System shut down with `HawkesEM` and `multiprocessing`

[rambor12]

Hello,

During the Monte Carlo simulation, I found that the HawkesEM with multiprocessing using apply_async leads to a sudden system shut down. This is still happening with another parallel processing package ray. Since I am not an expert in software engineering, I have no idea how to handle this problem. So I just ask here to get any hint to solve this issue. Here is the example:

import gc
import numpy as np
from tick.hawkes import SimuHawkesExpKernels, HawkesExpKern, HawkesEM

def run(seed):

    # Point generation
    hawkes = SimuHawkesExpKernels([[alpha/beta]], [[beta]], baseline, 
                                  end_time=T, verbose=False, seed=seed, 
                                  period_length=T)

    hawkes.simulate()
    points = hawkes.timestamps[0]
    del hawkes

    # Estimation    
    model = HawkesEM(kernel_support=10)
    model.fit([[points]])
    res = model.kernel.sum()

    del points, model
    gc.collect()
    print(end='.')

    return np.array(res)

N_sim = 6000
alpha = 1.6
beta = 2.0
T = 40000.
baseline = np.repeat(1,T).reshape((1,-1))

from functools import partial
from multiprocessing import Pool, cpu_count

def log_out(i, out):
    pool_out[i] = out

pool_out = {}
pool = Pool(cpu_count()-2)
try:
    for seed in range(N_sim):
        if seed not in pool_out:
            pool.apply_async(run, args=(seed,), 
                             callback=partial(log_out, seed))
    pool.close()
    pool.join()
except KeyboardInterrupt:
    pool.terminate()
pool_out=dict(sorted(pool_out.items()))

The run tine would be several minutes depends on your system. Here is my system info:

Linux-5.4.0-107-generic-x86_64-with-debian-buster-sid
Python: 3.7.3 (default, Mar 27 2019, 22:11:17) 
[GCC 7.3.0]
Intel(R) Core(TM) i9-7940X CPU @ 3.10GHz
tick Version: 0.6.0.0

Thank you for the reading.