Practical and efficient quantum circuit synthesis and transpiling with Reinforcement Learning

2024-05-21

arXiv.org

AI Summary

Quantum circuit compilation faces challenges in jointly optimizing synthesis and routing under stringent hardware constraints, while conventional approaches—such as SAT solvers—are computationally prohibitive. Method: This work pioneers the systematic integration of deep reinforcement learning (specifically PPO and DQN) into end-to-end quantum compilation, incorporating circuit representation learning, topology-aware action space design, and native gate set adaptation. Contribution/Results: The method achieves joint synthesis and routing optimization for circuits up to 65/133 qubits, supporting synthesis on Linear/Clifford/Permutation benchmarks with capacities of 9/11/65 qubits, respectively. It reduces two-qubit gate depth and count significantly compared to SABRE and accelerates compilation by several orders of magnitude over SAT-based methods. By breaking the real-time versus optimality trade-off in hardware-aware compilation, this approach establishes a scalable, efficient compilation paradigm tailored for large-scale noisy intermediate-scale quantum (NISQ) devices.

Technology Category

• Machine Learning: Quantum Machine Learning
• Search and Optimization: Learning to Search
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

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.

Problem

• Enhances quantum circuit synthesis using Reinforcement Learning
• Improves quantum circuit routing efficiency significantly
• Compatible with device constraints, outperforms traditional methods

Innovation

• Reinforcement Learning applied
• Quantum circuit synthesis enhanced
• Significant gate depth reduction