Joblib with objects

Question

I'm a newcomer w.r.t. parallel processing, and I'd like to ask for a bit of help here please.

I have an object that I need to replicate in memory, something like 50-100 times or even more. This is used in a biological virus simulator, where I try to simulate the replication and bursting of a virus inside a cell.

The code I am using right now uses deepcopy (please don't judge), and has the following logic:

from copy import deepcopy
from random import random
from datetime import datetime

import hashlib

class Virus(object):

    def __init__(self):
        self.id = None
        self.SetID()


    def SetID(self):
        random_number = str(random())
        current_time = str(datetime.now())

        unique_string = random_number + current_time

        unique_id = hashlib.new('sha512')
        unique_id.update(unique_string)

        self.id = unique_id.hexdigest()

    def GetID(self):
        return self.id

    def Replicate(self):
        new_virus = deepcopy(self)
        new_virus.SetID()

        return new_virus

Since my goal is to generate multiple copies of the virus, but I may potentially need to iterate many times, I thought of using joblib to speed things up. (I may be wrong, please correct me if this is so!) Therefore, I tried the following code:

h = Virus()
results = Parallel(n_jobs=2)(delayed(h.Replicate())(h) for i in range(10))

I expected results to be a list of 10 viruses, but instead, I get a TypeError saying:

TypeError: 'Virus' object is not callable

I can't seem to wrap my head around what is going on. How can I enable this to work?

Answer 1

delayed is to wrap a function (not an object, which would now return a replicated Virus). so do not call Replicate in delayed call

but also it seems you would not be able to wrap an instance method, so make replicate a regular function which would replicate its argument

def replicate(virus):
    return virus.Replicate()

h = Virus()
results = Parallel(n_jobs=2)(delayed(replicate)(h) for i in range(10))
print [x.id for x in results]

works for me