QAOA性能比较技术报告 (Qibo vs Qiskit)¶

测试环境配置¶

• 硬件: Windows 10 Pro, x64
• 量子比特数: 4-15
• QAOA层数: 1-5
• 优化器: COBYLA
• 采样数: 8192 shots

数据加载与预处理¶

In [2]:

!jupyter kernelspec list

!jupyter kernelspec list

Available kernels:
  python3    E:\quantum computing\venv\share\jupyter\kernels\python3
  qenv       C:\Users\Administrator\AppData\Roaming\jupyter\kernels\qenv
  sjenv      C:\Users\Administrator\AppData\Roaming\jupyter\kernels\sjenv
  venv       C:\Users\Administrator\AppData\Roaming\jupyter\kernels\venv

In [1]:

import pandas as pd
import matplotlib.pyplot as plt

# 加载数据
qibo_df = pd.read_csv('qaoa_qibo_benchmark_results.csv')
qiskit_df = pd.read_csv('qaoa_benchmark_results.csv')

# 添加模拟器标签
qibo_df['simulator'] = 'Qibo'
qiskit_df['simulator'] = 'Qiskit'

# 合并数据
combined_df = pd.concat([qibo_df, qiskit_df])

import pandas as pd import matplotlib.pyplot as plt # 加载数据 qibo_df = pd.read_csv('qaoa_qibo_benchmark_results.csv') qiskit_df = pd.read_csv('qaoa_benchmark_results.csv') # 添加模拟器标签 qibo_df['simulator'] = 'Qibo' qiskit_df['simulator'] = 'Qiskit' # 合并数据 combined_df = pd.concat([qibo_df, qiskit_df])

关键性能指标对比¶

1. 运行时间对比 (秒)¶

In [2]:

plt.figure(figsize=(10, 6))
for simulator in ['Qibo', 'Qiskit']:
    df = combined_df[combined_df['simulator'] == simulator]
    plt.plot(df['num_qubits'], df['runtime_sec'], 'o-', label=simulator)
    
plt.title('Runtime Comparison (Qibo vs Qiskit)')
plt.xlabel('Number of Qubits')
plt.ylabel('Runtime (seconds)')
plt.legend()
plt.grid()
plt.show()

plt.figure(figsize=(10, 6)) for simulator in ['Qibo', 'Qiskit']: df = combined_df[combined_df['simulator'] == simulator] plt.plot(df['num_qubits'], df['runtime_sec'], 'o-', label=simulator) plt.title('Runtime Comparison (Qibo vs Qiskit)') plt.xlabel('Number of Qubits') plt.ylabel('Runtime (seconds)') plt.legend() plt.grid() plt.show()

2. 内存使用对比 (MB)¶

In [3]:

plt.figure(figsize=(10, 6))
for simulator in ['Qibo', 'Qiskit']:
    df = combined_df[combined_df['simulator'] == simulator]
    plt.plot(df['num_qubits'], df['memory_usage_mb'], 's-', label=simulator)
    
plt.title('Memory Usage Comparison (Qibo vs Qiskit)')
plt.xlabel('Number of Qubits')
plt.ylabel('Memory Usage (MB)')
plt.legend()
plt.grid()
plt.show()

plt.figure(figsize=(10, 6)) for simulator in ['Qibo', 'Qiskit']: df = combined_df[combined_df['simulator'] == simulator] plt.plot(df['num_qubits'], df['memory_usage_mb'], 's-', label=simulator) plt.title('Memory Usage Comparison (Qibo vs Qiskit)') plt.xlabel('Number of Qubits') plt.ylabel('Memory Usage (MB)') plt.legend() plt.grid() plt.show()

3. 最优解概率对比¶

In [4]:

plt.figure(figsize=(10, 6))
for simulator in ['Qibo', 'Qiskit']:
    df = combined_df[combined_df['simulator'] == simulator]
    plt.plot(df['num_qubits'], df['best_probability'], 'd-', label=simulator)
    
plt.title('Optimal Solution Probability Comparison')
plt.xlabel('Number of Qubits')
plt.ylabel('Probability')
plt.legend()
plt.grid()
plt.show()

plt.figure(figsize=(10, 6)) for simulator in ['Qibo', 'Qiskit']: df = combined_df[combined_df['simulator'] == simulator] plt.plot(df['num_qubits'], df['best_probability'], 'd-', label=simulator) plt.title('Optimal Solution Probability Comparison') plt.xlabel('Number of Qubits') plt.ylabel('Probability') plt.legend() plt.grid() plt.show()

性能差异分析¶

1. Qibo优势:

• 专为量子计算优化的后端
• 更高效的电路编译和优化
• 原生支持参数化量子门

2. Qiskit特点:

• 更成熟稳定的模拟器
• 更好的大规模电路处理能力
• 内存管理更高效

结论与建议¶

1. 对于≤12 qubits的QAOA电路，推荐使用Qibo以获得更快运行速度
2. 对于大规模电路(≥15 qubits)，Qiskit表现更稳定
3. 两种模拟器都能正确解决MAX-CUT问题，选择取决于具体规模和要求

完整数据表格¶

In [9]:

# 原代码替换为：
display(
    qibo_df
    .groupby(['num_qubits', 'qaoa_layers'])
    .agg({
        'runtime_sec': 'mean',
        'memory_usage_mb': 'mean', 
        'best_probability': 'mean'
    })
)

# 原代码替换为： display( qibo_df .groupby(['num_qubits', 'qaoa_layers']) .agg({ 'runtime_sec': 'mean', 'memory_usage_mb': 'mean', 'best_probability': 'mean' }) )

		runtime_sec	memory_usage_mb	best_probability
num_qubits	qaoa_layers
4	1	0.53	21.25	0.264526
	2	0.11	0.23	0.270874
	3	0.23	0.31	0.487305
	4	0.25	0.22	0.470703
	5	0.33	0.42	0.482300
5	1	0.08	0.04	0.092285
	2	0.15	0.10	0.093384
	3	0.22	0.04	0.095459
	4	0.32	0.12	0.100220
	5	0.44	0.34	0.104736
6	1	0.08	0.03	0.139648
	2	0.15	0.00	0.167847
	3	0.26	0.09	0.274292
	4	0.39	0.07	0.351196
	5	0.43	0.43	0.473389
7	1	0.10	0.13	0.045898
	2	0.17	0.00	0.065308
	3	0.28	0.00	0.070312
	4	0.40	0.43	0.073853
	5	0.59	0.87	0.066040
8	1	0.13	0.00	0.078003
	2	0.18	0.06	0.103882
	3	0.41	0.19	0.211426
	4	0.60	0.41	0.439575
	5	0.75	1.04	0.240112
9	1	0.15	0.02	0.027100
	2	0.27	0.03	0.045044
	3	0.47	0.00	0.045044
	4	0.70	0.48	0.033813
	5	1.14	1.16	0.057251
10	1	0.25	0.12	0.039917
	2	0.39	0.00	0.050171
	3	0.58	0.27	0.160645
	4	1.09	1.03	0.202148
	5	1.20	1.19	0.133057
11	1	0.24	0.33	0.014526
	2	0.54	0.00	0.023804
	3	0.98	0.11	0.027710
	4	1.62	1.04	0.030884
	5	2.41	2.25	0.033691
12	1	0.45	0.98	0.020996
	2	0.76	0.26	0.027954
	3	1.58	0.06	0.089478
	4	2.20	1.08	0.180908
	5	3.17	2.14	0.164307
13	1	0.79	2.08	0.007324
	2	1.40	-0.47	0.008545
	3	2.72	1.26	0.017822
	4	3.76	1.82	0.021484
	5	5.46	3.28	0.033447
14	1	1.20	2.63	0.010498
	2	2.76	0.06	0.034912
	3	4.25	1.50	0.012451
	4	7.17	-7.80	0.018188
	5	11.13	2.08	0.047852
15	1	2.36	0.68	0.004272
	2	4.44	2.33	0.010498
	3	9.34	7.44	0.017700
	4	14.25	-3.36	0.009888
	5	23.22	11.94	0.012207

In [12]:

# 原代码替换为：
display(
    qiskit_df
    .groupby(['num_qubits', 'qaoa_layers'])
    .agg({
        'runtime_sec': 'mean',
        'memory_usage_mb': 'mean', 
        'best_probability': 'mean'
    })
)

# 原代码替换为： display( qiskit_df .groupby(['num_qubits', 'qaoa_layers']) .agg({ 'runtime_sec': 'mean', 'memory_usage_mb': 'mean', 'best_probability': 'mean' }) )

		runtime_sec	memory_usage_mb	best_probability
num_qubits	qaoa_layers
4	1	4.37	7.80	0.262939
	2	7.94	0.59	0.492676
	3	10.41	0.98	0.484619
	4	11.11	1.16	0.484863
	5	11.40	0.31	0.432007
5	1	3.81	0.00	0.094238
	2	6.34	0.00	0.098877
	3	7.65	-0.05	0.105713
	4	11.57	0.00	0.103394
	5	13.98	0.04	0.099976
6	1	5.48	0.01	0.143311
	2	6.65	0.00	0.265991
	3	10.77	0.00	0.199097
	4	12.74	0.00	0.445190
	5	13.40	-0.02	0.216675
7	1	4.88	0.00	0.047729
	2	6.49	0.00	0.058716
	3	8.62	-0.01	0.071167
	4	9.32	0.08	0.067505
	5	13.06	0.01	0.070190
8	1	5.10	0.03	0.080444
	2	5.77	0.00	0.122559
	3	7.70	0.00	0.084229
	4	11.19	0.00	0.130859
	5	11.61	0.01	0.270630
9	1	4.35	0.00	0.026855
	2	7.09	0.57	0.046875
	3	8.20	0.00	0.030884
	4	13.46	-0.01	0.049438
	5	14.12	-0.01	0.035522
10	1	4.31	0.02	0.039917
	2	5.98	0.56	0.051636
	3	10.73	0.02	0.165527
	4	9.42	0.05	0.162354
	5	13.16	0.00	0.141602
11	1	4.55	0.00	0.015259
	2	6.06	0.02	0.012573
	3	7.63	0.00	0.014526
	4	11.50	0.00	0.028564
	5	13.14	0.00	0.038208
12	1	5.52	0.00	0.021973
	2	7.11	0.00	0.053955
	3	9.56	0.00	0.080688
	4	12.53	0.00	0.047363
	5	16.59	0.10	0.081665
13	1	4.93	-0.07	0.008789
	2	8.04	0.00	0.020020
	3	8.81	0.00	0.021362
	4	15.58	0.00	0.023315
	5	13.46	0.14	0.015625
14	1	5.03	0.00	0.011719
	2	7.21	0.00	0.031128
	3	8.05	0.55	0.022339
	4	13.07	0.01	0.049072
	5	15.13	0.00	0.018677
15	1	5.24	0.06	0.005005
	2	5.60	0.20	0.006104
	3	8.52	-0.05	0.008179
	4	11.11	0.63	0.013916
	5	12.69	-0.34	0.008911