Using Reinforcement Learning to Perform Qubit Routing in Quantum Compilers

We show that the qubit-routing problem has a natural interpretation as a reinforcement learning problem. Our results show state-of-the-art performance when qubit routing is treated as abstracted problem, and suggest that reinforcement learning may lead to further gains being made when addressing backend optimisation more generally.

Matteo G. Pozzi, Steven J. Herbert, Akash Sengupta, Robert D. Mullins

Post by admin