Hybrid quantum optimization in the context of minimizing traffic congestion
AI Breakdown
Get a structured breakdown of this paper — what it's about, the core idea, and key takeaways for the field.
Abstract
Traffic optimization on roads is a highly complex problem, with one important aspect being minimization of traffic congestion. By mapping to an Ising formulation of the traffic congestion problem, we benchmark solutions obtained from the Quantum Approximate optimization Algorithm (QAOA), a hybrid quantum-classical algorithm. In principle, as the number of QAOA layers approaches infinity, the solutions should reach optimality. On the other hand, short-depth QAOA circuits are known to have limited performance. We show that using tailored initialization techniques encourages the convergence to the desired solution state at lower circuit depths with two and three QAOA layers, thus highlighting the importance of adapting quantum algorithms in the noisy intermediate scale (NISQ) quantum computing era. Moreover, for NISQ devices with limited qubit connectivity and circuit depth, we introduce a heuristic noise-resilient variant of QAOA predicated on the elimination of long-range 2-qubit interactions in the QAOA layers whilst the cost function is unaltered. Our results show that this QAOA variant is surprisingly effective, outperforming QAOA on a physical IBM Quantum computer device.