Expanding Graph Neural Networks for Ultra-Fast Optical Core Network Throughput Prediction to Large Node Scales

Using maximum achievable throughput as an objective, message passing neural networks (MPNN) are applied to larger optical networks (25-100 nodes), enabling physical properties-aware large-scale topology optimisation in record time, reducing computation time by 5 orders of magnitude, with close to perfect throughput correlation (ρ = 0.986).