Warfarin-Copy1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3964\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "from off_pol_eval_functions import * \n",
    "import sys\n",
    "import numpy as np\n",
    "import scipy\n",
    "import scipy.stats\n",
    "from scipy.spatial.distance import cdist\n",
    "import csv\n",
    "from sklearn import tree\n",
    "from sklearn import linear_model, neighbors, ensemble, tree\n",
    "\n",
    "from warfarin_functions import * \n",
    "\n",
    "ntrs = np.arange(100,2501,100)\n",
    "runsper = 4\n",
    "runsperper = 25\n",
    "params = [ntr for ntr in ntrs for run in range(runsper)]\n",
    "\n",
    "myid = 0#int(sys.argv[1])\n",
    "# np.random.seed(myid)\n",
    "ntr = 1000#params[myid-1]\n",
    "\n",
    "f = open('warfrin.csv','rU')\n",
    "csvr = csv.reader(f, dialect=csv.excel)\n",
    "header = np.array(csvr.next())\n",
    "data   = list(csvr)\n",
    "f.close()\n",
    "\n",
    "# filter only to subjects that reached stable dose of warfarin\n",
    "# and stable observed INR\n",
    "data = [x for x in data if x[37]=='1' and x[38]!='NA' and x[39]!='NA']\n",
    "\n",
    "agegroups = sorted(set(x[8].replace('NA','') for x in data))\n",
    "\n",
    "xmap = lambda x: \\\n",
    "[\n",
    "    # gender 0\n",
    "    ('Male?',x[3]=='male'),\n",
    "    # race 1:4\n",
    "    ('White?',x[5]=='White'),\n",
    "    ('Asian?',x[5]=='Asian'),\n",
    "    ('Black?',x[5]=='Black or African American'),\n",
    "    # ethnicity 4\n",
    "    ('Non-hispanic?',x[7]=='not Hispanic or Latino'),\n",
    "    # age 5:7\n",
    "    ('Age group',agegroups.index(x[8]) if x[8] in agegroups else 0),\n",
    "    ('No age?',x[8] not in agegroups or agegroups.index(x[8]) == 0),\n",
    "    # height 7\n",
    "    ('Height',float(x[9]) if x[9] not in ('NA', '') else 0.),\n",
    "        # x[9] in ('NA', ''), # NA indicator\n",
    "    # weight 8\n",
    "    ('Weight',float(x[10]) if x[10] not in ('NA', '') else 0.),\n",
    "        # x[10] in ('NA', ''), # NA indicator\n",
    "    # BMI\n",
    "    ('BMI',float(x[10])*100.*100./float(x[9])/float(x[9]) if x[10] not in ('NA', '') and x[9] not in ('NA', '') else 0.),\n",
    "] + [# Indication for Warfarin Treatment 9:17\n",
    "    ('Indication for Warfarin Treatment: '+str(i), str(i) in x[11])\n",
    "    for i in range(1,9)\n",
    "# ] + [# comorbidities\n",
    "#     c in (y.strip() for y in x[12].lower().split('; '))\n",
    "#     for c in comorbidities\n",
    "] + [# diabetes 17:19\n",
    "    ('Diabetes=0?',x[13]=='0'),\n",
    "    ('Diabetes=1?',x[13]=='1'),\n",
    "] + [#Congestive Heart Failure and/or Cardiomyopathy 19:21\n",
    "    ('Congestive Heart Failure and/or Cardiomyopathy=0?',x[14]=='0'),\n",
    "    ('Congestive Heart Failure and/or Cardiomyopathy=1?',x[14]=='1'),\n",
    "] + [#Valve Replacement 21:23\n",
    "    ('Valve Replacement=0?',x[15]=='0'),\n",
    "    ('Valve Replacement=1?',x[15]=='1'),\n",
    "# ] + [# medications\n",
    "#   x[16]\n",
    "] + [\n",
    "    # Aspirin 24:26\n",
    "    ('aspirin=0?',x[17]=='0'),\n",
    "    ('aspirin=1?',x[17]=='1'),\n",
    "    # Acetaminophen or Paracetamol (Tylenol) 26:28\n",
    "    ('Acetaminophen=0?',x[18]=='0'),\n",
    "    ('Acetaminophen=1?',x[18]=='1'),\n",
    "    # Was Dose of Acetaminophen or Paracetamol (Tylenol) >1300mg/day 28:30\n",
    "    ('Acetaminophen hi dose=0?',x[19]=='0'),\n",
    "    ('Acetaminophen hi dose=1?',x[19]=='1'),\n",
    "    # Simvastatin (Zocor) 30:32\n",
    "    ('Simvastatin=0?',x[20]=='0'),\n",
    "    ('Simvastatin=1?',x[20]=='1'),\n",
    "    # Atorvastatin (Lipitor) 32:34\n",
    "    ('Simvastatin=0?',x[21]=='0'),\n",
    "    ('Simvastatin=1?',x[21]=='1'),\n",
    "    # Fluvastatin (Lescol) 34:36\n",
    "    ('Fluvastatin=0?',x[22]=='0'),\n",
    "    ('Fluvastatin=1?',x[22]=='1'),\n",
    "    # Lovastatin (Mevacor) 36:38\n",
    "    ('Lovastatin=0?',x[23]=='0'),\n",
    "    ('Lovastatin=1?',x[23]=='1'),\n",
    "    # Pravastatin (Pravachol) 38:40\n",
    "    ('Pravastatin=0?',x[24]=='0'),\n",
    "    ('Pravastatin=1?',x[24]=='1'),\n",
    "    # Rosuvastatin (Crestor) 40:42\n",
    "    ('Rosuvastatin=0?',x[25]=='0'),\n",
    "    ('Rosuvastatin=1?',x[25]=='1'),\n",
    "    # Cerivastatin (Baycol) 42:43\n",
    "    ('Cerivastatin=0?',x[26]=='0'),\n",
    "    ('Cerivastatin=1?',x[26]=='1'),\n",
    "    # Amiodarone (Cordarone)\n",
    "    ('Amiodarone=0?',x[27]=='0'),\n",
    "    ('Amiodarone=1?',x[27]=='1'),\n",
    "    # Carbamazepine (Tegretol)\n",
    "    ('Carbamazepine=0?',x[28]=='0'),\n",
    "    ('Carbamazepine=1?',x[28]=='1'),\n",
    "    # Phenytoin (Dilantin)\n",
    "    ('Phenytoin=0?',x[29]=='0'),\n",
    "    ('Phenytoin=1?',x[29]=='1'),\n",
    "    # Rifampin or Rifampicin\n",
    "    ('Rifampin=0?',x[30]=='0'),\n",
    "    ('Rifampin=1?',x[30]=='1'),\n",
    "    # Sulfonamide Antibiotics\n",
    "    ('Sulfonamide Antibiotics=0?',x[31]=='0'),\n",
    "    ('Sulfonamide Antibiotics=1?',x[31]=='1'),\n",
    "    # Macrolide Antibiotics\n",
    "    ('Macrolide Antibiotics=0?',x[32]=='0'),\n",
    "    ('Macrolide Antibiotics=1?',x[32]=='1'),\n",
    "    # Anti-fungal Azoles\n",
    "    ('Anti-fungal Azoles=0?',x[33]=='0'),\n",
    "    ('Anti-fungal Azoles=1?',x[33]=='1'),\n",
    "    # Herbal Medications, Vitamins, Supplements\n",
    "    ('Herbal Medications, Vitamins, Supplements=0?',x[34]=='0'),\n",
    "    ('Herbal Medications, Vitamins, Supplements=1?',x[34]=='1'),\n",
    "] + [\n",
    "    #smoker\n",
    "    ('Smoker=0?',x[40]=='0'),\n",
    "    ('Smoker=0?',x[40]=='1'),\n",
    "] + [\n",
    "    # CYP2C9 consensus\n",
    "    ('CYP2C9 *1/*1',x[59]=='*1/*1'),\n",
    "    ('CYP2C9 *1/*2',x[59]=='*1/*2'),\n",
    "    ('CYP2C9 *1/*3',x[59]=='*1/*3'),\n",
    "    ('CYP2C9 NA',x[59]=='' or x[59]=='NA'),\n",
    "    # VKORC1 -1639 consensus\n",
    "    ('VKORC1 -1639 A/A',x[60]=='A/A'),\n",
    "    ('VKORC1 -1639 A/G',x[60]=='A/G'),\n",
    "    ('VKORC1 -1639 G/G',x[60]=='G/G'),\n",
    "    # VKORC1 497 consensus\n",
    "    ('VKORC1 497 T/T',x[61]=='T/T'),\n",
    "    ('VKORC1 497 G/T',x[61]=='G/T'),\n",
    "    ('VKORC1 497 G/G',x[61]=='G/G'),\n",
    "    # VKORC1 1173 consensus\n",
    "    ('VKORC1 1173 T/T',x[62]=='T/T'),\n",
    "    ('VKORC1 1173 C/T',x[62]=='C/T'),\n",
    "    ('VKORC1 1173 C/C',x[62]=='C/C'),\n",
    "    # VKORC1 1542 consensus\n",
    "    ('VKORC1 1542 C/C',x[63]=='C/C'),\n",
    "    ('VKORC1 1542 C/G',x[63]=='C/G'),\n",
    "    ('VKORC1 1542 G/G',x[63]=='G/G'),\n",
    "    # VKORC1 3730 consensus\n",
    "    ('VKORC1 3730 A/A',x[64]=='A/A'),\n",
    "    ('VKORC1 3730 A/G',x[64]=='A/G'),\n",
    "    ('VKORC1 3730 G/G',x[64]=='G/G'),\n",
    "    # VKORC1 2255 consensus\n",
    "    ('VKORC1 2255 C/C',x[65]=='C/C'),\n",
    "    ('VKORC1 2255 C/T',x[65]=='C/T'),\n",
    "    ('VKORC1 2255 T/T',x[65]=='T/T'),\n",
    "    # VKORC1 -4451 consensus\n",
    "    ('VKORC1 -4451 C/C',x[66]=='C/C'),\n",
    "    ('VKORC1 -4451 A/C',x[66]=='A/C'),\n",
    "    ('VKORC1 -4451 A/A',x[66]=='A/A')\n",
    "#     ,\n",
    "#     ('Therapeutic Dose',float(x[38]) if x[38] not in ('NA', '') else 0.),\n",
    "#     ('INR On Therapeutic Dose',float(x[39]) if x[39] not in ('NA', '') else 0.),\n",
    "#     ('Target INR', float(x[35]) if x[35] not in ('NA', '') else 0. )\n",
    "]\n",
    "\n",
    "X       = np.array([zip(*xmap(x))[1] for x in data])\n",
    "Xnames  = np.array(zip(*xmap(data[0]))[0])\n",
    "goodidx = np.where( (X.std(0)>=.05)  )[0] #np.sort(np.hstack((np.where(X.std(0)>=.05)[0],np.where(Xnames=='BMI')[0])))\n",
    "X       = X[:,goodidx]\n",
    "Xnames  = Xnames[goodidx]\n",
    "# filter out by where BMI is nonzero \n",
    "goodbmi = np.where( (X[:,9] > 0.003) )[0]\n",
    "X       = X[goodbmi,:]\n",
    "# ytarget = np.array([float(x[38]) for x in data])/7.\n",
    "\n",
    "target_INR = np.array([float(x[35]) if x[35] not in ('NA', '') else 0. for x in data])\n",
    "therapeut_dose = np.array([float(x[38]) if x[38] not in ('NA', '') else 0. for x in data])\n",
    "obs_INR = np.array([float(x[39]) if x[39] not in ('NA', '') else 0. for x in data])\n",
    "\n",
    "target_INR = target_INR[goodbmi]; therapeut_dose = therapeut_dose[goodbmi]; obs_INR = obs_INR[goodbmi]\n",
    "\n",
    "n = X.shape[0]\n",
    "print n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "International Normalised Ratio (INR) testing is well established as an integral part of warfarin treatment. INR has a critical role in maintaining the warfarin response within a therapeutic range, to provide the benefits of anticoagulation, while avoiding the risks of haemorrhage (Figure 1).\n",
    "\n",
    "The INR is defined as\n",
    "\n",
    "$$ INR =\\left( \\frac{\\text{Patient Prothrombin Time (sec)}}{\\text{Local Geometric Mean Prothrombin Time (sec)}}\\right)^{ISI} $$\n",
    "\n",
    "http://www.bpac.org.nz/BT/2010/November/inr.aspx\n",
    "\n",
    "In most situations the INR target is 2.5 (target range 2.0 – 3.0). This range is appropriate for the prophylaxis or treatment of venous thromboembolism and reduction of the risk of systemic embolism for people with atrial fibrillation and valvular heart disease.5 In some situations higher ranges are more appropriate. The target INR may vary depending on individual clinical situations. The target INR for mechanical prosthetic valves is dependent on the type of valve replacement used.6\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Semi-simulated dosages\n",
    "\n",
    "We can generate semi-simulated dosages from the true therapeutic dose by modeling dosage as a function of BMI and perturbing with some noise via a mixture model: \n",
    "\n",
    "$$T_i' = \\mu_{T^*} + \\sigma_{T^*} \\sqrt{\\theta} \\frac{x_{BMI}- \\mu_{BMI}}{\\sigma_{BMI}} + \\sqrt{(1 - \\theta)} \\sigma_{T^*} \\epsilon$$\n",
    "\n",
    "Suppose that the counterfactual outcome is given by a quadratic penalty to the INR outcome of the observed therapeutic dose: \n",
    "\n",
    "$$y_i' = y_i^* + \\frac{(T_i' - T_i)^2sgn(T_i' - T_i)}{max(T_i^2, T_i'^2)}$$\n",
    "\n",
    "This model also yields the propensity score for assigning the counterfactual dose. \n",
    "### Propensity score given our treatment generation model from BMI : \n",
    "   \n",
    "where $\\epsilon \\sim N(0,1)$ and $Z_{BMI} = \\frac{x_{BMI} - \\mu_{BMI}}{\\sigma_{BMI}}$\n",
    "\n",
    "$$ \\mathbb{P}[T_i'=t \\mid x_{bmi} ] = \\mathbb{P}\\left[\\epsilon = \\frac{t - \\mu_T^* - \\sigma_T^* \\sqrt{\\theta}Z_{BMI}}{\\sqrt{1 - \\theta}} \\right]$$\n",
    "\n",
    "\n",
    "### Dosing policy evaluation and optimization \n",
    "\n",
    "In any observational dataset, we will not have access to the full counterfactual distribution of outcomes. Previous analysis of the Warfarin dataset has focused on predicting being in a discrete bin of the \"correct\" therapeutic dose, e.g. Hamsa Bastani, Bertsimas/Kallus introducing 0-1 loss for being in the same bin. Other work (Chen, Zeng, Kosorok) look at a direct model of the outcome and directly penalize \n",
    "\n",
    "We leverage the rich previous study of the impacts of warfarin dosing, treating these models and analyses as \"oracle information\" we use to simulate realistic outcome data. We consider previous studies analyzing the effect that different INR levels have on the odds ratios for complications including ischemic stroke and hemorrhagic complications. The intensity of anticoagulation induced by warfarin therapy trades off between decreased stroke prevention at low levels of INR (less effective anticoagulation), and increased risk of intracranial bleeding in over-anticoagulation regimes. Various factors such as patient age and pre-existing health conditions (previous stroke) affect patient risk in ways that have been previously studied, but we do not directly account for in this simulated study.\n",
    "\n",
    "Proposed model: \n",
    "\n",
    "* Generate simulated doses as $T_i' = \\mu_{T^*} + \\sigma_{T^*} \\sqrt{\\theta} \\frac{x_{BMI}- \\mu_{BMI}}{\\sigma_{BMI}} + \\sqrt{(1 - \\theta)} \\sigma_{T^*} \\epsilon$\n",
    "* Simulate INR response to dose adjustment (inversion of dosage guidelines) <https://depts.washington.edu/anticoag/home/content/warfarin-maintenance-dosing-nomogram>: $\\hat{y}(T_i')$\n",
    "* Compute expected risk as a function of how the odds ratio is affected by INR [ACC/AHA/ESC 2006 guidelines for the management of patients with atrial fibrillation](https://academic.oup.com/europace/article/8/9/651/530495/ACC-AHA-ESC-2006-guidelines-for-the-management-of): $R(\\hat{y}(T_i')$\n",
    "* Compare $R(\\hat{y}(\\tau(X_i))$ across policies $\\tau$\n",
    "\n",
    "\n",
    "Optimization Question: \n",
    "\n",
    "If one generates a dataset of simulated dose adjustments and INR responses in this fashion, can you recover the therapeutic dose (or an optimal linear treatment policy which is similar) if you do off-policy optimization? What kind of treatment policy does the optimization return? \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "5"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "16.9843047536\n",
      "17.2106872797\n",
      "[  0.          37.3536892   15.98231102 ...,  41.29424242  35.91736714\n",
      "   5.91272953]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAh8AAAFkCAYAAACAUFlOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRETEV/8P/vKNrMlKV2cAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1efa9630>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Global values \n",
    "mu_dose = np.mean(therapeut_dose)\n",
    "std_dose = np.std(therapeut_dose)\n",
    "mu_bmi = np.mean(X[:,9])\n",
    "std_bmi = np.std(X[:,9])\n",
    "\n",
    "# regr = linear_model.LinearRegression()\n",
    "# regr.fit(X[:,9].reshape([n,1]), therapeut_dose.reshape([n,1]))\n",
    "\n",
    "theta = 0.5\n",
    "eps = np.random.normal(size=n)\n",
    "\n",
    "# eps = np.random.randn(n)\n",
    "bmi_Z = (X[:,9] - mu_bmi)/std_bmi\n",
    "\n",
    "# fit on centered model\n",
    "regr = linear_model.LinearRegression(fit_intercept = False)\n",
    "regr.fit(bmi_Z.reshape([n,1]), therapeut_dose.reshape([n,1]))\n",
    "\n",
    "\n",
    "sim_dose = mu_dose  + (np.sqrt(theta)*bmi_Z*std_dose + np.sqrt(1-theta)*std_dose*eps) \n",
    "sim_dose[sim_dose < 0 ] = - sim_dose[sim_dose < 0 ] \n",
    "print np.std(sim_dose)\n",
    "print std_dose \n",
    "\n",
    "plt.hist(therapeut_dose[therapeut_dose<150],alpha=0.5,color='g',label= 'true_dose', normed=True)\n",
    "plt.hist(sim_dose,alpha = 0.5, label = \"simulated dosage\", normed=True)\n",
    "plt.legend()\n",
    "\n",
    "y_counter = np.asarray(simulated_loss(sim_dose, therapeut_dose, obs_INR))\n",
    "print y_counter\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Propensity score given our treatment generation model from BMI : \n",
    "   \n",
    "where $\\epsilon \\sim N(0,1)$ and $Z_{BMI} = \\frac{x_{BMI} - \\mu_{BMI}}{\\sigma_{BMI}}$\n",
    "\n",
    "$$ \\mathbb{P}[T_i'=t \\mid x_{bmi} ] = \\mathbb{P}\\left[\\epsilon = \\frac{t - \\mu_T^* - \\sigma_T^* \\sqrt{\\theta}Z_{BMI}}{\\sqrt{1 - \\theta}} \\right]$$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8.56986448726e-07\n",
      "2.33967519703\n"
     ]
    }
   ],
   "source": [
    "# Add to param dictionary: \n",
    "# therapeutic dose (\"oracle values\") \n",
    "# observed INR  (\"oracle values\")\n",
    "#! FIXME will require global access to therapeutic doses/ \n",
    "def generated_Q(bmi, T, theta, NULL): \n",
    "    # centeredaround mu_dose with scale stddev of dose\n",
    "    theta = 0.5\n",
    "    Z_bmi = (bmi-mu_bmi)/std_bmi\n",
    "    return norm.pdf( T, loc = mu_dose + std_dose*np.sqrt(theta)*Z_bmi, scale = np.sqrt(1-theta) )\n",
    "\n",
    "print generated_Q(30,40,0.9,std_dose)\n",
    "\n",
    "d = 1\n",
    "t_lo =0\n",
    "t_hi = max(therapeut_dose)\n",
    "data = { 'n': n, 'y': obs_INR,'Q': generated_Q, 'x': X[:,9],'x_samp':X[:,9],  'd': d, 'T': therapeut_dose,'t_lo': t_lo ,'t_hi': t_hi  }\n",
    "data['tau'] = sim_dose\n",
    "data['h'] = 3; data['n_bins'] = 10\n",
    "data['kernel_int_func'] = epanechnikov_int\n",
    "data['kernel_func'] = epanechnikov_kernel\n",
    "data['oracle_func'] = counterfactual_y_eval\n",
    "data['threshold'] = 0.01 \n",
    "# print np.mean(counterfactual_y_eval(**data))\n",
    "\n",
    "print off_policy_evaluation(**data)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "23.659662705\n"
     ]
    }
   ],
   "source": [
    "'''\n",
    "Systematically evaluate over a treatment space defined by a linear treatment policy\n",
    "'''\n",
    "n_trials = 10\n",
    "n_treatments = 10\n",
    "t_lo = 0 ; t_hi = max(therapeut_dose) \n",
    "print t_hi / np.min(X[:,9])\n",
    "beta_space = np.linspace(0.5,2, n_treatments)\n",
    "off_pol_evals = np.zeros(n_treatments)\n",
    "oracle_evals = np.zeros(n_treatments)\n",
    "discrete_off_pol_evals = np.zeros(n_treatments)\n",
    "\n",
    "def simulated_loss(sim_dose, therapeut_dose, obs_INR):\n",
    "    \n",
    "    return (np.abs(therapeut_dose - sim_dose))+ np.random.normal()*0.1*mu_dose\n",
    "\n",
    "\n",
    "data['y'] = simulated_loss(sim_dose, therapeut_dose, obs_INR)\n",
    "data['y_samp'] = simulated_loss(sim_dose, therapeut_dose, obs_INR)\n",
    "data['T_samp'] = sim_dose \n",
    "\n",
    "# for beta_ind, beta in enumerate(beta_space):\n",
    "#     tau = np.clip(beta*data['x'] , t_lo, t_hi).flatten()\n",
    "#     data['tau'] = tau\n",
    "#     oracle_evals[beta_ind] = np.mean(simulated_loss(tau, therapeut_dose, obs_INR))\n",
    "#     off_pol_evals[beta_ind] = off_policy_evaluation(**data)\n",
    "#     discrete_off_pol_evals[beta_ind] = off_pol_disc_evaluation(discretize_tau_policy , **data)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x1d86e9e8>"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfwAAAFkCAYAAADFZ4k9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nlySpAP8fQvjWOvUbN64AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1d242828>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.scatter(beta_space, oracle_evals,color='b')\n",
    "plt.scatter(beta_space, off_pol_evals,color='g')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train a more sophisticated model with more covariates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\ipykernel\\__main__.py:7: DeprecationWarning: using a non-integer number instead of an integer will result in an error in the future\n"
     ]
    }
   ],
   "source": [
    "train_var = [5,9]+range(18,40)\n",
    "subframe = X[:,[5,9]+range(18,40)]\n",
    "subXnames = Xnames[[5,9]+range(18,40)]\n",
    "# subframe = X \n",
    "# subXnames = Xnames\n",
    "d = subframe.shape[1]\n",
    "trainind = np.random.choice(range(n),size =  round(0.8*n),replace = False)\n",
    "train = subframe[trainind]\n",
    "test_mask = np.ones(n, dtype=bool)\n",
    "test_mask[trainind] = False\n",
    "test = subframe[test_mask,:]\n",
    "\n",
    "y = simulated_loss(sim_dose[trainind], therapeut_dose[trainind], obs_INR[trainind])\n",
    "def generated_Q_multid_X(x,t,t_lo,t_hi): \n",
    "    return generated_Q(x,t,t_lo,t_hi)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": false
   },
   "source": [
    "## Load full dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "y = simulated_loss(sim_dose[trainind], therapeut_dose[trainind], obs_INR[trainind])\n",
    "h=4\n",
    "d = X.shape[1]\n",
    "train_data = { 'n': train.shape[0],'h': h, 'y': obs_INR,'Q': generated_Q_multid_X,'x_full': X, 'x': X,'x_samp': X[trainind,:],  'd': d, 'T': therapeut_dose[trainind],'t_lo': t_lo ,'t_hi': t_hi  }\n",
    "train_data['kernel_int_func'] = epanechnikov_int\n",
    "train_data['T_samp'] = np.array(sim_dose[trainind],order='F'); train_data['y_samp'] = np.array(simulated_loss(sim_dose[trainind], therapeut_dose[trainind], obs_INR[trainind]),order='F')\n",
    "train_data['kernel_func'] = epanechnikov_kernel\n",
    "train_data['DATA_TYPE'] = 'warfarin'\n",
    "train_data['BMI_IND'] = 9\n",
    "train_data['inds'] = trainind\n",
    "train_data['threshold'] = 0.01\n",
    "\n",
    "bounds = [(0, np.mean(therapeut_dose)/np.mean(X[:,i])) for i in range(d) ]\n",
    "bnds = tuple(tuple(x) for x in bounds)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### load subframe data (temporary)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'simulated_loss' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-1-8226e06ea673>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0my\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0msimulated_loss\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0msim_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtherapeut_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mobs_INR\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      3\u001b[0m \u001b[0md\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0msubframe\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mtrain_data\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;33m{\u001b[0m \u001b[1;34m'n'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mtrain\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;34m'h'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[1;36m4\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'y'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mobs_INR\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;34m'Q'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mgenerated_Q\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;34m'x_full'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0msubframe\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'x'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0marray\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtrain\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0morder\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'F'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;34m'x_samp'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0marray\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtrain\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0morder\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'F'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'T_samp'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0msim_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'd'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0md\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'T'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mtherapeut_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;34m't_lo'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mt_lo\u001b[0m \u001b[1;33m,\u001b[0m\u001b[1;34m't_hi'\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mt_hi\u001b[0m  \u001b[1;33m}\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0mtrain_data\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m'kernel_int_func'\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mepanechnikov_int\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'simulated_loss' is not defined"
     ]
    }
   ],
   "source": [
    "\n",
    "y = simulated_loss(sim_dose[trainind], therapeut_dose[trainind], obs_INR[trainind])\n",
    "d = subframe.shape[1]\n",
    "train_data = { 'n': train.shape[0],'h': 4, 'y': obs_INR,'Q': generated_Q,'x_full': subframe, 'x': np.array(train,order='F'),'x_samp': np.array(train,order='F'), 'T_samp': sim_dose[trainind], 'd': d, 'T': therapeut_dose[trainind],'t_lo': t_lo ,'t_hi': t_hi  }\n",
    "train_data['kernel_int_func'] = epanechnikov_int\n",
    "train_data['T_samp'] = np.array(sim_dose[trainind],order='F'); train_data['y_samp'] = np.array(simulated_loss(sim_dose[trainind], therapeut_dose[trainind], obs_INR[trainind]),order='F')\n",
    "train_data['kernel_func'] = epanechnikov_kernel\n",
    "train_data['DATA_TYPE'] = 'warfarin'\n",
    "train_data['BMI_IND'] = 1\n",
    "train_data['inds'] = trainind\n",
    "train_data['threshold'] = 0.01\n",
    "\n",
    "bounds = [(0, np.mean(therapeut_dose)/np.mean(subframe[:,i])) for i in range(d) ]\n",
    "bnds = tuple(tuple(x) for x in bounds)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pickle\n",
    "t_lo = 0 \n",
    "t_hi = max(therapeut_dose)\n",
    "\n",
    "beta_d = np.random.rand(d)\n",
    "h = 10\n",
    "n_train = train.shape[0]\n",
    "# train_data['oracle_func'] = counterfactual_y_eval\n",
    "\n",
    "def pol_eval_wrapper(beta, args): \n",
    "    data = dict(args)\n",
    "    t_lo = data['t_lo']; t_hi = data['t_hi']\n",
    "    tau = np.clip(np.dot(beta, data['x'].T) , t_lo, t_hi).flatten()\n",
    "    data['tau'] = tau\n",
    "    return off_policy_evaluation(**data)\n",
    "\n",
    "def oracle_eval_wrapper(beta, args): \n",
    "    data = dict(args)\n",
    "    t_lo = data['t_lo']; t_hi = data['t_hi']\n",
    "    tau = np.clip(np.dot(beta, data['x'].T) , t_lo, t_hi).flatten()\n",
    "    return np.mean( simulated_loss(tau,therapeut_dose,obs_INR) )\n",
    "\n",
    "# print off_pol_epan_lin_grad(beta_d, train_data)\n",
    "LAMBDA = 0.4\n",
    "def var_regularized_pol_eval_wrapper(beta, args): \n",
    "    data = dict(args)\n",
    "    t_lo = data['t_lo']; t_hi = data['t_hi']\n",
    "    tau = np.clip(np.dot(beta, data['x'].T) , t_lo, t_hi).flatten()\n",
    "    data['tau'] = tau\n",
    "    return off_policy_evaluation(**data) + LAMBDA * np.linalg.norm(beta,2) + LAMBDA * off_policy_variance(**data)\n",
    "\n",
    "def var_regularized_oracle_wrapper(beta, args): \n",
    "    data = dict(args)\n",
    "    t_lo = data['t_lo']; t_hi = data['t_hi']\n",
    "    tau = np.clip(np.dot(beta, data['x'].T) , t_lo, t_hi).flatten()\n",
    "    return np.mean( simulated_loss(tau,therapeut_dose,obs_INR) )\n",
    "\n",
    "def off_pol_var_lin_grad(beta, *args):\n",
    "    \"\"\"\n",
    "    Compute a gradient for special case of Epanechnikov kernel and linear policy tau\n",
    "    \"\"\"\n",
    "    \n",
    "    d = len(beta) \n",
    "    params = dict(args[0]); n = params['n']\n",
    "    THRESH = params['threshold']\n",
    "    y_out = params['y']; x = params['x']; h = params['h'];Q = params['Q']; n = params['n']; t_lo = params['t_lo'];  t_hi = params['t_hi']\n",
    "    kernel = params['kernel_func'];kernel_int =  params['kernel_int_func']\n",
    "    y_out = params['y_samp']; T = params['T_samp']; x = params['x_samp']\n",
    " \n",
    "    #! FIXME x vs xsamp\n",
    "    tau = np.dot(beta, params['x'].T)\n",
    "    params['tau'] = tau; params['beta'] = beta\n",
    "    clip_tau = np.clip(tau, t_lo, t_hi)\n",
    "    BMI_IND = params.get('BMI_IND'); THRESH = params['threshold'] # propensity score for warfarin data evaluations \n",
    "    \n",
    "\n",
    "    [f, g, nabla_f, nabla_g] = f_g(**params)\n",
    "    # compute gradient vector via quotient rule\n",
    "    if g < THRESH: \n",
    "        g = THRESH  \n",
    "    grad_estimator = np.asarray((g*nabla_f - f*nabla_g) / g**2 )\n",
    "    var_grad = -1*f *grad_estimator\n",
    "    var_grad_term1 = 0 \n",
    "    for i in range(n): \n",
    "        if (params.get('DATA_TYPE') == 'warfarin'): \n",
    "            Q_i = Q(x[i,BMI_IND], T[i], t_lo, t_hi)\n",
    "        else: \n",
    "            Q_i = Q(x[i], T[i], t_lo, t_hi)\n",
    "        if (abs(clip_tau[i] - t_lo) <= h):\n",
    "            alpha = kernel_int((t_lo-clip_tau[i])/h, 1)\n",
    "        elif (abs(clip_tau[i] - t_hi) <= h):\n",
    "            alpha = kernel_int(-1,  (t_hi - clip_tau[i])/h )\n",
    "        else:\n",
    "            alpha = 1\n",
    "#         Qs[i] = kernel( (clip_tau[i] - T[i])/h )/max(Q_i,THRESH)\n",
    "        if abs((clip_tau[i] - T[i])/h) >= 1:\n",
    "            var_grad_term1 += 0 # don't add anything to partial derivatives \n",
    "        else:\n",
    "            var_grad_term1 += 2* (kernel( (clip_tau[i] - T[i])/h )*1.0 * y_out[i]/max(Q_i,THRESH) * 1.0/alpha)*(.75*x[i]*y_out[i]/max(Q_i,THRESH)*(1 - 1.0*(clip_tau[i] - T[i])/h ))\n",
    "    var_grad_term1 = var_grad_term1*1.0/n\n",
    "    var_grad += var_grad_term1\n",
    "    return np.asarray((g*nabla_f - f*nabla_g) / g**2 ) + 2*LAMBDA*beta + 2*var_grad \n",
    "\n",
    "def plot_training_set_hist(beta, x, therapeut_dose, sim_dose, title, policy_label): \n",
    "    plt.hist(np.abs(np.dot(beta, x.T)  - therapeut_dose ),range=[0,80],alpha = 0.5,label=policy_label)\n",
    "    plt.hist(np.abs(sim_dose - therapeut_dose),alpha = 0.5,range=[0,80],color = 'g',label='original simulated dose' )\n",
    "    plt.legend()\n",
    "    plt.title(title,y=1.08)\n",
    "    plt.ylabel('frequencies')\n",
    "    plt.ylim((0, 1800))\n",
    "    plt.xlabel('Absolute value of distance between recommended and therapeutic dose')\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "train_data['y'] = train_data['y']-np.mean(train_data['y'])\n",
    "train_data['y_samp'] = train_data['y_samp']-np.mean(train_data['y_samp'])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n"
     ]
    }
   ],
   "source": [
    "import datetime\n",
    "d = X.shape[1]\n",
    "n_restarts = 1\n",
    "vals = np.zeros(n_restarts)\n",
    "betas = [None]*n_restarts\n",
    "i = 0 \n",
    "while i <= n_restarts: \n",
    "    print i\n",
    "\n",
    "    beta_d = 0.8*(mu_dose / (d*np.mean(X[trainind,:],axis=0))) +0.2*np.random.uniform(size=d,low=np.zeros(d), high = [bnds[k][1]*1.0/d for k in range(d) ])\n",
    "    \n",
    "#     beta_d = mu_dose / (d*np.mean(X[trainind,:],axis=0))\n",
    "#     oracle_res = minimize(pol_eval_wrapper, x0 = beta_d, jac = off_pol_epan_lin_grad, bounds = bnds , options={'disp':True,'gtol':1e-4}, args=train_data.items())  \n",
    "#     try: \n",
    "    oracle_res = minimize(var_regularized_pol_eval_wrapper, x0 = beta_d, bounds = bnds , options={'disp':True,'gtol':1e-3}, args=train_data.items())  \n",
    "\n",
    "#         oracle_res = minimize(var_regularized_pol_eval_wrapper, x0 = beta_d, jac = off_pol_epan_lin_grad, bounds = bnds , options={'disp':True,'gtol':1e-4}, args=train_data.items())  \n",
    "    betas[i] = oracle_res.x\n",
    "    print oracle_res.x\n",
    "    vals[i] = oracle_res.fun\n",
    "    print oracle_res.fun\n",
    "    i += 1\n",
    "#     except: \n",
    "#         print \"Gradient error\"\n",
    "    \n",
    "pickle.dump(oracle_res.x, open(str(datetime.datetime.now().strftime(\"%Y-%m-%d_%H-%M\")) + 'empirical_opt_beta-v2', 'wb'))\n",
    "best_off_pol_beta = betas[np.argmin(vals)]\n",
    "best_off_pol_val = vals[np.argmin(vals)]\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_off_pol_beta"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\axes\\_axes.py:5933: UserWarning: 2D hist input should be nsamples x nvariables;\n",
      " this looks transposed (shape is 81 x 3171)\n",
      "  'this looks transposed (shape is %d x %d)' % x.shape[::-1])\n"
     ]
    },
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-34-230883ddc2f2>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[1;31m# best_off_pol_beta = pickle.load(open('2017-05-16_23-32empirical_opt_beta-v2',\"rb\"))\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      2\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m \u001b[0mplot_training_set_hist\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbest_off_pol_beta\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtherapeut_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msim_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;34m\"Comparison of empirically optimal treatment policy vs original dose\"\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;34m\"empirically optimal training policy\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m \u001b[0mplot_training_set_hist\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbest_off_pol_beta\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtest_mask\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtherapeut_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtest_mask\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msim_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtest_mask\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;34m\"Comparison of empirically optimal treatment policy vs original dose\"\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;34m\"empirically optimal training policy\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-13-6ae331ebbed6>\u001b[0m in \u001b[0;36mplot_training_set_hist\u001b[0;34m(beta, x, therapeut_dose, sim_dose, title, policy_label)\u001b[0m\n\u001b[1;32m     37\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m     38\u001b[0m \u001b[1;32mdef\u001b[0m \u001b[0mplot_training_set_hist\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbeta\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtherapeut_dose\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msim_dose\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtitle\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mpolicy_label\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m---> 39\u001b[0;31m     \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mhist\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mabs\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdot\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbeta\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mx\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mT\u001b[0m\u001b[1;33m)\u001b[0m  \u001b[1;33m-\u001b[0m \u001b[0mtherapeut_dose\u001b[0m \u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mrange\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;36m80\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0malpha\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m0.5\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mlabel\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mpolicy_label\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     40\u001b[0m     \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mhist\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mabs\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0msim_dose\u001b[0m \u001b[1;33m-\u001b[0m \u001b[0mtherapeut_dose\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0malpha\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m0.5\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mrange\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;36m80\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mcolor\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;34m'g'\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mlabel\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'original simulated dose'\u001b[0m \u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m     41\u001b[0m     \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mlegend\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\pyplot.pyc\u001b[0m in \u001b[0;36mhist\u001b[0;34m(x, bins, range, normed, weights, cumulative, bottom, histtype, align, orientation, rwidth, log, color, label, stacked, hold, data, **kwargs)\u001b[0m\n\u001b[1;32m   2956\u001b[0m                       \u001b[0mhisttype\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mhisttype\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0malign\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0malign\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0morientation\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0morientation\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   2957\u001b[0m                       \u001b[0mrwidth\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mrwidth\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlog\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mlog\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mcolor\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mcolor\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlabel\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mlabel\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2958\u001b[0;31m                       stacked=stacked, data=data, **kwargs)\n\u001b[0m\u001b[1;32m   2959\u001b[0m     \u001b[1;32mfinally\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   2960\u001b[0m         \u001b[0max\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mhold\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mwashold\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\__init__.pyc\u001b[0m in \u001b[0;36minner\u001b[0;34m(ax, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1808\u001b[0m                     warnings.warn(msg % (label_namer, func.__name__),\n\u001b[1;32m   1809\u001b[0m                                   RuntimeWarning, stacklevel=2)\n\u001b[0;32m-> 1810\u001b[0;31m             \u001b[1;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0max\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1811\u001b[0m         \u001b[0mpre_doc\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0minner\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m__doc__\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   1812\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mpre_doc\u001b[0m \u001b[1;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\axes\\_axes.pyc\u001b[0m in \u001b[0;36mhist\u001b[0;34m(self, x, bins, range, normed, weights, cumulative, bottom, histtype, align, orientation, rwidth, log, color, label, stacked, **kwargs)\u001b[0m\n\u001b[1;32m   6075\u001b[0m                 patch = _barfunc(bins[:-1]+boffset, height, width,\n\u001b[1;32m   6076\u001b[0m                                  \u001b[0malign\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'center'\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlog\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mlog\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m-> 6077\u001b[0;31m                                  color=c, **{bottom_kwarg: bottom})\n\u001b[0m\u001b[1;32m   6078\u001b[0m                 \u001b[0mpatches\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mpatch\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   6079\u001b[0m                 \u001b[1;32mif\u001b[0m \u001b[0mstacked\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\__init__.pyc\u001b[0m in \u001b[0;36minner\u001b[0;34m(ax, *args, **kwargs)\u001b[0m\n\u001b[1;32m   1808\u001b[0m                     warnings.warn(msg % (label_namer, func.__name__),\n\u001b[1;32m   1809\u001b[0m                                   RuntimeWarning, stacklevel=2)\n\u001b[0;32m-> 1810\u001b[0;31m             \u001b[1;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0max\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1811\u001b[0m         \u001b[0mpre_doc\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0minner\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m__doc__\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   1812\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mpre_doc\u001b[0m \u001b[1;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\axes\\_axes.pyc\u001b[0m in \u001b[0;36mbar\u001b[0;34m(self, left, height, width, bottom, **kwargs)\u001b[0m\n\u001b[1;32m   2135\u001b[0m             \u001b[0mr\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mget_path\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_interpolation_steps\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m100\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   2136\u001b[0m             \u001b[1;31m#print r.get_label(), label, 'label' in kwargs\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m-> 2137\u001b[0;31m             \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0madd_patch\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mr\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   2138\u001b[0m             \u001b[0mpatches\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mr\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   2139\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\axes\\_base.pyc\u001b[0m in \u001b[0;36madd_patch\u001b[0;34m(self, p)\u001b[0m\n\u001b[1;32m   1761\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mget_clip_path\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   1762\u001b[0m             \u001b[0mp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mset_clip_path\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpatch\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1763\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_update_patch_limits\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mp\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1764\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpatches\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mp\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   1765\u001b[0m         \u001b[0mp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_remove_method\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mlambda\u001b[0m \u001b[0mh\u001b[0m\u001b[1;33m:\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpatches\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mremove\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mh\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\axes\\_base.pyc\u001b[0m in \u001b[0;36m_update_patch_limits\u001b[0;34m(self, patch)\u001b[0m\n\u001b[1;32m   1787\u001b[0m                 \u001b[0mxys\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mpatch_to_data\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtransform\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mxys\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   1788\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1789\u001b[0;31m             \u001b[0mupdatex\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mupdatey\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mpatch\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mget_transform\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0;31m\\\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1790\u001b[0m                 \u001b[0mcontains_branch_seperately\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtransData\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   1791\u001b[0m             self.update_datalim(xys, updatex=updatex,\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\patches.pyc\u001b[0m in \u001b[0;36mget_transform\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    193\u001b[0m         \u001b[0mto\u001b[0m \u001b[0mthe\u001b[0m \u001b[1;33m:\u001b[0m\u001b[1;32mclass\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m`\u001b[0m\u001b[0mPatch\u001b[0m\u001b[1;33m`\u001b[0m\u001b[1;33m.\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m    194\u001b[0m         \"\"\"\n\u001b[0;32m--> 195\u001b[0;31m         \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mget_patch_transform\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0martist\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mArtist\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mget_transform\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    196\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m    197\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mget_data_transform\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\patches.pyc\u001b[0m in \u001b[0;36mget_patch_transform\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    668\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m    669\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mget_patch_transform\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m--> 670\u001b[0;31m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_update_patch_transform\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    671\u001b[0m         \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_rect_transform\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m    672\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\patches.pyc\u001b[0m in \u001b[0;36m_update_patch_transform\u001b[0;34m(self)\u001b[0m\n\u001b[1;32m    663\u001b[0m         \u001b[0mbbox\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtransforms\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mBbox\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfrom_bounds\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mwidth\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mheight\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m    664\u001b[0m         \u001b[0mrot_trans\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtransforms\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mAffine2D\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m--> 665\u001b[0;31m         \u001b[0mrot_trans\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mrotate_deg_around\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_angle\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    666\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_rect_transform\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mtransforms\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mBboxTransformTo\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbbox\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m    667\u001b[0m         \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_rect_transform\u001b[0m \u001b[1;33m+=\u001b[0m \u001b[0mrot_trans\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\matplotlib\\transforms.pyc\u001b[0m in \u001b[0;36mrotate_deg_around\u001b[0;34m(self, x, y, degrees)\u001b[0m\n\u001b[1;32m   1924\u001b[0m         \u001b[1;32mand\u001b[0m \u001b[1;33m:\u001b[0m\u001b[0mmeth\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m`\u001b[0m\u001b[0mscale\u001b[0m\u001b[1;33m`\u001b[0m\u001b[1;33m.\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   1925\u001b[0m         \"\"\"\n\u001b[0;32m-> 1926\u001b[0;31m         \u001b[1;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtranslate\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m-\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m-\u001b[0m\u001b[0my\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mrotate_deg\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mdegrees\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtranslate\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0my\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1927\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   1928\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0mtranslate\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mself\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtx\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mty\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgsAAAFkCAYAAACuFXjcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nADARCwAAwEQsAAAAE7EAAABM/wWoA68vIMdUkAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1de484a8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# best_off_pol_beta = pickle.load(open('2017-05-16_23-32empirical_opt_beta-v2',\"rb\"))\n",
    "\n",
    "plot_training_set_hist(best_off_pol_beta, X[trainind,:], therapeut_dose[trainind], sim_dose[trainind],\"Comparison of empirically optimal treatment policy vs original dose\",\"empirically optimal training policy\")\n",
    "plot_training_set_hist(best_off_pol_beta, X[test_mask,:], therapeut_dose[test_mask], sim_dose[test_mask],\"Comparison of empirically optimal treatment policy vs original dose\",\"empirically optimal training policy\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[  3.34878260e-12]\n",
      " [  2.27233744e+00]\n",
      " [  2.10186436e-13]\n",
      " [  1.94613037e+00]\n",
      " [  4.93963436e-13]\n",
      " [ -4.59980522e-13]\n",
      " [  5.47827155e-11]\n",
      " [ -1.15325336e-13]\n",
      " [  1.48744485e-01]\n",
      " [ -1.05391182e-13]\n",
      " [  3.95693317e+00]\n",
      " [  3.06510535e+00]\n",
      " [ -7.63551671e-14]\n",
      " [  4.01937904e+00]\n",
      " [  1.01013936e-11]\n",
      " [  4.37414928e-12]\n",
      " [  2.58864883e+00]\n",
      " [  9.48284751e-01]\n",
      " [  1.41427592e-12]\n",
      " [  8.54196282e-12]\n",
      " [  1.81297162e-01]\n",
      " [  1.58129760e-13]\n",
      " [  1.38782100e-13]\n",
      " [  5.47813915e-11]\n",
      " [  1.86564553e-01]\n",
      " [  1.86913855e-12]\n",
      " [  1.14177992e-11]\n",
      " [  5.69451219e-12]\n",
      " [  5.14907981e-12]\n",
      " [  3.95146992e-01]\n",
      " [  2.09766884e-12]\n",
      " [  4.44294320e-12]\n",
      " [  4.10999861e-12]\n",
      " [  1.83330525e-12]\n",
      " [  2.01268876e-12]\n",
      " [  3.62105766e-11]\n",
      " [  1.86483697e-12]\n",
      " [  1.76396490e-11]\n",
      " [  1.79634632e-12]\n",
      " [  4.36252780e+00]\n",
      " [  6.86478941e-13]\n",
      " [  3.25328825e+00]\n",
      " [  2.71032694e-12]\n",
      " [  1.73112785e-12]\n",
      " [  1.25880200e+01]\n",
      " [  1.73112887e-12]\n",
      " [  1.53340836e+01]\n",
      " [  1.79634721e-12]\n",
      " [  1.39707498e-12]\n",
      " [  1.93685720e-12]\n",
      " [  1.93685829e-12]\n",
      " [  2.13552122e-11]\n",
      " [  2.22546695e+00]\n",
      " [  8.14146381e-13]\n",
      " [  3.21197396e-02]\n",
      " [  2.62757185e+00]\n",
      " [  7.85466746e+00]\n",
      " [  3.12960252e+00]\n",
      " [  2.73353289e-12]\n",
      " [  6.03574940e+00]\n",
      " [ -3.13894344e-13]\n",
      " [  2.90217705e+00]\n",
      " [  7.79963303e+00]\n",
      " [  5.36453104e-14]\n",
      " [  3.15295794e-12]\n",
      " [  2.36331980e-12]\n",
      " [ -9.59825904e-14]\n",
      " [  2.70707747e+00]\n",
      " [  4.52580093e+00]\n",
      " [ -2.68331030e-13]\n",
      " [  1.01199851e+00]\n",
      " [  5.66660826e+00]\n",
      " [  3.73786403e-12]\n",
      " [  4.33937273e-12]\n",
      " [ -2.96983686e-13]\n",
      " [  1.67093953e-12]\n",
      " [  1.54999388e-12]\n",
      " [ -1.41385616e-13]\n",
      " [  1.05543784e-12]\n",
      " [  4.75539444e-12]\n",
      " [  5.25852378e-12]]\n"
     ]
    }
   ],
   "source": [
    "import cvxpy as cvx\n",
    "d = X.shape[1]\n",
    "beta = cvx.Variable(d)\n",
    "# Implement a thresholded loss\n",
    "#cost = cvx.sum_entries(cvx.max_elemwise(cvx.abs(X[trainind,:]*beta - therapeut_dose[trainind]) - therapeut_dose[trainind]*0.1, 0))/n*1.0\n",
    "# normal l1 loss (median regression)\n",
    "cost = cvx.sum_entries(cvx.abs(X[trainind,:]*beta - therapeut_dose[trainind]))/n*1.0\n",
    "\n",
    "constraints = [beta >= 0]\n",
    "for i in range(d): \n",
    "    constraints += [beta[i] <= np.mean(therapeut_dose[trainind])/np.mean(X[trainind,i]) ]\n",
    "prob = cvx.Problem(cvx.Minimize(cost), constraints)\n",
    "\n",
    "prob.solve()\n",
    "print beta.value\n",
    "prescient_beta = beta.value.flatten().T\n",
    "prescient_beta = np.ravel(prescient_beta).T\n",
    "\n",
    "pickle.dump(prescient_beta, open(str(datetime.datetime.now().strftime(\"%Y-%m-%d_%H-%M\")) + 'prescient-beta-v2', 'wb'))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def oracle_eval_wrapper(beta, args): \n",
    "    data = dict(args)\n",
    "    INDS = data['inds']\n",
    "    t_lo = data['t_lo']; t_hi = data['t_hi']\n",
    "    tau = np.clip(np.dot(beta, data['x_samp'].T) , t_lo, t_hi).flatten()\n",
    "    return np.mean( simulated_loss(tau,therapeut_dose[INDS],obs_INR[INDS]) )\n",
    "\n",
    "beta_d = np.random.rand(d)\n",
    "train_data['inds'] = trainind\n",
    "oracle_actual_oracle_res = minimize(oracle_eval_wrapper, x0 = beta_d, bounds = bnds , options={'disp':True,'gtol':1e-4}, args=train_data.items())  \n",
    "pickle.dump(oracle_actual_oracle_res.x, open('oracle_opt_beta', 'wb'))\n",
    "print oracle_actual_oracle_res"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "print oracle_actual_oracle_res\n",
    "print \"off pol oracle\"\n",
    "print oracle_res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Build direct estimate and use 'regress and compare' "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestRegressor\n",
    "\n",
    "X_w_dose = np.column_stack((X, sim_dose))\n",
    "X_w_dose.shape\n",
    "\n",
    "clf = RandomForestRegressor(n_estimators=10)\n",
    "\n",
    "clf = clf.fit(X[trainind,:],y=therapeut_dose[trainind])\n",
    "train_data['x_full'] = X_w_dose\n",
    "train_data['rf'] = clf\n",
    "\n",
    "d = X.shape[1]\n",
    "# Regression and compare \n",
    "# Iterate through dataset and choose dose maximizing dose response \n",
    "\n",
    "# n = len(trainind)\n",
    "# ts = np.linspace(0,80,160)\n",
    "# best_taus_for_x = np.zeros(n)\n",
    "# for i in range(n): \n",
    "#     if (i%500 == 0): \n",
    "#         print i \n",
    "#     drf_vals = [clf.predict( np.append(X_w_dose[trainind[i],0:d-1], np.asarray(t)).reshape(1,-1)) for t in ts ]\n",
    "#     best_tau = np.argmin(drf_vals)\n",
    "#     best_taus_for_x[i] = ts[best_tau]\n",
    "    \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "AttributeError",
     "evalue": "'numpy.ndarray' object has no attribute 'append'",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mAttributeError\u001b[0m                            Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-23-6f727c87310b>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mXwnames\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mXnames\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mappend\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'simdose'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;31mAttributeError\u001b[0m: 'numpy.ndarray' object has no attribute 'append'"
     ]
    }
   ],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Male?', 'White?', 'Asian?', 'Black?', 'Non-hispanic?', 'Age group',\n",
       "       'No age?', 'Height', 'Weight', 'BMI',\n",
       "       'Indication for Warfarin Treatment: 1',\n",
       "       'Indication for Warfarin Treatment: 2',\n",
       "       'Indication for Warfarin Treatment: 3',\n",
       "       'Indication for Warfarin Treatment: 4',\n",
       "       'Indication for Warfarin Treatment: 5',\n",
       "       'Indication for Warfarin Treatment: 6',\n",
       "       'Indication for Warfarin Treatment: 7',\n",
       "       'Indication for Warfarin Treatment: 8', 'Diabetes=0?',\n",
       "       'Diabetes=1?', 'Congestive Heart Failure and/or Cardiomyopathy=0?',\n",
       "       'Congestive Heart Failure and/or Cardiomyopathy=1?',\n",
       "       'Valve Replacement=0?', 'Valve Replacement=1?', 'aspirin=0?',\n",
       "       'aspirin=1?', 'Acetaminophen=0?', 'Acetaminophen=1?',\n",
       "       'Acetaminophen hi dose=0?', 'Simvastatin=0?', 'Simvastatin=1?',\n",
       "       'Simvastatin=0?', 'Simvastatin=1?', 'Fluvastatin=0?',\n",
       "       'Lovastatin=0?', 'Lovastatin=1?', 'Pravastatin=0?',\n",
       "       'Pravastatin=1?', 'Rosuvastatin=0?', 'Rosuvastatin=1?',\n",
       "       'Cerivastatin=0?', 'Amiodarone=0?', 'Amiodarone=1?',\n",
       "       'Carbamazepine=0?', 'Carbamazepine=1?', 'Phenytoin=0?',\n",
       "       'Phenytoin=1?', 'Rifampin=0?', 'Sulfonamide Antibiotics=0?',\n",
       "       'Macrolide Antibiotics=0?', 'Anti-fungal Azoles=0?',\n",
       "       'Anti-fungal Azoles=1?',\n",
       "       'Herbal Medications, Vitamins, Supplements=0?',\n",
       "       'Herbal Medications, Vitamins, Supplements=1?', 'Smoker=0?',\n",
       "       'Smoker=0?', 'CYP2C9 *1/*1', 'CYP2C9 *1/*2', 'CYP2C9 *1/*3',\n",
       "       'CYP2C9 NA', 'VKORC1 -1639 A/A', 'VKORC1 -1639 A/G',\n",
       "       'VKORC1 -1639 G/G', 'VKORC1 497 T/T', 'VKORC1 497 G/T',\n",
       "       'VKORC1 497 G/G', 'VKORC1 1173 T/T', 'VKORC1 1173 C/T',\n",
       "       'VKORC1 1173 C/C', 'VKORC1 1542 C/C', 'VKORC1 1542 C/G',\n",
       "       'VKORC1 1542 G/G', 'VKORC1 3730 A/A', 'VKORC1 3730 A/G',\n",
       "       'VKORC1 3730 G/G', 'VKORC1 2255 C/C', 'VKORC1 2255 C/T',\n",
       "       'VKORC1 2255 T/T', 'VKORC1 -4451 C/C', 'VKORC1 -4451 A/C',\n",
       "       'VKORC1 -4451 A/A'], \n",
       "      dtype='|S49')"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Xnames"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['Weight' 'VKORC1 -1639 A/A' 'VKORC1 -1639 G/G' 'Age group' 'BMI' 'Height'\n",
      " 'CYP2C9 *1/*1' 'Anti-fungal Azoles=1?' 'VKORC1 1542 G/G' 'VKORC1 2255 T/T'\n",
      " 'Congestive Heart Failure and/or Cardiomyopathy=0?'\n",
      " 'Acetaminophen hi dose=0?' 'Carbamazepine=1?' 'VKORC1 1173 C/C'\n",
      " 'Indication for Warfarin Treatment: 3' 'aspirin=1?'\n",
      " 'Indication for Warfarin Treatment: 8' 'CYP2C9 NA' 'aspirin=0?'\n",
      " 'Indication for Warfarin Treatment: 2' 'White?'\n",
      " 'Indication for Warfarin Treatment: 1' 'Simvastatin=0?' 'VKORC1 3730 A/G'\n",
      " 'Diabetes=0?']\n"
     ]
    }
   ],
   "source": [
    "d=X.shape[1]\n",
    "sort_inds = np.argsort(clf.feature_importances_)[::-1][:d]\n",
    "\n",
    "pickle.dump(sort_inds, open('important_features_fm_regr.p','wb'))\n",
    "print Xnames[sort_inds[0:25]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "pickle.dump(best_taus_for_x, open(str(datetime.datetime.now().strftime(\"%Y-%m-%d_%H-%M\")) + 'dm_best_treatments-RC-train.p', 'wb'))\n",
    "pickle.dump(best_taus_for_x_test, open(str(datetime.datetime.now().strftime(\"%Y-%m-%d_%H-%M\")) + 'dm_best_treatments-RC-test.p', 'wb'))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "best_taus_for_x = pickle.load(open('2017-05-16_15-41dm_best_treatments-RC-train.p'))\n",
    "best_taus_for_x_test = pickle.load(open('2017-05-16_15-41dm_best_treatments-RC-test.p'))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Optimizing direct method "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "bounds = [ (0, np.mean(therapeut_dose)/(np.sqrt(d)*np.mean(X_w_dose[:,i]))) for i in range(d) ]\n",
    "bnds = tuple(tuple(x) for x in bounds)\n",
    "def dm_wrapper(beta, args): \n",
    "    data = dict(args)\n",
    "    INDS = data['inds']\n",
    "    t_lo = data['t_lo']; t_hi = data['t_hi']\n",
    "    clf = train_data['rf']\n",
    "    tau = np.clip(np.dot(beta, data['x_full'][trainind,:].T), t_lo, t_hi).flatten()\n",
    "    counterfactual_X = np.column_stack((data['x_full'][trainind,0:d-1], tau))\n",
    "    reg_lambda = 3\n",
    "    return 100* (np.mean( clf.predict( counterfactual_X )) + reg_lambda * np.linalg.norm( beta, 1 )) \n",
    "\n",
    "\n",
    "\n",
    "beta_d = np.random.uniform(low = np.zeros(d), high = [bnds[i][1] for i in range(d) ], size = d)\n",
    "epsilon = np.ones(d)*1e-2\n",
    "dm_res = minimize(dm_wrapper, x0 = beta_d, jac= None, bounds = bnds , options={'disp':True,'gtol':1e-8, 'eps':np.ones(d)*1e-1}, args=train_data.items())  \n",
    "\n",
    "pickle.dump(dm_res.x, open(str(datetime.datetime.now().strftime(\"%Y-%m-%d_%H-%M\")) + 'dm_beta-v2', 'wb'))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.dot(dm_res.x, X_w_dose[trainind,:].T)\n",
    "\n",
    "print dm_res.x\n",
    "print len(dm_res.x[0:len(dm_res.x)-1])\n",
    "print Xnames[np.where(dm_res.x[0:len(dm_res.x)-2] > 0)[0]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from statsmodels.nonparametric.kernel_regression import KernelReg\n",
    "\n",
    "list_data = [ subframe[trainind,i] for i in range(d) ] + [sim_dose[trainind]]\n",
    "print list_data\n",
    "print len(list_data)\n",
    "print [len(list_data[i]) for i in range(len(list_data))]\n",
    "kernel_regr = KernelReg(endog = [y.reshape([len(trainind),1])], exog=list_data, var_type=['c']*len(list_data) )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd \n",
    "df = pd.DataFrame(np.hstack([subframe[trainind], sim_dose[trainind].reshape([len(trainind),1])]) )\n",
    "\n",
    "df.to_csv(\"train_data.csv\", header = list(subXnames)+[\"sim_dose\"],index=None)\n",
    "ydf = pd.DataFrame(y) \n",
    "ydf.to_csv('y.csv',header='y',index =None)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Discrete off-policy evaluation and optimization "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-73-ab81b900c228>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m     36\u001b[0m     \u001b[1;32mreturn\u001b[0m \u001b[0mGPS\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m     37\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m---> 38\u001b[0;31m \u001b[0mintegrated_propensity_scores\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mintegrated_gps\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;36m9\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msim_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mmin_T\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mmax_T\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[0;32m<ipython-input-73-ab81b900c228>\u001b[0m in \u001b[0;36mintegrated_gps\u001b[0;34m(bmi, T, min_T, max_T)\u001b[0m\n\u001b[1;32m     29\u001b[0m             \u001b[0mGPS\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mscipy\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mintegrate\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mquad\u001b[0m\u001b[1;33m(\u001b[0m \u001b[0mgenerated_Q_wrapper\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m5\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m21\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0margs\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbmi\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtheta\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtheta\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m)\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m     30\u001b[0m         \u001b[1;32melif\u001b[0m \u001b[0mT\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m<\u001b[0m \u001b[1;36m49\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m---> 31\u001b[0;31m             \u001b[0mGPS\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mscipy\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mintegrate\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mquad\u001b[0m\u001b[1;33m(\u001b[0m \u001b[0mgenerated_Q_wrapper\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m21\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m49\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0margs\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbmi\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtheta\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtheta\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m)\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     32\u001b[0m         \u001b[1;32melif\u001b[0m \u001b[0mT\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m>=\u001b[0m \u001b[1;36m49\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m     33\u001b[0m             \u001b[0mGPS\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mscipy\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mintegrate\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mquad\u001b[0m\u001b[1;33m(\u001b[0m \u001b[0mgenerated_Q_wrapper\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m49\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m80\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0margs\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbmi\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtheta\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtheta\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m)\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\scipy\\integrate\\quadpack.pyc\u001b[0m in \u001b[0;36mquad\u001b[0;34m(func, a, b, args, full_output, epsabs, epsrel, limit, points, weight, wvar, wopts, maxp1, limlst)\u001b[0m\n\u001b[1;32m    309\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[1;33m(\u001b[0m\u001b[0mweight\u001b[0m \u001b[1;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m    310\u001b[0m         retval = _quad(func, a, b, args, full_output, epsabs, epsrel, limit,\n\u001b[0;32m--> 311\u001b[0;31m                        points)\n\u001b[0m\u001b[1;32m    312\u001b[0m     \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m    313\u001b[0m         retval = _quad_weight(func, a, b, args, full_output, epsabs, epsrel,\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\scipy\\integrate\\quadpack.pyc\u001b[0m in \u001b[0;36m_quad\u001b[0;34m(func, a, b, args, full_output, epsabs, epsrel, limit, points)\u001b[0m\n\u001b[1;32m    374\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[0mpoints\u001b[0m \u001b[1;32mis\u001b[0m \u001b[0mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m    375\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0minfbounds\u001b[0m \u001b[1;33m==\u001b[0m \u001b[1;36m0\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m--> 376\u001b[0;31m             \u001b[1;32mreturn\u001b[0m \u001b[0m_quadpack\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_qagse\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mfunc\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0ma\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mb\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mfull_output\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mepsabs\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mepsrel\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mlimit\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    377\u001b[0m         \u001b[1;32melse\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m    378\u001b[0m             \u001b[1;32mreturn\u001b[0m \u001b[0m_quadpack\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0m_qagie\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mfunc\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mbound\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0minfbounds\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mfull_output\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mepsabs\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mepsrel\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mlimit\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-73-ab81b900c228>\u001b[0m in \u001b[0;36mgenerated_Q_wrapper\u001b[0;34m(T, bmi, theta, NULL)\u001b[0m\n\u001b[1;32m     17\u001b[0m     \u001b[1;31m# centeredaround mu_dose with scale stddev of dose\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m     18\u001b[0m     \u001b[0mZ_bmi\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;33m(\u001b[0m\u001b[0mbmi\u001b[0m\u001b[1;33m-\u001b[0m\u001b[0mmu_bmi\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m/\u001b[0m\u001b[0mstd_bmi\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m---> 19\u001b[0;31m     \u001b[1;32mreturn\u001b[0m \u001b[0mnorm\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpdf\u001b[0m\u001b[1;33m(\u001b[0m \u001b[0mT\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mloc\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmu_dose\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0mstd_dose\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msqrt\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtheta\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0mZ_bmi\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mscale\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msqrt\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m-\u001b[0m\u001b[0mtheta\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     20\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m     21\u001b[0m \u001b[0mmin_T\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mmin\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0msim_dose\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;32mC:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\scipy\\stats\\_distn_infrastructure.pyc\u001b[0m in \u001b[0;36mpdf\u001b[0;34m(self, x, *args, **kwds)\u001b[0m\n\u001b[1;32m   1593\u001b[0m         \u001b[0mcond\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mcond0\u001b[0m \u001b[1;33m&\u001b[0m \u001b[0mcond1\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   1594\u001b[0m         \u001b[0moutput\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mzeros\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mshape\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mcond\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'd'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1595\u001b[0;31m         \u001b[0mputmask\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0moutput\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m(\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m-\u001b[0m\u001b[0mcond0\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m+\u001b[0m\u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0misnan\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mself\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mbadvalue\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m   1596\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0many\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mcond\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m   1597\u001b[0m             \u001b[0mgoodargs\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0margsreduce\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mcond\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m+\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m+\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mscale\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "def discretize_dose(dose): \n",
    "    discretized_dose = np.zeros(len(dose))\n",
    "    for i in range(len(dose)): \n",
    "        if dose[i] < 21: \n",
    "            discretized_dose[i] = 0 \n",
    "        elif dose[i] < 49: \n",
    "            discretized_dose[i] = 1\n",
    "        elif dose[i] >= 49: \n",
    "            discretized_dose[i] = 2\n",
    "    return discretized_dose\n",
    "\n",
    "disc_therapeutic_dose = discretize_dose(therapeut_dose)\n",
    "disc_sim_dose = discretize_dose(sim_dose) \n",
    "\n",
    "def generated_Q_wrapper(T, bmi, theta, NULL): \n",
    "    theta = 0.5\n",
    "    # centeredaround mu_dose with scale stddev of dose\n",
    "    Z_bmi = (bmi-mu_bmi)/std_bmi\n",
    "    return norm.pdf( T, loc = mu_dose + std_dose*np.sqrt(theta)*Z_bmi, scale = np.sqrt(1-theta) )\n",
    "\n",
    "min_T = min(sim_dose)\n",
    "max_T = max(sim_dose)\n",
    "# y is still the same loss information ( distance from treatment ) \n",
    "def integrated_gps(bmi, T, min_T, max_T ): \n",
    "    theta = 0.5\n",
    "    GPS = np.zeros(len(T))\n",
    "    for i in range(len(T)): \n",
    "        if T[i] < 21: \n",
    "            GPS[i] = scipy.integrate.quad( generated_Q_wrapper, 5, 21, args=(bmi[i], theta, theta) )[0]\n",
    "        elif T[i] < 49: \n",
    "            GPS[i] = scipy.integrate.quad( generated_Q_wrapper, 21, 49, args=(bmi[i], theta, theta) )[0]\n",
    "        elif T[i] >= 49: \n",
    "            GPS[i] = scipy.integrate.quad( generated_Q_wrapper, 49, 80, args=(bmi[i], theta, theta) )[0]\n",
    "    if i % 500 == 0: \n",
    "        print i \n",
    "    return GPS\n",
    "\n",
    "integrated_propensity_scores = integrated_gps(X[trainind,9], sim_dose[trainind], min_T, max_T)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'disc_sim_dose' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-72-9dc4171cc2d7>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      6\u001b[0m \u001b[0mXtrain\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m----> 7\u001b[0;31m \u001b[0mttrain\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mdisc_sim_dose\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mastype\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mint\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      8\u001b[0m \u001b[0mgps\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0masarray\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mgenerated_Q_wrapper\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0msim_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;36m9\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mi\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m0.5\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m0.5\u001b[0m \u001b[1;33m)\u001b[0m \u001b[1;32mfor\u001b[0m \u001b[0mi\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      9\u001b[0m \u001b[1;31m# all 0-1\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'disc_sim_dose' is not defined"
     ]
    }
   ],
   "source": [
    "# Impute POEM \n",
    "import sys\n",
    "sys.path.append('./POEM-norm/')\n",
    "import DatasetReader, Skylines\n",
    "\n",
    "Xtrain = X[trainind,:]\n",
    "ttrain = disc_sim_dose.astype(int)[trainind]\n",
    "gps = np.asarray([generated_Q_wrapper(sim_dose[i], X[trainind,9][i], 0.5, 0.5 ) for i in range(len(trainind))])\n",
    "# all 0-1\n",
    "# gps = integrated_propensity_scores\n",
    "ytrain = y\n",
    "Xtest = X[test_mask,:]\n",
    "\n",
    "\n",
    "\n",
    "mydata = DatasetReader.BanditDataset(None,False)\n",
    "mydata.trainFeatures        = np.hstack((Xtrain.copy(),np.ones((len(Xtrain),1))))\n",
    "mydata.sampledLabels        = np.zeros((len(ttrain),max(ttrain)+1))\n",
    "mydata.sampledLabels[range(len(ttrain)),ttrain] = 1.\n",
    "mydata.trainLabels          = np.empty(mydata.sampledLabels.shape)\n",
    "mydata.sampledLoss          = ytrain.copy()\n",
    "mydata.sampledLoss         -= mydata.sampledLoss.min()\n",
    "mydata.sampledLoss         /= mydata.sampledLoss.max()\n",
    "# computed on training set \n",
    "mydata.sampledLogPropensity = np.log(gps)\n",
    "#ones_like vs ones_line? \n",
    "mydata.testFeatures              = np.hstack((np.ones_like(Xtest),np.ones((len(Xtest),1))))\n",
    "mydata.testLabels                = np.array([])\n",
    "mydata.createTrainValidateSplit(validateFrac = 0.25)\n",
    "pool = None \n",
    "coef = None\n",
    "\n",
    "maj = Skylines.PRMWrapper(mydata, n_iter = 1000, tol = 1e-6, minC = 0, maxC = -1, minV = 0, maxV = -1,\n",
    "                            minClip = 0, maxClip = 0, estimator_type = 'Vanilla', verbose = True,\n",
    "                            parallel = pool, smartStart = coef)\n",
    "maj.calibrateHyperParams()\n",
    "maj.validate()\n",
    "Xtest1 = np.hstack((Xtest,np.ones((len(Xtest),1))))\n",
    "rec = Xtest1.dot(maj.labeler.coef_).argmax(1)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "rec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "### POEM-norm\n",
    "\n",
    "mydata = DatasetReader.BanditDataset(None,False)\n",
    "mydata.trainFeatures        = np.hstack((Xtrain.copy(),np.ones((len(Xtrain),1))))\n",
    "mydata.sampledLabels        = np.zeros((len(ttrain),max(ttrain)+1))\n",
    "mydata.sampledLabels[range(len(ttrain)),ttrain] = 1.\n",
    "mydata.trainLabels          = np.empty(mydata.sampledLabels.shape)\n",
    "mydata.sampledLoss          = ytrain.copy()\n",
    "mydata.sampledLoss         -= mydata.sampledLoss.min()\n",
    "mydata.sampledLoss         /= mydata.sampledLoss.max()\n",
    "mydata.sampledLogPropensity = np.log(gps)\n",
    "mydata.testFeatures              = np.hstack((np.ones_like(Xtest),np.ones((len(Xtest),1))))\n",
    "mydata.testLabels                = np.array([])\n",
    "mydata.createTrainValidateSplit(validateFrac = 0.25)\n",
    "pool = None \n",
    "coef = None\n",
    "\n",
    "maj = Skylines.PRMWrapper(mydata, n_iter = 1000, tol = 1e-6, minC = 0, maxC = -1, minV = 0, maxV = -1,\n",
    "                            minClip = 0, maxClip = 0, estimator_type = 'SelfNormal', verbose = True,\n",
    "                            parallel = pool, smartStart = coef)\n",
    "maj.calibrateHyperParams()\n",
    "maj.validate()\n",
    "Xtest1 = np.hstack((Xtest,np.ones((len(Xtest),1))))\n",
    "rec = Xtest1.dot(maj.labeler.coef_).argmax(1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "rec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LogisticRegression\n",
    "\n",
    "imputed_Q = LogisticRegression(multi_class = \"multinomial\",solver='lbfgs')\n",
    "# Impute probability of low/medium/high simulated dose\n",
    "imputed_Q.fit(subframe[trainind], disc_sim_dose[trainind])\n",
    "\n",
    "recentered_loss = discretized_loss - np.mean(discretized_loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "def off_pol_disc_given_disc_evaluation(policy, **params):\n",
    "    THRESH = 0.000001\n",
    "    y_out = params['y']; x = params['x_samp']; h = params['h']; Q = params['Q']; n = params['n']; t_lo = params['t_lo']; t_hi = params['t_hi']\n",
    "    imputed_Q = params['imputed_Q']\n",
    "    if ('y_samp' in params.keys()):\n",
    "        y_out = params['y_samp'].flatten()\n",
    "    if ('T_samp' in params.keys()):\n",
    "        T = params['T_samp'].flatten()\n",
    "    else:\n",
    "        T = params['T'].flatten()\n",
    "\n",
    "    t_lo = min(T)\n",
    "    t_hi = max(T)\n",
    "    \n",
    "    T_binned = discretize_dose(T)\n",
    "    bins = range(unique(T_binned)) + 1\n",
    "    bin_means = [T[T_binned == i].mean() for i in range(1, len(bins))]\n",
    "    tau_bins = discretize_dose(params['tau'])\n",
    "    loss = 0\n",
    "    treatment_overlap = np.where(np.equal(tau_bins, T_binned))[0]\n",
    "\n",
    "    for ind in treatment_overlap:\n",
    "        # use a discrete probability \n",
    "        loss += y_out[ind]/max(Q_i,THRESH)\n",
    "    n_overlap = len(treatment_overlap)\n",
    "    if n_overlap == 0:\n",
    "        print \"no overlap\"\n",
    "        return np.inf\n",
    "    return loss/(1.0*n_overlap)\n",
    "\n",
    "\n",
    "def oracle_eval_wrapper(beta, args): \n",
    "    data = dict(args)\n",
    "    INDS = data['inds']\n",
    "    t_lo = data['t_lo']; t_hi = data['t_hi']\n",
    "    tau = np.clip(np.dot(beta, data['x'].T) , t_lo, t_hi).flatten()\n",
    "    return np.mean( simulated_loss(tau,therapeut_dose[INDS],obs_INR[INDS]) )\n",
    "\n",
    "\n",
    "beta_d = np.random.rand(d)\n",
    "train_data['inds'] = trainind\n",
    "oracle_actual_oracle_res = minimize(oracle_eval_wrapper, x0 = beta_d, bounds = bnds , options={'disp':True,'gtol':1e-4}, args=train_data.items())  \n",
    "pickle.dump(oracle_actual_oracle_res.x, open('oracle_opt_beta', 'wb'))\n",
    "print oracle_actual_oracle_res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Histograms of performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### hist results of giving mu_dose amount of treatment"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAosAAAGgCAYAAADCaS1/AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nK8nJopmZmZmV9P8ByCf1b0yZeTQAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1ffa0438>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAosAAAGgCAYAAADCaS1/AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMzMzs5qcLJqZmZlZTf8fmvhuM15dLOAAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1ffa0a20>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "plt.hist(np.abs(np.mean(sim_dose[trainind]) - therapeut_dose[trainind] ).T,range=[0,80],alpha = 0.5,label='best-in-class treatment policy')\n",
    "plt.hist(np.abs(sim_dose[trainind] - therapeut_dose[trainind]),range=[0,80],alpha = 0.5,color = 'g',label='original simulated dose' )\n",
    "plt.title(\"Histogram of training set performance of assigning mean of simulated dose\",y=1.08)\n",
    "plt.legend()\n",
    "plt.ylabel('frequencies')\n",
    "plt.ylim((0,1800))\n",
    "plt.xlabel('Absolute value of distance between recommended and therapeutic dose')\n",
    "plt.show()\n",
    "\n",
    "plt.hist(np.abs(np.mean(sim_dose[test_mask]) - therapeut_dose[test_mask] ).T,range=[0,80],alpha = 0.5,label='best-in-class treatment policy')\n",
    "plt.hist(np.abs(sim_dose[test_mask] - therapeut_dose[test_mask]),range=[0,80],alpha = 0.5,color = 'g',label='original simulated dose' )\n",
    "plt.title(\"Histogram of training set performance of assigning mean of simulated dose\",y=1.08)\n",
    "plt.legend()\n",
    "plt.ylabel('frequencies')\n",
    "plt.ylim((0,1800))\n",
    "plt.xlabel('Absolute value of distance between recommended and therapeutic dose')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "ValueError",
     "evalue": "operands could not be broadcast together with shapes (3171,24) (81,) ",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-19-3474e9310210>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mhist\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mabs\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0masarray\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mtrain\u001b[0m\u001b[1;33m*\u001b[0m\u001b[0mprescient_beta\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mflatten\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m-\u001b[0m \u001b[0mtherapeut_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mT\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mrange\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;36m80\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0malpha\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m0.5\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mlabel\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'best-in-class treatment policy'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      3\u001b[0m \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mhist\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mabs\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0msim_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m \u001b[1;33m-\u001b[0m \u001b[0mtherapeut_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mrange\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;36m80\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0malpha\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;36m0.5\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mcolor\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;34m'g'\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mlabel\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'original simulated dose'\u001b[0m \u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mtitle\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m\"Histogram of training set performance of best-in-class treatment policy\"\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0my\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m1.08\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mlegend\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mValueError\u001b[0m: operands could not be broadcast together with shapes (3171,24) (81,) "
     ]
    }
   ],
   "source": [
    "\n",
    "plt.hist(np.abs(np.asarray((train*prescient_beta).flatten()) - therapeut_dose[trainind] ).T,range=[0,80],alpha = 0.5,label='best-in-class treatment policy')\n",
    "plt.hist(np.abs(sim_dose[trainind] - therapeut_dose[trainind]),range=[0,80],alpha = 0.5,color = 'g',label='original simulated dose' )\n",
    "plt.title(\"Histogram of training set performance of best-in-class treatment policy\",y=1.08)\n",
    "plt.legend()\n",
    "plt.ylabel('frequencies')\n",
    "plt.ylim((0,1800))\n",
    "plt.xlabel('L1 deviation from 10% range of therapeutic dose')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.ravel(prescient_beta).T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## histogram results of off-policy evaluation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "TypeError",
     "evalue": "plot_training_set_hist() takes exactly 6 arguments (5 given)",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-20-83ff1e25b67d>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mplot_training_set_hist\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbest_off_pol_beta\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtherapeut_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msim_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtrainind\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m\"Training set performance of empirically optimal treatment policy\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[1;32mprint\u001b[0m \u001b[1;34m\"test set performance\"\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mplot_training_set_hist\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbest_off_pol_beta\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtest_mask\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtherapeut_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtest_mask\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0msim_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtest_mask\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m\"Test set performance of empirically optimal treatment policy\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mTypeError\u001b[0m: plot_training_set_hist() takes exactly 6 arguments (5 given)"
     ]
    }
   ],
   "source": [
    "plot_training_set_hist(best_off_pol_beta, X[trainind,:], therapeut_dose[trainind], sim_dose[trainind], \"Training set performance of empirically optimal treatment policy\")\n",
    "\n",
    "print \"test set performance\"\n",
    "plot_training_set_hist(best_off_pol_beta, X[test_mask,:], therapeut_dose[test_mask], sim_dose[test_mask], \"Test set performance of empirically optimal treatment policy\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "plot_training_set_hist(prescient_beta, X[trainind,:], therapeut_dose[trainind], sim_dose[trainind], 'Training set performance of best-in-class treatment policy')\n",
    "\n",
    "print \"test set performance\"\n",
    "plot_training_set_hist(prescient_beta, X[test_mask,:], therapeut_dose[test_mask], sim_dose[test_mask], 'Test set performance of best-in-class treatment policy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "print sum(np.abs(sim_dose[trainind] - therapeut_dose[trainind]))\n",
    "print sum(np.abs(np.dot(oracle_actual_oracle_res.x, train.T) - therapeut_dose[trainind]))\n",
    "print sum(np.abs(np.dot(oracle_res.x, train.T) - therapeut_dose[trainind]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.asarray((np.dot(best_off_pol_beta, train.T) - sim_dose[trainind]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "np.dot(best_off_pol_beta, train.T)\n",
    "\n",
    "plt.hist(np.abs(np.dot(best_off_pol_beta, train.T)  - therapeut_dose[trainind] ),range=[0,80],alpha = 0.5,label='empirically optimal treatment policy')\n",
    "plt.hist(np.abs(sim_dose[trainind] - therapeut_dose[trainind]),alpha = 0.5,range=[0,80],color = 'g',label='original simulated dose' )\n",
    "plt.legend()\n",
    "plt.title('Histogram of training set performance of empirically optimal treatment policy',y=1.08)\n",
    "plt.ylabel('frequencies')\n",
    "plt.ylim((0, 1800))\n",
    "plt.xlabel('L1 deviation from 10% range of therapeutic dose')\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "print test.shape\n",
    "plt.hist(np.abs(np.dot(oracle_actual_oracle_res.x, test.T) - therapeut_dose[test_mask]),range=[0,50],alpha = 0.7,color='r', label='best-in-class treatment policy')\n",
    "plt.hist(np.abs(np.dot(oracle_res.x, test.T) - therapeut_dose[test_mask]),range=[0,50],alpha = 0.7,color='g', label='empirically optimal treatment policy')\n",
    "plt.hist(np.abs(sim_dose[test_mask] - therapeut_dose[test_mask]),range=[0,50],alpha = 0.5, color = 'b' ,label = 'original simulated dose')\n",
    "plt.title(\"Histogram of out-of-sample test performance (deviation from therapeutic dose)\",y=1.08)\n",
    "plt.legend(loc='upper right')\n",
    "plt.ylabel('frequency')\n",
    "plt.xlabel('L1 deviation from 10% range of therapeutic dose')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Direct method regress and compare "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "plot_training_set_hist(dm_res.x, X_w_dose[trainind,:], therapeut_dose[trainind], sim_dose[trainind], 'Test set performance of direct method (random forest) policy optimization', 'random forest policy')\n",
    "plot_training_set_hist(dm_res.x, X_w_dose[test_mask,:], therapeut_dose[test_mask], sim_dose[test_mask], 'Test set performance of direct method (random forest) policy optimization', 'random forest policy')\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n",
      "100\n",
      "200\n",
      "300\n",
      "400\n",
      "500\n",
      "600\n",
      "700\n",
      "[ 43.27044025  43.77358491  44.77987421  22.64150943  20.12578616\n",
      "  27.16981132  64.40251572  27.67295597  28.17610063  23.64779874\n",
      "  20.12578616  26.66666667  16.10062893  38.74213836  40.25157233\n",
      "  25.66037736  23.64779874  21.13207547  38.74213836  18.11320755\n",
      "  39.24528302  38.74213836  35.72327044  30.6918239   51.82389937\n",
      "  16.10062893  22.13836478  16.10062893  21.13207547  39.24528302\n",
      "  31.19496855  39.24528302  38.74213836  35.22012579  38.74213836\n",
      "  24.65408805  38.74213836  36.22641509  28.17610063  44.27672956\n",
      "  27.67295597  44.77987421  22.13836478  22.64150943  33.20754717\n",
      "  47.79874214  39.24528302  25.1572327   17.10691824  16.10062893\n",
      "  33.20754717  36.22641509  38.74213836  16.10062893  40.25157233\n",
      "  30.6918239   43.77358491  55.34591195  44.27672956  23.64779874\n",
      "  23.64779874  20.12578616  16.10062893  16.60377358  33.20754717\n",
      "  29.68553459  22.64150943  42.7672956   43.27044025  52.83018868\n",
      "  16.10062893  42.26415094  34.71698113  28.17610063  27.67295597\n",
      "  17.10691824  20.12578616  28.17610063  16.10062893  16.60377358\n",
      "  11.06918239  25.66037736  19.11949686  16.10062893  12.0754717\n",
      "  16.10062893  16.10062893  12.57861635  15.09433962  21.13207547\n",
      "  20.12578616  27.67295597  14.59119497  20.62893082  10.56603774\n",
      "  23.14465409  27.67295597  25.1572327   14.59119497  24.1509434\n",
      "  14.59119497  18.11320755  17.61006289  32.70440252  23.14465409\n",
      "  10.56603774  24.1509434   16.60377358  24.1509434   16.10062893\n",
      "  14.59119497  16.60377358  19.11949686  27.16981132  23.14465409\n",
      "  15.59748428  19.62264151  16.10062893  16.60377358   9.55974843\n",
      "  16.60377358  13.58490566  31.69811321  16.10062893  14.59119497\n",
      "  15.09433962  16.10062893  21.13207547  12.0754717    9.55974843\n",
      "  12.57861635  16.10062893  13.58490566  12.57861635   8.55345912\n",
      "  14.59119497  17.61006289  25.66037736  13.58490566  24.1509434\n",
      "  20.12578616  12.57861635  29.68553459  12.57861635  15.59748428\n",
      "  16.10062893  15.59748428  12.57861635  15.09433962  14.59119497\n",
      "  16.10062893  16.60377358  10.56603774  19.62264151  14.59119497\n",
      "  12.0754717   19.62264151  21.63522013  14.59119497  23.14465409\n",
      "  13.58490566  19.62264151  15.09433962  12.57861635  17.10691824\n",
      "  15.59748428  17.10691824  24.65408805  10.56603774  31.69811321\n",
      "  30.6918239   11.57232704  23.14465409  16.60377358  15.59748428\n",
      "  23.14465409  14.59119497  16.10062893  18.6163522   11.06918239\n",
      "  15.59748428  10.56603774  23.14465409  16.60377358  20.12578616\n",
      "  18.6163522   29.68553459  11.06918239  24.1509434   44.27672956\n",
      "  17.10691824  19.62264151  10.56603774  20.62893082  12.57861635\n",
      "  11.06918239  16.10062893  16.60377358  23.14465409  15.09433962\n",
      "  17.61006289  12.57861635  27.16981132  16.60377358  12.57861635\n",
      "  29.68553459  16.60377358  16.10062893  29.68553459  10.56603774\n",
      "  17.61006289  16.10062893  25.66037736  20.12578616  25.66037736\n",
      "  25.1572327   16.10062893  13.08176101  21.13207547  16.10062893\n",
      "  16.10062893  24.1509434   13.58490566  24.1509434   25.1572327\n",
      "  23.14465409  26.16352201  14.59119497  12.57861635  22.64150943\n",
      "  53.83647799  12.57861635  12.57861635  18.11320755  19.11949686\n",
      "  20.62893082  24.1509434   24.65408805  24.1509434   15.09433962\n",
      "  21.13207547  25.66037736  24.1509434   35.72327044  27.67295597\n",
      "  18.11320755  43.27044025  43.27044025  11.06918239  35.72327044\n",
      "  42.26415094  31.19496855  40.25157233  22.13836478  39.24528302\n",
      "  38.23899371  28.17610063  21.13207547  19.11949686  38.74213836\n",
      "  42.7672956   27.16981132  40.25157233  31.69811321  40.25157233\n",
      "  40.25157233  35.72327044  42.26415094  45.28301887  32.20125786\n",
      "  33.20754717  20.12578616  38.74213836  34.21383648  31.19496855\n",
      "  31.19496855  39.74842767  40.25157233  46.28930818  23.14465409\n",
      "  43.77358491  39.74842767  41.25786164  44.77987421  48.30188679\n",
      "  17.10691824  38.74213836  23.64779874  31.19496855  38.23899371\n",
      "  32.70440252  43.27044025  42.26415094  42.7672956   25.66037736\n",
      "  42.26415094  27.67295597   9.55974843  39.24528302  42.26415094\n",
      "  24.1509434   31.19496855  40.25157233  27.67295597  33.20754717\n",
      "  24.1509434   20.12578616  23.64779874  27.67295597  36.22641509\n",
      "  51.32075472  43.77358491  42.26415094  16.10062893  38.74213836\n",
      "  32.70440252  17.61006289  33.71069182  37.73584906  40.25157233\n",
      "  27.67295597  36.22641509  40.25157233  17.61006289  37.2327044\n",
      "  18.6163522   40.25157233  25.66037736  29.68553459  14.59119497\n",
      "  24.65408805  17.61006289  44.77987421  24.65408805  38.74213836\n",
      "  33.71069182  40.25157233  55.34591195  40.25157233  38.74213836\n",
      "  37.73584906  21.63522013  37.2327044   35.72327044  35.72327044\n",
      "  32.20125786  40.25157233  21.13207547  44.27672956  43.27044025\n",
      "  40.25157233  36.22641509  23.14465409  28.67924528  11.06918239\n",
      "  33.71069182  39.24528302  17.61006289  35.22012579  23.14465409\n",
      "  23.64779874  38.74213836  37.73584906  35.72327044  48.30188679\n",
      "  44.27672956  37.73584906  44.27672956  25.1572327   40.25157233\n",
      "  23.64779874  37.73584906  27.67295597  22.64150943  33.20754717\n",
      "  33.71069182  30.6918239   46.79245283  40.25157233  34.71698113\n",
      "  27.67295597  36.72955975  25.66037736  40.25157233  39.74842767\n",
      "  25.66037736  25.1572327   28.17610063  31.69811321  23.64779874\n",
      "  33.20754717  40.25157233  28.17610063  29.68553459  44.27672956\n",
      "  34.71698113  18.11320755  20.12578616  32.70440252  31.69811321\n",
      "  16.10062893  32.70440252  33.20754717  33.20754717  23.14465409\n",
      "  35.22012579  20.62893082  34.71698113  29.68553459  30.6918239\n",
      "  34.71698113  32.70440252  25.66037736  23.14465409  25.1572327\n",
      "  16.10062893  34.71698113  32.20125786  31.69811321  27.16981132\n",
      "  33.20754717  29.68553459  32.70440252  23.14465409  25.1572327\n",
      "  33.20754717  29.68553459  20.62893082  16.10062893  32.70440252\n",
      "  32.70440252  16.60377358  33.20754717  32.70440252  23.14465409\n",
      "  32.70440252  18.11320755  18.11320755  35.22012579  39.74842767\n",
      "  33.20754717  15.59748428  44.27672956  56.35220126  39.24528302\n",
      "  53.83647799  27.16981132  44.27672956  27.67295597  54.33962264\n",
      "  33.20754717  44.27672956  38.74213836  12.0754717   23.64779874\n",
      "  48.80503145  44.27672956  23.14465409  31.19496855  42.26415094\n",
      "  27.67295597  44.27672956  25.66037736  25.66037736  27.67295597\n",
      "  19.62264151  29.68553459  15.59748428  29.68553459  44.77987421\n",
      "  47.29559748  29.18238994  35.22012579  33.20754717  39.24528302\n",
      "  27.67295597  36.22641509  39.24528302  25.66037736  25.1572327\n",
      "  44.77987421  20.12578616  27.67295597  35.72327044  21.13207547\n",
      "  35.22012579  29.68553459  44.27672956  16.10062893  21.13207547\n",
      "  15.09433962  37.73584906  28.17610063  27.67295597  42.26415094\n",
      "  12.57861635  23.14465409  42.26415094  18.11320755  33.71069182\n",
      "  44.27672956  53.83647799  44.27672956  18.11320755  35.72327044\n",
      "  43.27044025  41.25786164  34.21383648  60.37735849  42.26415094\n",
      "  16.10062893  46.79245283  44.27672956  41.76100629  31.69811321\n",
      "  29.68553459  43.27044025  38.74213836  37.2327044   27.67295597\n",
      "  37.2327044   30.6918239   40.25157233  34.71698113  31.19496855\n",
      "  35.72327044  44.27672956  28.17610063  29.68553459  34.21383648\n",
      "  38.23899371  28.67924528  27.67295597  39.74842767  32.20125786\n",
      "  40.25157233  43.77358491  39.24528302  44.27672956  34.21383648\n",
      "  27.67295597  44.27672956  48.80503145  43.27044025  33.71069182\n",
      "  40.25157233  36.22641509  39.74842767  30.18867925  38.23899371\n",
      "  30.18867925  38.23899371  35.72327044  41.25786164  38.74213836\n",
      "  42.7672956   27.16981132  51.82389937  35.72327044  42.7672956\n",
      "  29.18238994  16.10062893  48.80503145  32.70440252  21.13207547\n",
      "  43.77358491  20.12578616  44.27672956  29.68553459  32.20125786\n",
      "  18.11320755  20.62893082  43.77358491  30.6918239   36.22641509\n",
      "  37.2327044   27.67295597  29.68553459  42.26415094  38.23899371\n",
      "  31.19496855  17.10691824  32.20125786  27.67295597  33.71069182\n",
      "  16.10062893  20.12578616  61.3836478   22.64150943  35.22012579\n",
      "  20.12578616  44.27672956  41.25786164  44.27672956  33.71069182\n",
      "  39.74842767  16.10062893  25.66037736  19.11949686  54.33962264\n",
      "  54.33962264  39.24528302  40.25157233  27.67295597  29.68553459\n",
      "  32.70440252  32.70440252  34.21383648  44.27672956  31.69811321\n",
      "  18.11320755  27.67295597  44.27672956  43.27044025  35.22012579\n",
      "  18.11320755  44.27672956  38.74213836  12.57861635  32.20125786\n",
      "  30.18867925  29.68553459  34.71698113  35.72327044  43.77358491\n",
      "  23.64779874  24.1509434   32.70440252  40.25157233  44.27672956\n",
      "  27.67295597  35.22012579  29.68553459  45.78616352  27.16981132\n",
      "  35.22012579  33.71069182  47.29559748  34.21383648  30.18867925\n",
      "  17.10691824  33.71069182  23.64779874  33.71069182  33.20754717\n",
      "  33.20754717  23.64779874  35.72327044  44.27672956  33.71069182\n",
      "  49.3081761   20.12578616  35.22012579  26.16352201  44.27672956\n",
      "  33.20754717  34.71698113  26.16352201  33.71069182  18.11320755\n",
      "  33.71069182  33.71069182  35.22012579  41.76100629  22.64150943\n",
      "  37.2327044   37.2327044   27.67295597  35.22012579  37.2327044\n",
      "  24.65408805  28.17610063  35.72327044  38.74213836  38.74213836\n",
      "  24.1509434   17.61006289  17.61006289  30.6918239   38.74213836\n",
      "  28.17610063  20.12578616  38.74213836  32.20125786  24.1509434\n",
      "  21.13207547  15.09433962  27.16981132  43.27044025  20.12578616\n",
      "  21.13207547  37.2327044   37.73584906  14.59119497  16.10062893\n",
      "  34.71698113  38.74213836  24.1509434   37.73584906  39.74842767\n",
      "  25.1572327   55.34591195  37.73584906  56.85534591  37.2327044\n",
      "  38.74213836  38.74213836  24.1509434   17.10691824  43.27044025\n",
      "  32.20125786  35.22012579   9.55974843  22.13836478  29.18238994\n",
      "  37.73584906  25.66037736  36.72955975  44.27672956  39.24528302\n",
      "  25.66037736  58.86792453  27.67295597  42.26415094  43.77358491\n",
      "  30.6918239   39.24528302  29.18238994  24.1509434   33.71069182\n",
      "  39.74842767  33.71069182  41.76100629  27.67295597  38.23899371\n",
      "  35.22012579  25.66037736  38.74213836  30.6918239   27.67295597\n",
      "  51.82389937  36.72955975  28.67924528  47.79874214  28.17610063\n",
      "  44.27672956  44.27672956  25.66037736  46.28930818  43.77358491\n",
      "  25.66037736  25.66037736  20.12578616  33.20754717  41.25786164\n",
      "  27.16981132  41.25786164  43.77358491  42.26415094  27.67295597\n",
      "  28.67924528  44.27672956  28.17610063  38.74213836  44.27672956\n",
      "  28.67924528  31.69811321  51.82389937  44.27672956  54.33962264\n",
      "  42.26415094  43.77358491  36.72955975  34.21383648  27.67295597\n",
      "  44.27672956  37.2327044   27.16981132  27.16981132  51.82389937\n",
      "  43.77358491  28.67924528  27.16981132  25.66037736  51.82389937\n",
      "  39.74842767  48.30188679  44.27672956  51.82389937  36.22641509\n",
      "  44.27672956  38.74213836  33.71069182]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgkAAAFkCAYAAACq4KjhAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAAPYQAAD2EBqD+naQAAIABJREFUeJzsnXd4W9X5xz9HsmzL25ZH9nL2IgMCYSRA2GWTHyVQwiy7\npYyySgstFCgtZZXSAmHTlD0LDbOBkAHEIWTaiZPYTuI95KEtnd8fx3I8JNsatpVwPs/jR/G55577\nyrF1v/c97xBSSjQajUaj0Wg6YxhoAzQajUaj0cQmWiRoNBqNRqMJiBYJGo1Go9FoAqJFgkaj0Wg0\nmoBokaDRaDQajSYgWiRoNBqNRqMJiBYJGo1Go9FoAqJFgkaj0Wg0moBokaDRaDQajSYgWiRoNBqN\nRqMJSEQiQQhxmxDCJ4T4a7ux51rH2n99GLmpGo1Go9Fo+pO4cE8UQhwCXAGsD3D4I+BiQLR+7wz3\nOhqNRqPRaAaGsDwJQogU4GXgcqAhwBSnlLJaSlnV+mWNxEiNRqPRaDT9T7jbDU8A70spPw9y/Ggh\nRKUQYqsQ4u9CiKwwr6PRaDQajWaACHm7QQhxHjADODjIlI+AN4GdQD5wP/ChEGKuDNCXWghhAU4E\ndgGOUO3RaDQajeZHTCIwClgmpayN9uIhiQQhxDDgEeA4KaU70Bwp5Wvtvt0khNgAFANHA18EOOVE\n4JVQ7NBoNBqNRtOBC4B/RXvRUD0Js4EcoEAI4Q9KNALzhBDXAQmdvQVSyp1CiBpgLIFFwi6Al19+\nmUmTJoVozv7FDTfcwMMPPzzQZvQLP5b3qt/ngYV+nwcWP4b3uWXLFn72s59B67002oQqEj4FpnUa\nex7YAjwQZDthGGAByoOs6QCYNGkSs2bNCtGc/Yv09PQD/j36+bG8V/0+Dyz0+zyw+LG8z1b6ZLs+\nJJEgpWwBNrcfE0K0ALVSyi1CiGTgLlRMQgXKe/AnoAhYFhWLNRqNRqM5QPG5fKwZuwZnWc+VA+aW\nz+1ze8Kuk9CO9t4DLzAdWAxkAHtR4uB3wWIYNBqNRqPRKJoKmnCWORn1h1EkDEnodm5cWpy6y/Yh\nEYsEKeWx7f7tAE6KdE2NRqPRaH6MNK5sxGA2MOK2ERhMA985YeAt+BGxaNGigTah3/ixvFf9Pg8s\n9Ps8sNgf36d1pZXUQ1JjQiAAiACxhv1rgBCzgLVr1679MQWYaDQajUbTASklq4asYtDFgxhz/5he\nnVNQUMDs2bMBZkspC6JtU2xIFY1Go9FofuQ4Shy4KlykHZE20Ka0oUWCRqPRaDQxQOPXjQCkHaZF\ngkaj0Wg0mnZYV1oxTzATnx0/0Ka0oUWCRqPRaDQxQOPKRtIPTx9oMzqgRYJGo9FoNAOMp8lD8w/N\npB0eO1sNoEWCRqPRaDQDTtM3TeBDexI0Go1Go9F0xPq1lbiMOJImJg20KR2IRllmjUaj0Wh+lEiv\npHl9M9ITWc2h+k/qSZubhjCInif3I1okaDQajUYTJpUvV7L14q1RWWvMn8bglZLTN2ygzNlzg6fP\nDzooKtftDi0SNBqNRqMJk7r/1pEyI4WJL0yMbCEDJE1MYm1TEx/W1fGzvDyy4rq/Rccb+j5iQIsE\njUaj0WjCQEpJ/ef1DL50MCnTU6Ky5if19aQYjTw7YQKmfhABPTHwFmg0Go1Gsx/SsrEFd5WbjAUZ\nUVvz0/p6js7IiAmBAFokaDQajUYTFvWf1SMSBOlHRCdt0eb18rXVyvGZmVFZLxpokaDRaDQaTRg0\nfNZA+uHpGM3GqKy3wmrFJSXHaZGg0Wg0Gs3+i8/jo2F5A5kLondD/7S+niHx8UxKip1aCVokaDQa\njUYTIk3fNuFt8kY1HuGT+nqOy8xEiNiplaBFgkaj0Wg0IVL/WT3GNCOpB6dGZb1ql4vvm5tjaqsB\ntEjQaDQajSZkGj5rIGN+Boa46NxGP29oAGBBjIkEXSdBo9FoND8afG4frkpXRGtIl8S60kr+g/lR\nskrFI0xOSmJIQkLU1owGWiRoNBqN5kfD1ku2UvVKVVTWyjw+k5VWK+ds2oTb54toLavXyzVDhkTF\nrmiiRYJGo9FofjS0bGzBcrqFIVdHdkOOy4gjeXIyK0tLsXo83DVqVETrGYDz8/IiWqMviEgkCCFu\nA+4DHpFS3thu/A/A5UAG8DVwtZRyeyTX0mg0Go0mUpylTnJ/movlJEtU1itxOhmTmMitI0ZEZb1Y\nI+yICyHEIcAVwPpO47cC17UemwO0AMuEEPER2KnRaDQaTUR4mj146j0kjkiM2pq7HA5GJkZvvVgj\nLJEghEgBXkZ5Cxo6Hb4euEdK+YGUciOwGBgCnBmJoRqNRqPRRIKzTLVfThgRveDAEoeDUVokdOEJ\n4H0p5eftB4UQo4FBwGf+MSllI7AGmBuukRqNRqPRREqbSBgeXZFwIHsSQo5JEEKcB8wADg5weBAg\ngcpO45WtxzQajUajGRAcpQ4QkDA0OiKhwe2m0evVIsGPEGIY8AhwnJTS3TcmaTQajUYTfZylTuIH\nx2MwRacAUolTeSZGxlhtg2gSqidhNpADFIh9xaWNwDwhxHXAREAAeXT0JuQB67pb+IYbbiA9vWO7\nzUWLFrFo0aIQTdRoNBqNpivOMmdUgxZLHA6AfvMkLF26lKVLl3YYs1qtfXrNUEXCp8C0TmPPA1uA\nB6SUO4QQFcAC4AcAIUQacCgqjiEoDz/8MLNmzQrRHI1Go9Foeoej1BH1oMV4IciL75/kvUAPzgUF\nBcyePbvPrhmSSJBStgCb248JIVqAWinlltahR4A7hRDbgV3APcBu4N2IrdVoNBqNJkycpU5SZqZE\nbb1dDgcjEhMxxFDXxmgTjYqLssM3Uj4ohEgC/okqpvQVcLKUMrJi2RqNRqPRhImUEkeZI+rbDX2a\n/mi1wqGHQm1t4OMbN/bdtVuJWCRIKY8NMHY3cHeka2s0Go1GEw3c1W6kU0Y3/dHp5KDk5Kit14VV\nq6CwEO64A1ICeED68tqt6N4NGo1Gozng8ddIiLYn4XRLdMo7B2TVKsjOhnvvhQHa0ohOHohGo9Fo\nNDGMo1RlIkQrcNHm9VLtdvdtZsPq1XDYYQMmEECLBI1Go9H8CHCWOjEkGjBlm6KyXmlfpz/6fLBm\njRIJA4gWCRqNRqM54HGUOUgYnoCI0lP5Lr9I6KtCSoWFKnBRiwSNRqPRaPoWZ6kz6kGLRmBYX4mE\n1avVNsMhh/TN+r1EiwSNRqPRHPD0RbXFoQkJxBn66Da6ejVMmQJpaX2zfi/RIkGj0Wg0Bzx9UW2x\nz4MW5w5882QtEjQajUZzQONz+XCVu/afFtFNTapQ0gDHI4AWCRqNRqM5wHHudYKMco0Ep7Pvgha/\n/VZlN2iRoNFoNBpN3+IsVYWUouVJcPl87HU6+86TsHq1ikWYOLFv1g8BXXFRo9FoNAc0/mqLCcMT\nkFJi9/kiWm+Xw4GPPqyRsHq16tnQV0GRIaBFgkaj0WgOaBylDuKy4ohLieOyrVt5tqIiKuvmm81R\nWacDHo8SCVddFf21w0CLBI1Go9HsV9iKbBQcWoDX7u3VfOmWpBykGiT9r6GB0y0WfpqbG5ENWXFx\nfSMS/vQn1fXxzDOjv3YYaJGg0Wg0mv2Khi8b8Fg9jH1kbK8j69IOTaPF62WHw8GdI0dyfl5e3xoZ\nDuvWwd13w223waxZA20NoEWCRqPRaPYzWn5owTzezLBfDgvpvG8aGwGY1g8tlkPG6YTFi1UBpbvu\nGmhr2tAiQaPRaDT7Fc3rm0mZnhLyeRtbWhDA5GiKhJYWeO01aO3lEDYrVkBhIfLbb7nr3nspLS3t\n8ZRHH300smv2Ai0SNBqNRrPfIKWkeX0zI349IuRzN7S0kG82k2Q0Rs+gX/0KnnkGTBF2lzQa4aGH\n2JGSwj333MP06dNJTU3t9hSPxxPZNXuBFgkajUaj2W9wljnxWr0kHxS6N2BjS0t0txpWr1YC4Ykn\n4JprorLkV88/jxCC//3vf2RmZvY4v6SkJCrXDcbAJ2FqNBqNRtNLmtc3A7RlK4TChubm6IkErxeu\nvRZmzoQrr4zOmsCKFSuYOnVqrwRCf6A9CRqNRhNr7N2rvgIhRPhjvZ0bFweTJwdfYwBpXt9MXEYc\nCcNCq55Y7XJR6XYzNVoi4amnoKAAVq5UWwVR4quvvuK4446L2nqRokWCRqPRxAoOB/zxjypX3u0e\nWFtuvRUeeGBgbQhAyw8tJB+UjAhRwGxsaQFg2g8/wDvvRG7IK6/ApZdGtVNjVVUVRUVF3H333VFb\nM1K0SNBoNJpYYNUquOgi2LULbr8dzjqr6xwpozsWbPxf/4JHHlHu9OHDuzW7v2le30zWSVkhn7eh\npYUEIRi7cCGkp0Ok7vw5c6Iuor7++msAjjrqqKiuGwlaJGg0Gk0scPHFkJoK69fDpEkDa8vEifDy\ny/D736vAvBjB2+LFvs1Oyi3hpT9Oqq0lzmhURYuyQhcafc1XX33FyJEjGTYstPoPfYkOXNRoNJqB\npqYGiorgppsGXiCAEit33gnPPQeFhQNtTRstm1pAElZmw4bKSqZ99x384Q8xKRBABS3GkhcBQhQJ\nQoirhBDrhRDW1q+VQoiT2h1/Tgjh6/T1YfTN1mg0mgOI1avVaxT3tyPmqqtg2DD47W8H2pI2mtc3\ngwGSp4QmEqTPx0a7nak2W1QzEaJJc3MzBQUFHHnkkQNtSgdC3W4oA24FtgECuBh4VwgxQ0q5pXXO\nR63j/qgSZ+RmajQazQHM6tWQmwujRg20JftISFDbDZdcoqL4+6uXgJRqiyOAB6P5s6kkZVgw/vbW\nkJYssdloPvdcpp12msrciEHWrFmD1+uNOU9CSD8tKeV/Og3dKYS4GjgM8IsEp5SyOhrGaTQazY+C\n1auVFyHWUg4vvFC55x9/XG099AfPPgtXXAHjxoGho7O7pXQCyXFb4YMPQlpyw7RpAEw74oiomRlt\nVqxYQVZWFhMnThxoUzoQtqQSQhiAc4EkYGW7Q0cLISqBeuBz4E4pZV1EVmo0Gs2BitcLa9bAb34z\n0JZ0xWiEn/8c7rkHHn4YMjL69nobN8IvfoG87Ap8j/+94zEJzUNWMuLWEXD7r3tcqtzp5H8NDUjg\nw9pa0mtrGZoQWm2F3uB0Ovn4449xR5iy+sEHH3DkkUdiMMRWqGDIIkEIMRVYBSQCTcBZUkq/X+gj\n4E1gJ5AP3A98KISYK2Ww/BuNRqP5EbN5MzQ3x1Y8QnsuuQR+9ztVF+Daa7uf29gIb76phE9ncnLg\nqKOCBw22tOA5ZzG7U6+l7I3T8S75KuC0lJndZzZIKXmlspJfbN9OQ7veBqdbLCHXVugNr732GosX\nL47KWk8++WRU1okm4XgStgIHAenAQuBFIcQ8KeVWKeVr7eZtEkJsAIqBo4Evulv0hhtuID09vcPY\nokWLWLRoURgmajQazX7CqlXqif3ggwfakjbe3fouW2q2tH1/1hETyPzrvfxrlosrDr6SJFNSwPOc\n9zzJ7r/sRBKsAuEnSiwEEAq++haqqn6PNz6VIVcPIm1OWpc5BrOBrOODZyY0ejxcsnUrb9XUcEFu\nLg+NHUtaazXEhD56Qt++fTuDBg1i8+bNEa1jMBi63AM7s3TpUpYuXdphzGq1RnTdnghZJEgpPcCO\n1m/XCSHmANcDVweYu1MIUQOMpQeR8PDDDzOrvwJjNBqNJlZYvRqmT4doNh6KAK/Py3lvnodRGEmM\nSwRgbb6L15c38e8lNzImK5/TJ5we8NzKl/dSJs4laXKAp323W7VVbm6BsgCueQF5P0lgxFNzSRgS\n3rbAcxUVfFBby+uTJ7MwNzesNUKlpKSEUaNG9UuvhUAPzgUFBcyePbvPrhmNME8DEPB/VAgxDLAA\n5VG4jkaj0Rx4rFoFxxwz0Fa0UdZYhsPj4KMLPuKksa0Z7j4f8rMxXFlQQnlTkI/zLVuwV5hIGSM5\neOOc/jO4HcV2O2PN5n4TCKBEwsiRI/vtev1NqHUS7hNCHCWEGCmEmCqEuB+YD7wshEgWQjwohDi0\n9fgC4B2gCFjWB7ZrNBrN/k19PWzdGlPxCIU1KsRsvGX8vkGDAfHzn3PeRqiv2Bn4xNdfx2YciXlW\n/92gO1PqcDAyMbFfr6lFQkdygRdQcQmfArOBE6SUnwNeYDrwLlAIPA18C8yTUg5wpxKNRqOJQdas\nUa+HHTawdrSjsLaQBGMCI9M73fguvRSDFMx4Icgz3+uvY48fQ9Kk0EsmR4sSp7NfRYLX66WsrOyA\nFgmh1km4vJtjDuCkYMc1Go1G04nVq8FigbFjB9qSNgprChmbNRajoVPw4eDBvHZGPoveXq8yMiZP\n3ndsyxY8G4txkYx5vLl/DW5HicPBuTk5/Xa98vJyPB7PAS0SYishU6PRaH4seDzw1ltwxBExVUSp\nqK6ICdkTAh776rzD2WtJgGuu6dg98vXXsSep7Ymk8YEzH/qaJo+Heo+nXz0Ju3btAtAiQaPRaDRR\n5p//VMWD7rxzoC3pQGFNIeOzxgc8lpM5jDvOToPly+Gll/YdeP11bDNOA8A8bmA8CSUOB0C/ioSS\nkhJ1TS0SNBqNRhMSW7bAL34BOwME+lVXK3Fw+eVwyCH9b1sQWlwtlDWWBfUkDEoZxBtDrcif/hRu\nuAHOOw/OOQc2bsQ+8ghMOSZMmaZ+tlpR6lRtgkb2QVXFYJSUlJCVlUVqamq/XbO/0SJBo9Fo+oJX\nXoG//U21fv7d78Bm23fsjjvUFsN99w2cfQHYVrcNgAmW4CLB6XVivf9uOPZY1eLaaoWf/hSbHDbg\n8QhxQjC4n0XCgexFgOjUSdBoNBpNZzZvVvEGRx8Nf/oTPPaY+n7qVFiyRAmI7OyBtrIDRbVFAN16\nEgDKUyQZr7/e4Zh9zlqSpw5cQagSh4NhCQkY+zG+48cgErQnQaPRaPqCzZth9my491619XDTTaqn\nwT/+oUowX3nlQFvYhcKaQixmC1nmwKWP/SKhormiw7iUEluRbWA9CU5nv241wI9DJGhPgkaj0UQb\npxO2b1f79gBjxsBvf6v+7W9+ZAzW32DgKKwtDOpFABicOhjoKhLc1W68Vu+AZTaAKqQ01tx/IkVK\nqUWCRqPRaMJg2zYlBtrXEvATg+LAT2FtIdNypwU9nhKfQrIpmfLmjqWZbUUq3mKgYxIW9EP/BD/V\n1dXY7XZGjRoV9bWrqqCurud5/VFeQ4sEjUajiTb+joCBREKMIqWkqLaIhZMWdjtvUMqgLp4Ee5Ed\nBJjzB0YkuHw+9rpcjOjnoEWIfvqjw+HhyScvJCNjT49zU1Pfjuq1A6FFgkaj0USbzZshL09VU9xP\nqGyppNHZ2LFnQwACiQRbkY2EEQkYzQPjJdntdCI5MGokbNmylPnz/43VuoiMjO5v0ZmZRioro3r5\nLmiREENICT5fTHsjNZr+xWaDW2+FCy+EOQPTWTAsNm3ar7wIsK+xU3cxCaDiEgJ5EsKJR3D7fACY\nDJHF0JcOUCGl5ORksrICB3mGg5Q+6uvvZ803J3PqwicYNrT7+YmJaVG7djC0SIghfv97eO89WLs2\npqq0ajQDQ0sLnHYafPEFNDfvXyJh8+aYav/cGwprCzEIA/mZ+d3OG5Q8qE1Q+LEV2cg8JrR4gGqX\ni8PXrWNeejpLJk4M2d72lLQWUho+ADUSRBQ/rGtq3sFg2MLLlHDHKz2Lj/KbgrTtjiJaJMQIUsIL\nL8CuXfDtt/vX56FGE3Wam+EnP4GCAlVn4MsvB9qi3uN2Q1ERXHvtQFsSEkW1RYzOGE1CXPc32kEp\ngzoELkqvxL7dzpArh/T6Wg6vlzM3bmS73c5up5OHx44lLS7821GJw0GuyYS5H92w0c5skFKybutN\nbKoXlOw8mjeuv7THczISM9jL3qjZEAgtEmKEdeuUQIiLgxdf1CJB8yNGSlXqd906+Phj2LMH/u//\n1OvQHvyvscD27ap5UwTbDdtqt7GzIUA55z5k9e7VPcYjgBIJNbYa3F43JqMJR5kD6ZS93m7wScnF\nW7dS0NzMm1OmsHDTJt6orubSwYPDtr3E4ejXrQZQzZ0OP/xw9jbt5d8b/43H54loPVfTCo5M3MWy\nDScyb+87nDN5YMpbdyZmRML990PnDp9HHgnnnz8w9vQ3b74JWVmweLHqm/LXv0J8/EBbpdEMAB99\npMTB++/D3Lm0RWZ99ZXqFRDrRJjZ4PV5OfzZw6mx1UTRqN5x51E9N5salDKI3IZcVllW4Wv2QWsz\nSPMEM+ubm7lp+3Y87TtEdqLJ66WguZk3pkzh7JwcFmRm8lJl5X4nEkpKSjjouIOY9uQ0WlwtJJmS\nGJ3k4dyhjrC2i8cneWmUQ9i19ENOPCF26hzGjEjYsAGS2gnRqirVRXXRogN/f15KeOMNOOMMuOwy\neOQR9Tl5xhnq+CefqG1ZPxdcAFOmDIytGk1IOJ0qGrczCQkQKFjN51N9DebNU9sNoLIEJkxQWw77\ni0iwWLo+9fSSb/d+S42thvfOe48Zg2ZE2bjuGZrWs6dmcOpghtUNw2f1Mfre0cRZ4ojPicc8yszr\nO3bwbVMTp/dQbvrG4cM5p/Xnc2FeHhdt3RrRjb7U6WRGSkqv5m6t2crKspVhXcePrcmG1WrlpZKX\nOHPOmTxz2jNYkizs3Hk3ZWUPkpm5IIxVBcOG3caunQbyuw8L6VdiRiT8618wa9a+7995B846C/bu\n3T88jJGwaZPawvzrX1VZ95kz1ZbDGWfAp5/CKadAbi4kJsLu3dDQAH//+0BbrdH0wKefwgknKBXc\nmUMOgRUrurrLXn0V1q+Hr7/u+HRw1FH7T1zCpk1KxYf5dPNx8cekJ6Rz8riTiTPEzEd0G4NSBpFm\nU1H1Q64dgiljn1u8oLmZo9LTeWnSpF6vd3Z2NlcbDLxSWckdYezx+6Sk1OFgRC8FxnmPn8f6NetD\nvk4HWtTLLT+5hQfOfaAteNHtriQpaQLTpr0f1rJlZUpXa5HQC2bPVq/ffXfgi4Q334S0NDjuOPX9\n4sVwyy3Ku3rOOWr8/fdVvMLxx6vGaxpNzPPDD0rZPvNMx/HGRhXU9/jjqp+BH7dblS4+7TQ4/PCO\n58ybp9aprY392gP+xk5hsqx4GceNOS4mBQJATlIOafY0pFESl77PRikla5uauHJI7wMYAVLi4jgn\nJ4cXKyq4fcSIkLMFqlwunFL2ygthc9tY/9R6jDVGEhMi257IGp7F7Wfd3sFel6sSkykv7DWLi9Wr\nFgkBqK5WXgM/BoN6el67dp/b/UDlzTfV56I/e2fRIrj5ZliwACZOVA9X/sBfi0X9rDSamKeiAoYM\nCRxYtHGjyvm94AIYpJoG8fTTsGMHvB2gity8eep1xYrY/kDweKCwEH7+87BOb3A0sGb3Gp78yZNR\nNix6mIwmBnkH4UnxdLhB7nW5qHK7mZ2aGvKaF+bl8VJlJa9VV4fcf2FLi3qs701zp/e/eR8q4L5/\n3MctV94Ssp094XJVYTaHf4cvLlYOqNGjo2hUhMSMSDjppK5jmZlKJBzIFBWpeIzf/37fWF4enHoq\nfPMN/Oc/ysvgJztbNZTTaGKeigoIFoz2hz/Av/8Nt90Gzz2n2ijfdBNccglMC9A7YORIGD5cbTnE\nikhwu5UoaM/27eByhR20+NmOz/BKLyfknxAFA/uOQe5B2FPtHcbWNjUBMKuXsQHtOTYzkxEJCZzn\nD/oMEZMQjO6FuHjx9RfBAJefe3lY1+kJt7uS9PS5YZ9fXAzDhu17YIwFYkYkPPZYx2YVy5apse++\nU1uaB2rw4iOPQHIynHhix/GXXlL9YTIyOo5bLMrjqtHEPOXl+7wEncnKgj/+Ea66SmUv/Pe/SiQ8\n8EDw9ebNU3twsUBRkcq8CNaFJ8zI4mXFy5hgmcDIjNjuLGhxWWhJaukwtrapiRyTiWFh3OGMQvDt\n7NnsbS2KFCpZJhPpvaizsPrz1WRPyiYrM3pVEtsTje2GWNpqgBgSCUcc0TFwMTMTHn1UZTns2aPU\n1YHGe+/Bk0/CE090zOwACOaxs1hUTMKBLJw0BwgVFWq/LBiXXw7//CcsX64ilxct6n69efOU96Gp\nKfgfSH9x771gNsMrr3Q9lpcXXBx1g5SSj4s/5owJMeIp6YYMRwbWRGuHsYLmZqY4HNx11114OntY\neskFF1zAlD5K3aqz1lG3uY7Trz29T9b3eu14vU3Ex0cmEmb0b0JLj8SMSOjMQQepuASfT205HGgi\nYc8euPRS5Tm9+uren5edraJfbTblgdBoYpbuthtANSlZtgzsdhgxouf1jjpKude+/jrw/mR/sX27\nEgePPhrVQi5FtUWUWEs4ceyJPU8eYFJsKZQllXUYW9vUxPSvv+aee+5h9OjRNDobaXQ2th03x5mx\nJAUPOq2uruabb77h008/7RObn3rtKfDCxede3Cfru91VAMTH54a9RnGxClaPJUISCUKIq4CrgVGt\nQ5uAP0gp/9tuzh+Ay4EM4Gvgainl9lANM5th0iQVx/Tdd7GzDRkJ5eXgcCgvwOWXq8DvJUtC8wj4\nA7tra7VI0MQwbrdyefX0RB1KLYGJE2H8eJXlMJAi4b77lLfg8ujuay8rXka8MZ75I+dHdd2+IKkl\nier0aqSUCCEodzopd7mYZ7ORl5fHtS9fy82f3MzCyQsZmT6S7XXbeb/ofYpuLCIvJfCT9tKlSzn/\n/PPZuHEjU6dOjbrNb7z9BiJXcNrc06K+NqitBiDs7Ya6OpXePmqUekhubTLZLYWFPc+JlFA9CWXA\nrcA2QAAXA+8KIWZIKbcIIW4FrgMWA7uAe4FlQohJUkpXqMbNnq3qAsRi8KKUcPLJUF+vHnDmz1e1\nX4I1M1u1qmNWlxDw2WehZ3P5a5TU1PTu4UujGRD8VRLDcLsHRQgVt3DVVeppvn0QU3+xc6cqYvKX\nvyiVH0U+Lv6YI0ccSXJ87Kt/U6OJutF1NLuaSU1IpaC5GYCk5makWXLzJzdzx5F3cO+x9yKEoMZW\nw6C/DOLNLW9yzSHXBFxz4cKF3HzzzTz22GM89dRTUbXX6/Wy4esNjJw3ss9SS10uvychPJHgT380\nmVT28C9Qd2fTAAAgAElEQVR+0fNnfBgxoiET0k9LSvmfTkN3CiGuBg4DtgDXA/dIKT8AEEIsBiqB\nM4HXQjVu1izl1Vu7Nvb24FeuVJ7SE09UKYoPPaTiC666KvD8detUGuN//6vex9ChqohcqLT3JGg0\nMUtFayvhaIoEUEVEfvtbVXmsPyqKNTSofUEp1ZPAypXqj/CKK6J+qW112zh13KlRXzdUCpqaeGz3\nbq4cMoS56eldjkspMTYaaTQ3csLLJ5AYl0hx6pEY0g7j5ZVP4fa5uXv+3fxu/u/aUiSzk7I5bsxx\nvLbptaAiwWQycc0113Dvvfdy3333kd1D1cZQWLV6Fa5GF8efdHzU1uyM2+33JIRnt18kNLbu0Nx1\nV2yUBAlbUgkhDMC5QBKwUggxGhgEfOafI6VsFEKsAeYSpkjwelXw4u7dKgMqVliyRLmFPvxQ3fSP\nPlqVmw8mErZtgzFjVO2DSGjvSdBoYha/SIigHn9AEhPVI9Yf/6jyhsMsfdwrHA61z7lhA0yfrtI1\nnU7485+7RhpHgRpbDTnJffN+GtxuVjY2djvH5fPxXEUF77U+gUgIKBK8zV7wwJwpc9icqVIWi1PH\nMsxgIykxn4xxGdx19F1dzvvplJ9y2XuXsbdpL0NSAxdcuuKKK7jnnnt4+umnuf3220N8l/twu908\n+OCDvPXWW0gp2VuxF5Lg/JP7rhmQy1VFXJwFgyG8xkzFxSrpp7hYlReJBYEAYYgEIcRUYBWQCDQB\nZ0kpC4UQc1G/V5WdTqlEiYeQaR/luXZt7IiEpiZ47TVVFdG/vTB/vnqw8fkCbzls2wbjxkV+7eRk\nVclWexI0MU15uVLPfXETv/pq1RHuiSfg7rvVWGOjioMAdQMPsSBPF7xeVejp229VeenDD1eiYdOm\niMPPa91ubF4vw9ttV3h8HursdWQnRe/puT137drFY3v29DhvnNnMixMnsrapibdqatpiDtrjrlU/\n56uOv4qs41Uq4fBVq1iUm8tyXzqTRgQuyXzmxDO58oMreWPzG/zy0F8GnJOTk8MFF1zAE088wc03\n34zJFPoNd/PmzVx00UWsW7eOQ044hMTkRGx5NmrTajlsxGEhr9dbXK7KiDMb8vP3adJYIRxPwlbg\nICAdWAi8KISYF6khTU3fY7Xau4yffDJ8+eUhfPppNZMm9fxLDpCQEM/IkQeFXN6zt7z6qsouuOSS\nfWPz58M996iKrIFiboqKVIGkSBFCeRO0J0ET01RUKIHQi9z1kLFYVCe0v/1NiZEvvlAq3E9qqiq6\nFMnN/Fe/gnffVdUf/cFEiYn76sVHwO927qSguZlV7XK+6+yq3kJOUt94EtY0NrIwJ4dHe4jjyIuP\nxygEWSYTj+7ZQ7HdzthOXhNPXWt6Y4YRu9dLjdvNbqeT2ampvFVbiyXII3CmOZMT8k/g1U2vBhUJ\nANdffz3PPvsskydPJjGMuI9t27YxevRojrj7CL707uv3cfaks0mM67tOkW53ZcSZDfn5Kn7t3HOj\naFiEhPwXLKX0ADtav10nhJiDikV4EBXMmEdHb0IesK6ndX/5y8u6BGEce6x6Wrda3+GJJ87giSd6\n38Th3nvf4Te/ObPX80NhyRIVi9Des3HYYerz8Msvu4oEt1vFO0XDkwC6oJJmP6Cn9MdIufFGVanx\nyy9Vc5O77tqX7nPXXXDhhcoLEE5w4TvvKAHy97+reulRpsbtZqvN1nGstS10X3gSPD4f61taOC83\nlyG9LHR0VHo6RuCLhoYuIsFdpzwJM4rXUdauntLBqanU1dUFFQmgthwWv7OYMmsZw9MDu4anT5/O\n448/zvbtISfFAXDhhRdyzLnHMOeFOSw5fQmXzLik55Mi4Ntv4f/+D26+uYqmpry25qWhUlmptGlJ\nSXBPwtKlS1m6dGmHMavVGnhylIiGzDcACVLKnUKICmAB8AOAECINOBR4oqdF/v73N5gxo2sp0+ee\ng8pKI6+/Xho0c6AzL71Ux733Hsepp37HQQcd3Pt30gs2bYLVq+H11zuOJyerxnbLl8M1neJydu1S\n1VvHj+/9dTyeJqT0YDJldjmmPQmamKeiIvpBi+0ZNQqs1sB7e/n5cPDBKsDxz38Obd3KStV34Ywz\nggcYRUiL10uDx0OD201Gqzu9ukU1ZOmLmITNNhsOny+kngppcXHMTk3li4YGft6pYZOnVnkSjFlx\nvDxJPflkm0yMjI+nvr6erKzg1QzPmHgGCcYE/rzyz5wy7pQOxwSCw4cfTmpCKtddd12vbQ3E+W+e\nz4j0EVw4/cI+8yj7KShQ3RtHj66kvn5q2DGtBsO+at7BRMKiRYtY1KnoWEFBAbOj4OEKRqh1Eu4D\nPgJKgVTgAmA+4C80/ggq42E7KgXyHmA38G5Pa5vNo0lO7rqXNXmycsNMmaLqJvSGBQuGMGlSNZdd\n5mblyibi40OvzrZhg/JiduaTT9RN+vQARbvmz1eipnMmht8TGoonobj4RpzOPUyf/mGXY9qToIl5\nystDU8XhEOypYdo0VRHx1lvVHt/8XtYd8BcwMRjgqaf6LJ2q2esFYJfDwYxWkdCXnoTvmpoQwMwQ\n8+WOzczkufLyLnEJ9hoXHiP835hBXJC3bw++rq4OKWW3noS0hDTOnHgmj3/zOI9/83iX49Nyp/H5\nRZ9H9HPYXredVze9ymMnPYbJGF4QYShYraq/TkZGJVOn5vKzn4W/1pNPKo90d4VK+5tQPQm5wAvA\nYMCK8hicIKX8HEBK+aAQIgn4J6qY0lfAyeHUSPAzc6Z6LSjovUhIT4/j6aclp546l/vue4m7776w\n19eTUnWwvflm9RkRaEv19ttV8GBn5s1Tpee3bev4+VhUpLyeoVSNtNkK2/JuO5OdvS9dRqOJSSoq\nen9z7gtuvFH1V//JT1Q7WVCPZ2+/Hfzmv2QJfPCBikXIDX9vuSf8ImGnw8GM1qf7als1RmEkIzGj\nu1PDYm1TExOSkkgJMT7kmIwMHigtZavNxqR2lds2lllxpsHFnbaTalufXLoTCQAvnfUSfznhL13G\n9zTu4fR/n86CFxfw+eLPu63O2B0PrHiA3ORcLp15aVjnh4rVCpmZHtzu2oj6NoCqjzBxYuD7y0AR\nap2EHkuMSSnvBu4O054uZGaq1MH33tuX/tcb4uKGcMopW3nwwTMZP/4bLJY5PZ4jpSol/847am/o\ngQdC68Z1xBHqIWT58o4iYds2Vfelt9slAE5nGR5PQ8Bj2pOgiWmk7Pvthp4wGlWE8T//qboy1tYq\n78Dy5SpfuTN2O9xxh6rDEMhNGEVafD5AiQQ/NbYaLEkWDCKED4lesrapidlhVN05Ij0dkxB80dDQ\nQSRs2dPIiHQDkzuVfPWLhO62G0C1mh6W1vWJaVjaMD5f/DnHvHAMx710HLcecSuC0Lw5Do+DF9e/\nyH0L7sNsijDDpZdYrTBkSA0gI8puACUSYimzAWK4d0N7jjoKXnhBpR2GxgSMRjfPP7+VTz7pWSSA\nEiXvvBNeGei0NFXbYfnyju3ki4pC87xK6cPp3IOUbnw+JwZDR6WiYxI0MU1jo7rpDqRIABU46U+R\nlFJ1kPzb3wKLhJdfVn9Uv/1tn5vV3pPgp7qluk8yG9ztghZDJdloZE5qKp/X13PNUBU0vtfppLba\nyeTsrgGhda0dMXvyJHTHlNwpfLb4M054+QQWvdlDw68gDE4ZzFUH9008SSCsVhg8WMXqR5LdIKXa\n5u5jjRoy+4VIeOYZlV4YOoKzzy6nuTmbnTutGI1dC4N0Jisrsp4I8+erB5j2cQnbtsF55/V+Dbe7\nGilVBLHLVUliYsfanBaLSsG02yNPB9dook5fFVKKBCHguuvgl79UUWbtU5N8PlUy9cwz+6XUc4tf\nJNj3pXzX2Gv6JB4hnKDF9hybmcnf9+zBJyUGIXipspKMRhiU1/WDp7eehJ6YljeN0l+V4vA4ep4c\ngMS4xH6JRfBjtcKIEWprOJLthpISVYNHexLCIC4u/EJKZ5yRyR//OA9YyvDhP+9xfqTMm6c+b3bt\ngtGjVf2V0tLQghYdjn3d1QKJBP+2S23tgdcdU3MA0FclmSPlwgtV1cR//ENVbPTzn/+oTjnPPtsv\nZjR7vcQL0dWT0AeZDWvDDFr0c0xGBveUlPDo7t3kxcfzTHk599hNmLO7bprX1tZiNpsxR+HJxWQ0\n9euNPhKsVsjOjtyT8MMP6nXatGhYFT2ivwHWx7zywyss276s1/N/8pM0HI4Uli0r6EOr9nHUUapB\nx7ut+RzFxcqrEMp2g9O5u+3f/s5i7dH9GzQxTayKhNRUVQHtqaeUevfz0EOq0En7Dmx9hMvnwy0l\nk5KS2OVwIKUEVExCtjn6noRwgxb9zE1LI89k4sbiYi7YsoUddjuDWgyYsrrewHuqkRAreL02Wlq2\nBPjaipTekNezWiEjowqjMQWjMfxy3T/8oLa7h/a+HFC/sF94Evy4vC6u++g6cpNz2Xrt1l7lvx50\nEGRnO1i+fCQXXLCBlJS+lWmZmXD22eph5frrVTwChOZJcDrLEMKElO62piHt0SJBE9OUl6t9sDBd\n3H3KtdfCY4+pAKfFi1Uf+uXL4Y03+uXy/q2GqcnJrG9pocrtJi8+nmpbdZ+lP4YTtOgn0Whk99y5\nuFvFjFEIvm1YhcnSVSTU1tZGvNXQH2zdehHV1YH/v8eMeYARI24NaT2rFdLTKyPObPCXY46lRoaw\nn4mEL3Z+QYOjgQZHA1+Xfc2RI47s8RyDAU48MZ4135zC7/9zKL8+dRe5yX2X3gSqtPzRR8P//qfi\nEVJTVfv53uJ07iYhYTgejzWgJ0E3edLENP5qi7H2aQfKpXfiiXDRReoLVPrUmX1TnbUz7UUCqLiE\nXJOpT5o7uX0+1jc3syjCdM44g6HtRiGlxF3nJi6r661jf/EkNDUVkJd3IUOGdAxu3LbtFzQ3rw95\nvbo6ePbZM0lMPIHMrrXves2nnxJRjYW+Yr8SCW9sfoMxmWOQUrJk3ZJeiQSAk04y8Mor05kqUina\nfgvNQTqQAaSkzCI3d2FEds6bp2o6PPkkpKcrL0Ion5d+kWAwJAYUCWlpKk5DexI0MclApz/2xFNP\nqapofg47TKVM9gPNnUWCw8E0cxwOjyPqnoTNNhtOKcMOWgyEt9ELXoJ6EmJdJHi9dhyOnYwc+RvS\n0ztuL6WkzKSlJTSRIKUKNly5cg6HHrohIl08Zw6c33dNKsNmvxEJHp+Ht7e+zeWzLifZlMwDXz/A\nYyc9RmpCz38AJ7TWg1yx5gRGDP6YKnvgwBqPxwp4yck5J6JSnkKoiq433aSCpUPtM+N0lpGQMAIh\nDAFFghBqy0F7EjQxSXl5bIuEESNUg6gBwF8jYWhCAplxcex0OKi2qSyHaKdAvl9TE1HQYiD8fRsC\neRJqa2sZPXp01K7VF9jt2wBJUlLXkoZJSeOorn49YOfLYLS0KKFgsVTz3HP/YMKEf0TZ4oFnvxEJ\ny3ctp9Zey8LJC8lLzuOu/93Fq5te5fJZPdZ3IjcXskbv4plPT+D4S0/j3CmBW2xVV7/Jpk0Lcbtr\niY+PTNUvXqwqM27dqpp/hILTuZu0tMMBLy5XRcA5uqCSJmapqFARvJou+D0JyUYjoxMT2elwUGNq\nAjqWZH6+vJwfWloCruEnXgjOyslhTmpqh5uaV0ru2LGD15eX8ejmVFKOjt7HvL8D5P4auGizbQEI\nKBLM5vF4vY243dW9zlLw91bKzi6PuJBSrLLfiIQ3t7zJyPSRzB48GyEEJ449kSXrlvRKJACYJ30F\ny0/gm5WrGVwXeI7NNoOGhuHY7YURi4SMDFi0SFV6Db2Q0m4SE4fj8zmC7pHpgkqamCXWtxsGEH9M\nQopfJNjtVBs6NneqdLm4vLCQoQkJJHezDVLndvOnsjJmpqRw8aBBZLZmMLxaVcVHdXW88VYymf9p\nwnmDk4ShIZSO7QZ3rfIk7K+BizbbVkymXEymrnaazeNa5xSFKBIkFktZROmPsUxMiYSC8gIeWvUQ\nPqlcchMtE/n1Eb8mwZjAW1ve4mfTf9ammC+beRn/9/r/8WXJl4zO6OjiMhlNDErp+CFlG/1v+PBC\nHrriDB4KakE+Q4Z8wbx5X5KefkTE7+e66+Cll/b1n+gN/kJKCQnD8HgaA243gPYkaGIUjweqq7VI\nCEIHT4LZzFvV1dSIjs2d/lVZiVEI1h18MFmm4LUCvFKyrK6OJ/fu5Ybt2/G1jmfFxfGfkZMxf7oF\nCdR+VMuQy4PHYYVCsO0Gl8tFU1NTzHsSWlq2kJQUuAmQ2ZwPCOz2bWRk9C7eTTUi9WKx7I44uyFW\niSmR8Ny65/ig6AMOGXIIEsnbW97mpR9e4rKZl1HZUsnCyfsCCk8bfxo5STnMfz5wE5n3znuP0yao\nXvBNzibqcz6EG4ZzxfRfcdPhNwU85+234bbb8qmvfz4qxeJmzICGhtCqIvoLKSUkDMPtrsXjqcPn\nc2MwqA+LlysqmJCURHZ2GuvWRW6jRhNVqqrUJm0sVVuMIfyehGSDgdGJiZQ6nVT6qkmJTyExTpU6\nfqGigtMslm4FAqh0xFMsFk6xWPD4fG0iwSgEVc9WUOiSmCeYqfuwLmoiwVPrQZgExpSOHo76+nog\nspLM/YHNtpW0tMMCHjMazSQkDMduL+r1ev7tBoulnPj4GKuCFCViSiRsqt7E8WOO541zVQ5rYU0h\nl79/OXd8fgfD0oYxZ+i+/gsJcQmsvGwlxXVd2yFe8u4lfF32dZtI2FG/AwBDxl4M2duDuv/9tVS2\nbWts6+vdHS9WVFDQpPYTjULwq2HDGJ7YsaZ5qMXH/IWUEhKGt8UjuN1VJCQMxeb1ckVREQtzchhq\nSdOeBE3ssUP9rWlPQmCavV4ShCCuVSR4pGSnvbnNi7C+uZn1LS3cE2IAYFyn7nGVL1eSuSCTjGMy\nKL2/FJ/ThyEh8tp5/vTHzoF90SrJ3JdI6cVuL2TQoIuDzklKGo/N1nuRUFsLPl8cFstevd3QH2yq\n3sRVs/flrk7InsDyi5fz/PfPk5OU06VD2tissYzN6lpr/eAhB7OuYt9jdnG9EhJTcqbQ4AzcWREg\nP1+9btvWu6pbN2zfTrzBQLbJxFabjbz4eG4ZMaLnE7vB6dyNEPGYTNlt7iuXq5KEhKF8Wl+P3eej\n0GbjIB2ToIkliorgL39Rndhyc/ulB8L+SIvXS0prnMHo1geKEoezTSS8UFFBjsnESRHcbB2lDhr+\n18DE5yeSMiOFnb/ZiXWFlcwFESTxt+Kp8wQNWoTY9iQ4HKX4fA6SkwNvN4CKS7BaV/R6zd2txXEt\nlvIDdrshZsoy19vrqWqpYkrulA7jBmHg0pmXtnkFesPMQTNZV76ureRpcV0xqfGpjM0aS729Puh5\ngwdDYqKbXbvMPZbndPt81Hk83Dt6NBsOOYQpSUkdarGHi0p/HIYQhrZoWX9cwjutqqDQZiPLImlq\nUl1wNZoBZdcuVXD+vffg979XfRDSe26m9mOk2ettC0Yc1SoS9nokOUk5uH0+Xqms5IK8PEyh9JXv\nROW/KjGYDWSfnU3y9GTih8ZT+5/ouB3dte6gQYsQ2yKhu8wGP2bzeOz2bUjpCzqnPX6RMHiwJC7u\nwPydjxmR4N8SmJo7NeK1Zg6eSbWtmr1NewHYXred/Kx8MhMzaXAE9yQIAWPGONi9e1SHJkuBqHWr\nAJ7c1n3DMWZzh65u4aIKKamuTX73lctVgcfn472aGiYlJWH1eonLUdfXWw6aAWfFCqVWN21SDZQy\nMgbaopilxedr8yQkGo0Mjo+nxhdHdlI2y+rqqHK7uSiU8qydkFJS+VIl2WdmE5eqtgUsp1io/TBK\nIiFItUW/SMiMpORgH2OzbcVgSCIhIXi3wKSkcfh8DpzOPb1a09+mZOzYyRHV1ollYma7obi+GJPB\nxLisEJocBGHmIJVOsK5iHUPThlJcX0x+Zj4ZiRndigSA/HwDe/aMxW4vwmweFXReVatIyGkVCaMT\nE3k/Cnds5UlQv8QGQwJxcZm4XJWsbGyk1uPhz/n5XFpYSFOGDYintlbHiGkGmLVr1V5dDD9Fxgrt\nPQmgPje+SZnBl8apuP5eyB/XxWHO2UthmOv77D5sm23k/yW/bSzrlCzKny7HXmzHnB9Zh0ZPrYek\niV2bGNXV1ZGWloaph2DLgcRm20JS0gSECP5sbDargDW7vYjExJ5bD1uttRgM6Ywd2/fNwQaKmBEJ\nO+p2MN4yPirtQUekjyAzMZN15es4dfypFNcX89MpPyUlPqVHkTB+vJm1a8dis31IVtYJQedV+z0J\n8apl6hizmRKHA6+UGAMoSp/PRVXVa/h8Xb0NyclT2kqE7iukpIiPz8PtruSdmhqGxMezKDeXywsL\nqUu2ARk6LkEz8BQUwKxZA23FfkFzu5gEgF8OG8biXctIzRrPoqe9pHkNNA9vjuga2Wdmk3n8vif6\nzOMyESZB7Ye1DPtFZL3l3XVu4iyBPQmxHLQIypMQLP3RT2LiKMCIzbaNzMwFPa7pdpeQlZVLVlbP\nc/dXYkYkbK/fzpSpU3qe2AuEEMwcPJN1FetweV2UWkvJz8zH6XX2KBLGjTNQWTmSxsbt3c6rag0G\naO9JcEnJXqezS4YDQEPDF2zdeqHfwnZHJGDgkEM2kZQ0HqdzT9t2A4DJlIfLVck7jTWckZ1NotHI\nqMREKuJsgN5u0AwwPh+sWwe/+c1AW7Jf0NLJk7Aw28KiDb/hV8c/TfoeCxOWjGPwJdF1DcalxJEx\nP4Pa9yMXCcECF/eHvg0tLVvIzAz+4AdgMJgwm8f0Og3S46kjLS0Rkymyn2ssE1MxCVNyoiMSoDV4\nsWIdJQ0l+KSP/Cy13eD0OrG7g8cO5OeD1xtHcXFjt+tXud0kGgxtTwVjWoXBjiDBi06nio+YN8/F\n0Uf72r7mzXOQkDCcnTvvaC2k5OogEuLj86iz72Wnw8GZ2dlICePNSez02TAY4Kuv4L//hc8+00GM\nmsi4/3548MEQT9q+XXW40Z6EXtHZk1Bnr0MiyanKAUlAV340yD0/l/pP6mla1xT2GtKnOkAGClyM\n9ZLMLlcNHk9tt0GLfszmca09HrrH5/PgdHoxm2PmNtonxIwnweqwRl0kPLTqIb7b+x0A+Zn52Nzq\n6bvB0YDZFHhvzp+5tX1799Gt1S4XuSZTW7CKP1J5p93O/ACBWy5XBXFxWW1FkfwYDAmMHn0vW7de\nSEbGsQAd98LicqizryPNaOTojAwWLADHQUOpPm87Y8bAo4+qL1DVHWOp1ahP+vi4+OM2UZabnMsR\nIyKvZKnpGz79VDVDvOWWEE5au1a9apHQK1o6iYRqmyrJnLknEy9ekib0jUjIuzCP0j+VsvOOnUz/\naHqP8yterMBWaOswJl0SfMGbOw2K4doY+zIbut9uUHPGU1v7YY/zmpq+ob4+l7S02A3WjAYxIxKA\nLumPkTBzsApefHPLm8Qb4xmWNozdjSpfpcHRwODUwC694cPBZPKya1cqXq8dozGwmKhyu9u2GkBF\nKg+Jjw/oSfBKyQ8NO3D4Mjjm+++7LiYnc6VhPM7tt5IALNrWQLNYx16Xizl2Bwup4mdD8og3GCgo\ngKEymZ1n2Kku8NFiVSp2+vR96Tixwmc7PuPkV05u+14gqLi5gtzkA7PoyP5OczOEnHlXUAAjR+qg\nxV7S7PWS3O6HXGNTQUXmUjPOXGdAV340MMQZGPPHMWxauIn6/9WTeXTwG1vZI2UU31BMwsgEhKFj\nfFXSpCRSZnTtKllXV8eUKdH7/I42NttWwEBSUs+B8WbzOByOHfh8HgyG4LfIurpl1NRcy+DB0W3x\nHWvEjEiIM8YFLIwULhMsEzDHmflw24eMzhiN0WAkI1E94XcXl2A0wsiRTvbuzcduLyYlJXBKZrXb\n3Ra06GeM2dylVsI71dX8ZudOzrEVMybOwtBO5/j5nhs5yXoVXkxkJOSRIQwclJLCPO8kMioaeWzs\naKxWVQY0sdiEF6gw2Jk0TPWlz8tTFXFjiRWlK7CYLRReV0hxfTGHPnMom6o2kTtai4RYpKlJpQGH\nxNq1MHt2n9hzINLFk9CiPAnGnUaSJvWNF8FP9tnZpB6cyo7bdjBr1ayAKXvVb1ZTfGMxw28ZTv6f\n8gOsEpiBClz0+dyUly+hqupfSOkJOs/pLMNsHoPB0HOjK7N5PFJ6cDh2kZQU/J5UVfUZVutd5OTo\n7YZ+YVTGKOK6UW2hYjQYmZ43nTV71pCfpX7ZeyMSAMaONbalQQYTCVUuF+M61VwenZhIcbtaCd83\nNXHWpk0syMjgOIOH7KSxXByk3rOUk1i//lWczjJenrLvmjU149lYIfG4aygtVd6PyjIj2I0U2mxM\nSlYiITc39kTCqt2rmDt8LpYkC2kJacQZ4thSs4VjRh8z0KZpAtAcalC9lMqT8Otf94k9ByLNXi+F\nVT9w3da/8tjJj1Fjq8EgDHi2eZDXPMLevfMYMuTKPrm2EIIxD4xh/XHr2fuPvaTNSetw3FHqYMvP\ntpBzbg5j7h8T0tr+wEWrdRVFRVf1WIwumH2jRt1NTs45vZpfVfUqO3b8BodjBxbLqZhMwZ/ok5Im\nkpl5XK/W9XsbKitfIDk5cD8GKd2UlpYipeGAr0Ae0l1ZCHE7cBYwEbADK4FbpZRF7eY8B1zU6dT/\nSilP6W7t/Mzeq9beMnPQTCUSMjuKhHpH8KqLAOPGxfPee+Ow2d4JOqfK7ebwTlXlxiQm8kn9vrU/\nra/HbDDw4fTpfP9dDfHxwZ+4hBBMnry0LcDRT/uqi36RAGDek0LhlH2CJNZEgtfnZc2eNdx6xK2A\n6sw53jKezdWbB9gyTTCaQo1p27FDuba0J6HXtPh8bK5cyxvfP0F6Qjpmk5nshGzshXbEqI+pr/f2\nmUgAyFyQSdZJWWy7JnBgXvpR6Ux8fmKXbYbusNvtOBwOLBYL5eVLcLtryM09N2Tbamreo7LyX70S\nCcjO+j4AACAASURBVA0NK9i8+TwsllOZOvVtUlKi11wpIWE48fFDKSm5t9t5dXWqU+SwAzexAQjd\nk3AU8DjwXeu59wMfCyEmSSnbpwx8BFzMvlw/Z08L94lIaI1L8G9jJJmSiDPE9cKTINi7dwyNjRtw\nuaoDznE6qxhiSEBK2ea2G202U+FyYfN6STIa+dJqZW5aGvEGAy5XRY+1vePj89pEQfsxALe7kh9+\naABSgDjy9qZTaNsXWJSbC9t6DsjtNzZXb6bR2cjcYXPbxiZlT2JLzZYBtEoTDCmVJ8HnU//u4Il+\n7DGYMwcO69Q9TwcthoRXShw+H3VNexieNpz7VtzHeMt4xrvG4xNNYKrB7e57pT/13am0bGrpekBA\n8pRkDKbQ3Of7qi1mUFv7AYMGXUh+/p9DtksIE1VVr/ZqrtX6JUZjGlOnvtttcaRwEMLAYYftxOfr\nvsx+cbEKVo+wXU/ME5JI6OwNEEJcDFQBs4H2XTGcUsrAd9cgjMkMzb3VG2YNVh9efpEghOixNDOo\nDAe3O57Nm7+itjbw/vmLACWwO+6vDB9+A7AvDXKXw8GEpCS+slq5YdgwfD4nHk898fGh+6XaN3n6\n7rtqoI74+BSSdqdRaCttmxdrnoRVu1dhFEYOGXpI29jknMk8U/DMAFqlCYbLBZ7WLV27HZL82+NS\nwh13QGoqbNgA2e1cugUF6jEqV8eY9AZ/m+jqplIeOeIWvq/4niXrlnBk45EwRHkQXa6+/yM2xBtI\nnZkatfX8zZ3M5irc7koslt732WlPaurBlJX9GZerqseOio2Na0hNPSTqAsGPwWDqkonWmdLWj18t\nEronA1UNqK7T+NFCiEqgHvgcuFNK2XlOB/rCkzB78GxeOfsVjh9zfNtYb0oz+9MgExKWMnVq15KG\nNS4XlxcV8nvzG1itX+4TCa0xCjsdDlw+Hw0eD/MzMtr+8Dt7CXqD0ZiI0ZiGy1VJYaEdqMNoNOIr\nmdXFk1BdrZ4Ee4pQ93rtNDauDrhvmJQ0nsTEyH/rV+1exfS86aTE74uEnpQ9ifLmchocDW1bP5rY\noP1WQ2NjO5FQWwstLerr8svh7bf3uRnWrtVehBBobhUJXk8L47LGccXsK9jTtIcZX85AjNqLhH7x\nJEQDh8PBH/7wB5qbm9m7VwkcIQqIi8vsUDE2FFJT1QNFU9O3WCw/CTpPSklj4xoGD74srOtEi7LW\n9j4x3K4iKoQtEoTysT8CrJBStt9o/gh4E9gJ5KO2JD4UQsyV/raMARiaNjRcU7qzkfOnnd9hrDci\nYdQodaOtrj6iw4OTn7KmJr4mh8TUSpobP2gbHxwfT4IQ7LDb2W63kyAEh6am4mpR1RvD8SSo8/Jw\nuSrYs8cIlOJ2N9O4Yy61Hg+1bjcWk4ncXPUk2NAAwYKMpZRUV79BcfGvcTpLAs5JTT2Y2bO/DcvO\n9qwqW8WC0R1LlU7KUTnKW6q3MHf43ECnaQaI9kGLTU3sC8Yqaf09+e1v4Z574Omn4YorlCItKIDr\nr+93W/dX/J4EvHbGZo0l3hjPh+d/SNFXRdQc9BhuwO2u6TH1LhZYtWoV999/PxMnTsRkMnHssceS\nkPAVGRknh217YuIo4uIsNDV9161IcDpLcbsrSUs7NFzzo0J5uXo90BueRvKb+HdgMtChOo6U8rV2\n324SQmwAioGjgS+CLXbKyad0aQ5y4okn8txzz0VgYlcyEjN6DFyMj1cupO1BKjP7mztlpM2iuupv\nuN31mEyZGIRgVGIiOx0OdjkcHJqWRqLRSIu7snXd8Lq7xccPpqzsEazWX5GWVkpj424qdxrBLSi0\n2Tg8Pb3N41tVFVgkeDyNbNhwOlbrciyW05k69W1Mpo4Ty8oeoqYmeLBmb6m11VJYW8id8+7sMD7B\nMgGBYHP1Zi0SYozOnoQ2du1Sr9dfD5WV8KtfwT//qQRCXByceGJ/mrlf4/ckGKWbkRkjAfUgY9ti\nw3heOepTReLx1Ib9WdFf7Gr9vVi3bh2JiYk4HLtZvXo4FsttYa8phCA19WAaG7t/SGlsXAMw4CKh\nUn2s96tIWLp0KUuXLu0wZrVa+/SaYYkEIcTfgFOAo6SU5d3NlVLuFELUAGPpRiSMGzeOtLR9KTkl\nJSW8//77HQIDo0FvRAKoLYflyyGQRqkY5QMBg9IOphpobv6ezEyV1jfabGa73c5Kq5WrhyrviMul\n+omaTDlh2Tx27CN8//37SDmESZMaWLNmM16vgN1m7h9UysSkJKyJAjKHUlWVwMQAlUfr6j7Gal3O\n1KnvkZ0deM8wMXEkHk/3XpbesHr3agAOH97R7Wg2mRmdOVoHL8YgnT0JbezaBSkpSnn+9a+qYldm\nphINxx+vW5CGgN+TMCQps0O6t22r7f/ZO+/4Kuvz/b+fs3Jm9iYBMiEgSzaIUgTFvQfO2tZvnbWt\n2pYO21qrdvCrVVsV69bixL2qCIKyDEhCIMyQvffJ2eP5/fE5JznJGZlo1Fyv1/MKPPuc84zrc9/X\nfd0o0mrQarOw24/5cvKjmyRUVFSQmpqK1qfDaml5B1ASH79yWPuNjp5Lbe3aiM/9zs4dREVN+Nq/\nI39zPWOwt9Rxw6pVq1i1alWvebt372b2cawwGjRJ8BGE84BTZFmuHMD6GUACEJFMPPzww5wYkN98\n5513OOecc6isrGTChAmDPc0gdLrd3F1ejkmbwLH2Y/2uv2QJ/P73sGNHqKWJKOdPp/i+GBQKHV1d\nX3aThGytlmcbGujyeDjFRzGdzgbU6sR+hTDhYDLNYu/eWkDJsmUT2LHjGQAWtadxyFrLIauVIzYb\nfE9DY2Poehy3uwVQRAzjqVSxeDzmYYc7t1VvI9mQTFZsVtCyKUlTxkjCKEQgMQgiCRMnCh2CwQDv\nvnvczuHuu+H553v+/5vfwLV9i6m/wfBHErKie4iVq9WFq9GFylRBQswK7PZj3whdQkVFRa/nckvL\nO8TEnIRaPbwEvck0F5frHhyO6rCtms3mnV97FAGgrQ2ioobgUvoNw6A+niRJ/wauBK4ALJIkpfgm\nrW+5QZKkv0qSNF+SpAmSJJ0KvAEcAj4czLH8zKiwsDDsOg6vlyankyank2ankwiSBza0tbGmuhqL\nbmK/mgSAu+4Cjyd4cjjg7DUNKJqjWLZMyaWXlnPDDQt55BExGsvSaunyeFBJEgu7SUL/5Y8AZrOZ\nTz75pHtqCWjx+NlnIje8bNksoBmTycrp5kwOzp/PwfnzydfrUWTYwlY4uFwtqNXxEdXAKpW4wT2e\n4YWvtlVvY2HGwpAjgYLEgjGvhFGIwEhCULph4sSv5BxefVWEbs8/X4zOHnzwKznsVwaLV/SDyYvp\nIfLWg1bQ2nArG7oFf19FhcNwUV5eTmZmKnZ7NTZbGe3tG8JGKAcDk2kOIMSLoeD1ujCbd33tJMHt\nFveMz8vuW43BcqAbgGhgE1AbMPmdMzzAdOBN4CDwOPAFcLIsy67BHCgtLY309HR2+WuxQ2Derl0k\nb91K8tatJG3dSua2bfzfwYO81dwcRBhKfZUAZnXKgEgCCIbYd9JoIPnMNma9eICPPoJzztlDRYWe\nm26CZ5/tqXCYazJ1t4R1OhsGJFr8zW9+w6mnnto9zZs3D7vP5nnPHnHO8+bNRaWC+PhySgMG5Dla\nLeqJkUmCShXZX1+lEhUHw0k5uL1udlTv6OWPEIiCxAIq2iuwOEPUaY/ha4OfJEhSmEjCcYYsi0Nd\ndpnoRPmLXwjZQ2W/scpvDjrd4hE4OW5i9zzrASuME9UBRuMMFAr9NyKSUF5+BJXqTbZvz2THjhy8\nXvuQSx8DERWVjkaTHpYkWCwleL22r50k+PUIppGrJB21GKxPQkRSIcuyHRheUioAs2fPDksSjlit\nFFss3D1xIjOMRlyyzOcdHbzX0sLjdXW8OnUqFyX1aAD8JKFFGUubrW1YWodGl4uUKA3Ll8OUKeUc\nOnQmP/yhk4oKBYt8ObrATpBOZ32v9s/hsHHjRlatWsU999xDbW0t3/ve91izZg2rV6+mrMyNXm8l\nOlpPQoIGvb6E0tIei+ccnY6P09toDGOo5HI1o1YPjCS4XG3oQve16hf7m/ZjcVlYkLEg5PIpSVOQ\nkTnYcrDbx2IMXz/MZkGA9fqASIL/zf0VkIS2NnEO/kOdcQao1fDWW3DLLcf98F8J6qxt4HUxOaGn\n3NtaakU9swkXoNPlotEkj/pIgsfjobq6jpQUOOGEt5AkNWp1PHp9/ojs32Sai9kcOoLc2bkDSVJh\nNA792bFtG7z55pA3B3o8ab7t5Y8wino3hMKcOXP45z//GfKF/mFbGypJ4qcZGZhU4mNclJTE/8vN\nJWv7drZ1dPQiCQd8JKFONuCRPVhcll41/INBk9PJVF+cyWicBXhIS+uiujqaPL2ePJ2OCwJqJ53O\nBkymyMKS1tZWSkpKuOOOO8jOziY7O5vbbruNe++9l/nz5+NwJJObK9xukpPjkeUvOHDgEjweCaVS\nkARXQi0NjTI9Rpc9EOmG/kiCuOKHE0nYVbsLCYnZ6aE/7+REoaosbSodIwmjCF1dIsRvMAREElpb\nxYKvgCT4iyj8ae6YGFi2DN5449tDEiq7msBjJzde9AZwtbloeacF1WUNeJXRqNVJqNXJOJ0NX/OZ\nRkZdXR1ut4fMzJQRSTH0hck0h+rqNSGf+2bzDgyG6WG78/YHqxXOO0/w3+FWJcTGisZ633aMapIw\ne/ZsWltbqaioYGKfB9UHra2cFBPTTRACMcNgoNjSE86WZZkDVivZWi0VdhkkNe329iGThEaXi+/5\nujmKBiBKkpMbqK6OxqBUcmh+71CYy9V/uuHzzz8HYMmSJd3z7rrrLp5//nmuuuoq4Cny80WUIjU1\ng5qaIhwOifJyyMkRJMGrkqm2OwBt0P7d7hb0+hBlDwEYiXTDrrpdTEqcFPa7jdHGkG5KHxMvjjKY\nzSJ0ajQGkIS+b+7jCL8dQ+Btfv75giC0tX07Rmw1llbwuMmKzcLV7qL4tGKcDU5iT2tHqc5FkiQ0\nmuRRn27wlz9mZxccl/1HR8/F7W7HZjsS1Nq5s3MHsbFLh7zvxx4T19OhQ5AVrKseFJYu/W50SB/V\nuky/eLFvysHh9fJJWxsrw7gGzTAaKQpQYlU7HHR5PKxKTsaDBIYJA9YlhEKj00myz9NBqdRiMBSQ\nkFBGdXXwuh6PHbe7vd9ynS1btpCenk5WwJUbHR3N/fffT0NDAxpNLrm5gphkZEzC4RAv2S++EKZ4\nyQ5BDOoUtuCdM1BNgihBdbv7LxENh111u/qNEExJmjImXhxl8EcSTKaAdIOfJHxFkQS9vrfr87nn\nCrHwcSyo+ErRYO9EhQelVUnxymJsR23M2DADt64cnU7YvIpIwugmCRU+Rpebe3zK7vzixdLSq9i7\n95zuqbj4bKzWA0PWI9hsQu9yzTXDJwggept9242UYJSThNTUVMaNGxdU4fB5RwdWr5fap57irLPO\nCprev+kmGj//nHqH6CvlTzVckpwsAvGGXNpsQ3sRWj0eLF4vSQHGT0bjLOLi9lJdLWyRA+EfFfRX\n3bBlyxaWLFkSFF675pprWLz4JGQ5s9sjPD19Il1d9URHW1m1SjxY540zwNYEmjWhm5IMJN0gSUqU\nypghRxLcXjdF9UXMTov88ChILGBX3S5eKnmJl0pe4su6L4d0vDGMHMzmHpLQK5JgMHwlw6XASks/\n0tNh/nyRcvg2oNlhQSfJlP+hHGuplRkfz8A004TNdqSbJHwTIgllZQeIjobk5OOTLlSrE8jM/KUv\n+qroniRJSVLSpcTHR2woHBaPPy6MQn/965E5z+8KSRjV6QYILV78oLWV+EOHePAPf2DZsmXE9Pml\nHJWV8Mtfcv3Bg7z44IOUWq1ESRInGAxkR2k4aswdciShyee2mOxLN4AgCTEx23E6hcFGYL8bv5FS\npHSD1WqlsLDQl1boDYVCwWuvfUJqqrqbJKSkpNDa6uWpp65Drxdd066+GrwVJmxxNpxOIULzQ5Y9\nuN1t/ZIEECmHoZKE0qZSbG5bvyRhceZiHtr5EJe/djkA0VHR1N1eh16tj7jdGI4furoEQYiODogk\nVFQEv7mPE8rLQ2c1zj8f7rlHjAKHKqYdKIqKYN++4PkKBZx5pvhuhoN2lx2jQYW9wk70wmhMJ5rw\neGw4HNXfqEjC0aNFpKSAwXDCcTtGTs79I7o/ux3+8he48kqRnh0JjJGEUYLZs2cHiRffb2lB88QT\nTJs2jY8++ghFHzcLj9eL/vbb+fDRR5m1ZQsLn3ySfL0epSQx02QcHklwOgG60w0gSEJi4guAMKTr\nTRL6t2TesWMHbrebk08+OeTyujpxLD9JSE1NxeXyEhPzCitXPoFKZSQ1FTo7dNgmNdPUBOMCWmGI\nl748IJKgVscNOd2wq06QOX+L7nC47ITLODv/bLyyl4qOCqY9Mo3XS1/nyulXDum4Yxg+AtMNvn49\nX6lHQkUFLArRF+j882H1atFbKpK5o9EoRoiB5HigMJuFcdPDDwtBWyjcdx/8auiOw8iyjNntIkWl\nwd3mRpMqTtRuLwNApxO5d40mGa/XgsdjQakcnUX45eVHSU0FvX7S130qA8YTT0B9vfidRwKyPEYS\nRg3mzJlDa2sr5eXlZGVlUeNwULJ5M+zcydq33goiCABKhYLZV19NwooVbL36ajb8618s+q3oIzDb\nFMNrxlzabGVDOh9/34akXpGEmSQlCUHChg3X43J9ypQpL2EyzfKRBCmiJfOWLVuIi4tj6tSpIZf7\na8X9I61UX/ed1laZxsYXMZlmER8/CaXZRlzWburqID19SjepcrmEKVN/mgSxztAjCbtqd5GfkE90\nVP9DLr1aS2vrRyian+Evs9J4as9TYyTha4S//NAfSehyu9FUVaFZvLjfbUcC5eVwxRXB8ydPhksu\nEQ0nw8HlgrIyWL4cTjqpZ35rK6xdKwzQwsHjgaefFrqeNWvg+uuDHfROPx327BnMpwlGk7UJt6Qi\nXqPH1epCXyCiZjabaBDTE0nwt4ZvQqcbnSShsrKW+fPjUSiivu5TGRAsFhGNuuIKyB+ZKk2sVnHt\njJGEUYBA8WJWVhYftLTAf/7D3AULOPvss8NuN8NoZEtiIj/72c/43Z/+RObPRDvnmUYjqAyU20ML\n/PpDoy+SEKhJUKtjmT//XlQqDzbbOXg871BXtxaT6RGf22JiRJvjLVu2sHjx4pCEBwRJ0Gh6IhQp\nvrqbjo5YDh263ncOb5LQJnHvxGvo6oL29o3ExS0FekjC8U437Krb1W+qAaCrq5ji4jNwOmtRKPTM\njlHwqy8/oaK9orvxzRi+WgRGEsxmOK24mJOWLOGv44ffNrw/tLeLUVm4oMXLL4ee74fZLMhNX+Ol\n22+HdesI2ck1ELNnwz//Gf74M2fCxx9H3kd/ONJ6BJQ6krUm3G1uVHHieWCzHUGhMHRHGjUacZO7\nXA3odGFO6GuELMvU1nYyYcL0r/tUBox//lOQwLvvHrl9+nsqfRdIwqgSLn7W3s7pRUWsCJiuqq8n\nKjmZn65bx4JXXuHOBx6Agwf56333RTRDmmEwcMBq5ZLrrwe1mkNPPw34SAJwZFD+jz1ocrmIUSrR\n9Hmhjxv3A8aNU2K3n0tKyrU0Nr6M1+v0lT+GTzW43W62bdvWq/SxLyoqIDOzZ4TjjyTExd3H7Nm7\nmT17N7m5i2iynsyPeRRZVmCz9bgqDY4kDC3d4Pa62VO/Z0AkoalpPV6vjRNP3ElOzhqUsg29Wscz\nRc8M+rhjGBn4hYvR0dAhu9jW2cnBlJSvxSNhsDCZRImkv4wS4OhReO45uP9+kQKMNL35ZuSPOX26\nKJmzDW1cAfhIgkJLii4Gd6sbdbwYZFith9HpcrufZWq1IAmjVZfQ0NCAw+ElOztyOfVoQUuL0CLc\ncMPIVDT48V0iCaMqkvB0fT1fdnWxLMCtECB51iyq1q+nZv16AOYsX87SpUsj7mu60YgH+MTthgsu\nYMOzz9J8992kJiSgdpup8gwtVNbocvUSLQYiMxOqqiAlZRVVVX+hre0jnM76btFiZ2cnjj6xz+Li\nYiwWS0SSUFnZo0cAMJlM6HQ6WlocmEwi/5+eDrbNXmqYhN2dgsPRU48pmjsd30jCgeYDQrQYxkQp\nEA5HBTpdPtHRc305WQ9XTr2Ap/c8zW9P/i2KCP0lxnB84Bcumkxgzha/f318/FfikTASlZbjx/eO\nJNx3n4gg/PjHwzkzgRkzRNXS/v0i6jBYONwOXtj7AsqkHxIvReHp8nRHEiyWkl65fbVahD2+7gqH\ne+4JnWLp7PQCKcet/HGkcf/94rf7bUDX+qNHYfv24e23zJetHiMJXzG+MJs5LzGRxyf1FsR0vvwy\nBw8e7P7/jBkz+t3XNJ8j4ouNjXDxxShef50HHniAe+65h2h3M/UMzXS7KcAjoS8yMsTIxGCYjl4/\nhYaGdTidDURFjefxxx/nhhtuwNu3RhIwGAwRW30ePgwBDTKRJImUlBTq6+u75yUnQ3uTGI20O1Nx\nOGq6l7lcLSgUhn5ziLIMSuXQSMKuWp9oMTWyaBHAbq9AqxUvH7/L45VTz2XtnhfYXLGZpROXDvr4\nYxge/OmG6GjwThe/f118/FcSSaioAK22t+B3sAgkCeXl8MwzYgQ5EhURU6eKAo+iIqg3vUt9V33E\n9Y0aIytzVxKjjcHitHDBSxewpXILpvG/IsZXXqqOV+Px2DCbvyA5+dLubRUKFSpVwlcWSZBlDyD1\navwmyyI0n58vnmmB2LAhBTiP/PzQIuuvA14vlJSALxPcjc5OeOghITgNvLZ+/GPYsGH4xzWZegvE\nv60YNSTB5vGwz2Lh5hDfenR0NHPnzh3U/kwqFTlaLVs6OpiYksLFN97IAw88wM6dO7E1V9IqaTnp\nby8EbTdrwQJujSCBLbfbe4kWA5GRAYWF/pf4FVRU3IdaHceXXyZx4403cu2113LeeecFbTdx4kQ0\nYfbpdouyrO9/v/f81NRUGhp67FuTksBikVC2amhS9I4kDMQjAUTnS7c7jpUrB59u2FW3i7z4PGK0\n/VNru72i2zBFpRKGWDOTc8mNz+Wmd2/ihGRRWnXp1Eu5eMrFgz6XMQwOXm/vSAKz2ojyeqmPj0dO\nSAhh8j2yCOWRMFhMmAAbN4p/33uvSD+MRBQBhFVEbi5sKjnAc1XhdVCB0Kq0nDfpPCo7KtnbuJf3\nr3yfsyoVRPs83lRxKszmnciyk5iY3i/cr8orwet1smfPUtTqBF8PBvEDtLQIMejdd8OFF/beJi3t\nGE1NC0lPHz2W6q+/DheHeUykpQltSiCKi0VkYbiVDiqVmL7tGDUf8bDVigeYM4JttWYYjRy12ynQ\n6/nVr35FZ2cnbW1tmOx1WBWxfO7r794Nm43P772Xh+PjI8YVb/URmTVb13SX/SkVSsbHraG6OhlZ\nhuTkVRw79lv27bPw0582csYZZ7B27VpUg7yqDh4UDLlv8CQ1NTUokgBgqDVSl5qIw9FjUDRQkiBy\ns7GcfroDj8eOUhls7xwOu+p2DSjVIMseHI7K7kiCv/+8x9POfafex9pda2m3t1PaXEqNuWaMJHwF\n8HmNUbD3ZTTOo3DZQhYfPsQneZNp9XhI8IlhZBmuu06Ea/vCYIAXXhia79JIVFqOHy8iEpWVolrh\nz38e2Ta+M2bA5rYXiU6Opv72erSq8PdGjbmG/+79L88WPUuLrYWPr/6YuePmYS3/FJPPg0IVr6K5\nfQtKZQxG47Re2w/FK8HrddPW9iF1dU/Q1vYxshwcsTQaZzJt2pvdz4Ly8t/T2bkT8FBXt5b0dMGq\n6urE+qFKTk2mnbS0LIooxB4MPvywh9wFQqEQJG8g2a5PPxV6A182uhcmTOjdqbGhQRgqzZwpoldj\n6B+jhiTss1qJ0us5YQTv7BlGI+ubmynQ60lISOCxxx4D4Fcbfs2zZTt46ZLXuteVZZm1u9by9s2V\npD73HGuvuy5stcFMoxGH28HqDavJS8gj1ZjK55Wfc5F8AXb7hbS2QkJCNpI0h9WrC8nOTmPdunWD\nJgggWC8I8VQgUlJS2LlzZ/f//b2sTE1GatPigzQJ/ZGE5mbYuxcMBvHSfrXkKZ4ueWvA51lYW8gF\nky/odz2How5ZdqPVTgQCm0q1cfGUi7tJwZ83/5k120I3eRnDyMJsBgkvJ77wc96elgcs5MpXX+KT\n1b+nzuEgwZde84fxV67sHb41m8Vobt8+CGP1ERHl5cJZcTiYMEFEQ958U0TfRiqK4Me0aTLrW9dx\n9eQL0Kkj5zAyojP4xeJfcOeiOwERWexyi+Zsen+6IU5NR/NmYmJOQpKUvbYPjCTIspeSkgu7/RTC\nwelsxOVqwGCYwfjxvwryWJBlD5WV91NUtIIZMzbQ1VVEZeVfyMq6F7u9jCNHbicu7jR0uqxun4z0\n9ODjKBSbcLmuoKVlZIw4b78damqC91VVJaIZf/tb//vYulWUvs6c2f+6JSXi77RpkdcbQw9GDUnY\nb7UyIzk5qGpgOJjuIxyT9b2d/BK0cdhad7PEJ5D0yl5uevcmXtj1GCyGzqfqqfngA64IVbjtw67a\nIlxeF/855z8szFzI3Mfn0tm+H7iQ6mpx0ZeVzaK5uZB3370Lo3FozaSKioQgsm+Dm77pBv9DO6ZT\nT11UHB5PB263GZXK5IskRE74fvqp+NvVJb6T9fueZX9TLXPTB5bmubDgwgGN+h0OIUGPivJrEkR6\nwuVq7bXe5MTJtNnbaLI2kWwYRrJ6DOKtuWWLaPThdws64QQ46yxAvFznUIiupYZ3b/kP8ZVWZhw5\nBECd04nfV2/rVvH3+ed7P9RragRJCOipNihUVMBllw1tWz/8wt5t2yAvb/juiH1hzPsS76FDrEh7\ncMDbBJJbi0+LpDOL718RI9NxdCsTJ94VtJ1anYzVegCAjo7PaWl5k5SUa7rvlVBQKPQkJV2MMiFU\nZwAAIABJREFUyTQ7LKmOi1vOnj3fo7j4NJzOBmJiljB+/J14PFZaW//HwYM/YMaMDVRXewA1RmMx\nXV293aUsFjHs37FDuFAOB263qBr5xz/g5pt7L7vmGvjkk/73YbWKZ+SPfjSwY5aUQFTUyLkufhcw\nekiCxcKZI5hqAFgQHU2iWs2iPhLUWG0sHfYODjSLG/HeLffywt4XWH3Sau7jPhavWMyvf/1rLrro\nIqKigsV+Xq+XR556BB6Bm9+6maWnLEWr1tKUvBsQLHjGDDhyJJr4+Chmzjx3yJ+huDg4igA9JMHr\n9aJQKLprwaMtWhqMglE4HDWoVJNxuVrQ6yN3bNu0SZARi0WQhA5bNRcXXMqa09cM+dxDwW4XJMGf\nbpAkpa+iorcOwt9S+kDzgTGSMFTU1IjE69tvC2chkwnUapG/crnEhZqUhNkMF7IeV1wSW1PjaH0t\nlqU/FC+1Y9ZO8DVS27oVJk0KHvX5OfhQSEJHh+jKNxLpBhD3ywB0zYPGQfWLYEnE1LRsSNt3+VKb\nUR1epCgJq6cIr9dCTExwVZNG05NuaGx8iaioDCZPfqqXuHAoMBqnM2PGxxQVLUOWZQoKnvPdfyYm\nT36SoqJT2bZtHNu2/ZCYmNvYu7f3FynLQq+g13eybVv0sEnC0aPiMiwI8Wj63vcEGW1t7b78QqKw\nUJCNhQsHdsySEpgyBZTK/tcdg8CoIQnldjtzR5gkpEZF0RTCMS7NlIaMTMG/xNWplJS8cOELXDr1\nUv79xb+ZffVstl+3nZtvvpk5c+b02tbpdPLkk09SVFSEaaqJqVOmsn79eioqKoi5MB6lku5ukLt2\nlbBgwaloNP24uURAcbFg1X2RkpKCx+OhpaWFpKQkoqLE6Mlkj6ICkXtwOKoxGARJcEgx/K2ykjsy\nM0OONDZtgtNOg8JCQTAs9gYyYzKHfN7hYLeXo1LFoVL1/NahvBly43NRSAoONB/g5AmjR0n9daOz\nU+Ta+33ItbTAihXiDXzjjXDBBaJERpLEsowM+M9/YPVquswyF/EaZedexhGXDfZkYTulBdxdFLVW\nQcZEQIzSQ1kn+zOEQyEJoVpEDwWpqYL/lJXB5ZcPb1994ZW9vF/1IurDl7A/Xs25A9Mu9oLFTxI6\nZaQ4NR0dW1AodJhMwToejSYFl6sJr9dFU9OrpKRcOWyC4IfJNIsTT/wCr9eGVttTVx0Xt4zJk5/D\nYinB7b4CSTJz2229q8xkWcZmO8Ts2Wa2bx9+qKbU1y0+FElYtkyQks2bhTV3OGzbJqpyThhgG4mS\nkoGvOwaBUUMSYGRFi5FwRu4ZbP/hdpweUTOTbkonJ17En+aNm0eZqozVq1dz//3388wzwQY/ixcv\nJv/OfBYvXsyT5z0JQMHcAg6UHCAj3Ut1tQJZlvniiy/4yU9+MuTzbGkRg8FQIyO/oVJDQwNJPkFC\ncjIYnBqanYmgAYejChCahGeanaxxlHFRUhLZferCmprEzfPLX0JNjYgkRClcZEYfD5JQ0a1H8EOl\nisPl6k0SolRR5MTldEd7xiAemgUFoob9uusirNjVJVIJzc3w2WfBXrQJCbBqFTzyCNx5J9K+UvI4\nwtpzzhHLS3R4TnGBo4UDZmf3LouKBN/oC41GqLyHQhKGa6Tkh0IhhHaVlaEjb8PBtqptVHVWcYJr\nFUVFQ9uHP5Kgavcixavo6NhMdPRCFIrgqiaRGvTS3Pw6LlcDycnDzMX0gV6fG3J+aqpoMNfe7qW9\n/X+ceOI4JvUpR1++/CyMxngeeEDYEg9nRF5aKnwGUkP0vpswAbKzRcohEknYulXoWQZyHv5Syb4V\nG2OIjFFDErQKRZB24HhBqVAyPyO0UmpBxgIeLXyUt+5+iz/96U8h13G4HZjuM/GTtB4CsOjkRRz4\n1wESptmorjZw7NgxWltbB126GYhwokXoIQn19fWc4KPGSUmAQ4lrzXTaVsfyyGf70R62smiRnSKH\nSJs0OJ1BJGHzZvF36VLYulWPx6PCqHKHjCTY7dXY7eVB8xUKNSbTnCARVvD2PR4JfoRzeZycOJnS\n5tKI+/suoaNDNF/avz/CSi4XXHSRWGnjxvBm9bfeCk89BW+9RdzGvXQQzbbcDKY4uqhSODADuFqp\nsomo086d4iEbKpIAIprgr5IYDMrLRY44JXIn9QEhNlaQhJFON6wrWUdGdAZLJizm001D24c/kqDs\n8PCkZRz7b72ep5/eHXJdvzVzTc1DaLVZmExDf4YMBZWVLqCGO++8k5UrVwYt37QJ/vQncYkNRwC4\nf78I/YfTJS9bFlmXIMsikjBQkWplpSC7Y5GEwWHUWNtN1utRjaBocaiYP24+TdYmjrUfC7vO3sa9\nuLwu5qT3pCLOPu1scADqGqqr4YsvvgAYNknQaoUQqy/8/Rv6lkHabDC/I5W2jnTMiYd4wL4DgPNT\nxE7q+zqOIN4lubkiAp2XJ2G2GjCqCBlJKCr6Hnv2LAmadu9eQFNTiBqkPnA4KrpFi36IzpOtQetO\nTpw8FkkIgD+NVVMTYaUXX4T//Q/eeCOyPeCsWbB4MTz8MGnb1/MOZ1PmdTDDaMQYLyT4UR4LjS7x\nctu2TYz6QoWGQZCESJGEzz8X1rh9UVEhRo0jceurVGJEOZLtJio7KllXso7Lpl7GjOkKDh4UbYcH\nC38kQWrzcECGLVvOQJZD6xv8IuOOjs9ISrr0K6/uEddXHePDfJFz54rfa7iuhaWl4a8nELqEfftE\n2WIolJWJKOhg9AgwRhIGi6//rezDlJEsah4G5o2bB8CO6h1h19lVuwulpGR6Ss8Q/4xTzgA1mG37\nqa6GnTt3MnHixO5UwFBQVCTc3kJVTur1emJjY7n66qtRKpVERUXhdFbT2ipu3jlZ2czQdBC9TBS1\nX5U+BSUiktAXmzaJKAIIQmK2a4hRK0gx9h7eud1mbLYjZGf/lblzS3tNanUyFsu+iJ9HluUwkYT4\noHQDQEFiARXtFVhdQxiifgsxIJKwdi2ceqoYhvWHW26BjRtJqC7mXc2FNLlcpGg06GIFScjQ6jCj\nQZZltm6FBQvCv8z1+sgk4bXX4LHHhHtoIHbtEp0eRwIOhyAJI/VObbO1ccYLZxAdFc2di+5k+nQR\nYn/8cVFquXFj+NbSfeEnCbR7aMGN16tix455Idf1RxKAEU819AevF1pa1EAtmZmh040Gg4hubts2\nvOMcONA/SQDxfAoFf7XNggUDO2ZJiSC6fV0kxxAZo4YkFHxFqYb+kGRIIjsumx014UlCYW0hJySf\n0KteWqvVYsgx0Na8l6oq2Lnzi2FFESB8ZYMfr7/+Oo899hiPPPIIRqMRs7mMpib/+WQwRdNB3gZR\n2aHRJJKi0QRFEhobBVsPJAldThUJan1QDwWrVcS54+KWYTBM7jXp9fndbW/DQYixbINKN8jIHG45\nHLTsu4h+SUJpqdAg/N//DWyHF14IaWm4VFq2x66kyeUiSa1GGy1IwiRTMl51DBXtVWFFi370F0k4\n4AsIvfdez7zWVnG6vkrMYaO1VRRuhODB/aLJ6eTmQ4dY+uWXLP3yS075cjenvHkH9V31fHDlB6QY\nU5g2TYiDf/ITkSdftkyU7w0EFo8HBeBpc9PmEL4TH3wQ2iZdqYxGkjTodHkYjQMo/h9BNDeDx6NA\nr+/AFEEjtnDh8EhCdbW4XiKRhLQ0sTxcymHbNkEwI1U/BGLvXhFFGLNdGRxGjSZh6ighCSB0CZFI\nQriWyONnjOfg+hK8Lti9u4w//OGcIZ+D34756qvDr7N06dLuRlfPP/88XV3lNDaKSoCoqAxczmrm\nK8VQR61OIEXTSr3TSWOjELa73T3ueX6SkJUFXTtkokPYkYlIgRSynFKny8ViiZQsDyx/nNhrvkg3\nhCYJAKXNpcxIPQ51bd8wBJIEWQ7xsHv8cdHVKIT1d0hoNPDXv7JhbQVSvY4Wl4tktZooH0mYnTCR\n95o8vPXlXtraxo8ISXj3XbjtNvHv998XI/MIHd8HDLu9JyxdXS1Eb33hCtE3xSPL/Keujt+Vl+Nw\nO4izHgJkGjUZeBVpbFn1NpMShXjPaBRuhP7P+Yc/wO9+JyQg/QkvuzwejEol7jY3beZodDobH3yg\nw+sNjs5IkrjHkpMv/8pTDZGMlAKxcKHQvd5+uzj/hAT4xS8Gnjby62qmTIm83rJlwpUxFLZtG3iq\nAUQkYTDrj0FgUCRBkqTVwAXAZMAGbAV+KcvyoT7r3Q38CIgFPgdulGU54jAzcxR5ZM4fN59X97+K\nw+0gStWb7dvddvY27uVHJwa7d8yYP4PSl8oBsFrjmDcvdDgxHN54A155RfQ/b2oSD76BKrXz8/PZ\nsOEQVqt4iEVFZeByNZObW4PbrQKiSdVoaHC5ePxx+P3vewyYzj67p1GJRgNW3KSEuDQsln1otVko\nlcGETqfLo7n5zYjn2NcjwQ8RSWhHlr29Sr3idHGkGFLGdAk++EmC3S4qG3uNoOx2YYf4gx8IJeBA\ncdVVvL0NtF4nMpCs0aA2CpJwYsIEaCrj3QMHkaSziHQ5RxIu2mxCoDhjhjDt8jeTevttmDOn/xfS\nQLB/vwhhA2x+tZy4uSKCpp2gRZet46m6On4Q0CQuEBLwo7Q03v7wfOJUSvIS8tgdvZyocctYlNmb\nGen1Pb4Q990n0g433yw+S6T3ucXrxaBQYGlx0+UysnLlZj744GS+/DK0dGT27C/6FQEfD/hJwoQJ\noZvY+bFihaioffddQfSOHBFahVNPHdhxSktF863+yNWyZfCvfwlCGXi9O50i0nrTTQM7nssliOr1\n1w9s/TH0YLCRhCXAQ0Chb9v7gP9JklQgy7INQJKkXwK3ANcA5cA9wIe+dcIGAhWjKAY0f9x8nB4n\nRQ1F3RoFP/Y27MXtdfcSLfqxaMEiXlR/Bi6A8Zx44uCaoLzyCvz3v4IhX3mlmDcYkrBunUjSNTVB\nTIxIvKWnF9PeHk9ZmUSqRkOp1cqWLcITITD0GwiLZEcrBTtEWiz7MBhCq350ujzc7raIfSIcjgoU\nCkN3Uyc/xP9l3O6O7l4OfoyJF3tQXS2qABoaxMO8F0l4/XURbx+o9VwAurogKlXcmslqNWqDIAlZ\nBvGiLemo7A61h0OkSMLhwyLy8fOfw7XXwscfC7e+998Pbr4zVPgrgQAO/tJGEeUASGqJufvmst3b\nSbZWy10hDBlmGAwkyh083naARy97g/Mmn8ffKyv5Q3k5XlkO+2yKjoaHHxYWFK+9Fr7JEMBBq5Vc\ntLS7RKv42NgDREefzHvvhSYJCkXkl3Q47Nu3j1tvvRWXyxW0bNq0aaxZswZdhNaYgiR4yc2NXI6e\nmir0JCB+2/x8YX40GJIwaVL/kYdTThGcN5xx00BtwI8cEcRiTLQ4eAyKJMiy3OunkiTp+0AjMBv4\nzDf7NuBPsiy/41vnGqABOB94eZjn+5VgZupMNEoNO6p3BJGEwtpCVApVL9GiH/lJ+ZBZB2UuUlPn\n0eppJe/veZid5gEd1755E4nTlEQrTuSee8TofqD+6Pn5+Vit/wUESUhOFiRBpyuisjKBvXshZZaG\njW3t7Nsq2qeGglf20iFb0SpCpRtKSE0N4eyEIAkANtuRsCTBbi9Hq50QFEL1EwO3uy0kSdhWPYzk\n57cI1dWiJvytt0TKodcDb+1a8UTtU9c+EJjNoEoSL5UkjQaFzozk1pGpFUPmRlUz5/QTptXrRXQj\nFPyphrPOEi+T994TkYTOTjh36GakvVBUBDl5HuoqrFR4wbTFREF8AUWnFnHst8co+7WbWUYj14Yq\nygde3S9i2v7S6CkGAxavlyqHgwkRopznny+mG28UBL8v8vJEy+ptnZ2skuNoR6T/7HaZFSvEd/G7\n3w3zwwdg/fr17Ny5k4v7MBav18vTTz9NcXExb731FvFhEvl1dSBJzWRlDVzdJ0lw1VXw97+LUf9A\nMsf9VTb4kZAgUqKtwcVPmEwDN+Hau1f8HSMJg8dwNQmxgAy0AkiSlAWkAt3dumVZ7pQkaQewkG8I\nSYhSRTEzdSbPFT9Hp0O0bTNoDKQZ03j/yPuckHxCyC5wufG5kO2BsioSE+fwfPHzmJ1m/rzsz0gD\naLj7y/sLaM6/j+0vpHPP6tTudMDDOx8mThvHldOvDLttfn4+gq8JMeKsWSJ/4HTuw2abx7FjkDpf\nQ73Did0ss2RJ6PNp6GrA7PaiVzt6maW4XO04nTUYDFNDbqfTCTMqq/Uw0dGhPShCGSlB7yZPfVGQ\nWMAzRc/g8XpQKr7bXqrV1cIDyU8SulFeLiTgzz8/pP12dYFick8kQaE1IzlNxKlUKPHijrFz4jwH\nED6NESmScOCA8PBISBBE4eWXRWlvZubIeRoUF0PiSa9jq1pImcZNc2Iz+lw9E++eyMEfHMS7UkPW\nyeHNGHZU72BCzARSjYJETPG96fZbLBFJAogX4x13CB+LQDQ3iwDPlbc6KLfbmYOOI4j6SbtdxWmn\niUhKczPdturDRWFhIYsWLeLpp58OWnbzzTdz9tlns3jxYp544onufjK5ubnofZ+3osKJLNeELX8M\nh6uuEhqNt97q3/FSlkV66PTTB7bvceN60qFDRUmJiH6M1Pf8XcKQSYIkhoMPAJ/JsuxXrKUiSEPf\nytYG37JvDFadsIr7PruPB3eKhi5mhxmb2wbATXNCJ8LGx4xHkaXASzlRUfm8uO9XnDfpPH664Kf9\nHq+tDX5qBuLK+LTuPZ566geAMG767Se/ZXrK9IgkIScnB2gBRCRBpTKiVMbg8XSgUCRQXAyXazTY\n8aKO9TB3buifvqqzii43GHQWKitlsrIEmfBXNuj1oUmCSmVCo0nFZgtfiWC3VxATE2yT7ScJocog\nJydOxu62U9lRSVZcVth9f9thNouXUHa20JL0IgmffCKGc0MsEzCbgVgXWoUCo1IJGjOYTUiSRLRX\nQVtULHHZpUB4pX1/JMFf5njWWaIi4MknRephJLKMsgx7imS0J/+FDPl1ahVqGiwNeL0QdU4q+ilV\nnP2wlagV4UnO9prtLMjoqaUbr9WiVyjYb7VyRj/hvPT00FGEykqRc19X3Al6mObWskupAA+43VHk\n5Ihz//DDnvTicFFYWMi1114bctn8+fP5/PPPWblyJYsD7OqvuuoqnnvuOQDKyhxEKn8Mh5wcIQp8\n7rn+SUJTk4gMDCSSMFIYs2MeOoYTSfg3MAUIfup/C/DTBT/t9XKXZZkORwf1XfVMiAmttlEr1UyY\nMgFLjhWbK4X9jSXcu+zeAR3PX2UwrUDHe4ff4wezBEnYcGwDHY6Ofp0Ho6KiyMoaR22tlcZGvW9e\nBlZrB1qtSDfc5mv5O+1kJ1ptGJLQIUiCUuHlyJEusrJEblJUNijQ68MXtet0ef2ShJSU4M6aarUI\nfYYzVALR6Om7TBL8pCAjQ4yqepGEjRuFOZKvq+lg0dUFumhR2SBJEl61Ga/dhCyD0WqgTZPAh63/\npm1XsA7Hj4NGaMiAtbtEA7VLplzSnVY6cEAIFAGWLBFhYrM5ONXglb2YHZFTc0qFEqOmt16mvh5a\nozeicRWzyCGzX23iUHkXp90On30m8dCdOcy6Zy/O7W4IoRtweVwU1hZy0akXdc9TSBKT9XpK+zCf\nVpeLDl/b53CIVqlIUKvJzBSj142NnWROiiK2XaJNI6PDjNUajdks2hv/979wxRXDJ0y1tbXU1tYG\n9ZsJRH5+PsXFxez3lRf861//YsuWLd3Lq6u9QC3jxw/eSvHqq4WRZ2Nj71bifRGpZ8PxQknJyFTR\nfBcxJJIgSdLDwJnAElmW6wIW1SPEwin0jiakAF9G2ufPfvYzYvp0a1y1ahWrVq0ayimOOCRJIlYb\nS6w28oM4LzGPqhPNVHws1j8t57QB7f+Ir/bj3AVTebD4jzg9TjRKDa/sfwWAZmszzdZmEvXh42X5\n+fk0NrbT1BRIEvYRG5vA0aNgcguf+IJFTkCs89xzYlS3UXSApaqzCqesAZwcO9YG9JAEnS4HpTJ8\n6FWny8NiKQ65zOVqx+PpCHJbBFAqTYAiZLohMyYTvVpPaXMpZ+SdEfbY33b4KxuCSIIsi1TDMHot\nd3WByugkyUcivSozOEw4HCA36VDqc3iq6EaeKnoi7D5kFcgL4YZ3ZGRkjFcYOTPvTLxeOHgQTj6n\nkjNe+DFurxvtpLtw7JvHKadowJeG83g9nP786Ww4tiHsMQAkJP6y/C/cufjO7nl79wKL/8IpzhWk\n4MBiG8+aa64lziBGt7f8PZ4fT4zlouur2Pm7xqB9uk93Y4+zM39c7zTZFIOB/QElG60uFxnbtmEL\nUUoZiChJ4uiCBYyLimL+fNik7GBldDTuA27a1RKxhiYUinjKyuDXv4ZLLxX3YahGboPBLp+SMBJJ\nADAajd2VV2VlZTz77LO0tLSQkJBAU5MKqCN9CCUnl14qyltffFF4SYRDaakwiMsN3UJixGGziefr\ntyGSsG7dOtatW9drXkffPNcIY9AkwUcQzgNOkWW5MnCZLMvHJEmqB04Fin3rRwPzgX9F2u8//vGP\nQVcDjEbkxOWwT7cXS9sqrs2+PKiEMhyOHhU524tOXMafC+/g88rPWTx+MW8ceIPzJ5/PGwfeoLSp\nlCUTglvL+pGXl8fmzfU0NYkbPCpKiI+SkkS4dPfHGkiDzBk9RSabNone8H5UdVRh0KYAVVRUtGOx\niNyk2byPqKgTeoWUFQpRxuSHTpdLU9NryLIcJE50OEJ7JABIkgKVKjZkukEhKchPyP/OVzj4SUJ6\nuiAJO3f6Fhw9Khb6jS6GALMZ9HoXEzSCRLoUgiR0doK5QkPczBNpussTcR///CesXg0Wi8yCJxbw\n961/58y8M6muFqWRb7f8DanlMLPTZzP10hfZtPtvrCu9gOtmiU5VD+18iE+OfcJDZzxEiiG8duDT\nik/5xce/IC8hj/Mni84/T+75D+T+jxuUr3MQK7JHQ0L+Pva8PwujEeasdPLw9mkUrGxgQkzvKIC9\n0o7+oToybstgVtqsXsum6PW83dzcfT3/r7UVm9fL61OnEh3KBhVwyzIXlJTwdH09v5kwgdkLvLyZ\nYmZBdDLuVjctChWxsY3ExcVRViZEi1dfLcwvlywRPiVDRWFhIUlJSYNKFfgJRWFhIcuXn05Hh464\nOBuqMJ8vEhISRBXCww+HF7ECfPSRIAia4N5WxwX79wsu/W0gCaEGzrt372Z2JAv2YWKwPgn/BlYB\n5wIWSZL8d3OHLMt+R/MHgN9KknQEUQL5J6AaiFxE/y1BTlwONQrxUU+JHfjQ4MgRkdebmTqTNGMa\n7x5+F4fHQbu9ndUnrebtg2+zv2l/RJKQn5+P3V5FQ8MsQOomCePGJaBQwFMPq+D3EnE5PSTh0CHB\ntB0OUWpU1VlFjC4DqOKdd9q515cteeWVfbz33g956qnex3z88Z6qO50uD4+nA5erJag9tt+Nsa9H\ngh/hXBdBCELL2srCfu7vAqqrRQg3KqpPJGHjRsHWloS/LvpDVxdERzlJ9jmIuhRmcCZRXQ0dxzTE\nLHH0uw+DQVxHsixxx8I7uPTVS9lVu4uWA+LhVaZ8l0/Pe7q77fd1b17HTz74Sff/f73h19wy7xZu\nmXdLxONcNOUi6rvquWr9VWy+bjP/3ftfXrKtIeHYDUxPPIHUCc187/rVKDN3kZj4PwDO/ncdh6+O\n46Y3Q4+OMxXpXFd4W5AYeYrBQIfHQ53TSXpUFO+3tjLdYOD8fqzWL01O5om6Om6OqWHa/IdA+jGp\nrdHYm5ppRUNcXCPJyeO7o4cPPSSarF19tfCRGGpnxcLCQubMmTMoA6acnBxiYmIoLCxkxozTkWUF\nqamRIyWRcOutokPp2rWR1/v+94d8iEHD37OhP+OmMYTGYOniDQhh4qY+868DngWQZfmvkiTpgccQ\n1Q9bgDMieSR8m5Abnwux5QAkuwZupnTkiGDXkiRxZt6ZvHf4Pdrt7eTG5zI3fS458Tn96hLy8/OR\n5QpqapxAFFFRQhKs1yeQnw/btkqouzR0Knt+Cr+Xfnu7qMGv6qxiRtJEYBu//30bDgdIUiuJiXWc\nc85UVqzoOd7atULZ7ScJer2/DPJwEEloanoNvX4KGk1o/apaHR+WJGTHZvNy7TeiMOa4obq6x3M+\nPV3kfZ1O0GzaJArt+6TqBgqXSxBEq8ZFkkbswyGbwZHNp58CLRo6lS7cXm/EBmz+1itWK1xYcCFZ\nsVms2baG+dXPI6lcLDsxr5sQADy48kE2V2zmyvVXolfrSTGmcO+p/et3FJKCZ85/hpOfPpn5/5mP\nV/aSffBB5ku3YN23F9M0AwUzO/m8qietUI2VOY+08YA0i76ZgvZ2OPdsJRs/uwZ7nQNtWk/kz1/h\nUGq1kqrR8EFrK9eFKaEMxPVpaTxdX8+uqnXESi+idvwA824jjUfrafPqmRTbSHp6HP8THIaYGJFu\nWLoUHnwQfvazfg8RBFmWKSws5McDbYnogyRJzJkzh8LCQs7wZfMyM4cuVTv1VCHYHE0oKRGCX2Ow\n9csYBoBB9W6QZVkhy7IyxPRsn/X+IMtyuizLelmWT+/PbfHbhJz4HDDVolB6qKoc+JDg6NGeHN2Z\neWdS2lzKiyUvdgvApiRNGRBJgEYaGsST0B9JUKsTulu6xsk9/Rs6O3usbNvbxd/KjkpSTMLTdtGi\ndlatgpUrReOmFSumsmoV3dMdd8CePfClT22i04kP0Fe86HZ30tz8Oqmp14Qd5ahUcbhcIYqhEd9p\nZUclTs93gmeGRCBJ8JeD1dXKsHEjncuXc2ZxMYt372bx7t2cWVyMuR9xnR9dXeKvWekk2adJsMtm\ncJr45BOIsmqQgaYQ5jyB8JMEi0WIC3++8Oe8vO9lXtpUhBx/kLuX/b7X+qYoE89f8DyFtYVsLN/I\n4+c8HiRIDHssjYE3L3+TFdkreGfVu7R/eCtTCiS69nZhmGYgxZhCg6VHEnXMbifHpGUqnxt8AAAg\nAElEQVTuXOEzETjNWdKC4tzr+ExO5nerunodJ0urRSNJ7LdY+LKri0aXq99KB4CF0dFM0etp7BA5\nofQ6F7t2KGitdNPp1hEb28bEiTqqqwVBAxEIuvxyeCK87CMiqquraWxs7FePEAp+kuB3W8zNHT0W\n+SOBscqG4WHUNHj6tiA/IZ8Lpp5LeoaH8vKBbWOxCBOTHGE1wPLs5agVaiwuCxdPEXLsgsQC9jdF\n7o0wfvx4lMo22tpUyDKYTLOJiTkFg2Fqt3NjpqGHJAR25GtrEyrvOnMdGdETUSj0uN3tvvPbByjR\n6/N7He/000UTFv+DTak0oNGkBTV6amp6Da/XQXJy+DqvSOmGnLgcvLKXivaKiJ9/JLGto4M/HDvW\nPT1SU8OmtjYanE7sHg92jwdnPwK2kUQoktCy/TDU1fHS0qV82NpKvl5PmkbD+62t7I3USCEAZjOg\n9mJVeEj2JYltHqFJ2LwZCpLFvLp+uiYFRhIArpt5HTHaGLbtaSd1YnuQvTHAwsyFPHLWI9x36n0s\nz14+oPP1IyM6g/eufI/ZMStpbYWp4104a5wYphmI1qfSZGnCK4vfp8xuJyuM18GOmh3Yp7/ED+fW\n8f8+jefOn3q45x645x64/14FcetyeP4fUbxc2k60UsmiSLaTPkiSxI9Sk4hzC1J/gsfBjh1gqXPR\n6TSSkNBFVpbIk1cEXNKXXCL6tYRxj46IwsJCoH/RYijMmTOH6upqSkvbAQ+TJsX1u803Cf7GTmMY\nGkZNg6dvCzRKDesvW8/3HmXAJMFf/uiPJERHRbNkwhLK28uZlSrEVAWJBVR3VmN2mDFFhbZMVSqV\npKYqqKlRY7GA0ZjMrFmbANFOVaOBySkaDvgcIA8FdNxob4dacy0yMpkxmagccd1CQqt1H3p9HgpF\nbxGmSiVq3R99VLitabWhyyAbGp4lNnYZWm14Fze1Og6b7VDIZTnxgj2VtZWRl5AXdh8jhUankzOK\ni1FIEnqFAhlodLlw9+kLLAHvTZvGyoHaYg4DVVXBJEH+ZCMolTwXH89ytZqnJk+m0+3mtc8+o9Ju\nZ9EAUhBdXUBMj5ESgMUlIgmdnTA3O4o9DJwk+LmJQWPgpjk3cU/zZM48LzyZun728Mz0/Y2CchQW\nWoGD47380pKHRx1Hi7UFozaeeqeTFKWXF0te7CYOfrx18C0S9Yk8+Ho8leMbefJfcUiSWEdpVNKp\nSGFnm4p6ex3Lb4xDPcAORpeY2jiCCBPMj7Xyx73QqfHgcEeRmGjrbkBVViZcKEGQbr0e1q8XItC+\n8HiEQLSqKnjZ558nYzA8xt/+Fr4qYdo0oRfoG8zzE4vNm48CaWRlDc4jYTSjrS2EO+kYBoUxkjBI\ndO7sRJOqQTs+sgvbxIk99cD9oS9JAHj0rEexuCzd4fmCJFFUfKD5AHPHhW9BPWGCnpoaYVgSmINb\nvlzcLA+aNXxaL0LHhw71mOC0tQk9AkBmdCaW1ljc7nas1kO0tLyDyRT6mNddB/ffL5pTXX65IAld\nXXu6l9vtFbS3b2Ly5GcifgeRIgkZ0RmoFCqOth2NuI+hoMXlot7pZKr/LQf87tgxJEni0Lx5JPpG\n106vlyM2GwesVqweofS/7cgRdprNx50kWK3CfMZPEuLiBCEzFm6kfMUKtnR18ZzPrShapSJGqaTS\n0b/YEHwkIdZnyaxWI8syXa4uJKcJGVg8Vc1/IKjFeF/4rXgDAxg3Tfs193TpOHVwfc4Ghf37BVmN\nbbPQppLYl+7BWSFB/DwaLA00+0p9dx59k2c2h24UcfX0q9GPi+KZJ7y0fSyiYB2fdmCab2L93438\n8ex4KsskfjfQnsSA0l6EjISEzOIMCx4PWG3iXk5MdJORIc67LECPq9OJ6oDXXgtNEl5+WTg0TpkS\n/KIvLx+HQpHBRx+FPh+vFx54QNhX/+MfvXsmTJgwgcTERPbsaQTkQbstjmbsE5nS7nTrGAaPMZIw\nCDS+3Mj+K/aTdGESU18O7Tzox8SJ8MEHA9vvkSPCYCZQNN13xOw3FdrftD8iScjPj2XrViFs61tO\nlZgIqQ4NDU4nXlnm8GGJGTNECWR7O9DhIwkxmRxRxdLW9hH19U+i0aSRlfWnMMeDk04SXguXXy7E\ni01Nr3SXjTU0vIBCoScx8cKI34FKFR9Wk6BSqJgYO5GjrSNPEv5SWckD1dW8OnUq5yYmssds5vG6\nOh7Ize0mCAAahYIpBgNTAsjEv2trKbPZRvyc+iLQSAnEC2Jcukz6wU3888Y/YVAouCDg4snUaqm0\n20PsKRhmMxAnSEKyRoPVZcUre9EpTNiABXMUpLVoeLy2lnkmE9PCqL/6RhIAKstEtcTk8P5bw8b+\n/eIatO+3oJ+sp8qvj46bS0NXA3ZlGgCtbftYMn4JH14V3HfYX9WQ9v000r4v1j/6q6M0PNvAFF0S\n3jQb1OhYGT/wTrVmcyEqbT52ZyO5yS3o9eC2ijd7UpIHpVI8I44d673dRRcJvU95ee++BB4P/PGP\nwrHynXd6byPLMomJs7ntttu46667wp7To4+KromtreJ+9QWOusWLH37oN1IaeqXMaENJiSBj+fn9\nrzuG0PhOahJc7S4a/ttA/bP1QVPLuy14LME14Y2vCoKgilXRvrkduU/ouS8mThROcAN5h/jLHyNV\nLhk1RsbHjO9XvDhrViLgobAwtNAsRaPBJcu0ud0cOiT6AcXGCpJQ1VlFdFQ00VHRqNVxWK37iY8/\nk9mzC9HrwzcO+sEPRGe/igohXvR4OnG5RH15ff2zJCVdiEoVWZSmVsfh8XQiy6Hr8XPicihrH/ky\nyAanE5csc9G+faxvauK2I0eYrNdz4wDMZLK1Wo4O8GU8HAQaKfkxL+EoRmsDz+fkcEFSEoaAurnx\nUVFUDSqSIF6sSWp1dzMyvcqEySQaFD1XUECr283MwkJuPHSIrhCiyL4koaUFfvtbEfEYygO6oSG8\nzXMgSkvFyNpSYsEwzdBDjuJOpK6rgWM2G2pJoqa5mPyEfHRqXdAUSkwbszAGZ52T/FY1jLOhqtWT\nMYh29mZzIYkx84jRpuF2NzB3tozD97j1uxFmZ/eOJIAgAVFRIuUQiHXrhFbhj38MPlZJSQmtra39\n6hFuuEEYHb30kkhZTZjQM23dug5ZXo5K1UTsEJ07RyNKSsQz7qvyZPg24jsVSZA9MnVP1HHsN8dw\nNYdXa0tREnHL4jDNNoECPF0eqv9ZTfJlySSvSqbknBJsR23oI6iA/aOAysr+G/MFVjZEQkFiwQBI\nwkRgA2vWTMZg+ASj0chFF13U/SBM9d0t9Q4nhw+rueACEb5uawNHZw0Z0eJNlJl5B0lJl5GScmW/\nddeXXAI33yzCoTfeKD7s1q09vqx5eQ/2+9l6mjy1h+wimROXw5bKLUHzh4sWl4sz4uOJViq5yBeb\n/HD69AHlnrN1Ojb6y0KOIw7WuEClZNy4nnNarNhG4aRJHFQoeDClt/nQeK2W7Z2dA9q3v2+DSaFE\np1RS7bNFNqhMTJstwtLL4uIomTuXf9fU8KuyMiZqtfyyT0g6ULi4a5cYEVss8PbbQys9W74c5s4V\nI95I2L8f/u96ma6Puog/I54qRyvZWi1lQGFXPUq9nYlaLUdbD3PJlIsi7ywA0QuFQDHxSyfKcTbc\nLRoslp7PGQler5OuriJSUq7E4ajG5Wpg2QI35i3i3ktKEkP47GzYvr33tiaTaOO+fr1orQ3gdHr5\n4x8lzj5bZvp0D/5Ck9raWv7617/yxBNPkJKSwsKF/bTqRLgijhsXHOU8eLCBV155mQkTdiJJP+z/\nQ35DMCZaHD6+MyTB1eqiaHkRXV92kXJtCln3ZKFJCaaX9mN2Wt5toeWdFuqfqe+en/ajNPIezsPT\n5QEJOrZ0DIgklJf3TxKOHBEhxv5QkFjAu4ffjbjOCSecgF6/mmPHHuW6634PVLJjx45uG1Y/SThY\n56K9XYzy/JEEh6OdeJ3Iu8bGntLv+dxfUcF5iYkUGA2ceqoIg95xx1SmTHm5uzJCpYomLm5FP3vq\n3eQpFEnIjsvmmaJnQro5DgetbjeT9Hoez88nXq3GI8ucNsDcc45OR63Tic3jQTdUB5x+IMsyv0nc\nhfamJPT6nO75M+3beXjlZaRpNJwa11uNPj4qipf7RDjsdvHC9r9gCgpEu4euLiDOSZJGvLj8kYQl\n80ycHJDx0igU/DQzkw/b2tjU3h5EEvyahMOHhZh1+nR49VUYSnq7s1OMAI8eFfnzcPrL1lYRrZua\n7MDT4RGRBEcdFyclsab8ILtsChLtdjLUSg47OoSHyQChSdagzdFi2WFmzZn/n73zjo+qTr//+04v\nmWQmFRISEjqhRaQjAuqCiNjWXtDvT9e67tp10cW26uraULGy6lrXyqJSFHERhdAJXSAhIYVAeplM\nn7m/Pz6Z1JnJpAprzus1L8jMvXduJnfuPfd5znNOEnc8KfZndOuE+Faoq9uLLDsxmcZRU5OJy3Wc\nP/8/D/f9Q01ERBUGgyAgaWkit0GWm1cRL7pIVOeKiyEmxkXfvg9QUfE82dlj0Wiymr1XdHQ0Cxcu\n5Lbbbmtlax8MU6eKR1MUFUXw2WcLGTz4bEBUaPzCa6USpkw5+XwGZFkcR7PCc8bvRRD8ZkhC6Zel\nWHdaOSXzFKImBf8yGYYYMAwxkHxnYIWvwqzAONpI9U/V9P2/vkG3k5QkvlxtTTg4naLaMHBg6OUA\n0uPSeWnzSzg8joBR1QAWi4Xi4lfp21fm7rsP8Pjjeg4dOtRAEhLqG5G7DggF9+DBjZUEr7MWkybw\n5ERLFDmd/CU3Fx8w3Ghk3jzR76yslIiPvySsbTRFY8hTkDHI6IHUues4Xne8Ic63K1DudhOjUqFS\nKHi1nXXxAfXl51yHo5lWoSuxtbaWCp0DzbRSZHlAA0HqX7WZpdMf4w/x8ShbkKYUnY4Kj4c6r7eh\nDfHnPzd3wVOr4bvvRCVBHeduGH/0BywtfMDEkAB6zOlRUTyRn9/KXEmpFGXyzZvFttetE62GjmDn\nTvGv3S7K7DffHHg5vzA4+Yc8PCYlERNNFO13kqbTYbFnc0DXh1q7nQFKQZjaQxJAtBxqMmu48qHB\n3IEg8+GQhNrarYCCiIgM1Op4rNZdKLweqlBjsZQ0EOIBAwQhqqgQlsZ+nHde44VZkhxUVDzMqFGH\nuPvu5mmyWq2WuXPnYjKF950NhcTERPr06dMgWpw+XYif/YiJgTvuEPbRJ0s34tgx8dn2VhI6h98M\nSaj8vhLTeFNIghAuok6LovLbEObkCLFMcnLbJCE3VzDesNoNccPxyT4OlR9iVEJwuW5kpIILL4TP\nPtORkNCH7OxG34IIlQqjQsGBg0JTMWiQ+NJXVoLsrAkZINUUa+rN2f0mO3PnCnHVypUdi71tbDcE\nMVSyNI5BhiIJsixTXv4NXm/rNEGNJhGLZUaz5yo8HqL9Cq52YmB9cMVhu73bSMKXZWUggyvWwd66\nOkZGREBdHTsSNdRa9FxgaP1ZpGjFqGqBw8Ewo5FDh4SXxdNPi5O81yvuVi+8UIzdKSY2Gin5KwnB\nyOJ0s5m/5Oay3WplQgvPAKNR9M1PO63jBAFg+3ax/owZYr9DkYTxUgXOpccY8voQKqPALcska7Uk\ne4+zUzkSq93OcK04Vv3HULiInBxJyb9LsBi9REYqyQ7TEq62ditG4wiUSgMaTQIu13HcDjdVCiVR\nUc1JAghdQlOSEB0tRIY7d8IPP2whKiqX5cuvJzm5+8Z/JUnik08+oV+/flitgiC88gpccIEgMq++\nKrwjnn0W1q+HEaF12ycE/HbMvSShc/hNkATZJ1O1poq+NwW/828PzNPMHF18FOcxJ9o+wQOcUlPb\nJgmBxh+DYXisGIPcX7Y/JEkAuPpq+PBDyMg4pxlJANFyyDukIDlZlIktFnGiUrpqSTOHlzDTQBLq\nR+OSkmDsWNFy6AxJCBTyBKLdAJBTkRPQmMeP8vKv2bPn/CCvSkyZUtJgGe2TZSrcbmI6SBL6aDTo\nFIp2ixezssQJOBwvpi/PL0W7Kx7fpHKWlZcLkrBtGx+edSbaw0rMMUZoobFMricJ+U4nw4xGFi4U\nkcW3394YyPXZZ6Lk/MknoJvtJr7e7dBfSQjmxTHOZMKgULC2qqoVSTAYxPH+f/8X/mcRCNu3izv2\nm24SRGbXrsB38AeyPNyvPID5dDN9b+zL5lqx7yk6HenKOnZKCuw+H157IQnGhKC/UzBETo5E9shY\nt9UycKCZ7GyZsmXlOPID/L0l4augilRRHbuZyHghItRoEvB4ynFX2KlSKrBYShqqZk1JwvgWA0vX\nXw9er5fExCv5wx/mk5zcdS22YDj9dGGb7TdzGjVKfK+TkkS+xIIFIvL71VeFHfuJgpwcUSlrOaVb\nVCSOyc6EZvXiN0ISrLusuMvcWM7sGiexqNNENaL652riLw4enJ6aCr/UhxcuXSp6tU2SZwFxodDr\nhR9/W4gxxBBniGvTeRGE8CshAVyuS8nOfqTZawkaDUWHlQyuvzHxaxLUzhoitW07ysmy3KqSADBv\nnpjFdrsbx6vChVIZASiDthuMGiMJxoQ2vRIKC1/EZJpIRsYPzZ632w+zdesorNbtREeLJmWNx4MP\niO5A4h2AQpJI0+naNQZZVSXKySAqTaFgi6ujOsrO8LyBxEyRWVZWxoP9+1O9ZQvLpk7F9W4yR5Ok\nVjPgSVotEpDvcJCVJRTtb77ZPLHTbBaEbtIkqIpyN8RE17pqkZAwqgNXRtQKBVOjovixqor7WggO\nlEpxog4USLm1poanCwrw1U8FTYmK4u4gH8D27aIaMXeuOIbff9nFwptbT2sk/6cIk+xm6JIMJEmi\noJ6sJWu1DDSYUNmL8OiTsNZkt7vVAGAcZURhVFCTWcOgQWZ+2ehmz1t7kLRSK12M7JORXTKoXbB8\nD3H2awFQq4Wo1FlTQqWUzDBzCSrVGED8DSyW1mOQfmzYsIGSkhIuuij0+HBXI9A0DQh31auvhiVL\n4PnnRXvpRMDnn8OaNeL80xRmsyCsYfpf9SIIfhMkoWpNFQq9okGx3Flok7To0nRU/9Q2SVi1SiiY\nr7wSfvc7ceJricGDwz+Qh8UO41DFoTaXU6mEGHLJkqlotc3PQn00GnblqZlzhvjZr0nQh3BzbIoD\nNhtFLhepOl0rkvDoo/DzzzBzZni/jx+SJIUMeQKhSwhFEqzWXVRV/Zf09H+jVDYXlYrybxS1tVsb\nSEJF/ShfR9sNIFoOOe0gCbfdJsq3u3a1Lep74kgZ2flKtr9p4fNSD9f88gtHnU5WVVTg1GiI2tqH\nFfGiZdAUaoWCRI2GfKeTpQ+K4yvQ3X1amtAOZBQ3aTc4a4nQRIQUh84wm3k6gC7B4xHHXaDU2jtz\ncsi12xkVEcFRp5MVFRXclpiIroXg02YTEwt/+pMgmjedZ2fiW1vZtqT1aOxoYN+swcxKE+wn3+nE\noFBgUalIiEjAV7wJki6itHwnIzpAEhQqBZETIqnJrGHAQB/rvpQxPPgdsf8PhN9mc8heGVddKcer\nPNhXpMApopIA4KgrpppBmM2lDVUzENWEL76onzJBTB+MERyCpUuX0rdvXyZOnNjufe8MmsaSt8S1\n18Izz8Dy5aJldSIgM1PoNz79bWfAdRt+EySh8vtKoqZFodR1nQI9aloU1T9Xh1zG75Uwb54o0336\naed6tSDEV3tK9oS17OWXw4svRmC1plFVVdUw/5yg0mDL17SqJLgc4VUS1lRVoZIkLoqN5dMm6qZT\nThF3G19/3X6SAKFdF0H0lEORhMLCRWg0SQGNmyRJwmQ6tV5UJlBRT3BiOlhJACFeXF0ZWp/ix0cf\nNT7CUf0vLS3lnOhodEol58TEoAS+Livjk8REzqioYNrVOp5+Gh58sHH23o8UnY7Mww6+XyEqCcF+\nxX6DvDiKfY3CRVfbRHG62cyDublkWa2Ma9JysNmEIVhLzrWuqoqfq6v5auRI5sXGsttqZfTWrayv\nqWk2meHz+Ni1S8Lnkxg7Vlx0Z23bTz5qlqSPQR/ReGGWZVi3RclTVzeSwQKnkxSdDkmSiDfG4yt6\nmjvH38x7W3Zz/sAzQv5OwRA5OZLifxYTVVdKiTeeqtOfw3MsBoUi+BdZU34KFYst+O71NZCEyp25\nVHlnYTY3thtA9Pz/+U9xTBw7JkTM778vqnVffvkl559/PooevhUuKBB/x0DnqvR0cS77179ODJIg\ny7BhA9x446+9J/+7+J8vxPicPqrWVWE5q2tDS6KmRWHNsuKpCZ625x+DtFiEbXFnCQIIkpBdEZ6C\nKiMDlEoZGE1OTuPF1VCpx+dQok0RYjir3oHPB1Zvf2qVMeytq2v1cDdpoK+prGRyZKSoJLhcDcZS\nCgWce25rR7hwESoJEoQuIZjrostVyvHjH5KU9EcUisCVAZNpHLW12xp+Lu+iSsJhu72hjB4MBQVw\nyy2iohTOuOsRh4NtVisXxsY27OM0s5nXjxxh7fDhXGMwcPvt4jNftKj1+lFOLf/d4+Tss4WPRTD4\nNSVNKwltTbiMN5nQKxT8WN1Ikj0eUSGxBPiaPXHkCKONRs6tV+eNNBqJV6v5vgW52n/lfqov3MIQ\npZWRI6Hg2QLcO2oovX4Y2tGR+AaYGh7yQBO/u87QrIpS4HA06DESjAlgL+Ryo5VyW1mH2g0gSIL7\nuBv96qPISBw7lkZGxlomTToc9DF66Do8xRLlX5ejVgv2VnakFJdXhdlc0qyS8NBDot2Qmyv0F/5c\nhqysLI4cOdLjrQZoHiYWCPPnw4oVzacffi0cPiz2Y0pwmVIvOon/eZJQs7EGn83X9SThtCjwQenn\npdTtrcN2wIbsa36hGDtWXBBWrGiuXu4MBkUPotJRSYU9+MXUD60WBg/2AWOaiRfVR0W/+Y+2nYzc\nsoV7j9crlYa8ySO2/ozcsqXV4/SsLBxeL15Z5r9VVZxpsRCrVuOUZazexlLwvHliVn7KFBF/O3u2\nuEsKJ7lYrW67knC87jhWl7XVa0ePvoEkKUhMDB4YZDKNw+nMx+UqARorCR3VJICoJDhluc0AJHF3\nGFjwZfN6Wz0+Ly1FI0mc0+TAOT8mhiyXC53LxUUZGURHC+KxeDE0uV5z7BhsWKpD2dfJJ5+EbmWV\n1H8Gce2oJGgUCqZERvJjEyOpHTvE1ETLWfotNTV8V1nJgv79G1oYkiRxlsXSjCTIskzVf6tQldpZ\n7NtG4cM55P41l+R7k/nTEjMff0yrxzvvNK+g5DudjSQhQtzBr89fD7R//NGPyEmiUjJ4mNj3oqJh\n6HShlXARIyOInBTJ0beOilA0uxHbAFHFjI62BiWxycmNJGHp0qWYzWZmBBJ4dDPaIglXXCF8HT7+\nuOf2KRgyM8W/kyb9uvvxv4z/+XZD5feVqGJURIzpWicQw1ADmkQNB65vzHUd9t4w+lzTOJJmMokL\nZFdicLToEWRXZDMhqe3knLFjlWRnjyU7e03Dc32LzShVMv+dM4yq8mJ+kI/xIsCWp3nu/25mSr/m\nzm3FLhdX7NvHzQcPcltSElUeD2dZLNjryUGZ242p/kI7a5ZQ0fsvWgUFYtrhwQfhD38IfKfpR2Ji\nNEbjanJzLwj4utpXweMjYMXK89ArWirr11FTM58lS4KzsX79TsVkgtrabcTEzKHc7UYjSc0sjdsL\n/xhkjt1OUgglV2amOJG1nDG/NyeHZwPF+gHnREcT2YTAnBcby505OVywcyems4XpzV13wUsvwWuv\nwQMPCKHs/PlAuhYpwUGESSZQ/9yPkpaVBFd4XhnTzWaeKyjAK8soJYm1a0VLo+VH+cSRIwzR67m4\naTAJcJbFwsclJcKnQq3GddSFu8zNh2npjNPXoHq6AONoI2mPhS9NL3A6mVtPqhKM9SShQJCE9o4/\n+qGJ1ZD2ZBqWOTHoJrsoKZkY9CLfFH3/0JcDNxyg6OUisFhwjRFtkdjY4HbZmZl/5PDhr0hJgdLS\nUi655BLUnahydRSFhaHvzGNjhbbqX/9qHE9VKlv/7XsCGzYIs7p2ZG/1op34TZAEyxkWJEXXjhBJ\nksTYTWNxFoov/e65u3Ec6X4ff39s8qHyQ2GRhNGj4d//HsWhQ280PLdlk4Kxp8C0uCgWvPA07762\nBtgE5ZVMMEUwKYBz2z+HDePq/fvZbrUSoVQywWRiX/2oRqnbTVr9xVKrFRetpsjKEmKnxx8PXVGY\nMOFK5s2rBoIspNSiHACH82ugrrkw0eOZzhtv3M+xY4FX9fnA50tj3ToLtbVbiYmZ0+CR0BkHx9T6\nHtJhu53Tg7jMyLIQr956a/Pns2prea6ggNuTkpgU2VoLMq3F32GAXs8Tq1ZxvrWxktK3rxAlPvus\nEMBt3SpO4g+/rONuu0yp201CCON6fyUhtmm7IQzx6gyzmYV5eVy2dy9GpZI13ySjGQoHLizg2nqT\nI68ss6y8nLeHDm1l+HSWxYIM/Leykovj47Fmid/pu4JIxr4Qz9jx8WiTtSi04RU7nT4fx1yuhkqC\nWWdGrVCzvmA9MfoYLPqOVxL7/6U/AElJeRw7NiasdeIujSP7jmyy78hG+W4clfVF27i4wNkksiyz\nffuHwAQuvngiJpPENddc0+F97gzaqiSAEDBeeGHjhENkJHzwQesJg+5GZiaE4Ubdi07gf4YkHPrz\nIWwHbK2er9lSw5D/654IMF0/Hbp+4iKh6aMJmQfRVYjURhJvjA9blzB6NPh8RvbubUzL2bCh8cuc\nk5NDVVWe+MFhDnoXeVVCArusVp4pKGBudDRqhaJhbK7phEMgZGSEW1GZW/8IDFmWMT1l4pEZl3PP\nlHtavf7QQ8G3vGcPjBol4fU26hIq6t0WOwO9UkmiRsPhEF4JOTlQVtb8ZCbLMndkZzPUYOC5gQPD\nyoqgrIwFzz0nhtab4P77hVg0KUlUE849F/a5tLBNjEG2JAl76+r4tqKCi+PiKHp+vlIAACAASURB\nVHW7iVapGt6/1lVLcmTw2czDh0U7w+eLZEZhMtl5XmQZjm3XY5x3HLvZQW6Tj+LSuDiubpEtASKp\ncqhez/dNSIIUqaKoRsvYsRA5sX2TSEX1gVZ+kuAXLxbVFjExqWumAxITD3D0aHgVCVWEivgr4il+\nsxhTajKla2QkyUdMTGBCWlRURF1dFXAbV1xxXivfhJ6C3S7CudoiCeedJ0ipP4Triy+ECPPVV4W/\nRU/AahVTQi3Jdy+6Fv8TJMGWbaPopSLMZ5hRRzcvzyVckUDsReG5CHYG6jh1j5AEqBcvVoZPEgCy\ns8Wd9/Hj4kTvv2Dl5uYC9b1huyXkdMOTAwbglWXm1Jd0/XefpW3047sKkiSR0SeDzUWb273u4MGi\nHFpefioazfuAEC52RrToR1tjkP6+adNJti/LyvixuppVYYZJAcL5CER6UhOkpTXGSfuRUq++L3A6\naXm9+fOhQ6ypquKenByiVaoGsgehhYterziexIVBATS/YF5iSGTVo4msKwzv1znLYmFVhdDWWLOs\n2BIjkGqlhhHA9sCfepnSRB2cEJFAUW1Rh/UITeH11tG37262bAn/tjX1kVTM081U9+lHaakKs7kW\nnS6w4+uuXbvq/zea/PzW5ko9hZax5MGgUDSfbrjySmHbfPPNwufCPznVUahUok3ZokvVDFu2iAph\nbyWhe/E/QRLKvylH0kqMXDYSVcSv8yupY3uOJAyOHsyB8gNtL4iYdY6IcFBVlYzVaiUzU2gzmpME\nNxqtG5fDHJIkKCWJZ5tYQ2oUCqKUyjYrCV2JGakzeHPbm+0OetJqhatldvY4YmL+jtN5jIr6u+jO\nYoBOx4E2SMKwYY16DIfXyz05OcyNjmZ2e5qp778PZ58d+sxZj2iVCoNC0RidXI+9dXWsqarijSFD\n0EgS7x0/zugmltKhhIv++ObXXhPe/k1hMAgO88UX4f86Z1ksLD56lFy7HWuWlQJjDMOGhZe02BL5\nTYyU/PDrErqCJNhsB0lMzKGgICJsszBtXy0JVyZgz4snO7sfSUkFqFSB/967d+/GZDLhdvcniESl\nRxDMSKktKJWizdi/P/z97+EJlUPBZhMTYT/8EFx4u2GDaHOkp3fuvXoRGv8zJMEy0/KrEQQQJMGR\n2/2aBBAnvW8OhjdnKEkwdKiTbdvEGGRm5hiSkoSS2mq1UlZWBoDWaMflsLTbujZOo+lxkvDET0+w\nr3QfI+LbZyCfng6bN49j4kSwWrdR4elLuiF4kme4GKjXs6L+jviDY8e46eBBbE09ly8VD2lt41Mq\nSeLbcNKC/Dh4EDZtaqwmtAFJkkjWasl3NhfKvVRYSF+Nhuv69EGjUHBd3+ZW5aEqCf67zIkTRYpk\nSxiNjeXnlnDkO6jJrCH+ssZxhBlmMwpgTUE5g7LtbOoXwfiO2RlQ4HQSo1JhaKKe8084dA1JOEBS\nUjYejyLsQDY/1OoEdu8+hRkz1jYbf2yKXbt2MWrUKMrKpBOCJCQltX9dSYJ77hGPzuK//4UzzhCd\ntT//OfAyfjFwr6Ni9+KkJwmeag/VP1Yz6MXOnwg6A01cz2gSQJz0yu3lVNorwxJknXqqhm3bxpCd\nvZsNG8YwebL4Quc28YNVa60onTFolMFFboEQp1b3KEmYkjwFtULN2ry1bZKEKkcV72a9yy3jbkGr\n0jJiBLz9dgp33hlDbe1Wyt1nd0m7YYBeT6nbzcLcXB4/coQr4+OZWS9idDiFe+DVV8Pp0xrXGRUR\nwZD2EJQPPhC3Te1QhqXodM0qCRVuN+8fP86ClBQ0Qc6soSoJbV1AjEZhyR3oTrvwxUIKFxUSPTe6\ngcyb1WrGm0zs3lTGIGBtYQT/CBa70QYKnE6SWxiRdG0l4Rf69xcjOzfeKMShwZCQICKu/XylsjKF\nY8fSGDbsL6jVgTNXdu/ezdSpUzl0SBgq/VooLBQVr27KKwsbM2eKKam//AXmzBGR9k0hy4Ik3H77\nr7N/vyWc9Bys4rsKZI9MzLldZETQQfRku8F/0msrx8CPCRN0wCB27y5g69bG8aamJEGhrkHlbruM\n3RJxanWPaRIADGoDE/tNZO2RtW0u+9cf/sqd397JUz8/BYhKwtGjEnr9OGprt3ZpuwHg8SNHWNi/\nPx8MH84NiYnckJjIiLxE5G8SuW9EYsNzNyQmMjHANENQyLIgCZdc0jyAoQ2ktKgkLCkuxivL3Bgk\nKMTr82Jz20JWEtTq4BdI/4WlZT4JQNW6KvBB7abm6ZxjIiKQd9vwKSXK9AbqJzvbjfwmRkp+dDVJ\nSEuL4NprxZ1rRUXgx9Gj4u5348bGdXfsEElO6ek/B6wkuFwu9u/fz6hRo0hJ4VevJLS31dBdeOop\nQUivvVYkfv7yi0jGfP99IVasqOjVI/QETvpKQvk35RhHGdH17wI7w05AHavGZ/PhtXlRGrp3YNh/\n0suuyGZc4rg2l8/IkACJ1avjcDia6xF0Oh0ulwtJUYXS2X6iFadWsztYjbmbMKP/DF7f9npIXcLB\n8oO8vu11hscO58mfnuSS9EtITxeVB5vtVJzOd6n0eDqcANkUI4xGxhiN3JaUxB9aXIA3bhR+GcMj\nCuD5z+DOO0UZpz1Yv15Y8rVzJC5Fp2N5fRvE4/PxSlERVyYkNFgwt4TfpCpYJaGoSJy0g5V3/YWR\nujpoOr3pqfVg3SG2Xb2+ulnQWrxGg/IXD0e1Rmado6Cj3Z8Cp5PTW4yMnjP4HIpqi4jRd/4Gwmb7\nhcjIibz7bujlvF6RuLl8uUjZBNi2rQ8JCUeIiysKqEk4cOAAHo+H0aNHc/w4fPtt89erqkS2SqD3\nqq4WF0ulUpTnO4sTiSQYjcKLYdq01rqDQYOEH0h9cGUvuhHtJgmSJE0D7gVOBfoCF8iy/FWT198B\nrm2x2ipZls/pzI4GguyVqVhRQd8/dE0EdGegjhUXG3eZG2VK95IEs85MrCE27DHI9HSQJB87d05G\nqxUZCyBIQmpqKiUlJciKSiRH8LCqYIjTaCht4rzXE5iROoO//fQ39pbuZWS8CIvfXrydEXEj0KrE\n3eSCNQtINCWSeX0mk/45ieu/up7vr1iPQqGkuDiV+PhjgLdLKgmRKhVZQeTomZmih69c8TXcfbeo\nnTZt6L/9Njz9dOPPEREi2KF/f2EnGBkJK1eK56ZNa/0GIZCi1XLM5WL0li04fT4KnE7+FKLZXOuq\nj4kOUkkoLAzdq/ZXElpyxprMGvCBfpCe6vXN807i1Wo0B33stkW0HNpoF/ztBqeziOzsu5FlUd26\nLhH27m29YYtlFklJN4e1bVn2YbcfpE+flqe11lAqxZ94+XJ48knx3KZNJtLTVwHCVbQl/JMNI0eO\n5MABKC4WaZoajfh3wIDAJKEl1q1r9yHSCoWFwin2RMGUKcLBtbhY/KxUChFwEEuSXnQDOnKGNAJZ\nwD+BL4MssxK4jkart+A2Y51AzaYa3GXuX73VAGIEEgRJ0KV0f1VjUPSgsNIgQVSoo6PLKC8fwJQp\njQYoubm5pKWl4fF4qPCVI9uHtns/elqTADA5eXKDLmFk/EiW7l/KRZ9exKj4UXxw0QdYXVa+2P8F\n713wHlG6KJbMW8K0d6bxxs4X6T/0BrKP6ImP9xGBtUs0CcHg75vecgsi2ABg2bLmJOG55wQR8Nvv\nVleLpvS334qB9ZoacDrhiSfardCaGxPD3f364arPlbhFp+MUU3Bhaq2zniS0UUkIhmAkofqnatSx\navpc14f8p/ORvTKSUpwa4hVqLIdhjd7AQ8EtMkKi1uOhyuMhWaulrOxzysq+wGKZFXR5t7ucQ4du\nw2I5E4Oh7Vk9p7MAn8+OwTAsrP2ZO1eUxAsKBM/bvl3BTTdtAQjYbti9ezcpKSmYzWaSk8Vxc/So\nyH7JyhIE4fXXxd1zU0hSY9z0nDnwyitdQxLO76AupLswYIB49OLXQbtJgizLq4BVAFLwGTSnLMvd\nHv9R/k056lh1u41XugMNlYTSHvRKCLOSADBgQB3l5TB+vBsQ+5qbm8u0adMoKyujtLQcn719kw0g\nvBKsXi8Or7dV7G93waA2MKnfJNbmreWi4Rdxw9c3cNaAszhuPc64N8eRaEoko08GV42+CoCpKVO5\ndfyt3LP6HrjsHv6thilAFNXN2g0HDoiRq65CTU0TE6WfmpCEBx4Q/9+7V+Qif/21cEAKBv9tZTsR\np9E0G1ltc3+dYh9DaRIyMoKvH0yTUPVTFVHToog6LQrvQ17q9tQ12KTH5/mQ3GCcqSEEfwmJ3Cbj\njzUVm4iIyGD06OVBl/d6HWzaNID8/CcZNuydNrdvs/0CEDZJmD1b3PEuXy5in10uiVNOEYQ+ULvB\nP9kAYuoIBMFITRVjflqtcNUMdQjcdpuw6G6LyIWCyyXGXE+UdkMvTgx0lyZhhiRJxxEuPT8AD8my\n3HYiURBk35VN0atFrZ6XXTIJ1yQ03JX8mmjabugJDI4ezHc534W9/IQJGrZsgbi4bGA4siyTm5vL\n/PnzycnJwVNchsfW/rN0U9fF5B40b5+ROoNXt7zKdf+5Dq1Sy8e//xiTxsRf//tXXtz4Im+f/zYK\nqfHO+4XZLzB74Gzefs/JL/u+goz3iaK6od0gy8K8Zfv2rvWg79OnniQsrycJmzYJy8I+fUR2eFQU\n/O53oTfSAYLQETS0GwJUEmRZXIBCXUACVRJ8Th81G2sY8NQATONNSCqJ6vXVDSRB3uRCAlIv6/jv\n+O6xY1hUKsaaTOw5sBmL5ayQyyuVOlJS7ic7+2769/8ren3o21Sb7QCSpEWnCyPfG3F3f9ppgiTU\n1YlKXnp6OXZ74HbD7t27GyyY/STBP+GQmSmimds6BK69FhYsgDffhEcfDWs3W+HoUfFvL0noRVN0\nx3TDSmA+cAZwHzAdWBGi6hASPqeP4n8WE3NuDAOfHdjsMfjlwaT9LfwAmO6E0qBEYVD06IRDSV1J\nw91fW/j97/sAddTWit5oeXk5VquVtLQ0LBYLXncZHoeO9nYOwrVm7mrMSJ1Bub2c1YdX8+4F7xJr\niEWr0vLM757BusDKGWnNVVxqpZp5Q+dx0bCLOZoprOLMVDVUEjZsgG3bxIndP8bXFY/i4vr+aXW1\nXxwisrRlWZCECy5o7P/8ymhoNwSoJFRXiwteqLvUpsLFhm1urUV2ykRNi0JpUBIxNqKZLuHYcgdH\n+8KgM3x0BFVuN28VF3NLYiIaXw12+wEiI9u2Ye7b9w+o1bHk5z/V5rI22y8YDEOQpPDZ47nnwpo1\nojI1fjwYDGIkRKVq3kyvrKyksLCwoZIQESHaB/4Jh3CzCSIjhZDvjTdERaAj6KiRUi/+t9HllQRZ\nlj9t8uNeSZJ2AznADOC/wda78847iWqhTr7iiiuYFTkLb42X1EdSiRjZtUmOXY1fYwwyuyKbsX3b\nVhrNnKlkzpz5bN1aDdzZMP6YlpZGdHQ0HreIT66qCsvQrwH+iOEe1yX0m0yEJoLrT7meWQOb959D\neT2kp4P12HC8MkRLVejr+/yLFolZ7Nmzu2mHa2pEYzUmRrQcJk4UM13PPddNb9h+hKok+I2U2qtJ\nqPqpCmWEkogM8d2NmhpF6ZeiE+k85iRmXQkrz4ERmo5d2d4oLsbl83F7UhK1tesAiIxsO/hMqTSQ\nknIvhw8/QP/+D6HT9Q+6rCAJ4bUa/Jg7F+69V8TEP/AAaDQJKJVRrYjG7t27ARjdxFjLHxldWCj+\nDZXI2BS33SayE774QsQ5txe9JOHEx8cff8zHLTK6q6urgyzdNej2EUhZlnMlSSoDBhGCJLzwwguM\nDSCr3X/dfgzDDBhHdMDdo7Cw8ZvT0UZdO6COVeMq7RnPgPaSBIAzzpjMwoULcblczUiCxWLB5xR9\n13aThB7Ob/BDr9Zz8I8HG1z1wsWwYeCrSqHWpyVWWYUkCYe7L78URKHb3NtqaoRH9owZIolq8GBR\nYjgrdGm8J1HrrEWlUKFVtq5shOPpr9eLQklTklD9UzWRkyNRqMQHGzU1isIXCnEWOcm5Lwe3rOCj\nKyX+3AGS6fT5WFRYyPw+feij1ZJXvAmVyoxeH15wQGLizeTn/52srJloNH2CLme17iA5uX02gsOG\niUyN3FxRCTAaR2AwtA6a27VrF2q1miFN3IJSUkS7wZ/5Ea4XQHq6GIN86imhLWiJkSNDH26FhaIi\n0VFtSC+6H1dccQVXtGCA27dv59RTT+229+x2kiBJUj8gBihu77o+l4/yZeUk/TGp/XG+Lpcwn9m4\nUSSRXNv2+FJn0ZOVhGh9NGadmdzK3LYXrseMGTOw2+1s2bKF3NxcoqKisFgs9SRB2DO3d5rRoFRi\nUCgo6+FKAkBfU/tHX41GSO2np9pnJFpZDsDixaLM262HSE2NuHKcf77wrX35ZWHD2EN6g3BQ6xKW\nzIG+a/67zCA+TIAgCAZDo3BR9spUr68m+Z7GVMnIqUJknPdoHiUflfC2YRhSRD4lHSCZHx8/TrHL\nxV31zKW2dhMm03gkKTymp1QaSU//N8ePfxhyOaNxBAkJV7dr3yRJtBxefllc5GNjbycx8TYAvvji\nC15//XViYmLYt28f6enpqJsIaJOTRftrwwYhXuwTnL+0wn33wWWXtU5E9flEwuNzzwmBYyAUFPRW\nEXrRGh3xSTAiqgL+M8kASZLGABX1j4eBL4Bj9cs9DRwEvm29tdCo/KEST5WHuIvb7wTIXXcJFZrF\nIuaIeoIkxKlxFnbLtGdAxOhjqHSEMUBdj4yMDCIjI/nxxx/Jz88nLU3oOcwWM3gEOwhnHrslfo0x\nyM4gPR1qfFFE+iqw2YTY6/rrBVHoNtTUiNu0QYPEDuzbB5de2o1v2H7UOoNbMhcViQpTW5ymaX5D\n3Z46vNVezNMa+/DaPlp0A3QUv1VMxNQolq5PIF15jJIwj5+lpaUNLpKvFhUxLyaG4UYjsixTU7OZ\nxMQbw9qOHxbLmVgsZ7ZrnXBx552ihSUqc4oG8vLVV1+xY8cOxowZgyRJXHXVVc3W87cbMjPDbzX4\nMXt2YKIvy0LYePfdYrL2b39r7el1Ihkp9eLEQUcqCeMQbQO5/uFvqv4LuBUYjRAumoGjCHKwUJbl\nts8CXq84mdbH5ZV+Xop+sB7j6ACtBp9PmOJv3Sp+liRxpzZ9utjG4sViuPi774SXZw9AHatucJbr\nCZh1Zqoc4d/6q1QqTjvtNNauXQvQQBL0Jj3+uOiO+CL1dMhTZ5GeDpVeC3q5kjPPFKK8P/6xm9/U\nTxIALr5YRCme2T0XJz+q11ez78p9ZPyQgX5g23bO/kpCIIQ7WteUJBS9UoQyUolpQvNtRp0WhbPA\nieLuIbBeoq8+PGvvWo+Hi/buRSNJqCQJjULBv1LExIHDcQS3uwSTqW09Qk8hLS3wcWWz2Rg3bhyr\nVq0KuF5KinBR3LZNFJu6ApIk2hCxsaKQtXJl67bCzp2tUsh70YsO+ST8SOipiA66ryMCyZcuhfx8\nfF4oW1pG4k2JgVsNjz0mtAb+kq3HAzt2CB9PWYbrrhNJLCUl8OKL4rmODViEjZ5sN4AgCe2pJIBo\nOTzyyCPEx8dzUX0gvCZCA1iRFD4qK9vflD/ZKglXXgkfH48mQbGfPn2EM15adw/JNCUJDz4ozOe7\nudWw4bMNPJD/AJOnT+b2FbczYnToQKym4U6OIw5QgC5ZGIO1Nf7oh8EgSELlfyspXlLM4NcGo9Q3\nF+ulLkwl4aoE1tkE+e8fpWGLK0DgQwv4/RDWZmQwuYXIubZ2MxCeaPHXhs1mwxDCf9o/BunxtL+S\n0BbuvluQkK+/bv1a//6iotaLXjTFiZPdcOAA8uLXsMsJ8J+t1Nr746nwEHdJgFbDypWCJDz2WOvm\nW2WlIAtTpwpSMGaMoOWFhY3fvm6CJk6Du9yN7JORFN3v3dDeSgLA9OnTsdls5OXlNVQS1AY1IGMw\nuKiqar9bZJxaTY7d3u71fi2ccgq8vTaGKE8tS5f2wBu6XOBwNJIEjUZEBXYzFi1bRJ6Ux6GiQ7wx\n5g2mTJnC8uXLMQfxtG0aE73n93uwZllJuCqBlAdSKCw0MjHEZOEvv/zCsGHDMBrBUe3lwB8OEHV6\nFIk3thYx6Afq0Q/Uk/eiEDv2j1KzvKjtSoKfJKTpWh+jNTWb0Gr7o9F0/+faWdhsNiyW4Omt/tOU\nwQDtSRMPF5dcIh696EU4OHFIwjPPUBB3K4dLLoLf24D96AfpxeiULAvVDYiL/VVXiRmjBQtab8di\naZ50MmaM+Hfnzm4nCepYNXhFfLXa0n12v35YdBYKatoXGTd27FgiIiIaPBIAFAZRPdAbnFRWdowk\nbKwJz6/hREEFFqIUNhweBzpV69/52LFjbN68OeQ2JEmiT58+pKWlERMTE1xc6/9s2pP82EkcOXKE\n1XmrWTBuAVefdzWfLvyU53c/zyWXXMKKFSuaCeX8KKgpYGjMUGRZxn7AjnmGmco1lRx//zj5mtOY\nO9EDtP6sVqxYwdy5c1m7di1G43SGb8rDWehk9IrRIclybq4Q5iXUt6t8sowiRLUvz+FAp1CQEKAC\nU1u7OSx/hBMBbVUSkpLE/c348dAF0SK96EWncOIcgllZVE14nihvEWmJK+GVV9AP1IsT73XXiTaC\nH2lp8N574c2rpaSIUbOdO0Nb33YBmloz9wRJ6Eglwa9LWLVqVQNJkAzixKzTOaiqigq1ekAE0yRU\nuN08mpdHjdcbcv3xJhO39sCIqh9eWaYMM3qFh4KqHAbHNi/DOxwOJk+eTF5eXtjbNBgMaAOYIsXF\nxbHt88+JgObRiN2MxYsXY5AMXP67yxny4BDmrZ2HeZeZu9bexe23385rr73WjNTY3Da2F2/nujHX\n4S5347V6SboliZhzY8h/5zgVt6hwv5HNzsPOBo2QQq2g3939ePjhhwH49ttvSfaNY/SBAlKfSMMw\nJHSkY26u+CrHq9X4EMdLbIgWTK7dTqpO14qM+Xxuamu3kZZ2QQc/rZ5FWyRBo4GBA2HmzB7cqV70\nIghOGJIgn3EmNVlq+k3VYl7xbxj9Cpi1YuD3ww9FSo4/vWTGjAZxY5vwtxyysrpt3/1oZs3ceiS6\ny9ERkgAwc+ZMvvvuO1JTUwFwSk5QglZr79B0Q6xaTaXHg9vnQ92EuP3p0CGWlZcz2hjc46LE7eaT\nkhJu7NsXVbeZFDRHtcdDJeL4Kara14okLFq0iMLCQjZv3kz//sFNdrxeL0ePHiU3N5f8/Hw8Hk+z\n10tLS3n22WfZvX07k6HHKgl1dXW89eZbzJXnEjsqFkkhMfTNoVQNreKpS5/i3jfuRaVSMXSoCPSa\nOHEi9ng7Hp+HqSlTcRwWZX1dmg6FVoE8W4yanvqXPnjXHab8GzE66jjs4IcjP7B161YGDRrEmjVr\nuLLq/2FTqUm+u+2qXV6eiPqNb2LIFZIkOBwBWw022y/4fPYTSrQYCm2RBBDp4D3IKXvRi6A4YUiC\n4/Lb8fzgIfL3I+Brr/A0/f3v4Z13hJn+3/4G0a3DUcJCRoawPutmNE2C7An4SYIsy+3ykbj11lsZ\nP358w4nK6rKCHjTquo5NN9SXrsvdbvrU302vLC/nw5IS3h02jGtDDHqvq6pielYW+202RgWZQbz9\n0CFGGY3cGGpIvx0od7upRpyBS2oONnvt+PHjPPHEEw2fUVvo27dvUCMTu93Oc889x/59+7qcJDid\nTsaOHcv06dN56qmnmrmVvv/++9TU1nABF2AYKv7G+oF6Eq5KYMr3U/jrg3/l2eefRZZl3G43Q4YM\n4eo3riZSG8mIuBGUrxUkQDegUbQIMOIqMyOebDTu2n/9fm764CamTZvG/Pnzuemmm7hHnU1WzCDO\n1QQnfC6XC6u1jtxcC/PnNx4/JW43w4OuJUjCaQGunA6H8AoJ10Tp10Y4JCG+/antvehFt+CEIQl1\nx/RoANN5g8Qo47ffwoUXwltviXnyjhIEEJWEl14Cq7Vbh+FV0eLj7KkkSLPOjMvrwu6xY1CHPuk0\nRUREBDOb1DJrnDVIegmFVNNhnwQQd4J9tFpqPR5uOniQWRYL8+sFetXV1SxZsgR3i7aE0+eDvDwe\n/e47xgW4iHplmdfz8his11PRHleZAEhKSuKaa66hwuNpIAnFZQfJz88npX6UbuHChahUqoYSemeg\n1+tJS0tj38F6ItKFJCEnJ4d9+/bxyy+/sGzZMl588UXG1OtvFi1axJxT59BnSx/0gxtHH/sv6M/x\n949zU9JNPGZ7DIAlS5Zw0003sS57HZP7TUapUGLPtaMyq1Cbxd81mCXzjiE7OOA6wNPznmbMWWPw\n+XwccG7l54jTQu77woULefvtd7Bas0hL60u8nySEGIOUZZk8h4NrAgg+HY58JEmNRnNyXFnDIQm9\n6MWJghOHJOyuwzzMLHr5Z58tDMjXrIHDh4X+oDMYM0aIH3fvDt/jtANQqBSoLKoerSQAVDmq2kUS\nWqLWVYvSqEShKGlIgmsP/CRhv82GRaXiifx8yt1u3hgypKHC8fbbb3PvvfcSHYDsKT0elksSawPE\nL3pkGY/Hw0FJ4tlOqLg8Hg/V1dVMmTKFCrO5gSR8+up/uWNZfyZOnMjcuXNZsmQJzz//fMD97AjS\n09PZ79c2dDFJAFi/fj1PPPEEl7YwZnp0/qNoijSoTI2fmWGogfjL48l/Kp++1/dFoVGQkZGBz+dj\n/bb13HfJfYBoI+jSGsv6RUXC/yAqSmQN7N+/n7KyMl751yuMNYxl0LZBpN6bSj9jPzb6dnHIF5qI\n//jjj5SWlgBXkZKymiiVCrUkhTRUqvB4qPV6A7YbnM4CtNrksJ0Wf23YbDb0+rZ9K3rRixMBJw5J\n2FNH5KT6k+js2cLb4N57YcSIzg8Lp6cLmfDOnd1KEkC0HHqKJFj0oq9e5agi0dTxUnyNswaNUYMk\n5XP0qEgvDCB+D4oEjQYlcNm+fQ3PvTBwIKlNToTffPMNs2fPZuXKla3W0kG5gAAAIABJREFUn79/\nPwdsNjYFKNu/UFDAXTk5GJRKSk87rf323PWora0lJiaGlStXYr74YhzocPkk9m44xqxZs9BoNDz6\n6KMMGjSIW265pUPvEQjDhw/ns8xM0TLrwgtDTk4OOp2OCRMm8NVXX5GVlUVtbX1Ak8mE5ikN7iGt\nj8P+D/Zny6gtHHvvGIk3JDJixAiUSiXWI1amJIvvmSPX0dBqADFQlJQEdruN8ePH43Q6UalUJCUl\n8cwNz1D2ehm2QzYy7Bls1e+AEJYHTqeT7du3M3nypWRmfs7nnz/O+PGPEK9Wh6wk5NZPN6UGJAn5\naLXdO7nUVXC73Xg8nt5KQi9OGpwwJMF+0E7kn+tJwvTpoNOJi/qiRZ03QdLpRAujB5wXezLkqWkl\noTOoddaiNWnx1BxGlsWdY72mMSyYVCq2nnoqx+vvBE1KJZOb3DVXV1ezbt06Fi1aFHD9cSYTn5aU\n4PL50LQQL66vrkYB1Hq9VHo8RLeHvTTBAVlGPWYM93/0EcZTTkGnUHIoW0d1qZ27776bWbNmUVpa\nikKhQKPRcOPXN9I/qj8Pnv5gh97Pj/T0dI5UVFAXFYWxC828srOzGThwIIr6z+uUU05p9vqWg1sa\nSXcTGEcYib0olsIXCkm8IRG9Xk9C/wSKjxczsZ8YIbTn2om7sNGfxO+2uH37dpxOJxs2bGDSpElI\nkoSrxEXmq5nsu3QfY+WxfFP3DUaKgMDTKjt27MDlcjFu3L1s3z6Sf/zjYWbNmkZ8VFTISkKe3yMh\nANFyOArQ6weE/sBOENjqgy16SUIvThacMPU52Ss3ntT0+kaicM01XfMGPTjh0J5KgizLOAocOI6I\nh7PYic/lC2vdriIJNa4aDCYDTqfonR850v5tZJhMzI6OZnZ0NFOioprd8X/77bd4PB7ODTKCOs5k\nwinL7G0aH4j4bNbX1HBm/SSL30ynI1hVUYF3/Hhc27dztdnM60OGsHmzCrVWGEyBGFeMiYlhQ8EG\n3tr+FssOLOvw+/kxfPhwZFnmQIA74M4gJyeHgQMHBnxN9snYD9qDjiDGzInBtt+G1y5GU7X9tBjK\nDURoIpC9Ms4jzlbthn79YOPGjRiNRsaPH9/w99XEa4iZF4M1y8oZZwl/EpvtB2QZDhwQHF+WG997\n48aN6HQ6nM7RDBmygJkzZ3LDDTcQS+gk0VyHA5NSSXSAlpO/3XAyoJck9OJkwwlDEhQ6BYYRTb44\njzwC//xn+KOObSEjQ2gSfOFdgDuK9pKEopeK2JiykY2p4pGZmMk67Tp+ivyJvZftxesI7jHQZSTB\nWYMx0ojVuh8QMbVdia+//prRo0c3iANbIiMiAgWwtb5c7keuw8Exl4ur6sVqeZ0gCXvq6hh55pl4\nnE5mFRRwbZ8+bNvkZcAokFSNhEaWZe7//n4A9pbuxSd37ngZPlzo9fd3sAISDKFIgrPIic/ua5hs\naAnDcAPIonoHUGOpwXXUhdfrxVnoRPbIDe0Gnw9ycgRJyMzMZPz48ahaXKj73iBGJIffOJyUlDHI\n8vf8/LMwPb3jDqiPCgEESTj11FPJz9cwYICSl19+mfz8fKr+85+QlYRchyOgR4Ise3E6i9DpAh9b\nJxp6SUIvTjacMO0GQ7qhIXMegEmTxKOrkJEhTOX37Oker9N6tFeTUPafMqJOj6L/Q2Ie3+fw4S53\n4ypyceRvR9h74V5GLB2BUtda1KdX6VEr1FTaOzCS0AS1zlpMUSaOVO0hNrZ1JSE3N7dN90GtVsuc\nOXNaGQp5PB5WrFgRss9vUCoZYTSytbaWPzR5fn11NQDzYmIwKhQNfemOYE9dHaePHElpSgorVqxg\nwoQJ7N9j4/c3QVFNEWkWYSy1/NByfs7/mTsn3ckLG18gryqPAZaOl7IjIyNJMhjY1/R2upPwer3k\n5eUFJQn+i79+SGANhGG4uEDV7aujbmAd5ZHl4BDEI+6oaDPo08S633wDx47BvHky776bybUB0lSj\n50QzZs0YzDPMZLx/Fvn5H3PmmZkMHQqRkbEsXjy4wRho48aNXHzxxXzzDcyZI9ox8+fP59M33iAp\nROBVXhCPBKezGPD2VhJ60YtuwglDEoyjghvudAmmTRNjlB991L0kIVYd9gikp8ZD9c/VDFo0iOjf\ntVbTR06KZPe83ey5YA8j/zOyFVGQJKnDhkpNUeOsIdIcidPpJD3dR35+I1mrqqpi0qRJlJSUtLmd\nyy67jI8//rjZ3d7GjRupqKgI2mrwY5zJ1KqS8HN1NekGA9FqNWl6fYcrCS6fjwN2O7clJeGbM4eV\nK1cyefJkfD6YNBEKawpJs6Th9Xl54PsHmJk6k7sm38ULG19gT8meTpEEgHSjkf1N7pKrHFXkV+cz\nOqFjx2FBQQFutzsoSbAdsCGppGYtg6ZQW9Ro+miw7bexI2MH1E+WZmVlMc0qDMu0/QXZe+45oRvu\n16+Q4uJiJgUg7pIkYTlDVPymTj2Hr756Drd7Cnv2gEKh4siRgxQWpqFUFnPkyBEmTpzE4sWNupdH\nHnmE9z/6iIKPPw5qM5jrcDA7QFXR6RRlr16S0ItedA9OmHZDxKju8y8AQKuFK66A998XkdTdBHWs\nGm+NNyxdQeWaSmSPTPScwON2ljMtjPpmFNXrqsldkBt4Gb2l88JFV21D4ExCgrNZu+GRRx6hrq6O\n7Oxsamtrgz4++eQTPvnkE55++ulm2/7666+Ji4tjwoTQbnjjTCZ21dXhaPK3WV9d3WCek6rTdViT\ncNBmwyPLjDQaOeecc8jOzmbRokUMH55Aal9BEgDe3/U+e0v38vez/k6SKQmzzszekr0des+mGK7V\nss/WKPl/7MfHmPLPKdjdHauM+Mcfg5KEgzZ0A3XNK3MtoB+u5/CWw/xjwz9ISUwhKSmJrKwsHLkO\nNEkalDolW7fCunUiOXDjxo2AcGcMhdtvn8k//pHNjh17yMrKwmIxo1Q+z5tvNm5jwIBJOByNyZv9\n+/dn5vz5OD76iGOlpa226fdICDzZILJLetsNvehF9+DEqSSM7OZKAsC118LixfD992LMshugiRO2\nsu5yN9q+rb38m6JiZQX6ofqG0m4gWM6wYJllwXYg8FxZV1US/L4AsbF1bN0q9mf37t288sorPPnk\nk0EvSH5ceuml7N69mwULFjB69GjOOeccQIw+zp07t0GFHwzjTCY8ssyuujomREZS6Xaz12bjvnod\nQ5pOx5qOOD0hWg0AI4xGNGecgUajYdOmTfzpT2cTpV7F9uoCcitzuWPVHVwx8gomJAlCMyJuBHtK\n93ToPZsiXalksdWKy+VCo9Hw9cGvqXPXsfrwas4bel7Dck//82nWrlpLn4jmplGSJHHzzTc3EK2c\nnBwUCkVQy2j7ATuGoQYOVx7mjlV34Pa1rmyNdY1l8J7BWM+zsvicxbz+/etkZWVhj7Y3HI/PPScy\nBM4/H+67byOpqan0acPQSq+XuOeexmPlT3+6nccf/zuvv76Q+fM30q9fP5xOkTndNJ77urvv5vsP\nP2T8uHHE1h+LDz74IBdffDHHXS4cPl+QyYZ8lEoTKtXJ4WHcSxJ6cbLhhKkk+C+u3Ypx44Rnwrvv\ndttbNA15CgVZlqlYWUHMnJg2t6mKUuGp9gR8zawzU+Xs/AhkXIzoRUdFVZOfDz6fzO23386gQYO4\n4447wtrOo48+yrnnnsvll1/OzJkzmT59Ovv27WPevHltrjvaaEQlSQ0th8z65MSpTSoJeQ4Hcgd6\n+3vq6kjUaIhWq4mIiOD0008HYNas01BKUFD1C5d9fhnR+mhenftqw3oj40eyp6TzJGG414tXljl0\n6BAHyw+SXZGNSqFi6S+NOdV1dXU8eMeDrFm7hkOHDjV7fPXVVzzzzDMNy+bk5JCSkoImSM6B7aAN\nwxAD7+18jx9yf8CgNrR6xIyKIbUqlazrszh3yLlkZGSISkK9kVJ+Pnz2mRAeKpWiChCo1dAWbrvt\nNtRqidLSxSxfLraRW18UazpmOyQxER55hCmzZzNlyhRKS0tZtkxMl4SKiBaTDSdHFQF6SUIvTj6c\nMJWEHoEkiWrCww9DVZVIh+xiNAt5CoG6vXU4C51BWw1NoYpS4a0O3CIx68xU2Cvav6P18Pq81Lnr\nGkhCREQ5NttA3nrrS3788Ue+/fbboBejllAoFHzwwQc89NBDVFSIfbrllluYM2dOm+vqlEpGGY18\nVVZGokbDZ6WlJKjVDKi/MKTpdNh8Pkrd7oZAoHCxp66OkU1Cpi6//HIOHz7MxImT2LMHvtr/HsVO\nJRv+34aGiREQJOGdrHdwe92olR2fTkh3OgHYv38/+TX56FQ6bjr1Jj7Y9QEenweVQsWCfyzAW+fF\n+/+8fPzYxyRHNfbYH3nkEV566SW8Xi9KpZKcnBwGDRoU8L18Th+OPAf6oXpWZa9i7pC5fHLxJ62W\nq4yrZOfrO3EcdmAcZiQjI4Pi4mKOeo4y4ncjeOklMJlEAKvL5WLbtm2tXB3DQUxMDDfccD2vvfYy\nBw44MBof44UXICZGbN+POLUaJkzghhtu4HfR0VRUVFBQIFoJfpIQqN3gcOSj050cegToJQm9OPlw\nwlQSegxXXw0uF3z6abdsXpOoQWFQUP1zdcjlKlZWoDAoiDq97TJpyEqC1typ6QarywpAn1hRRtZo\njgHw5JMfMG/ePGbNmtWu7UVGRvLSSy/xwQcf8MEHH/Dqq6+GbUF7htnMt5WVXLh3Lx+VlDA7OrpB\nBOm/QHREl9CSJFx//fXk5OSg14vf2ajy8vys5zk1sbnj48j4kbi8LrIrstv9nk0Ra7USazSyb98+\nlh9azszUmVw56krK7eX8nP8zLpeLJYuXYBhrQIqWWH14dbP1zzrrLCorK9m+fTsgKglpA9J4bsNz\nHLMea7as7ZANfOBJ8bC5aDNnDzw74D75Jxxs+8VFKyMjA4D9pfvRD9Dz1Vdw5ZUi6mTnzp04nc4O\nVRIA7rrrLmS5Glm24fFMwmiEW29tvkycRoOZSkodQiCbnJzcQBLyHA6iVSoig3oknFyVBJVKhbqL\nR2J70Yvuwm+PJCQmwqxZ8K9/dcvmlXolCVcmUPxWMT5PcPFixcoKzDPNAUcbW20zShmUJHRWuFjj\nFGV9S4SFiIgIVCqR5pOfD/fff3+Ht9sR/GPgQMqmTm14vDNsWMNr/lJzeycc6rxeDjsczUiCH2p1\nLAA3jrmUW8ff2ur1EXEiQnpvaSfEi2432O2k9+vHrj27WHdkHWf0PYNYVyxJpiSW7l/KW+++ha3M\nxk1/vonxSeNbkYSJEydiMplYvXo1siyTk5PDHs8e7ll9Dw/98FCzZY++ehSVRUVmVCYyMrMHBdbe\naBI0qCwqbPsESRgwYAARhgiyyUbRT8fhw41DQJmZmWi12laujuEiNTWVSy+9FLVaTWbmWH78ER57\nrPkyRqWSBTyDoehOQJCEwsJCfD5f0IhoOLmMlKA33KkXJx9+eyQBxJTDhg1QVtYtm0+8ORFnoZOK\nlYHbAJ5aMfoYjh4B6tsNtV5kX+t+fGeFi7UuoQGI1EYSHR2Ny3UUhcJJcvJpTOlsZkY7IUkSMWp1\nw0PRZJTSrFYTpVS22ythf10dMgQkCSqVaPVcOnx2wEyIOGMc8cb4dusScipyGn+o11ik9+/Pzxt/\nxvOVh4XzFjJo4CCiVkfxxbYveOzJx5CGSdx/4f3MGjCL1Tmrm5k4qdVqZsyYwerVqykrK6O2tpZM\naybTUqbx3s732JVbRGoqZK2yc/TNo6QsSGFl8UpGJ4wOmukhSRKG4Qbq9gtRp0KhYGTqSJaxjH98\n+QVeLwwZIpbduHEjY8eODbvtFAgvvPACy5YtC1lV6iOVYrT9jM/nITk5GZfLRWlpKbl2e0DRotdr\nx+0uPWkmG6CXJPTi5MNvkySMEHeI+NP5uhimU01EnBrB0TcCRypWfFuB7A4++tgSqigVyOCtba1L\n8JOEjgj6oLGSEKmNxGKxsH79z/h8eYwcOafDYUrdhY54JfgnG9IDkASFQo1KZcHtbj1250d7xYvb\ni7cz6OVB7D6+WzxRL8AcOWQIxwuOozqo4v777uf555/nyPojFD1aRMmREqZfNZ2EiARmDZxFub2c\nHcU7AKEZmfPhHPaa9vLT+p9YtEzkX1w27TKWX7mcCE0Ej3/xBUeOwPKFFWgTtSTelsiq7FVBWw1+\nGIYbGtoNAI+d9xhDpaG88MZy8R6XZRAREcHHH3/M5E4GoyUkJLSpTYmkBrVs5Y87P+fleh3HlWvX\nsqW2Nsj4oxhd7a0k9KIX3YffJknwy6q7iSSAqCZUrKjAcaT1Ra3o5SIip0SiHxBer14VJXqxgVoO\nZp0ZryzEhx1BrbM+OVBjIjo6mrVr16LRHEOnG9bGmj2Pjngl7KmrY4BOhzFADDWIloPbHbyiNDKu\n/SQBYOfx+jCxepJw7cW/J+qGKO749x08/PDD3HHHHfzyyy9oxmjgFLj/ctHamdRvEhGaCL7L+Q6A\nd7LeYVX2KvqO7ovX7eWJl58A4NWrX8WkNfGniX9i2SZBKPZt8ZD6WCp7qvdwvO44Zw8KTRKM6UZs\n+20NFao0ZxrPDHqGhx56H63WxR//+Hsef/xx3nrrLR544IGwP4OOQJZlIhE6njjXZuT4eAAqjx7l\ntKgoLo6La7XOyWakBL0koRcnH36bJCE6WiiyupEkxF8ej9Kk5OhbzasJNVtrqF5XTfL/b++846Oq\n0v//PlMyaZPeE5IAoROqUgTRFUQQBF0XEXTdtbvWRX6ylrWXtev6ta66Kq7gWrGwVlSKUhSUHjqp\npIf0mSQz9/fHmfRJMiEzSSac9+s1rzD3nrlznmSY87nPecpS17/Y9MFygWtLJMCJ929o6UkASE0N\nIiur9300+jvSIDtDy6DFlnQoEqJGcqD4AJY61963XlDsLZB9MHCUlj5MPqUJpcwb3lgXISEugcV3\nLybpsiTOHnC2nI/eyFn9z+Lrw19TainlzjV3cumoS1m/bD1x8XEY04xERkUSFiK9UDdPvBlKpbs9\nN8hMzB9j+PLglwQYA5iSOKXdufoP88debceSYUGza5RtKsN3gC9ZWQGMGOHDPffczZIlS7jqqquI\ndLJIuxObrQKh1QA6LvRLY82ZZ2IymfizXs/qUaOYGNS6o6XFIgMbTaYEj87NnSiRoPA2Or0SCCFO\nF0J8KoTIFkLYhRDznIx5QAiRI4SoEkJ8I4Rwnq/VUwghvQkeFAmGQAPRl0aT+3ou9trG/eWsp7Lw\nHehLxPwI16/VgScBOOEMh/qYBLNJehJMJhMzZgw+oU6QnqZeJNg7sbXSsUiIpKam/e0Gu2YnrTDN\npferD3JMK3KMd3gSNhzfQYAxgMn9mrvt/znrn/x05U/odY2ejpkDZvJjxo/87du/UVlbyaPTH0UI\nwcyzZ1JbU8uglEENY8P8whhllXUoMoJMCL3gy4NfMn3AdHz07ccQBAyXv5eqPVVkPpFJ2cYy+t3a\nj/37YcgQl8x1G7W1RQCEhJxBael6NK2OhISEhgwHZ1itGRiNUej17u2w6UmUSFB4GydyuxgA/AZc\nD7T6thZC/A24EbgGmABUAl8JIbqhWlIn6N/foyIBIO4vcdTk17Dvyn1oNg1LhoX89/NJ+GsCQu/6\nfn+9SHBWKyHUV979d8aToGkah0sOs79oP4eKD+Fn8MOgM3Ddddfx1ltvMWxYAPn50IV+Sh4h2deX\nGk0jt52Wwk0pqa0lu6amS56EEVGODAcXyzPvzt+NQDSKCodI+KViP6OiR2HQNU/jCzIFtQounDlw\nJrX2Wl7Z+gp3Tr2T+KB4AM4+W3obWla/9C+UW0MHcwVjXxrPj5k/dhiPAGDqZ0Lnr+PYv49x+K7D\nJN6RSNjMMPbt6wmRIP8GkZELsNkqKC/f2pDh0BZWa6ZXBS2CEgkK76PTxZQ0TfsS+BJAOI9suwV4\nUNO0zx1jLgPygPMBzxQnOBGSk2HNGo++ReDIQIavGM6eS/ag2TWMEUYMQQZiL4/t1HUMIQ5PwnH3\nbDe8veNt/rSqsZtfUrAs7zt+/HjGjx/f0No3M7Mxwr030L9JrYQ4U+uS12V1dTyVmcnLOTlU2e0N\nHof2RIK//2Dy81dgsaTj69u6zHGQKYjE4ER+zf2VS0Zd0u78SqpLOFZxjKmJU9mctVkWSiorA52O\nLcd3My35DJfsTAlLITkkGYBbJ9/acHy6o0tiS5GQe8xIvI+F7Bp/ouxjCfXNZO7g9htqAQidzHAo\n/LCQoClBJD+QzPHjkJ/f/X/3epEQFnYOen0gx49/T79+/Th8+DDp6Y8SEnI6wcHNt08slgyvikcA\nKRKCnGydKBS9FbdWXBRC9Ef2lGtYfTVNKxNCbAYm09tEwtGjoGly+8FDRC2MAh3sWbQHbJB4eyL6\ngI5rIzRF56dDGITT7YZgX1mMqTMi4YejPzA8cjgvz3kZgMTg5ndjjnYJZGT0LpFQH+G+uawMi93O\nUYsFq11u5RTU1vJ/WVlU2u1cHRvbIChCDYZ2RUJc3PVkZf2TgweXMnLkB40nsrPl5yIujrMHnM3H\naR/zxNlPtJvxUb/VcOGwC9mQsYEjJUcYVFaGFhREWtE+bpp4s0t2CiF4c/6bBJmC8DM2BrdGR0fz\n8ssvN3gUQH58s0sNzB1Uxnv7fPnbsNc46yyX3gaAwFGBWI5YGL5yODqDjv375fHu9iTU1cntBh+f\nGIKDT3eIhFP57rvPOXLkR3Q6P1JTPyM0tLGdtNWaSWho54p99TRVVVUd9r9QKHoT7i7LHIPcgshr\ncTyPhoa0vYTkZKiqkrUSPByUFbUgCqETpD+UTvxN8Z1+vRCizYJKvgZffA2+nRIJG7M2clbyWZye\ndLrT8wkJcn1s2g2yNxBoMBBtNLLU0QVRAEbHom0Qgkujo7knOZl4J16GtjAYzAwc+AR7915KcfG3\nhIXNkCcuvxw2boQXX2TRlEW8/uvrbMnewsSEtrsg7s7fjV7omT9kPku+WkJaYRqDysqoCfDFph1n\ndMxol+d1Rhteh2uvvbbZ8+IcG1V2PadPhQ8Pwv79dEokDHh8AEl3J+HbT4qqffvk8Z7wJOh0vuj1\n/oSE/I6jR+8jPHwUx46VEBNzPTU1h9m5cy4jR35CWNhMNE1zbDd4nydBbTcovImTq3dDU5qmQXpY\nJABEXhhJ5IUn/j4dNnlyUSSUVJeQVpjGnVPvbHOMjw/ExtIrgxc/T02lqK6Ogb6+JPr64tNBd0lX\niIpaTE7Oyxw8eDOnnLIdnc4Ihw9DQABcdhm/u+QSBg6LYuWulU5FgqZp1NTkcqRwIxNjkojx1RPs\nE8jewr2cV1ZGpZ8egSA1KrXLc23J3q8rgSCGn2FiwDoaPAGu4hPhA01iaPftg/h4yM+/i/LyFGJj\nL3frfNuitraooQJmSMjvsNuXUVv7AnY7BATcxqBBMezefSE7d87DbB6Lptmx2Sq8crtBiQSFN+Fu\nkZCLvMGLprk3IRr4tb0XLlmyhODg5n0MFi1axKJFi9w8RQdNRcKpp3rmPdxIR02eSiyuZTdsyd4C\nyHz89khMhHXr4J132h6j08G550JwN3bpPcUD+7lCCAYNep5ffhlHWtqf8fcfAtOOwtkzIcAfPn2f\nt+qCeKfodY4cCUUIKUw0rZaKih2UlW2itjaPWb4wazBs3pzEB5MElZaH2T0mDJ+NtQwKH0SAj/vb\noR9YbwGCGHyGL4P/23mR0JL9+6UXIS/vP+j1gd0oEgobREJg4Bj0+iCiowEsZGfn0q9fMiNHfkR6\n+kNYrTKt2Gw+pdn2gzegRIKiK6xcuZKVK1c2O1Za2n6foK7iVpGgadoRIUQuMB3YASCECAImAi+0\n99pnnnmGcePGuXM67RMaKtvQeTjDwV2050kI9XW9f8PGrI2E+4WTEtZ+VuqECfDcczQEMbbFY4/B\nsmUuvXWvJjBwNP37P0B29gscL14D59ogbBP4+sJFvojyQs4OgKNZ/4epIbVQ4O8/jNjYqzCbT2HR\nR5czd/Bc/jhqEf/afD+2mkwioo5Rt7SWKQbPCNHDv9VhEHbiEnQMHgyrV3ftevv2weTJdqzWbMBG\nVdV+/P09v/dQW1uIwSDLlOt0BgYPfol+/cKA2WRmZjJp0iR0OhP9+z/o8bl4EiUSFF3B2Y3ztm3b\nGD9+fBuv6DonUichQAgxWggxxnFogON5vd/vWeDvQojzhBCpwHIgC/jEPVN2E91QK8GdtNfkqTPb\nDZuyNjEpYVKHJZeffVamQLb3mDAB9uzptCm9lqSkuzjttBxOC/qa0xbAaf7/k88nZTL5iiBynwrh\nnZLz5bHTcjjttGzGjPmWAQMeQgRMZU3uceKiziM8/Fw083ye3mdh7JvjqDPYuSj8R6qqutZN0hlH\nDmrEBtnQ6aQH4PBh2VPqRLDb4cABGDiwFJBeq8LC7vlv23S7ASA6ejFJSefg7+/fbq0Eb0OJBIW3\ncSKehFOA75EBihrwlOP4W8AVmqY9LoTwB14BQoD1wGxN01xLbu9OvEgkGIINVB9yXrggxDekVctg\nZ9g1O5uyNnHbabd1OFYIeRPdHsOGQZprNYa8i/rc/HhHkGlQEOLqq/nzKy/w0NT3sZ77AiZD8+DI\n+joK9Z0jh0UMo8RSgv1AKRkvQtL9gWzZMgSdrnW5EJ3ODx+fGHx8YtDrg5wKOH//4YSHzyEoaCJC\nyOwYa66VY+UG+o3QsNtrGDTIh7o6+ZEeNKjVJTokM1OKv6Qk6c738xtCYeEqEhM7/rx0ldraQvz9\nhzU7JoRo1jLa29E0TYkEhddxInUS1tKBB0LTtPuA+05sSt1IN9RKcBcdxSS4UhFwX+E+Sq2lHcYj\nuMrQobBqlcezSLuf7GwZcNE0Ve3mm/F99lku2Gzhs/2f8Yfhf2j2kt0FuzHoDAwKl6vz0AhZ4Kiq\n6BjF/nDW8NUYqn9C01p7g+z2SmpqcrFac7DZKmldUNLGsWP/IiNODTQhAAAgAElEQVTjEYzGCEaO\nXEVw8BTKt5SThy+pg6vZsCGW6OjvgYns339iIqE+niEp6RDV1ZCQcBMHDtyE1ZqLyeTZ5CTpSWjd\nFbUviQSLo6S4EgkKb+LkzW6AbquV4A7ckd2wMWsjAsGE+AlumdPQobI1QV5e8/XU68nOhuhoMBob\njyUmIhYs4M41n3DK5zcwOWFyQyVEkJ6EweGDG0ohDwwbiF7o0cpKqQn0IzE0FRE26oSnpGk2ysq2\nsHfvpRw79gbBwVMo21xGgS6emKS92O3VaNrT+Pn9l/37Yc6czr/Hvn3S5MjIveTkhBEVdTEHDtxC\nUdGnxMVdc8Jz7whN05oFLjalX79+7N7tWrXLlqSnp/PYY49hdXSUPFH0ej1Lly5lSBeLR1RVyY6b\nSiQovAklErqpVsIJoWnw44/w6qvoay6grrT1nRa4nt2wKWsTI6NGYjaZ3TK9oY5Gkfv29TGRkJXV\nuNXQlKVLiXv3XVa9bCNj+Qhi4yegCwmFiAgmZH3ONF8zFD4A4eH4XH01A8MGElC9n6CI+C633RZC\nT3DwZMLCZlNSIjtEFm8qp9DuQ2io7DhZXPwRKSm1HDhgbO9SbbJvH6SkQF3dUXx9EzEawwkJmUZh\n4SqPigTpPbG26Un48ssvT+i6N9xwAxs3bmRwF4s+bN++ncjISB5++OEuXUeJBIU3cnKLhP795c9u\nqpXQKdavh5tugu3bQQgMw2Oxlc9Cs2sIXfMFJ9Q3lFJLKXbNjk60vRO0KWsTkxMmt3m+swwcCAaD\njEs4w7WKw95BdrasKNWSU06Be+4hZesGvj/6A7+WH2CcbSBi3z5mHM0iyOAP61+G3Fyw2RgRMZTA\n2v2Ex/R329RCQqaRk/MCluocjvxsxY4gOHgdUVGLKCz8mISEw+zf37k73i1b4Mkn4cMP4U9/qi93\nLMtuRkScz6FDt1FXV4bB4JlywvXVFtvyJOTm5lJTU4OPj+vtXzZs2MDq1at59913WbhwYZfmN3v2\nbHbu3Nnp161evZqHHnqIjRs3AkokKLyT3tcPuDtpWiuht3HTTfLnV1/BdddhKD8GGtSVOe/foKFR\nbi1v83Jl1jJ25e9yWzwCSNf0gAF9MHgxO9u5JwHg/vsJ+3wNZW+/xinnHEV/+nfof7+ThCV2vln/\nFuTkyGqNDz/MKUJeIybWfSmEwcGySmZRxvccK5ceg5CQnwgPn0dk5AIiI9ewf7+GpsF338kU1Trn\nu1QA3HorTJwIv/4KL7wgH1ZrRkPjpIiI89G0GoqK/uc2G1pS37ehLZGgaRo5OTmtzrWFpmnccccd\njBkzhgULFnR5fqmpqScsEjZt2kS1o1OaEgkKb+Tk9iSEhEBQUO8TCbt3Sw/CqlUwcyZs2YKh9AdA\ndoI0hjR3Jzdt8lTfy6ElW3O2oqG5VSSA3HLotSKhuBieeALuv1+WkXSV9kSCg8vHXk5MYAyZZTKo\nztfgy5xBjkCAe++F//yHCz+U+aH9+o04oek7w2SKxc9vEMcL1pLHWACiojIJDp6Kr28/YmNfJTPz\nekaPhvp1rbhYioWWfP45PPOMPLd0Kej1coG1WNIxmWSzK1/fRMzmieTmvkl09MVus6Mp9W2i29pu\nAMjMzCS5XtR3wBdffMGGDRv44osv0LmhImdqaipPPPEE5eXlmM2ub9Vt27YNgIKCAhITExvEghIJ\nCm/i5BYJvbVWwooVstjT7NnyeVwchlJ5J+UseLGpSEiidSdDgIxS2Yihf6j7XN8gRcJ7vadtV3M+\n+AAefRSmTIG5HXdFBGQOYHFxhyIBYPag2c5PJCbCX/7C4OeeAyA58cQDFp0RHDyNkqwN5HM7wcGV\nhIZG4uubgMkUz+jRj6PX20hO1vPss7BtG9x2G0yaBBdc0HiNvDy44gr5a7nttsa43bq6Umy28mYt\nmOPj/0Ja2p89VlipI08CwI4dO+jfv+PPrqZp3HnnnUybNo1zzjnHLfNLTZXltHft2sXkya5t19XV\n1bF9u4wVyc/PJzExUXkSFF7JyS0SwDWRUFUlb7M60TjohNE0KRL+8IfGu9/YWPRUAB2LhLbIq8wj\nxDcEX0MHxQ86ydChssdDdTX4+XU8vltZt07+/OQT10VCdrb86SwmoTPccQfitdegshJ9SGjXrtWC\nkJBp5B77N3kmHdHRWQ1bEEIITj99Ol995UdgoFxQp06FM898jj/+8TTeeONCEhPT0TS4/faVwDhe\ne615Yo/VKsVkfUwCQGTkQg4eXEpOzkukpDzjVlugsbmTTtd68TSbzURGRnLjjTdy4403unzNDRs2\ndDlYtJ5hw4ah1+vZuXOnyyIhLS2tIeWxoKAAUNsNCu9EiYTkZLnv7/iP3ICmyduw5cul2//88+Xi\n7Wk2bZKiZfHixmNxcRioBJyLhFA/uQi1l+GQV5FHdEC0W6cKUiRomqzUN8q9N8xdQ9Ng7Vop7D79\nFF5+WQq9jqgXCS54EtolOhr++ld4+GG5reVGgoOngdAoCKslImIfwcFTG87Fxl5JbW0hdrul4dgz\nz2zgggvGcd11HxEdXUpNDRw6FM/KlWlERw9tdm2LRXb1aupJ0Ot9iY29ipycl+nf/yH0evf2oKit\nLcJgCG9zUV+3bh3pneg2FhcX13D37w5MJhODBw/uVFxC/VYDKJGg8G6USEhJkU0KoqKcnx82DMaM\nabwr9TQrVsgFatq0xmOxsQ0iwVlBpWCTjEMori5u87J5lXlEB7pfJNSnjqel9TKRcOSITGW8/Xa5\n5bB5M5x2Wseva1ltsSvcdReMHw9xcV2/VhN8fZPQVcSQp9cYE51OcPCMhnN6fQD9+z/Q6jXffAPP\nPw91dYFomkZY2O2kph4F3m02zmrNQAgjPj7Nc1rj4q4jM/Nx8vJWEBd3tVvtaatGQj1Dhw5l6NCh\nbZ7vDlJTU9mxY4fL47dt28agQYPIy8sjPz8faBQJfr3O5aZQtI0SCVdeKVMhbU6qGSYmSoHwwQdw\n0UVyIzfa/QttA7W18N//yjy0pgFXERHo9DaEpjn1JBj1RiL8IzhWfqzNS+dW5HrEkxAWJvVVtwcv\nahpcfTVcd51MTWzJunXSj/7//h+8/rr0BrkiErKzZVvLwMCuz9HPr3kggJsQQqA7NJbcolhiYorw\n9+94AU1Jkf04HFcgKyuBgwefxGLJxNe3sd2yTH/s19Dpsh4/v2TCw+eSnf08sbFXuc2VD/UiwXkN\nkN5Camoq33zzDZqmuWT7tm3bGDduHFu3bm3mSfD19XVLMKVC0V2oT6u/v9yvnj+/9WPsWLnQ1HfY\nauJC9Ahr1shtj6ZbDQA6HSI2Br2pts2qi/HmeHLK204Ty6v0zHYD9FCGQ3q6XPxfaKO56Nq1MHo0\nhIfDvHkyLsEVXMhs6A0c+v5sqqvNDB3qc0ILdkzMn9DrA8jJebHZ8abpjy2Jj7+Bysod7N59IXv2\nXEpa2uWObpFdo2Vzp95IamoqJSUlLqVi2u12fv31V8aNG0dkZGQzT4LaalB4G0okuEL//nJfeetW\nz77PihVyxR0zpvW52FgMBmubIiHOHEdORTsiocIz2w0gtxy6XST8/LP8+dlnzgsBrF3buGUzf75s\nTODKJL1AJGiaxvI15xMdfZS5c0/MdW0wmImNvZKcnFew2aoajsv0R+ciITT0bGJi/kxdXQk1Ndnk\n579PVtZznX5vm60aq7XR69XRdkNvoD7GwZW4hIMHD1JRUdEgEpp6EpRIUHgbSiS4ghAwbpxnRUJV\nFXz8MVxyifM+EnFxGERVm02e4sxxbXoS6ux1FFYVetSTsG+fbDXcbfz8s6zmVFQEP/3U/FxmpoxJ\nqC8DOWOG9BitWtXxddsqydyL2LWplu8rkrnuwr0kJPzxhK8TH38TdXWl5OX9p+GYxZKBr6/zNFoh\ndAwd+gZjxnzPmDHfExt7Fbm5b2K3u96bWtM0du/+A7/+OgXN0cmqrs55c6feRHJyMgEBAS6JhPqg\nxbFjxxIVFaU8CQqvRokEVxk/3rMi4bPPoKICFi1yfj42FoO9rH1PQhsiobCqEA3NY56EoUOlxsnu\nuufZdbZskdtEsbGttxLqg0xPl6mB+PnBOee4tuXQVknmbkTTZAiMwSAfRiNcc02jw+SxRwXh1HD9\nxVPw8TnxO3A/v/5ERMwjM/NpNM2G3V5LTU1Om56ElshMinyKij53+T3z8pZTXPw/LJYjVFbuare5\nU29Cp9MxcuRIl0VCUlIS4eHhypOg8HqUSHCV8ePlHWrLVEl3sWKFrI87cKDz87Gx6GuPtysSjpUf\nw661vp3Pq8gD8KgnAWDvXo9cvjU2mxRsEyfCeefJxb9pf+W1a2H48Ob9OBYskOmlbcUw1F/32DGP\neBJsNpkm6gpffw3vvy8TM154Ae6+G/79b2lCWhq8+7mBhWQSPLDrdTsSE++gunofBQUfYLVmAVqb\nMQktCQxMxWyeyLFjr7k03mrN4eDBvxIZeRE6nT9FRaux26uw2y29XiSA6+WZ64MWASUSFF6PEgmu\n4sngxeJi+OILudXQFnFxGKxF1B1vWyTYNBv5lfmtzuVVOkSChzwJycky6eMEm/V1nn37pNdlwgRZ\nv+LQIVnKup5161p3nLr4YliyBG68Ef75T+fXzcuTq7kHRMJzz8nYjW++aX+c3Q533CGLID34IFx7\nLdxzj9wp+fJLOPVUCPG3M0eXg090J0pNt0FQ0ATCwmZx9OiDWCxHAVz2JID0JhQXf4nFktnuOE3T\n2L//LwhhYvDglwgNPZvi4tXtlmTubaSmprJ3717q2mmGoWlaM5EQFRVFZWUlVVVVSiQovBIlElxl\nwADZ58ETWw4ffCAXp4suanuMo1ZCXZHF6ek4s8zFd7bl4GlPgk4HCxfCu+86zyR1O1u2NGadnHWW\nTFes30rYulWKiKZ1JkCOf+opWLZMFjl66qnW13VXtUUnfPihdHYsXtxYisEZ778vmy394x/NQ1Pm\nzpU1vwwG+MvkEkLijQi9e9IQk5LuoapqN1lZMkfSVU8CQFTUxeh0fuTmvun0fE1NHrm5b7N79wKK\nij5l8OCXMBrDCA+fQ2npT1RXS/eKt3gSrFYrmzZt4tixY04fW7dupaSkpJknAWRBJSUSFN6IqpPg\nKjqd54IXV6yQwXXt1WBwVF20lTgPEmsqEsbFjmt2Lq8yD7OPGT+j54q4LF4s75bXrpXrtkf5+Wd5\nWx7kaF08a5YUCYMGyQ6MY8fCnDmtXyeELKxkNMr6CUVFsiJi/WrsrmqLLcjPl7GVjz0mCxpddBH8\n8EPrnlO1tfD3v8upT53a+jrTpslrHby6gOoK95UIDw6eTGjoDIqKPsVojECvd30hMxjMREUtJDf3\n3yQl3dWsvkJe3kr27r0E0AgMHE9KyrNERsq6EeHh5wJ28vLeAbxDJIwaNQqdTsfp9bEubSCEaCUS\n8vPzqaqqaniuUHgLSiR0hvHj5S2hO8nMlCvrm2+2Py42FgMV1JU7TyGICohCJ3RtehJiAmOcvMp9\nTJggnS0rVnhWJHz4IWxfNRYizkZ/v3THx5x/Plx6qXRnLFoEr70msxmcIQQ89JCsArV0qdxieOUV\neYuelSUFRIR7F6zVq+XPP/1JLvTTpsn4gtGjm487fFjunLT3ETMawZplxZTg3j4iSUn3UlLybae2\nGuqJjb2a3Nx/k5PzCvHxfwGkB+HAgRuIiPg9gwe/iI9P84qmJlM8gYFjKCh4HwCDofdvN4SHh7N5\n82by8vLaHRcVFUVMTEzDv6HRk6CqLSq8DSUSOsP48dJNXVQki/S4g/feA1/fjivzRUaiF9XYqgWa\nTWvlajboDMQExjgXCR4qydwUIeT6/PzzMtjOE72wNA2uuEJDX/Z7zNUGsh6QOw1Lr5wDI0bAZZc1\nb2nYHrfeKktFXn457NolCy7t3Cm9CG6uiPfJJzB5snQURUdLTfLAA/Dbb63H3nFHx+WtrVlWAlPd\nUBGyCSEhUwkLm4XR2PnPSXDwJOLibuDgwVsIDBxLcPAkDhy4GSH0DB78cpsZGGFhc8jIeBghTG7v\nB+EpTnFW3bMdIhyCU203KLwVFZPQGTwRvPjZZ3Krod513hZ6PYYQqenqyjuXBunJaotNWbwYSks9\nF8BYWAhlZYJXuZr0L9MYNkyWQyAkRC70y5a5JhDqufRSOdmICHj8cVkSOznZrXOuqpLZCvPnNx67\n/HI57/T01o+HH27/epqmecSTADBy5KcMHfr6Cb02JeVpzOZT2L37D+TkvEZBwXukpPyz3RTN8HC5\nJWQ0tt3cydsxmUwEBQU1bDcokaDwNpRI6AwpKWA2uy8uoaQENmxwuY2xIUK6KjtbUMlTHSBbMny4\nvAv2VLPMQ4fkz4GGDBg9mv79O+7y3SHTp8v9gKIi+OUXeOutrk6zGd9+K9toNxUJXcFWZsNeafeI\nSNDpjAjhQqdMp6/1YcSI99G0Wvbvv5qwsHOJimqj5oeDoKAJGI0RXhGP0BWioqKUJ0Hhtajths6g\n08GUKdJ/fPvtXb/eF1/IdABXRUJ0IBxw3i4aIC4wjk3Zm1od747thnoWL4b77oN//Uv+uoRo/Bkd\nDbNnn/i1G0TCSD8wmUhOhu+/d8eskTEJ9Z4iN/LJJzLGsr5bZlexZlkBPCISuorJFM+IER9w+PAd\nDB78UofeASH0REVdTG1tYTfNsGeo79+gRILCG1EiobNcd53Mzf/lF+fdBzvD55/LjAkXo+n1cXJL\nojNVF212GwWVBd3iSQApEh59VAYUOuPnn5v/2vJW5lH4cSHDVw7vMKXv4EGI0hdinjwSkDsDR4/K\nWIXe6K222eRu0uWXu++avVkkAISEnM64cRtcHp+S0vneD96G8iQovBm3bzcIIe4VQthbPPa4+316\njLlzZQvp9ir3uUJtrfQkuOhFADAkhAHti4T8ynxqbY1pkkXVRdg0W7d5Evr1k7somiYLA9lsspyw\nxSL7ZDW2KwabxcahpYcoeL+A7Jc6rul8aEclA237GwolJSdDZaXcKeiN/PCDLNDprq0GcIgEAT6x\nXS+k1BsQQvTZeIR6IiMjyc3NxWq1KpGg8Do8FZOwC4gGYhwPJ1nfXopeD9dfDytXdq1E848/wvHj\nsqywixj6yxzrjmol5FbkNhzzdCElZ1QfqkbTtIatBr1eZjvcfLOMDawvR5D7Zi41eTWETdA4ctcR\nrMes7V730PZyUsQhmDkTaIwx7HJcggewWGRxx4kTYdIk913XmmXFJ9oHnY8KJ/IWIiMjSU9PB1Ai\nQeF1eOqbpk7TtAJN0/Idj2IPvU/PcOWV8ufrTSLBf/tNVk585RV45BFZrOfyy2UEfUlJ62t8/jnE\nxMjtBhfRJcYgqKMuy8n1cF510dMlmVty7N/H2JyymdL1pa3OXXGF7LX04otgr7WTefdeIsVahm2/\nCJ1RcOj/HWr32gezfBnYrxZCQ4HeLRIefFDGULz+unszKj2V2aDwHFFRURQ53F1KJCi8DU+JhEFC\niGwhxCEhxH+EEP089D49Q0SELArw0kuyq9H8+bLK34IF0svw9NNyMzotTYb6f/pp62t89pncaujE\nCiLiHAWVso87PR8fJGMbmomEbvQklP5Yyv7r9gNQ8WtFq/NBQVIovPKKxtGzn8VSaCDpdzkYrYUM\n+H0e+SvyKVnjXACVF1rJt4YwcFJjxbqwMJlscuSI+23R7FpDK+PO8uuvsrri3XfL8g3uRIkE76Np\nlUUlEhTehidEwibgz8A5wHVAf2CdEMI7qqW4yk03QUaGXAV++02KgeJiGWtQWCj7B2zcKMXDt982\nf+3+/fLRia0GAOLi0FNJTUbrBRgg3C8co85Idnnj/n5eZR4BxgACfDz767dkWNj1+10ETTITEFtN\n5W+tPQkgtxyKi+HVtZMIH1VB4Ncvw+mnE5PxGkFTgjhyr/MV//CHvwKQct6whmNCNAYvupvdF+3m\nwPUutm1sQm2tFEIjRrgnAaYl1iwrPvF9Ix7hZEGJBIU34/bsBk3TvmrydJcQYguQDlwEvNHW65Ys\nWUJwcHCzY4sWLWLRovZzrXuMceNk1b7wcNldsK1yqzNmwPLlzUPwV62SVRanT+/ce0ZFEc5mclZf\nQNgXRYTPbl71UQjRKsMhr8Lz6Y+1RbXsnLcTnZ+OEbeWceiCdVT+eBbQ+jZ6wACYHb6XFwtPZb1e\noD9TwLF34eBBrMlDsaTXEDxN/q4uuQSuuUa+7uBne4FJDJzZvJV2S5Fgr7NT+FEh4fPC0fueWM6/\nzWKjeHUxhjADg14c1KnAujfegO3bZQ8qo7HJvGrtVB+qJmDoiYs1za5hybAQlRDV8WBFr6G+NDMo\nkaDoGitXrmTlypXNjpWWOr8hcxceT4HUNK1UCLEfSGlv3DPPPNPQFMVrcNZJsCUzZshqfnv2NPqe\n33tPdvEJ6OSCYTAwMPpDLCHT2H3hbkZ/M5rgKc2FVSuR4OFqi5Z0Cztm7aC2sJbR343G56OnCOAI\nhenCEbzYeoG9oXYtGM2EjXDsQkWHwaFMav3jKNYCCQv0Jy3Hh3/8A66+WmqrQ5sLMRuriYhsLsaS\nk2HNmsbnBe8XsHfxXvrd1o+BjzcXFK5S9lMZdoudmpwaqg9V45/i2he71SrbQixc2Do7NuMfGaQ/\nlM7knMn4RJyYJ6BsSxm2Mlurv7mid6M8CQp34ezGedu2bYz3QI2XejwuEoQQgUiBsNzT79UrmTJF\ntvv79lspEg4dkhUbly07ocvpJoxn+M9L2THmPXbM2UHc1XEgQBgE8TfFOxUJ7mzuVHWgCmuWFWOY\nkbryOvYs3IPOpGPsT2PxH+QPt6wlgCpsNQasGVZ8k3ybX8BqJboUnpywjWFv14eq+MKcFdhKXmWD\n8X5S5qawPTmeOXPgwAEYrDvIocIgBva3IkRrkdC0VkLOKzno/HVkPpVJ5IJIgk7toNy1E0q+LcEQ\naqCutI7SdaUui4RXX5WZG/fd1/y4rcpG1nNZaLUaRZ8VEXt5bKfnBFD0aRGGcAPBpymR4E1ENGkY\npkSCwttwu0gQQjwBfIbcYogH7gdqgZXtva7P4u8vhcK338Itt8D778utCWetjF3hX/9CP24cqfY7\n2TvlaQo/kdXqLOkWEBA3NY60wrSG4XkVeUyMn+gOSyj6XxG7LtiFVtMY0Bc4NpDU/6ViijHJW+mN\nGwkIjIMKqNxV2Uok2H/bTQUDiZ7S4sty8WL0l15K4CgjZVvKOONP8ZhMsrXCYPEFh8RwUka3bmqU\nnCz7IxQWgn9hJaVrSxn69lCynsli35X7GP/LeJfTBY8fhx07YN/HtfiMTaDwaC3f/wtIlzovLk4+\n6neW9HrpMTCZZOnlRx6RySwtqyvmvpFLXUkdfoP9KFxVeMIiofCTQsLnhndYdErRuzCZTAQHB1Na\nWqpEgsLr8IQnIQFYAYQDBcAGYJKmab205E03MGOGLENYWyu3GubO7fxWQz0xMfDhhxjOOIPUa16F\n1c8DkHZFGgUfFhA3y8l2gxtiEopWF7Hr97sImx3GwCcGUldSR11ZHcGTg9EHOPb+f/4ZLBZMt8xD\n/89KKrcWET6nedxExeq9aMQTdF6L3af588HPD7N/Bse3GAgIkC2Vv/gCbtZWc9C0gIuHtP649u8v\nfx49CiHv5GCMNBK1IIqAkQFsO3Ub6Y+kk3RXEgBCLxA65wvsnj0wa5bs3A1DwKGz9Ic1ojOhpkYK\nkZYkJsI998iCTvn58t9NsdfZyXwqk6iLojCfYubI349gq7Q1/s5cpOpgFVV7quj/UP9OvU7RO4iM\njFQiQeGVuD27QdO0RZqmJWia5qdpWqKmaYs1TfNAkpoXMWMGlJfLDIhff4WLLura9SZPhueek1Uf\nP/4YgMgLI6neV01SfhIllhKqa6uxa3byK/O7HJNQtFp6EMLnhDPivRH4D/InaEIQYTPCmi9269ZB\nUBDiT5cRwBEqfzrW6lrl6woQ1BEwObL5icBAmDcPc/Z3VKVVUVdax6xZ8MMPGqXfbyPTGsXAJiEG\nmk3jwC0HCDogS3Ac2mcn7608Yq6IQWfSYR5jpt/f+nHo/gy+9VnPtz7rWRfzE3VlratVbtggnT0h\nIbDmiRLeZjM7v7ey89+FfMVaDv1koaBAOkrS0xuTU7ZskYWSrroK/vY3WRZjYIswiMKPCrEcsdDv\ntn5EXBCB3WKn+MvOlw0p+rQIYRKEnh3a6dcqep6oqCiEEJg80UNdofAgqmxbdzB+PAQHyzgEf384\n99yuX/Paa+UKtVyGeoTOCEUfpCdqg4ykzinPoaS6hDp7XZc8CdWHqtlz8R7Czg1j+H+Ht++6X7sW\npk6FESMIEBlU7qlqNaRsLwSGFjvPPPj97wnK/Bo0KN9azqxZYLEIltcsxK7pSGnifMj/bz7Zz2WT\nsXAHZl87uz6toO54nYzRcHBwajLzfKZxDmdwDmfwu4LTSOoPZ54py1xccon8OWMGjBkjNU7CkQIG\npcDIM00MmReMHji+Vtal8PGRnoNBgyDWUoH/k7t5bVk527bJjNgHHmhujqZpZDyeQcj0EMzjzPgN\n8CNgVAAFH3e+Umfhp4WEzgjFEKjarXgjkZGR+Pv79/kS1Iq+hxIJ3YFeD2edJf3R550nhUJXEQLm\nzYNvvgGrFZ1JR/jccExfyzuVRR8uYt6784ATL6Rkr7Oz9497MUYZGbZ8GDpjOx+X2lpZanraNPDx\nISDWQmWOD5qtSUEiTaO8MBzzELvza8yahb8hF73JRtmWMoYNg36Bxbxg/CvQeJdur7Vz9N6jhM0J\nI+mORCItVez6uILQmaH4DZQBA1Yr3HCTjrGnCJYvl1rq7yOymeVXSGws5OXBsWPy57XXytiHkBAZ\ntBg6Q96tG8ONBKQGNIgEkEGIh24/xNZxWyl4v4D91+9nzBiN556D2Fgo31bOxn4bWW9ez3rzeiq2\nVpB4W2LD6yMuiKDo8yLsNW38Dpz9aotqKV1fSsS8vt1SufqwCSkAABCHSURBVC9TLxIUCm9D3ZZ0\nFzNmyK2Brm41NGXOHLjzTnkHP3Mmkb+PJH9FPsvil5ERmgHA8IjhjI0de0KXz3g4g7ItZYxdPxZD\nUAcflW3bZLclR/OlgGG+aDl6mUI4WH451u04TJW9H4mn1Ti/RlAQYvqZmDdnUL4lGoHGLL7k1drF\n+Pg0NsvMW55H9cFqhr8/HPMYM4O+spC73ZeEJQkNl3rySRmn8MknMHy4PJbv58OeBXs49R4zAcNa\nx4RYMixU769mwCMDGo6FnBFC0RcynKb6aDU7ztmBJd1C8n3JmE81s+OcHRS8V0DUwijsNXbSLk/D\nEGIgYamcizHcSOjMxi2CiPMjSL8/neM/HCdsZlj7v1MHRf8rAjuEnxfe8WBFr2TgwIHExLgvy0ih\n6C6USOguFi6U9YNPNKvBGampsu3i6tUwcyZhs8LQ+em4oegGEq9K7Pj17VC6qZSjDx4l6e9JBE92\nIeVu7VrpIXHk6wZMiYM1ULmjokEklK/aA5gxzx/S9nXmz8f89WbyNw+GPXuYVfEBr7KYAQOkQ8Zu\ntXP0gaNEXhSJeYwZgCFTffmm2kT4LOnKTU+Hhx+WNa7qBQJAxHkRGMIM5L6R21BDQdM0agtq0Wwa\nhasKQUDI70IaXhM8LZjs57Mp/raYtD+lofPTceqOUxtsCp8bzuE7DhNxfgQZT2RQubuS8VvGYx5n\ndmpe4OhAfJN9Kfy4sE2RYKu2kfl4JlUHqjCGGSndUIp5ohlTrNrP9laWLFnCNfWVwRQKL0JtN3QX\n4eHwxBMyX85dCCFFx+efg6ahD9ATNjuMgg8LqC2uJeP+/fw64FOqn1whI+060Yvg0K2HMI8zk/T3\nJNdesHYtnHZaQ5lB42nDMVBK5YashiFla/PRiyr8p7YjYObNI0jbgzWnFuvKbzjL9BN6nUacvYr8\n9/JJfzgda5aV5PuTG17Svz8cPSoazFuyRPaAuvvu5pfWmXREXxpN7vJc7LV2bJU2ds7ZyU/RP7Ex\nbiMHbzmI+VQzxrDGUokh06Rg2HHODozhRsZuGNsgEAAGPD4AS4aFA7ccIP3BdBJvS2xTIICsihn5\nh0hy38zlwM0HqD5c3ex86Y+l/DLmF9IfSceaYeX4D8epLagl7pq4Nq6o8AZMJhNhYa55jhSK3oTy\nJHg7c+bAyy/LXhFDhxJ5YSR7L9nLxoSNaNYa9HaNvbcVM/a2YYjYaFk3+Jxz2r1kxfYKyjaWMeKj\nEegMLujI0lKZInDbbQ2HxJjRBLCSyp8b6ySU79Uwh+a3mYYIQHw85lEm2AElHx6hMPRBpufmMTSr\nij0L5RZKzJ9jmpU3Tk6WdQp++016ED7+GN59VzZ/aknsFbFkP5dN3n/yyHklh6rdVQx5fQg+MT6g\ng4CRzbchfKJ9MJ9qBh2M+t+oZgICIGBYALFXxXLslWP4DfEj6d6ORVXSvUnoAnRkP59N9gvZmMeb\nEUaBZtMo31JO0MQgRq4a6XRLRKFQKLoTJRK8nbPOkn0gVq+GoUMJPy+csNlhBBWtI27rfVQ/+19+\nvXUkGZd9RVLuUzB7tiwJ+Pe/t9mBMuflHHzifAif68IeeFmZFB16PVx8cePx6GgC/HI5fmBQ49CC\nMGJOdd7BsimmBWfis7OYtLR56P3gnVUGIuYPoLa4H9UHqwkY0XzxrG8ZPWGCdNisXCl3d5wRODqQ\nwHGB7LtiH8YII2N+GIN5fNt3/gBjfhiDzlfXprhJvi+ZqrQqBjw6wKV+EYZAA/3v60/iskRyl+dS\nvrm84VzMH2OIuy5OFUxSKBS9AiUSvB1/fykUVq+GpUsxmA2MmvWDrO74r3/hc/V0EnMPc/SpTMI2\n/Rfz5GelSPjlF3nLrW++qNWV15H3nzwSbk1oP5sBZO2H2bNlS+w1a2iWowgEJGnk7POjfFs5WtFx\namxhmE/rePET588n7O63qCCF4V9Mx/8MGdVvDDNinGBsNT4lRQqFWbPgH/+QWQrtkbgskfRH0mXN\nhyEdR5zr/dtf+E0xJsb+0PngUL2/nvjr4mWvVIVCoeiFKJHQF5g7V/ZgfvNN+Vi7FpYuld2RgOT7\nkyn6ooi9l6Uxdt0dGMeNk+mTzz8vxUQT8t7Jw1ZlI/ZqJ6WDq6pk58s9e+Tz7GxZhvCbbxoCFpsS\ndEogpOnYOn5r47H5gzu2Z8QIhvT/COHvB2dc2+HwwEAZE+oqUQujiFqoOikqFApFR6jAxb7AnDlQ\nVydL/oH0tz/+eMNpnY+O4e8Mpya3hm2TtlE1+Cy44Qa46y6ZCuBA0zRyXs4h/LxwfBNaNGbKz4ff\n/Q7efltWFEpOljUR1qyRfn4nmM9OYhILGedzC6MMdzDmgu8wTR3WsT1CIF56EZ59trO/CYVCoVC4\nEeVJ6AskJsJnn8GAAc1z/poQMCKA8ZvHs3PuTrZN2saQ52/D771f4I/3wv/9HwhB1b4qKrdXMuDR\nAc1ffOCA9OVXVkovRcs+yG0xfTq+ZwzDd/ZsuOIKiIzs+DX1dBBcqVAoFArPo0RCX2Hu3A6H+A30\nY+zGsexZsIfdlxwG/gH5wJjG7QDfAb7N8/erqqRAMBph48bGjkquEB8PP/zg+niFQqFQ9CqUSDjJ\nMIYYGfXlKCq2V6DVaXD77fD99w3nfTV/RPFXEOEoAfzggzL2YOfOzgkEhUKhUHg9SiSchAi9aCz4\n88UzcgvBZpMND665BhYsgK+/llkLTz4J994ruxopFAqF4qRCiYSTHZMJZs5sfB4eDtOny7aG27dL\ncbBsWc/NT6FQKBQ9hhIJiuacfjq89BJcdZV8vn697JGsUCgUipMOJRIUrbnySpnyaLfD1Kk9PRuF\nQqFQ9BBKJCicc8cdPT0DhUKhUPQwqpiSQqFQKBQKpyiRoFAoFAqFwilKJCgUCoVCoXCKEgkKhUKh\nUCicokRCN7Jy5cqenkK3cbLYquzsWyg7+xYni52exGMiQQhxgxDiiBCiWgixSQhxqqfey1s4mT6w\nJ4utys6+hbKzb3Gy2OlJPCIShBALgaeAe4GxwHbgKyFEhCfeT6FQKBQKhfvxlCdhCfCKpmnLNU1L\nA64DqoArPPR+CoVCoVAo3IzbRYIQwgiMB9bUH9M0TQO+BSa7+/0UCoVCoVB4Bk9UXIwA9EBei+N5\nwBAn430B9u7d64Gp9C5KS0vZtm1bT0+jWzhZbFV29i2UnX2Lk8HOJmunryeuL+RNvhsvKEQskA1M\n1jRtc5PjjwHTNE2b3GL8YuAdt05CoVAoFIqTi0s0TVvh7ot6wpNQCNiA6BbHo4FcJ+O/Ai4BjgIW\nD8xHoVAoFIq+ii+QjFxL3Y7bPQkAQohNwGZN025xPBdABvCcpmlPuP0NFQqFQqFQuB1PdYF8GnhT\nCLEV2ILMdvAH3vTQ+ykUCoVCoXAzHhEJmqa956iJ8ABym+E34BxN0wo88X4KhUKhUCjcj0e2GxQK\nhUKhUHg/qneDQqFQKBQKpyiRoFAoFAqFwik9LhL6WiMoIcQdQogtQogyIUSeEOJjIcRgJ+MeEELk\nCCGqhBDfCCFSemK+7kAIcbsQwi6EeLrF8T5hoxAiTgjxthCi0GHLdiHEuBZjvNpWIYROCPGgEOKw\nw4aDQoi/OxnnVXYKIU4XQnwqhMh2fEbnORnTrk1CCJMQ4gXH379cCPGBECKq+6zomPbsFEIYhBCP\nCSF2CCEqHGPectS0aXoNr7bTydiXHWNubnG819sJLn92hwkhPhFCHHf8bTcLIRKanO+yrT0qEvpo\nI6jTgf8DJgIzACPwtRDCr36AEOJvwI3ANcAEoBJpt0/3T7drOETdNci/XdPjfcJGIUQI8CNgBc4B\nhgFLgZImY/qCrbcD1wLXA0OBZcAyIcSN9QO81M4AZOD09UCrACwXbXoWmANcCEwD4oAPPTvtTtOe\nnf7AGOB+5PfsBcjqt5+0GOftdjYghLgA+R2c7eS0N9gJHX92BwLrgT1IO1KBB2leb6jrtmqa1mMP\nYBPwzybPBZAFLOvJebnZxgjADkxtciwHWNLkeRBQDVzU0/PtpG2BwD7gLOB74Ok+aOOjwNoOxni9\nrcBnwKstjn0ALO8rdjr+H87rzN/O8dwKXNBkzBDHtSb0tE2u2ulkzCnIoncJfc1OIB5Zl2cYcAS4\nucXf16vsbMtWYCXwVjuvcYutPeZJECdPI6gQpAosBhBC9AdiaG53GbAZ77P7BeAzTdO+a3qwj9l4\nHvCLEOI9x/bRNiHEVfUn+5CtPwHThRCDAIQQo4EpwP8cz/uKnQ24aNMpyFTxpmP2IRchr7TbQf33\n0nHH8/H0ATuFEAJYDjyuaZqzhkB9yc45wAEhxJeO76ZNQoj5TYa5xdae3G5orxFUTPdPx/04/pDP\nAhs0TdvjOByD/M/p1XYLIS5GujDvcHK6T9joYADwF6THZCbwEvCcEOKPjvN9xdZHgf8CaUKIGmAr\n8Kymae86zvcVO5viik3RQI1DPLQ1xqsQQpiQf+8VmqZVOA7H0DfsvB1px/NtnO8rdkYhPbl/Qwr5\ns4GPgY+EEKc7xrjFVk9VXFRIXgSGI+/I+gyOwJhngRmaptX29Hw8jA7Yomna3Y7n24UQI4HrgLd7\nblpuZyGwGLgYucc5BvinECJH07S+ZOdJjRDCALyPFEfX9/B03IoQYjxwMzLuoq9Tf4O/StO05xz/\n3iGEOA353bTe3W/UE3S2EZRXIYR4HjgXOFPTtGNNTuUiYy+82e7xQCSwTQhRK4SoBc4AbnHchebh\n/TbWcwxo6bbcCyQ6/t0X/p4AjwOPapr2vqZpuzVNewd4hkZPUV+xsymu2JQL+AghgtoZ4xU0EQj9\ngJlNvAjQN+ycivxeymzyvZQEPC2EOOwY0xfsBLl+1tHxd1OXbe0xkeC4A90KTK8/5nDPT0fuj3ot\nDoEwH/idpmkZTc9pmnYE+QdqancQMhLXW+z+FhlJOwYY7Xj8AvwHGK1p2mG838Z6fkQG+zRlCJAO\nfebvCTIC3tbimB3Hd0QfsrMBF23aivwybjpmCPKLeGO3TbaLNBEIA4DpmqaVtBjSF+xcDoyi8Ttp\nNDIw9XFkZhL0DTvr18+faf3dNBjHdxPusrWHIzYvAqqAy5BpV68ARUBkT0eTdsGmF5HpcacjFVv9\nw7fJmGUOO89DLrargAOAT0/Pvwt2t8xu6BM2IgPXrMg76oFIl3w5cHFfshV4AxnQdC7y7usCIB94\nxJvtRKaRjUYKWjvwV8fzfq7a5Pg/fQQ4E+lF+xFY39O2uWonclv5E+Tikdrie8nYV+xsY3yz7AZv\nsdPFz+75yHTHqxzfTTcCNcBkd9raG34R1wNHkWlHG4FTenpOXbTHjrwja/m4rMW4+5AqtwrZBzyl\np+feRbu/o4lI6Es2IhfOHQ47dgNXOBnj1bY6vpCednyhVDoWyvsBgzfbidwGc/Z/8t+u2gSYkLVP\nCpEC8X0gqqdtc9VOpOhrea7++bS+Ymcb4w/TWiT0ejs78dn9M7Df8X92GzDX3baqBk8KhUKhUCic\n0uNlmRUKhUKhUPROlEhQKBQKhULhFCUSFAqFQqFQOEWJBIVCoVAoFE5RIkGhUCgUCoVTlEhQKBQK\nhULhFCUSFAqFQqFQOEWJBIVCoVAoFE5RIkGhUCgUCoVTlEhQKBQKhULhFCUSFAqFQqFQOOX/A8Po\nXpo5qw6nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1e61f748>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ts = np.linspace(0,80,160)\n",
    "n_test = sum(test_mask)\n",
    "best_taus_for_x_test = np.zeros(n_test)\n",
    "for i in range(n_test): \n",
    "\n",
    "    drf_vals = [clf.predict( np.append(X_w_dose[test_mask,:][i,0:d-1], np.asarray(t)).reshape(1,-1)) for t in ts ]\n",
    "    best_tau = np.argmin(drf_vals)\n",
    "    best_taus_for_x_test[i] = ts[best_tau]\n",
    "    \n",
    "    if (i%100 == 0): \n",
    "        print i \n",
    "        plt.plot(ts, drf_vals)\n",
    "    \n",
    "print best_taus_for_x_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "14.5911949686\n",
      "43.2704402516\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\paint_000\\Anaconda\\lib\\site-packages\\ipykernel\\__main__.py:10: DeprecationWarning: using a boolean instead of an integer will result in an error in the future\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgUAAAFkCAYAAACw3EhvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nCKBQICIiIin/HyJIR1K4RSE6AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x21bd7f28>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgUAAAFkCAYAAACw3EhvAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niKT8f263ns7kk4byAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x21af5d30>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "plt.figure()\n",
    "i=100\n",
    "drf_vals = [clf.predict( np.append(X_w_dose[test_mask][i,0:d-1], np.asarray(t)).reshape(1,-1)) for t in ts ]\n",
    "plt.plot(ts,drf_vals)\n",
    "print ts[np.argmin(drf_vals)]\n",
    "plt.figure()\n",
    "\n",
    "i = 100 \n",
    "drf_vals = [clf.predict( np.append(X_w_dose[test_mask[i],0:d-1], np.asarray(t)).reshape(1,-1)) for t in ts ]\n",
    "plt.plot(ts,drf_vals)\n",
    "\n",
    "print ts[np.argmin(drf_vals)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Off policy evaluation regress \\& compare "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "ts = np.linspace(0,80,160)\n",
    "n_test = sum(test_mask)\n",
    "best_taus_for_x_test = np.zeros(n_test)\n",
    "for i in range(n_test): \n",
    "\n",
    "    drf_vals = [clf.predict( np.append(X_w_dose[test_mask,:][i,0:d-1], np.asarray(t)).reshape(1,-1)) for t in ts ]\n",
    "    best_tau = np.argmin(drf_vals)\n",
    "    best_taus_for_x_test[i] = ts[best_tau]\n",
    "    \n",
    "    if (i%100 == 0): \n",
    "        print i \n",
    "        plt.plot(ts, drf_vals)\n",
    "    \n",
    "print best_taus_for_x_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "print np.median(np.abs(sim_dose[test_mask] - therapeut_dose[test_mask]))\n",
    "print np.median(np.abs(np.dot(oracle_actual_oracle_res.x, test.T) - therapeut_dose[test_mask]))\n",
    "print np.median(np.abs(np.dot(oracle_res.x, test.T) - therapeut_dose[test_mask]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": false
   },
   "source": [
    "## boxplot of results "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[u'seaborn-darkgrid', u'seaborn-notebook', u'classic', u'seaborn-ticks', u'grayscale', u'bmh', u'seaborn-talk', u'dark_background', u'ggplot', u'fivethirtyeight', u'seaborn-colorblind', u'seaborn-deep', u'seaborn-whitegrid', u'seaborn-bright', u'seaborn-poster', u'seaborn-muted', u'seaborn-paper', u'seaborn-white', u'seaborn-pastel', u'seaborn-dark', u'seaborn-dark-palette']\n"
     ]
    }
   ],
   "source": [
    "print(plt.style.available)\n",
    "\n",
    "plt.style.use('seaborn-notebook')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1da9dcc0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAbUAAAGACAYAAAAqHfzDAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/j9zYX/tAff2FQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1d64c400>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "{u'bodies': [<matplotlib.collections.PolyCollection at 0x1d515a90>,\n",
       "  <matplotlib.collections.PolyCollection at 0x45a57b8>,\n",
       "  <matplotlib.collections.PolyCollection at 0x45a5668>,\n",
       "  <matplotlib.collections.PolyCollection at 0x45a5630>,\n",
       "  <matplotlib.collections.PolyCollection at 0x1d4be198>],\n",
       " u'cbars': <matplotlib.collections.LineCollection at 0x1d239b70>,\n",
       " u'cmaxes': <matplotlib.collections.LineCollection at 0x1d5155c0>,\n",
       " u'cmins': <matplotlib.collections.LineCollection at 0x1d4bee10>}"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhUAAAFkCAYAAACXcsmHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1d246668>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "plt.style.use('seaborn-white')\n",
    "plt.figure()\n",
    "\n",
    "\n",
    "# prescient_norm = np.abs(np.dot(prescient_beta, X[test_mask,:].T)  - therapeut_dose[test_mask] )\n",
    "# off_pol_norm = np.abs(np.dot(best_off_pol_beta, X[test_mask,:].T)  - therapeut_dose[test_mask] )\n",
    "# mean_dose_pol_norm = np.abs(np.mean(sim_dose[test_mask]) - therapeut_dose[test_mask] )\n",
    "# dm_norms = np.abs(np.dot(dm_res.x, X_w_dose[test_mask,:].T)  - therapeut_dose[test_mask] )\n",
    "# original = np.abs(sim_dose[test_mask] - therapeut_dose[test_mask])\n",
    "\n",
    "prescient_norm =(np.dot(prescient_beta, X[test_mask,:].T)  - therapeut_dose[test_mask] )\n",
    "off_pol_norm = (np.dot(best_off_pol_beta, X[test_mask,:].T)  - therapeut_dose[test_mask] )\n",
    "mean_dose_pol_norm = (np.mean(sim_dose[test_mask]) - therapeut_dose[test_mask] )\n",
    "dm_norms = (best_taus_for_x_test  - therapeut_dose[test_mask] )\n",
    "original = (sim_dose[test_mask] - therapeut_dose[test_mask])\n",
    "\n",
    "\n",
    "norms = [prescient_norm, off_pol_norm, dm_norms, mean_dose_pol_norm, original]\n",
    "\n",
    "plt.figure(figsize=(4.5,3))\n",
    "plt.ylabel('Differences btn. policy and therapeutic dose')\n",
    "flierprops = dict(linestyle='-',color='black')\n",
    "\n",
    "# prefer to turn off mean line\n",
    "bp_dict = plt.boxplot(norms,sym='', showmeans = False, meanline=False)\n",
    "\n",
    "for whisker in bp_dict['whiskers']:\n",
    "    whisker.set(color='#000000',linestyle='solid')\n",
    "for box in bp_dict['boxes']:\n",
    "    # change outline color\n",
    "    box.set( color='#000000')\n",
    "for ind,line in enumerate(bp_dict['medians']):\n",
    "    # get position data for median line\n",
    "    x, y = line.get_xydata()[1] # top of median line\n",
    "    if ind == 4: \n",
    "        plt.text(x-0.18, y+4, '%.1f' % y, horizontalalignment='center') # draw above, centered\n",
    "    # overlay median value\n",
    "    else:\n",
    "        plt.text(x+0.2, y+4, '%.1f' % y, horizontalalignment='center') # draw above, centered\n",
    "\n",
    "top = 60\n",
    "bottom = -60\n",
    "plt.ylim(bottom, top)\n",
    "# plt.title('Boxplot of distances between policy-recommended doses and therapeutic doses',y = 1.2)\n",
    "plt.axhline(y=np.median(mean_dose_pol_norm), xmin=0,xmax=1,color='g', ls='dashdot',linewidth=0.7)\n",
    "plt.axhline(y=0, xmin=0,xmax=1,color='g',alpha=0.3, ls='dashed',linewidth=0.7)\n",
    "plt.xticks([1, 2, 3,4,5], ['Best-In-Class', 'Continuous OPE', 'DM (Random Forest R&C)', 'Mean dose', 'Original doses'])\n",
    "plt.xticks(rotation=25)\n",
    "plt.show()\n",
    "\n",
    "plt.violinplot(norms)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([  0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         4.52237224e+01,   1.40693419e-01,   5.41198528e-02,\n",
       "         3.84589475e-02,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   2.22407643e+01,\n",
       "         0.00000000e+00,   3.51258428e+00,   0.00000000e+00,\n",
       "         7.20597022e+01,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   4.15588041e+01,\n",
       "         0.00000000e+00,   5.91847734e+01,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         8.92863627e+01,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   1.14990322e-01,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   0.00000000e+00,\n",
       "         0.00000000e+00,   0.00000000e+00,   2.30327152e+00])"
      ]
     },
     "execution_count": 95,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_off_pol_beta\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "medians\n",
      "-0.315987717471\n",
      "-2.1888248526\n",
      "5.10023816108\n",
      "-0.786163522013\n",
      "1.62006993519\n",
      "means\n",
      "-2.14812783534\n",
      "-4.13316674065\n",
      "0.492798060197\n",
      "-2.425105681\n",
      "0.492798060197\n",
      "sum of absolute values of differences\n",
      "7322.33032243\n",
      "10352.858741\n",
      "10398.3378676\n",
      "10835.3081761\n",
      "12059.3834326\n",
      "standard deviations \n",
      "15.8304480498\n",
      "19.8645306839\n",
      "19.2959552669\n",
      "20.3006901124\n",
      "21.3480834224\n"
     ]
    }
   ],
   "source": [
    "print \"medians\"\n",
    "print np.median(prescient_norm)\n",
    "print np.median(off_pol_norm )\n",
    "print np.median(mean_dose_pol_norm )\n",
    "print np.median(dm_norms)\n",
    "print np.median(original)\n",
    "\n",
    "print \"means\"\n",
    "print_f_(np.mean)\n",
    "\n",
    "def print_f_(func): \n",
    "    print func(prescient_norm)\n",
    "    print func(off_pol_norm )\n",
    "    print func(mean_dose_pol_norm )\n",
    "    print func(dm_norms)\n",
    "    print func(original)\n",
    "               \n",
    "               \n",
    "print \"sum of absolute values of differences\"\n",
    "print_f_(lambda x: sum(abs(x)))\n",
    "print \"standard deviations \"\n",
    "print_f_(np.std)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'off_pol_dose_test' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-54-4eee0397ffc8>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0moff_pol_dose_test\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mscatter\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtest_mask\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;36m9\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mbest_taus_for_x_test\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mcolor\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'r'\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0malpha\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m0.3\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mlabel\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'treatment dose from DM'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mscatter\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtest_mask\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;36m9\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdot\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mbest_off_pol_beta\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mX\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtest_mask\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mT\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mcolor\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'g'\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0malpha\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m0.3\u001b[0m \u001b[1;33m,\u001b[0m \u001b[0mlabel\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'treatment dose from OPE'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mscatter\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtest_mask\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;36m9\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mtherapeut_dose\u001b[0m\u001b[1;33m[\u001b[0m\u001b[0mtest_mask\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0mcolor\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'b'\u001b[0m\u001b[1;33m,\u001b[0m\u001b[0malpha\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m0.1\u001b[0m \u001b[1;33m,\u001b[0m \u001b[0mlabel\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'therapeutic dose'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mNameError\u001b[0m: name 'off_pol_dose_test' is not defined"
     ]
    }
   ],
   "source": [
    "off_pol_dose_test = np.dot(best_off_pol_beta, X[test_mask,:].T\n",
    "plt.scatter(X[test_mask][:,9], best_taus_for_x_test,color='r',alpha=0.3,label='treatment dose from DM')\n",
    "plt.scatter(X[test_mask][:,9], np.dot(best_off_pol_beta, X[test_mask,:].T),color='g',alpha=0.3 , label='treatment dose from OPE')\n",
    "plt.scatter(X[test_mask][:,9], therapeut_dose[test_mask],color='b',alpha=0.1 , label='therapeutic dose')\n",
    "\n",
    "\n",
    "plt.xlabel(\"BMI\")\n",
    "plt.ylabel('dose')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Doubly Robust estimator from DM, IPW "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from copy import deepcopy \n",
    "def DR_wrapper(beta, args): \n",
    "    data = dict(args)\n",
    "    INDS = data['inds']\n",
    "    t_lo = data['t_lo']; t_hi = data['t_hi']\n",
    "    clf = train_data['rf']\n",
    "    # train beta which gets best treatment policy for direct estimator \n",
    "    tau = np.clip(np.dot(beta, data['x'][INDS,:].T), t_lo, t_hi).flatten()\n",
    "    # concatenate 1st (d) colunns of full X \n",
    "    counterfactual_X = np.column_stack((data['x_full'][INDS,0:-1], tau))\n",
    "    reg_lambda = 3\n",
    "    data = dict(args)\n",
    "    # evaluate rhat(\\tau_i, x_i) \n",
    "    dm = np.mean( clf.predict( counterfactual_X ))\n",
    "    t_lo = data['t_lo']; t_hi = data['t_hi']\n",
    "    tau = np.clip(np.dot(beta, data['x'].T) , t_lo, t_hi).flatten()\n",
    "    data['tau'] = tau\n",
    "    \n",
    "    # evaluate rhat(t_i, x_i) on observed treatments\n",
    "    T_X = np.column_stack((data['x_full'][INDS,0:-1], sim_dose[INDS]))\n",
    "    old_y = data_copy['y_samp']\n",
    "    data['y_samp'] = data['y_samp'] - clf.predict( T_X )\n",
    "    ipw = off_policy_evaluation(**data)\n",
    "    data['y_samp'] = old_y \n",
    "    \n",
    "    return 100*(dm + ipw) \n",
    "\n",
    "d = X.shape[1]\n",
    "beta_d = np.random.rand(d)\n",
    "train_data['inds'] = trainind\n",
    "\n",
    "bounds = [(0, np.mean(therapeut_dose)/(np.sqrt(d)*np.mean(X_w_dose[:,i]))) for i in range(d) ]\n",
    "bnds = tuple(tuple(x) for x in bounds)\n",
    "\n",
    "oracle_DR_res = minimize(DR_wrapper, x0 = beta_d, bounds = bnds , options={'disp':True,'gtol':1e-3}, args=train_data.items())  \n",
    "pickle.dump(DR_wrapper.x, open('oracle_DR_beta', 'wb'))\n",
    "print oracle_DR_res"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.13"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}