Crosstalk, caused by unwanted interactions from the surrounding environment, remains a fundamental challenge in existing superconducting quantum computers (SQCs). We propose a method for qubit placement, connectivity, and logical qubit allocation on tunable-coupler SQCs to eliminate unnecessary qubit connections and optimize resources while reducing crosstalk errors. Existing mitigation methods carry trade-offs, like increasing qubit connectivity or software-based gate scheduling. Our method, the Context-Aware COupler REconfiguration (CA-CORE) compilation method, aligns with application-specific design principles. It optimizes the qubit connections for improved SQC performance, leveraging tunable couplers. Through contextual analysis of qubit correlations, we configure an efficient coupling map considering SQC constraints. We then apply the SWAP-based Bidirectional Heuristic Search (SABRE) qubit mapping method and crosstalk-adaptive scheduling to further optimize the quantum circuit. Our architecture reduces depth by an average of 18% and 27%, and by up to 50% and 60%, compared to lattice and heavy-hex architectures, respectively. With crosstalk optimization through adaptive scheduling, we achieve performance improvements of 35%, 20%, and 160% on fully-enabled grid, lattice, and heavy-hex topologies, respectively.

中文翻译：

---

基于可调谐耦合器的超导量子计算机的上下文感知耦合器重新配置


由周围环境不必要的相互作用引起的串扰仍然是现有超导量子计算机 （SQC） 面临的基本挑战。我们提出了一种在可调谐耦合器 SQC 上放置、连接和逻辑量子比特分配的方法，以消除不必要的量子比特连接并优化资源，同时减少串扰错误。现有的缓解方法需要权衡取舍，例如增加量子比特连接或基于软件的门调度。我们的方法，即上下文感知 COupler REconfiguration （CA-CORE） 编译方法，与特定于应用程序的设计原则一致。它利用可调谐耦合器优化量子比特连接以提高 SQC 性能。通过对量子比特相关性的上下文分析，我们配置了一个考虑 SQC 约束的高效耦合映射。然后，我们应用基于 SWAP 的双向启发式搜索 （SABRE） 量子比特映射方法和串扰自适应调度来进一步优化量子电路。与晶格和重六边形架构相比，我们的架构分别将深度平均降低了 18% 和 27%，以及高达 50% 和 60%。通过自适应调度进行串扰优化，我们在完全启用的网格、晶格和重六边形拓扑上分别实现了 35%、20% 和 160% 的性能提升。