Experiments.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Try transfer learning for Fudenberg (HEAD part only?)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# is batchNormalization used properly?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting natsort\n",
      "  Using cached https://files.pythonhosted.org/packages/0f/65/81883897f4aaa1e53deaa65137318cfe80b36ce013c2e86f8fd0843cfa02/natsort-7.0.1-py3-none-any.whl\n",
      "Installing collected packages: natsort\n",
      "Successfully installed natsort-7.0.1\n",
      "Note: you may need to restart the kernel to use updated packages.\n"
     ]
    }
   ],
   "source": [
    "pip install natsort"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.14.0\n"
     ]
    }
   ],
   "source": [
    "# Try their model straightaway\n",
    "import os\n",
    "import json\n",
    "os.environ[\"CUDA_VISIBLE_DEVICES\"] = '-1' ### run on CPU\n",
    "import tensorflow as tf\n",
    "print(tf.__version__)\n",
    "if tf.__version__[0] == '1':\n",
    "    tf.compat.v1.enable_eager_execution()\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "from cooltools.lib.numutils import set_diag\n",
    "from basenji import dataset, dna_io, seqnn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas\n",
    "\n",
    "pandas.set_option('display.max_columns', 500)\n",
    "pandas.set_option('display.max_rows', 500)\n",
    "\n",
    "import h5py\n",
    "import random\n",
    "\n",
    "import math\n",
    "import tensorflow as tf\n",
    "import tensorflow.keras as keras\n",
    "from tensorflow.python.framework import ops\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras.models import Sequential\n",
    "\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "\n",
    "import cooler\n",
    "import cooltools as ct\n",
    "\n",
    "from Bio import SeqIO\n",
    "\n",
    "import pickle\n",
    "\n",
    "import scipy\n",
    "import os\n",
    "import re"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot_hic(matrix, use_log_scale = False):\n",
    "    fig = plt.figure(figsize=(10, 10))\n",
    "    ax = fig.add_subplot(111)\n",
    "    \n",
    "    if use_log_scale:\n",
    "        im = ax.matshow(np.log10(matrix), cmap='YlOrRd')\n",
    "        fig.colorbar(im)\n",
    "    else:\n",
    "        im = ax.matshow(matrix, cmap='YlOrRd')\n",
    "        fig.colorbar(im)\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def one_hot_dna(sequences = {}):\n",
    "    for k,v in sequences.items():\n",
    "        seq_array = np.array(list(v))\n",
    "\n",
    "        label_encoder = LabelEncoder()\n",
    "        integer_encoded_seq = label_encoder.fit_transform(seq_array)\n",
    "\n",
    "        integer_encoded_seq = integer_encoded_seq.reshape(len(integer_encoded_seq), 1)\n",
    "\n",
    "        onehot_encoder = OneHotEncoder(sparse = False)\n",
    "        result = onehot_encoder.fit_transform(integer_encoded_seq)\n",
    "        \n",
    "        # if Ns are present in the DNA sequence, result will have 5 columns. We delete 4th column which has Ns\n",
    "        # N row in the resulting training set will have all 0s\n",
    "        if result.shape[1] == 5:\n",
    "            result = np.delete(result, 3, 1)\n",
    "        \n",
    "        sequences[k] = result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Colab specific code\n",
    "# filepath = \"drive/My Drive/Colab Notebooks/S2-Wang2017-async.dm3.mapq_30.100.mcool\"\n",
    "\n",
    "filepath = \"S2-Wang2017-async.dm3.mapq_30.100.mcool\"\n",
    "\n",
    "resolution = \"::/resolutions/2000\" # 1 KB resolution\n",
    "c = cooler.Cooler(filepath + resolution)\n",
    "\n",
    "chroms = c.chromnames\n",
    "# don't use these small chromosomes\n",
    "chroms.remove(\"chrM\")\n",
    "chroms.remove(\"chr4\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "arr = c.matrix(balance=True).fetch('chr2R')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "arr_raw = c.matrix(balance=False).fetch('chr2R')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "sequences = {}\n",
    "for chrom in chroms:\n",
    "    fasta_sequence = list(SeqIO.parse(open(f\"./chromFa/{chrom}.fa\"),'fasta'))[0]\n",
    "    sequences[chrom] = str(fasta_sequence.seq)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "chr2R = sequences['chr2R']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_3\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "sequence (InputLayer)           [(None, 1048576, 4)] 0                                            \n",
      "__________________________________________________________________________________________________\n",
      "stochastic_reverse_complement_1 ((None, 1048576, 4), 0           sequence[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "stochastic_shift_1 (StochasticS (None, 1048576, 4)   0           stochastic_reverse_complement_1[0\n",
      "__________________________________________________________________________________________________\n",
      "re_lu_42 (ReLU)                 (None, 1048576, 4)   0           stochastic_shift_1[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_28 (Conv1D)              (None, 1048576, 96)  4224        re_lu_42[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_41 (BatchNo (None, 1048576, 96)  384         conv1d_28[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_11 (MaxPooling1D) (None, 524288, 96)   0           batch_normalization_41[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_43 (ReLU)                 (None, 524288, 96)   0           max_pooling1d_11[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_29 (Conv1D)              (None, 524288, 96)   46080       re_lu_43[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_42 (BatchNo (None, 524288, 96)   384         conv1d_29[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_12 (MaxPooling1D) (None, 262144, 96)   0           batch_normalization_42[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_44 (ReLU)                 (None, 262144, 96)   0           max_pooling1d_12[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_30 (Conv1D)              (None, 262144, 96)   46080       re_lu_44[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_43 (BatchNo (None, 262144, 96)   384         conv1d_30[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_13 (MaxPooling1D) (None, 131072, 96)   0           batch_normalization_43[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_45 (ReLU)                 (None, 131072, 96)   0           max_pooling1d_13[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_31 (Conv1D)              (None, 131072, 96)   46080       re_lu_45[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_44 (BatchNo (None, 131072, 96)   384         conv1d_31[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_14 (MaxPooling1D) (None, 65536, 96)    0           batch_normalization_44[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_46 (ReLU)                 (None, 65536, 96)    0           max_pooling1d_14[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_32 (Conv1D)              (None, 65536, 96)    46080       re_lu_46[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_45 (BatchNo (None, 65536, 96)    384         conv1d_32[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_15 (MaxPooling1D) (None, 32768, 96)    0           batch_normalization_45[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_47 (ReLU)                 (None, 32768, 96)    0           max_pooling1d_15[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_33 (Conv1D)              (None, 32768, 96)    46080       re_lu_47[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_46 (BatchNo (None, 32768, 96)    384         conv1d_33[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_16 (MaxPooling1D) (None, 16384, 96)    0           batch_normalization_46[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_48 (ReLU)                 (None, 16384, 96)    0           max_pooling1d_16[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_34 (Conv1D)              (None, 16384, 96)    46080       re_lu_48[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_47 (BatchNo (None, 16384, 96)    384         conv1d_34[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_17 (MaxPooling1D) (None, 8192, 96)     0           batch_normalization_47[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_49 (ReLU)                 (None, 8192, 96)     0           max_pooling1d_17[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_35 (Conv1D)              (None, 8192, 96)     46080       re_lu_49[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_48 (BatchNo (None, 8192, 96)     384         conv1d_35[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_18 (MaxPooling1D) (None, 4096, 96)     0           batch_normalization_48[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_50 (ReLU)                 (None, 4096, 96)     0           max_pooling1d_18[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_36 (Conv1D)              (None, 4096, 96)     46080       re_lu_50[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_49 (BatchNo (None, 4096, 96)     384         conv1d_36[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_19 (MaxPooling1D) (None, 2048, 96)     0           batch_normalization_49[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_51 (ReLU)                 (None, 2048, 96)     0           max_pooling1d_19[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_37 (Conv1D)              (None, 2048, 96)     46080       re_lu_51[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_50 (BatchNo (None, 2048, 96)     384         conv1d_37[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_20 (MaxPooling1D) (None, 1024, 96)     0           batch_normalization_50[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_52 (ReLU)                 (None, 1024, 96)     0           max_pooling1d_20[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_38 (Conv1D)              (None, 1024, 96)     46080       re_lu_52[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_51 (BatchNo (None, 1024, 96)     384         conv1d_38[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling1d_21 (MaxPooling1D) (None, 512, 96)      0           batch_normalization_51[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_53 (ReLU)                 (None, 512, 96)      0           max_pooling1d_21[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_39 (Conv1D)              (None, 512, 48)      13824       re_lu_53[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_52 (BatchNo (None, 512, 48)      192         conv1d_39[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_54 (ReLU)                 (None, 512, 48)      0           batch_normalization_52[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_40 (Conv1D)              (None, 512, 96)      4608        re_lu_54[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_53 (BatchNo (None, 512, 96)      384         conv1d_40[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_14 (Dropout)            (None, 512, 96)      0           batch_normalization_53[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_14 (Add)                    (None, 512, 96)      0           max_pooling1d_21[0][0]           \n",
      "                                                                 dropout_14[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_55 (ReLU)                 (None, 512, 96)      0           add_14[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_41 (Conv1D)              (None, None, 48)     13824       re_lu_55[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_54 (BatchNo (None, None, 48)     192         conv1d_41[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_56 (ReLU)                 (None, None, 48)     0           batch_normalization_54[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_42 (Conv1D)              (None, None, 96)     4608        re_lu_56[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_55 (BatchNo (None, None, 96)     384         conv1d_42[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_15 (Dropout)            (None, None, 96)     0           batch_normalization_55[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_15 (Add)                    (None, 512, 96)      0           add_14[0][0]                     \n",
      "                                                                 dropout_15[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_57 (ReLU)                 (None, 512, 96)      0           add_15[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_43 (Conv1D)              (None, None, 48)     13824       re_lu_57[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_56 (BatchNo (None, None, 48)     192         conv1d_43[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_58 (ReLU)                 (None, None, 48)     0           batch_normalization_56[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_44 (Conv1D)              (None, None, 96)     4608        re_lu_58[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_57 (BatchNo (None, None, 96)     384         conv1d_44[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_16 (Dropout)            (None, None, 96)     0           batch_normalization_57[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_16 (Add)                    (None, 512, 96)      0           add_15[0][0]                     \n",
      "                                                                 dropout_16[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_59 (ReLU)                 (None, 512, 96)      0           add_16[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_45 (Conv1D)              (None, None, 48)     13824       re_lu_59[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_58 (BatchNo (None, None, 48)     192         conv1d_45[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_60 (ReLU)                 (None, None, 48)     0           batch_normalization_58[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_46 (Conv1D)              (None, None, 96)     4608        re_lu_60[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_59 (BatchNo (None, None, 96)     384         conv1d_46[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_17 (Dropout)            (None, None, 96)     0           batch_normalization_59[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_17 (Add)                    (None, 512, 96)      0           add_16[0][0]                     \n",
      "                                                                 dropout_17[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_61 (ReLU)                 (None, 512, 96)      0           add_17[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_47 (Conv1D)              (None, None, 48)     13824       re_lu_61[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_60 (BatchNo (None, None, 48)     192         conv1d_47[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_62 (ReLU)                 (None, None, 48)     0           batch_normalization_60[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_48 (Conv1D)              (None, None, 96)     4608        re_lu_62[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_61 (BatchNo (None, None, 96)     384         conv1d_48[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_18 (Dropout)            (None, None, 96)     0           batch_normalization_61[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_18 (Add)                    (None, 512, 96)      0           add_17[0][0]                     \n",
      "                                                                 dropout_18[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_63 (ReLU)                 (None, 512, 96)      0           add_18[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_49 (Conv1D)              (None, None, 48)     13824       re_lu_63[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_62 (BatchNo (None, None, 48)     192         conv1d_49[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_64 (ReLU)                 (None, None, 48)     0           batch_normalization_62[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_50 (Conv1D)              (None, None, 96)     4608        re_lu_64[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_63 (BatchNo (None, None, 96)     384         conv1d_50[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_19 (Dropout)            (None, None, 96)     0           batch_normalization_63[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_19 (Add)                    (None, 512, 96)      0           add_18[0][0]                     \n",
      "                                                                 dropout_19[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_65 (ReLU)                 (None, 512, 96)      0           add_19[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_51 (Conv1D)              (None, None, 48)     13824       re_lu_65[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_64 (BatchNo (None, None, 48)     192         conv1d_51[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_66 (ReLU)                 (None, None, 48)     0           batch_normalization_64[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_52 (Conv1D)              (None, None, 96)     4608        re_lu_66[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_65 (BatchNo (None, None, 96)     384         conv1d_52[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_20 (Dropout)            (None, None, 96)     0           batch_normalization_65[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_20 (Add)                    (None, 512, 96)      0           add_19[0][0]                     \n",
      "                                                                 dropout_20[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_67 (ReLU)                 (None, 512, 96)      0           add_20[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_53 (Conv1D)              (None, None, 48)     13824       re_lu_67[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_66 (BatchNo (None, None, 48)     192         conv1d_53[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_68 (ReLU)                 (None, None, 48)     0           batch_normalization_66[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_54 (Conv1D)              (None, None, 96)     4608        re_lu_68[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_67 (BatchNo (None, None, 96)     384         conv1d_54[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_21 (Dropout)            (None, None, 96)     0           batch_normalization_67[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_21 (Add)                    (None, 512, 96)      0           add_20[0][0]                     \n",
      "                                                                 dropout_21[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_69 (ReLU)                 (None, 512, 96)      0           add_21[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv1d_55 (Conv1D)              (None, 512, 64)      30720       re_lu_69[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_68 (BatchNo (None, 512, 64)      256         conv1d_55[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_70 (ReLU)                 (None, 512, 64)      0           batch_normalization_68[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "average_to2d_1 (AverageTo2D)    (None, 512, 512, 64) 0           re_lu_70[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "concat_dist2d_1 (ConcatDist2D)  (None, 512, 512, 65) 0           average_to2d_1[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_71 (ReLU)                 (None, 512, 512, 65) 0           concat_dist2d_1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_13 (Conv2D)              (None, 512, 512, 48) 28080       re_lu_71[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_69 (BatchNo (None, 512, 512, 48) 192         conv2d_13[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "symmetrize2d_7 (Symmetrize2D)   (None, 512, 512, 48) 0           batch_normalization_69[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_72 (ReLU)                 (None, 512, 512, 48) 0           symmetrize2d_7[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_14 (Conv2D)              (None, 512, 512, 24) 10368       re_lu_72[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_70 (BatchNo (None, 512, 512, 24) 96          conv2d_14[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_73 (ReLU)                 (None, 512, 512, 24) 0           batch_normalization_70[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_15 (Conv2D)              (None, 512, 512, 48) 1152        re_lu_73[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_71 (BatchNo (None, 512, 512, 48) 192         conv2d_15[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_22 (Dropout)            (None, 512, 512, 48) 0           batch_normalization_71[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_22 (Add)                    (None, 512, 512, 48) 0           symmetrize2d_7[0][0]             \n",
      "                                                                 dropout_22[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "symmetrize2d_8 (Symmetrize2D)   (None, 512, 512, 48) 0           add_22[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_74 (ReLU)                 (None, 512, 512, 48) 0           symmetrize2d_8[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_16 (Conv2D)              (None, None, None, 2 10368       re_lu_74[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_72 (BatchNo (None, None, None, 2 96          conv2d_16[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_75 (ReLU)                 (None, None, None, 2 0           batch_normalization_72[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_17 (Conv2D)              (None, None, None, 4 1152        re_lu_75[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_73 (BatchNo (None, None, None, 4 192         conv2d_17[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_23 (Dropout)            (None, None, None, 4 0           batch_normalization_73[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_23 (Add)                    (None, 512, 512, 48) 0           symmetrize2d_8[0][0]             \n",
      "                                                                 dropout_23[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "symmetrize2d_9 (Symmetrize2D)   (None, 512, 512, 48) 0           add_23[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_76 (ReLU)                 (None, 512, 512, 48) 0           symmetrize2d_9[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_18 (Conv2D)              (None, None, None, 2 10368       re_lu_76[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_74 (BatchNo (None, None, None, 2 96          conv2d_18[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_77 (ReLU)                 (None, None, None, 2 0           batch_normalization_74[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_19 (Conv2D)              (None, None, None, 4 1152        re_lu_77[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_75 (BatchNo (None, None, None, 4 192         conv2d_19[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_24 (Dropout)            (None, None, None, 4 0           batch_normalization_75[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_24 (Add)                    (None, 512, 512, 48) 0           symmetrize2d_9[0][0]             \n",
      "                                                                 dropout_24[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "symmetrize2d_10 (Symmetrize2D)  (None, 512, 512, 48) 0           add_24[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_78 (ReLU)                 (None, 512, 512, 48) 0           symmetrize2d_10[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_20 (Conv2D)              (None, None, None, 2 10368       re_lu_78[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_76 (BatchNo (None, None, None, 2 96          conv2d_20[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_79 (ReLU)                 (None, None, None, 2 0           batch_normalization_76[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_21 (Conv2D)              (None, None, None, 4 1152        re_lu_79[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_77 (BatchNo (None, None, None, 4 192         conv2d_21[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_25 (Dropout)            (None, None, None, 4 0           batch_normalization_77[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_25 (Add)                    (None, 512, 512, 48) 0           symmetrize2d_10[0][0]            \n",
      "                                                                 dropout_25[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "symmetrize2d_11 (Symmetrize2D)  (None, 512, 512, 48) 0           add_25[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_80 (ReLU)                 (None, 512, 512, 48) 0           symmetrize2d_11[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_22 (Conv2D)              (None, None, None, 2 10368       re_lu_80[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_78 (BatchNo (None, None, None, 2 96          conv2d_22[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_81 (ReLU)                 (None, None, None, 2 0           batch_normalization_78[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_23 (Conv2D)              (None, None, None, 4 1152        re_lu_81[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_79 (BatchNo (None, None, None, 4 192         conv2d_23[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_26 (Dropout)            (None, None, None, 4 0           batch_normalization_79[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_26 (Add)                    (None, 512, 512, 48) 0           symmetrize2d_11[0][0]            \n",
      "                                                                 dropout_26[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "symmetrize2d_12 (Symmetrize2D)  (None, 512, 512, 48) 0           add_26[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_82 (ReLU)                 (None, 512, 512, 48) 0           symmetrize2d_12[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_24 (Conv2D)              (None, None, None, 2 10368       re_lu_82[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_80 (BatchNo (None, None, None, 2 96          conv2d_24[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "re_lu_83 (ReLU)                 (None, None, None, 2 0           batch_normalization_80[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_25 (Conv2D)              (None, None, None, 4 1152        re_lu_83[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_81 (BatchNo (None, None, None, 4 192         conv2d_25[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_27 (Dropout)            (None, None, None, 4 0           batch_normalization_81[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "add_27 (Add)                    (None, 512, 512, 48) 0           symmetrize2d_12[0][0]            \n",
      "                                                                 dropout_27[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "symmetrize2d_13 (Symmetrize2D)  (None, 512, 512, 48) 0           add_27[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "cropping2d_1 (Cropping2D)       (None, 448, 448, 48) 0           symmetrize2d_13[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "upper_tri_1 (UpperTri)          (None, 99681, 48)    0           cropping2d_1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 99681, 5)     245         upper_tri_1[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "switch_reverse_triu_1 (SwitchRe (None, 99681, 5)     0           dense_1[0][0]                    \n",
      "                                                                 stochastic_reverse_complement_1[0\n",
      "==================================================================================================\n",
      "Total params: 751,653\n",
      "Trainable params: 746,149\n",
      "Non-trainable params: 5,504\n",
      "__________________________________________________________________________________________________\n",
      "None\n",
      "model_strides [2048]\n",
      "target_lengths [99681]\n",
      "target_crops [-49585]\n"
     ]
    }
   ],
   "source": [
    "model_dir = \"./basenji/\"\n",
    "params_file = model_dir+'params.json'\n",
    "model_file  = model_dir+'model_best.h5'\n",
    "with open(params_file) as params_open:\n",
    "    params = json.load(params_open)\n",
    "    params_model = params['model']\n",
    "    params_train = params['train']\n",
    "\n",
    "seq_length = params_model['seq_length']\n",
    "target_length = params_model['target_length']\n",
    "target_crop = params_model['target_crop']\n",
    "seqnn_model = seqnn.SeqNN(params_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "successfully loaded\n"
     ]
    }
   ],
   "source": [
    "seqnn_model.restore(model_file)\n",
    "print('successfully loaded')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(99681, 448)"
      ]
     },
     "execution_count": 151,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target_crop = params_model['target_crop']\n",
    "hic_diags = 2\n",
    "target_length_cropped = target_length - 2*target_crop\n",
    "tlen = (target_length_cropped - hic_diags) * (target_length_cropped - hic_diags+1) // 2\n",
    "\n",
    "def from_upper_triu(vector_repr, matrix_len, num_diags):\n",
    "    z = np.zeros((matrix_len,matrix_len))\n",
    "    triu_tup = np.triu_indices(matrix_len,num_diags)\n",
    "    z[triu_tup] = vector_repr\n",
    "    for i in range(-num_diags+1,num_diags):\n",
    "        set_diag(z, np.nan, i)\n",
    "    return z + z.T\n",
    "\n",
    "tlen, target_length_cropped"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {},
   "outputs": [],
   "source": [
    "FACTOR = 15"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 181,
   "metadata": {},
   "outputs": [],
   "source": [
    "seq = chr2R[seq_length*FACTOR:seq_length*(FACTOR+1)]\n",
    "if len(seq) != seq_length: raise ValueError('len(seq) != seq_length')\n",
    "\n",
    "seq_1hot = dna_io.dna_1hot(seq, n_random=False)\n",
    "\n",
    "test_pred_from_seq = seqnn_model.model.predict(np.expand_dims(seq_1hot,0))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'GTGAAAATTGGAAAGTTGCCGCCTTCGAGAACTTTGAACCGCCGCCGATCTGCTCCGGATCTGTTCCACTGATCCAAAAATCGCCTTTGATGAACGCCGA'"
      ]
     },
     "execution_count": 164,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "seq[0:100]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[False, False,  True, False],\n",
       "       [False, False, False,  True],\n",
       "       [False, False,  True, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False, False, False,  True],\n",
       "       [False, False,  True, False],\n",
       "       [False, False,  True, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [False, False,  True, False],\n",
       "       [False, False, False,  True],\n",
       "       [False, False, False,  True],\n",
       "       [False, False,  True, False],\n",
       "       [False,  True, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False,  True, False],\n",
       "       [False,  True, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False, False, False,  True],\n",
       "       [False,  True, False, False],\n",
       "       [False, False,  True, False],\n",
       "       [ True, False, False, False],\n",
       "       [False, False,  True, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False, False, False,  True],\n",
       "       [False, False, False,  True],\n",
       "       [False, False,  True, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False,  True, False],\n",
       "       [False,  True, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False,  True, False],\n",
       "       [False,  True, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False,  True, False],\n",
       "       [ True, False, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False,  True, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False, False,  True, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False,  True, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False,  True, False],\n",
       "       [False, False,  True, False],\n",
       "       [ True, False, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False,  True, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False, False,  True, False],\n",
       "       [False, False, False,  True],\n",
       "       [False, False, False,  True],\n",
       "       [False,  True, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False, False,  True, False],\n",
       "       [ True, False, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False,  True, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False,  True, False, False],\n",
       "       [False, False,  True, False],\n",
       "       [False,  True, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False, False, False,  True],\n",
       "       [False, False, False,  True],\n",
       "       [False, False,  True, False],\n",
       "       [ True, False, False, False],\n",
       "       [False, False, False,  True],\n",
       "       [False, False,  True, False],\n",
       "       [ True, False, False, False],\n",
       "       [ True, False, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False,  True, False],\n",
       "       [False,  True, False, False],\n",
       "       [False,  True, False, False],\n",
       "       [False, False,  True, False],\n",
       "       [ True, False, False, False]])"
      ]
     },
     "execution_count": 161,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "seq_1hot[0:100, :] # ACGT Порядок"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 182,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,4))\n",
    "target_index = 0\n",
    "vmin=-2; vmax=2\n",
    "\n",
    "# plot pred\n",
    "plt.subplot(121) \n",
    "mat = from_upper_triu(test_pred_from_seq[:,:,target_index]  , target_length -2* target_crop , 2)\n",
    "im = plt.matshow(mat, fignum=False, cmap= 'RdBu_r', vmax=vmax, vmin=vmin)\n",
    "plt.colorbar(im, fraction=.04, pad = 0.05, ticks=[-2,-1, 0, 1,2]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 183,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/neil/anaconda3/lib/python3.7/site-packages/cooltools/lib/numutils.py:1317: RuntimeWarning: invalid value encountered in true_divide\n",
      "  val_cur = ar_cur / armask_cur\n",
      "/Users/neil/anaconda3/lib/python3.7/site-packages/cooltools/lib/numutils.py:1317: RuntimeWarning: divide by zero encountered in true_divide\n",
      "  val_cur = ar_cur / armask_cur\n"
     ]
    }
   ],
   "source": [
    "transformed_arr = transform_hic(arr[int(seq_length*FACTOR/2000):int(seq_length*(FACTOR+1)/2000), int(seq_length*FACTOR/2000):int(seq_length*(FACTOR+1)/2000)],\n",
    "                               arr_raw[int(seq_length*FACTOR/2000):int(seq_length*(FACTOR+1)/2000), int(seq_length*FACTOR/2000):int(seq_length*(FACTOR+1)/2000)])\n",
    "\n",
    "CROPPING_TARGET = 32\n",
    "cropped_transformed_arr = transformed_arr[CROPPING_TARGET:transformed_arr.shape[0] - CROPPING_TARGET, \n",
    "                                             CROPPING_TARGET:transformed_arr.shape[1] - CROPPING_TARGET]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 184,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQgAAADbCAYAAACRKw/KAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjAsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+17YcXAAAgAElEQVR4nOy9aaxl2XXf99vDGe/85qpXU1d1VY/sZqtbJJuDJFIRJEuyAztyIiu25ECIEMAOHERGICWIg3wJEiCIHQN2EAU2gjiBHdgWEJmWRSqkBFKmKFJUN8lu9lysqq7xjfe+O51hD/mwz7tVTbLJbord/RS8BVzcd88957x1z9l77bX+67/WEd57juVYjuVYvpPI91qBYzmWYzm6cmwgjuVYjuVN5dhAHMuxHMubyrGBOJZjOZY3lWMDcSzHcixvKscG4liO5VjeVN5zAyGE+CkhxEtCiFeFEL/2Hunwj4UQW0KI5+7ZtiSE+F0hxCvN+6DZLoQQf7/R92tCiB96F/Q7LYT4PSHEC0KI54UQf+so6SiESIUQXxJCfLXR779ttt8nhPijRr//WwgRN9uT5vOrzffn3kn97tFTCSGeEUJ88ijqdxTlPTUQQggF/APgzwEPA39FCPHwe6DK/w781Lds+zXgM977i8Bnms8QdL3YvH4F+F/eBf0M8Kve+4eADwF/o7lOR0XHEviE9/5x4P3ATwkhPgT8D8DfbfTbB3652f+XgX3v/f3A3232ezfkbwEv3PP5qOl39MR7/569gKeBT93z+deBX3+PdDkHPHfP55eAE83fJ4CXmr//V+CvfKf93kVd/x/gJ46ijkAO/AnwQWAH0N96r4FPAU83f+tmP/EO63WKYEQ/AXwSEEdJv6P6eq9DjE3g9Xs+X2+2HQVZ997fAmje15rt76nOjbv7BPBHR0nHxn1/FtgCfhd4DRh678130GGhX/P9CFh+J/UD/h7wXwCu+bx8xPQ7kvJeGwjxHbYdde73e6azEKIN/EvgP/PeH3y3Xb/DtndUR++99d6/n7BSfwB46Lvo8K7qJ4T4WWDLe/+Vezd/Fx3+LI7Ld0TeawNxHTh9z+dTwM33SJdvlTtCiBMAzftWs/090VkIERGMw//lvf/No6gjgPd+CPw+ASvpCyH0d9BhoV/zfQ/YewfV+gjwF4QQV4B/Rggz/t4R0u/IynttIL4MXGzQ5Bj4eeC33mOdDuW3gF9q/v4lQtx/uP0Xm0zBh4DRoZv/TokQQgD/CHjBe/8/HTUdhRCrQoh+83cG/DsEMPD3gJ97E/0O9f454LO+CfjfCfHe/7r3/pT3/hxhjH3We/8fHhX9jrS81yAI8NPAy4SY9b96j3T4p8AtoCasHr9MiDk/A7zSvC81+wpC5uU14OvAU++Cfh8luLhfA55tXj99VHQEHgOeafR7Dvg7zfbzwJeAV4F/DiTN9rT5/Grz/fl38V7/GPDJo6rfUXuJ5oIcy7Ecy7F8m7zXIcaxHMuxHGE5NhDHcizH8qZybCCO5ViO5U3l2EAcy7Ecy5vKO2IgjkIB1rEcy7H86eUHbiC+nwIsIcSv/KD1+EHKUdcPjr6Ox/r92ZR3woP4APCq9/6y974iMNf+3e9xzFG/OUddPzj6Oh7r92dQ3gkDcZQLsI7lWI7lbYj+3ru8bXlLhS6NS/crAFEcP3nx0cd9Him0AJy5u6PU4XDnwNnmVALwCO/wUt3zLz3CmrvbRDhe1EXYJgSYGiHl4hwA3tT4ukLECQgJSR6OURE4y5mT6zx56ZwXSQZC4IUEUyGExFcF3tnmVB6hFN5abFGF36klOI8HbFmB89jaIaRACIhaKbasUUmEdx6hJKrVRugIF2VYD665egelCew2IcgjSaIkylsKL1k5scnFRx/3UgiUCOcWgJKiud6gyinomBKNFIJIgjAFXmpqNEKAxsF0iPcOEIgoBu/BGpAy/O09RElzHe/ecl/NF9/hXfi7OebMxipPPv4+j1ThCFuB9+EaS0Xt7g4bdc+y5e4ZOfdy+qQA5W3zQVK5eyut7jmXCOp5D/PaUTvXHC+ImmujleD06VM8+cTj4UYtxhlQTPGmBiEQOoK0DYdjybm741HcHYOUM3xdhfHQ6uOl5plnntnx3q/yZ0zeCQPxlgqFvPe/AfwGwBM/9EP+k5/5HBujV8BW2N4md2QfJcIA78USURygpnu4OAuT2LvwLiReRQjvcWkHOdlGVnOAxb5ytg8qxusInEOWY6qNh5HlBL1/DTfcwpcFav0MZukcJspJ9q7g4gxhKuR4G28q/GCTee80pXX0zBg528e8+EVUbzkMGikRUoFUlC/+MfObW6g0JllZBh2x/cVnKHYPKPZndE4tEXVy8rUB3oZBa2tD5+J54ic+ge2s4dMuxsPNSc3ezPDf/+7LzCtLL4/40P3LfPTsgAeXM66PKw4Ky6y2tBNFL9E4D6kWtGNJLAVKCpLh6/g45yAeECtJ7KtwbXTKge4ghaBlxoiXvoA72AXnUMsbmNvXuPXZL3DiEx9Gn7wPX1folRN4nQS926vckX3W3RC1fRlz5xpqsAY6QiZZuN91hVs6hdfp3QXAO4SpcJ1VXp8Hw6GkYFOM8VJhki7D0lI7TyTFtxkL10zi0nhmtUNJgRSQaXlIq0Y0E9d7z51pzaQK/zvRiuUsQgjoJYo1MUU42+hUhMnvHdx8GTceIuIU2enj188z72wiBSTjW8FwAi7rhcXMGfTwOv5gF4D60sfYKyynljpXv+8Z9R7KO2EgFgVYwA1CccwvfK+DlhMBtsIsnUN4T2kciZZ459krLL2ke9cw6JhapUT1DK80wjuoS/ZKz1JnHV9OQGm8TtE7lxHOYHonEc7iVYSspqjJNnI+wk+GyN4KlPNwk51BubD6u/YqoprhOquIao4oJ6TFPqmQqP2rUJVhv/EQb8IxSIVs91GDNczl62Rrg+BlzKcLr2LwwEkAbFFRHczCYbGmffYkevkEjHcQ7VUK6xmVjmFh2CtqKuMWxuFMPyOPFLPasZprllLNXmHItKSXKCrrURLakUTYCmEdPgqTszd+PVyfKMe1QpuDjqvAOUQ9R62eDDoXU+z+Fmp1k9N/++8gqyk4E4xx0lncD2Er1qISUVf4qmD6wtfpPPURZGeATXth4iUOL4Khl9UUH+V4qYLxjXNmpoMAEg3ixlfRnQGsP4ASMZX3WAdahpUfYGtm+CdfCZHsWjfl0kqLPFJ0Yo06NAp4bo9Lxo1RuLCUs9qK8D54FlkUXJVOrBDTOcKZsOBEebiX3iE2H0R6B0rjpEaYirQcgTPI+QhsuKc+zoNnKSRmcAYGZwC4Mqq5NS7f7hw6MvIDNxDeeyOE+JuErjwK+Mfe++e/2zHSWeR0F9vbRHjPnuwwN4bKWYwLK6H10IpaZFogqxm18MS2RlTT4CUUY9Jel93CYl1GP1IkJoQWPm4hy3EIC3SMi1uo6S5+GlYGvAurXt6DNKwELushJ9sIU2H7mwhRgndhxfUOt3cHNx6iBmtULz8DOgII3gRgbl/DFiXeWg6u3MZWNdXBjHS5u/AYRONLF7sjADr3ncZs3wg3ZeMBaucprOO1vWBETi1ltFNNP41YyWO8h1FpySNJrAS7s5peGhGrEIJk0oOtEHWB8D5MAFM1hlLjdcLcK2IpiKpxmBDO4OM28sITCO8xz30ekaTY/iaMbiGrKU5IfJyFCW8qMBXq4PbCM0j6HWS7j0t7uCwYCGEKZDlF1LNgMKo9RJwGw+oMsyqECx6FHGzgvUOOt0m6p5gbsN6DE2jpiZRgVNR8/quhQHVjvcNqK2ZWBy+qFStEE1bszCq2pmESP7TaYilVGBcigsMQI5aE3+HDffFJ5+54Tlph4hP2EaZCFAcIWyNMiSvnwdvouoVXW8dtKutx3nNrPOPGuPi+5tJRkHfCg8B7/9vAb7+tg4TkjuxTGsfcGL54fciFQc5KKwaCxZ8bh/WCXjUlyzQuaSFcgldhn1gJnBfMvcc4T1rN8UkbH2XB4psKOd3Fx228ThF5F5zBRznRqfvxziBn+9ikHVa32uDjHDnbR+1exaftMMmcCc6tlLjRbjAyQLmzi57NSNp9ZKdP0j9gfG2L6a1dqvEMbz1RK6U6mKGyGCElOovRrRRbVJQ7uwgl0aubiOKAdj6gtJ7ro4J+FnF2pcUgjxcronGe3XmNEmFl/exru2gpuLjS4icvDJDT3eAV6RjqEjm8hc+6FCv343yYdNZ5vPwW2Mg7XlGnuHFQ8qPdJdz+NuIrn0Sunwn4jG+ce6lBGkRdI8sxbj4FqUg/9OewnfUQghziRELib7wcbvXph/Cvv4DMWojucrg/jcwqy7PqHGd7Mb3XPkc6uButWu9x7i6uAuCauONwW2kclQ3e56FIIXD3ABgqwEJ3YQMApeHQC3yzIWrrYFwbHMULgdBxg1n8/1PeEQPx/YoSkGhJ5SwXBjlfuj7k/pUWFwY53kPeuIQ+7QScwTu8ThCuwLWW0YTBoIQnlzZ4AfMRPkrDZOmuw1iGG20rXNLBp91ww8dbwYgUI/w4RjiDS3vBQMxHEKe41jKimgW3uAxxqsg7KCnxpqa1fgZz5xqTr36ZZGUJGWvKG2PS5S7tzVXq2ZxyOGG+e4CKNLqVUo9nVAczsrU+1XhG0m8jWh2EKRHlhMKk3L/cYlTUDGdhABvn2ZlVdGLFiXaMbmLvHz2/zNak5F8/d5vLezOeONnlfWuKWGlklNBevYCXCo1DuAqsQZgSOdzHXnkONVhDdJZwWY8HJi9yqZ7BmUeCGz3Zw02G+P3Qk0aeOA9JK1zXgz2KKy8iO330+hlsexU520eYAtvbDKGIilG9ZVwxBaXZeeAnmNSOC9NXEPWc1VaOFFA7zydf2ua+Qc6PX/oxhjNLZcPkvjKcszevUUJQO8+ff/os1/dmbPQzzg9yro3mjEqDFOF6JFqyksdIKXDO84VrwwWO8aHTPS7e+kIYT2WBPdglOn0J0lYIv+I2RdLDeo9vjEkuFfbys+jN87j2ajC+jXfh4+wN4zhu0NEPbraRos0vvgvz552QI2EgvFS41jKqcnjnMQ5WWjH3r7R4YLmFB0alIVYRsZR4oUE5hK3vYhCNezitA6CFt1ihkYBBIlqrKG+wg9PBi1ARarLDtmizUu+BM9ilM8jpbrjpzqEmO9jeBkU6ID38H1LiOqsw2EQVY6ovfwp0RHTyPuz+FuZghLeO+PyjyPSbFLsH2KqmdWIJWxlu/tuvYyuHmRdUjetrCsPw6pCTHzqPVBKZtnBph0K3mM5qChNWqEvrHYbzmvGkZF5bauvJtETJkEdYb8VMSsP2Qcnt4RZKCu5fyjEOwBNnvZAlAryQSFsjygliuo/dvoGQCp218FJTPvdFips3af/l/xRhK6QpAtYyGwNNFsLE+GKGn09x04OFJ4WKkdUUO9rFN7G48ECSIQFnDbtzy81xwYUoAJVpFnIPSgi2DgpSrbg9rUMCpVn9t6YVt5t4PosUF1daKCkY5BGDTHNtFDwIfY+HkWhJHimc87yyM+XWMADYj663qV5+NmRZCEZCLZ9AxgmimuNVjPeeyvpF9iSXGj8b42cTyAb4fLAIP1BxGB/O0CROQqZluosoJz+YifIeyJEwEALAVvTimL3Ckjaj+MIgxwN3JiWTyrLWisJqOR+H9JyKsTImmu2CCpOtFXWIlaDyMbGr8Doh2b8aADqpEbbCC4maj3Bph9XyNnK6i/AeRrcR1SS4wzoN8XM1J87isCI6E3ALqbD7WzjnSB59mvrqC1RXXkD1lolPn4fXLzP/2heRkaZzeo16VizAyO65DebbQ5J+m3oaYlNvHXE3Z/nxB9Crm/iqAGtIfYl1gkllUULwoVN9Pn91n+eGI/p5zLl+xlquuDU1TCpHogVr7ZgfeXCVB1bbPLza4rSa8nKZcW1U8L61FnPjuDosWG8nnGj36NczfGeV+Id/Er97PYCSdYXvLZNJhf2jf3X3PiVpyNI04oXAr1+EEw8RP/hR9N4V7HgfNXyd8qVncLMDqgc+Hrw/USOSDspZuPkyD+pv8lCc4iYzxEoMWUgYKgl/6X0nyLSiFUkmlVv8PylCNsY6zyCLeHitTe08m92UC2nFS1qxNb47GefGLbIWssnkHB4PNKlI0Csn0JfOgVT4yYj6+p8AkA1WSR77CYYmTBMT5SQPP429/FXEeB/xwIfxSSucC3BIlNSNrkGHUTzA6sGfYna8PRFCnAb+D2CD0KD3N7z3//P3e74jYSDwDllO8bKgl3SxDcocQDjDpLLszCpOMoapwUdZSGkWY5SrQgwrJaKakytP2YwpI2OiusBnfUQ1wUsdVocmWyHLKT5KQiiRdkMoITVeSMzK+bsrQl1gB6eQ422UCcCjWj6BUAo3PcDXNXr9TED+ywIZaUxRofIY1esg8ymTqzeZb+3T3lxFKkk5nJAtdzFFxcpj9yPiFJG2wnucYrrrIDWdOIRbr+xN+a0X7jCcVVzbnnJ2pUVtPaX1DAvbhByaO9OSWWXZn9eU1mO7PaK6ppfohbvsGuyhtj7gMbII4Fuc4p0DqRBpCykVOIvQMSJrUV9/FTceAqBPnEPEbXyDyQhngscwn+KXTqLXNnHzPlERWjnKIhh1IPzOBtS1o110u08yOMw8QB6pRVqzdj7gBcBaKyaPFNZ7eokmkoJBFtGJFSZOyaMDNvvpG4ZWOJck0ZLdXkoWK6zzpFqhN4J3o3rLsLS5GIuq+Y0iilGTbQZZM8Erh8sHAa+qK1xxgHQmpNCVQR7yQpxbMDEcKYV1vIty+AyVPxFCdICvCCF+13v/je/nZEfDQAgRJieAkLSisNIFdD5irRVxkjGiOMAOTmOQwYDohJHVRDKi5eZsyy6TsUUIONmO0ILgGZgCUZe41jKumfR26QxeSLYrRZYJWpHE5gP06EZQo5ohqinzbIVE0RiTFJv1Az5QTAL+kbZQvWW8swipiO5/DG8qqpefRfWWUeunqV79Gu3TG4xeu8HBlVuoNCZbHSCUREYRyWMfgc4K9vJXkacfwkUJwhqcitloSwrrmdWWv//Pv858XOFMzVMXlgDYmVue2xpzeXdGJ9WMC8P2QcGkqAFIVJ9YCc70kgbk9fSS4IlVzjNPuiRJCzW+g8i7yCQL4dfaKQC8TkPGIsopvvQ5Xv/sVwF48NQFVGuAnI8CnrO/xfhLnyNdW0U98qPQO4kwJdXv/KNgLCNN+ugHkP0VRGcJ26RX66/8HuiI9pkfCliHkEzjMNGMC+nucWVRAu5fykhUmHrWh+8fWsmprOfF3YKVPGatFbgZzodV/GwvJVaiyez0mDTZkm4iUY9+LPzGuM086VFZhxSCjopCpsUZ3Kt/ghsFTkN8/hH2N5+EtQFtaYhuPAduC+IkpEYPww0hQ/bIOxhcYla/ewbCh96jh49CGAshXiAwmf8MG4jDm5F2wDsyLbD+MAUlQ0w5DfjB569P2WgnLGUKJVIq65ERICXW3E1fFcaRaQlZD1EITHsVi0RlISXqVYyoCzKtqawn1cFrcWmPQrfIij28Tkh9iT8kYzkTACzVhCo6DaxDZ7G7t5FZCzc7oL78POgIkeaYW1eYXbmGSkOmJem3cdZRz+boNKE6mGJuvEZ0Icfub6MOtvCr9+GVZq+w7BeWa6OCa6OC5Y0OVb/GVI5TSzl5pDgoLePSUhnHvArvZQAdmNWW0nq0FBjnGRaG2npG5SFZ6B4Y35qGBxK4EKJJWXoVgTXIcpt6WiCUXKRpgTAZpEa2+yT9DiJtNdkNCbZe7OadW7j0h3gRKib9kb+E1wmVB3UYx+PewJp0zWy3LriWUgg8fpHKdD781kTLBU/iW6V2nvoeppWkYULKu9kOJcUb/m/YUS7Ysd7Zb6MEe2cRVYlQ8SKbFo7TDRv1vZNveYbK9yVHwkAIW4dBJSToGFnN6FVTfNrBC42cj/FRhkGy0U54YXtCL9Ws5DGdRKGlZKxi+qmktIKD0hErifVhpYmj4HZGs128TnD5ADW8gct6dOtt0ClieNAMakPa1fikRSlTIgmV9Uy8YpU5w/wEuZYk5RR/6xVkd5no5H3gbAgxpmO8s0TLZ6guP898e4itDLqV0ruwSXUwZb49JO7kiDwj6XcYPvNV7B9+iZWPf/wuCGgq5nXCK7sz/vj1IZe3JmRZhI4Vzjg6saZ2jjujEiWhn0eoxhAA3BoWxFry8GobWjGRFJTGEUnJSh6RR4pUBeMrp/tw82XExoWFIfTN/XD5AFHPkZNdBh/9MVobS1z99JfxpkaYkD3y3iF0Svzgk4FnUo7xcStgOdah05jo5H3IUw9is14ARm2FB8zKeSySuXGkWqKEZCkNn7dnhnFlAz3a3X3izbR2AQAUgoPaMiltSGU6QMKwqBdhqkmhFUl6iSSSAq8kmRbkkUDMbZjEtgqs0sYbWpCevEOeegBlisB9ye7BElSMz/uLlLcoJkhV4lWEy3oLY1Ea34DEb01Oi8wXvPkBO1TPA/cSK36jYSW/cU699WeofFc5EgbCNy4ZPqNWKbXwZJkOqUwVqK8u7aAELGWKXqrJI0VhXEDyuzFKCkrrmNehG++sDiHK3DiiuLHszaAPWINcrH6HxBevMlCaoYtZKu6QRjkHukM7klw9sKwmmo6GyoNZvR+ZD2Dv9YBDFDO8tbhZQPPNnWuYokJGmvzUBkIqZrf3SAcd6vGM3oVNhJKMXrtB58w6FYCpA76xfwOf95mpU+SR5PGTXS6utvjfPvMqzjiSLEIK2JvXvLA14ewgY7OTMq4MnUSzNyl57pUdhuOSj5xfRspAnFprxWRaspb6hvQzRRzMEHcuc+M3f5ONn/gEanUTZyrEyhlc0mIo2/TabdT4Di+d+QSbj/wk+nN/FTfaxY12sOsPhktbTplfPEsiLPrKH+P7J0JIVhvi/hrifT/GK7aPrTz3d1uo0S1UEchZdNbZnVs2WgJNTb53jaR3kmcPgoFwziNlqJ2IleC1/YJeGrGcKV7cmTIpDU+e7C34Dd/YnlBbv/AmHlxp8cHNNhutkB7NtERXk7sZh2KMLKdUX/k0Zm+H+Of/y4DfeJiqJYzzC880acZsaT0snQnkMu9Q157F7u4iohh//gQ2D8ZktF+xN695q1Li+PfDo06+o/xDf7Xw3j/13c7xJs9Q+b7kSBgIAFnPAwBUz4htjUtaDY23xqs4hAU6QYmUlTymMI77l1K+dmfKqFSkWpBpSWUtmZRoGfj3qRLULsSjMkqCWylkACxNGVBo7/BJOxgK7+nGEgoQ1YSxayGFYJAq8CCnu8TtVUQxCQxFpULuXMoAvDVYhJnsES+vUO3u4MsCZx0qjamnAcSM+13UINTuzG7vMdvap3P+FhrwrS60JCuZYi1v8wevj9ieVjjjWF9rk8WKq/vzRYpvJY/pxJrX9mdc3prw5Re3afcz3n+2T6IlvUSTaskruzPW2jFaxqQ6I89z5MTgT1zixN/+70I6rp4jJntwsIXfvc3SpR+GObhswNleTCTh/l/9VepTjzGykt7Ln0PoGL92H7XuIJQmasBNnKH15EdD+rOcciF2+LiFU+1AYDNluBfApLIUqSKOY7Y691EZz5lexLVR4D5Y68kjSTuS7M1rZrUFUm4cNMCnOCzS8uxMKpQUtBPN+aWctXbMpHZEMoQnlfOo6C6YKaoJdvsm0an7ic8/Sk2o4XDOYZxf8DCsDPwGQePNSB3Gjne4Ew8he/v47SshXG5iEQ+Nrm9NBBB/K3HtXvkep/ouz1D5vuRotJwTApv1m8IrHQAeZ0I4EGfU+XJYzQiYQydRKCH42p0p/+aFOzhPQ20Nbue9l/cw7jwMt12cB6wDQITMxwJc0jE+SiisD+BclNNPQ11DKwo4BCpGlBP03hXk6DYAvirQJ+4LKUDn8M6i+kv4Yko8GCCzFt454k4ejEO3Rbk3xO5vEy+vELVSBpdOYyeTkOKUEttawnrYKwxbk4qrO1MArPOMZjWjWcWstuSRYtK44cNZze6kopjVDFoxy+3kDXH3flGzNamYG8/cOCoHqBjXXuWG72D7m/goQ7T6IXPx4NMh+1DN8XFOYmbIaoZZu5/rU8+ticHtbwcQzwZX23vfVIC6UBjXXUK0+iE7sH8DNd7C+kBhdq1lXNajsB7bcA7mxnNnapowUZBoRaQkkZIhddhMvsI49uY188oGNqgPtReHP7c0AXBcyaNm4fBNRiRkcKzQi2Iz4T2+mCIHa/ilzYaPAXETgh2+onveA9fGLcaRzXq4fIBMQ9pTirvVpeq7zPfvMBWIpXjT11uQjwB/DfiEEOLZ5vXTb12DN8rR8CCkbsC/KCC/QgYQ0RUI7wJ2EGVEMkJGoKUk68aMSsVf/+EzaAl788AVmBsXwCvAuADEScGCNyGLg4CU52tkwoZqwjhHlsHlxDsy6XFZL2RLPMQKWiZgA9gqULCTLsQOZsNQmj0bI9IcEaeo938Cbr1CfeXFQKwaD4lPnEa9/xPEn/+XzK7fptyfMHz5ddLlHjLS2MqQnVgHwO1vo7o7pL02wwJuHxRsHZSsr7UZTSt2b4156nzIYhTGMS5Lro3mfP6lbc6vtviv/4PHKYxlf274/Ze3+VeFIdGSv/mx8yRKNBRrwbiyzFWHydTxwvaU/pkutr3Jp17b57H1Dj2lOLH/AnKwwUE8wDjox6Bf+yLFylN85eYIc+kvkGrJSh74EYkSIbtjK0Q9o3z28+gT5/CPfBy/fA4vJEXt2DcRle/xDz5zhbVuyicuLLMzM9wwjt/4wys8fqbPz15a5eJSymNroXhqVFr2rOeBlfDZefjQmQGT0vC1O+Nmm+fh9bAAJFrST0M6dFjYBT6jpKATS07mA4Stce1VZhuPYV0IS7rbr6Kb0CGfDhdgptvfZvr8MzgguXAJ8dTP4nUIXwsifNRFnX6Kynls6aidJ9WCc/27LMvvJYdkse9XvPd/wHduufB9ydEwEBBWc+9DzUDR3OzWcrDSqgIpabk5NICkkoJUCyrr2ZtbOomksK7xIoLnoJSgNB6pQauYuROkaRdpCjJhG0JUHICpKEVNdhb0WdPk3kelDZWl7cDPSHQB5RSvU4p0QEqXX9kAACAASURBVO5eDrUJTSpMZK3AwtvfDpyGKEL1lsOKfOcybjZD5yn5xjLdi2eo9oaMXrtB1EoptrZJViz69EVse4VRM6jzWDEpDEvtmCxW7N+ZkMeK84Oc10cFTnpALgbWzqwikoL9WcU3bhxwejlntZtyZ1KGisdEU1kbYurmGtbOMyotsRRsdlM6iaQdS8TSSWzWY1aHOHzuJK2TDxEZwZleRqQESkJtfQBJPcj26mJ1ji9NEO0+tdIYJJXxTGpHZUO9zBOn+2gpqG2gGgE8fqZPL9EclJZIapz3i8Iq6z2CAMYelIa6yagcegaHJd+H16Iw4bym8R7Czk2fCSGDnlI23wP4RdEasOC2iCRFJCnZhUu48RCZd74to+F8OId14W/V4Bb3enHfcxrwlj2Fd0WOhoGwocrQddbZKz1pr0usxBuUE1XgOVhDk60IacxEhcFwSEYJAKUnUZKCMBAnVXAXHQ4TSZRIiIVAi0DgOSy+cWknlIOP76A761QOMi0YpFGIW4Vo+hk4TJQTCZpqRYPcvLQoWHIv/FtkJxRs+bpCLZ+gevkZzJ1r6KVVqqtX0etnKK++gpkVxN0Ws619hFJkZ85hd28TJRmnNh7i2Znj/pUW93+0RSQFL+9MuXZ7zPmlnFPdlK1pRVsqpBQ8ca7P1kHJv/jydT58aYUbezMuf/UGT//c43zgTJ8vXN2nn0dcGOT0Ur2o/LQ+uPivj0pOdROeXvYIO4bCMF6+yLx2jOaWRAev48pQ8b41yVMn2xgXjq2bCX9QWrzqc3tcM64sZ098gExL+k0UN60dl/cK8kiRaMlffnCJmRU8vz2jtpBqyS+8b529ueX3r+zxsbMDYqUXhWWHk/igNHxje8JaK8Z5GDdVrRDGg/Ueaz070wrZTLi0qZ6V0uN9E84ClYiZF/Zu2rfBqBAyFGOZOjQaWjtL9fBPkO9dDlwa9cbpE+qAGmp4YxNiJd6WRyA4KnF/kKNhIATIyS4+SlnqrLNbWJwPgNK09rSiDrnyCxJUaQXz2lPZu2GF88E4nNEziqTHqAyUbe8DNz/TAk+IITNqwk+Xof/D+E4YDKbAtldBSPT2q8h8gMqX0bZAqBRdB7q0MCWVyqisZ6mJtcVhaXkRsAJvLSJJ8fMp9esvI1tdaHWD0bj8TczuLUxRMXz1BvPdCd0zKxS7I8zObVQxRZ++FKjkfsYDyy022pp/8tXbzCrLX//4BfJIcW00X7AOAWIdWIL7u1MSvcaDJ7u88OA6/+YLV/nUH17jP/6ZBxik0eKyJ0qwnGkyHQC8E+2YfqrwMmdYhxTdkhSMHby6N+PJEx06uWQ1Dw1p5vfk76QInkgWCXrVPmtyApnkVRMYv+04DLXVTLG24hD1FOEd35ysYB2s5HFzHhY1Jh8+PcA1Je0QWLWH1IVOovn4uQG3JhXf2J7w6edu8/GH1zjRThbl3RCYlLkKC8BeUS8Aw/25wvrg+gsRCsC0A6kEB4MLzcJiOd07iZxsh3sqJOn0TjjGO9TeNWx3Ax/nZMIjZ/vIG98guf9phnVjrCSh6cFbnQrie4CU77IcEQMh8XkfWc3x5QTrMubeo4RfpLbKe2r4D0qH955MSpwPrq8SMKs9RdJja1rjACUCyNSOQw5cC5qeDuNFN6rDhic+ynFJB2GqgLDrBDkfEUmNFwJtKkSDuiMkreqALO0gxiWinoUOQo0nIrMWSIUb7QbDkLVCKbSpsbu3sbWhHh2Q9Dt0zqyTLncZXDrN3ouh6ZDsLeOSDtM6xLHP3j5gVBieuz7iRD/jXD9jf15DHeJs64N3o6SgMo66tMwqy1o74fzJLp9/9Qrz4R3af/GRAPZJQSRDD4lYwrxxybUK7q1BMC5rCuvppwpHqB4VgmBcJUycXqD7oUYClActBWqyjb35Wri35zfuWflB1DOi3W/iyjlCKoatPkoI+undoWhcmFj9VLFfWCaVJVKSSRnA2DxS5JFiKQtNcqaVZTgumTcsyW/NGkQNSrjfZD+UENTWkWiFEsGI9BJFY4eYupAu35vXrLVy0qzfKFagZvvBuyBwH0THh74YdYGcj6ivvYw8+wQQcBLBG8vT34r8aTCIH7QcCQNx6M4BoDT9SIXYW1rwlsqH1eVkO6IwgQQ1qx2HJf+egDkkSjIqLQ74P5+5yZOnejyy2lq0IVNaBLwgyvA6boq0HC7tAYSmMt7hmrSnlzp4BUl70UnJS9XUjoxR9QxGt3FVSLXJtIWvCoSOQjHXdIw3NTLv4MZDiuvXiJf6VAfTUO6dxuRrfea7oTZj5YmHUaubqHOPgq2Z1C4AcDdGVMZRGcfz10f80Wu7PLLZ5UQ/Y7kV88KtAz58fpm9SRnqDPKIF28esNZJ+LUfv8ijp3q8cPOAtVaMEoJZbUNY4DyVCxmgnVnF7iyhNGEmn+0oZDFmu1LMa8+5fhbQ+Kat2iEJLUwwSTa8iv3654guPYEb7SDOvx8ftzmvGg6As9wsNdZFbG4+Tu2CK37WhtBhVrtFe7gQHrCoyiwaivhhKDPjriG4Nio40Un49Z95sDkWbk1CxWck5RsmW20d4zJwRQ7F3svY9FA5z6bZRk22qC4/h1YfDFjY4fh0JoyZrIddu7RYYGQxChT1D/4sRdTGVYdG6u1NdsmxB/FtIpwJlYFRitcpiSlIq9ACzgpN7CqMjNHNRLdNb4jDvoPGgVKCAkeqBUpInjzVY39ec2daM8h0SFU5QSxkk63QgQ3nXUMN1oFKKxU+7QavQOpQBQrhe6kWqLWY7uF11DSdcQgd4YtpeHc2FDsRQC5z6wqy0yfutsA5evedwFZmUc0Zd5rVJk4XzUd8nIVeie2Yyjj+6Pk7/Hsfuw/rPJ/++m12JxWnlvJFeDEpzQKkS7KI7VHB5e0Jj653GM5qJoVBCdGQpkKa2DdhQmU9eaQWMfTMeMYGkqSHNo5uIjnbT5seB4EUlaUDwIV+odUM4T3RfY8EIlrebRq9zhB1qOew+QBdWxIlELbCElFZT2mCobozrRv9INWNJybCq9d4F/vzmsI6imbu9dKI9VbMqDDszGrqBqQezurQMCYOxuawYcyZXsaJTkKiFZ04eCEAkQye2KGn6Ys8pLQP77uKmp4WEVanqNENRDkNJe8QUrwNC5jGuzj0GpR8mxjEcYjxHcS70IylnqN3Li9WbTkfhUmmE6KmbRpZD9sM7ECC8iQ6ZCtC7X7AHB5ZbXFnWrMzq0Ls6lNaURgEcdRBCEGcyrsErPFW0zJM4LIe87iLBFIzx2sWdRjCGtR4a9ElycUtMFvBMJg6IOI3rxCduYRzFtkZ4OZT7P42st1HJCmueB2dxgglEVFMuTckWeoH+nJTy+B1yqRyjApDZRw//uQphrOa63szvPNcvj7ilz54hhsHBQ9sdBaDPYsV/U7C5Vd3+dzenCzW/N6zN6nmBvuRc+RKEcWCSMlF5sK60K9xOQudrUdlzZdvTlACPnK6S1TPWKu3sZ2zCFOg96+RLCvipE105cuhTLq1xGTzCRzQmu+gRzdxuweIdg/XWma/6DJIFVE1QQ13iJbOMasdk9ry+qjg/315m3YakWjJZi9FS0GqFY+stVhp9Hpld8aNg2Jh4FItWFvKuDWp+We/9yrzyqKkYK0bMii9piS+dg4lBE+dbJNUB4CFhnYtvMOrmMpL8v0riPKAeuNhfP8U0dkCmw8wcRtPAFi3S81DQuK3riKjLCwgQoK+S7w6zF5AmOxvhwdxePxRkSNhIIQ16Fsv4PNe0xOxFcKAKMUgQz+HrB+qMgtBHKVEcUztAkosRTDck4oF5iAFDDLNqDSUTQGTFNCOJapJPXkV3GWftLDtlQXGMDOhwCkmtJwbVY5WlBKVgZLtklbI85cBkBR5F7RGVAUsbRL31/BSI53FbN9ADdZCMVeri5tPUf0l3Hi46AvRft+TzX6rqN4yprcBznFtr+T1UcHlrSkv3jjg4okOk8JQlYZON+WlnSmb3ZTPvLRFFitmlWVvUnH7zpjR1pDWoBNSj7VlMiz49EtbLLcTlpuy6ZU84mQnDOzCOByeXqxoxwmdRDEuLd/YmbPZSdiYhJ4ZPkow3Q2KqENlPEuANxVivE27miCcxY52cevn8Z015HgLYWp6XUU0uRMeJxAlxMWIlSjhoAzl2O00CgQm5znfZFmW0uDVWO9RjateNfUkhQlFat7DpDKLY++VeWV4fVRQO896K37Dd8jAtcFWWEJoup2cQqawWYzD9sFmyGw1h7QjSa8fw20Xem7C3bqVRrwQ3/6Mh7ch8jjN+e3idYxvhzjP9E4GLMCZgBG0VvFRiqgmiLrEtO8+WkCKwJAUpkCrmLgpxIlkoF1rKXA+pTSOz17eZbUV8/TpPu3YL1JiLgv4A01vQWEKEiVQrkJYQ6n1YvBFUuPiPIQ941sNkLkPUajkE10Nk11cdw22vhlapactRKuDuXWF6UvPkyz1EWmO7PRp95Zxk2GgacdpADnjpHFXNd3EMa5Crn+4PWU8yFBSECeaixttEh1c4j//6Aav7M5CGCEFVWmROiZuYu1zp/uMlluMZvXd3o2JRoq7/RLySGFs8MK6saQVhXM/e3tCL9H4tN0868Lh014IZYQIbejqKhi//a0QbiXB6HgVGgR7IdDVJDSGFRKX9kLbv7pkJe9gfbzoIZnFwTuQIrjnk8otCE55JDnZDffTOM/Lu1M6sebGuOD2/px2FpHFiqV2svidl7cnXN+fkcWhuG+91Wk4CoLYgyEiEbCmSl4uNK8fFCyd6pP6ErV3DZ900DpGVE0Gq+noLaI8XA9TIUSoRD7E0rQArUVYfITGvE2LcexBfKu4prGp0qEO31RIU+G66yhvmvhf41rLWGToICVDn0UIA3HuBGENDNkKpQPm0IqCN7HaivnajREPr7WxvnkGg1DEQgYew2wfTIVPu0DwapCSGEMv0cyNI1HhITwqaeOjHFlu45PuohGKy3pIQB5sBQKWVMjeMjiLbPfJlzdCQ5nDrAYg8g7e1OiT58J1KGYYGTfl6qFbkneevJtgnSfWkvvW2ygpGM5r2rFia1qx1oo5tZSjpGB3kJG1Y3rdMFH/84/fz0quefbOhEnTgCfRktI4rh8UrLcSNjoRk8qFUMQpEhWM7Ll+Fp43tHwuFHjZGmFrkuIWopxSXn4ekebEnQEiyULTmaVNRDVFlQeYlfsRpkSNtwK1Omkxc4okylHeMNi/RjI4B8D5peA5vLY/I48UgyxiZ1YtytfP9TM2uymjwnBlOOdff/02a92Eqzszbry2x30PrtLPIx5Yay8ISp95ccrXX9rh5ivXmP/C0zx9LjBQT3VTHl5JmdSO1MyZ/tP/kVN/7b/h9qTkt17e5aNn+qx96XdIHv4AdFYCJhW3KZfPY5MeMQa9f33Rw6IQCdMqeKlLdj+0LvQOs3bpDeng7yXyGIP4TuIRw1v4wWagWzcZBsYSOzgd3PlqjvMOlTXst6ZvpItzZHFAmnYxUdM7oqHJxiJgDu1Y8vTpPg+vtblxUFJbx4Orbaa1Y7MdYaUmlRpBFfgM0LRw14y9pBMFINTqGOI4pLVttSBJYavQlHWyg1dR0x/iFrITKvpEmqOWN3DDbUQUYUe7qOWNRes62eoilk/ht6/hq4KonjEXYXKfaCc8uNnFOs+tYTBEN3ZnCCl46uyA/XnAKE50Ug5Kw3BWUcwqTOUYOs/Xrg2pjGWtm/LIWptJafjU12+z1A4G5eJqi8IGoHJ/XlMYx+tNU5lWrFjJNYNUIYsRt30bITLW3TBkfvIBev10CKFOPRoaBCuNbS1jD58RgURHaWh7P76DKDLi/mlK63Fe0l4+x3BmaSeac4OMXqJ5ZXfGzqzi1rgMTM3Go5hUhtp5fufFO3TSiJ953wYHpSHWkq0zPdqpXjApIXhH/9EHz7D32An+4PJJLqy2Q6gpBaPCcGNcUzvQWc7yz/4i/uA6P7qU41djhBsjPvIXQ08LZ0IXdFOQ3n4OMziDqOfYV78SOou1uogmzew9b2BiltY3bM63Lsdpzm8R4V3oZFSOw3MX4hY+biNsHRrMCnm3TVwxDkUx81FDzw61G9IUKJGQUSNn40Wr+zgKcXg79livFtTccWlIVGBIek/APdJO8CQAlwZqdVuA8QGgasfBfTQeorSDqMuQ6cgksppj86XwcJn2EuKQjlsV2O0bTRu3UKvhpmPMrStEmxeCcdBReF7E0kY4n9LETcw9qy27k4pYy0Wqs5zX5J0kVDk6z9WdKUoKWnHwjEzlMLVFSMFwXPL8dc/1fM77N7rUzjM8KLDO088DUu+aisXCuKZVnWNWO9qxYinLA4fBGXYLg5aC9USGjJOQRIM1RGeASbqLwWRl3JRLe7T0KBV6fKp6K7So47DlPkxqF0ryVciuZE3Tl3ltqXAs5zFR0yTGeiiN5ZXbEy5utLmvn/Pa/ox+HrO2lBNrueBcHLrpJzsJp7oJqZZNZ6q7BqRocpyl8Zjlc+jdK8jtyyFV3VulXruEOriFKM2iMbLb30K2lhF1SbW/jWx1kVlrkXEJ9eNy0Q+ibpimb3kucOxBfJt4U+NVRLXxMGqyHR5qo9NAYFIRaj662yZOxajhjQAySY2cFIvCq1gIQC9qLPAa0bRIP8QcHlxtMy4NS1mE86HIK1aCPMsR5SQ8jQuH8aGtnfMB51jKFJENBV/qsJisnIbemCW4GERdLGLRkK2YwomL6M4Ac+UFZKcfejo6i1reCL0XJkPQUaBwq1BNKmf7iNY6eSSZ1ZbHz/Rpp5rffvYWB8M5pnb85Ic36MSa072U9XbCpDJ8+rk77E8rrHVEiUZpQTGr+JFH1rm42mpSeYp+N2W1m3B2pcWZXoaUghd3pigRSqQHjXs+qSx/4188x2Nn+vz84yd45taIREvWzg/oOYMqpyHrAsQHNxZP29LlAVXUAd8Qq4iZJn065UsIWTGtQ6VmOwoPQZromFe3Jjx9ekCiBX/1sfWmniIY5kO5NTHszmv+k4/dx6Sy/Mmt0AclixU/8uAqaYNj3JmUpFoRKcFLuzPu62c8faq7MBqKu71GAXyUUvmYauk80cp54tvfCGNrPgqdzfe3QqWqtWDqUOHqHfH5RwBw5Zy4GBFFAT8aRT0mWScYqtq9zTSnQN3zTI/3Wo6EgUCqponsJLip0yEi7+KSTiigSjuhJ2TDWDssqDrMOhwWXgUqfXBt5XQXVEyc3uXcK6GY1o5EBeOwM6vopxGVFXRjSaJjRD2ndIJYBc+mcgFVbs13qFurDWDmQ21H2lkAb/e+fNxCdZdC/j8Nuqr104g0D1mM5Q0A3MEusrscnrmRtELpuVUUvdNEMoB0W9OK3UnJreGcjX6Y2NfuTLDO0011SFdaxyCNmFcWU1mUkqStCB0prHHcblq9Z5FiXjvOruS0U02qFYV1KBdIRJ0soh1rOolmXBpc7Dm1FIDRGwcltXMsxxHGeUZoZNRjkLbvrpj3sFOjxt2WRfDmvMxCtkcETCht+BA4QytKeWijiyCs5t47lIREycWjmu9lI4bSb0thLLGSi9oLaCjfxmGdpxVrbhwUKBF6UPaS5uHQPtxbcfjAG+8ID/j1WC/wk2FzLhmenEVjHAjFeItjDjt8/3/svWmQndl53/c7593fu9/e0OhGY5sFs5JDzpBD0aREUbRkSZEsS1aiSpUtR2WVVfmaSpyUUylVUhV/yIfsSakS2ZbLUlVsKZKjNbIic6QhKQ055CwYDAczmAbQQO+37/7u5+TD8/YFhiLFGWqpsYqnCjXAHaDR6L7nvM95nv//9z/N8ZRUn3ssCGuxVr03rZQSuff7Zb0/Doj6muCe3MJO6zg8U2LDNoeqyUq2h/UCAcy6rmDitGDhVJ6IPNoVqMypZVvVII/T10zUwVeajaZcKwZJRTf0OE5EYNPwYpYiH+UKDyKvyUNJUVE5iqm/hFdZAke8H45WoH2M51MaS8PL5IlUlahc3J7KlDjTI1Q2xnbOYHaviwzb9eRJ1OwKqm4+RnmxMBJCAas6WvG522PmRcWNA6EmHdweUWQln/jYFpWxjNMSU7MWrx1MmUwzqtLgRy6bq026sUdWGl6+fsQLSUny7DlW2yEfv7yE59xTGU7yiqQwrC4H9CIp85cjiaj74SfWGWUlX9kdc7YdstkOiVzN/7c95GCa81OP1JmnniScqSpHlRUuBtfVOMeHwt9oRBKiQ21IyqYSm+iGtHTJ9z7QZ5wZjpOSk6SgGTisxJ5MlLSUcqeSerh3T8+rGi9oLLO8PjC0qn9d8uK2GNSKx87w1HqTTuAAGscN0VW52Ni+C6WVBnf60h+ighBv4zIqiIQv6nrizo3bkApaXzXa2OS+zIuaUnYatHP6ebyXloJSoP33YN74C17vjwNCaaqDHVQQSpCuNYtw1+ViIOlNYYcoEkMQbohKx4uDxfixXEe0I3TmKpcmmhYoyGl2gTIllXaxVvQTeSWa/JOkwNUivmq5PpOkolUH4HqOInRlTp5XFlc7ghtD3KQ+JQEyCzdRR5KqshmmuSJ9klqpqcoU63rYZIazcg6yuVwvNh6+p+2vpwS+G9ZSZsG7+67GL6XsnJ2M+NVf+n2e+sxH+Y6Hlol9h+t7E3aHKVlSYEqxNd85nHEYulSlSJjD2GOpGRDVMnbPgabvLExSR/Oc524O8LTmA/cJr3796h6d2OfBlQbGWpZil/jzv8hHn/4P+OLdCTtFKD6MQnHrKGWl4XEpPcYEDXLlw/IlUJpYKwqnw3FS8cc3R1zsRnSCPquBR14ZCmOkavI1vUgatBpF37foRJ7oc7+7gMo2fZdmPcYdpQVvDeZUxpKYis1OiFc7N31Xi26iMjx/S8jpWiuWY5+H6uqoa+eoP/ploq2HMUEL78OfkuogiKkaS8LXREx6en5S80t8qSBDyRodex2SQvoNWVktLN5FZYWp+a73gvp2BfEnVh2e665tQZZQ7t/C23xgEYenrMWGbRqeJnRBDWuGZNAE5Gm0kETfn86sXSEYNZcFhjs/IdQu1m8QR3E9749xNSy5Bdtzhyr0SasSpzALGvTRvMLRYhdOSkNlYFBUzPKCC12f0HXR+RwTtjEWPKWZBV2iuIeeHpJEF4mSI9Qpa8KLcEd3YXwsFRBiYiqXH+DYBCSZNO46gUtSVHzXlVVe2Rmys31C2G4zvHWNJCkYTDNupiVJXrHZj4h8h92jGYP9KWHdgCxr45J29aJMT2vaEoHYq8u6Cnn+jSOaocsTa9LtL4xhtR2y3PR5dKXJ/iwnKy3eQx+Wsa+ruTlM6YQuTd9he5hgrOVCS6CtWWGYVo70GzTMS7PoKVRWVJxN31nQpDSiQLz/zq7TyaJx7LTENOXWBr6GL9WbMZZJDcVxagWmsC40WexTVIaiMtw8STieZjhacXmlydmWHI7dyBU5fKONXkJ0LACOSxqvSBNVK+I4Fiyh0lIxedGiGT7JzWKcWRkW2o3Tr+O7XUqB4327gnjnMgZv6yF0dxUTdXDiziKQpepvwWgPnYyo4l5tGNJYJ5IqwvXFsl33KlSZYoIWOhO6NNZIr6LWVyhybNhCZVMC12cp8klKw/bc4flbQ57e6JCWp91ueSIEjuY4kfAZg8i6K2tp+WIac5RDYEp0maKURmVTYi9CpYn4FqyRxOhsIjqK+YkkjPdWMXXlYd2zAHQCB61qfmJbrMt745QXXtknnRc02gEPfeK72FprcjDOGExzHtts890PrqC14ldfustzt8cEkUcz8jiqUXUguLpTwxNeDVmxwla4uj/h4CShud7iQldk3vOi4ic/dJbAET1JN3S5djjjv7+qeGzjWMxfobAamr5AcYvKcuT2OTVMHCUVsStVWFpa2oHmMxe7BKqiVAKFOb2zGwBjmebSg2iebpS6UgTZeI6WK0bk6Xc0Med5Rav2bZya0aDGDtZ9ia+3SifE3biMioVEpazF1oeUrYOg5aFev/a1dwbXpyrufez7rxfAN0Txf6Olvl1BvHPZqhASdP+C9BDCjnjr05GoKfMpaBd3dEeuDlXti7BWNAu1u846HlVzZRHlLn4JtWA92FBETXp+Ij2CIkG5AS3Xpwp9nt7oMM8rdiYprbr81kq67f3QFQ9IIW/Q1chdhLdklWFkG/hG0XFKph1JbGqUmVx7ZscS2FNJNWR8CYjNzzyCzhOKsIM3O6RorOBg6JcnJGFf7qNKNvZjDy9zc39KUVQEkcdgmrPSDri02uB7Hl7lJClYjj1+6Il1rt0a0mn4JHnF+CQhiDz8wOF3Xtql0/B5bLPDie8wSsXoNM0rbhxMOb/WZLUd8NzNIaO0ZJwW/Osv3OLZR1f5zz91iVcPplw7mOJoxeNrLc61A6E+OYpBUrHWDGgHmr5X8bndnOuDGZ++2KfpywRo1YNUBezNSnxHEbmGTuDgmBwnP+TLeZ/XDqb8xAPCaUjdBrb0MDUhesm3dIOgpkzLuHLJt5xpNFlt+guSVOAqKiNXxg+daVJZy/UaUlPZGEcpVpsBbd8REZixuE9+D6bMwBpm0XJ9mMLBuGCYFnhac7EXsHR0i/Glv8Y4rzjj1cY/YDM+JYVnDNyeuIprtN17sXsrpXC+3YN451JRi2rrg5ROiGNy0C5V0MRO7nXGT0NSU7dB2BY0fduX1KlIW+mO1+AXGzTFsh22MVGHeWkXiUyn57yLITOKtLJMkoq0KklLORwudiOWYhdfy5Mv1hWjUmAskacXpGy5Bjj46QjjtOh44A7u0qr9HbZmbWLKxXTGLp3DRD2sH4sLspjjp2P0aZXUOYszPSQMGoysxxNrLdZbIS/sDFltB/iuw9WdEZv9iCvrbQ4mGevNgEeWYzqBw7WjhG47xHc146Sgyg3GM1SOZjpMqUrDdDmmuq/sHacFk7RkqenjaM32YM7NoxnzvOLGi9cAmH78AnfGGTuDOZHvsBR5NH15TmJeqQAAIABJREFUgmelIS0NrcAVHUM2YXdqeGVnxF+/vLSw5atsRtAIF+lWvqPpBv5inJjprqR3T4TXESw1oRJhnLJWrhtK43oB1gnJjThLA8ejHwZope47VOXp3w3kLx9EHllZUVm5KvYjoWkpJZbvsY5pNpqoKmeYSlVTWctxUjBKxWm61vTolwXjvOIkqViLXVT9vnRGu4vU+GC5j5PfS6v33sMBgVLfvmL8iVWVuIdv4vjSUzBRB6sdUaMZg3VDyuVLqHwuiVdBg366Dyk06pDd0oLbWsM9fLNuTDZRxZzEbwvEpPZWSIJXm9JqfEf8/63AWcyrW77DUuzy2uGc0NGsNISyFLoQehqlFMqU+NrF92WkZx2PXjWBXNSVOhnV1YJ87igN2RwVNSFPcLKZ0LHrQBZAhFRVjnd4naq1CsYQeYppAa8dTLl2d0wrdHn51ogsK3lyq8s0E+/FKCspjOHmKOP57QH9plQPjcDl7AN9JoOEsqhodkOatWmpMrbuPRiagcsHtrq8cGOAozOeudRfqBJ/9j/5QSoD/+vnbzKaF4KYM5abw4TShPzxnSFpaTie5lxaitnqRFzstdifHlAZuXqUBm7mDlutJVQ25VKvwcGsFLt3ZXEaK5jmClcKw4P9CJvuytftvmQqqxTqPhjbwbziF1/eZaMT0g4s42zO4SwnLw3nexFJYTiYZvzAw6tstDwebAOE3DiZLz6GsRaN4iSt+OM7IzbbIZ3QI3TFq6GRTX46Rh0kJWce/W46laUX1PaxeqxL/XPjN0gKQ53hg6N4T1MM1PvrivFNFRlKqZ9XSh0opV6977W+Uup3lVLX6//26teVUup/VEq9qZR6WSn1oXfzSSitqdrrGD+i7G8JOTqbiKnHVKJ7sEbGh45HpgVJf/qNKSVIjdwgSVB1Upf1YgFwUC68FWh3ATpRViYRpzhzrQR95mtF6Gh+7dVdZnUupKuVWMSVyGgXm9sa0VnU4BCVjtHpSEZ/Tm3yURobNmXurl10kci1KamDY/IUW0qUnzAnQqwjWLestOyNU84vNzie5sI8MJbBNKMZuJxpB9wYzMlKw/PbA77w5jFf/PJdBrOcZuhSZKX4UjyH2ThjOEzYHaZ1/0IoTKOkYLfWSrRCd6HIrIyoK9OyYpqWTNKSaVoyzysOZjm704zVRkA/klQvYyGt5KDqhh5XzoqvxanzJKaFIXVlIuAoed2t/9wgrUjr3o4JWtigwbxSlF4s8J6gSRF2SIMOqQoWDcF2nfkxySpagcuZVrB4+vuu6CjS0jC3kke62Q5Za8h1ZJIbRlnFOJMx7yApOJrlhIv0LU0n9GgGLi3fRSvFKBOVaVVLqqclDAtN1V6javQlaQypUmJPhGkd/90Ln5QCx3e+4Y+/7PVuKoh/CvzPSKT46fqHwO9Za/+xUuof1r/+z4C/ATxY//go8L/V//1Tl9Vu3RcIpOwu6/gzP6bqnBEBkSlJomVCm+FpGLstJqZBN3Bwam5h5CqceEkwcdrBeqHwHPyYzHXxKZlYTVOd0oNE0us5agFf1UquFSsNn594apPl2CWrLL5jidzTbpoE+pxmaqh8Kr6QunrIzzwiJefOVXS7T3V4V/Idl9brlLAOjPZQoRxyZjrE6SxRtFbFuVpKtdPxfUa+phuLL+Lm0Yx25HH1q0dsrDTk8zUSIzdKS57e7LDaCvgNrXjqQo9pWvD6G0e0ehF+6PLa81fx4w5FVuFHLv0axz5JS97en7K10uDSapOl2CfyXY6nOS/dGi7crPujFFMauq2AF7YHrLRD/v4zm0xysV6vNnzmRcVzbx3x6QdXON8NMfVIeS12ef1YxFabLR+lFK4WVWNSwetHCb3II3AVTtikKCzTvFwE72oFB7OMUVYSe4Kf2+iEPHWmRV4Zvrgz4tOXl+nHLl/ZndD0xeylkIPgjUHKpW7I2abAg46TiudvD0UsVlbMi4r5SExs37HZlAeAMdgoeAdR+3BeohTkniWIXW7WuakPL8f4OkIpRSdQhIEI95yT26gieRfbrF5Kva8qiG96QFhrn6tDQO9fPwx8V/3zfwb8W+SA+GHgF6ygnr6glOoqpdbrxOE/7S9BJWNJLGqtUHU3pEmZjEjDHn4kQbuBI+DQvLI0PdHsZ6XFd2Sc1As93CoVT349ArUujHLpYHcCV4xXtXza14qq1jkczSsCR9P0FLGu6IYOS5FoHq4dzfnkVpt5HYTr+zGnTWsvn2Kjnoh+oJ6aiLrSafUwboizcnZBGsKUcqC4vjhBs7Eg6fwGJmjKtcoaSr8pocKOpuE7VAbWuxHbhzMcR7NUk6ZuDmc8e6FPWhnZCPsTntzqsjdMGM4LVs+2adadfT/u0Fpq8tQjK/j1SNB3NavtgPVuxM5gzsE45Vx9cFTG8lv/8rO01y/wzLPnyJICayy0Am4czNgZJLx2occoLXllZ8iDS2fpRaK32GyHtH2Hr+xNiT2H9aZP4CqKHK4eihnLWHgzkpzQlYZPKxDx1q9cO2Q59vjIZvv07YFBfCknScFrBxmtwOFcJyKtJOj3I+e67E4zdibpO7wPsffOHkBlQdX/uzSWtKzu+3+nEsiakO64VOW9t6nvqIVLNC0tN8dClFpt+KivI5csLajG0qKR+W7WX5Ux59rpprfW7iql6sExG8Dt+37fTv3anzgglFI/Dfw0wNb6GsXtN3DXL4guQGU4xzfBDwmtQc9PqHqbIrd2PKbW4ea4ohc6NDxNo5wQNNvkxgp9uhQJL9pF1VLeykiaVGlE39+PHBrJEVN/iVEmOofjpKAfuoxK6TnIvN3yya0246zirZOUR5YjrBV1nygCJ7X1e4atcpSpcAfb2NkY64fYpsBlsIbq7tugNbrVQ8VNkZBXOaoedRoLk0ITuA0cC7vTgpf2JhxN5brw/LUDiqKivRxxPM3pxD7TtOQkLbg1SNjoRhyMMxyd8+adMXlaoB1NlrkorVg936fRDunUGonKWJK8ohP7bHQjpqmYv6Z5ydWdEfsHU2aHtwg7y2z2Iw4nGZHn8MylPm/sTZimJYOk4HheMJwX7E0zAkczSUt2xikQMi9EczHJzSKXQkJ6FVlZ1Q1A8B0JGFbAV26d8NhmRw7w+yaTYuiSct3Vin7kce1wRlZKH2V3kpGWFR8800brewj6vLCM0oJx7i1Gn6fj0ZYvY9oDm3NluSHCsRoCk1gH37G4Wg5TTytiV5FUMoo9SSuKBa5PDjFtJaz39O/JdfTeOPZK4byHK8lf9PrzblJ+vdro6w6f60TinwP48JOPWe/S49JzyKY1lKS5CM5RpkRPDqX8NiUrJKwErmScVPWb3dZk5mKOKrN7gFnHxcvGeNolcFwq1xf2YJVSNFbwakhK6EiGZWVl45w2JCPXMq8sb52kbA8TLvVCLNDS8s8yYQedikJPFdk9qXezi52PsduvYBttzGwsEuvTFPB0hvabqHSKLXMcJfi7QGmq1iqpI0/x5djnN16RiL/pMEVphR/Jty32HT641eMrt4YcjDO+eGNAkhTMxxmD/Snp+AhTFkTdNaJWwN/74UeZ5xUv3xq+43tRGctWP2KzH1MZy94447WX9xi8/Tpbz3yKMxd6fPRCn81+TCdw+di5Lq+faXEwy+lH4s3wXc3vXN0nLw1JXrE3TDjTjfj7z2xiazfsWwPJBn1m1WO3HTArBBuolMLaexmYndhb5ImevuZoWIrk9U4gaDpXw//xh9vs70/4yGNrrLRDWqHLA/0IV4tF/+YoZXeSMS8qvrw7JqgVllklwc+X+7HAZ45mfPxcm/D4BibtUDWW2B3lXOr46OlhnRSWobMJnhvQaK9D2KRWgS+WQfowlZWtcNoPedcbSIFy/t0/IPZPrw5KqXXgoH59Bzh33+/bBO5+04+Wp6Bdks45wvRE3IymlAaf1tjZEKfMJL/TcRnG67RcCdJVdTJ44KbS+QZpCtaAWVWVgonzY/FK+AK/tY6o8AJH4WpHsjWQvI2oPhx8JU8ZVyseWY641AupDLw1ytjqSLqX70RErVB6JzU7QKcTVJmiYuFaKtdHd89gD7YXUFrlh1J9+CGkYkQ77YhbP8ap31NZadjsR2z2YmLf4ebRnI1+xGheMEnlh1QNik7skSQF6xttrjy4xGh+lqSo2HnjmPloznOvS77DZJajtLgGI89ZjCmH84LYd3jsTIu//f0PszvcIvYdOrFfm57Uok9grEwBVhsC7VnvhlzfE43Eg2eaC8MUSGk+K1ikslvHJ3AqjNUL63VSWTYbHr6G/+JTlxZA3cUTxyiWo3uld+CKeOv8csxkJtUUQJJXHCdlbRGHtabPWlO+168dzjhJCmLPoem7rDcDNloelbU8sNTgtaOETnCOQCnsrCRyRT+jTAlKM+td5HfeOuFHlic4+2/Q2Hx6oULNSrtousK9Q+M04/NdLwWO9/45IL7Vz+RfA3+3/vnfBX7tvtf/Tj3NeBYYfdP+A4C1GC8iq8y9sVFdNWAMSjsSRFMKzjx2NbnVmObKYhyqslnd4MzeMXo6hdFWysUGTZya53Cqq0hKwySvyMp73+TQUfXkI0cVKb5mkc9xdyICoEFaMa8j5CpqPoIfoYqUqrV6L2ej0QVPRpo6bolBS2tUW7IvyFPJznB8quayTGxMiWtLAkdQcJu9+B1qvKWmXC0urzR5bL3NajugG3v4rmZjpUEn9lhtBzy51eXJcx280MGUOceDubAgKkNV1j/ue/zlpZjElmOfR8+0+OilPpdq09c0LdFKYLcaQeentWnN0xrfdYTgrBWr7ZBO7OPXAoivjZ5LSru4Otyd5ExzI70km6OTEY35AQ2TiJaBe0pEVysCfWraEuLVejei2woWLIisNIwyMXyNU4kXbHiaZp2l4dSE68DV9CIXX0kVsxx73Bql9SFSiSHva+aTo8zwr168g/VCzHS4aKCe6imMlQ2lUQs+hFO/b97tUii0o7/hj7/s9U0rCKXULyENyWWl1A7wXwH/GPi/lFI/BdwC/nb9238T+H7gTWAO/L1380nYIkUd3aITdXFObmIG++K/z1LobVCdHAi5J52KECWbUa48gEolZdsGbdFKeDG5E9HIx+hsImj6Wl/gT3ZlNFrleGELWzMsReVmGNRZEauRKzDcOv9BVQWFrQNftWWr49MMHNYa7gLb7mlL5NZR8K5P6jaI3RGqLOQgSqaCLWOGqd2qZrCHXhJ5tZmNMe0zMskJmuS4DFPprO/PMl6+PSQrDZ//wm06SzG/tzfhiSsrDJOCvDS8fGOAKQ2u71BkJZNhypVHVtnsR+wOU1Y2OyydabG50qAVuvSbAZUxjOo08M1+zMfOdXmr4ePVIqJ/8oWbvLF9QqMT8thGmx/7wFm6oYwKk9JwoRuxHHv83Oe3AYHJRr7DdzywxN+8slLrJWA5chhlhlf2J2SV4fWjmfgttCJwNJ+53GOaG17cnfDPv3TEziDh2QeWWG8lPLvpsxq6C9Pb7VmL40RES3kl8uf/8IPrPLXR4X/4vTfJshLtytTna9WL07Tkkxf7XDzXXbzmKNidG8Cw2Q640A2xVg40+RwF2W8CkWBHWvHjT29imh28lXOEg22CWqmLPs3NAOO1xAIAMunKct710v+OuTmttT/xDf7Xp7/O77XAf/yePwulJUdifgJ5JlAVLUg5J51gjJExoRuIRmD3OjruyZUhz8A3pGEPT0kITBS2cIo51pU8A5XNBDCbHYrjssgWsBe0T+RqZnlBy9cLpL7v11CYXEZUp283p34i5ZXl9iinspaNtk9lNS3XBTckrJIFUAbHx3bX0fkM21p5Z8lW5Zi5QE+UNSIfz6ZMdRuD5Ut3R/yzP9wmSwqUVnRXYm5+5RqdjfOM5gWv7kjvw3E0VSEciLS0jPeP2FtpLP6adiTE6GlaLriWp0/bylgOxykv7UlgsqMMu5N0ka+xUj+d92c5aWnohC6OUkzzknlhePBMiySvuHk049Jqk41WiKeVaAWMZVaohWahF8rE4muTr0AmFJHvstT0WWsGrLdCOoHzDjGZtfI5e1pTWQlVbvqSjnWqD1H30a0lxVuakLvDhMrKU10pmYyIKdcuAnuslV5HXkmF4yCEalVXpJGreXSlIWTu01Q2i4xFTvNMlLpvYmXeIfZ6l5sBpd8/V4z3h5JSaYrbbyx+6fRWMaNjVNwif+F3CB7/GGY2FnrRbCiQlcFtlOOA58N8SGzewEQd+tagJpnoDOK2+B4AnZzIWLHuUJ/CXozn41NyoeszL4SW3AmchQgKpfHy6UK45TsRoBeGrVcPJkzziAf6sZi2wg4OIurC8cVHYErIJpK+pB1B5IcN8YMsCy3a1toPPLlOTPOK5djnE4+s8n//3puks4J0PFxMFfLS0Apdovpp42jFNC15e5yx/+ofUKQzZg9c4tyFHivtgMDVvLp9guMIug6kOZmXhle3T3juj27zqY9t8cRmh3/zxuFi9LnRjUjLis/dOGa1HYqHIXQ5SQqMtfytR1Y5mpf84jTjex9c4UzLw1jLJJeD5lwnJCkMnqN5ar1JP3QoDOzNRJWZFCI8MhY+cLZNM3B5fKVB5Cma6ZEcsG4g18T7poW3Ril/8OYRn3l4lbQSzqSu+yqnAUIAHzjTYpJXvPD2YBGsc7r9pI8CnrYL8pjhHpymqlPAtZaLg5+OeNCMUaNMhHh+JAZAa+8ZyupqwtRENJULm+RdbwWlcPz3x7aE98kBYYocMxnibV7GTIbkb3wZ5Yc4WoPrUdy8hi0KnM4S1lR4Zy/KVOC+qDt1Cgo1BlXMJQ7PGAm1iduLPgBVLgzJDLFeG0tASei6OMohqwx+HaNmHRdV6xysF6PTEVErxNcaT0vlMM0j3hokrLdOk6c0vqNxow4497IYTNASXqV2RGUJNVwlo2ouL/gXqipohppZodmto/SKtGI2OGB+fJeod4aoJRqP3WHKk1td/uirR1hrmY0zpsOUsLvGyY2vUGUJUfMpDg8FarKx3l70BU5HnElWEsc+m0+c4WCc8dW9CZv9GN8RnclqU+L6HlttErjiwTic5QtexO40Z14Yntnq0Q4d9iYFKw2P1w+n7E4zznVCisry9smcUVrQizxCV3O5F9ENHCa59HG2OiFNX+Tdz98e0o88Hl/tk5aWPJfeUGEs3dBjteHiOYpfneYczHK0gksrDc4vN2iGLld3Rgu7tXhHXFbaIfOiYpTKKdMMHJYjd8GMrEyd5IVizcuwjiv5o8ZS6VCUn46PBrKwwziTHtRGX5CIKhmK1LqOUVBVLqzVumH9rpdSaO99sS2B98kBoVBSomkHW+YyDkRYld7Zi+Tb13DXtmpa9J6MClOZcLjrFzHzCXZ0jN54SFyb42P5uDUJCNddQERxfHSeYHxAaRqegGd1PicwJSPbwDgtetWktolLNaBPQ3LyOY4XErkuldU80I9Zb4VcPZiwHPtc7IbEniZ0PeJIGBSYUpqYde6GrXKoKmyjXwurCkwgyk/rhQtdRidwaYUuvbUGcIalc0JkqkrD9vUjXM/hYDDn/JkWO4cz5qOMIi3woib5dIAp5e5rjMUaGCUFnchjvRsuZNNZUuC4ms1+xM2jOTuDZNFobPguWc190LVu4XBm2R4mNeBFEzoSHhx7IuZaaQilS1B5OVWdwdkNPQkxqgxnmgGFESu4q6WsLyoB3WDgt6/u88jZNuc6IcfzQuzpSHSBqxWTTBqk692QSS6Rgv2mz0YnJPYcnpvmiyopyStW2wEPrrWYFxVHczncOqFLT6VkQcwoqzhOJULAd7hHRXc0eWXIrcV3RH5v6/7HtDDcneS0AodO0JBGuhcyKjUdVy9yNMrlS+91M0hl/C0updTPAz8IHFhrH/+WP1C93hcHhCmrRWzd6cqOjmmsbUmDsrMklYJ20JFUDbaqJCT38I5IlqlFS9oB7UiQbjqTOLs8RbVFjORMj6jivvQv7mNcmrCNLlOxbHtAnguZqn4i2HoOrqpCHJrW0KqrDt+Rzv/BLKflu5xt+fXsW9VlqCPVAyIrP0Xqcf/H92Ns7RMpTElSWDxHM88rqsqQJTll4VDmOX4cUuYVT31kbbEBXn3tgLAhB6vjR6xceZalzTUub0lTLi8Nr76yx3ylwYcu9GimBUleMagMeVoymhesd0OOpzl/8NIuj15e4vxyg8NpthjdXejHjNKSX/nSDllS4HoO/92PPsEkr7h+PKcXecxyxSsHEzkUYo9ZLh6Jj53rMkpl/Him6fHVo4T9WcZ3nu8yygy3RgnNVN6Or10/Ji8NP/bYWk2zlonCqVPzX7y0y1LD5wcePcOLd0ccz3OWmgHzOg39/h7EpdUmW/2IXujh1YdZYQwNT+PsvEx45mFGql1rNAKWIo+xH6MsODVOwNa2fuUodDahARC1OJwVCwTekY0YTSquH0/5jnNt2gBa82/eHnHt8D4s3TdZSimcP1sF8U/5k9aIb3m9Lw4IJ3BxOkuoenM7nSXc+Zxy/xbleIR/7hI2S/EeeBIzH2NnE2nuGTlYlB+i6jAam9Ybsb56oDX0N2B6LDmfjido/VAIzFQlqswwVppRHafEHdyVjZvPFx10ZSrZ1KeiLdcHNyQIO4DmYjek5bsM0oJRVvBAPyYprHThTYIJO5j2Os5kH50n9Xi2kgOjzHGmh2JYqw1a80KmGK3QJYg8zlzoMTycYYyHKQ1B5PL6tQMcR7N3d4wxlitXVrh9d8zOl++QnOxhTMVbTZ/xSUJRZ9uPjuf8P5+/SVVa0nnObJRRzMdce+6Ihz7+YYLI4+jOhMOVJuvdiO97eIUX7oz4n37hRX7yx5/kyfU2//5HzvFLX7jFdJjINaThsXS+x+1xyp1xyku3h/zRK/tkScETP/UM+7OMtwZzHlpqcKkfETiKrU7AatPnxd0pb53M+eXP30Jrhes7/PT3P0zTd3jlYEZRSbK35wg2DuBcN2J3nPLf/vZXcbRUD4+cbeO74qH8r3/gEZSCaVbx2uGU3XHKmwdT/s6Hz7HakLd86KgFYqBpLJ+62KXlAqbks3emNH2Xi91AMkGyKVSGuW7x6rzL5SBkySR8VO9wfX6R549nfO/5mJVyxOWzAWNgt/SxFj6xFfA9F1r8p+9hP/xZhFLfwBrxLa/3xQFRzjOq0THuxmXRCQBBs8v0pRewlYHbN9Ceiy1zihtXpVT3QynrTIXzwe+WmLdrzwOgo8YiZbu8u43fXcW0VyXxylTQ7Ev+hN9YXB+8mgQ17WzRqvkNOhXjlSpz3ME21HfKqrUqXIoqwUFK0djTnG35jLKC7zzX5DiTe22OS+BHYAzueFcapHFPaFhVjjKeVDaTAylL3RilhCz12t6EvKwWSsOw4dPqR0yHKemsQGuFFzgkU7lK+K5GaUWZTjFlQXLcZDZeZ3o8Jp+PaK2epcoNg/0Z1liKNCOfj0hP9pnubzMfP47nOVRlVQudDL3Iw3c0o9vXSPJHayWjSzvySCYZxloUmsiVScLxLGc4LzjZn5JNRxgrh93rd8esNQPKWtsQuBrXsRzNc3YGc4aHM4qsJGwI3q4wlhuDOZW1NRavWuSrRp70jnZvDeksx3Rib2FD9zScacqkZRY4vD3UTFIZ58aePABQmkq5jEsfVdbfP5Whh4dgDbuTmF5k2OoEtWBvJs1tt8WLu2POtgIUM8rrL1I8eoG3BnP0SoZzfEve0BsfZpzJYbZhT3COtt/1XlDfvAexrJT64n2//rlalfwXspR9DzLQv6j1gXNr9rd/5sdoP7iF01ul3LuFbnWxeYp/6XGSl7+AqQz+0rL0HZbWKfdvUU0nON2+JGmfHMqfqSqU52MmJzI6bbShvYwd3EW5PtXxLipsCCC3RtOb5gqzoEtcJcydiOZsTyoLU8nYSmnU8W1UfXhZLxYAbd3XMFGHufVqfJshdBXXB4JbP9cJFlmhjWqGMhWqmN/repfi37B+RNVYYmpcDuYlx/OC3UnGq3sTfvP5m2RpweHb2wzefJF45Rz9i08StXxMaeivtzCl4Uv/6hcBCLtrpMN9AB7/wR+nv9akWcNvB7Ocg9sjyqIinRUkw32CZp/V832+76NbbHRD3j6asVTnW764fcJ6N+Q7Li7Vxid5uq02fXxHsTPKeO1wyi9+9gb/6EceZzn2+Jcv3eVHnzjLWtNjUsfRFZXlZ3/rdfb2JzxwvscPPbnOMxtt/smX7jCa5+wOU9a7IZHv0Ao9Li3FfPhsm6sHMw5mGZ7WC+GRq9VCEHVlWaZUL+9N3iH6inyHfuRxdXdMK/T4+IU+T67GxKoAJNxnmFUo5MA6SSvONFwanqAD5qXlJK3YaGipIk0Jjk8adGjuvkQ1OsZuPSFAW61xD9+ELAE/4HDpcQ7mJQp42D3BOb6F++h3fsla+/Q33Qsbq/a3fuZHv+H/3/gv//dv+nHqCuLX/8r0IKgM1hiSuweUN3ao0oygO0b7Ljp8G+25OLEv15AwJr9xlTLN8ZeWsfMJxfbrIl1udSWwZnSMmU3QgDEV+r6KQ7d64p5M76HpdTIiinuoNKkxcS7KGowXonau4rR6Yryaj1FxW64HZbHQOeD4xJEPtVS7MtRGIZhkhtjTGGtQfoNIVcKurMndusgEkOOJ9+JUNTgvKnR99z/eHZMM90lOxJMxP7zNyoMfpMgqvMAhavjvUP0FrT7FfITf6jMfZbieJos8vMAlSwqqulR3fYeg2Ue7Hsk0Z1ADXV+/O+F7HotZbfjkmx2agWgJTicAWovkWpyNhthzuLTZIXR0HQ2hGGUFDd/haJajtWKalTRDl04vZr0bMspKbo4yDuuUr82+/Pt91+FCP6blu9ydiM6kF0pv5TTg9zQla5AUTPJK1KadkIO6ejltUBa+KxJ1zyErK/ZmxcKLUZpiQbY6pT5ZqEVy8v17BwlKu5JsbixVaxUVdmSEXh+Yp4G+Vju4GkHNKUWNe39tAAAgAElEQVQereE2lt79XlB8e8z5tUv7LuFSG6U10WoPW1VMbh2Q3ZmQHo9pnVvF6bRw1s5R7m6THA7Rnkt+fITf64l02fOwRY5NZuhGW9K68hTd6lEe3kGHDQnSRfoTrD9IFUrTEmvQ00NxZNYGL+uKv0K3+xhXXJl2+xWR0p4i8E51DiAqSD9iNXTJcTEETDLDV49nXFluoJVmkhkqT9OMe+g8kWtG7Tq1rk+OUKyTwnD1YCqJWWnB+M4bZJMB2vXRrocpC3a++LssP/QM7YtnmU9Egt5/4ENo1yPurnDx6Q9QZBW3vvwiJztN3KhBe2UFpRXWWLSrabQDuisx02HK7a+8wEsrDYLI48Yr+/zNp87ywfUWz2y0meSG33/7uJ5KiMbAUYpO4FJUlkv9mI//9YeYF0Y8G1rz3I0Bvqs52wk5nglX4kMXepxth6w2fEZZyfbJnBsHM84vxzx7oc+rexOSvOR7L/fYm5X8L3/4Np++ssqFXkToOgvmpKNhf5rzxVsnHIxTHltv853ne3z25gkH44zYd/BdTeBqnloXy/gX7474zdf2Gc6lgghcvdCQPL7R4fseWMJRwgU5Laq7oQMIMQxrKJyQnWGOq5cIAk1baYLTyqZ/HhDdRKtKaTmij/h/t+HzN0/e9V54F1eMv9T1/vhMLFRpTmNrA5unjLf3mO0eEy61qfKCYp6i4xn5my8z375FlZfEm5KUDWAmAlxxltZFcBU10HGLcndbkqx6q6hGC0yFCmOqwzu4rV6tshyLUSy6SGQNzuxYNm0teqkO7+KsnJXcjUYbVVO0SabY7vo9r78pZVphEgI/wtMesae5stzgTEOk0xUWpTSNyMd6Bj1JpAKpcqAhnXKgH7mstwJ2Rim7w5TOuSscv/Vlot4Z4uUNBm++SGNlC+36eIGziGpbvXSR+Tjj8PqLzIfnsKZiuv82a49/kuZSl2SSEbUCljfaGGMxpaGqDJ2lmNXv+266rUCe8h/ZZJpXvLw/XTxF+5H4MYbzHN91uDVKFt4GrcWG/cW7Y17bmzBNi3rzabY6IRe6EY+vtXh1f8JXdob4rsPhWL53P/bMJgDXj2aLBLA/vjvFUfCh8z1iz8EYSMuKyHWosGirFq5bR6s6e8LSDly2+hEfWGtTGMMoLQVAoxRXlhvcPkk4rvs1Zf1nT68lVU34dpRa4OIWy3GhHhlLtqvG15IBWhoZgfp7rwn8JxbNjHU8rJKJ1HtaSv2ZmpRfzxphrf0/v9WP9744IExVYSojGZXJjCovyCdzmhsrNNb75OM505t3aZ47gxP6uA25LpjKyJRj/RzK9cnf+DK60cYkM8xkiG51qU4OqY73JPeg2ZXYO+1Qbl/DWTuH7ZxBlSlRcoQNWhhr5M6ZzQUT5zginbUGMxuju/L7aS3L6DKbCCLNC+vpSEd4kq5cK7TScjjYe2zCpLREboiKOiLbdQMq5WKsQHDvTAueWBM7dSt08QMXU+Sko0O0dlBak00GNNcuyNfv1DUZuCQ6Jzm+u4iKs8agXR/XcyjTKWXo4TgapeSAKLIKr+3yyLkOw3mB72rOLzcYJYVcCwJ3UYJP6tHoqWozLw0N3138/UfznN1hQit0FyrNpi8OTkcrvnRnxMFYxF+DeqP+6AdDBknBi4M587rkv7o/oR269EIPc0r6QpFXwpYwVjbqQvFohfVhrMT1LccicEpLg0K4EE3fXVQWwDt+7mlVf3yLW+PxDGBrWbrVLsqtVZXq9FA5dW1ajFWYwR62yNEbsXh+6rhHrUri9+CtkDGn9943Ub3+FGvEt7TeFwcExjJ66w7p8YgqzcnHc2xlKeYJVS4ncHJwwuitOwB0Lm8w3xvghD5+K8b54Hej9m9Q7t+CRhvKgnTnFn67gW520Y02s69eJV46gxkeosJY+hVhjNm9jnU9CbXJJlCVkg0aNdFFgl1aBxDYi+thD7ZRcQtF7a1IR+KzqLX4pr2OO96lEUjPYZKZunKAyNW8dZJyuRfiKE3QWhPhVK1/aBcjrA1IS1e4j+NMOveuYuuZT/DWZ38dayq069PdegQ/Drnzxl36Z5cJGx5v/MG/XTQn3SCie/4xXN9hvLfDdH+bbDLg8PUR238oZbbSmsbKFu2Nh4ga/mKj39id8PYrOyQn+/yDn/l+bh7N+NznbjG68zb9rcv85I88yicu9OlFrgBt5jm/9vohqw2fv/WBs1TW8se3hgym8vnvTjM+d+OYzX7Mpx5eoRN6zAuZSlw/nvPy7SHPPX+Ln/j3rvDIapOf/ecv4nkOZ7e6/Ed/7QKd0MXTit9+84jX747ZGSQ8fanPP3h2i999a8Bnv3rIz//udcrC4AUOyz/8GM3ApR/JRtufFfzm6wc8vt7i2a1e3V/Qi6Cdlq9ZDS1vT2Sqcqkn8vC3ThIe6Me0AyGbFnlFL3JoesK6HKUVW84UfXKIKXKclbPcic8zziuKRL6W33uxzd9Yq/hH72E7/FXgQfy5riovqNKc9FiMS+FSG68Rkg2n3H3+FdoXztDcWGG8vUvQbZKPZ4S9FsUsFWn0H/wyZj7H7a9I1XC8h9/vigoyCDHJjKDfxWYpypOMRTMZYpKZCK+S2UI0pazFLp2DPKk1CmndJ9Ci9jw9CKZD0VXU2o3Tje5M9uXPmYpIVVSeXqDR3zpJhTodreFpRVJaup6Lyuf4fowJGqBd5nnGrVHC4Tjl+q0hh29+Fa/RxpQFSmusqSjSGaHpUuYJo6MZ6cynmI8WX9PkZI9o6SwmbGKKHGsqglYfU+aYUg4IawxFMqXKE6rSYI3F9R3arYDmUhsvjFhtBVTGsrTexI8epL/WpF1vcJCnalYZDiYZnVCky0VlSPKSyHf40t3Roow/nuaErsNqsxYsVdLg9F0tcYGDOUsNH89z8AKXTuxR1Ei5oq5IJmnJSjugFboiJquriCwtBYcHNZ9BPj/P0SLRjj1CV57kntZ4jqpBxPL1KpWLVsKKSGs7+oIUdcp20IqGo2h6YtTyY3eR26G2HqNsLJFmAqLJSoPWAlT23ODdb4Zv9yD+5FKOZnJ3StD26T18FlsZ8vGc5HhMlRtpSjoaJ/Qxlfy6mMzRnovfbjDf2cONQ/KbN7E33qYqSvLxjM7FdUx6G6fbl6ohalDVEw5MJdcN18NZOUfpRYvEKxP1cLIZOh0J5Sqfols9aW4CJp2JGhLEeGUqUVr6oci44x6qmKPLlGbcoxH5JKXlci+kH63x1mDO/tSpxVUebn2AWMeXN6bvELiaZuiSTHMmu2+ia/uwrfMswvYyjqNp9FcZ794icRzipQ1mh7cwZUGVp5zceAk3bBJ2VvBbPTprG4wPexy9/keLr70pc6oyxxpLVRmWopBPXllhsx8xzyseXm7w4FLMlbWmYOwDl1YdszcvZIIwTkuu701Yakqy1iQvieqS/hd+43VavYgfenaLl+qq4umNNvPC4WCa8+BSzHC1ietprt0asjNI6K+3ONuL+MRDy4Suw+Espxk4DOcFjlZ8+soqnlZcPZwyLwzdGqF3ekDMi4qiMkzzjKLO6XxgqUHg6oXrM7C69puIe3O/BsRErmR9WCxrTX8Rv6iAhqeJ0gFkgkF0Dm5SPv4ZDquIY1OSnRi0vtdzcJRib1rUBr93uRf+ihCl/nyXUngNj9Zmf/GSE/k4nkuZpATdJtlwSrTSo5gn+K2YzuUN/G6bbDAkO5kSn1nCXduiPN6lGI1xPJcqL3FDf9GPMMkMZ+kM5e62HA4AZQHZHHd0d5F4Zf1YXIRVDqM9cH1U3MTWB4NqL2EGMnJUYUOYelVVswwdmU5oLWadPMF6hsiVa4WnFftTZ8FnnBaGXi3fVkWK8kIanjAZIt8ljD3ipQ3KLKGsVaLNtYtYIzkXg/0pUW+NKk9IR4eL6sDxQ/xWn2I2IuqtEbWajA/3sKaivfGQ+FrKnDKZ4rg+q72o5lN65KVZoONe2Z9wME753LVD/NDl6Ys9fuARqYDy0vDCjQF7o5Tj3QmR77DZjxnNc54612U59ln+0cclncx3WKqj+V7dF+nxLC+5PUq5vjdhcnjMj3zmQT52vsd/8yuvkDR8VhsBX74zYjTPaYYe652QjW7EZ984XBC0Xr41IslKtra67O1PqErLV3aGNbxGE3kC3Tma5wyTYtHYXGn4fPhsh622j2tLJpVeYPd7obOoLHbGOUfzHK1ETn+hK+/RKOrgWYOTjFgNG8SeR+0e53BeohWcaXi8cZyyO83e0174dgXxNUs7wvz3WjFVmqMcjdIatxGSz3KKWUq01EY5GjcMUHGEcjROb4UAGL5xm/aDW2Q3r1OmOUG3hRv6FLMU5WiSwyHNesSptIO3cRlnZQMzPkY3u5KyPT7G6a2KcjKfy9UiS1BhjA3aApj1m2L7dkOBvVSSZSETjn6tjKzk9bCF9STsVU8SVNQhaK2RlJaLXak+AldzMCvphRqVjMRIlo4o1WptfjJErYD2xkMU6ZSg1Wd856tMdt+ic+4KQcPj0hNr3HnzmOPtuyxf/gC3Du8xg7V2KGrehONrhtuv4oZNehcex5Q56eiIfDLAmGpBgzpdB+OMwUjES7vDlJODKY6jednRfGCjw+Es52CcSiMycPE3OwSuZpoKwHZeGNJKmpRFZTieS9PzlENxev9/42AiyDw/YrUdiL4ikCbnKC24vjdh7yTB9xy+5/E1zrbFLyIULZ/hMMFUhisbbSLfIcmrheEs9h1Yby2Svg/GKaO6ERt5DoGr8KoUqpKkukfIDh1pVEYUXE1LtocJQX1VWY5dweR5Hrq1hok6ZJUoOJ0alDtIxDr+tSStd7WUkknZ+2S9Lw4IU5asf/Rh0TaMxQXnRj7F5F4KUpnmaM8jH88Iuq1Fw9JfWiZc6pAPhpTzlOGbd2htrRGviupReT5Btykkp7hFdXIgk45pzZUwFWw8jMoTjHZFn3CqdAwi8WFkY6FPp1OsH6LmAn018zFquSmuz9N8UO2ijAdljj6F2DqC7VdlStdzafke08JwMCvZm2ac73QIvEAOCWNoRppRvZkAinTKeOcNsvGR/JubPcKGR1VU7N8acvfVF0hP9sTLUi/t+pSZjA3z+QjTaVKmM8p0hnY/dK+pWuZUuWyc0/yLhu+ytz9hsD/DeXiZVujS6kWk84KTwZzfvrrPtBZNfeqRVTbaIY+tNnjhjow5K5Pxxe0BAHcG9zIhBocz/Mjl6YeWWW2Hi2kIwJmLS+wM5nha8/iFHpWx/P5XD9kfSVxgDuwOZeN/8soKqw2fD55p89zrhwyOZ/SbAZ98cJmW7/IvXri9qBTWGuL5ALh9kiz+nU3fYaPpYXGYWJ9r+1MC1yGsD4KV2KU13iH21ikqi1b3NvvpeNNEHa4epeyMUz660SZy4RSyOUgKbg0TrizH/P/svWmMZel53/d7z37OPXeve2uvrqlepnt6Vs6QHJIaUpRIyZZkKYpkyUhgWEYS50OcIEhgRDAQwB+VfDBgAU6MOHGgBHISK/EiW7IpiTLDZThDUrP1THdPb9O1L7fqrmdf8+E9dWdGpKQZyVCGAF+gUT23q2pu3brnPc/7PP//77/e/DAXvHifafH/7/WR2CA0y2R8b5/2Fcm7jc4msnqYBmRRRpkXLDx5CfPJz5Dt32f86uvUNxYJjobzXsTk/j5Go0Z45mF1G4RnCkbdIR6OcZ94VsJvsxSl1qgs4DpU6shS0WQmh90ktZoYkbzrkiZyI3HqCEV6QYikEKvwq88pC3J3AVGV9iJL5q7Rwm7KEWmeSBqWIhuSmlBo5yltS+FCs0lWlExSnUXNqpKgSrw452PrLe4eedQ6C5hum8HtlymLnPbm4ximhqIqmJbOhWc/JS/e7buohkWeRKTBlCzy0Kwaw3uvMLz3CnZ3Bd1yCcfHUlDVXaG5ehGhCB7cGtBZdplNI77x9W2CsQTcfu11KWXOs5JWr0aW5Lz44g71jo3l6PxeUXJ5SfIpjr2YMJEbW9eV4TtHY0mnsg2V7WGIYWv0G5aM9hOC5y60SddLTjbbbLRtmaTuJSy3LH7ssUVGFWJOVwUvvzPk9Z0xf/mZVVxTY2cS8lNPr6Aq8FjP5diXx4HPXF7A0VVWGya9KmqwaWo82ndZalioiqSFT+IcRYgqY1VHqVLDT4OUOCuYmesEk5iluokqpN07SIu5ArNpqiy7OhtNAz8t8NPqLUHJgqPTsXWWXR09erd5/CcuIRD6DyqI9y3FNGhdkqwDpZKZ5lGC3W8x3h5jNBzZFKwvoF90yL/5LRIgOBnRvrJOnmToNYvgZERjY4H2lXXisTznmp0W2WAfYVhoK5uI7hoiDSrWYA2hS6u4fGwF3R+gTI7mbAq12a0al03USkFZqAZFY0nKpbOkAtSqkkbtDVBnJzK8V7el65MauZCjTMOQ/z9UDRFOMHWTSarLOX69iTqastrWyEtZbl9dqbNXeSeUx56nKEpqTRPD1mi1bDprBjvHHratkyU5/mCH7pXnOLvzHcqiwOmuMt2XtK5z3UQ8OUWvNVANi3rHJokzxofHuC2LLM05fOOrmPUOutNkeOyjmyqmrVMzNXzAOz1EN9dQVYXjE4+8KFlu2fMjB0jJ9LnW4Pz4kudSmPXeo0zb1omzAqtiTujvIV6t1S06lj43bH1HEUyCFFNTKIqSWZKz4OjSpu3oHM5i4lwazNq2zkpdbgZFKTM1XONdNaajq4RZiapUJi9VzGXucZYTZzmngdQ91N+jYwjS4l3JeeVkNbOAUa6T5OV86qEKgaEiw6g/FDAGUL6PmJR/HisLQrIwQTE03Asr1B9ZJz49I5kFrDy/RfepRxFWjfzB6+SjAQuf/zxkKfWtQ3LPw15eJDoZIFSV6GzC8PY2C888hjAsyixFbffki14UlIMdRGdJlv1JSGnWEFlCtnAJgLTWQwknMuzGaZHW+xSmK0UykUydzt0FCYKpGprnMX+lopE3lhFJIPM6kW7OrCgpyoJGOqEwa5JXmUbyOYQTFjWLot7kteOQpxcWUWbHdKweqhBc7Ll8WREURYnbshCKzNlUFMHpwCPJHUIvIZjGTHZvU+tt4J/s4i5uYtQ72K1Fiiwh8Sek51VPtYosJYkzwlnC7OA+htNEMzTqK5forG/guCbDYw/HNXjsygL7ZwFPXmix/tOf5N7ehNhPydKCw/0p/3h/il03aDUsllsW26c+eVHScQ0mQcrDvQndZRfd1LhTTTwMTcWLM75x55TxNOLCUp2/+ol1rq7UZQ8kSimKUnpSsoJfeHoVVUjehGuorDUsFCE49GJ+/94pz19oc7lb49CLsVRlvjkoAuxqxClPbeU8q0ICbd6vpITznBX5OikVCXt/FvGvbh7xV55ZZdk1CLOCMCvIVZuskGAbXRWSH1GxJIapSqkvfOBrQQghK9yPyPpIbBCqrmE0HOz1NenUHOwjVAWz5aKoClpvVY4Q16+hTk8ogxllnqMBal3uzuZCjr2xSXYqpwtqb7WSVtdQm10wTMookLCZynhDrsoxZp5yVpg0TRWVgry5gj64K1mRVVjPLFUwz0GlWSzTw2MPdAeRp5S6RV4xDtGcueovTKVjsGmqlKU5Dw8WuiUDd4qCElBHU55eWOTeDK4YkrjdtqREeHW1wcRPSOKMPM1ZXW4wCVPiihBlLLp4foLZXEAoKsGZ7M+oRkg4Pp5rHZJgQlnklHku9RBpSlkskCUhVrNHnoSAjeE0KIuSJJZ3yiTO2D8L5pLuOw9HlGWJZijopkbkp0zOApyGdIBunwYEQUKeFThVed6o4vyajj5XWnpRyqcutImzgrf2Jjy10cLUVH7okS5pIaMS46qaODdqpZUFvShKjv2EjaasMpYaFhtNCbrV34OZP4/LE4LKtCWhMkEqQ3tVRTCLM3YmIXVDouxOfKnyVIRg2TXneL00L+m6JjdPPEahiWtIsvk51t7UhMTWv6eF4KUFH84w/YMN4ruW0A0an3wBsfgIzE7RQG4KtbokQyWRFDfpJmXvEQDU0T5lrSEBoaMB2vpl8rMj1MhHaXZRNyunq6KRNZcohUKmGOipvLsrwYiouY6X5BiaRRgXKKKgk40kvKXelxOKTMbmmVpNPmY4UGQkaHhKQ47wLIVJnJMWGXEmVZPnTMQwLejYGvteWikkpZmopitkoo9ry/HaaltDmR1zxTAo7CZWcIbYvcEn167zt75wmRM/qXwAcOvEYxIkbJ/KJu4XHl/ket/l78QZdt1kenGdg7sPme7dIfHeNQql/oRab4Px9pu8dzm9dfqPPoN3ekxZ5LSWe3ijkGDioVs2kZ/y4PVddMvm1jdvM7j5Ihc/9zMsb3VY7zq8czhj79Y9HrnawzZU3n7rmOlgQJEl1PsrdBZr/KVPSTPT4Tjk1ftDmnWTTtU8fGGzw/WlOpaqcOLFPL/WJMykXD3KZWbGecamnxYsV18XZiVXahmlZvHEokMznSAij36tS6Y7nAQZXVvDiifz+MZtIbg58PCSDC/JUQScBSlffXvA5aU6Xdfgzb0JmiJwLY2ffWKZXk1HETL4R1cF/92X7hBMI154apmuK6nfP3mlh1tF5kkQrtw07g7D+ZHkg10M4gdHjO9aQkHpLJLV+wi3B0uPIqKpJD1ZdcgzssYiIs8kSDZLKJ0W1BTyWge1cUruLqCbNtr6FZljkKeUhi3TtooCNI0oKwiFhYFA1Bbno6lz1L2uCEKrg2XW5J1d1RB5Rma4qCVEqo1ayhdtHOUUlHhJjp/KcjZM5Z2paaqcBSlBmvPWicdy3eSJxTpBkrMzCTE1qXNwdJVJnOHFOXlp0rF6GKqCFZzhmR2sh7cxNAOn/ew8ObsoS671XU78BLXSIhiqMn8TZkmOWdPpb64zvPcKqmGhGja13gbttU2JoIg8En8yB8sEg10eDnZpXXgcve1iOQbHd+4QnO2z/uxnyfOSaCo3C91yMdx2lW1U8OobR6iawvr1S8ymEWGY0ltrculaH9tQmQTSuLU3lM7NyUhualcflWX3KJKvU9vWZS4m8GAc0bV1Oraga+sceinryozEbmNqgmM/o2EqPOLqFMJAAFpFls6by0wyBT0vqRsKcV7geANJxkZWIq9sj3h8rclfvLRAkhcMo4ympfF7t0549eGIjmuQKwK7eP/kokQqLH/+4+sMqoZsnL2/p3K+zrM11psma40Pp6T8QZPyD62y1iJcfhJNQFQ581ynjYg9SW4qJWa8UA2GUU6YmgTqGgu2Sl6C1XSZRDlrS9coVdlR9lJpfPKSgp1hTMMssLV3f+GOLptnL+5OyUtZzioNEyFgUurYuqDIoWkYJHnJoZfOv9ZUFYI05w8OJiw4BodeTNPUJEMyzbl5NOOpFRkgu+hKV+aJn3AylfwD19KwDY28kNkRH1tv0bZ1GW1nCcTuDayHt9E++4uIk3tstQySQp6PJ3FOlJX807eOOBxHtBydMM1JKwt3s27y/KUuv/5bt1l++gscvPIlNMslizy2X/4S9eVLRJPBXHT13jXZvYnTXWGwO5xXGaP9XRRNJwt9MqdJkSWopo3bsojDlJ1Xv4XV7LF0ScJZRZWnEaaSSmVoCmGS89b2WBKjHIOPX+7y9JocQz8cBmx2HPTKZNW2VAaBiqMr7EwSXEOlY6sIzwO7TTufEZsuqgJGNEFEU0rdoiYUxO4NFH9G5/JziGhGtrBFgjTaZe11sqncjGQos9zU61Ui2MQ1ZRjxmY9rSbjOuyHEcs2SnN1JSNPSaFoaXiLTxs4zRL2kQAjZ1zhfrqFgaB88WesHY87vsdICpnFOw5QpTFFeEOclUWbhByl5IagbOUuuwiiSgFRHV+g7LuM4YxzJkv61oCAvA9KixIsz+q7BJJJqvVmSsdqwWHbNOe/xxE+qO77kMJwj1J9YrOOlkg05MSS5+fUjSa2OM+kfOPZjfu3rD3nhWp+8KKlb0mBVtzSSLOfNKojGi1IOxxF1S56n7+6MCb0Ey9Gx6/LOcvfI4+pKnYs9l6sLNT65dh1DMxAn98Cq4XgH1LIUMRvQqLVB0XhurcM4kGflS92aPIu3bAxNoe0YNDo2w90dyqIgnp7ONRTvPV4omo7VlorSIk2IpwMOXvmd9/1uhvdeAaCxeoVoeir7F2nCYPcUw3EIBrsEg13GD2/wsX//F1jtu+yfeLz9tW8RnO3z6I/8BFmaMx1M2HhshVbd5A/eGWEbGv26iW28G6RjqFK4VJQa94ZRFbJTcKGukrXX0YoMxRuwoowoqphDJZpAYpDXJNO0BJKv/gZqu4+mGtDZoNRMzP3Xyd1rnPgJn77YxdFVbg4Cri44c0IVyCZhWI0x86Lkq3dPsQ11rtk4HEds9Ws8vtLkE6stmVSuyAjB8yNgWpRz49tqw6Jhfogjww+EUt97xXnJgZcyjjLuDwP2JhGXujWiLMdLci62HaK8ZGcS8Z3dMU+tNPj67oQTL+FoGuEYKpcWajzarfHa0ZQ39idzyfCDCn5aFiVXVxuceQlPbbQ482IenPgYmsIPX+1zVNGNlusWN088jqbyDl0zVE69hN+6cTQP0n1jd0wcpvyzL98jjXLaizXyvMC0dYQQBNOYs8Mp0/07NNevSt2CJhjce5vZ4T2c7upcIVnrLLC3O+HLimB1tcHf+sJlnPazbLUMHO8Az12hcXwD/6XfRV9YRO0usb71RT6+0cZLcpZdk56j8+RGi65r4OgKkZ+SzIZ/7GsuFHVu5MqTkOb6Y/gDyVU8V2DOfz+zIWWR43RX6F+6SOSnZElGffki8WzI+rOfJY1z9o5mjAc+0UTqKPorDQmdNTX6HUmTygs55ThPADdUhcWaDNgtDIusSHn9aMYn15rSLxHIPoqIvSqMxkckHnl9kXz3Ftn+ffTNa8RVngpZinblOfzWBcwyRwz3SUcnxFtXCdIcS5NHj9Mg4eagnJu3/r1nVpjFOV++eTyvfrb3wF4AACAASURBVIJEuk4dQ2WpZfNj1xZltoeqEGfFXDI/iVJpNz/3e1TTEy+RTIoPtX5QQbx/nb/IwyBjGMlSvmXrTKq/q0Jwd+gTpDk7k4gHJx6XezUGfsL2qc/JNMaLMi79UI0lV2NSsQpefuuYH312jdv7U8YDH6dhzqPnXEvjcBzixRlGpnBjb8y3bxxz/dEFvr035tbBlAsLNWqGzHtwLflSnQfpxlmBUASRn+IPT4Al4jBhabMtI+IiyXuMZ0PO7r9KkSZsfPwF9FpjrnI8V0iabpsszSmKkokvw2D6NXmsqGWp3ByWnsRYvyn1HP01soJ3Q1/KimOgSQPSJM5I03yupPyjVlnIacb5x1pvHc2ukYX+d1Ua56Yu1bBpdB3SaEYGPPGFHyKJM1RNQdNVZsOQLC248sNfpNG2Zfp2lBH6CUenPl6UsdWvMZhKf8LzW10+vtpkxdWwkymE4Bouy3WTC00DS1MQY9m3EONDsByiG99E37yKqmik0zOEVSM7eIfUDymnPpplyMYyKQkGhl0jH0lUHbwLmomzgsNZjKOrtG2d1bpFWiv4mia1GO/tLZwnmV3qOJiaNHAdzRL04t0077x8t3KYv3YVv+IDL6H8oIL4wyvJC/713TN+580jkiru/sJCjXGQcGWxzvNrLX7z1jG/+hs36C7VsW2df/jlexRVWbjYd+m4Broi+N9fP+LNPVk9/NwLjzAOUi4v15m1bfKi5HAsxTy//dohSy2Lk12pctt7OCIKUraPPfqVnfju0YztU5/lls03bp3gjaM5/OObL+3S6jlE0zHB2QHd9VWyVGU88LFqBoN3HhKOjlA0A7u9RDQZcP///VdzM1UW+Zj1DvH0lMHtl1Eeex63ZZHEGWleEKS5zIKYDfBf+l2M9ZsoT38RdbJPEXhYDTF/Y683JJ79SpXHuVy3+LkvXuZX3/jO93i1ob58Eb3WRNF0dMsl9kYM773C4PZLc9do59LHUA2Lwc0X0SyXbvXfpq3hjyOCyYQ08ti+BfWOw8Jqg6OHI8qiZGGljqopeOOIrWdWmEUZqiK4sFDjrb0J26cBW335XNu2JG9pikDxz2B6wmKjz0+udHknzHH0kpXEJ68votiu1JAYFkIzKHQbYdVI9m9iXX4Cp7tEGQXksxHF4QNY2MIKBmRHO5RJRMNU2GzZWJpClElFZJRJR2rH1jmujpzPX+rSNDUudhy+sT1i+9Sfw2WSvMDRNY6DhH/w9Xf44vVFNpoWFzvOucqaUZjO08svdSya2oeL3hN/BmDMv+v1kdggQM63z92ErqXRrjra4zDla9sjxkFCOEtIWimaoVJkBYt9aUGe+Am2oXLn1CdIcpZbNm/tSQ7B3jDAq96g819yVjAdh/QaJmmc4Y8mWI0GtYZJmuYYmsobOxPSvKBh6zwc+KRpjqhm/ADNrgzS9Qc72NU5PksSikKn3rHnZ3dF03EWVlEqRL/kOcg3zHT/baC6kxclQhHkc85CySTOadTa6AuLMopwsk9ht0nffImVi58mL2oYqqDvqCjhhDsDn5WmRZoXJFlOc/UKwXvMW+drdngfoSi4i49gOE2EolBfvohRb5PMRui15vx52d0VVM3AcBwMW9Ktml2HOEyJfBunaeHUDcJZTFnIo4RVMxgde5i2xtl7nIxfeeOQTz/WZ28YcnmpzjhIcXRZ7qsUZO01VEUFzSA0GjSVihSl2xLLVxaUwwP0jSsItyVdtFmK6roUkS9T0mdjzGvPIfqblJUY7Typ7dw8lZfyIv7qvVM+tdXFNeVlcC6fOJ+mPByHzKrA4/Oqw3hPkthnriyw7Jo0TR0Bc3r5JMqIslwG9RQlufgQl9kPlJTfvRQh0BWFpqPz/KUuLUtnlmRkRcnMi3lzPGFn4FNkKVlSVMEx+nyMdnY4Y3TssXM045c+f5HNls3L98/4nRtHc8GPYWo8suiyfxYQh1IBuHPs8cKnNvjn/8e/Zbxziysv/DCmrfPW3oQ4ziiKkrfePkVVFRoLNoatsdqx6boGXz6a0Vy9gNVcwK675FmB4VgUWSErjd46wWCXIksZ3nsFoSgomjE3SbmLjzA7vI/htmlvPk6taaLpKqvLDW6deFzru0RZCYqG2l1C669RBB7pmy+hfPE/Rgi4qE0RWYQ4jSlqXerVMSgtZO5mrV2fPw/daWC3lxCKitVcwGq0MGwZygMtiqxgZaPF3TePSYIAf7CLZrs8/sUfQddVvHGEP42wuwY///wGX23bnIxCLiy69BsmD058FE2h13H4/GN9vCjjzIt5Y2dC09H5+FaHg1GIoam8cKXHZttmccOYTwlyFPRwUsUd5ljGDM1s0DQVhB+jpCHZwiVU40ia6aiOHOfn9SxFX7tEGfqI3oZkQhoOymgXsbyFWmTz5C0oeP1gym/8+ld47m//LBeaJuMorwAxAseVzM1ff2mXpZZF3ZKZoU1To2ureGlBz9H5Tz+2zCwtiCpEvqkJlpgSOi47k7yqBP8U6d4fIaHURyMX45mPlf/wX/weX98esdGy5zv4aZAQpjn7k4ggyQmTjLUKia4I2B6Fc1aAY6hsdZx5vPw3HkgK84O9CfWGxeUlF1UR3NidoCiCH39iaX5n+OrtAWGYsrHoMvQStvo1Oq7J0Is58xK6rjGHnSZZgRdlLLUsJu9BrD+8e0rkp5i2RhxKTYKqKex953ep9TaIZ0NaG9ckCaqxIAN/DAurpmOYGoatoSgCt2Xz5HqTrZ7LJEx5bq3JetMiK8DSBCuujlpBTswX/zGTN25QW+pgfv4X+e1xk1GY8uAsYDCN2BuG7O2MSeIM2zHmnookztB0FduW1ZqqCA5PfZ7a6vAHtwdc2+owCVKajk7L0edZnXvDgJNpTL9h8thKA1NVJNrN0jkNqomQpbPaMDkNUnRFzNPGll2Trq2xO4251JF297woaZgqWpFIoviN35cJ7sh8SrH4CKLIyB7cQO2tyqpB0cAbopg22ekhQjdk2JJhkbUkALe0GrLaEAra9BAR+2TdTXb8kjgrMTWZ73njeMYLF9q0LZUoK3ANlUmcc+Knc9n1+ThUINho6FjHN8kby5KEHk7IG0tkujxeqOEEdXCfvLNO5PS4cRLw7JKDOtzGWLv2gXIxnnviWvmtf/a//pH/rl7+1Af6Pv+u1p9YQQgh1pE5f0tAgUzy+XtCiA7wfwGbwEPgF8qyHAkhBPD3gJ8AAuCXyrJ85Y99EgK2WiYgKcbnWYh1QyXNSzZb9jzA1dFV0qJgGEq6UFCJULbaDmsNi51JCCkst2zWOg5/7ZMbvH3qY2oK4zDluQtthmFK3dBoWBrTKOPTVxYYejEn05hew+TqckOOSZdlBz7JCpqOgWOozKKUiz2XcZjy5t6EuqVxOI7QdJVnPrHI7VsnKIrArOmkcc7ClY+jaIb0RjgWVtFCVRXcloVZWbYVVaHVsjkdeMRhyvZpIC/accQ4SPj4Rpu0OA/IrXFRm2K+8m8oX/gP6S6/SHzzW5S6RZwVvLo7Zvs0oN8weWazzfW1BrcPZmz1a+wNQwaTiB9/bo1uzcDRpSz57sBnuWVhaAqLfRfHUPnkVgc/yZlFGd2awXrT4vn1Fr9//4yjKnLPNeXrpwrBY32XBVvDUEVlrlJZdjVcXWEUGXSVmMIw6duWhAKfL0+mYJeahdpbJR/sozS7lKFPeXBvTiJX7BpZc1XCeFrLFFWlIQyLvLOOZy1Qy31K0yUrYRSV9IxEUqYbywRZCWQYqqRcd2yZF5rk8ijn6EoFt1U4DSrAjS3Hm4q8DkiKEr3xLsm8VHVKzZBS+WAkow8iH5EnGAosuUY1eUn54EtIB/BHZH2QZ5IB/3VZlq8IIerAHwghfhf4JeDLZVn+ihDil4FfBv4b4C8Cl6s/nwT+x+rjH73yhEY64Wq3Ld1ycc5ZmLLsGtiaQt9RifOS01Dq548nMbdOPNqOXjEEC3YrMZKjS1xbtyablvvTiNWGha7IX/oolGfK9QpFVhQljqGyHWXz6uFkJgNk2o5AKQTbY9l9f3qjzSxKpTCmqhxsQ+XJjRYnw0ASn1UF3ZQ9Et1UaTyygl7NwffvHJAlIbVOn+Gxx9YTixzvjDEtnc6aQZI7NCvQapK9i1LzklySjmwdQxWILGLyxg26yy9SLF1GG+yDahCkGbMoYziJ5sQlL8poVlXA3tGMJMzYPvXxopTryw0apsZbexMePBjSX2nM3Zc39iZcWKhxfakOyC79sZ9gagpbfWnvDtOczZbNQvVaR3nBNClJ8pJTP2HZ1QizkjAriR2HYZCxUowqkrchLwQhAbDq9JAiCmSIc1FAUaAuXaAMPYrJGXm9T2o10c4zTfOEMgooFJXCaTONcmqK9LVoyHBekcdSDZsEOIYjj2xINXOcl5wFKd/YHtGtGfyFS12OkkxOIqqqWhHSeFUIUM4JYEYLP5VpYa4tdTlZnrKuq5SKitLocmgsMZskXLQzQCHrbHyAy+zdVYrvozFnWZaHwGH195kQ4hawCvwMkr8P8GvAV5AbxM8A/1spzy4vCSFaQojl6vt8zxWjcyepoaQJPUeb7+aaIkNSDv2McZTz5smMWZyjKnChLY8iXpIzi2MKpcRVVHRF3sFuHco5/qNLdb789gl/6fElTnwZC7996rPomqS5PJPePZoRJjnX1xp84dE+y67JJM54MAxwdMHzmx1GUcprO2NOpjGTIOWNB0PUilT08tunXFiq02+YHB1MCb2E/kYTu2YQzGLUyp7cWVlgcuozPdzBbksS1MGb3+bCs59i59gj9BKMRZcvPL6IoSqEac6lbo1l1yQvS9YbJn1HRZzG1JY6xDe/JTeHZ38S4Z+RFibXVho8tdHicrfGW8czbh8MGAw83qnSqRRN8NpNGfr76sKYXt3k4GiGaWtMxyFPXOrSdAxuH0zpuGbFR0hQheD+qVRfOoaKo6vsjkN+7cVt/vPPX2TBMXg4DnlysU6Y5UzijEMv49iLOQ0SrnRr9Gs6E7PLcZZxqS7zQLxCVhmqN6Dwp1XyeY5Sb5HtvC2Djw4fIkYDzM2rMj5xdCJjFptdFKeOMtxhqb2G4s3mmaq17hZTYVEXciPXj25ytbFMYdXZ8eXFHmUF1/puJXnPaVsqqjhXReacBilLroFVZWtoimxEdixp6hPxDFd3CSiZqi71loMyuMdpkHF/FNC/0KReRijxd6tW/8j1/ezFqDL/ngFeBhbPL/qyLA+FEP3q01aB97bO96rH3rdBCCH+BvA3APora9w+9Sv/vyxTFSFQhGzqeokUtTw4C+Z31tW6TNO2NEUeK1DmUXWqEHx6q4sXZzi6im2o3D0L6NcMlusVySjJaFs6t048DscRax2bH7ncYxSmXFtwSIuCy10Zdx/lBTvDUKLRqrCVIivIU+kGLMuSvYHPmzdP5nPwIiveF4cHYNV0It8gVFXyJOTs4QHR6IgoSLFtnWAa4/kJ1/sukygjLUo2mhY9R0dTkEnT4YSi1sX8/C9S6pa0jPtniDTk85tLFMDAT/GSjHGQ0muYjKcRtq2TVtOfes2g4xp0XYOmY6AqgoNRSLtm8PRGG0tT2BsGLDXkRCTNSz612WK9abE/jRj4CU1TY3Ozw/apT9PSqZsap0FC05Sl+VIFagkSKUFvmiptS2Ua56y4urTUpwENzYKkIL3zKsE79+Xr1O+hb12XiWj+jPGt+2RRzELkg6IS7h+g1yzMi4/Ns1b1xIMklhVIvQOllNsnqmBBUyinZ6hlAXmPvOgQV9VE3ZCy6XOGg63J911ellJCb2noikohSoQQ2IpALZI5eUyt1RFCyqwNS0VTZWrYNMqI81JuUHnyga+xEuaogI/C+sDPRAjhAv8P8F+WZTkV4o9Uh32vf/iuTmiVSPw/AVx5/KnSqnwMw0gG1/7+/TM+t9VlsWZgarLZVbe0eXDLLJG79DhI+drbA1QheGazhVFp64dePEe2BUmOF8kG5zTOqBkqv/PmMWGSM/Ni4lAaihRFsODoNE2V7UnMNx4OeW6tySwpWG3ZfOfBkKajEyQ5mqGiqgpelOFPY4JJjFXTuXq1h6Ep/JO/+z8AUk/Q33oEw9S487WvkAYTnO4q0WTAwsWnEIrKaPsuWZIz2b2N2Vzg71Q2a0VT6LRsntxoYWgKV3o17gx86pbGxU6TOCsI0oy0MPn85hLrN3+TMgp45JkvsOP2eGlnzONrTS72XWZRSt3S2T71eXazTdeRx4JzOtJ/8PF11hsW94YBTVPj45+7yFmYolY26dunProi+CtXm6j+GftajTAt+S9e2CIvZVzdFy92cQ0FRxMEWYmj5CxqpSSI+/uIIw+jvogIpUQ6P9qWEYnBjNErr+MfnaHXLA6+/iZ58nssPL6BVrMZvP4OWZSh6hqtKxvoNQu9vyJxgdXdtrDblHXJ2ShP3kHRbdZUjcLsIqIZ5cKGPNaoBvdPgzme3zU1msBiTaOpFYgkYKlWJ3c0NhomUV7IhiXSNm7MDmUwTllAlnB3GPHmscfnNtsceBnfGTRoWxmfWGuyOHuHwmlTfKhsThkW/VFZH2iDEELoyM3h18uy/KfVw8fnRwchxDJwUj2+B6y/58vXgIM/4fvj6CquqWJrCk1LR1MEJ14891Qc+zGzSiGZFSV1U+PBiceZl7BVCYTOU5tkdH05R50NvWRONTr3L4z8hCzJybOCIis5PPX5568f8NNPLHPrNOQbD4e8dO+Mft3kzrEEq4ZhShimrPZqpHFGlJW8M43xxhFppfrcPZgi3qPAUzRZGYRKMg+1OUfTnwNmVcPCH+xQ620gFBW7bpIlOc26tBJ3XcmHAFhpygnAKEx5dXfMLMq4ttKQSVBRgH75KRjusXRhmc9tdbl75lMzVNqOwTRKeXqjTb9mzuXClpZSdB28OOON4xmmpnDoxfRqBqeB5C3oqkBXVVxDI1MMFKHQMFQUpDnpHMPWMTQ0Rcg7d15SE5GkaykqjA7INj+GOj5AJJ4MLm52yQ4fks9GzHaPSf0I/1DyPs2GNQcQtx9d4ejbD8mihOhsQupHuLpGPjpBbXbJRydYq1uSRK4a5JMztPYSWW1N9jPKAjUckzVXKA2bppUziTJOg4zTQEJpuo6GrWmYME/pKquqAiE1EqYqKIUiWSLV6hY6jy+6dGyVOCvpOgbrDYsFW4NMQeTJhztifL81KaupxP8C3CrL8u++559+E/hrwK9UH//Fex7/m0KI/xPZnJz8cf0HkL2GpqXhGipNU8VQBZcXavzWm0cMpjGfvdojSHIG02huqhl6Md++PSAKUv7bX3yK0yDh//72HqMznzTOsRwd09Zp1U2Ojmckcc5Z2yYKErKkQp+pCoatUZQlw2OPr+5OubUzptWw6LgGD+6e8luK4MmNFqoiqU3Lqw1ZwYwjpsenHL/5NazWIrrtoho2e6/uk0UeVmsRs97BafUY3H2F8OzdPfJcTXm+8iSie+U5/JNdgrN9phfXMWtSE9J2pLdiEmcs16153+Sl7RHbpwHDScRTGy0Gfsojz3wBhntkS4+ie8c8t7zIWt1gb5ZUykybpim9JbYmKvy7RLMdeTF1Q+PQi7nclcCbjaaFo6sy7MaRcJS0KNGrJPIoL+hYKgu2jpfKiVOSywahDL21KFVdkr4bC3JzCMaUSUR2fFP+7GdHTO7ucHLjiCzMyKKM5WeXaF+7QJkXqJZBPPLoPNqnzAv8ozNqS120/qoMPwpmBAcnCO1FzKvPku3eQXEaZM0VCtMlyUvq07tkRzsUK08QF4LLHalbuD8KeOneGWsdm62ODWS4usv+OEatUP11Q6Lz1erIi1BIVYswKxC6wXp8xoVyhDgtyJ02n1jpUc89lGgmSejRbO4l+aDr+6pJCXwG+KvADSHEa9Vjfxu5MfwTIcR/BOwAf7n6t99GjjjvIcecf/2DPBFDVShKOd93dIUfv9jmwTDgaHzCoz2XUZjiVXfpw3HEm3dPcVs2y31XqtYUwaevLGBqfYIk5/bBlMEk4sG9MyYnYyl5dg2ypCBL80rxp7PWd9kf+PMNpVkz5hbl5c02z2y2ORqH3NufMjz2uHq5S79hcvVan6NejTTyGT14jcQb0rv6POHoiCJLicbHpMGER557Sgbp5jmaaROOjsiTCLUK3lE0gzSYcnbnO3Nm5MHdh/Q31/n137pNoyOBLWma83NfvEyS5YSJtFL3K4LT5W4NL8nYcXssXVhG944pjRrW6T3W631WVxp4lWDH1WVeB2WBKDIZSqsr9JwaLb3g2oKFoQoGgbSQr7gS6KuOdiVH0/cRRSYDgXSbQm+gTfZpxz4oGiKRLFBDKPJu/s4NhGaQ7t8HRZlnlvhHQzTLYP/rbxGOIk7ujuheaNC8IFWc03cOqS13SWYBvWcuk/kR9tICQtMpQh+hGRiXngRNx7yaIpo90sWriOVrpIbLMMrxJiltS8VcuoZYfBSRRth5Sm3/LXqGxcWrj/DTj14lyUu6tspJkLE7ibm6YFPXFbSzh2SNTbxUpmSJskT1BhiAVusS5EKSzDWTUrcISp0TP6Wh54g0Jmuucqp1+FDr+61JWZbl1/nefQWAH/0en18C/9mHeRKKkFZfQ5VTC1spUfwznlmRd+vHejXi/N3UJENTGM9inr7QouuajMKMUZCwPwy4utKg75r06yYPBh5fHYbU2nUpEW5YjCtJs6pJtmPL0RlUCkTX1gmTnGmYUjM1enVznvOQRCnR9JRJsMKFhRprHXkX9S9tkcchRZbQXVukKHLCMxfv+KEMronzuXqydeE6dneF0YPXMeodFEWtAnE8yqLAqHdQjZDp3h2G915h+ekvMNzdIZkNyeKQX33jOzRXr1Br17n4aI9nNts0HZ23jmeMg5SXdsZ8bqvLc8uLWKf3KAc7sjEnFBqASHxIq3TyirathBNEMKFn12WAUBJRJhHLy5dRgyHRiy+jLm7A5hPyTphElJFP9gdfpsxStN4q6WxEdHCA2aoz2znGXV0gixKCoyFZFNO6uMrw1jZCVUj9iHQWEE1i4mlMkeZ4xwG9rRb9JxZxVxdQdB3NNqhtrBKdDMijhOanPycnHIpK2V2X2oLq58jrfXy1xsTPyAuTyUyqM3VVYRhlnJYwiQqeWnTQlUgSziMfsfM6qxtPkTfaCOBMCFqWLqcq00PKw7uIWgfDaKAKUMKxbHYWmRyddi4gquzYXDHIswJTVSTsyDtDtNeIc+VD52OU309HjD+PJcnB0nLr6grkEh3/RF/lUsdhXfXJG01MtUWclzzWc/nMVldaa4uSr9wZcHN/yoPX97l1dZGtlQa//KOXeXyxjm1o73PlvbEzZjyLiSpmYpwV5JmsKk5PfaajkDwpWLnUYTyNuH3nlP5KA0VVKLKUMJWKzsOxfBOub7ax3WcAuLjR4r5r4E+X0YxPEExidl59Be/4HcqikB4Sy0WzXFJ/MrdUa1YNp7uK3VokHB+TeCNUw+LglS/NfRvnKxjs4vTWMe3nub7WwIsybh8M6DVMHl9rcvfMZ61usF7vo5YFWf8Kk6SgaSjyohIKpWYishglmlEODyhCHyZnMslc0xGaTnl4l+TsEKGoknB0eJciSylCX3I9i5w8CDCaXcokIjqbMrm3TxYl5FFMPPYQlQz65NW7xGMPVdeorS4w2xsx2Z5gtS2EoqDbGu6yS/PiKvXNFQCiwZD49Ayr38O49CSjRz5DM5PluhLPKHVnDoQZRznbw2Auprt96rPaMPnYcp27ZyE7k5BhKHM3W5aGtvxxmqaKPd6WisjYo9QtNp0c8gySTDYhASUYYVVhSEo0k9GNe/dR6i1Uoby7UbU0GopCU/hwcId0dIJW5Cxf+BiZ8oNcjD/TEkj0m6EI6fdPI9AMDFUjK+BObKOnKUYVuErNQKm08QD/MspY7zp86uef4l+/uM3X7j3k8TUZZ/9vXzsgS3M211v8V5+/RJLlvLVX8tnrixyNQ964e8p7JzKmrVPoBbNhiKoJ6m0b19KIYw27tcjenTOOd8b01po0bJ1ew2Qw8ObjzekoxDubcvnZC2i6wmjPZfHxz6JoBtOjPYo0wWr2sNuLgAy1Gd57hen+HYosIQ1lia4aNprlzkEv713BYJckfo7bBzOajs5g4DGeRlzsu9QMlb1ZwupKA4TCJCn4+y/t8lPX+jzZrSOKDGV6XIFWYspCnqXLNKGoKFNCMyizBMVtUWrBfJQoFJX87AgUhWQ4RjU00oe3mL2zS+pLa3ltuYtRd5g+PEK13r0wgtOQ1E9IfJkDKlRBkeaMt6c01urotkYexQhNR+2t4jh1suFAprA3ewC8kzqkqk3DViX/sYBpkrE/TXhpd4ytq5XDVY5WT4OM/VmEl+TUTdlryouSokIMWqaLb3VQADscSoVnkVFqpsxlVVQ5URDKXNQlDIsimEotxuhENkEVBev6J8lnI8hSuYmEPuF3voI5G2P31z7ExfDRalJ+JLYqIaQARVUEoiwQZYlI43lOwc4k4jRIpSW4+lxHV7A0+cfUFNY6Dp/YaEmS9PiYWwdT7lbKwempTFRacGRoS9PRudyrsdSy574JRVNQNAXDlMnSWZqTZyVGdfwQisCum0R+yOR4KHUQRTmnLZeFVD+mcU4STHBNaYTS7Bput0VjoU5W0aVV08Kuu7jdBq3l1fnrcM6JBKj1NjAqV+UfXrrTwHYMtvo1+g0Tw9JxHINZlJIX8hh23nNoGgo/da1P19Zl+pdmkbdWyVvrFJ01lI4MMC6TaH72zc4OpZKxuywfLwoocrKzQ8pIVhCqoRGPPaKTAUWS4a72cNd6OEsd7H4bu9eizAvySE6Nuo8usvUTT6PbGrW+S/dyB6EomA2DeBqjGipmqw6KilJroDS7KIaO2u5LNycwCjNunngMgrSCBJfkBVKUNQ45mcWcBclcIXvuDznP1UhzyZRURBWApWiEaUGUl1BkEE3AuAAAIABJREFUiMSj2L87r7Tmq3i3ohCqjE9IZgHh7h7h/gHRwQGFP6UMffKzIykTjwL8oyHZ8Q7F2R/bo/8eF4TyR//5c14fia1KiT0aOy9TtlfnmHllfIjbu4hhN3miXyMvpfV3HElQaL9mcPcsYBSl/M0Xtjj2Yl7cHvGf/OSjuD97nX7NQBWC/DOb/M7bJ0yClNeOPa73XZ5eamBW+Yzh8+t0XXMugPrS64d44wi3ZeFPY25+4y0Mp0n/Qoe//jOP8dXbA86GAWu9Gl6U8ebD0RxD/+YNidyv91cwNImEb/R6hLOYLPKIZ0PMeqdK2T5i/PBNssjH7q7gLm6S+lOSYELqT2ivbbL98pe+5+tlt5fQTVWasSq0XWqo1C2daZSiCnlBicRH5ClPdutQFvz+XkTdSLnStWladZRAbo7CsFC7y9JFqKgo9XaVPqYhnLokTwVTFLuGaPfJB/sYy+tonZDp7XsyJtGPKNIM1TIIjs5I/Qij7lAWBYouN5PJ/X1mBx7Ogk1wGmI2TQxXx1lwsHst7BWptSt8eYfWljdJ9+6jbzzOaZDz9ILKoadwoSnZpEE1Ti2Kkmc32tw4mHD3KOTWwZS6pbHWcWhYGo6uMgxTOo5Gz9GwVYFIAigyFhSPshAUtS5KOEHrr3FUf4SOpaKc/BuO/8F/j7d7wsaPfQI0HWPzGt7uEe6FFbztA5KpTx4lDG/9I/SaRff6I5y99U4VNt3C2HqcsrP6PX+P33MJ8WfuQQgh/gLSD6UC/3NZlr/yp/1eH4kNAs2g7K5XBGrZWS/tBqWioglZDpYVsSfNJSLM1pR55qKpSh1Fy9FpWzq6Wo2lFIGjqvMNwIszvFgqFF1DJUwL+g0Lu8o9yMuSZk02m9zqo+E0qXddag2LoBIVqZoyz3ZQVQVDU2jaOkGvxuQsIE8Khn5CHKbzyiOzdAa3JxRZwnTQpizyCibro1vu/KUoi7zSQ0B9+dJ3IepBouJMW2cwiUjCDEUTFJWE/OmNNk1TNtpIDVkWFxmFZlE3UnYmEV1HJzdV2m4PJfYRVgZZhlBVGSmoKLKaUA2ZKVKt8yOIUm/JY8hECpv0mkWRF6SzgGQakEfxPIS5SDOaqz3isUfqx+g1nTzJqS06ZGEGukI4ilCtALXZlc5MRaUMZqib12TModOmlihkisoPbchjZ5SVNGsqtiZTs9Jc8kSCJJ8nep3/Xi1N4uHOjctZCZqqIWLZ5BRliaimMIVZl0recIJQVZx+m3QWVEeokPLtV1FUBbXexqiPyaNE9omihDIvyKqPQlGwu00pG0//eLLXd/+CPySi7n1fKlTg7wNfRGqSvi2E+M2yLG/+ab7fR2KDiNG5R5+eodKc7SKKnGjhEhqyrNseRxRlSdPUmcSZjFuzSjTFoGvLaLa6qXGxLWPtFEUQpNLfrxuCbiX79ZKcL904YjyNaDUsLiw4fOZil6woiTIpKb6+1sRbkN/n0NJI45xnrvWqqPkxsypUpVPlIZybtpZbFh/bbPMvv7nN8NjnZHdCXnEwF1YbqKrCw6+nFFnK6e2Xaaxeob35OIr2MYmmm8heQ5nnjLffJIu8OdfxfNWXLzI7vI/VXCCJM378uTW2T31eu3lCvWbw7KYUQdUMVY4yz8/T02PK1ipXujbdygDWMUrU6SHCHyJUlayzLjeLPAfLlc3M2SnCbcrSdjamjCPJXXjkOugGhT/F6oHaXZKPj07Io5gsSijyAr1mk/oho1vbKIZG75lLHH/nLnkiHayarZF4KWbDoLbUoUzTuX1bffqzJPfeQN+8hvrg2ywtX+HetI0qBEFacPfMZxiaXO7IwKKjWcw4SPGijAsLNbquwWrDwtEV3jiaUasiHc/zSjTVQGQJudNGnRyS3/omPP1FKAvqr/2mFFstbVC78ij26gpKrY7iNCgiH8e0EIaF0ZACvSyIqC11KdIMf3+AUBWcfgv3+uMUszHl5OxDXA1/5griE8C9siwfAFR6pJ8Bvn83CEUIiQdXFRlnp0ptvCgSSqGw6EqWpFZNIxxdRWQJlmZTUpAXgiTPaVb9Al1R5n5+XVVkopKpyTwK1yAvSnoNE9eSzSxdlc8BHUaGOsfR24YqKUrau2c/oQhEIeafA5JrMIsy3Cglz0rKoiSryFBKZdQS4v2jrvfyIBVNR681AJm2Def9iPcr8PSapD9ZjRaartKtGXhRiqrKn6vrGLRtHVsT8sysaNW0YkIZNWhadXJTpWOU6Cd3ZMKY05KBREaNXLfR4ikomhQ4uZ35m1WYFopdI5+cgaa92+V3W2i9VcokovCnaIaFUGaohi4NZWOPdBaQJxnJ1JfiJ0PF6sifNwtPcVcXKPJCotY0nbLek+pL5a13R5JlwTTKmSUZllZxN4uSC01TGvTeo5zNi/+PvTePsezK7/s+59x9eXu9V1tX9UqyySaHw5nhzHBmJM0mS1G0WZIlwUgQAYqMBInzRywgCBAkMBAECWDAAQwkgRJbsBDIBiLLsmJb0lgaWRyJMxzOwiHZXLrZe1fXXm+77727npM/zq1HapshJSFoAXOARld1F6q73rv3d3+/7++7GLNi85CQjOYFgWMh6lzO03dzGa5su2Zz40WIxRg1NWxO4UfIuI0Mm8ukNn24gwybJsgZllsm6dhYvouqFNJxTJ6stEDKP5cj8GcdLYT52f/8syKEeLeX4C/VsoXT82dpob69mvrbnIeiQDgS+r4AZYJwtW3Ui1QlsipYj1uGpaeM2YdvSUQ5IwxDbGkxzQ1RquU7eDWfolBGH1Aok7QsBWSl4kw3pB06nF2J8G1ryZjDq23eUtNtlDVlu9sOlvgEsIyfG9cJ0FXt3pSkJYu8Ip3nFGlGVTrYrkXU9FClccF6t91cVeakddcQ9jawXN+sEYtTz0oDVkrbMXbudSGJV8/jBjZBHbxzZb3JN1dG9GpPTpNJKQwJql7NkWeIMkXOSzpx3+z4pY0KWqiwg8wXzK0A3xJG01BrFrTtIvMFWgjkyibUJCWdztHzqSkaYdP8+fgYqzOgmg4RlqRMM4p5SlWUVHmJsCTZKEFVGqu+/q3AxWv6OFGA26hDjcuCsr3xx64P4UdojC3hQS05V9oAxLOlSxQ0fPMQyEvFNC05mpsYg0adiOVIc23YAgPY+k0KBV7YwT53hcJywQmwB5voIjfYzOqW6aac0AQ81zwRXZp1rspLVKVMN2E7SEsiHRvh+cbq7v2SnmqK97c5R9/BMOY9aaHe63koCoQoUpyDa1S1sGWhLSqlEWWGyBLaxRztxiw889SxpUBM5sikxLdcFlYDzzZvfi+wcSXkymAX46xio2Eow/cnKY/Uuo3tVkBa5xishMY9udJ6mVBVKLXcUJwyKwECx6KqI+pPdR+LrCRbFJxUitk4W2ZgenGXdj+kqsx2I+pvUywSVJlTLhLy6QmqzGltXlymbGtlEqdO6dh+Zw1V5KgyN6lWYQuvdoJyLEGzJnS1aum7bxf0IwdKQ4LSJw8MUOgbuq/MZojZiekcwg4HmYUtY5J5ScuzaLU3DW6Rz8EN0HWnUHkxlpBYs6kxhj3eM5wJzze+DONj7DMXTdr53ACn+cTM7lopnMhHujZunJgZPnCxfQ+vHWP5LsHGwBQbabHAIZDaFCDPAK4VRp59b7QwnUK90jyqC3VWm/oAHNYp44vceH90QgdHSlMcdIkocrTjM0wrpBDk0sUfXDbrdr+JWD2PrHIT1NxYRbsBVCXW5HRDMUPlheF8FCW6Usi4jQgi0220eksfzNMV8ns/7wQL/wXP+9ZCfbvzUBQILS2qsAO2oR+7UqClQI6GiNmwbjlTPM/c3HI2ROzfRK8/ivIbJDODZldaE9iKRW3Jlld6mXLkSMlqZNR5p4nRljJpSWA6hklm8honaUHs2cvRYdA0N2Cl9FILcqYbcjhJee32kDB0sWxJnpYU8wnpcJ/m5qNI2yEZpbR6IU7Tprn5KFW+MPbxtotSFVWeIqRYpmyX+WLpI7l8fZRhY2bJsCYftbGkcYK6en/Mg73pkgymeiEbDR8ciZiPDS3ZcVGjA+MG7ddgpOMh8wW2jPnDu2OeXmsYboBl4Vr1Y+iUpKUVosrR0jYApeujwgZqfIw6NeP1fFM0AKRF89w65SwlG02J1rr4vRbR5SdQX/h9dKUIei0Wx2PTbdQcCmv9LCruE4/vwoNriPVzVGHXcDYwo+Vaw+P64Qy3aQKNrHpleRpeZEnBK3ffMcyJ/QLXkjgSJplCIHEsH1GyfCiA6TQDSyyl2doJUV5kfvbFGKFK4//p+djRNmoxw0pnuIuaOxIYngiOQzU+RjgOstkzblhx+73fC/D+bPL/9HkJeEQIcR7YAX4W+Nt/0W/2UBSISjrs2ysEtqChcpzcPIGq26+ZldqzP4Aoc6zpvmG6PbjGzq//Ouu/+D+xO1e8cThbZkQ4UixNTkLH4mI3JK3TsNYazrJwvHlkwnQWhfGaKJTm6v6UmwcJ07RcBuvc2k9YbwdstgO2uwFpXSCe22rzrb0pz794jzNPrXGmGzCeF7zx/BHJ/m0ufeLjLJKcey+/xOAHP8vjW8ZApioNcDnomNFlPC+4+cYho919pg9u4Lf6DB57htt1gVBFTpUbKvfJ29+gsX4RVSp2j4xN3M2bJ3iBzYPhgr/97BZJVrKXZPTDyNCnx8foIkfPJli9dShLyu4W2o2YWwHJvOTptQYNV7IoNXszE/TS8gIcKQzaX5WIbGqeqnEL7TWRyQi7v2k8G+zQyLZrzMReP0c+egW3GRJdOI9KRqi8ILn6GpbvUs5qo9pFjnRtivpzFfVQbgRuhBU2oCzqfNYW2nJZCR0cK+Zrd4b0Ypd+5JpszYXptmLPdFV3bg+xHYv+arzErTbbPi/eN9iCY0me6MdcCeaooEWJxBYgyhTKHFHlVH6DE6fHyuQm6sHbqHRmRq64jbV+Hm25MNxFxC2E1lTDA3RmRo/JG29Szhb0P/d5aA5Q1ns3odWwHGf/IkdrXQoh/kvgdzBrzn+itb76F/1+D0WB0FA/CcRSngtgdQZm/j6+b55+YRMVtBBrF1n7/s8isoT1dov2dpNxVnFvnLEeu9iW4HjuIYUZORSass5OHC4K0trMpRE4DFY8nr9zwh9dO1q6NPdil5dumlSq7X5UW+cXnOmGRpdRVtyoNyOfeW6bg0nGnaM5622fRz/5YeaTJ/nBj21zkmR8qx/RbniMalmxVpqqXsmdYhvd9Zi47eOGLap8QXK0T/vsk4zvvU42OaS19QRRf4vDN7+C2+iwsd3mwiDCtSWDjSaTkTF72Wr6vLI/peHatB0F2dx0C7XRirAdhGUhsxmVYzCHlmexKBWL0hTYJK8olHk6V1IT27YRdEnb4EOOh7Zc7M4AXVXIIKrFUw740XK8OEX4VTJicWDGm/mh+d0KXBbHY+IzfYpZiq4q8/SFWvA1Q4SGCapsj6q5ylw7JKOMtm/zyFqDi50QIaCoNF+8dohrS166dcLhJCNsenTbAZ95YrCMzHtrb8qTG4Z4prQBvGVyiHYDbNuvMZoh82gNd+0JZJbQLYeIwryGuD4iqH+mox0Dps6mWHEL5TWQcY7CbCwaF88tiWU2BuB9v/fDX+Zorf8tRjT5lz4PRYGwUfRkZnbxxcIAbIBodLGDyFRnpcw8qhTaMnRcigXWaIcq3sSVgjNNj7Zv4UpBVmosYVbKLdcyncOiJC1NkYg9m9i16QQWjjRJW/F6o1ZISiyZ0fBtLgxiDupIvqr2r2zU2w9LKJ460+KtvSn3TxYcJzle4OA4Fptt41x1ihecbkJs16IXGDYn1BuQSUpZVNiuDQRoVeF0YsLeBg++8QVmh3exgwhpu+TTIddf22e40WB1EOPakqcu9fjgdoe3T+ZLP4fHV3zsPF3OwqfmKmo2QVQVdjZBBR1a7U18y2JvZsar822Pk0XJMDUMxNXIJnRjHFWZ4pwlpsCEXWQ+AzlCBpFZgWK2GioZmdWnUpRHu7jNkDLNCfsdLN/F9l3Skwl26OM0QtxOx4wuRQYBqNhQq2U6RYUdpsomr52t3j5ZcHklIi0Vv3fzGEsIWqHLzYOE+4czqkrRavqc6Qasxh5SwHGSL9fRp695qbTRpBTZMsldOyEeBdPCpqkV+vUvQX8TubJpAntmQ9MpYbgoWkp0Ood0Dn5oHmKWhez0TUGuP+f92Njrv1wH8Vd9HooCwXyCfuk3TfvW30C7sUGZgxZa2lhFbsgzlrMsIKrMEckJMmrzOzeGbDZ9nutptAwpayB3XmrGWUHsejRdSaUs7mlNVik6UtDwbAJb8vRag6dWY861fZ6/M+L2yZxnL3SJXMsYgLQDkrxkb5JxZa3BSujSDRx2pym/e+2QM90QSwq+9K1djnamVGXFrcdWePPBlJuv7tP66BnOrkTc3J3SbHh87+X+8oKNXJs/+sM77L7yPI2NS7hhk/Z6Hz90Obz3TrZmWc+6TtQin895/MJZwjpUtlWDrC3vHT8H1xLoPDVdg+1SHu8iGx1DgvJj0xHYhkjlWuDZgkJJThYlgSM5mRpbOUsI7DypV28WUlUml0LaaGkjbAcZ9czcLSWq3vkL1zdqz5V1s/5sYmjbtgNlgd9tUhXmQaAWM5xzjxswsD4q6oG0mVkRRakIHcnRPOd27ag9nOd84/aQsyshu6OUb379AYPtFoNuyKDp0W+ap7bScPMg4XCScWEQ049cpDCWg/naWbyT2wDM2mcJ9Axrsk8z6pnrrNGm6m2jwo6RAAQtwx3BSL+RFvm1lylHJwQf/IQpGvX2Q7Z6qHMfMq9R+U540Hc6GpapXA/DeTgKhK7QixlKVeg8RV58huvWGR5L3iR77SvoVg/hR1iDMwhVmhd9ZRsmB5DN+cBqg4YnEdWUUQHTrOBsw2JawksPEhqeReRIPEvQ8pxlgMo0K1kJDBqelYokV4xT4/p8msUZ1ASbq/fHvP7KHn/rhx7jibUGv/yVO3VehIdrSQZNnycu9jjsx+Slohd7fP5KyI8/s0GSV4wXBbdevU/ca3KmG5CXioNJxt7+lPnoEK/Rpbu1jVaaZLhg/9q1JYuymE8Y3XmN7qUPATA7vMd4fomPXejy6v0xbz6YcP9kzrPfd3E5lx/OK9bXH0HvXkeXOdYpfboWIGnLMX6W+RwBtLwA3zIuzSfTnNvDBVIKLnZCVsKAlcB0YUFYYZWpab2rgqpWeFJL2oXrG/3EzChV1cKMNzqdGfFXnqKq1Ogs2l3jgrV50XQewEwGuBqcKkd5EUrpmgmpmRcVeWU4DrFv8+yFLr3IZb0dsHtswnSGNZEtcC2U1lw/nLGoXcErpZf6nt1pxstC8GyrgzXZJcgnxsY+aGFN9gxHYnAeXXMltFbgN6iEXIK2wm/hXXZwkhGytQJC4jz9fai7b6DGx8hs+hfSTzxEDcRDUiCoTTKUqkkxmp1JxqPFnPTBAwJpId+9TxYS5UXo4z3E5edoWbW6Ly3JSk1aaWQ6xfNaWAKmWbWMRYtci9i1SPIK5WpOO89CGWLNJC2YvwsfOF1n7h8knNx6k93RNtvdgGu3h3iBw3o7QApB5NqcXYlYbwdUygS/DiKXD643eGU/IclKFsN9HD+oQ4AqTsYpJ/t1ZxC2CGOPPCuZjxPmxzt/6lWyXN9QtIOYVugwyyvOrpiQn7Wmz/GiWNrEFUpjzU/Ij3dNZ9ZbNwQny60Zkl3TQbhmbHOkoJIGh2h5JsTnU1st8srwSYRWeJQG2W+sItIJ1mKEDBuo+dQUgcVsufu31s+DKql27yCihulidm6YtZ+qyE9GuKfisMMdrP4mym8R2oK3RzkgudR2ac73OXJWjHV+jd1kNc9h0PSWit71XsjecEGZVySW4aRIIf6Y/b9RcpoikVWKnUnKR1ZO/SVTCFpoIRHJMWrlAsJyjOkNZluiLRfsWrhluWiZIht9ZNBA2z7aNl8vggg1nCFOncP0e0/XOpUUPCzn4SgQjou9cc5IZYcHlK99ie9rdmH7CvHf+rtUL/6/hsVmGwszFXYYyZjuo88i50PWh28guhtMe4/QlYK2b3GYW9il4pNbTV4/WvDyXmIyHEKbbhDyX/zaa5zpBvzYU+v866t7DJo+P/ehDX7zK3e5+Y03+Pu/+MOkpeJbd0f81v/zB8wO77L97GcIXYtZXhG1fPoNj812wCB2yUrFYZLxg4/16QQO//PvXic/0+LZzabJ5PBs/rP//IcYNDweW4l4dX/KetvHemyFL30r4mR/xsm+IUc5fsDWh7+X4c69ZcYnwOHrLxD0Nnjy+z9LO3SYpiVX1hpYwmgRLMHSJm4jtklfeNG0vfac4tZVRNjAavUQsRndZL5Y8hxsyya2bVYjM1Zc7ITGrs2TvH64QAiP2HUItLnotRdT2S7KjZD5DJFOEUB161UIIkPAGu0a/KhWhyIlVmdA+vZ1nMjQlQEz9qgK7fhMC2NjZ0BTSPw+90cZv/ziXUZz82/npSJwLZK04Pk3D7Gk4Ew3oOHbZKUidC3OdENWQofOVpsLvZBv3huZoKWiqpm7FtutAC1typULvD4WPBJ7uLNDs5npbXPkrKAzjdIVgW1StqLdN9BRB217WPMTit55k+SlSqzRDuWL/wZn4zzOxQ9QttZw9t+iPHp/as6HqD48HAVCSMsw6Pbumog1z0cND6EyKyeoPQpqwoooFrTiGBbGfUd21qiCFotCMVWg0CwKTdOTOMWczYZ50ghhrOOVpvaZFIyzklboshK7eJbg408YRWGlWAKTzfVz+K0V1s51lnyIK5smZCYtTQq3Z5uE6pd2xriWZL3tE3s209zcgJ4tuXM0o1KaR3ohB5OU3VG6zNN0PIswdsmzknRWUFUa+S5wS9oOth9j2S6OY7CHU43J0TynqPQyKcu3LUP6Wd1GOC5VLTdebgrqtlefioK0QlRGJBe6MXaesBIGaA2vHxrDlUJp1mOXbmDTrH0mleUYMluRme1DlWOvboMXoO++RlVVhh9QFsuNRzU8wG2E2N0+IoiojvfeEYQJSVGZG7jlCU7SilIZgZ4lxfK1ejdJLahxmCXTVQrj9p0bY9p5YTI6Tk9RaWoNF44lyLGNSEunS79N8fgnyaM+kdIkhXG1zipo6Dl4AcpvGPfsIjfmNapCSwuhNdILkM0eCImcHpqfu/H+eBDf7SD+xNHSohoesPvFF9j6xf+eqr2J+Pq/NhmMpUHHRRAZJ+TsEJkcY033UUHHpCq5HeaFZryoePtkztE851w74GzbZ1AcspYco/2YsnfOkG5Uyc8+vc7OJOPl3QmP9COe6MdEjuS//cwFkk+e43/78p1lW/rsx7c40w342LkuO5OUJC35qac32J/lvHDzmCuDeDnb/qNf+Qbje2/wj//BL1Bpze/fOqYbODhS8MILd+mtx1xejXnhjUOGBwmNTkBVarzA4YlHV9g5nnPzW/dIJwnlYkZz81GymnHZu/Qh3DAkGaUMmj5bLX+Z5/DcubaxpreMwaw1vAfnnoLd6ybvUlWo2sGK6ciwFFc2qbzYzNPZFKSNU1/sK4ERfAnhUSjNc90K7WjuzCrcyKSKyywxG6e6C1FOgJiNYDZBrF1ETA6w1s4hyoz06osIv49UFfbqtsEiao4GAK5HVmluDFPavsOirPjAigeq4tpxyeV3xSDO7xo+Q1aqZcrXzYOEvFRLbkledxKv3R8v/TvzUuHaEqkMVyZ0LA5ryXvgmMAcy4lxvBgnn+PNjqkaZ8izCqVBuaG58fOFcT1LZ+ibxqbVXt0ynpX9TXTUBVWS/9G/xP/Qp9Gt1fd1P3wXg/iTp8gQfsT6Zz9hVmfjXeTqtiHY1Bdzcf9t0q8+TzFL6Xzq07y1/VnOtly8ck6pzDrTswUfXm/Uohzjc1k1zhrEWStEmbOnY47Tkm/ujimUYqPpo7Rmf5bT9m1eO0jYqdOzpmnJ/jglWxQcTjOzrRCCTujS9h3S0sjFPdsiKyvOdUN+7qc/wCJ/wjhRpyXHSU6SlkzTgvHOLdzgESpljGgsS5LOC9r9iMiz2TmeY9kSxw9w/IAybJFOjoy9W5ljuT5uYDObpNw/mfPxrTb7s5wbRzMTJSgFsWvcp7Fs5HyIKguEtCiHB4azkM7QWYoMjHbAEtJ0YZXRwaigBRhA0qMkdh3WYxftaERVcjAz6+P12GEmQ5pygWquIRdD00kAtNcQybHBJcaHaNsx2ZqHO9i9dZOade+6MZ7pDAynwPbxpOlS9mcF3drgZoHDegOO6vHi63eHNY26ot/0eP7qPnlWMh0uELXEP1sU7EYuD4YLiqykKhV3742JP7xRd4qSldBlLTaGMRqIXZs2KSiJ1rWVnJSEoiCIXIQqsd9+Ad0amHVvMTeiLWkhGh0ANArZGaBVaQDL9fP16v79bDH+0lTrv9LzcBSIsoCywN4wwJbMZ+CF6OGBEQVJCzUdce+L30JYkmity+aVH8CpAeK2CwslmeYVjVASUCxtu0SZmhwDrRBVgRABtjQjQc91ONP06YU2Wal543DGGwcJ90/mlPV4oWrmY+BYtDybTmD8Jk6t+gexRy9wOJwZHccH1ptL4AzMmDKa5yzyiu72RbqrMbFn85HzHV6xJMOTOWVeMQM+cNa0om98+U0cPzZqz6qqxVwBXmDjejZBz+VgkvHFG8dLvcjOJOVnL7copUuhNGI2gzxFnbocpTNEZ7B8vU9t0azZ1LTAcQvteIbnoCqCsEJ5EYEu6AY2d2YVBzPNM2sRti7ZnZfYUtDwG2aOb21iD+9BOkf6iRkz3BiRma7FvfAkxd1rS32Fyhb1xqNvPC+FJFNi6TM6LyoW2mNnWvA71w3OcDBJeeHVPbq9iLW2z6cuGd3KK28c1j9WhVamG3M8C6007abPLCuZT6Z/jAsha8Zt7FpIYQKkj1OPJFNcnF2nbJ9BBx1EkZpwYSERjS7V/bcep4gNAAAgAElEQVQQjkvy+OcJgw72g6sGmNTKqGNtD3beRPa34NGPUoHpON7r0SzlAQ/DeTgKBBiRT5EjtEaJ2pr9z/q6SnHnCy9hP/8fcenv/T3KwSXsG18h2nic2yOLfmgbIz1VIrMZ9vAuZXMN7bcQVcGqGrHqSQYXOpRKE9iS8Mu/ivPoh/lfr4p3SYY1lWvRbnjQ8Hj2Qpfnttp4tkAiWNRszKZvczgz+/lf//p9fuajW8sCIWsbvdO0r5/7m0/Q9B0arsV/+PgqT2+2+O2r+7zwwl2So122fvRjXLs95PD1F3DjDpYXoIqcwaWLNHshs1rX8VMf3+b6/pS90YILdRr34SzHmh0jhTRztCqX9nBICbZjkrMbbePnYNtGlZnODWfBa6It14i5ijlWmULDtMZNy8GNIpOXqkvsm1+F9Y8yzSs6vm9ac62RjT7O/MSEy9g+yg3QjQEin2FN97HPPU5x8zWE62P3jcuScFxE2EAFLao61b0bOPi2oFSaL98b8dr9MZXSrLd9vufpdV5864gTW3JcF97BeoNu7HJ3P6GqFD/y0S0DSOYVF3oh06zi917fJ/YdikojhSbJSsZZxUDMDI7gxnzhZs6Lt074hz/8BGhFriVufS2dekfYgzNUwwPi639gNCl5ajCW9hraDQyrUlqIMuNfPLD55Fab1ei9B+dovgtS/unjeNibF82TxWug3QANyPULS5DSXj/H5TMXAdBlgRofU5z5APdnmnTlIzil4KmBydZIlE2lIfA7eD2L1GkYVmO6a3j9jk9LlYyx+eLtER/7yM+yKBVXNo95crVBL3C4M1pwMMt56fYJNw9mXNub8uZaA6WN5+O5mjw1rN2SC6XIFgX/7Ct3aQYO/+MPP04ncLCE4O54QZKWfM+5LvOi4vZogSMFhzMzfjS6AY53hrfvj9Fac/H7fgytIG77HN47MrkY6ZT5eEy2KHi+E/C5JwZL3COsu5sdO6LpGgQuDjvGmr5eKVqujbu+hbBdcFzQCj2fUh3vocKG0VZ0BlRh1wCOxRyRTtBejFAVriVYjx125yWsf5Sub7GTKPZnJR3fwq8WoErKzrZxBYN3aNOnOEVZLguD1RmA66G9pgGj8zmVp5kVCs8yAbkauLwScbET8vLehDcfTDiYZLz99beIV1Zxbcn9vSllUTEaWSQjw0j91d++RtT0uLjd5uZBUhdpSV5WHM9y2qFjfDtzhePHhI0mjoRPbRc8vdZknCuyUjPKTEJ57Nho1zbPHXeKPvdBxP2rhuvQ6BhadXfbjCEH16jOPk3hRfzAqkdjchcxen9bjO+OGH/iCCkRQWzMSYQ0yULSBi+C0kULgXBjrMjMeqJMUeMjxpVkUZZ8/cGY7VbARzZiFuWpshNA4XoxeanxhDC2YmEHLSRWNkM6LQ6SnK89MJFzg8hlq+kRu5JS+Ugp6Dd97p+YG/xglqO0cZ9aCQ3h6jQq3pES27FIRgsW0wy39h9oecYTMS8VncC83POiqolSBoD1QwfLkmSzAtuVrF/ookpTcNwwpMxLSqBIE9JZwMFwgWdJYs9mUVTcGy04d67LotBIFGmliJwAXRZU8/nSYNbuLtDj4yWJ6dSzQI2PjelLVSHzmSGiVQXWYmRWmZaDzBJmMsSWgmlesZMoSgW2NBoX3IBUKWLfwsU8cWU6RhSZef+0pkpG6GxhfCOGB8ioCbbxINV2Y5leNVyY8eVUfu9Ygsi1aIUOB5OMZO82AAfDDum8oMwrpCyXHy9G90kHG3S6IYusxLIl3ch9R56fV8wLY3o7ySuU1gSO5ExsYQ0fcKy3WJSK/SSjF1goGyyVI8rM0LPrh5ZsdBBhbOTvwsYSykjf11pk2iJeHCFGu4aO/R6PUXN+t0D8+acGE5ElcnZs2uDVR9CqNLbkQqKiHtXqZVrXnqcxPKR89EdxLLG0EwMDUrZciXP7JboY0VB28yr26hZOZ4DwIzp+zM8/vsH9wufOKMXybSZZxaxQfHVnxCDy+IVnz/D6uY4xPQ0cBpHxt/ylL9/mkbUGP/H4gN0kx7ck/+Ann0IKgdKa++OMQhn9gFUTqb61N0Vp88R/6eYJean4zOMDfrcmYpWFwvFstnoh33xlj7vf/Crzw3s01i/y1Oc/xZ03IGz5nK1xjElacq4d8E9fuMOdoxn/1fdcQAjo+pZJvOpv4rZ6FLffoExzJm++jRP5+H2WTknC81HSMrv/IAI5QtgOVVkYEpQbmVWmKmnKBQ2/Qcf32Z+V2FLT9S3ePskI684lciSVtHGTHURVoMKOWasuRsYdqiwMkaruIJgbKbeO+0TlFFcGfGN3wkro8PRao07Xlrx6f48Lg5jdUcrWhz9BGHuk85y4zio9ejDBsgR25NDsnsMN7OXqM09LdurOwbUl87xiZ5ySVYqPbLTohza+JXh5P6UbbuBWClsK1mLPGB9rQLocYNFrd3DSMfrSxwGwD66h51Psr/0G9taj3Fv7CBvZjHhiuob9jWfZTf541OK3vfw1y3yPh+E8FAVCK5P+VMV9U53LHFEUqMmJYeetP27mwCp/h/aK4UYI18e3JZZ8pzWTwtjii3xukPRTfMMPjUK0Rp2NoWuFJaHl20Yubpk9eloqkrxkWu/Tj+fFsgA5dbDJIq84mpfM6135NK8YRA4CyZfvjQgdiwvdECnF0oY9qxSTtGRvnBJ7NptNn0fWYiql2d2ZkM4Kbu1OsWyJ3+ozP7xHNj0hz0oa3ZCw4TJoenR9M76sRC5/9zMXafnGm3NeKFYCB3u8QzEdovOU6a17qLzEDlycyMfqrZlW33ZMAVYV0g5RixkyiJBRD7WYoeZTQ4IqMqONaa6hpdmSdHyLJFe8fZKxGjvMCoVvCdJK0xIVqtYznNrXKb+FcALDBq3t2oSQxksiMKvWQ7vLNC95atWsLu+OU9YbHrvTjFbocjBJOZxkDM60iAOHbuyyyCvu7oxZTDOElFi2ZDacEnUarPRj0llOWShsR9ZcCmc5coSOZRLlFxWxK+lHjkl3E5BpvcyJzSuFFAJbgp0ntUV+7fnpxlgdjeyaGD5XGkWytn2UF5EUhqH7nu8FNIV6eFDKh6JAUJhWe1+2GTgZ1mQPmU1Jb7+Jmk1wL3/KCLSGB8a5x/aR2Qw9OI/obbESmuBYA0DV8X2jOwit0VEXMT1ERk3cRofqzJOUXhN3soO2XLTjYRUGzY5dycmiIi0Vx0nOaF4wzSpevT9iNC9wbUOAcm0zMtw5mvGrScaz2x1Cx+L68Zze2Q6BDb/6Bze5cKbFJ//Go4T1uuVfvXnIwTTj+t6U490p7pkWVwZGQrzeDvjVnQnj4zn333ibrSuXWLt0gdHtV9n68Pdi2ZKVzSaLacbNgxlH2zlPDGIcKVgJXRqejdbGmDcpLDrZjPTBA9LjCcVsQbxppNWqUlAWSx6CGh8bV6iyML6MWWp8J2v1p0iny42QXAwpW5tYWhvMwQ0Ms7RQ/NHdIY90I853fColiJwIz4uxJrugfcMbCDvImlC1jP8rC0TNks0rvbyZThYFK6HL0bzg9mjBI/2I568f4dqSyxsNdkcpi7xi92jGN/7FP8NrrhD2NpGOw/G1r9FYv8jjVz7PrcMZZaFY2WjQCh1WYhdHSpq+zXbLZ3eakZYmqHgtspekq1IpPFss/UMUColAByHadnEO30aNDlBnn6FqrqHdEFGmdC0BqaSKVxgSks3L5QPlPZ3vbjH+9NFlQXnrKqsfWEUU5uJRi9nSvcg+uU01Pmb61efx2g3cyx9m8chZCrux1M5b0owYBlGH6tXncc5fIdl8hjhPUMMDhBeY7ANAeY1atORy9yjl9mjBIHJZjT0ans2FXojSMIhcHultsJdk/M7Vfa7vJctNx4VBzA880qfpW1QKOoHDvUlKUSn+u7/5JL4lmReKrz2YcDTPGUQuLd+mF7sEroVnS17ambCfZBxMUoKGS9j0OH+5z7TGJz70Ez9NkVXYjsXe7SFaaaQtmRcVK4FNWilujxYczXO+/2KPrmubFlXaeO0G47eNpkMVJaooKaZznOEBajbB6gywz1xcrkHxTbE6jczTi5lRLla5IUEVGfbwHrLRB1WS1k863xI80o3YbnmM0opCKS51jNu0XWSIfPFOYpo2T1dRzA046froWqnrYlaPp8xH35ZQu41/8Y0DAtei3/QIXFOgr5xpcThOWbn8MZK920vD31PW6fE4pbUSGn6ELamUXm4xTolSZ5q+wTqyEqUN03aSG1C2LQsm2kW9q+U/TDWVlqwNHsXRigrQlo2z/yZa2hT9Syg3QBQp3WpIFbRxrfduW3vqvfmwnIeiQCDMGGAf3jRrOQBpGdouUE2H6MUMf9A3EXC2iycqhGWjtcazBKWGSVYROCaez3n0GYNXgImfV8pcjFb9I9dJRaLK6UcOqs7caHomc+PUs3JeVHQCB8+SS6aeJQWfuNRjs+Gz1nDYmxofyFku2JmkHM9yPnuxh1KGn/D63pTd0YKfeHqDor5Iz3RDkrTg9b0pi7wkLxXtpr8sPos6U2NzEHN/b8r0ZIFWGsez6XdDWr5phye55gOrDVqe6YBsaVpikSdM7+5TpjnReg+nEWL57jK3wnZ987qW7/gl6vnUqCrHx0YnYTtUt17FXt02DEmAdI4zP6HsbBP7RiWbVprzHZ9RWuFYgqN5yeG8JHYlQeuseT8kWPMh2qm7P6cWSbnvZIK4liD2LNJK0Yo9houCSmsGsYslBbcPZ6w0TAbpV2+c8G//8DZB7OLFXcb5GxRpguPHRP1tvEaHPCvxQ5fx8RzHsziYZPXra4Dj2JV0fMuAlZmi41sEtuBwbtbf2vahUAS2EWuBKVruYsi4ahGvP4nM54amfjr2VjmldLEdsKYHhN2uIa69x/NXYDn3V3oeigIh/AB7/Tzl/l1mb7yK127gf/w/oIr7JrxldA/d3cC68n2GN59NsW9/DacOltW2h4z7aKtNKx9iJYeo8REybBItjoxXgeeju5tUUY9KusbeXStEWXEhPeZco8WR3aXrVMhsyvlOgzvjjOdvHNGLPaa1a/UjazGDps+PX+7jSANI9iOH40XBqwdTvnXPjCPHSVYXE0mSFiZoR2uKSjHNS8ZzM8JUKluCaettnztHc966uk//jDGn3TlIGL2rTfYjl888MWCzaZy/80qzKCt6gUVoC5JCEdUjTby5QpVmuI0QJwqY7x0vcyuEnCIsaRygpIW9fg4wYC7Ua0hpQRCBF8BsAu01pJ+YLZMqcSmppE1LVFRKUCjFUU1dNoVK0/Jqo9iyRKRme6KFMGFJQmLNjctU5jY5nOYMFwXXj2e8dOuE//S5s7x1NGOz4fOTz2zy9vGMr7x9zG+8eI88LfBDBz9ymR3eo0wTLDdA2Tmtrcv4kYNWmv27I8pC4QY2t3annCQ5gWsiA1qeRZRPQECz0cSWht9icAfN26OcjdihKTJjdSgkIi3Q0jBNCwv82li5XLkASpnAnahvrjFphG/W+wjC0fU18rCch6JAoLXxLOgMaHzkk8i4TdVYNUnO+YzsrW9iDzahtWGQcDdCt9eNAvDUUNX22JsWDGRC9eAG4sIHjWHJ+AFq9QJgougqIckqTe40jOciykjHLRfSihd2c3YTxX5yQNt3+Nwjfc40fe5PUvZGi6WleqkMIDjNFW8eJhzMciqtefHVfYb7Cf/3f/NpxlnB8zdPlsSrr94dschLAtfmma22GT9un9CLXVzb4s7RjPk8Z3J4yKXHByyKire+9FXS8SGPfvr7sWzJcD8hSUuO5gW+bXE0yxlnJWuRy7zU9dNK4QpJmeZko4TJ7T2CfptillLVoTaW61CmhgLcPLdee0hGWL01QwCaTbDWz6PCDvrua0ZbkRwbhmRtLizKHDfZQUU9IifiUsfnsGZYblkzrs6NVqQf2hym0GpsEUtDYJOLsWEf1mCe1prQkdixMXR5bsuwSi90wuUN9tr9MXfvjTm4c8hiuM/3/Mgnubczprl+jnj1LFkyRKuK6e5NitYKrmczHyd4cWRyUl2LZFFQKc0gcnnrKGWrFRM7kpd3ZziW2ViEjkQpI7CzBFAps1FzQ/KozzgzAr1Saaqwg/AbIG2mheLGTNDHdE95+yzjtHzfmMJDtMT4zgVCCOEDzwNe/fW/prX+H2rX3H8OdIFvAP+x1joXQnjArwAfBo6Bn9Fa3/62/4jWhm4d+shGx5CZbA9RGsBMzSeoRds485xG2Dt1TJ9SZp6ldqgO6vnZjUHMUccTdGOAtlysbGK2JVov8wdsW5ILl6wOu71+MuPVmrl3eaPJ2bZfk4981tqB0VHYcqn23J2m7CYZx4lZo2WLgiwZsxo7RHUEHJjx5CTJCFyL0LVYCd2l7X7g2oSn7M1Socp8OWef+kI0OwHJKMULbI6TDEcK1mOb9dhmNzFP7VBWRCJFWz7acpnvnRhk33fJRgluI0RYEicKiNa6FPOUfDKnnKW4zTqXQimzigTDRh3tGlXm5MBshNwY5QYgbcNzqApEscDzYiqNuTEqzdW5z43hnCv9mHFmtgRZpYkdF9TYaBjK2lqwSPFiQaVglpvtEZh23rEEB0mOYwl2hgumJwuKNKG9dYHxomC4P2N2smcsCesMkdNEsmJ1hd6ZHmVesXP1LbjyGHHLp3I0d0YLHEuyEpqQpp2pGW0Dx7w3RaU4STXrsYPnuKi4j3JDhDaxf1orGp4FJQhh4WqYFYpXD6Y8t9XGt003kuTqfXYQ/LXrIDLgs1rrRAjhAH8ohPgt4L8G/qHW+p8LIf4P4OeB/73+fai1viSE+FngfwF+5jv9I8I1HP3KbxnBkFZUrU10Z5v8sc/gpCfkv/2PAUO3roqS6MOfQja7ZC9/CffRhLPrH+Xtcg0urHHBKhAFiLiFnB6g3Ihy5ZJxMJaagAKZTrGOD2HlAkllcbSo+Nz5Ln/jYs8Y0GDmwZf3EuaF4heefSfGfSWwmBWCrZZRVZqLu+Kpn38WpTXTXHE0y9lo+Wy3fGLXeBV8/cGYX/k3b7Lyk08SuzY7Jwv2RilunfoVNjwagw3G9dbksc/+EIONJmdXQi48s8FxkvHK3TE//fQmsWOcqPeTjHlus2prrOkB2jLYwWmHAAak1EoZoHK2IBslVMuPp0v36fJo19jELWZUu3eohgeIIMJaO4ceHyKyyZI+LYrM8BykjTXZxS4ygtZZWp5AabjSj+t1aEWlTaiPBqNX0ApqPEjbPiJLcGTAvDBbJM/W3D2e8WgvwrMlrx8mbPdC1CWN5UosS3Lz1X0GW02C+CxFVnLjD/7VOxft5Iigs0rRbNJdjZgf7+B6T3C+9h2tlGa4KPjlr+8waHo8vdbEtySbTZeVwOJwXiEETPKKk1RjCYewquh7Gq01i1JjW4r9TFFUijNNExb9SC9iIzbpZq8fpbiWXAYFvZdjQMqHp4X4jvsXbU5Sf+rUvzTwWeDX6j//p8CP1x//WP059d9/TojvUEK1Xj61hKoMqAjv+Bachq6muZmfLYnbHxggzXKx188h4jaBLfFtsRRxads3Vup+y/hZlhk2Zl+P5aIdo+03voyC0DbrzoYrl22hWwcDx66F1uDUNurjzIwXi8JEwk2yktA93asXJmJeCo5nec3LsDhJTXvb6BgXqtMnxSnwOZ4XWELQXY2WHURZx9dP05JpHerTCh1O0oJhWjFMK47muelGVGmezKpC2C7ti5tL30fLd5GOTeviJrOdIya3d5nc2mVxOML23aX7tCpKE2VfFoiogVbV0n5+GQZT06e1EEv7Om37oA3ByNYl/dAmdiVJXnHtZMHxvETVN5b24uUGCUCkE/N/FsajoeU7WALOt0O6gQEUk7RkvR1waTWmqj02qjwlzyoObt4iW7zj+QBmk+GGMelkwq1vvI5WFXE7YL3tE7jGRNizJe3Q8EmysmIltGm6Rgc08CqkMNqQUr0DHCaVXGaQaA2+bd5b1xL4tmAldCiUoYyb9/ada+a9nFMm5Z/36//v854wiDox+OvAJUxy8A1gpLU+fVfuYzIB4V3ZgLVH/xjoAUd/4nv+HeDvAGyv9U12Q5GDp0w+gZDGM1FDKOqnvWOblKaN84inPg3ZDK0V+spnKCybtjSyXaUBVVGFHYZpk1bTws4TrOmBUXfaPjOvjZYBRAGhFMSybmkr02reyc2bvhrarMcu01wxKxSzuvt+dd/YuzuW5NZwTtt3eG6rzY2TOW8+mHB5JSbJSr51d8STq0YL8sJNY+j6ox/fJnYtjmsJ86npye37Y5rtgB957iz3T+ZcvTNicjjG8WwWs3x5YT57oct67NGTGVkY8mgvouVZWLMdGD6A5grFzg1O3rhDNkqYHy3oPbZKNkrIRgnStZdxeE7k4/daqPw08i83BrPpzOg2lEJXFenVF5FBhHvhSZNPopWR0S9GKL+1NPdxJIiy5DA140alBauRy/16bdsNbJQrabjRO+Q3yzZ2+rnxttidpoSOxRsHCU+uNrh2POPGQcLbOxNuv3qXqsxp9Pp84tOP8sqre4zvvo4Xd4hXzwHgtVZwwxbrFzrMJiGut86D6x0+sNXiqY0W+0lGN3A41w4YRG6d4WpIXnszQ6lv+w6ebYp/Vd+cjnQJbGvJawGDrzhS4BRzkBK34XGSVssCEdfF470eXW/THpbzngqE1roCPiiEaAP/Enj8z/qy+vf3lA1YB47+EsCHn35Si3NPoWodhsxm6J1rxmnICxBeA1Gm+E9+1Lj4nLnM9arNRVdhDXfQvXOUf6IZepDZ2EVFx7dwkn1EmaOiHnK6j1Uc0MjeQoRNqtYmhdViXirSUjNwQGQzths9kkLx5nGGZ5ttxY2T+TIgOKsUHd/hmfWYcVowzkrGacmjvYjV2OPv/9abxL7Nh851eG1/ytd3xpzphhwnOd+6OzLBu1nJyeGMOycLqkrRW39n5XfzYEaRlWw/scGgG7J3NOPsSsS/f2WXB8MFP/PkGsr1OJmXDCKHjm8h9hLKcx/CGj0wobGWxHJsillOuNZjfGOHYpbRf+YS+WS2LBbR5SdIrr7G/HBI2O8Y4+C4TblzA6REBhHC71Me7lDcvYZ97nEoS6rE0KeFExjn5yLFmg8R6YRWY4us0viWZD02nccjXZ+TtCQpFHHgm9zQqjRqSmkT+H08WzKpeRCfPNvBsyWv7U/52su73PjDf4ff6pPPxkwf3KCz+jkcxyLqb7P78u8B0L30ISb3r9Hafpyzaw1ulorPfHCD3xil3DwwqsqDScYnLq3wWM/n390csjvNWG943DiZ89xWmy/eOObKWoOubzq1hmtIcHcDh2fWG5xpuLiW4HheMs1MvOP5dkiSK149SPierQalhq/uJLiWxLPfT3zvX2MthtZ6JIT498DHgbYQwq67iHfn/51mA94XQthACzj5s77f8kgLFXQM5z+dLEFHlc6QgFWPHLJt9P9V0KLKtdlmhG20kOSlaev6gYUo5lTKBPk6eWIIOrXRrUgDtCoRMl+OMMcLU/GbniQVHl5Uz8R2ZPQUNVVgELuEtV/Zm0czQsei61t0AoesUjiW4EI3oFTwj/antDohG02fl++POJhkfOaxPr5tcZJkFJXxm3AD2+gGKqPDaIUOu6MF4+EcP3RpNzzW2z5JWnL1/phPPDHAtS3uTTIGgc+GGjL2ekyyCrexijV6gJiPKMYTillKtLlCPssZvnmH6YMEJ3LY/9p1dKVQlcaNE9QXfh/LN+2+5buGgp2nCMd9R1ilKuzeujHvufmaEXdlC3RZYLk+UprEb+2YDUcsS2LHRQOLUtMNbE7Skr2p4TZEjk8jaCHLFKUjtNdgf1Zye7ig4ztc7Ibcn6TcGxtznKpSPPbZHyRu+Rw9mOB6Nl7gsH/XaDlOmZSnHp6zw3ucJE9RZBX/1//5BVYuXGBvf8rxaMGHLvUIHQOalnX48qIw719eafZGC/qxx8c2m9wdS4aLgif6MYUy9vuuqLvcMODepGCalcwKe8mbuT8t8Goy2yQrCdV7ByH+2oGUQog+UNTFIQA+jwEefx/4Kcwm4z8BThGi36w//3L991/U+juURI0BrWqzGKEqxNbjYNmoOmpP+BGi0TWtbZZwqRmhrJjCjUkLRVIobp6kDFYUzvEtNjefRlQ51ugI7Xgov8VcWbhtk2s6KxRSmHbnq3fMRfb959vLAJkLHcMqPNNwuXo4x7clzw6Ma/GiNBF/RhFoRpO12GMtNkVJA5fOdlhv+wwis8KslKblOwxiyUc2m7y2n5CVio88usKgJkhd25vS8G2+ecPU02cf6fH1W0Mq5XNhEHHnaM79kwXf82ifS13frAqLlP2yZCN2EIsUkSfGqcmxKaZzpvcN+9Jr+oQrAVVeUeUVlmthWSAtia4U5SytA3VdKAtUlaJVRfr2deMhubqNbLSRXrA0mz3lSug6rk6fUuCFQGYzs62wPUIvRrly6e+4M0npBg6VD52ghXYj4zPqtYk9exmjmNS5qevtgCtXVrm7Y96n04R1pTRRy+P4RoIbtUj2byFtB1UWlIuEOzeO/z/23izIkuy87/vlyT1v3rVu7dXV1Xv3bBgAgwEGGwUBNBAESYUo0VosKcJLWHpR6NmmQmGHFfazw6EHy5IdDoaWkCmZDIk0TCwksRLEYDBb98z03tVVXevdl9zP8cPJm90AIWLmgfQwYs5Ld9R662bmOd/3//4Ls8GszDZVnO5PyOKIf/ClKxxOEr6zOyJ0LJ5Zq5PkOqz5ZJby/HaLhmtV2p6iTBJfdh0miSQvFL7tUuTaDXy5piMdQ1vwyTMNTuY5cVJUf8uTnpg/a/15zMVYB/7PEocQwL9RSv0HwzBuAP/aMIx/DPwI+Ofl1/9z4NcNw7iNrhz++s/8DdEEa7hHUVvSDLu0j3r4FqdXfp5eVHDVuqcVhjVtbmoUKeboAOWGCLfGINcXKLBNjGyGTCgwJ6wAACAASURBVCIyCQU2dmcHpxzHuXZAUob8OqaBV/oqnmv5+iYwiqpfPJ7lZTKXzvk0DYODhotrFhSKCg84nGVcaPtkEt45jdhuuriW4JefW9dtR5JzMo7pT1PmJeA4z/SJ4lo6T2ORwbEUaklys+5y9UqX57da+I7Fg9NZGfxS49JanZ3244AZZTlcrDuIaISIRyg7ID+6weywTzxKGD0YYZgGTs1mfhpRWw0QpsDr6KR003fwl3QkXdQbEffHeJ0GwrFJ+0PsmofVWUalMfnDW8gkwlrerCoLc+FSJXOMbK6vn6VfjxJmSUZLqDs1Qt+jZnt0fJt/9dojPrXT5qWtBl46KLUZUHdMjqcJB5OEbmCT5JLjSUKa6wrLKlWjWanZGB31MYSJ5YUYkz4yz2hsXsZ0fIYHR1iOT5FGjI6OeeaT17i0FvLy7oBbh1Oe226x1nC1n0ehsE2hU91NwWv7I+qOVWlD3jyesFn3uNwNmOeKN44nlSfIQpLuWmb1mk/nKW3fRko4mr03y7k/V1RrpdTrwId/ysfvAi/+lI/HwK++p1dhmvoGUVKj8I6H8LUS7tEk5przRLahqctWM56g8gQJpKpJLvUubyiJIcyKZTjPJF1bR6yZKkcqPaEIbUPrA9IZTbfDKCnIDQvfkjhmaWAqwBIawU5y3YZIpU8v19T/5lLRcvWI9GiWsBI6WKbiY5sNHowS7g8eewEkuR6JHU9TZmleTS8Wc3LH0lbundCpvmel7nI4fGxZNpxnrG47j+PZhKUBwmxOcfgAUTpEW55DMk7w2h4yK5g8muI2XfIoxyp9KXTF4BL1RhRRSri1jBV4FFmO2ergLOLj/Jo2mE3jx1b1jqv9HBwXY+HfUaT6GpaOYEaeawBSWLo6jCfU/SaFB5/aafP26Ywr3YA1v4k53Mf3DETpEG6bgtCxAC0lX2m4TJMc0zDwAofpMML3bW3FF3aQeYpb75DNR1h+SDYb4dQa2vZfFixtrfPcdpNra9r81ncspnHGwBJkrn4vX9sbstLwaPo2W+1AHwxCU/fbnsZRjqYpzbIF6kcZd/tzhKHB6sCWVcbsYqx9NEvZLwHad7OU4ses8f7/Xu8PJqXtUdRXMbJYW4XnKUZjiQuzW1ywJXI6pxj1yH74e3if/RVNa1VaIi69Ov/k6/f58JkWv3q1w71pl2GtxdlCkeSKaVYwTky6QZ32YJdwaYdpJhHpXG9KlseKaxM6JuOkoOmatFyHpNABryaS277NKM50xWHAo0nKz1/QkvGoZFNaAn7ubItXDqaczlPu9uacjDUw9lc/tsVfed7jVm9OYJtcWgp4OIq5eTxhGue8cLZN27eZJjkvndU/dxBn3O/P8R2TtZbPJ84v0fZtAluwN465suTBVOMoU2nRsLQiMz+4T9E7ZP/b15FZgSEEwwdj3IaDE9pgC9JpRh6d4jY83FaIYYoqZduua8LUkyYnRe8QlWmm68JDUrna7IX56PEo1PHACbXmwnq8qRuL+IIiR+Qxbb/JS1sNrnQD3jmdI7o11tGltW0a5cNmcjxLEYau1pYbHm/tDukfTStNynSaaKm6bZPHUwxhsv78Fzi9+QO85jJ+q8309Ag7aJIlOS/f6XMwjLmzO8R2Lf76p3a4dzrl7vGU8yshfklsW4jq3j6dYQuBV45E1+suca6v9fEs45VHI1ZDl9AxKzOhwBZV2njHtxnH+XsmSn2QzfmTS5jaKdiro5xAz/NtX+cj5ilG18EKW6X7sUuBgPoqCogLxUrDwxIG80Kz8UxDS3YzqXg4irGFoFAObnuH4bzQDDfL0UCZyEkLWY6ytHOQiEaYtWUKBVGhDV4W7QZAw7W0ixIwz2RVbYwSyZ3BnL3+nDSXFFJxtqsfuH6U8frDIY4lGK6E3DqccDzWpWd2RpHkku/cPCXJZWVNt9MJmGcFjin42GazBNbKoBuZaqoyENoC0tJCbjJgdGuXaBAzPZpj+xaNrTrJOCHoBkSDGLfhEG52sWs+pueQz6IqZdtpt3VW5uYF8pN9HXazyK2w7MpDkiJ9HLLsa1s69UT6mXRrOjvC8rQGw7QQyUQDkk4NLx2w5jcR3RoPxwlrzQaFVBVIN8glbf9xLohvm9glMGnZAi9wSOMM23PJYqdy/h4+fIt0OsCtd0imemqRx1NayzVGvTnxLMUQBs26y+3jCdf3xjRLI5mDoW4FfdvkO3d7pXO2x0rDpTeXFZ/hdK6rx+E8YzV0yyBng6zQDlgLw+PTecakFOK926VQH1QQP7lyBMeJSVoYzPM68xKcWq4FeL4BPrhtg3D7I6QKolzSi4qSoaf4ixeWyArF9ZM53cCh5VnMM8nRLONrN08IPbtyfw5dC9s0uH085dpagy9e7JBJWSnuzPQEc3qKDJfpJwVvn0Ys1xwc09EMuhL5fuVgwjzTeQnbTY+sUOyOIv7t93YZnsy4enWZrY7PJ3Y63Dqd8Up/zje/s6uj523B5KSH6fisnVvieKeNZwmG45jreyOeXqtX7tkAqzWHjdCqKpoCgYj0KW2ODzCnJ2Q3f8TgldeYPDzi+I1Djm8NWD7fIlwPsX0L0zHxl1uY3pzaWgdZSJx6gL+xUl0HQ5ja1HbnWhmkvFllbuK4uiowhN7M0znKqqNKk59F6hlo4ZVSCjc0MJIp+C3NcxAWyq1rI5kyiWodWGs2GJh1TGClZuFaNYaxNuKxhWYnns5TXrrU5fvCYHspYDTPOO4rrj63xizpMjjaZDZOCFseMv84RSGRUmGaLbZ2nmWr4/PVr91mcnCH57/0aXzH5PremONHYyahw92Gy2ieMZql/NGrB9Q7Pl94foPRPOXBqdbPvHJ/wFLZ/g3nWoB3ZSXEMzUlfJIUHE8TlmoOUil+9FAL33znvVEpP6ggfmI9OeN4shhbTBmqTxcppqmTrNdqBrGnVXVaQSh50rjHKFV0oWdXb/j5TsBO2yewTV4608YAxonEFqURNvCjtEMiWlzNJIXSHg91V2CX6sStmo2jUn79h6f4jsWHNholOUsRxjrTMkty1lu6xH7zULtPz9OCv/FLV9nvz3lrd8hf/vlLrDRc9vpztktm5dm1Oh/abuGVuQ3dwMIprfT8dIyY9cjbW9jRCHX9mzrPMp4jZ2Pm9+4wO+yRzWLyKGfpbIOVZ1dpXtikiBPcVh1/YwWzuYTKMgxbP8wiaGCun9Vp2qA9JB0fCZXBL6XAzRWKRGpBU+GWVnr5hBNLS5odDBzT4GSSEtga67GFj2GASsH3ljma5TTdFgjwvfJnSYWJ9pWo9W7R8eoU3VUKBKOkoGYLRokGE18409JOX0nBYJ6y3vC4eTzhVs2hN4pZafu0ApveVDtez6OMeZSx14fVsy3aa88z7s8ZHs+YjWMsx0QWkluHUwqlGBxNNRhqm4xK12x4DEpffziiyCWOZwEeg5I1e/1gjCkE600Ps7QdnMQ5F1ZCVuruu34W5J8iBmEYxq8C/x2ax/SiUurln/U974sNYrFMYeBaoDCZp9pU1CwdohRo3oKSmIbAIsNx9Mhxv3xDPUvwJKtVCD0pKKTCd0yankXTtfAtTV5JckUvyjEdvQEgFTeOp/SjjEsdH1GG8ZiGUW1cjgARRez1I5ZCh9DVLkSU19RyTLyaU/azJlH6eMR1bSVkqeaw1494qcQdbCFwTQ3K/e0Xz+BaGsUHbV5iFzHS8SACxseYpaW69GoUJ/uaEl0Ch3bNY3YwJI9zmmebhJtd6jsbj1O90V6Uhq25DpT+kFqIVNNAYzkgmQmfwDKYZJKsKLgziFkPHVxLkJdUYmGAI3wmqY64Mw3t5zCIMqzQYZZqXoBtGlU84f1BROha1B0TUTIYldKVQ613i6K5gZj1ELMepszphMuIeIhnucgln2kqSXKdwl6ooDKcBbi0FlL3bPb6c0xh0Akd1lq6/WwGNi8nObZbMDqdk0Y589GIoNkkdYvKoCeeZTgliBulRVXNOpamWC8s7ETpPDWKMoRh8OB0Tt2z2FnSLWVF1S+t7d7L+lOsIN4EfgX4X9/tN7wvNghTQNMV+MkQY/81RHuNV80d/sM7JxyPY37l2XUC22SmQwroeIKgv8tx/RxHs5x/+r37fGi7xd98dpWs0Nz5Qik8y2Sz6VWS4TuDObd6c4Rh8LeeW0UpySDKaPt615+mkr9x0cecTFDxAdKtY3oh/+6tE17dHdAMbP7bz52nlsd84uISq6HLM8s1vvNwyFeuH3HjVo//+heu8NRyyO/f7bHTCfjihTZ/9GjK9aMJ//2vv4Jtm3TW6/zjf/cGtmvxzE6b3jQlzSVXN+p8+twSn9hqcncYczI3kcoilxmhE7LaWAHLqcp+0VzSobiywFtZ5tG33wRg/aNr+vNllWAubyJqDe1mLUxE2ELVl8lbG0TYhKNdcGoY6QwZLgPgKLg9TOl4+kFueTZHs4yszK1wTYNBlPPKwbjykBwlOXEhudWbIQyDaZoT55KmZ3MwiRknOW1PaxWOp0k1rcgKiWvV6Hh1xKzHPXuDC+O3yO7dqHwqCDvYjQs6NmAQlVMOE1sIBpFuAV7fHRJ6FlFaaLCx4XIwjNnq+DiWyRtf+y7j/ZtsvfhlJgd3MV2f4e5b+O01zjxzlfPn2piWYNKPgJg93yYqN75rGw0enM65//pN2ptb+KUr2Fde2Wc2TpC55MxOm6eWQ271tHT877y4zW+8us9rZVTgu1lSaTzqT2Mppd6Cx1jau1nviw3CQL9oJUysehulJGebDufaAZ5l4ltmBRCpEoNwmxukuQ51/dB2i6Zr0Y8K7f4joCjbjUUCljA02Hg6T4myokTHIXQf7+7mgq1dBr9gOWSZohvYPL3VpFn6Piq3zno9Yr3u4dsGHd/m2kaDNJeEjnYoOr8UUHcsDks+RcOzsEugbaPtE5U27ItAGNAU4EwqolyyVE4sbvdjXjucsF53+fLGEpHTwHMmCPOONvSVUuMG55+mSL+G2/BoXzvL+N4Blu8Qn/QJgjqATtqeTzCf/2wFKPpCwaObmEEdI9DcCFlbwi5SQKtFm65BlBd0fJ0n4Vk6t8ISWpwEVMh9M3T5wb1+5efgWjqqb3GKXugEFEpxMEmqacUglwzjnKK7ipj1uDB+i+T172Kt7+hpihCo2hL3hwnHs5RBnLEWuqyGLq8djnnncEyUFgxmKQ8eDKl3fExLcGmtzl4/KiMGkkoGLoRBbfkM6XyE11yu3KdOJwmn+2Msx0RJmCU5MpfIsnpYnOyNjs+59TquJXBLILXIJaFnMU1zkkKWgGXGVicA3oPt/Qdjzp++lFLkbgNWryAmJzTvfJPPX/4LHM4yarYOxFm4Sp/Mc14d52w3bWq24BcvLzNOCn7/fp9PnmnTKm9eYYBnaTq0KQzavs3BJCFFz6tdU7Ac2IiygQhtk9iq4S6FoCTzQp+CL241cISBZwk9AZEun9jSjkRhfMozKx3OND3+6tOrvHE8425/zqfPtng0Sfkn377HR862aXs2G9stmoHNZy53Wam5jOKM33vnhP/kqVW26h79OKtaoI5vsDtKCWyTj281Odt0uBcVNIXCchu4q+dQj25jrp0l332H/GSf7jPb5HGKKiS19SVq25skpz3y/okOxlnfwdy5Rnr7dRDXAZ2NYazv6HbDEFofISykW+Niy6FQ0I/LIF0liZRLLlWVQv2htTq7o5huoLGhQZTxX710FtAt325vxrlWwFvHUz51ts3eOGaaFnQDm9CxOJ5pQtE80+CrKfOqcpg+/UUCy8C88Q3M0T7N4Cp116p4BwBnmj7XnTGhZ7PZCbi+N+KwNyeJMt5+NCbNJa8/GDLpR5iuD9MB09NH2F6NWmeFWsPFEAaj0znzcYIhDGzXIp6nFIXE9SyEJfjtH+6TJjnXPvUsP3dthaWaw92TKR+/sETgmEziDN+x2B8nhI7FqMxRuditsfYeMAj4mcE5XcMwnsQO/mmpa9LX0zC+Bqz9lO/7NaXUb/2Uj/+J632xQUgF04U81nBwG1t47TMM50Xp1CzJ5OPSa5IWTNKC3VGEa5lc6njYwuIzZ9tIBYO4LMENeHqlVpa7mt1mmwZLgcOsDG11TYOOo/SDAajchqIkQ9lBxawsFKSFwkBXGiuepfn4pkOcK3rzrJJwF0px/XhGoRSfv7pCYJtIqfgvPr1DVkg8y+RH+yNuHU44GsUMooyOZyPLvzEudAUROibzTBO3PEsQ2Lq6aLpCjxhlgYqmiHobMZtg1XzcVl17T07mxMcneCvLiHoLs71CtncHUWtg71zTAA1oD46go3ENy0UGbWZmDSkVjfkRU29Zb8yyIMJmf5LxvYdDrnZrjOKcpmexXnd1uG6u//Z3TmecbwfYpsHlpRod3+KZVX3iPhzpXl9fy6TCjGxhMEoKOuFyVTkElkEqwRkco5II+8I1nPJnFkqV7c9jw500l6y3NG19MWZ2LEGz5pAmOW7YIY+m5NEUVeixaDoPEZZDkUYIy0FYNnlW4LgWtYZLnhYUWYEQBpZtUnMtbh1OOPQsQs+iFdj4tslxmfoepXllUfj67pBffn6Dldpj4tvPfhZ+5pjzVCn1wn/sk0qpL7zrX/Yu1vtig1DoNiOVilQpFtL+tHicdCyV3hik1FRUKRX9KMM2C55b0WCVY1qMEj3+1L2vRdfXnPq8fPgW0l7K32kKAxENEfOBzn4I2trQBDDR/oQ/uaGbhqFDfJREWS5pqiW6hZIl9iE4niW0PZudto8sMZGmp1uUk1nKaJ5yOIgoytdUKIUon5YkVxxMMzq+ydm6iZgPMIZzNtKZdpeeJeR338DwAuSoR35wn+Fbdzh57R7tKxskgynLH75EEac4F5/DaC4jbR97+xlk0Ma8+wMt5y5drEU8QnpNisYqE2mR5RLPEpzaXfaGWlh2s5ezXof/99YJb+6NuFBuALYQHEwS7g8jskIbzG7WNaZzPE1xLUGcS272Zrx5NGGvP2e95XM80QzJxXTg0lKNmi0QcdmvC4F54xs4g2Osyy8gH92m6er2TSFRimoM+tx6gx/tj4jSHMcSfOpSlwe9GYVUTOOcJJec7XbZfbPN9OgeWTwlTyPM2CGbjXHqbfxGhyIvUFISTZIyS8NEKlVGHOqb8nCacmoLvJpDp+7iWCYEeuyZ5pJ5WhClOjnt8GjCt2+fcnmt/p6ehw/GnD9lCUMbvRTyp5dYBjoti3JiIIRB8YSqxS6j2kCXn4v/GwaYTwxPRUmFhXJ09eTFKNWdhlLaDOXxryu/HpDlz1qECytZYRd68xHksqiyEDzLJM4LBDqKL5OK0DUJPRvHNknR7MGFxbtrCQJbcEZMMKZT8vaZ6uUV9VUoUkQWaeDRr1HUVzAGJ+RxQh7nHP7gPp0rK+SzmOYnf47BuU9V3386L6ilgrX1yxXJajEZVqbDXNmkhVYsJrk2d/k/vr+LKQyurjc4nWdVL/7q4ZiaY/LG3iHNwOHSco1vvHWMKQz+yoc3y+tgcONEe2jeOZ7y8qsHFIXk6adXSXPJSsNlueHhl9jQKHHxLBfCDqq2hDnaRyUR8tFtxMZFWq6AIiV0bKKF34Jjcjp3ONP2q0nKpaWAUZRx53haaVg+c7HLbzk+Mk8pEl0tKFlgAbYXsrrTYtybE88yhg9vo2TB4IFJ++wl3YbkBkmUMTzsYXk1XN9iWFYGC7Xt4p5Kcz1pcTyb7/9wnx+8F8MYpUjzdx+0816WYRh/GfhfgGXgtw3DeFUp9cU/6XveFxuEWpTvhtY+IPXDfr8M0BWGwUrN4WJHe0JK9IYQ2HqsOUr0G1qUyrlpoglU2iFZA0SBLSpxTaGofBwNA+ZOC7OuQbWlst0wEJrANUuqUd2Sb9P1Nb7xcFZHKUWRa4uw5ZqDYUBc6JtDGFobIgzwLZO0kHzl9ilpLhnOM9abHl94ZpWDYcT37/V5WRj8p8+XD5YwSP02+G0smWMkU4zhAcLX2EjevYilJHlzk8xr4u5cpRvPMG2rwiD8tS6G49HMJ9zLAgZRzvNdk1yY3B63GZf5FXEuCWwtMpqW1cLpXAvLkkJWhKDFhns8jllvebz9SDMQz6+EHI9jvnnrFN8xuX8y43Zvxpt7I/YHEdtLAestn9v7Y+58+6tc+YtfYndfm+BMk5y3StrzS5e6uhVb8rEbF7g/TGgGV7EvXCvp7wJ79xWye9dxBscE6zuoPMM+c5lnNj/MmYZbbdS/f39IUm5AgWOy3vIZRBnpfITMM5YufkQrVUenTI/uMe/tM95fw/JCTNejdeYi8XhIFs9IprOS2m0yOjrm5MZ3Wbr8Ams7V7Bsk9DT7lQngwghDM6uhrxzMEEIg6tnmlwvJNl7UHNq3c+f2hTj/0b7ubzr9b7YIAAkChOjjDhTSKmR8cOJto8PbBPXNFgEFDimoSnGMqdftiJS6anfIhMzLiT745g0l2w0PDYbXollFPRK1N00nrC0k4qW6yLKSqJQenSnE7wFTbcUOSHpRVl1IVueXYGiQKUf0D/foECXqW8/GjOJtUhrs+WzUcq8X9sdVnZzmdTOVa5l0C4miOmJ1jF4gU4C6z/CdA51wG48xhKmdn4SJq3L28S9EbPDXsV9EPMBmemzO4o4mAo+va2t2CepBtEeDiPW6i62GdLyLG73I+4PI9JCViQh0Mj6D3cHfPeNQz7zoXWOxwnH44SDoY7DcyzBcsOlW3f5w9s9dh+OmPQj5EWtabn/xi5ec7nK0rQcE9Mw6B9NcX2b7wujJEFp7sTxLKXuaqJYJhUUKdm964iP/SLm7msAqCRGBS1SqQ+XxTkty1P4yVJ9nhWkE22x7wQBSiri0Sl5rOnY06P7uI2u1nCEDsnU1DGBsiCdz1HSr76/vrLB2Y0GJ+W9WUhFluRVOE8SZYiShxGGLlE1HvvZS/FBi/FTl3iiDbDNH1c5FvIxFiEMg1kmuTOI6UcZUsGVboBRci6F0CO1OQVxqaOwhajINV95W6dj/b3PnKuQcNDjUFPoTcKyNQaxsGK/cayJS0239GQ0ynaibCNWahaTpOBfvHbAmZaPb+sAm0VmqFDaxHWvH7HccPn81RX+4OYJvWnKZ68u86sf3sS1BG8cT2j7Nuuhw9EsJ3FDNsQAI50Rv/E9DMfD3r6sZdXRFFrrGLKgGBwT7T/CrnnaJGZNZ2uqpTOIZELDN7nQCTjbdMil7t0XAiRTGNw6mfHygwGX1upc7WpQd1Emp2UM4Xx3SJpLOks1vv/OKbd/+A7Tw/uc+egnWdlqcnWjju+YPLPV5De//5DjBydk8RTTETy63deA4GykQ3ZL2rsXOCipqefbS4EeEeaS+4OIQawZipeXaigkoWPjDI4xd1/DWN7BSMYkr38X2/VwO+eJsgJVbhJroct+4JDkEt+xCEtvhxe//FnS4tNc/9br5NGMaHBYbQq15TPYnotlC/KsoLnawXYthsczsvmYLIbWmYtsPv008SyjP00ZHs8Y9+dYtsnGWr2KN1juBDr8KC24uqHxh2+9y+dA/WyQ8s90vS82CM1RMJhniuN5zijWp3smFb6tHZuarkWhtJhFGND07MoBGfSDPU5y6q71Y8w10zCI84Jc6pal7tlcWguZpgXCKNl/jlWGthqaj2F6HM8Loly3K3VXtxVuycA0DaMEUAt2R7H2M8wlSzWHg3FMLhVn2z6TtOBomlKUAqQXzneoe9rD0BQGKw2XlZpmY0qpCB097QBouHpzkY6PkU6xd67qsOJQt0LC9ZF5gipSzYOoedgrG4S2hbWyqb+2yFB2UHpDWvTjgjhX3OrNyqg5/f46DcFS6HChHRDnksE8JfQsJnFe3fSgmYVrLY++JQi7qwAEoUvo2xwMdaX2R3f6pHFGNDiideY8pikwfUF9aZnJozs4ZRWWZQXTYYTtWjr9qsxB1d4K2oBngSUppVWzQUmaMpIx2dpT8Oq3yPfu4J5/keUg0LiKoZ2cCqnYKKuVp5ZDBpHGT8wSQ/Lbq6TzEablYNca2J4+FAxhYFqCPC2oNVz9/1IyPusfE7a26ayHhJ7FiVLVweZYgsAx8R2TSayzQdJcloStJ+wK3sX6oIL4iWWqgkY2YkqdX//hQ771mo5O/6WXznKpW6tCavOylDQNgyVfZ1UsqohxknPjZMrndtp0fG300fRsPEuLaG72Zvz2G4d8+dk1zrUCXjkYE5dgUOha1BwHv8Q0Ugn/8nX9GjabHmeaPh3fxhI63n4RxZbkkm/dPuU3pynrLY8vP7XG//SVdzjYHfJf/vI1+lHGy7sDrTUQBn/vE9tEmeL6yZRmYNMKHJ5fa7A7ipikBVsNj6NZSpQrzoW2NrrJYor6KqawkLavJz7DA4rJUI854zmquYR74Slt1Ds4Rk6GOBefA5mXiU8ag5lnkmbNpB+5ZKV/RmCbeKaoMJSv3+3xyv0BHzvf0R4McVZVE8sNl09f7NKbpziW4HjQJp6nVcr201tNfufb9/ECm8/80qcYRRl33ziiSGM++Rcu0179PK5vV5ug79tMpwlprIVXg3lKoQJCx2Q11A9sx9OjXtAZriqJSV7/Lrz6LazP/S3E9ISJ6ZMkui3LpX6tm+Xm0PRsbvVmSKWJT0oqNq9cYDqMOXz9Ef7SBm7YYT7sYTk+vTt3uPzpl+gfzXA8Gz90UDKk0Q04vP2AeJZiuyZ3bvcQlsCyTRzXKnNDPS4t11gJXbJC0ptnXFoK3pNDlFKQf1BB/OT64++glKri1GdS0fZtrnUDlIJxVvD26Yz9sU54/sR2m6yQrNQcDqYp/ThndxSzWnNY6fgoBXXHYqXhMk5y7pQmLtoY5nEFMsukBimTGZtNj4Zr8eG1OnEheetkxj/79n3OdjXo9p89v07omPz8lRWOZymTNOeVRyNMYdDsBlzt1pikBcfjuMIjvnqnX1rVSV7fHTEcRnzz7RN+8fkNuoGNMAy2mx6XaznScDDiMSIebtjq9wAAIABJREFUUTx8i2zc02PJPNNGsvW2TsASJiKoU/QOQJiYzSXkfKK1FsIixWKc6nSntFD4ljaPPdt0mWVSTyYMSEv2n2kYnO0GZTSdxTffPqmCir95/QjQlcTe4YR4nhE2vSpl+2QU44cOXs3h4f6IwdGMlTMN0qTG628cYtsmR7sjak2X0VEf0/FRssD2XK4+t8Z6wyvdowWvHY4509R6GFvoOAL7zGVU0MJ2PfK9O4jpCXlnm1oyxPTauLm+rh/bbPDq4ZRBlHEwSdhp+1zrBvzGy3v0ezMMYTA8eKj/lt4jZmEHy68RDY6IBof0DqYMHryDZT9FPMso0pjD25qFubaj6diP7vZZWW6SJjnxPOXekZ6YjOYpV9cbND2LYZwxz4pqavaungRFtYG+H9b7Y4MoDWWTmfZ2WFvVfdtay6cd2Gw2POqOVm5KBdOkYFoiw44lqv9LBTdOpszSgvW6yyjOOZhmTNOc/UnMg9M5jiVoBZpHbxoGjikYxVl1UdZqIa5p03A1nyEt9Mw9ySVHRxMms5S9fsSHN5uY5dRClFVNrzxNmyX9OLBNnl5vlBwJxR+8o2+yVmATJTmykPR7M028Cl0OpgkdT2dMGKCVlKlDvn8Hw6uR7t/ADLXuwVs/R9E5gwzaiP5uxWkoBsfMHx3jXs0o6isM44L9sY7nk1JV8XKLkaueuGhA+Bs3T2gGDgfDmPWWVm0tXK8A0iTn9bdOWFmvk2cFeQli7u6PeOXf/iu6Vz+OG3aYnTyksb7DrH+IH57l+O49Rrs3qC3rMObeHW3u4oYdhG2TxQ6zpMvN4wm+rbUV7xyOue6M8R2L59YbnM4dntn8MKlUuJ3zuOdfZGL61JIhM7dFONkn+96/Jz48ZOtX/i7FSpeDacpq6LAdmohoRDOwGfQNsiTDb62y8+lfZvDgFqbrYTk+/duvYHkhtmvSPnuFyUmPPI0whEl3e0sTqGzdRpy5sITvmIyEQZFL0iQnKrGVb75zwqcud2l5NveHEWvhe2FSKuSfJ9PaP4uVKoO9uT5ZL3drLNd0mvP5dkDbt7jgpeSOx9u9uNqRP7rR1DJiBa8fTbS8Nin43TcPGU4S/psvX+V0nvGvf+82hVQcDiL27/Q53m6y0gn47NVlCqW1EHf68yqUZiV06Hgu42TO/rjg5b0RL57RZqovPr1KM3BwLMH//PXbZIVW9Z1frtEJHZZCl2sbDXKpeP1wwlbT4+fOtit9xf/+1Vsk5e/Z3m5xdbNBJ3R5ajlkKbD5xu1T1hoez64GWEpRMwRFbQl75xr5o3t4l55FxjNdRTgeU6/LOC5Ya29hp1Ok38bbPI9hfRejuczMrPGgP+cPHw45GEZ8tKy0DiePA4OTknEIerO9ezzlRz98xEFvzvpSwFY5Wr57PGUy0NZ3ndBhODQRIuf00ZhokuA2ukwP7zNK3yKPp4SrZ1FSj/iCzjpu2Obg1a/jNro4tSaWFyLzlDyeas7B0Sa3Sl6B72jBVejZBI5mnZ5p+5xpuBgGRFnBchCQJBLTaxNO9un76zQsm+h4QCgsln2rCkESsxNkbQnfOaTdCZiVRj1KKtauXsMPHWbjhOLyx1CyoNbwiOcpydQkm40xhElnNaR/NMV3TMZRxs5yjTSXZatoc+twylbH5/NXVrhXmgZ9+ak1DiYxdefdP2ZKQZF/UEH82FJKk5EW48yFUejuKGJ3BO9YJoE9phtoubGUmrRkYKCM0nlY6onD555aqcZzWTmqK6Qi9G3OXV2u5tZeudvPys+7lmCeFhWR6qQcs33+QpeD0kR1uQSbCqlIyhNZCIOz3RqbTU/Toi39wKW55HiW8gcPBjRcHeaTZxIlFUoqDo8m+I7JZy91OZqlHEwSPnG2zXbTo5lplqbx8A0MYZLc/BHZLCJYWkPOxthbF8lbW9SKGTUhEdMJpAmq7oDp4F79KNnqVUaznKzQQGTo2bzxaESUFgxLwlPds2gGDnVPI/0/uNdn72TGyrY2sT0cRNRLQ900lxiGQZ4V7B5NmQ5j4nmGaRoYQhAsbSLztJxc+CTTAcKyqzi8RahNsLTJ9OgexqSPW+9o/9A8ZTZO6I1iLq2FvL47ZDBL2ewEpLkkWsjJRekPYoChNObg5nOy7/17GpaN8bm/zer5V7SXRXlvpVJhWS5Inc4eZwXDw57+nSdjitWzJFGOaRo4QRPTMjEEZHFBfblNrd1ESTh5NKZIJb1hhOvbrLd87h5PORxpF6pRKRc/naeVo9ibRxNWag7r71GL8bNM4P8s1/tig1gsYVA5CSe5ZJTkpavxlM2Wx0rN1f2cWBAgS+6DUlUZvF6Wc4XSX7P4uO+YOmbtCVWeKFuMSBbVA6MnGfqBsH2DTmCxN4mJ84J6KSUupEJYAkMqTEsQenpyojMVBbagIkQdjxO2O772fHhCOVrkOo+h7liczlOSQnKpTMgy4ilFcx0xm2hQ0vFQ4xkqniMnQ63iBJQbokA7SOeZtngzHfKHNzHWr1FI/V4Etknds7h1qI1rpuV0YuGyZQpNPz8ZJxSFZKUTMJil5Km+Dos2QwgDJRVFIasWw6rZmJaopOW2FyKtFCULDPHjPgidix+hf/uVypo+m49Yf/4LDB++RdjyWGn71D2b0LN48GDI9b0R6y1PezE80cfrTUKQl+7P8eEh0fGA1fOvkJ//BEYyJckVg7jAtw0CU6e0TeOck70RRZ7iNzpkszHR8Ih0MqC+foEijcALGRzNOH7nh7hhh6WdC4QtjzwryJKC0/0xazttvnX9iKJsBaRvU2QFUZLTm6W8/WjC+ZUad0+mvPIge28bhNIBzu+X9b7YIAxD05DN0gWq5mjmoSZN/XGAZxhn3DiZcjrViTZPrdYrHOB4plmAB9OE4TxjpVGi4aHLlZWQBbv6aKqt1E1hsNX08KyFFb0mPJ1t+zq492BCoRTPrzW42PHpRdrKvhXYlVz7+t6Ib05TCqn4H758jbXQ4p//8BFpLgkckw+tNugGFt2/9DSZVMyzglf3huz1I/7FDx7yqUtdrTSdJtimwUptiVEu6Fx6gfSb/xfkGZbnUEwGuNdewFjeJvca5AosJCIeQb2DOr5HMeohggaZEzKaxLx9OqPtW5zrBLz1aEzgmJzt1iik5jicjGNe3415cH9A0HBplh6MAFNT540muWQ0z0iiDNe3+aUXz/Avv3KTaLhHo7PDbDChd/NlhGVTW96meeYqk4O7lbwadBzeeO8mQGVNb/khpzd/QDodIPOP0wq02UuUFtQ7Poc9DVJ/6lKXS0sBv39/iFSKtdCtDpGPbTbY+pW/SygsJGAkU8Ssx5I5YgkwpjPSV38fs7nEr33xF3jlYMzXbywTpQWH3YA8Lah3rnJxq0krsLl7PGOW5Jy99iXSOGd7NazarFbg0A0d9ocxgWOyVneZphoPW0yq2oHNX/rIBs+s1Nls6JZpNXgPLQa69Xm/rPfFBrFYi9LKMKhoyk+uxfumeQX6gizYjAvB05NrcUJCeUpW7EatkVg84LYpKi1HIbUeJCpDXsLST1AIA0uUJK4nmZLl+HVBblmMYRc+hI6lsxZSqQhdiyQvyIrHp7JujTTXw1tY6dsBdqEw4glmewXr8gtaSHZwF2NlR3s/KskgVtRsQW1Ju3wL28dqr5E3N+iXitbNhuYTSKmol+3VUuiQ5pJJnFebpGWbdFo+Wx1f6yNKos9WJ2Beio8Oag62qwVWtYZLvLKB41vU2nXq6xewvBC33sar2WRNnYKWjE/LIN0mze1rzE4ekke6DclmCz+GDkWhCVmmMHTFYAmSSI9YH/RmjErmapoX7AfaS2Oz6fHq4ZRipcuyrzeIJFcsmSOk30bc/kPS3ZvMd/dofvIcriXYbnr8tRe2GMQZfGQDgN1+xGbLJ7AFT200EIYOXd5sesTliLft2wgDnlmpM1st6Md6+lMozdKsOyZN12aUZNhCcHlJt4pGPMaYvwdthaKqTN4P6321QYAmQi2e88UDHpUP32LDqLIHXIuGZ1Rf9+PiLQEONAOnsn3LyocR9INecyxm5edcS1RgolLa8cixROWsbBqQS8qx2+PS/MllCoNpUjBzTTq+TeZYuJZgFGtnpU75s6ZpgilEpWbcbLhs1L1KS3A8z6k7grx7Hst0mLXO4pJB9zwqi1FOgDIEy06KUSSMDY9pJtkyLfLaFtINmY4ybFPwkfU6p/Oc03nKVifAt002Gx7zTKeCR2lOM7BZXg353FMrFf9goWLsBjajWFcSjwYRSirmacGF7RbtTkAhFd3lkGtPf4HeKCYtuQaOa5GtdvHbqzhByPr5NmfX6vSnz/LgTo/hwRFOrYHfapNMZ0ipW65O6LDScLm0VuftR+NqE79zPGWl4Wr8J5cVCWoQZRxM02ojH8QFS4C4/Yc8OPMZts99lPqrv4uxfolBlBG6FhdKt3BdXdksBY7etA3tGeKV1vdPBucEtknomtRdwVpocU651T1plYdEYAlMZZNi4c30SNg43dWxAO/lGfiggvjxJQydBD1ONMJ+OtetwwKUnKY5gW1ytrwp8pKYdr4T0A1sWp5FnCtOZ2m1ASwqip2Wz8NRzN2TKV9/e8Z//vFtNuou7/Tm7I9jXrk/wLEESaDLwY+shbRcwZevrFQy9MDWjL4Ho5jV0GG1DKOtOxYfWqtXZfH5lZAbJ1PuDQXXD8ZstQM+vN7AtQwWrpa2KcianvZRXK+zWnPLqY2ukBzTYMm3SApJigWdbVxVkOLgzU/AdBCDh2A5WppuWtQNSWoaSFdb0KWFou2Z9OOcW72I/Yme/jRKwDGwRSV0K6Qi9LKSam5ULditk5kumc+0mGcFb+6NyJKcVsPj/FLA3eMpUTlejmcp905mNLsBXuBwtDtkPpqytLVE1mgQj8fMxgF3c0mWFMwGMyzHxxAm01P9IJlmi3mUsdbyOBjGlROUYwmNHYy1PBz0lAMorewSVkOHuiNIpcK3Dd1W7N5k+9xHeTv2eXrnGtIw6JYuXgrF0UzfY0ku6QYOrx2OqTv6fc+kSVbICjCfZwU77QBbCPJCMVWSlqdfy8LifkGeQ0oUCkNJyFNUdxszeveWc0qpDyqIn7YWINRCRCQMDYpVJ74QlXlLzRZc7dZYCR18qzScRY8cA9Os/B5EKX7KpGJvMOeNd07pP7fOVsPlXMvHNDQnwRbaF3GRCQmwWbeJcx2KYwuDNFPVzfjYREYxSQv8koS13fE5GMdM4oy6Z//YZmUaVJ6OC6s12xSslDbqUlGRgrxkRDA9qfwmjP4+jl8jP9zV2RTr56HIUI11jFSTg7qWwIgnoCT18S3ctWucKj0JWsQDBLaJZwleP5xUKs12YOOUpcs0zulNU/3wp4VWay4FjBLdihS5ZJZoR2mzpCSncU6eSfJMTzlGvTl5JnHDGnla0FmtcW//AY67zuee3+Cf/W+/qzMs0kiTpIImeTxla+dZ5pHeqLY6fuUE1axpTcX5lVrFzQhdq/J+3Gn7mucwO8GyXAJTkL76+8x396i/+rs8vXONo/Y1lkTCfFAwSjJ2R7HGVaKM3jRhpeExTwsO8khnmSzVaHoWN06m1B0L3xbc6s0IHZP1ukeSL6ZkOntlp+VTc7REPi0kvp2xVlsjMBX2w1eRefqun4MPxpz/kZVJ9WPjHakUUqpqlOhaolL2NV3BxzdDpmVozTAuKju6J5dUmhm4WnPwHYtHt3b59t0NPEvw0laDhivInl4jLnQkHrnkVj+m79tcagCWzc2+Hl+NSlacPjH0w3QwjPjBvb6m2K7WaXs2t4+n9KYpX7q6SpIXvPxoxNVujdCx+J23j2kFNhfLnIfFapZZHb5lYhgg5gOU5ZK3z+Duv0Y2OKYYgEr1azFlTrZymXmuCJwA+/AGatyrTqv8cBdj9QqjWNKPMuquhWfpxKckl9QcSwfRlIYvtoDNlsc7h5MSuEx0eG6cV9kOKw2X3Ycj5uMJX79xhGMJOjWH/VmKZQu6G3WEpSc1jq8Dbvavv8O8t4+SBY9utfnNYUz3/HmUVIyOjlnaWtdVyXKNrY7PXp8yxMbkeJww6UekSc7ZbpfPXOwyiLIq33RQUuyvdQNENNK2/TIH08FsLtH85DmM9UtIw2BJJOylDjVHIgR85/4AUxicaflc6tYYJTmbTR3cW0hVJWsVUrHd9FmuOfSijKyQZfamSWCbVetlC03nr7smddei7Zp42QRyfc1kqQJ9t+uDFuMnVpJLXj+aEdhadXhtWbMCv7s75NapLnV7TY/AbuoTWBis1R6DjrlUSKXwTEE/zhiUF3O76fPChmYedgOH6G++xIXlkElaYBrQdE0+vB7ynd0RDwZRBRjqE8LDEjnnS0PZcWrzo4MxN0706/zsuU4lwlqUobYp+DsfPUNgC9qeyeEs43duHPFwEBE4Js+s1/EsPV4cRhnH45iHg4grK2HVQwe2SdHe0n/XOKMIr5Gcv4pnCb2hSZ03as0UkBPniquNdUwlUZZD3txAbjyLkcV8aDXQHgtKO151AqvCd4yyWnFMg3Ei+f7ekGc2NP/h/EpY9f4VizKXhB/d0P96diWnXoyOm+VU53iccO9gokN2n76C4z5F2PJ57kyTu8czDo8mnO5PeOaT13huu8nLd/qMenO++rXbrJ5t8XKS88bXvks8OsF0de7m7pttfsvxSecj0smAF7/8WUxhMEtyfuPlPZqBje8cEpUj3F/74i/glv6Y3ZrDfFBQcySbv/U/Mjvo8Q///j/SYT+2pc17hYeIBhSNdcT0BJEMdfXW9On7S0S55KnkAZgCddpHjnoYFz+CKHowSih6h4i2DjheCOrye9dRsmB8622OX37n3T8MH7QYf3zlUnE4TVipOSzXbDplfycVHAwjTKGnAos4eGXqB0WUpfuTa/GwTpKc9bqLm44BWK3VeWmnU9nOAVjlJgHQK7MoCqXFNXcHc7YaHhuhVpFmUuGa+qYbxznnzrSqfn0UaxCwUIqVmkXT1EE+rikYzjN6Uy1u+sS2zt3MCk2mGs0zRvOMtYZX/awklzwoMyQWvggLoGynLLEzqTjT8KpWR3p1KJZ1sLHjk0gDv8iwRUyrPAkLBcvluC2XqqSZG1gqx+DH/QqWaw6iDH+ZZwVZob0du2U7lBWK3iytMIK6Z9MNnWqy1J+mTCOt0zi3Xme95fFsufn0hhFZHHFpLeTaWoODYUw8S5kc3KG99jy2WzDe1+NQpgPyaMr06B4yT5G5VvmmxacxpeZk9HszBn2DdicgzgpO9ka8cjBmu+lpmbdUjJIMIWB20OPBN95m9Ut/hLr0IkqYmNNTlO1qC0F0cFDRWMOcHGMkM7zQIJcGatzDaC6j8pRi1MOZ9cju3UDOxszu3ie8fFlHFLoeyZvfIx1rrsqDr76K+Mmb9E9YH4w5f8ZSSk8LnvTYWJxmi+VbBr6lQSkWZCgJQuivMQ2DumvhWiYLUzW58JMogTgTWWrEBOIJvYFpGNW4ETRCbTxxvRYnMuhTWDzx8Sq2vQz4+clllhMQWZ7MTmkwYooFBqN72hsnU155MMASBp+8sIRXfp1niVIZKB8H+hiwO1MUssOd0zlNr+BSR1Hbvw5CYK1/TIOfAnzTYNHeCvQGaWQptunpUWhJPV+Ii4RhlOE2IKRRJZoLQ1cOi2AZPUIWCEOnZi8k4oWtWGlo38ajqTaY+cjFJf7Bl65olatS3NkdYgiD57/0acb9OaPTOVsvfhkhDKanj8ijKVk8pUgili5+BCcIuP6t1wHYvHIBQ2htxWycVAzJr99Y5q+9sMWFToBCsTuK+c79Af/w7/8jvTnkGTOvq8fdIRhFBmqgTXjkDDHrUdQ6WKNHhKfvUDMEcjbGSGPkZIio1Rn+P7+B6Tm4q+vUr10j3ntA8tbbhGfWQJjEvTHZPMJ0TNR7qQgUHxClfnIJw8C3TVxrEZKrT9NPnGnyTBnK4lkmDVcgMAhsAyudYtoehWVRd7QtvlKCQaQRaIC6Y5YmlJqJt9XwGJVaCGSOoSSm5dENHC4s69+zErrVOFIY0Iv0BjPLJEkhCcvRJTx2ogpdndpVswWeaVAYFhiQy6xKogrKkBfbNHCVllf7tknomHQDR7cW5Yk9TXOe2WrilGBmUvIm4lwH/bz26P9r781ibMvO+77fWmvPZ5+pplt1q+rOfXsk2SSbpDkItE3SkkNqYOAkAgLbcAw7QfwQwAjiSH7IQ56UBFAeEsAwrAAJICSBZSQRYkQSFYGWRJFUyFaTYs/dt+/YNZ46deY9rpWHb9e5txk22R2H1CVyP6CAOvM+++z1rW/4f///mH/zmU1WYo+8PpupEOcyyiqGWc16EOGyGd3QsGgGslQxxzMenmlYlp3F+RGqEs6Es3bdtKjYm+TN971faF1LAlm4ebWM1O6NMgJP02lQqIlvWG0FbLQCbp0ulszSK7HPp66tkfia/UnO6/tT4sDDDz267ZA4MJwezigWFZO9Gw2BSwtX11TFfQ5JZx3VYkbcP8f0NON07w5xT7gpzhCSi6Jm2NSMDmbFEg2q6gL32MeZRWuERjXt5JQ0BGd8zOxE1MUAnU+E6XsxA1uj4obgt7tKPdgjOreG6W8IWlRrXG1xtWVy821sWTE/HJJur+PHHtFK532th59KqLVSygDfAu45576klLoM/E/ACvA88Dedc4VSKgT+B+CjwAD4d5xzN3/oQWjFVhrSCaVi7GvZGR/b+xOK117AlQXe5gXMMz8D1qIWtRSkAOOFnE/6InZjPGoXN5GD7OrOaJwJCBw8tRZxb1KS1Q5VZihbo+uK6ysx59uyaDqBITCNtmJh+dqd02UBtKwdW2lIP/aWIBnrYC326KsMc/e71BeeZVwFFLVjWtQyNdpEC6EnTFOB0Xz0fJfQU2ynPqO8ZtFs7eOGlfuvX1vDaHjpaC5Q7MoyLWr+8I1j/vlvfpXHfvXLfHK3x2BeLklzytpyPJdx9qtPXEbd/g7x6S2iMJUL31aovEJVBTbuYqMOw6wm8jRPx3P09EhYujcv8oJS3BtnXOjG+EYW/mYqTkCYwxspgqYdeKEbCVYg0HRDw6vHGX6TkoWe5lIv5vHViLx2fO32iA9e6DHNSn7505eWKtuzccZ8NMKEUm9orWxQVwWmUe/ORsdko2MWw32K+Yj9774NwKXP/ALOOmZHY8rZmP21BD5ynleOJcwfLUp2e/Gy5mCUYE1ePp7z7LkWI6fQJqXTDVBlDlWGqgqqziZq+F2mz3+d9COfxJVSWFZRC9Momtk8wxUZ4eYmsM98f4DXiulf3yXspbQvnWd6Z/+9LrOG5/ThiSDeO1ke/EfAyw/c/jXg151zjwFD4O829/9dYOicuwb8evO892Q/cGxe6//HbffAE1WVS4j4gD2Y8ilnUXVBZR0nWU35/efe2XcAnlQTwZyF2ZEnbFL9yOeTuz2eWo+52DYNJZ68prSO3Euwm49TGRnVfvCrnKVHpnF8RsPlXsDFuMYgkPIlM3aTAhW1xdOKJ9aSZUSTVTWfvLLKr/3ql5caIF+7NXxAF9OjrB3feGPAPdfGXfgQLkyZRSscq8ZJeBF15xwuTCmtfM/aOWzcpV65gIt7hCc3+Vi35BevJDy77vPYitQ7zpikN9SMfmSIPMVz57tc6sUScVTSEm4VY3a7AR/YaPHm8Yy3x4LD+MqNIX/w1lAYozohq2nIW8fTpcq2FxiSbpf58T2qxYxWJ8SPRLfCC2KmB29xeut7KG0wXkC8KkjI4a3XiVo+6bmLmDBajqH3Y2k1D6Y5aWCWNYeonNAONM+eE3q906xmUToyFWKDGOfHS9FivbpF8uQHhd+z3Rct08kpKmljVrfQrTamv44OY4JOC+37uNqifY9sMAZbE/bSH3Bhv4s519RbfvDfT9rek4NQSu0AXwT+WXNbAX8V+K3mKf898EvN/7/Y3KZ5/HPqR4gBGq1Yb/l0Q0M7MCSeIjYKl0tbz1vbElXqIG3+ElAaVc7RxUxyfq0plLAiJb68j6+Fzr3WAaFRrKkFK7GhExicH+GCGLRHz825GNdspz5Ro4MxzGpOs3rJOnWmrVE7hzujxz+j6neOUV5zoDoU1hEYLaI8nuaZ7S7PbHd54lybdqBp+ZLueK6CumJSCVK0quX9J3nFYF5yklVLJiVf6+Y7eWx3Ij57qb9coKstSU+6kce5VsCFbiQ4gtpRNxofujnHZ47VjPYwkwNCm+E1E5IVIjfoTIMcHe/hHd8QgZ7GHE3Glk9IfE3qa670Q863w2WKVDbOMPUlYtrohMS+cG/sTXJuDxdLGcQ09BjNS+lCpAGtTkjSDYn7m0TdNZRWaC8gaPdpb11dIhKj7jpR/xwrlz9E7+IzeHGLOA3wAiODVysxt08WzMuafuyz0YkY5RVoD+eHqLok9eScjPKana/+t2yEtdR3TABegCozURhrrWIuPiXXwGAPe3qEMgbd6qCiRFrP2mAXM+osJ91ew29FKKMxkZAJ2/K9s1o7XJNK/eC/n7S91xTjvwb+E+BMAWQVOHXOnX3zu8B28/82cAfAOVcppUbN848ffEOl1N8H/j7A7s4OF+MaqgVqtkBVBRiPejzA5Rne9Uuwss0ilEp44Ap0PqM+ehuXzZhvfpDKOhaZ7KRnnYnQ0xROJOQ3TM7sf/yvWP3S36JavUThAlAQeKC++S+o9m7ibV/F++DnGeuEP703YlHaZQfh0BW8djzj2mqLtcRnpxloUkpRWxhkJW+ezPkrl3skKud2JqnSz11bFVEcFBuRo1IeB7OKSa1Z1DEvH0zphn5D5OK4PVrwh68e0Y08Rmkou7IRCPBZJHGa1YRGaP9/7toqo7zGOjjX8lhNPK6sxKzGZkk6Ey9OiG2Fba2i8hn1y19HeT7epaeJNp6gtE4KllWGMx6z3kXiYgxVxksjhXUZsa9JAw9PgQtSfA1xU3NJA5lxGOUV49wClWBRAAAgAElEQVTSaXd4YW/GvUnGhzY75JXIBmy1QxalzDaUTe3kykYqPBSdkNf3p0zGGbvPPEGRV4yO59TFgrizIroV9zaZHtxcEszOTwd4cQsviJmNc4xRtLeucm2ny3Yvpqwt39kfMy9qtrsavRg2gkdDnPHRJhUE5Bf/Q0a1x2BRshJ59HxxDKpcYMM2yovQ0xOM5+OqUgqWno+dnII2ci7P7VKPT6nLChMF1Jns9kfffpn1jz39HpcZEkGUP/lI4d3sRzoIpdSXgEPn3LeVUn/57O4f8FT3Hh67f4foCf5TgI8++yFHVaBsLTtWA1P1d69jVreW0vVFU6zTixHFt38Pf+caZmNHtDKcTGJ6FvJaagNpoEmGNzkKd3gt89j5m/8ZbnwXb3CTYuWK6IE6RXzhcdkRkja2yklbKTsd4bucjwSY88Rai0/vdnjpeMHtUcalXoRr2p8CFTdspCFtD/TpEZvpNo77kO+idrw1cWhVEnuarBJa/tAzaH0/1G8HHo9ttvn9lw+XYiy/9OHzbLeFr/JMCPdqP8Qo2C8q+pHs0F3PEnseUHE4rxgoxaWkRi9GqGIqi0N78OwXcGGL0gQopLuhqkwAWn5C7GYSScRdHkuFv7KoHT0yBlnI790o+MyFkp3U8MJBxnrLZ7MlnBf9Rhf1DM0aGc1OOyCrhZjHN2qp6v2du6fEgWHvNGM0L5co1iuX+xxPcubjXGTxqprxYC7DYJ21JcGsF8QshgecvPE89fWPESRd6mJBLxHhY6MU7cBjr1oICKqzJdee9jCzEzrdgCAMmTvNYFEyXFQsSovXDYj8BM+PhBM06aN6W+jpAOUHqCihPrgj6upViSsy4cu0Uqi0ZYXfijBpSry7swS4vSdz7i8kUng3ey8RxKeBX1BK/RtABHSQiKKnlPKaKGIHeLt5/l1gF7irlPKALnDyoz5EOSeOQT2Q9UQtdBDipkKgopWE/6ouqE6OCa48gw3bTb2gaWM2F2bRAKpUPkZHcGecsT/N+exKgj66gb92hdopPIW8x2rzmU3Nohv5lLVbdizWkoBocINuuItpjsNoISbRKAIDq7EvxVNnafma0t7HHHhayEQS3xB799maI3Of6+BMFnA1Dfizm0PhT1SC0itb96MZ3RQKha3aYZQm9hSqmBMCqZ9yZ5TTi3yoK7AV9cEdvI0dbNiW77gYgR9jog7UBVQF89YmISVmfICLuzilCWZH+H6M8VOwmmlu+eZbJ3xos4MZvs1Kcl4YwbWiHxliT7GoLLWjGQwL6ASa/VnFJ3d7FLXjaFZQWstGR7geTqYFo1nB8GBKNisxnub43rhJMXyctXJ/GBF11/F8jdKKwZtvshju40UpztYYz0CUcuNwxlPnO/Rjn7zBnNTWicZImaPsDLSHKnPCQIMXsBJ5LErLmydztFL0IhF9jrSWOpZzKGPA1uAF6HZPSINr4b1QgcFbWcM0YClXFQKa8gJMu/8elllz+eH+QmoN72Y/0kE4534F+BWAJoL4j51z/65S6p8DfwPpZPxt4Ew5+Leb219vHv8D96P6NrZq4MU+zk9wYZPJGA9VLCjvPo+ZnNJu8mMXJAS//KuUSLjSOXpj6VzG/avMrGW7OsJlCeXmU2xnE1Z2evz2awPceoArMoL9l3DTU7Lv/DH+R/8KtrOBco5ZvMZpVhN5jmsrMZ/aETUrbIXNuoRKcakXL3dVoxXn/BycZRwk/Kt7U/YmCb+8ItiDM7aswCiu9COySshx+5EhMhKBHM/LJb7/cFbwvbsjEV1pCGr+z5cO+CNP85eurUpLNDUczkqO59JtKWqJYjZbbRaV5d5pzhNrsQgLFZXUIXYeZ799mcAo2i/8NnZyKvT45y4D4nSDzaeYlB6d1ipmvI+aDqiO7qGe/DR+mOKc5ursdX79S08xKiwDt0tQi1J6bR3jouZo7gia9GctCViLDcoJu9YfvDlg/3TBsxd6hEbTjX02mnbvn76whx96BLHH5GSBFxj80KMqaxaTnNM7b9DbvUacBlRljfE01z/zSQZ7U/zQ0OpEKA3DA+Fz0Ep0SUpruLjaYqMVoPNT6s4mejaQ66vKUEqBs/R8D68bLF93d5zTDmJCL5B5F1tRp2uoMscZX9JgawmefA47PESnPZTnUw/2KY8P8FbWoCpRvo8K3gftvXPUD5GDeD9djO+3fwT8Q6XUG0iN4Tea+38DWG3u/4fAf/qj30o1i/CdLQYXpNi4e/9ZtkbZGpQmrx2llVYjzi5Tk6J2LEqHmRxiRvfkx60LIpfzmQs9eY/uOq7BAqgwWqYwTqllunK266sqE2Wrhh/SNanMmbYFgDOeODMgDTyRs6/cEqNwdpIXpaVqpgmF80KxnniC14AlPfrZZGXg6SXoqKjsUtnr9mjRDF9p0obQ5Wwg6+zY277GG+9J3p2NUFXGSmToVBPsaCC7X1nId9MedWsVnU/p1FNUPsHGXcpLH0M98UmK1jq6mKOLBVVvB5wlr6QwW1vIa8coF6HjXpNiWCtt16O5tKR7kc/Tm20+sNOjE3q8ciBq6t3I4/WDCe2VmLQX0epEBLFoZTgnY+N+5Eubc3zK6f5gqVtxcjBjeOtVyrwmmxeMjxccvvpttBY+h6ySGZtu5NGNPOlYTQ7lmtIaVRWoKkdPjlBVQeRpepHohL5yPGN/VjEqEbXzpC+bVyOqRFWi4xY09YL66J6kGVWJiUKwFrsQgWT3fha8++ksUgLgnPsq8NXm/xvAx3/AczLg33pfR9F48eVfY1nYxTlH3F9H+cES+4CTENZaS2AkBJT30UyLmpNFSXHjewD4FzPq/jbm5DYbf/o7qE9/mXLjOnoxQiuNv30VwgSMnIra0RQVAZSM7xqPhTPsjQpiT5P6RtCPSIFyaj0CozHOcbkXcqEbMsxqfK2WY8HOOt4cLkh8w7kGshxT0h7fZRLucjx/gOci8oitY5JVS6CR0YqrKwk3Txf85jfu8J9/6UlWY4NWiuN5yd4050InxDmph3iDm7i91wGwjQPUh7+DMgZv8wIqasnO1jCKn/irrFRD3Et/hGr3YOMyyvgc+2u0rCOcDaTFHPcpnOY0LziY5mymId3QEHqKni5xXsQbp6LqfZKJIxyUhtCzrEQ+n9juLGstZwtxUdR8/tnzjOaFUOrHPrO8oq4trU6I5xuGtwxlNsMEEX7o0eqEBJGP5z/F5GhAPjW01/uE6QpFVrHdjWgHQgX40tFUxry7MSqf4Y3eRlU5VWdTfpswRZULPD9iJTK0g5h+7EvL1td4WFzYQuUzVC7YCtVdx9UF+AG6twa9Nao7r4vsgOfj7VylvPkK1d5N/ItPvOel4H7aipQ/EcvnVK99C9NdRW0/gQsbCvcmfA4/+AXM9Aj7xvOiCbHzODOzIjJ52pHMTpeDMrvd82y0EjzzCdkZkz42auPCNuFTH4e6xIz3MNNjbL5AhTF1a5UsWReimHHJYCFsQd3Ix8UhdQWBcVzpBqhKCk7Oi3BIMbSwMubrkCKdstUDtRRxXoGnubaSLAlG7o4LXswqEn+L+Shf4i620pAvf0CKaXcbUdx509f/2q0hk6xisxeJQ9JSC9lMZfowa0hP2oGm6lxCtVbQ8+HyWA7+yX9BstGndf1xdNrDnNsVxXBnWRvfQJVzWN+mXr0gNQgT4HLHtLTU7R0SJbyXgYat1GM1PpMOEKWxsQugtJxPfYyCrdRnXMgE7DgTFqbbIzkWIaEViPZ6J1o6h3lRsyhrbGUJI4+qqLHO0b/4WCOkW3N6OMN4mjgNyGYlVbGgnI1p9busXrrKhXMpWXV/iK4deFxoBq+iVJEevyq7+/C76NUtbGtVuhVlRqQ1oRewnggyNhzeQjlHna5RJqsEddFAsisp+i4m93d2bahHA0y7L9239W3M6iYuPStwvTf7aStS/tjNVSX10T2UNuhGPg0kf3cOTiuPftyXKTpjMJXI253pZGCtYCaqEj09Iop72NYqzvhUgRC7el4A7bUlmrAeHsqHez4uaAkWwTpOs5JRVjbdB5YTo57WUuRqohgbtlFKo7Wm1hFFU3TU+RRVzKSfjkB4MR5Oe3RCgYRXVhS0b54uKGvHZjukHUg0kPiG9Za/PDdxk0IA3DqeUTfUcUZL6mIVRJ7C14JDQEn7cVpagqBD1KRk2IrpnUPKyZx4+zw66Qj6NIhRixH27Tekv7+2jU364lzRWCepS5HXxK1AojVbkfoe1pOIq6gltbO1EKd0VA61JfQDTjJHZWmiBsNwcb9bcaZfutEJuXV8HytglMJ6WsRwG5mDVifEWUcxn1POx40cXkpdZFIk1AZnIe0JDuTsnM3LmtgXaLv8xoqW0mBrps9/neTJD2IuBigvkm6Fs6hijm9aeEhx0sZdzOSQW2HIxagr11DDROaSLiqbSlEyiATpm7QldQ1F/Bf73nEQOPvTVaT8iZk2coLN/UP6QfAqV9dSc3jwMa3lB6keQFQqLQvg7GYxF9FbLxINx+a9lNewMTfISL+Z26gbPoozO8Pyv6PL4qSHYZQUISvroJZ6iFNaWopNy1Z5AE3dA0klQiMDTmddjOVAmbrPv2m0IgnMcnLybARbITqiuuG8sErU0U3Dial1gyhtagw4y4W/9nHK2ULQf+vbOD+BukLZSvQ2gghnxAk4J8XH2FPLtvHZ91JVgQs8jC1AN5OfnHGDAnWFng2w6TpG+eKfnOP1wZyn1lNCT/O9wwn9yCf0NIO5iOye1V2ePN/BaMW//PY9tBYYl6oUfmhwNqbMZJcVObwj1i7ssHIu5ejtMVWjy3mGorzUT3h9MGOwKHkqv4UbDwTHELdIP/JJ0AZ3fBc9PUH1tqSbZivSpI8LW9TpGmZyyCveDoejnM56QhJoQiXAMgAVtaGu8MIY09/AZTPsYoaOW7KpZfP3ugqaac6HB2r9UDgI5fl469vodg+rPSke1SWJNiTao/ITKCzBladxtsbGfSGZ1bKw7fAIFUaojYuyMJs0AOMzK61U85Gip6oyGVKqJYxTQYSqcpIkARSX+yHnUp+TRbXEJgRGipKqzJn1LzPKLbGWidIgG2FMQGQ0yijmuo3z2qTlGOeHlEYuIuugbCYfW02F/wx1eDY9em+SUdaObuThELaqaVax2YuX+hRGK7qhx4WOL9OsIMXUhhk8NALVVs6hF6fobHLfqXk+sJDowdaoci7diirDeQEqbqFmQ1TchagtMyxG03ZzbJDgvfEnqPYKddJHIzD3Qwyefqc6u0QmCTZISJrz7OuA27FPaS3rYcB2W87LVjvE14rnbw6X3+/W8VwKs3mF17By5YuS0cEhxWRIb/cas5ND9t+4BUBV1pwcTKkLobTrJYFgU0IZkEuDZoDPaFR3HdXgElxZoNt96sEexvMF52AMdbomSN18RpmscisMORzl/NHNE7baoUy9+i2iUKGzsYCqqgJVLlC6wE2FZMfZGjufoN/Pgnc/ZW3On4gFEfqpTwkHZFXIn62pb7yAm08In/okNukz3P7oMiIIkV3b04rZi39GfPU6xVNfIJodYOZDaUt5EUe5R7cXoJwjX71CtP897PBQWlBxC5V00PMhKhPeiNXj26xUJZtP/VVGueVoXuFrReIpdD7hd98c8lvP3+Pffm6Hp9ZbPGbHaKQDovMJ35v3eH5vzN95dovKOu6eFqLApRT92NAyijg74VJvhbWGn2FeWual5X9/aZ/VNFzK490ZLdg7zfhrT57j2kpCUUtRdjU2RAcv4Xe2mAY9fK2ItSKY7ImDVBozPcKNB9SjgThBYwguPYl79c+w2Qx3dE8APosZKozQjWq4q6RGUysNXkVr72UIY1n03Q3qu68KniKY4LyQ1V4fr5ji4oSjTEbSVVZStNZRDtZDx7TWxJ7hw1ttEl8zyS3X1xIOpsVy0Gw1DXjxzohsVnDzu8IH8eSnP0Ar9NifFpzuDzh66U8A2H76adLeBbJZweal/lIOb3C64PjemLU04JmNNu1QOCS32hFriY87PsFVhYxsd6UuYGdjEUCuyqYQXsvAFqDyGUFdcDHq0llP2GqHxL7izWHGYysxeeVIww6LytIxVmY4vBDds5Ji1gU66bwvoJRzjvrHVKRUSv2XwM8DBfAm8Heccz+UMPPhcBBKY+M+uZcQ5UIV7rTB276Cm0+pb3wHf+cabLwTcKJo8nDATk5JTm4AgpRTzmJG93hSadi32LBNHXap+hfQrVXZWQGyqYBmlAyBja98hnFR0230NJUSrMKidvheyJfXJvzSFzewaVcmQkc5edSlqB0t4GoYcb4dku59h7q9gadXAUkN0mZ+gfw+S7d1MC1E9v6XP7zNS4dTfu13X+NvfGyXbuRxZaNFP/YJPUXie0IFV1rWOltgK2alZSUyGFss23CliQgQajp7903sfAzWMr2zjzaaJIyWF67LZnitC5ity9jje0ui3rOOkmtJkVcXC2zcRfkB9fAQd+lZVF3gZyOwFc4LqJ0mWAxx2jDKa7JKaASNtiS+ICoDJXMn88rRjfwGQNawklfCbdnf3qGzEvPZJzd4fX/Csa/xohar15+jvXGebFaysiUamqZpBY8bzY7NS33unWbMztVsph5TZ8mrmrwy2JE4TN0SsWMVtbATma2wszEqSsALpG7U7PqumYJNAuHLeHOYkVWW40V1n3hIN0XrMEXlUyHuURp1NlZfvQ80wY8XSfkV4FeaEYhfQ/BN/+iHveAhcRDC2SDcs9VyOtOm6xD3UZMhrixIdVPsMQF57ZZyfeHV6+ikveR4AORizme4w1sCYvETAioJA8tcnnfWHjXBsr89LmqGi5p+KIjHwndklWNaWFqdLczBa9jpKf767rLOMM5rpqWFuM2qXaCYyc4ddQlDTaBl2TkQROh8SBx3wfffwaW5lQYMFyHzccbRNKcbeTxzvktkhFPioEFirif+8tjPyG90sVjet6gsXmsVVczR7R718JBiMie9eB7TqIK7qpDdtCixi5ngQjxfCn7OSb3FBEKaW2ZQF6hyzvTJz5O+/q9Qd18EwF37S6jC4h+9webGdUa1OEtBm9pmjF2+Y9BAumM/5M8PJ1xdSTicyW99Oi8JIg+tFXEacHmrzWorYD/yiFoBYeyxeelxLp7vcDItSCOPN98Y8PaNE3avrnJpXUht/+jFA5LAcJKVXHbhss2c+AZ17SMEswGn/8dvCZ+Dteh2T2D2ni/w6XZPQFBVKelIU5AMlab2Wzy2EnO8qLg5XDDMSp4736UTaiIPfC3zL9aPMYtTnFLiOPrvHUnJjxFJ6Zz7vQdufgMBMv5QezgcBKCKBWElwzSqynFKYeMuLumjHv8UNhvj3xNsg0t6sHJhWXxTz30JB9TGw5zcFum6jeuoukD7sZTPnMUb3qV+49sUwyOCK09LYbTVkTaoL3Rum77mXOItd9Ew8bg1LhhmNUQprZ3nCIwiOrkpO2wQMy8tb08KjmYln9B3qV5/Hv2hz2GDFh2llwXHUVYTJB7m8Ba+s+j2OWzcpRtKB2NRCSHNz3xoi0VRMS0qPr7dI68s+5OCf/LHb/Hp62v8+x/ZQh+PcMYnjTUqn0iLsrmwlB8wrxXJykWM0ujhIe50yvTW2wTtU4JO00bOcqqswGQzGO7hZhOc1nJeoi5OZ5j5Ca4ssNkMnXRIYnEwdjSQ8WcQlOnpIb6zpFvPUBrp1LRDg2fsklhnMK8giakry6VezMmi5Pm3R0uGbYiEQr6RBbxxNCWNPFbaIaetAM83HE1yTg9nHDmRP9xY7xI3Rdwbh1Pq2rLZDt8x8l/WQqSja6GJM1GA6W+ArZdTmcu0w9ZgLbrVwdXFkorOeRFRqMgrcX7DrOTZzQ6xrxruD0tf319O1o9lE9Ieqsrf8zpw8+PfLZ7/Z2s/5CmRUupbD9z+p81c0/u1fw/4n3/Ukx4aB4GzMjdQF4JP8ALJe5XGhS20rcAe3m8bPRABOC9Yvk3d2US13f3H/Ehak1UBddHM8N9n+HGLKSqaLx0EtkI1efy0glsjyR/LBs5cWSlahqqhoqsKtld82qEQkbw+v0z51CUeDyrQmlDrJanuBTPFVT7VM1/ALEbYuMuLxxlbqc9Gy2dRSbtwtZniPIP9WufwreILT59jKw2ZlJZOZxPnBQyzmtRPMa02r59krFqf3XwgEVTRRGNNuFyMZ8spQ2cttqioy4pgMUOlXUzaxWVzitdeIHzCR7fXKVcvo/MJ7sYLkva9/SK2yNDtPipJ8Q5fE8TrxQ9TA7qYE1U5ddKHCg5yS9QgRie55c64xNeKaVFx42TOuTTkXCOLOFyUjBYlv/P8PcLY5xNXV+k1LNcAaSSdjvHJHI2ogRV5xaghzd1vfitBlbL8vcqG0YtRjp2NCc9tobTB5sI47QZ7gqatSlxdEzz5nJDD+IHgHBJB8+psTBrKtfPc+S6xr/jqW0NuDOb8vY/tYONE+EmysdQgzlC472cZOPdz7+sF32dKqd8HNn/AQ//YOfe/Nc/5x0AF/OaPer+Hx0E82D5s2HreYSaAIEQVueTJTSvuHa+DpeqUqgtxOEpLTULJ/6rVkfZTvlguHHXmdB5wNDhLZTW3RxkbDeeCbaY3lWqOVzmUc5jFiG4D7vraYMabJ3Mef3YFkItUGKoUeniETfoc1TEbUYu8dtwdZ1zoBoTVnNrEBEYtkZORp9/BzXmhG9ENfbLKkUQJDqjqkjkOpRzfO5jyzLmUi05o8+VciqNSRlNnBc5aqrksJNvQpIF0PWzYhmxOdXqCPz1Fx21c1FnmxKrdlzZoVQpUO5vj5hNM31F3NnHGQ9UVKp+iojZKSffA10INP8qFTHi9JWpdZ1OwvtFERi+V1Wdj2XGTwBD7BhKWIr61dcvORhB6ZPOCurL0Ep+TaYGN/SUlnqcFOn+pF0vXa7DP7MZN2k8+KajQIoNWW1Clni9gO22kiA3o3pqIEGfyfWxLGK49DZ1QU9SOG4M53719yvpnZHrXBbHQ102PJYqo8vu/xU/AnHOf/2GPK6X+NvAl4HM/ckaKh8VBKI0zvqD3ggQ6ze4fxJIHA85UOD9BmQCVTTG3X8BuPUkdd8l4AFikhE5OZyOpETS9ahu1yVQo2AEFQTZaRhljv8uksNSlYydxmNEeAL3OOR5fS1CopeyfiKM4YbqyDSZjcYrShmMX87MXE/R6jjl6Axu2MSsXCfZfwp7sY8sCdeFpBrYi8X18DZ/Y7jArLcPap7KSj3/x+joARe14e1IwykQD4upKgkLIbNZig5kP2fUNY5MyLSyfvdRnJTaoY+GarHUAPY/o6U8QXBpz8vJ/B1lBa3MV7XvCgJT2UHGLeniITguIEuJnP4Xurkmf31YSOZzblZ/KWVxvk2rlArXy8L71v6JXFC5I8A9eEce6dgW0R+BgpyNCQ34553IvYVZ6FLUTTQ6jOZwVyynWF/fG3DqeYxuRnkkm0gCn81LUtYYLyrzi/GZ7Oafy1sGUIq94fX/KaJxRNziIdmCorAC3WoFmktfo/jrp9etkd2/haku4uSmK6YsZ3rld6WQEBp32qI/uQU8ifVfJZqOqQroVXkTkwbSw/L2P7bD+mW3eGFtG2Zyr/YjQC0lSua7NeF82oIfAlFI/hxQlP+uce0/gjIfEQaj7wKbvZ4R2FosgFsWRBGiTUw8G6K70zp3fWWpJ6rnUMZzx3xEROO0xK+yS0s33ZaAGrVmUbqn/eYbBEMaqjEBL6nF2RGcSefePvTkuP2I0qVmvRpjBbXnIT5bfwZUFZv08VWuVfChALaMVsQezUpyBjJcLAQvIBVjWFns2H9JoWYSewixGMgGrDe1eQhAZ3p5W5JUwSc2dT11ZOlpTT4a4xQy/FS35CkzUFCXjlqAQ8wzLqbREtZHv5PmY03tStHQOh8VpT9CXtsIoi7d7XdKZKpOKP4C1TErLrBTot1aA1kwLS1bbhm3KkPh2qdkhal2aduSxe6kv8yiBx7QRGIaGkbxxDElgWO9EHI1zFp5mZ0V+p0VeNVgRX7QylTj1dmhwRYbprpK//EoTOe0TdFrUWU49PsVZi7eyJtGEranuvL4E8HlhjCoX0soMU3xt6GsPG0skN8rm/MY3bvEPPnOZzdQn9oWVyoatd7By/QXbf4MgBL7SkLx9wzn3H/ywFzwcDqJBrwHLBbc0W2H0Ow/TGR/lB7ijm5joCLP7XDPFCPreS5S3X8N84ktLhmLXOCCt5KNQiCMy8lkPqoILx2WBDVpN+7OhgHfSNfF0oyJh77+vjbuMKs3rgylXzzfhZBDitKQlNumjtxPuJRfJcovW1fJ4z0ITrURQWD0AODpDi4aeXqI6Y0+zyRhz9BYum6E7q+ijN/BMwLeOOqwmAR8/v87hrJSRajWDqsSOT1h9+jJVVjC7d4StLdpoAU35vnQ1Gii7Kwv8D30WZwKqb/5LdBhj1rfR/Q2cF2LDttQesjl3Np8j0IoVoyjXr0lheDHizZnizw8nPNYwcAXtkD8/nDIvaz6122Et8RvyGxE3PpzmbHUjLq0mPLWeMi0q7o1zFkW1pNa/eO6+oE8cGB5bbzFqFMo+9/gGx/OCwaygn4jKtnF+o1VqaYee1LXCiHR3k8nNt5nvD8gGY9LtNeqywtUWU2TUg33Bg3RXqUcDsLZhsC6kq5NPl7+SqnJcEHO1H/EPPnOZD9z6CtmHf55BAathKtdyMfvXWR3/n1nDE/u+7OFwEMUcfeNb6JVNqv4FykBIPo16gJ7K3q832LiLuyI4gBoZlqobvH947ZPoix8m8++/h0M0IFbq4ZI3YuR3G6p8yCsZW1YKTrw+4doKi9LiLHRDOYLKQtbk6xqF9dvigBpS3a6n+dRuhzHA9keprfTH23XW4BMSxkW9pFo7mlecLJpxcdzyWPQDEUptpeZxRuM/XJSMsopFknJ5ZRdVF+wFmxzPK0Z5RT+q2O1EtOspHb+WYtnbr/p+60sAAAfsSURBVFGPBriyYPDiW7jaooxG+z7a95aLYPzyK7SvXkL316VLcftlVNzCP38Z3VnFtVZwtoJ7r6C1ob74Iexml/P5TH6bTGODmEoHqNY66whBzPnUp2wIg39mt83dScnRvFoWYU8WJZV1rLaCJa7g9cFsKTGwmoYsipqisry6NyFflKyvJEyyio005ImtDn/46hFvDefcPZnzytsTfvEj5/G1pkAQqbFf0g8NKu2Rf+/rS55IrxXjaksxmWOiAFtWKC+gPD7ARCHW86UtnLRx2Qw3PV2CoGxT1FbZGGd8Qi9kM/XJPvzzTIuaWWkJjaEioR/+67Aq/MXaw+Egypxq/za+50P/guT4NH1zZHdVsJwrcCaQKc2mUl3nEqri4LTUQIJdCro0YBZPoWcDmScwAdNYSGnOgFBVM+MwymtMISlFZDRR8+OKsK9afuZZ+mK9CFUXqGJOB9irAsb5mS6Hpm2K5cBWuXDLISLdfO7Jomx2U0XVRDK+VksYtm34KUAGo7Kq5vaoZmt7nUDDZFTw5nDOOKv4+E6XtdhDZxPBekwHlMNDwTlkc/LTKUprko0eQSeR+RXPR/k+1WwhNRVt0N1V7GiAHc7wr35wOeyF0uj1XVSVU4YtcmdIx3s4L2rIVDI8H2odkAaayFPEnsKWQpJTOYmGsrzmeF6QNirdddPlEeEeiaTOWpOTJsXIKxFn1lpU1jytllwPn76+xt2TOUUj8vvMRpvrqxHR7ADlLJutTaJyQvXWixTjGdlgzPxwSP/6Ltr3lgVcvyX4EG9lDazF27mKyzNUGIk+BixBUGZxKk7CBJjpMUkKsR8wKERD5Zt3R2y3Iz681ZLd5afUHgoH4Rzv6FrYZch/vzgID3QuHrAzWbwfqW521hpdfgYNyu/+C2vrsM3QlkYiAE/Bu4otW6GnW/JRaH1f+xIZmlLO4R44Nt2oaG22ZGe9fbpgJfYJzDuJO/0HjuvBc+AbUSL/88M5m2nA1bhi42KXvHacm7wFlcZGbaruNqq/g2drFt/6KrP9E+L1HvFql/TpZ6TO0Ayq6c4q65/7PNVgT6KHSx9phGNKqu4menJE8bX/BX/rMlz/OP/ibY+fPReSLoSH2IYthiSs1EPM5BBPexS9iywqy0vH92HGf3pPUow0FFKdvLKsxD7H85I/u3PKpOHg/Fsfv8AoKzmcFXz39in7BxOCyOeJ3S4rabCMKAbzktOspBf5fPGpTb53MOHG0ZTtTkC3FPwCVUFinAjpNtdYOV+Qbq8T9lKywVjqMYBJU0lDKgGPlTdfwVvfhjBCxy15fV2gTIBTSn73KhNH0dSsVsOU0JiGQzQnMilOJz/kwny47aFwEKrVxXziS5Ttc9wclexNpMD6ie0UMxswCvpYIuhfB6ToNBoWywUVNQxPwo14lpY0rExnrE/Ko9q43jBROUxpcU7aYGV9f2dvB0LXbpoWnBneQbVWKXS8rBsYrVANclEVFp1NqNau8PtvjfiZCyHbTlKZIj7H792E0lZoVfGzlztUaPanJa8NZOE8sZawlfoYW3BSGqal5fWTBbvdkDTQbHcipoUMjl1bESDRvLTstH2B9aJpu4y2stikLzWAbMKxt0Jea7YufoRwcorp3Sa48oyIvUxOxVNpsyRhpbOBB1KD0N4ynfMPXsXVNdFH/jJUktJ9erdHe3wbdbrHwfmPMS0t+byijnskKysirpNVTAvb8FZA6hsCoxnnFaV1WAsHs4JxVjFpRqevNgzXv/XCPXZWEq6ttfiFZ8/zx28c881v3+PF2pKmIU+cly7GY6sJ87Lm5umCvYm0o5+/JZ0glY1Rx7dxaxfw77yAKzLGr7/Cra+8gAkMfuzRvnSeeEO0Lo6+/TLx7o7wSPo+Zn2bau+m8DnYWiKq+URqNghwCu3JBlHlmPE+NmyB0lQkfHirRWTk+j02vR/b2vlx20PhIJz2KNNznGY1e5OcexNZPFoJtr32+mS1XepEVBZhnC5FpfuMI9IoBYZ36GyeRRaVEwhyVrmG6FXyFkkxLKU9E7BRS2WvQIMqFw2WgGayU4tobt6gFp1dksi8fDTl85famOObAHitVb5+SzotSWD46+dqfC8kMDF7jVjwbjeQeYYqw/lrOCdiwDudkMBTdBq2Juug61mR9QPMyS1UXVKtXEDnMwGYtVbR+UxIYmhSJx0Qb+zIwl/ZRpcLKUaenfuyQKc9rPHvpxxnyD9nqY730O0ernsOVeboYsG51gx1uofL5uxN5XfwtV4Sx5hGCsAohTH3VcRDT5FYwyivOJjl3BtnGKUoKkscGDYafY3v3D4F5my2QzZaAdc32/xfWlHmFYsGGLLRiRowlGIzDWkHHlvtkK12KEjYuVC0mcWpiM5Mhhx+61W0UbjaEq10mN7ZJ+yl2LJi/WNPS5fjASi6f/EJIXuxFSqbo60V7ESlcf3+8jw5L7zPyF7MpOZQWZxOODY9Xjt5H6zWD5mph0EHUCk1Ad6HRvpP3Nb4Pl2Ph9Ae9mP8//vxXXTOrf8Y3//HYg9FBAG86px77i/6IN7NlFLfepiPDx7+Y3x0fD+d9tPbf3lkj+yR/djtkYN4ZI/skb2rPSwO4v/NuOpP0h7244OH/xgfHd9PoT0URcpH9sge2cNpD0sE8cge2SN7CO2Rg3hkj+yRvas9chCP7JE9sne1Rw7ikT2yR/au9shBPLJH9sje1f5v4jEnfbY2OboAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 576x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(8,4))\n",
    "target_index = 0\n",
    "vmin=-2; vmax=2\n",
    "\n",
    "# plot pred\n",
    "plt.subplot(121) \n",
    "\n",
    "im = plt.matshow(cropped_transformed_arr, fignum=False, cmap= 'RdBu_r', vmax=vmax, vmin=vmin)\n",
    "plt.colorbar(im, fraction=.04, pad = 0.05, ticks=[-2,-1, 0, 1,2]);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "7864\n",
      "8388\n",
      "15728640\n",
      "16777216\n"
     ]
    }
   ],
   "source": [
    "print(int(seq_length*FACTOR/2000))\n",
    "print(int(seq_length*(FACTOR+1)/2000))\n",
    "\n",
    "print(seq_length*FACTOR)\n",
    "print(seq_length*(FACTOR+1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "8388.608"
      ]
     },
     "execution_count": 188,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "16777216/ 2000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "# combine all the transformations in one function\n",
    "def transform_hic(hic_matrix, hic_matrix_raw):\n",
    "    transformed_arr = ct.lib.numutils.adaptive_coarsegrain(hic_matrix, hic_matrix_raw)\n",
    "    transformed_arr, _, _, _ = ct.lib.numutils.observed_over_expected(transformed_arr, mask = ~np.isnan(transformed_arr))\n",
    "    transformed_arr = np.log(transformed_arr)\n",
    "    transformed_arr = ct.lib.numutils.interp_nan(transformed_arr)\n",
    "    transformed_arr = scipy.ndimage.gaussian_filter(transformed_arr, sigma = 1)\n",
    "    \n",
    "    return transformed_arr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Just Fudenberg model doesn't work well on Dropsophila Melanogaster => Try transfer learning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1048576"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "512 * 2048"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "512.0"
      ]
     },
     "execution_count": 96,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "512000 / 1000"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Входных нейронов должно быть 512.000, чтобы 512x512 в HEAD имело такой же смысл, как у Fudenberg'а\n",
    "# Еще надо убрать последний кропающий слой"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "500.0"
      ]
     },
     "execution_count": 101,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "512000 / 2**10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Надо уменьшать, но не так сильно"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "524288"
      ]
     },
     "execution_count": 105,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "2**10 * 512\n",
    "# наверное самое простое использовать такой входной массив"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# try both fine-tuning (not freezing any layers) and transfer learning (freezing some of the layers)\n",
    "# https://datascience.stackexchange.com/questions/41231/does-fine-tuning-of-transferred-layers-perform-better-than-frozen-transferred-la"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}