Optimization Tutorial¶

This tutorial demonstrates optimization techniques for quantum data embeddings.

Optimization Tutorial¶

This tutorial demonstrates optimization techniques for quantum data embeddings.

In [ ]:

import numpy as np
from quantum_data_embedding_suite.embeddings import AngleEmbedding
from quantum_data_embedding_suite.metrics import expressibility, trainability
import matplotlib.pyplot as plt

import numpy as np from quantum_data_embedding_suite.embeddings import AngleEmbedding from quantum_data_embedding_suite.metrics import expressibility, trainability import matplotlib.pyplot as plt

Hyperparameter Optimization¶

Learn how to optimize embedding hyperparameters for better performance.

In [ ]:

# Create sample data
X = np.random.randn(100, 4)

# Test different hyperparameters
results = {}
for n_qubits in [3, 4, 5]:
    embedding = AngleEmbedding(n_qubits=n_qubits)
    
    expr_score = expressibility(embedding, n_samples=200)
    train_score = trainability(embedding, data=X[:50])
    
    results[n_qubits] = {
        'expressibility': expr_score,
        'trainability': train_score
    }
    
    print(f"n_qubits={n_qubits}: expr={expr_score:.4f}, train={train_score:.4f}")

# Create sample data X = np.random.randn(100, 4) # Test different hyperparameters results = {} for n_qubits in [3, 4, 5]: embedding = AngleEmbedding(n_qubits=n_qubits) expr_score = expressibility(embedding, n_samples=200) train_score = trainability(embedding, data=X[:50]) results[n_qubits] = { 'expressibility': expr_score, 'trainability': train_score } print(f"n_qubits={n_qubits}: expr={expr_score:.4f}, train={train_score:.4f}")

Performance Analysis¶

Analyze the performance characteristics of different embeddings.

In [ ]:

# Plot results
n_qubits_list = list(results.keys())
expr_scores = [results[n]['expressibility'] for n in n_qubits_list]
train_scores = [results[n]['trainability'] for n in n_qubits_list]

plt.figure(figsize=(10, 4))

plt.subplot(1, 2, 1)
plt.plot(n_qubits_list, expr_scores, 'o-')
plt.xlabel('Number of Qubits')
plt.ylabel('Expressibility')
plt.title('Expressibility vs Qubits')
plt.grid(True)

plt.subplot(1, 2, 2)
plt.plot(n_qubits_list, train_scores, 'o-', color='red')
plt.xlabel('Number of Qubits')
plt.ylabel('Trainability')
plt.title('Trainability vs Qubits')
plt.grid(True)

plt.tight_layout()
plt.show()

# Plot results n_qubits_list = list(results.keys()) expr_scores = [results[n]['expressibility'] for n in n_qubits_list] train_scores = [results[n]['trainability'] for n in n_qubits_list] plt.figure(figsize=(10, 4)) plt.subplot(1, 2, 1) plt.plot(n_qubits_list, expr_scores, 'o-') plt.xlabel('Number of Qubits') plt.ylabel('Expressibility') plt.title('Expressibility vs Qubits') plt.grid(True) plt.subplot(1, 2, 2) plt.plot(n_qubits_list, train_scores, 'o-', color='red') plt.xlabel('Number of Qubits') plt.ylabel('Trainability') plt.title('Trainability vs Qubits') plt.grid(True) plt.tight_layout() plt.show()

In [ ]:

import numpy as np
from quantum_data_embedding_suite.embeddings import AngleEmbedding
from quantum_data_embedding_suite.metrics import expressibility, trainability
import matplotlib.pyplot as plt

import numpy as np from quantum_data_embedding_suite.embeddings import AngleEmbedding from quantum_data_embedding_suite.metrics import expressibility, trainability import matplotlib.pyplot as plt

Hyperparameter Optimization¶

Learn how to optimize embedding hyperparameters for better performance.

In [ ]:

# Create sample data
X = np.random.randn(100, 4)

# Test different hyperparameters
results = {}
for n_qubits in [3, 4, 5]:
    embedding = AngleEmbedding(n_qubits=n_qubits)
    
    expr_score = expressibility(embedding, n_samples=200)
    train_score = trainability(embedding, data=X[:50])
    
    results[n_qubits] = {
        'expressibility': expr_score,
        'trainability': train_score
    }
    
    print(f"n_qubits={n_qubits}: expr={expr_score:.4f}, train={train_score:.4f}")

# Create sample data X = np.random.randn(100, 4) # Test different hyperparameters results = {} for n_qubits in [3, 4, 5]: embedding = AngleEmbedding(n_qubits=n_qubits) expr_score = expressibility(embedding, n_samples=200) train_score = trainability(embedding, data=X[:50]) results[n_qubits] = { 'expressibility': expr_score, 'trainability': train_score } print(f"n_qubits={n_qubits}: expr={expr_score:.4f}, train={train_score:.4f}")

Performance Analysis¶

Analyze the performance characteristics of different embeddings.

In [ ]:

# Plot results
n_qubits_list = list(results.keys())
expr_scores = [results[n]['expressibility'] for n in n_qubits_list]
train_scores = [results[n]['trainability'] for n in n_qubits_list]

plt.figure(figsize=(10, 4))

plt.subplot(1, 2, 1)
plt.plot(n_qubits_list, expr_scores, 'o-')
plt.xlabel('Number of Qubits')
plt.ylabel('Expressibility')
plt.title('Expressibility vs Qubits')
plt.grid(True)

plt.subplot(1, 2, 2)
plt.plot(n_qubits_list, train_scores, 'o-', color='red')
plt.xlabel('Number of Qubits')
plt.ylabel('Trainability')
plt.title('Trainability vs Qubits')
plt.grid(True)

plt.tight_layout()
plt.show()

# Plot results n_qubits_list = list(results.keys()) expr_scores = [results[n]['expressibility'] for n in n_qubits_list] train_scores = [results[n]['trainability'] for n in n_qubits_list] plt.figure(figsize=(10, 4)) plt.subplot(1, 2, 1) plt.plot(n_qubits_list, expr_scores, 'o-') plt.xlabel('Number of Qubits') plt.ylabel('Expressibility') plt.title('Expressibility vs Qubits') plt.grid(True) plt.subplot(1, 2, 2) plt.plot(n_qubits_list, train_scores, 'o-', color='red') plt.xlabel('Number of Qubits') plt.ylabel('Trainability') plt.title('Trainability vs Qubits') plt.grid(True) plt.tight_layout() plt.show()