PennyLane and the Lightning plugins

PennyLane is an open-source software framework for quantum machine learning, quantum chemistry, and quantum computing, with the ability to run on all hardware. Maintained with ❤️ by Xanadu.

Documentation

For more information on PennyLane, including the demos, APIs and development guide, visit the PennyLane documentation site. You can also check the Lightning-plugins documentation:

• pennylane-lightning
• pennylane-lightning-gpu
• pennylane-lightning-kokkos

Download & install

Each image contains PennyLane and one of several high-performance plugins. Choose a version (e.g. v0.34.0) and append a device (plugin) name among the following:

• lightning-qubit: pennylane-lightning provides a fast state-vector simulator written in C++.
• lightning-gpu: pennylane-lightning-gpu is a plugin based on the NVIDIA cuQuantum SDK.
• lightning-kokkos-cuda: pennylane-lightning-kokkos parallelizes state-vector simulations using Kokkos' CUDA backend.
• lightning-kokkos-openmp: pennylane-lightning-kokkos parallelizes state-vector simulations using Kokkos' OpenMP backend.
• lightning-kokkos-rocm: pennylane-lightning-kokkos parallelizes state-vector simulations using Kokkos' HIP backend.

If you have Docker installed, download and spawn a container with pennylane-lightning as follows

docker run -v pwd:/io -it pennylaneai/pennylane:v0.34.0-lightning-qubit bash

On certain systems, there may be other solutions supporting Docker containers. For instance, NERSC computers (e.g. Perlmutter) have Shifter bringing containers to HPC. In this case, spawning a container is simple as

shifterimg pull pennylaneai/pennylane:v0.34.0-lightning-qubit
shifter --image=pennylaneai/pennylane:v0.34.0-lightning-qubit /bin/bash

where the first command downloads the image and the second spawns a container.

Test

You can test PennyLane and the lightning-qubit plugin, for example, with

pip install pytest pytest-mock flaky
pl-device-test --device default.qubit
pl-device-test --device default.qubit --shots 10000
pl-device-test --device lightning.qubit
pl-device-test --device lightning.qubit --shots 10000

Build

Decide on a target among:

• wheel-lightning-qubit
• wheel-lightning-gpu
• wheel-lightning-kokkos-cuda
• wheel-lightning-kokkos-openmp
• wheel-lightning-kokkos-rocm

For instance TARGET=wheel-lightning-qubit. Then the following command will build the target

docker build -f docker/Dockerfile --tag=${TARGET} --target ${TARGET} .

You may also feed a git tag or version as follows

docker build -f docker/Dockerfile --tag=${TARGET} --target ${TARGET} --build-arg="LIGHTNING_VERSION=${VERSION}" .

To start a container with a bash shell use

docker run -v `pwd`:/io -it ${TARGET} bash