DevQCC: Device-Aware Quantum Circuit Cutting framework with applications in quantum machine learning

Himanshu Sahu, Hari Prabhat Gupta, Vishnu Vardhan Puvvada, Rahul Mishra

Abstract: Quantum circuit cutting is a foundational technique that enables the execution of large quantum circuits on devices with limited capacity. Existing methods, however, are hardware-agnostic, focusing solely on minimizing wire cuts, gate cuts, and reconstruction costs. This often leads to suboptimal or infeasible solutions when executed on quantum hardware due to overlooked factors such as qubit connectivity, native gates, and device noise. To address this, we propose a Device-Aware Quantum Circuit Cutting (DevQCC) framework incorporating hardware-specific information into the circuit cutting process. DevQCC with its three-phase architecture provides a circuit cutting framework that improves fidelity of the solution. In Phase 1, DevQCC searches for all subcircuit sets which satisfies the hardware capacity and other constraints using mixed integer programming. In Phase 2, DevQCC identifies an optimal mapping subcircuit set to the device set using a device-aware partition mapping technique, considering noise, topology, and qubit connectivity. In Phase 3, distributed execution of subcircuits is performed using dependency graph structures, followed by classical post-processing for circuit reconstruction. DevQCC is further extended with DevQCC-state vector (SV) for scalable SV simulations and DevQCC quantum machine learning (QML) to efficiently train large QML models. Our experimental evaluations demonstrate that DevQCC significantly outperforms state-of-the-art circuit cutting methods on diverse quantum computing benchmarks. Notably, it improves fidelity, scalability, and execution feasibility across various hardware configurations. The results for DevQCC-SV and DevQCC-QML validate the effectiveness of the proposed methods.

External IDs:dblp:journals/qmi/SahuGPM25