{"id":"arxiv/2608.19797","url":"https://arxiv.org/abs/2608.19797","tags":["math.NA"],"score":0,"title":"Optimal Sobolev Approximation by Deterministic and Random Shallow Sigmoidal Networks","domain":"math.NA","source":"arxiv","authors":["Zhaohui Fu","Yangshuai Wang"],"reviews":0,"subject":"math.NA","summary":"Shallow networks with prescribed or randomly sampled hidden parameters are widely used as numerical trial spaces, yet their optimal Sobolev approximation power with standard smooth sigmoidal activations in general dimension remains unresolved. We establish the corresponding optimal rates for a class of smooth sigmoidal activations with Schwartz-class derivative decay, including $\\tanh$, the logistic sigmoid, and the error function $erf$. We first construct deterministic direction--offset dictionaries with $M$ features such that every $u\\in H^k(Ω)$ can be approximated with error of order…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-08-20","reactions":{}}