{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Transformation performance test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def transform_apply(tensor, transform):\n",
    "    feat_tensor = []\n",
    "    org_shape = tensor.shape\n",
    "    for plane in tensor[0]:\n",
    "        feat_tensor.append(transform(plane))\n",
    "    ret = np.concatenate(feat_tensor)\n",
    "    ret = ret.reshape(org_shape)\n",
    "    return ret"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "BD_transforms = [\n",
    "    lambda f:f, \n",
    "    lambda f:np.rot90(f,1), \n",
    "    lambda f:np.rot90(f,2), \n",
    "    lambda f:np.rot90(f,3), \n",
    "    lambda f:np.fliplr(f), \n",
    "    lambda f:np.flipud(f), \n",
    "    lambda f:np.transpose(f), \n",
    "    lambda f:np.fliplr(np.rot90(f, 1))\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "BD_fasttrans = [\n",
    "    lambda f: f,\n",
    "    lambda f: np.rot90(f, 1, axes=(2,3)),\n",
    "    lambda f: np.rot90(f, 2, axes=(2,3)),\n",
    "    lambda f: np.rot90(f, 3, axes=(2,3)),\n",
    "    lambda f: f[:, :, :, ::-1],\n",
    "    lambda f: f[:, :, ::-1, :],\n",
    "    lambda f: np.transpose(f, axes=(0, 1, 3, 2)),\n",
    "    lambda f: np.rot90(f, 1, axes=(2,3))[:, :, :, ::-1]\n",
    "]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True\n",
      "org   2.79942035675e-05\n",
      "fast  5.63967943192e-08\n",
      "times 496.379340448\n",
      "\n",
      "True\n",
      "org   0.00023686504364\n",
      "fast  4.14942979813e-06\n",
      "times 57.0837573266\n",
      "\n",
      "True\n",
      "org   0.000192616319656\n",
      "fast  3.43266010284e-06\n",
      "times 56.1128436505\n",
      "\n",
      "True\n",
      "org   0.000235383987427\n",
      "fast  4.21041965485e-06\n",
      "times 55.9051132007\n",
      "\n",
      "True\n",
      "org   6.21682882309e-05\n",
      "fast  3.42023849487e-07\n",
      "times 181.765945048\n",
      "\n",
      "True\n",
      "org   6.14371061325e-05\n",
      "fast  3.40197086334e-07\n",
      "times 180.592687593\n",
      "\n",
      "True\n",
      "org   5.45855045319e-05\n",
      "fast  5.8767914772e-07\n",
      "times 92.8831739966\n",
      "\n",
      "True\n",
      "org   0.000263137102127\n",
      "fast  4.56223011017e-06\n",
      "times 57.6772972368\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for i in range(0, 8):\n",
    "    t = np.random.randint(low=0, high=2, size=[1,49,19,19])\n",
    "    transform = BD_transforms[i]\n",
    "    fasttrans = BD_fasttrans[i]\n",
    "    \n",
    "    # sanity check\n",
    "    r1 = transform_apply(t, transform)\n",
    "    r2 = fasttrans(t)\n",
    "    sanity = (r1 == r2).all()\n",
    "    print sanity\n",
    "    assert sanity, \"The transformation is not same\"\n",
    "    \n",
    "    # time check\n",
    "    t1 = %timeit -oq transform_apply(t, transform)\n",
    "    t2 = %timeit -oq fasttrans(t)\n",
    "    \n",
    "    print \"org  \", t1.best\n",
    "    print \"fast \", t2.best\n",
    "    print \"times\", t1.best / t2.best    \n",
    "    print"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}