PennyLane Export

Model. This notebook exports a three-site transverse-field Ising DenseHamiltonian and its Pauli-term metadata to PennyLane.

Typical uses. Verifying interoperability, preparing VQE/QPE prototypes, checking wire ordering, and confirming that exported Pauli sums preserve the package matrix.

Parameters. n_sites, Ising coupling j, transverse field h, and PennyLane wire_order determine the demonstrated export. PennyLane remains an optional dependency.

Useful plots. Package Hamiltonian magnitude and the absolute matrix difference after PennyLane reconstruction.

import importlib.util

import matplotlib.pyplot as plt
import numpy as np

from quantum_lattice_models.export import to_pennylane_terms
from quantum_lattice_models.models import transverse_field_ising

pennylane_available = importlib.util.find_spec("pennylane") is not None
print(f"PennyLane available: {pennylane_available}")

PennyLane available: True

H = transverse_field_ising(n_sites=3, j=1.0, h=0.5)
print("Package Hamiltonian")
print(f"  shape: {H.shape}")
print(f"  Pauli terms: {len(H.terms)}")
for term in H.terms:
    print(f"  {term.coefficient:+.2f} * {''.join(term.operators)}")

Package Hamiltonian
  shape: (8, 8)
  Pauli terms: 5
  -1.00 * ZZI
  -1.00 * IZZ
  -0.50 * XII
  -0.50 * IXI
  -0.50 * IIX

if pennylane_available:
    import pennylane as qml

    qml_H = to_pennylane_terms(H)
    qml_matrix = qml.matrix(qml_H, wire_order=range(H.n_sites))
    difference = np.asarray(H) - qml_matrix
    print("Export validation")
    print(f"  PennyLane terms: {len(qml_H.coeffs)}")
    print(f"  matrices agree: {np.allclose(H, qml_matrix)}")
    print(f"  max absolute difference: {np.max(np.abs(difference)):.3e}")

    fig, axes = plt.subplots(1, 2, figsize=(8.5, 3.6))
    axes[0].imshow(np.abs(H), origin="lower", cmap="viridis")
    axes[0].set_title("Package |H|")
    axes[1].imshow(np.abs(difference), origin="lower", cmap="magma")
    axes[1].set_title("|H package - H PennyLane|")
    for ax in axes:
        ax.set_xlabel("Ket index")
        ax.set_ylabel("Bra index")
    fig.tight_layout()
else:
    print(
        "Install with `pip install quantum-lattice-models[pennylane]` "
        "to run the matrix comparison."
    )

Export validation
  PennyLane terms: 5
  matrices agree: True
  max absolute difference: 0.000e+00