A Fast and Adaptable Algorithm for Optimal Multi-Qubit Pathfinding in Quantum Circuit Compilation

Quantum computing has the potential to significantly enhance our ability to simulate and solve complex, classically intractable problems across various fields of research and industry. However, we are currently in the noisy intermediate-scale quantum (NISQ) era, where devices are relatively small and suffer from substantial noise levels, prohibiting large-scale computations. To achieve any quantum advantage in this regime and beyond, it is crucial to minimise the impact of noise from qubit decoherence and two-qubit gates. A direct approach is to improve the optimisation of quantum circuit compilation processes that map circuits onto physical devices, thereby reducing noisy gates and circuit execution times. This work focuses on multi-qubit pathfinding as a critical subroutine within the quantum circuit compilation mapping problem. We introduce an algorithm, modelled using binary integer linear programming, that navigates qubits on quantum hardware optimally with respect to circuit SWAP-gate depth, while also optimising for accumulated gate errors and can be flexibly adapted to various problem modifications. This multi-qubit pathfinding algorithm incorporates considerations for gate-error penalties, SWAP movement constraints, and configurable arrangements of source and target qubit locations and qubit teams. We have benchmarked the algorithm across a variety of quantum hardware layouts, assessing properties such as computational runtimes, solution SWAP depths, and accumulated SWAP-gate error rates. The results demonstrate the algorithm's practical runtimes on current quantum devices and compare its effectiveness across different hardware configurations, providing insights for future quantum hardware design.