Recursive Python Algorithm to Model Among Us смотреть онлайн

The final installment in the game theory of Among Us series. See code below

0:00 - Explanation of recursive algorithms
3:04 - Problems with simple decision tree models
3:54 - Model with 'simulated rounds'
5:38 - Problems with the final model
6:06 - Features of the final model
7:29 - Explanation of code
11:45 - Skip on 7?
13:09 - Clutch scenarios
14:32 - All minimum certainty results
15:32 - All random voting results

Code (note that less than and greater than symbols are not allowed by youtube so theyre replaced by lessthan and greaterthan respectively):
import numpy as np


def ev(amount_of_impostors, amount_of_crew, certainty=None, main=False):
"""Recursive function that returns expected value of voting someone off"""
impostors, crew, winrate = amount_of_impostors, amount_of_crew, certainty
kills = [-1, 0, 0, 1]
# [0, 1, 1, 2] with 1 impostor
# [1, 2, 2, 3] with 2 impostors

if (crew lessthan= 2 * impostors and not main) or crew == 1:
# We lost! (impostors can kill remaining crew)
return -1
if impostors lessthan= 0:
# We won! (no remaining impostors)
return 1

if certainty is None:
# If no certainty is given, we are guessing
winrate = impostors / (impostors + crew)

if main:
# If in the main branch, don't factor in kills, just calculate EVs
ev_correct = ev(impostors - 1, crew, certainty)
ev_wrong = ev(impostors, crew - 1, certainty)

return ev_correct * winrate + ev_wrong * (1 - winrate)
else:
# If we're not in the main branch, create a new branch for each possible kill
evs = []
for kill in kills:
ev_correct = ev(impostors - 1, crew - kill - impostors, certainty)
ev_wrong = ev(impostors, crew - 1 - kill - impostors, certainty)

evs.append(ev_correct * winrate + ev_wrong * (1 - winrate))

return np.average(evs)


def minimum_certainty(amount_of_impostors, amount_of_crew, accuracy):
"""Return minimum certainty to get EV = 0"""
impostors, crew = amount_of_impostors, amount_of_crew
certainty, step = 0, 1 / accuracy

for i in range(accuracy):
value = ev(impostors, crew, certainty, main=True)
if value greaterthan= 0:
return certainty
else:
certainty += step
return "Greater than 1"