How to make this simulation of which year a Solarflare hits the Earth faster?

Question

I wrote this code and try to learn a bit more how to code more efficiently and increase performance.

import random

def CalcAverageSolarFlareEvent(EventList):
    return sum(EventList) / len(EventList)

percentage_solar_flare = 12
decade_counting = 0
Event = []
CurrentYear = 2022

for Simulations in range(1, 999999):
    while True:
        if random.randint(1, 100) != percentage_solar_flare:
            decade_counting += 1
        else:
            Event.append(decade_counting)
            decade_counting = 0
            break

print("In the Year "+str(int(CalcAverageSolarFlareEvent(Event))*10+CurrentYear) +
      " we got a Solarflare")

I tried to calculate the decade_counting and adding current year at the end, to give more ram.

Answer 1

Python is not great for such a code, especially the standard CPython implementation. Consider using PyPy or Pyston or an embedded JIT (just-in-time compiler) like Numba, or alternatively a compiled language.

Moreover, you do not need to add items to a list so to count them or sum them: you can compute a partial sum on the fly.

Here is a modified code using the Numba JIT:

import random
import numba as nb

@nb.njit('()')
def compute():
    percentage_solar_flare = 12
    decade_counting = 0
    sum_Events = 0
    count_Event = 0
    CurrentYear = 2022

    for Simulations in range(1, 999_999):
        while True:
            if random.randint(1, 100) != percentage_solar_flare:
                decade_counting += 1
            else:
                sum_Events += decade_counting
                count_Event += 1
                decade_counting = 0
                break

    print("In the Year "+str(int(sum_Events/count_Event)*10+CurrentYear) +
          " we got a Solarflare")

compute()

The initial code takes 65 second on my machine while this one takes 4 seconds. Thus, it is about 16 times faster. Most of the time is spent in generating random values. SIMD instructions and multithreading can help to improve performance further (by 1~2 order of magnitude) but this is certainly not a good idea to use them is you are a beginner.