A fidelity-driven approach to quantum circuit partitioning via weighted hypergraphs for noise-resilient computation

TL;DR

Fidelipart is a fidelity-aware hypergraph-based quantum circuit partitioning framework that significantly reduces error-prone operations and enhances circuit fidelity on NISQ devices.

Contribution

It introduces a novel hypergraph model incorporating gate error rates for optimized quantum circuit partitioning, improving fidelity over existing heuristics.

Findings

Achieved up to 100% SWAP gate reduction

Reduced cut qubits by up to 52.2%

Estimated fidelity gains from 27.3% to over 250%

Abstract

Effective circuit partitioning is critical for Noisy Intermediate-Scale Quantum (NISQ) devices, which are hampered by high error rates and limited qubit connectivity. Standard partitioning heuristics often neglect gate-specific error impacts, leading to suboptimal divisions with significant communication overhead and reduced fidelity. This paper introduces Fidelipart, a novel framework that transforms quantum circuits into a fidelity-aware hypergraph. In this model, gate error rates and structural dependencies inform the weights of nodes (gates) and hyperedges (representing multi-qubit interactions and qubit timelines), guiding an Mt-KaHyPar partitioner to minimize cuts through error-prone operations. We evaluated Fidelipart against BQSKit's QuickPartitioner on 6-qubit/22-gate, 10-qubit/55-gate, and 24-qubit/88-gate benchmarks under a linear topology with a consistent local contiguous re-mapping strategy. Results demonstrate Fidelipart's superior performance. It achieved SWAP gate reductions ranging from 77.3% to 100% and up to a 52.2% decrease in cut qubits. These physical improvements directly translated to estimated fidelity gains ranging from 27.3% to over 250%. While Fidelipart showed a modest runtime increase of 8-13% and variable effects on maximum partition depth, its substantial enhancement of circuit fidelity highlights the significant benefits of integrating detailed error-awareness into the partitioning process for more reliable NISQ computations.