On the Distortion of Partitioning Performance by Random Quantum Circuits

TL;DR

This paper investigates how the use of random quantum circuits as benchmarks in hypergraph partitioning studies can misrepresent real-world performance, emphasizing the importance of benchmark selection.

Contribution

It demonstrates that random circuits distort partitioning evaluation metrics and strategy rankings, advocating for structured circuits in benchmarking.

Findings

Random circuits inflate cut costs and distort scaling trends.

Structured circuits better approximate real workload behaviour.

Benchmark choice significantly impacts research conclusions.

Abstract

Hypergraph partitioning is a central component of distributed quantum computing (DQC) compilers. However, due to the limited size of available quantum benchmark suites, many partitioning studies rely on random quantum circuits as evaluation workloads. In this work, we investigate whether such benchmarking practices provide a faithful assessment of partitioner performance. We evaluate a diverse set of state-of-the-art hypergraph partitioning strategies across three circuit origins: real algorithmic circuits, structured generated circuits, and fully random circuits. Our results show that random circuits significantly distort partitioning evaluation. They inflate cut costs, alter scaling trends across QPU counts and circuit sizes, and change the relative ranking of partitioning strategies. In contrast, structured generated circuits exhibit substantially lower distortion, more closely approximating real workload behaviour in cost, scaling, and strategy rankings. These findings demonstrate that benchmark selection directly influences methodological conclusions in DQC research and that random circuits may provide misleading guidance for compiler design.