python map reduce 并行_python的简单MapReduce实现：计算π

#coding=utf-8

import random

import multiprocessing

from multiprocessing import Process

class MapReduce(object):

def __init__(self, map_func, reduce_func, workers_num=None):

self.map_func = map_func

self.reduce_func = reduce_func

self.workers_num = workers_num

if not workers_num:

workers_num = multiprocessing.cpu_count()*2

self.pool = multiprocessing.Pool(workers_num)

def __call__(self, inputs):

map_result = self.pool.map(self.map_func, inputs)

reduce_result = self.reduce_func(map_result)

return reduce_result

def calculator(*args):

print multiprocessing.current_process().name,' processing'

points, circle_round = args[0]

points_in_circle = 0

for i in range(points):

# 这里其实只取了1/4圆

x = random.random()*circle_round

y = random.random()*circle_round

if (x**2 + y**2) < circle_round**2:

points_in_circle += 1

return points_in_circle

def count_circle_points(points_list):

return sum(points_list)

if __name__ == '__main__':

# 半径

CIRCLE_ROUND = 10

# 总点数

POINTS = 10000000

# 总进程数

WORKERS_NUM = 10

map_reduce = MapReduce(calculator, count_circle_points, WORKERS_NUM)

inputs = [(POINTS/WORKERS_NUM, CIRCLE_ROUND)] * WORKERS_NUM

all_points_in_circle = map_reduce(inputs)

ac_as = float(all_points_in_circle)/POINTS

print 'pi approach to:%7f'%(4*ac_as)