Pasqal has published new research exploring the performance of its neutral atom quantum processors on combinatorial optimization problems, with a focus on creating realistic, classically hard benchmarks. Recognizing the limitations of earlier speedup claims based on oversimplified problems, the study emphasizes the need for rigorous benchmarking against classical solvers used in production. The team applied its methodology to the Maximum Independent Set (MIS) problem on unit-disk graphs—a challenge that appears in fields like wireless networks and molecular design. This research is the first in a three-part series intended to clarify when and how quantum hardware can achieve a practical advantage.
The experiments were run on Pasqal’s Orion Alpha processor, using up to 100 qubits and collecting 100,000 experimental data points. These benchmarks tested both low-density and high-density graph instances to evaluate performance under more complex, industrially relevant conditions. The study compared these quantum results against classical solvers used by Pasqal’s industrial partners such as EDF, Thales, and BMW, providing a realistic picture of current performance gaps. The authors focused not only on outcome accuracy but also on wall-clock time, a key metric for real-world relevance.
Hardware limitations identified in the study include slow repetition rates—where MHz-scale quantum operations are bottlenecked by atom loading and imaging constraints, limiting execution rates to just a few Hz. Pasqal outlines a clear engineering roadmap to address these constraints, including improvements in reservoir loading, assembly algorithms, and fast imaging techniques aimed at reaching kHz-scale repetition rates. Additionally, the company notes the importance of scaling to larger systems; Pasqal has experimentally trapped over 1,100 atoms, and academic demonstrations have reached 6,100 atoms, suggesting a viable path to handling larger problem sizes that are currently intractable for classical solvers.
While quantum advantage for MIS on large, dense graphs has not yet been demonstrated, Pasqal’s results represent a critical step forward. With targeted hardware improvements and continued benchmarking of classically difficult problem instances, neutral atom quantum processors are expected to play a growing role in solving complex combinatorial optimization tasks. Future work from Pasqal will expand these benchmarks into additional application areas where quantum advantage may be even more pronounced.
Read the full research blog here and the corresponding preprint here.
March 21, 2025
Leave A Comment