cdist(..., "sqeuclidean") vs. broadcasting

euclid.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import logging as log\n",
    "import numpy as np\n",
    "from typing import Optional\n",
    "from scipy.spatial.distance import cdist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "np.set_printoptions(edgeitems=10)\n",
    "np.core.arrayprint._line_width = 180"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "FORMAT = \"{asctime} {name} {levelname:8s} {message}\"\n",
    "log.basicConfig(format=FORMAT, style='{', level=log.DEBUG)\n",
    "logger = log.getLogger(\"notebook\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def euclidean_square(x: np.ndarray, xp: Optional[np.ndarray]=None) -> np.ndarray:\n",
    "    \n",
    "    assert x.ndim == 2, \"x must be 2D\"\n",
    "    logger.debug(\"x.shape = {}\".format(x.shape))\n",
    "    \n",
    "    xs = np.sum(np.square(x), axis=1)\n",
    "    \n",
    "    if xp is None:\n",
    "        return np.reshape(xs, (-1, 1)) + np.reshape(xs, (1, -1)) - 2 * np.matmul(x, x.T)     \n",
    "    \n",
    "    assert xp.ndim == 2, \"xp must be 2D\"\n",
    "    assert x.shape[1] == xp.shape[1], \"x and xp must have the same number of columns\"\n",
    "    logger.debug(\"xp.shape = {}\".format(xp.shape))\n",
    "    \n",
    "    xps = np.sum(np.square(xp), axis=1)\n",
    "    \n",
    "    return np.reshape(xs, (-1, 1)) + np.reshape(xps, (1, -1)) - 2 * np.matmul(x, xp.T)     \n",
    "\n",
    "\n",
    "def euclidean(x: np.ndarray, xp: Optional[np.ndarray]=None) -> np.ndarray:\n",
    "    return np.sqrt(euclidean_square(x, xp)) \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2018-03-28 15:01:30,413 notebook DEBUG    x.shape = (7, 10)\n",
      "2018-03-28 15:01:30,414 notebook DEBUG    xp.shape = (3, 10)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "array([[5.4832288 , 3.96775002, 4.59193113],\n",
       "       [5.44005569, 6.46171414, 3.67261701],\n",
       "       [5.6897614 , 5.77721062, 6.12629625],\n",
       "       [2.4223504 , 6.31955375, 7.90340939],\n",
       "       [2.83818613, 5.30016691, 4.98072687],\n",
       "       [2.30973936, 4.85547768, 8.65337153],\n",
       "       [3.08661838, 2.52295335, 4.26490056]])"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "array([[5.4832288 , 3.96775002, 4.59193113],\n",
       "       [5.44005569, 6.46171414, 3.67261701],\n",
       "       [5.6897614 , 5.77721062, 6.12629625],\n",
       "       [2.4223504 , 6.31955375, 7.90340939],\n",
       "       [2.83818613, 5.30016691, 4.98072687],\n",
       "       [2.30973936, 4.85547768, 8.65337153],\n",
       "       [3.08661838, 2.52295335, 4.26490056]])"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "array([[False, False,  True],\n",
       "       [ True, False, False],\n",
       "       [ True,  True,  True],\n",
       "       [False, False, False],\n",
       "       [False, False, False],\n",
       "       [False,  True, False],\n",
       "       [ True,  True, False]])"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cols, x_rows, y_rows = np.random.randint(low=3, high=12, size=3)\n",
    "\n",
    "x = np.random.uniform(low=-1, high=1, size=(x_rows, cols))\n",
    "y = np.random.uniform(low=-1, high=1, size=(y_rows, cols))\n",
    "    \n",
    "a = cdist(x, y, 'sqeuclidean')\n",
    "b = euclidean_square(x, y)\n",
    "\n",
    "display(a,b)\n",
    "\n",
    "display(np.equal(a, b))\n",
    "display(np.allclose(a, b))"
   ]
  },
  {
   "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}