Graph optimisation on neutral atom array with local addressing

Neutral atoms have emerged as a powerful and scalable platform for quantum computing, offering the ability to generate large numbers of identical and high quality qubits in reconfigurable arrays. By coupling atoms to highly excited Rydberg states with strong, long-range dipole-dipole interactions this system can natively implement maximum independent set (MIS) graph problems on a unit disk graph, providing a route to performing analogue optimisation of real problems however with large systems required to reach a regime competitive against current classical optimisation protocols.

To extend this approach to explore optimisation of a wider class of problems including weighted graphs and quadratic unconstrained binary optimisation (QuBO), it is necessary to introduce locally addressed light-shifts to enable weighting of nodes during the optimisation process, enabling problems to be encoded with at worst a quadratic resource overhead.

In this talk we present work to develop a large-scale system for quantum computing and annealing, and show preliminary results highlighting our ability to implement small-scale demonstrations of weighted graph optimisation using programmable local light-shifts across an atom array. We introduce a hybrid annealing process combining global addressing with ramped light shifts, and outline prospects for scaling this approach to larger graph problems as a potential pathway to quantum utility.