CustomVQA

Divi

CustomVQA

Custom Variational Quantum Algorithm

CustomVQA lets you optimize any parameterized quantum circuit by wrapping a PennyLane QuantumScript or Qiskit QuantumCircuit. This is useful when you have a custom ansatz or problem-specific circuit that doesn’t fit neatly into VQE or QAOA.

PennyLane QuantumScript

import pennylane as qml
from divi.qprog import CustomVQA
from divi.qprog.optimizers import ScipyOptimizer, ScipyMethod
from divi import ParallelSimulator

# Build a parameterized QuantumScript
with qml.queuing.AnnotatedQueue() as q:
    qml.RX(0.5, wires=0)
    qml.RY(0.3, wires=1)
    qml.CNOT(wires=[0, 1])
    qml.RZ(0.1, wires=0)
    qml.expval(qml.PauliZ(0) @ qml.PauliZ(1))

tape = qml.tape.QuantumScript.from_queue(q)

vqa = CustomVQA(
    qscript=tape,
    optimizer=ScipyOptimizer(ScipyMethod.COBYLA),
    max_iterations=50,
    backend=ParallelSimulator(),
)

vqa.run()
print(f"Optimized energy: {min(vqa.losses[-1].values()):.4f}")

Qiskit QuantumCircuit

You can also pass a Qiskit QuantumCircuit with parameters. Divi will automatically convert it to a PennyLane script:

from qiskit import QuantumCircuit
from qiskit.circuit import Parameter

theta = Parameter("θ")
phi = Parameter("φ")

qc = QuantumCircuit(2)
qc.rx(theta, 0)
qc.ry(phi, 1)
qc.cx(0, 1)

vqa = CustomVQA(
    qscript=qc,
    optimizer=ScipyOptimizer(ScipyMethod.NELDER_MEAD),
    max_iterations=30,
    backend=ParallelSimulator(),
)

vqa.run()

Parameter Shape

By default, CustomVQA infers the parameter shape from the circuit’s trainable parameters. You can override this with the param_shape argument:

vqa = CustomVQA(
    qscript=tape,
    param_shape=(3,),  # Explicit shape
    ...
)

When to Use CustomVQA

You have a custom circuit design that doesn’t map to VQE or QAOA
You want to benchmark a specific ansatz against a known Hamiltonian
You’re prototyping a new variational algorithm
You want to optimize a circuit from a different framework (e.g., Qiskit) using Divi’s parallel execution

Time Evolution Checkpointing