Combinatorial Optimization on Rydberg Atom Arrays
End-to-end study of neutral-atom quantum optimization — from QUBO graph construction to error mitigation and approximation-resolved performance benchmarking.
Neutral-atom quantum processors, such as those built on Rydberg-blockade physics, provide a natural hardware graph for solving combinatorial optimization problems. This project spans three interconnected threads, all carried out in Prof. Jaewook Ahn’s group at KAIST.
QUBO Mapping via Rydberg Graphs
Investigated how QUBO (Quadratic Unconstrained Binary Optimization) problems can be encoded into Rydberg-atom graphs and contributed to optimizing atom arrangements under no-local-detuning hardware constraints.
Published: Advanced Quantum Technologies, 2024. DOI: 10.1002/qute.202300398
Deterministic Error Mitigation for MIS
Developed a DEM protocol using a physically informed binomial Hamming-shell model to evaluate Rydberg MIS experiments on Pasqal Fresnel hardware. Derived an entropy-controlled processing-cost scaling (2N H₂(p)) as a rigorous classical brute-force baseline, and identified a quantum–classical crossover around N ≈ 13.
Preprint: arXiv:2602.05432
Approximation-Dependent Performance Benchmarking
Developing an approximation-ratio-resolved shots-to-solution framework, STS(r), for hybrid neutral-atom optimization on QuEra Aquila. Post-processed outputs are modeled by a degeneracy-weighted Hamming-shell distribution governed by a single effective quality parameter β̃. Adiabatic annealing consistently exceeds an excitation-matched random baseline across system sizes N ∈ [30, 125], with measured advantage Δβ̃ = 0.23–0.40. This yields two distinct scaling regimes: exponential shots-to-solution advantage for near-exact targets (r → 1), vanishing for relaxed targets (r ≪ 1).
Status: In preparation
Keywords: Rydberg atoms · MIS · QUBO · Error mitigation · Hamming shells · STS(r) · QuEra Aquila · Pasqal Fresnel