diff --git a/notebooks/DeepEnsemble_linefit.ipynb b/notebooks/DeepEnsemble_linefit.ipynb index a1a658e..b42aa89 100644 --- a/notebooks/DeepEnsemble_linefit.ipynb +++ b/notebooks/DeepEnsemble_linefit.ipynb @@ -5,7 +5,7 @@ "id": "f774193d", "metadata": {}, "source": [ - "# Linefit Deep Ensemble" + "# Linear fit DeepEnsemble" ] }, { @@ -18,7 +18,7 @@ }, { "cell_type": "code", - "execution_count": 1, + "execution_count": 4, "id": "5e8c8f57", "metadata": {}, "outputs": [], @@ -34,7 +34,7 @@ }, { "cell_type": "code", - "execution_count": 2, + "execution_count": 5, "id": "2fa611df", "metadata": {}, "outputs": [], @@ -62,7 +62,7 @@ }, { "cell_type": "code", - "execution_count": 3, + "execution_count": 6, "id": "df535a12-c8bf-46eb-a575-12deafb1109b", "metadata": {}, "outputs": [], @@ -74,13 +74,14 @@ }, { "cell_type": "code", - "execution_count": 6, + "execution_count": 7, "id": "e420b77a", "metadata": {}, "outputs": [], "source": [ "loss_type = 'var_loss'\n", - "# options are 'no_var_loss' or 'var_loss'" + "# options are 'no_var_loss' or 'var_loss'\n", + "# var_loss is a gaussian negative log likelihood option that incorporates a variance term (for aleatoric uncertainty)" ] }, { @@ -89,62 +90,65 @@ "metadata": {}, "source": [ "## Generate line data\n", - "To do this make a dataframe and replicate a bunch of columns. There are 8 pendulums on two different planets. The planet_id and pendulum_id are integers denoting which pendulum and which planet each row of the dataframe belongs." + "Unlike in the case of inference, here we need to upfront generate a dataframe and replicate a bunch of columns. We will generate using random draws from distributions that are similar to the priors in the case of the likelihood-based inference analysis." ] }, { "cell_type": "code", - "execution_count": 7, + "execution_count": 8, "id": "674790fc-de90-4ed6-95d6-406afc2c02ea", "metadata": {}, "outputs": [], "source": [ - "def simulator(thetas):#, percent_errors):\n", - " # just plop the pendulum within here\n", - " x, m, b, sigma = thetas\n", - " #x = np.linspace(0, 100, 101)\n", + "# same simulator as in likelihood-based inference,\n", + "# but here we have \n", + "def simulator(thetas):\n", + " m, b = thetas\n", + " x_data = np.linspace(0, 100, 101)\n", " rs = np.random.RandomState()#2147483648)# \n", + " sigma = 10\n", " ε = rs.normal(loc=0, scale=sigma)#, size = len(x)) \n", - " return m * x + b + ε" + " return m * x_data + b + ε" ] }, { "cell_type": "code", - "execution_count": 8, + "execution_count": 27, "id": "42b76a6f-28b5-4f62-8d2f-85f8e39e13a1", "metadata": {}, "outputs": [], "source": [ - "length_df = 10000\n", + "x_data = np.linspace(0, 100, 101)\n", + "\n", + "length_df = 10\n", "#ms = np.zeros((length_df, 101))\n", - "ms = np.zeros((length_df, 1))\n", - "bs = np.zeros((length_df, 1))\n", - "sigmas = np.zeros((length_df, 1))\n", - "xs = np.zeros((length_df, 1))\n", - "inputs = np.zeros((length_df, 4))\n", - "ys = []\n", + "ms_list = []#np.zeros((length_df, 1))\n", + "bs_list = []#np.zeros((length_df, 1))\n", + "ys_list = []\n", "#true_sigmas = []\n", "\n", "for r in range(length_df):\n", " rs = np.random.RandomState()#2147483648)# \n", - " m = rs.normal(loc=1, scale=1)\n", - " b = rs.normal(loc=0, scale=10)\n", - " sigma = abs(rs.normal(loc=0, scale=10))\n", - " x = rs.uniform(low=0, high = 100)\n", - " thetas_1d = [x, m, b, sigma]\n", - " inputs[r,:] = thetas_1d\n", + " m = rs.uniform(low=0, high=10)\n", + " b = rs.uniform(low=-10, high=10) \n", + " ms_list.append(m)\n", + " bs_list.append(b)\n", " '''\n", " xs[r,:] = thetas_1d[0]#np.linspace(0, 100, 101)\n", " ms[r,:] = thetas_1d[1]#np.repeat(thetas_1d[0], 101)\n", " bs[r,:] = thetas_1d[2]#np.repeat(thetas_1d[1], 101)\n", " sigmas[r,:] = thetas_1d[3]#np.repeat(thetas_1d[2], 101)\n", " '''\n", - " ys.append(simulator(thetas_1d))" + " ys_list.append(simulator([m,b]))\n", + "ms_array = np.repeat(ms_list, len(x_data))\n", + "bs_array = np.repeat(bs_list, len(x_data))\n", + "xs_array = np.tile(x_data, length_df)\n", + "ys_array = np.reshape(ys_list, (length_df * len(x_data)))" ] }, { "cell_type": "code", - "execution_count": 45, + "execution_count": 33, "id": "b18cea86", "metadata": {}, "outputs": [ @@ -159,9 +163,9 @@ }, { "data": { - "image/png": "iVBORw0KGgoAAAANSUhEUgAAAz8AAAFzCAYAAAAQULd9AAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjguMCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy81sbWrAAAACXBIWXMAAA9hAAAPYQGoP6dpAABiFklEQVR4nO3de1yUZf4//hdnFJxB0JmBFZBKRRTFUHHSjIRlRHI16aAhYpkUC5br4WN8PWuKmprpopYVaOm62aoVKUoesOSgkpSnSFvdsXCgTWGEkuP9+8Pf3OsIKCpzfj0fj/vxYO7rmpnrYq65537f1+G2EwRBABERERERkZWzN3UBiIiIiIiIjIHBDxERERER2QQGP0REREREZBMY/BARERERkU1g8ENERERERDaBwQ8REREREdkEBj9ERERERGQTGPwQEREREZFNcDR1Ae5HY2MjSktL0aFDB9jZ2Zm6OGQmBEHA9evX4ePjA3t7w8X1bH/UHGO1P4BtkJpi+yNT428wmdK9tD+LDH5KS0vh6+tr6mKQmbp8+TK6dOlisNdn+6M7MXT7A9gGqWVsf2Rq/A0mU2pN+7PI4KdDhw4AblZQIpGYuDRkLrRaLXx9fcX2YShsf9QcY7U/gG2QmmL7I1PjbzCZ0r20P4sMfnTdnBKJhA2fmjB0NzjbH92JMYZhsA1SS9j+yNT4G0ym1Jr2xwUPiIiIiIjIJjD4ISIiIiIim8Dgh4iIiIiIbAKDHyIiIiIisgkMfoiIiIiIyCYw+CEiIiIiIpvA4IeIiIiIiGwCgx8iIiIiIrIJDH6IiIiIiMgmMPghIiIiIiKbwOCHLErXrl1hZ2fXZEtOThbzTJ8+HV5eXnB3d0dsbCzKysr0XkOtViMmJgbt27eHTCbDzJkzUV9fb+yqEBEREZGROZq6AMbU9Y0vxb8vLYsxYUnofh0/fhwNDQ3i49OnT+PPf/4znn32WXFfdnY2duzYAalUipSUFIwZMwZHjx4FADQ0NCAmJgYKhQJ5eXm4cuUKJkyYACcnJyxdutTo9bFkuu8Tv0tkamyL5o2fD9H94XfHMGwq+CHL17lzZ73Hy5Ytw8MPP4wnnngCP//8MwBgyZIlGDZsGAAgIyMDPXv2REFBAQYNGoT9+/fj7Nmz+OqrryCXyxESEoLFixdj1qxZWLBgAZydnY1eJ0vHiwpERERkKTjsjSxWbW0tPv74Y7z00kuws7NDcXExACA8PFzMExgYCD8/P+Tn5wMA8vPzERwcDLlcLuZRqVTQarU4c+ZMs+9TU1MDrVartxERERGR5WHwQxZr9+7dqKiowMSJEwEA5eXlAAAPDw+9fHK5HBqNBgCg0Wj0Ah9dui6tOWlpaZBKpeLm6+vbhrUgImq9X375BePHj4eXlxfatWuH4OBgnDhxQkwXBAHz5s2Dt7c32rVrh8jISJw/f17vNa5evYq4uDhIJBJ4eHhg0qRJqKqqMnZViIhMgsEPWawPPvgA0dHR8PHxMej7pKamorKyUtwuX75s0PcjImrOtWvXMHjwYDg5OWHv3r04e/YsVq1ahY4dO4p51qxZg7Vr12Ljxo0oLCyEm5sbVCoVbty4IeaJi4vDmTNnkJOTg6ysLBw5cgSJiYmmqBIRkdFxzg9ZpP/85z/46quvsHPnTnGfTCYDAFRUVEAikYj7y8rKoFAoAAAKhQLHjh3Tey3danC6PLdzcXGBi4tLm5afiOheLV++HL6+vsjIyBD3BQQEAIA4HHfDhg2YM2cORo0aBQDYsmUL5HI5du/ejbFjx+LcuXPIzs7G8ePH0b9/fwDAunXrMGLECKxcudLgF5OIiEyNPT9kkTIyMiCTyRAT878J9iEhIQCA3NxccV9JSQnUajWUSiUAQKlU4tSpU+IQOQDIycmBRCJBUFCQcQpPRHQfPv/8c/Tv3x/PPvssZDIZ+vXrh02bNunlKSsrQ2RkpPhYKpUiLCxMb96jh4eHGPgAQGRkJOzt7VFYWNjs+3Leo204cuQIRo4cCR8fH9jZ2WH37t1iWl1dHWbNmoXg4GC4ubnBx8cHEyZMQGlpaZPXefnll+84pPL777/H448/DldXV/j6+mLFihWGrprF6/rGl3qLC9GDYfBDFqexsREZGRlISEiAo+P/Oi+lUikAYPbs2Th06BCKiorw4osvQqlUYtCgQQCAqKgoBAUFIT4+Ht999x327duHOXPmIDk5mb07RGTW/v3vf2PDhg3o1q0b9u3bh6SkJLz22mvYvHmzXr7m5jXeOu9R10uu4+joCE9PT857tHHV1dXo27cv0tPTm6T9/vvv+PbbbzF37lx8++232LlzJ0pKSvCXv/ylSd4ffvihxSGVWq0WUVFR8Pf3R1FREd566y0sWLAA7733nkHrRnQrDnsji/PVV19BrVbjpZdeajZdpVIhNjYWNTU1UKlUWL9+vZjm4OCArKwsJCUlQalUws3NDQkJCVi0aJGxik9ED8CW73vR2NiI/v37i/ck69evH06fPo2NGzfi6aefNtj7pqamYtq0aeJjrVbLAMgKRUdHIzo6utk0qVSKnJwcvX1///vfMXDgQKjVavj5+aGkpAQAsHbtWoSFhQFoOqRy69atqK2txYcffghnZ2f06tULxcXFWL16NeedkdGw54csTlRUFARBQPfu3ZtNX7VqFa5evYrq6mrs3LmzyVwef39/7NmzB7///jt+/fVXrFy5Uq8HiehOuNoWmYq3t3eT4bk9e/aEWq3W26ebx3jr41vnPd467BcA6uvrcfXq1TvOe5RIJHobUWVlJezs7MQVVnXzaR999FExz+1DKvPz8zF06FC9e+qpVCqUlJTg2rVrzb4Ph11SW2PwQ0TUSlxti0xp8ODB4tV1nR9//BH+/v7iY7lcjgMHDoiPtVotCgsL9eY9VlRUoKioSMxz8OBBNDY2ilfrDUU3b4FzFyzfjRs3MGvWLIwbN04Mhm8PuoGmQyp5uwkyB7zcTUTUSlxti0zpb3/7Gx577DEsXboUzz33HI4dO4b33ntPb75EUlIS3nzzTXTr1g0BAQGYO3cufHx8MHr0aAA3e4qGDx+OyZMnY+PGjairq0NKSgrGjh3LtketUldXh+eeew6CIGDDhg0Gfz8Ou6S2xp4fIqJW4mpbZEoDBgzArl278I9//AO9e/fG4sWLsWbNGsTFxYl5pk6diilTpiAxMREDBgxAVVUVsrOz4erqKubZunUrAgMDERERgREjRmDIkCGccE6togt8/vOf/4grperc3qMDNB1SqVAomh2WqUtrDoddUltjz4+FseXJvkSmpltta9q0afh//+//4fjx43jttdfg7OysN+HcEKttLVy4sI1rY9lsdejUU089haeeeqrFdDs7OyxatOiOi7h4enpi27ZthigeWTFd4HP+/HkcOnQIXl5eeukDBw4EAJw8eRJPPPEEgKZDKpVKJWbPno26ujo4OTkBuHm7iR49eugNHyYyJPb8kFFwrDdZg8bGRjz66KNYunQp+vXrh8TERHH4kCGlpqaisrJS3C5fvmzQ9yMi21NVVYXi4mIUFxcDAC5evIji4mKo1WrU1dXhmWeewYkTJ7B161Y0NDRAo9FAo9GgtrYWANCjRw8AwGuvvYZjx47h6NGjTYZUvvDCC3B2dsakSZNw5swZ/POf/8Q777yjN6yNyNDY82Ol2ENEhnC34NXa211Lq23961//0ttXVlYGb29vvce6m/De72pbvA8VERnSiRMn8OSTT4qPdQFJQkICFixYgM8//xzA/24ornPo0CGEh4eLj7t3746IiAjY29sjNjYWa9euFdOkUin279+P5ORkhIaGolOnTpg3bx4XfCGjYvBDRNRK97Lalu4EQbfaVlJSEgD91bZCQ0MBGG+1LSKiloSHh0MQhBbT75R2qw8++OCO83L69OmDr7/++p7LR9RWGPwQAOu/Yk/UFrjaFhERkWVj8NOMW4f2MBgwHAZcZGl0q22lpqZi0aJFCAgIEFfb0q3ANnXqVDQ0NCAxMREVFRUYMmRIs6ttpaSktDg0hIiIiAyDwQ8R0T3galtERESWi6u9UZvham5EREREZM4Y/BARkVXgBRgiIrobBj9ERERkVAxUichUOOeHLAYXoiCyTTxJJiKitsKeHxvEK25EREREZItstueHyywbB4MsIiIiIjIX7PkhIiIiIiKbYLM9P0RERGQ47PknInPEnh9qEecGEREREZE1Yc8PGR0DKiIiIiIyBfb8PAD2jBARERERWQ4GP0REREREZBM47I2I2hxvSEtERETmiMHPLTiEzTR4okxEZB0e5HeU998jImNg8ENERERmgxfEiMiQOOeHiIiIiIhsAnt+iOiuOCSUiIiIrAGDnzbALnoiIiIiIvPHYW9ERERERGQTGPzQPeGNXYnIkvCYRUREt7qn4CctLQ0DBgxAhw4dIJPJMHr0aJSUlOjluXHjBpKTk+Hl5QV3d3fExsairKxML49arUZMTAzat28PmUyGmTNnor6+/sFrQ0RERBZDF5wyQCUiY7mn4Cc3NxfJyckoKChATk4O6urqEBUVherqajHP3/72N3zxxRfYsWMHcnNzUVpaijFjxojpDQ0NiImJQW1tLfLy8rB582ZkZmZi3rx5bVcrslr11/+L/36xEl5eXmjXrh2Cg4Nx4sQJvTxLliyBt7c32rVrh8jISJw/f14v/erVq4iLi4NEIoGHhwcmTZqEqqoqY1aDiIiIiEzgnhY8yM7O1nucmZkJmUyGoqIiDB06FJWVlfjggw+wbds2DBs2DACQkZGBnj17oqCgAIMGDcL+/ftx9uxZfPXVV5DL5QgJCcHixYsxa9YsLFiwAM7Ozm1XOzNhjTduM0WdGm5UQfPx/8HVrw/27t2Lzp074/z58+jYsaNevnfffRebN29GQEAA5s6dC5VKhbNnz8LV1RUAEBcXhytXrogB/IsvvojExERs27bNaHUhIiIiuhdcYKttPNBqb5WVlQAAT09PAEBRURHq6uoQGRkp5gkMDISfnx/y8/MxaNAg5OfnIzg4GHK5XMyjUqmQlJSEM2fOoF+/fk3ep6amBjU1NeJjrVb7IMWme2QuwxG0BZ/CUdIJnWKmYuDAgQCAgIAAMV0QBADAjBkzMGrUKADAli1bIJfLsXv3bowdOxbnzp1DdnY2jh8/jv79+wMA1q1bhxEjRmDlypXw8fExcq2IiIiIyFjue8GDxsZGTJ06FYMHD0bv3r0BABqNBs7OzvDw8NDLK5fLodFoxDy3Bj66dF1ac9LS0iCVSsXN19f3fotNFuyPC4VwVnTDr7vTIJPJ0K9fP2zatElMv3TpEgAgPDxc3CeVShEWFob8/HwAQH5+Pjw8PMTABwAiIyNhb2+PwsLCZt+3pqYGWq1WbyMiw+AcECIiMqT77vlJTk7G6dOn8c0337RleZqVmpqKadOmiY+1Wi0DIAMx5xOOugoN6k7ugWTAaOz7xzocP34cr732GpydnZGQkIDy8nIAgEwm03ve7cH37emOjo7w9PS8Y/C9cOFCA9SIiIiIiIzpvnp+UlJSkJWVhUOHDqFLly7ifoVCgdraWlRUVOjlLysrg0KhEPPcvvqb7rEuz+1cXFwgkUj0NrJBggAX+cPo+EQC+vXrh8TEREyePBkbN2406NumpqaisrJS3C5fvmzQ9yMiIiIiw7in4EcQBKSkpGDXrl04ePCg3nwLAAgNDYWTkxMOHDgg7ispKYFarYZSqQQAKJVKnDp1SrxKDwA5OTmQSCQICgp6kLoYBYdkmI6De0c4dfLT29ezZ0+o1WoA/+vxubVtAU2D79vT6+vrcfXqVQbfRERERFbunoKf5ORkfPzxx9i2bRs6dOgAjUYDjUaDP/74A8DN+RWTJk3CtGnTcOjQIRQVFeHFF1+EUqnEoEGDAABRUVEICgpCfHw8vvvuO+zbtw9z5sxBcnIyXFxc2r6GZDVc/hSEuqs/6+378ccf4e/vDwDo2rUrgJtLsutotVoUFhbqBd8VFRUoKioS8xw8eBCNjY0ICwszcA2IiIiIyJTuac7Phg0bAOhPKAduLmc9ceJEAMDbb78Ne3t7xMbGoqamBiqVCuvXrxfzOjg4ICsrC0lJSVAqlXBzc0NCQgIWLVr0YDUhq3CnJbQlA0ZB8/FMVOZ/ggsXeuDYsWN477338N577wEA7OzsAABvvfUWgoODxaWufXx8MHr0aAA3e4qGDx8uDperq6tDSkoKxo4dy5XeiIiIiKzcPQU/uqWE78TV1RXp6elIT09vMY+/vz/27NlzL299X6zx/jq2zMW7Ozo/PRsVuZvRu/c/ERAQgDVr1iAuLk4v3yuvvILExERUVFRgyJAhyM7OFu/xAwBbt25FSkoKIiIixEB97dq1xq4OERERERnZfS91TWQK7R8ZCJ9J6bhx4wbOnTuHyZMnN8kze/ZsaDQa3LhxA1999RW6d++ul+7p6Ylt27bh+vXrqKysxIcffgh3d3djVYGI6L4sWLAAdnZ2eltgYKBenunTp8PLywvu7u6IjY1tssCQWq1GTEwM2rdvD5lMhpkzZ6K+vt6Y1SAiMqkHuskpmRcuwkBEZN169eqFr776Snzs6Kj/M56dnY0dO3ZAKpUiJSUFY8aMwdGjRwEADQ0NiImJgUKhQF5eHq5cuYIJEybAyckJS5cuNWo9iIhMhT0/rWTqFd7a4v1NXQciInowjo6OUCgU4tapUycAQGVlJQBgyZIlGDZsGEJDQ5GRkYG8vDwUFBQAAPbv34+zZ8/i448/RkhICKKjo7F48WKkp6ejtrbWZHUi83DkyBGMHDkSPj4+sLOzw+7du/XSBUHAvHnz4O3tjXbt2iEyMhLnz59v8jovv/wyJBIJPDw8MGnSJFRVVemlf//993j88cfh6uoKX19frFixwpDVImrC6oIfc16K2lDlMtf6EhE9iPs9tpnz78CDOn/+PHx8fPDQQw8hLi5OXOq/uLgYgP6CRIGBgfDz80N+fj4AID8/H8HBwZDL5WIelUoFrVaLM2fOtPieNTU10Gq1ehtZn+rqavTt27fFOdsrVqzA2rVrsXHjRhQWFsLNzQ0qlQo3btzQy/fDDz8gJycHWVlZOHLkCBITE8U0rVaLqKgo+Pv7o6ioCG+99RYWLFggLlxEZAwc9mYCt/4gG3oxBmv88SciskVhYWHIzMxEjx49cOXKFSxcuBCPP/44Tp8+Ld6/zMPDQ+85crkcGo0GAKDRaPQCH126Lq0laWlpWLhwYRvWhMxRdHQ0oqOjm00TBAFr1qzBnDlzMGrUKADAli1bIJfLsXv3bowdOxYlJSUAgLVr14q3jli3bh1GjBiBlStXwsfHB1u3bkVtbS0+/PBDODs7o1evXiguLsbq1av1giQiQ7K6nh8iIkPhhHMypejoaDz77LPo06cPVCoV9uzZg4qKCnzyyScGfd/U1FRUVlaK2+XLlw36fmR+Ll68CI1Gg8jISHGfVCpFWFiY2LN47NgxAMCjjz4q5omMjIS9vT0KCwsB3Ox9HDp0KJydncU8KpUKJSUluHbtmjGqQsSeH1Oz9OW42bNEtoYTzslceHh4oHv37rhw4QIee+wxAEBFRQUkEomYp6ysDAqFAgCgUCjEE9Rb03VpLXFxceFNyG2crmewuZ5DXdrtF3qAm8dHT09Pvd7HgICAJq+hS+vYsWOT16ipqUFNTY34mMMu6UGx54eIWmSt8yYeBCeck7moqqrCTz/9BG9vb4SEhAAAcnNzxfSSkhKo1WoolUoAgFKpxKlTp8QhcgCQk5MDiUSCoKAgo5adqLXS0tIglUrFzdfX19RFIgvH4KeNWdLJojVPCiYyFE44J1OZMWMGcnNzcenSJeTl5eHpp5+Gg4MDxo0bB6lUCuDmfc4OHTqEoqIivPjii1AqlRg0aBAAICoqCkFBQYiPj8d3332Hffv2Yc6cOUhOTmbPDt2Rrmfw9t6dW3sWb+8VAoD6+npcvXpVr/exude49T1ux2GX1NZsPvhhAEBEraWbcJ6dnY0NGzbg4sWLePzxx3H9+nWDTzjnlU/6+eefMW7cOPTo0QPPPfccvLy8UFBQgM6dO4t5VCoVYmNjMXToUCgUCuzcuVNMc3BwQFZWFhwcHKBUKjF+/HhMmDABixYtMkV1yIIEBARAoVDgwIED4j6tVovCwkKxZ3HgwIEAgJMnT4p5Dh48iMbGRnEBBKVSiSNHjqCurk7Mk5OTgx49ejQ75A24OexSIpHobUQPgnN+iIha6daVkPr06YOwsDD4+/sbZcL5tGnTxMdarZYBkA3avn37XfOsWrUKmzZtajHd398fe/bsactikZWoqqrChQsXxMcXL15EcXExPD094efnh6lTp+LNN99Et27dEBAQgLlz58LHxwejR48GAPTo0QMA8Nprr2HTpk2oq6tDSkoKxo4dCx8fHwDACy+8gIULF2LSpEmYNWsWTp8+jXfeeQdvv/220etLtovBjxljbxSReeOEcyKyFidOnMCTTz4pPtZdcElISEBmZib+7//+D9XV1UhMTERFRQWGDBmC7OxsuLq66r1O9+7dERERAXt7e8TGxmLt2rVimlQqxf79+5GcnIzQ0FB06tQJ8+bN4zLXZFQMfiwUAyOyFJa+ouGd6Cacx8fH6004j4+PB9D8hPMlS5agvLwcMpkMACecE5F5CA8PhyAILabb2dlh0aJFdx0m+cEHH9xxaFqfPn3w9ddf33c5iR4Ugx8iolaaMWMGRo4cCX9/f5SWlmL+/PnihHNdz8zs2bPRpUsXSCQSTJkypcUJ5ytWrIBGo+GEcyIiIiNi8GMg7Jl5MLf+/6yxx4Ask27C+W+//YbOnTtjyJAh4oRz3QpsugnnNTU1UKlUWL9+vfh83YTzpKQkKJVKuLm5ISEhgRPOiYiIjITBDxFRK3HCORERkWWzieDH0nphLK28RERERESWwCaCH7JsDAaJiIiIqC0w+KE2x2CFiIiIiMyRvakLQEREREREZAzs+SEiPey5IyIiImvF4OcueCJIRERERGQdOOyNiIiIzFLXN77kRUgialPs+TETPLgTERERERkWgx8iIjI5XgAiIiJj4LA3IiIiIiKyCQx+iIiIiIjIJjD4ISIiIiIim8Dgh4iIiIiIbAKDHyIiIiIisgkMfoiIiIiIyCYw+CGLUvHNVvxn+VOws7MTt8DAQL0806dPh5eXF9zd3REbG4uysjK9dLVajZiYGLRv3x4ymQwzZ85EfX29MatBRET3QHezUy6JTkQPivf5IYvj1MkP6lOF4mNHR/1mnJ2djR07dkAqlSIlJQVjxozB0aNHAQANDQ2IiYmBQqFAXl4erly5ggkTJsDJyQlLly41aj2IiIiIyLgY/JDlsXeAQqFosruyshIAsGTJEgwbNgwAkJGRgZ49e6KgoACDBg3C/v37cfbsWXz11VeQy+UICQnB4sWLMWvWLCxYsADOzs5GrQoRERERGQ+HvZHFqb9WCh8fHzz00EOIi4uDWq0GABQXFwMAwsPDxbyBgYHw8/NDfn4+ACA/Px/BwcGQy+ViHpVKBa1WizNnzjT7fjU1NdBqtXobEREREVkeBj9kUVy8e8BrxN+QnZ2NDRs24OLFi3j88cdx/fp1lJeXAwA8PDz0niOXy6HRaAAAGo1GL/DRpevSmpOWlgapVCpuvr6+bVwrIiIiIjIGBj9kUdo93B9ugUPQp08fqFQq7NmzBxUVFfjkk08M9p6pqamorKwUt8uXLxvsvYiIiIjIcBj8kEXz8PBA9+7dceHCBchkMgBARUWFXp6ysjJxjpBCoWiy+pvucXPziADAxcUFEolEbyMiIiIiy8PghyxaVVUVfvrpJ3h7eyMkJAQAkJubK6aXlJRArVZDqVQCAJRKJU6dOiUOkQOAnJwcSCQSBAUFGbXstobL1JIpcalkIiICuNobWZhrBz9Au0cG4tKlXigtLcX8+fPh4OCAcePGwcXFBQAwe/ZsdOnSBRKJBFOmTIFSqcSgQYMAAFFRUQgKCkJ8fDxWrFgBjUaDOXPmIDk5WXw+ERGZP10ge2lZjIlLQkSWhD0/ZFHqr/8X//3iLfTo0QPPPfccvLy8UFBQgM6dO4t5VCoVYmNjMXToUCgUCuzcuVNMc3BwQFZWFhwcHKBUKjF+/HhMmDABixYtMkV1iIiIiMiI2PNDFqXzqFkA7nylb9WqVdi0aVOL6f7+/tizZ0+bl42IiAyPQxeJ6EGw54eIiIiIiGwCgx8iIiIiIrIJDH6IiIiIiCwIV6+8fwx+iIiIiIjIJnDBAyIiMhleuSQiImNizw8REZEFWrZsGezs7DB16lRx340bN5CcnAwvLy+4u7sjNjYWZWVles9Tq9WIiYlB+/btIZPJMHPmTNTX1xu59EREpsHgh4iIyMIcP34c7777Lvr06aO3PzU1FV988QV27NiB3NxclJaWYsyYMWJ6Q0MDYmJiUFtbi7y8PGzevBmZmZmYN2+esatARGQSDH6IiIgsSFVVFeLi4rBp0yZ07NhRL+2jjz7C6tWrMWzYMISGhiIjIwN5eXkoKCgAAOzfvx9nz57Fxx9/jJCQEERHR2Px4sVIT09HbW2tKapDRGRUDH6IiIgsSHJyMmJiYhAZGdkkra6uTm9/YGAg/Pz8kJ+fDwDIz89HcHAw5HK5mEelUkGr1eLMmTOGLzxZrIaGBsydOxcBAQFo164dHn74YSxevBiCIOjlW7JkCby9vdGuXTtERkbi/PnzeulXr15FXFwcJBIJPDw8MGnSJFRVVRmzKmTj7jn4OXLkCEaOHAkfHx/Y2dlh9+7deukTJ06EnZ2d3jZ8+HC9PGz4ROaHy2YSmb/t27fj22+/RVpaWrPpzs7O8PDw0Nsnl8uh0WgAABqNRi/w0aXr0ppTU1MDrVart5HtWb58OTZs2IC///3vOHfuHJYvX44VK1Zg3bp1evneffddbNy4EYWFhXBzc4NKpcKNGzfE9Li4OJw5cwY5OTnIysrCkSNHkJiYaOzqkA275+Cnuroaffv2RXp6eot5hg8fjitXrojbP/7xD710NnwisgaccE7GdPnyZbz++uvYunUrXF1djfa+aWlpkEql4ubr62u09ybzkZeXh1GjRiEmJgZdu3bFM888g6ioKBw7dgwAxB6gGTNmYNSoUejTpw+2bNmC0tJS8UL5uXPnkJ2djffffx9hYWEYMmQI1q1bh+3bt6O0tNRUVSMbc8/BT3R0NN588008/fTTLeZxcXGBQqEQt1vHJLPhE5E14IRzMraioiKUl5fj0UcfhaOjIxwdHZGbm4u1a9fC09MTAFBbW4uKigq955WVlUGhUAAAFApFk2Bc91iX53apqamorKwUt8uXL7dxzcgSPPbYYzhw4AB+/PFHAMB3332Hb775BtHR0QCAS5cuAQDCw8PF50ilUoSFhekNu/Tw8ED//v3FPJGRkbC3t0dhYWGz78ueR2prBpnzc/jwYchkMvTo0QNJSUn47bffxDQ2fCKydJxwTqYQERGBU6dOobi4WNz69++PuLg4fPPNNwAAJycnHDhwQHxOSUkJ1Go1lEolAECpVOLUqVMoLy8X8+Tk5EAikSAoKKjZ93VxcYFEItHbyPa88cYbGDt2LAIDA+Hk5IR+/fph6tSpiIuLAwCxTclkMr3n3T7s8vZ0R0dHeHp6tjjskj2P1NbaPPgZPnw4tmzZggMHDmD58uXIzc1FdHQ0GhoaALDhE5HlM/aEc14AIgDo0KEDevfurbe5ubnBy8tLDFzi4+Mxbdo0HDp0CEVFRXjxxRehVCoxaNAgAEBUVBSCgoIQHx+P7777Dvv27cOcOXOQnJwMFxcXU1aPzNwnn3yCrVu3Ytu2bfj222+xefNmrFy5Eps3bzbo+7LnkdqaY1u/4NixY8W/g4OD0adPHzz88MM4fPgwIiIi7us1U1NTMW3aNPGxVqtlAEREJqGbcH78+PFm0w0x4TwtLQ0LFy58wJKTLUhLS4OrqytiY2NRU1MDlUqF9evXi+kODg7IyspCUlISlEol3NzckJCQgEWLFpmw1GQJZs6cKfb+ADfP8f7zn/8gLS0NCQkJ4oXt8vJydO/eXXxeWVkZQkJCANwcWnlrryMA1NfX4+rVqy0Ou3RxcWFgTm2qzYOf2z300EPo1KkTLly4gIiICDZ8IrJYugnnOTk5Rp1wzgtA1JLDhw8DgNgb6OrqivT09DsuSuTv7489e/YYo3hkRX7//XfY2+sPGHJwcEBjYyMAoGvXrgCA3NxcDBkyBMDNdllYWIikpCQAN4ddVlRUoKioCKGhoQCAgwcPorGxEWFhYUaqCdk6gwc/P//8M3777Td4e3sDYMMnIst164RznYaGBhw5cgR///vfAfxvwvmtvT+3TzjXrY50a7ourTm8AETUOrrl+i8tizFxSazPyJEjsWTJEvj5+aFXr144efIkVq9ejZdeegkAYGdnBwB46623EBwcjICAAMydOxc+Pj4YPXo0AKBnz54YPnw4Jk+ejI0bN6Kurg4pKSkYO3YsfHx8TFU1sjH3POenqqpKnGgJABcvXkRxcTHUajWqqqowc+ZMFBQU4NKlSzhw4ABGjRqFRx55BCqVCoB+wz927BiOHj3Khk9EFsFUE86JiExt3bp1eOaZZ/DXv/4VPXv2xIwZM/DKK69g8eLFevleeeUVJCYmYsCAAaiqqkJ2drZeT/nWrVsRGBiIiIgIjBgxAkOGDMF7771n7OqQDbvnnp8TJ07gySefFB/rhmIkJCRgw4YN+P7777F582ZUVFTAx8cHUVFRWLx4sd5Vy61btyIlJQURERGwt7dHbGws1q5d2wbVISIyHN2E81u1NOHc09MTEokEU6ZMaXHC+YoVK6DRaDjhnIjMXocOHbBmzRqsWbPmjvlmz56N5cuXt5ju6emJbdu2tXHpiFrvnoOf8PBw8UZWzdm3b99dX4MNn4isFSecExERmS+Dz/khIrJmnHBORERkOQxyk1MiIiIiIiJzw+CHiIiIiIhsAoe9ERERERFZIN3y7gCXeG8t9vwQEREREZFNYM8PERERWaxbr3wTEd0Ngx8iIiIiIjPAYN7wGPwQkVFxfDIRERGZCuf8EBERERGRTWDwQ0RENqXrG19yaAkRkY1i8ENERERERDaBwQ9ZrGXLlsHOzg5Tp07V2z99+nR4eXnB3d0dsbGxKCsr00tXq9WIiYlB+/btIZPJMHPmTNTX1xux5ERERERkClzwgCzS8ePH8e6776JPnz5N0rKzs7Fjxw5IpVKkpKRgzJgxOHr0KACgoaEBMTExUCgUyMvLw5UrVzBhwgQ4OTlh6dKlxq4GkU3ikDMiIjIV9vyQxWms/QNxcXHYtGkTOnbsKO6vrKwEACxZsgTDhg1DaGgoMjIykJeXh4KCAgDA/v37cfbsWXz88ccICQlBdHQ0Fi9ejPT0dNTW1pqkPkRERERkHAx+yOJczdmAmJgYREZG6u0vLi4GAISHh4v7AgMD4efnh/z8fABAfn4+goODIZfLxTwqlQparRZnzpxp9v1qamqg1Wr1NiIiIiKyPBz2Rhal+mwuajU/IS2t6bCZ8vJyAICHh4fefrlcDo1GAwDQaDR6gY8uXZfWnLS0NCxcuPBBi05EZob3nCIisj3s+SGLUa/9FVcPbEKnkTPg6upqtPdNTU1FZWWluF2+fNlo701EREREbYfBD1mMWs0FNP5egSuZr8PR0RGOjo7Izc3F2rVr4ejoiM6dOwMAKioq9J5XVlYGhUIBAFAoFE1Wf9M91uW5nYuLCyQSid5GRERERJaHwQ9ZDFf/vvB+6e/wfnEtiouLUVxcjP79+yMuLg7FxcXo168fACA3N1d8TklJCdRqNZRKJQBAqVTi1KlT4hA5AMjJyYFEIkFQUJBxK0RERERERsU5P2Qx7F3aw7lzVwBA7969AQBubm7w8vJC7969xYUIZs+ejS5dukAikWDKlClQKpUYNGgQACAqKgpBQUGIj4/HihUroNFoMGfOHCQnJ8PFxcUk9SIiorbF+VxE1BL2/JDVUalUiI2NxdChQ6FQKLBz504xzcHBAVlZWXBwcIBSqcT48eMxYcIELFq0yIQlJiIiIiJjYM8PWbTDhw832bdq1Sps2rSpxef4+/tjz549BiwVEREREZkj9vwQEREREZFNYPBDREREREQ2gcEPERERERHZBAY/RERERERkExj8EBERERGRTWDwQ0RERERENoHBDxERERER2QQGP0REREREZBMY/BARERERkU1g8ENERERERDaBwQ8REREREdkEBj9EREQWYMOGDejTpw8kEgkkEgmUSiX27t2rl2f69Onw8vKCu7s7YmNjUVZWppeuVqsRExOD9u3bQyaTYebMmaivrzdmNYiITIrBDxERkQXo0qULli1bhqKiIpw4cQLDhg3DqFGjcObMGTFPdnY2duzYgdzcXJSWlmLMmDFiWkNDA2JiYlBbW4u8vDxs3rwZmZmZmDdvnimqQ0RkEgx+iIiILMDIkSMxYsQIdOvWDd27d8eSJUvg7u6OgoICVFZWAgCWLFmCYcOGITQ0FBkZGcjLy0NBQQEAYP/+/Th79iw+/vhjhISEIDo6GosXL0Z6ejpqa2tNWTWyEL/88gvGjx8PLy8vtGvXDsHBwThx4oReniVLlsDb2xvt2rVDZGQkzp8/r5d+9epVxMXFQSKRwMPDA5MmTUJVVZUxq0E2jsEPEVErcdgRmYuGhgZs374d1dXVUCqVKC4uBgCEh4eLeQIDA+Hn54f8/HwAQH5+PoKDgyGXy8U8KpUKWq1Wr/fodjU1NdBqtXqbJen6xpfo+saXpi6Gxbt27RoGDx4MJycn7N27F2fPnsWqVavQsWNHvXzvvvsuNm7ciMLCQri5uUGlUuHGjRtielxcHM6cOYOcnBxkZWXhyJEjSExMNHZ1yIY5mroARGS7dCckl5bFmLgkraMbdtStWzcIgoDNmzdj1KhROHnyJHx9fQH8b9iRVCpFSkoKxowZg6NHjwL437AjhUKBvLw8XLlyBRMmTICTkxOWLl1qyqqRhTh16hSUSiVu3LgBd3d37Nq1C0FBQWKA4+HhoZdfLpdDo9EAADQajV7go0vXpbUkLS0NCxcubMNakCVavnw5fH19kZGRIe4LCAgQ/xYEAQAwY8YMjBo1CgCwZcsWyOVy7N69G2PHjsW5c+eQnZ2N48ePo3///gCAdevWYcSIEVi5ciV8fHyMWCOyVez5ISJqJQ47IlPr0aMHiouLUVhYiKSkJCQkJODs2bMGfc/U1FRUVlaK2+XLlw36fmSePv/8c/Tv3x/PPvssZDIZ+vXrh02bNonply5dAqDf+yiVShEWFqbX++jh4SEGPgAQGRkJe3t7FBYWNvu+lt7zSOaHwQ8R0X0w5rAjIh1nZ2c88sgjCA0NRVpaGvr27Yt33nkHMpkMAFBRUaGXv6ysDAqFAgCgUCiaDMPUPdblaY6Li4s41FO3ke3597//jQ0bNqBbt27Yt28fkpKS8Nprr2Hz5s0AgPLycgAQ26LO7b2Pt6c7OjrC09Ozxd7HtLQ0SKVScdP1shPdLwY/RDZMNxae4+Fb79SpU3B3d4eLiwteffVVcdiR7offEMOOeOWTWtLY2IiamhqEhIQAAHJzc8W0kpISqNVqKJVKAIBSqcSpU6fEtgoAOTk5kEgkCAoKMmq5yfI0Njbi0UcfxdKlS9GvXz8kJiZi8uTJ2Lhxo0Hflz2P1NY454eI6B7ohh1VVlbi008/RUJCgt4JpyFwzgUBN08Co6Oj4efnh+vXr2Pbtm04fPgw9u3bB6lUCgCYPXs2unTpAolEgilTpkCpVGLQoEEAgKioKAQFBSE+Ph4rVqyARqPBnDlzkJycDBcXF1NWjSyAt7d3kyC5Z8+e+Ne//gXgfz0+5eXl6N69u5inrKxMDM4VCoVe8A0A9fX1uHr1aou9jy4uLmyf1KbY80NEdA9MMeyIVz4JuHlSOWHCBPTo0QMRERE4fvw49u3bhz//+c9iHpVKhdjYWAwdOhQKhQI7d+4U0xwcHJCVlQUHBwcolUqMHz8eEyZMwKJFi0xRHbIwgwcPRklJid6+H3/8Ef7+/gCArl27AtDvfdRqtSgsLNTrfayoqEBRUZGY5+DBg2hsbERYWJiBa0B0E3t+iIgeQHPDjuLj4wE0P+xoyZIlKC8vF4Ol1gw74pVPAoAPPvjgrnlWrVqlNwn9dv7+/tizZ09bFotsxN/+9jc89thjWLp0KZ577jkcO3YM7733Ht577z0AgJ2dHQDgrbfeQnBwMAICAjB37lz4+Phg9OjRAG72FA0fPlwcLldXV4eUlBSMHTuWK72R0TD4ISJqJQ47IiJbNWDAAOzatQupqalYtGgRAgICsGbNGsTFxenle+WVV5CYmIiKigoMGTIE2dnZcHV1FdO3bt2KlJQUREREwN7eHrGxsVi7dq2xq0M2jMEPEVEr6YYdXblyBVKpFH369BGHHekWIdANO6qpqYFKpcL69evF5+uGHSUlJUGpVMLNzQ0JCQkcdkREFuGpp57CU089dcc8s2fPxvLly1tM9/T0xLZt29q6aEStxuCHiKiVOOyIyHLduqqlpdxYmeheWNqNw02FCx4QEREREZFNYPBDREREREQ2gcEPERERERHZBM75ISIio7h1zgUREZEpsOeHiIhsXtc3vmRwRkRkA+45+Dly5AhGjhwJHx8f2NnZYffu3XrpgiBg3rx58Pb2Rrt27RAZGYnz58/r5bl69Sri4uIgkUjg4eGBSZMmoaqq6oEqQrbh+sk9KP0wBRKJBBKJBEqlEnv37tXLM336dHh5ecHd3R2xsbEoKyvTS1er1YiJiUH79u0hk8kwc+ZM1NfXG7MaRERERGQC9xz8VFdXo2/fvkhPT282fcWKFVi7di02btyIwsJCuLm5QaVS4caNG2KeuLg4nDlzBjk5OcjKysKRI0eQmJh4/7Ugm+HQwQsdn0hAUVERTpw4gWHDhmHUqFE4c+aMmCc7Oxs7duxAbm4uSktLMWbMGDGtoaEBMTExqK2tRV5eHjZv3ozMzEzMmzfPFNUhIiIiIiO65zk/0dHRiI6ObjZNEASsWbMGc+bMwahRowAAW7ZsgVwux+7duzF27FicO3cO2dnZOH78OPr37w8AWLduHUaMGIGVK1fCx8fnAapD1q79I2EAgG7dugEAlixZgg0bNqCgoAASiUTcN2zYMABARkYGevbsiYKCAgwaNAj79+/H2bNn8dVXX0EulyMkJASLFy/GrFmzsGDBAjg7O5umYkREZDS8HwqR7WrTOT8XL16ERqNBZGSkuE8qlSIsLAz5+fkAgPz8fHh4eIiBDwBERkbC3t4ehYWFzb5uTU0NtFqt3kbU0NCA7du3o7q6GkqlEsXFxQCA8PBwMU9gYCD8/Pz02l9wcDDkcrmYR6VSQavV6vUe3Yrtj4iIiMg6tGnwo9FoAEDvxFL3WJem0Wggk8n00h0dHeHp6SnmuV1aWhqkUqm4+fr6tmWxycLU/noJ7u7ucHFxwauvvopdu3YhKCgI5eXlAAAPDw+9/Le3v+bapy6tOWx/RERERNbBIlZ7S01NRWVlpbhdvnzZ1EUiE3Ly/BOKi4tRWFiIpKQkJCQk4OzZswZ7P7Y/IiIiIuvQpvf5USgUAICysjJ4e3uL+8vKyhASEiLm0V2h16mvr8fVq1fF59/OxcUFLi4ubVlUsmB2Dk545JFHAAChoaE4fvw43nnnHTz11FMAgIqKCnH+D3Cz/enalkKhwLFjx/ReT7caHNsfERERkXVr056fgIAAKBQKHDhwQNyn1WpRWFgIpVIJAFAqlaioqEBRUZGY5+DBg2hsbERYWFhbFodsRGNjI2pqasQAOzc3V0wrKSmBWq3Wa3+nTp3SC8BzcnIgkUgQFBRk1HITERERkXHdc89PVVUVLly4ID6+ePEiiouL4enpCT8/P0ydOhVvvvkmunXrhoCAAMydOxc+Pj4YPXo0AKBnz54YPnw4Jk+ejI0bN6Kurg4pKSkYO3YsV3qju7qWm4l2D/XHpUu9cP36dWzbtg2HDx/Gvn37IJVKAQCzZ89Gly5dIJFIMGXKFCiVSgwaNAgAEBUVhaCgIMTHx2PFihXQaDSYM2cOkpOT2btjQrfeXJKrLxEREZGh3HPwc+LECTz55JPi42nTpgEAEhISkJmZif/7v/9DdXU1EhMTUVFRgSFDhiA7Oxuurq7ic7Zu3YqUlBRERETA3t4esbGxWLt2bRtUh6xdQ3Ul/pu1Gj0+nQepVIo+ffpg3759+POf/yyuwqZSqRAbG4uamhqoVCqsX79efL6DgwOysrKQlJQEpVIJNzc3JCQkYNGiRaaqEpFVuzWwJSIiMrV7Dn7Cw8MhCEKL6XZ2dli0aNEdTyY9PT2xbdu2e31rInQa8TqAO/cOrFq1Cps2bWox3d/fH3v27GnzslkSnpASERGRLbKI1d6IiIiIiIgeFIMfIiIiIiKyCW261DUREREREZkOFxG6M/b8EBERERGRTWDwQ0RERERENoHBDxERERER2QQGP0REREREZBMY/BARERERkU1g8ENERERERDaBwQ8REREREdkEBj9ERERERGQTGPwQERGRTer6xpd6N4QkIuvH4IeIiIiIiGwCgx8iIiIiIrIJDH6IiIj+fxwGRURk3Rj8EBERERGRTWDwQ0RmhVfeiYiI2gZ/U5tyNHUBiIiIiEzp1pPDS8tiTFgSIjI09vwQERFZgLS0NAwYMAAdOnSATCbD6NGjUVJSopfnxo0bSE5OhpeXF9zd3REbG4uysjK9PGq1GjExMWjfvj1kMhlmzpyJ+vp6Y1aFiMhkGPwQERFZgNzcXCQnJ6OgoAA5OTmoq6tDVFQUqqurxTypqan44osvsGPHDuTm5qK0tBRjxowR0xsaGhATE4Pa2lrk5eVh8+bNyMzMxLx580xRJbJgy5Ytg52dHaZOnaq3f/r06Qy+yawx+CEiaiVeeSdTys7OxsSJE9GrVy/07dsXmZmZUKvVKCoqEvN89NFHWL16NYYNG4bQ0FBkZGQgLy8PBQUFAID9+/fj7Nmz+PjjjxESEoLo6GgsXrwY6enpqK2tNVXVyMIcP34c7777Lvr06dMkLTs7m8E3mTUGP0RErcQr72ROKisrAQCenp7ivrq6OkRGRoqPAwMD4efnh/z8fABAfn4+goODIZfLxTwqlQparRZnzpxp9n1qamqg1Wr1NmvGCeJ3VlVVhbi4OGzatAkdO3YU9+va45IlSxh8k1lj8ENkI3Q/6PxRv3+88k7morGxEVOnTsXgwYPRu3dvcb+zszM8PDz08srlcmg0GgCARqPRC3x06bq05qSlpUEqlYqbr69vG9aELE1ycjJiYmL0gmwAKC4uBgCEh4eL+xh8kzli8ENEdJ945Z1MJTk5GadPn8b27dsN/l6pqamorKwUt8uXLxv8Pck8bd++Hd9++y3S0tKapJWXlwMAg28yewx+iIjuA6+8k6mkpKQgKysLhw4dQpcuXfTSamtrUVFRobevrKwMCoUCAKBQKJrMQdM91uW5nYuLCyQSid5Gtufy5ct4/fXXsXXrVri6uhrtfRl8U1tj8ENEdB945Z2MTRAEpKSkYNeuXTh48CACAgKa5HFycsKBAwfExyUlJVCr1VAqlQAApVKJU6dOiVfpASAnJwcSiQRBQUGGrwRZrKKiIpSXl+PRRx+Fo6MjHB0dkZubi7Vr18LR0RGdO3cGAAbfZPZ4k1Mionuku/J+5MiRFq+839r7c/uP/7Fjx/Se05offxcXlzasgeFxblnbS05OxrZt2/DZZ5+hQ4cOYk+hVCoV88THx2PatGnw9PSERCLBlClToFQqMWjQIABAVFQUgoKCEB8fjxUrVkCj0WDOnDlITk62uDZGxhUREYFTp07p7XvxxRcRGBiIWbNmie0wNzcX8fHxAJoPvpcsWYLy8nLIZDIADL7J+Bj8EBG1kiAImDJlCnbt2oXDhw/f8cp7bGwsAP74U9vZsGEDAP0J5QCQkZEhriiYlpYGV1dXxMbGoqamBiqVCuvXrxfzOjg4ICsrC0lJSVAqlXBzc0NCQgIWLVpktHqQZerQoYPeEF8AcHNzg5eXF3r37i3ORZw9eza6dOnC4JvMFoMfIqJW4pV3MiVBEFpM0514urq6Ij09Henp6S3m9ff3x549e9q8fETAzQVcGHyTOWPwQ0TUSrzyTkT0P4cPH26yb9WqVdi0aVOLz2HwTabG4IeIqJV45Z2IiAyB8ySNh6u9kcWozP8EVzb/Deq3n4VMJsPo0aNRUlLSJN/06dPh5eUFd3d3xMbGNllZRq1WIyYmBu3bt4dMJsPMmTNRX19vrGoQERERGRVvcv4/DH7IYty4fBodHo2BYvxK5OTkoK6uDlFRUaiurtbLl52djR07diA3NxelpaXicCQAaGhoQExMDGpra5GXl4fNmzcjMzMT8+bNM3Z1iMiM6U4UeLJARGRdGPyQxZA/twjuwZFw7uyPvn37IjMzE2q1GkVFRQCAyspKAMCSJUswbNgwhIaGIiMjA3l5eSgoKAAA7N+/H2fPnsXHH3+MkJAQREdHY/HixUhPT0dtba3J6kZEREREhsfghyyWLtjx9PQEABQXFwPQn4weGBgIPz8/5OfnAwDy8/MRHBwMuVwu5lGpVNBqtThz5kyz71NTUwOtVqu3EREREZHlYfBDFqmxsRFTp07F4MGDxfsO6O5YfuvNJQFALpeLSxJrNBq9wEeXrktrTlpaGqRSqbj5+vq2ZVWIiIiIyEgY/JBFSk5OxunTp7F9+3aDv1dqaioqKyvF7fLlywZ/TyIiIiJqe1zqmizO1ZwNyCr/DkeOHEGXLl3E/TKZDABQUVEBiUQi7i8rK4NCoQAAKBQKHDt2TO/1dKvB6fLczsXFhTefJCIiIrIC7PkhiyEIAq7mbMDvP+bj4MGDCAgI0EsPCQkBAOTm5or7SkpKoFaroVQqAQBKpRKnTp0Sh8gBQE5ODiQSCYKCggxfCSIiIiIyGfb8kMW4mrMB1WdzIRszBx06dBDn6EilUrRr1w5SqRQAMHv2bHTp0gUSiQRTpkyBUqnEoEGDAABRUVEICgpCfHw8VqxYAY1Ggzlz5iA5OZm9O0RERERWjj0/ZDGqTu6BUFONsn+kwtvbW9z++c9/6uVTqVSIjY3F0KFDoVAosHPnTjHNwcEBWVlZcHBwgFKpxPjx4zFhwgQsWrTI2NUhIiIiIiNjzw9ZDP9ZWeLfl5bFtJhv1apV2LRpU8uv4++PPXv2tGnZiIiIiMj8MfghIrPU9Y0vxb/vFOyS+bj1MyMiIjJHDH6IiIiIbsMLMETWicEPEREREZENYFDPBQ+IiIiIiMhGMPghIiIiIiKbwOCHiIiIiIhsAoMfIiIiIiKyCQx+iIiIiIjIJnC1NyIiIqJW4EpZRJavzXt+FixYADs7O70tMDBQTL9x4waSk5Ph5eUFd3d3xMbGoqysrK2LQUT/v65vfMmbTxI9AH6HiG2AyHoYZNhbr169cOXKFXH75ptvxLS//e1v+OKLL7Bjxw7k5uaitLQUY8aMMUQxiIiIiIiIRAYZ9ubo6AiFQtFkf2VlJT744ANs27YNw4YNAwBkZGSgZ8+eKCgowKBBgwxRHCIiIiIiIsP0/Jw/fx4+Pj546KGHEBcXB7VaDQAoKipCXV0dIiMjxbyBgYHw8/NDfn5+i69XU1MDrVartxEREREREd2LNg9+wsLCkJmZiezsbGzYsAEXL17E448/juvXr0Oj0cDZ2RkeHh56z5HL5dBoNC2+ZlpaGqRSqbj5+vq2dbGJiIiIiMjKtfmwt+joaPHvPn36ICwsDP7+/vjkk0/Qrl27+3rN1NRUTJs2TXys1WoZABERERER0T0x+H1+PDw80L17d1y4cAEKhQK1tbWoqKjQy1NWVtbsHCEdFxcXSCQSvY2IiIiIiOheGDz4qaqqwk8//QRvb2+EhobCyckJBw4cENNLSkqgVquhVCoNXRQiIiIiIrJhbT7sbcaMGRg5ciT8/f1RWlqK+fPnw8HBAePGjYNUKsWkSZMwbdo0eHp6QiKRYMqUKVAqlVzpjYjIQvH+J0REZCnaPPj5+eefMW7cOPz222/o3LkzhgwZgoKCAnTu3BkA8Pbbb8Pe3h6xsbGoqamBSqXC+vXr27oYREREREREeto8+Nm+ffsd011dXZGeno709PS2fmsiIiIiImoFXa/9pWUxJi6JcRl8zg8REREREZE5YPBDREREREQ2gcEPERERERHZBAY/RERERPeo6xtfcqVDIgvE4IeIiMhCHDlyBCNHjoSPjw/s7Oywe/duvXRBEDBv3jx4e3ujXbt2iIyMxPnz5/XyXL16FXFxcZBIJPDw8MCkSZNQVVVlxFqQpUpLS8OAAQPQoUMHyGQyjB49GiUlJXp5bty4geTkZHh5ecHd3R2xsbEoKyvTy6NWqxETE4P27dtDJpNh5syZqK+vN2ZVyIYx+CGyIrorkbwaaRh3O/EEgCVLlvDEkwymuroaffv2bXHF1DVr1mDt2rXYuHEjCgsL4ebmBpVKhRs3boh54uLicObMGeTk5CArKwtHjhxBYmKisapAFiw3NxfJyckoKChATk4O6urqEBUVherqajFPamoqvvjiC+zYsQO5ubkoLS3FmDFjxPSGhgbExMSgtrYWeXl52Lx5MzIzMzFv3jxTVIlsEIMfIjJ75hLQ3e3EEwDeffddnniSwURHR+PNN9/E008/3Wz6hg0bMGfOHIwaNQp9+vTBli1bUFpaKgbq586dQ3Z2Nt5//32EhYVhyJAhWLduHbZv347S0lIj1oQsUXZ2NiZOnIhevXqhb9++yMzMhFqtRlFRkZjno48+wurVqzFs2DCEhoYiIyMDeXl5KCgoAADs378fZ8+exccff4yQkBBER0dj8eLFSE9PR21tramqRjaEwQ8RUSvd6cRTEAQAwIwZM3jiSSZTVlaGyMhI8bFUKkVYWBjy8/MBAPn5+fDw8ED//v3FPJGRkbC3t0dhYWGzr1lTUwOtVqu3EQFAZWUlAMDT01PcV1dXp9cGAwMD4efnp9cGg4ODIZfLxTwqlQparRZnzpxp8h5sf9TWGPwQEbWBS5cuAQDCw8PFfW1x4gnwx5/uza0nlbrHGo0GAKDRaCCTyfTSHR0d4enpKea5XVpaGqRSqbj5+voapuAWylx6po2tsbERU6dOxeDBg9G7d29xv7OzMzw8PPTy3t4Gm2ujurTbsf1RW3M0dQGIyDCs8cfYnO9GXV5eDgBNTiwf9MQTuPnjv3DhwjYu8YOzxjZ2J7fW1xzboKGkpqZi2rRp4mOtVssTUEJycjJOnz6Nb775xqDvw/ZHbY09P2RRblw+jfJPF3LCOdmU1NRUVFZWitvly5dNXSQyY7evrFVWVgaFQgEAUCgUYqCuU19fj6tXr4p5bufi4gKJRKK3kW1LSUlBVlYWDh06hC5duuil1dbWoqKiQm/f7W2wuTaqS7udNbc/LlJkGgx+yKIItTfgJHuIE87J7Oh6dG4/sXzQE0/Aun/8qW3J5XIcOHBAfKzValFYWAilUgkAUCqVqKio0JugfvDgQTQ2NiIsLMzo5SXLIggCUlJSsGvXLhw8eBABAQFN8jg5Oem1wZKSEqjVar02eOrUKb1jYU5ODiQSCYKCggxfCWrC1oIwDnsji9Lu4f5o93B/PP100yEnt084B4AtW7ZALpdj9+7dGDt2rDjh/Pjx4+K8i3Xr1mHEiBFYuXIlfHx8jFcZsipdu3YFcHMp2CFDhgD434lnUlISAP0Tz9DQUAA88aR7U1VVhQsXLoiPL168iOLiYjg5OQEAkpKS8Oabb6Jbt24ICAjA3Llz4ePjg9GjRwMAevbsieHDh2Py5MnYuHEj6urqkJKSgrFjx/L4R3eVnJyMbdu24bPPPkOHDh3E4bpSqVTMEx8fj2nTpsHT0xMSiQRTpkyBUqnEoEGDAABRUVEICgpCfHw8VqxYAY1Ggzlz5iA5ORkuLi4mqRfZFgY/ZDXuNuF87Nixd51w3twqXjU1NaipqREfc7K57WrpxNPT01Oc4PvWW28hODiYJ55kECdOnMCTTz4pPtbNhXjhhRcAAFOnTkVDQwMSExNRUVGBIUOGIDs7G66uruJztm7dipSUFERERMDe3h6xsbFYu3atcStCFmnDhg0A9H9nASAjI0O8l09aWhpcXV0RGxuLmpoaqFQqrF+/Xszr4OCArKwsJCUlQalUws3NDQkJCVi0aJHR6kG2jcEPWQ1DTTg318nmZHwtnXgmJCSIJ4+vvPIKTzzJYMLDw8Ve7ltptVps27YNdnZ2WLRo0R1PJD09PbFt2zZDFtMm2cKCGM21PR3dhUFXV1ekp6ffcXi6v78/9uzZ0+blI2oNBj9Ed8GVZkinpRNP4H8//LNnz8by5ctbfA2eeBIREZkOFzwgq2GoCeecbE5ERERkHRj8kNW4dcK5Dlc6IiJDsKWVkYiIrAmHvZFFaaz9A/XXrqC4uBgAJ5wTEZH5MecbMhPZOvb8kEWp1ZzHlczX0K9fPwA3J5z369cP8+bNE/PoJpwPGDAAVVVVzU44DwwMREREBEaMGIEhQ4bgvffeM3pdiIiIiMi42PNDFsXVrw/8Z2U1ezWNE86JiIiI6E7Y80NERERERDaBwQ8REREREdnEYi4MfoiIiIiIyCYw+CEiIiIiIpvA4IeIiIiIiGwCgx8iIiIiIrIJXOqaiIiIiMhIrH1BAXPH4IeILM6tPxy8gzoRmSseq4jMD4MfIivAq0hEROZNd5xmEERkWpzzQ0QWzRbuSUBERERtgz0/REREREQksuYhmwx+iIio1djLRkRElozD3oiIiIiIyCaw54eIiO6KPT7Ns+ahIURE1og9P0RERERE1CxrW1iIPT9EFsqaDkRERETWjL/Z5oM9P0RERG3A2q6OEhFZI/b8EBEREZnA7cEy542RObOWG/Wy54eIiIiIiGwCe36IyCpw1S0isgQcGklkWgx+iIioWTxJuz/WMjSEiMgaMfghIiIiIqJWsfSRFpzzQ0RERERENoE9P0QWhkORiIisU3NX1C39KjuRuWHwQ0REehhgExG1DWs/nlriHEcOeyMiIiIiIpvAnh8iC2DtV46IiIjIOph7bxCDHyKyOuZ+4CUiuhte9LJs/PzMF4MfIiIiA2ju5IcBObUFLoJA5saSgj0GP0RERERED8iSAgBbxuCHyIzxQPpgeHW09djWiMzfnb6nHO5L1DomXe0tPT0dXbt2haurK8LCwnDs2DFTFodsjDm3v65vfMmT0TZmbv9Tc25/ZBvYBq2T7lhnTse75lhT+7OE/7cpmGtbNFnPzz//+U9MmzYNGzduRFhYGNasWQOVSoWSkhLIZDJTFYtsBNsfmbJXyJzan7n9KFk7c5kHZE5tkGyPNbQ/HjvvjTn1TNoJgiCY4o3DwsIwYMAA/P3vfwcANDY2wtfXF1OmTMEbb7xxx+dqtVpIpVJUVlZCIpHopbEx2obmvjx3ahe3M1T7a6273cWbjKstDsbGan+tfS+2Mct0v23xXo9L/A22La1tVw9ygmpJv8FtgW39/tzp9+hBfovvpV2YpOentrYWRUVFSE1NFffZ29sjMjIS+fn5TfLX1NSgpqZGfFxZWQngZkVv11jzuwFKTOamuc9et+9u8bwh2x8A9J6/DwBweqGqxbRb+f1txx3LS4Z3r59Bc5+todofcG9tkG3MsrXms3qQ9gfwN9gWNdeummtHus/v1s+2uWOKMY+BD/obfGv5myv3nTRXd3owdzrGtfa36kGPgRBM4JdffhEACHl5eXr7Z86cKQwcOLBJ/vnz5wsAuHFr1Xb58mW2P24m29q6/bENcruX7W7tj8dAbobe+BvMzZRba46BFrHaW2pqKqZNmyY+bmxsxNWrV+Hl5QU7Oztxv1arha+vLy5fvmzSrlBjYp3/V2dBEHD9+nX4+Pi06fu1tv1ZMltsR63V2v+NodofYLo2aM7tgmXTZw7tz5w/E2Ox5f+BJfwGW9Lnw7Lem3tpfyYJfjp16gQHBweUlZXp7S8rK4NCoWiS38XFBS4uLnr7PDw8Wnx9iURi9g2lrbHON0ml0rs+z9Dtz5LZYjtqrdb8bwzR/gDTt0Fzbhcs2/+0pv0B/A02Blv9H1jKb7AlfT4sa+u19hhokqWunZ2dERoaigMHDoj7GhsbceDAASiVSlMUiWwI2x+ZEtsfmRrbIJkS2x+ZmsmGvU2bNg0JCQno378/Bg4ciDVr1qC6uhovvviiqYpENoTtj0yJ7Y9MjW2QTIntj0zJZMHP888/j19//RXz5s2DRqNBSEgIsrOzIZfL7/s1XVxcMH/+/Cbdo9aMdb4/hmh/lswW21FrGeJ/Yyntz5zbBcv2YPgbbBj8H7SOqY6BlvT5sKyGY7L7/BARERERERmTSeb8EBERERERGRuDHyIiIiIisgkMfoiIiIiIyCYw+CEiIiIiIptgVcFPeno6unbtCldXV4SFheHYsWOmLlKbWLBgAezs7PS2wMBAMf3GjRtITk6Gl5cX3N3dERsb2+TmYebuyJEjGDlyJHx8fGBnZ4fdu3frpQuCgHnz5sHb2xvt2rVDZGQkzp8/r5fn6tWriIuLg0QigYeHByZNmoSqqioj1sJyWet350Hc7XtnrS5duoRJkyYhICAA7dq1w8MPP4z58+ejtrZWL8/t/xs7OzsUFBQYpYzm0F7T0tIwYMAAdOjQATKZDKNHj0ZJSYlenvDw8Cb/o1dffdXoZTUWc/hcjKE1n701/C5bqrY8hu3YsQOBgYFwdXVFcHAw9uzZY7R6mPr71FbHOLVajZiYGLRv3x4ymQwzZ85EfX29MavSlGAltm/fLjg7OwsffvihcObMGWHy5MmCh4eHUFZWZuqiPbD58+cLvXr1Eq5cuSJuv/76q5j+6quvCr6+vsKBAweEEydOCIMGDRIee+wxE5b43u3Zs0eYPXu2sHPnTgGAsGvXLr30ZcuWCVKpVNi9e7fw3XffCX/5y1+EgIAA4Y8//hDzDB8+XOjbt69QUFAgfP3118IjjzwijBs3zsg1sTzW/N15EHf73lmrvXv3ChMnThT27dsn/PTTT8Jnn30myGQyYfr06WKeixcvCgCEr776Su//U1tba/DymUt7ValUQkZGhnD69GmhuLhYGDFihODn5ydUVVWJeZ544glh8uTJev+jyspKo5bTWMzlczGG1nz21vC7bKna6hh29OhRwcHBQVixYoVw9uxZYc6cOYKTk5Nw6tQpg9fBHL5PbXGMq6+vF3r37i1ERkYKJ0+eFPbs2SN06tRJSE1NNVo9mmM1wc/AgQOF5ORk8XFDQ4Pg4+MjpKWlmbBUbWP+/PlC3759m02rqKgQnJychB07doj7zp07JwAQ8vPzjVTCtnV78NPY2CgoFArhrbfeEvdVVFQILi4uwj/+8Q9BEATh7NmzAgDh+PHjYp69e/cKdnZ2wi+//GK0slsia/7uPIg7fe9szYoVK4SAgADxse7E4eTJk0Yvi7m21/LycgGAkJubK+574oknhNdff910hTIic/1cjOH2z94af5ct3f0cw5577jkhJiZGb19YWJjwyiuvGKqYInP8Pt3PMW7Pnj2Cvb29oNFoxH0bNmwQJBKJUFNTY8ji3pFVDHurra1FUVERIiMjxX329vaIjIxEfn6+CUvWds6fPw8fHx889NBDiIuLg1qtBgAUFRWhrq5Or+6BgYHw8/OzmrpfvHgRGo1Gr45SqRRhYWFiHfPz8+Hh4YH+/fuLeSIjI2Fvb4/CwkKjl9lS2MJ350G09L2zNZWVlfD09Gyy/y9/+QtkMhmGDBmCzz//3ODlMOf2WllZCQBN/k9bt25Fp06d0Lt3b6SmpuL33383RfEMypw/F2O4/bO3hd9lS3M/x7D8/Hy9zxAAVCqVwT9Dc/0+3c8xLj8/H8HBwXo3r1WpVNBqtThz5oxxCt4MR5O9cxv673//i4aGhiZ3BpbL5fjhhx9MVKq2ExYWhszMTPTo0QNXrlzBwoUL8fjjj+P06dPQaDRwdnaGh4eH3nPkcjk0Go1pCtzGdPVo7vPVpWk0GshkMr10R0dHeHp6Ws3/wRCs/bvzIO70vevQoYOpi2c0Fy5cwLp167By5Upxn7u7O1atWoXBgwfD3t4e//rXvzB69Gjs3r0bf/nLXwxWFnNtr42NjZg6dSoGDx6M3r17i/tfeOEF+Pv7w8fHB99//z1mzZqFkpIS7Ny502RlNQRz/VyMobnP3hZ+ly3J/R7DNBrNHc87DMUcv0/3e4xr6X+oSzMVqwh+rF10dLT4d58+fRAWFgZ/f3988sknaNeunQlLRmS97vS9mzRpkglLdn/eeOMNLF++/I55zp07p7eowy+//ILhw4fj2WefxeTJk8X9nTp1wrRp08THAwYMQGlpKd566y2DBj/mKjk5GadPn8Y333yjtz8xMVH8Ozg4GN7e3oiIiMBPP/2Ehx9+2NjFJANo6bOntsdjmOlY2zHOKoKfTp06wcHBoclKKmVlZVAoFCYqleF4eHige/fuuHDhAv785z+jtrYWFRUVeleZrKnuunqUlZXB29tb3F9WVoaQkBAxT3l5ud7z6uvrcfXqVav5PxiCrX13HsSt3ztLNH36dEycOPGOeR566CHx79LSUjz55JN47LHH8N5779319cPCwpCTk/Ogxbwjc2yvKSkpyMrKwpEjR9ClS5c75g0LCwNw80q0OZ8Y3Ctz/FyMoaXPXqFQWP3vsikY+ximUChM0qbN7fv0IMc4hULRZJU6Xb1M+V2wijk/zs7OCA0NxYEDB8R9jY2NOHDgAJRKpQlLZhhVVVX46aef4O3tjdDQUDg5OenVvaSkBGq12mrqHhAQAIVCoVdHrVaLwsJCsY5KpRIVFRUoKioS8xw8eBCNjY3il5GasrXvzoO49XtniTp37ozAwMA7bs7OzgBuXi0NDw9HaGgoMjIyYG9/95+K4uJig/9vzKm9CoKAlJQU7Nq1CwcPHkRAQMBdn1NcXAwAFtuGWmJOn4sx3O2zt4XfZVMw9jFMqVTqfYYAkJOTY/DP0Fy+T21xjFMqlTh16pTexemcnBxIJBIEBQUZpNytYrKlFtrY9u3bBRcXFyEzM1M4e/askJiYKHh4eOitMGGppk+fLhw+fFi4ePGicPToUSEyMlLo1KmTUF5eLgjCzSU1/fz8hIMHDwonTpwQlEqloFQqTVzqe3P9+nXh5MmTwsmTJwUAwurVq4WTJ08K//nPfwRBuLnUtYeHh/DZZ58J33//vTBq1Khml7ru16+fUFhYKHzzzTdCt27duNR1K1jzd+dB3O17Z61+/vln4ZFHHhEiIiKEn3/+WW8JU53MzExh27Ztwrlz54Rz584JS5YsEezt7YUPP/zQ4OUzl/aalJQkSKVS4fDhw3r/o99//10QBEG4cOGCsGjRIuHEiRPCxYsXhc8++0x46KGHhKFDhxq1nMZiLp+LMdztsxcE6/hdtlRtdQw7evSo4OjoKKxcuVI4d+6cMH/+fKMudW3q71NbHON0S11HRUUJxcXFQnZ2ttC5c2cudd2W1q1bJ/j5+QnOzs7CwIEDhYKCAlMXqU08//zzgre3t+Ds7Cz86U9/Ep5//nnhwoULYvoff/wh/PWvfxU6duwotG/fXnj66af1vuSW4NChQwKAJltCQoIgCDeXu547d64gl8sFFxcXISIiQigpKdF7jd9++00YN26c4O7uLkgkEuHFF18Url+/boLaWB5r/e48iLt976xVRkZGs9/FW6+VZWZmCj179hTat28vSCQSYeDAgXrL+hqaObTXlv5HGRkZgiAIglqtFoYOHSp4enoKLi4uwiOPPCLMnDnTau/zIwjm8bkYw90+e0Gwjt9lS9WWx7BPPvlE6N69u+Ds7Cz06tVL+PLLL41WD1N/n9rqGHfp0iUhOjpaaNeundCpUydh+vTpQl1dnVHrcjs7QRAEw/YtERERERERmZ5VzPkhIiIiIiK6GwY/RERERERkExj8EBERERGRTWDwQ0RERERENoHBDxERERER2QQGP0REREREZBMY/BARERERkU1g8GMlDh8+DDs7O1RUVJi6KEREBhUeHo6pU6eauhhkhSZOnIjRo0ebuhhE94TngPfG0dQFICIiIjIH77zzDnjvdyLrxuCHiIiICIBUKjV1EYjIwDjsrY39+uuvUCgUWLp0qbgvLy8Pzs7OOHDgQLPPeeyxxzBr1qwmr+Pk5IQjR44AAD766CP0798fHTp0gEKhwAsvvIDy8vIWy7FgwQKEhITo7VuzZg26du2qt+/9999Hz5494erqisDAQKxfv15Mq62tRUpKCry9veHq6gp/f3+kpaW15t9AVio8PBxTpkzB1KlT0bFjR8jlcmzatAnV1dV48cUX0aFDBzzyyCPYu3evqYtKVq6+vh4pKSmQSqXo1KkT5s6dyyv21GqffvopgoOD0a5dO3h5eSEyMhLV1dVNhr1dv34dcXFxcHNzg7e3N95+++0mwy67du2KN998ExMmTIC7uzv8/f3x+eef49dff8WoUaPg7u6OPn364MSJE+JzfvvtN4wbNw5/+tOf0L59ewQHB+Mf//iHEf8DZAhbtmyBl5cXampq9PaPHj0a8fHxzT6H54DGx+CnjXXu3BkffvghFixYgBMnTuD69euIj49HSkoKIiIimn1OXFwctm/frvfD/c9//hM+Pj54/PHHAQB1dXVYvHgxvvvuO+zevRuXLl3CxIkTH6isW7duxbx587BkyRKcO3cOS5cuxdy5c7F582YAwNq1a/H555/jk08+QUlJCbZu3drki0O2Z/PmzejUqROOHTuGKVOmICkpCc8++ywee+wxfPvtt4iKikJ8fDx+//13UxeVrNjmzZvh6OiIY8eO4Z133sHq1avx/vvvm7pYZAGuXLmCcePG4aWXXsK5c+dw+PBhjBkzptngedq0aTh69Cg+//xz5OTk4Ouvv8a3337bJN/bb7+NwYMH4+TJk4iJiUF8fDwmTJiA8ePH49tvv8XDDz+MCRMmiO9x48YNhIaG4ssvv8Tp06eRmJiI+Ph4HDt2zOD1J8N59tln0dDQgM8//1zcV15eji+//BIvvfRSs8/hOaAJCGQQf/3rX4Xu3bsLL7zwghAcHCzcuHGjxbzl5eWCo6OjcOTIEXGfUqkUZs2a1eJzjh8/LgAQrl+/LgiCIBw6dEgAIFy7dk0QBEGYP3++0LdvX73nvP3224K/v7/4+OGHHxa2bduml2fx4sWCUqkUBEEQpkyZIgwbNkxobGxsTZXJBjzxxBPCkCFDxMf19fWCm5ubEB8fL+67cuWKAEDIz883RRHJBjzxxBNCz5499Y5Ns2bNEnr27GnCUpGlKCoqEgAIly5dapKWkJAgjBo1ShAEQdBqtYKTk5OwY8cOMb2iokJo37698Prrr4v7/P39hfHjx4uPdcfAuXPnivvy8/MFAMKVK1daLFdMTIwwffr0B6gZmYOkpCQhOjpafLxq1SrhoYceavFciueAxseeHwNZuXIl6uvrsWPHDmzduhUuLi4t5u3cuTOioqKwdetWAMDFixeRn5+PuLg4MU9RURFGjhwJPz8/dOjQAU888QQAQK1W31f5qqur8dNPP2HSpElwd3cXtzfffBM//fQTgJur3hQXF6NHjx547bXXsH///vt6L7Iuffr0Ef92cHCAl5cXgoODxX1yuRwA7tglT/SgBg0aBDs7O/GxUqnE+fPn0dDQYMJSkSXo27cvIiIiEBwcjGeffRabNm3CtWvXmuT797//jbq6OgwcOFDcJ5VK0aNHjyZ5bz0u6o6BdzouNjQ0YPHixQgODoanpyfc3d2xb9+++/5NJ/MxefJk7N+/H7/88gsAIDMzExMnTtQ7Xt2K54DGx+DHQH766SeUlpaisbERly5dumv+uLg4fPrpp6irq8O2bdsQHBwsHjirq6uhUqkgkUiwdetWHD9+HLt27QJwc0xmc+zt7Zt04dfV1Yl/V1VVAQA2bdqE4uJicTt9+jQKCgoAAI8++iguXryIxYsX448//sBzzz2HZ5555p7/F2RdnJyc9B7b2dnp7dMd4BsbG41aLiKi1nBwcEBOTg727t2LoKAgrFu3Dj169MDFixfv+zWbOwbe6bj41ltv4Z133sGsWbNw6NAhFBcXQ6VStfibTpajX79+6Nu3L7Zs2YKioiKcOXPmrkPUeA5oXFztzQBqa2sxfvx4PP/88+jRowdefvllnDp1CjKZrMXnjBo1ComJicjOzsa2bdswYcIEMe2HH37Ab7/9hmXLlsHX1xcA9CZONqdz587QaDQQBEE86BYXF4vpcrkcPj4++Pe//613deF2EokEzz//PJ5//nk888wzGD58OK5evQpPT8/W/CuIiAyisLBQ73FBQQG6desGBwcHE5WILImdnR0GDx6MwYMHY968efD39xdPKHUeeughODk54fjx4/Dz8wMAVFZW4scff8TQoUMf6P2PHj2KUaNGYfz48QBuBkU//vgjgoKCHuh1yTy8/PLLWLNmDX755RdERkaK524t4TmgcTH4MYDZs2ejsrISa9euhbu7O/bs2YOXXnoJWVlZLT7Hzc0No0ePxty5c3Hu3DmMGzdOTPPz84OzszPWrVuHV199FadPn8bixYvvWIbw8HD8+uuvWLFiBZ555hlkZ2dj7969kEgkYp6FCxfitddeg1QqxfDhw1FTU4MTJ07g2rVrmDZtGlavXg1vb2/069cP9vb22LFjBxQKBTw8PB74f0RE9CDUajWmTZuGV155Bd9++y3WrVuHVatWmbpYZAEKCwtx4MABREVFQSaTobCwEL/++it69uyJ77//XszXoUMHJCQkYObMmfD09IRMJsP8+fNhb2/f4hCm1urWrRs+/fRT5OXloWPHjli9ejXKysoY/FiJF154ATNmzMCmTZuwZcuWu+bnOaBxcdhbGzt8+DDWrFmDjz76CBKJBPb29vjoo4/w9ddfY8OGDXd8blxcHL777js8/vjj4lUm4GYEn5mZiR07diAoKAjLli3DypUr7/haPXv2xPr165Geno6+ffvi2LFjmDFjhl6el19+Ge+//z4yMjIQHByMJ554ApmZmQgICABw88C/YsUK9O/fHwMGDMClS5ewZ88e2Nuz2RCRaU2YMAF//PEHBg4ciOTkZLz++utITEw0dbHIAkgkEhw5cgQjRoxA9+7dMWfOHKxatQrR0dFN8q5evRpKpRJPPfUUIiMjMXjwYHFp4AcxZ84cPProo1CpVAgPD4dCodBbYpssm1QqRWxsLNzd3Vv9ufIc0HjshNsHBRIRERFRE9XV1fjTn/6EVatWYdKkSaYuDpmxiIgI9OrVC2vXrjV1Ueg2HPZGRERE1IyTJ0/ihx9+wMCBA1FZWYlFixYBuDlHg6g5165dw+HDh3H48GG9m4aS+WDwQ0RERNSClStXoqSkBM7OzggNDcXXX3+NTp06mbpYZKb69euHa9euYfny5c0ui06mx2FvRERERERkEyx/1hIREREREVErMPghIiIiIiKbwOCHiIiIiIhsAoMfIiIiIiKyCQx+iIiIiIjIJjD4ISIiIiIim8Dgh4iIiIiIbAKDHyIiIiIisgkMfoiIiIiIyCb8f8KkBqTX3SPHAAAAAElFTkSuQmCC", + "image/png": "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", "text/plain": [ - "
" + "
" ] }, "metadata": {}, @@ -171,21 +175,18 @@ "source": [ "plt.clf()\n", "fig = plt.figure(figsize = (10,4))\n", - "ax0 = fig.add_subplot(151)\n", - "ax0.hist(inputs[:,0], bins=50)\n", + "ax0 = fig.add_subplot(141)\n", + "ax0.hist(xs_array, bins=50)\n", "ax0.set_xlabel('x values')\n", - "ax1 = fig.add_subplot(152)\n", - "ax1.hist(inputs[:,1], bins=50)\n", + "ax1 = fig.add_subplot(142)\n", + "ax1.hist(ms_array, bins=50)\n", "ax1.set_xlabel('m')\n", - "ax2 = fig.add_subplot(153)\n", - "ax2.hist(inputs[:,2], bins=50)\n", + "ax2 = fig.add_subplot(143)\n", + "ax2.hist(bs_array, bins=50)\n", "ax2.set_xlabel('b')\n", - "ax3 = fig.add_subplot(154)\n", - "ax3.hist(inputs[:,3], bins=50)\n", - "ax3.set_xlabel('sigma')\n", - "ax4 = fig.add_subplot(155)\n", - "ax4.hist(ys, bins=50)\n", - "ax4.set_xlabel('y values')\n", + "ax3 = fig.add_subplot(144)\n", + "ax3.hist(ys_array, bins=50)\n", + "ax3.set_xlabel('y values')\n", "plt.show()" ] },