EPOC: An Efficient Pulse Generation Framework with Advanced Synthesis for Quantum Circuits

Quantum optimal control has been explored by researchers due to its capability to greatly reduce circuit latency. However, it is also known for its significant computational overhead. Previous works have proposed various methods to accelerate quantum optimal control, such as utilizing GPUs, pre-compilation techniques, and improved circuit libraries. These pulse generation frameworks focus on generating pulses from unitary matrices derived from quantum circuits, without exploring better unitary matrices through equivalent representations. Consequently, they overlook many optimization opportunities by adopting coarse-grained methods. In this work, we propose a novel approach that combines ZX-Calculus, circuit partitioning and circuit synthesis to accelerate pulse generation. Our contribution lies in employing finer granularity in pulse generation, enabling increased parallelism and decreased latency in quantum pulses. Finer granularity is achieved by grouping quantum gates and decomposing the resulting unitary matrices into smaller unitary matrices using synthesis techniques. Additionally, we explore further optimization possibilities by continuously optimizing the circuit through the identification of equivalent representations. By adopting these techniques, we achieve a further reduction in circuit latency while only requiring quantum optimal control for relatively small-sized unitary matrices. For the first time, circuit synthesis is introduced into the workflow of quantum optimal control. We are able to achieve 31.74% reduction in latency compared to previous work and a 76.80% reduction compared with the gatebased method to create pulses. Our approach demonstrates the potentials for significant performance improvements in quantum circuits while minimizing the computational overhead associated with quantum optimal control.