Speaker
Description
Neural quantum states are among the most expressive variational ansätze for quantum many-body problems, yet evaluating them incurs a digital computational cost that grows with network size. We propose realizing the ansatz directly in analog hardware, using experimentally demonstrated devices as variational wavefunctions within the variational Monte Carlo (VMC) framework. We study two platforms: the microring resonator meshgrid (MRM) and the optical reservoir (OR), a disordered lithium niobate slab that maps phase-encoded spin configurations to high-dimensional speckle features. Benchmarked against a restricted Boltzmann machine (RBM) on frustrated two-dimensional spin models and the H$_2$ molecule, both platforms match the digital baseline. These results establish analog hardware as a viable substrate for neural quantum states.
| Topic | Tensor methods and quantum-inspired machine learning |
|---|---|
| Additional Topic | Quantum state reconstruction from data |