This plugin allows users of Pennylane to use Orquestra as a backend for simulating and executing quantum circuits. This enables Pennylane to run larger problems on Orquestra that might otherwise have been infeasible on a local simulator. As Antal Száva from Xanadu put it, “The plugin really excels with batched execution of multiple deep circuits. Combined with PennyLane's ability to compute quantum gradients, PennyLane-Orquestra allows us to explore larger quantum machine learning workflows in parallel.”
The PennyLane-Orquestra plugin is built on top of the Orquestra workflow API. The plugin auto-generates the workflow file and submits it to Orquestra’s Quantum Engine to run. In doing this, the Pennylane-Orquestra plugin allows users to take advantage of Orquestra’s ability to deploy and scale circuit simulations without having to compose a workflow by hand. For people who want to use Pennylane for differentiable programming and run large, batched operations, this is the tool for the job!
“PennyLane's approach of quantum differentiable programming and built-in support for quantum gradients allows you to use libraries such as TensorFlow, PyTorch, and Jax to drive and optimize remote Orquestra workflows.” - Josh Izaac at Xanadu
“We look forward to more closely integrating Orquestra's workflow management and algorithm libraries with PennyLane.” - Max Radin at Zapata
At Zapata, many scientists write our workflows by hand and run them directly. One aspect we find interesting about this project is how it leverages Orquestra’s scaling skills but automates the workflow creation. Depending on needs, you can call Pennylane from Orquestra or now call Orquestra from Pennylane (or do both for extra credit! 😉). We’re excited to add this tool to our arsenal, as is Xanadu: “Orquestra offers a glimpse into using on-demand quantum cloud platforms, just like the one that we are building at Xanadu. It is really exciting to have PennyLane connect to it!”