Analysis Operator Inference - Further computations.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Analysis Sparsity"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Additional computations"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imports:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script><script>requirejs.config({paths: { 'plotly': ['https://cdn.plot.ly/plotly-latest.min']},});if(!window._Plotly) {require(['plotly'],function(plotly) {window._Plotly=plotly;});}</script>"
      ],
      "text/vnd.plotly.v1+html": [
       "<script type=\"text/javascript\">window.PlotlyConfig = {MathJaxConfig: 'local'};</script><script type=\"text/javascript\">if (window.MathJax) {MathJax.Hub.Config({SVG: {font: \"STIX-Web\"}});}</script><script>requirejs.config({paths: { 'plotly': ['https://cdn.plot.ly/plotly-latest.min']},});if(!window._Plotly) {require(['plotly'],function(plotly) {window._Plotly=plotly;});}</script>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "import torch\n",
    "import torch.nn.functional as F\n",
    "import torchvision\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import plotly.offline as ply\n",
    "import plotly.graph_objs as go\n",
    "ply.offline.init_notebook_mode(connected=True)\n",
    "\n",
    "import os\n",
    "import time\n",
    "from collections import defaultdict\n",
    "from copy import deepcopy\n",
    "import skimage.restoration as skir\n",
    "\n",
    "\n",
    "from IPython.display import clear_output\n",
    "%load_ext autoreload\n",
    "\n",
    "from PIL import Image\n",
    "import bilevelsurrogates as Sur\n",
    "\n",
    "#ymmv:\n",
    "torch.backends.cudnn.benchmark = True"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### GPU Setup:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "device = torch.device('cuda:3')\n",
    "dtype = torch.float"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Sur.deterministic()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Data Generation:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "noise = 25/255\n",
    "clip = False\n",
    "batch_size = 200\n",
    "\n",
    "patch_cropping = torchvision.transforms.RandomCrop((64,64))\n",
    "#patch_cropping = None\n",
    "\n",
    "path = os.path.dirname(os.getcwd()) + '/data/BSDS300/' \n",
    "datasetTrain = Sur.data.BSDS300_for_denoising(path, noise_std=noise, augmentations=patch_cropping, flip=True,\n",
    "                                              grayscale=True, clip_to_realistic=clip)\n",
    "datasetTest = Sur.data.BSDS300_for_denoising(path, split = 'testing68', noise_std=noise,\n",
    "                                             augmentations=patch_cropping,\n",
    "                                             flip=True, grayscale=True, clip_to_realistic=clip)\n",
    "datasetTestFull = Sur.data.BSDS300_for_denoising(path, split = 'testing68', noise_std=noise,\n",
    "                                             augmentations=None, flip=True, grayscale=True,\n",
    "                                             clip_to_realistic=clip)\n",
    "datasetTest12 = Sur.data.Test12_for_denoising(os.path.dirname(os.getcwd()) + '/data/test12/', noise_std=noise,\n",
    "                                             augmentations=None, clip_to_realistic=clip)\n",
    "samples = Sur.data.Samples(datasetTrain, batch_size, device=device, dtype=dtype)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def energy_setup(model_setup):\n",
    "    dictionary = Sur.DCTConvolution(in_channels=1, out_channels=48, kernel_size=7, mean=False,\n",
    "                                    bias=model_setup['bias']).to(**model_setup['data'])\n",
    "    torch.nn.init.orthogonal_(dictionary.weight, gain=1)\n",
    "    #torch.nn.init.uniform_(dictionary.weight,-0.1, 0.1)\n",
    "    energy = Sur.model.AnalysisSparsity(dictionary, model_setup)\n",
    "    return energy"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define Higher-level loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "loss = Sur.loss.PSNR()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1) Load Model and validate model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "%autoreload 2\n",
    "model_setup, training_setup, iterative_setup, state = torch.load(\n",
    "    './energies/predictive-primal_large2019-05-26.pt', map_location=device)\n",
    "model_setup['data'] = dict(device=device, dtype=dtype)\n",
    "energy = energy_setup(model_setup)\n",
    "energy.load_state_dict(state)\n",
    "energy.to(device=device, dtype=dtype)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Validate training loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TRAINING | MSE: 0.000902 | avg PSNR for these samples: 28.320,  Time: 00:17 mins\n",
      "TRAINING | MSE: 0.000902 | avg PSNR for these samples: 28.313,  Time: 00:17 mins\n"
     ]
    }
   ],
   "source": [
    "img_denoised, _ = Sur.training_error(energy, loss, samples);\n",
    "samples.redraw_noise()\n",
    "img_denoised, _ = Sur.training_error(energy, loss, samples);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TRAINING | MSE: 0.000885 | avg PSNR for these samples: 28.355,  Time: 00:18 mins\n"
     ]
    }
   ],
   "source": [
    "samples.reset()\n",
    "samples.step()\n",
    "img_denoised, _ = Sur.training_error(energy, loss, samples);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "_, _ = Sur.testing_error(energy, loss, datasetTest);\n",
    "_, _ = Sur.testing_error(energy, loss, datasetTestFull);\n",
    "_, _ = Sur.test_twice_error(energy, loss, datasetTestFull, noise_est);\n",
    "_, _ = Sur.testing_error(energy, loss, datasetTest12, batch_size=1);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(1,3,figsize=(20, 10))\n",
    "axes[0].imshow(samples.x[0,:,:,].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "axes[1].imshow(samples.y[0,:,:,].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "axes[2].imshow(img_denoised[0,:,:,:].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "fig.canvas.draw()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1a) Density function estimation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Noisy samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "filter_responses = energy.operator(samples.y).detach()\n",
    "plt.hist(filter_responses.view(-1).cpu().numpy(), range=(-0.5, 0.5), bins = 250, density=True, log=True);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Clean samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "filter_responses = energy.operator(samples.x).detach()\n",
    "plt.hist(filter_responses.view(-1).cpu().numpy(), range=(-0.5, 0.5), bins = 250, density=True, log=True);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1b) Sort filters by importance "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dictionary norm is 2.271.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 49 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "Sur.visualize(energy.operator)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TESTING | MSE: 0.000893 | avg PSNR for the given testset: 27.936, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000894 | avg PSNR for the given testset: 27.926, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000890 | avg PSNR for the given testset: 27.950, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000895 | avg PSNR for the given testset: 27.907, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000894 | avg PSNR for the given testset: 27.922, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000890 | avg PSNR for the given testset: 27.955, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000892 | avg PSNR for the given testset: 27.923, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000891 | avg PSNR for the given testset: 27.933, Time: 01:27 mins---\n",
      "TESTING | MSE: 0.000896 | avg PSNR for the given testset: 27.896, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000892 | avg PSNR for the given testset: 27.939, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000890 | avg PSNR for the given testset: 27.941, Time: 01:25 mins---\n",
      "TESTING | MSE: 0.000894 | avg PSNR for the given testset: 27.903, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000890 | avg PSNR for the given testset: 27.941, Time: 01:27 mins---\n",
      "TESTING | MSE: 0.000891 | avg PSNR for the given testset: 27.951, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000896 | avg PSNR for the given testset: 27.902, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000890 | avg PSNR for the given testset: 27.942, Time: 01:27 mins---\n",
      "TESTING | MSE: 0.000890 | avg PSNR for the given testset: 27.950, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000898 | avg PSNR for the given testset: 27.883, Time: 01:25 mins---\n",
      "TESTING | MSE: 0.000890 | avg PSNR for the given testset: 27.945, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000899 | avg PSNR for the given testset: 27.884, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000895 | avg PSNR for the given testset: 27.902, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000890 | avg PSNR for the given testset: 27.942, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000894 | avg PSNR for the given testset: 27.908, Time: 01:25 mins---\n",
      "TESTING | MSE: 0.000891 | avg PSNR for the given testset: 27.942, Time: 01:25 mins---\n",
      "TESTING | MSE: 0.000892 | avg PSNR for the given testset: 27.927, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000895 | avg PSNR for the given testset: 27.909, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000893 | avg PSNR for the given testset: 27.934, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000891 | avg PSNR for the given testset: 27.940, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000892 | avg PSNR for the given testset: 27.926, Time: 01:25 mins---\n",
      "TESTING | MSE: 0.000895 | avg PSNR for the given testset: 27.915, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000895 | avg PSNR for the given testset: 27.908, Time: 01:27 mins---\n",
      "TESTING | MSE: 0.000889 | avg PSNR for the given testset: 27.948, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000892 | avg PSNR for the given testset: 27.933, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000898 | avg PSNR for the given testset: 27.880, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000893 | avg PSNR for the given testset: 27.921, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000900 | avg PSNR for the given testset: 27.879, Time: 01:27 mins---\n",
      "TESTING | MSE: 0.000898 | avg PSNR for the given testset: 27.893, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000891 | avg PSNR for the given testset: 27.950, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000891 | avg PSNR for the given testset: 27.930, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000904 | avg PSNR for the given testset: 27.845, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000892 | avg PSNR for the given testset: 27.937, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000890 | avg PSNR for the given testset: 27.952, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000893 | avg PSNR for the given testset: 27.918, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000893 | avg PSNR for the given testset: 27.930, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000892 | avg PSNR for the given testset: 27.933, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000891 | avg PSNR for the given testset: 27.943, Time: 01:25 mins---\n",
      "TESTING | MSE: 0.000894 | avg PSNR for the given testset: 27.913, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000894 | avg PSNR for the given testset: 27.913, Time: 01:26 mins---\n"
     ]
    }
   ],
   "source": [
    "filters = energy.operator.weight.clone().detach()\n",
    "psnr_vals = []\n",
    "for i in range(filters.shape[0]):\n",
    "    energy.operator.weight[i, :, :, :].zero_()\n",
    "    #_, metric = Sur.training_error(energy, loss, samples)\n",
    "    _, metric = Sur.testing_error(energy, loss, datasetTestFull)\n",
    "    psnr_vals.append(metric)\n",
    "    energy.operator.weight[i, :, :, :] = filters[i, :, :, :].clone()\n",
    "indices = np.argsort(psnr_vals)\n",
    "energy.operator.weight = torch.nn.Parameter(energy.operator.weight[indices, :, :, :])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dictionary norm is 2.271.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 49 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "Sur.visualize(energy.operator)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1c) Remove filters by order of importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TESTING | MSE: 0.000888 | avg PSNR for the given testset: 27.961, Time: 01:25 mins---\n",
      "TESTING | MSE: 0.000890 | avg PSNR for the given testset: 27.945, Time: 01:25 mins---\n",
      "TESTING | MSE: 0.000893 | avg PSNR for the given testset: 27.926, Time: 01:27 mins---\n",
      "TESTING | MSE: 0.000893 | avg PSNR for the given testset: 27.913, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000895 | avg PSNR for the given testset: 27.892, Time: 01:26 mins---\n",
      "TESTING | MSE: 0.000898 | avg PSNR for the given testset: 27.875, Time: 01:25 mins---\n",
      "TESTING | MSE: 0.000900 | avg PSNR for the given testset: 27.845, Time: 01:25 mins---\n",
      "TESTING | MSE: 0.000905 | avg PSNR for the given testset: 27.812, Time: 01:25 mins---\n",
      "TESTING | MSE: 0.000915 | avg PSNR for the given testset: 27.745, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.000933 | avg PSNR for the given testset: 27.631, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.000962 | avg PSNR for the given testset: 27.472, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001016 | avg PSNR for the given testset: 27.193, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001029 | avg PSNR for the given testset: 27.125, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001086 | avg PSNR for the given testset: 26.851, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001117 | avg PSNR for the given testset: 26.714, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001176 | avg PSNR for the given testset: 26.465, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001186 | avg PSNR for the given testset: 26.425, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001197 | avg PSNR for the given testset: 26.376, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001210 | avg PSNR for the given testset: 26.320, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001254 | avg PSNR for the given testset: 26.148, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001272 | avg PSNR for the given testset: 26.079, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001296 | avg PSNR for the given testset: 25.989, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001373 | avg PSNR for the given testset: 25.722, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001486 | avg PSNR for the given testset: 25.359, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001552 | avg PSNR for the given testset: 25.157, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001631 | avg PSNR for the given testset: 24.934, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001661 | avg PSNR for the given testset: 24.851, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001795 | avg PSNR for the given testset: 24.503, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001839 | avg PSNR for the given testset: 24.394, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001904 | avg PSNR for the given testset: 24.238, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.001991 | avg PSNR for the given testset: 24.038, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.002029 | avg PSNR for the given testset: 23.953, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.002152 | avg PSNR for the given testset: 23.694, Time: 01:23 mins---\n",
      "TESTING | MSE: 0.002209 | avg PSNR for the given testset: 23.579, Time: 01:17 mins---\n",
      "TESTING | MSE: 0.002289 | avg PSNR for the given testset: 23.419, Time: 01:12 mins---\n",
      "TESTING | MSE: 0.002376 | avg PSNR for the given testset: 23.255, Time: 01:08 mins---\n",
      "TESTING | MSE: 0.002439 | avg PSNR for the given testset: 23.141, Time: 01:05 mins---\n",
      "TESTING | MSE: 0.002616 | avg PSNR for the given testset: 22.833, Time: 01:01 mins---\n",
      "TESTING | MSE: 0.002798 | avg PSNR for the given testset: 22.541, Time: 00:58 mins---\n",
      "TESTING | MSE: 0.002910 | avg PSNR for the given testset: 22.368, Time: 00:54 mins---\n",
      "TESTING | MSE: 0.003148 | avg PSNR for the given testset: 22.026, Time: 00:43 mins---\n",
      "TESTING | MSE: 0.003288 | avg PSNR for the given testset: 21.836, Time: 00:42 mins---\n",
      "TESTING | MSE: 0.003405 | avg PSNR for the given testset: 21.683, Time: 00:39 mins---\n",
      "TESTING | MSE: 0.003577 | avg PSNR for the given testset: 21.469, Time: 00:38 mins---\n",
      "TESTING | MSE: 0.003810 | avg PSNR for the given testset: 21.194, Time: 00:37 mins---\n",
      "TESTING | MSE: 0.004068 | avg PSNR for the given testset: 20.910, Time: 00:36 mins---\n",
      "TESTING | MSE: 0.004221 | avg PSNR for the given testset: 20.749, Time: 00:36 mins---\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/vsa_jonas/Dropbox/Documents_Hyperion/Python/BilevelSurrogates/bilevelsurrogates/subfunctions/normest.py:38: UserWarning:\n",
      "\n",
      "Normest did not convergence to tolerance 1e-07 within 5000 iterations. Final tol was nan\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TESTING | MSE: nan | avg PSNR for the given testset: nan, Time: 01:39 mins---\n"
     ]
    }
   ],
   "source": [
    "filters = energy.operator.weight.clone().detach()\n",
    "psnr_vals = []\n",
    "for i in reversed(range(filters.shape[0])):\n",
    "    energy.operator.weight[i, :, :, :].zero_()\n",
    "    #_, metric = Sur.training_error(energy, loss, samples)\n",
    "    _, metric = Sur.testing_error(energy, loss, datasetTestFull)\n",
    "    psnr_vals.append(metric)\n",
    "indices = np.argsort(psnr_vals)\n",
    "\n",
    "# Return filters\n",
    "energy.operator.weight = torch.nn.Parameter(energy.operator.weight[indices, :, :, :])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "config": {
        "linkText": "Export to plot.ly",
        "plotlyServerURL": "https://plot.ly",
        "showLink": false
       },
       "data": [
        {
         "name": "PSNR",
         "type": "scatter",
         "uid": "70ed80c9-29ad-4e99-82ac-7eb8a904d167",
         "y": [
          27.961323764589096,
          27.945317533281113,
          27.925787369410198,
          27.913349045647514,
          27.89153085814582,
          27.8745403819614,
          27.844605604807537,
          27.81153631210327,
          27.744841390185886,
          27.63089871406555,
          27.47155910068088,
          27.193348593182034,
          27.1246239344279,
          26.851372904247707,
          26.71383113331265,
          26.46538699997796,
          26.424799071417915,
          26.37645806206597,
          26.319500764211018,
          26.147851361168755,
          26.079386234283447,
          25.989042308595444,
          25.722151041030884,
          25.359497997495865,
          25.157336923811172,
          24.934314515855576,
          24.850515551037258,
          24.503166410658096,
          24.393762323591446,
          24.237975067562527,
          24.03808018896315,
          23.952907933129204,
          23.69428515434265,
          23.578515238232082,
          23.419383260938858,
          23.255290057924057,
          23.140897750854492,
          22.833070039749146,
          22.540548880894978,
          22.368042442533707,
          22.025528536902534,
          21.836037900712753,
          21.68335639105903,
          21.46885559293959,
          21.19376712375217,
          20.90957424375746,
          20.749237563874985,
          null
         ]
        }
       ],
       "layout": {
        "title": {
         "text": "PSNR value after removing the n-th least important filter"
        },
        "xaxis": {
         "automargin": true,
         "autorange": true,
         "title": {
          "text": "Iterations"
         },
         "type": "linear"
        },
        "yaxis": {
         "automargin": true,
         "autorange": true,
         "title": {
          "text": "PSNR"
         },
         "type": "log"
        }
       }
      },
      "text/html": [
       "<div id=\"6a8dcb30-ee49-4be6-bf92-6cda863a36d9\" style=\"height: 525px; width: 100%;\" class=\"plotly-graph-div\"></div><script type=\"text/javascript\">require([\"plotly\"], function(Plotly) { window.PLOTLYENV=window.PLOTLYENV || {};window.PLOTLYENV.BASE_URL=\"https://plot.ly\";Plotly.newPlot(\"6a8dcb30-ee49-4be6-bf92-6cda863a36d9\", [{\"name\": \"PSNR\", \"y\": [27.961323764589096, 27.945317533281113, 27.925787369410198, 27.913349045647514, 27.89153085814582, 27.8745403819614, 27.844605604807537, 27.81153631210327, 27.744841390185886, 27.63089871406555, 27.47155910068088, 27.193348593182034, 27.1246239344279, 26.851372904247707, 26.71383113331265, 26.46538699997796, 26.424799071417915, 26.37645806206597, 26.319500764211018, 26.147851361168755, 26.079386234283447, 25.989042308595444, 25.722151041030884, 25.359497997495865, 25.157336923811172, 24.934314515855576, 24.850515551037258, 24.503166410658096, 24.393762323591446, 24.237975067562527, 24.03808018896315, 23.952907933129204, 23.69428515434265, 23.578515238232082, 23.419383260938858, 23.255290057924057, 23.140897750854492, 22.833070039749146, 22.540548880894978, 22.368042442533707, 22.025528536902534, 21.836037900712753, 21.68335639105903, 21.46885559293959, 21.19376712375217, 20.90957424375746, 20.749237563874985, null], \"type\": \"scatter\", \"uid\": \"70ed80c9-29ad-4e99-82ac-7eb8a904d167\"}], {\"title\": {\"text\": \"PSNR value after removing the n-th least important filter\"}, \"xaxis\": {\"automargin\": true, \"autorange\": true, \"title\": {\"text\": \"Iterations\"}, \"type\": \"linear\"}, \"yaxis\": {\"automargin\": true, \"autorange\": true, \"title\": {\"text\": \"PSNR\"}, \"type\": \"log\"}}, {\"showLink\": false, \"linkText\": \"Export to plot.ly\", \"plotlyServerURL\": \"https://plot.ly\"})});</script><script type=\"text/javascript\">window.addEventListener(\"resize\", function(){window._Plotly.Plots.resize(document.getElementById(\"6a8dcb30-ee49-4be6-bf92-6cda863a36d9\"));});</script>"
      ],
      "text/vnd.plotly.v1+html": [
       "<div id=\"6a8dcb30-ee49-4be6-bf92-6cda863a36d9\" style=\"height: 525px; width: 100%;\" class=\"plotly-graph-div\"></div><script type=\"text/javascript\">require([\"plotly\"], function(Plotly) { window.PLOTLYENV=window.PLOTLYENV || {};window.PLOTLYENV.BASE_URL=\"https://plot.ly\";Plotly.newPlot(\"6a8dcb30-ee49-4be6-bf92-6cda863a36d9\", [{\"name\": \"PSNR\", \"y\": [27.961323764589096, 27.945317533281113, 27.925787369410198, 27.913349045647514, 27.89153085814582, 27.8745403819614, 27.844605604807537, 27.81153631210327, 27.744841390185886, 27.63089871406555, 27.47155910068088, 27.193348593182034, 27.1246239344279, 26.851372904247707, 26.71383113331265, 26.46538699997796, 26.424799071417915, 26.37645806206597, 26.319500764211018, 26.147851361168755, 26.079386234283447, 25.989042308595444, 25.722151041030884, 25.359497997495865, 25.157336923811172, 24.934314515855576, 24.850515551037258, 24.503166410658096, 24.393762323591446, 24.237975067562527, 24.03808018896315, 23.952907933129204, 23.69428515434265, 23.578515238232082, 23.419383260938858, 23.255290057924057, 23.140897750854492, 22.833070039749146, 22.540548880894978, 22.368042442533707, 22.025528536902534, 21.836037900712753, 21.68335639105903, 21.46885559293959, 21.19376712375217, 20.90957424375746, 20.749237563874985, null], \"type\": \"scatter\", \"uid\": \"70ed80c9-29ad-4e99-82ac-7eb8a904d167\"}], {\"title\": {\"text\": \"PSNR value after removing the n-th least important filter\"}, \"xaxis\": {\"automargin\": true, \"autorange\": true, \"title\": {\"text\": \"Iterations\"}, \"type\": \"linear\"}, \"yaxis\": {\"automargin\": true, \"autorange\": true, \"title\": {\"text\": \"PSNR\"}, \"type\": \"log\"}}, {\"showLink\": false, \"linkText\": \"Export to plot.ly\", \"plotlyServerURL\": \"https://plot.ly\"})});</script><script type=\"text/javascript\">window.addEventListener(\"resize\", function(){window._Plotly.Plots.resize(document.getElementById(\"6a8dcb30-ee49-4be6-bf92-6cda863a36d9\"));});</script>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "data = [go.Scatter(y=psnr_vals, name = 'PSNR')]\n",
    "layout = go.Layout(\n",
    "         xaxis=dict(\n",
    "             type='linear',\n",
    "             autorange=True,\n",
    "             title='Iterations',\n",
    "             automargin=True\n",
    "             ),\n",
    "         yaxis=dict(\n",
    "             type='linear',\n",
    "             autorange  = True,\n",
    "             title='PSNR',\n",
    "             automargin=True\n",
    "         ),\n",
    "         title = (f'PSNR value after removing the n-th least important filter')\n",
    ")\n",
    "fig = go.Figure(data=data, layout=layout)\n",
    "ply.iplot(fig)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 1d) Run filters through a non-convex model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "nc_setup = deepcopy(model_setup)\n",
    "nc_setup['alpha'] = 0.01\n",
    "nc_setup['epsilon'] = 1\n",
    "nc_setup['bias'] = False\n",
    "nc_setup['norm'] = 'aniso'\n",
    "# Possible regularizers are\n",
    "# 'huber', 'l1', 'log-square', 'sqrt', log-abs\n",
    "nc_setup['regularizer'] = 'log-abs' \n",
    "nc_setup['learn_scales'] = False\n",
    "nc_energy = Sur.model.AnalysisNonconvex(energy.operator, nc_setup)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TRAINING | MSE: 0.000905 | avg PSNR for these samples: 28.259,  Time: 03:46 mins\n"
     ]
    }
   ],
   "source": [
    "img_denoised, _ = Sur.training_error(nc_energy, loss, samples);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 2) Compare with optimal TV"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Current alpha is 0.049019607843137254\n",
      "Average psnr of denoised image is 26.63237494468689\n",
      "Current alpha is 0.04961984793917567\n",
      "Average psnr of denoised image is 26.683459968566893\n",
      "Current alpha is 0.050220088035214085\n",
      "Average psnr of denoised image is 26.733455085754393\n",
      "Current alpha is 0.0508203281312525\n",
      "Average psnr of denoised image is 26.782364015579223\n",
      "Current alpha is 0.051420568227290915\n",
      "Average psnr of denoised image is 26.830165691375733\n",
      "Current alpha is 0.05202080832332933\n",
      "Average psnr of denoised image is 26.876861925125123\n",
      "Current alpha is 0.052621048419367746\n",
      "Average psnr of denoised image is 26.922443313598635\n",
      "Current alpha is 0.05322128851540616\n",
      "Average psnr of denoised image is 26.966912212371827\n",
      "Current alpha is 0.053821528611444576\n",
      "Average psnr of denoised image is 27.01026651382446\n",
      "Current alpha is 0.05442176870748299\n",
      "Average psnr of denoised image is 27.052496061325073\n",
      "Current alpha is 0.055022008803521406\n",
      "Average psnr of denoised image is 27.093607845306398\n",
      "Current alpha is 0.05562224889955982\n",
      "Average psnr of denoised image is 27.133581142425538\n",
      "Current alpha is 0.05622248899559824\n",
      "Average psnr of denoised image is 27.17241069793701\n",
      "Current alpha is 0.05682272909163665\n",
      "Average psnr of denoised image is 27.210100917816163\n",
      "Current alpha is 0.05742296918767507\n",
      "Average psnr of denoised image is 27.246663904190065\n",
      "Current alpha is 0.05802320928371348\n",
      "Average psnr of denoised image is 27.28210252761841\n",
      "Current alpha is 0.0586234493797519\n",
      "Average psnr of denoised image is 27.316409425735472\n",
      "Current alpha is 0.05922368947579031\n",
      "Average psnr of denoised image is 27.3495862865448\n",
      "Current alpha is 0.05982392957182873\n",
      "Average psnr of denoised image is 27.381641035079955\n",
      "Current alpha is 0.06042416966786714\n",
      "Average psnr of denoised image is 27.41259171485901\n",
      "Current alpha is 0.06102440976390556\n",
      "Average psnr of denoised image is 27.442433519363405\n",
      "Current alpha is 0.061624649859943974\n",
      "Average psnr of denoised image is 27.471167898178102\n",
      "Current alpha is 0.06222488995598239\n",
      "Average psnr of denoised image is 27.498794355392455\n",
      "Current alpha is 0.06282513005202081\n",
      "Average psnr of denoised image is 27.525332279205323\n",
      "Current alpha is 0.06342537014805923\n",
      "Average psnr of denoised image is 27.55077127456665\n",
      "Current alpha is 0.06402561024409764\n",
      "Average psnr of denoised image is 27.575136613845824\n",
      "Current alpha is 0.06462585034013606\n",
      "Average psnr of denoised image is 27.598436059951784\n",
      "Current alpha is 0.06522609043617447\n",
      "Average psnr of denoised image is 27.620679607391356\n",
      "Current alpha is 0.06582633053221289\n",
      "Average psnr of denoised image is 27.641884422302248\n",
      "Current alpha is 0.0664265706282513\n",
      "Average psnr of denoised image is 27.662064237594606\n",
      "Current alpha is 0.06702681072428972\n",
      "Average psnr of denoised image is 27.681228942871094\n",
      "Current alpha is 0.06762705082032813\n",
      "Average psnr of denoised image is 27.699387016296388\n",
      "Current alpha is 0.06822729091636655\n",
      "Average psnr of denoised image is 27.716544694900513\n",
      "Current alpha is 0.06882753101240496\n",
      "Average psnr of denoised image is 27.732734451293947\n",
      "Current alpha is 0.06942777110844338\n",
      "Average psnr of denoised image is 27.747962636947634\n",
      "Current alpha is 0.0700280112044818\n",
      "Average psnr of denoised image is 27.762245464324952\n",
      "Current alpha is 0.07062825130052021\n",
      "Average psnr of denoised image is 27.77560070991516\n",
      "Current alpha is 0.07122849139655862\n",
      "Average psnr of denoised image is 27.788052444458007\n",
      "Current alpha is 0.07182873149259704\n",
      "Average psnr of denoised image is 27.799617061614992\n",
      "Current alpha is 0.07242897158863545\n",
      "Average psnr of denoised image is 27.810311050415038\n",
      "Current alpha is 0.07302921168467387\n",
      "Average psnr of denoised image is 27.820141048431395\n",
      "Current alpha is 0.07362945178071229\n",
      "Average psnr of denoised image is 27.82914281845093\n",
      "Current alpha is 0.0742296918767507\n",
      "Average psnr of denoised image is 27.837322225570677\n",
      "Current alpha is 0.07482993197278912\n",
      "Average psnr of denoised image is 27.844699535369873\n",
      "Current alpha is 0.07543017206882753\n",
      "Average psnr of denoised image is 27.85128481864929\n",
      "Current alpha is 0.07603041216486595\n",
      "Average psnr of denoised image is 27.85711350440979\n",
      "Current alpha is 0.07663065226090436\n",
      "Average psnr of denoised image is 27.86219882965088\n",
      "Current alpha is 0.07723089235694278\n",
      "Average psnr of denoised image is 27.866559133529663\n",
      "Current alpha is 0.07783113245298119\n",
      "Average psnr of denoised image is 27.8701873588562\n",
      "Current alpha is 0.0784313725490196\n",
      "Average psnr of denoised image is 27.873102598190307\n"
     ]
    }
   ],
   "source": [
    "tv_setup = deepcopy(model_setup)\n",
    "tv_setup['norm'] = 'iso2'\n",
    "\n",
    "# Define Model\n",
    "grad = Sur.Convolution(1, 2, 3).initialize_gradient().to(device)\n",
    "grad.weight.requires_grad = False\n",
    "tv_energy = Sur.model.AnalysisSparsity(grad, tv_setup)\n",
    "\n",
    "\n",
    "def grid_search(img_noisy):\n",
    "    psnr_vals = []\n",
    "    alpha_range = np.linspace(0.5*noise,0.8*noise,50)\n",
    "    for alpha_test in alpha_range:\n",
    "        tv_energy.setup['alpha'] = alpha_test\n",
    "        output = tv_energy.minimize(samples.y)\n",
    "        psnr = Sur.psnr_compute(output, samples.x)\n",
    "        psnr_vals.append(psnr)\n",
    "        print(f'Current alpha is {alpha_test}')\n",
    "        print(f'Average psnr of denoised image is {psnr}')\n",
    "    alpha = alpha_range[np.argmax(psnr_vals)]\n",
    "    return alpha\n",
    "    \n",
    "tv_energy.setup['alpha'] = grid_search(samples.y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Optimal alpha by grid search is 0.078\n",
      "TRAINING | MSE: 0.001068 | avg PSNR for these samples: 27.873,  Time: 00:01 mins\n",
      "TESTING | MSE: 0.001051 | avg PSNR for the given testset: 27.407, Time: 00:06 mins---\n"
     ]
    }
   ],
   "source": [
    "print(f'Optimal alpha by grid search is {tv_energy.setup[\"alpha\"]:.3f}')\n",
    "img_denoised, _ = Sur.training_error(tv_energy, loss, samples);\n",
    "_, _ = Sur.testing_error(tv_energy, loss, datasetTestFull);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(1,3,figsize=(20, 10))\n",
    "axes[0].imshow(samples.x[0,:,:,].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "axes[1].imshow(samples.y[0,:,:,].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "axes[2].imshow(img_denoised[0,:,:,:].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "fig.canvas.draw()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3) Fixed Point Property Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "import skimage.restoration as skir\n",
    "\n",
    "# If using test data:\n",
    "samples = Sur.data.Samples(datasetTest, batch_size, device=device, dtype=dtype)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "psnr_est = defaultdict(list)\n",
    "noise_est = defaultdict(lambda: [noise])\n",
    "\n",
    "output = dict()\n",
    "output['tv'], output['dict'] = [samples.y]*2\n",
    "\n",
    "energies = dict()\n",
    "energies['tv'], energies['dict'] = deepcopy(tv_energy), deepcopy(energy)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Estimated Noise for tv is 0.1000 lead to psnr 27.28\n",
      "Estimated Noise for dict is 0.1000 lead to psnr 28.16\n",
      "Estimated Noise for tv is 0.0104 lead to psnr 27.17\n",
      "Estimated Noise for dict is 0.0069 lead to psnr 28.20\n",
      "Estimated Noise for tv is 0.0076 lead to psnr 27.05\n",
      "Estimated Noise for dict is 0.0044 lead to psnr 28.17\n",
      "Estimated Noise for tv is 0.0061 lead to psnr 26.93\n",
      "Estimated Noise for dict is 0.0035 lead to psnr 28.13\n",
      "Estimated Noise for tv is 0.0053 lead to psnr 26.82\n",
      "Estimated Noise for dict is 0.0029 lead to psnr 28.09\n",
      "Estimated Noise for tv is 0.0047 lead to psnr 26.73\n",
      "Estimated Noise for dict is 0.0026 lead to psnr 28.05\n",
      "Estimated Noise for tv is 0.0043 lead to psnr 26.64\n",
      "Estimated Noise for dict is 0.0023 lead to psnr 28.01\n",
      "Estimated Noise for tv is 0.0039 lead to psnr 26.56\n",
      "Estimated Noise for dict is 0.0021 lead to psnr 27.98\n",
      "Estimated Noise for tv is 0.0037 lead to psnr 26.48\n",
      "Estimated Noise for dict is 0.0019 lead to psnr 27.94\n",
      "Estimated Noise for tv is 0.0035 lead to psnr 26.42\n",
      "Estimated Noise for dict is 0.0018 lead to psnr 27.90\n",
      "Estimated Noise for tv is 0.0032 lead to psnr 26.35\n",
      "Estimated Noise for dict is 0.0017 lead to psnr 27.87\n",
      "Estimated Noise for tv is 0.0031 lead to psnr 26.29\n",
      "Estimated Noise for dict is 0.0016 lead to psnr 27.84\n",
      "Estimated Noise for tv is 0.0029 lead to psnr 26.24\n",
      "Estimated Noise for dict is 0.0015 lead to psnr 27.81\n",
      "Estimated Noise for tv is 0.0028 lead to psnr 26.18\n",
      "Estimated Noise for dict is 0.0014 lead to psnr 27.78\n",
      "Estimated Noise for tv is 0.0027 lead to psnr 26.13\n",
      "Estimated Noise for dict is 0.0014 lead to psnr 27.75\n",
      "Estimated Noise for tv is 0.0026 lead to psnr 26.09\n",
      "Estimated Noise for dict is 0.0013 lead to psnr 27.72\n",
      "Estimated Noise for tv is 0.0025 lead to psnr 26.04\n",
      "Estimated Noise for dict is 0.0013 lead to psnr 27.69\n",
      "Estimated Noise for tv is 0.0024 lead to psnr 26.00\n",
      "Estimated Noise for dict is 0.0012 lead to psnr 27.67\n",
      "Estimated Noise for tv is 0.0023 lead to psnr 25.96\n",
      "Estimated Noise for dict is 0.0012 lead to psnr 27.64\n",
      "Estimated Noise for tv is 0.0023 lead to psnr 25.92\n",
      "Estimated Noise for dict is 0.0011 lead to psnr 27.62\n",
      "Estimated Noise for tv is 0.0022 lead to psnr 25.88\n",
      "Estimated Noise for dict is 0.0011 lead to psnr 27.60\n",
      "Estimated Noise for tv is 0.0021 lead to psnr 25.84\n",
      "Estimated Noise for dict is 0.0011 lead to psnr 27.57\n",
      "Estimated Noise for tv is 0.0021 lead to psnr 25.81\n",
      "Estimated Noise for dict is 0.0010 lead to psnr 27.55\n",
      "Estimated Noise for tv is 0.0020 lead to psnr 25.77\n",
      "Estimated Noise for dict is 0.0010 lead to psnr 27.53\n",
      "Estimated Noise for tv is 0.0020 lead to psnr 25.74\n",
      "Estimated Noise for dict is 0.0010 lead to psnr 27.51\n",
      "Estimated Noise for tv is 0.0019 lead to psnr 25.71\n",
      "Estimated Noise for dict is 0.0010 lead to psnr 27.49\n",
      "Estimated Noise for tv is 0.0019 lead to psnr 25.68\n",
      "Estimated Noise for dict is 0.0009 lead to psnr 27.47\n",
      "Estimated Noise for tv is 0.0018 lead to psnr 25.65\n",
      "Estimated Noise for dict is 0.0009 lead to psnr 27.45\n",
      "Estimated Noise for tv is 0.0018 lead to psnr 25.62\n",
      "Estimated Noise for dict is 0.0009 lead to psnr 27.43\n",
      "Estimated Noise for tv is 0.0017 lead to psnr 25.60\n",
      "Estimated Noise for dict is 0.0009 lead to psnr 27.41\n",
      "Estimated Noise for tv is 0.0017 lead to psnr 25.57\n",
      "Estimated Noise for dict is 0.0008 lead to psnr 27.40\n",
      "Estimated Noise for tv is 0.0017 lead to psnr 25.54\n",
      "Estimated Noise for dict is 0.0008 lead to psnr 27.38\n",
      "Estimated Noise for tv is 0.0016 lead to psnr 25.52\n",
      "Estimated Noise for dict is 0.0008 lead to psnr 27.36\n",
      "Estimated Noise for tv is 0.0016 lead to psnr 25.49\n",
      "Estimated Noise for dict is 0.0008 lead to psnr 27.35\n",
      "Estimated Noise for tv is 0.0016 lead to psnr 25.47\n",
      "Estimated Noise for dict is 0.0008 lead to psnr 27.33\n",
      "Estimated Noise for tv is 0.0016 lead to psnr 25.45\n",
      "Estimated Noise for dict is 0.0008 lead to psnr 27.31\n",
      "Estimated Noise for tv is 0.0015 lead to psnr 25.42\n",
      "Estimated Noise for dict is 0.0008 lead to psnr 27.30\n",
      "Estimated Noise for tv is 0.0015 lead to psnr 25.40\n",
      "Estimated Noise for dict is 0.0007 lead to psnr 27.28\n",
      "Estimated Noise for tv is 0.0015 lead to psnr 25.38\n",
      "Estimated Noise for dict is 0.0007 lead to psnr 27.27\n",
      "Estimated Noise for tv is 0.0015 lead to psnr 25.36\n",
      "Estimated Noise for dict is 0.0007 lead to psnr 27.25\n",
      "Estimated Noise for tv is 0.0014 lead to psnr 25.34\n",
      "Estimated Noise for dict is 0.0007 lead to psnr 27.24\n",
      "Estimated Noise for tv is 0.0014 lead to psnr 25.32\n",
      "Estimated Noise for dict is 0.0007 lead to psnr 27.23\n",
      "Estimated Noise for tv is 0.0014 lead to psnr 25.30\n",
      "Estimated Noise for dict is 0.0007 lead to psnr 27.21\n",
      "Estimated Noise for tv is 0.0014 lead to psnr 25.28\n",
      "Estimated Noise for dict is 0.0007 lead to psnr 27.20\n",
      "Estimated Noise for tv is 0.0014 lead to psnr 25.26\n",
      "Estimated Noise for dict is 0.0007 lead to psnr 27.19\n",
      "Estimated Noise for tv is 0.0013 lead to psnr 25.24\n",
      "Estimated Noise for dict is 0.0007 lead to psnr 27.17\n",
      "Estimated Noise for tv is 0.0013 lead to psnr 25.22\n",
      "Estimated Noise for dict is 0.0006 lead to psnr 27.16\n",
      "Estimated Noise for tv is 0.0013 lead to psnr 25.20\n",
      "Estimated Noise for dict is 0.0006 lead to psnr 27.15\n",
      "Estimated Noise for tv is 0.0013 lead to psnr 25.19\n",
      "Estimated Noise for dict is 0.0006 lead to psnr 27.13\n",
      "Estimated Noise for tv is 0.0013 lead to psnr 25.17\n",
      "Estimated Noise for dict is 0.0006 lead to psnr 27.12\n"
     ]
    }
   ],
   "source": [
    "repetitions = 50\n",
    "\n",
    "for iterations in range(repetitions):\n",
    "    for key in output.keys():\n",
    "        noise_est[key].append(skir.estimate_sigma(output[key].permute(2,3,0,1).cpu().numpy(),\n",
    "                                    multichannel=True, average_sigmas=True))\n",
    "        energies[key].setup['alpha'] *= noise_est[key][-1] / noise_est[key][-2]\n",
    "        output[key] = energies[key].minimize(output[key])\n",
    "        psnr_est[key].append(Sur.psnr_compute(output[key], samples.x))\n",
    "        print(f'Estimated Noise for {key} is {noise_est[key][-1]:.4f} lead to psnr {psnr_est[key][-1]:.2f}')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.plotly.v1+json": {
       "config": {
        "linkText": "Export to plot.ly",
        "plotlyServerURL": "https://plot.ly",
        "showLink": false
       },
       "data": [
        {
         "name": "Repeating total variation (testing68 sample)",
         "type": "scatter",
         "uid": "8a7339f5-a22f-4d61-88fd-82268c6cfa3b",
         "x": [
          1,
          2,
          3,
          4,
          5,
          6,
          7,
          8,
          9,
          10,
          11,
          12,
          13,
          14,
          15,
          16,
          17,
          18,
          19,
          20,
          21,
          22,
          23,
          24,
          25,
          26,
          27,
          28,
          29,
          30,
          31,
          32,
          33,
          34,
          35,
          36,
          37,
          38,
          39,
          40,
          41,
          42,
          43,
          44,
          45,
          46,
          47,
          48,
          49
         ],
         "y": [
          27.281665381263284,
          27.17259474361644,
          27.04571185392492,
          26.92840270435109,
          26.822207478915942,
          26.72610928030575,
          26.63851255529067,
          26.55863442140467,
          26.484779414008646,
          26.416136685539694,
          26.35257863998413,
          26.293179820565616,
          26.237188872169046,
          26.184353996725644,
          26.134306767407587,
          26.08674888049855,
          26.04146463730756,
          25.998295307159424,
          25.9570811215569,
          25.917506330153522,
          25.879604199353388,
          25.84323277192957,
          25.80844309750725,
          25.774785967434156,
          25.742367099313174,
          25.710944344015683,
          25.68077502531164,
          25.651534725637998,
          25.623157052432788,
          25.595646100885727,
          25.568953093360452,
          25.543107818154727,
          25.51798439025879,
          25.493562502019547,
          25.469660141888788,
          25.44642260495354,
          25.423733290504007,
          25.401517503401813,
          25.37982688230627,
          25.358599129845114,
          25.337856012232162,
          25.317570209503174,
          25.297678554759305,
          25.278220513287714,
          25.25915002822876,
          25.24043344048893,
          25.222065308514765,
          25.204084396362305,
          25.18648885278141,
          25.169242045458624
         ]
        },
        {
         "name": "Repeating dictionary approach (testing68 sample)",
         "type": "scatter",
         "uid": "81334626-5343-4e41-aca5-bac69fb1e2ca",
         "x": [
          1,
          2,
          3,
          4,
          5,
          6,
          7,
          8,
          9,
          10,
          11,
          12,
          13,
          14,
          15,
          16,
          17,
          18,
          19,
          20,
          21,
          22,
          23,
          24,
          25,
          26,
          27,
          28,
          29,
          30,
          31,
          32,
          33,
          34,
          35,
          36,
          37,
          38,
          39,
          40,
          41,
          42,
          43,
          44,
          45,
          46,
          47,
          48,
          49
         ],
         "y": [
          28.159235870136936,
          28.19779292274924,
          28.17179396573235,
          28.134004648993997,
          28.09334129445693,
          28.052827807033765,
          28.013511797961065,
          27.97550173366771,
          27.93902857163373,
          27.904045441571405,
          27.870404523961685,
          27.83811330795288,
          27.807220290688907,
          27.77757209890029,
          27.749017743503344,
          27.72146396075978,
          27.69484581666834,
          27.66901787589578,
          27.64400083878461,
          27.61971397960887,
          27.59620949801277,
          27.573491797727698,
          27.55148688484641,
          27.530112687279196,
          27.50941214841955,
          27.489234980414896,
          27.46963663662181,
          27.45056132709279,
          27.431999430936926,
          27.413897710687973,
          27.396267947028665,
          27.3789961197797,
          27.362120740553912,
          27.34564750334796,
          27.32953985999612,
          27.313739271724927,
          27.298269973081702,
          27.283125035903034,
          27.26832128973568,
          27.253849141737994,
          27.239675521850586,
          27.225793417762308,
          27.212146310245288,
          27.198726794298956,
          27.185527997858383,
          27.17255948571598,
          27.159780418171604,
          27.147205044241513,
          27.13486090828391,
          27.122717100031235
         ]
        }
       ],
       "layout": {
        "title": {
         "text": "PSNR value vs number of repetitions"
        },
        "xaxis": {
         "automargin": true,
         "autorange": true,
         "title": {
          "text": "Iterations"
         },
         "type": "log"
        },
        "yaxis": {
         "automargin": true,
         "autorange": true,
         "title": {
          "text": "PSNR"
         },
         "type": "log"
        }
       }
      },
      "text/html": [
       "<div id=\"579eea32-f757-481e-8b96-116cdd5e1b00\" style=\"height: 525px; width: 100%;\" class=\"plotly-graph-div\"></div><script type=\"text/javascript\">require([\"plotly\"], function(Plotly) { window.PLOTLYENV=window.PLOTLYENV || {};window.PLOTLYENV.BASE_URL=\"https://plot.ly\";Plotly.newPlot(\"579eea32-f757-481e-8b96-116cdd5e1b00\", [{\"name\": \"Repeating total variation (testing68 sample)\", \"x\": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49], \"y\": [27.281665381263284, 27.17259474361644, 27.04571185392492, 26.92840270435109, 26.822207478915942, 26.72610928030575, 26.63851255529067, 26.55863442140467, 26.484779414008646, 26.416136685539694, 26.35257863998413, 26.293179820565616, 26.237188872169046, 26.184353996725644, 26.134306767407587, 26.08674888049855, 26.04146463730756, 25.998295307159424, 25.9570811215569, 25.917506330153522, 25.879604199353388, 25.84323277192957, 25.80844309750725, 25.774785967434156, 25.742367099313174, 25.710944344015683, 25.68077502531164, 25.651534725637998, 25.623157052432788, 25.595646100885727, 25.568953093360452, 25.543107818154727, 25.51798439025879, 25.493562502019547, 25.469660141888788, 25.44642260495354, 25.423733290504007, 25.401517503401813, 25.37982688230627, 25.358599129845114, 25.337856012232162, 25.317570209503174, 25.297678554759305, 25.278220513287714, 25.25915002822876, 25.24043344048893, 25.222065308514765, 25.204084396362305, 25.18648885278141, 25.169242045458624], \"type\": \"scatter\", \"uid\": \"8a7339f5-a22f-4d61-88fd-82268c6cfa3b\"}, {\"name\": \"Repeating dictionary approach (testing68 sample)\", \"x\": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49], \"y\": [28.159235870136936, 28.19779292274924, 28.17179396573235, 28.134004648993997, 28.09334129445693, 28.052827807033765, 28.013511797961065, 27.97550173366771, 27.93902857163373, 27.904045441571405, 27.870404523961685, 27.83811330795288, 27.807220290688907, 27.77757209890029, 27.749017743503344, 27.72146396075978, 27.69484581666834, 27.66901787589578, 27.64400083878461, 27.61971397960887, 27.59620949801277, 27.573491797727698, 27.55148688484641, 27.530112687279196, 27.50941214841955, 27.489234980414896, 27.46963663662181, 27.45056132709279, 27.431999430936926, 27.413897710687973, 27.396267947028665, 27.3789961197797, 27.362120740553912, 27.34564750334796, 27.32953985999612, 27.313739271724927, 27.298269973081702, 27.283125035903034, 27.26832128973568, 27.253849141737994, 27.239675521850586, 27.225793417762308, 27.212146310245288, 27.198726794298956, 27.185527997858383, 27.17255948571598, 27.159780418171604, 27.147205044241513, 27.13486090828391, 27.122717100031235], \"type\": \"scatter\", \"uid\": \"81334626-5343-4e41-aca5-bac69fb1e2ca\"}], {\"title\": {\"text\": \"PSNR value vs number of repetitions\"}, \"xaxis\": {\"automargin\": true, \"autorange\": true, \"title\": {\"text\": \"Iterations\"}, \"type\": \"log\"}, \"yaxis\": {\"automargin\": true, \"autorange\": true, \"title\": {\"text\": \"PSNR\"}, \"type\": \"log\"}}, {\"showLink\": false, \"linkText\": \"Export to plot.ly\", \"plotlyServerURL\": \"https://plot.ly\"})});</script><script type=\"text/javascript\">window.addEventListener(\"resize\", function(){window._Plotly.Plots.resize(document.getElementById(\"579eea32-f757-481e-8b96-116cdd5e1b00\"));});</script>"
      ],
      "text/vnd.plotly.v1+html": [
       "<div id=\"579eea32-f757-481e-8b96-116cdd5e1b00\" style=\"height: 525px; width: 100%;\" class=\"plotly-graph-div\"></div><script type=\"text/javascript\">require([\"plotly\"], function(Plotly) { window.PLOTLYENV=window.PLOTLYENV || {};window.PLOTLYENV.BASE_URL=\"https://plot.ly\";Plotly.newPlot(\"579eea32-f757-481e-8b96-116cdd5e1b00\", [{\"name\": \"Repeating total variation (testing68 sample)\", \"x\": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49], \"y\": [27.281665381263284, 27.17259474361644, 27.04571185392492, 26.92840270435109, 26.822207478915942, 26.72610928030575, 26.63851255529067, 26.55863442140467, 26.484779414008646, 26.416136685539694, 26.35257863998413, 26.293179820565616, 26.237188872169046, 26.184353996725644, 26.134306767407587, 26.08674888049855, 26.04146463730756, 25.998295307159424, 25.9570811215569, 25.917506330153522, 25.879604199353388, 25.84323277192957, 25.80844309750725, 25.774785967434156, 25.742367099313174, 25.710944344015683, 25.68077502531164, 25.651534725637998, 25.623157052432788, 25.595646100885727, 25.568953093360452, 25.543107818154727, 25.51798439025879, 25.493562502019547, 25.469660141888788, 25.44642260495354, 25.423733290504007, 25.401517503401813, 25.37982688230627, 25.358599129845114, 25.337856012232162, 25.317570209503174, 25.297678554759305, 25.278220513287714, 25.25915002822876, 25.24043344048893, 25.222065308514765, 25.204084396362305, 25.18648885278141, 25.169242045458624], \"type\": \"scatter\", \"uid\": \"8a7339f5-a22f-4d61-88fd-82268c6cfa3b\"}, {\"name\": \"Repeating dictionary approach (testing68 sample)\", \"x\": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49], \"y\": [28.159235870136936, 28.19779292274924, 28.17179396573235, 28.134004648993997, 28.09334129445693, 28.052827807033765, 28.013511797961065, 27.97550173366771, 27.93902857163373, 27.904045441571405, 27.870404523961685, 27.83811330795288, 27.807220290688907, 27.77757209890029, 27.749017743503344, 27.72146396075978, 27.69484581666834, 27.66901787589578, 27.64400083878461, 27.61971397960887, 27.59620949801277, 27.573491797727698, 27.55148688484641, 27.530112687279196, 27.50941214841955, 27.489234980414896, 27.46963663662181, 27.45056132709279, 27.431999430936926, 27.413897710687973, 27.396267947028665, 27.3789961197797, 27.362120740553912, 27.34564750334796, 27.32953985999612, 27.313739271724927, 27.298269973081702, 27.283125035903034, 27.26832128973568, 27.253849141737994, 27.239675521850586, 27.225793417762308, 27.212146310245288, 27.198726794298956, 27.185527997858383, 27.17255948571598, 27.159780418171604, 27.147205044241513, 27.13486090828391, 27.122717100031235], \"type\": \"scatter\", \"uid\": \"81334626-5343-4e41-aca5-bac69fb1e2ca\"}], {\"title\": {\"text\": \"PSNR value vs number of repetitions\"}, \"xaxis\": {\"automargin\": true, \"autorange\": true, \"title\": {\"text\": \"Iterations\"}, \"type\": \"log\"}, \"yaxis\": {\"automargin\": true, \"autorange\": true, \"title\": {\"text\": \"PSNR\"}, \"type\": \"log\"}}, {\"showLink\": false, \"linkText\": \"Export to plot.ly\", \"plotlyServerURL\": \"https://plot.ly\"})});</script><script type=\"text/javascript\">window.addEventListener(\"resize\", function(){window._Plotly.Plots.resize(document.getElementById(\"579eea32-f757-481e-8b96-116cdd5e1b00\"));});</script>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "iter_range = np.arange(1, repetitions)\n",
    "data = [go.Scatter(x = iter_range, y=psnr_est['tv'],\n",
    "                   name = f'Repeating total variation ({samples.dataset.split} sample)')]\n",
    "data.append(go.Scatter(x = iter_range, y=psnr_est['dict'],\n",
    "                       name = f'Repeating dictionary approach ({samples.dataset.split} sample)'))\n",
    "\n",
    "layout = go.Layout(\n",
    "         xaxis=dict(\n",
    "             type='log',\n",
    "             autorange=True,\n",
    "             title='Iterations',\n",
    "             automargin=True\n",
    "             ),\n",
    "         yaxis=dict(\n",
    "             type='log',\n",
    "             autorange  = True,\n",
    "             title='PSNR',\n",
    "             automargin=True\n",
    "         ),\n",
    "         title = (f'PSNR value vs number of repetitions')\n",
    "         )\n",
    "fig = go.Figure(data=data, layout=layout)\n",
    "ply.iplot(fig)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(1,3,figsize=(20, 10))\n",
    "axes[0].imshow(samples.x[0,:,:,].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "axes[1].imshow(output['tv'][0,:,:,].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "axes[2].imshow(output['dict'][0,:,:,:].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "fig.canvas.draw()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 4) Self-referential Denoising"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "status: iteration    0, loss: 0.00006 (avg PSNR est: 39.411), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00400 (avg PSNR est: 20.965), tol: 0.031%\n",
      "status: iteration 2000, loss: 0.00392 (avg PSNR est: 21.057), tol: 0.016%\n",
      "status: iteration 3000, loss: 0.00392 (avg PSNR est: 21.054), tol: 0.024%\n",
      "status: iteration 4000, loss: 0.00394 (avg PSNR est: 21.030), tol: 0.047%\n",
      "status: iteration 5000, loss: 0.00392 (avg PSNR est: 21.055), tol: 0.016%\n",
      "status: iteration 6000, loss: 0.00393 (avg PSNR est: 21.050), tol: 0.137%\n",
      "status: iteration 7000, loss: 0.00393 (avg PSNR est: 21.041), tol: 0.013%\n",
      "status: finished  8000  iterations in 4 minutes and 30.35 seconds. tol: 0.009%\n",
      "-- Iteration 0 done, tol: 100.000%, MSE: 0.003232, avg PSNR: 21.900\n",
      "status: iteration    0, loss: 0.00196 (avg PSNR est: 24.066), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00212 (avg PSNR est: 23.719), tol: 0.021%\n",
      "status: iteration 2000, loss: 0.00216 (avg PSNR est: 23.645), tol: 0.045%\n",
      "status: iteration 3000, loss: 0.00214 (avg PSNR est: 23.681), tol: 0.052%\n",
      "status: iteration 4000, loss: 0.00217 (avg PSNR est: 23.627), tol: 0.028%\n",
      "status: finished  4028  iterations in 2 minutes and 17.64 seconds. tol: 0.000%\n",
      "-- Iteration 1 done, tol: 13.599%, MSE: 0.002793, avg PSNR: 22.543\n",
      "status: iteration    0, loss: 0.00337 (avg PSNR est: 21.713), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00337 (avg PSNR est: 21.712), tol: 0.030%\n",
      "status: iteration 2000, loss: 0.00335 (avg PSNR est: 21.744), tol: 0.009%\n",
      "status: iteration 3000, loss: 0.00338 (avg PSNR est: 21.696), tol: 0.009%\n",
      "status: iteration 4000, loss: 0.00334 (avg PSNR est: 21.747), tol: 0.055%\n",
      "status: iteration 5000, loss: 0.00336 (avg PSNR est: 21.724), tol: 0.103%\n",
      "status: iteration 6000, loss: 0.00336 (avg PSNR est: 21.727), tol: 0.109%\n",
      "status: finished  6693  iterations in 3 minutes and 48.52 seconds. tol: 0.000%\n",
      "-- Iteration 2 done, tol: 0.701%, MSE: 0.002773, avg PSNR: 22.574\n",
      "status: iteration    0, loss: 0.00348 (avg PSNR est: 21.580), tol: 100.000%\n",
      "status: finished   150  iterations in 0 minutes and 5.11 seconds. tol: 0.000%\n",
      "-- Iteration 3 done, tol: 3.430%, MSE: 0.002868, avg PSNR: 22.432\n",
      "status: iteration    0, loss: 0.00399 (avg PSNR est: 20.977), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00390 (avg PSNR est: 21.082), tol: 0.026%\n",
      "status: finished  1538  iterations in 0 minutes and 52.53 seconds. tol: 0.000%\n",
      "-- Iteration 4 done, tol: 2.662%, MSE: 0.002792, avg PSNR: 22.546\n",
      "status: iteration    0, loss: 0.00359 (avg PSNR est: 21.438), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00354 (avg PSNR est: 21.499), tol: 0.017%\n",
      "status: finished  1547  iterations in 0 minutes and 53.17 seconds. tol: 0.000%\n",
      "-- Iteration 5 done, tol: 0.355%, MSE: 0.002782, avg PSNR: 22.561\n",
      "status: iteration    0, loss: 0.00351 (avg PSNR est: 21.533), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00348 (avg PSNR est: 21.574), tol: 0.017%\n",
      "status: iteration 2000, loss: 0.00348 (avg PSNR est: 21.573), tol: 0.023%\n",
      "status: iteration 3000, loss: 0.00347 (avg PSNR est: 21.587), tol: 0.035%\n",
      "status: iteration 4000, loss: 0.00348 (avg PSNR est: 21.569), tol: 0.019%\n",
      "status: finished  4120  iterations in 2 minutes and 22.37 seconds. tol: 0.000%\n",
      "-- Iteration 6 done, tol: 0.073%, MSE: 0.002784, avg PSNR: 22.558\n",
      "status: iteration    0, loss: 0.00349 (avg PSNR est: 21.560), tol: 100.000%\n",
      "status: finished   583  iterations in 0 minutes and 20.12 seconds. tol: 0.000%\n",
      "-- Iteration 7 done, tol: 0.439%, MSE: 0.002796, avg PSNR: 22.538\n",
      "status: iteration    0, loss: 0.00348 (avg PSNR est: 21.576), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00347 (avg PSNR est: 21.581), tol: 0.032%\n",
      "status: finished  1096  iterations in 0 minutes and 37.83 seconds. tol: 0.000%\n",
      "-- Iteration 8 done, tol: 0.347%, MSE: 0.002787, avg PSNR: 22.553\n",
      "status: iteration    0, loss: 0.00349 (avg PSNR est: 21.557), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00348 (avg PSNR est: 21.579), tol: 0.041%\n",
      "status: iteration 2000, loss: 0.00348 (avg PSNR est: 21.576), tol: 0.027%\n",
      "status: iteration 3000, loss: 0.00347 (avg PSNR est: 21.581), tol: 0.013%\n",
      "status: iteration 4000, loss: 0.00346 (avg PSNR est: 21.597), tol: 0.002%\n",
      "status: iteration 5000, loss: 0.00348 (avg PSNR est: 21.577), tol: 0.039%\n",
      "status: iteration 6000, loss: 0.00346 (avg PSNR est: 21.599), tol: 0.016%\n",
      "status: iteration 7000, loss: 0.00348 (avg PSNR est: 21.572), tol: 0.014%\n",
      "status: finished  7846  iterations in 4 minutes and 31.08 seconds. tol: 0.000%\n",
      "-- Iteration 9 done, tol: 0.023%, MSE: 0.002786, avg PSNR: 22.554\n",
      "status: iteration    0, loss: 0.00353 (avg PSNR est: 21.508), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00351 (avg PSNR est: 21.541), tol: 0.015%\n",
      "status: iteration 2000, loss: 0.00351 (avg PSNR est: 21.540), tol: 0.016%\n",
      "status: iteration 3000, loss: 0.00350 (avg PSNR est: 21.555), tol: 0.115%\n",
      "status: iteration 4000, loss: 0.00351 (avg PSNR est: 21.536), tol: 0.010%\n",
      "status: finished  4159  iterations in 2 minutes and 23.59 seconds. tol: 0.000%\n",
      "-- Iteration 10 done, tol: 0.277%, MSE: 0.002778, avg PSNR: 22.566\n",
      "status: iteration    0, loss: 0.00349 (avg PSNR est: 21.562), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00344 (avg PSNR est: 21.620), tol: 0.043%\n",
      "status: finished  1629  iterations in 0 minutes and 56.23 seconds. tol: 0.000%\n",
      "-- Iteration 11 done, tol: 0.139%, MSE: 0.002774, avg PSNR: 22.572\n",
      "status: iteration    0, loss: 0.00348 (avg PSNR est: 21.578), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00344 (avg PSNR est: 21.619), tol: 0.047%\n",
      "status: iteration 2000, loss: 0.00344 (avg PSNR est: 21.623), tol: 0.037%\n",
      "status: iteration 3000, loss: 0.00347 (avg PSNR est: 21.588), tol: 0.059%\n",
      "status: finished  3233  iterations in 2 minutes and 0.72 seconds. tol: 0.000%\n",
      "-- Iteration 12 done, tol: 0.282%, MSE: 0.002782, avg PSNR: 22.559\n",
      "status: iteration    0, loss: 0.00347 (avg PSNR est: 21.587), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00343 (avg PSNR est: 21.631), tol: 0.003%\n",
      "status: iteration 2000, loss: 0.00346 (avg PSNR est: 21.602), tol: 0.011%\n",
      "status: iteration 3000, loss: 0.00343 (avg PSNR est: 21.638), tol: 0.032%\n",
      "status: finished  3236  iterations in 3 minutes and 4.26 seconds. tol: 0.000%\n",
      "-- Iteration 13 done, tol: 0.156%, MSE: 0.002787, avg PSNR: 22.553\n",
      "status: iteration    0, loss: 0.00352 (avg PSNR est: 21.521), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00350 (avg PSNR est: 21.554), tol: 0.034%\n",
      "status: iteration 2000, loss: 0.00350 (avg PSNR est: 21.554), tol: 0.021%\n",
      "status: iteration 3000, loss: 0.00350 (avg PSNR est: 21.546), tol: 0.014%\n",
      "status: finished  3210  iterations in 3 minutes and 11.02 seconds. tol: 0.000%\n",
      "-- Iteration 14 done, tol: 0.036%, MSE: 0.002788, avg PSNR: 22.551\n",
      "status: iteration    0, loss: 0.00349 (avg PSNR est: 21.556), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00347 (avg PSNR est: 21.592), tol: 0.015%\n",
      "status: finished  1254  iterations in 1 minutes and 18.13 seconds. tol: 0.000%\n",
      "-- Iteration 15 done, tol: 0.297%, MSE: 0.002796, avg PSNR: 22.538\n",
      "status: iteration    0, loss: 0.00353 (avg PSNR est: 21.514), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00350 (avg PSNR est: 21.552), tol: 0.029%\n",
      "status: iteration 2000, loss: 0.00352 (avg PSNR est: 21.528), tol: 0.012%\n",
      "status: iteration 3000, loss: 0.00350 (avg PSNR est: 21.546), tol: 0.032%\n",
      "status: iteration 4000, loss: 0.00350 (avg PSNR est: 21.544), tol: 0.016%\n",
      "status: finished  4339  iterations in 4 minutes and 35.37 seconds. tol: 0.000%\n",
      "-- Iteration 16 done, tol: 0.804%, MSE: 0.002773, avg PSNR: 22.574\n",
      "status: iteration    0, loss: 0.00348 (avg PSNR est: 21.579), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00344 (avg PSNR est: 21.630), tol: 0.010%\n",
      "status: iteration 2000, loss: 0.00345 (avg PSNR est: 21.613), tol: 0.036%\n",
      "status: iteration 3000, loss: 0.00345 (avg PSNR est: 21.615), tol: 0.000%\n",
      "status: finished  3002  iterations in 3 minutes and 4.31 seconds. tol: 0.000%\n",
      "-- Iteration 17 done, tol: 0.192%, MSE: 0.002779, avg PSNR: 22.566\n",
      "status: iteration    0, loss: 0.00349 (avg PSNR est: 21.566), tol: 100.000%\n",
      "status: finished   597  iterations in 0 minutes and 35.74 seconds. tol: 0.000%\n",
      "-- Iteration 18 done, tol: 1.222%, MSE: 0.002813, avg PSNR: 22.511\n",
      "status: iteration    0, loss: 0.00342 (avg PSNR est: 21.644), tol: 100.000%\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "status: finished   721  iterations in 0 minutes and 36.10 seconds. tol: 0.000%\n",
      "-- Iteration 19 done, tol: 0.700%, MSE: 0.002793, avg PSNR: 22.543\n",
      "status: iteration    0, loss: 0.00350 (avg PSNR est: 21.547), tol: 100.000%\n",
      "status: finished   867  iterations in 0 minutes and 39.98 seconds. tol: 0.000%\n",
      "-- Iteration 20 done, tol: 0.340%, MSE: 0.002783, avg PSNR: 22.558\n",
      "status: iteration    0, loss: 0.00350 (avg PSNR est: 21.546), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00349 (avg PSNR est: 21.564), tol: 0.022%\n",
      "status: finished  1381  iterations in 1 minutes and 18.73 seconds. tol: 0.000%\n",
      "-- Iteration 21 done, tol: 0.198%, MSE: 0.002778, avg PSNR: 22.567\n",
      "status: iteration    0, loss: 0.00354 (avg PSNR est: 21.499), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00351 (avg PSNR est: 21.542), tol: 0.003%\n",
      "status: iteration 2000, loss: 0.00350 (avg PSNR est: 21.553), tol: 0.088%\n",
      "status: iteration 3000, loss: 0.00351 (avg PSNR est: 21.542), tol: 0.021%\n",
      "status: iteration 4000, loss: 0.00351 (avg PSNR est: 21.537), tol: 0.055%\n",
      "status: iteration 5000, loss: 0.00353 (avg PSNR est: 21.508), tol: 0.009%\n",
      "status: iteration 6000, loss: 0.00350 (avg PSNR est: 21.545), tol: 0.022%\n",
      "status: iteration 7000, loss: 0.00351 (avg PSNR est: 21.532), tol: 0.100%\n",
      "status: finished  8000  iterations in 5 minutes and 20.43 seconds. tol: 0.003%\n",
      "-- Iteration 22 done, tol: 0.148%, MSE: 0.002774, avg PSNR: 22.574\n",
      "status: iteration    0, loss: 0.00356 (avg PSNR est: 21.478), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00351 (avg PSNR est: 21.542), tol: 0.034%\n",
      "status: iteration 2000, loss: 0.00353 (avg PSNR est: 21.514), tol: 0.005%\n",
      "status: iteration 3000, loss: 0.00353 (avg PSNR est: 21.512), tol: 0.003%\n",
      "status: iteration 4000, loss: 0.00354 (avg PSNR est: 21.498), tol: 0.013%\n",
      "status: finished  4329  iterations in 4 minutes and 38.04 seconds. tol: 0.000%\n",
      "-- Iteration 23 done, tol: 0.280%, MSE: 0.002782, avg PSNR: 22.560\n",
      "status: iteration    0, loss: 0.00351 (avg PSNR est: 21.539), tol: 100.000%\n",
      "status: finished   722  iterations in 0 minutes and 42.56 seconds. tol: 0.000%\n",
      "-- Iteration 24 done, tol: 1.392%, MSE: 0.002820, avg PSNR: 22.500\n",
      "status: iteration    0, loss: 0.00352 (avg PSNR est: 21.526), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00347 (avg PSNR est: 21.582), tol: 0.008%\n",
      "status: finished  1061  iterations in 0 minutes and 58.21 seconds. tol: 0.000%\n",
      "-- Iteration 25 done, tol: 1.475%, MSE: 0.002779, avg PSNR: 22.565\n",
      "status: iteration    0, loss: 0.00350 (avg PSNR est: 21.549), tol: 100.000%\n",
      "status: finished   761  iterations in 0 minutes and 52.42 seconds. tol: 0.000%\n",
      "-- Iteration 26 done, tol: 0.365%, MSE: 0.002789, avg PSNR: 22.550\n",
      "status: iteration    0, loss: 0.00348 (avg PSNR est: 21.576), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00345 (avg PSNR est: 21.615), tol: 0.013%\n",
      "status: iteration 2000, loss: 0.00345 (avg PSNR est: 21.614), tol: 0.026%\n",
      "status: iteration 3000, loss: 0.00346 (avg PSNR est: 21.599), tol: 0.066%\n",
      "status: iteration 4000, loss: 0.00345 (avg PSNR est: 21.609), tol: 0.001%\n",
      "status: finished  4204  iterations in 4 minutes and 43.98 seconds. tol: 0.000%\n",
      "-- Iteration 27 done, tol: 0.437%, MSE: 0.002777, avg PSNR: 22.569\n",
      "status: iteration    0, loss: 0.00351 (avg PSNR est: 21.541), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00347 (avg PSNR est: 21.585), tol: 0.036%\n",
      "status: finished  1923  iterations in 2 minutes and 12.72 seconds. tol: 0.000%\n",
      "-- Iteration 28 done, tol: 1.105%, MSE: 0.002807, avg PSNR: 22.520\n",
      "status: iteration    0, loss: 0.00351 (avg PSNR est: 21.534), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00348 (avg PSNR est: 21.570), tol: 0.005%\n",
      "status: finished  1421  iterations in 1 minutes and 38.00 seconds. tol: 0.000%\n",
      "-- Iteration 29 done, tol: 0.388%, MSE: 0.002796, avg PSNR: 22.539\n",
      "status: iteration    0, loss: 0.00364 (avg PSNR est: 21.382), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00359 (avg PSNR est: 21.433), tol: 0.012%\n",
      "status: iteration 2000, loss: 0.00362 (avg PSNR est: 21.397), tol: 0.006%\n",
      "status: iteration 3000, loss: 0.00360 (avg PSNR est: 21.423), tol: 0.021%\n",
      "status: iteration 4000, loss: 0.00362 (avg PSNR est: 21.406), tol: 0.017%\n",
      "status: iteration 5000, loss: 0.00358 (avg PSNR est: 21.450), tol: 0.051%\n",
      "status: iteration 6000, loss: 0.00361 (avg PSNR est: 21.420), tol: 0.016%\n",
      "status: iteration 7000, loss: 0.00361 (avg PSNR est: 21.418), tol: 0.036%\n",
      "status: finished  7614  iterations in 8 minutes and 41.64 seconds. tol: 0.000%\n",
      "-- Iteration 30 done, tol: 0.410%, MSE: 0.002785, avg PSNR: 22.555\n",
      "status: iteration    0, loss: 0.00347 (avg PSNR est: 21.583), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00344 (avg PSNR est: 21.619), tol: 0.018%\n",
      "status: finished  1293  iterations in 1 minutes and 20.89 seconds. tol: 0.000%\n",
      "-- Iteration 31 done, tol: 0.552%, MSE: 0.002800, avg PSNR: 22.531\n",
      "status: iteration    0, loss: 0.00346 (avg PSNR est: 21.599), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00342 (avg PSNR est: 21.654), tol: 0.001%\n",
      "status: finished  1094  iterations in 1 minutes and 11.13 seconds. tol: 0.000%\n",
      "-- Iteration 32 done, tol: 0.329%, MSE: 0.002791, avg PSNR: 22.545\n",
      "status: iteration    0, loss: 0.00347 (avg PSNR est: 21.591), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00343 (avg PSNR est: 21.633), tol: 0.006%\n",
      "status: iteration 2000, loss: 0.00344 (avg PSNR est: 21.626), tol: 0.026%\n",
      "status: finished  2801  iterations in 3 minutes and 13.17 seconds. tol: 0.000%\n",
      "-- Iteration 33 done, tol: 0.185%, MSE: 0.002786, avg PSNR: 22.554\n",
      "status: iteration    0, loss: 0.00352 (avg PSNR est: 21.527), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00349 (avg PSNR est: 21.566), tol: 0.016%\n",
      "status: iteration 2000, loss: 0.00349 (avg PSNR est: 21.566), tol: 0.081%\n",
      "status: iteration 3000, loss: 0.00350 (avg PSNR est: 21.546), tol: 0.013%\n",
      "status: finished  3551  iterations in 4 minutes and 4.67 seconds. tol: 0.000%\n",
      "-- Iteration 34 done, tol: 0.212%, MSE: 0.002792, avg PSNR: 22.546\n",
      "status: iteration    0, loss: 0.00358 (avg PSNR est: 21.449), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00355 (avg PSNR est: 21.487), tol: 0.005%\n",
      "status: iteration 2000, loss: 0.00355 (avg PSNR est: 21.490), tol: 0.002%\n",
      "status: iteration 3000, loss: 0.00355 (avg PSNR est: 21.493), tol: 0.074%\n",
      "status: finished  3162  iterations in 3 minutes and 33.66 seconds. tol: 0.000%\n",
      "-- Iteration 35 done, tol: 0.279%, MSE: 0.002800, avg PSNR: 22.534\n",
      "status: iteration    0, loss: 0.00361 (avg PSNR est: 21.418), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00355 (avg PSNR est: 21.485), tol: 0.028%\n",
      "status: iteration 2000, loss: 0.00356 (avg PSNR est: 21.477), tol: 0.101%\n",
      "status: iteration 3000, loss: 0.00358 (avg PSNR est: 21.448), tol: 0.021%\n",
      "status: iteration 4000, loss: 0.00361 (avg PSNR est: 21.417), tol: 0.117%\n",
      "status: iteration 5000, loss: 0.00358 (avg PSNR est: 21.457), tol: 0.019%\n",
      "status: finished  5159  iterations in 5 minutes and 33.95 seconds. tol: 0.000%\n",
      "-- Iteration 36 done, tol: 0.391%, MSE: 0.002789, avg PSNR: 22.550\n",
      "status: iteration    0, loss: 0.00352 (avg PSNR est: 21.520), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00353 (avg PSNR est: 21.508), tol: 0.064%\n",
      "status: iteration 2000, loss: 0.00349 (avg PSNR est: 21.563), tol: 0.019%\n",
      "status: iteration 3000, loss: 0.00350 (avg PSNR est: 21.550), tol: 0.028%\n",
      "status: iteration 4000, loss: 0.00351 (avg PSNR est: 21.539), tol: 0.061%\n",
      "status: iteration 5000, loss: 0.00349 (avg PSNR est: 21.564), tol: 0.037%\n",
      "status: finished  5976  iterations in 6 minutes and 40.93 seconds. tol: 0.000%\n",
      "-- Iteration 37 done, tol: 0.162%, MSE: 0.002793, avg PSNR: 22.542\n",
      "status: iteration    0, loss: 0.00351 (avg PSNR est: 21.533), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00348 (avg PSNR est: 21.579), tol: 0.091%\n",
      "status: iteration 2000, loss: 0.00348 (avg PSNR est: 21.576), tol: 0.133%\n",
      "status: iteration 3000, loss: 0.00347 (avg PSNR est: 21.591), tol: 0.075%\n",
      "status: iteration 4000, loss: 0.00347 (avg PSNR est: 21.582), tol: 0.026%\n",
      "status: iteration 5000, loss: 0.00345 (avg PSNR est: 21.616), tol: 0.345%\n",
      "status: iteration 6000, loss: 0.00348 (avg PSNR est: 21.577), tol: 0.007%\n",
      "status: finished  6036  iterations in 6 minutes and 53.86 seconds. tol: 0.000%\n",
      "-- Iteration 38 done, tol: 0.598%, MSE: 0.002777, avg PSNR: 22.569\n",
      "status: iteration    0, loss: 0.00350 (avg PSNR est: 21.551), tol: 100.000%\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "status: iteration 1000, loss: 0.00348 (avg PSNR est: 21.576), tol: 0.004%\n",
      "status: iteration 2000, loss: 0.00345 (avg PSNR est: 21.608), tol: 0.077%\n",
      "status: finished  2921  iterations in 3 minutes and 20.03 seconds. tol: 0.000%\n",
      "-- Iteration 39 done, tol: 0.220%, MSE: 0.002783, avg PSNR: 22.558\n",
      "status: iteration    0, loss: 0.00344 (avg PSNR est: 21.619), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00342 (avg PSNR est: 21.653), tol: 0.038%\n",
      "status: iteration 2000, loss: 0.00343 (avg PSNR est: 21.639), tol: 0.002%\n",
      "status: iteration 3000, loss: 0.00341 (avg PSNR est: 21.667), tol: 0.078%\n",
      "status: iteration 4000, loss: 0.00342 (avg PSNR est: 21.653), tol: 0.006%\n",
      "status: iteration 5000, loss: 0.00342 (avg PSNR est: 21.652), tol: 0.015%\n",
      "status: finished  5511  iterations in 6 minutes and 18.46 seconds. tol: 0.000%\n",
      "-- Iteration 40 done, tol: 0.149%, MSE: 0.002787, avg PSNR: 22.553\n",
      "status: iteration    0, loss: 0.00353 (avg PSNR est: 21.517), tol: 100.000%\n",
      "status: finished   618  iterations in 0 minutes and 42.56 seconds. tol: 0.000%\n",
      "-- Iteration 41 done, tol: 1.145%, MSE: 0.002819, avg PSNR: 22.503\n",
      "status: iteration    0, loss: 0.00343 (avg PSNR est: 21.636), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00341 (avg PSNR est: 21.661), tol: 0.018%\n",
      "status: iteration 2000, loss: 0.00341 (avg PSNR est: 21.667), tol: 0.002%\n",
      "status: iteration 3000, loss: 0.00342 (avg PSNR est: 21.649), tol: 0.038%\n",
      "status: finished  3190  iterations in 3 minutes and 35.47 seconds. tol: 0.000%\n",
      "-- Iteration 42 done, tol: 1.241%, MSE: 0.002784, avg PSNR: 22.557\n",
      "status: iteration    0, loss: 0.00346 (avg PSNR est: 21.604), tol: 100.000%\n",
      "status: finished   579  iterations in 0 minutes and 39.84 seconds. tol: 0.000%\n",
      "-- Iteration 43 done, tol: 1.541%, MSE: 0.002827, avg PSNR: 22.489\n",
      "status: iteration    0, loss: 0.00345 (avg PSNR est: 21.615), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00341 (avg PSNR est: 21.659), tol: 0.021%\n",
      "status: iteration 2000, loss: 0.00338 (avg PSNR est: 21.701), tol: 0.042%\n",
      "status: iteration 3000, loss: 0.00343 (avg PSNR est: 21.641), tol: 0.014%\n",
      "status: iteration 4000, loss: 0.00348 (avg PSNR est: 21.571), tol: 0.084%\n",
      "status: iteration 5000, loss: 0.00341 (avg PSNR est: 21.664), tol: 0.041%\n",
      "status: finished  5185  iterations in 5 minutes and 47.75 seconds. tol: 0.000%\n",
      "-- Iteration 44 done, tol: 1.399%, MSE: 0.002787, avg PSNR: 22.552\n",
      "status: iteration    0, loss: 0.00348 (avg PSNR est: 21.574), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00346 (avg PSNR est: 21.603), tol: 0.026%\n",
      "status: iteration 2000, loss: 0.00344 (avg PSNR est: 21.619), tol: 0.021%\n",
      "status: iteration 3000, loss: 0.00343 (avg PSNR est: 21.639), tol: 0.108%\n",
      "status: iteration 4000, loss: 0.00345 (avg PSNR est: 21.618), tol: 0.054%\n",
      "status: iteration 5000, loss: 0.00345 (avg PSNR est: 21.612), tol: 0.050%\n",
      "status: iteration 6000, loss: 0.00345 (avg PSNR est: 21.616), tol: 0.012%\n",
      "status: iteration 7000, loss: 0.00344 (avg PSNR est: 21.621), tol: 0.016%\n",
      "status: finished  7393  iterations in 8 minutes and 25.32 seconds. tol: 0.000%\n",
      "-- Iteration 45 done, tol: 0.113%, MSE: 0.002784, avg PSNR: 22.557\n",
      "status: iteration    0, loss: 0.00352 (avg PSNR est: 21.522), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00349 (avg PSNR est: 21.562), tol: 0.027%\n",
      "status: finished  1420  iterations in 1 minutes and 32.93 seconds. tol: 0.000%\n",
      "-- Iteration 46 done, tol: 0.053%, MSE: 0.002782, avg PSNR: 22.560\n",
      "status: iteration    0, loss: 0.00351 (avg PSNR est: 21.535), tol: 100.000%\n",
      "status: finished   320  iterations in 0 minutes and 22.04 seconds. tol: 0.000%\n",
      "-- Iteration 47 done, tol: 4.158%, MSE: 0.002898, avg PSNR: 22.387\n",
      "status: iteration    0, loss: 0.00416 (avg PSNR est: 20.799), tol: 100.000%\n",
      "status: iteration 1000, loss: 0.00407 (avg PSNR est: 20.890), tol: 0.061%\n",
      "status: iteration 2000, loss: 0.00407 (avg PSNR est: 20.899), tol: 0.018%\n",
      "status: iteration 3000, loss: 0.00411 (avg PSNR est: 20.853), tol: 0.099%\n",
      "status: finished  3705  iterations in 3 minutes and 41.19 seconds. tol: 0.000%\n",
      "-- Iteration 48 done, tol: 3.869%, MSE: 0.002786, avg PSNR: 22.555\n",
      "status: iteration    0, loss: 0.00358 (avg PSNR est: 21.451), tol: 100.000%\n",
      "status: finished   457  iterations in 0 minutes and 20.53 seconds. tol: 0.000%\n",
      "-- Iteration 49 done, tol: 1.143%, MSE: 0.002818, avg PSNR: 22.505\n",
      "status: finished  50  iterations in 160 minutes and 43.72 seconds. tol: 1.143%\n"
     ]
    }
   ],
   "source": [
    "batch_small = 200\n",
    "small_samples = Sur.data.Samples(datasetTrain, batch_small, device=device, dtype=dtype)\n",
    "self_samples = Sur.data.Samples(datasetTrain, batch_small, device=device, dtype=dtype)\n",
    "self_samples.x = small_samples.y.clone()\n",
    "self_samples.redraw_noise()\n",
    "\n",
    "energy = energy_setup(model_setup)\n",
    "    \n",
    "training_setup = Sur.training.default_setup('DiscriminativeLearning')\n",
    "\n",
    "subroutine = Sur.training.DiscriminativeLearning(energy, loss, self_samples, training_setup,\n",
    "                                                 algorithm='predictive-primal')\n",
    "iterative_setup = dict(linearizer_iterations=50, extrapolation_choice='n-margin',\n",
    "                       backtracking=False, check_loss=True, verbose=True, tolerance=1e-4, stochastic=False)\n",
    "\n",
    "optimizer =  Sur.training.IterativeLearning(subroutine, iterative_setup)\n",
    "optimizer.run();\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "21.934726610183716\n",
      "TRAINING | MSE: 0.002786 | avg PSNR for these samples: 22.555,  Time: 00:07 mins\n",
      "TRAINING | MSE: 0.001197 | avg PSNR for these samples: 26.560,  Time: 00:07 mins\n"
     ]
    }
   ],
   "source": [
    "print(Sur.psnr_compute(optimizer.x, small_samples.x))\n",
    "img_denoised, _ = Sur.training_error(energy, loss, self_samples);\n",
    "img_denoised, _ = Sur.training_error(energy, loss, small_samples);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Dictionary norm is 1.446.\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x720 with 49 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "Sur.visualize(energy.operator)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABIEAAAFrCAYAAABG0ZmCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzsvVnQXdV1qDuWQIAxMo0AISE69X37q0MCJNHIsUPAsWMbHKejEldSpyq3Elea85aq+5Cbh9yXpHLjqjSQuAwhdtGDMCCEJFDf982POhAC0VkGJCS87wO4TrTGN8USxjLS+r6qUz4a2XuvuWYzxlyb/X+z6nQ6ISIiIiIiIiIipzc9ftUNEBERERERERGRXz5+CSQiIiIiIiIi0gL8EkhEREREREREpAX4JZCIiIiIiIiISAvwSyARERERERERkRbgl0AiIiIiIiIiIi3AL4FERERERERERFqAXwKJiIiIiIiIiLSAX+hLoKqqvlhV1ZaqqrZXVfVXn1ajRETk9MA6ISIix8M6ISJycqk6nc4ne2NVnRERWyPi5ojYGxHLIuKOTqez8dNrnoiInKpYJ0RE5HhYJ0RETj5n/gLvnRwR2zudTndERFVV90bEbRFRTNrnnXde56KLLjomRl9CnXPOObmhZ3JTDx8+nGI9e/ZMsQ8++CDFevTIP4Siz6Nr0+dFRJx99tkpdvTo0RSj+z5y5EiKff7zn0+xn/70p43eG8Ftp2tTG6l/6L6bjuH777+PbSTo2tTGM844I8XOPffcFHvvvfcavS6iedsPHjyYYk3n2aFDh1KsNMepf3v16tXodXTfdJ2qqlKM1lEp3vQ6Z511VqP30thQf0fwmqP+pc+ke2k6T0tfoNNn0vqs3/fBgwfjvffeywNxavOJ6kTv3r2Pif3sZz9Lr6O1T/MrgufY5z73uRSjHENrg+YI5Y1SXqZ2Nq0J1Mam64X6sXRtum96P+W3pu+lMXj33XexjTTetNbefPPNFDv//PNTjPJG0/1EBPc5vf8nP/lJilH/0BicSHuof2lOUj9SnzfNYxdeeCG2h17bdP9AY0Pz+bzzzksxyv0Rzfd79JkUo30Y3V+prtPcp7zyzjvvHPPvn/zkJ/Huu+9aJ+B5gtZV0zUQwXOWXktrja7dtJ6U8nLTZxl6P72O+qI+vyLKeyDqC1q/dO2m902fR/WS+rZE02cZWqtNnwfodRGcy2ieUX77RfYjdM+l99NzJo1D0+tQrJQHaY5TDm9a/ymnn8g+g9pJ9031n9p4IvOUoM+k+VO/n4MHD8ahQ4c+tk78Il8CXR4Re/7Hv/dGxJTjveGiiy6KP//zPz8mRklg2LBh+F5i27ZtKda/f/8Ue+utt1KMJsbOnTtTjDY5b7/9NrZnwIABKfbGG2+kGE3U1157LcWmTMldOn/+/BTbt28ftueSSy5JMepzujYlhtdff73R5w0ZMiTFdu/ejW1s+iUHtfELX/hCinV1daXY+vXrU2zs2LHYHmr7yy+/nGLPPvtsitEDCPXjli1bUuyCCy7A9lCxmj17dopRslm7dm2KXXzxxSlGiaZv377Ynn79+qXYhg0bUozu54orrkixjRvzPm/cuHEptmDBAmwPrTlqD82LPn36pNjevXtTrOlGp/SZtD7r933ffffh553inHCd6N27d/zlX/7lMTHKl/Rgf+WVV+Jnrlu3LsVGjx6dYq+++mqK0QaU6sSIESNSjPJGBNcommM0Fw8cOJBiEyZMSDFaL7Thj+D+pfumzRltcilGX8QNHz48xVavXo1tpDp82WWXpdgPf/jDFJszZ06KDRo0KMVefPHFFCvlQZo/3d3dKfb000+nWNOHku3bt6cY3XME9+/QoUNTjPLyqlWrUuzyyy9PsZdeeinFvv71r2N76LX0hRhtsCmnz5s3L8Wuu+66FKM9YQTPSVrHU6dObXSdpuurtHcdNWpUiu3atSvFVqxYccy/7777bvy8U5xP9DzxF3/xF8fE6AGOnidofxnBdZrWC6012kPRfrnpfwiL4P07fVlAeZn2orS3XbZsWYqV9upUcylv0bWbfrFLdZDqJdXBEvSsR89wtDemGkV1mXJtRMRVV12VYrTPoOc6euahLz4oj5XmOM3TyZMnpxg9T9C8oC/IaY6X6tall16aYps3b04xWof0Xmoj7c1K+wyaA3v27Ekxqv+U65vO09IXwbTm9u/fn2L1+6F9EPGLOIHoG6a0equq+qOqqpZXVbWc/suJiIictlgnRETkeFgnREROMr/Il0B7I+J//mf9/hGR/rNnp9P5XqfT6ep0Ol30jaGIiJy2WCdEROR4WCdERE4yv8ifgy2LiMFVVV0TES9FxDcj4s7jvaGqqvRTNvr5Pv2EvvSTWvrTC/o5GP3Uin6ySD/xWr58eYqVfiJOf/pF16GfItJPIylGP6ss/XyTfkJHfUF/DkA/gbvjjjtS7KmnnkqxE/kZO/10kH4OePXVV6dY059q0k+CS39+ReNNP6Ml6Kd71B76U7SSo4j+ixetkbpHJYJ/5kl/Gkc/Oy79uRP9zJ/ukf5cjvqR1jatIxrDCP7JNfUZtZH+LIN+Lks/O6U2RvDfGtPPf+t/m01/q30acMJ1otPppLk3ePDg9Dr6CXzpJ7W03rZu3ZpilGPozxVpLtLPi+mnxRH852l0bcpRNN/pzyDoT2WXLl2K7bn11ltTjOYj/Vk1/enlrFmzUmzlypUptmbNmhSjnB7Bf2pDf1Z77bXXphjNn02bNqUY9TftCSI4j9KfMdFP0SnvUN2heym5DGg90JyiPzmiP9PdsWNHitHY1P9c6edQX9C+gPIy/dyd/hyQ6i3VvAj+Uz0aBxpXyvWUa+ja9Kcapc+kPVe9NpecUKc4n+h5or53vOaaa9LrKF/SnweV4vTnHKW9UR16PqE/OaV5HMHPCbT+6TmB8mVTP1npT1jo2aOpy4yegyi/0fqj/VPpmZDa+Morr6QYaQ0oZ1G76U9lS/t3qjO0n6RcRDmP+mLkyJEpdiLuOJoX9OdklP+pnpyIg4fyMsWoTtB6pz6j+Vx6HqU1R2uppGBpAu0zSv1D16b7qe8Bmz5PfOIvgTqdztGqqv5XRMyNiDMi4l87nU7eDYqISCuxToiIyPGwToiInHx+kV8CRafTeSwiHvuU2iIiIqcZ1gkRETke1gkRkZPLL+IEEhERERERERGRUwS/BBIRERERERERaQEVSb1+WfTr169z1113HRMj6SIJv1577TX8TJJxkdjp85//fIqRGJREUySkIjFY6dpNIaEZiQJJfEWy3ggWfpFYjKTCJAYjSRW9jk5uoNdFRLz//vspRlJTkv1Su5vKmUlyGsHzr+m4luRedUjsVZLIUl/Q+5vKnek6JHK7//77sT0ktSOpKYnJjh49mmLf+ta3Uoxkc6UxIMHbVVddlWKUQ2gdU04k8SL1Q0TEkSNHUozaXhcgPvroo3HgwIHT0g59IvTr16/zne9855jY7Nmz0+tonFatWoWfSYJckquTIJfE0OPHj08xkj2XBKTnnHNOipEwdODAgSlG+X/t2rUpRmu/JMMnwTKJhkmaTnmH3lsSedYhyXUE57xJkyalGImKqR8pb9AYUP6NYOEv1TIaa/pMGgPKG6NGjcL2UI6i8aI5STmL1gwd4vDf//3f2B4a7yVLljT6TNqHTZ8+PcVILE77qAiW1ZIIlvZMdBBIU4lwabzotZTTnnvuuWP+PXfu3HjjjTdaXyf69+/f+dM//dNjYiSupZxX2vtR7qE9L0l3aTybSmpLYmjaT1JebyqqJskx5Sx6RojgPTjlKNr/N8079JxAz2+l5wnqC3pOpL0o5Sy6Nu0lS3t/6gvqR6qZTZ8nmj7zRvCcpPaU9gpNrkM174knnsD3U/6nvE65kZ4xvvzlL6cYSfhp/CM419Ne4b333ksx2guRiJvWXGk+0xqh+6k/Tzz22GPx+uuvf2yd8JdAIiIiIiIiIiItwC+BRERERERERERagF8CiYiIiIiIiIi0AL8EEhERERERERFpAWxg/CXRq1evmDlz5jExkk+98MILKTZo0CD8TJIzkhiQxFckgCJBG0mcSvJBElWRiGv37t2NrkMCO3pvSeJFsrH9+/en2LBhw1Js27ZtKUYyXJJcksSLBFcREf369UuxV155JcVovKh/SJBJwk+Sl5U+c9y4cSn2X//1Xyk2efLkFKOxIdlsSZxK7aRxINEtCZKpH0mGWWoPjQ3JL0nESetm3rx5KUZrgcYgIuLKK69MMeofkkCTeJ3EgkOHDk2xhQsXYntmzJiRYiRarK+bHj38Tj7iQ9HgxIkTj4mRiLMuTI0o14nFixenGK0DEtfSeFLdoTxGAr8IFjmSbJrWBuUDWvskJKR8EMHrkvJo//79U4yExiVpeh0SF5dqGY0X9S+9jvqC+ozqRKlukSxywoQJKfZP//RPKXbnnXemWH1vFPGhBLgOiWojIkaMGJFiJNhsKnwmOTPtPUjGW7o2zT/aP9AcX716dYpt3rw5xb75zW9ieyiHUI2j/N/d3Y2fWefGG29MMdonRETccsstKUbjXe8L68SHnHvuuSln0l6b8hMdKBLBeZn2sjRHSABMc45eV4KuQ2u1VGfq0N6WngdKUO6hz6TDR0im3HTtk7C7dKgRrWmS+NJcoVz/+uuvpxjVDnr2K30mHUCxdOnSFKMDDZoeIFQSQ1Of03qgtUCvo7GmeVKSn+/bty/FBg8enGJUwylGeziSko8ZMwbbQ/dIc4Xqf0moXocE0vS8HcFzoEn+ob0VYTUREREREREREWkBfgkkIiIiIiIiItIC/BJIRERERERERKQF+CWQiIiIiIiIiEgL8EsgEREREREREZEWcFJPBzt8+HA6XeLAgQPpdWSHP3r0KH4mmbPJDt6nT58UI3s6GdHJfk8m9wg+gYjs8nQ/9DoywdPpACtXrsT2kFWfTvOg06rI5k8xOnmNzOl0KklExPLly1OM7ptOmxowYECK0QkmZMSnEwMieE4+9thjKUansVC7X3rppRRbv359itFJEhERv/Ebv5FidEIXGfmpb2nek2W/dDoYnSRGJ3TROr7//vtTjNbCd7/73cbtob6k0wDefffdRtemkwAoB9B6jeDTDekz6+uQ2txG3n///XT6Bq0rOmGhVCemT5+eYps2bUqxrq6uFKOTV3r16pVilBvptJ+IiFtvvTXF6AQzOmWL5nH9NLUIzhGPPPIItodOVaM1TSc00clJ27dvTzE6QZBO4qC+jYhYsWJFitEJGHTSJeU8OmGDYnv37sX2UH6jk6DoxCiaK3SSJ+1H6HURfOIM5R0aLzrhhfYE1Bd0Kl4EnwRG85lOoXvwwQdTjE7F+uu//usUK52KQifo0H1T/ad9FO096NSgN998E9uzaNGiFKN9av2Utvnz5+PntY2jR4+mHEd7oNK+iqCTCWk/ee655za6Du2Nab6XThWiHEN7Ndob0wmrdFIV5TxaKxF8IlfT049pv025jNY5rQt6xojgHNV0DlA9orGhZzDq7wieA2vXrk0xyo104hjtRWm8SidY016B9vQ0XnR6ZtMTdkv7W2on1TLq82eeeSbFaC38/u//foqV6hbNH1o3dB26b9rPND0JPIJPN6ST1urfhZTGv46/BBIRERERERERaQF+CSQiIiIiIiIi0gL8EkhEREREREREpAX4JZCIiIiIiIiISAs4qWLoo0ePJskWCdZI9kzCvQgWJJE4c8yYMSlGMmVqz3nnnZdiJXklyZhIKkViKBINb926NcVIpkbSzYiI559/PsWof0i6S4JNElWSaIzEl7t27cI2kuCVBHQkTiMp3ebNm1OMhGZ03Qgem6ZiURKiUn/TvCXxWQSLBOuyyAieFyT3JHnihAkTUoz6MSJi6tSpKUbz79lnn02xoUOHphj1I8n0SuI0kufR+iSRJ0k2SQ5J86wkuqPXDh48OMXqotuSWLCN1Nf/hg0b0mumTJmSYrRWIjiP0viRGJLE0CR7p/x0ww03YHtoLlOdodxx0003pdh//ud/phiJrydNmoTtIZHnunXrUozk/iTNJZHn6NGjU2zZsmUpVpKkUv/QeFOf0YENJJqm+nbttddie2hPQrmD2kO1kPYOJOzesWMHtoeEsZRvae5RrifJ+te+9rUUo0MTIljG3rdv3xR78sknU4z2a7Reqd5Sno9g+SnNXRoHkjvTHoek1JRTIlhMSnWiLiqlcW4jR48eTYczkKyVchHl9Ijmwl+S5pOIn/YmJwLNJ9ov0VqlXET3QvOpJM3duXNnitHaoH4kkTPdH9UY2tuSwD2Cn+HoOrSnp7VP+bvpYRERPF6Ud2gcKMfQ8yQdfFOSYdP76VmIxNC076HX9e/fP8VKBxqMHTs2xYYMGZJitFegwwdorCm2b98+bA+NDdUekrZT/9C8oLlXep6gdUzrs/59Bo0z4S+BRERERERERERagF8CiYiIiIiIiIi0AL8EEhERERERERFpAX4JJCIiIiIiIiLSAk6qGPqss85KgkaSK5Lwi0RKEVmaF8EyxTVr1qQYSdtIHrh8+fIUI9FkBMu4SPhFsrBVq1alGEnOCBLpRkTcfvvtKUaCLhJnkYiThHh79uxJMRJklfqMxH0kOiPJMY1Xjx75u02SV5ekfVdccUWKzZ07N8VIfkefSeIzEmSSyDWCZYfbtm1LMZKa0jyjNj7++OMpVhJsLl68OMVIVEZzkmS+JFijeymJ0y6++OIUoznw8MMPpxjNXco1JIubM2cOtofWEomJ63OXxqqNVFWV1haJ3alPSSgfwbJhytWrV69OMVp/JK8kCT8J9yNYuk8CQZIXk8x8xowZKXYiovHrr78+xUi6TOuK8jqJIWkMR44cmWI0rhEsryeo9tDYUI2hdlMuioiYPHlyij366KMpRu2mWj979uwUGz58eIpRXY7gvNPd3Z1iJJGluUf3fc8996TYuHHjsD3U57fddluK0R5n5syZKUbjRXvF0iEidOAIsXbt2hQjQSsJY6nWlw7tIBnrAw88kGL1gwYoH7WRHj16pDlBOYtE6LRWIlhyTOuN5hjt82mtNd1zlD6T8jrtHUh8S59He5sSlPPoM6lO0GEodJgC7elJxFxqN12HDuugPQHNH7oO5ZLS8wTlVjroguYK7VHo2YHuef/+/dgeuk7TvmgqJaf7K4mzqUbRvKDnVto/UJ0geXVJnE3tJMkyfadAe4+mhwBR7YjgXEV7qU/6POEvgUREREREREREWoBfAomIiIiIiIiItAC/BBIRERERERERaQF+CSQiIiIiIiIi0gJOqhj60KFDSWhLsleS3pJEL4KlbSROI/EViZhIDEfSTZJzRUQsXLgwxUh0Nnjw4BTbvXt3ipFAiuSnpfaQtIvaQ6+75pprUoxkkST2JXEayRUjWLp16623phiNKwk2SaZJsq+SOJXGoSQlrkPzlCR5JIEuyQqJnTt3ptikSZMavY644YYbUozWTAQLo0lURiI3ktXRHKd+LInOSJRJ8mqSC9J6J4kgzXESPEZE7Nq1K8VIkl0XWp+IoPF05tChQ2l9UJ6nsaPcFsHrnwTwdNAAiccJkj2T4DCCcyGJy6n2LFq0KMWoL0iQWGrPBRdckGJUhy+77LIUIzEw3R9Jgf/jP/4jxUq5liTAVCdILD1v3rwUI2EjSZxLIt5XXnklxah/SExLYnGqE0uWLEkxOpwhgsebBJ007ylnUT265ZZbGr03goXhJNSlvcfrr7+eYlQzad/Tu3dvbA/VexIxk6ia1gIdukE16p133sH20N6FBMF1ATWJeNvI4cOH07jQ2NE+pCSpJVksrX8SzVJupdfR+isdALN9+/YUoxpH90P7HboOzdmSfJ6g9UtzlHIevZfyIN1f6TAcYujQoY2uTbWD8i31T2ld0vyhPS9BOZ32iTTWZ599Nn4mtZMk4ldddVWKNT18hsaacnoE77epppComvbldH9U60uHiNB40TMhPR/RYQG0n6HaSmMYwc8ZtN+r1w5aR4S/BBIRERERERERaQF+CSQiIiIiIiIi0gL8EkhEREREREREpAX4JZCIiIiIiIiISAs4qSbSd955J5YtW3ZMjATCJOIsScBIulRVVYqREI0kYCTIJAlYSbpE1yYJFEkOSSBYEp3WIYFcRMT48eNTjER5JMkiyRWJU0kM1lSwFsHiTJKfkkCaBFkkhiQhJQn6IljwR31G7SFRGQna6BolUdlFF12UYiQqozaOHDmy0bVpDEp0d3en2H333Zdi06ZNSzGSjdMYkohz4sSJ2B6SSDcVVX/5y19OMZL2kbyOJJIREc8880yK0RiuXr36mH/Temsjhw4dio0bNx4To7GnvEpzKSKiT58+ja5NdYbWOdUtWucl2TdJjkmmSHmQPpM+j6SkJTn/lClTUowktVRnSPZOUmBqI61TEmxH8PqdP39+o2uTxJkE9zt27EixkrSVaiuJM6n20Nyl/iHpdj1v/BySiFLeoVxPh2R88YtfTLGXXnopxUoC0hdeeCHFHn300RTr169fipGUlISd1N+l+UPzmeoM5YA777wzxWhvRn1Le8oIlrHSOqzngJKovG289957sXbt2mNitMek+UXzJuLEhMh1aP3S80kpnxAkM6d6RDm46T6fKMnMqX+aCpqpHjU93IH6tnRgAwmfN2/enGLUjyQfJik19Q/l2gjuM5Ik09jQ8wTNKdqzlg6aoXxNfU7vp/VF99308KEI3gOsWLEixYYMGZJitLZpbGi9kyw8gucV1T0SUI8ePTrFaAxLNYGgvUuT7xRo/RP+EkhEREREREREpAX4JZCIiIiIiIiISAvwSyARERERERERkRbgl0AiIiIiIiIiIi3gpIqhq6pKQiMSrpJcmcRXESxNIsEqXeeyyy5LMZJmkUCsLi79OdR2EnnS6/r3759iJFiePHlyipVEbiQqJKEViaqo3SQBe/DBB1OMxNkkbI5gQTONIcndSLrVVEBHQuEI7rOlS5emGIlcST5J90IyPRJsR7AwjoTGJAwlESzJ3UiIV5e4/xy6R5JA02eS4G/Pnj0pRhI5GusIFryRCJiE2CSl69u3b4qtX78+xUhKFxExa9asFHvjjTdSrH6PJYlw2zjzzDNTrqDcSHmZXhfBwl+CagLlLZI9knxw5cqVeB2SD9J6oRw1ePDgFHviiSdS7NZbb00xElJGsICUahTljhEjRqQY1aOHH3640efRdSO4RlG9JvkwvXfQoEEptmrVqhQjEWdExM6dO1OM6jXlS6rh9HnPP/98itFeJoJz5q//+q+n2OLFi1OMJJLUHhJkkrw8gtfSpEmTUozqDB0gQPJbyqsl6SYJXs8///wUI8H2hg0bUuz2229PMRKiUj9EsASaDnyoz1Pai7SRTqeT9rKUD2hdlA52ob0f7b9I7kzrnK5Ddat0MEdTkTO9jtpD16G9UikHN5VN016Nnjtoz9N0P1iSeNN6o3Y3Fe7S8wTV6lJ76P2076RaT7WH+rvpPCkxatSoFKNxoOcW2stQf5cOL6K5Rs9W9BxF+Z/mOM0zqmURPE9pLdHhBTQv6NqU5+kaEby/ovuuj01JDF7HXwKJiIiIiIiIiLQAvwQSEREREREREWkBfgkkIiIiIiIiItIC/BJIRERERERERKQFfKyJtKqqf42IX4+IVzudzqiPYhdFxH0RcXVE7IyIr3c6nWzerV/szDOTWJgESePGjaN24Gdu27YtxS6//PIUI1ElSRxJKkZCqpKImeIkXiPRLIkLe/bsmWIkNBswYAC2h0S8JKAi4ReJGF999dUUI9kjye9IFhbBIkYS/H31q19NMRKikoiNpG10jQgWYpJsksaru7s7xUjQRYLtkqiaxGJ79+5NMRovug712YQJE1KMhMsREffee2+KkciThN8kqiZ5IsnQSLwbwX1BY0NSXJoDJPckCXRJ5Ez5osmaK4nBTwU+zTpx1llnJWHwggUL0uu6urqoHfiZlN8o75BYmqSwW7ZsSTHKyyVJLb2Wxp9ilJ8ox9DhBTfccAO25+mnn04xkk0uXLgwxagmkDSRRMyHDx9OsdIhENSXJEm+8847U+xf/uVfUqypAHL8+PHYnh/84AcpRnsXqnskoKa8QdLNkkCS4uvWrUsxEn5SfvvXf/3XFLv55ptTjMY6IuKpp55KMZoD1L80p2j8hw8fnmIlCTzJT+kz6QAKOlSA6gnNx5KMnQSmlH+effbZY/5N+5tThU+zTvTo0SMJiGm/S2J/2p9G8B6D1hXlZdr703tPZG9M+ZGky7RnoZpA9ZGuUapb1E7aB1H/kDSX1gDlE+ozyt8RzQXwtNaWL1+eYiSlpueykoiX5hTVOHo/zWe6Ns2zkqia+pL25U2fHanPqJ7QuERErFmzJsUoj9JzPdVM2tNT/qZ5H8H3TWuOrk3PMvQ6En6XZPW0lqg99ftpKgtv8kugf4+IL9ZifxURT3c6ncER8fRH/xYRkXby72GdEBGRMv8e1gkRkc8EH/slUKfTeS4i6ueR3RYRd3/0/787IvJZmSIi0gqsEyIicjysEyIinx0+qROoT6fT2RcR8dH/5t9afURVVX9UVdXyqqqWl37yKCIipx2fqE6U/tRWREROOz5RnSj9OYeIiDTjly6G7nQ63+t0Ol2dTqeL/o5NRETazf+sEyXniYiItJf/WSdKTkcREWnGJ/0SaH9VVX0jIj7632zxExGRNmOdEBGR42GdEBH5FfCxp4MVeCgifjci/vaj/32wyZuOHDmSTkUZMmRIet2LL76YYqWffpJlm2zuZGMnW/nLL7+cYmRTL5nXqT1k5Kdr0+lQ5557bqNrlEz577//fqP30399f/PNfEAD2fzHjh2bYjQGixYtwjbSSQS33HJLijX9+S+Z7ulXaCUbO7WdTOv79u1LMTopoX7SUUTEwIEDG31eBJ928tJLL6XYlClTPvF76XSgq6++GtszcuTIFKP5R6fikeGf1hKN4ejRo7E99Fo6dZDGm04OofWxcuXKFCv1D81TOnWwfjpV6XSJU5hPVCfeeuutePjhh4+JXX/99el1dCpc6U+OKb/RyXl06heNM+V0OqlyxowZ2B46UWNGWrmPAAAgAElEQVTixIkpRuuSToKk2kF9QWs/gucsndIybdq0FKMTn5YuXZpidDIZ1VY6nSmC8yPVUTr1rempJnSKCOWSCD6ZavHixZ+4jbNnz04x+tPI0p9L0mfS6WCUl2muUOx73/teilEOjYiYPn16itHJVlQL6fQ8qsGbNm1KsWuvvRbbQ6f8EWPGjEkxWu/UHjqhk07OjOBaSCeJ1d9PJ8ud4nyiOvHBBx+kUx9pL0l9WjqJk/ZvtB+kvQTNBzppiF5X2ouW9vV1aO/QdM97In+hQf1Gz1ZNT0miHEN7Jbq/7du3YxvpuWXw4MH42jrUbnrmoeey0mlMNH8on1BupLGhk84oRv0dweuBTmNsmp/omfn5559PsdIzM53SRntwqlvUj/QLQZpnpRPwqL6WTnisQ/sMGn+qb3QSaATvkZqcBlfKKXU+9lVVVf0gIl6IiKFVVe2tququ+DBZ31xV1baIuPmjf4uISAuxToiIyPGwToiIfHb42F8CdTqdOwr/pxs/5baIiMgpiHVCRESOh3VCROSzwy9dDC0iIiIiIiIiIr96/BJIRERERERERKQFfFIx9CfirLPOSpI7EjaR5PLCCy/EzyQJFEmy1q5dm2IkpCIBGQmtSmI5knutX78+xUg+2N3djZ9Zh9pdEleR6LIu04tgqVlTARmNIcnUSGgaweIsEj6SOJOkmyTDpvv7yU9+gu0hwSbJYakfSZpM4jSaE6U5RRK6vn37phjNcRKvEyRYK4nFSHRJa5akiCSbJbFzz549U4xksxEseCOxHMXWrFmTYiRjHDp0aIrRuEawUO/+++9PsVmzZh3z7xMRNJ7OfOELX4ibb775mFhT0WRJPkjrhYR7JI+l2kMSaFoXJXklCUjnzZuXYiTdfeONN1KMJNmUL0kUHMHyeqoplCdIcjh16tQUI2EnxUrrgGSIlOtpHEheSjWqaX2L4BxFuYhyB81TykXUF3TdCF4P9FoSYtNcmTBhQort3Lkzxah+R/A+hdYczYERI0ak2IoVK1KM9hQl+TmNI9Ujqs1Uo2hPQFJbOlQggvtnyZIlKTZnzpxj/m2d+JAzzzwzyf1pLtKenvLBibyW9o70LELCfbpGaQ2RpJauTXmH8lvT+yvJ52kv2vQgFboO7RvpvdQequml95Pcn57BmtYY2o+U+oxyMNV12ndSPqE8T2LnEjQOVI/oM6ndlFepz2jvEMH75abXode9/fbbjdqzZ88ebE/Tmkuvo30Y5QWSQJdE3rS2n3rqqRSr1+umdcJfAomIiIiIiIiItAC/BBIRERERERERaQF+CSQiIiIiIiIi0gL8EkhEREREREREpAWcVDF0p9NJojQSLpHEqSTdIpnuvn37UoxkWCTsI+nmoEGDUowkXqVrk2yMRIN0L9dee22KPfjggylGYscIFpiSBIyE1iRdJBEXiRhJINfV1YVtXL16dYotWrQoxUiIR3LGhQsXphjJx0hoGsECYRLqkUSUZMEkmqZ7obkTwaJTEtOR5JLkgDT3aM2VIMEbfSbdN4kSSQxKskHqhwi+RxLdkqiaxpDmM82f4cOHY3tIBkzvr1+b5lgbef/995O0r1+/ful1JNzbtWsXfibJnWn9k4jxn//5n1OM8jKNH+X+CJbP9unTJ8VIfE/iy1tvvTXF7rnnnhT7yle+gu1ZtmxZitE9btmyJcXGjx+fYiQ+vPfee1OM1sXMmTOxjVSvX3zxxRQjQT7JmRcsWJBiNKfoniN4TpKslPIJ5TwSBdMcLUltqa6TJJPGi+YjtZFyKOX0CK7NlG+pblHtoLVEdbAkIKWxoTVH40B7FNoDnsihJtTndLBE/f0klW8r9VxIuZH2EiUJK40pfSYJzmlPR9ehuV2CPpPW/8GDB1OM9nSUb6lm0pqM4MMCKGdSPzYVEtOBK/Reei6LYAn01q1bU4zmBeUnOrCHxpD2oRHlfFSHaj31RVPhb6k99AxHkJCYnhOpxpQOuSHoOwC6Ds0VWockkKb5SPdXujbldbpvys379+9PMZo/tGeK4GdKem19/pQO9qnjL4FERERERERERFqAXwKJiIiIiIiIiLQAvwQSEREREREREWkBfgkkIiIiIiIiItICTrphri7JIrkSyTlJUhjB8mIS7JLsiUSeJPEjEdfGjRuxPZdeemmKkViOJFckn9q0aVOKjR49OsVIhhnBojOKkUSOBLevvvpqoxhJ29asWYNt3Lt3b4rddtttKUZSOhrrr371qym2YcOGFCvJ70jGSfI7mhd0LySqGzJkSIpNmDAB20OycupzEp2RqIxEY3RtEqdGRAwbNizFmsoBqS+ojXTPJcEeCaNJ2kz3TdJW6tsdO3akWEkOT3Of1lJTcVvbOPPMM1MeJgEgSV1LBwiQ7J3eXxdSR3D+JukizVnK3xGc/6n20ByhHEPyyjvuuCPFnnzySWwPifypz0l0StJcEh+SVJjkwcuXL8c2kshz+vTpja5D+Z9k2nRtkkpHcD6idT5q1KgUo3lGglaq9YMHD8b20Fyj8aI9CtWjv//7v0+xWbNmpRgdxBDBa5H2ayQ/nTdvXopNnDgxxZYuXZpiVJ8iuCZQ3aJ1TPujnTt3phhJSS+66CJsD9Xcp59+OsXqIlfF0GVISEtSYRqnCK4TtIaoJtB+gKS3lNNLzzfUdso79CxDOZgO8KB9I8lsI3jPS/dD64XukfqWYiThLx2kQjLtsWPHphjlJ3ovifTpGYHGOoKl3ZR3aO5RbqRnWRpDkoCX3k8x2hPQvKc+Iwl/6dAO+kzKcVT/qW8JmlOlQztInE33Q++nOd70cB56vo3g7wWof+o5oKmQ3KcQEREREREREZEW4JdAIiIiIiIiIiItwC+BRERERERERERagF8CiYiIiIiIiIi0gJNumKuL27q6utJrSHK1atUq/DwSuw4cODDFSHJFQiuSUpNgqSQfJGETScRI4kgSOBKDTZ48OcVI9hwR0atXrxQjiSjJC0l8RfdCAjJqIwlEI1iIunXr1hSjsSHBH0l8qR9LojK6H5Ls0ViTvPLiiy9u1EYSgEWwPJ3EsiSWo/EiITbJVGkdRfB4kXiN2kjSTZr39Hnjxo3D9jzwwAMptn79+hQbMWJEipGYlsSrJK8jUV0Ez3NaD3W5fElW2TbOPPPMJK8l+STl4GeeeQY/k0SXtM5pLtLY0VjR+itJfElgSjWBRIMk56T8TUJCmu8RXHOfe+65FOvfv3+KHT58OMVIckiCzTlz5qQYyXEjeA7QWqM8QdJO6m/Kl6W6NWnSpBQj+TDlPOoLGhsSWpfyBM2LphLhzZs3p9gVV1yRYjQfSfYcwXuPZcuWNXod1XoaG2o37SkjIu65554Uo5pAY0i5gmS11G6qZRERc+fOTbFf+7VfS7HFixcf82/ay7SRqqqSDJXmLM0vkutH8F6L9n4kYiYxNAmA6b2l/Ts9e9B8onxL+ZLqDklvS3mD8igJlikX0X3TXKb+pnEtiaGpzyl30PMN1TLaG5NImQ4UieD5R+NN7aGxoX0G5WW6bqmdNF60F6JnZnoeoDlRmuN0bZqnTaXtNEdJVE4HykTwYQPUZ5T/6RmOagLN0RIvvvhiitHet37oDt0z4S+BRERERERERERagF8CiYiIiIiIiIi0AL8EEhERERERERFpAX4JJCIiIiIiIiLSAk6qGPqMM86I3r17HxMjsTMJEkePHo2fSUIsEieSGJoEkvR5JFg6ETnjlVdemWIkPib5LMm0tmzZkmIktIuIGDVqVIpNmTIlxUhASWJfEimSZG/FihUpRqKxCL5HkmGRTIuklCTJo/eS7LF0bRKVTZ06NcVILEeiMpIKk7w2gu+R2rhz584Uo7m3bt26FKN5snLlSmxPXVQZwdK/CRMmNHodCWxJfEby24hIOaV07bqIOSKSgDiCheE07wcNGoTtIREsiQnrwjiaJ23k3XffjTVr1hwTo7GnOUJrMoLHhNYbiY+bCulpjA8ePIjtobo3dOjQFHvttddSjHIHzWMSH5fklWPGjEmxb37zm40+k0TztNboXujAh5Kwn/YFVAtJxn3dddel2PLly1OMxp9yY0TEs88+i/E6Q4YMSTHK6SSVfPLJJ1OMBKIRLE+lPn/hhRcatbFPnz4pNnLkyBT78Y9/jO0hjhw5kmJ0AMHLL7/c6PNuu+22FCsJSKmG07XXrl2bYlSjaF9IdYLyRwSvxXrei8hzsqns+3SnR48eSX5LolhaVyTCj+C8TsJnki5TjaLPo3lIe//SdUicTHWGPpM+j2pj6dAL2sNT7aH1QntwkgqTdJkk0KV1QM9g1J6mgm6aUyTsLuVlkinTXKF5SmNIgmR6/iuJgemgGqq51Ge0buh5i3Ie7XlK16YxpBiNA8nG6VmvtN+m61Btpfuh+UPri/auNC4RPH/oHuvPQU3rhL8EEhERERERERFpAX4JJCIiIiIiIiLSAvwSSERERERERESkBfglkIiIiIiIiIhIC/iVG+ZI7ETS23feeQff31TkRZIrknORSJHaWBLL0fupPdOmTWv0uj179qQYyX5LYmgSWm3evDnFSEpFUjvqR5KX1YV9ESw+i2CJNEnWSEpKElAS1T3zzDMpRhLQCBZi3nzzzSlGY03zgmSPJDml8Y9gMRj1OfUZzR+Sts2fPz/FSOQXwZJ2ei2tG5pn69evTzG6PxK5RrCscPv27SlGgscDBw6kGPU3jVdJXjpjxowUo3VTz12l/m4b5513XsqP1Ncks6e5FMGSPBJDkhSwaT0hSOAXwbmV5uytt96aYrSmaR7fcMMNKdbd3Y3tGT9+fIrNnTs3xSjfUv+QdJEOZ6A5P2vWLGwjrX8SPpJsmmSadEDCE088kWI7duzA9tx+++0pNmnSpBSjOkHzguoEiY8pr0awmJzmFO0JSIZMB0MsWrQoxUp5i95PNZcE7VSjli1blmJUT+h1ESy/pvVAdYKE1iS1pb0DycsjWKhOa6n+OuvEh1RVlXI47RtORHxMfUtjTzHKefQ6EkOXDk0huS9dh9YVXYfmF+3VS1Jh2t9SLmvajzQONIZUv0sC+N27d6cY1Xqqo5RPaF9NeYP6MYJF3iQGprxDdYvyJeUiGv8Ifpama1OdoYOTaBzokBqSkkfw8yztFWhe0H6N9kL03tJeiO6b+pzuhwTttJboWbj0HQcdTEEHENXnadM64S+BRERERERERERagF8CiYiIiIiIiIi0AL8EEhERERERERFpAX4JJCIiIiIiIiLSAvwSSERERERERESkBZzU08GqqkqWdzKq0+lVr7/+On7mhAkTUoxObiGrOZ2SMWjQoBQj2zyd7hLB5n46hWzXrl0pRicE0CkZ7733XoqVTPBkRSeTPPUZnYpC5vRrrrkmxVavXp1iAwcOxDbSKSJk5N+3b1+K0ekn1LdNT7WJ4BNn6CQBsuLTCS2lkx/qkAU+gk/L6927d6Nr06kItJbodDky9Jc+k04XINs9nb5ARn3q25kzZ2J7nn322RTbu3dvin3xi19MMTpJiE5KonspnfxE5n46Xaz0/rbTo0ePdOoD5Wpa+6VTXyZOnJhiNOdfeOGFFOvq6kqxoUOHphidsFQaYzodgu7xqaeeSjE6EYNitM5LJ0ZQv9GpHVRHqV7TqSZ0OgydQNa3b19sI+UOqnuUq2msH3nkkRSjnEenu0Tw6SB0giX1OZ0iRXWCah6dchXBOZPmBZ2WRKcL0akmdHLitm3bsD1Tp05NMdp7UG6lGkNjSLm2VEcp19O1v/GNb6QYnZ5JezOae1SLIrgebdy4McXqpyDS/q+NVFWV9tu01ui0INpzRDQ/JZfmHa01yicnMn6UYyi3Nj2hkXIE5X5af6X20MlddI9UO6gmUG2k00FLtb7pyZRUHynf0pyiU9JKz2CU86h/6FmY5in1N90zzeWIiJdeeqnRtSk30j1SG5vOxwiea3Qdqke0Fuh19IxJe7iIiHXr1qUY1T06JZmePWkfRfO+lBco19DY0mc2wV8CiYiIiIiIiIi0AL8EEhERERERERFpAX4JJCIiIiIiIiLSAvwSSERERERERESkBZxUMXSPHj2SZItkRiTILElqSQ5KkisSQ5EwjgRkY8aMSTESKZfeT8InEjuROIv6gkSMpfaQwIykXSNGjEixq666KsW2bNmSYiTOJBEuCZcjWIZI7yfpJkn7SLBJryvJvV977bUUI1kwScDovSSqJIkgCVYjIpYuXdro/cOGDUux7u7uFCNJdlNhYATPNZKDU7vpMydPnpxitN5JiBfBfXHrrbemGAnem8pmKU+VxHvU56+++mqK1eXnJLRrI0ePHk3yWpoPJNwvsWbNmhQjAeW0adNSjAT5FKN1RbUoIsteIyL279+fYnSPVMtIFkzzfeXKldgeymUkex85cmSKUa4mWfucOXNSjESc48aNwzauX78+xSi3Ul2ntTV27NgUI5F3KQ+SsP9LX/pSitE9/uAHP0ixr33taylGUsiSiHnFihUpRrWZYrSPIpk6SaVvv/12bA+1ncbh+eefTzGShc6ePTvFaLxK8nM6TIHavnz58hTr169filGdIFEp1cYIlqLTnqR+GAetyzZSVVUaa1qrlC/pwJQIFsDSXKS9Os13Gk/K6fTeUpzaTgepUBtp7dOaJilwqT20N6LnDtpj0n6Z2kjvLdV/+kxa+5S/mwq/T0TCSzmP2k77ahoHEtLTM0/pmZBqPR34Q89rJNNuWmNKNO1z2nNRLiQRM+0fS/JzujYdDkKic9rv0TqkOUHXjeBnBxrb+vNR0+cJfwkkIiIiIiIiItIC/BJIRERERERERKQF+CWQiIiIiIiIiEgL+NgvgaqquqKqqnlVVW2qqmpDVVV/+lH8oqqqflxV1baP/vfCX35zRUTks4Z1QkREjod1QkTks0MTMfTRiPjzTqezsqqqXhGxoqqqH0fE70XE051O52+rqvqriPiriPjL431Qjx49kjCKRMMk7CxJWEleTKIpkoWRBI6kWyR6JflUBMu0SpK1Oq+88kqKkYiLpGsDBgzAzyRxIglISSJKMjaSLlJ7SF5XkqmReLupvOzgwYMpRvJSkk+T7Cviw3na5DokJSMBHYn3SCBJEvAIHq+m0s0ZM2akGElpSV5JwuaIiH/8x39MMZKa9e3bN8VI0EbrkATrNJcjIiZNmpRiJMqj+UftJmEcydlorCMi9u7dm2Ik7a5/ZklAe4rwqdWJnj17pjxM+ZvmbCkPUg6mdU4SX8rpJPsjcX1pDZHknGrZnj17UozWPq0Nkk/SAQCl+NNPP51iJEOkXN2/f/8Ue/vtt1OM+oHyUwSPA8k9Kd9SjGo9yX5nzpyJ7aGaSSJYyvU333xzitGBDVRbqd2l6+zcuTPFqC/+4A/+IMWeeOKJFKN6QjL1iIi/+Zu/STESLFOdoHxJ9Z9koaUDH5qKwGlOUp9Tu0leXjrUhNY27V3q+8KS1PgU4VN9nqjXb9pDU/+XJKwkkaacQPOG9l+0X6b3ktg5gusM7Ttpz0IH0jSV8JbEvvSZtH+j5xbqC2p309eVuPTSSxu1h6A9GPUF7QdLAnjaJ1I/0vyhnE77WMqNNJdL7aHaQ59JeZlqMD3L0AESEbzPaHo/lKtpDGn8S8/ldMAHXafpwQAEXbtUJ6jtffr0SbH63ozaR3zsL4E6nc6+Tqez8qP//8GI2BQRl0fEbRFx90cvuzsi+IgIERE5rbFOiIjI8bBOiIh8djghJ1BVVVdHxPiIWBIRfTqdzr6IDxN7ROSvX0VEpFVYJ0RE5HhYJ0REfrU0/hKoqqrzIuKHEfF/dTqd/Pu/8vv+qKqq5VVVLW/6Z1EiInLq8WnUidKf/YmIyKnPp1En6E9YRESkOY2+BKqqqmd8mLC/3+l0fvRReH9VVX0/+r/3jYj8R5IR0el0vtfpdLo6nU5X6e9eRUTk1ObTqhOlv2UXEZFTm0+rTpQcGiIi0oyPFUNXHxqa/iUiNnU6nb//H/+nhyLidyPibz/63wc/7rM6nU6SWpEosKn4KoIlaSQBJvnZW2+9lWIkACaJF0khIyLGjRvX6DokFSSZJgnIxo8fn2Lr1q3D9tCvr0h0RuLVCy/MBzRQ/9B7aVxJShcRsXTp0hQjyRr9lx96YCSZJm0YSgJJErSSgJKkryRJI4kXSWRJsBrB4mOShZG0lcaLfmlB62PRokXYHpKf0X2TCJbmBY0hyf0oFsECtDVr1qQYCfFIzknzZ8KECSl24MABbM/QoUNTjPqnLn08lYWfn2adOHLkSBJ8kiySxrgkhiZIQEw5j2TRdFABCTJXrlyJ16Y58uijj6bYnDlzUozmJ+X/W265JcVIFh0RsWLFihSjww/Wrl2bYlSPmh4MQP1dWlfURpoXJIaktU+yXxLLkpg/gg9O2Lp1a4pR/qZcTf+R7KWXXkoxEjtH8Jyk+6YcTPfdq1cvvE6dhx9+GOM0DtQ/tGeiPRyJwQcNGpRidLBIBB82QPdN9ZoOByEJNI013XME5wDqs/q+yTrxf6j3Ba0hyjt0KECJUj6qQ2JfgsaY9mkRvOehfTDlUZJK03upJpREynSPJM2lvTU9d1B7KEbjWupvytd0UAiNAz0n0LXp80rQ/pbmFImTKddTXm56eEwEP9/QXp2ee2kt0TjQcyc9Q5WgvEzjRXsh+k6AxrVU3+h5gp4TaFzpWZ/mD9Wo0i8b6bsC6p/6fVM/EE1OB5seEd+OiHVVVa3+KPa/48Nk/V9VVd0VEbsj4rcaXVFERE43rBMiInI8rBMiIp8RPvZLoE6nszAi+Gc4ETd+us0REZFTDeuEiIgcD+uEiMhnhxM6HUxERERERERERE5N/BJIRERERERERKQFNHECfWr07NkzSY4WLFiAr6tD4toIFq6SSOmMM85IMZKILlmyJMUGDhyYYsOGDcP2kMiZBJskKaTPJDHksmXL8NoESa7o2iSVJLkbyRB/8zd/M8VI4lUSGpIkiwRb1D8kpSMRF72O5kQE9w/JlOl1JNSbMmVKipFEliSeESx9HTFiRIo1FUNTf+/cuTPFRo0ahe2hfqM+J1khrTmS5JEQce/evdgektWSbJTaTfOM2v3888+nGElAIyKuuOKKFKO5Uh9vmqNtpKqqVANobpM0sSTNpXEeMmRIipHQkOoJyXDvuOOOFKO5FMFrmsSQ1G4SeZIokNY0yR4jWCJJfU5yzy1btqQY5X/KJ/R5VP8juH9Ickk5hsSXtKeg9lDfRjQ/bID6kYThf/Inf5JilL9LeyESdJL8evPmzSnWNKfTnCrVLZJ20jiQoJPGkPYeJIYuydiXL1+eYjNmzEgxOoSEDnKg/E+C1lJev/7661Ps3/7t31Js8ODBx/y7tD7aRo8ePdLcof1y6b0E5dump5CRLJjmA62L0hoiES+9n/IE1Qnax9LBLnTdCJbNNs2t9F7qb3reoveWxNC03qi+UZ/T2qJxpXxZWpfU5/QsRM8EdI/0rEZSYXpdBM9Jmis0rnQdOiiGah69LoLXIs0f6t+mguWmEucIfoanekRjSPtHOrxi27ZtKUaHIUU0F4bX+7fp84S/BBIRERERERERaQF+CSQiIiIiIiIi0gL8EkhEREREREREpAX4JZCIiIiIiIiISAs4qSbSI0eOJHHb5Zdfnl5HUimSPUWwyJEEwiSPJYkjSXxJkPXUU09he6ZNm9bo/SQVfPzxx1PshhtuSDESO5YgqVmfPn1SjERwJJWeOnVqipEkjcRX559/PraR5kBT6SvJr2j+kBisJKom2eiuXbtSjCRgX/nKV1KMhJb03pKAlMS0ixYtSjHqHxK+kTiT1gxJ1yNY+tqvX78Uo/nTVEA9b968FCMJXATnAJLxkaCNpGurVq1KMZIxLl26FNtDfUFtLInJ206PHj2SGJCE65RPSnWC5J79+/dv9H6Sbn7rW99Ksd27d6cYCaAjWNo4ZsyYFCNp7ty5c1Ps29/+doo99NBDKVbqH5qzJKUkmSaJdEnsTsJ0WufUlgiWADeV+JMUuC7cjeA6SAc7RLDQsmktJCkwyedJUkn9HcHizcWLF6cYycEpP5H8/LbbbksxWh8RLCCl+6bX0TyjfQLJVKnmRUR0dXWlGI33jh07UowOYqD73rhxY4pRjYngnEYy7vp1SJLbRo4ePZoOlaA8TzmP9v4RvJ+kOkNjR9DejfLqnj178P0052lfRdehvRbVPLqX0gECtGemPqc5Su8lcT3lVTo8hJ5tIlgqTH1G7aHXkTib9pIl2XhTkTO1hw44ofbQeFHtiOB+I/E9iZOp9tA8oxpO7Y7g5wna9xCU/+n5dvv27SlGuTaC62hp31SH6ij1D127lNfpM2mOf9LnCX8JJCIiIiIiIiLSAvwSSERERERERESkBfglkIiIiIiIiIhIC/BLIBERERERERGRFnBSxdAkciNJ2t69e1OMRFwRLEgiaS4JpEj4Rp9HcjaSoUWwlJIkWSR3JnEmCeOaSqoi+L5JfnfhhRemWM+ePVOMJNf0XhLQlWRzNF4kBiM5L8mwSO5G84ekkBEsaCORN4mYly1blmIkICXx5RNPPIHtISkeycJIsEZzhcaLZJrUj6U4ibPrEviIiJEjR6YYyWFpzZXEeyTtJGi8SHRHOYByBa2PCBZqU//W76ckfG0b77//fhL+Uk0gAXxpLtB6o7VKeYfkwzQ/KeeVZH1UE0ggSfmfxPWrV69OMZIzk7w6gnMh1WZqI4lFqT0kn6dYqb6RYJ9yHq3fgQMHphjJYWmdltYl1dZZs2alGPUjxUiSSvPkySefxPZ0d3enWNPcSLls+vTpKVbKwQTtZ2iuLFmyJMVof0VybxKnUm2MiJgwYUKKUZ/RmqW1Te2hPEPzMYLnM9WZ+mfS/q2NfPDBB6kGHD58OL2O6kSpDylO+xhavyTcpXVFh5SUJMe016I5QntEEtySzJzybUlSS/1L0KvLMUsAACAASURBVLVpHVAOpucEGpdSW0iITPmE+pzyG41BabwIyick8iao/lOOoflYqvV0P1TjqC9orlDfNpWFR/B405qlZ0/ah5HYn/Z1pb0Z7XEI6gvqR5ortD5K/UMHBlGNq6+Rps8T/hJIRERERERERKQF+CWQiIiIiIiIiEgL8EsgEREREREREZEW4JdAIiIiIiIiIiItwC+BRERERERERERawEk9ZqCqqmTKphM2BgwYkGJkC4/g0zPIOE+xffv2pRidpvHMM8+kWMkOv3///hT70pe+1Kg9dPoVnX5CJ2wMHToU20MmebLq04kF11xzTYr9+Mc/btQeuj96XQSfikI2d7Kn03XI+v7WW2+l2LRp07A99H468YnGi04xopOzaE5NnDgR20P3SKdO0HXI0k8npdGcKJ3SQ31Jtnoy99PpOzSutL7odIDSZzY99euSSy5JMVpLNP6l/qGcRNTnSum0urZx1llnpdN4aM716dMnxejUhQg+gaLpaS50yhFde/v27SlWOk3pgQceSLG77rorxWjejR8/PsXodJhVq1al2OjRo7E9tFY///nPpxitwVGjRqUYnVZCpyFR3inVCTr1heob5fW1a9emWNOTaSjPl9i6dWuj97/22mspRjnr+9//forNnj0br3311VenGJ2ISHOFYlSXqXaUxovyOp28+du//dsp9s4776QY7aMoB1OuiOCTW+k6dN/U7i9/+csptmLFihQrnbJJ64ty0qZNm475N83btlJfM1RD6TTUUh/SmND40b6cPpPyE53aVDqpiO6H9lq01mi/Q+0+kZxXOhG1Du07qWbW53YEr1PqB1q7Edw/9HzT9HmCTiaj9pROlaLcQTmT9i703Et7Aqr1tLctQXOFnmVp/KkmUN/SniCC+5LGi/qXnhNpj0J9W9q/Ux2me6Q1S+uG6jKdOEhrM4LHka5dP0Ws9Hl1/CWQiIiIiIiIiEgL8EsgEREREREREZEW4JdAIiIiIiIiIiItwC+BRERERERERERawEkVQ7///vuxZ8+eY2IkeyL5VEkKS1IpEnmSiHH69OkpRsJeEo2RPDAiYuzYsSm2aNGiFOvq6koxuseDBw+mGInu6HURLMQiURnJFEmGReNFIi2iJOIaMmRIilG7ly1blmIkICO531VXXZViJPGMiBgzZkyKkZSURGc0V0gCN3jwYLw2MXDgwBRbt25dipFkfdCgQY2uTTK+kvCTpG0kWZ8xY0aK0TwjgRlJybds2YLtqeeUCJYQkvCZYiSRJUFfSbxHMmCa++PGjTvm3z/60Y/w89rGoUOHkjDy/PPPT6+jnE75oATlcJIuzpkzJ8Uef/zxRp9XEmxeeeWVKUb3M2XKlBQjqeDLL7+cYiNHjkwx6seIiB07dqQY5X+qUSQppBzc3d2dYiTCrQsOf85v/dZvNWrPwoULU4xqFO0z6F5Kcu++ffumGK1zEotS3xKUQ6lvI1gYft9996UYrRF6L90fjU2pPSSHpTpDc5I+c+PGjSlG64gk1xG8tknuTDLfuqg+ImLBggWNrk01OILnH+17brzxxmP+PXfuXPy8tnHkyJEkr6V5Q2uaZMgRvDZIPkvrnJ47KFdTzivVCbo27XdI7N9UfE/tKfUP5UJ6nqC9Mc132ks2FXGXDtKgZxQSH9PY0PhT/qbDSE5EVE15kO6R+ozGhu65dHgR5dumomLKt7TXpjEszSnqS+ofOoCC5gA9Y9D6ome6CH6WprxCa476h+oOrRnqhwieA7S+6odalQ5JqeMvgUREREREREREWoBfAomIiIiIiIiItAC/BBIRERERERERaQF+CSQiIiIiIiIi0gJOqhj6yJEjSWBJ4jOSMJVEzCRoI8Hm4cOHU4xkWCTxIpFWSUBK1yGBLF3njTfeSDESSJIYes2aNdge6h8SPpJEbMOGDZ+4jfR5JMOOYIEdjSEJ/vr165diu3fvTjESw82cORPbQ8Iwkv7RHKDrULtpDEkWFsESMBJ5kryYJJckKiOh2S233ILtoTn1yCOPpBjJZkleTQJRms8DBgzA9pB4k0TVJLCjeUrzkca11D8kHCe59xNPPHHMvyl3tJF33303Vq1adUyM+m/y5MkpNn/+fPxMWm8kPiSZHglghw0blmIkZ7700kuxPbSGdu7cmWKUW6k+0jqnOUtC+QgWflLdony7a9euFKN1Re05kbxM+Y36jITYJJCmPEj9QALICB4HWsMkr6SaQDJMmrc01hERq1evTrEJEyY0ujbtRxYvXpxi3/jGN1KM9gQRPH8efPDBFKO5QmuJavBzzz2XYl/96lexPbS2ae5SXqExpPsjcXapbtXl9xE83nX5MeWtNnL06NF0eEVT2XtJmkrrn4S/tPej11F+o2tTGyN43pFMmT6T2kPXof3X3r17sT1Ut0i6S3vZ+jyOYFkwSfwpRvkggu+xdHBOHXoepbxM4uNSnaD9JI0htZtyI81xmmcktI7g/EGHxdA9Up2g5xPKjbT/juA5tXnz5hSjuUIx6gs65KbUHhpHaiN9J0Hzh9YCvY7qTgQ/y9C1ly9ffsy/m9YJfwkkIiIiIiIiItIC/BJIRERERERERKQF+CWQiIiIiIiIiEgL8EsgEREREREREZEWcFLF0D179kyiTJJukZSMZFgREUOGDEkxkkqRROytt95KsTFjxqQYCZZKIjeSV5Kccc+ePSlGsjmST5LwqyQVJikZyStJ+EXvJWnno48+mmLjxo1LMRrrCJa+0WtJfkdy7zlz5qQY3XNdpPVzSCxGomoaVxJDklic5lRJ+Llt27YUO3DgQIpNmzYtxWgtkWSXxqAk0yOhJa0vukeSg9NY0zxbv349tofkudSehx56KMVuuummFKO+GDFiRIqVJLuUGygv1NtN120jZ599dlxzzTXHxEjWRzJyEqZH8Hqh/u7fv3+KUa6m+UD1hESjEXw/JFMkMSRdp6mknnJ6BOcYyqNUZ6geUc6bN29eil1//fUpRiLEiIiLLrqoUay7uzvFSH5K+ZvE13RAQgSPDclYSQ5OsmASypdqQtP20BwfO3ZsitE9Uq6msaFDBSIiNm7cmGJ03yS0pvxPcs8/+7M/S7FSnZg1a1aKkXj33nvvTTGqEwTtR2nvEMF1i+YK5R/5MLfW94QkcG16mEkErzfKt5R3aL9Dh7BQTSg9T9AejHIUHbhBUE0gsS+tiwjObyS5LUmJ69TrfETEli1bUozqZanPqMZRPzbd8w4dOjTF6LmM9v4R/OxK43X11VenWFPxMY1haY5THiUoV9P401qgsSkdpkSHNlAepHVD84eE1rQfIVF5BM9dEnnXDy+JiBg1alSKUV/QfC7l+aay+k/6POEvgUREREREREREWoBfAomIiIiIiIiItAC/BBIRERERERERaQF+CSQiIiIiIiIi0gJOqhg6IkuSSJBF8kmSu0VE7N69O8VIAnnBBRekGMm9SFJLYriSDLGpgJTEaySAIskZxUoiNhJakSSL5GcXX3xxipFAlER19HkleR31GUmySEBHMsyFCxemWJ8+fVKM2h3B901SahpDEn7RtUnYvGLFCmwPSdLoOhdeeGGKvfjiiylGskIam2XLlmF7aByojSQRJYEd3QuJlEloHcFiOpqnNF4kESUhHonF6RoREZMmTUoxEg7WxauUj9pIVVVpvdE8JvEhyfoiOE9QDqdxprEjkS5JSWluR0SsXbs2xWje0f1QnaC6RfmgVCfo2pdddlmKUR6kQwVIPjl16tQUI/lwSUhP13nsscfwtXVIpvnyyy+nGNWiQYMG4WeS8JP2IzRedKABycZp70D1LYLrFs1xEmJSvb3xxhsbvY7uOYLnColOBw8enGKU/2nfQvOeDveI4D0kCTZJ8Ep9TvOR8gyt9YiIm2++OcXofq677rpj/v0P//AP+Hlto6qq9DxBOZgoCfKp/2md0zNG0wNbqI30eRE8PyknkHSXhLS0/yIxL8l1IziPUp0hyTHtESkH0/6UBMDUNxGcO6im0LjS+qX8RgdI0P1FNJ8rBD23Ut6heUZ9G8HPhE2vQ3OX9t+0vkribJqn1MbSmq1D85HWR+n5j8aLnmWoz2mu0PMf3QvtCSK4PtI6Hj58+DH/pvpL+EsgEREREREREZEW4JdAIiIiIiIiIiItwC+BRERERERERERawMd+CVRV1TlVVS2tqmpNVVUbqqr6m4/i11RVtaSqqm1VVd1XVVX+Y0oRETntsU6IiMjxsE6IiHx2aGKnOhwRszudzk+rquoZEQurqno8Iv4sIv7fTqdzb1VV/19E3BUR/3S8Dzpy5EiSEl5yySXpdSQAJFFUBAubjh49mmIkhqLrUIyEVk899RS2h9pJAqm6xCmCBZIkJCTJWUm6TGJIkpqR8I2EXzReJD8lCePOnTuxjSTeJhkWyeqob0kYR3LPkjhry5YtKUbzh9pIsjGSpL766qspVpLIDhs2LMVIIkax6dOnpxj1N60ZkvtF8BwnoR4J0ejatIapv0vz54orrkgxmqff/e53U+yBBx5IMVoLdG16XUTEunXrUmzkyJEpVr9HmsunEJ9anTh69Gi89tprx344CJvrwtQInksRnFtJnEgy5AkTJqQYSXxXrlyZYt///vexPZQn6vccEXHttdemGOWTXbt2pdjq1atTjGSYESyBbAr1La19En5SHly0aBFeh8TpkydPTjGqPVQze/funWJUG0uiapJpkliW6ijNqe3btzf6vJKAlNq+Y8eOFBs7dmyKfe1rX0sxkjjTPCERawTXV5K5NxXiU12mdbR582Z8P/Ub3c+dd96ZYps2bUoxGn+S7A4ZMgTbQ/sUWjenGZ9anfjggw/S2qT5RfuV0t6PhKskriWxL12H2kN7oJJcnfI1HXBRygl1qE7Q+qV+iGDpbtN1Rc9WtAZIcE/3XDosiPqXcjBBezDaGzc9NCGCRfy0v6X8TftY6jPaH5VySdPPJHkxPUfR2ND6KLWHXksxWgtUo+hZmPq79MxM85Tm8+23355itAekeUH7npIsvLu7O8XombD+DEYCeeJjfwnU+ZCf70R6fvT/OhExOyL++6P43RGRe0RERE57rBMiInI8rBMiIp8dGjmBqqo6o6qq1RHxakT8OCJ2RMRbnU7n51+R7o0I/hmDiIic9lgnRETkeFgnREQ+GzT6EqjT6XzQ6XTGRUT/iJgcEflvmT78Nj9RVdUfVVW1vKqq5fSzMREROfX5tOpE6afoIiJyauPzhIjIZ4MTOh2s0+m8FRHPRsTUiLigqqqf/xFb/4jIgoUP3/O9TqfT1el0uko+AhEROT34ResE+aVEROT0wecJEZFfLR8rhq6q6pKIONLpdN6qqupzEXFTRPw/ETEvIr4WEfdGxO9GxIMf91nnnHNOkumuWLEivW706NEpRoLECJYukbyMxFckC6PrkASMpKIRLJuiYkWvIwEUyQfffffdFCPxbARLrUlWR8I4EuJRu0k0Sf1Yerhbu3ZtilFfkAj2/vvvb/Q6mhN0zxEsRCN53t69e1OMZHXnnntuivXq1SvFrr/+emwPSYlpTpHAlOYpjQ1JoEtiMeq3gQMHphj1I80f+i96JF2mMYxgOSj1GY3hV77ylRSjcaUcQOuj9P4BAwakWF1Wdypvaj/NOnH22WfHNddcc0xs2bJl6XUkPqZcFME5huY8CftI2Lh+/foUI0E5yZ4jeB2QgJKkuSQeJ9kjrQGS2UewmJRE/CQaJvk8iXSnTp2aYiTSpc+L4Fq2atWqFCPp8vz581OMRPw01tQPEREHDhxIMcoxtM+gOkFzgmSRM2bMwPaQgJQkqzTPqN6SJJXyGAm2I/igAmo7CZ+b3gvVvNIBC7RvohjtKb/zne+kGAnMabxIShrBc4D2vvV7LH3eqcCnWSfOPPPMtLboYADaQ5PMPoLzLQmfaQzoM5vubUqHQtBzC72W8ijFqI0UK4mm6dq0v6Vr09qgPSLVMnpdaT9IeZTaQ/WfxL50CATNiZJomNpOImfKB03nD8mMSwe7UF+QYJl+kU3vpXpCubq0X6YcTs9HdB3qR6rX9CxTOtiFrk11ncZ1ypQpKUZ7Sqo7pbxOonOqcfX1VRJNp/c1eE3fiLi7qqoz4sNfDv1Xp9N5pKqqjRFxb1VV/3dErIqIf2l0RREROd2wToiIyPGwToiIfEb42C+BOp3O2ogYD/Hu+PDveUVEpMVYJ0RE5HhYJ0REPjuckBNIREREREREREROTfwSSERERERERESkBTQzB31KHD58OLZv335MjER4JHbq27cvfmZXV1eKkbST5IwkldqzZ0+KDRo0KMVKYjkSS5Lc6/HHH0+xPn36pNjEiRNTjARiJHuOiBg/Pv3yFiW3JJp68cUXU4ykwiQB3bx5c4qR4CoiYvbs2SlG0i0SSNJ7qb9JkDxmzBhsz9atW1OMZKM0z2g+Np17GzduxPaQpI/WA72fRG4kvvzCF76QYiUZe1PROa0Fem89J0SwJK0k/CRBMK1ZEtORtO35559v1J5hw4Zhe0gkuG3bthSrj4NH3n7IoUOHUv6gPqXxLImPSVRIomGSO1NuJdE0tZHWeQTLdCn2gx/8IMVIuPv1r3+90eeVZJF1EXcEt/2mm25KMRKhk4yV5N50LyS0joj45je/mWIkryS55+TJ+S9NSHBP/UP1MiLihz/8YYpde+21KbZ06dIUGz48n4p91VVXpRj1LdXGCB5v2l8tWLAgxUiwTTmU8vfrr7+O7SF5Kgk6ad2QbJYk63QgAdW8iOZCZdpnkCyUZOMk/CzJ6mlvtmbNmhSrjysJUtvIBx98kPaUtEegA0lKMnNagzTvSCBNzy20V6f9V0kMTeuAxp/2rHQdOkSDpMA03yNYQEyvpZpJeYLE1/RsRc9lpecJ2gPQ2qf6P3To0Ebtof1uSabd9LAJmmeUQ6lvaU7QwQURzeXn1G6SOFN76BqlvEXrk+o6fSbt62he0N6a5nIEjwM9w5F4mZ6ZaJ9BdaK0z6Br09ytr/emdcJfAomIiIiIiIiItAC/BBIRERERERERaQF+CSQiIiIiIiIi0gL8EkhEREREREREpAX4JZCIiIiIiIiISAs4qaeDnXXWWXHllVceEyOzPNm56bSRiIhFixalGJmz6SQfMpDTqQEPPvhgipVOKqKTKcgkT+8nYzid+EQnBpD1P4IN72SNp9Oh6mMVEdHd3Z1iZM8nkzuZ5SP45C46SYD6hwztNFfodIENGzZge6677roU2717d4pRn1O7SycE1SmdWEGnX7355pspRuNF7aETXmhODBgwANtD1nk6YYJOl6PTMs4555wUo9OKaAwjImbNmpVitLbJ3E+nBtx+++0p9uSTT6ZY6YTASy65JMVoHdb7sXRCSNs455xzUr6mtUZ5uXRCB52mR3ORTgejE0zoxIZHHnkEr01Qjdq3b1+K0QmGlFupjXTSTWnOXnjhhSlGa2jHjh0pRqdsUK7ev39/ilE/0nsjuM/oJBA6pYfq8rhx41Ls7rvvTjHq74iIOXPmpNjcuXNTjPYelA9oPtKJIaU6QfmW9g90QhedBEb3TXsZyncRXIepPXQ6DOVlGlfaP5bqOp3IQ/dNpyoR3/nOd1KMTvOj+4vgE3noxMP62m56ytnpTs+ePdP6oDlLp+nRniOC97e0b6U6Q/mSoFNlS6f70t6RYk1PK6P9IM3P0vMWrUtqD81RWlc0NtSPdC+l01Qpj1JNoT0r1Uc6gYr6kU66jOATDOkeqc+pf0qnftWh58QIPnGS9sb0TEjtppxF7S7NKYLmFM0B2vdQe2gMqZZF8HjRZ1Jf0LynfUbT5/oI7jfaz3zSUyP9JZCIiIiIiIiISAvwSyARERERERERkRbgl0AiIiIiIiIiIi3AL4FERERERERERFrASRVDn3322Uk2S4I2kmGSVDKCJVkkC168eHGKkbyKpMAkviQhVQTfDwl7ScRJQjOSB5IUeOfOnY3bQ0Isuja1kfp7+/btKUYSOBJ7RbAkk8R7JAwncSZJ20iSWhLxkqCV5MN033SPNHfpnkmSGsESsAkTJqTY888/n2LTp09PMbo/EjGXhNZr1qxJMZJu0hiS/IzEhLt27UqxkjB206ZNKUYiYcoLtL5oXCdPnpxipf6hOUn3uGXLlmP+rfDzQ84444wkJRw5cmR6Ha2XkizyRz/6UYqR+PzRRx9t9Jk0H0hcXBIxk8iT5IVNpek0Z0koSIL7iIjhw4enGElzSaZJ4lSqMdTGrq6uFCvV1nvuuSfFZs+enWJNBZQkmqd+KEHtvOuuu1Ksvs4jeO6uXbs2xUj2SHksImLhwoUpNnPmzBSjejR27NgUW7JkSYrR/qgkUqY914gRI/C1dag2036ERL4lUTXJPWk/Q9cePHhwij3++OMpRvOxJAymmkAS0VWrVn1s+9pIz549k8ifDr2gGl86pIT2WrSn27p1a6PX0d6f5mFpb0z3Q88jVCfo+aaphJ2E+6X20LMD7WXo2vReynm0hkrrij6TBPL0OtqzUv9QP5RqPY0tHRZAz3X0XspZVG9LsnGa+3QgBtVHkibTvpz2MqU5TkJtWrNNhf1N83xpv930AAui6fqgvVUJ6kvqi7rwu2md8JdAIiIiIiIiIiItwC+BRERERERERERagF8CiYiIiIiIiIi0AL8EEhERERERERFpASdVDH306NEkHCN52Y4dO1KsJGYi4S+JgUkq3KdPnxQjORcJtui6ESwHI9HU/PnzU2z06NEptnr16hSrS1Mjyv1D/UsxajfJxkgYd9NNN6UYyXpLYt+qqlLsxRdfbNRGkl9dfvnlKUZSOhKsRnBfkgSYhF0vv/xyipEQjfpxw4YN2B4SXZJsbOLEiSlGAlqSFY4aNSrFSAAdwdJluh8ar5/+9KcpRqJT6sfSeJEk+7HHHkux3r17pxjlCnodSUlLojsS4BJ1OVxJ5Nc2PvjggyTd/bu/+7v0OpIrlubIlClTUowEgpSr6XUkBSa5YqlO0GdSjlmwYEGK0VqlfDlkyJAUo1wbwbmZalRTYSMdVPB7v/d7KbZt27YUK4l9SbpI16FcT31BdZTGn/JdBAufSZxJ1yE5J8nPly1blmKlMSTpPkk3qX/oM2nN0BynPUoE51GSrNIY0pyimkCvo31dBM8rWl80XosWLUoxWodN62AEi1fpfoYOHXrMv0kC3EY6nU7ay9K6or0/7YsieO9Hexa6Dq0rEuk3lT1H8Nyh2kNziSTHlN8ob5TmLD07EE0PuSC5M0lzKUdQDS1dmw5toPdTjPIY9VlpXdL76To0ByifkDib5gQdABPRXNpMMnyq/9Ru2vfQYS+lz6Q1R7WD+oLmKD3Llp6Z6fmRns2o3TQOTfuM7iWC1yytz3ptpnxE+EsgEREREREREZEW4JdAIiIiIiIiIiItwC+BRERERERERERagF8CiYiIiIiIiIi0gJMqhu7Ro0eSFZEgiyRVJcnV8OHDU4zkTOPGjUuxF154IcUuu+yyFCNpIsmaIiL69++fYiRouu6661KMxIdTp05NMZKSkVQ0goVWI0aMSDGSAJMgccaMGSlG/d3d3Z1iJWEjSbvWr1+fYtOmTUsxEnGSiI0Ea6U+I3Efvb+p9I+kq03l1RE8p+g6JDskSR6JmJcsWZJiJEMuxZuKwEm8fsstt6QYSeAGDBiA7Vm6dGmKUf9QjK5Dc4qkonv37sX2kOCZ2l5f7yXha9s466yzkuSOxo7W37p16/Azu7q6UoxyFMn5n3zyyRSj+UCSSzrkIILrHknuSc5Lsl+S89M6J0FmBNdXqplUJ0h8ePXVV6cYyatJzF4SpJN4k8aQ6huNP8kVaVyo3RERkyZNSjG6HxoHquE0z0h0TkLqiIjf+Z3fSTESqtKceuqppxpde+XKlSlGtSyCZbVUJ0jQSa+j+k85mMSyERHz5s1LMZpTJOimgxgoB9CcorwQwcJxEm/X5x/libZS72+qEzQmtJ+LYGkuvZ+eE3bv3p1ivXr1SjGq81TLIngPTjmK8i3tv+hQEOqz0qEXr776aorR+qd6QmuN2kOCXNo3Ug6N4L6kmkkHS9C40vNE03kWweNF4mNa1/Re6p8LLrggxUryc3oOo7Gh/qU20nymZ1nqswh+7qF8S/dDY035n+YzidMjuGbSfdOzFcWoTtC6pufgCJ5XlKfqsvqmzxP+EkhEREREREREpAX4JZCIiIiIiIiISAvwSyARERERERERkRbgl0AiIiIiIiIiIi3gpIqhf/aznyVZWd++fdPrSChYktQuX748xUjkRcImkoA1lUWXZIgkbaR7JPExiaoOHDiQYiSaJFlYBIvc3n777RQ7dOhQitF9kwSa5GV0zyWhIYlAmwqfSUD33HPPpdgf//EfpxhJTiNYDEYCYRqvpiI/kmHS6yJYfr569eoUI9H1+PHjU4zWDElgSXIawaJLkrHRuE6YMCHFmq4Zmrel15J4j3IASbdJ0EbvPf/887E9q1atatSe+pqlOdZGDh06lIT2NMavvfZailGOiOD1Qp9JosFRo0al2LPPPptitKZpXUXwWm+a62ku0pwlkSJJnCNYmktrgwSSVJtJKkntoQMWaK1E8Hqjz2wqkN66dWuKff3rX0+xktyb9ik03kOGDEkxymUkKqV+HDNmDLaH5j7d97Zt21Ls29/+doo99NBDKUb7ntKcor0C1Ufau8yePTvFSO5K62Pjxo3YnqFDh6YYidLrgs2IiMGDB6cY7R9oDEuCzsWLF6cY9W9JdN12Op1OqpmUD0gCXZIK036ShLY07ygP0pylNtK6iOAaR3mQ9mqf+9znUoxk0U0PzIjg+kifSa+jPEFrn2oj9VnpeYI+k8aGZMrUbsqXN954Y4rRWEc0PyyA5gBJkylG/VM6YIHqOs172udTfqKDE0hUTrLnCB4vkinTvKD20HVo7tG8jeDcQLWMoHlG/21JAQAAIABJREFUUmqq/zQnInheUdvra7b0nUAdfwkkIiIiIiIiItIC/BJIRERERERERKQF+CWQiIiIiIiIiEgL8EsgEREREREREZEWcFLF0BFZfkRyPRJskrg4ggXCS5cuTbFrr702xUj2RxI+EnuR0LL02jlz5uBr6+zduzfFurq6UozEhyQ5i4gYMGBAijUVRpGAlKRbJUl2HRL0RbCgk2SsJD4mmdof/uEfphhJ5EoyPnotSTdJXkaQtG3EiBEptn37dnw/zX0SlU2ePDnFlixZkmJvvvlmipGAcMGCBdiekSNHphhJbal/SDhI65BEvrQWIiKeeeaZFKO8QONNkrzSPK1DMt4IlsP+/+2deawe1Zmnf8d2MDFgg23sNot3vIH3fbfBBtNpCJ1AwqRbiUZR54+0Jj09kWgykaKZ0YxEzx/pbrVG3UonIUhhS0iACGggGJvFG97Xa/vi3Rh8ARMCNCTB1Pzhjxlc73NMuTHX3731eyTL3DffV3Xq1DnnPVW5fl6aN+W1sEsXv5OXTggAy4I9WiNo3JSF0h9CY3Ht2rUh9tWvfjXEtmzZEmJf/OIXQ4yExE888QS2h+S+N9xwQ4iRZJzGJ62XJFwn2b/E6xHlYSo0UFWkToLEI0eOhBgJ9yXpnnvuCTG6DySBJEkqSaBJkrpw4UJsD0kgaa0mkSedhz43ffr0EMuJj0ksSfuCIUOGhNhPf/rTECNZ6CWXXBJi9957L7aH9lyUU0giSrme5jvtt3LC/ocffjjExo8fH2K0tyOJPMlhSV5KY1ySvvKVr4QY5euygJyErXUkpRRyJom+aV3OiY9pzaNCIQMHDgwxEvFecMEFlc6dk+GTAJbmAUGCc5q/NN5zUmGaWzQeaa2nGLWR+oz25PRdiccA7fMImr8zZ84MMZK95/qM1nXKUVWLxdA9oPUp1z/UlxTr379/iB08eDDE6FmExvPOnTuxPbQ3oz0FjQvaZ1A/0hpMeTDXzr59+4YYPU/QcyLtCajdlPMkHrs5qfW/Bz91GGOMMcYYY4wxxtQAvwQyxhhjjDHGGGOMqQF+CWSMMcYYY4wxxhhTA/wSyBhjjDHGGGOMMaYGtKsYukuXLurRo8dJMZLMrly5MsQWLFiAxyQ5E0m3XnrppRA7cOBAiJGIkUSVJE3M8eyzz4YYyepIKNva2hpiJJkl+bDEfUHXnZMXVoFkodTGnLya2k4SORKIUT+SYJk+RzJkiUVuJEkrj2WJxyOJmEkOSLJQiUV3JDCnPqP2kMiPYpMmTcL2kISO+oJE51u3bg2xqjJNkmZKPG9IVkcC83379oUYieGoH6kfJL5GGn/l+UD3z5yAxuIzzzwTYosWLcLvU//TMWlt3LNnT4iRFJbGEuU3ieW+NM9JXkhzf8WKFSFG0lyaaxKvb6tXrw6xd999t9J3aQ0leTzdl1xupTxM92bs2LEhRn1BckZal3O5kdYJEn6ToJOum+TcVMSB2i1xHiZRJa3VJNIn+SkJO+l4Eud1KlRBayutodQ/JG2nsSdJc+bMwXgZkueSiJOk7yQgpzEhSY899lil7+f6t+6klMK9JvH93r17Q4zGocRjh+YGzQPa85LUlfZ5uSITJHsnOS+t67QuUxtpTafz5uJUqICuh9YTitG1VF1Dc9+nohs010ju/Oabb4YY7dVysl46N7Wd5nlV2TiJnXOiatob07MVXSONXboWylskU5f4GZXGGe0LSNpO56YxmpOx076ejkl9Qe2hfR1dX+4ZntpJ3y/n0aoFoPybQMYYY4wxxhhjjDE1wC+BjDHGGGOMMcYYY2qAXwIZY4wxxhhjjDHG1IDKL4FSSl1TShtTSo80fh6SUlqTUmpNKd2fUuJ/iG2MMaYWOE8YY4w5Fc4Txhhz9jmd3wT6K0ktH/n5byX9XVEUV0h6Q9LXz2TDjDHGdDicJ4wxxpwK5wljjDnLVKoOllK6TNLnJP0vSf8lnVC6Xy3pK42P3CXpv0n6p1Md54MPPghmcjJnT506NcSomorERnUynVPVD7JxU7WRXr16hdigQYOwPfTZlpaWEBs1alSIUYUmqmpCFRCowobEVdHo+1RFpG/fviFGlVKoigxVaKEqBBJXZCDz+qFDh0KMrmX8+PF4njJk/Ze4igiN04suuijEyKhPVbLovl5xxRXYHqqqQxVHqCLD7373uxCjqkh0jrVr12J7aJxSJYpVq1aFGFVLouPR3MzZ/KmaIH2fqupQJTkaj1SlL1c1iKqR0HguV3nKzY+OwpnKE8ePHw8V9ejezZw5M8So+onE84Cqg1B1B6rGQXmCqlzkKv7NmDEjxOgaL7/88hCjNZ3OQxVwcmvM+vXrQ+yqq64KMRrzlHuoAsmvfvWrELvuuutCLFdF8stf/nKl9hC0vlHVFqoiQp+TuBLdpk2bQozuIa0xlEepHym3SpyjKE8cPXo0xKj6DlXAowo4VCVN4jFFfUnXTe1+8sknQ4wqPuaqLNKcrTq3aW2manV0D7Zt24btoTFJ469cnapq1Zdm5UzliaIowt6B5gvt1XOVYWl80hihfEIVeygf0dzPrXlVq9rSWKRnnvfeey/E6PpylaVoD0brP10jPavR8aj6GfUP7RslvocUo3P37t07xOh5ko5Ha6jEVbGomhftR+hzdB661/RsJPEaTs8ydEzao9AaTH27efNmbA89h9H+nfbgVH2P1kca9zSHJR5/NB9oHlJupXtIz2q5yq10Hhp/5XUuV60uHKvSp6S/l3SbpA9HQB9JvymK4sPePiwpjgRjjDF1wXnCGGPMqXCeMMaYJuBjXwKllP5EUltRFB/9v3USfDS+7jrx/W+klNallNZ19P+n2xhjTORM5gn6f22MMcZ0bPw8YYwxzUOVfw42W9KNKaU/lnSupJ468Sb/wpRSt8bb+8sk4e9lF0XxA0k/kKT+/fvjwm6MMaZDc8byRN++fZ0njDGm83HG8kS/fv2cJ4wx5hPwsb8JVBTFd4qiuKwoisGSbpX0dFEUfyZpmaSbGx/7mqSHP7VWGmOMaVqcJ4wxxpwK5wljjGkeKomhM/yNpPtSSv9T0kZJP/q4L3Tv3l1Dhgw5KUYiRZKKkXhWYmEfCfdI9kRCKzoeCT9J4iVJ+/btC7ExY8aE2PPPPx9iJDQuy2Nz5GSBJPIiOei6desqnYdkWnR9JKUiabbE93vkyJEhRoJFkm6SSIskZyTdk1jsS9IuOg9J9mickSyMxHkSS8novtI4JYkcyQpJDEfSZIml1rl7W4YkztRuas/s2bPxmD/5yU9CjO4DSX+pz6lv6R4cOHAA20MiQZo3ZflxTmjawTntPNG7d2/9+Z//+Umxe++9N3yO+rmtrQ2PSQLB5557LsRISEjrxJYtW0JswYIFIZYTbJJMd/r06SH22GOPhdi8efMqtZHW1ZzE94Sb9WSoMADNIZLh0vp28803hxiJnWnuSrwuT5w4McRWrFgRYiRxpvWN9hnUNxLvFWhfQIJukmHS/KcCErn20P0iSCJOElHKmZQ7KB9I0vLly0OM9h6UW5ctWxZitBeigiE33ngjtofmEvUl3UPaz9BaQxJiynkSS2jp3OWCGDkhbgfntPNEt27dwnwlqTetRbTOSyyQpX0ejXmSF1Pepz1Hbi9K6wTNVSoWQMVeaMzRPKc+k3hvRGJomi8ks6XzkOSa1tpc0QW6D9QXdN0k9ybJMbU7ty7TPaR5TmOP1kYaP7TPyI0pOibdG+rfqv1DUIEkifcPtHehdY/GPX2X9hRU+EKSnnnmmRCrsi5L3Gc0LqqKyiV+PqLn/bJsuurzxGm9BCqKYrmk5Y3/3itp2ul83xhjTOfGecIYY8ypcJ4wxpizS9XqYMYYY4wxxhhjjDGmA+OXQMYYY4wxxhhjjDE1wC+BjDHGGGOMMcYYY2rAJxFDnzZFUQRZEUm8nnrqqRC75ppr8Ji7d+8OMZKIESSzLctaJRbAklRMYlkpia/mzp0bYiQGq9o/JJCUWIhKwkYSWm3evDnE5s+fH2Ik4t6wYUOI/eY3v8E2Up+T1GzKlCkhRuI9EoORNDMnw6Z7OGDAgBAjOSCJV0lKThJYEl/nzk3C57IYTOJ+pPv/zjvvhFhODkjyVBK8kuCP5gJJ7ajPcvOa5lJOilfm5ZdfDjGa2zTGlyxZgsekuURjvCwMzIkO68a//du/aePGjSfFqG8eeuihEMtJYVtbW0OM1ltaO0j2SnPt6NGjIZaT9ZKokOb/4sWLQ4zG7IgRI0KMxNc0JyW+bpJf01zdvn17iJG8moSmJHLNCfIJkkWT8JHmHxUfIIknrVmSNGzYsBCj4g50HyhPkOxx8ODBIUYyTInHFB2T1jIazyS+pHMcPnwY20NjkvYAVNyB9lw07kk0ff/992N7JkyYEGIkOqX5TnsFmu8kxCUpucT3hsZ++X45T/x/yqJiEj7TXMvtl2nfQXstkusStK8iKWxub0xSWBLIXnLJJSFG+xhqN+3dSKScaw+tEyTDpf1yVQk0xXKCdOpzWrdoDNC9IWie5gT5tI8m0TDlHjomrTuUJ6iIg8R9QWsrrWV0X+kZnNpNRQWk6vtlmtvUblrTad+SEydTTskVqyhD87iKxFniIhcSjzV67i2vC9QPhH8TyBhjjDHGGGOMMaYG+CWQMcYYY4wxxhhjTA3wSyBjjDHGGGOMMcaYGuCXQMYYY4wxxhhjjDE1oF3F0O+8845WrVp1UozkjCRmamtrw2OSqIxkyiSGIkEuyStJupYTz5IYqmfPniH26quvhlhZciexIJHExSS/k7gvKEbtmTx5coitXr06xIYMGRJi1Gc56RpJskj6RsI3kmaRqJIEYnRfJBaBkhDt0ksvDTESvpH4kmRhJNKUWGBWnkcSi8Coz+haqH9yAkoSwZEkdfjw4ZWOSYJeOh6J6iSWBlYVXefGQBkSsW3duhU/S3OWrqcsAsyJDuvG73//+yDovPLKK/FzZUhmL/HaQ3mGjkmiWZK10zwnmaHE6xbFSCxNMk0qFkDjneauxGJJWpdJ7jh79uwQe/HFF0OM5iTd15wAniTS1D+UR0n4SoJ7kpzSOpY7Jl3jwIEDQ4xEkzT2KC/T8SReR5944okQo3FG+yjac9HalhN5k4ydxhmtwVQEYteuXZU+l1tHaVzRGKe9FI2ziy66KMRofuSKiOzbty/EqKBGubhH1cInnZ333nsvjAlal2lPT+JZqbrEl/ZatD7RXpu+S+eQeN9Bn6W5T+sgyYfpWSYnOaa2U1EB2qvRekt7bcqZtD7lIHE23VeK0f6U+oL6MSfipXWU9uD0XEd5gvqC2p0bU7Qe0b6AnjvovlKM5NU5ETN9luYs5T0q7kTXR89quTxB44LyBN1DyoV03TSvcwUWqH/o+bp8PfSMSfg3gYwxxhhjjDHGGGNqgF8CGWOMMcYYY4wxxtQAvwQyxhhjjDHGGGOMqQF+CWSMMcYYY4wxxhhTA9pVDN2jRw9NmjTppBjJOUk+uG7dOjwmCXJJ4kfC5gkTJoTY/v37Q6x3794hlhNDk4yNRJ4kziR5FUkTScSVkzPSZ0mSSte4adOmEKPrJkEmSRPpHkjSa6+9FmI0LkgYRyJAEn5RjNotSW+99VaIUdtJ0Lhly5YQmz59Op6nzI4dOyq3hwSkJN6j+0oS6C9/+cshRuIzSbr77rtDbNCgQSFG94b6cezYsZViS5cuxfZQ/5CkLydULUNrCsnmcnNu5syZIZaT4n2Url27Vmhd5+fcc88NBQNIPk9yT5LHStKKFStCrKoofMmSJSHW0tISYiTxzUlhaXySJJMKGpAocPz48ZWOlxNV9+nTJ8RIdE1rFK2jtO5QLqJ+IFm0JD366KMhRrmerqUs15VYnEix559/HttD6xtJO0kgT/sMKpJBYlC6rxLLKykn0Bin/Dhu3LgQo3lI/S1Jjz/+eIjRWCHxNuU3EtDSHo7kyhLPT8ozNL9GjRoVYpTraQ3PiZxp3zNr1qwQK0tEcwUb6kb37t3DPpGEqfQ8kSvqQFJwWjNJ7EoyW5IU0xpBuUjivToJf2lfRe2m9YTmWm4vUlU2S3mLBMu0x6T20PFIhi3xvaHrodyTy49laA7SfJZ4/tPehdZvkgXTPpbaQ/JqifuHPkvjlNZ6GrsLFy6sdF6J92a096habIg+RwUJ1q9fj+2hOUvjh6TvBD3rU77NPU9QHqbrLre76vOEfxPIGGOMMcYYY4wxpgb4JZAxxhhjjDHGGGNMDfBLIGOMMcYYY4wxxpga4JdAxhhjjDHGGGOMMTWgXcXQUhSBkciTxIckFZOk0aNHhxgJw0g0RXKm+fPnh9iRI0fw3MSwYcNCbNu2bSFGwlASEpIsjNpNUmlJOnbsWIiR3I0klyTyIjEY3Rs6B4nYJBbp0rkpRsJdkm6SnJsEYhLL6qqKvOi+rlmzJsQuvvjiECOBoST16tUrxEjuSXI4ug8kaCX5KZ1DYgFiW1tbiJEElMR7JJEl0SGNZYklezSeaR7T2CU55y9/+csQozVFYtkorUnl8VxVStjZ6dKlS5ASvvTSS+FzNEb27duHxySxK60dtE7QekLiQ5I45iSutPaQaJakiSROJhku5YScqPrZZ58NMRrfU6ZMCbEnnngixK666qoQo/lH0syVK1diG2+66aYQq1r4gKTChw4dCjG65pwAkvI13YcvfOELIUZFF+i7tI7R9Uncl5SjaNxTfqPPkVCX5ubptKe1tTXEaL2kvRWNZxLL5tqzfPnyEJs4cWKI0fyg+XXnnXeG2F//9V9jexYtWhRiJN4uF3K477778Hh1pDxGaf2n8ZnbS9DehuYGrQkkyKUxQvMqVziCRM5U2IMK0lB7aA7QHpHk01Je7luG9jJ0HlqX6Rz03VzhEupz6p+q56FnMOqf3H6Z1mvKW7Tm0b6aCqFQf+dE1SR8poI/NCap3bTe0nN9TlRM/UZyZipKQc+UNKaqFtfJnZvaSPtP6p8RI0aEGBW6yu3/aS9FlMd41ecJ/yaQMcYYY4wxxhhjTA3wSyBjjDHGGGOMMcaYGuCXQMYYY4wxxhhjjDE1wC+BjDHGGGOMMcYYY2pAu4qh33rrLT399NMnxUgKS9JEkgJK1YVoJOciCRhJCklmSOeQWAJMstH+/fuHGElEST5JwqecOJvEWXTMvn37hhgJ0ajP6B5WFchJLPYlGd9jjz0WYiRXJLEXCTbnzJmD7aHz0DEfeuihSp8jCewLL7wQYtOmTcP20FgjAdmLL74YYiQpJqkdSdtykl0S05Kkj0RudF9JfkvXTFJaieWgdD0k2aN1geb7t7/97RDLjXH6PgknywL8nDivbrz77rtBkkxSYVprhw4dWvk8JD4kQSKt1S0tLZWORxJGiec/ff+KK66o9DlaDyiX0fEknr90jd27d8fvl6G1iPIbrQc5KSl9lopIkPCR1izqn9mzZ1duD63/gwcPDrF77rknxKi/33zzzRCjXE3XIkmTJ08OMVqDSRhKAloSX9JeJifYpLFCkksSYtO6TCJvGhM0FyTp5ptvDrGNGzeGGMlzSTZLsX/+538OMRL5Spyb6XrK/Ut73jry9ttvh4IWtMbs2bMnxGjPKuWLl1SB9iw0h2g9yRUQoLFD368qgaU9C+07SKQr8f6/6jFpXtHeraqkOPcMRsekffnOnTtDjNYsmm90D0gALEmbN28OMVqjqAAFtZvGLu2PBg0ahO2hsUb9S8ekZxkSMROvvvoqxumYNH6of2i+Ul6n440cORLbQ/ebxiTNERp7tH+ktT+3BlDOpXxdbk/ueGX8m0DGGGOMMcYYY4wxNcAvgYwxxhhjjDHGGGNqgF8CGWOMMcYYY4wxxtQAvwQyxhhjjDHGGGOMqQF+CWSMMcYYY4wxxhhTA9q1OliXLl2C1ZoM7ePGjQux9957r/J5qlbEGj58eIht27YtxCZOnBhiPXv2xHOTAZ8qmND36Rp79eoVYlTJiyqqSWw1J5s7VYIiAz5VK6B2UzW3vXv3YhupygYdk6qiUGUCss1TpbPvf//72J5vfvObIUbVfI4dOxZiVA2CKgHQ/SdDv8RVB9auXRtiV199dYhRdaJf/OIXIfalL30pxAYOHIjtob7s06dPiFFlA5pz1Gdk1M9BVQNmzpwZYlT5Z9euXSFG10LjLLcGUHsIGqfmRHWG8jpKlSrmzp0bYlQ1QeK5SveeKsnMmjUrxNavXx9iVNHodKrNUAWL3BgrQ9WmqCJdbg2mqhZ0TLoPl112WYjR3Kf1bciQISG2e/dubCPlLZqXkyZNCjFaY+i+UsU5uteSdNttt4XYAw88EGLUP1u3bg0xquZxOpVJqVoJ5XCaIzROqYoYVSq57rrrsD1UmYyginXURhq7F110UYhRf0tcNeaaa64JMaoOQ5Xbqo7HXJUe2ivSPqy8B3QVyTwHDx4MMaoMm9trVa0iTBXxLr744hCjiki0pucq+dA+mOY5jQm6RlrT6dxUsVfi/qH5Qutt1Yph1N/UZ7kqgFQprWrVT8qDdG6a548//ji259prrw0x2ntQP1IerVqtjK5Z4uetw4cPhxhVcqR8RNXPpk6dGmK5KpJVK4TSueke0ri48MILQyxX6ZSeb0aPHl3pmPSsT+OxamVaid9d0H62nJtprhP+TSBjjDHGGGOMMcaYGuCXQMYYY4wxxhhjjDE1wC+BjDHGGGOMMcYYY2qAXwIZY4wxxhhjjDHG1IB2F0OXZXokbVu6dGmIkZxZYnHygQMHQoykSyTdIokTiZlIpJU7D8l5SZxG7Sbx1XPPPRdiJKmSpLa2thAj8TZJIOm6Sc5Mkio6HolKJZbnkfyahHEkaCOhGQm//+Iv/gLbQ3JGkojR9dD9J0k29W1OFEyyOhoXJBsjkfcNN9wQYnTNU6ZMwfaQmJRkyiSBJnEjzTmSCObk8AsXLgwxks317t07xOh+0ZgieSIJuyWW2tH9LgtDc6K6OlIey7T+/+xnPwsxEg1L0oQJE0LspZdeCjES5G7fvj3ESHJIInUSQEosGiQhIUmSSZBLc3rlypUhlhNV0xozZsyYECMx9KpVq0Js/vz5IUa5iNanUaNGYRs3btwYYiSgJKlwVcEy5Zg77rgD27Nu3boQGzp0aIjRfaW5TqJqWvNI7CnxXorkkNQXtJ+hfqRxkstbixcvDjHaCxEk16+6PtKeUOL1mtZlErk++uijIUa5Y/DgwSFGexmJ5zaN/csvv/ykn6v2YR0oi4rp3pMQPFf0gvYitL+lOUR7DlpvSeycEzHTvoPkzHRMkpnT9dEafDp5gvZVtIeitYzE7iQ+JoF0TqRLYmC6blobqY20z6dcNmfOHGzP66+/HmI0fuheU7tp7NLYI1G5VP1Zhs5N93XGjBkhRvsjyo0S7/VpD033lfqWjkdjJSdjp8ISNOdo3FctYEVFBXIFDWis0PuD8rigcxD+TSBjjDHGGGOMMcaYGuCXQMYYY4wxxhhjjDE1wC+BjDHGGGOMMcYYY2qAXwIZY4wxxhhjjDHG1IBEcsRPi/PPP78oS3tJZkgSp5w4rWfPniFGAslbb701xEg+SLIokjCRpEpimSKJE59//vkQI+EjCZ9JXvnKK69ge0gYRp8lUSldC8m0CZIh5r5LMj86N32OJInr168PMZIckxAv91kS4pH0laRkJLWlMZ5rz/jx40OM7ivJr0nOTOcmoRkJTSWWttGcI8Efyfj69esXYjRW6Foknl80Z0mUR+OexLsvvvhiiJXFzh9CQkYSLZZlfNdff702b94cO6hmnH/++cXYsWNPitF6Qmsj9b3EeWLPnj0htmjRohAjMTRBOSonhaU5SOs6CXspR9HcoL7IySJpLaMcR9JEEkjSHKK1MSfxJWg9orWxqvCZChKQhJEklxKLQGkdpVw/cuTIEKN1lSSguVxPIm9ag0mUW1U+OXv27BBbvXo1tofWx1/+8pchRsUmSEBO95CKXOQEmzTnaA9IcnmSmtJ8p7WC5ofE95bEq2Wh9ezZs7Vhw4ba54nPfvazRVnETesJrTu0f5I4T9N8oz0dzRcSylIBD9pfSix3pfFE84XmdNW8k9urUztJ2kzC55yItwzdL7oWarfE10h7TMqF1LdUQIIE8CTnlvL71jKU6+laSHBP/ZMbUySBppxJ0ny6rzRn6Ny59tAcoXWZ5g21m9YA+m6uiAgV3aE8Ss+eJIumfQv1LZ03931aF8rFeaZPn65169Z9bJ7wbwIZY4wxxhhjjDHG1AC/BDLGGGOMMcYYY4ypAX4JZIwxxhhjjDHGGFMD4j/mA1JK+yW9Jem4pPeLopiSUuot6X5JgyXtl/SloihY3GOMMaZT4zxhjDHmVDhPGGNMc1BJDN1YtKcURfHaR2L/W9KxoijuSCndLumioij+5lTH6dWrV1EWC5JwieRTJMiSWIZL0jYS0pLw7bXXXgux6dOnh1hOmktCKxJLzps3L8RWrlwZYiTnJKEZyb4kFuCReG3p0qUhdvnll4cYidNI4kyispy0lY5Jsl+SsbW1tYVYVWFsTu5dltJKLN4jId7UqVNDjMRpNCZyYjASxpHojoRo1G6KUZ+VRWMfQqI8Gmd0PSTPIxkmCZtJzilxX5Kgje43XQuNZ5Iakshd4uuh85Tvw7e+9S21trZ2WOHnmcoTffv2LW688caTYgcPHgyfI6FsTnz8ox/9KMTK55CkZ555JsSqykZnzJgRYiQKllg0TmOWxNKPP/54iJFUmNZvkhlKLCqkNYakwrSHoLWDroXWjVwbaZ5T3qNcT/JJuq/UDzkBfPfu3UOM1i3aoyxevDjEWlpaKrUxJ7R89NFHQ4zE2ZSP6L5SnqD9GsnCJWnfvn0hRv17xRVXhBiNZ8qDtE8gYbdUXWpL/UPjlGSjNIdpHkrcF1u3bg2xssD8u9/9rvbu3Vv7PHHeeecVNN/K0LNDTlJMzwQkGqfv096GxheogZQjAAAgAElEQVStqzSvJBY00/6CxhgVGqE5QP1D809iGTOt/wcOHAgxeiYgwS21hz6XE0NT/qA1nNYOytfUF3QPcu0ZNmxYiL3++ushRmsMjW/aq9NalCtKQc9CdI1ViwXQGKX5kRtT9CxN94HakxO8lyEJf27/Tm2nMUVzm/YotE+g/SPd1xzHjh0LsXKeuOWWW7Rt27ZPVQz9eUl3Nf77Lkk3fYJjGWOM6Xw4TxhjjDkVzhPGGNPOVH0JVEh6MqW0PqX0jUasf1EUL0tS4+/4etsYY0xdcJ4wxhhzKpwnjDGmCajkBJI0uyiKIymlfpJ+nVLaWfUEjUX+G1L+n3QZY4zp8JyRPEH/PNQYY0yn4IzkCfonQsYYY6pT6TeBiqI40vi7TdKDkqZJOppSGiBJjb+jnOXEd35QFMWUoiimeNE2xpjOyZnKE/4/C4wxpnNypvIEuZuMMcZU52PF0Cml8yR1KYrircZ//1rS/5B0jaTXPyJy610UxW2nOlbv3r2La6+99qQYyZBIpNizZ8/cMU/Z/g8hkRKJa0kMTYLEkSNH4nlIFkVyR5IckviWhFTr168PMZKk5tpDL+PKUqlcjEScJG2lfsw93JEYcsKECSFGYl+ShVGf0f3PidxILL5mzZoQo+shGRsJ6Ohe58RpJKEjqSTNZfocic7pfuXaQxI66vMdO3aEGF0LieBJnEcyNYnvw5YtW0KMhIy0kSQpHY1RWhckHrskmy7Lc5csWaLNmzd3SOHnmcwTffr0Ka677rqTYjRf6B7n5jRJ5WnMPvfccyFG84XmwPbt20MsJy4kUSHJiym/kZyR5gCde8yYMdgeWutJuk9zcOPGjSG2YMGCEHvyySdDjPI6CVYlFhDTXCMhMbWb+mLv3r0hlvs/r0j4uWHDhhAj6SsJzBctWhRiNJ5p7Em8tlLeoz0BjT1qD+09coJNkhxPnjw5xGi9pUIV5aIiEgvWSVQqsaz86aefDrGc6LoMjSnaH9GYkqQRI0aEGIlgy/vHxYsXa9OmTbXPExdccEFRnv+0H6Q8TbJWiddRus+0ttJ+ieYa7WNpbyKxyJmukdZMeu6gtYz2iPTd04HaQ9ddVbBNwt3cS0ASeVOxF7qvVeXVtGbl1mV6VqR1i34DesCAASFGz6207uT27zSmSM5MzxN0D0lKTp/L7d8pP9JnqbgK5Uf6Lo2f3PMoFZYgyTrt6atCz4SUvyV+9qB9Ybk9M2fO1Pr16z82T1R5ld5f0oONjukm6Z6iKB5PKa2V9LOU0tclHZR0S4VjGWOM6Xw4TxhjjDkVzhPGGNMkfOxLoKIo9koKdUaLonhdJ97eG2OMqTHOE8YYY06F84QxxjQPn6REvDHGGGOMMcYYY4zpIPglkDHGGGOMMcYYY0wNaHe9flk2deDAgfAZEoORrEli8RqJZkkCSoJNipEsLCclIxkXyb1ILEeSKxLYkUCM+kGS3njjjUrnIfkZnYfuDUku6Rw5URlJ1kg2PXXq1BDbtWtXiFHfkthr0KBB2J6dO2PFUpKx0X195ZVXQozuIYlYc2OcpJ3URmoPSdtISkeSNBLDSiwNJEkvCdporJCck8ZPTmpLfU59QYK1TZs2hRhdH90bOp7E60UVwevHSfrrwvHjx4PIb8WKFeFzNKfHjw//0kCS9Oijj4bY3LlzQ4xyB61FJC4nAemSJUuwPfT9OXPmhBjNDZJcE1R8gPpM4nxGskjqC5IPkySbiheQxDeXJw4dOhRilKNorSZR5bp160KM9iM5SSrlQlrzypJzSXrkkUdCjK6P+uJ09h5r164NMZIzU0EDkkBTLqIcLLH8ksTZb775ZohR3rrvvvtCjHIrzSNJamlpCTESbNL42bx5c4hVleeSJF3ioig0BsrS4Nw+wXCBC1pjcrmW5KyUz2ke0BpDe1v6bk4qTOLkqvsLGid0PIrlqCpEpmukdYu+S3tRui+5dpP4mMbF8OHDQ4yel+g89JxIeweJcwo9j5LQmvqHcgy1OyfIp7FP0mUq2kDtpv03jT26L1J1WTmNcRorVJCGzk39LfFYo9xKOZP2gNRn1D8k986dm/ri3/v84N8EMsYYY4wxxhhjjKkBfglkjDHGGGOMMcYYUwP8EsgYY4wxxhhjjDGmBvglkDHGGGOMMcYYY0wNaFcxdFEUQdq7f//+8DkS+z311FN4TJJkklSKJMD79u0LMZKFkSiKxEwSS7JIVFxVXr169eoQGz16dIiRaFhieeEHH3wQYiRI3L17d4iRqGzjxo0hVlVyJrGUbOjQoSF21113hdhVV10VYiToo2uheyWxCG727Nkh1qdPnxCbPn16iC1btizESD6ZG1Pbtm2r9H3qXxrjdK9JSEniU0m69NJLQ4yEaHQfSEw7bNiwEKMxSvI7iQVtJCZsbW0NsXHjxoUYSeTKomIpL4am+U7zsDxnc/e/bhw/fjyIF8tyVEmaOHFiiJEAWuK1lXIPrf8klKVxQ5JiEmlK0t69e0OMxizlN7oWkr3TGkGSSon7l4SEs2bNqnRuEvbSGkzyYIpJ3Jckhnz44YdDjMYKiRTpXufk3suXLw+x66+/PsRImn/rrbeGGOVwylE0RiXpoYceCjHKR5S3KMeQZJ3Wb5KkSiz8pFxPaz2JXCnv0FzIjXHak9C4oKILixcvDjHKebT25/ZmJAelvV15DOQEq3WjKIowdki4X7U4hsR5msY3Sb1JpF9VHp6TutIeivZ5NA9oDaYxS/snOq/E6y21nSTQNDdo7tOad8EFF2B7CBLN05yhoguUe+j66BkzV9CAxgCtO3SN9NxKewe617RnlXh9pO/T+kR7cBKVV70HEu+bqj5HU46huUCfy7WHhNi0N6O1JldsqAzN4ZxYnD5b5Xmiap7wbwIZY4wxxhhjjDHG1AC/BDLGGGOMMcYYY4ypAX4JZIwxxhhjjDHGGFMD/BLIGGOMMcYYY4wxpgb4JZAxxhhjjDHGGGNMDWjX6mDdunULVUjIIr9hw4YQGzJkCB6TjOpr1qwJMTKd03ep0hUZ8SkmSRMmTAgxqkQwYsSIEKOKGFOnTg2xXbt24bkJMttTX1IVMjo3VRcg0/3AgQNDjO61xFZ9qmpFFvqRI0eG2DPPPBNiF198cYhRNQeJq1XRuTdt2hRiVOGJ+puuj8aExKZ8MtOPGTMmxKi6EPUPVdCgikESV3Qgoz5VY6GqClRx4POf/3yI5aqVUXuooh99jsYzVYj4oz/6oxCj6lISm/upmkh5ncpVc6gb5557bqi8RhVannvuuRCjuStJo0aNCrF77rknxKiqBd07qvhE1StyFd8of9D4pgoPtHbQ/KO5lstblHvomNu3bw+xK6+8MsTa2tpCjCofde3aNcRorZV4LaM1eMCAASFG+YjyKF3zypUrsT0LFiwIMaoYSPOa+oeqgND9ogpbEle1pGukKiRU1eSHP/xhiNG6TJVXJK46RFUoaW9GFbVofsyYMSPEaCxLvK5Tn1Peoj0c7T1o/NBaIXGVTtqTlKvQ5fJg3ejatWuoBER5msYcVRCSeCxSRTxa62l/S+emz+WqX9GYpcpdVN2J2kj9Q7k1V62s6nPU4cOHQ4zWZRrLtP7TmpV7nqB7S9dIFZ/oOYGqBdI5aM8pcf9UHSu0plNOoOcBupbcZ2mfQvmW1izaB9MeOFc9jaAxQOemuUB7CnqWzVXPopxQNUbrd9VK4LRfk3j/Sfdr7dq1J/2cq/BXxr8JZIwxxhhjjDHGGFMD/BLIGGOMMcYYY4wxpgb4JZAxxhhjjDHGGGNMDfBLIGOMMcYYY4wxxpga0K5i6N///vdBfkWiMhKaHTx4EI9J0raqAjKSOJHcq3///iFGgjSJZUwktNq2bVuIkSxs69atIUaiKRKf5qBjkkyLhF9vv/12iFF/k+SM+kFi8RVJJWkMPProoyFGAjKSV7e2tmJ7SLxGwlj63IEDB0KM7jXdLxK+Siw1+9d//dcQo36k8Th27NgQI0leTixGolMSGJJc8NJLLw0xkpyRyC8nRKW202dpTFIbSZJO6wddsyQNHz48xMrSNimKqklcXUeOHz8e5huJy0nYSPNPqj5f6J7SunPeeeeFGBUFIHGxJC1atCjEli1bFmI07mi805ykuZaTvY8bNy7E/vEf/zHEKBdS3jp27FiIkVSY5N4k3JX4flO7KSds3rw5xEgMecstt1Q6r8TXvXv37hCjsUKiYcp5JBXN5VGSUq9YsSLESBhK92H27NkhRmOPYhLPRdo30dwmcSbtC2nc58TJJESl+UnrMO2FaEzR/mjo0KHYHpKD0z4jl2fqDuUJEunSPKf7JHH/0/x99913Q4zGJ+0vKO/koPNUFalTX5A0nfbLp9NGOiatjVWft2jNo71oTu5ddY2i+0WFD6gfac3KiX3pOYGuh8YKPRNQMRPK9bl1mdYTWsvoPlR9Pqb7n9vf0n2gZ1yi6nlo7c8VEKDv0/yiGF3Le++9V+lz9C5E4rFGa0D5mHQOwr8JZIwxxhhjjDHGGFMD/BLIGGOMMcYYY4wxpgb4JZAxxhhjjDHGGGNMDfBLIGOMMcYYY4wxxpga0K5i6HPPPTcIcUmaS8KvIUOG4DFJcvjGG2+EGMmVSJxEsr5du3aF2Lx587A9JAcjCdiwYcNCjORV48ePDzGStu3ZswfbQzK2+fPnhxjdB5IuTp48OcRIzk2fo+uTWCJK8jMSLJJMi0RcJPx75ZVXsD0kTyUJGIncSLBNYjn6bk7aSqJLknaS7LBPnz4hRn1G92D06NHYnnXr1oUYiTNp3NN8JVE5yXhJ2idxv9F9mDZtWojRWkES4VmzZoUY3UOJZXUUKwsD6TN15DOf+UyQzT///PPhcySkz8n1SHJIawKtlz169AgxEiQ+++yzITZw4EBsD63XJPujc59zzjkhRvJJ6guSpErSXXfdFWLf+ta3QuzHP/5xiJH0fMmSJSG2Y8eOECOZcS6XkUz/iSeeCLHrr78+xCjXk2CZzp2TyNJ8pXWHxinJr6lQAa2XJJCWOIfTukVrK7WH9hndu3cPsZz4mK6HcvOqVatCjESnJNOmAgKUTySpV69eIUY56oYbbggx2j/eeeedIfa5z30uxKjwgcTiVlqnymJiGmN1pFu3bkFKS2Ob1tCePXviMUmSTPeP5gYJZWkPTWMuN6fps3QeWidorafP0TXTvJI47w0aNCjE6Dnq+PHjIUbzquq+kdqS+z7dB1qXc9L9MjQm6BwSr3m076R2U46hfSe1J3ctdL8pt9Ixad7QeSjfkmBd4sILtDejfqS5Tc+OlNdzxaZobtNaTXsXaveWLVtCjPJtTuRN95ti5f7NHS+ct9KnjDHGGGOMMcYYY0yHxi+BjDHGGGOMMcYYY2qAXwIZY4wxxhhjjDHG1AC/BDLGGGOMMcYYY4ypASknN/006N27d3HttdeeFCN5EYmdSI4rsUiPBMRjx44NsZwYuMo5chIwkrGRoJOkVBMmTAgxkmmtX78+xEgALEnDhw+vdG4SbJGAjvqMxNAkGtu9eze2kQRtJFMuSwAllpKRZI/E4g8//DC2h65n5syZIUaStMGDB4cYCVFJzjZ37lxsD4kgN2zYEGI5UV4Zkv6RBC4n8l69enWIkbSNxu6YMWNCjMbeZZddFmIkJZVYIlq1f6hv6dwk6KNrlrjf6Nzl7z/wwANqa2uLduCa0a9fv+KLX/ziSTHqaxqHJH+VuP9p3SFI2EgsXLgwxF566SX87N69e0OMhM9V1yKaQ4888kiI3XTTTdgeWlvXrFkTYpdcckmIkeSS7te+fftCjNZLEk1LfL+uvPLKEKOccNFFF4UYybRJQHrPPfdge6jPqJADCb/pWlpaWkKM9hPUZ5I0ZcqUEFu+fHmI0d6F8i2dm+4/CZslzq80xml+US6kfEvjjPZ6Eu/jXnjhhRCjPTHlTDpPTrxOVO3zstT0wQcf1Kuvvlr7PHHBBRcUkyZNOilGQlq6d7n9O90TEhrTHpOOSWOJ9jE5iSvtyyjH0TygNYb2ZLSnp/2pxNdN6z/tq+h5gqT7VJyBzpuTV1Pbqc9p/aYYrRuUTyhfSnxvaB9MeYtir732WohR/i/PjQ+h+0DPZnQf6Lv0OZoLVFxHkjZv3hxi1GckbCYpOeUJyls5wT7dW3pmpr6oejzKg++++y5+n/Ij5Yly7O6779bRo0c/Nk/4N4GMMcYYY4wxxhhjaoBfAhljjDHGGGOMMcbUAL8EMsYYY4wxxhhjjKkBfglkjDHGGGOMMcYYUwOi7fRT5A9/+EMQXZLQmORcOcHmm2++GWLz5s0Lsba2tkrnIakria9I9pg7z9ChQ0OMRGUkOiOp1MCBA0OMBNC5dpKomiRZW7duDTESkF1xxRUhRgI6EqxJLOOj/iHZNPU3yQE3btwYYnQtkjRy5MhKn+3Xr1+I7d+/P8TofpH47NChQ9gekgZS/yxbtizESCK6c+fOSsfbtm0btodkbMOGDQuxqVOnhhhdN4mdScR2+PBhbA9J6EjQunLlyhAjuSfNQ5Kz5cSEJCGmNeDYsWMn/UzzoI787ne/C+JkWqsvv/zyEKNxI3HfVl1PSEpJc3XXrl0hlpMPkiz+mmuuCTHKR7TGLFq0KMRoDuSEnyRdpPWaJKK0dpA0ccSIESFGEsdcLqN1gs5Dcmea+zfeeGOIkaSSCjZIXKyiLPGV+N7QWCHhK8k0c4L8nFC7DN1DWtO3b99e6btbtmzB89B9oDH11a9+NcRoHtJ3W1tbQ2zdunXYHpoPJE+lsVIuaCKxVHr06NEhlsvrR48eDbHJkyeH2Lhx4076+fHHH8fj1Y33338/SHJpn3b++eeHGMmMJRauUqEI2mOSaJ7EzrRm5SSzVQW5BF0L7SVp3aG5m4PWcILWMhLk0jMh5e/cfonidG9IFkz3gZ6hqGBDrh9I0E19ThJo2kvS8xu1m9YXiccujbNNmzaFGO0fDhw4EGLU7tz+ncYardUUo4IttF+jvEWC7Vyc9u8k06bPUd9WvQcSryG0JpWfg6o+T/g3gYwxxhhjjDHGGGNqgF8CGWOMMcYYY4wxxtQAvwQyxhhjjDHGGGOMqQF+CWSMMcYYY4wxxhhTA9pVDP3+++8HGerbb78dPkeSwpwMkaRvJDQkcRbJXgmSM5PkTGKBKQl2SXJIYrGyIFViEW5O4kvySuqLWbNmhdiQIUNCjIRf1BckPiMRm8RyOJK+Ul/QMUnkTZAYVooiRklav359iJGgjSTQ1GckFT5y5Ai2h2S3x48fD7F33303xEgWSoI2kvHlRHckY6M5QtB4JGkbCVpprZBY+kdjioSKJOMj4TetKXSvJR6TJHIr9+P999+Px6sb7733XpDukUiPxHc0JyWWO9J8I6nsxIkTQ4zWPBoPuTxBOY7WcLoeGsfPPvtsiFGf5eTBzz33XIhRjiKROolvSXJI6wnN6c997nOV2zht2rQQo/VtwYIFIbZ69eoQo/0EFQqQWHz/L//yLyFG95AErbSu0thraWnB9qxatSrEiqIIMco9S5cuDTGS51JezgktSahN+wzKCQQJSGlPQOJ0iftiyZIlIUb9Q5JdKkBChUpye1cShlORjfK5qehGHTl+/HgYO7Tm0XpJc03inELjgeYv7Tnoc3QO2rvljkniWtoD0TVSPqKxnZvT1L9U/IDWPBq3NCfp+YbOkduf0mfpuYyeE6jdtH+n+0pzV+LnDBJL0/6B7iHdL7oWWi8lztdUgIK+T3OBxhTdaxqjEu8zSO5Mc4HaQ89GJHHO5Z1c8Ywqn6N201ihvqXCFxI/j9D+sdy/P/3pT/F4oX2VPmWMMcYYY4wxxhhjOjR+CWSMMcYYY4wxxhhTA/wSyBhjjDHGGGOMMaYGVHoJlFK6MKX0QEppZ0qpJaU0M6XUO6X065RSa+PvalIQY4wxnQ7nCWOMMafCecIYY5qDqmLof5D0eFEUN6eUzpHUQ9J/lbS0KIo7Ukq3S7pd0t+c8mTdugVxEgllSc6VE2ySVIqkZiQQJOEjSZxIXklyt1ycRGc5qVkZEqeR0CrXHuqLYcOGhdjzzz8fYsOHDw8xug8XX3xxiJEYmiReEvcPCTqp3VUhiReJxiRpz549ITZ48OAQIxkbCZZJ5Hb06NEQI3mZxGPyhRdeCDESi9G5BwwYEGIkT7z66quxPSSRo3FB95WkhiRT27lzZ4jl5Ikkq12xYkWIzZ8/P8Tef//9EHvqqadCjMYKCQgl6cEHHwyx6667LsTK6xy1pYNxRvLEOeecE0TatJ4MHTo0xDZt2oTHJFngSy+9FGK0xpDYlaSwtB7kJK4k9qM1mNYtGnc9evQIMZovORkiSRspN5N0uaqoktbG0aNHh9jKlSuxjZTL9u3bF2K0llFfkIiT1qxdu3Zhe0gYevPNN4cY7T2q5kfKCXSvJR5rlMsoT5A4neYH7bduu+02bE9ra2uIbdiwIcRo7NH9IkkqjYncOkrxH//4xyG2cOHCEKMxRd+dMWNGiOUKGixbtizEaG6XC3TQvepgnJE80bVr1zCPaN9Isdw9oTFWdQ9Oe3q6nzQOaY+Xi5OcmeYQrcsUy/UFQfOS5gaJb6kIB623VffqucI+VYumkOSe+pti1N85ATyNFZL90n2gvqXxQ/vqF198EdvT1tYWYjt27Agxmjc0fuj5j0TcVHBH4jFARQmof8qFpiTuM3qeoHkkcUEkKhgyadKkEKN2U8EPel6n/ZEkbdy4McTmzJkTYuU9ZdXniY/9TaCUUk9J8yT9SJKKovh9URS/kfR5SXc1PnaXpJsqndEYY0ynwnnCGGPMqXCeMMaY5qHKPwcbKulVSXemlDamlH6YUjpPUv+iKF6WpMbf8dWmMcaYOuA8YYwx5lQ4TxhjTJNQ5SVQN0mTJP1TURQTJb2jE7+qWYmU0jdSSutSSuvoV8mMMcZ0eJwnjDHGnIozlifon/0YY4ypTpWXQIclHS6KYk3j5wd0YhE/mlIaIEmNv+M/NJRUFMUPiqKYUhTFlJy3xhhjTIfGecIYY8ypOGN5IufkM8YYU42PfQlUFMUrkg6llD60KF8jaYekX0n6WiP2NUkPfyotNMYY09Q4TxhjjDkVzhPGGNM8VK0O9p8k3d0w+e+V9B914gXSz1JKX5d0UNItH3uybt1CxRGqaEHVh8jGnoMs9FXt8ocOHQoxamPOBE+VO6g9dO6DBw+GGFXjoF+DzVXeovaQrXzUqFEhRhVeqD3E9u3bQyx3DylORv61a9eGGJngJ06cGGJkls/12a233hpiZM9fv359pXOTjZ/aTYZ/SWppaQkxqnZA10OVwObOnRtiVCnhlVdewfbQeegeUvWcI0eOhBhVeaL+oWp+ElfaoflOlY2oggJVEqCKMR988AG2h+YIVXMrVxPJVW7qQJyRPNGlS5ewblEFuZdffjnEqFKFxOstrTG0BlOlC5oDW7ZsCbGpU6die6gKDVWvW716dYhR9RRadyiX5SpLjRgxIsSoShf1BX2Xqh/SPKUKSeXKcB9C94vWDlqrKUZzumolIIkrxFG1MqqUc9VVV4UYVX2iNfThh/n5mHLCm2++GWJUjYVy/Te/+c0Qo3uwf/9+bA9Vl5k+fXqIUfUTytdUXYbm4bRp07A9VLmF5hJVRaMYVZeh32K85JJLKreH8kS5aiDtjzsYZ+x5orxPoH0IVZDL7d/p/tGYr1pFjOYf/QZT7reaaB+TW8PL0JymdtPzxOlUT6O1nvJw1RxM7aG9KLVFylf9LUN7UXpWo2pRVHkpV712woQJIUbrCe07aTzT/Ke9KFWGlPj5j+439S+tZbT3oOeJXB6lHEfzgcYPVQejPQrtyXPPspS3qD20ftMzHFUHnTx5cohRP0g8nrdu3Rpi5WcZmv9EpZdARVFskjQF/qdrKp3FGGNMp8Z5whhjzKlwnjDGmOagihPIGGOMMcYYY4wxxnRw/BLIGGOMMcYYY4wxpgb4JZAxxhhjjDHGGGNMDagqhj4jdO/eXcOGDTsp9oc//CF8jgRQOXEafZ/EYiRDJAkYSaVI7paTbpGUcMOGDSE2Y8aMECNxLYmmBg8eHGIk+5Kkfv36hRjJFEk2RqKqSy+9NMRIsEb3gERaEgsxSSJKElASWu/atSvESPaVE5CSnLEsNJdYVkuyMRqjP//5z0OM+lHi+zB79uwQu/POO0OMpHYk1CXRWA4SxlFfEGPGjAkxmu8kyV2zZk2ISdw/JNQjWSHJ/GbNmhViAwcODDGS9kosyqU27t69+6SfSTRaR3r06BHEeTR/ab7kZOa0ZpKckeTjNJbK907idSw3Rq65JuovHnzwwRC77rrrQozmH4kdr7zyyhCjdkvcduoL6t9nnnkmxG644YYQIwEwXQvJHnOfpTxMuePmm28OMRJfkzC2T58+2J5Vq1aFGM1hkvNXlcg++eSTIZYTw1IBgilTonrlvvvuC7GZM2eGGBU+oHGSE6KSPLWqFJf6jKTtJF2lfZTEuZnmDeUykppSgQXao+Ty1qBBg0KMBK+tra0n/Ux7qzpyzjnnhD6kMVdVzixx39IxCZJK01ylvQC1UeL9Nj0TkLCXxiztRWm9fPXVV7E99MxE10MyburHcnEMifuC5kVu3aH7QIVGqG+r5jwSyucE8FR8pKqomMYprcFPPfVUiOWE/bTfpkIFy5cvDzHaR9E4oyI+uXlEeZ32ewT1OZ1n9OjRIUbCZomvh+43jSk6N4mzKb/lnuGpz2mfURaL50TTZfybQMYYY4wxxhhjjDE1wC+BjDHGGGOMMcYYY2qAXwIZY4wxxhhjjDHG1AC/BDLGGGOMMcYYY4ypAe0qhv7ggw+CzOvQoUPhcyRxzomiSOJHwieSK9HnSNZHMrWpU6die0gE94UvfCHESBZN8mESQJG8mvpRYtEdxc4555wQI8khCb9IsEYy1ZwYmq6b5I4k5yU53PDhw0Ns3759IYy7FGcAAAhaSURBVJaTe1999dUhRrI6ug+33357iJGcmdpDkkqJZaUkB/zTP/3TEKM+ozFK4rPcmCKpGYkESQ5I94sEm5s2bQoxmoeS9Nvf/jbEaAzQ90lUTTI+ktWOGzcO20N9SXLZsriPhH91JKUU1iOSHD700EMhRkJxiWX4JEikuUZiR8o7NCdJCiix6PKOO+4IsXvvvTfEKCfQeWiNyUllW1paQozmKq3VJKAmyTXlYJKoU39L0nnnnRdiVSWpNPdpXV66dGmI0Roh8V6BZOOUHylP0NiltTon96YxdeDAgRCjohTz5s0LMWo35QTaE+SOSTJtWvdI+k7SVpJX58YPCXDHjh0bYiT3nDRpUoiRQJTk1dSPEgtRSeZZlvHmpMZ1oyiK0F8k8aV9CO13JV5jqGgKjSXaI5A8lqB1VaouIKb9Mn2XxjZdS05U3aVL/L0BOiZ9jgpz0HpJ7SHZM+VviQXL9OxRLlIkcX6k/j569GiI5Qq70Hmof2mt/973vhdi1I87duwIsZzcm/Y49OxBeYJyJvU33a9ce0gOTvs1Gme0XlbdC+Wet2gM0LMw5Vt67qC1hp5ZqMiRxO8+6HminPeqPk/4N4GMMcYYY4wxxhhjaoBfAhljjDHGGGOMMcbUAL8EMsYYY4wxxhhjjKkBfglkjDHGGGOMMcYYUwNSURTtd7KUXpV0QFJfSdHo2HHpTNfja2leOtP1+Foig4qiiNa/muE80SHwtTQnnelapM51Pc4TZ5BOmic607VInet6fC3NS2e6nnbNE+36Euj/nTSldUVRTGn3E39KdKbr8bU0L53penwt5uPobP3ama7H19KcdKZrkTrX9XSma2kmOlO/dqZrkTrX9fhampfOdD3tfS3+52DGGGOMMcYYY4wxNcAvgYwxxhhjjDHGGGNqwNl6CfSDs3TeT4vOdD2+lualM12Pr8V8HJ2tXzvT9fhampPOdC1S57qeznQtzURn6tfOdC1S57oeX0vz0pmup12v5aw4gYwxxhhjjDHGGGNM++J/DmaMMcYYY4wxxhhTA9r9JVBKaUlKaVdK6cWU0u3tff5PQkrpxymltpTSto/EeqeUfp1Sam38fdHZbGNVUkqXp5SWpZRaUkrbU0p/1Yh31Os5N6X0Qkppc+N6/nsjPiSltKZxPfenlM45222tSkqpa0ppY0rpkcbPHfJaUkr7U0pbU0qbUkrrGrEOOc4kKaV0YUrpgZTSzsb8mdmRr6cZcZ5oDpwnmh/niebDOaJ9cJ5oDjpTnnCOaG6cJ84s7foSKKXUVdL/kXS9pDGS/kNKaUx7tuET8hNJS0qx2yUtLYriCklLGz93BN6X9O2iKEZLmiHpLxv3oqNez+8kXV0UxXhJEyQtSSnNkPS3kv6ucT1vSPr6WWzj6fJXklo+8nNHvpaFRVFM+Ejpw446ziTpHyQ9XhTFKEnjdeIedeTraSqcJ5oK54nmx3mi+XCO+JRxnmgqOlOecI5ofpwnzhRFUbTbH0kzJT3xkZ+/I+k77dmGM3ANgyVt+8jPuyQNaPz3AEm7znYb/53X9bCkxZ3heiT1kLRB0nRJr0nq1oifNP6a+Y+kyxoLwNWSHpGUOvC17JfUtxTrkONMUk9J+9TwqXX062nGP84TzfvHeaK5/jhPNN8f54h262fniSb901nyhHNE8/1xnjizf9r7n4NdKunQR34+3Ih1ZPoXRfGyJDX+7neW23PapJQGS5ooaY068PU0fuVxk6Q2Sb+WtEfSb4qieL/xkY403v5e0m2SPmj83Ecd91oKSU+mlNanlL7RiHXUcTZU0quS7mz8eu0PU0rnqeNeTzPiPNGEOE80Jc4TzYdzRPvgPNGEdIY84RzR1DhPnEHa+yVQgpjLk51FUkrnS/qFpP9cFMVvz3Z7PglFURwvimKCTrz5niZpNH2sfVt1+qSU/kRSW1EU6z8aho82/bU0mF0UxSSd+LXtv0wpzTvbDfoEdJM0SdI/FUUxUdI76ji/etpR6MhjvVPiPNF8OE80Lc4R7UNHHuudks6SJ5wjmhrniTNIe78EOizp8o/8fJmkI+3chjPN0ZTSAElq/N12lttTmZTSZ3Riwb67KIpfNsId9no+pCiK30harhP/NvnClFK3xv/UUcbbbEk3ppT2S7pPJ36N8+/VMa9FRVEcafzdJulBnUiqHXWcHZZ0uCiKNY2fH9CJhbyjXk8z4jzRRDhPNC3OE82Jc0T74DzRRHTGPOEc0Xw4T5xZ2vsl0FpJVzTM5OdIulXSr9q5DWeaX0n6WuO/v6YT/xa26UkpJUk/ktRSFMX3P/I/ddTruTildGHjvz8raZFOSLaWSbq58bEOcT1FUXynKIrLiqIYrBNz5OmiKP5MHfBaUkrnpZQu+PC/JV0raZs66DgriuIVSYdSSiMboWsk7VAHvZ4mxXmiSXCeaF6cJ5oT54h2w3miSehMecI5onlxnvh0GtLeMqQ/lrRbJ/6N5Xfb+/yfsO33SnpZ0h904i3e13Xi31culdTa+Lv32W5nxWuZoxO/ArhF0qbGnz/uwNczTtLGxvVsk/S9RnyopBckvSjp55K6n+22nuZ1LZD0SEe9lkabNzf+bP9wznfUcdZo+wRJ6xpj7SFJF3Xk62nGP84TzfHHeaJj/HGeaK4/zhHt1s/OE03wpzPlCeeI5v3jPHHm/6RGQ4wxxhhjjDHGGGNMJ6a9/zmYMcYYY4wxxhhjjDkL+CWQMcYYY4wxxhhjTA3wSyBjjDHGGGOMMcaYGuCXQMYYY4wxxhhjjDE1wC+BjDHGGGOMMcYYY2qAXwIZY4wxxhhjjDHG1AC/BDLGGGOMMcYYY4ypAX4JZIwxxhhjjDHGGFMD/i9/xRzyZ/7gXAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 1440x720 with 3 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, axes = plt.subplots(1,3,figsize=(20, 10))\n",
    "axes[0].imshow(self_samples.x[0,:,:,].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "axes[1].imshow(self_samples.y[0,:,:,].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "axes[2].imshow(img_denoised[0,:,:,:].permute(1, 2, 0).cpu().numpy().squeeze(), cmap='gray')\n",
    "fig.canvas.draw()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TESTING | MSE: 0.001272 | avg PSNR for the given testset: 26.311, Time: 00:05 mins---\n",
      "TESTING | MSE: 0.001159 | avg PSNR for the given testset: 26.571, Time: 01:24 mins---\n",
      "TESTING | MSE: 0.000970 | avg PSNR for the given testset: 27.664, Time: 02:46 mins---\n"
     ]
    }
   ],
   "source": [
    "_, _ = Sur.testing_error(energy, loss, datasetTest);\n",
    "_, _ = Sur.testing_error(energy, loss, datasetTestFull);\n",
    "_, _ = Sur.test_twice_error(energy, loss, datasetTestFull);"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Plotting"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## save filters to pdf:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# fig.savefig(\"filters_168_13_13.pdf\", bbox_inches='tight', dpi=1200)"
   ]
  }
 ],
 "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": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}