多处理“池”挂起正在运行的程序

Question

I have written a program to use the "Pool" from multiprocessing python package,but it keeps hanging the kernel.(I am using Jupyter with Anaconda,by the way)

I tried using a function from an imported package like "math",which works perfectly,but, the moment i use a function i created in the same python file it doesn't work.

import multiprocessing as mp

def f(x):
   for i in x:
      return i+1

ar=[1,2,3,4,5]
def main():
    pool=mp.Pool(processes=4)
    pool.map(f,ar)
if __name__== '__main__':
    main()

It shows that the kernel is working,but the cpu usage sits at 0%.

Answer 1

The problem is ar is not within the local scope of your main function, it's above it. Put it inside and it should work.

from multiprocessing import Pool

def f(x):
    return x+1

if __name__ == '__main__':
    ar=[1,2,3,4,5]
    with Pool(4) as p:
        print(p.map(f, ar))

Also consider map is taking your iterable and assigning it across the processes, in essence it is doing the for loop for you. Therefore you can access the x directly.

And then last, using with as a context manager ensures your Pool closes when it is complete. [docs]

Edit: I edited and tested the above code and mine sends the correct values and closes properly.

This code also works for me in both Spyder 3.2.6 and standard terminal. Both snippets close out properly and don't hang on my machine.

import multiprocessing as mp

def f(x):
    return x+1

def main():
    ar = [1,2,3,4,5]
    pool = mp.Pool(processes=4)
    print(pool.map(f, ar))

if __name__ == '__main__':
    main()