Search papers, labs, and topics across Lattice.
The paper introduces QFlowNet, a novel framework for unitary synthesis that combines Generative Flow Networks (GFlowNets) with Transformers to efficiently learn from sparse reward signals and generate diverse quantum gate sequences. QFlowNet leverages GFlowNets to sample solutions proportional to their reward, addressing the limitations of RL-based approaches that often converge to a single policy. By using Transformers as encoders, the framework captures the non-local structure of unitary matrices, enabling the policy network to learn from dense latent representations.
Ditch RL's single-solution rut: QFlowNet leverages Generative Flow Networks to discover a diverse set of high-performing quantum circuits for unitary synthesis.
Unitary Synthesis, the decomposition of a unitary matrix into a sequence of quantum gates, is a fundamental challenge in quantum compilation. Prevailing reinforcement learning(RL) approaches are often hampered by sparse reward signals, which necessitate complex reward shaping or long training times, and typically converge to a single policy, lacking solution diversity. In this work, we propose QFlowNet, a novel framework that learns efficiently from sparse signals by pairing a Generative Flow Network (GFlowNet) with Transformers. Our approach addresses two key challenges. First, the GFlowNet framework is fundamentally designed to learn a diverse policy that samples solutions proportional to their reward, overcoming the single-solution limitation of RL while offering faster inference than other generative models like diffusion. Second, the Transformers act as a powerful encoder, capturing the non-local structure of unitary matrices and compressing a high-dimensional state into a dense latent representation for the policy network. Our agent achieves an overall success rate of 99.7% on a 3-qubit benchmark(lengths 1-12) and discovers a diverse set of compact circuits, establishing QFlowNet as an efficient and diverse paradigm for unitary synthesis.
