Autoregressive Neural Quantum State studies of the Su-Schrieffer-Heeger Model

Autoregressive Neural Quantum States (AR-NQS) offer a promising route to studying the ground states of quantum lattice systems. However, their performance is often limited by the underlying sequence model's ability to handle long-range dependencies. We propose a new AR-NQS architecture that leverages recent advances in State-Space Models (SSMs), which are designed to capture long-range correlations more efficiently than traditional RNNs or Transformers. We apply this SSM-based wavefunction ansatz, combined with variational Monte Carlo, to investigate the phase diagram of the 2D Su-Schrieffer-Heeger model, which has only recently begun to be explored. In this talk, we will introduce the SSM-NQS architecture, and present our findings on the emergent phases and quantum critical points.