26th Australian Institute of Physics Congress and the Australian and New Zealand Conference on Optics and Photonics

Optical Stochstic Simulator for Demonstration of Unbounded Quantum Memory Advantage

Not scheduled

20m

Belinda Hutchinson Building (The University of Sydney )

Contributed Oral AIP | Quantum Science and Technology (QST)

Description

Stochastic processes are everywhere -- weather, financial markets, traffic -- and simulating them faithfully requires sufficient memory of past observations to predict future behaviour. For processes with a long historical dependence, the memory requirement grows with complexity posing a hard barrier for classical simulators. Quantum simulators offer an alternative through encoding memory in quantum states. States with overlapping future statistics can share memory resources rather than occupying orthogonal dimensions. A quantum simulator can handle more complex processes within the same memory dimension as a classical one. For certain process classes this advantage is unbounded – classical memory grows without limit as process complexity increases, while the quantum memory dimension remains fixed.

So far, experimental realisation of this quantum advantage has been practically limited by an exponential increase in resources with the process complexity. Here, we propose and demonstrate a new experimental architecture that utilizes an optical switch to loop photons through a single simulation gate. This enables any simulation complexity without introducing any new optical elements, opening the door to demonstrating larger quantum advantages, limited only by noise and experimental losses. We demonstrate this approach in an optical quantum simulator where memory is encoded onto the polarisation qubit of a single photon, and the stochastic process is implemented with a reconfigurable two-qubit unitary gate using a displaced Sagnac interferometer.

We demonstrate the quantum advantage by sampling output statistics for multiple processes of increasing complexity. Comparing against the optimal classical simulator restricted to the same memory dimension via Kullback-Leibler divergence rate, we show that where the optimal classical simulator accuracy degrades, our platform provides faithful statistics regardless of complexity. This work establishes a scalable photonic platform for quantum simulation of stochastic processes, paving the way towards quantum simulators capable of tackling processes whose memory requirements place them entirely beyond classical reach.