Speaker
Description
Approximating the dynamics generated by complex many-body Hamiltonians with simpler, experimentally feasible models is a central challenge at the intersection of quantum simulation and quantum learning. Quantum generative adversarial networks (QGANs) have recently emerged as a powerful framework for learning unitary dynamics, often achieving significantly more compact representations than standard Trotter-based approaches. However, their practical performance is frequently limited by training instabilities, including optimization plateaus and convergence to suboptimal local minima, especially as system size increases. In this work, we introduce an entanglement-assisted learning strategy that enhances the training of QGANs through the controlled use of an ancillary qubit. Our approach dynamically enlarges the generator’s Hilbert space by coupling a randomly initialized ancilla to the system mid-training. We show that the combination of entanglement, randomization, and adaptive circuit extension reshapes the optimization landscape, enabling the learning process to escape stagnation and achieve higher fidelities. We analyze these behaviors by computing the bipartite ancilla-system von Neumann entanglement entropy during training, and utilizing Common Random Numbers to demonstrate the amplification of gradient variance upon ancilla insertion. Through numerical simulations, we demonstrate that this method substantially improves the approximation of nontrivial target dynamics (including Heisenberg models) using hardware-efficient ansatzes with limited locality. We further analyze the roles of ancilla connectivity, timing of introduction, and parameter initialization, identifying the key ingredients responsible for the observed performance gains. Our results highlight a general, theoretically grounded, and potentially scalable strategy for mitigating optimization bottlenecks in variational quantum algorithms, with applications in quantum simulation and machine learning.
| Topic | Theoretical foundations of quantum learning |
|---|---|
| Additional Topic | Quantum variational circuits and their applications |