06 - Neural ODE - PINN Approach.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "298c5ec4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Collecting torchdiffeq\n",
      "  Downloading torchdiffeq-0.2.3-py3-none-any.whl.metadata (488 bytes)\n",
      "Requirement already satisfied: torch>=1.3.0 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torchdiffeq) (2.2.1)\n",
      "Requirement already satisfied: scipy>=1.4.0 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torchdiffeq) (1.11.4)\n",
      "Requirement already satisfied: numpy<1.28.0,>=1.21.6 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from scipy>=1.4.0->torchdiffeq) (1.26.2)\n",
      "Requirement already satisfied: filelock in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (3.13.1)\n",
      "Requirement already satisfied: typing-extensions>=4.8.0 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (4.10.0)\n",
      "Requirement already satisfied: sympy in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (1.12)\n",
      "Requirement already satisfied: networkx in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (3.2.1)\n",
      "Requirement already satisfied: jinja2 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (3.1.3)\n",
      "Requirement already satisfied: fsspec in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (2024.2.0)\n",
      "Requirement already satisfied: nvidia-cuda-nvrtc-cu12==12.1.105 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (12.1.105)\n",
      "Requirement already satisfied: nvidia-cuda-runtime-cu12==12.1.105 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (12.1.105)\n",
      "Requirement already satisfied: nvidia-cuda-cupti-cu12==12.1.105 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (12.1.105)\n",
      "Requirement already satisfied: nvidia-cudnn-cu12==8.9.2.26 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (8.9.2.26)\n",
      "Requirement already satisfied: nvidia-cublas-cu12==12.1.3.1 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (12.1.3.1)\n",
      "Requirement already satisfied: nvidia-cufft-cu12==11.0.2.54 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (11.0.2.54)\n",
      "Requirement already satisfied: nvidia-curand-cu12==10.3.2.106 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (10.3.2.106)\n",
      "Requirement already satisfied: nvidia-cusolver-cu12==11.4.5.107 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (11.4.5.107)\n",
      "Requirement already satisfied: nvidia-cusparse-cu12==12.1.0.106 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (12.1.0.106)\n",
      "Requirement already satisfied: nvidia-nccl-cu12==2.19.3 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (2.19.3)\n",
      "Requirement already satisfied: nvidia-nvtx-cu12==12.1.105 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (12.1.105)\n",
      "Requirement already satisfied: triton==2.2.0 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from torch>=1.3.0->torchdiffeq) (2.2.0)\n",
      "Requirement already satisfied: nvidia-nvjitlink-cu12 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from nvidia-cusolver-cu12==11.4.5.107->torch>=1.3.0->torchdiffeq) (12.4.99)\n",
      "Requirement already satisfied: MarkupSafe>=2.0 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from jinja2->torch>=1.3.0->torchdiffeq) (2.1.5)\n",
      "Requirement already satisfied: mpmath>=0.19 in /system/conda/miniconda3/envs/cloudspace/lib/python3.10/site-packages (from sympy->torch>=1.3.0->torchdiffeq) (1.3.0)\n",
      "Downloading torchdiffeq-0.2.3-py3-none-any.whl (31 kB)\n",
      "Installing collected packages: torchdiffeq\n",
      "Successfully installed torchdiffeq-0.2.3\n"
     ]
    }
   ],
   "source": [
    "! pip install torchdiffeq"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "initial_id",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:36:33.307542Z",
     "start_time": "2024-02-22T04:36:33.297854Z"
    },
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "from scipy.integrate import solve_ivp\n",
    "from matplotlib import pyplot as plt\n",
    "from sklearn.model_selection import train_test_split\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "from torch.utils.data import TensorDataset, DataLoader\n",
    "from torchdiffeq import odeint"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "dbf0b54a53fad777",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T03:14:42.235184Z",
     "start_time": "2024-02-22T03:14:42.223786Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def pbpk_three_comp(t,y,Qk,Qt,Kpk,Kpt,CL,Vv,Vk,Vt,Dose):\n",
    "    V,K,T = y\n",
    "    dVdt= (Dose/Vv) - (Qk/Vv)*V + (Qk/Vk)*Kpk*K - (Qt/Vv)*V + (Qt/Vt)*Kpt*T \n",
    "    dKdt= (Qk/Vk) * Kpk * V - (Qk/Vk)*K - (CL/Vk)*K\n",
    "    dTdt= (Qt/Vt) * Kpt * V - (Qt/Vt)*T\n",
    "    return [dVdt,dKdt,dTdt]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "e8991205ee28893a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T03:16:53.168928Z",
     "start_time": "2024-02-22T03:16:53.155801Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "y0=[0,0,0]\n",
    "tspan = (0,100)\n",
    "Dose= 100\n",
    "Qk,Qt,Kpk,Kpt,CL,Vv,Vk,Vt=2, 1, 3, 0.5, 5, 10, 5, 8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "525d85227eb4bb7a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T03:17:05.616127Z",
     "start_time": "2024-02-22T03:17:05.603110Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "sol = solve_ivp(pbpk_three_comp, tspan, y0, args=(Qk,Qt,Kpk,Kpt,CL,Vv,Vk,Vt,Dose), t_eval=np.linspace(0,100,100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "d111903f64d84779",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T03:17:41.876852Z",
     "start_time": "2024-02-22T03:17:41.749146Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(sol.t, sol.y[0], label='Vein')\n",
    "plt.plot(sol.t, sol.y[1], label='Kidney')\n",
    "plt.plot(sol.t, sol.y[2], label='Tumor')\n",
    "plt.xlabel('Time')\n",
    "plt.ylabel('Concentration')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "80bbedb2f270465",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:11:59.593651Z",
     "start_time": "2024-02-22T04:11:59.583962Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "#Adding some non linearity by adding Michaelis-Menten kinetics\n",
    "def michaelis_menten(C, Vmax, Km):\n",
    "    return Vmax * C / (Km + C)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "ca2afc387508aa42",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:12:37.391560Z",
     "start_time": "2024-02-22T04:12:37.379192Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def metabol_pbpk_three_comp(t, y, Qk, Qt, Kpk, Kpt, Vmax, Km, Vv, Vk, Vt, Dose):\n",
    "    V, K, T = y\n",
    "    metabolism_rate = michaelis_menten(K, Vmax, Km)\n",
    "    dVdt = (Dose/Vv) - (Qk/Vv)*V + (Qk/Vk)*Kpk*K - (Qt/Vv)*V + (Qt/Vt)*Kpt*T \n",
    "    dKdt = (Qk/Vk) * Kpk * V - (Qk/Vk)*K - metabolism_rate\n",
    "    dTdt = (Qt/Vt) * Kpt * V - (Qt/Vt)*T\n",
    "    return [dVdt, dKdt, dTdt]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "c44f5118ef2c5674",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:13:12.927458Z",
     "start_time": "2024-02-22T04:13:12.915909Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "params = {\n",
    "    \"Qk\": 1.0, \"Qt\": 0.5,\n",
    "    \"Kpk\": 0.8, \"Kpt\": 0.5,\n",
    "    \"Vmax\": 0.1, \"Km\": 0.1,\n",
    "    \"Vv\": 1.0, \"Vk\": 0.5, \"Vt\": 0.3,\n",
    "    \"Dose\": 1.0\n",
    "}\n",
    "y0 = [0, 0, 0]\n",
    "\n",
    "t_span = [0, 24] \n",
    "t_eval = np.linspace(t_span[0], t_span[1], 100)\n",
    "\n",
    "\n",
    "sol = solve_ivp(lambda t, y: metabol_pbpk_three_comp(t, y, **params), t_span, y0, t_eval=t_eval)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "80dfaf6c2445352c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:32:14.475277Z",
     "start_time": "2024-02-22T04:32:14.466457Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def simulate_pbpk_model(y0, params, t_span, t_eval):\n",
    "    sol = solve_ivp(lambda t, y: metabol_pbpk_three_comp(t, y, **params), t_span, y0, t_eval=t_eval)\n",
    "    return sol.y\n",
    "initial_conditions_range = {\n",
    "    \"V\": (0, 1),\n",
    "    \"K\": (0, 1),\n",
    "    \"T\": (0, 1)\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "d37df70a9d67af6c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:50:52.773151Z",
     "start_time": "2024-02-22T04:50:38.821383Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "data_samples = []\n",
    "for _ in range(10000):\n",
    "    y0 = [np.random.uniform(low, high) for low, high in initial_conditions_range.values()]\n",
    "    sol = solve_ivp(lambda t, y: metabol_pbpk_three_comp(t, y, **params), t_span, y0, t_eval=t_eval)\n",
    "    data_samples.append(sol.y)  \n",
    "data_samples = np.array(data_samples)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "a8d478d98523fa0a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T05:06:04.905887Z",
     "start_time": "2024-02-22T05:06:04.893709Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Average Concentration')"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "average_concentrations = np.mean(data_samples, axis=0)\n",
    "\n",
    "plt.figure(figsize=(10, 6))\n",
    "plt.plot(t_eval, average_concentrations[0], label='Vein', linestyle='-', marker='', color='blue')\n",
    "plt.plot(t_eval, average_concentrations[1], label='Kidney', linestyle='-', marker='', color='red')\n",
    "plt.plot(t_eval, average_concentrations[2], label='Tissue', linestyle='-', marker='', color='green')\n",
    "plt.title('Average Drug Concentration in Compartments Over Time')\n",
    "plt.xlabel('Time')\n",
    "plt.ylabel('Average Concentration')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "dc41d294659400c1",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:51:04.237731Z",
     "start_time": "2024-02-22T04:51:04.227174Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "data_samples_reshaped = data_samples.reshape(data_samples.shape[0], -1)  \n",
    "train,test=train_test_split(data_samples_reshaped, test_size=0.2,random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "1a7ba2c9243f81e5",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:51:11.606520Z",
     "start_time": "2024-02-22T04:51:11.598130Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "train_tensor = torch.tensor(train, dtype=torch.float32)\n",
    "test_tensor = torch.tensor(test, dtype=torch.float32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "9f6f165de2b217b6",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:51:14.108716Z",
     "start_time": "2024-02-22T04:51:14.102337Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "class NLPBPKModel(nn.Module):\n",
    "    def __init__(self):\n",
    "        super(NLPBPKModel, self).__init__()\n",
    "        self.net = nn.Sequential(\n",
    "            nn.Linear(3,128),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(128,128),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(128,128),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(128,3)\n",
    "                )\n",
    "    def forward(self, y):\n",
    "        return self.net(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "d0515798ac588a47",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T05:04:01.419414Z",
     "start_time": "2024-02-22T05:04:01.409234Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "class HybridODELSTMModel(nn.Module):\n",
    "    def __init__(self, ode_func, param_dims, hidden_size):\n",
    "        super().__init__()\n",
    "        self.ode_func = ode_func\n",
    "        self.ode_solver = odeint \n",
    "        self.param_dims = param_dims\n",
    "        self.lstm = nn.LSTM(input_size=3 + param_dims, hidden_size=hidden_size)  \n",
    "\n",
    "    def forward(self, t, x0, params):\n",
    "        ode_pred = self.ode_solver(self.ode_func, x0, t, method=\"dopri5\", options=dict(args=params))         \n",
    "        lstm_out, _ = self.lstm(ode_pred.unsqueeze(1))\n",
    "        return lstm_out[:, -1, :]       \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "bd34594d3d6318eb",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:51:23.021021Z",
     "start_time": "2024-02-22T04:51:23.013770Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "pbpk_model= NLPBPKModel()\n",
    "optimizer = torch.optim.Adam(pbpk_model.parameters(), lr=1e-3)\n",
    "criterion= nn.MSELoss()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "3739b12a9927b9cd",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:51:26.350556Z",
     "start_time": "2024-02-22T04:51:26.343153Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "train_dataset = TensorDataset(train_tensor)\n",
    "test_dataset = TensorDataset(test_tensor)\n",
    "train_loader = DataLoader(train_dataset, batch_size=64, shuffle=True)\n",
    "test_loader = DataLoader(test_dataset, batch_size=64, shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fe17d2e7e118ebdb",
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "e09fbb14b2e6cbdb",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:51:45.160553Z",
     "start_time": "2024-02-22T04:51:45.156183Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def train_and_validate(model, train_loader, test_loader, optimizer, criterion, epochs):\n",
    "    model.train() \n",
    "    train_losses = []\n",
    "    val_losses = []\n",
    "    for epoch in range(epochs):\n",
    "        total_train_loss = 0\n",
    "        for batch in train_loader:\n",
    "            y_batch = batch[0]\n",
    "            optimizer.zero_grad()\n",
    "            y_pred = model(y_batch)\n",
    "            loss = criterion(y_pred, y_batch)            \n",
    "            loss.backward()\n",
    "            optimizer.step()           \n",
    "            total_train_loss += loss.item()\n",
    "        avg_train_loss = total_train_loss / len(train_loader)\n",
    "        train_losses.append(avg_train_loss)\n",
    "        model.eval()\n",
    "        total_val_loss = 0\n",
    "        with torch.no_grad():\n",
    "            for batch in test_loader:\n",
    "                y_batch = batch[0]\n",
    "                y_pred = model(y_batch)\n",
    "                loss = criterion(y_pred, y_batch)           \n",
    "                total_val_loss += loss.item()\n",
    "        avg_val_loss = total_val_loss / len(test_loader)\n",
    "        val_losses.append(avg_val_loss)      \n",
    "        model.train() \n",
    "        print(f'Epoch {epoch+1}, Train Loss: {avg_train_loss}, Val Loss: {avg_val_loss}') \n",
    "    return train_losses, val_losses\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "f58bc91e1a13ee10",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T04:51:46.728177Z",
     "start_time": "2024-02-22T04:51:46.655109Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "RuntimeError",
     "evalue": "mat1 and mat2 shapes cannot be multiplied (64x300 and 3x128)",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mRuntimeError\u001b[0m                              Traceback (most recent call last)",
      "\u001b[1;32m/teamspace/studios/this_studio/rl_for_theranostics/06 - Neural ODE - PINN Approach.ipynb Cell 21\u001b[0m line \u001b[0;36m1\n\u001b[0;32m----> <a href='vscode-notebook-cell://vscode-01hseqhj02003qxgccaa58mamm.studio.lightning.ai/teamspace/studios/this_studio/rl_for_theranostics/06%20-%20Neural%20ODE%20-%20PINN%20Approach.ipynb#X25sdnNjb2RlLXJlbW90ZQ%3D%3D?line=0'>1</a>\u001b[0m train_losses, val_losses \u001b[39m=\u001b[39m train_and_validate(pbpk_model, train_loader, test_loader, optimizer, criterion, \u001b[39m100\u001b[39;49m)\n\u001b[1;32m      <a href='vscode-notebook-cell://vscode-01hseqhj02003qxgccaa58mamm.studio.lightning.ai/teamspace/studios/this_studio/rl_for_theranostics/06%20-%20Neural%20ODE%20-%20PINN%20Approach.ipynb#X25sdnNjb2RlLXJlbW90ZQ%3D%3D?line=2'>3</a>\u001b[0m plt\u001b[39m.\u001b[39mfigure(figsize\u001b[39m=\u001b[39m(\u001b[39m10\u001b[39m, \u001b[39m6\u001b[39m))\n\u001b[1;32m      <a href='vscode-notebook-cell://vscode-01hseqhj02003qxgccaa58mamm.studio.lightning.ai/teamspace/studios/this_studio/rl_for_theranostics/06%20-%20Neural%20ODE%20-%20PINN%20Approach.ipynb#X25sdnNjb2RlLXJlbW90ZQ%3D%3D?line=3'>4</a>\u001b[0m plt\u001b[39m.\u001b[39mplot(train_losses, label\u001b[39m=\u001b[39m\u001b[39m'\u001b[39m\u001b[39mTraining Loss\u001b[39m\u001b[39m'\u001b[39m)\n",
      "\u001b[1;32m/teamspace/studios/this_studio/rl_for_theranostics/06 - Neural ODE - PINN Approach.ipynb Cell 21\u001b[0m line \u001b[0;36m1\n\u001b[1;32m      <a href='vscode-notebook-cell://vscode-01hseqhj02003qxgccaa58mamm.studio.lightning.ai/teamspace/studios/this_studio/rl_for_theranostics/06%20-%20Neural%20ODE%20-%20PINN%20Approach.ipynb#X25sdnNjb2RlLXJlbW90ZQ%3D%3D?line=7'>8</a>\u001b[0m y_batch \u001b[39m=\u001b[39m batch[\u001b[39m0\u001b[39m]\n\u001b[1;32m      <a href='vscode-notebook-cell://vscode-01hseqhj02003qxgccaa58mamm.studio.lightning.ai/teamspace/studios/this_studio/rl_for_theranostics/06%20-%20Neural%20ODE%20-%20PINN%20Approach.ipynb#X25sdnNjb2RlLXJlbW90ZQ%3D%3D?line=8'>9</a>\u001b[0m optimizer\u001b[39m.\u001b[39mzero_grad()\n\u001b[0;32m---> <a href='vscode-notebook-cell://vscode-01hseqhj02003qxgccaa58mamm.studio.lightning.ai/teamspace/studios/this_studio/rl_for_theranostics/06%20-%20Neural%20ODE%20-%20PINN%20Approach.ipynb#X25sdnNjb2RlLXJlbW90ZQ%3D%3D?line=9'>10</a>\u001b[0m y_pred \u001b[39m=\u001b[39m model(y_batch)\n\u001b[1;32m     <a href='vscode-notebook-cell://vscode-01hseqhj02003qxgccaa58mamm.studio.lightning.ai/teamspace/studios/this_studio/rl_for_theranostics/06%20-%20Neural%20ODE%20-%20PINN%20Approach.ipynb#X25sdnNjb2RlLXJlbW90ZQ%3D%3D?line=10'>11</a>\u001b[0m loss \u001b[39m=\u001b[39m criterion(y_pred, y_batch)            \n\u001b[1;32m     <a href='vscode-notebook-cell://vscode-01hseqhj02003qxgccaa58mamm.studio.lightning.ai/teamspace/studios/this_studio/rl_for_theranostics/06%20-%20Neural%20ODE%20-%20PINN%20Approach.ipynb#X25sdnNjb2RlLXJlbW90ZQ%3D%3D?line=11'>12</a>\u001b[0m loss\u001b[39m.\u001b[39mbackward()\n",
      "File \u001b[0;32m/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/torch/nn/modules/module.py:1511\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1509\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_compiled_call_impl(\u001b[39m*\u001b[39margs, \u001b[39m*\u001b[39m\u001b[39m*\u001b[39mkwargs)  \u001b[39m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m   1510\u001b[0m \u001b[39melse\u001b[39;00m:\n\u001b[0;32m-> 1511\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49m_call_impl(\u001b[39m*\u001b[39;49margs, \u001b[39m*\u001b[39;49m\u001b[39m*\u001b[39;49mkwargs)\n",
      "File \u001b[0;32m/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/torch/nn/modules/module.py:1520\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1515\u001b[0m \u001b[39m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1516\u001b[0m \u001b[39m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m   1517\u001b[0m \u001b[39mif\u001b[39;00m \u001b[39mnot\u001b[39;00m (\u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_backward_hooks \u001b[39mor\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_backward_pre_hooks \u001b[39mor\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_forward_hooks \u001b[39mor\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m   1518\u001b[0m         \u001b[39mor\u001b[39;00m _global_backward_pre_hooks \u001b[39mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m   1519\u001b[0m         \u001b[39mor\u001b[39;00m _global_forward_hooks \u001b[39mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1520\u001b[0m     \u001b[39mreturn\u001b[39;00m forward_call(\u001b[39m*\u001b[39;49margs, \u001b[39m*\u001b[39;49m\u001b[39m*\u001b[39;49mkwargs)\n\u001b[1;32m   1522\u001b[0m \u001b[39mtry\u001b[39;00m:\n\u001b[1;32m   1523\u001b[0m     result \u001b[39m=\u001b[39m \u001b[39mNone\u001b[39;00m\n",
      "\u001b[1;32m/teamspace/studios/this_studio/rl_for_theranostics/06 - Neural ODE - PINN Approach.ipynb Cell 21\u001b[0m line \u001b[0;36m1\n\u001b[1;32m     <a href='vscode-notebook-cell://vscode-01hseqhj02003qxgccaa58mamm.studio.lightning.ai/teamspace/studios/this_studio/rl_for_theranostics/06%20-%20Neural%20ODE%20-%20PINN%20Approach.ipynb#X25sdnNjb2RlLXJlbW90ZQ%3D%3D?line=12'>13</a>\u001b[0m \u001b[39mdef\u001b[39;00m \u001b[39mforward\u001b[39m(\u001b[39mself\u001b[39m, y):\n\u001b[0;32m---> <a href='vscode-notebook-cell://vscode-01hseqhj02003qxgccaa58mamm.studio.lightning.ai/teamspace/studios/this_studio/rl_for_theranostics/06%20-%20Neural%20ODE%20-%20PINN%20Approach.ipynb#X25sdnNjb2RlLXJlbW90ZQ%3D%3D?line=13'>14</a>\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49mnet(y)\n",
      "File \u001b[0;32m/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/torch/nn/modules/module.py:1511\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1509\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_compiled_call_impl(\u001b[39m*\u001b[39margs, \u001b[39m*\u001b[39m\u001b[39m*\u001b[39mkwargs)  \u001b[39m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m   1510\u001b[0m \u001b[39melse\u001b[39;00m:\n\u001b[0;32m-> 1511\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49m_call_impl(\u001b[39m*\u001b[39;49margs, \u001b[39m*\u001b[39;49m\u001b[39m*\u001b[39;49mkwargs)\n",
      "File \u001b[0;32m/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/torch/nn/modules/module.py:1520\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1515\u001b[0m \u001b[39m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1516\u001b[0m \u001b[39m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m   1517\u001b[0m \u001b[39mif\u001b[39;00m \u001b[39mnot\u001b[39;00m (\u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_backward_hooks \u001b[39mor\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_backward_pre_hooks \u001b[39mor\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_forward_hooks \u001b[39mor\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m   1518\u001b[0m         \u001b[39mor\u001b[39;00m _global_backward_pre_hooks \u001b[39mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m   1519\u001b[0m         \u001b[39mor\u001b[39;00m _global_forward_hooks \u001b[39mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1520\u001b[0m     \u001b[39mreturn\u001b[39;00m forward_call(\u001b[39m*\u001b[39;49margs, \u001b[39m*\u001b[39;49m\u001b[39m*\u001b[39;49mkwargs)\n\u001b[1;32m   1522\u001b[0m \u001b[39mtry\u001b[39;00m:\n\u001b[1;32m   1523\u001b[0m     result \u001b[39m=\u001b[39m \u001b[39mNone\u001b[39;00m\n",
      "File \u001b[0;32m/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/torch/nn/modules/container.py:217\u001b[0m, in \u001b[0;36mSequential.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m    215\u001b[0m \u001b[39mdef\u001b[39;00m \u001b[39mforward\u001b[39m(\u001b[39mself\u001b[39m, \u001b[39minput\u001b[39m):\n\u001b[1;32m    216\u001b[0m     \u001b[39mfor\u001b[39;00m module \u001b[39min\u001b[39;00m \u001b[39mself\u001b[39m:\n\u001b[0;32m--> 217\u001b[0m         \u001b[39minput\u001b[39m \u001b[39m=\u001b[39m module(\u001b[39minput\u001b[39;49m)\n\u001b[1;32m    218\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39minput\u001b[39m\n",
      "File \u001b[0;32m/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/torch/nn/modules/module.py:1511\u001b[0m, in \u001b[0;36mModule._wrapped_call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1509\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_compiled_call_impl(\u001b[39m*\u001b[39margs, \u001b[39m*\u001b[39m\u001b[39m*\u001b[39mkwargs)  \u001b[39m# type: ignore[misc]\u001b[39;00m\n\u001b[1;32m   1510\u001b[0m \u001b[39melse\u001b[39;00m:\n\u001b[0;32m-> 1511\u001b[0m     \u001b[39mreturn\u001b[39;00m \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49m_call_impl(\u001b[39m*\u001b[39;49margs, \u001b[39m*\u001b[39;49m\u001b[39m*\u001b[39;49mkwargs)\n",
      "File \u001b[0;32m/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/torch/nn/modules/module.py:1520\u001b[0m, in \u001b[0;36mModule._call_impl\u001b[0;34m(self, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1515\u001b[0m \u001b[39m# If we don't have any hooks, we want to skip the rest of the logic in\u001b[39;00m\n\u001b[1;32m   1516\u001b[0m \u001b[39m# this function, and just call forward.\u001b[39;00m\n\u001b[1;32m   1517\u001b[0m \u001b[39mif\u001b[39;00m \u001b[39mnot\u001b[39;00m (\u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_backward_hooks \u001b[39mor\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_backward_pre_hooks \u001b[39mor\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_forward_hooks \u001b[39mor\u001b[39;00m \u001b[39mself\u001b[39m\u001b[39m.\u001b[39m_forward_pre_hooks\n\u001b[1;32m   1518\u001b[0m         \u001b[39mor\u001b[39;00m _global_backward_pre_hooks \u001b[39mor\u001b[39;00m _global_backward_hooks\n\u001b[1;32m   1519\u001b[0m         \u001b[39mor\u001b[39;00m _global_forward_hooks \u001b[39mor\u001b[39;00m _global_forward_pre_hooks):\n\u001b[0;32m-> 1520\u001b[0m     \u001b[39mreturn\u001b[39;00m forward_call(\u001b[39m*\u001b[39;49margs, \u001b[39m*\u001b[39;49m\u001b[39m*\u001b[39;49mkwargs)\n\u001b[1;32m   1522\u001b[0m \u001b[39mtry\u001b[39;00m:\n\u001b[1;32m   1523\u001b[0m     result \u001b[39m=\u001b[39m \u001b[39mNone\u001b[39;00m\n",
      "File \u001b[0;32m/home/zeus/miniconda3/envs/cloudspace/lib/python3.10/site-packages/torch/nn/modules/linear.py:116\u001b[0m, in \u001b[0;36mLinear.forward\u001b[0;34m(self, input)\u001b[0m\n\u001b[1;32m    115\u001b[0m \u001b[39mdef\u001b[39;00m \u001b[39mforward\u001b[39m(\u001b[39mself\u001b[39m, \u001b[39minput\u001b[39m: Tensor) \u001b[39m-\u001b[39m\u001b[39m>\u001b[39m Tensor:\n\u001b[0;32m--> 116\u001b[0m     \u001b[39mreturn\u001b[39;00m F\u001b[39m.\u001b[39;49mlinear(\u001b[39minput\u001b[39;49m, \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49mweight, \u001b[39mself\u001b[39;49m\u001b[39m.\u001b[39;49mbias)\n",
      "\u001b[0;31mRuntimeError\u001b[0m: mat1 and mat2 shapes cannot be multiplied (64x300 and 3x128)"
     ]
    }
   ],
   "source": [
    "\n",
    "train_losses, val_losses = train_and_validate(pbpk_model, train_loader, test_loader, optimizer, criterion, 100)\n",
    "\n",
    "plt.figure(figsize=(10, 6))\n",
    "plt.plot(train_losses, label='Training Loss')\n",
    "plt.plot(val_losses, label='Validation Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Loss')\n",
    "plt.title('Training and Validation Loss Over Epochs')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "id": "781e7941c20d7a26",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T05:29:25.872445Z",
     "start_time": "2024-02-22T05:29:25.055922Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "# Three-Compartment ODE Model \n",
    "def three_comp_ode(t, y, Qk=Qk, Qt=Qt, Kpk=Kpk, Kpt=Kpt, Vv=Vv, Vk=Vk, Vt=Vt):\n",
    "    V, K, T = y\n",
    "    dVdt = - (Qk/Vv)*V + (Qk/Vk)*Kpk*K - (Qt/Vv)*V + (Qt/Vt)*Kpt*T\n",
    "    dKdt = (Qk/Vk) * Kpk * V - (Qk/Vk)*K\n",
    "    dTdt = (Qt/Vt) * Kpt * V - (Qt/Vt)*T\n",
    "    return [dVdt, dKdt, dTdt]\n",
    "\n",
    "# Data Generation\n",
    "def generate_data(num_samples, t_span, t_eval, Qk, Qt, Kpk, Kpt, Vv, Vk, Vt, noise_std=0.1):\n",
    "    data = []\n",
    "    for _ in range(num_samples):\n",
    "        x0 = np.random.rand(3)\n",
    "        sol = solve_ivp(three_comp_ode, t_span, x0, t_eval=t_eval, args=(Qk, Qt, Kpk, Kpt, Vv, Vk, Vt))\n",
    "        data.append(sol.y.T + np.random.randn(*sol.y.T.shape) * noise_std)\n",
    "    return np.array(data)\n",
    "\n",
    "# Parameter Values\n",
    "Qk, Qt, Kpk, Kpt = 1.0, 0.5, 0.8, 0.3 \n",
    "Vv, Vk, Vt = 2.0, 1.0, 0.5\n",
    "t_span = [0, 24]\n",
    "t_eval = np.linspace(t_span[0], t_span[1], 100)\n",
    "\n",
    "# Generate Data\n",
    "data = generate_data(1000, t_span, t_eval, Qk, Qt, Kpk, Kpt, Vv, Vk, Vt)\n",
    "train_data, test_data = data[:800], data[800:] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "id": "bff430c81d346318",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T05:31:14.337238Z",
     "start_time": "2024-02-22T05:31:14.325033Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(800, 100, 3)"
      ]
     },
     "execution_count": 69,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data.shape\n",
    "#(800, 100, 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "id": "de173553bab1a70a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T05:51:44.007620Z",
     "start_time": "2024-02-22T05:51:44.000160Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "class ODEFunc(nn.Module):\n",
    "    def __init__(self, hidden_size):\n",
    "        super(ODEFunc, self).__init__()\n",
    "        self.net = nn.Sequential(\n",
    "            nn.Linear(hidden_size, hidden_size),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(hidden_size, hidden_size),\n",
    "        )\n",
    "    \n",
    "    def forward(self, t, y):\n",
    "        return self.net(y)\n",
    "\n",
    "class HybridODELSTMModel(nn.Module):\n",
    "    def __init__(self, input_size, hidden_size, output_size, seq_len, Qk, Qt, Kpk, Kpt, Vv, Vk, Vt):\n",
    "        super(HybridODELSTMModel, self).__init__()\n",
    "        self.seq_len = seq_len\n",
    "        self.lstm = nn.LSTM(input_size=input_size, hidden_size=hidden_size, batch_first=True)\n",
    "        self.ode_func = ODEFunc(hidden_size)\n",
    "        self.fc = nn.Linear(hidden_size, output_size)\n",
    "        self.Qk = nn.Parameter(torch.tensor(Qk, dtype=torch.float32))\n",
    "        self.Qt = nn.Parameter(torch.tensor(Qt, dtype=torch.float32))\n",
    "        self.Kpk = nn.Parameter(torch.tensor(Kpk, dtype=torch.float32))\n",
    "        self.Kpt = nn.Parameter(torch.tensor(Kpt, dtype=torch.float32))\n",
    "        self.Vv = nn.Parameter(torch.tensor(Vv, dtype=torch.float32))\n",
    "        self.Vk = nn.Parameter(torch.tensor(Vk, dtype=torch.float32))\n",
    "        self.Vt = nn.Parameter(torch.tensor(Vt, dtype=torch.float32))\n",
    "    \n",
    "    def forward(self, x, t):\n",
    "        lstm_out, (h_n, c_n) = self.lstm(x)\n",
    "        h_n = h_n[-1]\n",
    "        t = torch.linspace(0., self.seq_len, steps=self.seq_len, device=x.device)\n",
    "        ode_sol = odeint(self.ode_func, h_n, t, method=\"dopri5\")\n",
    "        out = self.fc(ode_sol[-1])\n",
    "        return out\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "id": "37418297ad304ece",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T05:51:45.942818Z",
     "start_time": "2024-02-22T05:51:45.934724Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from torch.utils.data import TensorDataset, DataLoader\n",
    "\n",
    "batch_size = 32  \n",
    "train_dataset = TensorDataset(torch.tensor(train_data[:, :-1]), torch.tensor(train_data[:, 1:]))  \n",
    "train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)\n",
    "# Similarly create test_dataset and test_loader\n",
    "test_dataset = TensorDataset(torch.tensor(test_data[:, :-1]), torch.tensor(test_data[:, 1:]))\n",
    "test_loader = DataLoader(test_dataset, batch_size=batch_size, shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "id": "62c91a3fbe0177ae",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T05:51:46.567060Z",
     "start_time": "2024-02-22T05:51:46.561316Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "input_size = 3  # Since each timestep contains concentrations from 3 compartments\n",
    "hidden_size = 128  # Example hidden size for LSTM\n",
    "output_size = 3  # Predicting concentrations for 3 compartments at the next timestep\n",
    "seq_len = 100  # Length of each input sequence\n",
    "model = HybridODELSTMModel(input_size=3, hidden_size=128, output_size=3, seq_len=100, Qk=1.0, Qt=0.5, Kpk=0.8, Kpt=0.3, Vv=2.0, Vk=1.0, Vt=0.5)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "id": "3e9ad49018aa6085",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T06:05:32.314637Z",
     "start_time": "2024-02-22T06:03:06.465583Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1: Train Loss = 174.9932, Val Loss = 3.2629\n",
      "Validation loss decreased (inf --> 3.262929).  Saving model ...\n",
      "Epoch 2: Train Loss = 4.1251, Val Loss = 0.3167\n",
      "Validation loss decreased (3.262929 --> 0.316665).  Saving model ...\n",
      "Epoch 3: Train Loss = 0.3486, Val Loss = 0.0734\n",
      "Validation loss decreased (0.316665 --> 0.073402).  Saving model ...\n",
      "Epoch 4: Train Loss = 0.0587, Val Loss = 0.0333\n",
      "Validation loss decreased (0.073402 --> 0.033308).  Saving model ...\n",
      "Epoch 5: Train Loss = 0.0316, Val Loss = 0.0368\n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 6: Train Loss = 0.0319, Val Loss = 0.0400\n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 7: Train Loss = 0.0287, Val Loss = 0.0326\n",
      "Validation loss decreased (0.033308 --> 0.032617).  Saving model ...\n",
      "Epoch 8: Train Loss = 0.0310, Val Loss = 0.0333\n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 9: Train Loss = 0.0310, Val Loss = 0.0365\n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 10: Train Loss = 0.0296, Val Loss = 0.0401\n",
      "EarlyStopping counter: 3 out of 20\n",
      "Epoch 11: Train Loss = 0.0292, Val Loss = 0.0349\n",
      "EarlyStopping counter: 4 out of 20\n",
      "Epoch 12: Train Loss = 0.0343, Val Loss = 0.0649\n",
      "EarlyStopping counter: 5 out of 20\n",
      "Epoch 13: Train Loss = 0.0347, Val Loss = 0.0424\n",
      "EarlyStopping counter: 6 out of 20\n",
      "Epoch 14: Train Loss = 0.0359, Val Loss = 0.0409\n",
      "EarlyStopping counter: 7 out of 20\n",
      "Epoch 15: Train Loss = 0.0370, Val Loss = 0.0296\n",
      "Validation loss decreased (0.032617 --> 0.029621).  Saving model ...\n",
      "Epoch 16: Train Loss = 0.0353, Val Loss = 0.0398\n",
      "EarlyStopping counter: 1 out of 20\n",
      "Epoch 17: Train Loss = 0.0355, Val Loss = 0.0417\n",
      "EarlyStopping counter: 2 out of 20\n",
      "Epoch 18: Train Loss = 0.0394, Val Loss = 0.0410\n",
      "EarlyStopping counter: 3 out of 20\n",
      "Epoch 19: Train Loss = 0.0373, Val Loss = 0.0490\n",
      "EarlyStopping counter: 4 out of 20\n",
      "Epoch 20: Train Loss = 0.0444, Val Loss = 0.0480\n",
      "EarlyStopping counter: 5 out of 20\n",
      "Epoch 21: Train Loss = 0.0304, Val Loss = 0.0362\n",
      "EarlyStopping counter: 6 out of 20\n",
      "Epoch 22: Train Loss = 0.0435, Val Loss = 0.0789\n",
      "EarlyStopping counter: 7 out of 20\n",
      "Epoch 23: Train Loss = 0.0576, Val Loss = 0.1053\n",
      "EarlyStopping counter: 8 out of 20\n",
      "Epoch 24: Train Loss = 0.0548, Val Loss = 0.0587\n",
      "EarlyStopping counter: 9 out of 20\n",
      "Epoch 25: Train Loss = 0.0703, Val Loss = 0.0527\n",
      "EarlyStopping counter: 10 out of 20\n",
      "Epoch 26: Train Loss = 0.0516, Val Loss = 0.0300\n",
      "EarlyStopping counter: 11 out of 20\n",
      "Epoch 27: Train Loss = 0.0612, Val Loss = 0.0446\n",
      "EarlyStopping counter: 12 out of 20\n",
      "Epoch 28: Train Loss = 0.0547, Val Loss = 0.0474\n",
      "EarlyStopping counter: 13 out of 20\n",
      "Epoch 29: Train Loss = 0.0681, Val Loss = 0.0326\n",
      "EarlyStopping counter: 14 out of 20\n",
      "Epoch 30: Train Loss = 0.0395, Val Loss = 0.0795\n",
      "EarlyStopping counter: 15 out of 20\n",
      "Epoch 31: Train Loss = 0.0583, Val Loss = 0.0654\n",
      "EarlyStopping counter: 16 out of 20\n",
      "Epoch 32: Train Loss = 0.0882, Val Loss = 0.1082\n",
      "EarlyStopping counter: 17 out of 20\n",
      "Epoch 33: Train Loss = 0.0775, Val Loss = 0.0616\n",
      "EarlyStopping counter: 18 out of 20\n",
      "Epoch 34: Train Loss = 0.1214, Val Loss = 0.0859\n",
      "EarlyStopping counter: 19 out of 20\n",
      "Epoch 35: Train Loss = 0.0929, Val Loss = 0.1856\n",
      "EarlyStopping counter: 20 out of 20\n",
      "Early stopping\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "class EarlyStopping:\n",
    "    def __init__(self, patience=10, verbose=False, delta=0):\n",
    "        self.patience = patience\n",
    "        self.verbose = verbose\n",
    "        self.delta = delta\n",
    "        self.best_score = None\n",
    "        self.early_stop = False\n",
    "        self.counter = 0\n",
    "        self.best_loss = np.Inf\n",
    "\n",
    "    def __call__(self, val_loss):\n",
    "        score = -val_loss\n",
    "\n",
    "        if self.best_score is None:\n",
    "            self.best_score = score\n",
    "            self.save_checkpoint(val_loss)\n",
    "        elif score < self.best_score + self.delta:\n",
    "            self.counter += 1\n",
    "            print(f'EarlyStopping counter: {self.counter} out of {self.patience}')\n",
    "            if self.counter >= self.patience:\n",
    "                self.early_stop = True\n",
    "        else:\n",
    "            self.best_score = score\n",
    "            self.save_checkpoint(val_loss)\n",
    "            self.counter = 0\n",
    "\n",
    "    def save_checkpoint(self, val_loss):\n",
    "        if self.verbose:\n",
    "            print(f'Validation loss decreased ({self.best_loss:.6f} --> {val_loss:.6f}).  Saving model ...')\n",
    "        self.best_loss = val_loss\n",
    "\n",
    "optimizer = torch.optim.Adam(model.parameters())\n",
    "criterion = nn.MSELoss()\n",
    "early_stopping = EarlyStopping(patience=20, verbose=True)\n",
    "\n",
    "train_losses = []\n",
    "val_losses = []\n",
    "\n",
    "for epoch in range(100):\n",
    "    model.train()\n",
    "    epoch_losses = []\n",
    "\n",
    "    for x, y in train_loader:\n",
    "        x, y = x.float(), y.float()\n",
    "        t_eval_tensor = torch.linspace(0., model.seq_len, steps=model.seq_len, device=x.device)\n",
    "        \n",
    "        optimizer.zero_grad()\n",
    "        preds = model(x, t_eval_tensor)\n",
    "        y_true_adjusted = y[:, -1, :]\n",
    "        loss = criterion(preds, y_true_adjusted)\n",
    "        loss.backward()\n",
    "        optimizer.step()\n",
    "\n",
    "        epoch_losses.append(loss.item())\n",
    "    \n",
    "    train_loss = np.mean(epoch_losses)\n",
    "    train_losses.append(train_loss)\n",
    "\n",
    "    model.eval()\n",
    "    val_epoch_losses = []\n",
    "    with torch.no_grad():\n",
    "        for x, y in test_loader:\n",
    "            x, y = x.float(), y.float()\n",
    "            preds = model(x, t_eval_tensor)\n",
    "            y_true_adjusted = y[:, -1, :]\n",
    "            val_loss = criterion(preds, y_true_adjusted)\n",
    "            val_epoch_losses.append(val_loss.item())\n",
    "    \n",
    "    val_loss = np.mean(val_epoch_losses)\n",
    "    val_losses.append(val_loss)\n",
    "    print(f\"Epoch {epoch+1}: Train Loss = {train_loss:.4f}, Val Loss = {val_loss:.4f}\")\n",
    "\n",
    "    early_stopping(val_loss)\n",
    "    if early_stopping.early_stop:\n",
    "        print(\"Early stopping\")\n",
    "        break\n",
    "\n",
    "plt.figure(figsize=(10, 6))\n",
    "plt.plot(train_losses, label='Training Loss')\n",
    "plt.plot(val_losses, label='Validation Loss')\n",
    "plt.xlabel('Epoch')\n",
    "plt.ylabel('Loss')\n",
    "plt.title('Training and Validation Loss Over Epochs')\n",
    "plt.legend()\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "id": "a5c57a04793c2599",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T06:14:07.016437Z",
     "start_time": "2024-02-22T06:14:07.003592Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'y_preds' 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[106], line 1\u001b[0m\n\u001b[0;32m----> 1\u001b[0m \u001b[43my_preds\u001b[49m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'y_preds' is not defined"
     ]
    }
   ],
   "source": [
    "y_preds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "id": "967896791644a52f",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T06:14:33.476359Z",
     "start_time": "2024-02-22T06:14:32.840921Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Loss: 0.1856\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA2AAAAIhCAYAAAAo4dnZAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8pXeV/AAAACXBIWXMAAA9hAAAPYQGoP6dpAACFXElEQVR4nOzdd1QU9/7G8ffu0pEigmBBwN6xKyiWFLvRVMuNKTfNVMvNTTPVJNdUo8mNRtM1iZpimhqNKSr2WLB3pYgggkqVtju/P0j4XWM0ShvK8zpnz9FhdubBTYCH7+xnLIZhGIiIiIiIiEi5s5odQEREREREpKZQARMREREREakgKmAiIiIiIiIVRAVMRERERESkgqiAiYiIiIiIVBAVMBERERERkQqiAiYiIiIiIlJBVMBEREREREQqiAqYiIiIiIhIBVEBExGRKsNisVzSY+XKlaU+V05ODs8+++wlHys2NhaLxcJrr71W6nOLiEj15WR2ABERkUu1fv36c/7+/PPP8+uvv/LLL7+cs71169alPldOTg7PPfccAH379i318UREREAFTEREqpAePXqc8/eAgACsVut520VERCorXYIoIiLVSn5+Pi+88AItW7bE1dWVgIAAbr/9dk6ePHnOfr/88gt9+/alTp06uLu706hRI66//npycnKIjY0lICAAgOeee6740sbbbrut1Pni4+O5+eabqVu3Lq6urrRq1YrXX38dh8Nxzn6zZs0iPDycWrVq4eXlRcuWLXniiSeKP56Tk8PDDz9MWFgYbm5u+Pn50aVLF+bPn1/qjCIiUn60AiYiItWGw+Fg+PDhREdH88gjjxAZGUlcXBzPPPMMffv2ZfPmzbi7uxMbG8uQIUOIiorigw8+wNfXl8TERJYtW0Z+fj716tVj2bJlDBw4kDvuuIM777wToLiUldTJkyeJjIwkPz+f559/ntDQUBYvXszDDz/M4cOHmTlzJgALFizgvvvu48EHH+S1117DarVy6NAh9uzZU3ysSZMmMW/ePF544QU6duxIdnY2u3btIi0trVQZRUSkfKmAiYhItfH555+zbNkyvvrqK6677rri7eHh4XTt2pWPPvqIe++9ly1btpCbm8urr75KeHh48X5jxowp/nPnzp0BaNiwYZld4jht2jQSExPZuHEj3bp1A2DAgAHY7XbeeecdJkyYQPPmzVm7di2+vr68+eabxc+98sorzznW2rVr6d+/PxMnTizeNmTIkDLJKSIi5UeXIIqISLWxePFifH19GTZsGIWFhcWPDh06EBQUVDzRsEOHDri4uHD33Xfz8ccfc+TIkQrJ98svv9C6devi8vWH2267DcMwioeJdOvWjTNnzjB69Gi+/fZbUlNTzztWt27d+OGHH3jsscdYuXIlZ8+erZDPQURESkcFTEREqo0TJ05w5swZXFxccHZ2PueRnJxcXGSaNGnCTz/9RN26dbn//vtp0qQJTZo0YcaMGeWaLy0tjXr16p23vX79+sUfBxg7diwffPABcXFxXH/99dStW5fu3buzYsWK4ue8+eabPProo3zzzTf069cPPz8/RowYwcGDB8v1cxARkdJRARMRkWrD39+fOnXq8Ntvv/3l44/3WAFERUXx/fffk56ezoYNG4iIiGDChAksWLCg3PLVqVOHpKSk87YfP368OP8fbr/9dtatW0d6ejpLlizBMAyGDh1KXFwcAJ6enjz33HPs27eP5ORkZs2axYYNGxg2bFi55RcRkdJTARMRkWpj6NChpKWlYbfb6dKly3mPFi1anPccm81G9+7defvttwHYunUrAK6urgBlemnflVdeyZ49e4rP8Ye5c+disVjo16/fec/x9PRk0KBBTJ48mfz8fHbv3n3ePoGBgdx2222MHj2a/fv3k5OTU2aZRUSkbGkIh4iIVBujRo3i008/ZfDgwYwfP55u3brh7OzMsWPH+PXXXxk+fDjXXnst77zzDr/88gtDhgyhUaNG5Obm8sEHHwBw1VVXAeDl5UVISAjffvstV155JX5+fvj7+xMaGnrRDDt37uTLL788b3vXrl2ZOHEic+fOZciQIUyZMoWQkBCWLFnCzJkzuffee2nevDkAd911F+7u7vTs2ZN69eqRnJzM1KlT8fHxoWvXrgB0796doUOH0r59e2rXrs3evXuZN28eEREReHh4lOG/qoiIlClDRESkirr11lsNT0/Pc7YVFBQYr732mhEeHm64ubkZtWrVMlq2bGncc889xsGDBw3DMIz169cb1157rRESEmK4uroaderUMfr06WN899135xzrp59+Mjp27Gi4uroagHHrrbdeMMvRo0cN4IKPDz/80DAMw4iLizPGjBlj1KlTx3B2djZatGhhvPrqq4bdbi8+1scff2z069fPCAwMNFxcXIz69esbN910k7Fjx47ifR577DGjS5cuRu3atQ1XV1ejcePGxsSJE43U1NRS/quKiEh5shiGYZhX/0RERERERGoOvQdMRERERESkgqiAiYiIiIiIVBAVMBERERERkQqiAiYiIiIiIlJBVMBEREREREQqiAqYiIiIiIhIBdGNmEvI4XBw/PhxvLy8sFgsZscRERERERGTGIZBZmYm9evXx2q9+BqXClgJHT9+nODgYLNjiIiIiIhIJZGQkEDDhg0vuo8KWAl5eXkBRf/I3t7eJqcRERERERGzZGRkEBwcXNwRLkYFrIT+uOzQ29tbBUxERERERC7prUkawiEiIiIiIlJBVMBEREREREQqiAqYiIiIiIhIBdF7wEREREREKpBhGBQWFmK3282OIpfIZrPh5ORUJrefUgETEREREakg+fn5JCUlkZOTY3YUuUweHh7Uq1cPFxeXUh1HBUxEREREpAI4HA6OHj2KzWajfv36uLi4lMmKipQvwzDIz8/n5MmTHD16lGbNmv3tzZYvRgVMRERERKQC5Ofn43A4CA4OxsPDw+w4chnc3d1xdnYmLi6O/Px83NzcSnwsDeEQEREREalApVk9EfOU1eumV19ERERERKSCqICJiIiIiIhUEBUwERERERGRCqICJiIiIiIiF2SxWC76uO2220p87NDQUKZPn15m+1UFmoIoIiIiIiIXlJSUVPznhQsX8vTTT7N///7ibe7u7mbEqrK0AiYiIiIiYhLDMMjJLzTlYRjGJWUMCgoqfvj4+GCxWM7Ztnr1ajp37oybmxuNGzfmueeeo7CwsPj5zz77LI0aNcLV1ZX69evz0EMPAdC3b1/i4uKYOHFi8WpaSc2aNYsmTZrg4uJCixYtmDdv3jkfv1AGgJkzZ9KsWTPc3NwIDAzkhhtuKHGOS6EVMBERERERk5wtsNP66eWmnHvPlAF4uJSuDixfvpybb76ZN998k6ioKA4fPszdd98NwDPPPMOXX37JG2+8wYIFC2jTpg3Jycls374dgEWLFhEeHs7dd9/NXXfdVeIMX3/9NePHj2f69OlcddVVLF68mNtvv52GDRvSr1+/i2bYvHkzDz30EPPmzSMyMpJTp04RHR1dqn+Tv6MCJiIiIiIiJfLiiy/y2GOPceuttwLQuHFjnn/+eR555BGeeeYZ4uPjCQoK4qqrrsLZ2ZlGjRrRrVs3APz8/LDZbHh5eREUFFTiDK+99hq33XYb9913HwCTJk1iw4YNvPbaa/Tr1++iGeLj4/H09GTo0KF4eXkREhJCx44dS/mvcnEqYCIiIiL/I/HMWc7m22lat5bZUaQGcHe2sWfKANPOXVpbtmzht99+48UXXyzeZrfbyc3NJScnhxtvvJHp06fTuHFjBg4cyODBgxk2bBhOTmVXQ/bu3Vu86vaHnj17MmPGDICLZrj66qsJCQkp/tjAgQO59tpr8fDwKLN8f6b3gImIiIj8LvHMWQbPiGbwjGgOnMg0O47UABaLBQ8XJ1MepXnP1R8cDgfPPfccMTExxY+dO3dy8OBB3NzcCA4OZv/+/bz99tu4u7tz33330bt3bwoKCsrgX+///flzMQyjeNvFMnh5ebF161bmz59PvXr1ePrppwkPD+fMmTNlmu9/qYCJiIiIAHaHwaSFMaSfLSDf7uD5xXsueUiBSE3VqVMn9u/fT9OmTc97WK1FVcPd3Z1rrrmGN998k5UrV7J+/Xp27twJgIuLC3a7vVQZWrVqxZo1a87Ztm7dOlq1alX894tlcHJy4qqrruKVV15hx44dxMbG8ssvv5Qq08XoEkQRERERYPbqw2w8egoPFxuFdoPog6ms3H+Sfi3rmh1NpNJ6+umnGTp0KMHBwdx4441YrVZ27NjBzp07eeGFF/joo4+w2+10794dDw8P5s2bh7u7OyEhIUDR/b1Wr17NqFGjcHV1xd/f/4LnSkxMJCYm5pxtjRo14t///jc33XQTnTp14sorr+T7779n0aJF/PTTTwAXzbB48WKOHDlC7969qV27NkuXLsXhcNCiRYty+zfTCpiIiIjUeNsTzjDtxwMAPHtNG27rGQrAC0v2UGB3mJhMpHIbMGAAixcvZsWKFXTt2pUePXowbdq04oLl6+vLu+++S8+ePWnfvj0///wz33//PXXq1AFgypQpxMbG0qRJEwICAi56rtdee42OHTue8/juu+8YMWIEM2bM4NVXX6VNmzbMnj2bDz/8kL59+/5tBl9fXxYtWsQVV1xBq1ateOedd5g/fz5t2rQpt38zi6G19RLJyMjAx8eH9PR0vL29zY4jIiIiJZSdV8jQt9ZwNDWbwe2CeHtMJzLzCun36krSsvN5dlhrbusZZnZMqQZyc3M5evQoYWFhuLm5mR1HLtPFXr/L6QZaARMREZEa7fnFeziamk2Qtxv/ubYdFosFbzdnJl7dHIDpPx/kTE6+ySlFpLpQARMREZEaa9muJBb8loDFAtNGhuPr4VL8sVFdg2kR6MWZnAJm/HzQxJQiUp2ogImIiEiNlJyey2OLiqag3d27MZFNzn3zv5PNypNDi6aozVsfx+GTWRWeUUSqHxUwERERqXEcDoN/fRHDmZwC2jbw5l9X//XEs6hmAVzZsi6FDoP/LNlbwSlFpDpSARMREZEa5/01R1l7KA03ZyszRnXExenCPxI9MaQVTlYLP+9LIfrgyQpMKdWVZuBVTWX1uqmAiYiISI2y+3g6ryzfB8DTQ9vQJKDWRfdvElCLsRFFI7VfWLyXQo2llxJydnYGICcnx+QkUhJ/vG5/vI4lpRsxi4iISI1xNt/O+AUxFNgNrm4dyOhuwZf0vPFXNuPrbYnsP5HJgt8SuLlHSDknlerIZrPh6+tLSkoKAB4eHlgsFpNTyd8xDIOcnBxSUlLw9fXFZrOV6ngqYCIiIlJj/GfpXg6lZBHg5crL17e/5B9+fT1cmHBlM579fg/TVhzgmg718XYr3W/BpWYKCgoCKC5hUnX4+voWv36loQImIiIiNcLPe08wb0McAK/fGI6fp8vfPONc/+gRwrwNcRw+mc1/fznEE4NblUdMqeYsFgv16tWjbt26FBQUmB1HLpGzs3OpV77+oAImIiIi1V5KZi6PfLkDgDt6hdG7ecBlH8PZZuXJIa25/aPf+HDtUcZ0a0Sov2dZR5UawmazldkP9FK1aAiHiIiIVGsOh8G/v9hBWnY+LYO8+PeAvx45fyn6tgigd/MACuwGU3/QWHoRuXwqYCIiIlKtfbw+llUHTuLqZOXN0R1xcy75qoPFYuHJIa2wWS0s332CdYdTyzCpiNQEKmAiIiJSbe1LzmDqD0Uj558Y3IrmgV6lPmbzQC/GdGsEFI2ltzt0TycRuXQqYCIiIlIt5RbYGT8/hvxCB/1aBHBLRNmNjp94dXO83JzYk5TBl1sSyuy4IlL9qYCJiIhItfTysn3sP5GJfy0XXrkhvEzvt+Tn6cL4K5sB8OryA2TlFZbZsUWkelMBExERkWpn5f4UPlwbC8CrN4QT4OVa5ue4JSKU0DoepGblMfPXQ2V+fBGpnlTAREREpFpJy8rj4S+KRs7fGhFCv5Z1y+U8Lk7W4nuBvbfmKAmncsrlPCJSvaiAiYiISLVhGAaPfrWD1Kw8mtWtxePlfLPkq1sHEtmkDvmFDl5atq9czyUi1YMKmIiIiFQbn26M56e9KbjYrMwYVbqR85fCYrHw1NDWWC2wZEcSv8WeKtfziUjVpwImIiIi1cKhlExeWLIHgEcGtqB1fe8KOW+ret6M7BoMwPOL9+DQWHoRuQgVMBEREany8grtPDQ/htwCB1HN/Plnz7AKPf+kq1tQy9WJHcfS+XpbYoWeW0SqFhUwERERqfJe//EAe5IyqO3hzGs3hmO1lt3I+UsR4OXK/f2aAvDK8n3k5GssvYj8NRUwERERqdLWHExlzuojALx0fXsCvd1MyXF7z1CC/dw5kZHHO6uOmJJBRCo/FTARERGpsk5n5/OvL2IAGN2tEQPaBJmWxc3ZxuODiqYuzll9mONnzpqWRUQqLxUwERERqZIMw+CxRTs4kZFH4wBPnhpaviPnL8WgtkF0C/Ujt8DBKxpLLyJ/QQVMREREqqTPNyewfPcJnG0W3hzVEQ8XJ7MjFY+lt1jgm5jjbIs/bXYkEalkVMBERESkyjlyMotnvysaOf+v/i1o28DH5ET/r11DH67v1BCAKYv3YBgaSy8i/08FTERERKqUAruDCQtjOFtgJ6JxHe6Oamx2pPP8e0ALPFxsbIs/w3fbj5sdR0QqERUwERERqVKm/3SAHcfS8XF35vWbKn7k/KUI9Hbj3j5NAHj5h33kFthNTiQilYUKmIiIiFQZG4+kMXPlYQD+c2076vu6m5zowu7q3Zj6Pm4cT8/l3dUaSy8iRVTAREREpEpIP1vAxIUxGAbc2LkhQ9rXMzvSRbk523h0UEsAZq06zImMXJMTiUhlYHoBmzlzJmFhYbi5udG5c2eio6MvuO+iRYu4+uqrCQgIwNvbm4iICJYvX37efl999RWtW7fG1dWV1q1b8/XXX5fqvCIiImIuwzCY/PVOjqfnElLHg2euaWN2pEtyTXh9OjbyJSffzqvL95sdR0QqAVML2MKFC5kwYQKTJ09m27ZtREVFMWjQIOLj4/9y/9WrV3P11VezdOlStmzZQr9+/Rg2bBjbtm0r3mf9+vWMHDmSsWPHsn37dsaOHctNN93Exo0bS3xeERERMdfX2xJZvCMJm9XC9JEdqOVq/sj5S2GxWHh6aGsAvtxyjJ3H0k1OJCJmsxgmzkbt3r07nTp1YtasWcXbWrVqxYgRI5g6deolHaNNmzaMHDmSp59+GoCRI0eSkZHBDz/8ULzPwIEDqV27NvPnzy+z82ZkZODj40N6ejre3t6X9BwRERG5fPFpOQx+M5qsvEL+dXVzHryymdmRLtuEBdv4JuY43UL9WHhPDyyWyjc4RERK7nK6gWkrYPn5+WzZsoX+/fufs71///6sW7fuko7hcDjIzMzEz8+veNv69evPO+aAAQOKj1nS8+bl5ZGRkXHOQ0RERMpXod3B+IXbyMorpGtobe7r19TsSCXyyMCWuDlb2RR7ih92JZsdR0RMZFoBS01NxW63ExgYeM72wMBAkpMv7QvT66+/TnZ2NjfddFPxtuTk5Ises6TnnTp1Kj4+PsWP4ODgS8ooIiIiJffWL4fYFn8GL1cnpt3UAVslHDl/Ker7unN376Kx9FN/2Kux9CI1mOlDOP68BG8YxiUty8+fP59nn32WhQsXUrdu3cs+5uWe9/HHHyc9Pb34kZCQ8LcZRUREpOS2xJ3irV8OAvDCtW0J9vMwOVHpjOvTmEBvVxJOneXDtbFmxxERk5hWwPz9/bHZbOetOqWkpJy3OvVnCxcu5I477uDzzz/nqquuOudjQUFBFz1mSc/r6uqKt7f3OQ8REREpH5m5BYxfEIPDgGs7NmB4hwZmRyo1DxcnHhlQNJb+7V8PcTIzz+REImIG0wqYi4sLnTt3ZsWKFedsX7FiBZGRkRd83vz587ntttv47LPPGDJkyHkfj4iIOO+YP/74Y/ExS3peERERqTjPfLubY6fP0rC2O88Nrxoj5y/FtR0b0L6hD1l5hUxbobH0IjWRqTNcJ02axNixY+nSpQsRERHMmTOH+Ph4xo0bBxRd9peYmMjcuXOBovJ1yy23MGPGDHr06FG8iuXu7o6Pjw8A48ePp3fv3rz88ssMHz6cb7/9lp9++ok1a9Zc8nlFRETEPN/GJLJoWyJWC0wf2QFvN2ezI5UZq7VoLP0N76xnwW8JjO0RSuv6uqpGpCYx9T1gI0eOZPr06UyZMoUOHTqwevVqli5dSkhICABJSUnn3Jtr9uzZFBYWcv/991OvXr3ix/jx44v3iYyMZMGCBXz44Ye0b9+ejz76iIULF9K9e/dLPq+IiIiY49jpHJ78ZhcAD1zRjC6hfn/zjKqnS6gfQ9rXwzDg+cV7MPGOQCJiAlPvA1aV6T5gIiIiZcvuMBg9ZwObYk/RsZEvX9wTgZPN9Hlh5SLhVA5XTltFfqGDOWM7079NkNmRRKQUqsR9wERERET+1zurDrMp9hSeLjamj+xQbcsXQLCfB3f2CgPgP0v3kl/oMDmRiFSU6vuVTURERKqMmIQzvLHiAADPDW9LSB1PkxOVv/v6NcW/liuxaTnMXR9rdhwRqSAqYCIiImKq7LxCJizYRqHDYEj7elzfqeqPnL8UtVydeGRACwBm/HyQU9n5JicSkYqgAiYiIiKmeu773cSm5VDPx43/jGiHxWIxO1KFub5zQ1rX8yYzt7B4BVBEqjcVMBERETHN0p1JfL75GBYLvDGyAz4e1Wfk/KWwWS08NbQ1AJ9ujOPAiUyTE4lIeVMBExEREVMkpZ/l8UU7Abi3TxN6NK5jciJzRDSpw4A2gTg0ll6kRlABExERkQrncBhMWrid9LMFtG/ow4SrmpsdyVRPDG6Fs81C9MFUVu4/aXYcESlHKmAiIiJS4d6NPsL6I2m4OxeNnHdxqtk/koTU8eT2nkVj6V9YsocCu8bSi1RXNfurnYiIiFS4XYnpvPbjfgCeGdaaxgG1TE5UOTxwRVPqeLpw+GQ2n26IMzuOiJQTFTARERGpMGfz7Ty0YBsFdoMBbQIZ2TXY7EiVhrebM5P6F12KOf3ng5zJ0Vh6kepIBUxEREQqzAtL9nDkZDaB3q68dF37GjVy/lKM7BJMi0AvzuQUMOPng2bHEZFyoAImIiIiFWLFnhN8ujEegNdv7EBtTxeTE1U+TjYrTw5tBcC89XEcPpllciIRKWsqYCIiIlLuUjJyefSrHQDcFRVGr2b+JieqvKKaBXBly7oUOgz+s2Sv2XFEpIypgImIiEi5cjgM/vXFdk5l59OqnjcPD2hhdqRK74khrXCyWvh5XwrRBzWWXqQ6UQETERGRcvXhuliiD6bi6mTlzVEdcHWymR2p0msSUIuxESEAvLB4L4UaSy9SbaiAiYiISLnZm5TByz/sA+DJoa1pFuhlcqKqY/yVzfD1cGb/iUwW/JZgdhwRKSMqYCIiIlIucgvsjF+wjXy7gytb1uXm7o3MjlSl+Hq4MPGqorH001YcICO3wOREIlIWVMBERESkXLz0wz4OnMjCv5YrL9+gkfMlMaZ7I5oEeHIqO5///nLI7DgiUgZUwERERKTM/bovhY/WxQLw2o3t8a/lam6gKsrZZuXJoa0B+HDtUWJTs01OJCKlpQImIiIiZSo1K49/f7kdgNsiQ+nboq7Jiaq2fi3q0rt5AAV2g6k/aCy9SFWnAiYiIiJlxjAMHvlyB6lZ+bQI9OKxQS3NjlQtPDmkFTarheW7T7DucKrZcUSkFFTAREREpMx8siGOX/al4OJkZcboDrg5a+R8WWge6MWYbkVDTF5YvBe7wzA5kYiUlAqYiIiIlImDJzJ5YUnRJXKPDWxJyyBvkxNVLxOvbo6XmxN7kjL4covG0otUVSpgIiIiUmp5hXYeWhBDXqGD3s0DuC0y1OxI1Y6fpwvjr2wGwKvLD5CVV2hyIhEpCRUwERERKbXXlu9nb1IGfp4uvHZDe6xWjZwvD7dEhBLm70lqVh4zf9VYepGqSAVMRERESiX64EnejT4KwMvXt6eut5vJiaovFycrTwxuBcB7a46ScCrH5EQicrlUwERERKTETmXn86/Pi0bO39yjEVe3DjQ5UfV3Vau6RDapQ36hg5eW7TM7johcJhUwERERKRHDMHj0qx2kZObRJMCTyYNbmx2pRrBYLDw1tDVWCyzZkcRvsafMjiQil0EFTEREREpkwW8JrNhzAmebhRmjOuLuopHzFaVVPW9Gdg0G4PnFe3BoLL1IlaECJiIiIpft8Mkspny/B4B/D2hB2wY+JieqeSZd3YJark7sOJbO19sSzY4jIpdIBUxEREQuS36hgwkLYjhbYCeySR3u7NXY7Eg1UoCXKw9c0RSAV5bvIydfY+lFqgIVMBEREbksb/x0gJ2J6fi4OzPtpg4aOW+i23uGEuznzomMPN5ZdcTsOCJyCVTARERE5JKtP5zGO6sOA/DSde0I8tHIeTO5Otl4YlDRWPo5qw9z/MxZkxOJyN9RARMREZFLkp5TwKTPYzAMGNklmEHt6pkdSYCBbYPoFuZHboGDVzSWXqTSUwETERGRv2UYBk98vZOk9FxC63jw9DCNnK8sLBYLTw1pjcUC38QcZ1v8abMjichFqICJiIjI3/pqayJLdibhZC0aOe/p6mR2JPkf7Rr6cH2nhgBMWbwHw9BYepHKSgVMRERELiouLZtnvt0FwMSrmxMe7GtuIPlL/x7QAg8XG9viz/Dd9uNmxxGRC1ABExERkQsqsDsYvyCG7Hw73cL8GNenidmR5AICvd24r2/R6/PyD/vILbCbnEhE/ooKmIiIiFzQWz8fJCbhDF5uTrwxsgM2jZyv1O6MakwDX3eOp+fy7mqNpRepjFTARERE5C/9FnuK//56CID/XNuOBr7uJieSv+PmbOPRQS0BmLnyMCcyck1OJCJ/pgImIiIi58nILWDCghgcBlzXqQHDwuubHUku0bD29ejUyJezBXZeXb7f7Dgi8icqYCIiInKep7/ZReKZswT7ufPcNW3MjiOXwWKx8NTQotsEfLnlGDuPpZucSET+lwqYiIiInOObbYl8E3Mcm9XC9JEd8XJzNjuSXKaOjWozokPRquXzGksvUqmogImIiEixhFM5PPVN0cj5B69oSueQ2iYnkpJ6ZGBL3JytbIo9xQ+7ks2OIyK/UwETERERAArtDiYujCEzr5BOjXx5oF9TsyNJKdT3deee3kVj6af+sFdj6UUqCRUwERERAWDWysNsjjtNLVcnpo/siJNNPyZUdff0aUygtysJp87y4dpYs+OICCpgIiIiAmyLP830nw8CMGV4GxrV8TA5kZQFDxcnHh1YNJb+7V8PcTIzz+REIqICJiIiUsNl5RUyYWEMdofBsPD6XNuxgdmRpAyN6NCA9g19yMorZNoKjaUXMZsKmIiISA337He7iUvLoYGvOy+MaIvFYjE7kpQhq9XC07+PpV/wWwJ7jmeYnEikZlMBExERqcEW7zjOl1uOYbXAGyM74OOukfPVUZdQP4a0r4dhaCy9iNlUwERERGqo42fO8sSinQDc17cp3cL8TE4k5emxgS1xcbKy/kgaK/acMDuOSI2lAiYiIlID2R0GExfGkJFbSHhDH8Zf1czsSFLOgv08uCsqDID/LN1LfqHD5EQiNZPpBWzmzJmEhYXh5uZG586diY6OvuC+SUlJjBkzhhYtWmC1WpkwYcJ5+xQUFDBlyhSaNGmCm5sb4eHhLFu27Jx9nn32WSwWyzmPoKCgsv7UREREKq05q4+w8egpPFxsTB/VEWeNnK8R7u3blAAvV2LTcpi7PtbsOCI1kqlfbRcuXMiECROYPHky27ZtIyoqikGDBhEfH/+X++fl5REQEMDkyZMJDw//y32efPJJZs+ezVtvvcWePXsYN24c1157Ldu2bTtnvzZt2pCUlFT82LlzZ5l/fiIiIpXRzmPpvP5j0TS8Z4e1Iczf0+REUlFquTrx7/4tAJjx80HSsjSWXqSimVrApk2bxh133MGdd95Jq1atmD59OsHBwcyaNesv9w8NDWXGjBnccsst+Pj4/OU+8+bN44knnmDw4ME0btyYe++9lwEDBvD666+fs5+TkxNBQUHFj4CAgDL//ERERCqbnPxCxi/YRqHDYFDbIG7s0tDsSFLBru/ckNb1vMnMLWT6TwfNjiNS45hWwPLz89myZQv9+/c/Z3v//v1Zt25diY+bl5eHm5vbOdvc3d1Zs2bNOdsOHjxI/fr1CQsLY9SoURw5cuRvj5uRkXHOQ0REpKp5fvFejqRmE+TtxtTr2mnkfA1ks1p46vex9J9ujOPAiUyTE4nULKYVsNTUVOx2O4GBgedsDwwMJDk5ucTHHTBgANOmTePgwYM4HA5WrFjBt99+S1JSUvE+3bt3Z+7cuSxfvpx3332X5ORkIiMjSUtLu+Bxp06dio+PT/EjODi4xBlFRETMsHx3MvM3xWOxwLSbwvH1cDE7kpgkokkdBrQJxKGx9CIVzvR33P75N2+GYZTqt3EzZsygWbNmtGzZEhcXFx544AFuv/12bDZb8T6DBg3i+uuvp127dlx11VUsWbIEgI8//viCx3388cdJT08vfiQkJJQ4o4iISEU7kZHLY1/tAODuqMZENvU3OZGY7YnBrXC2WYg+mMrK/SfNjiNSY5hWwPz9/bHZbOetdqWkpJy3KnY5AgIC+Oabb8jOziYuLo59+/ZRq1YtwsLCLvgcT09P2rVrx8GDF74O2tXVFW9v73MeIiIiVYHDYfDwF9s5nVNAm/re/Ov3IQxSs4XU8eSfPYt+PnphyR4K7BpLL1IRTCtgLi4udO7cmRUrVpyzfcWKFURGRpb6+G5ubjRo0IDCwkK++uorhg8ffsF98/Ly2Lt3L/Xq1Sv1eUVERCqbD9YeJfpgKm7OVmaM6oiLk+kXwEglcf8VTanj6cLhk9l8uiHO7DgiNYKpX4EnTZrEe++9xwcffMDevXuZOHEi8fHxjBs3Dii67O+WW2455zkxMTHExMSQlZXFyZMniYmJYc+ePcUf37hxI4sWLeLIkSNER0czcOBAHA4HjzzySPE+Dz/8MKtWreLo0aNs3LiRG264gYyMDG699daK+cRFREQqyO7j6byyrGjk/FNDW9O0bi2TE0ll4u3mzKT+zQF446eDnMnJNzmRSPXnZObJR44cSVpaGlOmTCEpKYm2bduydOlSQkJCgKIbL//5nmAdO3Ys/vOWLVv47LPPCAkJITY2FoDc3FyefPJJjhw5Qq1atRg8eDDz5s3D19e3+HnHjh1j9OjRpKamEhAQQI8ePdiwYUPxeUVERKqDs/l2xi+IId/u4KpWgYzp1sjsSFIJjewSzNx1cew/kcmMnw/yzLA2ZkcSqdYshsbelEhGRgY+Pj6kp6fr/WAiIlIpPf3tLuaujyPAy5Vl46OoU8vV7EhSSUUfPMnY9zfhZLWwfGJvmgRopVTkclxON9BF4CIiItXQz3tPMHd90Xt6XrsxXOVLLiqqWQBXtapLocPgP0v2mh1HpFpTARMREalmTmbm8ciXRSPn/9kzjD7NA0xOJFXBE4Nb4WS18PO+FKIPaiy9SHlRARMREalGDMPg319uJy07n5ZBXjwyUCPn5dI0DqjFLRGhALyweC+FGksvUi5UwERERKqRuevjWLn/JC5ORSPn3ZxtZkeSKmT8lc3w9XBm/4lMFvyWYHYckWpJBUxERKSaOHAikxeXFr1/54lBLWkR5GVyIqlqfDycmXhV0Vj6aSsOkH62wOREItWPCpiIiEg1kFtg56H528gvdNCneQC3RoaaHUmqqDHdG9EkwJNT2fm8/eshs+OIVDsqYCIiItXAq8v3sy85kzqeLrx2YzgWi8XsSFJFOdusPDm0NQAfrj1KbGq2yYlE/lp+oYOM3Kq3SqsCJiIiUsWtPnCS99ccBeDVG9sT4KWR81I6/VrUpU/zAArsBlN/0Fh6qXwSz5zlptnrGT9/Gw5H1bqtsQqYiIhIFZaWlce/vtgOwC0RIVzRMtDkRFJdPDmkFTarheW7T7DucKrZcUSK/bovhSFvRhOTcIYtcac5mla1VmlVwERERKoowzB49KudnMzMo2ndWjwxuJXZkaQaaRboxT+6NwKKxtLbq9gqg1Q/hXYHryzbx+0f/caZnALaN/RhyUNRNAmoZXa0y6ICJiIiUkV9timen/aewMVmZcaoDho5L2VuwlXN8XJzYk9SBl9u0Vh6Mc+JjFzGvLeRmSsPA3BrRAhfjIsg2M/D5GSXTwVMRESkCjqUksXzi/cA8MjAFrSp72NyIqmO/DxdGH9lMwBeXX6ArLxCkxNJTbT2UCpD3oxm09FTeLrYeGt0R54b3hZXp6r5SycVMBERkSomv9DB+AXbyC1w0KupP//sGWZ2JKnGbokIJczfk9SsPGZqLL1UILvDYMZPB7n5/Y2kZuXTMsiL7x/sxbDw+mZHKxUVMBERkSrm9RX72X08A18PZ16/KRyrVSPnpfy4OFmL31/43pqjJJzKMTmR1ASpWXnc9uEm3vjpAIYBI7sE8839PWlcxd7v9VdUwERERKqQdYdSmbP6CAAvXdeeQG83kxNJTXBVq7r0bFqH/EIHLy3bZ3YcqeY2HT3FkDejiT6YipuzldduDOflG9pXm/e5qoCJiIhUEWdy8pn0+XYMA0Z3C2Zg2yCzI0kNYbFYeHJIa6wWWLIjid9iT5kdSaohh8PgnVWHGf3uBk5k5NEkwJNv7+/FDZ0bmh2tTKmAiYiIVAGGYfD4op0kZ+TS2N+Tp4a2NjuS1DCt6nkzsmvRWPrnF++pcje/lcrtTE4+d8/bzEs/7MPuMBjeoT7fPdCLFkFeZkcrcypgIiIiVcAXW47xw65knKwWpo/qgIeLk9mRpAaadHVzark6seNYOl9vSzQ7jlQTMQlnGPLmGn7am4KLk5UXr23L9JEd8HStnl/nVMBEREQqudjUbJ79bjcA/+rfgvYNfc0NJDVWgJcrD1zRFIBXlu8jJ19j6aXkDMPgo7VHufGddSSeOUtIHQ8W3RvJP7qHYLFU3+FCKmAiIiKVWIHdwfiFMeTk2+nR2I+7ezc2O5LUcLf3DCXYz50TGXm8s+qI2XGkisrILeD+z7by7Pd7KLAbDGwTxPcP9qJtg+p/T0MVMBERkUpsxk8H2Z5wBm83J6bd1AGbRs6LyVydbDwxqGgs/ZzVhzl+5qzJiaSq2X08nWveWsPSnUWXVT89tDWzbu6Et5uz2dEqhAqYiIhIJbXxSBpvryy68e3U69pT39fd5EQiRQa2DaJbmB+5BQ5e0Vh6uUSGYTB/UzzXzlxHbFoODXzd+XxcBP/sFVatLzn8MxUwERGRSij9bEHxyPkbOjdkSPt6ZkcSKWaxFK1aWCzwTcxxtsWfNjuSVHLZeYVM+nw7jy/aSX6hg34tAlj8YC86NaptdrQKpwImIiJSyRiGwZPf7CLxzFka+Xnw7DVtzI4kcp62DXy4oVPR/ZmmLN6DYWgsvfy1gycyGf72Wr7elojNauHRgS15/9au1PZ0MTuaKVTAREREKplvYhL5fvtxbL+PnK9VTUcxS9X37wEt8HCxsS3+DN9tP252HKmEvt52jGv+u5ZDKVnU9XLlszu7c2/fJlhr8PtZVcBEREQqkYRTOTz1TdHI+fFXNquRl+dI1VHX2437+jYB4OUf9nE2325yIqkscgvsPL5oBxMXbudsgZ2eTeuwdHwU3RvXMTua6VTAREREKolCu4MJC2PIyiukS0jt4h9sRSqzO6Ma08DXnePpubwXrbH0AkdTs7l25jrmb0rAYin6ZdLcf3bHv5ar2dEqBRUwERGRSuLtXw+zJe40Xq5OvDGyA042fZuWys/N2cajg1oCMHPlYU5k5JqcSMy0ZEcSw95aw96kDOp4ujD3n92YeHVz3ULjf+gru4iISCWwJe40b/5yEIDnR7Ql2M/D5EQil25Y+3p0auTL2QI7ry7fb3YcMUFeoZ1nv9vN/Z9tJSuvkK6htVnyUBRRzQLMjlbpqICJiIiYLDO3gAkLt2F3GIzoUJ8RHRuYHUnkslgsFp4eVjSt88stx9h5LN3kRFKREk7lcNM76/loXSwA4/o0Yf5dPQjycTM3WCWlAiYiImKyZ7/bQ8KpszTwdWfKiLZmxxEpkQ7Bvlz7+y8PntdY+hrjpz0nGPrWGrYfS8fH3Zn3b+3CY4Na6hLqi9C/jIiIiIm+336cr7Yew2qB6aM64O3mbHYkkRJ7ZGAL3JytbIo9xQ+7ks2OI+WowO5g6g97uXPuZtLPFhAe7MuSh3pxZatAs6NVeipgIiIiJkk8c5Ynvt4JwAP9mtI11M/kRCKlU8/HnXt6F03v/M/SveQWaCx9dZScnsuYdzcwe1XR1Mvbe4byxT0RNKyt965eChUwERERE9gdBhMXxpCZW0iHYF8evLKZ2ZFEysQ9fRoT6O3KsdNn+XBtrNlxpIytPnCSwW9G81ts0cTWWf/oxDPD2uDipFpxqfQvJSIiYoJ3Vh1m09FTeLrYmDGqA856v4RUEx4uTjw6sGgs/du/HuJkZp7JiaQs2B0G01Yc4NYPN3EqO5/W9bz5/sFeDGpXz+xoVY6+2ouIiFSw7QlneGPFAQCevaYNIXU8TU4kUrZGdGhAeEMfsvIKmbZCY+mrupOZedzywUbe/PkghgGjuzVi0X2RhPrra1dJqICJiIhUoOy8QiYsjKHQYTCkXT1u6NzQ7EgiZc5qtfDU0NYALPgtgT3HM0xOJCW14Ugag9+MZu2hNDxcbEwf2YGp17XDzdlmdrQqSwVMRESkAj2/eA9HU7Op5+PGi9e2xWKxmB1JpFx0CfVjaPt6GIbG0ldFDofB278eYsy7GziZmUezurX47oGeuk9hGVABExERqSDLdiWx4LcELBZ4/aZwfD1czI4kUq4eG9QSFycr64+ksWLPCbPjyCU6nZ3PHR//xqvL9+Mw4LqODfj2gZ40retldrRqQQVMRESkAiSn5/LYoqKR8/f0bkJkE3+TE4mUv4a1PbgrKgwoGkufX+gwOZH8na3xpxnyZjS/7j+Jq5OVl69vx+s3hePh4mR2tGpDBUxERKScORwG//oihjM5BbRr4MOkq5ubHUmkwtzbtykBXq7EpuUwd32s2XHkAgzD4P01R7npnfUcT88lzN+Tr+/ryciujXSpdBlTARMRESln7685ytpDabg725g+qoPulyM1Si1XJ/7dvwUAM34+SFqWxtJXNulnCxj3yRaeX7yneEDQdw/0pHV9b7OjVUv6DiAiIlKOdh9P55Xl+wB4elhrmgTUMjmRSMW7vnND2tT3JjO3kOk/HTQ7jvyPXYnpDHtrDct3n8DZZuG5a9rw3zEd8XJzNjtataUCJiIiUk7O5tt5aP42CuwG/VsHMqprsNmRRExh+5+x9J9ujGN/cqbJicQwDD7ZEMd1s9YRfyqHBr7ufDkuklsjQ3XJYTlTARMRESknLy7dw+GT2dT1cuWl69vrhxqp0Xo0rsPANkE4DHhhicbSmykrr5DxC2J48ptd5Bc6uKpVXZY+FEV4sK/Z0WoEFTAREZFy8NOeE3yyIR4oGjnv56mR8yKPD26Ji81K9MFUVu4/aXacGml/cibX/HcN320/js1q4YnBLXn3li74eOiSw4qiAiYiIlLGUjJzeeSrHQDc2SuMqGYBJicSqRxC6nhye89QAJ5fsocCu8bSV6Qvtxxj+NtrOHIymyBvNxbe3YO7ezfR6nwFUwETEREpQw6HwcNf7OBUdj4tg7z498AWZkcSqVTuv6IpdTxdOHIym083xJkdp0Y4m2/nkS+38/AX28ktcBDVzJ8lD/WiS6if2dFqJBUwERGRMvTx+lhWHyi6gembozvi6mQzO5JIpeLt5syk/kX3wnvjp4Ocyck3OVH1dvhkFtfOXMvnm49htcC/rm7Ox7d3o04tV7Oj1VgqYCIiImVkX3IGU38oGjk/eUgrmgd6mZxIpHIa2SWYlkFepJ8tYMbPGktfXr7ffpxr3lrDvuRM/Gu58skd3XnwymZYrbrk0EwqYCIiImUgt8DO+Pkx5Bc6uKJlXcb2CDE7kkil5WSz8uSQorH089bHcSgly+RE1UteoZ2nvtnFg/O3kZ1vp3uYH0sf6kVkU3+zowkqYCIiImXi5WX72H8iE/9aLrxyg0bOi/ydXs38uapVXQodBv9ZutfsONVGfFoON8xaz7zf3193f78mfHpnd+p6u5mcTP5gegGbOXMmYWFhuLm50blzZ6Kjoy+4b1JSEmPGjKFFixZYrVYmTJhw3j4FBQVMmTKFJk2a4ObmRnh4OMuWLSvVeUVERC5m5f4UPlwbC8CrN4bjr/dWiFySJwa3wslq4Zd9Kaw+oLH0pfXj7mSGvBXNzsR0fD2c+fD2rvx7QEucbKb/yC//w9RXY+HChUyYMIHJkyezbds2oqKiGDRoEPHx8X+5f15eHgEBAUyePJnw8PC/3OfJJ59k9uzZvPXWW+zZs4dx48Zx7bXXsm3bthKfV0RE5ELSsvJ4+IuikfO3RYbSr0VdkxOJVB2NA2pxS0QoUHRz5kKNpS+RAruDFxbv4e55W8jMLaRTI1+WPhSlr0eVlMUw8Tbk3bt3p1OnTsyaNat4W6tWrRgxYgRTp0696HP79u1Lhw4dmD59+jnb69evz+TJk7n//vuLt40YMYJatWrxySeflPi8eXl55OXlFf89IyOD4OBg0tPT8fb2vuTPWUREqg/DMLjz4838vC+F5oG1+O6BXrg5a+qhyOVIzymgz2u/ciangBdGtOVmvX/yshw/c5YHPtvK1vgzQNG9Bx8d1BJnrXpVqIyMDHx8fC6pG5j2yuTn57Nlyxb69+9/zvb+/fuzbt26Eh83Ly8PN7dzr3F1d3dnzZo1pTrv1KlT8fHxKX4EBweXOKOIiFQP3+9I4ud9KbjYrMwY1VHlS6QEfDycmXhV0Vj6aSsOkH62wOREVcfK/SkMeTOarfFn8HJz4p2bO/Pk0NYqX5Wcaa9OamoqdrudwMDAc7YHBgaSnJxc4uMOGDCAadOmcfDgQRwOBytWrODbb78lKSmpVOd9/PHHSU9PL34kJCSUOKOIiFR9hmEw89dDANzXrwmt6ulqCJGSGtO9EU3r1uJUdj5v//7/lVxYod3Ba8v3c9uHv3E6p4C2DbxZ8mAUA9sGmR1NLoHp9fjPU6IMwyjV5KgZM2bQrFkzWrZsiYuLCw888AC33347Ntu5v5W83PO6urri7e19zkNERGqu1QdT2ZeciYeLjdsiQ82OI1KlOdusTB7SCoAP1x4lNjXb5ESVV0pGLje/v5H//l5Ux/YI4ctxkTSq42FyMrlUphUwf39/bDbbeatOKSkp561OXY6AgAC++eYbsrOziYuLY9++fdSqVYuwsLByPa+IiNQss1cdBmBU10b4eriYnEak6uvXoi59mgdQYNdY+gtZdziVwW+uYcORU3i62HhzdEeeH9FWlz9XMaYVMBcXFzp37syKFSvO2b5ixQoiIyNLfXw3NzcaNGhAYWEhX331FcOHD6+Q84qISPW381g66w6nYbNauCMqzOw4ItXGk0NaYbNa+HHPCdYdTjU7TqXhcBi89fNBbn5vI6lZebQM8uK7B3txTXh9s6NJCTiZefJJkyYxduxYunTpQkREBHPmzCE+Pp5x48YBRe+7SkxMZO7cucXPiYmJASArK4uTJ08SExODi4sLrVsX3U1948aNJCYm0qFDBxITE3n22WdxOBw88sgjl3xeERGRi5m9umj165rw+jTwdTc5jUj10SzQi390b8Tc9XE8v3gvix/shc1as29qfio7nwkLY4rvk3Zj54ZMGd4WdxetelVVphawkSNHkpaWxpQpU0hKSqJt27YsXbqUkJCi8aNJSUnn3ZurY8eOxX/esmULn332GSEhIcTGxgKQm5vLk08+yZEjR6hVqxaDBw9m3rx5+Pr6XvJ5RURELiQ+LYelO4sGO93du7HJaUSqnwlXNeebbYnsTcrgyy0JjOzayOxIptkce4oHPttGckYubs5Wpgxvy01dNIm7qjP1PmBV2eXM+hcRkerj6W93MXd9HH2aB/DxP7uZHUekWnov+ggvLNmLfy1XVv67L7VcTV0zqHCGYfBe9FFeWrYPu8OgcYAnM//RiZZB+pmzsqoS9wETERGpatKy8vh8c9FtSO7po9UvkfJyS0QoYf6epGblFd/uoaZIzyng7nlbeHHpXuwOg2Hh9fnugV4qX9WICpiIiMglmrs+jtwCB+0b+hDRuI7ZcUSqLRcnK08MLhpL/96aoyScyjE5UcXYcewMQ96KZsWeE7jYrLwwoi1vjupQ41YAqzsVMBERkUuQk1/I3PWxANzTu0mp7lkpIn/vqlZ16dm0DvmFDl76YZ/ZccqVYRjMXR/LDbPWc+z0WYL93Fl0XyQ39wjR15pqSAVMRETkEnyx+Rincwpo5OfBwLZBZscRqfYsFgtPDmmN1QJLdibxW+wpsyOVi8zcAh6Yv42nv91Nvt1B/9aBLH4wirYNfMyOJuVEBUxERORvFNodvBt9BIC7osJq/FhskYrSqp538RTEKd/vweGoXrPj9iZlcM1/17JkRxJOVgtPDW3N7LGd8XF3NjualCMVMBERkb+xdFcyx06fxc/ThRs6awS0SEX6V//m1HJ1YmdiOl9vSzQ7TpkwDIPPf0tgxNtrOZqaTT0fNxbeE8EdvcJ0yWENoAImIiJyEYZhMHtV0Y2Xb40I1c1PRSqYfy1XHriiKQCvLN9HTn6hyYlKJye/kIe/2MEjX+0gr9BB3xYBLHkois4htc2OJhVEBUxEROQi1h5KY/fxDNydbdwSEWJ2HJEa6faeoTTy8+BERh7vrDpidpwSO5SSyYi31/LV1mNYLfDvAS344Nau+Hm6mB1NKpAKmIiIyEXMXl20+jWyazC19UOSiClcnWw8PqglALNXHSbxzFmTE12+b2MSuea/azlwIosAL1c+u6sH9/drilXvKa1xVMBEREQuYPfxdKIPpmKzWrijV5jZcURqtIFtg+gW5kdeoYNXllWdsfS5BXae+Hon4xfEkJNvJ7JJHZY+FEUP3UuwxlIBExERuYA5q4sudRrSrh7Bfh4mpxGp2SwWC08PbY3FAt/GHGdr/GmzI/2tuLRsrpu5js82xmOxwENXNmPeHd0J8HI1O5qYSAVMRETkLyScymHxjiQA7u7d2OQ0IgLQtoEPN3RqCMDzi/dgGJV3LP2yXUkMfXMNe5Iy8PN04ePbuzHp6ua6jYWogImIiPyV99ccxe4wiGrmrxuiilQi/x7QAg8XG9viz/Dd9uNmxzlPfqGD577fzbhPtpKZV0iXkNoseagXvZsHmB1NKgkVMBERkT85nZ3Pwt8SALindxOT04jI/6rr7cZ9fYv+v3z5h32czbebnOj/JZ45y02z1/Ph2lgA7unTmPl396Cej7u5waRSUQETERH5k3kb4jhbYKdNfW96NtUb5UUqmzujGtPA153j6bm8F105xtL/su8EQ96MJibhDN5uTrx7SxceH9QKZ5t+3JZz6b8IERGR/5FbYOejdbEA3NOnCRaL3q8hUtm4Odt49Pex9DNXHuZERq5pWQrtDl5eto9/frSZMzkFhDf0YclDUVzdOtC0TFK5qYCJiIj8jy+2HONUdj4Na7szuG2Q2XFE5AKGta9Hp0a+nC2w88qy/aZkOJGRy5j3NjJrZdH9Am+LDOXzcRGamioXpQImIiLyO7vD4N3fR8/f2SsMJ106JFJpWSwWnh7WBoCvth5j57H0Cj3/moOpDJ4Rzaajp6jl6sTbYzrx7DVtcHWyVWgOqXr0nUVEROR3y3YlE38qB18PZ27qGmx2HBH5Gx2Cfbm2YwMApizeXSFj6e0Og+k/HWDsBxtJy86nVT1vvn+wF0Pa1yv3c0v1oAImIiICGIbBO6uKLiO6JSIUDxcnkxOJyKV4ZGAL3Jyt/BZ7mh92JZfruVKz8rjtw01M/+kghgGjuwXz9X2RhPl7lut5pXpRARMREQHWH0ljZ2I6rk5Wbo0IMTuOiFyiej7uxbeL+M/SveQWlM9Y+k1HTzHkzWiiD6bi7mxj2k3hTL2uPW7OuuRQLo8KmIiICDB7VdF7v27qEkydWq4mpxGRy3FPn8YEebtx7PTZ4ntwlRWHo2h1fPS7GziRkUfTurX49oGeXNepYZmeR2oOFTAREanx9iZlsOrASawWuDMqzOw4InKZPFyceGRgCwDe/vUQKZllM5b+TE4+d83dzEs/7MPuMLi2YwO+vb8nzQO9yuT4UjOpgImISI33x+TDQe3qEVJH7+UQqYpGdGhAeEMfsvIKmfbjgVIfb1v8aYa8uYaf96Xg4mRl6nXtmHZTOJ6uen+olI4KmIiI1GiJZ87y3fbjANzTu7HJaUSkpKxWC08NbQ3Aws0J7D5esrH0hmHw4dqj3DR7PYlnzhJax4Ov74tkdLdGujG7lAkVMBERqdE+WHOUQodBZJM6tG/oa3YcESmFLqF+DG1fD8OAFxbvveyx9Bm5Bdz36Vae+34PBXaDQW2D+O7BXrSp71NOiaUmKlEBS0hI4NixY8V/37RpExMmTGDOnDllFkxERKS8pecUMH9TPAD39GlichoRKQuPDWqJi5OV9UfSWLHnxCU/b/fxdK55aw0/7ErG2Wbh2WGtmfmPTni7OZdjWqmJSlTAxowZw6+//gpAcnIyV199NZs2beKJJ55gypQpZRpQRESkvHyyMY6cfDstg7zo3czf7DgiUgYa1vbgrt+H6fxn6V7yCx0X3d8wDOZviufameuITcuhga87X4yL5LaeYbrkUMpFiQrYrl276NatGwCff/45bdu2Zd26dXz22Wd89NFHZZlPRESkXOQW2Plw7VEAxvVpoh+0RKqRe/s2JcDLldi0HOauj73gftl5hUz6fDuPL9pJfqGDK1vWZclDvegQ7FthWaXmKVEBKygowNW16B4pP/30E9dccw0ALVu2JCkpqezSiYiIlJNFWxNJzcqnga87Q9rXMzuOiJShWq5O/Lt/0Vj6GT8fJC0r77x9Dp7IZPjba/l6WyI2q4XHBrXk3Vu64OvhUtFxpYYpUQFr06YN77zzDtHR0axYsYKBAwcCcPz4cerUqVOmAUVERMqa3WHwbnTR6Pl/9grD2aaZVCLVzfWdG9KmvjeZuYW88dO5Y+kXbT3GNf9dy6GULAK9XVlwdw/G9WmC1aqVcCl/JfqO8/LLLzN79mz69u3L6NGjCQ8PB+C7774rvjRRRESkslqxJ5mjqdn4uDszqmuw2XFEpBzY/mcs/Wcb49mfnElugZ3HvtrBpM+3c7bATq+m/ix5KIquoX4mp5WapER3kuvbty+pqalkZGRQu3bt4u133303Hh4eZRZORESkrBmGwaxVRatfY3uE6KaqItVYj8Z1GNgmiGW7k5n89U6y8grZl5yJxQITrmzOA1c0xaZVL6lgJfquc/bsWQzDKC5fcXFxfP3117Rq1YoBAwaUaUAREZGytOnoKbYnnMHFycqtkaFmxxGRcvb44Jb8si+FzXGnAfCv5cKMUR3p2VSTT8UcJboEcfjw4cydOxeAM2fO0L17d15//XVGjBjBrFmzyjSgiIhIWZq9umj164bODQnwcjU5jYiUt5A6nozr0xiAbmF+LHkoSuVLTFWiArZ161aioqIA+PLLLwkMDCQuLo65c+fy5ptvlmlAERGRsnLgRCa/7EvBYoG7ohqbHUdEKsjEq5vz87/6MP+uHgR6u5kdR2q4El2CmJOTg5eXFwA//vgj1113HVarlR49ehAXF1emAUVERMrKnN9Xvwa2CSLM39PkNCJSUSwWC00CapkdQwQo4QpY06ZN+eabb0hISGD58uX0798fgJSUFLy9vcs0oIiISFlISj/LtzGJANzdW6tfIiJijhIVsKeffpqHH36Y0NBQunXrRkREBFC0GtaxY8cyDSgiIlIWPlwbS4HdoHuYHx0b1f77J4iIiJSDEl2CeMMNN9CrVy+SkpKK7wEGcOWVV3LttdeWWTgREZGykH62gM82xgMwrk8Tk9OIiEhNVuKbnwQFBREUFMSxY8ewWCw0aNBAN2EWEZFK6bON8WTlFdIi0Iu+LQLMjiMiIjVYiS5BdDgcTJkyBR8fH0JCQmjUqBG+vr48//zzOByOss4oIiJSYnmFdj5YexQoeu+XxaKbroqIiHlKtAI2efJk3n//fV566SV69uyJYRisXbuWZ599ltzcXF588cWyzikiIlIi32xL5GRmHvV83BgWXt/sOCIiUsOVqIB9/PHHvPfee1xzzTXF28LDw2nQoAH33XefCpiIiFQKDodRfOPlf/YMw8WpRBd+iIiIlJkSfSc6deoULVu2PG97y5YtOXXqVKlDiYiIlIWf9p7gyMlsvNycGNUt2Ow4IiIiJStg4eHh/Pe//z1v+3//+1/at29f6lAiIiJl4Y/Vr5t7hODl5mxyGhERkRJegvjKK68wZMgQfvrpJyIiIrBYLKxbt46EhASWLl1a1hlFREQu2+bYU2yJO42LzcrtkaFmxxEREQFKuALWp08fDhw4wLXXXsuZM2c4deoU1113Hbt37+bDDz8s64wiIiKX7Z1VRatf13VqQF1vN5PTiIiIFLEYhmGU1cG2b99Op06dsNvtZXXISisjIwMfHx/S09Px9vY2O46IiPyPQymZXDVtNRYL/DSpD00CapkdSUREqrHL6QYaByUiItXOu6uL7vt1datAlS8REalUVMBERKRaOZGRy9fbEgG4p08Tk9OIiIicy/QCNnPmTMLCwnBzc6Nz585ER0dfcN+kpCTGjBlDixYtsFqtTJgw4S/3mz59Oi1atMDd3Z3g4GAmTpxIbm5u8cefffZZLBbLOY+goKCy/tRERMQEH66NJd/uoGtobTqH1DY7joiIyDkuawriddddd9GPnzlz5rJOvnDhQiZMmMDMmTPp2bMns2fPZtCgQezZs4dGjRqdt39eXh4BAQFMnjyZN9544y+P+emnn/LYY4/xwQcfEBkZyYEDB7jtttsAznlOmzZt+Omnn4r/brPZLiu7iIhUPpm5BXy6IQ6Ae3pr9UtERCqfyypgPj4+f/vxW2655ZKPN23aNO644w7uvPNOoGjlavny5cyaNYupU6eet39oaCgzZswA4IMPPvjLY65fv56ePXsyZsyY4ueMHj2aTZs2nbOfk5OTVr1ERKqZ+ZviycwrpGndWlzRsq7ZcURERM5zWQWsLEfM5+fns2XLFh577LFztvfv359169aV+Li9evXik08+YdOmTXTr1o0jR46wdOlSbr311nP2O3jwIPXr18fV1ZXu3bvzn//8h8aNG1/wuHl5eeTl5RX/PSMjo8QZRUSk7OUXOnh/TdHwjbt7N8ZqtZicSERE5HwluhFzWUhNTcVutxMYGHjO9sDAQJKTk0t83FGjRnHy5El69eqFYRgUFhZy7733nlP0unfvzty5c2nevDknTpzghRdeIDIykt27d1OnTp2/PO7UqVN57rnnSpxLRETK17cxiZzIyCPQ25XhHeqbHUdEROQvmT6Ew2I59zeUhmGct+1yrFy5khdffJGZM2eydetWFi1axOLFi3n++eeL9xk0aBDXX3897dq146qrrmLJkiUAfPzxxxc87uOPP056enrxIyEhocQZRUSkbDkcBnNWF914+Z89w3B10vt6RUSkcjJtBczf3x+bzXbealdKSsp5q2KX46mnnmLs2LHF7ytr164d2dnZ3H333UyePBmr9fzO6enpSbt27Th48OAFj+vq6oqrq2uJc4mISPn5dX8KB1OyqOXqxOju5w9xEhERqSxMWwFzcXGhc+fOrFix4pztK1asIDIyssTHzcnJOa9k2Ww2DMPAMIy/fE5eXh579+6lXr16JT6viIiYZ/aqotWvf3RvhLebs8lpRERELsy0FTCASZMmMXbsWLp06UJERARz5swhPj6ecePGAUWX/SUmJjJ37tzi58TExACQlZXFyZMniYmJwcXFhdatWwMwbNgwpk2bRseOHenevTuHDh3iqaee4pprrikeNf/www8zbNgwGjVqREpKCi+88AIZGRnnDeoQEZHKb0vcaTbFnsLZZuH2nmFmxxEREbkoUwvYyJEjSUtLY8qUKSQlJdG2bVuWLl1KSEgIUHTj5fj4+HOe07Fjx+I/b9myhc8++4yQkBBiY2MBePLJJ7FYLDz55JMkJiYSEBDAsGHDePHFF4ufd+zYMUaPHk1qaioBAQH06NGDDRs2FJ9XRESqjjmrDwMwokMDgnzcTE4jIiJycRbjQtflyUVlZGTg4+NDeno63t7eZscREamRjpzM4sppqzAMWDGxN80CvcyOJCIiNdDldAPTpyCKiIiU1LvRRzEMuKpVXZUvERGpElTARESkSkrJzOWrrccAuKdPE5PTiIiIXBoVMBERqZI+XhdLfqGDTo186RJS2+w4IiIil0QFTEREqpysvELmrY8Dila/LBaLyYlEREQujQqYiIhUOQs2xZORW0hjf0+ubhVodhwREZFLpgImIiJVSoHdwftrjgJwd+/GWK1a/RIRkapDBUxERKqU77cfJyk9lwAvV0Z0bGB2HBERkcuiAiYiIlWGYRjMXnUEgNt7huLmbDM5kYiIyOVRARMRkSpj5YGT7D+RiaeLjX90DzE7joiIyGVTARMRkSpj9qrDAIzu1ggfd2eT04iIiFw+FTAREakSYhLOsOHIKZysFv7ZK8zsOCIiIiWiAiYiIlXCnNVFq1/XdKhPfV93k9OIiIiUjAqYiIhUerGp2fywKxkoGj0vIiJSVamAiYhIpfdu9BEMA/q1CKBlkLfZcUREREpMBUxERCq11Kw8vtxyDIB7+jQxOY2IiEjpqICJiEilNnddLHmFDsKDfeke5md2HBERkVJRARMRkUorO6+Qj9fHATCud2MsFovJiUREREpHBUxERCqtzzcnkH62gNA6HvRvE2R2HBERkVJTARMRkUqpwO7gveijANzVuzE2q1a/RESk6lMBExGRSmnpziQSz5zFv5YL13dqaHYcERGRMqECJiIilY5hGLyz6ggAt0WG4uZsMzmRiIhI2VABExGRSif6YCp7kzLwcLFxc48Qs+OIiIiUGRUwERGpdGavPgzAyK7B+Hq4mJxGRESk7KiAiYhIpbLzWDprD6Vhs1q4o1eY2XFERETKlAqYiIhUKn+sfg1rX4+GtT1MTiMiIlK2VMBERKTSiE/LYenOJADu7t3E5DQiIiJlTwVMREQqjffWHMFhQO/mAbSu7212HBERkTKnAiYiIpXCqex8Pt+cAMC43o1NTiMiIlI+VMBERKRSmLs+ltwCB+0a+BDRpI7ZcURERMqFCpiIiJjubL6dj9fFAnBPn8ZYLBZzA4mIiJQTFTARETHdF1sSOJ1TQCM/Dwa2CTI7joiISLlRARMREVMV2h28G30EgLuiwnCy6VuTiIhUX/ouJyIipvphVzIJp87i5+nCDZ2DzY4jIiJSrlTARETENIZhFN94+daIUNxdbCYnEhERKV8qYCIiYpp1h9PYlZiBm7OVsREhZscREREpdypgIiJimndWFa1+jewSjJ+ni8lpREREyp8KmIiImGL38XSiD6ZitcCdUbrxsoiI1AwqYCIiYoo5q4smHw5pX59gPw+T04iIiFQMFTAREalwCadyWLwjCYB7emv1S0REag4VMBERqXDvrzmK3WHQq6k/bRv4mB1HRESkwqiAiYhIhTqdnc/C3xIAuKePVr9ERKRmUQETEZEK9cmGOM4W2Gldz5teTf3NjiMiIlKhVMBERKTC5BbY+WhdLFC0+mWxWMwNJCIiUsFUwEREpMJ8ueUYadn5NPB1Z0i7embHERERqXAqYCIiUiHsDoN3o4tGz98VFYaTTd+CRESk5tF3PxERqRDLdycTl5aDr4czN3UNNjuOiIiIKVTARESk3BmGwexVhwG4JSIUDxcnkxOJiIiYQwVMRETK3YYjp9h+LB1XJyu3RoSYHUdERMQ0KmAiIlLuZq8uWv26sUtD6tRyNTmNiIiIeVTARESkXO1NymDl/pNYLXBnL914WUREajYVMBERKVfvri6afDiobT1C/T1NTiMiImIuFTARESk3iWfO8t324wDc3VurXyIiIqYXsJkzZxIWFoabmxudO3cmOjr6gvsmJSUxZswYWrRogdVqZcKECX+53/Tp02nRogXu7u4EBwczceJEcnNzS3xeEREpmQ/WHKXQYRDRuA7hwb5mxxERETGdqQVs4cKFTJgwgcmTJ7Nt2zaioqIYNGgQ8fHxf7l/Xl4eAQEBTJ48mfDw8L/c59NPP+Wxxx7jmWeeYe/evbz//vssXLiQxx9/vMTnFRGRy5eeU8D8TUVfV+/po9UvERERAIthGIZZJ+/evTudOnVi1qxZxdtatWrFiBEjmDp16kWf27dvXzp06MD06dPP2f7AAw+wd+9efv755+Jt//rXv9i0aVPxKldpzvuHjIwMfHx8SE9Px9vb+5KeIyJSk7z96yFeXb6flkFe/DA+CovFYnYkERGRcnE53cC0FbD8/Hy2bNlC//79z9nev39/1q1bV+Lj9urViy1btrBp0yYAjhw5wtKlSxkyZEipzpuXl0dGRsY5j8piS9wpPlp7lIzcArOjiIgAkFtg58O1sUDR6pfKl4iISBEns06cmpqK3W4nMDDwnO2BgYEkJyeX+LijRo3i5MmT9OrVC8MwKCws5N577+Wxxx4r1XmnTp3Kc889V+Jc5Wn2qiP8uOcEryzfz4iODbglIoSWQVqVExHzfL0tkdSsPOr7uDG0fX2z44iIiFQapg/h+PNvRQ3DKNVvSleuXMmLL77IzJkz2bp1K4sWLWLx4sU8//zzpTrv448/Tnp6evEjISGhxBnLWu/mATStW4ucfDufbYxn4PRobnxnHd9tP05+ocPseCJSw9gdRvHo+TuiGuNsM/1bjYiISKVh2gqYv78/NpvtvFWnlJSU81anLsdTTz3F2LFjufPOOwFo164d2dnZ3H333UyePLnE53V1dcXV1bXEucrTzT1C+Ef3Rmw4cop5G2JZvvsEv8We5rfY0/jXcmV0t2BGd2tEfV93s6OKSA2wYs8JjqRm4+PuzKiuwWbHERERqVRM+7Wki4sLnTt3ZsWKFedsX7FiBZGRkSU+bk5ODlbruZ+WzWbDMAwMwyi385rNYrEQ0aQOM//RmbWPXsH4K5tR18uV1Kw83vrlEFGv/Mo98zaz9lAqJs5dEZFqzjAM3ll1GICxPULwdDXt93wiIiKVkqnfGSdNmsTYsWPp0qULERERzJkzh/j4eMaNGwcUXfaXmJjI3Llzi58TExMDQFZWFidPniQmJgYXFxdat24NwLBhw5g2bRodO3ake/fuHDp0iKeeeoprrrkGm812Seet6oJ83Jh4dXMeuKIpP+4+wdz1sWw8eorlu0+wfPcJGgd4MrZHCNd1aoiPu7PZcUWkGvkt9jQxCWdwcbJya2So2XFEREQqHVML2MiRI0lLS2PKlCkkJSXRtm1bli5dSkhICFB04+U/35urY8eOxX/esmULn332GSEhIcTGxgLw5JNPYrFYePLJJ0lMTCQgIIBhw4bx4osvXvJ5qwtnm5Uh7esxpH09DpzIZN76OBZtPcaRk9k89/0eXllWNLRjbI8QWtfX0A4RKb3Zv69+Xd+pIQFelfOybRERETOZeh+wqqyq3gcsK6+Qr7ceY+76OA6mZBVv7xJSm7ERIQxqWw8XJ71hXkQu34ETmfR/YzUWC/w8qQ+NA2qZHUlERKRCXE430MX5NUwtVyfGRoRyc48QNh49xbz1cSzfnczmuNNsjjvN87X2MKprI0Z3b0QDDe0Qkcsw5/fJhwNaB6l8iYiIXIBWwEqoqq6A/ZUTGbnM3xTP/E3xnMjIA8BqgataBTI2IoSeTfyxWnUTVRG5sKT0s/R+5VcK7AZf3xdJx0a1zY4kIiJSYbQCJpcl0NuNCVc15/5+TVmx5wTz1sex/kgaP+45wY97TtDY35N/9Ajhhs4a2iEif+3DtbEU2A26hfmpfImIiFyEVsBKqDqtgP2VgycymbchjkVbE8nKKwTA3dnGiI71ublHCG3q+5icUEQqi4zcAiKn/kJWXiEf3NaFK1qW/F6OIiIiVdHldAMVsBKq7gXsD1l5hXy9LZF562M5cOL/h3Z0DqnN2B4hDGoXhKuTzcSEImK2d1Yd5qUf9tE8sBbLxvfWJcsiIlLjqIBVgJpSwP5gGAabjp5i3oY4lu1KptBR9J9NHU8XRnYN5h89QjS0Q6QGyiu0E/Xyr6Rk5vHajeHc0Lmh2ZFEREQqnN4DJmXOYrHQvXEdujeuQ0pGLgt+S+CzjfEkZ+Qyc+Vh3ll1mCtbBTK2Rwi9mmpoh0hN8e2246Rk5hHk7cY14fXNjiMiIlLpaQWshGraCthfKbA7+GnPCeZtiGPd4bTi7WH+nvyjeyNu7ByMj4eGdohUVw6HwdVvrOLwyWwmD27FXb0bmx1JRETEFLoEsQKogJ3rUEomn2yI56stx8j8fWiHm7OV4eENGBsRQtsGGtohUt2s2HOCu+ZuxsvNiXWPXYGXm37hIiIiNZMKWAVQAftr2XmFfBOTyLz1cexLzize3rGRL7dEhDC4XT0N7RCpJm6YtY7Ncae5t28THh3Y0uw4IiIiplEBqwAqYBdnGAab404zd30cy3YlUWAv+s/M74+hHd0b0bC2h8kpRaSkNsee4oZ31uNis7Lm0X7U9XYzO5KIiIhpNIRDTGexWOga6kfXUD9SMluxcFMCn22KJyk9l1krDzN71WGuaFmXsRGhRGloh0iVM3v1EQCu7dhA5UtEROQyaAWshLQCdvkK7Q5+2pvCvA2xrD30/0M7Qut4cHOPEG7o3BBfDxcTE4rIpTiUksVV01YB8NOkPjStW8vkRCIiIubSJYgVQAWsdA6lZPHJhrhzhna4OlkZ3qE+Y3uE0q6hhnaIVFaPfrmDhZsTuLp1IO/e0sXsOCIiIqZTAasAKmBlIye/kG+2HWfu+thzhnZ0CPZlbI8QhrSvh5uzhnaIVBYnMnKJevlX8u0Ovro3gs4hfmZHEhERMZ0KWAVQAStbhmGwJe408zbEsXTnuUM7bupSNLQj2E9DO0TM9tIP+3hn1WG6hNTmy3sjzY4jIiJSKaiAVQAVsPJzMjOPhb/F89nGeI6n5wJgscAVLepyc0QIfZoFaGiHiAkycwuInPoLmXmFvHtLF65uHWh2JBERkUpBUxClSgvwcuWBK5oxrk8Tft6Xwicb4og+mMrP+1L4eV8Kjfw8uLlHI27qEqyhHSIVaMGmBDLzCmkS4MmVLeuaHUdERKRK0gpYCWkFrGIdPpnFpxvi+WJLApm5/z+0Y1h4fW6JCKF9Q19zA4pUc/mFDnq/8ivJGbm8cn17buoabHYkERGRSkOXIFYAFTBz5OQX8m3Mceauj2NvUkbx9vDfh3YM1dAOkXLx5ZZjPPzFdup6uRL9aD9cnfT/mYiIyB9UwCqACpi5DMNga/xp5q2PY+nOZPLtDgBqezhzU5dgbu4RoqEdImXE4TAYOGM1B05k8diglozr08TsSCIiIpWKClgFUAGrPFKz8lj4WwKfbYwn8cxZoGhoR9/mAdwSEUqf5hraIVIav+w7wT8/2kwtVyfWPX4F3m7OZkcSERGpVDSEQ2oU/1qu3N+vKeP6NOGXfSnMXR9L9MFUft1/kl/3nyTYz52bu4dwU5dgantqaIfI5Xpn1REA/tG9kcqXiIhIKWkFrIS0Ala5HTmZxacb4/licwIZvw/tcHGyMqx90dCO8GBfcwOKVBFb409z3cx1ONssRD9yBUE+bmZHEhERqXR0CWIFUAGrGs7m2/lueyJz18ex+/j/D+1o39CHsT1CGBZeX0M7RC5i3LwtLNudzA2dG/LajeFmxxEREamUVMAqgApY1WIYBtsSzjBvfRxLdiQVD+3w/WNoR/cQGtXR0A6R/3XkZBZXTluFYcCPE3vTPNDL7EgiIiKVkgpYBVABq7pSs/L4fHMCn244d2hHn+YB3BIRQp/mdbFpaIcIjy/ayfxN8VzZsi7v39bV7DgiIiKVlgpYBVABq/rsDoNf96Uwd0Mcqw+cLN4e7OfOP34f2uGnoR1SQ6Vk5tLr5V/JL3Tw+T0RdAvzMzuSiIhIpaUCVgFUwKqX2NRsPtkQx+d/GtoxtH09bokIJbyhDxaLVsWk5nh1+T7e/vUwHRv5sujeSP33LyIichEqYBVABax6Optv5/vtx5m7IZZdif8/tKNdAx/GRoRwjYZ2SA2QlVdI5NSfycgt5J2bOzOwbZDZkURERCo1FbAKoAJWvRmGQczvQzsW/8/QDh93Z27q0pB/dA8h1N/T5JQi5eO96CO8sGQvjf09WTGpj94TKSIi8jdUwCqACljNkZaVx+ebj/HJhrjioR1QNLRjbI8Q+rXU0A6pPgrsDvq88ivH03OZel07RndrZHYkERGRSk8FrAKogNU8dofByv0pzF0fx6r/GdrRsPYfQzsaUqeWq4kJRUrv623HmLhwO/61XFnzaD9dcisiInIJVMAqgApYzRaX9sfQjmOkny0AwMVWNLTj5ogQOgb7amiBVDmGYTBoRjT7kjP594AW3N+vqdmRREREqgQVsAqgAiYAuQV2vtt+nHnr49iZmF68vW0Db8b2COGa8Aa4u2gFQaqGlftTuO3D3/B0sbHusSvx8XA2O5KIiEiVoAJWAVTA5M+2J5xh7vo4vt9xnPzC/x/acWPnhvyjRwhhGtohldzoORtYfySNO3uF8eTQ1mbHERERqTJUwCqACphcyKnsfL7YnMAnG+NIOPX/Qzt6/z604woN7ZBKaHvCGYa/vRYnq4XVj/Sjvq+72ZFERESqjMvpBk4VlEmkxvDzdOGePk24M6oxqw+cZO76WFYeOMnq3x8NfN0Z070Ro7oGa2iHVBpzVh8B4Jrw+ipfIiIi5UgrYCWkFTC5HPFpOXy6MY6FmxM4k/P/QzsGtwtibEQonRppaIeYJzY1myteX4nDgGUTomgZpK9pIiIil0OXIFYAFTApidwCO4t3JDFvfSzbj/3/0I429YuGdgzvoKEdUvGe/GYnn2yIp2+LAD66vZvZcURERKocFbAKoAImpbU94QzzNsTx/fbj5P0+tMPbzYkbOgdzc49GNA6oZXJCqQlSs/Lo+dIv5BU6mH9XDyKa1DE7koiISJWjAlYBVMCkrJzOzueLLQl8siGe+FM5xdujmvlzf7+m9GisH4il/Ez7cT9v/nKI8IY+fHN/T10KKyIiUgKX0w2sFZRJRC6gtqcLd/duwsqH+/Lh7V25omVdLBaIPpjKqDkbeObbXZzNt5sdU6qh7LxCPl4fB8A9fZqofImIiFQAFTCRSsJqtdCvRV0+uK0rq//dj9HdggH4eH0cg9+MZmv8aZMTSnXz+eYE0s8WEFLHgwFtgsyOIyIiUiOogIlUQsF+Hky9rj1z/9mNIG83jqZmc8OsdbyybB95hVoNk9IrtDt4L/ooAHdFNda96URERCqICphIJda7eQDLJ/Tm2o4NcBgwc+Vhhv93LXuOZ5gdTaq4JTuTSDxzljqeLtzQuaHZcURERGoMFTCRSs7Hw5k3RnbgnZs74efpwr7kTIa/vYa3fz1Eod1hdjypggzDYPaqohsv3xYZipuzbn0gIiJSUVTARKqIgW3rsXxCb65uHUiB3eDV5fu5cfZ6jpzMMjuaVDFrDqWyJykDd2cbYyNCzI4jIiJSo6iAiVQhAV6uzBnbmddvDMfL1Ylt8WcY/GY0H609isOhO0rIpflj9WtUt2B8PVxMTiMiIlKzqICJVDEWi4XrOzdk+cTe9GrqT26Bg2e/38PN728k8cxZs+NJJbcrMZ01h1KxWS3c0SvM7DgiIiI1jgqYSBVV39eduf/sxvPD2+DubGPd4TQGvrGaLzYnoPury4XMXl20+jWsfT0a1vYwOY2IiEjNowImUoVZrRbGRoSydHwUnRr5kplXyL+/3MFdc7eQkplrdjypZBJO5bBkx3EA7u7dxOQ0IiIiNZPpBWzmzJmEhYXh5uZG586diY6OvuC+SUlJjBkzhhYtWmC1WpkwYcJ5+/Tt2xeLxXLeY8iQIcX7PPvss+d9PChINyGVqivM35MvxkXy6MCWuNis/LT3BAPeWM3SnUlmR5NK5L3oIzgMiGrmT+v63mbHERERqZFMLWALFy5kwoQJTJ48mW3bthEVFcWgQYOIj4//y/3z8vIICAhg8uTJhIeH/+U+ixYtIikpqfixa9cubDYbN9544zn7tWnT5pz9du7cWeafn0hFslkt3Nu3Cd892JNW9bw5nVPAfZ9uZfyCbZzJyTc7npjsVHY+CzcnADCuj1a/REREzGJqAZs2bRp33HEHd955J61atWL69OkEBwcza9asv9w/NDSUGTNmcMstt+Dj4/OX+/j5+REUFFT8WLFiBR4eHucVMCcnp3P2CwgIKPPPT8QMLYO8+fb+njzQrylWC3wbc5wB01fz6/4Us6OJieaujyW3wEHbBt5ENqljdhwREZEay7QClp+fz5YtW+jfv/852/v378+6devK7Dzvv/8+o0aNwtPT85ztBw8epH79+oSFhTFq1CiOHDly0ePk5eWRkZFxzkOksnJxsvLwgBZ8dW8kjQM8OZGRx+0f/sbji3aSlVdodjypYGfz7Xy8LhaAe3o3wWKxmBtIRESkBjOtgKWmpmK32wkMDDxne2BgIMnJyWVyjk2bNrFr1y7uvPPOc7Z3796duXPnsnz5ct59912Sk5OJjIwkLS3tgseaOnUqPj4+xY/g4OAyyShSnjo2qs2SB6O4vWcoAPM3xTNoxmo2HT1lbjCpUF9sSeB0TgHBfu4Maqv3u4qIiJjJ9CEcf/5NrGEYZfbb2ffff5+2bdvSrVu3c7YPGjSI66+/nnbt2nHVVVexZMkSAD7++OMLHuvxxx8nPT29+JGQkFAmGUXKm7uLjWeGteGzu7rTwNedhFNnGTlnPS8u2UNugd3seFLOCu0O3os+CsBdUY1xspn+ZV9ERKRGM+07sb+/Pzab7bzVrpSUlPNWxUoiJyeHBQsWnLf69Vc8PT1p164dBw8evOA+rq6ueHt7n/MQqUoim/izbEIUI7sEYxjwbvRRhr61hh3HzpgdTcrRst3JxJ/KobaHMzd21sq9iIiI2UwrYC4uLnTu3JkVK1acs33FihVERkaW+viff/45eXl53HzzzX+7b15eHnv37qVevXqlPq9IZebl5szLN7Tn/Vu74F/LlUMpWVw7cx1vrDhAgd1hdjwpY4ZhMHtV0ftbb40Mxd3FZnIiERERMfValEmTJvHee+/xwQcfsHfvXiZOnEh8fDzjxo0Dii77u+WWW855TkxMDDExMWRlZXHy5EliYmLYs2fPecd+//33GTFiBHXqnD/t6+GHH2bVqlUcPXqUjRs3csMNN5CRkcGtt95aPp+oSCVzZatAVkzszZD29bA7DGb8fJDrZq7j4IlMs6NJGVp/OI2diem4OVu5JSLU7DgiIiICOJl58pEjR5KWlsaUKVNISkqibdu2LF26lJCQEKDoxst/vidYx44di/+8ZcsWPvvsM0JCQoiNjS3efuDAAdasWcOPP/74l+c9duwYo0ePJjU1lYCAAHr06MGGDRuKzytSE9T2dOHtMZ0Y0OY4T32zi52J6Qx5aw0P92/OHb0aY7NqUl5V987qotWvkV2C8fN0MTmNiIiIAFgMwzDMDlEVZWRk4OPjQ3p6ut4PJlXeiYxcHv1qByv3nwSgW6gfr90YTqM6HiYnk5LaczyDwW9GY7XAqn/3I9hPr6WIiEh5uZxuoHFYIkKgtxsf3taVl65rh6eLjU2xpxg4YzWfboxDv6OpmuasPgzAkPb1Vb5EREQqERUwEQGKbgkxqlsjlk3oTfcwP3Ly7Uz+ehe3ffgbyem5ZseTy3DsdA7f70gC4J7ejU1OIyIiIv9LBUxEzhHs58H8u3rw5JBWuDhZWXXgJP3fWMW3MYlaDasi3l9zFLvDoGfTOrRt4GN2HBEREfkfKmAich6r1cKdUY1Z+lAvwhv6kJFbyPgFMdz36VbSsvLMjicXcTo7nwWbim4Uf0/vJianERERkT9TARORC2pa14uv7o1k0tXNcbJa+GFXMgOmr2bFnhNmR5ML+GRDHGcL7LSq501UM3+z44iIiMifqICJyEU52aw8dGUzvrm/J80Da5Galc9dczfz8BfbycgtMDue/I/cAjsfrYsFYFyfxlgsupWAiIhIZaMCJiKXpG0DH757oBf39G6MxQJfbjnGwDdWs/ZQqtnR5HdfbjlGWnY+DXzdGdyuntlxRERE5C+ogInIJXNztvH44FZ8fk8Ejfw8OJ6eyz/e28gz3+7ibL7d7Hg1mt1h8G500Y2X74wKw9mmL+8iIiKVkb5Di8hl6xrqxw/jo7i5RyMAPl4fx+A3o9kaf9rkZDXXj7uTiUvLwdfDmZFdg82OIyIiIhegAiYiJeLp6sQLI9ox95/dCPJ242hqNjfMWscry/aRV6jVsIpkGAbvrCq68fItPULwcHEyOZGIiIhciAqYiJRK7+YBLJ/Qm+s6NsBhwMyVhxn+37XsOZ5hdrQaY+PRU2w/lo6rk5VbIkPNjiMiIiIXoQImIqXm4+HMtJEdeOfmTtTxdGFfcibD317D278eotDuMDtetTf799WvG7s0xL+Wq8lpRERE5GJUwESkzAxsW4/lE3vTv3UgBXaDV5fv58bZ6zlyMsvsaNXWvuQMft1/EqsF7uzV2Ow4IiIi8jdUwESkTPnXcmX22M68fmM4Xq5ObIs/w+A3o/lo7VEcDsPseNXOnNVFkw8Hta1HqL+nyWlERETk76iAiUiZs1gsXN+5Icsn9qZXU39yCxw8+/0ebn5/I8dO55gdr9o4fuYs38UcB+Du3lr9EhERqQpUwESk3NT3dWfuP7vx/PA2uDvbWHc4jYHTo/l8cwKGodWw0vpgzVEKHQY9GvsRHuxrdhwRERG5BCpgIlKurFYLYyNCWTo+is4htcnKK+SRL3dw19wtpGTmmh2vykrPKWD+pngA7unTxOQ0IiIicqlUwESkQoT5e/L5PRE8OrAlLjYrP+09wYA3VrN0Z5LZ0aqkTzbGkZ1vp0WgF32bB5gdR0RERC6RCpiIVBib1cK9fZvw3YM9aV3Pm9M5Bdz36VbGL9jGmZx8s+NVGbkFdj5cGwvAPX0aY7FYzA0kIiIil0wFTEQqXMsgb765vycPXtEUqwW+jTnOgOmr+XV/itnRqoSvtyWSmpVHfR83hoXXNzuOiIiIXAYVMBExhYuTlX/1b8FX90bSOMCTExl53P7hbzy+aCdZeYVmx6u0HA6Dd38fPf/PXmE42/RlXEREpCrRd24RMVXHRrVZ8mAUt/cMBWD+pngGzVjNxiNp5garpFbsPcGR1Gy83ZwY1a2R2XFERETkMqmAiYjp3F1sPDOsDZ/d1Z0Gvu4knDrLqHc38MLiPeQW2M2OV2kYhsE7qw4DMDYihFquTiYnEhERkculAiYilUZkE3+WTYhiZJdgDAPeW3OUoW+tYcexM2ZHqxQ2x51mW/wZXJys3BoZanYcERERKQEVMBGpVLzcnHn5hva8f2sXArxcOZSSxbUz1/HGigMU2B1mxzPV7N9Xv67v1JC6Xm4mpxEREZGSUAETkUrpylaB/DihN0Pa18PuMJjx80Gum7mOgycyzY5mioMnMvlpbwoWC9wVFWZ2HBERESkhFTARqbRqe7rw9phOvDW6I74ezuxMTGfIW2uYs/owdodhdrwKNef3yYcDWgfROKCWyWlERESkpFTARKTSGxZenx8n9KZfiwDyCx38Z+k+Rs/ZQHxajtnRKkRyei7fxCQCcHefxianERERkdJQARORKqGutxsf3NaVl65rh6eLjU2xpxg4YzWfbozDMKr3atiHa49SYDfoFupHp0a1zY4jIiIipaACJiJVhsViYVS3Riyb0JvuYX7k5NuZ/PUubvvwN5LTc82OVy4ycgv4dGM8APdo9UtERKTKUwETkSon2M+D+Xf14MkhrXBxsrLqwEn6v7GKb2MSq91q2Gcb48nKK6RZ3Vr0a1HX7DgiIiJSSipgIlIlWa0W7oxqzNKHehHe0IeM3ELGL4jhvk+3kpaVZ3a8MpFXaOeDNUcBuLt3Y6xWi8mJREREpLRUwESkSmta14uv7o1k0tXNcbJa+GFXMgOmr2bFnhNmRyu1b7cdJyUzj0BvV4Z3aGB2HBERESkDKmAiUuU52aw8dGUzvrm/J80Da5Galc9dczfz8BfbycgtMDteiTgcBrNXF914+Y5eYbg46cu1iIhIdaDv6CJSbbRt4MP3D/binj6NsVjgyy3HGPjGatYeSjU72mX7ZV8Kh09m4+XqxOhujcyOIyIiImVEBUxEqhVXJxuPD2rFF/dEEFLHg+PpufzjvY088+0uzubbzY53yf5Y/fpHjxC83JxNTiMiIiJlRQVMRKqlLqF+LH0oipt7FK0efbw+jsFvRrM1/rTJyf7elrhT/BZ7Ghebldt7hpodR0RERMqQCpiIVFuerk68MKIdc//ZjSBvN46mZnPDrHW8smwfeYWVdzVs9qojAFzbsQGB3m4mpxEREZGypAImItVe7+YBLJ/Qm+s6NsBhwMyVhxn+37XsOZ5hdrTzHErJYsXeogmOd/XWjZdFRESqGxUwEakRfDycmTayA+/c3Ik6ni7sS85k+NtrePvXQxTaHWbHK/Ze9BEMA65uHUjTurXMjiMiIiJlTAVMRGqUgW3rsXxib/q3DqTAbvDq8v3cOHs9R05mmR2NlIxcFm1NBGBcH61+iYiIVEcqYCJS4/jXcmX22M5MuykcLzcntsWfYfCb0Xy09igOh2Farg/XxZJvd9A5pDadQ/xMyyEiIiLlRwVMRGoki8XCdZ0asnxCb6Ka+ZNb4ODZ7/dw8/sbOXY6p8LzZOYW8MmGOADu0Xu/REREqi0VMBGp0er7ujP3n914fngb3J1trDucxsDp0Xy+OQHDqLjVsAWbEsjMLaRxgCdXtQqssPOKiIhIxVIBE5Eaz2KxMDYilKXjo+gcUpusvEIe+XIHd83dQkpmbrmfP7/QwftrjgJFq19Wq6XczykiIiLmUAETEfldmL8nn98TwaMDW+Jis/LT3hMMeGM1S3cmlet5v9t+nOSMXAK8XBnRsUG5nktERETMpQImIvI/bFYL9/ZtwncP9qR1PW9O5xRw36dbGb9gG2dy8sv8fA6HwZzVhwH4Z88wXJ1sZX4OERERqTxUwERE/kLLIG++ub8nD17RFJvVwrcxxxkwfTW/7k8p0/OsPJDCgRNZ1HJ1Ykz3RmV6bBEREal8VMBERC7AxcnKv/q34Kt7I2kc4MmJjDxu//A3Hl+0k6y8wjI5x+xVRwAY070RPu7OZXJMERERqbxUwERE/kaHYF+WPhTFP3uGATB/UzyDZqxm45G0Uh13W/xpNh49hbPNwu09Q8sgqYiIiFR2KmAiIpfAzdnG08NaM/+uHjTwdSfh1FlGvbuBFxbvIbfAXqJjzlldtPo1vEMD6vm4l2VcERERqaRML2AzZ84kLCwMNzc3OnfuTHR09AX3TUpKYsyYMbRo0QKr1cqECRPO26dv375YLJbzHkOGDCnxeUVE/hDRpA7LJkQxskswhgHvrTnK0LfWsOPYmcs6ztHUbJbtTgbgbt14WUREpMYwtYAtXLiQCRMmMHnyZLZt20ZUVBSDBg0iPj7+L/fPy8sjICCAyZMnEx4e/pf7LFq0iKSkpOLHrl27sNls3HjjjSU+r4jI//Jyc+blG9rz/q1dCPBy5VBKFtfOXMcbKw5QYHdc0jHejT6CYcCVLevSPNCrnBOLiIhIZWExDMMw6+Tdu3enU6dOzJo1q3hbq1atGDFiBFOnTr3oc/v27UuHDh2YPn36RfebPn06Tz/9NElJSXh6epb6vH/IyMjAx8eH9PR0vL29L+k5IlL9nM7O58lvd7FkR9G9wto18GHaTeE0u0ipOpmZR8+XfyG/0MHn90TQLcyvouKKiIhIObicbmDaClh+fj5btmyhf//+52zv378/69atK7PzvP/++4waNaq4fJX0vHl5eWRkZJzzEBGp7enC22M68dbojvh6OLMzMZ0hb61hzurD2B1//futj9fFkl/ooEOwL11Da1dwYhERETGTaQUsNTUVu91OYGDgOdsDAwNJTk4uk3Ns2rSJXbt2ceedd5b6vFOnTsXHx6f4ERwcXCYZRaR6GBZenx8n9KZfiwDyCx38Z+k+Rs/ZQHxazjn7ZecVMnd9LADj+jTGYrGYkFZERETMYvoQjj//8GEYRpn9QPL+++/Ttm1bunXrVurzPv7446Snpxc/EhISyiSjiFQfdb3d+OC2rrx0XTs8XWxsij3FwBmr+XRjHH9c7b3gtwQycgsJ8/fk6tZBJicWERGRimZaAfP398dms5236pSSknLe6lRJ5OTksGDBgnNWv0pzXldXV7y9vc95iIj8mcViYVS3Riyb0JvuYX7k5NuZ/PUubv3wNxJO5fB+dNHo+buiGmOzavVLRESkpjGtgLm4uNC5c2dWrFhxzvYVK1YQGRlZ6uN//vnn5OXlcfPNN1foeUVEAIL9PJh/Vw+eGtoaVycrqw+c5IrXV3I8PRf/Wi5c16mB2RFFRETEBE5mnnzSpEmMHTuWLl26EBERwZw5c4iPj2fcuHFA0WV/iYmJzJ07t/g5MTExAGRlZXHy5EliYmJwcXGhdevW5xz7/fffZ8SIEdSpU+eyzysiUhasVgt39AqjT3N//vX5drYfSwfg9p5huDnbTE4nIiIiZjC1gI0cOZK0tDSmTJlCUlISbdu2ZenSpYSEhABFN17+8725OnbsWPznLVu28NlnnxESEkJsbGzx9gMHDrBmzRp+/PHHEp1XRKQsNa3rxVf3RvLB2qMcTc3mtshQsyOJiIiISUy9D1hVpvuAiYiIiIgIVJH7gImIiIiIiNQ0KmAiIiIiIiIVRAVMRERERESkgqiAiYiIiIiIVBAVMBERERERkQqiAiYiIiIiIlJBVMBEREREREQqiAqYiIiIiIhIBVEBExERERERqSAqYCIiIiIiIhVEBUxERERERKSCqICJiIiIiIhUEBUwERERERGRCqICJiIiIiIiUkFUwERERERERCqICpiIiIiIiEgFUQETERERERGpICpgIiIiIiIiFcTJ7ABVlWEYAGRkZJicREREREREzPRHJ/ijI1yMClgJZWZmAhAcHGxyEhERERERqQwyMzPx8fG56D4W41JqmpzH4XBw/PhxvLy8sFgspmbJyMggODiYhIQEvL29Tc0iZUeva/Wj17R60uta/eg1rZ70ulY/lek1NQyDzMxM6tevj9V68Xd5aQWshKxWKw0bNjQ7xjm8vb1N/49Pyp5e1+pHr2n1pNe1+tFrWj3pda1+Kstr+ncrX3/QEA4REREREZEKogImIiIiIiJSQVTAqgFXV1eeeeYZXF1dzY4iZUiva/Wj17R60uta/eg1rZ70ulY/VfU11RAOERERERGRCqIVMBERERERkQqiAiYiIiIiIlJBVMBEREREREQqiAqYiIiIiIhIBVEBqwZmzpxJWFgYbm5udO7cmejoaLMjSSmsXr2aYcOGUb9+fSwWC998843ZkaSUpk6dSteuXfHy8qJu3bqMGDGC/fv3mx1LSmHWrFm0b9+++OafERER/PDDD2bHkjI0depULBYLEyZMMDuKlMKzzz6LxWI55xEUFGR2LCmlxMREbr75ZurUqYOHhwcdOnRgy5YtZse6ZCpgVdzChQuZMGECkydPZtu2bURFRTFo0CDi4+PNjiYllJ2dTXh4OP/973/NjiJlZNWqVdx///1s2LCBFStWUFhYSP/+/cnOzjY7mpRQw4YNeemll9i8eTObN2/miiuuYPjw4ezevdvsaFIGfvvtN+bMmUP79u3NjiJloE2bNiQlJRU/du7caXYkKYXTp0/Ts2dPnJ2d+eGHH9izZw+vv/46vr6+Zke7ZBpDX8V1796dTp06MWvWrOJtrVq1YsSIEUydOtXEZFIWLBYLX3/9NSNGjDA7ipShkydPUrduXVatWkXv3r3NjiNlxM/Pj1dffZU77rjD7ChSCllZWXTq1ImZM2fywgsv0KFDB6ZPn252LCmhZ599lm+++YaYmBizo0gZeeyxx1i7dm2VvuJLK2BVWH5+Plu2bKF///7nbO/fvz/r1q0zKZWI/J309HSg6Ad2qfrsdjsLFiwgOzubiIgIs+NIKd1///0MGTKEq666yuwoUkYOHjxI/fr1CQsLY9SoURw5csTsSFIK3333HV26dOHGG2+kbt26dOzYkXfffdfsWJdFBawKS01NxW63ExgYeM72wMBAkpOTTUolIhdjGAaTJk2iV69etG3b1uw4Ugo7d+6kVq1auLq6Mm7cOL7++mtat25tdiwphQULFrB161ZdQVKNdO/enblz57J8+XLeffddkpOTiYyMJC0tzexoUkJHjhxh1qxZNGvWjOXLlzNu3Dgeeugh5s6da3a0S+ZkdgApPYvFcs7fDcM4b5uIVA4PPPAAO3bsYM2aNWZHkVJq0aIFMTExnDlzhq+++opbb72VVatWqYRVUQkJCYwfP54ff/wRNzc3s+NIGRk0aFDxn9u1a0dERARNmjTh448/ZtKkSSYmk5JyOBx06dKF//znPwB07NiR3bt3M2vWLG655RaT010arYBVYf7+/thstvNWu1JSUs5bFRMR8z344P+1d2exMbYNGMevvq1S1VaMJp2iqipRtKRTKRWMICLWSGyhTCJF1CsllhDrxHJgmzpTETuVCLGmDK1WCIKMLU3VUo2aaioEsbf9DsTk68vXVwfPfCP/X/IczDMz9311jnrN/dzz/K1jx46poKBAbdu29XUc/KTg4GDFx8crJSVF69atU/fu3ZWdne3rWPDS9evXVVVVJYvFoqCgIAUFBamwsFBbtmxRUFCQampqfB0Rv0BoaKgSExNVWlrq6yjwktls/uaLroSEBL/6AToKmB8LDg6WxWKR0+msd97pdCotLc1HqQD8U11dnWbPnq3Dhw8rPz9fHTp08HUk/AZ1dXX68OGDr2PASwMHDtTt27flcrk8R0pKiiZNmiSXy6XAwEBfR8Qv8OHDBxUXF8tsNvs6CrzUp0+fb27lcu/ePbVv395HiRqPSxD93Lx585Senq6UlBT17t1bOTk5Ki8v18yZM30dDV568+aN7t+/73n86NEjuVwutWrVSjExMT5MBm9lZmZq//79Onr0qMLCwjyr1hEREQoJCfFxOnhjyZIlGjp0qNq1a6fXr18rNzdX58+fV15enq+jwUthYWHf7MsMDQ2VyWRiv6Yfmz9/vkaMGKGYmBhVVVVp9erVevXqlaZOnerraPDS3LlzlZaWprVr12rcuHG6evWqcnJylJOT4+toP4wC5ufGjx+v58+fy263y+12q1u3bjp16pRffQuA+q5du6YBAwZ4Hn+9Rn3q1KnauXOnj1LhZ3y9TYTVaq13fseOHbLZbMYHwk979uyZ0tPT5Xa7FRERoaSkJOXl5Wnw4MG+jgbgvzx58kQTJ05UdXW1IiMj1atXL12+fJn/k/xYz549deTIES1evFh2u10dOnSQw+HQpEmTfB3th3EfMAAAAAAwCHvAAAAAAMAgFDAAAAAAMAgFDAAAAAAMQgEDAAAAAINQwAAAAADAIBQwAAAAADAIBQwAAAAADEIBAwAAAACDUMAAAPAzNptNo0eP9nUMAIAXKGAAAL8TEBDQ4GGz2bweOzY2Vg6H44de93W+wMBARUdHa9q0aXrx4kWj5rNarcrKyvIuLADA71DAAAB+x+12ew6Hw6Hw8PB657Kzsw3JYbfb5Xa7VV5ern379qmoqEhz5swxZG4AgH+igAEA/E5UVJTniIiIUEBAQL1zRUVFslgsatasmeLi4rRq1Sp9/vzZ8/6VK1cqJiZGTZs2VXR0tKc0Wa1WPX78WHPnzvWsbjUkLCxMUVFRatOmjQYMGKApU6boxo0bnuefP3+uiRMnqm3btmrevLkSExN14MABz/M2m02FhYXKzs72zFdWViZJunv3roYNG6bw8HCFhYWpb9++evDgQb35N2zYILPZLJPJpMzMTH369OlnP1oAwG8W5OsAAAD8SqdPn9bkyZO1ZcsWT2mZPn26JGnFihU6dOiQNm/erNzcXHXt2lWVlZW6efOmJOnw4cPq3r27pk+froyMjEbNW1FRoRMnTig1NdVz7v3797JYLFq0aJHCw8N18uRJpaenKy4uTqmpqcrOzta9e/fUrVs32e12SVJkZKQqKirUr18/Wa1W5efnKzw8XBcvXqxXIgsKCmQ2m1VQUKD79+9r/Pjx6tGjR6NzAwCMFVBXV1fn6xAAAHhr586dysrK0suXLyVJ/fr109ChQ7V48WLPa/bu3auFCxfq6dOn2rRpk7Zu3ao7d+6oSZMm34wXGxurrKysf92XFRsbK7fbrSZNmqimpkbv379Xamqq8vLy1LJly//5vmHDhikhIUEbNmyQ9GXVrUePHvX2nS1ZskS5ubkqKSn5bkabzabz58/rwYMHCgwMlCSNGzdOf/31l3JzcxvMDQDwLS5BBAD8Ua5fvy673a4WLVp4joyMDLndbr19+1Zjx47Vu3fvFBcXp4yMDB05cqTeylJjLFiwQC6XS7du3dK5c+ckfSlYNTU1kqSamhqtWbNGSUlJMplMatGihc6cOaPy8vIGx3W5XOrbt+93y9dXXbt29ZQvSTKbzaqqqvLq7wAAGIdLEAEAf5Ta2lqtWrVKY8aM+ea5Zs2aqV27diopKZHT6dTZs2c1a9YsrV+/XoWFhQ0Wnu9p3bq14uPjJUmdOnWSw+FQ7969VVBQoEGDBmnjxo3avHmzHA6HEhMTFRoaqqysLH38+LHBcUNCQv517n9mDQgIUG1tbaPyAwCMRwEDAPxRkpOTVVJS4ilG3xMSEqKRI0dq5MiRyszMVOfOnXX79m0lJycrODjYs4LVWF9XpN69eydJunDhgkaNGqXJkydL+lIOS0tLlZCQ4HnP9+ZLSkrSrl279OnTp0aXQgDA/zcuQQQA/FGWL1+u3bt3a+XKlbp7966Ki4t18OBBLV26VNKXPWPbt2/XnTt39PDhQ+3Zs0chISFq3769pC97u4qKilRRUaHq6uoG53r9+rUqKyvldrt19epVLViwQK1bt1ZaWpokKT4+Xk6nU5cuXVJxcbFmzJihysrKemPExsbqypUrKisrU3V1tWprazV79my9evVKEyZM0LVr11RaWqo9e/aopKTkN3xiAAAjUcAAAH+UIUOG6MSJE3I6nerZs6d69eqlTZs2eQpWy5YttW3bNvXp00dJSUk6d+6cjh8/LpPJJOnLvb3KysrUsWNHRUZGNjjX8uXLZTabFR0dreHDhys0NFROp9Mz1rJly5ScnKwhQ4bIarUqKipKo0ePrjfG/PnzFRgYqC5duigyMlLl5eUymUzKz8/Xmzdv1L9/f1ksFm3bto3VMAD4A/AriAAAAABgEFbAAAAAAMAgFDAAAAAAMAgFDAAAAAAMQgEDAAAAAINQwAAAAADAIBQwAAAAADAIBQwAAAAADEIBAwAAAACDUMAAAAAAwCAUMAAAAAAwCAUMAAAAAAzyH4ENpz5Up8U8AAAAAElFTkSuQmCC",
      "text/plain": [
       "<Figure size 1000x600 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "test_losses = []\n",
    "model.eval()\n",
    "ys = []\n",
    "with torch.no_grad():\n",
    "    for x, y in test_loader:\n",
    "        x = x.float()\n",
    "        t_eval_tensor = torch.linspace(0., model.seq_len, steps=model.seq_len, device=x.device)\n",
    "        y_true = y.float()\n",
    "\n",
    "        preds = model(x, t_eval_tensor)\n",
    "        y_true_adjusted = y_true[:, -1, :]\n",
    "        ys.append(y_true_adjusted)\n",
    "        loss = criterion(preds, y_true_adjusted)\n",
    "\n",
    "        test_losses.append(loss.item())\n",
    "\n",
    "test_loss = np.mean(test_losses)\n",
    "print(f\"Test Loss: {test_loss:.4f}\")\n",
    "\n",
    "plt.figure(figsize=(10, 6))\n",
    "plt.plot(test_losses, label='Test Loss')\n",
    "plt.xlabel('Test Batch')\n",
    "plt.ylabel('Loss')\n",
    "plt.title('Test Loss')\n",
    "plt.legend()\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "id": "7e2aef157668fcaa",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T06:14:38.381146Z",
     "start_time": "2024-02-22T06:14:38.354082Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[tensor([[ 0.6426,  0.4318,  0.2220],\n",
       "         [ 0.3605,  0.3719,  0.1443],\n",
       "         [ 0.3024,  0.2718,  0.2453],\n",
       "         [ 0.7975,  0.5448,  0.1181],\n",
       "         [ 0.7607,  0.4650,  0.1904],\n",
       "         [ 0.4489,  0.4339,  0.1630],\n",
       "         [ 1.0127,  0.5219,  0.2902],\n",
       "         [ 0.3525,  0.3797,  0.4070],\n",
       "         [ 0.6124,  0.5185,  0.2511],\n",
       "         [ 0.5871,  0.3506,  0.1145],\n",
       "         [ 0.5247,  0.1258,  0.0710],\n",
       "         [ 0.8625,  0.7125,  0.2338],\n",
       "         [ 0.6842,  0.3904,  0.0579],\n",
       "         [ 0.5771,  0.4930,  0.0797],\n",
       "         [ 0.1574,  0.0969,  0.0101],\n",
       "         [ 0.3596, -0.0264, -0.1449],\n",
       "         [ 0.1738,  0.4750, -0.0094],\n",
       "         [ 0.7058,  0.6284,  0.3473],\n",
       "         [ 0.5521,  0.4833,  0.2496],\n",
       "         [ 0.3463,  0.2399,  0.1716],\n",
       "         [ 0.4525,  0.1772,  0.0951],\n",
       "         [ 0.5219,  0.4312,  0.1987],\n",
       "         [ 0.2677,  0.2892, -0.0029],\n",
       "         [ 0.4366,  0.0855,  0.2556],\n",
       "         [ 0.3368,  0.5808,  0.1295],\n",
       "         [ 0.2820,  0.1640, -0.0649],\n",
       "         [ 0.2436,  0.3225,  0.0032],\n",
       "         [ 0.2405,  0.3017,  0.1154],\n",
       "         [ 0.7153,  0.3907,  0.2022],\n",
       "         [ 0.4208,  0.2702,  0.0774],\n",
       "         [ 0.5848,  0.4859,  0.1564],\n",
       "         [ 0.5941,  0.5840,  0.4222]]),\n",
       " tensor([[ 3.8607e-01,  4.7439e-01,  3.1843e-01],\n",
       "         [ 4.3222e-01,  2.3785e-01,  2.7078e-01],\n",
       "         [ 4.8185e-01,  3.3654e-01,  1.3522e-01],\n",
       "         [ 3.0413e-01,  3.1732e-01,  3.5557e-02],\n",
       "         [ 5.3474e-01,  3.3026e-01,  2.1535e-01],\n",
       "         [ 5.6590e-01,  4.2599e-01,  3.7825e-02],\n",
       "         [ 5.2015e-01,  2.8635e-01,  8.6063e-02],\n",
       "         [ 3.6538e-01,  2.6617e-01, -9.9985e-02],\n",
       "         [ 7.1232e-01,  7.6203e-01,  1.0518e-01],\n",
       "         [ 4.1461e-01,  4.5046e-01,  1.4728e-01],\n",
       "         [ 1.6183e-01,  2.4677e-01,  8.2283e-02],\n",
       "         [ 4.1332e-01,  4.0533e-01,  9.0856e-02],\n",
       "         [ 4.9893e-02, -1.6649e-01,  9.2931e-02],\n",
       "         [ 5.3688e-01,  3.4166e-01,  6.7573e-03],\n",
       "         [ 5.3696e-01,  4.7160e-01,  9.2155e-02],\n",
       "         [ 2.6681e-01,  3.7386e-01,  2.6599e-01],\n",
       "         [ 5.4540e-01,  4.3816e-01,  6.9183e-03],\n",
       "         [ 5.4132e-01,  5.0075e-01,  2.6525e-01],\n",
       "         [ 6.0085e-01,  1.9001e-01,  4.7078e-02],\n",
       "         [ 9.3150e-01,  7.2327e-01,  3.3805e-01],\n",
       "         [ 3.3466e-01,  1.5680e-01,  2.2784e-01],\n",
       "         [ 7.5124e-01,  4.2458e-01,  1.3207e-01],\n",
       "         [ 2.8333e-01,  2.1448e-01,  1.0698e-01],\n",
       "         [ 2.5120e-01,  4.5617e-01,  2.4160e-01],\n",
       "         [ 7.8790e-01,  8.2092e-01,  2.8733e-01],\n",
       "         [ 3.4288e-01,  2.6396e-01,  3.1958e-01],\n",
       "         [ 2.9074e-01,  2.1167e-01,  1.7652e-01],\n",
       "         [ 8.9006e-02,  9.9067e-02,  6.6722e-02],\n",
       "         [ 1.5916e-01,  3.2103e-01,  4.8116e-02],\n",
       "         [ 3.3987e-01,  2.8635e-01, -3.6448e-04],\n",
       "         [ 3.4604e-01,  4.2915e-01,  2.7389e-02],\n",
       "         [ 4.7194e-01,  3.9404e-01,  2.0252e-01]]),\n",
       " tensor([[ 0.3185,  0.2025,  0.0314],\n",
       "         [ 0.6204,  0.4958,  0.2029],\n",
       "         [ 0.0587,  0.0505, -0.0401],\n",
       "         [ 0.7013,  0.3547,  0.2693],\n",
       "         [ 0.3690,  0.5482,  0.0802],\n",
       "         [ 0.2251,  0.1055,  0.1581],\n",
       "         [ 0.3289,  0.3368,  0.0902],\n",
       "         [ 0.5782,  0.4081,  0.1528],\n",
       "         [-0.0571,  0.1107,  0.1845],\n",
       "         [-0.0441,  0.1604,  0.0250],\n",
       "         [ 0.4156,  0.5926,  0.2039],\n",
       "         [ 0.5226,  0.4956,  0.1223],\n",
       "         [ 0.6455,  0.4271,  0.1249],\n",
       "         [ 0.4010,  0.3435, -0.0024],\n",
       "         [ 0.4513,  0.1258,  0.0667],\n",
       "         [ 0.3959,  0.2640,  0.3086],\n",
       "         [ 0.4143,  0.2899,  0.0694],\n",
       "         [ 0.5748,  0.4196,  0.1803],\n",
       "         [ 0.2109,  0.3209,  0.2277],\n",
       "         [ 0.7639,  0.5177,  0.1948],\n",
       "         [ 0.7757,  0.5698,  0.1977],\n",
       "         [ 0.5080,  0.5109,  0.1244],\n",
       "         [ 0.3373,  0.4622,  0.2157],\n",
       "         [ 0.2124,  0.1290,  0.1336],\n",
       "         [ 0.5283,  0.2302,  0.0579],\n",
       "         [ 0.3794,  0.2726,  0.1004],\n",
       "         [ 0.5066,  0.3289,  0.2170],\n",
       "         [ 0.2852,  0.1591,  0.0915],\n",
       "         [ 0.2071,  0.0883,  0.1943],\n",
       "         [ 0.7825,  0.5556,  0.0841],\n",
       "         [ 0.7760,  0.4972,  0.0732],\n",
       "         [ 0.4236,  0.4355,  0.0645]]),\n",
       " tensor([[ 0.1790,  0.4366,  0.2035],\n",
       "         [ 0.4446,  0.3793,  0.2565],\n",
       "         [ 0.7139,  0.6100,  0.2960],\n",
       "         [ 0.3093,  0.3387, -0.0013],\n",
       "         [ 0.1927,  0.3993,  0.0264],\n",
       "         [ 0.2469,  0.2071,  0.2022],\n",
       "         [ 0.5250,  0.4711,  0.1543],\n",
       "         [ 0.6446,  0.3323,  0.0372],\n",
       "         [ 0.5722,  0.5776,  0.3457],\n",
       "         [ 0.2826,  0.4159, -0.0049],\n",
       "         [ 0.4708,  0.3720,  0.2748],\n",
       "         [ 0.5169,  0.6123,  0.2074],\n",
       "         [ 0.4285,  0.1687,  0.1294],\n",
       "         [ 0.4895,  0.2876,  0.1206],\n",
       "         [ 0.6240,  0.5147,  0.4157],\n",
       "         [ 0.7103,  0.5238,  0.1249],\n",
       "         [ 0.8605,  0.6010,  0.3036],\n",
       "         [ 0.3140,  0.1408, -0.0593],\n",
       "         [ 0.8037,  0.7134,  0.3972],\n",
       "         [ 0.1362,  0.5639, -0.0092],\n",
       "         [ 0.0723,  0.1895,  0.0549],\n",
       "         [ 0.5514,  0.4191,  0.2789],\n",
       "         [ 0.2024,  0.1018,  0.1587],\n",
       "         [ 0.4189,  0.6109, -0.0256],\n",
       "         [ 0.3657,  0.3383,  0.0691],\n",
       "         [ 0.6715,  0.5586,  0.2767],\n",
       "         [ 0.4610,  0.3682,  0.0079],\n",
       "         [ 0.4948,  0.2538,  0.1070],\n",
       "         [ 0.0876,  0.2262,  0.0521],\n",
       "         [ 0.3370,  0.3305, -0.0406],\n",
       "         [ 0.4703,  0.2896,  0.1948],\n",
       "         [ 0.5416,  0.5101, -0.0523]]),\n",
       " tensor([[ 0.3392,  0.1527,  0.0802],\n",
       "         [ 0.4709,  0.5794,  0.1504],\n",
       "         [ 0.8121,  0.6190,  0.2590],\n",
       "         [ 0.3811,  0.3791,  0.0158],\n",
       "         [ 0.4198,  0.5753,  0.0216],\n",
       "         [ 0.3174,  0.1823,  0.0509],\n",
       "         [ 0.5637,  0.5296,  0.1213],\n",
       "         [ 0.5740,  0.5755,  0.2925],\n",
       "         [ 0.2264,  0.4164, -0.0187],\n",
       "         [ 0.6770,  0.4123,  0.2345],\n",
       "         [ 0.7628,  0.4445,  0.1449],\n",
       "         [ 0.7252,  0.6673,  0.3249],\n",
       "         [ 0.4241,  0.3148,  0.2908],\n",
       "         [ 0.1899,  0.4734, -0.1736],\n",
       "         [ 0.6898,  0.6187,  0.2235],\n",
       "         [ 0.2842,  0.2863,  0.0298],\n",
       "         [ 0.7492,  0.7798,  0.3875],\n",
       "         [ 0.7056,  0.4837,  0.2479],\n",
       "         [ 0.3319,  0.3064,  0.1706],\n",
       "         [ 0.4248,  0.2945,  0.3843],\n",
       "         [ 0.2930,  0.1331,  0.0339],\n",
       "         [ 0.7895,  0.4637,  0.2492],\n",
       "         [ 0.5138,  0.5543,  0.2600],\n",
       "         [ 0.5099,  0.4694,  0.1548],\n",
       "         [ 0.5770,  0.4782, -0.0046],\n",
       "         [ 0.7025,  0.4988,  0.1939],\n",
       "         [ 0.6492,  0.4597, -0.0687],\n",
       "         [ 0.3172,  0.2369,  0.2681],\n",
       "         [ 0.1346,  0.4907, -0.0240],\n",
       "         [ 0.7661,  0.5145,  0.0411],\n",
       "         [ 0.3954,  0.1661,  0.0436],\n",
       "         [ 0.7349,  0.5775,  0.2210]]),\n",
       " tensor([[ 0.7498,  0.5245,  0.1583],\n",
       "         [ 0.3500,  0.1928,  0.0411],\n",
       "         [ 0.5456,  0.2893,  0.2477],\n",
       "         [ 0.6404,  0.3505,  0.0456],\n",
       "         [ 0.3578,  0.3437,  0.1560],\n",
       "         [ 0.5994,  0.5080,  0.3781],\n",
       "         [ 0.4434,  0.3313, -0.0409],\n",
       "         [ 0.4845,  0.4885,  0.2285],\n",
       "         [ 0.1529,  0.1150,  0.0975],\n",
       "         [ 0.6926,  0.6640,  0.4946],\n",
       "         [ 0.9130,  0.5581,  0.2868],\n",
       "         [ 0.7054,  0.2787,  0.0230],\n",
       "         [ 0.2107,  0.0275,  0.1805],\n",
       "         [ 0.4887,  0.1089,  0.1585],\n",
       "         [ 0.6815,  0.6482,  0.3834],\n",
       "         [ 0.5692,  0.2710,  0.1630],\n",
       "         [ 0.7425,  0.6128,  0.1309],\n",
       "         [ 0.5175,  0.2549,  0.0849],\n",
       "         [ 0.4407,  0.1900,  0.1889],\n",
       "         [ 0.4712,  0.3077,  0.0850],\n",
       "         [ 0.7056,  0.8476,  0.2454],\n",
       "         [ 0.3709,  0.1944,  0.1233],\n",
       "         [ 0.4228,  0.4677,  0.2456],\n",
       "         [ 0.5839,  0.6248,  0.1213],\n",
       "         [ 0.5162,  0.4890, -0.0680],\n",
       "         [ 0.2347, -0.0021,  0.0165],\n",
       "         [ 0.7041,  0.5548,  0.0980],\n",
       "         [ 0.5404,  0.2213,  0.0246],\n",
       "         [ 0.2715,  0.2059,  0.0799],\n",
       "         [ 0.6217,  0.2468,  0.1807],\n",
       "         [ 0.7120,  0.5780,  0.1141],\n",
       "         [ 0.6929,  0.8294,  0.1336]]),\n",
       " tensor([[0.5624, 0.4442, 0.1650],\n",
       "         [0.9143, 0.5824, 0.1445],\n",
       "         [0.4692, 0.3171, 0.0393],\n",
       "         [0.7850, 0.5532, 0.0856],\n",
       "         [0.1259, 0.0142, 0.2441],\n",
       "         [0.4134, 0.5543, 0.1246],\n",
       "         [0.2789, 0.0421, 0.1390],\n",
       "         [0.2763, 0.4996, 0.0320]])]"
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ys"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "id": "590ebf4bfb7dcaf6",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T06:46:19.158138Z",
     "start_time": "2024-02-22T06:46:19.141316Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(800, 100, 3)"
      ]
     },
     "execution_count": 127,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "id": "53775489e55a806e",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T06:44:25.170115Z",
     "start_time": "2024-02-22T06:44:25.158873Z"
    },
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "class ODEFunc(nn.Module):\n",
    "    def __init__(self, hidden_size):\n",
    "        super(ODEFunc, self).__init__()\n",
    "        self.net = nn.Sequential(\n",
    "            nn.Linear(hidden_size, hidden_size),\n",
    "            nn.Tanh(),\n",
    "            nn.Linear(hidden_size, hidden_size),\n",
    "        )\n",
    "    \n",
    "    def forward(self, t, y):\n",
    "        return self.net(y)\n",
    "\n",
    "class HybridODELSTMModel(nn.Module):\n",
    "    def __init__(self, input_size, hidden_size, output_size, seq_len):\n",
    "        super(HybridODELSTMModel, self).__init__()\n",
    "        self.seq_len = seq_len\n",
    "        self.lstm = nn.LSTM(input_size=input_size, hidden_size=hidden_size, batch_first=True)\n",
    "        self.ode_func = ODEFunc(hidden_size)\n",
    "        self.fc = nn.Linear(hidden_size, output_size)\n",
    "    \n",
    "    def forward(self, x, t_span):\n",
    "        lstm_out, (h_n, c_n) = self.lstm(x)\n",
    "        h_n = h_n[-1]\n",
    "        ode_sol = odeint(self.ode_func, h_n, t_span, method=\"dopri5\")\n",
    "        out = self.fc(ode_sol[-1])\n",
    "        return out\n",
    "\n",
    "def three_comp_ode(t, y, Qk, Qt, Kpk, Kpt, Vv, Vk, Vt):\n",
    "    V, K, T = y.unbind()\n",
    "    dVdt = - (Qk/Vv)*V + (Qk/Vk)*Kpk*K - (Qt/Vv)*V + (Qt/Vt)*Kpt*T\n",
    "    dKdt = (Qk/Vk) * Kpk * V - (Qk/Vk)*K\n",
    "    dTdt = (Qt/Vt) * Kpt * V - (Qt/Vt)*T\n",
    "    return torch.stack([dVdt, dKdt, dTdt])\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "id": "9cd953db9f6125fd",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2024-02-22T06:57:54.914752Z",
     "start_time": "2024-02-22T06:57:54.702837Z"
    },
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "too many values to unpack (expected 3)",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "Cell \u001b[0;32mIn[130], line 50\u001b[0m\n\u001b[1;32m     46\u001b[0m t_span_tensor \u001b[38;5;241m=\u001b[39m torch\u001b[38;5;241m.\u001b[39mlinspace(\u001b[38;5;241m0.\u001b[39m, seq_len, steps\u001b[38;5;241m=\u001b[39mseq_len, device\u001b[38;5;241m=\u001b[39m\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mcuda\u001b[39m\u001b[38;5;124m'\u001b[39m \u001b[38;5;28;01mif\u001b[39;00m torch\u001b[38;5;241m.\u001b[39mcuda\u001b[38;5;241m.\u001b[39mis_available() \u001b[38;5;28;01melse\u001b[39;00m \u001b[38;5;124m'\u001b[39m\u001b[38;5;124mcpu\u001b[39m\u001b[38;5;124m'\u001b[39m, requires_grad\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)\u001b[38;5;241m.\u001b[39mfloat()  \n\u001b[1;32m     49\u001b[0m \u001b[38;5;66;03m# Train the model with PINN loss\u001b[39;00m\n\u001b[0;32m---> 50\u001b[0m model \u001b[38;5;241m=\u001b[39m \u001b[43mtrain_pinn\u001b[49m\u001b[43m(\u001b[49m\u001b[43mmodel\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mtrain_loader\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43moptimizer\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mcriterion\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[38;5;241;43m100\u001b[39;49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mQk\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mQt\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mKpk\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mKpt\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mVv\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mVk\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mVt\u001b[49m\u001b[43m,\u001b[49m\u001b[43mt_span_tensor\u001b[49m\u001b[43m)\u001b[49m\n",
      "Cell \u001b[0;32mIn[130], line 22\u001b[0m, in \u001b[0;36mtrain_pinn\u001b[0;34m(model, train_loader, optimizer, criterion, num_epochs, Qk, Qt, Kpk, Kpt, Vv, Vk, Vt, t_span_tensor)\u001b[0m\n\u001b[1;32m     17\u001b[0m d_preds \u001b[38;5;241m=\u001b[39m torch\u001b[38;5;241m.\u001b[39mautograd\u001b[38;5;241m.\u001b[39mgrad(outputs\u001b[38;5;241m=\u001b[39mpreds, inputs\u001b[38;5;241m=\u001b[39mt_span_tensor, \n\u001b[1;32m     18\u001b[0m                           grad_outputs\u001b[38;5;241m=\u001b[39mtorch\u001b[38;5;241m.\u001b[39mones_like(preds), \n\u001b[1;32m     19\u001b[0m                           create_graph\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m, retain_graph\u001b[38;5;241m=\u001b[39m\u001b[38;5;28;01mTrue\u001b[39;00m)[\u001b[38;5;241m0\u001b[39m] \n\u001b[1;32m     21\u001b[0m \u001b[38;5;66;03m# Evaluate your PBPK ODE with predictions\u001b[39;00m\n\u001b[0;32m---> 22\u001b[0m ode_evals \u001b[38;5;241m=\u001b[39m \u001b[43mthree_comp_ode\u001b[49m\u001b[43m(\u001b[49m\u001b[43mt_span_tensor\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mpreds\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mQk\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mQt\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mKpk\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mKpt\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mVv\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mVk\u001b[49m\u001b[43m,\u001b[49m\u001b[43m \u001b[49m\u001b[43mVt\u001b[49m\u001b[43m)\u001b[49m \n\u001b[1;32m     24\u001b[0m \u001b[38;5;66;03m# Physics loss (compare predicted derivatives to ODE output)\u001b[39;00m\n\u001b[1;32m     25\u001b[0m physics_loss \u001b[38;5;241m+\u001b[39m\u001b[38;5;241m=\u001b[39m criterion(d_preds, ode_evals) \n",
      "Cell \u001b[0;32mIn[125], line 33\u001b[0m, in \u001b[0;36mthree_comp_ode\u001b[0;34m(t, y, Qk, Qt, Kpk, Kpt, Vv, Vk, Vt)\u001b[0m\n\u001b[1;32m     32\u001b[0m \u001b[38;5;28;01mdef\u001b[39;00m \u001b[38;5;21mthree_comp_ode\u001b[39m(t, y, Qk, Qt, Kpk, Kpt, Vv, Vk, Vt):\n\u001b[0;32m---> 33\u001b[0m     V, K, T \u001b[38;5;241m=\u001b[39m y\u001b[38;5;241m.\u001b[39munbind()\n\u001b[1;32m     34\u001b[0m     dVdt \u001b[38;5;241m=\u001b[39m \u001b[38;5;241m-\u001b[39m (Qk\u001b[38;5;241m/\u001b[39mVv)\u001b[38;5;241m*\u001b[39mV \u001b[38;5;241m+\u001b[39m (Qk\u001b[38;5;241m/\u001b[39mVk)\u001b[38;5;241m*\u001b[39mKpk\u001b[38;5;241m*\u001b[39mK \u001b[38;5;241m-\u001b[39m (Qt\u001b[38;5;241m/\u001b[39mVv)\u001b[38;5;241m*\u001b[39mV \u001b[38;5;241m+\u001b[39m (Qt\u001b[38;5;241m/\u001b[39mVt)\u001b[38;5;241m*\u001b[39mKpt\u001b[38;5;241m*\u001b[39mT\n\u001b[1;32m     35\u001b[0m     dKdt \u001b[38;5;241m=\u001b[39m (Qk\u001b[38;5;241m/\u001b[39mVk) \u001b[38;5;241m*\u001b[39m Kpk \u001b[38;5;241m*\u001b[39m V \u001b[38;5;241m-\u001b[39m (Qk\u001b[38;5;241m/\u001b[39mVk)\u001b[38;5;241m*\u001b[39mK\n",
      "\u001b[0;31mValueError\u001b[0m: too many values to unpack (expected 3)"
     ]
    }
   ],
   "source": [
    "def train_pinn(model, train_loader, optimizer, criterion, num_epochs, Qk, Qt, Kpk, Kpt, Vv, Vk, Vt,t_span_tensor):\n",
    "    for epoch in range(num_epochs):\n",
    "        model.train()\n",
    "        total_loss = 0\n",
    "        for batch_idx, (x, y) in enumerate(train_loader):\n",
    "            x, y = x.float(), y.float()\n",
    "            optimizer.zero_grad()\n",
    "            \n",
    "            # Forward pass: compute predictions\n",
    "            preds = model(x, t_span_tensor)\n",
    "            y_true_adjusted = y[:, -1, :]\n",
    "            \n",
    "            # Compute the MSE loss (data-fidelity term)\n",
    "            mse_loss = criterion(preds, y_true_adjusted)\n",
    "            \n",
    "            # Calculate derivatives with respect to time\n",
    "            d_preds = torch.autograd.grad(outputs=preds, inputs=t_span_tensor, \n",
    "                                      grad_outputs=torch.ones_like(preds), \n",
    "                                      create_graph=True, retain_graph=True)[0] \n",
    "\n",
    "            # Evaluate your PBPK ODE with predictions\n",
    "            ode_evals = three_comp_ode(t_span_tensor, preds, Qk, Qt, Kpk, Kpt, Vv, Vk, Vt) \n",
    "\n",
    "            # Physics loss (compare predicted derivatives to ODE output)\n",
    "            physics_loss += criterion(d_preds, ode_evals) \n",
    "            \n",
    "            # Total loss is the sum of data-fidelity and physics-fidelity terms\n",
    "            loss = mse_loss #+ physics_loss\n",
    "            loss.backward()\n",
    "            optimizer.step()\n",
    "            \n",
    "            total_loss += loss.item()\n",
    "        \n",
    "        print(f\"Epoch {epoch+1}/{num_epochs}, Loss: {total_loss/len(train_loader)}\")\n",
    "    \n",
    "    return model\n",
    "\n",
    "# Instantiate the model\n",
    "model = HybridODELSTMModel(input_size, hidden_size, output_size, seq_len)\n",
    "\n",
    "# Define optimizer and loss function\n",
    "optimizer = torch.optim.Adam(model.parameters())\n",
    "criterion = torch.nn.MSELoss()\n",
    "\n",
    "# Define the time span tensor outside the loop for efficiency\n",
    "t_span_tensor = torch.linspace(0., seq_len, steps=seq_len, device='cuda' if torch.cuda.is_available() else 'cpu', requires_grad=True).float()  \n",
    "\n",
    "\n",
    "# Train the model with PINN loss\n",
    "model = train_pinn(model, train_loader, optimizer, criterion, 100, Qk, Qt, Kpk, Kpt, Vv, Vk, Vt,t_span_tensor)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "607758518d25d3d8",
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}