clustering 1.1.ipynb

{
 "metadata": {
  "name": "",
  "signature": "sha256:d043d2b646c25c84cf2aad2ac9a7188ad6afcf429ab8f36b3e4a4a59653aeabc"
 },
 "nbformat": 3,
 "nbformat_minor": 0,
 "worksheets": [
  {
   "cells": [
    {
     "cell_type": "raw",
     "metadata": {},
     "source": [
      "choosing gaps :\n",
      "1) total range = (least value greater than cut,greatest value)\n",
      "2) start with the longest gap\n",
      "3) at each step use the longest gap with the center at opposite sign of sum[centers(position from the center of total interval)]\n",
      "4) stop when average of \"var\" starts(?) to increase\n",
      "        var for a bunch = (#gaps)*sum(gap_size**2)/bunch_size**2 (if bunch_size != 0)\n",
      "        ? = when average(var) increases two times in a row\n",
      "        don't stop if number of hits in a bunch is greater than number of layers\n",
      "        I have assumed that each track has at least 2 hits ( to be a track ! )\n",
      "input -> an array of NN output values (sorted ascendingly from cut) + # of layers\n",
      "we also need a matrix for all gaps (may be large though, in the real case)\n",
      "        each row of which is : [length(end-start),center((start+end)/2.0),starting line index]\n",
      "clustering_coefficient (positive) -> greater values for more clusters and 0 for no clustering (one cluster)\n",
      "output -> a matrix -> each row : [starting index of a bunch,ending index of a bunch]"
     ]
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "import numpy as np\n",
      "import matplotlib.pyplot as plt"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 2
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "in_array = []\n",
      "for i in range (7):\n",
      "    for j in range (int(uniform(2,5))):\n",
      "        in_array.append(numpy.random.normal(10*(i+1),2))\n",
      "in_array = sorted(in_array)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [],
     "prompt_number": 3
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "clustering_coefficient = 1\n",
      "n_layers = 5\n",
      "n_lines = len(in_array)\n",
      "n_gaps = n_lines - 1\n",
      "interval_center = (in_array[n_lines - 1] + in_array[0])/2.0\n",
      "\n",
      "gap_matrix = np.zeros((n_gaps,3),dtype=float32)\n",
      "for i in range (n_gaps) :\n",
      "    gap_matrix[i][2] = i\n",
      "    gap_matrix[i][1] = (in_array[i+1] + in_array[i])/2.0 - interval_center\n",
      "    if gap_matrix[i][1] < 0 :\n",
      "        gap_matrix[i][0] = in_array[i+1] - in_array[i]\n",
      "    else :\n",
      "        gap_matrix[i][0] = in_array[i] - in_array[i+1]\n",
      "# sorting gap_matrix ascendingly by gap lengths\n",
      "gap_matrix.view('f32,f32,f32').sort(order=['f0'], axis=0)\n",
      "\n",
      "out_matrix = np.zeros((int(n_lines),2),dtype=int32)\n",
      "\n",
      "# var receives indices of first and last line in the bunch and returns val as defined above\n",
      "def var (start1 = 0,end1 = n_gaps) :\n",
      "    start = int(start1)\n",
      "    end = int(end1)\n",
      "    sum_square = 0.0\n",
      "    for i in range (start,end) :\n",
      "        sum_square += (in_array[i+1] - in_array[i])**2\n",
      "    return (end-start) * sum_square / (in_array[end] - in_array[start])\n",
      "\n",
      "sgn = 0\n",
      "# some unreadable code : !!!        (maybe I add some comments later)\n",
      "\n",
      "# finds the interval(start and ending line indices) that we are going to check whether can be separated or not\n",
      "def find_interval (accepted_gaps) :\n",
      "    interval = np.zeros(2,dtype=int8)\n",
      "    interval[1] = n_gaps\n",
      "    Gl_sgn = (-1)**(accepted_gaps%2)\n",
      "    Gl_index = int(Gl_sgn*(accepted_gaps/2) + (accepted_gaps%2)*(n_gaps-1))\n",
      "    for i in range (accepted_gaps) :\n",
      "        sgn = (-1)**(i%2)\n",
      "        index = int(sgn*i + (i%2)*(n_gaps-1))\n",
      "        if gap_matrix[index][1] > gap_matrix[Gl_index][1] and in_array[int(gap_matrix[index][2])] < in_array[interval[1]] :\n",
      "            interval[1] = int(gap_matrix[index][2])\n",
      "        if gap_matrix[index][1] < gap_matrix[Gl_index][1] and in_array[int(gap_matrix[index][2])+1] > in_array[interval[0]] :\n",
      "            interval[0] = int(gap_matrix[index][2] + 1)\n",
      "    return interval\n",
      "\n",
      "def adding_gaps () :\n",
      "    refused_gaps = 0\n",
      "    accepted_gaps = 0\n",
      "    while refused_gaps < 2 and accepted_gaps < n_gaps:\n",
      "        interval = np.zeros(2,dtype=int8)\n",
      "        interval = find_interval(accepted_gaps)\n",
      "        Gl_sgn = (-1)**(accepted_gaps%2)\n",
      "        Gl_index = int(Gl_sgn*(accepted_gaps/2) + (accepted_gaps%2)*(n_gaps-1))\n",
      "        un_spr = var(interval[0],interval[1]) # var for unseperated interval\n",
      "        piece1 = piece2 = 2*un_spr + 1\n",
      "        if in_array[int(interval[0])] != in_array[int(gap_matrix[Gl_index][2])] :\n",
      "            piece1 = var(interval[0],gap_matrix[Gl_index][2])\n",
      "        if in_array[int(gap_matrix[Gl_index][2])+1] != in_array[int(interval[1])] :\n",
      "            piece2 = var(gap_matrix[Gl_index][2]+1,interval[1])\n",
      "        if (piece1+piece2) >= clustering_coefficient * 2 * un_spr and (interval[1] - interval[0]) < n_layers :\n",
      "            refused_gaps += 1\n",
      "            gap_matrix[Gl_index][1] = in_array[n_gaps]\n",
      "            continue\n",
      "        else :\n",
      "            accepted_gaps += 1\n",
      "    return accepted_gaps\n",
      "\n",
      "accepted_gaps = adding_gaps()\n",
      "\n",
      "for i in range (accepted_gaps-1,n_gaps) :\n",
      "    sgn = (-1)**(i%2)\n",
      "    index = sgn*i/2 + (i%2)*(n_gaps-1)\n",
      "    gap_matrix[index][1] = in_array[n_gaps]\n",
      "gap_matrix.view('f32,f32,f32').sort(order=['f1'], axis=0)\n",
      "out_matrix [0] = [0,gap_matrix[0][2]]\n",
      "out_matrix [accepted_gaps] = [gap_matrix[accepted_gaps-1][2]+1,n_gaps]\n",
      "for bunch in range (1,accepted_gaps) :\n",
      "    out_matrix [bunch] = [int(gap_matrix[bunch-1][2]+1),int(gap_matrix[bunch][2])]\n",
      "print \"number of bunches \\t\" , accepted_gaps + 1\n",
      "for i in range (len(out_matrix)) :\n",
      "    for j in range (2) :\n",
      "        if i == 0 or out_matrix[i][j] != 0 :\n",
      "            print in_array[out_matrix[i][j]]\n",
      "    print \"\\t\""
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "output_type": "stream",
       "stream": "stdout",
       "text": [
        "number of bunchs \t7\n",
        "7.27920204551\n",
        "10.11141006\n",
        "\t\n",
        "16.8399165506\n",
        "17.8595134472\n",
        "\t\n",
        "28.0598476276\n",
        "31.6043401452\n",
        "\t\n",
        "36.8603934378\n",
        "38.8920199596\n",
        "\t\n",
        "52.1786551527\n",
        "52.4462989112\n",
        "\t\n",
        "59.965424374\n",
        "62.5444942526\n",
        "\t\n",
        "66.8054760679\n",
        "71.6909060584\n",
        "\t\n",
        "\t\n",
        "\t\n",
        "\t\n",
        "\t\n",
        "\t\n",
        "\t\n",
        "\t\n",
        "\t\n",
        "\t\n",
        "\t\n"
       ]
      }
     ],
     "prompt_number": 4
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "plt.hist([val for val in in_array],bins = 100)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": [
      {
       "metadata": {},
       "output_type": "pyout",
       "prompt_number": 4,
       "text": [
        "(array([ 1.,  1.,  1.,  0.,  1.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
        "         0.,  0.,  0.,  1.,  0.,  0.,  1.,  0.,  0.,  0.,  0.,  2.,  0.,\n",
        "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  1.,  0.,  0.,  1.,  0.,\n",
        "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  1.,  0.,  0.,  0.,  1.,\n",
        "         1.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
        "         0.,  0.,  1.,  1.,  0.,  1.,  0.,  0.,  0.,  0.,  0.,  0.,  1.,\n",
        "         0.,  0.,  0.,  1.,  0.,  0.,  1.,  0.,  0.,  0.,  0.,  0.,  0.,\n",
        "         0.,  0.,  0.,  1.,  0.,  0.,  0.,  1.,  1.]),\n",
        " array([  7.18615959,   7.8197743 ,   8.45338901,   9.08700372,\n",
        "          9.72061842,  10.35423313,  10.98784784,  11.62146255,\n",
        "         12.25507726,  12.88869196,  13.52230667,  14.15592138,\n",
        "         14.78953609,  15.4231508 ,  16.0567655 ,  16.69038021,\n",
        "         17.32399492,  17.95760963,  18.59122433,  19.22483904,\n",
        "         19.85845375,  20.49206846,  21.12568317,  21.75929787,\n",
        "         22.39291258,  23.02652729,  23.660142  ,  24.29375671,\n",
        "         24.92737141,  25.56098612,  26.19460083,  26.82821554,\n",
        "         27.46183025,  28.09544495,  28.72905966,  29.36267437,\n",
        "         29.99628908,  30.62990379,  31.26351849,  31.8971332 ,\n",
        "         32.53074791,  33.16436262,  33.79797733,  34.43159203,\n",
        "         35.06520674,  35.69882145,  36.33243616,  36.96605087,\n",
        "         37.59966557,  38.23328028,  38.86689499,  39.5005097 ,\n",
        "         40.13412441,  40.76773911,  41.40135382,  42.03496853,\n",
        "         42.66858324,  43.30219795,  43.93581265,  44.56942736,\n",
        "         45.20304207,  45.83665678,  46.47027149,  47.10388619,\n",
        "         47.7375009 ,  48.37111561,  49.00473032,  49.63834503,\n",
        "         50.27195973,  50.90557444,  51.53918915,  52.17280386,\n",
        "         52.80641857,  53.44003327,  54.07364798,  54.70726269,\n",
        "         55.3408774 ,  55.97449211,  56.60810681,  57.24172152,\n",
        "         57.87533623,  58.50895094,  59.14256565,  59.77618035,\n",
        "         60.40979506,  61.04340977,  61.67702448,  62.31063919,\n",
        "         62.94425389,  63.5778686 ,  64.21148331,  64.84509802,\n",
        "         65.47871273,  66.11232743,  66.74594214,  67.37955685,\n",
        "         68.01317156,  68.64678627,  69.28040097,  69.91401568,  70.54763039]),\n",
        " <a list of 100 Patch objects>)"
       ]
      },
      {
       "metadata": {},
       "output_type": "display_data",
       "png": "iVBORw0KGgoAAAANSUhEUgAAAXUAAAEACAYAAABMEua6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYqlLUiOWuiQ1YqlLUiP/B/CGlBax8ev+AAAAAElFTkSuQmCC\n",
       "text": [
        "<matplotlib.figure.Figure at 0x7f153dae6550>"
       ]
      }
     ],
     "prompt_number": 4
    },
    {
     "cell_type": "code",
     "collapsed": false,
     "input": [
      "def separator (length_array,x1,x2) :\n",
      "    for i in range (len(length_array)):\n",
      "        if abs(length_array[i] - x1) > abs(x2 - length_array[i]) :\n",
      "            return i\n",
      "mean1 = length_array[0]\n",
      "mean2 = length_array[len(length_array)-1]\n",
      "everything_is_ok = True\n",
      "while everything_is_ok :\n",
      "    sum1 = 0\n",
      "    sum2 = 0\n",
      "    for i in range (separator(length_array,mean1,mean2)) :\n",
      "        sum1 += length_array[i]\n",
      "    x1 = sum1 / float(separator(length_array,mean1,mean2))\n",
      "    for i in range (separator(length_array,mean1,mean2),len(length_array)) :\n",
      "        sum2 += length_array[i]\n",
      "    x2 = sum2 / float(len(length_array)-separator(length_array,mean1,mean2))\n",
      "    if mean1 != x1 or mean2 != x2 :\n",
      "        mean1 = x1\n",
      "        mean2 = x2\n",
      "    else :\n",
      "        everything_is_ok = False\n",
      "print len(length_array)-separator(length_array,mean1,mean2)"
     ],
     "language": "python",
     "metadata": {},
     "outputs": []
    }
   ],
   "metadata": {}
  }
 ]
}