{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "import numpy as np\n",
    "import tensorly as tl\n",
    "tl.set_backend('pytorch')\n",
    "\n",
    "from tensorly.decomposition import parafac\n",
    "from tensorly.cp_tensor import cp_to_tensor, CPTensor"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Generate tensor\n",
    "rank = 70\n",
    "t_shape = (10,  20, 30)\n",
    "\n",
    "factors = [torch.randn(l, rank) for l in t_shape]\n",
    "weights = tl.ones(rank)\n",
    "\n",
    "t = cp_to_tensor(CPTensor((weights, factors)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "approx error: 114.89116668701172, non-zero elements in approx: 4200\n"
     ]
    }
   ],
   "source": [
    "# Parafac\n",
    "r = rank\n",
    "\n",
    "cp = parafac(t, r, cvg_criterion = 'rec_error')\n",
    "approx = cp_to_tensor(cp)\n",
    "\n",
    "approx_error = tl.norm(t - approx)\n",
    "nonzero_elems_count = sum(t_shape)*r\n",
    "\n",
    "print('approx error: {}, non-zero elements in approx: {}'.format(approx_error,\\\n",
    "                                                                 nonzero_elems_count))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "approx error: 112.28209686279297, non-zero elements in approx: 4200.0\n"
     ]
    }
   ],
   "source": [
    "# Sparse parafac\n",
    "r = rank-5\n",
    "sparsity = .05\n",
    "\n",
    "cp, sparse = parafac(t, r, sparsity=sparsity, cvg_criterion = 'rec_error')\n",
    "approx_sp = cp_to_tensor(cp) + sparse\n",
    "\n",
    "approx_error = tl.norm(t - approx_sp)\n",
    "nonzero_elems_count = sum(t_shape)*r + sparsity*np.prod(t_shape)\n",
    "\n",
    "print('approx error: {}, non-zero elements in approx: {}'.format(approx_error,\\\n",
    "                                                                 nonzero_elems_count))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}