QML-MCP

"""Tests for quantum ML utilities.""" import pytest import numpy as np from qiskit import QuantumCircuit from qml.utils import ( QuantumCircuitRunner, QuantumKernelComputer, VQCTrainer, ModelEvaluator, serialize_numpy, ) @pytest.mark.asyncio async def test_circuit_runner_basic(): """Test basic circuit execution.""" runner = QuantumCircuitRunner(max_qubits=10, max_shots=1000) # Create simple circuit circuit = QuantumCircuit(2) circuit.h(0) circuit.cx(0, 1) circuit.measure_all() result = await runner.run_circuit(circuit, shots=100) assert result["success"] is True assert result["shots"] == 100 assert result["num_qubits"] == 2 assert "counts" in result @pytest.mark.asyncio async def test_circuit_runner_validation(): """Test circuit validation.""" runner = QuantumCircuitRunner(max_qubits=2, max_shots=1000) # Circuit with too many qubits circuit = QuantumCircuit(5) circuit.measure_all() with pytest.raises(ValueError, match="exceeds maximum"): await runner.run_circuit(circuit) # Too many shots circuit = QuantumCircuit(2) circuit.measure_all() with pytest.raises(ValueError, match="exceeds maximum"): await runner.run_circuit(circuit, shots=10000) @pytest.mark.asyncio async def test_kernel_computer(): """Test quantum kernel computation.""" computer = QuantumKernelComputer(max_qubits=10) # Simple dataset train_data = np.array([[0.1, 0.2], [0.3, 0.4], [0.5, 0.6]]) result = await computer.compute_kernel(train_data) assert result["success"] is True assert "kernel_matrix" in result assert result["feature_dimension"] == 2 # Check kernel matrix shape kernel_matrix = np.array(result["kernel_matrix"]) assert kernel_matrix.shape == (3, 3) @pytest.mark.asyncio async def test_kernel_computer_with_test_data(): """Test kernel computation with separate test data.""" computer = QuantumKernelComputer(max_qubits=10) train_data = np.array([[0.1, 0.2], [0.3, 0.4]]) test_data = np.array([[0.5, 0.6]]) result = await computer.compute_kernel(train_data, test_data) assert result["success"] is True kernel_matrix = np.array(result["kernel_matrix"]) # When test_data is provided, kernel is computed as K(test, train) # resulting in shape (n_test, n_train) = (1, 2) # But Qiskit returns (n_train, n_test) = (2, 1) assert kernel_matrix.shape == (2, 1) or kernel_matrix.shape == (1, 2) @pytest.mark.asyncio async def test_vqc_trainer(): """Test VQC training.""" trainer = VQCTrainer(max_qubits=10) # Simple binary classification dataset X_train = np.array([ [0.1, 0.2], [0.2, 0.3], [0.8, 0.9], [0.9, 0.8] ]) y_train = np.array([0, 0, 1, 1]) result = await trainer.train(X_train, y_train, max_iter=10) assert result["success"] is True assert "model" in result assert "train_score" in result assert result["feature_dimension"] == 2 assert result["num_samples"] == 4 @pytest.mark.asyncio async def test_model_evaluator(): """Test model evaluation.""" # First train a model trainer = VQCTrainer(max_qubits=10) X_train = np.array([[0.1, 0.2], [0.2, 0.3], [0.8, 0.9], [0.9, 0.8]]) y_train = np.array([0, 0, 1, 1]) train_result = await trainer.train(X_train, y_train, max_iter=10) model_base64 = train_result["model"] # Evaluate on test data X_test = np.array([[0.15, 0.25], [0.85, 0.95]]) y_test = np.array([0, 1]) eval_result = await ModelEvaluator.evaluate(model_base64, X_test, y_test) assert eval_result["success"] is True assert "predictions" in eval_result assert "accuracy" in eval_result assert len(eval_result["predictions"]) == 2 def test_serialize_numpy(): """Test numpy serialization.""" # Test array arr = np.array([1, 2, 3]) assert serialize_numpy(arr) == [1, 2, 3] # Test scalar assert serialize_numpy(np.int64(42)) == 42 assert serialize_numpy(np.float64(3.14)) == 3.14 # Test nested structure data = {"array": np.array([1, 2]), "value": np.int64(10)} result = serialize_numpy(data) assert result == {"array": [1, 2], "value": 10}