gym_wait_sim.py

gym_wait_sim.py

import numpy as np
import csv
from tqdm import tqdm
import seaborn as sns
import matplotlib.pyplot as plt
import pandas as pd

def simulate_wait_times(num_machines, num_simulations, mean, std_dev):
    wait_times = []

    for _ in range(num_simulations):
        total_workout_times = []
        for _ in range(num_machines):
            sampled_time = -1
            while sampled_time < 0:
                sampled_time = np.random.normal(mean, std_dev)
            total_workout_times.append(sampled_time)

        remaining_times = [np.random.uniform(0, workout_time) for workout_time in total_workout_times]
        wait_time = np.min(remaining_times)
        wait_times.append(wait_time)

    wait_times_array = np.array(wait_times)
    mean_wait = np.mean(wait_times_array)

    quantiles = [50, 75, 90, 95]
    quantile_values = np.percentile(wait_times_array, quantiles)

    return {
        'wait_times': wait_times_array,
        'mean': mean_wait,
        'p50': quantile_values[0],
        'p75': quantile_values[1],
        'p90': quantile_values[2],
        'p95': quantile_values[3],
        'num_machines': num_machines,
        'num_simulations': num_simulations,
        'mean_param': mean,
        'std_dev_param': std_dev
    }

param_sets = [{'num_machines': 10, 'num_simulations': 10000, 'mean': 30, 'std_dev': 10},
              {'num_machines': 10, 'num_simulations': 10000, 'mean': 25, 'std_dev': 10},
              {'num_machines': 10, 'num_simulations': 10000, 'mean': 20, 'std_dev': 10},
             {'num_machines': 10, 'num_simulations': 10000, 'mean': 30, 'std_dev': 5},
              {'num_machines': 10, 'num_simulations': 10000, 'mean': 20, 'std_dev': 5},
              {'num_machines': 10, 'num_simulations': 10000, 'mean': 25, 'std_dev': 5},
             {'num_machines': 10, 'num_simulations': 10000, 'mean': 30, 'std_dev': 15},
              {'num_machines': 10, 'num_simulations': 10000, 'mean': 20, 'std_dev': 15},
              {'num_machines': 10, 'num_simulations': 10000, 'mean': 25, 'std_dev': 15},]

results = []

for param_set in tqdm(param_sets, desc='Simulations'):
    num_machines = param_set['num_machines']
    num_simulations = param_set['num_simulations']
    mean = param_set['mean']
    std_dev = param_set['std_dev']
    
    simulation_result = simulate_wait_times(num_machines, num_simulations, mean, std_dev)
    results.append(simulation_result)
    # Generate and save figure for each param set
    fig, axes = plt.subplots()
    sns.distplot(simulation_result['wait_times'], ax=axes)
    axes.set_title(f'num_machines={num_machines}, num_simulations={num_simulations}, mean={mean}, std_dev={std_dev}')
    plt.savefig(f'num_machines={num_machines}_num_simulations={num_simulations}_mean={mean}_std_dev={std_dev}.png')
    plt.close()

# Generate final figure
fig, axes = plt.subplots()

for param_set, simulation_result in zip(param_sets, results):
    num_machines = param_set['num_machines']
    num_simulations = param_set['num_simulations']
    mean = param_set['mean']
    std_dev = param_set['std_dev']
    
    sns.distplot(simulation_result['wait_times'], ax=axes, 
                label=f'num_machines={num_machines}, num_simulations={num_simulations}, mean={mean}, std_dev={std_dev}')
    
axes.set_title('Combined Simulation Results')
plt.legend()
plt.show()

csv_filename = 'simulation_results.csv'
with open(csv_filename, mode='w', newline='') as file:
    fieldnames = ['num_machines', 'num_simulations', 'mean_param', 'std_dev_param', 'mean', 'p50', 'p75', 'p90', 'p95']
    writer = csv.DictWriter(file, fieldnames=fieldnames)
    writer.writeheader()
    writer.writerows([{k: v for k, v in i.items() if k in fieldnames} for i in results])