beSnake: A routing algorithm for scalable spin-qubit architectures

TL;DR

beSnake is a scalable qubit routing algorithm for spin-qubit architectures that optimizes execution time and fidelity by incorporating shuttle operations, managing diverse topologies, and dynamically adjusting to noise and parallel tasks.

Contribution

This work introduces beSnake, a novel routing algorithm that combines shuttle operations with traditional methods, tailored for scalable spin-qubit architectures, improving speed and efficiency.

Findings

Achieves up to 80% reduction in gate overhead

Reduces circuit depth by up to 54%

Routes up to 1,000 qubits with 8.33 times faster times

Abstract

As quantum computing devices increase in size with respect to the number of qubits, two-qubit interactions become more challenging, necessitating innovative and scalable qubit routing solutions. In this work, we introduce beSnake, a novel algorithm specifically designed to address the intricate qubit routing challenges in scalable spin-qubit architectures. Unlike traditional methods in superconducting architectures that solely rely on SWAP operations, beSnake also incorporates the shuttle operation to optimize the execution time and fidelity of quantum circuits and achieves fast computation times of the routing task itself. Employing a simple breadth-first search approach, beSnake effectively manages the restrictions created by diverse topologies and qubit positions acting as obstacles, for up to 72\% qubit density. It also has the option to adjust the level of optimization and to dynamically tackle parallelized routing tasks, all the while maintaining noise awareness. Our simulations demonstrate beSnake's advantage over an existing routing solution on random circuits and real quantum algorithms with up to $1,000$ qubits, showing an average improvement of up to $80\%$ in gate overhead and $54\%$ in depth overhead, and up to $8.33$ times faster routing times.