customtopograd.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import numpy as np\n",
    "from topograd import topoclustergrad\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "final = np.loadtxt('finalxx.txt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0\n",
      "tensor(0.8885, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "1\n",
      "tensor(0.8378, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "2\n",
      "tensor(0.8033, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "3\n",
      "tensor(0.8110, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "4\n",
      "tensor(0.7298, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "5\n",
      "tensor(0.7299, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "6\n",
      "tensor(0.6918, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "7\n",
      "tensor(0.6344, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "8\n",
      "tensor(0.6038, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "9\n",
      "tensor(0.5929, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "10\n",
      "tensor(0.5829, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "11\n",
      "tensor(0.5550, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "12\n",
      "tensor(0.5311, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "13\n",
      "tensor(0.5131, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "14\n",
      "tensor(0.5187, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "15\n",
      "tensor(0.5042, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "16\n",
      "tensor(0.4850, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "17\n",
      "tensor(0.4531, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "18\n",
      "tensor(0.4356, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "19\n",
      "tensor(0.4283, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "20\n",
      "tensor(0.4165, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "21\n",
      "tensor(0.3956, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "22\n",
      "tensor(0.3937, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "23\n",
      "tensor(0.3780, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "24\n",
      "tensor(0.3684, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "25\n",
      "tensor(0.3550, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "26\n",
      "tensor(0.3670, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "27\n",
      "tensor(0.3613, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "28\n",
      "tensor(0.3736, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "29\n",
      "tensor(0.3549, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "30\n",
      "tensor(0.3358, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "31\n",
      "tensor(0.3467, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "32\n",
      "tensor(0.3269, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "33\n",
      "tensor(0.3288, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "34\n",
      "tensor(0.3624, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "35\n",
      "tensor(0.3340, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "36\n",
      "tensor(0.3101, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "37\n",
      "tensor(0.3171, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "38\n",
      "tensor(0.3079, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "39\n",
      "tensor(0.3174, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "40\n",
      "tensor(0.3100, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "41\n",
      "tensor(0.2833, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "42\n",
      "tensor(0.3145, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "43\n",
      "tensor(0.3096, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "44\n",
      "tensor(0.2808, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "45\n",
      "tensor(0.2683, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "46\n",
      "tensor(0.2725, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "47\n",
      "tensor(0.2820, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "48\n",
      "tensor(0.2620, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "49\n",
      "tensor(0.2808, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "50\n",
      "tensor(0.2606, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "51\n",
      "tensor(0.2540, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "52\n",
      "tensor(0.2567, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "53\n",
      "tensor(0.2745, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "54\n",
      "tensor(0.2731, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "55\n",
      "tensor(0.2780, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "56\n",
      "tensor(0.2640, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "57\n",
      "tensor(0.2586, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "58\n",
      "tensor(0.2556, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "59\n",
      "tensor(0.2712, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "60\n",
      "tensor(0.2528, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "61\n",
      "tensor(0.2488, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "62\n",
      "tensor(0.2665, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "63\n",
      "tensor(0.2656, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "64\n",
      "tensor(0.2592, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "65\n",
      "tensor(0.2538, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "66\n",
      "tensor(0.2525, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "67\n",
      "tensor(0.2455, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "68\n",
      "tensor(0.2381, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "69\n",
      "tensor(0.2362, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "70\n",
      "tensor(0.2332, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "71\n",
      "tensor(0.2224, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "72\n",
      "tensor(0.2197, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "73\n",
      "tensor(0.2144, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "74\n",
      "tensor(0.2202, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "75\n",
      "tensor(0.2204, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "76\n",
      "tensor(0.2237, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "77\n",
      "tensor(0.2147, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "78\n",
      "tensor(0.2098, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "79\n",
      "tensor(0.2091, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "80\n",
      "tensor(0.2042, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "81\n",
      "tensor(0.2057, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "82\n",
      "tensor(0.2072, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "83\n",
      "tensor(0.2029, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "84\n",
      "tensor(0.1967, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "85\n",
      "tensor(0.1909, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "86\n",
      "tensor(0.1750, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "87\n",
      "tensor(0.1692, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "88\n",
      "tensor(0.1638, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "89\n",
      "tensor(0.1822, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "90\n",
      "tensor(0.1627, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "91\n",
      "tensor(0.1595, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "92\n",
      "tensor(0.1518, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "93\n",
      "tensor(0.1490, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "94\n",
      "tensor(0.1459, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "95\n",
      "tensor(0.1515, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "96\n",
      "tensor(0.1462, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "97\n",
      "tensor(0.1450, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "98\n",
      "tensor(0.1396, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "99\n",
      "tensor(0.1360, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor(0.1330, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "101\n",
      "tensor(0.1311, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "102\n",
      "tensor(0.1465, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "103\n",
      "tensor(0.1258, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "104\n",
      "tensor(0.1270, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "105\n",
      "tensor(0.1414, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "106\n",
      "tensor(0.1255, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "107\n",
      "tensor(0.1210, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "108\n",
      "tensor(0.1361, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "109\n",
      "tensor(0.1274, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "110\n",
      "tensor(0.1286, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "111\n",
      "tensor(0.1267, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "112\n",
      "tensor(0.1235, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "113\n",
      "tensor(0.1300, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "114\n",
      "tensor(0.1288, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "115\n",
      "tensor(0.1365, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "116\n",
      "tensor(0.1144, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "117\n",
      "tensor(0.1563, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "118\n",
      "tensor(0.1345, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "119\n",
      "tensor(0.1344, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "120\n",
      "tensor(0.1491, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "121\n",
      "tensor(0.1305, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "122\n",
      "tensor(0.1258, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "123\n",
      "tensor(0.1280, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "124\n",
      "tensor(0.1219, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "125\n",
      "tensor(0.1084, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "126\n",
      "tensor(0.1208, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "127\n",
      "tensor(0.1297, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "128\n",
      "tensor(0.1105, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "129\n",
      "tensor(0.1215, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "130\n",
      "tensor(0.1057, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "131\n",
      "tensor(0.1035, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "132\n",
      "tensor(0.1066, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "133\n",
      "tensor(0.1048, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "134\n",
      "tensor(0.1219, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "135\n",
      "tensor(0.1017, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "136\n",
      "tensor(0.1010, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "137\n",
      "tensor(0.1157, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "138\n",
      "tensor(0.1013, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "139\n",
      "tensor(0.1046, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "140\n",
      "tensor(0.0978, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "141\n",
      "tensor(0.0978, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "142\n",
      "tensor(0.1001, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "143\n",
      "tensor(0.1110, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "144\n",
      "tensor(0.0992, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "145\n",
      "tensor(0.0976, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "146\n",
      "tensor(0.0945, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "147\n",
      "tensor(0.0908, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "148\n",
      "tensor(0.0898, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "149\n",
      "tensor(0.0907, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "150\n",
      "tensor(0.1000, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "151\n",
      "tensor(0.0937, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "152\n",
      "tensor(0.0870, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "153\n",
      "tensor(0.0854, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "154\n",
      "tensor(0.0859, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "155\n",
      "tensor(0.0817, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "156\n",
      "tensor(0.0966, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "157\n",
      "tensor(0.0820, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "158\n",
      "tensor(0.0801, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "159\n",
      "tensor(0.0794, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "160\n",
      "tensor(0.0807, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "161\n",
      "tensor(0.0788, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "162\n",
      "tensor(0.0743, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "163\n",
      "tensor(0.0760, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "164\n",
      "tensor(0.0862, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "165\n",
      "tensor(0.0748, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "166\n",
      "tensor(0.0847, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "167\n",
      "tensor(0.0729, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "168\n",
      "tensor(0.0713, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "169\n",
      "tensor(0.0714, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "170\n",
      "tensor(0.0812, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "171\n",
      "tensor(0.0669, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "172\n",
      "tensor(0.0682, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "173\n",
      "tensor(0.0646, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "174\n",
      "tensor(0.0753, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "175\n",
      "tensor(0.0635, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "176\n",
      "tensor(0.0738, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "177\n",
      "tensor(0.0589, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "178\n",
      "tensor(0.0616, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "179\n",
      "tensor(0.0577, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "180\n",
      "tensor(0.0597, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "181\n",
      "tensor(0.0556, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "182\n",
      "tensor(0.0578, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "183\n",
      "tensor(0.0560, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "184\n",
      "tensor(0.0581, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "185\n",
      "tensor(0.0548, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "186\n",
      "tensor(0.0572, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "187\n",
      "tensor(0.0536, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "188\n",
      "tensor(0.0587, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "189\n",
      "tensor(0.0522, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "190\n",
      "tensor(0.0616, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "191\n",
      "tensor(0.0564, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "192\n",
      "tensor(0.0532, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "193\n",
      "tensor(0.0506, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "194\n",
      "tensor(0.0505, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "195\n",
      "tensor(0.0531, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "196\n",
      "tensor(0.0476, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "197\n",
      "tensor(0.0498, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "198\n",
      "tensor(0.0490, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n",
      "199\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor(0.0514, grad_fn=<topoclustergradBackward>)\n",
      "ssssssssssssssssssssssssssssss\n"
     ]
    }
   ],
   "source": [
    "final = torch.tensor(final)\n",
    "# topo = topograd.apply\n",
    "for i in range(200):\n",
    "#     print('sssssssssssssssssssssssssssssss')\n",
    "    topo = topoclustergrad.apply\n",
    "    print(i)\n",
    "    final.requires_grad = True\n",
    "#     print(final.grad)\n",
    "    x = topo(final,10,10,0.1,2)\n",
    "    print(x)\n",
    "    x.backward()\n",
    "#     print(final.grad)\n",
    "#     final1 = copy.deepcopy(final)\n",
    "#     fg= copy.deepcopy(final.grad)\n",
    "#     print(final)\n",
    "#     fffs = copy.deepcopy(final.grad)\n",
    "#     print(final)\n",
    "    final = final.detach() - 0.05 * final.grad\n",
    "#     print('this is final')\n",
    "#     print(final)\n",
    "\n",
    "    print('ssssssssssssssssssssssssssssss')\n",
    "#     sa,sd,ss = major1(final.numpy(),10,10,0.1,1,0.1,1)\n",
    "#     final.detach_()\n",
    "#     print(final.grad)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7f0c991fc590>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize = (10,10))\n",
    "final = final.numpy()\n",
    "plt.scatter(final[:,0],final[:,1],s = 4 )"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "clustering",
   "language": "python",
   "name": "clustering"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}