{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "336b1d26-e539-4fce-83d0-961245e46486",
   "metadata": {},
   "source": [
    "## Theory: the rules of probability as logic\n",
    "\n",
    "### Notation\n",
    "\n",
    "A **proposition** is a statement that is either true or false: $A \\equiv$ \"the spectrum contains a line\", $T \\equiv$ \"the line amplitude lies in $[T, T+dT]$\".\n",
    "\n",
    "- $p(A \\mid B)$: the probability that $A$ is true **given** that $B$ is true. Everything to the right of the bar is assumed true.\n",
    "- $A, B$: \"$A$ and $B$ are both true\". $\\bar A$: \"$A$ is false\".\n",
    "- $I$: the **background information**: everything we know before the data (instrument, theory, noise model, allowed parameter ranges). Every probability is conditional on some $I$, and results can only be as good as $I$.\n",
    "\n",
    "### The two basic rules\n",
    "\n",
    "$$\n",
    "\\textbf{Sum rule:}\\qquad p(A\\mid I) + p(\\bar A\\mid I) = 1\n",
    "$$\n",
    "\n",
    "$$\n",
    "\\textbf{Product rule:}\\qquad p(A,B\\mid I) = p(A\\mid I)\\,p(B\\mid A,I) = p(B\\mid I)\\,p(A\\mid B,I)\n",
    "$$\n",
    "\n",
    "Everything else follows from these two.\n",
    "\n",
    "### Bayes' theorem\n",
    "\n",
    "Equating the two forms of the product rule, with $H$ a hypothesis and $D$ the data:\n",
    "\n",
    "$$\n",
    "\\underbrace{p(H\\mid D,I)}_{\\text{posterior}} \\;=\\; \\frac{\\overbrace{p(H\\mid I)}^{\\text{prior}}\\;\\overbrace{p(D\\mid H,I)}^{\\text{likelihood}}}{\\underbrace{p(D\\mid I)}_{\\text{evidence}}}\n",
    "$$\n",
    "\n",
    "- **Prior** $p(H\\mid I)$: what we believed before the data.\n",
    "- **Likelihood** $p(D\\mid H,I)$, often written $\\mathcal{L}(H)$: the probability of getting the data we actually got, *if* $H$ were true. As a function of $H$ it is **not** a probability distribution over $H$ and does not integrate to 1.\n",
    "- **Evidence** (marginal or global likelihood) $p(D\\mid I)$: the normalising constant. It is irrelevant for parameter estimation and central for model selection.\n",
    "- **Posterior** $p(H\\mid D,I)$: what we believe after the data.\n",
    "\n",
    "For parameter estimation we usually just write\n",
    "\n",
    "$$\n",
    "p(\\theta\\mid D,I) \\;\\propto\\; p(\\theta\\mid I)\\;p(D\\mid\\theta,I).\n",
    "$$\n",
    "\n",
    "### Useful rules\n",
    "\n",
    "**Marginalisation (extended sum rule).** If $\\{B_i\\}$ are mutually exclusive and exhaustive,\n",
    "$$\n",
    "p(A\\mid I) = \\sum_i p(A,B_i\\mid I), \\qquad\\text{or for a continuous parameter}\\qquad\n",
    "p(A\\mid I) = \\int p(A,\\phi\\mid I)\\,d\\phi .\n",
    "$$\n",
    "This is how nuisance parameters are removed: to get $p(T\\mid D,I)$ when $\\nu_0$ is unknown, integrate the joint posterior $p(T,\\nu_0\\mid D,I)$ over $\\nu_0$. **Integrate, don't maximise:** fixing $\\nu_0$ at its best-fit value gives a posterior for $T$ that is too narrow.\n",
    "\n",
    "**Normalisation.** Probability densities integrate to 1: $\\int p(\\theta\\mid D,I)\\,d\\theta = 1$. So a density $p(\\theta)$ has units of $1/[\\theta]$, and $p(\\theta)\\,d\\theta$ is a probability.\n",
    "\n",
    "**Independent data.** If the measurement errors are independent, the likelihood factorises:\n",
    "$$\n",
    "p(D\\mid\\theta,I) = \\prod_{i=1}^{N} p(d_i\\mid\\theta,I)\n",
    "\\quad\\Longrightarrow\\quad\n",
    "\\ln p(D\\mid\\theta,I) = \\sum_{i=1}^{N} \\ln p(d_i\\mid\\theta,I).\n",
    "$$\n",
    "For Gaussian errors this gives $\\ln\\mathcal{L} = \\text{const} - \\chi^2/2$, so maximising the likelihood is the same as minimising $\\chi^2$.\n",
    "\n",
    "**Sequential updating.** Today's posterior is tomorrow's prior:\n",
    "$$\n",
    "p(\\theta\\mid D_2,D_1,I) \\propto p(\\theta\\mid D_1,I)\\;p(D_2\\mid\\theta,D_1,I).\n",
    "$$\n",
    "Analysing two independent data sets one after the other gives the same answer as analysing them together.\n",
    "\n",
    "**Change of variables.** Densities transform with the Jacobian: if $u = g(\\theta)$ then $p(u) = p(\\theta)\\,|d\\theta/du|$. For example, a prior uniform in $\\ln T$ is a prior $\\propto 1/T$ in $T$:\n",
    "$$\n",
    "p(\\ln T\\mid I) = \\frac{1}{\\ln(T_{\\max}/T_{\\min})}\n",
    "\\quad\\Longleftrightarrow\\quad\n",
    "p(T\\mid I) = \\frac{1}{T\\,\\ln(T_{\\max}/T_{\\min})}.\n",
    "$$\n",
    "\"Uniform\" is not a property of a prior; it depends on the parameter you choose.\n",
    "\n",
    "**Work in logs.** Likelihoods are products of many small numbers (here $\\sim10^{-37}$). Always compute $\\ln p$, and add in log space with `scipy.special.logsumexp`.\n",
    "\n",
    "### Two kinds of problem\n",
    "\n",
    "1. **Parameter estimation**: given a model $M$, what are the parameter values? Answer: the posterior $p(\\theta\\mid D,M,I)$. The evidence cancels.\n",
    "2. **Model selection**: which model is more probable? Answer: the posterior odds, which **require** the evidence.\n",
    "\n",
    "### Odds ratio and Bayes factor\n",
    "\n",
    "For two models $M_1$ and $M_0$, apply Bayes' theorem to each and take the ratio. The unknown $p(D\\mid I)$ cancels:\n",
    "\n",
    "$$\n",
    "O_{10} \\;\\equiv\\; \\frac{p(M_1\\mid D,I)}{p(M_0\\mid D,I)}\n",
    "\\;=\\; \\underbrace{\\frac{p(M_1\\mid I)}{p(M_0\\mid I)}}_{\\text{prior odds}}\n",
    "\\;\\times\\;\n",
    "\\underbrace{\\frac{p(D\\mid M_1,I)}{p(D\\mid M_0,I)}}_{\\text{Bayes factor } B_{10}}\n",
    "$$\n",
    "\n",
    "**Posterior odds = prior odds × Bayes factor.** The Bayes factor is the part that comes from the data.\n",
    "\n",
    "**The evidence of a model is its prior-weighted average likelihood:**\n",
    "$$\n",
    "p(D\\mid M_1,I) = \\int p(\\theta\\mid M_1,I)\\;p(D\\mid\\theta,M_1,I)\\,d\\theta .\n",
    "$$\n",
    "It is an *average* over the prior, not the maximum. A model with no free parameters has evidence equal to its likelihood.\n",
    "\n",
    "**From odds to probabilities.** With only two models,\n",
    "$$\n",
    "p(M_1\\mid D,I) = \\frac{O_{10}}{1+O_{10}}.\n",
    "$$\n",
    "With several models $M_0,\\dots,M_K$,\n",
    "$$\n",
    "p(M_i\\mid D,I) = \\frac{O_{i0}}{\\sum_{j=0}^{K} O_{j0}}, \\qquad O_{00}=1.\n",
    "$$\n",
    "\n",
    "*Example (weak line, Jeffreys prior, $B_{10}\\approx 11$).* With even prior odds, $p(M_1\\mid D) = 11/12 \\approx 0.92$. If you were sceptical of the new theory and set prior odds of $1{:}10$, then $O_{10}\\approx1.1$ and $p(M_1\\mid D)\\approx 0.52$. The same data leave the question open.\n",
    "\n",
    "### The Occam factor\n",
    "\n",
    "If the likelihood is sharply peaked (width $\\delta\\theta$) compared to a prior of width $\\Delta\\theta$,\n",
    "\n",
    "$$\n",
    "B_{10} \\;\\approx\\;\n",
    "\\underbrace{\\frac{p(D\\mid\\hat\\theta,M_1,I)}{p(D\\mid M_0,I)}}_{\\text{max-likelihood ratio}\\;\\ge1}\n",
    "\\;\\times\\;\n",
    "\\underbrace{\\frac{\\delta\\theta}{\\Delta\\theta}}_{\\text{Occam factor}\\;\\le1}.\n",
    "$$\n",
    "\n",
    "- The maximum-likelihood ratio can **never** favour the simpler nested model: extra parameters always fit at least as well.\n",
    "- The Occam factor penalises the complex model for the parameter space the data rule out. **Each marginalised parameter contributes its own Occam factor.**\n",
    "- The prior range $\\Delta\\theta$ therefore matters for model selection, even when it barely affects parameter estimation. Doubling $\\Delta\\theta$ halves $B_{10}$ (for a uniform prior).\n",
    "\n",
    "### How strong is the evidence?\n",
    "\n",
    "A common convention (Kass & Raftery 1995), for guidance only:\n",
    "\n",
    "| $B_{10}$ | $\\ln B_{10}$ | Evidence for $M_1$ |\n",
    "|---|---|---|\n",
    "| 1 – 3 | 0 – 1.1 | barely worth mentioning |\n",
    "| 3 – 20 | 1.1 – 3 | positive |\n",
    "| 20 – 150 | 3 – 5 | strong |\n",
    "| > 150 | > 5 | very strong |\n",
    "\n",
    "These are labels, not thresholds. What matters is the posterior probability, and what it would cost to be wrong.\n",
    "\n",
    "### Checklist and common pitfalls\n",
    "\n",
    "1. **Write down Bayes' theorem first**, then identify prior, likelihood and evidence for your problem.\n",
    "2. **State $I$ explicitly:** noise model, parameter ranges, what is known exactly and what is not.\n",
    "3. **Priors for model selection must be proper** (normalised). With an improper prior such as $p(T)\\propto 1/T$ on $(0,\\infty)$, the evidence depends on an arbitrary constant and $B_{10}$ is meaningless.\n",
    "4. **Check prior sensitivity.** If the conclusion changes between reasonable priors, report that. It means the data alone do not decide.\n",
    "5. **Don't do model selection with a parameter estimate.** \"The 95% interval for $T$ excludes zero\" is not the same statement as \"$M_1$ is more probable than $M_0$\".\n",
    "6. **A Bayes factor is not a p-value.** A p-value is $p(\\text{data this extreme}\\mid M_0)$; $B_{10}$ compares how well each model predicted the data you actually have.\n",
    "7. **Fix your choices before looking at the answer.** Trying several priors or noise treatments and quoting the most favourable $B_{10}$ is the Bayesian version of p-hacking."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6a5916d8",
   "metadata": {},
   "source": [
    "# Homework 3 — Is there a spectral line?\n",
    "**Observational Astrophysics, NKUA** · Bayesian model comparison\n",
    "\n",
    "You will compute the **Bayes factor** and the **odds ratio** for two models of a radio spectrum:\n",
    "\n",
    "- $M_0$: no line, $d_i = e_i$ (no free parameters)\n",
    "- $M_1$: line, $d_i = T f_i + e_i$ (one free parameter, the line strength $T$)\n",
    "\n",
    "with independent Gaussian noise $e_i \\sim N(0,\\sigma^2)$, $\\sigma = 1$ mK.\n",
    "\n",
    "Read the homework sheet first. Cells marked **TODO** are for you. Work in **log space** throughout (see Step 2).\n",
    "\n",
    "*Adapted from Gregory (2005), Ch. 3, and Gregory (2010), Introduction to Bayesian Data Analysis.*"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "389dd832",
   "metadata": {},
   "source": [
    "## 0. Setup and data (given — just run)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "6b14a258",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from scipy.special import logsumexp\n",
    "from scipy import stats\n",
    "\n",
    "plt.rcParams.update({\"figure.dpi\": 110, \"axes.grid\": True, \"grid.alpha\": 0.3})\n",
    "rng = np.random.default_rng(2010)\n",
    "\n",
    "N      = 64                                # spectrometer channels\n",
    "nu     = np.arange(1, N + 1, dtype=float)  # channel number = frequency\n",
    "NU0    = 37.0                              # predicted line centre\n",
    "SIG_L  = 2.0                               # line width: exp{-(nu-nu0)^2/8}  ->  sigma_L = 2\n",
    "SIGMA  = 1.0                               # known noise, mK\n",
    "\n",
    "def profile(nu0, x=nu, sig_l=SIG_L):\n",
    "    # Line shape f_i (peak = 1). Broadcasts over nu0.\n",
    "    return np.exp(-0.5 * ((x - nu0) / sig_l) ** 2)\n",
    "\n",
    "def log_trapz(logy, x, axis=-1):\n",
    "    # log of the trapezoid integral of exp(logy) dx, computed stably.\n",
    "    logy = np.moveaxis(logy, axis, -1)\n",
    "    pair = np.logaddexp(logy[..., :-1], logy[..., 1:])\n",
    "    return logsumexp(pair + np.log(np.diff(x) / 2), axis=-1)\n",
    "\n",
    "f = profile(NU0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "119fc5e8",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABYoAAAHjCAYAAACNXCsAAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjYuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8o6BhiAAAACXBIWXMAABDrAAAQ6wFQlOh8AAEAAElEQVR4nOzdd3hT5dsH8O9JRzrTXfa2lC2zyJA9lY2IyCurLEVEwQHKFgRluZguQFErIEsUCgKKoyBgkb1BRrpX2qYrOe8f/SU0dmWcrPb7uS4vmzOec+dpSdM797kfQRRFEURERERERERERERUacnsHQARERERERERERER2RcTxURERERERERERESVHBPFRERERERERERERJUcE8VERERERERERERElRwTxURERERERERERESVHBPFRERERERERERERJUcE8VERERERERERERElRwTxURERERERERERESVHBPFRERERERERERERJUcE8VERERERERERERElRwTxURERERERERERESVHBPFRERERERERERERJUcE8VERBI7duwYBEHA5s2b7R2K3SQlJWHMmDGoXr06BEFAt27dSpwXW83ViRMn0KdPH4SGhkIQBDz99NNWvR4RERER2UfdunXRrVs3e4dBROSUXO0dABGRszh27Bi6d++Ot99+G3PnzrV3OA5t1qxZiIqKwltvvYX69eujSpUqdovlwYMH6NmzJ2rXro2FCxdCoVCgadOmdouHiIiIiMgUx44dw7Fjx/Dyyy/D39/f3uEQUQXGRDERkcS6dOkCtVoNNzc3e4diN4cOHULfvn0xf/58/TatVmuXedmyZQuysrKwdetWtG3b1qbXJiIiIiKy1LFjx7Bo0SKMGzeOiWIisiq2niAikphMJoOHhwdcXFzsHYpFRFFEdna2WefGxcUhMDDQYJu95uX48ePw9/dnkpiIiIjIAWVlZdk7hArHkvfxRFS5MVFMRCSxkvrubt68GYIg4OjRo3j//ffRsGFDyOVy1K9fH2vWrCk2Rl5eHt577z20aNECnp6eUCgU6NWrF3799Vej49Bd8/Dhw1iyZAnq1asHuVyORo0aYe3ataUeu2zZMn187733nv6YtLQ0zJw5Uz9OlSpVMGrUKFy7dk1/zLhx4yAIAkRRxJYtWyAIgn4uTO1HbOkcLF26FIIg4KeffkJaWpo+lnfffdeo84mIiIisRffe6+eff8Y777yD+vXrw8PDA48++ih++uknAMDFixcxYMAA+Pn5wd/fH+PGjUNmZmaxseLj4zF9+nTUrVsX7u7uqFKlCv7v//4Pt2/fNjhOpVJh3rx5eOyxxxASEgJ3d3fUrVsXL730EtLS0oqNu23bNnTo0AGBgYHw9PREnTp1MHz4cFy+fFl/TLdu3VC3bt0Sn6MgCBg3bpz+8e3btyEIAhYuXIidO3ciIiICXl5eePLJJ/XH/P3333jqqacQGhoKd3d31K9fH7Nnzy4x6Xn69Gn06tUL3t7eCAgIwPDhw4s95/IY8xx172+Tk5MxYcIEhISEwNPTEx07dsSRI0dKHPfo0aPo378/AgICIJfL0bhxY7z77rvQaDTFjr116xYmTZqEOnXq6N9j9+nTB4cOHQJQOMeLFi0CANSrV0//nnbhwoUAyn8fX1rP5qLfD52i79c3bdqEJk2awMPDAw0bNsTWrVsBAPfv38czzzyDoKAgeHt7Y8iQIYiLizNp3onIcbH1BBGRDb355pvIyMjA+PHj4ePjg61bt2LmzJmoVq0annnmGQBAQUEBnnjiCfzyyy8YNWoUpk6diuzsbHz11Vfo0aMHdu/ejQEDBhh9zdmzZyM9PR2TJk2CXC7H119/jRdffBGJiYkGbwwB4LXXXoNarcbYsWMREhKCWrVqASj8w6JTp064ePEiRo8ejY4dO+LGjRtYt24dDhw4gN9//x1NmjTBlClT0KtXLzz33HN4/PHHMXnyZABAx44d8e+//xodsxRz0Lt3b/j6+mLGjBl44oknMGrUKADA448/bnQcRERERNY0Z84c5Obm4vnnn4eLiws++OADDB48GDt27EBkZCSefvppDBw4EH/++Se2bNkCuVyOjRs36s+/e/cuOnbsiMzMTERGRqJhw4a4f/8+1q9fj+joaJw6dQq1a9cGUJjg27RpE4YNG4aRI0fCw8MDJ0+exPr16/Hbb7/hxIkT+hZh27Ztw//93/+hU6dOWLBgAXx8fHD//n0cOXIE165dQ6NGjcx+znv27MH777+PqVOnYtKkSRBFEQBw4MABDBkyBLVq1cL06dNRpUoVnD17FqtXr8bvv/+Oo0ePwtW1MIVx5swZdOnSBS4uLnjppZdQs2ZNREdHo1u3bkZXKJv6HPv27QuFQoF58+YhJSUFGzZsQN++ffHTTz+hV69e+uM+//xzTJw4Ea1atcLs2bPh7++P33//HXPmzMHff/+Nb7/9Vn/smTNn0LNnT2RnZ2P8+PF49NFHoVKpEBMTg8OHD6N379546623EBgYiF27dmHNmjUIDg4GALRo0cIgvtLex5tj3bp1SExMxKRJk+Dj44NPP/0UY8eOhZubG+bMmYPHH38cS5YsweXLl7F27VqMHTsWBw8eNPt6RORARCIiMsrRo0dFAOLbb79t1HFffPGFftsXX3whAhBbtGgh5uTk6LdnZmaKQUFBYocOHfTb3n//fRGA+P333xuMm5eXJ7Zq1UqsV6+eUfHqrlmzZk0xNTVVv12tVott2rQRXVxcxFu3bhkcGxYWJmZmZhYba968eSIA8d133zXYfuzYMRGA2LNnT4PtAMSxY8cabCtpXkraJuUcHDx4UAQgfvbZZ0YdT0RERGQLpb03/Pvvv0UAoiAIYlRUlME5gwcPFt3c3ESVSqXfNmTIEDEgIEC8ceOGwbE3b94UfXx8xHHjxum35ebminl5ecVi+eSTT0QA4vbt2/Xbhg4dKvr6+pZ4fFFdu3YV69SpU+K+/74fvHXrlghAdHV1FS9cuGBwrFqtFqtWrSpGREQYzIcoiuL27dtFAOLmzZv12x5//HFRJpOJp0+fNjh2ypQpIgCxa9euZcZtynMcO3asCEAcOHCgqNFo9Ntv374tenl5iWFhYaJWqxVFURSVSqXo4eEhDhkyRL9NZ8WKFSIA8dixY6IoiqJWqxWbNWsmurq6iidOnCh23aLXWrBggQhA/969qPLex9epU6fE+dB9PxYsWKDfpntvXrVqVTElJUW/PS4uTpTL5aIgCMX+HpgxY4YIQLxy5UqxaxCR82HrCSIiG3rxxRchl8v1j729vdGhQwdcvXpVv+3LL79E3bp18fjjjyMpKUn/X3p6OgYNGoRbt24ZHF+eF154wWDRCw8PD8ycORMajQZ79uwpdqy3t3exMXbu3Ak/Pz/MmDHDYHvXrl3RvXt3HDlyBKmpqUbHVB6p5uDMmTMAgDZt2hhs79KlC6KioiSLl4iIiMgc/31v2LJlSygUClSrVg1PP/20wbFdu3ZFfn6+vr1Ceno69u7diyeeeAIKhcLgPZOvry8ee+wxgypPd3d3fcVwQUEB0tLSkJSUhJ49ewIAYmJi9Mf6+/sjOzsb+/bt01f8SuXJJ59EkyZNDLYdPnwYcXFxGDduHFQqlcFz6dq1K7y9vfXPJTExEcePH0e/fv3QunVrg3HmzZtndBymPsc5c+ZAJnuYQqlTpw5Gjx6Na9eu4Z9//gEA7NixAzk5OZg4cSKSk5MNnsfAgQMBQP88zp49i/Pnz2P06NGIiIgodr2i1zJGae/jzTFhwgQEBAToH1epUgXh4eEQBAEvvfSSwbFdu3YFAJP+PiEix8XWE0RENlS/fv1i24KCgpCcnKx/fOnSJWRnZyMkJKTUceLj49GwYUOjrvnfN+JFt12/ft1ge2lj3rx5E02bNjX4Q0anefPmOHr0KG7dumXwhtISUs3B6dOnIZfL0bRpU4PtpvR6JiIiIrKWkt4bBgQElNg2QPc+S/e+8erVq9Bqtdi2bRu2bdtW4vj/TTZ+8sknWLduHc6fP4+CggKDfSkpKfqv33rrLRw/fhzDhw9HUFAQOnbsiO7du+PZZ59FlSpVTHuS/1HS+7dLly4BKEx2vvDCCyWeFx8fDwC4ceMGgJLf49aoUQN+fn5GxWHqcyzvPfWjjz6qfx5ltUjTPQ9dYvW/yW5zGfu3gTFK+7msXr06PDw8im0HYPD3DBE5LyaKiYhsyMXFpdxjtFotwsPD8fHHH5d6TLNmzSSJRxAEg8deXl5GH2tNUs3BmTNn0KJFC30/OyIiIiJHUtp7w7LeM+qqX7VaLQDg6aefxqRJk8q91gcffICXX34ZPXv2xLp161C9enXI5XJoNBr069dPPx4ANGjQABcvXsTRo0dx5MgRHD9+HK+++irmzZuHn376Sb/mQ2nvD/+bhC6qpPebumsvXbq0xOpaAJIVJOgY+xyNoZsH3fP49NNPUadOnRKPrV69OgBIXqld2vt4c75HlvxcEpFz41/OREQOpmHDhrh79y66desmSYLz4sWLGDx4cLFtQOEbZGPUr18f165dQ25ubrGq4vPnz0MQBNSrV8/iWHWkmIO0tDTcvHkTvXv3NtgeHR2NqVOn4ubNm1KESkRERGQXjzzyCGQyGdRqtcFiaqXZsmUL6tati+joaINKY10V7H+5ubmhT58+6NOnDwDg77//RkREBBYtWoTDhw8DAAIDA3H69Oli55r6PktXDevh4VHuc9G9f9W9ny3q/v37SE9PN/q6xjxHnYsXL6JDhw7FthWNSfc8AgICyn0e4eHh+muWx5KCjcDAQINqcR2+FyaikrBHMRGRgxkzZgxSU1OxdOnSEvfrblcz1rp165CWlqZ/nJubi9WrV8PFxQWDBg0yaoxhw4YhPT0dH330kcH23377DUeOHEH37t0lrfKQYg5K60985swZtGrVyvIgiYiIiOwoKCgITzzxBPbv34+jR4+WeEzR90y6D9+LVg6LoojFixcXOy8xMbHYtqZNm8LT09OgxUB4eDhUKhVOnjxpcOyKFStMei59+/ZFlSpVsGLFCsTFxRXbX1BQoE92hoSEoHPnzjhw4ID+/Z7OkiVLjL6msc9RZ9myZQZzd+fOHWzbtg2PPPIIWrRoAaCwutvDwwMLFy5EZmZmsTHUajVUKhUA4NFHH0WzZs3w1Vdf4dSpU8WOLXotHx8fACgx4Vue8PBwXL58Gffv3zcYe9WqVSaPRUQVHyuKiYhM9Msvv5S6b+7cuRaPP2PGDPz8889YuHAhfv31V/Tp0weBgYG4e/cufv/9d9y6dcukCoAqVaqgXbt2mDBhAtzd3fH111/jzJkzmDt3rtFVwK+//jq+//57vPbaazh79iw6duyIGzduYN26dfDz8yuWQLaUFHOgq275b9+3M2fOSNYLjoiIiMieNmzYgM6dO6N379549tln0a5dO8hkMty5cwf79+9Hu3btsHnzZgDAU089hTfeeAN9+/bFU089hezsbOzatQu5ubnFxu3bty98fX3RtWtX1K5dG5mZmfj222+hUqkwbtw4/XFTpkzBqlWrMGTIEMyYMQNeXl7Yv3+/QZGCMby8vPDll19i8ODBaNy4McaPH49GjRpBpVLh+vXr+P777/Huu+/qr71mzRp06dIF3bp1w7Rp01CzZk1ER0cjNjYWwcHBRl3T2Oeo8+DBA/Tq1QtDhw5FSkoK1q9fj9zcXHz88cf6it8aNWpg48aNmDBhAsLDwzF27FjUr18fKSkpuHTpEnbt2oXdu3ejW7duEAQBmzdvRo8ePdCpUydMmDABjz76KLKysnDixAnUq1cP7777LgDgscceAwC88cYbGD16NDw8PNCsWTOjWrG99NJL+Oabb9CjRw9MnToVoihi+/btRs0REVU+TBQTEZno8OHDxW5F05EiUezq6op9+/Zh06ZN2LJlC5YsWYKCggJUrVoVrVu3xvLly00ab/ny5fjzzz+xadMmPHjwAHXq1MGHH36I6dOnGz2Gr68vfvvtNyxevBi7du1CVFQU/Pz8MHjwYCxatEjSxTMAaebgzJkzcHNzQ/PmzYttHz9+vKTxEhEREdlDjRo1cObMGbz33nvYvXs3vvvuO7i7u6NGjRp4/PHHERkZqT/21VdfBVDYP/eVV15BUFAQBg8ejKVLlyIwMNBg3BdeeAE7duzAp59+iuTkZPj5+aFJkyb47rvvMGLECP1xderUwb59+/Dmm29i/vz58PPzw/Dhw7F8+XL4+/ub9Fx69+6NM2fOYPny5di+fTvi4+OhUChQt25dREZGomfPnvpj27Zti19++QVvvPEGPvzwQ7i5uaFnz5745Zdf0K1bN6OuZ+xz1Dl48CBeffVVLF68GCqVCi1btsRXX31VrMXEmDFjEB4ejpUrV+Lzzz9HSkoKAgMDUb9+fbz66qv66mOg8M6306dPY8mSJdizZw8+++wzBAYGomXLlvp2GADQqVMnvPvuu9iwYQMmTZqEgoICLFiwwKhE8WOPPYZt27ZhyZIleOONNxAaGooxY8Zg7NixaNSokVFzRUSVhyCy4zgRUYW0efNmjB8/HkePHjX6DXNFl56ejoCAADx48ABVq1a1dzhERERE5ODGjRuHLVu2cLE2IqoU2KOYiIgqjTNnzqBq1apMEhMRERERERH9BxPFRERUabA/MREREREREVHJmCgmIqJKY9asWfjhhx/sHQYRERERERGRw2GPYiIiIiIiIiIiIqJKjhXFRERERERERERERJUcE8VERERERERERERElZyrvQNwFDk5OTh37hxCQkLg6sppISIiIqqsCgoKkJiYiObNm8PDw8Pe4ZiF722JiIiICDDtvS3fNf7PuXPnEBERYe8wiIiIiMhBnDx5Eu3atbN3GGbhe1siIiIiKsqY97ZMFP9PSEgIgMJJq1atmsXjaTQaJCcnIygoCC4uLhaPR+XjnNsH5932OOe2xzm3Pc657XHOH1IqlYiIiNC/P3RGfG9bMXDebY9zbnuccyD7zBkAgFfr1ja5Hufc9jjntsc5f8iU97ZMFP+P7pa8atWqoWbNmhaPp9FoIJfLERISUul/IG2Fc24fnHfb45zbHufc9jjntsc5L86ZWzbwvW3FwHm3Pc657XHOAUjwOm0Kzrntcc5tj3NenDHvbbmYHRERERERERGRneQrlchXKu0dBhERK4qJiIiIiIiIiOzl/qxXAQB1v95m50iIqLJjopiIiIiIiIiIyE6CJk20dwhERACYKCYiIiIiIiIishvf7t3tHQIREQD2KCYiIiIiIiIispuC1FQUpKbaOwwiIiaKiYiIiIgcxebNmyEIQrH/Zs+ebe/QiIjISu5NexH3pr1o7zCIiNh6goiIiIjI0Rw4cAB+fn76xzVq1LBjNEREZE2BY56zdwhERACYKCYiIiIicjht2rRBcHCwvcMgIiIbUPTrZ+8QiIgAsPUEEREREREREZHdaNVqaNVqe4dBRMSKYiIiIiIiR9O0aVMkJSWhTp06mDRpEl5//XW4uLiUenxGRgYyMjL0j5VKJQBAo9FAo9FYHI9Go4FWq5VkLDIe5932OOe2xzkH/p0QCQCo/dWXNrke59z2OOe2xzl/yJQ5cJhE8fXr17Fy5UrExMTg/PnzaNSoEc6fP2/UuVu2bMGyZctw+/ZtPPLII1iwYAFGjBhh5YiJiIiIiKRVrVo1LFq0CO3bt4cgCNi7dy/mzp2L+/fv4+OPPy71vNWrV2PRokXFticnJ0Mul1scl1arRXp6OgBAJuNNibbCebc9zrntcc4BWd8+AIDExMRSj9HlR5o1a2bx9Tjntsc5tz3O+UPJyclGH+swieILFy5g//79aN++PbRaLbRarVHn7dixA+PGjcPs2bPRp08f7N69GyNHjoSfnx/69Olj5aiJiIiIiKTTt29f9O3bV/+4T58+8PT0xJo1a/DWW2+hWrVqJZ43c+ZMTJw4Uf9YqVQiIiICQUFBCAkJsTguXSVKcHBwmZXNJC3Ou+1xzm2Pcw7gubIXs8vJycGQIUMAAPHx8fDw8LDocpxz2+Oc2x7n/KHc3Fyjj3WYRPHAgQMxePBgAMC4ceNw6tQpo86bN28eRowYgWXLlgEAunfvjsuXL2P+/PlMFBMRERGR03v66aexcuVKxMbGlpooVigUUCgUxba7uLhI9seRTCaTdDwyDufd9jjntsc5L9vhw4eRlZUFAPj5558xaNAgi8fknNse59z2OOeFTHn+DlN7bU4Z+K1bt3D58mWMGjXKYPuzzz6LkydPIikpSarwiIjIgeT8m4P81Hx7h0FEZBOiKNo7BCIisqLbz47G7WdHG2xTqVQQBAGCIGDixImQyWSQyWSIjIzUb8/MzLRTxERUUTlMRbE5Ll26BABo3LixwfYmTZpAFEVcvnwZnTt3LvFcLvhR8XDO7YPzbnuVfc6zr2bjTIszcK/pjnaX20FwFax+zco+5/bAObc9zvlDjjYHUVFRcHFxQatWrewdChERWYHfoIHFtsXFxem/LloEV/RrpVKJsLAw6wZHRJWKUyeKU1NTAQD+/v4G2wMCAgAAKSkppZ7LBT8qHs65fXDeba+yz3nKjhSIBSJyb+fi3q/34NHcsh5txqjsc24PnHPb45w/ZMqCH1Lr27cvevbsqV+saO/evdi0aRNmzJiBqlWr2i0uIiKynoBnnim2LSwsDGq1Gnv37sXIkSMN9kVFRWHQoEEW9yomIvovp04U6wiCYTWZ7va8/24vigt+VDycc/vgvNteZZ/zpPMPqyhcrrggpIflr9nlqexzbg+cc9vjnD9kyoIfUmvUqBE+/fRT3Lt3D1qtFg0bNsT777+P6dOn2y0mIiKyDw8PD7i7uxfbLpfLmSQmIqtw6kSxrnI4NTUVVapU0W9PS0sz2F8SLvhRMXHO7YPzbnuVec5VMSr915knMuHyom3moDLPub1wzm2Pc17Ins//gw8+wAcffGC36xMRke39O2ECAKD2558X27dgwQIAQGhoKNLS0pCXl4f58+dj8ODBNo2RiCoHp76vUNebWNerWOfixYsQBAGNGjWyR1hERGQlufdzkfvvw0q/9D/S7RgNEREREZHlvB/vAu/Hu5S4LyYmBhs2bEBcXBzCw8OxfPlyxMTE2DhCIqosnDpRXK9ePTRq1AhRUVEG27/55htEREQgODjYTpEREZE1pP9pmBjOuZmDvIQ8O0VDRERERGS5oPHjEDR+XIn7PD09cenSJZw9exZ+fn4YNGgQPD09bRsgEVUaDtN6Ijs7Gz/++CMA4M6dO8jIyMCOHTsAAF27dkVISAgiIyOxZcsWFBQU6M9bvHgxRo4ciQYNGqB3797Ys2cPoqOjceDAAbs8DyIisp6MPzMAAJ5hnlBfVwNi4bbgwfxgkIiIiIgqnqysLERFRWHZsmU4duxYpW8PRUTW5TCJ4oSEBIwYMcJgm+7x0aNH0a1bN2g0Gv1CK0WPyc7OxjvvvIOVK1fikUceQVRUFPr06WOz2ImIyDZ0ieKAPgGQecqQ9U8W0v9MZ6KYiIiIiJzW3RemAQBqrVtbbN/evXvRp08feHp6Yvv27WjRogXCw8NtHSIRVRIOkyiuW7cuRFEs85jNmzdj8+bNxbaPHTsWY8eOtVJkRETkCLS5WqhOFy5k59fBD2KBiKx/spDxR4adIyMiIiIiMp9ni+al7qtTpw5mzJgBADh8+DBcXFyYKCYiq3GYRDEREVFZVGdUEPMKP1BUdFBALBCh3KiE6pQK2nwtZG5O3XafiIiIiCqp4KlTS9yekZGBatWqoV69egAAX19fZGSwSIKIrId/VRMRkVPQtZ1wC3WDRz0PKDooAABatRaZZzPtGRoRERERkeSioqKwatUq/eNatWrB1ZX1fkRkPXyFISIip6BLFPt19IMgCPAM84RrkCsKkguQ8WcGFG0Vdo6QiIiIiMh092fOBADUWL3aYPvXX3+NJUuW6B/rWlAQEVkLK4qJiMgppP+ZDgD6SmJBEODXwQ8A2KeYiIiIiJyWW+3acKtd22BbUlIS/v33X3To0EG/7dixYyWu20REJBUmiomIyOHl3M1B3v08AA8TxUW/1iWRiYiIiIicTejLLyP05ZcNtgUHByM2NhYy2cO0TWJiIk6cOGHj6IioMmGimIiIHJ6uYlhwFeDb1le/XZcozr2Ti1xlrl1iIyIiIiKS2muvvWaQJAYAhULBxeyIyKqYKCYiIoenqxj2aeUDF08X/Xbfdr7632S6HsZERERERM7kwZtv4cGbb+kfX7t2DT/99BO8vLwMjqtZsyYaNWpk6/CIqBLhYnZEROTwdEngom0nAMDVxxU+j/og8+9MpP+RjpBhIfYIj4iIiIjIbC6+vgaPv/nmGzz77LMQBMFge9OmTdG0aVNbhkZElQwriomIyKFp1Bpk/p0JoHiiuOg2VhQTERERkTOqMmc2qsyZrX/s6emJUaNGFTsuPj4eI0aMsGVoRFTJMFFMREQOTXVaBTFfBAD4dfArtl+XKFadVkGbp7VpbEREREREUsrKysKsWbNQr169YvvkcjliY2NtHxQRVRpMFBMRkUPTVQq7V3eHvLa82H5d8ljMFfWVx0REREREziJu8WLELV4MAFi4cCE+/fTTEo/z9fXlYnZEZFVMFBMRkUMr2p/4v33aAMCjvgfcQt0APFz0joiIiIjIWYhaLUStFlqtFtu3b8fgwYNLPM7FxQUTJ060cXREVJlwMTsiInJYoijqE8UltZ0AAEEQoOigQPKeZGT8kQG8bMMAiYiIiIgsVG3hQgDAr7/+ioYNG6JKlSqlHrt06VIbRUVElREriomIyGHl3MlBXlwegJIXstPRJZG5oB0REREROatWrVph/fr1ZR7TokULJCcn2ygiIqpsWFFMREQOK+OPwsSv4C7Ap7VPqcfpksi593KRcy8HHjU9bBIfEREREZGlzr36GrRaDU42boxJkyaVeaxcLodKpUJQUJCNoiOiyoQVxURE5LB0FcK+rX3h4uFS6nG+bX0huAoG5xARERERObqcnBwciIrCwe+244cffij3eIVCwQXtiMhqmCgmIiKHpVucrqy2EwDg4uUCn5aFFce6KmQiIiIiIkd38OBBvH7vLt64fw+NGjUq9/hXXnmlzB7GRESWYKKYiIgckiZLg8zYTADlJ4qLHqNLLhMREREROSKVSgVBECAIAiZOnAiZrDA18/nnn+u3Z2Zmlnhux44d4eNTeks2IiJLMFFMREQOSXVKBWgKvzYlUZx5JhOaHI01QyMiIiIiMltcXJz+66SkJLwQEIgXg4KRlJSk365UKks8d/78+Ua1qCAiMgcTxURE5JB0lcHyWnKjFqfTJYrFfBGZZ0quwCAiIiIisrewsDCo1WpERUUBAOq7u6O+uzsAICoqCmq1GmFhYSWeyx7FRGRNrvYOgIiIqCS6RemMqSYGAI86HnCv5o48ZR7S/0iHX0c/a4ZHRERERGQ2Dw8PuP8vOTxT+UC/XS6Xw8Oj9CIJJoqJyJpYUUxERA5HFEWTE8WCIOiP1Z1LREREROSoFixYAKAw+Tt69GgAha0lyvLMM89g+PDhVo+NiConVhQTEZHDybmZg/zEfAAwqTJY0UGBpO+TkPFnBkRRhCAI1gqRiIiIiMgiMTEx2LJlCwKOHkXPLl3w+OOPY8yYMWWe4+npiZSUFBtFSESVDSuKiYjI4aT/UdifWOYhg09L41d19utQmFTOU+Yh999cq8RGRERERCQFT09PTJgwAR38/aE+fQZTpkyBp6dnmefExMRg1apVNoqQiCobJoqJiMjh6FpH+LTxgczd+F9VPm18ILgVVhHrks1ERERERI7q66+/xgc+Pqi1cYNRx7NHMRFZExPFRETkcHSJYl2FsLFcPFzg09rHYAwiIiIiIkd17tw5NG3a1OjjfX19mSgmIqthopiIiBxKQWYBMv/JBAAoOhq3kF1RuuQyE8VERERE5OjOnTuH1g8eIGXrVqOOb968ObYaeSwRkamYKCYiIoeiOqkCtIVfKzqYnijWnZMZmwlNtkbK0IiIiIiIJPXGG2/A78oVqH4+YtTxoiji0KFDVo6KiCorJoqJiMih6CqBPep6QF5VbvL5ukSxWCBCdUolaWxERERERFLJyspCcHAw6n35Jeps2WzUOVqtFm+++aZ1AyOiSouJYiIicijpfxYuQmdONTEAeNTygLxmYYKZ7SeIiIiIyFH99ddfWLRokUnneHp6Ijs720oREVFlx0QxERE5DFEUkRFTmNw1pz+xji7JrEs6ExERERE5mnPnzqFZs2ZIjfoOqVHfGXWOIAiQy+UQRdHK0RFRZcREMREROQz1NTUKkgsAPFyUzhy6RHHGnxl8E01EREREDuncuXNo3rw50vfsQfqePUafd+/ePQiCYMXIiKiycrV3AERERDrpfxRWAMs8ZfBu4W32OLpEcX5CPnJu5sCzgack8RERERERSWXFihVwc3OD59ChJp333nvvYfLkyfD397dOYERUabGimIiIHEbmmUwAgG9bX8jczP8V5dvKF4J7YZWF6gwXtCMiIiIix6LVavHll1/Cy8sLgqsrBFfj6/h2796NhIQEK0ZHRJUVE8VEROQwcu/nAgA86ntYNI5MLoO8RuGCdnkP8iyOi4iIiIhISv/++y8+++wzAED63r1I37vX6HMVCgUyMrhoMxFJj4liIiJyGHnKwqSue1V3i8fSjZGrzLV4LCIiIiIiKen6EwNA6rdRSP02yuhzmzRpYq2wiKiSY49iIiJyGHlxhYlieTW5xWO5V3M3GJOIiIiIyFHExcWhdevWAIBamzaZdO7q1autERIRESuKiYjIMYiiqE/qSllRzEQxETmzzMxM1KxZE4Ig4NSpU/YOh4iIJDJp0iS8/PLLAAAXH2+4+Bi/kPOnn36K6OhoK0VGRJUZE8VEROQQNBkaaNVaABInipVMFBOR83r77bdRUFBg7zCIiEhi06ZNQ2Zm4ULOGdHRyDAh8Xv37l1cu3bNWqERUSXGRDERETmEopW/urYRlmDrCSJydpcvX8batWuxaNEie4dCREQSysvLw44dO+DtXVhFnLJ5C1I2bzH6fIVCAZVKZa3wiKgSY49iIiJyCEUXnZOyojg/MR/aAi1krvxslIicy0svvYSpU6ciPDzc3qEQEZGErly5gvDwcAiCAACo+dGHJp3frFkzpKWlWSEyIqrsmCgmIiKHoKv8lXnK4OLrYvF4+mSzCOQn5ENe3fIF8oiIbGXHjh04e/YsduzYgTNnzpR7fEZGBjIyMvSPlUolAECj0UCj0Vgcj0ajgVarlWQsMh7n3fY457ZXGedcrVZj2LBh+ucs+PsDgNFz0KtXL5OO/6/KOOf2xjm3Pc75Q6bMARPFRETkEPQL2VVz11dXWEJe7WFiOC8uj4liInIa2dnZmDlzJpYtWwaFQmHUOatXry6xRUVycjLkcstf/7RaLdLT0wEAMhnv0LAVzrvtcc5trzLOee3atVG7dm0kJiYCAPJPnAQAuLWPMOr8P//8Ez/99BMWL15s1vUr45zbG+fc9jjnDyUnJxt9LBPFRETkEHSLzknRdgIA3ELdAAGAyAXtiMi5LFmyBFWqVMG4ceOMPmfmzJmYOHGi/rFSqURERASCgoIQEhJicUy6SpTg4GC4uFh+1wcZh/Nue5xz26uMc/7cc8/h1VdfxaOPPgoA+PfbbwEA1Qc8adT5gYGBiI+PN/v1vTLOub1xzm2Pc/5Qbm5u+Qf9DxPFRETkEIpWFEtB5iaDW7Ab8hPzuaAdETmNO3fuYNWqVdi1a5e+lURmZqb+/5mZmfDx8Sl2nkKhKLH62MXFRbI/jmQymaTjkXE477bHObe9yjbnf/75J+rXr69/vjVWrAAAo59/QEAAMjMzLZqvyjbnjoBzbnuc80KmPH8miomIyCHoE8USVRTrxmKimIicya1bt5CXl4cnnyxeVda9e3e0b98eMTExdoiMiIikkJGRAa1WC///9SUGAPeaNUwao3r16hgxYoTEkRERMVFMREQOQurWE7qxss5lIVdp/K02RET21LJlSxw9etRgW2xsLF555RVs2LAB7dq1s1NkREQkhYKCAixdutRgW9bJwh7F3hHG9SgODg42aDdERCQVJoqJiMgh6Kp+iy5CZyldGwtWFBORs/D390e3bt1K3NemTRu0bt3atgEREZGkBEEoVg2c+P4HAADvr7cZNUZ+fj7q1KmDBw8eSB4fEVVuTBQTEZHdafO1yE/MByB9RTHAxeyIiIiIyDHMnz8f7du3x//93//pt1V7e7FJY7i5uSE/P1/q0IiImCgmIiL7y094+EbXKoliVhQTkRPr1q0bRFG0dxhERCSBc+fOFWsbIW/QwORxZDIZNBpNpV+ki4ikJbN3AEREREUTubp2EVIo2nqCSRYiIiIisidRFHH58mU0atTIYLv63Dmoz50zaay1a9dCq9VKGR4RESuKiYjI/vSLzQmAW6ibZOPqKoq12VpoVBq4Kvhrj4iIiIjsZ9euXZDLDdfkiF+2HABQ18gexQDQtm1bJoqJSHKsKCYiIrvTVRS7hbhB5irdr6aiC+Ox/QQRERER2dO///4LHx+fYturvPkmqrz5pkljjRs3DtevX5cqNCIiAKwoJiIiB6BL4krZn/i/4+XF5cGroZek4xMRERERGWvHjh1Qq9Vo3ry5wXbPZk1NHkuhUCAjI0Oq0IiIALCimIiIHECe0jqJYheFC2QeMoNrEBERERHZw7lz54oliQEg5+pV5Fy9atJYvr6+TBQTkeSYKCYiIrvTVxRLuJAdAAiCYLCgHRERERGRvVy4cKHERHHcwkWIW7jIpLEWLlyIiIgIqUIjIgLA1hNEROQArFVRrBsz51bOwwXziIiIiIjs4Pjx48UWsgOA0FkzTR5Lo9EgNTUVAQEBUoRGRASAFcVEROQArNWjuOiYrCgmIiIiInu5d+8etm3bBkEQiu3zatMGXm3amDTejh07sHfvXqnCIyIC4ECJ4qtXr6Jfv37w9vZGaGgoZsyYAbVaXe55WVlZmD17Nho0aAAvLy+EhYVh4cKFyM1l5RgRkTMQRVGfxJVXK15hYSm2niAiIiIieztx4gT+/PPPEvfl3b2LvLt3TRqPi9kRkTU4ROuJtLQ09OjRA3Xq1MHOnTuRkJCAmTNnIjk5GV999VWZ5z7//PPYvXs3li5dimbNmuHkyZOYN28eUlJS8OGHH9roGRARkbk0GRpo1VoAVq4o5mJ2RERERGQnpS1kBwAP3pgNAKj79Tajx1MoFLhrYnKZiKg8DpEo3rhxI1JTUxEbG4vg4GAAgKurK0aPHo233noLjRs3LvG8goICbN++Ha+//jqmT58OAOjevTvu3LmDqKgoJoqJiJxA0UpfqRezKzomK4qJiIiIyF7y8vLQunXrEvcFv/CCyeONHDkSWq3W0rCIiAw4ROuJH3/8Eb169dIniQFg+PDhkMvl+PHHH0s9TxRFFBQUwM/Pz2C7v78/RFG0WrxERCQdg0SxFSuK8xPzoS3gm2kiIiIisr133nkHjz/+eIn7fDp3gk/nTiaNl5SUhEOHDkkRGhGRnkNUFF+6dAkTJkww2CaXy9GgQQNcunSp1PPc3Nwwfvx4fPTRR+jUqROaNm2Kv/76C5988om+wrg0GRkZBv18lEolgMKVQzUajQXPBvpxtFqtJGORcTjn9sF5t72KNufq+4X96GWeMsALkj8v15D//aoTgRxlDuTVTe+DXNHm3Blwzm2Pc/4Q54CIiKSkVqsxfvx4fPvttyXuz09IAAC4hYYaPeaDBw/wySefYMiQIVKESEQEwEESxampqfD39y+2PSAgACkpKWWeu379ekydOhWPPfaYftv06dMxf/78Ms9bvXo1Fi1aVGx7cnIy5HLLF1PSarVIT08HAMhkDlG4XeFxzu2D8257FW3OU6+nAgBcQl2QlJQk+fj57vn6rxMuJcDDzcPkMSranDsDzrntcc4fSk5OtncIRERUgVy8eLHM97n3X34FgOk9irmYHRFJzSESxQAgCEKxbaIolri9qNmzZ+OHH37Apk2bEB4ejtOnT2PBggUICAgoMRGsM3PmTEycOFH/WKlUIiIiAkFBQQgJCTH/ifyPrhIlODgYLi4uFo9H5eOc2wfn3fYq2pxnqjIBAJ7VPSV5/f0vrb8WN3ADAOCd443AkECTx6hoc+4MOOe2xzl/KDc3194hEBFRBXL+/PlSF7IDgKDICaXuK42fn1+l/2CXiKTnEInigIAApKamFtuelpZW6kJ2QOGL7cqVK7Fnzx4MGjQIANClSxfIZDK8+uqrmDZtGkJLuXVDoVBAoVAU2+7i4iLZH0cymUzS8ah8nHP74LzbXkWa8/yEwopf92ruVnk+Li4ucAt2Q35SPgoSC8y+RkWac2fBObc9znmhyv78iYhIWiEhIejfv3+p+3179jR5zOrVq+OXX36xJCwiomIc4uOnxo0bF+tFnJubixs3bpSZKL548SIAoGXLlgbbW7ZsiYKCAty5c0fyWImISFq6xezcq0m/kJ2ObuyiC+cREREREdlC//790adPn1L3azIyoDGjjcQrr7xiSVhERMU4RKL4iSeewM8//2zQD27Xrl3Izc3FE088Uep5derUAQCcPn3aYPupU6cAAHXr1pU+WCIiklSe8n+J4qpWTBT/b+xcJW8nJyIiIiLbiY2NRe3atZGTk1PqMXenPo+7U583eeyvvvrKktCIiIpxiNYTU6ZMwUcffYTBgwdj3rx5SEhIwMyZMzF69GiDiuLIyEhs2bIFBQUFAIC2bdsiIiICU6dORXx8PMLDw/HXX39h8eLFGDlypFV6XRIRkbR0Vb7yapYvJFoaVhQTERERka3l5OSgU6dOUKvVZR4X8OyzZo0vl8uRm5sLudx676OJqHJxiESxv78/jhw5gunTp2PYsGHw8vLCqFGj8O677xocp9Fo9AutAIX94/bt24d58+bh3XffRVxcHGrVqoXp06fjrbfesvXTICIiE2nztchP+l+PYhtUFDNRTERERES2cvDgQWRnZwMAoqOj9Wsr/ZffgCfNGr9Tp07IyclhopiIJOMQiWIAaNiwIQ4ePFjmMZs3b8bmzZsNtoWGhmLjxo1WjIyIiKwlPyEfEAu/tkmiWMlEMRERERFZj0qlgkKhAAAEBwdDJpNBFEVERkYiKSlJf4yPj4/+HDGv8D2q4G7a++GoqCiJoiYiKuQQPYqJiKhyKlrha6vF7ERRtNp1iIiIiKhyi4uL03+dlJQErVYLURT1SWIAUCqVBufcGTced8aNN/lar732Gq5cuWJ+sERE/8FEMRER2Y0+USwAbiFuVruOrqJYm62FJlNTztFEREREROYJCwuDWq0usdo3KioKarUaYWFhBtv9hw+D//BhJl/r5s2biI+PNztWIqL/cpjWE0REVPnkKnMBAG7BbpC5We+zy6JtLfKUeXD15a8/IiIiIrIODw8PuJfQRkIul8PDw6PYdv/hw826jkKhQEZGhlnnEhGVhBXFRERkN7qKYmu2nQAAebWHC3xwQTsiIiIisrYFCxYAAPz8/BAZGQkAmD9/vqTX6NatG0JDQyUdk4gqN5ZUERGR3egWl7PmQnYA4KJwgcxDBm2OlgvaEREREZHVxcTE4PPPP0fv3r0RFhaGdu3aYcyYMSUee+f/ngMA1PnqS5OuMXbsWIvjJCIqioliIiKyG31FsZUTxYIgwL2qO3Ju57CimIiIiIisztPTE4899hi0Wi0EQcCUKVNKPda3Xz+zrrF+/XoUFBRg+vTp5oZJRGSArSeIiMhubNV6oug1mCgmIiIiIlvYtGkT/vnnn3KPC/y/0Qj8v9Emjy8IAhISEswJjYioREwUExGR3diq9UTRa+gW0CMiIiIisqaLFy+iadOmVhvf19eXi9kRkaTYeoKIiOxCFEV9dW/RxeashRXFRERERGRLd+/eRVhYWLnH/TtpMgCg9iebTBq/TZs2CAgIMCs2IqKSMFFMRER2oVFpoFVrAdi2opiJYiIiIiKyhZs3b0ImK/9Gbu/2EWaN37BhQ9SqVcusc4mISsLWE0REZBe6thOAjRPFSiaKiYiIiMi6YmNj8fXXXxt1bNDEiQiaONHka5w9exZDhw41+TwiotKwopiIiOyiaGWvLRezy0/Mh7ZAC5krPyslIiIiIus4fvw4kpKSrHoNhUIBlUpl1WsQUeXCv5KJiMgudIvKyTxlcPF1sfr19FXLIpCfkG/16xERERFR5XXhwgWjF7K7N3067k2fbvI1FAoFF7MjIkmxopiIiOxCV1HsXtUdgiBY/XpF21vkxeVBXt36C+gRERERUeXk5eWFFi1aGHWsPLyRWdfw9/fHihUrzDqXiKgkTBQTEZFd6BPFNmg7AQDuVQwTxURERERE1rJ69Wqjjw15cZpZ13Bzc0P16tXNOpeIqCRsPUFERHahW1TOFgvZAYDMTQa3YDeDaxMRERERSS0+Ph7PPvusTa7Vu3dvaLVam1yLiCo+JoqJiMgubF1RXPRarCgmIiIiImu5cOEC8vKMf795/7XXcf+11826lo+PD7Kyssw6l4jov5goJiIiuyjao9hWdNdiopiIiIiIrOXChQto0qSJ0ce7VQmFW5VQs66lUCigUqnMOrcssbGxiI2NlXxcInJs7FFMRER2YevWE0WvlavMtdk1iYiIiKhyqV+/Ptq0aWP08aGvvmr2tbZv347g4GCzzy9JTk4OOnfuDABISkqCh4eHpOMTkeNiRTEREdmcNl+L/KR8AIC8mtxm12XrCSIiIiKytj59+qBjx442uVZ8fDzS0tIkHfPgwYPIyspCVlYWoqOjJR2biBwbE8VERGRz+Qn5gFj4tV1aT3AxOyJyUAcPHkTXrl0REhICuVyO+vXrY+bMmUhPT7d3aEREZARRFFG3bl3k5+cbfY5y3nwo580363pr167FP//8Y9a5RalUKgiCAEEQMHHiRMhkMshkMkRGRuq3Z2ZmWnwdInJsbD1BREQ2V7Si1149ikVRhCAINrs2EZExUlJS0LFjR7z88ssICAjA+fPnsXDhQpw/f55VXURETiAhIQEKhQJubm5GnyPIzb/DTqFQICMjw+zzdeLi4vRfJyUllfi1UqlEWFiYxdciIsdlcqJYo9Fg//79OHToEE6cOIG4uDio1WoEBQUhPDwcjz/+OIYPH4569epZI14iIqoA9IliAXALNf5NtKV0rSe02VpoMjVw9eXnpUTkWEaNGoVRo0bpH3fr1g1yuRyTJ0/GgwcPUL16dTtGR0RE5TF1ITsAqDr3LbOvJ9VidmFhYVCr1di7dy9GjhxpsC8qKgqDBg1ir2KiSsDov5AzMzOxcuVKrF+/HikpKWjatClatmyJLl26wMPDA6mpqbh16xbee+89zJ49G926dcOiRYvQqVMna8ZPREROSLeYnFuwG2RutuuCVLR6OU+Zx0QxETmFoKAgADDpNmYiIrKPhg0b4vXXX7fZ9ZYuXQpXV2ne03p4eMDdvfjdfnK5nEliokrC6FeTevXqoUmTJnj33XcxdOhQ+Pn5lXrsyZMn8e2332LAgAFYunQpXnjhBUmCJSKiikFXUayr8LWVogvn5cXlwauhl02vT0RkLI1Gg/z8fFy8eBGLFy/GwIEDUadOnVKPz8jIMLj1WKlU6sfRaDSSxKPVaiUZi4zHebc9zrntVbQ5z8jIQOPGjU16PgnLlgMAQufMNvl6Z86cQVZWFrp37270OWXN+YIFCwAA/fv3x08//QQAmD9/PgYMGGBybPRQRfs5dwac84dMmQOjE8W7d+82ujo4IiICERERWLRoEe7cuWN0MEREVDnoE8U27E8MAC4KF8g8ZNDmaA36JBMROZo6derg/v37AIB+/frhm2++KfP41atXY9GiRcW2JycnQ25B70sdrVarX1BPJuN62LbCebc9zrntVbQ5j4yMxDvvvIPGjRsbfU52SgoAIDEx0eTr/fnnn7h58yaaNWtm9Dllzfnu3buxfft2BAYGwtfXF23atMGIESPMio0eqmg/586Ac/5QcnKy0ccanSg2pYXEtWvXEBYWBl9fX5NerIiIqHLIU9onUSwIAtyruiPndo4+BiIiR/Tjjz8iMzMTFy5cwNtvv42BAwfi0KFDcHFxKfH4mTNnYuLEifrHSqUSERERCAoKQkhIiMXx6CpRgoODS42BpMd5tz3Oue1VtDm/ffs22rdvb9KHdCEr3jP7ejVq1MC1a9dMeq0vb85nzZqFzz77DAcPHsTatWsREBBgdnxUqKL9nDsDzvlDubm5Rh9rdKJ43Lhx2Lx5c7nH/f333+jfv7/BiplERERF2av1hO6aObdzWFFMRA6tRYsWAICOHTuidevWaNu2LXbt2oWnnnqqxOMVCgUUCkWx7S4uLpL9cSSTySQdj4zDebc9zrntVZQ5T0hIgJ+fH7y8bNfeLDQ0FHK53OS5K2/OJ06ciM8//xwJCQkIDg6WItRKr6L8nDsTznkhU56/0Ynib775BlqtFlu2bIEgCCUe8+uvv2LQoEGoWbOm0QEQEVHlY6+K4qLX1C2oR0Tk6Fq2bAkXFxdcv37d3qEQEVEZAgMDER0dbfJ5CatWAwBCZ800+dzu3bub1J/YWAcOHIBSqTTof09EFZ/RTTq+++47fPfdd3juuecgimKx/fv27UO/fv0QHh6OX3/9VdIgiYio4hBF0W49iotekxXFROQs/vzzT2g0GtSvX9/eoRARURnOnDmj74lqinylEvn/W4TUVA8ePCixR72lvv/+e4iiiIKCAsnHJiLHZXRF8eDBg7Fjxw6MGDECGo0G27Zt0zeD/vLLLzFhwgR069YNu3fvhre3t9UCJiIi56ZRaaBVawEA8mqWL7BkKl27CyaKicgRDRs2DG3btkWLFi3g6emJs2fP4r333kOLFi0wZMgQe4dHRERl2Lx5M7p06YLmzZubdF6NlSvMvmZBQQH27t2LBQsWmD1GSVJSUvDVV1+hc+fOko5LRI7NpGX/BgwYgO+//x67d+/GqFGjoNVq8f7772PcuHEYPHgw9u/fzyQxERGVqegicnatKOZidkTkgCIiIrB9+3Y8++yzGDx4MD7//HNMnjwZx48fh7u77V8ziYjIeBcvXkSTJk1sek2FQmGV9hByuRw///wzDh8+LPnYROS4jK4o1unfvz/27NmDoUOHonnz5rh8+TImTJiAjRs36iuMiYiISlO0ktdei9kBQH5iPrQFWshc+buLiBzH7NmzMXv2bHuHQUREZkhISEB4eLjJ5yWuXQsACJk2zeRzfX198eijj5p8Xnm++eYbLFmyBBcuXECvXr0kH5+IHJPRieIzZ87ovw4ODsaCBQswe/Zs9OvXD88//zxiY2MNjm/durVkQRIRUcWhSxTLPGVw8bX96rP6KmaxMFlsj/YXRERERFTxXLhwAYIgmHxezqVLZl/TxcUFO3bsMPv80kydOhXh4eFm9VwmIudldKK4bdu2Jb7gHThwAAcPHtQ/FkURgiBAo9FIEyEREVUoucpcAIUJW3PeSFuqaLuLPGUeE8VEREREZLHY2FicOnUKEydONPncWh9/bNG1+/fvj71798LNzc2icYrauXMntm7digcPHkg2JhE5PqMTxUePHrVmHEREVEnoKort0XYCANyrFEkUc0E7IiIiIpLAH3/8AaVSaZdr37x5EyqVCoGBgZKMl52dDU9PT/Tv31+S8YjIeRidKO7atas14yAiokpCt4icPRayAwCZmwxuwW7IT8rngnZEREREJIkLFy6gc+fOZp2b/NnnAICgyAlmne/r64uMjAzJEsU5OTno168f/vrrL2zfvh3vvfeeJOMSkePjCj5ERGRT+opiOyWKi16bFcVEREREJIWAgAC0bNnSrHOz/vwTWX/+afa1R40aBblcunZqgYGB2LRpE0RRxPnz5yUbl4gcn9EVxUXl5+dj1apV2L59O+7evYucnJxix2RkZFgcHBERVTz2bj2hu3bW+SwmiomIiIhIEkuWLDH73NqffmLRtWfNmmXR+f8VExOD3bt3Y9y4ccztEFUyZiWKX3jhBWzduhWDBg1Cv3794O5uvz/2iYjIudi79UTRa+sW1iMiIiIiMldiYiIiIyOxd+9eu1x/6tSpGDp0KPr27SvJePfv30daWhpCQ0MxcOBAScYkIudgVqJ4586dWLNmDV544QWp4yEiogpMm69FflI+AMdIFLOimIiIiIgsdfHiRbi4uJh9fsq2bQCAwNGjzTrf1dUVaWlpZl//v5KTkxEYGIjAwEC88cYbko1LRI7PrB7Fvr6+qF+/vtSxEBFRBZefmA+IhV/Lq0nXR81UurYXTBQTERERkaUuXLiAJk2amH2+6qcDUP10wOzzdYvZSSUsLAxdu3aFKIqoXbu2ZOMSkeMzq6J41qxZWLt2LXr16gVXV7OGICKiSkjXdgJwkIpiZR5EUYQgCHaLhYiIiIicW3h4OFq3bm32+XW++tKi6w8aNAienp4WjVFU9+7d9V+r1Wq+XyaqRMzK8r700kt48OABHnnkEXTp0gX+/v4G+wVBwAcffCBFfEREVIHoK3gFwC3UzW5x6CqKtdlaaDI1cPXlh55EREREZJ7HHnsM3t7edrt+y5YtkZcn3Z1ykydPxpAhQ/DEE0/A09MTOTk5kiaiichxmfWX8bZt27By5UoIgoCff/652GJ2TBQTEVFJdIvHuQW7QeZmVvcjSRStZs5T5jFRTERERERmCwsLw82bN+Hh4WHW+Wk7vwcA+A8fZtb527dvR2xsLFavXm3W+f+VkJAAhUIBAFi6dCmriYkqEbP+Mp4zZw6eeuopbNq0Sf/iQUREVB5dRbE920789/p5cXnwauhlx2iIiIiIyFklJibCx8fH7CQxAKTt3AnA/ESxQqGQtEdxSkoKAgMDAQD9+vWDTGa/Ag8isi2z/rWnpqZi0qRJTBITEZFJ9IniavZNFLv6uULmUfgrkAvaEREREZG5Ll68aNFCdgBQe/MXqL35C7PPVygUUKlUFsVQ1LRp01CzZk0AwJgxY3Du3DnJxiYix2ZWorhv3744ceKE1LEQEVEFp1vMzt4VxYIgGCxoR0RERERkjvDwcMyfP9+iMWTu7pC5m//+uGXLlnjttdcsiqGoTp066QsDfX19Ja1WJiLHZlbriUmTJmHatGnIyspCz549iy1mB8CiFT+JiKhicpTWE7oYcm7nsKKYiIiIiMyWmJiIBg0aWDRG+v79AAC/J58063wvLy/J2kOIoohWrVohMTERgPRtLYjIsZmVKO7fvz8AYNmyZVi2bJlBY3NRFCEIAjQajTQREhFRheEorSeKxsBEMRERERGZ66WXXsKaNWvQsmVLs8dI3fY1APMTxampqRg3bhz++ecfs2PQycrKgre3t/7xzJkz9f2KiajiMytRfPToUanjICKiCk4URYdpPVE0hlxlrp0jISIiIiJndeXKFYSHh1s0Rq316yw6X8qq36IL2QGAt7c3cnP5fpmosjArUdy1a1ep4yAiogpOo9JAq9YCAOTV5HaOhhXFRERERGSZpKQkeHt7w9PT06JxXPz8LDrfy8sLWVlZFo2hExwcjI0bN+of79y5E7m5uZgzZ44k4xORYzO6iU1+fr5ZFzD3PCIiqliKLhrnSBXFXMyOiIiIiMzx77//YsOGDRaPozpyBKojR8w+XxAEnDhxwuI4ACAzMxNeXl76x+xRTFS5GJ0orlu3LtasWYOUlBSjjv/tt98wdOhQvPvuu2YHR0REFUfRyl1HShTnJ+ZDW6C1czRERERE5ExycnLQqVMnDBw4EDk5ORaNlfzpZ0j+9DOLxvjrr7+Ql2d5AcQvv/yCtWvX6h8zUUxUuRjdemLjxo2YO3cuZs+eja5du6JTp05o3rw5QkJCIJfLkZaWhlu3buH06dM4cOAAkpKS8Pzzz2Pq1KnWjJ+IiJyELlEs85TBReFi52iKLKgnFiaLHaEdBhERERE5h4MHD+oTxNHR0Rg0aJDZY9VYs9rieN5++21069YNVapUsWic5ORkgx7FTzzxBHr27GlpeETkJIxOFA8YMAADBgzAkSNHsHXrVnz66ae4f/8+gMLbHERRhLu7O9q0aYNXXnkFzz33HIKDg60WOBERORfdonHuVd0hCIKdozGsas5T5jFRTERERERlUqlUUCgUAKDPdwiCgMjISCQlJemP8fHxMWlcNwuTu8DDyl9LE8UpKSkICgrSP1ar1Thx4gQGDBhgaYhE5ARMXsyuR48e6NGjBwAgLi4OSqUSOTk5CAwMRL169eDubv/biYmIyPHoKoodoe0EALiHFkkUc0E7IiIiIipHXFyc/mtdYlgURf3XAKBUKhEWFmbSuJm//w4A8OnUyezYFAoFVCqV2efrTJ8+HaIo6h/Hx8dj5cqVTBQTVRImJ4qLqlq1KqpWrSpVLEREVIHpE8XVHCNRLHOXwS3YDflJ+UwUExEREVG5wsLCoFarsXfvXowcOdJgX1RUFAYNGgQPDw+Tx01auw6AZYnin376SZK79n7//Xc0bdpUXzktVQKaiJyD0YvZWdvVq1fRr18/eHt7IzQ0FDNmzIBarTbq3JSUFLzwwguoVq0aPDw80LBhQ2zcuNHKERMRkSnylI5VUQw8jEUXGxERERFRWTw8PEq8k1oul5uVJAaA6sveQfVl71gU165du/DPP/9YNAYArF+/Hnfv3tU/VigUyM7OtnhcInIOFlUUSyUtLQ09evRAnTp1sHPnTiQkJGDmzJlITk7GV199Vea5mZmZ6Nq1Kzw9PfHBBx8gNDQU165dQ35+vo2iJyIiYzhaRTFQGEvW+SxWFBMRERGR0RYsWAAACA0NxcCBA/HZZ59h/vz5GDx4sFnjudepY3FMx48fR3Z2Nlq0aGHROP9dzM7Pzw+XLl2yNDwichIOkSjeuHEjUlNTERsbq28I7+rqitGjR+Ott95C48aNSz33nXfegVqtxsmTJ+Hp6QkA6Natmy3CJiIiE+gTxVUcKFH8v4pi3UJ7RERERETliYmJwcKFC3HhwgV88sknaNeuHcaMGWP2eNlnzgAAvFq3NnsM3WJ2ltKtQaUjCALmzZuHt99+2+KxicjxOUTriR9//BG9evXSJ4kBYPjw4ZDL5fjxxx/LPPfzzz9HZGSkPklMRESOR1ugRX5i4Z0eDlVR/L9EcX4870IhIiIiIuN4enriqaeeQseOHSEIAqZMmWJRTiJh5SokrFxlUUy1atWCXC63aAwAOHXqFEJDQw22ffbZZ8jL4x14RJWBQ1QUX7p0CRMmTDDYJpfL0aBBgzJvcbh16xbi4+MREBCAAQMG4NChQ/Dx8cEzzzyDlStXlvlCnZGRYfBpm1KpBABoNBpoNBoLn1HhOFqtVpKxyDicc/vgvNueM855Xlwe8L/Fk11DXB0mdtfQwl+DuXG5ZcbkjHPu7Djntsc5f4hzQERE5WnXrh3atWsnyVhVF8y3eIyJEydaPIYoipg9ezbeffddg+26Be2CgoIsvgY5ltjYWABAy5Yt7RoHOQ6zE8UajQYnTpzAvXv3kJOTU2y/KbddpKamwt/fv9j2gIAApKSklHpeXFwcAOC1117DiBEj8OOPP+LixYuYM2cO8vLy8Mknn5R67urVq7Fo0aJi25OTkyX5FE6r1SI9PR0AIJM5ROF2hcc5tw/Ou+0545znXHr4e0LlpkJOYvHfG/aQ41UYR54yD4mJiaUe54xz7uw457bHOX8oOTnZ3iEQEZGD27p1KwDTch+l8QgPt3iMP/74A2fPnsXzzz9v9hiZmZnYvn17sURxixYtkJvLVm0VTU5ODjp37gwASEpKMnsxRqpYzEoUnzlzBsOGDcPdu3chimKx/YIgmPxiKQhCsW2iKJa4XUer1QIAGjdujM8//xwA0LNnT+Tn5+O1117D22+/japVq5Z47syZMw0+cVMqlYiIiEBQUBBCQkJMir0kukqU4OBguLi4WDwelY9zbh+cd9tzxjlPyX34oV/VxlUhc3eMJJTrI65QQgltlhaBnoFw8Sl5Pp1xzp0d59z2OOcP8Y9hIiIqz5UrV1C7dm1Jxsq5eBEA4NGkidljpKWlISYmxqJEcUpKikF/Yp3vvvvO7DHJcR08eBBZWVkAgOjoaAwaNMjOEZEjMCtR/Pzzz8PX1xfR0dFo0qQJ3N0t6zcZEBCA1NTUYtvT0tLKXMhO9wLWo0cPg+09evSAVqvFpUuXSk0UKxQKKBSKYttdXFwk++NIJpNJOh6Vj3NuH5x323O2OS9ILAAAuAa6ws3Tzc7RPORZ42GLIk2SBu5+pf8+c7Y5rwg457bHOS9U2Z8/ERGVLyUlRbLb9eOWLAUA1P16m9lj+Pr6WryYXV5eHtq2bVts+7Jly/Dkk0+iRYsWFo1P9qdSqfS5sODgYP1dZJGRkUhKStIf4+PjY7cYyb7MShRfuHAB3333HXr27ClJEI0bNy7Wizg3Nxc3btwo1ru4qAYNGpSYpNZVOVf22yaJiBxFXnzh4he6xeMcRdF48uLz4NmAC6MSERERUfnat2+PJhZUABdV5Y3XLR6jevXqaN68uUVjhIWFYcOGDcW2X7lyBS1atGCiuALQtXAFoE8M//drpVKJsLAwm8ZFjsOsTGrDhg2hUqkkC+KJJ57Azz//bNAPbteuXcjNzcUTTzxR6nnu7u7o3bs3fv75Z4PtP//8M1xdXSV70SYiIsvkxTlmotg1wBWCW2GLI12MRERERETlGTduHJo2bSrJWJ6PPgrPRx+1aIwGDRpg8eLFFo3x888/Y+3atcW2KxQKi6uVyTGEhYVBrVYjKiqq2L6oqCio1WomiSs5sxLFa9aswbJly3D58mVJgpgyZQr8/f0xePBgHDx4EF9++SWmT5+O0aNHG7SeiIyMhKurYRH0/PnzcfbsWYwZMwbR0dF4//33sWDBArz44ouS9BomIiLLOWqiWJAJcK9SGBMTxURERERkrA4dOkCpVEoyVu7Nm8i9edOiMdLT08sstDPG1atXDSpOddq1a8f8SgXi4eFR4t35crmcC9qR8a0nmjdvbrCwnFKpRLNmzVC9enX4+/sbHCsIAs6ePWt0EP7+/jhy5AimT5+OYcOGwcvLC6NGjSq20qZGo9EvtKITERGB/fv3Y86cORg4cCCCgoIwffp0vP3220Zfn4iIrMtRE8VAYUy593KZKCYiIiIio92/f79YLsRcyrnzAFjWo9jT09OkPExJUlJSEBQUVGz7c889Z9G45HgWLFig/3rw4MHYs2cP5s+fj8GDB9sxKnIERieK27RpY5AollrDhg1x8ODBMo/ZvHkzNm/eXGx779690bt3bytFRkRElnL0RDHAimIiIiIiMp5arYanpzTrW4S89JLFY7i7uyM/P9+iMRo0aIAqVaoU2x4VFYV79+5h1qxZFo1PjiMmJgZbt27FsWPHMHbsWPTv3x9jxoyxd1jkAIxOFJeUoCUiIjKGPlFcxfESxW5V3AAwUUxERERExhFFEe+8845k43k/1l6ScZ5//nmIomh2kd8zzzxT4naNRoN///3XktDIwXh6ekImkyEmJgaRkZGYMmWKvUMiB2FWj+IJEybg1q1bJe67c+cOJkyYYFFQRERUcWhyNNCkF7YNYkUxERERETm7goICDBkyRLLx8h88QP6DBxaPs3DhQovOHzNmDP7+++9i27mYXcV069YtzJkzB7169bJ3KORAzEoUb968GYmJiSXuS0pKwpYtWywKioiIKo78+Ie3wDl0ojieiWIiIiIiKt+lS5cwcuRIyca7/+pruP/qaxaP0759e9y7d8/s82/cuAEvL69i2xs1aoQ+ffpYEho5oLi4OJw+fRrr1q2zdyjkQIxuPfFfpd3KcO3atRKbnxMRUeVUtFLXoRPFcXkW3apHRCSF7du3Y9u2bTh9+jRSUlLQoEEDPP/885gyZQpkMrNqPIiISGIpKSkIDAyUbLzgKZMlGcfT0xMqlcrs85OTk0vM5zzyyCOoX7++JaGRA+rbty/u37+P2NhYe4dCDsToRPH69euxfv16AIVJ4meffbZY4/acnBzcvn0bI0aMkDZKIiJyWvpEsQvgFuRm32BKoEsUi3kiCtIK4BbgeDESUeWxatUq1KlTBytWrECVKlVw9OhRvPTSS7h58yZWrFhh7/CIiAilJ1TN5dO1qyTjWNoiYsqUKQgICCi2/datW3jmmWdw4sQJS8IjBzNw4ECcOnUKx48ft3co5ECMThRXr14dbdq0AQCcP38e4eHhCAkJMTjG3d0djRs3RmRkpLRREhGR09IvZBfqDsHF8ap1i1Y558XlMVFMRHa1b98+g/fY3bt3R2ZmJj7++GMsWbIEcrncjtEREREAdOrUCU2aNJFsvIKUFACAq4VVytOnT0ft2rXNOlcURUyZMgUuLi7F9vn6+rJHcQX0yCOP4Pjx4+jYsaO9QyEHYnSiePDgwRg8eLD+8bx583jrARERlUufKHbAthNA8USxd2NvO0ZDRJXdfwsxAKBVq1bIyclBSkoKqlWrZoeoiIioqKysLCgUCsnGu/fidABA3a+3WTROmzZt4OZmXtFDRkYGWrdujRs3bhTbx0RxxaPRaKDRaNCgQQO89prl/bGp4jCrR/EXX3whdRxERFRB6RPFVRwzUezq4wqZlwzabK1BP2UiIkdx/PhxBAYGIjQ0tNRjMjIyDP6IVyqVAB7+IWgpjUYDrVYryVhkPM677XHObc8Z53zdunVo0qQJxo0bJ8l4/s89BwAWz8Hy5csRHh6O8ePHl3lcSXOekJCAwMDAEmNwdXXF8uXLnep75Ggc7ec8Li4OISEhyM/PR/PmzXHhwgV7hyQ5R5tzezJlDsxKFE+YMKHUfTKZDH5+fmjVqhWGDRtW4oqZRERUeTh6RTFQGFvOzRwmionI4Zw6dQpffPEFFixYUOLtwDqrV6/GokWLim1PTk6WpF2FVqtFeno6AHBRPRvivNse59z2nHHOHzx4gBYtWiAxMVGaAVu1BADkWDieTCaDUqksN66S5vzGjRvw8fEp9dywsDDEx8c7zffI0Tjaz3lOTg7mz5+P5ORkJCcnQ6lUwtXVrBShw3K0Oben5ORko48166fgr7/+Qnx8PJKSkuDn54eQkBAkJiYiPT0dwcHB8PT0xPvvv4+33noLR44cQYMGDcy5DBERVQB58U6UKI5nopiIHEdcXByGDx+OiIgIvPHGG2UeO3PmTEycOFH/WKlUIiIiAkFBQSW2szCVrhIlODi4zIQ1SYvzbnucc9tzxjnPyspCvXr1JHl9BQBtdjYAQGZhoV21atWQnp5eblwlzXnLli2xcOHCUs/t1KkT/vrrL/j5+VkUY2XlaD/nMpkM7dq1Q0hICEJCQiCTyST7eXYUjjbn9pSbm2v0sWYlileuXIlp06Zh+/bt6Fpkdc6jR49i0qRJ+Pjjj1G/fn0MGDAAr7/+Onbu3GnOZYiIqAJwlopiAKwoJiKHkZ6ejv79+8PLywt79+4tt+ekQqEosV+mi4uLZH8cyWQySccj43DebY9zbnvONueffvopFAqFZPHenTIVgOU9ikeMGIH8/Hyj4vrvnLu4uCAsLKzUcxUKBbKyshBo4YJ7lZkj/Zzv378fFy9exIoVK/Dmm2/C09PTIeKSmiPNuT2Z8vzNqr1+7bXXsGjRIoMkMVC4KvOCBQvw6quvomHDhpgzZw6OHj1qziWIiKgCEEWRiWIiIhPl5ORg0KBBiI+Px4EDBxAUFGTvkIiIqIjo6GgIgiDZeAEjn0bAyKctHsfd3V1/q72pvvvuO2zatKnU/X5+flzQrgJRKpX6BXKHDBkCT09PO0dEjsKsRPG1a9fg7+9f4r6AgAD9KpkNGjSAWq02OzgiInJuGpUGWrUWABPFRETGKCgowNNPP42zZ8/iwIEDqFOnjr1DIiKi/3j11VclrVD0GzwYfoMHWzxObGwsli9fbta5KSkpZVYLf/TRR/ydVIHExcWhatWqAAp/nvft22fniMhRmNV6olGjRli5ciW6d+9usFhdVlYWVqxYgSZNmgAobPCu+8EjIqLKp2jilYliIqLyTZs2Dfv27cN7772H7OxsxMTE6Pc1adKkxPYSRERkO6IoQqPRSLrwl6gtLKwQLFxwS6FQQKVSmXVucnIy6tWrV+p+jUYDlUoFHx8fc8MjBzJ79mx9Pi8kJARJSUl2jogchVmvbB999BH69++PmjVronv37vrF7I4cOYKCggIcOHAAAPDPP/9g+PDhkgZMRETOo2ji1a1K2f017cm9SmGiOD8xH6JGhOAi3a2ERESmOHjwIADg9ddfL7bv6NGj6Natm40jIiKiojIzMxEcHCzpmHf+7zkAlvcoVigUZreHeO211wwKAf9r48aN6NGjB0aMGGFueORAzpw5g+7duwOAPqdHBJiZKO7cuTOuXbuG1atX49SpU7h48SKqVauGyZMn45VXXtFXEb/zzjuSBktERM5FlygW5AJc/aSrupCavtpZC+Ql5kFeVW7fgIio0rp9+7a9QyAiojL4+vri0qVLko4pRdsJAHjkkUewc+dOs869fPky2rRpU+p+9iiuWF555RWcPHkS3t7eGDRoEHJycuwdEjkIs/9qr1q1Kt577z0pYyEiogqm6EJ2Ui74IbWibTHy4pgoJiIiIqKS3bp1C4cOHcLkyZMlG1OKhewAQBAE7Nu3DxMmTDD53Llz52Lr1q2lrkdlSbUyOZ7U1FT99zooKAh37tyxb0DkMCxrgENERFSGvPiHiWJHpms9AQD58fl2jISIiIiIHNm1a9dw7Ngxe4dRIplMhtmzZ5t1bkpKCoKCgkrd/9JLL2Hq1KnmhkYOJDMzEz4+PvpCnhs3buDll1+2b1DkMMxKFKvVarz55pto2LAhvLy84OLiYvCflE3diYjIeRWtKHZkMrkMrgGFv7u4oB0RERFR2WJjYxEbG2vvMOwiOTm5zISqOe6MH48748dbPI5MJkN+fr5Z35u0tDT4+fmVuv/evXv4448/LIiOHIWPjw+uX7+uf8wexVSUWRndadOm4euvv8aIESMwfvx4uLs7dgKAiIjsw1kSxUBhjAWpBUwUExEREZUhJycHnTt3BgAkJSXBw8PDzhHZlpubG8LDwyUd01eihUpzcnKQnp6Ozp07m/y9uXPnDlxcXErdf+nSJfz000/o2bOnFKGSHV29ehWnT5/GqFGjAADBwcFISkqyc1TkKMxKFO/btw8rVqzA9OnTpY6HiIgqEGdLFGdfymaimIiIiKgMBw8eRFZWFgAgOjoagwYNsnNEtvXUU09JPmbg2LGSjHPw4EGIooisrCyTvjdZWVn4/PPPy8zxKBQKpKenSxIn2dfZs2fx66+/6hPFcrkcx48ft3NU5CjMaj3h4uIi+SdoRERU8egTxVWcIFH8vxiZKCYiIiIypFKpIAgCBEHAxIkTIZPJIJPJEBkZqd+emZlp7zBt4sMPP8Svv/5q7zD0/vu9EQTB5O9NXFwcvvrqqzKP8ff3hyiKUoZOdqJUKlGtWjWDbWfOnIFarbZTRORIzEoUT506FV9++aXUsRARUQUiakWnWcwOeBgjE8VEREREhuLi4vRfJyUlQavVQqvVGtyurlQq7RGazZ08eRLZ2dmSjnl36vO4O/V5s8797/dGFEWTvzcpKSkIDAws85g2bdrgu+++MytGciyCIKBBgwYG21auXIn79+/bKSJyJGa1nvD29sbx48fRoUMH9O7dG/7+/gb7BUHAK6+8IkV8RETkpPKT8wFN4ddMFBMRERE5r7CwMKjVauzduxcjR4402BcVFYVBgwZVml7F1ljMzrNVK7PPleJ7k5ubi0ceeaTMY1QqFVauXIlFixaZHSs5hpJajOgWtCvv54AqPrMSxW+88QYA4N9//8WJEyeK7WeimIiIdNXEgJMliuOZKCYiIiL6Lw8PjxIXspfL5ZUmSQwAERERqFGjhqRjBk+ZbNH5ln5vOnfurF+gsDQymQybN29morgCmDt3LiIjI1GvXj39tpCQEKSkpNgxKnIUZrWe0N1mUtp/Go1G6jiJiMjJFK3MdYoexf9LFBekFkCbq7VzNERERESOZ8GCBQCACRMmYPz48QCA+fPn2zMkm1u0aBGqV69u7zCK0X1vIiMjERkZCcD4783+/fuxa9euMo/x8vLSL2JIzm3//v1wdTWsG/3ss8/w5JNP2ikiciRmJYqJiIjKo0sUuyhc4OLlYudoyle06plVxURERETFxcTEYN68eWjWrBn27duHDRs2ICYmxt5h2VTLli0lX9Tt3iuv4J6Fd2XHxMRgw4YNWLRoEbRarUnfm1OnTuHevXtlHiMIAlq3bs0F7SqAuLg4VKlSxWDbb7/9hoMHD9opInIkZieK8/PzsWHDBkRGRqJPnz64du0agMIeOJcuXZIsQCIick66RLEztJ0A/pMoZp9iIiIiomI8PT3h5uaG3NxcAMDkyZPh6elp56hsp6CgAEqlEoIgSDque926cK9b16IxPD09MWXKFISEhGD//v0mfW+Sk5PLXcwOAKKjoyV/7mRboiiiT58+xVqVXL9+HYcOHbJTVORIzEoU37x5E+Hh4Xjttddw5coV/Pzzz1CpVACAX3/9Fe+9956kQRIRkfPRJ4qdoO0EALgFu+l/KzJRTERE5oqNjUVsbKy9wyCymh9++AEDBgyAh4cHcnJy7B2OTaWmphqVUDVV6IwZCJ0xQ5Kx3N3dERAQgISEBKPPqV27Nho0aFDucRMnTkRiYqIl4ZGdCYKALVu2FNuuW8yOyKxE8UsvvYSQkBDcunULx44dM7j1oGvXrvj1118lC5CIiJyTs1UUCy4C3ELcADBRTERE5snJydEvClXZEmhUOWg0GrRt2xZNmzbFmjVrIJNVrm6Wrq6uePnll+0dRrleeOEF5OUZ/3721VdfxWOPPVbucZcuXUJqaqoloZGdnTt3Tt9fvKiaNWvC19fXDhGRozHrVf3YsWOYO3cugoODi912ULVqVSiVSkmCIyIi5+VsiWLgYaxMFBMRkTkOHjyIrKwsZGVlITo62t7hEElOo9Fg7dq1EAQB7du3r3T9an19fTFhwgTJx30wew4ezJ4j2XgvvfQSatSoYfTxzz77LO7fv1/ucQqFAhkZGZaERnZ27949FBQUFNvesmVLfPzxx3aIiByNWYliV1fXUn8hxMfHw8fHx6KgiIjI+eXH5wNw0kQxF7MjIiIjqVQqCIIAQRAwceJEyGQyyGQyREZG6rdnZmbaO0wiSYwaNUp/B/GECRMq3fpEP/30EyZPniz5uC7+/nDx95dsvO+//96kyue//voLXl5e5R7Xr18/5nucnFKpRLVq1Yptz83NxZgxY+wQETkasxLFXbt2xapVq5Cfn6/fJggCRFHEpk2b0LNnT8kCJCIi58SKYiIiqgzi4uL0XyclJUGr1UKr1SIpKUm/nXdcUkWQm5uLP/74AxEREQAKq2t1axVVFsnJyQgKCpJ83Cqz30CV2W9INl69evVw4cIFo49PS0uDn59fucfNmDEDjRo1siQ0srNGjRqhf//+xba7u7vjxx9/tENE5GhczTlp+fLl6NSpExo3bozBgwdDEASsXbsW58+fx7Vr13Dy5Emp4yQiIieizdciP8mJK4qZKCYiIiOFhYVBrVZj7969GDlypMG+qKgoDBo0CB4eHnaKjkg6v/76K9q3b6//eXbWRLFuscmWLVuafK61EsVSCw8Px+XLl40+ftKkSUb1m16yZAnCwsKKvdaR82jbti3c3NyKbRcEAR4eHlCr1fD09LRDZOQozKoobty4MU6fPo1OnTrhm2++gYuLC3744Qc88sgjOHnypFGrZRIRUcWVn/DwjhMmiomIqKLz8PCAu3vx33dyuZxJYqowmjVrhiVLlugfz50716gF0ByJpQtO9uvXD0OHDpU8LuXChVAuXCjZeF5eXlixYoXRPaTfeecdo47TaDRITEy0JDSys9GjR+PYsWMl7nv66aeRm5tr24DI4ZicKM7NzcWqVauQmZmJLVu24MGDB8jLy0NcXBy++uorJomJiMgg0epexYkSxVUeJoor2+IsRERkmQULFgAAZDKZvtfn/Pnz7RkSkWREUcRff/2Fpk2bGmxLT0+3Y1Sms3TBSW9vb5MWibOn7t27Iy0trdzjrl27hm7duhk1Jhezc35xcXGoWrVqiftWr14Nfwl7ZZNzMjlRLJfLMW/ePKSmplojHiIiqgCKJordQovf2uSodBXF2mwtNJkaO0dDRETOJCYmBgsXLkTbtm2xZMkSrFu3DjExMfYOi0gSV65cweLFiyEIgn7bzp07sWvXLjtGZRwpF5x87bXX8Ndff0keY7WFC1FNwopioDDpt3PnznKPS0lJMXqBuscffxwdO3a0NDSyo9IWswOAWbNmlVptTJWHWT2KW7ZsiYsXL6JLly5Sx0NERBWALlHsFuwGmZtZXY7somibjLy4PLj6mvVrkoiIKiFPT088+uijSEpKwtSpU6HRaNjnkSqMH374AQMGDDDY5uvra7Boo6P674KTJX2tVCoRFhZW7ljO0qMYAJo0aYLz58+Xe1xKSgoCAwONGrNVq1bIy2OLNme2fPnyUhcuLCgowP37920cETkas/56/+CDD7BmzRp8//33UKvVUsdEREROLi++8A2kM/UnBv6TKI7nm2AiIjLNkCFD8MEHH2Dp0qXYunWrvcMhkszvv/+OJ5980mCbsyxmp1twMioqqti+qKgoqNVqo5LEQGFS1RqJ4vjl7yJ++buSjtmkSRNcvHix3OPCw8MxYcIEo8Y8fPgwJk2aZGloZCf5+flo1qyZwZ0BRYWEhLAHNZlXUdyjRw/k5eVhxIgRAAobpRf9QRMEwel6FRERkXR0FcXOlih29XeF4C5AzBO5oB0REZlswYIFmDFjBlq0aIEjR47YOxwiyezcubNYcunpp5/G8OHD7RSRaaRacPLrr78utb+rJTRG9BI2VcuWLfHxxx+Xe1xAQABCQkKMGpM9ip3bnTt3MH78ePz5558l7tfl+qhyMytRPGvWrFI/gSAiInLWRLEgCHCv6o7cf3OZKCYiIpNoNBqsX78e8+bNQ4sWLfD+++/bOyQiSfz444/IyMjAM888Y7A9OTkZp0+fxuDBg+0UmWl0C04OHToUGo0Ge/fuxfz5802K/8qVK2jSpInksVVfvkzyMeVyOc6dO4dq1arpF9gsyfvvv4/Q0FBMmzat3DGZKHZuSqWyzA86OnTowEQxmZcoXihxk3UiIqpYnDVRDICJYiIiMsuVK1cQFhYGV1dXPPLII1i5cqW9QyKSxNdff42nnnqq2Pb79+9j06ZNTpMojomJwZYtWzB79mycOnUKTzzxBMaMGWP0+Wq1GtOnT8fQoUOtGKW01q1bh5o1a6Jt27alHpOSkoJGjRoZNV6dOnXw1ltvSRUe2VhcXFypC9kBwG+//aZvM0uVl1k9inv06IHLly+XuO/q1avo0aOHRUEREZFzc/ZEMQAmiomIyCTp6ekYNmwYAMDV1RWurq5ITk62c1RElikoKMCRI0fQq1evYvucpUexjqenJ5555hm4urqib9++mDJlikkLTlqrPzEAJHzwARI++EDycY3pU2zK8/Ly8rJK6w2yjV69emHOnDml7g8ODmaPYjIvUXzs2LFSbzfIyMjAr7/+alFQRETk3HRJVrcqbnaOxHTuVZgoJiIi03Xo0AGzZs3SP964cSP/LiKnl5aWhueffx4+Pj7F9ikUCqdrSXn37l00a9YMKpUKGo3GpHOTk5OtlijOu30bebdvSz5ukyZNcOvWrTKPmT17Ntq1a2fUeKIolvihATmHGzduQKvVlrqfi9kRYGaiGECpvxD++OMPhIaGmh0QERE5N022BhpV4RtvVhQTEVFlMXToUNy9e1f/uEWLFvjnn3/sGBGR5by9vTFv3rwS99WsWRO//PKLjSOyTLVq1bB48WJUr14dDx48MOnc5s2bY//+/VaJq+aaNai5Zo3k406cOFHfm7k0GRkZZfYwLsrFxaXMRCM5tnXr1uHs2bOl7g8ICMA777xjw4jIERmdKF62bBkUCoX+U8Pu3bvrH+v+k8vleOWVV5xm5VMiIpJeXvzDBKszJ4rz4/PtHAkRETmLgoIC/P7776hevbp+GxPFVBFERETg/v37Je4TRdGgit4Z5OXloVGjRvjmm28QEhJi0rnnzp3DsWPHrBOYlWg0Grz55ptlHjN27Fjk5OQYPaa7uztyc3MtDY3sQKlUltmj2MXFBY0aNeKHAZWc0YvZdezYEbNmzYIoili8eDFGjRqFmjVrGhzj7u6Oxo0bY+DAgZIHSkREzqFoJa4zJ4rz4vMgakUIMue6pZKIiGzv0qVLCA8Ph4uLi35b586d0alTJztGRWSZW7duQRRF1KhRo8T9giBg69atWLVqlY0jM9+GDRtQtWpVdOvWDXfu3EF4eLjR58bExOD27dvo37+/5HElbdwEAAieMlnScd3c3PDFF19g/vz58PDwKPGY9PR0KBQKo8eMjo6Gq6vRqSRyIHFxceX2mH7qqadw7NgxdgqoxIz+1921a1d07doVQOEvhIkTJ5b6C4OIiCovXaJYcBXgFuiEPYr/lygW80UUpBbALcj5ngMREdmWl5cXZs6cabDNzc0Nq1evxuTJk0vs71qRxcbGAgBatmxp1zjIMvv378eTTz5Z5jEuLi4oKChwmsTh3bt30a5dO0RHR0OlUpXaVqMk1uxRrP77b6uMCwDh4eG4evUqWrRoUWyfRqOBTCYzqdd0XFwcatWqZVJymRzDiRMnyv23qutTzERx5WVWj+IFCxYUSxLfvn0bhw8fRkpKiiSBERGRcyq6kJ0zVuMWrYJmn2IiIjKGn58fBg8eXGz7oUOHcOHCBTtEZD85OTno3LkzOnfubNLt7OR4nnzySUybNq3MYzp06OBU3+e7d++iZs2aqFu3Lm6bsXhc0fYyUqq1YT1qbVhvlbF79+6NjIyMEvfJZDL9BzvGWr58OW7evClBZGRLGo0G77//PmSystOAXNCOzEoUz5o1Cy+//LL+8a5duxAeHo4+ffogLCwMp0+flio+IiJyMrrkqjO2nQAA9ypMFBMRkWkGDBiAK1euFNteGfsUHzx4EFlZWcjKykJ0dLS9wyEzxMbG4o8//sClS5dQu3btMo/dtWuXU1XMb9q0CU2aNDErUTxnzhyMGjXKOoFZ0VtvvYXOnTuXuC85ORl//PGHSeMpFIpSE8/kuBITE7F169Zyj3vvvffQunVrG0REjsqsRPGuXbvQtm1b/eM333wTTzzxBP755x9ERERg7ty5kgVIRETORZ8oruKciWIXbxe4+BT2mGSimIiIypOfn49bt26V2Ou0ffv2UKlUdojKtlQqFQRB0LcolMlkkMlkiIyM1G/PzMy0d5hkBF1FeI8ePfDll1+We/zMmTNx48YNG0RmOVEUcfLkScjlcjRv3hw//fSTSecvX77cas81ZcsWpGzZYpWx//nnn2KtcXSuX7+OzZs3mzQeE8XOqbyF7HTy8vJKXcCSKgezEsVKpVL/yeKNGzdw5coVzJ07F82aNcNLL72EU6dOSRokERE5D2evKAaKLGjHRDEREZXj4sWLaNSoUYm38w4bNqzUBE1FEhcXp/86KSkJWq0WWq0WSUlJ+u1KpdIeoZGJdBXhubm5Rq1JdPXqVae5TT01NRXz588HUNhbefXq1dBoNEaf/8MPPyA/P98qsamOHYPq2DGrjB0SElJqdX9ycjICAwNNGm/9+vVWWdCPrCs7OxtNmjQp97jo6Ghs377dBhGRozIrUezn54eEhAQAhX23AgMD0aZNGwCAu7s71Gq1dBESEZFTyYuvQInieCaKiYiobPXq1cPatWtL3CeKIoYMGQJRFG0clW2FhYVBrVYjKiqq2L6oqCio1WqEhYXZITIyRkkV4QCwefPmcivCFQqF01TN6/oT63z77bcmfYBhzcXs6nzxBep88YVVxq5atSoSExNLTHJnZ2ebvGjZyZMnWRzohDp16oQPPvig3OOCg4Od5sMfsg6zEsVdunTB/PnzsXbtWrz77rsYMmSIft+VK1fK7WNEREQVFyuKiYioMvnnn39KvZ1XEATcvn0bd+/etXFUtufh4YGsrKxi2+VyOTw8POwQERmrpIpwoDAxqlNaQrV79+4mV6Tay4MHD1CrVi39Y1P7FLdo0QIBAQFWiMy6BEHAzJkzkZ2dXWzfiBEjsGLFCpPGO3HiBH755RepwiMb+fbbb41qtxISEmJwNwhVPmYlitesWYOqVati9uzZqF27NpYuXarf9+WXX+Lxxx+XLEAiInIeoigyUUxERJXKK6+8gpSUlFL3t2jRAmfPnrVhRPbzxhtvAADGjx8PT09PANDf6k+Oy5KK8EmTJunvLnZ0/fv3x+eff65/3LJlS5OqoaOiouDq6mqN0JAa9R1So76zytgA8Prrr8Pb27vY9j179nAxu0rijz/+QGpqarnHde/eHVus1C+bnINZr3I1atTAkSNHStx38OBBfmJMRFRJFaQXQMwtvL2WiWIiIqro8vLy8O+//6JBgwalHjNs2DD4+vraMCr7uH37Nm7duoWvvvoKkydPRkBAABo2bIgxY8bYOzQygoeHB9zdi793K68i/IMPPoC3tzcmTpxozfAksWfPHtSqVQutW7cGACxcuNDoczMyMvDcc89hz549Vokt/X/jBox82irjf/TRR1CpVHjrrbcMth8+fBhdu3Y1aayQkBD2HHdCxi5ml5ubiy+++AIvvviiDaIiR2RWRXFZFApFib9giIio4iuaWGWimIiIKrrLly+jWbNmJS5kpzNkyBB06dLFhlHZ3oYNGzB58mR4e3tjypQpEAQBq1atwpQpU/SVxeT4FixYAADo0aMHIiMjAZRfEa7RaPTrFzm6nTt3GiQ4z5w5g5UrVxp1blJSklWTo3W++hJ1vvrSauOHhYXh4sWLxbanpKSY3Dpk2LBhJiXZyTGEh4ejbt265R7n6uqKd955x/oBkcOSPFFMRESVl0GiuIrzJordqrgBAPKT8qHN19o5GiIiclQtWrQot+djUlKSvoKxIjp+/DhWrlyJr776ymB7//79cfnyZTtFReb4888/8dJLL+HQoUP45JNPsGHDBsTExJR5jjMtZnfv3j2DxewEQcCBAweMOjclJcVqC9kBgCCTQSjjAydLNWnSBJcuXSq2PTQ0FNWrVzdprGvXrmHVqlVShUY2smTJEtSrV6/c4+RyOfLy8ir8IqxUOodJFF+9ehX9+vWDt7c3QkNDMWPGDKjVapPG2LVrFwRBQLNmzawUJRERlUWXKJZ5yuDi62LnaMynr4YWgfzE4itEExFZ0/Xr1zF16lS0bNkSrq6ufG/rwD7//HPcuXOnzGOCgoIQHx9v8t82jio2NhaxsbH6x7/88guioqIQGhpqcJyXl5fTVJoSoFar8f777+P999+HTCaDIAhGVYRHRESgW7dutgnSQt7e3mYvZufn54ennnrKSpEVtp5It1JbCwCoVasW3nzzzWLb16xZg6ZNm5o0VnZ2tlGLopFj6dChg9HHhoeHl7g4KVUO1unEbqK0tDT06NEDderUwc6dO5GQkICZM2ciOTm52CfTpVGr1Zg5cyaqVKli5WiJiKg0+fGFSVX3qu4QBMHO0ZivaNuMvPg8uFZxiF+XRFRJXLhwAfv370f79u2h1Wqh1fLOBke1du3acvt7CoKApk2b4uLFi06z6FdpcnJy0LlzZwDA3bt38dtvv2Hu3LklHhsSEoLExERbhkcWWLFiBRITE01+/9asWTM0bNjQSlFJa9++fQaP/f398fjjjxt1blhYWKkL+klBt5Cd3+DBVhlfJpOhdevWUKlUBj3TR48ejc2bN5u0zhQXs3M+mZmZJrVO+f33360YTfl0H0a2bNnSrnFUVg5RUbxx40akpqZiz5496NevH8aMGYMPP/wQ27ZtK/H2iJIsW7YMtWvXRr9+/awcLRERlUZXUezM/YkBwD20SKKYfYqJyMYGDhyIu3fvYseOHRW6ZYGzy8nJwb1791C/fv1yj507dy6qVq1qg6is6+DBg8jKykJWVhaGDh2K48ePl3rs6NGj0aRJExtGR+a6c+cOPv30UyxatMjkc3///XeMGjXKClFJKyUlBc8//7zBNkEQ8MUXXxh1/tq1a7F8+XJrhAYAqP3pJ6j96SdWGx8o7EH9yy+/GGz7+eefIZfLTRonICDA5Cpksi9jF7LTWb16dYk9rW1B94Fk586dkZOTY5cYKjuHSBT/+OOP6NWrF4KDg/Xbhg8fDrlcjh9//LHc82/cuIFVq1bhww8/tGaYRERUjoqSKJa5y+AaVFhFzEQxEdlaWQujkeOIj4/H0KFDjarAjIiIQF6ec/4+UalUEAQBgiBg4sSJ+rYEf/zxB1asWAFBEJCZmVnsvLZt21aI5HhlkJmZiQ8//NDkRc0AwNfX1yl6FN+5cwfXr18vtn3q1Kn4888/yz0/MTERPj4+1ggNACDz8oLMy8tq4wPF+xQXFBTAxcXF5Cpyf39/oxPs5Bg8PT0xceJEo48/d+5cif9ebKHoB5LR0dF2iaGyM/pe2ubNmxv9AiIIAs6ePWt0EJcuXcKECRMMtsnlcjRo0MCoiuIZM2ZgzJgxePTRR42+ZkZGhsHtEroyfI1GA41GY/Q4pdFoNNBqtZKMRcbhnNsH5932HHnOc5W5AAoXg3PE+EzhXtUdBckFyH2Q69BzXlFxzm2Pc/6QM84B39vaXs2aNbF27Vqj5uTixYt4/fXXcejQIZOu4Qjzfv/+ff3XSUlJ+q/z8x/28L93716x2/J3796N48eP46OPPrJ+kBJyhDm3pdjYWAQEBGDgwIFmPWdvb29kZGRYNF+2mPM7d+6gRo0axa7h7u6O69evIyIioszzk5OT8cgjj1gtRlV04WuDb5/eVhkfKOw7u3fvXv3vhZycHDzzzDNmPae+ffvi4MGDVoiy4rLna0toaCjGjh1r9LWDgoKQkJBgs1hVKhUCAgIAAMHBwfoPzCMjI/W/d9LS0kz+sMZRX8/t0VrDlDkwOlHcpk0bq/WbTE1Nhb+/f7HtAQEBSElJKfPcffv24Y8//sDVq1dNuubq1atLvLUmOTnZ5FsvSqLVapGeng6AVSG2wjm3D8677TnynGffzwYA5PvmO39fwv8V1aTfSod7krvDznlF5cg/5xUV5/yh5ORke4dgMr63tb2lS5eiT58+aNeuXbnHBgcH4/z580hISDDpbypHmHd/f3/cunUL0dHRmDJlisG+jRs3ok+fPvDw8Cj2e9/NzQ337t1zuvcDjjDntlJQUIDnnnsOy5cvh5eZ1awuLi6YMWOGRd9nW8y5VqtF27Zti8UZFBSECxculBt/t27dULNmTav9PGd89hkAIKdVS6uMDwCPPvooGjRogMTERGi1WuTn5+P111836zmdPXsWcXFxcHFx3sWrbc2ery2ffvopMjIyMHPmTKOOb9CgAQRBsNnr982bN/VfF/1AsujXFy5cMKrVU1GO+Hqek5ODLl26ACh8Tqb0B7eEKe9tjU4Ub9682ZxYjFbSGyZRFMt8I5WTk4OXX34ZixYtMmhbYYyZM2calN4rlUpEREQgKCgIISEhJo1VEl22Pjg4mC+eNsI5tw/Ou+058pzfSLoBAPCv5y/Ja6k9JddMRjay4ZLhov8d44hzXlE58s95RcU5fyg3N9feIZiM721t7/jx43j55ZeNnt+QkBAUFBSgevXqRl/Dkea9pL+3QkJCUKtWrRKPDwsLQ3p6utO9H3CkObe2tWvXolWrVnjyySfNHkMURWRkZFj0fbbFnA8uZZG4gQMHIiEhodz4H3vsMQQHB8PV1ToLHAesXwcAcDWj/YexQkJCsHXrVjz33HPQarX4+++/sWPHDnz66acmj+Xv7w8PD48SC/6oZGfOnEFqaioaNGhg89eWzMxM1K9f3+h/p5MnT7ZyRIZCQkKQmZmJffv2Fet5/s0332DgwIFmJVQd8fV87969yM4uLK6KjY3FwIEDbXJdU97bOsQy7gEBAUhNTS22PS0tDY0bNy71vPfffx8ymQyjRo1CWloaACAvLw9arRZpaWnw8vKCu3vJfTIVCgUUCkWx7S4uLpL9AMlkMknHo/Jxzu2D8257jjjnokZEfkLhbage1T0cKjZzyKsVVuDlx+fDxcXFIee8ouOc2x7nvJAzPn++t7UttVqN+Ph41K9f3+gK4X379qFKlSomz5+jzLuuYt3DwwOjRo3CF198gYULF2Lo0KElHh8WFob33nvP7nGbw1Hm3BqK3vJ89epVi79Hoiiid+/eFlceWnvOX3jhBYwaNQqPP/64wfZWrVohMzOz3OsOHToU3377rckVjcZysdEHKmvWrEGfPn1QtWpVJCcnQxRFs+Z8zJgxFfbfiDXk5OSge/fuEEURCQkJpeaprCU+Ph6dOnUy+vv1xx9/4IcffsA777xj5cge8vb2LjEZ7OnpCW9vb7PHdYTXc5VKpX+PVrS1xqRJk/RV0yqVyqp90E15/mYnirVaLY4cOYKrV6+WuBKhsSXtANC4ceNivYhzc3Nx48aNYr2Li7p8+TKuX79e4qciAQEBWL9+PaZOnWp0HEREZL785HxAW/i1sy9mBzx8Dnnxzrn4EBERWU92djbeeustk9pIFBQU4NSpU+jYsaMVI7OemJgYrF+/Hp9//jk+++wztG/fHmPGjCn1eC8vL65Y72BycnLQuXNnAMDp06fx8ccfWzym7t9AeXcD29uFCxf0PVCLSk1NRfv27XHt2rUyz09OTkZQUJC1wkPmL78AAHy6drXaNYDC3MvFixdRtWpVpKWlmbWAIQDMmzdP4sgqNt0CbQBw6NAhDBkyxKbXHzt2LMLDw40+3sXFxaR1x6SyYMECAIVrlvXr1w979uzB3LlzS70jwFnExcXpvy6ttYZSqSzW699ezEoUx8XFoVu3brh69SoEQYAoigAM20eYkih+4okn8Pbbbxu8+O7atQu5ubl44oknSj1v9uzZGDdunMG25cuX48qVK/jiiy/QsGFDE54VERFZIi/uYUK1QiWK45goJiIiQx4eHpg2bZpJ51y/fh3btm1z2kSxp6cnZs6ciZdffhmCIBTrV1ySp59+Gg8ePGDVoYMomqzq2bMn7ty5I8n3xtvbG9nZ2RZV/Vnb3bt3S2yToru7WavVltnDNCMjo8S7NqSStHETAOsnips2bYo7d+4AABo2bIgmTZqYNc6zzz6L6dOno0OHDlKGV6GUVUWquxPD2lWkOjVq1EBoaKjRx4eEhNilv3xMTAw2bdqE1atXY9euXZg4cWKJ6y84m7CwMKjVauzduxcjR4402BcVFYVBgwbZrFexMczq5jxz5kwEBQXh7t27EEURJ06cwO3bt/H2228jLCzM5IXlpkyZAn9/fwwePBgHDx7El19+ienTp2P06NEGrSciIyMNegI1atQI3bp1M/ivatWq8Pb2Rrdu3Uzq/0VERJYpmlB1q+Jmx0ikoUsUa9I10Kgda6VcIqrYsrOzsWPHDuzYsQN37txBRkaG/rGzLQxWUb344ovYu3evSee0aNEC//zzj5Uiso1jx47hwIEDRh8fFBTklItDViQqlQqCIEAQBEycOFGfrFKpVHB1dYUgCMjMzLToGj/88IMki2Za07Rp00pM9AqCgOrVq0OpVJZ5/o4dO6xaMV1j5QrUWLnCauPrzJs3T9/PPjw8XL+olqlkMpl+kTAq2X+rSLVaLbRabbEqUlvo0qWLvl+vMapWrYpBgwZZMaKSeXp6YurUqfj6668hCAI+++wz3Lt3D3/99ZfNY5Gah4dHiS1H5HK5QyWJATMTxb/++itmzZqFatWqASi8zaR27dp488038dxzz+HFF180aTx/f38cOXIE3t7eGDZsGGbOnIlRo0bhk08+MThOo9GY9MNNRES2o0sUu/q7wsXD+SuHilZF58fn2zESIqpsEhISMGLECIwYMQLHjh3D3bt39Y8vXLhg7/AIwKlTp9C6dWuTzqlRowaysrJQUFBgpaisLzo6Gv/++6/Rx9urKo0eKilZBRRWyOpYmqxKTEzUVyo7ooKCAkydOrXURO/atWvLrBZWq9VW7ynrVr063GxQ6JaSkoLFixcDKOw7vmvXLrPG8fPzM/gZouJ0VaRRUVHF9kVFRUGtVtuk1YBGo4FWqzXpZ9jLywtvvfWWFaMq3X9bouTl5WH8+PFOudDwf+laaygUCkRGRgIA5s+fb8+QSmRWoli3eq1MJoNCoUBCQoJ+X4cOHfDbb7+ZPGbDhg31t8IkJibiww8/hKenp8Exmzdv1re5KM3mzZtx/vx5k69PRESW0SWKK0LbCcDwebD9BBHZUt26dSGKYon/devWzd7hVXpZWVlITk5GzZo1TTpPEATcuHHD4A5JZ3Pt2jWTEhvbt29nO0A7s0Wyavny5bh+/bpFY1jT2bNnMWzYsFL316tXr8zq2Dt37uC1116zRmh6WTEnkBVzwqrXAAAfHx+sW7cOQGF/5pL6NhtjxIgRaNasmZShVUiOUEWamJhoUtsJnUceeQRqtdoKEZXt8OHDWL9+vf5xly5d0KVLFyxdutTmsUgtJiYGS5cuxciRI/HJJ59gw4YNiImJsXdYxZiVKK5Xr57+U8emTZviyy+/1O/btWuX2Q3RiYjIeemSqRWh7QQAuAW5Af8rjOaCdkREpCOTyRAVFWXWbeh79uzB4cOHrRCVbSgUCpMWRLp27Vq5i4SR9Vk7WeXr6wuVSmXxONZy9+7dMj/Y+e6777Bly5ZS91t7ITsASPzwQyR++KFVrwEUfs/9/PyQmJho0WJ27du3L7HnMxWnqyIteheKLatIAwIC8O2335p8nkKhMGiTYStKpbJYG9nly5dbtUe4rXh6euLNN9/Epk2b9L3+/1sg6wjMShQ/+eSTiI6OBgDMnTsXu3btQmhoKGrUqIF169Zh+vTpkgZJRESOr6JVFAsyAe6hhc8lP46tJ4iIqNDt27fRoEEDs85NSkrCwYMHJY7Idj777DOTKqkPHTqEI0eOWDEiMpYuWRUeHi75Lc8KhcLhE8VlJTXr1q2L27dvl7rfFoniakveRrUlb1v1GjqDBw9GSkoKpk+fbvZr2WeffYbVq1dLHFnFFBMTgw0bNqBdu3aYPHkyXnjhBZtWkcbHxyMvz/SiF3u1Dnrw4IG+za2OQqHAq6++ih07djh1+yYA2L17d7E2u47GrPueli1bpv+6f//++P3337F7926o1Wr07t0b/fv3lyxAIiJyDrqq24qSKAYKn0ueMg958XlwNe9XJhERVTDLly/H8OHDzVrop0WLFti+fbsVorI+pVKJOXPmYPPmzUafExoaatAjl+wnJiYGffv2xZQpU/Dss8+iXbt2GDNmjCRjr1271uo9fC3Rr1+/MvfXrVsXd+7cKXV/nz590KlTJ6nDMiCvX9+q4xf13nvvQaPRQKlUwtfX16wxFAoFbt26JXFkFZOnpyemTJmCOXPmoGvXrmjXrp1N/70cPnwYly9fRqtWrUw675VXXimWsLWFBQsWwMWl5PVu9u7di5s3b+L1119HbGwsAKBly5a2C04C58+fh4+Pj73DKJMkf/W2a9cO7dq1k2IoIiJyUhWtohh4+Fzy4vPgBS87R0NERPYWGxuL3377De+8845Z5zdt2tSsXpGO4MqVKyYvXlWlShW2nnAQnp6emDx5Mnr16qW/5Vkqp06dglwuR0REhGRjSik3Nxd169YtdX+zZs3K/ADk3LlzcHNzs2pVsfrsWQCA56OPWu0aOj///DN+/fVXfPzxx4iPjzdrDIVCUWZfZypuyZIlSExMxDPPPIO3334bjz32mE2uGxcXh6pVq5p8XocOHcxqsWSpH3/8EU8++WSJ+9asWYM2bdqgf//+6Ny5M4DCO3Vs1e9ZCvfv30f37t3tHUaZzGo9ofPgwQOcPHkSv/76a7H/iIiocqnIiWK2niAiopycHHTu3Bm3bt0yu6+nt7e3wfouzsTUhewAYPjw4VizZo2VIiJTZGdnY8CAAahSpYrkY//xxx/45ZdfJB9XKs8//zzu3r1b6n53d3dER0dDq9WWuH/Xrl04efKktcIDAMS/+x7i333PqtfQCQgIwL59+yCKotmJwA4dOmDatGkSR1axDR06FBqNBr169cKBAwdsdt24uDizKoM//vhjfPHFF1aIqGzz5s2DRqMpcV9QUBA++OAD7Ny5E1lZWcjKytK3xXUWLVu2dPiFIM2qKL558yaee+45fV8VURQN9guCUOo3loiIKh5trhYFKYX9oipioliXBCciosrr4MGDyMrKAlDYe9ec1hMAMHv2bPTr1w/dunWTMDrrq1WrFuqbeHt8YmIivvjiC7z++utWioqMtXv3bvzyyy/YuHGj5GP7+vo6dIuR8hazA4DVq1ejT58+xRbRAgp7FLdp08Za4QEAqs59y6rjF1W3bl38/fffEAQBOTk58Pb2NnkMX1/fUtsDUHEFBQU4efIkXFxc0LdvXzz//PNYuHChTa69atWqYjk7Y4SEhODff/+1QkSlE0URqamp8Pf3L7ZPpVLpF7QLDg6GIAgQBAGRkZH6RfdUKpXDt3WQ8m4OazGronjSpEn4999/sWnTJvz222/466+/DP6z9qdtRETkWPISHiZSK2SiOJ6JYiKiykilUun/GJ04caLBH6a6rzMzM00a09vbG2fOnLFSxNbTt29f9OzZ06RzBEHA559/bqWIyBTHjh2z2u3OISEhDps01Gg0yM7OLrcXb506dUpd0E6tViM4ONgK0T3k0aQJPJo0seo1dI4fPw6gMCl36NAhs8a4d+8enn/+eSnDqtCKtn9o1qwZ1q9fb7Nrb968GdnZ2SafZ4/F7DIzM1G9evUSK92LfhiVlJQEURSh1Wr1SWKgsJd+UbGxsTh//rz1AjZD48aNzVpc0JbMqig+efIktmzZgmHDhkkdDxEROaGiFbfuVSpQorjKw9YT5nwST0REzu2/f5iW9LVSqTSpJUOLFi2wa9cuaQK0oVatWiE6OtqkHsuBgYFITk62YlRkrGPHjmHx4sVWGXvEiBFWGVcKgiAYVchWv379Uquit27dKnVYxeRcuQIA8AgPt8r4/63GlMlkEEURkyZNwtChQ/XHGFuNyR7Fpnv22WcBQP8h499//23yAnPmePfdd/HUU0+ZfF7Xrl3RuHFjK0RUOl9fX5w7d67EfWFhYVCr1di7dy9GjhxpsO/bb79F//799T/jQGG7qK5du0IURSQkJJhVOS+1goICpKSkOPTin4CZFcU1atRw2E8MiYjI9vSJYgFwC3GzbzAS0lUUa3O00KpK7ltHREQVl+4P06ioqGL7oqKioFarTe7bGxERga5du0oVok1otVrcv38fISEhJp0nk8nQqlWrUnu/ku3s3LnTrAWtjHHlyhV88MEHVhnbUg8ePMD169fLPe79998vtRDu7bffhkqlkjo0A3GLFiNukXUS+UDxD720Wi1EUSyzGrMsCoXC5MUtK7OaNWti1qxZ+seXL1/Gxx9/bJNrp6WlldjKoTxeXl5ISUmRPqAy/PPPP2X2Rfbw8CgxyZqWloZGjRph0qRJ+Pnnn6HRaPTtorKzs82unJdaQkKCVfrES82sRPGSJUuwfPlym//QEBGRY9K1ZnALcYPM1aJ1Uh1K0TYamkT23iciqoxK+8NULpebtdJ6tWrV0KhRI5w+fVqK8GxC1+PVnIWvoqOjIZNVnPcGzsjaP2sZGRk2XZzLFCdOnMB3331X7nEXL14stU3Kxo0b4epq1s3YRgt9dRZCX51V/oFmkvpDL09PTxw5ckTKECu0devWYfXq1frHvXr1wuHDh61+x6JarcYjjzxi1mt3VlYWIiMjrRBV6c6ePYuzZ8+WecyCBQsAAJGRkfr41q1bh0uXLqFt27bo1asXXF1dMXbsWAiCAJlMhkmTJpndLkpKHh4emDt3rt2ubyyzXu22bNmCe/fuoW7dumjZsmWxTycEQcCePXukiI+IiJyArqK4IvUnBgyfT0FCgR0jISIieyr6hykAfPbZZ5g/fz4GDx5s8lg5OTno3Lkz5HI5UlJSzEo225qPjw+WLl1q1rlvvPEGJkyYgHAr3VJP5Vu/fj369u2L5s2bW2V8X19fq1fcmuvevXvlLmQHFP67jIqKwoQJE0rc5+npaY3w9Lxat7bq+IC0H3rpWno0aNDArCRkZXPnzh00aNBA/zgoKAjh4eG4d+8eatWqZbXrenp6IiYmxqxzAwICbF4c+uDBgxIXlCwqJiYGW7duxeTJkwEA7dq1w5gxY+Dp6YkePXroj9O1Rvn/9u47vKmyDQP4ne5BJx1UCqVA2cieFpC9QbaCsgrUASogiqhMFQdDRfwKqKCIWBBRkFGmCGjBQZG9yuyge9KmbXK+P2piQ5I24yQnbe/fdXHRnPGeJ2+b5OQ573leXSPnjb0TSCxOTk4YOHCgJMc2hkmXdvPy8tCwYUO0a9cO9vb2yM3N1fjHWxCIiKqXqpootve0h51L6UdlSSoTxURE1VVsbCy6dOmCV155BRs2bEBUVJTJX75jYmKgVCpRUFCAAwcOiBypZRQXF+Oxxx4zad/bt2/j7t27IkdExvjll18sWu7E19cXTZo0sVj75khNTTUoEVevXj2dk9kVFxejffv2FohMU9Ht2yi6fdvix1Fd9Jo6daq6Zu7ChQtNamvBggUmTZJWHSUkJGglQA8cOGDRJDEAnDt3zuSyMPb29pDJZCgpsd53IHd39wrrIru6uiIyMlI9QjgyMlJ9IccS5aLEtHnzZnzwwQeSHd9QJo0oPnr0qNhxEBFRJVZVE8UymQxOtZxQeKuQI4qJiKoxmUyG+Ph41K9fX/3F1BgPTySlGoEXERGhHulkzERS1vb++++jdevWmDRpktH7BgQEICUlxQJRkSHS09MREBBg1CSExgoICMDnn39usfbN8fbbbxt0e7+vry9CQ0MhCILGCFlHR0erlNVIfH0BAKDet1ssehzVaMyIiAikpqYiPDwckydPNqktVZ1iW5gkzNaNHj0arVq10lh27949fPDBB/jkk08sdtxLly7h8uXLJu9/7tw5q85PNnPmTLPbELtclJgMGTFtC1gsioiIzFZVE8XAf89JkcIaxURE1dWJEyfQqVMnk2cqf3giKUEQzJpIytquXbtm8iis7t27V4rJe6qqmjVr4uTJkxY9hiAIGDBggEWPYar3338fmZmZFW4nk8mwf/9+rTIKV69exfz58y0VnprfC8/D74XnLX6ch0djzpgxw+SyGpzQznCdO3dGUFCQxrJatWph+/btFh2VnZycrHVcY5w4ccKqF/pmzJiBhIQEs9spO3K+RYsWAEwfOS+mhIQE1K5dW+owKmTSiOKlS/XPxmlnZwcvLy+0bt0a3bp1MzkwIiKqPNSJ4sCqlyh2DHQEwNITRETVmZ+fH1566SWT91fdDrtr1y6MGzdOY110dDSGDRsm+Uin8ly9ehWNGjUyad/Ro0eLHA0Z491338XQoUMtVp8YKE2y2urkjGvXrjV4lOKCBQswevRotC1TLzghIQHx8fGWCk+thomlXaT066+/cqJKAwiCgBYtWmhcMAQABwcHhIeH49ixYxarW+vr64uQkBCT9//222/h6+trtYt9hw4dwooVK8xup+zI+b59++LVV1/F4sWLzQ/QTC+//HLVTRR/+OGHKCkpgVwuB1B6O0ZxcTGA0uHcJSUlUCqVaNu2Lfbu3Qt/f3/xIiYiIptTHUYUs/QEEVH15ePjg9atW5vVhi3fDluRbdu2oWbNmibtGxMTg/3792P16tUiR0WG+PzzzzF9+nSLH8fR0RFFRUUmj7q3BIVCgcLCQoNLI+Tk5ODKlSsaieL09HT4+vpaKkS14vv3AQCOlWj0/fbt29G6dWtOVFmBnJwceHl56Zz0LyIiAs7OzhY79tNPP23W/v7+/khNTRUpmvIJgoCcnBx4eHiY3ZZq5LxCoUBAQAC6dOli8QkpDZGTk4NmzZpJHUaFTLr888svvyA4OBgbN25ERkYG5HI5MjIy8MUXXyA4OBjHjh1DTEwM7t27h3nz5okdMxER2ZCSvBIo85UAqniimCOKiYiqpYSEBPTt21eUtlS3w0ZERKhrFtvC7bDlSUtLQ2Zmps4khyHc3d1x8+ZNkaMiQ9y6dQvu7u5WGbg1ZcoUq056ZYjs7Gw8/vjjBm8fGhqqNaFdQEAAunfvLm5gOiTMnoOE2XMsfhwxHT58GBcvXpQ6DJtXXrmBAQMGoEOHDhY79owZM8yqURwUFIS8vDwRI9KvqKgIU6dONfmzRp82bdpYNBlvjJEjR0odgkFMGlH8wgsv4JVXXtGYzMDb2xtTpkxBYWEhZs+ejVOnTuHNN98st0wFERFVfsX3i9U/V+VEsSKVNYqJiKqjmJgY0eqvqm6HnTFjBm7fvo0OHTrgrbfeMqvNuLg4ADB7xLM+p0+fxpYtW9CzZ0+T9rfmiDTSlJeXJ8rkUIZ45513rHIcY/j6+mLbtm0Gb9+1a1et+qjWSBIDQM1pEVY5jphYo9gwwcHB5d5RER4ejh9//BH16tUT9bhxcXE4duyYWZ8x1ryQ6ezsjA8++ED0didPnmwTVQ4KCwvh6OgIBweT0rBWZdKI4rNnz+qtcxIaGopz584BAFq0aIHs7GzToyMiIpsnvydX/+wUVPUSxc5BpVegS1JKoCxWShwNERFZW0xMDPr37y9KW2Unkvr222+xZMkSs26HLSwsRHh4OMLDw1FYWChKjA+7du2ayfWJAaB27doaA4zIepo0aYLIyEirHOuJJ57A+fPnrXIsQx08eBCffvqpwdt36dIFo0aN0li2ePFi7NixQ+zQtHj06gWPXr0sfhwx1a1b1+CkV1xcnPqilrnEbMsaMjIyyi1f0rt3b8TExIh6TNVnw7Vr1+Dl5WVyO2fOnMHmzZtFjEy/Q4cOWeT96tdff8Xy5ctFb9dYSUlJeOSRR6QOwyAmJYpDQkLw+eef61y3fv16dRI5PT0dfn5+pkdHREQ2r+BGAQDAwdsBjj6OEkcjPpf6/9aNVALy2/LyNyYioipn5cqV6N27t+jtJiQkmD0BWExMDPLz85Gfn48DBw6IFJmm+Ph4hIWFmbx/jRo18OSTT4oYERlCEAQ0bdrUaiM+BUGwuUFi586dMyqmzMxMtGnTRmPZ7du3rVJ3WZGdDYWN9V9FXnzxRUyYMKHC7cS8oGWNi2Ni++abb/Djjz/qXd+/f3/RE8WqzwZBEPDLL7+Y3E5aWhr27t0rXmDluHfvHmrUqCF6uzKZDGfPnhW9XWPVqVMHe/bskToMg5g05nn58uUYO3YsGjdujCFDhqhvJ/r5558RHx+P7du3AyitWdOjRw9RAyYiIttScL00UezaUPoJAizBtf5/z6vgegFqNBb/BIaIiGzT5cuXcefOHfTr10/0ts+ePYuzZ8+ic+fORu2Xm5urrm/s5+cHO7vSsT8RERFIS0tTbyPWF+6PPvoICoV55ZeaNm2K+Ph4m6kTWR3cvHkTbm5u6r8VS/Pw8EBubq5VjmWou3fvaiV+y+Pt7Y2EhAQIgqCuk5qenm7yRI7GuPvc8wCAet9usfixxPLLL7/g5s2bmDJlSrnbqZKWAHDgwAEMGzbM5GOK2Za1JCQklFu6p3v37qIMsLTEZ4M1SwdZasStv78/7v87WaSULl++jKysLAQEBEgdSoVMGlE8YsQInD59Gq1atcLOnTuxZMkS7Ny5E61bt8Yff/yBJ554AgCwdu1abNlSed7oiIjIeKoRxS4NbHvGdlPZu9urS2oUxleOkQtERCSO7777Dv/8849F2g4JCcHt27eN3i85OVn9c1paGpRKJZRKpToRAJR+4RbL0qVLzZ5ciHWKre+XX34xua60KZ5++mk0bNjQasczRLNmzdCqVSuDt5fJZAgICNBIKo0ePRoNGjSwRHgafCaMh8+E8RY/jpjS0tJw+vRpnetyc3Mhk8kgk8kwbdo02NnZwc7ODhEREerlhk6SJmZbUihvMjsAcHEp/Q51/fp1s45jic+GWrVqITg42Ky4DNWtWzfRJo4tKygoyCIXe4116NAhs0Z3W5NJiWKgdObAbdu2IT4+HgUFBYiPj0d0dLTFJlEgIiLbVNVHFAP/JcFVz5WIiKoHMesTPywkJMSksgBhYWEoKChAdHS01rro6GgUFBSYVSqiLLlcjnXr1sHe3t6sdgICApCSkiJKTGSYxx57DM8995zVjtejRw+bq785ffp0oxLFALBs2TKNUhOjRo2yyghAr8GD4TV4sMWPI6byJrMTM2lp7YtjYtu4cSPat29f7ja//vorNm3aZNZxLPHZUKtWLbPjMlSTJk3QrFkz0dv18vLC/PnzRW/XWImJiTb3HqmPyYliIiIiQRDUI4pdG1TdRLHquXFEMRFR9SGXy+Ho6IgWLVpYpP3Q0FAcPHjQpH1dXFx01k11dnZWj04TQ3x8POrXr292O0uWLNE7GTqJTxAEJCcno3HjxlY75sqVK7F+/XqrHc8QLVu2hFxu3PwSffr00RhB36BBA7NLrxhCWVQEZVGRxY8jpuDgYDRt2lTnOjGTlta8OGYJ27dvr7DOdf/+/bF//36zj2WJz4Zhw4ZBEARzQ6vQ0KFDcfPmTYu03aRJE8lrWufk5JQ7styWGFyjeNiwYVi5ciXCwsIqrAMjk8nw008/mR0cERHZtuL0YiiyS0+eq8WI4hscUUxEVF04Ozvj119/tegxXnvtNSxevBiursZ/hi5atEj9c8uWLXHu3DksXLgQw4cPFy2+zMxMhIeHm91OnTp1UFTJkmCV2Y0bN/Dyyy/jzJkzVjumh4eH1SbOM0RJSQlSUlKMrou9fv16lJSUYP78+RAEAQqFAg4OJk3tZJQ7k0vr/FamGsXNmjUrdwSomElLa10cE1txcTEWL16MZ599ttztmjRpArlcjtzcXHh4eJh1TNVnw5AhQ1CjRg189913Zn02XLhwATk5OfDy8jIrrookJiYiKCjIIm17eHggJSUFdevWtUj7hoiKipLs2MYyeERxbm6u+kpaTk4OcnNz9f6zpQ8IIiKynMIb/12ZrcqJYtVzK4wvhKC0/BV1IiKS3oIFC/DHH39Y9BjHjx/HnTt3TNo3NjYWUVFR2L9/Pxo2bIioqCjExsaKGl/Xrl3x3nvvmd3O119/ja1bt4oQERnC2vWJAdubzM7U27xDQkJw69YtAEB2drbFk2Mq3qNGwXvUKKscSywpKSkYMWJEuduokpbdunVDREQEAGDhwoUmHa9sW6qkp6ltWUtSUpJByU+ZTIZ//vnH7CQxUPrZsGbNGixduhTffvut2Z8N1qgxLwgC8vPzRZuE9WEP1x6Xwvz58yUf1Wwogy+NHT16VP1zZSnATERElqWq2WvnZgenWuXfUlWZudQvHakgyAXIE+RwqWO7IxeIiMh8giBgy5YtmDt3rkWPU7duXdy5c8ekEgGurq4YM2YMFi1ahLS0NERGRooe33vvvYeePXuiU6dOZrUTEBBg9kRNlU1cXBwASDKHz/HjxzF69GirHrNv377o2LGjVY9ZHldXV7z++utG71evXj11otjZ2RmffvqpyJHp5j1qpFWOIyYXF5cKJ/uMjY1F69atsXz5crRu3Rq7d+/GiRMnTDre0aNH0bBhQ8TExMDFxQXr16/HxIkTTWrLWtLS0gyeDPHKlSv4/vvv8eabb5p1TFdXV7Rp0wYLFizAvn37zP5sGD16tMVH1SuVSnz//fcWa//FF1+UtD6wIAjYsGEDli9fLlkMxmCNYiIiMlnZ+sTmzohuy1SlJwCWnyAiqg4uXLiAoKAg1KxZ06LHeeyxx8za/9KlSzh79iy6dOkiUkSaDhw4IMpt3dYYkWZLCgsLER4ejvDwcElGkG3YsMFikzDq4+XlZZUSDYZycXHBkCFDjN6vefPmeP/99wGUlg1o1KiR2KFVGTVq1KhwFHlOTg7kcjm6dOkCd3d3dOnSBcePHzfpePv378eQIUNw8OBBzJs3D5GRkSaV7bGmtm3bGpwADQoKQlRUFP7++2/1hSZTnTx50uzPF5VXXnkF9erVE6UtfXJycuDr62ux9rt27Qp3d3eLtV+R3NxceHp6VprvywYnitPS0nReLfrnn38wevRoNG/eHL1798bu3btFDZCIiGyXakRxVZ7IDgAcfRxh51P6kal6zkREVHXZ2dnhtddes/hxZs2ahb59+5q8f3x8PNq2batObInt2rVraNiwodnt9OrVC8uWLRMhosohJiYG+fn5yM/Px4EDB6x67Fu3bmHTpk0VTp4ltnPnzlnlNWOodevW4ZNPPjF6Pzc3NyQlJUEQBPz2229WK21w++lncPvpZ6xyLLHY2dlVOFq1pKQEq1evhp1d6Xn0xIkT8fXXX5t0vI0bN2LKlCnw9fWtNHco7Nu3z+D3AC8vL9StW1eUi0y//fabaInid955B5s3bxalLX1OnTqFDz74wGLtf/755/jss88s1n5FcnNz0bt3b8mObyyDE8Wvv/46Jk+erLHs9u3b6NatG3766Se4urri/PnzGDFihMUnfSAiItugHlFchesTqzjVK/3CVbYuMxERVU1+fn4V1t4Uw6lTp8y6zbiwsBAtW7ZEu3btRJ8sTqlUYsSIEaKMwrK3t8fp06dFiMp25ebmQiaTQSaTYdq0abCzs4OdnR0iIiLUy/Py8iweR0xMDK5evWrx4zzM09PTpmoU3717F8HBwSbtO3fuXKSkpCA9Pd3idxWoeAwcAI+BA6xyLDEtXboUgqB//o7MzEw88cQT6seDBw/Gyy+/bNKxVq9ejR49eiA4OBh37941qQ1rO3r0qFGxdu7cGQUFBWZfZNq0aZNoiWInJyckJiaK0pY+ptYUN1StWrUkrVFcu3ZtfP7555Id31gGJ4pPnjyJCRMmaCxbvXo18vLysGfPHvz555+4desWOnfubLEr2kREZFtUo2vLlmaoqlSJYo4oJiKq2vLz89GqVSv1RN6W5OrqipMnT5q8//Tp0xEREYGSkhLRSzvY2dmZNCJTl5KSEsyZM0eUtiwtLi4O58+fN3q/5ORk9c9paWlQKpVQKpVIS0tTL09KShIlxvJIMZEdYHuT2d29exd16tQxaV9VneL09HSL3g5flu+ECfB9KN9SGbRq1Urjb7ys5ORkDB06VCOR7OzsjJo1axr9vrdr1y64uLjAzs4OtWvXxrPPPmtW3NaSkJCA2rVrl7tN2YtMmzdvNvsi0507d7Bv3z7R7irw8/OzeOmgxMREgyb9M5XUk9nt3r1b0hHNxjI4UZyQkIAWLVpoLNu9ezdat26Nfv36ASg90Zk1a1aFBc2JiKjyK8ktQXFKMYDqMaLYsZ4jANYoJiKq6o4dO4Zu3brB3t7e4scKCQnBnTt3TN7/5ZdfRnJyMgICApCSkiJiZMDOnTsxb948Udry8PBATk6OKG1ZUmFhIXr06IHhw4cbfdt3WFgYCgoKEB0drbUuOjoaBQUFCAsLEytULXFxcYiLi4OdnR3Cw8Mtdhx9ateurfO5S2XBggUmTySoShQ/9dRTFpkksipxc3PT+9r+4YcfMHLkSHXZCZW0tDSjJhoUBAFz585Vvyc7OjpixIgRVrmYZ66AgACEhoaWu43YF5kOHTqEU6dOGR+sHl26dDGrRJIhXnrpJURERFis/fbt2+ONN96wWPsVuXjxIvLz8yU7vrEMThSrrmSo3L9/Hzdv3kSPHj00tqtdu7beK0pERFR1lE2YVvUaxYDmiOLybrEjIqLKbf/+/RgwwDq3gHt5eWHUqFEm7//999/Dy8sLM2bMEP0W+StXriAgIECUtmQyGdzd3VFQYNsXW1W1hR88eICDBw8avb+Li4vOUXzOzs6iTAqoT9nJ87744gt4eXlZ7Fj6ODg4iJqcMteDBw9M7of3338fo0ePxo0bN1BSUiJyZLrdmTYdd6ZNt8qxxOTp6ak3Ubxz506MGTNGa3nnzp2RnJyMmzdvGnSM48ePo06dOhoJ18GDB+PWrVsmxWxNq1evRuPGjcvdRuyLTGLWJwaARo0aoVOnTqK1p8uhQ4cgl8st1n6NGjUkzVMmJCRYtLSG2AxOFDdu3BiHDh1SP/75558hk8nUo4lVkpKS4O/vL16ERERkk1S1emWOMjjXcZY4GstTjShW5CpQnFYscTRERGQpr7zyilnJW2O9//77Jl2AVCVdXV1d8cQTT4iW1FW5evWqqCNg4+Pj4epqexeW9dUWnj59ukm3fS9atAhAaQJt3LhxAGDxCdHKTp5nav1Xc8lkMrz44ouSHPthRUVF6r43xf3797Fr1y588sknuHjxooiR6efepQvcu3SxyrHENHv2bL0JsB07duhMMKpea3/++adBx7hx4waee+45jWV16tSx+TrFgiBg4MCBBm0r5kWm7OxsdO3a1ah9ynPr1i2tvJ/Yli9fbtG7TmQymVnvCeZq1aoVHn30UcmObywHQzd88cUXMXHiRGRmZqJWrVr43//+h4YNG6JPnz4a28XExKBly5aiB0pERLZFXZ+4ngvsHAy+7lhpqUYUA6XP3cnfurOJExGR5SUmJuL8+fMYNGiQ1Y45ZswYvPLKK+jcubNR++Xl5WHixIkAgBUrVsDZ2VnUOsAjRoxAu3btRGvvm2++Qffu3Su8DdvaHr7tW9fPSUlJBifNY2NjMXbsWHTo0AFvvfUWWrRogaeeekq8gP+Vm5sLT09PAKU1RO3s7KBUKrF161asW7dOvU2NGjVEP7Y+MpkMSqVSq9SAtZk7MVZ6ejrWr18PpVJptRrFNSOmWuU4YuvYsaPOBOfOnTvh5+eHbt266dzv1VdfNah9pVKJyZMna9zdDqBSTGiXmZmJa9euGby96iLTxIkTkZ2djZ9++gkLFy7E8OHDjTru9u3bjdq+Iv7+/pW+RrG9vT3s7e1RVFRkcu3muLg4ADCppI0ly2pYgsHv4BMmTMA777yDffv2YfXq1WjevDl++OEHODj8l2tOSUnB7t27MXToUIsES0REtkOVKK4O9YkBwN7fHnbupR+bnNCOiKhq2rVrFw4fPmzVYwYEBOD27dtG7+fv7493331X3YbYNYpbt24t6q2yhw8fxqVLl0RrryxVfV5TiH3bt6urK2QyGSZOnAiZTAYnJyds2rTJpNjKo6uuKQCNUXnWmDyvrBo1alisDqcxv2NzJrIDNCezE7ukS1WzbNky7Nq1S2v56tWr4exc/h2HM2fORGxsbLnbfPXVV+oEalnPP/+8JJM2GsOQiezKio2NRVRUFDZt2oSFCxfi7bffrrB/Hnbo0CGsWrXK2FDLVaNGDWRmZpr8HlsRQRDg5OQENzc3i7SvYs7nZNnyPsbWrweAli1b4sGDByYdWwpGXeqbP38+7t27h7y8PBw7dgzNmzfXWK+aSbCyzEBJRESmU9Uorg71iYHSUTKq58oJ7YiIqqaYmBir1SdWqVu3rkkT2kVFRWHNmjUASpPGYiaKc3Jy9I4ENJXYMaqY+wUeEL+28K5du1CvXj0AwNixY7Ft2zbR5zeQevI8Xa5evQoPDw/R2zX2d9ypUyd8/vnnJh/P398fzs7O+O233xASEmJyO8a4O3Mm7s6caZVjiUlXjeLExETcu3cPHTp0KHffVq1aYfPmzeVus3HjRvTv319ruZ+fHzIyMowP2IpcXFzw5JNPGry9q6srIiMjIZPJ8Msvv6CwsNDocj2HDh0SvR66XC6HXC436z22PDKZDPHx8aK3+7DDhw+bPGq5bHmfAwcOGLWvUqlEcnKyxRPhYqr69woTEZFFVLcRxQDg0qD0xIsjiomIqpa4uDicOXMGCQkJoidIKzJ06FA8/vjjRu93+fJleHt7AyhNjM2YMUO0mK5duyZ6kjEsLEzr9nExmPMFvizVqMVmzZrBx8cHgGm1hefOnYvffvtN/bhevXqoWbOmUbegG8rFxUXjDl8VS0+ep8+mTZtw//590ds19nf8559/mjWaWiaTYdOmTXjllVesVkbDpWlTuDRtapVjiUlfonj27NkVvt7HjBmDH3/8EUVFRTrXX79+HSkpKTrr7V65cgULFiwwPXArCAsL06qtbKghQ4bg559/Nnq/kydPijqRHVD6+issLDT7PVaf+Ph4rF27VvR2H3b27Flcv37d4O311a+PiIgwqn59eno6/Pz8zAnd6gyuUUxERKSiKFRAfq90ZlpV8rQ6cK1fmhRXTeRHRESVn2q0IlB6C7+1E2yNGjUyKakVHx+PMWPGAAB8fX1FTdBZIlEs5l2nuurzAqV1IFW1hY2tzxsbG4tOnTrhvffeQ0FBAf7++2+8+eabRsX14MEDbN26FUuWLNFYfvTo0QpvwzdFSUkJpk2bBgCYPHky7O3t8cUXX5hU11QMO3bsQKtWrRAYGGh2W+b8jrdu3YquXbuaPHdSYWEhOnXqhKKiInzwwQdWeU/wf+EFix/DEsaNGweFQqGx7NFHH0X79u0r3Nfb2xtr1qzRWzfWz88P33zzjc6Ec2WoUbx8+XLUrFnTpIt4jRo1AgBkZGQYVSe7R48eaNGihdHHe9jDrz+g9AKKOe+x+ly6dAmnTp3CCxZ+Dezfvx9NmzZF48aNDdperPr1jo6OOsun2DKOKCYiIqMV3iwE/r2DslqNKG7IEcVERFVN2dGKK1assPrx79+/b9IcL6NHj0bTf0cgyuVyDBs2TLSYevbsiblz54rWHlA6alus/tVVn1epVGp9gTdGSUkJ0tLS0KtXLzz22GOYOHEi0tPTjWrjxx9/RN++fbWSJ/b29oiMjBS9/MT333+PgQMHYu3atfjyyy+xYcMGREVFGV3XVCyenp7Izc0VpS1zfsd3795FcHCwyceOiYlRj3K1xAjKqsTZ2VljVGVCQgI6duxo8P5Dhw7F2bNntZYrFArs27dPb8I5KCjI6jW4jXX79m2zJkP866+/jNpfLpdj6dKlsLe3N/mYKrqSpIIgmPUeq4+5k08aKjAw0KgLqmKV93FwcMDgwYONilVqTBQTEZHR1DV6ZYBraPVJFKtqFBenFaMku0TiaIiIyFS6bikFgFWrVhl1S6kYTEl4CIKAIUOGqJMIqknExEpE3rhxw6wEhy5FRUU4evSoKG1Zoj6vk5MTfv75Zzg6OgIoTRaqJgs0VKNGjfDKK69oLXdwcMDFixfx559/GtWeysMTueXn5+PHH3/EuHHjsHnzZjz//PPqv9vIyEij65qKpV69eqL9DZrzO87NzTV6MruH3xNUjL3N3FQJr8xDwivzLNa+pfz555/48MMP1Y937NiBfv36GdXGk08+qXVR5siRI/j666/17uPg4GCxydXEYuxkdg+7d+8eXnvtNYO3//TTT7F8+XKTj1eWNWugOzs7mzz63xi1a9c2ekI5MerXf/fdd3jnnXeMOq7UmCgmIiKjqUbUOtdxhp1z9fkoKVtmgxPaEZGlXL16FQMGDIC7uzsCAgLw0ksvoaCA7zli0jVaEQAyMzPVy601Ws3e3h7u7u5adT7Lc//+fa1ayjVq1BAtkTVnzhykpqaK0paK2JPZiT0B3Z49ezQSH5MnT8bu3bsNjjktLQ01atTQm/AYN26czqRLRR6eyC0zMxN9+/bFmTNnIJPJrFZD1xDvv/8+evfuLVp7+n7Hjo6O5f6ODx8+rJ5M0FCG3mZuKY5BQXA0caItKT1co3jbtm0YO3aswfs7Ojpi5MiRWq+NjRs3YsqUKeXuGxsba5Ga2GKZNGmSwWUOdKlVqxa+/vprgz//T548iU6dOpl8vIfpe/2dPXtW1HIsEydOxIQJE0RrT5+nnnoKy5YtM3o/VdmIJk2aqJcZU7/eWiOmxWQ7nypERFRpqGr0VqeyEwDgHOwMmVNpnTQmionIErKystCrVy/k5uZix44dWLFiBbZs2YLp06dLHZpNenikpaGsOVrKEDt37jTqi/eNGzdQv359jWUHDhwQbSTpzZs3ERoaKkpbKgEBAWjXrp2obaq+wE+YMAH9+/cHYNoEdPn5+XjhhRc0kiLu7u6YNm0aDh06ZFAbX3zxBb755hu960ePHm3ShEZlS6Ps2bMHvXv3xujRo7XqINuC77//Hnv37hW1zbK/44iICPUyfSUu5HK5UaMwVaR+TwiYOwcBc+dYrH1LebjcyDvvvGP063zixInYs2eP+rFCocC9e/cqLKcTHR2Nc+fOGRewCUz9nOnatatZd2Y4OTmhW7duBt2JIQgCTp06JWqiGPjv9RcREaF+/a1atQobN24U7Rhz587F5cuXRWtPn+TkZLz33ntG7xcbG4uoqCgoFAqkpKQYXd7H3JHlUuBkdkREZDTViGJVKYbqQmYvg0uoCwquFLBOMRFZxLp165CZmYm4uDh1UsnBwQETJkzAG2+8oa5JS+ZPQif2iFRz2NnZITk5GXXr1jVo+/j4eK1EcWJiItzd3c0euZSeng4PDw91CQaxuLu7IyoqStQ2Y2Nj8fXXXyM4OBgdOnRA48aN8fPPPxvdzv79+9GrVy84OztrTMy1cOFCyGQyCIKgc0ItFUEQ8NVXX+Gnn37Su02tWrXw6quvIjc3Fx4eHuXGo28it2effRZpaWk4c+YMZsyYIcpEUmK6efMmlEolBg0aJFqbsbGxGD16NJo0aYI33ngDHTp0QI0aNTBkyBDs379f6+LIvXv3TC5xYkvvCZVF06ZN8d133wEoHdEaGhpa7mtFl7Zt2+Knn35SJ2Nbt26NX3/9tcL96tSpY/EJ7Uz9nJHL5Wjbtq3GSHVTqJKzFVEqlfjqq68qfG8xluo9VjUhX4cOHdCvXz8MHjwY7u7uRo0e1+fYsWOYNWuW2e1URCaTYfPmzZg/f75R+7m6umL69Ol46623cPz4cTRp0sSoi7KzZs0SZYJPa+KIYiIiMppqNG11G1EM/PecOaKYiCxh79696NOnj8bIw1GjRsHZ2Vn0kXqVXdmRlqZOOKUaLRUeHo42bdoAMG1Eqrm2bdumMaKuIj169MDzzz+vsezrr782uQZuWTVr1tQ5uZQYVKPlxeLq6orIyEi8/vrrKC4uxtSpU/HDDz8Y3c7Zs2cxZswYreUymQyfffYZ1q5dW+7+V69ehb+/f4UjTjdt2oQ333yzwngsMVmfNYg5mZ2Kg4MDzp8/r64VHBkZifHjx6Nz584YM2YMiouLNba/e/eu0fWJy9I1gtIa7wmJb76JRAP+NmyNIAjqizOvvvqqSSVrZDIZjhw5gk6dOiE8PByDBg1CQkJChfuFhoZqlAuyBFM/Z5KSkkQZRdq/f39069atwtrf169fR6NGjcw+3sNU77Fla6CHhobi4MGDaNeuncaFNVNHXicmJqJWrVoiRq2bn5+fxnuoMQoKChAREYHU1FTs27fPqH2zsrLg7e1t0nGlwhHFRERkFGWJEoU3/y09Uc1GFAP/PWeOKCYiS7h06RKmTp2qsczZ2RkNGjTApUuX9O6Xk5OjUSdSlURSKBQaX+RMpVAooFQqRWnLHLm5ufDx8QGgOdIyIiJC/QUwKyvL4JGWJ0+exObNm5GUlARfX19Mnz4dzzzzjNWfZ506dRAfH691XH39npqaivr162ss9/PzQ3Jystmxx8bGoqCgAD179jSrHV1yc3ORnJwMNzc3UdtNTk6Gn58fXnjhBWRmZhrdB6rkoOr1UrbPe/fujcGDB2P69OlwcND99blhw4aIiYmp8LiDBg3C4sWL8eGHH8Le3l7vdvXr10deXh52796Np556SmPd1q1bMXToULi4uEj+enxYaGgoBEEwOq7y3l8OHjyI9u3bIyAgQGP9O++8g9dffx3x8fFo2LChevnly5cRHBxsct+o3hNU5X7atWtnlfcEmUvp+aW1fqdivacrlUq88sorCAgIwP3799GqVSuT2rx16xaKiopQVFSECxcuIDAwsMJ2nn32WQDi95kYnzN37tzBI488ohGbqX0+YMAAfPLJJ+VO+LZy5Uo8/vjjGDdunFFtm0qV2B09ejSeeeYZ9O/fXz3y+v79+waPvBYEAaNGjYKjo6NF/vbL9rm9vT3at2+P4uJio2u7u7i44O2338Yff/yB7du3GxXrmDFjEB8fL3k9eWNilgliTUtayd27d09960JwcLDZ7SkUCqSmpsLf37/ckwASD/tcGux365O6zwtuFuBU/VMAgPZx7VGjlW3d9mgJZfs86bMkXH/xOpyDndHlbhepQ6uypP47r47Y5/8R+7zQGI6Ojli2bJnWrZHh4eEICAjQO1py8eLFOmuWent7i/b7rOj2e2tQKBTIysoqdxtTnnNeXh6cnJx03nZuDcXFxSgsLNR527Cufs/OzkaNGjU0nqdqwiNz6xQ/ePAAdnZ2FrnVPicnB25ubnoTrqbKzMxUJ3YUCoVRk7wVFxejpKREo98e7vPc3Fw4OTnB2dlZZxuGlJNQycnJgaurq0GlPYqKirRG6Hp4eEj2d2pJ5b2/VLSusLBQ/ftLT08HUDoynson1nu6qs9dXV2NuggkCAIyMjIAQF3i5WG+vr56Y1QqlZDL5aLVZlcR43NGEAQolUqtbUzpc0Pe27OysuDp6Wn1ZKRSqUR2djacnZ3Vcdrae5QYf+dyuRwKhQJubm5Gvd8DQEZGhlm1qsWi+rs25NyWI4qJiMgoZUfSutSvfvXaVCOK5ffkUBQoYO9avRNqRCQ+XV9oKvqiM2fOHEybNk39OCkpCR07dkRcXJxogyDS0tLg5+cn+YWEwsLCCkdaGuvw4cNo1KiRWbesm6O4uBgKhUIrdn39HhISgkuXLmkkZUpKSmBvb2/2F+Lx48dj0qRJ6snhxPTdd9+hffv2GiNAxaAaLQYA69evx7lz57BmzRqD9p0yZQoGDhyorrWpq89v3LgBBwcHhISEaO2/Y8cO/PDDD9iyZYtBx7t58yb8/f0NGvXerFkz5Obmqu8y+PLLLxEaGoq///7boGNZ299//41vvvkGq1atMmo/fX/nN27cwObNm7F48WK9+xYVFWHUqFFo0aIFunTpgpEjRwIAPv/8cwwdOtSk51EdiPWevmvXLnWfb9mypcIJ6Mq6du2auu6+vvGLJ0+e1FvSJTk5Gf3797dIqRxzP2euXbsGZ2dnjbrzpvb52bNnMXPmTBw/flzn+pycHLRp0wY3btwwuE0xlB15rVAo1ElqZ2dng0de//XXX/jwww/Vda7F9nCfz5s3D1OmTEGzZs2Mamf58uVwdnbGnDnGTTh57949DB8+HH/99ZdR+1nCvXv3UK9ePYO2ZaKYiIiMoqrN6xjoCAeP6vcxUrYuc+HNQrg3c5cwGiKqanx8fHTWXMzKyip3IjtPT0/15Fdl2dvbi5bYtbOzE7U9U7m7u+v8ku7q6gp3d9Pekxs3bow6depIdmuonZ0dVq5ciVdeeUXnurL9XlBQAEEQtEY03blzBydOnMAzzzxjVixKpRJNmza1yO95/PjxGkldMVy+fBlnzpxRJ3TGjx+PZs2aYeXKlRWONCwqKsKhQ4ewdu1ajZge7vNGjRrh1KlTSE9PR4cOHTTa2Lx5M1544QWDn1PDhg3xv//9r9xSFkDpKE25XI633npLfbdAx44dMXHiRMlfg/rY2dnh8uXLJsWn6/3lyy+/hK+vb7ntlZSUYN++fdi3b586ISWTyTB9+nR1sio3N9fmJv57WPLb7wAAar35htWOaep7uq7JFpVKJWbMmKFOGhvS502aNEFBQQF27dqlVTIhOjoaw4YNKzchGxQUhPv371vk9WDu58wXX3yBZs2aYcqUKRrLTenzNm3aYMWKFXr38fHxwcWLF63+vlC2HnVRUZH657J1gFNSUuDl5aW3jeTkZK27Y8RWts9TU1Nx7969cst46HLnzh0MGDAA9vb2WLZsGVq3bm3Qhajg4GAcOHDAJt6zjYmBk9kREZFRVCOKq2N9YgBwqeei/vRknWIiElvTpk21ahHL5XLcuHGj3ERxdSP2hFNdu3bVmhTLmmQyGVasWGFQDM7OzjpHJ2VnZ+Orr74yO5bt27cjNDTU7HZ0WbVqFVasWCFqm3///Td+++039WMvLy8MHjzYoBFcDx48wNKlSw26jTg9PV1rIjpBENC1a1f06dPHqJh3796NY8eO6V0vCAKmT5+OWbNmYenSpRoTSYl9m72YxJzMrqioCFu2bNFKtD2s7MR/eXl5AEr7z9Yn/nuYIJdDkMulDsMguiZbVP2sYmifu7i46CxT4OzsXOGoXTs7O3Tv3h1yC/Wb6sLd1KlTMX78eACGf84kJCSIMpkdUPr54Ofnp/c9bdu2bbh27ZooxzJGWFgYCgoKEB0drbUuOjoaBQUFFU7wmZSUhEceecRSIWoJDAzE/fv3jd5v+PDh6ouEvr6+iI2NNWi/K1eulDu/hK1iopiIiIxSeOPfiewa2u4XFUuyc7KDS93SE1fV6GoiIrEMGjQIhw8fVtd8BICdO3dCLpdj0KBBEkZmW2JjYzF+/Hh4enqiV69eiIqKMviL28NKSkqgUCj01p+1ltq1ayMhIaHC7fR98QwICEBKSopZMaSmpuKll14yq43y+Pn5aYxCE0NCQoJWeZV169apJ1YqT0ZGhtbkkfoMHDgQiYmJGre53717F/PmzTO65vK4ceN0JldU0tPTERAQgNmzZxvVrtT8/f3xxBNPiNJWdnY25s+fD39//3K3EyNZZQuCli1F0LKlUodhELH73JwLfzt27LDYe3dYWBiee+45rF+/HocOHcKaNWsM/pwRM1EMlJayeP/993Wu+9///ifZxJbmJPqB0skihw8fbonQdHr88cdNKsfVpk0b9X5t27Y1uPzPkSNHcOTIEaOPJzUmiomIyCjVfUQxALg0+DdRzBHFRCSyyMhIeHt7Y/jw4YiJicHmzZsxa9YsTJgwgSOKy3B1dYVcLoeDgwN2795t1kjL+/fvIzAwUOQIjdeqVSuNCwT6HDp0SOdoVD8/P53lR4yxe/dui9T7VBEjmf0wOzs7NGrUSGOZTCbDhAkTcPPmTb37lZSUoHv37sjJyTHoODKZDG+99RYuX76MuLg4xMXFYejQobh69arRMT/xxBN6R1zGx8ejsLAQUVFRkpVCMZW3tzdeffVVUdrKzs7GCy+8YNC25iaryHhi9nlsbCyioqKwYcMGbNiwwagLf2+//bbe2r3mOH/+PG7duoVPP/0U9vb2aNeuHdq0aWPw58wPP/yAxo0bixZPr169cPz4ca27ToqLi3Hx4kU8+uijoh3LWGUT/fXr1wdgeKI/NDQUbdq0sVhsDxsyZIjRd4AolUqN/m3VqhUGDx5s0L6JiYlWHTEtlsr1yUNERJISBEE9ira6jigG/nvuTBQTkdi8vb1x5MgRuLu7Y+TIkZgzZw6eeuopbNiwQerQbM4ff/yBESNG4N69e2a14+vrK0rJBnN9+eWXaNeuXYXbxcfHq7+Ml+Xk5IQTJ06YfPzCwkI8//zz+O2331BYWGhyO+Vp27atesSgWObOnYsRI0ZoLW/WrBk2bdqkd7/jx4/j0Ucfhbe3t8HHGjt2LIYOHYrw8HB07doVDg4ORk+KBJSWx9i9e7fW8uLiYowbNw5//vmn0W3aAkEQRJkQ8vLlyxg2bJjeCc50EbscjbWlrFiBFJHLsliaWH3u6uqKyMhIk0qs5OXl4cqVK0YfsyL+/v7YtGmT+mJNeHi4wQlpQRAQHR1t9J0G5XF1dUV4eDguXLigsTwpKQnDhw8X9VjGKpvov379ulGJ/okTJ1p1ordjx44ZfBeJiqo8hmqiWDc3N0ydOtWgz8nc3FxRJhS2NiaKiYjIYEVJRVAWlNYhq84jilXPnaUniMgSGjVqhJiYGOTn5yM1NRWffPKJTdcllUJxcTGee+45tGjRArVq1TKrrbS0NDg6OooUmekOHDiArVu3VrhdZmYmGjRooHPdSy+9pHMyREPExMRALpejuLgYBw4cMKmNivj5+Yk+umrWrFk6RylPmjQJmzdv1ntL9g8//IDRo0cbfby5c+ciPz8fBQUFaN++vdH7q5w8eRIzZ85Uj04GgMWLF6NNmzailW+wNplMhsLCQqMSvLqsX78eM2bMUCdmDGHOqFRbUHw/BcX3xR1tb2m20OfBwcG4e/euqG2qSsx06tRJvSwiIgITJ040aP+MjAx89NFHosYElNYibt26tcayunXrYv369aIfyxhlE/1JSUk4c+aMwecs1h5x6+3tjdu3bxu1z61bt1CvXj2NZc8995xBn5Nr1qwxePSxLWGimIiIDFZ2BC1HFAOFtwqhLFZKHA0RUfUjl8sxb948eHh4YPv27Wa19fPPPxuUoLW0nJwc/PLLLxVut2nTJjz22GM618XFxRk1cVdubq56FN+0adNgZ2cHOzs7REREqJerJggTQ1FREXr16iVaewDw008/oUaNGlrLg4OD8c4776CoqEjnfu+++y6efPJJg45Rtp9Uf292dnb44YcfTO6nRx99FDt27EB4eDjCw8Px4MED5OTkYPXq1Ua1Y2tcXFzMmlyspKQEO3fuNDgpp2LOqFRbUPvDD1D7ww+kDsMottDnnTp1MmlUf3k++ugjrRq0gYGB+PXXX9UT95VH7PrEKnK5HE8//bTGMtXFJlsRGBiIH3/8EQUFhg2mSU5ONvtirzFMmcyuadOmeOeddzSWtW3b1qCR0AsWLEB+fr5Rx7MFTBQTEZHBVCNo7b3s4eAr3S1OUlOPplYA8juVY4ZqIqKqZNmyZepyHJMnTzZrgjRbqSEYEhJS4UgnQRAwc+ZMveuNrQGcnJys/jktLQ1KpRJKpRJpaWnq5cYknivi6uoqyohTFYVCgcLCQri5uelcP3r0aJw5c0ZreVxcHA4ePKh3v4eV7SdVHWlz+8nDwwOhoaHIz89Hfn4+1q5dizVr1sDd3d2odmzNu+++a9RI4Ic5ODjg3Llz8PX1FTEqqqo6dOiAsWPHitZednY2vv32Wzz77LNa6z755BOdE4k+LCMjAw0bNhQtJhUXFxecP38e165dA1D6ebBr1y6EhoaKfixT2dvbo3Pnzvjtt98M2v6bb77RWevaUvz9/TFr1iyj9klOTtZ6PzJ0QrsNGzZUyjrpTBQTEZHB1BPZNXQ160tAZVe27AbrFBMRWd/p06fRsWNHAKWjt8y59dhWEsWhoaEICwsrd5uUlBQcPXpU7/oePXoY9aU0LCwMBQUFiI6O1loXHR2NgoKCCmMylpeXl8ETyFWkoKAA48eP17teEASMHTsWGRkZGsu/+uoro0aVidlPZUcnX7p0SX0+tXjxYouM4ra2/v37m3WO+PLLL2tN2FUdpH66FqmfrpU6jEonIyND1MnQzp07p55U9mHh4eEG1YF//PHHLTavwJAhQ7Bnzx4AwO3bt+Hj4wMvLy+LHMtU06ZNM+hzKC8vz6ga8WKwt7fHyJEjDRoZrrJ69WqcPn1aY1m7du0qrMf94MEDuLi4wN7e3qRYpcREMRERGYwT2ZWyd7eHU1Dp1W/WKSYisi6FQoGLFy+iZcuWAIDatWsjISHB5PbWrFmDgQMHihWeyfz8/LBmzZpyt4mPjy939NjMmTPRuXNno47r4uKic0SXs7OzRUZC/fjjjwaP5K1IjRo1yq0F6uTkhDFjxuDbb79VLxMEAT/++KPOCfDKI1Y/lR2dnJWVpR5d/eDBA/VyMUdxW9v48eMNGnWpy/nz53HkyBGrJ49sgfzKZcivXJY6jErHx8cH9+7dE+UuhZKSEnTu3BlvvfWWzvXh4eE4f/58he18//33+PXXX82OR5exY8ciICBA/fj111+3yHHMMWTIEK1ayrpcvHhRb19bUq9evTTehyuiq0axm5sbiouLkZ2drXe/nJwc9OnTx9QwJWUzieKrV69iwIABcHd3R0BAAF566aUK65rk5ORg8eLF6NSpE7y9veHv748BAwYYNASciIiMpx5RXI0nslNRT2jHEcVERFYlCAJ++OEH9QR0AwcORM2aNU1u75tvvjFqdJElDRo0SF3aQJeSkhL07NlT7/pt27Zh5cqVRh930aJFAIDu3bsjIiICACocLWWqBw8elPscjbF3717Mnz+/3G2mTp2Kn3/+Wf04NTUV/fv3N6kupqqfIiIiTO4nKUZxW5OHh4fRI8bj4uJw/vx5bNiwAc8++2y1vGsteM0aBFdwoYi0yWQyeHt7Iysry+y2vv76a7z66qt61w8dOrTCi3lA6cSk5pRDKs+jjz6KkSNH4vTp07h16xbGjRtnkeOYQxAENG3aFLm5ueVuJ9XdPMbWKdaVKAaATz/9FCdPntS7X61atbBx40ZTQpScTSSKs7Ky0KtXL+Tm5mLHjh1YsWIFtmzZgunTp5e73507d7Bu3Tr06dMH0dHR2LhxIxQKBbp27cpkMRGRyARB0Cg9Ud2p+oAjiomIrOvq1asIDg5WPx43bhy6du1qcnsLFy6Eg4Nt1N0vLi4ut05xt27dMHfuXL3r7ezs1PUrjREbG4uPP/4YMTEx2LBhA6KiohAbG2t0O4ZYv349Tp06JUpbN2/eVF8w0Kdly5bYs2cP4uLiEBcXB39/f0RFRZl0vNjYWERFRWHDhg1m9ZO1R3Fbk6enZ4UJorIKCwvRo0cPDB8+HEFBQZgwYYIFo6OqaPny5Wbf3q9UKrFy5UpMmTJF7zYymQyLFi1CYmJiuW1ZajI7lcjISHTv3h29e/fGlStXLHYcU8lkMnTt2hXHjx8vd7vExEQEBQVZKar/1KpVy6hEcWxsrM67HNq1a1du3nHfvn345JNPTAlRcjaRKF63bh0yMzPx008/YcCAAZg4cSI++eQTbNmypdzbVkJDQ3Hjxg2888476N+/P4YMGYLdu3fD19fXoCs9RERkuJKMEiiyFQA4ohgAXBqUfpHjiGIiIut6uF7g4cOHERkZaVJbRUVFsLOzqzDZaC1169bFnTt39K6fP38+fv/9d73rjZ3MTsXV1RVFRUVYvXo1ZDIZIiMj4epqmc96U2PUJSEhQeOigT4nT55Ex44dER4ejjZt2pg8+tDV1RWRkZHqWsLm9JMYo5Nt0auvvor27dsbvH1MTAzy8/Px4MEDNGnSxObqrVpL+uefI/3zz6UOo1Lq06eP2aUn9u7di5CQEHVJI30SExMrTIDWrl0bdevWNSue8tSqVQtyuRxKpdImE8VAaXmHI0eOlLvNpEmTyr3waSmvvfaawXWtMzIysHfvXp13ObRt2xZxcXF697106VKlrTdvE4nivXv3ok+fPvDz81MvGzVqFJydnbF37169+7m7u2vVt3JxcUHTpk0rvMpDRETGKTtyliOK/+uDwvhCCEpxZm8nIqKKxcbGolOnTurHvr6+uHHjhkltpaeno2nTpmKFZrbRo0eXm/j87bff4OnpqXd9w4YNMWjQIJOOre/2WrGJmSju1q0bwsPDK9zu3r17KC4uRn5+PgoKCmyiBq5Yo5NtjZubG+RyebnblJ3Qb9q0abCzK01LTJs2rUpM6GeK/FOnkX/qdMUbkpYVK1Zg69atZrXRp08fgyagM2RCu/Xr14teUqHsa+aLL75QL7fV18yIESPKHZ0NAL/++qtRdx+Ixc/Pz+C+unTpEr7//nud67p161bu311CQoJNTJRrCpu4x+rSpUuYOnWqxjJnZ2c0aNDA6EL4+fn5OHPmDCZOnFjudjk5ORq1k1QTBigUCigUCqOOqYtCoYBSqRSlLTIM+1wa7Hfrk6rP86/mAwDsXO1gH2BfrX7nuvrcuZ4zAEBZqETB3QI4BztLFV6VxPcW62Of/4d9YDrV6BpDJrIxxYMHD5CVlYU6deqolwUHB5s8mV1QUBAOHz4sVnhmGzBgAIqLi/Wuv3nzZrmT2T3yyCOYPHmySce+deuWVW77f+6550SbBb5bt25wd3fXuS43N1edVC87ICkhIUE9Oiw3Nxc1atQQJRZjqUYnq5g6Kt7WbN26Fa6urpg9e7bebcpOJJWWlqb+uexI76SkpEpdq9lYdTeslzqESis4OBh37941ej/V51VJSQmuX7+OJ598ssJ9evTogfj4eL3r5XI5Ro8ejd27dxsdT3nKvmbK1ngv+/qxpdeMv78/Ll68iOzsbL13CXz88cdYtmyZxue5NRw4cAD//POPQfX8y7uA6uDggI8//hjPPPMMfH19tdY/+uijaNWqlbnhSsImEsWZmZk6r+r6+PggIyPDqLbefPNNPHjwADNnzix3u1WrVmHJkiVay9PT0+HsbP6XfaVSqZ4BUXWFlCyLfS4N9rv1SdXn6f+UnpQ4hjhqnJRUB7r6XOH9XyIp+a9kuDmLM4M7leJ7i/Wxz/8j1kRb1U1hYaF6dGdaWppFaq26ubnh2rVrGreB+vn54aWXXjKpvd9++w03btzAM888I1aIZjl27BjWrl2L7du3a60TBAFTp07VuqPy4W0eeeQRk0bsLliwAC1atDB6P2OlpKQgNjYWY8eONbut9u3b4/jx4/D399dapy8ZmZ+fr/7ZlhIrVYWHh0eF3+FVE/rt2rVLazKu6OhoDBs2rNLXaibrqVOnTrkleXQp+3nVu3dvPP300wbtV69ePSxZsgSCIOgsR5CYmKgehCimyvia+fLLLzF8+HCMHDlS5/qkpCRJRtzWqlULBw8eNGjb7OxsNGrUSO/6kydPomnTpujXr5/WukmTJpkco9RsIlEMQOeLTN+LT59vv/0WH330EdauXYuGDRuWu+2cOXMwbdo09eOkpCR07NgRNWvW1HmiYSzVSBQ/Pz/RrphT+djn0mC/W59UfZ6RXHrSX6NRDVHeJysTnX3uD9z0uYmSzBI4pztXuz6xNL63WB/7/D8V3TZNuqlqjQKlI3aGDRsm+jF++ukn1KpVS6P0hEwmw+jRo1FSUmL0pHR///23TV38rFOnjt7J7JRKJRYvXlzu/jKZDA4ODigqKtI5WZo+giBAEAR4eHgYE65JMjIysH79erMTxYIgIC0tTWO0cFmVMbFSFXh6epY7IaNKVZ7QzxQZ32wBAPg+zcn8jPXYY4+hfv36Ru1T9vPqjz/+wA8//GDwvi+88AKGDx+uMzloyYnsKttrRlWnWF+iOC8vD4GBgVaOCggMDDR4Mrvnn3++3PWqCe10/S20bt0av/76a7nlomyVTSSKfXx8kJmZqbU8KyvL4JphBw8exJQpUzBv3rwKf5lA6QeYrl+Yvb29aF+O7OzsRG2PKsY+lwb73fqk6PPC+EIAgGuYa7X8Xevqc9eGrsj9Ixfym/Jq2SeWxvcW62Ofl6ruz98YD9/erxqNHhERoU7Ainl7/6ZNmzBr1iyt5SNGjEBUVBSaN29uVHuJiYlWqctrqODgYNy7d0/nupiYGGzfvh0bN24st42AgACkpqYalazIyMjAlClTcPXqVaPiNYW/vz9SU1PNbicnJwceHh7lDiyqbImVqmDkyJEGXyRSTejXuXNn1K1bF9u2bcPChQsxfPhwS4Zok3L37wfARLEpvLy88Mcff6BZs2blbqfr80oQBOTl5akvMhryedW4cWOcOHFCZ3LQy8sLo0aNMvGZVKzsJJgA8MUXX9jsa6Znz55Yu3at3vWmzi1grmbNmuHrr782aNvFixdjxowZekc+d+nSBUePHtVaLggC7t27Z5WLr5ZgE/cVNm3aVKsWsVwux40bNwxKFJ8+fRojR47EmDFj8P7771sqTCKiaq3geulkdq4NOJGdiqovVH1DRFTdPHx7v1KphFKp1KqbKAZBEPDnn3+iffv2WutMrVPs5+dnU5PZOTs7Y82aNRAE7UlS4+PjUbdu3QrbWL58udGJeWtNZAeUJoqNHfmti5OTk0GTT5VNrKiSKwsXLjT7+KRbbm4uTp82bFK22NhY/O9//8O1a9ewcuVKfPbZZ1ViQj9ThHyzGSHfbJY6jEpJJpNh/PjxOt83y9L1eSUIgsaEaoZ8XpU3oV3Lli0rnC/LHJVpEsy6devit99+07nu/v37+OCDD6wcUSknJyfs//fCTEW++eYbuLrq/+7bvXt39WdMWdnZ2fD29jaqQoItsYlE8aBBg3D48GGNenA7d+6EXC6vcNbeS5cuYdCgQXjsscewcePGSvuLICKyZSV5JSi+Xzq5jmtDJopVVH1RcIOJYiKqnlS390dHR2uti46ORkFBgWg1YHNzc9GlSxeddwWWNxK3PHPmzEG3bt3ECE80rVu31qijqxIfH48GDRoYtH9FCZOHWTNR7OTkhL/++svsdvLz8w3qj8qUWKkK7ty5g08//dSgbV1dXTF69GjUrFkTDg4OmDFjRrlJGSJdZDIZ/Pz8KiwjJNbn1aOPPoqpU6fqXPfqq6/iu+++MyxwE6gmwZTJZJDJZIiMjLTp18zPP/+ssx5wfHw8fv31VwkiKvX666+jpKSk3G0UCoU64VueJ598UqtCgp2dHZYuXWpumJKxiURxZGQkvL29MXz4cMTExGDz5s2YNWsWJkyYoHGFPyIiQuPqc0pKCvr37w9HR0fMmzcPf/31F2JjYxEbG4szZ85I8VSIiKqkwhuF6p85ovg/Lg1Kb1stuF5g9JdyIqKqwlq393t6emLbtm0610VGRqJv375Gtzly5EidJfCktGDBAp2JzNGjR6Nnz54V7v/hhx9i9+7dRh1z0KBBWL58uVH7mGPevHnqyTNN9fPPPxuUkKxsiZXKzsPDAzk5OQZvf//+fZ13CVQ3WTt2IGvHDqnDqLTq1KmDu3fvVridGJ9XDg4O6NOnj87E9J07d1CzZk2D2qkO5HK5zs/txMRESSayU/Hz86uwBFJubi6GDx9e4WDUoqIirfyjnZ0dBg8ebHacUrGJRLG3tzeOHDkCd3d3jBw5EnPmzMFTTz2ldSuRQqFQT7QCABcvXsTdu3eRnJyMPn36oEuXLup/I0aMsPbTICKqslQjZmUOMjjXdZY4GtuhGlGsyFGgOL1Y4miIiKRT9vZ+1eSeYt/e/+677+q9XdTX1xcZGRlGt3nixAmbm2imbt26uHPnjtZyb29vg+oOBwQEICUlxahjHj9+XOP2a0v77bffNG4DN0VCQgKCg4NFiojE4uHhgcLCwoo3/Ffz5s0NrhdalWXt+AFZOwyfUI00ff/992jVqpVB26o+r3x9fdWTahr7ebV+/Xqdf7eWnMyuMnr88cd11vC1t7dH69atrR/Qv4KDgzUqGuji7e2Nzz//vMK22rZti7///ltj2Y4dOziiWAyNGjVSzzyZmpqKTz75ROtK76ZNmzRGbD3++OPqGXof/nfr1i0rPwMioqpLVYPXpZ4L7Bxs5qNDcmVHV7NOMRFVZ2Vv709OTrbI7f379+9HUFCQznVXr17FG2+8YVR7BQUFcHJysrnJC0NCQnD79m2NZYIgoHv37gbtHxAQYPCM7ipr1qxBYmKiUfuYw5Rk9sOYkLFNdevWxalTpwzefsuWLThw4IAFI6ocQjZtRMim8ieqJP2uXLli8N+d6vPql19+wdatW036vNJXp/j5559HSEiIUW1VZYGBgejduzcePHigsfyJJ57A888/L1FUwL59+9CiRYtyt9m5cydWrVpVYVsDBw5EaGioxjKpR0ybi9/2iYioQuqJ7FifWINTLSfYuZV+lDJRTETVWdnb++/cuYOLFy+Kent/SUkJLl++jObNm+tcb0qN4qKiIsycOVOM8ET11FNPaX2BTklJQUBAAOzsKv76NmzYMKOflzVrFANA165dzZ7QbtGiRRgwYIBIEZFYBEHAm2++afD2hw8fRkEBz6FkTk6Q6SiJQIb5559/sGvXLoO2dXV1Rbdu3VBQUAA7OzuTytF06tQJp06d0hjIKAgCunXrBnd3d6PaqurWrVun1b9LlizBH3/8IVFEpaWL9E1IqHLhwgWDPnPbtWuHoUOHaiyr7BcymSgmIqIKqUpPMFGsSSaTqfukbB1nIqLqrHbt2ti2bZvWCCJzPHjwAG+++abe5GKtWrWMLmXg6emJ+fPnixGeqJydnbW+QMfHx6N+/foG7e/k5IT4+HijjvngwQPUqlXLqH3MMW/ePHTp0sWsNo4fP66z1ihJSyaTISoqyuDtjfnbrsqyf96D7J/3SB1GpRUcHGxQjWKVLVu24PTp0yYfz93dHb///rvGsvT0dPTp08fkNquqY8eOISIiQmPZyZMnJb2b5+LFi1q/v4cZcwG1adOmGrXZn3/+efTq1cucECXFRDEREVVIXXqigXiTElUVqvITHFFMRFTK0dER3bt3x5EjR0RrUyaTlTtK1sHBwejJrLdv347Zs2ebG5roiouL8fLLL2sse/TRR7FmzRqD9s/IyMArr7xi1DFv3Lhh0MgpsWzfvh2fffaZWW3MmjXLqjGT4ezt7VFSUmLQtk2aNNG6bbs6yvz2W2R++63UYVRaISEhkMvlBm9//PhxdOvWzaxjpqamaiQbExISKnW5AUt59NFHcfDgQY3R14mJiXpLSVlDYGBghSWawsPDDa573bx5c8TFxakfZ2ZmwsvLy5wQJcVPViIiKpdSroT8bumJF0cUa1P1iWrUNRERATNmzBD19tsXX3wRP/74Y7nbnDx5ssJZzMtKTEy0ydnpvb29kZWVBaVSqV524cIFjS/Z5fH39zeqH65du4ZPP/3U6DjNoVAocOnSJZP3LyoqAgCOKLZRbdq0MXhCu6ioKNSoUcPCEdm+OlH/Q52o/0kdRqXVtGlTfP/99wZtq1AokJ6eXmGN2orcunULGzZsUD+u7OUGLMXHxweBgYG4evWqelnfvn0REBAgWUz169ev8H3n6aefNvgiVrt27fDXX3+pHz/55JNQKBRmxSglJoqJiKhcBTcLgH+/m5advI1KcUQxEZG23r17o3Xr1gYnNyty+vRpdOzYsdxtvv32W1y8eNHgNm11shmZTIagoCCNyd7WrVuHCxcuGLS/u7s78vPzDe77f/75B//8849JsZrK39/frMnsUlJSEBwcLGJEJKb9+/cblPy9du0annrqKStEZPvsPT1h7+kpdRiV2nPPPadxgU0fe3t7nD9/3uzSB4899phGnduOHTsaPalqdfHBBx/As8zf9+rVqyUtPdG9e3csXrxY73qFQmHUZ8yECRPQr18/AIBSqURhYSHc3NzMDVMyTBQTEVG51LV3ZYBLKEtPPEw1org4tRglOYbdZklEVB0MGzYMZ8+eNbudnJwc5ObmVvilrXbt2kZNaDd06FD06NHD3PAsIi4uTqNmsLF1XG/evGnwtrdv37bqRHYAEBYWhscff9zk/YODg3Hq1CnxAiJRvf7667h161aF212/fl3SZJEtyT18GLmHD0sdRqV2/Phxgy5Aff311zh06JDZx6tVqxaCgoKQm5sLALh//z48PDzMbrcq6tatG9LS0gCUfj717dtX0niysrIwdepUvesTExONqtvfoEED9d9BamqqpKOlxcBEMRERlUs1UtY52Bn2LjyZf1jZus0sP0FE9J+BAwdizx7zJ2dydHTE9u3bK9yuXr16yMrKMrjdevXqISQkxIzILGffvn0aEy3dvXvXqDquO3fuNLj8hKurq8F1GMVSt25dPPvssybvf/r0aezevVvEiEhM//zzT4X1PwFOZFdW+hdfIv2LL6UOo1IzdEK76OhouLiIM/jl119/VSeHV69ezQtYehQWFqJ///5QKpVISEiQfLStu7s79u3bp3e9MRPZqYwaNQr5+fnw8/PD0aNHzYxQWkwUExFRuVTJT9Yn1s2ljgtkjjIATBQTEZU1ZMgQ/Pzzz2a3c/nyZdStW7fC7ebOnYsXXnjB4Hb79u1rVGLZms6dO4dff/1V/fjy5ctGfbHevn07bty4YdC2zz33HAYPHmx0jOZQKBRmjWI+efIkzp07J15AJCoPDw/16LryNG/eHAMHDrRCRLav9kerUfuj1VKHUan16tWrwpI7CoUCf/75Jzp06CDKMU+ePImVK1cCKK1RzJI4unl4eCA0NBTnz59HUlKS5GWfHB0doVQq9dYRrl+/PubNm2dwezKZDK1atcLZs2dx7do1nD9/XqxQJcFEMRERlUs1opj1iXWT2cvUJTlYp5iI6D8tW7ZEVFSU2e0sXrxYYxIcfW7evImPP/7Y4HbT0tJscjI7AAgJCcGdO3cAlN4C++WXxo00DAwMNLgG8NixY5GZmWl0jOawt7dHQUGByZP9cNIo2xYeHg4vL68Kt+vQoQM6d+5shYhsn2NAABwr+e3qUnv11VcrrGWfnZ2NKVOmiDai2M/PD9HR0QD4vlSR3r1748iRI2jatCmefPJJqcPBM888g4IC3d/dlEql0ZMdtmvXDn///TeOHTsmSmkTKTFRTERE5eKI4oqp+kZdz5mIiCCTyWBvb4/Y2FiT2xAEAX/++SfatWtX4bYKhQLffvutQe3m5eWhRo0akMlkJsdmSQ0bNoSDgwMA4NKlSzh58qRR+4eGhur9AlyWIAg4evSoQUk9sdWsWRMZGRkm7evi4oKGDRuKHBGJZebMmQaN2OzSpQsSExOtEJHtyztxEnknjHudk6a9e/fi/fffL3cbb29vvPfee6Ids1GjRrh37x7y8/Pxyy+/ICgoSLS2q5q5c+fiscceQ0ZGBrp06SJ1OFixYoXeSTcXL16MX375xaj2XnnlFUybNs1mJ8o1BhPFRESkl6AQUHizNPlZthYvaVKNtuaIYiIiTXfu3FHflmuKhIQE+Pn56f0yV1bt2rWRkJBgULtOTk7YsWOHyXFZWseOHbFq1SoApZPNGVOfGADeeOMNg0ZsZWZmwtvbG3Z21v9auGzZMjg7O5u079tvv41u3bqJHBGJZe3atfjqq6/K3UYQBCQmJjKx9q+0zz5D2mefSR1Gpebk5IQzZ86Uu83EiRNx7Ngx0Y4pk8kwd+5c/PLLL1i9erUk76WVhbOzM8LDw9GrVy/s3btX6nAwa9YsvbWETalR7Orqii+++AK5ubmoU6eOCBFKh3/FRESkV+HdQgjFpbW+OKJYP1XfsEYxEZGmnj174sSJEygqKjJp/+DgYIMnB3J1dUXr1q0rrFEJAOnp6bC3t90JWgVBwJgxYwCU3gLbsmVLo/Y/ffo0vvjiiwq3Ky4uNmtSOXP07NnT5KTKhAkTUFjIu3hsVWFhYYUjhVNSUuDv78/E2r8eef89PPK+eCNdq6OKJrMTBAHHjx8XffLOF154AWPGjMG7777L96VyxMTEoKioCAqFArdu3ZI6HNjb2+u9uHzr1i2jJ7t1dHTEokWLMG7cOAwfPlyMECXDd2UiItKr7AhZ1ijWT9U38ntyKApNq7dIRFQVubq6onfv3rhy5YpJ+2/ZssXgSdkA4OeffzaonMTRo0eNrvtrTTKZDH/88Qdyc3Mxbdo0jBo1yqj9c3JyDKqRGBAQgLlz55oaplneeuutcmed10cQBOzfv9/k0chkeYZMZufu7o7//e9/VorI9jnVqQOnSj4KUWp169Ytd2LO27dvw9vbG97e3qIe93//+x8KCgqgVCpx4MABUduu7HJzcyGTySCTyTBt2jT15/Pbb7+tXp6XlydJbIGBgUhOTta5bteuXUb/ncjlcmRlZeGxxx4zeI4AW8VEMRER6aWquesY4AgHDweJo7Fd6tHWAtSlOoiIqNQ333xj9IhYlc8++8ygEcIqr732GuLi4ircrjLUEKxbty7u3LmDl19+GdnZ2UbtGxAQYNAX1Y8++ggrVqwwNUSzGBrjw9LS0uDv72+z9aUJaN++Pbp27VruNllZWQgLC7NSRLbvwV9/4cFff0kdRqXm5uaG+fPn6/3McHd3V5f0MVfZBOi7776rXh4RESF5AtSWlE3EpqWlqX836enp6uVJSUlWjwsA+vbtq3Pyw6ysLIPLWJUVExMDhUIBhUKBEydOiBGiZJgoJiIivfLOlp7gsOxE+VzquQD/3sGs6jMiIipVUFCAESNGGJXwBUrLIly/fh1NmzY1eJ/MzEyDRiBXhtqoI0aMgEwmw4EDB+Dh4WHUvkFBQQZN9nbr1i3J+sHf3x+pqalG71dcXGxQ/WWSTtu2bTFo0KByt/nyyy+xfft2K0Vk+1JWrkLKSnGSmNVZhw4d9Cb5UlJSKryAYaiHE6C6fpYqAWpLwsLCUFBQgOjoaK110dHRKCgokOyCUdu2bdGmTRut5efOnTP4boeHR0yrSulERkZW6gsGTBQTEZFOgiAgY1/pbOTej3tLG4yNs3O2g1fX0hnjM/abNoM7EVFV5erqilu3buHq1asG7xMXF4fDhw9j2LBhRtUSDg4Oxr179yrc7p133sHYsWMNblcKs2fPhp+fH2rWrGl0HVd/f3+sW7euwu1MmbBHLCNGjEC7du0MGgFe1iOPPILFixdbJCYSx/HjxzF69Ohyt4mPj0eDBg2sFJHtq7V4EWotXiR1GJVerVq19NYpHjNmjMZIVnPYcgLU1ri4uMDJyUlrubOzM1xcpJss/cyZMzrLOhnzufjwBQOlUqn+WaUyXjBgopiIiHR6cOUBCm+VllGoOaimxNHYPt9BvgCAjH0ZEJTGjZojIqrqhgwZgj179hi0bWFhIcLDwzF69GisWbPGqON069YN9evXr3C76OhoyOVyo9q2tujoaLzwwgvo3r27Sfv36dMHCkX5dfOnTp2KZs2amdS+uezt7TFu3DiEh4cbNQHUV199hQ8//NCCkZG5atSoUWGN4vj4eINeq9WFS6NGcGnUSOowKr06deroTBSnpaWhsLAQwcHBoh3LVhOgtmjRotKLIBEREYiIiAAALFy4UMqQEBgYiPv372stv3v3LkJDQw1qo6peMGCimIiIdFKNJnbwdoBHJ+Nuea2Oag4sTaYXpxQj70zlu8WIiMiSnnzySYNH6MTExCA/Px/5+flGJ4p79uyJoUOHVrjd22+/bXQpDGtzdXXF7t278d133xmVSFVJSEhARob+u1wEQUDLli3h4+NjTpgm2759OwoKCpCfn2/UBFDx8fFwc3OzYGRkLkMms/v222/RiIlRtYLzF1Bw/oLUYVR6U6dORevWrbWWnzx5EuHh4aIfzxYToLYoNjYWUVFR2LBhAzZs2ICoqCjExsZKGpO+Ovmvv/46nn/+eYPbqYoXDJgoJiIinTL2ln659OnvAzsHflxUxP1Rdzg9UnqSkL5XnNvaiIiqiubNm6N3794oKCjQuV5fnb/33nvPqDp/V69eRd++fcvdRhAEZGVlGT2jubXduXMHJSUlePDgAQ4ePGj0/hVNFpeVlYV+/fqZE6LRyv6eVeUj7OzsjJoA6t69e6hdu7YVoiVTBQUF4bXXXtO7vrCwEMeOHYOjo6MVo7Jt9999F/fLTIpGpmnevLnOC0mDBw/GRx99JPrxbDEBaotcXV016vZGRkbC1VXaOXCcnJywYcMGreWffvopsrKyjGqrql0w4Dd/IiLSUpJXgqxfswD8N1KWyieTyeA78N/yE3tZp5iI6GGzZ8/G7t27da7TV+ev7IhYQ+r8BQYGIj4+vtxt8vPzUb9+fchkMkPCtqqyidQlS5YAKP18mT59utET44wfP77ckbdS1Ccu+3tW1QpVKpVG1XMcNGiQzgmIyHa4ubmhZcuWetffvHnT4MmiqovA1+cj8PX5UodR6Z0+fRovv/yy1vKoqCiLjO60xQQoGS4kJETrAvbq1avh4OBgVDtV7YIBE8VERKQl60gWhKLSW3J9B/hKHE3loarlnHMqB0VpRRJHQ0RkWwYPHoyff/5Z5zpVnb+tW7dqrTOmzp+npyeys7PLLStRo0YNnD592vDArejhhDlQOgLalIlxIiMjy00ES5EoFqOeY+/evVG3bl1LhUgi6dKli951rE+szbVlS7iWk1wnw+iqUZyfn493332XCVzS8uKLL+LatWvqxyUlJcjLyzP6jqOqdsGAiWIiItKiKp3g0d4DToHaNZdIN58+PpA5yAAByDyQKXU4REQ2pW/fvjh06JDeCdYKCgrw/vvvay03ps6fTCbDihUrUFJSonebuLg4rF+/3rCgrUzMiXHee+89rF27Vu/67t27Y8GCBSbHaipz6zk2atRIPeKcbJNqtL6+CzZJSUmVcoInS5LfuAH5jRtSh1HpBQcH4969exrLTp8+jY4dO9rkXSQkrYcntMvMzMRjjz0mYUS2gYliIiLSIAiCeiI7VSkFMoyDpwO8wr0A/DcZIBERlfL09MSRI0dw9uxZxMXFaay7du0aunTpov7CZk6dv6FDh6K4uFjv+nPnzuHSpUvGBW9FYk2M4+XlVW6N4hs3bkg2KZyqnmPPnj3VE5oZ8nt+8OABXFxcYG9vb9H4yHyurq56a5JPmzYNb775ppUjsm1Jby1E0luVt6aprXBxcdEqcZSQkID+/ftLFBHZsocTxf7+/ti5c6eEEdkGJoqJiEjDg4sPIL8jBwD4DmKi2FiqPsvYnwFBof/WZyKi6kgmk+Gxxx5DeHg4CgsLAQA5OTmoWbMmPvvsM9y4ccPsOn9z5szBkSNH9K5PTEzEI488YtbzsDRVInXq1KkYP348AOMT5hVNZvfOO+/g+vXrpgdpBlU9x61bt6K4uNjg33NCQgInsqskduzYoXeyulWrVmmN+qzu/F9+Cf4vvyR1GFVCTk6OxmRkTz/9NJ577jnpAiKbtWDBAgwdOlT9eM+ePdi4caOEEdkGJoqJiEiDquyEQ00HeHbwlDiaykc1Crs4rRi5f+ZKHA0RVTYHDx7E+PHj0aBBA8hkMsycOVPqkER19OhRFBYWIj8/H/v378fSpUsxduxY+Pr6olevXqLU+dN163FZ3t7eaNGihblPxaJUidR169ZhxYoV+Oyzz4xOmHfo0AGjR4/Wu16KGsUqqt9zYGAgmjdvjrFjxxr0ew4ICMDKlSutECGZq6SkRH0x6GFffvml0ZNFVXXuHTvCvWNHqcOoEj766COcP38eQOnf4bBhw8qtW0/V14MHD3DhwgX14z///BO5ufz+xndnIiLSoC470d8XMnvW8jKWe3N3ONdxhvyuHOn70uHZicl2IjLcvn37EBcXhx49eiAjo2qUsMnNzYWnZ+l7oZ+fH4DSkcXjxo1DUVHpxJ95eXmoUaOGKMcLDg5GQkKC3vWRkZGiHMeSVIlUhUIBmUyGGTNmGF1uoXbt2rCz0z8uKDExEUFBQeaGaraHbxMvT15eHurUqWPBaEgsixYtwnvvvYe2bdtqLBcEAffv30etWrUkisw2Fd0rfc9yCuaIeXMFBwerJ7SLi4tDfn4+6xOTThcuXMD27dvRtWtXAKUXUNu0aSNxVNLjiGIiIlIrySlB9vFsAEDNQTUljqZykslk/5Wf2Fs1kjxEZD0rVqzAxYsX8eWXX8LLy0vqcESRnJys/jktLQ1AabJIlSQGSie3EssTTzyByZMn610/ceLEanHbe05ODnr27Kl3/e3bt22i1u+ZM2cwd+5cg7b95ptvsHXrVgtHRGLw8PDQOTIvNzcXvXv3ZuLuIYmvvorEV1+VOowqoU6dOupE8fHjxxEeHi5xRGSratWqpVGjuFmzZmjWrJmEEdkGJoqJiEgt83AmhBIBkAE+/X2kDqfSqjmwNMme+2cuilKKKtiaiOg/5Y0ArazCwsJQUFCA6OhorXXR0dEoKChAWFiYaMfz8vLSSE4/7Pjx41UmCV8eb29vZGZm6lx3+/Zt/PTTT1aOSLfQ0FBs27bNoFvDWaO48tCXKPb09MR3330nQUS2ze+5Z+H33LNSh1ElvPzyy+qLTyEhIRg+fLjEEZGtCgwMVF/ABoB58+ahYcOGEkZkG1h6goiI1FQjYD06esDJT3vGdTKMd29vyBxlEIoFZMRkoNYzvL2SiCwrJycHOTk56seqEboKhQIKhcLs9hUKBZRKpcltOTo66qxJ6ujoCEdHR1FiVMnOzsb06dNx7tw5rXWCICA3Nxdubm6iHtNSzO13Z2dn5Ofnw8XFRWP5X3/9hUOHDmHs2LFihGkWDw8P1KpVCxcvXkSTJk3K3fbu3bsYPny4RX935vY5lVq1ahVcXFy0+nHXrl1ISEjQmFyMfQ64/nvru7X6oCr3eWpqKg4ePIhnnnkGffv2tZn3+6rc57aqoj4PCgrC77//DoVCgZKSEnTs2BF///23laO0DmP+7pgoJiIiAKVfntP3lU5kpxoRS6ZxqOEAr+5eyDqchYy9TBQTkeWtWrUKS5Ys0Vqenp4OZ2dns9tXKpXIzi4tTWTqqOe33noLADB+/HgAwLfffos333wTXbp0MTu+slQ1UFNTU7XWyeVyTJo0Sec6W2Ruv69fvx4ZGRlwdHTUWH7+/Hn4+fnZTD/MmDEDmZmZFcazePFieHp6WjRuMf7WqXRSKGdnZ7Rs2VJj+W+//QZvb2+N3yH7HFD+O/rfzsc6d/RV5T5PSkrCRx99hLCwMLz00kvYs2eP1CEBqNp9bqsM6fOVK1di9uzZSExMhCAINvO5KLb09HSDt2WimIiIAAD55/JRlFBaJkFVY5dMV3NQzdJEcUwGBIXAiQGJqqns7GyD6u+GhoaaldCdM2cOpk2bpn6clJSEjh07ombNmvD39ze5XRXVSBQ/Pz+T69qeOnUKmzdvxvTp0wEA4eHheOaZZ+Dq6mp2fA/z9PSEs7OzehI9FUEQ8MEHH4h+PEsxt98DAwPh7OyMmjU1LwA7ODigbdu2ovxtiGHq1KlIT0/XivNh+/btwzPPPGPR+rZi/K0T8M8//8DBwQG9evXSWJ6cnIxu3bpp/O2xz4E7s+cAAOp+s9kqx6vKfe7l5YWEhATExMQgPDzcZt7nqnKf2ypD+nz37t2YO3cucnNz0bBhQ5v5exGbXC43eFsmiomICACQvrf0KqOjvyM82nlIHE3l5zvIFzfm3kBJZglyTuXAq2vVr4dJRNp27tyJKVOmVLjdmTNn0Lp1a5OP4+npqZUUBQB7e3vRvpDa2dmZ1V6NGjU0bjcv+7PY/vzzT3h5eWmNIIqJiUF0dDS+/PJLix1bbOb0+8cff4zx48dj4MCBGssXLFggVniiePDgAVq3bo179+7pHfVVUlKC1157zaDXk7nM/Vun0mRdSkqKVh926tQJrVu31lpe3fu85pTJAGDV519V+1wmkyEzMxPvvfceNm/ebFPPr6r2uS2rqM9VdYqDg4Mxffr0Kvu7MeZ5cbw7EREBADL2ldYn9h3gC5kdR7+ay62xG1zqldaEVPUtEVU/kydPhiAIFf4zJ0lM2v744w9cv35da3liYmKFo1arkoCAAJ230c6cOVN9O64tcHNzQ/369XXWlVa5f/8+atViKafKwsfHR2dNzBdeeAF169aVICLb5tmvHzz79ZM6jCohJiYGSqUSJSUlBk2SSdVb69atUVBQgICAAPTp00fqcGwCE8VERITirGJknyz9wsiyE+KQyWTqvlSN1iYiIus4dOgQTp48qbU8MTERjzzyiAQRSSMgIAApKSlay6Ojo1GjRg0JItKvV69eOHLkiN719+/fZ4KxEhk/fjzefvttjWUPHjxAo0aNJIrItiny8qHIy5c6jEorNzcXMpkMMpkM06ZNU//84osvqn/Oy8uTOkyyQR9//DE6dOiAl19+Gfv27ZM6HJvA0hNERITMg5mAAoAd4NuPiWKx+A70ReJnicj7Ow/yZDmca5k/oRQRVW23b9/GH3/8AaA0qXLjxg18//33AIDRo0dLGVqlEhwcjISEBK3lffr0gYdH9Smv9Pzzz2uVcsjKyoKnp6fN3V773HPPlTv6r23btti9e7cVIyJzXLt2DUeOHEFkZKR62c2bNxEUFCRhVLbr7owZAIB6326ROJLKKTk5Wf1zWlqazp+TkpIQFhZm1bjI9v3www8oKSnBrVu3UK9ePanDsQkcUUxEROrSCJ6dPeHo61jB1mQon54+kDmXlvHI2M/yE0RUsaNHj2LMmDEYM2YMUlNTsX//fvVjMlxISIjO0WPBwcHVKlGQlZWFo0ePaizLz8+3yb+noKAgnDhxAiUlJTrXHzp0CIcPH7ZyVGSqzMxM/PTTTxrL4uPj0aBBA4kism0+T46Dz5PjpA6j0goLC0NBQQGio6O11kVHR6OgoKBavfeT4TIyMvDXX3/h1q1bCAkJkTocm8BEMRFRNScoBXWiuOag6lO30Rrs3e3h/bg3ACBjLxPFRFSx8moak+FGjBiB9957T2v5yJEjcefOHQkikkZSUhI+++wzjWW1a9fW2Te2YP369fjrr790rjt8+DBu3rxp5YjIVB4eHsjNzdVY1qRJE0yfPl2iiGyb17Bh8Bo2TOowKjUXFxc4OTlpLXd2doaLi4sEEVFlEBgYiPv372PDhg3w8uLk4wATxURE1V7e2TwUJRcBKC2VQOKqObA0+Z5xIAPKEqXE0RARVQ85OTl4/fXXtZYnJiZWq1vfAwMDtWoUr1+/HuvWrZMoovKVV6f43r17qF27tpUjIlP5+Phojc5zdXVFmzZtJIrItgklJRD0jKYnwy1atAgAEBERgYiICADAwoULpQyJbFxYWBiCg4Or1US3FWGimIiomlONdHWq5YQarW1rYpuqQDWhnSJbgZzfcySOhoioenB1dcXGjRs1limVSsjlcri5uUkUlfUFBASgsLBQY9nFixdttg969uyJf/75R+c6T09P1o+sRGrVqoVvvvlGY9mzzz6Ls2fPShSRbbs9cRJuT5wkdRiVXmxsLKKiorBhwwZs2LABUVFRiI2NlTossmFNmjRBvXr18Morr0gdis3gZHZERNVc+r50AIDvAF/I7GQSR1P1uIW5waWBCwpvFCJjXwa8u3lLHRIRUZXn6OgIe3t7yOVyODuXTiQqCEK1mwzN1dUV58+f11h2+/ZtjBo1SqKIyte5c2d07txZ57q1a9daORoyhyAIGDlyJHbu3KleduPGDdSvX1/CqGyX1/DhUodQJbi6umpMoFj2ZyJdcnNzMX36dDg6OqKwsJBlSsARxURE1VpxRrF6lKtq5CuJT1X7OX1vusSREBFVH4MGDUJOzn93cmRmZsLe3l7CiKSxcOFCjVqxQ4YMQZMmTSSMSD+ZTIa1a9fi9OnTGssFQcCIESMkiopMIZPJcOLECfVjpVKJ1NRU+Pv7SxiV7fIZNxY+48ZKHQZRtaMqd1RcXIwDBw5IHI1tYKKYiKgayziQASgB2AM+fX2kDqfKUiXh88/mQ54glzgaIqLqYcOGDRpJqVOnTmHNmjUSRiSNgwcPIjk5Wf34qaeesulknVwux969ezWWZWdn49y5cxJFRKZycHBASZm6uwcPHoRMxrvXiEhaubm5kMlkkMlkmDZtGgDAzs4OERER6uV5eXkSRykdJoqJiKqxjH2l9Ym9unrB0dtR4miqLu8e3rBzKf3IzdifIXE0RETVw+rVqzVKTSQmJuKRRx6RMCJpBAQEIDU1FQCQlZWFli1bShxR+XRNaJeQkMCJ7Cqhp556CkVFpRMmJyYmaiSNSdPtSZNxe9JkqcMgqhbKXjxNS0sDUHrXg+pnAEhKSrJ6XLaCiWIiompKUArqRDHLTliWvas9vHt5A2D5CSIia5HL5bh8+bL6cXVNFLdv3x6CIAAorU9s6xPCtWrVCi1btlTHDJSOTB05cqSEUZEpVq1apZ448dChQ/juu+8kjsh2efTuBY/evaQOg6haCAsLQ0FBAaKjo7XWRUdHo6CgAGFhYRJEZhs4mR0RUTWV+3cuilOLAQA1B9aUOJqqz3egLzL2ZiDzYCaUxUrYOfJaLRGRJQUHB+PPP/9UP37ttdegVColjEgab731lvrnW7duISQkRMJoKmZnZ4e1a9dCoVCoa0o3btwYjRs3ljgyMtbYsWOxZMkSNG3aFPHx8ZzIrhy+EydKHQJRteLi4gInJyet5c7OztV+Qjt+SyUiqqYy9paOJnZ6xAnuj7pLHE3Vp5rQTpGrQPbJbImjISKq+tq0aYO2bduqH+/cubNa1hzcuXMnvvjiCwClffLiiy9KHFHFvvrqK7z55pvqxx988AG+/PJLCSMiUxQUFCArKwsAcPPmTSaKicimLFq0CAAQERGBiIgIAKUTwFZ3HFFMRFQNKfIVuL/lPoDSBCYnFrE81/qucG3sioIrBUhanwTvHt7sdyIiC2revDmaNWumfrxixQp06dJFwoikUVBQgLNnzwIoLccRGhoqcUQV69ixI9auXat+fOPGjUoRN2ny9PREbm4uAGDjxo0SR2Pb7kY+CwCosy5K4kiIqo/Y2Fh8/fXXmDFjBgCgQ4cOmMjR/RxRTERU3QiCgMsRl1FwtQCQAYGTAqUOqdoImhIEAEjZmoLE/yVKHA0RUdVWWFiIhg0bqh8nJSUhKChIwoikERAQgJSUFACl5TdUSWNb1qRJEyQmJqpHoyYkJCA4OFjaoMho48aNU5c6ee+999SlREiba7t2cG3XTuowiKoVV1dXREZGQiaTQSaTITIyEq6urlKHJTkmiomIqpm7K+4iNbp09vPQZaHwDveWNqBqpM4rdeA7sHTiwOsvXUfW8SxpAyIiqsJcXFyQm5sLhUIBpVIJHx8fODs7Sx2W1TVo0AAdO3YEUHr7v61PZgcAMpkMX331lTqx+NRTT6FRo0YSR0XGGjRoEEJDQ5Gbm4v169fzTqpy+M2YDr8Z06UOg4iIiWIiouok42AG4ufHAwD8Rvqh7oK6EkdUvcjsZWi6pSlcGrhAKBFwYfQFFN4rlDosIqIqq1atWrh//z7s7Oxw4cIFqcORRGhoKObMmQOgdGTuI488InFEhunYsSOuX78OABgwYABq1uTEu5XNu+++i6ioKNYnJiKqRJgoJiKqJgriC3Bx3EVACbg1c0OTTU04skMCjj6OaPFjC9i526E4pRgXRl6AolAhdVhERFXSsmXL4OrqiqtXr+Kjjz6SOhxJKBQKNGnSBIIg4Pfff4eDQ+WYpiYtLQ2TJ0+GXC5Hy5YtpQ6HTODh4YHc3FxkZWWpR7WTbvdeno17L8+WOgwiIiaKiYiqA0W+AuefOI+SzBLYe9mjxY8t4OBROb4oVkU1WtRA06+aAgBy/8jFteevQRAEiaMiIqp6unfvDplMhsuXL+PcuXNShyMJe3t7ZGRkIDExUT1CtzJQlSzYvn07vL29pQ6HTKCazK579+744IMPpA7HpjnXD4VzfU7YSETSY5aAiKiKU01el38uH5ABzbY2g1uYm9RhVXv+o/xRd0Fd3Hn3DpI3JsOjnQdqv1Bb6rCIiKqUTz75BD4+PnBycqo0JRcswc/PD/v27cPJkyfRv39/qcMxWLdu3TB16lQolUoUFhbCxcVF6pDICL169ULbtm3x0UcfoXPnzujcubPUIdks/xdflDoEIiIAHFFMRFTlaUxe93Yoag5kjT9bEbo09L/J7V6+jqxfs6QNiIioigkODkZCQgJq1KiBVq1aSR2OZBYsWICEhIRKMZFdWYMGDUJxcTEUCgUOHDggdThkJH9/f3h7eyMmJgZ2dkw9EBFVBny3JiKqwrQmr3udk9fZEq3J7cZcQOFdTm5HRCSW4OBg3Lt3D08//TRGjx4tdTiSeeKJJ5CVlYXQUNu/tT03NxcymQwymQwzZ84EANjZ2SEiIkK9PC8vT+IoyRBxcXGYPXs24uPjOZldBRJfPVp60AAAHd9JREFUm4/E1+ZLHQYRERPFRERVFSevqxy0JrcbxcntiIjE0rlzZ7z55pt4/vnnceXKFanDkUxERAQCAwPxzDPPSB1KhZKTk9U/p6WlAQCUSqX6ZwBISkqyelxkPNVkdt26dUPNmryjrTz2NWvCnn1ERDaAiWIioiqIk9dVLlqT2z3Hye2IiMTg4eGBO3fu4OTJk6hRo4bU4UiisLAQO3fuxOuvv66RbLVVYWFhKCgoQHR0tNa66OhoFBQUICwsTILIyFiqyew+//xzDlaoQOCr8xD46jypwyAisp1E8dWrVzFgwAC4u7sjICAAL730EgoKCgza96uvvkKTJk3g4uKCFi1aYPv27RaOlojINgmCgLx/8nDxyYucvK6SUU1uBwDJm5IRPz8e8gS5xFEREVVuMpkMTz75JO7evYvAwECpw5FETEwMiouLAQAnTpyQOBrDuLi4wMnJSWu5s7MzJ7SrRGrXro3+/ftjzJgxUodCREQGsonhZVlZWejVqxdCQkKwY8cOpKSkYM6cOUhPT8c333xT7r7ff/89Jk+ejPnz56Nfv3748ccfMW7cOHh5eaFfv35WegZERNLKv5SPlOgUpEan4sHlB+rlnLyucgldGoq8M3nI2JeBux/cxd0P78Ir3Av+Y/3hP9ofzrWcpQ6RiKhSkcvlyM7Ohr29PUpKSuDgYBNffywuNzcXnp6eAAA/Pz/IZDIIgoAZM2Zg5MiR6m1seZT1okWLAJSWzQCAL774AgsXLsTw4cOlDIuMoFAosHz5ctjZ2aGwsJBJ/nIkLV4MAAj6938iIqnYxJnSunXrkJmZibi4OPj5+QEAHBwcMGHCBLzxxhto2rSp3n3feustjBkzBsuXLwcA9OzZE5cvX8bChQuZKCaiKq3gWgHSvk9D6rbU0tHDZTg94oRHnnuEk9dVMjJ7GZptbYbrs68jdXsqFHkKZB/PRvbxbFx/6Tq8e3gjYFwA/Eb5wclPe6QVERFpiomJgSAIKCkpwYEDBzBs2DCpQ7IKXXV+H/45KSnJpks4xMbG4uuvv8aMGTMAAB06dMDEiRMljoqMceDAASgUCigUimr1+jOFzM5mbvYmomrOJhLFe/fuRZ8+fdRJYgAYNWoUpk6dir179+pNFN+8eROXL1/Gu+++q7F8/PjxmDJlCtLS0jTaJCKqTJRyJYozi1GSUYLijP/+L7xdiOTvkyE/r1mWwDHAEf6j/REwLgBe4V6Q2bEWXGXk4OWAJl82QdjaMGTszUBKdArSf06HskCJrKNZyDqahasvXIVPbx/49veFY4AjHH0d4eDroP7fwdsBdg78wkFE1ZO+0bQRERHqRKmtj6Y1l6rO765duzBu3DiNddHR0Rg2bJjNj+50dXVFZGSk+nHZn8l2Pfz6A0pLwFSn158pai1cKHUIREQAbCRRfOnSJUydOlVjmbOzMxo0aIBLly6Vux8ArURys2bNIAgCLl++jPDwcJ375uTkICcnR/1YNXOu6oqnuRQKBZRKZbltJXyagOxfss0+FpUSBAFFRUVIcUrhZAlWxH4HIGj/rJ6IrOw6JSAoBPU/1WMo/l2uFCAUCSjJLE0IK/OVFR7aoaYD/Eb6wW+0H7x7eEPmUPo7UApKwPy3MvqXIe/ponMCfJ/whe8TvlDkKZCxNwOp21KRsS8DglxA5oFMZB7I1Lu7vZc9HH0dYe9tD5m9TP0PdvjvZ/t/f7aTac5a8O9LWf2almmvszS+t1ifLfd5zaE1ETjJevVlrfpaJ9FVhdG0YmCdX5KCrtefIAjV7vVHRFRZ2USiODMzE97e3lrLfXx8kJGRUe5+ALT29fHxAYBy9121ahWWLFmitTw9PR3OzubXgFQqlcjOLk0C2+m5jST993Rk/8hEMREZyA6w97aHnbcd7FvZo+bomqjRrQZkjjKUoARpmbY/k3llZch7usX1BPx7+sM31xd5MXnI3Z0L+WU5FFkKKHO0LyooshVQZDPZRVWDUEuA3SDrvfbS09OtdiwSX1UYTSsW1vkla+PrzzT3l78HAAh8fb7EkRBRdWcTiWIAOkeuCIJg0IiWh7dRjeQrb985c+Zg2rRp6sdJSUno2LEjatasCX9/f0PD1ks1EsXPzw/29va6t+mjgIsbPyTFIgiCepIEWxsJVZVVi3434GlpPHd9oy9l0DmqUz2y004GmZMMDj4OpWUEfP4tJ1Cz9H97D3vI7GRQKBTq0jr63l9IXIa8p1uNP4Dn/v33L6FEQEnWfyVKSjJK1CPTS7JK/hu1/vAI9rIj21Wj33WMkFc/FB5aoIsBmxiiWry32Bhb7nOfHj6inJ8ZSi6XV7wR2TSOpi3FOr8kBb7+jKfIzZU6BCIiADaSKPbx8VGPDi4rKyur3InsVCOHMzMzERj43+2IWVlZGut18fT0VNdOKsve3l60JICdnV257dWeVhu1p9UW5VhUmshJTU2Fv7+/9ImcaoT9Lo2K3l9IfDbd5/aAQ6ADXAKrzpcvvrdYH/v8P9X9+VcVHE3LOr8kHb7+jPPIu+9IHQIREQAbSRQ3bdpUqxaxXC7HjRs3tGoXP7wfUFqruEmTJurlFy9ehEwm01hGRERERETVB0fTEkmHrz8iosrJJqZEHzRoEA4fPqxRD27nzp2Qy+UYNGiQ3v1CQ0PRpEkTREdHayzfunUrOnbsqJ5llYiIiIiIqhfVaFqZTAaZTIbIyEi4urpKHRZRtcDXn3FSPvoIKR99JHUYRES2MaI4MjISa9aswfDhw/HWW28hJSUFc+bMwYQJEzRKT0REROCrr75CSUmJetnSpUsxbtw4NGjQAH379sVPP/2EAwcOYP/+/VI8FSIiIiIiIiIigxXfuSN1CEREAGwkUezt7Y0jR45g1qxZGDlyJNzc3PDUU0/h/fff19hOoVCoJxRSGTNmDB48eIB3330XK1asQMOGDREdHY1+/fpZ8ykQERERERERERmt9qpVUodARATARhLFANCoUSPExMSUu82mTZuwadMmreWTJk3CpEmTLBQZERERERERERERUdVmEzWKiYiIiIiIiIiqo7SoKKRFRUkdBhGR7YwoJiIiIiIiIiKqbgr+OSd1CEREAJgoJiIiIiIiIiKSTJ3P1kodAhERAJaeICIiIiIiIiIiIqr2mCgmIiIiIiIiIpJI+sZNSN+4SeowiIiYKCYiIiIiIiIikkr+8V+Rf/xXqcMgImKNYiIiIiIiIiIiqdT98kupQyAiAsARxURERERERERERETVHhPFREREREREREQSyfzuO2R+953UYRARsfSESklJCQAgKSlJlPYUCgXS09Mhl8thb28vSptUPva5NNjv1sc+tz72ufWxz62Pff4f1fmg6vywMuK5bdXAfrc+9rn1sc+Bu99FAwDqhIdb5Xjsc+tjn1sf+/w/xpzbMlH8r9TUVABAx44dJY6EiIiIiGxBamoq6tWrJ3UYJuG5LRFRJfTTj1JHQERVmCHntjJBEATrhGPbCgsLce7cOfj7+8PBwfz8eVJSEjp27IjTp08jKChIhAipIuxzabDfrY99bn3sc+tjn1sf+/w/JSUlSE1NRcuWLeHi4iJ1OCbhuW3VwH63Pva59bHPrY99bn3sc+tjn//HmHNbjij+l4uLCzp06CB6u0FBQQgODha9XdKPfS4N9rv1sc+tj31ufexz62Ofl6qsI4lVeG5btbDfrY99bn3sc+tjn1sf+9z62OelDD235WR2RERERERERERERNUcE8VERERERERERERE1RwTxRbi6emJRYsWwdPTU+pQqg32uTTY79bHPrc+9rn1sc+tj31O5eHfhzTY79bHPrc+9rn1sc+tj31ufexz03AyOyIiIiIiIiIiIqJqjiOKiYiIiIiIiIiIiKo5JoqJiIiIiIiIiIiIqjkmiomIiIiIiIiIiIiqOSaKiYiIiIiIiIiIiKo5JoqJiIiIiIiIiIiIqjkmikV29epVDBgwAO7u7ggICMBLL72EgoICqcOqMq5fv45nn30WrVu3hoODA1q0aKFzu71796JNmzZwcXFBw4YN8dlnn1k50qpj+/bteOKJJ1CnTh24u7vj0Ucfxf/+9z8olUqN7djn4omJiUGPHj3g7+8PZ2dn1K9fH3PmzEF2drbGduxzy8nLy0NwcDBkMhn+/PNPjXXsd3Fs2rQJMplM69/8+fM1tmN/W8YXX3yBVq1awcXFBQEBARg2bJjGevY7lcXzW8vi+a118dzW+nhuKz2e21oHz2+lw3Nb8ThIHUBVkpWVhV69eiEkJAQ7duxASkoK5syZg/T0dHzzzTdSh1clXLhwAXv27EGnTp2gVCq1TugA4Pfff8fw4cMxceJErFq1CidPnsSsWbPg5OSEadOmSRB15bZy5UqEhITgww8/RGBgII4ePYoXX3wR8fHx+PDDDwGwz8WWkZGBrl274uWXX4aPjw/Onz+PxYsX4/z58zhw4AAA9rmlLVu2DCUlJVrL2e/i279/P7y8vNSPa9eurf6Z/W0ZixcvxurVq/HGG2+gU6dOyMjIwP79+9Xr2e9UFs9vLY/nt9bFc1vr47mt9Hhua108v7UuntuKTCDRvPfee4Kbm5uQmpqqXrZlyxYBgHDx4kUJI6s6FAqF+udJkyYJzZs319pmwIABQseOHTWWTZ8+XQgKCtLYnwyTkpKitWz27NmCi4uLUFhYKAgC+9wa1q9fLwAQEhISBEFgn1vSpUuXBHd3dyEqKkoAIPzxxx/qdex38WzcuFEAoPGZ+TD2t/guXrwo2NvbCzExMXq3Yb9TWTy/tTye31oXz21tA89trYfnttbD81vr47mt+Fh6QkR79+5Fnz594Ofnp142atQoODs7Y+/evRJGVnXY2ZX/JyuXy3HkyBE8+eSTGssnTJiApKQknDlzxpLhVUn+/v5ay9q0aYPCwkJkZGSwz62kZs2aAIDi4mL2uYW9+OKLePbZZ9G4cWON5ex362J/W8amTZtQv3599OvXT+d69js9jOe3lsfzW+viua1t4Lmt9fDc1nawz8XHc1vxMVEsokuXLqFp06Yay5ydndGgQQNcunRJoqiqlxs3bqCoqEjr99CsWTMA4O9BJMePH4evry8CAgLY5xakUChQWFiIv//+G0uXLsXQoUMREhLCPreg77//HmfPnsXChQu11rHfLaN58+awt7dH/fr1sXz5cigUCgDsb0uJjY1Fy5YtsWzZMgQEBMDJyQk9evRAXFwcAPY7aeP5rfT4urQ8nttaB89trY/nttLg+a318NxWfKxRLKLMzEx4e3trLffx8UFGRob1A6qGMjMzAUDr9+Dj4wMA/D2I4M8//8TGjRuxaNEi2Nvbs88tKCQkBAkJCQCAAQMGYOvWrQD4d24pDx48wJw5c7B8+XJ4enpqrWe/iysoKAhLlixBp06dIJPJsGvXLrz55ptISEjAp59+yv62kOTkZPz999+4cOECoqKi4OTkhCVLlqBv3764du0a+5208PxWenxdWhbPba2H57bWxXNb6+P5rfXx3FZ8TBSLTCaTaS0TBEHncrIcff3N34N5kpOTMWrUKHTs2BGvvfaaxjr2ufj27t2LvLw8XLhwAcuWLcPQoUNx8OBB9Xr2ubjefvttBAYGYvLkyeVux34XR//+/dG/f3/14379+sHV1VU9EYUK+1tcSqUSeXl52LFjB5o3bw4AaNeuHUJDQ7F+/Xo89thjANjvpInnt7aBr0vx8dzWunhua108t7U+nt9aH89txcfSEyLy8fFRX60oKysrS321gixL1c8P/x5Uj/l7MF12djYGDhwINzc37Nq1C46OjgDY55b06KOPomvXrpg+fTp27tyJo0ePYufOnexzC7h9+zZWrlyJJUuWICcnB1lZWcjLywMA5OXlIS8vj/1uBWPHjoVCoUBcXBz720J8fX0RGBioPpEGSke/NGnSBBcuXGC/kxae30qPr0vL4Lmt9fHc1np4bms7eH5rWTy3FR8TxSJq2rSpVn0TuVyOGzduaNVDIcto0KABnJyctH4PFy9eBAD+HkxUWFiIYcOG4f79+9i/f7968gmAfW4trVu3hr29Pa5fv84+t4CbN2+iqKgIgwcPho+PD3x8fDB06FAAQM+ePdGnTx/2uxUIgqD+mf1tGfr6TRAE2NnZsd9JC89vpcfXpfh4bis9nttaFs9tbQfPby2L57biY6JYRIMGDcLhw4eRnp6uXrZz507I5XIMGjRIwsiqD2dnZ/Tq1Qvbtm3TWL5161YEBQWhTZs2EkVWeZWUlGDs2LE4e/Ys9u/fj5CQEI317HPr+P3336FQKFC/fn32uQW0bt0aR48e1fi3evVqAEBUVBQ+++wz9rsVREdHw97eHm3atGF/W8iQIUNw//59nD9/Xr0sISEBly9fRqtWrdjvpIXnt9Lj61JcPLe1DTy3tSye29oOnt9aFs9tLUAg0WRmZgq1a9cWHnvsMWH//v3C119/Lfj5+QkTJkyQOrQqIz8/X9i+fbuwfft24fHHHxfq1KmjfpySkiIIgiD89ttvgoODgzBt2jTh6NGjwttvvy3Y2dkJGzZskDj6ymnGjBkCAOGDDz4Qfv/9d41/2dnZgiCwz8U2YsQI4Z133hF2794tHDp0SFi5cqUQGBgoPProo4JcLhcEgX1uDUePHhUACH/88Yd6GftdPP369RPef/99Yc+ePcKePXuEyMhIQSaTCS+//LJ6G/a3+EpKSoS2bdsKYWFhQnR0tLBz506hTZs2Qu3atYW8vDxBENjvpInnt5bH81vr4rmt9fHc1jbw3NbyeH5rfTy3FR8TxSK7cuWK0K9fP8HNzU3w8/MTZs2aJTx48EDqsKqMmzdvCgB0/jt69Kh6uz179gitWrUSnJychPr16wuffvqpdEFXciEhIexzK1u+fLnQunVrwcPDQ3B3dxeaN28uvPXWW+ovLyrsc8vSdTItCOx3sbz44otCWFiY4OrqKjg7OwstW7YUPv74Y0GpVGpsx/4W3/3794Xx48cLXl5egpubmzBw4EDh8uXLGtuw36ksnt9aFs9vrYvnttbHc1vbwHNby+P5rTR4bisumSCUKZhCRERERERERERERNUOaxQTERERERERERERVXNMFBMRERERERERERFVc0wUExEREREREREREVVzTBQTERERERERERERVXNMFBMRERERERERERFVc0wUExEREREREREREVVzTBQTERERERERERERVXNMFBMRERERERERERFVc0wUExGVY+/evRgwYABq1qwJJycnhISE4Pnnn8eNGzfU29SrVw8zZ86UMErzyGQyrFixQuowDPL4449jyJAhUodBREREVCnx3Na28NyWiGwNE8VERHq8+eabGDx4MNzc3LBu3TocOnQIy5Ytw5UrV9CnTx+pwyMiIiIiMhjPbYmIqCIOUgdARGSL9u/fj3feeQevv/463n33XfXy7t27Y+LEidi9e7eE0ZGlFBYWwsXFReowiIiIiETFc9vqiee2RGQsjigmItJhxYoVCAwMxJIlS3SuHzp0qNayTz/9FCEhIfDy8sITTzyB1NRU9br8/HzMnDkTjRs3hpubG+rVq4dnn30W2dnZGm2obvUrr61ffvkFMpkMBw4cwPjx4+Hh4YGQkBB88MEHWjH9/vvv6NWrF9zd3eHl5YXx48cjJSXF6P4wJK5NmzZBJpMhLS1NY98WLVpg8uTJ6seTJ09GixYtEBMTg5YtW8LV1RXdunXDzZs3kZGRgXHjxsHT0xMNGjRAdHS0zni+/vprNGjQAK6urnj88cdx5coVjfWCIGDFihVo1KgRnJ2dUb9+faxevVpjm8WLF6NGjRo4ffo0unTpAhcXF6xZs8boviEiIiKydTy31cRzWyIi3ZgoJiJ6SElJCU6ePIk+ffrA0dHRoH127dqF3bt3Y+3atfj444/xyy+/YNasWer1Dx48gEKhwDvvvIN9+/bh7bffxrFjxzBixAij21J57rnn0KhRI+zcuRODBw/Ga6+9hv3796vX//7773j88cfh5eWF6OhorF+/Hn/88QeGDRtmQq8YHpchkpKS8Nprr+Gtt97Cli1bcPPmTUyYMAFPPvkkWrRogR07dqBdu3Z4+umncfv2bY19//77byxfvhzvvfcevv76ayQlJaF///6Qy+XqbV566SUsXLgQkyZNwp49ezB58mS89tpriIqK0mirqKgIEyZMwDPPPIP9+/ejX79+Jj0fIiIiIlvFc1vTnqMxeG5LRFWGQEREGpKTkwUAwvz58w3aPiQkRAgODhYKCwvVy9544w3B0dFRUCgUOvcpLi4WTpw4IQAQrly5YlRbR48eFQAI8+bNU2+jUCiEOnXqCBEREepl3bt3F7p27SoolUr1svPnzwsymUzYs2ePehkA4cMPPzT7OW7cuFEAIKSmpmrs27x5c2HSpEnqx5MmTRJkMplw8eJF9bI1a9YIAITXXntNvSwzM1Owt7cXPvroI/WyHj16CHZ2dsLVq1fVy65evSrY2dkJ69atEwRBEK5fvy7IZDL1Y5V58+YJtWrVUse7aNEiAYCwbdu2cp87ERERUWXGc1vTniPPbYmoOuKIYiKihwiCAKB0xmRD9ejRA87OzurHzZo1Q3FxscatcJs3b0abNm1Qo0YNODo6Ijw8HABw9epVo9sCoDFCwM7ODk2aNMG9e/cAlI7yOHnyJMaMGQOFQoGSkhKUlJSgcePGCAoKwh9//GHwczM2LkM88sgjaNq0qfpxo0aNAEBjIhVvb28EBATg7t27Gvu2aNECYWFh6sdhYWFo0aIFYmNjAQCHDh0CAIwaNUr9vEtKStC7d28kJydrtTdo0CCj4yciIiKqLHhua/pzNBTPbYmoqmCimIjoIX5+fnBxccGdO3cM3sfb21vjsZOTE4DSCSQAYOfOnZg4cSI6duyIbdu2ITY2Fjt37tTYxtC2yttOtU1mZiYUCgVmz54NR0dHjX+JiYlaJ5RiPEcx2irvOakEBARotRcQEICkpCQAQFpaGgRBgJ+fn8bzHjBgAABoPHc3Nze4u7sbHT8RERFRZcFzW9Oeoxht8dyWiCobB6kDICKyNQ4ODggPD8ehQ4dQXFxscC238mzfvh2tW7fGunXr1MuOHTtmdrv6eHt7QyaTYcGCBXjiiSe01vv5+Yl+TNWMykVFRRrLMzIyRD2OrlEeKSkpaNeuHQDA19cXMpkMJ06cUJ+kl9W4cWP1z8aMrCEiIiKqjHhuaxqe2xJRdcQRxUREOsydOxf379/H0qVLda7/+eefjWqvoKBA68Ruy5YtJsdXEXd3d3Tp0gWXLl1C+/bttf7Vq1dP9GMGBwcDAC5duqRedv78eSQnJ4t6nPPnz+PatWvqx9euXcP58+fRqVMnAEDv3r0BAOnp6Tqfu4eHh6jxEBEREdk6ntsaj+e2RFQdcUQxEZEOAwYMwBtvvIG3334bly5dwlNPPYWAgADcvn0bmzdvxtWrVzFkyBCD2+vbty9eeOEFLF26FF27dsW+fftw+PBhCz4D4MMPP0SvXr0wbtw4PPnkk/Dx8cG9e/dw8OBBTJkyBY8//riox+vUqRPq1KmD2bNnY/ny5cjJycF7770HX19fUY8TGBiIYcOGYdmyZRAEAW+99RZq166NSZMmASitCffCCy/gmWeewbx589CpUycUFxfj6tWrOHr0KH788UdR4yEiIiKydTy3NR7PbYmoOmKimIhIj7fffhtdunTBJ598gunTpyM3NxePPPII+vbti9WrVxvVVmRkJOLj4/Hpp59ixYoV6N+/P7799lt07tzZQtEDXbt2xYkTJ7Bo0SJMmTIFRUVFCA4ORu/evdGwYUPRj+fo6IidO3fiueeew5gxY9CwYUN89NFHePHFF0U9Ttu2bTFq1CjMmzcPSUlJ6NSpE6KiotS3BwLAJ598gsaNG2PdunVYunQp3N3d0bhxY4wdO1bUWIiIiIgqC57bGofntkRUHckE1RSoRERERERERERERFQtsUYxERERERERERERUTXHRDERERERERERERFRNcdEMREREREREREREVE1x0QxERERERERERERUTXHRDERERERERERERFRNcdEMREREREREREREVE1x0QxERERERERERERUTXHRDERERERERERERFRNcdEMREREREREREREVE1x0QxERERERERERERUTXHRDERERERERERERFRNcdEMREREREREREREVE193+DmaxyoZUnZAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 1430x495 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "GREGORY_STRONG = np.array([-0.3784699703803287,-0.9567626659181726,-0.8179874873507014,-0.020671723923978647,-1.6087130585080849,0.08127903798852154,0.6956693233271531,1.544195356876418,0.5927346860388676,-0.8489485385905176,0.43858977862804016,-1.559186899389141,-0.18384435268829924,0.9939475414195276,1.2059195024486082,-0.8284688624125628,-1.4945657874326943,0.11493909870515462,-0.33164725260926375,0.1949388032470849,-0.1784684930899738,0.28226439065466113,-0.2228581136479476,0.01836419639097997,0.179072704844774,0.8593338897793665,1.1517664699408345,1.7295627959197244,-1.5999379538051328,-0.3197255194522214,0.10444885990116792,-0.07873366670852244,1.4502840090705629,0.07442736532651861,2.4156518913009744,3.4470945391962067,4.42733578071102,4.917354468433997,1.7268682183139688,0.6729781434892117,3.0580693292363144,-0.08116233205055323,0.4307808218166238,-0.03775658686836669,0.41565927775274636,-1.1065200209775305,-0.3823833125281616,-0.009454658265808158,-0.7414136191397724,0.8827430327293673,0.26783865034753185,-0.3104363177060627,0.43565144811387047,2.2354076952054545,0.12054098372764588,-0.055324523758522814,-0.7214125850365256,0.9759053943257279,0.7122267936151523,0.6534601833317317,-1.241895954440375,0.1261295731369516,-0.7448438873418364,0.9232148792684454])\n",
    "# GREGORY_STRONG = None   # uncomment to use a simulated spectrum instead\n",
    "\n",
    "d = (GREGORY_STRONG if GREGORY_STRONG is not None\n",
    "     else 4.3 * f + SIGMA * rng.standard_normal(N))   # simulated stand-in\n",
    "\n",
    "fig, axes = plt.subplots(1, 2, figsize=(13, 4.5))\n",
    "axes[0].plot(nu, f, \"m\"); axes[0].set_title(r\"Line profile $f_i$\")\n",
    "axes[1].plot(nu, d, \"k*--\", lw=0.6, ms=5); axes[1].axhline(0, c=\"k\", lw=0.6)\n",
    "axes[1].axvline(NU0, c=\"C3\", ls=\":\", lw=1); axes[1].set_title(\"measured spectrum\")\n",
    "for ax in axes: ax.set_xlabel(\"Channel number\")\n",
    "axes[0].set_ylabel(\"Signal strength (mK)\")\n",
    "plt.tight_layout()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3535a673",
   "metadata": {},
   "source": [
    "## 1. The likelihood\n",
    "For independent Gaussian noise\n",
    "$$\n",
    "\\ln p(D\\mid M_1,T,I) = -\\frac N2\\ln(2\\pi\\sigma^2) - \\frac{\\chi^2(T)}{2},\\qquad\n",
    "\\chi^2(T)=\\sum_{i=1}^{N}\\frac{(d_i-Tf_i)^2}{\\sigma^2}.\n",
    "$$\n",
    "$M_0$ is the special case $T = 0$; since it has no free parameter, its evidence **is** its likelihood: $p(D\\mid M_0,I) = p(D\\mid M_1, T=0, I)$.\n",
    "\n",
    "**TODO 1.** Complete `loglike(T)`. It must accept a NumPy **array** of $T$ values (hint: broadcasting, `T[:, None]`)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "72d98040",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ln p(D|M0) = [-116.10166024]   ->  p(D|M0) = [3.78172214e-51]\n"
     ]
    }
   ],
   "source": [
    "def loglike(T, d=d, f=f, sigma=SIGMA):\n",
    "    # ln p(D | M1, T, I) for an array of T values\n",
    "    T = np.atleast_1d(T)\n",
    "    chi2 = (((d[None, :] - T[:, None] * f[None, :]) / sigma) ** 2).sum(axis=1)\n",
    "    return -0.5 * len(d) * np.log(2 * np.pi * sigma ** 2) - 0.5 * chi2\n",
    "\n",
    "ln_Z0 = loglike([0])         # TODO: ln p(D | M0, I)\n",
    "# print(\"ln p(D|M0) =\", ln_Z0)\n",
    "print(\"ln p(D|M0) =\", ln_Z0, \"  ->  p(D|M0) =\", np.exp(ln_Z0))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "398ff406",
   "metadata": {},
   "source": [
    "**TODO 2.** Find the best-fit $\\hat T$ (maximum likelihood = minimum $\\chi^2$) on a fine grid of $T$, and the maximum likelihood $p(D\\mid \\hat T, M_1, I)$. Plot $\\ln p(D\\mid M_1,T,I)$ against $T$.\n",
    "\n",
    "*Question:* the likelihoods are of order $10^{-37}$ and $10^{-51}$. Why do we work with logarithms?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a63c94bb",
   "metadata": {},
   "outputs": [],
   "source": [
    "T_fine = np.linspace(0, 10, 20001)\n",
    "lnL = ...            # TODO\n",
    "T_hat = ...          # TODO\n",
    "ln_Lmax = ...        # TODO\n",
    "print(f\"T_hat = {T_hat:.2f} mK,  max likelihood = {np.exp(ln_Lmax):.2e}\")\n",
    "\n",
    "# TODO: plot lnL against T_fine"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f2d10e98-adcd-4d2c-ad59-1b51f37caeab",
   "metadata": {},
   "source": [
    "## Errors on $\\hat T$: how can we compute them?\n",
    "\n",
    "> **Note:** this part will be done in class. You can skip it for the homework.\n",
    "\n",
    "For this problem all four methods give the same answer, $\\delta T \\approx 0.53$ mK. The model is linear in $T$, so $\\ln L(T)$ is an exact parabola. In general they can differ, and seeing why is part of the next lecture.\n",
    "\n",
    "**1. Error propagation (analytic).**\n",
    "Setting $d\\chi^2/dT = 0$ gives the best fit as a weighted sum of the data:\n",
    "$$\n",
    "\\hat T = \\frac{\\sum_i d_i f_i}{\\sum_i f_i^2} = \\sum_i w_i\\, d_i, \\qquad w_i = \\frac{f_i}{\\sum_j f_j^2}.\n",
    "$$\n",
    "Because $\\hat T$ is a weighted sum of independent measurements, each with variance $\\sigma^2$, the rule for the variance of a linear combination (Lectures 1–2) gives\n",
    "$$\n",
    "\\mathrm{Var}(\\hat T) = \\sum_i w_i^2\\,\\sigma^2 = \\frac{\\sigma^2 \\sum_i f_i^2}{\\left(\\sum_i f_i^2\\right)^2} = \\frac{\\sigma^2}{\\sum_i f_i^2}\n",
    "\\quad\\Rightarrow\\quad \\delta T = \\frac{\\sigma}{\\sqrt{\\sum_i f_i^2}} .\n",
    "$$\n",
    "\n",
    "**2. Curvature of $\\ln L$ at the peak.**\n",
    "Near its maximum, any smooth $\\ln L$ is approximately a parabola:\n",
    "$$\n",
    "\\ln L(T) \\approx \\ln L_{\\max} + \\tfrac12 \\left.\\frac{d^2 \\ln L}{dT^2}\\right|_{\\hat T} (T-\\hat T)^2 ,\n",
    "$$\n",
    "so $L(T)$ is approximately a Gaussian with\n",
    "$$\n",
    "\\delta T^2 = -\\left(\\left.\\frac{d^2 \\ln L}{dT^2}\\right|_{\\hat T}\\right)^{-1}.\n",
    "$$\n",
    "A sharp peak means a small error, and a flat peak means a large error. Numerically, take a finite difference on the grid:\n",
    "```python\n",
    "h = T_fine[1] - T_fine[0]; k = np.argmax(lnL)\n",
    "curv = (lnL[k+1] - 2*lnL[k] + lnL[k-1]) / h**2\n",
    "dT_curv = 1 / np.sqrt(-curv)\n",
    "```\n",
    "\n",
    "**3. From $\\Delta\\chi^2 = 1$.**\n",
    "Since $\\ln L = \\text{const} - \\chi^2/2$,\n",
    "$$\n",
    "\\Delta\\chi^2(T) = \\chi^2(T) - \\chi^2_{\\min} = -2\\,[\\ln L(T) - \\ln L_{\\max}] .\n",
    "$$\n",
    "For one parameter, the 68% (1σ) interval is where $\\Delta\\chi^2 \\le 1$. Equivalently, it's where $\\ln L$ is within ½ of its maximum. ($\\Delta\\chi^2 = 4$ gives 2σ, and $9$ gives 3σ.)\n",
    "```python\n",
    "dchi2 = -2 * (lnL - lnL.max())          # Δχ²(T) relative to the minimum\n",
    "inside = T_fine[dchi2 <= 1.0]\n",
    "print(f\"68% interval: [{inside.min():.2f}, {inside.max():.2f}] mK\")\n",
    "```\n",
    "\n",
    "**4. Monte Carlo simulations (what the error bar means).**\n",
    "A frequentist error bar is the scatter of the estimate over repeated experiments. To get it:\n",
    "- simulate many spectra from the best-fit model plus noise;\n",
    "- refit each one;\n",
    "- take the standard deviation of the fitted values.\n",
    "```python\n",
    "n_sim = 5000\n",
    "sims  = T_hat * f + SIGMA * rng.standard_normal((n_sim, N))   # fake data sets\n",
    "T_sim = sims @ f / (f @ f)                                     # best fit of each\n",
    "print(\"MC scatter of T_hat:\", T_sim.std())\n",
    "```\n",
    "\n",
    "**Questions for class.** Do all four methods agree? Which ones would still work if the model were not linear in $T$, or if the data were Poisson counts?"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "d7883bbf",
   "metadata": {},
   "source": [
    "## 2. Priors for the line strength\n",
    "Prior information: theory predicts $0.1 \\le T \\le 100$ mK.\n",
    "\n",
    "| prior | $p(T\\mid M_1,I)$ |\n",
    "|---|---|\n",
    "| uniform | $1/(T_{\\max}-T_{\\min})$ |\n",
    "| Jeffreys | $1/\\left[T\\ln(T_{\\max}/T_{\\min})\\right]$ |\n",
    "\n",
    "**TODO 3.** Complete `log_prior(T, kind)`. Check numerically that each prior integrates to 1 over $[T_{\\min}, T_{\\max}]$.\n",
    "\n",
    "*Question:* what fraction of the prior probability lies in 10–100 mK for each prior?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "58f28c58",
   "metadata": {},
   "outputs": [],
   "source": [
    "TMIN, TMAX = 0.1, 100.0\n",
    "\n",
    "def log_prior(T, kind, Tmin=TMIN, Tmax=TMAX):\n",
    "    T = np.asarray(T, float)\n",
    "    if kind == \"uniform\":\n",
    "        return ...    # TODO\n",
    "    if kind == \"jeffreys\":\n",
    "        return ...    # TODO\n",
    "    raise ValueError(kind)\n",
    "\n",
    "# A logarithmic grid resolves the narrow likelihood peak and the 3-decade range\n",
    "T_grid = np.geomspace(TMIN, TMAX, 20001)\n",
    "for kind in [\"uniform\", \"jeffreys\"]:\n",
    "    print(kind, \"normalisation =\", ...)   # TODO: should be 1"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6a0b5fcc",
   "metadata": {},
   "source": [
    "## 3. Evidence, Bayes factor and odds ratio\n",
    "$$\n",
    "p(D\\mid M_1,I) = \\int p(T\\mid M_1,I)\\,p(D\\mid M_1,T,I)\\,dT,\n",
    "\\qquad\n",
    "B_{10} = \\frac{p(D\\mid M_1,I)}{p(D\\mid M_0,I)},\n",
    "\\qquad\n",
    "O_{10} = \\frac{p(M_1\\mid I)}{p(M_0\\mid I)}\\,B_{10}.\n",
    "$$\n",
    "\n",
    "**TODO 4.** Compute $\\ln p(D\\mid M_1,I)$ for both priors with `log_trapz`, then $B_{10}$ and $O_{10}$ for equal prior odds."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "52ef7d94",
   "metadata": {},
   "outputs": [],
   "source": [
    "prior_odds = 1.0      # p(M1|I)/p(M0|I): neither theory favoured in advance\n",
    "\n",
    "for kind in [\"uniform\", \"jeffreys\"]:\n",
    "    ln_Z1 = ...       # TODO: log of the integral of prior x likelihood\n",
    "    B10   = ...       # TODO\n",
    "    O10   = ...       # TODO\n",
    "    print(f\"{kind:9s}: p(D|M1) = {np.exp(ln_Z1):.2e}   B10 = {B10:.2e}   O10 = {O10:.2e}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dab8b223",
   "metadata": {},
   "source": [
    "**TODO 5 — the Occam factor.** Check the back-of-the-envelope estimate\n",
    "$$\n",
    "B_{10} \\approx \\underbrace{\\frac{p(D\\mid \\hat T,M_1,I)}{p(D\\mid M_0,I)}}_{\\text{max-likelihood ratio}} \\times \\underbrace{\\sqrt{2\\pi}\\,\\delta T\\;p(\\hat T\\mid M_1,I)}_{\\text{Occam factor } \\Omega_T},\n",
    "\\qquad \\delta T = \\sigma/\\sqrt{\\textstyle\\sum_i f_i^2}.\n",
    "$$\n",
    "Compute the max-likelihood ratio, $\\Omega_T$ for each prior, and their product. Compare with TODO 4."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "703d526f",
   "metadata": {},
   "outputs": [],
   "source": [
    "dT = ...              # TODO\n",
    "ML_ratio = ...        # TODO\n",
    "for kind in [\"uniform\", \"jeffreys\"]:\n",
    "    Omega = ...       # TODO\n",
    "    print(f\"{kind:9s}: ML ratio = {ML_ratio:.2e}   Omega = {Omega:.3g}   ML x Omega = {ML_ratio * Omega:.2e}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "23829504",
   "metadata": {},
   "source": [
    "## 4. Questions (answer in a short text cell)\n",
    "1. Which model is favoured, and how strongly (use the Kass–Raftery scale)? Does the choice of prior matter here? Why?\n",
    "2. The max-likelihood ratio is $\\sim 10^{14}$, but $B_{10}$ is only $\\sim 10^{12}$. Explain the difference in one sentence.\n",
    "3. Repeat TODO 4 with $T_{\\max} = 1000$ mK and $T_{\\max} = 10$ mK. How does $B_{10}$ change for each prior? Why does the uniform prior react more?\n",
    "4. How small would the prior odds $p(M_1\\mid I)/p(M_0\\mid I)$ have to be for the posterior odds to be 1 : 1? Would a sceptic of theory 1 be convinced by these data?\n",
    "5. *(Optional)* Plot the normalised posterior $p(T\\mid D,M_1,I)$ for both priors. Does the prior matter for **parameter estimation**?\n",
    "\n",
    "*Next lecture:* the same calculation for a **weak** line — where the answer changes completely."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}
