pytket-pennylane 0.19.0

Pytket extension and Pennylane plugin.

pytket-pennylane

Pytket extension and PennyLane plugin which allows pytket backends and compilation to be used as a PennyLane device.

Pytket is a quantum SDK python package which provides state of the art compilation for quantum circuits and a unified interface for execution on a number of "backends" (devices and simulators). PennyLane is a package for differentiable programming of quantum computer, which also provides a way to execute circuits on a variety of "devices". This package allows users to easily leverage the differentiablecircuits of PennyLane combined with the compilation available in Pytket.

The package is available for python 3.10, 3.11 and 3.12, and can be installed by cloning and installing from source, or via pip:

pip install pytket-pennylane

API documentation is here.

See the PennyLane documentation and Pytket documentation to get an intro to PennyLane and Pytket.

To use the integration once installed, initialise your pytket backend (in this example, an AerBackend which uses Qiskit Aer), and construct a PennyLane PytketDevice using this backend:

import pennylane as qml
from pytket.extensions.qiskit import AerBackend

# initialise pytket backend
pytket_backend = AerBackend()

# construct PennyLane device
dev = qml.device(
    "pytket.pytketdevice",
    wires=2,
    pytket_backend=pytket_backend,
    shots=1000
)

# define a PennyLane Qnode with this device
@qml.qnode(dev)
def my_quantum_function(x, y):
    qml.RZ(x, wires=0)
    qml.RX(y, wires=1)
    return qml.expval(qml.PauliZ(0) @ qml.PauliZ(1))

# call the node
print(my_quantum_function(0.1, 0.2))

The example above uses the Pytket default compilation pass for the backend, you can change the optimisation level of the default backend pass (0, 1 or 2) by setting the optimisation_level parameter:

dev = qml.device(
    "pytket.pytketdevice",
    wires=2,
    pytket_backend=pytket_backend,
    optimisation_level=2,
    shots=1000
)

You can also use any Pytket compilation pass using the compilation_pass parameter, which is used instead of the default pass:

from pytket.passes import PauliSimp, SequencePass

# use a Chemistry optimised pass before the backend's default pass

custom_pass = SequencePass([PauliSimp(), pytket_backend.default_compilation_pass()])

dev = qml.device(
    "pytket.pytketdevice",
    wires=2,
    pytket_backend=pytket_backend,
    compilation_pass=custom_pass,
    shots=1000
)