVDvrM9hzX/Fxl1RpmkiuumqivPyPV6RxZKTTjzNBLli7p2279sIKkLrZRgHynr179qjqm0z9NFeTqSHGxR2XzhCG7KWtQ4zRkATrYjJ2gW2RDDIM3mmPol7jIUUuOZAqh5IKJKfgQrFzwDYSWH80GSCzQDpAhTp9+rR8/J//Alevq4wbP96tU8M8ILo/8be/yZBhw4vBJPLySgY+W09F1znSkBiZ9jdt2q3y5N+flnCg47hCOw5pl7iP7du3F4Zc6WYbBdKAKsViaCwlHAQbIhca47Nt6zZJh/ZExmgvbR1SpYvFk6JnahTkJaNbBcumU1lyPClPBYDnIIV33fEMOZXmLZ0j/STUzxOMs0issY0WFR69Y/t21EJJEUqKLHz01NNPSSRUFmdWqSp8KCsPIGT/cy88r16IdevXSV5+rlvYUct6fE6Ck6dOKVixe++7X264aapLlTndhlKu+QWFkBjbllkHu6xn19tFgafQdBQeXh+mpLOSCmeL0ZYt/UP8AgIlullTu51ZDjJG41YuffvAIz2gWZBEBvnItpNZkplfKOcv1JG41Hw5m1UgnSFZxob5ig/CeiqbP65dsxYvWbBwNdm3d6+cRShHI6hV9FrVhhYaGiovvvKyPPjAA2r1pLrhro0gtFSfn37mGRl35QT121WeleOyDTXFaWcMh52xNizclT022wDS7AdPfkFBnhw8eLC4e+bKr1m9Rjp27GC3tMjOHGKMhipdfFdmP2LDvCUiyFO2nMiSE6l5kotsmVxIjxvisiQeTLJrlL+EQXpkUHhlyI/M/NiAIkZdID7/+e675LGHH5F/IwvkhZdfdlvbohm5i382iY6WF158UZkOqpIxUmKz1/NNRsDFyhGGkInIghdefEE6dOrkcpX8TGrgfnjPe102eTlmjtCl+EVw4x+Muti5c6cKyZp49dUShQXGaBSIfJAK2rVb98toaxxjzbdDjNGkSlMqscza/CE9DowNkuMpPrId6nVarkl6PJ1egJjHdGld31faNvAVf5+64lEel7XiSRi6cez4Mbn6mmslGszh7nvvkWcR47do4W9y3aTrXUrNsuJxyz2kQ8eO5e6vjJ1cpRkbSuZYzOjK6ZhvCY/lJwA2wW4IqaHEZ29r3bq1vafW+HnbYfKhGtilG2xgIZfsi2SK6enpalHTZVvLHibaFs9CYxg+YsRlzI/lPs7jHWvbtq2KbSy7l/L3OMYYrdTUmoV6S8NAD9l+Ohe2xxzJgfQI84rsTsiBNJkvXaICpDGkSx/Punar12tWr1bOnW49uqmgzlGjRysE508QK8bCRh0hWdQWlbr8Ia+cvbuwYr/5+uuSmJiEiRwgnTp1FH+oNmU1vqxkpGeQqRMYECTfzv1OGjVqVNbhbrtdqdFwENaD2SMSZh4vby/1rNzOmjObgQ7UDfV8WHNGN8sUoBrN96l161YSgMBuo3HR3QoTRXR0E6kXFuqQ5O0YY7QsKBr3WeLbD2E9fZsGSLMQb9l6OktSEO9YCPQJphSuPJwmjbC9SyN/CfX3QOyj9ZkzVOkoWq9Yvkyim0bDK4uQlaJ287RpshRoK++8+bY8/9KLaiLyRdSqikEh+79ZMOrI4cPyzcyZyD7IlauuuVomTLgSzPES/Uv3zsm/cMECefOfbypEoNL73f1vBnCnpqTKRsTVMobRE8HoNAlwQjPp4JP/fawQx4nOrtvlFOBcZ1YXy3mEhYUXm4s4nxncHXc8Dp7q/dKzZ+8S+y7vqeItHs+hVXxYySN2ns5WiDp+cKC0buinMl9KHlH2X0HIlIlBTrUnmF9G3nkpuMgwnjqSmXdBjqYUYFuh+CMe0ktJjxUHhzMsZ97PP8uyZctwjpeEhocJ6wcTimv+L7+oWDEeQ2nF199PGbxrg+RIJlSVCwD77tmrF+i7X47HHZN1a2DfhWTeGNX4yqIvz2kFFfgQKvL17tNbAlHEy9ZW1c9l6/3Ycvxu4GbO+PprlL7dC4aYjTzvs7IDpSl++3WhfPXllyhsli5jxo+rlZK0NXQ8jSiEXzDX169ZA7jDOnjXGin7Ik0ye3bvVllvhw4eQo68H2KZm6oAcPP4RmuuYRxTBy+alQqxcYrIjC2JCkQi1M9DxrevJ552updpc9wJ9fpEaq7kQ72+AAbJBp4oTev5SLtGfhIC+yOZqJ2XuHTTteQXpQ/GcxrpZWRSfgBsLYtZOUoWgjg8cO99UGE2S9PoZvLOe+9KG4SglHe9FcuXS2uUgKW6aIB8VHQf5s/FV5YvfFU+V0X3U1n7Ofsu4B+bJ2Fl3YCb9EPOURcLL/6vlOYQYwzz95Rx7ZCc7SDXSgB02a7TOcixRuwjMtkuMci6EhkMQICGvhIW4CXQxh120lQK1Zy4EwYOz4OkzMT5/Xv3Kcl5yk03qqB3e1fPih53DzIQHnnwITl0+KD07dtf/vHqPyQKdp7KlFiPHjkqC+bPU2AA9DwyCoGpdIQYK48JV3TvNbmfwkAWtKaELGhOzKfVzW4KEPGwMXhFkA8iXcCPHGWQTsEYSQ2unCfT8xWDTEFUOMFwDQZZF5JkwxAPaYHc6ygEkRPajEHilbQ42D0YznYi7Sz333Ov/PXhh6Rlq1aKYfz5jjtgqL4or7/xuoqZs+aeKZHRbktGai3TYQ3n5597VhJOJwjzUx/EPdCOVhnMkdLvvXchBAvZPcx0YJ933DpdOdleAhO2N+3LGlpU5THLD6er+F4CrlysA62oKi/m5n0j3kFJjG0a+KksPH9vx6jpmPOF3KySGjl8EzhgIoO9FYPceyZXkjKB+4jgcLw2kpBeiE+GhCD2sVk9LwBU+IgvREiuDg4KrJX0BDXfDavjUbVlHWWDoTEX91c4PAzV2pq7JFOc+933iAXrqmyC1kiaI0aNlOMIl/ro3/+GrWeWNIuJkSlTb1CSqzXXLO8YhrdkZmSpvgzVm+r6imXLbXqu8q5RE/soI3IRqgOm6EEIP0fFnJp4CCe4Jml44SIYoWJHdfDlOF9yiDFyHB2/hZKUJZOLBoNsAgaZgHjH/edyJAEwZpQgsbBKGkoq7MBn77lcZNd4gkH6Az28LqRIvFy1XIrcvHmTCm7li0I7IydaCjD/hg9HDjViB61tuTm58uEH78tf/nK3xMTGlkjQL6+P6bffDkdMnPw0d6786513YBiPEkKjORKvyOtt2bRZvH28FBMxnovQUkOHDVPex/LuyXwf6UKvJj+8J2IfUsomth+RefipKebUIMBbgvzwDms9yHzIrPqdDBzY5OwCOV/ko7DqpAoOcowxVuEgkuk2DvFSnxQ8+IHEXJUxk1sItB4wSMZBHocX+3hKGqRID4UeHhPuDY+2B1ZfGGJrIZM8CyTvgsIC4M+dklXAnWT+dDuAFIwZO1YqCgEh0+Aqx9X2AiIF2NRKDGcOHR9kGBUxDUqpjz/xhJxGBMBqxJW+8sJLSnIk7JMh6VXwPlrczVxYSrwn4uPVczHkqkuXrjJ6zBibJNIc1MFZMH8+GPcPMm7CBJlw1ZXCeMyP//tfGY9Qo3ETxjvMxC0+gBUbWyLRIQbxvl58eXWziQI74Z/IYOpxJZZWd4gxKr5Y2SKjBZKEwsnTp2kg8qzPy7HkAjmSnAfJkSuEFEmR52VbTrbsSshFGqKHNAv1VYZYBowrJqkmtYWO3WwTQ5TIzOitZfYPx4fQXEyf6olqgkShKYu5ZWWaygSwj/TMDCVJEQ392JGjiln4BwYoL7KhopdFOiIpP/v88/Kn2+9Qif1Eqmb8mSOMkeezUeplyuOFC+dlCwJ5d+/aqTJoynsu8/vkcddPmiTJCKT+/rs5Et6gviQjQP0mgJ22QaaEzjYxp1bt/u0QY6RUV52NQeLtIjykDbzULMJ1JDlXTiJzJo9qNhhCISSbk2n8FCgPdoNAL4lG2E9EIJDGYYxl2iElSarr7trI+Oj0IJwVWwSyS269+WZ5+NHHZCwkRx9fH4uPvmrlSnkVWIZpqWmKuWZlZ8nH//ufAsDlCUOGDpVngd5DoIqKGpljNPJX7/zLn1VsY2UxHIb3EFKOrT5Qy/90223yxP89KSNGjlBqcEX3Zewfiawoxg1uBzzVXffcrdFtDMLo72IKOMQYi3up5h9kbI2DoWbjQ5zHI4A4O5qSh1KuzMWmFImUQ9ShOQUGyQ9s21LPx0Ma8Zwgb2BGMrvmEpNk/JO7ttjYWAQMN1bB7oMGDyqTMXbr3k2ef+FFJSkStu1FMMExo8dKv4H9Ybf0kQYRDZHKV3bKn0E/2gDfe/ddaQ90k5GjRinp1dhXmd/MlQ5F9sP8X36Wfv372cQYmzZtqjInmkL6tMaxVJn3rftyDQo4xhhpl6rhxsJbHRAI3j7CTxIBa0Z4M+ZfZ4FhXgCDLMQ9XoRnOwVI4vzQ2007DoPTG4JJNgjwBKOE80YFkZukycoMFK1h8qjLh4QEq9Qzqtn80PngURdpl/h4EpINCwMlS37YKDW+8fprSr0cRscNVHNrGiGfvgNiejKw8e66+26b7H/W9F/6mHrGc2GcSzd61impWjId0FZJx8uePbtl9LixylRg6TjaNUkf0slSo+2Vaj6Zq6XzLZ3jjNtoe+UzGItEXi7iiVEegymLdEg5YgZxxue15p4H2Ow3AABAAElEQVQcY4xOJGnxVqg689MNkGZJcNicglebqnY6gmgpRXL+kFmy9OvZzEL1IZHAEyXEFwwywEPq+3tL/YC6UL3hxKHajf2UOMFGyD9conFCkwFysvL3WdSE6Qk0GzIDorqwSD2lpYvY3w0I5z7w0JZoeE4eW5E90fwcMom1a9fKkt8WyTPPP2d1zKR5HxX9Nn8uMr5z5xJlyODB8Fh7X3bqWmDy9ULaIe2KbGRgBiNbv2GDjB47Rv5YvER5pAvyCyQMqaSln3fTxk2oF9JWhT9ZYnxEwmEqWnfQ1lHP+2UPUE0b6JX/6osvlcmjK+rO0EP/3bdzhCFSffr2Qf77BFV9r5pux2kuY3kptPL2nJVPUOJrgEwZllsY166ejG8XKr3gvGGaIYF0LxZ5XY3HpPpNl/+hc3kKbXzenjT5eXeKrDicAUQgFvpCSEc+HBvGCU7+TeglqrRshHlnDYyJAHlgic5vZ3+rcki7AsHlpx9+UIW0yETNGyWtwSgq1aRZtJIszfeV9fssPMf/euttuROB2FRVq0LK2LJ5s5J2eQ+bNmxUvydMZK2by1V85sua3wPLCGxG2M/PP/0k3cEA+vTtK3En4mUzAB3WrFldTC/z5/OFPdYDklRZjXQirL4rm2I49vEIsSJGJBeORcjbJo0iYDppA3OFNyTG2tjKHnUnpgYHkxIDC61zSlPcp4TDiUCGwH10APBvDvbFvExpGugtrcIDwSQyJRsS46mUXDmdlispudjvCXXBwxMM05S3yoybnIKLcgI1a06kkRA5ihq+ALeoBxAMSpehqG1Tz6+OBOM3Pd+UKGj7xFeNS5fMQyYTSU5KVirRZ19+UQyesXLFChmMovRUm+ik2YwcZ3pnzdVlSll0tFjbCJf1zFNPyy3Tb0XR+x5WeXeZ1tccWSwcN2tby5at1HMlQlI8D8/053iuSBRCMmeAlCr5jOzXXMo7Ae/83Lnfy7XXXQ+AgWZKTbwev8kwb4Rzypy58v1aD9BjMgumHJr3Y9wrJas/lvyunqEsVds4tqq/eS+kgTEHeP+0E/NvMm/upzR8EdpSXn6euh2Ot7H/iSf/T7xxXCGk6tMJp/FMzVFcbaJ0gJ2Y74maazC/UOrmcdQwDOma6EC8NunOXDQuSDTV8Dxuo0TK37wef3Mf6UkJm/dmibZVTS9r+neMMYIJ1ERjXNuK5StkCVRCMsKBcCrQ0E+o+MWLFgGB4xcUh3pCYoGywxf/hWefUzFv4wGB/9T//U26du0mmzZukBPY1zy2uUy4dpLEdugqaQWeKlA0FZzThPpDRlkUzwdmmQtmmYAAyoQMBFGKSSJDRJAEwrET7FsXTJPMEowTH4L0klEa9srqZJgsAFQWkOu5xETxxULCF9vby1sBoxrhMPaMpXK2vPMvQMl3tDqYm+d89tln8vjjj1+mcmdCuiWuozmzM+6Lha5YI6W8xslGxJ/2HdqXmHT9Bw5Aic0BJU796yMPl/jb+IOTNQG1iMkoyGQsNS4eNB20a+9YjKalvm3d9t+PPgL+YJhceeWVKj2SjOzF51+QYcOGokDaMPnX2+8g1KoBGFwB7nmN5GTlyKQbpiC+dYwyNfDYIUOGqAJyv8O8kIiF7uSJk/IkaiZxEWHJkKVL/5DTpxJQ5rWrjBg9SpUkJXN79eVXlBmC9ZZ8fPxk+m23ytzv5yrn255du2UfkISaNouVyZMnIb72pCyCqYXM9UpoMLxmkB0IS7bSx57jHVSlq58zchV69plnMFBLVU7utOnT1er+zJNPSRIkiSuA6ksnAAFqEwA39haw/wqwSg6CLYqrFsEIPvv0U+nbr5/88803pT5CQN76x4tSJ/OcdIv0lZGtQ+SazqEyBt+9mwVLawTe0kkDAVHlaCNiSKH/1EUKFxuDzVPh1IlDsPkOqN0rUDL2l92pMndnsizcny6rj2bIjlPZchTxl3QOZcHAmYeTCCBgyuaxPPHsGUxrzqmPCVSAlfsiJGlKFWH4myu6PY1jMeOrr5Fdk4L4wOuVBEKJoKwPr0cb55cIlUmC3dMSz/n8s88Vs7bnfngOGRkx+Vq3bqPux95+tm3bhlrkncv0dtNhkY/3ql690GLpyd5rOXreaeSnk54cDzbS4PixoygMZ1KPCbn3ycefKMnx/gcfROxnd/ni00+EEF3UqOKQypkK6ZhpnVw8WOP6gQcfUOmgX3/5lcyYMUMBPd93/314Z8/L36EdbIVGQgmQyPk///gj7Kw95cqrr1IOtyNHDiOI/kclsDz06KNy6kS8PPrww3L02HG5+557UIepsXzz9QwVY+vos1fV+fbNiKq6Gyv6pc1s9+498s67/1ISIVd31ov+0+23qXKKffr1lUefeFz+et8D8o+XXpYdwMB7++23IZlcKqs59capKtCX6L+P49jrsPJvBPhlFBCVfcE8mX9NNJ8wmq6QTcNGUF3iR6ZAmqR3OxWhQczIIXPji8hcEU509RvSJR083J+SrU7HPyZ1nI6eQCCAQLNXSCAR+BEd6qWuaRxp7zelGGXzgmpTVuuD0JYTJ05II9AsBRUVGdhMlcrWRnQbqqxcZCipLIGkUZ6tjRk1Rw4fgc1zq+zfvx/nXAX7HVaZoka6UeW7BYg5fAbzpp4LkoUlKdL8OP5OSkyCGSUIDM32ZzL6ImNPS0+TcIQDGSojGQiZIe+F7xztuDQFeMIE4wpt2PBhKnedgf/UKDasW6tMBWRuRiMYBzOk0kLSEEjfFPnpGbJy5QoIIFOF2hafnQ6t++89DrvsGmkJGyTb0OFXyOQpk6URsDjJhDmW115/nRJSaNJa/sdSZfKYAimVICBU53ft3AVtzzQnjOs707eDo1q90g4Jdxio0ecxmK+8+BJg4S+9/LwTqs0caAY3cyA/w6o4ffrt0qpN6xJSUefOnYvPpX2S6sLphAS14pIxWmpkliz9yo/ReM0sOGUyci5IMhhleu55SQMzzMi/APALsEocUJphgp+ofO809U4UooriBcRXmqCSjH7t/b7tT3eoZyivlu70226T+fPmgamtlMFQZQivbw3DKX1Px44dk3++/gbsmImKOZbeX9HfqrQlVCqjcdw++uBDqF1NMXZAAsckNNpfUNyMam1FaY08fhOcKQTMvczTbnRmxTdh1NrATmvu7ab2MWvmN3LHn+9U90G0aIbyENWFjMBZbWXG4zbGOPv5mt5tMis6lShhkuGX1ZhaWogU3JjYmOLxoKkiNjZWhXRxIWNr1CjiMidNA2hiNNWwMYWTAMaca2zeiItlooUzt0uz3K675MORPVTfQ3p5eaoVj5UAmf3AK3NwOUgtWrZUxGd6G3NgyeR27NiuvLJ8GYyXNxdxWnyZ2XgubV4cNFsZBK8diLAefpjXbTRm4SiGCTMkJct0lEfMgJOHn3xMcEwjfEzhQ0GwT1bWO2KNvYbH3DB1qpoUfMntbauRKcMJQueHPa0LbFWcIEbjvaSmpcrE7lcXTyBjX2kJ0thu/s1JzjFlbOK1KLtg7kwyP66832TOfB9YU6T/gAEl+iAj6YzQJkO6njR5cgmJt7x+q3of32tzOzGdkuUxPGvvh89KmtJmaTB/fufl5aqxM6Rpjp3x2+hb/W3GFnieKzXHGCMe1mA21fXQ7ZESlo28aKowLcEIOXhxx4/L7Fmz5dZbb1XG5y/hreSK/jJKp7K+yM8//Ih82JuLVTSGZ3SC1BhSL0SplZR+pgLM1XyiOvI8TD0MhrocjHkfFUISm+IE+W7kYAVOJ5NEKQcimDNdkdJodTdHmCLvlR5ofiqrMaiYNkh7bXZbUWCqEJ5qvh80rdi6yPE5qOqfgVeWEjcXWXMa+WORHX7F8OJ+zc0AlUUDe/vhvdGDzrAs3vtBVNEjLR1tMVj4OEe2b9uuzFZMB01JTpED+w8IA/8pbLhrc4wxKnmteklDpJahw4bD0/aWDBwyVBpDjF+I2CvGnDFtjYP2/XffyZ133glD8kCsbvkKAqtn717AF+ymbnbB/AWSC3tR8xYtVXEmlhslyAJF/qps4JcK/YcIQI1UuZNLElNVXtcV+qZkT3tnWbncFT0DHQqUbCgp2sMU2T/tX6wjQttz6T6skVoruseq2k9J9gt4+efMmgWot2hZvmypJJ5LgvnIZMM1JDpDiOE31VzjGY39vD9KenTG8RhqF2PHjUPA97fwUscDKSkWNseVKrZ1AOYWTRvm5/L80n1zG/tjqJBxfV6XZjDj+jzG2ZrHc2i23pRRDItqYPNwqKCc8dXY+sHB0gII1UexwhP/b9jQobD9/FnF5S2Dt7p1q9ZyzXXXqdg0IllTzaKU2bx5cxXVP2XKFDVIRyEpMn/4LqjlLGdpDFw1Pkqtv5Sh8s1DiBVrTRN8orRaZg2ReA61B0fGkBOVJhVH+rDmXo1jjiG/n3ZpmlaaokAcQ72YbWVra9mqpQSBmbPeMs1aN8C5GBkVqQA36BBh7G5MTAwC75spuylpno34w46dO6lg/wwU4WrTto0qLEU1nFI77fK081IQ6YTSwymIJqBEzd+0+cY2j1WMLSU5Gce2UaFyZJJG3x2K+iZNKclGRDTGwtNcmbe4gEE3V55ugio72pjFdg4RH4Uwl9aH0zQCOK1M8XWk2VXaYNa2JOWNbYQwlmEtg2pEFbT3oSeMHac8n1ddfXUJG5K9/enzHKMAEccZD0eJh+Vuy3McOXYl5zt7OUK7TqTmqVjZfjFBQKbXeIz2jBLr0+85m63ijFujtEHHRr4qpdeevoxz7FKl2+LiBRCNWQrV9vXNuHTNfBO81d64vZq5Y+uvytWaRm57JC7rr1K5R55hNACcHtNQwyUo2PZyqpV7N7b15ii9o4BUHwTHHZ11IUgQsENYtO2G3fTocADBtAaSfz5CPhoEQnNwUFokmeySGF2Zvj8hGJXqAbNiKPq7UyOTYcgSazebe+Hd6Rmd5VnoBT569KiKaGCWkeGtdpb70/fhGAUcU8Qdu3aNnD0RKjRzid2NKZKY9K7/DWl2c2Z9C0Sds2rSmodx1AjB3eiilMZpr2Z+8D6Upn38kUeRQfKZChVzo8fUjwIK1DqJ0d1HnRkZXyCtLgdZBYMGDRYC0NIAT4eCMnvw2/yjCII9NtpELjscfZbVeOU60BNpiOeH1+d3WSo/VVTjQ2akVFZuw+8yW4ldjBK1sRWfYDqXXlT1H65pxOBlIwSGDo7Vq1YhkHyTjBhxhUy58UblwKguh42NT6UPt5MCdjNGxkkxJ7n4fbLzBqryNL7QzvzCUmql46EsBuEIbQ4fOiQbN22SQ0i/O4Y4zzMJZ1RpBy94XeltJAqKr48vUIUU0iTohKsVMUxet4iNKpBW7jOnI3+rjxk3NRieqSPVlekYHMvQkFDgHdYLDpGQ0HoqAiAcOdrMxiAN+Pw8X2ViQEVNgqeTCDqpSFlkfB5zflNTkpGilokMGOR5FzFI9VXEuIxtvHcW8+Lf5ttMf2Mnt/MgNMLPGcexL5KAf6soBrzfuUgNZOIAFxlvhJxEYIGJjWkunbt1kf79+6syD+Z0UZ3a+A+DyplqqCV7GwlndjjfH5oyHI1KMOvSfonxjddeU/mOhH9yxlYIFByW2CRIAie/M7YeiJ1k6BAlOjKGqmyc8AyiZtgSVcHsrGwEAeMbE59ArQZDU/dADlHUjJ/mDMDSb4OR8rTi/eSoRY0/yXyycO14IGiTeTFOLiw0TGUwkVnv379PjVkK0F2CEcZBwIsQhI4wvpTnFjezP/hcbJfYnekog+GZ/jIxvOLfl/3ABvRDqdbPzx/oPvgUfQeoRcS/ymyIB7BwffbJpxKPOEFnbcyLZ1Pv6KUhdZrbZRjSgAH9AVox2Kq0UWtu3G6JkQPK1dx4Ma25WHUew2T2j/79kTz08ENOi0AcDUBXFniqComxOmlty7WYfnnDpMnywUf/liikE1JSIlrSoQMH5eHHHoXq390t7b9l0YhADXTiUGp0xsZF88DBA8iz9lUVGp3RycTFLCY2plIhzOwK1+EA0oHhzI2YfPUQmd8F2S6NimqZOPP91pp7g2SWnJwEIBCTpsFF4aVXXqk1j1/6QQMhNTM91RkbA7GJz7hi+XKUOLgS4BxdK5X5OOMzG/dkN2M0OnDWb1XMBylRxWqds95oLbsvc424lj26Sz0uodcWLVwoP/30s0JhGo48cWtASlzqIcu5WTdmjAVAP3Fe6PRyxkTv0hSoUQqQKS6YNx8YBL8qFHDCwNWmjCQS360ZIz2eWmKs0Tl22cWLx8PMgXLZQXpDjVGAudI///yzLPv9D7li5EiUBBldaQ6NGnsoOy7stoyRXmlCQ5l7S+2gjz6lyijghO7NKntW5++YTlRGKLBC4Dqgc48cPUaGjxhe5bXBnZUybssYCwpRLB2qNAM5dHNCCmi+6DSDQqbIGtk/Abd069YtqnDc4KFD3RpvsSLiuy9jRPA58eiKVbeKKKH3VwsFjPFw1jCvaiGCE12E45CMgPof5/4ge1HSdtz48Sp4nZiUtbm5FWNkNg7x9BQwJgbc3/9SGU4alGlzrOpA6tr8Mulndy0KMNWSsG9zv/tejiGW8krUSerVp09xfRfXeprKvduqTbew414Z8MvMDGatEOAyHwzNknTB7UzdMt/HYkUbUO2Pg83A6UGDBilIK66ItJ0QVNP8eDtuT5+iKeAWFCBTZLGr2UD9PoFMpImo89wHJYXNi5C5xYPa+RBOJTEyb3Q76vnO/f572bF9h7DUI2tLsNYtwwWMvGIO6oIFC1TgdnegPhvR+KFIIXvwvvuA3B2kCq7z71mzvlEF2PtgJWTNaUOVs5Ne+jQHKWDQXy9QDhLSgdMpfMQD+Z7zjALINddei0JfXYrnkQNdu82pTsUYyRSf/NvfVP4s1eHNmzfJKtSYCAgIlFFjR8vVV1+jmCVr/v7w/Vx58pmnS4DODhw0UNXrOIG8U36MRvV51OjRtdqYbNBCf9duCpApskTBXNRFykG1v2tRAqRDh45VXu/I1ajuVIxx4YJf5WFg3A0dPkzZClnp79cF8+UX1AOZB1WYBuL6KAqehbCCUaNGKYnR3GZI9Zk1o4lFmI8iWEaLimoi3XtekiyN7fpbU6A2UYBMcf++fapYHCV3ApiwJhJt77qVpIBTMcbmLZortdkozM3i63+5+2657Y47ZP26dYivWqvsjv2IpAH7oaUUpSHDhsua1atLMMax48ZaPLYkKfRf1UUBTkqtSlcXtU3XoZlq9+7d8v23c2BH9JPrUBebxeFqE4CJLRR3KsY49aabLN47bYiDACnET0WNVf/eRahBKpB/GMLI1ZBqtD0F2Cu6lt6vKeAKFCBT3LF9u3w3Z46E1guTyVOnKJOUubblCs9RnffodF5pRx/eUKcNhwwxDwnv5a4FsByllz7fvSlAMOnNACz+ZsYMCa9fH5UYb5FmzZrpsLUKht3tGCOfd+jwKwA06qc80AxYJfCobpoCtY0CDGdbv269zARTjGoSLTfffDNQ0xvXNjLY9bxuyRgHFXmnGa4zAKE+DPPRzTkoYITr8G60nbHqxoRMcfXKVTLj66+Vg2XqTTdKhMYltZrgTmVjtPquKziwYUSE8k6Hh9dX9UXMJ2MFp+rdmgIuTwECzK5Ytky+nTVbuqAY2mR4nxs0aODyz1WdD+CWjJEEpDrdsmULHclfnW+TvlaNU0Chbi9ZAqY4S/r27SfXT5ks4Qhx0802CrgtYySOHNObtOfNtheiqo82pHetRlc+pZkmu2jRIhXvywiOq6+5VsJQnVE32yngtowxsJajg9j+KugzXJkCBEn5FWmyRN4eMmyYAoSoFxrqyo9Uo/futoyxRqmqL64pUI0UIFP8GbVZlv6+REaMHCGjx46tlajblUnyWs0Y+UKxnCczaOxVuakSsg+mW2nVveJXU6vSFdPIliNYdpXIUauWr5QRSJMdgXIEwcHBtnShj7VAAbcM17HwnBY3/Th3rtx7112SmJhocX9FG4nyk4RzP/7vf+XF559XSCUVnVPr92tA9Up5BbggE5bv+znfyVqUIhgFm/pIMEbNFCuFvO5bDMsa8swFQk9oWKh4orZx6UamRw8fUwrLyifli7lx40ZZMJ8QaBFyEefoVhEFNGesiEIV7SdTTElJgZNlLvKf98jYceNk4EATslRF5+r91lHAbSRG5oNSnS3duM3SdjK9xMRzMnHi1cKi56UbwW0ZHBt3/LjF83l8SEiIeilZiFwjlJSmYBl/A0CCTXuly6BPBZu5YBOI+dvZs2Xf3n0yAajb9EDX9lIEFZDN5t1uwRgZprANWI47kSjPSmfmk24PEEX4AjETwLyR8XVo30F69u5lEaAzOSlJqciHgV1nibGa98W8bMN2Zr5d/9YUqEwKkCkmJCTInNnfStyx43LV1VdLX426XZkkLu7LLRjjOkCSzf9lntxx222yeNFiVQ6BT0iG9vijj8kXn34qrAdj3rJQPmHK1KklbDJkqHz5eB6/2fhdept5P/q3fRTQRQJto9sFvJPxwCdl4HbC6dMy8dprpBcWdY0aZRsdrT3aLbzSfr5+8rcn/09OAK7966++RMR/X+SF+qraMbm5OaiPO+KyDJgWLVsKP+aNTphzALklYzyOFZnqOSHg9+zZI15AFA9DBkEj5JuWZXM070v/1hSoLArwfTx6BKjbsIlnZWfJNdddK50AyKzNN5VF4cv7cQvG2KtPb/VkEyZeJS89/4Ic2H9ARfyTyXXu3KVMdbk0OehE+eR//5XMjEwlJVLK/NfbbxdDlk0CuOe9D9xfQsos3Yf+u3wKGCYHc3NH+WfU7r1kigcPHACW4nfKRDQZ72DrNm00U6zi18ItGKNBo+EonPXRh/9W5RA6d+0sO3fskFtvv83qYNeRo0ZKu3ZtlcQYHxcvr7/2qky/7Xbp3MW0OhOcQldRM6jt6LdWpiuiIDUW1nqeC6boBTv29ZMnKdRtjS1aEeUc3+9WjDEIga1kjqwgmHA6QTLhiGndqrXVq2tkZKTwwxYevl+d17ZdO2XgNoBvHSe57kFRQPPFcl8EMkUu7N+jaFVQYBBQt2/QALPlUqxyd7qF88WcJGPHj4MHOl/eefMt6datm/j62YfFSJWPtWesyYihWqhVQ/NRKPu3VqXLpo2xh0xxy+YtMmvmTJQiCFWo2zExMVa9i0Yf+tsxCriVxEhSdOzUSdojDIeB21FRUcX2QVvJFII61tdedz1qYzQp94VkPCRTApmalZubpxikMfltvaY+XlOAYWUbN2xESM5saRrTTKYicqJRY426Xd1vhtsxRhLwpltuliaAcrdURdBaArN2zAN/fbDcw7myb1i/XiJgewyBGr916xYEiwdabdMst3O9s9ZRgEyRFS5nA2CWtu4pN9wgtGvrVv0UcEvGOGTo0GqhJI3gLOPKT21p9JLaYjagKcLcHGFI07b0URtoS81j5YoVCnWbzr7JYIoadbvmRt4tGWPNkdP9r7wXmUSZCGOylrE1b9FCTXBz5uj+VLLtCckUl/7xByTFWdKrV2+ZBNTt+qjop1vNUUAzxpqjvUte+Z9vvCG7du0uzgyq6CGefe55GTNujHJkGccaUqPxd23+zs/LR7bWIvn++++lf//+yq6tUbdr/o1wC8ZI6YUfZ5FKmELoLPdS2a/YY3/7m2RnIh8d/1nTmjdvbjFcylqJ05prVPcxvHfal5mjjyR59XzMQrHE8OmYY3SDpX3EA1244FeZN2+eDB02VK666ipVvK26n0df73IKuAVjpCqSlJiENMCIGk/XowH9zJkzKnXQHQNxW0A1pp3R2mZtyJO1/TnDcampqQoDceWKlQAg8ULOcm/p0bOncrrxeY2UUb4LPwAabMKVV17mCCRTnI8yBIt/+02uuGI4pOpxUg+RELo5BwXcgjFy9f5+zhy54capUh9lIi2tztVBbjKMXTt3qbzW0WPH2B0qVB33au81CMV26sQJuQCpyZo2EQgwHTp2tCg1WnO+sx1DbeDdd96RRWBoJEEecvHnfvc9mGI9GTN+rEy8aqI0aRqtmONGRCwcO3JUSZfmz0Ep8ucff0SJ0+VyBUoREGCWEHa6OQ8F3IIxBgQEQAWpJ3/AgD2AgJ34m6qsJQZZR5ByYZZ1wWMo2ZkD0nKlL8Tn/PkLJVVGzARL7MBA4Dl54qTCybvrrr+4LerJrh075eDBg3Iei5E1jYAe7Tt0KHGopXEpcYAT/3H2zFk5AkCHd959Dwy/AyDt9sqK5ctl4cKFSN2bI3NmfSvt2rdT7+CRI4fl5VdeEaMwG1Vwghv/glIEGzZskJFjxshwSIuOhJU5Malc+tbqYLAszXWXeyhKa/Nhq+FKnJ6WLvUQ4O3l6S11PcAKwfzq1jExSqo5detymymMJCg4SNq0aSvdenRXAeEFUMt37twJB8MuOXE8XvLyEbSN/9j/hQv4BYmBf5ugyABNxpc9PV1SoF61x4SYNv02adO2jVtKi5X1UgwDsOpH//2ftGrdyuVssZs3bZKTJ0+q2irmefO5ObmyZs1q+W3hb7IPaExMJZ18wxTk2XdRiy6nGTFAv5kxA4z1KKTEkTJo0GAV91pZdNX9VB4F3IYxmpOENqBTp06BQaYVbzZJKSYmCT6pmCW38UMUHQKANkae9DFIA/5+/sXqEF9o08dAnSZbNDUySUqbDRtGKPumr6996YfFN1lLfrgyY7RniLiIEulp5tcz5MSpkzJ+3Hjp06+vkirt6U+fU/UUcAtVujSZaMS2xZBNxkcGyaJW4yZMUI4Td/Uql6ZVTfxtWqRq4srVf00yxVOQMMkUk1OS5cabboJm0d5tTS3VT+GquaJbMkZbSXXu7Dnx8/eT6bffrlVgW4ln5/FcjNy90fzCEgQzoT7nwOFy87RpMNtoLEVXGPdazxg5QRcv+k0hdg9BLBltQjSW1yapxhVeVFe7RzLFQ3BSzZwxE/btOrA936qxFF1oEN0OdsxW2pMBsqAQPYOf/O9jBfVEexA90yoew9YO9fEVUsDdFx2Gj7EcxpeffyHeCPy+9bbpwvhPd4xrrXCwXfSAWi8xctxY++V+IOkw3Ie1ehmOMfyKK6RPnz6qtKq2N1bu200l2l1VaTLFndt3yDffzJT6qBFE9ZmwYfodqtx3qKp704zRjMJE/6YNaCEQwL/47DO16l+NutONIhurEqvuLumYkaJqfyrzovvZGKllbN68WWbPnCXRTZsA/u4WRCw01GaZqn2bqqT3Wq9Kl6YqwW3vuPNOuff++1XID7McFv76qzAEiHYj3RyngDsuMGSK69aula++/FKat2gut91xh8LpdMdndfwNcP4etMRYxhj1A9IJpcdfwRRnfPm17Nu3T65EzmssQBEYr6hf+DIIZ+Vms+QjK89w3sPIFImlOGvmNypR4BZIioGwWevmuhTQEmM5Y8c60jfdfLP839NPKiSVDz/4QH764QeVwUBbkm72U8BdbIwGluJXX3wpvWCTngabomaK9r8XznKmlhitGIlu3burWr4EDmApywMHDsr4CeOR+tdWZS9o6dEKIrrhIcRSXIRQr29RimAYnHWTATDLPH3dXJ8CWmK0cgz5wl9z7bXyf888LQF+fvKf//xHvvt2jiQmJppCe6zsRx9mSsd0dTqYYMPmyYyvvlJ501OQF62ZoquP6qX71xLjJVpY9YvpXG1at5YlS5YouKnDhw8p2CiWaqUKpaVHq8jo0uE6ZIo///iTfAc0nWuuu1aumjhRzAElrKOAPsqZKaAZox2j4wHgiNGAjKJzhmCjn33yiezp208mXj1RwlGroyw0Zzsu5canuKb7hViKP3w/V+b98otCzxk3frzOe3bDt1Sr0g4MakxsLMJ67pPpt90u+/bvk3fefFv++P13hepD8ADdLFPAVaVqMsU5334rvyFSgaDI4wE44gezim7uRwEtMVbCmDLHujUwGDlhmAZmhPY0iY7WgeFl0PcSeFsZBzjZ5pycHJn1zTeyGuUMpoApMjOKZQx0c08KaImxksa1MdK+iM7z4EN/lbPItX7/vfcUcK4ODK8kAtdQNwwrygTq9tcI3F6/bh0kxZvkihEjNFOsofGorstqibGSKd0HUP4M42G2zOwZ3yjpkXVAYpENQbXLVdXISiaTS9CBTDElJUVhKR44sF9uApZiXwT+04asm3tTQDNGK8aXqYAM5LU2KJkq1pQbbpB27drLfBjp3//gfRk8aJCMHD1aAehqlBXnB5GgjZihWASYjY+PVwCzrASomaIVE8YNDtGM0YpBPHb0qPJCZmdnW3G0SAgQxG9GWliXrl0QGI7QnsWLVczjgQMHVJnMDigOxZg3LT1aRc5qP4hM8fTp0zLzq6/lHJjjzTffJB07d9ZMsdpHouYuqBmjFbSn5LAcpS5ToVZZ0xo1aiSTJk9Wh1J9vhKF1FsjtGfBL/PkE5RP6Nuvv1w5EcXVwUApPdYWBknQ7vP8BxiYBecvqo8nKuh4oDiZszSFuh0Xp4pWZWVmyS3TblGFrbSk6CwjVD334ZbFsKqHdLZfhar4kkWLZS4wH8PDwmQEKsUx3ZAgue6O11eICovpuYUSl1Ig9984QW585CXUPukgDYN8pGk9L/HzRvXGGuaPZIqHDx2Wb2bOkAsonXsjJEVidWrTh+3vuqufoSVGK0aQ6CmsJGitjZFMzhKzo2Q4cvQoJT3+umC+fIrA8L5AembmBHH73DUwnExx75lc2X0mW3Lyz0tB4UVJzQETSsqRo2CU+895SK+mARIR6C2eNRQnwTrZ+/btV5Kir58vTCHTJCY2RlhuV7faRwEtMVox5ps2bpTXX31NYTJacbgqpfrOv95FFkx4uYcTqupHoPV41PWQwYiFHDhgoASHBLuV9HgeTHHbiUzZA6Z4XsBk6lyU/z48Va574Hmp36wlFWlIZwXi7+MlQ5qHQIL0rHbVmkhJu1BLnPVZ6oeHyVQgKhGX092l+HJfzlq+U0uMVrwA4YAfGwSvMoN8rWlkbt4+3hUeOgiF55lWuGDBrzLzy69k3959MhG2x2gEhvu4AeYjeKIcTMqTVVv3SMGFuhIYGi5hQf7SolVL6RwZJIU+BRJ3NlmSziVIw2atZc3xTBncPFDqB3pJVWjVlPiZ58yoAcOuS6a4ZcsWmQWmGNUkCurzzap8rrG/wkHUB7glBbTE6CTDunH9Bvn5559VOuHgIYMVjFVISIhLq3JnMvJl+ZFMSTiTILNeelCaxraQof17SLOmTST+WJxs3bYV9VG2y6i/PCVRbbuLh6e3NA72koGxgRIAm2NltwwEav8O8I8RI0eqSpA0kWxYv15ltLRq2VpuvvUWCQ0NLWaalX193Z/rUEBLjE4yVr369FZphcR8nDP7W9m/f7+MB2J4q5atVM1rV5Ng6HXeejJHsgvOS3BoQ4lu3kq2r/1DNi3/rQTF+Xyj+nSW4wU+sD9elNPpBXI0uUDaNKwrXpXojaG0+MtPP6uCZwMGDFAo7GtWrYKj5Rvp1KUzAGZvlaBgjbpdYnBq8R81ZOquxRQv59EpITLM56m/PwMPbV356MN/A9rsO0lKShJXQww/nJwnSdksQVtHvDzryITxYyTA3/+yp78CnvmO0WESE+oLx4upQNbehBzJyqvc+jqZmZkye9YsWbtmjQrYPgiwYf7drUcPmT59umaKl41M7d7g8Rxa7SaB8z19REQEYPJ7Q432lAXz56u0woAAfxU47u3t7fSqXgFCXTbGZUlmHss/1JG2Df2kX7sYWTDvF8SCJhcTnGEwjz3+uLLtNYRHOj61QHLhsS6AcdLXq+7/t3cl8FFX1/oQskySIYRskEBYgglh3yTsKItahfr6rK1isZs/pbVgxa1oV23R10Xte7Z1wac+bau41LaCBUTRJztiICwSlpAEEhISQvZMNuj33WSSyZ4AyWzntuNs/+Xe7w5fzr3nnO9IWNDlccTQWqQqzvp/vS/lZeVy6lQWQoVGyegxY4wSu2opNkyJvqhHQC1GF/0p0EGw6MuLUG/mxxIOT+nq518wFk5eXp5RDG8eOsT3DDlp/rkzhselcJEN1qL4GIIbERkgoX37yMxZs5rIdI0ZM1Zihww2cYIkwngQqF/vOqvxaJ5NbFWXR7qNe4trXn8DpFhm4Ni+bas8/qtVUmHDUh/ZTF1J93QGnnrPnkdALcaex7xLd6RHnMIUVqtVPv5os+yFw4LLbAriOlqP/Ad+8OBB4zxwZpYG9xZpLZbUW4sjoiwyKDQAy+Repr/Uq7SnVn7zW9+WSVdONp8TlL7+PpJRBKuxui4rxuLXW8KDYTUi/vNiG/9QvLlmjXywYb3YbJUNl+H2xMb1GyQ9LV2CgS090hrI3QCP179Qi9FNfgKUulr5yMMyePAQaD6+IqxKd+rkKRN+wn/8DER+YtXjkoogZWfuR2acq2ywFgMQrR0f4Q8rsI7YrpwyRWJiBqKvsCSRKjn7qtlNLEh/WI3Dwyxid0gfzbeZgPBLmaLi4mJTrIp7jPZGrJirPnBQrKkBHR+v2S12bPS5DgElRjf6JVD4dun3vyd3fe97kpGZIc/8/vfywcaNJisnOytbsrJOya+feEJOZ2c7ZUldA2sx9YxNqhjAiDY8wiJBsPrs9h4t2bnz5hpSnDIlyRSkbx5EHY9ld1A9M5bCAcPMGO45XkzjH4y/oQxBNnA5j2vw/sxIGgtBiIdWrpR3/v43ue/BB2TwkCFuHRZ1MdjoOe0joOE67ePjkt/OnD2rQTH8L1CAYRGuAykpUlFeIbt37UKq4UuyDCUXWBe7J8N80o21CIcLPNEB8ERfEQ5rsVmO3zxYviw3eg280YGteKktOH54uEVSsktBiL2Ee43D+vmJn6XrP9WiwiIUrHrT7MkyPnEyyPjrt3xNZiFYn9sQ2hSBthDo+q+trSvp5z2KQGRkpCxBcffpEE4tPFco77z9jpSWlpg+/PXPf0YlwxFy41dulCAsGXuicW+x0VrsJXEgN2tAo7Vo7wO9wePGj5fZc2abWEL7547P8REBkoaMmYLyKni2xViNo/r37nJc4/vwQl+Ah/yrN39Nvr74VomPj1fpMEeg9XWbCCgxtgmN639Bbyr3y7h8zs7OwvK5rs9Uifntb34jQ4cNgZU0pUfIIB1xi4V0uNRbiyQ37i2yL9Q3pGOD1isfK+6/zziP2kKYHuq48ACQIpfRSCuk1QgFHr/Azv9cuYyOioiUl197VfhHhEt2fqZNEegMArrH2BmUXPAYkiJFKO6/d4UcPHTQyGQ5drOoqFBW/XKVnMw82e2EwLjFI3mVUM2pC6+Jw96iFfuE3Fs8BE/5FvSTOcr2dq6wUPYiP7nK4TP7d/ZnEqsVwhIXLpwHQXZur5HER8dTNbDh/aZOn2acO1RGohOmqKio27Gw91+f3RuBzv8Jdu9xelzvSYz0rF77peug5jNAvjj8hZTgH351dU0dOSAP+IsvDskf/ucZWfnjh43V1F37jWkFVVJYH7dIT3SC2Vusc7mwBvNO5CO/hGBqC4Qx2F6DV70Aor9UIKJYRmvNWI0gx9KsOqvxSF4F9hqxZxnYdg51QUGBnDhxQmqqquVM3hn57LPPJDcnF+RogeV6HtsL8fItFCyjA0abItAeAkqM7aHjwt8xrnHa9OnmwW5SEIGlE/bv2y8p+5LxSJE8kMPmjz6EGO5EuenmrxoivdxD4t7ikTPlUOOuu/IVkRYjAGH3RF/K/RJgeR7Pr4B243kpQ7B3Gpw7o/wDEc7T8uoMx1n54INw6lxn1NGpnD533jx5+KEfGauReeeTIQrcmsPnUvqo53omAkqMHjKvDEVhLRk+br3tVrNkzM7KMgR57NhRs5Rk6tvlthqPIdawuD5DxWSvhGMvEHGK7bXKyqr2vm74zni2IwLhoS6TKhDvkTOVyKn2E//AllX63oUqemZGpnHqkBTZaBneuXSpfH/pXTJq5GiZjLxobYpAZxBQYuwMSm54DAlw4KBB5tFd3a/CnuLR/Lq9RbhVJD4yUIIZt9jSoGvoAp0xzDrx9evcTy8BTpjj8FBXwUNNTYrjSDccO8C3hdX46SefSD+kTvr6NiXNcVDOCQ+PkA8/3AS1ooVNAsobOqUvFIFmCLT/p73ZwfpWEbAjQP/ukbNVUmLPiUY63/BW4habp9nRGeTYuAVg9xaz+BQ92I6N2TAJ2Gv0x94lG/caWR6huYOZzhbGLdbUUriisfEPRBjq6zANsbbZtRuP0leKQFMElBib4qHvOomADXE0x/Kwt4gg7F4oUDAiMhhZLi1NxagB/aUcyucl8ApzH/D48WOSBOWgC8hEKSsrNUKxdo/1c88+azzHzbtwBfYaQwJgYcJDXYk86lRaqdjbdGzjx0+Q3DO5kn3qlAkRsn9HLzXFI8aOG6tB3XZQ9LlDBFREokOIPOMAEgQ92T6woHohps/euLS1x0N2dv+RlHQot0KyiqqE2XpWkNaU2EAJdEj/s1/fZrNJFnK6Dx8+jBCdZBk1eozEwzu8efNmeIxzTLA3ZdbYP1YNDO0X2mIflOVVSbm5ZdWw+gS52LUQpvBHZUEf8znvFTc8TnZu3wECLjHlThnfybFRd3Hnjh1yNzKBoqOjW1zb3k99VgQcEdDSBo5oeOhrLlWTk5Mlec8eU7smzqEk6GFUKfz/T7fIYmSG9AkJ6RQCJYgr3JhaBAUdsBSKW00eaJXE/pZWM1NefuklWbToy2YpGxkVafb4SMCFiGUkGTJVjyS29p/vwaueKnfceadQsLd5Y6XBD1KL5QxSYS5Aziwe3u9JA4NAxo0kz2tu+mCTZCBkh2o5FksgMn+CZBKcLlxON8/Lbn4Pfa8I2BHo3A64/Wh9dksEihHfePDAAdmwfj2k/TfLU08/JVGw0khQb0CncBtUrRcuvKFTxFgLkt2XVW5KFlCEti+sxWHcW2yjDMHJzEzZl7JPZkOL0TFmMTQ0tAmWZxGDOA3yanaPcpMv8YayZYmQMCtCNgy3NdPO2iS2r78M7OvXUFWQ17z5azebUythqfo7FL1qfj19rwi0h0Djn9v2jtLv3BoB7vGxdvV3vnsHlpap8vmeZLN85qDy8/OMIG5IK1Za80FzCX0SXuHMokosaXuZZfm46GCh8ENbjSEzuxHgbWsny4UW7X6IYAyLG97uPuCQsACJDPbHfSHKC2M1ObscWTHnYUG2bJ5QZbHlqPSTnkKg7V90T/VA79PtCHBvjcvTefPnSXRMjKx/f51xhHApW1Fhk0U33miEcDvqSAn29vYyphCOXzpcBiKmcFBfX2PNtXXu12+5BSION0t75QO4L0iNRqqTs09tNRqlE2MQEoR0Q/quz5XXyEHInNns0eVtnaifKwJdRECJsYuAufPhtKKuv/562bV7t+Tl5soXBw/Jf950U6fqKFfBC7zrZKkUG2VukWAo50yICWoIo2kLF8ZSJqB2dnuq4hak7N26eDGcJokdqmiHBfuhgmCgWOozA4/l2iT9XFVD5k1b/dDPFYGuIKDE2BW0PODYRUiN8/f3k40bP5BPPv5YZkH+qz1rjkOm4yMZ+4o5xfRC98KensDxYcX+YvvB3J2Fi9qIE5Gu11mNxFH9AyUa+4s+vc7LeSyr2bczpVVmed3Ze+pxikB7CCgxtoeOB35HFfCkqVPlnbfeNsrVbTk77EOvBSkehlV2DDnLNfX7irTYYuH0oEPEGY23TYoNRj0Yis2ex9L+gmzPKJOCsrrwIWf0Se/pWQgoMXrWfHZqNP/xla/I2LFjZMasmcJqhG01eqCpnJNyukwQoWNCcwZCF3HsAFTzg0K3Mxv3GUmOIUbZu06abFtGOcRtWSnRmT3Te3sCAhrH6Amz2MUx0MFxNj/fiMU2T9mzX4qWYkZhlezKKJXymlrjgY5E7efZQ+1kZD/Suc8Z2F/cmVGC8KE6NmRVwZlDrRIKUVsnGbTOBUTvflkQUGK8LDB61kVIiumwFHefAinC48tsmX6wzOZcESKhAcg2wXtXaqkQyd17qkQq6h3aJMfpQ6wSHkTVcFfqqfbFXRDQpbS7zFQP9ZOOFirm7MpsSoqzhtHZ4tNC5KEr3WKKHtW0mwtFdOUajsfyOoyBHIHKgmNigiWwfnl/tqxGtpwokdwSm3A7QJsi0FUElBi7ipgHH09hhgOnK2QPLMVKvKalGGbxk5kgxX6wvphNcvzYcSOKezEwpB0/Lr9DLRqm7l1qIymmnzghJSUlhhxHw1M9FuSI7ppWUFIpaz78XLIKKKLLqEdtikDnEVBi7DxWHnvkeVhVrKuyPb3UOFoYL83skoggP5kVB+8vnkmSr/3fq/K/L642ZHQxYOTl5ZuURBsycS61FUDTcemdd0l6WlqDBToK+drM2w7y7yUFOZny2uMPyNo9J+BRr5YKqAGp8XipqHvP+Zor7T1z3WKkJAoWs88uqkZ6XakUoYRATXWV1FTapK/FR8b1s0qgD93RvmKzVRo1nAqQGnOvmZdcBhkx5iMz/pD7jrTiaMGxFg2dOudhqZVXlNep+kDRpxrXdszf4zKYkmNU4OH59JDzwdcmKwcaigEBFqmsgswYdBt5TV6bYhD5IFmWiy1G1gzPt6uTU1yClQ+O78vDfaEuDo3GbcfzJbskRCZEQ74Me6W+OEC3Hlv8HPQDBwRUdswBDG95CS40QdtnEdqyG3uJB3LKjTDDhfM1kvLxWtn59guy7R9/lk82bTBaigkJCRCg2IDYx7eEy2HKh81FeuGPHnzIEOOQIUMMaeVARmzZ9++WcRPGG9WcY5D8+uMzz8gf8NiE8ytsFZAfS5Vbbr1FrMihPgU5svfXrZM/oijWhg3roZtYjsJe/Q35US7s0V/83Fh5a954wxT12rl9uyQkjjDk+bOf/lSOHj0qXxw4ZMQvKDtm97CXFeTIX//0lJxMT5PTRw9KSEQ0tNEiJLO4FkK2F6QP0maMlBkIWAnSW371XRunWoxdw8utj6a3mdttjPWjEnZmIUoGMGgb9MClckl+tmx5+0VZ8cPlJiNmPwpq/ffvn5aJEyfJvHlzZW/y58hnzpd77llucq+zUFOmpLioYSlLq+7kyUypgnVJcdif/+xnEhYeLo8+9phkZ2XL6tUvQH6szGBIJ8xDD9wP4Yg4+eF9K+Qcqga++vIrshvpir947FFYkjYjLEGy/DYq+9301ZvksUcfk9f/8hfUpb5ffrBsuVEMWr7iXklKmmLI0j451Hf87h3flSOph2XJ8kckYMAQuYCSBwwEP5BTAWWeKhke4S/xEMC1wGHji1QeDe2xo6fPRECJ0YN/B7QMuX/IZzogThXXyAlksJwprTapfZSC8MVeIjNYBiLFzr9Xb3kr2NKw9L167tUyesxoU56VS1VrsNXoKvaHKIXdOmsLPuo/nkE+9qonHoeI7PC6pTYUuJ9+8ilzCtMR6YS5/ZvflAG4HpsFy+gHH3hAToNE7Z7rZcuXy+QpV5pc6/nz58sByKedh3c7KirK6Dj2xzN1JB1DiJiXHQ5Cps7jDVPipDY40ljFFJ3gHweGIO2Hkyk1zyaDQvwNSYZhH5U4kCD5RwL/1+bFCCgxeuDkU/CBYTeFCOzLLUaNZRBCPsiwUYSmzkL09YGDBUHbY+DR7Q+CkNoxchNEJV55+SU4WV6UCVgSX40SpNded12XUTqJZfKA6AFNSGs01Lv9sB/JdgJOk2LsR7K8qS+IjI1kyJCe01iSh4bWidUyhdEuQNGnjxXfdy2zhfw2BPWoY0D8R1Hm9QhCkUpRM4b4UCUoraDSPIKhBh4JgYqoEF/pj2c6cFjt0FlpjwYQ/Y/TEFBidBr03XNjhtxsSSuBswFOFPpNHButQ5hC3F+L6gOVGjgqovHM92zVsOi+c8cdsvi2b2BJu0s+3LRJ/mvV41J0rkgWL7nN8UrmNZ0gtQ43qcQSmg4VtkAo+dCx4igjRifKBZAfm5+fv8QOijUlB8Khrs3GY+mxjseeZiYEbtns1zNvLuE/FPoeNSBIWD/mGJbSxyF0W1wByxlWM7cXWLe6DP1LR+1qtiCcMBrqQaxS6NdKHetL6Iqe6gYIaLiOG0xSV7rIf8SlMA2rUdqUmolcFvpCogtC29IPaXIjQQ7XJoTI3OF9oKXo30CKvMeez/bITx75sdkHnIdl6y9XrZKr5s6V9PQTxivMWjFVqAnNinskLKs12Fh39E7zs0NY5rJ+DNvI0aOkEPuGR+BsKUYhLB6zdcsWkGXd9yxORf3FAFiQQ4YONcttXmPNmjXm+I4IsVc9mbPAlmOlQXNz/MdexoBajzXVjRqPrDbIsJ4bEvvKnOGhMgzit1aQIMrVwDqsC1PiNcoR3lPH4XVEb7+uPnsHAmoxeuA8hyIkpQIWEMsNhGNZGGnFEtHaW7iPVs8nrY46AUWqTmdny29+/WuZNn26cYiwJMKyZctMyQE6NTZ/9JG8tPpF4zCZMXOWrH3vPfHHUrg36kRv3vShISmSEj3Z1y9cKE/+9rcy75oFUgqC+hj7ijU11WY/cCrKGEyYMEEe/9UqLNfnSjD2MP+17n0ZkZiIglj9TNiPPXTH3tneCNfhspr7iQwXojr4a6++ii2CGpmB/jJ0yN5Y44We7+ee/ZMswT7meGwL2JfkPIZWciwEMfjg/mt+aa0ptsX910JkzqB6gkl/tBOs/br67B0IaLiOB85zX4Sj0OM6DsWiaBFFWn2xZ9axdmIgyGnm7FniA6dFyt59WPbWyjduX2JCc0hSw4bFgfhQSxokxwJTM2bOBNH6mLCZAf0HyF1L7wKBBsmVKI9K0qK8WUREhKTCaoyNHSTfWLJEIsLCzfeUO5s9Zw6uOUzSkE3DeMRFN34ZJHa7hMCZwhrQXKYnTZuKolYWM0vMvAnpGypj4BAKtlplEJbidOBEREbIUFzH0SHEUg10wJwrPGfuTcFcOmNaayRJhvAMwLbC8HCLJEBWLSbETyKQc63L6NYQ8/zPVETC8+dYR6gIKAJdRECX0l0EzB0Op9PDHu7iDv11xT7SurQv212xf9qn7kVAibF78XXK1blXl5Ob0+ABdkon3PymI0eOxDJ+WodlH9x8mNr9NhBQYmwDGHf+mDnERnWmPjTGncfirL5XwrvuGGrkrH7ofZ2DgO4xOgd3vasioAi4MAIax+jCk6NdUwQUAecgoMToHNy9+q50DKlzyKt/Ai4/eN1jdPkp8rwOZp3KQk7OBYmJiWkSe+h5I9URuSsCajG668y5cb//8e678nc8KsovXcnbjWHQrrswAkqMLjw52jVFQBFwDgJKjM7B3evv2pFIhNcDpAA4FQHdY3Qq/F56c6icQQrWSwevw3YHBNRidIdZ8rA+qrXoYRPqgcNRYvTASXWHIdWVIlCtQ3eYK2/soxKjN866k8esFqOTJ0Bv3yECSowdQqQHdAcCjsWruuP6ek1F4FIQUGK8FPT03ItCQC3Gi4JNT+pBBJQYexBsvVUjAmoxNmKhr1wPASVG15sTj+9RfSFBjx+nDtB9EVBidN+5c+ueq8Xo1tPn8Z1XYvT4KXa9AXKPUQN1XG9etEeNCCgxNmKhr3oIAZPzAmbsrNVIIq2tre2h3ultFAHUF1cQFIGuIGAnqfY8y8YibGcjsb1zW+tLSXGxHE09IlUoN6BNEegJBJQYewJlD7mHDXWdc06flv3790t2VrapK0Oyak50ZWVlkp+X166V11lrkdDt2bNHXnj+OXM/D4FSh+HiCKiIhItPkKt0j4WhXv/r6/Lqyy9L/tl88fHxlWnTkuS666+X6TNmSF8UuPf39zdkuHHDBvN8w8KFEhwc3GIIzYm0xQEOH/C+vP6MGTPFP8Df4Rt9qQh0HwJKjN2HrUddmVbiPyEuO/+aBRIYGCgp+1JkX0qKbNu6TcLCwuXaL10n4UlRegAABc9JREFUs2bPksyMTHn3nXfkF798TAICAtrEoLMWY+rhw/LZ7t0yYeJEGTlqlCHfNi+qXygClwkBJcbLBKSnX2bXjh1y97JlMufqqxoILz09Xda9t1beX7cOZPi2vPXmm+KLQvW3Lr5NhsXFtVu2oLNWI69z370rDDH2xrW1KQI9gYASY0+g7AH3mDZ9hiSOTGwgRQ5p6NCh8oPly+TOpXfJAViU3H+MHzHCfO7n59dk1NXV1YYoaSn6+PiII8k5ftfkJLzJAPnGRMdINOrDOJ7T/Dh9rwhcTgSUGC8nmh58rfETxrc5Ou4tTpo8uc3vSXw7t++QESBW7jlGREWKJcAidObUIAxnP5bkU5KSxGKxtLjG1q1bZeLkSWb53uJL/UAR6CYElBi7CVi9bCMCtB43btwozz37Jxk7brwkJo6Q8ooKeerJJyX582SZOGkiPh/XghhZYnXn9u1y2+23i38zC7Tx6vpKEbj8CCgxXn5M9YqtILAATpsH1/9LdsBydGxcVt9734pWvdfnzp2T06dzpDeW32Xl5eIHy5THa1MEuhsB/ZV1N8J6fYPA1GnTJKp/fxBb01ovgwYOknHjx0vzPUmeZINVmZiYKP5YYvfp00dJUX9LPYaAWow9BrV334ihO3PmzDHOlPKy8gYw5i9YIFarteG944uBgwbJ755+yvEjfa0I9AgCajH2CMx6EyKw4JprmiyZ6WVmXGRrThdFTBFwJgJKjM5E38vuzSVzbOxghN3U/ewSE0fK8PgrWl1Gexk0OlwXQ0CJ0cUmxJO7Qwtx3vx5EhgUZIbJbJkgZNFoUwRcDQElRlebEQ/vz3wsp7mnyOXzVVdfLRYlRg+fcfccnjpf3HPe3LbX8fHxwocPrMfo6AGazeK2M+nZHVdi9Oz5dcnRXXPttbAa+0hQK8o7Ltlh7ZTXIdALyfyqMu910+7cAefm5MA7bZVga3CnVbyd22O9u7choMTobTOu41UEFIEOEVDnS4cQ6QGKgCLgbQgoMXrbjOt4FQFFoEMElBg7hEgPUAQUAW9DQInR22Zcx6sIKAIdIqDE2CFEeoAioAh4GwJKjN42404aL6PCaqDk7RgdxgqAFKPVpgi4GgJKjK42Ix7YHxLg6ezTsuXTT6UcgrP2thvV//i5kqMdEX12FQSUGF1lJjy4H0eOHJHVz78g995zjxw+dNjUnOZwVz/3vLyCOtWlpaUePHodmjsioMTojrPmZn0ehXrQD//kEQmPjJSMzHShBcm24v77ZQSqCmr1PzebUC/orhKjF0yyKwyRlQQjwiOksLDQEGMtqgMeOHBAZsycoRUAXWGCtA9NEFBibAKHvulOBPr16wdiLJLzIMXUw4elf1SUhIWHay2X7gRdr31RCCgxXhRsetLFINAvLEyKYTEWoPpfSsp+SZratJZ0ZWWlVFVVNfFcX8x99BxF4FIRUGK8VAT1/E4jEAZi5FJ67XvvyYL5C6Cu07QI1sebN8uePXukGmE92hQBZyKgxOhM9L3s3qGhoZKRkSEzZsyU0LDQJpJjDNlhidWkpCThfqQ2RcCZCCgxOhN9L7o3nS1W1IZe+fBKGT1mtPj6NtVIzsvLk3vu/oGcyT3jRajoUF0Vgaa/TlftpfbLLRFglovNZjPhOCfS0mRwbKxMmDSpVYuwtqZW4hMSUAPG4pZj1U57FgJKjJ41ny41msJzhfLWW29K3LA4uYD/zZo9WwICAlrt47atW2TchPFtft/qSfqhItBNCOhSupuA1cuK5OXnyVGE5dBqnD1nTrvxisnJeyUiIkJ8USRLmyLgbAS0tIGzZ0DvbxBYt3atjBo9WgYPHqyZMPqbcDoCSoxOnwLtgCKgCLgaArqUdrUZ0f4oAoqA0xFQYnT6FGgHFAFFwNUQUGJ0tRnR/igCioDTEVBidPoUaAcUAUXA1RBQYnS1GdH+KAKKgNMRUGJ0+hRoBxQBRcDVEFBidLUZ0f4oAoqA0xH4N/3wIPDTTxaNAAAAAElFTkSuQmCC\n", "text/plain": [ "" ] }, "execution_count": 5, "metadata": { "image/png": { "width": 400 } }, "output_type": "execute_result" } ], "source": [ "Image('images/fig18_4.png', width=400)" ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "with pm.Model() as model:\n", " \n", " zbeta0 = pm.Normal('zbeta0', mu=0, sd=2)\n", " #zbeta1 = pm.Normal('zbetaj1', mu=0, sd=2)\n", " #zbeta2 = pm.Normal('zbetaj2', mu=0, sd=2)\n", " #zmu = zbeta0 + (zbeta1 * X[:,0]) + (zbeta2 * X[:,1])\n", " zbetaj = pm.Normal('zbetaj', mu=0, sd=2, shape=(2))\n", " zmu = zbeta0 + pm.math.dot(zbetaj, zX.T)\n", " \n", " nu = pm.Exponential('nu', 1/29.)\n", " zsigma = pm.Uniform('zsigma', 10**-5, 10)\n", " \n", " likelihood = pm.StudentT('likelihood', nu=nu, mu=zmu, lam=1/zsigma**2, observed=zy)" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ "Auto-assigning NUTS sampler...\n", "Initializing NUTS using jitter+adapt_diag...\n", "/home/xian/anaconda3/envs/pymcv4beta/lib/python3.10/site-packages/pymc/model.py:984: FutureWarning: Model.initial_point has been deprecated. Use Model.recompute_initial_point(seed=None).\n", " warnings.warn(\n", "Multiprocess sampling (2 chains in 2 jobs)\n", "NUTS: [zbeta0, zbetaj, nu, zsigma]\n", "/home/xian/anaconda3/envs/pymcv4beta/lib/python3.10/site-packages/pymc/model.py:984: FutureWarning: Model.initial_point has been deprecated. Use Model.recompute_initial_point(seed=None).\n", " warnings.warn(\n" ] }, { "data": { "text/html": [ "\n", "