PennyLane

0.44.1 · active · verified Wed Apr 15

PennyLane is a cross-platform Python library for differentiable programming of quantum computers, quantum machine learning, and quantum chemistry. It enables users to build, optimize, and deploy hybrid quantum-classical applications by seamlessly integrating with popular machine learning frameworks like NumPy, PyTorch, TensorFlow, and JAX. The library is under active development, with new versions and features released every few months, aiming to make quantum computing accessible for research and application development.

Warnings

breaking Maintenance support for NumPy versions less than 2.0 is deprecated as of PennyLane v0.44.x and will be completely dropped in v0.45. Future versions of PennyLane will only work with NumPy >= 2.0.
Fix: Upgrade your NumPy installation to version 2.0 or newer: `pip install "numpy>=2.0"`.
breaking Support for Intel MacOS platforms (x86) has been removed as of PennyLane v0.44.x, impacting Macs running on Intel processors. Support for Python 3.10 was deprecated in v0.42 and removed in subsequent releases. Minimum Python version for current PennyLane is 3.11.
Fix: If on MacOS, migrate to an ARM-based Mac or build x86 wheels manually if absolutely necessary. Ensure your Python environment is 3.11 or newer.
deprecated The `custom_decomps` keyword argument for `qml.device` is deprecated and will be removed in v0.45. New decomposition rules should be defined as quantum functions with registered resources using `qml.decomposition.enable_graph`.
Fix: Refactor custom decomposition logic to use the new graph-based decomposition system and `qml.decomposition` module.
deprecated Accessing hermiticity via `pennylane.operation.Operator.is_hermitian` is deprecated. Use `is_verified_hermitian` for better reflection of functionality or the `is_hermitian()` function for thorough verification.
Fix: Update code to use `operator.is_verified_hermitian` property or call the `qml.is_hermitian(operator)` function for a more comprehensive check.
breaking The argument `level=None` is no longer valid for functions like `qml.specs()`, `qml.draw()`, `workflow.get_transform_program()`, `workflow.construct_batch()`, `drawer.draw_mpl()`. It must be replaced with `level='device'` to apply all transforms.
Fix: Change any instances of `level=None` to `level='device'` when calling these introspection and workflow functions.

Install

pip install pennylane Stable release
pip install pennylane[all] With all optional dependencies (e.g., JAX, PyTorch, TensorFlow)

Imports

pennylane
```
import pennylane as qml
```
The official documentation and community widely adopt `qml` as the standard alias for PennyLane.

Quickstart

This quickstart demonstrates how to define a quantum device, create a quantum circuit (QNode) with parameters, execute it, and compute gradients using PennyLane's automatic differentiation capabilities. This is a fundamental workflow for variational quantum algorithms.

import pennylane as qml
from pennylane import numpy as np

# Define a quantum device
dev = qml.device("default.qubit", wires=2)

# Define a QNode (quantum function)
@qml.qnode(dev)
def circuit(phi, theta):
    qml.RX(phi[0], wires=0)
    qml.RY(phi[1], wires=1)
    qml.CNOT(wires=[0, 1])
    qml.RX(theta, wires=0)
    return qml.expval(qml.PauliZ(0))

# Define parameters with automatic differentiation enabled
phi_params = np.array([0.54, 0.12], requires_grad=True)
theta_param = np.array(0.9, requires_grad=True)

# Execute the circuit
result = circuit(phi_params, theta_param)
print(f"Circuit output: {result}")

# Compute gradients
grad_fn = qml.grad(circuit)
gradients = grad_fn(phi_params, theta_param)
print(f"Gradients: {gradients}")

view raw JSON →