pennylane-lightning

What it is and what it does

PennyLane-Lightning is a plugin that extends PennyLane with fast quantum simulators written in C++. It provides multiple backends: lightning.qubit for CPU-based state-vector simulation with optional OpenMP parallelization, lightning.kokkos for multi-platform parallelism (OpenMP, CUDA, HIP, MPI), lightning.gpu for NVIDIA GPU acceleration via cuQuantum SDK, lightning.amdgpu for AMD GPUs, and lightning.tensor for tensor-network simulation. The package depends on PennyLane and scipy-openblas32, and is designed to accelerate hybrid quantum-classical computations on modern hardware from laptops to HPC clusters.

You install it as a PennyLane device plugin and select a backend (e.g., 'lightning.qubit' or 'lightning.gpu') when creating a device. Most backends are available via pip for Linux x86, Linux ARM, macOS ARM, and Windows; GPU and MPI variants may require source builds on some platforms. The plugin is actively maintained, supports Python 3.11 and above, and has no known security vulnerabilities.

Use it for:

• Accelerate quantum machine learning training by using GPU-backed simulation instead of CPU-only backends.
• Run large-scale quantum circuit simulations on HPC clusters using MPI-distributed state-vector backends.
• Prototype quantum algorithms with fast state-vector simulation on consumer hardware before deploying to quantum hardware.
• Simulate tensor-network quantum circuits using Matrix Product State or Exact Tensor Network methods for structured problems.
• Develop hybrid quantum-classical optimization loops where the simulator is the bottleneck and GPU acceleration is needed.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

PennyLane-Lightning provides high-performance quantum state-vector and tensor-network simulators as a plugin for PennyLane, with backends optimized for CPUs, GPUs (NVIDIA and AMD), and distributed computing via MPI.

Yes. PennyLane-Lightning is a mature, actively maintained plugin with low install friction, no security vulnerabilities, and permissive licensing. Install it if you use PennyLane and need faster simulation than the default backend, especially for GPU-accelerated or distributed workloads. The CPU-only lightning.qubit backend is available on all major platforms via pip; GPU backends require appropriate hardware and drivers but are well-documented.

Install

pennylane-lightning on PyPI

pip install pennylane-lightning

uv add pennylane-lightning

poetry add pennylane-lightning

Installing pennylane-lightning

Before you install

Low installation friction with a pure-wheel distribution. Actively maintained with recent commits and a stable release cadence; supports current Python versions (3.11+).

License in practice

Released under Apache License 2.0 (permissive), allowing commercial use and modification. GPU backends use NVIDIA cuQuantum SDK headers under their own respective licenses.

Quickstart

pip install pennylane-lightning

import pennylane as pnl
dev = pnl.device('lightning.qubit', wires=2)

@pnl.qnode(dev)
def circuit():
    pnl.PauliX(wires=0)
    return pnl.expval(pnl.PauliZ(0))

print(circuit())

Requires Python 3.11 or above. GPU backends (lightning.gpu, lightning.amdgpu, lightning.kokkos with CUDA/HIP) require corresponding GPU drivers and toolkits; most are available via pip for common platforms, but some require source builds.

Verify before relying

• Performance benchmarks comparing lightning.qubit, lightning.gpu, and lightning.tensor backends on typical workloads.
• Whether scipy-openblas32 is automatically installed or must be explicitly configured for optimal performance.
• Compatibility and performance characteristics of MPI-enabled backends on specific HPC clusters.

Package facts

Evidence: pennylane_lightning-0.45.0-py3-none-any.whl

Tags

More Physics packages

Further reading