好吧，感谢@SIGHUP和@Frank Yellin，我能够找到这个问题，所以如果有人遇到类似问题，我会在这里分享.

Python seems unable to 100 anything when there are too many concurrent processes running.

例如，判断您的程序是否有效的解决方案是将其写入. text文件.一旦流程过多，100 statements will NOT appear in the Python console.

我不知道这print个声明是否会冻结整个程序，或者是否会继续运行.然而，我建议删除任何印刷声明，以避免任何意外.

以下是一种让我的示例中的代码轻松工作的方法(小心，long0 000行或更多will take a 100 time行):

#minimal example
#changes according to SIGHUP and Frank Yellin

import time
import pandas as pd
import random as rd
import multiprocessing as mp

import psutil #RAM usage

import sys

def create_datafile(nrows):
    """
    create a random pandas dataframe file
    To visualize this rather simple example,
    let's say that we are looking at a pool of 0.1*nrows products across different stores,
    that can have different values of the attribute "price"
    (in the list "stores").
    """
    
    price = [rd.randint(0,300) for i in range(nrows)]
    stores = [i%(0.1*nrows) for i in range(nrows)]

    data=zip(stores,price)
 
    return pd.DataFrame(data=data, columns=["stores", "price"])

            
def task(store):
    global data
    global alive_file
    """
    the task we want to accomplish: compute mean price
    for each store in the dataframe.
    """

    #print('I am alive!',flush=True) DO NOT put a print statement

    with open(alive_file, 'a') as f:
        f.write("I am alive !")

    product_df = data.groupby('stores', as_index = False).agg(mean_price = ("price", 'mean'))

    selected_store = product_df[product_df['stores'] == store] #select the mean for a given store

    return (store, selected_store['mean_price'])

def pinit(_df, _alive_file):
    global data 
    global alive_file
    data = _df
    alive_file = _alive_file

def get_ram_usage_pct(): #source: https://www.pragmaticlinux.com/2020/12/monitor-cpu-and-ram-usage-in-python-with-psutil/
    """
    Obtains the system's current RAM usage.
    :returns: System RAM usage as a percentage.
    :rtype: float
    """
    return psutil.virtual_memory().percent


if __name__ == "__main__":
    ##
    nrows= int(sys.argv[1]) #number of rows in dataframe
    nb_jobs= int(sys.argv[2]) #number of jobs


    print('Creating data...')
    tic=time.time()
    df = create_datafile(nrows)
    toc=time.time()
    print(f'Data created. Took {round((toc-tic)/60,1)} minutes (rounded to 0.1)')

    print('RAM usage after data creation is {} %'.format(get_ram_usage_pct()))

    #print(data_df)

    #create parameters for multiprocessing task
    stores_list = [(i % (0.1 * nrows),) for i in range(nrows)]
    #dics_stores=[{} for _ in stores_list]
    #parameters = [(df, stores_list[i]) for i in range(nrows)]

    dic_mean={}
    #launch multiprocessing tasks with starmaps
    tic=time.time()
    print(f'Max number of jobs: {mp.cpu_count() - 1}')
    print(f'Running: {min(nb_jobs, mp.cpu_count() - 1)} jobs...')
    with mp.Pool(initializer=pinit, initargs=(df,"alive.txt",), processes=min(nb_jobs, mp.cpu_count() - 1)) as pool:
        for store,result in pool.starmap_async(task, stores_list).get():
            dic_mean[store] = result[store]
    toc=time.time()
    print(f'Processed data in {round((toc-tic)/60,1)} minutes (rounded to 0.1).') 

    #print(dic_mean)
    #dic_means now contains all the means computed by each program.

感谢大家花时间判断我的问题并改进我的代码，帮助我识别真正的问题.