{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "f785959f-8601-4d0c-9542-0478b3776622",
   "metadata": {},
   "source": [
    "Via sockets, programs communicate with each other.  Use of the multiprocessing module is illustrated."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "665a86d7-a070-41cc-8e0d-27a90fec3953",
   "metadata": {},
   "source": [
    "All code in this notebook is for viewing only as it should be executed in a terminal window."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c9758698-4e36-4238-b98e-29d25db261c4",
   "metadata": {},
   "source": [
    "# 1. Client/Server Computing to Estimate $\\pi$"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dabedb7d-9a0d-422b-a8c3-e63b10b7307b",
   "metadata": {},
   "source": [
    "The code below implements a server for Monte Carlo method to estimate pi,\n",
    "with two clients.  The server dispatches the seeds and collects results.\n",
    "The code should be executed in a terminal window and is included below for viewing only."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "127def0c-f7f7-41be-b8f3-975a6f175608",
   "metadata": {},
   "source": [
    "```\n",
    "from socket import socket as Socket\n",
    "from socket import AF_INET, SOCK_STREAM\n",
    "\n",
    "HOSTNAME = ''   # blank so any address can be used\n",
    "NUMBER = 11267  # number for the port\n",
    "BUFFER = 80     # size of the buffer\n",
    "\n",
    "SERVER_ADDRESS = (HOSTNAME, NUMBER)\n",
    "SERVER = Socket(AF_INET, SOCK_STREAM)\n",
    "SERVER.bind(SERVER_ADDRESS)\n",
    "SERVER.listen(2)\n",
    "\n",
    "print('server waits for connections...')\n",
    "\n",
    "FIRST, FIRST_ADDRESS = SERVER.accept()\n",
    "SECOND, SECOND_ADDRESS = SERVER.accept()\n",
    "\n",
    "FIRST.send('1'.encode())\n",
    "SECOND.send('2'.encode())\n",
    "\n",
    "print('server waits for results...')\n",
    "\n",
    "PART1 = FIRST.recv(BUFFER).decode()\n",
    "PART2 = SECOND.recv(BUFFER).decode()\n",
    "\n",
    "RESULT = 2*(float(PART1)+float(PART2))\n",
    "print('approximation for pi =', RESULT)\n",
    "\n",
    "SERVER.close()\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1633aca6-199c-4ff9-b795-45d7c6e53530",
   "metadata": {},
   "source": [
    "The clients connecting to the server are *compute clients* or *workers*.  The code for the client script is included below."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "209adbb2-9269-4251-987b-5a5503a7de7c",
   "metadata": {},
   "source": [
    "```\n",
    "from random import seed, uniform\n",
    "from socket import socket as Socket\n",
    "from socket import AF_INET, SOCK_STREAM\n",
    "\n",
    "HOSTNAME = 'localhost'  # on same host\n",
    "NUMBER = 11267          # same port number\n",
    "BUFFER = 80             # size of the buffer\n",
    "\n",
    "SERVER_ADDRESS = (HOSTNAME, NUMBER)\n",
    "CLIENT = Socket(AF_INET, SOCK_STREAM)\n",
    "CLIENT.connect(SERVER_ADDRESS)\n",
    "\n",
    "print('client is connected')\n",
    "DATA = CLIENT.recv(BUFFER).decode()\n",
    "print('client received %s' % DATA)\n",
    "\n",
    "seed(int(DATA))\n",
    "\n",
    "N = 10**7\n",
    "CNT = 0\n",
    "for i in range(0, N):\n",
    "    xpt = uniform(0, 1)\n",
    "    ypt = uniform(0, 1)\n",
    "    if xpt**2 + ypt**2 <= 1:\n",
    "        CNT = CNT + 1\n",
    "R = float(CNT)/N\n",
    "print('client computes %.12f' % R)\n",
    "\n",
    "CLIENT.send(str(R).encode())\n",
    "\n",
    "CLIENT.close()\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6d22d6e4-76d6-4df6-be15-32b4b06e3522",
   "metadata": {},
   "source": [
    "The code in this section is to illustrate the sockets, but for practical purposes, this way of working is rather tedious.  Therefore, the `multiprocessing` module is introduced next."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c8a7ce83-4811-4fc4-b4e7-dbdc0678718d",
   "metadata": {},
   "source": [
    "# 2. Multiprocessing"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "39a7c481-d783-4f5a-b26a-c1db313b0f33",
   "metadata": {},
   "source": [
    "The code below should be executed in a terminal window.  It is included for viewing only."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "20523bbd-7bc5-4dad-9814-e0534f219ab9",
   "metadata": {},
   "source": [
    "```\n",
    "from multiprocessing import Process\n",
    "from time import sleep\n",
    "\n",
    "def say_hello(name, tslp):\n",
    "    \"\"\"\n",
    "    Process with name says hello,\n",
    "    prints its process id, and\n",
    "    sleeps for tslp seconds.\n",
    "    \"\"\"\n",
    "    from os import getpid\n",
    "    prt = 'hello from ' + name \\\n",
    "        + ' with process id ' + str(getpid())\n",
    "    print(prt)\n",
    "    print(name, 'sleeps', tslp, 'seconds')\n",
    "    sleep(tslp)\n",
    "    print(name, 'wakes up')\n",
    "\n",
    "def main():\n",
    "    \"\"\"\n",
    "    Defines two processes.\n",
    "    \"\"\"\n",
    "    from os import getpid\n",
    "    print('the process id of main is', getpid())\n",
    "    apr = Process(target=say_hello, args = ('A', 4))\n",
    "    bpr = Process(target=say_hello, args = ('B', 1))\n",
    "    print('starting two processes...')\n",
    "    apr.start()\n",
    "    sleep(1) # to make printing look nice\n",
    "    bpr.start()\n",
    "    print('waiting for processes to wake up...')\n",
    "    apr.join()\n",
    "    bpr.join()\n",
    "    print('processes are done')\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "8d6dbaff-450a-434c-9fbd-b654113d982f",
   "metadata": {},
   "source": [
    "# 3. Multiprocessing to Estimate $\\pi$"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a0f6a9b9-8008-4082-8655-e20960fa93d2",
   "metadata": {},
   "source": [
    "The main advantage of using the `multiprocessing` module over the explicit client/server communication is that one single program defines all computations."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "c1d0b218-7fa7-46ba-971a-786665ffee3e",
   "metadata": {},
   "source": [
    "```\n",
    "from multiprocessing import Process, Queue\n",
    "from multiprocessing import freeze_support\n",
    "from math import pi\n",
    "from random import uniform as u\n",
    "from random import seed\n",
    "\n",
    "def monte_carlo4pi(nbr, nsd, result):\n",
    "    \"\"\"\n",
    "    Estimates pi with nbr samples,\n",
    "    using nsd as seed.\n",
    "    Adds the result to the queue q.\n",
    "    \"\"\"\n",
    "    seed(nsd)\n",
    "    cnt = 0\n",
    "    for _ in range(nbr):\n",
    "        (x, y) = (u(-1, 1), u(-1, 1))\n",
    "        if x**2 + y**2 <= 1:\n",
    "            cnt = cnt + 1\n",
    "    result.put(cnt)\n",
    "\n",
    "def main():\n",
    "    \"\"\"\n",
    "    Prompts the user for the number of samples\n",
    "    and the number of processes.\n",
    "    \"\"\"\n",
    "    nbr = int(input('Give the number of samples : '))\n",
    "    npr = int(input('Give the number of processes : '))\n",
    "    queues = [Queue() for _ in range(npr)]\n",
    "    procs = []\n",
    "    for k in range(1, npr+1):\n",
    "        procs.append(Process(target=monte_carlo4pi, \\\n",
    "            args=(nbr, k, queues[k-1])))\n",
    "    for process in procs:\n",
    "        process.start()\n",
    "    for process in procs:\n",
    "        process.join()\n",
    "    app = 4*sum([q.get()/nbr for q in queues])/npr\n",
    "    print(app, f\"error : {abs(app - pi):.3e}\")\n",
    "\n",
    "if __name__ == '__main__':\n",
    "    freeze_support()\n",
    "    main()\n",
    "```"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a0025037-f9a8-4c2b-adc6-0cb7e3be8ce7",
   "metadata": {},
   "source": [
    "The `freeze_support()` is critical to make the code run on Windows computers."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}