Practical and efficient quantum circuit synthesis and transpiling with Reinforcement Learning

About

This paper demonstrates the integration of Reinforcement Learning (RL) into quantum transpiling workflows, significantly enhancing the synthesis and routing of quantum circuits. By employing RL, we achieve near-optimal synthesis of Linear Function, Clifford, and Permutation circuits, up to 9, 11 and 65 qubits respectively, while being compatible with native device instruction sets and connectivity constraints, and orders of magnitude faster than optimization methods such as SAT solvers. We also achieve significant reductions in two-qubit gate depth and count for circuit routing up to 133 qubits with respect to other routing heuristics such as SABRE. We find the method to be efficient enough to be useful in practice in typical quantum transpiling pipelines. Our results set the stage for further AI-powered enhancements of quantum computing workflows.

David Kremer, Victor Villar, Hanhee Paik, Ivan Duran, Ismael Faro, Juan Cruz-Benito · 2024

Related benchmarks

Task	Dataset	Result
Topology-aware CNOT Synthesis	4-L hardware topology	Average CNOT Count10	6
Topology-aware CNOT Synthesis	4-Y hardware topology	Average CNOT Count8.1	6
Topology-aware CNOT Synthesis	5-L hardware topology	Average CNOT Count16.1	6
Topology-aware CNOT Synthesis	5-T hardware topology	Average CNOT count13.9	6
Topology-aware CNOT Synthesis	6-L hardware topology	Average CNOT Count25.4	6
Topology-aware CNOT Synthesis	6-T hardware topology	Average CNOT Count22.5	6
Topology-aware CNOT Synthesis	6-Y hardware topology	Average CNOT Count21.6	6
Topology-aware CNOT Synthesis	7-Y hardware topology	Avg CNOT Count31	5
Topology-aware CNOT Synthesis	7-L hardware topology	Average CNOT Count37.5	5
Topology-aware CNOT Synthesis	7-T hardware topology	Average CNOT Count34.3	5