{"id":"arxiv/2605.27300","url":"https://arxiv.org/abs/2605.27300","tags":["quant-ph"],"score":0,"title":"Deterministic Mapping of Topological Phases via Autoregressive Exogenous Neural Networks","domain":"quant-ph","source":"arxiv","authors":["Graciana Puentes"],"reviews":0,"subject":"quant-ph","summary":"We report a comparative analysis of three dynamic neural network (NN) architectures -- NAR, NARX, and NIO -- to evaluate their efficiency in estimating the critical-measurement-strength parameter ($c_{crit}$) characterizing topological phase transitions in geometric phases induced by weak measurements. Our results demonstrate that the NARX architecture achieves superior predictive fidelity, reaching a Mean Squared Error (MSE) of $10^{-27}$ -- the limit of numerical precision -- at an optimal delay of $d=1$. This exceptional performance implies the identification of a perfect functional…","category":"research","comments":0,"disputed":false,"division":"physics","mentions":0,"discussed":0,"published":"2026-05-26","reactions":{}}