EchoStateNetwork-Optimization_gh.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Create data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from numpy import *\n",
    "from matplotlib.pyplot import *\n",
    "import scipy.linalg\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def getData(size=5100, numChanges = 10):\n",
    "    #maxchanges = 50\n",
    "    #numChanges = random.randint(10, maxchanges)\n",
    "    #numChanges = 10\n",
    "    t = np.arange(0,size*0.8,0.8)\n",
    "    rlist = list()\n",
    "    while numChanges:\n",
    "        r = random.randint(1,size)\n",
    "        if r not in rlist:\n",
    "            rlist.append(r)\n",
    "        numChanges -= 1\n",
    "    rlist.sort()\n",
    "\n",
    "    for i in range(len(t)):\n",
    "        if i in rlist:\n",
    "            for j in range(i, len(t)):\n",
    "                t[j] = -t[j]\n",
    "\n",
    "    p=5\n",
    "    data = 1/2 * p * t % (2 * p)\n",
    "    \n",
    "    y1 = list()\n",
    "    y2 = list()\n",
    "    for i in range(len(t)):\n",
    "        if t[i] < 0:\n",
    "            y1.append(1)\n",
    "            y2.append(0)\n",
    "        else:\n",
    "            y1.append(0)\n",
    "            y2.append(1)\n",
    "            \n",
    "    return data, y1, y2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "data, y1, y2 = getData()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# ESN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def trainESN(data, y1, y2, initLen, trainLen, testLen, inSize, outSize, resSize, a, reg, spec, scale):\n",
    "    random.seed(42)\n",
    "    Win = random.rand(resSize,1+inSize) - scale\n",
    "    W = random.rand(resSize,resSize) - scale\n",
    "\n",
    "    # normalizing and setting spectral radius (correct, slow):\n",
    "    rhoW = max(abs(linalg.eig(W)[0]))\n",
    "    #print(\"Spectral Radius: \", rhoW)\n",
    "    W *= spec / rhoW\n",
    "\n",
    "    # allocated memory for the design (collected states) matrix\n",
    "    X = zeros((1+inSize+resSize,trainLen-initLen))\n",
    "    # set the corresponding target matrix directly\n",
    "    Yt = [y1[initLen+1:trainLen+1], y2[initLen+1:trainLen+1]]\n",
    "\n",
    "    # run the reservoir with the data and collect X\n",
    "    x = zeros((resSize,1))\n",
    "    for t in range(trainLen):\n",
    "        u = data[t]\n",
    "        x = (1-a)*x + a*tanh( dot( Win, vstack((1,u)) ) + dot( W, x ) )\n",
    "        if t >= initLen:\n",
    "            X[:,t-initLen] = vstack((1,u,x))[:,0]\n",
    "\n",
    "    # train the output by ridge regression\n",
    "    X_T = X.T\n",
    "    Wout = dot( dot(Yt,X_T), linalg.inv( dot(X,X_T) + \\\n",
    "        reg*eye(1+inSize+resSize) ) )\n",
    "\n",
    "    # average absolute size of learned output weights\n",
    "    avgout = sum(abs(Wout))/(len(Wout[0]) * outSize)\n",
    "    #print(\"Average size output weights: \", avgout)\n",
    "\n",
    "\n",
    "    # run the trained ESN with training data\n",
    "    Y = zeros((outSize,trainLen))\n",
    "    u = data[initLen] # starting point\n",
    "    for t in range(trainLen):\n",
    "        x = (1-a)*x + a*tanh( dot( Win, vstack((1,u)) ) + dot( W, x ) ) # update internal activations\n",
    "        y = dot( Wout, vstack((1,u,x)) ) # calculate outputs\n",
    "        # save the outputs\n",
    "        Y[:,t][0] = y[0]\n",
    "        Y[:,t][1] = y[1]\n",
    "        # get new input\n",
    "        u = data[initLen+t+1] \n",
    "\n",
    "\n",
    "\n",
    "    # compute RMSE for the first errorLen time steps\n",
    "    rmse1 = (sum( square( y1[initLen+1:initLen+trainLen+1] - \n",
    "        Y[0,0:trainLen] ) ) / trainLen)**(1/2.0)\n",
    "    #print('RMSE = \\t' + str( rmse1 ))\n",
    "    nrmse1 = rmse1 / (max(y1) - min(y1))\n",
    "    #print('train NRMSE = ', str(nrmse1) )\n",
    "\n",
    "    # compute RMSE for the first errorLen time steps\n",
    "    rmse2 = (sum( square( y2[initLen+1:initLen+trainLen+1] - \n",
    "        Y[1,0:trainLen] ) ) / trainLen) **(1/2.0)\n",
    "    #print('RMSE = \\t' + str( rmse2 ))\n",
    "    nrmse2 = rmse2 / (max(y2) - min(y2))\n",
    "    #print('NRMSE = ' + str( nrmse2 ))\n",
    "    \n",
    "    \n",
    "    #-------------------------------------------------------------------TEST-------------------------------------\n",
    "    # run the trained ESN with test data\n",
    "    Y = zeros((outSize,testLen))\n",
    "    u = data[trainLen] # starting point\n",
    "    for t in range(testLen):\n",
    "        x = (1-a)*x + a*tanh( dot( Win, vstack((1,u)) ) + dot( W, x ) ) # update internal activations\n",
    "        y = dot( Wout, vstack((1,u,x)) ) # calculate outputs\n",
    "        # save the outputs\n",
    "        Y[:,t][0] = y[0]\n",
    "        Y[:,t][1] = y[1]\n",
    "        # get new input\n",
    "        u = data[trainLen+t+1] \n",
    "    \n",
    "    # compute RMSE for the first errorLen time steps\n",
    "    errorLen = testLen\n",
    "    rmse3 = (sum( square( y1[trainLen+1:trainLen+errorLen+1] - \n",
    "        Y[0,0:errorLen] ) ) / errorLen)**(1/2.0)\n",
    "    #print('RMSE = \\t' + str( rmse3 ))\n",
    "    nrmse3 = rmse3 / (max(y1) - min(y1))\n",
    "    #print('test NRMSE  =  ' + str( nrmse3 ))\n",
    "    \n",
    "    # compute RMSE for the first errorLen time steps\n",
    "    rmse4 = (sum( square( y2[trainLen+1:trainLen+errorLen+1] - \n",
    "        Y[1,0:errorLen] ) ) / errorLen) **(1/2.0)\n",
    "    #print('RMSE = \\t' + str( rmse4 ))\n",
    "    nrmse4 = rmse4 / (max(y2) - min(y2))\n",
    "    #print('NRMSE = ' + str( nrmse4 ))\n",
    "    \n",
    "    return avgout, nrmse1, nrmse2, nrmse3, nrmse4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "initLen = 100\n",
    "trainLen = 2500\n",
    "testLen = 2500"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# generate ESN reservoir\n",
    "inSize = 1\n",
    "outSize = 2\n",
    "resSize = 50\n",
    "a = 0.9 # leaking rate\n",
    "reg = 1e-4  # regularization coefficient\n",
    "spec = 1.4\n",
    "scale = 0.5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train: \t mean: \t 0.05152847029652667 \t std: \t 0.003601219852924342\n",
      "Test:  \t mean: \t 0.03197361798171307 \t std: \t 0.01081963621958958\n"
     ]
    }
   ],
   "source": [
    "traineval = list()\n",
    "testeval = list()\n",
    "for times in range(10):\n",
    "    data, y1, y2 = getData(numChanges = 5)\n",
    "    avg, n1, n2, n3, n4 = trainESN(data, y1, y2, initLen, trainLen, testLen, inSize, outSize, resSize, a, reg, spec, scale)\n",
    "    traineval.append(n1)\n",
    "    testeval.append(n3)\n",
    "print(\"Train: \\t mean: \\t\", mean(traineval), \"\\t std: \\t\", std(traineval))\n",
    "print(\"Test:  \\t mean: \\t\", mean(testeval), \"\\t std: \\t\", std(testeval))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# generate ESN reservoir\n",
    "inSize = 1\n",
    "outSize = 2\n",
    "resSize = 300\n",
    "a = 0.9 # leaking rate\n",
    "reg = 1e-1  # regularization coefficient\n",
    "spec = 1\n",
    "scale = 0.5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train: \t mean: \t 0.04481090331247683 \t std: \t 0.0\n",
      "Test:  \t mean: \t 0.0025664933778220818 \t std: \t 0.0\n"
     ]
    }
   ],
   "source": [
    "traineval = list()\n",
    "testeval = list()\n",
    "for times in range(10):\n",
    "    data, y1, y2 = getData(numChanges = 5)\n",
    "    avg, n1, n2, n3, n4 = trainESN(data, y1, y2, initLen, trainLen, testLen, inSize, outSize, resSize, a, reg, spec, scale)\n",
    "    traineval.append(n1)\n",
    "    testeval.append(n3)\n",
    "print(\"Train: \\t mean: \\t\", mean(traineval), \"\\t std: \\t\", std(traineval))\n",
    "print(\"Test:  \\t mean: \\t\", mean(testeval), \"\\t std: \\t\", std(testeval))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Input Scaling "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# generate ESN reservoir\n",
    "inSize = 1\n",
    "outSize = 2\n",
    "resSize = 300\n",
    "a = 0.9 # leaking rate\n",
    "reg = 1e-2  # regularization coefficient\n",
    "spec = 1\n",
    "#scale = 0.7\n",
    "trainmean = list()\n",
    "testmean = list()\n",
    "avgoutmean = list()\n",
    "tries = [ 0, 0.1, 0.2, 0.3, 0.5, 0.7, 0.9, 1, 1.1, 1.2, 1.3, 1.4, 1.5]\n",
    "for r in tries:\n",
    "    traineval = list()\n",
    "    testeval = list()\n",
    "    avgout = list()\n",
    "    for times in range(10):\n",
    "        data, y1, y2 = getData(numChanges = 5)\n",
    "        avg, n1, n2, n3, n4 = trainESN(data, y1, y2, initLen, trainLen, testLen, inSize, outSize, resSize, a, reg, spec, r)\n",
    "        traineval.append(n1)\n",
    "        testeval.append(n3)\n",
    "        avgout.append(avg)\n",
    "    trainmean.append(mean(traineval))\n",
    "    testmean.append(mean(testeval))\n",
    "    avgoutmean.append(mean(avgout))\n",
    "plt.plot(tries, trainmean)\n",
    "plt.plot(tries, testmean)\n",
    "plt.legend(['train', 'test'])\n",
    "plt.show()\n",
    "\n",
    "plt.plot(tries, avgoutmean)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Spectral Radius"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# generate ESN reservoir\n",
    "inSize = 1\n",
    "outSize = 2\n",
    "resSize = 300\n",
    "a = 0.9 # leaking rate\n",
    "reg = 1e-2  # regularization coefficient\n",
    "#spec = 1.4\n",
    "scale = 0.5\n",
    "trainmean = list()\n",
    "testmean = list()\n",
    "avgoutmean = list()\n",
    "tries = [0.001, 0.05, 0.1, 0.5, 1, 5]\n",
    "for r in tries:\n",
    "    traineval = list()\n",
    "    testeval = list()\n",
    "    avgout = list()\n",
    "    for times in range(10):\n",
    "        data, y1, y2 = getData(numChanges = 5)\n",
    "        avg, n1, n2, n3, n4 = trainESN(data, y1, y2, initLen, trainLen, testLen, inSize, outSize, resSize, a, reg, r, scale)\n",
    "        traineval.append(n1)\n",
    "        testeval.append(n3)\n",
    "        avgout.append(avg)\n",
    "    trainmean.append(mean(traineval))\n",
    "    testmean.append(mean(testeval))\n",
    "    avgoutmean.append(mean(avgout))\n",
    "plt.plot(tries, trainmean)\n",
    "plt.plot(tries, testmean)\n",
    "plt.legend(['train', 'test'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD8CAYAAACMwORRAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4wLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvqOYd8AAAGPlJREFUeJzt3X+Q3HWd5/Hnq7vnBwkJIcwPMT8ISCDJsiI4RsqICxm0AmeBnu5J0D31wNSVP/fYOo+9vcI9rKu6de9O727h3CyyFN4uFAW7bk6zi0JAFAQzAUFDCIQQYAhrJgkEkpDJ/HjfH9OT9PT86M5M93ynv/16VE3N9Lc/6X63Fq/vuz/fHx9FBGZmli6ZpAswM7PKc7ibmaWQw93MLIUc7mZmKeRwNzNLIYe7mVkKOdzNzFLI4W5mlkIOdzOzFMol9cYtLS2xZMmSpN7ezKwmbdmyZW9EtJYal1i4L1myhK6urqTe3sysJkl6qZxxnpYxM0shh7uZWQo53M3MUsjhbmaWQg53M7MUcribmaWQw93MLIVqOtwP9vZz9+ZX8FKBZmYj1XS43/qznXz93qd5ce+hpEsxM5tRajbcBweDe5/oBuDw0YGEqzEzm1lqNtw379rPK/vfBqC33+FuZlaoZsP9ni3dx/4+0jeYYCVmZjNPTYb74aP9bPz1a5zbPgdw525mVqwmw/2+rf/MoaMDfPqixYA7dzOzYjUZ7vds6WbR/JNYdXYL4M7dzKxYzYX7q2+8zaMv7OMTFy7kpIYs4M7dzKxYzYV71679ZCU+ceFCmvPh3tvnzt3MrFDJcJd0m6Q9kn5TYtz7JA1I+mTlyhvtqvcsYPOfXMai+bNoyg2Vf6TfnbuZWaFyOvfbgTUTDZCUBf4MuK8CNZV06uxGgGPh3utpGTOzEUqGe0Q8DOwvMewrwL3AnkoUVa5cNkMuIx9QNTMrMuU5d0kLgI8D3y1j7DpJXZK6enp6pvrWADQ3ZH1A1cysSCUOqH4H+A8RUbJ9joj1EdERER2tra0VeOuhqRl37mZmI+Uq8BodwF2SAFqAKyT1R8QPKvDaJblzNzMbbcrhHhFnDv8t6Xbgh9MV7ODO3cxsLCXDXdKdwCVAi6Ru4BtAA0BElJxnr7Ymd+5mZqOUDPeIWFvui0XE56ZUzSS4czczG63mrlAt1tyQ8XnuZmZFaj7cm3JZd+5mZkVqPtybGzKeczczK1Lz4e7O3cxstJoPd3fuZmaj1Xy4u3M3Mxut5sPdnbuZ2Wg1H+7DnXtEJF2KmdmMUfPh3tyQYTCgb8DhbmY2rObDvSmXX2rP8+5mZsfUfLg3N+SX2vO8u5nZMTUf7u7czcxGq/1wd+duZjZK7Ye7O3czs1FqPtw9525mNlrNh7s7dzOz0Wo+3Ic7d9/T3czsuJLhLuk2SXsk/Wac5z8t6en8z6OSzq98meNz525mNlo5nfvtwJoJnn8R+L2IeDfwTWB9Beoqm+fczcxGK2cN1YclLZng+UcLHj4GLJx6WeVranDnbmZWrNJz7tcC/1jh15xQc86du5lZsZKde7kkXcpQuH9wgjHrgHUAixcvrsj7unM3MxutIp27pHcDtwJXRcS+8cZFxPqI6IiIjtbW1kq8tTt3M7MxTDncJS0G/g74g4h4buolnZhcNkM2I3fuZmYFSk7LSLoTuARokdQNfANoAIiI7wI3AqcBt0gC6I+IjmoVPJbmnFdjMjMrVM7ZMmtLPH8dcF3FKpqEpgavo2pmVqjmr1AFd+5mZsVSEe5DnbvD3cxsWDrCPZeht8/TMmZmw9IR7g1ZjrhzNzM7JhXh3uzO3cxshFSEuzt3M7ORUhHu7tzNzEZKRbj7bBkzs5HSEe7u3M3MRkhFuDc3ZDznbmZWIBXh3pTLunM3MyuQinB3525mNlIqwr0pl2VgMOgfcMCbmUFKwv3YItnu3s3MgJSEe1Muv9Se593NzICUhLs7dzOzkVIR7u7czcxGSkW4H+vcvWCHmRlQRrhLuk3SHkm/Ged5SfpfknZIelrShZUvc2LHOncvtWdmBpTXud8OrJng+cuBpfmfdcD/mXpZJ6bJnbuZ2Qglwz0iHgb2TzDkKuCOGPIYME/S6ZUqsBzu3M3MRqrEnPsC4JWCx935bdPGc+5mZiNVItw1xrYYc6C0TlKXpK6enp4KvPUQd+5mZiNVIty7gUUFjxcCu8caGBHrI6IjIjpaW1sr8NZDhjv3XnfuZmZAZcJ9A/Cv82fNXAQciIjXKvC6ZXPnbmY2Uq7UAEl3ApcALZK6gW8ADQAR8V1gI3AFsAM4DHy+WsWOx3PuZmYjlQz3iFhb4vkAvlSxiibBnbuZ2UipuEK1ISsycuduZjYsFeEuaWg1JnfuZmZASsId8qsxuXM3MwNSFO7u3M3MjktNuLtzNzM7LjXh7s7dzOy41IS7O3czs+NSE+7u3M3MjktPuLtzNzM7Jj3hnsvS6wWyzcyAFIV7c0PGC2SbmeWlJtzduZuZHZeacG9uyPiAqplZXmrCvSmX9QFVM7O81IS7O3czs+NSE+5NuSx9A8HA4JjLt5qZ1ZXUhPuxdVTdvZuZpSfcm3Jeas/MbFhZ4S5pjaTtknZIumGM5xdLelDSk5KelnRF5UudWHODl9ozMxtWMtwlZYGbgcuBFcBaSSuKhv0n4O6IuAC4Gril0oWW0uRFss3Mjimnc18J7IiInRFxFLgLuKpoTABz83+fAuyuXInlac4vkn3EV6mamZUV7guAVwoed+e3FfpT4DOSuoGNwFcqUt0JWDR/FgBbd7853W9tZjbjlBPuGmNb8fmGa4HbI2IhcAXwfUmjXlvSOkldkrp6enpOvNoJ/M4759I+t4lNz/62oq9rZlaLygn3bmBRweOFjJ52uRa4GyAifgE0Ay3FLxQR6yOiIyI6WltbJ1fxOCSxelkbDz+3l6O+x4yZ1blywn0zsFTSmZIaGTpguqFozMtAJ4Ck5QyFe2Vb8zJ0LmvnYG8/v3xx/3S/tZnZjFIy3COiH/gycB+wjaGzYrZKuknSlflhfwR8QdJTwJ3A5yJi2i8VXXV2C025DA94asbM6lyunEERsZGhA6WF224s+PsZYFVlSztxJzVmWXV2Cw9s28ONH12BNNbhAjOz9EvNFarDVi9r4+X9h3mh52DSpZiZJSaV4Q7wwLY9CVdiZpac1IX7O+edxIrT5zrczayupS7cATqXt9H10n7eOHw06VLMzBKRynBfvayNwYCHtk/72ZhmZjNCKsP9/IXzaDm5kQee9dSMmdWnVIZ7JiMuPbeNn27fQ9+Ar1Y1s/qTynAH6FzezptH+una9XrSpZiZTbvUhvsHl7bQmM34RmJmVpdSG+4nN+V4/1nzPe9uZnUpteEOcNnydnb2HOLFvYeSLsXMbFqlOtyPX63qqRkzqy+pDvdF82dxbvscX61qZnUn1eEOsHp5G5t37efA231Jl2JmNm1SH+6dy9roHwx+9ryvVjWz+pH6cL9g8amcOqvBUzNmVldSH+7Z/NWqD27fw8DgtC8OZWaWiNSHOwxdrfrG4T6eeNlXq5pZfSgr3CWtkbRd0g5JN4wz5l9JekbSVkl/W9kyp+bic1rIZeSpGTOrGyXDXVIWuBm4HFgBrJW0omjMUuCPgVUR8TvAH1ah1kmb29zAyjPn+1YEZlY3yuncVwI7ImJnRBwF7gKuKhrzBeDmiHgdICJmXIvcubyd5357kFf2H066FDOzqisn3BcArxQ87s5vK3QOcI6kRyQ9JmlNpQqslE5frWpmdaSccNcY24pPO8kBS4FLgLXArZLmjXohaZ2kLkldPT3Te975kpbZvKt1tm8kZmZ1oZxw7wYWFTxeCOweY8w/RERfRLwIbGco7EeIiPUR0RERHa2trZOtedI6l7fz+M79HOztn/b3NjObTuWE+2ZgqaQzJTUCVwMbisb8ALgUQFILQ9M0OytZaCWsXtbG0YFBfu6rVc0s5UqGe0T0A18G7gO2AXdHxFZJN0m6Mj/sPmCfpGeAB4F/HxH7qlX0ZHWccSpzm3Pc71MizSzlcuUMioiNwMaibTcW/B3A9fmfGSuXzXDJuW08+OweBgeDTGaswwlmZrWvLq5QLdS5vI19h47yVPcbSZdiZlY1dRfuv3dOK1lfrWpmKVd34T5vViPvPeNUnxJpZqlWd+EOcNnyNra99iavvvF20qWYmVVFXYb76mXtAGxy925mKVWX4f6u1tmccdosNvlWBGaWUnUZ7pLoXNbOIy/s4/BRX61qZulTl+EOQ6dEHu0f5JEdM+5aKzOzKavbcH/fkvnMacr5LpFmlkp1G+6NuQwfOqeVTfmrVc3M0qRuwx2GbiS2561etu5+M+lSzMwqqq7D/dJlbUhwv6dmzCxl6jrc589u5MLFp/p8dzNLnboOdxg6a+bXrx7gt28eSboUM7OKcbj7alUzS6G6D/dz2k9mwbyTfJdIM0uVug93SVy2vI2f7+jhSN9A0uWYmVVE3Yc7wOrl7RzpG+QXL/hqVTNLh7LCXdIaSdsl7ZB0wwTjPikpJHVUrsTqe/+Z85nVmOWBZ31KpJmlQ8lwl5QFbgYuB1YAayWtGGPcHOCrwOOVLrLamhuyXLy0hU3b9jC0HKyZWW0rp3NfCeyIiJ0RcRS4C7hqjHHfBL4F1OQ5hZ3L2tl94AjbXnsr6VLMzKasnHBfALxS8Lg7v+0YSRcAiyLihxWsbVpduqwNwDcSM7NUKCfcNca2Y3MXkjLAt4E/KvlC0jpJXZK6enp6yq9yGrTOaeL8RfO8tqqZpUI54d4NLCp4vBDYXfB4DnAe8JCkXcBFwIaxDqpGxPqI6IiIjtbW1slXXSWdy9p4qvsNet7qTboUM7MpKSfcNwNLJZ0pqRG4Gtgw/GREHIiIlohYEhFLgMeAKyOiqyoVV1Hn8jYi4MHt7t7NrLaVDPeI6Ae+DNwHbAPujoitkm6SdGW1C5xOK06fy+mnNLPJV6uaWY3LlTMoIjYCG4u23TjO2EumXlYyJLF6WRs/ePJVevsHaMplky7JzGxSfIVqkc7lbRw6OsDjO/cnXYqZ2aQ53It84F0tNDdkfJdIM6tpDvcizQ1ZPnh2C/dv+62vVjWzmuVwH8PqZe10v/42z+85mHQpZmaT4nAfw+pjV6t6asbMapPDfQzvOKWZ8xbM9a0IzKxmOdzHsXpZO0+8/Dr7Dx1NuhQzsxPmcB/HZcvbGAx4yFermlkNcriP47x3nkLrnCbfSMzMapLDfRyZjFh9bhsPb++hb2Aw6XLMzE6Iw30CncvbeKu3n80v+mpVM6stDvcJfHBpC425jKdmzKzmONwnMKsxxwfedRoP+GpVM6sxDvcSOpe1sWvfYXbuPZR0KWZmZXO4lzC8tqrv8W5mtcThXsLCU2ex7B1zuN9Xq5pZDXG4l6FzeRtdL73OgcN9SZdiZlYWh3sZOpe3MzAY/PT5nqRLMTMrS1nhLmmNpO2Sdki6YYznr5f0jKSnJT0g6YzKl5qc8xfO47TZjb6RmJnVjJLhLikL3AxcDqwA1kpaUTTsSaAjIt4N3AN8q9KFJimbEZec28ZD23vo99WqZlYDyuncVwI7ImJnRBwF7gKuKhwQEQ9GxOH8w8eAhZUtM3mXLW/jwNt9bHnp9aRLMTMrqZxwXwC8UvC4O79tPNcC/ziVomai4atV/+Pf/5pNz/qiJjOb2coJd42xbcxkk/QZoAP483GeXyepS1JXT09tHZyc09zAX37mvQwG/Jvbu7h6/WM8+bK7eDObmcoJ925gUcHjhcDu4kGSLgP+BLgyInrHeqGIWB8RHRHR0draOpl6E3XpsjZ+/O8+xDc/dh4v9Bzi47c8yhf/Zgs7e7zWqpnNLCo1vSApBzwHdAKvApuBayJia8GYCxg6kLomIp4v5407Ojqiq6trsnUn7lBvP3/1s52sf3gnvf2DrF25iK92LqVtTnPSpZlZiknaEhEdJceVM3cs6QrgO0AWuC0i/oukm4CuiNgg6X7gd4HX8v/k5Yi4cqLXrPVwH9bzVi//e9Pz/O3jL9OYy3DdxWex7kNncXJTLunSzCyFKhru1ZCWcB/24t5D/Lcfb+dHT7/GabMb+WrnUtauXExjzteJmVnllBvuTp4KObNlNjdfcyH/8KVVLG0/mW9s2MqHv/1T/t9Tuxkc9Jk1Zja9HO4Vdv6iedz5hYv468+/j5Masnzlzif52C2P8OgLe5MuzczqiMO9CiRx6blt/OirF/Pff/989r7VyzV/9Tifve2XbHvtzaTLM7M64Dn3aXCkb4A7frGLmx98gTeP9PHx9yzg+o+cw8JTZyVdmpnVGB9QnYEOHO7jlp/u4K8f2QUBn/3AGXzxkrM5dXZj0qWZWY1wuM9gu994m2//5DnueaKbk5tyfPGSs/n8qiU0N2STLs3MZjiHew3Y/s9v8a1/epYHnt3DO+Y2c/2Hz+ET711INjPWHR/MzHwqZE049x1z+N7n3sdd6y6i/ZRmvn7v06z5zsPc/4xvTGZmU+NwnwEuOus0fvDFD3DLpy+kfzC47o4uPvWXj/GEb0xmZpPkcJ8hJHHF755+7MZkO/ce4l/e8ij/9vtbeME3JjOzE+Q59xnqUG8/t/7sRdY//AJH+gf5yIp25s9upCmXpakhQ3P+d1MuM7Qtl6G5Yej30PaJt3le36w2lTvn7rtbzVCzm3J87bKlXPP+xfzFpue5f9sejvQN0Ns/SG//AH0DU9sp5zLKh34+/At3BIU7jsJt+Z3E8R3LGP92rNcr2pbLCMk7F7Nqcudeo/oHBjk6MEhv3+CxwD/SN/S7t394e/G2od8jtvUP0Ns3yJGC5wtf40j++cLtR/unto5sRozcgeSyNDeM3IGM2pbfSTQX7SyaCnc0JV5v+N94x2K1zJ17yuWyGXLZDLMSuP5pcDAKdizDO4yRO4tR28bd4YzedqRvkANv9x3fVrQDm6rCncXIbyzHdxJTnfYa9Q0o/1zG02E2TRzudsIyGdGcyeYvumqY1veOyO9Y+sfZuYy5wyn4VlK47di3lpE7kEOH+sd9vane4LMxmxm5Myjz20nzCXw78XEWA4e71RhJ+cDKQgKLXvUNDI7aSRwZY8cx2W8yB97uY884O6JKH2cZ6xvLWN82Jv52Ms7r+ThL4hzuZiegIZuhIZtJZKWtco+zTHaHc7C3n30Hj07LcZbmKRyobxpn2mu84zaN2fo8zuJwN6sRM+04y4idywkd2B//OMtY02SVPs4y6kD9CUx7lZwKK9qBNWaTO85SVrhLWgP8T4bWUL01Iv5r0fNNwB3Ae4F9wKciYldlSzWzpNT0cZbiM8HG+Nay71B/0Y6lusdZrlm5mOsuPqsy/wONo2S4S8oCNwMfBrqBzZI2RMQzBcOuBV6PiLMlXQ38GfCpahRsZvVlJh5nOfbNY6xtZXyTaTm5qep1l9O5rwR2RMROAEl3AVcBheF+FfCn+b/vAf5CksJ3vzKzGpfkcZapKOfeMguAVwoed+e3jTkmIvqBA8BpxS8kaZ2kLkldPT09k6vYzMxKKifcxzoaUNyRlzOGiFgfER0R0dHa2lpOfWZmNgnlhHs3sKjg8UJg93hjJOWAU4D9lSjQzMxOXDnhvhlYKulMSY3A1cCGojEbgM/m//4ksMnz7WZmySl5hCAi+iV9GbiPoVMhb4uIrZJuAroiYgPwPeD7knYw1LFfXc2izcxsYmUd/o2IjcDGom03Fvx9BPj9ypZmZmaT5ZWYzMxSyOFuZpZCiS3WIakHeGmS/7wF2FvBcmqBP3N98GeuD1P5zGdERMlzyRML96mQ1FXOSiRp4s9cH/yZ68N0fGZPy5iZpZDD3cwshWo13NcnXUAC/Jnrgz9zfaj6Z67JOXczM5tYrXbuZmY2gZoLd0lrJG2XtEPSDUnXU22SbpO0R9Jvkq5lukhaJOlBSdskbZX0taRrqjZJzZJ+Kemp/Gf+z0nXNB0kZSU9KemHSdcyHSTtkvRrSb+S1FXV96qlaZn8qlDPUbAqFLC2aFWoVJH0IeAgcEdEnJd0PdNB0unA6RHxhKQ5wBbgYyn//1nA7Ig4KKkB+DnwtYh4LOHSqkrS9UAHMDciPpp0PdUmaRfQERFVP6+/1jr3Y6tCRcRRYHhVqNSKiIeps9snR8RrEfFE/u+3gG2MXiAmVWLIwfzDhvxP7XRekyBpIfAvgFuTriWNai3cy1kVylJE0hLgAuDxZCupvvwUxa+APcBPIiLtn/k7wNeBwaQLmUYB/FjSFknrqvlGtRbuZa34ZOkg6WTgXuAPI+LNpOuptogYiIj3MLQgzkpJqZ2Gk/RRYE9EbEm6lmm2KiIuBC4HvpSfdq2KWgv3claFshTIzzvfC/xNRPxd0vVMp4h4A3gIWJNwKdW0CrgyPwd9F7Ba0v9NtqTqi4jd+d97gL9naKq5Kmot3MtZFcpqXP7g4veAbRHxP5KuZzpIapU0L//3ScBlwLPJVlU9EfHHEbEwIpYw9N/xpoj4TMJlVZWk2fkTBJA0G/gIULWz4Goq3COiHxheFWobcHdEbE22quqSdCfwC+BcSd2Srk26pmmwCvgDhrq5X+V/rki6qCo7HXhQ0tMMNTE/iYi6OD2wjrQDP5f0FPBL4EcR8U/VerOaOhXSzMzKU1Odu5mZlcfhbmaWQg53M7MUcribmaWQw93MLIUc7mZmKeRwNzNLIYe7mVkK/X8jCXjqjeVmEwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(tries, avgoutmean)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Leaking rate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# generate ESN reservoir\n",
    "inSize = 1\n",
    "outSize = 2\n",
    "resSize = 300\n",
    "#a = 0.9 # leaking rate\n",
    "reg = 1e-2  # regularization coefficient\n",
    "spec = 1\n",
    "scale = 0.5\n",
    "trainmean = list()\n",
    "testmean = list()\n",
    "avgoutmean = list()\n",
    "tries = [1, 0.9, 0.8, 0.7, 0.6, 0.5, 0.4, 0.3, 0.2]\n",
    "for r in tries:\n",
    "    traineval = list()\n",
    "    testeval = list()\n",
    "    avgout = list()\n",
    "    for times in range(10):\n",
    "        data, y1, y2 = getData(numChanges = 5)\n",
    "        avg, n1, n2, n3, n4 = trainESN(data, y1, y2, initLen, trainLen, testLen, inSize, outSize, resSize, r, reg, spec, scale)\n",
    "        traineval.append(n1)\n",
    "        testeval.append(n3)\n",
    "        avgout.append(avg)\n",
    "    trainmean.append(mean(traineval))\n",
    "    testmean.append(mean(testeval))\n",
    "    avgoutmean.append(mean(avgout))\n",
    "plt.plot(tries, trainmean)\n",
    "plt.plot(tries, testmean)\n",
    "plt.legend(['train', 'test'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(tries, avgoutmean)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Reservoir Size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# generate ESN reservoir\n",
    "inSize = 1\n",
    "outSize = 2\n",
    "#resSize = 50\n",
    "a = 0.9 # leaking rate\n",
    "reg = 1e-2  # regularization coefficient\n",
    "spec = 1\n",
    "scale = 0.5\n",
    "trainmean = list()\n",
    "testmean = list()\n",
    "avgoutmean = list()\n",
    "tries = [1, 5, 10, 50, 100, 300, 500, 700, 900]\n",
    "for r in tries:\n",
    "    traineval = list()\n",
    "    testeval = list()\n",
    "    avgout = list()\n",
    "    for times in range(10):\n",
    "        data, y1, y2 = getData(numChanges = 5)\n",
    "        avg, n1, n2, n3, n4 = trainESN(data, y1, y2, initLen, trainLen, testLen, inSize, outSize, r, a, reg, spec, scale)\n",
    "        traineval.append(n1)\n",
    "        testeval.append(n3)\n",
    "        avgout.append(avg)\n",
    "    trainmean.append(mean(traineval))\n",
    "    testmean.append(mean(testeval))\n",
    "    avgoutmean.append(mean(avgout))\n",
    "plt.plot(tries, trainmean)\n",
    "plt.plot(tries, testmean)\n",
    "plt.legend(['train', 'test'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(tries, avgoutmean)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Ridge regression regularization parameter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAX0AAAEACAYAAABfxaZOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4wLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvqOYd8AAAIABJREFUeJzt3Xt8VNW99/HPyp2EJECAIARIVLyAIBBEEFEUL2Drpa03eqyeHnqobbF9ntaqnCoerT1qz3nUYm2tWltbrxxbK61R1EpaJIgkeIFwEcQMCcg9CSQk5LaeP/YEQwjJJHPZeybf9+s1r8zsWXvP72fwNztrr72WsdYiIiK9Q5zbAYiISOSo6IuI9CIq+iIivYiKvohIL6KiLyLSi6joi4j0Iir6IiK9iIq+iEgvoqIvItKLqOiLiPQiCW4H0N7AgQNtbm6u22F0qra2lrS0NLfDCIlYySVW8gDl4lVez6WkpGSvtXZQV+08V/Rzc3MpLi52O4xOFRYWMmPGDLfDCIlYySVW8gDl4lVez8UY4wuknbp3RER6ERV9EZFeREVfRKQX8VyffkcaGxupqKigvr7e7VAAyMzMZMOGDWE5dkpKCjk5OSQmJobl+CLSu0VF0a+oqCA9PZ3c3FyMMW6Hw8GDB0lPTw/5ca217Nu3j4qKCvLy8kJ+fBGRqOjeqa+vJysryxMFP5yMMWRlZXnmLxoRiT1RUfSBmC/4rXpLniJytH9+sodNOw+G/XOipui7raqqil/96lfd3u+yyy6jqqoqDBGJSCxZ8Oe1LHpnc9g/R0U/QMcr+s3NzZ3uV1BQQL9+/cIVlojEgM+r69heVUf+iP5h/6youJDrBXfccQeffvop48ePJy4ujszMTE444QQ+/PBD1q9fz1VXXUV5eTn19fX84Ac/YN68ecAXdxjX1NQwe/Zszj33XIqKihg2bBivvvoqffr0cTkzEXFbia8SgEm5KvrHuOevpazfcSCkxxw9NIO7Lx/TaZsHHniAdevW8eGHH1JQUMA111zDunXrjoyyefrppxkwYAB1dXWcddZZfO1rXyMrK+uoY2zevJkXXniBJ598kmuvvZY//elP3HDDDSHNRUSiT3FZJX0S4zn9hIywf1bUFX2vmDx58lHDKhctWsQrr7wCQHl5OZs3bz6m6Ofl5TF+/HgA8vPzKSsri1i8IuJdJb5Kxg/vR2J8+Hvco67od3VGHiltZ9srLCzk7bffZuXKlaSmpjJjxowOh10mJycfeR4fH09dXV1EYhUR76o93MT6zw/wnfNPisjn6UJugNLT0zl4sOPhVNXV1fTv35/U1FQ2btzIe++9F+HoRCRafVRRRXOLJT8C/fkQhWf6bsnKymLatGmcccYZJCUlMXTo0CPvzZo1i8cff5xx48Zx6qmnMmXKFBcjFZFoUlLmXMSdGIGRO6Ci3y3PP/88cOw0DMnJybz++usd7tPabz9w4EDWrVt3ZPutt94avkBFJGoU+yo5JbsvmX0iM9+WundERFzS0mJZs62S/JEDIvaZKvoiIi7ZvLuGg/VNTBoZma4dUNEXEXFNsW8/EJmbslqp6IuIuKSkrJKBfZMYMSA1Yp+poi8i4pKSbZXkj+wf0dl1VfRFRFyw5+BhfPsOMSmCF3FBRT9gPZ1aGeCRRx7h0KFDIY5IRKJZib8/P1I3ZbVS0Q+Qir6IhFJxWSVJCXGMGRr+Sdba0s1ZAWo7tfL5559PTk4Oixcv5vDhw3zlK1/hnnvuoba2lmuvvZaKigqam5u566672LVrFzt27OCCCy5g4MCBLFu2zO1URMQDSrZVcmZOJskJ8RH93ICKvjFmFvALIB54ylr7QLv3zwMeAcYB11trX27z3k3Anf6X91lrnwkq4tfvgJ1rgzrEMYaMhdkPdNqk7dTKf/nLXygoKOD999/HWssVV1zBP//5T/bs2cPQoUN57bXXAGdOnszMTB566CGWLVvGwIEDQxu3iESl+sZm1m2vZu65J0b8s7vs3jHGxAOPAbOB0cAcY8zods22Af8KPN9u3wHA3cDZwGTgbmNMZDuwwuCdd97hzTffZMKECUycOJGNGzeyefNmxo4dy9tvv83tt9/O8uXLyczMdDtUEfGgjyuqaWy25EfwpqxWgZzpTwa2WGu3AhhjXgSuBNa3NrDWlvnfa2m376XAW9ba/f733wJmAS/0OOIuzsgjwVrLggUL+Pa3v33MeyUlJRQUFLBgwQIuueQSFi5c6EKEIuJlrTdluVH0A7mQOwwob/O6wr8tEMHs6yltp1aeOXMmTz/9NDU1NQBs376d3bt3s2PHDlJTU7nhhhu49dZbWbNmzTH7iois8VVy4qA0BqQlRfyzAznT7+iuARvg8QPa1xgzD5gHkJ2dTWFh4VHvZ2Zmul40k5KSmDx5MqNHj2bmzJl89atf5eyzzwacBVWefPJJtm7dyl133UVcXBwJCQk8/PDDHDx4kBtvvJFLL72UIUOGHOnv70x9ff0x/w3CpaamJmKfFU6xkgcoF68KVS7WWt7bcogJgxPc+W9jre30AUwFlrZ5vQBYcJy2vweubvN6DvCbNq9/A8zp7PPy8/Nte+vXrz9mm5sOHDgQ1uNHMt9ly5ZF7LPCKVbysFa5eFWoctm866Adefvf7Ivv+0JyvFZAse2inltrA+reWQ2MMsbkGWOSgOuBJQF+pywFLjHG9PdfwL3Ev01EpFda43MWTYnkdMptdVn0rbVNwHycYr0BWGytLTXG3GuMuQLAGHOWMaYCuAb4jTGm1L/vfuCnOF8cq4F7/dtERHqlYt9++qcmctKgtK4bh0FA4/SttQVAQbttC9s8Xw3kHGffp4Gng4hRRCRmFPsiP8laW1EzDYPTZRX7ekueIr3R/toGtu6pZaILQzVbRUXRT0lJYd++fTFfEK217Nu3j5SUFLdDEZEwaO3Pj/TMmm1Fxdw7OTk5VFRUsGfPHrdDAZwhleEqzCkpKeTkdNhTJiJRrthXSWK8YVyOe3frR0XRT0xMJC8vz+0wjigsLGTChAluhyEiUabEt58xQzNJSYzsJGttRUX3johItDvc1MxHFdURXQS9Iyr6IiIRULrjAA1NLRFdBL0jKvoiIhFQUuZcxHVz5A6o6IuIRESxbz8jBqQyON3d0Xkq+iIiYWatpcRX6Xp/Pqjoi4iE3bb9h9hb0xDxRdA7oqIvIhJmxWXu35TVSkVfRCTMin2VpKckMGpwX7dDUdEXEQm3Et9+Jo7oT1ycO5OstaWiLyISRtV1jXyyq8YTF3FBRV9EJKzWbGtdNEVFX0Qk5pWUVRIfZxg/op/boQAq+iIiYVXiq2T0CRmkJnljfksVfRGRMGlsbuHD8irPdO2Air6ISNhs+PwAdY3NKvoiIr3BkZuyPHAnbisVfRGRMCnZVsmwfn04IbOP26EcoaIvIhIG1lpKyio91bUDKvoiImGxvaqOnQfqVfRFRHqDEp+3bspqpaIvIhIGJb5K0pLiOW1IutuhHEVFX0QkDIrLKhk/oh8J8d4qs96KRkQkBtQcbmLjzgPke2D+/PYCKvrGmFnGmE3GmC3GmDs6eD/ZGPOS//1Vxphc//ZEY8wzxpi1xpgNxpgFoQ1fRMR7PthWSYvFMzNrttVl0TfGxAOPAbOB0cAcY8zods3mApXW2pOBh4EH/duvAZKttWOBfODbrV8IIiKxqsRXiTEwwSOTrLUVyJn+ZGCLtXartbYBeBG4sl2bK4Fn/M9fBmYaYwxggTRjTALQB2gADoQkchERjyrxVXJqdjrpKYluh3KMQIr+MKC8zesK/7YO21hrm4BqIAvnC6AW+BzYBvyPtXZ/kDGLiHhWc4vlg21Vnpp6oa1A5vrsaH0vG2CbyUAzMBToDyw3xrxtrd161M7GzAPmAWRnZ1NYWBhAWO6pqanxfIyBipVcYiUPUC5eFWgu2w40U3O4idRDuygs3Bf+wLopkKJfAQxv8zoH2HGcNhX+rpxMYD/wdeANa20jsNsYswKYBBxV9K21TwBPAEyaNMnOmDGj+5lEUGFhIV6PMVCxkkus5AHKxasCzeWPK8uAUr4x6xyGD0gNc1TdF0j3zmpglDEmzxiTBFwPLGnXZglwk//51cA71lqL06VzoXGkAVOAjaEJXUTEe4p9lQxOTyanv3cmWWury6Lv76OfDywFNgCLrbWlxph7jTFX+Jv9FsgyxmwBfgi0Dut8DOgLrMP58vidtfbjEOcgIuIZxWWVTMrtjzOWxXsCWr/LWlsAFLTbtrDN83qc4Znt96vpaLuISCzaWV3P9qo6/u3cPLdDOS7dkSsiEiJenWStLRV9EZEQKfbtJyUxjjFDM9wO5bhU9EVEQqTEV8mZOf1I9Ngka215NzIRkShyqKGJ0h0HPHtTVisVfRGREPiovJrmFuvp/nxQ0RcRCYkSnzPDzMQRKvoiIjGv2FfJqMF96Zea5HYonVLRFxEJUkuLZY2v0vP9+aCiLyIStC17ajhQ3+T5rh1Q0RcRCVpxmXNT1qRc7y2P2J6KvohIkEp8lWSlJZGb5b1ZNdtT0RcRCVKJbz8TR3p3krW2VPRFRIKw5+BhyvYd8uQi6B1R0RcRCULrJGvRMHIHVPRFRIKyZlslSfFxnDEs0+1QAqKiLyIShOKy/YzNySQ5Id7tUAKioi8i0kP1jc2s234gavrzQUVfRKTH1m6vpqG5JTSTrL33a1jffvnx0FPRFxHpoZCtlNXSAv94EDYvDUFUnYupor+/tsHtEESkFykuqyRvYBpZfZODO9CejVBXCSOnhSawTsRM0f+ovIppD7zDU8u3Yq11OxwRiXHWWtZsqwxN145vhfNz5DnBH6sLMVP0c7PSOHfUQO57bQM3P1tCdV2j2yGJSAzbureW/bUNobmI6yuCjGHQb2Twx+pCzBT9zNREnvhGPnd+6XT+vmE3X350OR9XVLkdlojEqJD151vrFP2R50AEpnGImaIPYIzhW9NPZPHNU2lutlz965X8YWWZuntEJORKyirJ7JPISYP6Bneg/VuhZmdEunYgxop+q4kj+vPa96dz7qiBLHy1lPnPf8DBenX3iEjoFPv2kz+yP3FxQZ6d+4qcnxG4iAsxWvQB+qcl8dSNk7hj9mm8UbqTyx99l9Id1W6HJSIxoLK2gU/31IboIm4RpGbBwFOCP1YAYrboA8TFGW4+/yRenDeFusZmvvKrIp5b5VN3j4gEZc22EPXngzNyJ0L9+RBg0TfGzDLGbDLGbDHG3NHB+8nGmJf8768yxuS2eW+cMWalMabUGLPWGJMSuvADc1buAAq+P52z8wbwk1fW8X9e+pDaw02RDkNEYkSxr5KEOMOZOf2CO1B1BVT5Ita1AwEUfWNMPPAYMBsYDcwxxoxu12wuUGmtPRl4GHjQv28C8Cxws7V2DDADcKVzPatvMs98czI/uvgU/vrRDi7/5bts3HnAjVBEJMqVlFUyZlgmfZKCnGTNt9L5GaGLuBDYmf5kYIu1dqu1tgF4EbiyXZsrgWf8z18GZhpnCZlLgI+ttR8BWGv3WWubQxN698XFGW6ZOYpnv3U2B+ubuOqxFSwuLncrHBGJQg1NLXxUURWi8fkrIDkDss8I/lgBCqToDwPaVsYK/7YO21hrm4BqIAs4BbDGmKXGmDXGmNuCDzl455w0kILvT2fiiP7c9vLH/GjxRxxqUHePiHStdEc1h5tCNMmarwhGTIG4yE3LnBBAm46uLrS/Enq8NgnAucBZwCHg78aYEmvt34/a2Zh5wDyA7OxsCgsLAwgreN862TKIRP68poKVn2xn/vgUhvbt+nuwpqYmYjGGW6zkEit5gHLxqtZc3vjM6aFu2L6Bwn2beny8xIYqpu3dxKcZZ1Meyf9G1tpOH8BUYGmb1wuABe3aLAWm+p8nAHtxvgiuB37fpt1dwI87+7z8/Hwbaf/8ZLedeO+b9rQ7X7d/XlPeZftly5aFP6gIiZVcYiUPa5WLV7Xm8u0/FNtzH/x78AcsfdXauzOs3bYq+GNZa4Fi20U9t9YG1L2zGhhljMkzxiT5C3n7SZ+XADf5n18NvOMPYikwzhiT6r+oez6wvgffTWE1fdQgCn4wnbE5mfzflz7ijj99TH2ja5ceRMSjrLWUbKskf0SIunYS+sAJ44M/Vjd0WfSt00c/H6eAbwAWW2tLjTH3GmOu8Df7LZBljNkC/BC4w79vJfAQzhfHh8Aaa+1roU8jeNkZKTz/rbP57oyTeHF1OVc9toKte2rcDktEPKR8fx17Dh4mP3dA8AfzrYDhZ0FCUvDH6oZA+vSx1hYABe22LWzzvB645jj7PoszbNPzEuLjuG3WaZyVN4AfvvQhlz/6Lg98bRyXnznU7dBExAOKffsBgh+5U1cFO9fCjGNuewq7mL4jt6cuOHUwr31/OqedkMEtL3zAnX9Zq+4eEaHYV0l6cgKnZKcHd6DyVYCN6Pj8Vir6xzG0Xx9enDeFeeedyLPvbePqx4vw7at1OywRcdEaXyXjR/QjPuhJ1lZAXCLknBWawLpBRb8TifFx/Mdlp/PUjZMo31/Hlxe9y+trP3c7LBFxQW2jZdOug0waGYr+/CIYlg+JfYI/Vjep6AfgotHZ/O2WczlxcF++89wanttwmIamFrfDEpEI+rSqGWthUm6Q/fkNtbDjA1e6dkBFP2DDB6Tyv9+eyr9Ny+MtXxPXPF5E+f5DboclIhGyuaqFOAPjhwc5yVrFamhpiugka22p6HdDUkIcCy8fzfzxyWzdW8uXFi3nrfW73A5LRCJgS2Uzp5+QQVpyQIMej89XBCYOhk8OTWDdpKLfA5OGJPDaLdMZkZXKv/+hmJ+9tp7GZnX3iMSqpuYWtla3hG4R9CHjICUj+GP1gIp+D43ISuXlm8/hG1NG8uTyz7juNyvZUVXndlgiEgYbPj/I4WaCvymr6bDTveNS1w6o6AclJTGen151Bo/OmcAnu2r40qLlLNu02+2wRCTESvw3ZQU9s+aOD6Cp3rWLuKCiHxKXnzmUJfOnkZ2Rwjd/t5oH39hIk7p7RGJGsa+SASmGYf2CHGLpW+H8HDE1+KB6SEU/RE4c1Je/fG8acyYP59eFn/L1p1ax60C922GJSAiU+Co5uV8IyqWvCAadDmlZwR+rh2Kr6De4O4QyJTGe+786joevO5O1FdVc9ovlLN+8x9WYRKTnyvbW8tiyLXxeXc+o/kEudNLcBNtWudq1AwFOuBYVyt+H56+Df/lfyJnkaihfmZDD2GGZfPe5Ndz49PvccuEofjBzVPC3botIWFlrKd1xgKWlO1laupNPdjkz7Y4f3o+zhhwO7uC71kLDQRX9kBl8OsQnQcGP4Vt/hzh3/4g5eXA6f/neNBa+Wsqiv2+muGw/j1w/nsHpKa7GJSJHa2puYXVZJUtLd/LW+l1sr6ojzsBZuQNY+OXRXDImm5z+qcGvAOYrcn6q6IdIcjpcfC+8Mg8+fA4mfsPtiEhNSuB/rjmTs/MGcNer67jsF++yaM54zjlpoNuhifRq9Y3NvLt5L0tLd/L2hl1UHmokKSGO80YN5AcXjWLmaYPJ6psc2g/1FUH/PMhwd6r22Cn6AOOuheLfwtv/CadfDn2CvF06RK6ZNJxxOf347nMl3PDUKv7PRacw/4KTiVN3j0jEVNc1smzjbpaW7uQfn+zhUEMz6SkJXHjaYC4dM4TzTxkU/N22x9PS4hT9Uy8Lz/G7IbaKvjEw++fwxAz4x4Mw6363Izri1CHpLJl/Lj95ZS0PvfUJq8v288h140N/NiEiR+w6UM+b63fxZulOVn66j6YWy+D0ZL4yYRiXjhnClBOzSEqIQFfw3k1Qt9/1rh2ItaIPMHQ85P8rrPoNTLzR6ev3iLTkBB6+bjxnn5jF3UtKuWzRch6dM5HJeSGYqlVEANi6p4Y31+9iaelOPthWBUDewDTmTs/j0jFDGJ/TL/J/ZbeOz1fRD5ML74LSV+D12+HGV52/ADzCGMOcySMYl5PJ955bw5wn3+PWS07l2+edqO4ekR6w1rJu+xcjbjbvdkbcjB2WyY8uPoVLzxjCqMF9MW7WAV8RpA+F/rnuxeAXm0U/LQsuvBMKboUNS2D0lW5HdIwxQzP56y3ncsef1/LgGxt5/7N9PHTtePqnRXaRZBGvs9Zy8HATu6rr2XXgMDsP1LPrQD27D9Sz80A967YfODLiZnLeAL5+9mguGTMk+LtnQ8Vap+iPnOaJE9DYLPoA+d+E4t/B0p/AyRdDUqrbER0jPSWRX86ZwJS8Afz0bxv40qLlPPr1icHP7yESJeobm9l94DC7Dtazs9pfzA8ePvLceRymroM1qjNSEsjOSGHM0Ax+cNEoLjo9mwFePGmq/AwOfu6Jrh2I5aIfnwCX/Rx+/yVY8Qu4YIHbEXXIGMM3puYyfnh/vvt8Cdf9ZiV3zD6NuefmufvnqEgAWlosTS2W5hZLU0sLTc1fvG5sbuGz6mYa1+9iZ+uZeXU9uw4ePnKWXnWo8ZhjJifEkZ2RwpCMFM4YlsnM053ngzOSGZKRQrb/eWpSlJSvstb+fPdm1mwrSv6r9VDuuTDmq7DiERj/deg/0u2IjmtsTiZ/u2U6t738Efe9toGCtZ9H5EauPXvrebG8JGzHt9iwHbutvXvreaG8OKyfYQNMpTsZd3TMffvqedbXPpfAjtqdGJ3i3EJjc2vRtjQ1txz1/IsC7rQ9UuCbnSLfEsjnrXRyiTMwKD2Z7IwUhg9IZVJuf38xTzlS5LMzksnskxhbJzy+IkjNgkGnuh0JEOtFH+CS++CTN+DNn8B1z7odTacy+yTy+A35/L6ojJdWl/PZ3tqwf2ZtbQs1hPdzIvH/b01tC7XGO8tXdqdotW9ZU2c53MHaDIEeMtB2CXFxJMQZ4uMMfRLjiY8zJMQZEuINCXFxR72O97d13uv4dWK8abNPHPHGULF1ExdNm0R2RgpZaUkkxMfWdF8B8a1wZtX0yBdZ7Bf9zGEw/Ufwzk/h02Vw0gVuR9QpYwzfnJbHN6flReTzCgsLmTHjvIh8VjjFSh7Qmst0t8MIicLaTxmX442bJF1RXQFVPpjyHbcjOaJ3fO1One8MlXr9dmg+tg9RRCQsfCudnx65iAu9pegnpsCsB5y74t5/wu1oRKS38K2A5AzIPsPtSI4IqOgbY2YZYzYZY7YYY+7o4P1kY8xL/vdXGWNy270/whhTY4y5NTRh98Aps+Dki6DwAajRkoYiEgG+IhgxBeKCnIs/hLos+saYeOAxYDYwGphjjBndrtlcoNJaezLwMPBgu/cfBl4PPtwgGOOc7TfWwdv3uBqKiPQCNXuc3gUPde1AYGf6k4Et1tqt1toG4EWg/S2uVwLP+J+/DMw0/uELxpirgK1AaWhCDsLAUc4FlQ+fhYrwDVMUEWFba3++N8bntwqk6A8Dytu8rvBv67CNtbYJqAayjDFpwO2Ad06tz78N+mY7UzS0aPFyEQkTXxEk9IETxrsdyVECGbLZ0eDS9rdkHK/NPcDD1tqazsYtG2PmAfMAsrOzg1+hpgvZOXM4feMjbHxpITtPuKjb+9fU1IQ9xkiJlVxiJQ9QLl7V3VzyS5fS1HcUH71bFL6gesJa2+kDmAosbfN6AbCgXZulwFT/8wRgL84XwXKgzP+oAvYD8zv7vPz8fBt2LS3WPnmRtT8/ydq6qm7vvmzZstDH5JJYySVW8rBWuXhVt3Kpq7L27kxrl90ftnjaA4ptF/XcWhtQ985qYJQxJs8YkwRcDyxp12YJcJP/+dXAO/44pltrc621ucAjwH9Za3/Z7W+mUDMGLvtvqN3rjOYREQmlbasA67mLuBBAn751+ujn45zNbwAWW2tLjTH3GmOu8Df7LU4f/hbgh8Axwzo9Z+h4yL/JWWxl90a3oxGRWOJbAXGJMGyS25EcI6BpGKy1BUBBu20L2zyvB67p4hj/2YP4wuvChf7FVm7z3GIrIhLFfEUwbKInp3TvHXfkHk9aFlxwJ3z2D9jwV7ejEZFY0HAIdqzxZNcO9PaiDzDp32DwGGexlQbvzNIoIlGqYjW0NHlufH4rFf3WxVaqt0HRIrejEZFo5ysCEwfDJ7sdSYdU9OGLxVbefRgqfW5HIyLRzLcChoyFlEy3I+mQin6rS37qfDu/eafbkYhItGpqcLp3PNq1Ayr6X8jMgek/hA1LYGuh29GISDTa8QE01Xv2Ii6o6B9t6i1abEVEes7nXwR9xFR34+iEin5biSlw6f2wZyO8/6Tb0YhItPGtgEGnQdpAtyM5LhX99k6d7V9s5X4ttiIigWtucqZf8HDXDqjoH6vtYit/986M0CLicbvWQsNBT1/EBRX9jrUutvKBFlsRkQD5/FMo60w/Sp33Y2exldd/rMVWRKRrviLonwcZQ92OpFMq+seTkgEX3wvbS+Cj592ORkS8rKXFKfoe79oBFf3Ojb0WcibD2/8J9dVuRyMiXrV3E9Tt93zXDqjody4uzpmXp3YvFD7odjQi4lWt4/NV9GPA0Akw8UZ4X4utiMhx+Iogfahzc6fHqegHYuZCSEqDN24H235NeBHp1az19+efExULManoByJtoLPYytZC2Pg3t6MRES+p/AwOfh4VXTugoh+4I4ut/AdxzYfdjkZEvOLI+Hzvj9wBFf3AxSfA7AehahvDy19xOxoR8QpfEaRmwaBT3Y4kICr63ZE3HcZ8hRHb/gRV29yORkS8wLfCmVUzCvrzQUW/+y65z/m59CfuxiEi7qveDpVlUdO1Ayr63ZeZw7YR12ixFRGBbSudn1FyERdU9HukfPhV0G+kFlsR6e18KyAp3VkTN0qo6PdAS3wSzPIvtrL6KbfDERG3+IpgxBSIi3c7koCp6PfUqZfBSTNh2f1Qs8ftaEQk0mr3Oid+UdS1Ayr6PWeMM4SzsVaLrYj0Rkf686PnIi6o6Aen7WIr27XYikiv4iuChBRnfq4oElDRN8bMMsZsMsZsMcbc0cH7ycaYl/zvrzLG5Pq3X2yMKTHGrPX/vDC04XvAebdB38FQcJsWWxHpTXwrIOcsSEhyO5Ju6bLoG2PigceA2cBoYI4xZnS7ZnOBSmvtycDDQOs8xHuBy621Y4GbgD+GKnDPSMmAi+6B7cX8fby8AAAK+UlEQVTw0QtuRyMikVBfDTvXRl3XDgR2pj8Z2GKt3WqtbQBeBK5s1+ZK4Bn/85eBmcYYY639wFq7w7+9FEgxxiSHInBPGXedf7GVu7XYikhvsG0V2Jaou4gLgRX9YUB5m9cV/m0dtrHWNgHVQFa7Nl8DPrDWxt5sZW0XW/nHz92ORkTCzbcC4hKd7p0okxBAm44mlGg/qXynbYwxY3C6fC7p8AOMmQfMA8jOzqawsDCAsNxTU1PTYYynnHAxQ977NcVNp3EobXjkA+uB4+USbWIlD1AuXtU2lwlrX4e+J/FB0fvuBtUT1tpOH8BUYGmb1wuABe3aLAWm+p8n4PTlG//rHOATYFpXn2WtJT8/33rdsmXLOn6jZo+19w+39pkrrG1piWhMPXXcXKJMrORhrXLxqiO5HK619p4B1r51t4vRHAsotgHU2EC6d1YDo4wxecaYJOB6YEm7NktwLtQCXA28Y621xph+wGv+L4kVPftaiiJpA+GCn2ixFZFYVrEaWpqi8iIuBNCnb50++vk4Z/MbgMXW2lJjzL3GmCv8zX4LZBljtgA/BFqHdc4HTgbuMsZ86H8MDnkWXjJpLgweDUv/Axrr3I5GRELNVwQmDoZPdjuSHgmkTx9rbQFQ0G7bwjbP64FrOtjvPuC+IGOMLvEJMPvn8MyXYcUimHG72xGJSCj5VjgTrKVkuh1Jj+iO3HDwL7bCuw9psRWRWNLU4HTvRGnXDqjoh8/FPwUMvHmn25GISKjs+ACa6qNyfH4rFf1w6Tccpv8I1r8KW//hdjQiEgo+/3iUEVPdjSMIKvrhdM4tWmxFJJb4imDQac5IvSiloh9OiSn+xVY2aLEVkWhnm2Hbe1HdtQMq+uF36mVw0oVabEUkyvWtKYOGg1F9ERdU9MPPGJilxVZEol2/qlLnSRT354OKfmQMOgXOvlmLrYhEsczqUuifC5nt55uMLir6kXL+7VpsRSRaWeuc6Ud51w6o6EdO28VWPn7R7WhEpDv2bCKx6WDUX8QFFf3IGnedM//2W1psRSSqtI7PV9GXbomLc+blqd2jxVZEoomviMNJA6B/ntuRBE1FP9KGTYSJ34BVj8OeTW5HIyJdsRZ8RVT1G+OMxotyKvpumHk3JKbBG3c4/6BExJvqKuGf/wMHd1CdOcbtaEJCRd8NaQPhgv+AT9+Bja+5HY2ItLd/KxT8GB4aDcvug5MuZPfg6W5HFRIBzacvYXDWt6Dk97B0AZw8ExL7uB2RSO9mLZSvgqJHnZOxuAQYdy1M+S4MOYOmGFnrV0XfLfEJcNnP4ZnLnX9k59/mdkQivVNzE2xYAit/6dw82ae/M0Pu5H+H9CFuRxdyKvpuyjsPRl8Fyx+CM6+HfiPcjkik96g/AGv+4AyqqC6HASfBl/4fnDkHktLcji5sVPTddsl98MlSePMuuPYZt6MRiX1V5U6hL3nmiwnUZv8cTpnlDKuOcSr6bus3HKb/EJb9zFls5cTz3Y5IJDZtL4GiXzoLG4GzpOnU7znDqHsRFX0vOOf7zmRsr98ON7/r9PeLSPBammHT605//baVkJwBU7/rTICYmeN2dK5QdfGCxBS49L/gpX9xFluZcrPbEYlEt4Za+PB5eO9XzvDLzBFw6f3OjZHJ6W5H5yoVfa847Uv+xVb+C8ZeHdXLsYm45sDn8P4TUPw01FfBsElwzUI47XL9Be2n/wpe0brYyq+nOoutXPGo2xGJRI+da2HlY7D2ZWdZw9O+DFPnw4iz3Y7Mc1T0vaR1sZWVj0H+N3vdBSbppayF5kZoqoPGemjyPxrroOnwsdub6v2v/e/7iuCzfzhTm5w11/l/aED0T4wWLir6XnP+7fDxYvjTXBiW/8X2DufoabftmDZdvQ+j9+yB3b/rUagB69YkVQG2bXfM03fvhr1/CNnxOmkYYLueO33XLtj/fNg/5yhH/dvo7N9NN/7NWcsZuz6H8kVdF3EbxMJCGTnOWhX5Nzk3VkmnVPS9JiXD6dp5664OllZsV3COKVQdFKQu2qQdqgW7tyeRBqgbE8oFPPncse3SDx2C5s9DdryOm0VmcryMujpo9IX3Q6zt/N9GT99r935KfQPUN0NCCqQOcH4mpDiDFxL6QEKyMwVJT7cnpPSKsfWhFFDRN8bMAn4BxANPWWsfaPd+MvAHIB/YB1xnrS3zv7cAmAs0A9+31i4NWfSx6tRZziMCVhcWMmPGjIh8Vji9HyN5AKyKoVyKYyiXWNHlV6QxJh54DJgNjAbmGGNGt2s2F6i01p4MPAw86N93NHA9MAaYBfzKfzwREXFBIH8XTQa2WGu3WmsbgBeBK9u1uRJonUPgZWCmMcb4t79orT1srf0M2OI/noiIuCCQoj8MKG/zusK/rcM21tomoBrICnBfERGJkED69DsartD+itbx2gSyL8aYecA8gOzsbAo9Pm91TU2N52MMVKzkEit5gHLxqljJJZCiXwEMb/M6B9hxnDYVxpgEIBPYH+C+WGufAJ4AmDRpkvX6hZ/CGLo4FSu5xEoeoFy8KlZyCaR7ZzUwyhiTZ4xJwrkwu6RdmyXATf7nVwPvWGutf/v1xphkY0weMAp4PzShi4hId3V5pm+tbTLGzAeW4gzZfNpaW2qMuRcottYuAX4L/NEYswXnDP96/76lxpjFwHqgCfietbY5TLmIiEgXAhqnb60tAArabVvY5nk9cM1x9v0Z8LMgYhQRkRAxNkJ3GQbKGLMHCPPtiEEbCITzNtZIipVcYiUPUC5e5fVcRlprB3XVyHNFPxoYY4qttZPcjiMUYiWXWMkDlItXxUoumrRCRKQXUdEXEelFVPR75gm3AwihWMklVvIA5eJVMZGL+vRFRHoRnemLiPQiKvoiIr2Iir6ISC+ioh9ixpjRxpjFxphfG2OudjuenjLGTDfGPG6MecoYU+R2PMEwxswwxiz35zPD7XiCYYw53Z/Hy8aY77gdTzCMMScaY35rjHnZ7Vi6K5pjV9FvwxjztDFmtzFmXbvts4wxm4wxW4wxd3RxmNnAo9ba7wA3hi3YToQiD2vtcmvtzcDf+GKBnIgL0e/EAjVACs7Mr64I0e9lg//3ci3g2o1CIcplq7V2bngjDVx3cvJa7N1irdXD/wDOAyYC69psiwc+BU4EkoCPcJaNHItTENs+BvsfjwH/DayI1jza7LcYyIjy30mcf79s4LlozsW/zxVAEfD1aM/Fv9/LbuXR05y8Fnt3HgFNuNZbWGv/aYzJbbf5yHKRAMaYF4ErrbX3A18+zqG+518L+M/hirUzocrDGDMCqLbWHghjuJ0K4e8EoBJIDkecgQhVLtaZ2XaJMeY14PnwRXx8If69eEJ3csKZOTgqqXuna91a8tEYk2uMeQL4A87Zvlf0ZOnKucDvwhZRz3X3d/JVY8xvgD8CvwxzbN3V3VxmGGMW+fMpOF47l3Q3lyxjzOPABGPMgnAH10Md5hQlsXdIZ/pdC2jJxyNvWFuGf+lHj+lWHgDW2rvDFEuwuvs7+TMu/dUVgO7mUggUhiuYIHU3l33AzeELJyQ6zClKYu+QzvS7FtCSj1EgVvIA5eJVsZRLq5jLSUW/a4EsFxkNYiUPUC5eFUu5tIq9nNy+kuylB/AC8DnQiPMNP9e//TLgE5yr+D9xO87ekody8e4jlnKJ5Zw6emjCNRGRXkTdOyIivYiKvohIL6KiLyLSi6joi4j0Iir6IiK9iIq+iEgvoqIvItKLqOiLiPQiKvoiIr3I/wc1MuX0zYalqwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# generate ESN reservoir\n",
    "inSize = 1\n",
    "outSize = 2\n",
    "resSize = 300\n",
    "a = 0.9 # leaking rate\n",
    "#reg = 1e-4  # regularization coefficient\n",
    "spec = 1\n",
    "scale = 0.5\n",
    "trainmean = list()\n",
    "testmean = list()\n",
    "avgoutmean = list()\n",
    "tries =  [100, 10, 1, 1e-1, 1e-2, 1e-4, 1e-6, 1e-7, 1e-8, 1e-10]\n",
    "for r in tries:\n",
    "    traineval = list()\n",
    "    testeval = list()\n",
    "    avgout = list()\n",
    "    for times in range(10):\n",
    "        data, y1, y2 = getData(numChanges = 5)\n",
    "        avg, n1, n2, n3, n4 = trainESN(data, y1, y2, initLen, trainLen, testLen, inSize, outSize, resSize, a, r, spec, scale)\n",
    "        traineval.append(n1)\n",
    "        testeval.append(n3)\n",
    "        avgout.append(avg)\n",
    "    trainmean.append(mean(traineval))\n",
    "    testmean.append(mean(testeval))\n",
    "    avgoutmean.append(mean(avgout))\n",
    "plt.semilogx(tries, trainmean)\n",
    "plt.semilogx(tries, testmean)\n",
    "plt.grid()\n",
    "plt.legend(['train', 'test'])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.semilogx(tries, avgoutmean)\n",
    "plt.grid()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}