uniqc.algorithms.core.training.vqe_torch module

Variational Quantum Eigensolver (VQE) with TorchQuantum backend.

Provides VQESolver for finding ground state energies using TorchQuantum’s native PyTorch autograd. Includes built-in H2 molecule Hamiltonian.

class uniqc.algorithms.core.training.vqe_torch.VQESolver(hamiltonian, nuclear_repulsion=0.0, n_qubits=4, ansatz_fn=<function build_hea_circuit>, n_params=16, lr=0.05, device='cpu')

Bases: object

Variational Quantum Eigensolver with PyTorch optimization.

Parameters:

• hamiltonian – List of (pauli_string, coefficient).

• nuclear_repulsion – Constant energy offset.

• n_qubits – Number of qubits.

• ansatz_fn – Callable(params, n_qubits) -> (opcodes, n_qubits, overrides).

• n_params – Number of variational parameters.

• lr – Learning rate.

• device – “cpu” or “cuda”.

solve(n_iters=100, verbose=True)

Run VQE optimization.

Returns:

(final_energy, optimal_params)

step()

Single optimization step. Returns energy value.

uniqc.algorithms.core.training.vqe_torch.build_h2_hamiltonian(bond_length=0.735)

H2 molecule Hamiltonian in STO-3G basis (4 qubits, Bravyi-Kitaev).

Returns:

(pauli_terms, nuclear_repulsion) where pauli_terms is a list of (pauli_string, coefficient) tuples.

uniqc.algorithms.core.training.vqe_torch.build_hea_circuit(params, n_qubits, depth=2)

Build HEA circuit with torch.Tensor parameters.

Returns (opcode_list, n_qubits, param_overrides) for TorchQuantumSimulator.