misc/notebooks/Dataset_example.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "227f3a01-a1fd-42f0-af24-c70bf5ec3c3e",
   "metadata": {},
   "source": [
    "# Several Sentences Dataset example"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b1830a5e-b0b9-484c-9991-08d65195d0f5",
   "metadata": {},
   "outputs": [],
   "source": [
    "import sys\n",
    "sys.path.append(\"../../my_lib/src/Quantum\")\n",
    "sys.path.append(\"../../my_lib/src/DataPreparation\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "582432a7-e81e-4808-bcaa-99d7f649e55c",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "import pandas as pd\n",
    "import json\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import random\n",
    "import dictionary\n",
    "import sentence\n",
    "import Optimizer as optimizer\n",
    "import loader\n",
    "import Circuit as circuit"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "26966b08-f5d5-4ff2-a89f-7af75f3f912d",
   "metadata": {},
   "source": [
    "## Processing the .json file and preparing the dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fd293c4f-eb3b-4afa-b69c-c141b658225f",
   "metadata": {},
   "source": [
    "Several datasets are found in the Dataset folder. \n",
    "In the complete dataset, three types of sentences are found:\n",
    "* NOUN-TVERB-NOUN\n",
    "* NOUN-IVERB\n",
    "* NOUN-IVERB-PREP-NOUN\n",
    "\n",
    "In the dataset used in this notebook example, only transitive sentences are used and the vocabulary is modified from the original Data Preparation module files.\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "9a692b38-6a05-40ae-9809-c12b16bc5740",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "filename = 'Expanded_Transitive_dataset.json'\n",
    "Dftrain, Dftest = loader.createdf(\"../../my_lib/src/Quantum/Datasets/\"+filename)#Dataframe form Json file\n",
    "#Several datasets are found in the Dataset folder. \n",
    "#In the complete dataset, three types of sentences are found:\n",
    "# \n",
    "\n",
    "Myvocab = loader.getvocabdict(Dftrain, Dftest)#A dictionary with word and categories found in dataset\n",
    "\n",
    "MyDict = dictionary.QuantumDict(qn=1,#Number of qubits for noun category. \n",
    "                                qs=1)#Number of qubits for sentence category\n",
    "MyDict.addwords(myvocab=Myvocab)\n",
    "MyDict.setvocabparams()#Randomly assign params to words following and ansatz structure."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "1025e6b9-40e8-483a-9fae-25fa60f0678e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<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>sentence</th>\n",
       "      <th>sentence_type</th>\n",
       "      <th>truth_value</th>\n",
       "      <th>sentence_type_code</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>dog eats bone</td>\n",
       "      <td>NOUN-TVERB-NOUN</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>mouse eats fish</td>\n",
       "      <td>NOUN-TVERB-NOUN</td>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>cat eats bone</td>\n",
       "      <td>NOUN-TVERB-NOUN</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>bird flees cat</td>\n",
       "      <td>NOUN-TVERB-NOUN</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>dog flees mouse</td>\n",
       "      <td>NOUN-TVERB-NOUN</td>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "          sentence    sentence_type  truth_value  sentence_type_code\n",
       "0    dog eats bone  NOUN-TVERB-NOUN         True                   0\n",
       "1  mouse eats fish  NOUN-TVERB-NOUN        False                   0\n",
       "2    cat eats bone  NOUN-TVERB-NOUN         True                   0\n",
       "3   bird flees cat  NOUN-TVERB-NOUN         True                   0\n",
       "4  dog flees mouse  NOUN-TVERB-NOUN        False                   0"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Dftrain.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "67aadd04-ba23-4840-a2fa-aeceb6a438bd",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<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>sentence</th>\n",
       "      <th>sentence_type</th>\n",
       "      <th>truth_value</th>\n",
       "      <th>sentence_type_code</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>dog chases mouse</td>\n",
       "      <td>NOUN-TVERB-NOUN</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>bird eats bird</td>\n",
       "      <td>NOUN-TVERB-NOUN</td>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>cat eats dog</td>\n",
       "      <td>NOUN-TVERB-NOUN</td>\n",
       "      <td>False</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>cat chases mouse</td>\n",
       "      <td>NOUN-TVERB-NOUN</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>dog eats meat</td>\n",
       "      <td>NOUN-TVERB-NOUN</td>\n",
       "      <td>True</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "           sentence    sentence_type  truth_value  sentence_type_code\n",
       "0  dog chases mouse  NOUN-TVERB-NOUN         True                   0\n",
       "1    bird eats bird  NOUN-TVERB-NOUN        False                   0\n",
       "2      cat eats dog  NOUN-TVERB-NOUN        False                   0\n",
       "3  cat chases mouse  NOUN-TVERB-NOUN         True                   0\n",
       "4     dog eats meat  NOUN-TVERB-NOUN         True                   0"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Dftest.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "64956963-71fc-4422-ac01-1c696702dcf5",
   "metadata": {},
   "source": [
    "-----\n",
    "For creating the sentences the following ansatz was used.\n",
    "* qn = 1\n",
    "* qs = 1\n",
    "\n",
    "More qubits are shown in the circuit figure for a better understanding of Ansatz shape\n",
    "<img src=\"SentenceAnsatz.png\">"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "fc34a034-49cf-4736-9fab-69a51f41f6b1",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "def createsentencelist(dftrain, mydict):\n",
    "    sentences_list = []\n",
    "    for i, DataInstance in dftrain.iterrows():\n",
    "        a_sentence = sentence.Sentence(DataInstance,\n",
    "                                       dataset=True,\n",
    "                                       dictionary=mydict)\n",
    "\n",
    "        a_sentence.getqbitcontractions()\n",
    "        a_sentence.setparamsfrommodel(mydict)\n",
    "        sentences_list.append(a_sentence)\n",
    "\n",
    "    return sentences_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "51127958-9c07-4c4d-98e5-c5e3f6f49bff",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "iteration 10 \n",
      " Cost: 0.4249097822351392\n",
      "iteration 20 \n",
      " Cost: 0.4044928751089384\n",
      "iteration 30 \n",
      " Cost: 0.5966064079766388\n",
      "iteration 40 \n",
      " Cost: 0.37999511293833466\n",
      "iteration 50 \n",
      " Cost: 0.3650454453377588\n",
      "iteration 60 \n",
      " Cost: 0.3784365859980766\n",
      "iteration 70 \n",
      " Cost: 0.34802458981037054\n",
      "iteration 80 \n",
      " Cost: 0.31853631993430365\n",
      "iteration 90 \n",
      " Cost: 0.3662310739525147\n",
      "iteration 100 \n",
      " Cost: 0.30522736449653187\n",
      "iteration 110 \n",
      " Cost: 0.30039293049231514\n",
      "iteration 120 \n",
      " Cost: 0.2733249978302871\n",
      "iteration 130 \n",
      " Cost: 0.26477023809905254\n",
      "iteration 140 \n",
      " Cost: 0.22538271111985617\n",
      "iteration 150 \n",
      " Cost: 0.22373424945826334\n",
      "iteration 160 \n",
      " Cost: 0.2381827227566488\n",
      "iteration 170 \n",
      " Cost: 0.21966755133512111\n",
      "iteration 180 \n",
      " Cost: 0.22123338520459068\n",
      "iteration 190 \n",
      " Cost: 0.21093640903698044\n",
      "iteration 200 \n",
      " Cost: 0.21008400217920892\n",
      "iteration 210 \n",
      " Cost: 0.20917975795126517\n",
      "iteration 220 \n",
      " Cost: 0.19763485625962812\n",
      "iteration 230 \n",
      " Cost: 0.1983702717620301\n",
      "iteration 240 \n",
      " Cost: 0.1968089276348701\n",
      "iteration 250 \n",
      " Cost: 0.19703558510161595\n",
      "iteration 260 \n",
      " Cost: 0.19328728960437794\n",
      "iteration 270 \n",
      " Cost: 0.1904646842163195\n",
      "iteration 280 \n",
      " Cost: 0.19010488698717035\n",
      "iteration 290 \n",
      " Cost: 0.18333597211138836\n",
      "iteration 300 \n",
      " Cost: 0.18020638892973218\n",
      "iteration 310 \n",
      " Cost: 0.17896777654635615\n",
      "iteration 320 \n",
      " Cost: 0.17902191437540999\n",
      "iteration 330 \n",
      " Cost: 0.1790102085607546\n",
      "iteration 340 \n",
      " Cost: 0.17984266949969202\n",
      "iteration 350 \n",
      " Cost: 0.17929999710927097\n",
      "iteration 360 \n",
      " Cost: 0.17195121632183644\n",
      "iteration 370 \n",
      " Cost: 0.17458954190465675\n",
      "iteration 380 \n",
      " Cost: 0.1705431973036917\n",
      "iteration 390 \n",
      " Cost: 0.1702453814386779\n",
      "iteration 400 \n",
      " Cost: 0.16501775648426642\n",
      "iteration 410 \n",
      " Cost: 0.1645773905062744\n",
      "iteration 420 \n",
      " Cost: 0.16312628905587653\n",
      "iteration 430 \n",
      " Cost: 0.16463454507949146\n",
      "iteration 440 \n",
      " Cost: 0.16672333313430132\n",
      "iteration 450 \n",
      " Cost: 0.16443455507114843\n",
      "iteration 460 \n",
      " Cost: 0.16293136016629717\n",
      "iteration 470 \n",
      " Cost: 0.16052446751616184\n",
      "iteration 480 \n",
      " Cost: 0.16075832440373572\n",
      "iteration 490 \n",
      " Cost: 0.16012158861296572\n",
      "iteration 500 \n",
      " Cost: 0.1598169987944293\n"
     ]
    }
   ],
   "source": [
    "    SentencesList = createsentencelist(Dftrain, MyDict)\n",
    "    SentencesList = [x for x in random.sample(SentencesList, 40)]#For some dataset that may be too big,\n",
    "                                                                #we can select just a few sentences to try out\n",
    "                                                                #the model parameters.\n",
    "    par, ix = MyDict.getindexmodelparams()\n",
    "    myopt = optimizer.ClassicalOptimizer()\n",
    "    result = myopt.optimizedataset(SentencesList, par, MyDict,\n",
    "                                   tol=1e-5,\n",
    "                                   options={'maxiter':500, 'rhobeg': 1},\n",
    "                                   method=\"COBYLA\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "fe5b7099-c572-4d23-bec1-45e6cb7faada",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Cost function (Cross entropy)')"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(myopt.itercost)\n",
    "plt.xlabel('Iterations')\n",
    "plt.ylabel('Cost function (Cross entropy)')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f09bcd8c-088c-48ce-b4f8-7221915c5b05",
   "metadata": {},
   "source": [
    "-----------------\n",
    "## Re-running the sentences of the training dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "ddfef635-dcf6-4976-b94f-3199357aeb45",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "MyDict.updateparams(result.x)\n",
    "SentencesList = createsentencelist(Dftrain, MyDict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "ee2a448c-0b4e-41e1-b937-0d392495e3cb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sentence: dog eats bone\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: mouse eats fish\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: cat eats bone\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: bird flees cat\n",
      "label = True\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: dog flees mouse\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: mouse chases dog\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: mouse flees cat\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: mouse eats mouse\n",
      "label = False\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: mouse eats cat\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: mouse flees fish\n",
      "label = False\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: cat eats mouse\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: bird flees dog\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: dog chases fish\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: dog flees bird\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: bird eats cat\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: fish flees bird\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: bird eats mouse\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: bird chases fish\n",
      "label = True\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: bird eats fish\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: dog eats dog\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: mouse eats dog\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: cat flees dog\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: cat chases cat\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: cat eats cat\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: dog eats cat\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: bird chases mouse\n",
      "label = True\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: dog flees fish\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: dog chases dog\n",
      "label = False\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: mouse flees dog\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: dog chases bird\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: cat eats fish\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: cat chases bird\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: bird eats dog\n",
      "label = False\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: mouse chases bird\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: mouse chases fish\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: bird eats seed\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: bird chases bird\n",
      "label = False\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: dog eats mouse\n",
      "label = False\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: cat flees mouse\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: mouse eat seed\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: cat flees fish\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: bird flees mouse\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: cat chases dog\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: mouse chases mouse\n",
      "label = False\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: mouse chases cat\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: fish eats seed\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: mouse flees bird\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: bird chases dog\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n"
     ]
    }
   ],
   "source": [
    "Dftrain_result = Dftrain.copy()\n",
    "Result=[]\n",
    "for i,a_sentence in enumerate(SentencesList):\n",
    "    mycirc = circuit.CircuitBuilder()\n",
    "    mycirc.createcircuit(a_sentence, dataset=True)\n",
    "    mycirc.executecircuit()\n",
    "\n",
    "    probs = [0, 0]\n",
    "    # print('sentence: {}'.format(mysentence.sentence))\n",
    "    for sample in mycirc.result:\n",
    "        state = sample.state.bitstring\n",
    "        postselectedqubits = ''.join(state[x] for x in range(len(state)) if x != a_sentence.sentencequbit)\n",
    "        if postselectedqubits == '0' * (mycirc.qlmprogram.qbit_count - 1):\n",
    "            if state[a_sentence.sentencequbit] == '0':\n",
    "                probs[0] = sample.probability\n",
    "                # print(\n",
    "                #    \"State %s: probability %s, amplitude %s\" % (sample.state, sample.probability, sample.amplitude))\n",
    "            elif state[a_sentence.sentencequbit] == '1':\n",
    "                probs[1] = sample.probability\n",
    "                # print(\n",
    "                #    \"State %s: probability %s, amplitude %s\" % (\n",
    "                #   sample.state, sample.probability, sample.amplitude))\n",
    "    prob0 = probs[0] / sum(probs)\n",
    "    prob1 = probs[1] / sum(probs)\n",
    "    print('Sentence: {}'.format(a_sentence.sentence))\n",
    "    if a_sentence.label == 0:\n",
    "        print('label = False')\n",
    "    elif a_sentence.label == 1:\n",
    "        print('label = True')\n",
    "    if prob0 >= 0.5:\n",
    "        print('I found this sentence to be FALSE')\n",
    "        if a_sentence.label==0:\n",
    "            Result.append('Correct')\n",
    "        elif a_sentence.label==1:\n",
    "            Result.append('Error')\n",
    "    else:\n",
    "        print('I found this sentence to be TRUE')  \n",
    "        if a_sentence.label==0:\n",
    "            Result.append('Error')\n",
    "        elif a_sentence.label==1:\n",
    "            Result.append('Correct')\n",
    "    print('------')\n",
    "Dftrain_result['Result']=Result    \n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "256216c7-171f-402e-b077-27b4b469167e",
   "metadata": {},
   "source": [
    "---------\n",
    "## Testing the model in the training dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "4a41e1f0-81fa-40f3-b495-2ead207b1e32",
   "metadata": {},
   "outputs": [],
   "source": [
    "MyDict.updateparams(result.x)\n",
    "SentencesList = createsentencelist(Dftest, MyDict)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "29ec4040-8509-4056-aadd-99b75feb0ee9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Sentence: dog chases mouse\n",
      "label = True\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: bird eats bird\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: cat eats dog\n",
      "label = False\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: cat chases mouse\n",
      "label = True\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: dog eats meat\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: cat eats bird\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: cat chases fish\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: dog chases cat\n",
      "label = True\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: dog flees cat\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: cat eats meat\n",
      "label = True\n",
      "I found this sentence to be TRUE\n",
      "------\n",
      "Sentence: bird chases cat\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: bird flees fish\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n",
      "Sentence: mouse eats bird\n",
      "label = False\n",
      "I found this sentence to be FALSE\n",
      "------\n"
     ]
    }
   ],
   "source": [
    "Dftest_result = Dftest.copy()\n",
    "Result=[]\n",
    "for a_sentence in SentencesList:\n",
    "    mycirc = circuit.CircuitBuilder()\n",
    "    mycirc.createcircuit(a_sentence, dataset=True)\n",
    "    mycirc.executecircuit()\n",
    "\n",
    "    probs = [0, 0]\n",
    "    # print('sentence: {}'.format(mysentence.sentence))\n",
    "    for sample in mycirc.result:\n",
    "        state = sample.state.bitstring\n",
    "        postselectedqubits = ''.join(state[x] for x in range(len(state)) if x != a_sentence.sentencequbit)\n",
    "        if postselectedqubits == '0' * (mycirc.qlmprogram.qbit_count - 1):\n",
    "            if state[a_sentence.sentencequbit] == '0':\n",
    "                probs[0] = sample.probability\n",
    "                # print(\n",
    "                #    \"State %s: probability %s, amplitude %s\" % (sample.state, sample.probability, sample.amplitude))\n",
    "            elif state[a_sentence.sentencequbit] == '1':\n",
    "                probs[1] = sample.probability\n",
    "                # print(\n",
    "                #    \"State %s: probability %s, amplitude %s\" % (\n",
    "                #   sample.state, sample.probability, sample.amplitude))\n",
    "    prob0 = probs[0] / sum(probs)\n",
    "    prob1 = probs[1] / sum(probs)\n",
    "    print('Sentence: {}'.format(a_sentence.sentence))\n",
    "    if a_sentence.label == 0:\n",
    "        print('label = False')\n",
    "    elif a_sentence.label == 1:\n",
    "        print('label = True')\n",
    "    if prob0 >= 0.5:\n",
    "        print('I found this sentence to be FALSE')\n",
    "        if a_sentence.label==0:\n",
    "            Result.append('Correct')\n",
    "        elif a_sentence.label==1:\n",
    "            Result.append('Error')\n",
    "    else:\n",
    "        print('I found this sentence to be TRUE')  \n",
    "        if a_sentence.label==0:\n",
    "            Result.append('Error')\n",
    "        elif a_sentence.label==1:\n",
    "            Result.append('Correct')\n",
    "    print('------')\n",
    "Dftest_result['Result']=Result    \n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f3ebe15c-a680-4387-a93a-f5beb3b88e87",
   "metadata": {},
   "source": [
    "------------\n",
    "We now plot an histogram for the test and train dataset and see how many sentences instances have been correctly labelled and in how many the model has failed. \n",
    "We can see that we were successful with the train dataset but now with the test, which may be a sign of overfitting. \n",
    "To improve this, we need to further investigate:\n",
    "* Different Ansaetze\n",
    "* Different classical optimization methods\n",
    "* Different datasets"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "87a11ae6-1b1e-4a18-9b96-83fe146872c4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "train accuracy = 0.7916666666666666\n",
      "test accuracy = 0.6923076923076923\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fb429e7c3a0>"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "trainaccuracy = (Dftrain_result['Result']=='Correct').sum()/len(Dftrain_result['Result'])\n",
    "testaccuracy = (Dftest_result['Result']=='Correct').sum()/len(Dftest_result['Result'])\n",
    "\n",
    "print('train accuracy = {}'.format(trainaccuracy))\n",
    "print('test accuracy = {}'.format(testaccuracy))\n",
    "\n",
    "plt.hist([Dftest_result['Result'],Dftrain_result['Result']], label =['Test Dataset', 'Train Dataset'])\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "40904492-c49c-4ef5-b04f-52d4c0c500f3",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}