Neutral Atom Quantum Computing for Unit Commitment

With increasing penetration of renewables and distributed energy resources, combinatorialpower system optimisation problems are becoming increasingly computationally challenging. Quantumcomputing offers a potential alternative approach for exploring these problems and neutral atom systemsin particular represent a promising architecture due to their scalability and flexible connectivity. This workpresents the first investigation and hardware demonstration of neutral-atom quantum computing for thepower system unit commitment problem. We encode the problem on neutral-atom hardware using twoapproaches. First, a static analogue approach using van der Waals interactions between atoms to represent aquadratic unconstrained binary optimisation that approximates the original mixed integer linear constrainedunit commitment formulation. We demonstrate unit-commitment instances on static analogue neutral-atomhardware and show that direct static embeddings are limited to very small systems. Second, a dynamicshuttling gate-based approach implements the problem via the Quantum Approximate Optimisation Algo-rithm (QAOA). For larger problems, we analyse the gate-based approach, simulating QAOA solutions oninstances of up to 28 qubits, exhibiting a median approximation ratio of 0.83. Extrapolating to a 24-time-step, 66-generator instance requiring 7,932 qubits, we estimate that the runtime of a classical solver mayexceed that of a single-shot, 3-layer QAOA implementation; however, the solution quality achievable atthis scale remains uncertain. Our results indicate that solving real-world unit commitment problems using agate-based method will require substantial improvements in hardware fidelity or quantum error correction.We estimate that distance-19 surface-code quantum error correction would require 3.44 million physicalqubits to solve this problem.