Experimental quantum optimization on globally driven neutral atom arrays  

Neutral atom quantum annealers (QAs) are promising platforms for solving combinatorial optimization problems [1]. In particular, NP-hard unit disk maximum weight independent set (UD-MWIS) problems can be natively mapped to the QA. To solve more generic combinatorial optimization problems, one should first encode them as UD-MWIS and then run the optimization on the QA. One of the existing encodings is based on the Parity architecture [2-3], which allows optimization for a wide range of problems. This approach involves two compilation steps: first, compiling a given problem into one with local fields and local constraints using the Parity mapping; then, compiling the constraints into atomic gadgets to enforce them during quantum annealing. This compilation process can be used with QAs employing either local [4] or global [5] addressing driving sweeps. In the local addressing approach, atoms require different detunings during the sweep to encode the problem fields and compensate for unwanted long-range interactions. The global addressing approach uses only global driving sweeps and weakly interacting anchor atoms, avoiding the need for local detunings. These anchor atoms encode the desired fields and compensations.

This study focuses on the experimental implementation of the Parity compilation for a neutral atom QA using global addressing driving sweeps. We demonstrate the performance of those atomic gadgets both separately and in larger layouts on the Pasqal Fresnel device [6]. We investigate the dependence of the optimization results on the annealing sweep parameters, positional noise, and compilation-specific parameters.

References

    1. L. Henriet, et al. Quantum 4 (2020): 327.

    2. W. Lechner, et al. Sci. Adv. 7, e1500838 (2015).

    3. K. Ender, et al. Quantum 7 (2023): 950.

    4. M. Lanthaler, et al. Phys. Rev. Lett. 130, 220601 (2023).

    5. M. Lanthaler, et al. arXiv:2410.03902 (2024).

    6. “Pasqal cloud services documentation,” https://docs.pasqal.com/