Léo Monbroussou of the University of Edinburgh, and colleagues at the École Polytechnique Fédérale de Lausanne (EPFL) in Lausanne, Switzerland, and Sorbonne Université, have investigated passive linear optics as a restricted model of quantum computation. Passive linear optics offers complexity-theoretic evidence of quantum advantage for sampling tasks and possesses low losses, making it attractive for near-term algorithms. This is particularly relevant as building large-scale, fault-tolerant quantum computers remains a significant engineering challenge. Passive linear optics, utilising photons and linear optical elements like beam splitters and phase shifters, presents a potentially viable pathway to demonstrate quantum effects with fewer physical resources. A growing body of work in qubit architectures has revealed a close connection between barren plateaus, regions in the parameter space where gradients vanish, hindering optimisation, and classical simulability. However, whether an analogous tradeoff exists for bosonic systems, such as those employing photons, remains largely unexplored. The team are building on a recently developed representation-theoretic framework to address this gap in understanding, aiming to characterise the limits of classical simulation for these systems. Polynomial scaling of expectation values unlocks improved quantum verification Researchers from University of Edinburgh, Sorbonne University, PSL University, Terra Quantum AG, and Institute of Physics have identified a pathway to potentially exceed the capabilities of existing classical simulation methods for quantum computation. Here, ‘n’ represents the number of modes in the photonic circuit. This improvement is significant because the concentration of expectation values dictates how easily a quantum state can be distinguished from a classical probability distribution; a highly concentrated signal makes verification easier. The ability to move from exponential to polynomial scaling in the signal component represents a substantial reduction in the computational resources required to verify quantum advantage. This advancement is key as it addresses a vital barrier in verifying quantum advantage, where demonstrating a