vectorized_h3.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Experiments with vectorized functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/ajfriend/work/h3-py/uber/h3-py/env/lib/python3.8/site-packages/h3/unstable/__init__.py:4: UserWarning: Modules under `h3.unstable` are experimental, and may change at any time.\n",
      "  warnings.warn(\n"
     ]
    }
   ],
   "source": [
    "from h3.unstable import vect\n",
    "\n",
    "import h3.api.numpy_int as h3i\n",
    "import h3.api.basic_str as h3s\n",
    "\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# `geo_to_h3` times"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "N = 1_000_000\n",
    "\n",
    "lats = np.random.uniform(0, 90, N) \n",
    "lons = np.random.uniform(0, 90, N)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.1 s ± 57.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit\n",
    "[h3s.geo_to_h3(lat, lon, 10) for lat, lon in zip(lats, lons)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.63 s ± 7.51 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit\n",
    "[h3i.geo_to_h3(lat, lon, 10) for lat, lon in zip(lats, lons)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "802 ms ± 13.5 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit\n",
    "vect.geo_to_h3(lats, lons, 9)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Haversine distances between cells\n",
    "\n",
    "compute haversine distances between h3 cells, using\n",
    "integer representation (avoiding conversion to/from `str` representation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def rand_hexes(N):\n",
    "    lats = np.random.uniform(0, 90, N) \n",
    "    lons = np.random.uniform(0, 90, N)\n",
    "\n",
    "    out = np.array(\n",
    "        [h3i.geo_to_h3(a, o, 9) for a,o in zip(lats, lons)],\n",
    "        dtype = 'uint64',\n",
    "    )\n",
    "    \n",
    "    return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "N = 1_000_000\n",
    "\n",
    "a = rand_hexes(N)\n",
    "b = rand_hexes(N)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1 s ± 22.9 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit\n",
    "vect.cell_haversine(a, b)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Multithreading\n",
    "\n",
    "Since these functions release the Python GIL, we can acheive some speedups via parallelism on multicore machines.\n",
    "\n",
    "Timings below are run on a quad-core MacBook."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## `geo_to_h3`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "from concurrent.futures import ThreadPoolExecutor\n",
    "\n",
    "N = 100_000\n",
    "K = 10\n",
    "\n",
    "geos = [\n",
    "    # lat/lng vectors\n",
    "    (np.random.uniform(0, 90, N), np.random.uniform(0, 90, N))\n",
    "    for _ in range(K)\n",
    "]\n",
    "\n",
    "def foo(geo):\n",
    "    lats, lngs = geo\n",
    "    out = vect.geo_to_h3(lats, lngs, 9)\n",
    "    return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "821 ms ± 30.9 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit\n",
    "out = list(map(foo, geos))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "248 ms ± 10.8 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit\n",
    "with ThreadPoolExecutor(max_workers=4) as ex:\n",
    "    out = list(ex.map(foo, geos))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## `cell_haversine`\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "N = 100_000\n",
    "K = 10\n",
    "\n",
    "pairs = [\n",
    "    (rand_hexes(N), rand_hexes(N))\n",
    "    for _ in range(K)\n",
    "]\n",
    "\n",
    "def foo(ab):\n",
    "    a, b = ab\n",
    "    out = vect.cell_haversine(a, b)\n",
    "    return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.03 s ± 19.4 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit\n",
    "out = list(map(foo, pairs))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "339 ms ± 21.1 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n"
     ]
    }
   ],
   "source": [
    "%%timeit\n",
    "with ThreadPoolExecutor(max_workers=4) as ex:\n",
    "    out = list(ex.map(foo, pairs))"
   ]
  },
  {
   "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": "3.8.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}