Quantum circuit partition as a maze: emerging percolation transition via path finding

Abstract:In quantum circuit optimization, circuit partitioning enables the optimization process to be parallelized across multiple devices. Each device is responsible for either reducing the number of selected gates or simplifying the local circuit structure. Most existing approaches to circuit partitioning are quantum-distribution-oriented and rely on splitting CNOT gates by introducing mid-circuit measurements and qubit resets. Currently, there is no criterion to determine how a circuit can be optimally partitioned without removing the CNOT gates for circuit optimization purposes. To address this challenge, we formalize the partition problem as a cutting path through a maze, where the CNOT gates represent the walls. We show that the existence of such a path separates quantum circuits into two classes through a percolation phase transition. In particular, it distinguishes a partitionable regime from a nonpartitionable one, arising from qubit permutations. Such permutations are generated by simulated annealing. We analyze its effect on the CNOT cluster from the perspective of network science and distribution analysis. Our results show that partitioning into two CNOT clusters is possible when the number of CNOTs is almost equal to the number of qubits. Based on this observation, we provide a scalable and practical criterion for identifying whether such a partition exists. Overall, our framework provides theoretical and numerical insight into circuit partitioning within quantum circuit optimization, forming the basis for algorithmic development.

Main text: 16 pages, 6 figures. Supplementary material: 16 pages, 10 figures. M. Guatto and F. Preti contributed equally to this work

arXiv:2606.04070 [quant-ph]

(or arXiv:2606.04070v1 [quant-ph] for this version)

https://doi.org/10.48550/arXiv.2606.04070 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.