Speaker
Description
Determining Hamiltonian ground states is central to physics, chemistry, and combinatorial optimization, but is often classically intractable due to the exponential growth of Hilbert space. While convergence-guaranteed quantum eigensolvers require high-overlap initial states, standard variational quantum eigensolver (VQE) used to prepare such states requires instance-specific optimization and often suffers from barren plateaus. To address these limitations, we propose an end-to-end VQE framework that combines a graph autoencoder with a classical neural network to generate ansatz parameters that generalize across Hamiltonian families. Numerical experiments demonstrate improved generalization and trainability, with reduced test error and a significantly milder decay of gradient variance.
| Topic | Quantum variational circuits and their applications |
|---|