timeit.repeat - 透過重複來理解模式

import timeit import matplotlib.pyplot as plt import json import os import statistics import numpy as np def run_experiment(number_of_repeats, number_of_runs=1000): execution_time = timeit.repeat( "from functools import reduce; reduce((lambda x, y: x * y), range(1, 2000))", repeat=number_of_repeats, number=number_of_runs ) return execution_time def save_result(result, repeats): filename = f'execution_time_results_{repeats}.json' with open(filename, 'w') as f: json.dump(result, f) def load_result(repeats): filename = f'execution_time_results_{repeats}.json' if os.path.exists(filename): with open(filename, 'r') as f: return json.load(f) return None def truncated_mean(data, percentile=95): data = np.array(data) lower_bound = np.percentile(data, (100 - percentile) / 2) upper_bound = np.percentile(data, 100 - (100 - percentile) / 2) return np.mean(data[(data >= lower_bound) & (data <= upper_bound)]) # List of number_of_repeats to test repeat_values = [5, 20, 100, 500, 1000, 3000] # Run experiments and collect results results = [] for repeats in repeat_values: result = load_result(repeats) if result is None: print(f"Running experiment for {repeats} repeats...") try: result = run_experiment(repeats) save_result(result, repeats) print(f"Experiment for {repeats} repeats completed and saved.") except KeyboardInterrupt: print(f"\nExperiment for {repeats} repeats interrupted.") continue else: print(f"Loaded existing results for {repeats} repeats.") # Print time taken per repetition avg_time = statistics.mean(result) print(f"Average time per repetition for {repeats} repeats: {avg_time:.6f} seconds") results.append(result) trunc_means = [truncated_mean(r) for r in results] medians = [np.median(r) for r in results] mins = [np.min(r) for r in results] maxs = [np.max(r) for r in results] # Create subplots fig, axs = plt.subplots(2, 2, figsize=(15, 12)) fig.suptitle('Execution Time Analysis for Different Number of Repeats', fontsize=16) metrics = [ ('Truncated Mean (95%)', trunc_means), ('Median', medians), ('Min', mins), ('Max', maxs) ] for (title, data), ax in zip(metrics, axs.flatten()): ax.plot(repeat_values, data, marker='o') ax.set_title(title) ax.set_xlabel('Number of Repeats') ax.set_ylabel('Execution Time (seconds)') ax.set_xscale('log') ax.grid(True, which="both", ls="-", alpha=0.2) # Set x-ticks and labels for each data point ax.set_xticks(repeat_values) ax.set_xticklabels(repeat_values) # Rotate x-axis labels for better readability ax.tick_params(axis='x', rotation=45) plt.tight_layout() # Save the plot to a file plt.savefig('execution_time_analysis.png', dpi=300, bbox_inches='tight') print("Plot saved as 'execution_time_analysis.png'") # Create histograms for data distribution with 10 bins fig, axs = plt.subplots(2, 3, figsize=(20, 12)) fig.suptitle('Data Distribution Histograms for Different Number of Repeats (10 bins)', fontsize=16) for repeat, result, ax in zip(repeat_values, results, axs.flatten()): ax.hist(result, bins=10, edgecolor='black') ax.set_title(f'Repeats: {repeat}') ax.set_xlabel('Execution Time (seconds)') ax.set_ylabel('Frequency') plt.tight_layout() # Save the histograms to a file plt.savefig('data_distribution_histograms_10bins.png', dpi=300, bbox_inches='tight') print("Histograms saved as 'data_distribution_histograms_10bins.png'") # Create histograms for 1000 and 3000 repeats with 30 bins fig, axs = plt.subplots(1, 2, figsize=(15, 6)) fig.suptitle('Data Distribution Histograms for 1000 and 3000 Repeats (30 bins)', fontsize=16) for repeat, result, ax in zip([1000, 3000], results[-2:], axs): ax.hist(result, bins=100, edgecolor='black') ax.set_title(f'Repeats: {repeat}') ax.set_xlabel('Execution Time (seconds)') ax.set_ylabel('Frequency') plt.tight_layout() # Save the detailed histograms to a file plt.savefig('data_distribution_histograms_detailed.png', dpi=300, bbox_inches='tight') print("Detailed histograms saved as 'data_distribution_histograms_detailed.png'") plt.show()