Entanglement-assisted Hamiltonian dynamics learning

Approximating the dynamics of a complex many-body Hamiltonian with a simpler effective model lies at the interface of quantum Hamiltonian learning and quantum simulation. In this context, quantum generative adversarial networks (QGANs) have been shown to outperform standard Trotter-based approximations. However, their performance is often hindered by training plateaus that become increasingly severe with system size. To overcome these limitations, we propose an entanglement-assisted learning strategy that couples a single randomly initialized auxiliary qubit to the learning system,  strongly increasing the expressivity of the model and the whole learning process.