CIFAR2/methods/04_if/01_IF_val_5000-0.5.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "31e0858d-12e5-4eaf-8cc3-f461d2aff756",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "974ba093-b42b-4ee9-a8c0-cd79d161435c",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8c2f5efb-631c-4083-84ae-dc219c6a75a1",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fde1d3d3-3ae3-4194-b2e8-596809be9636",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9e3a5e70-ec86-4af6-ad75-bdc4351b4d40",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "723df846-36f4-433c-99cf-044b96c870ec",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cf1854ba-184b-4c9f-9a5c-9158e786e404",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "c4cadde9-8859-435a-b6ba-09aa42bb468d",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "os.environ[\"CUDA_DEVICE_ORDER\"]=\"PCI_BUS_ID\"   \n",
    "os.environ[\"CUDA_VISIBLE_DEVICES\"]='1'\n",
    "\n",
    "os.environ[\"HF_HOME\"]=\"~/codes/.cache/huggingface\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "2f7cd96b-ca27-4ce1-ac2d-c3dd67e069b7",
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import random\n",
    "import numpy as np\n",
    "\n",
    "def set_seeds(seed):\n",
    "    random.seed(seed)\n",
    "    np.random.seed(seed)\n",
    "    torch.manual_seed(seed)\n",
    "    torch.backends.cudnn.deterministic = True\n",
    "    torch.backends.cudnn.benchmark = False\n",
    "    \n",
    "set_seeds(42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "4ff495b8-8df0-49e8-a409-4ed9601c8143",
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "def show_images(images, scores, test_artist, train_artists):\n",
    "    n: int = len(images)\n",
    "    f = plt.figure(figsize=(16, 2))\n",
    "    for i in range(n):\n",
    "        # Debug, plot figure\n",
    "        ax = f.add_subplot(1, n, i + 1)\n",
    "        if i==0:\n",
    "            pass\n",
    "            ax.title.set_text(test_artist)\n",
    "        else:\n",
    "            ax.title.set_text(str(np.round(scores[i-1], 4))+'\\n'+train_artists[i-1])\n",
    "            ax.axis('off')\n",
    "        if images[i]==None:\n",
    "            pass\n",
    "        else:\n",
    "            plt.imshow(images[i])\n",
    "\n",
    "    plt.show(block=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "c4e89a09-d6ff-46a1-8a9e-9c6f9c167114",
   "metadata": {},
   "outputs": [],
   "source": [
    "from datasets import load_dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "2682e2c4-9275-4bae-a732-74b981a358d2",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "fb2dc562-293c-4e74-8e97-141fc7465938",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "bda41875-8976-4852-823a-91dc1f42aebf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5000"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "with open('../../data/indices/5000-0.5/idx-train.pkl', 'rb')  as handle:\n",
    "    idx_train = pickle.load(handle)\n",
    "len(idx_train)   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "5ffbb813-1636-44e3-b943-400824fde5e7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1000"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "with open('../../data/indices/5000-0.5/idx-val.pkl', 'rb')  as handle:\n",
    "    idx_val = pickle.load(handle)\n",
    "len(idx_val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "773b1a12-d60d-4ed8-8f61-b03a3af61e71",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "40c86797-31d5-45d0-b46f-52f8c84dd130",
   "metadata": {},
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'load_dataset' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[1], line 3\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[38;5;66;03m# from datasets import load_dataset\u001b[39;00m\n\u001b[0;32m----> 3\u001b[0m dataset \u001b[38;5;241m=\u001b[39m \u001b[43mload_dataset\u001b[49m(\n\u001b[1;32m      4\u001b[0m     \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mcifar10\u001b[39m\u001b[38;5;124m'\u001b[39m,\n\u001b[1;32m      5\u001b[0m     \u001b[38;5;66;03m# args.dataset_config_name,\u001b[39;00m\n\u001b[1;32m      6\u001b[0m     \u001b[38;5;66;03m# cache_dir=args.cache_dir,\u001b[39;00m\n\u001b[1;32m      7\u001b[0m     \u001b[38;5;66;03m# split=\"train\",\u001b[39;00m\n\u001b[1;32m      8\u001b[0m )\n",
      "\u001b[0;31mNameError\u001b[0m: name 'load_dataset' is not defined"
     ]
    }
   ],
   "source": [
    "from datasets import load_dataset\n",
    "\n",
    "dataset = load_dataset(\n",
    "    'cifar10',\n",
    "    # args.dataset_config_name,\n",
    "    # cache_dir=args.cache_dir,\n",
    "    # split=\"train\",\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c6632d56-9134-4720-99a9-1853fe349039",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b2e50aae-335f-457d-bfb7-e3991f19fb50",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5df420e1-7166-4e5b-baa3-a012fe7e32e1",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "f2ce6b6b-ff45-4156-9606-7c87046e751a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<PIL.PngImagePlugin.PngImageFile image mode=RGB size=32x32>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from datasets import DatasetDict, Dataset, load_dataset, Image\n",
    "\n",
    "train_dataset = dataset['train'].select(idx_train)\n",
    "train_dataset[0][\"img\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "26b8b800-7147-4c3b-9f62-b99110392974",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "087ea1db-01c9-4e6a-a402-8b8a9aa93e8b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<PIL.PngImagePlugin.PngImageFile image mode=RGB size=32x32>"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val_dataset =  dataset['test'].select(idx_val)\n",
    "val_dataset[0][\"img\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "e16837b7-8ddd-46ab-9d0a-c988f2ca9514",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[7405, 5226, 1363, 6615, 7612]"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "idx_val[0:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0dd5e770-1fed-4f65-b4e7-5373e2adcb10",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "317f972c-b763-40d3-8186-759118c12237",
   "metadata": {},
   "outputs": [],
   "source": [
    "class Args():\n",
    "            \n",
    "    dataset_name=\"cifar10\"\n",
    "\n",
    "    output_dir=\"../../saved/5000-0.5/ddpm/checkpoint-8000\"\n",
    "    \n",
    "    # K=10\n",
    "    Z=65536\n",
    "\n",
    "    used_dim=32768\n",
    "    \n",
    "    t_strategy='uniform'\n",
    "   \n",
    "args = Args()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "bb9d8f5f-5370-4157-be84-f77949771866",
   "metadata": {},
   "outputs": [],
   "source": [
    "import diffusers\n",
    "from diffusers import DDPMPipeline, DDPMScheduler, UNet2DModel\n",
    "from diffusers.optimization import get_scheduler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "3ffe8c0c-a6df-41e9-8cd9-4dfa6eaa4d89",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "0740ed2c-c98c-494d-bc1f-381653a77964",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2728"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.random_projection import johnson_lindenstrauss_min_dim\n",
    "johnson_lindenstrauss_min_dim(n_samples=1e3, eps=0.15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "58c5c4e8-7171-4847-9fba-51ea76b2c794",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3002"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.random_projection import johnson_lindenstrauss_min_dim\n",
    "johnson_lindenstrauss_min_dim(n_samples=2e3, eps=0.15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "32c7cc29-8267-4fc3-915f-2628006b3af9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3364"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.random_projection import johnson_lindenstrauss_min_dim\n",
    "johnson_lindenstrauss_min_dim(n_samples=5e3, eps=0.15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "5396c1cc-03c9-407d-b4ba-d16f4b56bc25",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3638"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.random_projection import johnson_lindenstrauss_min_dim\n",
    "johnson_lindenstrauss_min_dim(n_samples=1e4, eps=0.15)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a8c9ec30-0da2-4b68-9f5f-8e9e0faab1bc",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "04be809b-6fd9-4a8a-9d64-94c6ef3b4f85",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ad307d24-057d-43bc-9772-5ae5d62c9fa1",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "c37c6ab7-8c07-416a-b7b1-4b2c03a82cfa",
   "metadata": {},
   "outputs": [],
   "source": [
    "# load ground truth\n",
    "my_list = [\n",
    "    0,1,2,3,\n",
    "    4,5,6,7,\n",
    "    8,9,10,11,\n",
    "    12,13,14,15,\n",
    "    16,17,18,19,\n",
    "    20,21,22,23,\n",
    "    24,25,26,27,\n",
    "    28,29,30,31,\n",
    "    32,33,34,35,\n",
    "    36,37,38,39,\n",
    "    40,41,42,43,\n",
    "    44,45,46,47,\n",
    "    48,49,50,51,\n",
    "    52,53,54,55,\n",
    "    56,57,58,59,\n",
    "    60,61,62,63,\n",
    "          ]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "7e5b270b-c5df-4b5d-a8fd-2a924a6edad6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(64, 1000, 1000)"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "loss_array_list = []\n",
    "\n",
    "for i in my_list:\n",
    "    for seed in [\n",
    "        0,\n",
    "                 1,\n",
    "                 2,\n",
    "                 # 3,\n",
    "                 # 4,\n",
    "                ]:\n",
    "        for e_seed in [\n",
    "            0, \n",
    "                       1, \n",
    "                       2\n",
    "                      ]:\n",
    "            with open('../../saved/5000-0.5/lds-val/ddpm-sub-{}-{}/e-{}-val.pkl'.format(i, seed, e_seed), 'rb')  as handle:\n",
    "                loss_list = pickle.load(handle)\n",
    "            margins = np.concatenate(loss_list, axis=-1) # -logp\n",
    "            ####\n",
    "            if (seed==0) and (e_seed)==0:\n",
    "                loss_array = margins\n",
    "            else:\n",
    "                loss_array += margins\n",
    "            \n",
    "    loss_array = loss_array/(3*3)\n",
    "    \n",
    "    loss_array_list.append(loss_array)\n",
    "lds_loss_array = np.stack(loss_array_list)\n",
    "lds_loss_array.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "d823ef7b-9d53-4f31-ab11-864274554493",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for i in range(1):\n",
    "    for j in range(8):\n",
    "        plt.plot(lds_loss_array[i, :, j], color=\"C{}\".format(j))\n",
    "    # break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "874a332c-29b1-46fe-bf83-12e7c8f64e28",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(64, 5000)"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mask_array_list = []\n",
    "\n",
    "for i in my_list:\n",
    "    # print(i)\n",
    "    with open('../../data/indices/5000-0.5/lds-val/sub-idx-{}.pkl'.format(i), 'rb')  as handle:\n",
    "        sub_idx_train = pickle.load(handle)\n",
    "    # print(len(sub_idx_train))\n",
    "    mask_array = np.in1d(idx_train, sub_idx_train)\n",
    "        \n",
    "    mask_array_list.append(mask_array)\n",
    "    \n",
    "lds_mask_array = np.stack(mask_array_list)\n",
    "lds_mask_array.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "41ebfcfb-5789-44f0-99d0-0dddc7c6f32f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(64, 1000)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lds_testset_correctness = lds_loss_array.mean(axis=1)\n",
    "lds_testset_correctness.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "46afee77-65eb-4023-8077-d11ed785d406",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for j in range(4):\n",
    "    plt.plot(lds_testset_correctness[:, j], color=\"C{}\".format(j))\n",
    "    # break\n",
    "# plt.ylim(0.15, 0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8d92c4b4-5b21-434f-b6cc-96e968cbe495",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "649a172c-968c-4aa3-b086-b32d64d22424",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "27"
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from scipy.stats import spearmanr, pearsonr\n",
    "\n",
    "lamb_list = [\n",
    "        1e-2, 2e-2, 5e-2,\n",
    "        1e-1, 2e-1, 5e-1,\n",
    "        1e0, 2e0, 5e0,\n",
    "        1e1, 2e1, 5e1,\n",
    "        1e2, 2e2, 5e2,\n",
    "        1e3, 2e3, 5e3, \n",
    "        1e4, 2e4, 5e4, \n",
    "        1e5, 2e5, 5e5, \n",
    "        1e6, 2e6, 5e6, \n",
    "    ]\n",
    "\n",
    "len(lamb_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e1b7c56d-0212-4aca-ba16-819db366b19a",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "effb0646-37be-44c2-8b2e-7bf3c67ea049",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5000, 65536)\n",
      "(5000, 32768)\n",
      "(1000, 65536)\n",
      "(1000, 32768)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_4009/1445120343.py:32: UserWarning: The given NumPy array is not writable, and PyTorch does not support non-writable tensors. This means writing to this tensor will result in undefined behavior. You may want to copy the array to protect its data or make it writable before converting it to a tensor. This type of warning will be suppressed for the rest of this program. (Triggered internally at ../torch/csrc/utils/tensor_numpy.cpp:206.)\n",
      "  gen_dstore_keys = torch.from_numpy(gen_dstore_keys).cuda()\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([32768, 32768])\n",
      "tensor(39.2496, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: -0.006 (avg p value 0.492188)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(39.1747, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.007 (avg p value 0.493739)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(18.8999, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.084 (avg p value 0.439958)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(8.6607, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.087 (avg p value 0.441724)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(4.2605, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.088 (avg p value 0.440291)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(1.6972, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.088 (avg p value 0.439886)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.8484, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.088 (avg p value 0.440005)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.4245, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.089 (avg p value 0.439770)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.1703, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.089 (avg p value 0.437719)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0855, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.090 (avg p value 0.435699)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0431, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.092 (avg p value 0.431199)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0175, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.096 (avg p value 0.425575)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0089, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.100 (avg p value 0.418670)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0046, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.104 (avg p value 0.407926)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0019, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.110 (avg p value 0.394776)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0010, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.113 (avg p value 0.385295)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0005, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.115 (avg p value 0.380145)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0002, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.115 (avg p value 0.386765)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(9.9095e-05, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.114 (avg p value 0.388968)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(4.9731e-05, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.112 (avg p value 0.392252)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(1.9948e-05, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.105 (avg p value 0.398832)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(9.9850e-06, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.097 (avg p value 0.406524)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(4.9957e-06, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.087 (avg p value 0.419626)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(1.9992e-06, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.073 (avg p value 0.451035)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(9.9977e-07, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.065 (avg p value 0.463086)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(4.9993e-07, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.059 (avg p value 0.465625)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(1.9999e-07, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.054 (avg p value 0.468292)\n",
      "(5000, 65536)\n",
      "(5000, 32768)\n",
      "(1000, 65536)\n",
      "(1000, 32768)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(60.7976, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: -0.012 (avg p value 0.499995)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(51.3385, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.010 (avg p value 0.523637)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(17.5794, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.194 (avg p value 0.244985)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(8.5538, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.197 (avg p value 0.242606)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(4.2536, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.197 (avg p value 0.241266)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(1.7027, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.199 (avg p value 0.237770)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.8545, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.202 (avg p value 0.233116)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.4303, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.205 (avg p value 0.225925)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.1750, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.213 (avg p value 0.215066)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0892, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.221 (avg p value 0.202207)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0456, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.230 (avg p value 0.188052)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0188, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.238 (avg p value 0.179244)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0096, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.240 (avg p value 0.176403)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0049, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.237 (avg p value 0.182372)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0020, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.222 (avg p value 0.202746)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0010, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.206 (avg p value 0.226456)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0005, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.188 (avg p value 0.253716)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(0.0002, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.164 (avg p value 0.297000)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(9.9846e-05, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.147 (avg p value 0.326063)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(4.9955e-05, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.130 (avg p value 0.358008)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(1.9991e-05, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.106 (avg p value 0.392117)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(9.9973e-06, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.088 (avg p value 0.406547)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(4.9991e-06, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.074 (avg p value 0.421729)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(1.9998e-06, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.061 (avg p value 0.436548)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(9.9993e-07, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.056 (avg p value 0.440353)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(4.9998e-07, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.054 (avg p value 0.443044)\n",
      "torch.Size([32768, 32768])\n",
      "tensor(2.0000e-07, device='cuda:0')\n",
      "torch.Size([1000, 5000])\n",
      "Correlation: 0.052 (avg p value 0.445051)\n"
     ]
    }
   ],
   "source": [
    "# for loop over functions\n",
    "lds_list = []\n",
    "for K in [10, 100, \n",
    "          # 1000\n",
    "         ]:\n",
    "    ####\n",
    "    ####\n",
    "    dstore_keys_list = []\n",
    "    for split in range(5):\n",
    "        dstore_keys = np.memmap('{}/features-0/ddpm-train-keys-{}-{}-{}-{}-{}.npy'.format(\n",
    "            args.output_dir, split, K, args.Z, 'loss', args.t_strategy), \n",
    "                                dtype=np.float32, \n",
    "                                mode='r',\n",
    "                                shape=(1000, args.Z))\n",
    "        dstore_keys_list.append(dstore_keys) \n",
    "        \n",
    "    dstore_keys = np.vstack(dstore_keys_list)\n",
    "    print(dstore_keys.shape)  \n",
    "    dstore_keys = dstore_keys[:, :args.used_dim]\n",
    "    print(dstore_keys.shape)\n",
    "    ####        \n",
    "    gen_dstore_keys = np.memmap('{}/features-0/ddpm-val-keys-{}-{}-{}-{}-{}.npy'.format(\n",
    "                args.output_dir, 0, K, args.Z, 'loss', args.t_strategy), \n",
    "                                dtype=np.float32, \n",
    "                                mode='r',\n",
    "                                shape=(1000, args.Z))\n",
    "    print(gen_dstore_keys.shape)    \n",
    "    gen_dstore_keys = gen_dstore_keys[:, :args.used_dim]\n",
    "    print(gen_dstore_keys.shape)\n",
    "    ####\n",
    "    dstore_keys = torch.from_numpy(dstore_keys).cuda()\n",
    "    gen_dstore_keys = torch.from_numpy(gen_dstore_keys).cuda()\n",
    "    # dstore_keys = torch.from_numpy(dstore_keys).cuda() * 1000\n",
    "    # gen_dstore_keys = torch.from_numpy(gen_dstore_keys).cuda() * 1000\n",
    "    ####\n",
    "    kernel = dstore_keys.T@dstore_keys\n",
    "    ####\n",
    "    rs_list = []\n",
    "    ps_list = []\n",
    "\n",
    "    best_scores = None\n",
    "    best_lds = -np.inf\n",
    "    for lamb in lamb_list:\n",
    "        # kernel_ = kernel + lamb*np.eye(kernel.shape[0])\n",
    "        kernel_ = kernel + lamb*torch.eye(kernel.shape[0]).cuda()\n",
    "        # kernel_ = torch.from_numpy(kernel_).cuda()\n",
    "        ####\n",
    "        kernel_ = torch.linalg.inv(kernel_)  \n",
    "        # kernel_ = kernel_.cpu().numpy()\n",
    "        ####\n",
    "        print(kernel_.shape)\n",
    "        print(torch.mean(kernel_.diagonal()))\n",
    "        ####\n",
    "        # scores = gen_dstore_keys.dot((dstore_keys@kernel_).T)\n",
    "        scores = gen_dstore_keys@((dstore_keys@kernel_).T)\n",
    "        print(scores.size())\n",
    "        scores = scores.cpu().numpy()\n",
    "        \n",
    "        # compute lds\n",
    "\n",
    "        margins = lds_testset_correctness\n",
    "        infl_est_ = -scores\n",
    "        preds = lds_mask_array @ infl_est_.T\n",
    "####\n",
    "        rs = []\n",
    "        ps = []\n",
    "\n",
    "        for ind in range(1000):\n",
    "           r, p = spearmanr(preds[:, ind], margins[:, ind])\n",
    "           # r, p = pearsonr(preds[:, ind], margins[:, ind])\n",
    "           rs.append(r)\n",
    "           ps.append(p)\n",
    "    \n",
    "        rs, ps = np.array(rs), np.array(ps)\n",
    "        print(f'Correlation: {rs.mean():.3f} (avg p value {ps.mean():.6f})')\n",
    "\n",
    "        rs_list.append(rs.mean())   \n",
    "        ps_list.append(ps.mean())\n",
    "\n",
    "        if rs.mean()>best_lds:\n",
    "            best_scores = scores\n",
    "            best_lds = rs.mean()\n",
    "\n",
    "    with open('./val_5000_0.5_{}_{}_{}_{}.pkl'.format('loss', K, lamb, best_lds), 'wb') as handle:\n",
    "        pickle.dump(best_scores, handle)\n",
    "            \n",
    "    lds_list.append((rs_list, ps_list))\n",
    "\n",
    "    # break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c3380ec5-8479-45b3-81c8-78d83cb98f3d",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "47e17534-05f6-4f8d-a9b8-7ea2c69258fb",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "c08ec20a-9e82-4f58-b342-95145208f5b3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[-0.0060387 ,  0.00709208,  0.08383473,  0.08743059,  0.08801133,\n",
       "         0.08831383,  0.08843255,  0.08867818,  0.08936378,  0.09036001,\n",
       "         0.09234519,  0.0958943 ,  0.09956214,  0.10417619,  0.11004785,\n",
       "         0.11309124,  0.1152296 ,  0.11517811,  0.11402312,  0.11157539,\n",
       "         0.10470684,  0.09677891,  0.08700027,  0.07285058,  0.06462047,\n",
       "         0.05914236,  0.05444904],\n",
       "       [ 0.49218787,  0.4937387 ,  0.4399579 ,  0.44172443,  0.44029094,\n",
       "         0.43988609,  0.44000519,  0.4397705 ,  0.43771866,  0.43569874,\n",
       "         0.43119857,  0.42557527,  0.41866983,  0.4079264 ,  0.39477587,\n",
       "         0.38529549,  0.38014474,  0.38676466,  0.38896755,  0.39225188,\n",
       "         0.39883151,  0.40652357,  0.41962602,  0.45103456,  0.46308593,\n",
       "         0.46562519,  0.4682921 ]])"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lds_array = np.array(lds_list)\n",
    "lds_array[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "c7d99b73-0b7e-4685-af5e-20038c48751d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2, 2, 27)"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lds_array.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "bb765d6a-70b5-435d-baed-e456e650ea69",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "3798369f-a4c3-41c1-a02e-84ba7c5395c0",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open('./val_5000_0.5_{}.pkl'.format('loss'), 'wb') as handle:\n",
    "    pickle.dump(lds_array, handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "af9b280d-3e7b-472f-9a68-8417a27e7bc6",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "8122043b-7cc0-4d50-a1a5-98e6eed9a4ed",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.1152296 , 0.23996492])"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lds_array[:, 0].max(axis=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "80995dab-f837-4566-9ea9-ccb5b1c5b0fd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAiwAAAGfCAYAAAB8wYmvAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABXP0lEQVR4nO3de1iUdf7/8efMAANyRhQEUfGcecwDWXkoWdG2g2WldtDc1rayvpXbyX6ltbVrVtvawXLX7WBtlrWVHdcySrNCTc1KUzyfBUHlrBxm7t8fN6AkKoPAPQOvx3Xd19wzc8/Nm2liXn5Ot80wDAMRERERL2a3ugARERGR01FgEREREa+nwCIiIiJeT4FFREREvJ4Ci4iIiHg9BRYRERHxegosIiIi4vUUWERERMTrKbCIiIiI11NgEREREa/nV5sXzZ49m6eeeoqMjAx69erF888/z4ABA6o9du7cubz++uusW7cOgL59+/K3v/2tyvE33ngj8+bNq/K6lJQUFi1aVKN63G43+/btIzQ0FJvNVptfSURERBqYYRjk5+cTFxeH3X6aNhTDQ2+//bYREBBgvPLKK8b69euNSZMmGREREUZmZma1x1977bXG7NmzjR9//NHYsGGDceONNxrh4eHGnj17Ko+ZMGGCMWLECGP//v2V26FDh2pc0+7duw1AmzZt2rRp0+aD2+7du0/7XW8zDM8ufpiUlET//v154YUXALN1IyEhgTvuuIMHHnjgtK93uVxERkbywgsvMH78eMBsYcnJyWHhwoWelFIpNzeXiIgIdu/eTVhYWK3OISIiIg0rLy+PhIQEcnJyCA8PP+WxHnUJlZSUsHr1aqZOnVr5mN1uJzk5mbS0tBqdo6ioiNLSUqKioqo8vmTJElq2bElkZCQXXXQRjz/+OM2bN6/2HMXFxRQXF1fez8/PByAsLEyBRURExMfUZDiHR4Nus7OzcblcxMTEVHk8JiaGjIyMGp3j/vvvJy4ujuTk5MrHRowYweuvv05qaiozZ85k6dKljBw5EpfLVe05ZsyYQXh4eOWWkJDgya8hIiIiPqZWg25r64knnuDtt99myZIlBAYGVj4+duzYyv0ePXrQs2dPOnTowJIlSxg2bNgJ55k6dSpTpkypvF/RpCQiIiKNk0ctLNHR0TgcDjIzM6s8npmZSWxs7Clf+/TTT/PEE0/wxRdf0LNnz1Me2759e6Kjo9myZUu1zzudzsruH3UDiYiINH4eBZaAgAD69u1Lampq5WNut5vU1FQGDhx40tc9+eSTPPbYYyxatIh+/fqd9ufs2bOHgwcP0qpVK0/KExERkUbK44XjpkyZwty5c5k3bx4bNmzg1ltvpbCwkIkTJwIwfvz4KoNyZ86cycMPP8wrr7xCu3btyMjIICMjg4KCAgAKCgq49957Wb58OTt27CA1NZXLL7+cjh07kpKSUke/poiIiPgyj8ewjBkzhqysLKZNm0ZGRga9e/dm0aJFlQNxd+3aVWXxl5deeomSkhKuuuqqKueZPn06jzzyCA6Hg59//pl58+aRk5NDXFwcw4cP57HHHsPpdJ7hryciIiKNgcfrsHijvLw8wsPDyc3N1XgWERERH+HJ97euJSQiIiJeT4FFREREvJ4Ci4iIiHg9BRYRERHxegosIiIi4vUadGl+EWnkDAMKMuHwjvJtJ9hs0PY8aN0f/LRUgYjUjgKLiHimON8MIod3QM7OquEkZyeUHa3+dX5B0OZcSBwMiUOgVS9w6E+QiNSM/lqISPX2roGMX04MJkUHT/06mwPC4yGyHUS0hdIi2L4MCg/Atq/NDcAZDu3OPxZgWp5ltsaIiFRDgUVEqsrbB/+7HzZ8dPJjgqLMQBLZtvy2nRlOIttBeGtw+Fc93jAgKx22L4Xt38COZXA0F9I/MzeA4BbQbpAZYNoPgchEBRgRqaSVbkXE5HbBDy9D6l+gJN9sKWk/FKLanxhMAs/w/zO3C/b/ZIaX7d/ArjSzJeZ44QnHWl8SB0OYLoYq0th48v2twCIiZtfPx3fC3tXm/fh+cOksiO3RMD+/rAT2rjLDy7alsOcHcJcee95mh5S/wbm3Nkw9ItIgFFhEpGZKCmHJE5A2GwwXOMNg2DTo9wewO6yta9dyswtp2xKzNQbgqleg+2jr6hKROuXJ97fGsIg0VZu+gE//DLm7zPvdLocRM72j6yUgGDoOMzfDMMfUrPwnfHALhMSag3VFpEnRwnEiTU1+BrwzAeZfbYaV8AQYtwCued07wspv2WwwYgacdSm4SuDtcXBgg9VViUgDU2ARaSrcbvjh3/BCf/h1oTmoduDtcNty6DLC6upOze6AK+dCQpI5u+g/V0HefqurEpEGpMAi0hRkrodXhptdQMV5ENcHbv4aUv4KzhCrq6sZ/yAY9zY07wR5e+DNq+FontVViUgDUWARacxKimDxdPjnYHPmTUAIjHwS/phqrjTra5pFwfX/heCWkPkLvHODOcNIRBo9BRaRxmrzl/DiufDdLHCXQddLYPJKSPqTtTOAzlRkO7juHfAPNmcQffx/5sBcEWnUFFhEGpuCLPjvH+DN0eaS+mHxMHY+jH3TXDK/MYjrA9fMM8fh/PQWfPW41RWJSD1TYBFpTDZ+ZraqrHvPXGzt3Ntg8gro+nurK6t7nX4Hlz5r7i97Gla9Ym09IlKvtA6LSGNQnA+fPwhrXjfvtzwbLn8B4s+xtq76ds4NkLcXlswwBxSHxnn/jCcRqRW1sIj4ul0rYM4F5WHFBufdAZO+avxhpcKQ+6HP9WC44b8TYc9qqysSkXqgwCLiq1ylkPoYvDoCDu8wF4Cb8DEMfxz8A62uruHYbHDJLOiYbF5Acf41cHCr1VWJSB1TYBHxRVnp8O9kc+yG4YaeY+HW7yBxkNWVWcPhD1fPM6dqF2XDm1dBYbbVVYlIHVJgEfElbjes+Ke5rsr+tRAUCVe/Blf+EwLDra7OWs4QuPZdiGgDh7bB/DHmOjQi0igosIj4irx95lTl/90HZUehw0VwaxqcfYXVlXmP0Bi4/n0zyO1dBe/dBG6X1VWJSB1QYBHxBevehxcHwtavwC8QLn7a/GL2xosVWi26k7mEv18gpH8Gn92rheVEGgEFFhFvdiQH3r/ZnP1yNMdcMO1Py2DAJHOwqVSvzbnmxRKxwaqX4dt/WF2RiJwhBRYRb7V9Gbx0Pvy8wFwEbvB9cNNiaNHZ6sp8Q7fLYORMcz/1UfhpgbX1iMgZ0cJxIt6m9Ch89RikzQYMiEyEK/8FCQOsrsz3JP0JcnfD98/Dh5PNMS7th1pdlYjUglpYRLxJxi8w9yJIewEwoO+NcMu3CitnIvkv0H00uEthwXhzSriI+BwFFhFvUFwAXzwE/xwCB9ZDcAsYt8C8Vo4zxOrqfJvdDqNegjYDoTjXXFiu8KDVVYmIhxRYRKxkGLDhE5idZHZbGC446zJzurKuiVN3/Jww5k2IbGeuCrzgOigrtroqEfGAAouIVQ7vhLfGml+eeXsgoq258NmYNyCkhdXVNT7BzeHad8AZDrvS4OM7Nd1ZxIcosIg0tLISWPaM2aqyaRHY/WHQn+G25dB5uNXVNW4tusA1r4HNAT+9Bd8+Y3VFIlJDCiwiDWnHd/DPQeY027Ij0PYC8xpAw6ZBQDOrq2saOlwEFz9l7qf+BX790Np6RKRGNK1ZpCEUZsMXD8NP8837zaIh5a/Qc4wWgLNC/5sgezOseAne/5N5pev4c6yuSkROQS0sIvXJ7YbVr8HzfY+Flb4T4fYfoNdYhRUrpfwVOg03W7reGge5e62uSEROQYFFpL5krINXUszBnUdzIKYH3PQlXDoLmkVZXZ3YHTD6ZWjZDQoy4K0x5vRyEfFKCiwida04Hz7/f/DPwbBnJQSEQMrf4OYlkNDf6urkeIFh5oUSg1uYi/a9P0lXdxbxUgosInXFMODXj8zZP2kvmGuqdLscJq+EgZPBoSFjXimyLYydDw6neXXnLx+xuiIRqYYCi0hdOLgV5o+Bd26AvL3H1lS55nUIj7e6OjmdhAEw6kVz//vnYM3r1tYjIifQP/lEzkRxASz7u9mi4iox11Q5/05zXRVNU/YtPa4yZw4tfQI+udu86GTiIKurEpFyCiwitWEYsO49c6py/j7zsfYXwsgnoUVna2uT2hv6ABzcbP63XXA9TPoKmnewuioRQYFFxHMZv8Bn98Gu7837EW0gZQZ0/b2mKfs6mw0un21eNmHvKnjzavjjl5rVJeIFNIZFpKaKDsEnU8zZP7u+B78guPAhc1DtWZcorDQW/kHmINzwBDi0Fd4Zb15OQUQspcAicjpuF/zwb3j+HFj1MhhuOPsKc/G3IfeaX3DSuITGmNOdA0JgxzL4dIoulChiMQUWkVPZ+T38cwh8+mc4cthcZGzCJ3D1axCRYHV1Up9iu8NVr4DNDj++YQ6sFhHLKLCIVCdvH/z3Jnh1JGT+AoHhMPIp+NMyzRxpSjqnwPC/mvtfPAwbP7O2HpEmTIFF5HhlxeY05ef7wbr/AjboeyPcsQaSbtbib03Rubea13/CgPf+CPt/troikSZJf31FKqQvgkUPwOHt5v2EJHOaclxvS8sSi9lscPFT5udi2xJ4ayz86RsIjra6MpEmRS0s0rSVlcDWr83pq2+NMb+UQmLhin/BHz5XWBGTwx+ungfNO5krGX/0fxqEK9LA1MIiTU/hQdiyGNL/B1u/guI883G7v3nNn8H3gDPU2hrF+wRFmINw514E6Z/Cmnlmd6GINAgFFmn8DAOy0mHT/8xunz0rzanJFYJbQOcRcP5dEN3RsjLFB7TqCcMehsXTYNFUaHuBPjMiDUSBRRqnshLY+R1s+twMKod3VH0+poc5A6TLSIg7B+zqHZUaGngHbF5srs/y/iS46Quzy0hE6pUCizQeRYdg8xcndvUAOAIgcbDZktJ5hNZQkdqz2+GKOfDSebBvDSydCRc9ZHVVIo1erf5ZOXv2bNq1a0dgYCBJSUmsXLnypMfOnTuXQYMGERkZSWRkJMnJySccbxgG06ZNo1WrVgQFBZGcnMzmzZtrU5o0JaVHYe8a+PYf8MoIeKoDfPAn+HWhGVaCW0Cf62HMf+C+7XD9ezBgksKKnLnw1nDJP8z9ZX+HXcutrUekCfC4hWXBggVMmTKFOXPmkJSUxKxZs0hJSSE9PZ2WLVuecPySJUsYN24c5513HoGBgcycOZPhw4ezfv164uPjAXjyySd57rnnmDdvHomJiTz88MOkpKTw66+/EhgYeOa/pfi+woOQ8bN54cHMdeZtVjoYrqrHqatHGkr30bDpC/j5bbNr6JbvIDDM6qpEGi2bYXg2Ny8pKYn+/fvzwgvmMtVut5uEhATuuOMOHnjggdO+3uVyERkZyQsvvMD48eMxDIO4uDj+/Oc/c8899wCQm5tLTEwMr732GmPHjj3tOfPy8ggPDyc3N5ewMP3B8Glutzm1OOOXqlv+vuqPD4qC+L5mSFFXjzS0o7nw0gWQuwt6jTO7ikSkxjz5/vaohaWkpITVq1czderUysfsdjvJycmkpaXV6BxFRUWUlpYSFWVern379u1kZGSQnJxceUx4eDhJSUmkpaVVG1iKi4spLi6uvJ+Xl3fCMeIDSo/AgV+rBpPM9VBSUP3xUe0htkf51tO8DW2lqySLdQLD4cp/wWsXw09vQafh0P1Kq6sSaZQ8CizZ2dm4XC5iYmKqPB4TE8PGjRtrdI7777+fuLi4yoCSkZFReY7fnrPiud+aMWMGjz76qCelixVKj0DObsjZBTk7Ibdiv3wryKz+dX6B0PKsqsEk5mytjSLeqe1AuGAKLHsaPrnbXCE5PN7qqkQanQadJfTEE0/w9ttvs2TJkjMamzJ16lSmTJlSeT8vL4+EBHUFNLiSwmOBJHdX1TCSswsKs05/jmbNj4WSitvmHXXNHvEtQx+Aramw70dYeAvc8KHGT4nUMY++FaKjo3E4HGRmVv2XcWZmJrGxsad87dNPP80TTzzBl19+Sc+ePSsfr3hdZmYmrVq1qnLO3r17V3sup9OJ0+n0pHQ5GcOA4nw4mgNHcsw++cr98vvV7RcdNLfTCQiBiLbm2JKINse28PL7zZqrS0d8n8Mfrvw3/HMQbP8Gls+G8+6wuiqRRsWjwBIQEEDfvn1JTU1l1KhRgDnoNjU1ldtvv/2kr3vyySf561//yueff06/fv2qPJeYmEhsbCypqamVASUvL48VK1Zw6623evbbyInKiiF7M2RtNMeLHNhoXgulIpgcza266qunnOHHBZFqQklQpAKJNA3RHSHlb/DJXZD6F2g/1GwxFJE64XG7+5QpU5gwYQL9+vVjwIABzJo1i8LCQiZOnAjA+PHjiY+PZ8aMGQDMnDmTadOmMX/+fNq1a1c5LiUkJISQkBBsNht33XUXjz/+OJ06daqc1hwXF1cZiqQGykrg0FY4sMHcsjaY4eTQthOn/lbH4TSvlRIYYQ4kPN1+UKS5FkVQRL39SiI+p++N5YsXfgbvTYKbvwb/IKurEmkUPA4sY8aMISsri2nTppGRkUHv3r1ZtGhR5aDZXbt2YT+u7/all16ipKSEq666qsp5pk+fziOPPALAfffdR2FhITfffDM5OTlccMEFLFq0SGuwVMdVZoaQikBy4Fez9eTgFnCXVf+awHBocRa07GreRrY7LnhEmM/rj6rImbPZ4LLn4cVV5v+jXz4CI2daXZVIo+DxOizeqEmsw1JwABbcYC4F7iqp/piA0PJQ0tWcZdOiK7TsBqGx6pYRaUibvoD5V5v7178HHZNPfbxIE1Vv67CIhbZ+BbvLl//2b/abUFJ+G95awUTEG3QeDv0nwQ9zYeFtcGsaBDe3uioRn6bA4iuOHDZvu14C17yhKZMi3u53fzFnDGWnw8f/Z17TSv+gEKk1fev5iorAEhqrsCLiCwKawei5YPeHjZ/Aj29YXZGIT9M3n6+oCCxBkdbWISI116oXXPSQuf+/B+DgVmvrEfFhCiy+QoFFxDeddwe0vQBKC82rOrtKra5IxCcpsPgKBRYR32R3mFdxdobD3tXwzVNWVyTikxRYfIUCi4jvikiAS54x9795CnatsLYeER+kwOIrFFhEfFuPq6DHNealMN6fZF7DS0RqTIHFVyiwiPi+3z8N4W0gZyd89VerqxHxKQosvsDtNq+SDAosIr4sMBwu/Ye5v+plyNllbT0iPkSBxRcczQHKr6AQGGFhISJyxjoMg8TB5iU2ljxhdTUiPkOBxRdUdAcFhIBfgLW1iMiZsdlg2HRz/6e3zIuYishpKbD4AnUHiTQurfuZl9kw3PDVY1ZXI+ITFFh8QeWA2whLyxCROnTRw2Czm8v271lldTUiXk+BxRdohpBI49OyK/QaZ+5/+QgYhqXliHg7BRZfoMAi0jgNfQAcAbBjGWz72upqRLyaAosvUGARaZwi2kC/m8z9Lx9VK4vIKSiw+AIFFpHGa9CfzRmA+9fCrx9aXY2I11Jg8QUKLCKNV0gLGDjZ3P/qcXCVWVuPiJdSYPEFCiwijdvA2yEoCg5uhp/mW12NiFdSYPEFlYElyto6RKR+BIaZXUNgrn5besTaekS8kAKLL1ALi0jj1/+PEBYPeXvhh39bXY2I11Fg8QUKLCKNn3+gOc0ZYNkzcDTX2npEvIwCi7czDAUWkaai17XQvBMcOQTfv2B1NSJeRYHF2xXng+Ey97U0v0jj5vCDYQ+b+2mzoSDL2npEvIgCi7eraF3xCwL/IGtrEZH6d9ZlENcHSgth2dNWVyPiNRRYvJ26g0SaFpsNhk0391e9Aod3WluPiJdQYPF2CiwiTU+HCyFxMLhKzGnOIqLA4vUUWESapmGPmLc/vw0HNlhaiog3UGDxdpWBJcLSMkSkgbXuC10vAcNtLtkv0sQpsHg7tbCINF0XPQw2O2z8BHb/YHU1IpZSYPF2CiwiTVfLrubaLACpj5rrMok0UQos3u5IjnmrwCLSNA19ABwBsGMZbP3K6mpELKPA4u3UwiLStEUkmNcZAkj9C7jd1tYjYhEFFm+nwCIig/4MASGwfy1s+NDqakQsocDi7Y4cMm8VWESaruBoGHi7uf/V4+Aqs7YeEQsosHg7tbCICMDAyRAUBQe3wNo3ra5GpMEpsHgzXalZRCoEhsHge8z9pTOh9Ii19Yg0MAUWb1ZaZC7NDQosIgL9boKw1pC3F374t9XViDQoBRZvVtG6YveHgGBraxER6/kHmtOcAZb9HY7mWluPSANSYPFmx3cH2WzW1iIi3qHXOIjubP59+P55q6sRaTAKLN5M41dE5LccfnDRQ+b+t7Ngz2pLyxFpKAos3kyBRUSqc9ZlcNal4C6Fd8ZD4UGrKxKpdwos3kyBRUSqY7PB5S9C846Qtwfe+wO4XVZXJVKvFFi8mQKLiJxMYBhc8wb4N4NtS2DJDKsrEqlXCizerCKwNIuytg4R8U4x3eDS58z9b56C9EXW1iNSjxRYvFllC0uEpWWIiBfreTUMuNnc/+BmOLTd2npE6okCizdTl5CI1MTwv0Lr/ua6LO/coFVwpVFSYPFmR3LMWwUWETkVvwC4eh40aw4Zv8Bn91hdkUidU2DxZmphEZGaCo+Hq14Bmx1+/A+snmd1RSJ1SoHFmymwiIgn2g89tqjcZ/fCvh8tLUekLimweDMFFhHx1Pl3Q+eR4Co2F5UrOmR1RSJ1QoHFW5UeNa/WDAosIlJzdjtcMQci20HOLnj/ZnC7ra5K5IwpsHiroznmrc0BzjBLSxERHxMUYS4q5xcIWxaba7SI+DgFFm91/BosulKziHiqVU+45B/m/pIZsOVLa+sROUO1CiyzZ8+mXbt2BAYGkpSUxMqVK0967Pr16xk9ejTt2rXDZrMxa9asE4555JFHsNlsVbauXbvWprTGQ+NXRORM9b4W+t4IGPDeH80uIhEf5XFgWbBgAVOmTGH69OmsWbOGXr16kZKSwoEDB6o9vqioiPbt2/PEE08QGxt70vOeffbZ7N+/v3L79ttvPS2tcVFgEZG6MGImxPUx/6a8Mx7Kiq2uSKRWPA4szzzzDJMmTWLixIl069aNOXPm0KxZM1555ZVqj+/fvz9PPfUUY8eOxel0nvS8fn5+xMbGVm7R0dGelta4KLCISF3wD4RrXjf/luz7Ef53v9UVidSKR4GlpKSE1atXk5ycfOwEdjvJycmkpaWdUSGbN28mLi6O9u3bc91117FrVxNvulRgEZG6EtEGRv8bsMHqV2HtfKsrEvGYR4ElOzsbl8tFTExMlcdjYmLIyMiodRFJSUm89tprLFq0iJdeeont27czaNAg8vPzqz2+uLiYvLy8Klujo8AiInWpYzIMnWruf3K3uYS/iA/xillCI0eO5Oqrr6Znz56kpKTw2WefkZOTwzvvvFPt8TNmzCA8PLxyS0hIaOCKG0DFYk8KLCJSVwbfCx1/B2VHYcENx65XJuIDPAos0dHROBwOMjMzqzyemZl5ygG1noqIiKBz585s2bKl2uenTp1Kbm5u5bZ79+46+9leQy0sIlLX7Ha48l8Q3gYOb4cPbtGicuIzPAosAQEB9O3bl9TU1MrH3G43qampDBw4sM6KKigoYOvWrbRq1ara551OJ2FhYVW2RkeBRUTqQ7MoGPM6OJyw6X/w3T+srkikRjzuEpoyZQpz585l3rx5bNiwgVtvvZXCwkImTpwIwPjx45k6dWrl8SUlJaxdu5a1a9dSUlLC3r17Wbt2bZXWk3vuuYelS5eyY8cOvv/+e6644gocDgfjxo2rg1/RRymwiEh9iesDF5evfvvV47BtqbX1iNSAn6cvGDNmDFlZWUybNo2MjAx69+7NokWLKgfi7tq1C7v9WA7at28fffr0qbz/9NNP8/TTTzNkyBCWLFkCwJ49exg3bhwHDx6kRYsWXHDBBSxfvpwWLVqc4a/nwyr6lhVYRKQ+nDMedq+Etf8xF5W7ZRmE1l3XvkhdsxmGYVhdxJnKy8sjPDyc3NzcxtM99LfWUJIPd6yB5h2srkZEGqOSIvh3MhxYD20vgPEfgsPjf8eK1Jon399eMUtIfsNVaoYVUAuLiNSfgGZwzTwICIGd38LXf7W6IpGTUmDxRsdPNQwMt6wMEWkCojvBZc+b+98+A5u+sLYekZNQYPFGFQNuA8PB7rC2FhFp/LpfCQNuNvc/uBlyGuFSEeLzFFi8kWYIiUhDG/44xJ1j/v1590YoK7G6IpEqFFi8UWVgibK2DhFpOvyccPVrZsvu3lWweJrVFYlUocDijdTCIiJWiGwLV/zT3F/xEqxfaGk5IsdTYPFGCiwiYpUuI+H8O839D2+Hg1utrUeknAKLN1JgERErXTQN2pxnLq/wzgQoPWJ1RSIKLF5JgUVErOTwg6tehmbRkPkL/O8+qysSUWDxSgosImK1sDgY/W/ABmteh7VvWV2RNHEKLN5IgUVEvEGHC2Fo+cVsP50CBzZYW480aQos3kiBRUS8xeB7oP2FUFoE74yH4gKrK5ImSoHFGymwiIi3sDvMrqHQOMjeBJ/cBb5/zVzxQQos3kiBRUS8SXA0XPUK2Bzwy7uw6hWrK5ImSIHF27hdcDTX3FdgERFv0XYgJD9i7i96APattbIaaYIUWLzN0VygvLk1KMLKSkREqjrvDujye3CVmONZjr+yvEg9U2DxNhXdQQGh4PC3thYRkePZbDBqNkS0gZyd8OFkjWeRBqPA4m0q/sWi7iAR8UZBkXDN6+AIgI2fQNpsqyuSJkKBxdtUDriNsLQMEZGTiusDI2aY+19Oh10rrK1HmgQFFm+jGUIi4gv63QTdR4O7DN69EQqzra5IGjkFFm9z5JB5q8AiIt7MZoNLn4XmnSB/H3z+oNUVSSOnwOJt1MIiIr7CGQpX/svc/3mBpjpLvVJg8TYKLCLiS+LPgR7XmPuLH9asIak3CizeRoFFRHzNRQ+Zs4a2fwObF1tdjTRSCizeRoFFRHxNZFtI+pO5v3iauWK3SB1TYPE2Ciwi4osG/RkCIyBrA6x90+pqpBFSYPE2Ciwi4ouCImHwveb+V3+FkkJr65FGR4HF2yiwiIivGjAJItpCQQakvWh1NdLIKLB4E7f7WGBpFmVtLSIinvJzwrBp5v53s6DggKXlSOOiwOJNSvLBcJv7gRGWliIiUitnX2ku3V9SAEuesLoaaUQUWLxJReuKfzPwD7S2FhGR2rDbYfjj5v7q1yBrk6XlSOOhwOJNNH5FRBqDdhdA55FguCD1UaurkUZCgcWbKLCISGPxu0fBZoeNn8DO762uRhoBBRZvosAiIo1Fiy5wznhz/wst2S9nToHFm1QGlghLyxARqRNDHwT/YNi7Cn5daHU14uMUWLyJWlhEpDEJjYHz/8/c//JRKCuxth7xaQos3uRIjnmrwCIijcXA2yEkBg5vh1UvW12N+DAFFm+iFhYRaWycITB0qrm/dOaxf5iJeMjP6gLkOAosIo2e222QXViM2w02G9gAbGDDVnnfZrOV35qPY6Pa5wL9HNjtNit/nZrpcwMsfwmy0+Hbf5gziEQ8pMDiTRRYRBoNl9tgz+EiNmcWsPlAAZsP5LPlQAFbDhRQVOKqk5/hZ7cRFRxAdIiT5iEBtAhxEh3qJDokgObBx/ZbhDiJCg7Az2FRo7rDD373F3hrjBlc+v8RIhKsqUV8lgKLN1FgEfE5ZS43uw4Vsbk8jGzKzGdzZgFbswooLnNX+xqbDRw2GwZgGEb5bS1+ttvgQH4xB/KLa3R8ZDN/okOc5hbqpHlwAHERgXSNDePsuDCahzg9L6KmOqdA2wtg57fw1eNw5T/r72dJo6TA4k0UWES8lmEY7D50hHX7cstbTcwWk21ZhZS4qg8mAX52OrQIoVPLEDrHhNCxZSidYkJoG9XspK0dhmFgGJwQZgzKHy/fdxuQf7SU7PwSsguLyc4vJrughOyCYg4WHNvPLijmUGEJbgMOF5VyuKiUzQcKqv3ZMWFOurUK4+y4cLrFhdGtVRhtoprVTbeTzQbDH4O5F8LPC2DgbdCq15mfV5oMBRZvYRgKLCJepMzlZmNGPj/sOMSqHYf5Ycehk7ZkBPrb6dgyhE7lgaRTy1A6tQwhIaoZDg+/7G02cyxL+b1THhvi9KNVeNBpz+lyGxwuqggz5m1WecDZfaiIX/fnsT27kMy8YjLzsvg6PavKzzirVSjdWoWVh5hwOseG4PRzePR7ARB/DnS/Ctb9FxZPgxsWctwvK3JKCizeorQIXOVrFCiwiDS4opIy1u7K4Ycdh1m18xBrdh6m8DdjTfwdNs5qFUbnGDOQVIST+Iggrx786rDbKruCTqaguIz0jDzW78vj1315/Lo/j40Z+RQUl/HDjsP8sONw5bF+dhsdW4ZUhpiz48Lp3y6yZmNkhj0MGz6CbUtgSyp0Sq6D31CaAgUWb1HRuuIIMK/WLCL1Kiu/mNU7D5kBZcch1u3Lw+WuOpAk1OnHOW0j6d8ukn7toujVOoKggFq0LPiAEKcffdtG0bdtVOVjZS4327ILWb8vtzLErN+XR05RKRsz8tmYkc/7P+4FoEOLYP48vAsju8diO1WrSWQ7GHAzpL0Aix+GDheCvXG+p1K3FFi8RdEh8zYoUk2kIvVg96Ei0rYeNLt4dh5me3bhCce0Cg+kf7so+reLpG/bKLrEhnrcpdOY+DnsdI4JpXNMKFf0MR8zDIOMvKOs32sGmF/35ZG27SBbswq57c019Gwdzr0pXbigY/TJg8ugP8OPb8CBX+Gnt6DP9Q33S4nPUmDxFhq/IlLnDMNg+bZDzF22ja82HqjynM0GXWJC6dcukv7toujXLor4iNOPB2nqbDYbrcKDaBUeRHK3GMAc/Dt32XZeXraNn/fkcsPLKxnYvjn3jehCnzbV/E1rFgWD7jFbWL56HM6+EgLUsiynpsDiLRRYROpMmcvNZ+symPvNNn7ZmwuYAaVf28jyFpQozmkTSXgzf4srbRxCA/2Z8rvOjB/Yltlfb+HN5btI23aQK178nuHdYrgnpQudY0KrvmjAzbByLuTuguWzYfC91hQvPkOBxVsosIicsYLiMt5euYtXv9vB3pwjgDmD56q+rbnpgvYkRgdbXGHjFh3iZPqlZ3PTBYk8++Vm3luzhy9+zWTxhkyu6BPP3cmdSYgqb0nxD4Rh0+D9P8K3z8I5N0JIC0vrF++mwOItFFhEam1/7hFe+24H81fsIr+4DIDmwQFMOK8d15/blqjgAIsrbFpaRzbjqat78ach7Xn6800sWp/B+2v28vFP+7guqS2TL+xIi1AndB9tDr7dv9a8ztDvn7a6dPFiCizeQoFFxGPr9+Xy72Xb+finfZSVz/Dp0CKYSYPaM6pPPIH+mn1ipY4tQ5lzQ19+2p3DU5+n8+2WbF77fgfvrNrNH85P5OYh7Qkb/hjMuxRWvwpJt0B0R6vLFi+lwOItKgNLhKVliHg7wzBYuimLfy/bzrdbsisfP7d9FJMGtefCLi29ek2UpqhXQgT/+WMS323J5slFG/lpTy4vfL2FN5bv5LahHfhjx+E4tnxhDsIdO18zJaVaCizeQi0sIqdUXObio7X7+Pey7aRn5gPmgmgX92jFpEGJ9GwdYW2Bclrnd4xm4eTz+Xx9Jk9/kc6WAwXM+N9Gvg4ZyXxbKvb0z+Cnt6H3OKtLFS+kwOItjuSYt0FRpzxMpKk5kH+Ud1ft4bXvd5BVvjR+cICDsQPaMPH8drSO1HRYX2Kz2RjRPZbfdYvh/TV7mPXlZpbnxPB3x2ju9X8H1yd/xtHmXIhKtLpU8TIKLN5CLSwilQqKy/h8XQYL1+7luy3ZVCxAGxsWyI3nt2PcgDaEB2lKsi9z2G1c3S+By3rHMX/FLl5MdTCk7CcGlKWzfe51BN/yBS3DQ6wuU7yIAou3UGCRJq6kzM3STVksXLuXL3/NpLjs2BWQeydEMH5gWy7pGUeAXw2uVyM+w+nnYOL5iVx5TmvmfuxH1/XXk3hkPbOfuYuQlIe4/ty2TXq1YTmmVv/nz549m3bt2hEYGEhSUhIrV6486bHr169n9OjRtGvXDpvNxqxZs874nI2SAos0QW63wcrth3jwg18Y8LcvmfT6Kj79eT/FZW7aRwdzd3JnltwzlIWTz+fKc1orrDRi4UH+3HNNMjkXzQTgFv7Lhx9/wOWzv2Xt7hxrixOv4HELy4IFC5gyZQpz5swhKSmJWbNmkZKSQnp6Oi1btjzh+KKiItq3b8/VV1/N3XffXSfnbHRKj0CZuciVAos0BekZ+Sxcu5eP1u6rXOANoEWok8t6xTGqdzzd48NOfRE9aZTaDJmAO/tbHL+8w3POFxmx929c8WIe1w5ow30pXbU6cRNmMwzDOP1hxyQlJdG/f39eeOEFANxuNwkJCdxxxx088MADp3xtu3btuOuuu7jrrrvq7JwAeXl5hIeHk5ubS1hYmCe/jnfI2w/PdAWbA6Yd1JQ+aZT25Rzho5/2sfDHvWzMyK98PMTpx4jusYzqHc/ADs3V/C9wNBfmXAA5u/ghPIWrMycA5mKAUy8+i9HnxCvMNhKefH971MJSUlLC6tWrmTp1auVjdrud5ORk0tLSalVsbc5ZXFxMcXFx5f28vLxa/WyvcXx3kP4nlEbkcGEJ/ysfPLty+6HKx/0dNi7s0pJRfeK5qGtLLfAmVQWGw5Vz4dWR9M/9nMW/u5jbfkpk84EC7nn3J975YTePjepOl9jQ059LGg2PAkt2djYul4uYmJgqj8fExLBx48ZaFVCbc86YMYNHH320Vj/PK2n8ijQSLrfBz3tyWJKexdJNWfy0J4fj23CTEqMY1Seei7u3UtO+nFqbc80LIi6dSaeV0/j05mW8sq6MZ7/czModh/j9c8u46YJE/m9YJ4Kdmj/SFPjkf+WpU6cyZcqUyvt5eXkkJCRYWNEZUmARH3Yg/yjfbMpm6aYslm3OIqeotMrz3VqFcVnvOC7rFUdcRJBFVYpPGnwfbP0K9vxAwEe3csuEj7m0Vxx/+Xg9n6/P5J/fbOOjn/Yx/dJupJwdq26iRs6jwBIdHY3D4SAzM7PK45mZmcTGxtaqgNqc0+l04nQ6a/XzvJICi/iQUpebNTsPs3ST2Yqyfl/VLtnQQD8Gd2rBkM4tGNy5BbHhgRZVKj7P4QdX/gvmDIKd38F3s4gf9Gf+eUM/vtqYyfSP1rP70BFu+c8ahnZpwaOXnU3b5roid2PlUWAJCAigb9++pKamMmrUKMAcIJuamsrtt99eqwLq45w+R4FFvNy+nCNmQEnP4rst2ZVXRK7Qs3U4QzqbIaV3QgR+Dk0/ljoS1R5GPgkf3gZf/w3aD4X4vlzUNYbzOkTz4tdbmLN0G0vSs/jdP77h9gs7cuvQDvjrM9joeNwlNGXKFCZMmEC/fv0YMGAAs2bNorCwkIkTJwIwfvx44uPjmTFjBmAOqv31118r9/fu3cvatWsJCQmhY8eONTpno6fAIl7maKmLVTsOs3TTAZakZ7H5QEGV56OCAxjcKZohXVowqFMLokMaUYuneJ/e18LmL+DXhfDeJPjTN+AMIdDfwZThXbi8TzzTP1zPt1uyeWbxJr7ZlMXz1/ahVbi6IBsTjwPLmDFjyMrKYtq0aWRkZNC7d28WLVpUOWh2165d2O3Hku2+ffvo06dP5f2nn36ap59+miFDhrBkyZIanbPRU2ARixmGQXpmPt9uzuabzdms2Hawykqzdhv0aRNZ2YrSIz5cV0SWhmOzwSX/gD0/wKGt8PlUuOz5yqc7tAjhjZsG8NFP+3jog3Ws2nmY3z/3Lc9c04uhXZrAWl5NhMfrsHgjn1+H5Z0J5r8cRj4FSTdbXY00EVn5xXy3JZtvNmfx7eZsDuQXV3k+JszJoE4tGNqlBRd0jCaiWYBFlYqU2/4NzLsMMOCaN6DbZSccsvNgIZPnr2HdXnNs1eQLO3B3cmd1U3qpeluHReqJWlikARwtdbF652G+2ZzFsk3Z/Lq/6mDZQH87SYnNGdQpmsGdW9CpZYhmXYh3SRwM598J382Cj/8PWveDsLgqh7RtHsx7t57HXz/dwOtpO5n99VZ+2HGY58f1ISZMA8B9mQKLN1BgkXpgGAabMgtYtjmLZZuzWbH9IEdL3VWOOTsujAs6RTO4Uwv6to3UAm7i/S78f7Dta9j/Eyy8Fa7/AOxVW0+cfg7+cnl3BiRG8cB7v7By+yEufnYZ/xjTm8GdW1hUuJwpBRZvcCTHvFVgkVpyuQ32Hj7C1uwCth4o4Nf9edV287QMNbt5BneO5vyO0RosK77HLwBGv2xOdd62BJa/COdVP6P0kp5xnB0XzuQ31/Dr/jwmvLqS2y/syF3JnXUJCB+kMSze4G+toSQf7lgDzTtYXY14sdyiUrZmF7Atq5BtWeW32QXsOFhESZn7hOOdfnaS2jdncKdoBnVqQecYdfNII7HqVfjkLnAEwB9ToVXPkx56tNTFY5/8ypsrdgFwbvsonhvbh5bqIrKcJ9/fCixWKyuBx8ubKO/bDs2irK1HLFfqcrP7UFFlGNl6wLzdllXIwcKSk74uwM9OYvNg2rcIpkOLEM5t35x+7dTNI42UYcDb10H6pxDdBW5eAgHNTvmSD9fu5cH3f6GwxEV0SACzxvThgk7RDVOvVEuDbn3J0ZzyHZt5wS9pdI6WujhUWMLhohIOF5ZyqKiEnKIS87HCEg4XlZrPlT+fmXeUMvfJ/x0RE+akfXQI7VsE075FCB3KA0pcRJCauaXpsNnMqc0vrYLsdFg8DX7/9ClfcnnveHrEh3Pbm2vYmJHPDa+s4I6LOnHnsE76f8cHqIXFalnpMHsABEbAAzutrkZ+w+02KCwpo7DYRUFxKQXFLgqLy8g/WkZhcRkFx22FxWUcLio9IYwcKXV5/HMD/e0kloeSDuWhpH10CIktggnRhd5EjtmSCv+50twftwC6jDjtS46Wunj04195a6XZRTSwfXOeHdeblqHqImpoamHxJY14hpBhGJS6DEpdbkpdbkpcbsqOv19m4HIbuI3jNzMkuAwDwwC3YR5jGBx3LFWOd7ndlLoMylwGZW535e3xj5n7bsrcx44pPe74EpebwvLQkV9+W3C0jMISz8NGdfzsNiKDA4hs5k9kswCiggOIaBZAVLB5/9hj/rQMC6RVWKAWZhOpiY7D4NzbzMG3H06G29Ig5NSLxQX6O5hxZQ+SEqN48INfSNt2kIuf/ZbnxvbmvI7qIvJWCixWa+DAYhgGxWVuikpcFJWUld+6KCo29wtLyjhS4qKwxMWRkrLyW7NVoaj02HFHS12UHBc+SsvcVe+7zJDQWDjsNkKcfpVbsNNBSKA/IU5H+X3z8fAgf6KCA8rDSQBRzQKIDPYnxOmnwa4i9WXYdNi2FA6sh4W3wXXvml1GpzGqTzzd481ZROmZ+Vz38gruHNaJOy5SF5E3UmCxWgMElkOFJSzbnMU3m7JZtjnrhKmuDSnAz06Aw46/w4afw46f3YbdZsNux7y12bDbjtu3H3+f8vtVj3HYbfg5bPjZzfM67Db8y8/tV/Gz7PbyY8ofO+65iuOPhRE/QgOr7jv97AocIt7KPxBG/xv+NRS2LDbHswx/rEYv7dgyhIWTz+eRj9azYNVuZn25mR92HOK5sX1ormn/XkWBxWr1EFhKXW7WlK9o+s2mbNbty6W6kUpOPzvBTj+C/B0EOx0EBfgRHOCgWYCDZgF+NAtwEBTgIDjAr/y2/HGngyB/BwF+dvwd5hbgsOPvZzu27zADwrH7ZjDQl76I1IuYbnDpLHMxue+fg+Boc1XcGggKcDDzqp4MSIzioYXr+G7LQa5/eSVv33wu4UH+9Vu31JgCi9UqAssZTmfedbCIpZuz+GZTFmlbD1JQXFbl+a6xoQzp3ILBnVvQo3U4wQF+avIUkcal97VQmGW2sCyeBs2ioc91NX756L6t6dE6nGvnrmDD/jwmzVvF6zcN0NIAXkKBxWq1bGEpKC4jbetBvtmUxTebs9h5sKjK81HBAeY1YTq1YFCnaC2QJCJNw/l3QmG22cry0R3m39auF9f45Z1jQnn9DwMY8680Vu44xO3z1zDn+r66eKIXUGCxmgeBJT0jny83ZPLNpizW7DpcZVCrn93GOW0jzVaUTi04Oy5Ms0xEpGn63V+g6CCsfRP+OxGufx/anV/jl3eLC+PlCf254eUVfLnhAPe/9wtPXdVTf1MtpsBitRoGltU7DzH6pbQqj7Vt3ozBncxunoEdmmt9DhERMGcIXfqc+fc1/TN4axxM/BRie9T4FAMSo5h97Tn86T+reW/NHqKC/Xnw4rM0Ds9C+oazWg0Dy4rthwDo0CKYG89rx+DOLWjbPLi+qxMR8U0OP7jqFXjjStj1PfxnNPzhc4hKrPEpkrvFMHN0T+559yfmLttOVLCTW4fqem9WUaec1WoYWDZl5ANw5TmtuWFgO4UVEZHT8Q+CcW9BTHcoyIQ3roD8TI9OcVXf1vy/i88CYOaijSz4YVd9VCo1oMBitZoGlswCADq1DKnvikREGo+gCLj+PYhoC4e3w5uj4WiuR6eYNLg9twwxW1amvv8Li9Zl1EOhcjoKLFZyu479j3OKwOJyG2zJMgNL55jQhqhMRKTxCI2FGz6A4JaQ8Qu8dS2UHvXoFPeP6MKYfgm4Dfi/t3/k+63Z9VSsnIwCi5WOT/mBESc9bOfBQkrK3AT620mIOvXl00VEpBrNO8D1/wVnGOz8Ft67CVxlp39dOZvNxl+v6M7wbjGUlLm5+fXVrNvrWUuNnBkFFitVdAc5w8wBYidR0R3UsWWIFnsTEamtVr3MMS0OJ2z8BD65i2qXAT8JP4ed58b14dz2URQUlzHhlZVsK2/9lvqnwGKlyvErEac8bFOmOeBW3UEiImeo3QXm7CGbHX58A1If9ejlgf4O5o7vR/f4MA4WlnDDyyvJyPWse0lqR4HFSjUecKvAIiJSZ866BC591tz/9h/w/QsevTw00J/XJg4gMTqYvTlHGP/KCnKKSuqhUDmeAouVahhYNmdWDLjVDCERkTpxznhIfsTc/+L/wdq3PHp5dIiT1/8wgJgwJ5syC/jDaz9QVFLzMTHiOQUWK9UgsJS63GzL1gwhEZE6d/5dMPB2c//DyZC+yKOXJ0Q14/U/JBEe5M+aXTnc9uYaSl3uuq9TAAUWa9UgsOzILqTUZRAc4CA+IqiBChMRaQJsNvjdY9BzLBgueHcC7Ew7/euO0yU2lFdu7Eegv50l6Vnc8+5PuN01H8grNafAYqUaBJbKGUIxobqGhYhIXbPb4fIXoFMKlB2Ft8ZA5nqPTtG3bRQvXd8XP7uND9fu4y+f/IrhwewjqRkFFivVILCklw+47aLxKyIi9cPhD1e/BgnnmutjvXElHNru0Sku7NKSp6/uBcBr3+/g+a+21EOhTZsCi5VqEFg2a4aQiEj9C2gG174NLbtBQQa89nvI3uzRKUb1iWf6pd0AeGbxJv69bJtaWuqQAouViswrMJ+6S8gMLJ0UWERE6ldQpLmEf3QXyNsLr46EjHUenWLi+Yn830UdAXj80w3c8PJKdh4srI9qmxwFFiudpoWluMzFjoNFAHRRYBERqX+hsTDxM4jtAYVZZkvL3tUeneLu33Vm6siuOP3sfLslm+H/+IaXlmzVDKIzpMBipdMElm1ZhbjcBqGBfsSEORuwMBGRJiw4GiZ8Aq37w9EcmHc57Piuxi+32Wz8aUgHPr9rMOd3bE5xmZuZizZy6fPfsnZ3Tr2V3dgpsFjF7Tb/R4CTBpbjV7jVDCERkQYUFGF2D7UbBCX58J/RsCXVo1O0iw7mPzcl8ferexHZzJ+NGflc8eJ3PPrxegqKtcicpxRYrFKcB0Z58+BJrtSsJflFRCzkDIXr3oVOw6HsCLw1FjZ84tEpbDYbo/u25sspQ7iyTzyGAa9+t4PhzywldUNmPRXeOCmwWKWiO8i/GfgHVnvIJi3JLyJiLf8gGPMmnHUZuErgnfHw87sen6Z5iJNnxvTm9T8MICEqiH25R7lp3iomv7mGA3m6eGJNKLBYpXL8StRJD9lUuQaLWlhERCzjFwBXvQq9xpkr4r4/CVbPq9WpBnduwRd3DeFPQ9rjsNv49Jf9DHtmKfNX7NIKuaehwGKV0wy4PVLiYtchc4aQpjSLiFjM4QeXvwj9bgIM+Pj/IO3FWp0qKMDB1JFn8dHt59OzdTj5R8t48INfGPOvNLYcyK/buhsRBRarVAaWiGqf3ppVgGFAZDN/okMCGq4uERGpnt0Ov/87nHeHef/zqfDNU1DLxeHOjgvng9vOZ9ol3WgW4OCHHYe5+NlvmfXlJorLXHVYeOOgwGKV07SwpGdohpCIiNepuGDi0AfN+189DqmP1jq0OOw2/nBBIounDOGiri0pcbmZ9eVmLn52GSu3H6rDwn2fAotVjuSYtyeb0nxAM4RERLySzQZD74fhj5v3v/0H/O8+c7mKWoqPCOLlCf144do+RIc42ZpVyDX/TOOB937WSrnlFFiscpoWls2aISQi4t3OuwMu+Qdgg5X/go9uB3ftu3JsNhuX9IwjdcoQxg1IAODtH3Yz9OklTHx1JV+nH2jSA3P9rC6gyfKgS0hERLxUvz+Yy1MsvBXWvgmlRXDlXPMK0LUU3syfGVf25Io+rXlxyRaWpGfxdfnWtnkzbji3LVf3TSC8We1/hi9SYLHKKQJLYXEZe3OOAAosIiJer9dYM7T89w+w/gMoPQpXv3bSNbZqakBiFAMSB7A9u5D/LN/JO6t2s/NgEY9/uoG/f7GJUX3iGT+wLWe1Cqub38PLqUvIKqcILJsPmN1B0SFOIoM1Q0hExOt1uwzGvQ1+gbDpfzD/GiguqJNTJ0YH8/Al3Vjx4DD+dkUPusaGcqTUxVsrdzHy2WVcMyeNT37e1+gvrqjAYpVTBJZN5d1BXWI1fkVExGd0Sobr34OAENi+FP45CJb9HXL31MnpmwX4cW1SG/535yAW3Hwuv+/RCofdxsodh7h9/o9cMPMrnv1yMwfyG+fKuQosVjlVYClf4bZTS3UHiYj4lHYXwPgPzVXMD22D1L/AP7rDa5fAj29C8ZkvDGez2Uhq35zZ153Dd/dfxP8N60R0iJPMvGL+8eUmzn/iK/7vrR9ZvfMQRi2nW3sjm9EIfpu8vDzCw8PJzc0lLMwH+vIMAx5rAe5SuPtXCI+v8vT4V1byzaYs/nZFD65NamNRkSIiUmtH8+DXD+Gnt2Hnt8ce9wuCsy4xx70kDjVX0K0DJWVu/rduP6+n7WT1zsOVj58dF8b4gW25tFcczQK8b9iqJ9/fCixWKC6AGeUh5cH9ENCsytPn/i2VjLyjvHfrQPq2Pfm1hkRExAcc3gm/vAM/LYCDm489HhIDPa42w0tsjzr7cev25vJ62g4+XLuP4jJzXEuzAAcpZ8dyRZ94zuvQHD+Hd3SwKLB4u5zdMKs7OJzwUKa5CFG53COl9Hr0CwB+mj6c8KCmNW1NRKTRMgzYuwZ+fht++S8cOW4l25ju0HOMGWDCWtXJjztcWMI7q3Yzf+Uudh4sqny8RaiTy3rFcUWfeM6OC7N0NXUFFm+3/2dzMFZILNyTXuWp1TsPMfqlNFqFB5I2dZhFBYqISL0qK4EtX8JPb8GmReAqMR+32aH9UPPK0F1/DwHBZ/yjDMNgza4cFv64l49/3kdOUWnlc51ahnDFOfFc3jue+IigM/5ZnvLk+9v7OrSaglMMuE3PMKfB6QrNIiKNmF8AdL3Y3I4cNtdv+WkB7F4OW78yt4AQ6DwCWvczu4xie0BguMc/ymaz0bdtJH3bRvLwJd1YuimLD37cw5cbDrD5QAFPLkrnyUXpnNs+iiv6xDOyRyvCAr2vdV+BxQo1mCHUuaWmNIuINAlBkeaKuf3+YM4s+vkds+Xl8A5Y919zqxDRFlr1hNiKrQeExVUZWnAqAX52ftctht91iyH3SCmL1u3n/TV7WbH9EMu3mdvDH67nd2fFMKpPPEM6tyDAzzvGuyiwWKEmgSVWLSwiIk1OVHsY+gAMuR92rzRbWjJ+gYyfIXc35Ow0tw0fH3tNs+bHWmBie5m30Z3A7jjljwoP8mdM/zaM6d+GvTlHWPjjXj74cS9bDhTw6S/7+fSX/UQ28+eSnnFccU48fRIiLB3vosBihVMGloqLHiqwiIg0WTYbtEkytwpFh8rDS3mAyfgFstKh6CBsW2JuFfyCIKbbsVaY5h0hKhHC4qsNMvERQUy+sCO3De3A+n15fPDjXj5cu4/sgmLeWL6TN5bvpG3zZiy4eSCx4Wd2yYHaUmCxQsXI8KCIKg8fKiwhu6AYMAdCiYiIVGoWBe2HmFuF0iNwYEPVEJOxDkoLYe9qczue3R8i2kBkOzPARCYe249oi80ZQvf4cLrHhzN1ZFe+23qQhT/uZdG6DNyGQUyYsyF/4yoUWKxwkhaWiu6g1pFBBDv1n0ZERE7DPwjizzG3Cm4XHNoOGT8dCzCHt5vrwbhL4dBWc9tazfmCW1YGGL/IdgyJTGTIwET+mtydHUdDfa9LaPbs2Tz11FNkZGTQq1cvnn/+eQYMGHDS4999910efvhhduzYQadOnZg5cyYXX3xx5fM33ngj8+bNq/KalJQUFi1aVJvyvN+RHPP2N4Flc8X4FXUHiYhIbdkdEN3R3LqPPva42wV5+8rDyw4z1Bzecez+kcNQeMDc9qyscspmQDf/ZnDv1hMWO20oHgeWBQsWMGXKFObMmUNSUhKzZs0iJSWF9PR0WrZsecLx33//PePGjWPGjBlccsklzJ8/n1GjRrFmzRq6d+9eedyIESN49dVXK+87ndY1O9W7k7awVExpVneQiIjUMbsDIhLMLXHwic8fyakaYI4PNLl7zDVhLAorUIvA8swzzzBp0iQmTpwIwJw5c/j000955ZVXeOCBB044/tlnn2XEiBHce++9ADz22GMsXryYF154gTlz5lQe53Q6iY2Nre3v4VtOEljSy1tYuqiFRUREGlpQBAT1hrjeJz7nKoWCAw1cUFUeTa4uKSlh9erVJCcnHzuB3U5ycjJpaWnVviYtLa3K8WB29/z2+CVLltCyZUu6dOnCrbfeysGDB09aR3FxMXl5eVU2n1JNYDEMQ11CIiLinRz+J1yot6F5FFiys7NxuVzExMRUeTwmJoaMjIxqX5ORkXHa40eMGMHrr79OamoqM2fOZOnSpYwcORKXy1XtOWfMmEF4eHjllpCQ4MmvYb2KwNLs2IUNswtKOFxUis0GHVqoS0hEROR4XjEVZezYsZX7PXr0oGfPnnTo0IElS5YwbNiJ19OZOnUqU6ZMqbyfl5fnO6Gl9AiUHTX3j2thqZgh1DaqGUEBp17sR0REpKnxqIUlOjoah8NBZmZmlcczMzNPOv4kNjbWo+MB2rdvT3R0NFu2bKn2eafTSVhYWJXNZ1S0rtj9zOtElKsILLqGkIiIyIk8CiwBAQH07duX1NTUysfcbjepqakMHDiw2tcMHDiwyvEAixcvPunxAHv27OHgwYO0alU3l9j2KsePXzluPnvFDCENuBURETmRx1c0mjJlCnPnzmXevHls2LCBW2+9lcLCwspZQ+PHj2fq1KmVx995550sWrSIv//972zcuJFHHnmEVatWcfvttwNQUFDAvffey/Lly9mxYwepqalcfvnldOzYkZSUlDr6Nb3IaRaN05RmERGRE3k8hmXMmDFkZWUxbdo0MjIy6N27N4sWLaocWLtr1y7s9mM56LzzzmP+/Pk89NBDPPjgg3Tq1ImFCxdWrsHicDj4+eefmTdvHjk5OcTFxTF8+HAee+yxxrkWy0lmCG3SDCEREZGTshmGYVhdxJnKy8sjPDyc3Nxc7x/PsuZ1+OgO6DwCrl0AwP7cIwyc8RUOu41f/5KC00+DbkVEpPHz5Pvb4y4hOUPVtLBUjF9p17yZwoqIiEg1FFgaWjWBRQvGiYiInJoCS0OrJrCkZyiwiIiInIoCS0OrrkvogNklpMAiIiJSPQWWhvabwOJ2G2yp7BLSlGYREZHqKLA0tMrAEgHA3pwjFJa48HfYaBcdbF1dIiIiXkyBpaEdyTFvy1tYNh8wW1faR4fg79B/DhERkeroG7Kh/aZLqGJKs1a4FREROTkFloZUVgIlZkCpDCzlM4R0DSEREZGTU2BpSEdzynds4AwHYNMBXaVZRETkdBRYGtLxA27tdlxugy2VU5rVJSQiInIyCiwN6TfjV3YfKuJoqZsAPzttm2uGkIiIyMkosDSkEwbcmt1BHVuE4LDbrKpKRETE6ymwNKSiQ+btbwJLl1iNXxERETkVBZaGpCnNIiIitaLA0pBO0iXUuaVaWERERE5FgaUhHRdYylxutmUVAuoSEhEROR0FloZ0XGDZcbCIEpebIH8H8RFB1tYlIiLi5RRYGlJlYIlic2bFgnEh2DVDSERE5JQUWBrScS0s6RXjV7TCrYiIyGkpsDSk4wLL5kytcCsiIlJTCiwN6UiOeRsUWTlDSNcQEhEROT0FlobiKoPiXABKAsLZnl0+Q0iBRURE5LQUWBrK0dzK3e0FfpS5DUKdfrQKD7SwKBEREd+gwNJQKsavOMPZlHUEgI4xIdhsmiEkIiJyOgosDaVywG3EsWsIqTtIRESkRhRYGspxM4Q04FZERMQzCiwNpZopzWphERERqRkFloZSHlhcgRHsOGjOENIaLCIiIjWjwNJQygNLLiG4DQgP8qdFqNPiokRERHyDAktDKQ8sB8rMCx12iQnVDCEREZEaUmBpKOWBZV+xGVg6qTtIRESkxhRYGkp5YNlRGADooociIiKeUGBpKOWBZVOeH6DAIiIi4gkFloZSHli25PsDmiEkIiLiCQWWhlIeWHIIoXlwAM1DNENIRESkphRYGoLbDUdzAMg1QtQdJCIi4iEFloZQnAeGG4BcgtUdJCIi4iEFloZw5BAAR22BlOCvawiJiIh4SIGlIVSMXzHMlpUusQosIiIinlBgaQjlgeWQOxiAzi0VWERERDyhwNIQjuQAkGMEExPmJLyZv7X1iIiI+BgFloZQ3sJyGM0QEhERqQ0FloZQcaVmI4RO6g4SERHxmAJLQzhu0bgusZrSLCIi4ikFloZQOUsoWFOaRUREakGBpb4dycG9+UsA9hnRdGqpFhYRERFPKbDUtyVPYC/KYos7jl9CBxEaqBlCIiIinlJgqU+Z62HlvwCYXjaBxNhIiwsSERHxTQos9cUw4LN7wXDxS/hQvnP30JRmERGRWlJgqS/r3oOd31FmD+SWA1cCcE4btbCIiIjUhgJLfSjOhy8eAmBWyWXsNaK54dy2pJwdY3FhIiIivkmBpT588xTk72eHEcPcsou5LqkNf7n8bGw2m9WViYiI+CQFlrqWtQn397MBeKR0PFcO6Mhjl3dXWBERETkDCix1yTDIevdO7EYZi13nENv3Mv46qjt2u8KKiIjImVBgqUOrF71OiwPfU2z4s7rrvfztih4KKyIiInWgVoFl9uzZtGvXjsDAQJKSkli5cuUpj3/33Xfp2rUrgYGB9OjRg88++6zK84ZhMG3aNFq1akVQUBDJycls3ry5NqVZ5vMftxG7/C8ALG1xLfeNG6mwIiIiUkc8DiwLFixgypQpTJ8+nTVr1tCrVy9SUlI4cOBAtcd///33jBs3jptuuokff/yRUaNGMWrUKNatW1d5zJNPPslzzz3HnDlzWLFiBcHBwaSkpHD06NHa/2YNaNG6DDa9/xjxtmwO+cUwbNIMhRUREZE6ZDMMw/DkBUlJSfTv358XXngBALfbTUJCAnfccQcPPPDACcePGTOGwsJCPvnkk8rHzj33XHr37s2cOXMwDIO4uDj+/Oc/c8899wCQm5tLTEwMr732GmPHjj1tTXl5eYSHh5Obm0tYWJgnv84Z+2J9Bk+8+T/+538vTlsprqtfx3H25Q1ag4iIiC/y5PvboxaWkpISVq9eTXJy8rET2O0kJyeTlpZW7WvS0tKqHA+QkpJSefz27dvJyMiockx4eDhJSUknPWdxcTF5eXlVNit8+Wsmk+ev4UHHPJy2UtztL8LR7TJLahEREWnMPAos2dnZuFwuYmKqLoAWExNDRkZGta/JyMg45fEVt56cc8aMGYSHh1duCQkJnvwadeKrjZnc9uYaBhmrSXb8iGH3x37xk6DpyyIiInXOJ2cJTZ06ldzc3Mpt9+7dDfrzv04/wC1vrMHmOsoTwfMBsA28DaI7NWgdIiIiTYVHgSU6OhqHw0FmZmaVxzMzM4mNja32NbGxsac8vuLWk3M6nU7CwsKqbA1l6aYs/vTGakpcbp6M+4aWpfsgtBUMvrfBahAREWlqPAosAQEB9O3bl9TU1MrH3G43qampDBw4sNrXDBw4sMrxAIsXL648PjExkdjY2CrH5OXlsWLFipOe0yrLNmdx8+urKClzM7YzXJb3lvnE8MfBqSsxi4iI1Bc/T18wZcoUJkyYQL9+/RgwYACzZs2isLCQiRMnAjB+/Hji4+OZMWMGAHfeeSdDhgzh73//O7///e95++23WbVqFf/6178AsNls3HXXXTz++ON06tSJxMREHn74YeLi4hg1alTd/aZn6Lst2fxx3iqKy9wknxXD35zPYis7Am0vgO6jrS5PRESkUfM4sIwZM4asrCymTZtGRkYGvXv3ZtGiRZWDZnft2oXdfqzh5rzzzmP+/Pk89NBDPPjgg3Tq1ImFCxfSvXv3ymPuu+8+CgsLufnmm8nJyeGCCy5g0aJFBAYG1sGveOa+35rNTfN+oLjMzbCuLXlxYA72+R+BzQEaaCsiIlLvPF6HxRvV5zosy7cdZOKrP3Ck1MWFXVow59oeOOcOhuxNkHQrjHyiTn+eiIhIU1Fv67A0Nev35fKH18ywMqRzC166vi/OVXPNsBLcAoaeuFCeiIiI1D2Pu4Sako4tQxjYvjklLjf/vKEvgUcOwNKZ5pPJj0JQhKX1iYiINBUKLKfg9HPw4vXnYBgQ6O+Ajx6GkgJo3R96jbO6PBERkSZDgeU0nH4Oc2fHd/DLu4ANLn4a7OpNExERaSj61q0JVxl8Vr4wXL+JENfb0nJERESaGgWWmlj1MhxYD0GRcNHDVlcjIiLS5CiwnE5BFnz1V3N/2DRoFmVtPSIiIk2QAsvpfPkIFOdCq15wzgSrqxEREWmSFFhOZf/PsPY/5v7FT4PdYW09IiIiTZRmCZ1KTHe47AXIToeEAVZXIyIi0mQpsJyK3Q7n3GB1FSIiIk2euoRERETE6ymwiIiIiNdTYBERERGvp8AiIiIiXk+BRURERLyeAouIiIh4PQUWERER8XoKLCIiIuL1FFhERETE6ymwiIiIiNdTYBERERGvp8AiIiIiXk+BRURERLxeo7has2EYAOTl5VlciYiIiNRUxfd2xff4qTSKwJKfnw9AQkKCxZWIiIiIp/Lz8wkPDz/lMTajJrHGy7ndbvbt20doaCg2m61Oz52Xl0dCQgK7d+8mLCysTs/d1Om9rR96X+uP3tv6o/e2/njze2sYBvn5+cTFxWG3n3qUSqNoYbHb7bRu3bpef0ZYWJjX/YduLPTe1g+9r/VH72390Xtbf7z1vT1dy0oFDboVERERr6fAIiIiIl5PgeU0nE4n06dPx+l0Wl1Ko6P3tn7ofa0/em/rj97b+tNY3ttGMehWREREGje1sIiIiIjXU2ARERERr6fAIiIiIl5PgUVERES8ngLLacyePZt27doRGBhIUlISK1eutLokn/bII49gs9mqbF27drW6LJ/0zTffcOmllxIXF4fNZmPhwoVVnjcMg2nTptGqVSuCgoJITk5m8+bN1hTrY0733t54440nfI5HjBhhTbE+ZMaMGfTv35/Q0FBatmzJqFGjSE9Pr3LM0aNHmTx5Ms2bNyckJITRo0eTmZlpUcW+oybv7dChQ0/43N5yyy0WVew5BZZTWLBgAVOmTGH69OmsWbOGXr16kZKSwoEDB6wuzaedffbZ7N+/v3L79ttvrS7JJxUWFtKrVy9mz55d7fNPPvkkzz33HHPmzGHFihUEBweTkpLC0aNHG7hS33O69xZgxIgRVT7Hb731VgNW6JuWLl3K5MmTWb58OYsXL6a0tJThw4dTWFhYeczdd9/Nxx9/zLvvvsvSpUvZt28fV155pYVV+4aavLcAkyZNqvK5ffLJJy2quBYMOakBAwYYkydPrrzvcrmMuLg4Y8aMGRZW5dumT59u9OrVy+oyGh3A+OCDDyrvu91uIzY21njqqacqH8vJyTGcTqfx1ltvWVCh7/rte2sYhjFhwgTj8ssvt6SexuTAgQMGYCxdutQwDPMz6u/vb7z77ruVx2zYsMEAjLS0NKvK9Em/fW8NwzCGDBli3HnnndYVdYbUwnISJSUlrF69muTk5MrH7HY7ycnJpKWlWViZ79u8eTNxcXG0b9+e6667jl27dlldUqOzfft2MjIyqnx+w8PDSUpK0ue3jixZsoSWLVvSpUsXbr31Vg4ePGh1ST4nNzcXgKioKABWr15NaWlplc9t165dadOmjT63Hvrte1vhzTffJDo6mu7duzN16lSKioqsKK9WGsXFD+tDdnY2LpeLmJiYKo/HxMSwceNGi6ryfUlJSbz22mt06dKF/fv38+ijjzJo0CDWrVtHaGio1eU1GhkZGQDVfn4rnpPaGzFiBFdeeSWJiYls3bqVBx98kJEjR5KWlobD4bC6PJ/gdru56667OP/88+nevTtgfm4DAgKIiIiocqw+t56p7r0FuPbaa2nbti1xcXH8/PPP3H///aSnp/P+++9bWG3NKbBIgxo5cmTlfs+ePUlKSqJt27a888473HTTTRZWJlJzY8eOrdzv0aMHPXv2pEOHDixZsoRhw4ZZWJnvmDx5MuvWrdMYtnpwsvf25ptvrtzv0aMHrVq1YtiwYWzdupUOHTo0dJkeU5fQSURHR+NwOE4YnZ6ZmUlsbKxFVTU+ERERdO7cmS1btlhdSqNS8RnV57dhtG/fnujoaH2Oa+j222/nk08+4euvv6Z169aVj8fGxlJSUkJOTk6V4/W5rbmTvbfVSUpKAvCZz60Cy0kEBATQt29fUlNTKx9zu92kpqYycOBACytrXAoKCti6dSutWrWyupRGJTExkdjY2Cqf37y8PFasWKHPbz3Ys2cPBw8e1Of4NAzD4Pbbb+eDDz7gq6++IjExscrzffv2xd/fv8rnNj09nV27dulzexqne2+rs3btWgCf+dyqS+gUpkyZwoQJE+jXrx8DBgxg1qxZFBYWMnHiRKtL81n33HMPl156KW3btmXfvn1Mnz4dh8PBuHHjrC7N5xQUFFT5l9H27dtZu3YtUVFRtGnThrvuuovHH3+cTp06kZiYyMMPP0xcXByjRo2yrmgfcar3NioqikcffZTRo0cTGxvL1q1bue++++jYsSMpKSkWVu39Jk+ezPz58/nwww8JDQ2tHJcSHh5OUFAQ4eHh3HTTTUyZMoWoqCjCwsK44447GDhwIOeee67F1Xu30723W7duZf78+Vx88cU0b96cn3/+mbvvvpvBgwfTs2dPi6uvIaunKXm7559/3mjTpo0REBBgDBgwwFi+fLnVJfm0MWPGGK1atTICAgKM+Ph4Y8yYMcaWLVusLssnff311wZwwjZhwgTDMMypzQ8//LARExNjOJ1OY9iwYUZ6erq1RfuIU723RUVFxvDhw40WLVoY/v7+Rtu2bY1JkyYZGRkZVpft9ap7TwHj1VdfrTzmyJEjxm233WZERkYazZo1M6644gpj//791hXtI0733u7atcsYPHiwERUVZTidTqNjx47Gvffea+Tm5lpbuAdshmEYDRmQRERERDylMSwiIiLi9RRYRERExOspsIiIiIjXU2ARERERr6fAIiIiIl5PgUVERES8ngKLiIiIeD0FFhEREfF6CiwiIiLi9RRYRERExOspsIiIiIjXU2ARERERr/f/AQ+Ks+p43/ZGAAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "for k in range(lds_array.shape[0]):\n",
    "    plt.plot(lds_array[k, 0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "58582fc9-190f-42c7-a62c-1f7b2111d74a",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "9bb0b736-25aa-4368-a617-ba6efa313cd1",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "48b22ec8-2d00-497b-8992-7442d7626802",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "381858e5-e107-480e-a471-60097744cdb7",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d86133a7-5da3-4beb-8b47-d599bfce3150",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8d2709ed-3499-4651-98eb-0cf583cff5b2",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ffcf2bce-eecd-419b-94d9-7fb8d6475cf4",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}