PennyLane Qiskit Plugin

PennyLane is a cross-platform Python library for quantum machine learning, automatic differentiation, and optimization of hybrid quantum-classical computations.

Qiskit is an open-source compilation framework capable of targeting various types of hardware and a high-performance quantum computer simulator with emulation capabilities, and various compiler plug-ins.

This PennyLane plugin allows to use both the software and hardware backends of Qiskit as devices for PennyLane.

Features

• Provides three devices to be used with PennyLane: qiskit.basicaer, qiskit.aer and qiskit.ibmq. These devices provide access to the various backends.
• Supports a wide range of PennyLane operations and expectation values across the providers.
• Combine Qiskit's high performance simulator and hardware backend support with PennyLane's automatic differentiation and optimization.

Installation

This plugin requires Python version 3.5 and above, as well as PennyLane and Qiskit. Installation of this plugin, as well as all dependencies, can be done using pip:

pip install pennylane-qiskit

To test that the PennyLane Qiskit plugin is working correctly you can run

$ make test

in the source folder. Tests restricted to a specific provider can be run by executing make test-basicaer, make test-aer make test-ibmq.

Note

Tests on the IBMQ device can only be run if a ibmqx_token for the IBM Q experience is configured in the PennyLane configuration file.

If this is the case, running make test also executes tests on the ibmq device. By default tests on the ibmq device run with ibmq_qasm_simulator backend and those done by the basicaer and aer device are run with the qasm_simulator backend. At the time of writing this means that the test are "free". Please verify that this is also the case for your account.

Getting started

Once the PennyLane-Qiskit plugin is installed, the three provided Qiskit devices can be accessed straightaway in PennyLane.

You can instantiate a 'qiskit.aer' device for PennyLane with:

import pennylane as qml
dev = qml.device('qiskit.aer', wires=2)

This device can then be used just like other devices for the definition and evaluation of QNodes within PennyLane. A simple quantum function that returns the expectation value of a measurement and depends on three classical input parameters would look like:

@qml.qnode(dev)
def circuit(x, y, z):
    qml.RZ(z, wires=[0])
    qml.RY(y, wires=[0])
    qml.RX(x, wires=[0])
    qml.CNOT(wires=[0, 1])
    return qml.expval(qml.PauliZ(wires=1))

You can then execute the circuit like any other function to get the quantum mechanical expectation value.

circuit(0.2, 0.1, 0.3)

You can also change the default device's backend with

dev = qml.device('qiskit.aer', wires=2, backend='unitary_simulator')

To get a current overview what backends are available you can query this by

dev.capabilities()['backend']

While the device 'qiskit.aer' is the standard go-to simulator that is provided along the Qiskit main package installation, there exists a natively included python simulator that is slower but will work usually without the need to check out other dependencies (gcc, blas and so on) which can be used by 'qiskit.basicaer'.

Another important difference between the two is that while 'qiskit.aer' supports a simulation with noise, 'qiskit.basicaer' does not.

Noise models

You can instantiate a noise model and apply it to the device by calling

import pennylane as qml

import qiskit
from qiskit.providers.aer.noise.device import basic_device_noise_model

qiskit.IBMQ.load_account()
provider = qiskit.IBMQ.get_provider(group='open')
ibmq_16_melbourne = provider.get_backend('ibmq_16_melbourne')
device_properties = ibmq_16_melbourne.properties()

noise_model = basic_device_noise_model(device_properties)

dev = qml.device('qiskit.aer', wires=2, noise_model=noise_model)

Please refer to the Qiskit documentation for more information on noise models.

IBM Q Experience

PennyLane-Qiskit supports running PennyLane on IBM Q hardware via the qistkit.ibmq device. You can choose between different backends - either simulators or real hardware.

import pennylane as qml
dev = qml.device('qiskit.ibmq', wires=2, backend='ibmq_16_melbourne')

By default, the qiskit.ibmq device will attempt to use an already active or stored IBM Q account. If none are available, you may also directly pass your IBM Q API token, as well as an optional URL:

import pennylane as qml
dev = qml.device('qiskit.ibmq', wires=2, backend='ibmq_qasm_simulator', ibmqx_token="XXX")

In order to avoid accidentally publishing your token, it is best to store it using the qiskit.IBMQ.save_account() function. Alternatively, you can specify the token or URL via the PennyLane configuration file by adding a section such as

[qiskit.global]

  [qiskit.ibmq]
  ibmqx_token = "XXX"
  ibmqx_url = "XXX"

Note that, by default, the qiskit.ibmq device uses the simulator backend ibmq_qasm_simulator, but this may be changed to any of the real backends as given by

dev.capabilities()['backend']

Please refer to the plugin documentation as well as to the PennyLane documentation for further reference.

How to cite

If you are doing research using PennyLane, please cite our whitepaper:

Ville Bergholm, Josh Izaac, Maria Schuld, Christian Gogolin, Carsten Blank, Keri McKiernan and Nathan Killoran. PennyLane. arXiv, 2018. arXiv:1811.04968

Contributing

We welcome contributions - simply fork the repository of this plugin, and then make a pull request containing your contribution. All contributers to this plugin will be listed as authors on the releases.

We also encourage bug reports, suggestions for new features and enhancements, and even links to cool projects or applications built on PennyLane.

Authors

Shahnawaz Ahmed, Carsten Blank, Sebastian Boerakker, Christian Gogolin, Josh Izaac.

Support

• Source Code: https://github.com/XanaduAI/pennylane-qiskit
• Issue Tracker: https://github.com/XanaduAI/pennylane-qiskit/issues

If you are having issues, please let us know by posting the issue on our Github issue tracker.

License

The PennyLane qiskit plugin is free and open source, released under the Apache License, Version 2.0.