Metodi_Monte_Carlo_RL_ITA.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "# Introduzione ai metodi Monte Carlo (MC)\n",
        "\n",
        "Mario Fiorino\n",
        "\n",
        "\n",
        "\n",
        "Il metodo Monte Carlo è una tecnica statistica utilizzata per trovare soluzioni approssimate attraverso il campionamento casuale (*random sampling*). In sostanza, gli approcci Monte Carlo, nelle loro diverse declinazioni, impiegano il campionamento casuale per generare dati, i quali vengono successivamente utilizzati per approssimare e stimare un valore. Per capirci di più vedere esempio : https://github.com/PacktPublishing/Deep-Reinforcement-Learning-with-Python/blob/master/04.%20Monte%20Carlo%20Methods/.ipynb_checkpoints/4.01.%20Understanding%20the%20Monte%20Carlo%20Method-checkpoint.ipynb , dove si vuole calcolre il valore atteso (expected value): $\\mathbb{E}[X]$, quando non si conosce esplicitamente la funzione di probabilità $p(x_{i})$.\n",
        "\n",
        "$\\;$\n",
        "\n",
        "**Metodi Monte Carlo per il Reinforcement Learning**\n",
        "\n",
        "Nel contesto Reinforcement Learning (RL), un algoritmo model-free (in contrapposizione a uno model-based) è un algoritmo che non presuppone una conoscenza completa della dinamica dell'ambiente: concretamente non necessitano della funzione di transizione di stato (transition model) e della funzione di ricompensa che definiscono la dinamica dell'ambiete markoviano (MDP).\n",
        "\n",
        "Un algoritmo RL model-free può essere pensato come un algoritmo esplicito *trial-and-error*; in quanto\n",
        "gli agenti sfruttano esclusivamente i dati derivanti dall'interazione con l'ambiente, senza la necessità di un modello del sistema. Monte Carlo Control, SARSA, Q-learning e Actor-Critic sono algoritmi RL model-free.\n",
        "\n",
        "I metodi Monte Carlo nel RL si basano sulla generazione di numerosi (per sfruttare la legge dei grandi numeri) episodi o traiettorie all'interno dell'ambiente, per ottenere stime della funzione valore (value function), o dei parametri della policy ottimale. Questi metodi richiedono solo \"esperienza\", ovvero sequenze di esempi di stati, azioni e ricompense derivanti dall'interazione reale o simulata con l'ambiente, senza necessitare di una conoscenza preliminare dello stesso.\n",
        "\n",
        "Nota 1: Apprendere da una simulazione necessita di un modello-campionario che serve solo a generare campioni, non le distribuzioni di probabilità complete di tutte le possibili transizioni necessarie per la programmazione dinamica. In molti casi, è sorprendentemente facile generare esperienza campionata secondo le distribuzioni di probabilità desiderate, ma impossibile ottenere tali distribuzioni in forma esplicita.\n",
        "\n",
        "Nota 2: I metodi MC assumono che l'esperienza sia suddivisa in episodi, ovvero sequenze di interazioni con l'ambiente che terminano in stati finali (detti anche *terminali* o *stati di assorbimento* , in quanto una volta raggiunti, non permettono più all'agente di effettuare transizioni verso altri stati).\n",
        "\n",
        "Nota 3: A ogni episodio, i valori della funzione di valore o i parametri della policy (policy) vengono aggiornati in base ai \"Return\" (somma cumulativa delle ricompense ottenute - vedi notazioni successive) campionati, migliorando progressivamente le stime dei valori ottimali cercati.\n"
      ],
      "metadata": {
        "id": "IqE26vl7t2vP"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Concetti di base\n",
        "\n",
        "Una **traiettoria** $\\tau$ è una sequenza di stati e azioni in un certo ambiente:\n",
        "\n",
        "$\\tau = (s_0, a_0, s_1, a_1, s_2,...)$\n",
        "\n",
        "Dove $s_0$ è lo stato di partenza, $a_0$ l'azione esegutia in tale stato, mentre $s_1$ è lo stato ottenuto dopo esecuzione dell'azione $a_0$, ect ...\n",
        "\n",
        "Le traiettorie sono spesso chiamate anche **episodi** o rolluot.\n",
        "\n",
        "La funzione di ricompensa (o reward function) $R$ dipende dallo stato attuale, dall’azione intrapresa in quello stato, e dal prossimo stato ottenuto:\n",
        "\n",
        "$r_t = R(s_t, a_t, s_{t+1})$\n",
        "\n",
        "L’obiettivo dell’agente è accumulare il maggior numero di ricompense ricevibili in un certo ambiente, durante un episodio: ovvero trovare tra le tante traiettorie $\\tau$ possibili, che partono da un certo stato iniziale al tempo $t=0$, quella che  massimizza il valore:\n",
        "\n",
        "$R(\\tau) = \\sum_{t=0}^{T} \\gamma^{t} r_{t}$\n",
        "\n",
        "$R(\\tau)$ è detto anche **Return**.\n",
        "\n",
        "Spesso si trova l'epressione *Return parziale* :\n",
        "\n",
        "$R_t(\\tau) = \\sum_{t'=t}^{T} \\gamma^{t'-t} r_{t'}$\n",
        "\n",
        "per indicare la somma scontata delle ricompense a partire dal passo temporale $t$ fino alla fine della traiettoria.  In modo equivalente, si parla di \"Return da uno stato $s_k$\" per riferirsi alla somma associata a una traiettoria che ha come stato iniziale $s_k$."
      ],
      "metadata": {
        "id": "8-Npod943Rak"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "$\\gamma$ : detto **discount factor**, è un parametro che assume valori nell'intervallo $(0,1)$, e determina il peso delle ricompense future rispetto a quelle immediate (ovvero in prossimità dello step corrente).\n",
        "\n",
        "Una variazione di $\\gamma$ modifica la percezione dell'agente riguardo alla natura del problema di RL. Se $\\gamma$ è basso, l'agente tende a focalizzarsi maggiormente sulle ricompense immediate, attribuendo meno importanza a quelle future.\n",
        "\n",
        "Si dice che questo porta a una riduzione *dell'orizzonte temporale del problema*, concentrandosi solo su un numero limitato di step futuri (e spesso velocizzando il training).\n",
        "\n",
        "La scelta del valore di $\\gamma$ dipende dalla natura del problema.\n",
        "\n",
        "La tabella sotto illustra l'effetto dello \"sconto\" applicato a una ricompensa ricevuta dopo $k$ step, per diversi valori di $\\gamma$. In sostanza espicita il valore nuemrico di $\\gamma^k$.\n",
        "\n",
        "Questa rappresentazione aiuta a comprendere l'orizzonte temporale del problema per un agente in funzione del valore di $\\gamma$\n",
        "\n",
        "Ad esempio, se $\\gamma = 0.8$ , l'orizzonte effettivo del problema si estende a circa $10$ step; mentre se $\\gamma = 0.99$, l'orizzonte rimane significativo anche a circa $200-500$ step.\n",
        "\n",
        "\n",
        "| $\\gamma$     | $\\gamma^{10}$   | $\\gamma^{50}$    | $\\gamma^{100}$   | $\\gamma^{200}$   | $\\gamma^{500}$   | $\\gamma^{1000}$  | $\\gamma^{2000}$  |\n",
        "|-------|--------|--------|--------|--------|--------|--------|--------|\n",
        "| 0.8   | 0.11   | 0.00   | 0.00   | 0.00   | 0.00   | 0.00   | 0.00   |\n",
        "| 0.9   | 0.35   | 0.01   | 0.00   | 0.00   | 0.00   | 0.00   | 0.00   |\n",
        "| 0.95  | 0.60   | 0.08   | 0.01   | 0.00   | 0.00   | 0.00   | 0.00   |\n",
        "| 0.99  | 0.90   | 0.61   | 0.37   | 0.13   | 0.07   | 0.00   | 0.00   |\n",
        "| 0.995 | 0.95   | 0.78   | 0.61   | 0.37   | 0.08   | 0.01   | 0.00   |\n",
        "| 0.999 | 0.99   | 0.95   | 0.90   | 0.82   | 0.61   | 0.37   | 0.14   |\n",
        "| 0.9997| 0.997  | 0.985  | 0.970  | 0.942  | 0.861  | 0.741  | 0.549  |\n",
        "\n"
      ],
      "metadata": {
        "id": "uUADxpqOrKoP"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "**Tipologie di Return:**\n",
        "\n",
        "Una possibile formulazione del return è il *finite-horizon undiscounted return*, che corrisponde semplicemente alla somma delle ricompense ottenute all'interno di un orizzonte temporale fisso, senza l'applicazione del *discount factor*:\n",
        "\n",
        "$R(\\tau) = \\sum_{t=0}^T r_t$.\n",
        "\n",
        "Nel testo di Sutton & Barto troverete la seguente simbologia per il return:\n",
        "$G_{t} =  r_{t+1} + r_{t+2} + r_{t+3} + ... + r_{T}$\n",
        " , cioè la somma di reward ricevuti dopo lo step $t$.\n",
        "\n",
        "\n",
        "Un altro tipo di return, utilizzato per i \"continuing tasks\", cioè quelli che non hanno uno stato terminale, è il *infinite-horizon undiscounted return*: che è la somma inifinita di tutti i reward ottenuti dall'agente, ma scontati di un fattore  $\\gamma$ :\n",
        "\n",
        "$R(\\tau) = \\sum_{t=0}^{\\infty} \\gamma^t r_t.$\n",
        "\n",
        "$\\;$\n",
        "\n",
        "Nota 1: La seguente proprietà che esprime il return  in modo ricorsivo, collegandolo al return ottenuto in passaggi temporali successivi.\n",
        "\n",
        "$R(\\tau) =  \\gamma^0 r_0 + \\gamma^1 r_1 + \\gamma^2 r_2 + \\gamma^3 r_3 + ... .$\n",
        "\n",
        "$R(\\tau) =   r_0 + \\gamma( r_1 + \\gamma^1 r_2 + \\gamma^2 r_3 + ... ).$\n",
        "\n",
        "$R(\\tau) =   r_0 + \\gamma (R_{t=t_{1}}(\\tau))$\n",
        "\n",
        "$\\;$\n",
        "\n",
        "**Funzioni valore**\n",
        "\n",
        "**State-value function**, $V^{\\pi}(s)$, fornisce l' *expected return* associato a uno stato $s$, nel caso in cui l'agente inizi la sua interazione con ambiente in tale stato, e segua una data policy $\\pi$ durante il processo decisionale:\n",
        "\n",
        "$V^{\\pi}(s) = \\mathbb{E}_{s_0 = s,\\ \\tau \\sim \\pi} \\; [R(\\tau)]$\n",
        "\n",
        "**Action-value function**, $Q^{\\pi}(s,a)$, fornisce l' *expected return* associato a una coppia stato-azione $(s, a)$, se  l'agente inizia nello stato s, qui esegue l'azione $a$ (che potrebbe non provenire dalla policy $\\pi$),  e, successivamente, dopo tale l'azione, adotta la politica $\\pi$ in modo permanente:\n",
        "\n",
        "$Q^{\\pi}(s,a) = \\mathbb{E}_{s_0 = s,\\ a_0 = a,\\ \\tau \\sim \\pi} \\; [R(\\tau)]$\n",
        "\n",
        "In seguito vedremo come $Q^{\\pi}(s,a)$ ha il vantaggio rispetto a $V^{\\pi}(s)$ di fornire agli agenti un metodo diretto per determinare l'azione da intraprendere.\n",
        "\n",
        "$Q^{\\pi}(s,a)$ può essere anche pensata come una forma di archiviazione del *one-step lookahead* per ogni azione $a$ in ogni stato $s$\n",
        "\n",
        "$\\;$\n",
        "\n",
        "**Relazioni tra due funzioni valore**\n",
        "\n",
        "La funzione valore di stato $V^{\\pi}(s)$ può essere espressa in termini della funzione azione-valore $Q^{\\pi}(s,a)$ :\n",
        "\n",
        "$V^{\\pi}(s) = \\mathbb{E}_{a\\sim \\pi(s)}{[Q^{\\pi}(s,a)]}$\n",
        "\n",
        "ovvero : $V^{\\pi}$  è un *expectation* su $Q^{\\pi}$ per tutte le azioni $a$ disponibili in un particolare stato $s$, seguendo la policy $π$\n",
        "\n",
        "$\\;$\n",
        "\n",
        "Inoltre, si consideri la policy ottimale $\\pi^*$, la relativa *Optimal Value Function* :\n",
        "\n",
        "$V^{\\pi^*}(s) = \\underset{\\pi}\\max \\; \\mathbb{E}_{s_0 = s,\\ \\tau \\sim \\pi}[{R(\\tau)]}$\n",
        "\n",
        "e *Optimal Action-Value Function* :\n",
        "\n",
        "$Q^{\\pi^*}(s,a) = \\underset{\\pi}\\max \\; \\mathbb{E}_{s_0 = s,\\ a_0 = a,\\ \\tau \\sim \\pi} \\; [R(\\tau)]$\n",
        "\n",
        "Dove il termine \" $\\underset{\\pi}\\max$ \" indica l'operazione di massimizzazione sull'insieme delle policy disponibili. In pratica, dice di selezionare la policy che produce il valore massimo dell'*expectation*.\n",
        "\n",
        "Vale, chiaramente, la relazione:\n",
        "\n",
        "$V^{\\pi^*}(s) = \\underset{a} \\max \\; Q^{\\pi^*}(s,a)$.\n",
        "\n",
        "$\\;$\n",
        "\n",
        "Ref:\n",
        "\n",
        "Reinforcement Learning: An Introduction; Richard S. Sutton and Andrew G. Barto, Second Edition - MIT Press, Cambridge, 2018.\n",
        "\n",
        "Spinning Up in Deep RL by OpenAI:\n",
        "\n",
        "https://spinningup.openai.com/en/latest/spinningup/rl_intro.html\n",
        "\n",
        "Foundations of Deep Reinforcement Learning: Theory and Practice in Python di Laura Graesser, Wah Loon Keng."
      ],
      "metadata": {
        "id": "ZpAMAZLV9oNA"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Simbologia matematica usata:\n",
        "\n",
        "<div>\n",
        "<img src=\"data:image/png;base64,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\", width=\"300\">\n",
        "</div>"
      ],
      "metadata": {
        "id": "Lwyza0w3dUIi"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Monte Carlo Prediction (policy evaluation)\n",
        "\n",
        "Iniziamo considerando i metodi Monte Carlo per ricavare la funzione valore per una policy arbitraria.\n"
      ],
      "metadata": {
        "id": "b9JaKymp3hzg"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Comprendere il gioco del BlackJack (BJ) in Gym\n",
        "\n",
        "\n",
        "Regole del gioco : https://www.gymlibrary.dev/environments/toy_text/blackjack/\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "id": "DJ8hcU5wbSYC"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "import gym\n",
        "import pandas as pd\n",
        "from collections import defaultdict\n",
        "import numpy as np\n",
        "import random\n",
        "\n",
        "import matplotlib\n",
        "from matplotlib import pyplot as plt\n",
        "from mpl_toolkits.mplot3d import Axes3D\n",
        "from mpl_toolkits.axes_grid1 import make_axes_locatable\n",
        "%matplotlib inline\n",
        "from IPython import display\n",
        "import pygame\n",
        "\n",
        "\n",
        "env = gym.make('Blackjack-v1')"
      ],
      "metadata": {
        "id": "P34fckmDdPyh",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a7b4dc31-3a13-46b1-aff0-85d493ceab46"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/gym/core.py:317: DeprecationWarning: \u001b[33mWARN: Initializing wrapper in old step API which returns one bool instead of two. It is recommended to set `new_step_api=True` to use new step API. This will be the default behaviour in future.\u001b[0m\n",
            "  deprecation(\n",
            "/usr/local/lib/python3.10/dist-packages/gym/wrappers/step_api_compatibility.py:39: DeprecationWarning: \u001b[33mWARN: Initializing environment in old step API which returns one bool instead of two. It is recommended to set `new_step_api=True` to use new step API. This will be the default behaviour in future.\u001b[0m\n",
            "  deprecation(\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def query_environment(name):\n",
        "    env = gym.make(name)\n",
        "    spec = gym.spec(name)\n",
        "    print(f\"Action Space: {env.action_space}\")\n",
        "    print(f\"Dimention Action Space: {env.action_space.n}\")\n",
        "    print(f\"Observation Space: {env.observation_space}\")\n",
        "    print(f\"Max Episode Steps: {spec.max_episode_steps}\")\n",
        "    print(f\"Reward Range: {env.reward_range}\")\n",
        "\n",
        "query_environment('Blackjack-v1')"
      ],
      "metadata": {
        "id": "aclZPsO4dgrY"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Inizializza il gioco\n",
        "state = env.reset()\n",
        "done = False\n",
        "\n",
        "# Stampa lo stato attuale del gioco : state\n",
        "# Questo consiste in un consiste in una tupla di 3 elementii:\n",
        "# state[0] : la somma  attuale del giocatore,\n",
        "# state[1] : il valore della carta scoperta del banco (1-10 dove 1 è un asso)\n",
        "# state[2] : Se il giocatore ha un asso “usable” (True o False)\n",
        "# Nota: Nel BJ gli assi possono contare come 11 (sono detti “usable”) o 1.\n",
        "print(\"Stato attuale: \", state)\n",
        "\n",
        "\n",
        "# Ripeti il gioco fino alla sua fine\n",
        "while not done:\n",
        "\n",
        "    # Ci sono due tipi di azione:\n",
        "    # Richiedi carte aggiuntive, una per una (hit = 1)\n",
        "    # Stai (stick = 0)\n",
        "    action = env.action_space.sample() # campionamento casuale di un'azione\n",
        "    if action == 0:\n",
        "      print(\" Azione 0: Stick!\")\n",
        "    else:\n",
        "      print(\" Azione 1: Hit!\")\n",
        "    #print(f\"Taking action: {action}\")\n",
        "\n",
        "    # Esegui l'azione ed osserva il risultato\n",
        "    next_state, reward, done, _ = env.step(action)\n",
        "\n",
        "\n",
        "    # Stato del gioco dopo la mossa effettuata\n",
        "    print(f\"Stato ottenuto dopo la mossa: {next_state}\")\n",
        "\n",
        "    # Update the state\n",
        "    state = next_state\n",
        "\n",
        "    #Reward\n",
        "    #win game: +1\n",
        "    #lose game: -1\n",
        "    #draw game: 0\n",
        "    print(f\" Reward : {reward}\")\n",
        "    print(f\" Il gioco è terminato? : {done}\") # se arriviamo allo stato terminale allora done == True\n",
        "\n",
        "\n",
        "#Il gioco è finito\n",
        "env.close()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "j8FD__RVesqc",
        "outputId": "fc61e8ff-f8fb-4607-d27f-b6ede435ba07"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Stato attuale:  (8, 7, False)\n",
            " Azione 0: Stick!\n",
            "Stato ottenuto dopo la mossa: (8, 7, False)\n",
            " Reward : 1.0\n",
            " Il gioco è terminato? : True\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#Rendering\n",
        "pygame.init()\n",
        "plt.rcParams[\"figure.figsize\"] = [6.00, 3.00]\n",
        "plt.imshow(env.render(mode='rgb_array'))\n",
        "plt.axis('off')\n",
        "plt.show()\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        },
        "id": "eEV8TJ0Vhin5",
        "outputId": "8d1c9fe3-da34-4df8-be38-2c3c3f2c9036"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 600x300 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Note sul gioco del BJ:\n",
        "\n",
        "1.\n",
        "\n",
        "Nell'impostazione data da Gym, tutti i reward ottenibili durante (cioè fin tanto il gioco non termina)  di una partita di BJ sono zero;  di fatto le ricompense ottenute negli stati terminali coincidono con il \"return\". Per tanto non ha senso applicare un di fattore sconto minore di uno, quindi gamma sara esattamente = 1.\n",
        "\n",
        "2.\n",
        "\n",
        "Anche se si dispone di una conoscenza buona dell'ambiente in cui si svolge il gioco del blackjack(regole, stati,...) , non sarebbe facile applicare metodi di programmazione dinamica (DP) per calcolare la funzione valore. I metodi DP richiedono la conoscenza della dinamica dell'ambiente, in particolare della  della funzione di distribuzione della transizione di stato (transition model); e non è facile determinare questa funzione per il blackjack. Per capirci, se la somma del giocatore è 14 e decide di restare, qual è la sua probabilità di terminare con una ricompensa di +1 in funzione della carta che il banco mostra? Tutte le probabilità devono essere calcolate prima di poter applicare la DP e tali calcoli sono spesso complessi e soggetti a errori.\n",
        "\n",
        "Al contrario, generare i campioni di giochi richiesti dai metodi Monte Carlo è molto facile!"
      ],
      "metadata": {
        "id": "ZbNQhV2sdqTD"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### First-visit Monte Carlo Prediction\n",
        "\n",
        "Algoritmo First-visit MC:\n",
        "\n",
        "0.\n",
        "Inizializza delle strutture dati in modo di avere un mapping, per ogni stato:\n",
        "\n",
        "total_return : state -> Somma totale dei return di ogni state, ottenuta durante tutto il training\n",
        "\n",
        "N : state -> Numero di volte che lo state è apparso durante tutti gli episodi\n",
        "\n",
        "\n",
        "1.\n",
        "Input: definisci una policy da valutare\n",
        "\n",
        "2.\n",
        "Genera un certo numero di episodi che seguono la policy prima definita\n",
        "\n",
        "3.\n",
        "Per ogni episodio generato:\n",
        "ricava il return di ogni stato incontrato, se incontri lo stesso stato più di una volta considera solo la prima visita\n",
        "\n",
        "4.\n",
        "Stima la vaule fuction di ogni stato  dividendo:\n",
        "\n",
        "( Somma totale dei return di un certo stato ottenuta durante tutto il training)  $\\div$   ( Numero di volte che tale stato è apparso durante tutti gli episodi )\n"
      ],
      "metadata": {
        "id": "hi_uiyQM9TsI"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Il codice sottostante è stato ispirato ed ha come riferimento:\n",
        "# https://github.com/PacktPublishing/Deep-Reinforcement-Learning-with-Python/tree/master/04.%20Monte%20Carlo%20Methods\n",
        "# di Sudharsan Ravichandiran\n"
      ],
      "metadata": {
        "id": "8Dw8H9wzjfUx"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Definiamo una policy di input per la fase di evaluation\n",
        "#\n",
        "# Come si comporta questa semplice policy:\n",
        "# Se il valore totale del giocatore: state[0], è superiore a 18,\n",
        "# meglio non chiedere un'altra carta: stick = 0, in quanto si rischia di perdere la partita\n",
        "# In tutti gli altri casi, chiede carta hit = 1\n",
        "\n",
        "def policy(state):\n",
        "    if state[0] > 18:\n",
        "       return 0\n",
        "    return 1"
      ],
      "metadata": {
        "id": "rA6zjwYO61Vm"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Generiamo un singolo episodio utilizzando la policy data come input\n",
        "#\n",
        "# \" Generate an episode following policy : S_0, A_0, R_1, S_1, A_1, R_2,...,S_{T-1}, A_{T-1}, R_T \"\n",
        "\n",
        "def generate_episode(policy):\n",
        "\n",
        "    # Definiamo una lista chiamato episodio gli esiti della partita\n",
        "    episode = []\n",
        "\n",
        "    #inizializziamo lo stato reimpostando l'ambiente\n",
        "    state = env.reset()\n",
        "\n",
        "    #print(\"Stato di partenza: \", state)\n",
        "\n",
        "    # Ripeti il gioco fino alla sua fine\n",
        "    while True:\n",
        "\n",
        "        #selezioniamo l'azione in base alla politica specificata\n",
        "        action = policy(state)\n",
        "\n",
        "        #eseguiamo l'azione e memorizziamo le informazioni della partita\n",
        "        next_state, reward, done, _ = env.step(action)\n",
        "\n",
        "        #memorizza lo stato di partenza, l'azione eseguita e la ricompensa otteunta, nella lista \"episode\"\n",
        "        episode.append((state, action, reward))\n",
        "\n",
        "        if done:\n",
        "            break\n",
        "\n",
        "        state = next_state\n",
        "\n",
        "    return episode"
      ],
      "metadata": {
        "id": "wmI73eLL_NY6"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#Come possiamo osservare, l'output di ogni step della partita è sotto forma di\n",
        "# (state, action, reward) : (tupla, numero intero, numero float)\n",
        "generate_episode(policy)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sOIqMAciC6z-",
        "outputId": "8b800f88-208b-4cf9-fc83-9798a6de5a45"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[((15, 2, True), 1, 0.0), ((15, 2, False), 1, -1.0)]"
            ]
          },
          "metadata": {},
          "execution_count": 108
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "##### Premessa tecnica del codice che seguirà"
      ],
      "metadata": {
        "id": "tZrFJj4wNcHp"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Nota tecnica :\n",
        "#\n",
        "# per maggiori info su Defaultdict\n",
        "# https://www.geeksforgeeks.org/defaultdict-in-python/\n",
        "# Di base\n",
        "# \"The functionality of both dictionaries and defaultdict are almost same except for the fact that\n",
        "# defaultdict never raises a KeyError.\n",
        "# It provides a default value for the key that does not exists.\"\n",
        "\n",
        "defaultdict_demo = defaultdict(float)\n",
        "defaultdict_demo[1] = 99.99\n",
        "# un classico dict = {}, nelle due righe successive darebbe errore\n",
        "#  in quanto nel dizionario demo non esiste la Key = 3 o la Key = 7\n",
        "defaultdict_demo[3]\n",
        "defaultdict_demo[7]\n",
        "print(defaultdict_demo)\n",
        "print(\"---\")\n",
        "print(defaultdict_demo[3])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "mDxl3zDRPpa2",
        "outputId": "5bc0f453-404b-4895-d5d4-4b69a6903314"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "defaultdict(<class 'float'>, {1: 99.99, 3: 0.0, 7: 0.0})\n",
            "---\n",
            "0.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#a = generate_episode(policy)\n",
        "a = [((9, 4, False), 1, 0.0), ((13, 4, False), 1, 0.0), ((21, 4, False), 0, 1.0)]\n",
        "print(a)\n",
        "\n",
        "# la funzione zip(*) scoppatta i dati in ingresso:\n",
        "# maggiori info : https://www.geeksforgeeks.org/zip-in-python/\n",
        "xs,xa,xr = zip(*a)\n",
        "print(\" -- \")\n",
        "print(xs)\n",
        "print(xa)\n",
        "print(xr)\n",
        "\n",
        "print(\"\\n*\\n\")\n",
        "\n",
        "# La funzione enumerate() aggiunge un contatore a un iterabile, e lo restituisce\n",
        "\n",
        "for t, st in enumerate(xs):\n",
        "  print(\"Contatore:\",t)\n",
        "  print(\"st : \", st)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "C8Z8-ri9K0UX",
        "outputId": "95677c0c-22e5-4719-9972-e9a96b0ce280"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[((9, 4, False), 1, 0.0), ((13, 4, False), 1, 0.0), ((21, 4, False), 0, 1.0)]\n",
            " -- \n",
            "((9, 4, False), (13, 4, False), (21, 4, False))\n",
            "(1, 1, 0)\n",
            "(0.0, 0.0, 1.0)\n",
            "\n",
            "*\n",
            "\n",
            "Contatore: 0\n",
            "st :  (9, 4, False)\n",
            "Contatore: 1\n",
            "st :  (13, 4, False)\n",
            "Contatore: 2\n",
            "st :  (21, 4, False)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#Esplicitazione tecnica del codice usato\n",
        "#\n",
        "# Innanzitutto, definiamo due dizionarii per memorizzare:\n",
        "# l'insime di \"return\" ottenuti in funzione degli state.\n",
        "total_return = defaultdict(float)\n",
        "total_return[(20, 10, False)] =  3.0 # gli diamo anche alcuni valori di partenza\n",
        "total_return[(20, 18, False)] += 5.0  # se la chiave non è presente gli da valore zero ed in questo caso poi incrementa di 5.0\n",
        "# il numero di volte in cui lo stato viene visitato nel corso degli episodi \"N[state]\"\n",
        "N = defaultdict(int)\n",
        "\n",
        "#Imposta il numero di episodi che vogliamo generare:\n",
        "num_iterations = 4\n",
        "\n",
        "for i in range(num_iterations):\n",
        "\n",
        "    #Generiamo un singolo episodio utilizzando la policy specificata\n",
        "    episode = generate_episode(policy)\n",
        "\n",
        "    #Memorizza in un certo modo tutti gli stati, le azioni, i rewards ottenuti dall'episodio\n",
        "    states, actions, rewards = zip(*episode)\n",
        "\n",
        "    print(\"\\n EP\",i+1)\n",
        "    print(\"states: \", states)\n",
        "    print(\"rewards\", rewards)\n",
        "    print(\"--\")\n",
        "    #Poi per ogni step dell'episodio generato\n",
        "    for t, state in enumerate(states):\n",
        "        print(\"t :\", t)\n",
        "        gh = states[0:t]\n",
        "        print(\"states[0:t] \", gh)\n",
        "        print(\"state ;\", state)\n",
        "\n",
        "        #Se lo state non è già stato visitato nel segunte episodio, fai i sucessivi aggiornamenti\n",
        "        if state not in states[0:t]:\n",
        "\n",
        "            R = (sum(rewards[t:]))\n",
        "            print(\"rewards[t:] \", rewards[t:])\n",
        "            print(\"R = sum(rewards[t:] : \",R)\n",
        "\n",
        "            total_return[state] =  total_return[state] + R\n",
        "            print(f\"total_return[{state}] : \",total_return[state])\n",
        "\n",
        "            # Aggiorna il numero di volte in cui un certo stato viene visitato\n",
        "            N[state] =  N[state] + 1\n",
        "\n",
        "print(\"\\nFINALE\")\n",
        "print(total_return)\n",
        "print(N)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "PEa2x8uZelFc",
        "outputId": "7614fc61-c47e-454f-8ad5-92817c50c020"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\n",
            " EP 1\n",
            "states:  ((13, 9, True), (14, 9, True), (18, 9, True), (19, 9, True))\n",
            "rewards (0.0, 0.0, 0.0, 0.0)\n",
            "--\n",
            "t : 0\n",
            "states[0:t]  ()\n",
            "state ; (13, 9, True)\n",
            "rewards[t:]  (0.0, 0.0, 0.0, 0.0)\n",
            "R = sum(rewards[t:] :  0.0\n",
            "total_return[(13, 9, True)] :  0.0\n",
            "t : 1\n",
            "states[0:t]  ((13, 9, True),)\n",
            "state ; (14, 9, True)\n",
            "rewards[t:]  (0.0, 0.0, 0.0)\n",
            "R = sum(rewards[t:] :  0.0\n",
            "total_return[(14, 9, True)] :  0.0\n",
            "t : 2\n",
            "states[0:t]  ((13, 9, True), (14, 9, True))\n",
            "state ; (18, 9, True)\n",
            "rewards[t:]  (0.0, 0.0)\n",
            "R = sum(rewards[t:] :  0.0\n",
            "total_return[(18, 9, True)] :  0.0\n",
            "t : 3\n",
            "states[0:t]  ((13, 9, True), (14, 9, True), (18, 9, True))\n",
            "state ; (19, 9, True)\n",
            "rewards[t:]  (0.0,)\n",
            "R = sum(rewards[t:] :  0.0\n",
            "total_return[(19, 9, True)] :  0.0\n",
            "\n",
            " EP 2\n",
            "states:  ((21, 8, True),)\n",
            "rewards (1.0,)\n",
            "--\n",
            "t : 0\n",
            "states[0:t]  ()\n",
            "state ; (21, 8, True)\n",
            "rewards[t:]  (1.0,)\n",
            "R = sum(rewards[t:] :  1.0\n",
            "total_return[(21, 8, True)] :  1.0\n",
            "\n",
            " EP 3\n",
            "states:  ((14, 3, False),)\n",
            "rewards (-1.0,)\n",
            "--\n",
            "t : 0\n",
            "states[0:t]  ()\n",
            "state ; (14, 3, False)\n",
            "rewards[t:]  (-1.0,)\n",
            "R = sum(rewards[t:] :  -1.0\n",
            "total_return[(14, 3, False)] :  -1.0\n",
            "\n",
            " EP 4\n",
            "states:  ((17, 7, False), (18, 7, False), (20, 7, False))\n",
            "rewards (0.0, 0.0, 1.0)\n",
            "--\n",
            "t : 0\n",
            "states[0:t]  ()\n",
            "state ; (17, 7, False)\n",
            "rewards[t:]  (0.0, 0.0, 1.0)\n",
            "R = sum(rewards[t:] :  1.0\n",
            "total_return[(17, 7, False)] :  1.0\n",
            "t : 1\n",
            "states[0:t]  ((17, 7, False),)\n",
            "state ; (18, 7, False)\n",
            "rewards[t:]  (0.0, 1.0)\n",
            "R = sum(rewards[t:] :  1.0\n",
            "total_return[(18, 7, False)] :  1.0\n",
            "t : 2\n",
            "states[0:t]  ((17, 7, False), (18, 7, False))\n",
            "state ; (20, 7, False)\n",
            "rewards[t:]  (1.0,)\n",
            "R = sum(rewards[t:] :  1.0\n",
            "total_return[(20, 7, False)] :  1.0\n",
            "\n",
            "FINALE\n",
            "defaultdict(<class 'float'>, {(20, 10, False): 3.0, (20, 18, False): 5.0, (13, 9, True): 0.0, (14, 9, True): 0.0, (18, 9, True): 0.0, (19, 9, True): 0.0, (21, 8, True): 1.0, (14, 3, False): -1.0, (17, 7, False): 1.0, (18, 7, False): 1.0, (20, 7, False): 1.0})\n",
            "defaultdict(<class 'int'>, {(13, 9, True): 1, (14, 9, True): 1, (18, 9, True): 1, (19, 9, True): 1, (21, 8, True): 1, (14, 3, False): 1, (17, 7, False): 1, (18, 7, False): 1, (20, 7, False): 1})\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "#### Calcolo della funzione valore"
      ],
      "metadata": {
        "id": "PXhMmBTYOSP5"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Innanzitutto, definiamo due dizionarii per memorizzare:\n",
        "# l'insime di \"return parziali\" ottenuti in funzione degli state.\n",
        "# il numero di volte in cui lo stato viene visitato nel corso degli episodi \"N[state]\"\n",
        "total_return = defaultdict(float)\n",
        "N = defaultdict(int)\n",
        "\n",
        "#Imposta il numero di episodi che vogliamo generare:\n",
        "num_iterations = 1_000_000\n",
        "\n",
        "for i in range(num_iterations):\n",
        "\n",
        "    #Generiamo un singolo episodio utilizzando la policy specificata\n",
        "    episode = generate_episode(policy)\n",
        "\n",
        "    #Memorizza in un certo modo tutti gli stati, le azioni, i rewards ottenuti dall'episodio\n",
        "    states, actions, rewards = zip(*episode)\n",
        "\n",
        "    #Poi per ogni step dell'episodio generato\n",
        "    for t, state in enumerate(states):\n",
        "\n",
        "        #Se lo state non è già stato visitato nel segunte episodio, fai i sucessivi aggiornamenti\n",
        "        if state not in states[0:t]:\n",
        "\n",
        "            # Calcoliamo il return R di un certo stato come somma dei reward ottenuti.\n",
        "            # Di fatto il reward ottenuto nello stato terminale coincidono con il \"return\".\n",
        "            # E' una somma di 0.0 che finisce con 1.0 oppure -1.0 oppure 0.0\n",
        "            R = (sum(rewards[t:]))\n",
        "\n",
        "            #Aggiorna il dizionario total_return, mappando gli stati con i return ottenuti\n",
        "            total_return[state] =  total_return[state] + R\n",
        "\n",
        "            # Aggiorna il numero di volte in cui un certo stato viene visitato\n",
        "            N[state] =  N[state] + 1\n"
      ],
      "metadata": {
        "id": "NaHkh0A-FT0k"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print(len(total_return))\n",
        "print(len(N))\n",
        "# in genere gia con 10_000 episodi si raggiungono i 280 stati"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jadIRtzLcG_l",
        "outputId": "53e85b02-6cbd-4c68-d88f-a76de3b78a31"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "280\n",
            "280\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Per una migliore comprensione dei dati, possiamo convertirli in un dataframe Pandas.\n",
        "# Tuttavia, questa conversione non è necessaria per implementare l'algoritmo in modo efficiente.\n",
        "\n",
        "# Converti il dizionario total_returns in un dataframe pandas:\n",
        "total_return = pd.DataFrame(total_return.items(),columns=['state', 'total_return'])\n",
        "\n",
        "# Converti il dizionario N in un dataframe pandas:\n",
        "N = pd.DataFrame(N.items(),columns=['state', 'N'])\n",
        "\n",
        "# Unisci i due dataframe usando come base comune gli state\n",
        "df = pd.merge(total_return, N, on=\"state\")\n",
        "\n",
        "# Dai un occhiata alle prime righe del dataframe pandas\n",
        "#df.head(10)"
      ],
      "metadata": {
        "id": "C2A1bGW2ZzP4"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Qui ricaviamo la funzione valore (expected retur) per ogni stato,\n",
        "# approssimandola con la media empirica dei returns ottenuti\n",
        "\n",
        "df['value'] = df['total_return'] / df['N']\n",
        "\n",
        "df.head(15)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 519
        },
        "id": "-i8AKBjFdCk5",
        "outputId": "53948632-a608-4216-cf3d-50ff7caa0ca8"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "              state  total_return      N     value\n",
              "0    (13, 1, False)       -5520.0   9370 -0.589114\n",
              "1    (19, 1, False)       -1193.0  10645 -0.112071\n",
              "2    (15, 1, False)       -6371.0   9889 -0.644251\n",
              "3     (6, 5, False)        -550.0   1466 -0.375171\n",
              "4    (15, 5, False)       -5235.0   9992 -0.523919\n",
              "5    (18, 8, False)       -6318.0  10612 -0.595364\n",
              "6    (21, 10, True)       16794.0  18243  0.920572\n",
              "7   (10, 10, False)       -1703.0  15049 -0.113164\n",
              "8   (19, 10, False)        -772.0  42349 -0.018229\n",
              "9    (17, 8, False)       -5884.0  10363 -0.567789\n",
              "10   (13, 3, False)       -4154.0   9238 -0.449664\n",
              "11   (16, 3, False)       -5601.0  10012 -0.559429\n",
              "12   (21, 3, False)        5997.0   6724  0.891880\n",
              "13  (12, 10, False)      -17097.0  35584 -0.480469\n",
              "14  (18, 10, False)      -28370.0  42523 -0.667168"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-424be4aa-daac-48f8-a73e-2d7bf7f1d170\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>state</th>\n",
              "      <th>total_return</th>\n",
              "      <th>N</th>\n",
              "      <th>value</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>(13, 1, False)</td>\n",
              "      <td>-5520.0</td>\n",
              "      <td>9370</td>\n",
              "      <td>-0.589114</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>(19, 1, False)</td>\n",
              "      <td>-1193.0</td>\n",
              "      <td>10645</td>\n",
              "      <td>-0.112071</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>(15, 1, False)</td>\n",
              "      <td>-6371.0</td>\n",
              "      <td>9889</td>\n",
              "      <td>-0.644251</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>(6, 5, False)</td>\n",
              "      <td>-550.0</td>\n",
              "      <td>1466</td>\n",
              "      <td>-0.375171</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>(15, 5, False)</td>\n",
              "      <td>-5235.0</td>\n",
              "      <td>9992</td>\n",
              "      <td>-0.523919</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>(18, 8, False)</td>\n",
              "      <td>-6318.0</td>\n",
              "      <td>10612</td>\n",
              "      <td>-0.595364</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>(21, 10, True)</td>\n",
              "      <td>16794.0</td>\n",
              "      <td>18243</td>\n",
              "      <td>0.920572</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>(10, 10, False)</td>\n",
              "      <td>-1703.0</td>\n",
              "      <td>15049</td>\n",
              "      <td>-0.113164</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>(19, 10, False)</td>\n",
              "      <td>-772.0</td>\n",
              "      <td>42349</td>\n",
              "      <td>-0.018229</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>(17, 8, False)</td>\n",
              "      <td>-5884.0</td>\n",
              "      <td>10363</td>\n",
              "      <td>-0.567789</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>(13, 3, False)</td>\n",
              "      <td>-4154.0</td>\n",
              "      <td>9238</td>\n",
              "      <td>-0.449664</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>(16, 3, False)</td>\n",
              "      <td>-5601.0</td>\n",
              "      <td>10012</td>\n",
              "      <td>-0.559429</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>(21, 3, False)</td>\n",
              "      <td>5997.0</td>\n",
              "      <td>6724</td>\n",
              "      <td>0.891880</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>(12, 10, False)</td>\n",
              "      <td>-17097.0</td>\n",
              "      <td>35584</td>\n",
              "      <td>-0.480469</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>(18, 10, False)</td>\n",
              "      <td>-28370.0</td>\n",
              "      <td>42523</td>\n",
              "      <td>-0.667168</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-424be4aa-daac-48f8-a73e-2d7bf7f1d170')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-424be4aa-daac-48f8-a73e-2d7bf7f1d170 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-424be4aa-daac-48f8-a73e-2d7bf7f1d170');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-a462bc26-8995-45d8-b180-a25cd86e481e\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-a462bc26-8995-45d8-b180-a25cd86e481e')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-a462bc26-8995-45d8-b180-a25cd86e481e button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "##### Studiamo un pò più da vicino le informazioni ottenute dal nostro algoritmo Monte Carlo  "
      ],
      "metadata": {
        "id": "3xNb6JeRx2Hc"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Riordina i dati in base alla funzoine valore con valore maggiore\n",
        "kf = df.sort_values(by=['value'],ascending=False)\n",
        "\n",
        "kf.head(30)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 989
        },
        "id": "bi__2R7riYhD",
        "outputId": "1755d09c-a707-49af-e633-963b9bbac15d"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "               state  total_return      N     value\n",
              "115    (21, 9, True)        4440.0   4495  0.987764\n",
              "92     (21, 7, True)        4392.0   4448  0.987410\n",
              "169    (21, 8, True)        4553.0   4623  0.984858\n",
              "141    (21, 6, True)        4496.0   4575  0.982732\n",
              "170    (21, 5, True)        4416.0   4509  0.979375\n",
              "129    (21, 2, True)        4422.0   4516  0.979185\n",
              "118    (21, 3, True)        4458.0   4554  0.978920\n",
              "54     (21, 4, True)        4505.0   4604  0.978497\n",
              "153   (21, 9, False)        6209.0   6616  0.938482\n",
              "167   (21, 8, False)        6151.0   6572  0.935940\n",
              "179   (21, 7, False)        6092.0   6598  0.923310\n",
              "6     (21, 10, True)       16794.0  18243  0.920572\n",
              "133   (21, 6, False)        5923.0   6579  0.900289\n",
              "84    (21, 4, False)        5918.0   6635  0.891937\n",
              "12    (21, 3, False)        5997.0   6724  0.891880\n",
              "60    (21, 5, False)        5914.0   6635  0.891334\n",
              "48   (21, 10, False)       23712.0  26693  0.888323\n",
              "140   (21, 2, False)        5969.0   6798  0.878052\n",
              "105    (20, 8, True)        1489.0   1840  0.809239\n",
              "63    (20, 8, False)       11105.0  14144  0.785139\n",
              "36    (20, 7, False)       11132.0  14245  0.781467\n",
              "142    (20, 7, True)        1350.0   1742  0.774971\n",
              "114   (20, 9, False)       10791.0  14221  0.758807\n",
              "31     (20, 9, True)        1336.0   1780  0.750562\n",
              "150    (20, 5, True)        1232.0   1764  0.698413\n",
              "123   (20, 6, False)        9796.0  14043  0.697572\n",
              "208    (20, 6, True)        1177.0   1744  0.674885\n",
              "22     (21, 1, True)        3002.0   4456  0.673698\n",
              "90    (20, 5, False)        9320.0  13976  0.666857\n",
              "78    (20, 4, False)        9379.0  14208  0.660121"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-e0eaae70-74aa-4348-9ef6-dbff27c3ec9c\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>state</th>\n",
              "      <th>total_return</th>\n",
              "      <th>N</th>\n",
              "      <th>value</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>115</th>\n",
              "      <td>(21, 9, True)</td>\n",
              "      <td>4440.0</td>\n",
              "      <td>4495</td>\n",
              "      <td>0.987764</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>92</th>\n",
              "      <td>(21, 7, True)</td>\n",
              "      <td>4392.0</td>\n",
              "      <td>4448</td>\n",
              "      <td>0.987410</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>169</th>\n",
              "      <td>(21, 8, True)</td>\n",
              "      <td>4553.0</td>\n",
              "      <td>4623</td>\n",
              "      <td>0.984858</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>141</th>\n",
              "      <td>(21, 6, True)</td>\n",
              "      <td>4496.0</td>\n",
              "      <td>4575</td>\n",
              "      <td>0.982732</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>170</th>\n",
              "      <td>(21, 5, True)</td>\n",
              "      <td>4416.0</td>\n",
              "      <td>4509</td>\n",
              "      <td>0.979375</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>129</th>\n",
              "      <td>(21, 2, True)</td>\n",
              "      <td>4422.0</td>\n",
              "      <td>4516</td>\n",
              "      <td>0.979185</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>118</th>\n",
              "      <td>(21, 3, True)</td>\n",
              "      <td>4458.0</td>\n",
              "      <td>4554</td>\n",
              "      <td>0.978920</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>54</th>\n",
              "      <td>(21, 4, True)</td>\n",
              "      <td>4505.0</td>\n",
              "      <td>4604</td>\n",
              "      <td>0.978497</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>153</th>\n",
              "      <td>(21, 9, False)</td>\n",
              "      <td>6209.0</td>\n",
              "      <td>6616</td>\n",
              "      <td>0.938482</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>167</th>\n",
              "      <td>(21, 8, False)</td>\n",
              "      <td>6151.0</td>\n",
              "      <td>6572</td>\n",
              "      <td>0.935940</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>179</th>\n",
              "      <td>(21, 7, False)</td>\n",
              "      <td>6092.0</td>\n",
              "      <td>6598</td>\n",
              "      <td>0.923310</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>(21, 10, True)</td>\n",
              "      <td>16794.0</td>\n",
              "      <td>18243</td>\n",
              "      <td>0.920572</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>133</th>\n",
              "      <td>(21, 6, False)</td>\n",
              "      <td>5923.0</td>\n",
              "      <td>6579</td>\n",
              "      <td>0.900289</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>84</th>\n",
              "      <td>(21, 4, False)</td>\n",
              "      <td>5918.0</td>\n",
              "      <td>6635</td>\n",
              "      <td>0.891937</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>(21, 3, False)</td>\n",
              "      <td>5997.0</td>\n",
              "      <td>6724</td>\n",
              "      <td>0.891880</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>60</th>\n",
              "      <td>(21, 5, False)</td>\n",
              "      <td>5914.0</td>\n",
              "      <td>6635</td>\n",
              "      <td>0.891334</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>48</th>\n",
              "      <td>(21, 10, False)</td>\n",
              "      <td>23712.0</td>\n",
              "      <td>26693</td>\n",
              "      <td>0.888323</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>140</th>\n",
              "      <td>(21, 2, False)</td>\n",
              "      <td>5969.0</td>\n",
              "      <td>6798</td>\n",
              "      <td>0.878052</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>105</th>\n",
              "      <td>(20, 8, True)</td>\n",
              "      <td>1489.0</td>\n",
              "      <td>1840</td>\n",
              "      <td>0.809239</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>63</th>\n",
              "      <td>(20, 8, False)</td>\n",
              "      <td>11105.0</td>\n",
              "      <td>14144</td>\n",
              "      <td>0.785139</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>36</th>\n",
              "      <td>(20, 7, False)</td>\n",
              "      <td>11132.0</td>\n",
              "      <td>14245</td>\n",
              "      <td>0.781467</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>142</th>\n",
              "      <td>(20, 7, True)</td>\n",
              "      <td>1350.0</td>\n",
              "      <td>1742</td>\n",
              "      <td>0.774971</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>114</th>\n",
              "      <td>(20, 9, False)</td>\n",
              "      <td>10791.0</td>\n",
              "      <td>14221</td>\n",
              "      <td>0.758807</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>31</th>\n",
              "      <td>(20, 9, True)</td>\n",
              "      <td>1336.0</td>\n",
              "      <td>1780</td>\n",
              "      <td>0.750562</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>150</th>\n",
              "      <td>(20, 5, True)</td>\n",
              "      <td>1232.0</td>\n",
              "      <td>1764</td>\n",
              "      <td>0.698413</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>123</th>\n",
              "      <td>(20, 6, False)</td>\n",
              "      <td>9796.0</td>\n",
              "      <td>14043</td>\n",
              "      <td>0.697572</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>208</th>\n",
              "      <td>(20, 6, True)</td>\n",
              "      <td>1177.0</td>\n",
              "      <td>1744</td>\n",
              "      <td>0.674885</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>22</th>\n",
              "      <td>(21, 1, True)</td>\n",
              "      <td>3002.0</td>\n",
              "      <td>4456</td>\n",
              "      <td>0.673698</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>90</th>\n",
              "      <td>(20, 5, False)</td>\n",
              "      <td>9320.0</td>\n",
              "      <td>13976</td>\n",
              "      <td>0.666857</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>78</th>\n",
              "      <td>(20, 4, False)</td>\n",
              "      <td>9379.0</td>\n",
              "      <td>14208</td>\n",
              "      <td>0.660121</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-e0eaae70-74aa-4348-9ef6-dbff27c3ec9c')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-e0eaae70-74aa-4348-9ef6-dbff27c3ec9c button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-e0eaae70-74aa-4348-9ef6-dbff27c3ec9c');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-4f71d60a-4266-457e-801a-11e98de15171\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-4f71d60a-4266-457e-801a-11e98de15171')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-4f71d60a-4266-457e-801a-11e98de15171 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Qui i dati sono ordinati in base al numero di volte che\n",
        "# si verificano. Dal minore al maggiore.\n",
        "# Lo stato (20,10, False) si è verificato 56179 volte\n",
        "df.sort_values(by=['N'])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "id": "yeZNtmgKiuJK",
        "outputId": "34bd7f06-5c18-4718-aa3b-979f8965e561"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "               state  total_return      N     value\n",
              "173    (4, 3, False)        -127.0    424 -0.299528\n",
              "222    (4, 8, False)        -145.0    435 -0.333333\n",
              "275    (12, 6, True)           4.0    441  0.009070\n",
              "236    (4, 2, False)        -106.0    441 -0.240363\n",
              "272    (12, 4, True)         -31.0    443 -0.069977\n",
              "..               ...           ...    ...       ...\n",
              "53   (16, 10, False)      -25055.0  40504 -0.618581\n",
              "20   (17, 10, False)      -26973.0  42001 -0.642199\n",
              "8    (19, 10, False)        -772.0  42349 -0.018229\n",
              "14   (18, 10, False)      -28370.0  42523 -0.667168\n",
              "21   (20, 10, False)       24275.0  56179  0.432101\n",
              "\n",
              "[280 rows x 4 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-43b7fd4c-7d7f-49f1-8be1-0335448a18f2\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>state</th>\n",
              "      <th>total_return</th>\n",
              "      <th>N</th>\n",
              "      <th>value</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>173</th>\n",
              "      <td>(4, 3, False)</td>\n",
              "      <td>-127.0</td>\n",
              "      <td>424</td>\n",
              "      <td>-0.299528</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>222</th>\n",
              "      <td>(4, 8, False)</td>\n",
              "      <td>-145.0</td>\n",
              "      <td>435</td>\n",
              "      <td>-0.333333</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>275</th>\n",
              "      <td>(12, 6, True)</td>\n",
              "      <td>4.0</td>\n",
              "      <td>441</td>\n",
              "      <td>0.009070</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>236</th>\n",
              "      <td>(4, 2, False)</td>\n",
              "      <td>-106.0</td>\n",
              "      <td>441</td>\n",
              "      <td>-0.240363</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>272</th>\n",
              "      <td>(12, 4, True)</td>\n",
              "      <td>-31.0</td>\n",
              "      <td>443</td>\n",
              "      <td>-0.069977</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>53</th>\n",
              "      <td>(16, 10, False)</td>\n",
              "      <td>-25055.0</td>\n",
              "      <td>40504</td>\n",
              "      <td>-0.618581</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>20</th>\n",
              "      <td>(17, 10, False)</td>\n",
              "      <td>-26973.0</td>\n",
              "      <td>42001</td>\n",
              "      <td>-0.642199</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>(19, 10, False)</td>\n",
              "      <td>-772.0</td>\n",
              "      <td>42349</td>\n",
              "      <td>-0.018229</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>(18, 10, False)</td>\n",
              "      <td>-28370.0</td>\n",
              "      <td>42523</td>\n",
              "      <td>-0.667168</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21</th>\n",
              "      <td>(20, 10, False)</td>\n",
              "      <td>24275.0</td>\n",
              "      <td>56179</td>\n",
              "      <td>0.432101</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>280 rows × 4 columns</p>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-43b7fd4c-7d7f-49f1-8be1-0335448a18f2')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-43b7fd4c-7d7f-49f1-8be1-0335448a18f2 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-43b7fd4c-7d7f-49f1-8be1-0335448a18f2');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-4bfb133f-b198-40ca-9f1b-bc6603bdae71\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-4bfb133f-b198-40ca-9f1b-bc6603bdae71')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-4bfb133f-b198-40ca-9f1b-bc6603bdae71 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "df.sort_values(by=['total_return'],ascending=False)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 423
        },
        "id": "GsQp4AV8iyG1",
        "outputId": "6bbed3df-515e-459f-ba54-df94d23acea9"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "              state  total_return      N     value\n",
              "21  (20, 10, False)       24275.0  56179  0.432101\n",
              "48  (21, 10, False)       23712.0  26693  0.888323\n",
              "6    (21, 10, True)       16794.0  18243  0.920572\n",
              "36   (20, 7, False)       11132.0  14245  0.781467\n",
              "63   (20, 8, False)       11105.0  14144  0.785139\n",
              "..              ...           ...    ...       ...\n",
              "25  (14, 10, False)      -21433.0  38349 -0.558893\n",
              "30  (15, 10, False)      -23248.0  39804 -0.584062\n",
              "53  (16, 10, False)      -25055.0  40504 -0.618581\n",
              "20  (17, 10, False)      -26973.0  42001 -0.642199\n",
              "14  (18, 10, False)      -28370.0  42523 -0.667168\n",
              "\n",
              "[280 rows x 4 columns]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-83cab67f-06a7-44cc-a2fa-30d859298b63\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>state</th>\n",
              "      <th>total_return</th>\n",
              "      <th>N</th>\n",
              "      <th>value</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>21</th>\n",
              "      <td>(20, 10, False)</td>\n",
              "      <td>24275.0</td>\n",
              "      <td>56179</td>\n",
              "      <td>0.432101</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>48</th>\n",
              "      <td>(21, 10, False)</td>\n",
              "      <td>23712.0</td>\n",
              "      <td>26693</td>\n",
              "      <td>0.888323</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>(21, 10, True)</td>\n",
              "      <td>16794.0</td>\n",
              "      <td>18243</td>\n",
              "      <td>0.920572</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>36</th>\n",
              "      <td>(20, 7, False)</td>\n",
              "      <td>11132.0</td>\n",
              "      <td>14245</td>\n",
              "      <td>0.781467</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>63</th>\n",
              "      <td>(20, 8, False)</td>\n",
              "      <td>11105.0</td>\n",
              "      <td>14144</td>\n",
              "      <td>0.785139</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25</th>\n",
              "      <td>(14, 10, False)</td>\n",
              "      <td>-21433.0</td>\n",
              "      <td>38349</td>\n",
              "      <td>-0.558893</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>30</th>\n",
              "      <td>(15, 10, False)</td>\n",
              "      <td>-23248.0</td>\n",
              "      <td>39804</td>\n",
              "      <td>-0.584062</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>53</th>\n",
              "      <td>(16, 10, False)</td>\n",
              "      <td>-25055.0</td>\n",
              "      <td>40504</td>\n",
              "      <td>-0.618581</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>20</th>\n",
              "      <td>(17, 10, False)</td>\n",
              "      <td>-26973.0</td>\n",
              "      <td>42001</td>\n",
              "      <td>-0.642199</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>(18, 10, False)</td>\n",
              "      <td>-28370.0</td>\n",
              "      <td>42523</td>\n",
              "      <td>-0.667168</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>280 rows × 4 columns</p>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-83cab67f-06a7-44cc-a2fa-30d859298b63')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-83cab67f-06a7-44cc-a2fa-30d859298b63 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-83cab67f-06a7-44cc-a2fa-30d859298b63');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-022c67b8-85e6-4609-893f-41c677676483\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-022c67b8-85e6-4609-893f-41c677676483')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-022c67b8-85e6-4609-893f-41c677676483 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#\n",
        "\n",
        "df[df['state']==(21,9,False)]['value'].values"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dFqBmtPbeFdx",
        "outputId": "7bf42a33-c903-4286-cd4d-8e7827f2959a"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([0.93848247])"
            ]
          },
          "metadata": {},
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "df[df['state']==(5,8,False)]['value'].values"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "NBzAXVHJeL9I",
        "outputId": "31d843a0-5bc4-44f3-a168-a696f3a42444"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([-0.30371204])"
            ]
          },
          "metadata": {},
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "#### Every visit v.s. First visit:\n",
        "\n",
        "Si noti: gli stati possono essere visitati più volte nello stesso episodio.\n",
        "Chiamiamo \"First-visit\", quando un certo stato si verifica in un singolo episodio per la prima visita.\n",
        "Nella stima della funzione valore di un certo stato,\n",
        "il metodo MC First-visit, considera solo la prima visita di uno stato per ricavare la media dei return. Il metodo MC Every-visit, invece, ad ogni visita (che puo essersi verificata due, tre, quattro ... volte) calcola la media dei return.\n",
        "\n",
        "Concretamente, nel codice MC First-visit presentato sopra, l'unica differenza è nel rigo :\n",
        "\n",
        "`if state not in states[0:t]:`\n",
        "\n",
        "che nel MC Every visit non è presente; per tutto il resto, i due codici sono identici.\n",
        "\n",
        "Per ulteriori approfondimenti:\n",
        "\n",
        "https://ai.stackexchange.com/questions/10812/what-is-the-difference-between-first-visit-monte-carlo-and-every-visit-monte-car\n",
        "\n",
        "*\n",
        "\n",
        "In genere :\n",
        "\n",
        "MC First-visit : Less bias, more variance\n",
        "\n",
        "MC Every-visit : More bias, less variance\n",
        "\n",
        "*\n",
        "\n",
        "Convervenza\n",
        "\n",
        "Entrambi,  First-visit MC e Every-visit MC,  convergono alla funzione valore $V^{\\pi}(s)$  quando il numero di visite (o prime visite) a $s$ tende all'infinito.  \n",
        "\n",
        "Nel caso First-visit MC, i return sono indipendenti e distribuiti in modo identico (independently and identically distributed, o IID: cioè ogni return ha la stessa distribuzione di probabilità, e le variabili sono indipendenti tra loro). Ogni media dei return è un'approssimazione dell \"expected value\", la deviazione standard del suo errore è pari a $1/√n$.\n",
        "\n",
        "Le stime di Every-visit MC convergono quadraticamente.\n"
      ],
      "metadata": {
        "id": "KTuCHRjldrdm"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Monte Carlo control"
      ],
      "metadata": {
        "id": "9hXCQy4_AMmv"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Quale azione selezionare?\n",
        "\n",
        "Problema:\n",
        "\n",
        "senza un'esplicita  funzione di transizione di stato (transition model) i valori della value fuction non sono sufficienti per determinare una \" policy milgiore \" di quella valutata in partenza (vedi: https://github.com/MarioFiorino/Tutorial-Reinforcement-Learning-ITA-Python/blob/main/Programmazione_Dinamica_ITA.ipynb)\n",
        "\n",
        "Soluzione:\n",
        "\n",
        "Per la fase di policy control, i metodi MC necessitano di lavorare con la funzione action-value : $Q^{\\pi}(s,a)$.\n",
        "\n",
        "Per quello che riguarda i metodi Every visit e First visit MC, nella fase di evaluation, usare la funzione action-value  $Q^{\\pi}(s,a)$ significa essenzialmete fare la stessa cosa che si è fatto con la value function; la diffrezna principale sta nel trattare visite di coppie stato-azione, e non di stati singoli. Una coppia stato-azione $(s,a)$ si verifica in un episodio quando lo stato $s$ viene visitato e l'azione $a$ viene intrapresa in esso. Questi metodi convergono quadraticamente, quando il numero di visite a ciascuna coppia stato-azione si avvicina all’infinito.\n",
        "\n",
        "\n",
        "Il principale problema che i metodi MC possono incontrare nella fase di evaluation, usando la funzione action value, è che *molte coppie $(s,a)$ potrebbero non essere mai visitate*. Ad esmepio, se $\\pi$ è una politica deterministica, allora si osserveranno return solo per una specifica azione per ogni stato; ignorando tutte le altre azioni. Le quali potrebbere potenzilmente produrre politiche ottimali. Quindi abbiamo bisogno di stimare il valore di tutte le azioni disponibili  (o per lo meno, quante più azioni possibili) di ogni stato.\n",
        "\n",
        "Allora, come garantire un’esplorazione continua?\n",
        "\n",
        "Una soluzione : *nella fase iniziale fare in modo che ogni coppia stato-azione $(s,a)$ abbia una probabilità maggiore di zero di essere selezionata come inizio.* Questo in linea teorica, garantisce che tutte le coppie stato-azione siano visitate un numero infinito di volte nel limite di un numero infinito di episodi. Chiamiamo questa assunzione : **exploring starts**.\n",
        "\n",
        "L'assunzione \"exploring starts\" è talvolta utile, ma ovviamente non è possibile fare affidamento su questa in generale, in particolare quando si apprende direttamente dall'interazione con un ambiente reale. In tal caso è improbabile che le condizioni di partenza siano così utili. L’approccio alternativo più comune per garantire che tutte le coppie stato-azione siano visitate è quello di considerare policy stocastiche, per le quali, in ogni stato, tutte le azioni disponibili hanno una probabilità maggiore di zero di essere selezionate."
      ],
      "metadata": {
        "id": "jQfi1DE_QOZ4"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "Nei metodi MC, la fase di policy improvement consiste nel rendere la policy \"greedy\" in base alla funzione action-value (l'utilizzo di tale funzione rende non necessario il modello di transizione di stato per la costruzione di una policy greedy come per la DP). In questo caso una policy greedy sceglie un'azione che massimizza $q(s,a)$, in formula:\n",
        "\n",
        "\n",
        "$\\pi_{greedy}(s) = \\arg \\max_\\limits{a} \\; q(s,a),  \\:\\; ∀s \\in S$\n",
        "\n",
        "\n",
        "$\\;$\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "id": "hQPz-nqL3Fxm"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Monte Carlo policy optimization\n",
        "\n",
        "\n",
        "Nel processo di policy iteration, cioè:\n",
        "\n",
        " $\\pi_{0}$ - policy evaluation ->\n",
        " $q_{\\pi_{0}}(s,a) $ - policy improvement -> $\\pi_{1}$ - policy evaluation -> $q_{\\pi_{1}}(s,a)$ - policy improvement -> $\\pi_{2}$ - ect ...\n",
        "\n",
        " che porta ad una policy ottimale, nei metodi MC sono necessarie due assunzioni per garantirne la convergenza :\n",
        " 1. supponiamo di poter osservare effettivamente un numero infinito di episodi\n",
        " 2. gli episodi siano generati in base alla \"exploring starts\"\n",
        "\n",
        "Per ottenere un algoritmo praticabile dovremo rimuovere entrambe le ipotesi.\n",
        "\n",
        "Concentriamoci sulla prima:\n",
        "\n",
        "la policy evaluation opera su un numero infinito di epsidi; questa ipotesi è relativamente facile da rimuovere. Infatti, anche i metodi DP classici, convergono solo asintoticamente alla vera funzione del valore. L'idea è di accontentarsi  di approssimazioni della $q_{\\pi_{k}}(s,a) $ in certi limiti di errore:\n",
        "\n",
        "cioè durante ciascuna fase di policy evaluation vengono adottate misure sufficienti per garantire che questi limiti siano rispettati. Questo approccio consente una corretta convergenza fino ad un certo livello di approssimazione. Tuttavia, tale solzione è probabile che richieda un numero eccessivo di episodi per essere utile nella pratica.\n",
        "\n",
        "Una alternativa può essere quella di porre un limite alla iterazioni nella fase di policy evaluation, e poi passare alla policy improvement. In pratica alla fine ogni fase di policy evaluation spostiamo di un pò il valore della funzione action-value $q$ \"in avanti\" , ma non ci aspettiamo di essere vicini effettivemente al valor ottimale, se non dopo molti cicli di policy evaluation<-->policy improvement.\n"
      ],
      "metadata": {
        "id": "IHAecST8ZqT-"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Monte Carlo with Exploring Starts\n",
        "\n",
        "Per la stima della policy ottimale.\n",
        "\n",
        "*\n",
        "\n",
        "Note sull'algoritmo :\n",
        "\n",
        "I  cicli policy evaluation e policy improvement si alternano \"episode-by-episode\" : dopo ogni episodio $k$: i returns campionati sono usati per ricavare $q_{\\pi_{k}}(s,a)$; subito dopo la policy $\\pi_{k}$ viene migliorata (greedy) in base ai valori di $q_{\\pi_{k}}(s,a)$ ottenuti"
      ],
      "metadata": {
        "id": "Ee5hjgXfW9_O"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Codice sottostante prende ispirazione\n",
        "# dal codice https://twice22.github.io/rl-part4/\n",
        "# e dal codice https://github.com/aditya1702/Machine-Learning-and-Data-Science/blob/master/Implementation%20of%20Reinforcement%20Learning%20Algorithms/Tensorflow%20Implementations/Monte%20Carlo%20Methods/Monte%20Carlo%20Control%20with%20Exploring%20Starts.ipynb\n",
        "\n",
        "\n",
        "\n",
        "def generate_episode_Exploring_Starts(policy):\n",
        "    state = env.reset()\n",
        "    start = True # flag per lo stato iniziale\n",
        "\n",
        "    episode = [] # Definiamo una lista che tiene traccia gli esiti della partita\n",
        "\n",
        "    while True:\n",
        "\n",
        "        # Condizione Exploring Starts:\n",
        "        # se lo stato è iniziale, genera un azione random\n",
        "        if start:\n",
        "             action = np.random.choice([0,1])\n",
        "             start=False\n",
        "        # Altrimenti, se lavori con un stato successivo a quello iniziale:\n",
        "        #   se lo stato corrente è compreso dalla policy che l'agente sta costruendo episodio per episodio,\n",
        "        #   allora seleziona l'azione in base a tale policy;\n",
        "        #   altrimenti, esegui una policy arbitraria iniziale che produce sempre l'azione = 1 , cioè \"hit\"\n",
        "        else :\n",
        "           action = policy[state] if state in policy else 1\n",
        "\n",
        "        next_state, reward, terminate, _ = env.step(action)\n",
        "        episode.append((state, action, reward))\n",
        "\n",
        "        if terminate:\n",
        "            break\n",
        "\n",
        "        state = next_state\n",
        "\n",
        "    return episode\n"
      ],
      "metadata": {
        "id": "8w6xrb4AJfU_"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "#### Esplicitazione tecnica del codice usato"
      ],
      "metadata": {
        "id": "dqhBayidTm0P"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Fase di inizializzazione delle strutture dati utilizzate\n",
        "policy = {}\n",
        "\n",
        "Q_table = defaultdict(lambda: np.zeros(env.action_space.n)) #  Note : lambda: np.zeros(2) = [0. 0.], lambda da in unscita un array di zeri a due elementi\n",
        "# La Q_table ha due valori per ogni stato : una per l'azione stick, altra per l'azione hit\n",
        "# Esempio : Q_table:  {(14, 6, False): [-1.,  5.]}\n",
        "# per lo state : (14, 6, False), per azione(stick) = 0 ha valore = -1.; per hit il valore di q = 5.\n",
        "\n",
        "\n",
        "returns_sum = defaultdict(float) # Nota:  returns(state, action)\n",
        "returns_count = defaultdict(int)\n",
        "\n",
        "# Parametri\n",
        "gamma = 1\n",
        "\n"
      ],
      "metadata": {
        "id": "xt8e_xR3fKkq"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "#\n",
        "# Da riterarsi più volte con i trajectory diferenti\n",
        "#\n",
        "\n",
        "#trajectory = generate_episode_Exploring_Starts(policy)\n",
        "\n",
        "###campiomaneti giudati, di fantasia ma servono a capire il codice\n",
        "\n",
        "#trajectory = [((7, 6, False), 1, 0.0),((7, 6, False), 0, 0.0),((7, 6, False), 1, 0.0),((14,6 , False), 0, -1.0)]\n",
        "\n",
        "#trajectory = [((7,6 , False), 1, 97.0)]\n",
        "\n",
        "#trajectory = [((7,6 , False), 1, -1.0),((11, 4, False), 0, -11.0),((11, 4, False), 0, 5.0),((7,6 , False), 1, -60.0)]\n",
        "\n",
        "#trajectory = [((17, 3, False), 1, 0.0), ((20, 3, False), 1, 0.0), ((21, 3, False), 1, -1.0)]\n",
        "\n",
        "trajectory = [((17, 3, False), 1, -5.0), ((17, 3, False), 1, 10.0), ((21, 3, False), 1, 5.0)]\n",
        "\n",
        "G, T = 0, len(trajectory)\n",
        "print(trajectory)\n",
        "visited_state = []\n",
        "for i in range(T):\n",
        " #print(trajectory[i][0])\n",
        " visited_state.append((trajectory[i][0],trajectory[i][1]))\n",
        "print(\"visited_state : \", visited_state)\n",
        "\n",
        "for i in range(T-1, -1, -1):\n",
        "       print(\"\\n  i:\",i)\n",
        "       state, action, reward = trajectory[i]\n",
        "       G += gamma**i * reward # ha solo un senso formale, di fatto, in questo caso, G = reward\n",
        "       print(\"  G : \",G)\n",
        "\n",
        "       visited_state.pop()\n",
        "       print(\"  ^ visited_state ^ 2: \", visited_state)\n",
        "       print(\"\\t (state, action) not in visited_state : \", (state, action) not in visited_state)\n",
        "       if (state, action) not in visited_state: # A meno che la coppia (state, action) non sia stata già visitata in questo episodio, fai:\n",
        "\n",
        "                #Per ricavare la media dei Return\n",
        "                returns_sum[(state, action)] += G\n",
        "                returns_count[(state, action)] += 1\n",
        "                print(\"returns_sum = \",dict(returns_sum)) # to print defaultdict variable without its type\n",
        "                print(\"returns_cont = \",dict(returns_count))\n",
        "\n",
        "                Q_table[state][action] = returns_sum[(state, action)] / returns_count[(state, action)]\n",
        "                print(f\"Q_table[{state}][{action}] = returns_sum[(state, action)] / returns_count[(state, action) : \", Q_table[state][action])\n",
        "                print(\"Q_table\", dict(Q_table))\n",
        "\n",
        "                policy[state] = np.argmax(Q_table[state]) # update policy greedily\n",
        "                print(f\" Azione greedy della policy[{state}] : \",policy[state])\n",
        "                print(\"policy aggiornata :\", policy)\n"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "93b9b020-889f-4f09-919f-3d36ae9d7b4e",
        "id": "X7yDSA62CZiC"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[((17, 3, False), 1, -5.0), ((17, 3, False), 1, 10.0), ((21, 3, False), 1, 5.0)]\n",
            "visited_state :  [((17, 3, False), 1), ((17, 3, False), 1), ((21, 3, False), 1)]\n",
            "\n",
            "  i: 2\n",
            "  G :  5.0\n",
            "  ^ visited_state ^ 2:  [((17, 3, False), 1), ((17, 3, False), 1)]\n",
            "\t (state, action) not in visited_state :  True\n",
            "returns_sum =  {((14, 6, False), 0): -1.0, ((7, 6, False), 0): -1.0, ((7, 6, False), 1): 29.0, ((11, 4, False), 0): -66.0, ((21, 3, False), 1): 4.0, ((20, 3, False), 1): -1.0, ((17, 3, False), 1): -1.0}\n",
            "returns_cont =  {((14, 6, False), 0): 1, ((7, 6, False), 0): 1, ((7, 6, False), 1): 3, ((11, 4, False), 0): 1, ((21, 3, False), 1): 2, ((20, 3, False), 1): 1, ((17, 3, False), 1): 1}\n",
            "Q_table[(21, 3, False)][1] = returns_sum[(state, action)] / returns_count[(state, action) :  2.0\n",
            "Q_table {(14, 6, False): array([-1.,  0.]), (7, 6, False): array([-1.        ,  9.66666667]), (11, 4, False): array([-66.,   0.]), (21, 3, False): array([0., 2.]), (20, 3, False): array([ 0., -1.]), (17, 3, False): array([ 0., -1.])}\n",
            " Azione greedy della policy[(21, 3, False)] :  1\n",
            "policy aggiornata : {(14, 6, False): 1, (7, 6, False): 1, (11, 4, False): 1, (21, 3, False): 1, (20, 3, False): 0, (17, 3, False): 0}\n",
            "\n",
            "  i: 1\n",
            "  G :  15.0\n",
            "  ^ visited_state ^ 2:  [((17, 3, False), 1)]\n",
            "\t (state, action) not in visited_state :  False\n",
            "\n",
            "  i: 0\n",
            "  G :  10.0\n",
            "  ^ visited_state ^ 2:  []\n",
            "\t (state, action) not in visited_state :  True\n",
            "returns_sum =  {((14, 6, False), 0): -1.0, ((7, 6, False), 0): -1.0, ((7, 6, False), 1): 29.0, ((11, 4, False), 0): -66.0, ((21, 3, False), 1): 4.0, ((20, 3, False), 1): -1.0, ((17, 3, False), 1): 9.0}\n",
            "returns_cont =  {((14, 6, False), 0): 1, ((7, 6, False), 0): 1, ((7, 6, False), 1): 3, ((11, 4, False), 0): 1, ((21, 3, False), 1): 2, ((20, 3, False), 1): 1, ((17, 3, False), 1): 2}\n",
            "Q_table[(17, 3, False)][1] = returns_sum[(state, action)] / returns_count[(state, action) :  4.5\n",
            "Q_table {(14, 6, False): array([-1.,  0.]), (7, 6, False): array([-1.        ,  9.66666667]), (11, 4, False): array([-66.,   0.]), (21, 3, False): array([0., 2.]), (20, 3, False): array([ 0., -1.]), (17, 3, False): array([0. , 4.5])}\n",
            " Azione greedy della policy[(17, 3, False)] :  1\n",
            "policy aggiornata : {(14, 6, False): 1, (7, 6, False): 1, (11, 4, False): 1, (21, 3, False): 1, (20, 3, False): 0, (17, 3, False): 1}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "#### Training Loop Monte Carlo Exploring Start"
      ],
      "metadata": {
        "id": "uPN_4MdIWo21"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Fase di inizializzazione delle strutture dati utilizzate\n",
        "policy = {}\n",
        "\n",
        "Q_table = defaultdict(lambda: np.zeros(env.action_space.n))\n",
        "\n",
        "returns_sum = defaultdict(float) # Nota:  returns(state, action)\n",
        "returns_count = defaultdict(int)\n",
        "\n",
        "# Parametri\n",
        "gamma = 1\n",
        "num_episodes = 500_000\n",
        "\n",
        "for episode in range(num_episodes):\n",
        "  trajectory = generate_episode_Exploring_Starts(policy)\n",
        "  G, T = 0, len(trajectory)\n",
        "\n",
        "  visited_state = []\n",
        "  for i in range(T):\n",
        "    visited_state.append((trajectory[i][0],trajectory[i][1]))\n",
        "\n",
        "  for i in range(T-1, -1, -1):\n",
        "       state, action, reward = trajectory[i]\n",
        "       G += gamma**i * reward # ha solo un senso formale, di fatto, in questo caso, G += reward\n",
        "\n",
        "       visited_state.pop()\n",
        "       if (state, action) not in visited_state: # A meno che la coppia (state, action) non sia stata già visitata in questo episodio, fai:\n",
        "\n",
        "                returns_sum[(state, action)] += G\n",
        "                returns_count[(state, action)] += 1\n",
        "\n",
        "                Q_table[state][action] = returns_sum[(state, action)] / returns_count[(state, action)]\n",
        "\n",
        "                policy[state] = np.argmax(Q_table[state]) # update policy greedily\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "id": "At4IIII8VZyh"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "##### Analisi dei dati di training"
      ],
      "metadata": {
        "id": "qsaU_lvzXui-"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "print(len(policy))\n",
        "print(len(Q_table))\n",
        "\n",
        "# Converti il dizionario total_returns in un dataframe pandas:\n",
        "Q_table_pd = pd.DataFrame(Q_table.items(),columns=['state','action'])\n",
        "\n",
        "# Converti il dizionario N in un dataframe pandas:\n",
        "policy_pd = pd.DataFrame(policy.items(),columns=['state', 'action'])\n",
        "\n",
        "\n",
        "policy_pd.head(5)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 241
        },
        "id": "vV7SltVfblHf",
        "outputId": "3e1868ed-8060-4f63-a6cc-e9b4241780bd"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "280\n",
            "280\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "             state  action\n",
              "0   (20, 9, False)       0\n",
              "1    (7, 5, False)       1\n",
              "2  (13, 10, False)       1\n",
              "3    (17, 9, True)       1\n",
              "4  (20, 10, False)       0"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-6b510c9d-901e-49ec-9d87-b67b00d0c2fc\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>state</th>\n",
              "      <th>action</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>(20, 9, False)</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>(7, 5, False)</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>(13, 10, False)</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>(17, 9, True)</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>(20, 10, False)</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-6b510c9d-901e-49ec-9d87-b67b00d0c2fc')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-6b510c9d-901e-49ec-9d87-b67b00d0c2fc button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-6b510c9d-901e-49ec-9d87-b67b00d0c2fc');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-328263f4-2c10-496e-9a15-be4b944ceee4\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-328263f4-2c10-496e-9a15-be4b944ceee4')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-328263f4-2c10-496e-9a15-be4b944ceee4 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 98
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "Q_table_pd.head(5)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "hm9fBT8clACX",
        "outputId": "84ce17ca-1b9d-4754-ca96-afa389f8c8ac"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "             state                                        action\n",
              "0   (20, 9, False)     [0.7618500430910657, -0.8630681818181818]\n",
              "1    (7, 5, False)  [-0.13023255813953488, 0.035940803382663845]\n",
              "2  (13, 10, False)    [-0.5690338311490399, -0.4634963862993611]\n",
              "3    (17, 9, True)  [-0.45136186770428016, -0.23595505617977527]\n",
              "4  (20, 10, False)     [0.4326483189310871, -0.8565229563269877]"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-e351429d-76a9-4397-8360-5a8fc7b622d1\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>state</th>\n",
              "      <th>action</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>(20, 9, False)</td>\n",
              "      <td>[0.7618500430910657, -0.8630681818181818]</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>(7, 5, False)</td>\n",
              "      <td>[-0.13023255813953488, 0.035940803382663845]</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>(13, 10, False)</td>\n",
              "      <td>[-0.5690338311490399, -0.4634963862993611]</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>(17, 9, True)</td>\n",
              "      <td>[-0.45136186770428016, -0.23595505617977527]</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>(20, 10, False)</td>\n",
              "      <td>[0.4326483189310871, -0.8565229563269877]</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-e351429d-76a9-4397-8360-5a8fc7b622d1')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-e351429d-76a9-4397-8360-5a8fc7b622d1 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-e351429d-76a9-4397-8360-5a8fc7b622d1');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-956999cc-1b58-493e-8f4e-9506505105e9\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-956999cc-1b58-493e-8f4e-9506505105e9')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-956999cc-1b58-493e-8f4e-9506505105e9 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 99
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Crea una funzione valore dalla action-value function q, selezionando l'azione migliore in ogni stato\n",
        "\n",
        "V = defaultdict(float)\n",
        "for state, actions in Q_table.items():\n",
        "    action_value = np.max(actions)\n",
        "    V[state] = action_value\n"
      ],
      "metadata": {
        "id": "41vGZoPOXOey"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "V_pd = pd.DataFrame(V.items(),columns=['state','Value'])\n",
        "V_pd = V_pd.sort_values(by=['Value'],ascending=False)\n",
        "V_pd.head(5)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        },
        "id": "mTyyoJXCoAfy",
        "outputId": "a3360fe5-321e-46c7-9d51-7db7e2f0291e"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "             state     Value\n",
              "165  (21, 9, True)  0.992091\n",
              "181  (21, 7, True)  0.988304\n",
              "98   (21, 8, True)  0.985087\n",
              "112  (21, 2, True)  0.980886\n",
              "173  (21, 5, True)  0.976974"
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-0553b672-629d-49c6-8bb1-9e5ee6bc4fa8\" class=\"colab-df-container\">\n",
              "    <div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>state</th>\n",
              "      <th>Value</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>165</th>\n",
              "      <td>(21, 9, True)</td>\n",
              "      <td>0.992091</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>181</th>\n",
              "      <td>(21, 7, True)</td>\n",
              "      <td>0.988304</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>98</th>\n",
              "      <td>(21, 8, True)</td>\n",
              "      <td>0.985087</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>112</th>\n",
              "      <td>(21, 2, True)</td>\n",
              "      <td>0.980886</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>173</th>\n",
              "      <td>(21, 5, True)</td>\n",
              "      <td>0.976974</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "    <div class=\"colab-df-buttons\">\n",
              "\n",
              "  <div class=\"colab-df-container\">\n",
              "    <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-0553b672-629d-49c6-8bb1-9e5ee6bc4fa8')\"\n",
              "            title=\"Convert this dataframe to an interactive table.\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\" viewBox=\"0 -960 960 960\">\n",
              "    <path d=\"M120-120v-720h720v720H120Zm60-500h600v-160H180v160Zm220 220h160v-160H400v160Zm0 220h160v-160H400v160ZM180-400h160v-160H180v160Zm440 0h160v-160H620v160ZM180-180h160v-160H180v160Zm440 0h160v-160H620v160Z\"/>\n",
              "  </svg>\n",
              "    </button>\n",
              "\n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    .colab-df-buttons div {\n",
              "      margin-bottom: 4px;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "    <script>\n",
              "      const buttonEl =\n",
              "        document.querySelector('#df-0553b672-629d-49c6-8bb1-9e5ee6bc4fa8 button.colab-df-convert');\n",
              "      buttonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "      async function convertToInteractive(key) {\n",
              "        const element = document.querySelector('#df-0553b672-629d-49c6-8bb1-9e5ee6bc4fa8');\n",
              "        const dataTable =\n",
              "          await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                    [key], {});\n",
              "        if (!dataTable) return;\n",
              "\n",
              "        const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "          '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "          + ' to learn more about interactive tables.';\n",
              "        element.innerHTML = '';\n",
              "        dataTable['output_type'] = 'display_data';\n",
              "        await google.colab.output.renderOutput(dataTable, element);\n",
              "        const docLink = document.createElement('div');\n",
              "        docLink.innerHTML = docLinkHtml;\n",
              "        element.appendChild(docLink);\n",
              "      }\n",
              "    </script>\n",
              "  </div>\n",
              "\n",
              "\n",
              "<div id=\"df-03941c64-7f62-4d33-82e1-480fc483cb86\">\n",
              "  <button class=\"colab-df-quickchart\" onclick=\"quickchart('df-03941c64-7f62-4d33-82e1-480fc483cb86')\"\n",
              "            title=\"Suggest charts\"\n",
              "            style=\"display:none;\">\n",
              "\n",
              "<svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "     width=\"24px\">\n",
              "    <g>\n",
              "        <path d=\"M19 3H5c-1.1 0-2 .9-2 2v14c0 1.1.9 2 2 2h14c1.1 0 2-.9 2-2V5c0-1.1-.9-2-2-2zM9 17H7v-7h2v7zm4 0h-2V7h2v10zm4 0h-2v-4h2v4z\"/>\n",
              "    </g>\n",
              "</svg>\n",
              "  </button>\n",
              "\n",
              "<style>\n",
              "  .colab-df-quickchart {\n",
              "      --bg-color: #E8F0FE;\n",
              "      --fill-color: #1967D2;\n",
              "      --hover-bg-color: #E2EBFA;\n",
              "      --hover-fill-color: #174EA6;\n",
              "      --disabled-fill-color: #AAA;\n",
              "      --disabled-bg-color: #DDD;\n",
              "  }\n",
              "\n",
              "  [theme=dark] .colab-df-quickchart {\n",
              "      --bg-color: #3B4455;\n",
              "      --fill-color: #D2E3FC;\n",
              "      --hover-bg-color: #434B5C;\n",
              "      --hover-fill-color: #FFFFFF;\n",
              "      --disabled-bg-color: #3B4455;\n",
              "      --disabled-fill-color: #666;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart {\n",
              "    background-color: var(--bg-color);\n",
              "    border: none;\n",
              "    border-radius: 50%;\n",
              "    cursor: pointer;\n",
              "    display: none;\n",
              "    fill: var(--fill-color);\n",
              "    height: 32px;\n",
              "    padding: 0;\n",
              "    width: 32px;\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart:hover {\n",
              "    background-color: var(--hover-bg-color);\n",
              "    box-shadow: 0 1px 2px rgba(60, 64, 67, 0.3), 0 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "    fill: var(--button-hover-fill-color);\n",
              "  }\n",
              "\n",
              "  .colab-df-quickchart-complete:disabled,\n",
              "  .colab-df-quickchart-complete:disabled:hover {\n",
              "    background-color: var(--disabled-bg-color);\n",
              "    fill: var(--disabled-fill-color);\n",
              "    box-shadow: none;\n",
              "  }\n",
              "\n",
              "  .colab-df-spinner {\n",
              "    border: 2px solid var(--fill-color);\n",
              "    border-color: transparent;\n",
              "    border-bottom-color: var(--fill-color);\n",
              "    animation:\n",
              "      spin 1s steps(1) infinite;\n",
              "  }\n",
              "\n",
              "  @keyframes spin {\n",
              "    0% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "      border-left-color: var(--fill-color);\n",
              "    }\n",
              "    20% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    30% {\n",
              "      border-color: transparent;\n",
              "      border-left-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    40% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-top-color: var(--fill-color);\n",
              "    }\n",
              "    60% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "    }\n",
              "    80% {\n",
              "      border-color: transparent;\n",
              "      border-right-color: var(--fill-color);\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "    90% {\n",
              "      border-color: transparent;\n",
              "      border-bottom-color: var(--fill-color);\n",
              "    }\n",
              "  }\n",
              "</style>\n",
              "\n",
              "  <script>\n",
              "    async function quickchart(key) {\n",
              "      const quickchartButtonEl =\n",
              "        document.querySelector('#' + key + ' button');\n",
              "      quickchartButtonEl.disabled = true;  // To prevent multiple clicks.\n",
              "      quickchartButtonEl.classList.add('colab-df-spinner');\n",
              "      try {\n",
              "        const charts = await google.colab.kernel.invokeFunction(\n",
              "            'suggestCharts', [key], {});\n",
              "      } catch (error) {\n",
              "        console.error('Error during call to suggestCharts:', error);\n",
              "      }\n",
              "      quickchartButtonEl.classList.remove('colab-df-spinner');\n",
              "      quickchartButtonEl.classList.add('colab-df-quickchart-complete');\n",
              "    }\n",
              "    (() => {\n",
              "      let quickchartButtonEl =\n",
              "        document.querySelector('#df-03941c64-7f62-4d33-82e1-480fc483cb86 button');\n",
              "      quickchartButtonEl.style.display =\n",
              "        google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "    })();\n",
              "  </script>\n",
              "</div>\n",
              "    </div>\n",
              "  </div>\n"
            ]
          },
          "metadata": {},
          "execution_count": 101
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_policy(policy):\n",
        "\n",
        "    def get_Z(player_hand, dealer_showing, usable_ace):\n",
        "        if (player_hand, dealer_showing, usable_ace) in policy:\n",
        "            return policy[player_hand, dealer_showing, usable_ace]\n",
        "        else:\n",
        "            return 1\n",
        "\n",
        "    def get_figure(usable_ace, ax):\n",
        "        x_range = np.arange(1, 11)\n",
        "        y_range = np.arange(11, 22)\n",
        "        X, Y = np.meshgrid(x_range, y_range)\n",
        "        Z = np.array([[get_Z(player_hand, dealer_showing, usable_ace) for dealer_showing in x_range] for player_hand in range(21, 10, -1)])\n",
        "        surf = ax.imshow(Z, cmap=plt.get_cmap('Pastel2', 2), vmin=0, vmax=1, extent=[1, 11, 11, 22])\n",
        "        plt.xticks(x_range, ('A', '2', '3', '4', '5', '6', '7', '8', '9', '10'))\n",
        "        plt.yticks(y_range)\n",
        "        ax.set_xlabel('Dealer Showing')\n",
        "        ax.set_ylabel('Player Hand')\n",
        "        ax.grid(color='black', linestyle='-', linewidth=1)\n",
        "        divider = make_axes_locatable(ax)\n",
        "        cax = divider.append_axes(\"right\", size=\"5%\", pad=0.1)\n",
        "        cbar = plt.colorbar(surf, ticks=[0, 1], cax=cax)\n",
        "        cbar.ax.set_yticklabels(['0 (STICK)','1 (HIT)'])\n",
        "        cbar.ax.invert_yaxis()\n",
        "\n",
        "    fig = plt.figure(figsize=(8, 8))\n",
        "    ax = fig.add_subplot(121)\n",
        "    ax.set_title('Usable Ace', fontsize=12)\n",
        "    get_figure(True,ax)\n",
        "    ax = fig.add_subplot(122)\n",
        "    ax.set_title('No Usable Ace', fontsize=12)\n",
        "    get_figure(False, ax)\n",
        "    plt.show()"
      ],
      "metadata": {
        "id": "SsldSwUYsg7t"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plot_policy(policy)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 387
        },
        "id": "pK72wu_2eTkt",
        "outputId": "99087412-4a2e-48e1-ad8c-7e2bb6e0d8ec"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 800x800 with 4 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def test_optimal_policy(policy, numero_partite=1_000):\n",
        "\n",
        "    tot_rew = 0\n",
        "    tot_lost = 0\n",
        "    state = env.reset()\n",
        "\n",
        "    for _ in range(numero_partite):\n",
        "\n",
        "        done = False\n",
        "        while not done:\n",
        "            next_state, reward, done,_ = env.step(policy[state])\n",
        "\n",
        "            state = next_state\n",
        "            if reward > 0 :\n",
        "               tot_rew += reward\n",
        "\n",
        "            if reward < 0 :\n",
        "               tot_lost += reward\n",
        "\n",
        "            if done:\n",
        "                state = env.reset()\n",
        "\n",
        "\n",
        "    print(\"Agente RL_MC\")\n",
        "    print(\"Sono state vinte : \",int(tot_rew),\" su \", numero_partite, \"partite giocate\")\n",
        "    print(\"Sono state perse : \",int(tot_lost),\". Premio accumualto Tot : \", (tot_rew+tot_lost))\n",
        "\n",
        "\n",
        "def test_optimal_random_policy(numero_partite=1_000):\n",
        "\n",
        "    tot_rew = 0\n",
        "    tot_lost = 0\n",
        "    state = env.reset()\n",
        "\n",
        "    for _ in range(numero_partite):\n",
        "\n",
        "        done = False\n",
        "        while not done:\n",
        "            next_state, reward, done,_ = env.step(np.random.choice([0,1]))\n",
        "\n",
        "            state = next_state\n",
        "            if reward > 0 :\n",
        "               tot_rew += reward\n",
        "\n",
        "            if reward < 0 :\n",
        "               tot_lost += reward\n",
        "\n",
        "            if done:\n",
        "                state = env.reset()\n",
        "\n",
        "\n",
        "    print(\"\\nAgente Random\")\n",
        "    print(\"Sono state vinte : \",int(tot_rew),\" su \", numero_partite, \"partite giocate\")\n",
        "    print(\"Sono state perse : \",int(tot_lost),\". Premio accumualto Tot : \", (tot_rew+tot_lost))\n",
        "\n",
        "\n",
        "test_optimal_policy(policy)\n",
        "\n",
        "test_optimal_random_policy(numero_partite=1_000)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uFIwFXWxNQPx",
        "outputId": "d1dcbde6-87d6-4f4d-fa31-e8932f0c0f06"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Agente RL_MC\n",
            "Sono state vinte :  433  su  1000 partite giocate\n",
            "Sono state perse :  -482 . Premio accumualto Tot :  -49.0\n",
            "\n",
            "Agente Random\n",
            "Sono state vinte :  310  su  1000 partite giocate\n",
            "Sono state perse :  -645 . Premio accumualto Tot :  -335.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "def plot_value_function(V, title = \"Value Function\"):\n",
        "\n",
        "    min_x = 11 # min(k[0] for k in V.keys())\n",
        "    max_x = max(k[0] for k in V.keys())\n",
        "    min_y = min(k[1] for k in V.keys())\n",
        "    max_y = max(k[1] for k in V.keys())\n",
        "\n",
        "    x_range = np.arange(min_x, max_x + 1)\n",
        "    y_range = np.arange(min_y, max_y + 1)\n",
        "    X, Y = np.meshgrid(x_range, y_range)\n",
        "\n",
        "    # Find value for all (x, y) coordinates\n",
        "    Z_noace = np.apply_along_axis(lambda _: V[(_[0], _[1], False)], 2, np.dstack([X, Y]))\n",
        "    Z_ace = np.apply_along_axis(lambda _: V[(_[0], _[1], True)], 2, np.dstack([X, Y]))\n",
        "\n",
        "    def plot_surface(X, Y, Z, title):\n",
        "        fig = plt.figure(figsize = (9, 6))\n",
        "        ax = fig.add_subplot(111, projection = '3d')\n",
        "        surf = ax.plot_surface(X, Y, Z, rstride = 1, cstride = 1,\n",
        "                               cmap = matplotlib.cm.coolwarm, vmin = -1.0, vmax = 1.0)\n",
        "        ax.set_xlabel('Player Sum')\n",
        "        ax.set_ylabel('Dealer Showing')\n",
        "        ax.set_zlabel('Value')\n",
        "        ax.set_title(title)\n",
        "        ax.view_init(ax.elev, -120)\n",
        "        fig.colorbar(surf)\n",
        "        plt.show()\n",
        "\n",
        "    plot_surface(X, Y, Z_noace, \"{} (No Usable Ace)\".format(title))\n",
        "    plot_surface(X, Y, Z_ace, \"{} (Usable Ace)\".format(title))"
      ],
      "metadata": {
        "id": "mS0F1BGysc8D"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plot_value_function(V, title=\"Optimal Value Function\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "id": "UWDVRYgPYpzF",
        "outputId": "ab682ea1-b825-4f4c-e589-2f95cb7befff"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 900x600 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 900x600 with 2 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "\n",
        "Nota :\n",
        "\n",
        "I grafici della politica ottimale e della funzione di valore trovati sono molto simili al quelli esposti nel testo di Reinforcement Learning di Sutton e Barto, Second Edition, pag PDF : 122"
      ],
      "metadata": {
        "id": "rHzkYiF-tH8R"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "\n",
        "\n",
        "###  Monte Carlo Control without Exploring starts\n",
        "\n",
        "Fare i conti con seconda assunzione, ovvero come evitare l'improbabile assunzione Exploring starts?\n",
        "\n",
        "L'unico modo generale per assicurarsi che tutte le azioni siano selezionate infinite volte è che l'agente continui a selezionarle durante il training.\n",
        "\n",
        "Due approcci di fondo:\n",
        "-  Metodi on-policy: per valutare o migliorare la politica utilizzata per prendere decisioni (e produrre dati)\n",
        "-  Metodi off-policy: valutare o migliorare una politica diversa da quella utilizzata per generare i dati.\n",
        "\n",
        "\n",
        "$\\;$\n",
        "\n",
        "Nei metodi di controllo on-policy la policy usata è generalmente \"soft\".\n",
        "\n",
        "Una **soft-policy** è una policy che fa sì che per ogni stato, esista una probabilità positiva (anche minima) di selezionare ciascuna delle azioni disponibili (praticamente in ogni stato esiste sempre la possibilità di selezionare una qualsiasi azione).\n",
        "\n",
        "\n",
        "\n",
        "**ϵ-greedy policy**\n",
        "\n",
        "Premessa: la caridinalità $|Actions(state)|$  indica il numero di azioni totali disponibili in un certo stato.\n",
        "\n",
        "Le ϵ-greedy policy sono policy che selezionano:\n",
        "\n",
        "\n",
        "- con la probabilità: $1 -ϵ + \\frac{ϵ}{|Action(state)|} $,  l'azione greedy che massimizza la funzione costo (ad esempio $q_{\\pi}(s,a)$) in un certo stato.\n",
        "\n",
        "- con la probabilità $\\frac{ϵ}{|Actions(state)|}$ , una delle azioni non-greedy disponibili in un certo stato;  al fine di esplorare nuove possibilità.\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "**L' ϵ-greedy-policy sono un esempio di soft-policy**, in quanto in ogni stato, si ha la probabilità minima: $\\frac{ϵ}{|Actions(state)|}$ , di selezionare una qualsiasi azione disponibile.\n",
        "\n",
        "Spesso, per semplificare il formalismo, e quando l'ambiente in cui si svolge il training lo consente, l'implementazione di tale policy si riduce a definire con $1-ϵ$ la probabilità di selezionare l'azione greedy e con $ϵ$ la probabilità di selezionare una delle azioni non-greedy.\n",
        "\n",
        "\n",
        "\n",
        "$\\;$\n",
        "\n",
        "Convergenza:\n",
        "\n",
        "Il teorema \"Policy improvement\" garantisce che qualsiasi politica ϵ-greedy rispetto a $q_{\\pi}(s,a)$, sia un miglioramento rispetto a qualsiasi politica ϵ-soft di $\\pi$. Per dimostrazione vedi pag PDF 123, Sutton e Barto.\n",
        "\n",
        "\n",
        "\n"
      ],
      "metadata": {
        "id": "BbIom_QbWW42"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "#### On-policy first-visit MC control (using ϵ-soft policies)\n",
        "\n",
        "Per la stima della policy ottimale."
      ],
      "metadata": {
        "id": "Eoo1bDMDdZeE"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def generate_episode_soft_policy(s_pi):\n",
        "    state = env.reset()\n",
        "\n",
        "    episode = [] # Definiamo una lista che tiene traccia gli esiti della partita\n",
        "\n",
        "    while True:\n",
        "\n",
        "        action =  1 # sceglie sempre azione 1, ma se ...\n",
        "\n",
        "        if state in s_pi:\n",
        "            probs = s_pi[state]\n",
        "            #print(f\"\\tprobs = s_pi[{state}] = \",probs)\n",
        "            action = np.random.choice(np.arange(env.action_space.n), p=probs)\n",
        "            # Note Tecniche:\n",
        "            # 1. np.arange(4)  produce un array del tipo [0, 1, 2, 3]\n",
        "            #    nel nostro caso : np.arange(env.action_space.n) = [0 1]\n",
        "            # 2. p è la probabilità associata a ciascun elemento nell'array. Se non fornito, il campione presuppone una distribuzione uniforme su tutte le voci.\n",
        "            # Nel nostro caso, la probabilità con cui np.random.choice sceglie non è uniforme, ma definita da s_pi[state]\n",
        "            #\n",
        "            #print(\"\\taction =\", action )\n",
        "        next_state, reward, terminate, _ = env.step(action)\n",
        "        episode.append((state, action, reward))\n",
        "\n",
        "        if terminate:\n",
        "            break\n",
        "\n",
        "        state = next_state\n",
        "\n",
        "    return episode"
      ],
      "metadata": {
        "id": "wFtrUsRtdW84"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "##### Test delle parti nuove del codice"
      ],
      "metadata": {
        "id": "9MtJscJPwBvQ"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "#\n",
        "#Test\n",
        "#\n",
        "n_actions = env.action_space.n\n",
        "#Q_table = defaultdict(lambda: np.zeros(n_actions))\n",
        "EPS = 0.6\n",
        "\n",
        "e_policy = defaultdict(lambda: np.zeros(n_actions))\n",
        "e_policy[(14, 6, False)][0] = 0.8\n",
        "e_policy[(14, 6, False)][1] = 0.2\n",
        "print(\"\\tANTE e_policy = \", dict(e_policy))\n",
        "\n",
        "#e = generate_episode_soft_policy(e_policy)\n",
        "\n",
        "e = [((14, 6, False), 1, -5.0), ((21, 3, False), 0, 10.0),((11, 4, False), 0, 10.0) ]\n",
        "\n",
        "Q_table = {(14, 6, False): [-1.,  0.], (11, 4, False): [-66.,   0.], (21, 3, False): [0.1, 0.]}\n",
        "print(\"e =\",e)\n",
        "print(\"Q_table=\", Q_table)\n",
        "print(\"\\n\")\n",
        "\n",
        "for i in range((len(e))-1, -1, -1):\n",
        "      state, action, reward = e[i]\n",
        "      best_action = np.argmax(Q_table[state])\n",
        "      print(i)\n",
        "      print(\"Q best_action =\",best_action)\n",
        "\n",
        "      for a_index in np.arange(n_actions):\n",
        "         if a_index == best_action:\n",
        "                  e_policy[state][a_index] = 1 - EPS + (EPS/n_actions)\n",
        "                  print(f\"e_policy[{state}][{a_index}] =  \", e_policy[state][a_index] )\n",
        "         else:\n",
        "                  e_policy[state][a_index] = EPS/n_actions\n",
        "                  print(f\"e_policy[{state}][{a_index}] = \", e_policy[state][a_index] )\n",
        "\n",
        "print(\"\\tPOST e_policy = \", e_policy)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "z32UWSReq8T_",
        "outputId": "54b01c32-0ad7-4baf-a364-ced51e4fdebc"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\tANTE e_policy =  {(14, 6, False): array([0.8, 0.2])}\n",
            "e = [((14, 6, False), 1, -5.0), ((21, 3, False), 0, 10.0), ((11, 4, False), 0, 10.0)]\n",
            "Q_table= {(14, 6, False): [-1.0, 0.0], (11, 4, False): [-66.0, 0.0], (21, 3, False): [0.1, 0.0]}\n",
            "\n",
            "\n",
            "2\n",
            "Q best_action = 1\n",
            "e_policy[(11, 4, False)][0] =  0.3\n",
            "e_policy[(11, 4, False)][1] =   0.7\n",
            "1\n",
            "Q best_action = 0\n",
            "e_policy[(21, 3, False)][0] =   0.7\n",
            "e_policy[(21, 3, False)][1] =  0.3\n",
            "0\n",
            "Q best_action = 1\n",
            "e_policy[(14, 6, False)][0] =  0.3\n",
            "e_policy[(14, 6, False)][1] =   0.7\n",
            "\tPOST e_policy =  defaultdict(<function <lambda> at 0x7a10b32d2950>, {(14, 6, False): array([0.3, 0.7]), (11, 4, False): array([0.3, 0.7]), (21, 3, False): array([0.7, 0.3])})\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "##### Training loop"
      ],
      "metadata": {
        "id": "q3z8FfsZDW5c"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "# Training loop\n",
        "\n",
        "n_actions = env.action_space.n  # numero di azioni disponibili\n",
        "\n",
        "# Fase di inizializzazione delle strutture dati utilizzate\n",
        "e_policy = defaultdict(lambda: np.zeros(n_actions)) # definisce la probabilità di azione dato un certo stato,\n",
        "#key (state,action) -> value probability\n",
        "\n",
        "Q_table = defaultdict(lambda: np.zeros(env.action_space.n))\n",
        "\n",
        "returns_sum = defaultdict(float) # Nota:  returns(state, action)\n",
        "returns_count = defaultdict(int)\n",
        "\n",
        "# Parametri\n",
        "gamma = 1\n",
        "num_episodes = 500_001\n",
        "EPS = 0.25\n",
        "\n",
        "for episode in range(num_episodes):\n",
        "  trajectory = generate_episode_soft_policy(e_policy)\n",
        "  G, T = 0, len(trajectory)\n",
        "\n",
        "  visited_state = []\n",
        "  for i in range(T):\n",
        "    visited_state.append((trajectory[i][0],trajectory[i][1]))\n",
        "\n",
        "  for i in range(T-1, -1, -1):\n",
        "       state, action, reward = trajectory[i]\n",
        "       G = gamma*G + reward # ha solo un senso formale, di fatto, in questo caso, G += reward\n",
        "\n",
        "       visited_state.pop()\n",
        "       if (state, action) not in visited_state: # A meno che la coppia (state, action) non sia stata già visitata in questo episodio, fai:\n",
        "\n",
        "                returns_sum[(state, action)] += G  # returns_sum[(state, action)] = returns_sum[(state, action)] + G\n",
        "                returns_count[(state, action)] += 1\n",
        "\n",
        "                Q_table[state][action] = returns_sum[(state, action)] / returns_count[(state, action)]\n",
        "\n",
        "                best_action = np.argmax(Q_table[state])\n",
        "                # alternativa:\n",
        "                # best_action = max(list(range(env.action_space.n)), key = lambda x: Q_table[state][x])\n",
        "\n",
        "                # aggiorna la soft_policy\n",
        "                for a_index in np.arange(n_actions):\n",
        "                   if a_index == best_action:\n",
        "                       e_policy[state][a_index] = 1 - EPS + (EPS/n_actions)\n",
        "                   else:\n",
        "                       e_policy[state][a_index] = EPS/n_actions\n",
        "\n",
        "  if (episode!= 0) and (episode % 100_000 == 0) :\n",
        "               print(\"Q_TABLE                                              COUNT    0         1   \")\n",
        "               print(\"Q_table[(17, 5, True)] = \", Q_table[(17, 5, True)],\" Count : \",returns_count[((17, 5, True), 0)], \" -  \",returns_count[((17, 5, True), 1)] )\n",
        "               print(\"Q_table[(17, 8, True)] = \", Q_table[(17, 8, True)],\" Count : \",returns_count[((17, 8, True), 0)], \" -  \",returns_count[((17, 8, True), 1)] )\n",
        "               print(\"Q_table[(18, 9, True)] = \", Q_table[(18, 9, True)],\" Count : \",returns_count[((18, 9, True), 0)], \" -  \",returns_count[((18, 9, True), 1)] )\n",
        "\n",
        "               print(\"Q_table[(13, 5, False)] = \", Q_table[(13, 5, False)],\" Count : \",returns_count[((13, 5, False), 0)], \" -  \",returns_count[((13, 5, False), 1)] )\n",
        "               print(\"Q_table[(13, 6, False)] = \", Q_table[(13, 6, False)],\" Count : \",returns_count[((13, 6, False), 0)], \" -  \",returns_count[((13, 6, False), 1)] )\n",
        "               print(\"Q_table[(13, 7, False)] = \", Q_table[(13, 7, False)],\" Count : \",returns_count[((13, 7, False), 0)], \" -  \",returns_count[((13, 7, False), 1)] )\n",
        "\n",
        "               print(\"Q_table[(11, 10, False)] = \", Q_table[(11, 10, False)],\" Count : \",returns_count[((11, 10, False), 0)], \" -  \",returns_count[((11, 10, False), 1)] )\n",
        "               #print(\"\\ne_policy\")\n",
        "               #print(pd.DataFrame(e_policy.items(),columns=['state','action']).head(4))\n",
        "               print(\" \\n\")\n"
      ],
      "metadata": {
        "id": "u4quWwYWjWws",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "f71b4f98-4a78-47ed-e714-e3676f267a2f"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Q_TABLE                                              COUNT    0         1   \n",
            "Q_table[(17, 5, True)] =  [0.19587629 0.        ]  Count :  97  -   16\n",
            "Q_table[(17, 8, True)] =  [-0.5        -0.07142857]  Count :  20  -   84\n",
            "Q_table[(18, 9, True)] =  [-0.04347826 -0.27777778]  Count :  92  -   36\n",
            "Q_table[(13, 5, False)] =  [-0.17647059 -0.39423077]  Count :  697  -   104\n",
            "Q_table[(13, 6, False)] =  [-0.13636364 -0.36111111]  Count :  660  -   108\n",
            "Q_table[(13, 7, False)] =  [-0.48812665 -0.28671329]  Count :  379  -   429\n",
            "Q_table[(11, 10, False)] =  [-0.52252252 -0.07848444]  Count :  222  -   1478\n",
            " \n",
            "\n",
            "Q_TABLE                                              COUNT    0         1   \n",
            "Q_table[(17, 5, True)] =  [ 0.03351955 -0.10526316]  Count :  179  -   38\n",
            "Q_table[(17, 8, True)] =  [-0.61764706 -0.14136126]  Count :  34  -   191\n",
            "Q_table[(18, 9, True)] =  [-0.18099548 -0.35483871]  Count :  221  -   62\n",
            "Q_table[(13, 5, False)] =  [-0.16014235 -0.46296296]  Count :  1405  -   216\n",
            "Q_table[(13, 6, False)] =  [-0.1201748 -0.25     ]  Count :  1373  -   216\n",
            "Q_table[(13, 7, False)] =  [-0.46311475 -0.3032146 ]  Count :  488  -   1151\n",
            "Q_table[(11, 10, False)] =  [-0.51196172 -0.0673456 ]  Count :  418  -   3044\n",
            " \n",
            "\n",
            "Q_TABLE                                              COUNT    0         1   \n",
            "Q_table[(17, 5, True)] =  [-0.02430556 -0.03389831]  Count :  288  -   59\n",
            "Q_table[(17, 8, True)] =  [-0.51923077 -0.14052288]  Count :  52  -   306\n",
            "Q_table[(18, 9, True)] =  [-0.17280453 -0.2987013 ]  Count :  353  -   77\n",
            "Q_table[(13, 5, False)] =  [-0.18667917 -0.42580645]  Count :  2132  -   310\n",
            "Q_table[(13, 6, False)] =  [-0.12511759 -0.27002967]  Count :  2126  -   337\n",
            "Q_table[(13, 7, False)] =  [-0.44312394 -0.3175813 ]  Count :  589  -   1968\n",
            "Q_table[(11, 10, False)] =  [-0.53090333 -0.06081974]  Count :  631  -   4538\n",
            " \n",
            "\n",
            "Q_TABLE                                              COUNT    0         1   \n",
            "Q_table[(17, 5, True)] =  [-0.04304636  0.0754717 ]  Count :  302  -   159\n",
            "Q_table[(17, 8, True)] =  [-0.44444444 -0.11990408]  Count :  63  -   417\n",
            "Q_table[(18, 9, True)] =  [-0.17078189 -0.27      ]  Count :  486  -   100\n",
            "Q_table[(13, 5, False)] =  [-0.1795422  -0.40139211]  Count :  2796  -   431\n",
            "Q_table[(13, 6, False)] =  [-0.12504403 -0.28971963]  Count :  2839  -   428\n",
            "Q_table[(13, 7, False)] =  [-0.45868946 -0.33630528]  Count :  702  -   2804\n",
            "Q_table[(11, 10, False)] =  [-0.52291422 -0.04768234]  Count :  851  -   6019\n",
            " \n",
            "\n",
            "Q_TABLE                                              COUNT    0         1   \n",
            "Q_table[(17, 5, True)] =  [-0.04923077  0.14716981]  Count :  325  -   265\n",
            "Q_table[(17, 8, True)] =  [-0.43037975 -0.11931818]  Count :  79  -   528\n",
            "Q_table[(18, 9, True)] =  [-0.19063005 -0.20661157]  Count :  619  -   121\n",
            "Q_table[(13, 5, False)] =  [-0.16600567 -0.34823091]  Count :  3530  -   537\n",
            "Q_table[(13, 6, False)] =  [-0.13969963 -0.31962617]  Count :  3529  -   535\n",
            "Q_table[(13, 7, False)] =  [-0.45185185 -0.35020862]  Count :  810  -   3595\n",
            "Q_table[(11, 10, False)] =  [-0.52222222 -0.04764405]  Count :  1080  -   7619\n",
            " \n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# Per il grafico\n",
        "# converti la stochastic policy ottenuta in una deterministic policy\n",
        "#\n",
        "deter_pi = {state: np.argmax(values) for state, values in e_policy.items()}\n",
        "plot_policy(deter_pi)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 387
        },
        "outputId": "6bf07b06-52b5-427e-c519-37e01455a744",
        "id": "yZHIUfiizCgY"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 800x800 with 4 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAvgAAAFyCAYAAACA4a7AAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjcuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/bCgiHAAAACXBIWXMAAA9hAAAPYQGoP6dpAABX1klEQVR4nO3de1wU9f4/8NeCutwEBEVAEVdRIVQkL3kpASUV71ZqZoqaeSxMjNIkjgmHY2anvJu3Sq3U7GSoWVoeE0hTExU18wK24v2WwrqsXITP7w9/7Nfluos77DK+no/HPmg+M/N5v4fwPW9mZweFEEKAiIiIiIhkwcbSCRARERERkfmwwSciIiIikhE2+EREREREMsIGn4iIiIhIRtjgExERERHJCBt8IiIiIiIZYYNPRERERCQjbPCJiIiIiGSEDT4RERERkYywwadaZe3atVAoFEhLS6ty29DQUISGhkqfFBHRYyI+Ph4KhQK3bt2qctvmzZtj3Lhx0idFRGWwwSeTVVXg27ZtK/vGesSIEVAoFHjnnXcsnQoRWVjJhQc7Oztcvny5zPrQ0FC0bdvWbPHGjRsHJyenCtc7OTnJvrHu0qULFAoFli9fbulUiKwSG3wiE2k0Gnz//fdo3rw5Nm7cCCGEpVMiIiuQn5+PDz74wNJpyF5GRgYOHTqE5s2bY/369ZZOh8gqscEnMtHmzZtRVFSEzz//HBcvXkRqaqqlUyIiK9ChQwesXr0aV65csXQqsvbVV1/Bw8MDH3/8MX777TecP3/e0ikRWR02+FQjlixZgsDAQDg4OKBBgwbo1KkTNmzYoF+flZWF119/HW3atIG9vT3c3d0xfPjwCgu3TqfDP/7xD7i7u8PZ2Rljx47FnTt3qswjPz8fs2fPhp+fH5RKJXx8fDBjxgzk5+cbfSzr16/Hs88+i7CwMAQEBFR4Ben06dMYMWIEGjVqBHt7e7Rp0wZxcXEG21y+fBkTJkxA48aNoVQqERgYiM8//9zoXIjIerz77rsoKioy6ir+/fv3kZiYiJYtW0KpVKJ58+Z49913TapFxiosLERCQgJatWoFOzs7uLu74+mnn8auXbv02xw/fhzjxo1DixYtYGdnB09PT0yYMAF///13uXPeunULI0aMgLOzM9zd3REdHY28vLwqc8nOzsa0adPg4+MDpVIJPz8/zJs3D8XFxUYfz4YNG/DCCy9g4MCBcHFxMTiXPOzgwYPo378/GjRoAEdHR7Rv3x6LFi0y2Ob06dN44YUX4ObmBjs7O3Tq1Anbtm0zOhcia1XH0gmQ/K1evRpTp07FCy+8oD8JHD9+HAcPHsRLL70EADh06BB+++03vPjii2jatCnOnz+P5cuXIzQ0FH/++SccHBwM5pwyZQpcXV0RHx+PM2fOYPny5cjKykJycjIUCkW5eRQXF2Pw4MHYu3cvJk2ahICAAJw4cQILFizA2bNnsWXLliqP5cqVK9izZw/WrVsHABg1ahQWLFiApUuXol69evrtjh8/jmeeeQZ169bFpEmT0Lx5c5w7dw7ff/895syZAwC4fv06unbtCoVCgSlTpqBRo0bYsWMHXnnlFWg0GkybNq0a320ishSVSoWxY8di9erVmDlzJry9vSvcduLEiVi3bh1eeOEFvPXWWzh48CDmzp2LU6dOISkpyax5xcfHY+7cuZg4cSK6dOkCjUaDtLQ0HDlyBM8++ywAYNeuXfjrr78wfvx4eHp64uTJk1i1ahVOnjyJAwcOlKmrI0aMQPPmzTF37lwcOHAAixcvxp07d/DFF19UmIdOp0NISAguX76Mf/zjH2jWrBl+++03xMbG4urVq1i4cGGVx3Lw4EFkZmZizZo1qFevHp577jmsX78e7777rsF2u3btwsCBA+Hl5YXo6Gh4enri1KlT2L59O6KjowEAJ0+eRI8ePdCkSRPMnDkTjo6O+OabbzB06FBs3rwZw4YNM/E7TWRFBJGJZs+eLQCImzdvlrs+MDBQhISE6JeHDBkiAgMDK51Tp9OVGdu/f78AIL744gv92Jo1awQA0bFjR1FQUKAf//DDDwUAsXXrVv1YSEiIQR5ffvmlsLGxEb/++qtBnBUrVggAYt++fZXmKIQQH330kbC3txcajUYIIcTZs2cFAJGUlGSwXc+ePUX9+vVFVlaWwXhxcbH+v1955RXh5eUlbt26ZbDNiy++KFxcXMr9nhCR9SmpS4cOHRLnzp0TderUEVOnTtWvDwkJMaiB6enpAoCYOHGiwTxvv/22ACB++eWXSuNFRkYKR0fHCtc7OjqKyMhI/XJQUJAYMGBApXOWV282btwoAIjU1FT9WEn9Hzx4sMG2r7/+ugAgjh07ph/z9fU1yCMxMVE4OjqKs2fPGuw7c+ZMYWtrKy5cuFBpjkIIMWXKFOHj46OvpT///LMAII4eParf5v79+0KlUglfX19x584dg/0frsG9e/cW7dq1E3l5eQbru3fvLlq1alVlLkTWjLfokORcXV1x6dIlHDp0qMJt7O3t9f9dWFiIv//+G35+fnB1dcWRI0fKbD9p0iTUrVtXv/zaa6+hTp06+PHHHyuM8d///hcBAQHw9/fHrVu39K9evXoBAPbs2VPlsaxfvx4DBgxA/fr1AQCtWrVCx44dDW7TuXnzJlJTUzFhwgQ0a9bMYP+Sq2BCCGzevBmDBg2CEMIgn759+yInJ6fc4yYi69aiRQuMGTMGq1atwtWrV8vdpqROxcTEGIy/9dZbAIAffvjBrDm5urri5MmTyMjIqHCbh2twXl4ebt26ha5duwJAubUoKirKYPmNN94AgCpr8DPPPIMGDRoY1Lzw8HAUFRVV+Xmm+/fvY9OmTRg5cqS+lvbq1QseHh4GNfjo0aNQq9WYNm0aXF1dDeYo2e/27dv45ZdfMGLECNy9e1efy99//42+ffsiIyOj3CciEdUWbPBJEg+/nfvOO+/AyckJXbp0QatWrRAVFYV9+/YZbH/v3j289957+vsyGzZsiEaNGiE7Oxs5OTll5m/VqpXBspOTE7y8vCr9sFVGRgZOnjyJRo0aGbxat24NALhx40alx3Tq1CkcPXoUPXr0QGZmpv4VGhqK7du3Q6PRAAD++usvAKj0sXg3b95EdnY2Vq1aVSaf8ePHG5UPEVmnf/7zn7h//36F9+JnZWXBxsYGfn5+BuOenp5wdXVFVlbWI+fwcA3+17/+hezsbLRu3Rrt2rXD9OnTcfz4cYPtb9++jejoaDRu3Bj29vZo1KgRVCoVABhVg1u2bAkbG5sqa/DOnTvL1Lzw8HAAVde8n3/+GTdv3kSXLl309VetViMsLAwbN27U38d/7tw5AJXX4MzMTAghMGvWrDL5zJ4926h8iKwZ78Enk9nZ2QF40JSXR6fT6bcBgICAAJw5cwbbt2/Hzp07sXnzZnzyySd47733kJCQAODB1Z81a9Zg2rRp6NatG1xcXKBQKPDiiy+a9OGryhQXF6Ndu3aYP39+uet9fHwq3f+rr74CALz55pt48803y6zfvHmzvjk3JhcAePnllxEZGVnuNu3btzdqLiKyLi1atMDLL7+MVatWYebMmRVuV9HnhapiZ2eH/Px8CCHKzCGEQF5enkEN7tmzJ86dO4etW7fi559/xqeffooFCxZgxYoVmDhxIoAH99T/9ttvmD59Ojp06AAnJycUFxejX79+RtVgY46luLgYzz77LGbMmFHu+pKLLRUpuUo/YsSIctenpKQgLCysyjxKcgGAt99+G3379i13m9K/gBHVJmzwyWS+vr4AgDNnzpRpinU6HS5evIg+ffoYjDs6OmLkyJEYOXIkCgoK8Nxzz2HOnDmIjY2FnZ0dvv32W0RGRuLjjz/W75OXl4fs7Oxyc8jIyDAo5FqtFlevXkX//v0rzLtly5Y4duwYevfubfKJVQiBDRs2ICwsDK+//nqZ9YmJiVi/fj3Gjx+PFi1aAAD++OOPCudr1KgR6tevj6KiIv3VKyKSj3/+85/46quvMG/evDLrfH19UVxcjIyMDAQEBOjHr1+/juzsbH2NrYivry/u37+Pc+fOlWlCMzMzUVRUVGYONzc3jB8/HuPHj4dWq0XPnj0RHx+PiRMn4s6dO9i9ezcSEhLw3nvv6fep7JaejIwM/RX+krjFxcVo3rx5hfu0bNkSWq22WjUvNzcXW7duxciRI/HCCy+UWT916lSsX78eYWFhaNmyJYAHNbiiWCV1um7duqzBJEu8RYdM1rt3b9SrVw/Lly8vc2Vn1apVuH//PiIiIvRjpR+zVq9ePTzxxBMQQqCwsBAAYGtrW+YPRi1ZsgRFRUXl5rBq1Sr9vgCwfPnyMnFLGzFiBC5fvozVq1eXWXfv3j3k5uZWuO++fftw/vx5jB8/Hi+88EKZ18iRI7Fnzx5cuXIFjRo1Qs+ePfH555/jwoULBvOUHKOtrS2ef/55bN68udxfBG7evFlhLkRk/Vq2bImXX34ZK1euxLVr1wzWlVyIKP3UmJJ3FwcMGFDp3CV1bunSpWXWLVu2zGAboGwNdnJygp+fn/6RnLa2tgBQpgZX9lSbkjgllixZUiZuaSNGjMD+/fvx008/lVmXnZ2N+/fvV7hvUlIScnNzERUVVW4NHjhwIDZv3oz8/Hw8+eSTUKlUWLhwYZmLRCXH6OHhgdDQUKxcubLcz0qwBlNtxyv4ZDIPDw+89957+Oc//4mePXti8ODBcHBwwG+//YaNGzeiT58+GDRokH77Pn36wNPTEz169EDjxo1x6tQpLF261ODDqgMHDsSXX34JFxcXPPHEE9i/fz/+97//wd3dvdwcCgoK0Lt3b4wYMQJnzpzBJ598gqeffhqDBw+uMO8xY8bgm2++weTJk7Fnzx706NEDRUVFOH36NL755hv89NNP6NSpU7n7rl+/Hra2thWeeAcPHoy4uDh8/fXXiImJweLFi/H000/jySefxKRJk6BSqXD+/Hn88MMPSE9PBwB88MEH2LNnD5566im8+uqreOKJJ3D79m0cOXIE//vf/3D79m1j/ncQkZWKi4vDl19+iTNnziAwMFA/HhQUhMjISKxatQrZ2dkICQnB77//jnXr1mHo0KFV3mbSoUMHTJw4EYsWLUJGRobBoy5//PFHTJw4EUFBQfrtn3jiCYSGhqJjx45wc3NDWloavv32W0yZMgUA4OzsjJ49e+LDDz9EYWEhmjRpgp9//hlqtbrCHNRqNQYPHox+/fph//79+Oqrr/DSSy8ZxC1t+vTp2LZtGwYOHIhx48ahY8eOyM3NxYkTJ/Dtt9/i/PnzaNiwYbn7rl+/Hu7u7ujevXu56wcPHozVq1fjhx9+wHPPPYfly5dj0KBB6NChA8aPHw8vLy+cPn0aJ0+e1P+CsWzZMjz99NNo164dXn31VbRo0QLXr1/H/v37cenSJRw7dqzS/w9EVs1CT+8hGfjqq69E165dhaOjo1AqlcLf318kJCQYPHJMCCFWrlwpevbsKdzd3YVSqRQtW7YU06dPFzk5Ofpt7ty5I8aPHy8aNmwonJycRN++fcXp06fLPGat5HF0KSkpYtKkSaJBgwbCyclJjB49Wvz9998GcUs/JlMIIQoKCsS8efNEYGCgUCqVokGDBqJjx44iISHBIJ/S+7i7u4tnnnmm0u+HSqUSwcHB+uU//vhDDBs2TLi6ugo7OzvRpk0bMWvWLIN9rl+/LqKiooSPj4+oW7eu8PT0FL179xarVq2qNBYRWY+HH5NZWmRkpABQ5lHBhYWFIiEhQahUKlG3bl3h4+MjYmNjy9TPihQVFYlFixaJoKAgYWdnJ+zs7ERQUJBYvHixKCoqMtj23//+t+jSpYtwdXUV9vb2wt/fX8yZM8fgUcOXLl3S1ysXFxcxfPhwceXKFQFAzJ49W79dyWMy//zzT/HCCy+I+vXriwYNGogpU6aIe/fuGcQtXb+FEOLu3bsiNjZW+Pn5iXr16omGDRuK7t27i48++sggn4ddv35d1KlTR4wZM6bC74dOpxMODg5i2LBh+rG9e/eKZ599VtSvX184OjqK9u3biyVLlhjsd+7cOTF27Fjh6ekp6tatK5o0aSIGDhwovv322wpjEdUGCiFKvSdHRERERES1Fu/BJyIiIiKSETb4REREREQywgafiKiWKCgogJ+fH3777TdLp2KUW7duwcPDA5cuXbJ0KkREAB6fOsoGn4jIApYtW4bmzZvDzs4OTz31FH7//fcq91mxYgVUKpXBk0RSUlLQq1cvuLm5wcHBAa1atUJkZCQKCgowbtw4KBSKCl8lzywPDQ3FtGnTDGJlZmZi/PjxaNq0KZRKJVQqFUaNGoW0tDT9NgqFAlu2bNEvFxYWYtSoUWjSpAn++OMPNGzYEGPHjtX/ZVAiInNiHa0YG3wiohq2adMmxMTEYPbs2Thy5AiCgoLQt29f3Lhxo8J9hBBYunQpXnnlFf3Yn3/+iX79+qFTp05ITU3FiRMnsGTJEtSrVw9FRUVYtGgRrl69qn8BwJo1a/TLhw4dKjdWWloaOnbsiLNnz2LlypX4888/kZSUBH9/f7z11lvl7qPT6TB48GAcOnQIe/fuRdu2bQEA48ePx/r16/nYVyIyK9bRKlj2IT5ERI+fLl26iKioKP1yUVGR8Pb2FnPnzq1wn0OHDgkbGxuh0Wj0YwsWLBDNmzc3Oi4AkZSUVGY8JCREREdHCyGEKC4uFoGBgaJjx45lHrcoxINH2pae786dO6J79+6iffv24urVq2X2UalU4tNPPzU6TyKiqrCOVk72f+iquLgYV65cQf369aFQKCydDhFZiBACd+/ehbe3N2xsKn/zMi8vDwUFBSbPX7rGKJVKKJVKg7GCggIcPnwYsbGx+jEbGxuEh4dj//79Fc7/66+/onXr1vo/DgcAnp6euHr1KlJTU9GzZ0+T8q1Ieno6Tp48iQ0bNui/Tw/XURsbG2g0Gv32arUa//znP+Ho6Ijvv/8eDg4OBusBIDg4GLt378bw4cPNkiMRWQbrqHHKq6MPc3V1NVi+du0aQkJC4OTkhJSUlDLrAaBLly749ddfDd59qIzsG/wrV67Ax8fH0mkQkZW4ePEimjZtWuH6vLw8eDdrijs3/zZpXicnJ2i1WoOx2bNnIz4+3mDs1q1bKCoqQuPGjQ3GGzdujNOnT1c4f1ZWFry9vQ3Ghg8fjp9++gkhISHw9PRE165d0bt3b4wdOxbOzs4m5V8iIyMDAODv768fq6yOxsTE6P/b19e30rk3btxYrZyIyLqwjlauvDpamejoaLRo0QK7du2Cg4NDudt4e3vj6NGjRucg+wa/5Le02esW4YlOHSSL89efZxA7fBLm/ncVWjzRhnGsKI6cjoVxqk+nzcX4p/obXLkpT0FBAe7c/BtrDv4IBydHk+a+ePGiwQmh9FWnR3Hv3j3Y2dkZjNna2mLNmjX497//jV9++QUHDx7E+++/j3nz5uH333+Hl5eXyXFEOX/7kHWUceQWR07HUpNxWEeNU14drczAgQOxZcsWrFy5Em+++Wa529jb20On0xk9p+wb/JK3euwc7OFQ30myOHb//zcuOwcHxrGyOHI6FsZ5dMbequfg5GhyPs7OzlVe8WnYsCFsbW1x/fp1g/Hr16/D09Oz0v1OnDhR7romTZpgzJgxGDNmDBITE9G6dWusWLECCQkJJuUPAK1btwYAnD59GsHBwQBYRxlHfnHkdCw1GacE62jlyqujlRkzZgwGDx6MCRMmQAhh8M5oidu3b6NRo0ZG58Cn6BAR1aB69eqhY8eO2L17t36suLgYu3fvRrdu3SrcLzg4GKdPn67yylCDBg3g5eWF3NzcauXXoUMHPPHEE/j4449RXFxcrTmIiKRU2+todnZ2mbHIyEisXbsWM2bMwEcffVRm/R9//GHULwslZH8Fn4jI2sTExCAyMhKdOnVCly5dsHDhQuTm5mL8+PEV7hMWFgatVouTJ0/qH522cuVKpKenY9iwYWjZsiXy8vLwxRdf4OTJk1iyZEm1clMoFFizZg3Cw8PxzDPPIC4uDk2aNKnWXEREUqltddTf3x9arRbff/89fv75Z6SkpJTZb8yYMbCxsUFkZCSEEJg+fTqAB4/PPHz4MN5//32jc2CDT0RUw0aOHImbN2/ivffew7Vr19ChQwfs3LmzzAfGHubu7o5hw4Zh/fr1mDt3LoAHT1XYu3cvJk+ejCtXrsDJyQmBgYHYsmULQkJCqp1fly5dkJaWhjlz5uDVV1/FzZs3qz0XEZEUalsdvXXrFry8vNC9e3csXLiwwv1Gjx4NGxsbjBkzBsXFxXjnnXewdetWNGvWDM8884zR8dngExFZwJQpUzBlyhST9omLi8Ozzz6LuLg4ODk5ITg4GF9++aXR+1f0tnRycnKZsdatW2PdunUAAI1GAxcXF5NyJSKSWm2qo6bMN2rUKIwaNUq/vGjRIrz33ntG5wjwHnwiolqjffv2mDdvHtRqtaVTISKqlWpbHb116xaee+45g4bfGLyCT0RUi4wbN87SKRAR1Wq1qY42bNgQM2bMMHk/XsEnIiIiIpIRNvhERERERDLCBp+IiIiISEbY4BMRERERyQgbfCIiIiIiGWGDT0REREQkI2zwiYiIiIhkhA0+EREREZGMsMEnIiIiIpIRNvhERERERDLCBp+IiIiISEbY4BMRERERyQgbfCIiIiIiGbFogz937lx07twZ9evXh4eHB4YOHYozZ84YbLNq1SqEhobC2dkZCoUC2dnZlkmWiMgKsY4SEVFpFm3wU1JSEBUVhQMHDmDXrl0oLCxEnz59kJubq99Gp9OhX79+ePfddy2YKRGRdWIdJSKi0upYMvjOnTsNlteuXQsPDw8cPnwYPXv2BABMmzYNAJCcnFzD2RERWT/WUSIiKs2iDX5pOTk5AAA3N7dqz5Gfn4/8/Hz9skajAQBc+isLdg4Oj5ZgJS5mqg2+Mo71xJHTsTBO9eXpdJLOby1YRxmHcawjhhzjPC51VA4UQghh6SQAoLi4GIMHD0Z2djb27t1bZn1ycjLCwsJw584duLq6VjhPfHw8EhISJMyUiGqznJwcODs7V7heo9HAxcUFm06mwKG+k1Fz6u5qMTIwpMq5pcY6SkQ1Qc51VC6s5gp+VFQU/vjjj3JPSqaIjY1FTEyMflmj0cDHxwdR8+Lg1zbgUdOs0MVMNeZHz0LMokT4+KkYx4riyOlYGKf68nQ6xA6fJNn81oB1lHEYx3piyDHO41BH5cIqGvwpU6Zg+/btSE1NRdOmTR9pLqVSCaVSWWa8aQtf+LWT7sRUwsdPxThWGkdOx8I4ptPd1Uo2tzVgHWUcxrHOGHKKI/c6KicWbfCFEHjjjTeQlJSE5ORkqFTS/dZJRCRHrKNERFSaRRv8qKgobNiwAVu3bkX9+vVx7do1AICLiwvs7e0BANeuXcO1a9eQmZkJADhx4gTq16+PZs2aPdKHyIiI5IB1lIiISrPoc/CXL1+OnJwchIaGwsvLS//atGmTfpsVK1YgODgYr776KgCgZ8+eCA4OxrZt2yyVNhGR1WAdJSKi0ix+i05V4uPjER8fL30yRES1EOsoERGVZtEr+EREREREZF5s8ImIiIiIZIQNPhERERGRjLDBJyIiIiKSETb4REREREQywgafiIiIiEhG2OATEREREckIG3wiIiIiIhlhg09EREREJCNs8ImIiIiIZIQNPhERERGRjLDBJyIiIiKSETb4REREREQywgafiIiIiEhG2OATEREREcmIRRv8uXPnonPnzqhfvz48PDwwdOhQnDlzxmCbvLw8REVFwd3dHU5OTnj++edx/fp1C2VMRGRdWEeJiKg0izb4KSkpiIqKwoEDB7Br1y4UFhaiT58+yM3N1W/z5ptv4vvvv8d///tfpKSk4MqVK3juuecsmDURkfVgHSUiotLqWDL4zp07DZbXrl0LDw8PHD58GD179kROTg4+++wzbNiwAb169QIArFmzBgEBAThw4AC6du1qibSJiKwG6ygREZVm0Qa/tJycHACAm5sbAODw4cMoLCxEeHi4fht/f380a9YM+/fvL/fElJ+fj/z8fP2yRqMBAFz6Kwt2Dg6S5X4xU23wlXGsJ46cjoVxqi9Pp5N0fmvBOso4jGMdMeQY53Gpo3KgEEIISycBAMXFxRg8eDCys7Oxd+9eAMCGDRswfvx4gxMNAHTp0gVhYWGYN29emXni4+ORkJBQIzkTUe2Tk5MDZ2fnCtdrNBq4uLhg08kUONR3MmpO3V0tRgaGVDm31FhHiagmyLmOyoXVXMGPiorCH3/8oT8pVVdsbCxiYmL0yxqNBj4+PoiaFwe/tgGPmmaFLmaqMT96FmIWJcLHT8U4VhRHTsfCONWXp9Mhdvgkyea3BqyjjMM41hNDjnEehzoqF1bR4E+ZMgXbt29HamoqmjZtqh/39PREQUEBsrOz4erqqh+/fv06PD09y51LqVRCqVSWGW/awhd+7aQ7MZXw8VMxjpXGkdOxMI7pdHe1ks1tDVhHGYdxrDOGnOLIvY7KiUWfoiOEwJQpU5CUlIRffvkFKpXhb50dO3ZE3bp1sXv3bv3YmTNncOHCBXTr1q2m0yUisjqso0REVJpFr+BHRUVhw4YN2Lp1K+rXr49r164BAFxcXGBvbw8XFxe88soriImJgZubG5ydnfHGG2+gW7dufPIDERFYR4mIqCyLNvjLly8HAISGhhqMr1mzBuPGjQMALFiwADY2Nnj++eeRn5+Pvn374pNPPqnhTImIrBPrKBERlWbRBt+YB/jY2dlh2bJlWLZsWQ1kRERUu7COEhFRaRa9B5+IiIiIiMyLDT4RERERkYywwSciIiIikhE2+EREREREMsIGn4iIiIhIRtjgExERERHJCBt8IiIiIiIZYYNPRERERCQjbPCJiIiIiGSEDT4RERERkYywwSciIiIikhE2+EREREREMsIGn4iIiIhIRtjgExERERHJiEUb/NTUVAwaNAje3t5QKBTYsmWLwfrr169j3Lhx8Pb2hoODA/r164eMjAzLJEtEZIVYR4mIqDSLNvi5ubkICgrCsmXLyqwTQmDo0KH466+/sHXrVhw9ehS+vr4IDw9Hbm6uBbIlIrI+rKNERFRaHUsGj4iIQERERLnrMjIycODAAfzxxx8IDAwEACxfvhyenp7YuHEjJk6cWJOpEhFZJdZRIiIqzWrvwc/PzwcA2NnZ6cdsbGygVCqxd+9eS6VFRFRrsI4SET2eLHoFvzL+/v5o1qwZYmNjsXLlSjg6OmLBggW4dOkSrl69WuF++fn5+pMaAGg0GgDApb+yYOfgIFm+FzPVBl8Zx3riyOlYGKf68nQ6See3RqyjjMM4loshxziPYx2trRRCCGHpJABAoVAgKSkJQ4cO1Y8dPnwYr7zyCo4dOwZbW1uEh4fDxsYGQgjs2LGj3Hni4+ORkJBQQ1kTUW2Tk5MDZ2fnCtdrNBq4uLhg08kUONR3MmpO3V0tRgaGVDm31FhHiagmyLmOyoXVXsEHgI4dOyI9PR05OTkoKChAo0aN8NRTT6FTp04V7hMbG4uYmBj9skajgY+PD6LmxcGvbYBkuV7MVGN+9CzELEqEj5+KcawojpyOhXGqL0+nQ+zwSZLNb61YRxmHcSwTQ45xHtc6WhtZdYNfwsXFBcCDD4ylpaUhMTGxwm2VSiWUSmWZ8aYtfOHXTroTUwkfPxXjWGkcOR0L45hOd1cr2dy1Aeso4zCOZWLIKc7jXkdrE4s2+FqtFpmZmfpltVqN9PR0uLm5oVmzZvjvf/+LRo0aoVmzZjhx4gSio6MxdOhQ9OnTx4JZExFZD9ZRIiIqzaINflpaGsLCwvTLJW8JR0ZGYu3atbh69SpiYmJw/fp1eHl5YezYsZg1a5al0iUisjqso0REVJpFG/zQ0FBU9hnfqVOnYurUqTWYERFR7cI6SkREpVntc/CJiIiIiMh0bPCJiIiIiGSEDT4RERERkYywwSciIiIikhE2+EREREREMsIGn4iIiIhIRtjgExERERHJCBt8IiIiIiIZYYNPRERERCQjbPCJiIiIiGSEDT4RERERkYywwSciIiIikhE2+EREREREMsIGn4iIiIhIRtjgExERERHJiEUb/NTUVAwaNAje3t5QKBTYsmWLwXqtVospU6agadOmsLe3xxNPPIEVK1ZYJlkiIivEOkpERKVZtMHPzc1FUFAQli1bVu76mJgY7Ny5E1999RVOnTqFadOmYcqUKdi2bVsNZ0pEZJ1YR4mIqLQ6lgweERGBiIiICtf/9ttviIyMRGhoKABg0qRJWLlyJX7//XcMHjy4hrIkIrJerKNERFSaUVfwNRqN0S9z6t69O7Zt24bLly9DCIE9e/bg7Nmz6NOnj1njEBFJjXWUiIhqilFX8F1dXaFQKIyasKio6JESetiSJUswadIkNG3aFHXq1IGNjQ1Wr16Nnj17VrhPfn4+8vPz9cslJ8tLf2XBzsHBbLmVdjFTbfCVcawnjpyOhXGqL0+nk3T+qrCOVk1uP3OMY71x5HQsNRnH0nWUjKcQQoiqNkpJSdH/9/nz5zFz5kyMGzcO3bp1AwDs378f69atw9y5cxEZGVm9RBQKJCUlYejQofqxjz76CKtXr8ZHH30EX19fpKamIjY2FklJSQgPDy93nvj4eCQkJFQrByKSv5ycHDg7O1e4XqPRwMXFBZtOpsChvpNRc+ruajEyMKTSuVlHiUguLFVHyXhGNfgP6927NyZOnIhRo0YZjG/YsAGrVq1CcnJy9RIpdWK6d+8eXFxckJSUhAEDBui3mzhxIi5duoSdO3eWO095V558fHwQFxeHgICAauVmDLVajVmzZiExMREqlYpxrCiOnI6FcapPp9Nh0qRJVnFiqq11NGpeHPzaSldHL2aqMT96FmIWJcLHT7qfBcZhHDkdS03GydPpEDvcOuooVc7kD9nu37+/3EesderUCRMnTjRLUgBQWFiIwsJC2NgYfkzA1tYWxcXFFe6nVCqhVCrLjPv6+sLf399s+VVEpVIxjpXGkdOxMI7ptFqtZHObqrbW0aYtfOHXTroGv4SPn4pxGKdG4sjpWGoiju6u9dRRqpzJj8n08fHB6tWry4x/+umn8PHxMWkurVaL9PR0pKenA3hwJS89PR0XLlyAs7MzQkJCMH36dCQnJ0OtVmPt2rX44osvMGzYMFPTJiKyGqyjREQkJZOv4C9YsADPP/88duzYgaeeegoA8PvvvyMjIwObN282aa60tDSEhYXpl2NiYgAAkZGRWLt2Lb7++mvExsZi9OjRuH37Nnx9fTFnzhxMnjzZ1LSJiKwG6ygREUnJ5Aa/f//+yMjIwCeffILTp08DAAYNGoTJkyebfOUpNDQUlX0EwNPTE2vWrDE1RSIiq8Y6SkREUqrWH7pq2rQp3n//fXPnQkT02GAdJSIiqVSrwc/Ozsbvv/+OGzdulPmg1tixY82SGBGRnLGOEhGRVExu8L///nuMHj0aWq0Wzs7OBn+4RaFQ8MRERFQF1lEiIpKSyU/ReeuttzBhwgRotVpkZ2fjzp07+tft27elyJGISFZYR4mISEomN/iXL1/G1KlT4SDhnysnIpIz1lEiIpKSyQ1+3759kZaWJkUuRESPBdZRIiKSksn34A8YMADTp0/Hn3/+iXbt2qFu3boG6wcPHmy25IiI5Ih1lIiIpGRyg//qq68CAP71r3+VWadQKFBUVPToWRERyRjrKBERScnkBr/049yIiMg0rKNERCQlk+/BJyIiIiIi61WtP3SVm5uLlJQUXLhwAQUFBQbrpk6dapbEiIjkjHWUiIikYnKDf/ToUfTv3x86nQ65ublwc3PDrVu34ODgAA8PD56YiIiqwDpKRERSMvkWnTfffBODBg3CnTt3YG9vjwMHDiArKwsdO3bERx99JEWORESywjpKRERSMrnBT09Px1tvvQUbGxvY2toiPz8fPj4++PDDD/Huu+9KkSMRkaywjhIRkZRMbvDr1q0LG5sHu3l4eODChQsAABcXF1y8eNG82RERyRDrKBERScnkBj84OBiHDh0CAISEhOC9997D+vXrMW3aNLRt29akuVJTUzFo0CB4e3tDoVBgy5YtBusVCkW5r//85z+mpk1EZDVYR4mISEomN/jvv/8+vLy8AABz5sxBgwYN8Nprr+HmzZtYtWqVSXPl5uYiKCgIy5YtK3f91atXDV6ff/45FAoFnn/+eVPTJiKyGqyjREQkJZOfotOpUyf9f3t4eGDnzp3VDh4REYGIiIgK13t6ehosb926FWFhYWjRokW1YxIRWRrrKBERSalaz8G3hOvXr+OHH37AunXrKt0uPz8f+fn5+mWNRgMAyMrKgoODg2T5qdVqg6+MYz1x5HQsjFN9Op1O0vlrg0eto5f+yoKdhHX0Yqba4CvjME5tjiHHOHmso7WGQgghjNkwODgYCoWiyu2OHDlSvUQUCiQlJWHo0KHlrv/www/xwQcf4MqVK7Czs6twnvj4eCQkJFQrByKSv5ycHDg7O1e4XqPRwMXFBZtOpsChvpNRc+ruajEyMKTKuVlHiUgOLFlHyThGX8F/+IQhhMDcuXMxefJkuLm5SZFXGZ9//jlGjx5d6UkJAGJjYxETE6Nf1mg08PHxQVxcHAICAiTLT61WY9asWUhMTIRKpWIcK4pTEiNmUSJ8/KQ7louZasyPlsf3TI5xdDodJk2aJNn8xmAdrVxN/1tlHOuMoz0iv/OCXOLk6XSIHW7ZOkrGMbrBnz17tsHyxx9/jOjo6Bq5j/PXX3/FmTNnsGnTpiq3VSqVUCqVZcZ9fX3h7+8vRXoGVCoV41hpHB8/FfzaSdeclJDT90xOcbRarWRzG4t11Dg19W+VcawzTs6dB1/ldF6QSxzdXcvXUTKOyU/RsYTPPvsMHTt2RFBQkKVTISKqlVhHiYgeHxb9kK1Wq0VmZqZ+Wa1WIz09HW5ubmjWrBmAB28N//e//8XHH39sqTSJiKwW6ygREZVm0QY/LS0NYWFh+uWSez4jIyOxdu1aAMDXX38NIQRGjRpliRSJiKwa6ygREZVmdIO/ePFig+X79+9j7dq1aNiwocH41KlTjQ4eGhqKqh7iM2nSJIt/MI6IyBxYR4mIqCYY3eAvWLDAYNnT0xNffvmlwZhCoTDpxERE9DhhHSUioppgdIMv9R+hISKSO9ZRIiKqCbXiKTpERERERGQcNvhERERERDLCBp+IiIiISEbY4BMRERERyYhJz8G/f/8+NmzYgL59+6Jx48ZS5UREZHGNbwo43av88ZMltFrjtgNYR4no8SFVHaWqmXQFv06dOpg8eTLy8vKkyoeISNZYR4mISGom36LTpUsXpKenS5AKEdHjgXWUiIikZNItOgDw+uuvIyYmBhcvXkTHjh3h6OhosL59+/ZmS46ISI5YR4mISEomN/gvvvgiAMM/pa5QKCCEgEKhQFFRkfmyIyKSIdZRIiKSkskNPv8SIxHRo2EdJSIiKZnc4Pv6+kqRBxHRY4N1lIiIpFSt5+B/+eWX6NGjB7y9vZGVlQUAWLhwIbZu3WrW5IiI5Ip1lIiIpGJyg798+XLExMSgf//+yM7O1t8r6urqioULF5o7PyIi2WEdJSIiKZnc4C9ZsgSrV69GXFwcbG1t9eOdOnXCiRMnTJorNTUVgwYNgre3NxQKBbZs2VJmm1OnTmHw4MFwcXGBo6MjOnfujAsXLpiaNhGR1WAdJSIiKZnc4KvVagQHB5cZVyqVyM3NNWmu3NxcBAUFYdmyZeWuP3fuHJ5++mn4+/sjOTkZx48fx6xZs2BnZ2dq2kREVoN1lIiIpGTyh2xVKhXS09PLfEhs586dCAgIMGmuiIgIREREVLg+Li4O/fv3x4cffqgfa9mypWkJExFZGdZRIiKSkskNfkxMDKKiopCXlwchBH7//Xds3LgRc+fOxaeffmq2xIqLi/HDDz9gxowZ6Nu3L44ePQqVSoXY2FgMHTq0wv3y8/ORn5+vX9ZoNACArKwsODg4mC2/0koeeyf14+8Yp/oxLmZKeywl88vheybHODqdTtL5TcE6Wr6a/rfKONYZRyvD84Jc4uRZUR2lyimEEMLUndavX4/4+HicO3cOAODt7Y2EhAS88sor1U9EoUBSUpL+pHPt2jV4eXnBwcEB//73vxEWFoadO3fi3XffxZ49exASElLuPPHx8UhISKh2HkQkbzk5OXB2dq5wvUajgYuLC5KTk+Hk5GTUnFqtFqGhoVXO/TDWUSKqrayljlLFqtXgl9DpdNBqtfDw8Hj0REqdmK5cuYImTZpg1KhR2LBhg367wYMHw9HRERs3bix3nvKuPPn4+CAuLs7kt75NoVarMWvWLCQmJkKlUjGOFcWR07EwTvXpdDpMmjTJ6k5MrKP/p+RnIWZRInz8pPtZuJipxvxoxnmc45TEkEt9q6l/O3k6HWKHW18dpbJMvkVn9uzZmDBhAnx9feHg4CDZ27UNGzZEnTp18MQTTxiMBwQEYO/evRXup1QqoVQqy4z7+vrC39/f7HmWplKpGMdK48jpWBjHdFqtVrK5TcU6WjkfPxX82kn3iwTjME4JudS3ElJ/z3R3raeOUuVMforO1q1b0bJlS/Tu3RsbNmwwuMpjTvXq1UPnzp1x5swZg/GzZ8/yr0ASUa3GOkpERFIyucFPT0/HoUOHEBgYiOjoaHh6euK1117DoUOHTA6u1WqRnp6O9PR0AA/eYkpPT9c/n3n69OnYtGkTVq9ejczMTCxduhTff/89Xn/9dZNjERFZC9ZRIiKSkskNPgAEBwdj8eLFuHLlCj777DNcunQJPXr0QPv27bFo0SLk5OQYNU9aWhqCg4P1z4OOiYlBcHAw3nvvPQDAsGHDsGLFCnz44Ydo164dPv30U2zevBlPP/10ddImIrIarKNERCSVajX4JYQQKCwsREFBAYQQaNCgAZYuXQofHx9s2rSpyv1DQ0MhhCjzWrt2rX6bCRMmICMjA/fu3UN6ejqGDBnyKCkTEVkV1lEiIjK3ajX4hw8fxpQpU+Dl5YU333wTwcHBOHXqFFJSUpCRkYE5c+Zg6tSp5s6ViEg2WEeJiEgqJjf47dq1Q9euXaFWq/HZZ5/h4sWL+OCDD+Dn56ffZtSoUbh586ZZEyUikgvWUSIikpLJj8kcMWIEJkyYgCZNmlS4TcOGDVFcXPxIiRERyRXrKBERScnkBn/WrFlS5EFE9NhgHSUiIimZ3OADwKVLl7Bt2zZcuHABBQUFBuvmz59vlsSIiOSMdZSIiKRicoO/e/duDB48GC1atMDp06fRtm1bnD9/HkIIPPnkk1LkSEQkK6yjREQkJZM/ZBsbG4u3334bJ06cgJ2dHTZv3oyLFy8iJCQEw4cPlyJHIiJZYR0lIiIpmdzgnzp1CmPHjgUA1KlTB/fu3YOTkxP+9a9/Yd68eWZPkIhIblhHiYhISiY3+I6Ojvr7Rb28vHDu3Dn9ulu3bpkvMyIimWIdJSIiKZl8D37Xrl2xd+9eBAQEoH///njrrbdw4sQJfPfdd+jatasUORIRyQrrKBERScnkBn/+/PnQarUAgISEBGi1WmzatAmtWrXikx+IiIzAOkpERFIyucFv0aKF/r8dHR2xYsUKsyZERCR3rKNERCQlk+/BJyIiIiIi62XUFfwGDRpAoVAYNeHt27cfKSEiIjliHSUioppiVIO/cOFCidMgIpI31lEiIqopRjX4kZGRkgRPTU3Ff/7zHxw+fBhXr15FUlIShg4dql8/btw4rFu3zmCfvn37YufOnZLkQ0QkFdZRIiKqKUbfg19cXIx58+ahR48e6Ny5M2bOnIl79+49UvDc3FwEBQVh2bJlFW7Tr18/XL16Vf/auHHjI8UkIrIU1lEiIqoJRj9FZ86cOYiPj0d4eDjs7e2xaNEi3LhxA59//nm1g0dERCAiIqLSbZRKJTw9Pasdg4jIWrCOEhFRTTC6wf/iiy/wySef4B//+AcA4H//+x8GDBiATz/9FDY20j2MJzk5GR4eHmjQoAF69eqFf//733B3d69w+/z8fOTn5+uXNRoNACArKwsODg6S5alWqw2+Mo71xJHTsTBO9el0OknnNwbraOVKfgYuZkr7s1AyP+M8vnFK5pZLfaupfzt5VlBHyTgKIYQwZkOlUonMzEz4+Pjox+zs7JCZmYmmTZs+eiIKRZl7R7/++ms4ODhApVLh3LlzePfdd+Hk5IT9+/fD1ta23Hni4+ORkJDwyPkQkTzl5OTA2dm5wvUajQYuLi5ITk6Gk5OTUXNqtVqEhoZWOTfrKBHJgSXrKBnH6Cv49+/fh52dncFY3bp1UVhYaPakSrz44ov6/27Xrh3at2+Pli1bIjk5Gb179y53n9jYWMTExOiXNRoNfHx8EBcXh4CAAMlyVavVmDVrFhITE6FSqRjHiuLI6VgYp/p0Oh0mTZok2fzGYB2tXMnPQsyiRPj4SfezcDFTjfnRjGOtcbRHeF6w1jjWUEfJOEY3+EIIjBs3DkqlUj+Wl5eHyZMnw9HRUT/23XffmTfDh7Ro0QINGzZEZmZmhScmpVJpkGMJX19f+Pv7S5ZbCZVKxThWGkdOx8I4ptNqtZLNbSzWUeP4+Kng1066XyQYx7rj5Nx58JXnBeuLYw11lIxjdINf3iPeXn75ZbMmU5VLly7h77//hpeXV43GJSIyB9ZRIiKqCUY3+GvWrDF7cK1Wi8zMTP2yWq1Geno63Nzc4ObmhoSEBDz//PPw9PTEuXPnMGPGDPj5+aFv375mz4WISGqso0REVBOMbvClkJaWhrCwMP1yyT2fkZGRWL58OY4fP45169YhOzsb3t7e6NOnDxITE8t965iI6HHEOkpERKVZtMEPDQ1FZQ/x+emnn2owGyKi2od1lIiISpPuwctERERERFTj2OATEREREckIG3wiIiIiIhlhg09EREREJCNs8ImIiIiIZIQNPhERERGRjLDBJyIiIiKSETb4REREREQywgafiIiIiEhG2OATEREREckIG3wiIiIiIhlhg09EREREJCNs8ImIiIiIZIQNPhERERGRjLDBJyIiIiKSEYs2+KmpqRg0aBC8vb2hUCiwZcuWCredPHkyFAoFFi5cWGP5ERFZO9ZRIiIqzaINfm5uLoKCgrBs2bJKt0tKSsKBAwfg7e1dQ5kREdUOrKNERFRaHUsGj4iIQERERKXbXL58GW+88QZ++uknDBgwoIYyIyKqHVhHiYioNIs2+FUpLi7GmDFjMH36dAQGBhq1T35+PvLz8/XLGo0GAJCVlQUHBwdJ8gQAtVpt8JVxrCeOnI6FcapPp9NJOr+1qo119GKmtD8LJfMzjnXG0fK8YLVxHtc6WhsphBDC0kkAgEKhQFJSEoYOHaofmzt3Lvbs2YOffvoJCoUCzZs3x7Rp0zBt2rQK54mPj0dCQoL0CRNRrZSTkwNnZ+cK12s0Gri4uCA5ORlOTk5GzanVahEaGlrl3FJjHSWimiDnOioXVnsF//Dhw1i0aBGOHDkChUJh9H6xsbGIiYnRL2s0Gvj4+CAuLg4BAQFSpArgwW/Ns2bNQmJiIlQqFeNYURw5HQvjVJ9Op8OkSZMkm98a1dY6GrMoET5+0v0sXMxUY34041hrHO0RnhesNc7jWEdrK6tt8H/99VfcuHEDzZo1048VFRXhrbfewsKFC3H+/Ply91MqlVAqlWXGfX194e/vL1W6eiqVinGsNI6cjoVxTKfVaiWb21rV1jrq46eCXzvpfpFgHOuOk3PnwVeeF6wvzuNYR2srq23wx4wZg/DwcIOxvn37YsyYMRg/fryFsiIiqj1YR4mIHk8WbfC1Wi0yMzP1y2q1Gunp6XBzc0OzZs3g7u5usH3dunXh6emJNm3a1HSqRERWiXWUiIhKs2iDn5aWhrCwMP1yyT2fkZGRWLt2rYWyIiKqPVhHiYioNIs2+KGhoTDlIT4V3S9KRPS4Yh0lIqLSLPqXbImIiIiIyLzY4BMRERERyQgbfCIiIiIiGWGDT0REREQkI2zwiYiIiIhkhA0+EREREZGMsMEnIiIiIpIRNvhERERERDLCBp+IiIiISEbY4BMRERERyQgbfCIiIiIiGWGDT0REREQkI2zwiYiIiIhkhA0+EREREZGMsMEnIiIiIpIRizb4qampGDRoELy9vaFQKLBlyxaD9fHx8fD394ejoyMaNGiA8PBwHDx40DLJEhFZIdZRIiIqzaINfm5uLoKCgrBs2bJy17du3RpLly7FiRMnsHfvXjRv3hx9+vTBzZs3azhTIiLrxDpKRESl1bFk8IiICERERFS4/qWXXjJYnj9/Pj777DMcP34cvXv3ljo9IiKrxzpKRESlWbTBN0VBQQFWrVoFFxcXBAUFVbhdfn4+8vPz9csajQYAkJWVBQcHB8nyU6vVBl8Zx3riyOlYGKf6dDqdpPPXBrWljl7MlPZnoWR+xrHOOFqeF6w2Duto7aEQQghLJwEACoUCSUlJGDp0qMH49u3b8eKLL0Kn08HLywtbtmxB586dK5wnPj4eCQkJEmdLRLVVTk4OnJ2dK1yv0Wjg4uKC5ORkODk5GTWnVqtFaGholXNLjXWUiGqCnOuoXFj9FfywsDCkp6fj1q1bWL16NUaMGIGDBw/Cw8Oj3O1jY2MRExOjX9ZoNPDx8UFcXBwCAgIky1OtVmPWrFlITEyESqWSTZyYRYnw8ZMuzsVMNeZHS3s8cv1/wzim0el0mDRpkmTzW7PaVkdrqu4wjnXG0R6RvibIrb6xjlJpVt/gOzo6ws/PD35+fujatStatWqFzz77DLGxseVur1QqoVQqy4z7+vrC399f6nShUqlkFcfHTwW/dtKd0EvUxPHI7f8N45hGq9VKNre1q211tKbqDuNYZ5ycOw++8rxgfXEe5zpa29S65+AXFxcb3BtKRESmYR0lIpI3i17B12q1yMzM1C+r1Wqkp6fDzc0N7u7umDNnDgYPHgwvLy/cunULy5Ytw+XLlzF8+HALZk1EZD1YR4mIqDSLNvhpaWkICwvTL5fc8xkZGYkVK1bg9OnTWLduHW7dugV3d3d07twZv/76KwIDAy2VMhGRVWEdJSKi0iza4IeGhqKyh/h89913NZgNEVHtwzpKRESl1bp78ImIiIiIqGJs8ImIiIiIZIQNPhERERGRjLDBJyIiIiKSETb4REREREQywgafiIiIiEhG2OATEREREckIG3wiIiIiIhlhg09EREREJCNs8ImIiIiIZIQNPhERERGRjLDBJyIiIiKSETb4REREREQywgafiIiIiEhG2OATEREREcmIRRv81NRUDBo0CN7e3lAoFNiyZYt+XWFhId555x20a9cOjo6O8Pb2xtixY3HlyhXLJUxEZGVYR4mIqDSLNvi5ubkICgrCsmXLyqzT6XQ4cuQIZs2ahSNHjuC7777DmTNnMHjwYAtkSkRknVhHiYiotDqWDB4REYGIiIhy17m4uGDXrl0GY0uXLkWXLl1w4cIFNGvWrCZSJCKyaqyjRERUmkUbfFPl5ORAoVDA1dW1wm3y8/ORn5+vX9ZoNACArKwsODg4SJabWq02+CqXOBczpY1TMr+UxyPX/zeMYxqdTifp/LVFbaijNVV3GMc642hroCbIrb6xjlJpCiGEsHQSAKBQKJCUlIShQ4eWuz4vLw89evSAv78/1q9fX+E88fHxSEhIkChLIqrtcnJy4OzsXOF6jUYDFxcXJCcnw8nJyag5tVotQkNDq5xbaqyjRFQT5FxH5aJWXMEvLCzEiBEjIITA8uXLK902NjYWMTEx+mWNRgMfHx/ExcUhICBAshzVajVmzZqFxMREqFQqxrGiOHI6FsapPp1Oh0mTJkk2v7VjHWUcxqnZGHKM87jX0drE6hv8kpNSVlYWfvnllyp/q1MqlVAqlWXGfX194e/vL1WaeiqVinGsNI6cjoVxTKfVaiWb29qxjjIO41guhpziPM51tLax6ga/5KSUkZGBPXv2wN3d3dIpERHVKqyjRESPH4s2+FqtFpmZmfpltVqN9PR0uLm5wcvLCy+88AKOHDmC7du3o6ioCNeuXQMAuLm5oV69epZKm4jIarCOEhFRaRZt8NPS0hAWFqZfLrnnMzIyEvHx8di2bRsAoEOHDgb77dmzB6GhoTWVJhGR1WIdJSKi0iza4IeGhqKyh/hYyQN+iIisFusoERGVZtG/ZEtERERERObFBp+IiIiISEbY4BMRERERyQgbfCIiIiIiGWGDT0REREQkI2zwiYiIiIhkhA0+EREREZGMsMEnIiIiIpIRNvhERERERDLCBp+IiIiISEbY4BMRERERyQgbfCIiIiIiGWGDT0REREQkI2zwiYiIiIhkhA0+EVEt0rNnT2zYsEHyOLdu3YKHhwcuXbokeSwiopr0ONRRizb4qampGDRoELy9vaFQKLBlyxaD9d999x369OkDd3d3KBQKpKenWyRPIiJzqqr2VWTbtm24fv06XnzxRf2Yp6cnAgMDy8wVHx+PDh066Ouog4ODvo42b94cCoWiwte4cePQsGFDjB07FrNnz5bgO0BE9GjMWUebN2+OhQsXltm2pI6Wt2ztddSiDX5ubi6CgoKwbNmyCtc//fTTmDdvXg1nRkQknapqX0UWL16M8ePHw8bm/0q3EAJNmjSpso6Gh4frxw4dOoSrV6/i6tWr2Lx5MwDgzJkz+rFFixYBAMaPH4/169fj9u3bph4iEZGkzFlHq8Pa62idGo1WSkREBCIiIipcP2bMGADA+fPnaygjIiLpVVX7ynPz5k388ssv+pNGCXt7e/Tv3x/Dhg0rd7+SOjpt2jT9WKNGjfT/7ebmBgDw8PCAq6urwb4l7wxs377dpFyJiKRmzjpaHabW0aSkJLzyyiuPHNdYFm3wpZCfn4/8/Hz9ck5ODgDg7NmzksbNysoCAJw6dQo6nY5xrCiOnI6Fcarv3r17AB5c8TZGbm6u0XOXbKvRaAzGlUollEql0fNUZu/evXBwcEBAQIBZ5qvMw3U0ODgYycnJAFhHGUc+ceR0LDUZh3W0erp06YJff/21Rht8CCsBQCQlJZW7Tq1WCwDi6NGjVc4ze/ZsAYAvvvjiq9zXxYsXK60h9+7dE56enibP6+TkVGZs9uzZj1T7HrZgwQLRokWLMuO+vr6iXr16wtHRUQAQSqVSODo6irp164qgoCD9dtHR0QIoW0f37NkjAIg7d+7ox1hH+eKLr8pecq6jD79K19HZs2cbLJcor44+7M033xShoaFV5mdOsruCHxsbi5iYGP1ydnY2fH19ceHCBbi4uEgWV6PRwMfHBxcvXoSzszPjWFEcOR0L41SfEAJ3796Ft7d3pdvZ2dlBrVajoKDA5PkVCoXBmLmuOgEPrpzZ2dmVu2769OkYN24cWrVqhQULFuDZZ5/F4sWLkZqaWq1YD9fRf/3rX9i1axeOHz/OOso4sokjp2OpyTiPQx192KPU0YfZ29tL+s5KeWTX4Ff0Vo6Li4ukP/QlnJ2dGcdK48jpWBineoxtTu3s7Co8CVhKw4YNcefOnQrX+fn5AQC8vLzg5+envye0Oh6uo7m5ufD09MTx48dZRxlHdnHkdCw1FedxqKMlHqWOPuz27dsG9+zXBD4Hn4ioFggODsa1a9cqPDlJ5Y8//kBQUFCNxiQikoIl62hwcHCNxrRog6/VapGenq5/vr1arUZ6ejouXLgA4MFvPOnp6fjzzz8BPHj8UHp6Oq5du2aplImIHllVta88wcHBaNiwIfbt22cwXlxcjMuXL5eZq+QBAyV19ObNmwBMq6M6nQ6HDx9Gr169qnGURETSMWcdlVJJHe3Tp0+NxQQAi37ItuRDCaVfkZGRQggh1qxZU+56Yz5wUSIvL0/Mnj1b5OXlSXMQjGP1ceR0LIwjD1XVvorMmDFDvPjiiwZjjRs3LneuoKAgERQUVGUdrezDYRs2bBBt2rSR3c8C4zCOnI6lJuNYE3PWUV9fX7FgwYIy25b+UG11PmRbUkdrmkIII591REREFnXt2jUEBgbiyJEj8PX1lTxe165dMXXqVLz00kuSxyIiqgmPSx3lPfhERLWEp6cnPvvss0rfgjaXW7du4bnnnsOoUaMkj0VEVFMelzrKK/hERERERDLCK/hERERERDLCBp+IiIiISEZk3eDv378ftra2GDBggGQx5s6di86dO6N+/frw8PDA0KFDcebMGbPGWL58Odq3b6//AxbdunXDjh07zBqjPB988AEUCgWmTZtm1nnj4+OhUCgMXv7+/maNUeLy5ct4+eWX4e7uDnt7e7Rr1w5paWlmjdG8efMyx6NQKBAVFWXWOEVFRZg1axZUKhXs7e3RsmVLJCYmQoq77O7evYtp06bB19cX9vb26N69Ow4dOvRIc6ampmLQoEHw9vaGQqHAli1bDNYLIfDee+/By8sL9vb2CA8PR0ZGxiPFJPOQupbWRB0FLFNLWUeNwzpqHNZRMpasG/zPPvsMb7zxBlJTU3HlyhVJYqSkpCAqKgoHDhzArl27UFhYiD59+iA3N9dsMZo2bYoPPvgAhw8fRlpaGnr16oUhQ4bg5MmTZotR2qFDh7By5Uq0b99ekvkDAwNx9epV/Wvv3r1mj3Hnzh306NEDdevWxY4dO/Dnn3/i448/RoMGDcwa59ChQwbHsmvXLgDA8OHDzRpn3rx5WL58OZYuXYpTp05h3rx5+PDDD7FkyRKzxgGAiRMnYteuXfjyyy9x4sQJ9OnTB+Hh4bh8+XK158zNzUVQUBCWLVtW7voPP/wQixcvxooVK3Dw4EE4Ojqib9++yMvLq3ZMMg+pa2lN1FGg5msp66jxWEeNwzpKRqvxB3PWkLt37wonJydx+vRpMXLkSDFnzpwaiXvjxg0BQKSkpEgap0GDBuLTTz+VZO67d++KVq1aiV27domQkBARHR1t1vkreo6sub3zzjvi6aefljxOadHR0aJly5aiuLjYrPMOGDBATJgwwWDsueeeE6NHjzZrHJ1OJ2xtbcX27dsNxp988kkRFxdnlhgARFJSkn65uLhYeHp6iv/85z/6sezsbKFUKsXGjRvNEpOqxxK1tKbqqBDS1VLW0UfDOlo11lGqjGyv4H/zzTfw9/dHmzZt8PLLL+Pzzz+X5C240kr+eqSbm5sk8xcVFeHrr79Gbm4uunXrJkmMqKgoDBgwAOHh4ZLMDwAZGRnw9vZGixYtMHr0aEkeV7Vt2zZ06tQJw4cPh4eHB4KDg7F69Wqzx3lYQUEBvvrqK0yYMAEKhcKsc3fv3h27d+/G2bNnAQDHjh3D3r17ERERYdY49+/fR1FREezs7AzG7e3tJblCCDz4C4TXrl0z+JlzcXHBU089hf3790sSk4xjiVoqdR0FpK+lrKPVxzpaPayjZMDSv2FIpXv37mLhwoVCCCEKCwtFw4YNxZ49eySNWVRUJAYMGCB69Ohh9rmPHz8uHB0dha2trXBxcRE//PCD2WMIIcTGjRtF27Ztxb1794QQQpIrTz/++KP45ptvxLFjx8TOnTtFt27dRLNmzYRGozFrHKVSKZRKpYiNjRVHjhwRK1euFHZ2dmLt2rVmjfOwTZs2CVtbW3H58mWzz11UVCTeeecdoVAoRJ06dYRCoRDvv/++2eMIIUS3bt1ESEiIuHz5srh//7748ssvhY2NjWjdurVZ5kepK0/79u0TAMSVK1cMths+fLgYMWKEWWJS9dR0LZWyjgpRM7WUdfTRsI4ah3WUKiPLBv/06dOiTp064vr16/qxqKgo8fLLL0sad/LkycLX11dcvHjR7HPn5+eLjIwMkZaWJmbOnCkaNmwoTp48adYYFy5cEB4eHuLYsWP6MSlOTKXduXNHODs7m/1t8rp164pu3boZjL3xxhuia9euZo3zsD59+oiBAwdKMvfGjRtF06ZNxcaNG8Xx48fFF198Idzc3CQ50WZmZoqePXsKAMLW1lZ07txZjB49Wvj7+5tlfp6YagdL1FIp66gQ0tdS1tFHxzpqHNZRqowsG/zp06fr/0GVvGxsbIS9vb3Izs6WJGZUVJRo2rSp+OuvvySZv7TevXuLSZMmmXXOpKSkMt83AEKhUAhbW1tx//59s8Z7WKdOncTMmTPNOmezZs3EK6+8YjD2ySefCG9vb7PGKXH+/HlhY2MjtmzZIsn8TZs2FUuXLjUYS0xMFG3atJEknhBCaLVa/clixIgRon///maZt/SJ6dy5cwKAOHr0qMF2PXv2FFOnTjVLTDJdTdfSmq6jQpi/lrKOPhrWUeOxjlJlZHcP/v379/HFF1/g448/Rnp6uv517NgxeHt7Y+PGjWaNJ4TAlClTkJSUhF9++QUqlcqs81ekuLgY+fn5Zp2zd+/eOHHihMH3rVOnThg9ejTS09Nha2tr1ngltFotzp07By8vL7PO26NHjzKP2jt79ix8fX3NGqfEmjVr4OHhIdmjBHU6HWxsDP/J2traori4WJJ4AODo6AgvLy/cuXMHP/30E4YMGSJJHJVKBU9PT+zevVs/ptFocPDgQck+a0KVq8laaqk6Cpi/lrKOPhrW0epjHSUDlv4Nw9ySkpJEvXr1yr26NGPGDNGpUyezxnvttdeEi4uLSE5OFlevXtW/dDqd2WLMnDlTpKSkCLVaLY4fPy5mzpwpFAqF+Pnnn80WoyJSvLX81ltvieTkZKFWq8W+fftEeHi4aNiwobhx44ZZ4/z++++iTp06Ys6cOSIjI0OsX79eODg4iK+++sqscYR4cF9ns2bNxDvvvGP2uUtERkaKJk2aiO3btwu1Wi2+++470bBhQzFjxgyzx9q5c6fYsWOH+Ouvv8TPP/8sgoKCxFNPPSUKCgqqPefdu3fF0aNHxdGjRwUAMX/+fHH06FGRlZUlhBDigw8+EK6urmLr1q3i+PHjYsiQIUKlUunvY6aaVZO1tCbqqBCWq6Wso8ZhHa0a6ygZS3YN/sCBAyt8++vgwYMCgMG9kY8KQLmvNWvWmC3GhAkThK+vr6hXr55o1KiR6N27d40090JIc2IaOXKk8PLyEvXq1RNNmjQRI0eOFJmZmWaNUeL7778Xbdu2FUqlUvj7+4tVq1ZJEuenn34SAMSZM2ckmV8IITQajYiOjhbNmjUTdnZ2okWLFiIuLk7k5+ebPdamTZtEixYtRL169YSnp6eIiop65Fsy9uzZU+6/lcjISCHEg0e8zZo1SzRu3FgolUrRu3dvSb+fVLmarKU1UUeFsFwtZR01Duto1VhHyVgKIWrg2ZFERERERFQjZHcPPhERERHR44wNPhERERGRjLDBJyIiIiKSETb4REREREQywgafiIiIiEhG2OATEREREckIG3wiIiIiIhlhg08W17x5cyxcuNAisceNG4ehQ4fWaEyFQoEtW7bUaEwikjfWUSJ6GBt8KmPcuHFQKBRQKBSoW7cuGjdujGeffRaff/45iouLLZ2eSVavXo2goCA4OTnB1dUVwcHBmDt3rkVzunr1KiIiIiyaAxFJi3VUWqyjRJVjg0/l6tevH65evYrz589jx44dCAsLQ3R0NAYOHIj79+9bOj0DBQUF5Y5//vnnmDZtGqZOnYr09HTs27cPM2bMgFarreEMDXl6ekKpVFo0ByKSHuuodFhHiSrHBp/KpVQq4enpiSZNmuDJJ5/Eu+++i61bt2LHjh1Yu3atfrvs7GxMnDgRjRo1grOzM3r16oVjx47p1587dw5DhgxB48aN4eTkhM6dO+N///tfpbGrmjM+Ph4dOnTAp59+CpVKBTs7u3Ln2bZtG0aMGIFXXnkFfn5+CAwMxKhRozBnzpwy23700Ufw8vKCu7s7oqKiUFhYqF93584djB07Fg0aNICDgwMiIiKQkZEBABBCoFGjRvj222/123fo0AFeXl765b1790KpVEKn0wEwfGv5/PnzUCgU+O677xAWFgYHBwcEBQVh//79BvmtXr0aPj4+cHBwwLBhwzB//ny4urpW+n0kIstiHWUdJbIUNvhktF69eiEoKAjfffedfmz48OG4ceMGduzYgcOHD+PJJ59E7969cfv2bQCAVqtF//79sXv3bhw9ehT9+vXDoEGDcOHChQrjVDUnAGRmZmLz5s347rvvkJ6eXu48np6eOHDgALKysio9rj179uDcuXPYs2cP1q1bh7Vr1xqcfMeNG4e0tDRs27YN+/fvhxAC/fv3R2FhIRQKBXr27Ink5GQAD05ip06dwr1793D69GkAQEpKCjp37gwHB4cKc4iLi8Pbb7+N9PR0tG7dGqNGjdJf4du3bx8mT56M6OhopKen49lnny335EpE1o91lHWUqEYIolIiIyPFkCFDyl03cuRIERAQIIQQ4tdffxXOzs4iLy/PYJuWLVuKlStXVjh/YGCgWLJkiX7Z19dXLFiwwOg5Z8+eLerWrStu3LhR6XFcuXJFdO3aVQAQrVu3FpGRkWLTpk2iqKjI4Fh9fX3F/fv39WPDhw8XI0eOFEIIcfbsWQFA7Nu3T7/+1q1bwt7eXnzzzTdCCCEWL14sAgMDhRBCbNmyRTz11FNiyJAhYvny5UIIIcLDw8W7776r3x+ASEpKEkIIoVarBQDx6aef6tefPHlSABCnTp0SQjz4ng8YMMDg2EaPHi1cXFwqPX4ishzWUdZRIkviFXwyiRACCoUCAHDs2DFotVq4u7vDyclJ/1Kr1Th37hyAB1ee3n77bQQEBMDV1RVOTk44depUhVeejJkTAHx9fdGoUaNKc/Xy8sL+/ftx4sQJREdH4/79+4iMjES/fv0MPuQWGBgIW1tbg/1u3LgBADh16hTq1KmDp556Sr/e3d0dbdq0walTpwAAISEh+PPPP3Hz5k2kpKQgNDQUoaGhSE5ORmFhIX777TeEhoZWmmv79u0N4gPQ53DmzBl06dLFYPvSy0RUe7COso4SSa2OpROg2uXUqVNQqVQAHpx0vLy89G+rPqzkvsa3334bu3btwkcffQQ/Pz/Y29vjhRdeqPADXcbMCQCOjo5G59y2bVu0bdsWr7/+OiZPnoxnnnkGKSkpCAsLAwDUrVvXYHuFQmHSUy7atWsHNzc3pKSkICUlBXPmzIGnpyfmzZuHQ4cOobCwEN27d690jodzKDnx17YnbRCRcVhHy2IdJTIvNvhktF9++QUnTpzAm2++CQB48sknce3aNdSpUwfNmzcvd599+/Zh3LhxGDZsGIAHJ57z589XGMOYOR/FE088AQDIzc01avuAgADcv38fBw8e1J9c/v77b5w5c0Y/l0KhwDPPPIOtW7fi5MmTePrpp+Hg4ID8/HysXLkSnTp1MulEWlqbNm1w6NAhg7HSy0RUO7COso4S1QTeokPlys/Px7Vr13D58mUcOXIE77//PoYMGYKBAwdi7NixAIDw8HB069YNQ4cOxc8//4zz58/jt99+Q1xcHNLS0gAArVq10n+A69ixY3jppZcqvaJizJzGeu2115CYmIh9+/YhKysLBw4cwNixY9GoUSN069bNqDlatWqFIUOG4NVXX8XevXtx7NgxvPzyy2jSpAmGDBmi3y40NBQbN25Ehw4d4OTkBBsbG/Ts2RPr169HSEiISXmX9sYbb+DHH3/E/PnzkZGRgZUrV2LHjh36K1REZJ1YR6HPn3WUqGaxwady7dy5E15eXmjevDn69euHPXv2YPHixdi6dav+PkuFQoEff/wRPXv2xPjx49G6dWu8+OKLyMrKQuPGjQEA8+fPR4MGDdC9e3cMGjQIffv2xZNPPllhXGPmNFZ4eDgOHDiA4cOHo3Xr1nj++edhZ2eH3bt3w93d3eh51qxZg44dO2LgwIHo1q0bhBD48ccfDd4ODgkJQVFRkcE9oqGhoWXGqqNHjx5YsWIF5s+fj6CgIOzcuRNvvvlmhY+1IyLrwDr6f1hHiWqWQgghLJ0EEZnm1VdfxenTp/Hrr79aOhUiolqJdZTkjPfgE9UCH330EZ599lk4Ojpix44dWLduHT755BNLp0VEVGuwjtLjhFfwiWqBESNGIDk5GXfv3kWLFi3wxhtvYPLkyZZOi4io1mAdpccJG3wiIiIiIhnhh2yJiIiIiGSEDT4RERERkYywwSciIiIikhE2+EREREREMsIGn4iIiIhIRtjgExERERHJCBt8IiIiIiIZYYNPRERERCQjbPCJiIiIiGTk/wHYGsuLw/t0ywAAAABJRU5ErkJggg==\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "##### Alternativa ϵ-greedy policy\n",
        "\n",
        "\n",
        "Sotto, propongo un codice alternativo per l'implementazione di una policy epsilon-greedy, che risulta essere più intuitivo di quello proposto in precedenza.\n",
        "\n"
      ],
      "metadata": {
        "id": "XTc4HZsADrBu"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "def generate_episode_soft_policy(s_pi):\n",
        "    state = env.reset()\n",
        "\n",
        "    episode = []\n",
        "\n",
        "    while True:\n",
        "\n",
        "        action = np.random.choice([0, 1])\n",
        "\n",
        "        if state in s_pi:\n",
        "            supp = [s_pi[state], int(not(s_pi[state]))] # [0,1] oppure [1,0]\n",
        "            action = np.random.choice(supp, p=[0.8, 0.2])  # p = probabilità di estrazione, nel 80% dei casi azione è \"greedy\"\n",
        "\n",
        "\n",
        "        next_state, reward, terminate, _ = env.step(action)\n",
        "        episode.append((state, action, reward))\n",
        "\n",
        "        if terminate:\n",
        "            break\n",
        "\n",
        "        state = next_state\n",
        "\n",
        "    return episode\n",
        "\n",
        "#\n",
        "# Training loop\n",
        "#\n",
        "\n",
        "n_actions = env.action_space.n  # numero di azioni disponibili\n",
        "\n",
        "# Fase di inizializzazione delle strutture dati utilizzate\n",
        "e2_policy =  {}\n",
        "\n",
        "Q_table = defaultdict(lambda: np.zeros(env.action_space.n))\n",
        "\n",
        "returns_sum = defaultdict(float) # Nota:  returns(state, action)\n",
        "returns_count = defaultdict(int)\n",
        "\n",
        "# Parametri\n",
        "num_episodes = 500_000\n",
        "\n",
        "\n",
        "for episode in range(num_episodes):\n",
        "  trajectory = generate_episode_soft_policy(e2_policy)\n",
        "  G, T = 0, len(trajectory)\n",
        "\n",
        "  visited_state = []\n",
        "  for i in range(T):\n",
        "    visited_state.append((trajectory[i][0],trajectory[i][1]))\n",
        "\n",
        "  for i in range(T-1, -1, -1):\n",
        "       state, action, reward = trajectory[i]\n",
        "       G += reward\n",
        "\n",
        "       visited_state.pop()\n",
        "       if (state, action) not in visited_state: # A meno che la coppia (state, action) non sia stata già visitata in questo episodio, fai:\n",
        "\n",
        "                returns_sum[(state, action)] += G\n",
        "                returns_count[(state, action)] += 1\n",
        "\n",
        "                Q_table[state][action] = returns_sum[(state, action)] / returns_count[(state, action)]\n",
        "\n",
        "                best_action = np.argmax(Q_table[state])\n",
        "                e2_policy[state] = best_action # update\n",
        "                # alternativa:\n",
        "                # best_action = max(list(range(env.action_space.n)), key = lambda x: Q_table[state][x])\n",
        "\n",
        "  if (episode!= 0) and (episode % 100_000 == 0) :\n",
        "               print(\"Q_TABLE                                              COUNT    0         1   \")\n",
        "               print(\"Q_table[(17, 5, True)] = \", Q_table[(17, 5, True)],\" Count : \",returns_count[((17, 5, True), 0)], \" -  \",returns_count[((17, 5, True), 1)] )\n",
        "               print(\"Q_table[(17, 8, True)] = \", Q_table[(17, 8, True)],\" Count : \",returns_count[((17, 8, True), 0)], \" -  \",returns_count[((17, 8, True), 1)] )\n",
        "               print(\"Q_table[(18, 9, True)] = \", Q_table[(18, 9, True)],\" Count : \",returns_count[((18, 9, True), 0)], \" -  \",returns_count[((18, 9, True), 1)] )\n",
        "\n",
        "               print(\"Q_table[(13, 5, False)] = \", Q_table[(13, 5, False)],\" Count : \",returns_count[((13, 5, False), 0)], \" -  \",returns_count[((13, 5, False), 1)] )\n",
        "               print(\"Q_table[(13, 6, False)] = \", Q_table[(13, 6, False)],\" Count : \",returns_count[((13, 6, False), 0)], \" -  \",returns_count[((13, 6, False), 1)] )\n",
        "               print(\"Q_table[(13, 7, False)] = \", Q_table[(13, 7, False)],\" Count : \",returns_count[((13, 7, False), 0)], \" -  \",returns_count[((13, 7, False), 1)] )\n",
        "\n",
        "               print(\"Q_table[(11, 10, False)] = \", Q_table[(11, 10, False)],\" Count : \",returns_count[((11, 10, False), 0)], \" -  \",returns_count[((11, 10, False), 1)] )\n",
        "               #print(\"\\ne_policy\")\n",
        "               #print(e2_policy)\n",
        "               print(\" \\n\")\n",
        "\n",
        "\n",
        "plot_policy(e2_policy)"
      ],
      "metadata": {
        "id": "KNtcfyD2GfJ6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "69b2c974-8b22-4074-f275-f74ad3b9ac26"
      },
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.10/dist-packages/ipykernel/ipkernel.py:283: DeprecationWarning: `should_run_async` will not call `transform_cell` automatically in the future. Please pass the result to `transformed_cell` argument and any exception that happen during thetransform in `preprocessing_exc_tuple` in IPython 7.17 and above.\n",
            "  and should_run_async(code)\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Q_TABLE                                              COUNT    0         1   \n",
            "Q_table[(17, 5, True)] =  [-0.0375     -0.09090909]  Count :  80  -   33\n",
            "Q_table[(17, 8, True)] =  [-0.28125     0.00900901]  Count :  32  -   111\n",
            "Q_table[(18, 9, True)] =  [-0.17142857 -0.38709677]  Count :  105  -   31\n",
            "Q_table[(13, 5, False)] =  [-0.14237856 -0.3045977 ]  Count :  597  -   174\n",
            "Q_table[(13, 6, False)] =  [-0.15072934 -0.34841629]  Count :  617  -   221\n",
            "Q_table[(13, 7, False)] =  [-0.4484127  -0.51226158]  Count :  504  -   367\n",
            "Q_table[(11, 10, False)] =  [-0.49700599 -0.11310861]  Count :  334  -   1335\n",
            " \n",
            "\n",
            "Q_TABLE                                              COUNT    0         1   \n",
            "Q_table[(17, 5, True)] =  [-0.06521739 -0.28571429]  Count :  184  -   56\n",
            "Q_table[(17, 8, True)] =  [-0.43396226 -0.12871287]  Count :  53  -   202\n",
            "Q_table[(18, 9, True)] =  [-0.16587678 -0.54545455]  Count :  211  -   55\n",
            "Q_table[(13, 5, False)] =  [-0.16282764 -0.34029851]  Count :  1259  -   335\n",
            "Q_table[(13, 6, False)] =  [-0.15756136 -0.34126984]  Count :  1263  -   378\n",
            "Q_table[(13, 7, False)] =  [-0.4867664 -0.4303653]  Count :  869  -   876\n",
            "Q_table[(11, 10, False)] =  [-0.55869873 -0.12338157]  Count :  707  -   2626\n",
            " \n",
            "\n",
            "Q_TABLE                                              COUNT    0         1   \n",
            "Q_table[(17, 5, True)] =  [-0.11029412 -0.22093023]  Count :  272  -   86\n",
            "Q_table[(17, 8, True)] =  [-0.31578947 -0.17218543]  Count :  76  -   302\n",
            "Q_table[(18, 9, True)] =  [-0.21221865 -0.4875    ]  Count :  311  -   80\n",
            "Q_table[(13, 5, False)] =  [-0.1383812 -0.3497053]  Count :  1915  -   509\n",
            "Q_table[(13, 6, False)] =  [-0.15458422 -0.40213523]  Count :  1876  -   562\n",
            "Q_table[(13, 7, False)] =  [-0.48653846 -0.41703751]  Count :  1040  -   1573\n",
            "Q_table[(11, 10, False)] =  [-0.56966825 -0.12496857]  Count :  1055  -   3977\n",
            " \n",
            "\n",
            "Q_TABLE                                              COUNT    0         1   \n",
            "Q_table[(17, 5, True)] =  [-0.11813187 -0.23300971]  Count :  364  -   103\n",
            "Q_table[(17, 8, True)] =  [-0.33018868 -0.13874346]  Count :  106  -   382\n",
            "Q_table[(18, 9, True)] =  [-0.21770335 -0.47115385]  Count :  418  -   104\n",
            "Q_table[(13, 5, False)] =  [-0.1520125  -0.33783784]  Count :  2559  -   666\n",
            "Q_table[(13, 6, False)] =  [-0.14516129 -0.39178082]  Count :  2480  -   730\n",
            "Q_table[(13, 7, False)] =  [-0.49174917 -0.4057322 ]  Count :  1212  -   2233\n",
            "Q_table[(11, 10, False)] =  [-0.57562724 -0.11504095]  Count :  1395  -   5372\n",
            " \n",
            "\n"
          ]
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 800x800 with 4 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Concludendo\n",
        "\n",
        "\n",
        "Un fatto importante dei metodi Monte Carlo è che le stime per ogni stato sono indipendenti tra loro. Questo significa che la stima per uno stato non dipende dalla stima di nessun altro stato. In altre parole, i metodi Monte Carlo non richiedono il \"bootstrap\" come nella programmazione dinamica.\n",
        "\n",
        "*Si noti che l'onere computazionale per la stima del valore di un singolo stato è indipendente dal numero di stati totale.Questo rende i metodi Monte Carlo particolarmente attraenti quando si desidera stimare solo il valore di un sottoinsieme di stati. È possibile generare molti episodi campione partendo dagli stati di interesse e calcolare i return medi solo di questi stati, ignorando tutti gli altri.*\n",
        "\n",
        "$\\;$\n",
        "\n",
        "\"Caveat\", ovvero, fattori da tenere in considerazione quando si utilizzano questi metodi:\n",
        "\n",
        "1. Può essere applicato solo agli MDP con episodi terminali.\n",
        "\n",
        "2.  È necessario attendere la fine dell'episodio per calcolare il return\n",
        "\n",
        "3. In generale, presentano una varianza elevata, ma consistente in condizioni moderatamente stabili.\n",
        "\n"
      ],
      "metadata": {
        "id": "Zc9-zMAObK_X"
      }
    }
  ]
}