QAOA Parameter Transfer for Quantum Optimization in Au-tonomous Mobility
Abstract
Jeremy Mange and Paramsothy Jayakumar
This paper presents an approach for solving complex optimization problems in autonomous mobility through the transfer of optimal parameters in the Quantum Approximate Optimization Algorithm (QAOA) from small to large problem instances. Autonomous mobility problems, such as path planning and resource assignment, can often be formulated as graph problems, enabling the application of QAOA and parameter transfer. We focus on the Max-Cut formulation as a broadly applicable proxy for these graph-based tasks, and demonstrate that QAOA parameters optimized for small graphs can be effectively used for execution on larger graphs with similar graph structures, achieving high-quality solutions with significantly reduced computational effort. Experiments are conducted on both high-performance computing simulators and real quantum hardware (Rigetti Ankaa-2), with graphs ranging from 10 to 80 vertices. Results show that transferred parameters achieve approximation ratios within 3% of full QAOA optimization on average, and in some cases exceed it, confirming the practical viability of this approach for autonomous systems.
