{"id":"arxiv/2610.09357","url":"https://arxiv.org/abs/2610.09357","tags":["math.NA"],"score":0,"title":"Fully Deterministic Shallow Neural Networks with Localized Activations. Part II: Condition Number Estimates","domain":"math.NA","source":"arxiv","authors":["Ran Bi","Weibing Deng"],"reviews":0,"subject":"math.NA","summary":"We prove positive definiteness and quantitative condition-number bounds for continuous Sobolev Gram matrices of deterministic shallow networks with Gaussian and tanh-difference activations. The prescribed feature families include the optimal Sobolev approximation constructions of Part I. On every fixed nonempty bounded Lipschitz domain in $\\R^d$, $d\\ge2$, and for every fixed integer $m\\ge0$, the raw and column-normalized Gram matrices in the full $H^m$ inner product have spectral condition numbers bounded by $\\exp(CN^{2/d}/\\log(N+2))$ for sufficiently large $N$, where $N$ is the feature…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-07","reactions":{}}