Large scale Hybrid Quantum Workflows with PennyLane

Abstract

Understanding the interoperability of classical and quantum workflows will be essential for when quantum computers become widely available. Using PennyLane, the library for differentiable quantum computation, we can already build rich workflows that seamlessly integrate classical and quantum components. This talk will introduce our vision for PennyLane, focusing on differentiable quantum programming in heterogeneous environments. I will discuss our work to date in developing HPC-focused tooling for Pennylane, and introduce our results from simulating a distributed QAOA graph optimization workflow on Perlmutter using Ray and NVIDIA’s cuQuantum library. This work is outlined in our recent preprint https://arxiv.org/abs/2207.14734